Are Deep Learning Structural Models Sufficiently Accurate for Virtual Screening? Application of Docking Algorithms to AlphaFold2 Predicted Structures

Machine learning-based protein structure prediction algorithms, such as RosettaFold and AlphaFold2, have greatly impacted the structural biology field, arousing a fair amount of discussion around their potential role in drug discovery. While there are few preliminary studies addressing the usage of these models in virtual screening, none of them focus on the prospect of hit-finding in a real-world virtual screen with a model based on low prior structural information. In order to address this, we have developed an AlphaFold2 version where we exclude all structural templates with more than 30% sequence identity from the model-building process. In a previous study, we used those models in conjunction with state-of-the-art free energy perturbation methods and demonstrated that it is possible to obtain quantitatively accurate results. In this work, we focus on using these structures in rigid receptor-ligand docking studies. Our results indicate that using out-of-the-box Alphafold2 models is not an ideal scenario for virtual screening campaigns; in fact, we strongly recommend to include some post-processing modeling to drive the binding site into a more realistic holo model.

Structure-Activity Relationship Studies of Antimalarial Plasmodium Proteasome Inhibitors─Part II

With increasing reports of resistance to artemisinins and artemisinin-combination therapies, targeting the Plasmodium proteasome is a promising strategy for antimalarial development. We recently reported a highly selective Plasmodium falciparum proteasome inhibitor with anti-malarial activity in the humanized mouse model. To balance the permeability of the series of macrocycles with other drug-like properties, we conducted further structure-activity relationship studies on a biphenyl ether-tethered macrocyclic scaffold. Extensive SAR studies around the P1, P3, and P5 groups and peptide backbone identified compound TDI-8414. TDI-8414 showed nanomolar antiparasitic activity, no toxicity to HepG2 cells, high selectivity against the Plasmodium proteasome over the human constitutive proteasome and immunoproteasome, improved solubility and PAMPA permeability, and enhanced metabolic stability in microsomes and plasma of both humans and mice.

Are Deep Learning Structural Models Sufficiently Accurate for Free-Energy Calculations? Application of FEP+ to AlphaFold2-Predicted Structures

The availability of AlphaFold2 has led to great excitement in the scientific community─particularly among drug hunters─due to the ability of the algorithm to predict protein structures with high accuracy. However, beyond globally accurate protein structure prediction, it remains to be determined whether ligand binding sites are predicted with sufficient accuracy in these structures to be useful in supporting computationally driven drug discovery programs. We explored this question by performing free-energy perturbation (FEP) calculations on a set of well-studied protein-ligand complexes, where AlphaFold2 predictions were performed by removing all templates with >30% identity to the target protein from the training set. We observed that in most cases, the ΔΔG values for ligand transformations calculated with FEP, using these prospective AlphaFold2 structures, were comparable in accuracy to the corresponding calculations previously carried out using crystal structures. We conclude that under the right circumstances, AlphaFold2-modeled structures are accurate enough to be used by physics-based methods such as FEP in typical lead optimization stages of a drug discovery program.

Design, Synthesis, and Optimization of Macrocyclic Peptides as Species-Selective Antimalaria Proteasome Inhibitors

With over 200 million cases and close to half a million deaths each year, malaria is a threat to global health, particularly in developing countries. Plasmodium falciparum, the parasite that causes the most severe form of the disease, has developed resistance to all antimalarial drugs. Resistance to the first-line antimalarial artemisinin and to artemisinin combination therapies is widespread in Southeast Asia and is emerging in sub-Saharan Africa. The P. falciparum proteasome is an attractive antimalarial target because its inhibition kills the parasite at multiple stages of its life cycle and restores artemisinin sensitivity in parasites that have become resistant through mutation in Kelch K13. Here, we detail our efforts to develop noncovalent, macrocyclic peptide malaria proteasome inhibitors, guided by structural analysis and pharmacokinetic properties, leading to a potent, species-selective, metabolically stable inhibitor.

Development of a Highly Selective Plasmodium falciparum Proteasome Inhibitor with Anti-malaria Activity in Humanized Mice

Plasmodium falciparum proteasome (Pf20S) inhibitors are active against Plasmodium at multiple stages-erythrocytic, gametocyte, liver, and gamete activation stages-indicating that selective Pf20S inhibitors possess the potential to be therapeutic, prophylactic, and transmission-blocking antimalarials. Starting from a reported compound, we developed a noncovalent, macrocyclic peptide inhibitor of the malarial proteasome with high species selectivity and improved pharmacokinetic properties. The compound demonstrates specific, time-dependent inhibition of the β5 subunit of the Pf20S, kills artemisinin-sensitive and artemisinin-resistant P. falciparum isolates in vitro and reduces parasitemia in humanized, P. falciparum-infected mice.

Rigorous Free Energy Simulations in Virtual Screening

Virtual high throughput screening (vHTS) in drug discovery is a powerful approach to identify hits: when applied successfully, it can be much faster and cheaper than experimental high-throughput screening approaches. However, mainstream vHTS tools have significant limitations: ligand-based methods depend on knowledge of existing chemical matter, while structure-based tools such as docking involve significant approximations that limit their accuracy. Recent advances in scientific methods coupled with dramatic speedups in computational processing with GPUs make this an opportune time to consider the role of more rigorous methods that could improve the predictive power of vHTS workflows. In this Perspective, we assert that alchemical binding free energy methods using all-atom molecular dynamics simulations have matured to the point where they can be applied in virtual screening campaigns as a final scoring stage to prioritize the top molecules for experimental testing. Specifically, we propose that alchemical absolute binding free energy (ABFE) calculations offer the most direct and computationally efficient approach within a rigorous statistical thermodynamic framework for computing binding energies of diverse molecules, as is required for virtual screening. ABFE calculations are particularly attractive for drug discovery at this point in time, where the confluence of large-scale genomics data and insights from chemical biology have unveiled a large number of promising disease targets for which no small molecule binders are known, precluding ligand-based approaches, and where traditional docking approaches have foundered to find progressible chemical matter.

Abstract 4142: Conformational dynamics of the chimeric kinase DNAJB1-PRKACA, the driver for fibrolamellar hepatocellular carcinoma

In fibrolamellar hepatocellular carcinoma (FLC) there is a single common genetic alteration in all tumors: A deletion in one copy of chromosome 19 resulting in the fusion of the first exon of DNAJB1, encoding the J-domain of heat shock protein 40, with exons 2-10 of the catalytic subunit of protein kinase A, PRKACA [1]. This produces an enzymatically active chimeric protein. Expression of the chimera in mouse liver either by recreating the deletion by CRISPR/Cas9, or expression, in trans, by a transposon is sufficient to produce FLC [2]. We used molecular dynamics simulations, NMR, and SAXS to analyze the conformational dynamics of the native and chimeric kinase. The most significant differences were in the amino domain with the chimera in an ensemble of conformations. Some structures had the J-domain tucked under the large lobe of the kinase, similar to that reported in the crystal structure, and in others the J-domain was dislodged from the core of the kinase, swinging free in solution. These simulated dislodged states were captured experimentally both by NMR and small angle X-ray scattering. Modeling of the different conformations onto the RIIβ holoenzyme revealed no obvious steric interactions suggesting that the J-domain does not preclude formation of the holoenzyme. The flexible conformations appear to be independent of the nucleotide/substrate binding mode as they were observed in ATP, ADP, Apo, and ATP-PKI bound states.

[1] Honeyman, J. N. et al. Detection of a recurrent DNAJB1-PRKACA chimeric transcript in fibrolamellar hepatocellular carcinoma. Science 343, 1010-1014 (2014).

[2] Kastenhuber, E. R. et al. DNAJB1-PRKACA fusion kinase interacts with β-catenin and the liver regenerative response to drive fibrolamellar hepatocellular carcinoma. Proc. Natl. Acad. Sci. 201716483 (2017). doi:10.1073/pnas.1716483114

Citation Format: Michael D. Tomasini, Yingiie Wang, Adak Karamafrooz, James Hall, Susan S. Taylor, Thijs Beuming, Gianluigi Veglia, Sanford M. Simon. Conformational dynamics of the chimeric kinase DNAJB1-PRKACA, the driver for fibrolamellar hepatocellular carcinoma [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 4142.

Identification of a Mycothiol-Dependent Nitroreductase from Mycobacterium tuberculosis

The success of Mycobacterium tuberculosis (Mtb) as a pathogen depends on the redundant and complex mechanisms it has evolved for resisting nitrosative and oxidative stresses inflicted by host immunity. Improving our understanding of these defense pathways can reveal vulnerable points in Mtb pathogenesis. In this study, we combined genetic, structural, computational, biochemical, and biophysical approaches to identify a novel enzyme class represented by Rv2466c. We show that Rv2466c is a mycothiol-dependent nitroreductase of Mtb and can reduce the nitro group of a novel mycobactericidal compound using mycothiol as a cofactor. In addition to its function as a nitroreductase, Rv2466c confers partial protection to menadione stress.

Conformational Landscape of the PRKACA-DNAJB1 Chimeric Kinase, the Driver for Fibrolamellar Hepatocellular Carcinoma

In fibrolamellar hepatocellular carcinoma a single genetic deletion results in the fusion of the first exon of the heat shock protein 40, DNAJB1, which encodes the J domain, with exons 2-10 of the catalytic subunit of protein kinase A, PRKACA. This produces an enzymatically active chimeric protein J-PKAcα. We used molecular dynamics simulations and NMR to analyze the conformational landscape of native and chimeric kinase, and found an ensemble of conformations. These ranged from having the J-domain tucked under the large lobe of the kinase, similar to what was reported in the crystal structure, to others where the J-domain was dislodged from the core of the kinase and swinging free in solution. These simulated dislodged states were experimentally captured by NMR. Modeling of the different conformations revealed no obvious steric interactions of the J-domain with the rest of the RIIβ holoenzyme.

Investigating Protein-Peptide Interactions Using the Schrödinger Computational Suite

The Schrödinger software suite contains a broad array of computational chemistry and molecular modeling tools that can be used to study the interaction of peptides with proteins. These include molecular docking using Glide and Piper, relative binding free energy predictions with FEP+, conformational searches using MacroModel and Desmond, and structural refinement using Prime and PrimeX. In this review we provide a comprehensive overview of these tools and describe their potential application in the identification and optimization of peptide ligands for proteins.

Relative Binding Free Energy Calculations Applied to Protein Homology Models

A significant challenge and potential high-value application of computer-aided drug design is the accurate prediction of protein-ligand binding affinities. Free energy perturbation (FEP) using molecular dynamics (MD) sampling is among the most suitable approaches to achieve accurate binding free energy predictions, due to the rigorous statistical framework of the methodology, correct representation of the energetics, and thorough treatment of the important degrees of freedom in the system (including explicit waters). Recent advances in sampling methods and force fields coupled with vast increases in computational resources have made FEP a viable technology to drive hit-to-lead and lead optimization, allowing for more efficient cycles of medicinal chemistry and the possibility to explore much larger chemical spaces. However, previous FEP applications have focused on systems with high-resolution crystal structures of the target as starting points-something that is not always available in drug discovery projects. As such, the ability to apply FEP on homology models would greatly expand the domain of applicability of FEP in drug discovery. In this work we apply a particular implementation of FEP, called FEP+, on congeneric ligand series binding to four diverse targets: a kinase (Tyk2), an epigenetic bromodomain (BRD4), a transmembrane GPCR (A2A), and a protein-protein interaction interface (BCL-2 family protein MCL-1). We apply FEP+ using both crystal structures and homology models as starting points and find that the performance using homology models is generally on a par with the results when using crystal structures. The robustness of the calculations to structural variations in the input models can likely be attributed to the conformational sampling in the molecular dynamics simulations, which allows the modeled receptor to adapt to the "real" conformation for each ligand in the series. This work exemplifies the advantages of using all-atom simulation methods with full system flexibility and offers promise for the general application of FEP to homology models, although additional validation studies should be performed to further understand the limitations of the method and the scenarios where FEP will work best.

Predicting Binding Affinities for GPCR Ligands Using Free-Energy Perturbation

The rapid growth of structural information for G-protein-coupled receptors (GPCRs) has led to a greater understanding of their structure, function, selectivity, and ligand binding. Although novel ligands have been identified using methods such as virtual screening, computationally driven lead optimization has been possible only in isolated cases because of challenges associated with predicting binding free energies for related compounds. Here, we provide a systematic characterization of the performance of free-energy perturbation (FEP) calculations to predict relative binding free energies of congeneric ligands binding to GPCR targets using a consistent protocol and no adjustable parameters. Using the FEP+ package, first we validated the protocol, which includes a full lipid bilayer and explicit solvent, by predicting the binding affinity for a total of 45 different ligands across four different GPCRs (adenosine A_2AAR, β₁ adrenergic, CXCR4 chemokine, and δ opioid receptors). Comparison with experimental binding affinity measurements revealed a highly predictive ranking correlation (average spearman ρ = 0.55) and low root-mean-square error (0.80 kcal/mol). Next, we applied FEP+ in a prospective project, where we predicted the affinity of novel, potent adenosine A_2A receptor (A_2AR) antagonists. Four novel compounds were synthesized and tested in a radioligand displacement assay, yielding affinity values in the nanomolar range. The affinity of two out of the four novel ligands (plus three previously reported compounds) was correctly predicted (within 1 kcal/mol), including one compound with approximately a tenfold increase in affinity compared to the starting compound. Detailed analyses of the simulations underlying the predictions provided insights into the structural basis for the two cases where the affinity was overpredicted. Taken together, these results establish a protocol for systematically applying FEP+ to GPCRs and provide guidelines for identifying potent molecules in drug discovery lead optimization projects.

What can crystal structures of aminergic receptors tell us about designing subtype-selective ligands?

G protein-coupled receptors (GPCRs) are integral membrane proteins that represent an important class of drug targets. In particular, aminergic GPCRs interact with a significant portion of drugs currently on the market. However, most drugs that target these receptors are associated with undesirable side effects, which are due in part to promiscuous interactions with close homologs of the intended target receptors. Here, based on a systematic analysis of all 37 of the currently available high-resolution crystal structures of aminergic GPCRs, we review structural elements that contribute to and can be exploited for designing subtype-selective compounds. We describe the roles of secondary binding pockets (SBPs), as well as differences in ligand entry pathways to the orthosteric binding site, in determining selectivity. In addition, using the available crystal structures, we have identified conformational changes in the SBPs that are associated with receptor activation and explore the implications of these changes for the rational development of selective ligands with tailored efficacy.

Accurate and reliable prediction of relative ligand binding potency in prospective drug discovery by way of a modern free-energy calculation protocol and force field

Designing tight-binding ligands is a primary objective of small-molecule drug discovery. Over the past few decades, free-energy calculations have benefited from improved force fields and sampling algorithms, as well as the advent of low-cost parallel computing. However, it has proven to be challenging to reliably achieve the level of accuracy that would be needed to guide lead optimization (∼5× in binding affinity) for a wide range of ligands and protein targets. Not surprisingly, widespread commercial application of free-energy simulations has been limited due to the lack of large-scale validation coupled with the technical challenges traditionally associated with running these types of calculations. Here, we report an approach that achieves an unprecedented level of accuracy across a broad range of target classes and ligands, with retrospective results encompassing 200 ligands and a wide variety of chemical perturbations, many of which involve significant changes in ligand chemical structures. In addition, we have applied the method in prospective drug discovery projects and found a significant improvement in the quality of the compounds synthesized that have been predicted to be potent. Compounds predicted to be potent by this approach have a substantial reduction in false positives relative to compounds synthesized on the basis of other computational or medicinal chemistry approaches. Furthermore, the results are consistent with those obtained from our retrospective studies, demonstrating the robustness and broad range of applicability of this approach, which can be used to drive decisions in lead optimization.

Docking and Virtual Screening Strategies for GPCR Drug Discovery

Progress in structure determination of G protein-coupled receptors (GPCRs) has made it possible to apply structure-based drug design (SBDD) methods to this pharmaceutically important target class. The quality of GPCR structures available for SBDD projects fall on a spectrum ranging from high resolution crystal structures (<2 Å), where all water molecules in the binding pocket are resolved, to lower resolution (>3 Å) where some protein residues are not resolved, and finally to homology models that are built using distantly related templates. Each GPCR project involves a distinct set of opportunities and challenges, and requires different approaches to model the interaction between the receptor and the ligands. In this review we will discuss docking and virtual screening to GPCRs, and highlight several refinement and post-processing steps that can be used to improve the accuracy of these calculations. Several examples are discussed that illustrate specific steps that can be taken to improve upon the docking and virtual screening accuracy. While GPCRs are a unique target class, many of the methods and strategies outlined in this review are general and therefore applicable to other protein families.

Docking and free energy perturbation studies of ligand binding in the kappa opioid receptor

The kappa opioid receptor (KOR) is an important target for pain and depression therapeutics that lack harmful and addictive qualities of existing medications. We present a model for the binding of morphinan ligands and JDTic to the JDTic/KOR crystal structure based on an atomic level description of the water structure within its active site. The model contains two key interaction motifs that are supported by experimental evidence. The first is the formation of a salt bridge between the ligand and Asp 138(3.32) in transmembrane domain (TM) 3. The second is the stabilization by the ligand of two high energy, isolated, and ice-like waters near TM5 and TM6. This model is incorporated via energetic terms into a new empirical scoring function, WScore, designed to assess interactions between ligands and localized water in a binding site. Pairing WScore with the docking program Glide discriminates known active KOR ligands from large sets of decoy molecules much better than Glide's older generation scoring functions, SP and XP. We also use rigorous free energy perturbation calculations to provide evidence for the proposed mechanism of interaction between ligands and KOR. The molecular description of ligand binding in KOR should provide a good starting point for future drug discovery efforts for this receptor.

Selecting an optimal number of binding site waters to improve virtual screening enrichments against the adenosine A2A receptor

A major challenge in structure-based virtual screening (VS) involves the treatment of explicit water molecules during docking in order to improve the enrichment of active compounds over decoys. Here we have investigated this in the context of the adenosine A2A receptor, where water molecules have previously been shown to be important for achieving high enrichment rates with docking, and where the positions of some binding site waters are known from a high-resolution crystal structure. The effect of these waters (both their presence and orientations) on VS enrichment was assessed using a carefully curated set of 299 high affinity A2A antagonists and 17,337 decoys. We show that including certain crystal waters greatly improves VS enrichment and that optimization of water hydrogen positions is needed in order to achieve the best results. We also show that waters derived from a molecular dynamics simulation - without any knowledge of crystallographic waters - can improve enrichments to a similar degree as the crystallographic waters, which makes this strategy applicable to structures without experimental knowledge of water positions. Finally, we used decision trees to select an ensemble of structures with different water molecule positions and orientations that outperforms any single structure with water molecules. The approach presented here is validated against independent test sets of A2A receptor antagonists and decoys from the literature. In general, this water optimization strategy could be applied to any target with waters-mediated protein-ligand interactions.

Discovery and characterization of a G protein-biased agonist that inhibits β-arrestin recruitment to the D2 dopamine receptor

A high-throughput screening campaign was conducted to interrogate a 380,000+ small-molecule library for novel D2 dopamine receptor modulators using a calcium mobilization assay. Active agonist compounds from the primary screen were examined for orthogonal D2 dopamine receptor signaling activities including cAMP modulation and β-arrestin recruitment. Although the majority of the subsequently confirmed hits activated all signaling pathways tested, several compounds showed a diminished ability to stimulate β-arrestin recruitment. One such compound (MLS1547; 5-chloro-7-[(4-pyridin-2-ylpiperazin-1-yl)methyl]quinolin-8-ol) is a highly efficacious agonist at D2 receptor-mediated G protein-linked signaling, but does not recruit β-arrestin as demonstrated using two different assays. This compound does, however, antagonize dopamine-stimulated β-arrestin recruitment to the D2 receptor. In an effort to investigate the chemical scaffold of MLS1547 further, we characterized a set of 24 analogs of MLS1547 with respect to their ability to inhibit cAMP accumulation or stimulate β-arrestin recruitment. A number of the analogs were similar to MLS1547 in that they displayed agonist activity for inhibiting cAMP accumulation, but did not stimulate β-arrestin recruitment (i.e., they were highly biased). In contrast, other analogs displayed various degrees of G protein signaling bias. These results provided the basis to use pharmacophore modeling and molecular docking analyses to build a preliminary structure-activity relationship of the functionally selective properties of this series of compounds. In summary, we have identified and characterized a novel G protein-biased agonist of the D2 dopamine receptor and identified structural features that may contribute to its biased signaling properties.

A single glycine in extracellular loop 1 is the critical determinant for pharmacological specificity of dopamine D2 and D3 receptors

Subtype-selective agents for the dopamine D3 receptor (D3R) have been considered as potential medications for drug addiction and other neuropsychiatric disorders. Medicinal chemistry efforts have led to the discovery of 4-phenylpiperazine derivatives that are >100-fold selective for the dopamine D3 receptor over dopamine D2 receptor (D2R), despite high sequence identity (78% in the transmembrane domain). Based on the recent crystal structure of D3R, we demonstrated that the 4-phenylpiperazine moiety in this class of D3R-selective compounds binds to the conserved orthosteric binding site, whereas the extended aryl amide moiety is oriented toward a divergent secondary binding pocket (SBP). In an effort to further characterize molecular determinants of the selectivity of these compounds, we modeled their binding modes in D3R and D2R by comparative ligand docking and molecular dynamics simulations. We found that the aryl amide moiety in the SBP differentially induces conformational changes in transmembrane segment 2 and extracellular loop 1 (EL1), which amplify the divergence of the SBP in D3R and D2R. Receptor chimera and site-directed mutagenesis studies were used to validate these binding modes and to identify a divergent glycine in EL1 as critical to D3R over D2R subtype selectivity. A better understanding of drug-dependent receptor conformations such as these is key to the rational design of compounds targeting a specific receptor among closely related homologs, and may also lead to discovery of novel chemotypes that exploit subtle differences in protein conformations.

Improved docking of polypeptides with Glide

Predicting the binding mode of flexible polypeptides to proteins is an important task that falls outside the domain of applicability of most small molecule and protein-protein docking tools. Here, we test the small molecule flexible ligand docking program Glide on a set of 19 non-α-helical peptides and systematically improve pose prediction accuracy by enhancing Glide sampling for flexible polypeptides. In addition, scoring of the poses was improved by post-processing with physics-based implicit solvent MM-GBSA calculations. Using the best RMSD among the top 10 scoring poses as a metric, the success rate (RMSD ≤ 2.0 Å for the interface backbone atoms) increased from 21% with default Glide SP settings to 58% with the enhanced peptide sampling and scoring protocol in the case of redocking to the native protein structure. This approaches the accuracy of the recently developed Rosetta FlexPepDock method (63% success for these 19 peptides) while being over 100 times faster. Cross-docking was performed for a subset of cases where an unbound receptor structure was available, and in that case, 40% of peptides were docked successfully. We analyze the results and find that the optimized polypeptide protocol is most accurate for extended peptides of limited size and number of formal charges, defining a domain of applicability for this approach.

Ligand-dependent activation and deactivation of the human adenosine A(2A) receptor

G-protein-coupled receptors (GPCRs) are membrane proteins with critical functions in cellular signal transduction, representing a primary class of drug targets. Acting by direct binding, many drugs modulate GPCR activity and influence the signaling pathways associated with numerous diseases. However, complete details of ligand-dependent GPCR activation/deactivation are difficult to obtain from experiments. Therefore, it remains unclear how ligands modulate a GPCR's activity. To elucidate the ligand-dependent activation/deactivation mechanism of the human adenosine A2A receptor (AA2AR), a member of the class A GPCRs, we performed large-scale unbiased molecular dynamics and metadynamics simulations of the receptor embedded in a membrane. At the atomic level, we have observed distinct structural states that resemble the active and inactive states. In particular, we noted key structural elements changing in a highly concerted fashion during the conformational transitions, including six conformational states of a tryptophan (Trp246(6.48)). Our findings agree with a previously proposed view that, during activation, this tryptophan residue undergoes a rotameric transition that may be coupled to a series of coherent conformational changes, resulting in the opening of the G-protein binding site. Further, metadynamics simulations provide quantitative evidence for this mechanism, suggesting how ligand binding shifts the equilibrium between the active and inactive states. Our analysis also proposes that a few specific residues are associated with agonism/antagonism, affinity, and selectivity, and suggests that the ligand-binding pocket can be thought of as having three distinct regions, providing dynamic features for structure-based design. Additional simulations with AA2AR bound to a novel ligand are consistent with our proposed mechanism. Generally, our study provides insights into the ligand-dependent AA2AR activation/deactivation in addition to what has been found in crystal structures. These results should aid in the discovery of more effective and selective GPCR ligands.

Structure-based virtual screening of MT2 melatonin receptor: influence of template choice and structural refinement

Developing GPCR homology models for structure-based virtual screening requires the choice of a suitable template and refinement of binding site residues. We explored this systematically for the MT2 melatonin receptor, with the aim to build a receptor homology model that is optimized for the enrichment of active melatoninergic ligands. A set of 12 MT2 melatonin receptor models was built using different GPCR X-ray structural templates and submitted to a virtual screening campaign on a set of compounds composed of 29 known melatonin receptor ligands and 2560 drug-like decoys. To evaluate the effect of including a priori information in receptor models, 12 representative melatonin receptor ligands were placed into the MT2 receptor models in poses consistent with known mutagenesis data and with assessed pharmacophore models. The receptor structures were then adapted to the ligands by induced-fit docking. Most of the 144 ligand-adapted MT2 receptor models showed significant improvements in screening enrichments compared to the unrefined homology models, with some template/refinement combinations giving excellent enrichment factors. The discriminating ability of the models was further tested on the 29 active ligands plus a set of 21 inactive or low-affinity compounds from the same chemical classes. Rotameric states of side chains for some residues, presumed to be involved in the binding process, were correlated with screening effectiveness, suggesting the existence of specific receptor conformations able to recognize active compounds. The top MT2 receptor model was able to identify 24 of 29 active ligands among the first 2% of the screened database. This work provides insights into the use of refined GPCR homology models for virtual screening.

Steric hindrance mutagenesis in the conserved extracellular vestibule impedes allosteric binding of antidepressants to the serotonin transporter

The serotonin transporter (SERT) controls synaptic serotonin levels and is the primary target for antidepressants, including selective serotonin reuptake inhibitors (e.g. (S)-citalopram) and tricyclic antidepressants (e.g. clomipramine). In addition to a high affinity binding site, SERT possesses a low affinity allosteric site for antidepressants. Binding to the allosteric site impedes dissociation of antidepressants from the high affinity site, which may enhance antidepressant efficacy. Here we employ an induced fit docking/molecular dynamics protocol to identify the residues that may be involved in the allosteric binding in the extracellular vestibule located above the central substrate binding (S1) site. Indeed, mutagenesis of selected residues in the vestibule reduces the allosteric potency of (S)-citalopram and clomipramine. The identified site is further supported by the inhibitory effects of Zn(2+) binding in an engineered site and the covalent attachment of benzocaine-methanethiosulfonate to a cysteine introduced in the extracellular vestibule. The data provide a mechanistic explanation for the allosteric action of antidepressants at SERT and suggest that the role of the vestibule is evolutionarily conserved among neurotransmitter:sodium symporter proteins as a binding pocket for small molecule ligands.

Molecular determinants of selectivity and efficacy at the dopamine D3 receptor

The dopamine D3 receptor (D3R) has been implicated in substance abuse and other neuropsychiatric disorders. The high sequence homology between the D3R and D2R, especially within the orthosteric binding site (OBS) that binds dopamine, has made the development of D3R-selective compounds challenging. Here, we deconstruct into pharmacophoric elements a series of D3R-selective substituted-4-phenylpiperazine compounds and use computational simulations and binding and activation studies to dissect the structural bases for D3R selectivity and efficacy. We find that selectivity arises from divergent interactions within a second binding pocket (SBP) separate from the OBS, whereas efficacy depends on the binding mode in the OBS. Our findings reveal structural features of the receptor that are critical to selectivity and efficacy that can be used to design highly D3R-selective ligands with targeted efficacies. These findings are generalizable to other GPCRs in which the SBP can be targeted by bitopic or allosteric ligands.

Thermodynamic analysis of water molecules at the surface of proteins and applications to binding site prediction and characterization

Water plays an essential role in determining the structure and function of all biological systems. Recent methodological advances allow for an accurate and efficient estimation of the thermodynamic properties of water molecules at the surface of proteins. In this work, we characterize these thermodynamic properties and relate them to various structural and functional characteristics of the protein. We find that high-energy hydration sites often exist near protein motifs typically characterized as hydrophilic, such as backbone amide groups. We also find that waters around alpha helices and beta sheets tend to be less stable than waters around loops. Furthermore, we find no significant correlation between the hydration site-free energy and the solvent accessible surface area of the site. In addition, we find that the distribution of high-energy hydration sites on the protein surface can be used to identify the location of binding sites and that binding sites of druggable targets tend to have a greater density of thermodynamically unstable hydration sites. Using this information, we characterize the FKBP12 protein and show good agreement between fragment screening hit rates from NMR spectroscopy and hydration site energetics. Finally, we show that water molecules observed in crystal structures are less stable on average than bulk water as a consequence of the high degree of spatial localization, thereby resulting in a significant loss in entropy. These findings should help to better understand the characteristics of waters at the surface of proteins and are expected to lead to insights that can guide structure-based drug design efforts.

The binding sites for benztropines and dopamine in the dopamine transporter overlap

Analogs of benztropines (BZTs) are potent inhibitors of the dopamine transporter (DAT) but are less effective than cocaine as behavioral stimulants. As a result, there have been efforts to evaluate these compounds as leads for potential medication for cocaine addiction. Here we use computational modeling together with site-directed mutagenesis to characterize the binding site for BZTs in DAT. Docking into molecular models based on the structure of the bacterial homolog LeuT supported a BZT binding site that overlaps with the substrate-binding pocket. In agreement, mutations of residues within the pocket, including(2) Val152(3.46) to Ala or Ile, Ser422(8.60) to Ala and Asn157(3.51) to Cys or Ala, resulted in decreased affinity for BZT and the analog JHW007, as assessed in [(3)H]dopamine uptake inhibition assays and/or [(3)H]CFT competition binding assay. A putative polar interaction of one of the phenyl ring fluorine substituents in JHW007 with Asn157(3.51) was used as a criterion for determining likely binding poses and establish a structural context for the mutagenesis findings. The analysis positioned the other fluorine-substituted phenyl ring of JHW007 in close proximity to Ala479(10.51)/Ala480(10.52) in transmembrane segment (TM) 10. The lack of such an interaction for BZT led to a more tilted orientation, as compared to JHW007, bringing one of the phenyl rings even closer to Ala479(10.51)/Ala480(10.52). Mutation of Ala479(10.51) and Ala480(10.52) to valines supported these predictions with a larger decrease in the affinity for BZT than for JHW007. Summarized, our data suggest that BZTs display a classical competitive binding mode with binding sites overlapping those of cocaine and dopamine.

Hydration Site Thermodynamics Explain SARs for Triazolylpurines Analogues Binding to the A2A Receptor

A series of triazolylpurine analogues show interesting and unintuitive structure-activity relationships against the A2A adenosine receptor. As the 2-substituted aliphatic group is initially increased to methyl and isopropyl, there is a decrease in potency; however, extending the substituent to n-butyl and n-pentyl results in a significant gain in potency. This trend cannot be readily explained by ligand-receptor interactions, steric effects, or differences in ligand desolvation. Here, we show that a novel method for characterizing solvent thermodynamics in protein binding sites correctly predicts the trend in binding affinity for this series based on the differential water displacement patterns. In brief, small unfavorable substituents occupy a region in the A2A adenosine receptor binding site predicted to contain stable waters, while the longer favorable substituents extend to a region that contains several unstable waters. The predicted binding energies associated with displacing water within these hydration sites correlate well with the experimental activities.

High-energy water sites determine peptide binding affinity and specificity of PDZ domains

PDZ domains have well known binding preferences for distinct C-terminal peptide motifs. For most PDZ domains, these motifs are of the form [S/T]-W-[I/L/V]. Although the preference for S/T has been explained by a specific hydrogen bond interaction with a histidine in the PDZ domain and the (I/L/V) is buried in a hydrophobic pocket, the mechanism for Trp specificity at the second to last position has thus far remained unknown. Here, we apply a method to compute the free energies of explicit water molecules and predict that potency gained by Trp binding is due to a favorable release of high-energy water molecules into bulk. The affinities of a series of peptides for both wild-type and mutant forms of the PDZ domain of Erbin correlate very well with the computed free energy of binding of displaced waters, suggesting a direct relationship between water displacement and peptide affinity. Finally, we show a correlation between the magnitude of the displaced water free energy and the degree of Trp-sensitivity among subtypes of the HTRA PDZ family, indicating a water-mediated mechanism for specificity of peptide binding.

A comprehensive structure-based alignment of prokaryotic and eukaryotic neurotransmitter/Na+ symporters (NSS) aids in the use of the LeuT structure to probe NSS structure and function

The recently elucidated crystal structure of a prokaryotic member of the neurotransmitter/sodium symporter (NSS) family (Yamashita et al., 2005) is a major advance toward understanding structure-function relationships in this important class of transporters. To aid in the generalization of these results, we present here a comprehensive sequence alignment of all known prokaryotic and eukaryotic NSS proteins, based on the crystal structure of the leucine transporter from Aquifex aeolicus (LeuT). Regions of low sequence identity between prokaryotic and eukaryotic transporters were aligned with the aid of a number of bioinformatics tools, and the resulting alignments were validated by comparison with experimental data. In a number of regions, including the transmembrane segments 4, 5, and 9 as well as extracellular loops 2, 3, and 4, our alignment differs from the one proposed previously [Nature (Lond) 437: 215-223, 2005]. Important similarities and differences among the sequences of NSS proteins in regions likely to determine selectivity in substrate binding and mechanisms of transport regulation are discussed in the context of the LeuT structure and the alignment.

State-dependent conformations of the translocation pathway in the tyrosine transporter Tyt1, a novel neurotransmitter:sodium symporter from Fusobacterium nucleatum

The gene of a novel prokaryotic member (Tyt1) of the neurotransmitter:sodium symporter (NSS) family has been cloned from Fusobacterium nucleatum. In contrast to eukaryotic and some prokaryotic NSSs, which contain 12 transmembrane domains (TMs), Tyt1 contains only 11 TMs, a characteristic shared by approximately 70% of prokaryotic NSS homologues. Nonetheless upon heterologous expression in an engineered Escherichia coli host, Tyt1 catalyzes robust Na+-dependent, highly selective l-tyrosine transport. Genetic engineering of Tyt1 variants devoid of cysteines or with individually retained endogenous cysteines at positions 18 or 238, at the cytoplasmic ends of TM1 and TM6, respectively, preserved normal transport activity. Whereas cysteine-less Tyt1 was resistant to the inhibitory effect of sulfhydryl-alkylating reagents, N-ethylmaleimide inhibited transport by Tyt1 variants containing either one or both of the endogenous cysteines, and this inhibition was altered by the substrates sodium and tyrosine, consistent with substrate-induced dynamics in the transport pathway. Our findings support a binding model of Tyt1 function in which an ordered sequence of substrate-induced structural changes reflects distinct conformational states of the transporter. This work identifies Tyt1 as the first functional bacterial NSS member putatively consisting of only 11 TMs and shows that Tyt1 is a suitable model for the study of NSS dynamics with relevance to structure/function relationships of human NSSs, including the dopamine, norepinephrine, serotonin, and gamma-aminobutyric acid transporters.

A pincer-like configuration of TM2 in the human dopamine transporter is responsible for indirect effects on cocaine binding

The second transmembrane segment (TM2) of DAT and other neurotransmitter transporters has been proposed to play a role in oligomerization as well as in cocaine binding. In an attempt to determine whether TM2 contributes to the binding site and/or transport pathway of DAT, we mutated to cysteine, one at a time, 25 residues in TM2 - from Phe98 to Gln122 - in an appropriate DAT background construct. Four of the mutants, F98C, G110C, P112C, and E117C, did not express at the cell surface, and G121C was inactive, despite its presence on the cell surface. Of the 21 mutants that expressed, none of the substituted cysteines reacted with MTSEA biotin-CAP, and none of the 20 functional mutants was sensitive to MTSEA or MTSET. Thus, TM2 does not appear to be water-accessible, based both on the lack of functional effects of charged MTS derivatives, and on the biochemical determination of lack of reaction with a biotinylated MTS derivative. This leads to the conclusion that TM2 does not contribute directly to the substrate-binding site or the transport pathway, and suggests that the observed effect of mutations in this region on cocaine binding is indirect. Three mutants, M106C, V107C and I108C, were crosslinked by treatment with HgCl(2). This crosslinking was inhibited by the presence of the cocaine analogue MFZ 2-12, likely due to a conformational rearrangement in TM2 upon inhibitor binding. However, the lack of crosslinking of cysteines substituted for Leu99, Leu113 and Leu120 - three of the residues that along with Met106 form a leucine heptad repeat in TM2 - makes it unlikely that this leucine repeat plays a role in symmetrical TM2 dimerization. Importantly, a high-resolution structure of LeuT, a sodium-dependent leucine transporter that is sufficiently homologous to DAT to suggest a high degree of structural similarity, became available while this manuscript was under review. We have taken advantage of this structure to explore further and interpret our experimental results in a rigorous structural context.

Modeling membrane proteins based on low-resolution electron microscopy maps: a template for the TM domains of the oxalate transporter OxlT

The availability of both EM and high-resolution crystallographic data for several membrane proteins (MPs) permits a detailed evaluation of the ability of molecular modeling techniques to complement EM data in the development of models of MPs. A protocol for this purpose is presented, consisting of (1) identifying transmembrane (TM) domains from sequence; (2) assigning buried and lipid-exposed faces of the TM domains; and (3) assembling the TM domains into a bundle, based on geometric restraints obtained from the EM data. The protocol is validated by predicting the structures of several 7- and 12-TM MPs to within 3-5 A r.m.s.d. from their crystal structures. The protocol is applied to generate a model of the oxalate transporter OxlT, for which a high-resolution structure is not yet available.

Molecular determinants for the complex binding specificity of the PDZ domain in PICK1

PICK1 (protein interacting with C kinase 1) contains a single PDZ domain known to mediate interaction with the C termini of several receptors, transporters, ion channels, and kinases. In contrast to most PDZ domains, the PICK1 PDZ domain interacts with binding sequences classifiable as type I (terminating in (S/T)XPhi; X, any residue) as well as type II (PhiXPhi; Phi, any hydrophobic residue). To enable direct assessment of the affinity of the PICK1 PDZ domain for its binding partners we developed a purification scheme for PICK1 and a novel quantitative binding assay based on fluorescence polarization. Our results showed that the PICK1 PDZ domain binds the type II sequence presented by the human dopamine transporter (-WLKV) with an almost 15-fold and >100-fold higher affinity than the type I sequences presented by protein kinase Calpha (-QSAV) and the beta(2)-adrenergic receptor (-DSLL), respectively. Mutational analysis of Lys(83) in the alphaB1 position of the PDZ domain suggested that this residue mimics the function of hydrophobic residues present in this position in regular type II PDZ domains. The PICK1 PDZ domain was moreover found to prefer small hydrophobic residues in the C-terminal P(0) position of the ligand. Molecular modeling predicted a rank order of (Val > Ile > Leu) that was verified experimentally with up to a approximately 16-fold difference in binding affinity between a valine and a leucine in P(0). The results define the structural basis for the unusual binding pattern of the PICK1 PDZ domain by substantiating the critical role of the alphaB1 position (Lys(83)) and of discrete side chain differences in position P(0) of the ligands.

PDZBase: a protein-protein interaction database for PDZ-domains

SUMMARY

PDZBase is a database that aims to contain all known PDZ-domain-mediated protein-protein interactions. Currently, PDZBase contains approximately 300 such interactions, which have been manually extracted from > 200 articles. The database can be queried through both sequence motif and keyword-based searches, and the sequences of interacting proteins can be visually inspected through alignments (for the comparison of several interactions), or as residue-based diagrams including schematic secondary structure information (for individual complexes).

A knowledge-based scale for the analysis and prediction of buried and exposed faces of transmembrane domain proteins

MOTIVATION

The dearth of structural data on alpha-helical membrane proteins (MPs) has hampered thus far the development of reliable knowledge-based potentials that can be used for automatic prediction of transmembrane (TM) protein structure. While algorithms for identifying TM segments are available, modeling of the TM domains of alpha-helical MPs involves assembling the segments into a bundle. This requires the correct assignment of the buried and lipid-exposed faces of the TM domains.

RESULTS

A recent increase in the number of crystal structures of alpha-helical MPs has enabled an analysis of the lipid-exposed surfaces and the interiors of such molecules on the basis of structure, rather than sequence alone. Together with a conservation criterion that is based on previous observations that conserved residues are mostly found in the interior of MPs, the bias of certain residue types to be preferably buried or exposed is proposed as a criterion for predicting the lipid-exposed and interior faces of TMs. Applications to known structures demonstrates 80% accuracy of this prediction algorithm.

AVAILABILITY

The algorithm used for the predictions is implemented in the ProperTM Web server (http://icb.med.cornell.edu/services/propertm/start).

8R-Lisuride Is a Potent Stereospecific Histamine H1-Receptor Partial Agonist

The human histamine H1 receptor (H1R) is an important, well characterized target for the development of antagonists to treat allergic conditions. Many neuropsychiatric drugs are known to potently antagonize the H1R, thereby producing some of their side effects. In contrast, the tolerability and potential therapeutic utility of H1R agonism is currently unclear. We have used a cell-based functional assay to evaluate known therapeutics and reference drugs for H1R agonist activity. Our initial functional screen identified three ergot-based compounds possessing heretofore-unknown H1R agonist activity. 8R-lisuride demonstrated potent agonist activity in various assays including receptor selection and amplification technology, inositol phosphate accumulation, and activation of nuclear factor-kappaB with pEC50 values of 8.1, 7.9, and 7.9, respectively, and with varying degrees of efficacy. Based on these assays, 8R-lisuride is the most potent stereospecific partial agonist for the human H1R yet reported. Investigation of the residues involved in histamine and lisuride binding, using H1R mutants and molecular modeling, have revealed that although these ligands are structurally different, the lisuride-binding pocket in the H1R closely corresponds to the histamine-binding pocket. The discovery of a potent stereospecific partial H1R agonist provides a valuable tool to further characterize this important therapeutic target in vitro.

Probing conformational changes in neurotransmitter transporters: a structural context

The Na+/Cl-dependent neurotransmitter transporters, a family of proteins responsible for the reuptake of neurotransmitters and other small molecules from the synaptic cleft, have been the focus of intensive research in recent years. The biogenic amine transporters, a subset of this larger family, are especially intriguing as they are the targets for many psychoactive compounds, including cocaine and amphetamines, as well as many antidepressants. In the absence of a high-resolution structure for any transporter in this family, research into the structure-function relationships of these transporters has relied on analysis of the effects of site-directed mutagenesis as well as of chemical modification of reactive residues. The aim of this review is to establish a structural context for the experimental study of these transporters through various computational approaches and to highlight what is known about the conformational changes associated with function in these transporters. We also present a novel numbering scheme to assist in the comparison of aligned positions between sequences of the neurotransmitter transporter family, a comparison that will be of increasing importance as additional experimental data is amassed.