ACS Catalysis

L-3,4-dihydroxyphenylalanine (L-Dopa) is an important pharmaceutical for the treatment of Parkinsons disease and a precursor to numerous catechol-containing compounds. The flavin-dependent monooxygenase HpaBC is a promising biocatalyst for microbial L-Dopa production but exhibits limited native activity toward L-tyrosine. Although structure-based machine learning (ML) models have become increasingly popular for protein engineering, relatively few studies have systematically compared their performance or evaluated their integration into iterative engineering workflows. Here, we benchmarked multiple ML models for their ability to predict activity enhancing mutations in HpaBC. Experimentally validated single mutants were used to seed combinatorial design with EVOLVEpro, generating progressively improved higher-order variants. We next evaluated how expanding the EVOLVEpro training set with directed evolution derived variants influenced combinatorial predictions and finally explored an expanded sequence space by allowing combinations of both machine learning derived and directed evolution derived mutations. This workflow produced HpaBC variants with substantially improved activity. Although incorporating directed evolution data substantially altered EVOLVEpros predicted mutational trajectories, both training strategies converged on variants with comparable activities, demonstrating that distinct regions of sequence space can yield similarly optimized enzymes. Together, these results provide a systematic comparison of zero-shot ML models and establish an iterative framework for integrating machine learning with directed evolution to accelerate enzyme engineering.

Enzymatic sulfation offers a green-chemistry route to the selective sulfation of phenolic and other hydroxylated compounds. Bacterial arylsulfate sulfotransferases (ASSTs) are particularly attractive biocatalysts because they use simple aryl sulfate donors, such as p-nitrophenyl sulfate (pNPS), instead of the complex and labile cofactor PAPS required by conventional sulfotransferases. However, their rational application and engineering remain limited because the molecular principles controlling substrate recognition, catalytically competent donor and acceptor positioning, covalent sulfo-enzyme formation, and sulfo transfer are not fully understood. In particular, it has remained unclear how conserved active-site histidines cooperate during catalysis and how active-site preorganization promotes efficient sulfo transfer. Here, we address these questions for the newly characterized ASST (ASTB) from Desulfitobacterium hafniense clade 6 (Dh_ASST_6) by combining X-ray crystallography, steady-state enzyme kinetics, and quantum mechanics/molecular mechanics (QM/MM) simulations. Crystal structures of Dh_ASST_6 in the apo state and donor-soaked state reveal a monomeric {beta}-propeller enzyme with a histidine-centered active site and directly identify His350 as the transient sulfo carrier. Kinetic analysis shows efficient turnover and supports a ping-pong bi-bi mechanism, consistent with formation of a covalent sulfo-enzyme intermediate. Steered QM/MM simulations of the second catalytic step show that sulfuryl transfer from sulfated His350 to the phenolic acceptor proceeds through a proton-coupled, multistep pathway in which His283 assists acceptor activation by proton transfer. Active-site preorganization precedes the main bond-reorganization event and generates a near-attack geometry in which acceptor alignment, proton transfer, and stabilization of the sulfo group by a conserved hydrogen-bonding network jointly promote sulfo transfer. Free-energy reconstruction gives a low barrier of approximately 9.6 kcal/ mol, in agreement with efficient enzymatic turnover. The structural, kinetic, and computational data support a complete catalytic cycle for Dh_ASST_6 based on a cooperative two-histidine mechanism. This model establishes catalytically competent acceptor positioning and preservation of the histidine-centered hydrogen-bonding network as key design principles for ASST engineering. These features provide concrete targets for expanding substrate scope and tuning donor or acceptor preference while maintaining efficient sulfuryl transfer.

Enzymes can be computationally designed for an increasing range of reactions, yet the catalytic efficiencies achieved are typically well below those of natural biocatalysts. Directed evolution can narrow this gap, often improving activity through mutations with non-obvious effects on catalysis. Here, we dissect how directed evolution improved a de novo designed Kemp eliminase by comparing the ground-state and transition-state ensembles of the designed and evolved variants. Extensive molecular dynamics simulations and QM/MM reaction barrier calculations identify subtle but important differences that modulate reactivity. Simulations show that reaction proceeds via a specific transition-state-like conformation, characterized by desolvation of the catalytic base and an organized active-site electric field. Barriers correlate strongly with the reaction energy of proton transfer, indicating that desolvating the catalytic base in the ligand-bound complex raises its proton affinity and thereby lowers the barrier. Evolution enriches the ground-state ensemble in reactive conformations, minimizing the reorganization required to reach the transition state. After evolution, formation of the transition-state-like conformation causes less disruption to the protein, which maintains coordinated motions. This is consistent with the emergence of a negative activation heat capacity with evolution. Efficient catalysis requires a ground-state ensemble that resembles the transition state, an objective that enzyme design should pursue alongside geometric and electrostatic complementarity for the transition state.

Biocatalysis offers sustainable solutions to pressing challenges in chemical synthesis by exploiting the remarkable efficiency and selectivity of enzymes. Importantly, enzymes are able to accommodate non-native substrates and mediate transformations outside of their natural repertoire. Enzymes can be engineered for diverse applications by harnessing these promiscuous activities and optimizing them using directed evolution (DE). The success of a DE campaign, however, depends on the availability of a protein starting point that displays detectable levels of the desired function. To find a starting point, researchers often screen libraries of protein variants for novel activities, typically with low rates of success. Here, instead, we diversified the active site of a desirable parent protein and applied machine learning to generate informed, promiscuous libraries of protein variants. Specifically, we tested 26 different carbene and nitrene transfer reactions and used active learning-assisted directed evolution (ALDE) to generate optimized protoglobin variants with high activity across multiple reactions. We observed improvements in activity and selectivity for every reaction performed by the parent enzyme in at least one member of the ALDE-predicted libraries. Moreover, variants from these libraries can catalyze 5 out of 10 reactions not catalyzed by the parent protoglobin. These results indicate that supervised machine learning can help guide the construction of high-value enzyme libraries with expanded catalytic scope. O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=99 SRC="FIGDIR/small/740427v1_ufig1.gif" ALT="Figure 1"> View larger version (31K): org.highwire.dtl.DTLVardef@6d5cfborg.highwire.dtl.DTLVardef@1f38e0aorg.highwire.dtl.DTLVardef@f25fa2org.highwire.dtl.DTLVardef@64a0dc_HPS_FORMAT_FIGEXP M_FIG C_FIG

Engineered glycosynthases (GSs) are powerful biocatalysts for custom glycan synthesis, yet their optimization via directed evolution is severely constrained by bottlenecks in high-throughput screening for activated azido-sugar donors. Here, we demonstrate that chemical rescue (CR)--the azide-mediated restoration of hydrolytic activity in nucleophile-deficient mutants--serves as a predictive, high-throughput proxy for glycosynthase activity. Applying an azide-responsive Escherichia coli biosensor screen to a site-saturation mutagenesis library of Thermotoga maritima -L-fucosidase (TmAfc), we established a strong rank-order correlation between CR and GS activities in both crude lysates ({rho} = 0.73) and purified enzymes ({rho} = 0.95). Transition path sampling and QM/MM umbrella sampling revealed that both pathways proceed through a shared oxocarbenium-ion-like transition state ({Delta}G{ddagger} {approx} 8.7 kcal/mol), providing a structural and thermodynamic rationale for using CR to select for transition-state-stabilizing mutations. Biochemical characterization of top-performing variants yielded an engineered fucosynthase (TmAfc_D224G_N70D_T392S) exhibiting a nearly 100-fold enhancement in Vmax alongside altered regioselectivity. This two-tiered screening framework leverages cost-effective chemical rescue assays to streamline glycosynthase engineering for tailored glycans synthesis.

De novo-designed enzymes have recently achieved high catalytic activity and stereoselectivity while demonstrating exceptional thermostability in entirely novel protein scaffolds. Among these, -helical barrel protein scaffolds are attractive structures for biocatalysis due to their structural simplicity, high thermostability, and rationalizable sequence patterning. However, enabling major structural reengineering of these scaffolds while maintaining the structure, stability and catalytic activity while also improving soluble protein production remain major challenges and pose the fundamental question how engineerable a de novo backbone-sequence pair is. Here, we combine deep learning based and classic computational protein design to modify and optimize de novo -helical barrel biocatalysts. Using the previously reported six-helical barrel 6H5L as a model scaffold, AlphaFold2-guided RosettaRemodel enabled the design of a truncated variant, whose crystal structure closely matches the design model. Additional sequence-redesign using ProteinMPNN generated a variant with a tenfold increase of soluble protein yield in Escherichia coli. Biochemical, biophysical, and structural analyses showed that both variants retained the overall barrel architecture, high thermal stability, and catalytic activity for both purified protein and whole-cell systems. Detailed kinetic analysis on the variants showed both variation in kcat and Km, reflecting changes in catalytic turnover and substrate binding. Together, these approaches provide new insights and possibilities for the further engineering of functional de novo -helical barrels, their ability to withstand dramatically large sequence changes and their broader application in biocatalysis and biotechnology.

Enzymes catalyze multistep reactions by stabilizing successive transition states within well organized, yet dynamic active sites. However, computational enzyme design typically targets a single transition state using rigid active-site models. Here, we introduce multistate enzyme design, which uses conformational ensembles to optimize active sites across an entire reaction coordinate. Applied to a de novo Morita--Baylis--Hillmanase, multistate enzyme design outperformed conventional single-state design, with the most active variant achieving >100-fold higher bi-substrate catalytic efficiency and surpassing an extensively optimized enzyme from directed evolution in both efficiency and enantioselectivity. Structural and kinetic analyses revealed that multistate design preserved catalytic preorganization and conformational plasticity, distributed stabilization across the reaction coordinate and avoided kinetic bottlenecks created by single-state optimization. By contrast, single-state design compromised preorganization, destabilized upstream states and shifted rate limitation away from the targeted transition state. Multistate enzyme design provides a framework for designing catalytic landscapes rather than static active sites, opening a route to efficient de novo enzymes for complex multistep chemistry.

Enzyme catalysts are increasingly used for sustainable pharmaceutical manufacturing, but engineering industrial biocatalysts remains challenging as multiple catalytic and developability properties must be optimized simultaneously from limited experimental data. Here, we develop a machine learning-guided multi-objective design framework that integrates sparse functional measurements with complementary evolutionary and structural information to engineer the ketoreductase Gre2. The resulting designs achieved simultaneous improvements in catalytic performance, protein yield, and thermal stability, with selected designs retaining improved performance under process-relevant conditions. More broadly, our results demonstrate that integrating complementary biological information enables efficient multi-objective enzyme engineering from sparse experimental data, providing a general strategy for accelerating industrial biocatalyst development.

Despite advances in de novo enzyme design, success has been largely limited to low energy barrier model reactions. Amide bonds such as those linking amino acids along the peptide backbone are stable for hundreds of years in neutral aqueous solution because of the high energy barrier to hydrolysis1. Here we describe the de novo design of enzymes which utilize an activated cysteine nucleophile to hydrolyze the polypeptide backbone in a sequence-dependent manner, with a success rate of 13/69=19% and rate enhancements over the background reaction (kcat/kuncat) of up to 3 x 107. The designed proteases have folds very different from proteases in nature (TM score < 0.50), and six crystal structures are very close to the design models (C RMSDs < 1.2 [A]), highlighting the capacity for generalization and the accuracy of the design methodology. Experimental and computational analyses suggest that the remaining gap in activity to the most active native cysteine proteases arises from imperfections in active site preorganization and substrate positioning. The designed proteases efficiently cleave their targets in mammalian cells, opening the door to a wide range of synthetic biology applications.

2-Deoxyribose-5-Phosphate Aldolase (DERA) is a key enzyme in the pentose phosphate pathway. Due to its C-C bond formation and stereoselective capabilities, DERA has been widely used for biocatalytic applications including the synthesis of chiral intermediates for antiviral and anticancer drugs. While protein engineering has expanded its substrate pool, improved yield, and enhanced stereoselectivity, the molecular basis of stereoselectivity remains unclear. Here, we determined the crystal structures of wildtype DERA from Geobacillus sp. and two of its variants with opposite stereoselectivity. Using a combination of structural biology, biochemistry, organic synthesis and molecular dynamic simulations, we show that the catalytic Lysine adopts two conformations and the Lysine conformation is a key determinant of DERA stereoselectivity. We also identified a mechanism of regulating stereoselectivity via a key amino acid. Using DERA from E. coli, we show that these findings are most likely conserved among bacteria.

Engineering proteins at the scale of modern high-throughput screening technologies remains challenging because most experimental libraries sample only limited regions of sequence space while maintaining relatively low levels of functional diversity. Although recombination-based strategies can introduce substantially larger sequence changes than conventional mutagenesis approaches, it is unclear whether highly recombined enzymes can retain catalytic competence when library sizes approach millions to billions of variants. To investigate this question, we constructed a recombination library by combining ten sequence blocks from ten bacterial PETases, generating a theoretical diversity of 10^8 chimeric enzymes. Screening of yeast-displayed libraries using a fluorophosphonate probe for catalytic-serine reactivity identified substantial populations of catalytically competent variants despite extensive sequence recombination and dozens of amino acid substitutions per chimera. Approximately 2% of sampled variants possessed active-site serine reactivity, corresponding to an estimated 10 catalytically competent enzymes within the full theoretical library. Deep sequencing showed that functional populations remained broadly distributed across sequence space following selection and contained multiple enriched recombination patterns spanning the PETase structure. Secondary lysate screening and enzyme-normalized activity measurements further identified multiple purified PETase chimeras with soluble esterase activity comparable to or greater than the wild type Ideonella sakaiensis PETase on a chromogenic surrogate substrate. Together, these results demonstrate that large-scale recombination can generate millions of diverse and catalytically competent enzymes while remaining compatible with modern high-throughput screening workflows. More broadly, this work establishes recombination-based diversity generation as a scalable strategy for protein engineering and large-scale sequence-function data generation.

Sulfation plays an important role in many metabolic processes such as substrate detoxification, solubilization or activation. The enzymes responsible for these reactions, sulfotransferases, occur in all domains of life and can be classified in two major, evolutionary independent groups, PAPS dependent and PAPS independent or arysulfate sulfotransferases (ASST) enzymes. Besides the fact that they play key roles in biological functions, sulfotransferases can be interesting enzymatic tools to selectively sulfate specific positions of various metabolites. To this end we need to understand the relation between sequence, structure and function. Actually, eukaryotic PAPS dependent sulfotransferases have extensively been studied, while we largely lack knowledge about the diversity of bacterial PAPS-independent ASSTs. We have conducted a multidisciplinary study of a bacterial sulfotransferase from Salmonella typhimurium, revealing the importance of active site residues for activity and guiding substrate specificity. Based on the crystal structure in complex with substrate, we used molecular docking combined with Quantum Mechanics/Molecular Mechanics (QM/MM) free energy calculations to investigate the reaction catalytic mechanism with a focus on the role of specific active site residues in substrate specificity. Our results show that the catalytic mechanism involving three conserved histidines is common with previously determined ASSTs, and highlight a specific pair of aromatic residues, Phe174 and Tyr557, that play an important role in defining the substrate specificity in ASST clade 5. These findings enhance our general understanding of how substrate specificity is attained and has the potential to guide further enzyme engineering to expand the acceptor scope of ASSTs.

D-Ribitol-5-phosphate (Rbo5P) plays vital roles in bacterial and mammalian development. In mammals, Rbo5P is an essential component of the O-mannosyl glycan on -dystroglycan, and its defective biosynthesis causes a group of dystroglycanopathies. Supplementation with cytidine diphosphate ribitol (CDP-ribitol, CDP-Rbo) and its analogues has shown therapeutic potential for certain dystroglycanopathies. However, facile and scalable strategies to prepare CDP-Rbo remain unavailable. Here, we report a practical one-pot enzymatic strategy for gram-scale CDP-Rbo production. Starting from inexpensive ribitol, Rbo5P is first generated by L-ribulokinase (AraB) and then converted to CDP-Rbo by CDP-ribitol pyrophosphorylase (TarI) in 90% yield. To streamline the process, AraB and TarI were fused into a single bifunctional biocatalyst, AraB-TarI, which converts ribitol directly into CDP-Rbo on a gram-scale in a one-step reaction. Furthermore, a nucleotide recycling strategy was developed to lower the cost and increase the atom economy from 47% to 75%. This work provides a green and scalable route to CDP-Rbo and a reliable material supply for developing therapeutics against dystroglycanopathies.

Protein engineering requires efficient navigation of vast sequence spaces under limited evaluation budgets, especially when multiple properties must be optimized simultaneously. We developed Evolution-inspired Multi-Objective Bayesian Optimization (EvoMOBO), an active-learning framework that integrates path-dependent sequence generation, global competition among generated variants, and explicit multi-objective optimization. Benchmarking against state-of-the-art methods on complete steroid receptor DNA-binding domain and ParD3 antitoxin landscapes demonstrated robust target-region enrichment, Pareto-front advancement, and sequence diversity across two- and three-objective tasks. In the DBD landscape, simulation-derived geometric descriptors served as labels for both initialization and iterative updating, enriching variants with favorable measured activities without experimental labels. Building on this validation, we applied EvoMOBO to two enzyme-engineering tasks using simulation-derived mechanistic descriptors, with experiments reserved for final validation. For an old yellow enzyme (GkOYE), 16 of 26 tested variants outperformed the wild type, and the best increased non-native oxidative dehydrogenation conversion from 17.5% to 95%. For a formate oxidase (AoFOx), EvoMOBO identified aggregation-resistant variants, two of which nearly doubled diethyl phthalate degradation in a photoenzymatic cascade. Together, these results establish EvoMOBO as a modular framework for multi-objective protein engineering using experimental or mechanism-derived labels.

Despite their intriguing native metalloenzymology, naturally occurring copper enzymes remain largely underexploited for new-to-nature biocatalytic reactions. Herein, we report the systematic investigation and reprogramming of natural copper enzymes to catalyze unnatural free radical (pseudo)halogenation reaction in a highly enantioselective fashion. Evaluating Cu enzymes in the decarboxylative azidation of redox-active esters revealed activity across multiple Cu enzyme families, with type III binuclear Cu enzymes, particularly the Bacillus megaterium tyrosinase (BmTyr), exhibiting superior activity and enantioselectivity across both stabilized and unstabilized alkyl radicals upon further engineering. The strong halide binding affinity of the binuclear Cu system also enabled challenging enantioconvergent bromination, chlorination and isothiocyanation reactions, which remained inaccessible to repurposed nonheme Fe enzymes. Further EPR and UV-visible spectroscopic analyses confirmed the coupled binuclear nature of wild-type and engineered bacterial tyrosinases and provided insights into the origin of their enhanced activity. Collectively, this study establishes binuclear copper enzymes as a powerful platform for new-to-nature stereoselective radical reactions, expanding the scope of metalloenzyme catalysis beyond mononuclear systems.

Poly(ethylene terephthalate) (PET) is one of the most widely produced plastics, and enzymatic depolymerization offers a promising route to closed-loop recycling under mild conditions. However, most known bacterial PET hydrolases belong to a conserved canonical-fold cutinase family, leaving much of alpha/beta-hydrolase diversity unexplored. Here, we mapped bacterial cutinase sequence space by combining bioinformatics-guided sequence selection with high-throughput secretion screening in Bacillus subtilis. A library of 1,120 genes encoding 954 unique bacterial cutinases, spanning canonical- and minimal-fold families, was screened for activity on Impranil DLN and semicrystalline PET. We identified 156 secreted cutinases with polyester activity, broadly distributed across sequence space, but only ten showed detectable PET hydrolysis, all from the canonical-fold family. These PET hydrolases were active at 40-50{degrees}C, preferred alkaline pH, and showed moderate thermostability. Our results demonstrate that PET activity is rare among bacterial cutinases and provide a scalable workflow for discovering diverse enzyme starting points.

Homologous protein structures contain valuable information about tolerated sequence variation. However, translating this information into practical enzyme design strategies remains challenging. Here we present EnZight, a user-friendly web server that integrates homologous structural alignment with intuitive visualization to identify substitution hotspots in protein cores. EnZight exploits structurally aligned homologs to identify positions where the surrounding structural environment is conserved while the residue at the position varies across homologs. This enables prediction of substitutions that preserve fold integrity while modulating function and thermostability. The approach further provides interactive structural outputs that allow users to inspect and prioritize substitutions manually. To validate the use of EnZight, we used a polyurethane-degrading amidase as a proof of concept. We constructed 34 variants, and 97% were successfully expressed, indicating high foldability of the predicted substitutions. Several substitutions improved both catalytic turnover and thermostability, and, importantly, beneficial substitutions combined additively, enabling stepwise accumulation of improvements. The best triple mutant variant exhibited a six-fold increase in catalytic turnover and 2{degrees}C increase in apparent melting temperature. Enhanced activity toward the pharmaceutical micropollutant flutamide further demonstrates EnZight's broad applicability in identifying substitutions that enable enzyme optimization across diverse substrates.

The standard numbering scheme for class B beta-lactamases (metallo-beta-lactamases, MBLs), the BBL scheme, was proposed by Galleni et al. in 2001 and updated by Garau et al. in 2004 from the small number of MBL structures then solved. Hundreds of additional class B sequences have since been catalogued, but most have never been structurally solved, so whether BBL correspondences hold across this larger diversity has not been directly tested. Following the precedent set for the sibling class D enzymes, we used ColabFold (AlphaFold2/MMseqs2) to model 87 class B families across subclasses B1, B2, and B3 lacking an experimental structure, after validating the pipeline against an eight-enzyme benchmark spanning all three subclasses (RMSD to experimental structures 0.35-0.46 A). Applying BBL zinc-ligand numbering to the 87 models via paired sequence- and structure-based residue-transfer methods, the canonical zinc-coordination motif transferred cleanly in the large majority of families, and 11 of the 16 sequence-level exceptions were accounted for by a literature-supported biological cause (a subclass-appropriate substitution or a previously proposed, unconfirmed fourth subgroup), while two motif-retaining families showed predicted local structural divergence at the active-site loop, flagged by structural superposition, that could not be resolved to exact zinc-ligand numbering and warrants direct experimental follow-up. These results show that the BBL scheme, formulated before metagenomic sequencing or deep-learning structure prediction existed, generalizes robustly into the structural space these technologies have since opened, and identify a small set of variants whose active sites merit experimental attention.

The biosynthetic machineries of ribosomally synthesized and post-translationally modified peptides (RiPPs) are often substrate tolerant. A remarkable example is the class II lanthipeptide synthetase ProcM, which naturally functions as a generalist enzyme that has not evolved to use a specific substrate during its evolutionary history. Although ProcM has been studied extensively, the sequence features associated with productive modification remain underexplored. In this study, we use the ultrahigh-throughput mRNA display technique to map the sequence compatibility of ProcM across a focused library. This approach expands the landscape of ProcM reactivity beyond native substrates and individually characterized variants. Machine learning (ML) is used as a tool to demonstrate that the selected dataset contains learnable signatures and classification architectures revealed a balanced accuracy of 0.73. This performance contrasts sharply with the near-perfect accuracy of specialized enzyme models as the sequence-fitness landscape of the generalist enzymes are characterized by class imbalance and limited by intrinsic dataset features. Our results provide a high-throughput view of ProcM reactivity and highlight differences with previous high-throughput studies on substrate selectivity of RiPP modification enzymes. Future studies will need to assess whether these differences are common when comparing generalist with specialist enzymes.

Although significant progress has been made in creating de novo metalloenzymes that hydrolyze activated esters1,2, the energetically demanding cleavage of amide bonds has remained a major challenge for enzyme design: amide bonds are significantly more stable than ester bonds, the amine leaving groups in proteins are not activated, and peptide substrates are flexible making them difficult to bind precisely. Here, we report the de novo design of zinc proteases from minimal catalytic motifs using RoseTTAFold Diffusion 2 for Molecular Interfaces3, optimized for both enzyme and protein-protein interaction design. Of 135 computational designs experimentally tested, 36% had activity and cleaved precisely at the intended site. The most active design accelerates peptide bond hydrolysis by more than 108-fold relative to the uncatalyzed reaction, and by over 1010-fold following the introduction of four point mutations that enhanced active-site preorganization4. Building on these capabilities, we designed metalloproteases that specifically cleave human TDP-43, the amyloid-{beta} peptide, and serum amyloid A with rate accelerations up to 9.2x108-fold over background. We illustrate the potential of our approach for bio-orthogonal control over cell state by generating caged cytokines and caged receptor antagonists that are selectively unmasked by our designed proteases. These results demonstrate that de novo enzyme design has advanced well beyond model reactions with activated substrates and open the door to design of proficient metallohydrolases for a wide range of applications in medicine and bioremediation.