Although it is known that the phenotypic and genotypic characteristics of P. aeruginosa populations within the CF lung fluctuate over time [9, 16], the factors that are responsible for this diversification are not fully understood. When P. aeruginosa LESB58 was grown in ASM with and without sub-inhibitory concentrations of antibiotics, we observed differential effects of antibiotics commonly used to treat CF patients on the diversity of LESB58 populations in the ASM model. In particular, increased levels of phenotypic diversification occurred in LESB58 populations grown in ASM when sub-inhibitory concentrations of colistin, ceftazidime and azithromycin were present. However, extensive diversification of the P. aeruginosa populations was not seen in the presence of sub-inhibitory concentrations of meropenem.
There are a number of mechanisms by which sub-inhibitory concentrations of antibiotics could potentially enhance bacterial diversification. One potential mechanism could involve the antibiotics inducing mutagenesis within bacterial populations, causing variation and/or promoting the hypermutability phenotype [31–34]. A second potential mechanism could involve the antibiotics acting as signalling molecules, altering the QS systems within bacterial populations and subsequently promoting social evolution and diversification [35, 36, 38]. Antibiotic exposure has been shown to induce mutagenesis by triggering the SOS response and thus increasing the expression of error-prone DNA polymerases, which could give rise to diversity within bacterial populations [31–34]. It is possible that ceftazidime induced mutagenesis in the LESB58 populations through the induction of the SOS response. It has been suggested that this increase in mutagenesis is an adaptive strategy that favours the acquisition of antimicrobial resistance and survival in harsh environments [33, 39]. However, it has been argued that the generation of genetic variants within the CF lung does not require the SOS response, and that starvation and oxidative stress caused by antibiotic exposure can promote diversity within P. aeruginosa biofilms [31, 40–42].
The hypermutable phenotype occurs as a consequence of defects in error avoidance or DNA repair genes, typically termed anti-mutator genes . It has been suggested that hypermutability, promoted by extrinsic and intrinsic factors, is the driver of P. aeruginosa adaptation and survival in the CF lung [44, 45]. Although phenotypic diversification of LESB58 was observed following culture in ASM, especially when sub-inhibitory concentrations of colisitin, ceftazidime or azithromycin were present, no hypermutable isolates were detected. In our previous study using LES isolates from multiple CF patients, we found hypermutable sub-types but only at low frequency . In this study we found no evidence that hypermutability was driving this diversification and adaptation process. This supports work by Ciofu et al. who found that the hypermutability phenotype was not essential for the acquisition of mucoidy and loss of QS. Other studies have also suggested that spontaneous mutation and mutator strains are not required to produce genetic variants in populations of P. aeruginosa within the CF lung [40, 46].
It has been shown that sub-inhibitory concentrations of antibiotics can act as signalling molecules that regulate bacterial gene transcription, physiology and virulence [36, 38, 47–51]. In particular, tobramycin, colistin and azithromycin at sub-inhibitory concentrations have been shown to modulate the QS networks in bacterial populations [35, 36, 38]. These antibiotics are commonly used to treat CF patients and, therefore, the signalling activities of these antibiotics could increase bacterial fitness for survival in the harsh environment of the CF lung , suggesting that the classical view of antibiotics acting only to reduce bacterial fitness and virulence is not always the case. In the current study, across all the ASM cultures, no single dominant phenotypic variant emerged. Some patterns in the diversification process were evident. For example, isolates lacking the pyocyanin production phenotype occurred following culture in ASM with ceftazidime or colistin. However this was only evident in two out of the three biological replicates (ASM + Ceftazidime: 27.5% and 40% of the isolates; ASM + Colistin: 42.5% and 40% of the isolates), highlighting the variability between replicates. A previous study by Cummins et al. has shown that sub-inhibitory concentrations of colisitin actually increases pyocyanin production. Pyocyanin production is regulated by QS, which relies upon small diffusible signal molecules interacting with transcriptional activators to couple gene expression with cell population density. Although QS-deficiency is a common feature amongst P. aeruginosa CF isolates [16, 52, 53], QS regulates a number of factors of relevance to CF, including pyocyanin and LasA production . Our previous studies suggested that LES populations in CF comprise a mixture of QS-positive and QS-deficient bacteria [7, 9, 54], which is what we have observed in this study in ASM. The QS-deficient populations could benefit at the cost of QS-positive populations.
The main phenotypic variations involved changes in colony morphology, pyocyanin production and antimicrobial susceptibilities. A high diversity in the antimicrobial susceptibility profiles of CF isolates within adult sputum samples has been demonstrated previously , highlighting the limitations of performing antimicrobial susceptibility tests on a single isolate from a CF patient sputum sample. It was also shown that using one antibiotic could lead to enhanced resistance to a different, unrelated antibiotic . A similar pattern was observed in this study, when exposure to one antibiotic altered the antibiotic susceptibility profiles to unrelated antibiotics. In particular, exposure to azithromycin, tobramycin or ceftazidime led to an increase in resistance to tazobactam/piperacillin. This could have serious clinical consequences for the CF patient, in terms of the generation of antimicrobial resistant P. aeruginosa populations, because CF patients are regularly exposed to a number of different antibiotics. In our study, the presence of meropenem had a weaker effect on diversification compared to the other antibiotics, despite having a similar mechanism of action to ceftazidime. However, it is possible that cell death was occurring in these populations, since the numbers of cells obtained following culture were generally lower. This is despite the meropenem concentration in ASM being 8-fold less than the minimum inhibitory concentration of this antibiotic. Therefore, the apparent reduction in diversity could be attributed to the populations exhibiting cell death. This suggests that there may be a clinical advantage to using some antibiotics (eg. meropenem) rather than others. It would also be interesting to analyse combinations of two antibiotics, since it is often the case that dual therapy is used clinically. The identification of individual mutations within the LESB58 populations to explain the changes in individual phenotypic traits would have been beyond the scope of this work.