Best practices for artificial intelligence in life sciences research.

We describe 11 best practices for the successful use of artificial intelligence and machine learning in pharmaceutical and biotechnology research at the data, technology and organizational management levels.

How computational chemistry develops: a tribute to Peter Goodford.

This editorial discusses the foundation of aspects of computational chemistry and is a tribute to Peter Goodford, one of those founders, who recently passed away. Several colleagues describe Professor Goodford's work and the person himself.

Leaving us with fond memories, smiles, SMILES and, alas, tears: a tribute to David Weininger, 1952-2016.

David Weininger's career, accomplishments, genius, and friendship are warmly remembered by several of his colleagues, friends, and admirers.

Urgency and austerity as drivers of success.

This piece describes the approach by which even a small CADD (Computer-Aided Drug Design) group with limited resources and limited time can achieve substantial success given short budgets and the compressed, urgent environment of a biotech. Some comparisons are made with CADD operations in big pharma.

Biological Testing of Organophosphorus-Inactivated Acetylcholinesterase Oxime Reactivators Identified via Virtual Screening.

There is a pressing need for new therapeutics to reactivate covalently inactivated acetylcholinesterase (AChE) due to exposure to organophosphorus (OP) compounds. Current reactivation therapeutics (RTs) are not broad-spectrum and suffer from other liabilities, specifically the inability to cross the blood-brain-barrier. Additionally, the chemical diversity of available therapeutics is small, limiting opportunities for structure-activity relationship (SAR) studies to aid in the design of more effective compounds. In order to find new starting points for the development of oxime-containing therapeutic reactivators and to increase our base of knowledge, we have employed a combination of computational and experimental procedures to identify additional compounds with the real or potential ability to reactivate AChE while augmenting and complementing current knowledge. Computational methods were used to identify previously uninvestigated oxime-containing molecules. Experimentally, six compounds were found with reactivation capabilities comparable to, or exceeding, those of 2-pralidoxime (2-PAM) against a panel of AChE inactivated by paraoxon, diisopropylfluorophosphate (DFP), fenamiphos, and methamidophos. One compound showed enhanced reactivation ability against DFP and fenamiphos, the least tractable of these OPs to be reactivated.

Outcome of the First wwPDB/CCDC/D3R Ligand Validation Workshop.

Crystallographic studies of ligands bound to biological macromolecules (proteins and nucleic acids) represent an important source of information concerning drug-target interactions, providing atomic level insights into the physical chemistry of complex formation between macromolecules and ligands. Of the more than 115,000 entries extant in the Protein Data Bank (PDB) archive, ∼75% include at least one non-polymeric ligand. Ligand geometrical and stereochemical quality, the suitability of ligand models for in silico drug discovery and design, and the goodness-of-fit of ligand models to electron-density maps vary widely across the archive. We describe the proceedings and conclusions from the first Worldwide PDB/Cambridge Crystallographic Data Center/Drug Design Data Resource (wwPDB/CCDC/D3R) Ligand Validation Workshop held at the Research Collaboratory for Structural Bioinformatics at Rutgers University on July 30-31, 2015. Experts in protein crystallography from academe and industry came together with non-profit and for-profit software providers for crystallography and with experts in computational chemistry and data archiving to discuss and make recommendations on best practices, as framed by a series of questions central to structural studies of macromolecule-ligand complexes. What data concerning bound ligands should be archived in the PDB? How should the ligands be best represented? How should structural models of macromolecule-ligand complexes be validated? What supplementary information should accompany publications of structural studies of biological macromolecules? Consensus recommendations on best practices developed in response to each of these questions are provided, together with some details regarding implementation. Important issues addressed but not resolved at the workshop are also enumerated.

Discovery and Development of LX7101, a Dual LIM-Kinase and ROCK Inhibitor for the Treatment of Glaucoma.

The structure of LX7101, a dual LIM-kinase and ROCK inhibitor for the treatment of ocular hypertension and associated glaucoma, is disclosed. Previously reported LIM kinase inhibitors suffered from poor aqueous stability due to solvolysis of the central urea. Replacement of the urea with a hindered amide resulted in aqueous stable compounds, and addition of solubilizing groups resulted in a set of compounds with good properties for topical dosing in the eye and good efficacy in a mouse model of ocular hypertension. LX7101 was selected as a clinical candidate from this group based on superior efficacy in lowering intraocular pressure and a good safety profile. LX7101 completed IND enabling studies and was tested in a Phase 1 clinical trial in glaucoma patients, where it showed efficacy in lowering intraocular pressure.

Exploring the physicochemical properties of oxime-reactivation therapeutics for cyclosarin, sarin, tabun, and VX inactivated acetylcholinesterase.

The inactivation of acetylcholinesterase (AChE) by organophosphorus agent (OP) compounds is a serious problem regardless of how the individual was exposed. The reactivation of OP-inactivated AChE is dependent on the OP conjugate, and commonly a specific oxime is better at reactivating a specific OP conjugate than several diverse OP conjugates. The presented research explores the physicochemical properties needed for the reactivation of OP-inactivated AChE. Four different OPs, cyclosarin, sarin, tabun, and VX, were analyzed using the same set of oxime reactivators. A trial descriptor pool of semiempirical, traditional, and molecular interaction field descriptors was used to construct an ensemble of QSAR models for each OP-conjugate pair. Based on the molecular information and the cross-validation ability, individual QSAR models were selected to be part of an OP-conjugate consensus model. The OP-conjugate specific models provide important insight into the physicochemical properties required to reactivate the OP conjugates of interest. The reactivation of AChE inactivated with either cyclosarin or tabun requires the oxime therapeutic to possess an overall polar-positive surface area. Oxime therapeutics for the reactivation of sarin-inactivated AChE are conformationally dependent while oxime reverse therapeutics for VX require a compact region with a highly hydrophilic region and two positively charged pyridine rings.

The errors of our ways: taking account of error in computer-aided drug design to build confidence intervals for our next 25 years.

The future of the advancement as well as the reputation of computer-aided drug design will be guided by a more thorough understanding of the domain of applicability of our methods and the errors and confidence intervals of their results. The implications of error in current force fields applied to drug design are given are given as an example. Even as our science advances and our hardware become increasingly more capable, our software will be perhaps the most important aspect in this realization. Some recommendations for the future are provided. Education of users is essential for proper use and interpretation of computational results in the future.

An atomic and molecular view of the depth dependence of the free energies of solute transfer from water into lipid bilayers.

Molecular interactions and orientations responsible for differences in 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC) bilayer partitioning of three structurally related drug-like molecules (4-ethylphenol, phenethylamine, and tyramine) were investigated. This work is based on previously reported molecular dynamics (MD) simulations that determined their transverse free energy profiles across the bilayer. Previously, the location where the transfer free energy of the three solutes is highest, which defines the barrier domain for permeability, was found to be the bilayer center, while the interfacial region was found to be the preferred binding region. Contributions of the amino (NH2) and hydroxyl (OH) functional groups to the transfer free energies from water to the interfacial region were found to be very small both experimentally and by MD simulation, suggesting that the interfacial binding of these solutes is hydrophobically driven and occurs with minimal loss of hydrogen-bonding interactions of the polar functional groups which can occur with either water or phospholipid head groups. Therefore, interfacial binding is relatively insensitive to the number or type of polar functional groups on the solute. In contrast, the relative solute free energy in the barrier domain is highly sensitive to the number of polar functional groups on the molecule. The number and types of hydrogen bonds formed between the three solutes and polar phospholipid atoms or with water molecules were determined as a function of solute position in the bilayer. Minima were observed in the number of hydrogen bonds formed by each solute at the center of the bilayer, coinciding with a decrease in the number of water molecules in DOPC as a function of distance away from the interfacial region. In all regions, hydrogen bonds with water molecules account for the majority of hydrogen-bonding interactions observed for each solute. Significant orientational preferences for the solutes are evident in certain regions of the bilayer (e.g., within the ordered chain region and near the interfacial region 20-25 Å from the bilayer center). The preferred orientations are those that preserve favorable molecular interactions for each solute, which vary with the solute structure.

Functional group dependence of solute partitioning to various locations within a DOPC bilayer: a comparison of molecular dynamics simulations with experiment.

Atomic-level molecular dynamics simulations of 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC) bilayers containing small, amphiphilic, drug-like molecules were carried out to examine the influence of polar functionality on membrane partitioning and transport. Three related molecules (tyramine, phenethylamine, and 4-ethylphenol) were chosen to allow a detailed study of the isolated effects of the amine and hydroxyl functionalities on the preferred solute location, free energies of transfer, and the effect of combining both functional groups in a same molecule. Transfer free energy profiles (from water) generated from molecular dynamics (MD) simulations as a function of bilayer depth compared favorably to comparable experimental results. The simulations allowed the determination of the location of the barrier domain for permeability where the transfer free energy is highest and the preferred binding region at which the free energy is a minimum for each of the three solutes. Comparisons of the free energy profiles reveal that the hydrocarbon chain interior is the region most selective to chemical structure of different solutes because the free energies of transfer in that region vary to a significantly greater extent than in other regions of the bilayer. The contributions of the hydroxyl and amino groups to the free energies of solute transfer from water to the interfacial region were close to zero in both the MD simulations and experimental measurements. This suggests that the free energy decrease observed for solute transfer into the head group region occurs with minimal loss in solvation by hydrogen bonding to polar functional groups on the solute and is largely driven by hydrophobicity. Overall, the joint experimental and simulation studies suggest that the assumption of additivity of free energy contributions from multiple polar functional groups on the same molecule may hold for predictions of passive bilayer permeability coefficients providing that the groups are well isolated. However, this assumption does not hold for predictions of relative liposome-binding affinities.

Mechanism of Inhibition of Novel Tryptophan Hydroxylase Inhibitors Revealed by Co-crystal Structures and Kinetic Analysis.

Trytophan Hydroxylase Type I (TPH1), most abundantly expressed in the gastrointestinal tract, initiates the synthesis of serotonin by catalyzing hydroxylation of tryptophan in the presence of biopterin and oxygen. We have previously described three series of novel, periphery-specific TPH1 inhibitors that selectively deplete serotonin in the gastrointestinal tract. We have now determined co-crystal structures of TPH1 with three of these inhibitors at high resolution. Analysis of the structural data showed that each of the three inhibitors fills the tryptophan binding pocket of TPH1 without reaching into the binding site of the cofactor pterin, and induces major conformational changes of the enzyme. The enzyme-inhibitor complexes assume a compact conformation that is similar to the one in tryptophan complex. Kinetic analysis showed that all three inhibitors are competitive versus the substrate tryptophan, consistent with the structural data that the compounds occupy the tryptophan binding site. On the other hand, all three inhibitors appear to be uncompetitive versus the cofactor 6-methyltetrahydropterin, which is not only consistent with the structural data but also indicate that the hydroxylation reaction follows an ordered binding mechanism in which a productive complex is formed only if tryptophan binds only after pterin, similar to the kinetic mechanisms of tyrosine and phenylalanine hydroxylase.