Interactions between host epithelial cells and Acinetobacter baumannii promote the emergence of highly antibiotic resistant and highly mucoid strains

ABSTRACT

Acinetobacter baumannii is an important nosocomial pathogen. Upon colonizing a host, A. baumannii are subjected to selective pressure by immune defenses as they adapt to the host environment. However, the mechanism of this pathoadaptation is unknown. Here, we established an in vitro system to evolve A. baumannii driven by the continuous selective pressure exerted by epithelial cells, and we used a combination of experimental evolution, phenotypic characterization and multi-omics analysis to address the underlying mechanism. When continuously exposed to selective pressure by pulmonary epithelial cells, A. baumannii showed ptk mutation-mediated mucoid conversion (reduced adhesion and increased anti-phagocytic ability) by enhancement of capsular exopolysaccharide chain length; rsmG mutation-mediated deficiency of 7-methylguanosine modification in the 524th nucleotide of 16S rRNA, which increased ribosome translation efficiency; and rnaseI mutation-mediated changes in outer membrane permeability and efflux pump expression. Together, these mutations altered susceptibility to a variety of antimicrobial agents, including the novel antibiotic cefiderocol, by regulating siderophore and siderophore-receptor biosynthesis. In conclusion, pulmonary epithelial cells modulate A. baumannii pathoadaptation, implicating the host–microbe interaction in the survival and persistence of A. baumannii.

KEYWORDS:

• Acinetobacter baumannii
• pathoadaptation
• epithelial cell
• experimental evolution
• antibiotic resistance
• mucoid phenotype

Introduction

Upon colonizing a host, bacterial pathogens rapidly adapt to survive under multiple selective pressures. Invading bacteria must overcome host defenses, including nutrient limitation and immune response [1]. Bacterial pathogens have evolved numerous strategies to this end [2], for example, by acquiring new gene by horizontal gene transfer or by undergoing pathoadaptive mutations [3]. Pathogen mutations during persistent host infection are likely to be pathoadaptive [4,5].

Acinetobacter baumannii, an opportunistic pathogen, is increasing in incidence, and is prevalent in hospital environments, particularly in intensive care units and in patients with pneumonia and bacteremia [6]. Translocation of A. baumannii from the lungs to the bloodstream is associated with severe complications (e.g. sepsis), particularly in immunosuppressed patients or those on mechanical ventilation, leading to a high mortality rate [7]. Virulence factors such as OmpA, lipopolysaccharides, and capsule [8,9] mediate host colonization and damage. Especially, capsular exopolysaccharide and associated mucoid phenotype have emerged as a universal virulence factor owing to several observations [10]. Previous evidence supports a role for capsule structures in contributing to defenses against antibiotics such as antimicrobial peptides [11–13]. Because of the variability of the Acinetobacter genome, adaptive evolution under host stress can also promote persistence [14].

The lung is the main colonization and infection site of A. baumannii, and the airway epithelium is the first line of defense against invasion by pathogenic microorganisms [15,16]. Epithelial cells trigger the host inflammatory responset [17]. Although epithelial cells are nonspecific phagocytes, they are important in the pathogenesis of A. baumannii [18]. A. baumannii changes at multiple levels as it interacts with a host, for example, undergoing transcriptome remodelling and gene mutation [19–21]. Experimental evolution in controlled environments that mimic host selective pressures, combined with advanced DNA sequencing techniques, enables the investigation of adaptive evolution. To the best of our knowledge, the host cell-mediated adaptive evolution of A. baumannii has not been investigated using an experimental evolution model.

We established an in vitro system to evolve A. baumannii under continuous selective pressure from epithelial cells. Epithelial cell-adapted clones displayed not only the ability to evade adherence and phagocytosis due to a mucoid phenotype but also increased resistance to several antibiotics.

Materials and methods

Bacterial strains, cells, plasmids, and infection procedure

The wild-type (WT) A. baumannii strain ATCC17978 was used, and ASD9 and ACD9 were obtained from a laboratory evolution experiment (Table S1). One-step chromosomal gene mutation based on recombination-mediated genetic engineering (pMo130-Hyg) was performed to establish mutant strains; the pYMAb2-Hyg shuttle vector was used to establish gene-complemented strains (Table S1). Strains were cultured in Mueller-Hinton (MH) Broth (Oxoid; Thermo Scientific, Waltham, MA); gene-complemented strains were cultured in MH Broth supplemented with hygromycin (100 μg/mL). All strains were grown at 37°C.

The A549 human pulmonary adenocarcinoma cell line (#CCL185; ATCC, Manassas, VA) and HBE human bronchial epithelial cell line (#CRL2741; ATCC, Manassas, VA) were cultivated in Dulbecco’s modified Eagle’s medium (Cat. No: 10-013-CV; Corning Inc., Corning, NY) supplemented with 10% fetal bovine serum (35-081-CV; Corning Inc., Corning, NY) and 1% penicillin/streptomycin (Cat. No: 30-002-CI; Corning Inc., Corning, NY) and incubated at 37°C in an atmosphere of 5% CO₂. THP-1 human monocytes were cultivated in Roswell Park Memorial Institute-1640 medium (10-040-CV; Corning Inc., Corning, NY) with 10% fetal bovine serum and 1% penicillin/streptomycin. THP-1 cells were differentiated into macrophage-like cells (THP-1 Mø) by incubation in the presence of phorbol myristate acetate (Cat. No: HY-18739; MCE, Monmouth Junction, NJ) at 200 ng/mL for 48 h. For cell infection, fetal bovine serum-free medium without penicillin/streptomycin containing bacteria at a multiplicity of infection (MOI) of 100 was added to cell plates and incubated at 37°C in a 5% CO₂ atmosphere.

Laboratory evolution experiment

A single colony of ATCC17978 was cultured in 2 mL MH broth overnight at 37°C. Fetal bovine serum-free medium without penicillin/streptomycin containing bacteria at an MOI of 1:1 (10⁶ A. baumannii to 10⁶ cells) was added to cell flasks and incubated for 24 h at 37°C in an atmosphere of 5% CO_2, which was considered a passage. Bacteria were subsequently bottlenecked to start the next passage. As a control, equal numbers of bacteria evolved under identical conditions in the absence of cells. All populations evolved for 9 days, and populations on days 3, 6, and 9 were collected. The populations that evolved with and without cells were designated A1/A2/A3 and H1/H2/H3, and A4/A5/A6 and H4/H5/H6, respectively.

Genomic DNA sequencing and analysis

The genomic DNA of ACD9 and HCD9 was extracted using a QIAamp DNA Minikit (Qiagen Valencia, CA) following the manufacturer’s protocol. Libraries were prepared using the TruePrepTM DNA Library Prep Kit V2 for Illumina (Vazyme, Nanjing, China). Using a single “transposase” enzymatic reaction, sample DNA is simultaneously fragmented and tagged with adapters, an optimized, limited-cycle PCR protocol amplifies tagged DNA and adds sequencing indexes. Individual libraries were assessed on the QIAxcel Advanced Automatic nucleic acid analyzer and then were quantitated through qPCR by the use of KAPA SYBR FAST qPCR Kits (Kapabio systems, Boston, US). The bacterial genome was sequenced on an Illumina HiSeq platform (Illumina, San Diego, CA, USA) as described previously [22]. Breseq v. 0.33.0 was used for comparative genome analyses [23], referencing the original genome of ATCC17978 (GenBank accession NZ_CP018664.1). Mutations were confirmed by PCR and Sanger sequencing.

Cloning, expression, and recombinant RsmG protein purification

The full-length coding sequences of A. baumannii ATCC17978 rnaseI (AUO97_RS04110) and rsmG (AUO97_RS15310) were amplified using DNA from A. baumannii ATCC17978 and cloned into the pYMAb2 vectors to generate pYMAb2-rnaseI and pYMAb2-rsmG. Plasmids were verified by DNA sequencing and are listed in Table S2.

Bacterial expression and purification of His-tagged RsmG (RsmG-WT, RsmGΔT, and RsmG(TTG)_2→3) were performed as described previously [24]. Briefly, the rsmG sequence (AUO97_RS15310) was amplified using DNA from A. baumannii ATCC17978/ACD9/HCD9, cloned into the pET28A vector to generate the pET28A-rsmG-WT/ pET28A-rsmGΔT/ pET28A-rsmG(TTG)_2→3 constructs, and transformed into competent BL21 (DE3) cells. Isopropyl-β-d-thiogalactoside was added to the bacterial suspension to a final concentration of 1 mM to induce target protein production. Details of protein purification are provided in the Supplementary Methods.

Minimum inhibitory concentration determination

The minimum inhibitory concentration (MIC) of the following antibiotics was determined by the broth microdilution method with cation-adjusted MH broth (Oxoid, Hampshire, UK): cefoperazone/sulbactam (1:1, CFU/SU), ceftazidime (CAZ), cefepime (FEP), cefiderocol (CFDC), imipenem (IPM), meropenem (MEM), colistin (COL), amikacin (AMK), and gentamicin (GEN). MIC values were interpreted according to the Clinical and Laboratory Standards Institute guidelines (CLSI) [25].

Quantitative biofilm assay

Biofilm formation on polystyrene was measured via crystal violet staining [26]. Each strain was cultivated overnight at 37°C, and the cultures were diluted 1:100 into 150 µL MH broth in 96-well plates and incubated at 37°C for 24 h. The cultures were washed three times with PBS and stained with 0.1% crystal violet, washed with PBS, and 200 µL 30% acetic acid was added to each well. The optical density at 550 nm (OD₅₅₀) of each well was quantified using a microplate reader.

Bacterial adherence and invasion assay

For bacterial adherence assay, A. baumannii was added to a monolayer of epithelial cells at an MOI of 100 and incubated in a 5% CO₂ atmosphere at 37°C for 1 h. For bacterial invasion assay, A. baumannii was added to THP-1 Mø cells at an MOI of 100 and incubated in a 5% CO₂ atmosphere at 37°C for 1 h. The cells were washed three times with PBS and a fresh culture medium containing 100 μg/mL gentamicin was added, followed by incubation for 2 h. The cells were washed three times with PBS and lysed with 0.2% Triton X-100 for 20 min, and the lysate was plated on MH agar after gradient dilution. Details are provided in the Supplementary Methods.

Membrane permeability assay

Membrane permeability was measured using To-Pro-3 (Invitrogen, Thermo Fisher Scientific) and DiOC₂(3) (3, 3′-diethyloxacarbocyanineiodide) as described previously [27]. Membrane permeability was expressed as the ratio of To-Pro-3 fluorescence to green fluorescence or the frequency of To-Pro-3⁺ cells. Details are provided in the Supplementary Methods.

Bacterial capsule detection, surface polysaccharide extraction, and Alcian blue staining

Bacterial capsules were stained with Congo red and observed under a microscope (Olympus, Tokyo, Japan). Surface polysaccharides were extracted using the hot aqueous-phenol method [27]. Polysaccharide extracts were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and stained with Alcian blue (Sangon Biotech, Co., Ltd.).

Direct RNA sequencing and EpiNano-based m⁷G-modified site prediction

m⁷G-modified sites were predicted as described previously [28]. Briefly, A. baumannii total RNA samples were used to prepare RNA libraries for direct RNA sequencing; the sequence data were analyzed using EpiNano. Modification scores at each site were obtained for ATCC17978 and the rsmG mutant strains (rsmGΔT and rsmG(TTG)_2→3). Details are provided in the Supplementary Methods.

In vitro m⁷G methyltransferase activity by LC-MS/MS

RNA probes derived from 16S rRNA of A. baumannii ATCC17978 were purchased from Beijing Tsingke Biotechnology Co., Ltd. In vitro methyltransferase activity assay was performed as described previously [29]. Briefly, a 50 μL reaction mixture containing the RNA probe, recombinant protein (RsmG-WT, RsmGΔT, and RsmG(TTG)_2→3), and SAM was incubated at 37°C for 3 h, and the reaction products were digested by nuclease P1 and alkaline phosphatase for QQQ LC-MS/MS analysis [30]. Total m⁷G and G contents were quantified based on standard curves (Figure S2 A–D) to calculate the m⁷G/G ratio. Details are provided in the Supplementary Methods.

In vitro translation assay

Cell-free translation of eGFP mRNA was performed as described previously [31]. Briefly, a homemade Escherichia coli in vitro transcription-translation PURE system was used to assess 70S ribosome activity. PURE reaction mix (15 μL), which translates the eGFP gene, was added to 70S ribosome (rsmG-WT, rsmGΔT, and rsmG(TTG)_2→3). Protein synthesis was performed by incubation at 37°C (lid temperature 42°C), and eGFP fluorescence intensity was read at 0.5, 1, 1.5, and 2 h on a GloMax 20/20 luminometer (Promega).

Growth assay

Growth rates were determined in microplates using a Bioscreen C Microplate Reader (Oy Growth Curves Ab Ltd., Finland). The strains were cultured overnight at 37°C, and the cultures were diluted 1:100 into 200 µL MH broth with or without the ferrous iron chelator, 2,2′-dipyridyl (DIP), and incubated at 37°C with shaking. Absorbance at 600 nm was measured at 5 min intervals. Each strain was assayed in three independent cultures and three separate experiments. Growth was determined by calculating the area under the growth curve [32].

Transcriptome analysis of A. baumannii rsmG and rnaseI mutants

To investigate the role of rnaseI and rsmG in antibiotic resistance, RNA-seq of A. baumannii ATCC17978, rnaseIΔA, rsmGΔT, and rsmG(TTG)_2→3 was performed as described previously [33]. Briefly, A. baumannii ATCC17978, rnaseIΔA, rsmGΔT, and rsmG(TTG)_2→3 total RNA were subjected to RNA-sequencing, and genes differentially expressed in the rsmG and rnaseI mutants were screened. Details are provided in the Supplementary Methods.

RNA preparation, quantitative reverse transcription-PCR, and mRNA stability measurement

Total RNA was extracted, and RT–PCR was performed following the manufacturer’s instructions (Cat. No. RR037A; TaKaRa, Tokyo, Japan). TB Green Premix Ex Taq II (Cat. No. RR820A; TaKaRa, Tokyo, Japan) was used for signal detection, followed by real-time PCR analysis (Roche LightCycler480; Roche Molecular Systems, Basel, Switzerland) of gene expression levels; the primer sequences are listed in Table S3. For mRNA stability measurement, bacterial cultures were harvested at 0, 1, and 3 min after rifampicin addition (1 mg/mL), and total RNA samples were prepared and subjected to qRT-PCR. Details are provided in the Supplementary Methods.

Statistical analysis

Data are expressed as means ± standard deviations. Comparisons between two groups were performed using two-tailed unpaired Student’s t-tests. One-way analysis of variance was used for multiple-group comparisons. Two-way repeated measures analysis of variance was used to compare changes in a single variable over time in each group. Statistical analyses were performed using Prism v. 9.0 (GraphPad Software, San Diego, CA). P < 0.05 was considered to indicate statistical significance.

Results

The emergence of morphological and phenotypic diversity

We generated four evolved A. baumannii populations (all from the same ancestral colony of ATCC17978) that adapted to the antagonistic interaction with A549 and HBE cells. The populations (A1 to A3 and H1 to H3) evolved in a complete culture medium with A549 or HBE cells at an MOI of 1:1 (10⁶ A. baumannii to 10⁶ cells). Populations (A4 to A6 and H4 to H6) evolved under identical conditions in the absence of cells and were used as the controls. Bacteria from A1-A6 and H1-H6 were streaked on plates, and a randomly selected colony was cultured in liquid medium (Figure 1A). These colonies are hereafter referred to as ASD3/ASD6/ASD9/ACD3/ACD6/ACD9 and HSD3/HSD6/HSD9/HCD3/HCD6/HCD9 (Table S4).

Figure 1. Evolution experiment and phenotypic characterization. (A) Pathogen–host cell co-culture-based laboratory evolution experiment. (B) Representative images of the morphological diversity of A. baumannii populations adapted to pulmonary epithelial cells (A549 and HBE). (C) Representative Congo red-stained images of capsules of colonies (ASD9/ACD9/HSD9/HCD9) derived from evolved strains. Scale bar, 10 μm (D) OD₆₀₀ values of culture supernatants. (E) Biofilm formation is revealed by crystal violet staining. (F) Adherence to epithelial cells and internalization by macrophages. Data are from at least three independent experiments; error bars represent standard deviations. **P < 0.01 vs. blank group, ^##P < 0.01, ^#P < 0.05.

Adaptation of A. baumannii in the presence of epithelial cells was characterized by mucoid phenotype emergence. After 3 days, mucoid colonies were observed in populations co-cultured with A549 and HBE cells. This was not observed in the control groups (Figure 1 and Figure S1). Congo red capsule staining showed a thickened capsule in colonies from evolved populations (ACD9 and HCD9) compared to those from control populations (ASD9 and HSD9). A sedimentation assay of microviscosity showed that the OD₆₀₀ of the supernatant of evolved strains (ACD9 and HCD9) was higher than that of the control strains (ASD9 and HSD9). Crystal violet staining showed that the evolved strains had considerably weaker biofilm formation than the control strains. Regarding cell adherence and internalization, the evolved strains showed less adhesion to epithelial cells and less internalization into macrophages than the control strains. The evolved strains showed significantly different antibiotic susceptibilities, and their resistance to cephalosporins increased significantly (Table 1).

Table 1. Antimicrobial susceptibility of ATCC17978 and in vitro evolved strains

Strains	MIC (mg/L)
	AMK	GEN	FEP	CFU/SU	CAZ	CFDC	MEM	IPM	COL
ASD9	2	1	8	1	4	0.125	0.25	0.125	0.5
ACD9	4	1	>128	64	>128	8	2	0.5	0.25
HSD9	2	1	8	1	4	0.125	0.25	0.25	0.5
HCD9	2	2	64	16	32	2	1	0.25	0.5
ATCC17978	2	0.5	4	1	2	0.125	0.25	0.25	0.5
ATCC25922	2	0.5	0.03	0.25	0.125	0.25	0.03	0.125	0.06

Note: MIC: Minimum inhibitory concentration; AMK: amikacin; GEN: gentamicin; FEP: cefepime; CFU/SU: cefoperazone/sulbactam (1:1); CAZ: ceftazidime; CFDC: cefiderocol; MEM: meropenem; IPM: imipenem; COL: colistin.

Identification and investigation of genetic mutations associated with morphological and phenotypic diversity

To detect mutations, the genomes of ACD9 and HCD9 were subjected to whole genome sequencing (WGS). Three and two mutations were detected in ACD9 and HCD9, respectively, and confirmed by Sanger sequencing (Table 2). ptk (AUO97_RS06965) and rsmG (AUO97_RS15310) were mutated in ACD9 and HCD9, and rnaseI (AUO97_RS04110) was mutated only in ACD9. The mutation in ptk resulted in a substitution of valine for glutamic acid at amino acid 653 (ptk V653E). rsmG harboured a missense mutation (584/633nt) and glutamine insertion (175/633nt). The mutation in rnaseI (rnaseIΔA) resulted in premature termination of translation (rnaseI^M122*).

Table 2. Mutations acquired by evolved clones identified through whole genome re-sequencing (WGS).

Strains	Gene	Mutation	Annotation
ACD9	rnaseI	Coding (320/663 nt)	Δ1 bp
	rsmG	Coding (175/633 nt)	(TTG)2→3
	ptk	T→A	V653E (GTG→GAG)
HCD9	rsmG	Coding (584/633 nt)	Δ1 bp
	ptk	T→A	V653E (GTG→GAG)

Next, the impact of each mutation on morphological and phenotypic diversity was investigated (Figure 2). We introduced four mutations into the genome of the parental strain ATCC17978 to generate four mutant strains. The effect of each mutation was determined by evaluating mucoid phenotype and antibiotic susceptibility. Only ptk V653E showed a distinct mucoid phenotype, as indicated by a thickened capsule and increased microviscosity, decreased adhesion to epithelial cells and internalization into macrophages, and decreased biofilm formation. The rsmG and rnaseI mutations but not ptk mutations significantly altered antibiotic susceptibility (Table S5). ATCC17978 and the rsmG and rnaseI mutations (rsmGΔT, rsmG(TTG)_2→3, and rnaseIΔA) were transformed with a plasmid encoding WT rsmG and rnaseI, which restored their susceptibility to cephalosporins (Table S6).

Figure 2. ptk mutation induces the mucoid phenotype and reduces adhesion and internalization. (A) Representative images of morphological diversity of ATCC17978 and its mutant strains (ptk V653E, rsmGΔT, rsmG(TTG)_2→3, and rnaseIΔA). (B) Adherence to epithelial cells and internalization by macrophages of ATCC17978 and its mutant strains. (C) OD₆₀₀ values of culture supernatants of ATCC17978 and its mutant strains. (D) Biofilm formation by ATCC17978 and its mutant strains as revealed by crystal violet staining. (E) Representative images of morphological diversity of the evolved and ptk-WT complemented strains. (F) Representative Congo red-stained images of the capsules of evolved strains and ptk-WT complemented strains. Scale bar, 10 μm. (G) Adherence to epithelial cells and internalization by macrophages of evolved strains and ptk-WT complemented strains. (H) Capsular exopolysaccharide stained with Alcian blue (sup supernatant, arrows HMW exopolysaccharides). (I–K) IL-6, IL-8, and TNF-α mRNA levels in A549 cells (I), HBE cells (J), and THP-1 derived macrophages (K) at 3 h after infection. **P < 0.01 vs. ATCC17978 group.

Enhancement of capsular exopolysaccharide chain length in ptk mutation mediates mucus phenotype and reduces bacterial adhesion and internalization

Next, we replaced the mutated ptk in ACD9 and HCD9 with the WT ptk, thus generating the revertant mutants ACD9/ptk-WT and HCD9/ptk-WT (Figure 2). Compared to their parental strains, the two revertant mutants lacked a mucoid phenotype and had a thinner capsule and increased epithelial cell adhesion and macrophage internalization. SDS-PAGE and Alcian blue staining showed that the lysates of ACD9, HCD9, and ptk V653E had a very high-molecular-weight (HMW) polysaccharide that replaced the capsular polysaccharide band of the WT. This polysaccharide was also present in the culture supernatant. The polysaccharide bands of the two revertant mutants were similar to those of the WT. Therefore, ptk mutation increased the capsule polymer chain length. To investigate the effect of the extended polymer chain on host cells, we determined the mRNA levels of proinflammatory cytokines, including IL-6, IL-8, and TNF-α in A549, HBE, and THP-1 Mø. The IL-6, IL-8, and TNF-α transcript levels were significantly upregulated in the ptk V653E-infected groups, compared to the control and WT-infected groups.

The m⁷G524 deficiency caused by rsmG mutation induces antibiotic resistance

The m⁷G modification of 16S rRNA in the presence or absence of rsmG mutation was detected by Oxford Nanopore Technologies (ONT), a third-generation sequencing technology. m⁷G modification of G524 and G526 (replacement of guanosine with adenosine at nucleotide 520 in the 16S rRNA of A. baumannii) in 16S rRNA was missing in the rsmG mutant (Figure 3A). RsmG mediates m⁷G modification of nucleotide 527 in the 16S rRNA of E. coli (corresponds to G524 of A. baumannii) (Figure 3B). Consistent with previous studies, strains harbouring the rsmG mutation exhibited elevated MICs to streptomycin (Table S7). Next, an in vitro methyltransferase activity assay was performed to determine m⁷G activity and to verify modification sites for synthetic RNA probes with or without potential modification sites (obtained from ONT). Total m⁷G and G contents were measured via LC-MS/MS and quantified based on standard curves (Figure 3C). The RsmG mutant protein (RsmGΔT, RsmG(TTG)_2→3) showed no methyltransferase activity with RNA probe 1 (WT), whereas WT RsmG exhibited methyltransferase activity (Figure 3D). Therefore, rsmG mutation results in the loss of RsmG methylation activity. RNA probes 2 and 4 (524G to 524A in the consensus sequence) resulted in inefficient methylation compared to RNA probes 1 and 3 (524G) (Figure 3E and Figure S2). In addition, the m⁷G modification was downregulated by 526G to 526A (RNA Probe 3) but was significantly higher than that of RNA probes 2 and 4 (Figure S2F). The secondary structure of RNA Probe 3 (526G to 526A) changed significantly compared to RNA Probe 1(WT), potentially affecting methylation and indicating that RsmG has site specificity (Figure S1B). These results led us to hypothesize that rsmG mutations increase protein synthesis. Therefore, we measured the in vitro translational activity of ribosomes from the WT and rsmG mutant strains, using eGFP mRNA as a template. Ribosomes from the rsmG mutants (rsmGΔT, rsmG(TTG)_2→3) exhibited higher eGFP synthetic activity than the WT (Figure 3F). Ribosomes from rsmG-complemented strains (rsmGΔT/pYMAb2-rsmG and rsmG(TTG)_2→3/pYMAb2-rsmG) showed decreased eGFP synthetic activity compared to the corresponding rsmG mutant strains (rsmGΔT/pYMAb2 and rsmG(TTG)_2→3/pYMAb2) (Figure 3G and H). Therefore, rsmG mutations resulted in m⁷G524 deficiency, which increased translation efficiency and acted as a global regulator.

Figure 3. m⁷G524 deficiency caused by rsmG mutation increases ribosomal translation efficiency. (A) Prediction of m⁷G modification site in rsmG mutant strains by third-generation sequencing. (B) RsmG mediated m⁷G modification of nucleotide 524 in the 16S rRNA of A. baumannii. (C) In vitro methyltransferase activity using 16S rRNA probes and RsmG protein, with detection of m⁷G and G by LC-MS/MS. (D and E) m⁷G/G ratio. (F–H) 70S ribosome activity of rsmG mutant (F) and complemented strains (G and H) by in vitro translation assay. **P < 0.01 vs. RsmG-WT (Probe 1, ATCC17978, rsmGΔT/pYMAb2 or rsmG(TTG)_2→3/pYMAb2) groups, ^##P < 0.01.

The rnaseI mutation affects outer membrane permeability and efflux pump expression

Transcriptomic analysis was performed to explore the mechanism of rnaseI mutation-mediated antibiotic resistance. The transcriptome of the WT and rnaseIΔA strains revealed upregulation of four genes encoding RND efflux transporters, which mediate antibiotic transport, and upregulation of five genes involved in outer membrane formation (Figure 4A and B). We hypothesized that modulation of nucleotide homeostasis mediates some of the transcriptional and phenotypic changes observed in rnaseIΔA, therefore, we performed mRNA stability assays. The mRNA stability of RND efflux transporter-related genes and outer membrane protein/lipoprotein synthesis-related genes in rnaseIΔA were significantly higher than that of the WT after transcriptional repression by rifampicin (Figure 4C and D). Next, we examined the effect of carbonyl cyanide 3-chlorophenylhydrazone (CCCP; efflux pump inhibitor) on MIC values. CCCP reduced the MIC values of rnaseIΔA (Table S8). Scanning electron microscopy (SEM) showed that the outer membrane morphologies of the ACD9 and rnaseIΔA strains were significantly changed compared to the ASD9 and ATCC17978 WT strains. Evolved and mutant cells were of heterogeneous shapes rather than the homogenous shape of the parental isolate (Figure 4E). In addition, the outer membrane permeability was reduced in ACD9 compared to ASD9, and in the rnaseIΔA strain compared to the WT strain, as measured according to the positive cell frequency (Figure 4F) and fluorescence intensity (Figure 4G). Therefore, overexpression of RND efflux transporters and changes in outer membrane morphology and permeability are implicated in the antibiotic resistance of the rnaseIΔA strain.

Figure 4. rnaseI mutation affects outer membrane permeability and efflux pump expression. (A) Heatmap of RND efflux transporter gene expression. (B) Heatmap of outer membrane formation gene expression. (C and D) mRNA stability of RND efflux transporter-related genes (C) and outer membrane protein/lipoprotein synthesis-related genes (D) in rnaseIΔA compared to ATCC17978. (E) Representative images of the morphology of the evolved and rnaseI WT/ΔA strains by scanning electron microscopy (SEM). (F and G) Outer membrane permeability was measured via flow cytometry – the percentage of TO-PRO-3 positive cells (F) and adjusted fluorescence intensity (G). **P < 0.01, *P < 0.05 vs. ATCC17978 or ASD9 group, ^##P < 0.01, ^#P < 0.05.

The rsmG and rnaseI mutation impairs iron acquisition, which may mediate resistance to cefiderocol

The rsmG and rnaseI mutants (rsmGΔT, rsmG(TTG)_2→3, and rnaseIΔA) showed downregulation of bauB, bauE, bauF, basB, basC, basD, basE, basF, basG, and basJ (Figure 5A). These genes are linked to siderophore and siderophore-receptor biosynthesis. We hypothesized that downregulated expression of these genes may suppress iron acquisition and utilization. In the completed medium, the growth of the rsmG and rnaseI mutants was not inhibited compared to the WT (Figure 5B and H), whereas the rsmG and rnaseI mutants displayed significant growth restriction in an iron-limited medium (containing the ferrous iron chelator 2,2′-dipyridyl; DIP) (Figure 5C, D, and H). In the presence of 100 μM DIP, only the rsmGΔT/pYMAb2-rsmG strain partially recovered from growth restriction (Figure 5F, and I-K). In the presence of 200 μM DIP, the growth of the rnaseI and rsmG-complemented strains partially recovered (Figure 5G, and I-K). Mutations in these genes impaired bacterial utilization of iron, possibly explaining the reduction in cefiderocol susceptibility.

Figure 5. rsmG and rnaseI mutations impair bacterial iron acquisition and utilization. (A) Siderophore and siderophore receptor gene expression data in rsmG and rnaseI mutants. (B–G) Growth curves of the mutants (B–D) and their complemented strains (E–G) in the presence or absence of the ferrous iron chelator 2,2′-dipyridyl (DIP). (H–K) Areas under the growth curves (AUC) of the mutants (H) and their complemented (I–K) strains in the presence or absence of DIP. **P < 0.01, *P < 0.05 vs. ATCC17978 or ATCC17978/pYMAb2 group, ^##P < 0.01, ^#P < 0.05.

Figure 6. Proposed mechanism of A. baumannii pathoadaptation to epithelial cells. In A. baumannii, mutation of ptk mediates mucoid conversion and reduces adhesion and internalization. Mutations of rsmG and rnaseI involve in antibiotic resistance, and m⁷G524 deficiency in 16S rRNA caused by rsmG mutation increases ribosomal translation efficiency and may act as global regulators, and rnaseI mutation affects outer membrane permeability and efflux pump expression by modulating RNA metabolism. (By Figdraw, www.figdraw.com).

Discussion

Bacterial evolution from a commensal to a pathogen may occur by the acquisition of new gene via horizontal gene transfer (gain of function) or accumulation of pathoadaptive mutations (change in function) [34,35]. For example, macrophage-adapted E. coli exhibit increased intracellular survival and delayed phagosome maturation due to a single IS1 insertion upstream of yrfF [3]. The loss of function of mucA in Pseudomonas aeruginosa results in an increased ability to escape phagocytosis and pulmonary clearance [36]. The lung is the main colonization and infection site of A. baumannii, an opportunistic pathogen mainly in patients on mechanical ventilation [7]. The airway epithelium is the first line of defense against invasion by pathogenic microorganisms, and epithelial cells trigger the host inflammatory response [17]. Therefore, the acquisition of adaptive mutations is essential for A. baumannii persistence. It is shown that A. baumannii can very easily acquire and lose genes [37]. Several pathogenic bacteria have evolved to survive and indeed take advantage of the host environment [38]. In this study, we induced the evolution of A. baumannii in vitro with two pulmonary epithelial cell lines. Epithelial cell-adapted A. baumannii exhibited a mucoid phenotype and enhanced antibiotic resistance, and several gene mutations and deletions (Table S9) were found. The mucoid phenotype was characterized by reduced adhesion to epithelial cells and internalization by macrophages. The antibiotic resistance adaptation was associated with the overproduction of efflux pumps and reduced outer-membrane permeability (Figure 6).

In a previous study, we found that clinical isolates from sputum showed enhanced resistance and morphological changes after within-host evolution [39]. In that study, we identified mutations in ptk that caused mucoid conversion and decreased adherence and internalization, promoting dissemination. Macrophage- and neutrophil-adapted A. baumannii also undergo mucoid conversion and ptk mutation (unpublished data). The K locus, which encompasses ptk, is responsible for capsule biosynthesis, and ptk mutants produce very HMW polysaccharides [40]. Our data confirm that ptk mutations trigger the production of HMW exopolysaccharides, which have a pro-inflammatory effect on host cells. Capsular exopolysaccharide is important for soft tissue infection, defense against serum factor-mediated killing, and biofilm formation [41]. Therefore, the acquisition of a mucoid phenotype promotes the dissemination and persistence of A. baumannii in the host.

Antibiotic resistance contributes to the persistence of A. baumannii in the host [39] and is an adaptive response to stress [21]. In this study, epithelial cell-adapted A. baumannii showed enhanced resistance to ceftazidime, cefoperazone/sulbactam, cefepime, and cefiderocol, as a result of mutations to rsmG and rnaseI (linked to RNA modification and metabolism). rsmG, which encodes S-adenosylmethionine (SAM)-dependent 16S rRNA methyltransferase, is involved in 7-methylguanosine modification (m⁷G) of the 530 loops of 16S rRNA in a variety of Gram-negative bacteria [42,43]. In this study, mutations in rsmG resulted in the loss of the m⁷G modification of 16S rRNA in A. baumannii, and the modification site was G524 of 16S rRNA. We excluded G526 as a potential methylation modification site because the RNA probe retained the m⁷G modification after the change from 526G to 526A. Therefore, the reduction of the m⁷G/G ratio was likely related to the altered secondary structure of RNA probe 3. RsmG mutation-associated loss of m⁷G modification mediates low-level streptomycin resistance [42,43]. In this study, rsmG mutation reduced susceptibility to several β-lactams, however, the resistance mechanism is unknown. RNA-seq revealed no resistance mechanism. The universally conserved 530 loops of 16S rRNA is crucial for translation [44,45]. Loss of the m⁷G modification may affect ribosomal function [46], and the translation efficiency and fitness of bacterial pathogens [47]. Therefore, we hypothesized that the RsmG mutation enhances β-lactam resistance by affecting ribosome function and protein translation. Indeed, rsmG mutation increased ribosomal translation efficiency, a rapid-response mechanism to environmental stress [48,49]. Stress can reduce the overall protein translation capacity of bacteria [50], hampering their survival. Maintenance of translation efficiency guarantees timely synthesis of stress-responsive proteins, enabling long-term survival during colonization and infection [51], Therefore, rsmG mutation-mediated upregulation of protein translation efficiency may promote bacterial survival under β-lactam stress.

β-lactam tolerance can also be mediated by overexpression of genes encoding efflux pumps and outer-membrane changes. Indeed, our data suggest that these were altered by rnaseI mutation. RNase I regulate bacterial biofilm formation, motility, acid resistance, β-lactam tolerance, and nucleotide metabolism [52]. Under stress conditions, periplasmic RNase I enter the cell, leading to extensive RNA degradation [53]. In this study, the mRNA stability of RND efflux pumps and outer membrane genes was upregulated by rnaseI mutation, enhancing the expression of RND efflux pumps, changing the morphology of the outer membrane, and decreasing its permeability, leading to β-lactam resistance. In addition, impaired iron utilization caused by the rsmG and rnaseI mutations promoted resistance to cefiderocol. This is unlike previously reported cefiderocol resistance mechanisms, such as mutations in genes encoding iron transport-related proteins and the transfer of an extended-spectrum β-lactamase-encoding gene [54,55]. This study demonstrates a mutagenic role for host factors previously thought to be insusceptible to resistance by mutation.

Conclusion

A. baumannii adapted to host-associated selective pressures during 9 days of co-culture with epithelial cells. Mucoid clones with β-lactam resistance and the ability to evade adherence and phagocytosis rapidly emerged. Our data reveal the complex influence of pulmonary epithelial cells on A. baumannii pathoadaptation, implicating the host-microbe interaction in A. baumannii survival and persistence. This pathoadaptation could explain the emergence of antibiotic resistance and tissue translocation in patients infected with A. baumannii. Our findings provide insight into the effect of the host response on the evolution of A. baumannii.

Authors’ contributions

W. Z. and Y.S.Y. wrote the main manuscript text, X.T.H., H.Z., and Z.H.Z. designed this study, Y.Y., J.T.H, J.F.W., L.L., M.S.G., Y.F., and Y.S. participated in part of the experiments, H.C., J.Z.L., X.T.H., and J.Z.L. revised the manuscript, Y.Y. and X.C.L. analyzed the data.

Acknowledgements

The work was supported by the National Natural Science Foundation of China (82072313, 31770142, 81971897) and the National Science Foundation of China International Cooperation and Exchange Programme (81861138054). We are grateful to Dr. Fengqin Wang from College of Animal Sciences, Zhejiang University for her great support to this work.

Data availability

The complete genome sequences of the evolved strains (ACD9 and HCD9) were deposited in NCBI under BioProject No. PRJNA867505. The RNA-seq data of A. baumannii ATCC17978, rnase I, and rsmG mutants were deposited in NCBI under BioProject No. PRJNA867119. The direct RNA-seq (ONT) data of A. baumannii rsmG WT/mutants were deposited in NCBI under BioProject No. PRJNA882509.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by National Natural Science Foundation of China [grant number: 31770142,81861138054, 81971897,82072313].