Search

Comparative molecular field analysis (CoMFA) and quantum chemical calculations were performed on cycloguanil (Cyc) derivatives of the wild type and the quadruple mutant (Asn51Ile, Cys59Arg, Ser108Asn, Ile164Leu) of Plasmodium falciparum dihydrofolate reductase (PfDHFR). The represented CoMFA models of wild type (r(2) = 0.727 and r(2) = 0.985) and mutant type (r(2) = 0.786 and r(2) = 0.979) can describe the differences of the Cyc structural requirements for the two types of PfDHFR enzymes and can be useful to guide the design of new inhibitors. Moreover, the obtained particular interaction energies between the Cyc and the surrounding residues in the binding pocket indicated that Asn108 of mutant enzyme was the cause of Cyc resistance by producing steric clash with p-Cl of Cyc. Consequently, comparing the energy contributions with the potent flexible WR99210 inhibitor, it was found that the key mutant residue, Asn108, demonstrates attractive interaction with this inhibitor and some residues, Leu46, Ile112, Pro113, Phe116, and Leu119, seem to perform as second binding site with WR99210. Therefore, quantum chemical calculations can be useful for investigating residue interactions to clarify the cause of drug resistance.

The goal of a quantitative structure-activity relationship (QSAR) model is to encode the relationship between molecular structure and biological activity or physical property. Based on this encoding, such models can be used for predictive purposes. Assuming the use of relevant and meaningful descriptors, and a statistically significant model, extraction of the encoded structure-activity relationships (SARs) can provide insight into what makes a molecule active or inactive. Such analyses by QSAR models are useful in a number of scenarios, such as suggesting structural modifications to enhance activity, explanation of outliers and exploratory analysis of novel SARs. In this paper we discuss the need for interpretation and an overview of the factors that affect interpretability of QSAR models. We then describe interpretation protocols for different types of models, highlighting the different types of interpretations, ranging from very broad, global, trends to very specific, case-by-case, descriptions of the SAR, using examples from the training set. Finally, we discuss a number of case studies where workers have provide some form of interpretation of a QSAR model.

We introduce PHASE, a highly flexible system for common pharmacophore identification and assessment, 3D QSAR model development, and 3D database creation and searching. The primary workflows and tasks supported by PHASE are described, and details of the underlying scientific methodologies are provided. Using results from previously published investigations, PHASE is compared directly to other ligand-based software for its ability to identify target pharmacophores, rationalize structure-activity data, and predict activities of external compounds.

Three-dimensional quantitative structure-activity relationship (3D-QSAR) modelling using comparative molecular similarity indices analysis (CoMSIA) was applied to a series of 406 structurally diverse dihydrofolate reductase (DHFR) inhibitors from Pneumocystis carinii (pc) and rat liver (rl). X-ray crystal structures of three inhibitors bound to pcDHFR were used for defining the alignment rule. For pcDHFR, a QSAR model containing 6 components was selected using leave-10%-out cross-validation (n= 240, q2 = 0.65), while a 4-component model was selected for rlDHFR (n= 237, q2 = 0.63); both include steric, electrostatic and hydrophobic contributions. The models were validated using a large test set, designed to maximise its diversity and to verify the predictive accuracy of models for extrapolation. The pcDHFR model has r2 = 0.60 and mean absolute error (MAE) = 0.57 for the test set after removing 4 outliers, and the rlDHFR model has r2 = 0.60 and MAE = 0.69 after removing 4 test set outliers. In addition, classification models predicting selectivity for pcDHFR over rlDHFR were developed using soft independent modelling by class analogy (SIMCA), with a selectivity ratio of 2 (IC50,rlDHFR/ IC50,pcDHFR) used for delimiting classes. A 5-component model including steric and electrostatic contributions has cross-validated and test set classification rates of 0.67 and 0.68 for selective inhibitors, and 0.85 and 0.72 for unselective inhibitors. The predictive accuracy of models, together with the identification of important contributions in QSAR and classification models, offer the possibility of designing potent selective inhibitors and estimating their activity prior to synthesis.

Genetic algorithms have properties which make them attractive in de novo drug design. Like other de novo design programs, genetic algorithms require a method to reduce the enormous search space of possible compounds. Most often this is done using information from known ligands. We have developed the ADAPT program, a genetic algorithm which uses molecular interactions evaluated with docking calculations as a fitness function to reduce the search space. ADAPT does not require information about known ligands. The program takes an initial set of compounds and iteratively builds new compounds based on the fitness scores of the previous set of compounds. We describe the particulars of the ADAPT algorithm and its application to three well-studied target systems. We also show that the strategies of enhanced local sampling and re-introducing diversity to the compound population during the design cycle provide better results than conventional genetic algorithm protocols.

Free energy force field (FEFF) 3D-QSAR analysis was used to construct ligand-receptor binding models for a set of 18 structurally diverse antifolates including pyrimethamine, cycloguanil, methotrexate, aminopterin and trimethoprim, and 13 pyrrolo[2,3-d]pyrimidines. The molecular target ('receptor') used was a 3D-homology model of a specific mutant type of Plasmodium falciparum (Pf) dihydrofolate reductase (DHFR). The dependent variable of the 3D-QSAR models is the IC50 inhibition constant for the specific mutant type of PfDHFR. The independent variables of the 3D-QSAR models (the descriptors) are scaled energy terms of a modified first-generation AMBER force field combined with a hydration shell aqueous solvation model and a collection of 2D-QSAR descriptors often used in QSAR studies. Multiple temperature molecular dynamics simulation (MDS) and the genetic function approximation (GFA) were employed using partial least square (PLS) and multidimensional linear regressions as the fitting functions to develop FEFF 3D-QSAR models for the binding process. The significant FEFF energy terms in the best 3D-QSAR models include energy contributions of the direct ligand-receptor interaction. Some changes in conformational energy terms of the ligand due to binding to the enzyme are also found to be important descriptors. The FEFF 3D-QSAR models indicate some structural features perhaps relevant to the mechanism of resistance of the PfDHFR to current antimalarials. The FEFF 3D-QSAR models are also compared to receptor-independent (RI) 4D-QSAR models developed in an earlier study and subsequently refined using recently developed generalized alignment rules.

A series of 5-(arylthio)-2,4-diaminoquinazolines are known as selective inhibitors of dihydrofolate reductase (DHFR) from Candida albicans. We have performed docking and molecular dynamics simulations of these inhibitors with C. albicans and human DHFR to understand the basis for selectivity of these agents. Study was performed on a selected set of 10 compounds with variation in structure and activity. Molecular dynamics simulations were performed at 300 K for 45 ps with equilibration for 10 ps. Trajectory data was analyzed on the basis of hydrogen bond interactions, energy of binding and conformational energy difference. The results indicate that hydrogen bonds formed between the compound and the active site residues are responsible for inhibition and higher potency. The selectivity index, i.e the ratio of I50 against human DHFR to I50 against fungal DHFR, is mainly determined by the conformation adapted by the compounds within the active site of two enzymes. Since the human DHFR active site is rigid, the compound is trapped in a higher energy conformation. This energy difference between the two conformations deltaE mainly governs the selectivity against fungal DHFR. The information generated from this analysis of potency and selectivity should be useful for further work in the area of antifungal research.

Dihydrofolate reductase has successfully been used as a drug target in the area of anti-cancer, anti-bacterial and anti-malarial chemotherapy. Little has been done to evaluate it as a drug target for treatment of the trypanosomiases and leishmaniasis. A crystal structure of Leishmania major dihydrofolate reductase has been published. In this paper, we describe the modelling of Trypanosoma cruzi and Trypanosoma brucei dihydrofolate reductases based on this crystal structure. These structures and models have been used in the comparison of protozoan, bacterial and human enzymes in order to highlight the different features that can be used in the design of selective anti-protozoan agents. Comparison has been made between residues present in the active site, the accessibility of these residues, charge distribution in the active site, and the shape and size of the active sites. Whilst there is a high degree of similarity between protozoan, human and bacterial dihydrofolate reductase active sites, there are differences that provide potential for selective drug design. In particular, we have identified a set of residues which may be important for selective drug design and identified a larger binding pocket in the protozoan than the human and bacterial enzymes.

Binding energy calculations for complexes of mutant and wild-type human dihydrofolate reductases with 2,4-diaminopteridine and 2,4-diaminoquinazoline inhibitors are reported. Quantitative insight into binding energetics of these molecules is obtained from calculations based on force field energy evaluation and thermal sampling by molecular dynamics simulations. The calculated affinity of methotrexate for wild-type and mutant enzymes is reasonably well reproduced. Truncation of the methotrexate glutamate tail results in a loss of affinity by several orders of magnitude. No major difference in binding strength is predicted between the pteridines and the quinazolìnes, while the N-methyl group present in methotrexate appears to confer significantly stronger binding. The recent improvement, which is used here, of our linear interaction energy method for binding affinity prediction, as well as problems with treating charged and flexible ligands are discussed. This approach should be suitable in a drug discovery context for prediction of binding energies of new inhibitors prior to their synthesis, when some information about the binding mode is available.

A central problem in forming accurate regression equations in QSAR studies is the selection of appropriate descriptors for the compounds under study. We describe a novel procedure for using inductive logic programming (ILP) to discover new indicator variables (attributes) for QSAR problems, and show that these improve the accuracy of the derived regression equations. ILP techniques have previously been shown to work well on drug design problems where there is a large structural component or where clear comprehensible rules are required. However, ILP techniques have had the disadvantage of only being able to make qualitative predictions (e.g. active, inactive) and not to predict real numbers (regression). We unify ILP and linear regression techniques to give a QSAR method that has the strength of ILP at describing steric structure, with the familiarity and power of linear regression. We evaluated the utility of this new QSAR technique by examining the prediction of biological activity with and without the addition of new structural indicator variables formed by ILP. In three out of five datasets examined the addition of ILP variables produced statistically better results (P < 0.01) over the original description. The new ILP variables did not increase the overall complexity of the derived QSAR equations and added insight into possible mechanisms of action. We conclude that ILP can aid in the process of drug design.

This paper describes the further development of the functionality of our in-house de novo design program, PRO&#95;LIGAND. In particular, attention is focused on the implementation and validation of the 'direct tweak' method for the construction of conformationally flexible molecules, such as peptides, from molecular fragments. This flexible fitting method is compared to the original method based on libraries of prestored conformations for each fragment. It is shown that the directed tweak method produces results of comparable quality, with significant time savings. By removing the need to generate a set of representative conformers for any new library fragment, the flexible fitting method increases the speed and simplicity with which new fragments can be included in a fragment library and also reduces the disk space required for library storage. A further improvement to the molecular construction process within PRO&#95;LIGAND is the inclusion of a constrained minimisation procedure which relaxes fragments onto the design model and can be used to reject highly strained structures during the structure generation phase. This relaxation is shown to be very useful in simple test cases, but restricts diversity for more realistic examples. The advantages and disadvantages of these additions to the PRO&#95;LIGAND methodology are illustrated by three examples: similar design to an alpha helix region of dihydrofolate reductase, complementary design to the active site of HIV-1 protease and similar design to an epitope region of lysozyme.

A new automated procedure to improve the predictive quality of CoMFA models for both training and test sets is described. A model of greater consistency is generated by performing small reorientations of the underlying molecules for which too low activities are calculated. In order to predict activities of test compounds, the most similar molecules in the previously optimized model are identified and used as a basis for the prediction. This method has been applied to two independent sets of dihydrofolate reductase inhibitors (80 compounds each, serving as training sets), resulting in a significant increase of the cross-validated r2 value. For both models, the predictive r2 value for a test set consisting of 70 compounds was improved substantially.

The steric descriptors commonly used in CoMFA--Lennard-Jones 6-12 potential-derived interaction energies calculated between a probe atom and the molecules under investigation--have been replaced by variables indicating the presence of an atom of a particular molecule in predefined volume elements (cubes) within the region enclosing the ensemble of superimposed molecules. The resulting 'atom indicator vectors' were used as steric fields in the subsequent PLS analyses, with and without inclusion of electrostatic Coulomb interaction-derived fields. Application of this method to five training sets (80 compounds each) and five test sets (60 compounds each), randomly selected from an ensemble of 256 dihydrofolate reductase inhibitors, leads to models of significantly higher consistency, as indicated by the cross-validated r2 values for the training sets and the predictive r2 values for the test sets.

We present a system, FLOG (Flexible Ligands Oriented on Grid), that searches a database of 3D coordinates to find molecules complementary to a macromolecular receptor of known 3D structure. The philosophy of FLOG is similar to that reported for DOCK [Shoichet, B.K. et al., J. Comput. Chem., 13 (1992) 380]. In common with that system, we use a match center representation of the volume of the binding cavity and we use a clique-finding algorithm to generate trial orientations of each candidate ligand in the binding site. Also we use a grid representation of the receptor to assess the fit of each orientation. We have introduced a number of novel features within this paradigm. First, we address ligand flexibility by including up to 25 explicit conformations of each structure in our databases. Nonhydrogen atoms in each database entry are assigned one of seven atom types (anion, cation, donor, acceptor, polar, hydrophobic and other) based on their local bonded chemical environments. Second, we have devised a new grid-based scoring function compatible with this 'heavy atom' representation of the ligands. This includes several potentials (electrostatic, hydrogen bonding, hydrophobic and van der Waals) calculated from the location of the receptor atoms. Third, we have improved the fitting stage of the search. Initial dockings are generated with a more efficient clique-finding algorithm. This new algorithm includes the concept of 'essential points', match centers that must be paired with a ligand atom. Also, we introduce the use of a rapid simplex-based rigid-body optimizer to refine the orientations. We demonstrate, using dihydrofolate reductase as a sample receptor, that the FLOG system can select known inhibitors from a large database of drug-like compounds.