ABSTRACT
Ceftazidime/avibactam is a combination of beta-lactam/beta-lactamases inhibitor, which use is restricted to some clinical cases including cystic fibrosis patients infected with multidrug resistant Pseudomonas aeruginosa, in which mutation is the main driver of resistance. This study aims to predict the mechanisms of mutation-driven resistance that are selected for when P. aeruginosa is challenged with either ceftazidime or ceftazidime/avibactam. For this purpose, P. aeruginosa PA14 was submitted to experimental evolution in the absence of antibiotics and in the presence of increasing concentrations of ceftazidime or ceftazidime/avibactam for 30 consecutive days. Final populations were analysed by whole-genome sequencing. All evolved populations reached similar levels of ceftazidime resistance. Besides, all of them were more susceptible to amikacin and produced pyomelanin. A first event in the evolution was the selection of large chromosomal deletions containing hmgA (involved in pyomelanin production), galU (involved in β-lactams resistance) and mexXY-oprM (involved in aminoglycoside resistance). Besides mutations in mpl and dacB that regulate β-lactamase expression, mutations related to MexAB-OprM overexpression were prevalent. Ceftazidime/avibactam challenge selected mutants in the putative efflux pump PA14_45890-45910 and in a two-component system (PA14_45870-45880), likely regulating its expression. All populations produce pyomelanin and were more susceptible to aminoglycosides likely due to the selection of large chromosomal deletions. Since pyomelanin-producing mutants, presenting similar deletions are regularly isolated from infections, the potential aminoglycosides hyper-susceptiblity and reduced β-lactams susceptibility of pyomelanin-producing P. aeruginosa should be taken into consideration for treating infections by these isolates.
INTRODUCTION
Pseudomonas aeruginosa is an opportunistic pathogen widely distributed in nature (1), which is a major cause of lung and airway infections in hospitalized patients, as well as chronic infections in patients with cystic fibrosis (CF) and chronic obstructive pulmonary disease (2, 3). This bacterial species presents a characteristic low susceptibility to antibiotics, including β-lactams, which is mainly the consequence of its low permeability and the presence in its genome of different intrinsic resistance genes, including those encoding multidrug (MDR) efflux pumps (4) and beta-lactamases. In addition, an increasing number of P. aeruginosa isolates has acquired several resistance genes through horizontal gene transfer (HGT), including different classes of carbapenemases. Finally, P. aeruginosa is able to develop resistance through mutation, particularly when causing chronic infections, to nearly any available antibiotic. Under this situation, the emergence and spread of MDR resistant global clones is of special concern (5).
The use of β-lactam/β-lactamase inhibitor combinations, such as amoxicillin/clavulanic acid or ceftolozane/tazobactam, has proven to be effective against class A β-lactamases (which include narrow and extended-spectrum β-lactamases and some carbapenemases); whereas effective combinations against class B, C (extended spectrum cephalosporinases) and D β-lactamases (6-8) have not been available until recently. One of them is the ceftazidime/avibactam combination, whose use was approved in 2015 by the FDA (9).
Avibactam, formerly known as NXL104, belongs to a new class of β-lactamase inhibitors, the diazabicyclooctanes (10). This inhibitor has a potent activity against most Class A, Class C and some Class D β-lactamases (11). Avibactam has been mainly used for restoring the activity of the third generation cephalosporin ceftazidime (12). Thus far, it has been used for the treatment of patients with complicated urinary tract infections, including pyelonephritis; and community-acquired intra-abdominal infections, usually in combination with metronidazole (13). Besides, future studies are likely to expand the use of ceftazidime/avibactam to include other cases, such as cystic fibrosis patients with MDR resistant P. aeruginosa infections (13).
Given the fact that this treatment is currently reserved for patients who have no alternative therapeutic options, a judicious use of antibiotic stewardship should be applied in order to prevent the incidence of drug resistance. Nevertheless, and although there are numerous studies on the activity of ceftazidime/avibactam against pathogens resistant to other antibiotics (14-17), analysis for predicting potential mechanisms of resistance to this antimicrobial combination are still scarce.
In the present study, experimental evolution followed by whole-genome sequencing (WGS) was used to examine the evolutionary trajectories taken by P. aeruginosa towards resistance against the combination ceftazidime/avibactam compared to the ones followed in absence of avibactam. This may throw light upon the different mechanisms of resistance that are selected for in P. aeruginosa when its β-lactamase activity is inhibited by the presence of this novel inhibitor. In addition, the present work may allow us to elucidate whether the presence of avibactam modifies the resistance level acquired by the bacterial populations in comparison to the one developed when ceftazidime is used alone. Thus, these results may give rise to strategies for predicting, managing, and eventually reducing resistance to ceftazidime/avibactam. This is widely important, as this treatment is strictly restricted to few clinical cases in which resistant strains would be of major concern.
MATERIALS AND METHODS
Growth conditions and antibiotic susceptibility assays
Unless otherwise stated, bacteria were grown in Luria Bertani (LB) Broth at 37°C with shaking at 250 rpm. The susceptibility to tigecycline, tetracycline, aztreonam, ceftazidime, imipenem, meropenem, ciprofloxacin, levofloxacin, norfloxacin, tobramycin, streptomycin, amikacin, gentamycin, colistin, polymyxin B, chloramphenicol, fosfomycin and erythromycin was determined by disk diffusion in Mueller Hinton Agar (MHA) (Sigma) at 37°C. For a set of antibiotics, MICs were determined using E-test strips (MIC Test Strip, Liofilchem®). MICs of ceftazidime and ceftazidime/avibactam were determined in LB by double dilution in microtiter plates.
Experimental evolution procedure
Twelve bacterial populations from a stock P. aeruginosa PA14 culture (four controls without antibiotic, four populations challenged with ceftazidime, and four populations challenged with ceftazidime/avibactam) were grown in parallel in LB for 30 consecutive days. Each day, the cultures were diluted (1/250) in fresh LB. The concentrations of ceftazidime used for selection were increased over the evolution experiment from the concentration that hinders the growth of P. aeruginosa PA14 under these culture conditions (4 μg/ml) up to 128 μg/ml, doubling them every 5 days. The avibactam concentration was maintained constant, as used in clinical tests (18), at 4 μg/ml. In some occasions, the cultures did not grow when antibiotic concentration increased, in which case, the selection was kept at the concentration that allows growth. Every five days, samples from each culture were preserved at -80 °C for further research.
Whole-genome sequencing (WGS)
Gnome® DNA kit (MP Biomedicals) was used to extract genomic DNA. WGS was performed by Sistemas Genómicos S.L. The quality of the extracted material was analysed via a 4200 Tape Station, High Sensitivity assay and the DNA concentration ascertained by real-time PCR using a LightCycler 480 device (Roche). Libraries were obtained without amplification following Illumina protocols and were pair-end sequenced (100 x 2) in an Illumina HiSeq 2500 sequencer. The average number of reads per sample was 7,178,870, which represents a 200x coverage, on average.
Bioinformatics analysis of WGS and confirmation of genetic changes
Mutations in the evolved populations were identified using CLC Genomics Workbench 9.0 (QIAGEN). P. aeruginosa UCBPP-PA14 reference chromosome (NC_008463.1) was used to align the reads obtained from WGS data (previously trimmed). Sanger sequencing was used to verify and to settle the order of appearance of the putative mutations found via WGS (Table S1). Thirty-two pairs of primers, which amplified 200-400 base pair regions containing each genetic modification, were designed (Table S2). After PCR amplification, the corresponding amplicons were purified using the QIAquick PCR Purification Kit (QIAGEN) and sequenced at GATC Biotech.
RESULTS
Stepwise evolution of P. aeruginosa towards ceftazidime and ceftazidime/avibactam resistance
To determine the potential evolutionary trajectories that can lead to either ceftazidime or ceftazidime/avibactam resistance, four biological replicates were allowed to evolve in parallel in each of the following conditions (Figure 1): under selective pressure with ceftazidime (populations 1-4), ceftazidime/avibactam (populations 5-8), and in the absence of any selective pressure (populations 9-12). The susceptibility of each population to the selecting antibiotic was determined every 5 days by E-test. However, after 20 days of evolution, MICs reached the highest limits of the E-test strips and MICs were again determined, for each evolutionary step, by double dilution (Table S3). Stepwise evolutionary trajectories were observed for both treatments, in which the selected populations reached quite similar levels of resistance (Figure 2). These results suggest that avibactam inhibition may not be a guarantee of impeding P. aeruginosa to acquire high-level ceftazidime resistance. An increase in the MIC of an antibiotic after experimental evolution does not necessarily imply that antibiotic-resistant mutants have been selected for: resistance may have arisen due to a phenotypic (inducible) adaptation to the presence of ceftazidime rather than to mutations (19-21). To address this possibility, the evolved populations were cultured in the absence of selection pressure (three sequential passages on LB) and the MICs again determined. These were found not to vary, indicating that the observed modifications were due to the selection of stable mutants.
Cross-resistance and collateral sensitivity of the evolved populations
Taking into consideration the few therapeutic options for patients submitted to ceftazidime/avibactam therapy, knowing whether or not acquisition of resistance to this combination might alter the susceptibility to other antibiotics is of crucial importance. To that end, the susceptibility to a range of representative antibiotics was tested by disk diffusion assay (Table S4). From these results, a set of antibiotics was chosen for determining their MICs against the different evolved populations. Every evolved replicate showed altered susceptibility to antimicrobials belonging to different structural families (Table 1), implying that at least some resistance mutations were not ceftazidime- or ceftazidime/avibactam-specific. All populations evolved in the presence of either ceftazidime or ceftazidime/avibactam presented decreased susceptibility to other β-lactams, to chloramphenicol and to erythromycin and they were more susceptible to fosfomycin and to amikacin. Notably, while populations evolving in the presence of ceftazidime were less susceptible to tetracycline and did not present changes in the susceptibility to tigecycline, populations evolved in the presence of ceftazidime/avibactam were hyper-susceptible to both antibiotics.
Analysis of mutations associated with the acquisition of resistance
To know the genetic events associated with the acquisition of resistance in the evolved populations, the genomes of each, as well as that of the original PA14 strain, were sequenced on the last day of the experiment. Table 2 encompasses the resulting mutated genes and their functional significance, whereas Table S1 shows the locations of all 40 genetic changes that were unveiled and were not present in control populations evolving in the absence of antibiotics. A total of 37 single nucleotide variants (SNVs) and 3 multi-nucleotide variants (MNVs; deletions and substitutions of various nucleotides) were found, 36 located in genes and 4 in intergenic regions. Most mutations located in genes resulted in amino acid alterations, stop codons or frameshifts. In addition, all the populations evolved in the presence of antibiotics contained large chromosomal deletions (55 to 443 kbp) representing from 0,88% to 7,09% of the P. aeruginosa PA14 genome. Five different deletions were selected, they all presenting a 55 kbp common region (Figure 3).
To verify the presence and the order of appearance of the genetic changes identified by WGS, the regions holding these mutations were amplified using specific oligonucleotides (Table S2) and the amplicons Sanger-sequenced in each evolutionary step (Figure 4). Regarding the large chromosomal deletions, primers located at the flanking sequences were used to verify their presence. In all cases, these analyses confirmed the information obtained from WGS.
Common mutations selected upon either ceftazidime or ceftazidime/avibactam selection pressure
Upon one day of experimental evolution, all P. aeruginosa PA14 cultures challenged with antibiotic produced a brown pigment, which appeared to be pyomelanin, whose accumulation is normally due to the lack of homogentisate 1,2-dioxygenase activity provided by the enzyme HmgA (22). All the chromosomal large deletions selected during evolution presented hmgA (Figure 3). In addition, the deletions included as well galU (involved in LPS biosynthesis), whose inactivation reduces ceftazidime susceptibility (23); and the MDR efflux pump mexXY-oprM, which contributes to aminoglycoside resistance in P. aeruginosa (24). Deletion of the latter is likely the cause of the observed amikacin hyper-susceptibility of all evolved populations (Table 1).
Another common element in both evolutions is nalD, which encodes a secondary repressor of MexAB-OprM (25, 26). Three out of eight replicates showed the same T11N change, which has been previously found in XDR P. aeruginosa high-risk clones (27) that overexpress MexAB-OprM. Notably, four replicates (including the three presenting mutations in nalD) also presented mutations in mexB, indicating this efflux system to be a relevant element in the acquisition of resistance. Two other elements that are selected for in both treatments are ftsI and clpA. The first encodes PBP3, the target of different β-lactams (28, 29), which has been already found to be mutated in numerous resistant P. aeruginosa isolates. Indeed, the mutations R504C/H found in populations 1 and 5 are also present among isolates from widespread nosocomial P. aeruginosa clones (29-31). clpA encodes for an intracellular protease involved in different aspects of P. aeruginosa physiology, in addition to aztreonam resistance (28, 32).
Mutations selected by ceftazidime
In addition to the observed mutations in nalD, which would allow mexAB-oprM overexpression, we found as well mutations that should lead to the overexpression of this system in the populations evolving under ceftazidime challenge. Two populations carried mutants in mexR, which encodes a local repressor of mexAB-oprM expression. Another population presented a mutation upstream mexA that might prevent the interaction of NalD with its operator (25), thus allowing mexAB-oprM overexpression.
Other mutations specifically selected for by ceftazidime were found in mpl and dacB. The proteins encoded by these genes are involved in the recycling of peptidoglycan muropeptides. Besides, they control the activity of AmpR and consequently the level of AmpC expression (23, 33), which is known to be a main element in P. aeruginosa resistance to β-lactams. Interestingly, mpl V124G (replicate 2, Table S1) has been found before in a clinical isolate (P. aeruginosa NCGM1984). These finding, along with the ftsI and nalD mutations aforementioned, validate our experimental evolution approach as a valuable predictive model for the in vivo selection of antibiotic resistance.
Finally, mutations at orfN, pitA, infB, grpE, clpP and dnaK were selected for in populations challenged with ceftazidime. orfN codes for a putative glycosyl transferase of type A flagellins (34). Mutations on this gene have been found in ciprofloxacin P. aeruginosa resistant strains (35), and also in P. aeruginosa populations submitted to tigecycline and tobramycin experimental evolutions (Sanz-García et al. submitted). pitA encodes a phosphate transporter, infB the translation initiation factor IF-2 and dnaK, grpE and clpP encode proteins involved in regulatory gene networks involved in response to stress. None of them has been previously related to ceftazidime resistance, excepting dnaK, whose inactivation leads to stronger susceptibility to various antimicrobials in Escherichia coli (36).
Mutations selected by ceftazidime/avibactam
The challenge with ceftazidime/avibactam selected mutants in a predicted efflux pump (PA14_45890-45910), as well as in the two-components system (TCS) encoded by the operon PA14_45870-45880, likely regulating its expression. Previous studies have shown this efflux pump to be involved in P. aeruginosa intrinsic resistance (37) and susceptibility to carbapenems (38). Regarding the substrate recognition profile this pump might display, it is remarkable that populations 5 and 7, which present the aforementioned mutations, show a much lower susceptibility to imipenem than any other replicate (Table 1), suggesting this pump to have certain specificity to carbapenems.
Other mutations that were selected upon ceftazidime/avibactam treatment were found in pepA, spoT, dnaJ and flgF. pepA encodes for a protease necessary for P. aeruginosa cytotoxicity, virulence and, consequently, lung infection (39, 40). Although its implication in antibiotic resistance has not been studied in detail, it has been reported that its inactivation confers meropenem resistance in P. aeruginosa (41). Moreover, pepA mutants are selected in the presence of aztreonam (28). SpoT has been related to piperacillin resistance (42); while DnaJ, a chaperone protein and FlgF, a flagellar basal body rod protein (43), have been reported to modify the susceptibility of E. coli to a range of antibiotics when they are inactivated (36).
The other mutations that were selected for in populations under ceftazidime/avibactam challenge; namely those occurring in ctpA, an essential gene for the transition between acute and chronic P. aeruginosa infection (44), pcm, that encodes for a L-isoaspartate carboxylmethyltransferase type II that participates in protein repair and degradation and glnD, which is implicated in N2 metabolism, (45) have not been reported to be involved in antibiotic resistance.
Discussion
The use of β-lactamase inhibitors has re-emerged as a fruitful strategy for fighting infections by MDR bacteria. Among them, ceftazidime/avibactam can be a useful combination for treating infections by different organisms, including P. aeruginosa. The analysis of the mechanisms of resistance to previous β-lactam/β-lactamase inhibitor combinations as amoxicillin/clavulanate have shown that the main mechanisms selected along their use have been increased expression or mutation of pre-existing β-lactamases and acquisition of new ones by HGT (46-50). P. aeruginosa has already acquired different carbapenemases that might be important elements in ceftazidime/avibactam resistance. In addition, resistance can be achieved through mutations, particularly in the case of P. aeruginosa causing chronic infections. To identify potential mutations involved in the acquisition of either ceftazidime or ceftazidime/avibactam resistance, bacterial populations were submitted to increasing selective concentrations of these antimicrobials. In both cases, the first event in the evolution seems to be the deletion of large regions of P. aeruginosa chromosome that comprise, among several other genes, hmgA, galU and mexXY. A similar situation has been previously reported in other P. aeruginosa experimental evolution assays in the presence of β-lactams, such as piperacillin (42) and meropenem (51). Additionally, pyomelanin-producing mutants are regularly isolated from infections; up to 13% of CF patients harbour pyomelanin-producing mutants (52), likely because the production of pyomelanin increases resistance to oxidative stress and persistence in chronic lung infections (22). Recent work has also shown that these mutations can be selected to prevent bacteriophage predation (53). Notably, melanogenic clinical isolates of P. aeruginosa present large chromosomal deletions, similar to those reported in the present work (54). Our results then support that ceftazidime selects for these genome deletions, and the presence of avibactam cannot prevent them from happening. It might be possible that deletions are the consequence of increased recombination triggered by the presence of the antibiotic. However, the fact that P. aeruginosa evolving in the presence of ciprofloxacin do not produce pyomelanin (a marker of these deletions) and that pyomelanin-producing mutants are selected when a recA P. aeruginosa defective strain is challenged with either ceftazidime or ceftazidime/avibactam (data not shown), goes against this possibility. Besides the already known effect of the lack of galU on the susceptibility to β-lactams, the absence of other genes located in the deletion, such as mexXY, may affect P. aeruginosa susceptibility to antibiotics. Deletion of this pump is likely the cause of the observed hyper-susceptibility to amikacin of the evolved populations. In addition, it might have an indirect effect on the decreased susceptibility to beta-lactams, particularly in the case of those strains carrying mutations in the repressors of mexAB-oprM.
MexAB-OprM is an important determinant of intrinsic P. aeruginosa resistance to different antibiotics, including β-lactams (24). Further, mutants overexpressing this efflux pump are regularly isolated from infections and it has been shown its expression to be prevalent among resistant P. aeruginosa clinical isolates (55-58). MexAB and MexXY share the outer membrane protein OprM, which produces antagonistic interactions when both systems are expressed (51, 59). Hence, MexXY-OprM elimination might favour the efficiency of β-lactams efflux, reducing the competition of both efflux pumps for OprM.
Important elements in the acquisition of ceftazidime resistance are efflux pumps, particularly MexAB-OprM, since mutations either in the elements regulating its expression or in the efflux pump itself were found in six out of eight evolved populations, whereas the two remaining populations harboured mutants in the putative PA14_45890-45910 and in its potential TCS regulator. While the substrates of MexAB-OprM are known and include β-lactams, the substrates of PA14_4590-45910 are unknown. Nevertheless, it is remarkable that populations presenting mutations on this determinant display a much lower susceptibility to imipenem than any other replicate, suggesting this pump to have certain specificity to β-lactams.
Mutations in elements involved in the regulation of AmpC expression were selected when just ceftazidime was used for selection and not in the presence of ceftazidime/avibactam. This suggests that, at least in the P. aeruginosa PA14 background, the efficient inhibition by avibactam of intrinsic β-lactamases, preclude the emergence of mechanisms based on their overexpression and other mechanisms, including the above mentioned large deletions and modifications in the activity of efflux pumps are preferentially selected. This does not necessarily mean that resistance to ceftazidime/avibactam cannot be associated to changes in the activity of AmpC, particularly if the challenged isolate is already resistant to ceftazidime. Indeed, avibactam resistant mutants presenting changes in the avibactam binding pocket of AmpC are selected in vitro at low frequency from Amp C-overexpressing ceftazidime resistant P. aeruginosa isolates (60).
Although most of the mutants here reported have been previously associated to be involved in antibiotic resistance, it is still possible that some of the mutations might be selected for compensating the fitness costs associated with the acquisition of resistance. This might be the case of ctpA, pcm or the mutations at structural elements of efflux pumps that were selected after mutations in the regulators of their expression. For the latter, it might also be possible that these mutations increase the capability of extruding the antibiotic substrates, as described for AcrB (61). The fact that in all evolved populations mutants in efflux pumps are selected, provides an explanation of the cross-resistance phenotype observed in all resistant strains. This situation might be of concern, since both ceftazidime and ceftazidime/avibactam might select for resistance to other antibiotics, at least along chronic infections in which mutation is the main cause of acquisition of resistance.
P. aeruginosa evolution in chronic infections frequently involves large genome deletions (62), usually linked to the production of pyomelanin (54). Whether these deletions are selected by antibiotic treatment or are just the consequence of the adaptation to the environment of the lungs of the CF patient remains to be established. However, this evolution provides a link between antibiotic resistance and virulence for this relevant pathogen. In any case, and given that deletions containing galU and hmgA appear to be a first step on the evolution towards ceftazidime/avibactam resistance, pyomelanin production could be considered as a marker in the selection of the antibiotic of choice for treating P. aeruginosa infections. Both in vitro work, including the results here shown, and the analysis of clinical pyomelanin-producers, have shown that these isolates are hyper-susceptible to aminoglycosides probably because the deletions they present include mexXY. It would then be judicious using aminoglycosides, and not β-lactams for treating infections by pyomelanin-producing P. aeruginosa.
Acknowledgments
Work in our laboratory is supported by grants from the Instituto de Salud Carlos III (Spanish Network for Research on Infectious Diseases [RD16/0016/0011]), from the Spanish Ministry of Economy and Competitivity (BIO2017-83128-R) and from the Autonomous Community of Madrid (B2017/BMD-3691). The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication. FSG is the recipient of a FPU fellowship.