Exposure of Salmonella enterica Serovar Typhimurium to High Level Biocide Challenge Can Select Multidrug Resistant Mutants in a Single Step

Background Biocides are crucial to the prevention of infection by bacteria, particularly with the global emergence of multiply antibiotic resistant strains of many species. Concern has been raised regarding the potential for biocide exposure to select for antibiotic resistance due to common mechanisms of resistance, notably efflux. Methodology/Principal Findings Salmonella enterica serovar Typhimurium was challenged with 4 biocides of differing modes of action at both low and recommended-use concentration. Flow cytometry was used to investigate the physiological state of the cells after biocide challenge. After 5 hours exposure to biocide, live cells were sorted by FACS and recovered. Cells recovered after an exposure to low concentrations of biocide had antibiotic resistance profiles similar to wild-type cells. Live cells were recovered after exposure to two of the biocides at in-use concentration for 5 hours. These cells were multi-drug resistant and accumulation assays demonstrated an efflux phenotype of these mutants. Gene expression analysis showed that the AcrEF multidrug efflux pump was de-repressed in mutants isolated from high-levels of biocide. Conclusions/Significance These data show that a single exposure to the working concentration of certain biocides can select for mutant Salmonella with efflux mediated multidrug resistance and that flow cytometry is a sensitive tool for identifying biocide tolerant mutants. The propensity for biocides to select for MDR mutants varies and this should be a consideration when designing new biocidal formulations.


Introduction
Salmonella enterica serovar Typhimurium is a major cause of gastrointestinal disease and, as with many bacteria, Salmonella infections are becoming harder to treat because a high percentage of isolates are now resistant to commonly used antibiotics. Due to this increasing prevalence of resistance in pathogenic isolates there is a greater than ever requirement for effective cleaning and disinfection regimes to prevent infections and contain outbreaks. These regimes rely on the effective use of biocides, there is concern that the resulting increased use of biocides in farming, food production, hospital settings and the home is contributing to the selection of antibiotic resistant strains as some mechanisms of biocide resistance also confer antibiotic resistance [1]. Biocides incorporate disinfectants, antiseptics and preservatives, compounds that are often composed of a mixture of ingredients that act upon a wide range of cellular mechanisms and targets. This wide target base makes it difficult for bacteria to become resistant to biocides [1]. However, biocides are not always used at the correct concentration and can become compromised by coming into contact with organic material [1]. Bacteria that survive a low level dose of biocide are more likely to be resistant to antibiotics [2]. A single exposure to some biocides has previously been found to be insufficient to select for multidrug resistant (MDR) strains [3], however, repeated, sub-inhibitory exposure to biocides does result in selection of MDR bacteria. The increased accumulation of biocides in the environment at low levels has the potential to provide environments which will favour the selection of mutants with increased tolerance to biocides and antibiotics [4].
Low-level biocide-antibiotic cross resistance can result from the increased expression of efflux pumps, particularly in Enterobacteriaceae the tri-partite AcrAB-TolC system [5,6]. AcrAB-TolC is the major multidrug efflux pump in many Gram-negative bacteria, including Salmonella, and has been shown to be over-expressed in Salmonella in response to exposure to low doses of biocide [3,7]. Non-specific, low-level resistance to antibiotics and biocides can also be mediated by decreasing permeability of the membrane often due to repression of major porins [8]. As well as its role in antimicrobial resistance, the AcrAB-TolC efflux pump also has a role in Salmonella pathogenicity, all three of the components being required for virulence [9,10,11]. Salmonella, as with all bacteria, has a large complement of transporters with nine known multidrug efflux systems [12]. Whilst AcrAB is the major system, expression of other RND pumps (AcrEF and AcrD) is co-ordinately regulated with AcrAB in Salmonella and AcrEF can functionally complement AcrAB [13]. AcrAB-TolC production is tightly regulated, three homologous global activators, MarA, SoxS and Rob, and one local repressor, AcrR, have been shown to regulate acrAB in Escherichia coli [14,15,16,17,18]. An additional activator, RamA, also homologous to MarA, SoxS and Rob, is involved in acrAB regulation in Salmonella but is absent in E. coli [19]. Overexpression of RamA leads to an increase in expression of both acrAB and acrEF [20,21].
Flow cytometry (FCM) and fluorescence-activated cell sorting (FACS) are techniques that can investigate and separate individual bacteria from large populations depending on their light scattering and fluorescent properties. Fluorescent dyes can be used to monitor membrane integrity and potential and thus monitor bacterial physiology; a major advantage of the technique is that it does not rely upon bacterial growth on agar plates or in broth, and so is suitable for studying bacteria having a viable but nonculturable (VNC) phenotype. In this study we have utilised FCM to analyse the physiological response of the model food borne pathogen Salmonella to both low-level and ''in-use'' concentrations of biocides. Using FACS we isolated mutants able to survive challenge with ''in-use'' concentrations of two biocides after one exposure. These mutants were multidrug resistant and overexpressed the AcrEF efflux pump and MarA, demonstrating that biocide exposure can select for mutants with a broad, low-level antibiotic resistance.

Isolation of viable cells from low and in-use concentrations of biocide
Exposure to biocide concentrations of 0.4% (V), 0.005% (SK), 0.005% (AQ) and 0.002% (TR) were used to provoke a ''low'' dose challenge of each biocide, where bacterial growth was inhibited but bacteria were not necessarily killed. Each biocide was also tested at the recommended working concentration of 1%, after 5 hours incubation in biocide each culture was analysed using FCM and PIX staining ( Figure 1). After low dose challenge, the biocides generated different physiological effects as revealed by FCM. The majority of cells in Superkill or Trigene were in the lower left (''healthy'') quadrant, 74% and 60% respectively, the majority of the remaining cells being in the upper left quadrant (disrupted membrane potential but not ruptured membrane). The biocide AQAS had a greater antimicrobial effect and only 39% of cells remained in the ''healthy'' quadrant. Virkon had a much more pronounced effect with only 0.3% of cells in the ''healthy'' quadrant after a low dose exposure, the majority of cells in the upper right ''dead'' quadrant. The healthiest cells in each population were sorted by FACS (Figure 1a), each sorted population gave rise to viable colonies on LB agar which were retained for further investigation.
After an ''in-use'' dose challenge the majority of cells were in the upper right ''dead'' quadrant (V 98%, SK 88%, AQ 98%, TR 70%) (figure 1b). These cells were stained by both PI and BOX, therefore had ruptured membranes, no membrane potential and were presumed dead. Surprisingly there were cells in each biocide not stained by PI or BOX, maintaining an intact membrane/ membrane potential, present at a low frequency. These populations (circled) were sorted, cells sorted from Virkon and AQAS gave no culturable colonies but cells sorted after high-level Superkill and Trigene treatment were recovered and representatives retained (referred to as SK 1% and TR 1%).
Colonies isolated from in-use concentrations of biocide have reduced susceptibility to multiple antibiotics but not to biocides The colonies isolated from the low dose of biocide had an antibiotic susceptibility profile identical to SL1344. The two mutants isolated from in-use concentrations of Superkill and Trigene had a broad, low-level reduced susceptibility to multiple antibiotics, nalidixic acid (4-fold), chloramphenicol (3-fold), tetracycline (2-fold), and ciprofloxacin (2-fold) ( Table 1). They also had a 2-fold reduced susceptibility to the biocide triclosan.
Populations of healthy cells recovered after exposure to a low dose of biocide were challenged again with the same level of biocide and re-analysed by FCM. The percentage of cells present in the lower-left ''healthy'' quadrant was not significantly different to that of SL1344 when first challenged. This suggests that the potential to survive exposure to a low dose of biocide is randomly distributed amongst a population, reflecting heterogeneity in terms of growth phase and/or gene expression amongst the population; cells able to survive a single exposure to a low biocide concentration were not pre-disposed to biocide survival upon repeat challenge. The mutants recovered from the in-use biocide challenge were however stably multidrug resistant although they showed no increase in MIC of the biocide from which they were recovered.

Biocides have varying effects on cell growth rate
Cells sorted from low dose challenge with Virkon and Trigene grew at a similar rate to wild-type cells in broth when no biocide was added ( Figure 2a). Cells sorted from low dose challenge with Superkill had a slightly longer lag phase than wild-type cells and cells sorted from AQAS did not achieve as high an optical density after the same length of time. The mutants sorted from in-use concentrations of Superkill and Trigene had a longer lag phase than SL1344 and only achieved a final OD 600 of 0.9, significantly lower than the OD 600 of 1.5 achieved by SL1344 after 16 hours growth ( Figure 2b). This shows a fitness cost has been incurred by both mutants.

Cells isolated from in-use concentrations of biocide have increased efflux activity
The efflux activity of the sorted cells was assessed by measuring intracellular accumulation of Hoescht 33342, a known substrate of MDR efflux pumps. The cells exposed to a low-dose of biocide accumulated levels of Hoescht dye similar to SL1344 (data not shown). The two mutants recovered from the in-use biocide challenge both accumulated significantly less dye than SL1344 ( Figure 3) and in the presence of an efflux pump inhibitor PAßN, accumulation of Hoescht was elevated, confirming that these mutants have increased efflux activity compared with SL1344 ( Figure 3).

AcrEF and MarA are up-regulated in high level biocide mutants
Comparative RT-PCR was used to determine the level of expression of genes relevant to multidrug resistance in the highlevel biocide mutants. In both mutants the expression of ompF, soxS and acrB was statistically significantly down-regulated in comparison with SL1344 ( Figure 4). The ramA gene was significantly down-regulated in SK 1% but not TR 1%. The down-regulation of acrB in both mutants was a surprise as this is usually considered the major efflux system in Salmonella. In both SK 1% and TR 1%, marA (2 fold) and acrF (10 fold and 7 fold respectively) expression was however significantly increased compared to SL1344 ( Figure 4) indicating that the efflux pump AcrEF-TolC is mediating the MDR phenotype in these mutants. Sequencing showed upregulation of MarA was not due to a mutation in the marA promoter or within marR.
To confirm that AcrEF over-expression was responsible for the MDR phenotype seen in these mutants, we inactivated acrEF in these strains, this resulted in a broad loss of MDR (Table 1) thus   confirming that the AcrEF pump is integral to the MDR phenotype seen. As acrEF has been reported to be regulated by FNR in E. coli [26] we investigated the level of fnr expression in our mutants but the level of expression was not significantly different between the mutants and wild-type.

Effect of acrB, tolC and ramR mutations on SL1344 survival in low dose biocide
To further explore the role of efflux in biocide tolerance a series of isogenic mutants (table 2) lacking components of AcrAB-TolC or over-expressing this system were examined for their ability to survive biocide exposure. The AcrAB-TolC system was studied as this is constitutively expressed whereas AcrEF is cryptic in wildtype cells. After 5 hours incubation in low dose biocide, the physiology of the wild-type strain SL1344 and isogenic mutants lacking a functional acrB, tolC and ramR were assessed by FCM ( Figure 5). The striking result was that the ramR mutant (which over-expresses AcrAB-TolC) had a significantly greater proportion of cells in the lower-left ''healthy'' quadrant than SL1344 for all 4 of the biocides tested. The biggest effects were seen with Superkill and AQAS where percentages in the lower-left ''healthy'' quadrant for SL1344 were 5% and 36% respectively whereas for the ramR mutant 98% of cells in each biocide were in this quadrant. In each case, the wild-type had a larger proportion of BOX+ bacteria, lacking membrane potential, than the ramR mutant. A smaller but still consistent effect was seen for the Trigene and Virkon biocides where the ramR mutant had higher ''healthy'' percentages, 99% compared to 95% in Trigene and 1.8% to 0.2% in Virkon, where the majority of the remaining bacteria were in the upper-right ''dead'' quadrant. Virkon in particular seemed not to be affected by efflux pump loss or overexpression; this was also suggested by the addition of the efflux pump inhibitor (EPI) PAbN, which decreased the percentage of cells of SL1344 in the lower-left quadrant for all biocides bar Virkon ( Figure 5).

Discussion
Successful prevention and eradication of outbreaks of bacterial infection relies on biocide use, as a result biocides must be able to rapidly and completely kill large populations of bacteria. Previously we have shown that repeated exposure of Salmonella to low levels of biocide can select for multidrug resistant mutants, in our previous work we were unable using conventional culture based techniques to isolate mutants with increased biocide tolerance. In this study we show that FCM is a very useful technique for identifying populations of bacteria that can withstand biocide challenge; FACS has the added advantage of allowing the physical isolation of those biocide-resistant bacteria for further investigation. The fact that FACS succeeded in isolating resistant mutants where plate-based screening failed suggests that the FACS method aids bacterial survival in this context; possibly as a result of the decoupling of the selection and growth processes.
Using this approach we were able to show that a single, lowlevel exposure to 2-4X the MIC of biocide, whilst preventing growth in broth did not rupture the membranes of a large proportion of a population (with the exception of Virkon) and that sorted 'healthy' cells were not MDR. This supports the previous findings that more than one exposure to a low level of biocide is required to select antibiotic resistant mutants. This might reflect the fact that most cells can survive this low-level stress (even if not able to grow) and that the natural variation within a population includes cells relatively unaffected by low-level biocide exposure. Therefore there is little pressure for selection of mutants with increased resistance. The flow cytometer was however also able to identify and recover mutants from populations that had been exposed to an extreme 1% biocide challenge. Subsequent, replicate experiments with traditional culture methods did not recover live mutants after incubation with 1% biocide. This indicates the potential for FACS to identify and select mutants that occur at a very low frequency. Typically biocides are expected to provoke a 5-log reduction in viable bacterial numbers; in our experiments the proportion of cells in a population that were able to maintain a membrane potential after 5 hours exposure to the working concentration of biocide was 2% in Virkon and AQAS, 12% in Superkill and 30% in Trigene. After sorting, the vast majority of these populations did  not grow and viable colonies were only recovered from Superkill and Trigene. However, we made no efforts to improve this recovery rate with the addition of a richer recovery medium than LB. The isolation of some immediately viable cells from a larger population of cells that maintain a membrane potential might suggest this population is highly heterogeneous and comprises efflux mutants as characterised here and possibly a larger viable but non-culturable population. Further work to investigate this is warranted.
The two resistant mutants characterised had a classic, low-level MDR phenotype with a slightly increased resistance to a broad range of drugs from different classes of antibiotics. The AcrAB-TolC efflux pump has been shown to be the most important efflux pump in expelling antibiotics from Salmonella. It was therefore expected that expression of this pump would be up-regulated in our recovered cells. However, the down regulation of acrB and the up-regulation of acrF indicate that the AcrEF-TolC efflux pump is the main driver in the MDR phenotype of these two mutants. This was confirmed by inactivation of AcrEF-TolC in these strains. Previous work has indicated a relationship between AcrAB-TolC and AcrEF-TolC; over-expression of AcrEF-TolC can complement AcrB deletion [13]. The fact that both the mutants recovered from two different biocides had this phenotype suggests a common mechanism of cell survival to biocides with de-repression of AcrEF (mediated by up-regulation of MarA) favoured over de-repression of AcrAB. The MarA regulon has not been shown to include acrEF in Salmonella; however in E. coli a putative MarA binding site in the acrEF promoter has been described [22]. Study of the S. Typhimurium acrEF promoter region confirms that this putative MarA binding site is also present and is a better match, by 1 base, to the putative binding site consensus sequence. This suggests that AcrEF de-repression in the mutants is a result of MarA overexpression. Sequencing of the mutant's marA promoter region showed no mutations. The genetic basis for marA de-repression in the mutants studied here is therefore currently unknown.
Of the isogenic mutants tested for ability to survive a low dose of biocide, only a ramR mutant had a discernable effect on the survival of the cells. This mutant is MDR and over-expresses AcrAB-TolC [23]. This suggests therefore that AcrAB-TolC can recognise some biocides as substrates (excluding Virkon) and that de-repression of AcrAB-TolC will be protective against biocide stress. Surprisingly, loss of acrB or tolC did not have a very  detrimental effect on the cells. This might indicate that other efflux systems can complement the loss of the AcrAB-TolC pump. Chemical inactivation of efflux with PAbN also decreased the survival of the cells in Superkill, AQAS and Trigene. This again suggests that efflux is an intrinsic mechanism by which biocides are expelled from the cell; once again cells grown in Virkon were less affected by the addition of PAbN; more evidence efflux is of little importance in the elimination of this biocide. This study has shown that Salmonella can survive challenge with in-use concentrations of some biocides, that this is due to derepression of the AcrEF MDR efflux system and that these mutants are MDR. Flow cytometry is a sensitive tool for evaluating the cidal potential of biocides which offers significant advantages over culture based methods. Our work suggests the potential for different biocides to select for MDR varies; the development of novel biocide formulations should bear this in mind.

Isolation of biocide tolerant mutants
Salmonella enterica serovar Typhimurium strain SL1344 [24] was the parent strain used in all experiments, for biocide exposure overnight cultures (5 ml) of SL1344 were grown in LB broth at 37uC and used to inoculate fresh 5 ml LB broths containing the desired concentration of different biocides. After 5 hours samples were removed and analysed by FCM. An ethanol killed sample of bacteria was used as a control. For each experiment 10,000 cells were analysed and fluorescence from two dyes, PI (stains the DNA of bacteria that have lost membrane integrity) and BOX (stains bacteria with a collapsed membrane potential) used to identify cells with maintaining a membrane potential which were sorted by FACS into FACSFlow buffer and aliquoted onto LB agar plates. Colonies that grew were stored on Protect Beads (Technical Service Consultants Ltd). Full details of the methodology are given in Material S1.

Phenotypic characterisation of biocide tolerant mutants
Cells recovered after biocide exposure were characterised in terms of their antimicrobial susceptibility using broth microdilution and agar dilution methods following the recommendations of the British Society for Antimicrobial Chemotherapy [25]. The growth of sorted biocide tolerant mutants compared to the parental strain SL1344 was also analysed measuring the optical density at 600 nm of replicate cultures. Full details of the methodology are given in Material S1.

The role of efflux
Efflux activity was determined phenotypically by measuring the accumulation of Hoescht 33342, a substrate of MDR efflux systems according to the method of Webber & Coldham [26]. Accumulation of Hoescht 33342 was measured in the presence and absence of PAbN, an efflux pump inhibitor, added to a final concentration of 100 mg.ml-1. Expression of efflux pump genes acrB, acrEF, known regulators marA. soxS, ramA and fnr, porin ompF and 16S rRNA (rrsH) was also determined using comparative RT-PCR as previously described [13]. Primers and full details are listed in Table S1. To confirm its role in multidrug resistance, the acrEF locus was disrupted in the SK 1% and TR 1% mutants using bacteriophage P22. Full details of the methodology are given in Material S1.

Supporting Information
Material S1 Full description of methodology. (DOC)