ACS Catalysis

Enzymes offer unparalleled selectivity and sustainability for chemical synthesis, yet their widespread industrial application is often hindered by the slow and uncertain process of discovering and optimizing suitable biocatalysts. While directed evolution remains the gold standard for enzyme optimization, its success hinges on the availability of a starting enzyme with measurable activity, a persistent bottleneck for many desired functions. Designing libraries likely to contain such functional starting points remains a major challenge. In this work, we use the GenSLM protein language model (PLM) along with a series of filters to generate novel sequences of the {beta}-subunit of tryptophan synthase (TrpB) that express in Escherichia coli, are stable, and are catalytically active in the absence of a TrpA partner. Many generated TrpBs also demonstrated significant substrate promiscuity, accepting non-canonical substrates typically inaccessible to natural TrpBs. Remarkably, several outperformed both natural and laboratory-optimized TrpBs on native and non-canonical substrates. Comparative analysis of the most active and promiscuous generated TrpB and its closest natural homolog confirmed that this enhanced functional versatility does not stem from the natural enzyme, highlighting the creative potential of generative models. Our results demonstrate that the model can generate enzymes which not only preserve natural structure and function but also acquire non-natural properties, establishing PLMs as powerful tools for biocatalyst discovery and engineering, with the potential in some cases to bypass further optimization.

White-rot fungi secrete a repertoire of high-redox potential oxidoreductases to efficiently decompose lignin. Of these enzymes, versatile peroxidases (VPs) are the most promiscuous biocatalysts. VPs are attractive enzymes for research and industrial use, but their recombinant production is extremely challenging. To date, only a single VP has been structurally characterized and optimized for recombinant functional expression, stability and activity. Computational enzyme optimization methods can be applied to many enzymes in parallel, but they require accurate structures. Here, we demonstrate that model structures computed by deep-learning based ab initio structure prediction methods are reliable starting points for one-shot PROSS stability-design calculations. Four designed VPs encoding as many as 43 mutations relative to the wild type enzymes are functionally expressed in yeast whereas their wild type parents are not. Three of these designs exhibit substantial and useful diversity in reactivity profile and tolerance to environmental conditions. The reliability of the new generation of structure predictors and design methods increases the scale and scope of computational enzyme optimization, enabling efficient discovery and exploitation of the functional diversity in natural enzyme families.

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.

Glycoside hydrolases (GHs) play central roles in carbohydrate metabolism and are widely exploited for industrial and biomedical applications. However, they are often not optimal for applications due to their constrained function and strict stereochemical specificity, necessitating the discovery and optimization of distinct enzymes for each glycosidic configuration. Members of glycoside hydrolase family 1 (GH1) are archetypal retaining {beta}-glycosidases, while -specific activity is rare within this family. Here, I demonstrate that a retaining GH1 enzyme can be engineered to hydrolyze both {beta}- and -configured substrates without altering its canonical catalytic residues. Using a well-characterized {beta}-glycosidase and computational protein design strategies targeting second-shell residues surrounding the active site, a bifunctional {beta}-/-glycosidase containing 45 mutations was generated. The engineered variant acquired the ability to hydrolyze the -configured substrate 4-nitrophenyl--D-glucopyranoside while retaining activity toward the originals {beta}-substrates, with reduced catalytic efficiency and thermostability. Structural modeling and docking analyses reveal that the engineered enzyme preserves the original fold and accommodates substrates within the catalytic pocket in a similar manner to the wild type. These findings provide direct evidence that stereochemical constraint in retaining GH is more flexible than previously appreciated and can be modulated through targeted engineering.

Bacterial aryl malonate decarboxylase is a cofactor-free enzyme that generates a wide spectrum of -chiral carboxylic acids in outstanding optical purity, including several non-steroidal anti-inflammatory drugs and chiral building blocks. The well-characterized AMDase from Bordetella bronchiseptica (BbAMDase) and related enzymes of the same family have three main limitations: (i) low stability, both operational and thermal, and (ii) limited substrate spectrum regarding the size of the smaller substituent on the -C-atom and (iii) low stereoselectivity towards -alkenyl--alkyl malonic acids. To address these limitations, we expanded the structural diversity of the AMDase family by ancestral sequence reconstruction (ASR). The phylogenetic analysis of the decarboxylase revealed conserved structural motifs and key amino acids in the hydrophobic active-site cavity, a catalytic motif crucial for activity and selectivity of the enzyme. The analysis highlighted the natural distribution of amino acid exchanges that had been previously identified in enzyme engineering campaigns. AMDase ancestors showed higher stability, activity, and, in one case, also stereoselectivity than BbAMDase. While the up to 10 {degrees}C higher unfolding temperature of AMDase ancestors is a frequent result in ASR, the improvement of the half-life time of 294-fold of ancestor N131 was surprising. Ancestor N31 formed 2-methyl-but-3-enoic acid from its corresponding malonic acid in an optical purity of 99.7% eeR. The extant BbAMDase produces this compound in much lower optical purity (96.8% eeR), which corresponds to a 1.4 kcal{middle dot}mol-1 difference of the transition state free energy of the two reaction paths leading to the different enantiomers. Furthermore, the stereoselectivity of the ancestors was completely inverted by switch of the catalytic cysteine residues G74C/C188G.

Enzyme reactions, both in Nature and technical applications, commonly occur at the interface of immiscible phases. Nevertheless, stringent descriptions of interfacial enzyme catalysis remain sparse, and this is partly due to a shortage of coherent experimental data to guide and assess such work. We have produced and kinetically characterized 83 cellulases, which revealed a conspicuous linear free energy relationship (LFER) between the strength of substrate binding and the activation barrier. This common scaling occurred despite the investigated enzymes were structurally and mechanistically diverse. We suggest that the scaling reflects basic physical restrictions of the hydrolytic process and that evolutionary selection have condensed cellulase phenotypes near the line. One consequence of the LFER is that the activity of a cellulase can be estimated from substrate binding strength, irrespectively of structural and mechanistic details, and this appears promising for in silico selection and design within this industrially important group of enzymes. On a more general note, the LFER may identify a link to inorganic heterogeneous catalysis, and hence open for the implementation of approaches from this field within interfacial enzymology.

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.

O_SCPLOWLC_SCPLOW-Threonine aldolases (LTAs) are attractive biocatalysts for synthesizing {beta}-hydroxy--amino acids (HAAs) via C-C bond formation in pharmaceuticals, although their industrial applications suffer from low activity and diastereoselectivity. Herein, we describe the discovery of a new LTA from Neptunomonas marine (NmLTA) that displays both ideal enzymatic activity (64.8 U/mg) and diastereoselectivity (89.5% diastereomeric excess; de) for the desired product O_SCPLOWLC_SCPLOW-threo-4-methylsulfonylphenylserine (O_SCPLOWLC_SCPLOW-threo-MPTS). Using X-ray crystallography, site-directed mutagenesis, and computational modeling, we propose a "dual-conformation" mechanism for the diastereoselectivity control of NmLTA, whereby the incoming 4-methylsulfonylbenzaldehyde (4-MTB) could potentially bind at the NmLTA active site in two distinct orientations, potentially forming two diastereoisomers (threo- or erythro-form products). Importantly, two key NmLTA residues H140 and Y319 play critical roles in fine-tuning the binding mode of 4-MTB, supported by our site-mutagenesis assays. Uncovering of the catalytic mechanism in NmLTA guides us to further improve the diastereoselectivity of this enzyme. A triple variant of NmLTA (N18S/Q39R/Y319L; SRL) exhibited both improved diastereoselectivity (de value > 99%) and enzymatic activity (95.7 U/mg) for the synthesis of O_SCPLOWLC_SCPLOW-threo-MPTS compared with that of wild type. The preparative gram-scale synthesis for O_SCPLOWLC_SCPLOW-threo-MPTS with the SRL variant produced a space-time yield of up to 9.0 g L-1h-1, suggesting a potential role as a robust C-C bond synthetic tool for industrial synthesis of HAAs at a preparative scale. Finally, the SRL variant accepted a wider range of aromatic aldehyde derivatives as substrates and exhibited improved diastereoselectivity toward para-site substituents. This work provides deep structural insights into the molecular mechanism underlying the catalysis in NmLTA and pinpoints the key structural motifs responsible for regulating the diastereoselectivity control, thereby guiding future attempts for protein engineering of various LTAs from different sources.

Ketosynthase domains govern chain transfer and substrate selectivity in modular polyketide synthases (PKS), yet their functional tunability in native contexts remains poorly understood. We performed phylogenetically guided mutagenesis of the KS5 domain from the Streptomyces cinnamonensis monensin PKS and evaluated 72 variants in vivo across wild-type and reductive-loop-null backgrounds. This revealed discrete active-site motifs that control productivity, redox-state specificity, and extender-unit selection, functions traditionally ascribed to other PKS domains. AlphaFold3 structural mapping linked these motifs to substrate-tunnel and catalytic-core features, providing a mechanistic basis for the observed phenotypes. Our findings demonstrate that KS domains can be rationally re-tuned to overcome productivity bottlenecks and alter specificity in intact PKSs, offering a route to improved yields and expanded chemical diversity in engineered polyketides.

Fluorinases have high potential for industrial biofluorination but any applications have been precluded by low catalytic efficiency and resistance to active site engineering. In this work, we employed PRIZM, a computational workflow utilizing an existing low-N dataset and zero-shot models for in silico prediction of activity-enhancing mutations at distal sites. The combination of these predictions with expert opinion led to the identification of 21 fluorinase mutants with enhanced relative activities, while 3 variants showed increased melting temperatures. A mutation in the hexameric interface, K237R, resulted in the largest stability gain, a more than 3.2-fold improvement in catalytic efficiency at 57{degrees}C, and an 8-fold increase in relative activity at 62{degrees}C. These results highlight the potential of distal fluorinase engineering for improving properties required to realize its industrial applications.

We present a fully computational workflow for de novo design of efficient enzymes using backbone fragments from natural proteins and without recourse to iterative experimental optimization. The best designed Kemp eliminase exhibits >140 mutations from any natural protein, high stability (>85 {degrees}C) and unprecedented catalytic efficiency (12,700 M-1s-1), surpassing previous computational designs by two orders of magnitude. We find that mutations both inside and outside the active site contribute synergistically to the high observed activity and stability. Mutation of an aromatic residue used in all prior Kemp eliminase designs increases efficiency to >105 M-1s-1. Our approach addresses critical limitations in design methodology, generating stable, high-efficiency, new-to-nature enzymes in complex folds and enables testing hypotheses on the fundamentals of biocatalysis through a limited experimental effort.

Cyclohexanone monooxygenases (CHMO) consume molecular oxygen and NADPH to catalyze the valuable oxidation of cyclic ketones. However, CHMO usage is restricted by poor thermostability and stringent specificity for NADPH. Efforts to engineer CHMO have been limited by the sensitivity of the enzyme to perturbations in conformational dynamics and long-range interactions that cannot be predicted. We demonstrate a pair of aerobic, high-throughput growth selection platforms in Escherichia coli for oxygenase evolution, based on NADPH or NADH redox balance. We utilize the NADPH-dependent selection in the directed evolution of thermostable CHMO and discover the variant CHMO GV (A245G-A288V) with a 2.7-fold improvement in residual activity compared to the wild type after 40 {degrees}C incubation. Addition of a previously reported mutation resulted in A245G-A288V-T415C which has further improved thermostability at 45 {degrees}C. We apply the NADH-dependent selection to alter the cofactor specificity of CHMO to accept NADH, a less expensive cofactor than NADPH. We identified the variant CHMO DTNP (S208D-K326T-K349N-L143P) with a 21-fold cofactor specificity switch from NADPH to NADH compared to the wild type. Molecular modeling indicates that CHMO GV experiences more favorable residue packing and backbone torsions, and CHMO DTNP activity is driven by cooperative fine-tuning of cofactor contacts. Our introduced tools for oxygenase evolution enable the rapid engineering of properties critical to industrial scalability.

Acetyl-coenzyme A (CoA) is a central metabolic intermediate that links carbon and energy metabolism across all domains of life. The conversion of acetate and acetyl-CoA is carried out by three enzyme pathways: acetate kinase/phosphotransacetylase, ADP-forming acetyl-CoA synthetase, and AMP-forming acetyl-CoA synthetase (Acs). Acs enzymes serve critical physiological roles across diverse organisms generally by catalyzing a reversible two-step reaction forming acetyl-CoA and AMP from acetate and ATP. Isolated from the wastewater reclamation facility in Norman, Oklahoma, Syntrophus aciditrophicus strain SB (Sa) relies on an AMP-forming acetyl-CoA synthetase (SaAcs1) that favors synthesizing acetate and ATP from acetyl-CoA and AMP, in contrast to all previously characterized Acs enzymes. The origin of this preference and the structural determinants of both the thioester-forming step and catalytic directionality remain poorly understood. Here, we report a 2.2 [A] crystal structure of full-length SaAcs1 in the adenylation conformation with acetyl-AMP bound in the active site. Structural comparison to the extensively characterized Acs enzymes from Salmonella enterica (SeAcs) and Cryptococcus neoformans (CnAcs) revealed a displaced CoA-binding loop in SaAcs1. Enzymatic assays confirmed that SaAcs1 preferentially catalyzes the ATP-forming reaction. Site-directed mutagenesis demonstrated that reversion of two residues, G196 and T197, at the beginning of the CoA-binding loop to the consensus sequence repositions the loop and shifts catalytic preference toward the AMP-forming direction. Together, these results establish the CoA-binding loop and G196 and T197 as the primary structural determinants of directional preference in SaAcs1.

The fluorinase enzyme, the only known biocatalyst forming stable carbon-fluorine bonds, operates with extremely low efficiency, catalyzing one reaction every 2-12 minutes. This severely limits its utility for sustainable biofluorination, and its sluggish activity remains poorly understood. We suppressed its aggregation through directed mutagenesis and elucidated the kinetic mechanism using a novel mathematical framework that fits complex kinetic and oligomerization data. This analysis revealed that >80% of enzyme molecules are inactive under standard conditions due to two dead-end pathways. The designed W50F+A279R mutant preferentially formed hexamers and displayed enhanced catalytic efficiency in this oligomeric state. When coupled with mechanism-based optimization of the reaction medium, including enzymatic removal of the inhibitory product, the catalytic turnover rate reached 12.5 {+/-} 2.1 min-{superscript 1}, representing [~]60-fold increase compared with previously reported turnover rates of the wild-type enzyme. Our work provides a mechanistic blueprint for fluorinase enhancement and a generalizable mathematical framework for analyzing kinetics of multimeric enzymes.

Directed evolution (DE) inspired by natural evolution (NE) has been achieving tremendous successes in protein/enzyme engineering. However, the conventional one-protein-for-one-task DE cannot match the multi-proteins-for-multi-tasks NE in terms of screening throughput and efficiency, thus often failing to meet the fast-growing demands for biocatalysts with desired properties. In this study, we design a novel multi-enzyme-for-multi-substrate (MEMS) DE model and establish the proof-of-concept by running a NE-mimicking and higher-throughput screening on the basis of two-P450s-against-seven-substrates (2Px7S) in one pot. With the significantly improved throughput and hit-rate, we witness a series of convergent evolution events of the two archetypal cytochrome P450 enzymes (P450 BM3 and P450cam) in laboratory. Further structural analysis of the two functionally convergent P450 variants provide important insights into how distinct active-sites can reach a common catalytic goal.

Enzyme-directed evolution increasingly relies on computational tools to prioritize mutations, yet their practical value is difficult to assess because kinetic data are often aggregated across heterogeneous assay conditions, inflating apparent generalization. Here we introduce EnzyArena, a curated benchmark that groups kinetic parameters (kcat, Km, kcat/Km) into condition-matched experimental subsets to enable realistic evaluation. Using this resource, we benchmark 10 representative models from two arising strategy families--zero-shot fitness prediction and supervised kinetic-parameter prediction--across BRENDA- and SABIO-RK-derived subsets and 25 independent mutagenesis datasets. Kinetic-parameter predictors perform strongly on database-derived subsets but lose their advantage on independent datasets, whereas zero-shot predictors show more consistent generalization. A simple consensus of multiple zero-shot models further improves the precision of identifying beneficial mutants. We prospectively validated these findings in a wet-lab campaign (150 mutants) comparing random mutants, UniKP-prioritized mutants and ESM-1v-prioritized mutants (representing supervised kinetic-parameter prediction and zero-shot fitness prediction, respectively), where ESM-1v achieved the highest utility and UniKP underperformed the random baseline. Together, this study establishes realistic baselines for computational mutant prioritization and highlights consensus zero-shot strategies as a practical starting point for enzyme engineering.

Polyurethanes (PURs) represent a significant challenge in plastic waste management due to their chemical resilience and limited recycling options. In this study, we report the identification and characterization of six novel bacterial urethanases, expanding the enzymatic repertoire for targeted PUR depolymerization. These enzymes demonstrated carbamate-cleaving activity optimally under alkaline conditions, maintaining stability across a pH range of 7 to 10 and varying thermal and solvent tolerances. Among the candidate enzymes, u17, u10, and u15 collectively exhibited high activity, catalytic efficiency, and thermostability, establishing a strong foundation for further optimization. Building on these results, u15 emerged as particularly notable for its catalytic efficiency on the carbamate model substrate di-urethane ethylene methylenedianiline, DUE-MDA, with a kcat/KM of 51.8 {+/-} 0.1 (s-1mM-1). and this motivated its selection for detailed structural analysis. High-resolution crystallography of u15 revealed key active-site architecture, including the conserved amidase signature catalytic triad and flexible loop regions that influence substrate binding and specificity. Molecular docking and molecular dynamics simulations further elucidated substrate binding determinants of u15 during urethane bond hydrolysis. Docking of DUE-MDA revealed two distinct substrate orientations (Pose A and Pose B) differing in the positioning of the carbamate group relative to Ser177. Pose A was more stable and catalytically competent, maintaining the substrate within the oxyanion hole and sustaining optimal geometry for nucleophilic attack by Ser177. Comparable behavior was observed for the partially hydrolyzed intermediate mono-urethane ethylene methylenedianiline, MUE-MDA, indicating a conserved binding mode across substrates. Collectively, these findings highlight amidase signature urethanases as valuable scaffolds for advancing sustainable and scalable biocatalytic recycling of polyurethanes. TOC O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=87 SRC="FIGDIR/small/705263v1_ufig1.gif" ALT="Figure 1"> View larger version (15K): org.highwire.dtl.DTLVardef@127bf23org.highwire.dtl.DTLVardef@75c29corg.highwire.dtl.DTLVardef@13bbf30org.highwire.dtl.DTLVardef@18504a4_HPS_FORMAT_FIGEXP M_FIG C_FIG

Evolutionary constraints significantly limit the diversity of naturally occurring enzymes, thereby reducing the sequence repertoire available for enzyme discovery and engineering. Recent breakthroughs in protein structure prediction and de novo design, powered by artificial intelligence, now enable us to create enzymes with desired functions without relying on traditional genome mining. Here, we demonstrate a computational strategy for creating new-to-nature PET hydrolases by leveraging the known catalytic mechanisms and implementing multiple deep learning algorithms and molecular computations. This strategy includes the extraction of functional motifs from a template enzyme (here we use leaf-branch compost cutinase, LCC), regeneration of new protein scaffolds, computational screening, experimental validation, and sequence refinement. We successfully replicate PET hydrolytic activity with designer enzymes that are at least 30% shorter in sequence length than LCC. Among them, RsPETase 1 stands out due to its robust expressibility. It exhibits comparable activity to IsPETase and considerable thermostability with a melting temperature of 56 {degrees}C, despite sharing only 34% sequence similarity with LCC. This work suggests that enzyme diversity can be expanded by recapitulating functional motifs with computationally built protein scaffolds, thus generating opportunities to acquire highly active and robust enzymes that do not exist in nature.

Phenylalanine ammonia-lyases (PALs) deaminate L-phenylalanine to trans-cinnamic acid and ammonium and have idespread application in chemo-enzymatic synthesis, agriculture, and medicine. In particular, the PAL from Anabaena variabilis (Trichormus variabilis) has garnered significant attention as the active ingredient in Pegvaliase(R), the only FDA-approved drug treating classical phenylketonuria (PKU). Although an extensive body of literature exists on structure, substrate-specificity, and catalytic mechanism, protein-wide sequence determinants of function remain unknown, which limits the ability to rationally engineer these enzymes. Previously, we developed a high-throughput screen (HTS) for PAL, and here, we leverage it to create a detailed sequence-function landscape of PAL by performing deep mutational scanning (DMS). Our method revealed 79 hotspots that affected a positive change in enzyme fitness, many of which have not been reported previously. Using fitness values and structure-function analysis, we picked a subset of residues for comprehensive single- and multi-site saturation mutagenesis to improve the catalytic activity of PAL and identified combinations of mutations that led to improvement in reaction kinetics in cell-free and cellular contexts. To understand the mechanistic role of the most beneficial mutations, we performed QM/MM and MD and observed that different mutants confer improved catalytic activity via different mechanisms, including stabilizing first transition and intermediate states and improving substrate diffusion into the active site, and decreased product inhibition. Thus, this work provides a comprehensive sequence-function relationship for PAL, identifies positions that improve PAL activity when mutated and assesses their mechanisms of action.