Peptide MSI‐1 inhibited MCR‐1 and regulated outer membrane vesicles to combat immune evasion of Escherichia coli

Abstract

Polymyxin resistance is conferred by MCR‐1 (mobile colistin resistance 1)‐induced lipopolysaccharide (LPS) modification of G⁻ bacteria. However, the peptide MSI‐1 exerts potent antimicrobial activity against mcr‐1‐carrying bacteria. To further investigate the potential role of MCR‐1 in improving bacterial virulence and facilitating immune evasion, and the immunomodulatory effect of peptide MSI‐1, we first explored outer membrane vesicle (OMV) alterations of mcr‐1‐carrying bacteria in the presence and absence of sub‐MIC MSI‐1, and host immune activation during bacterial infection and OMV stimulation. Our results demonstrated that LPS remodelling induced by MCR‐1 negatively affected OMV formation and protein cargo by E. coli. In addition, MCR‐1 diminished LPS‐stimulated pyroptosis but facilitated mitochondrial dysfunction, further aggravating apoptosis in macrophages induced by OMVs of E. coli. Similarly, TLR4‐mediated NF‐κB activation was markedly alleviated once LPS was modified by MCR‐1. However, peptide MSI‐1 at the sub‐MIC level inhibited the expression of MCR‐1, further partly rescuing OMV alteration and attenuation of immune responses in the presence of MCR‐1 during both infection and OMV stimulation, which can be exploited for anti‐infective therapy.

1. LPS remodeling induced by MCR‐1 negatively affected outer membrane vesicle (OMV) formation and protein cargo by E coli.2. MCR‐1 diminished pyroptosis, but facilitated mitochondrial dysfunction during OMV stimulation.3. Antimicrobial peptide MSI‐1 at the sub‐MIC level inhibited the expression of MCR‐1, further partly rescuing OMV alteration and attenuation of immune responses during both OMV stimulation and bacterial infection.

INTRODUCTION

At present, the increasing prevalence of antibiotic resistance has aroused more and more public concern since the 1990s. Polymyxins, one of cationic lipopeptide antibiotics, were first discovered in 1947 and are currently used as last‐resort therapies against multidrug‐resistant Gram‐negative (G⁻) bacteria by selectively binding to bacterial membranes via electrostatic interactions and efficiently killing microbial pathogens (Yahav et al., 2012). Unfortunately, the emergence and rapid spread of mcr‐1 gene inexorably threatened the effectiveness of polymyxins as the ‘last line of refuge’ against G⁻ bacteria because of their resistance to all currently available antibiotics. The plasmid‐mediated gene mcr‐1 was first described in Enterobacteriaceae in China (Liu et al., 2016), which assisted rapid dissemination of polymyxin resistance via horizontal gene transfer. Recently, polymyxin resistance conferred by mcr‐1 has already been detected in diverse species, including but not limited to turkeys, cattle, pigs and humans (Kieffer et al., 2017; Ovejero et al., 2017). MCR‐1 is a phosphoethanolamine (pEtN) transferase that modifies the lipid A moiety of lipopolysaccharide (LPS) with phosphatidylethanolamine, which reduces negative charge of LPS and enables bacterial pathogens to develop resistance to polymyxin (Son et al., 2019).

LPS is mainly anchored on the outer membrane (OM) of G⁻ bacteria, of which the properties affect the membrane fluidity and curvature, OM budding and consequent biogenesis of outer membrane vesicles (Henderson et al., 2016; Li et al., 2012; McMahon & Gallop, 2005). Outer membrane vesicles (OMVs), derived from the OM of almost all types of Gram‐negative (G⁻) bacteria under normal and stress conditions (Tashiro et al., 2011), are 20–250 nm in diameter and involved in encapsulation and transportation of proteins, nucleic acids, toxins and other virulence factors in addition to LPS (Kulp & Kuehn, 2010), thus giving them an important role in bacteria‐bacteria and bacteria‐host interactions, including biofilm formation, pathogenesis and the activation of host immune responses. Recently, Sinha et al. reported that the addition of pEtN to LPS regulated by PmrC negatively affected the spontaneous biogenesis of liposomal OMVs produced by Citrobacter rodentium (Sinha et al., 2019). Additionally, PagL activity and penta‐acylation of LPS have been confirmed to increase the size and number of OMVs in Salmonella species (Elhenawy et al., 2016). Therefore, in the present study, we first explored the OMV alterations while LPS is modified by MCR‐1, which has been reported to be implicated in polymyxin resistance.

As a typical subset of pathogen‐associated molecular patterns (PAMPs), LPS is responsible to evoke extremely acute immunostimulatory effects and an overwhelming release of cytokines including interleukin (IL)‐6 or tumour necrosis factor α (TNF‐α) in mammalian organisms by aggregating with LPS‐binding protein (LBP) and subsequently binding to toll‐like receptor 4 (TLR4). Alternatively, LPS associated with OMVs can be translocated into the host cytosol and trigger pyroptosis by activating the noncanonical caspase 4/11‐dependent inflammasome (Shi et al., 2014; Vanaja et al., 2016) through the lipid A moiety. Nevertheless, modifications of the lipid A component of LPS in G⁻ bacteria such as hydroxylation (Bartholomew et al., 2019), and PagP‐dependent palmitoylation (Bishop, 2005) as a pathogenic strategy have evolved to avoid innate immune signalling and assist intracellular bacterial survival. For instance, Bartholomew et al. revealed that lipid A 2‐hydroxylation controlled by LpxO in Acinetobacter baumannii contributed to the decreased recognition mediated via TLR4‐myeloid differentiation factor 2 (MD2) receptor complex and potentially increased ability of immune evasion (Bartholomew et al., 2019). Weakly stimulatory tetra‐acylated lipid A was produced by Yersinia pestis in a temperature‐dependent remodelling manner and eventually resulted in invasion of host innate immune defences in the early stage of infection (Montminy et al., 2006). It is plausible that structural modification of lipid A attenuated the host inflammatory response, thus hindering either recognition of LPS by TLR4/MD2/CD14 complex or intracellular detection by inflammatory caspases and finally contributing to virulence. Herein, we further explored the immunostimulatory activity of pEtN‐modified LPS by LBP binding and OMV delivery.

As one of effective antimicrobial candidates, antimicrobial peptide (AMP) contributed to form the first line of host immune system and had broad antimicrobial activity against infections such as bacteria, fungi and virus (Diamond et al., 2009). To our knowledge, most AMPs were cationic, and the electrostatic interactions of AMPs with negatively charged LPS of G⁻ bacteria, lipoteichoic acid (LTA) of Gram‐positive (G⁺) bacteria, and with lipid bilayers, have been proposed to enhance AMP selectivity and stability, thereby resulting in membrane permeability. In our previous study, a novel cationic AMP, MSI‐1, was designed by truncation and amino acid substitution based on magainin 2, which displayed potent and broad‐spectrum antimicrobial activity against drug‐resistant bacteria and fungi, including penicillin‐resistant E. coli, methicillin‐resistant S. aureus and fluconazole‐resistant C. albicans (Ma, et al., 2020a; Ma et al., 2020b; Ye et al., 2021). We have clearly demonstrated the electrostatic interactions of AMPs with negatively charged LPS of G⁻ bacteria, which have been proposed to enhance AMP selectivity and stability towards biological membranes, thereby resulting in membrane permeability. Consequently, we hypothesized that MSI‐1 at sub‐MIC level potentially also affects biogenesis and biological functions of secreted OMVs by disturbing bacterial membranes and causing stress condition.

To investigate this hypothesis, we first evaluated the biogenesis and immunostimulatory activity of OMVs by mcr‐1‐carrying E. coli in the presence and absence of sub‐MIC MSI‐1. Our results suggested that the pEtN decoration of LPS induced by MCR‐1 negatively affected the biogenesis of OMV formation by E. coli and caspase 11‐dependent effector responses upon OMV stimulation. Owing to the key role of OMVs in modulating immune response, MCR‐1‐induced OMV alteration presented here provided a novel mechanism for MCR‐1 function in facilitating immune evasion of G⁻ infections. However, exposure to MSI‐1 at sub‐MIC level could promote OMV formation by mcr‐1‐carrying E. coli, further increasing caspase 11‐dependent pyroptosis activation and attenuating mitochondrial dysfunction in macrophages stimulated by OMVs derived from mcr‐1‐carrying E. coli, which can be exploited for anti‐infective therapy.

EXPERIMENTAL PROCEDURES

Reagents

Peptide MSI‐1 (purity > 98%) were synthesized by GL Biochem Co., Ltd. Polymyxin B sulphate (CAS1405‐20‐5; purity 99%) and Cytoplasmic and Mitochondrial Protein Extraction Kit were purchased from Sangon Biotech Co., Ltd. Isopropyl β‐D‐Thiogalactoside (IPTG) and 1,6‐diphenyl‐1,3,5‐hexatirene (DPH) were purchased from Aladdin Biochem Technology Co., Ltd. Specific TLR4 inhibitor Resatorvid (TAK₂₄₂; HY‐11109), caspase‐11 inhibitor wedelolactone (Wed; HY‐N0551), JC‐1 Assay Kit and Cytotoxicity LDH Assay Kit were purchased from MCE Co., Ltd. 1,1′‐dioctadecyl‐3,3,3′,3′‐tetramethylindocarbocyanine perchlorate (Dil), DAPI, actin‐tracker green, LysoTracker red and 2, 7‐dichlorofluorescein diacetate (DCFH‐DA) were purchased from Beyotime. TNF‐α, IL‐6, IL‐1β and IL‐18 ELISA Kits were purchased from MultiSciences Biotech Co., Ltd.

Animals

Approximately six‐week‐old BALB/c female mice were purchased from the Laboratory Animal Center of Yangzhou University (Yangzhou, China) and housed under standard conditions of light and temperature. Mice had free access to standard laboratory and water prior to and in the duration of experiment. All procedures involving animals were approved by approved by Science and Technology Department of Jiangsu Province (Approval code: SYXK 2018‐0019).

Bacterial strains and cells

The recombinant plasmid pET28a‐mcr‐1 (GenBank accession no. KY283125.1) was constructed by Genewiz, Inc. and transformed into E. coli BL21 (designated as BL21/pMCR‐1). Meanwhile, E. coli BL21 transformed with pET28a‐vector was termed as BL21/pEmpty and served as negative control (Figure S1). BL21/pMCR‐1 was grown in LB broth (1% tryptone, 0.5% yeast extract, 0.5% NaCl, pH 7.2–7.4) with 50 μg/mL kanamycin and treated with 0.5 mM IPTG to induce MCR‐1 expression at 37°C with constant shaking (220 rpm).

THP‐1 cells (American Type Culture Collection, ATCC TIB‐202™) and RAW 264.7 cells (ATCC TIB‐71™) were maintained with RPMI‐1640 (containing 0.05 mM of 2‐mercaptoethanol) or DMEM medium supplemented with 10% heat‐inactivated foetal bovine serum (Gibico) at 37°C with 5% CO₂. For all experiments, THP‐1 cells were differentiated into macrophage cells by addition of 10 ng/mL phorbol myristate acetate for 48 h.

For infection, the overnight cultures of bacteria at logarithmic phase were harvested and washed two to three times in cold PBS by centrifugation and resuspended in serum‐free culture medium supplemented with kanamycin and IPTG. Differentiated THP‐1 cells and murine macrophage cell line RAW 267.4 cells cultivated directly on 24‐well plates were washed and infections were performed at an multiplicity of infection (MOI) of 50 bacteria per cell. For LPS and OMV stimulation, cells seeded in 24‐well plates were challenged with purified LPS (1 μg/mL) or isolated OMVs (20 μg protein/mL) for 4 h at 37°C in a 5% CO₂ incubator. To specifically inhibit TLR4, the cellular ligand for LPS, infected or LPS‐stimulated cells were pre‐treated with TAK₂₄₂ (100 nM) for 1 h.

Antimicrobial activity of MSI‐1 against BL21/pMCR‐1 in vitro

The IZDs, MIC and minimum bactericidal concentration (MBC) values of MSI‐1 and polymyxin B were determined by disk diffusion assay following our protocol described elsewhere (Ma et al., 2020b) and microdilution method according to Clinical and Laboratory Standards Institute susceptibility testing protocol (Brown‐Elliott et al., 2012), respectively. In addition, the time‐killing kinetics assay was performed by sub‐culturing bacterial suspension (1 × 10⁵ CFU/mL) into culture tubes supplemented with IPTG in the presence or absence of MSI‐1 or polymyxin B at 4 × MIC at 37°C. At fixed time intervals, bacterial enumeration was performed by agar plate dilution method. Bacterial growth assay of BL21/pEmpty, BL21/pMCR‐1 with or without 2 μg/mL of MSI‐1 were determined in LB medium supplemented with kanamycin and IPTG and OD_{600 nm} were measured at the indicated time points.

OMV isolation and quantification

OMVs were obtained from bacterial culture supernatants as described previously (Vanaja et al., 2016). Briefly, bacterial cultures were grown in LB broth at 37°C for 8 h to achieve abundant OMV production. Subsequently, supernatants of BL21/pEmpty or BL21/pMCR‐1 at early stationary phase were obtained by centrifuging and filtering to remove residual cells. The sterile filtrates were subsequently concentrated by ultrafiltration using a 100‐kDa Amicon centrifugal filters (Millipore), and the collected retentates were transferred to ultracentrifuge tubes and ultracentrifuged at 150,000 g for 3 h using a 90 Ti rotor (Beckman). The vesicle‐free supernatant fraction was collected, and the vesicle pellet was washed and resuspended in sterile PBS for use or stored at −80°C for further analysis. Each OMV sample was visualized by transmission electron microscope (TEM) according to standard procedures, of which particle size distribution and zeta potential were determined using a Nano ZS instrument (Malvern), the protein and LPS contents were quantified by BCA assay and Limulus Amebocyte lysate assay, respectively. The protein and LPS productions of OMVs were normalized by dividing by corresponding OD_600nm value of each culture per millilitre, and these values were further divided by the OMV production of BL21/pEmpty to give rise to relative folds of OMV productions.

Proteomics analysis of OMVs

The protein profiles of OMVs were resolved by 12% sodium dodecyl sulfate‐polyacrylamide gel electrophoresis (SDS‐PAGE) and gels were stained with Coomassie brilliant blue R‐250. After SDS‐PAGE analysis, total proteins of each OMV sample (20 μg) were added with 10 mM DTT and incubated at 56°C for 1 h to reduce disulphide bonds. For the alkylation reaction, each sample solution mixed with 50 mM iodoacetamide was incubated at room temperature in the dark for 40 min. Subsequently, acetone was added to each sample (6:1, v/v). The samples were frozen at −20°C overnight and then centrifuged at 8000 g for 10 min. The obtained precipitates were reconstituted in 100 μL of 50 mM NH₄HCO₃ and digested into peptides using trypsin (1 μg). The tryptic peptides were vacuum‐dried for LC–MS/MS analysis.

The peptides were separated by Easy‐nLC 1200 system (Thermo Fisher Scientific). Samples were injected onto a C18 reversed‐phase nano‐LC column (Acclaim PepMap, 100 Å, 150 μm × 15 cm). The flow rate was set as 600 nL/min, and the injection volume was 5 μL. LC linear gradient elution was conducted using mobile phase A composed of 0.1% (v/v) formic acid and mobile phase B composed of 80% (v/v) acetonitrile and 0.1% (v/v) formic acid. Then the eluted components were analysed by MS/MS spectrometry on a Q Exactive™ Hybrid Quadrupole‐Orbitrap™ Mass Spectrometer (Thermo Fisher Scientific). Mass spectrometry proteomics data were analysed and searched against a E. coli BL21 (DE3) protein sequence database obtained from UniPort (www.uniprot.org). As for bioinformatics analysis of detected proteins, pSORTb web servers were used to predict the subcellular location (Yu et al., 2010). The virulent proteins in proteomes were assessed by VirulentPred (Garg & Gupta, 2008). The predicted gene ontology (GO) terms of obtained differentially expressed proteins were classified according to the COG annotation terms (www.ncbi.nlm.nih.gov/COG).

OMVs phagocytosis assays

To monitor internalization of OMVs in THP‐1 cells, the isolated OMVs were incubated with Dil, one of lipophilic membrane dyes, at the final concentration of 25 μM for 30 min at 37°C. Subsequently, the Dil‐labelled OMVs were pelleted and the supernatants containing unbound dye were washed out by ultracentrifugation (150,000 g, 2 h). THP‐1 cells were plated onto glass coverslips and exposed to labelled OMVs (20 μg protein/mL) for 4 h in the dark prior to fixation in 4% polyformaldehyde and permeabilization with 0.1% Triton X‐100. Nuclei were labelled with DAPI and cytoskeletons were visualized using actin‐tracker green. Fluorescence images were taken using LSM as above.

Flow cytometry

THP‐1 cells were seeded at a concentration of 4 × 10⁵ cells/mL and subsequently stimulated with OMVs derived from BL21/pEmpty, BL21/pMCR‐1 with or without extra MSI‐1 for 4 h as described above. For evaluation of apoptotic cell death, cells challenged with OMVs were harvested and stained with Annexin V‐FITC and PI in Annexin V‐FITC binding buffer for 15 min. Moreover, mitochondrial rupture was confirmed visually using a Transwell insert approach and detected by flow cytometry using a JC‐1 Assay Kit. THP‐1 cells stimulated with OMVs were incubated with JC‐1 dyes or DCFH‐DA to observe changes in the mitochondrial membrane potential (MMP) and intracellular ROS production, respectively. Above samples were further measured by flow cytometry (BD Biosciences).

Isolation of LPS and lipid A and phenotypic characterization of BL21/pMCR‐1 strain

LPS derived from BL21/pEmpty or BL21/pMCR‐1 were extracted using the hot phenol‐water microextraction procedure (Galanos et al., 1969). The lyophilized LPS were further labelled with FITC to determine the interaction between LPS and LBP by microscale thermophoresis (MST) assay. Briefly, FITC‐labelled LPS was incubated with LBP (ranged from 100 to 0.00305 μg/mL with 2‐fold dilution) for 5 min in an assay buffer containing 20 mM Tris (pH 8.0) and 150 mM NaCl. Subsequently, the samples were loaded into hydrophilic glass capillaries and detected using a Monolith NT.115 instrument (NanoTemper Technologies) to calculate the values of EC₅₀ (half maximal effective concentration). Moreover, LPS solutions with varying concentrations (from 1000 to 0.0305 μg/mL) were incubated with isovolumetric FITC‐MSI‐1 (16 μg/mL) and EC₅₀ values were also estimated by MST assay, which enabled us to determine the binding characteristics of peptide MSI‐1 to LPS derived from BL21/pEmpty or BL21/pMCR‐1.

In order to confirm the lipid A structure derived from BL21/pMCR‐1, lipid A species were isolated from LPS of BL21/pMCR‐1 and BL21/pEmpty by mild acetate hydrolysis as described earlier with slight modification (El Hamidi et al., 2005). Briefly, purified LPS from BL21/pMCR‐1 or BL21/pEmpty were hydrolysed with 9 mL acetate buffer (50 mM, pH 4.5) containing 1% SDS for 1 h at 100°C. The resulting hydrolysate was further extracted with 20 mL of 1:1 chloroform/methanol mixture and the lower phase containing lipid A sample was dried. Lipid A sample solubilized in chloroform/methanol (2:1, v/v) was mixed with an equal portion of norharmane MALDI matrix, loaded onto a MALDI plate and dried at room temperature. The full spectrum of the lipid A isolated from each strain was determined by matrix‐assisted laser desorption ionization–time of flight mass spectrometry (MALDI‐TOF MS) using Bruker UltrafleXtreme MALDI‐TOF MS instrument, equipped with a nitrogen laser (335 nm) in positive linear mode (Bruker). Data acquisition and analysis were performed using flexAnalysis software.

Membrane fluidity assay

Membrane fluidity of BL21/pEmpty, BL21/pMCR‐1 with or without extra MSI‐1 was determined using DPH. Briefly, bacterial cultures were grown as described above to an OD₆₀₀ of 0.3 in LB broth media in the presence or absence of MSI‐1 and washed in PBS. Bacterial suspensions were subsequently incubated with DPH dissolved in tetrahydrofuran for 30 min at 37°C. The sample containing tetrahydrofuran was considered as negative control. The unbounded fluorescent probe was washed out by centrifugation and the fluorescence polarization was measured using a spectrofluorometer at an excitation and emission wavelength of 360 and 426 nm, respectively. The degree of polarization was calculated and the relative change in membrane fluidity was determined as described previously (Mykytczuk et al., 2007).

RNA‐seq analysis

Total RNA was extracted from differentiated THP‐1 cells infected with BL21/pMCR‐1 or BL21/pEmpty. After quality checking and polyA selection, mRNA samples for each of 3 biological replicates were subjected to library generation and sequenced on the Illumina Solexa GAIIx sequencer (Illumina). RNA‐seq assay was performed with three independent RNA samples for each condition. The threshold set for differentially expressed genes (DEGs) was defined as fold change > 2 and FDR < 0.01. Hierarchical Clustering analysis was performed to show the distinguishable expression patterns among samples. Signalling pathway enrichment of these DEGs was conducted using KEGG analysis (www.kegg.jp/). All data analysis and visualization of DEGs were conducted using R v.3.6.1 (r‐project.org).

qPCR experiments

For quantitative RT‐PCR analysis, infected THP‐1 or RAW 264.7 cells were collected for RNA isolation using TRIzol Reagent (Vazyme) according to the manufacturer's standard protocol. Bacterial total RNA was extracted using RNeasy columns (Qiagen). The extracted RNA was then reversely transcribed to cDNA using the HiScript® II Reverse Transcriptase (Vazyme) and the cDNA was used for the qRT‐PCR analysis. The thermocycling regimen was 95°C for 30 s followed by 40 cycles of 95°C for 10 s, 60°C for 30 s. Primers of the genes of interest are listed in Table S2 and the expression level of target gene was normalized to the level of 16S rRNA expression for bacteria or GAPDH for mammalian cell samples using the comparative C _T method (2^−ΔΔCT).

Western blot analysis

Proteins in macrophages lysates were collected and diluted in 5 × SDS‐PAGE sample buffer. Mitochondrial and cytosolic subcellular fractions were separated using a Cytoplasmic and Mitochondrial Protein Extraction Kit. Bacterial suspensions of BL21/pMCR‐1 induced by IPTG were collected and the supernatants contained soluble MCR‐1 were obtained by sonicating. After BCA assay, protein samples were loaded on 10% or 12% SDS‐PAGE gels and underwent routine western blot. The primary antibodies were listed in Table S3. Finally, the blots were probed with corresponding secondary antibody (1:3000) for visualization by chemiluminescence and quantitation using a Bio‐Rad ChemiDoc™ MP imaging system.

ELISA and LDH (lactate dehydrogenase) assay

Conditioned media were collected at the indicated time points and assessed for LDH released using a Cytotoxicity LDH Assay Kit. The contents of IL‐6, TNF‐α, IL‐1β or IL‐18 in culture supernatants or sera from mice were quantified using ELISA kit.

Mouse in vivo experiments

Inflammasome activation by OMVs in vivo

Ten mice per group received intraperitoneal injections with isolated OMVs (100 μg proteins) from BL21/pEmpty or BL21/pMCR‐1. Mice injected with PBS were served as negative control. Cytokine levels in the plasma were assessed at 4 h post OMV injection. In addition, three mice per group were sacrificed and lung samples were excised and stored at −80°C for western blot analysis, or fixed in 4% paraformaldehyde for haematoxylin and eosin (H&E) staining or histological analysis. For survival experiments, mice were observed for 7 days after OMV stimulation, and a survival curve was plotted. The weights and temperature alterations of the animals in each group were also recorded.

Bacteraemia model of infection

In the bacteraemia model, groups of 12 mice received intraperitoneal injections of IPTG as described above. An overnight bacterial culture of BL21/pMCR‐1 or BL21/pEmpty was diluted to a density of 1 × 10⁹ CFU/mL and used to infect mice to establish bacteraemia model. Following treatment with MSI‐1 (single dose of 5 mg/kg) or PBS (negative control), five mice in each group were sacrificed after 4 h of infection, of which the liver, spleen, kidney and lung samples were collected and homogenized in sterile PBS for a CFU assay on MacConkey agar plates to evaluate bacterial translocation. Survival of remaining mice was recorded daily for 7 days.

Statistical analysis

For statistical analysis, all data are presented as means ± SD (standard deviation) and representative of three independent experiments with the exception of proteomic data and RNA‐seq data described above. The p‐values were calculated using unpaired two‐way Student test and statistical significance was defined as a p‐value <0.05.

RESULTS

OMV formation of E. coli was dampened by MCR‐1 and promoted by peptide MSI‐1

MCR‐1 can confer polymyxin resistance. Confirmedly, the expression of MCR‐1 decreased the sensitivity of bacteria to polymyxin B, with minimum inhibitory concentration (MIC) value of polymyxin B against BL21/pMCR‐1 increased 8‐fold (Table S1) but inhibition zone diameter (IZD) decreased to 13.2 mm (Figure 1A), respectively, when compared to that against BL21/pEmpty. Importantly, peptide MSI‐1 displayed a potent bacteriostatic effect on both BL21/pEmpty and BL21/pMCR‐1 (Table S1). In the disk diffusion assay, these two strains were found susceptible to MSI‐1 with IZDs of 15.7 and 15.0 mm, respectively (Figure 1A). Besides, dominant zones were also exhibited by two clinical polymyxin‐resistant isolates (Figure S2), demonstrating the efficiency of MSI‐1 against mcr‐1‐carrying bacteria. Moreover, the outcome of time‐kill kinetics showed that MSI‐1 eradicated all bacteria of BL21/pMCR‐1 in 1 h at 16 μg/mL, whereas polymyxin B did so after 4 h of incubation at 16 μg/mL (Figure 1B). Interestingly, there is growing evidence that resistance to polymyxins conferred by MCR‐1 in E. coli is due to its pEtN transferase activity in the process of covalent modification of LPS, which is a major constituent in OMVs. It has been reported that OMVs are bacterial defensive weapons that combine with environmental stressors such as antimicrobial agents (Maredia et al., 2012) and the chemical modification of LPS has been implicated in vesicle formation (Elhenawy et al., 2016). Here, we explored OMV production in mcr‐1‐carrying bacteria with or without MSI‐1 at the sub‐MIC level. First, BL21/pEmpty and BL21/pMCR‐1 were grown under similar conditions in the absence or presence of 1/2 MIC of MSI‐1. BL21/pEmpty grew more rapid than BL21/pMCR‐1, and the growth rate of the latter did not visibly differ from that grew in the presence of MSI‐1 (Figure S3A). Therefore, OMVs derived from BL21/pEmpty, BL21/pMCR‐1 and BL21/pMCR‐1 grew under sub‐MIC MSI‐1 (designated as sBL21/pMCR‐1) were isolated from normalized cell‐free supernatants of early stationary growth phase cultures and visualized by TEM. As shown in Figure 1C, dynamic light scattering revealed that BL21/pEmpty‐OMVs and BL21/pMCR‐1‐OMVs revealed characteristic double‐membrane spherical buds with average diameters of 43.43 nm (PdI = 0.252) and 39.58 nm (PdI = 0.187). Thus, OMVs formed by the BL21/pMCR‐1 strain were comparable to that of BL21/pEmpty in size. However, the average diameter of the sBL21/pMCR‐1‐OMVs was 131.8 nm (PdI = 0.380), revealing the effect of MSI‐1 on OMV morphology and particle size released from BL21/pMCR‐1. Zeta potentials of vesicles shown in Table S4 revealed high stability of each sample. Interestingly, there was considerable variability in protein and LPS contents in OMVs from three strains; that is, BL21/pMCR‐1 exhibited 36.9% and 50.7% decreases in protein and LPS amounts associated with OMVs with respect to BL21/pEmpty (**p < 0.01), while MSI‐1 treatment apparently resulted in 2.7‐fold and 1.1‐fold increases in the levels of proteins and LPS enriched in OMVs compared to that of BL21/pMCR‐1 without MSI‐1 treatment (Figure 1D). The amount of LPS was also reduced in BL21/pMCR‐1‐OMVs following standardization with the protein amounts (Figure S4B). Additionally, the markedly decreased membrane fluidity (***p < 0.001) and mRNA expression of vesiculation‐related genes indicated that the presence of MCR‐1 presumably affected bacterial membrane deformation and repressed vesicle budding from E. coli (Figure 1E,F). In this case, MSI‐1, which acts as a cell membrane‐interrupting AMP (Ma et al., 2020b), marginally increased the membrane fluidity of BL21/pMCR‐1 at the sub‐MIC level, thus being capable of stimulating OMV budding. Overall, these data supported the hypothesis that the process of vesiculation is negatively affected by MCR‐1 in E. coli but promoted by peptide MSI‐1 at the sub‐MIC level (Figure 1G).

FIGURE 1.

OMV formation of E. coli was dampened by MCR‐1 and promoted by peptide MSI‐1. (A) Disk diffusion assay against BL21/pEmpty, BL21/pMCR‐1 for the bacteriostatic activities of MSI‐1. 1: normal saline; 2: MSI‐1 (5 μg); 3: polymyxin B (5 μg). (B) Time‐killing kinetics of MSI‐1 and polymyxin against BL21/pMCR‐1. (C) Transmission electron micrograph (upper) and corresponding size distribution (bottom) of isolated OMVs from BL21/pEmpty, BL21/pMCR‐1 and sBL21/pMCR‐1 strains. Scale Bar: 100 nm. Dynamic light scattering was applied to analyse the size distribution by volume. (D) Protein (upper) and LPS (bottom) fold production associated with isolated OMVs from OD_600nm‐normalized cultures of BL21/pEmpty, BL21/pMCR‐1 and sBL21/pMCR‐1 strains by BCA assay and LAL assay. (E) The membrane fluidity of the BL21/pEmpty, BL21/pMCR‐1 and sBL21/pMCR‐1 strains was determined using a TMA‐DPH probe and presented as % fluidity compared to the fluorescence polarization of BL21/pEmpty. (F) Expression changes of vesiculation‐specific genes in three strains by qRT‐PCR. (G) Putative model of OMVs formation in the presence of MCR‐1. Data are representative of three independent experiments. The data are reported as the mean ± SD. ns, not significant; *p < 0.05; **p < 0.01; *** p < 0.001.

Peptide MSI‐1 decreased protein constituents, especially many established virulent proteins, enriched in OMVs of BL21/pMCR‐1

To further assess whether the protein contents packaged in OMVs were affected by MCR‐1, the vesicle samples were separated by 12% SDS‐PAGE and visualized by Coomassie blue staining, which displayed different protein contents in the three types of OMVs (Figure 2A). The fractionated constituents of OMVs were subjected to in‐gel tryptic digestion for further LC–MS/MS analysis, which could identify and quantify the protein contents of OMVs. The number of identified proteins is summarized in Table S5, with a total of 74, 134 and 36 proteins with different sources. According to in silico analysis results of subcellular localization by the bioinformation tool PSORTb v3.0.3 (https://www.psort.org/psortb/), the cytoplasm proteins accounted for a large proportion of BL21/pEmpty‐OMVs (77.0% of 57). The number of encoded cytoplasmic proteins packaged in BL21/pMCR‐1‐OMVs was increased (82.1% of 110). In addition, the general enrichment of OM components in these three types of samples further confirmed the OM origin of OMVs (Figure S3C). Furthermore, VirulentPred (http://bioinfo.icgeb.res.in/virulent/) was used to screen out the vesicle proteins that were predicted to be high scoring virulent proteins in Table S5.

FIGURE 2.

Peptide MSI‐1 decreased protein constituents, especially many established virulent proteins, enriched in OMVs of BL21/pMCR‐1. (A) SDS‐PAGE results of protein profiles in OMVs from BL21/pEmpty, BL21/pMCR‐1 and sBL21/pMCR‐1 strains. (B, C) Samples were separated by 12% SDS‐PAGE and visualized by Coomassie blue staining. The content of key virulent proteins (B) and MCR‐1 (C) in BL21/pEmpty‐OMVs, BL21/pMCR‐1‐OMVs and sBL21/pMCR‐1‐OMVs by label‐free quantitative proteomics. (D) Venn graph of differentially expressed proteins (DEPs) overlapping between BL21/pEmpty‐OMVs and BL21/pMCR‐1‐OMVs (D, left) and the numbers of upregulated and downregulated DEPs from BL21/pMCR‐1‐OMVs compared with BL21/pEmpty‐OMVs (D, right). (E) The DEPs between BL21/pEmpty‐OMVs and BL21/pMCR‐1‐OMVs were categorized based on the related biological process, cellular component and molecular function. (F) Venn graph of DEPs overlapping between BL21/pMCR‐1‐OMVs and sBL21/pMCR‐1‐OMVs (F, left) and the numbers of upregulated (red) and downregulated (green) DEPs from BL21/pMCR‐1‐OMVs and BL21/pMCR‐1‐OMVs compared with sBL21/pMCR‐1‐OMVs (F, right). (G) The DEPs between BL21/pMCR‐1‐OMVs and sBL21/pMCR‐1‐OMVs were categorized based on the gene ontology terms biological process, cellular component and molecular function.

Subsequently, to obtain the detailed composition of differentially expressed proteins (DEPs), of which the abundances between BL21/pEmpty‐OMVs and BL21/pMCR‐1‐OMVs were more than a twofold intensity difference, we identified 117 upregulated proteins and 49 downregulated proteins in BL21/pMCR‐1‐OMVs versus BL21/pEmpty‐OMVs in Table S6. Of these, only 28 proteins were common to both BL21/pMCR‐1‐OMVs and BL21/pEmpty‐OMVs, 106 were unique to BL21/pMCR‐1‐OMVs and 49 were unique to BL21/pEmpty‐OMVs only (Figure 2D). Afterwards, we performed GO analysis to categorize DEPs with respect to their biological processes, cellular processes and molecular functions (Figure 2E). For biological process, translation, primary metabolic process, organonitrogen compound metabolic process, metabolic process and cellular protein metabolic process were involved. For cellular component and molecular function, cytoplasmic part and protein binding dominated, respectively. Among these DEPs, many of the major OM‐ and OM‐anchored proteins were abundant in BL21/pMCR‐1‐OMVs but not in BL21/pEmpty‐OMVs, and several lipoproteins were inimitably harboured in OMVs derived from BL21/pMCR‐1, including Lpp (Nguyen & Götz, 2016) and BamC in the Bam complex (Hussain & Bernstein, 2018), and the σ^E regulon YraP responded to cell envelope stress (Onufryk et al., 2005). Virulent proteins were also incorporated into BL21/pMCR‐1‐OMVs, such as periplasmic protein TolB, which is responsible for cell division in the Tol‐Pal system (Li et al., 2022). These specific constituents in addition to LPS in OMVs might accelerate the process of cell death during infection and OMV stimulation. However, the lipoprotein RcsF which is related to Rcs stress, and the OM protein OmpX were abolished in BL21/pMCR‐1‐OMVs (Figure 2B). Subsequently, we compared the OMV constituents from BL21/pMCR‐1 grown in the presence or absence of sub‐MIC MSI‐1 (BL21/pMCR‐1‐OMVs versus sBL21/pMCR‐1‐OMVs). Considering that reports have described misfolded protein enrichment accompanied by antibiotic‐stimulated vesiculation (Schwechheimer & Kuehn, 2015), it is conceivable that the amount of MCR‐1 in sBL21/pMCR‐1‐OMVs was greater than that in BL21/pMCR‐1‐OMVs in our results (Figure 2C). However, the number of detected proteins, especially many of the established virulent proteins, drastically decreased (Figure 2B,F). Results of GO analysis showed that proteins associated with translation and primary metabolic process were also abundant in the biological process perspective. The cellular component for the DEPs included ribosomal subunit, cytoplasmic part, macromolecular complex, outer membrane and other cell component. Further analysis revealed that proteins associated with structural function of ribosome were most abundant in the molecular function group (Figure 2G). Herein, we surmised that the alteration of vesicle productions or virulence factors might be attributed to the changes in the composition of the OM and underlying cytoplasmic membrane in response to environmental stimuli. However, more specific evidences are necessary to relate these proteins with alterative abundance to OMVs formation and virulence in our further study.

Peptide MSI‐1 increased caspase 11‐dependent pyroptosis activation in macrophages stimulated by BL21/pMCR‐1‐OMVs

As a well‐characterized PAMP, extracellular LPS is recognized by the TLR4/MD2 receptor complex and initiates inflammatory responses while cytoplasmic LPS evokes caspase 4/11 activation and a robust signal cascade. Therefore, chemical modification of LPS possibly affected host immune activation during bacterial infection and OMV stimulation (Akira et al., 2006; Kayagaki et al., 2013). Herein, we explored the effects of LPS modification on intracellular inflammatory sensing by OMV delivery. To investigate whether OMVs from BL21/pMCR‐1 might facilitate the delivery of LPS and other proteins into cultured human cells, we carried out confocal microscopy analysis to observe the internalization of OMVs labelled by fluorescein via endocytic pathways in differentiated macrophage‐like THP‐1 cells. The bright fluorescent dot appeared in the cytoplasm indicated that OMVs (20 μg protein/mL) from BL21/pEmpty, BL21/pMCR‐1, sBL21/pMCR‐1 all could be internalized and accumulated in THP‐1 cells after 4 h of co‐incubation, and the uptake did not show obvious differences despite the different origin of OMVs (Figure 3A). However, the internalized OMVs derived from BL21/pMCR‐1 failed to induce pyroptosis and subsequent caspase‐1 activation (Figure 3B). To further define the role of modified LPS associated with OMVs in caspase 11 inflammasome activation, THP‐1 cells and RAW 264.7 cells were pre‐treated with DMSO or Wed for 2 h prior to stimulating with three types of isolated OMVs or isometric vesicle‐free supernatant. Cytosolic LPS sensing by OMV delivery was sufficient to induce caspase 1 downstream signalling events and the release of mature IL‐1β as well as IL‐18 in both the BL21/pEmpty‐OMV‐ and sBL21/pMCR‐1‐OMV‐treated groups, but not in the BL21/pMCR‐1‐OMV‐treated group (Figure 3C–E). Notably, modified LPS in OMVs strongly attenuated the proteolytic processing of pro‐caspase 1 and GSDMD in THP‐1 cells. However, the results of sBL21/pMCR‐1‐OMV‐treated groups suggested that peptide MSI‐1 treatment in the process of OMV formation abolished the suppressive effects of intracellular inflammatory sensing induced by BL21/pMCR‐1‐OMVs, which was highly specific to noncanonical caspase 11‐dependent inflammasome pathway because macrophage cell death and caspase‐1 processing were both markedly reduced in the presence of Wed. Regarding RAW 264.7 cells, we found a similar decrease in inflammasome responses through treatment with BL21/pMCR‐1‐OMVs (Figure S4A,B). Consistently, western blot analysis confirmed that LPS modification mediated by MCR‐1 diminished LPS‐induced caspase 11 activation, NT‐GSDMD (p31) formation and proteolytic IL‐1β maturation (p17) in OMVs stimulated RAW 264.7 cells (Figure S4C). Taken together, our data demonstrated that peptide MSI‐1 was capable of increasing caspase 11‐dependent pyroptosis activation in macrophages stimulated by BL21/pMCR‐1‐OMVs.

FIGURE 3.

Peptide MSI‐1 increased caspase 11‐dependent pyroptosis activation in macrophages stimulated by BL21/pMCR‐1‐OMVs. (A) Confocal microscopy images of THP‐1 stimulated with OMVs (20 μg protein/mL) pre‐labelled with Dil (red) for 4 h. The nucleus and cytoskeleton were stained with DAPI (blue) and Actin‐tracker (green). Scale bars, 10 μm. (B) LDH were released from THP‐1 cells treated with DMSO or Wed (50 μM) prior to OMV stimulation. The cell death rate was presented as % cytotoxicity relative to the LDH released from the 0.1% Triton X‐100‐treated group. (C, D) The released IL‐1β (C) and IL‐18 (D) from THP‐1 cells treated with DMSO or Wed (50 μM) prior to OMV stimulation. (E) Cell lysates of THP‐1 cells treated with DMSO or Wed prior to OMV stimulation were analysed via western blot. Relative expression levels of caspase 1 (p10) and N‐Terminal GSDMD are shown below each blot. Data are representative of three independent experiments. The data are reported as the mean ± SD. ns, not significant; *p < 0.05; **p < 0.01; *** p < 0.001.

Peptide MSI‐1 diminished mitochondrial dysfunction, further blocking apoptosis in macrophages stimulated by BL21/pMCR‐1‐OMVs

Since TLR4 activation in response to cytoplasmic recognition of LPS was implicated in mitochondrial reactive oxygen species (ROS) generation and more severe cell apoptosis (West et al., 2011), the relationship between intracellular sensing of LPS by OMV delivery and mitochondrial damage is not well understood. Herein, we first detected the apoptotic proportion of immune cells under the stimulation of OMVs from different sources. As shown in Figure 4A, following stimulation with BL21/pMCR‐1‐OMVs for 4 h, the percentage of PI⁺ THP‐1 cells remarkably increased to 21.77% compared to the control group. In contrast, OMVs from BL21/pEmpty and sBL21/pMCR‐1 at the same concentrations (20 μg protein/mL) did not obviously induce apoptosis. We then sought to evaluate the extent of mitochondrial dysfunction while THP‐1 cells were stimulated with secreted OMVs without direct cell–cell contact using a Transwell insert approach (Sutaria et al., 2017). Cells were plated in 12‐well plates containing coverslips at a density of ∼4 × 10⁵ cells/mL in the presence of BL21/pEmpty, BL21/pMCR‐1 and sBL21/pMCR‐1 (MOI = 50) in the top chamber of Transwell filters. As can be seen in Figure 4B, normal cells from negative control group exhibited red fluorescence (JC‐1 aggregates) due to the integrity of the mitochondrial membrane and the existence of a high membrane potential gradient, while BL21/pMCR‐1‐OMVs obviously prevented the accumulation of JC‐1 in the mitochondria and showed dispersed green fluorescence (JC‐1 monomers) after 4 h of incubation. Subsequently, isolated OMVs were directly added for 4 h and intracellular ROS production of stimulated THP‐1 cells were detected by flow cytometry. Consistent with the morphological characteristics observed in confocal experiments, exposure to BL21/pMCR‐1‐OMVs did result in the disruption of MMP, as the JC‐1 monomer/aggregate ratio was 1.75, while the values were 0.06 and 0.29 in the BL21/pEmpty‐OMV‐ and sBL21/pMCR‐1‐OMV‐treated groups, respectively (Figure S4D). Besides, results showed a concomitant increase of intracellular ROS generation following BL21/pMCR‐1‐OMV stimulation (Figure 4C), which indicated that internalized OMV of BL21/pMCR‐1 tended to augment ROS production and mitochondrial damage. Next, we considered other hallmarks of mitochondrial dysfunction and found a decreased abundance of anti‐apoptotic BCL‐2 proteins, along with significantly increased release of cytochrome C into the cytosol and enhanced expression of BCL‐2 associated agonist of cell death (BAD) induced by BL21/pMCR‐1‐OMVs (Figure 4D). Taken together, peptide MSI‐1 at the sub‐MIC level partly rescued mitochondrial dysfunction and aggravated apoptosis induced by OMVs from BL21/pMCR‐1.

FIGURE 4.

Peptide MSI‐1 diminished mitochondrial dysfunction, further blocking apoptosis in macrophages stimulated by BL21/pMCR‐1‐OMVs. (A) Detection of THP‐1 apoptotic cells treated with OMVs for 4 h. (B) THP‐1 cells were seeded in the bottom chamber of a Transwell plate. Following differentiation for 48 h, fresh RPMI 1640 was added. Bacteria suspended in RPMI 1640 without serum were inoculated in the top chamber of 0.4‐μm Transwell filters to separate them from THP‐1 cells. After 4 h of coculture, bacteria were removed thoroughly, and cells were stained with JC‐1 and visualized by confocal microscopy. Scale bars, 10 μm. (C) ROS production of OMVs‐treated THP‐1 cells were determined by staining of DCFH‐DA and flow cytometry analysis. (D) Western blot analysis of BCL‐2, cytochrome c (CYTO C), and BAD in cell extracts (left) and the level of CYTO C in the cytosolic fraction (right) of THP‐1 cells treated with OMVs as indicated. β‐Actin and COX IV are shown as cell and mitochondrial protein controls, respectively. Relative expression level of CYTO C are shown below corresponding blots. Data are representative of three independent experiments.

MCR‐1 inhibited OMV‐associated systemic inflammatory responses during LPS shock

Subsequently, the virulence potential of BL21/pMCR‐1‐OMVs was further investigated in vivo. In lethal sepsis caused by OMVs from BL21/pMCR‐1 or BL21/pEmpty (Figure 5A), the survival rate of both OMV‐stimulated groups was only 14% at 72 h after injection. And the inoculation of mice with BL21/pMCR‐1‐OMVs (100 μg proteins) resulted in comparative weight loss and temperature loss at 24 h post‐injection (Figure 5B,C). However, histopathology images similarly revealed less infiltration of IL‐1β positive cells in mice injected with BL21/pMCR‐1‐OMVs. H&E staining showed that OMVs from BL21/pMCR‐1 caused weaker inflammation than OMVs from BL21/pEmpty, as evidenced by marked infiltration of neutrophils into the alveolar space, alveolar septal thickening and vascular congestion in the lungs of mice injected with BL21/pEmpty‐OMVs but not BL21/pMCR‐1‐OMVs (Figure 5D). Consistent with our in vitro data, the plasma levels of IL‐1β and IL‐18 release in the BL21/pMCR‐1‐OMV‐stimulated group were markedly blunted relative to the group stimulated with BL21/pEmpty‐OMVs (Figure 5E). In line with decreased cytokine release, the activation of the noncanonical cytosolic LPS sensing pathway in the lung tissues was also significantly decreased in BL21/pMCR‐1‐OMV‐stimulated mice (Figure 5F). Furthermore, the systemic inflammatory responses evoked by BL21/pMCR‐1‐OMVs was markedly attenuated supporting our hypothesis that MCR‐1‐mediated modification diminished LPS‐stimulated pyroptosis by OMV delivery in vitro and in vivo, which might facilitate bacterial immune evasion of G⁻ infections.

FIGURE 5.

MCR‐1 inhibited OMV‐associated systemic inflammatory responses during LPS shock. (A) The survival of mice, which received intraperitoneal injections of OMVs (100 μg) from BL21/pEmpty or BL21/pMCR‐1. n = 7 for each group. (B, C) Mice received intraperitoneally injection with each indicated OMVs (100 μg) or PBS and the total body weight (B) and temperature (C) were recorded over time. n = 5 for each group. (D) Serum IL‐1β (left) and IL‐18 (right) contents in mice after injected with OMVs or PBS for 4 h. n = 5 for each group. (E) Haematoxylin and eosin (H&E) staining (upper) and immunohistochemistry (IHC) staining of IL‐1β (bottom) in mouse lung sections stimulated with each indicated OMV at 4 h. Hyperaemia and inflammatory cell infiltration in H&E images and positive cells in IHC images are marked by black arrows. Scale bar, 100 μm. (F) Western blot analysis of different proteins in homogenized mouse lung tissues. Relative expression levels of caspase 1 (p10) and N‐Terminal GSDMD are shown below each blot. The data are reported as the mean ± SD. ns, not significant; *p < 0.05; ***p < 0.001.

MCR‐1 in E. coli attenuated the immune response of macrophages during bacterial infection

Having demonstrated a role of modified LPS in caspase 4/11 signalling activation, we further examined the complete transcriptomic response of THP‐1 cells during infection with either BL21/pMCR‐1 or BL21/pEmpty to gain greater insight into the effect of MCR‐1 on the immune response of the infected host. In brief, 390 DEGs were identified (differential regulation was defined as a fold change > 2 and FDR < 0.01) in the BL21/pMCR‐1‐infected group with respect to the BL21/pEmpty‐infected group, with 180 upregulated and 210 downregulated genes, respectively (Figure 6A). The detailed information of various genes in the heatmap was provided in Table S7. And KEGG analysis showed the top 20 notable pathways enriched with these DEGs, including the TNF and NF‐κB signalling pathways, cytokine‐cytokine receptor interaction (Figure 6B). Differences in RNA‐seq between the BL21/pMCR‐1‐infected group and the BL21/pEmpty‐infected groups were further confirmed by qRT‐PCR analysis. To this end, 11 randomly selected DEGs were analysed at the mRNA level, which presented a similar differential expression trend compared with the RNA‐seq data (Figure 6C). The expression levels of partial inflammatory‐related genes in NF‐κB signalling pathways, including ccl4, il‐1β, cd14 and cd40, from the BL21/pMCR‐1‐infected group were significantly lower than those from the BL21/pEmpty‐infected group. An obviously increased expression of CAMK2β was observed in the BL21/pMCR‐1‐infected group, which is involved in the negative modulation of IFN signalling and represents a potential virulence strategy of BL21/pMCR‐1 to evade immune recognition and promote intracellular survival (Alphonse et al., 2022). However, further research is needed to elucidate underlying mechanism of MCR‐1 to modulate CAMK2β expression, thus facilitating bacterial evasion capability during BL21/pMCR‐1 infection.

FIGURE 6.

MCR‐1 in E. coli attenuated the immune response of THP‐1 cells during bacterial infection. (A) Hierarchical cluster analysis of differentially expressed genes (DEGs) identified as fold change>2 and FDR <0.01 from microarray analysis in THP‐1 cells infected with BL21/pEmpty or BL21/pMCR‐1 at an MOI of 50 for 4 h. (B) The top 20 differentially expressed pathways by dot mapping. More information on mRNA‐seq is available in Table S7. (C) Analysis of mRNA expression of inflammatory DEGs by qRT‐PCR to validate microarray results. (D) EC₅₀ values for binding of LBP (100 μg/mL) to BL21/pEmpty‐LPS, BL21/pMCR‐1‐LPS and sBL21/pMCR‐1‐LPS with FITC labelling by MST assay. Data points are mean values ±SD for n = 2 technical replicates each. (E, F) THP‐1 cells were treated with or without TAK₂₄₂ (1 μM) for 1 h, followed by infection with BL21/pEmpty and BL21/pMCR‐1 at an MOI of 50 for 4 h, the levels of released IL‐6 and TNF‐α were determined by ELISA (E), and NF‐κB expression and activation in cell extracts were detected by western blot analysis (F). (G, H) THP‐1 cells were treated with or without TAK₂₄₂ (100 nM) for 1 h, followed by stimulated with BL21/pEmpty‐LPS, BL21/pMCR‐1‐LPS (1 μg/mL) for 4 h, levels of released IL‐6 and TNF‐α were determined by ELISA (G), NF‐κB expression and activation in cells extracts were detected by western blot analysis (H). Relative expression levels of p‐P65 and p‐IκB in G and I are shown below each blot. Data are representative of three independent experiments. The data are reported as the mean ± SD. **p < 0.01; *** p < 0.001.

As a pivotal molecule to initiate TLR4 signalling activation during bacterial infection or LPS stimulation, LBP has been reported to interact with LPS via hydrophobic lipid A (Akira et al., 2006). Herein, we verified the structural change of lipid A while MCR‐1 was overexpressed in E. coli BL21, the spectrum of lipid A derived from BL21/pEmpty was obtained by MALDI‐TOF MS and showed a predominant hexa‐acylated lipid A as indicated by [M + H + Na]⁺ at m/z 1820.18 in the positive mode, in which the peaks at m/z 1740.23 arose by loss of a phosphate group (Figure S5A). The base peak at m/z 1739.46 also presented in the spectrum of BL21/pMCR‐1. Besides, a minor peak at m/z 1864.16 was obtained, which was consistent with the addition of pEtN (m/z 124) to the hexa‐acylated species m/z 1739.46, demonstrating the presence of the pEtN modification while MCR‐1 was overexpressed (Figure S5B). We subsequently determined if MCR‐1‐mediated structural modification prevented the interplay of LPS and LBP by MST assay and found an apparent affinity between LBP and LPS extracted from BL21/pEmpty (designated as BL21/pEmpty‐LPS), with an EC₅₀ value of 3.93 ± 1.97 μg/mL yielded by hill fit (Figure 6D). However, MCR‐1‐modified LPS (designated as BL21/pMCR‐1‐LPS) was not bound to LBP at all, which was partially reversed if LPS was extracted from BL21/pMCR‐1 that grew in medium supplemented with sub‐MIC MSI‐1 (designated as sBL21/pMCR‐1‐LPS). As a result, we hypothesized that pEtN modifications of lipid A obviously reduced the affinity of LBP and the stimulatory response of TLR4. We, therefore, sought to determine whether MCR‐1‐mediated lipid A modification could affect TLR4‐dependent IL‐6 and TNF‐α secretion in response to bacterial infection or stimulation with pEtN‐LPS, and results showed indeed reduction of released IL‐6 and TNF‐α in THP‐1 cells (Figure 6E,G). Additionally, to further elucidate the response of the TLR4 signalling pathway during BL21/pMCR‐1 infection or BL21/pMCR‐1‐LPS stimulation, we analysed the expressions or phosphorylation levels of some crucial components, including nuclear factor (NF)‐κB p65 (P65) and IκB. Consequently, we observed markedly reduced amounts of phosphorylated p65 (p‐P65) and phosphorylated IκB (p‐IκB) yet roughly constant P65 total protein following infection with BL21/pMCR‐1 or stimulation with BL21/pMCR‐1‐LPS (Figure 6F,H), which could be inhibited by simultaneous TAK₂₄₂ treatment. As expected, the attenuated immune response induced by MCR‐1‐mediated LPS modification was also observed in RAW 264.7 cells (Figure S5C–E).

Peptide MSI‐1 disturbed MCR‐1 synthesis in BL21/pMCR‐1 and protected against BL21/pMCR‐1‐induced infection in vitro and in vivo

Subsequently, to demonstrate the role of MSI‐1 in the process of MCR‐1‐catalysed reaction in E. coli, we first performed MST binding assay as described above to explore the binding characteristics of peptide MSI‐1 to pEtN‐LPS. Surprisingly, MSI‐1 exhibited comparative affinity to LPS from either BL21/pEmpty or BL21/pMCR‐1 (BL21/pEmpty‐LPS, BL21/pMCR‐1‐LPS) as assessed by MST assay (Figure 7A). Considering that AMP MSI‐1 possessed potent antimicrobial activity against a wide spectrum of microorganisms with multiple mechanisms, including modulating host gene expression, such as that of DNA synthesis‐related genes and β lactam resistance genes, and bacterial protein synthesis in our previous study (Ma et al., 2020b), it was reasonable that MSI‐1 negatively affected the expression of MCR‐1 in a dose‐ and time‐dependent manner in our results (Figure 7B). As a result, we speculated that peptide MSI‐1 might affect the LPS‐stimulated downstream signalling pathways by acting on MCR‐1 expression in BL21/pMCR‐1 at the sub‐MIC level. As expected, we observed that immune responses reflected in the level of released cytokines (IL‐6 and TNF‐α) and evoked activation of key molecules, including P65 and IκB, were reversely elevated if BL21/pMCR‐1 was preincubated with MSI‐1 at the sub‐MIC concentration (Figure 7C,D), which correspondingly increased in the group treated with sBL21/pMCR‐1‐LPS, yet MSI‐1 caused no significant change in LPS‐induced cytokines (Figure 7E,F). Besides, we further investigated in vivo virulence and immunostimulatory activity of BL21/pMCR‐1 and the therapeutic effect of the peptide MSI‐1. Mice were challenged with BL21/pMCR‐1 (5 × 10⁸ CFU), which led to 100% lethality in 2 days and displayed a higher mortality rate and bacterial loads in different organs than mice infected with BL21/pEmpty (Figure 7G), which might contribute to the significantly reduced circulating levels of cytokines induced by BL21/pMCR‐1 (Figure 7I) and immune evasion ability of polymyxin‐resistant E. coli in vivo. However, treatment with 5 mg/kg MSI‐1 obviously prolonged survival. Additionally, bacterial counts in different organs and the production of pro‐inflammatory cytokines decreased significantly in the presence of MSI‐1 (Figure 7H,I). Collectively, these findings suggested that peptide MSI‐1 acts as an inhibitor of MCR‐1 synthesis and comparatively protects against BL21/pMCR‐1‐induced infection in vitro and in vivo.

FIGURE 7.

Peptide MSI‐1 disturbed MCR‐1 synthesis in BL21/pMCR‐1 and protects against BL21/pMCR‐1‐induced infection in vitro and in vivo. (A) MST analysis of the binding affinity between FITC‐MSI‐1 (16 μg/mL) and BL21/pEmpty‐LPS, BL21/pMCR‐1‐LPS. (B) The expression of His‐tagged MCR‐1 in BL21/pMCR‐1 after treatment with MSI‐1 at a concentration of 2 μg/mL (1/2 MIC) at the indicated time points (upper) and in the presence of serial concentrations of MSI‐1 for 4 h (bottom) was assessed by western blot analysis. Relative expression levels are shown below each blot. (C, D) THP‐1 cells were infected with BL21/pEmpty, BL21/pMCR‐1 and sBL21/pMCR‐1 at an MOI of 50 for 4 h, NF‐κB expression and activation (C) in cells extracts were detected by western blot, IL‐6 (D, left) and TNF‐α (D, right) in cell supernatant were measured by ELISA. THP‐1 cells were stimulated with BL21/pEmpty‐LPS, BL21/pMCR‐1‐LPS and sBL21/pMCR‐1‐LPS (1 μg/mL), and for 4 h. (E, F) NF‐κB expression and activation in cells extracts were detected by western blot (E), IL‐6 (F, left) and TNF‐α (F, right) in cell supernatant were measured by ELISA. Relative expression levels of p‐P65 and p‐IκB in (C and E) are shown below each blot. (G) The survival of mice challenged with 1 × 10⁹ CFU/mL BL21/pEmpty or BL21/pMCR‐1 and treated with MSI‐1 was recorded on 7 consecutive days. n = 7 for each group. (H) CFU counts in mice livers, spleens, kidneys and lungs (n = 5) were detected by the agar plate dilution method. n = 5 for each group. (I) Cytokines in the plasma of mice were measured by ELISA. (n = 3 mice per group) 4 h post infection. Data are representative of three independent experiments. The data are reported as the mean ± SD. ns, not significant differences; *p < 0.05; **p < 0.01; ***p < 0.001.

DISCUSSION

The mcr‐1 gene product MCR‐1 is a pEtN transferase that remodels the lipid A moiety of LPS with phosphatidylethanolamine, which results in a decreased negative charge of LPS and confers polymyxin resistance to bacteria (Son et al., 2019). As a major component of OM of G⁻ bacteria, LPS properties affect the curvature, fluidity, permeability and integrity of OM. LPS is a glycan‐based Gram‐negative PAMP that is recognized by both the cell surface receptor TLR4 and intracellular caspase 4/11 (Akira et al., 2006; Kayagaki et al., 2013). As a result, we speculated that MCR‐1 of pathogens mediates chemically modified LPS and phenotypes involved in bacterial fitness, virulence and recognition by the host immune system. However, the properties of OM budding and regulation of inflammatory sensing in the context of LPS modification are poorly defined.

The work presented in Figure 8 established that the pEtN decoration of LPS induced by MCR‐1 negatively affected the membrane fluidity and the rate of OMV formation by E. coli. Vesicle budding was significantly inhibited, as evidenced by the reduced protein and LPS contents in OMVs. According to the results of proteomic analysis, more diverse protein constituents, especially many established virulent proteins, were observed in BL21/pMCR‐1‐OMVs, which might contribute to antibiotic resistance and bacterial pathogenesis of mcr‐1‐carrying bacteria. In addition, OMVs functioned as a vehicle to deliver LPS and other virulent proteins into the cytosol. Our results demonstrated that caspase 11‐dependent effector responses were insufficiently triggered by intracellular pEtN‐modified LPS upon OMV stimulation in vitro and in vivo. Beyond caspase 11 processing, RNA‐seq analysis displayed significantly different gene expression profiles during infection. As expected, TLR4‐mediated NF‐κB activation was dramatically alleviated once LPS was modified by MCR‐1. The MST binding assay further confirmed the attenuated affinity between LBP and pEtN‐decorated LPS. However, AMP MSI‐1 at the sub‐MIC level partly rescued OMV alteration and the immune responses attenuation in the presence of MCR‐1 during infection. Therefore, our data provide insights into the potential role of MCR‐1 in facilitating immune evasion of G⁻ infections.

FIGURE 8.

Proposed model of the key role of MSI‐1 in inhibiting MCR‐1 expression and LPS modification, subsequently promoting OMV budding by BL21/pMCR‐1 and inflammatory signalling in macrophages. LPS modification mediated by MCR‐1 in E. coli hindered host immune activation due to a failure to interact with both the cell surface receptor TLR4 and intracellular caspase 11. Moreover, AMP MSI‐1 at the sub‐MIC level partly rescued OMV alteration and the immune responses attenuation in the presence of MCR‐1 during infection (see text for detailed discussion).

Since LPS is a major constituent to be recruited faithfully from the OM of bacteria in the process of OMV assembly, chemical alterations to LPS have also been implicated in vesicle formation and cargo protein sorting, which has been evidenced by many retrospective studies. For instance, the endogenously produced Pseudomonas quinolone signal by Pseudomonas aeruginosa binds to the 4′‐phosphate group of LPS and increases the release of OMVs (Mashburn‐Warren et al., 2009). Another example of how LPS structure affects OMV formation is that the deacylation of lipid A by PagL leads to hypervesiculation and increased OMV production in Salmonella enterica serovar Typhimurium (Elhenawy et al., 2016). Indeed, our results demonstrated that the addition of pEtN to LPS regulated by MCR‐1 negatively affected vesicle production and cargo packaging simultaneously, which further demonstrated that the LPS structure was responsible for vesicle formation. In this study, we first reported the protein profile of OMVs derived from mcr‐1‐carrying E. coli and identified 106 unique proteins in secreted OMVs of BL21/pMCR‐1 when compared with vesicular proteins from BL21/pEmpty. Of these, we identified proteins that are expected to be involved in the biogenesis and functions of OMVs, and several proteins associated with multidrug resistance, bacterial virulence and the interaction of bacteria and host. Recent studies have indicated that G⁻ bacteria potentially altered the productions and contents of OMVs in response to various environmental stimuli such as temperature change, oxidative stress and treatment with envelope‐targeting antibiotics (Schwechheimer & Kuehn, 2015), it is reasonable that treatment of BL21/pMCR‐1 with AMP MSI‐1 at the sub‐MIC level caused an increase in OMV production and size, but reduced vesicular protein species. In other words, peptide MSI‐1 could exert potent antimicrobial activities against drug‐resistant bacteria including polymyxin‐resistant E. coli (Figure 1A,B). Moreover, MSI‐1, even at sub‐MIC concentration, affected the process of vesiculation and protein cargo by BL21/pMCR‐1 and further rescued attenuation of immune responses during both infection and OMV stimulation, reflecting a more intricate role for AMP as regulators of the innate immune system. Therefore, MSI‐1 is promoted to be exploited for anti‐infective therapy as a novel antimicrobial agent with immunomodulatory effects.

Besides, the results of the current study also showed the differences in the role of LPS extracted from BL21/pMCR‐1 and unmodified LPS in macrophages. LPS modification mediated by MCR‐1 in E. coli contributed to the change in OMVs and the low response of both the host pattern recognition receptor TLR4 during bacterial infection and noncanonical caspase 4/11‐dependent inflammasomes upon OMV stimulation. Nonetheless, OMV exposure derived from BL21/pMCR‐1 facilitated mitochondrial dysfunction and aggravated apoptosis in macrophages. Our results and those of Yang et al. (2017) conclusively demonstrate that MCR‐1‐mediated modification diminished the release of IL‐6 and TNF‐α induced by pEtN‐LPS. In turn, after incubation with MSI‐1, LPS produced by BL21/pMCR‐1 evoked a comparatively intense inflammatory response. We proposed that LPS modification mediated by MCR‐1 might be weakened in the presence of sub‐MIC MSI‐1, which resulted from the inhibitory effect of MSI‐1 on MCR‐1 expression (Figure 7). These findings were supported by the observation in the MST assay that pEtN‐modified LPS deficiently aggregated with native LBP, which was implicated in enhancing the host immune response to endotoxin (Figure 6D). In addition, Kopp et al. (2016) revealed that LBP colocalizes with LPS in both the cytoplasmic membrane and the close vicinity of activated caspases in intracellular compartments, which inspired us that the accessory protein LBP might also facilitate intracellular LPS localization in the process of OMV delivery. We presumed that low affinity for LBP might also be associated with an attenuated response of the caspase 4/11‐dependent inflammasome by pEtN‐LPS in our results (Figure 3). Considering that mitochondrial dysfunction was responsible for LPS‐induced apoptosis (Kuwabara & Imajoh‐Ohmi, 2004), it is conceivable that pEtN‐LPS delivered by OMVs tends to impede favoured pyroptosis, which facilitates the elimination of intracellular replication niches and clearance of bacterial pathogens, but promotes mitochondrial apoptosis as a default backup mechanism, which reflects the precise control of programmed death of infected cells.

AUTHOR CONTRIBUTIONS

Xinyue Ye: Methodology (equal); writing – original draft (equal); writing – review and editing (equal). Jian Wang: Formal analysis (equal); software (equal); validation (equal); writing – review and editing (equal). Pengfei Xu: Methodology (equal); validation (equal); writing – review and editing (equal). Xiaoqian Yang: Validation (equal); writing – review and editing (equal). Qixue Shi: Methodology (equal); writing – review and editing (equal). Genyan Liu: Investigation (equal); methodology (equal). Zhaoshi Bai: Funding acquisition (equal); visualization (equal); writing – review and editing (equal). Changlin Zhou: Conceptualization (equal); funding acquisition (equal); supervision (equal); visualization (equal); writing – review and editing (equal). Lingman Ma: Conceptualization (equal); funding acquisition (equal); supervision (equal); visualization (equal); writing – review and editing (equal).

CONFLICT OF INTEREST STATEMENT

None declared.

ETHICS STATEMENT

All procedures in this study involving animals were compliant with institutional animal health and well‐being policies, approved by the Experimental Animal Ethics Committee of China Pharmaceutical University (2019‐03‐002) and conducted following protocols approved by Science and Technology Department of Jiangsu Province (SYXK 2016‐0011). All the experiments were complied with all relevant ethical regulations.

Supporting information

Figure S1

Figure S2

Figure S3

Figure S4

Figure S5

Table S1

Table S2

Table S3

Table S4

Table S5

Table S6

Table S7

Ye, X. , Wang, J. , Xu, P. , Yang, X. , Shi, Q. , Liu, G. et al. (2023) Peptide MSI‐1 inhibited MCR‐1 and regulated outer membrane vesicles to combat immune evasion of Escherichia coli . Microbial Biotechnology, 16, 1755–1773. Available from: 10.1111/1751-7915.14297

DATA AVAILABILITY STATEMENT

The data supporting the findings of this study are available from the corresponding author upon reasonable request