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, low-level 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 
. 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
however in E. coli
a putative MarA binding site in the acrEF
promoter has been described 
. 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 over-expression. 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 
. 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
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 PAβN 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 PAβN; 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 de-repression 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.