Protein Binding Pocket Dynamics - Accounts of Chemical Research

Lilian Hernández Alvarez , Diego Enry Barreto Gomes , Jorge Enrique .... Elena Segala , Marta Amaral , Dong Guo , Dorothee Andres , Victoria Georgi , ...
1 downloads 0 Views 3MB Size
This is an open access article published under an ACS AuthorChoice License, which permits copying and redistribution of the article or any adaptations for non-commercial purposes.

Article pubs.acs.org/accounts

Protein Binding Pocket Dynamics Antonia Stank,†,‡ Daria B. Kokh,† Jonathan C. Fuller,†,∇ and Rebecca C. Wade*,†,⊥,§ †

Heidelberg Institute for Theoretical Studies (HITS), Schloss-Wolfsbrunnenweg 35, 69118 Heidelberg, Germany Heidelberg Graduate School of Mathematical and Computational Methods for the Sciences and §Interdisciplinary Center for Scientific Computing (IWR), Heidelberg University, Im Neuenheimer Feld 368, 69120 Heidelberg, Germany ⊥ Center for Molecular Biology of the University of Heidelberg (ZMBH), DKFZ-ZMBH Alliance, Im Neuenheimer Feld 282, 69120 Heidelberg, Germany ‡

CONSPECTUS: The dynamics of protein binding pockets are crucial for their interaction specificity. Structural flexibility allows proteins to adapt to their individual molecular binding partners and facilitates the binding process. This implies the necessity to consider protein internal motion in determining and predicting binding properties and in designing new binders. Although accounting for protein dynamics presents a challenge for computational approaches, it expands the structural and physicochemical space for compound design and thus offers the prospect of improved binding specificity and selectivity. A cavity on the surface or in the interior of a protein that possesses suitable properties for binding a ligand is usually referred to as a binding pocket. The set of amino acid residues around a binding pocket determines its physicochemical characteristics and, together with its shape and location in a protein, defines its functionality. Residues outside the binding site can also have a long-range effect on the properties of the binding pocket. Cavities with similar functionalities are often conserved across protein families. For example, enzyme active sites are usually concave surfaces that present amino acid residues in a suitable configuration for binding low molecular weight compounds. Macromolecular binding pockets, on the other hand, are located on the protein surface and are often shallower. The mobility of proteins allows the opening, closing, and adaptation of binding pockets to regulate binding processes and specific protein functionalities. For example, channels and tunnels can exist permanently or transiently to transport compounds to and from a binding site. The influence of protein flexibility on binding pockets can vary from small changes to an already existent pocket to the formation of a completely new pocket. Here, we review recent developments in computational methods to detect and define binding pockets and to study pocket dynamics. We introduce five different classes of protein pocket dynamics: (1) appearance/disappearance of a subpocket in an existing pocket; (2) appearance/disappearance of an adjacent pocket on the protein surface in the direct vicinity of an already existing pocket; (3) pocket breathing, which may be caused by side-chain fluctuations or backbone or interdomain vibrational motion; (4) opening/closing of a channel or tunnel, connecting a pocket inside the protein with solvent, including lid motion; and (5) the appearance/disappearance of an allosteric pocket at a site on a protein distinct from an already existing pocket with binding of a ligand to the allosteric binding site affecting the original pocket. We suggest that the class of pocket dynamics, as well as the type and extent of protein motion affecting the binding pocket, should be factors considered in choosing the most appropriate computational approach to study a given binding pocket. Furthermore, we examine the relationship between pocket dynamics classes and induced fit, conformational selection, and gating models of ligand binding on binding kinetics and thermodynamics. We discuss the implications of protein binding pocket dynamics for drug design and conclude with potential future directions for computational analysis of protein binding pocket dynamics.



site adapts by “induced fit” to bind the respective ligand. A further model, “conformational selection”, in which the protein may adopt different conformations in its unbound state and a ligand binds selectively to one of these pre-existing conformations, was first employed to explain conformational changes of a protein binding site arising from the binding of an

INTRODUCTION As early as 1894, Fischer introduced one of the first models of protein−ligand binding using the analogy of a lock for a rigid protein binding pocket and a key for a rigid and specific ligand to explain the interaction between an enzyme and its substrate.1 The limitations of this model became clear when protein crystal structures were reported that showed a variety of pocket shapes for the same receptor cocrystallized with different ligands. Indeed, a model taking receptor flexibility into account was introduced in 1958 by Koshland2 in which the protein binding © 2016 American Chemical Society

Received: November 20, 2015 Published: April 25, 2016 809

DOI: 10.1021/acs.accounts.5b00516 Acc. Chem. Res. 2016, 49, 809−815

Article

Accounts of Chemical Research

HSP90. It is lined by an unstable α-helix3 (blue) that undergoes distortion, converting to two short helices connected by a loop (orange). The inset shows a purine-based inhibitor that occupies both the ADP/ATP binding site and a hydrophobic transient subpocket formed under α-helix3. Figure 2B shows an adjacent pocket in interleukin 2 (IL-2), which has a highly adaptive protein−protein binding site that can be blocked by a small molecule. Arkin et al.14 detected a flexible hydrophobic subpocket on the surface adjacent to the protein− protein binding site, which provides an additional space for binding a small molecule inhibitor.15 This adjacent binding site is formed due to side-chain rotation accompanied by backbone adaptation. Pocket breathing motion is illustrated in Figure 2C for the B-cell lymphoma-extra-large (BCL-XL) protein, where considerable variation of the binding pocket shape is caused by movement of the α-helices lining the binding site. A set of NMR structures of nonspecific lipid transfer protein (ns-LTP) in complex with prostaglandin B2 demonstrates high plasticity of the hydrophobic binding pocket, including the opening and closing of a channel that enables ligand binding, as shown in Figure 2D for two models from the NMR ensemble.16 Figure 2E shows the formation of an allosteric pocket in P38 mitogenactivated protein kinase (P38 MAPK) due to motion of the highly conserved Asp-Phe-Gly motif.17 Opening of the allosteric binding pocket requires flipping of the Phe sidechain toward the ATP/ADP binding site. This reduces the volume of this binding site, and thus, binding of an inhibitor at the allosteric site inhibits ADP/ATP binding.

allosteric ligand.3 This model has been supported by numerous experiments for both allosteric and nonallosteric ligands.4,5 Protein dynamics occur over spatiotemporal scales ranging from atomic fluctuations (∼ femtoseconds) to protein folding and subunit association (∼ seconds to hours).6 The size and flexibility of the individual components of a protein (e.g., amino acid side-chains or domains) influence the time scale on which their motions occur. The motions of different protein elements are often coupled, and both intrinsic protein flexibility as well as conformational adjustment to the interacting ligand may contribute to the binding process. The importance of the time scale of protein conformational transitions to distinguish between the two limiting cases, induced-fit and conformational selection mechanisms, has been highlighted in several studies.7−9 The binding of glucose to human glucokinase and of geldanamycin to heat shock protein 90 (HSP90) are two examples where the predominant roles of conformational selection10 and induced fit11 mechanisms, respectively, were demonstrated experimentally. Further examples are reviewed by Copeland.12 More generally, protein−ligand binding can be considered to involve both mechanisms, but in any particular case, their relative importance varies, as does their effect on binding thermodynamics and kinetics.13 Five Classes of Protein Binding Pocket Dynamics

We introduce a classification of protein binding pocket dynamics with five classes: subpocket, adjacent pocket, breathing motion, channel/tunnel, and allosteric pocket. These classes are illustrated schematically in Figure 1 relative

Detecting and Defining Protein Binding Pockets and Channels

To be able to bind one or more ligands, a protein pocket must possess or acquire a number of features that complement those of potential binders. Specifically, the volume should correspond to or exceed that of a ligand, the shape should enable the ligand to fit in, and the physicochemical properties should complement those of the ligand. Therefore, the main properties for characterizing a protein binding pocket are the overall geometry, the composition of amino acid residues, the type of solvation, the hydrophobicity, the electrostatics, and the chemical fragment interactions.18,19 These characteristics are also important for evaluation of the pocket’s “druggability”, i.e. its ability to bind a drug.20 The position of a ligand in the holo-structure of a protein determined experimentally can be used to define the binding pocket and channels. Alternatively, computational methods can be applied for this purpose. Henrich et al.18 reviewed the most popular computational approaches and tools to identify protein pockets and binding sites. These methods are based on either geometric or energetic analyses of the target protein structure, and some additionally use protein structure and/or sequence comparison. Often, the use of a combination of techniques improves predictions. This is exemplified by the Metapocket21 Web server, which makes predictions of binding sites from a consensus from eight different shape-based binding site detection tools. Shape-based methods may fail if only unbound protein structures are available, and in this case, protein dynamics should be considered. Furthermore, many protein channels and tunnels are only partially open at any given moment, and thus approaches involving analysis of the protein flexibility are required for their detection, for example, using crystallographic thermal factors22,23 or molecular dynamics

Figure 1. Cartoon representation of five different classes of pocket dynamics: subpocket, adjacent pocket, breathing motion, channel/ tunnel, allosteric pocket. Regions colored in pink indicate pocket variation relative to the reference structure (shown in the center); the red dotted lines show the pocket shapes. For allostery, the shape of the original binding site is affected by a molecule binding at a distinct binding site.

to a reference binding pocket. The five classes are distinct but nonexclusive, meaning that overlaps between these classes are possible. An allosteric pocket, for example, may also be in close vicinity to the reference pocket and, thus, considered as an adjacent pocket. An important feature of an adjacent pocket is that it is positioned such that one bivalent ligand could bind in both the adjacent and the reference pockets simultaneously. Breathing motion refers to the enlargement or contraction of the original pocket, roughly retaining the original pocket shape. Such breathing motion may precede subsequent motions to form a distinct subpocket. Examples of each of the five classes are shown in Figure 2. The first case, subpocket formation (Figure 2A), is illustrated by the binding site of the N-terminal ATP-binding domain of 810

DOI: 10.1021/acs.accounts.5b00516 Acc. Chem. Res. 2016, 49, 809−815

Article

Accounts of Chemical Research

Figure 2. Examples of protein binding sites that illustrate the five different classes of pocket dynamics represented in Figure 1. For each case, two structures with different pocket conformations are shown in cartoon representation with flexible elements responsible for pocket changes shown in orange and blue. The binding pocket variations are visualized in the insets in cross sections of the protein structures going through the pocket of interest. Protein interiors are shown in gray. Transient opening of protein cavities is highlighted in red. In A, an unstable part of α-helix3 is shown in blue. In B, a flexible loop (Gln74-Leu80 shown in light blue and light orange) is missing in the crystal structures and was modeled using PRIME software (Schrödinger LLC, version 4.1).59,60 In 2E, the flipping Phe and Asp residues are shown in orange (open active site pocket with ADP bound) and blue (occupied allosteric pocket and blocked ADP-binding pocket). The structures (protein name, PDB ID) are (A) HSP90, 1yer,61 1uyd;62 (B) IL-2, 1pw6,63 1m4a;14 (C) BCL-XL, 3zln,64 3qkd;65 (D) ns-LTP, 1cz2 (models 2 and 8);16 (E) P38 MAPK, 1kv1,17 1ny3;66 highlighted in orange and blue, respectively. See text for details.

(MD) simulations.24 These methods have recently been reviewed by Brezovsky et al.25

IL-2, and successfully identified transient pockets.26 Although MD simulation can be used as a sampling method for all of the pocket classes shown in Figure 1, it is computationally expensive, and the binding pocket dynamics may not be adequately sampled during the simulation time. Therefore, other methods that are more computationally efficient, although less accurate, have been used to study protein binding pocket dynamics arising from large-scale protein motions. In particular, normal mode analysis (NMA) provides a means to quickly explore the possible motion of a protein around a given input structure and is particularly useful for low-frequency interdomain harmonic breathing motions. It may be applied to atomic-detail molecular mechanics models or to coarse-grained elastic network models. For example, Ahmed et al. used a normal mode-based geometric simulation approach on an adenylate kinase structure to generate a pathway of conformational changes that describe domain movements leading to binding site closure.27 However, NMA may not be appropriate when higher frequency or anharmonic motions have an important influence on binding pocket dynamics. Another

Prediction of Protein Pocket Dynamics and Computational Detection of Transient Pockets

Considering multiple conformations of a protein can increase the accuracy of ligand docking calculations and enable the detection of novel binding pockets. In addition, it can give insights into the kinetics of ligand binding or transition channel opening. Experiments may not be able to access all the conformations that could affect compound selectivity. Computational methods to simulate pocket dynamics can fill these gaps. Here, different computational sampling methods and pocket analysis tools applicable for detecting the five classes of pocket dynamics depicted in Figure 1 are discussed. Algorithms for Sampling of Protein Pocket Conformations. MD simulation is often used to explore variations in protein pocket shape and physicochemical properties due to protein dynamics. For example, Eyrisch and Helms applied 10 ns MD simulations to several proteins, including BCL-XL and 811

DOI: 10.1021/acs.accounts.5b00516 Acc. Chem. Res. 2016, 49, 809−815

Article

Accounts of Chemical Research method for exploring protein mobility is tCONCOORD,28 which performs geometric constraint-based sampling of protein conformations. This method enables fast sampling of large-scale motions (such as loop or domain motions). Ashford et al.29 showed that the ensemble of BCL-XL structures generated by tCONCOORD revealed a transient binding subpocket, and Seeliger and de Groot30 were able to generate transitions from apo-to-holo conformations for several proteins displaying interdomain pocket breathing motions using a combination of tCONCOORD and MD refinement. The framework rigidity optimized dynamic algorithm, FRODA,31 is another sampling algorithm for examining the internal mobility of proteins in a short time while respecting the stereochemistry and defined constraints. Metz et al. applied FRODA to sample hydrophobic transient pockets in IL-2 with better results than a comparable MD simulation, which shows the applicability to the subpocket and adjacent pocket classes.32 A disadvantage of the latter methods is that the generated structures are not energetically validated, and additional MD equilibration is generally required before using them further, for example, in a ligand docking procedure. Protein Binding Pocket Analysis Algorithms. Simulations or experiments may provide a large ensemble of protein structures with a variety of binding pocket conformations. To distinguish a suitable pocket for ligand docking, the analysis of pocket dynamics in multiple structures is required. MDPocket33 is one of the first methods designed for analysis of an ensemble of structures or MD trajectories. It provides the user with a pocket frequency map for visualization of pocket opening and for tracing some of the characteristics (e.g., pocket volume and accessible surface area) of a selected cavity along an MD trajectory. In another approach, EPOSBP,26 protein cavities defined by pocket lining atoms are clustered to identify conserved or transient pocket regions. A deficiency of this approach is that the numbering of subpockets changes from snapshot to snapshot, hindering analysis of the pocket dynamics. Principal component analysis of pocket shapes mapped on a 3D grid is used in the PocketAnalyzer34 approach aimed at detection of transient pocket regions in MD trajectories. In TRAPP (TRAnsient binding Pockets in Proteins),35 conserved and transient regions are defined with respect to a starting reference structure. TRAPP can be applied to the detection of larger conformational changes, such as might be revealed by a tCONCOORD or NMA calculation. This is achieved by using only binding site residues for structure alignment together with a robust algorithm for pocket detection that does not require parameter adjustment when going from buried to shallow cavities. Unlike previously discussed methods, where pocket shapes are used for the analysis of pocket dynamics, the PPIAnalyzer method32 clusters protein conformations with respect to the RMSD of heavy atoms, followed by the identification of transient binding sites using the PocketAnalyzer program. All of the methods described above are in principle applicable to exploring the formation of adjacent pockets, subpockets, and pocket breathing motions. Detecting allosteric pockets and analyzing the structural and dynamic relationship between allosteric and orthosteric pockets poses further requirements. Methods that reveal energy transport or networks in proteins that might relate to allosteric effects have been reported, such as SPACER36 and MCPath.37 Furthermore, the FRODA method has been applied for sampling structures to reveal the relationship between allosteric and orthosteric pockets.31 For

channel detection, further tools are available. One example is Caver 3.0,24 which uses a Voronoi diagram to describe tunnels in an MD trajectory. Other tools are Try_cavity,38 Cavity analysis,39 and MOLE,40 which employ MD trajectories to generate conformations for analysis. Effect of Protein Binding Pocket Dynamics on the Thermodynamics and Kinetics of Ligand Binding

Target flexibility is one of the main factors that affects receptordrug binding kinetics (see reviews by Pan et al.,41 Romanowska et al.,42 and Klebe et al.43). Here, we focus on the influence of protein dynamics on the thermodynamics and kinetics of ligand binding in the context of the different binding pocket classes. Protein−ligand binding is often described by the one- or two-step models illustrated in Figure 3. These models are the

Figure 3. (A) Schematic illustration of free energy profiles for protein−ligand binding. Binding free energy and transition state energy are denoted by ΔG0 and ΔG#, respectively. kon and koff are association and dissociation rate constants; Kd is the equilibrium dissociation constant for receptor R and ligand L in the one-step binding model with or without conformational selection. RC describes the conformational ensemble of the receptor, and kc and k−c describe the corresponding transition rate constants. (B) Two-step binding model with induced fit. RF describes the receptor conformation in the free, unbound form and k1 − k4 indicate the respective rate constants for transitions between the states.

most commonly used, but other models with more barriers or with a downhill binding free energy landscape are sometimes applicable. The observed rate constants depend on the kinetics of all the steps involved in the binding/dissociation process.42 The free energy barriers to binding and unbinding may in part arise from the conformational rearrangement of the ligand and the protein. In particular, a barrier can be associated with stochastic protein motion causing closing and opening of the pocket itself (e.g., breathing motion) or the pocket entrance (in the case of a channel or tunnel). These cases can be considered with a conformation selection binding mechanism, and the association kinetics can be described using a gating model.7,44,45 This model describes how the binding rate is modulated by the relation between the time scale of accessibility (or gating) of 812

DOI: 10.1021/acs.accounts.5b00516 Acc. Chem. Res. 2016, 49, 809−815

Article

Accounts of Chemical Research

protein exterior show distinct features that might be exploited in inhibitor design. Bartolowits and Davisson recently discussed and encouraged the consideration of subpockets in fragment-based drug design to find ligands with higher specificity.55 Some examples in which experimental and computational methods were combined to account for subpocket dynamics are the design of an inhibitor that discriminates between Plasmodium falciparum and human HSP90s,56 the detection of a transient subpocket that made it possible to distinguish between human and parasitic phosphodiesterase,57 and the exploitation of a flexible, partially transient allosteric surface cavity for the design of an inhibitor of HIV-1 protease subtype B.58

the binding site and the time scale for ligand binding. The frequency of gate opening and the fraction of time the gate is open are influenced by the type of structural dynamics necessary to open the gate. The closed state is often ascribed to a lid motion that blocks ligand entrance. Movement of a lid can be energetically expensive, explaining the long residence times of many ligands bound through a gating mechanism (see reviews 12 and 46). For example, the rate of enzyme−inhibitor complex formation of Mycobacterium tuberculosis enoyl-ACPreductase (InhA) was found to correlate with motion of the substrate binding loop.47 In the two-step model (Figure 3B), a weakly bound transient complex (an encounter complex) is formed after passing the first barrier to binding, which increases the chance of a ligand to bind to a specific protein pocket that only opens stochastically or enables slow conformational changes required for the ligand binding (by induced fit). For example, binding of a bivalent compound to an adjacent pocket may proceed in two steps: first to the original pocket and then to the adjacent one due to induced fit. Channel and breathing pocket dynamics often fall into the category of gating or conformational selection processes, whereas the induced-fit mechanism often plays a leading role in the formation of small subpockets. On the other hand, ligand-induced stabilization of breathing motions can alter binding kinetics and thermodynamics, as was observed for the protein kinase inhibitor Gleevec, which showed different binding kinetics to Abl and Src driven by its different ability to stabilize the P-loop.48 The case of an allosteric pocket is the most complex, as both the allosteric and orthosteric binding sites can be considered to be gated, and the binding rates of the two pockets are not independent of each other. Ligand binding in the allosteric pocket can increase (activate) or decrease (inhibit) the kon of the ligand in the orthosteric binding site and vice versa for kof f.



CONCLUSIONS We have here introduced a classification of protein binding pocket dynamics into five classessubpocket, adjacent pocket, breathing motion, channel/tunnel, and allosteric pocketand highlighted the importance of considering protein pocket dynamics in the drug design process. A range of computational methods is available for simulating and analyzing protein pocket dynamics. However, the treatment of protein dynamics in structure-based drug design projects is generally very limited, and better computational tools are required for studying binding pocket dynamics for this purpose. An automated tool that combines simulations of protein mobility using a variety of approaches with the analysis of pocket dynamics would help users to choose the most appropriate strategy for designing new inhibitors. The assessment of the druggability of a protein pocket should include measures of how the physicochemical properties of a pocket are affected by the pocket dynamics derived from analysis of simulations and NMR or crystal structures. Treatment of protein pocket dynamics may increase the space of potential binders, e.g., through ligand binding in transient subpockets. However, it may also provide a basis for identifying specific compounds that can bind to the structurally conserved parts of a pocket despite its mobility. We suggest that consideration of the class of protein binding pocket dynamics can facilitate the choice of appropriate computational approaches to sample and analyze the pocket dynamics as well as ligand design strategies.

Implications for Drug Design

Consideration of the protein pocket dynamics can increase the accuracy of pocket identification and elucidate alternative shapes or even reveal a new transient pocket. For example, a new pocket, a trench, adjacent to the active site was identified in MD simulations49 of HIV integrase, which was exploited in the discovery of HIV integrase inhibitors by Merck & Co., leading to the development of the drug Raltegravir.50 We suggest that drug design approaches should be adapted for particular types of pocket dynamics. The dynamics of channels and adjacent pockets (particularly allosteric pockets) are relevant for the design of bivalent ligands. Such ligands enable simultaneous occupation of two binding sites, for example, the allosteric and orthosteric binding sites in Gprotein-coupled receptors.51 Differences in the dynamics of structurally similar proteins can alter binding kinetics and, thus, be used for improving ligand binding selectivity. For instance, differences in the pocket breathing motions in bacterial and human thymidylate synthases were found to affect active site accessibility and provide an explanation for the antibacterial selectivity of a set of inhibitors.52 Another example where the protein dynamics appears to influence drug selectivity is the cytochrome P450 CYP51.53 MD simulations showed that CYP51 had a more rigid active site but greater tunnel flexibility than other CYPs studied, which may facilitate the design of selective inhibitors of parasitic CYP51s.54 Furthermore, the dynamic tunnels in CYP51 from the buried active site to the



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Present Address ∇

KNIME.com AG, Technoparkstr. 1, 8005 Zurich, Switzerland

Author Contributions

The manuscript was written through the contributions of all authors. Funding

The authors’ work is supported by the Klaus Tschira Foundation, the German Federal Ministry of Education and Research (BMBF) Virtual Liver Network [Grant 0315749], the EU/EFPIA Innovative Medicines Initiative (IMI) Joint Undertaking K4DD project [Grant 115366], and the EU FEP Flagship Program Human Brain Project [Grant 604102]. This paper reflects only the authors’ views, the BMBF, the IMI, and the European Commission are not liable for any use that may be made of the information contained herein. 813

DOI: 10.1021/acs.accounts.5b00516 Acc. Chem. Res. 2016, 49, 809−815

Article

Accounts of Chemical Research Notes

(13) Greives, N.; Zhou, H.-X. Both Protein Dynamics and Ligand Concentration Can Shift the Binding Mechanism between Conformational Selection and Induced Fit. Proc. Natl. Acad. Sci. U. S. A. 2014, 111, 10197−10202. (14) Arkin, M. R.; Randal, M.; DeLano, W. L.; Hyde, J.; Luong, T. N.; Oslob, J. D.; Raphael, D. R.; Taylor, L.; Wang, J.; McDowell, R. S.; Wells, J. A.; Braisted, A. C. Binding of Small Molecules to an Adaptive Protein-Protein Interface. Proc. Natl. Acad. Sci. U. S. A. 2003, 100, 1603−1608. (15) Hyde, J.; Braisted, A. C.; Randal, M.; Arkin, M. R. Discovery and Characterization of Cooperative Ligand Binding in the Adaptive Region of Interleukin-2. Biochemistry 2003, 42, 6475−6483. (16) Tassin-Moindrot, S.; Caille, A.; Douliez, J. P.; Marion, D.; Vovelle, F. The Wide Binding Properties of a Wheat Nonspecific Lipid Transfer Protein. Solution Structure of a Complex with Prostaglandin B2. Eur. J. Biochem. 2000, 267, 1117−1124. (17) Pargellis, C.; Tong, L.; Churchill, L.; Cirillo, P. F.; Gilmore, T.; Graham, A. G.; Grob, P. M.; Hickey, E. R.; Moss, N.; Pav, S.; Regan, J. Inhibition of p38 MAP Kinase by Utilizing a Novel Allosteric Binding Site. Nat. Struct. Biol. 2002, 9, 268−272. (18) Henrich, S.; Salo-Ahen, O. M. H.; Huang, B.; Rippmann, F.; Cruciani, G.; Wade, R. C. Computational Approaches to Identifying and Characterizing Protein Binding Sites for Ligand Design. J. Mol. Recognit. 2010, 23, 209−219. (19) Schreyer, A.; Blundell, T. CREDO: A Protein-Ligand Interaction Database for Drug Discovery. Chem. Biol. Drug Des. 2009, 73, 157−167. (20) Hopkins, A. L.; Groom, C. R. The Druggable Genome. Nat. Rev. Drug Discovery 2002, 1, 727−730. (21) Huang, B. MetaPocket: A Meta Approach to Improve Protein Ligand Binding Site Prediction. OMICS 2009, 13, 325−330. (22) Carugo, O.; Argos, P. Accessibility to Internal Cavities and Ligand Binding Sites Monitored by Protein Crystallographic Thermal Factors. Proteins: Struct., Funct., Genet. 1998, 31, 201−213. (23) Lüdemann, S. K.; Carugo, O.; Wade, R. C. Substrate Access to Cytochrome P450cam: A Comparison of a Thermal Motion Pathway Analysis with Molecular Dynamics Simulation Data. J. Mol. Model. 1997, 3, 369−374. (24) Chovancova, E.; Pavelka, A.; Benes, P.; Strnad, O.; Brezovsky, J.; Kozlikova, B.; Gora, A.; Sustr, V.; Klvana, M.; Medek, P.; Biedermannova, L.; Sochor, J.; Damborsky, J. CAVER 3.0: A Tool for the Analysis of Transport Pathways in Dynamic Protein Structures. PLoS Comput. Biol. 2012, 8, e1002708. (25) Brezovsky, J.; Chovancova, E.; Gora, A.; Pavelka, A.; Biedermannova, L.; Damborsky, J. Software Tools for Identification, Visualization and Analysis of Protein Tunnels and Channels. Biotechnol. Adv. 2013, 31, 38−49. (26) Eyrisch, S.; Helms, V. Transient Pockets on Protein Surfaces Involved in Protein-Protein Interaction. J. Med. Chem. 2007, 50, 3457−3464. (27) Ahmed, A.; Rippmann, F.; Barnickel, G.; Gohlke, H. A Normal Mode-Based Geometric Simulation Approach for Exploring Biologically Relevant Conformational Transitions in Proteins. J. Chem. Inf. Model. 2011, 51, 1604−1622. (28) Seeliger, D.; Haas, J.; de Groot, B. L. Geometry-Based Sampling of Conformational Transitions in Proteins. Structure 2007, 15, 1482− 1492. (29) Ashford, P.; Moss, D. S.; Alex, A.; Yeap, S. K.; Povia, A.; Nobeli, I.; Williams, M. A. Visualisation of Variable Binding Pockets on Protein Surfaces by Probabilistic Analysis of Related Structure Sets. BMC Bioinf. 2012, 13, 39. (30) Seeliger, D.; de Groot, B. L. Conformational Transitions upon Ligand Binding: Holo-Structure Prediction from Apo Conformations. PLoS Comput. Biol. 2010, 6, e1000634. (31) Wells, S.; Menor, S.; Hespenheide, B.; Thorpe, M. F. Constrained Geometric Simulation of Diffusive Motion in Proteins. Phys. Biol. 2005, 2, S127−S136. (32) Metz, A.; Pfleger, C.; Kopitz, H.; Pfeiffer-Marek, S.; Baringhaus, K. H.; Gohlke, H. Hot Spots and Transient Pockets: Predicting the

The authors declare no competing financial interest. Biographies Antonia Stank studied Bioinformatics at Ludwig-MaximiliansUniversity and the Technical University Munich and received her MSc in 2012. She is currently working on her doctoral studies at HITS on computational studies on macromolecular dynamics and the relation to protein binding and function. Daria Kokh obtained her doctorate in Physics and Mathematics in 1995 at St. Petersburg University in Russia. Her research interests are in the development and application of computational methods for modeling physical processes in molecular systems. Currently, she is working at HITS on the simulation of protein−drug binding kinetics. Jonathan Fuller received his doctorate in Bioinformatics in 2010 from the University of Leeds, UK, having studied protein−protein interaction inhibitors by molecular modeling and simulation. As a postdoctoral researcher at HITS from 2010 to 2015, he applied knowledge of protein structure to systems biology problems. He is now an applications scientist at KNIME.com AG. Rebecca Wade studied at Oxford University and received her doctorate in Molecular Biophysics in 1988. She leads the Molecular and Cellular Modeling group at HITS and holds a professorship in “Computational Structural Biology” at Heidelberg University. Her research is focused on the development and application of computeraided methods to model and simulate biomolecular interactions.



REFERENCES

(1) Fischer, E. Einfluss Der Configuration Auf Die Wirkung Der Enzyme. Ber. Dtsch. Chem. Ges. 1894, 27, 2985−2993. (2) Koshland, D. E. Application of a Theory of Enzyme Specificity to Protein Synthesis. Proc. Natl. Acad. Sci. U. S. A. 1958, 44, 98−104. (3) Monod, J.; Wyman, J.; Changeux, J. P. On the Nature of Allosteric Transitions: A Plausible Model. J. Mol. Biol. 1965, 12, 88− 118. (4) Changeux, J.-P.; Edelstein, S. Conformational Selection or Induced Fit? 50 Years of Debate Resolved. F1000 Biol. Rep. 2011, 3, 19. (5) Vogt, A. D.; Di Cera, E. Conformational Selection or Induced Fit? A Critical Appraisal of the Kinetic Mechanism. Biochemistry 2012, 51, 5894−5902. (6) Watanabe, M.; Becker, O. Dynamics Methods. In Computational Biochemistry and Biophysics; Becker, O. M., MacKerell, A. D., Roux, B., Watanabe, M., Eds.; CRC Press: New York, 2001; pp 39−68. (7) Zhou, H.-X. From Induced Fit to Conformational Selection: A Continuum of Binding Mechanism Controlled by the Timescale of Conformational Transitions. Biophys. J. 2010, 98, L15−L17. (8) Hammes, G. G.; Chang, Y.-C.; Oas, T. G. Conformational Selection or Induced Fit: A Flux Description of Reaction Mechanism. Proc. Natl. Acad. Sci. U. S. A. 2009, 106, 13737−13741. (9) Gianni, S.; Dogan, J.; Jemth, P. Distinguishing Induced Fit from Conformational Selection. Biophys. Chem. 2014, 189, 33−39. (10) Kim, Y. B.; Kalinowski, S. S.; Marcinkeviciene, J. A Pre-Steady State Analysis of Ligand Binding to Human Glucokinase: Evidence for a Preexisting Equilibrium. Biochemistry 2007, 46, 1423−1431. (11) Gooljarsingh, L. T.; Fernandes, C.; Yan, K.; Zhang, H.; Grooms, M.; Johanson, K.; Sinnamon, R. H.; Kirkpatrick, R. B.; Kerrigan, J.; Lewis, T.; Arnone, M.; King, A. J.; Lai, Z.; Copeland, R. a; Tummino, P. J. A Biochemical Rationale for the Anticancer Effects of Hsp90 Inhibitors: Slow, Tight Binding Inhibition by Geldanamycin and Its Analogues. Proc. Natl. Acad. Sci. U. S. A. 2006, 103, 7625−7630. (12) Copeland, R. A. Conformational Adaptation in Drug-Target Interactions and Residence Time. Future Med. Chem. 2011, 3, 1491− 1501. 814

DOI: 10.1021/acs.accounts.5b00516 Acc. Chem. Res. 2016, 49, 809−815

Article

Accounts of Chemical Research

(52) Ferrari, S.; Costi, P. M.; Wade, R. C. Inhibitor Specificity via Protein Dynamics: Insights from the Design of Antibacterial Agents Targeted Against Thymidylate Synthase. Chem. Biol. 2003, 10, 1183− 1193. (53) Yu, X.; Cojocaru, V.; Mustafa, G.; Salo-Ahen, O. M. H.; Lepesheva, G. I.; Wade, R. C. Dynamics of CYP51: Implications for Function and Inhibitor Design. J. Mol. Recognit. 2015, 28, 59−73. (54) Lepesheva, G. I.; Ott, R. D.; Hargrove, T. Y.; Kleshchenko, Y. Y.; Schuster, I.; Nes, W. D.; Hill, G. C.; Villalta, F.; Waterman, M. R. Sterol 14alpha-Demethylase as a Potential Target for Antitrypanosomal Therapy: Enzyme Inhibition and Parasite Cell Growth. Chem. Biol. 2007, 14, 1283−1293. (55) Bartolowits, M.; Davisson, V. J. Considerations of Protein Subpockets in Fragment-Based Drug Design. Chem. Biol. Drug Des. 2016, 87, 5−20. (56) Wang, T.; Bisson, W. H.; Mäser, P.; Scapozza, L.; Picard, D. Differences in Conformational Dynamics between Plasmodium Falciparum and Human Hsp90 Orthologues Enable the StructureBased Discovery of Pathogen-Selective Inhibitors. J. Med. Chem. 2014, 57, 2524−2535. (57) Jansen, C.; Wang, H.; Kooistra, A. J.; de Graaf, C.; Orrling, K. M.; Tenor, H.; Seebeck, T.; Bailey, D.; de Esch, I. J. P.; Ke, H.; Leurs, R. Discovery of Novel Trypanosoma Brucei Phosphodiesterase B1 Inhibitors by Virtual Screening against the Unliganded TbrPDEB1 Crystal Structure. J. Med. Chem. 2013, 56, 2087−2096. (58) Kunze, J.; Todoroff, N.; Schneider, P.; Rodrigues, T.; Geppert, T.; Reisen, F.; Schreuder, H.; Saas, J.; Hessler, G.; Baringhaus, K. H.; Schneider, G. Targeting Dynamic Pockets of HIV-1 Protease by Structure-Based Computational Screening for Allosteric Inhibitors. J. Chem. Inf. Model. 2014, 54, 987−991. (59) Jacobson, M. P.; Friesner, R. A.; Xiang, Z.; Honig, B. On the Role of the Crystal Environment in Determining Protein Side-Chain Conformations. J. Mol. Biol. 2002, 320, 597−608. (60) Jacobson, M. P.; Pincus, D. L.; Rapp, C. S.; Day, T. J. F.; Honig, B.; Shaw, D. E.; Friesner, R. A. A Hierarchical Approach to All-Atom Protein Loop Prediction. Proteins: Struct., Funct., Genet. 2004, 55, 351− 367. (61) Stebbins, C. E.; Russo, A. A.; Schneider, C.; Rosen, N.; Hartl, F. U.; Pavletich, N. P. Crystal Structure of an Hsp90-Geldanamycin Complex: Targeting of a Protein Chaperone by an Antitumor Agent. Cell 1997, 89, 239−250. (62) Wright, L.; Barril, X.; Dymock, B.; Sheridan, L.; Surgenor, A.; Beswick, M.; Drysdale, M.; Collier, A.; Massey, A.; Davies, N.; Fink, A.; Fromont, C.; Aherne, W.; Boxall, K.; Sharp, S.; Workman, P.; Hubbard, R. E. Structure-Activity Relationships in Purine-Based Inhibitor Binding to HSP90 Isoforms. Chem. Biol. 2004, 11, 775−785. (63) Thanos, C. D.; Randal, M.; Wells, J. A. Potent Small-Molecule Binding to a Dynamic Hot Spot on IL-2. J. Am. Chem. Soc. 2003, 125, 15280−15281. (64) Lessene, G.; Czabotar, P. E.; Sleebs, B. E.; Zobel, K.; Lowes, K. N.; Adams, J. M.; Baell, J. B.; Colman, P. M.; Deshayes, K.; Fairbrother, W. J.; Flygare, J. a; Gibbons, P.; Kersten, W. J. a; Kulasegaram, S.; Moss, R. M.; Parisot, J. P.; Smith, B. J.; Street, I. P.; Yang, H.; Huang, D. C. S.; Watson, K. G. Structure-Guided Design of a Selective BCL-X(L) Inhibitor. Nat. Chem. Biol. 2013, 9, 390−397. (65) Sleebs, B. E.; Czabotar, P. E.; Fairbrother, W. J.; Fairlie, W. D.; Flygare, J. A.; Huang, D. C. S.; Kersten, W. J. A.; Koehler, M. F. T.; Lessene, G.; Lowes, K.; Parisot, J. P.; Smith, B. J.; Smith, M. L.; Souers, A. J.; Street, I. P.; Yang, H.; Baell, J. B. Quinazoline Sulfonamides as Dual Binders of the Proteins B-Cell Lymphoma 2 and B-Cell Lymphoma Extra Long with Potent Proapoptotic Cell-Based Activity. J. Med. Chem. 2011, 54, 1914−1926. (66) Underwood, K. W.; Parris, K. D.; Federico, E.; Mosyak, L.; Czerwinski, R. M.; Shane, T.; Taylor, M.; Svenson, K.; Liu, Y.; Hsiao, C. L.; Wolfrom, S.; Maguire, M.; Malakian, K.; Telliez, J. B.; Lin, L. L.; Kriz, R. W.; Seehra, J.; Somers, W. S.; Stahl, M. L. Catalytically Active MAP KAP Kinase 2 Structures in Complex with Staurosporine and ADP Reveal Differences with the Autoinhibited Enzyme. Structure 2003, 11, 627−636.

Determinants of Small-Molecule Binding to a Protein-Protein Interface. J. Chem. Inf. Model. 2012, 52, 120−133. (33) Schmidtke, P.; Bidon-Chanal, A.; Luque, F. J.; Barril, X. MDpocket: Open-Source Cavity Detection and Characterization on Molecular Dynamics Trajectories. Bioinformatics 2011, 27, 3276− 3285. (34) Craig, I. R.; Pfleger, C.; Gohlke, H.; Essex, J. W.; Spiegel, K. Pocket-Space Maps to Identify Novel Binding-Site Conformations in Proteins. J. Chem. Inf. Model. 2011, 51, 2666−2679. (35) Kokh, D. B.; Richter, S.; Henrich, S.; Czodrowski, P.; Rippmann, F.; Wade, R. C. TRAPP: A Tool for Analysis of Transient Binding Pockets in Proteins. J. Chem. Inf. Model. 2013, 53, 1235−1252. (36) Goncearenco, A.; Mitternacht, S.; Yong, T.; Eisenhaber, B.; Eisenhaber, F.; Berezovsky, I. N. SPACER: Server for Predicting Allosteric Communication and Effects of Regulation. Nucleic Acids Res. 2013, 41, W266−W272. (37) Kaya, C.; Armutlulu, A.; Ekesan, S.; Haliloglu, T. MCPath: Monte Carlo Path Generation Approach to Predict Likely Allosteric Pathways and Functional Residues. Nucleic Acids Res. 2013, 41, W249−W255. (38) Paramo, T.; East, A.; Garzón, D.; Ulmschneider, M. B.; Bond, P. J. Efficient Characterization of Protein Cavities within Molecular Simulation Trajectories: Trj_cavity. J. Chem. Theory Comput. 2014, 10, 2151−2164. (39) Parulek, J.; Turkay, C.; Reuter, N.; Viola, I. Visual Cavity Analysis in Molecular Simulations. BMC Bioinf. 2013, 14, S4. (40) Petřek, M.; Košinová, P.; Koča, J.; Otyepka, M. MOLE: A Voronoi Diagram-Based Explorer of Molecular Channels, Pores, and Tunnels. Structure 2007, 15, 1357−1363. (41) Pan, A. C.; Borhani, D. W.; Dror, R. O.; Shaw, D. E. Molecular Determinants of Drug-Receptor Binding Kinetics. Drug Discovery Today 2013, 18, 667−673. (42) Romanowska, J.; Kokh, D. B.; Fuller, J. C.; Wade, R. C. Computational Approaches for Studying Drug Binding Kinetics. In Thermodynamics and Kinetics of Drug Binding; Wiley-VCH Verlag GmbH & Co. KGaA: Weinheim, 2015; pp 211−235. (43) Klebe, G. The Use of Thermodynamic and Kinetic Data in Drug Discovery: Decisive Insight or Increasing the Puzzlement? ChemMedChem 2015, 10, 229−231. (44) McCammon, J. A.; Northrup, S. H. Gated Binding of Ligands to Proteins. Nature 1981, 293, 316−317. (45) Zhou, H. X.; Wlodek, S. T.; McCammon, J. a. Conformation Gating as a Mechanism for Enzyme Specificity. Proc. Natl. Acad. Sci. U. S. A. 1998, 95, 9280−9283. (46) Garvey, E. P. Structural Mechanisms of Slow-Onset, Two-Step Enzyme Inhibition. Curr. Chem. Biol. 2010, 4, 64−73. (47) Li, H. J.; Lai, C. T.; Pan, P.; Yu, W.; Liu, N.; Bommineni, G. R.; Garcia-Diaz, M.; Simmerling, C.; Tonge, P. J. A Structural and Energetic Model for the Slow-Onset Inhibition of the Mycobacterium Tuberculosis ENoyl-ACP Reductase InhA. ACS Chem. Biol. 2014, 9, 986−993. (48) Wilson, C.; Agafonov, R. V.; Hoemberger, M.; Kutter, S.; Zorba, A.; Halpin, J.; Buosi, V.; Otten, R.; Waterman, D.; Theobald, D. L.; Kern, D. Using Ancient Protein Kinases to Unravel a Modern Cancer Drug’s Mechanism. Science 2015, 347, 882−886. (49) Schames, J.; Henchman, R.; Siegel, J.; Sotriffer, C.; Haihong, N.; McCammon, J. A. Discovery of a Novel Binding Trench in HIV Integrase. J. Med. Chem. 2004, 47, 1879−1881. (50) Summa, V.; Petrocchi, A.; Bonelli, F.; Crescenzi, B.; Donghi, M.; Ferrara, M.; Fiore, F.; Gardelli, C.; Paz, O. G.; Hazuda, D. J.; Jones, P.; Kinzel, O.; Laufer, R.; Monteagudo, E.; Muraglia, E.; Nizi, E.; Orvieto, F.; Pace, P.; Pescatore, G.; Scarpelli, R.; Stillmock, K.; Witmer, M. V.; Rowley, M. Discovery of Raltegravir, a Potent, Selective Orally Bioavailable HIV-Integrase Inhibitor for the Treatment of HIV-AIDS Infection. J. Med. Chem. 2008, 51, 5843−5855. (51) Shonberg, J.; Scammells, P. J.; Capuano, B. Design Strategies for Bivalent Ligands Targeting GPCRs. ChemMedChem 2011, 6, 963− 974. 815

DOI: 10.1021/acs.accounts.5b00516 Acc. Chem. Res. 2016, 49, 809−815