Predicting enzymatic reactions with a molecular transformer

The use of enzymes for organic synthesis allows for simplified, more economical and selective synthetic routes not accessible to conventional reagents. However, predicting whether a particular molecule might undergo a specific enzyme transformation is very difficult. Here we used multi-task transfer learning to train the molecular transformer, a sequence-to-sequence machine learning model, with one million reactions from the US Patent Office (USPTO) database combined with 32 181 enzymatic transformations annotated with a text description of the enzyme. The resulting enzymatic transformer model predicts the structure and stereochemistry of enzyme-catalyzed reaction products with remarkable accuracy. One of the key novelties is that we combined the reaction SMILES language of only 405 atomic tokens with thousands of human language tokens describing the enzymes, such that our enzymatic transformer not only learned to interpret SMILES, but also the natural language as used by human experts to describe enzymes and their mutations.

Alkylated phenanthrene distributions in black rockfish (Sebastes schlegelii) and biotransformation into hydroxylated metabolites after intragastric administration

Marine organisms such as fish are at risk of exposure to petrogenic polycyclic aromatic hydrocarbons (PAHs) released in oil spills. PAH toxicities are affected by the rates of PAH biotransformation and elimination in fish tissues, but little information on these rates is available. In this study, the biotransformation and tissue distribution of methylated phenanthrenes-typical petrogenic PAHs found after oil spills-in black rockfish (Sebastes schlegelii) were investigated. Two groups of fish were used. Each fish in one group was given a single intragastric dose of 3-methylphenanthrene, and each fish in the other group was given a single intragastric dose of 3,6-dimethylphenanthrene. The fish were allowed to recover in purified sea water for 196 h. Methylated phenanthrenes were detected in only blood and liver for 24 h after dosing, but the concentrations decreased over time and > 98% had been eliminated by the end of the study. Four mono-hydroxylated metabolites of 3,6-dimethylphenanthrene and six mono-hydroxylated metabolites of 3-methylphenanthrene were tentatively identified for the first time from tandem mass spectrometry analyses of fish bile. The concentrations of these metabolites in bile remained constant for 192 h, suggesting that the metabolites could be used as biomarkers of rockfish exposure to petrogenic PAHs.

Chemoenzymatic o-Quinone Methide Formation

Generation of reactive intermediates and interception of these fleeting species under physiological conditions is a common strategy employed by Nature to build molecular complexity. However, selective formation of these species under mild conditions using classical synthetic techniques is an outstanding challenge. Here, we demonstrate the utility of biocatalysis in generating o-quinone methide intermediates with precise chemoselectivity under mild, aqueous conditions. Specifically, α-ketoglutarate-dependent non-heme iron enzymes, CitB and ClaD, are employed to selectively modify benzylic C-H bonds of o-cresol substrates. In this transformation, biocatalytic hydroxylation of a benzylic C-H bond affords a benzylic alcohol product which, under the aqueous reaction conditions, is in equilibrium with the corresponding o-quinone methide. o-Quinone methide interception by a nucleophile or a dienophile allows for one-pot conversion of benzylic C-H bonds into C-C, C-N, C-O, and C-S bonds in chemoenzymatic cascades on preparative scale. The chemoselectivity and mild nature of this platform is showcased here by the selective modification of peptides and chemoenzymatic synthesis of the chroman natural product (-)-xyloketal D.

In vitro metabolism of naphthalene and its alkylated congeners by human and rat liver microsomes via alkyl side chain or aromatic oxidation

Mineral oils are widely applied in food production and processing and may contain polycyclic aromatic hydrocarbons (PAHs). The PAHs that may be present in mineral oils are typically alkylated, and have been barely studied. Metabolic oxidation of the aromatic ring is a key step to form DNA-reactive PAH metabolites, but may be less prominent for alkylated PAHs since alkyl substituents would facilitate side chain oxidation as an alternative. The current study investigates this hypothesis of preferential side chain oxidation at the cost of aromatic oxidation using naphthalene and a series of its alkyl substituted analogues as model compounds. The metabolism was assessed by measuring metabolite formation in rat and human liver microsomal incubations using UPLC and GC-MS/MS. The presence of an alkyl side chain markedly reduced aromatic oxidation for all alkyl-substituted naphthalenes that were converted. 1-n-Dodecyl-naphthalene was not metabolized under the experimental conditions applied. With rat liver microsomes for 1-methyl-, 2-methyl-, 1-ethyl-, and 2-ethyl- naphthalene, alkyl side chain oxidation was preferred over aromatic oxidation. With human liver microsomes this was the case for 2-methyl-, and 2-ethyl-naphthalene. It is concluded that addition of an alkyl substituent in naphthalene shifts metabolism in favor of alkyl side chain oxidation at the cost of aromatic ring oxidation. Furthermore, alkyl side chains of 6 or more carbon atoms appeared to seriously hamper and reduce overall metabolism, metabolic conversion being no longer observed with the C12 alkyl side chain. In summary, alkylation of PAHs likely reduces their chances of aromatic oxidation and bioactivation.

Carcinogenic Metabolic Activation Process of Naphthalene by the Cytochrome P450 Enzyme 1B1: A Computational Study

The metabolic activation and transformation of naphthalene by the cytochrome P450 enzyme (CYP 1B1) plays an important role in its potential carcinogenicity. The process has been explored by a quantum mechanics/molecular mechanics (QM/MM) computational method. Molecular dynamic simulations were performed to explore the interaction between naphthalene and CYP 1B1. Naphthalene involves α- and β-carbon, the electrophilic addition of which would result in different reaction pathways. Our computational results show that both additions on α- and β-carbon can generate naphthalene 1,2-oxide. The activation barrier for the addition on β-carbon is higher than that for the α-carbon by 2.6 kcal·mol^-1, which is possibly caused by the proximity between β-carbon and the iron-oxo group of Cpd I in the system. We also found that naphthalene 1,2-oxide is unstable and the O-C bond cleavage easily occurs via cellular hydronium ion, hydroxyl radical/anion; then it will convert to the potential ultimate carcinogen 1,2-naphthoquinone. The results demonstrate and inform a detailed process of generating naphthalene 1,2-oxide and new predictions for its conversion.

Structural and functional characterisation of the cytochrome P450 enzyme CYP268A2 from Mycobacterium marinum

Members of the cytochrome P450 monooxygenase family CYP268 are found across a broad range of Mycobacterium species including the pathogens Mycobacterium avium, M. colombiense, M. kansasii, and M. marinum. CYP268A2, from M. marinum, which is the first member of this family to be studied, was purified and characterised. CYP268A2 was found to bind a variety of substrates with high affinity, including branched and straight chain fatty acids (C10–C12), acetate esters, and aromatic compounds. The enzyme was also found to bind phenylimidazole inhibitors but not larger azoles, such as ketoconazole. The monooxygenase activity of CYP268A2 was efficiently reconstituted using heterologous electron transfer partner proteins. CYP268A2 hydroxylated geranyl acetate and trans-pseudoionone at a terminal methyl group to yield (2E,6E)-8-hydroxy-3,7-dimethylocta-2,6-dien-1-yl acetate and (3E,5E,9E)-11-hydroxy-6,10-dimethylundeca-3,5,9-trien-2-one, respectively. The X-ray crystal structure of CYP268A2 was solved to a resolution of 2.0 Å with trans-pseudoionone bound in the active site. The overall structure was similar to that of the related phytanic acid monooxygenase CYP124A1 enzyme from Mycobacterium tuberculosis, which shares 41% sequence identity. The active site is predominantly hydrophobic, but includes the Ser99 and Gln209 residues which form hydrogen bonds with the terminal carbonyl group of the pseudoionone. The structure provided an explanation on why CYP268A2 shows a preference for shorter substrates over the longer chain fatty acids which bind to CYP124A1 and the selective nature of the catalysed monooxygenase activity.

The Oxidation of Hydrophobic Aromatic Substrates by Using a Variant of the P450 Monooxygenase CYP101B1

The cytochrome P450 monooxygenase CYP101B1, from a Novosphingobium bacterium is able to bind and oxidise aromatic substrates but at a lower activity and efficiency than norisoprenoids and monoterpenoid esters. Histidine 85 of CYP101B1 aligns with tyrosine 96 of CYP101A1, which, in the latter enzyme forms the only hydrophilic interaction with its substrate, camphor. The histidine residue of CYP101B1 was mutated to phenylalanine with the aim of improving the activity of the enzyme for hydrophobic substrates. The H85F mutant lowered the binding affinity and activity of the enzyme for β-ionone and altered the oxidation selectivity. This variant also showed enhanced affinity and activity towards alkylbenzenes, styrenes and methylnaphthalenes. For example the rate of product formation for acenaphthene oxidation was improved sixfold to 245 nmol per nmol CYP per min. Certain disubstituted naphthalenes and substrates, such as phenylcyclohexane and biphenyls, were oxidised with lower activity by the H85F variant. Variants at H85 (A and G) designed to introduce additional space into the active site so as to accommodate these larger substrates did not improve the oxidation activity. As the H85F mutant of CYP101B1 improved the oxidation of hydrophobic substrates, this residue is likely to be in the substrate binding pocket or the access channel of the enzyme. The side chain of the histidine might interact with the carbonyl groups of the favoured norisoprenoid substrates of CYP101B1.

Potassium 3-formyl-[1,1′-biphenyl]-4-olate monohydrate

The title salt, K+·C13H9O2−·H2O, was synthesized from 5-bromosalicylaldehyde and a phenylboronic acid derivative using the Suzuki–Miyaura cross-coupling reaction (Miyaura & Suzuki, 1979). In addition to the intermolecular interactions between the charged species, two O—H...O hydrogen bonds involving the isolated water molecules further stabilize the crystal packing of the title salt leading to the formation of a three-dimensional framework structure.