Biocatalysis@TUDelft

Nature Biotechnology, Published online: 02 October 2025; doi:10.1038/s41587-025-02836-0

A protein foundation model represents protein sequence, structure and function.

Science Advances, Volume 11, Issue 40, October 2025.

Nature, Published online: 01 October 2025; doi:10.1038/s41586-025-09519-5

A two-phase machine-learning-based tool making use of high-throughput experimentation is introduced to examine the connections between chemical and protein sequence space and predict productive biocatalytic reactions among substrate and enzyme pairs.

Nature, Published online: 01 October 2025; doi:10.1038/d41586-025-03161-x

An approach that identifies, and predicts compatibility between, chemical and enzyme-sequence spaces can streamline and reduce risk in the discovery of enzymes that can catalyse a desired reaction. The strategy uses high-throughput experiments to generate data about enzyme-mediated reactions, and there is a tool that can predict compatible substrate–enzyme pairs.

Nature Chemical Biology, Published online: 06 October 2025; doi:10.1038/s41589-025-02034-8

In plants, oxidosqualene cyclases (OSCs) perform a highly complex single reaction to generate the basis of all triterpenoid diversity. Here the authors leverage genome mining and transient expression to uncover multiple evolutionary and mechanistic insights for OSCs across the plant kingdom.

Hemin immobilized within hydrophobic SBA15 to develop a novel catalyst, Alkyl-SBA15-amideHemin. This bioinspired enzyme mimic, highly stable, and recyclable heterogeneous system facilitates highly trans-diastereoselective cyclopropanation of styrene in aqueous media via a concerted carbene transfer mechanism.

Abstract

Metal-porphyrin-based biocatalysts are gaining popularity in sustainable chemical synthesis. Heme-based enzymes like myoglobins, cytochrome P450s are known for their regio- and stereo-selective cyclopropanation reactions in aqueous media. However, their practical application is challenged by complex, multistep synthesis procedures and inherent structural instability. To address these limitations, we developed a stable, enzyme-mimicking heterogeneous catalyst using porous SBA15 material. The inner core of SBA15 was functionalized with an alkyl chain and protoporphyrin IX iron (III) (hemin), resulting in alkyl-SBA15-amideHemin. Characterization through TEM, UV–vis and FT-IR spectroscopy, CHN analysis etc., confirmed the successful immobilization of hemin within the hydrophobic pores of functionalized SBA15. The catalytic performance of alkyl-SBA15-amideHemin was evaluated in the cyclopropanation reaction of styrene with ethyl diazoacetate (EDA), achieving a turnover number (TON) of 144—nearly double that of free hemin (TON 75). The catalyst exhibited high trans diastereoselectivity (∼70%) for 2-(cyclopropyl benzene) esters in aqueous media, outperforming other known heterogeneous catalysts. Its catalytic activity and selectivity were maintained over four recycling cycles, highlighting its stability and reusability. Steady-state emission studies, as well as the kinetics of the iron-carbene complex, provided insights into the catalytic mechanism. These findings suggest that immobilized hemin within a hydrophobic environment serves as a robust and selective catalyst for chemical transformations in aqueous media.

This study offers structural insights into α-galactosidase (AGal)-catalyzed hydrolysis and transglycosylation using QM/MM free-energy simulations. Hotspot identification via machine learning and free-energy profiling revealed key residues in Saccharomyces cerevisiae GH27 AGal, enabling rational, computer-aided engineering of acceptor subsites to enhance galactooligosaccharide synthesis from lactose and sucrose.

Abstract

Microbial α-galactosidases (AGals) are widely used in agriculture and food industries for degrading raffinose family oligosaccharides and synthesizing galactooligosaccharides (GOSs). While rational engineering of AGals is ongoing, limited understanding of substrate specificity and the determinants of hydrolysis and transglycosylation hinders progress. Here, we apply quantum mechanics/molecular mechanics (QM/MM) simulations to investigate the catalytic mechanism and substrate specificity of Saccharomyces cerevisiae glycoside hydrolase family 27 (GH27) AGal. The enzyme catalyzes hydrolysis via a Koshland double-displacement mechanism and cleaves linear galactomannans in an exo-mode. Free-energy calculations indicate glycosylation is the rate-determining step with a barrier (ΔG ^‡) of 17.8 kcal·mol⁻¹, consistent with experimental data. A key 4-OH···nucleophile interaction stabilizes the transition state, particularly for deglycosylation. Machine learning identified Trp188 and Phe235 at positive subsites as mutational hotspots. Six AGal variants were evaluated for in silico transglycosylation activity. Aromatic substitutions at Phe235 (F235Y and F235W) favored nucleophilic attack (NA) with sucrose, while W188A, W188R, and F235S showed low reaction barriers for lactose. The W188A variant showed improvement with a 10 kcal·mol⁻¹ decrease in ΔG ^‡, a pronounced 0.3 Å shortening of NA distance, and an increased solvent exposure of ∼500–600 Ų. These results highlight the potential of computer-aided subsite engineering to enhance AGal performance in GOS production.

The yield and turnover of in vitro natural product biosynthesis, and even its feasibility, are determined by intermolecular interactions. Reaction engineering allows to improve the performance of such complex biocatalytic systems.

Natural products are widely used as pharmaceuticals and agrochemicals, or as active ingredients in food and cosmetics. Their biosynthesis typically involves a series of enzyme-controlled reactions in dedicated liquid environments. The reconstruction of these multistep transformations under in vitro conditions bears significant potential for technical utilization. However, the concurrent operation of multiple enzymes in a single reaction flask or reactor is often associated with major challenges. Herein, the difficulties in reaching high substrate conversion and product yields with in vitro enzyme cascades are summarized. Furthermore, both established and emerging concepts for improving their performance are discussed.

ThiF-like enzymes are a widespread protein family found in disparate biosynthetic pathways. They are unified by their use of an NTP to modify a carboxylate, generating an activated species prone to nucleophilic addition. This common intermediate is then targeted by diverse nucleophiles, including persulfide or amino acid side chains, to yield a variety of structural scaffolds. This review highlights the diversity of chemical modifications made by the ThiF-like enzyme family, ranging from its involvement in the biosynthesis of universal enzyme cofactors to the formation of exotic bioactive RiPP natural product scaffolds.

ThiF-like proteins are members of the widespread E1-like enzyme superfamily. The eponymous ThiF enzyme was first described in thiamin biosynthesis as part of Escherichia coli's primary metabolism, and homologous proteins have been subsequently discovered in secondary metabolism. These ThiF-like enzymes are united in their defining ability to perform nucleotidylation of a carboxyl group to generate an activated, electrophilic intermediate, a feature it shares with the structurally related ubiquitin-activating enzymes. From here, an array of different nucleophiles are used across distinct biosynthetic pathways to yield diverse structural scaffolds. In this review, we discuss various ThiF-like enzymes that perform nucleotidylation to facilitate a diverse array of interesting and rare chemistry on different types of substrates, as well as showcase some of the shared structural features.

The marine bacterial flavoenzymes Clz9 and Tcz9 can process cannabigerolic acid (CBGA) to the minor cannabinoid, cannabichromenic acid (CBCA), however, the mechanistic details of this extrinsic transformation are still obscure. Here, we report a thorough analysis of CBCA-formation by Clz9 and Tcz9 through high-resolution crystallographic characterization, biochemical analysis, and spectroscopic interrogation. Our work reveals that Clz9 and Tcz9 use different biochemical mechanisms from Cannabis cyclases and each other in their production of CBCA. Collection of a high-resolution substrate-bound structure, the first for any cannabinoid cyclase, provides key insights into how active site architecture affects substrate binding and stereoselectivity. Engineering approaches improve the stereoselectivity of CBCA formation by Clz9 and Tcz9, providing access to (R) and (S)-CBCA. Collectively, our work advances understanding of enzymatic cannabinoid formation and cements Clz9 and Tcz9 as two unique members of the BBE-like enzyme family with encouraging potential for biocatalytic cannabinoid production applications.

Proximity labeling (PL) identifies endogenous proteins in specific subcellular regions. APEX2 is a peroxidase that enables PL with high temporal resolution by oxidizing biotin phenol (BP) into a radical that tags nearby proteins. However, the BP radical has a relatively large diffusion radius, limiting spatial resolution. Replacing the phenol in BP with nitrophenol (NP) could potentially increase spatial resolution by generating shorter-lived radicals, but APEX2 cannot efficiently oxidize phenols with high redox potentials. Here, we report the directed evolution of APOX, a quadruple mutant of APEX2 with a higher reduction potential that exhibits 6-fold and 2.5-fold faster oxidation of NPs and BP, respectively. Using APOX with a membrane-permeable alkyne-NP probe, we demonstrate PL in living mammalian cells, including proteomic mapping with excellent subcellular compartment specificity. APOX expands peroxidase PL by accessing high redox potential probes, opening opportunities to further tune the diffusion radius and enhance the tagging of biomolecules beyond proteins.

The 2-oxindole class of heterocycles are privileged structural components in natural products and biologically active compounds. One of the most attractive methods for accessing 2- oxindoles is through direct oxidation of indoles, but current methods rely on the use of chemical oxidizing agents that lead to the generation of harmful waste products or biocatalytic methods using enzymes with limited substrate scope. Herein, we describe the development and application of a general biocatalytic platform for oxidative rearrangement of indoles using enzymatic halide recycling with vanadium-dependent haloperoxidases (VHPOs) facilitated by a catalytic quantity of halide salt and hydrogen peroxide as the terminal oxidant. This catalytic system is effective for the oxidative rearrangement of indoles into 2-oxindoles and 2-spirooxindoles. The developed protocol has been applied in multi-enzymatic and chemoenzymatic synthesis, late-stage functionalization of biologically active molecules, tryptophan-selective peptide modification, and gram-scale syntheses of coerulescine and horsfiline.

Enzymes catalyze complex chemical transformations with remarkable efficiency and selectivity, yet their atomistic mechanisms remain challenging to capture because conventional simulations trade accuracy for efficiency. Here we introduce a reactive machine learning/molecular mechanics (ML/MM) framework that bridges quantum chemistry with long-timescale sampling, enabling direct exploration of enzymatic transition states and free-energy landscapes. Coupled with metadynamics, this approach achieves nanosecond sampling of bond-forming reactions and quantitatively predicts activation barriers, mutational effects, and stereoselectivity. Applied to Diels–Alderases, the framework not only reproduces experimental activity andendo/exo preferences with sub-kcal/mol accuracy but also uncovers how pathway dynamics and local electrostatics preorganize substrates for selective outcomes. By uniting reactivity, conformational dynamics, and predictive power, this work establishes reactive ML/MM as a broadly applicable strategy for mechanistic enzymology and a foundation for the rational design of new biocatalysts.

Tryptophan is an essential amino acid, and its fermentation-based synthesis is well-described with double-digit titers. Therefore, tryptophanase may serve as the gateway enzyme towards indole-derived bio-based molecules. They are widespread PLP-dependent enzymes with a conserved active site. Even though their function and structure are well described, the number of characterized tryptophanases is rather low. Here, we characterized a panel of tryptophanases mined from public databases, aimed to explore their industrial amenability. We used the machine learning method variational autoencoding to visualize the tryptophanase sequence space and ensure that the natural sequence diversity is covered in our panel. After initial experiments, the gathered information was used to identify 2 additional regions of interest in the visualized latent space. The solubly expressed enzymes (18 of 21) were biochemically characterized for optimal reaction conditions and kinetic parameters. The characterized panel encompassed a diverse set of tryptophanases with distinct optimal conditions, facilitating the selection of an appropriate candidate. The most suitable tryptophanase, PreTIL, was used in an enzymatic cascade towards biobased indigo synthesis. Optimal cosubstrate and enzyme concentrations were determined by Design of Experiments, resulting in 3.0 ± 0.8 mM indigo formation in situ. Though this titer is comparable to reports on indigo synthesis through fermentation, a fundamental phenomenon appears to halt biocatalytic indigo synthesis.

A cytochrome P450 enzyme from the thermophile Thermosporothrix hazakensis was characterised. We modified the heme domain of the enzyme through protein engineering to enable it to function as a peroxygenase biocatalyst. We demonstrated the oxidation of fatty acids and aromatic compounds and identified the metabolites.

ABSTRACT

The cytochrome P450 monooxygenase enzymes (CYPs) of the CYP102 family are versatile, self-sufficient biocatalysts. The archetypal example is CYP102A1 (P450_BM3) from the bacterium Bacillus (Priestia) megaterium, and variants of this enzyme can oxidise many substrates with high activity and selectivity. However, this enzyme has relatively low thermal stability. Here, we identify and characterise a CYP102 family enzyme from the moderately thermophilic bacterium Thermosporothrix hazakensis. We were able to produce this enzyme using Escherichia coli and demonstrate the in vivo oxidation of fatty acids. However, the activity of the isolated holoenzyme was low, so we generated a peroxygenase variant by introducing the E278Q and T279E mutations into the heme domain (‘HazakQE’). This isolated variant was able to catalyse the oxidation of a range of substrates using hydrogen peroxide as the oxidant. The product distributions arising from fatty acid oxidation using the holoprotein monooxygenase and heme domain peroxygenase variants of this enzyme were broadly similar to those obtained with P450_BM3. For fatty acids, the oxidation occurred predominantly at the ω-1 through to ω-3 positions. Styrene was epoxidised and tetralone hydroxylated at the benzylic carbon. The oxidation of 1-methoxynaphthalene generated the dimeric Russig's blue, enabling colorimetric assays of the enzyme activity. Although the HazakQE heme peroxygenase was more thermostable than the mesophilic CYP199A4 enzyme from Rhodopseudomonas palustris, it was not more resistant to heating than the heme domain of P450_BM3. These peroxygenase variants offer a simple platform for metabolite identification and biocatalysts for oxidation reactions, which could be enhanced through protein engineering.

In this study, we confirm that PlqH is the hydroxylase operating in plastoquinone biosynthesis in photosynthetic cyanobacteria (Cyanobacteriia). Our phylogenetic analyses demonstrate that cyanobacterial PlqH homologues originated from hydroxylases involved in ubiquinone biosynthesis in bacteria. Plastoquinone production in Escherichia coli was achieved by expressing two heterologous genes, one of which was PlqH. However, plastoquinone was unable to replace ubiquinone in several cellular processes in E. coli, suggesting that the structure of quinones influences their function.

Isoprenoid quinones constitute a class of redox lipids that are indispensable for electron transfer in a variety of cellular functions. For instance, plastoquinone, an integral component of plants, algae and Cyanobacteriota, plays a pivotal role in photosynthesis. Isoprenoid quinones are biosynthesised via evolutionary-related pathways, in which some steps are still incompletely characterised. In this study, we confirm the identity of the PlqH enzyme, a flavin-dependent monooxygenase (FMO) conserved in photosynthetic cyanobacteria, which possesses a regioselective hydroxylase activity required for plastoquinone biosynthesis. Phylogenetic analyses demonstrate that cyanobacterial PlqH homologues originated from FMOs involved in bacterial ubiquinone biosynthesis. The synthesis of plastoquinone by Escherichia coli was achieved by expressing two heterologous genes in a genetically engineered strain, which was optimised to produce plastoquinone levels comparable to those of natural ubiquinone. However, plastoquinone was unable to replace ubiquinone in several cellular processes in E. coli, suggesting that fine structural and thermodynamic constraints both play a significant role in the function of quinones.

Nature Chemistry, Published online: 02 October 2025; doi:10.1038/s41557-025-01954-1

Pseudokinases are non-canonical protein-kinase-like proteins deficient in kinase activity, few of which have enzymatic activity that differs from phosphorylation. Now a pseudokinase-enabled cyclization activity for the biosynthesis of ribosomally synthesized and post-translationally modified peptides has been observed. Here pseudokinases can catalyse a Michael addition for (ene)thioether crosslinking through a sandwich-like substrate-assisted process.