Exposure of Escherichia coli and Salmonella enterica serovar Typhimurium to triclosan induces a species-specific response, including drug detoxification

Abstract

Objectives

The use of triclosan within various environments has been linked to the development of multiple drug resistance (MDR) through the increased expression of efflux pumps such as AcrAB–TolC. In this work, we investigate the effect of triclosan exposure in order to ascertain the response of two species to the presence of this widely used biocide.

Methods

The transcriptomes of Salmonella enterica serovar Typhimurium SL1344 and Escherichia coli K-12 MG1655 after exposure to the MIC of triclosan (0.12 mg/L) were determined in microarray experiments. Phenotypic validation of the transcriptomic data included RT–PCR, ability to form a biofilm and motility assays.

Results

Despite important differences in the triclosan-dependent transcriptomes of the two species, increased expression of efflux pump component genes was seen in both. Increased expression of soxS was observed in Salmonella Typhimurium, however, within E. coli, decreased expression was seen. Expression of fabBAGI in Salmonella Typhimurium was decreased, whereas in E. coli expression of fabABFH was increased. Increased expression of ompR and genes within this regulon (e.g. ompC, csgD and ssrA) was seen in the transcriptome of Salmonella Typhimurium. An unexpected response of E. coli was the differential expression of genes within operons involved in iron homeostasis; these included fhu, fep and ent.

Conclusions

These data indicate that whilst a core response to triclosan exposure exists, the differential transcriptome of each species was different. This suggests that E. coli K-12 should not be considered the paradigm for the Enterobacteriaceae when exploring the effects of antimicrobial agents.

biocide, transcriptome, efflux

Introduction

Triclosan (2,4,4′-trichloro-2′hydroxydiphenylether) is a member of the bisphenol biocide family, which exhibit a broad spectrum of activity against many Gram-negative and Gram-positive bacteria.¹ At concentrations <1000 mg/L, triclosan is bacteriostatic, whereas at higher concentrations, such as those commonly found in soaps and handwashes, it is bactericidal.² Due to these antimicrobial properties, and the ease of its incorporation into plastics, triclosan is now found in a wide range of personal care products, household cleaning products, domestic appliances such as refrigerators, surgical apparatus and increasingly within food processing plants in order to reduce microbial load.³

Triclosan interferes with the biosynthesis of fatty acids through non-competitive inhibition of the NADH-dependent enoyl-(acyl carrier protein) reductase (FabI) in both Escherichia coli and Salmonella enterica serovar Typhimurium,^4,5 or the homologue of FabI, InhA, in Mycobacterium tuberculosis.⁶ Point mutations within fabI have been shown to confer resistance to triclosan within E. coli⁶ and Salmonella Typhimurium.⁷ Other resistance mechanisms also play a role, including the overproduction of the target protein (through overexpression of the fabI gene)^7,8 and growth within a biofilm.⁹ The bactericidal activity of triclosan cannot be explained by inhibition of fabI alone¹⁰ and the identification of multiple mechanisms of resistance to triclosan suggests that other targets for triclosan exist.^7,11

In E. coli and Salmonella Typhimurium, deletion of the AcrAB–TolC efflux system leads to a 4- to 10-fold increase in susceptibility to triclosan, whereas overexpression of this efflux pump confers a 2- to 4-fold decrease in susceptibility.¹² The AcrAB–TolC pump is also required for the development of multiple drug resistance (MDR), including triclosan resistance.¹² Known regulators of the AcrAB–TolC efflux pump include the AraC/XylS-type regulatory genes marA,^13–15soxRS^16,17 and rob.^18,19 A further AraC family regulator found in Salmonella and some other Enterobacteriaceae, but not E. coli or Shigella spp., ramA, can also regulate acrAB and tolC.^20–23 Therefore, the increasing use of triclosan in the household, hospital and food processing setting has raised concerns that the development of triclosan resistance may involve de-repression of efflux systems and a selection for cross-resistance to multiple antibiotic classes.^24–26

The major role of AcrAB–TolC in the development of MDR suggests that one of the cellular responses to the presence of triclosan could be induction of this efflux pump. Furthermore, both E. coli^27–29 and Salmonella Typhimurium^30–32 undergo a global SOS stress response when exposed to certain bactericidal antimicrobials and agents that cause the formation of hydroxyl radicals.³³ To explore the hypothesis that triclosan induces a stress response, transcriptomic analyses of E. coli MG1655 and Salmonella Typhimurium SL1344 in the presence and absence of triclosan were performed. Global changes in each individual transcriptome and the Salmonella proteome were detected and a common triclosan-dependent signature revealed. However, there were also differences in the response to triclosan between Salmonella Typhimurium and E. coli, demonstrating a species-specific response. These data indicate that not all species of the Enterobacteriaceae behave the same as E. coli K-12 in the presence of triclosan, strongly suggesting that this organism should not be the paradigm for modelling bacterial responses to antimicrobial agents.

Materials and methods

Bacterial strains and culture medium

E. coli K-12 MG1655³⁴ and S. enterica serovar Typhimurium SL1344,³⁵ LT2 (ATCC 19585) and 14028S (ATCC 14028) were used in this study. The strains used were stored at −80°C on Protect™ beads until required. Cultures were grown in MOPS-based minimal medium (Teknova Inc., CA, USA) supplemented with 700 mg/L l-histidine for SL1344.³⁶

Growth kinetics

To quantify any growth differences upon exposure to triclosan, growth kinetics in the presence and absence of triclosan were determined by two methods. For spectrophotometric determination, a 4% inoculum of an overnight culture was added to at least six wells of a 96-well microtitre plate containing 200 µL of MOPS minimal medium. To half of the wells, triclosan was added to a final concentration of 0.12 mg/L (the previously defined MIC) and the plate was then incubated for 18 h with shaking at 37°C. Optical density readings at 600 nm were recorded automatically every 10 min using a FLUOstar Optima (BMG Labtech, UK) spectrophotometer. For viable count determination, a 4% inoculum of an overnight culture was added to 50 mL of MOPS minimal medium in 250 mL Erlenmeyer flasks and incubated at 37°C, with shaking at 220 rpm. Upon reaching an OD₆₀₀ of 0.7, the culture was split into two 25 mL aliquots. Triclosan was added to one aliquot to 0.12 mg/L. The other aliquot was reserved as a triclosan-free control. Subsequently, at half hourly intervals for 5 h, aliquots were taken from both cultures, serially diluted (ranging from 10⁻⁶ to 10⁻⁹ cfu/mL) in PBS and 100 µL inoculated onto Luria–Bertani (LB) agar plates. After overnight incubation at 37°C, colonies were counted. All data were analysed in Microsoft Excel with standard deviations, and paired, two-tailed Student's t-tests (P < 0.05) applied to define significance.

Antimicrobial susceptibility

The MIC of triclosan for both strains was determined using the BSAC standardized broth microdilution method.³⁷ The lowest concentration of antimicrobial that caused no visible growth was determined to be the MIC of that compound.

Triclosan exposure and RNA extraction

Overnight cultures of either E. coli MG1655 or Salmonella Typhimurium were grown in MOPS minimal medium at 37°C. A 4% inoculum was added to six 25 mL aliquots of defined MOPS minimal medium. Strains were grown to mid-logarithmic phase, defined as OD₆₀₀ = 0.7 ± 0.02, on a Biomate 3 spectrophotometer (Thermo Spectromic, UK). In order to induce a triclosan exposure response, rather than a triclosan-induced cell death effect, an exposure time of 30 min was selected. This time was based upon analyses of the published literature^29,38 of antibiotic exposure studies, which demonstrated that a 30–60 min exposure time was optimal for induction of a wide range of antibiotic-responsive genes. Therefore, upon reaching mid-logarithmic phase, three aliquots of each strain were exposed to triclosan for both species (0.12 mg/L) for 30 min at 37°C with shaking at 220 rpm. The three remaining unexposed aliquots were also incubated for a further 30 min at 37°C. Following E. coli exposure to triclosan, RNA was extracted as described previously.³⁹ Following Salmonella Typhimurium exposure to triclosan, RNA was also extracted as described previously.⁴⁰ Extracted RNA was quantified for yield and quality on an Agilent 2100 Bioanalyser Automated Analysis System (Agilent Technologies, USA).

Microarray construction, transcriptomic experiments and data validation

The Operon Array Ready E. coli 1.0 microarray oligonucleotide set (Qiagen-Operon) was printed and treated as described previously.^41,42 Microarray slide pre-treatment, probe labelling and microarray hybridizations were performed as previously described,³⁹ except that microarray slides were scanned using an Axon Genepix 4000B scanner and Genepix 5.1 software (Axon Scientific, UK). The Salmonella microarray and protocols for probe synthesis, hybridization and washing were as described previously.⁴³ All data generated by microarray experiments were analysed with Bioconductor,⁴⁴ using P < 0.05 as the only significance cut-off. Validation of the transcriptomic data was performed for Salmonella Typhimurium SL1344 and E. coli MG1655 by random selection of six differentially expressed genes from both species. RT–PCR experimental design was as described previously⁴⁵ with primers listed in Table 1, and preliminary experiments to ascertain the level of cDNA dilution required to prevent PCR saturation were performed as described previously.⁴⁵

Table 1

Primers used in this study

Primer name	Species	Sequence (5′ to 3′)
csgF F	Salmonella Typhimurium SL1344	GACGTTCCAGTTCGCTAATC
csgF R	Salmonella Typhimurium SL1344	ATCGTTGGTCACCATACGTC
fabB F	Salmonella Typhimurium SL1344	TTCTCTCAGGAGCTGAAGGA
fabB R	Salmonella Typhimurium SL1344	TCCAGCTCTTGCACAACAAC
cysC F	Salmonella Typhimurium SL1344	TGGAAGAGGCGTTACATCAG
cysC R	Salmonella Typhimurium SL1344	TAAGATGAACCTGTGGCGAG
guaC F	Salmonella Typhimurium SL1344	CGTTCTTATCCGTCCTAAGC
guaC R	Salmonella Typhimurium SL1344	GAACTCATGCCGTAGAACAG
moaA F	Salmonella Typhimurium SL1344	TTCGTTGCACCTACTGTCTG
moaA R	Salmonella Typhimurium SL1344	GGTAATCGCATGGAACTGAC
narY F	Salmonella Typhimurium SL1344	TGTATTGGCTGTCACACCTG
narY R	Salmonella Typhimurium SL1344	CATCGCACTGACGTTCATAC
phoH F	Salmonella Typhimurium SL1344	CGTGTGTTGAGACGAGATTC
phoH R	Salmonella Typhimurium SL1344	GATCCTCTCGACGTCCTTAT
trpE F	Salmonella Typhimurium SL1344	TGACACTGTCACCATTCAGG
trpE R	Salmonella Typhimurium SL1344	ACAGAGTCCAGCACAATTCC
ompA F	E. coli MG1655	GATAACACCTGGTACACTGG
ompA R	E. coli MG1655	ACGAGTGTAGATGTCCAGGT
csgF F	E. coli MG1655	TTACTCAGGCCATCCAGTCA
csgF R	E. coli MG1655	GATCTGTCACGTTCAACTGC
fliC F	E. coli MG1655	CGGTACTAACTCTGAGTCTG
fliC R	E. coli MG1655	CACCGGAAGAGTCAGTATAG
narG F	E. coli MG1655	AATAGAGACAGGTGCTCGAC
narG R	E. coli MG1655	CTGAACCACTGGTATCACCT
nrdE F	E. coli MG1655	ACAGTGTGACCTTCAGTAGC
nrdE R	E. coli MG1655	ACTCGCTCTACGTCATAAGG
dam F	E. coli MG1655	ACTATCAGTTCCGCGAAGAG
dam R	E. coli MG1655	CATAAGGCGGATCGCAATAG
fabB F	E. coli MG1655	CAGCATCGGTAATAACCAGC
fabB R	E. coli MG1655	CAGCTCTTCAACCACTACCA
glcF F	E. coli MG1655	GCCTCACTTGCCGTAATTGT
glcF R	E. coli MG1655	TTCACTGACCTGACGTGCTT

Salmonella Typhimurium proteomic analyses

In order to further validate gene expression changes observed through transcriptomic experiments, the effect upon the proteome of triclosan exposure was investigated. Protein extracts were obtained from cultures of Salmonella Typhimurium SL1344 grown exactly as for RNA extraction. The extracts were prepared and analysed by two-dimensional liquid chromatography/multiple-stage mass spectrometry (2D-LC-MSn) as described previously.⁴⁶ The relative abundance of proteins was determined using the spectrum count method following published guidelines and denotes the number of peptide counts (‘hits’) detected for each protein.⁴⁶ Expression analyses were limited to those proteins common to all three replicates from control or test cultures. The statistical significance of percentage changes in protein expression was determined using a paired, two-tailed Student's t-test (P < 0.05).

Biofilm formation assays

Biofilms were established in polystyrene microtitre trays by inoculating 200 µL of LB broth with ∼10³ cfu of each strain as appropriate (dilutions were made from overnight cultures grown in LB broth) and triclosan serially diluted as appropriate, starting from 0.12 mg/L. After incubation at 30°C for 24–72 h, all liquid was removed from all wells and adherent cells were stained with 1% Crystal Violet for 15 min. After staining, unbound dye was removed and 200 µL of 70% ethanol was added to solubilize bound dye. The absorbance of each well was determined at 570 nm using a FLUOstar Optima (BMG Labtech). All biofilm assays were repeated on at least three separate occasions each with eight technical replicates. A paired, two-tailed Student's t-test was used to assess significance.

Motility assays

The effect of triclosan exposure upon bacterial motility was quantified as described previously.¹¹ with triclosan (0.12 mg/L) added prior to inoculation with bacteria.

Pyruvate utilization assays

To ascertain whether the gene expression changes observed by microarray analysis correlated with changes in pyruvate utilization, a pyruvate utilization assay was performed as described previously.¹¹ Triclosan was added (0.12 mg/L) prior to inoculation with either Salmonella Typhimurium SL1344 or E. coli MG1655. Pyruvate utilization in the presence of triclosan was then compared with a triclosan-free control after 30 min and 16 h; the former exposure in order to correlate best with microarray experiments. Standard deviations were calculated and paired, two-tailed Student's t-tests used to assess significance (P < 0.05).

Microarray data

In accordance with the practices of the Sanger Institute and ExGen, Salmonella Typhimurium SL1344 data have been submitted to ArrayExpress (accession E-MEXP-2078) and those for E. coli MG1655 submitted to GEO (accession GSE15059).

Results

Exposure to triclosan confers a ‘core’ gene expression signature in E. coli K-12 MG1655 and Salmonella Typhimurium SL1344

The MIC (0.12 mg/L) of triclosan conferred no inhibition upon growth with no significant differences in the growth rates or final population density of either species seen during the time period of the microarray experiment (data not shown). After exposure to the MIC of triclosan for 30 min, the transcriptomes of Salmonella Typhimurium SL1344 and E. coli MG1655 contained many differentially expressed genes (Figure 1). Despite there being differences between the response of Salmonella Typhimurium and E. coli to triclosan exposure, a common transcriptomic signature was observed for the two species (Figure 2). This comprised: (i) genes, and regulators thereof, that encode proteins which efflux solutes from the bacterial cell (e.g. the increased expression of emrR, emrA, acrA, E. coli acrB, Salmonella Typhimurium acrE and tolC); (ii) differential expression of genes that encode proteins that allow the entry of solutes into the cell (e.g. ompC, ompF); (iii) increased expression of detoxification genes known to be within the soxS regulon, such as zwf and nfnB; (iv) decreased expression of members of ribosomal biosynthetic operons (rpsGL and rplCD); and (v) decreased expression of fabG.

Figure 1

Venn diagram of all significant (P < 0.05) gene expression changes in Salmonella Typhimurium SL1344 and E. coli MG1655 after triclosan exposure. Boxed numbers indicate the total number of significant genes differentially expressed for each species.

Figure 2

Gene expression changes of all 47 ‘core’ triclosan-dependent genes with altered expression in response to triclosan in both species after triclosan treatment. Dark grey bars, Salmonella Typhimurium SL1344; light grey bars, E. coli MG1655. Negative values correspond to a gene/protein expression decrease.

Exposure to triclosan confers altered, but divergent, expression of the same genes in E. coli and Salmonella Typhimurium

For some genes, differential expression was observed in both E. coli and Salmonella Typhimurium, but with an increase for one species and a decrease for the other. For example, expression of the genes ompA and ompC encoding outer membrane proteins was increased and that of ompF decreased in Salmonella Typhimurium, whereas the opposite was found for E. coli (Tables 2 and 3). The expression of the regulatory gene ompR was also increased in Salmonella Typhimurium. The expression of many members of the flg, flh and fli operons, which encode flagellar biosynthesis, was decreased in Salmonella Typhimurium, but increased for E. coli. For Salmonella Typhimurium, exposure to triclosan conferred an 8-fold decrease in SL1344 motility (Figure 3). For E. coli, the microarray expression changes did not correlate with any significant change in motility. Further experimentation revealed that for fliC, the changes in expression were not confirmed by RT–PCR (Figure 4b).

Figure 3

Motility of Salmonella Typhimurium SL1344 and E. coli MG1655 after exposure to triclosan. *P < 0.05 significance when compared with the unexposed control.

Figure 4

Fold change in expression of (a) six selected Salmonella SL1344 genes and (b) six selected E. coli MG1655 genes to confirm the accuracy of the transcriptomic data. *P < 0.05 significance when compared with the unexposed control.

Table 2

Selected groups of Salmonella Typhimurium SL1344 genes with significant (P < 0.05) differential expression in response to triclosan exposure

Gene name	Product	Fold changea
Drug detoxification
acrE	acriflavin resistance protein E	1.63
acrR	acrAB operon repressor	1.87
ompC	outer membrane porin C	2.31
ompF	outer membrane porin F	0.83
ompH	outer membrane protein OmpH precursor	1.27
ompR	two-component response regulator	1.43
sodA	manganese superoxide dismutase	1.58
sodC	copper-zinc superoxide dismutase	1.46
tolC	outer membrane protein	1.50
fumC	fumarate hydratase class II	2.24
acnA	aconitate hydratase 1 (citrate hydrolyase 1)	1.99
fpr	ferredoxin-NADP reductase	1.45
Curli biosynthesis
csgA	major curlin subunit precursor	1.63
csgDEFG	curli biosynthesis and regulation thereof	3.11–6.06
SPI-1
sicP	chaperone (associated with virulence)	1.27
sipF	probable acyl carrier protein	1.40
sitA–D	iron transport proteins	2.33–3.42
sprA	possible AraC-family transcriptional regulator	1.52
STM2904	hypothetical protein	1.77
STM2905	hypothetical protein	1.79
stpA	tyrosine phosphatase (associated with virulence)	1.40
SPI-2
spiC	putative pathogenicity island 2 secreted effector protein	1.74
ssaD	putative pathogenicity island protein	2.62
ssaE	putative secretion system protein	2.08
ssaI	putative pathogenicity island protein	1.30
ssaJ	putative pathogenicity island lipoprotein	1.32
ssaL	putative secretion system protein	1.63
ssaN–Q	putative type III secretion proteins	1.34–1.72
ssaV	putative type III secretion protein	1.90
sscA	putative type III secretion system chaperone protein	1.85
sscB	putative pathogenicity island protein	1.77
sseA–E	putative pathogenicity island effector proteins	1.81–2.32
ssrA	putative two-component sensor kinase	1.61
STM1389	putative pathogenicity island protein	3.79
STM1410	putative pathogenicity island protein	1.45
Fatty acid biosynthesis
fabI	enoyl-(acyl carrier protein) reductase (NADH)	0.82
fabA	d-3-hydroxydecanoyl-(acyl carrier protein)	0.47
fabB	3-oxoacyl-(acyl carrier protein) synthase I	0.18
fabG	3-oxoacyl-(acyl carrier protein) reductase	0.73
Global regulators
fis	DNA-binding protein	0.40
fnr	fumarate and nitrate reduction regulatory protein	0.84
phoP	transcriptional regulatory protein PhoP, regulator of virulence determinants	0.83
phoQ	sensor protein PhoQ, regulator of virulence determinants	0.72
rpoS	stress response sigma factor	2.30
soxS	superoxide stress response regulator	1.70
Motility
flgA	flagellar biosynthesis	0.39
flgI–L	flagellar P-ring and hook-associated proteins	0.43–0.71
flgN	flagellar synthesis protein	0.76
flhB	flagellar biosynthesis/transcriptional activator	0.76
Ribosomal biosynthesis
rplA–E	50S ribosomal subunit protein L1, 2, 3, 4 and 5	0.34–0.52
rplI	50S ribosomal subunit protein	0.42
rplK–W	50S ribosomal subunit protein L11–23	0.29–0.49
rpmA	50S ribosomal subunit protein L27	0.54
rpmB	50S ribosomal subunit protein L28	0.41
rpoA	DNA-directed RNA polymerase alpha chain	0.67
rpoC	DNA-directed RNA polymerase, beta-subunit	0.74
rpsA–C	30S ribosomal protein S1, 2 and 3	0.32–0.74
rpsE–M	30S ribosomal subunit protein S5–13	0.32–0.74
rpsO–R	30S ribosomal subunit protein S15–18	0.29–0.42

^aCompared with Salmonella Typhimurium SL1344 unexposed to triclosan.

Table 3

Selected groups of E. coli MG1655 genes with significant (P < 0.05) differential expression in response to triclosan exposure

Gene name	Product	Fold changea
Drug detoxification
acrB	acridine efflux pump	2.21
imp	organic solvent tolerance	1.96
fldA	flavodoxin 1	3.11
emrD	2-module integral membrane pump	0.60
marR	multiple antibiotic resistance regulator	0.54
sodC	copper-zinc superoxide dismutase	0.29
ompC	outer membrane porin C	0.05
ompF	outer membrane porin F	4.98
Curli biosynthesis
csgA	major curlin subunit precursor	0.56
csgDEFG	curli biosynthesis and regulation thereof	0.28–0.41
Global regulator
fis	site-specific DNA inversion stimulation factor	5.91
rpoS	stress response sigma factor	0.53
soxS	superoxide stress response regulator	0.18
Motility
flgM	anti-FliA (anti-sigma) factor	5.19
fliC	flagellar biosynthesis; flagellin, filament structural protein	215.55
fliD	flagellar biosynthesis	13.48
fliG–I	flagellar biosynthesis	3.02–8.32
fliL–N	flagellar biosynthesis	2.26–6.41
fliS	flagellar biosynthesis	6.40
fliT	flagellar biosynthesis	9.98
fliZ	orf, hypothetical protein	16.55
NADH dehydrogenase
nuoA–N	NADH dehydrogenase I chain A–N	2.49–6.90
lrhA	NADH dehydrogenase transcriptional regulator, LysR family	2.02
Fatty acid biosynthesis
fabA	beta-hydroxydecanoyl thioester dehydrase, trans-2-decenoyl-ACP isomerase	3.58
fabB	3-oxoacyl-(acyl carrier protein) synthase I	7.16
fabF	3-oxoacyl-(acyl carrier protein) synthase II	2.64
fabG	3-oxoacyl-(acyl carrier protein) reductase	0.34
fabH	3-oxoacyl-(acyl carrier protein) synthase III	3.39
Iron transport
fepA–D	ferric enterobactin (enterochelin) transport	0.13–0.61
fepG	ferric enterobactin transport protein	0.16
fes	enterochelin esterase	0.17
fhuA–F	hydroxamate iron uptake	0.12–0.46
Ribosomal biosynthesis
rplC	50S ribosomal subunit protein L3	0.38
rplD	50S ribosomal subunit protein L4, regulates expression of S10 operon	0.33
rpmJ	50S ribosomal subunit protein L36	0.28
rpoS	RNA polymerase, sigma S (sigma38) factor	0.53
rpsG	30S ribosomal subunit protein S7, initiates assembly	0.23
rpsL	30S ribosomal subunit protein S12	0.30
rpsV	30S ribosomal subunit protein S22	0.02

^aCompared with E. coli MG1655 unexposed to triclosan.

The expression of the fatty acid biosynthetic genes was also divergent between the species, although fabG expression was decreased in both. The increased expression of fabB in E. coli and decreased expression in Salmonella Typhimurium was confirmed by RT–PCR (Figure 4a and b). However, these gene expression changes did not significantly affect utilization of pyruvate within either species (data not shown). The expression of the transcriptional activator soxS, but not soxR, was also changed in both species; increased expression in Salmonella Typhimurium and decreased expression in E. coli. The expression of sodC, a superoxide dismutase, and fis, a global regulator, was also divergent between the species (Tables 2 and 3). Increased expression of the sigma stress response gene rpoS was seen within Salmonella Typhimurium, whereas expression was decreased within E. coli.

Figure 5

Comparison of the expression of the Salmonella Typhimurium SL1344 genes found to have a corresponding protein significantly (P < 0.05) differentially expressed upon exposure to triclosan. Dark grey bars, microarray fold changes; light grey bars, proteomic fold changes. Negative values correspond to a gene/protein expression decrease.

Expression of the curli biosynthetic genes csgADEFG was increased in Salmonella Typhimurium, but decreased in E. coli (Tables 2 and 3, and Figure 4a and b). SL1344 forms biofilms poorly; nonetheless biofilm production was reduced (data not shown). In the biofilm model used, there was no observable change in biofilm production in E. coli (data not shown). Again for the representative gene tested, csgF, the microarray data were not confirmed (Figure 4a and b).

Responses to triclosan exposure specific to Salmonella Typhimurium SL1344

In addition to genes encoding components of efflux pumps, expression of the gene encoding the outer membrane protein TolC was increased in Salmonella Typhimurium (Table 2). Expression of additional genes within the soxS regulon (sodA, sodC, fumC, acnA and fpr) was also increased (Table 2). There was also increased expression of many genes within the Salmonella pathogenicity islands 1 and 2. This included genes from the ssa and sse operons (Table 2). RT–PCR experiments confirmed these data from the Salmonella microarray experiments; increased expression of phoH and narY and decreased expression of cysC and trpE (Figure 4a). Salmonella Typhimurium had decreased expression of considerably more ribosomal biosynthetic genes after triclosan exposure than E. coli, mainly from the rpl, rpm, rpo and rps operons (Table 2).

The effect of triclosan on the proteome of Salmonella Typhimurium was also determined. In total, 41 of 304 total detected proteins (13%) were found to be significantly (P < 0.05) differentially expressed after exposure to triclosan. Of those 41 proteins, 19 corresponding genes were also differentially expressed. Correlation between gene and protein expression was high, with expression of 15/19 (79%) matching the gene expression levels observed on the microarray. Of the four proteins whose expression did not match their gene expression, three (PflB, SeqA and TolC) had decreased expression compared with increased expression of the corresponding gene, whereas increased expression of one protein (ArgS) was seen compared with decreased expression of the corresponding gene (Figure 5).

Responses to triclosan exposure specific to E. coli MG1655

Few genes within detoxification pathways and the regulation thereof were found to be triclosan dependent in E. coli. However, expression of imp, which is involved in organic solvent tolerance, was increased (Table 3). Expression of marR, the local repressor of the transcriptional activator marA, was decreased by 0.5-fold, although no differential expression of marA was detected (Table 3). The expression of genes within the nuo (NADH dehydrogenase subunits) and the pur (purine biosynthesis) operons was increased. Expression of the efflux pump component emrD was decreased.

Differential expression of several operons that have been previously identified as responsive to the level of intracellular iron⁴⁷ was observed in response to triclosan exposure. These include the fep, fes and fhu operons. Upon further analysis, a strong correlation between operons regulated by increased free iron concentration and by triclosan exposure was observed (Table 4). This phenomenon was not seen with Salmonella Typhimurium.

Table 4

Comparison of operons regulated by free iron and also by exposure to triclosan in E. coli MG1655

Gene/operon	Expression increased/decreased by free iron?a	Expression increased/decreased by triclosan exposure?b
fhuACDBEF	decreased	decreased
entABCDEF	decreased	decreased
fepABCDG	decreased	decreased
fes	decreased	decreased
tonB	decreased	decreased
exbBD	decreased	NC
feoAB	decreased	increased
fecACDIR	decreased	decreased
cydAB	increased	increased
narKGHIJ	increased	increased
narWYU	increased	decreased
napABCDFGH	increased	increased
nuoABCDEFGHIJKLMN	increased	increased
nirBCD	increased	increased
nikACD	increased	increased
cspBI	increased	increased
sodB	increased	increased
nrdEFHI	decreased	decreased
cyoBC	increased	NC
purCDEHLMNT	decreasedc	increased
cheZ	increased	NC

NC, no significant gene expression change.

^aData from McHugh et al.⁴⁷

^bData from this study.

^cPreliminary data mentioned in McHugh et al.⁴⁷

Discussion

Characterization of triclosan-resistant mutants by our team has been published previously.¹¹ Based on those findings, we hypothesized that transient exposure of E. coli and Salmonella Typhimurium to an inhibitory concentration of triclosan may induce a protective response, rather than induce cell death. Within this protective response, we hypothesized that altered expression of efflux pump genes, and regulators thereof, would be seen, so allowing intracellular concentrations of triclosan to be reduced to subinhibitory levels. A core transcriptomic response to triclosan exposure was observed for both species. This included increases in expression of genes encoding components of efflux pumps, and decreases in expression of ribosomal subunit genes (e.g. rpl and rps) and fatty acid biosynthetic genes. Previous work with E. coli has shown a similar, yet more extensive, down-regulation of ribosomal biosynthesis upon oxygen deprivation and acid stress,⁴⁸ but no significant retardation of growth in either species was observed in the present study.

The most surprising observation in this study was the divergent gene expression changes between such closely related species. Both species exhibited decreased expression of fabG, a fatty acid biosynthetic gene; however, Salmonella Typhimurium showed decreased expression of other fatty acid biosynthetic genes, including fabI, whereas E. coli showed increased expression of fabABFH. Previous work showed that triclosan-resistant mutants of Salmonella Typhimurium shift metabolism from the use of products of the fab gene cluster and increase production of pyruvate; this was postulated to bypass the inhibitory effects of triclosan.⁷ However, in the current study, no change in pyruvate production was observed in either species, suggesting that triclosan exposure does not confer this metabolic pathway shift under the experimental conditions used here. Both species also showed increased expression of the glucose dehydrogenase encoded by zwf and the NAD(P)H nitroreductase encoded by nfnB, which are both documented members of the soxS regulon in both species.^31,49 However, expression of soxS was opposite between the species, suggesting that induction of zwf and nfnB in response to triclosan exposure may be via different regulatory pathways.

No differential expression of ompR was observed in E. coli, whereas expression was increased in Salmonella Typhimurium. This observation probably explains other gene expression differences in the divergent response. Yoshida et al.⁵⁰ reported that an increased level of phosphorylated ompR regulates a preferential shift to increased expression of ompC over ompF. For molecules such as bile salts that gain entry via OmpF, this results in exclusion from the bacterium whilst entry of essential solutes is maintained via OmpC.⁵¹ Prigent-Combaret et al.⁵² suggested that E. coli OmpR is also a regulator of the csgD regulator, which in turn activates curli formation⁵³ and thus biofilm formation.⁵⁴ The latter is considered to be important in triclosan resistance.⁹ Consequently, it is unsurprising that expression of genes involved in the biosynthesis of curli, a major component of the extracellular polysaccharide associated with biofilms, was also increased in response to triclosan exposure within Salmonella. Previous work with E. coli has indicated that increased expression of csgD decreases the formation of biofilm,⁵⁵ suggesting that this is not as important a step in the early formation of a biofilm compared with Salmonella. Increased expression of ompR may also have a role in the increased expression of genes within the two Salmonella pathogenicity islands, SPI-1 and SPI-2, despite the decrease in expression of the virulence regulator genes phoPQ. Kim and Falkow⁵⁶ showed that phosphorylated OmpR can activate expression of SPI-2 in the absence of PhoPQ.⁵⁶ Furthermore, the local activator for SPI-2, ssrA, was increased in expression after triclosan exposure, and previous work in Salmonella has shown that phosphorylated OmpR can bind to either sensor kinase gene ssrA or ssrB within SPI-2 and activate expression.^57,58

Other divergent regulatory gene expression changes were also observed, including motility genes. Decreased expression of Salmonella Typhimurium motility genes and motility was seen after triclosan exposure, whereas no difference in motility was observed for E. coli despite increased gene expression. However, energy to drive flagellar rotation is also required, suggesting that altered expression of flagellar genes alone may not be sufficient to confer phenotypic motility changes. Differential expression of flagellar genes has been previously observed; decreased expression in the presence of heavy metal within E. coli³⁹ and increased expression in the presence of sub-MICs of tetracycline in Salmonella Typhimurium.⁵⁹ Therefore, it is hypothesized that motility gene expression is dependent upon the nature of the toxic compound, the condition of the flagellae present and the uncoupling of energy sources to drive motility.

Expression of the regulatory genes fis and soxS was also divergent. The action of E. coli fis in the presence and absence of other transcriptional regulators such as marA and rob has been demonstrated to be far more complex than a simple binding activation pathway.⁶⁰ Whether the differential regulation of both fis and soxS is a direct response to triclosan exposure or an indirect result of altered expression of higher order regulators is not known. Given that within E. coli 37 transcriptional regulators (8 putative) and within Salmonella Typhimurium 24 transcriptional regulators (14 putative), were increased in expression, the regulatory networks responding to triclosan exposure are likely to be considerably more complex than a single response regulator whose action is conserved between both species.

An unexpected gene expression signature observed solely in E. coli encompassed the differential expression of operons responsible for iron homeostasis, such as the fhu, fep and ent operons. Previous transcriptomic work in E. coli identified a large regulon of genes under the control of the iron homeostatic regulator fur,⁴⁷ which bore similarity to the transcriptome obtained post-triclosan exposure. As no altered expression of fur was seen and there is no evidence that triclosan acts as an iron chelator, it is hypothesized that intracellular triclosan accumulation within E. coli is activating a fur-like pathway. Kohanski et al.³³ showed that bactericidal antibiotics induce the intracellular formation of toxic hydroxyl radicals in E. coli, which damage proteins and DNA.³³ Furthermore, the production of hydroxyl radicals through the Fenton reaction has been demonstrated to be partially dependent on the availability of ferrous iron.⁶¹ Given that the transcriptome in the presence of triclosan was similar to that of increased intracellular free iron, it is hypothesized that the bacterium is responding to hydroxyl formation by triclosan by attempting to change the amount of available intracellular iron. Whilst triclosan is known to be bactericidal at high concentrations, at the concentration used and over a 30 min exposure, no killing by triclosan was detected; nonetheless these data suggest that triclosan induced a similar response to other bactericidal antimicrobials, which Kohanski et al. indicated as resultant to hydroxyl formation. Further analysis of the microarray data suggested that very few genes common to the response to bactericidal antibiotics³³ were common to the E. coli triclosan response.

In summary, data obtained from this study allow a better understanding of the transcriptomic response in E. coli and Salmonella Typhimurium to a commonly used biocide. Our results also suggest that, as hypothesized, certain efflux pumps and regulators thereof form part of a core response to triclosan, supporting the potential for triclosan to induce MDR. Furthermore, these data indicate that triclosan may interact with multiple metabolic pathways. Finally, our data indicate that E. coli, often offered as a paradigm for the Enterobacteriaceae when exploring the effects of antimicrobial agents, responds quite differently to triclosan compared with at least one other member of the same taxonomic family.

Funding

A. M. B. and M. I. G. were supported by the Bristol-Myers Squibb Foundation Non-Restrictive Grant in Infectious Diseases to L. J. V. P., M. A. W. by BBSRC Grant BB/DD20476/1, A. I. and J. W. by The Wellcome Trust, M. J. W. and N. C. by Defra and C. C. and J. L. H. by BBSRC Grant EGA16107.

Transparency declarations

None to declare.

Acknowledgements

We thank Mala Patel, University of Birmingham, and Maria Fookes, The Wellcome Trust Sanger Institute, for help with the transcriptomic experiments. We also thank Professor Jay Hinton for helpful discussion.

Author notes

†

Present address: Fios Genomics Ltd, ETTC, King's Buildings, Edinburgh EH9 3JL, UK.

‡

Present address: Infection Prevention Team, The Chestnuts, The Royal Wolverhampton Hospitals NHS Trust, Wednesfield Road, Wolverhampton WV10 0QP, UK.

§

Present address: School of Biosciences, Faculty of Science, Sutton Bonington LE12 5RD, UK.

¶

Present address: Health Protection Agency, Centre for Infections, 61 Colindale Avenue, London NW9 5EQ, UK.