Molecular basis of human noradrenaline transporter reuptake and inhibition.

Noradrenaline, also known as norepinephrine, has a wide range of activities and effects on most brain cell types1. Its reuptake from the synaptic cleft heavily relies on the noradrenaline transporter (NET) located in the presynaptic membrane2. Here we report the cryo-electron microscopy (cryo-EM) structures of the human NET in both its apo state and when bound to substrates or antidepressant drugs, with resolutions ranging from 2.5 Å to 3.5 Å. The two substrates, noradrenaline and dopamine, display a similar binding mode within the central substrate binding site (S1) and within a newly identified extracellular allosteric site (S2). Four distinct antidepressants, namely, atomoxetine, desipramine, bupropion and escitalopram, occupy the S1 site to obstruct substrate transport in distinct conformations. Moreover, a potassium ion was observed within sodium-binding site 1 in the structure of the NET bound to desipramine under the KCl condition. Complemented by structural-guided biochemical analyses, our studies reveal the mechanism of substrate recognition, the alternating access of NET, and elucidate the mode of action of the four antidepressants.

Protein-DNA binding sites prediction based on pre-trained protein language model and contrastive learning.

Protein-DNA interaction is critical for life activities such as replication, transcription and splicing. Identifying protein-DNA binding residues is essential for modeling their interaction and downstream studies. However, developing accurate and efficient computational methods for this task remains challenging. Improvements in this area have the potential to drive novel applications in biotechnology and drug design. In this study, we propose a novel approach called Contrastive Learning And Pre-trained Encoder (CLAPE), which combines a pre-trained protein language model and the contrastive learning method to predict DNA binding residues. We trained the CLAPE-DB model on the protein-DNA binding sites dataset and evaluated the model performance and generalization ability through various experiments. The results showed that the area under ROC curve values of the CLAPE-DB model on the two benchmark datasets reached 0.871 and 0.881, respectively, indicating superior performance compared to other existing models. CLAPE-DB showed better generalization ability and was specific to DNA-binding sites. In addition, we trained CLAPE on different protein-ligand binding sites datasets, demonstrating that CLAPE is a general framework for binding sites prediction. To facilitate the scientific community, the benchmark datasets and codes are freely available at https://github.com/YAndrewL/clape.

Enhancing Protein Solubility via Glycosylation: From Chemical Synthesis to Machine Learning Predictions.

Glycosylation is a valuable tool for modulating protein solubility; however, the lack of reliable research strategies has impeded efficient progress in understanding and applying this modification. This study aimed to bridge this gap by investigating the solubility of a model glycoprotein molecule, the carbohydrate-binding module (CBM), through a two-stage process. In the first stage, an approach involving chemical synthesis, comparative analysis, and molecular dynamics simulations of a library of glycoforms was employed to elucidate the effect of different glycosylation patterns on solubility and the key factors responsible for the effect. In the second stage, a predictive mathematical formula, innovatively harnessing machine learning algorithms, was derived to relate solubility to the identified key factors and accurately predict the solubility of the newly designed glycoforms. Demonstrating feasibility and effectiveness, this two-stage approach offers a valuable strategy for advancing glycosylation research, especially for the discovery of glycoforms with increased solubility.

DeepP450: Predicting Human P450 Activities of Small Molecules by Integrating Pretrained Protein Language Model and Molecular Representation.

Cytochrome P450 enzymes (CYPs) play a crucial role in Phase I drug metabolism in the human body, and CYP activity toward compounds can significantly affect druggability, making early prediction of CYP activity and substrate identification essential for therapeutic development. Here, we established a deep learning model for assessing potential CYP substrates, DeepP450, by fine-tuning protein and molecule pretrained models through feature integration with cross-attention and self-attention layers. This model exhibited high prediction accuracy (0.92) on the test set, with area under the receiver operating characteristic curve (AUROC) values ranging from 0.89 to 0.98 in substrate/nonsubstrate predictions across the nine major human CYPs, surpassing current benchmarks for CYP activity prediction. Notably, DeepP450 uses only one model to predict substrates/nonsubstrates for any of the nine CYPs and exhibits certain generalizability on novel compounds and different categories of human CYPs, which could greatly facilitate early stage drug design by avoiding CYP-reactive compounds.

Susceptibilities of Human ACE2 Genetic Variants in Coronavirus Infection.

The coronavirus disease 2019 (COVID-19) pandemic, caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), has resulted in more than 235 million cases worldwide and 4.8 million deaths (October 2021), with various incidences and mortalities among regions/ethnicities. The coronaviruses SARS-CoV, SARS-CoV-2, and HCoV-NL63 utilize the angiotensin-converting enzyme 2 (ACE2) as the receptor to enter cells. We hypothesized that the genetic variability in ACE2 may contribute to the variable clinical outcomes of COVID-19. To test this hypothesis, we first conducted an in silico investigation of single-nucleotide polymorphisms (SNPs) in the coding region of ACE2. We then applied an integrated approach of genetics, biochemistry, and virology to explore the capacity of select ACE2 variants to bind coronavirus spike proteins and mediate viral entry. We identified the ACE2 D355N variant that restricts the spike protein-ACE2 interaction and consequently limits infection both in vitro and in vivo. In conclusion, ACE2 polymorphisms could modulate susceptibility to SARS-CoV-2, which may lead to variable disease severity. IMPORTANCE There is considerable variation in disease severity among patients infected with SARS-CoV-2, the virus that causes COVID-19. Human genetic variation can affect disease outcome, and the coronaviruses SARS-CoV, SARS-CoV-2, and HCoV-NL63 utilize human ACE2 as the receptor to enter cells. We found that several missense ACE2 single-nucleotide variants (SNVs) that showed significantly altered binding with the spike proteins of SARS-CoV, SARS-CoV-2, and NL63-HCoV. We identified an ACE2 SNP, D355N, that restricts the spike protein-ACE2 interaction and consequently has the potential to protect individuals against SARS-CoV-2 infection. Our study highlights that ACE2 polymorphisms could impact human susceptibility to SARS-CoV-2, which may contribute to ethnic and geographical differences in SARS-CoV-2 spread and pathogenicity.

Leveraging structure for enzyme function prediction: methods, opportunities, and challenges.

The rapid growth of the number of protein sequences that can be inferred from sequenced genomes presents challenges for function assignment, because only a small fraction (currently <1%) has been experimentally characterized. Bioinformatics tools are commonly used to predict functions of uncharacterized proteins. Recently, there has been significant progress in using protein structures as an additional source of information to infer aspects of enzyme function, which is the focus of this review. Successful application of these approaches has led to the identification of novel metabolites, enzyme activities, and biochemical pathways. We discuss opportunities to elucidate systematically protein domains of unknown function, orphan enzyme activities, dead-end metabolites, and pathways in secondary metabolism.

Computational-guided discovery and characterization of a sesquiterpene synthase from Streptomyces clavuligerus.

Terpenoids are a large structurally diverse group of natural products with an array of functions in their hosts. The large amount of genomic information from recent sequencing efforts provides opportunities and challenges for the functional assignment of terpene synthases that construct the carbon skeletons of these compounds. Inferring function from the sequence and/or structure of these enzymes is not trivial because of the large number of possible reaction channels and products. We tackle this problem by developing an algorithm to enumerate possible carbocations derived from the farnesyl cation, the first reactive intermediate of the substrate, and evaluating their steric and electrostatic compatibility with the active site. The homology model of a putative pentalenene synthase (Uniprot: B5GLM7) from Streptomyces clavuligerus was used in an automated computational workflow for product prediction. Surprisingly, the workflow predicted a linear triquinane scaffold as the top product skeleton for B5GLM7. Biochemical characterization of B5GLM7 reveals the major product as (5S,7S,10R,11S)-cucumene, a sesquiterpene with a linear triquinane scaffold. To our knowledge, this is the first documentation of a terpene synthase involved in the synthesis of a linear triquinane. The success of our prediction for B5GLM7 suggests that this approach can be used to facilitate the functional assignment of novel terpene synthases.

Defining the Product Chemical Space of Monoterpenoid Synthases.

Terpenoid synthases create diverse carbon skeletons by catalyzing complex carbocation rearrangements, making them particularly challenging for enzyme function prediction. To begin to address this challenge, we have developed a computational approach for the systematic enumeration of terpenoid carbocations. Application of this approach allows us to systematically define a nearly complete chemical space for the potential carbon skeletons of products from monoterpenoid synthases. Specifically, 18758 carbocations were generated, which we cluster into 74 cyclic skeletons. Five of the 74 skeletons are found in known natural products; some of the others are plausible for new functions, either in nature or engineered. This work systematizes the description of function for this class of enzymes, and provides a basis for predicting functions of uncharacterized enzymes. To our knowledge, this is the first computational study to explore the complete product chemical space of this important class of enzymes.

Chiral N,O-Ligand/[Cu(OAc)2 ]-Catalyzed Asymmetric Construction of 4-Aminopyrrolidine Derivatives by 1,3-Dipolar Cycloaddition of Azomethine Ylides with α-Phthalimidoacrylates.

A protocol to access useful 4-aminopyrrolidine-2,4-dicarboxylate derivatives has been developed. A variety of chiral N,O-ligands derived from 2,3-dihydroimidazo[1,2-a]pyridine motifs have been evaluated in the asymmetric 1,3-dipolar cycloaddition of azomethine ylides to α-phthalimidoacrylates. Reactions catalyzed by copper in combination with ligand 7-Cl-DHIPOH provided the highest level of stereoselectivity for the 1,3-dipolar cycloaddition reaction. The reaction tolerates both ß-substituted and ß-unsubstituted α-phthalimidoacrylate as dipolarophiles, affording the corresponding quaternary 4-aminopyrrolidine cycloadducts with excellent diastereo- (>98:2 d.r.) and enantioselectivities (up to 97 % ee). Removal of the phthalimido protecting group can be accomplished by a simple NaBH4 reduction. Theoretical calculations employing DFT methods show this cycloaddition reaction is likely to proceed through a stepwise mechanism and the stereochemistry was also theoretically rationalized.

Predicting the functions and specificity of triterpenoid synthases: a mechanism-based multi-intermediate docking approach.

Terpenoid synthases construct the carbon skeletons of tens of thousands of natural products. To predict functions and specificity of triterpenoid synthases, a mechanism-based, multi-intermediate docking approach is proposed. In addition to enzyme function prediction, other potential applications of the current approach, such as enzyme mechanistic studies and enzyme redesign by mutagenesis, are discussed.

Synthesis and enzymatic studies of bisubstrate analogues for farnesyl diphosphate synthase.

Farnesyl diphosphate synthase catalyzes the sequential chain elongation reactions between isopentenyl diphosphate (IPP) and dimethylallyl diphosphate (DMAPP) to form geranyl diphosphate (GPP) and between IPP and GPP to give farnesyl diphosphate (FPP). Bisubstrate analogues containing the allylic and homoallylic substrates were synthesized by joining fragments for IPP and the allylic diphosphates with a C-C bond between the methyl group at C3 in IPP and the Z-methyl group at C3 in DMAPP (3-OPP) and GPP (4-OPP), respectively. These constructs placed substantial limits on the conformational space available to the analogues relative to the two substrates. The key features of the synthesis of bisubstrate analogues 3-OPP and 4-OPP are a regioselective C-alkylation of the dianion of 3-methyl-3-buten-1-ol (5), a Z-selective cuprate addition of alkyl groups to an α,ß-alkynyl ester intermediate, and differential activation of allylic and homoallylic alcohols in the analogues, followed by a simultaneous displacement of the leaving groups with tris(tetra-n-butylammonium) hydrogen diphosphate to give the corresponding bisdiphosphate analogues. The bisubstrate analogues were substrates for FPP synthase, giving novel seven-membered ring analogues of GPP and FPP. The catalytic efficiencies for cyclization of 3-OPP and 4-OPP were similar to those for chain elongation with IPP and DMAPP.

Design of reversible, cysteine-targeted Michael acceptors guided by kinetic and computational analysis.

Electrophilic probes that covalently modify a cysteine thiol often show enhanced pharmacological potency and selectivity. Although reversible Michael acceptors have been reported, the structural requirements for reversibility are poorly understood. Here, we report a novel class of acrylonitrile-based Michael acceptors, activated by aryl or heteroaryl electron-withdrawing groups. We demonstrate that thiol adducts of these acrylonitriles undergo ß-elimination at rates that span more than 3 orders of magnitude. These rates correlate inversely with the computed proton affinity of the corresponding carbanions, enabling the intrinsic reversibility of the thiol-Michael reaction to be tuned in a predictable manner. We apply these principles to the design of new reversible covalent kinase inhibitors with improved properties. A cocrystal structure of one such inhibitor reveals specific noncovalent interactions between the 1,2,4-triazole activating group and the kinase. Our experimental and computational study enables the design of new Michael acceptors, expanding the palette of reversible, cysteine-targeted electrophiles.

PointGAT: A Quantum Chemical Property Prediction Model Integrating Graph Attention and 3D Geometry.

Predicting quantum chemical properties is a fundamental challenge for computational chemistry. While the development of graph neural networks has advanced molecular representation learning and property prediction, their performance could be further enhanced by incorporating three-dimensional (3D) structural geometry into two-dimensional (2D) molecular graph representation. In this study, we introduce the PointGAT model for quantum molecular property prediction, which integrates 3D molecular coordinates with graph-attention modeling. Comparison with other current models in molecular prediction tasks showed that PointGAT could provide higher predictive accuracy in various benchmark data sets from MoleculeNet, including ESOL, FreeSolv, Lipop, HIV, and 6 out of 12 tasks of the QM9 data set. To further examine PointGAT prediction of quantum mechanical (QM) energies, we constructed a C10 data set comprising 11,841 charged and chiral carbocation intermediates with QM energies calculated at the DM21/6-31G*//B3LYP/6-31G* levels. Notably, PointGAT achieved an R2 value of 0.950 and an MAE of 1.616 kcal/mol, outperforming even the best-performing graph neural network model with a reduction of 0.216 kcal/mol in MAE and an improvement of 0.050 in R2. Additional ablation studies indicated that incorporating molecular geometry into the model resulted in markedly higher predictive accuracy, reducing the MAE value from 1.802 to 1.616 kcal/mol. Moreover, visualization of PointGAT atomic attention weights suggested its predictions were interpretable. Findings in this study support the application of PointGAT as a powerful and versatile tool for quantum chemical property prediction that can facilitate high-accuracy modeling for fundamental exploration of chemical space as well as drug design and molecular engineering.

Accurate prediction of CDR-H3 loop structures of antibodies with deep learning.

Accurate prediction of the structurally diverse complementarity determining region heavy chain 3 (CDR-H3) loop structure remains a primary and long-standing challenge for antibody modeling. Here, we present the H3-OPT toolkit for predicting the 3D structures of monoclonal antibodies and nanobodies. H3-OPT combines the strengths of AlphaFold2 with a pre-trained protein language model and provides a 2.24 Å average RMSDCα between predicted and experimentally determined CDR-H3 loops, thus outperforming other current computational methods in our non-redundant high-quality dataset. The model was validated by experimentally solving three structures of anti-VEGF nanobodies predicted by H3-OPT. We examined the potential applications of H3-OPT through analyzing antibody surface properties and antibody-antigen interactions. This structural prediction tool can be used to optimize antibody-antigen binding and engineer therapeutic antibodies with biophysical properties for specialized drug administration route.

EvoAI enables extreme compression and reconstruction of the protein sequence space.

Designing proteins with improved functions requires a deep understanding of how sequence and function are related, a vast space that is hard to explore. The ability to efficiently compress this space by identifying functionally important features is extremely valuable. Here, we first establish a method called EvoScan to comprehensively segment and scan the high-fitness sequence space to obtain anchor points that capture its essential features, especially in high dimensions. Our approach is compatible with any biomolecular function that can be coupled to a transcriptional output. We then develop deep learning and large language models to accurately reconstruct the space from these anchors, allowing computational prediction of novel, highly fit sequences without prior homology-derived or structural information. We apply this hybrid experimental-computational method, which we call EvoAI, to a repressor protein and find that only 82 anchors are sufficient to compress the high-fitness sequence space with a compression ratio of 1048. The extreme compressibility of the space informs both applied biomolecular design and understanding of natural evolution.

Predicting enzyme-substrate specificity with QM/MM methods: a case study of the stereospecificity of (D)-glucarate dehydratase.

The stereospecificity of d-glucarate dehydratase (GlucD) is explored by QM/MM calculations. Both the substrate binding and the chemical steps of GlucD contribute to substrate specificity. Although the identification of transition states remains computationally intensive, we suggest that QM/MM computations on ground states or intermediates can capture aspects of specificity that cannot be obtained using docking or molecular mechanics methods.

Catalysts or initiators? Beckmann rearrangement revisited.

The catalytic mechanism of the organo-mediated Beckmann rearrangement has been modeled using DFT calculations. Five representative promoters were shown to be initiators rather than catalysts. A self-propagating mechanism is shown to be energetically much more favored than the previously proposed mechanisms involving a Meisenheimer complex.

Mechanistic insight into self-propagation of organo-mediated Beckmann rearrangement: a combined experimental and computational study.

Organo-mediated Beckmann rearrangement in the liquid phase, which has the advantage of high efficiency and straightforward experimental procedures, plays an important role in the synthesis of amides from oximes. However, the catalytic mechanisms of these organic-based promoters are still not well understood. In this work, we report a combined experimental and computational study on the mechanism of Beckmann rearrangement mediated by organic-based promoters, using TsCl as an example. A novel self-propagating cycle is proposed, and key intermediates of this self-propagating cycle are confirmed by both experiments and DFT calculations. In addition, the reason why cyclohexanone oxime is not a good substrate of the organo-mediated Beckmann rearrangement is discussed, and a strategy for improving the yield is proposed.

SARS-CoV-2 main protease cleaves MAGED2 to antagonize host antiviral defense.

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is the agent causing the global pandemic of COVID-19. SARS-CoV-2 genome encodes a main protease (nsp5, also called Mpro) and a papain-like protease (nsp3, also called PLpro), which are responsible for processing viral polyproteins to assemble a functional replicase complex. In this study, we found that Mpro of SARS-CoV-2 can cleave human MAGED2 and other mammalian orthologs at Gln-263. Moreover, SARS-CoV and MERS-CoV Mpro can also cleave human MAGED2, suggesting MAGED2 cleavage by Mpro is an evolutionarily conserved mechanism of coronavirus infection in mammals. Intriguingly, Mpro from Beta variant cleaves MAGED2 more efficiently than wild type, but Omicron Mpro is opposite. Further studies show that MAGED2 inhibits SARS-CoV-2 infection at viral replication step. Mechanistically, MAGED2 is associated with SARS-CoV-2 nucleocapsid protein through its N-terminal region in an RNA-dependent manner, and this disrupts the interaction between SARS-CoV-2 nucleocapsid protein and viral genome, thus inhibiting viral replication. When MAGED2 is cleaved by Mpro, the N-terminal of MAGED2 will translocate into the nucleus, and the truncated MAGED2 is unable to suppress SARS-CoV-2 replication. This work not only discovers the antiviral function of MAGED2 but also provides new insights into how SARS-CoV-2 Mpro antagonizes host antiviral response. IMPORTANCE Host factors that restrict severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection remain elusive. Here, we found that MAGED2 can be cleaved by SARS-CoV-2 main protease (Mpro) at Gln-263. SARS-CoV and MERS-CoV Mpro can also cleave MAGED2, and MAGED2 from multiple species can be cleaved by SARS-CoV-2 Mpro. Mpro from Beta variant cleaves MAGED2 more efficiently efficiently than wild type, but Omicron is the opposite. MAGED2 depletion enhances SARS-CoV-2 infection, suggesting its inhibitory role in SARS-CoV-2 infection. Mechanistically, MAGED2 restricts SARS-CoV-2 replication by disrupting the interaction between nucleocapsid and viral genomes. When MAGED2 is cleaved, its N-terminal will translocate into the nucleus. In this way, Mpro relieves MAGED2' inhibition on viral replication. This study improves our understanding of complex viral-host interaction and provides novel targets to treat SARS-CoV-2 infection.

Structural changes of Listeria monocytogenes Sortase A: a key to understanding the catalytic mechanism.

Listeria monocytogenes is one of the most virulent foodborne pathogens. L. monocytogenes Sortase A (SrtA) enzyme, which catalyzes the cell wall anchoring reaction of the leucine, proline, X, threonine, and glycine proteins (LPXTG, where X is any amino acid), is a target for the development of antilisteriosis drugs. In this study, the structure of the L. monocytogenes SrtA enzyme-substrate complex was obtained using homology modeling, molecular docking and molecular dynamics simulations. Explicit enzyme-substrate interactions in the inactive and active forms of the enzyme were compared, based on 30 ns simulations on each system. The active site arginine (Arg 197) was found to be able change its hydrogen donor interactions from the LP backbone carbonyl groups of the LPXTG substrate in the inactive form, to the TG backbone carbonyls in the active form, which could be of importance for holding the substrate in position for the catalytic process.