Biocatalysis@TUDelft

Antibody junctional diversity in jawed vertebrates arose from RAG transposon domestication, generating CDR3-focused V(D)J recombination with lymphocyte specificity--a constrained system limiting engineered antibody diversity. To overcome these limitations, we developed ARESEC (Antibody Reprogram and Expression System in Engineered Cells), a synthetic platform that enables programmable V(D)J recombination in non-lymphoid HEK293T cells. Through engineered recombination signal sequences (RSS) and optimized RAG1/2 expression, ARESEC enables RSS-guided DNA recombination across all three complementarity-determining regions (CDR1/2/3). Co-expression of terminal deoxynucleotidyl transferase (TdT) enhanced junctional sequence diversity by 41.5-86.3% across 3 CDRs. The platform enables native IgG production by coupling mammalian surface display with FACS-based functional screening. Using clinical-grade antibodies (Nivolumab and Durvalumab) as high-affinity starting scaffolds, we achieved further efficient affinity maturation through focused CDR diversification (library complexity >103). From these diversification libraries, we identified optimized variants including the anti-PD-L1 Dur1 with 2.01-fold enhanced binding, demonstrating the systems capacity for rapid antibody optimization from minimal diversity sampling. Collectively, ARESEC establishes a synthetic paradigm that transcends natural V(D)J constraints, generating multi-CDR diversification in non-lymphoid cells to enable rapid discovery of affinity-matured antibodies, effectively bridging immune evolution with modern antibody engineering demands.

In fungal bioluminescence, light production generates oxyluciferin, which can inhibit the luciferase enzyme. This study shows that the new enzyme caffeylpyruvate hydrolase (CPH) recycles oxyluciferin back into metabolic intermediates, relieving inhibition and sustaining light emission. This recycling step helps maintain efficient bioluminescence in plants and represents a potential target for pathway optimisation.

In the fungal bioluminescence pathway, caffeylpyruvate (fungal oxyluciferin) is recycled to caffeic and pyruvic acids through hydrolysis. The reaction is thought to be catalysed by a putative caffeylpyruvate hydrolase, possibly encoded by the cph gene, a member of the biosynthetic gene cluster encoding bioluminescence. Here, we experimentally confirmed the role of CPH as caffeylpyruvate hydrolase in vitro and in vivo. We show that oxyluciferin is an inhibitor of the bioluminescence reaction and that CPH orthologues from various bioluminescent fungi differ in their ability to hydrolyse oxyluciferin. This suggests a possible role of caffeylpyruvate hydrolases in bioluminescent fungi and highlights CPH as a target for directed evolution to develop autoluminescent reporter tools.

Nature Chemical Biology, Published online: 21 May 2026; doi:10.1038/s41589-026-02213-1

Cells reduce the S–S bond in cystine to obtain cysteine. A robust alternative pathway, initiating with enzymatic cleavage of a cystine C–S bond followed by nonenzymatic reactions, is here shown to support survival of reductase-deficient cells.

We report the first stereoselective mechano-biocatalytic reduction of prochiral ketones with Baker's Yeast under solvent minimized ball milling conditions. Both dry and fresh yeast afford chiral alcohols with up to >99% ee and substantially lower E factors. Scale up via twin-screw extrusion demonstrates a sustainable, operationally simple asymmetric platform.

Mechanochemistry and biocatalysis offer complementary routes to greener synthesis, yet their combination remains challenging, as enzymes and whole-cell systems often lose activity under low-solvent and mechanically demanding conditions. Here we demonstrate that Baker's yeast, a widely used whole-cell biocatalyst, retains catalytic activity and high stereoselectivity under ball-milling conditions. This enables a highly stereoselective whole-cell mechano-biocatalytic reduction of α-keto esters and aryl–alkyl ketones, affording chiral alcohols with enantiomeric excesses up to >99% under solvent-minimized conditions. Both lyophilized and fresh yeast operate effectively, delivering E-factors substantially lower than those obtained under conventional aqueous conditions (>1000 in solution vs. 104–76 under mechano-biocatalytic conditions). The process is scalable, as demonstrated by translation to a twin-screw extrusion protocol, and significantly reduces waste compared with traditional whole-cell reductions. These results establish whole-cell mechanochemical biocatalysis as a practical and sustainable platform for asymmetric synthesis.

Supervised machine learning was used to identify wild-type imine reductases for challenging arylamine reductive aminations. A K-Nearest Neighbors model enriched highly active enzymes from untested sequence space, delivering improved catalysts for seven reactions and revealing a link between activity and evolutionary plausibility.

Reductive amination with arylamines is a valuable but challenging transformation in pharmaceutical biocatalysis. To address this gap area in the biocatalytic toolkit, supervised machine learning was used to identify untested enzymes capable of catalyzing these reactions. A dataset comprising 269 enzymes tested on 9 different reductive amination reactions was constructed and used to train a K-Nearest-Neighbors regression model. The model was applied to generate predictions for 6588 untested wild-type enzymes, from which 42 top candidates were identified. It was observed that these top 42 predicted enzymes were substantially enriched in active variants compared to the training dataset, and improved variants for 7 of the 9 reactions were identified. It was also found that these top-predicted enzymes were significantly enriched in sequences with lower ESM1b pseudoperplexity (a measure of evolutionary plausibility) despite the model not being trained on such information. The results demonstrate that supervised machine learning is an effective tool for identifying highly active wild-type biocatalysts for difficult transformations and suggest that protein language models may also be useful in identifying such enzymes.

Bacillus subtilis PdaC is the first peptidoglycan deacetylase able to act in reverse, acetylating a glucosamine oligomer (chitopentaose) with a pattern of acetylation mirroring the natural deacetylation reaction with unique final regioselectivity, acetylating all but the reducing end of D5.

ABSTRACT

Partially acetylated chitosan oligosaccharides play key biological roles and have numerous applications, with their bioactivities being strongly influenced by the acetylation pattern of their glucosamine monomers. However, chemical de-N-acetylation reactions of chitooligosaccharides (COS) lack the regioselectivity required to control these patterns. Chitin deacetylases (CDAs) exhibit diverse deacetylation patterns, becoming powerful biocatalysts for the tailored production of functional partially-acetylated chitosan oligosaccharides. However, known CDAs do not cover all necessary patterns. CDAs have also been shown to catalyze the reverse reaction by acetylating glucosamine oligomers in the presence of excess acetate, which complements the portfolio of sequence-defined partially acetylated chitosan oligomers (paCOS). In this study, the regioselectivity of the peptidoglycan de-N-acetylase BsPdaC from Bacillus subtilis during the N-acetylation of chitopentaose (GlcNH₂)₅ (D5) was investigated. Remarkably, the enzyme is able to acetylate all but the reducing end GlcNH₂ units, mirroring the natural deacetylation reaction, giving access to new patterns of acetylation not found with known CDAs. The enzyme primarily produced A3D2 and A4D1 (A: GlcNAc, D: GlcNH₂), and MS sequencing revealed a novel acetylation pattern (DAAAD and AAAAD). Different conditions were tested, but the product profile remained unchanged. This new specificity, producing AAAAD, expands the diversity of sequence-defined paCOS to evaluate their bioactivities.

Enzymes catalyze reactions with remarkable specificity and can unlock recalcitrant feedstocks that are dilute, complex, and variable in their constituent molecules. While characterized enzymatic reactions cover a wide range of chemistries, there are an undetermined number of cryptic activities for every known one. These cryptic activities can be elicited through rational design, adaptive laboratory evolution, and increasingly, generative models of proteins. However, prior to tuning a catalyst one must efficiently predict viable novel reactions. In this work we leverage the growing amount of mechanistic enzyme information, specifically the Mechanism and Catalytic Site Atlas, to construct a set of reaction rules that can meet this demand. By explicitly utilizing mechanistic information, the rule sets developed here more accurately identify molecular structures required for catalysis compared to existing curated and heuristically constructed rules. The 899 Distilled rules are constructed directly from characterized mechanisms and cover 62.5% of reactions from Rhea. The Learned rule set is generated from a classifier trained on mechanistic data, allowing full coverage of Rhea and precise identification of mechanism-required atoms (ROC-AUC = 0.98). Additionally, our Learned rules exhibit a more favorable tradeoff between novelty and feasibility and provide users with fine-grained control over this tradeoff. The rules are compatible with all SMARTS-based reaction network expansion and retrosynthesis software.

tRNA wobble base uridines are heavily modified to influence anticodon:codon base pairing and tune the structure of the anticodon stem loop for efficient and accurate translation. Both Gram-positive and negative bacteria, as well as certain archaea, modify wobble uridines to contain either a 5-carboxymethylaminomethyl (cmnm5) or a 5-methylaminomethyl (mnm5) moiety, as well as in some instances a 2-thio (s2) moiety. Bacteria utilize the conserved MnmEG complex to produce cmnm5U, which is further modified in some tRNAs to mnm5U. The steps to synthesize the latter are catalyzed by non-orthologous enzymes in distantly related bacteria. Escherichia coli utilizes a single bifunctional enzyme, MnmC, to both demodify cmnm5U to nm5U and subsequently methylate nm5U to mnm5U, while Bacillus subtilis relies on the radical SAM (rSAM) enzyme MnmL, followed by the stand-alone MnmM methylase for mnm5U production. It was previously noted that E. coli and related bacteria that encode MnmC can also contain homologs of MnmL. As an E. coli mnmC mutant accumulates cmnm5U, the function of the MnmL homolog in this organism, YhcC, was unknown. Here, we find YhcC is necessary for cmnm5s2U demodification in vivo under anaerobic growth, and that the same MnmC-mediated demodification activity requires O2 and only occurs under aerobic growth. In vitro reaction experiments demonstrate that purified YhcC, reconstituted to its [4Fe-4S] form, is able to bind tRNA and catalyze the nm5s2U-tRNA synthesis from cmnm5s2U-tRNA. Together, these results define the heretofore unknown biochemistry of the E. coli rSAM enzyme YhcC and show this enzyme replaces MnmC under anaerobic conditions to carry out the synthesis of nm5s2U. These parallel tRNA modification pathways highlight how E. coli has adapted to maintain biosynthesis of a critical wobble base modification under both aerobic and anaerobic growth. General audience summaryTransfer RNA (tRNA) molecules serve as critical components in protein synthesis through their direct interaction with the ribosome and messenger RNA (mRNA). During protein synthesis, tRNA molecules utilize an anticodon composed of three bases that recognize a complementary mRNA codon. At the 3CCA end of tRNA the amino acid corresponding to the anticodon sequence is incorporated into the growing polypeptide chain. While canonical Watson-Crick base pairing, pairing between U:A and G:C, occurs between the second and third bases of the anticodon and the corresponding bases of the codon, there is increased flexibility in the ribosome at the first position of the anticodon. This results in non-canonical base pairing and allows a single tRNA molecule to recognize multiple codons. Modification at this site restricts or enhances this "wobble" base pairing. Bacteria often install mnm5s2U34 to various tRNAs to facilitate proper and efficient translation. In E. coli, three proteins are responsible for this hypermodification pathway: MnmE and MnmG convert s2U to cmnm5s2U, while MnmC subsequently removes the carboxymethyl group to generate nm5s2U and methylates this intermediate into the ultimate product, mnm5s2U. We found that an additional enzyme, YhcC, is required for mnm5s2U synthesis in anaerobic conditions, specifically at the cmnm5s2U demodification step. Previous studies showed that MnmC-catalyzed demodification is dependent on FAD for the initial oxidation of tRNA substrate, generating FADH2. We propose that FADH2 recycling is dependent on O2 to regenerate FAD, allowing multiple catalytic turnovers. We show that, in vitro, O2 is required for cmnm5s2U demodification by MnmC, supporting a new model of mnm5s2U synthesis with alternative aerobic and anaerobic routes. Conservation of multiple enzymes that perform similar chemistry, albeit under differing environmental conditions, highlights the importance of maintaining wobble base modifications as well as the ability of bacteria to adapt to their surroundings.

A substrate-guided strategy was developed to IRED-M5 mutant for tertiary amine synthesis. Pocket remodeling of M5 variant enabled efficient two-step cascade or direct reductive amination of secondary amines, delivering diverse N-heterocyclic tertiary amines in up to 99% conversion across 34 examples.

ABSTRACT

The enzymatic synthesis of N‑heterocyclic tertiary amines remains underdeveloped, largely due to the time and resource demands of traditional enzyme engineering methods. Guided by key substrate features including substrate size and the positive charge of the imine intermediate, we employed a straightforward and accessible computational design to engineer the imine reductase variant M5. We successfully constructed the combinatorial mutant M5-W234I-M203A-I149G, which exhibits significantly enhanced catalytic activity toward N‑heterocyclic secondary amines, with improved substrate binding and kinetics. The variant maintained high activity across a range of secondary amines (30 mM), whether generated via a two‑step synthesis or direct use. Our results indicate that interactions between residues are a critical consideration in the construction of highly active combinatorial mutants. This work provides an accessible design strategy and establishes a rational design framework for imine reductases engineering.

The pseudokinase-like nitrile-forming enzyme CalN was identified through comparative analysis of the biosynthetic gene clusters (BGCs) for calyculin derivatives. In vitro reactions of the recombinant CalN revealed that CalN converts calyculinamide A into calyculin A through a substrate-adenylation mechanism in an ATP- and Mg²⁺-dependent manner. The pseudokinase-like nitrile-forming enzyme CalN was identified through comparative analysis of the biosynthetic gene clusters for calyculin derivatives. CalN converts calyculinamide A into calyculin A through a substrate-adenylation mechanism in an ATP-dependent manner.

ABSTRACT

Several pseudokinases, previously regarded as dead enzymes due to the lack of catalytic residues, catalyze nucleotidylation. While they often utilize macromolecular substrates such as proteins and RNAs in primary metabolism, those acting on non-macromolecules in specialized metabolisms are limited. Calyculin A, a cytotoxic natural product produced by an uncultured sponge symbiont, possesses a unique nitrile group at the end of its tetraene tail. Even though its biosynthetic gene cluster (BGC) has been identified, the enzyme responsible for nitrile formation remains unknown. Herein, through a comparative analysis of the BGCs for calyculin derivatives in symbiotic bacteria from distinct sources, we identified a novel nitrile-forming enzyme, CalN. While CalN lacks sequence homology with other known nitrile-forming enzymes, it is structurally similar to pseudokinases. In vitro enzymatic reactions demonstrated that CalN specifically catalyzes nitrile formation through the adenylation of an amide substrate, calyculinamide A. In silico analyses and mutational experiments showed that CalN's structure features a unique insertion that plays critical roles in ATP recognition and the spatial coordination of catalytic residues. This study not only identifies a new family of nitrile-forming enzymes but also expands the variety of chemical reactions mediated by pseudokinases in nature.

Engineering an acyl-CoA ligase shifts thiol specificity toward inexpensive synthetic thiols such as N-acetylcysteamine (SNAC), enabling enzymatic generation of acyl-SNAC starter units that support polyketide biosynthesis.

Fatty acyl-coenzyme A (CoA) thioesters are indispensable intermediates in primary metabolism and essential building blocks for biosynthetic pathways yielding fuels, fine chemicals, and pharmaceuticals. However, large-scale production of acyl-CoAs is frequently constrained by the intracellular availability and high cost of CoA, motivating the development of alternative strategies for precursor generation. Here, we report the engineering of the acylCoA ligase AcsA _Pc to preferentially utilize inexpensive, membrane-permeable synthetic CoA surrogates, with a focus on N-acetylcysteamine (SNAC). Building on a broad specificity variant (D449E), structure-guided saturation mutagenesis and random mutagenesis, coupled with high-throughput colorimetric screening, progressively shifted thiol specificity from CoA to pantetheine and SNAC. The resulting double mutant, F430W/D449E, exhibits a 26-fold improvement in SNAC utilization relative to wildtype while concomitantly reducing CoA activity. Kinetic analyses reveal that these gains arise from increased reaction rates with synthetic thiols and altered substrate preferences across a representative acid panel. The functional relevance of AcsA-generated acyl-SNACs was demonstrated in a reconstituted polyketide system, where enhanced availability of SNAC-linked starter units drove an ∼8-fold increase in pyrone formation. This work establishes AcsA as a tunable platform for orthogonal generation of polyketide precursors and a general framework for alleviating CoA-dependent biosynthetic bottlenecks.

Nature, Published online: 18 May 2026; doi:10.1038/d41586-026-00525-9

If the thought of ‘working’ a room packed with strangers fills you with dread, academic careers coach Caroline Dunne has some advice.

In lignin peroxidase and versatile peroxidase, low FRET from buried Trp to heme promotes long-range electron transfer (LRET) from substrate to heme. This efficient electron flow is facilitated by reduced FRET competition. This mechanism likely evolved to enable the effective degradation of bulky, high-redox-potential lignin that cannot reach the enzyme's active site, thereby ensuring efficient polymer breakdown.

A key step in the evolution of lignin-degrading enzymes is revealed by the observation that, unlike other heme proteins studied to date, ClassII peroxidases exhibit minimal energy transfer from tryptophan to heme residues. Bioinformatics analyses and molecular dynamics simulations (MDS) of ClassII peroxidases (manganese peroxidase, MnP and versatile peroxidase, VP) and the ClassIII enzyme horseradish peroxidase indicate that the tryptophan residue in horseradish peroxidase has the highest orientational factor and corresponding fluorescence resonance energy transfer (FRET) propensity. In contrast, tryptophan residues in manganese peroxidase and versatile peroxidase display lower FRET propensity due to unfavorable orientation factors, despite their proximity to the heme. Steady-state fluorescence experiments support this prediction, showing strong emission in manganese peroxidase and versatile peroxidase but weak emission in horseradish peroxidase. This decreased tryptophan-to-heme energy transfer appears to minimize competition between direct fluorescence resonance energy transfer and long-range electron transfer (LRET), allowing electrons to flow from bulky lignin substrates to the heme center. Such a mechanism likely provided a selective advantage during the evolution of ClassII peroxidases, facilitating efficient lignin degradation at the enzyme surface.

Nature Communications, Published online: 16 May 2026; doi:10.1038/s41467-026-73260-4

Many bacteria sequester enzymes within protein cages for efficient vitamin B2 production. Cryo-EM captures rare cage assembly intermediates, revealing multilayered mechanisms behind architectural precision.

Nature Synthesis, Published online: 15 May 2026; doi:10.1038/s44160-026-01074-9

The enantiodiscrimination between two minimally different alkyl substituents is synthetically challenging. Here a photobiocatalytic strategy combines thiamine diphosphate-dependent enzymes with visible-light-driven photoredox catalysis for the enantioselective synthesis of all-carbon quaternary stereocentres. The process can distinguish between methyl and ethyl groups, through the enantioselective cross-coupling of prochiral alkyl radicals with enzymatic thiamine-derived ketyl radicals.

High-throughput screening (HTS) in asymmetric catalysis is now limited by analytical speed rather than reaction setup. This review surveys recent advances in chiral chromatography, optical assays, mass spectrometry (MS), and nuclear magnetic resonance (NMR) for rapid enantiomeric excess (ee) determination, highlighting direct strategies that transform crude mixture analysis into predictive maps of catalyst generality and chemical space.

ABSTRACT

The widespread adoption of high-throughput experimentation (HTE) has transformed asymmetric catalysis by enabling the parallel execution of numerous reactions to systematically explore chemical space instead of relying on isolated trial and error. Currently, the overall efficiency of this process is limited by analytical readouts rather than reaction setups. Specifically, the rapid and reliable determination of enantiomeric excess (ee) from complex crude mixtures remains a primary bottleneck. To address this bottleneck, recent advancements in chiral analytical methodologies offer solutions through either serial analysis to shorten individual measurement times and increase information density or parallel analysis to multiplex signals for formats using microplates. These techniques span chromatography, spectroscopy, mass spectrometry, and nuclear magnetic resonance. Key factors in applying these methods include their compatibility with multiple substrates, robustness against matrix interference, and the ability to balance speed with accuracy, which allow for earlier evaluation of substrate generality and minimizes bias towards a single substrate. This proactive approach ultimately generates high-dimensional datasets that effectively guide the optimization of catalysts and reaction conditions toward universally applicable asymmetric transformations. In this review, we summarize these recent advancements in high-throughput chiral analytical methodologies and highlight their profound impact on accelerating discovery in the field of asymmetric catalysis.