Expanding the antibacterial selectivity of polyether ionophore antibiotics through diversity-focused semisynthesis

Abstract

Polyether ionophores are complex natural products capable of transporting cations across biological membranes. Many polyether ionophores possess potent antimicrobial activity and a few selected compounds have the ability to target aggressive cancer cells. Nevertheless, ionophore function is believed to be associated with idiosyncratic cellular toxicity and, consequently, human clinical development has not been pursued. Here, we demonstrate that structurally novel polyether ionophores can be efficiently constructed by recycling components of highly abundant polyethers to afford analogues with enhanced antibacterial selectivity compared to a panel of natural polyether ionophores. We used classic degradation reactions of the natural polyethers lasalocid and monensin and combined the resulting fragments with building blocks provided by total synthesis, including halogen-functionalized tetronic acids as cation-binding groups. Our results suggest that structural optimization of polyether ionophores is possible and that this area represents a potential opportunity for future methodological innovation.

Extended Data Fig. 1. Strategies for the syn-thesis of protected 5-(halo)methylidene tetronic acid building blocks and subsequent coupling

(a) Chemical structure of the targeted 3-acyltetronic acid-derivatives found in nonthmicin/ecteinamycin and 6. No previous syntheses of the halogenated variants have been reported. (b) Examples of known methods used to prepare non-halogenated variants. (c) Failed attempts and model studies toward 5-chloromethylidene tetronate. I) Unsuccessful synthesis of 5-tributylstannylmethylidene tetronate. II) Unsuccessful chlorination of 5-dimethylaminomethylidene tetronate. III) Direct and indirect chlorination of 5-methylidene tetronate methyl ether. In practice, the removal of the methyl group from the chlorinated tetronate product using LiCl/DMSO or BBr₃ was difficult. Thus, we moved to synthesize the tetronate derivatives bearing a TMSE group, which can be removed under much milder condition (TBAF/DMF). DMSO = dimethylsulfoxide DCC = N,N’-dicyclohexylcarbodiimide, DMAP = 4-dimethylaminopyridine, Ms = methanesulfonyl, MOM = methoxymethyl, DCE = 1,2-dichloroethane, DBU = 1,8-diazabicyclo[5.4.0]undec-7-ene, TMS = trimethylsilyl, NCS = N-chlorosuccinimide, AIBN = azobisisobutyronitrile

Extended Data Fig. 2. Synthesis of hybrid polyether HL324 (32)

Preparation of analog HL324 (32) was performed through anti-selective (Evans-Saksena) reduction of the ketone-functionality of 26 followed by DCC-mediated coupling with 5-chloromethylidene tetronate. rt = room temperature, TBAF = tetrabutylammonium fluoride, DMF = N,N-dimethylformamide, DCC = N,N’-dicyclohexylcarbodiimide, DMAP = 4-dimethylaminopyridine, HPLC = high pressure liquid chromatography.

Extended Data Fig. 3. Droplet screen for anti-bacterial activity in B. cereus

a) Representative images of inhibition zones of all com-pounds from B. cereus. b) Heatmap representing no (red), slight (yellow) or large (green) inhibition zones. All compounds were tested at 10 mM in DMSO except monensin (1 mM) and HL160 (7 mM). c) Chemical struc-tures of acid-containing synthetic fragments. SL382 = 27; SL415 = 26 – see Fig. 3.

Extended Data Fig. 4. Use of morphological profiling via cell painting to determine bioactivity threshold in U-2OS cells

a) Plots of cell count (grey) and Mahalanobis distance (‘Bioactivity’; red and green) against concentration of compound. Green points indicate significant activity (mp-value < 0.01) while red points indicate no significant activity (mp-value > 0.01). Many ionophores induce an active profile without loss of cell viability while bioactivity of the hybrid ionophores is correlated with a loss of cells, indicating a profile representing toxicity. The bioactivity threshold is determined as the first concentration to reach significance (mp-value < 0.01) b) Dose-dependent morphological profiles from which the bioactivity threshold is determined. Note that some ionophores change profile at higher concentrations, e.g. calcimycin (Cal) and nan-changmycin (Nan), typically correlated with a loss of cells. See Supplementary Fig. 3–9 for representative images, all dose-response plots, profiles and correlation matrices in both U-2OS and Vero cells.

Fig. 1. Accessing structural diversity within the polyether ionophores.

(a) Flowchart depicting the overall concept of reconstructing new polyether scaffolds by recycling elements from abundant feedstock polyether ionophores. The resulting “hybrid” molecules (blue squares) are plotted in a hypothetical structure and bioactivity space to illustrate the relation of these compounds to the natural polyethers (red squares). The compounds that possess ionophore activity constitute a sub-space of a larger bioactivity-space that can be explored using hybrid polyethers. (b) Chemical structures of polyether ionophores nonthmicin and ecteinamycin. Both are active against gram-positive bacteria, with especially strong potency against C. difficile reported for ecteinamycin. The compounds bear resemblance to lysocellin/ferensimycin but the chlorinated methylidene tetronic acid group of nonthmicin is unprecedented. The X-ray structure depicts ecteinamycin bound to a single sodium-ion and the chemical groups on the hydrophobic periphery that have been altered in the target hybrid polyether 6 have been circled in pink. No crystal structure of nonthmicin is available. (c) Chemical structure of the hybrid polyether 6 with indication of the required fragments and the origin of these fragments. The main fragment, ketone 4, can be obtained in a single synthetic step from lasalocid.

Fig. 2. Building block synthesis.

(a) Route to aldehyde 7 proceeding in ten steps from optically active salt 9. (b) Different synthesis strategies attempted to construct the protected 5-halomethylidene tetronic acid building block. A dihalogenation-dehydrohalogenation of the corresponding 5-methylidene tetronic acid was successful (see also Extended Data Fig. 1). (c) Construction of the required TMSE-protected precursor 19 was carried out in 6 steps from the commercial racemic acetonide-protected glycerate 15. This sequence allowed for preparation of both the chlorine and bromine-variants (8a and 8b). THF = tetrahydrofuran, TBSCl = tert-butyldimethylsilyl chloride, THF = tetrahydrofuran, rt = room temperature, TEMPO = (2,2,6,6-tetramethylpiperidin-1-yl)oxyl, DMAP = N,N-dimethylpyridine-4-amine, DIBAL-H = diisobutylaluminium hydride. Tr = trityl, TMSE = 2-(trimethylsilyl)ethyl, DCC = N,N’-dicyclohexylcarbodiimide, BRSM = based on recovered starting material, Ms = methanesulfonyl, DBU = 1,8-Diazabicyclo[5.4.0]undec-7-ene, TMS = trimethylsilyl.

Fig. 3. Fragment-coupling via boron trifluoride-mediated Mukaiyama-aldol reaction.

(a) The (Z)-TES-enolate 20a could be readily obtained, but special procedures had to be developed to access mixtures of (E)- and (Z)-TES-enolates. Purification could be used to further enrich the (E)-TES-enolate 20b. (b) Aldol reaction affords two major products (22 and 23) depending on the configuration of the silyl-enolate derived from ketone 4. Compound 22 was confirmed by X-ray analysis of derivative 24 to be the initially targeted aldol-product (c) Both aldol products 22 and 23 could be processed towards the final fragment coupling in two high-yielding steps. TESOTf = triethylsilyltrifluoromethanesulfonate, THF = tetrahydrofuran, rt = room temperature, DCE = 1,2-dichloroethane.

Fig. 4. Final coupling of tetronic acid derivatives.

Mild conditions were developed for effecting the coupling of unprotected 5-(halo)methylidene tetronates to carboxylic acids 26 or 27. The structures of HL201 (6), HL204 (29, see Supplementary Information) and HL160 (31), all as the sodium salts, were solved by X-ray diffraction. TMSE = 2-(trimethylsilyl)ethyl, TBAF = tetrabutylammonium fluoride, DMF = N,N-dimethylformamide, DCC = N,N’-dicyclohexylcarbodiimide, DMAP = 4-dimethylaminopyridine, rt = room temperature, HPLC = high pressure liquid chromatography.

Fig. 5. Oxidative deconstruction of monensin for hybrid polyether synthesis.

Single-step Cr-mediated conversion of monensin to complex bis-lactone 33 can be used to generate novel derivatives. Selective protection of the δ-lactone in 33 allows reconstruction of a THF-unit through an oxa-Michael cyclization and subsequent coupling of the 5-chloromethylidene tetronate. Assignment of stereochemistry at C8 in 39a/39b could not be unambiguously done, see Supplementary Information. rt = room temperature, CSA = camphorsulfonic acid, DIBAL-H = Diisobutylaluminium hydride, TBAF = tetrabutylammonium fluoride, THF = tetrahydrofuran, pTSA = p-toluenesulfonic acid, DCE = 1,2-dichloroethane, DCC = N,N’-dicyclohexylcarbodiimide, DMAP = 4-dimethylaminopyridine, HPLC = high pressure liquid chromatography