Structural and time-resolved mechanistic investigations of protein hydrolysis by the acidic proline-specific endoprotease from Aspergillus niger.

Proline-specific endoproteases have been successfully used in, for example, the in-situ degradation of gluten, the hydrolysis of bitter peptides, the reduction of haze during beer production, and the generation of peptides for mass spectroscopy and proteomics applications. Here we present the crystal structure of the extracellular proline-specific endoprotease from Aspergillus niger (AnPEP), a member of the S28 peptidase family with rarely observed true proline-specific endoprotease activity. Family S28 proteases have a conventional Ser-Asp-His catalytic triad, but their oxyanion-stabilizing hole shows a glutamic acid, an amino acid not previously observed in this role. Since these enzymes have an acidic pH optimum, the presence of a glutamic acid in the oxyanion hole may confine their activity to an acidic pH. Yet, considering the presence of the conventional catalytic triad, it is remarkable that the A. niger enzyme remains active down to pH 1.5. The determination of the primary cleavage site of cytochrome c along with molecular dynamics-assisted docking studies indicate that the active site pocket of AnPEP can accommodate a reverse turn of approximately 12 amino acids with proline at the S1 specificity pocket. Comparison with the structures of two S28-proline-specific exopeptidases reveals not only a more spacious active site cavity but also the absence of any putative binding sites for amino- and carboxyl-terminal residues as observed in the exopeptidases, explaining AnPEP's observed endoprotease activity.

Computational prediction of ω-transaminase selectivity by deep learning analysis of molecular dynamics trajectories.

We previously presented a computational protocol to predict the enzymatic (enantio)selectivity of an ω-transaminase towards a set of ligands (Ramírez-Palacios et al. (2021) Journal of Chemical Information and Modeling 61(11), 5569-5580) by counting the number of binding poses present in molecular dynamics (MD) simulations that met a defined set of geometric criteria. The geometric criteria consisted of a hand-crafted set of distances, angles and dihedrals deemed to be important for the enzymatic reaction to take place. In this work, the MD trajectories are reanalysed using a deep-learning approach to predict the enantiopreference of the enzyme without the need for hand-crafted criteria. We show that a convolutional neural network is capable of classifying the trajectories as belonging to the 'reactive' or 'non-reactive' enantiomer (binary classification) with a good accuracy (>0.90). The new method reduces the computational cost of the methodology, because it does not necessitate the sampling approach from the previous work. We also show that analysing how neural networks reach specific decisions can aid hand-crafted approaches (e.g. definition of near-attack conformations, or binding poses).

Super High-Throughput Screening of Enzyme Variants by Spectral Graph Convolutional Neural Networks.

Finding new enzyme variants with the desired substrate scope requires screening through a large number of potential variants. In a typical in silico enzyme engineering workflow, it is possible to scan a few thousands of variants, and gather several candidates for further screening or experimental verification. In this work, we show that a Graph Convolutional Neural Network (GCN) can be trained to predict the binding energy of combinatorial libraries of enzyme complexes using only sequence information. The GCN model uses a stack of message-passing and graph pooling layers to extract information from the protein input graph and yield a prediction. The GCN model is agnostic to the identity of the ligand, which is kept constant within the mutant libraries. Using a miniscule subset of the total combinatorial space (204-208 mutants) as training data, the proposed GCN model achieves a high accuracy in predicting the binding energy of unseen variants. The network's accuracy was further improved by injecting feature embeddings obtained from a language module pretrained on 10 million protein sequences. Since no structural information is needed to evaluate new variants, the deep learning algorithm is capable of scoring an enzyme variant in under 1 ms, allowing the search of billions of candidates on a single GPU.

Computational Prediction of ω-Transaminase Specificity by a Combination of Docking and Molecular Dynamics Simulations.

ω-Transaminases (ω-TAs) catalyze the conversion of ketones to chiral amines, often with high enantioselectivity and specificity, which makes them attractive for industrial production of chiral amines. Tailoring ω-TAs to accept non-natural substrates is necessary because of their limited substrate range. We present a computational protocol for predicting the enantioselectivity and catalytic selectivity of an ω-TA from Vibrio fluvialis with different substrates and benchmark it against 62 compounds gathered from the literature. Rosetta-generated complexes containing an external aldimine intermediate of the transamination reaction are used as starting conformations for multiple short independent molecular dynamics (MD) simulations. The combination of molecular docking and MD simulations ensures sufficient and accurate sampling of the relevant conformational space. Based on the frequency of near-attack conformations observed during the MD trajectories, enantioselectivities can be quantitatively predicted. The predicted enantioselectivities are in agreement with a benchmark dataset of experimentally determined ee% values. The substrate-range predictions can be based on the docking score of the external aldimine intermediate. The low computational cost required to run the presented framework makes it feasible for use in enzyme design to screen thousands of enzyme variants.

Computational Redesign of an ω-Transaminase from Pseudomonas jessenii for Asymmetric Synthesis of Enantiopure Bulky Amines.

ω-Transaminases (ω-TA) are attractive biocatalysts for the production of chiral amines from prochiral ketones via asymmetric synthesis. However, the substrate scope of ω-TAs is usually limited due to steric hindrance at the active site pockets. We explored a protein engineering strategy using computational design to expand the substrate scope of an (S)-selective ω-TA from Pseudomonas jessenii (PjTA-R6) toward the production of bulky amines. PjTA-R6 is attractive for use in applied biocatalysis due to its thermostability, tolerance to organic solvents, and acceptance of high concentrations of isopropylamine as amino donor. PjTA-R6 showed no detectable activity for the synthesis of six bicyclic or bulky amines targeted in this study. Six small libraries composed of 7-18 variants each were separately designed via computational methods and tested in the laboratory for ketone to amine conversion. In each library, the vast majority of the variants displayed the desired activity, and of the 40 different designs, 38 produced the target amine in good yield with >99% enantiomeric excess. This shows that the substrate scope and enantioselectivity of PjTA mutants could be predicted in silico with high accuracy. The single mutant W58G showed the best performance in the synthesis of five structurally similar bulky amines containing the indan and tetralin moieties. The best variant for the other bulky amine, 1-phenylbutylamine, was the triple mutant W58M + F86L + R417L, indicating that Trp58 is a key residue in the large binding pocket for PjTA-R6 redesign. Crystal structures of the two best variants confirmed the computationally predicted structures. The results show that computational design can be an efficient approach to rapidly expand the substrate scope of ω-TAs to produce enantiopure bulky amines.

Thermostable D-amino acid decarboxylases derived from Thermotoga maritima diaminopimelate decarboxylase.

Diaminopimelate decarboxylases (DAPDCs) are highly selective enzymes that catalyze the common final step in different lysine biosynthetic pathways, i.e. the conversion of meso-diaminopimelate (DAP) to L-lysine. We examined the modification of the substrate specificity of the thermostable decarboxylase from Thermotoga maritima with the aim to introduce activity with 2-aminopimelic acid (2-APA) since its decarboxylation leads to 6-aminocaproic acid (6-ACA), a building block for the synthesis of nylon-6. Structure-based mutagenesis of the distal carboxylate binding site resulted in a set of enzyme variants with new activities toward different D-amino acids. One of the mutants (E315T) had lost most of its activity toward DAP and primarily acted as a 2-APA decarboxylase. We next used computational modeling to explain the observed shift in catalytic activities of the mutants. The results suggest that predictive computational protocols can support the redesign of the catalytic properties of this class of decarboxylating PLP-dependent enzymes.

Semiprocessive Hyperglycosylation of Adhesin by Bacterial Protein N-Glycosyltransferases.

Processivity is an important feature of enzyme families such as DNA polymerases, polysaccharide synthases, and protein kinases, to ensure high fidelity in biopolymer synthesis and modification. Here, we reveal processive character in the family of cytoplasmic protein N-glycosyltransferases (NGTs). Through various activity assays, intact protein mass spectrometry, and proteomics analysis, we established that NGTs from nontypeable Haemophilus influenzae and Actinobacillus pleuropneumoniae modify an adhesin protein fragment in a semiprocessive manner. Molecular modeling studies suggest that the processivity arises from the shallow substrate binding groove in NGT, which promotes the sliding of the adhesin over the surface to allow further glycosylations without temporary dissociation. We hypothesize that the processive character of these bacterial protein glycosyltransferases is the mechanism to ensure multisite glycosylation of adhesins in vivo, thereby creating the densely glycosylated proteins necessary for bacterial self-aggregation and adherence to human cells, as a first step toward infection.

Protein-ligand binding with the coarse-grained Martini model.

The detailed understanding of the binding of small molecules to proteins is the key for the development of novel drugs or to increase the acceptance of substrates by enzymes. Nowadays, computer-aided design of protein-ligand binding is an important tool to accomplish this task. Current approaches typically rely on high-throughput docking essays or computationally expensive atomistic molecular dynamics simulations. Here, we present an approach to use the recently re-parametrized coarse-grained Martini model to perform unbiased millisecond sampling of protein-ligand interactions of small drug-like molecules. Remarkably, we achieve high accuracy without the need of any a priori knowledge of binding pockets or pathways. Our approach is applied to a range of systems from the well-characterized T4 lysozyme over members of the GPCR family and nuclear receptors to a variety of enzymes. The presented results open the way to high-throughput screening of ligand libraries or protein mutations using the coarse-grained Martini model.