The microbiome-derived antibacterial lugdunin acts as a cation ionophore in synergy with host peptides

Abstract

Lugdunin is a microbiome-derived antibacterial agent with good activity against Gram-positive pathogens in vitro and in animal models of nose colonization and skin infection. We have previously shown that lugdunin depletes bacterial energy resources by dissipating the membrane potential of Staphylococcus aureus. Here, we explored the mechanism of action of lugdunin in more detail and show that lugdunin quickly depolarizes cytoplasmic membranes of different bacterial species and acidifies the cytoplasm of S. aureus within minutes due to protonophore activity. Varying the salt species and concentrations in buffers revealed that not only protons are transported, and we demonstrate the binding of the monovalent cations K⁺, Na⁺, and Li⁺ to lugdunin. By comparing known ionophores with various ion transport mechanisms, we conclude that the ion selectivity of lugdunin largely resembles that of 15-mer linear peptide gramicidin A. Direct interference with the main bacterial metabolic pathways including DNA, RNA, protein, and cell wall biosyntheses can be excluded. The previously observed synergism of lugdunin with dermcidin-derived peptides such as DCD-1 in killing S. aureus is mechanistically based on potentiated membrane depolarization. We also found that lugdunin was active against certain eukaryotic cells, however strongly depending on the cell line and growth conditions. While adherent lung epithelial cell lines were almost unaffected, more sensitive cells showed dissipation of the mitochondrial membrane potential. Lugdunin seems specifically adapted to its natural environment in the respiratory tract. The ionophore mechanism is refractory to resistance development and benefits from synergy with host-derived antimicrobial peptides.

Importance: The vast majority of antimicrobial peptides produced by members of the microbiome target the bacterial cell envelope by many different mechanisms. These compounds and their producers have evolved side-by-side with their host and were constantly challenged by the host's immune system. These molecules are optimized to be well tolerated at their physiological site of production, and their modes of action have proven efficient in vivo. Imbalancing the cellular ion homeostasis is a prominent mechanism among antibacterial natural products. For instance, over 120 naturally occurring polyether ionophores are known to date, and antimicrobial peptides with ionophore activity have also been detected in microbiomes. In this study, we elucidated the mechanism underlying the membrane potential-dissipating activity of the thiazolidine-containing cycloheptapeptide lugdunin, the first member of the fibupeptides discovered in a commensal bacterium from the human nose, which is a promising future probiotic candidate that is not prone to resistance development.

Keywords: Staphylococcus aureus; antibacterial agent; bacteriocin; dermcidin; mechanism of action; membrane potential; microbiome interaction; natural product; synergism.

Fig 1

Lugdunin dissipates the membrane potential in different bacterial species. Upon lugdunin addition (black arrow), the cytoplasmic membranes of S. aureus NCTC8325 (A) and B. subtilis 168 (B) depolarize quickly and in a concentration-dependent manner. Lugdunin, albeit more slowly and at higher concentration, is able to dissipate the membrane potential of an efflux-deficient E. coli mutant (E. coli BW25113 ∆acrA) in the presence of the outer membrane permeabilizing agent PMBN (15 µg/mL) (C). The assays were conducted in phosphate-buffered saline (PBS) buffer with 0.1% glucose and 3,3′-diethyloxacarbocyanine iodide [DiOC₂(3)] as the membrane potential probe. The highest lugdunin concentration shown corresponds to 2× MIC for each of the three strains. CCCP (5 µM) was used as the positive control. All data are normalized to an untreated DMSO control, which was set to a fluorescence ratio (red/green) of 1. The experiment was performed with at least three independent biological replicates with technical duplicates each. Error bars show the standard error of the mean (SEM).

Fig 2

Fluorescence microscopy of lugdunin-treated B. subtilis 168 trpC2. (A) Lugdunin-mediated (10 µg/mL, 1.6× MIC) membrane depolarization caused delocalization of the cell division proteins YFP-FtsA and GFP-MinD similar to the positive control CCCP (50 µM, 5× MIC), while vancomycin (10 µg/mL, 20× MIC) served as a negative control. (B) Co-staining with the membrane-permeant SYTO 9 and the membrane-impermeant propidium iodide (PI) to monitor cytoplasmic membrane integrity. Both dyes emit fluorescence only after binding to DNA. Lugdunin (25 µg/mL, 4×MIC) did not lead to large membrane lesions or pore formation, as PI was not able to enter the cytoplasm. Crude nisin (100 µg/mL) was used as a positive control for the formation of pores. (C) Lugdunin (25 µg/mL, 4× MIC) does not cause formation of membrane blebs in B. subtilis. Membrane blebbing was observed after treatment with vancomycin (5 µg/mL, 10× MIC) or gramicidin S (12 µg/mL, 4× MIC). Scale bars, 5 µm.

Fig 3

Lugdunin shows protonophore activity in S. aureus cells. In S. aureus NCTC8325 cells loaded with the pH-sensitive dye BCECF, lugdunin caused acidification of the cytoplasm in a concentration-dependent manner (A). Acidification was also observed upon the addition of gramicidin A (B), but not valinomycin (C). Concentrations used correspond to 1–8× MIC for lugdunin (3.13–25 µg/mL), 0.6–2.5× MIC for gramicidin A (12.5–50 µg/mL), and 0.6–2.5× MIC for valinomycin (12.5–50 µg/mL). The protonophore CCCP (50 µM, 5× MIC) was used side-by-side as a positive control in all experiments. The addition of the compounds is indicated by a black arrow. Ten minutes after compound addition, nigericin (20 µM) was added to all samples (gray arrow) as a control that allows quick acidification of the cytoplasm. The experiment was performed with at least four independent biological replicates with technical duplicates each. Error bars show the SEM.

Fig 4

Dissipation of the membrane potential is affected by different salt species and salt concentrations. Lugdunin (3 µg/mL, 3.83 µM, 1× MIC) treatment led to membrane depolarization of S. aureus NCTC8325 in a 50 mM Tris buffer containing different concentrations of NaCl and KCl, while hyperpolarization was observed when no additional salts were present (A). Depolarization caused by lugdunin was stronger with KCl than NaCl. Contrarily, CCCP (5 µM, 0.5× MIC) led to depolarization under all buffer conditions (B). Gramicidin A (5 µM, 0.5× MIC) caused depolarization in the presence of NaCl and KCl and slight hyperpolarization in the absence of salts (C). Valinomycin (5 µM, 0.25× MIC) caused membrane depolarization only when KCl was added to the Tris buffer, while hyperpolarization occurred without added salts as well as in the presence of NaCl (D). DiOC₂(3) was used as a membrane potential probe in all assays. All data were normalized to the measured 0 min value of the DMSO control, which was set to a fluorescence ratio (red/green) of 1. The experiment was performed with two independent biological replicates (different cultures on different days) with technical duplicates (two parallel aliquots each day). Error bars show the standard deviation (SD).

Fig 5

In vitro assays show binding and transport of monovalent cations by lugdunin. (A) The picric acid assay shows binding of lugdunin to Li⁺, Na⁺, and K⁺ ions. The derivative N-acetyl-lugdunin with strongly decreased antimicrobial activity (MIC S. aureus NCTC8325 ≥100 µg/mL) showed reduced binding of the respective monovalent ions. The ionophore valinomycin strongly interacted with all of the three tested monovalent ions. For daptomycin and erythromycin, no significant interaction was detected under these conditions. H₂O was employed as a negative control. The assay was performed with three replicates; error bars show the SD. (B) Proton transport rates into artificial phospholipid vesicles obtained from the normalized initial slopes of fluorescence time traces (compare Fig. S6) depending on the salt concentration. Lugdunin was added to POPC vesicles filled with the pH-sensitive dye pyranine (peptide-to-lipid ratio 1:250) and exposed to a pH gradient (pH_in = 7.4, pH_out = 6.4). At the start of each experiment, salt concentrations (ranging from 0 to 500 mM KCl, NaCl, or LiCl) were equal between the lumen of the vesicles and the surrounding buffer. From fitting equation E1 (see Fig. S6) to the data, the transport constant K_T was determined reporting on the transport affinity of lugdunin for K⁺, Na⁺, and Li⁺. Data represent the mean ± SEM from greater than or equal to five measurements per salt concentration.

Fig 6

Synergistic action of lugdunin and its enantiomer with dermcidin-derived peptide DCD-1. In membrane potential assays with DiOC₂(3), the addition of 5 µg/mL DCD-1 lowered the concentration of lugdunin (A) or its enantiomer (B) necessary to cause membrane depolarization of S. aureus NCTC8325 by more than fivefold. The synergistic effect emerged immediately after compound addition (black arrows) and established itself prominently and stable (compare right panels after 30 and 60 min of compound exposure). The concentration of DCD-1 (5 µg/mL) employed in the assay does not cause membrane depolarization by itself. The experiment was performed in PBS buffer with at least three biological replicates with technical duplicates each. Error bars show the SEM.

Fig 7

Activity of lugdunin on eukaryotic cells is strongly cell line- and growth condition-dependent. (A) When treated with compounds immediately after seeding, the epithelial cell lines A549 and HEK293T were less susceptible to lugdunin than the suspension cell line THP-1. The ionophore gramicidin A showed notably higher cytotoxic activity against all tested cell lines. (B) When the cells were allowed to settle overnight before treatment, the adherent epithelial cell lines A549, HEK293T, and HeLa were increasingly resistant to lugdunin, while THP-1 cells remained susceptible. Half-maximal inhibitory concentrations (IC₅₀) were calculated by nonlinear regression in GraphPad Prism. IC₅₀ values (in micrograms per milliliter) are depicted for each cell line in their respective graph, and colors reflect the compound used for treatment. Cell viability assays were performed with minimum of three biological replicates; error bars represent the SEM. (C) Mitochondrial membrane potential of A549 and HeLa cells after compound treatment for 10–20 min was visualized using the JC-1 dye. Mitochondria of A549 cells appeared largely unaffected up to the tested concentration of 50 µg/mL. The observed depolarization of mitochondria in HeLa was notable from 12.5 µg/mL. CCCP (6.25 µg/mL) was used as a positive control to depolarize mitochondrial membranes. Numbers in white in the pictures represent the respective compound concentrations in micrograms per milliliter. Scale bars, 10 µm.