Differential impact of ramRA mutations on both ramA transcription and decreased antimicrobial susceptibility in Salmonella Typhimurium

Objectives: This study was focused on analysing the heterogeneity of mutations occurring in the regulators of efﬂux-mediated MDR in Salmonella Typhimurium. Moreover, the impact of such mutations on impairing the transcription of ramA , acrB , tolC and acrF was also assessed as was the impact on the resistance or decreased susceptibility phenotype. Methods: Strains were selected in vitro under increasing ciproﬂoxacin concentrations. Etest and broth micro-dilution tests were used to determine the MICs of several unrelated compounds. Screening of mutations in the quinolone target genes and MDR regulators was performed. RT–PCR analysis was used to detect the levels of expression of acrB , tolC , ompF , acrF , emrB , acrR , ramA , soxS and marA . Results: All mutant strains showed increased MICs of most of the antimicrobials tested, with the exception of kanamycin. Mutations in the quinolone target genes did not occur in all the mutants, which all harboured mutations in the ramRA regulatory region. All the mutants overexpressed ramA , tolC and acrB (only tested in 60-wt derivatives), whereas differential results were seen for the remaining genes. Conclusions: Mutations in the ramRA region related to resistance and/or decreased susceptibility to antimicrobials predominate in Salmonella . There is heterogeneity in the types of mutations, with deletions affecting RamR-binding sites having a greater impact on ramA expression and the MDR phenotype.


Introduction
The ever-increasing levels of resistance to antimicrobial compounds are of great concern, particularly for pathogens of clinical relevance. Salmonella enterica serovar Typhimurium is a pathogen distributed worldwide that typically causes gastroenteritis in humans. 1 Fluoroquinolones and cephalosporins are the current first-line treatments; however, recent data have revealed that in particular geographical areas, such as China, high percentages of resistance to compounds such as nalidixic acid (61.9%) and cefepime, cefotaxime and ceftazidime (90%) have already been detected. 2 Quinolone resistance has been widely studied in Enterobacteriaceae, particularly in Escherichia coli and S. enterica. 3 In E. coli, the mechanism that largely contributes to resistance and/or decreased susceptibility to quinolones is the acquisition of mutations located in the genes encoding the two quinolone targets: DNA gyrase (gyrA and gyrB) and topoisomerase IV (parC and parE). 4,5 These mutations are usually acquired in the quinolone resistance-determining regions (QRDRs) detected in each of the target genes. 3 On the other hand, increased drug extrusion by means of the overexpression of AcrAB-TolC, the main efflux pump described in Enterobacteriaceae, 3 is also of great concern since it confers cross-resistance to several unrelated compounds, including antimicrobial drugs. 6,7 To a lesser extent, other efflux systems, such as AcrEF and EmrAB, have been reported to participate in the extrusion of antimicrobial compounds. 8,9 In Salmonella, increased efflux has been described as the primary mechanism in quinolone resistance acquisition. 10 Alternatively, decreased production of the OmpF porin has at times been related to the MDR phenotype 11,12 despite controversial data suggesting no clear role in S. enterica. 13 Several regulators have been reported to influence the expression of the acrAB operon in Salmonella. AcrR is the local repressor encoded upstream of the acrAB genes and mutations within its coding sequence have been associated with increased expression of the pump. 14 In addition, three homologous transcriptional activators, RamA, SoxS and MarA, have been associated with increased acrB and tolC expression levels. While clear associations have been reported for enhanced production of SoxS and RamA and overexpression of acrAB, 11,15,16 only indirect results have associated greater production of MarA with increased levels of resistance, supposedly mediated by higher levels of AcrAB. 11,17 In terms of regulation, each of these three activators has its own regulator: RamR, SoxR and MarR, respectively. 3 In terms of the MDR phenotype, the clinical relevance of mutations located in the genes encoding these latter regulators has been clearly shown for RamR, 18,19 while there have been few reports for mutations located in the soxRS region. 11,15 Concerning MarA, even though its overexpression has been detected in MDR S. enterica strains, 8,20 the putative responsible mutations in the marRAB region have not been mapped. Naturally occurring mutations in this region have been widely reported in E. coli, 21,22 whereas, to our knowledge, such mutations in S. enterica have only been reported in a single study, associating it with high MarA overexpression and an MDR phenotype. 12 The aim of this study was to determine the mechanisms involved in increasing the MICs of different antimicrobial agents for a collection of Salmonella Typhimurium mutants selected in vitro, particularly when studying strains with low MICs of ciprofloxacin and their derivative mutants selected at the initial steps of drug exposure following a stepwise procedure. The mechanisms studied included target gene mutations and the expression of several genes involved in decreasing the intracellular concentration of the drug. Moreover, and as a novel approach, we also assessed the role and heterogeneity of ramRA mutations and their impact on increasing the expression of ramA and the phenotype of decreased susceptibility to multiple antibiotics or MDR.

Bacterial strains and selection of resistant mutants
Two Salmonella Typhimurium clinical isolates, strains 59-wt and 60-wt, were recovered from independent stool samples in the Department of Clinical Microbiology at the Hospital Clinic of Barcelona, Spain. Strain 59-wt has previously been characterized, as have its derivative mutants displaying increasing ciprofloxacin MICs, including the highly resistant mutant 59-64. 23 As indicated, the clinical isolate 59-wt was grown at 378C on MacConkey agar plates in the presence of ciprofloxacin (Fluka) in a multistep selection process with doubling concentrations of the drug. 23 Single colonies were randomly selected at different steps and previously characterized. In the present study, we characterized additional randomly selected colonies during the process (59-mut1, 59-mut2 and 59-mut3) to assess the occurrence of heterogeneity in the mechanisms of resistance. Likewise, strain 60-wt was similarly treated and exposed to increasing ciprofloxacin concentrations and two different mutants were randomly selected (60-mut1 and 60-mut2).

Susceptibility testing
The MICs of several quinolones and unrelated antimicrobial compounds were determined by Etest (AB Biodisk) according to the manufacturer's recommendations and interpreted according to CLSI guidelines. 24 The broth microdilution method was used to evaluate the MICs of ciprofloxacin, norfloxacin and nalidixic acid when maximum Etest values were reached. The compounds tested were ciprofloxacin, norfloxacin, nalidixic acid, chloramphenicol, tetracycline, erythromycin, amoxicillin, ceftriaxone, kanamycin and cefoxitin.

Detection of mutations within the QRDRs and regulatory loci
Mutations acquired in the QRDRs of the gyrA, gyrB, parC and parE genes as well as in the MDR regulatory loci soxRS, marRAB, acrR and ramR were screened by PCR amplification as described previously. 25 Amplicons were purified and sent to Beckman Coulter Genomics (Essex, UK) for sequencing reactions. Detection of mutations was carried out using BioEdit w software (Ibis Biosciences, Carlsbad, CA, USA) by comparison with the genome of Salmonella Typhimurium LT2 as the reference strain (RefSeq: NC_003197.1).

RNA extraction and real-time PCR
Bacterial pellets were obtained as described previously. 25 Briefly, strains were grown in LB broth at 378C with shaking to reach exponential phase (OD 600 ¼ 0.6). Four millilitres of bacterial cells was treated with 8 mL of RNA Protect Bacteria Reagent (Qiagen) and subsequently incubated with Tris-EDTA buffer supplemented with lysozyme. RNA extractions were obtained using the Maxwell w 16 Research Instrument (Promega) and the Maxwell w 16 LEV simplyRNA Blood Kit (Promega) following the manufacturer's recommendations. Five independent RNA extractions were made.
The acrB, tolC, ompF, acrF, emrB, ramA, marA, soxS and acrR genes were analysed by RT-PCR following previously described conditions. 26 The 16S rRNA gene was used as an internal control for normalization and susceptible strains 59-wt and 60-wt were the reference strains for their respective derived mutants. The 2 2DDCT method was used for relative gene expression calculations. 27 Five independent assays were performed and each RNA sample was tested in triplicate. The primers used are reported in Table 1. Mean (+SD) values are detailed in Table 2.

Results and discussion
Quinolone resistance and the MDR phenotype Three and two derivative mutants were selected from the quinolone-susceptible clinical isolates 59-wt and 60-wt, respectively. Susceptibility testing to several unrelated compounds was used to determine the acquisition of the quinolone resistance and MDR phenotypes ( Table 3). The term MDR has been defined as resistance to one agent in three or more antimicrobial categories 28 or to four or more antimicrobials in the particular case of non-typhoidal Salmonella. 29 In the present study, we used instead the term decreased susceptibility to multiple antibiotics when increased MICs of more than four antimicrobial compounds were seen even though the resistance breakpoints were not reached. Strain 59-64, already characterized in a previous study, 23 was also included in the present work for comparison with the mutants. The results showed that in comparison with their WT strain, all selected mutants had increased MICs (1.5-to .8-fold) of all the drugs tested, except for kanamycin, for which no increase was recorded. Only 59-wt derivative mutants showed the acquisition of QRDR mutations (Table 4). Strains 59-mut1 and 59-mut2 showed a similar genetic background in terms of target gene mutations. However, higher MIC values were seen for 59-mut2 concerning all the drugs (except for amoxicillin and chloramphenicol, which had already shown maximum Etest values for 59-wt, and tetracycline). Likewise, on comparing strains 60-mut1 and 60-mut2 a similar conclusion was obtained, with higher MIC results seen for 60-mut2 despite having background similarity.
In accordance with the fact that strains 59-mut3 and 59-64 were selected at higher ciprofloxacin concentrations, these strains showed the highest MICs, mostly concerning quinolones, being maximal for strain 59-64.
Taking into account the increased MICs of most of these compounds for all the mutants, and the fact that increased efflux confers a cross-resistance phenotype by means of increased AcrAB or even a hitherto uncharacterized efflux pump, 6,23 enhanced extrusion activity was the most likely mechanism underlying this phenotype. Moreover, the results obtained from 60-wt and its derivative mutants strengthen the idea that efflux is selected at primary stages of the process of quinolone resistance acquisition as suggested previously 10,25 and this mechanism is selected even before target gene mutations. On the contrary, it should be noted that mutants selected in a single step-selection process, usually performed at concentrations higher than the initial MIC, may follow a different pattern of acquisition of resistance mechanisms.

Expression of structural genes involved in MDR
Gene expression analysis was performed to determine the expression patterns of genes related to bacterial efflux and permeability. The results were interpreted after comparison of the expression  ramRA mutations and impact on MDR levels of each clinical isolate with their respective mutant derivatives. The genes studied were acrB, tolC, ompF, acrF and emrB ( Figure 1 and Table 2). Overexpression of the AcrAB-TolC efflux pump has been reported as the most relevant mechanism in terms of efflux. 3 In the present study, acrB was only analysed in 60-wt and its derivatives, which all overexpressed this gene (5.2-to 9.5-fold), since it was reported that 59-wt has a mutation inactivating the acrAB operon. 23 The tolC gene was found to be consistently overexpressed in all the mutants (≥2.3-fold), particularly for strains 59-mut2 and 60-mut2 (6.2-and 5.4-fold, respectively). In contrast, ompF always showed decreased expression with the strongest results being seen in strains 59-64 (23.3-fold) and 60-mut2 (22.2-fold). From these results, we suggest that AcrAB-TolC was involved in the phenotype of decreased susceptibility to multiple antibiotics in the case of 60-wt derivatives, whereas an unknown efflux system, likely acting in conjunction with TolC, participated in the case of 59-wt derivatives. Next, we assessed other efflux-related genes, such as acrF and emrB ( Figure 1 and Table 2), which may play a secondary role in antibiotic resistance. 8,9 Our results showed that only two strains clearly overexpressed acrF [59-mut2 (6-fold) and 60-mut2 (4.9-fold)], whereas emrB showed a slightly decreased expression in all the mutants (21.2-to 21.9-fold). Thus, we can only suggest a role in increasing the MICs mentioned for the AcrEF efflux system in these two particular mutants, one of which is also an AcrAB overproducer (60-mut2).

Expression of the MDR regulators: the key role of ramA
In addition to the analysis of these structural genes, we also studied the levels of expression of the AcrAB regulators: acrR, ramA, soxS and marA ( Figure 1 and Table 2). We could not find a clear interpretation for acrR expression. In contrast, ramA was overexpressed in all the mutants, thereby suggesting this regulator as the cause of the increased MICs for both the mutants overexpressing acrB and those overexpressing an unknown efflux system. Similar results have also highlighted the greater importance and prevalence of increased RamA over that of the other regulators. 16,30 Maximal ramA expression levels were seen for 59-mut2 and 60-mut2 (66-and 74.2-fold, respectively) above the levels detected for the remaining mutants (13.4-to 19.6-fold). In line with these results, these two strains also showed higher MICs and acrB (only reported for 60-wt derivatives) and tolC expression values in comparison with their closely related mutants 59-mut1 and 60-mut1, respectively. In addition, as mentioned above, 59-mut2 and 60-mut2 were also reported to clearly overexpress  acrF. This latter association between high ramA expression (.60-fold in the present study) and tolC and acrF overexpression agrees with a previously reported study. 31 The soxS expression values detected in the present study were ,2-fold higher in most of the mutants versus the expression levels seen in the two clinical isolates (Figure 1). Only two mutants, strains 59-64 and 60-mut2, showed overexpression of .4-fold. However, it was not possible to consistently associate this trait with higher expression values of ramA or acrF in both mutants. In contrast, these two strains did show the minimum levels of ompF expression (23.3-and 22.2-fold, respectively). Similarly, marA transcription also showed ≤2-fold increased expression in three mutant strains: 59-mut1, 59-mut3 and 60-mut1. In contrast, the highest levels were seen in 59-mut2 (4.3-fold), 59-64 (3.7-fold) and 60-mut2 (3.6-fold).
To understand our results, it is worth mentioning that RamA-binding sites have already been reported in Salmonella concerning the acrAB and tolC promoters. 32 The 20 bp sequences recognized by this regulator resemble those initially reported to be present in all members of the marA/soxS/rob regulon in E. coli. 33 It has been described that most of the residues of the two helix-turn-helix motifs (important for DNA sequence recognition) of MarA from E. coli are conserved in RamA from S. enterica serovar Paratyphi B. 34 Moreover, it has previously been reported that the marRAB promoter contains its own marbox sequence. 33 In agreement with this, RamA from Salmonella Paratyphi B has been shown to bind the MarA operator of E. coli. 34 Thus, the binding sites characterized for MarA and SoxS in E. coli, equally termed marbox or soxbox, are similar to the already-mentioned rambox in Salmonella. 31 Unravelling the mutations leading to the phenotype of decreased susceptibility to multiple antibiotics In order to determine the mutations underlying the resistance phenotypes, sequencing and detection of mutations was performed in all the strains for all known regulators of MDR (acrRA, ramRA, soxRS, marRAB and acrSE). The results revealed the acquisition of mutations in the ramRA loci for all the mutants (Table 4). Mutations were located within the ramR coding sequence, either leading to a single amino acid substitution (Gln-19 Pro, strain 60-mut1) or even deletions of 44 and 6 nt (strains 59-mut1 and 59-mut3, respectively). Surprisingly, the two strains (59-mut2 and 60-mut2) with the highest ramA overexpression values harboured a similar genotype: 6 nt deletion and 16 nt deletion, respectively, in the ramA promoter. Lastly, and as previously reported, 23 strain 59-64 showed a single nucleotide change also located in the ramA promoter.
Previous reports have revealed that mutations or gene interruptions can be either acquired within ramR or in the ramA promoter. 11,16,30 However, no association has ever been made between the type of mutation and transcription levels of ramA. The results observed in the present study point out that severe ramRA mutations and impact on MDR 621 JAC nucleotide deletions located in the ramA promoter have a higher impact on increasing the expression of this regulator, whereas mutations within ramR or single nucleotide changes in the ramA promoter have a lesser effect. We performed an exhaustive analysis of the literature looking for studies that determined both ramA transcription levels and ramRA mutations in strains with resistance or decreased susceptibility to fluoroquinolones. Studies conducted in serovars Typhimurium, 36,37 Enteritidis, 11 Kentucky 38 and other serovars 30 were found to report similar results (Table 4). In order to understand this situation, it is necessary to note that RamR has been reported to bind as a homodimer to two RamR-binding sites located in the ramA promoter ( Figure 2). 37 Thus, taking into account all this information, we hypothesize that important deletions occurring in these binding sites seriously impair the RamR repressive activity by preventing RamR binding and lead to high levels of ramA expression (.60-fold). In contrast, mutations or deletions occurring in RamR or single nucleotide modifications affecting one binding site do not seem to abolish repression to the same extent and lead to moderate levels of ramA transcription (less than 40-fold). This latter situation would be supported by the capacity of the mutated form of RamR to partially preserve its repressive activity or by the existence of other regulators capable of binding to the ramA promoter even in the absence of a functional RamR protein.
Nonetheless, to our knowledge, two exceptions have been reported: one Salmonella Kentucky strain 38 and one Salmonella Paratyphi B mutant (Table 4). 30 The former situation might be explained by a large deletion detected at the very beginning of the repressor (affecting the protein sequence from the amino acid at position 14), whereas no clear explanation could justify the latter situation. Therefore, in order to elucidate the role of these mutations and strengthen or not our hypothesis, a larger number of strains need to be analysed in further studies.
In no strain did we find any mutation in any of the other regulatory sequences analysed in the present study. Consequently, we are unable to explain the increased soxS transcription reported in 59-64 and 60-mut2. Concerning acrF overexpression, previous results have associated it with mutations within the acrS gene or in the acrEF promoter. 9 However, in the present study, no mutation in the acrSE regulatory region could explain our findings. Instead, and as previously mentioned and reinforced by our results, overexpression of this efflux component is related to the levels of ramA transcription. 31 High levels of ramA expression trigger acrF overexpression, whereas intermediate levels do not. In line with these results, a previous study has also associated nucleotide deletions in the ramA promoter with acrEF overexpression. 36 In view of these findings, the regulatory network that controls the expression of genes involved in the phenotype of decreased susceptibility to multiple antibiotics or MDR still needs further research to completely understand the bacterial response for survival under antimicrobial exposure. Nonetheless, we must keep in mind that our observations have arisen from mutants selected in a stepwise process, which may harbour additional mutations with unknown influence. Additional experiments are required in order to validate these results.

Conclusions
The results of our study indicate that RamA overexpression leads to the phenotype of decreased susceptibility to multiple antibiotics by using two different efflux-related strategies: overexpression of AcrAB and overexpression of a hitherto uncharacterized efflux pump. Moreover, we provide further evidence of the prevalence of ramRA mutations versus other acrB regulators in the acquisition of MDR. However, heterogeneity was observed in the types of mutations acquired, which may be associated with different levels of ramA transcription. Large deletions affecting the RamR-binding sites in the ramA promoter were observed in strains with higher ramA transcription levels, a trait that may account for the highest expression levels of acrB, tolC, marA and acrF, hence related to a major contribution to the phenotype of decreased susceptibility to multiple antibiotics.