Directed Evolution Approach Research Articles

Navigating directed evolution efficiently: optimizing selection conditions and selection output analysis.

Directed evolution is a powerful tool that can bypass gaps in our understanding of the sequence-function relationship of proteins and still isolate variants with desired activities, properties, and substrate specificities. The rise of directed evolution platforms for polymerase engineering has accelerated the isolation of xenobiotic nucleic acid (XNA) synthetases and reverse transcriptases capable of processing a wide array of unnatural XNAs which have numerous therapeutic and biotechnological applications. Still, the current generation of XNA polymerases functions with significantly lower efficiency than the natural counterparts and retains a significant level of DNA polymerase activity which limits their in vivo applications. Although directed evolution approaches are continuously being developed and implemented to improve XNA polymerase engineering, the field lacks an in-depth analysis of the effect of selection parameters, library construction biases and sampling biases. Focusing on the directed evolution pipeline for DNA and XNA polymerase engineering, this work sets out a method for understanding the impact of selection conditions on selection success and efficiency. We also explore the influence of selection conditions on fidelity at the population and individual mutant level. Additionally, we explore the sequencing coverage requirements in directed evolution experiments, which differ from genome assembly and other -omics approaches. This analysis allowed us to identify the sequencing coverage threshold for the accurate and precise identification of significantly enriched mutants. Overall, this study introduces a robust methodology for optimizing selection protocols, which effectively streamlines selection processes by employing small libraries and cost-effective NGS sequencing. It provides valuable insights into critical considerations, thereby enhancing the overall effectiveness and efficiency of directed evolution strategies applicable to enzymes other than the ones considered here.

A combinatorially complete epistatic fitness landscape in an enzyme active site

Protein engineering often targets binding pockets or active sites which are enriched in epistasis-nonadditive interactions between amino acid substitutions-and where the combined effects of multiple single substitutions are difficult to predict. Few existing sequence-fitness datasets capture epistasis at large scale, especially for enzyme catalysis, limiting the development and assessment of model-guided enzyme engineering approaches. We present here a combinatorially complete, 160,000-variant fitness landscape across four residues in the active site of an enzyme. Assaying the native reaction of a thermostable β-subunit of tryptophan synthase (TrpB) in a nonnative environment yielded a landscape characterized by significant epistasis and many local optima. These effects prevent simulated directed evolution approaches from efficiently reaching the global optimum. There is nonetheless wide variability in the effectiveness of different directed evolution approaches, which together provide experimental benchmarks for computational and machine learning workflows. The most-fit TrpB variants contain a substitution that is nearly absent in natural TrpB sequences-a result that conservation-based predictions would not capture. Thus, although fitness prediction using evolutionary data can enrich in more-active variants, these approaches struggle to identify and differentiate among the most-active variants, even for this near-native function. Overall, this work presents a large-scale testing ground for model-guided enzyme engineering and suggests that efficient navigation of epistatic fitness landscapes can be improved by advances in both machine learning and physical modeling.

Directed evolution of bacteriophages: impacts of prolific prophage.

Directed evolution is a common strategy for evolving phages to expand host range, often targeting pathogenic strains of bacteria. In this study we investigated phage host-range expansion using directed evolution in the Pseudomonas aeruginosa system. We show that prophage are active players in directed evolution and can contribute to observation of host-range expansion. Since prophage are prevalent in bacterial hosts, particularly pathogenic strains of bacteria, and all directed evolution approaches involve iteratively propagating phage on one or more bacterial hosts, the presence of prophage in phage preparations is a factor that needs to be considered in experimental design and interpretation of results. These results highlight the importance of screening for prophages either genetically or through intraspecies antagonism assays during selection of bacterial strains and will contribute to improving experimental design of future directed evolution studies.

Comparative in vivo characterization of newly discovered myotropic adeno-associated vectors

BackgroundAdeno-associated virus (AAV)-based gene therapy is a promising strategy to treat muscle diseases. However, this strategy is currently confronted with challenges, including a lack of transduction efficiency across the entire muscular system and toxicity resulting from off-target tissue effects. Recently, novel myotropic AAVs named MyoAAVs and AAVMYOs have been discovered using a directed evolution approach, all separately demonstrating enhanced muscle transduction efficiency and liver de-targeting effects. However, these newly discovered AAV variants have not yet been compared.MethodsIn this study, we performed a comparative analysis of these various AAV9-derived vectors under the same experimental conditions following different injection time points in two distinct mouse strains.ResultsWe highlight differences in transduction efficiency between AAV9, AAVMYO, MyoAAV2A and MyoAAV4A that depend on age at injection, doses and mouse genetic background. In addition, specific AAV serotypes appeared more potent to transduce skeletal muscles including diaphragm and/or to de-target heart or liver.ConclusionsOur study provides guidance for researchers aiming to establish proof-of-concept approaches for preventive or curative perspectives in mouse models, to ultimately lead to future clinical trials for muscle disorders.

Machine learning model of the catalytic efficiency and substrate specificity of acyl-ACP thioesterase variants generated from natural and in vitro directed evolution.

Modulating the catalytic activity of acyl-ACP thioesterase (TE) is an important biotechnological target for effectively increasing flux and diversifying products of the fatty acid biosynthesis pathway. In this study, a directed evolution approach was developed to improve the fatty acid titer and fatty acid diversity produced by E. coli strains expressing variant acyl-ACP TEs. A single round of in vitro directed evolution, coupled with a high-throughput colorimetric screen, identified 26 novel acyl-ACP TE variants that convey up to a 10-fold increase in fatty acid titer, and generate altered fatty acid profiles when expressed in a bacterial host strain. These in vitro-generated variant acyl-ACP TEs, in combination with 31 previously characterized natural variants isolated from diverse phylogenetic origins, were analyzed with a random forest classifier machine learning tool. The resulting quantitative model identified 22 amino acid residues, which define important structural features that determine the catalytic efficiency and substrate specificity of acyl-ACP TE.

Directed-evolution approach to empower EGFR targeting for immunotherapy

Advances in directed-evolution technologies are enabling new strategies to isolate binding proteins that recognize disease-associated states of a target protein. In this issue of Cell Reports Methods, Dobersberger etal. devised a yeast display-based selection scheme to discover proteins that engage the cancer-associated activated state of a receptor to enable design of safe and effective immunotherapies.

Electrical-Driven Directed-Evolution of Copper Nanowires Catalysts for Efficient Nitrate Reduction to Ammonia.

The electrocatalytic conversion of nitrate (NO3 -) to NH3 (NO3RR) at ambient conditions offers a promising alternative to the Haber-Bosch process. The pivotal factors in optimizing the proficient conversion of NO3 - into NH3 include enhancing the adsorption capabilities of the intermediates on the catalyst surface and expediting the hydrogenation steps. Herein, the Cu/Cu2O/Pi NWs catalyst is designed based on the directed-evolution strategy to achieve an efficient reduction of NO3‾. Benefiting from the synergistic effect of the OV-enriched Cu2O phase developed during the directed-evolution process and the pristine Cu phase, the catalyst exhibits improved adsorption performance for diverse NO3RR intermediates. Additionally, the phosphate group anchored on the catalyst's surface during the directed-evolution process facilitates water electrolysis, thereby generating Hads on the catalyst surface and promoting the hydrogenation step of NO3RR. As a result, the Cu/Cu2O/Pi NWs catalyst shows an excellent FE for NH3 (96.6%) and super-high NH3 yield rate of 1.2molh-1gcat. -1 in 1m KOH and 0.1m KNO3 solution at -0.5V versus RHE. Moreover, the catalyst's stability is enhanced by the stabilizing influence of the phosphate group on the Cu2O phase. This work highlights the promise of a directed-evolution approach in designing catalysts for NO3RR.

Exploring the Protein Sequence Space with Global Generative Models.

Recent advancements in specialized large-scale architectures for training images and language have profoundly impacted the field of computer vision and natural language processing (NLP). Language models, such as the recent ChatGPT and GPT-4, have demonstrated exceptional capabilities in processing, translating, and generating human language. These breakthroughs have also been reflected in protein research, leading to the rapid development of numerous new methods in a short time, with unprecedented performance. Several of these models have been developed with the goal of generating sequences in novel regions of the protein space. In this work, we provide an overview of the use of protein generative models, reviewing (1) language models for the design of novel artificial proteins, (2) works that use non-transformer architectures, and (3) applications in directed evolution approaches.

Targeted In Vivo Mutagenesis in Yeast Using CRISPR/Cas9 and Hyperactive Cytidine and Adenine Deaminases.

Directed evolution is a preferred strategy to improve the function of proteins such as enzymes that act as bottlenecks in metabolic pathways. Common directed evolution approaches rely on error-prone PCR-based libraries where the number of possible variants is usually limited by cellular transformation efficiencies. Targeted in vivo mutagenesis can advance directed evolution approaches and help to overcome limitations in library generation. In the current study, we aimed to develop a high-efficiency time-controllable targeted mutagenesis toolkit in the yeast Saccharomyces cerevisiae by employing the CRISPR/Cas9 technology. To that end, we fused the dCas9 protein with hyperactive variants of adenine and cytidine deaminases aiming to create an inducible CRISPR-based mutagenesis tool targeting a specific DNA sequence in vivo with extended editing windows and high mutagenesis efficiency. We also investigated the effect of guide RNA multiplexing on the mutagenesis efficiency both phenotypically and on the DNA level.

The G285S mutation in nsP1 is sufficient to render Sindbis virus as a stable vector for gene delivery.

Neuroscience, gene therapy, and vaccine have all benefited from the increased use of viral vectors. Sindbis virus (SINV) is a notable candidate among these vectors. However, viral vectors commonly suffer from a loss of expression of the transgene, especially RNA viral vectors. In this study, we used a directed evolution approach by continuous passage of selection to identify adaptive mutations that help SINV to stably express exogenous genes. As a result, we found two adaptive mutations that are located at aa 285 (G to S) of nsP1 and aa 422 (D to G) of nsP2, respectively. Further study showed that G285S was sufficient for SINV to stabilize the expression of the inserted gene, while D422G was not. Combined with AlphaFold2 and sequence alignment with the genus Alphavirus, we found that G285S is conserved. Based on this mutation, we constructed a new vector for the applications in neural circuits mapping. Our results indicated that the mutant SINV maintained its anterograde transsynaptic transmission property. In addition, when the transgene was replaced by another gene, granulocyte-macrophage colony-stimulating factor (GM-CSF), the vector still showed stable expression of the inserted gene. Hence, using SINV as an example, we have demonstrated an efficient approach to greatly augment the gene delivery capacity of viral vectors, which will be useful to neuroscience and oncolytic therapy.

Group Selection as a Basis for Screening Mutagenized Libraries of Public Goods (Bacillus thuringiensis Cry Toxins).

The pesticidal toxins of Bacillus thuringiensis (Bt) supply the active proteins for genetically modified insect-resistant crops. There is therefore keen interest in finding new toxins, or improving known toxins, in order to increase the mortality of various targets. The production and screening of large libraries of mutagenized toxins are among the means of identifying improved toxins. Since Cry toxins are public goods, and do not confer advantages to producers in competition, conventional directed evolution approaches cannot be used here. Instead, thousands of individual mutants have to be sequenced and assayed individually, a costly and time-consuming process. In this study, we tested a group selection-based approach that could be used to screen an uncharacterized pool of Cry toxin mutants. This involved selecting for infectivity between subpopulations of Bt clones within metapopulations of infected insects in three rounds of passage. We also tested whether additional mutagenesis from exposure to ethyl methanesulfonate could increase infectivity or supply additional Cry toxin diversity during passage. Sequencing of pools of mutants at the end of selection showed that we could effectively screen out Cry toxin variants that had reduced toxicity with our group selection approach. The addition of extra mutagenesis during passage decreased the efficiency of selection for infectivity and did not produce any additional novel toxin diversity. Toxins with loss-of-function mutations tend to dominate mutagenized libraries, and so a process for screening out these mutants without time-consuming sequencing and characterization steps could be beneficial when applied to larger libraries. IMPORTANCE Insecticidal toxins from the bacterium Bacillus thuringiensis are widely exploited in genetically modified plants. This application creates a demand for novel insecticidal toxins that can be used to better manage resistant pests or control new or recalcitrant target species. An important means of producing novel toxins is via high-throughput mutagenesis and screening of existing toxins, a lengthy and resource-intensive process. This study describes the development and testing of an efficient means of screening a test library of mutagenized insecticidal toxins. Here, we showed that it is possible to screen out loss-of-function mutations with low infectivity within a pool without the need to characterize and sequence each mutant individually. This has the potential to improve the efficiency of processes used to identify novel proteins.

Restricting Promiscuity of Plant Flavonoid 3'-Hydroxylase and 4'-O-Methyltransferase Improves the Biosynthesis of (2S)-Hesperetin in E. coli.

Enzyme promiscuity is evolutionarily advantageous to plants for gaining new enzyme functions when adapting to environmental challenges. However, this promiscuity can negatively affect the expression of genes encoding for plant enzymes in microorganisms. Here, we show that refining the promiscuity of flavonoid 3'-hydroxylase (F3'H) and 4'-O-methyltransferase (F4'OMT) improves (2S)-hesperetin production in Escherichia coli. First, we employed inverse molecular docking to screen a highly substrate-specific ThF3'H from Tricyrtis hirta, which could selectively convert 100 mg L-1 (2S)-naringenin to (2S)-eriodictyol but not (2S)-isosakuranetin, with a cytochrome P450 reductase from Arabidopsis thaliana. Second, we employed a directed evolution approach to restrict the promiscuity of MpOMT from Mentha × piperita. The strain harboring the MpOMTS142V mutant presented a remarkably increased preference for (2S)-eriodictyol. Finally, 27.5 mg L-1 (2S)-hesperetin was produced, while only minor amounts of (2S)-eriodictyol and (2S)-isosakuranetin accumulated as byproducts. This value represents a 14-fold increase in (2S)-hesperetin compared to the parental strain, along with a dramatic reduction in side products. Our work highlights the benefit of alleviating the promiscuity of plant enzymes when engineering production of natural products by microbial cell factories.

Machine learning optimization of candidate antibody yields highly diverse sub-nanomolar affinity antibody libraries

Therapeutic antibodies are an important and rapidly growing drug modality. However, the design and discovery of early-stage antibody therapeutics remain a time and cost-intensive endeavor. Here we present an end-to-end Bayesian, language model-based method for designing large and diverse libraries of high-affinity single-chain variable fragments (scFvs) that are then empirically measured. In a head-to-head comparison with a directed evolution approach, we show that the best scFv generated from our method represents a 28.7-fold improvement in binding over the best scFv from the directed evolution. Additionally, 99% of designed scFvs in our most successful library are improvements over the initial candidate scFv. By comparing a library’s predicted success to actual measurements, we demonstrate our method’s ability to explore tradeoffs between library success and diversity. Results of our work highlight the significant impact machine learning models can have on scFv development. We expect our method to be broadly applicable and provide value to other protein engineering tasks.

Toward high-throughput engineering techniques for improving CAR intracellular signaling domains.

Chimeric antigen receptors (CAR) are generated by linking extracellular antigen recognition domains with one or more intracellular signaling domains derived from the T-cell receptor complex or various co-stimulatory receptors. The choice and relative positioning of signaling domains help to determine chimeric antigen receptors T-cell activity and fate in vivo. While prior studies have focused on optimizing signaling power through combinatorial investigation of native intracellular signaling domains in modular fashion, few have investigated the prospect of sequence engineering within domains. Here, we sought to develop a novel in situ screening method that could permit deployment of directed evolution approaches to identify intracellular domain variants that drive selective induction of transcription factors. To accomplish this goal, we evaluated a screening approach based on the activation of a human NF-κB and NFAT reporter T-cell line for the isolation of mutations that directly impact T cell activation in vitro. As a proof-of-concept, a model library of chimeric antigen receptors signaling domain variants was constructed and used to demonstrate the ability to discern amongst chimeric antigen receptors containing different co-stimulatory domains. A rare, higher-signaling variant with frequency as low as 1 in 1000 could be identified in a high throughput setting. Collectively, this work highlights both prospects and limitations of novel mammalian display methods for chimeric antigen receptors signaling domain discovery and points to potential strategies for future chimeric antigen receptors development.

Calcium-dependent protein folding in a designed molecular switch.

Allosteric regulation of protein activity is a fundamental principle for the temporal and spatial control of cellular function. Transfer of such principles to rational design of proteins will pave the way for biotechnological applications were control of a given process with external cues is desirable. Through a semi-rational and directed evolution approach we have been able to design a calcium dependent molecular switch with distinct “on” and “off” states decided by the presence and absence of calcium, respectively. The design was established in the context of protein-protein interactions with antibodies which is a massive biotechnological application linked to purification of therapeutic antibodies. Our approach was to introduce a randomized calcium binding loop into the C2 domain of Streptococcal Protein G. The large ensemble of different sequences was displayed on the surface of E. coli and subjected to selective pressures for binding to a human FAB in the presence but not in the absence of calcium. From this directed evolution experiment we discovered evolved variants that contained calcium switches with distinct “on” and “off” behavior towards FAB binding. The molecular mechanism underlying the calcium switch was discovered from quantification of both structure and dynamics with NMR spectroscopy. We found the designed protein was partially disordered in the absence of calcium, and that the disordered segment corresponded to the calcium loop and part of the FAB interaction surface of the parental C2 domain. In presence of calcium both the calcium binding loop and the FAB surface became fully structured and as a consequence the FAB binding activity was restored. Therefore, the calcium-switch in our designed protein is dictated by a “coupled folding and binding” mechanism, a principle that has evolved over and over again under natural selection in for instance intrinsically disordered proteins.

A trans-amplifying RNA simplified to essential elements is highly replicative and robustly immunogenic in mice

Trans-amplifying RNA (taRNA) is a split-vector derivative of self-amplifying RNA (saRNA) and a promising vaccine platform. taRNA combines a non-replicating mRNA encoding an alphaviral replicase and a transreplicon (TR) RNA coding for the antigen. Upon translation, the replicase amplifies the antigen-coding TR, thereby requiring minimal amounts of TR for immunization. TR amplification by the replicase follows a complex mechanism orchestrated by genomic and subgenomic promoters (SGPs) and generates genomic and subgenomic amplicons whereby only the latter are translated into therapeutic proteins. This complexity merits simplification to improve the platform. Here, we eliminated the SGP and redesigned the 5' untranslated region to shorten the TR (STR), thereby enabling translation of the remaining genomic amplicon. We then applied a directed evolution approach to select for faster replicating STRs. The resulting evolved STR (eSTR) had acquired A-rich 5' extensions, which improved taRNA expression thanks to accelerated replication. Consequently, we reduced the minimal required TR amount by more than 10-fold without losing taRNA expression invitro. Accordingly, eSTR-immunized mice developed greater antibody titers to taRNA-encoded influenza HA than TR-immunized mice. In summary, this work points the way for further optimization of taRNA by combining rational design and directed evolution.

A phylogeny-based directed evolution approach to boost the synthetic applications of glycosyltransferases

Enzymatic catalysis in mild conditions plays a pivotal role in green chemistry, and the mining and the directed evolution of novel enzymes are crucial for the utilization of efficient biocatalysts.

Protein engineering via Bayesian optimization-guided evolutionary algorithm and robotic experiments.

Directed protein evolution applies repeated rounds of genetic mutagenesis and phenotypic screening and is often limited by experimental throughput. Through in silico prioritization of mutant sequences, machine learning has been applied to reduce wet lab burden to a level practical for human researchers. On the other hand, robotics permits large batches and rapid iterations for protein engineering cycles, but such capacities have not been well exploited in existing machine learning-assisted directed evolution approaches. Here, we report a scalable and batched method, Bayesian Optimization-guided EVOlutionary (BO-EVO) algorithm, to guide multiple rounds of robotic experiments to explore protein fitness landscapes of combinatorial mutagenesis libraries. We first examined various design specifications based on an empirical landscape of protein G domain B1. Then, BO-EVO was successfully generalized to another empirical landscape of an Escherichia coli kinase PhoQ, as well as simulated NK landscapes with up to moderate epistasis. This approach was then applied to guide robotic library creation and screening to engineer enzyme specificity of RhlA, a key biosynthetic enzyme for rhamnolipid biosurfactants. A 4.8-fold improvement in producing a target rhamnolipid congener was achieved after examining less than 1% of all possible mutants after four iterations. Overall, BO-EVO proves to be an efficient and general approach to guide combinatorial protein engineering without prior knowledge.

Improvement of thermoalkaliphilic laccase (CtLac) by a directed evolution and application to lignin degradation

Thermoalkaliphilic laccase (CtLac) from the Caldalkalibacillus thermarum strain TA2.A1 has advantageous properties with potential industrial applications, such as high enzyme activity and stability at 70 °C and pH 8.0. In the present study, a directed evolution approach using a combination of random and site-directed mutagenesis was adopted to enhance the laccase activity of CtLac. Spectrophotometric assay and real-time oxygen measurement techniques were employed to compare and evaluate the enzyme activity among mutants. V243 was targeted for site-directed mutagenesis based on library screening. V243D showed a 25–35% higher laccase activity than wild-type CtLac in the spectrophotometric assay and oxygen consumption measurement results. V243D also showed higher catalytic efficiency than wild-type CtLac with decreased Km and increased kcat values. In addition, V243D enhanced oxidative degradation of the lignin model compound, guaiacylglycerol-β-guaiacyl ether (GGGE), by 10% and produced a 5–30% increase in high-value aldehydes than wild-type CtLac under optimal enzymatic conditions (i.e., 70 °C and pH 8.0). Considering the lack of protein structural information, we used the directed evolution approach to predict Val at the 243 position of CtLac as one of the critical amino acids contributing to the catalytic efficiency of the enzyme. Moreover, it found that the real-time oxygen measurement technique could overcome the limitations of the spectrophotometric assay, and apply to evaluate oxidase activity in mutagenesis research.Key points• CtLac was engineered for enhanced laccase activity through directed evolution approach• V243D showed higher catalytic efficiency (kcat/Km) than wild-type CtLac• V243D produced higher amounts of high-value aldehydes from rice straw than wild-type CtLac

Removing carbon catabolite repression in Parageobacillus thermoglucosidasius DSM 2542.

Parageobacillus thermoglucosidasius is a thermophilic bacterium of interest for lignocellulosic biomass fermentation. However, carbon catabolite repression (CCR) hinders co-utilization of pentoses and hexoses in the biomass substrate. Hence, to optimize the fermentation process, it is critical to remove CCR in the fermentation strains with minimal fitness cost. In this study, we investigated whether CCR could be removed from P. thermoglucosidasius DSM 2542 by mutating the Ser46 regulatory sites on HPr and Crh to a non-reactive alanine residue. It was found that neither the ptsH1 (HPr-S46A) nor the crh1 (Crh-S46A) mutation individually eliminated CCR in P. thermoglucosidasius DSM 2542. However, it was not possible to generate a ptsH1 crh1 double mutant. While the Crh-S46A mutation had no obvious fitness effect in DSM 2542, the ptsH1 mutation had a negative impact on cell growth and sugar utilization under fermentative conditions. Under these conditions, the ptsH1 mutation was associated with the production of a brown pigment, believed to arise from methylglyoxal production, which is harmful to cells. Subsequently, a less directed adaptive evolution approach was employed, in which DSM 2542 was grown in a mixture of 2-deoxy-D-glucose(2-DG) and xylose. This successfully removed CCR from P. thermoglucosidasius DSM 2542. Two selection strategies were applied to optimize the phenotypes of evolved strains. Genome sequencing identified key mutations affecting the PTS components PtsI and PtsG, the ribose operon repressor RbsR and adenine phosphoribosyltransferase APRT. Genetic complementation and bioinformatics analysis revealed that the presence of wild type rbsR and apt inhibited xylose uptake or utilization, while ptsI and ptsG might play a role in the regulation of CCR in P. thermoglucosidasius DSM 2542.