Flattening the curve-How to get better results with small deep-mutational-scanning datasets.

Proteins are used in various biotechnological applications, often requiring the optimization of protein properties by introducing specific amino-acid exchanges. Deep mutational scanning (DMS) is an effective high-throughput method for evaluating the effects of these exchanges on protein function. DMS data can then inform the training of a neural network to predict the impact of mutations. Most approaches use some representation of the protein sequence for training and prediction. As proteins are characterized by complex structures and intricate residue interaction networks, directly providing structural information as input reduces the need to learn these features from the data. We introduce a method for encoding protein structures as stacked 2D contact maps, which capture residue interactions, their evolutionary conservation, and mutation-induced interaction changes. Furthermore, we explored techniques to augment neural network training performance on smaller DMS datasets. To validate our approach, we trained three neural network architectures originally used for image analysis on three DMS datasets, and we compared their performances with networks trained solely on protein sequences. The results confirm the effectiveness of the protein structure encoding in machine learning efforts on DMS data. Using structural representations as direct input to the networks, along with data augmentation and pretraining, significantly reduced demands on training data size and improved prediction performance, especially on smaller datasets, while performance on large datasets was on par with state-of-the-art sequence convolutional neural networks. The methods presented here have the potential to provide the same workflow as DMS without the experimental and financial burden of testing thousands of mutants. Additionally, we present an open-source, user-friendly software tool to make these data analysis techniques accessible, particularly to biotechnology and protein engineering researchers who wish to apply them to their mutagenesis data.

Click, Compute, Create: A Review of Web-based Tools for Enzyme Engineering.

Enzyme engineering, though pivotal across various biotechnological domains, is often plagued by its time-consuming and labor-intensive nature. This review aims to offer an overview of supportive in silico methodologies for this demanding endeavor. Starting from methods to predict protein structures, to classification of their activity and even the discovery of new enzymes we continue with describing tools used to increase thermostability and production yields of selected targets. Subsequently, we discuss computational methods to modulate both, the activity as well as selectivity of enzymes. Last, we present recent approaches based on cutting-edge machine learning methods to redesign enzymes. With exception of the last chapter, there is a strong focus on methods easily accessible via web-interfaces or simple Python-scripts, therefore readily useable for a diverse and broad community.

Enzyme Machinery for Bacterial Glucoside Metabolism through a Conserved Non-hydrolytic Pathway.

The flexible acquisition of substrates from nutrient pools is critical for microbes to prevail in competitive environments. To acquire glucose from diverse glycoside and disaccharide substrates, many free-living and symbiotic bacteria have developed, alongside hydrolysis, a non-hydrolytic pathway comprised of four biochemical steps and conferred from a single glycoside utilization gene locus (GUL). Mechanistically, this pathway integrates within the framework of oxidation and reduction at the glucosyl/glucose C3, the eliminative cleavage of the glycosidic bond and the addition of water in two consecutive lyase-catalyzed reactions. Here, based on study of enzymes from the phytopathogen Agrobacterium tumefaciens, we reveal a conserved Mn2+ metallocenter active site in both lyases and identify the structural requirements for specific catalysis to elimination of 3-keto-glucosides and water addition to the resulting 2-hydroxy-3-keto-glycal product, yielding 3-keto-glucose. Extending our search of GUL-encoded putative lyases to the human gut commensal Bacteroides thetaiotaomicron, we discover a Ca2+ metallocenter active site in a putative glycoside hydrolase-like protein and demonstrate its catalytic function in the eliminative cleavage of 3-keto-glucosides of opposite (α) anomeric configuration as preferred by the A. tumefaciens enzyme (ß). Structural and biochemical comparisons reveal the molecular-mechanistic origin of 3-keto-glucoside lyase stereo-complementarity. Our findings identify a basic set of GUL-encoded lyases for glucoside metabolism and assign physiological significance to GUL genetic diversity in the bacterial domain of life.

Sterol interactions influence the function of Wsc sensors.

The Wsc1, Wsc2, and Wsc3 proteins are essential cell surface sensors that respond to cell wall perturbation by activating the cell wall integrity pathway (CWIP). We show here that in situ production of cholesterol (in place of ergosterol) induces hyper-phosphorylation of Slt2, the MAPK of the CWIP, and upregulates cell wall biosynthesis. Deletion of all three Wsc genes in K. phaffii reverts these phenotypes. In the cholesterol-producing strain, both Wsc1 and Wsc3 accumulate in the plasma membrane. Close inspection of the transmembrane domains of all three Wsc proteins predicted by AlphaFold2 revealed the presence of CRAC sterol-binding motifs. Experiments using a photoreactive cholesterol derivative indicate intimate interaction of this sterol with the Wsc transmembrane domain, and this apparent sterol binding was abrogated in Wsc mutants with substitutions in the CRAC motif. We also observed cholesterol interaction with CRAC-like motifs in the transmembrane domains of mammalian integrins, analogs of Wsc proteins. Our results suggest that proper signaling of the Wsc sensors requires highly specific binding of the native endogenous terminal sterol, ergosterol.

Derivatives of Natural Organocatalytic Cofactors and Artificial Organocatalytic Cofactors as Catalysts in Enzymes.

Catalytically active non-metal cofactors in enzymes carry out a variety of different reactions. The efforts to develop derivatives of naturally occurring cofactors such as flavins or pyridoxal phosphate and the advances to design new, non-natural cofactors are reviewed here. We report the status quo for enzymes harboring organocatalysts as derivatives of natural cofactors or as artificial ones and their application in the asymmetric synthesis of various compounds.

Computational backbone design enables soluble engineering of transferrin receptor apical domain.

Supply of iron into human cells is achieved by iron carrier protein transferrin and its receptor that upon complex formation get internalized by endocytosis. Similarly, the iron needs to be delivered into the brain, and necessitates the transport across the blood-brain barrier. While there are still unanswered questions about these mechanisms, extensive efforts have been made to use the system for delivery of therapeutics into biological compartments. The dimeric form of the receptor, where each subunit consists of three domains, further complicates the detailed investigation of molecular determinants responsible for guiding the receptor interactions with other proteins. Especially the apical domain's biological function has been elusive. To further the study of transferrin receptor, we have computationally decoupled the apical domain for soluble expression, and validated the design strategy by structure determination. Besides presenting a methodology for solubilizing domains, the results will allow for study of apical domain's function.

Computational design of a homotrimeric metalloprotein with a trisbipyridyl core.

Metal-chelating heteroaryl small molecules have found widespread use as building blocks for coordination-driven, self-assembling nanostructures. The metal-chelating noncanonical amino acid (2,2'-bipyridin-5yl)alanine (Bpy-ala) could, in principle, be used to nucleate specific metalloprotein assemblies if introduced into proteins such that one assembly had much lower free energy than all alternatives. Here we describe the use of the Rosetta computational methodology to design a self-assembling homotrimeric protein with [Fe(Bpy-ala)3]2+ complexes at the interface between monomers. X-ray crystallographic analysis of the homotrimer showed that the design process had near-atomic-level accuracy: The all-atom rmsd between the design model and crystal structure for the residues at the protein interface is ∼1.4 Å. These results demonstrate that computational protein design together with genetically encoded noncanonical amino acids can be used to drive formation of precisely specified metal-mediated protein assemblies that could find use in a wide range of photophysical applications.

Improved molecular replacement by density- and energy-guided protein structure optimization.

Molecular replacement procedures, which search for placements of a starting model within the crystallographic unit cell that best account for the measured diffraction amplitudes, followed by automatic chain tracing methods, have allowed the rapid solution of large numbers of protein crystal structures. Despite extensive work, molecular replacement or the subsequent rebuilding usually fail with more divergent starting models based on remote homologues with less than 30% sequence identity. Here we show that this limitation can be substantially reduced by combining algorithms for protein structure modelling with those developed for crystallographic structure determination. An approach integrating Rosetta structure modelling with Autobuild chain tracing yielded high-resolution structures for 8 of 13 X-ray diffraction data sets that could not be solved in the laboratories of expert crystallographers and that remained unsolved after application of an extensive array of alternative approaches. We estimate that the new method should allow rapid structure determination without experimental phase information for over half the cases where current methods fail, given diffraction data sets of better than 3.2 Å resolution, four or fewer copies in the asymmetric unit, and the availability of structures of homologous proteins with >20% sequence identity.

Structural and functional characterization of NikO, an enolpyruvyl transferase essential in nikkomycin biosynthesis.

Nikkomycins are peptide-nucleoside compounds with fungicidal, acaricidal, and insecticidal properties because of their strong inhibition of chitin synthase. Thus, they are potential antibiotics especially for the treatment of immunosuppressed patients, for those undergoing chemotherapy, or after organ transplants. Although their chemical structure has been known for more than 30 years, only little is known about their complex biosynthesis. The genes encoding for proteins involved in the biosynthesis of the nucleoside moiety of nikkomycins are co-transcribed in the same operon, comprising the genes nikIJKLMNO. The gene product NikO was shown to belong to the family of enolpyruvyl transferases and to catalyze the transfer of an enolpyruvyl moiety from phosphoenolpyruvate to the 3'-hydroxyl group of UMP. Here, we report activity and inhibition studies of the wild-type enzyme and the variants C130A and D342A. The x-ray crystal structure revealed differences between NikO and its homologs. Furthermore, our studies led to conclusions concerning substrate binding and preference as well as to conclusions about inhibition/alkylation by the antibiotic fosfomycin.

Vascular bioactivation of nitroglycerin by aldehyde dehydrogenase-2: reaction intermediates revealed by crystallography and mass spectrometry.

Aldehyde dehydrogenase-2 (ALDH2) catalyzes the bioactivation of nitroglycerin (glyceryl trinitrate, GTN) in blood vessels, resulting in vasodilation by nitric oxide (NO) or a related species. Because the mechanism of this reaction is still unclear we determined the three-dimensional structures of wild-type (WT) ALDH2 and of a triple mutant of the protein that exhibits low denitration activity (E268Q/C301S/C303S) in complex with GTN. The structure of the triple mutant showed that GTN binds to the active site via polar contacts to the oxyanion hole and to residues 268 and 301 as well as by van der Waals interactions to hydrophobic residues of the catalytic pocket. The structure of the GTN-soaked wild-type protein revealed a thionitrate adduct to Cys-302 as the first reaction intermediate, which was also found by mass spectrometry (MS) experiments. In addition, the MS data identified sulfinic acid as the irreversibly inactivated enzyme species. Assuming that the structures of the triple mutant and wild-type ALDH2 reflect binding of GTN to the catalytic site and the first reaction step, respectively, superposition of the two structures indicates that denitration of GTN is initiated by nucleophilic attack of Cys-302 at one of the terminal nitrate groups, resulting in formation of the observed thionitrate intermediate and release of 1,2-glyceryl dinitrate. Our results shed light on the molecular mechanism of the GTN denitration reaction and provide useful information on the structural requirements for high affinity binding of organic nitrates to the catalytic site of ALDH2.

The structure of glycerol trinitrate reductase NerA from Agrobacterium radiobacter reveals the molecular reason for nitro- and ene-reductase activity in OYE homologues.

In recent years, Old Yellow Enzymes (OYEs) and their homologues have found broad application in the efficient asymmetric hydrogenation of activated C=C bonds with high selectivities and yields. Members of this class of enzymes have been found in many different organisms and are rather diverse on the sequence level, with pairwise identities as low as 20 %, but they exhibit significant structural similarities with the adoption of a conserved (αß)(8)-barrel fold. Some OYEs have been shown not only to reduce C=C double bonds, but also to be capable of reducing nitro groups in both saturated and unsaturated substrates. In order to understand this dual activity we determined and analyzed X-ray crystal structures of NerA from Agrobacterium radiobacter, both in its apo form and in complex with 4-hydroxybenzaldehyde and with 1-nitro-2-phenylpropene. These structures, together with spectroscopic studies of substrate binding to several OYEs, indicate that nitro-containing substrates can bind to OYEs in different binding modes, one of which leads to C=C double bond reduction and the other to nitro group reduction.

Fusion of binding domains to Thermobifida cellulosilytica cutinase to tune sorption characteristics and enhancing PET hydrolysis.

A cutinase from Thermomyces cellullosylitica (Thc_Cut1), hydrolyzing the synthetic polymer polyethylene terephthalate (PET), was fused with two different binding modules to improve sorption and thereby hydrolysis. The binding modules were from cellobiohydrolase I from Hypocrea jecorina (CBM) and from a polyhydroxyalkanoate depolymerase from Alcaligenes faecalis (PBM). Although both binding modules have a hydrophobic nature, it was possible to express the proteins in E. coli . Both fusion enzymes and the native one had comparable kcat values in the range of 311 to 342 s(-1) on pNP-butyrate, while the catalytic efficiencies kcat/Km decreased from 0.41 s(-1)/ µM (native enzyme) to 0.21 and 0.33 s(-1)/µM for Thc_Cut1+PBM and Thc_Cut1+CBM, respectively. The fusion enzymes were active both on the insoluble PET model substrate bis(benzoyloxyethyl) terephthalate (3PET) and on PET although the hydrolysis pattern was differed when compared to Thc_Cut1. Enhanced adsorption of the fusion enzymes was visible by chemiluminescence after incubation with a 6xHisTag specific horseradish peroxidase (HRP) labeled probe. Increased adsorption to PET by the fusion enzymes was confirmed with Quarz Crystal Microbalance (QCM-D) analysis and indeed resulted in enhanced hydrolysis activity (3.8× for Thc_Cut1+CBM) on PET, as quantified, based on released mono/oligomers.

Essential Functional Interplay of the Catalytic Groups in Acid Phosphatase.

The cooperative interplay between the functional devices of a preorganized active site is fundamental to enzyme catalysis. An in-depth understanding of this phenomenon is central to elucidating the remarkable efficiency of natural enzymes and provides an essential benchmark for enzyme design and engineering. Here, we study the functional interconnectedness of the catalytic nucleophile (His18) in an acid phosphatase by analyzing the consequences of its replacement with aspartate. We present crystallographic, biochemical, and computational evidence for a conserved mechanistic pathway via a phospho-enzyme intermediate on Asp18. Linear free-energy relationships for phosphoryl transfer from phosphomonoester substrates to His18/Asp18 provide evidence for the cooperative interplay between the nucleophilic and general-acid catalytic groups in the wild-type enzyme, and its substantial loss in the H18D variant. As an isolated factor of phosphatase efficiency, the advantage of a histidine compared to an aspartate nucleophile is ∼104-fold. Cooperativity with the catalytic acid adds ≥102-fold to that advantage. Empirical valence bond simulations of phosphoryl transfer from glucose 1-phosphate to His and Asp in the enzyme explain the loss of activity of the Asp18 enzyme through a combination of impaired substrate positioning in the Michaelis complex, as well as a shift from early to late protonation of the leaving group in the H18D variant. The evidence presented furthermore suggests that the cooperative nature of catalysis distinguishes the enzymatic reaction from the corresponding reaction in solution and is enabled by the electrostatic preorganization of the active site. Our results reveal sophisticated discrimination in multifunctional catalysis of a highly proficient phosphatase active site.

Structure of the Reductase Domain of a Fungal Carboxylic Acid Reductase and Its Substrate Scope in Thioester and Aldehyde Reduction.

The synthesis of aldehydes from carboxylic acids has long been a challenge in chemistry. In contrast to the harsh chemically driven reduction, enzymes such as carboxylic acid reductases (CARs) are considered appealing biocatalysts for aldehyde production. Although structures of single- and didomains of microbial CARs have been reported, to date no full-length protein structure has been elucidated. In this study, we aimed to obtain structural and functional information regarding the reductase (R) domain of a CAR from the fungus Neurospora crassa (Nc). The NcCAR R-domain revealed activity for N-acetylcysteamine thioester (S-(2-acetamidoethyl) benzothioate), which mimics the phosphopantetheinylacyl-intermediate and can be anticipated as the minimal substrate for thioester reduction by CARs. The determined crystal structure of the NcCAR R-domain reveals a tunnel that putatively harbors the phosphopantetheinylacyl-intermediate, which is in good agreement with docking experiments performed with the minimal substrate. In vitro studies were performed with this highly purified R-domain and NADPH, demonstrating carbonyl reduction activity. The R-domain was able to accept not only a simple aromatic ketone but also benzaldehyde and octanal, which are typically considered to be the final product of carboxylic acid reduction by CAR. Also, the full-length NcCAR reduced aldehydes to primary alcohols. In conclusion, aldehyde overreduction can no longer be attributed exclusively to the host background.

A local platform for user-friendly FAIR data management and reproducible analytics.

Collaborative research is common practice in modern life sciences. For most projects several researchers from multiple universities collaborate on a specific topic. Frequently, these research projects produce a wealth of data that requires central and secure storage, which should also allow for easy sharing among project participants. Only under best circumstances, this comes with minimal technical overhead for the researchers. Moreover, the need for data to be analyzed in a reproducible way often poses a challenge for researchers without a data science background and thus represents an overly time-consuming process. Here, we report on the integration of CyVerse Austria (CAT), a new cyberinfrastructure for a local community of life science researchers, and provide two examples how it can be used to facilitate FAIR data management and reproducible analytics for teaching and research. In particular, we describe in detail how CAT can be used (i) as a teaching platform with a defined software environment and data management/sharing possibilities, and (ii) to build a data analysis pipeline using the Docker technology tailored to the needs and interests of the researcher.

A Novel High-Throughput Nanopore-Sequencing-Based Strategy for Rapid and Automated S-Protein Typing of SARS-CoV-2 Variants.

Rapid molecular surveillance of SARS-CoV-2 S-protein variants leading to immune escape and/or increased infectivity is of utmost importance. Among global bottlenecks for variant monitoring in diagnostic settings are sequencing and bioinformatics capacities. In this study, we aimed to establish a rapid and user-friendly protocol for high-throughput S-gene sequencing and subsequent automated identification of variants. We designed two new primer pairs to amplify only the immunodominant part of the S-gene for nanopore sequencing. Furthermore, we developed an automated "S-Protein-Typer" tool that analyzes and reports S-protein mutations on the amino acid level including a variant of concern indicator. Validation of our primer panel using SARS-CoV-2-positive respiratory specimens covering a broad Ct range showed successful amplification for 29/30 samples. Restriction to the region of interest freed sequencing capacity by a factor of 12-13, compared with whole-genome sequencing. Using either the MinION or Flongle flow cell, our sequencing strategy reduced the time required to identify SARS-CoV-2 variants accordingly. The S-Protein-Typer tool identified all mutations correctly when challenged with our sequenced samples and 50 deposited sequences covering all VOCs (December 2021). Our proposed S-protein variant screening offers a simple, more rapid, and low-cost entry into NGS-based SARS-CoV-2 analysis, compared with current whole-genome approaches.

Epoxide-hydrolase-initiated hydrolysis/rearrangement cascade of a methylene-interrupted bis-epoxide yields chiral THF moieties without involvement of a "cyclase".

In contrast with electrophilic enzyme-catalysed cyclisations in terpenoid biosynthesis, cyclisations of tetrahydrofuran moieties found in several groups of natural products, such as annonaceous acetogenins, neurofurans and phytooxylipins, appear to proceed through a nucleophilic cascade mechanism starting from bis-epoxy fatty acid precursors. This hypothesis was verified by epoxide-hydrolase-catalysed hydrolytic ring-opening/cyclisation cascades starting from a methylene-interrupted meso-bis-epoxide model substrate, which furnished the corresponding THF products with excellent de and ee values. Molecular modelling showed that the points of enzyme attack were consistent with the stereospecificities of the enzymes, whereas the stereochemical courses of the cyclisation were solely governed by Baldwin's rules and did not invoke the involvements of a "cyclase".

De novo design of a homo-trimeric amantadine-binding protein.

The computational design of a symmetric protein homo-oligomer that binds a symmetry-matched small molecule larger than a metal ion has not yet been achieved. We used de novo protein design to create a homo-trimeric protein that binds the C3 symmetric small molecule drug amantadine with each protein monomer making identical interactions with each face of the small molecule. Solution NMR data show that the protein has regular three-fold symmetry and undergoes localized structural changes upon ligand binding. A high-resolution X-ray structure reveals a close overall match to the design model with the exception of water molecules in the amantadine binding site not included in the Rosetta design calculations, and a neutron structure provides experimental validation of the computationally designed hydrogen-bond networks. Exploration of approaches to generate a small molecule inducible homo-trimerization system based on the design highlight challenges that must be overcome to computationally design such systems.

De novo design of a non-local ß-sheet protein with high stability and accuracy.

ß-sheet proteins carry out critical functions in biology, and hence are attractive scaffolds for computational protein design. Despite this potential, de novo design of all-ß-sheet proteins from first principles lags far behind the design of all-α or mixed-αß domains owing to their non-local nature and the tendency of exposed ß-strand edges to aggregate. Through study of loops connecting unpaired ß-strands (ß-arches), we have identified a series of structural relationships between loop geometry, side chain directionality and ß-strand length that arise from hydrogen bonding and packing constraints on regular ß-sheet structures. We use these rules to de novo design jellyroll structures with double-stranded ß-helices formed by eight antiparallel ß-strands. The nuclear magnetic resonance structure of a hyperthermostable design closely matched the computational model, demonstrating accurate control over the ß-sheet structure and loop geometry. Our results open the door to the design of a broad range of non-local ß-sheet protein structures.

De novo design of self-assembling helical protein filaments.

We describe a general computational approach to designing self-assembling helical filaments from monomeric proteins and use this approach to design proteins that assemble into micrometer-scale filaments with a wide range of geometries in vivo and in vitro. Cryo-electron microscopy structures of six designs are close to the computational design models. The filament building blocks are idealized repeat proteins, and thus the diameter of the filaments can be systematically tuned by varying the number of repeat units. The assembly and disassembly of the filaments can be controlled by engineered anchor and capping units built from monomers lacking one of the interaction surfaces. The ability to generate dynamic, highly ordered structures that span micrometers from protein monomers opens up possibilities for the fabrication of new multiscale metamaterials.