Machine learning models and performance dependency on 2D chemical descriptor space for retention time prediction of pharmaceuticals.

The predictive modeling of liquid chromatography methods can be an invaluable asset, potentially saving countless hours of labor while also reducing solvent consumption and waste. Tasks such as physicochemical screening and preliminary method screening systems where large amounts of chromatography data are collected from fast and routine operations are particularly well suited for both leveraging large datasets and benefiting from predictive models. Therefore, the generation of predictive models for retention time is an active area of development. However, for these predictive models to gain acceptance, researchers first must have confidence in model performance and the computational cost of building them should be minimal. In this study, a simple and cost-effective workflow for the development of machine learning models to predict retention time using only Molecular Operating Environment 2D descriptors as input for support vector regression is developed. Furthermore, we investigated the relative performance of models based on molecular descriptor space by utilizing uniform manifold approximation and projection and clustering with Gaussian mixture models to identify chemically distinct clusters. Results outlined herein demonstrate that local models trained on clusters in chemical space perform equivalently when compared to models trained on all data. Through 10-fold cross-validation on a comprehensive set containing 67,950 of our company's proprietary analytes, these models achieved coefficients of determination of 0.84 and 3 % error in terms of retention time. This promising statistical significance is found to translate from cross-validation to prospective prediction on an external test set of pharmaceutically relevant analytes. The observed equivalency of global and local modeling of large datasets is retained with METLIN's SMRT dataset, thereby confirming the wider applicability of the developed machine learning workflows for global models.

Unequivocal identification of two-bond heteronuclear correlations in natural products at nanomole scale by i-HMBC.

HMBC is an essential NMR experiment for determining multiple bond heteronuclear correlations in small to medium-sized organic molecules, including natural products, yet its major limitation is the inability to differentiate two-bond from longer-range correlations. There have been several attempts to address this issue, but all reported approaches suffer various drawbacks, such as restricted utility and poor sensitivity. Here we present a sensitive and universal methodology to identify two-bond HMBC correlations using isotope shifts, referred to as i-HMBC (isotope shift detection HMBC). Experimental utility was demonstrated at the sub-milligram / nanomole scale with only a few hours of acquisition time required for structure elucidation of several complex proton-deficient natural products, which could not be fully elucidated by conventional 2D NMR experiments. Because i-HMBC overcomes the key limitation of HMBC without significant reduction in sensitivity or performance, i-HMBC can be used as a complement to HMBC when unambiguous identifications of two-bond correlations are needed.

DELTA50: A Highly Accurate Database of Experimental 1H and 13C NMR Chemical Shifts Applied to DFT Benchmarking.

Density functional theory (DFT) benchmark studies of 1H and 13C NMR chemical shifts often yield differing conclusions, likely due to non-optimal test molecules and non-standardized data acquisition. To address this issue, we carefully selected and measured 1H and 13C NMR chemical shifts for 50 structurally diverse small organic molecules containing atoms from only the first two rows of the periodic table. Our NMR dataset, DELTA50, was used to calculate linear scaling factors and to evaluate the accuracy of 73 density functionals, 40 basis sets, 3 solvent models, and 3 gauge-referencing schemes. The best performing DFT methodologies for 1H and 13C NMR chemical shift predictions were WP04/6-311++G(2d,p) and ωB97X-D/def2-SVP, respectively, when combined with the polarizable continuum solvent model (PCM) and gauge-independent atomic orbital (GIAO) method. Geometries should be optimized at the B3LYP-D3/6-311G(d,p) level including the PCM solvent model for the best accuracy. Predictions of 20 organic compounds and natural products from a separate probe set had root-mean-square deviations (RMSD) of 0.07 to 0.19 for 1H and 0.5 to 2.9 for 13C. Maximum deviations were less than 0.5 and 6.5 ppm for 1H and 13C, respectively.

A Distributional Model of Bound Ligand Conformational Strain: From Small Molecules up to Large Peptidic Macrocycles.

The internal conformational strain incurred by ligands upon binding a target site has a critical impact on binding affinity, and expectations about the magnitude of ligand strain guide conformational search protocols. Estimates for bound ligand strain begin with modeled ligand atomic coordinates from X-ray co-crystal structures. By deriving low-energy conformational ensembles to fit X-ray diffraction data, calculated strain energies are substantially reduced compared with prior approaches. We show that the distribution of expected global strain energy values is dependent on molecular size in a superlinear manner. The distribution of strain energy follows a rectified normal distribution whose mean and variance are related to conformational complexity. The modeled strain distribution closely matches calculated strain values from experimental data comprising over 3000 protein-ligand complexes. The distributional model has direct implications for conformational search protocols as well as for directions in molecular design.

One-Step Regio- and Stereoselective Electrochemical Synthesis of Orexin Receptor Antagonist Oxidative Metabolites.

Synthesis of drug metabolites, which often have complex structures, is an integral step in the evaluation of drug candidate metabolism, pharmacokinetic (PK) properties, and safety profiles. Frequently, such synthetic endeavors entail arduous, multiple-step de novo synthetic routes. Herein, we present the one-step Shono-type electrochemical synthesis of milligrams of chiral α-hydroxyl amide metabolites of two orexin receptor antagonists, MK-8133 and MK-6096, as revealed by a small-scale (pico- to nano-mole level) reaction screening using a lab-built online electrochemistry (EC)/mass spectrometry (MS) (EC/MS) platform. The electrochemical oxidation of MK-8133 and MK-6096 was conducted in aqueous media and found to produce the corresponding α-piperidinols with exclusive regio- and stereoselectivity, as confirmed by high-resolution nuclear magnetic resonance (NMR) characterization of products. Based on density functional theory (DFT) calculations, the exceptional regio- and stereoselectivity for this electrochemical oxidation are governed by more favorable energetics of the transition state, leading to the preferred secondary carbon radical α to the amide group and subsequent steric hindrance associated with the U-shaped conformation of the cation derived from the secondary α-carbon radical, respectively.

Deuterated Modifiers in Sub/Supercritical Fluid Chromatography for Streamlined NMR Structure Elucidation.

Isolation and chemical characterization of target components in fast-paced pharmaceutical laboratories can often be challenging, especially when dealing with mixtures of closely related, possibly unstable species. Traditionally, this process involves intense labor and manual intervention including chromatographic method development and optimization, fraction collection, and drying processes prior to NMR analyses for unambiguous structure elucidation. To circumvent these challenges, a foundational framework for the proper utilization of supercritical carbon dioxide (scCO2) and deuterated modifiers (CD3OD) in sub/supercritical fluid chromatography (SFC) is herein introduced. This facilitates a streamlined multicomponent isolation with minimized protic residues, further enabling immediate NMR analysis. In addition to bypassing tedious drying processes and minimizing analyte degradation, this approach (complementary to traditional reversed-phase liquid chromatography, RPLC) delivers highly efficient separations and automated fraction collection using readily available analytical/midscale SFC instrumentation. A series of diverse analytes across a wide spectrum of chemical properties (acid, basic, and neutral), combined with different stationary-phase columns in SFC are investigated using both a protic organic modifier (CH3OH) and its deuterated counterpart (CD3OD). The power of this framework is demonstrated with pharmaceutically relevant applications in the context of target characterization and analysis of complex multicomponent reaction mixtures from modern synthetic chemistry, demonstrating high isolation yields while reducing both the environmental footprint and manual intervention. This workflow enables unambiguous fast-paced structure elucidation on the analytical scale, providing results that are comparable to traditional, but time-consuming, RPLC purification approaches.

Application of 1,1-ADEQUATE and DFT to correct 13 C misassignments of carbonyl chemical shifts for carbapenem antibiotics.

Prior to the development of sensitive proton-detected 2D NMR experiments, assigning 13 C signals could be a significant challenge, and mistakes have occurred even for prominent compound classes. In this study, 1,1-ADEQUATE data were used to unambiguously reassign the 13 C chemical shifts for the ß-lactam carbonyl at the C-7 position and the proximal carboxylate at the C-10 position of the carbapenems, meropenem and imipenem. Density functional theory (DFT) was then investigated to provide sufficiently accurate 13 C chemical shift predictions, allowing for the carbonyl signal reassignment of thienamycin.

Discovery and Structure-Based Design of Macrocyclic Peptides Targeting STUB1.

Recent evidence suggests that deletion of STUB1─a pivotal negative regulator of interferon-γ sensing─may potentially clear malignant cells. However, current studies rely primarily on genetic approaches, as pharmacological inhibitors of STUB1 are lacking. Identifying a tool compound will be a step toward validating the target in a broader therapeutic sense. Herein, screening more than a billion macrocyclic peptides resulted in STUB1 binders, which were further optimized by a structure-enabled in silico design. The strategy to replace the macrocyclic peptides' hydrophilic and solvent-exposed region with a hydrophobic scaffold improved cellular permeability while maintaining the binding conformation. Further substitution of the permeability-limiting terminal aspartic acid with a tetrazole bioisostere retained the binding to a certain extent while improving permeability, suggesting a path forward. Although not optimal for cellular study, the current lead provides a valuable template for further development into selective tool compounds for STUB1 to enable target validation.

Driving to a Better Understanding of Acyl Glucuronide Transformations Using NMR and Molecular Modeling.

Acyl glucuronide (AG) metabolites of carboxylic acid-containing drugs and products of their transformations have long been implicated in drug-induced liver injury (DILI). To inform on the DILI risk arising from AG reactive intermediates, a comprehensive mechanistic study of enzyme-independent AG rearrangements using nuclear magnetic resonance (NMR) and density functional theory (DFT) was undertaken. NMR spectroscopy was utilized for structure elucidation and kinetics measurements of nine rearrangement and hydrolysis products of 1ß-O-acyl glucuronide of ibufenac. To extract rate constants of rearrangement, mutarotation, and hydrolysis from kinetic data, 11 different kinetic models were examined. Model selection and estimated rate constant verification were supported by measurements of H/D kinetic isotope effects. DFT calculations of ground and transition states supported the proposed kinetic mechanisms and helped to explain the unusually fast intramolecular transacylation rates found for some of the intermediates. The findings of the current study reinforce the notion that the short half-life of parent AG and slow hydrolysis rates of AG rearrangement products are the two key factors that can influence the in vivo toxicity of AGs.

Conformational Strain of Macrocyclic Peptides in Ligand-Receptor Complexes Based on Advanced Refinement of Bound-State Conformers.

Macrocyclic peptides are an important modality in drug discovery, but molecular design is limited due to the complexity of their conformational landscape. To better understand conformational propensities, global strain energies were estimated for 156 protein-macrocyclic peptide cocrystal structures. Unexpectedly large strain energies were observed when the bound-state conformations were modeled with positional restraints. Instead, low-energy conformer ensembles were generated using xGen that fit experimental X-ray electron density maps and gave reasonable strain energy estimates. The ensembles featured significant conformational adjustments while still fitting the electron density as well or better than the original coordinates. Strain estimates suggest the interaction energy in protein-ligand complexes can offset a greater amount of strain for macrocyclic peptides than for small molecules and non-peptidic macrocycles. Across all molecular classes, the approximate upper bound on global strain energies had the same relationship with molecular size, and bound-state ensembles from xGen yielded favorable binding energy estimates.

Identification and characterization of a residual host cell protein hexosaminidase B associated with N-glycan degradation during the stability study of a therapeutic recombinant monoclonal antibody product.

Host cell proteins (HCPs) are process-related impurities derived from host organisms, which need to be controlled to ensure adequate product quality and safety. In this study, product quality attributes were tracked for several monoclonal antibodies (mAbs) under the intended storage and accelerated stability conditions. One product quality attribute not expected to be stability indicating is the N-glycan heterogeneity profile. However, significant N-glycan degradation was observed for one mAb under accelerated and stressed stability conditions. The root cause for this instability was attributed to hexosaminidase B (HEXB), an enzyme known to remove terminal N-acetylglucosamine (GlcNAc). HEXB was identified by liquid chromatography-mass spectrometry (LC-MS)-based proteomics approach to be enriched in the impacted stability batches from mAb-1. Subsequently, enzymatic and targeted multiple reaction monitoring (MRM) MS assays were developed to support process and product characterization. A potential interaction between HEXB and mAb-1 was initially observed from the analysis of process intermediates by proteomics among several mAbs and later supported by computational modeling. An improved bioprocess was developed to significantly reduce HEXB levels in the final drug substance. A risk assessment was conducted by evaluating the in silico immunogenicity risk and the impact on product quality. To the best of our knowledge, HEXB is the first residual HCP reported to have impact on the glycan profile of a formulated drug product. The combination of different analytical tools, mass spectrometry, and computational modeling provides a general strategy on how to study residual HCP for biotherapeutics development.

"One-Shot" Measurement of Residual Chemical Shift Anisotropy Using Poly-γ-benzyl-l-glutamate as an Alignment Medium.

A method for the measurement of residual chemical shift anisotropy in one experiment using a biphasic isotropic/anisotropic lyotropic liquid crystalline medium based on poly-γ-benzyl-l-glutamate as the alignment medium is presented. This approach is demonstrated on the model compound strychnine and neotricone, a depsidone natural product with a questionable structural assignment based on comparison with the closely related excelsione and in-depth density functional theory calculations.

Denigrins and Dactylpyrroles, Arylpyrrole Alkaloids from a Dactylia sp. Marine Sponge.

Seven new arylpyrrole alkaloids (1-7), along with four known compounds, were isolated from an extract of a Dactylia sp. nov. marine sponge, and their structures were elucidated by interpretation of NMR and MS spectroscopic data. Denigrins D-G (1-4) have highly substituted pyrrole or pyrrolone rings in their core structures, while dactylpyrroles A-C (5-7) have tricyclic phenanthrene cores. Due to the proton-deficient nature of these scaffolds, key heteronuclear correlations from 1H-15N HMBC and LR-HSQMBC NMR experiments were used in the structure assignment of denigrin D (1). Dictyodendrin F (8), a previously described co-metabolite, inhibited transcription driven by the oncogenic PAX3-FOXO1 fusion gene with an IC50 value of 13 µM.

Direct regioisomer analysis of crude reaction mixtures via molecular rotational resonance (MRR) spectroscopy.

Direct analyses of crude reaction mixtures have been carried out using molecular rotational resonance (MRR) spectroscopy. Two examples are presented, a demonstration application in photocatalytic CH-arylation as well as generation of an intermediate in a natural product synthesis. In both cases, the reaction can proceed at more than one site, leading to a mixture of regioisomers that can be challenging to distinguish. MRR structural parameters were calculated for the low lying conformers for the desired compounds, and then compared to the experimental spectra of the crude mixtures to confirm the presence of these species. Next, quantitation was performed by comparing experimentally measured line intensities with simulations based on computed values for the magnitude and direction of the molecular dipole moment of each species. This identification and quantification was performed without sample purification and without isolated standards of the compounds of interest. The values obtained for MRR quantitation were in good agreement with the chromatographic values. Finally, previously unknown impurities were discovered within the photocatalytic CH-arylation work. This paper demonstrates the utility of MRR as a reaction characterization tool to simplify analytical workflows.

XGen: Real-Space Fitting of Complex Ligand Conformational Ensembles to X-ray Electron Density Maps.

We report a new method for X-ray density ligand fitting and refinement that is suitable for a wide variety of small-molecule ligands, including macrocycles. The approach (called "xGen") augments a force field energy calculation with an electron density fitting restraint that yields an energy reward during the restrained conformational search. The resulting conformer pools balance goodness-of-fit with ligand strain. Real-space refinement from pre-existing ligand coordinates of 150 macrocycles resulted in occupancy-weighted conformational ensembles that exhibited low strain energy. The xGen ensembles improved upon electron density fit compared with the PDB reference coordinates without making use of atom-specific B-factors. Similarly, on nonmacrocycles, de novo fitting produced occupancy-weighted ensembles of many conformers that were generally better-quality density fits than the deposited primary/alternate conformational pairs. The results suggest ubiquitous low-energy ligand conformational ensembles in X-ray diffraction data and provide an alternative to using B-factors as model parameters.

Analysis of Benzenoid Substitution Patterns in Small Molecule Active Pharmaceutical Ingredients.

Previous analyses have revealed that benzenoid rings are prevalent scaffolds in active pharmaceutical ingredients (APIs). Here, we analyze the substitution patterns of benzenoid rings in small molecule APIs approved by the FDA through 2019 and show that only a few substitution patterns (1-, 1,2-, 1,4-, and 1,2,4-) prevail, and the distribution has remained relatively constant over time. We postulate the connection between available synthetic methods and the occurrence of a few benzenoid substitution patterns by providing an overview of synthetic methods that elaborate existing substitution patterns and those that create new substitution patterns, including those of the former that are favored by medicinal chemists. Finally, we calculated medicinal chemistry properties of benzenoid containing APIs that are often used by practitioners as design elements, including "druglikeness", shape, complexity, and similarity/diversity and discuss these properties in the context of synthesis.

Deep Dive into Machine Learning Models for Protein Engineering.

Protein redesign and engineering has become an important task in pharmaceutical research and development. Recent advances in technology have enabled efficient protein redesign by mimicking natural evolutionary mutation, selection, and amplification steps in the laboratory environment. For any given protein, the number of possible mutations is astronomical. It is impractical to synthesize all sequences or even to investigate all functionally interesting variants. Recently, there has been an increased interest in using machine learning to assist protein redesign, since prediction models can be used to virtually screen a large number of novel sequences. However, many state-of-the-art machine learning models, especially deep learning models, have not been extensively explored. Moreover, only a small selection of protein sequence descriptors has been considered. In this work, the performance of prediction models built using an array of machine learning methods and protein descriptor types, including two novel, single amino acid descriptors and one structure-based three-dimensional descriptor, is benchmarked. The predictions were evaluated on a diverse collection of public and proprietary data sets, using a variety of evaluation metrics. The results of this comparison suggest that Convolution Neural Network models built with amino acid property descriptors are the most widely applicable to the types of protein redesign problems faced in the pharmaceutical industry.

The merger of decatungstate and copper catalysis to enable aliphatic C(sp3)-H trifluoromethylation.

The introduction of a trifluoromethyl (CF3) group can dramatically improve a compound's biological properties. Despite the well-established importance of trifluoromethylated compounds, general methods for the trifluoromethylation of alkyl C-H bonds remain elusive. Here we report the development of a dual-catalytic C(sp3)-H trifluoromethylation through the merger of light-driven, decatungstate-catalysed hydrogen atom transfer and copper catalysis. This metallaphotoredox methodology enables the direct conversion of both strong aliphatic and benzylic C-H bonds into the corresponding C(sp3)-CF3 products in a single step using a bench-stable, commercially available trifluoromethylation reagent. The reaction requires only a single equivalent of substrate and proceeds with excellent selectivity for positions distal to unprotected amines. To demonstrate the utility of this new methodology for late-stage functionalization, we have directly derivatized a broad range of approved drugs and natural products to generate valuable trifluoromethylated analogues. Preliminary mechanistic experiments reveal that a 'Cu-CF3' species is formed during this process and the critical C(sp3)-CF3 bond-forming step involves the copper catalyst.

Combination of HDX-MS and in silico modeling to study enzymatic reactivity and stereo-selectivity at different solvent conditions.

The higher-order structure of a protein defines its function, and protein structural dynamics are often essential for protein binding and enzyme catalysis. Methods for protein characterization in solution are continuously being developed to understand and explore protein conformational changes with regards to function and activity. The goal of this study was to survey the use of combining HDX-MS global conformational screening with in silico modeling and continuous labeling peptide-level HDX-MS as an approach to highlight regions of interest within an enzyme required for biocatalytic processes. We surveyed in silico modeling correlated with peptide level HDX-MS experiments to characterize and localize transaminase enzyme structural dynamics at different conditions. This approach was orthogonally correlated with a global Size-Exclusion-HDX (SEC-HDX) screen for global conformational comparison and global alpha-helical content measurements by circular dichroism. Enzymatic activity and stereo-selectivity of transaminases were compared at different reaction-solution conditions that forced protein conformational changes by increasing acetonitrile concentration. The experimental peptide-level HDX-MS results demonstrated similar trends to the modeling data showing that certain regions remained folded in transaminases ATA-036 and ATA-303 with increasing acetonitrile concentration, which is also associated with shifting stereoselectivity. HDX modeling, SEC-HDX and CD experimental data showed that transaminase ATA-234 had the highest level of global unfolding with increasing acetonitrile concentration compared to the other two enzymes, which correlated with drastically reduced product conversion in transamination reaction. The combined HDX modeling/experimental workflow, based on enzymatic reactions studied at different conditions to induce changes in enzyme conformation, could be used as a tool to guide directed evolution efforts by identifying and focusing on the regions of an enzyme required for reaction product conversion and stereoselectivity.

Unequivocal structure confirmation of a breitfussin analog by anisotropic NMR measurements.

Structural features of proton-deficient heteroaromatic natural products, such as the breitfussins, can severely complicate their characterization by NMR spectroscopy. For the breitfussins in particular, the constitution of the five-membered oxazole central ring cannot be unequivocally established via conventional NMR methods when the 4'-position is halogenated. The level of difficulty is exacerbated by 4'-iodination, as the accuracy with which theoretical NMR parameters are determined relies extensively on computational treatment of the relativistic effects of the iodine atom. It is demonstrated in the present study, that the structure of a 4'-iodo breitfussin analog can be unequivocally established by anisotropic NMR methods, by adopting a reduced singular value decomposition (SVD) protocol that leverages the planar structures exhibited by its conformers.