Anti-biofilm Agents against Pseudomonas aeruginosa: A Structure-Activity Relationship Study of C-Glycosidic LecB Inhibitors

Anti-biofilm Agents against Pseudomonas aeruginosa: A Structure− Activity Relationship Study of C‑Glycosidic LecB Inhibitors Roman Sommer, †,‡ Katharina Rox, ‡,§ Stefanie Wagner, †,‡ Dirk Hauck, †,‡ Sarah S. Henrikus, †,‡,# Shelby Newsad, †,‡ Tatjana Arnold, ‡,§ Thomas Ryckmans, ∥ Mark Bronstrup, ‡,§ Anne Imberty, ⊥ Annabelle Varrot, ⊥ Rolf W. Hartmann, ‡,#,¶ and Alexander Titz* ,†,‡,# † Chemical Biology of Carbohydrates, Helmholtz Institute for Pharmaceutical Research Saarland (HIPS), Helmholtz Centre for Infection Research, D-66123 Saarbrucken, Germany ‡ Deutsches Zentrum fur Infektionsforschung (DZIF), Standort Hannover, D-38124 Braunschweig, Germany § Chemical Biology, Helmholtz Centre for Infection Research, D-38124 Braunschweig, Germany ∥ Roche Pharmaceutical Research and Early Development, Roche Innovation Center Basel, CH-4070 Basel, Switzerland ⊥ Univ. Grenoble Alpes, CNRS, CERMAV, F-38000 Grenoble, France # Department of Pharmacy, Saarland University, D-66123 Saarbrucken, Germany ¶ Drug Design and Development, Helmholtz Institute for Pharmaceutical Research Saarland (HIPS), Helmholtz Centre for Infection Research, D-66123 Saarbrucken, Germany * SSupporting Information ABSTRACT: Biofilm formation is a key mechanism of antimicrobial resistance. We have recently reported two classes of orally bioavailable C-glycosidic inhibitors of the Pseudomonas aeruginosa lectin LecB with antibiofilm activity. They proved efficient in target binding, were metabolically stable, nontoxic, selective, and potent in inhibiting formation of bacterial biofilm. Here, we designed and synthesized six new carboxamides and 24 new sulfonamides for a detailed structure−activity relationship for two clinically representative LecB variants. Sulfonamides generally showed higher inhibition compared to carboxamides, which was rationalized based on crystal structure analyses. Substitutions at the thiophenesulfonamide increased binding through extensive contacts with a lipophilic protein patch. These metabolically stable compounds showed a further increase in potency toward the target and in biofilm inhibition assays. In general, we established the structure−activity relationship for these promising antibiofilm agents and showed that modification of the sulfonamide residue bears future optimization potential. ■INTRODUCTION Pseudomonas aeruginosa is an opportunistic Gram-negative bacterium with high clinical importance and classified as a critical priority 1 pathogen by the WHO in 2017. 1−4 Especially for cystic fibrosis (CF) patients, chronic infections result in recurrent pneumonia, sepsis, and lung damage. 5 Challenges in treating P. aeruginosa infections result from its intrinsic antimicrobial resistance and acquired resistances that often lead to multidrug-resistant MDR or XDR strains. 6 In addition, the bacterium’s antimicrobial tolerance is further enhanced by the self-formation of biofilms, a protective enclosure against host immune defense and antibiotic treatment. 7,8 Because bacteria residing in a biofilm are up to 1000-fold more resistant toward antibiotics, 7 targeting biofilm formation has been an emerging therapeutic approach in recent years to overcome the resistance problem (reviewed in refs 9−11). The two virulence factors LecA 12 and LecB 13 (initially called PA-IL and PA-IIL 14 ) are regulated by quorum sensing 15 and have decisive roles in biofilm formation. It is currently anticipated that both tetravalent carbohydrate-binding proteins cross-link glycoconjugates on host cells or tissue with bacterial lipopolysaccharide and exopolysaccharides to stabilize the matrix and integrity of the biofilm. 9,16 Thus, blocking this process with exogenous compounds could prevent the formation or even destroy established biofilms. Received: July 11, 2019 Published: September 25, 2019 Article pubs.acs.org/jmc Cite This: J. Med. Chem. 2019, 62, 9201−9216 © 2019 American Chemical Society 9201 DOI: 10.1021/acs.jmedchem.9b01120 J. Med. Chem. 2019, 62, 9201−9216 Downloaded via HELMHOLTZ CTR FOR INFECTION RSRCH on October 28, 2019 at 14:19:36 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles. Both carbohydrate-binding proteins, so-called lectins, were first isolated from the clinical isolate P. aeruginosa PAO1 by Gilboa-Garber et al. 14,17,18 Onehurdlefortherapeutic intervention is the fact that P. aeruginosa has a high genomic diversity among different isolates. 19−21 The protein sequence of LecA is rather conserved among P. aeruginosa strains but for LecB isolates are grouped into either PAO1or PA14-like LecB protein sequence families. 22,23 Despite these sequence variations in the two LecB variants, they surprisingly have a conserved binding specificity for similar glycoconjugates, which paves the way for the simultaneous targeting of a broad range of clinical isolates with one single compound. In its quarternary structure, LecB forms noncovalent homotetramers where two Ca2+-ions are present in each monomer, 23,24 mediating the recognition of its carbohydrate ligands, L-fucose and Dmannose (e.g., methyl α-D-mannoside (1), Figure 1). Because LecB is localized extracellularly, 13 the Gram-negative bacterial cell envelope which usually imposes a stringent hurdle for many antibiotics with intracellular targets, is not problematic for targeting LecB. Besides its role in biofilm formation and bacterial adhesion, LecB was also shown to carbohydrate-dependently block human ciliary beating, 25 interfere with tissue repair processes 26 and, recently, activate B-cells. 27 Furthermore, a direct involvement of LecA and LecB in infection and host colonization by P. aeruginosa using a murine infection model revealed the suitability of both lectins as therapeutic targets. 28,29 Inhalation of an aerosol of fucose and galactose, the ligands of LecB and LecA, resulted in a reduction of bacterial load in human infected airways 30,31 and in mice. 32 Because fucosides display higher affinities for LecB than mannosides, inhibitor development generally centered around fucose-based inhibitors presented on a multivalent scaffold to further increase affinity/avidity. 33,34 Following this strategy, multivalent glycopeptide dendrimers have been developed which efficiently inhibit the formation and disperse established biofilms of P. aeruginosa. 35 Another example of multivalent fucosides on a calixarene scaffoldshowedverypotent nanomolar affinities for LecB but surprisingly required millimolar concentrations (5 mM) for biofilm inhibition and, in contrast to the desired properties, the compound induced bacterial aggregation. 28 Therefore, multivalent presentation of carbohydrates could mimic bacterial exopolysaccharides and, thus, stabilize the biofilm rather than inducing its desired disintegration. In addition, these multivalent presentations of native carbohydrates may be immunogenic and interfere with the patient’s immune system, e.g., by binding to the various innate immunity pattern recognition receptors. To overcome these disadvantages intrinsic to multivalently displayed lectin ligands, we have embarked on the development of monovalent glycomimetic small molecules as competitive inhibitors of LecB. 36−41 Our small molecules possess drug-like properties that resulted in oral bioavailability of the two tested C-glycosides with systemic distribution, which is impossible for the high molecular weight multivalent compounds. We have started with the weak LecB ligand methyl α-Dmannoside (1) and transformed it into C-6 modified amide and sulfonamide derivatives that led to an increase in potency up to a factor of 20 (e.g., 2and 3,Figure 1). 36,38 These compounds showed good receptor binding kinetics and proved efficient in the prevention of bacterial adhesion. Because fucose and mannose are recognized by LecB, we merged the necessary functional groups that were shown to elicit attractive interactions with the protein into one molecule and obtained the first of our C-glycosides lacking the O-glycosidic linkage. 37 After the exploration of additional interactions of heptose39 orfluoroglycomimetics 40 with LecB, we then combined the initial C-glycosides with the amidoand sulfonamido substituents to obtain the glycomimetics 4−7. 41 Especially the sulfonamides 6and 7displayed favorable profiles in target binding potency and selectivity, ADME/Tox parameters and Figure 1. Design approach for C-glycosidic LecB inhibitors and extension of the structural space for extended SAR studies of compounds 4−37. Derivatives of methyl α-D-mannoside 1−3and their inhibitory potency for the binding with LecBPAO1. 36 C-Glycosides simultaneously derived of Dmannosides and L-fucosides are hybrid-type LecB ligands 4and 5and 6and 7. 41 Journal of Medicinal Chemistry Article DOI: 10.1021/acs.jmedchem.9b01120 J. Med. Chem. 2019, 62, 9201−9216 9202 oral bioavailability in a murine pharmacokinetics model. Importantly, 6and 7possessed excellent antibiofilm activity in a P. aeruginosa biofilm formation assay as monitored by confocal fluorescence light microscopy. Here,wehavesignificantly increased the number of derivatives of the C-glycoside ligands of LecB to 30 new derivatives yielding a detailed structure−activity relationship for this class of drug-like antibiofilm compounds. We have determined key interactions of the new more potent derivatives from crystallographic analyses, deduced the structural basis of the increased binding potency for the PA14 variant of LecB, and addressed a new binding pocket on the protein surface. The C-glycosidic dimethylthiophene sulfonamide 22 was identified as front runner with good potency in a biofilm assay and excellent ADME/Tox properties. ■RESULTS AND DISCUSSION Design. C-Glycosides of the amide series, e.g., 4and 5, and the sulfonamide series, e.g., 6and 7, have previously been reported by us as potent glycomimetic inhibitors of LecB (Figure 1). 41 To expand the structure−activity relationship of these potent compound classes, we aimed to further diversify the amide and sulfonamide substituents and explore their interaction with LecB. In the amide series, the cinnamide and dimethoxycinnamide derivative of a mannoside and its C-glycoside analogue are better binders than benzamides or aliphatic amides. 36,38,41 We therefore aimed at analyzing a potential rigidification of the cinnamide by ring closure between the ortho-position of the phenyl group and the α-carbon by introducing heteroatom linkers in benzothiophene-, benzofuran-, and indole-2-carboxamides (9−11). These substitutions affect rigidity, hydrogenbond donating or accepting properties, and total polar surface area of the resulting molecules while maintaining the original cinnamide pharmacophore. Furthermore, we substituted the double bond in the cinnamoyl group with 5-membered heterocycles, i.e., a thiazole and a thiophene (12,13). To address the very potent sulfonamide series of Cglycosides and expand the highly orally bioavailable thiophenyl derivative 7, we included a number of different 5-membered heterocycles (14−28), e.g., furan, oxazole, pyrazoles, and the regioisomer of the original thiophene. In addition, various substituents were attached to those heterocycles, and the focus was set to methyl groups that showed good potency increase in the previous mannose-series (IC50s for LecBPA14: phenylsulfonamide 16 μM→trimethylphenylsulfonamide 1 μM, see Figure 2, compounds 47 and 3). 36 Inspection of the crystal structure 41 of the complex of 7with LecB indicated a possible second shallow cleft between the two loops formed by residues Val69−Asp75 and Glu95−Ala105 Figure 2. Competitive binding assay of inhibitors with LecBPAO1 and LecBPA14 based on fluorescence polarization. Means and standard deviations were determined from a minimum of three independent experiments. n.s.: not soluble at 1 mM in TBS/Ca containing 1% DMSO. IC50 values for 1−7,38, and 41−46 with LecBPAO1 and LecBPA14 were previously published. 23,36−38,41 Journal of Medicinal Chemistry Article DOI: 10.1021/acs.jmedchem.9b01120 J. Med. Chem. 2019, 62, 9201−9216 9203 within reach of the thiophene (Supporting Information, Figure S1). Therefore, we designed short spacers of directly attached to the thiophene moiety in position 5 with the aim to tether additional substituents targeting this cleft (29−37). Synthesis of C-Glycoside LecB Inhibitors. For the synthesis of the β-C-glycosides 8−28,weobtainedthe precursor 39 from L-fucose (38) by Henry addition of nitromethane with in situ ring closure to the β-anomer followed by a reduction to yield amine 40 (Scheme 1). 37,41,42 Then, diversification to give amide or sulfonamide substituted LecB antagonists was generated by coupling with different electrophiles (Scheme 1). The final coupling step yielded amides 8−13 and sulfonamides 14−28 in moderate to good yields (17−76%, over 2 steps). To probe the potential additional binding pocket on LecB (see Supporting Information, Figure S1), the bromothiophene 28 was further transformed in palladium-catalyzed Sonogashira cross coupling reactions to the substituted alkyne derivatives 29−36 in acceptable yields of 24−74% for these protectinggroup free syntheses. Z-Styryl 37 was obtained from the Sonogashira product 29 following hydrogenation with a Lindlar catalyst in 59% yield. Inhibition of LecB Carbohydrate Binding Function. All synthesized structures were then analyzed for their capacity to inhibit both representative lectin variants of the two clinically relevant bacterial strain clades, LecBPAO1 and LecBPA14, using established competitive binding assays 23,36 (Figure 2). Previously, the impact of modifications at the cinnamide substituent in 2was found negligible and a dimethoxy substitution (43,Figure 2) only marginally increased potency. 38 Because rigidification and extension of the cinnamide to a naphthalene carboxamide was also tolerated by the protein for the mannose-series, 38 we tested new derivatives of C-glycosidic cinnamide 4: benzothiophene (9), benzofuran (10), and indole (11). Despite the introduction of isosteric changes in polarity and hydrogen bonding properties, 9−11 showed similar or slightly decreased activities in this series with benzothiophene 9as the best inhibitor of both LecB variants (IC50 4.28 and 2.34 μM, for LecBPAO1 and LecBPA14, respectively). A reduction in affinity which was especially pronounced for the PAO1-type lectin was also observed for carboxamide-linked thiazole 12. The corresponding thiophene derivative 13 was insoluble under the assay conditions. In analogy to the mannose-series, C-glycosidic sulfonamide derivatives 6and 7showed superior affinities over the amidegroup (4,5). While a relatively extended SAR was described for the cinnamides in the mannose series, 38 the previously synthesized mannose-derived sulfonamides 36 occupy a narrow Scheme 1. Synthesis of the Amides 8−13 and Sulfonamides 14−38 a a Reagents and conditions: (a) MeNO2, DBU, molecular sieves 3 Å, 1,4-dioxane, 50 °C, 3 d; (b)Pt/C, H2, HCl, MeOH, rt, 2 d; (c) acyl/sulfonyl chloride or carboxylic acid/EDC·HCl, Et3N, DMF, 0 °C; (d) CuI, Pd(PPh3)2Cl2, RCCH, Et3N, DMF, 50 °C, 16−42 h; (e) 1 atm H2, Lindlar’s catalyst, quinoline, rt, 46 h. Yields for 8−28 are given over two steps from the nitro derivative 39. Journal of Medicinal Chemistry Article DOI: 10.1021/acs.jmedchem.9b01120 J. Med. Chem. 2019, 62, 9201−9216 9204 structural diversity: substitution of the phenyl moiety in 46 with methyl groups (→3) leads to increased affinities and isosteric replacement of the phenyl substituent by thiophene (46 →45) improved potency toward LecB. To extend the SAR of this potent LecB inhibitor class, we tested a set of C-glycosidic 5-membered heteroatomsubstituted sulfonamides with additional substituents varying in constitution and heteroatom position (14−28). Despite the rather high structural diversity of the sulfonamide substituents, all tested compounds potently inhibited both LecB variants in the low microto nanomolar range (IC50[LecBPAO1] 1.32−7.01 μM; IC50[LecBPA14] 0.20−1.04 μM). While not affected by constitutional change of the heteroatom position (7→21,or 19 →20), the affinity dropped slightly if the ring sulfur was substituted by oxygen in furan derivatives (7→14 or 22 → 15) or nitrogen in pyrazoles (18−20). Enlargement of the aromatic core by addition of alkyl groups or halogen substituents in 7generally had a beneficial impact on binding especially in the thiophene series with low microto nanomolar activities of 22−25 for both LecB variants. Dimethyl 22 and monomethyl 25 (IC50[LecBPAO1] 1.32−1.87 μM; IC50[LecBPA14] 0.20−0.33 μM) were the best ligands among these thiophenes, both presenting a methyl group in ortho-position to the sulfonamide linker. A similar phenomenon was observed for the pyrazoles, where the best compound 18 (IC50[LecBPAO1] 2.42 μM; IC50[LecBPA14] 0.68 μM) also had such an ortho-methyl substituent compared to the less active derivatives 19 and 20 (IC50[LecBPAO1] 5.88−7.01 μM; IC50[LecBPA14] 0.73−0.96 μM), whose methyl groups were in different positions. Extension of the initial and unsubstituted thiophene 7to address the second cleft on LecB yielded the set of 10 alkyne and alkene derivatives 29−37. Again, all compounds showed potent inhibition of both lectin variants in the submicromolar range for the PA14-type and a 1−3μMaffinity range for the PAO1-type. Phenylacetylene derivative 29 stands out as most potent inhibitor of both LecB variants (IC50[LecBPAO1] 1.52 μM; IC50[LecBPA14] 0.14 μM). Further modifications of 29 were tolerated by both proteins although with a moderate reduction in affinity of up to 2-fold: replacement of the phenyl with other groups (34,35) or elongated spacing (36) was possible. The position of a methyl substituent at the phenyl group in ortho-, meta-, or para-position (30−32) did not have an influence on activity, and transformation of the acetylene into Z-alkene 37 was also not altering potency dramatically. The thermodynamics of binding of the potent 2,5-dimethyl thiophene derivative 22 to both LecB variants was then further studied by isothermal titration calorimetry (Figure 3, Supporting Information, Table S1). The ligand showed a 1 ligand to 1 LecB monomer binding stoichiometry and affinities for LecB in the low microto nanomolar range, Kd= 1.2 μM for PAO1 and Kd= 0.32 μM for PA14, and thus confirmed the obtained IC50 data. As observed for 7, 41 also this thiophenecontaining ligand 22 showed an enhanced enthalpy driven binding (ΔH−47.3 to −40.4 kJ/mol) compared to the carbocyclic derivatives 4and 6, which was partially compensated by disfavored entropic contributions (−TΔS 13.5−3.3 kJ/mol). Structure of LecB in Complex with Dimethylthiophene 22. To analyze the impact of the methyl substituents in this inhibitor class, we performed crystallization of dimethylthiophene 22 in complex with LecBPA14 using hanging drop cocrystallization. The complex of the lectin crystallized in the P61 space group with four protomers per asymmetric unit. The resulting structure was solved to 1.45 Å resolution, and all carbohydrate-binding sites were occupied with compound 22 (Figure 4, Supporting Information, Table S2). In previous structures of LecB with sulfonamide ligands, we have observed two different rotamers for the sulfonamides resulting in two types of interaction: a specific interaction 41 with the protein as described for 6or 7with the cleft enclosed between the two loops Val69−Asp75 and Asp96−Asp104 and a likely unspecific interaction 36 induced by crystal contacts for the mannose-derivative 3. Interestingly, in the structure of LecB with dimethylthiophene derivative 22, both binding poses can be observed (Figure 4A,B,C). In this respect, one binding mode of 22 resembles the structure of LecB with thiophene 7in a way that the thiophene residue is oriented between the same loops of the protein without disturbing crystal contacts (Figure 4B, D). This supports the previous argumentation for the relevance of this binding mode compared to the second observed pose of 22, which is similar to the reported structure 36 of the mannose analogue 3where crystal packing interactions likely favored the altered orientation of the sulfonamide substituent (Figure 4A). Furthermore, the crystal structure of 7in complex with LecBPA14 shows a tight coordination of the thiophene moiety to the CH2group of Ser97 (S-CH2distance 4.3 Å, or the calculated S-CH2distance 3.3 Å, sum of H and S van der Waals radii is 3.05 Å). This Ser97 is part of the loop Asp96−Asp104 adjacent to the carbohydrate-binding site. A highly similar interaction is seen in the complex of 22 and LecBPA14 (S-CH2 distance 3.9 Å, or the calculated S-CH2distance 3.2 Å, sum of H and S van der Waals radii is 3.05 Å). The methyl group of 22 in ortho-position to the sulfonamide linker enters deeply into this still rather shallow pocket and establishes numerous lipophilic contacts with Gly24, Val69, and the CH2of Asp96. This extensive interaction pattern serves as an explanation for the beneficial effect on the binding affinity of the ortho-methyl group. Coordination of the carbohydrate-derived ligands by LecBPAO1 and LecBPA14 is largely similar. 23 Subtle differences are introduced by the sequence variations between those two type strains and result in the observed difference in activity of Figure 3. Isothermal titration microcalorimetry of LecBPAO1 and LecBPA14 with dimethylthiophene 22. Means and standard deviations were determined from a minimum of three independent titrations. One representative titration graph is depicted for LecBPAO1 only. Journal of Medicinal Chemistry Article DOI: 10.1021/acs.jmedchem.9b01120 J. Med. Chem. 2019, 62, 9201−9216 9205 the LecB variants. Sulfonamides are generally recognized very well, and the PA14 variant of LecB has a very high affinity for the sulfonamides, 3−10-fold higher than the PAO1 variant. Interestingly, in the cinnamide series of the carboxamides, these differences are vanished and sometimes selectivity is even inverted to the benefit of LecBPAO1 (e.g., 20 vs 37 μM for 43 and 2.34 vs 2.8 μM for 5, respectively). An explanation for this observation results from the presence of Ser97 in LecBPA14 that introduces further steric bulk compared to Gly97 in its PAO1 relative. The LecB structures reveal this Gly to Ser variation for LecBPA14 at position 97 pointing into the ligand binding site Figure 4. Crystal structure of LecBPA14 with C-glycoside ligand 22 (1.45 Å resolution, PDB 5MAZ), (A) Observed binding pose of 22 with crystal contacts. (B) Observed binding pose of 22 without crystal contacts. (C) Superposition of both observed binding poses. (D) Relevant binding pose of 22 with 2Fobs −Fcalc electron density displayed at 1σ. Ligands and amino acids of the carbohydrate recognition domain (CRD) are depicted as sticks colored by elements (C, gray; N, blue; O, red; S, yellow); protein surface in transparent blue and two Ca2+-ions in the binding sites are shown as green spheres. Figure 5. (A) Crystal structure of the complex of LecBPA14 with dimethylthiophene 22 (PDB 5MAZ) reveals the hydrophobic interaction of the ortho-methyl group attached to the thiophene residue with a hydrophobic patch on the protein surface. Furthermore, the thiophene interacts with the side chain CH2of Ser97. (B) Crystal structure of the complex of LecBPAO1 with manno-cinnamide 2(PDB 5A3O) reveals the hydrophobic interaction of the cinnamoyl group with a hydrophobic patch formed by the loop carrying Gly97 that lacks the serine side chain present in LecBPA14. (C) superposition of the two structures indicating the steric clash between carboxamide substituents, conformationally fixed through a hydrogen bond of their NH group with the carboxylate of Asp96, and the bulk of the side chain of Ser97 in LecBPA14. This serves as an explanation for the increased selectivity of the amides for LecBPAO1, which contrasts the selectivity of the sulfonamides for LecBPA14. Electron density 2Fobs − Fcalc is displayed at 1σ. Ligands are depicted as sticks colored by elements (C, gray; N, blue; O, red; S, yellow); protein surface in blue and two Ca2+-ions in the binding sites are shown as green spheres. Journal of Medicinal Chemistry Article DOI: 10.1021/acs.jmedchem.9b01120 J. Med. Chem. 2019, 62, 9201−9216 9206 (Figure 5). In case of LecBPAO1, this serine is absent and a glycine moiety occupies its position. The position of the entire loop incorporating either serine or glycine is not affected by this substitution. Now, the complex of LecBPAO1 with the cinnamide of the mannose series (2,Figure 5B) shows a tight lipophilic interaction of the cinnamide moiety with the CH2 group of glycine. 38 Furthermore, the cinnamide residue is additionally conformationally fixed through a hydrogen bond of the cinnamide NH with the side chain of Asp96. Because the carboxamides are conformationally fixed in their orientation through this hydrogen bond of the amide-NH with Asp96, they would experience a steric repulsion from this Ser97 in the PA14 variant of LecB and therefore possess the observed less pronounced affinity increase compared to the sulfonamides with LecBPA14 (see overlay in Figure 5C). The latter ones are positioned differently due to the different steric orientation of the linking sulfonamide (planar and trans for the carboxamides, Figure 5B, compared to three-dimensional and cis for the sulfonamides, Figure 5A). Therefore, the selectivity differences of LecBPA14 and LecBPAO1 for the cinnamide series of the carboxamides depend on the amino acid present at position 97. Inhibition of Bacterial Biofilm Formation. The two most promising compounds selected from the competitive binding assay, dimethyl thiophene 22 and Sonogashira product 29, were then tested in a biofilm assay. We used genetically modified P. aeruginosa PA14 constitutively and intracellulary expressing the fluorescent protein mCherry from the pMP7605 plasmid, 43 followed by quantification of fluorescence using confocal light scanning microscopy (CLSM) as previously described. 41 The prerequisite for this in situ imaging assay that fluorescence intensity directly correlates with cell density was established earlier. 41 The desired absence of bactericidal or bacteriostatic effects was also confirmed for these two selected compounds by measuring total fluorescence intensities of the bacterial cultures after growth in the presence of 100 μM compounds for 23 h. Bacteria reached comparable densities with compounds as the DMSO control (Figure 6). For biofilm inhibition experiments, bacterial cultures were grown in the presence of 100 μM compounds for 48 h when biofilm mass was quantified by CLSM (Figure 7). The two sulfonamide C-glycosides 22 and 29 showed very potent inhibition of P. aeruginosa biofilm formation by >80% and >90%, respectively. This statistically significant reduction of biofilm formation compared to the DMSO control is contrasting the nonsignificant effect of the natural carbohydrate ligands methyl α-D-mannoside (1) and methyl α-Lfucoside (42), previously reported (data for the latter two compounds from Sommer et al. 41 ). This inactivity of 1and 42 is in contrast to their good biophysical protein binding activity and may result from continuous depletion in the complex biofilm experiment where bacterial factors could degrade these molecules and/or employ these natural glycosides as nutrients, whereas the C-glycosides remain active. In Vitro Metabolic Stability and Toxicity. In vitro metabolic stability of this compound class was assessed for a number of sulfonamide-based LecB inhibitors in the presence of mouse or human liver microsomes and murine plasma (Table 1,Figure 8A, Supporting Information, Table S3). Compounds were selected based on structural diversity of the sulfonamide substituent and included oxazole 17, dimethylthiophene 22, methylthiophene 23, dichlorothiophene 27, and the Sonogashira product 29. A low intrinsic clearance (CLint) by mouse and human liver microsomes for all tested compounds with values at 10 μL/min/mg protein or below revealed a very high stability (Table 1). In murine plasma, those five compounds were also fully stable over a period of 120 min without any detectable degradation (Figure 8A). These data reveal a high degree of in vitro metabolic stability for this compound class and support our recent results on sulfonamide-linked C-glycosides bearing trimethylphenyl 6and Figure 6. Bacterial growth of mCherry-expressing P. aeruginosa quantified by fluorescence intensity (FI) and normalized on the DMSO control in the presence of 100 μM lectin inhibitors 22 or 29. Figure 7. Inhibition of biofilm formation by P. aeruginosa after 48 h growth in the presence of compounds 22 or 29. Depicted data for methyl α-L-fucoside (42), 1, and the DMSO control have been published. 41 (A) Quantification of biofilm biomass. Averages and standard deviations of biofilm formation from three independent assays. Statistical significance was calculated using the Student’sttest. (B) Raw data of confocal fluorescence microscopy 3D images show one representative z-stack per condition. Table 1. Metabolic Stability of Selected Sulfonamides against Mouse and Human Liver Microsomes a CLint compd mouse human 17 <10 <10 22 <10 <10 23 10 <10 27 <10 <10 29 <10 <10 a Intrinsic clearance (CLint)inμL/min/mg. Journal of Medicinal Chemistry Article DOI: 10.1021/acs.jmedchem.9b01120 J. Med. Chem. 2019, 62, 9201−9216 9207 thiophene 7that also showed a good metabolic stability against liver microsomes and murine plasma. 41 The five thiophenes were further studied, and toxicity of compounds was assessed in vitro using the immortalized human hepatocyte cell line Hep G2 (Figure 8B). Only the alkyne derivative 29 displayed a weak toxicity at the highest concentration tested resulting in 64% hepatocyte viability. No toxicity was detected for all other compounds tested up to a concentration of 100 μM, which is consistent with previous in vitro data on 6and 7and allows to generally classify this compound class as nontoxic against Hep G2 cells. Additionally, the sulfonamides 6and 7are nontoxic in a murine pharmacokinetics model as reported. Binding to plasma proteins reduces the amount of free drug, which is the necessary state for binding to the drug’s target. On the contrary, plasma protein binding can positively result in prolonged plasma half-lives (Figure 9). Therefore, balancing the plasma protein binding is necessary, and we tested this property for a set of selected LecB-targeting sulfonamides, the C-glycosides 6,7,17,22,23, and 29, and the O-glycosides 3 and 45 (which are the O-linked analogues of the C-glycosides 6and 7). All tested compounds showed approx 70−80% plasma protein binding with the exception of the alkyne 29, which was highly bound by plasma proteins. The two sulfonamides trimethylphenyl 6and thiophenyl 7show a plasma protein binding of 73% and 81%, respectively. Both compounds had been analyzed for their pharmakokinetics in mice, and the compound with higher plasma protein binding, 7, also showed an increased plasma half-life of 34 min vs 17 min for 6. 41 ■CONCLUSIONS In summary, two clinically relevant variants of LecB were tested with >30 newly synthesized glycomimetic inhibitors to extend the known structure−activity relationship of Cglycosides. Potent inhibition for LecB of both strain types was observed, and sulfonamides generally performed better in inhibition experiments than the carboxamides. Interestingly, the affinity difference of the carboxamides for the two lectins variants was rather small and usually only approximately 2-fold better for LecBPA14. In contrast, the sulfonamides displayed a higher affinity increase for LecBPA14 over LecBPAO1, and the difference was up to 10-fold for the Sonogashira products 29− 36. The crystal structure of the dimethylthiophene 22 with LecBPA14 explains its increased potency due to a lipophilic interaction of the methyl group in ortho-position to the Figure 8. (A) Stability of LecB ligands in mouse plasma. The error bars show the standard deviation of minimum three assays. (B) Toxicity of LecB ligands to human liver Hep G2 cells. Measured OD560nm/670nm was normalized to the controls. Untreated cells served as negative control (normalized OD560nm/670nm = 1), and Triton X-100 treated cells served as positive control (normalized OD560nm/670nm = 0). The error bars show the standard deviation of minimum three assays. Figure 9. Plasma protein binding of LecB inhibitors C-glycosides 6− 29 and the O-glycosides 3and 45 and naproxen as a control. Box plots show mean and confidence intervals. Journal of Medicinal Chemistry Article DOI: 10.1021/acs.jmedchem.9b01120 J. Med. Chem. 2019, 62, 9201−9216 9208 sulfonamide with hydrophobic protein residues. Furthermore, by comparison of this LecBPA14/22 structure with the previously reported structure of cinnamide 2with LecBPAO1, the selectivity differences of LecBPA14 and LecBPAO1 for the cinnamide series of the carboxamides could be assigned to the amino acid present at position 97: serine in LecBPA14 induces a steric clash with the cinnamide residue that is absent in case of Gly97 for LecBPAO1. The two promising compounds, dimethylthiophene 22 and phenylacetylene bearing thiophene 29, were further tested in a biofilm assay. Both compounds showed a strong inhibition of biofilm formation by P. aeruginosa. Further compound profiling for ADME and toxicity parameters showed that both compounds and others of this class had highest metabolic stability in murine plasma and with mouse and human liver microsomes. However, the dimethyl thiophene 22 was finally prioritized as the front-runner due to a very high plasma protein binding and mainly the moderate mammalian cytotoxicity observed at 100 μM for the otherwise very potent extended structure 29. In general, C-glycosidic sulfonamides showed highest potency toward both LecB strain-types and their specific structure−activity relationship was rationalized by crystal structure analyses. Nevertheless, the surface exposed nature of the rather shallow binding site where the sulfonamide substituent resides on LecB allows a certain degree of modification at the sulfonamide substituents. This will be a beneficial trait for future optimization of compounds or attachment of other cargo molecules, e.g., imaging probes 44 or antibiotics for targeted delivery. ■EXPERIMENTAL SECTION Chemical Synthesis. Thin layer chromatography (TLC) was performed on Silica Gel 60 coated aluminum sheets containing fluorescence indicator (Merck KGaA, Darmstadt, Germany) and developed under UV light (254 nm) and aqueous KMnO4solution or a molybdate solution (a 0.02 M solution of ammonium cerium sulfate dihydrate and ammonium molybdate tetrahydrate in aqueous 10% H2SO4). Prepacked Silica Gel 60 columns from Interchim and a Teledyne Isco Combiflash Rf200 system were used for preparative medium pressure liquid chromatography (MPLC). Nuclear magnetic resonance (NMR) spectroscopy was performed on a Bruker Avance III 500 UltraShield spectrometer or on a Bruker Avance III 400 UltraShield spectrometer at 500 MHz/400 MHz (1H), 126 MHz/101 MHz (13C), or 376 MHz (19F). Chemical shifts are given in parts per million (ppm) and were calibrated on residual solvent peaks as internal standard. 45 Multiplicities were specified as s (singlet), d (doublet), t (triplet), q (quartet), or m (multiplet). The signals were assigned with the help of 1H, 1H-COSY, and DEPT-135-edited 1H, 13C-HSQC experiments. Assignment numbering of the C-glycoside atoms and groups corresponds to the numbering in fucose. High resolution mass spectra (HRMS) were obtained on a Bruker maxis 4G hr-QqToF spectrometer, and the data were analyzed using DataAnalysis (Bruker Daltonics, Bremen, Germany). Commercial chemicals and solvents were used without further purification. Deuterated solvents were purchased from Eurisotop (Saarbrucken, Germany). C-Glycoside 39 was synthesized following the procedure described by Phiasivongsa et al., 42 and reduction toward amine 40 was described previously. 37 The purity of the final compounds was further analyzed by HPLC-UV, and all UV active compounds had a purity of at least 95%. Chromatographic separation was performed on a Dionex Ultimate 3000 HPLC (Thermo Scientific, Germany) with UV detection at 254 nm using a RP-18 column (100/2 Nucleoshell RP18plus, 2.7 μm, from Machery Nagel, Germany) as stationary phase. LCMS grade distilled MeCN and double distilled H2O were used as mobile phases. In a gradient run, an initial concentration of 5% MeCN in H2O was increased to 95% during 7 min at a flow rate 600 μL/min. The injection volume was 10 μL of 1 mM compound in H2O/DMSO = 100:1. General Procedure for Amide and Sulfonamide Couplings. β-L-fucopyranosyl methylamine (40) (1 equiv) and triethylamine (1.5 equiv) were dissolved in dry DMF (30 mL per gram substrate) and cooled to 0 °C. The corresponding chloride (1.2 equiv) dissolved in DMF (0.08 M) was added dropwise under nitrogen. In the case of carboxylic acids, EDC·HCl (1.2 equiv) was added. The reaction was allowed to warm to rt and was stirred for further 1−24 h. Saturated aqueous NH4Cl was added and extracted with EtOAc. The combined organic layers were dried over Na2SO4,filtered, and concentrated in vacuo. Unless otherwise indicated, the residue was purified by chromatography on silica (CH2Cl2to CH2Cl2/EtOH = 10:1 or CH2Cl2/MeOH = 10:1). N-β-L-Fucopyranosylmethyl benzamide (8). Compound 8was obtained following the general procedure from 40 and benzoyl chloride as a colorless solid (70.9 mg, 0.251 mmol, 55%). 1H NMR (500 MHz, MMeOH-d4)δ7.82−7.81 (m, 2H, CHphenyl), 7.56−7.51 (m, 1H, ArCH), 7.78−7.44 (m, 2H, ArCH), 3.71 (m, 2H, CH2NH), 3.67−3.64 (m, 1H, H-4), 3.64−3.60 (m, 1H, H-5), 3.50−3.48 (m, 2H, H-1, H-3), 3.35−3.32 (m, 1H, H-2), 1.25 (d, J= 3.61 Hz, CH3). 13C NMR (126 MHz, MeOH-d4)δ170.9 (C = O), 135.9 (ArC), 132.7 (ArCH), 129.6 (ArCH), 128.5 (ArCH), 80.0 (C-2), 76.3 (C3), 75.8 (C-5), 73.7 (C-4), 69.9 (C-1), 42.6 (CH2), 17.2 (C-6) ppm. HRMS calcd C14H20NO5+: 282.1336, found 282.1345. N-β-L-Fucopyranosylmethyl Benzo[b]thiophene-2carboxamide (9). Compound 9was obtained following the general procedure from 40 and benzo[b]thiophene-2-carbonyl chloride as a yellowish solid (12 mg, 0.036 mmol, 21%). 1H NMR (500 MHz, MeOH-d4)δ7.99 (s, 1H, ArCH), 7.93−7.88 (m, 2H, ArCH), 7.46−7.39 (m, 2H, ArCH), 3.77−3.72 (m, 1H, CH2NH), 3.70−3.61 (m, 3H, CH2NH, H-4, H-5), 3.54−3.46 (m, 2H, H-1, H-3), 3.38−3.33 (m, 1H, H-2), 1.28 (d, J= 6.31 Hz, 3H, CH3). 13C NMR (126 MHz, MeOH-d4)δ 165.4 (CO), 142.6 (ArC), 140.9 (ArC), 140.0 (ArC), 127.6 (ArCH), 126.8 (ArCH), 126.4 (ArCH), 126.1 (ArCH), 123.3 (ArCH), 80.0 (C-2), 76.4 (C-3), 76.0 (C-5), 73.8 (C-4), 70.0 (C1), 42.8 (CH2), 17.3 (CH3). HRMS calcd C16H20NO5S+: 338.1057, found 338.1045. N-β-L-Fucopyranosylmethyl Benzo[b]furan-2-carboxamide (10). Compound 10 was obtained following the general procedure from 40 and benzo[b]furan-2-carbonyl chloride as a colorless solid (32 mg, 0.10 mmol, 58%). 1H NMR (500 MHz, MeOH-d4)δ7.74−7.71 (m, 1H, ArCH), 7.61−7.59 (m, 1H, ArCH), 7.50 (d, J= 0.80 Hz, olefinH), 7.48−7.44 (m, 1H, ArCH), 7.34−7.30 (m, 1H, ArCH), 3.83− 3.79 (m, 1H, CH2NH), 3.67−3.61 (m, 3H, CH2NH, H-4, H-5), 3.50−3.48 (m, 2H, H-1, H-3), 3.37−3.33 (m, 1H, H-2), 1.28 (d, J= 6.31 Hz, 3H, CH3). 13C NMR (126 MHz, MeOH-d4)δ161.7 (C O), 156.6 (ArC), 150.0 (ArC), 129.0 (ArC), 128.4 (ArCH), 125.0 (ArCH), 123.9 (ArCH), 113.0 (ArCH), 111.6 (olefin-CH), 80.0 (C2), 76.4 (C-3), 76.0 (C-5), 73.8 (C-4), 70.3 (C-1), 42.2 (CH2) 17.3 (CH3). HRMS calcd C16H20NO6+: 322.1285, found 322.1300. N-β-L-Fucopyranosylmethyl 1H-Indole-2-carboxamide (11). Compound 11 was obtained following the general procedure from 40,1H-indole-2-carboxylic acid and EDC·HCl as a colorless solid (28 mg, 0.09 mmol, 31%). 1H NMR (500 MHz, DMSO-d6)δ11.57 (d, J = 2.1 Hz, 1H, NHindole), 8.41 (t, J= 5.7 Hz, 1H, NHCO), 7.60 (d, J= 8.0 Hz, 1H, ArCH), 7.43 (d, J= 8.3 Hz, 1H, ArCH), 7.29−7.12 (m, 2H, 2 ×ArCH), 7.03 (ddd, J= 8.0, 6.9, 1.0 Hz, 1H, ArCH), 4.96 (dd, J= 4.5, 1.5 Hz, 1H, OH), 4.68 (d, J= 5.2 Hz, 1H, OH), 4.39 (d, J= 4.9 Hz, 1H, OH-4), 3.70 (ddd, J= 13.9, 4.6, 2.4 Hz, 1H, CH2NH), 3.53−3.44 (m, 2H, H-5, H-4), 3.36−3.25 (m, 3H, CH2NH, H-2, H3), 3.24−3.18 (m, 1H, H-1), 1.12 (d, J= 6.3 Hz, 3H, H-6). 13C NMR (126 MHz, DMSO-d6)δ161.5 (C = O), 136.4 (ArC), 131.6 (ArC), 127.1 (ArC), 123.3 (ArCH), 121.5 (ArCH), 119.7 (ArCH), 112.3 (ArCH), 102.9 (ArCH), 78.7 (C-1), 74.5 (C-2 or C-3), 73.8 (C-5), 71.6 (C-4), 68.7 (C-2 or C-3), 41.2 (CH2NH), 17.2 (C-6). HRMS calcd C16H21N2O5+: 321.1445, found 321.1429. N-β-L-Fucopyranosylmethyl 2-Phenyl-1,3-thiazol-4-carboxamide (12). Compound 12 was obtained following the general Journal of Medicinal Chemistry Article DOI: 10.1021/acs.jmedchem.9b01120 J. Med. Chem. 2019, 62, 9201−9216 9209 (34) Bernardi, A.; Jimenez-Barbero, J.; Casnati, A.; De Castro, C.; Darbre, T.; Fieschi, F.; Finne, J.; Funken, H.; Jaeger, K.-E.; Lahmann, M.; Lindhorst, T. K.; Marradi, M.; Messner, P.; Molinaro, A.; Murphy, P. V.; Nativi, C.; Oscarson, S.; Penades, S.; Peri, F.; Pieters, R. J.; Renaudet, O.; Reymond, J.-L.; Richichi, B.; Rojo, J.; Sansone, F.; Schaffer, C.; Turnbull, W. B.; Velasco-Torrijos, T.; Vidal, S.; Vincent, S.; Wennekes, T.; Zuilhof, H.; Imberty, A. Multivalent glycoconjugates as anti-pathogenic agents. Chem. Soc. Rev. 2013,42, 4709−4727. (35) Johansson, E. M. V.; Crusz, S. A.; Kolomiets, E.; Buts, L.; Kadam, R. U.; Cacciarini, M.; Bartels, K.-M.; Diggle, S. P.; Camara, M.; Williams, P.; Loris, R.; Nativi, C.; Rosenau, F.; Jaeger, K.-E.; Darbre, T.; Reymond, J.-L. Inhibition and dispersion of Pseudomonas aeruginosa biofilms by glycopeptide dendrimers targeting the fucosespecific lectin LecB. Chem. Biol. 2008,15, 1249−1257. (36) Hauck, D.; Joachim, I.; Frommeyer, B.; Varrot, A.; Philipp, B.; Moller, H. M.; Imberty, A.; Exner, T. E.; Titz, A. Discovery of two classes of potent glycomimetic inhibitors of Pseudomonas aeruginosa LecB with distinct binding modes. ACS Chem. Biol. 2013,8, 1775− 1784. (37) Sommer, R.; Exner, T. E.; Titz, A. A biophysical study with carbohydrate derivatives explains the molecular basis of monosaccharide selectivity of the Pseudomonas aeruginosa lectin LecB. PLoS One 2014,9, No. e112822. (38) Sommer, R.; Hauck, D.; Varrot, A.; Wagner, S.; Audfray, A.; Prestel, A.; Moller, H. M.; Imberty, A.; Titz, A. Cinnamide derivatives of D-mannose as inhibitors of the bacterial virulence factor LecB from Pseudomonas aeruginosa. ChemistryOpen 2015,4, 756−767. (39) Hofmann, A.; Sommer, R.; Hauck, D.; Stifel, J.; GottkerSchnetmann, I.; Titz, A. Synthesis of mannoheptose derivatives and their evaluation as inhibitors of the lectin LecB from the opportunistic pathogen Pseudomonas aeruginosa. Carbohydr. Res. 2015,412,34− 42. (40) Beshr, G.; Sommer, R.; Hauck, D.; Siebert, D. C. B.; Hofmann, A.; Imberty, A.; Titz, A. Development of a competitive binding assay for the Burkholderia cenocepacia lectin BC2L-A and structure activity relationship of natural and synthetic inhibitors. MedChemComm 2016, 7, 519−530. (41) Sommer, R.; Wagner, S.; Rox, K.; Varrot, A.; Hauck, D.; Wamhoff, E.-C.; Schreiber, J.; Ryckmans, T.; Brunner, T.; Rademacher, C.; Hartmann, R. W.; Bronstrup, M.; Imberty, A.; Titz, A. Glycomimetic, orally bioavailable LecB inhibitors block biofilm formation of Pseudomonas aeruginosa. J. Am. Chem. Soc. 2018,140, 2537−2545. (42) Phiasivongsa, P.; Samoshin, V.; Gross, P. Henry condensations with 4,6-O-benzylidenylated and non-protected D-glucose and Lfucose via DBU-catalysis. Tetrahedron Lett. 2003,44, 5495−5498. (43) Lagendijk, E. L.; Validov, S.; Lamers, G. E. M.; de Weert, S.; Bloemberg, G. V. Genetic tools for tagging Gram-negative bacteria with mCherry for visualization in vitro and in natural habitats, biofilm and pathogenicity studies. FEMS Microbiol. Lett. 2010,305,81−90. (44) Wagner, S.; Hauck, D.; Hoffmann, M.; Sommer, R.; Joachim, I.; Muller, R.; Imberty, A.; Varrot, A.; Titz, A. Covalent lectin inhibition and application in bacterial biofilm imaging. Angew. Chem., Int. Ed. 2017,56, 16559−16564. (45) Gottlieb, H. E.; Kotlyar, V.; Nudelman, A. NMR Chemical shifts of common laboratory solvents as trace impurities. J. Org. Chem. 1997,62, 7512−7515. (46) Wilkins, M. R.; Gasteiger, E.; Bairoch, A.; Sanchez, J. C.; Williams, K. L.; Appel, R. D.; Hochstrasser, D. F. Protein identification and analysis tools in the ExPASy server. Methods Mol. Biol. 1998,112, 531−552. (47) Kabsch, W. XDS. Acta Crystallogr., Sect. D: Biol. Crystallogr. 2010,66, 125−132. (48) Winn, M. D.; Ballard, C. C.; Cowtan, K. D.; Dodson, E. J.; Emsley, P.; Evans, P. R.; Keegan, R. M.; Krissinel, E. B.; Leslie, A. G. W.; McCoy, A.; McNicholas, S. J.; Murshudov, G. N.; Pannu, N. S.; Potterton, E. A.; Powell, H. R.; Read, R. J.; Vagin, A.; Wilson, K. S. Overview of the CCP4 suite and current developments. Acta Crystallogr., Sect. D: Biol. Crystallogr. 2011,67, 235−242. (49) McCoy, A. J.; Grosse-Kunstleve, R. W.; Adams, P. D.; Winn, M. D.; Storoni, L. C.; Read, R. J. Phaser crystallographic software. J. Appl. Crystallogr. 2007,40, 658−674. (50) Murshudov, G. N.; Skubak, P.; Lebedev, A. A.; Pannu, N. S.; Steiner, R. A.; Nicholls, R. A.; Winn, M. D.; Long, F.; Vagin, A. A. REFMAC5 for the refinement of macromolecular crystal structures. Acta Crystallogr., Sect. D: Biol. Crystallogr. 2011,67, 355−367. (51) Emsley, P.; Lohkamp, B.; Scott, W. G.; Cowtan, K. Features and development of Coot. Acta Crystallogr., Sect. D: Biol. Crystallogr. 2010,66, 486−501. (52) Schneider, C. A.; Rasband, W. S.; Eliceiri, K. W. NIH Image to ImageJ: 25 years of image analysis. Nat. Methods 2012,9, 671−675. (53) Heydorn, A.; Nielsen, A. T.; Hentzer, M.; Sternberg, C.; Givskov, M.; Ersbøll, B. K.; Molin, S. Quantification of biofilm structures by the novel computer program COMSTAT. Microbiology 2000,146, 2395−2407. Journal of Medicinal Chemistry Article DOI: 10.1021/acs.jmedchem.9b01120 J. Med. Chem. 2019, 62, 9201−9216 9216