Epitopes recognition of SARS-CoV-2 nucleocapsid RNA binding domain by human monoclonal antibodies.

Coronavirus nucleocapsid protein (NP) of SARS-CoV-2 plays a central role in many functions important for virus proliferation including packaging and protecting genomic RNA. The protein shares sequence, structure, and architecture with nucleocapsid proteins from betacoronaviruses. The N-terminal domain (NPRBD) binds RNA and the C-terminal domain is responsible for dimerization. After infection, NP is highly expressed and triggers robust host immune response. The anti-NP antibodies are not protective and not neutralizing but can effectively detect viral proliferation soon after infection. Two structures of SARS-CoV-2 NPRBD were determined providing a continuous model from residue 48 to 173, including RNA binding region and key epitopes. Five structures of NPRBD complexes with human mAbs were isolated using an antigen-bait sorting. Complexes revealed a distinct complement-determining regions and unique sets of epitope recognition. This may assist in the early detection of pathogens and designing peptide-based vaccines. Mutations that significantly increase viral load were mapped on developed, full length NP model, likely impacting interactions with host proteins and viral RNA.

Structure-based design of SARS-CoV-2 papain-like protease inhibitors.

The COVID-19 pandemic is caused by SARS-CoV-2, an RNA virus with high transmissibility and mutation rate. Given the paucity of orally bioavailable antiviral drugs to combat SARS-CoV-2 infection, there is a critical need for additional antivirals with alternative mechanisms of action. Papain-like protease (PLpro) is one of the two SARS-CoV-2 encoded viral cysteine proteases essential for viral replication. PLpro cleaves at three sites of the viral polyproteins. In addition, PLpro antagonizes the host immune response upon viral infection by cleaving ISG15 and ubiquitin from host proteins. Therefore, PLpro is a validated antiviral drug target. In this study, we report the X-ray crystal structures of papain-like protease (PLpro) with two potent inhibitors, Jun9722 and Jun9843. Subsequently, we designed and synthesized several series of analogs to explore the structure-activity relationship, which led to the discovery of PLpro inhibitors with potent enzymatic inhibitory activity and antiviral activity against SARS-CoV-2. Together, the lead compounds are promising drug candidates for further development.

Dual domain recognition determines SARS-CoV-2 PLpro selectivity for human ISG15 and K48-linked di-ubiquitin.

The Papain-like protease (PLpro) is a domain of a multi-functional, non-structural protein 3 of coronaviruses. PLpro cleaves viral polyproteins and posttranslational conjugates with poly-ubiquitin and protective ISG15, composed of two ubiquitin-like (UBL) domains. Across coronaviruses, PLpro showed divergent selectivity for recognition and cleavage of posttranslational conjugates despite sequence conservation. We show that SARS-CoV-2 PLpro binds human ISG15 and K48-linked di-ubiquitin (K48-Ub2) with nanomolar affinity and detect alternate weaker-binding modes. Crystal structures of untethered PLpro complexes with ISG15 and K48-Ub2 combined with solution NMR and cross-linking mass spectrometry revealed how the two domains of ISG15 or K48-Ub2 are differently utilized in interactions with PLpro. Analysis of protein interface energetics predicted differential binding stabilities of the two UBL/Ub domains that were validated experimentally. We emphasize how substrate recognition can be tuned to cleave specifically ISG15 or K48-Ub2 modifications while retaining capacity to cleave mono-Ub conjugates. These results highlight alternative druggable surfaces that would inhibit PLpro function.

A Structural Systems Biology Approach to High-Risk CG23 Klebsiella pneumoniae.

Klebsiella pneumoniae is a leading cause of antibiotic-resistant-associated deaths in the world. Here, we report the deposition of 14 structures of enzymes from both the core and accessory genomes of sequence type 23 (ST23) K1 hypervirulent K. pneumoniae.

Dual domain recognition determines SARS-CoV-2 PLpro selectivity for human ISG15 and K48-linked di-ubiquitin.

The Papain-like protease (PLpro) is a domain of a multi-functional, non-structural protein 3 of coronaviruses. PLpro cleaves viral polyproteins and posttranslational conjugates with poly-ubiquitin and protective ISG15, composed of two ubiquitin-like (UBL) domains. Across coronaviruses, PLpro showed divergent selectivity for recognition and cleavage of posttranslational conjugates despite sequence conservation. We show that SARS-CoV-2 PLpro binds human ISG15 and K48-linked di-ubiquitin (K48-Ub 2 ) with nanomolar affinity and detect alternate weaker-binding modes. Crystal structures of untethered PLpro complexes with ISG15 and K48-Ub 2 combined with solution NMR and cross-linking mass spectrometry revealed how the two domains of ISG15 or K48-Ub 2 are differently utilized in interactions with PLpro. Analysis of protein interface energetics predicted differential binding stabilities of the two UBL/Ub domains that were validated experimentally. We emphasize how substrate recognition can be tuned to cleave specifically ISG15 or K48-Ub 2 modifications while retaining capacity to cleave mono-Ub conjugates. These results highlight alternative druggable surfaces that would inhibit PLpro function.

Data collection from crystals grown in microfluidic droplets.

Protein crystals grown in microfluidic droplets have been shown to be an effective and robust platform for storage, transport and serial crystallography data collection with a minimal impact on diffraction quality. Single macromolecular microcrystals grown in nanolitre-sized droplets allow the very efficient use of protein samples and can produce large quantities of high-quality samples for data collection. However, there are challenges not only in growing crystals in microfluidic droplets, but also in delivering the droplets into X-ray beams, including the physical arrangement, beamline and timing constraints and ease of use. Here, the crystallization of two human gut microbial hydrolases in microfluidic droplets is described: a sample-transport and data-collection approach that is inexpensive, is convenient, requires small amounts of protein and is forgiving. It is shown that crystals can be grown in 50-500 pl droplets when the crystallization conditions are compatible with the droplet environment. Local and remote data-collection methods are described and it is shown that crystals grown in microfluidics droplets and housed as an emulsion in an Eppendorf tube can be shipped from the US to the UK using a FedEx envelope, and data can be collected successfully. Details of how crystals were delivered to the X-ray beam by depositing an emulsion of droplets onto a silicon fixed-target serial device are provided. After three months of storage at 4°C, the crystals endured and diffracted well, showing only a slight decrease in diffracting power, demonstrating a suitable way to grow crystals, and to store and collect the droplets with crystals for data collection. This sample-delivery and data-collection strategy allows crystal droplets to be shipped and set aside until beamtime is available.

Functional and Structural Characterization of Diverse NfsB Chloramphenicol Reductase Enzymes from Human Pathogens.

Phylogenetically diverse bacteria can carry out chloramphenicol reduction, but only a single enzyme has been described that efficiently catalyzes this reaction, the NfsB nitroreductase from Haemophilus influenzae strain KW20. Here, we tested the hypothesis that some NfsB homologs function as housekeeping enzymes with the potential to become chloramphenicol resistance enzymes. We found that expression of H. influenzae and Neisseria spp. nfsB genes, but not Pasteurella multocida nfsB, allows Escherichia coli to resist chloramphenicol by nitroreduction. Mass spectrometric analysis confirmed that purified H. influenzae and N. meningitides NfsB enzymes reduce chloramphenicol to amino-chloramphenicol, while kinetics analyses supported the hypothesis that chloramphenicol reduction is a secondary activity. We combined these findings with atomic resolution structures of multiple chloramphenicol-reducing NfsB enzymes to identify potential key substrate-binding pocket residues. Our work expands the chloramphenicol reductase family and provides mechanistic insights into how a housekeeping enzyme might confer antibiotic resistance. IMPORTANCE The question of how new enzyme activities evolve is of great biological interest and, in the context of antibiotic resistance, of great medical importance. Here, we have tested the hypothesis that new antibiotic resistance mechanisms may evolve from promiscuous housekeeping enzymes that have antibiotic modification side activities. Previous work identified a Haemophilus influenzae nitroreductase housekeeping enzyme that has the ability to give Escherichia coli resistance to the antibiotic chloramphenicol by nitroreduction. Herein, we extend this work to enzymes from other Haemophilus and Neisseria strains to discover that expression of chloramphenicol reductases is sufficient to confer chloramphenicol resistance to Es. coli, confirming that chloramphenicol reductase activity is widespread across this nitroreductase family. By solving the high-resolution crystal structures of active chloramphenicol reductases, we identified residues important for this activity. Our work supports the hypothesis that housekeeping proteins possessing multiple activities can evolve into antibiotic resistance enzymes.

Improving deconvolution methods in biology through open innovation competitions: an application to the connectivity map.

MOTIVATION: Do machine learning methods improve standard deconvolution techniques for gene expression data? This article uses a unique new dataset combined with an open innovation competition to evaluate a wide range of approaches developed by 294 competitors from 20 countries. The competition's objective was to address a deconvolution problem critical to analyzing genetic perturbations from the Connectivity Map. The issue consists of separating gene expression of individual genes from raw measurements obtained from gene pairs. We evaluated the outcomes using ground-truth data (direct measurements for single genes) obtained from the same samples. RESULTS: We find that the top-ranked algorithm, based on random forest regression, beat the other methods in accuracy and reproducibility; more traditional gaussian-mixture methods performed well and tended to be faster, and the best deep learning approach yielded outcomes slightly inferior to the above methods. We anticipate researchers in the field will find the dataset and algorithms developed in this study to be a powerful research tool for benchmarking their deconvolution methods and a resource useful for multiple applications. AVAILABILITY AND IMPLEMENTATION: The data is freely available at clue.io/data (section Contests) and the software is on GitHub at https://github.com/cmap/gene_deconvolution_challenge. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Sensor Domain of Histidine Kinase VxrA of Vibrio cholerae- A Hairpin-swapped Dimer and its Conformational Change.

VxrA and VxrB are cognate histidine kinase (HK) - response regulator (RR) pairs of a two-component signaling system (TCS) found in Vibrio cholerae, a bacterial pathogen that causes cholera. The VxrAB TCS positively regulates virulence, the Type VI Secretion System, biofilm formation, and cell wall homeostasis in V. cholerae, providing protection from environmental stresses and contributing to the transmission and virulence of the pathogen. The VxrA HK has a unique periplasmic sensor domain (SD) and, remarkably, lacks a cytoplasmic linker domain between the second transmembrane helix and the dimerization and histidine phosphotransfer (DHp) domain, indicating that this system may utilize a potentially unique signal sensing and transmission TCS mechanism. In this study, we have determined several crystal structures of VxrA-SD and its mutants. These structures reveal a novel structural fold forming an unusual ß hairpin-swapped dimer. A conformational change caused by relative rotation of the two monomers in a VxrA-SD dimer could potentially change the association of transmembrane helices and, subsequently, the pairing of cytoplasmic DHp domains. Based on the structural observation, we propose a putative scissor-like closing regulation mechanism for the VxrA HK.IMPORTANCE V. cholerae has a dynamic life cycle, which requires rapid adaptation to changing external conditions. Two-component signal transduction (TCS) systems allow V. cholerae to sense and respond to these environmental changes. The VxrAB TCS positively regulates a number of important V. cholerae phenotypes, including virulence, the Type Six Secretion System, biofilm formation, and cell wall homeostasis. Here, we provide the crystal structure of the VxrA sensor histidine kinase sensing domain and propose a mechanism for signal transduction. The cognate signal for VxrAB remains unknown, however, in this work we couple our structural analysis with functional assessments of key residues to further our understanding of this important TCS.

Structure of papain-like protease from SARS-CoV-2 and its complexes with non-covalent inhibitors.

The pandemic caused by Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) continues to expand. Papain-like protease (PLpro) is one of two SARS-CoV-2 proteases potentially targetable with antivirals. PLpro is an attractive target because it plays an essential role in cleavage and maturation of viral polyproteins, assembly of the replicase-transcriptase complex, and disruption of host responses. We report a substantive body of structural, biochemical, and virus replication studies that identify several inhibitors of the SARS-CoV-2 enzyme. We determined the high resolution structure of wild-type PLpro, the active site C111S mutant, and their complexes with inhibitors. This collection of structures details inhibitors recognition and interactions providing fundamental molecular and mechanistic insight into PLpro. All compounds inhibit the peptidase activity of PLpro in vitro, some block SARS-CoV-2 replication in cell culture assays. These findings will accelerate structure-based drug design efforts targeting PLpro to identify high-affinity inhibitors of clinical value.

Crystal structures of SARS-CoV-2 ADP-ribose phosphatase: from the apo form to ligand complexes.

Among 15 nonstructural proteins (Nsps), the newly emerging Severe Acute Respiratory Syndrome coronavirus 2 (SARS-CoV-2) encodes a large, multidomain Nsp3. One of its units is the ADP-ribose phosphatase domain (ADRP; also known as the macrodomain, MacroD), which is believed to interfere with the host immune response. Such a function appears to be linked to the ability of the protein to remove ADP-ribose from ADP-ribosylated proteins and RNA, yet the precise role and molecular targets of the enzyme remain unknown. Here, five high-resolution (1.07-2.01 Å) crystal structures corresponding to the apo form of the protein and its complexes with 2-(N-morpholino)ethanesulfonic acid (MES), AMP and ADP-ribose have been determined. The protein is shown to undergo conformational changes to adapt to the ligand in the manner previously observed in close homologues from other viruses. A conserved water molecule is also identified that may participate in hydrolysis. This work builds foundations for future structure-based research on ADRP, including the search for potential antiviral therapeutics.

Development of a Deep Learning Algorithm for Periapical Disease Detection in Dental Radiographs.

Periapical radiolucencies, which can be detected on panoramic radiographs, are one of the most common radiographic findings in dentistry and have a differential diagnosis including infections, granuloma, cysts and tumors. In this study, we seek to investigate the ability with which 24 oral and maxillofacial (OMF) surgeons assess the presence of periapical lucencies on panoramic radiographs, and we compare these findings to the performance of a predictive deep learning algorithm that we have developed using a curated data set of 2902 de-identified panoramic radiographs. The mean diagnostic positive predictive value (PPV) of OMF surgeons based on their assessment of panoramic radiographic images was 0.69(± 0.13), indicating that dentists on average falsely diagnose 31% of cases as radiolucencies. However, the mean diagnostic true positive rate (TPR) was 0.51(± 0.14), indicating that on average 49% of all radiolucencies were missed. We demonstrate that the deep learning algorithm achieves a better performance than 14 of 24 OMF surgeons within the cohort, exhibiting an average precision of 0.60(± 0.04), and an F1 score of 0.58(± 0.04) corresponding to a PPV of 0.67(± 0.05) and TPR of 0.51(± 0.05). The algorithm, trained on limited data and evaluated on clinically validated ground truth, has potential to assist OMF surgeons in detecting periapical lucencies on panoramic radiographs.

The crystal structure of nsp10-nsp16 heterodimer from SARS-CoV-2 in complex with S-adenosylmethionine.

SARS-CoV-2 is a member of the coronaviridae family and is the etiological agent of the respiratory Coronavirus Disease 2019. The virus has spread rapidly around the world resulting in over two million cases and nearly 150,000 deaths as of April 17, 2020. Since no treatments or vaccines are available to treat COVID-19 and SARS-CoV-2, respiratory complications derived from the infections have overwhelmed healthcare systems around the world. This virus is related to SARS-CoV-1, the virus that caused the 2002-2004 outbreak of Severe Acute Respiratory Syndrome. In January 2020, the Center for Structural Genomics of Infectious Diseases implemented a structural genomics pipeline to solve the structures of proteins essential for coronavirus replication-transcription. Here we show the first structure of the SARS-CoV-2 nsp10-nsp16 2'-O-methyltransferase complex with S-adenosylmethionine at a resolution of 1.80 Å. This heterodimer complex is essential for capping viral mRNA transcripts for efficient translation and to evade immune surveillance.

Crystal structure of Nsp15 endoribonuclease NendoU from SARS-CoV-2.

Severe Acute Respiratory Syndrome coronavirus 2 (SARS-CoV-2) is rapidly spreading around the world. There is no existing vaccine or proven drug to prevent infections and stop virus proliferation. Although this virus is similar to human and animal SARS-CoVs and Middle East Respiratory Syndrome coronavirus (MERS-CoVs), the detailed information about SARS-CoV-2 proteins structures and functions is urgently needed to rapidly develop effective vaccines, antibodies, and antivirals. We applied high-throughput protein production and structure determination pipeline at the Center for Structural Genomics of Infectious Diseases to produce SARS-CoV-2 proteins and structures. Here we report two high-resolution crystal structures of endoribonuclease Nsp15/NendoU. We compare these structures with previously reported homologs from SARS and MERS coronaviruses.

3D domain swapping in the TIM barrel of the α subunit of Streptococcus pneumoniae tryptophan synthase.

Tryptophan synthase catalyzes the last two steps of tryptophan biosynthesis in plants, fungi and bacteria. It consists of two protein chains, designated α and ß, encoded by trpA and trpB genes, that function as an αßßα complex. Structural and functional features of tryptophan synthase have been extensively studied, explaining the roles of individual residues in the two active sites in catalysis and allosteric regulation. TrpA serves as a model for protein-folding studies. In 1969, Jackson and Yanofsky observed that the typically monomeric TrpA forms a small population of dimers. Dimerization was postulated to take place through an exchange of structural elements of the monomeric chains, a phenomenon later termed 3D domain swapping. The structural details of the TrpA dimer have remained unknown. Here, the crystal structure of the Streptococcus pneumoniae TrpA homodimer is reported, demonstrating 3D domain swapping in a TIM-barrel fold for the first time. The N-terminal domain comprising the H0-S1-H1-S2 elements is exchanged, while the hinge region corresponds to loop L2 linking strand S2 to helix H2'. The structural elements S2 and L2 carry the catalytic residues Glu52 and Asp63. As the S2 element is part of the swapped domain, the architecture of the catalytic apparatus in the dimer is recreated from two protein chains. The homodimer interface overlaps with the α-ß interface of the tryptophan synthase αßßα heterotetramer, suggesting that the 3D domain-swapped dimer cannot form a complex with the ß subunit. In the crystal, the dimers assemble into a decamer comprising two pentameric rings.

Structural and biochemical analysis of the metallo-ß-lactamase L1 from emerging pathogen Stenotrophomonas maltophilia revealed the subtle but distinct di-metal scaffold for catalytic activity.

Emergence of Enterobacteriaceae harboring metallo-ß-lactamases (MBL) has raised global threats due to their broad antibiotic resistance profiles and the lack of effective inhibitors against them. We have been studied one of the emerging environmental MBL, the L1 from Stenotrophomonas maltophilia K279a. We determined several crystal structures of L1 complexes with three different classes of ß-lactam antibiotics (penicillin G, moxalactam, meropenem, and imipenem), with the inhibitor captopril and different metal ions (Zn+2 , Cd+2 , and Cu+2 ). All hydrolyzed antibiotics and the inhibitor were found binding to two Zn+2 ions mainly through the opened lactam ring and some hydrophobic interactions with the binding pocket atoms. Without a metal ion, the active site is very similarly maintained as that of the native form with two Zn+2 ions, however, the protein does not bind the substrate moxalactam. When two Zn+2 ions were replaced with other metal ions, the same di-metal scaffold was maintained and the added moxalactam was found hydrolyzed in the active site. Differential scanning fluorimetry and isothermal titration calorimetry were used to study thermodynamic properties of L1 MBL compared with New Deli Metallo-ß-lactamase-1 (NDM-1). Both enzymes are significantly stabilized by Zn+2 and other divalent metals but showed different dependency. These studies also suggest that moxalactam and its hydrolyzed form may bind and dissociate with different kinetic modes with or without Zn+2 for each of L1 and NDM-1. Our analysis implicates metal ions, in forming a distinct di-metal scaffold, which is central to the enzyme stability, promiscuous substrate binding and versatile catalytic activity. STATEMENT: The L1 metallo-ß-lactamase from an environmental multidrug-resistant opportunistic pathogen Stenotrophomonas maltophilia K279a has been studied by determining 3D structures of L1 enzyme in the complexes with several ß-lactam antibiotics and different divalent metals and characterizing its biochemical and ligand binding properties. We found that the two-metal center in the active site is critical in the enzymatic process including antibiotics recognition and binding, which explains the enzyme's activity toward diverse antibiotic substrates. This study provides the critical information for understanding the ligand recognition and for advanced drug development.

Advancing computational biology and bioinformatics research through open innovation competitions.

Open data science and algorithm development competitions offer a unique avenue for rapid discovery of better computational strategies. We highlight three examples in computational biology and bioinformatics research in which the use of competitions has yielded significant performance gains over established algorithms. These include algorithms for antibody clustering, imputing gene expression data, and querying the Connectivity Map (CMap). Performance gains are evaluated quantitatively using realistic, albeit sanitized, data sets. The solutions produced through these competitions are then examined with respect to their utility and the prospects for implementation in the field. We present the decision process and competition design considerations that lead to these successful outcomes as a model for researchers who want to use competitions and non-domain crowds as collaborators to further their research.

Conservation of the structure and function of bacterial tryptophan synthases.

Tryptophan biosynthesis is one of the most characterized processes in bacteria, in which the enzymes from Salmonella typhimurium and Escherichia coli serve as model systems. Tryptophan synthase (TrpAB) catalyzes the final two steps of tryptophan biosynthesis in plants, fungi and bacteria. This pyridoxal 5'-phosphate (PLP)-dependent enzyme consists of two protein chains, α (TrpA) and ß (TrpB), functioning as a linear αßßα heterotetrameric complex containing two TrpAB units. The reaction has a complicated, multistep mechanism resulting in the ß-replacement of the hydroxyl group of l-serine with an indole moiety. Recent studies have shown that functional TrpAB is required for the survival of pathogenic bacteria in macrophages and for evading host defense. Therefore, TrpAB is a promising target for drug discovery, as its orthologs include enzymes from the important human pathogens Streptococcus pneumoniae, Legionella pneumophila and Francisella tularensis, the causative agents of pneumonia, legionnaires' disease and tularemia, respectively. However, specific biochemical and structural properties of the TrpABs from these organisms have not been investigated. To fill the important phylogenetic gaps in the understanding of TrpABs and to uncover unique features of TrpAB orthologs to spearhead future drug-discovery efforts, the TrpABs from L. pneumophila, F. tularensis and S. pneumoniae have been characterized. In addition to kinetic properties and inhibitor-sensitivity data, structural information gathered using X-ray crystallo-graphy is presented. The enzymes show remarkable structural conservation, but at the same time display local differences in both their catalytic and allosteric sites that may be responsible for the observed differences in catalysis and inhibitor binding. This functional dissimilarity may be exploited in the design of species-specific enzyme inhibitors.

Use of Crowd Innovation to Develop an Artificial Intelligence-Based Solution for Radiation Therapy Targeting.

IMPORTANCE: Radiation therapy (RT) is a critical cancer treatment, but the existing radiation oncologist work force does not meet growing global demand. One key physician task in RT planning involves tumor segmentation for targeting, which requires substantial training and is subject to significant interobserver variation. OBJECTIVE: To determine whether crowd innovation could be used to rapidly produce artificial intelligence (AI) solutions that replicate the accuracy of an expert radiation oncologist in segmenting lung tumors for RT targeting. DESIGN, SETTING, AND PARTICIPANTS: We conducted a 10-week, prize-based, online, 3-phase challenge (prizes totaled $55 000). A well-curated data set, including computed tomographic (CT) scans and lung tumor segmentations generated by an expert for clinical care, was used for the contest (CT scans from 461 patients; median 157 images per scan; 77 942 images in total; 8144 images with tumor present). Contestants were provided a training set of 229 CT scans with accompanying expert contours to develop their algorithms and given feedback on their performance throughout the contest, including from the expert clinician. MAIN OUTCOMES AND MEASURES: The AI algorithms generated by contestants were automatically scored on an independent data set that was withheld from contestants, and performance ranked using quantitative metrics that evaluated overlap of each algorithm's automated segmentations with the expert's segmentations. Performance was further benchmarked against human expert interobserver and intraobserver variation. RESULTS: A total of 564 contestants from 62 countries registered for this challenge, and 34 (6%) submitted algorithms. The automated segmentations produced by the top 5 AI algorithms, when combined using an ensemble model, had an accuracy (Dice coefficient = 0.79) that was within the benchmark of mean interobserver variation measured between 6 human experts. For phase 1, the top 7 algorithms had average custom segmentation scores (S scores) on the holdout data set ranging from 0.15 to 0.38, and suboptimal performance using relative measures of error. The average S scores for phase 2 increased to 0.53 to 0.57, with a similar improvement in other performance metrics. In phase 3, performance of the top algorithm increased by an additional 9%. Combining the top 5 algorithms from phase 2 and phase 3 using an ensemble model, yielded an additional 9% to 12% improvement in performance with a final S score reaching 0.68. CONCLUSIONS AND RELEVANCE: A combined crowd innovation and AI approach rapidly produced automated algorithms that replicated the skills of a highly trained physician for a critical task in radiation therapy. These AI algorithms could improve cancer care globally by transferring the skills of expert clinicians to under-resourced health care settings.

Natural separation of the acyl-CoA ligase reaction results in a non-adenylating enzyme.

Acyl-coenzyme A (CoA) ligases catalyze the activation of carboxylic acids via a two-step reaction of adenylation followed by thioesterification. Here, we report the discovery of a non-adenylating acyl-CoA ligase PtmA2 and the functional separation of an acyl-CoA ligase reaction. Both PtmA1 and PtmA2, two acyl-CoA ligases from the biosynthetic pathway of platensimycin and platencin, are necessary for the two steps of CoA activation. Gene inactivation of ptmA1 and ptmA2 resulted in the accumulation of free acid and adenylate intermediates, respectively. Enzymatic and structural characterization of PtmA2 confirmed its ability to only catalyze thioesterification. Structural characterization of PtmA2 revealed it binds both free acid and adenylate substrates and undergoes the established mechanism of domain alternation. Finally, site-directed mutagenesis restored both the adenylation and complete CoA activation reactions. This study challenges the currently accepted paradigm of adenylating enzymes and inspires future investigations on functionally separated acyl-CoA ligases and their ramifications in biology.