Design of rigid protein-protein interaction inhibitors enables targeting of undruggable Mcl-1.

The structure-based design of small-molecule inhibitors targeting protein-protein interactions (PPIs) remains a huge challenge as the drug must bind typically wide and shallow protein sites. A PPI target of high interest for hematological cancer therapy is myeloid cell leukemia 1 (Mcl-1), a prosurvival guardian protein from the Bcl-2 family. Despite being previously considered undruggable, seven small-molecule Mcl-1 inhibitors have recently entered clinical trials. Here, we report the crystal structure of the clinical-stage inhibitor AMG-176 bound to Mcl-1 and analyze its interaction along with clinical inhibitors AZD5991 and S64315. Our X-ray data reveal high plasticity of Mcl-1 and a remarkable ligand-induced pocket deepening. Nuclear Magnetic Resonance (NMR)-based free ligand conformer analysis demonstrates that such unprecedented induced fit is uniquely achieved by designing highly rigid inhibitors, preorganized in their bioactive conformation. By elucidating key chemistry design principles, this work provides a roadmap for targeting the largely untapped PPI class more successfully.

Enhancing torsional sampling using fully adaptive simulated tempering.

Enhanced sampling algorithms are indispensable when working with highly disconnected multimodal distributions. An important application of these is the conformational exploration of particular internal degrees of freedom of molecular systems. However, despite the existence of many commonly used enhanced sampling algorithms to explore these internal motions, they often rely on system-dependent parameters, which negatively impact efficiency and reproducibility. Here, we present fully adaptive simulated tempering (FAST), a variation of the irreversible simulated tempering algorithm, which continuously optimizes the number, parameters, and weights of intermediate distributions to achieve maximally fast traversal over a space defined by the change in a predefined thermodynamic control variable such as temperature or an alchemical smoothing parameter. This work builds on a number of previously published methods, such as sequential Monte Carlo, and introduces a novel parameter optimization procedure that can, in principle, be used in any expanded ensemble algorithms. This method is validated by being applied on a number of different molecular systems with high torsional kinetic barriers. We also consider two different soft-core potentials during the interpolation procedure and compare their performance. We conclude that FAST is a highly efficient algorithm, which improves simulation reproducibility and can be successfully used in a variety of settings with the same initial hyperparameters.

Virtual Screening in the Cloud Identifies Potent and Selective ROS1 Kinase Inhibitors.

ROS1 rearrangements account for 1-2% of non-small cell lung cancer patients, yet there are no specifically designed, selective ROS1 therapies in the clinic. Previous knowledge of potent ROS1 inhibitors with selectivity over TrkA, a selected antitarget, enabled virtual screening as a hit finding approach in this project. The ligand-based virtual screening was focused on identifying molecules with a similar 3D shape and pharmacophore to the known actives. To that end, we turned to the AstraZeneca virtual library, estimated to cover 1015 synthesizable make-on-demand molecules. We used cloud computing-enabled FastROCS technology to search the enumerated 1010 subset of the full virtual space. A small number of specific libraries were prioritized based on the compound properties and a medicinal chemistry assessment and further enumerated with available building blocks. Following the docking evaluation to the ROS1 structure, the most promising hits were synthesized and tested, resulting in the identification of several potent and selective series. The best among them gave a nanomolar ROS1 inhibitor with over 1000-fold selectivity over TrkA and, from the preliminary established SAR, these have the potential to be further optimized. Our prospective study describes how conceptually simple shape-matching approaches can identify potent and selective compounds by searching ultralarge virtual libraries, demonstrating the applicability of such workflows and their importance in early drug discovery.

Alkynyl Benzoxazines and Dihydroquinazolines as Cysteine Targeting Covalent Warheads and Their Application in Identification of Selective Irreversible Kinase Inhibitors.

With a resurgence in interest in covalent drugs, there is a need to identify new moieties capable of cysteine bond formation that are differentiated from commonly employed systems such as acrylamide. Herein, we report on the discovery of new alkynyl benzoxazine and dihydroquinazoline moieties capable of covalent reaction with cysteine. Their utility as alternative electrophilic warheads for chemical biological probes and drug molecules is demonstrated through site-selective protein modification and incorporation into kinase drug scaffolds. A potent covalent inhibitor of JAK3 kinase was identified with superior selectivity across the kinome and improvements in in vitro pharmacokinetic profile relative to the related acrylamide-based inhibitor. In addition, the use of a novel heterocycle as a cysteine reactive warhead is employed to target Cys788 in c-KIT, where acrylamide has previously failed to form covalent interactions. These new reactive and selective heterocyclic warheads supplement the current repertoire for cysteine covalent modification while avoiding some of the limitations generally associated with established moieties.

ProtoCaller: Robust Automation of Binding Free Energy Calculations.

ProtoCaller is a Python library distributed through Anaconda which automates relative protein-ligand binding free energy calculations in GROMACS. It links a number of popular specialized tools used to perform protein setup and parametrization, such as PDB2PQR, Modeller, and AmberTools. ProtoCaller supports commonly used AMBER force fields with additional cofactor parameters, and AM1-BCC is used to derive ligand charges. ProtoCaller also comes with an extensive PDB parser, an enhanced maximum common substructure algorithm providing improved ligand-ligand mapping, and a light GROMACS wrapper for running multiple molecular dynamics simulations. ProtoCaller is highly relevant to most researchers in the field of biomolecular simulation, allowing a customizable balance between automation and user intervention.

Heteroarylamide smoothened inhibitors: Discovery of N-[2,4-dimethyl-5-(1-methylimidazol-4-yl)phenyl]-4-(2-pyridylmethoxy)benzamide (AZD8542) and N-[5-(1H-imidazol-2-yl)-2,4-dimethyl-phenyl]-4-(2- pyridylmethoxy)benzamide (AZD7254).

Aberrant hedgehog (Hh) pathway signaling is implicated in multiple cancer types and targeting the Smoothened (SMO) receptor, a key protein of the Hh pathway, has proven effective in treating metastasized basal cell carcinoma. Our lead optimization effort focused on a series of heteroarylamides. We observed that a methyl substitution ortho to the heteroaryl groups on an aniline core significantly improved the potency of this series of compounds. These findings predated the availability of SMO crystal structure in 2013. Here we retrospectively applied quantum mechanics calculations to demonstrate the o-Me substitution favors the bioactive conformation by inducing a dihedral twist between the heteroaryl rings and the core aniline. The o-Me also makes favorable hydrophobic interactions with key residue side chains in the binding pocket. From this effort, two compounds (AZD8542 and AZD7254) showed excellent pharmacokinetics across multiple preclinical species and demonstrated in vivo activity in abrogating the Hh paracrine pathway as well as anti- tumor effects.

Water Sites, Networks, And Free Energies with Grand Canonical Monte Carlo.

Water molecules play integral roles in the formation of many protein-ligand complexes, and recent computational efforts have been focused on predicting the thermodynamic properties of individual waters and how they may be exploited in rational drug design. However, when water molecules form highly coupled hydrogen-bonding networks, there is, as yet, no method that can rigorously calculate the free energy to bind the entire network or assess the degree of cooperativity between waters. In this work, we report theoretical and methodological developments to the grand canonical Monte Carlo simulation technique. Central to our results is a rigorous equation that can be used to calculate efficiently the binding free energies of water networks of arbitrary size and complexity. Using a single set of simulations, our methods can locate waters, estimate their binding affinities, capture the cooperativity of the water network, and evaluate the hydration free energy of entire protein binding sites. Our techniques have been applied to multiple test systems and compare favorably to thermodynamic integration simulations and experimental data. The implications of these methods in drug design are discussed.

A molecular dynamics study of CaCO3 nanoparticles in a hydrophobic solvent with a stearate co-surfactant.

Stearates containing overbased detergent nanoparticles (NPs) are used as acid neutralising additives in automotive and marine engine oils. Molecular dynamics (MD) simulations of the self-assembly of calcium carbonate, calcium stearate as a co-surfactant and stabilising surfactants of such NPs in a model explicit molecular hydrophobic solvent have been carried out using a methodology described first by Bodnarchuk et al. [J. Phys. Chem. C, 2014, 118, 21092]. The cores and particles as a whole become more elongated with stearate, and the surfactant molecules are more spaced out in this geometry than in their stearate-free counterparts. The rod dimensions are found to be largely independent of the surfactant type for a given amount of CaCO3. The corresponding particles without stearate were more spherical, the precise shape depending to a greater extent on the chemical architecture of the surfactant molecule. The rod-shaped stearate containing nanoparticles penetrated a model water droplet to a greater depth than the corresponding near-spherical particle, which is possibly facilitated by the dissociation of nanoparticle surfactant molecules onto the surface of the water in this process. These simulations are the first to corroborate the nanoparticle-water penetration mechanism proposed previously by experimental groups investigating the NP acid neutralisation characteristics.

Strategies to calculate water binding free energies in protein-ligand complexes.

Water molecules are commonplace in protein binding pockets, where they can typically form a complex between the protein and a ligand or become displaced upon ligand binding. As a result, it is often of great interest to establish both the binding free energy and location of such molecules. Several approaches to predicting the location and affinity of water molecules to proteins have been proposed and utilized in the literature, although it is often unclear which method should be used under what circumstances. We report here a comparison between three such methodologies, Just Add Water Molecules (JAWS), Grand Canonical Monte Carlo (GCMC), and double-decoupling, in the hope of understanding the advantages and limitations of each method when applied to enclosed binding sites. As a result, we have adapted the JAWS scoring procedure, allowing the binding free energies of strongly bound water molecules to be calculated to a high degree of accuracy, requiring significantly less computational effort than more rigorous approaches. The combination of JAWS and GCMC offers a route to a rapid scheme capable of both locating and scoring water molecules for rational drug design.

Water network perturbation in ligand binding: adenosine A(2A) antagonists as a case study.

Recent efforts in the computational evaluation of the thermodynamic properties of water molecules have resulted in the development of promising new in silico methods to evaluate the role of water in ligand binding. These methods include WaterMap, SZMAP, GRID/CRY probe, and Grand Canonical Monte Carlo simulations. They allow the prediction of the position and relative free energy of the water molecule in the protein active site and the analysis of the perturbation of an explicit water network (WNP) as a consequence of ligand binding. We have for the first time extended these approaches toward the prediction of kinetics for small molecules and of relative free energy of binding with a focus on the perturbation of the water network and application to large diverse data sets. Our results support a qualitative correlation between the residence time of 12 related triazine adenosine A(2A) receptor antagonists and the number and position of high energy trapped solvent molecules. From a quantitative viewpoint, we successfully applied these computational techniques as an implicit solvent alternative, in linear combination with a molecular mechanics force field, to predict the relative ligand free energy of binding (WNP-MMSA). The applicability of this linear method, based on the thermodynamics additivity principle, did not extend to 375 diverse A(2A) receptor antagonists. However, a fast but effective method could be enabled by replacing the linear approach with a machine learning technique using probabilistic classification trees, which classified the binding affinity correctly for 90% of the ligands in the training set and 67% in the test set.

Discovery of AZD4747, a Potent and Selective Inhibitor of Mutant GTPase KRASG12C with Demonstrable CNS Penetration.

The glycine to cysteine mutation at codon 12 of Kirsten rat sarcoma (KRAS) represents an Achilles heel that has now rendered this important GTPase druggable. Herein, we report our structure-based drug design approach that led to the identification of 14, AZD4747, a clinical development candidate for the treatment of KRASG12C-positive tumors, including the treatment of central nervous system (CNS) metastases. Building on our earlier discovery of C5-tethered quinazoline AZD4625, excision of a usually critical pyrimidine ring yielded a weak but brain-penetrant start point which was optimized for potency and DMPK. Key design principles and measured parameters that give high confidence in CNS exposure are discussed. During optimization, divergence between rodent and non-rodent species was observed in CNS exposure, with primate PET studies ultimately giving high confidence in the expected translation to patients. AZD4747 is a highly potent and selective inhibitor of KRASG12C with an anticipated low clearance and high oral bioavailability profile in humans.

Enhancing Ligand and Protein Sampling Using Sequential Monte Carlo.

The sampling problem is one of the most widely studied topics in computational chemistry. While various methods exist for sampling along a set of reaction coordinates, many require system-dependent hyperparameters to achieve maximum efficiency. In this work, we present an alchemical variation of adaptive sequential Monte Carlo (SMC), an irreversible importance resampling method that is part of a well-studied class of methods that have been used in various applications but have been underexplored in computational biophysics. Afterward, we apply alchemical SMC on a variety of test cases, including torsional rotations of solvated ligands (butene and a terphenyl derivative), translational and rotational movements of protein-bound ligands, and protein side chain rotation coupled to the ligand degrees of freedom (T4-lysozyme, protein tyrosine phosphatase 1B, and transforming growth factor ß). We find that alchemical SMC is an efficient way to explore targeted degrees of freedom and can be applied to a variety of systems using the same hyperparameters to achieve a similar performance. Alchemical SMC is a promising tool for preparatory exploration of systems where long-timescale sampling of the entire system can be traded off against short-timescale sampling of a particular set of degrees of freedom over a population of conformers.

Discovery of AZD4625, a Covalent Allosteric Inhibitor of the Mutant GTPase KRASG12C.

KRAS is an archetypal high-value intractable oncology drug target. The glycine to cysteine mutation at codon 12 represents an Achilles heel that has now rendered this important GTPase druggable. Herein, we report our structure-based drug design approach that led to the identification of 21, AZD4625, a clinical development candidate for the treatment of KRASG12C positive tumors. Highlights include a quinazoline tethering strategy to lock out a bio-relevant binding conformation and an optimization strategy focused on the reduction of extrahepatic clearance mechanisms seen in preclinical species. Crystallographic analysis was also key in helping to rationalize unusual structure-activity relationship in terms of ring size and enantio-preference. AZD4625 is a highly potent and selective inhibitor of KRASG12C with an anticipated low clearance and high oral bioavailability profile in humans.

Drugging the undruggable: a computational chemist's view of KRASG12C.

In recent years, the emergence of targeted covalent inhibitors which bind to the G12C mutant of KRAS have offered a solution to this previously intractable target. Inhibitors of KRASG12C tend to be structurally complex, displaying features such as atropisomerism, chiral centres and a reactive covalent warhead. Such molecules result in lengthy and challenging syntheses, and as a consequence critical decisions need to be made at the design level to maximise the chances of success. Here we take a retrospective look into how computational chemistry can help guide and prioritise medicinal chemistry efforts in the context of a series of conformationally restricted tetracyclic quinolines.

Sensitivity of Binding Free Energy Calculations to Initial Protein Crystal Structure.

Binding free energy calculations using alchemical free energy (AFE) methods are widely considered to be the most rigorous tool in the computational drug discovery arsenal. Despite this, the calculations suffer from accuracy, precision, and reproducibility issues. In this publication, we perform a high-throughput study of more than a thousand AFE calculations, utilizing over 220 µs of total sampling time, on three different protein systems to investigate the impact of the initial crystal structure on the resulting binding free energy values. We also consider the influence of equilibration time and discover that the initial crystal structure can have a significant effect on free energy values obtained at short timescales that can manifest itself as a free energy difference of more than 1 kcal/mol. At longer timescales, these differences are largely overtaken by important rare events, such as torsional ligand motions, typically resulting in a much higher uncertainty in the obtained values. This work emphasizes the importance of rare event sampling and long-timescale dynamics in free energy calculations even for routinely performed alchemical perturbations. We conclude that an optimal protocol should not only concentrate computational resources on achieving convergence in the alchemical coupling parameter (λ) space but also on longer simulations and multiple repeats.

Virtual Screening Approach to Identifying a Novel and Tractable Series of Pseudomonas aeruginosa Elastase Inhibitors.

Novel therapies are required to treat chronic bacterial infections in cystic fibrosis (CF) sufferers. The most common pathogen responsible for these infections is Pseudomonas aeruginosa, which persists within the lungs of CF sufferers despite intensive antibiotic treatment. P. aeruginosa elastase (also known as LasB or pseudolysin) is a key virulence determinant that contributes to the pathogenesis and persistence of P. aeruginosa infections in CF patients. The crucial role of LasB in pseudomonal virulence makes it a good target for the development of an adjuvant drug for CF treatment. Herein we discuss the discovery of a new series of LasB inhibitors by virtual screening and computer assisted drug design (CADD) and their optimization leading to compounds 29 and 39 (K i = 0.16 µM and 0.12 µM, respectively).

Utilizing Grand Canonical Monte Carlo Methods in Drug Discovery.

The concepts behind targeting waters for potency and selectivity gains have been well documented and explored, although maximizing such potential gains can prove to be challenging. This problem is exacerbated in cases where there are multiple interacting waters, wherein perturbation of one water can affect the free energy landscape of the remaining waters. Knowing the right modification a priori is challenging, and computational approaches are ideally suited to help answer the key question of which substitution is best to try. Here, we use Grand Canonical Monte Carlo and the recent Grand Canonical Alchemical Perturbation methods to both understand and predict the effect of ligand-mediated water displacement when more than one water molecule is involved, as well as to understand how exploiting water networks can help govern selectivity.

Structure-Based Design and Pharmacokinetic Optimization of Covalent Allosteric Inhibitors of the Mutant GTPase KRASG12C.

Attempts to directly drug the important oncogene KRAS have met with limited success despite numerous efforts across industry and academia. The KRASG12C mutant represents an "Achilles heel" and has recently yielded to covalent targeting with small molecules that bind the mutant cysteine and create an allosteric pocket on GDP-bound RAS, locking it in an inactive state. A weak inhibitor at this site was optimized through conformational locking of a piperazine-quinazoline motif and linker modification. Subsequent introduction of a key methyl group to the piperazine resulted in enhancements in potency, permeability, clearance, and reactivity, leading to identification of a potent KRASG12C inhibitor with high selectivity and excellent cross-species pharmacokinetic parameters and in vivo efficacy.

Lowering Lipophilicity by Adding Carbon: AzaSpiroHeptanes, a logD Lowering Twist.

We have conducted an analysis of azaspiro[3.3]heptanes used as replacements for morpholines, piperidines, and piperazines in a medicinal chemistry context. In most cases, introducing a spirocyclic center lowered the measured logD 7.4 of the corresponding molecules by as much as -1.0 relative to the more usual heterocycle. This may seem counterintuitive, as the net change in the molecule is the addition of a single carbon atom, but it may be rationalized in terms of increased basicity. An exception to this was found with N-linked 2-azaspiro[3.3]heptane, where logD 7.4 increased by as much as +0.5, consistent with the addition of carbon. During our investigation, we also concluded that azaspiro[3.3]heptanes are most likely not suitable bioisosteres for morpholines, piperidines, and piperazines, when not used as terminal groups, due to significant changes in their geometry.

DPP1 Inhibitors: Exploring the Role of Water in the S2 Pocket of DPP1 with Substituted Pyrrolidines.

A series of pyrrolidine amino nitrile DPP1 inhibitors have been developed and characterized. The S2 pocket structure-activity relationship for these compounds shows significant gains in potency for DPP1 from interacting further with target residues and a network of water molecules in the binding pocket. Herein we describe the X-ray crystal structures of several of these compounds alongside an analysis of factors influencing the inhibitory potency toward DPP1 of which stabilization of the water network, demonstrated using Grand Canonical Monte Carlo simulations and free energy calculations, is attributed as a main factor.