Discovery of Clinical Candidate GLPG3970: A Potent and Selective Dual SIK2/SIK3 Inhibitor for the Treatment of Autoimmune and Inflammatory Diseases.

The salt-inducible kinases (SIKs) SIK1, SIK2, and SIK3 belong to the adenosine monophosphate-activated protein kinase (AMPK) family of serine/threonine kinases. SIK inhibition represents a new therapeutic approach modulating pro-inflammatory and immunoregulatory pathways that holds potential for the treatment of inflammatory diseases. Here, we describe the identification of GLPG3970 (32), a first-in-class dual SIK2/SIK3 inhibitor with selectivity against SIK1 (IC50 of 282.8 nM on SIK1, 7.8 nM on SIK2 and 3.8 nM on SIK3). We outline efforts made to increase selectivity against SIK1 and improve CYP time-dependent inhibition properties through the structure-activity relationship. The dual activity of 32 in modulating the pro-inflammatory cytokine TNFα and the immunoregulatory cytokine IL-10 is demonstrated in vitro in human primary myeloid cells and human whole blood, and in vivo in mice stimulated with lipopolysaccharide. Compound 32 shows dose-dependent activity in disease-relevant mouse pharmacological models.

Optimization of Selectivity and Pharmacokinetic Properties of Salt-Inducible Kinase Inhibitors that Led to the Discovery of Pan-SIK Inhibitor GLPG3312.

Salt-inducible kinases (SIKs) SIK1, SIK2, and SIK3 are serine/threonine kinases and form a subfamily of the protein kinase AMP-activated protein kinase (AMPK) family. Inhibition of SIKs in stimulated innate immune cells and mouse models has been associated with a dual mechanism of action consisting of a reduction of pro-inflammatory cytokines and an increase of immunoregulatory cytokine production, suggesting a therapeutic potential for inflammatory diseases. Following a high-throughput screening campaign, subsequent hit to lead optimization through synthesis, structure-activity relationship, kinome selectivity, and pharmacokinetic investigations led to the discovery of clinical candidate GLPG3312 (compound 28), a potent and selective pan-SIK inhibitor (IC50: 2.0 nM for SIK1, 0.7 nM for SIK2, and 0.6 nM for SIK3). Characterization of the first human SIK3 crystal structure provided an understanding of the binding mode and kinome selectivity of the chemical series. GLPG3312 demonstrated both anti-inflammatory and immunoregulatory activities in vitro in human primary myeloid cells and in vivo in mouse models.

Crystal structure of adenosine A2A receptor in complex with clinical candidate Etrumadenant reveals unprecedented antagonist interaction.

The Gs protein-coupled adenosine A2A receptor (A2AAR) represents an emerging drug target for cancer immunotherapy. The clinical candidate Etrumadenant was developed as an A2AAR antagonist with ancillary blockade of the A2BAR subtype. It constitutes a unique chemotype featuring a poly-substituted 2-amino-4-phenyl-6-triazolylpyrimidine core structure. Herein, we report two crystal structures of the A2AAR in complex with Etrumadenant, obtained with differently thermostabilized A2AAR constructs. This led to the discovery of an unprecedented interaction, a hydrogen bond of T883.36 with the cyano group of Etrumadenant. T883.36 is mutated in most A2AAR constructs used for crystallization, which has prevented the discovery of its interactions. In-vitro characterization of Etrumadenant indicated low selectivity versus the A1AR subtype, which can be rationalized by the structural data. These results will facilitate the future design of AR antagonists with desired selectivity. Moreover, they highlight the advantages of the employed A2AAR crystallization construct that is devoid of ligand binding site mutations.

Cryo-EM structures of a LptDE transporter in complex with Pro-macrobodies offer insight into lipopolysaccharide translocation.

Lipopolysaccharides are major constituents of the extracellular leaflet in the bacterial outer membrane and form an effective physical barrier for environmental threats and for antibiotics in Gram-negative bacteria. The last step of LPS insertion via the Lpt pathway is mediated by the LptD/E protein complex. Detailed insights into the architecture of LptDE transporter complexes have been derived from X-ray crystallography. However, no structure of a laterally open LptD transporter, a transient state that occurs during LPS release, is available to date. Here, we report a cryo-EM structure of a partially opened LptDE transporter in complex with rigid chaperones derived from nanobodies, at 3.4 Å resolution. In addition, a subset of particles allows to model a structure of a laterally fully opened LptDE complex. Our work offers insights into the mechanism of LPS insertion, provides a structural framework for the development of antibiotics targeting LptD and describes a highly rigid chaperone scaffold to enable structural biology of challenging protein targets.

Apo and ligand-bound high resolution Cryo-EM structures of the human Kv3.1 channel reveal a novel binding site for positive modulators.

Kv3 ion-channels constitute a class of functionally distinct voltage-gated ion channels characterized by their ability to fire at a high frequency. Several disease relevant mutants, together with biological data, suggest the importance of this class of ion channels as drug targets for CNS disorders, and several drug discovery efforts have been reported. Despite the increasing interest for this class of ion channels, no structure of a Kv3 channel has been reported yet. We have determined the cryo-EM structure of Kv3.1 at 2.6 Å resolution using full-length wild type protein. When compared to known structures for potassium channels from other classes, a novel domain organization is observed with the cytoplasmic T1 domain, containing a well-resolved Zinc site and displaying a rotation by 35°. This suggests a distinct cytoplasmic regulation mechanism for the Kv3.1 channel. A high resolution structure was obtained for Kv3.1 in complex with a novel positive modulator Lu AG00563. The structure reveals a novel ligand binding site for the Kv class of ion channels located between the voltage sensory domain and the channel pore, a region which constitutes a hotspot for disease causing mutations. The discovery of a novel binding site for a positive modulator of a voltage-gated potassium channel could shed light on the mechanism of action for these small molecule potentiators. This finding could enable structure-based drug design on these targets with high therapeutic potential for the treatment of multiple CNS disorders.

Discovery and Optimization of Orally Bioavailable Phthalazone and Cinnolone Carboxylic Acid Derivatives as S1P2 Antagonists against Fibrotic Diseases.

Idiopathic pulmonary fibrosis (IPF) is a chronic and progressive lung disease. Current treatments only slow down disease progression, making new therapeutic strategies compelling. Increasing evidence suggests that S1P2 antagonists could be effective agents against fibrotic diseases. Our compound collection was mined for molecules possessing substructure features associated with S1P2 activity. The weakly potent indole hit 6 evolved into a potent phthalazone series, bearing a carboxylic acid, with the aid of a homology model. Suboptimal pharmacokinetics of a benzimidazole subseries were improved by modifications targeting potential interactions with transporters, based on concepts deriving from the extended clearance classification system (ECCS). Scaffold hopping, as a part of a chemical enablement strategy, permitted the rapid exploration of the position adjacent to the carboxylic acid. Compound 38, with good pharmacokinetics and in vitro potency, was efficacious at 10 mg/kg BID in three different in vivo mouse models of fibrotic diseases in a therapeutic setting.

Mechanistic Implications of Reductive Co-C Bond Cleavage in B12-Dependent Methylmalonyl CoA Mutase.

Vitamin B12-dependent enzymes catalyze several difficult radical reactions. There are fundamental open questions that need to be addressed to fully understand the formation of highly reactive radical species, its dynamics, and interaction with the substrate and enzyme. In this work, ab initio molecular dynamics was performed within a QM/MM framework on a reduced AdoCbl cofactor, which was taken as a post proton-coupled electron transfer initial step for the activation of the AdoCbl-dependent methylmalonyl CoA mutase enzyme. The calculated free-energy profile reveals two possible pathways, stepwise (I) and concerted (II) for the reductive Co-C cleavage and subsequent H-abstraction. The computed activation barrier from metadynamics for both the pathways is comparable (78.5 and 76.2 kJ/mol, respectively); however, the concerted pathway may be preferred kinetically because it avoids the formation of a high-energy radical intermediate with possibly a larger recrossing rate. Our results are consistent with the previous conductor hypothesis, indicating the explicit role of cob(II)alamin in stabilizing the radical intermediate involved in the H-atom transfer.

Shedding Light on Important Waters for Drug Design: Simulations versus Grid-Based Methods.

Water molecules play an important role in the association of drugs with their pharmaceutical targets. For this reason, calculating the energetic contribution of water is essential to make accurate predictions of compounds' affinity and selectivity. Water molecules can also modify the binding mode of compounds by forming water bridges, or clusters, that stabilize a particular orientation of the ligand. Several computational methods have been developed for solvent mapping, but few studies have attempted to compare them in a drug design context. In this paper, four commercially available solvent mapping tools (SZMAP, WaterFLAP, 3D-RISM, and WaterMap) are evaluated on three different protein targets. The methods were compared by looking at their ability to predict the structure-activity relations of lead compounds. All methods were found to be useful to some degree and to improve the predictions from docking alone. However, the only simulation-based approach tested, WaterMap, was found in some cases to be more accurate than grid-based methods.

Membranes serve as allosteric activators of phospholipase A2, enabling it to extract, bind, and hydrolyze phospholipid substrates.

Defining the molecular details and consequences of the association of water-soluble proteins with membranes is fundamental to understanding protein-lipid interactions and membrane functioning. Phospholipase A2 (PLA2) enzymes, which catalyze the hydrolysis of phospholipid substrates that compose the membrane bilayers, provide the ideal system for studying protein-lipid interactions. Our study focuses on understanding the catalytic cycle of two different human PLA2s: the cytosolic Group IVA cPLA2 and calcium-independent Group VIA iPLA2. Computer-aided techniques guided by deuterium exchange mass spectrometry data, were used to create structural complexes of each enzyme with a single phospholipid substrate molecule, whereas the substrate extraction process was studied using steered molecular dynamics simulations. Molecular dynamic simulations of the enzyme-substrate-membrane systems revealed important information about the mechanisms by which these enzymes associate with the membrane and then extract and bind their phospholipid substrate. Our data support the hypothesis that the membrane acts as an allosteric ligand that binds at the allosteric site of the enzyme's interfacial surface, shifting its conformation from a closed (inactive) state in water to an open (active) state at the membrane interface.

Improved Reweighting of Accelerated Molecular Dynamics Simulations for Free Energy Calculation.

Accelerated molecular dynamics (aMD) simulations greatly improve the efficiency of conventional molecular dynamics (cMD) for sampling biomolecular conformations, but they require proper reweighting for free energy calculation. In this work, we systematically compare the accuracy of different reweighting algorithms including the exponential average, Maclaurin series, and cumulant expansion on three model systems: alanine dipeptide, chignolin, and Trp-cage. Exponential average reweighting can recover the original free energy profiles easily only when the distribution of the boost potential is narrow (e.g., the range ≤20kBT) as found in dihedral-boost aMD simulation of alanine dipeptide. In dual-boost aMD simulations of the studied systems, exponential average generally leads to high energetic fluctuations, largely due to the fact that the Boltzmann reweighting factors are dominated by a very few high boost potential frames. In comparison, reweighting based on Maclaurin series expansion (equivalent to cumulant expansion on the first order) greatly suppresses the energetic noise but often gives incorrect energy minimum positions and significant errors at the energy barriers (∼2-3kBT). Finally, reweighting using cumulant expansion to the second order is able to recover the most accurate free energy profiles within statistical errors of ∼kBT, particularly when the distribution of the boost potential exhibits low anharmonicity (i.e., near-Gaussian distribution), and should be of wide applicability. A toolkit of Python scripts for aMD reweighting "PyReweighting" is distributed free of charge at http://mccammon.ucsd.edu/computing/amdReweighting/.

Dipeptide Aggregation in Aqueous Solution from Fixed Point-Charge Force Fields.

The description of aggregation processes with molecular dynamics simulations is a playground for testing biomolecular force fields, including a new generation of force fields that explicitly describe electronic polarization. In this work, we study a system consisting of 50 glycyl-l-alanine (Gly-Ala) dipeptides in solution with 1001 water molecules. Neutron diffraction experiments have shown that at this concentration, Gly-Ala aggregates into large clusters. However, general-purpose force fields in combination with established water models can fail to correctly describe this aggregation process, highlighting important deficiencies in how solute-solute and solute-solvent interactions are parametrized in these force fields. We found that even for the fully polarizable AMOEBA force field, the degree of association is considerably underestimated. Instead, a fixed point-charge approach based on the newly developed IPolQ scheme [Cerutti et al. J. Phys. Chem.2013, 117, 2328] allows for the correct modeling of the dipeptide aggregation in aqueous solution. This result should stimulate interest in novel fitting schemes that aim to improve the description of the solvent polarization effect within both explicitly polarizable and fixed point-charge frameworks.

Insertion of the Ca²âº-independent phospholipase A2 into a phospholipid bilayer via coarse-grained and atomistic molecular dynamics simulations.

Group VI Ca²âº-independent phospholipase A2 (iPLA2) is a water-soluble enzyme that is active when associated with phospholipid membranes. Despite its clear pharmaceutical relevance, no X-ray or NMR structural information is currently available for the iPLA2 or its membrane complex. In this paper, we combine homology modeling with coarse-grained (CG) and all-atom (AA) molecular dynamics (MD) simulations to build structural models of iPLA2 in association with a phospholipid bilayer. CG-MD simulations of the membrane insertion process were employed to provide a starting point for an atomistic description. Six AA-MD simulations were then conducted for 60 ns, starting from different initial CG structures, to refine the membrane complex. The resulting structures are shown to be consistent with each other and with deuterium exchange mass spectrometry (DXMS) experiments, suggesting that our approach is suitable for the modeling of iPLA2 at the membrane surface. The models show that an anchoring region (residues 710-724) forms an amphipathic helix that is stabilized by the membrane. In future studies, the proposed iPLA2 models should provide a structural basis for understanding the mechanisms of lipid extraction and drug-inhibition. In addition, the dual-resolution approach discussed here should provide the means for the future exploration of the impact of lipid diversity and sequence mutations on the activity of iPLA2 and related enzymes.

Fluoroketone inhibition of Ca(2+)-independent phospholipase A2 through binding pocket association defined by hydrogen/deuterium exchange and molecular dynamics.

The mechanism of inhibition of group VIA Ca(2+)-independent phospholipase A(2) (iPLA(2)) by fluoroketone (FK) ligands is examined by a combination of deuterium exchange mass spectrometry (DXMS) and molecular dynamics (MD). Models for iPLA(2) were built by homology with the known structure of patatin and equilibrated by extensive MD simulations. Empty pockets were identified during the simulations and studied for their ability to accommodate FK inhibitors. Ligand docking techniques showed that the potent inhibitor 1,1,1,3-tetrafluoro-7-phenylheptan-2-one (PHFK) forms favorable interactions inside an active-site pocket, where it blocks the entrance of phospholipid substrates. The polar fluoroketone headgroup is stabilized by hydrogen bonds with residues Gly486, Gly487, and Ser519. The nonpolar aliphatic chain and aromatic group are stabilized by hydrophobic contacts with Met544, Val548, Phe549, Leu560, and Ala640. The binding mode is supported by DXMS experiments showing an important decrease of deuteration in the contact regions in the presence of the inhibitor. The discovery of the precise binding mode of FK ligands to the iPLA(2) should greatly improve our ability to design new inhibitors with higher potency and selectivity.

Accelerated Molecular Dynamics Simulations with the AMOEBA Polarizable Force Field on Graphics Processing Units.

The accelerated molecular dynamics (aMD) method has recently been shown to enhance the sampling of biomolecules in molecular dynamics (MD) simulations, often by several orders of magnitude. Here, we describe an implementation of the aMD method for the OpenMM application layer that takes full advantage of graphics processing units (GPUs) computing. The aMD method is shown to work in combination with the AMOEBA polarizable force field (AMOEBA-aMD), allowing the simulation of long time-scale events with a polarizable force field. Benchmarks are provided to show that the AMOEBA-aMD method is efficiently implemented and produces accurate results in its standard parametrization. For the BPTI protein, we demonstrate that the protein structure described with AMOEBA remains stable even on the extended time scales accessed at high levels of accelerations. For the DNA repair metalloenzyme endonuclease IV, we show that the use of the AMOEBA force field is a significant improvement over fixed charged models for describing the enzyme active-site. The new AMOEBA-aMD method is publicly available (http://wiki.simtk.org/openmm/VirtualRepository) and promises to be interesting for studying complex systems that can benefit from both the use of a polarizable force field and enhanced sampling.

Exploring the Photophysical Properties of Molecular Systems Using Excited State Accelerated ab Initio Molecular Dynamics.

In the present work, we employ excited state accelerated ab initio molecular dynamics (A-AIMD) to efficiently study the excited state energy landscape and photophysical topology of a variety of molecular systems. In particular, we focus on two important challenges for the modeling of excited electronic states: (i) the identification and characterization of conical intersections and crossing seams, in order to predict different and often competing radiationless decay mechanisms, and (ii) the description of the solvent effect on the absorption and emission spectra of chemical species in solution. In particular, using as examples the Schiff bases formaldimine and salicylidenaniline, we show that A-AIMD can be readily employed to explore the conformational space around crossing seams in molecular systems with very different photochemistry. Using acetone in water as an example, we demonstrate that the enhanced configurational space sampling may be used to accurately and efficiently describe both the prominent features and line-shapes of absorption and emission spectra.

The elusive 5'-deoxyadenosyl radical in coenzyme-B12-mediated reactions.

Vitamin B(12) and its biologically active counterparts possess the only examples of carbon-cobalt bonds in living systems. The role of such motifs as radical reservoirs has potential application in future catalytic and electronic nanodevices. To fully understand radical generation in coenzyme B(12) (dAdoCbl)-dependent enzymes, however, major obstacles still need to be overcome. In this work, we have used Car-Parrinello molecular dynamics (CPMD) simulations, in a mixed quantum mechanics/molecular mechanics (QM/MM) framework, to investigate the initial stages of the methylmalonyl-CoA-mutase-catalyzed reaction. We demonstrate that the 5'-deoxyadenosyl radical (dAdo(•)) exists as a distinct entity in this reaction, consistent with the results of extensive experimental and some previous theoretical studies. We report free energy calculations and first-principles trajectories that help understand how B(12) enzymes catalyze coenzyme activation and control highly reactive radical intermediates.

Induced fit or conformational selection? The role of the semi-closed state in the maltose binding protein.

A full characterization of the thermodynamic forces underlying ligand-associated conformational changes in proteins is essential for understanding and manipulating diverse biological processes, including transport, signaling, and enzymatic activity. Recent experiments on the maltose binding protein (MBP) have provided valuable data about the different conformational states implicated in the ligand recognition process; however, a complete picture of the accessible pathways and the associated changes in free energy remains elusive. Here we describe results from advanced accelerated molecular dynamics (aMD) simulations, coupled with adaptively biased force (ABF) and thermodynamic integration (TI) free energy methods. The combination of approaches allows us to track the ligand recognition process on the microsecond time scale and provides a detailed characterization of the protein's dynamic and the relative energy of stable states. We find that an induced-fit (IF) mechanism is most likely and that a mechanism involving both a conformational selection (CS) step and an IF step is also possible. The complete recognition process is best viewed as a "Pac Man" type action where the ligand is initially localized to one domain and naturally occurring hinge-bending vibrations in the protein are able to assist the recognition process by increasing the chances of a favorable encounter with side chains on the other domain, leading to a population shift. This interpretation is consistent with experiments and provides new insight into the complex recognition mechanism. The methods employed here are able to describe IF and CS effects and provide formally rigorous means of computing free energy changes. As such, they are superior to conventional MD and flexible docking alone and hold great promise for future development and applications to drug discovery.

Accessing a hidden conformation of the maltose binding protein using accelerated molecular dynamics.

Periplasmic binding proteins (PBPs) are a large family of molecular transporters that play a key role in nutrient uptake and chemotaxis in Gram-negative bacteria. All PBPs have characteristic two-domain architecture with a central interdomain ligand-binding cleft. Upon binding to their respective ligands, PBPs undergo a large conformational change that effectively closes the binding cleft. This conformational change is traditionally viewed as a ligand induced-fit process; however, the intrinsic dynamics of the protein may also be crucial for ligand recognition. Recent NMR paramagnetic relaxation enhancement (PRE) experiments have shown that the maltose binding protein (MBP) - a prototypical member of the PBP superfamily - exists in a rapidly exchanging (ns to µs regime) mixture comprising an open state (approx 95%), and a minor partially closed state (approx 5%). Here we describe accelerated MD simulations that provide a detailed picture of the transition between the open and partially closed states, and confirm the existence of a dynamical equilibrium between these two states in apo MBP. We find that a flexible part of the protein called the balancing interface motif (residues 175-184) is displaced during the transformation. Continuum electrostatic calculations indicate that the repacking of non-polar residues near the hinge region plays an important role in driving the conformational change. Oscillations between open and partially closed states create variations in the shape and size of the binding site. The study provides a detailed description of the conformational space available to ligand-free MBP, and has implications for understanding ligand recognition and allostery in related proteins.

On the Use of Accelerated Molecular Dynamics to Enhance Configurational Sampling in Ab Initio Simulations.

We have implemented the accelerated molecular dynamics approach (Hamelberg, D.; Mongan, J.; McCammon, J. A. J. Chem. Phys. 2004, 120 (24), 11919) in the framework of ab initio MD (AIMD). Using three simple examples, we demonstrate that accelerated AIMD (A-AIMD) can be used to accelerate solvent relaxation in AIMD simulations and facilitate the detection of reaction coordinates: (i) We show, for one cyclohexane molecule in the gas phase, that the method can be used to accelerate the rate of the chair-to-chair interconversion by a factor of ∼1 × 10(5), while allowing for the reconstruction of the correct canonical distribution of low-energy states; (ii) We then show, for a water box of 64 H(2)O molecules, that A-AIMD can also be used in the condensed phase to accelerate the sampling of water conformations, without affecting the structural properties of the solvent; and (iii) The method is then used to compute the potential of mean force (PMF) for the dissociation of Na-Cl in water, accelerating the convergence by a factor of ∼3-4 compared to conventional AIMD simulations.(2) These results suggest that A-AIMD is a useful addition to existing methods for enhanced conformational and phase-space sampling in solution. While the method does not make the use of collective variables superfluous, it also does not require the user to define a set of collective variables that can capture all the low-energy minima on the potential energy surface. This property may prove very useful when dealing with highly complex multidimensional systems that require a quantum mechanical treatment.