Faraday Discussions最新文献

Biocatalysis is a rapidly evolving field with increasing impact in organic synthesis, chemical manufacturing and medicine. The Faraday Discussion reflected the current state of biocatalysis, covering the design of de novo enzymatic activities, but especially methods for the improvement of enzymes targeting a broad range of applications (i.e., hydroxylations by P450 monooxygenases, enzymatic deprotection of organic compounds under mild conditions, synthesis of chiral intermediates, plastic degradation, silicone polymer synthesis, and peptide synthesis). Central themes have been how to improve an enzyme using methods of rational design and directed evolution, informed by computer modelling and machine learning, and the incorporation of new catalytic functionalities to create hybrid and artificial enzymes.

The electrochemical leaf enables the electrification and control of multi-enzyme cascades by exploiting two discoveries: (i) the ability to electrify the photosynthetic enzyme ferredoxin NADP⁺ reductase (FNR), driving it to catalyse the interconversion of NADP⁺/NADPH whilst it is entrapped in a highly porous, metal oxide electrode, and (ii) the evidence that additional enzymes can be co-entrapped in the electrode pores where, through one NADP(H)-dependent enzyme, extended cascades can be driven by electrical connection to FNR, via NADP(H) recycling. By changing a critical active-site tyrosine to serine, FNR’s exclusivity for NADP(H) is swapped for unphosphorylated NAD(H). Here we present an electrochemical study of this variant FNR, and show that in addition to the intended inversion of cofactor preference, this change to the active site has altered FNR’s tuning of the flavin reduction potential, making it less reductive. Exploiting the ability to monitor the variant’s activity with NADP(H) as a function of potential has revealed a trapped intermediate state, relieved only by applying a negative overpotential, which allows catalysis to proceed. Inhibition by NADP⁺ (very tightly bound) with respect to NAD(H) turnover was also revealed and interestingly, this inhibition changes depending on the applied potential. These findings are of critical importance for future exploitation of the electrochemical leaf.

The prenylated-flavin mononucleotide-dependent decarboxylases (also known as UbiD-like enzymes) are the most recently discovered family of decarboxylases. The modified flavin facilitates the decarboxylation of unsaturated carboxylic acids through a novel mechanism involving 1,3-dipolar cyclo-addition chemistry. UbiD-like enzymes have attracted considerable interest for biocatalysis applications due to their ability to catalyse (de)carboxylation reactions on a broad range of aromatic substrates at otherwise unreactive carbon centres. There are now ∼35 000 protein sequences annotated as hypothetical UbiD-like enzymes. Sequence similarity network analyses of the UbiD protein family suggests that there are likely dozens of distinct decarboxylase enzymes represented within this family. Furthermore, many of the enzymes so far characterized can decarboxylate a broad range of substrates. Here we describe a strategy to identify potential substrates of UbiD-like enzymes based on detecting enzyme-catalysed solvent deuterium exchange into potential substrates. Using ferulic acid decarboxylase (FDC) as a model system, we tested a diverse range of aromatic and heterocyclic molecules for their ability to undergo enzyme-catalysed H/D exchange in deuterated buffer. We found that FDC catalyses H/D exchange, albeit at generally very low levels, into a wide range of small, aromatic molecules that have little resemblance to its physiological substrate. In contrast, the sub-set of aromatic carboxylic acids that are substrates for FDC-catalysed decarboxylation is much smaller. We discuss the implications of these findings for screening uncharacterized UbiD-like enzymes for novel (de)carboxylase activity.

Data-driven techniques for establishing quantitative structure property relations are a pillar of modern materials and molecular discovery. Fuelled by the recent progress in deep learning methodology and the abundance of new algorithms, it is tempting to chase benchmarks and incrementally build ever more capable machine learning (ML) models. While model evaluation has made significant progress, the intrinsic limitations arising from the underlying experimental data are often overlooked. In the chemical sciences data collection is costly, thus datasets are small and experimental errors can be significant. These limitations of such datasets affect their predictive power, a fact that is rarely considered in a quantitative way. In this study, we analyse commonly used ML datasets for regression and classification from drug discovery, molecular discovery, and materials discovery. We derived maximum and realistic performance bounds for nine such datasets by introducing noise based on estimated or actual experimental errors. We then compared the estimated performance bounds to the reported performance of leading ML models in the literature. Out of the nine datasets and corresponding ML models considered, four were identified to have reached or surpassed dataset performance limitations and thus, they may potentially be fitting noise. More generally, we systematically examine how data range, the magnitude of experimental error, and the number of data points influence dataset performance bounds. Alongside this paper, we release the Python package NoiseEstimator and provide a web-based application for computing realistic performance bounds. This study and the resulting tools will help practitioners in the field understand the limitations of datasets and set realistic expectations for ML model performance. This work stands as a reference point, offering analysis and tools to guide development of future ML models in the chemical sciences.

The structure of ionic liquids (ILs), which a decade or two ago was the subject of polite but heated debate, is now much better understood. This has opened opportunities to ask more sophisticated questions about the role of structure in transport, the structure of systems with ions that are not prototypical, and the similarity between ILs and other dense ionic fluids such as the high-temperature inorganic molten salts; let alone the fact that new areas of research have emerged that sprung from our collective understanding of the structure of ILs such as the deep eutectic solvents, the polymerized ionic liquids, and the zwitterionic liquids. Yet, our understanding of the structure of prototypical ILs may not be as complete as we think it to be, given that recent experiments appear to show that in some cases there may be more than one liquid phase resulting in liquid–liquid (L–L) phase transitions. This article presents a perspective on what we think are key topics related to the structure and structural dynamics of ILs and to some extent high-temperature molten salts.

In our recent study, we demonstrated using ⁷Li solid-state Nuclear Magnetic Resonance (ssNMR) and single-crystal X-ray diffraction that the cathode LiFeV₂O₇ possesses a defect associated with the positioning of vanadium atoms. We proposed that this defect could be the source of extra signals detected in the ⁷Li spectra. In this context, we now apply density functional theory (DFT) calculations to assign the experimental signals observed in ⁷Li NMR spectra of the pristine sample. The calculation results are in strong agreement with the experimental observations. DFT calculations are a useful tool to interpret the observed paramagnetic shifts and understand how the presence of disorder affects the spectra behavior through the spin-density transfer processes. Furthermore, we conducted a detailed study of the lithiated phase combining operando synchrotron powder X-ray diffraction (SPXRD) and DFT calculations. A noticeable volume expansion is observed through the first discharge cycle which likely contributes to the enhanced lithium dynamics in the bulk material, as supported by previously published ssNMR data. DFT calculations are used to model the lithiated phase and demonstrate that both iron and vanadium participate in the redox process. The unusual electronic structure of the V⁴⁺ exhibits a single electron on the 3d_xy orbital perpendicular to the V–O–Li bond being a source of a negative Fermi contact shift observed in the ⁷Li NMR of the lithiated phase.

Computational crystal structure prediction (CSP) is an increasingly powerful technique in materials discovery, due to its ability to reveal trends and permit insight across the possibility space of crystal structures of a candidate molecule, beyond simply the observed structure(s). In this work, we demonstrate the reliability and scalability of CSP methods for small, rigid organic molecules by performing in-depth CSP investigations for over 1000 such compounds, the largest survey of its kind to-date. We show that this highly-efficient force-field-based CSP approach is superbly predictive, locating 99.4% of observed experimental structures, and ranking a large majority of these (74%) as among the most stable possible structures (to within uncertainty due to thermal effects). We present two examples of insights such large predicted datasets can permit, examining the space group preferences of organic molecular crystals and rationalising empirical rules concerning the spontaneous resolution of chiral molecules. Finally, we exploit this large and diverse dataset for developing transferable machine-learned energy potentials for the organic solid state, training a neural network lattice energy correction to force field energies that offers substantial improvements to the already impressive energy rankings, and a MACE equivariant message-passing neural network for crystal structure re-optimisation. We conclude that the excellent performance and reliability of the CSP workflow enables the creation of very large datasets of broad utility and explanatory power in materials design.

Epoxide hydrolase StEH1, from potato, is similar in overall structural fold and catalytic mechanism to haloalkane dehalogenase DhlA from Xanthobacter autotrophicus. StEH1 displays low (promiscuous) hydrolytic activity with (2-chloro)- and (2-bromo)ethanebenzene producing 2-phenylethanol. To investigate possibilities to amplify these very low dehalogenase activities, StEH1 was subjected to targeted randomized mutagenesis at five active-site amino acid residues and the resulting protein library was challenged for reactivity towards a bait chloride substrate. Enzymes catalyzing the first half-reaction of a hydrolytic cycle were isolated following monovalent phage display of the mutated proteins. Several StEH1 derived enzymes were identified with enhanced dehalogenase activities.

One of the main limitations in supporting experimental characterization of Al siting/pairing via modelling is the high computational cost of ab initio calculations. For this reason, most works rely on static or very short dynamical simulations, considering limited Al pairing/siting combinations. As a result, comparison with experiment suffers from a large degree of uncertainty. To alleviate this limitation we have developed neural network potentials (NNPs) which can dynamically sample across broad configurational and chemical spaces of sodium-form aluminosilicate zeolites, preserving the level of accuracy of the ab initio (dispersion-corrected metaGGA) training set. By exploring a wide range of Al/Na arrangements and a combination of experimentally relevant Si/Al ratios, we found that the ²³Na NMR spectra of dehydrated high-silica CHA zeolite offer an opportunity to assess the distribution and pairing of Al atoms. We observed that the ²³Na chemical shift is sensitive not only to the location of sodium in 6- and 8MRs, but also to the Al–Si_n–Al sequence length. Furthermore, neglect of thermal and dynamical contributions was found to lead to errors of several ppm, and has a profound influence on the shape of the spectra and the dipolar coupling constants, thus necessitating the long-term dynamical simulations made feasible by NNPs. Finally, we obtained a predictive regression model for the ²³Na chemical shift in CHA (Si/Al = 35, 17, 11) that circumvents the need for expensive NMR density functional calculations and can be easily extended to other zeolite frameworks. By combining NNPs and regression methods, we can expedite the simulations of NMR properties and capture the effect of dynamics on the spectra, which is often overlooked in computational studies despite its clear manifestation in experimental setups.

The pair distribution function (PDF) is an important metric for characterising structure in complex materials, but it is well known that meaningfully different structural models can sometimes give rise to equivalent PDFs. In this paper, we discuss the use of model likelihoods as a general approach for discriminating between such homometric structure solutions. Drawing on two main case studies—one concerning the structure of a small peptide and the other amorphous calcium carbonate—we show how consideration of model likelihood can help drive robust structure solution, even in cases where the PDF is particularly information-poor. The obvious thread of these individual case studies is the potential role for machine-learning approaches to help guide structure determination from the PDF, and our paper finishes with some forward-looking discussion along these lines.