Microbial evolution experiments are powerful tools for investigating bacterial adaptation to antibiotic stress under controlled laboratory conditions and are commonly used to assess the causes and dynamics of the evolution of antibiotic resistance in bacteria[7]. According to our previous experiments in a lung infection model in rats, we were only able to find plasmid-mediated resistance instead of QRDR mutations, which were highly detected in clinical studies[8]. In this study, we developed resistance of P.aeruginosa in liquid medium, in vitro, under sub-lethal concentrations of antibiotics 0.5-fold MIC of the parental strain (ATCC27853) to 64-fold MIC, which is completely different from the stepwise selection of resistant mutants in animal models; we were effectively able to block the communication between P.aeruginosa and the environment, and thus focus on the development of P.aeruginosa resistance under increasing antibiotic pressure. Furthermore, in previous reports, many colleagues used agar plates as their medium and then chose a single colony for subsequent passages and sequencing[9]; this method could exclude the emerging mutations which existed in the minorities[10]. Therefore, we incubated bacteria in liquid medium and extracted DNA and RNA from the collected liquids containing the resistant bacteria obtained from the various resistant mechanisms[4], as previously demonstrated in finding the number of QRDR mutations that increased progressively in Streptococcus pneumoniae. This method also helped us detect the low frequency mutation in the process of evaluation, such as parE45 7(S457C) and gyrB466 (S466F).
We found five different mutations in QRDR of which three were consistent with the published results of clinical detection in P.aeruginosa-resistant populations[3, 11], namely gyrA83 (T83I), gyrA87 (D87N) and gyrB466 (S466F); we detected the replacement of the serine residue 457 by a cysteine in parE, which was similar to the clinical results showing an alteration arginine[12]. However, we did not find a parC mutation, which is different from the clinical results, but this was also demonstrated by a study that sequenced both the clinical isolates and the laboratory-derived mutations of Mycoplasma bovis, showing different abilities in the development of resistance[13]. We speculate that the parC mutation could be related to the drug combination or the hospital environment, which is why we used the in vitro model to investigate the mechanisms of P.aeruginosa resistance to overcome the uncertainty of clinical treatment.
Our results showed that the evolutionary trajectories of fluoroquinolone resistance are more complicated than previously described. Firstly, ciprofloxacin and levofloxacin develop resistance through different mechanisms. In our experiments, mutation in gyrA83 occurred in five lineages for levofloxacin and two lineages for ciprofloxacin, while there were two kinds of gyrA87 mutations only under the ciprofloxacin pressure, and the three different gyrA mutations did not appear at the same time. The mutation that occurred in parE was only found in one lineage that evolved under levofloxacin pressure and was found in combination with gyrA83; the mutation in gyrB was always found in combination with the gyrA mutations, and could also disappeared in the evolutionary process. These experiment phenomena were coincident with the clinical results showing that the gyrA mutation is the main resistance mechanism of P.aeruginosa, while the gyrB mutation can reinforce the resistant abilities, and it was rare to find parE mutations[11, 14]. Our study showed that the main mechanism contributing to the resistance of P.aeruginosa, which involved efflux pump expression, was overexpression of MexCD-OprJ, and MexEF-OprN and MexXY-OprM were also up-regulated. These results were in accordance with the clinical detection[12, 15, 16], so our study verified the clinical phenomenon in vitro.
When we analyzed the lineages separately from the perspective of the two different mechanisms, we found a relationship between mutations and overexpression of efflux pumps, in that the gyrA83 mutation could down-regulate the expression of mexC, while gyrA87 and gyrB466 could not, and mexC and mexE may cooperate with each other.
Although levofloxacin and ciprofloxacin are both FQ, they have different abilities in their bactericidal action and induction of resistance. Our study showed that ciprofloxacin had a stronger ability to kill the bacteria, while the mutations and overexpression of efflux pumps suggest that it may render bacteria more susceptible to resistance.