FOCUS--Development of a Global Communication and Modeling Platform for Applied and Computational Medicinal Chemists.

Communication of data and ideas within a medicinal chemistry project on a global as well as local level is a crucial aspect in the drug design cycle. Over a time frame of eight years, we built and optimized FOCUS, a platform to produce, visualize, and share information on various aspects of a drug discovery project such as cheminformatics, data analysis, structural information, and design. FOCUS is tightly integrated with internal services that involve-among others-data retrieval systems and in-silico models and provides easy access to automated modeling procedures such as pharmacophore searches, R-group analysis, and similarity searches. In addition, an interactive 3D editor was developed to assist users in the generation and docking of close analogues of a known lead. In this paper, we will specifically concentrate on issues we faced during development, deployment, and maintenance of the software and how we continually adapted the software in order to improve usability. We will provide usage examples to highlight the functionality as well as limitations of FOCUS at the various stages of the development process. We aim to make the discussion as independent of the software platform as possible, so that our experiences can be of more general value to the drug discovery community.

Pocket-space maps to identify novel binding-site conformations in proteins.

The identification of novel binding-site conformations can greatly assist the progress of structure-based ligand design projects. Diverse pocket shapes drive medicinal chemistry to explore a broader chemical space and thus present additional opportunities to overcome key drug discovery issues such as potency, selectivity, toxicity, and pharmacokinetics. We report a new automated approach to diverse pocket selection, PocketAnalyzer(PCA), which applies principal component analysis and clustering to the output of a grid-based pocket detection algorithm. Since the approach works directly with pocket shape descriptors, it is free from some of the problems hampering methods that are based on proxy shape descriptors, e.g. a set of atomic positional coordinates. The approach is technically straightforward and allows simultaneous analysis of mutants, isoforms, and protein structures derived from multiple sources with different residue numbering schemes. The PocketAnalyzer(PCA) approach is illustrated by the compilation of diverse sets of pocket shapes for aldose reductase and viral neuraminidase. In both cases this allows identification of novel computationally derived binding-site conformations that are yet to be observed crystallographically. Indeed, known inhibitors capable of exploiting these novel binding-site conformations are subsequently identified, thereby demonstrating the utility of PocketAnalyzer(PCA) for rationalizing and improving the understanding of the molecular basis of protein-ligand interaction and bioactivity. A Python program implementing the PocketAnalyzer(PCA) approach is available for download under an open-source license ( http://sourceforge.net/projects/papca/ or http://cpclab.uni-duesseldorf.de/downloads ).

Ensemble docking into multiple crystallographically derived protein structures: an evaluation based on the statistical analysis of enrichments.

Docking into multiple receptor conformations ("ensemble docking") has been proposed, and employed, in the hope that it may account for receptor flexibility in virtual screening and thus provide higher enrichments than docking into single rigid receptor structures. The statistical analyses presented in this paper provide quantitative evidence that in some cases docking into a crystallographically derived conformational ensemble does indeed yield better enrichment than docking into any of the individual members of the ensemble. However, these "successful" ensembles account for only a minority of those examined and it would not have been possible to prospectively predict their identity using only protein structural information. A more frequently observed outcome is that the ensemble enrichment is higher than the mean of the enrichments provided by its individual members. An additional and promising finding is that, if a set of known active compounds is available, an approach based on induced-fit docking appears to be a reliable way to construct ensembles which provide relatively high enrichments.

Inhibition of collagen-induced discoidin domain receptor 1 and 2 activation by imatinib, nilotinib and dasatinib.

Imatinib, nilotinib and dasatinib are protein kinase inhibitors which target the tyrosine kinase activity of the Breakpoint Cluster Region-Abelson kinase (BCR-ABL) and are used to treat chronic myelogenous leukemia. Recently, using a chemical proteomics approach another tyrosine kinase, the collagen receptor Discoidin Domain Receptor1 (DDR1) has also been identified as a potential target of these compounds. To further investigate the interaction of imatinib, nilotinib and dasatinib with DDR1 kinase we cloned and expressed human DDR1 and developed biochemical and cellular functional assays to assess their activity against DDR1 and the related receptor tyrosine kinase Discoidin Domain Receptor2 (DDR2). Our studies demonstrate that all 3 compounds are potent inhibitors of the kinase activity of both DDR1 and DDR2. In order to investigate the question of selectivity among DDR1, DDR2 and other tyrosine kinases we have aligned DDR1 and DDR2 protein sequences to other closely related members of the receptor tyrosine kinase family such as Muscle Specific Kinase (MUSK), insulin receptor (INSR), Abelson kinase (c-ABL), and the stem cell factor receptor (c-KIT) and have built homology models for the DDR1 and DDR2 kinase domains. In spite of high similarity among these kinases we show that there are differences within the ATP-phosphate binding loop (P-loop), which could be exploited to obtain kinase selective compounds. Furthermore, the potent DDR1 and DDR2 inhibitory activity of imatinib, nilotinib and dasatinib may have therapeutic implications in a number of inflammatory, fibrotic and neoplastic diseases.

Solution NMR structure of a designed metalloprotein and complementary molecular dynamics refinement.

We report the solution NMR structure of a designed dimetal-binding protein, di-Zn(II) DFsc, along with a secondary refinement step employing molecular dynamics techniques. Calculation of the initial NMR structural ensemble by standard methods led to distortions in the metal-ligand geometries at the active site. Unrestrained molecular dynamics using a nonbonded force field for the metal shell, followed by quantum mechanical/molecular mechanical dynamics of DFsc, were used to relax local frustrations at the dimetal site that were apparent in the initial NMR structure and provide a more realistic description of the structure. The MD model is consistent with NMR restraints, and in good agreement with the structural and functional properties expected for DF proteins. This work demonstrates that NMR structures of metalloproteins can be further refined using classical and first-principles molecular dynamics methods in the presence of explicit solvent to provide otherwise unavailable insight into the geometry of the metal center.

Parameterization of azole-bridged dinuclear platinum anticancer drugs via a QM/MM force matching procedure.

Azole-bridged diplatinum compounds are promising new anticancer drugs designed to induce small distortions upon DNA alkylation, able to circumvent resistance problems of existing platinum drugs. Hybrid quantum classical (QM/MM) molecular dynamics (MD) simulations of different azole-bridged platinum drugs have recently revealed the nature of the local deformations at the DNA binding site. However, the description of global slow converging rearrangements cannot be addressed by QM/MM MD due to the short time scale accessible. Extensive classical MD simulations are therefore mandatory to describe accurately the structural distortions in the DNA double helix. This issue is now addressed by developing a new set of accurate force field parameters of the platinated moiety via a recently proposed force matching procedure of the classical forces to ab initio forces obtained from QM/MM trajectories. The accuracy of our force field parameters is validated by comparison of structural properties from classical MD and hybrid QM/MM simulations. The structural characteristics of the Pt-lesion are well reproduced during classical MD compared with QM/MM simulations and available experimental data. The global distortions in the DNA duplex upon binding of dinuclear Pt-compounds are very small and rather opposite to those induced by cisplatin. Thus, the force match approach significantly extends the potentialities of molecular simulations in the study of anticancer drugs and of the interactions with their biological targets.

Azole-bridged diplatinum anticancer compounds. Modulating DNA flexibility to escape repair mechanism and avoid cross resistance.

Dinuclear azole-bridged Pt compounds bind to DNA helices, forming intrastrand crosslinks between adjacent guanines in a similar way to cisplatin. Their cytotoxic profile is, however, different from that of first and second generation Pt drugs in that they lack cross resistance in cisplatin-resistant cell lines. In contrast to cisplatin, which induces a large kink in DNA duplex, structural NMR studies and molecular dynamics simulations have shown that azole-bridged diplatinum compounds induce only small structural changes in double-stranded DNA. These structural differences have been invoked to explain the different cytotoxic profile of these compounds. Here, we show that in addition to the small structural changes in DNA, dinuclear Pt compounds also affect DNA minor groove flexibility in a different way than cisplatin. Free-energy calculations on azole-bridged diplatinum DNA adducts reveal that opening of the minor groove requires a higher free-energy cost (DeltaG ~ 7-15 kcal/mol) than in the corresponding cisplatin-DNA adduct (DeltaG ~ 0 kcal/mol). This could prevent minor groove binding proteins from binding to diplatinum-DNA adducts thus leading to a different cellular response than cisplatin and possibly decreasing the activity of excision repair enzymes. Although the development of drug resistance is a highly complex mechanism, our findings provide an additional rationale for the improved cytotoxic activity of these compounds in cell lines resistant to cisplatin.

Modeling the charge distribution at metal sites in proteins for molecular dynamics simulations.

Almost half of the proteome of living organisms is constituted of metalloproteins. Unfortunately, the ability of the current generation of molecular dynamics pairwise-additive forcefields to properly describe metal pockets is severely lacking due to the intrinsic difficulty of handling polarization and charge transfer contributions. In order to improve the description of metalloproteins, a simple reparameterization strategy is proposed herein that does not involve artificial constraints. Specifically, a non-bonded quantum mechanical-based model is used to capture the mean polarization and charge transfer contributions to the interatomic forces within the metal site. The present approach is demonstrated to provide enough accuracy to maintain the integrity of the metal pocket for a variety of metalloproteins during extended (multi-nanosecond) molecular dynamics simulations. The method enables the sampling of small conformational changes and the relaxation of local frustrations in NMR structures.

Structural and dynamical properties of manganese catalase and the synthetic protein DF1 and their implication for reactivity from classical molecular dynamics calculations.

There is a pressing need for accurate force fields to assist in metalloprotein analysis and design. Recent work on the design of mimics of dimetal proteins highlights the requirements for activity. DF1 is a de novo designed protein, which mimics the overall fold and active site geometry of a series of diiron and dimanganese proteins. Specifically, the dimanganese form of DF1 is a mimic of the natural enzyme manganese catalase, which catalyzes the dismutation reaction of hydrogen peroxide into water and oxygen. During catalytic turnover, the active site has to accommodate both the reduced and the oxidized state of the dimanganese core. The biomimetic protein DF1 is only stable in the reduced form and thus not active. Furthermore, the synthetic protein features an additional bridging glutamate sidechain, which occupies the substrate binding site. The goal of this study is to develop classical force fields appropriate for design of such important dimanganese proteins. To this aim, we use a nonbonded model to represent the metal-ligand interactions, which implicitly takes into account charge transfer and local polarization effects between the metal and its ligands. To calibrate this approach, we compare and contrast geometric and dynamical properties of manganese catalase and DF1. Having demonstrated a good correspondence with experimental structural data, we examine the effect of mutating the bridging glutamate to aspartate (M1) and serine (M2). Classical MD based on the refined forcefield shows that these point mutations affect not only the immediate coordination sphere of the manganese ions, but also the relative position of the helices, improving the similarity to Mn-catalase, especially in case of M2. On the basis of these findings, classical molecular dynamics calculations with the active site parameterization scheme introduced herein seem to be a promising addition to the protein design toolbox.

Modeling anticancer drug-DNA interactions via mixed QM/MM molecular dynamics simulations.

The development of anticancer drugs started over four decades ago, with the serendipitous discovery of the antitumor activity of cisplatin and its successful use in the treatment of various cancer types. Despite the efforts made in unraveling the mechanism of the action of cisplatin, as well as in the rational design of new anticancer compounds, in many cases detailed structural and mechanistic information is still lacking. Many of these drugs exert their anticancer activity by covalently binding to DNA inducing a distortion or simply impeding replication, thus triggering a cellular response, which eventually leads to cell death. A detailed understanding of the structural and electronic properties of drug-DNA complexes and their mechanism of binding is the key step in elucidating the principles of their anticancer activity. At the theoretical level, the description of covalent drug-DNA complexes requires the use of state-of-the-art computer simulation techniques such as hybrid quantum/classical molecular dynamics simulations. In this review we provide a general overview on: drugs which covalently bind to DNA duplexes, the basic concepts of quantum mechanics/molecular mechanics (QM/MM), molecular dynamics methods and a list of selected applications of these simulations to the study of drug-DNA adducts. Finally, the potential and the limitations of this approach to the study of such systems are critically evaluated.

Ab initio calculations of intramolecular parameters for a class of arylamide polymers.

Using DFT methods, we have determined intramolecular parameters for an important class of arylamide polymers displaying antimicrobial and anticoagulant inhibitory properties. A strong link has been established between these functions and the conformation that the polymers adopt in solution and at lipid bilayer interfaces. Thus, it is imperative for molecular dynamics simulations designed to probe the conformational behavior of these systems to accurately describe the torsional degrees of freedom. Standard force fields were shown to be deficient in this respect. Therefore, we have computed the relevant torsional energy profiles using a series of constrained geometry optimizations. We have also determined electrostatic parameters using our results in combination with standard RESP charge optimization. Force constants for bond and angle potentials were calculated by iteratively matching quantum and classical normal modes via a Monte Carlo scheme. The resulting new set of parameters accurately described the conformation and dynamical behavior of the arylamide polymers.

Binding of novel azole-bridged dinuclear platinum(II) anticancer drugs to DNA: insights from hybrid QM/MM molecular dynamics simulations.

Dinuclear Pt-containing compounds might be used to overcome the intrinsic and acquired cell resistance of widely used anticancer drugs such as cisplatin. Recently, the complexes [[cis-Pt(NH3)2]2(mu-OH)(mu-pz)](NO3)2 (with pz = pyrazolate) (1), [[cis-Pt(NH3)2]2(mu-OH)(mu-1,2,3-ta-N(1),N(2))](NO3)2 (with ta = 1,2,3-triazolate) (2), and the binding of 1 to d(CpTpCpTpG*pG*pTpCpTpCp) have been characterized. Here we provide the structural and electronic properties of the free drugs, of the intermediates of binding to guanine bases, and of the products, by performing DFT calculations. Our results show that in 2 an isomerization of the Pt-coordination sphere from N(2) to N(3) of the triazolate unit determines a thermodynamic stabilization of approximately 20 kcal/mol as a consequence of the formation of an allylic structure. In addition, hybrid quantum-classical molecular dynamics simulations of 1 and 2 DNA adducts have shed light on the structural distortions that the drugs induce to the DNA duplex. Our calculations show that the rise and the tilt of the two adjacent guanines are identical in the presence of 1 and 2, but they markedly increase when 2 binds in the N(1),N(3) fashion. In addition, the drugs do not provoke any kink upon binding to the double-stranded DNA, suggesting that they may act with a mechanism different than that of cisplatin. The accuracy of our calculations is established by a comparison with the NMR data for the corresponding complex with 1.

Duocarmycins binding to DNA investigated by molecular simulation.

Duocarmycins are a potent class of antitumor agents, whose activity arises through their covalent binding to adenine nucleobases of DNA.(1-3) Here, we perform molecular dynamics (MD) and hybrid Car-Parinello QM/MM simulations to investigate aspects of duocarmycin binding to the d(pGpApCpTpApApTpTpGpApC) oligonucleotide. We focus on the derivatives (+)-duocarmycin SA (DSA) and (+)-duocarmycin SI (DSI), for which structural information of the covalent complex with the oligonucleotide is available, as well as on the related, but less reactive, NBOC-duocarmycin SA (NBOC-DSA), interacting with the same oligonucleotide. Comparison is made with adenine alkylation reaction in water performed by the smallest of these compounds (NBOC-DSA). The MD calculations suggest that, in noncovalent complexes, (i) drug binding causes a partial dehydration of the minor groove, without inducing a significant conformational changes, and (ii) DSA and DSI occupy a more favorable position for nucleophilic attack than NBOC-DSA, consistently with the lower reactivity of the latter. The QM/MM calculations, which are used to investigate the first step of the alkylation reaction, turn out to provide strongly underestimated free energy barriers. Within these approximations, our calculations suggest that an important ingredient for the experimentally observed DNA catalytic power is the polarization of the drug by the biomolecular scaffold.