(C) PLOS One
This story was originally published by PLOS One and is unaltered.
 . . . . . . . . . .



The activation cascade of the broad-spectrum antiviral bemnifosbuvir characterized at atomic resolution [1]

['Aurélie Chazot', 'Aix Marseille Université', 'Cnrs', 'Afmb', 'Umr', 'Marseille', 'Claire Zimberger', 'Mikael Feracci', 'Adel Moussa', 'Atea Pharmaceuticals']

Date: 2024-09 

Abstract Bemnifosbuvir (AT-527) and AT-752 are guanosine analogues currently in clinical trials against several RNA viruses. Here, we show that these drugs require a minimal set of 5 cellular enzymes for activation to their common 5′-triphosphate AT-9010, with an obligate order of reactions. AT-9010 selectively inhibits essential viral enzymes, accounting for antiviral potency. Functional and structural data at atomic resolution decipher N6-purine deamination compatible with its metabolic activation. Crystal structures of human histidine triad nucleotide binding protein 1, adenosine deaminase-like protein 1, guanylate kinase 1, and nucleoside diphosphate kinase at 2.09, 2.44, 1.76, and 1.9 Å resolution, respectively, with cognate precursors of AT-9010 illuminate the activation pathway from the orally available bemnifosbuvir to AT-9010, pointing to key drug–protein contacts along the activation pathway. Our work provides a framework to integrate the design of antiviral nucleotide analogues, confronting requirements and constraints associated with activation enzymes along the 5′-triphosphate assembly line.

Citation: Chazot A, Zimberger C, Feracci M, Moussa A, Good S, Sommadossi J-P, et al. (2024) The activation cascade of the broad-spectrum antiviral bemnifosbuvir characterized at atomic resolution. PLoS Biol 22(8): e3002743. https://doi.org/10.1371/journal.pbio.3002743 Academic Editor: Chaitan Khosla, Stanford University, UNITED STATES OF AMERICA Received: February 19, 2024; Accepted: July 9, 2024; Published: August 27, 2024 Copyright: © 2024 Chazot et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability: The coordinates and structure factors for HINT1, ADALP1, GUK1, and NDPK structures have been deposited in the PDB with accession codes 8PWK, 8QCH, 8PTS, and 8PIE, respectively. Structural data can be found at https://doi.org/10.2210/pdb8pwk/pdb, https://doi.org/10.2210/pdb8qch/pdb, https://doi.org/10.2210/pdb8pts/pdb and https://doi.org/10.2210/pdb8pie/pdb. The data underlying figures, supplementary figures, and tables, including the raw blots for S2 Fig, can be found at https://zenodo.org/records/12606239. Funding: This project has received funding through the 2023 Louis Pasteur Bicentenary Prize, the Grand Prix Scientifique de la Fondation Simone et Cino Del Duca – Institut de France awarded to La Fondation CNRS for BC, and Research-Action-ANR Covid-19 -PHOTONS (2020-2022)- PHOsphorylation Tools for Nucleotides Synthesis awarded to KA. AC received a doctoral fellowship from the Ministère de l'Education Nationale, de l'Enseignement Supérieur et de la Recherche. This work was supported by the ‘Agence Nationale pour la Recherche’ through the French Infrastructure for Integrated Structural Biology (FRISBI) [ANR-10-INSB-05-01].The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing interests: S.G., A.M. and J.P.S. are employees of ATEA Pharmaceuticals, Inc. Abbreviations: ADALP1, adenosine deaminase-like protein 1; CatA, Cathepsin A; CES1, carboxylesterase 1; COVID-19, Coronavirus Disease 2019; GUK1, guanylate kinase 1; HBV, hepatitis B virus; HCV, hepatitis C virus; HINT, histidine triad nucleotide-binding protein; HIV-1, human immunodeficiency virus type 1; NA, nucleoside/nucleotide analogue; NDPK, nucleoside diphosphate kinase; RdRp, RNA-dependent RNA polymerase; RDV, remdesivir; SARS-CoV-2, Severe Acute Respiratory Syndrome Coronavirus 2; SOF, sofosbuvir; TAF, tenofovir alafenamide; TEAB, triethylammonium bicarbonate

Introduction The recent Coronavirus Disease 2019 (COVID-19) crisis due to Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) has evidenced the need for safe and potent antivirals, in conjunction with accurate and rapid diagnostics. First, at the public level, prophylaxis around index cases may significantly curb emergence by cutting viral load and transmissibility. Second, at the individual level, early therapy may limit virus spread to vital organs, avoiding complications. Nucleoside/nucleotide analogues (NAs) represent the first family of compounds that have been used as direct-acting antivirals several decades ago [1–3] against DNA viruses such as herpesvirus, hepatitis B virus (HBV), the retrovirus human immunodeficiency virus type 1 (HIV-1), and RNA viruses such as hepatitis C virus (HCV) and SARS-CoV-2. Upon reaching their target cells, antiviral drugs often act after intracellular activation. Early biologically active nucleoside analogues (NAs) such as ara-C (cytarabine) penetrate the target cell efficiently and are transformed into the nucleoside analogue 5′-triphosphate (NA-TP) by a series of kinase reactions (reviewed in [2–4]). ProTide prodrug technology has been implemented to deliver NA carrying a 5′-phosphate or phosphonate [5]. Aryloxy phosphoramidate prodrugs of, e.g., tenofovir alafenamide (TAF) for HIV-1 and HBV, sofosbuvir (SOF) for HCV, and remdesivir (RDV) for SARS-CoV-2 have met clinical success [5–7]. They achieve cell penetration through shielding the phosphate charges, and preferential hydrolysis at the P-N bond bypasses the first—often limiting—nucleoside kinase reaction, allowing significant building up of NA-TP pools. NA-TPs poison viral RNA synthesis specifically, provoking premature viral RNA chain termination or chemical/genetic corruption of the viral nucleic acid [8]. Two main parameters determine NA potency: the concentration ratio of NA-TP over its natural NTP competitor, and the selectivity of the viral RNA-dependent RNA polymerase (RdRp) for use of the NA-TP [9]. NA-TP pools must be formed efficiently upon intracellular activation, and the differences of the NA-TP scaffold relative to its natural NTP counterpart must remain “below the radar” of the viral RdRp. General knowledge, structural and functional data about cellular enzymes along a given NA activation pathway are fragmented. Numerous NAs have been designed showing appropriate poisoning of viral RNA synthesis through their 5′-TP in vitro, but clinical development has failed for lack of transmembrane permeability, intracellular metabolic activation, and/or cellular toxicity. Clearly, NA drug-design needs an integrated view (i.e., structural, functional, and mechanistic) from the delivered NA up to the ultimate inhibited viral enzyme reaction accounting for antiviral effect, as well as the molecular basis for (non)interaction with cellular enzymes. Such a global picture is emerging for 2 FDA-approved NA prodrugs directed against RNA viruses, SOF and RDV (S1 Fig). The activation pathway of RDV has been elucidated and profiled in a variety of tissues [10]. RDV is an aryloxy phosphoramidate prodrug (GS-5734, Veklury) converted in 2 steps to the 5′-monophosphate NA by esterases belonging to 2 families (Cathepsin A (CatA) and/or carboxylesterase 1 (CES1), followed by the cleavage of the P-N bond by a histidine triad nucleotide (HINT) phosphoramidase [11]. The resulting monophosphate NA is then converted to RDV-TP by the subsequent action of 2 cellular phosphotransferases. However, although detailed studies at the atomic level exist to understand accommodation of the 1′-CN group at the SARS-CoV-2 RdRp active site [12,13], structural insight of the relevant activation intermediates is lacking for all enzymes in the activation pathway. SOF (PSI-7851, Sovaldi) undergoes the same CatA/CES1 deprotection pathway as RDV but is activated to the 5′-triphosphate by the UMP-CMP kinase and the nucleoside diphosphate kinase (NDPK) sequentially [14]. Here again, although the mechanism of chain-termination at the HCV NS5b RdRp has been elucidated at atomic resolution [15], the structural basis of activation remains uncharacterized. Bemnifosbuvir (AT-527 (hemisulfate salt), AT-511 (free base)) is a guanosine analogue currently in clinical trials against SARS-CoV-2 and HCV. It showed in Phase II clinical trials a 71% risk reduction in outpatients with moderate COVID-19 (MORNINGSKY; NCT04396106) [16,17], but the associated Phase III study ended prematurely as it did not meet its primary end point [17]. It is currently under investigation in a global Phase III clinical trial in outpatients at high risk for disease progression (SUNRISE-3; NCT05629962), as well as evaluated as an anti-HCV drug [18] in combination with the NS5A inhibitor ruzasvir (NCT05904470). Its epimer AT-752 is currently in clinical Phase II against Dengue virus (NCT05466240) [19,20]. Both bemnifosbuvir and AT-752, once processed by the CatA/CES1 pathway, converge to the same precursor AT-551 (Fig 1A) [18]. These 2 analogues are among the few antiviral purine NAs devoid of significant cellular toxicity. Once the P-N bond of AT-551 is hydrolyzed (presumably) by HINT1 [21], giving rise to the monophosphate AT-8003, the diamino purine base is (presumably) converted to a natural guanosine base through specific N6-deamination carried out by the adenosine deaminase-like protein 1 (ADALP1) enzyme [22,23]. This reaction is believed to skirt a cellular step responsible for the toxicity of unprotected, “natural” guanosine analogues [18,24]. The resulting 2′-F-2′-C-methyl guanosine 5′-monophosphate (AT-8001) is (presumably) consecutively phosphorylated twice to yield AT-9010 by means of guanylate kinase 1 (GUK1) [25] nucleoside diphosphate kinase (NDPK) [26,27]. AT-9010 accumulates in various cell types [18]. The HCV RNA synthesis is likely halted through RNA chain termination [18]. SARS-CoV-2 RNA synthesis is halted through targeting of the replicase complex at 2 distinct sites [28], and the Dengue virus RNA synthesis is also halted through targeting 2 sites of the NS5 protein [19,20]. PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 1. Validation of the activation pathway of bemnifosbuvir. (A) The activation pathway of AT-511, the free base form of bemnifosbuvir (AT-527), into the active form AT-9010. AT-281, the diastereoisomer of AT-511 and free base form of AT-752, follows the same pathway. (B) Typical RP-HPLC chromatograms of the enzymatic assays after 2 min (gray) and the last kinetic point (black). For each enzyme CatA, CES1, HINT1, ADALP1, GUK1, and NDPK, the successful validation of activity results in conversion of the substrate into the corresponding product. All compounds were identified by coinjection with authentic samples. The data underlying this figure can be found at https://zenodo.org/records/12606239. ADALP1, adenosine deaminase-like protein 1; CatA, Cathepsin A; CES1, carboxylesterase 1; GUK1, guanylate kinase 1; HINT1, histidine triad nucleotide-binding protein type 1; NDPK, nucleoside diphosphate kinase; RP-HPLC, reversed-phase high-performance liquid chromatography. https://doi.org/10.1371/journal.pbio.3002743.g001 Most information relative to activation pathways has been obtained through measurements of intracellular concentrations of NA intermediates [10,11,14,24]. Studies described herein aim to ascertain which individual enzymes are involved in the activation pathway, as well as clarify their structural and functional mode of interaction with activation intermediates. Interaction maps of NAs with both activating partners and viral enzymes at atomic resolution would guide improved design of novel NAs. In this work, we have identified a minimal set of 5 human enzymes involved in the activation of bemnifosbuvir and AT-752 to their active 5′-TP form AT-9010. We demonstrate distinct stereo-preference of CatA and CES1 for AT-527 and its epimer AT-752, respectively. We elucidate an ordered activation pathway with potential limiting steps, including substrate specificity requirements at the N6-purine position for ADALP1. Structural analysis with key human enzymes HINT1, ADALP1, GUK1, and NDPK in complex with the bemnifosbuvir/AT-752 intermediate NA at 2.09, 2.44, 1.76, and 1.9 Å resolution informs on both the N6-purine and 2′-ribose chemosteric requirements leading to active NAs targeting viral RdRps.

Discussion Here, we cast light on the activation pathway of bemnifosbuvir and its epimer AT-752. Our work quantitates the differential activation of these substrates relative to their natural counterpart GTP. We visualize and measure essential interactions these 2 compounds and their metabolites engage with specific enzymes involved in their sequential activation. Turnover measurements (Table 1) show that the rate-limiting steps are generally occurring early in the pathway (CatA/CES1, HINT1, and ADALP1) rather than at the level of the kinases. On these slow, early steps, we note a de-esterification rate approximately 14-fold higher for AT-511 than AT-281 into the common intermediate AT-551 by CatA. Comparatively, CES1 reacts approximately 300-fold slower than CatA with AT-511 but shows an opposite stereo-preference with AT-281 being preferred approximately 8-fold over AT-511. These results are aligned with those obtained with PSI-7851, a racemic mixture of PSI-7976 and PSI-7977 (SOF), which shows both the same CatA and stereoisomer preference [14]. A number of NA prodrugs (e.g., SOF, TAF, RDV) need CatA/CES1 [10,11,37,38]-mediated activation. However, CatA is inhibited by the antiviral drug telaprevir (an HCV protease inhibitor) [10,14], an observation that casts light on drug–drug interaction at the molecular level. Murakami and colleagues [14] have shown that HINT1 is the rate-limiting step along the SOF activation pathway. Its metabolic intermediate PSI-352707 (S1 Fig) binds poorly to HINT1, precluding saturating substrate conditions in infected cells and determination of kcat and Km in enzyme assays. In the case of AT-551, HINT1 is also the rate-limiting step (Table 1). Despite intense crystal soaking efforts using AT-551, we observed only weak positive partial density orthogonal to the expected, “natural substrate” position where phosphate of AT551 faces the catalytic His 112. Co-crystallization studies resulted in a clean HINT1:AT-8003 complex structure, which proposes a possible structural basis for the kinetic bottleneck along the SOF activation line [14]. It may well be that the higher HINT1 conversion activity of AT-551 as compared to the corresponding intermediate metabolite of SOF, PSI-352707, which differ only by their nucleobase, results from the higher stacking power of purines versus uracil onto Ile 44. The HINT1 limiting step, however, does not seem to impact accumulation of the SOF 5′-triphosphate form (nor AT-9010) in cells. The compound pools are then readily available for selective use by the NS5B viral RdRp in HCV-infected liver cells. These results indicate that cell types and their relative expression of these “early step” enzymes CatA/CES1 and HINT1 are expected to play a major role in drug activation: (i) the esterase-mediated deprotection step may well govern the overall rate of conversion of the drug into its active metabolite; (ii) the relative abundance of CatA- and CES1-like enzymes should influence these early steps; and (iii) further optimization in antiviral activity might be possible through optimization of the prodrug part in the 5′ portion of the NA rather than focusing on optimizing the kinase steps. The choice of targeted infected cells may ultimately point to the most appropriate epimer and prodrug. In the case of ADALP1, unlike HINT1, the substrate and products are superimposable in the crystal structure, indicating that this complex should be reliable and useful for drug design. Ideally, the NA activation line must be proven functional in cells, organs, or animal models supporting preclinical tests before commitment to full clinical trials in humans. AlphaFold2 [39]-generated structural models of homologous animal enzymes did not point to obvious polymorphisms potentially deleterious for activation. The very high degree of structural conservation along species suggests a vital role in mammals for these activation enzymes. As such, monitoring differential expression of HINT1 should remain a sufficient and essential asset in predicting P-N hydrolysis in a given tissue or cell. The situation is somewhat similar for ADALP1, which shows significant polymorphisms across organisms, although mapping them invariably points to external, solvent exposed loops, with no direct impact on the NA binding site. In the case of exposition of cells to bemnifosbuvir, its corresponding 5′-triphosphate AT-9010 has been demonstrated in cellular concentrations up to approximately 700 μM in human primary cells and cell types incubated in vitro with AT-511, including bronchial and nasal epithelial cells [18], hepatocytes, and Huh7 cells [24]. Thus, none of the enzymes studied here limit the formation of the active compound AT-9010 in these cells. Bemnifosbuvir, though, shows reduced antiviral activity in some cell lines, e.g., in VeroE6, HeLa, or MRC-5 cells [18]. We note that cynomolgus monkey hepatocytes have been reported to activate bemnifosbuvir to levels 50-fold lower than Huh7 cells [18]. The superimposition of human and cynomolgus HINT1 and ADALP1 structures does not point to any polymorphism that could substantiate differences in substrate binding or catalysis. Bemnifosbuvir shares 2 activation enzymes (HINT1 and NDPK) with SOF and RDV. Hence, the HINT1 structure should represent the most relevant enzyme to optimize the prodrug part of such antivirals. Two other enzymes ADALP1 and GUK1 are specific for diamino purine- and guanine-containing NAs, respectively. With the abacavir success in mind [22], the ADALP1 structure presented here should guide further exploration of chemical diversity at the N6-amino purine group. A wealth of structural data already exists regarding NAs in complex with RNA virus replicases, e.g., HCV [15]; picornavirus [40]; Reovirus [41]; SARS-CoV-2 [13,28,42]; for a repository of stacked RdRp structures, see [43]. Much like along the NA activation line, analysis of selectivity of viral polymerase in ternary complexes also identifies proximal amino acid side chains potentially able to accommodate chemical decorations without impacting the NA-5′-triphophate incorporation efficiency. Finally, an NA drug potential is also conversely determined by its innocuity toward host enzymes that could misuse their 5′-triphosphate form and alter cell metabolism or host nucleic acid. Human mitochondrial RNA polymerase (POLRMT) inhibition accounts, in part, for cellular toxicity of NAs [44]. Much like SOF-TP, AT-9010 is a poor substrate for POLRMT, which uses GTP >69,000-fold better [18]. Comparative examination of ternary complexes of viral RdRps and POLRMT [45] with their RNA and NA-TP substrates may further document the strategy to rationally integrate NA drug design from chemistry to selective inhibition of viral growth [44]. Our work casts light on a whole set of previously ill-defined reactions involved in drug activation and contributes to the integration of NA drug-design from bioavailability to precise and selective mechanism of action as direct-acting antivirals.

Materials and methods Materials Recombinant human CatA, cathepsin L (CatL), CES1, and trans-epoxy-succinyl-L-leucylamino(4-guanidino)butane (E-64) were purchased from R&D Systems. The respective genes coding for the human recombinant enzymes HINT1, ADALP1, GUK1, and NDPK were coded into a pNC-ET28 expression vector with an N-terminal hexahistidine tag (6xHis) and a TEV protease recognition site, purchased from Twist Bioscience. In humans, the NDPK gene is expressed as several isoforms, NDPK-A and NDPK-B being both in synthesis of nucleoside triphosphates other than ATP. The sequence of NDPK-B [26,27,46] was used throughout the study. Control compound AMP-NH 2 was from Biosynth, N6-Me-AMP was purchased from BioLog Life Science Institute, and AMP, IMP, GMP, GDP, and GTP were from Sigma Aldrich. Other compounds with a name starting with “AT” were synthesized by TopPharm, USA. Protein expression The DNA sequence encoding for ADALP1, NDPK, GUK1, and HINT1 were synthesized and cloned into a pET-28a (+) vector for HINT1 and GUK1, into a pNC-ET28 vector for NDPK, and into a pET-28 for ADALP1 (Twist Biosciences). All proteins were expressed in E. coli NEB C2566 cells (New England Biolabs). The cells were grown at 37°C in TB medium for HINT1, GUK1, and NDPK and LB Broth for ADALP1 containing 50 μg/mL Kanamycin until the absorbance at 600 nm reached 0.6 to 0.8. At this stage, induction was started with 0.5 mM IPTG and the cells were grown overnight at 16°C. Then, the cells were harvested and the pellets were suspended in lysis buffer (50 mM Tris (pH = 8.0), 300 mM NaCl, 10 mM imidazole, 1 mM PMSF, 0.25 mg/mL lysozyme, and 10 μg/mL DNase) for HINT1, GUK1, and NDPK, before storing them at −80°C. For ADALP1, the pellet was freeze-dried and stored at −80°C. HINT1, GUK1, and NDPK purification The lysate was disrupted and cleared by centrifugation at 12,000g for 30 min at 10°C. Then, the supernatant was loaded onto a Ni-NTA-beads (Thermo Fisher) batch and washed with 50 mM Tris (pH = 8.0), 300 mM NaCl, 20 mM imidazole buffer. The recombinant protein was then eluted with 50 mM Tris (pH = 8.0), 300 mM NaCl, 250 mM imidazole buffer. About 10% glycerol was added for HINT1. TEV protease was used to remove the N-terminus His-tag during an overnight dialysis in a buffer (50 mM Tris (pH = 8.0), 150 mM NaCl, 1 mM DTT) at 4°C for HINT1, at 37°C for GUK1, and at room temperature for NDPK. A second purification step on Ni-NTA beads was performed to remove any uncleaved protein before to achieve a SEC purification onto a Superdex 75 16/600 GE column (GE Healthcare) in a final buffer containing 10 mM Tris (pH = 8.0), 50 mM NaCl for HINT1; 20 mM Tris (pH = 8.0), 100 mM NaCl, 1 mM DTT for GUK1; and 50 mM Tris (pH = 8.0), 150 mM NaCl, 1 mM DTT for NDPK. The fractions were analyzed on SDS-PAGE and the ones containing the pure target protein were pooled. The purified protein was then concentrated until 10 mg/mL, aliquoted, and stored at −80°C. Same protocol was used to express and purify uncleaved NDPK for activity assays. ADALP1 purification Bacterial pellets were lysed in half buffer (50 mM HEPES (pH = 7.5), 300 mM NaCl, 10% glycerol, 0.5 mM TCEP) and half Master Mix Bug Buster (Merck) supplemented with a tablet of EDTA-free antiprotease cocktail (Roche) per 50 ml of lysate before sonication. The lysate was cleared by centrifugation at 12,000g for 30 min at 10°C, and the supernatant was applied onto a Talon Superflow beads batch (Merck). The immobilized proteins were washed with 50 mM HEPES (pH 7.5), 300 mM NaCl, 12.5 mM imidazole, 10% glycerol, 0.5 mM TCEP, then washed with 50 mM HEPES (pH 7.5), 1.5M NaCl, 10% glycerol, 0.5 mM TCEP, and eluted in buffer (50 mM HEPES (pH 7.5), 300 mM NaCl, 150 mM imidazole, 10% glycerol, 0.5 mM TCEP). TEV protease was used to remove the N-terminus His-tag during an overnight dialysis in a buffer 50 mM HEPES (pH = 7.5), 300 mM NaCl, 10% glycerol, 1 mM DTT. The untagged ADALP1 was further purified by a Talon Superflow beads batch (Merck). Finally, ADALP1 was purified by size exclusion chromatography using a S75 16/60 GE Column (GE Healthcare) equilibrated in buffer containing 10 mM Tris (pH = 8.0), 40 mM NaCl, 1 mM TCEP. The single ADALP1 peak corresponded to the monomeric form. It was concentrated to 10 mg/mL, flash-frozen in liquid nitrogen, and stored at −80°C. CatA enzyme assay Human recombinant CatA was first activated by following the manufacturer’s instructions (R&D Systems), incubating at 37°C 10 μg/mL CatA with 1 μg/mL CatL in an activation buffer (25 mM MES (pH = 6.0), 5 mM DTT) during 30 min. About 10 μM E-64 (CatL inhibitor) was then added before aliquoting and storage of activated CatA at −80°C. AT-511 and AT-281 hydrolysis by activated CatA was measured by incubating 100 μM compound in a reaction buffer containing 25 mM MES (pH 6.5), 100 mM NaCl, 1 mM DTT, 0.1% NP-40, and 20 nM enzyme at 37°C for 45 min. The reaction was started by adding the enzyme. At various time points, 10 μL aliquots were collected from the reaction mixtures mixed with EDTA 10 mM (final concentration) and stopped by heating the samples at 95°C for 5 min. The sample were filtered on centrifugal filters Microcon—10 (Sigma Aldrich). The filtrates were mixed with triethylammonium bicarbonate (TEAB) 1M (1:1) and analyzed onto a C18 reverse phase column (2.5 μm, 4.6 by 100 mm, Xbridge C18 BEH, Waters) equipped with a guard column; substrates and products were detected at 260 nm and eluted at a flow rate of 1 ml/min at 35°C with a gradient of 0% to 4% of MeCN in TEAB 50 mM (pH = 7) in 10 min. CES1 enzyme assay AT-511 and AT-281 hydrolysis by human recombinant CES1 (100 nM) was assayed by incubating at 37°C the enzyme with 100 μM compound in 50 mM Tris buffer (pH = 7.5), 0.1% NP-40 and 1 mM DTT for 2 h. The reaction was started by adding the substrate. At various time points, 20 μL aliquots were collected from the reaction mixtures mixed with EDTA 10 mM (final concentration) and stopped by heating the samples at 95°C for 5 min. The sample were filtered on centrifugal filters Microcon—10 (Sigma–Aldrich). The filtrates were mixed with TEAB 1M (1:1) and analyzed onto a C18 reverse phase column (2.5 μm, 4.6 by 100 mm, Xbridge C18 BEH, Waters) equipped with a guard column; substrates and products were detected at 260 nm and eluted at a flow rate of 1 ml/min at 35°C with a gradient of 0% to 4% of MeCN in TEAB 50 mM (pH = 7) in 10 min. HINT1 enzyme assay The hydrolytic reactions with HINT1 (100 nM) were performed on AMP-NH 2 and AT-551 (200 μM) in 20 mM HEPES buffer (pH 7.2), 20 mM KCl, 1 mM MgCl 2 , and 1 mM DTT, at 37°C for 2 h. The reaction was started by adding the enzyme. At various time points, 10 μL aliquots were collected from the reaction mixtures mixed with EDTA 10 mM (final concentration) and stopped by heating the samples at 95°C for 5 min. The sample are filtered on AcroPrep Advance 96-Well Filter Plates with 3K Omega membrane (Pall). The filtrates were mixed with TEAB 1M and analyzed onto a C18 reverse phase column (3 μm, 3 by 150 mm, Acclaim Polar Advantage II, Thermo Fisher) equipped with a guard column; substrates and products were detected at 260 nm and eluted at a flow rate of 0.5 ml/min at 30°C with a gradient of 0% to 15% of MeCN in TEAB 50 mM (pH = 7) in 17 min. ADALP1 enzyme assay ADALP1 (100 nM) activity was measured on several substrates (N6-Me-AMP, AT-8003, AT-8002, AT-8004, AT-8010, AT-551, AT-259, AT-229) at 200 μM concentration in a reaction buffer containing BTP (pH = 6.8), 100 mM NaCl, and 1 mM DTT. Reactions were incubated at 37°C for 2 h. About 1 μM ADALP1 was also tested with AT-8002, AT-551, AT-229, and AT-259. The reaction was started by adding the substrate. At various time points, 10 μL aliquots were collected from the reaction mixtures diluted with H 2 O and stopped by heating the samples at 95°C for 5 min. The sample were filtered on AcroPrep Advance 96-Well Filter Plates with 3K Omega membrane (Pall). The filtrates were mixed with TEAB 1M and analyzed onto a C18 reverse phase column (2.5 μm, 4.6 by 100 mm, Xbridge C18 BEH Premier, Waters) equipped with a guard column; N6-Me-AMP, AT-8002, and products were detected at 260 nm and eluted at a flow rate of 1 ml/min at 35°C with a gradient of 0% to 10% of MeCN in TEAB 50 mM (pH = 7) in 10 min. AT-8003, AT-551, and products were eluted at a flow rate of 1 ml/min at 35°C with a gradient of 0% to 15% of MeCN in TEAB 50 mM (pH = 7) in 10 min. AT-8004, AT-8010, AT-229, AT-259, and products were eluted at a flow rate of 1 ml/min at 35°C with a gradient of 0% to 25% of MeCN in TEAB 50 mM (pH = 7) in 10 min. GUK1 enzyme assay GUK1 (40 nM) was assayed on GMP, AT-8003, and AT-8001 (200 μM) in a reaction buffer containing 50 mM Tris buffer (pH = 8.0), 50 mM KCl, 5 mM MgCl 2 , 1 mM ATP, and 1 mM DTT. Reactions were incubated at 37°C for 1 h. The reaction was started by adding the enzyme. At various time points, 10 μL aliquots were collected from the reaction mixtures, mixed with EDTA 10 mM (final concentration), and stopped by heating the samples at 95°C for 5 min. The sample were filtered on AcroPrep Advance 96-Well Filter Plates with 3K Omega membrane (Pall). The filtrates were mixed with TEAB 1M and analyzed onto a C18 reverse phase column (3 μm, 3 by 150 mm, Acclaim Polar Advantage II, Thermo Fisher) equipped with a guard column and equilibrated with TEAB 50 mM (pH = 7) (buffer A). Substrates and products were detected at 260 nm and eluted at a flow rate of 0.5 ml/min at 30°C with a gradient of 0% to 10% of MeCN in TEAB 50 mM (pH = 7) in 12 min. NDPK enzyme assay NDPK (20 nM) was assayed on GDP and AT-8500 (200 μM) in a reaction buffer containing 50 mM Tris buffer (pH = 8.0), 50 mM KCl, 5 mM MgCl 2 , 1 mM ATP, and 1 mM DTT. Reactions were incubated at 37°C for 1 h. The reaction was started by adding the enzyme. At various time points, 10 μL aliquots were collected from the reaction mixtures mixed with EDTA 10 mM and stopped by heating the samples at 95°C for 5 min. The samples were filtered on AcroPrep Advance 96-Well Filter Plates with 3K Omega membrane (Pall). The filtrates were mixed with TEAB 1M and analyzed onto a C18 reverse phase column (3 μm, 3 by 150 mm, Acclaim Polar Advantage II, Thermo Fisher) equipped with a guard column; substrates and products were detected at 260 nm and eluted at a flow rate of 0.5 ml/min at 30°C with a gradient of 0% to 10% of MeCN in TEAB 50 mM (pH = 7) in 12 min. Crystallization of HINT1 Crystallization conditions were adapted from [21]. Briefly, co-crystals with AT-8003 were grown at 293.15 K, using a 1:1 ratio of HINT1 at 10 mg/mL with AT-8003 (final concentration 25 mM) to precipitant solution 0.1 M MES (pH 6.1 to 6.5), 27% to 30% PEG 8000. Crystals grew in a few days and were cryo-protected with reservoir solution supplemented with 20% PEG 400, and flash-frozen in liquid nitrogen at 100 K. Crystallization of ADALP1 Crystallization assay were set up using the sitting drop vapor diffusion method, using a 2:1 ratio of protein solution to the reservoir solution at 293.15 K. The co-crystals with AT-8001 were obtained by mixing ADALP1 at 17 mg/mL with AT-8003 (final concentration 10 mM) and 1:10,000 w/w dilution a-Chymotrypsin prior to crystallization and were grown from 1.1 to 2.1 M ammonium sulfate, 0.1 M sodium cacodylate/HCl (pH 5.3 to 6.3), and 0.2 M sodium chloride for 2 mo. Crystals were cryo-protected with reservoir solution supplemented with 20% glycerol and flash-frozen in liquid nitrogen at 100 K. Crystallization of GUK1 Large co-crystals of GUK1 (17 mg/mL) with AT-8001 (2 mM) supplemented with 5 mM MgCl2 were grown using the sitting drop method with a 3:1 ratio of protein solution to precipitant solution consisting of 66 mM sodium cacodylate (pH 6.5 to 7.5), 12.2% to 22.2% PEG 3350, 8.3% PEG 4000, 33 mM MES (pH 6.5), and 66 mM magnesium chloride over a period of 10 d at 293.15 K. Crystals were cryo-protected with reservoir solution supplemented with 20% glycerol and flash-frozen in liquid nitrogen at 100 K. Crystallization of NDPK Crystallization conditions were adapted from [46] using the sitting-drop vapor diffusion method using crystallization buffer. All crystals were grown at 277.15 K, using a 1:1 ratio of protein (5 mg/mL) mixed with AT-9010 (final concentration 1 mM) to precipitant solution (12% PEG 4000, 50 mM Tris (pH = 8.4), 16% glycerol, 1 mM DTT). Crystals grew in 3 d and were flash-frozen directly in liquid nitrogen at 100 K. Structure determination of the human HINT1:AT-8003 complex The dataset of HINT1 in complex with AT-8003 was collected on the Proxima-1 beamline at the Synchrotron SOLEIL. Dataset was processed using AUTOPROC [47]. The phase was obtained using Molecular Replacement using MOLREP [48] with the PDB entry 6N3V as a model. The ligand AT-8003 and geometry description files were generated using the GRADE2 server [49]. Structure handling and refinement were done using COOT [50] and BUSTER [51]. Structure determination of the human ADALP 1:AT-8001 complex The dataset of ADALP 1 in complex with AT-8001 was collected on the Proxima-2 beamline at the Synchrotron SOLEIL. Dataset was processed using AUTOPROC [47], and processed data were sent to the CCP4 cloud suite [52]. The pipeline is fully automatic. The phase was obtained using Molecular Replacement–MORDA [52] using PDB entry 6IV5 as a model. The ligand AT-8001 and geometry description files were generated using the GRADE2 server [49]. Minor correction on the structure and further refinement were done using COOT [50] and REFMAC5 [53], respectively. Structure determination of the human GUK 1:AT-8001 complex The dataset of GUK1 in complex with AT-8001 was collected on the ESRF ID23-1 beamline. Dataset was processed using AUTOPROC [47]. The phase was obtained using Molecular Replacement–MORDA [52] using PDB entry 4F4J as a search model and optimized with ARP/WARP [54]. Structure refinement was done using COOT [50] and BUSTER [51], respectively. Structure determination of the human NDPK:AT-8500 complex The dataset of NDPK in complex with AT-8500 was collected on the Proxima-2 beamline at the Synchrotron SOLEIL. Dataset was handled using the CCP4 suite [55]. Images were processed by XDS [56] and AIMLESS [57]. The phase was obtained using Molecular Replacement–PHASER [58] with the PDB entry 1NUE as a model. The ligand AT-8500 corresponding to the diphosphate form of AT-9010 was generated using the AceDRG program [59]. Structure handling and refinement were done using COOT [50] and REFMAC5 [53] software, respectively. All electron-density maps were inspected using COOT [50]. Extra density accounting for ions, and/or compounds were observed for all complexed structures. The structures were evaluated using MOLPROBITY [60] and PROCHECK [61]. Structural analysis and high-resolution figures were done with UCSF ChimeraX [62]. Facilities, statistics of data collections, refinements, and PDB deposition code are given in Table 2. Modeling of GUK 1 closed conformation Sequence of human GUK 1 was submitted to an homology modeling process using modeler V10 and the structure of mouse GUK complexed with GMP and ADP (PDB 1LVG) as template. The 2 sequences have 88% identity, which makes us confident that the resulting closed conformation model is reasonable. Docking of AT-8003 onto HINT1 The 3D model of HINT1 was energy minimized by the steepest gradient method of energy minimization followed by conjugate gradient minimization, using the MMTK and AMBERpackages [63–65]. Mol2 and PDB files format of the ligands and receptor were converted to PDBQT format using CHIMERA prior to docking. All the water and solvent atoms of the protein were removed, and the polar hydrogen and polar charge were added onto the ions and ligand prior to docking. The protein was kept rigid, while the ligand was allowed to rotate and explore more flexible binding pockets. Docking of the respective ligands into the cavity were performed iteratively using AUTODOCK VINA—version 1.1.2 [66]⁠. The best poses from the first round of docking were used as seed for the second round. The resulting first round of docking were carefully analyzed to retain the best poses. The grid box size dimensions was designed to include the long binding cleft and the catalytic site; its dimensions were 37.87 × 16.12 × 20.89. The default scoring function was used for docking. Ten binding modes of the docked complexes were obtained and sorted based on their binding energy, and amino acid residues present at a distance less of 3 Å were considered as the binding partners of the ligands. Six binding modes with the phosphate oriented toward the catalytic site were kept and compared to the experimental structure. One binding mode is overlapped with the structure. A control experiment with AMP was also performed. Figures representing the docked complexes have been generated using CHIMERA [62]. Docking of GMP and AT-8001 in GUK1 closed conformation The 3D model of GUK 1 was energy minimized by the steepest gradient method of energy minimization followed by conjugate gradient minimization, using the MMTK and AMBER packages [63–65]. Mol2 and PDB files format of the ligands and receptor were converted to PDBQT format using chimera prior to docking. All the water and solvent atoms of the protein were removed, and the polar hydrogen and polar charge were added onto the ions and ligand prior to docking. The protein was kept rigid, while the ligand was allowed to rotate and explore more flexible binding pockets. Docking of the respective ligands into the cavity were performed iteratively using AUTODOCK VINA [66]. The best poses from the first round of docking were used as seed for the second round. The resulting first round of docking were carefully analyzed to retain the best poses. The grid box size dimensions were first 40 × 40 × 40, to verify that our ligands will preferentially bind in the catalytic site. The grid box size was further optimized to 23.2 × 25.6 × 21.2, thus covering the binding pockets; the default scoring function was used for docking. As control for the procedure, GMP was docked following the same protocol, and the final pose is virtually identical to the one measure in the experimental structure (PDB 1LVG). Binding modes of the docked complexes were obtained and sorted based on their binding energy; ions and amino acid residues present at a distance less of 3 Å were considered as the binding partners of the ligands. Binding modes were compared to theses of the native structure. The interaction figures representing the docked complexes have been generated using CHIMERAX [67].

Acknowledgments We thank Veronique Fattorini, Camille Falcou, and Pierre Gauffre for synthetic gene design and help in protein purification. We thank Nancy Agrawal, Alex Vo, and Qi Huang for critical reading of the manuscript, together with Ashleigh Shannon and numerous colleagues for helpful suggestions along the project. The authors acknowledge La Fondation CNRS for funding management, SOLEIL and ESRF for provision of synchrotron radiation facilities, and we would like to thank beamline teams of Proxima-1, Proxima-2 and ID23-eh1 for their assistance during data collection.

[END]
---
[1] Url: https://journals.plos.org/plosbiology/article?id=10.1371/journal.pbio.3002743

Published and (C) by PLOS One
Content appears here under this condition or license: Creative Commons - Attribution BY 4.0.

via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/