Exploring Graph Traversal Algorithms in Graph-Based Molecular Generation.

Here, we explore the impact of different graph traversal algorithms on molecular graph generation. We do this by training a graph-based deep molecular generative model to build structures using a node order determined via either a breadth- or depth-first search algorithm. What we observe is that using a breadth-first traversal leads to better coverage of training data features compared to a depth-first traversal. We have quantified these differences using a variety of metrics on a data set of natural products. These metrics include percent validity, molecular coverage, and molecular shape. We also observe that by using either a breadth- or depth-first traversal it is possible to overtrain the generative models, at which point the results with either graph traversal algorithm are identical.

pICalculax: Improved Prediction of Isoelectric Point for Modified Peptides.

The isoelectric point of a peptide is a physicochemical property that can be accurately predicted from the sequence of the peptide when the peptide is built from natural amino acids. Peptides can however have chemical modifications, such as phosphorylations, amidations, and unnatural amino acids, which can result in erroneous predictions if not accounted for. Here we report on an open source program, pICalculax, which in an extensible way can handle pI calculations of modified peptides. Tests on a database of modified peptides and experimentally determined pI values show an improvement in pI predictions when taking the modifications into account. The correlation coefficient improves from 0.45 to 0.91, and the root-mean-square deviation likewise improves from 3.3 to 0.9. The program is available at https://github.com/EBjerrum/pICalculax.

Machine learning optimization of cross docking accuracy.

Performance of small molecule automated docking programs has conceptually been divided into docking -, scoring -, ranking - and screening power, which focuses on the crystal pose prediction, affinity prediction, ligand ranking and database screening capabilities of the docking program, respectively. Benchmarks show that different docking programs can excel in individual benchmarks which suggests that the scoring function employed by the programs can be optimized for a particular task. Here the scoring function of Smina is re-optimized towards enhancing the docking power using a supervised machine learning approach and a manually curated database of ligands and cross docking receptor pairs. The optimization method does not need associated binding data for the receptor-ligand examples used in the data set and works with small train sets. The re-optimization of the weights for the scoring function results in a similar docking performance with regard to docking power towards a cross docking test set. A ligand decoy based benchmark indicates a better discrimination between poses with high and low RMSD. The reported parameters for Smina are compatible with Autodock Vina and represent ready-to-use alternative parameters for researchers who aim at pose prediction rather than affinity prediction.

Rigid body essential X-ray crystallography: distinguishing the bend and twist of glutamate receptor ligand binding domains.

The ligand-binding domain (LBD) from the ionotropic glutamate receptor subtype 2 (GluR2) has been shown to adopt a range of ligand-dependent conformational states. These states have been described in terms of the rotation required to fit subdomain (lobe) 2 following superposition of subdomain (lobe) 1. The LBD has a closed-cleft conformation for full agonists, but partial agonists induce a range of closure, which in turn controls the open probability of discrete subconductance states in the full-length receptor. Although this description is useful, it may not account for all physiologically important motions that the receptor undergoes. We have used an approach that combines the methods of essential dynamics and rigid-body dynamics to analyze 124 monomer domains from 55 crystal structures of the GluR2 LBD. We are able to show that partial agonists also induce a significant amount of twist that would not be anticipated using one rotational descriptor between apo and full-agonist-bound states. Furthermore, one of the crystal structures (chain B from 1P1U, the GluR2 L650T-AMPA complex), which has been suggested to represent an agonist-bound inactive form of the receptor, lies at the extreme of this twist motion. We suggest that partial agonists not only prevent full closure but also move the receptor closer to this inactive state. We demonstrate additionally how the method can be used to compare the results of molecular dynamics simulations with the crystallographic data and the extent to which the conformational space explored by both overlaps.

Tweaking agonist efficacy at N-methyl-D-aspartate receptors by site-directed mutagenesis.

The structural basis for partial agonism at N-methyl-D-aspartate (NMDA) receptors is currently unresolved. We have characterized several partial agonists at the NR1/NR2B receptor and investigated the mechanisms underlying their reduced efficacy by introducing mutations in the glutamate binding site. Key residues were selected for mutation based on ligand-protein docking studies using a homology model of NR2B-S1S2 built from the X-ray structure of NR1-S1S2 in complex with glycine. Wild-type and mutant forms of NR2B were coexpressed with NR1 in Xenopus laevis oocytes and characterized by two-electrode voltage-clamp electrophysiology. By combining mutagenesis of residues His486 or Val686 with activation by differently substituted partial agonists, we introduce varying degrees of steric clash between the ligand and the two binding domains S1 and S2. In cases where ligand-protein docking predicts increased steric clashes between agonists and the residues forming the S1-S2 interface, the agonists clearly show decreased relative efficacy. Furthermore, we demonstrate that the mutation S690A affects both potency and efficacy in an agonist-specific manner. The results indicate that essential residues in the ligand binding pocket of NR2B may adopt different conformations depending on the agonist bound. Together, these data indicate that agonist efficacy at the NR2B subunit can be controlled by the extent of steric clashes between the agonist and the ligand binding domains and by ligand-dependent arrangements of residues within the binding pocket.

Design, synthesis, and pharmacology of a highly subtype-selective GluR1/2 agonist, (RS)-2-amino-3-(4-chloro-3-hydroxy-5-isoxazolyl)propionic acid (Cl-HIBO).

On the basis of structural studies, chloro-homoibotenic acid (Cl-HIBO) was designed and synthesized. Cl-HIBO was characterized in binding and electrophysiology experiments on native and cloned subtypes of GluRs. Electrophysiological selectivities ranged from 275 to 1600 for GluR1/2 over GluR3/4. The potent AMPA receptor activity was strongly desensitizing and the neurotoxicity similar to AMPA. Thus, Cl-HIBO is the most subtype selective agonist reported to date on GluR1/2, and offers a new standard for selectively studying subtypes of AMPA receptors.