DeepMind presented remarkably accurate predictions at the recent CASP14 protein structure prediction assessment conference. We explored network architectures incorporating related ideas and obtained the best performance with a three-track network in which information at the 1D sequence level, the 2D distance map level, and the 3D coordinate level is successively transformed and integrated. The three-track network produces structure predictions with accuracies approaching those of DeepMind in CASP14, enables the rapid solution of challenging X-ray crystallography and cryo-EM structure modeling problems, and provides insights into the functions of proteins of currently unknown structure. The network also enables rapid generation of accurate protein-protein complex models from sequence information alone, short circuiting traditional approaches which require modeling of individual subunits followed by docking. We make the method available to the scientific community to speed biological research.

Hydratases provide access to secondary and tertiary alcohols by regio- and/or stereospecifically adding water to carbon-carbon double bonds. Thereby, hydroxy groups are introduced without the need for costly cofactor recycling, and that makes this approach highly interesting on an industrial scale. Here we present the first crystal structure of a recombinant oleate hydratase originating from Elizabethkingia meningoseptica in the presence of flavin adenine dinucleotide (FAD). A structure-based mutagenesis study targeting active site residues identified E122 and Y241 as crucial for the activation of a water molecule and for protonation of the double bond, respectively. Moreover, we also observed that two-electron reduction of FAD results in a sevenfold increase in the substrate hydration rate. We propose the first reaction mechanism for this enzyme class that explains the requirement for the flavin cofactor and the involvement of conserved amino acid residues in this regio- and stereoselective hydration.

S-adenosyl-L-methionine (AdoMet)-dependent methylation is central to the regulation of many biological processes: more than 50 AdoMet-dependent methyltransferases methylate a broad spectrum of cellular compounds including nucleic acids, proteins and lipids. Common to all AdoMet-dependent methyltransferase reactions is the release of the strong product inhibitor S-adenosyl-L-homocysteine (AdoHcy), as a by-product of the reaction. S-adenosyl-L-homocysteine hydrolase is the only eukaryotic enzyme capable of reversible AdoHcy hydrolysis to adenosine and homocysteine and, thus, relief from AdoHcy inhibition. Impaired S-adenosyl-L-homocysteine hydrolase activity in humans results in AdoHcy accumulation and severe pathological consequences. Hyperhomocysteinemia, which is characterized by elevated levels of homocysteine in blood, also exhibits a similar phenotype of AdoHcy accumulation due to the reversal of the direction of the S-adenosyl-L-homocysteine hydrolase reaction. Inhibition of S-adenosyl-L-homocysteine hydrolase is also linked to antiviral effects. In this review the advantages of yeast as an experimental system to understand pathologies associated with AdoHcy accumulation will be discussed.

DeepMind presented remarkably accurate protein structure predictions at the CASP14 conference. We explored network architectures incorporating related ideas and obtained the best performance with a 3-track network in which information at the 1D sequence level, the 2D distance map level, and the 3D coordinate level is successively transformed and integrated. The 3-track network produces structure predictions with accuracies approaching those of DeepMind in CASP14, enables rapid solution of challenging X-ray crystallography and cryo-EM structure modeling problems, and provides insights into the functions of proteins of currently unknown structure. The network also enables rapid generation of accurate models of protein-protein complexes from sequence information alone, short circuiting traditional approaches which require modeling of individual subunits followed by docking. We make the method available to the scientific community to speed biological research.

The utilization of CO2 as a carbon source for organic synthesis meets the urgent demand for more sustainability in the production of chemicals. Herein, we report on the enzyme‐catalyzed para‐carboxylation of catechols, employing 3,4‐dihydroxybenzoic acid decarboxylases (AroY) that belong to the UbiD enzyme family. Crystal structures and accompanying solution data confirmed that AroY utilizes the recently discovered prenylated FMN (prFMN) cofactor, and requires oxidative maturation to form the catalytically competent prFMNiminium species. This study reports on the in vitro reconstitution and activation of a prFMN‐dependent enzyme that is capable of directly carboxylating aromatic catechol substrates under ambient conditions. A reaction mechanism for the reversible decarboxylation involving an intermediate with a single covalent bond between a quinoid adduct and cofactor is proposed, which is distinct from the mechanism of prFMN‐associated 1,3‐dipolar cycloadditions in related enzymes.

Background: Conjugative plasmid transfer is the prevalent means for spreading antibiotic resistance genes among bacteria.Results: Surface exposure of transfer protein TraM from the Gram-positive (G+) plasmid pIP501 was confirmed, and its crystal structure was solved.Conclusion: Structural relations to type IV secretion (T4S) proteins provide a novel classification scheme.Significance: The novel classification will help elucidate structure-function relationships in G+ T4S systems.

Background: Berberine bridge enzyme-like proteins are a multigene family in plants.Results: Members of the berberine bridge enzyme-like family were identified as monolignol oxidoreductases.Conclusion: Berberine bridge enzyme-like enzymes play a role in monolignol metabolism and lignin formation.Significance: Our results indicate a novel and unexpected role of berberine bridge enzyme-like enzymes in plant biochemistry and physiology.

The structure determination of major allergens is a prerequisite for analyzing surface exposed areas of the allergen and for mapping conformational epitopes. These may be determined by experimental methods including crystallographic and NMR-based approaches or predicted by computational methods. In this review we summarize the existing structural information on allergens and their classification in protein fold families. The currently available allergen-antibody complexes are described and the experimentally obtained epitopes compared. Furthermore we discuss established methods for linear and conformational epitope mapping, putting special emphasis on a recently developed approach, which uses the structural similarity of proteins in combination with the experimental cross-reactivity data for epitope prediction.