The reorganization energy of compounds upon binding to proteins, from dynamic and solvated bound and unbound states.

The intramolecular reorganization energy (ΔEReorg) of compounds upon binding to proteins is a component of the binding free energy, which has long received particular attention, for fundamental and practical reasons. Understanding ΔEReorg would benefit the science of molecular recognition and drug design. For instance, the tolerable strain energy of compounds upon binding has been elusive. Prior studies found some large ΔEReorg values (e.g. > 10 kcal/mol), received with skepticism since they imply excessive opposition to binding. Indeed, estimating ΔEReorg is technically difficult. Typically, ΔEReorg has been approached by taking two energy-minimized conformers representing the bound and unbound states, and subtracting their conformational energy. This is a drastic oversimplification, liable to conformational collapse of the unbound conformer. Instead, the present work applies extensive molecular dynamics (MD) and the modern OPLS3 force-field to simulate compounds bound and unbound states, in explicit solvent under physically relevant conditions. The thermalized unbound compounds populate multiple conformations, not reducible to one or a few energy-minimized conformers. The intramolecular energies in the bound and unbound states were averaged over pertinent conformational ensembles, and the reorganization enthalpy upon binding (ΔHReorg) deduced by subtraction. This was applied to 76 systems, including 43 approved drugs, carefully selected for i) the quality of the bioactive X-ray structures and ii) the diversity of the chemotypes, their properties and protein targets. It yielded comparatively low ΔHReorg values (median = 1.4 kcal/mol, mean = 3.0 kcal/mol). A new finding is the observation of negative ΔHReorg values. Indeed, reorganization energies do not have to oppose binding, e.g. when intramolecular interactions stabilize preferentially the bound state. Conversely, even with competing water molecules, intramolecular interactions can occur predominantly in the unbound compound, and be replaced by intermolecular counterparts upon protein binding. Such disruption of intramolecular interactions upon binding gives rise to occasional larger ΔHReorg values. Such counterintuitive larger ΔHReorg values may be rationalized as a redistribution of interactions upon binding, qualitatively compatible with binding.

The Effect of Core Replacement on S64315, a Selective MCL-1 Inhibitor, and Its Analogues.

Following the identification of thieno[2,3-d]pyrimidine-based selective and potent inhibitors of MCL-1, we explored the effect of core swapping at different levels of advancement. During hit-to-lead optimization, X-ray-guided S-N replacement in the core provided a new vector, whose exploration led to the opening of the so-called deep-S2 pocket of MCL-1. Unfortunately, the occupation of this region led to a plateau in affinity and had to be abandoned. As the project approached selection of a clinical candidate, a series of core swap analogues were also prepared. The affinity and cellular activity of these compounds showed a significant dependence on the core structure. In certain cases, we also observed an increased and accelerated epimerization of the atropoisomers. The most potent core replacement analogues showed considerable in vivo PD response. One compound was progressed into efficacy studies and inhibited tumor growth.

Design and Synthesis of Pyrrolo[2,3-d]pyrimidine-Derived Leucine-Rich Repeat Kinase 2 (LRRK2) Inhibitors Using a Checkpoint Kinase 1 (CHK1)-Derived Crystallographic Surrogate.

Inhibitors of leucine-rich repeat kinase 2 (LRRK2) and mutants, such as G2019S, have potential utility in Parkinson's disease treatment. Fragment hit-derived pyrrolo[2,3-d]pyrimidines underwent optimization using X-ray structures of LRRK2 kinase domain surrogates, based on checkpoint kinase 1 (CHK1) and a CHK1 10-point mutant. (2R)-2-Methylpyrrolidin-1-yl derivative 18 (LRRK2 G2019S cKi 0.7 nM, LE 0.66) was identified, with increased potency consistent with an X-ray structure of 18/CHK1 10-pt. mutant showing the 2-methyl substituent proximal to Ala147 (Ala2016 in LRRK2). Further structure-guided elaboration of 18 gave the 2-[(1,3-dimethyl-1H-pyrazol-4-yl)amino] derivative 32. Optimization of 32 afforded diastereomeric oxolan-3-yl derivatives 44 and 45, which demonstrated a favorable in vitro PK profile, although they displayed species disconnects in the in vivo PK profile, and a propensity for P-gp- and/or BCRP-mediated efflux in a mouse model. Compounds 44 and 45 demonstrated high potency and exquisite selectivity for LRRK2 and utility as chemical probes for the study of LRRK2 inhibition.

Discovery of S64315, a Potent and Selective Mcl-1 Inhibitor.

Myeloid cell leukemia 1 (Mcl-1) has emerged as an attractive target for cancer therapy. It is an antiapoptotic member of the Bcl-2 family of proteins, whose upregulation in human cancers is associated with high tumor grade, poor survival, and resistance to chemotherapy. Here we report the discovery of our clinical candidate S64315, a selective small molecule inhibitor of Mcl-1. Starting from a fragment derived lead compound, we have conducted structure guided optimization that has led to a significant (3 log) improvement of target affinity as well as cellular potency. The presence of hindered rotation along a biaryl axis has conferred high selectivity to the compounds against other members of the Bcl-2 family. During optimization, we have also established predictive PD markers of Mcl-1 inhibition and achieved both efficient in vitro cell killing and tumor regression in Mcl-1 dependent cancer models. The preclinical candidate has drug-like properties that have enabled its development and entry into clinical trials.

Exploring IDP-Ligand Interactions: tau K18 as A Test Case.

Over the past decade intrinsically disordered proteins (IDPs) have emerged as a biologically important class of proteins, many of which are of therapeutic relevance. Here, we investigated the interactions between a model IDP system, tau K18, and nine literature compounds that have been reported as having an effect on tau in order to identify a robust IDP-ligand system for the optimization of a range of biophysical methods. We used NMR, surface plasmon resonance (SPR) and microscale thermophoresis (MST) methods to investigate the binding of these compounds to tau K18; only one showed unambiguous interaction with tau K18. Several near neighbors of this compound were synthesized and their interactions with tau K18 characterized using additional NMR methods, including 1D ligand-observed NMR, diffusion-ordered spectroscopy (DOSY) and 19F NMR. This study demonstrates that it is possible to detect and characterize IDP-ligand interactions using biophysical methods. However, care must be taken to account for possible artefacts, particularly the impact of compound solubility and where the protein has to be immobilized.

Modelling the binding mode of macrocycles: Docking and conformational sampling.

Drug discovery is increasingly tackling challenging protein binding sites regarding molecular recognition and druggability, including shallow and solvent-exposed protein-protein interaction interfaces. Macrocycles are emerging as promising chemotypes to modulate such sites. Despite their chemical complexity, macrocycles comprise important drugs and offer advantages compared to non-cyclic analogs, hence the recent impetus in the medicinal chemistry of macrocycles. Elaboration of macrocycles, or constituent fragments, can strongly benefit from knowledge of their binding mode to a target. When such information from X-ray crystallography is elusive, computational docking can provide working models. However, few studies have explored docking protocols for macrocycles, since conventional docking methods struggle with the conformational complexity of macrocycles, and also potentially with the shallower topology of their binding sites. Indeed, macrocycle binding mode prediction with the mainstream docking software GOLD has hardly been explored. Here, we present an in-depth study of macrocycle docking with GOLD and the ChemPLP scores. First, we summarize the thorough curation of a test set of 41 protein-macrocycle X-ray structures, raising the issue of lattice contacts with such systems. Rigid docking of the known bioactive conformers was successful (three top ranked poses) for 92.7% of the systems, in absence of crystallographic waters. Thus, without conformational search issues, scoring performed well. However, docking success dropped to 29.3% with the GOLD built-in conformational search. Yet, the success rate doubled to 58.5% when GOLD was supplied with extensive conformer ensembles docked rigidly. The reasons for failure, sampling or scoring, were analyzed, exemplified with particular cases. Overall, binding mode prediction of macrocycles remains challenging, but can be much improved with tailored protocols. The analysis of the interplay between conformational sampling and docking will be relevant to the prospective modelling of macrocycles in general.

Establishing Drug Discovery and Identification of Hit Series for the Anti-apoptotic Proteins, Bcl-2 and Mcl-1.

We describe our work to establish structure- and fragment-based drug discovery to identify small molecules that inhibit the anti-apoptotic activity of the proteins Mcl-1 and Bcl-2. This identified hit series of compounds, some of which were subsequently optimized to clinical candidates in trials for treating various cancers. Many protein constructs were designed to identify protein with suitable properties for different biophysical assays and structural methods. Fragment screening using ligand-observed NMR experiments identified several series of compounds for each protein. The series were assessed for their potential for subsequent optimization using 1H and 15N heteronuclear single-quantum correlation NMR, surface plasmon resonance, and isothermal titration calorimetry measurements to characterize and validate binding. Crystal structures could not be determined for the early hits, so NMR methods were developed to provide models of compound binding to guide compound optimization. For Mcl-1, a benzodioxane/benzoxazine series was optimized to a K d of 40 µM before a thienopyrimidine hit series was identified which subsequently led to the lead series from which the clinical candidate S 64315 (MIK 665) was identified. For Bcl-2, the fragment-derived series were difficult to progress, and a compound derived from a published tetrahydroquinone compound was taken forward as the hit from which the clinical candidate (S 55746) was obtained. For both the proteins, the work to establish a portfolio of assays gave confidence for identification of compounds suitable for optimization.

Structure-Guided Discovery of a Selective Mcl-1 Inhibitor with Cellular Activity.

Myeloid cell leukemia 1 (Mcl-1), an antiapoptotic member of the Bcl-2 family of proteins, whose upregulation when observed in human cancers is associated with high tumor grade, poor survival, and resistance to chemotherapy, has emerged as an attractive target for cancer therapy. Here, we report the discovery of selective small molecule inhibitors of Mcl-1 that inhibit cellular activity. Fragment screening identified thienopyrimidine amino acids as promising but nonselective hits that were optimized using nuclear magnetic resonance and X-ray-derived structural information. The introduction of hindered rotation along a biaryl axis has conferred high selectivity to the compounds, and cellular activity was brought on scale by offsetting the negative charge of the anchoring carboxylate group. The obtained compounds described here exhibit nanomolar binding affinity and mechanism-based cellular efficacy, caspase induction, and growth inhibition. These early research efforts illustrate drug discovery optimization from thienopyrimidine hits to a lead compound, the chemical series leading to the identification of our more advanced compounds S63845 and S64315.

Energy windows for computed compound conformers: covering artefacts or truly large reorganization energies?

The generation of 3D conformers of small molecules underpins most computational drug discovery. Thus, the conformer quality is critical and depends on their energetics. A key parameter is the empirical conformational energy window (ΔEw), since only conformers within ΔEw are retained. However, ΔEw values in use appear unrealistically large. We analyze the factors pertaining to the conformer energetics and ΔEw. We argue that more attention must be focused on the problem of collapsed low-energy conformers. That is due to artificial intramolecular stabilization and occurs even with continuum solvation. Consequently, the conformational energy of extended bioactive structures is artefactually increased, which inflates ΔEw. Thus, this Perspective highlights the issues arising from low-energy conformers and suggests improvements via empirical or physics-based strategies.

Using Spin-Coated Silver Nanoparticles/Zinc Oxide Thin Films to Improve the Efficiency of GaInP/(In)GaAs/Ge Solar Cells.

We synthesized a silver nanoparticle/zinc oxide (Ag NP/ZnO) thin film by using spin-coating technology. The treatment solution for Ag NP/ZnO thin film deposition contained zinc acetate (Zn(CH3COO)2), sodium hydroxide (NaOH), and silver nitrate (AgNO3) aqueous solutions. The crystalline characteristics, surface morphology, content of elements, and reflectivity of the Ag NPs/ZnO thin film at various concentrations of the AgNO3 aqueous solution were investigated using X-ray diffraction, scanning electron microscopy, energy-dispersive X-ray spectroscopy, atomic force microscopy, and ultraviolet⁻visible⁻near infrared spectrophotometry. The results indicated that the crystalline structure, Ag content, and reflectance of Ag NP/ZnO thin films depended on the AgNO3 concentration. Hybrid antireflection coatings (ARCs) composed of SiNx and Ag NPs/ZnO thin films with various AgNO3 concentrations were deposited on GaInP/(In)GaAs/Ge solar cells. We propose that the optimal ARC consists of SiNx and Ag NP/ZnO thin films prepared using a treatment solution of 0.0008 M AgNO3, 0.007 M Zn(CH3COO)2, and 1 M NaOH, followed by post-annealing at 200 °C. GaInP/(Al)GaAs/Ge solar cells with the optimal hybrid ARC and SiNx ARC exhibit a conversion efficiency of 34.1% and 30.2% with Voc = 2.39 and 2.4 V, Jsc = 16.63 and 15.37 mA/cm², and fill factor = 86.1% and 78.8%.

S55746 is a novel orally active BCL-2 selective and potent inhibitor that impairs hematological tumor growth.

Escape from apoptosis is one of the major hallmarks of cancer cells. The B-cell Lymphoma 2 (BCL-2) gene family encodes pro-apoptotic and anti-apoptotic proteins that are key regulators of the apoptotic process. Overexpression of the pro-survival member BCL-2 is a well-established mechanism contributing to oncogenesis and chemoresistance in several cancers, including lymphoma and leukemia. Thus, BCL-2 has become an attractive target for therapeutic strategy in cancer, as demonstrated by the recent approval of ABT-199 (Venclexta™) in relapsed or refractory Chronic Lymphocytic Leukemia with 17p deletion. Here, we describe a novel orally bioavailable BCL-2 selective and potent inhibitor called S55746 (also known as BCL201). S55746 occupies the hydrophobic groove of BCL-2. Its selectivity profile demonstrates no significant binding to MCL-1, BFL-1 (BCL2A1/A1) and poor affinity for BCL-XL. Accordingly, S55746 has no cytotoxic activity on BCL-XL-dependent cells, such as platelets. In a panel of hematological cell lines, S55746 induces hallmarks of apoptosis including externalization of phosphatidylserine, caspase-3 activation and PARP cleavage. Ex vivo, S55746 induces apoptosis in the low nanomolar range in primary Chronic Lymphocytic Leukemia and Mantle Cell Lymphoma patient samples. Finally, S55746 administered by oral route daily in mice demonstrated robust anti-tumor efficacy in two hematological xenograft models with no weight lost and no change in behavior. Taken together, these data demonstrate that S55746 is a novel, well-tolerated BH3-mimetic targeting selectively and potently the BCL-2 protein.

Design of Leucine-Rich Repeat Kinase 2 (LRRK2) Inhibitors Using a Crystallographic Surrogate Derived from Checkpoint Kinase 1 (CHK1).

Mutations in leucine-rich repeat kinase 2 (LRRK2), such as G2019S, are associated with an increased risk of developing Parkinson's disease. Surrogates for the LRRK2 kinase domain based on checkpoint kinase 1 (CHK1) mutants were designed, expressed in insect cells infected with baculovirus, purified, and crystallized. X-ray structures of the surrogates complexed with known LRRK2 inhibitors rationalized compound potency and selectivity. The CHK1 10-point mutant was preferred, following assessment of surrogate binding affinity with LRRK2 inhibitors. Fragment hit-derived arylpyrrolo[2,3-b]pyridine LRRK2 inhibitors underwent structure-guided optimization using this crystallographic surrogate. LRRK2-pSer935 HEK293 IC50 data for 22 were consistent with binding to Ala2016 in LRRK2 (equivalent to Ala147 in CHK1 10-point mutant structure). Compound 22 was shown to be potent, moderately selective, orally available, and brain-penetrant in wild-type mice, and confirmation of target engagement was demonstrated, with LRRK2-pSer935 IC50 values for 22 in mouse brain and kidney being 1.3 and 5 nM, respectively.

The design and SAR of a novel series of 2-aminopyridine based LRRK2 inhibitors.

Leucine-rich repeat kinase 2 (LRRK2) has attracted considerable interest as a therapeutic target for the treatment of Parkinson's disease. Compounds derived from a 2-aminopyridine screening hit were optimised using a LRRK2 homology model based on mixed lineage kinase 1 (MLK1), such that a 2-aminopyridine-based lead molecule 45, with in vivo activity, was identified.

The MCL1 inhibitor S63845 is tolerable and effective in diverse cancer models.

Avoidance of apoptosis is critical for the development and sustained growth of tumours. The pro-survival protein myeloid cell leukemia 1 (MCL1) is overexpressed in many cancers, but the development of small molecules targeting this protein that are amenable for clinical testing has been challenging. Here we describe S63845, a small molecule that specifically binds with high affinity to the BH3-binding groove of MCL1. Our mechanistic studies demonstrate that S63845 potently kills MCL1-dependent cancer cells, including multiple myeloma, leukaemia and lymphoma cells, by activating the BAX/BAK-dependent mitochondrial apoptotic pathway. In vivo, S63845 shows potent anti-tumour activity with an acceptable safety margin as a single agent in several cancers. Moreover, MCL1 inhibition, either alone or in combination with other anti-cancer drugs, proved effective against several solid cancer-derived cell lines. These results point towards MCL1 as a target for the treatment of a wide range of tumours.

Towards understanding the unbound state of drug compounds: Implications for the intramolecular reorganization energy upon binding.

There has been an explosion of structural information for pharmaceutical compounds bound to biological targets, but the conformations and dynamics of compounds free in solution are poorly characterized, if at all. Yet, knowledge of the unbound state is essential to understand the fundamentals of molecular recognition, including the much debated conformational intramolecular reorganization energy of a compound upon binding (ΔEReorg). Also, dependable observation of the unbound compounds is important for ligand-based drug discovery, e.g. with pharmacophore modelling. Here, these questions are addressed with long (⩾0.5µs) state-of-the-art molecular dynamics (MD) simulations of 26 compounds (including 7 approved drugs) unbound in explicit solvent. These compounds were selected to be chemically diverse, with a range of flexibility, and good quality bioactive X-ray structures. The MD-simulated free compounds are compared to their bioactive structure and conformers generated with ad hoc sampling in vacuo or with implicit generalized Born (GB) aqueous solvation models. The GB conformational models clearly depart from those obtained in explicit solvent, and suffer from conformational collapse almost as severe as in vacuo. Thus, the global energy minima in vacuo or with GB are not suitable representations of the unbound state, which can instead be extensively sampled by MD simulations. Many, but not all, MD-simulated compounds displayed some structural similarity to their bioactive structure, supporting the notion of conformational pre-organization for binding. The ligand-protein complexes were also simulated in explicit solvent, to estimate ΔEReorg as an enthalpic difference ΔHReorg between the intramolecular energies in the bound and unbound states. This fresh approach yielded ΔHReorg values⩽6kcal/mol for 18 out of 26 compounds. For three particularly polar compounds 15⩽ΔHReorg⩽20kcal/mol, supporting the notion that ΔHReorg can be substantial. Those large ΔHReorg values correspond to a redistribution of electrostatic interactions upon binding. Overall, the study illustrates how MD simulations offer a promising avenue to characterize the unbound state of medicinal compounds.

Tackling the conformational sampling of larger flexible compounds and macrocycles in pharmacology and drug discovery.

Computational conformational sampling underpins much of molecular modeling and design in pharmaceutical work. The sampling of smaller drug-like compounds has been an active area of research. However, few studies have tested in details the sampling of larger more flexible compounds, which are also relevant to drug discovery, including therapeutic peptides, macrocycles, and inhibitors of protein-protein interactions. Here, we investigate extensively mainstream conformational sampling methods on three carefully curated compound sets, namely the 'Drug-like', larger 'Flexible', and 'Macrocycle' compounds. These test molecules are chemically diverse with reliable X-ray protein-bound bioactive structures. The compared sampling methods include Stochastic Search and the recent LowModeMD from MOE, all the low-mode based approaches from MacroModel, and MD/LLMOD recently developed for macrocycles. In addition to default settings, key parameters of the sampling protocols were explored. The performance of the computational protocols was assessed via (i) the reproduction of the X-ray bioactive structures, (ii) the size, coverage and diversity of the output conformational ensembles, (iii) the compactness/extendedness of the conformers, and (iv) the ability to locate the global energy minimum. The influence of the stochastic nature of the searches on the results was also examined. Much better results were obtained by adopting search parameters enhanced over the default settings, while maintaining computational tractability. In MOE, the recent LowModeMD emerged as the method of choice. Mixed torsional/low-mode from MacroModel performed as well as LowModeMD, and MD/LLMOD performed well for macrocycles. The low-mode based approaches yielded very encouraging results with the flexible and macrocycle sets. Thus, one can productively tackle the computational conformational search of larger flexible compounds for drug discovery, including macrocycles.

Drug-like bioactive structures and conformational coverage with the LigPrep/ConfGen suite: comparison to programs MOE and catalyst.

Computational conformational sampling underpins many aspects of small molecule modeling and design in pharmaceutical work. This work examined in detail the widely distributed LigPrep/ConfGen software suite and the conformational models it produces for drug-like compounds. We also compare LigPrep/ConfGen to MOE and Catalyst. Tests of the conformational sampling protocols included the reproduction of known bioactive structures of ligands, characterization of the size, coverage and diversity of the output conformational models, and relative computation times. The present tests will help the user to make informed choices among the predefined ConfGen protocols (Very fast, Fast, Intermediate, and Comprehensive), and the adjustable input parameters. The parameters governing the initial compound preparation (LigPrep) and the subsequent conformational sampling were explored. This analysis has led to a new protocol called "ConfGen Optimized", which improves upon the predefined protocols. ConfGen Optimized is computationally tractable and reproduced 80% of the bioactive structures within 1 A, versus 66% for the default ConfGen Fast protocol. We also addressed the issue of the reproduction of compact/folded bioactive structures by ConfGen. It involved the compilation of a new set of 50 folded diverse drug-like bioactive structures. This indicates that heuristics penalizing folded conformers hinder reproduction of some binding modes. Overall, ConfGen offers great flexibility of use and provides a valuable addition to the molecular modeling toolbox.