Incorporation of protein flexibility and conformational energy penalties in docking screens to improve ligand discovery.

Proteins fluctuate between alternative conformations, which presents a challenge for ligand discovery because such flexibility is difficult to treat computationally owing to problems with conformational sampling and energy weighting. Here we describe a flexible docking method that samples and weights protein conformations using experimentally derived conformations as a guide. The crystallographically refined occupancies of these conformations, which are observable in an apo receptor structure, define energy penalties for docking. In a large prospective library screen, we identified new ligands that target specific receptor conformations of a cavity in cytochrome c peroxidase, and we confirm both ligand pose and associated receptor conformation predictions by crystallography. The inclusion of receptor flexibility led to ligands with new chemotypes and physical properties. By exploiting experimental measures of loop and side-chain flexibility, this method can be extended to the discovery of new ligands for hundreds of targets in the Protein Data Bank for which similar experimental information is available.

SAMPL4 & DOCK3.7: lessons for automated docking procedures.

The SAMPL4 challenges were used to test current automated methods for solvation energy, virtual screening, pose and affinity prediction of the molecular docking pipeline DOCK 3.7. Additionally, first-order models of binding affinity were proposed as milestones for any method predicting binding affinity. Several important discoveries about the molecular docking software were made during the challenge: (1) Solvation energies of ligands were five-fold worse than any other method used in SAMPL4, including methods that were similarly fast, (2) HIV Integrase is a challenging target, but automated docking on the correct allosteric site performed well in terms of virtual screening and pose prediction (compared to other methods) but affinity prediction, as expected, was very poor, (3) Molecular docking grid sizes can be very important, serious errors were discovered with default settings that have been adjusted for all future work. Overall, lessons from SAMPL4 suggest many changes to molecular docking tools, not just DOCK 3.7, that could improve the state of the art. Future difficulties and projects will be discussed.

Ligand pose and orientational sampling in molecular docking.

Molecular docking remains an important tool for structure-based screening to find new ligands and chemical probes. As docking ambitions grow to include new scoring function terms, and to address ever more targets, the reliability and extendability of the orientation sampling, and the throughput of the method, become pressing. Here we explore sampling techniques that eliminate stochastic behavior in DOCK3.6, allowing us to optimize the method for regularly variable sampling of orientations. This also enabled a focused effort to optimize the code for efficiency, with a three-fold increase in the speed of the program. This, in turn, facilitated extensive testing of the method on the 102 targets, 22,805 ligands and 1,411,214 decoys of the Directory of Useful Decoys-Enhanced (DUD-E) benchmarking set, at multiple levels of sampling. Encouragingly, we observe that as sampling increases from 50 to 500 to 2000 to 5000 to 20,000 molecular orientations in the binding site (and so from about 1×10(10) to 4×10(10) to 1×10(11) to 2×10(11) to 5×10(11) mean atoms scored per target, since multiple conformations are sampled per orientation), the enrichment of ligands over decoys monotonically increases for most DUD-E targets. Meanwhile, including internal electrostatics in the evaluation ligand conformational energies, and restricting aromatic hydroxyls to low energy rotamers, further improved enrichment values. Several of the strategies used here to improve the efficiency of the code are broadly applicable in the field.

Chemical informatics uncovers a new role for moexipril as a novel inhibitor of cAMP phosphodiesterase-4 (PDE4).

PDE4 is one of eleven known cyclic nucleotide phosphodiesterase families and plays a pivotal role in mediating hydrolytic degradation of the important cyclic nucleotide second messenger, cyclic 3'5' adenosine monophosphate (cAMP). PDE4 inhibitors are known to have anti-inflammatory properties, but their use in the clinic has been hampered by mechanism-associated side effects that limit maximally tolerated doses. In an attempt to initiate the development of better-tolerated PDE4 inhibitors we have surveyed existing approved drugs for PDE4-inhibitory activity. With this objective, we utilised a high-throughput computational approach that identified moexipril, a well tolerated and safe angiotensin-converting enzyme (ACE) inhibitor, as a PDE4 inhibitor. Experimentally we showed that moexipril and two structurally related analogues acted in the micro molar range to inhibit PDE4 activity. Employing a FRET-based biosensor constructed from the nucleotide binding domain of the type 1 exchange protein activated by cAMP, EPAC1, we demonstrated that moexipril markedly potentiated the ability of forskolin to increase intracellular cAMP levels. Finally, we demonstrated that the PDE4 inhibitory effect of moexipril is functionally able to induce phosphorylation of the small heat shock protein, Hsp20, by cAMP dependent protein kinase A. Our data suggest that moexipril is a bona fide PDE4 inhibitor that may provide the starting point for development of novel PDE4 inhibitors with an improved therapeutic window.

ZINC: a free tool to discover chemistry for biology.

ZINC is a free public resource for ligand discovery. The database contains over twenty million commercially available molecules in biologically relevant representations that may be downloaded in popular ready-to-dock formats and subsets. The Web site also enables searches by structure, biological activity, physical property, vendor, catalog number, name, and CAS number. Small custom subsets may be created, edited, shared, docked, downloaded, and conveyed to a vendor for purchase. The database is maintained and curated for a high purchasing success rate and is freely available at zinc.docking.org.

Ligand discovery from a dopamine D3 receptor homology model and crystal structure.

G protein-coupled receptors (GPCRs) are intensely studied as drug targets and for their role in signaling. With the determination of the first crystal structures, interest in structure-based ligand discovery increased. Unfortunately, for most GPCRs no experimental structures are available. The determination of the D(3) receptor structure and the challenge to the community to predict it enabled a fully prospective comparison of ligand discovery from a modeled structure versus that of the subsequently released crystal structure. Over 3.3 million molecules were docked against a homology model, and 26 of the highest ranking were tested for binding. Six had affinities ranging from 0.2 to 3.1 µM. Subsequently, the crystal structure was released and the docking screen repeated. Of the 25 compounds selected, five had affinities ranging from 0.3 to 3.0 µM. One of the new ligands from the homology model screen was optimized for affinity to 81 nM. The feasibility of docking screens against modeled GPCRs more generally is considered.

Protein pockets: inventory, shape, and comparison.

The shape of the protein surface dictates what interactions are possible with other macromolecules, but defining discrete pockets or possible interaction sites remains difficult. First, there is the problem of defining the extent of the pocket. Second, one has to characterize the shape of each pocket. Third, one needs to make quantitative comparisons between pockets on different proteins. An elegant solution to these problems is to sort all surface and solvent points by travel depth and then collect a hierarchical tree of pockets. The connectivity of the tree is determined via the deepest saddle points between each pair of neighboring pockets. The resulting pocket surfaces tessellate the entire protein surface, producing a complete inventory of pockets. This method of identifying pockets also allows one to easily compute important shape metrics, including the problematic pocket volume, surface area, and mouth size. Pockets are also annotated with their lining residue lists and polarity and with other residue-based properties. Using this tree and the various shape metrics pockets can be merged, grouped, or filtered for further analysis. Since this method includes the entire surface, it guarantees that any pocket of interest will be found among the output pockets, unlike all previous methods of pocket identification. The resulting hierarchy of pockets is easy to visualize and aids users in higher level analysis. Comparison of pockets is done by using the shape metrics, avoiding the complex shape alignment problem. Example applications show that the method facilitates pocket comparison along mutational or time-dependent series. Pockets from families of proteins can be examined using multiple pocket tree alignments to see how ligand binding sites or how other pockets have changed with evolution. Our method is called CLIPPERS for complete liberal inventory of protein pockets elucidating and reporting on shape.

Shape and evolution of thermostable protein structure.

Organisms evolved at high temperatures must maintain their proteins' structures in the face of increased thermal disorder. This challenge results in differences in residue utilization and overall structure. Focusing on thermostable/mesostable pairs of homologous structures, we have examined these differences using novel geometric measures: specifically burial depth (distance from the molecular surface to each atom) and travel depth (distance from the convex hull to the molecular surface that avoids the protein interior). These along with common metrics like packing and Wadell Sphericity are used to gain insight into the constraints experienced by thermophiles. Mean travel depth of hyperthermostable proteins is significantly less than that of their mesostable counterparts, indicating smaller, less numerous and less deep pockets. The mean burial depth of hyperthermostable proteins is significantly higher than that of mesostable proteins indicating that they bury more atoms further from the surface. The burial depth can also be tracked on the individual residue level, adding a finer level of detail to the standard exposed surface area analysis. Hyperthermostable proteins for the first time are shown to be more spherical than their mesostable homologues, regardless of when and how they adapted to extreme temperature. Additionally, residue specific burial depth examinations reveal that charged residues stay unburied, most other residues are slightly more buried and Alanine is more significantly buried in hyperthermostable proteins.

Finding and characterizing tunnels in macromolecules with application to ion channels and pores.

We describe a new algorithm, CHUNNEL, to automatically find, characterize, and display tunnels or pores in proteins. The correctness and accuracy of the algorithm is verified on a constructed set of proteins and used to analyze large sets of real proteins. The verification set contains proteins with artificially created pores of known path and width profile. The previous benchmark algorithm, HOLE, is compared with the new algorithm. Results show that the major advantage of the new algorithm is that it can successfully find and characterize tunnels with no a priori guidance or clues about the location of the tunnel mouth, and it will successfully find multiple tunnels if present. CHUNNEL can also be used in conjunction with HOLE, with the former used toprime HOLE and the latter to track and characterize the pores. Analysis was conducted on families of membrane protein structures culled from the Protein Data Bank as well as on a set of transmembrane proteins with predicted membrane-aqueous phase interfaces, yielding the first completely automated examination of tunnels through membrane proteins, including tunnels that exit in the membrane bilayer.

Structure-based maximal affinity model predicts small-molecule druggability.

Lead generation is a major hurdle in small-molecule drug discovery, with an estimated 60% of projects failing from lack of lead matter or difficulty in optimizing leads for drug-like properties. It would be valuable to identify these less-druggable targets before incurring substantial expenditure and effort. Here we show that a model-based approach using basic biophysical principles yields good prediction of druggability based solely on the crystal structure of the target binding site. We quantitatively estimate the maximal affinity achievable by a drug-like molecule, and we show that these calculated values correlate with drug discovery outcomes. We experimentally test two predictions using high-throughput screening of a diverse compound collection. The collective results highlight the utility of our approach as well as strategies for tackling difficult targets.

Structure-based identification of small molecule binding sites using a free energy model.

We separately have shown that the maximal druglike affinity of a given binding site on a protein can be calculated on the basis of the binding-site structure alone by using a desolvation-based free energy model along with the notion that druglike ligands fall into certain physiochemical property ranges. Here, we present an approach where we reformulate the calculated druggability affinity as an additive free energy to facilitate the searching of whole protein surfaces for druglike binding sites. The highest-scoring patches in many cases represent known ligand-binding sites for druggable targets, but not for difficult targets. This approach differs from other approaches in that it does not simply identify pockets with the greatest volume but instead identifies pockets that are likely to be amenable to druglike small-molecule binding. Combining the method with a functional residue prediction method called SCA (statistical coupling analysis) results in the prediction of potentially druggable allosteric binding sites on p38alpha kinase.

Travel depth, a new shape descriptor for macromolecules: application to ligand binding.

Depth is a term frequently applied to the shape and surface of macromolecules, describing for example the grooves in DNA, the shape of an enzyme active site, or the binding site for a small molecule in a protein. Yet depth is a difficult property to define rigorously in a macromolecule, and few computational tools exist to quantify this notion, to visualize it, or analyze the results. We present our notion of travel depth, simply put the physical distance a solvent molecule would have to travel from a surface point to a suitably defined reference surface. To define the reference surface, we use the limiting form of the molecular surface with increasing probe size: the convex hull. We then present a fast, robust approximation algorithm to compute travel depth to every surface point. The travel depth is useful because it works for pockets of any size and complexity. It also works for two interesting special cases. First, it works on the grooves in DNA, which are unbounded in one direction. Second, it works on the case of tunnels, that is pockets that have no "bottom", but go through the entire macromolecule. Our algorithm makes it straightforward to quantify discussions of depth when analyzing structures. High-throughput analysis of macromolecule depth is also enabled by our algorithm. This is demonstrated by analyzing a database of protein-small molecule binding pockets, and the distribution of bound magnesium ions in RNA structures. These analyses show significant, but subtle effects of depth on ligand binding localization and strength.

Protonation of excited state pyrene-1-carboxylate by phosphate and organic acids in aqueous solution studied by fluorescence spectroscopy.

Pyrene-1-carboxylic acid has a pK of 4.0 in the ground state and 8.1 in the singlet electronic excited state. In the pH range of physiological interest (pH approximately 5-8), the ground state compound is largely ionized as pyrene-1-carboxylate, but protonation of the excited state molecule occurs when a proton donor reacts with the carboxylate during the excited state lifetime of the fluorophore. Both forms of the pyrene derivatives are fluorescent, and in this work the protonation reaction was measured by monitoring steady-state and time-resolved fluorescence. The rate of protonation of pyrene-COO(-) by acetic, chloroacetic, lactic, and cacodylic acids is a function of DeltapK, as predicted by Marcus theory. The rate of proton transfer from these acids saturates at high concentration, as expected for the existence of an encounter complex. Trihydrogen-phosphate is a much better proton donor than dihydrogen- and monohydrogen-phosphate, as can be seen by the pH dependence. The proton-donating ability of phosphate does not saturate at high concentrations, but increases with increasing phosphate concentration. We suggest that enhanced rate of proton transfer at high phosphate concentrations may be due to the dual proton donating and accepting nature of phosphate, in analogy to the Grotthuss mechanism for proton transfer in water. It is suggested that in molecular structures containing multiple phosphates, such as membrane surfaces and DNA, proton transfer rates will be enhanced by this mechanism.

An intuitive approach to measuring protein surface curvature.

A natural way to measure protein surface curvature is to generate the least squares fitted (LSF) sphere to a surface patch and use the radius as the curvature measure. While the concept is simple, the sphere-fitting problem is not trivial and known means of protein surface curvature measurement use alternative schemes that are arguably less straightforward to interpret. We have developed an approach to solve the LSF sphere problem by turning the sphere-fitting problem into a solvable plane-fitting problem using a transformation known as geometric inversion. The approach works on any arbitrary surface patch, and returns a radius of curvature that has direct physical interpretation. Additionally, it is flexible in its ability to find the curvature of an arbitrary surface patch, and the "resolution" can be adjusted to highlight atomic features or larger features such as peptide binding sites. We include examples of applying the method to visualization of peptide recognition pockets and protein conformational change, as well as a comparison with a commonly used solid-angle curvature method showing that the LSF method produces more pronounced curvature results.