Optimizing Excipient Properties to Prevent Aggregation in Biopharmaceutical Formulations.

Excipients are included within protein biotherapeutic solution formulations to improve colloidal and conformational stability but are generally not designed for the specific purpose of preventing aggregation and improving cryoprotection in solution. In this work, we have explored the relationship between the structure and antiaggregation activity of excipients by utilizing coarse-grained molecular dynamics modeling of protein-excipient interaction. We have studied human serum albumin as a model protein, and we report the interaction of 41 excipients (polysorbates, fatty alcohol ethoxylates, fatty acid ethoxylates, phospholipids, glucosides, amino acids, and others) in terms of the reduction of solvent accessible surface area of aggregation-prone regions, proposed as a mechanism of aggregation prevention. Polyoxyethylene sorbitan had the greatest degree of interaction with aggregation-prone regions, decreasing the solvent accessible surface area of APRs by 20.7 nm2 (40.1%). Physicochemical descriptors generated by Mordred are employed to probe the structure-property relationship using partial least-squares regression. A leave-one-out cross-validated model had a root-mean-square error of prediction of 4.1 nm2 and a mean relative error of prediction of 0.077. Generally, longer molecules with a large number of alcohol-terminated PEG units tended to interact more, with qualitatively different protein interactions, wrapping around the protein. Shorter or less ethoxylated compounds tend to form hemimicellar clusters at the protein surface. We propose that an improved design would feature many short chains of 5 to 10 PEG units in many distinct branches and at least some hydrophobic content in the form of medium-length or greater aliphatic chains (i.e., six or more carbon atoms). The combination of molecular dynamics simulation and quantitative modeling is an important first step in an all-purpose protein-independent model for the computer-aided design of stabilizing excipients.

Theoretical analysis of divalent cation effects on aptamer recognition of neurotransmitter targets.

Aptamer-based sensing of small molecules such as dopamine and serotonin in the brain, requires characterization of the specific aptamer sequences in solutions mimicking the in vivo environment with physiological ionic concentrations. In particular, divalent cations (Mg2+ and Ca2+) present in brain fluid, have been shown to affect the conformational dynamics of aptamers upon target recognition. Thus, for biosensors that transduce aptamer structure switching as the signal response, it is critical to interrogate the influence of divalent cations on each unique aptamer sequence. Herein, we demonstrate the potential of molecular dynamics (MD) simulations to predict the behaviour of dopamine and serotonin aptamers on sensor surfaces. The simulations enable molecular-level visualization of aptamer conformational changes that, in some cases, are significantly influenced by divalent cations. The correlations of theoretical simulations with experimental findings validate the potential for MD simulations to predict aptamer-specific behaviors on biosensors.

AI4Green: An Open-Source ELN for Green and Sustainable Chemistry.

An Electronic Laboratory Notebook (ELN) combining features, including data archival, collaboration tools, and green and sustainability metrics for organic chemistry, is presented. AI4Green is a web-based application, available as open-source code and free to use. It offers the core functionality of an ELN, namely, the ability to store reactions securely and share them among different members of a research team. As users plan their reactions and record them in the ELN, green and sustainable chemistry is encouraged by automatically calculating green metrics and color-coding hazards, solvents, and reaction conditions. The interface links a database constructed from data extracted from PubChem, enabling the automatic collation of information for reactions. The application's design facilitates the development of auxiliary sustainability applications, such as our Solvent Guide. As more reaction data are captured, subsequent work will include providing "intelligent" sustainability suggestions to the user.

Predicting bioactivity of antibiotic metabolites by molecular docking and dynamics.

Antibiotics enter the environment through waste streams, where they can exert selective pressure for antimicrobial resistance in bacteria. However, many antibiotics are excreted as partly metabolized forms, or can be subject to partial breakdown in wastewater treatment, soil, or through natural processes in the environment. If a metabolite is bioactive, even at sub-lethal levels, and also stable in the environment, then it could provide selection pressure for resistance. (5S)-penicilloic acid of piperacillin has previously been found complexed to the binding pocket of penicillin binding protein 3 (PBP3) of Pseudomonas aeruginosa. Here, we predicted the affinities of all potentially relevant antibiotic metabolites of ten different penicillins to that target protein, using molecular docking and molecular dynamics simulations. Docking predicts that, in addition to penicilloic acid, pseudopenicillin derivatives of these penicillins, as well as 6-aminopenicillanic acid (6APA), could also bind to this target. MD simulations further confirmed that (5R)-pseudopenicillin and 6APA bind the target protein, in addition to (5S)-penicilloic acid. Thus, it is possible that these metabolites are bioactive, and, if stable in the environment, could be contaminants selective for antibiotic resistance. This could have considerable significance for environmental surveillance for antibiotics as a means to reduce antimicrobial resistance, because targeted mass spectrometry could be required for relevant metabolites as well as the native antibiotics.

Krein support vector machine classification of antimicrobial peptides.

Antimicrobial peptides (AMPs) represent a potential solution to the growing problem of antimicrobial resistance, yet their identification through wet-lab experiments is a costly and time-consuming process. Accurate computational predictions would allow rapid in silico screening of candidate AMPs, thereby accelerating the discovery process. Kernel methods are a class of machine learning algorithms that utilise a kernel function to transform input data into a new representation. When appropriately normalised, the kernel function can be regarded as a notion of similarity between instances. However, many expressive notions of similarity are not valid kernel functions, meaning they cannot be used with standard kernel methods such as the support-vector machine (SVM). The Krein-SVM represents generalisation of the standard SVM that admits a much larger class of similarity functions. In this study, we propose and develop Krein-SVM models for AMP classification and prediction by employing the Levenshtein distance and local alignment score as sequence similarity functions. Utilising two datasets from the literature, each containing more than 3000 peptides, we train models to predict general antimicrobial activity. Our best models achieve an AUC of 0.967 and 0.863 on the test sets of each respective dataset, outperforming the in-house and literature baselines in both cases. We also curate a dataset of experimentally validated peptides, measured against Staphylococcus aureus and Pseudomonas aeruginosa, in order to evaluate the applicability of our methodology in predicting microbe-specific activity. In this case, our best models achieve an AUC of 0.982 and 0.891, respectively. Models to predict both general and microbe-specific activities are made available as web applications.

Machine learning for yield prediction for chemical reactions using in situ sensors.

Machine learning models were developed to predict product formation from time-series reaction data for ten Buchwald-Hartwig coupling reactions. The data was provided by DeepMatter and was collected in their DigitalGlassware cloud platform. The reaction probe has 12 sensors to measure properties of interest, including temperature, pressure, and colour. Colour was a good predictor of product formation for this reaction and machine learning models were able to learn which of the properties were important. Predictions for the current product formation (in terms of % yield) had a mean absolute error of 1.2%. For predicting 30, 60 and 120 min ahead the error rose to 3.4, 4.1 and 4.6%, respectively. The work here presents an example into the insight that can be obtained from applying machine learning methods to sensor data in synthetic chemistry.

Cysteine-Selective Modification of Peptides and Proteins via Desulfurative C-C Bond Formation.

The site-selective modification of peptides and proteins facilitates the preparation of targeted therapeutic agents and tools to interrogate biochemical pathways. Among the numerous bioconjugation techniques developed to install groups of interest, those that generate C(sp3 )-C(sp3 ) bonds are significantly underrepresented despite affording proteolytically stable, biogenic linkages. Herein, a visible-light-mediated reaction is described that enables the site-selective modification of peptides and proteins via desulfurative C(sp3 )-C(sp3 ) bond formation. The reaction is rapid and high yielding in peptide systems, with comparable translation to proteins. Using this chemistry, a range of moieties is installed into model systems and an effective PTM-mimic is successfully integrated into a recombinantly expressed histone.

N-Hydroxyethyl acrylamide as a functional eROP initiator for the preparation of nanoparticles under "greener" reaction conditions.

N-Hydroxyethyl acrylamide was used as a functional initiator for the enzymatic ring-opening polymerisation of ε-caprolactone and δ-valerolactone. N-Hydroxyethyl acrylamide was found not to undergo self-reaction in the presence of Lipase B from Candida antarctica under the reaction conditions employed. By contrast, this is a major problem for 2-hydroxyethyl methacrylate and 2-hydroxyethyl acrylate which both show significant transesterification issues leading to unwanted branching and cross-linking. Surprisingly, N-hydroxyethyl acrylamide did not react fully during enzymatic ring-opening polymerisation. Computational docking studies helped us understand that the initiated polymer chains have a higher affinity for the enzyme active site than the initiator alone, leading to polymer propagation proceeding at a faster rate than polymer initiation leading to incomplete initiator consumption. Hydroxyl end group fidelity was confirmed by organocatalytic chain extension with lactide. N-Hydroxyethyl acrylamide initiated polycaprolactones were free-radical copolymerised with PEGMA to produce a small set of amphiphilic copolymers. The amphiphilic polymers were shown to self-assemble into nanoparticles, and to display low cytotoxicity in 2D in vitro experiments. To increase the green credentials of the synthetic strategies, all reactions were carried out in 2-methyl tetrahydrofuran, a solvent derived from renewable resources and an alternative for the more traditionally used fossil-based solvents tetrahydrofuran, dichloromethane, and toluene.

Alchemical Free Energy Methods Applied to Complexes of the First Bromodomain of BRD4.

Accurate and rapid predictions of the binding affinity of a compound to a target are one of the ultimate goals of computer aided drug design. Alchemical approaches to free energy estimations follow the path from an initial state of the system to the final state through alchemical changes of the energy function during a molecular dynamics simulation. Herein, we explore the accuracy and efficiency of two such techniques: relative free energy perturbation (FEP) and multisite lambda dynamics (MSλD). These are applied to a series of inhibitors for the bromodomain-containing protein 4 (BRD4). We demonstrate a procedure for obtaining accurate relative binding free energies using MSλD when dealing with a change in the net charge of the ligand. This resulted in an impressive comparison with experiment, with an average difference of 0.4 ± 0.4 kcal mol-1. In a benchmarking study for the relative FEP calculations, we found that using 20 lambda windows with 0.5 ns of equilibration and 1 ns of data collection for each window gave the optimal compromise between accuracy and speed. Overall, relative FEP and MSλD predicted binding free energies with comparable accuracy, an average of 0.6 kcal mol-1 for each method. However, MSλD makes predictions for a larger molecular space over a much shorter time scale than relative FEP, with MSλD requiring a factor of 18 times less simulation time for the entire molecule space.

Effect of Oriented Electric Fields on Biologically Relevant Iron-Sulfur Clusters: Tuning Redox Reactivity for Catalysis.

Enzyme-based iron-sulfur clusters, exemplified in families such as hydrogenases, nitrogenases, and radical S-adenosylmethionine enzymes, feature in many essential biological processes. The functionality of biological iron-sulfur clusters extends beyond simple electron transfer, relying primarily on the redox activity of the clusters, with a remarkable diversity for different enzymes. The active-site structure and the electrostatic environment in which the cluster resides direct this redox reactivity. Oriented electric fields in enzymatic active sites can be significantly strong, and understanding the extent of their effect on iron-sulfur cluster reactivity can inform first steps toward rationally engineering their reactivity. An extensive systematic density functional theory-based screening approach using OPBE/TZP has afforded a simple electric field-effect representation. The results demonstrate that the orientation of an external electric field of strength 28.8 MV cm-1 at the center of the cluster can have a significant effect on its relative stability in the order of 35 kJ mol-1. This shows clear implications for the reactivity of iron-sulfur clusters in enzymes. The results also demonstrate that the orientation of the electric field can alter the most stable broken-symmetry state, which further has implications on the directionality of initiated electron-transfer reactions. These insights open the path for manipulating the enzymatic redox reactivity of iron-sulfur cluster-containing enzymes by rationally engineering oriented electric fields within the enzymes.

Kernel Methods for Predicting Yields of Chemical Reactions.

The use of machine learning methods for the prediction of reaction yield is an emerging area. We demonstrate the applicability of support vector regression (SVR) for predicting reaction yields, using combinatorial data. Molecular descriptors used in regression tasks related to chemical reactivity have often been based on time-consuming, computationally demanding quantum chemical calculations, usually density functional theory. Structure-based descriptors (molecular fingerprints and molecular graphs) are quicker and easier to calculate and are applicable to any molecule. In this study, SVR models built on structure-based descriptors were compared to models built on quantum chemical descriptors. The models were evaluated along the dimension of each reaction component in a set of Buchwald-Hartwig amination reactions. The structure-based SVR models outperformed the quantum chemical SVR models, along the dimension of each reaction component. The applicability of the models was assessed with respect to similarity to training. Prospective predictions of unseen Buchwald-Hartwig reactions are presented for synthetic assessment, to validate the generalizability of the models, with particular interest along the aryl halide dimension.

Machine-Learning-Enabled Virtual Screening for Inhibitors of Lysine-Specific Histone Demethylase 1.

A machine learning approach has been applied to virtual screening for lysine specific demethylase 1 (LSD1) inhibitors. LSD1 is an important anti-cancer target. Machine learning models to predict activity were constructed using Morgan molecular fingerprints. The dataset, consisting of 931 molecules with LSD1 inhibition activity, was obtained from the ChEMBL database. An evaluation of several candidate algorithms on the main dataset revealed that the support vector regressor gave the best model, with a coefficient of determination (R2) of 0.703. Virtual screening, using this model, identified five predicted potent inhibitors from the ZINC database comprising more than 300,000 molecules. The virtual screening recovered a known inhibitor, RN1, as well as four compounds where activity against LSD1 had not previously been suggested. Thus, we performed a machine-learning-enabled virtual screening of LSD1 inhibitors using only the structural information of the molecules.

Structural variation of protein-ligand complexes of the first bromodomain of BRD4.

The bromodomain-containing protein 4 (BRD4), a member of the bromodomain and extra-terminal domain (BET) family, plays a key role in several diseases, especially cancers. With increased interest in BRD4 as a therapeutic target, many X-ray crystal structures of the protein in complex with small molecule inhibitors are publicly available over the recent decade. In this study, we use this structural information to investigate the conformations of the first bromodomain (BD1) of BRD4. Structural alignment of 297 BRD4-BD1 complexes shows a high level of similarity between the structures of BRD4-BD1, regardless of the bound ligand. We employ WONKA, a tool for detailed analyses of protein binding sites, to compare the active site of over 100 of these crystal structures. The positions of key binding site residues show a high level of conformational similarity, with the exception of Trp81. A focused analysis on the highly conserved water network in the binding site of BRD4-BD1 is performed to identify the positions of these water molecules across the crystal structures. The importance of the water network is illustrated using molecular docking and absolute free energy perturbation simulations. 82% of the ligand poses were better predicted when including water molecules as part of the receptor. Our analysis provides guidance for the design of new BRD4-BD1 inhibitors and the selection of the best structure of BRD4-BD1 to use in structure-based drug design, an important approach for faster and more cost-efficient lead discovery.

Benzene, Toluene, and Monosubstituted Derivatives: Diabatic Nature of the Oscillator Strengths of S1 ← S0 Transitions.

For benzene, toluene, aniline, fluorobenzene, and phenol, even sophisticated treatments of electron correlation, such as MRCI and XMS-CASPT2 calculations, show oscillator strengths typically lower than experiment. Inclusion of a simple pseudo-diabatization approach to perturb the S1 state with approximate vibronic coupling to the S2 state for each molecule results in more accurate oscillator strengths. Their absolute values agree better with experiment for all molecules except aniline. When the coupling between the S1 and S2 states is strong at the S0 geometry, the simple diabatization scheme performs less well with respect to the oscillator strengths relative to the adiabatic values. However, we expect the scheme to be useful in many cases where the coupling is weak to moderate (where the maximum component of the coupling has a magnitude less than 1.5 au). Such calculations give an insight into the effects of vibronic coupling of excited states on UV/vis spectra.

Near-Ultraviolet Circular Dichroism and Two-Dimensional Spectroscopy of Polypeptides.

A fully quantitative theory of the relationship between protein conformation and optical spectroscopy would facilitate deeper insights into biophysical and simulation studies of protein dynamics and folding. In contrast to intense bands in the far-ultraviolet, near-UV bands are much weaker and have been challenging to compute theoretically. We report some advances in the accuracy of calculations in the near-UV, which were realised through the consideration of the vibrational structure of the electronic transitions of aromatic side chains.

A Machine Learning Protocol for Predicting Protein Infrared Spectra.

Infrared (IR) absorption provides important chemical fingerprints of biomolecules. Protein secondary structure determination from IR spectra is tedious since its theoretical interpretation requires repeated expensive quantum-mechanical calculations in a fluctuating environment. Herein we present a novel machine learning protocol that uses a few key structural descriptors to rapidly predict amide I IR spectra of various proteins and agrees well with experiment. Its transferability enabled us to distinguish protein secondary structures, probe atomic structure variations with temperature, and monitor protein folding. This approach offers a cost-effective tool to model the relationship between protein spectra and their biological/chemical properties.

Interaction of the maturation protein of the bacteriophage MS2 and the sex pilus of the Escherichia coli F plasmid.

One promising strategy to combat antimicrobial resistance is to use bacteriophages that attach to the sex pili produced by transmissible antimicrobial resistance (AMR) plasmids, infect AMR bacteria and select for loss of the AMR plasmids, prolonging the life of existing antimicrobials. The maturation protein of the bacteriophage MS2 attaches to the pili produced by Incompatibility group F plasmid-containing bacteria. This interaction initiates delivery of the viral genetic material into the bacteria. Using protein-protein docking we constructed a model of the F pilus comprising a trimer of subunits binding to the maturation protein. Interactions between the maturation protein and the F pilus were investigated using molecular dynamics simulations. In silico alanine scanning and in silico single-point mutations were explored, with the longer term aim of increasing the affinity of the maturation protein to other Incompatibility group pili, without reducing the strength of binding to F pilin. We report our computational findings on which residues are required for the maturation protein and F pilin to interact, those which had no effect on the interaction and the mutations which led to a stronger interaction.

Force Fields for Macromolecular Assemblies Containing Diketopyrrolopyrrole and Thiophene.

Utilizing a force-matching procedure, we parametrize new force fields systematically for large conjugated systems. We model both conjugated polymers and molecular crystals that contain diketopyrrolopyrrole, thiophene, and thieno[3,2-b]thiophene units. These systems have recently been found to have low band gaps, which exhibit high efficiency for photovoltaic devices. The equilibrium structures, forces, and energies of the building block chromophores, diketopyrrolopyrrole, thiophene, and thieno[3,2-b]thiophene computed using our parameters are comparable to those computed using the reference electronic structure method. We assess the suitability of this new force field for electronic property calculations by comparing the electronic excitation properties computed along classical and ab initio molecular dynamics trajectories. For both trajectories, we find similar distributions of TDDFT-calculated excitation energies and oscillator strengths for the building block chromophore diketopyrrolopyrrole-thieno[3,2-b]thiophene. The structural, dynamic, and electronic properties of the macromolecular assemblies built upon these chromophores are characterized. For both polymers and molecular crystals, pronounced peaks around 0° or 180° are observed for the torsions between chromophores under ambient conditions. The high planarity in these systems can promote local ordering and π-π stacking, thereby potentially facilitating charge transport across these materials. For the model conducting polymers, we found that the fluctuations in the density of states per chain per monomer is negligibly small and does not vary significantly with chains comprising 20-40 monomers. Analysis of the electron-hole distributions and the transition density matrices indicates that the delocalized length is approximately 4-6 monomers, which is in good agreement with other theoretical and experimental studies of different conducting polymers. For the molecular crystals, our investigation of the characteristic time scale of the fluctuation in the excitonic couplings shows that a low-frequency vibration below 100 cm-1 is observed for the nearest neighbors. These observations are in line with previous studies on other molecular crystals, in which low-frequency vibrations are believed to be responsible for the large modulation of the excitonic coupling. Thus, our approach and the new force fields provide a direct route for studying the structure-property relations and the molecular level origins of the high efficiency of these classes of materials.