Educational guide
Differential membrane lipid disruption by lipopeptide ...
Introduction A current estimate is that by 2050, antimicrobial-resistant (AMR) infections will cause 39 million deaths annually 1 . Pathogenic bacteria are associated with over 550,000 neonate deaths worldwide yearly, of which Gram-negative bacteria Klebsiella
This guide cannot diagnose a condition or recommend a personal treatment plan. Discuss medical questions with a qualified professional.
Introduction
A current estimate is that by 2050, antimicrobial-resistant (AMR) infections will cause 39 million deaths annually1. Pathogenic bacteria are associated with over 550,000 neonate deaths worldwide yearly, of which Gram-negative bacteria Klebsiella pneumoniae, Escherichia coli, and Acinetobacter baumannii are classified as urgent concerns2,3. The lipopeptide colistin is a last-line treatment for these infections, but increasing resistance, dose-limiting toxicity, and a limited antimicrobial spectrum threaten its use4,5,6. For that reason, there has been significant effort both in discovering new lipopeptide natural products and in making better colistin analogs7,8,9,10,11. Several of these are currently in clinical trials12,13,14. However, an incomplete understanding of mechanisms of action, resistance, and toxicity hinders development efforts.
Turnercyclamycin lipopeptides are similar to colistin in that they require biosynthesis of lipopolysaccharide (LPS) for their mechanism of action, as ∆lpxC knockouts of Acinetobacter, one of the few Gram-negative pathogens that can survive without LPS, are fully resistant to the agents15,16,17. However, turnercyclamycins retain activity against colistin-resistant Acinetobacter strains without the cytotoxicity and hemolytic activity that is common in lipopeptides. Turnercyclamycins are active against K. pneumoniae, E. coli, and A. baumannii in vitro and controlled infection in an A. baumannii mouse model. Moreover, turnercyclamycins A and B and colistin exhibit very different resistance profiles; turnercyclamycins are active against colistin-resistant mcr-1 strains, while turnercyclamycin B is uniquely blocked by mutations in outer membrane (OM) transporter MlaA. This resistance difference is striking because turnercyclamycins A and B differ simply by the presence of an additional ethylene in the middle of the lipid chain. All three agents are inhibited by aminoarabinosylation of LPS.
Although seemingly straightforward, the colistin/polymyxin mechanism of action remains enigmatic4,5. Polycationic residues in colistin interact with the negatively charged phosphate groups of the outer membrane (OM) lipopolysaccharide (LPS), displacing divalent cations, causing membrane weakness, lysis and cell death5,6,18. Many antimicrobial peptides damage the OM19,20,21,22, causing rapid lysis and early onset of cell death23,24. After promoting OM changes, colistin rapidly damages the CM, which is thought to be its major mechanism of action25,26,27 (among other possible mechanisms28,29,30,31,32,33). For example, damage to P. aeruginosa was amplified in the presence of murepavadin, a drug that blocks P. aeruginosa LptD25. Without a complete mechanistic understanding, it has been challenging to define the problems facing lipopeptide drug development at the molecular level.
We hypothesized that turnercyclamycins would similarly impact CM lipids or proteins. To test this hypothesis, here we measure the impact of colistin and turnercyclamycins A and B on E. coli membrane integrity using cellular dyes. Surprisingly, in contrast to colistin, little CM damage is found with turnercyclamycins, calling the hypothesis into question. Further, we use LPS transport mutants in the non-pathogenic model system Acinetobacter baylyi34,35,36 and transmission electron microscopy (TEM) to directly observe membrane effects. As expected, significant CM damage is observed with colistin, but again surprisingly, not with turnercyclamycins. To further investigate these observations, we directly measure the interactions of lipopeptides with a variety of lipids and also measure lipopeptide effects on the whole cell phospholipodome. Both colistin and turnercyclamycins A and B differently impact the phospholipid content of the bacterial cell, and the activity of each is differently affected by specific phospholipids. Further, direct binding of LPS to colistin is observed by isothermal titration calorimetry (ITC), but turnercyclamycins do not bind LPS in the assay. These results reveal a substantially different mechanism than other clinically used lipopeptides, emphasizing the promising potential of new lipopeptides as therapeutic agents.
Results
Significant data for polymyxin/colistin activity have been gathered over the course of nearly 80 years of rigorous study. Therefore, all experiments in this study compare the individual actions of test compounds turnercyclamycins A and B against those of colistin, enabling robust conclusions in comparison to prior literature in the field. We used three different bacterial species for this work: E. coli, A. baumannii, and A. baylyi.
Antimicrobial spectrum of action
In previous work, we showed that turnercyclamycins target E. coli, K. pneumoniae, and Acinetobacter strains (including colistin-resistant) at 1, 2, and 4–8 µg mL−1, respectively15. We previously tested turnercyclamycins against E. coli K strains. Here, we treated E. coli B strain BL21(DE3), which is notoriously resistant to colistin37, with colistin and turnercyclamycins A (lipid tail: myristic acid, C14:0) and B (lipid tail: palmitoleic acid, C16:1) (Fig. 1). These results show that turnercyclamycins are broadly effective (MIC90 = 2 µg mL−1) against E. coli, including strains such as mcr-1 and BL21(DE3) known to carry colistin resistance, while colistin was much less effective against BL21(DE3) (MIC90 = 8 µg mL−1) in comparison to E. coli K lines (MIC90 = 0.5 µg mL−1). We also tested the efficacy of turnercyclamycins against Pseudomonas aeruginosa. While colistin was active against P. aeruginosa, turnercyclamycins were not.
Fig. 1: Chemical structures of colistin and turnercyclamycins15,16, with MIC90 values against various Gram-negative bacteria.nt is not tested.
Density dependence of antibiotic activity
As found with most lipopeptides, the efficacies of both colistin and turnercyclamycins depend upon the starting density of bacterial culture. While this is sometimes considered to be a disadvantage, it also has an advantage: starting density assays provide a smooth range of conditions that are useful for mechanism of action studies. In initial assays using different concentrations of lipopeptides at different starting densities of culture, we found that applying a concentration of drugs four times higher than the MIC90 seemed potentially ideal. We therefore performed measurements in which lipopeptide concentrations in E. coli were fixed at 4× the MIC90, and the starting density of the inoculum was varied between OD600s of ~0.5 down to «0.01 (Supplementary Fig. 1). Turnercyclamycins and colistin exhibited 90% killing at starting inoculum OD600 0.004, and 0.015, respectively. Turnercyclamycins and colistin showed a gradual increase in percent of dead cells in a concentration range from OD600 0.5 until their 90% kill ODs. This enabled the drugs to be rigorously and reproducibly compared under identical starting OD600s in which the drugs had similar levels of impact on all cell types. Thus, most experiments below use 4× the MIC unless otherwise indicated. In order to ensure that there were still viable cells in experiments, the preceding experiments used starting OD600s that had a substantial population of viable cells treated at 4× MIC.
Colistin causes rapid OM and CM leakage in tandem with cell death, but turnercyclamycins do not
To determine the time-kill curve of turnercyclamycins, a fixed number of E. coli cells were treated with test compounds, incubated, and plated at different time points. Colistin showed the fastest killing curve with significant population reduction as early as 30 min with complete killing at 4 h (Fig. 2a). Turnercyclamycins exhibited a slower killing rate, with turnercyclamycin A significantly reducing cell counts at 6 h with complete inhibition at 10 h, while turnercyclamycin B achieved the lowest counts at 10 h (Fig. 2a).
Fig. 2: Effect of turnercyclamycins on membrane permeabilization and depolarization, and interaction with LPS molecules.All experiments were performed with duplicate replicates. A single legend at the top right is used for (a–c) Col, colistin (2 μg/mL); TurA, turnercyclamycin A (8 μg/mL); Tur B, turnercyclamycin B (8 μg/mL); DMSO, 1% v/v; compound + dye ctrl, compounds tested with the dye but without cells. a Turnercyclamycins are slower to kill E. coli in comparison to colistin. Time-kill curves are shown for E. coli C600-treated cells; y-axis, total bacterial counts (CFU/mL); x-axis, incubation time. b Colistin causes almost immediate OM permeabilization, while turnercyclamycins permeabilize the OM more slowly. NPN fluorescence (y-axis) increases as the OM is permeabilized over time (x-axis). Starting OD600 inoculum = 0.5. c Turnercyclamycins cause significantly less inner membrane damage compared to colistin. Inner membrane damage assessment using Sytox Green for E. coli C600 cells treated with 2 μg/mL colistin, 8 μg/mL turnercyclamycin A, 8 μg/mL turnercyclamycin B, and 1% DMSO v/v. y-Axis shows Sytox green fluorescence and x-axis shows incubation time. Starting OD600 inoculum = 0.5. d Representative fluorescent microscopy images of E. coli C600 cells untreated and treated with 1% DMSO v/v, 8 μg/mL turnercyclamycin A, 8 μg/mL turnercyclamycin B, and 2 μg/mL colistin, for 10 min, and stained with DiSC3(5). Scale bar indicates 10 µm. e, f Micro-ITC profile of turnercyclamycin A, turnercyclamycin B, and colistin binding to LPS analyzed by MicroCal’s ORIGIN software. e Curve generated for colistin showing computed KD values. f Raw titration curves from both colistin and turnercyclamycin interaction experiment with LPS. Curve shown is for colistin while turnercyclamycin A and B binding curve could not be generated. The highest concentrations titrated to 100 µM turnercyclamycin A, turnercyclamycin B, and colistin is 100 µM LPS.
OM permeability was investigated using N-phenyl-1-naphthylamine (NPN) dye, which increases in fluorescence as the OM becomes permeable38,39. As expected, colistin immediately increased OM permeability (Fig. 2b). Turnercyclamycins also caused increased membrane permeability, but at a slower pace and to a much lesser extent than found with colistin.
Damage to the E. coli CM was assessed by monitoring Sytox Green, whose fluorescence increases upon binding to DNA when the CM has been compromised40,41. The dye reflects direct CM permeabilization and damage, but not other possible effects such as alterations to membrane fluidity, domains, or transport. Colistin caused CM damage in a time course paralleling the killing curve, as shown by a large increase in fluorescence (Fig. 2c). However, turnercyclamycins showed a much lower readout, consistent with the minor fluorescence increase that occurs with delayed cell lysis. This suggested that substantial CM damage may not be a central part of turnercyclamycin mechanism of action (Fig. 2c).
Membrane depolarization in E. coli was also assessed using a potentiometric probe, 3,3’-dipropylthiadicarbocyanine iodide (DiSC3(5))42,43,44. Fluorescence microscopy of cells treated with DiSC3(5) revealed that, in comparison to controls, lipopeptide-treated cells were more fluorescent and retained fluorescence over a 2-h period (Fig. 2d and Supplementary Fig. 2), suggesting that the dye penetrated the cells, but membrane depolarization did not occur. Differences between turnercyclamycins and colistin are seen at early time points, when less fluorescence is observed in turnercyclamycin-treated cells. Assessment of changes in Gram-negative bacterial membrane potential remains challenging42. The loss of DiSC3(5) fluorescence was not observed for treated samples, and based on previous observations by Buttress et al.42, the most likely explanation was that treatments caused the OM to become more permeable, allowing more dye to enter the periplasm. As such, further work is required to fully evaluate potential membrane depolarization events.
Unlike colistin, turnercyclamycins do not directly bind to LPS
The actions of colistin on both the OM and CM are thought to be potentiated by binding to LPS and related compounds4,5,6. We thus sought to determine turnercyclamycin binding affinity with LPS using isothermal titration calorimetry (ITC). Similar to previous reports45,46,47, we found that colistin binds to LPS with low micromolar affinity (KD 10.17 μM) (Fig. 2e). However, turnercyclamycins did not show any signs of binding to LPS up to the maximum obtainable experimental concentration of 100 μM (Fig. 2f). These data suggested that, rather than binding directly to LPS molecules as does colistin, turnercyclamycins impact earlier steps in the biosynthesis or transport of LPS or other lipids.
Depletion of LPS biosynthesis, but not transport, abolishes lipopeptide effectiveness
To determine how OM LPS affects the potency of turnercyclamycins, we targeted the Lpt system that transports LPS molecules synthesized in the CM to the OM48,49. lptD knockout mutants are thought to lack LPS in the OM50,51, while lptE encodes a protein complexed with LptD, which disaggregates LPS molecules before OM transport52. We inhibited LPT function in A. baylyi using knockout mutants in the lptD and lptE genes. Colistin and turnercyclamycins retained efficacy against both mutant strains (Fig. 3a and Supplementary Table S2), suggesting that OM LPS is not required for activity of either drug. We previously showed that LPS biosynthetic gene lpxC was required for both colistin and turnercyclamycin efficacy16. Here, we used knockout mutants of lpxA, encoding the first enzyme in LPS biosynthesis53,54. Both colistin and turnercyclamycins lost activity against lpxA mutants (Fig. 3a and Supplementary Table S2). Thus, it is clear that LPS synthesis but not transport is required for colistin and turnercyclamycin potency.
Fig. 3: Effects of LPS, lipids, and OMVs on lipopeptide activity.a MIC90s of colistin, turnercyclamycin A, and turnercyclamycin B against A. baylyi WT, ∆lptD, ∆lptE, and ∆lpxA strains. *indicates concentration ≥32 µg/mL. Darker shades of blue show a higher MIC90 value while lighter blue shows a lower MIC90 value. Experiment was done with two biological replicates. b OMVs decrease the potency of turnercyclamycins, but not colistin. MIC90s of colistin, turnercyclamycin A, and turnercyclamycin B combined with 9.62 × 108 OMV particles/mL/well against A. baumannii ATCC 19606. Light red indicates lower MIC90 and dark red indicates higher MIC90. *indicates concentration ≥32 µg/mL. Experiment was done with two biological replicates. In contrast to the other panels in this figure, the dose of compounds is varied in these experiments. c Percent inhibition of colistin at 4 μg/mL upon interaction with an increased amount of OMVs, against A. baumannii ATCC 19606. 1× is 9.62 × 108 OMV particles/mL/well. Experiment was done with three biological replicates. * indicates p < 0.05; ** indicates p < 0.01. d Percent inhibition of turnercyclamycin A and turnercyclamycin B at 8 μg/mL upon interaction with a reduced amount of OMVs, against A. baumannii ATCC 19606. 1× is 9.62 × 108 OMV particles/mL/well. Experiment was done with three biological replicates. e Turnercyclamycin analogs are detected in isolated OMVs. Extracted ion chromatogram highlighting the presence of turnercyclamycin signal in OMVs isolated from turnercyclamycin A and turnercyclamycin B treated ∆tolA E. coli BW25113. f MIC90 fold change of turnercyclamycin A, turnercyclamycin B, and colistin after combination with various phospholipids against E. coli C600. The phospholipids used are as follows: cardiolipin (CL); hexanoylceramide (HC); phosphatidic acid (PA); phosphatidylcholine (PC); phosphatidylethanolamine (PE); phosphatidylglycerol (PG). A * indicates fold change ≥8. Experiment was done with two biological replicates.
Turnercyclamycins remained potent against single transport and porin mutants
We tested whether single transporter mutants might impact transport of turnercyclamycins and colistin across membranes. A panel of 17 different porin and efflux pump-deficient E. coli and A. baumannii mutants were tested but had minimal effects on drug potency (Supplemental Figs. 3 and 4). This result indicates that the porin and efflux pump mutants tested were not solely responsible for lipopeptide drug transport.
Fig. 4: Turnercyclamycins and colistin induce cellular damages to both WT and ΔlptD A. baylyi.Representative transmission electron micrographs of untreated, colistin, turnercyclamycin A, and turnercyclamycin B-treated WT and ΔlptD A. baylyi. Images shown are either magnified at ×6000 or ×12,000. The experiment was done once using duplicates for each sample.
TEM reveals cellular and membrane damage over a treatment time course
TEM was used to observe morphological changes caused by the lipopeptide antibiotics in A. baylyi wild type and ΔlptD cells over 24 h (Fig. 4 and Supplementary Fig. 5). Colistin damaged wild type cells within 5 h, a time point consistent with the time-kill curves shown in Fig. 2a. Observable damage included extensive cell lysis, increased membrane vesiculation, and membrane breaking. Colistin-induced damage was exacerbated in ΔlptD cells, with heavy damage observed as early as 1 h. Colistin treatment led to visibly damaged OM and CM. In ΔlptD cells, colistin caused obvious and extensive CM damage in cells that still had largely intact OM. Overall, these results provide further support to the literature in the field demonstrating that OM LPS is not central to colistin action.
When treated with turnercyclamycins A and B, cells lysed after 10 h and 12 h, respectively (Fig. 4 and Supplementary Fig. 5). This timing is much later than observed with colistin and is consistent with the time-kill curves in Fig. 2a. Unlike what was observed with colistin, CM damage could not be visualized under any condition over the time course, and instead, modest OM damage followed by complete cell lysis at later timepoints were observed. These differences in action reflect our findings using cell permeability dyes, in which colistin caused extensive CM permeability while turnercyclamycins did not. Alternatively, it should be noted that only a relatively major disruption of membranes is easily visualized by this method.
Also noteworthy was that treated cells exhibited darkly stained granules associated with the CM. While these were found in all treatments, they were more noticeable in turnercyclamycin-treated cells. We cannot speculate from the data as to the identity of these granules. Additionally, it should be noted that only a relatively major disruption of membranes is easily visualized by this method.
Diverse phospholipid species and membrane constituents differentially modulate lipopeptide activity
In previous work, addition of LPS to culture media greatly decreased the efficacy of turnercyclamycins A and B, while lipid A had a lesser impact (both lipids added at 100 μg mL−1)16. By contrast, colistin was equally impacted by both lipid species. Above, we showed that LPS binds to colistin, but not to turnercyclamycins. Moreover, mutations of lipid transporter MlaA, which cycles mislocalized phospholipids out of the OM, caused resistance only to turnercyclamycin B, but not turnercyclamycin A or colistin16. Nonetheless, by multiple measures in this study, we showed that all three agents cause OM damage. Therefore, we sought to determine how exogenous lipids might more broadly impact the activity of the agents.
We first used outer membrane vesicles (OMV) as a proxy for OM lipids and proteins55. We isolated OMVs from a ∆tolA E. coli mutant, which produces high levels of OMVs that contain mostly OM content56,57. A fixed amount (109 mL−1) of OMVs was mixed with the compounds for periods derived from their time-kill kinetics. The resulting compound-OMV mixture was used as a test agent for the preceding antimicrobial assay using A. baumannii as the test pathogen. Both turnercyclamycins exhibited reduced potency after OMV exposure, with MIC90s increased by more than four-fold to >32 μg mL−1 (Fig. 3b and Supplementary Table S3). Colistin retained its potency in the presence of OMVs at the fixed concentration (Fig. 3b and Supplementary Table S3). Because these experiments required extensive amounts of limited compounds, we then switched to using variable OMV concentrations at fixed antibiotic concentrations. OMVs were used at concentrations between 0.16-4.0 × 109 mL−1. Colistin was inactivated in the presence of 3 × 109 mL−1, indicating that at higher concentrations, OMVs blocked colistin (Fig. 3c). By contrast, turnercyclamycins A and B were relatively inactive above an OMV concentration of 0.25 × 109 mL−1 (Fig. 3d), indicating that OMVs much more potently block turnercyclamycins (and especially turnercyclamycin B) in comparison to colistin. Similarly, a previous study showed that E. coli OMVs block colistin action58.
To determine whether turnercyclamycins were directly bound to vesicles, we treated ∆tolA E. coli mutants with both turnercyclamycin analogs and isolated the resulting OMVs. Both turnercyclamycin analogs were associated with the OMVs as determined by mass spectrometry (Fig. 3e). These data suggested that lipids in the OMV bind to turnercyclamycins.
Since turnercyclamycins did not bind to LPS as gauged by ITC, these findings prompted us to investigate the impact of other phospholipids on antibiotic efficacy. Phospholipids cardiolipin (CL) and phosphatidylglycerol (PG) impact the efficacy of colistin59,60,61. Moreover, the lipopeptide brevicidine B also binds to CL and PG, with nanomolar affinity62. Data showing that only turnercyclamycin B was affected by ∆mlaA mutation16 led us to test phosphatidylethanolamine (PE), a major lipid recycled from the OM by MlaA63,64. However, addition of PE to the culture media did not change the MICs of colistin or turnercyclamycins against E. coli C600 (Fig. 3f).
We combined commercially available phospholipids with antibiotics. Some of these phospholipids were pure compounds with defined acyl groups, while others were mixtures isolated from biological sources. We then used the resulting phospholipid/drug mixture in antimicrobial assays against E. coli C600 (Fig. 3f). Neither phosphatidylcholine (PC) nor hexanoylceramide (HC) impacted any compound. CL affected the potency of colistin but not turnercyclamycins, while phosphatidic acid (PA) reduced the effectiveness of both turnercyclamycins but not that of colistin. Interestingly, PG greatly decreased the potency of turnercyclamycin B specifically, with detectable but comparatively minimal impact on colistin. This implies that PG specifically interacts with turnercyclamycin B, which might be related to the sensitivity of the compound to ∆mlaA.
Lipidomic profiles of turnercyclamycin-treated cells resemble those of ∆mlaA mutants
To further define the impact of lipids on turnercyclamycins, we treated E. coli wild type and ∆mlaA cells with the drugs and compared them to untreated controls using a mass spectrometry-based phospholipidomics protocol. Partial least squares discriminant analysis (PLS-DA) clustering resulted in the triplicate replicates for each condition being correlated, reflecting the reproducibility of the experiment. Colistin-treated cells, both wild type and ∆mlaA, clustered separately from the remaining conditions (Fig. 5a). The much different growth and death characteristics of colistin- versus turnercyclamycin-treated cells made it difficult to interpret this difference. However, lipidomics has been performed previously using colistin-treated bacteria, revealing many significant differences reflecting the wide impact of colistin on bacterial lipid homeostasis65,66,67,68.
Fig. 5: Lipopeptide antibiotics alter phospholipid homeostasis differently.a Partial least squares-discriminant analysis (PLS-DA) plot of phospholipid composition of untreated, colistin, turnercyclamycin A, and turnercyclamycin B-treated WT and ∆mlaA E. coli BW25113 taken from analysis by MetaboAnalyst 6.0. A red line divides the colistin treatments from turnercyclamycin treatment and ΔmlaA. Turnercyclamycins A and B treatment is at 8 µg/mL and colistin is at 2 µg/mL. b–f Volcano plot analysis showing significant lipids at a cut off value fold change (FC) = 1.5 and p-value = 0.05. P-value was calculated via MetaboaAnalyst 6.0 statistical package using two-sided t-test. Groups compared are b turnercyclamycin A-treated WT cells vs. turnercyclamycin B-treated WT cells; c turnercyclamycin A-treated ΔmlaA cells vs. turnercyclamycin B-treated ΔmlaA cells; d untreated ΔmlaA cells vs. untreated WT cells; e turnercyclamycin A-treated WT cells vs. untreated ΔmlaA cells; f turnercyclamycin B-treated WT cells vs. untreated ΔmlaA cells.
Remarkably, conditions lacking colistin were more closely clustered in the PLS-DA analysis (Fig. 5a). Among these, turnercyclamycin A- and B-treated WT cells and untreated ΔmlaA cells occupied a relatively similar space, with turnercyclamycin B-treated wild type cells overlapping the untreated ΔmlaA cell cluster, indicative of their strikingly similar lipidomic profiles. Colistin and turnercyclamycins modify lipid homeostasis in different ways, leading to different lipid profiles over the course of treatment. This reflects findings for turnercyclamycins and colistin, which each bind to and are inhibited by a different set of lipids, and each interacts differently with lipid-associated proteins such as MlaA.
Decrease in PG and diacylated phospholipids as a consequence of turnercyclamycin treatment
Upon closer comparison using volcano plots, three key trends were apparent (Fig. 5b–f and Supplementary Tables 4–8). First, the smallest number of significant differences was observed when comparing turnercyclamycin A- and turnercyclamycin B-treated samples, suggesting that both molecules have a similar impact on lipid homeostasis (Fig. 5b, c and Supplementary Tables 4 and 5). Second, in untreated ΔmlaA samples, monoacylated lipids were generally increased, while diacylated lipids were decreased when compared to untreated wild type cells (Fig. 5d and Supplementary Table 6). We speculate that this may be caused by a change in the regulation of lipid biosynthesis; for example, incompletely acylated phospholipids that end up in the OM may no longer be recycled into the periplasm for further synthetic modification. The same trend, although to a lesser degree, was observed when turnercyclamycin-treated wild-type cells were compared with untreated ΔmlaA cells (Fig. 5e, f and Supplementary Tables 7 and 8). Third, PG was significantly decreased in turnercyclamycin-treated wild type cells compared with untreated ΔmlaA cells (Fig. 5e, f and Supplementary Tables 7 and 8), implying that PG might play a direct role in turnercyclamycin potency.
Discussion
The data described here define a general model of turnercyclamycin action (Fig. 6). At early time points, turnercyclamycins penetrate and interact with the OM, (Fig. 6, step 1). Acting on the inner leaflet of the OM or on targets exposed to the periplasm, turnercyclamycins alter phospholipid composition over a series of hours (Fig. 6, step 2). OM damage occurs at late time points (Fig. 6, step 3), leading to cell lysis and death (Fig. 6, step 4). This stands in contrast to colistin, which directly binds to LPS and causes both CM and OM damage, resulting in much faster bacterial killing. Thus, despite similarities in structure and requirement for LPS biosynthesis (both LpxA and LpxC are required), the mechanisms of action of turnercyclamycins and colistin are fundamentally distinct.
Fig. 6: Schematic of cellular changes induced in Gram-negative bacteria after treatment with turnercyclamycins.(Created in BioRender. Lim, A. (2025) https://BioRender.com/sfza2um.).
This general mechanistic model does not define the molecular targets of either colistin or turnercyclamycins, but it suggests some testable hypotheses and eliminates many possibilities. The most important remaining questions include turnercyclamycin entry to the cell, their molecular targets of action, and why is MlaA seemingly important to the action of the turnercyclamycins.
Previous work with colistin suggests that entry is aided by binding to OM LPS, followed by internalization through the membrane4,5,6,25. However, even in depleted or absent OM LPS in P. aeruginosa models, colistin is still very effective25. We exploited the unique growth of Acinetobacter in the absence of OM LPS, an unusual property. A. baylyi LPS transport knockouts retained susceptibility to colistin. Turnercyclamycins acted similarly in LPS assays with LPS transport knockouts. In addition, like colistin, turnercyclamycins caused OM damage, albeit more slowly. Increased NPN and DiSC3(5) fluorescence upon drug treatment, most likely due to increased uptake after OM damage, suggests permeabilization of the OM as a potential common feature of the drugs. Thus, it is likely that turnercyclamycins penetrate the outer membrane in a manner that is similar to that of colistin. An interesting difference is that, in contrast to colistin, turnercyclamycins appear not to directly bind to LPS. The net neutral charge of turnercyclamycins contrasts with the cationic nature of colistin15,16, which is thought to be important in binding to anionic LPS/LPO. This difference implies that another chemical feature is responsible for membrane penetration in the turnercyclamycins. Our hypothesis is that direct interactions with the lipid bilayer drive turnercyclamycin entry. Indeed, turnercyclamycins have a longer lipid chain than colistin, they are more inhibited by OMVs, and they directly bind to OMVs. Alternatively, transport by MlaA or as-yet unknown proteins may be involved. Fluorescent analogs of turnercyclamycins would be useful to differentiate between these possibilities.
The precise molecular targets of colistin are controversial, and those of turnercyclamycins are unknown. However, the phospholipidomics, TEM, and lipid inhibition experiments performed in this study suggest several hypotheses for the turnercyclamycin mode of action. By TEM and fluorescence microscopy, turnercyclamycin caused the most damage to the OM and not the CM, in contrast to colistin. Phospholipidomics of E. coli whole cells revealed a decrease in PG lipids as the major change in turnercyclamycin treatment. PG and LPS are linked both in their biosynthesis and transport. In particular, LpxC and LpxA are at the interface between LPS and fatty acid biosynthesis, directly interacting with AcpP that also interacts with the PG biosynthetic pathway54,69,70,71,72,73. One hypothesis is that turnercyclamycins are acylated AcpP mimics that interact with one or more of the proteins along the LPS or PG pathways. This would explain all the observed features of turnercyclamycins—the dependence upon LPS biosynthesis (blocking late biosynthetic steps is toxic to cells; regulation post early intermediates is important in cell survival), the modification of PG, and the resulting damage likely on the inner leaflet of the OM. In addition, it would explain why MlaA alterations provide similar phospholipidomics phenotypes to turnercyclamycins, since issues with retrograde transport of lipids might resemble OM damage. An alternative hypothesis is that turnercyclamycins impact lipid homeostasis via more direct mechanisms, for example, by directly binding to specific lipids or by blocking lipid transport, both of which are observed properties of turnercyclamycins.
In summary, turnercyclamycins exhibit potent activity against clinical drug-resistant pathogens and have significantly different killing and resistance mechanisms in comparison to those of colistin, despite superficial similarities. Turnercyclamycins A and B themselves also exhibit surprising differences in action, considering that they differ by a very small change in the lipid tails. In the lipopeptide antibiotics, small changes in lipidome interactions lead to profound effects on therapeutic applications, suggesting a path forward to solve some of the major challenges in antibiotic development.
Methods
Bacterial strains
The bacterial strains used in this study are listed in Supplementary Table 1. Transposon mutant strains were generated in a previous study74. All strains were cryopreserved as single-use cultures at −80 °C.
Antibiotics
Colistin (J660915.03) was obtained from Thermo Scientific. Turnercyclamycins A and B were produced by fermentation in the shipworm symbiotic bacterial strain Teredinibacter turnerae T7901 and purified to homogeneity as previously described15. Their chemical identities and high levels of purity were demonstrated using HPLC, mass spectrometry, NMR, and other experiments as previously reported.
Antimicrobial assay (broth microdilution)
Assessment of antimicrobial activity by broth microdilution assay was performed following the Clinical and Laboratory Standards Institute (CLSI) M100 method with some modifications75. Cryopreserved strains were streaked onto Luria-Bertani (LB) agar for E. coli and Mueller–Hinton (MH) agar for A. baylyi, A. baumannii, and P. aeruginosa. The plates were then incubated at 37 °C for 8–12 h. Single colonies were subsequently transferred into LB or MH broth and incubated for 6–8 h at 37 °C with shaking at 150 rpm. For A. baylyi ΔlptD mutants, glycerol stocks were thawed and immediately suspended in M9 medium supplemented with 0.2% succinate and 10 μg/mL kanamycin and incubated for 18–24 h at 37 °C, 150 rpm before assay use. Cultures were adjusted to achieve a density corresponding to the 0.5 McFarland standard (approximately 1 × 108 cells/mL), diluted 200-fold, and used as the inoculum for the assays. Each well of a 96-well plate (Costar® 3474) received 200 µL of the test organisms, followed by the addition of compounds dissolved in 100% DMSO at a two-fold dilution scheme starting at 32 µg/mL, with eight dilutions per compound (for colistin, the solvent is 1× Phosphate buffered saline (PBS) and the highest dilution starts at 8 μg/mL). For E. coli knockouts and A. baylyi mutants, 10 μg/mL kanamycin was added, while for A. baumannii mutants, 10 μg/mL tetracycline was supplemented. The solvent (DMSO and 1× PBS) was maintained at 1%. The plates were then incubated at 37 °C, 150 rpm for 18–20 h. Subsequently, 3-(4,5-dimethylthiazol-2-yl)−2,5-diphenyltetrazolium bromide (MTT, InvitrogenTM M6494, 10 µL of a 5 mg/mL solution in 1× PBS) was added to each well and incubated for 2 h. MTT formazan crystals were solubilized in 100 µL of DMSO for 1 h, and absorbance at 570 nm was measured using a Biotek-Synergy 2 Microplate Reader (Biotek). Each dilution was done in triplicate with three biological replicates unless stated otherwise. Percent inhibition was computed using the formula below:
$$\%{inhibition}=(1-\left(\frac{{{{A}}}_{570}{test}{agent}-\,{{{A}}}_{570}{media}}{{{{A}}}_{570}{solvent}-\,{{{A}}}_{570}{media}}\right)x\,100)$$
Density dependence of drug effectiveness
Density dependence of drug effectiveness was assessed via the previously discussed broth microdilution antimicrobial assay with some modifications. The compound concentrations that were used were at 4× MIC against E. coli C600 (2 μg/mL colistin, 8 μg/mL turnercyclamycin A, 8 μg/mL turnercyclamycin B) and the starting inoculum was set at decreasing OD600 values, with the densest inoculum at OD600 = 0.5.
Time-kill assay
An overnight culture of E. coli C600 was diluted to ~106 CFU/mL. The resulting bacterial suspension was then treated with 4× MIC90 of the compounds (2 μg/mL colistin, 8 μg/mL turnercyclamycin A, 8 μg/mL turnercyclamycin B). They were then incubated for 0.5, 1, 2, 4, 6, 8, 10, 12, 17, and 24 h at 37 °C, 150 rpm, of which 400 μL aliquots were collected for each time point, spun down at 12,000 × g for 2 min and resuspended in sterile 1× PBS. Aliquots (100 µL) of 10-fold serial dilutions were then made and plated on LB agar plates, followed by overnight incubation at 37 °C. The resulting colonies were then counted, and the resulting CFU/mL per experimental treatment was recorded. The experiment was performed in two biological replicates.
Outer membrane vesicle isolation and quantification
E. coli ∆tolA was grown in LB supplemented with 10 μg/mL kanamycin for 18–24 h. Afterward, outer membrane vesicles were isolated using SBI System Biosciences ExoBacteriaTM OMV Isolation Kit (EXOBAC100A-1) following the manufacturer’s protocols using 25 mL of the bacterial suspension at OD600 = 0.8. Isolated OMVs were quantified using the SBI System Biosciences ExoELISA-ULTRA Complete Kit (EXEL-ULTRA-GroEL-1) following the manufacturer’s protocols.
OMV-test compounds co-treatment antimicrobial assay
Various OMV concentrations (1× = 9.62 × 108 particles/mL/well) were combined with various concentrations of the test compounds, followed by incubation at room temperature. For colistin, the incubation time was 2 h and for turnercyclamycin was 6 h. The mixture was used as a test treatment agent for the antimicrobial assay described above with A. baumannii ATCC19606 as the test pathogen.
Turnercyclamycin detection in OMVs via mass spectrometry
An overnight culture of ∆tolA E. coli BW25113 grown in LB broth with 10 μg/mL kanamycin at 37 °C, 150 rpm, was diluted to an OD600 = 0.5. Afterward, 8 μg/mL turnercyclamycin A and B were added to the cell suspension and further incubated at 37 °C, 150 rpm for 20 h. Upon incubation, an OD600 = 0.8 suspension (10 mL) was used for OMV isolation and quantification following protocols described above. The isolated OMVs were lyophilized overnight. The lyophilized solids were extracted with 100 µL of methanol with gentle sonication for 1 h. Samples were centrifuged (13,000 × g) for 10 min to remove particulates, and the sample was diluted 1:10 in methanol for LC-HRMS analysis. MS spectra were collected on a Waters Xevo G2-XS Q-ToF MS with an Acquity UPLC system using an Acquity UPLC BEH C18 column (1.8 µm, 2.1 × 50 mm). The column flow rate was 0.6 µL/min with mobile phase A as water with 0.1% formic acid and mobile phase B as acetonitrile with 0.1% formic acid. The gradient method consisted of a 0.1 min hold at 5% B, a linear gradient to 100% B over 7.4 min, a hold at 100% B for 1 min, a linear gradient back to 5% B over 0.4 min, and a re-equilibration at 5% B for 0.6 min. One microlitre of each sample was injected for analysis.
Membrane depolarization assay
Assessment of membrane depolarization was done following methods adapted from Buttress et al.42. E. coli C600 was grown in LB broth at 37 °C, 150 rpm until it reached an OD600 of 0.3. These were then transferred to 2 mL Eppendorf tubes with perforated lids and test compounds (2 µg/mL colistin, 8 μg/mL turnercyclamycin A, 8 μg/mL turnercyclamycin B) and 1% DMSO was added prior to incubation for 5 min, 55 min and 115 min, at 37 °C, 150 rpm. Once the chosen time point was reached, 1 µM of DiSC3(5) (InvitrogenTM D306) was added to the sample and incubated for 5 min. Five hundred microlitres of the sample was then placed on Teflon-coated multi-spot microscope slides (EprediaTM FIS 9991090) covered with 1.2% agarose in H2O. Imaging was done immediately on a Nikon Automated Widefield CCD Fluorescence Microscope. Resulting raw images were viewed and processed using Nikon NIS-Elements Viewer. Each treatment was tested in duplicates and two biological replicates. Fluorescence microscopy was done at the University of Utah Cell Imaging Core Facility.
Membrane permeability assay
E. coli C600 was cultured in LB broth at 37 °C with shaking at 150 rpm until reaching an OD600 of 0.5. The culture was subsequently centrifuged at 12,000 × g for 5 min and washed twice with 1× PBS. The resulting pellet of cells was then suspended in M9 medium. NPN (ThermoScientific 147160250) dye (10 μM) was then added to the suspension, followed by dispensation in a black, clear-bottom, 96-well plate. Afterwards, 4× MIC (2 μg/mL colistin, 8 μg/mL turnercyclamycin A, 8 μg/mL turnercyclamycin B) of the test compounds was added to the plate. Fluorescence was then read at 350/420 nm, at 2-min intervals for 5 h at 37 °C using a Molecular Devices SpectraMax M5 plate reader. One percent DMSO was used as a solvent control. Each test compound was tested in quadruplicate wells and two biological replicates.
Sytox green assay
E. coli C600 was cultured in LB broth at 37 °C with shaking at 150 rpm until it reached an OD600 of 0.5. The culture was then centrifuged at 12,000 × g for 5 min and washed twice with 1× PBS. The resulting cell pellet was suspended in an M9 medium. Next, Sytox Green (InvitrogenTM S7020) (2 μM) was added to the suspension, which was transferred to a black, clear-bottom 96-well plate. Following this, the plate received test compounds at 4× MIC (2 μg/mL colistin, 8 μg/mL turnercyclamycin A, 8 μg/mL turnercyclamycin B). Fluorescence was measured at 488/523 nm at 2-min intervals over 20 h at 37 °C using a Molecular Devices SpectraMax M5 plate reader, with 1% DMSO serving as the solvent control. Each test compound was tested in quadruplicate wells and two biological replicates.
Transmission electron microscopy (TEM)
WT and ΔlptD A. baylyi were grown in LB broth and M9 medium supplemented with 0.2% succinate and 10 μg/mL kanamycin, respectively at 37 °C, 150 rpm overnight. Afterward, the culture was diluted with the appropriate media to an OD600 = 0.3. Test compounds were at 4× MIC (2 μg/mL colistin, 8 μg/mL turnercyclamycin A, 8 μg/mL turnercyclamycin B). This was then incubated at 37 °C, 150 rpm. Aliquots (2.0 mL) were then collected after 1, 2, 5, and 8 h for colistin-treated cells, 2, 5, 10, and 24 h for turnercyclamycin A-treated cells, and 2, 7, 12, and 24 h for turnercyclamycin B-treated cells. The aliquots were then spun down at 12,000 × g for 15 min and washed thrice with 1× PBS. The resulting cell pellets were resuspended in 1.5 mL of fixative agent (2.5% glutaraldehyde, 1.0% paraformaldehyde). Upon fixation, the samples were repelleted, removing the fixative agent, and rinsed with 1× PBS. They were then postfixed with 2% osmium tetroxide for 1 h, followed by an H2O rinse and en bloc staining for 1 h. After staining, they were dehydrated with multiple washes of ethanol and acetone before infiltration with plastic. The blocks were then cut using a microtome with a diamond edge and placed in TEM grids. Finally, the samples were viewed and imaged using a JEOL JEM-1400 electron microscope. All electron microscopy and sample preparation were done at the University of Utah Electron Microscopy Core Laboratory. Imaging was done on two sets of samples treated with the compounds.
Micro-isothermal titration calorimetry
Micro-ITC used the Malvern MicroCal iTC200 system at 25 °C. Colistin, turnercyclamycin A, and turnercyclamycin B, dissolved in 1× PBS with 1.8% DMSO at a final concentration of 100 µM was placed in the sample cell chamber. LPS (Sigma-Aldrich L2018) dissolved in 1× PBS with 1.8% DMSO at 1000 µM was then placed in the injector and titrated at a rate of 1.8 µL at 3-min intervals. All samples were degassed prior to the start of interaction. The KD values were calculated using MicroCal’s ORIGIN software.
Sample preparation for lipidomic analysis
Wild type E. coli BW25113 and ∆mlaA E. coli strains were grown in LB broth supplemented with 10 μg/mL kanamycin at 37 °C, 150 rpm overnight. The bacterial suspension was adjusted to an OD600 = 0.3. Afterwards, 4× MIC (2 μg/mL colistin, 8 μg/mL turnercyclamycin A, 8 μg/mL turnercyclamycin B) was added to the bacterial suspensions and incubated at 37 °C, 150 rpm. Aliquots amounting to an OD600 = 0.5 was collected for each treated sample at the following time points: untreated, 20 h; turnercyclamycin A and B treated, 20 h; colistin treated, 3, 6, and 12 h. The suspension was pelleted at 12,000 × g for 15 min and washed thrice with 1× PBS.
Sample extraction procedures followed the protocol from Matyash et al.76. Briefly, extraction was performed using a biphasic solvent system comprising cold methanol, methyl tert-butyl ether (MTBE), PBS, and water, with some adjustments. Each sample received 225 µL of methanol (MeOH) containing internal standards, followed by 750 µL of MTBE. The samples were sonicated for 60 s and then kept on ice with periodic vortexing for 1 h. To induce phase separation, 188 µL of PBS was added, followed by a 10-s vortex, a 15-min rest at room temperature, and centrifugation at 15,000 × g for 10 min at 4 °C. The upper organic layer was collected, while the lower aqueous layer was re-extracted using 1 mL of a mixture containing 10:3:2.5 (v/v/v) MTBE/MeOH/dd-H2O. After brief vortexing, incubation at room temperature, and centrifugation under the same conditions, the upper phases were combined and dried in a SpeedVac. The lipid extracts were then reconstituted in 600 µL of 4:1:1 (v/v/v) IPA (isopropyl alcohol)/ACN (acetonitrile)/water and transferred into LC-MS vials for analysis. Simultaneously, a procedural blank sample was prepared, and pooled quality control (QC) samples were generated by combining equal volumes of each sample after final resuspension. The experiment was done on three technical replicates for each sample with one biological replicate.
Mass spectrometry and lipidomic analysis
Lipid extracts were separated using an Acquity UPLC CSH C18 column (2.1 × 100 mm; 1.7 µm) along with an Acquity UPLC CSH C18 VanGuard precolumn (5 × 2.1 mm; 1.7 µm) (Waters, Milford, MA). The system was maintained at 65 °C and linked to an Agilent HiP 1290 Sampler, Agilent 1290 Infinity pump, and an Agilent 6545 Accurate Mass Q-TOF dual AJS-ESI mass spectrometer (Agilent Technologies, Santa Clara, CA). Samples were analyzed in a random order in separate experiments for both positive and negative ionization modes, with a scan range of m/z 100–1700. In positive mode, the source gas temperature was set to 225 °C with a drying gas flow of 11 L/min, a nebulizer pressure of 40 psi, a sheath gas temperature of 350 °C, and a sheath gas flow of 11 L/min. The VCap voltage was 3500 V, the nozzle voltage 500 V, the fragmentor 110 V, the skimmer 85 V, and the octopole RF peak 750 V. In negative mode, the source gas temperature was 300 °C with a drying gas flow of 11 L/min, a nebulizer pressure of 30 psi, a sheath gas temperature of 350 °C, and a sheath gas flow of 11 L/min. The VCap voltage was 3500 V, the nozzle voltage 75 V, the fragmentor 175 V, the skimmer 75 V, and the octopole RF peak 750 V. Mobile phase A was ACN (60:40, v/v) with 10 mM ammonium formate and 0.1% formic acid, while mobile phase B was IPA:ACN (90:9:1, v/v/v) with 10 mM ammonium formate and 0.1% formic acid. For negative mode analysis, the modifiers were switched to 10 mM ammonium acetate. The chromatography gradient for both modes began at 15% mobile phase B, increased to 30% B over 2.4 min, then to 48% B from 2.4 to 3.0 min, then to 82% B from 3 to 13.2 min, then to 99% B from 13.2 to 13.8 min, held until 16.7 min, and then returned to the initial conditions for a 5-min equilibration. The flow rate was 0.4 mL/min throughout, with injection volumes of 5 µL for positive mode and 10 µL for negative mode. Tandem mass spectrometry used iterative exclusion with the same LC gradient and collision energies of 20 V and 27.5 V in positive and negative modes, respectively.
For data processing, Agilent MassHunter (MH) B.07.00 Workstation, along with the MH Qualitative and MH Quantitative software packages, was used. To ensure the reliability of the lipidomics data, pooled QC samples (n = 8) and process blanks (n = 4) were injected throughout the sample queue. Lipid annotation was performed using accurate mass and MS/MS matching with LipidMatch 3.177. Data exported from MH Quantitative was examined in Excel, where initial lipid targets were filtered based on specific criteria. Only lipids with relative standard deviations (RSD) less than 30% in QC samples were included in the data analysis. Additionally, only lipids with background AUC counts in process blanks that were less than 30% of the QC counts were considered. The filtered Excel data tables were then normalized to the ratio of class-specific internal standards, adjusted to tissue mass, and summed prior to statistical analysis. Multivariate analysis was conducted using MetaboAnalyst 6.078. Statistical models were developed for the normalized data following logarithmic transformation (base 10) and Pareto scaling. The initial pass for the volcano plot applied a fold change (FC) cut-off of 1.5, an adjusted p-value cut-off of 0.05. The experiment was done on three replicates of each treatment.
Phospholipid-test compound co-treatment antimicrobial assay
Phospholipids cardiolipin (100 μg/mL) (Sigma CAR-201), phosphatidic acid (100 μg/mL) (Sigma P4013), phosphatidylglycerol (100 μg/mL) (Sigma P6412), phosphatidylcholine (100 μg/mL) (Sigma P3556), phosphatidylethanolamine (50 µg/mL) (European Pharmacopeia Y0001953), and hexanoylceramide (10 μg/mL) (Larodan 56-1049) were combined with varying concentrations of colistin, turnercyclamycin A, and turnercyclamycin B for 1 h. The resulting mixture was used as test compounds for the antimicrobial assay. Afterwards, the broth microdilution antimicrobial assay proceeded as described in the “Antimicrobial assay (broth microdilution)” section of “Methods” above, using E. coli C600 as the test pathogen. The experiment was done on two biological replicates.
Quantification and statistical analysis
MICs were plotted using GraphPad Prism 10.1.2. Error bars indicate mean ± SD of three replicate wells, with experiments conducted in either two or three biological replicates. In-panel text and axes were adjusted for readability in figures using Adobe Illustrator 2025.
Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.
Data availability
The DiSC3(5) microscopy data generated in this study have been deposited in FigShare (https://doi.org/10.6084/m9.figshare.30865277). Lipidomics data are available at the NIH Common Fund’s National Metabolomics Data Repository (NMDR)79 website, the Metabolomics Workbench, https://www.metabolomicsworkbench.org where it has been assigned Study ID ST004466. The data can be accessed directly via its Project https://doi.org/10.21228/M84Z8T. This work is supported by NIH grant U2C-DK119886 and OT2-OD030544 grants. All other data are available in the article and its Supplementary files. Source data are provided with this paper.
References
Naghavi, M. et al. Global burden of bacterial antimicrobial resistance 1990–2021: a systematic analysis with forecasts to 2050. Lancet 404, 1199–1226 (2024).
WHO, https://www.who.int/teams/maternal-newborn-child-adolescent-health-and-ageing/newborn-health/newborn-infections (2025).
Wen, S. C. H. et al. Gram-negative neonatal sepsis in low- and lower-middle-income countries and WHO empirical antibiotic recommendations: a systematic review and meta-analysis. PLoS Med. 18, e1003787 (2021).
Andrade, F. F., Silva, D., Rodrigues, A. & Pina-Vaz, C. Colistin update on its mechanism of action and resistance, present and future challenges. Microorganisms 8, 1716 (2020).
Trimble, M. J., Mlynárčik, P., Kolář, M. & Hancock, R. E. Polymyxin: alternative mechanisms of action and resistance. Cold Spring Harb. Perspect. Med. 6, https://doi.org/10.1101/cshperspect.a025288 (2016).
Ledger, E. V. K., Sabnis, A. & Edwards, A. M. Polymyxin and lipopeptide antibiotics: membrane-targeting drugs of last resort. Microbiology 168, https://doi.org/10.1099/mic.0.001136 (2022).
Lepak Alexander, J., Wang, W. & Andes David, R. Pharmacodynamic evaluation of MRX-8, a novel polymyxin, in the neutropenic mouse thigh and lung infection models against gram-negative pathogens. Antimicrob. Agents Chemother. 64, e01517–e01520 (2020).
Duncan Leonard, R., Wang, W. & Sader Helio, S. In vitro potency and spectrum of the novel polymyxin MRX-8 tested against clinical isolates of gram-negative bacteria. Antimicrob. Agents Chemother. 66, e00139–00122 (2022).
Bruss, J. et al. Single- and multiple-ascending-dose study of the safety, tolerability, and pharmacokinetics of the polymyxin derivative SPR206. Antimicrob. Agents Chemother. 65, 00739–00721 (2021).
Rodvold, K. A., Bader, J., Bruss, J. B. & Hamed, K. Pharmacokinetics of SPR206 in plasma, pulmonary epithelial lining fluid, and alveolar macrophages following intravenous administration to healthy adult subjects. Antimicrob. Agents Chemother. 67, e0042623 (2023).
Zurawski, D. V. et al. SPR741, an antibiotic adjuvant, potentiates the in vitro and in vivo activity of rifampin against clinically relevant extensively drug-resistant Acinetobacter baumannii. Antimicrob. Agents Chemother. 61, https://doi.org/10.1128/aac.01239-17 (2017).
Götze, S. et al. Structure, biosynthesis, and biological activity of the cyclic lipopeptide anikasin. ACS Chem. Biol. 12, 2498–2502 (2017).
Yamanaka, K. et al. Direct cloning and refactoring of a silent lipopeptide biosynthetic gene cluster yields the antibiotic taromycin A. Proc. Natl. Acad. Sci. USA 111, 1957–1962 (2014).
Bekiesch, P. et al. Viennamycins: lipopeptides produced by a Streptomyces sp. J. Nat. Prod. 83, 2381–2389 (2020).
Miller, B. W. et al. Shipworm symbiosis ecology-guided discovery of an antibiotic that kills colistin-resistant Acinetobacter. Cell Chem. Biol. 28, 1628–1637.e1624 (2021).
Lim, A. L. et al. Resistance mechanisms for Gram-negative bacteria-specific lipopeptides, turnercyclamycins, differ from that of colistin. Microbiol. Spectr. 11, e02306–e02323 (2023).
Simpson, B. W. et al. Acinetobacter baumannii can survive with an outer membrane lacking lipooligosaccharide due to structural support from elongasome peptidoglycan synthesis. mBio 12, e03099–03021 (2021).
Huang, H. W. DAPTOMYCIN, its membrane-active mechanism vs. that of other antimicrobial peptides. Biochim. Biophys. Acta Biomembr. 1862, 183395 (2020).
Benfield, A. H. & Henriques, S. T. Mode-of-action of antimicrobial peptides: membrane disruption vs. intracellular mechanisms. Front. Med. Technol. 2, https://doi.org/10.3389/fmedt.2020.610997 (2020).
Li, Q. et al. Outer-membrane-acting peptides and lipid II-targeting antibiotics cooperatively kill Gram-negative pathogens. Commun. Biol. 4, 31 (2021).
Stephani, J. C., Gerhards, L., Khairalla, B., Solov’yov, I. A. & Brand, I. How do antimicrobial peptides interact with the outer membrane of Gram-negative bacteria? Role of lipopolysaccharides in peptide binding, anchoring, and penetration. ACS Infect. Dis. 10, 763–778 (2024).
Epand, R. M., Walker, C., Epand, R. F. & Magarvey, N. A. Molecular mechanisms of membrane targeting antibiotics. Biochim. Biophys. Acta Biomembr. 1858, 980–987 (2016).
Zou, W. et al. Exploring the active core of a novel antimicrobial peptide, palustrin-2LTb, from the Kuatun frog, Hylarana latouchii, using a bioinformatics-directed approach. Comput. Struct. Biotechnol. J. 20, 6192–6205 (2022).
Giacometti, A., Cirioni, O., Greganti, G., Quarta, M. & Scalise, G. In vitro activities of membrane-active peptides against Gram-positive and Gram-negative aerobic bacteria. Antimicrob. Agents Chemother. 42, 3320–3324 (1998).
Sabnis, A. et al. Colistin kills bacteria by targeting lipopolysaccharide in the cytoplasmic membrane. eLife 10, e65836 (2021).
Cerny, G. & Teuber, M. Differential release of periplasmic versus cytoplasmic enzymes from Escherichia coli B by polymyxin B. Arch. Mikrobiol. 78, 166–179 (1971).
Schindler, P. R. G. & Teuber, M. Action of polymyxin B on bacterial membranes: morphological changes in the cytoplasm and in the outer membrane of Salmonella typhimurium and Escherichia coli B. Antimicrob. Agents Chemother. 8, 95–104 (1975).
Nakajima, K. & Kawamata, J. Studies on the mechanism of action of colistin. IV. Activation of “latent” ribonuclease in Escherichia coli by colistin. Biken J. 9, 115–123 (1966).
Teuber, M. Precipitation of ribosomes from E. coli B by polymyxin B. Naturwissenschaften 54, 71 (1967).
McCoy, L. S. et al. Polymyxins and analogues bind to ribosomal RNA and interfere with eukaryotic translation in vitro. Chembiochem 14, 2083–2086 (2013).
Saugar, J. M. et al. Activities of polymyxin B and cecropin A-melittin peptide CA(1-8)M(1-18) against a multiresistant strain of Acinetobacter baumannii. Antimicrob. Agents Chemother. 46, 875–878 (2002).
Dwyer, D. J., Kohanski, M. A., Hayete, B. & Collins, J. J. Gyrase inhibitors induce an oxidative damage cellular death pathway in Escherichia coli. Mol. Syst. Biol. 3, 91 (2007).
Brochmann, R. P. et al. Bactericidal effect of colistin on planktonic Pseudomonas aeruginosa is independent of hydroxyl radical formation. Int. J. Antimicrob. Agents 43, 140–147 (2014).
Bailey, J. et al. Essential gene deletions producing gigantic bacteria. PLoS Genet. 15, e1008195 (2019).
Elliott, K. T. & Neidle, E. L. Acinetobacter baylyi ADP1: transforming the choice of model organism. IUBMB Life 63, 1075–1080 (2011).
Machovina, M. M. et al. Enabling microbial syringol conversion through structure-guided protein engineering. Proc. Natl. Acad. Sci. USA 116, 13970–13976 (2019).
Xu, F. et al. Critical role of 3’-downstream region of pmrB in polymyxin resistance in Escherichia coli BL21(DE3). Microorganisms 9, https://doi.org/10.3390/microorganisms9030655 (2021).
Helander, I. M. & Mattila-Sandholm, T. Fluorometric assessment of gram-negative bacterial permeabilization. J. Appl Microbiol. 88, 213–219 (2000).
Muheim, C. et al. Increasing the permeability of Escherichia coli using MAC13243. Sci. Rep. 7, 17629 (2017).
Mood, E. H. et al. Antibiotic potentiation in multidrug-resistant Gram-negative pathogenic bacteria by a synthetic peptidomimetic. ACS Infect. Dis. 7, 2152–2163 (2021).
Heesterbeek, D. A. C. et al. Complement-dependent outer membrane perturbation sensitizes Gram-negative bacteria to Gram-positive specific antibiotics. Sci. Rep. 9, 3074 (2019).
Buttress, J. A. et al. A guide for membrane potential measurements in Gram-negative bacteria using voltage-sensitive dyes. Microbiology 168, https://doi.org/10.1099/mic.0.001227 (2022).
Boix-Lemonche, G., Lekka, M. & Skerlavaj, B. A rapid fluorescence-based microplate assay to investigate the interaction of membrane active antimicrobial peptides with whole Gram-positive bacteria. Antibiotics 9, https://doi.org/10.3390/antibiotics9020092 (2020).
Strahl, H. & Hamoen, L. W. Membrane potential is important for bacterial cell division. Proc. Natl. Acad. Sci. USA 107, 12281–12286 (2010).
Howe, J. et al. Thermodynamic analysis of the lipopolysaccharide-dependent resistance of Gram-negative bacteria against polymyxin B. Biophys. J. 92, 2796–2805 (2007).
Wouters, C. L. et al. Breaking membrane barriers to neutralize E. coli and K. pneumoniae virulence with PEGylated branched polyethylenimine. Biochim. Biophys. Acta Biomembr. 1865, 184172 (2023).
Srimal, S., Surolia, N., Balasubramanian, S. & Surolia, A. Titration calorimetric studies to elucidate the specificity of the interactions of polymyxin B with lipopolysaccharides and lipid A. Biochem. J. 315, 679–686 (1996).
Botte, M. et al. Cryo-EM structures of a LptDE transporter in complex with Pro-macrobodies offer insight into lipopolysaccharide translocation. Nat. Commun. 13, 1826 (2022).
Sherman, D. J. et al. Lipopolysaccharide is transported to the cell surface by a membrane-to-membrane protein bridge. Science 359, 798–801 (2018).
Bos, M. P., Tefsen, B., Geurtsen, J. & Tommassen, J. Identification of an outer membrane protein required for the transport of lipopolysaccharide to the bacterial cell surface. Proc. Natl. Acad. Sci. USA 101, 9417–9422 (2004).
Putker, F., Bos, M. P. & Tommassen, J. Transport of lipopolysaccharide to the Gram-negative bacterial cell surface. FEMS Microbiol. Rev. 39, 985–1002 (2015).
Wu, T. et al. Identification of a protein complex that assembles lipopolysaccharide in the outer membrane of Escherichia coli. Proc. Natl. Acad. Sci. USA 103, 11754–11759 (2006).
Anderson, M. S., Bulawa, C. E. & Raetz, C. R. The biosynthesis of gram-negative endotoxin. Formation of lipid A precursors from UDP-GlcNAc in extracts of Escherichia coli. J. Biol. Chem. 260, 15536–15541 (1985).
Williams, A. H. & Raetz, C. R. H. Structural basis for the acyl chain selectivity and mechanism of UDP-N-acetylglucosamine acyltransferase. Proc. Natl. Acad. Sci. USA 104, 13543–13550 (2007).
Lieberman, L. A. Outer membrane vesicles: a bacterial-derived vaccination system. Front. Microbiol. 13, https://doi.org/10.3389/fmicb.2022.1029146 (2022).
Reimer, S. L. et al. Comparative analysis of outer membrane vesicle isolation methods with an Escherichia coli tolA mutant reveals a hypervesiculating phenotype with outer-inner membrane vesicle content. Front. Microbiol. 12, 628801 (2021).
Bernadac, A., Gavioli, M., Lazzaroni, J. C. & Raina, S. & Lloubès, R. Escherichia coli tol-pal mutants form outer membrane vesicles. J. Bacteriol. 180, 4872–4878 (1998).
Li, X. et al. The attenuated protective effect of outer membrane vesicles produced by a mcr-1 positive strain on colistin sensitive Escherichia coli. Front. Cell Infect. Microbiol. 11, 701625 (2021).
Few, A. V. The interaction of polymyxin E with bacterial and other lipids. Biochim. Biophys. Acta 16, 137–145 (1955).
Araki, T., Hiroguchi, J., Matsunaga, H. & Kimura, Y. Potentiating effect of lysophosphatidylcholine on antibacterial activity of polymyxin antibiotics. Chem. Pharm. Bull. 33, 1729–1733 (1985).
Imai, M., Inoue, K. & Nojima, S. Effect of polymyxin B on liposomal membranes derived from Escherichia coli lipids. Biochim. Biophys. Acta 375, 130–137 (1975).
Zhao, X. et al. Elucidating the mechanism of action of the Gram-negative-pathogen-selective cyclic antimicrobial lipopeptide brevicidine. Antimicrob. Agents Chemother. 67, e00010–e00023 (2023).
Giordano Nicole, P., Cian Melina, B. & Dalebroux Zachary, D. Outer membrane lipid secretion and the innate immune response to Gram-negative bacteria. Infect. Immun. 88, https://doi.org/10.1128/iai.00920-19 (2020).
Coudray, N. et al. Structure of bacterial phospholipid transporter MlaFEDB with substrate bound. eLife 9, e62518 (2020).
Han, M.-L. et al. Comparative metabolomics and transcriptomics reveal multiple pathways associated with polymyxin killing in Pseudomonas aeruginosa. mSystems 4, 00149–00118 (2019). 10.1128/msystems.
Mahamad Maifiah, M. H. et al. Integrated metabolomic and transcriptomic analyses of the synergistic effect of polymyxin–rifampicin combination against Pseudomonas aeruginosa. J. Biomed. Sci. 29, 89 (2022).
Tao, Y. et al. Colistin treatment affects lipid composition of Acinetobacter baumannii. Antibiotics 10, 528 (2021).
Zhu, Y. et al. Polymyxins bind to the cell surface of unculturable Acinetobacter baumannii and cause unique dependent resistance. Adv. Sci. 7, 2000704 (2020).
Wyckoff, T. J. O. & Raetz, C. R. H. The active site of Escherichia coli UDP-N-acetylglucosamine acyltransferase: chemical modification and site-directed mutagenesis. J. Biol. Chem. 274, 27047–27055 (1999).
Mohan, S., Kelly, T. M., Eveland, S. S., Raetz, C. R. & Anderson, M. S. An Escherichia coli gene (FabZ) encoding (3R)-hydroxymyristoyl acyl carrier protein dehydrase. Relation to fabA and suppression of mutations in lipid A biosynthesis. J. Biol. Chem. 269, 32896–32903 (1994).
Zhang, Y.-M. & Rock, C. O. in (eds Ridgway, N. D. & McLeod, R. S.) Biochemistry of Lipids, Lipoproteins and Membranes, Sixth Edition, 73–112 (Elsevier, 2016).
Bertani, B. & Ruiz, N. Function and biogenesis of lipopolysaccharides. EcoSal Plus 8, https://doi.org/10.1128/ecosalplus.ESP-0001-2018 (2018).
Thomanek, N. et al. Intricate crosstalk between lipopolysaccharide, phospholipid and fatty acid metabolism in Escherichia coli modulates proteolysis of LpxC. Front. Microbiol. 9, 2018 (2019).
Gallagher, L. A. et al. Resources for genetic and genomic analysis of emerging pathogen Acinetobacter baumannii. J. Bacteriol. 197, 2027–2035 (2015).
Lewis II, J. S. et al. CLSI M100-Ed35 Performance Standards for Antimicrobial Susceptibility Testing Standards. 35th Edition, Clinical and Laboratory Standards Institute (2025).
Matyash, V., Liebisch, G., Kurzchalia, T. V., Shevchenko, A. & Schwudke, D. Lipid extraction by methyl-tert-butyl ether for high-throughput lipidomics. J. Lipid Res. 49, 1137–1146 (2008).
Koelmel, J. P. et al. LipidMatch: an automated workflow for rule-based lipid identification using untargeted high-resolution tandem mass spectrometry data. BMC Bioinform. 18, 331 (2017).
Xia, J., Sinelnikov, I. V., Han, B. & Wishart, D. S. MetaboAnalyst 3.0—making metabolomics more meaningful. Nucleic Acids Res. 43, W251–W257 (2015).
Sud, M. et al. Metabolomics Workbench: an international repository for metabolomics data and metadata, metabolite standards, protocols, tutorials and training, and analysis tools. Nucleic Acids Res. 4, D463–D470 (2016).
Acknowledgements
We thank Colin Manoil and Jeannie Bailey, from the University of Washington, for providing us with wild type and knockout mutants of both Acinetobacter baumannii 5075UW and Acinetobacter baylyi ADP1. We would like to thank Matthew Mulvey, from the University of Utah, for supplying E. coli C600. We would also like to thank Basil Mathew of the University of Utah Department of Pathology for worthwhile discussions. We are grateful for the assistance of John Alan Maschek of University of Utah Metabolomics Core for the lipidomic analysis, Xiang Wang of the University of Utah Cell Imaging Core for the fluorescence microscopy experiments, and to Nancy Chandler and David Belnap of the University of Utah Electron Microscopy Core for the transmission electron microscopy experiments. We would like to thank Amy Barrios of the University of Utah for the use of her plate reader for fluorescence spectroscopy experiments. We would also like to acknowledge the help of Debbie Eckert and Michael Kay for assistance in the isothermal calorimetry experiments. Funding for this project was provided by the National Institute of Allergy and Infectious Disease, National Institutes of Health (NIH R01AI162943). Albebson Lim was also funded by the University of Utah L.S. Skaggs Graduate Student Fellowship provided by the Skaggs Institute for Research.
Author information
Authors and Affiliations
Department of Medicinal Chemistry, University of Utah, Salt Lake City, UT, USA
Albebson L. Lim, Bailey W. Miller, Margo G. Haygood & Eric W. Schmidt
Department of Pathology and ARUP Laboratories, University of Utah, Salt Lake City, UT, USA
Mark A. Fisher
Department of Pharmacology and Toxicology, University of Utah, Salt Lake City, UT, USA
Louis R. Barrows
Authors
- Albebson L. Lim
- Bailey W. Miller
- Mark A. Fisher
- Margo G. Haygood
- Louis R. Barrows
- Eric W. Schmidt
Contributions
A.L.L. and E.W.S. conceptualized and designed the study. A.L.L. and B.W.M. performed experiments and analyzed data. A.L.L., B.W.M., and E.W.S. wrote the manuscript. A.L.L., B.W.M., M.A.F., M.G.H., L.R.B., and E.W.S. reviewed and edited the manuscript. E.W.S. and A.L.L. acquired funding for the project.
Corresponding author
Correspondence to Eric W. Schmidt.
Ethics declarations
Competing interests
The authors declare the following competing interests: E.W.S., B.W.M., and M.G.H. have submitted a patent application for turnercyclamycins. All other authors declare no competing interests.
Peer review
Peer review information
Nature Communications thanks the anonymous reviewers for their contribution to the peer review of this work. A peer review file is available.
Additional information
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.
About this article
Cite this article
Lim, A.L., Miller, B.W., Fisher, M.A. et al. Differential membrane lipid disruption by lipopeptide antibiotics, colistin and turnercyclamycins. Nat Commun 17, 1880 (2026). https://doi.org/10.1038/s41467-026-68681-0
Received: 14 June 2025
Accepted: 13 January 2026
Published: 19 January 2026
Version of record: 20 February 2026
DOI: https://doi.org/10.1038/s41467-026-68681-0