Dataset Information

Enhanced Sequence-Activity Mapping and Evolution of Artificial Metalloenzymes by Active Learning.

ABSTRACT: Tailored enzymes are crucial for the transition to a sustainable bioeconomy. However, enzyme engineering is laborious and failure-prone due to its reliance on serendipity. The efficiency and success rates of engineering campaigns may be improved by applying machine learning to map the sequence-activity landscape based on small experimental data sets. Yet, it often proves challenging to reliably model large sequence spaces while keeping the experimental effort tractable. To address this challenge, we present an integrated pipeline combining large-scale screening with active machine learning, which we applied to engineer an artificial metalloenzyme (ArM) catalyzing a new-to-nature hydroamination reaction. Combining lab automation and next-generation sequencing, we acquired sequence-activity data for several thousand ArM variants. We then used Gaussian process regression to model the activity landscape and guide further screening rounds. Critical characteristics of our pipeline include the cost-effective generation of information-rich data sets, the integration of an explorative round to improve the model's performance, and the inclusion of experimental noise. Our approach led to an order-of-magnitude boost in the hit rate while making efficient use of experimental resources. Search strategies like this should find broad utility in enzyme engineering and accelerate the development of novel biocatalysts.

SUBMITTER: Vornholt T

PROVIDER: S-EPMC11273458 | biostudies-literature | 2024 Jul

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Enhanced Sequence-Activity Mapping and Evolution of Artificial Metalloenzymes by Active Learning.

Vornholt Tobias T Mutný Mojmír M Schmidt Gregor W GW Schellhaas Christian C Tachibana Ryo R Panke Sven S Ward Thomas R TR Krause Andreas A Jeschek Markus M

ACS central science 20240522 7

Tailored enzymes are crucial for the transition to a sustainable bioeconomy. However, enzyme engineering is laborious and failure-prone due to its reliance on serendipity. The efficiency and success rates of engineering campaigns may be improved by applying machine learning to map the sequence-activity landscape based on small experimental data sets. Yet, it often proves challenging to reliably model large sequence spaces while keeping the experimental effort tractable. To address this challenge ...[more]

PMID: 39071060

Similar Datasets

Project description:Artificial metalloenzymes (ArMs) result from anchoring a metal-containing moiety within a macromolecular scaffold (protein or oligonucleotide). The resulting hybrid catalyst combines attractive features of both homogeneous catalysts and enzymes. This strategy includes the possibility of optimizing the reaction by both chemical (catalyst design) and genetic means leading to achievement of a novel degree of (enantio)selectivity, broadening of the substrate scope, or increased activity, among others. In the past 20 years, the Ward group has exploited, among others, the biotin-(strept)avidin technology to localize a catalytic moiety within a well-defined protein environment. Streptavidin has proven versatile for the implementation of ArMs as it offers the following features: (i) it is an extremely robust protein scaffold, amenable to extensive genetic manipulation and mishandling, (ii) it can be expressed in E. coli to very high titers (up to >8 g·L-1 in fed-batch cultures), and (iii) the cavity surrounding the biotinylated cofactor is commensurate with the size of a typical metal-catalyzed transition state. Relying on a chemogenetic optimization strategy, varying the orientation and the nature of the biotinylated cofactor within genetically engineered streptavidin, 12 reactions have been reported by the Ward group thus far. Recent efforts within our group have focused on extending the ArM technology to create complex systems for integration into biological cascade reactions and in vivo. With the long-term goal of complementing in vivo natural enzymes with ArMs, we summarize herein three complementary research lines: (i) With the aim of mimicking complex cross-regulation mechanisms prevalent in metabolism, we have engineered enzyme cascades, including cross-regulated reactions, that rely on ArMs. These efforts highlight the remarkable (bio)compatibility and complementarity of ArMs with natural enzymes. (ii) Additionally, multiple-turnover catalysis in the cytoplasm of aerobic organisms was achieved with ArMs that are compatible with a glutathione-rich environment. This feat is demonstrated in HEK-293T cells that are engineered with a gene switch that is upregulated by an ArM equipped with a cell-penetrating module. (iii) Finally, ArMs offer the fascinating prospect of "endowing organometallic chemistry with a genetic memory." With this goal in mind, we have identified E. coli's periplasmic space and surface display to compartmentalize an ArM, while maintaining the critical phenotype-genotype linkage. This strategy offers a straightforward means to optimize by directed evolution the catalytic performance of ArMs. Five reactions have been optimized following these compartmentalization strategies: ruthenium-catalyzed olefin metathesis, ruthenium-catalyzed deallylation, iridium-catalyzed transfer hydrogenation, dirhodium-catalyzed cyclopropanation and carbene insertion in C-H bonds. Importantly, >100 turnovers were achieved with ArMs in E. coli whole cells, highlighting the multiple turnover catalytic nature of these systems.

Project description:The biotechnological revolution has made it possible to create enzymes for many reactions by directed evolution. However, because of the immense number of possibilities, the availability of enzymes that possess a basal level of the desired catalytic activity is a prerequisite for success. For new-to-nature reactions, artificial metalloenzymes (ARMs), which are rationally designed hybrids of proteins and catalytically active transition-metal complexes, can be such a starting point. This Account details our efforts toward the creation of ARMs for the catalysis of new-to-nature reactions. Key to our approach is the notion that the binding of substrates, that is, effective molarity, is a key component to achieving large accelerations in catalysis. For this reason, our designs are based on the multidrug resistance regulator LmrR, a dimeric transcription factor with a large, hydrophobic binding pocket at its dimer interface. In this pocket, there are two tryptophan moieties, which are important for promiscuous binding of planar hydrophobic conjugated compounds by π-stacking. The catalytic machinery is introduced either by the covalent linkage of a catalytically active metal complex or via the ligand or supramolecular assembly, taking advantage of the two central tryptophan moieties for noncovalent binding of transition-metal complexes. Designs based on the chemical modification of LmrR were successful in catalysis, but this approach proved too laborious to be practical. Therefore, expanded genetic code methodologies were used to introduce metal binding unnatural amino acids during LmrR biosynthesis in vivo. These ARMs have been successfully applied in Cu(II) catalyzed Friedel-Crafts alkylation of indoles. The extension to MDRs from the TetR family resulted in ARMs capable of providing the opposite enantiomer of the Friedel-Crafts product. We have employed a computationally assisted redesign of these ARMs to create a more active and selective artificial hydratase, introducing a glutamate as a general base at a judicious position so it can activate and direct the incoming water nucleophile. A supramolecularly assembled ARM from LmrR and copper(II)-phenanthroline was successful in Friedel-Crafts alkylation reactions, giving rise to up to 94% ee. Also, hemin was bound, resulting in an artificial heme enzyme for enantioselective cyclopropanation reactions. The importance of structural dynamics of LmrR was suggested by computational studies, which showed that the pore can open up to allow access of substrates to the catalytic iron center, which, according to the crystal structure, is deeply buried inside the protein. Finally, the assembly approaches were combined to introduce both a catalytic and a regulatory domain, resulting in an ARM that was specifically activated in the presence of Fe(II) salts but not Zn(II) salts. Our work demonstrates that LmrR is a privileged scaffold for ARM design: It allows for multiple assembly methods and even combinations of these, it can be applied in a variety of different catalytic reactions, and it shows significant structural dynamics that contribute to achieving the desired catalytic activity. Moreover, both the creation via expanded genetic code methods as well as the supramolecular assembly make LmrR-based ARMs highly suitable for achieving the ultimate goal of the integration of ARMs in biosynthetic pathways in vivo to create a hybrid metabolism.

Dataset Information

Enhanced Sequence-Activity Mapping and Evolution of Artificial Metalloenzymes by Active Learning.

Publications

Enhanced Sequence-Activity Mapping and Evolution of Artificial Metalloenzymes by Active Learning.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets