The bacterial strains and plasmids used in this study are listed in Supplementary Table S1. The primers used in this study are listed in Supplementary Table S2. Unless otherwise stated, all A. baumannii strains were grown in Luria–Bertani (LB) broth or on LB agar plates at 37°C. Antibiotics and other chemicals were used at the following final concentrations: gentamicin, 10 μg/mL; colistin, 4 μg/mL; kanamycin, 50 μg/mL.

To generate pMMB67EH-NP-mcr-1.1 and pMMB67EH-NP-mcr-4.3, the mcr-1.1/mcr-4.3 gene and their native promoter were amplified from an mcr-1-carrying IncX4 (Feng et al., 2018) plasmid and mcr-4.3 carrying plasmid (Ma et al., 2019) and then cloned into pMMB67EH-spy, a low-copy number plasmid (containing the RSF1010 origin of replication). The map of plasmids and recombinant plasmids used for constructing mcr-positive A. baumannii strains are listed into Supplementary Figure S1.

The A. baumannii strains were grown to mid-log phase (OD₆₀₀ = 0.6–0.8) in LB medium. The cultures were pelleted, washed once with PBS, adjusted to an OD₆₀₀ of 0.5 with PBS, and diluted 1: 10. Then, 10 μL of the diluted culture was added into 90ul of LB. The OD₆₀₀ of the culture within 24 h was determined using a microplate reader (BioTek). Three replicates were analyzed for each strain.

The MIC of the antimicrobial agents against A. baumannii with or without MCR were determined using the standard broth microdilution method, according to the CLSI 2020 guidelines. In brief, all antibiotics were 2-fold diluted in Mueller-Hinton broth (MHB), and 90 μL of this solution was mixed with 10 μL of diluted culture, containing approximately 1.5 × 10⁶ CFU/mL in a 96-well microtiter plate. After 16–20 h of incubation at 37°C, the MIC values were defined as the lowest concentration of antibiotics with no visible bacterial growth. Experiments were performed with 3 biological replicates.

In vitro competition experiments were used to measure the competitiveness of the ATCC 17978 (mcr-1.1/pMMB67EH), ATCC 17978 (mcr-4.3/pMMB67EH), and ATCC 17978 (pMMB67EH only). The mcr-positive strains (both resistant to gentamicin and colistin) were competed against ATCC 17978 (pMMB67EH only, only resistant to gentamicin) and plating assays on LB agar plates with colistin/gentamicin or gentamicin was used to measure changes in the CFU of the two strains during competition. All competitions were carried out in M9 medium with three replicates per strain. The bacteria were cultured overnight in LB supplemented with the appropriate antibiotics. After the cell density was normalized (OD₆₀₀ = 0.5), the bacteria solution was diluted 1: 100 in M9 broth and mixed at 1: 1 ratio. Cultures were incubated at 37°C with shaking (220 rpm). After 12, 24, 48, 72 h, aliquots were serially diluted 10-fold in PBS, and 10 μL of each dilution was spotted onto LB agar with colistin/gentamicin (CFU _mcr ) or gentamicin (CFU _total). The plates were incubated at 37°C and enumerate the rate by CFU count after 16 h: rate (%) = CFU _mcr /CFU _total * 100%.

Bacterial cells were grown overnight in LB broth at 37°C. After the cell density was normalized (OD₆₀₀ = 0.5), the resulting cultures were serially diluted, and 10 μL of each dilution was spotted onto LB agar or LB agar supplemented with the indicated concentrations of sodium dodecyl sulfate (SDS) and ethylenediaminetetraacetic acid (EDTA) (Figure 1A: 0.1% SDS + 0.2 mM EDTA or 0.1% SDS + 1.0 mM EDTA; Figure 1B: 0.1% SDS + 0.4 mM EDTA or 0.5% SDS + 0.4 mM EDTA). The plates were incubated at 37°C and photographed after ∼24 h (Feng et al., 2022). Experiments were performed with 3 biological replicates.

The genomic DNA of the F7-AB was extracted using the Qiagen Blood and Tissue kit (Qiagen, Hilden, Germany). DNA libraries were constructed with 350 bp paired-end fragments and sequenced using an Illumina HiSeq 2000 platform. The Illumina reads of the strain were mapped against the reference genome sequence of A. baumannii ATCC 19606 (GenBank accession: GCF_009035845.1) using the program MAQ. In our WGS analysis, we used a variety of programs for different steps. We started with FastQC (Babraham Bioinformatics, 2010) and Trimmomatic (Bolger et al., 2014) for quality control and preprocessing of raw sequencing data. Trimmed reads with length > 30 and Phred scores > 20 were retained for subsequent analyzes. Then, we employed BWA (Li and Durbin, 2009) for aligning the reads to a reference genome (A. baumannii ATCC 19606; GenBank accession: GCF_009035845.1). Variant calling was done using Genome Analysis Toolkit (GATK) (McKenna et al., 2010; DePristo et al., 2011). After obtaining the variant calls from the WGS data, we further filtered and annotated the SNPs using VCFtools (Danecek et al., 2011) and SnpEff (Cingolani et al., 2014) for SNP analysis. We excluded ambiguous calls as wells as indels and MNVs. We also excluded calls with low coverage using a minimum depth of either 10% of the mean coverage and calls with mapping quality or base quality scores < 30. Consistent with previous findings (Oikonomou et al., 2015; Nhu et al., 2016; Zhang et al., 2017; Nurtop et al., 2019; Jovcic et al., 2021; Kabic et al., 2023), four mutations [LpxD (E117K), PmrA (I13N), PmrB (A138T and A444V)] linked to colistin resistance were detected among the screened SNPs (Supplementary Table S6).

Synergy measurement was utilized to determine the antimicrobial effect of colistin and macromolecular hydrophobic antibiotics (rifampicin, erythromycin and cefoxitin) upon colistin-resistance A. baumannii (MCR-1.1-expressing cells, MCR-4.3-expressing cells, pig feces-derived MCR-4.3-expressing A. baumannii strain AB18PR065, clinical A. baumannii strain F7-AB). The checkerboard assay was set up in a 96-well plate. Briefly, columns 1 to 8 contained 2-fold serial dilutions of colistin, and rows A to H contained 2-fold serial dilutions of macromolecular hydrophobic antibiotics. Then the strains were cultured to exponential phase and adjusted to a density of OD₆₀₀ = 0.5 by diluting with LB broth. After dilution at a ratio of 1: 10, 20 μL of diluted culture was added to 180 μL of LB broth containing colistin and another indicated antibiotics in a 96-well plate. A nanophotometer (NP80, IMPLEN) was utilized to measure the optical density at 600 nm (OD₆₀₀) in each well before and after incubation at 37°C for 16 h. The fractional inhibitory concentration index (FICI) was calculated according to a previously published protocol (Pál et al., 2023). Briefly, FICI = (MIC of drug A in combination) / (MIC of drug A alone) + (MIC of drug B in combination) / (MIC of drug B alone), Synergy is defined as FICI ≤ 0.5, Indifference is defined as 0.5 < FICI ≤ 4, Antagonism is defined as FICI > 4 (Jenkins and Schuetz, 2012).

Exponentially growing bacterial cultures (optical density at OD₆₀₀ = 0.4–0.6) of ATCC 17978 (mcr-1.1/pMMB67EH), ATCC 17978 (mcr-4.3/pMMB67EH) and ATCC 17978 (pMMB67EH only, as a control group) were pelleted. After the supernatant was removed, the cell pellet was resuspended in RNA-easy Isolation Reagent (Vazyme). The total RNA was precipitated by the addition of isopropanol and collected by centrifugation. The supernatant was discarded, and the RNA pellet was washed with 75% ethanol. After the pellet was air-dried, the mRNA was dissolved in RNase-free H₂O. Any contaminating genomic DNA was digested with gDNA wiper Mix (Vazyme). The purified mRNA was reverse transcribed to cDNA with HiScript II qRT SuperMix (Vazyme). The cDNA levels of the target genes were then quantified with quantitative real-time PCR (qPCR) on CFX Opus 96 Real-Time PCR Instrument (Bio-Rad) using AceQ Universal SYBR qPCR Master Mix (Vazyme). All qPCR primers were determined to be >95% efficient, and the cDNA molecular masses were experimentally confirmed to be within the linear dynamic range of the assay. The signals were normalized to those of the housekeeping 16S rRNA transcript and quantified with the ΔΔCt method. The error bars are the 95% confidence intervals of three technical replicates.

The A. baumannii strains were grown to mid-log phase (OD₆₀₀ = 0.6–0.8) in LB medium. The cultures were pelleted, washed once with PBS, adjusted to an OD₆₀₀ of 0.5 with PBS, and diluted 1: 10. Then, 200 μL of the diluted culture was added into 1800ul of LB. Cultures were incubated at 37°C with shaking (220 rpm). After 0, 1, 2, 3 day, bright-field imaging was performed on an inverted microscope (Olympus BX63). The ImageJ software was used to analyze the cell length.

Statistical analysis was performed using Prism (version 8.3.0, GraphPad Software). Data were analyzed using the paired Student’s t test, and for comparison of data from three or more conditions, analysis of variance (ANOVA) was used. p value of 0.05 or less was considered statistically significant.

Although evidence of our previous study showed that expression of MCR-1 induces cell shrinkage and death in E. coli and K. pneumonia during the stationary phase (Feng et al., 2022), the impact of MCR on the viability of A. baumannii remains unclear. To study the impact of MCR on the physiology of A. baumannii, the plasmids pMMB67EH-NP-mcr-1.1 and pMMB67EH-NP-mcr-4.3 carrying mcr-1.1 and mcr-4.3 and their native promoters, respectively, were generated, and the low-copy number pMMB67EH-spy plasmid served as the empty vector control. We then constructed three strains, A. baumannii strain ATCC 17978 carrying pMMB67EH-NP-mcr-1.1 (MCR-1.1-expressing cells), strain ATCC 17978 carrying pMMB67EH-NP-mcr-4.3 (MCR-4.3-expressing cells) and ATCC 17978 carrying the empty vector pMMB67EH-spy (Vector cells). Minimum inhibitory concentration (MIC) of colistin against the three strains above were then determined (Table 1), and the growth of MCR-1.1-expressing cells, MCR-4.3-expressing cells and Vector cells in LB medium were monitored (Figures 2A,B). Like previously observed in the E. coli (Feng et al., 2022), the expression of MCR did not affect the growth of A. baumannii ATCC 17978 (Figures 2A,B). However, when we extended the incubation time, we found that the MCR-expressing cells have shorter length compared to vector cells in the stationary phase (Figure 2E). Meanwhile, in vitro competition experiments were also performed to determine the effect of mcr-1.1 expression on bacterial growth in the stationary phase. As shown in Figures 2C,D, after 24 h, the growth rates of A. baumannii strains expressing MCR showed significant decrease when compared to Vector cells (pMMB67EH-spy, without mcr). These results suggest that the viability and competitiveness of mcr-positive A. baumannii strains was impaired in the stationary phase.

To understand how MCR damage the viability of A. baumannii strains, we investigated the mechanism underlying this phenomenon. Overexpression of mcr-1 results in significant degradation in cell membrane and cytoplasmic structures (Yang et al., 2017). Therefore, we wonder whether the bacterial death caused by MCR in A. baumannii strains results from increased OM permeability. To test this, we employed SDS/EDTA sensitivity assays to measure the OM permeability of MCR-1.1-expressing cells, MCR-4.3-expressing cells, vector cells (Figures 3A,B). As expected, all the MCR-expressing cells displayed increased sensitivity to detergents SDS/EDTA compared with the vector cells, indicating that MCR-1.1 and MCR-4.3 both increased the permeability of OM in A. baumannii. Moreover, MCR-1.1-expressing cells were more sensitive to SDS/EDTA than MCR-4.3-expressing cells, which may be associated with the different expression level of MCR protein (Figure 3C). These data demonstrated that the increased OM permeability affected the viability of mcr-positive A. baumannii strains in the stationary phase.

Acinetobacter is intrinsically resistant toward many antibiotics because of the low permeability of its outer membrane (Zahn et al., 2016), especially resistant to OM-impermeable large molecule antimicrobials due to the absence of large-channel general porins such as OmpF and OmpC (Sugawara and Nikaido, 2012). Therefore, we speculated that mcr-positive A. baumannii might be more sensitive to certain antimicrobials. To do it, we screened nearly 10 antibiotics (Supplementary Table S3), covering most of macromolecular hydrophobic antibiotics for change in MIC with or without the presence of MCR protein. As we had expected, we found that MCR-expressing cells were more sensitive to rifampicin (RIF), erythromycin (ERY), and MCR-1.1-expressing cells were more sensitive to cefoxitin (FOX), in contrast to MCR-negative cells (Figure 1A). Next, we aimed to investigate whether antibiotics could potentiate this effect. In the presence of colistin, several antibiotics were highly potentiated. Checkerboard broth microdilution assays showed potentiation of RIF, ERY, and FOX by colistin against MCR-expressing A. baumannii (Figures 1B–D). By calculating their Fractional Inhibitory Concentration Index (FICI), in which ≤0.5 indicating synergistic, we found that there was potent synergy of colistin combination with rifampicin, erythromycin and cefoxitin versus MCR-expressing A. baumannii (Figures 1B–D).

To investigate whether colistin potentiation is conserved beyond laboratory-generated MCR-expressing A. baumannii, we tested the FICI of a MCR-4.3-expressing A. baumannii strain (AB18PR065) derived from pig feces (Ma et al., 2019) in colistin combination with rifampicin, erythromycin and cefoxitin. As expected, the FICI was less than or equal to 0.5 for all three drugs (Figures 1D,E), indicating that MCR-4.3 also rendered the potentiation of large molecule antimicrobials in clinical A. baumannii expressing MCR.

Given the therapeutic potential of combination therapy in mcr-positive A. baumannii, we wondered whether this potentiation could be applicated to clinical chromosomally mediated colistin resistance A. baumannii, since the complete loss or modifications of the target LPS are two main mechanisms of chromosomally mediated colistin resistance A. baumannii (Novović and Jovčić, 2023), which is similar to the acquired mechanism of mcr encoding a phosphoethanolamine (pEtN) transferase that adds pEtN to the lipid A component of LPS. To test this, we fist the FICI of a clinical A. baumannii strain F7-AB which is intrinsically resisted to colistin [LpxD (E117K), PmrA (I13N), PmrB (A138T and A444V)] (Oikonomou et al., 2015; Nhu et al., 2016; Zhang et al., 2017; Nurtop et al., 2019; Jovcic et al., 2021; Kabic et al., 2023). Encouragingly, the result of the checkerboard assay was similar with that of mcr-positive A. baumannii except erythromycin and cefoxitin (Figure 4). These observations indicate that the combination of colistin with rifampicin is not only able to overcome mcr mediated acquired colistin resistance, but also able to treat chromosomal colistin resistance.