Advertisement

Synthesis, testing, and computational modeling of pleuromutilin 1,2,3-triazole derivatives in the ribosome

  • Logan M. Breiner
    Affiliations
    Department of Chemistry, Virginia Polytechnic Institute and State University (Virginia Tech), Blacksburg, VA, 24061, United States

    Center for Emerging, Zoonotic, and Arthropod-borne Pathogens, Virginia Polytechnic Institute and State University (Virginia Tech), Blacksburg, VA, 24061, United States
    Search for articles by this author
  • Anthony J. Briganti
    Affiliations
    Department of Biochemistry, Virginia Polytechnic Institute and State University (Virginia Tech), Blacksburg, VA, 24061, United States
    Search for articles by this author
  • Jennifer P. McCord
    Affiliations
    Department of Chemistry, Virginia Polytechnic Institute and State University (Virginia Tech), Blacksburg, VA, 24061, United States
    Search for articles by this author
  • Moriah E. Heifetz
    Affiliations
    Department of Chemistry, Virginia Polytechnic Institute and State University (Virginia Tech), Blacksburg, VA, 24061, United States
    Search for articles by this author
  • Sophia Y. Philbrook
    Affiliations
    Department of Chemistry, Virginia Polytechnic Institute and State University (Virginia Tech), Blacksburg, VA, 24061, United States
    Search for articles by this author
  • Carla Slebodnick
    Affiliations
    Department of Chemistry, Virginia Polytechnic Institute and State University (Virginia Tech), Blacksburg, VA, 24061, United States
    Search for articles by this author
  • Anne M. Brown
    Affiliations
    Center for Emerging, Zoonotic, and Arthropod-borne Pathogens, Virginia Polytechnic Institute and State University (Virginia Tech), Blacksburg, VA, 24061, United States

    Department of Biochemistry, Virginia Polytechnic Institute and State University (Virginia Tech), Blacksburg, VA, 24061, United States

    Research & Informatics and Department of Biochemistry, Virginia Polytechnic Institute and State University (Virginia Tech), Blacksburg, VA, 24061, United States
    Search for articles by this author
  • Andrew N. Lowell
    Correspondence
    Corresponding author. Department of Chemistry, Virginia Polytechnic Institute and State University (Virginia Tech), Blacksburg, VA, 24061, United States.
    Affiliations
    Department of Chemistry, Virginia Polytechnic Institute and State University (Virginia Tech), Blacksburg, VA, 24061, United States

    Center for Emerging, Zoonotic, and Arthropod-borne Pathogens, Virginia Polytechnic Institute and State University (Virginia Tech), Blacksburg, VA, 24061, United States

    Faculty of Health Sciences, Virginia Polytechnic Institute and State University (Virginia Tech), Blacksburg, VA, 24061, United States
    Search for articles by this author
Open AccessPublished:November 17, 2022DOI:https://doi.org/10.1016/j.tchem.2022.100034

      Abstract

      Pleuromutilin antimicrobials have given rise to the most recently FDA approved class of antibiotics for systemic human use. In this work, we describe a synthesis, assay, modeling approach to pleuromutilin development for the highly complex bacterial ribosome. Libraries of substituted 1,2,3-triazole derivatives were synthesized at the pleuromutilin C20 position by applying a recent anti-Markovnikov hydroazidation protocol to directly install an azido group, and at the C22 position through established methods. To learn about the interactions of these libraries with the ribosome and assess the potential for subsequent derivatization, an unbiased computational modeling method was used to biochemically rationalize binding modes of the C20 and C22 pleuromutilin derivatives. A pattern emerged where the triazole and its pendant chain, be it off the C20 or C22 position, moved to occupy the space vacated by the C22 sulfide group of clinical pleuromutilin compounds. Subsequent activity testing and comparative ranking of the computationally docked derivatives to the in vitro activity results showed a high predictability rating for the C22 substituted compounds. These combined investigations reveal potential restrictions and sites for expansion, paving the way for the development of future pleuromutilin derivates and other ribosome targeting antibiotics.

      Graphical abstract

      Keywords

      1. Introduction

      Antimicrobial resistance (AMR) is a global health crisis, directly responsible for killing 1.27 million people in 2019 [
      Antimicrobial Resistance Collaborators
      Global burden of bacterial antimicrobial resistance in 2019: a systematic analysis.
      ], with the number of people succumbing to antibiotic-resistant pathogens projected to increase to 10 million yearly by 2050 [
      ]. The number of antibiotics approved by the Food & Drug Administration (FDA) has declined in recent decades [
      • Ventola C.L.
      The antibiotic resistance crisis: part 1: causes and threats.
      ], making effective new treatments more urgently needed. Natural products are an incredibly rich source of pharmaceuticals, especially antibiotics [
      • Newman D.J.
      • Cragg G.M.
      Natural products as sources of new drugs over the nearly four decades from 01/1981 to 09/2019.
      ], and a strong focus on cultivating those that exhibit low rates of resistance is particularly relevant for the development of effective antimicrobials that help suppress AMR.
      Pleuromutilin (2, Fig. 1), a tricyclic diterpenoid antibiotic, was first isolated from Pleurotus mutilus (now Clitopilus scyphoides) in 1951 [
      • Kavanagh F.
      • Hervey A.
      • Robbins W.J.
      Antibiotic substances from basidiomycetes: VIII. Pleurotus multilus (Fr.) sacc. And Pleurotus passeckerianus pilat.
      ]. Comprised of a mutilin core (1) with a glycolic ester side chain at the C14 position, pleuromutilin was shown to have activity against Gram-positive bacteria and mycoplasmas [
      • Egger H.
      • Reinshagen H.
      New pleuromutilin derivatives with enhanced antimicrobial activity. II. Structure-activity correlations.
      ]. Mutilin drugs exert their effect by interfering with peptide bond formation in the peptidyl transfer center (PTC). The mutilin core binds in a hydrophobic pocket in the A-site of the 50S subunit of the ribosome [
      • Schlünzen F.
      • Pyetan E.
      • Fucini P.
      • Yonath A.
      • Harms J.M.
      Inhibition of peptide bond formation by pleuromutilins: the structure of the 50S ribosomal subunit from Deinococcus radiodurans in complex with tiamulin.
      ] while the C14 sidechain interacts with nucleotides from the P-site [
      • Schlünzen F.
      • Pyetan E.
      • Fucini P.
      • Yonath A.
      • Harms J.M.
      Inhibition of peptide bond formation by pleuromutilins: the structure of the 50S ribosomal subunit from Deinococcus radiodurans in complex with tiamulin.
      ,
      • Davidovich C.
      • Bashan A.
      • Auerbach-Nevo T.
      • Yaggie R.D.
      • Gontarek R.R.
      • Yonath A.
      Induced-fit tightens pleuromutilins binding to ribosomes and remote interactions enable their selectivity.
      ]. Early investigations showed that a wide range of functionality could be installed on the ester side chain [
      • Egger H.
      • Reinshagen H.
      New pleuromutilin derivatives with enhanced antimicrobial activity. II. Structure-activity correlations.
      ]. Thus, semi-synthetic derivatization of pleuromutilin focused mainly on modification of the C22 position of the glycolic ester [
      • Goethe O.
      • Heuer A.
      • Ma X.
      • Wang Z.
      • Herzon S.B.
      Antibacterial properties and clinical potential of pleuromutilins.
      ,
      • Fazakerley N.J.
      • Procter D.J.
      Synthesis and synthetic chemistry of pleuromutilin.
      ].
      Fig. 1
      Fig. 1Pleuromutilin and its commercial derivatives.
      Despite initial concerns about potentially unfavorable ADMET (absorption, distribution, metabolism, excretion, and toxicity) properties, empirical results demonstrated that the mutilin class could be optimized to avoid these issues [
      • Prince W.
      • Obermayr F.
      • Ivezic-Schoenfeld Z.
      • Lell C.
      • Wicha W.
      • Strickmann D.
      • Tack K.
      • Novak R.
      In A Phase 2 Study Comparing the Safety and Efficacy of Two Doses of BC-3781 versus Vancomycin in Acute Bacterial Skin and Skin Structure Infections (ABSSSI), Abstr. L-966. Abstr. 51st Intersci, Program and Abstracts of the 51st Interscience Conference on Antimicrobial Agents and Chemotherapy (Chicago).
      ,
      • Rittenhouse S.
      • Biswas S.
      • Broskey J.
      • McCloskey L.
      • Moore T.
      • Vasey S.
      • West J.
      • Zalacain M.
      • Zonis R.
      • Payne D.
      Selection of retapamulin, a novel pleuromutilin for topical use.
      ]. The first approved drug was the veterinary compound tiamulin (3) [
      • Czok R.
      • Meingassner J.G.
      • Mieth H.
      • Schutze E.
      Antibiotic compositions for treating coccidiosis.
      ] with subsequent approvals including the veterinary drug valnemulin (6) [
      • Burch D.G.S.
      • Ripley P.H.
      • Zeisl E.
      ] and the topical agent retapamulin (5) [
      • Jacobs M.R.
      Retapamulin: a semisynthetic pleuromutilin compound for topical treatment of skin infections in adults and children.
      ]. Recent work by Nabriva Therapeutics generated lefamulin (4), which was approved by the FDA for systemic use in humans in 2019 [
      • Hunt A.
      FDA Approves New Antibiotic to Treat Community-Acquired Bacterial Pneumonia.
      ]. The mutilins are an attractive class for further improvement because of slow resistance development due to their unique binding site within the PTC, slow rates of spontaneous mutation [
      ,
      • Paukner S.
      • Riedl R.
      Pleuromutilins: potent drugs for resistant bugs—mode of action and resistance.
      ,
      • Yan K.
      • Madden L.
      • Choudhry A.E.
      • Voigt C.S.
      • Copeland R.A.
      • Gontarek R.R.
      Biochemical characterization of the interactions of the novel pleuromutilin derivative retapamulin with bacterial ribosomes.
      ,
      • Gentry D.R.
      • Rittenhouse S.F.
      • McCloskey L.
      • Holmes D.J.
      Stepwise exposure of Staphylococcus aureus to pleuromutilins is associated with stepwise acquisition of mutations in rplC and minimally affects susceptibility to retapamulin.
      ], and a lack of shared resistance mechanisms from other antibiotic classes [
      • Yan K.
      • Madden L.
      • Choudhry A.E.
      • Voigt C.S.
      • Copeland R.A.
      • Gontarek R.R.
      Biochemical characterization of the interactions of the novel pleuromutilin derivative retapamulin with bacterial ribosomes.
      ].
      Building on the success of first-generation pleuromutilin derivatives (Fig. 1), several groups are undertaking a variety of improvement strategies focused on taking advantage of pi-stacking–type interactions with the ribosomal RNA nucleotides. One variation of this approach is the use of click-conjugates, especially triazoles via copper-catalyzed alkyne-azide cycloadditions (CuAAC), which is attractive because of its ease and the increased acceptance of 1,2,3-triazole–containing leads in medicinal chemistry [
      • Bozorov K.
      • Zhao J.
      • Aisa H.A.
      1,2,3-Triazole-containing hybrids as leads in medicinal chemistry: a recent overview.
      ,
      • Jain A.
      • Piplani P.
      Exploring the chemistry and therapeutic potential of triazoles: a comprehensive literature review.
      ,
      • Marzi M.
      • Farjam M.
      • Kazeminejad Z.
      • Shiroudi A.
      • Kouhpayeh A.
      • Zarenezhad E.
      A recent overview of 1,2,3-triazole-containing hybrids as novel antifungal agents: focusing on synthesis, mechanism of action, and structure-activity relationship (SAR).
      ]. Initial investigations in this area by Nielsen and coworkers [
      • Lolk L.
      • Pøhlsgaard J.
      • Jepsen A.S.
      • Hansen L.H.
      • Nielsen H.
      • Steffansen S.I.
      • Sparving L.
      • Nielsen A.B.
      • Vester B.
      • Nielsen P.
      A click chemistry approach to pleuromutilin conjugates with nucleosides or acyclic nucleoside derivatives and their binding to the bacterial ribosome.
      ,
      • Dreier I.
      • Kumar S.
      • Søndergaard H.
      • Rasmussen M.L.
      • Hansen L.H.
      • List N.H.
      • Kongsted J.
      • Vester B.
      • Nielsen P.
      A click chemistry approach to pleuromutilin derivatives, part 2: conjugates with acyclic nucleosides and their ribosomal binding and antibacterial activity.
      ,
      • Dreier I.
      • Hansen L.H.
      • Nielsen P.
      • Vester B.
      A click chemistry approach to pleuromutilin derivatives. Part 3: extended footprinting analysis and excellent MRSA inhibition for a derivative with an adenine phenyl side chain.
      ,
      • Heidtmann C.V.
      • Voukia F.
      • Hansen L.N.
      • Soerensen S.H.
      • Urlund B.
      • Nielsen S.
      • Pedersen M.
      • Kelawi N.
      • Andersen B.N.
      • Pedersen M.
      • Reinholdt P.
      • Kongsted J.
      • Nielsen C.U.
      • Klitgaard J.K.
      • Nielsen P.
      Discovery of a potent adenine-benzyltriazolo-pleuromutilin conjugate with pronounced antibacterial activity against MRSA.
      ] produced compounds that rivaled the activity of tiamulin (3) with additional, conformationally restricted analogs (such as 7, Fig. 2) surpassing it [
      • Dreier I.
      • Kumar S.
      • Søndergaard H.
      • Rasmussen M.L.
      • Hansen L.H.
      • List N.H.
      • Kongsted J.
      • Vester B.
      • Nielsen P.
      A click chemistry approach to pleuromutilin derivatives, part 2: conjugates with acyclic nucleosides and their ribosomal binding and antibacterial activity.
      ,
      • Heidtmann C.V.
      • Voukia F.
      • Hansen L.N.
      • Soerensen S.H.
      • Urlund B.
      • Nielsen S.
      • Pedersen M.
      • Kelawi N.
      • Andersen B.N.
      • Pedersen M.
      • Reinholdt P.
      • Kongsted J.
      • Nielsen C.U.
      • Klitgaard J.K.
      • Nielsen P.
      Discovery of a potent adenine-benzyltriazolo-pleuromutilin conjugate with pronounced antibacterial activity against MRSA.
      ]. Other groups have likewise derivatized various spacers, such as 2-aminothiophenol (8) [
      • Zhang Z.
      • Li K.
      • Zhang G.-Y.
      • Tang Y.-Z.
      • Jin Z.
      Design, synthesis and biological activities of novel pleuromutilin derivatives with a substituted triazole moiety as potent antibacterial agents.
      ] or piperazine (9) [
      • Zhang G.-Y.
      • Zhang Z.
      • Li K.
      • Liu J.
      • Li B.
      • Jin Z.
      • Liu Y.-H.
      • Tang Y.-Z.
      Design, synthesis and biological evaluation of novel pleuromutilin derivatives containing piperazine and 1,2,3-triazole linker.
      ], with CuAAC chemistry to generate novel triazole derivatives. Not only were these compounds more effective than tiamulin at reducing bacterial load and increasing survival in murine MRSA models, but they also showed only moderate inhibition of CYP3A4 [
      • Stresser D.M.
      • Broudy M.I.
      • Ho T.
      • Cargill C.E.
      • Blanchard A.P.
      • Sharma R.
      • Dandeneau A.A.
      • Goodwin J.J.
      • Turner S.D.
      • Erve J.C.L.
      • Patten C.J.
      • Dehal S.S.
      • Crespi C.L.
      Highly selective inhibition of human CYP3A in vitro by azamulin and evidence that inhibition is irreversible.
      ], a metabolic enzyme responsible for the breakdown of common drugs.
      Fig. 2
      Fig. 2Triazole containing pleuromutilin derivatives.
      In contrast to C22 substitutions, derivatization of the C19–C20 vinyl group is just starting to be explored. Nabriva Therapeutics has functionalized the vinyl group through hydroboration-oxidation (not shown) or ozonolysis followed by reductive amination resulting in diamines (such as 10, Fig. 3), including C12 epimers (11) [
      • Thirring K.
      • Heilmayer W.
      • Riedl R.
      • Kollmann H.
      • Ivezic-Schoenfeld Z.
      • Wicha W.
      • Paukner S.
      • Strickmann D.
      Preparation of 12-epi-pleuromutilin derivatives as antimicrobial agents.
      ]. Other work by Xianfeng and coworkers using C22-benzoxaboroles (inset) demonstrated that a pendant phenyl group (12), alkyl amines (13), and 5-membered heterocycles (14 and 15) were well tolerated in place of the vinyl group, increasing activity over the parent by an order of magnitude [].
      Fig. 3
      Fig. 3C20 modifications of pleuromutilin.
      Computational docking methods that can be applied to aid in the development of pleuromutilin derivatives, and for ribosome-targeting antibiotics generally, are lacking. The difficulty is in assessing the extraordinarily large ribosome (2.5 ​MDa) while also maintaining a sufficiently sensitive approach that enables analysis of the interaction of antibiotic ligands with the underlying nucleotide and protein structure of the ribosome. Beyond assisting in the specific development of new pleuromutilin or other antibiotic derivatives, the application of computational methods could also eventually be used to answer questions about the activity (or lack thereof) of previously reported compounds.
      Thus, the opportunity to apply computational methodologies to the ribosome coupled with these synthetic findings spurred us to explore triazole derivatives of the glycolic ester and of the vinyl group. Established substitution conditions for the C22 alcohol would enable facile installation of an azide and subsequent CuAAC chemistry would enable access to a variety of flexible functionality capable of being further elaborated beyond the rotationally restricted aromatic and saturated rings that have been used previously. Activation of the C20 position, however, would require optimization of recent methodological breakthroughs. Accordingly, we sought to incorporate azide functionality at the C20 and C22 positions and subsequently derivatize them into a library of substituted triazoles. These compounds could then be tested for activity and used as a training set to assess computational approaches for analyzing ribosome-binding antibiotics.

      2. Results and discussion

      2.1 C22 derivative synthesis by tosylation and direct displacement

      In the first set of derivatives, the C22-triazoles (Fig. 4) were synthesized via the well-known intermediate 22-O-tosylpleuromutilin (16) []. This electrophile underwent conversion to the α-azidoester (17) via SN2 displacement with sodium azide as previously described [
      • Lolk L.
      • Pøhlsgaard J.
      • Jepsen A.S.
      • Hansen L.H.
      • Nielsen H.
      • Steffansen S.I.
      • Sparving L.
      • Nielsen A.B.
      • Vester B.
      • Nielsen P.
      A click chemistry approach to pleuromutilin conjugates with nucleosides or acyclic nucleoside derivatives and their binding to the bacterial ribosome.
      ]. The triazoles (1828) were produced using Sharpless's technique [
      • Himo F.
      • Lovell T.
      • Hilgraf R.
      • Rostovtsev V.V.
      • Noodleman L.
      • Sharpless K.B.
      • Fokin V.V.
      Copper (I)-catalyzed synthesis of azoles. DFT study predicts unprecedented reactivity and intermediates.
      ] with heating of 17 and the appropriate alkyne to approximately 70 ​°C. Most compounds were obtained in high yields but increased steric bulk near the alkyne reduced transformation efficacy (2426).
      Fig. 4
      Fig. 4Synthesis of C22 triazoles.
      During our preparation of 22-O-tosylpleuromutilin (16), we isolated and identified 22-deoxy-22-chloro-pleuromutilin (29, Fig. 5), a material that has been intentionally generated [
      • Riedl K.
      Studies on pleuromutilin and some of its derivatives.
      ], but never to our knowledge reported as a byproduct of the tosylation reaction. In a procedure using 4-dimethylaminopyridine as a catalyst, we observed that the mother liquor from the recrystallization of 16 showed a non-trivial amount of mutilin-containing material. Chromatographic separation and characterization, including X-ray crystallography, confirmed this material to be 29, which we isolated in a 20% yield. This material must arise from the nucleophilic displacement of the tosylate in 16 by the chloride anion formed during tosylation. The displacement of sulfonates by chloride during sulfonylation is known, especially with benzylic substrates [
      • Chappe B.
      • Musikas H.
      • Marie D.
      • Ourisson G.
      Synthesis of three acyclic all-trans-tetraterpene diols, putative precursors of bacterial lipids.
      ,
      • Ding R.
      • He Y.
      • Wang X.
      • Xu J.
      • Chen Y.
      • Feng M.
      • Qi C.
      Treatment of alcohols with tosyl chloride does not always lead to the formation of tosylates.
      ,
      • Osorio L.S.
      • Ionta M.
      • Demuner A.J.
      • Sousa B.L.d.
      • Ferraz G.O.
      • Varejão E.V.
      • Ferreira-Silva G.A.
      • Pilau E.J.
      • Silva E.
      • Santos M.H.d.
      Synthesis of 1,2,3-triazole derivatives of hydnocarpic acid isolated from Carpotroche brasiliensis seed oil and evaluation of antiproliferative activity.
      ]. As 16 is an intermediate in nearly all synthetic schemes that involve pleuromutilin modifications at C22, suppression of 29 or designing synthetic strategies that take advantage of its co-generation is an important consideration in the semisynthesis of pleuromutilin derivatives.
      Fig. 5
      Fig. 522-Deoxy-22-chloropleuromutilin and its X-ray crystal structure.

      2.2 C20 derivative synthesis using a modified anti-Markovnikov hydroazidation protocol

      Toward installation of a C20 azido component, we sought to modify the vinyl group into an azide directly instead of using a multi-step protocol. Previous studies describing the stoichiometric [
      • Li X.
      • Chen P.
      • Liu G.
      Iodine(III) reagent (ABX—N3)-induced intermolecular anti-Markovnikov hydroazidation of unactivated alkenes.
      ] or catalytic [
      • Li H.
      • Shen S.-J.
      • Zhu C.-L.
      • Xu H.
      Direct intermolecular anti-Markovnikov hydroazidation of unactivated olefins.
      ] hydroazidation of vinyl compounds suggested that this approach was achievable on pleuromutilin (2, Fig. 6). The resulting novel intermediate (30) could then be decorated with diverse functionality through CuAAC chemistry.
      Fig. 6
      Fig. 6Synthesis of C20 triazoles.
      Our initial trials focused on the catalytic hydroazidation of 2 using 2-iodosobenzoic acid (IBA), TMSN3, trifluoroacetic acid (TFA), and water [
      • Li H.
      • Shen S.-J.
      • Zhu C.-L.
      • Xu H.
      Direct intermolecular anti-Markovnikov hydroazidation of unactivated olefins.
      ]. This approach created small amounts of 30, but the acid and water present in the reaction mixture also hydrolyzed 2 to 1. Removal of the acid from the reaction mixture eliminated hydrolysis, but yields were low and silylation of the primary and secondary alcohols was observed. Intentional silylation prior to hydroazidation significantly reduced the yield of 30, likely due to steric hindrance around the vinyl group. Further complicating a catalytic approach, the active catalyst species degraded over time, restricting conversion.
      After experimenting with ratios of reagents, we determined that appreciable conversion of 2 to 30 could be achieved using a super-stoichiometric amount of IBA and TMSN3 along with omission of the acid co-catalyst and inclusion of excess water. This combination afforded the highest yield of 30 without the production of silylated byproducts. Even under these optimized conditions, full conversion of the starting material was not observed without degradation of the catalyst. While the overall yield of 30 was modest (32%), this one-step protocol avoided a longer sequence of protection, hydroboration, activation, displacement, and deprotection.
      Concerningly, the polarity of the desired hydroazide (30) prevented its separation from pleuromutilin (2) under standard chromatographic conditions (normal phase, SiO2). However, the triazole achieved after CuAAC was easily separable from 2, preserving this atom economic approach. Thus, a library of C20-triazoles were synthesized by reacting the resulting 80%/20% mixture of 30/2 with a variety of alkynes, giving 3141 in 26–83% yield. As with the C22 derivatives, steric bulk near the alkyne generally decreased yields.
      Azides are an increasingly popular functional group due to their ease of derivatization into triazoles [
      • Himo F.
      • Lovell T.
      • Hilgraf R.
      • Rostovtsev V.V.
      • Noodleman L.
      • Sharpless K.B.
      • Fokin V.V.
      Copper (I)-catalyzed synthesis of azoles. DFT study predicts unprecedented reactivity and intermediates.
      ] and the acceptance of triazoles in drugs [
      • Lengerli D.
      • Ibis K.
      • Nural Y.
      • Banoglu E.
      The 1,2,3-triazole “all-in-one” ring system in drug discovery: a good bioisostere, a good pharmacophore, a good linker, and a versatile synthetic tool.
      ]. However, azide chemistry can be hazardous because the presence of metals or halogenated solvents can cause an explosion and hydrazoic acid is both potentially explosive and toxic [
      • Treitler D.S.
      • Leung S.
      How dangerous is too dangerous? A perspective on azide chemistry.
      ]. Of these potential dangers, the use of halogenated solvents and generation of hydrazoic acid are concerns for the current work. During the hydroazidation reaction (which is performed behind a blast shield), hydrazoic acid is generated transiently from TMS-azide and rapidly used up in the catalytic cycle. Excess hydrazoic acid contributes to the breakdown of IBA, which liberates the acid as nitrogen gas [
      • Li H.
      • Shen S.-J.
      • Zhu C.-L.
      • Xu H.
      Direct intermolecular anti-Markovnikov hydroazidation of unactivated olefins.
      ]. Although the reaction is performed in dichloromethane, the neutral or acidic reaction conditions exclude the formation of azide anions needed to displace the chlorides and generate diazidomethane. The workup basifies any remaining hydrazoic acid, converting it to sodium azide that could potentially react with dichloromethane to form an explosive substance. However, the generated azide anions are only briefly exposed to a solution of dichloromethane during extraction. Formation of diazidomethane is slow, taking weeks at room temperature, even under concentrated conditions [
      • Hassner A.
      • Stern M.
      • Gottlieb H.E.
      • Frolow F.
      Synthetic methods. 33. Utility of a polymeric azide reagent in the formation of di- and triazidomethane. Their NMR spectra and the x-ray structure of derived triazoles.
      ]. Thus, the risk of generating dangerously unstable or toxic azide species in this reaction is minimal, but care should be taken as outlined by the cautionary statements in the experimental.

      2.3 Derivatization location impacts antibacterial activity

      The minimum inhibitory concentration (MIC) of our libraries of C22 and C20 derivatives was measured against a series of Gram-positive and Gram-negative bacteria (Table 1, Table 2). Pleuromutilin was used as a positive control.
      Table 1C22 Triazole derivatives. MICs are reported in μg/mL. Broth dilution assay performed in duplicate. NorA and TolC strains are efflux pump knockouts of their respective organisms. All tested compounds were inactive against A. baumanii at 64 ​μg/mL.
      Table thumbnail fx1
      Table 2C20 Triazole derivatives. MICs are reported in μg/mL. Broth dilution assay performed in duplicate. All tested compounds were inactive against A. baumanii at 64 ​μg/mL.
      Table thumbnail fx2

      2.3.1 C22 pleuromutilin derivatives

      A portion of the C22-triazole derivatives (Table 1), namely the homologated alcohol series (18–20), the hexyl chain (27), and the benzyl ether (28), showed comparable activity to pleuromutilin in Staphylococcus aureus, both methicillin-resistant S. aureus (MRSA) and S. aureus with the NorA efflux pump removed (NorA) [
      • Patel D.
      • Kosmidis C.
      • Seo Susan M.
      • Kaatz Glenn W.
      Ethidium bromide MIC screening for enhanced efflux pump gene expression or efflux activity in Staphylococcus aureus.
      ]. The homologated alcohols were also comparable to pleuromutilin in vancomycin-resistant Enterococcus (VRE) growth inhibition. In addition, azido derivative 17 was comparable or slightly outperformed pleuromutilin against all species, showing broad spectrum activity corresponding to that reported by Reidl [
      • Riedl K.
      Studies on pleuromutilin and some of its derivatives.
      ]. The homologated amines (21 and 22) showed diminished activity with all species, although their positive charge under physiological conditions likely resulted in their relative retention of activity against Escherichia coli. The other moderate-to-weakly active compounds against E. coli were methyl and ethyl alcohols 18 and 19, branched alcohols 23 and 24, and methyl benzyl ether 28, suggesting that a polar atom close to the triazole makes critical contacts in the binding pocket. However, the mixed results with alcohols with longer alkyl chains (25 versus 26) indicate the binding in this region is complicated. The activity of the other C22-triazole compounds was moderate to weak against E. coli TolC (the efflux pump knockout strain of E. coli [
      • Zgurskaya H.I.
      • Krishnamoorthy G.
      • Ntreh A.
      • Lu S.
      Mechanism and function of the outer membrane channel TolC in multidrug resistance and physiology of enterobacteria.
      ]) and no compound had activity at 64 ​μg/mL against Acinetobacter baumannii, which may be an issue of permeability in this Gram negative pathogen or differences in its ribosome binding site [
      • Gordon N.C.
      • Wareham D.W.
      Multidrug-resistant Acinetobacter baumannii: mechanisms of virulence and resistance.
      ]. However, there was no general correlation with the chemical property data (Tables S11 and S12) suggesting that all derivatives acted on the strains in the same basic way. While only 19 was active against E. coli, the entire series of short, homologated alcohols (1820) were equally efficacious against the two S. aureus species and showed a slight increase in potency with increasing chain length against VRE, suggesting that the C22 triazole was accommodated by the binding site in all these species but subtler binding pocket difference beyond the triazole changed the activity due to differences in hydrogen bonding contacts. Despite having the same length as 18, secondary and tertiary alcohols 23 and 24 were less active against MRSA and variable against the NorA knockout, indicating that a combination of binding site differences and efflux are responsible for the change in activity. Likewise, secondary alcohols attached to a longer alkyl chain (25 and 26) showed variable activity. With the butyl chain (25), efflux may be responsible for the differences in inhibition as activity was maintained against the NorA knockout but lost against MRSA. With the hexyl chain (26), activity between MRSA and the NorA knockout strain were identical suggesting that it is not prone to efflux. The alkyl derivative (27) showed similar potency to pleuromutilin against S. aureus, but lost efficacy against VRE and was inactive against E. coli. As the E. coli strain is efflux deficient, the change in activity is likely due to subtle differences between the S. aureus and E. coli binding sites. Lastly, the benzyl ether derivative (28) showed a marginal reduction in activity for the Gram-positive strains compared to pleuromutilin and matched previously reported activity against E. coli [
      • Dreier I.
      • Kumar S.
      • Søndergaard H.
      • Rasmussen M.L.
      • Hansen L.H.
      • List N.H.
      • Kongsted J.
      • Vester B.
      • Nielsen P.
      A click chemistry approach to pleuromutilin derivatives, part 2: conjugates with acyclic nucleosides and their ribosomal binding and antibacterial activity.
      ].

      2.3.2 C20 pleuromutilin derivatives

      Unexpectedly, and in contrast to previous results where heterocycles attached to the C20 position showed activity [], few of the C20-triazole derivatives demonstrated inhibition of the test organisms (Table 2). However, our results do align with the Nabriva findings, where C20 derivatives (such as 10, Fig. 3) lost activity compared to the parent compound in vitro [
      • Thirring K.
      • Heilmayer W.
      • Riedl R.
      • Kollmann H.
      • Ivezic-Schoenfeld Z.
      • Wicha W.
      • Paukner S.
      • Strickmann D.
      Preparation of 12-epi-pleuromutilin derivatives as antimicrobial agents.
      ]. One possible explanation is that without a C22 sidechain or an intact double bond at the C19–C20 position [], our C20 functionalized compounds did not bind or bound in an ineffective configuration. Of the compounds tested, only two showed an appreciable level of activity, the hexyl derivative (41) against E. coli and the azide precursor (30) against everything except VRE. Compound 41 being active in contrast to its C22 partner 27, which was inactive, indicates that uptake of alkyl-substituted pleuromutilins is probably not a critical issue; instead, this result suggests that changes in the binding pocket among the different ribosomes influences binding and hence activity. This insight into binding site differences strongly supports the continued application and development of computational methods to help us learn about binding site differences in the ribosome and guide the development of future derivatives. Although these compounds were generally inactive in our assays, their synthesis demonstrates the synthetic utility of hydroazidating vinyl-containing natural products.

      2.4 In-silico exploration of triazole derivative binding modes and efficacy predication

      Predicting antibiotic derivative efficacy via computational methods is a growing area of research in rational antibiotic redesign [
      • König G.
      • Sokkar P.
      • Pryk N.
      • Heinrich S.
      • Möller D.
      • Cimicata G.
      • Matzov D.
      • Dietze P.
      • Thiel W.
      • Bashan A.
      • Bandow Julia E.
      • Zuegg J.
      • Yonath A.
      • Schulz F.
      • Sanchez-Garcia E.
      Rational prioritization strategy allows the design of macrolide derivatives that overcome antibiotic resistance.
      ]; however, determining protocols and acquiring training datasets are critical advancements for a large, complex cellular target, such as the bacterial ribosome. Historically, molecular docking programs have been optimized for small-molecule–protein interactions. With the continued emergence of nucleic acid polymers as therapeutic targets, these programs are finding new, more demanding applications. Evaluations of currently used docking programs to predict small molecule binding modes (poses) in structures partially or fully comprised of nucleic acids [
      • Tessaro F.
      • Scapozza L.
      How ‘protein-docking’ translates into the new emerging field of docking small molecules to nucleic acids?.
      ] suggests that AutoDock Vina [
      • Trott O.
      • Olson A.J.
      Autodock Vina: improving the speed and accuracy of docking with a new scoring function, efficient optimization, and multithreading.
      ] outperforms Schrödinger Glide [
      • Friesner R.A.
      • Murphy R.B.
      • Repasky M.P.
      • Frye L.L.
      • Greenwood J.R.
      • Halgren T.A.
      • Sanschagrin P.C.
      • Mainz D.T.
      Extra precision Glide: docking and scoring incorporating a model of hydrophobic enclosure for Protein−Ligand complexes.
      ] in predictive ability [
      • Ruiz-Carmona S.
      • Alvarez-Garcia D.
      • Foloppe N.
      • Garmendia-Doval A.B.
      • Juhos S.
      • Schmidtke P.
      • Barril X.
      • Hubbard R.E.
      • Morley S.D.
      rDock: a fast, versatile and open source program for docking ligands to proteins and nucleic acids.
      ]. Accurate binding pose prediction was considered one of the most important criteria when selecting a molecular docking program for our exploration into prokaryotic ribosomal small molecule inhibitors. Accurate pose prediction gives invaluable insight on drug positioning within the ribosomal binding cavity, and therefore insight into drug repositioning through modification, which then can be applied to future drug design. Thus, we worked to create a molecular docking protocol that could reliably dock antibiotics and novel derivatives into small or large ribosomal subunits while accurately ranking them as compared to experimental MIC values.
      Although inherently limited by their focus on the binding target (other important factors such as cellular penetration and efflux of drugs cannot be taken into account by docking), previous studies established that clipping the ribosomal binding cavity to a subset of relevant binding nucleotides and residues minimizes the effects of structural inaccuracies [
      • Kufareva I.
      • Abagyan R.
      Methods of protein structure comparison.
      ]. Within the 2.5 ​MDa ribosome complex, ribonucleoprotein elements within 3.5 ​Å of the co-crystalized ligand have the strongest interatomic contacts and are responsible for most direct binding interactions. Thus, all ribonucleoprotein elements that contained nucleotides or residues within 15 ​Å of the crystal structure ligand were considered relevant for antibiotic binding.

      2.4.1 Computational method design

      In order to establish a robust protocol to apply computational design to ribosomal drug discovery, ribosome structures with antibiotics that target various regions were collected from the RCSB PDB (PDB IDs: 1FJG [
      • Carter A.P.
      • Clemons W.M.
      • Brodersen D.E.
      • Morgan-Warren R.J.
      • Wimberly B.T.
      • Ramakrishnan V.
      Functional insights from the structure of the 30S ribosomal subunit and its interactions with antibiotics.
      ], 4V56 [
      • Borovinskaya M.A.
      • Shoji S.
      • Holton J.M.
      • Fredrick K.
      • Cate J.H.D.
      A steric block in translation caused by the antibiotic spectinomycin.
      ], 5HL7 [
      • Eyal Z.
      • Matzov D.
      • Krupkin M.
      • Paukner S.
      • Riedl R.
      • Rozenberg H.
      • Zimmerman E.
      • Bashan A.
      • Yonath A.
      A novel pleuromutilin antibacterial compound, its binding mode and selectivity mechanism.
      ], 4V9Q [
      • Svidritskiy E.
      • Ling C.
      • Ermolenko D.N.
      • Korostelev A.A.
      Blasticidin S inhibits translation by trapping deformed tRNA on the ribosome.
      ], 6CZR [
      • Serrano C.M.
      • Kanna Reddy H.R.
      • Eiler D.
      • Koch M.
      • Tresco B.I.C.
      • Barrows L.R.
      • VanderLinden R.T.
      • Testa C.A.
      • Sebahar P.R.
      • Looper R.E.
      Unifying the aminohexopyranose- and peptidyl-nucleoside antibiotics: implications for antibiotic design.
      ], 4U56 [
      • Garreau De Loubresse N.
      • Prokhorova I.
      • Holtkamp W.
      • Rodnina M.V.
      • Yusupova G.
      • Yusupov M.
      Structural basis for the inhibition of the eukaryotic ribosome.
      ], 6B4V [
      • Svidritskiy E.
      • Korostelev A.A.
      Mechanism of inhibition of translation termination by blasticidin S.
      ], 1KC8 [
      • Hansen J.L.
      • Moore P.B.
      • Steitz T.A.
      Structures of five antibiotics bound at the peptidyl transferase center of the large ribosomal subunit.
      ], and 1XBP [
      • Schlünzen F.
      • Pyetan E.
      • Fucini P.
      • Yonath A.
      • Harms J.M.
      Inhibition of peptide bond formation by pleuromutilins: the structure of the 50S ribosomal subunit from Deinococcus radiodurans in complex with tiamulin.
      ]) and overlayed. From these overlayed structures, we selected a receptor grid box that was large enough (30 ​× ​30 ​× ​30 ​Å) to accommodate all of the parent antibiotic binding sites in order to remove the variable of constantly changing box sizes, center coordinates, and relevant ribonucleoprotein elements. We then removed the various co-crystallized antibiotics from their crystal structures and redocked them using standard methods (Tables S1–S9, Figures S1-S9, SI Computation Methods pages S3–S4). Further, interactions between redocked antibiotics and nearby nucleotides were used as an additional metric to determine the ability to replicate co-crystal antibiotic positioning, in addition to root-mean-square deviation (RMSD) calculations comparing atom coordinate overlap.
      For these redocked antibiotics, AutoDock Vina [
      • Trott O.
      • Olson A.J.
      Autodock Vina: improving the speed and accuracy of docking with a new scoring function, efficient optimization, and multithreading.
      ] showed better consistency and accuracy than Schrödinger Glide [
      • Friesner R.A.
      • Murphy R.B.
      • Repasky M.P.
      • Frye L.L.
      • Greenwood J.R.
      • Halgren T.A.
      • Sanschagrin P.C.
      • Mainz D.T.
      Extra precision Glide: docking and scoring incorporating a model of hydrophobic enclosure for Protein−Ligand complexes.
      ] using an analogous protocol based on the same center and docking coordinate space in terms of scoring interactions between the small molecule and nucleotides and by conformational ensemble quality. For example, redocking of tiamulin from 1XBP (Fig. 7A) using Schrödinger Glide [
      • Friesner R.A.
      • Murphy R.B.
      • Repasky M.P.
      • Frye L.L.
      • Greenwood J.R.
      • Halgren T.A.
      • Sanschagrin P.C.
      • Mainz D.T.
      Extra precision Glide: docking and scoring incorporating a model of hydrophobic enclosure for Protein−Ligand complexes.
      ] did not generate a minimum of nine ligand poses, while AutoDock Vina did so reliably (Fig. 7B). Further, those poses predicted by Schrödinger Glide were highly clustered into one area (Fig. 7C), missing key ribosome interactions, while AutoDock Vina sampled more diversity throughout the antibiotic binding cavity space. Quantitatively, Schrödinger Glide resulted in RMSD values ranging between 5.17 ​Å to 5.52 ​Å (Fig. 7D), a lack of sampling that indicated this docking program was not suitable for these systems. On the other hand, AutoDock Vina [
      • Trott O.
      • Olson A.J.
      Autodock Vina: improving the speed and accuracy of docking with a new scoring function, efficient optimization, and multithreading.
      ] consistently generated nine redocked poses with more diversity throughout the antibiotic binding cavity space as measured by RMSD values (Tables S1–S9). A portion of these poses were lower than 2 ​Å for most of PDB structures tested during protocol development. Thus, the increased number of redocked poses produced by the AutoDock Vina [
      • Trott O.
      • Olson A.J.
      Autodock Vina: improving the speed and accuracy of docking with a new scoring function, efficient optimization, and multithreading.
      ] protocol allowed for more sampling of the binding cavity, conferring a conformational ensemble with a wider range of RMSD values including some sub-2 Å results that were sufficient for use in our study.
      Fig. 7
      Fig. 7(A) Tiamulin in PDB ID: 1XBP (grey, colored by element) with individual nucleotides of the binding pocket that contact (<3.5 ​Å) the redocked tiamulin and derivatives. Adenosines are shown in red, cytidines shown in blue, guanosines shown in yellow, and uridines shown in green. Low interactive nucleotides are greyed out. (B) The nine poses of redocked tiamulin (purple lines) using AutoDock Vina overlayed with the native tiamulin ligand (grey sticks). (C) The eight poses of redocked tiamulin (gold lines) using Schrödinger Glide overlayed with the native tiamulin ligand (grey sticks). (D) RMSD values for the Autodock Vina and Schrödinger Glide poses shown in (B) and (C), respectively. Vina reliably furnished nine poses while Glide could only return a maximum of eight. While the average RMSD was approximately the same, the thorough sampling of the receptor grid box and broader range of RMSD values provided by Autodock Vina instilled confidence that it would provide better correlations. (E) Parameters for the two programs excerpted from the docking protocol (SI Computation Methods pages S3–S4). (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)
      Among the various antibiotic/ribosome complexes studied, the average RMSD values (obtained by averaging all nine poses produced by AutoDock Vina) ranged from 4.3 ​Å to 12.7 ​Å (Tables S1-S9, SI Computational Methods, pages S3–S4). These values are larger than traditionally observed for small molecule protein modeling [
      • Kufareva I.
      • Abagyan R.
      Methods of protein structure comparison.
      ], but are critical to providing a baseline for establishing a widely applicable protocol to successfully model these highly complex systems.

      2.4.2 Method validation for pleuromutilin-containing co-crystal structures

      Among the co-crystal structures used to establish our docking method, the Deinococcus radiodurans (PDB ID: 1XBP [
      • Schlünzen F.
      • Pyetan E.
      • Fucini P.
      • Yonath A.
      • Harms J.M.
      Inhibition of peptide bond formation by pleuromutilins: the structure of the 50S ribosomal subunit from Deinococcus radiodurans in complex with tiamulin.
      ]) and Staphylococcus aureus (PDB ID: 5HL7 [
      • Eyal Z.
      • Matzov D.
      • Krupkin M.
      • Paukner S.
      • Riedl R.
      • Rozenberg H.
      • Zimmerman E.
      • Bashan A.
      • Yonath A.
      A novel pleuromutilin antibacterial compound, its binding mode and selectivity mechanism.
      ]) 50S ribosomal subunit structures co-crystallized with pleuromutilin derivatives tiamulin and lefamulin, respectively, offered the best potential platforms for the triazole derivatives under study. Redocking using the S. aureus and D. radiodurans 50S ribosomal subunits centered on the same coordinates and using the established 30 ​× ​30 ​× ​30 ​Å box size (Fig. 7E) reliably produced binding poses with RMSDs of less than 2 ​Å (Tables S3 and S9). Lefamulin (Table S3, Fig. S3) was successfully redocked into the S. aureus 50S ribosomal subunit with an RMSD of 1.5 ​Å for the lowest energy pose, but there were three major sub-pocket areas sampled during the redocking, resulting in an average RMSD for all poses of 12.7 ​Å. The affinity of lefamulin for these sub-pockets dramatically lowered the frequency of shared base pair interactions that the redocked poses shared with the co-crystalized ligand, conserving only five of the crystal structure/lefamulin interactions at ≥70% frequency. Tiamulin (Table S9, Fig. S9) was successfully redocked into the D. radiodurans 50S ribosomal subunit with an RMSD of 1.8 ​Å for the lowest energy pose and an average RMSD of 5.7 ​Å for all poses. Redocking of tiamulin in the D. radiodurans PTC and lefamulin in the S. aureus PTC sampled the same volume of space, but the redocking of tiamulin produced a higher frequency of shared base pair interactions between redocked poses and the co-crystalized ligand in the crystal structure, conserving all nine base pair interactions vital for coordination of the co-crystalized ligand at ≥70% frequency. Thus, redocking procedures based on D. radiodurans using tiamulin were selected over S. aureus with lefamulin because of their lower average RMSD and the observed high degree of alignment with the authentic tiamulin ligand from the co-crystal structure. Further, compounds 115, which bear a variety of C20 and C22 substituents, could also be reliably docked to 1XBP (Fig. S10 and Table S10). These successes gave us confidence that docking triazole derivatives to the PTC would be feasible.

      2.4.3 Application of computational methods to triazole pleuromutilin derivatives

      C22-Functionalized triazole-pleuromutilin derivatives 1828 (Figs. S12–S39) and C20-functionalized derivatives 3141 (Figs. S40–S67) were docked using the tiamulin docking protocol (SI Computation Methods pages S3–S4) and parameters (Fig. 7E). Each derivative's predicted lowest free energy pose (LFE, Table 4, Table 3) and average free energy of all poses (AFE) were calculated using AutoDock Vina [
      • Trott O.
      • Olson A.J.
      Autodock Vina: improving the speed and accuracy of docking with a new scoring function, efficient optimization, and multithreading.
      ]. Qikprop [
      Maestro.
      ] was used to calculate ligand physicochemical properties including total heavy atoms, molecular weight, volume, logP of octanol/water, solvent accessible surface area, and human oral absorption (Tables S11–S12). The predicted ranking of the triazole-pleuromutilin derivatives were based on each compounds' ligand efficiency, defined as the free energy of binding divided by the number of non-hydrogen atoms in the ligand (heavy atom count, HAC). Ranks were assigned using the ligand efficiency for the lowest energy pose (LE-LP) in ascending order (most negative to least negative). Ranks from the ligand efficiency of all poses (LE-AP) gave a similar order. Top ranked derivatives showed the most consistent pose clustering and a high frequency of interaction with the nucleotides that contact tiamulin (Figs. S12–S39) while lower ranks displayed poor clustering and a lower frequency of interaction with the proximal nucleotides, results that support the modeling approach. The docked poses for all derivatives displayed a similar range of predicted free energies of binding (−10.7 ​kcal/mol to −9.1 ​kcal/mol). Overall, the C22 derivatives ranked better in predicted efficacy based on computational methods than the C20 derivatives.
      Table 3Docking results of the C22 triazoles.
      IDLFEAFEHACLE-LPLE-AP
      18−9.8−9.133−0.297−0.28
      19−10.6−9.734−0.312−0.29
      20−9.4−8.835−0.269−0.25
      21−9.6−8.933−0.291−0.27
      22−9.8−9.134−0.288−0.27
      23R

      23S
      −10.0

      −10.0
      −9.6

      −9.4
      34−0.294

      −0.294
      −0.28

      −0.28
      24−10.2−9.435−0.291−0.27
      25R

      25S
      −9.5

      −9.6
      −9.1

      −9.0
      36−0.264

      −0.267
      −0.25

      −0.25
      26R

      26S
      −9.5

      −9.1
      −9.3

      −8.7
      38−0.250

      −0.239
      −0.24

      −0.23
      27−9.4−9.137−0.254−0.25
      28−10.7−10.040−0.268−0.25
      ID – Compound Number; LFE – Vina Lowest Free Energy Pose (kcal·mol-1); AFE – Vina Free Energy, Average of All Poses (kcal·mol-1); HAC – Heavy Atom Count; LE-LP – Ligand Efficiency for Lowest Energy Pose; LE-AP – Ligand Efficiency from All Poses.
      Table 4Docking results of the C20 triazoles.
      IDLFEAFEHACLE-LPLE-AP
      31−9.3−8.734−0.274−0.26
      32−9.1−8.235−0.260−0.23
      33−9.4−8.636−0.261−0.24
      34−9.2−8.734−0.271−0.25
      35−9.1−8.035−0.260−0.23
      36R

      36S
      −9.5

      −9.5
      −8.9

      −8.8
      35−0.271

      −0.271
      −0.25

      −0.25
      37−9.6−8.936−0.267−0.25
      39R

      39S
      −9.4

      −9.3
      −8.7

      −8.7
      37−0.254

      −0.251
      −0.24

      −0.24
      40R

      40S
      −9.4

      −9.1
      −8.7

      −8.4
      39−0.241

      −0.233
      −0.22

      −0.21
      41−9.1−8.438−0.239−0.22
      38−10.6−9.841−0.259−0.24
      ID – Compound Number; LFE – Vina Lowest Free Energy Pose (kcal·mol-1); AFE – Vina Free Energy, Average of All Poses (kcal·mol-1); HAC – Heavy Atom Count; LE-LP – Ligand Efficiency for Lowest Energy Pose; LE-AP – Ligand Efficiency from All Poses.
      Considering the computational analysis of the C22 and C20 groups separately, substitutions of short carbon chains terminated by an electronegative group, such as 18, 21, 31, and 34 (Tables 3 and 4, Figures S12-S13, S18-S19, S40-S41, and S46-47), gave the most favorable predicted free energy of binding results. A main contributor to the more favorable binding modes of these compounds was their ability to interact favorably with G2044 (Figs. S12–S67), the nucleotide that engages the ester on the thioether chain of tiamulin. Interaction with the G2044 nucleotide is essential for targeting derivatives to bind within the ribosomal PTC, showing interactions with over 70% of all docked derivative poses from both the C22 and C20 groups (Figs. S12–S67). As the carbon chain-length off the triazole rings increased beyond an ethyl group, less favorable predicted free energies of binding and binding modes with poorer relative fitness to the cavity (e.g., less similar position to tiamulin) were generally observed. This connection between longer chain length and worse predicted binding rank is illustrated by the difference between 19 and 20. Compound 20, with one additional carbon atom, did not perform as well as 19 with respect to predicted free energy of binding (Table 3, LFE and AFE), ligand efficiency measurements (LE-LP and LE-AP), and nucleotide interactions (Figs. S15 and S17). This effect was accentuated in longer chain derivatives, such as 25, 26, 39, and 40. Compared to other long-chain derivatives, compounds 28 and 38 performed better, likely because of favorable pi-stacking interactions between the aromatic group and A2045 (Figs. S39 and S67). Polar functional groups (alcohols and ethers) positioned closer to the triazole mitigated some of the negative effects of length by mimicking the interaction of the hydroxyl groups of the best-performing derivatives, 18 and 19. Indeed, high (>70%) interaction levels between nucleotides A2430, C2431, U2483, and G2484 and the alcohols of derivatives 18 and 19 (as well as polar groups from other derivatives, Figs. S12–S17) support the conclusion that polarity at the α- and β-carbon of the triazole chain plays a crucial role in coordinating triazole-derived pleuromutilins to the ribosome.
      Comparing the predicted poses of C22 (Table 3) and C20 (Table 4) triazole derivatives with the same substituent from the other series showed that while they bound with similar energies, the C22 variants universally had lower predicted free energies of binding for the lowest energy pose and for the average energy of all poses. Energy minimizations of the C20 derivatives with Schrödinger Maestro [
      Maestro.
      ] independent of docking showed no intramolecular interactions that would preclude successful docking. Pattern analysis of the binding modes for the two series were correlated to each other and to the tiamulin ligand by using fingerprint graphing (Figs. S12–S67) of the docking simulations to quantify ligand-nucleotide interactions. The differences in interactions were visualized using PyMOL 2.5.0 [
      The PyMOL Molecular Graphics System.
      ] (Figs. S12–S67). Nucleotide interactions placed the triazole ring of all derivatives in the same area, that which is normally occupied by the sulfide-containing chain of tiamulin.

      2.5 Unifying computational analysis and in vitro activity

      To rationalize these energetic and orientational differences revealed by the docking studies, close inspection of the two series was undertaken using PyMOL 2.5.0 and revealed that each grouped with separate orientations of the triazole ring. As the C22 triazole derivatives are substituted directly onto the C22 position, they and their pendant chains maximize contacts normally made by the thioether chain while the mutilin core is easily accommodated within its binding pocket (Fig. 8A, Figs. S12–S39). In contrast, when the triazole ring derivatives are attached to the C20 position, they must rotate and bend inward to effectively contact the nucleotides that usually interact with the thioether substituent of tiamulin (Fig. 8B, Figs. S40–S67). Adding to the binding inadequacies the C20 triazole series, these derivatives still contain a C22 hydroxyl group; electron repulsion between it and the C20 triazole prevent either from assuming as favorable of a contact position. The resulting shift in orientation of the triazole in the C20 series also diminishes the contacts of the mutilin core to its binding pocket relative to those of the C22 series. Thus, we hypothesize that the C20 series cannot reach as far into the PTC while retaining the vital mutilin core interactions needed for effective inhibitory activity, while the C22-triazole derivatives can maintain these important contacts.
      Fig. 8
      Fig. 8Comparison of docking poses and rank orders for the C20 and C22 pleuromutilin triazole derivatives. (A) The best poses of the highest (brown) and lowest (yellow) ranked C22 derivatives docked to the ribosome. Tiamulin (grey) is shown for reference. The triazoles make the same contacts as the tiamulin sulfide when not perturbing core mutilin contacts. (B) The best poses of the highest (light blue) and lowest (yellow) ranked C20 derivatives docked to the ribosome. Tiamulin (grey) is shown for reference. The triazole still makes contact with the same residues as the tiamulin sulfide, but must push the glycolic acid residue out of the way and perturb the mutilin binding position to do so. (C) Scatterplot comparing average MIC values (Table 1, Table 2) to ligand efficiency (LE-LP, ) for C22 derivatives (blue circles) and C20 derivatives (red diamonds). The activity of the C22 derivatives tracks well with the docking results while the lack of activity among the C20 derivatives prevents assessment of the docking fidelity. (D) Comparison overlay of C22 derivative 19 (brown) and C20 derivative 32 (light blue) highlighting the perturbation of the mutilin core (left) and overlay of the triazoles (right) with the site usually occupied by the tiamulin thioether. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)
      To compare the computational analysis to the MIC assay results, the averages of the MIC values for each triazole derivative were compared to the independently derived computational ranks of the C20 and C22 subgroups. A limitation of this computational/MIC comparison is that we are assuming there is no difference in permeability or efflux of the derivatives, which range in cLogP from −0.8 to 2.5 and solvent exposed surface area from 695 to 942 ​Å2 (Tables S11 and S12). By averaging the MIC values among the various species tested, we sought to minimize organism-specific effects that would interfere with MIC/computational comparisons. Overall, high agreement between the MIC values and computational results was achieved by considering ligand efficiency for the active C22-substituted derivatives (Fig. 8C, blue circles). For the inactive C20 subgroup, distinguishing the ranking agreement was hampered by low variance in the MIC assay results (Fig. 8C, red diamonds). However, alignment for the general activity predictions comparing one series to the other by computational modeling and MIC assays were in accord: The C22 triazole series were more active than the C20 triazole series. The triazole group was tolerated as a viable substitution off the C22 position with shorter chains and polar functionality being more active. The C20 triazoles series lost activity compared to pleuromutilin likely due to the inability to maintain sufficient contacts of the mutilin core and the side chain simultaneously (Fig. 8D).

      3. Conclusions

      We synthesized libraries of pleuromutilin derivatives containing triazoles at the C20 and C22 positions. To activate the vinyl as an azide, recently published hydroazidation techniques were used [
      • Li X.
      • Chen P.
      • Liu G.
      Iodine(III) reagent (ABX—N3)-induced intermolecular anti-Markovnikov hydroazidation of unactivated alkenes.
      ,
      • Li H.
      • Shen S.-J.
      • Zhu C.-L.
      • Xu H.
      Direct intermolecular anti-Markovnikov hydroazidation of unactivated olefins.
      ] showing that a direct anti-Markovnikov hydroazidation can be performed on an unprotected, complex natural product. Subsequent CuAAC reactions on this C20 azido (30) and a traditionally prepared C22 azido intermediate (17) resulted in libraries of pleuromutilin-triazole derivatives that were tested against both Gram-positive and -negative strains, demonstrating that triazoles are well tolerated as a C22 sidechain modification and a viable pathway to rapidly create antibiotic derivative libraries. The underexplored C20 triazole derivatives did not show activity against the tested organisms, but contribute additional data to an ongoing debate about whether this site is suitable for functionalization. Some reports suggest that pairing C20 heterocycles with more complex C22 sidechains produces active compounds [] while others indicate that chain extension in place of the C20 vinyl position is deleterious to activity [
      • Thirring K.
      • Heilmayer W.
      • Riedl R.
      • Kollmann H.
      • Ivezic-Schoenfeld Z.
      • Wicha W.
      • Paukner S.
      • Strickmann D.
      Preparation of 12-epi-pleuromutilin derivatives as antimicrobial agents.
      ].
      To gain further insight to the binding patterns responsible for the observed activity and to establish protocols for docking and developing antibiotic derivatives generally, we completed an unbiased computational docking study with of our libraries of compounds based on the tiamulin/D. radiodurans co-crystal ribosome structure (PDB ID: 1XBP [
      • Schlünzen F.
      • Pyetan E.
      • Fucini P.
      • Yonath A.
      • Harms J.M.
      Inhibition of peptide bond formation by pleuromutilins: the structure of the 50S ribosomal subunit from Deinococcus radiodurans in complex with tiamulin.
      ]). The redocked ligand had a low RMSD and all derivatives were tightly clustered, indicating a high model validity. The computational ranks predicting derivative effectiveness based on in silico ligand efficiency measurements largely agreed with the ranks based on average minimum inhibitory concentration for the C22 derivatives. The poor activity of the C20 derivatives in the MIC assays prevented their differentiation, thus precluding a meaningful correlation to the computational results for this series. However, computational comparison between the C20 series and C22 series indicated that the latter should be more active, a result which was confirmed by the activity assay. Discrepancies in ranks could also be due to issues beyond the computationally assessed interactions, such as cellular penetration, an issue that will be addressed for future derivatives by using a cell-free transcription/translation assay to independently assess binding alongside the current MIC assay that determines actual efficacy. Overall, the C22 triazole-pleuromutilin derivates performed better in the activity assays and scored better computationally than the C20 derivatives, a result which we rationalize is caused by the inability of the latter to successfully bind to the underlying nucleotides with their substituted chain and mutilin core simultaneously.
      This combined synthetic/assay/modeling approach serves as a basis for the continued development of pleuromutilin triazole derivatives and the guided evolution of ribosome-targeting antibiotics generally. The computational work establishes a critical starting point for the application of computational methods for analysis of antibiotic-ribosome binding, work that will help in unraveling the activity of functionalized C20 vinyl pleuromutilin derivatives, both with and without C12 epimerization. Further exploration of these derivatives is needed to understand the possibilities of substitution at the vinyl substituent, with or without C22 substitution. Future work will continue to use a combined computational/medicinal chemistry approach to guide development of additional potent pleuromutilin derivatives and for the development of other classes of antibiotics.

      Data availability

      Our PDBQT files and configuration files are available on the Brown Lab Open Science Framework page for community use. Please visit https://osf.io/82n73/

      Declaration of competing interest

      The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

      Acknowledgements

      We thank the support of the National Science Foundation under 1726077 for crystallography experiments. LMB thanks the Center of Emerging, Zoonotic, and Arthropod-Borne Pathogens at Virginia Polytechnic Institute for the ID IGEP Fellowship in 2021. We also thank the Mevers group for use of their polarimeter. We would also like to thank C. Luo and the Reviewers for their time and helpful suggestions during revision.

      Appendix A. Supplementary data

      The following are the Supplementary data to this article:

      References

        • Antimicrobial Resistance Collaborators
        Global burden of bacterial antimicrobial resistance in 2019: a systematic analysis.
        Lancet. 2022; 399: 629-655https://doi.org/10.1016/s0140-6736(21)02724-0
      1. No Time to Wait: Securing the Future from Drug-Resistant Infections. Interagency Coordination Group on Antimicrobial Resistance, 2019
        • Ventola C.L.
        The antibiotic resistance crisis: part 1: causes and threats.
        P T. 2015; 40: 277-283
        • Newman D.J.
        • Cragg G.M.
        Natural products as sources of new drugs over the nearly four decades from 01/1981 to 09/2019.
        J. Nat. Prod. 2020; 83: 770-803https://doi.org/10.1021/acs.jnatprod.9b01285
        • Kavanagh F.
        • Hervey A.
        • Robbins W.J.
        Antibiotic substances from basidiomycetes: VIII. Pleurotus multilus (Fr.) sacc. And Pleurotus passeckerianus pilat.
        Proc. Natl. Acad. Sci. U.S.A. 1951; 37: 570https://doi.org/10.1073/pnas.37.9.570
        • Egger H.
        • Reinshagen H.
        New pleuromutilin derivatives with enhanced antimicrobial activity. II. Structure-activity correlations.
        J. Antibiot. 1976; 29: 915-922https://doi.org/10.7164/antibiotics.29.923
        • Schlünzen F.
        • Pyetan E.
        • Fucini P.
        • Yonath A.
        • Harms J.M.
        Inhibition of peptide bond formation by pleuromutilins: the structure of the 50S ribosomal subunit from Deinococcus radiodurans in complex with tiamulin.
        Mol. Microbiol. 2004; 54: 1287-1294https://doi.org/10.1111/j.1365-2958.2004.04346.x
        • Davidovich C.
        • Bashan A.
        • Auerbach-Nevo T.
        • Yaggie R.D.
        • Gontarek R.R.
        • Yonath A.
        Induced-fit tightens pleuromutilins binding to ribosomes and remote interactions enable their selectivity.
        Proc. Natl. Acad. Sci. U.S.A. 2007; 104: 4291-4296https://doi.org/10.1073/pnas.0700041104
        • Goethe O.
        • Heuer A.
        • Ma X.
        • Wang Z.
        • Herzon S.B.
        Antibacterial properties and clinical potential of pleuromutilins.
        Nat. Prod. Rep. 2019; 36: 220-247https://doi.org/10.1039/c8np00042e
        • Fazakerley N.J.
        • Procter D.J.
        Synthesis and synthetic chemistry of pleuromutilin.
        Tetrahedron. 2014; 70: 6911-6930https://doi.org/10.1016/j.tet.2014.05.092
        • Prince W.
        • Obermayr F.
        • Ivezic-Schoenfeld Z.
        • Lell C.
        • Wicha W.
        • Strickmann D.
        • Tack K.
        • Novak R.
        In A Phase 2 Study Comparing the Safety and Efficacy of Two Doses of BC-3781 versus Vancomycin in Acute Bacterial Skin and Skin Structure Infections (ABSSSI), Abstr. L-966. Abstr. 51st Intersci, Program and Abstracts of the 51st Interscience Conference on Antimicrobial Agents and Chemotherapy (Chicago).
        American Society for Microbiology, Washington, DC2011
        • Rittenhouse S.
        • Biswas S.
        • Broskey J.
        • McCloskey L.
        • Moore T.
        • Vasey S.
        • West J.
        • Zalacain M.
        • Zonis R.
        • Payne D.
        Selection of retapamulin, a novel pleuromutilin for topical use.
        Antimicrob. Agents Chemother. 2006; 50: 3882-3885https://doi.org/10.1128/AAC.00178-06
        • Czok R.
        • Meingassner J.G.
        • Mieth H.
        • Schutze E.
        Antibiotic compositions for treating coccidiosis.
        US patent. 1979; 4: 890
        • Burch D.G.S.
        • Ripley P.H.
        • Zeisl E.
        Veterinary use of a pleuromutilin derivative.
        US Patent. 2000; 6: 250
        • Jacobs M.R.
        Retapamulin: a semisynthetic pleuromutilin compound for topical treatment of skin infections in adults and children.
        Future Microbiol. 2007; 2: 591-600https://doi.org/10.2217/17460913.2.6.591
        • Hunt A.
        FDA Approves New Antibiotic to Treat Community-Acquired Bacterial Pneumonia.
        2019
        • Paukner S.
        • Riedl R.
        Pleuromutilins: potent drugs for resistant bugs—mode of action and resistance.
        Cold Spring Harb. Perspect. Med. 2017; 7: a027110https://doi.org/10.1101/cshperspect.a027110
        • Yan K.
        • Madden L.
        • Choudhry A.E.
        • Voigt C.S.
        • Copeland R.A.
        • Gontarek R.R.
        Biochemical characterization of the interactions of the novel pleuromutilin derivative retapamulin with bacterial ribosomes.
        Antimicrob. Agents Chemother. 2006; 50: 3875-3881https://doi.org/10.1128/AAC.00184-06
        • Gentry D.R.
        • Rittenhouse S.F.
        • McCloskey L.
        • Holmes D.J.
        Stepwise exposure of Staphylococcus aureus to pleuromutilins is associated with stepwise acquisition of mutations in rplC and minimally affects susceptibility to retapamulin.
        Antimicrob. Agents Chemother. 2007; 51: 2048-2052https://doi.org/10.1128/aac.01066-06
        • Bozorov K.
        • Zhao J.
        • Aisa H.A.
        1,2,3-Triazole-containing hybrids as leads in medicinal chemistry: a recent overview.
        Biorg. Med. Chem. 2019; 27: 3511-3531https://doi.org/10.1016/j.bmc.2019.07.005
        • Jain A.
        • Piplani P.
        Exploring the chemistry and therapeutic potential of triazoles: a comprehensive literature review.
        Mini-Rev. Med. Chem. 2019; 19: 1298-1368https://doi.org/10.2174/1389557519666190312162601
        • Marzi M.
        • Farjam M.
        • Kazeminejad Z.
        • Shiroudi A.
        • Kouhpayeh A.
        • Zarenezhad E.
        A recent overview of 1,2,3-triazole-containing hybrids as novel antifungal agents: focusing on synthesis, mechanism of action, and structure-activity relationship (SAR).
        J. Chem. 2022; 2022: 1-50https://doi.org/10.1155/2022/7884316
        • Lolk L.
        • Pøhlsgaard J.
        • Jepsen A.S.
        • Hansen L.H.
        • Nielsen H.
        • Steffansen S.I.
        • Sparving L.
        • Nielsen A.B.
        • Vester B.
        • Nielsen P.
        A click chemistry approach to pleuromutilin conjugates with nucleosides or acyclic nucleoside derivatives and their binding to the bacterial ribosome.
        J. Med. Chem. 2008; 51: 4957-4967https://doi.org/10.1021/jm800261u
        • Dreier I.
        • Kumar S.
        • Søndergaard H.
        • Rasmussen M.L.
        • Hansen L.H.
        • List N.H.
        • Kongsted J.
        • Vester B.
        • Nielsen P.
        A click chemistry approach to pleuromutilin derivatives, part 2: conjugates with acyclic nucleosides and their ribosomal binding and antibacterial activity.
        J. Med. Chem. 2012; 55: 2067-2077https://doi.org/10.1021/jm201266b
        • Dreier I.
        • Hansen L.H.
        • Nielsen P.
        • Vester B.
        A click chemistry approach to pleuromutilin derivatives. Part 3: extended footprinting analysis and excellent MRSA inhibition for a derivative with an adenine phenyl side chain.
        Bioorg. Med. Chem. Lett. 2014; 24: 1043-1046https://doi.org/10.1016/j.bmcl.2014.01.019
        • Heidtmann C.V.
        • Voukia F.
        • Hansen L.N.
        • Soerensen S.H.
        • Urlund B.
        • Nielsen S.
        • Pedersen M.
        • Kelawi N.
        • Andersen B.N.
        • Pedersen M.
        • Reinholdt P.
        • Kongsted J.
        • Nielsen C.U.
        • Klitgaard J.K.
        • Nielsen P.
        Discovery of a potent adenine-benzyltriazolo-pleuromutilin conjugate with pronounced antibacterial activity against MRSA.
        J. Med. Chem. 2020; 63: 15693-15708https://doi.org/10.1021/acs.jmedchem.0c01328
        • Zhang Z.
        • Li K.
        • Zhang G.-Y.
        • Tang Y.-Z.
        • Jin Z.
        Design, synthesis and biological activities of novel pleuromutilin derivatives with a substituted triazole moiety as potent antibacterial agents.
        Eur. J. Med. Chem. 2020; 204112604https://doi.org/10.1016/j.ejmech.2020.112604
        • Zhang G.-Y.
        • Zhang Z.
        • Li K.
        • Liu J.
        • Li B.
        • Jin Z.
        • Liu Y.-H.
        • Tang Y.-Z.
        Design, synthesis and biological evaluation of novel pleuromutilin derivatives containing piperazine and 1,2,3-triazole linker.
        Bioorg. Chem. 2020; 105104398https://doi.org/10.1016/j.bioorg.2020.104398
        • Stresser D.M.
        • Broudy M.I.
        • Ho T.
        • Cargill C.E.
        • Blanchard A.P.
        • Sharma R.
        • Dandeneau A.A.
        • Goodwin J.J.
        • Turner S.D.
        • Erve J.C.L.
        • Patten C.J.
        • Dehal S.S.
        • Crespi C.L.
        Highly selective inhibition of human CYP3A in vitro by azamulin and evidence that inhibition is irreversible.
        Drug Metab. Dispos. 2004; 32: 105-112https://doi.org/10.1124/dmd.32.1.105
        • Thirring K.
        • Heilmayer W.
        • Riedl R.
        • Kollmann H.
        • Ivezic-Schoenfeld Z.
        • Wicha W.
        • Paukner S.
        • Strickmann D.
        Preparation of 12-epi-pleuromutilin derivatives as antimicrobial agents.
        WO2015110481A1, 2015
        • Xianfeng L.
        • Lunde C.S.
        • Jacobs R.T.
        • Yasheen Z.
        Boron-containing small molecules.
        2017 (WO2017151492A1)
        • Himo F.
        • Lovell T.
        • Hilgraf R.
        • Rostovtsev V.V.
        • Noodleman L.
        • Sharpless K.B.
        • Fokin V.V.
        Copper (I)-catalyzed synthesis of azoles. DFT study predicts unprecedented reactivity and intermediates.
        J. Am. Chem. Soc. 2005; 127: 210-216https://doi.org/10.1021/ja0471525
        • Riedl K.
        Studies on pleuromutilin and some of its derivatives.
        J. Antibiot. 1976; 29: 132-139https://doi.org/10.7164/antibiotics.29.132
        • Chappe B.
        • Musikas H.
        • Marie D.
        • Ourisson G.
        Synthesis of three acyclic all-trans-tetraterpene diols, putative precursors of bacterial lipids.
        Bull. Chem. Soc. Jpn. 1988; 61: 141-148https://doi.org/10.1246/bcsj.61.141
        • Ding R.
        • He Y.
        • Wang X.
        • Xu J.
        • Chen Y.
        • Feng M.
        • Qi C.
        Treatment of alcohols with tosyl chloride does not always lead to the formation of tosylates.
        Molecules. 2011; 16: 5665-5673https://doi.org/10.3390/molecules16075665
        • Osorio L.S.
        • Ionta M.
        • Demuner A.J.
        • Sousa B.L.d.
        • Ferraz G.O.
        • Varejão E.V.
        • Ferreira-Silva G.A.
        • Pilau E.J.
        • Silva E.
        • Santos M.H.d.
        Synthesis of 1,2,3-triazole derivatives of hydnocarpic acid isolated from Carpotroche brasiliensis seed oil and evaluation of antiproliferative activity.
        J. Braz. Chem. Soc. 2020; 31: 2500-2510https://doi.org/10.21577/0103-5053.20200125
        • Li X.
        • Chen P.
        • Liu G.
        Iodine(III) reagent (ABX—N3)-induced intermolecular anti-Markovnikov hydroazidation of unactivated alkenes.
        Sci. China Chem. 2019; 62: 1537-1541https://doi.org/10.1007/s11426-019-9628-9
        • Li H.
        • Shen S.-J.
        • Zhu C.-L.
        • Xu H.
        Direct intermolecular anti-Markovnikov hydroazidation of unactivated olefins.
        J. Am. Chem. Soc. 2019; 141: 9415-9421https://doi.org/10.1021/jacs.9b04381
        • Lengerli D.
        • Ibis K.
        • Nural Y.
        • Banoglu E.
        The 1,2,3-triazole “all-in-one” ring system in drug discovery: a good bioisostere, a good pharmacophore, a good linker, and a versatile synthetic tool.
        Expet Opin. Drug Discov. 2022; (In press)https://doi.org/10.1080/17460441.2022.2129613
        • Treitler D.S.
        • Leung S.
        How dangerous is too dangerous? A perspective on azide chemistry.
        J. Org. Chem. 2022; 87: 11293-11295https://doi.org/10.1021/acs.joc.2c01402
        • Hassner A.
        • Stern M.
        • Gottlieb H.E.
        • Frolow F.
        Synthetic methods. 33. Utility of a polymeric azide reagent in the formation of di- and triazidomethane. Their NMR spectra and the x-ray structure of derived triazoles.
        J. Org. Chem. 1990; 55: 2304-2306https://doi.org/10.1021/jo00295a014
        • Patel D.
        • Kosmidis C.
        • Seo Susan M.
        • Kaatz Glenn W.
        Ethidium bromide MIC screening for enhanced efflux pump gene expression or efflux activity in Staphylococcus aureus.
        Antimicrob. Agents Chemother. 2010; 54: 5070-5073https://doi.org/10.1128/AAC.01058-10
        • Zgurskaya H.I.
        • Krishnamoorthy G.
        • Ntreh A.
        • Lu S.
        Mechanism and function of the outer membrane channel TolC in multidrug resistance and physiology of enterobacteria.
        Front. Microbiol. 2011; 2: 189https://doi.org/10.3389/fmicb.2011.00189
        • Gordon N.C.
        • Wareham D.W.
        Multidrug-resistant Acinetobacter baumannii: mechanisms of virulence and resistance.
        Int. J. Antimicrob. Agents. 2010; 35: 219-226https://doi.org/10.1016/j.ijantimicag.2009.10.024
        • König G.
        • Sokkar P.
        • Pryk N.
        • Heinrich S.
        • Möller D.
        • Cimicata G.
        • Matzov D.
        • Dietze P.
        • Thiel W.
        • Bashan A.
        • Bandow Julia E.
        • Zuegg J.
        • Yonath A.
        • Schulz F.
        • Sanchez-Garcia E.
        Rational prioritization strategy allows the design of macrolide derivatives that overcome antibiotic resistance.
        Proc. Natl. Acad. Sci. U.S.A. 2021; 118e2113632118https://doi.org/10.1073/pnas.2113632118
        • Tessaro F.
        • Scapozza L.
        How ‘protein-docking’ translates into the new emerging field of docking small molecules to nucleic acids?.
        Molecules. 2020; 25https://doi.org/10.3390/molecules25122749
        • Trott O.
        • Olson A.J.
        Autodock Vina: improving the speed and accuracy of docking with a new scoring function, efficient optimization, and multithreading.
        J. Comput. Chem. 2010; 31: 455-461https://doi.org/10.1002/jcc.21334
        • Friesner R.A.
        • Murphy R.B.
        • Repasky M.P.
        • Frye L.L.
        • Greenwood J.R.
        • Halgren T.A.
        • Sanschagrin P.C.
        • Mainz D.T.
        Extra precision Glide: docking and scoring incorporating a model of hydrophobic enclosure for Protein−Ligand complexes.
        J. Med. Chem. 2006; 49: 6177-6196https://doi.org/10.1021/jm051256o
        • Ruiz-Carmona S.
        • Alvarez-Garcia D.
        • Foloppe N.
        • Garmendia-Doval A.B.
        • Juhos S.
        • Schmidtke P.
        • Barril X.
        • Hubbard R.E.
        • Morley S.D.
        rDock: a fast, versatile and open source program for docking ligands to proteins and nucleic acids.
        PLoS Comput. Biol. 2014; 10e1003571https://doi.org/10.1371/journal.pcbi.1003571
        • Kufareva I.
        • Abagyan R.
        Methods of protein structure comparison.
        Methods Mol. Biol. 2012; 857: 231-257https://doi.org/10.1007/978-1-61779-588-6_10
        • Carter A.P.
        • Clemons W.M.
        • Brodersen D.E.
        • Morgan-Warren R.J.
        • Wimberly B.T.
        • Ramakrishnan V.
        Functional insights from the structure of the 30S ribosomal subunit and its interactions with antibiotics.
        Nature. 2000; 407: 340-348https://doi.org/10.1038/35030019
        • Borovinskaya M.A.
        • Shoji S.
        • Holton J.M.
        • Fredrick K.
        • Cate J.H.D.
        A steric block in translation caused by the antibiotic spectinomycin.
        ACS Chem. Biol. 2007; 2: 545-552https://doi.org/10.1021/cb700100n
        • Eyal Z.
        • Matzov D.
        • Krupkin M.
        • Paukner S.
        • Riedl R.
        • Rozenberg H.
        • Zimmerman E.
        • Bashan A.
        • Yonath A.
        A novel pleuromutilin antibacterial compound, its binding mode and selectivity mechanism.
        Sci. Re. 2016; 639004https://doi.org/10.1038/srep39004
        • Svidritskiy E.
        • Ling C.
        • Ermolenko D.N.
        • Korostelev A.A.
        Blasticidin S inhibits translation by trapping deformed tRNA on the ribosome.
        Proc. Natl. Acad. Sci. U.S.A. 2013; 110: 12283-12288https://doi.org/10.1073/pnas.1304922110
        • Serrano C.M.
        • Kanna Reddy H.R.
        • Eiler D.
        • Koch M.
        • Tresco B.I.C.
        • Barrows L.R.
        • VanderLinden R.T.
        • Testa C.A.
        • Sebahar P.R.
        • Looper R.E.
        Unifying the aminohexopyranose- and peptidyl-nucleoside antibiotics: implications for antibiotic design.
        Angew. Chem. Int. Ed. 2020; 59: 11330-11333https://doi.org/10.1002/anie.202003094
        • Garreau De Loubresse N.
        • Prokhorova I.
        • Holtkamp W.
        • Rodnina M.V.
        • Yusupova G.
        • Yusupov M.
        Structural basis for the inhibition of the eukaryotic ribosome.
        Nature. 2014; 513: 517-522https://doi.org/10.1038/nature13737
        • Svidritskiy E.
        • Korostelev A.A.
        Mechanism of inhibition of translation termination by blasticidin S.
        J. Mol. Biol. 2018; 430: 591-593https://doi.org/10.1016/j.jmb.2018.01.007
        • Hansen J.L.
        • Moore P.B.
        • Steitz T.A.
        Structures of five antibiotics bound at the peptidyl transferase center of the large ribosomal subunit.
        J. Mol. Biol. 2003; 330: 1061-1075https://doi.org/10.1016/S0022-2836(03)00668-5
      2. Maestro.
        Schrödinger, LLC, New York, NY2021
      3. The PyMOL Molecular Graphics System.
        Schrödinger, LLC, New York, NY2021
        https://pymol.org/2/
        Version: Version 8