Comparison of structure fingerprint and molecular interaction field based methods in explaining biological similarity of small molecules in cell-based screens

In this work, we calculated the pair wise chemical similarity for a subset of small molecules screened against the NCI60 cancer cell line panel. Four different compound similarity calculation methods were used: Brutus, GRIND, Daylight and UNITY. The chemical similarity scores of each method were related to the biological similarity data set. The same was done also for combinations of methods. In the end, we had an estimate of biological similarity for a given chemical similarity score or combinations thereof. The data from above was used to identify chemical similarity ranges where combining two or more methods (data fusion) led to synergy. The results were also applied in ligand-based virtual screening using the DUD data set. In respect to their ability to enrich biologically similar compound pairs, the ranking of the four methods in descending performance is UNITY, Daylight, Brutus and GRIND. Combining methods resulted always in positive synergy within a restricted range of chemical similarity scores. We observed no negative synergy. We also noted that combining three or four methods had only limited added advantage compared to combining just two. In the virtual screening, using the estimated biological similarity for ranking compounds produced more consistent results than using the methods in isolation.     

Molecular dynamics simulations for human CAR inverse agonists

Constitutive androstane receptor (CAR), along with pregnane x receptor (PXR), is an important metabolic sensor in the hepatocytes. Like all other nuclear receptors (NRs), CAR works in concert with coregulator proteins, coactivators, and corepressors which bind to the NRs. The main basis for the receptor to distinguish between coactivators and corepressors is the position of the C-terminal helix 12 (H12), which is determined by the bound NR ligand. CAR, having constitutive activity, can be repressed or further activated by its ligands. Crystal structure of human CAR bound to an agonist and a coactivator peptide is available, but no structural information on an inverse agonist-bound human CAR and a corepressor exists. In our previous molecular dynamics (MD) studies, no corepressor peptide was included. Therefore, probably due to the strong interactions which keep the relatively short H12 of CAR in the active position, the structural changes elicited by inverse agonists were very subtle, and H12 of CAR seemed to more or less retain its active conformation. Here, we have run a series of MD simulations to study the movement of H12 in the presence of both activating and repressing ligands as well as a corepressor peptide. The presence of the corepressor on the coregulator surface of CAR induced a clear shift of H12 of the inverse agonists-bound CAR. In general, H12 moved toward H10 and not away from the ligand binding domain, as seen in some other NRs. However, H12 of CAR is short enough that this movement seems to be adequate to accommodate the binding of the corepressor.     

A comparative molecular field analysis of cytochrome P450 2A5 and 2A6 inhibitors

Structure-activity relationships of 23 P450 2A5 and 2A6 inhibitors were analysed using the CoMFA [1] and GOLPE/GRID with smart region definition (SRD) [2]. The predictive power of the resulting models was validated using five compounds not belonging to the model set. All models have high internal and external predictive power and resulting 3D-QSAR models are supporting each other. Both Sybyl and GOLPE highlight properties near lactone moiety to be important for 2A5 and 2A6 inhibition. Another important feature for pIC₅₀ was the size of the substituent in the 7-positon of coumarin. The models suggest that the 2A5 binding site is larger that that of 2A6 due to larger steric regions in the CoMFA coefficient maps and corresponding GOLPE maps. In addition, the maps reveal that 2A6 disfavours negative charge near the lactone moiety of coumarin.     

Comparing the quality and predictiveness between 3D QSAR models obtained from manual and automated alignment

A set of 113 flexible cyclic urea inhibitors of human immunodeficiency virus protease (HIV-1 PR) was used to compare the quality and predictive power of CoMFA and CoMSIA models for manually or automatically aligned inhibitor set. Inhibitors that were aligned automatically with molecular docking were in agreement with information obtained from existing X-ray structures. Both alignment methods produced statistically significant CoMFA and CoMSIA models, with the best q(2) value being 0.649 and the best predictive r(2) being 0.754. The manual alignment gave statistically higher values, whereas the automated alignment gave more robust models for predicting the activities of an external inhibitor set. Both models utilized similar amino acids in the HIV-1 PR active site, supporting the idea that hydrogen bonds form between an inhibitor and the backbone carbonyl oxygens of Gly48 and Gly48' and also the backbone NH group of Asp30, Gly48, Asp29', and Gly48' of the enzyme. These results suggest that an automated inhibitor alignment can yield predictive 3D QSAR models that are well comparable to manual methods. Thus, an automated alignment method in creating 3D QSAR models is encouragable when a well-characterized structure of the target protein is available.     

Comparison of Experimental Methods and Theoretical Calculations on Crystal Energies of ‘Isoenergetic’ Polymorphs of Cimetidine

The thermodynamic energy relationship between two crystal modifications of cimetidine was investigated and compared with differences in their processing properties with respect to transformation from one modification to the other.The crystal energies of the two modifications A and D were found to be almost identical and therefore the polymorphs are regarded as virtually isoenergetic crystals. This statement is based on DSC measurements of the melting points and of the enthalpies of fusion for the two crystal forms, which enable the calculation of the Gibbs free energy functions. Furthermore, the statement is supported by measurements of the enthalpies of solution in two different solvents. Both DSC and solution experiments reveal a slightly higher stability of the D modification with respect to the A form. In addition, tribomechanical treatment also indicates modification D to be the more stable one, as well as the higher density of the D form. No transformation during DSC at low heating rate was found which could be used in a stability consideration.As the explicit crystal structures of the two modifications are resolved, it was possible to calculate crystal energies theoretically as well. The theoretical results showed a remarkable difference in the crystal energies at zero degree Kelvin. Furthermore, they were just contradicting experimental findings by stating A being more stable than D. Possible reasons for this discrepancy and the feasibility of today's calculation methods with respect to prediction of stability properties are discussed.This revised version was published online in November 2005 with corrections to the Cover Date.     

BRUTUS: optimization of a grid-based similarity function for rigid-body molecular superposition. II. Description and characterization

Finding novel lead molecules is one of the primary goals in early phases of drug discovery projects. However, structurally dissimilar compounds may exhibit similar biological activity, and finding new and structurally diverse lead compounds is difficult for computer algorithms. Molecular energy fields are appropriate for finding structurally novel molecules, but they are demanding to calculate and this limits their usefulness in virtual screening of large chemical databases. In our approach, energy fields are computed only once per superposition and a simple interpolation scheme is devised to allow coarse energy field lattices having fewer grid points to be used without any significant loss of accuracy. The resulting processing speed of about 0.25 s per conformation on a 2.4 GHz Intel Pentium processor allows the method to be used for virtual screening on commonly available desktop machines. Moreover, the results indicate that grid-based superposition methods could be efficiently used for the virtual screening of compound libraries.     

Characterization of the sulfhydryl-sensitive site in the enzyme responsible for hydrolysis of 2-arachidonoyl-glycerol in rat cerebellar membranes

We have previously reported that the endocannabinoid, 2-arachidonoyl-glycerol (2-AG), is hydrolyzed in rat cerebellar membranes by monoglyceride lipase (MGL)-like enzymatic activity. The present study shows that, like MGL, 2-AG-degrading enzymatic activity is sensitive to inhibition by sulfhydryl-specific reagents. Inhibition studies of this enzymatic activity by N-ethylmaleimide analogs revealed that analogs with bulky hydrophobic N-substitution were more potent inhibitors than hydrophilic or less bulky agents. Interestingly, the substrate analog N-arachidonylmaleimide was found to be the most potent inhibitor. A comparison model of MGL was constructed to get a view on the cysteine residues located near the binding site. These findings support our previous conclusion that the 2-AG-degrading enzymatic activity in rat cerebellar membranes corresponds to MGL or MGL-like enzyme and should facilitate further efforts to develop potent and more selective MGL inhibitors.