Roger S. Armen

A consensus view of fold space: Combining SCOP, CATH, and the Dali Domain Dictionary

We have determined consensus protein‐fold classifications on the basis of three classification methods, SCOP, CATH, and Dali. These classifications make use of different methods of defining and categorizing protein folds that lead to different views of protein‐fold space. Pairwise comparisons of domains on the basis of their fold classifications show that much of the disagreement between the classification systems is due to differing domain definitions rather than assigning the same domain to different folds. However, there are significant differences in the fold assignments between the three systems. These remaining differences can be explained primarily in terms of the breadth of the fold classifications. Many structures may be defined as having one fold in one system, whereas far fewer are defined as having the analogous fold in another system. By comparing these folds for a nonredundant set of proteins, the consensus method breaks up broad fold classifications and combines restrictive fold classifications into metafolds, creating, in effect, an averaged view of fold space. This averaged view requires that the structural similarities between proteins having the same metafold be recognized by multiple classification systems. Thus, the consensus map is useful for researchers looking for fold similarities that are relatively independent of the method used to compare proteins. The 30 most populated metafolds, representing the folds of about half of a nonredundant subset of the PDB, are presented here. The full list of metafolds is presented on the Web.

The role of α‐, 310‐, and π‐helix in helix→coil transitions

The conformational equilibrium between 310‐ and α‐helical structure has been studied via high‐resolution NMR spectroscopy by Millhauser and coworkers using the MW peptide Ac‐AMAAKAWAAKA AAARA‐NH2. Their 750‐MHz nuclear Overhauser effect spectroscopy (NOESY) spectra were interpreted to reflect appreciable populations of 310‐helix throughout the peptide, with the greatest contribution at the N and C termini. The presence of simultaneous αN(i,i + 2) and αN(i,i + 4) NOE cross‐peaks was proposed to represent conformational averaging between 310‐ and α‐helical structures. In this study, we describe 25‐nsec molecular dynamics simulations of the MW peptide at 298 K, using both an 8 Å and a 10 Å force‐shifted nonbonded cutoff. The ensemble averages of both simulations are in reasonable agreement with the experimental helical content from circular dichroism (CD), the 3JHNα coupling constants, and the 57 observed NOEs. Analysis of the structures from both simulations revealed very little formation of contiguous i → i + 3 hydrogen bonds (310‐helix); however, there was a large population of bifurcated i → i + 3 and i → i + 4 α‐helical hydrogen bonds. In addition, both simulations contained considerable populations of π‐helix (i → i + 5 hydrogen bonds). Individual turns formed over residues 1–9, which we predict contribute to the intensities of the experimentally observed αN(i,i + 2) NOEs. Here we show how sampling of both folded and unfolded structures can provide a structural framework for deconvolution of the conformational contributions to experimental ensemble averages.

Structural Implications for Selective Targeting of PARPs.

Poly(ADP-ribose) polymerases (PARPs) are a family of enzymes that use NAD+ as a substrate to synthesize polymers of ADP-ribose (PAR) as post-translational modifications of proteins. PARPs have important cellular roles that include preserving genomic integrity, telomere maintenance, transcriptional regulation, and cell fate determination. The diverse biological roles of PARPs have made them attractive therapeutic targets, which have fueled the pursuit of small molecule PARP inhibitors. The design of PARP inhibitors has matured over the past several years resulting in several lead candidates in clinical trials. PARP inhibitors are mainly used in clinical trials to treat cancer, particularly as sensitizing agents in combination with traditional chemotherapy to reduce side effects. An exciting aspect of PARP inhibitors is that they are also used to selectivity kill tumors with deficiencies in DNA repair proteins (e.g., BRCA1/2) through an approach termed “synthetic lethality.” In the midst of the tremendous efforts that have brought PARP inhibitors to the forefront of modern chemotherapy, most clinically used PARP inhibitors bind to conserved regions that permits cross-selectivity with other PARPs containing homologous catalytic domains. Thus, the differences between therapeutic effects and adverse effects stemming from pan-PARP inhibition compared to selective inhibition are not well understood. In this review, we discuss current literature that has found ways to gain selectivity for one PARP over another. We furthermore provide insights into targeting other domains that make up PARPs, and how new classes of drugs that target these domains could provide a high degree of selectivity by affecting specific cellular functions. A clear understanding of the inhibition profiles of PARP inhibitors will not only enhance our understanding of the biology of individual PARPs, but may provide improved therapeutic options for patients.

Resistance of Akt kinases to dephosphorylation through ATP-dependent conformational plasticity

Phosphorylation of a threonine residue (T308 in Akt1) in the activation loop of Akt kinases is a prerequisite for deregulated Akt activity frequently observed in neoplasia. Akt phosphorylation in vivo is balanced by the opposite activities of kinases and phosphatases. Here we describe that targeting Akt kinase to the cell membrane markedly reduced sensitivity of phosphorylated Akt to dephosphorylation by protein phosphatase 2A. This effect was amplified by occupancy of the ATP binding pocket by either ATP or ATP-competitive inhibitors. Mutational analysis revealed that R273 in Akt1 and the corresponding R274 in Akt2 are essential for shielding T308 in the activation loop against dephosphorylation. Thus, occupancy of the nucleotide binding pocket of Akt kinases enables intramolecular interactions that restrict phosphatase access and sustain Akt phosphorylation. This mechanism provides an explanation for the “paradoxical” Akt hyperphosphorylation induced by ATP-competitive inhibitor, A-443654. The lack of phosphatase resistance further contributes insight into the mechanism by which the human Akt2 R274H missense mutation may cause autosomal-dominant diabetes mellitus.

Identification of the Functional Binding Pocket for Compounds Targeting Small-Conductance Ca2+-Activated Potassium Channels

Small- and intermediate-conductance Ca2+-activated potassium channels, activated by Ca2+-bound calmodulin, play an important role in regulating membrane excitability. These channels are also linked to clinical abnormalities. A tremendous amount of effort has been devoted to developing small molecule compounds targeting these channels. However, these compounds often suffer from low potency and lack of selectivity, hindering their potentials for clinical use. A key contributing factor is the lack of knowledge of the binding site(s) for these compounds. Here we demonstrate by X-ray crystallography that the binding pocket for the compounds of the 1-EBIO class is located at the calmodulin-channel interface. We show that, based on structure data and molecular docking, mutations of the channel can effectively change the potency of these compounds. Our results provide insight into the molecular nature of the binding pocket and its contribution to the potency and selectivity of the compounds of the 1-EBIO class.

A scalable and accurate method for classifying protein-ligand binding geometries using a MapReduce approach

We present a scalable and accurate method for classifying protein-ligand binding geometries in molecular docking. Our method is a three-step process: the first step encodes the geometry of a three-dimensional (3D) ligand conformation into a single 3D point in the space; the second step builds an octree by assigning an octant identifier to every single point in the space under consideration; and the third step performs an octree-based clustering on the reduced conformation space and identifies the most dense octant. We adapt our method for MapReduce and implement it in Hadoop. The load-balancing, fault-tolerance, and scalability in MapReduce allow screening of very large conformation spaces not approachable with traditional clustering methods. We analyze results for docking trials for 23 protein-ligand complexes for HIV protease, 21 protein-ligand complexes for Trypsin, and 12 protein-ligand complexes for P38alpha kinase. We also analyze cross docking trials for 24 ligands, each docking into 24 protein conformations of the HIV protease, and receptor ensemble docking trials for 24 ligands, each docking in a pool of HIV protease receptors. Our method demonstrates significant improvement over energy-only scoring for the accurate identification of native ligand geometries in all these docking assessments. The advantages of our clustering approach make it attractive for complex applications in real-world drug design efforts. We demonstrate that our method is particularly useful for clustering docking results using a minimal ensemble of representative protein conformational states (receptor ensemble docking), which is now a common strategy to address protein flexibility in molecular docking.

Computational multiscale modeling in protein--ligand docking

The term multiscale modeling usually refers to solving physical problems along multiple spatial or temporal scales. This definition can also be extended to include other nonorthogonal descriptive scales, to allow a hierarchical approach to accurate and efficient problem solving. Therefore, the scales proposed in this article are not the traditional scales currently used in life sciences. We deal with computational rather than experimental multi- scales, and we extend this language to describe the problem of protein-ligand docking.

Different disease-causing mutations in transthyretin trigger the same conformational conversion.

Transthyretin (TTR)-containing amyloid fibrils are deposited in cardiac tissue as a natural consequence of aging. A large number of inherited mutations lead to amyloid diseases by accelerating TTR deposition in other organs. Amyloid formation is preceded by a disruption of the quaternary structure of TTR and conformational changes in the monomer. To study conformational changes preceding the formation of amyloid, we performed molecular dynamics simulations of the wild-type monomer, amyloidogenic variants (V30M, L55P, V122I) and a protective variant (T119M) at neutral and low pH. At low pH, the D strand dissociated from the beta-sheet to expose the A strand, consistent with experimental studies. In amyloidogenic variants and in the wild-type at low pH, there was a conformational change in the beta-sheets into alpha-sheet via peptide bond flips that was not observed at neutral pH in the wild-type monomer. The same residues participated in conversion in each amyloidogenic variant simulation, originating in the G strand between residues 106 and 109, with accelerated conversion at low pH. The T119M protective variant changed the local conformation of the H strand and suppressed the conversion observed in amyloidogenic variants.

Structural basis for selective inhibition of Cyclooxygenase-1 (COX-1) by diarylisoxazoles mofezolac and 3-(5-chlorofuran-2-yl)-5-methyl-4-phenylisoxazole (P6).

The diarylisoxazole molecular scaffold is found in several NSAIDs, especially those with high selectivity for COX-1. Here, we have determined the structural basis for COX-1 binding to two diarylisoxazoles: mofezolac, which is polar and ionizable, and 3-(5-chlorofuran-2-yl)-5-methyl-4-phenylisoxazole (P6) that has very low polarity. X-ray analysis of the crystal structures of COX-1 bound to mofezolac and 3-(5-chlorofuran-2-yl)-5-methyl-4-phenylisoxazole allowed the identification of specific binding determinants within the enzyme active site, relevant to generate structure/activity relationships for diarylisoxazole NSAIDs.

Systematic and efficient side chain optimization for molecular docking using a cheapest‐path procedure

Molecular docking of small‐molecules is an important procedure for computer‐aided drug design. Modeling receptor side chain flexibility is often important or even crucial, as it allows the receptor to adopt new conformations as induced by ligand binding. However, the accurate and efficient incorporation of receptor side chain flexibility has proven to be a challenge due to the huge computational complexity required to adequately address this problem. Here we describe a new docking approach with a very fast, graph‐based optimization algorithm for assignment of the near‐optimal set of residue rotamers. We extensively validate our approach using the 40 DUD target benchmarks commonly used to assess virtual screening performance and demonstrate a large improvement using the developed side chain optimization over rigid receptor docking (average ROC AUC of 0.693 vs. 0.623). Compared to numerous benchmarks, the overall performance is better than nearly all other commonly used procedures. Furthermore, we provide a detailed analysis of the level of receptor flexibility observed in docking results for different classes of residues and elucidate potential avenues for further improvement.