Pseudomonas aeruginosa is a significant opportunistic pathogen and a serious burden to public health and economy [1]. Especially dangerous are carbapenem-resistant strains of P. aeruginosa regarded by WHO as critical priority pathogens [2]. This breeds the need for understanding mechanisms underlying bacterial resistance to carbapenems. Research into the molecular genetic underpinnings of carbapenem resistance focuses mostly on β-lactamase-associated mechanisms that are determined by plasmid genes and therefore can be acquired through horizontal gene transfer. However, there is another contributor whose role should not be overlooked: induced mutations in the core genome of P. aeruginosa resulting in high-level carbapenem resistance [3]. There are two approaches to the study of mutations conferring resistance to carbapenems. The first involves the analysis of drug-resistant isolates obtained from clinical, agricultural or environmental sources. In the second approach, the evolution of antibiotic resistance is modeled in vitro. For that, bacteria are grown in antibiotic concentration gradients. A smart method for studying mutational resistance has been proposed in [4]. Its authors created a spatiotemporal model that enabled migration of Escherichia coli in trimethoprim and ciprofloxacin gradients and generated a variety of mutants for further analysis. Interestingly, some of the E. coli clones carried mutations that were not linked to trimethoprim or ciprofloxacin resistance [4]. So, we became curious to explore the direction and implications of such mutations. Specifically, we were interested in the clinically significant phenomenon of cross-resistance, in which a mutation induced by exposure to an antibiotic could confer resistance to other antibiotics [5, 6]. The aim of this study was to test the hypothesis that P. aeruginosa can develop crossresistance to other antibiotics while adapting to meropenem.

METHODS

Bacteriological study

In our experiment, we used the spatiotemporal model of antibiotic resistance in motile bacteria [4]. The reference ATCC 27853 strain of P. aeruginosa was precultured in semi-solid LB agar (0.28% agarose) in a Petri dish at 37 °С for 24 h. After 24 h, the cells were harvested from the propagating colony front and seeded onto another Petri dish with semi-solid LB agar. The procedure was repeated 3 times. Then, 10 µl of the grown bacterial culture was picked up with an inoculation loop and introduced into the top layer (semi-solid agar) of the culture medium contained in a device shown in the figure. The medium had a sandwich composition. The bottom layer was LB Miller broth (Becton Dickinson; USA) supplemented with 1.6% agarose, 30 µg/ml kanamycin sulfate, 100 µg/ml cycloheximide, and meropenem taken at one of the concentrations shown in the figure. The optimum thickness of the bottom layer equaled three-fifths of the total medium thickness (~2.0 cm). The bottom layer was distributed into 5 isolated compartments of the dish containing different concentrations of meropenem. The middle layer (one-fifth of the total medium thickness) was LB Miller broth supplemented with 2.0% agarose, 30 µg/ml kanamycin sulfate, 100 µg/ml cycloheximide, and ink (4.0 ml per 1L culture medium) added as a contrasting background for photography purposes. The middle layer spread over the bottom layer was solid. The top layer (one-fifth of the total medium thickness) was semi-solid agar (Miller LB broth) with 0.3% agarose, 30 µg/ml kanamycin sulfate and 100 µg/ml cycloheximide.

The cells were incubated in air at 37 °С for 216 h. Every 12 h, P. aeruginosa samples were collected from the propagating colony front and reseeded on Mueller–Hinton agar (Becton Dickinson; USA) to obtain a sufficient amount of bacteria for the subsequent analysis of their phenotypic traits (antibiotic resistance profiles) and genomic changes.

Resistance to meropenem and imipenem was tested using the agar dilution method described in [7]. Resistance to colistin was assessed using the broth microdilution method following the guidelines of the Clinical and Laboratory Standards Institute (CLSI) [8].

Bacterial DNA was isolated from the 24-h culture of P. aeruginosa grown on Mueller–Hinton agar (Becton Dickinson; USA) using a QIAamp DNA Mini Kit (Qiagen; Germany) according to the manufacturer’s protocol. The obtained DNA samples were stored at –20 °C.

To prepare genomic DNA libraries, 400 g of the isolated bacterial DNA was sheared in an ultrasonicator (Covaris; USA). The fragments were then end-repaired and ligated to MGI adapters (MGI; China). The libraries were purified on Agencourt AMPure XP beads (Beckman; USA). Concentrations of the bacterial DNA and DNA libraries were measured using a Qubit 4 fluorometer (Thermo Fisher Scientific; USA). Whole-genome sequencing was performed using the MGISEQ-2000 platform (MGI; China). Read length was 250 bp.

The quality of the raw sequence data was tested in FASTQC; the reads were trimmed in Trimmomatic v.0.38. Bacterial genomes were assembled de novo using SPAdes 3.14 [9]. The assembled sequences were tested for contamination using Contest16S.  The obtained genome assemblies were evaluated in QUAST 5.0 [10]. Genetic similarity between the assembled genomes was assessed in MUMmer [11]. The genomes were annotated using RAST [12] and Prokka software [13]. To detect the presence of single nucleotide polymorphisms (SNPs), the short reads were mapped to the reference genome in Snippy [14]. ATCC 27853 was used as a reference genome. The detected variants were annotated and their influence on the genes was predicted in SnpEff [15]. The search for antibiotic resistance genes in the genomes assembled de novo, their analysis and validation of the detected SNPs were all carried out in BLASTn. The analysis of resistance determinants was aided by ResFinder and the AMRFinderPlus algorithm included in the NCBI Pathogen Detection pipeline [16, 17]

RESULTS

A total of 93 P. aeruginosa mutants with various phenotypic traits (colony color, antibiotic resistance profile, mucoid/non-mucoid phenotype) were harvested during 216 h of incubation. Among those isolates, two strains (Е62 obtained at 192 h of incubation and Е74 obtained at 216 h of incubation) demonstrated significantly reduced (four- to eightfold) susceptibility to colistin and high resistance to meropenem and imipenem. Phenotypic and genotypic characteristics of these 2 strains are shown in the table.

For E62, meropenem and imipenem MICs were 16 µg/ml and 128 µg/ml, respectively; for E74, they were 16 µg/ml and 256 µg/ml, respectively, which satisfied the CLSI criteria for antibiotic resistance. According to CLSI criteria, the E62 isolate was characterized as susceptible to colistin at increased exposure; for this strain, colistin MIC was 4 times higher than for the baseline strain. According to the CLSI criteria, the Е74 strain was characterized as resistant to colistin (MIC: 4 µg/ml).

Both strains carried a mutation in the porin gene (oprD) resulting in the substitution of glycine with aspartic acid at position 307 of the protein. Besides, both isolates had a missense mutation in the mexD gene (this gene encodes the subunit of the MexCD-OprJ efflux pump). Also, both E62 and E74 had a nonsense mutation in the phoQ gene resulting in the premature termination of protein synthesis (289 out of 448 amino acids).

DISCUSSION

P. aeruginosa strains with simultaneous resistance to carbapenems and polymyxins are not rare. For example, among multidrug resistant P. aeruginosa representatives, 22.2% of meropenem-resistant isolates were unsusceptible to colistin [18]. The evolution of such isolates is rarely described in the literature. It is possible that they acquire their resistance profiles through consecutive or simultaneous therapeutic exposure to carbapenems and colistin. The phenomenon observed in our study proves that P. aeruginosa can reduce their susceptibility to colistin following exposure to meropenem. The hypothetical mechanisms underlying induction of cross-resistance to colistin by meropenem fall into the “all roads lead to resistance” concept, meaning that in P. aeruginosa any stressor causes hypermutability and leads to the emergence of multiple clones with novel properties [3]. Such a mutational explosion can lead to the emergence of persisting mutations disrupting synthesis of lipopolysaccharides, the primary target of colistin.

Genomes of the Е62 and Е74 isolates carried mutations that can explain their resistance to meropenem/imipenem and reduced susceptibility to colistin. The missense mutation in the oprD gene reported in this study may have caused a structural change in the OprD porin, which transports meropenem and imipenem inside the bacterial cell [19]. The search of GeneBank (https://www.ncbi.nlm.nih.gov/genbank) identified only one clinical isolate with a similar amino acid sequence of the OprD porin (GCA_003194245.1). Similar to our mutant, this isolate obtained in 2013 was also resistant to meropenem and imipenem (MIC > 32 µg/ml). Another mutation that could have reduced susceptibility to carbapenems was the missense mutation in the mexD gene encoding the subunit of the MexCD-OprJ efflux pump. The MexCD-OprJ system is involved in the efflux of β-lactams; its hyperexpression is correlated with carbapenem resistance in P. aeruginosa [20]. The phoQ gene codes for the sensor histidine kinase, which is part of the two-component regulatory PhoPQ system. Mutations in phoQ were reported to cause resistance to polymyxins in P. aeruginosa, including specimens isolated from patients with cystic fibrosis [21, 22].

Thus, all phenotypic characteristics of carbapenemresistant isolates of P. aeruginosa with reduced susceptibility to colistin observed in our study were associated with mutations.

CONCLUSIONS

The phenomenon of cross-resistance described in this paper may be due to the fact that the rate of point mutations in P. aeruginosa, specifically in the genes implicated in antimicrobials resistance, increases under stress conditions. Our findings prove that exposure to meropenem can lead to resistance not only to other β-lactams but also to colistin used as a last resort drug for P. aeruginosa infections, which seriously complicates the treatment strategy and limits its options.