Characterization of Triclosan-Resistant Mutants Reveals Multiple Antimicrobial Resistance Mechanisms in Rhodospirillum rubrum S1H

Abstract

Antimicrobial resistance mechanisms were identified in 11 spontaneous high- and low-level triclosan resistance (Tcs^r) mutants of Rhodospirillum rubrum S1H by genotyping complemented with transcriptional analyses, antibiotic resistance screening, and membrane permeability analyses. High-end Tcs^r (MIC = 8 mg/liter) was the result of a FabI1(G98V) mutation. This point mutation led to an even higher level of Tcs^r (MIC ≥ 16 mg/liter) in combination with constitutive upregulation of mexB and mexF efflux pump homologs. Hence, a mechanistic synergy of constitutive efflux pump expression and a FabI1 point mutation could prevent TCS-induced cell permeabilization, which was shown to occur between 4 and 8 mg/liter TCS in the R. rubrum S1H parent strain. Low-level Tcs^r mutants constitutively upregulated the emrAB, mexAB, and/or mexF homolog. The mutants that overexpressed emrAB also derepressed the micropollutant-upregulated factors mufA1 and mufM. In some cases, low-level Tcs^r decreased innate resistance to ampicillin and tetracycline, while in others, a triclosan-induced antibiotic cross-resistance was shown for chloramphenicol and carbenicillin. This study showed that the TCS resistance degree is dependent of the initial exposure concentration in Rhodospirillum rubrum S1H and that similar resistance degrees can be the result of different defense mechanisms, which all have distinct antibiotic cross-resistance profiles.

Triclosan (TCS) is a broad-spectrum antimicrobial biocide that is incorporated in a multitude of contemporary personal care products due to its low toxicity in humans (22). In view of its widespread use and chemical rigidity, TCS and its derivatives have been detected in various matrices across the globe. While some bacteria, like the opportunistic pathogen Pseudomonas aeruginosa PAO1, possess innate high-level resistance to triclosan (8, 10), some bacterial species, such as the model species Escherichia coli, Salmonella enterica, Staphylococcus aureus, and Mycobacterium tuberculosis, can become more resistant through mutagenesis of given resistance mechanisms (3, 39). The mechanisms of conferred TCS resistance can take various forms, including target mutation, increased target expression, induction of efflux pumps, decreased influx or membrane permeability, and TCS transformation or degradation (37). However, little is known of the effects of TCS on environmental strains.

The purple nonsulfur α-proteobacterium Rhodospirillum rubrum is commonly found in soil and surface waters and has been reported to occur in wastewater-treating anaerobic bioreactors (17, 35). R. rubrum S1 has become increasingly valuable in biotechnological applications, such as hydrogen- and poly-β-hydroxyalkanoate production (13), because of its wide metabolic spectrum. The l-threonine-resistant derivative strain R. rubrum S1H has been chosen as a carbon- and nitrogen-mineralizing microorganism in the Micro-Ecological Life Support System Alternative (MELiSSA) (21, 34). This system aims to sustain astronauts during long-haul space missions by providing an alternative food source and producing oxygen, while biodegrading the organic waste and sequestering carbon dioxide (21). R. rubrum is known to be remarkably resistant to rare heavy metal oxyanions by detoxifying these compounds, a trait that is typical for purple nonsulfur bacteria (33). However, the mechanisms conferring resistance to antimicrobials in R. rubrum have not been described to date. In a previous study, we showed that very low concentrations of TCS (≥0.025 mg/liter) caused severe growth inhibition by extension of the lag phase but had little effect on growth rate. The whole-genome transcriptomic approach yielded a global view of the response mechanisms of R. rubrum S1H after TCS-induced inhibition. The results suggested the potential involvement of several efflux systems and the micropollutant-upregulated factor (muf) genes in intrinsic TCS resistance (35a).

In the present study, TCS was used at higher concentrations, as a selective agent, to generate Tcs^r variants of R. rubrum S1H, which could subsequently be used to identify antimicrobial resistance mechanisms. The study aimed at (i) determining the intrinsic TCS resistance degree and the mutation rate of R. rubrum in response to TCS under aerobic chemoorganotrophic and anaerobic photoheterotrophic conditions, (ii) confirming the stability of the conferred resistance phenotypes during long-term aerobic and anaerobic cultivation, (iii) assessing the nature of the conferred resistance mechanisms through gene sequencing and differential gene expression analysis of the resistant mutants compared to the level for the parent strain, and (iv) screening for antibiotic cross-resistance in the Tcs^r mutants.

MATERIALS AND METHODS

Strains and growth conditions.

Rhodospirillum rubrum S1H was obtained from the American Type Culture Collection (ATCC 25903). R. rubrum S1H is a derivative strain of R. rubrum S1 (34), of which the genome sequence is publically available. For dark aerobic (DAE) growth, R. rubrum S1H was cultured in Sistrom medium A (SIS) with 0.2% (wt/vol) succinate as the sole carbon source and incubated in the dark (40). For photoheterotrophic (light anaerobic [LAN]) growth, R. rubrum S1H was grown in SIS with 0.2% succinic acid (40) or in the basal salt medium of Segers and Verstraete (BSV) with either 0.2% (wt/vol) acetate or lactate as a carbon source for metabolic fitness assessment (15). (NH₄)₂SO₄ (0.5 g/liter) was supplied as a nitrogen source. LAN cultures were grown in a Certomat BS-1 shake incubator (Braun B. Biotech International). SIS-agar was prepared by adding 1.5% (wt/vol) agar prior to sterilization. All liquid cultures were grown at 150 rpm and incubated at 30°C.

Determination of the MIC for TCS under DAE and LAN conditions.

Prior to the plating experiments, SIS-agar was spiked from a triclosan (97.0% high-performance liquid chromatography [HPLC] grade; Fluka) stock solution in methanol. The different TCS concentrations were supplemented with equal final volumes of methanol to nullify the potential effect of differential solvent concentrations. In addition, a solvent control (with only methanol) and growth control (without methanol) were prepared. The final TCS concentrations ranged between 0.01 and 0.8 mg/liter. Cells from triplicate stationary-phase cultures (under DAE and LAN conditions) were harvested and washed with 0.85% NaCl. A 10-fold dilution series was made in 0.85% NaCl, which was spotted in triplicate 50-μl drops onto SIS-agar with and without TCS. For the concentrations of ≥0.075 mg/liter TCS, the drops were spotted 12-fold to increase the resolution of the assay. For DAE conditions, the petri dishes were incubated aerobically at 30°C in the dark for 8 days. For LAN cultivation, the plates were placed in BBL GasPak anaerobic containers, using one BBL GasPack Plus system (Becton, Dickinson, and Company) per container and using tungsten lamps as a light source. An Anaerotest (Merck) anaerobic indicator was included in each container to monitor the anaerobic conditions during the incubation for 2 weeks at 28°C. The viable count (CFU/ml) was determined as a function of the TCS concentration. The MIC was defined as the intersection of the extrapolated initial inhibition slope with the x axis (i.e., the TCS concentration), as previously described (31).

Cell permeability assay.

To determine the concentration at which TCS elicits cell permeabilization in R. rubrum S1H, exponential-phase cultures grown under DAE conditions (optical density at 680 nm [OD₆₈₀] = 0.4) were harvested, spun, and resuspended in SIS at 30°C supplemented with 1 to 12 mg/liter TCS or methanol-supplemented SIS as a solvent control. Acute exposure of R. rubrum S1H to these concentrations was limited to 30 min at 30°C, after which the cells were washed in prewarmed, 0.2-μm-filtered SIS. The washed cell suspensions were subsequently diluted 100-fold in phosphate-buffered saline (PBS) for flow cytometric analyses with a Beckman Coulter Epics XL-MCL flow cytometer. An L-7007 LIVE/DEAD BacLight bacterial viability kit (Invitrogen) was used in accordance with the manufacturer's instructions, with a cutoff set to 10,000 recorded events per analysis. All analyses were performed in technical duplicates for each of the five biological replicas.

TCS-resistant-mutant selection and mutation frequency.

Six colonies were isolated from each of the 10 TCS concentrations tested under DAE conditions. Thus, 60 colonies were arbitrarily isolated and purified by cultivation under DAE conditions for 8 days (i.e., ∼30 generations) in liquid SIS without selective pressure. Subsequently, the isolates were rechallenged to TCS by using an 8- by 6-array replica plater (Sigma-Aldrich). The isolates that were able to grow beyond the MIC of the parent strain were considered Tcs^r variants. For the TCS concentrations that yielded Tcs^r mutants, the mutation frequency was determined. The mutation frequency was determined by calculating the ratio of the average number of CFU/ml at 200, 400, and 800 μg/liter triclosan to the total number of seeded CFU/ml (i.e., the number of unexposed cells).

Isolation of TCS-resistant mutants.

Colonies grown under DAE conditions were arbitrarily picked up from SIS-agar supplemented with 0.05 to 0.2 mg/liter TCS, after which they were purified and screened for decreased TCS susceptibility. The stability of the Tcs^r of the isolates was examined after cultivation for 40 consecutive days in SIS without TCS under DAE conditions, with serial subculturing every 4 days (upon the reaching of the cell density maximum, 2 × 10⁸ cells/ml). In parallel, the isolates were verified for metabolic fitness by growing them for 14 days under LAN conditions in liquid SIS with succinate or liquid BSV supplemented with acetate and lactate (16). If no difference in MIC was observed after the long-term DAE or LAN experiment, the Tcs^r variants were considered stable. The Tcs^r mutants were stored at −80°C in 15% glycerol-0.85% NaCl. A stationary-phase culture of mutant M36 grown under DAE conditions was plated onto SIS-agar containing 16 mg/liter TCS. The plate was incubated for 7 days at 30°C under DAE conditions. One colony was isolated from the petri plate, purified, and stored to yield the derivate mutant M36⁺.

Primer design.

The primers for DNA sequencing and for differential gene expression analysis were designed using the online Primer3 website, version 0.4.0 (http://frodo.wi.mit.edu/), or Primer Express version 3.0 (Applied Biosystems). Potential secondary structures of the primer pairs were screened using the online primer analysis software NetPrimer (Premier Biosoft). All primers used in the present study are presented in Table S1 in the supplemental material. The genes mufA2, mufB, mexC, and mexE were not included in the study, because no appropriate primers could be constructed.

Sequencing of the presumed enoyl-ACP reductase gene (fabI1) in R. rubrum S1H.

Stationary-phase cultures of the parent strain and the Tcs^r variants grown under DAE conditions were subjected to a phenol-chloroform DNA extraction protocol (7). Genomic DNA concentration and purity were determined with a NanoDrop ND-1000 spectrophotometer (NanoDrop Technologies). A fragment of the gene encoding the putative enoyl-acyl carrier protein (enoyl-ACP) reductase (fabI1) in R. rubrum S1H was amplified using 1.25 units of proofreading Pfu DNA polymerase (Fermentas), 5 μl of 10× Pfu buffer (Fermentas), 1 μM each sequencing primer presented in Table S1 in the supplemental material, 5 μl of 2 mM deoxynucleoside triphosphate (dNTP) mixture (Fermentas), and 10 ng template DNA in a total reaction volume of 50 μl. The PCR was performed with a model 2700 Geneamp PCR system (Applied Biosystems). The PCR program consisted of an initial denaturation for 5 min at 95°C, followed by 35 cycles of 30 s at 95°C, 30 s at 55°C, and 1 min at 72°C, followed by a final extension of 10 min at 72°C. The reaction yielded an amplicon of 319 bp that covered the two domains that constitute the FabI1 active site, i.e., the 98th and 158th amino acids. Duplicate PCR mixtures were pooled and purified using the Wizard SV gel and PCR clean-up system (Promega), and 30 μl of the 50-ng/μl purified PCR product was sequenced by GATC Biotech (Germany), using the forward and the reverse primers listed in Table S1 in the supplemental material. The obtained sequences were compared to the publicly available sequence of R. rubrum S1 using the sequence alignment tool ClustalW2 (http://www.ebi.ac.uk/Tools/clustalw2/index.html).

RNA extraction and reverse transcription.

Triplicate cultures were grown in SIS without TCS under DAE conditions. Ten-milliliter aliquots of the cultures were harvested at OD₆₈₀s of 0.35 to 0.4 and immediately put on ice. The cultures were centrifuged (for 5 min at 9,820 × g) at 4°C, decanted, and stored at −80°C overnight. The following day, the pelleted cells were subjected to an RNA isolation procedure using the SV total RNA isolation system (Promega). RNA quality and concentration were determined using a NanoDrop ND-1000 spectrophotometer (NanoDrop Technologies). The total RNA extracts were stored at −80°C. The reverse transcription reaction to cDNA and subsequent quantitative PCR (Q-PCR) analysis were performed as previously described (12), using the primers listed in Table S1 in the supplemental material. The data processing was performed using the delta-delta threshold cycle (ΔΔC_T) method as previously described (12). To correct for differences in the amount of starting material, several presumed housekeeping genes were tested as potential normalization genes (i.e., the genes encoding 16S rRNA, 23S rRNA, and glyceraldehyde-3-phosphate dehydrogenase). The 23S rRNA gene was chosen as a normalization gene, as its expression showed the least variation between the parent strain and the different mutants and between replicas. The targeted genes were regarded as differentially regulated only when P was ≤0.05 in a two-tailed t test. Fold induction/repression values of ≥2.0 and ≤0.5 were used as cutoff values.

Determination of antibiotic MICs.

Triclosan-induced antibiotic cross-resistance in the Tcs^r mutants was scrutinized by comparison with the innate resistance degree of the R. rubrum S1H parent strain. For the exposure experiments, SIS-agar was supplemented with 2-fold-concentration gradations for tetracycline, ampicillin, carbenicillin, nalidixic acid, chloramphenicol, kanamycin, streptomycin, and gentamicin (BioChemica; AppliChem). The antibiotics were chosen to cover a range of antibiotics with different physicochemical properties and modes of action. In addition, some structurally related antibiotics were also included in the study. For these compounds, similar cross-resistance profiles were expected, due to similar chemical properties. The early-stationary-phase Tcs^r mutants grown under DAE conditions were spotted in triplicate using a replica plating approach. Six early-stationary-phase cultures of the parent strain were also spotted, as controls. For DAE conditions, the petri dishes were incubated aerobically at 30°C in the dark for 8 days.

RESULTS

Triclosan-induced growth inhibition under DAE and LAN conditions.

Upon exposure of R. rubrum S1H to TCS in SIS-agar, a biphasic inhibition curve was observed under both DAE and LAN conditions. An initial ca. 5-log decrease in viable count was observed between 0.025 and 0.05 mg/liter TCS. By extrapolation of the slope to the intersection with the x axis (31), the MIC of R. rubrum S1H for TCS was determined to be 0.071 mg/liter under DAE conditions and 0.092 mg/liter under LAN conditions (Fig. 1). At 0.05 mg/liter TCS, a deviation from the initial exponential slope was observed under both growth conditions (Fig. 1). This ca. 3-log decrease in viable count was observed between 0.05 and 0.15 mg/liter TCS for DAE conditions and between 0.05 and 0.2 mg/liter TCS for LAN conditions. Between 0.15 and 0.8 mg/liter TCS, highly similar counts in viable cells were observed under DAE conditions, which could be indicative of mutational adaptation to TCS. Under LAN conditions, however, no colonies that could grow above 0.2 mg/liter TCS after a single round of exposure to TCS were observed. From this experiment, it was apparent that the cellular response to TCS was complex and concentration dependent, especially between 0.05 and 0.2 mg/liter TCS, but relatively similar between cells grown under DAE and LAN conditions.

FIG. 1.

The biphasic growth inhibition profiles of R. rubrum S1H grown on SIS-agar supplemented with triclosan under dark aerobic (DAE) (open triangles, dotted line) or light anaerobic (LAN) (open circles, full line) conditions. The extrapolation of the initial slopes illustrates the determination of the MIC at the intersection with the x axis (y = 1 × 10⁰). Error bars indicate the standard deviations for biological triplicates.

Triclosan-induced cell permeabilization.

Complementary to the growth inhibition assay, a flow cytometric assay was performed to screen for TCS-induced membrane permeabilization in DAE-condition-grown R. rubrum S1H upon acute exposure (30 min) to high levels of TCS. The fraction of the total population that suffered a loss of outer membrane integrity increased significantly (≥4 mg/liter TCS) (see Fig. S1 in the supplemental material). After short exposure to 8 mg/liter TCS, a total of 93% of the wild-type cells were scored as having a compromised membrane. After exposure to 12 mg/liter TCS, >99% of the wild-type cells had a permeabilized cellular envelope. Hence, any triclosan-resistant mutant with a MIC above 4 mg/liter is likely to express a resistance mechanism that circumvents membrane permeabilization.

Triclosan-resistant-mutant selection and mutation frequency.

A concentration-dependent increase in Tcs^r variants was observed, as the isolates originally exposed to ≥0.075 mg/liter TCS (i.e., above the MIC under DAE conditions) could grow beyond the concentrations from which they were isolated after the nonselective treatment, with 1.6 mg/liter being the highest TCS concentration tested during the rechallenge. Two out of six DAE-condition-grown colonies isolated from 0.075-mg/liter TCS could grow with up to 0.075 mg/liter TCS, of which one could grow with up to 0.2 mg/liter TCS (ca. 3× MIC). All six DAE-condition-grown colonies isolated from 0.1 mg/liter TCS could grow with up to 0.075 mg/liter, of which two could grow with up to 1.6 mg/liter TCS (ca. 22× MIC). All six DAE-condition-grown colonies isolated from 0.15 mg/liter TCS could grow with up to 0.15 mg/liter TCS, of which five could grow with up to 1.6 mg/liter TCS (ca. 22× MIC). All six DAE-condition-grown colonies isolated from 0.2 mg/liter TCS could grow with up to at least 1.6 mg/liter TCS. Thus, a stable mutation frequency of 2.39 × 10⁻⁸ was observed between 0.15 and 0.8 mg/liter TCS under DAE conditions (Fig. 1). This experiment illustrated that although a stable mutation frequency was observed at ≥0.15 mg/liter TCS, nuances in TCS resistance degrees exist between 0.075 and 0.2 mg/liter TCS, which most likely originated from distinct mechanistic origins.

On the basis of their MICs, several isolates were selected for further experimentation in order to have at least one representative of each of the different Tcs^r degrees (Table 1). All Tcs^r variants were shown to be genetically stable after long-term serial subcultivation (for 40 days) under DAE conditions. In addition, all had retained the ability to grow under LAN conditions with succinate, acetate, or lactate as a carbon source, while preserving their distinctive Tcs^r phenotypes.

TABLE 1.

MICs of a selection of 10 triclosan-resistant isolates for triclosan and selected antibiotics^a

Strain	TCS level (mg/liter) at origin	MIC (mg/liter)
		TCS	Cm	Amp	Car	Te	Nal	Km	Sm	Gm
S1H	NA	0.050	4	400	200	0.400	320	1	4	0.8
M26	0.075	0.075	2	−	400	0.100	160	−	−	−
M17	0.075	0.100	>8	50	100	0.200	−	−	2	0.4
M3	0.075	0.100	8	−	−	0.800	−	−	−	0.4
M13	0.075	0.100	8	>400	800	0.800	−	−	−	−
M27	0.100	0.100	8	>400	800	−	−	−	−	−
M21	0.100	0.250	>8	200	100	0.200	>320	−	2	0.4
M28	0.100	0.250	−	200	400	−	−	>1	−	−
M34	0.100	0.500	8	−	800	−	−	−	−	−
M33	0.100	1.000	2	200	100	−	160	−	−	0.4
M36	0.200	8.000	−	−	−	−	−	−	−	−
M36+	NA	>16.000	ND	ND	ND	ND	ND	ND	ND	ND

^aAbbreviations: Te, tetracycline; Amp, ampicillin; Car, carbenicillin; Nal, nalidixic acid; Cm, chloramphenicol; Km, kanamycin; Sm, streptomycin; Gm, gentamicin; NA, not applicable; ND, not determined. The isolates are sorted based on the MICs for TCS and their antibiotic cross-resistance profile, and for every mutant, the level of triclosan observed at the origin of the isolate is given. Mutant M36⁺ is a derivative of mutant M36. Increased MICs are indicated in bold. A minus sign indicates that the MIC of the isolate is identical to that of the parent strain.

Bioinformatic search for potential antimicrobial resistance mechanisms in R. rubrum S1H.

Two genes that show homologies with the enoyl-acyl carrier protein (enoyl-ACP) reductases of Rhodobacter sphaeroides, namely, fabI1 and fabI2, were found in the R. rubrum S1 genome (see Table S1 in the supplemental material). Six RND efflux pump systems that showed homologies with the multidrug resistance (MDR) efflux pumps of Pseudomonas aeruginosa (MexAB-OprM, MexCD, MexEF, and PA4375-PA4374) and E. coli K-12 (AcrAB and EmrAB) were found in the genome. Also, genes encoding two E. coli K-12 protein diffusion channels (porins), OmpC and OmpF, were found in the R. rubrum S1 genome. The R. rubrum genome possesses six genes annotated as β-lactamase-like genes and one gene as a β-lactamase gene (Rru_A1250), and the latter gene was targeted in the present study. No homologs of the E. coli AmpC protein were found in the R. rubrum S1 genome. Our previous study, which aimed at identifying genes in R. rubrum S1H that respond to exposure to low TCS concentrations (0.01 mg/liter and 0.025 mg/liter TCS), showed important upregulation of the genes encoding micropollutant-upregulated factors (mufA1 and mufM), phage shock proteins (pspA and pspC), the oxidative stress-related glutathione peroxidase (gpo), and exopolysaccharide biosynthesis proteins (capD and exoD) (35a). Therefore, these genes were also scrutinized in the present study. In addition, the genes encoding the putative short-chain fatty acid transporter (atoE), two efflux pump components (acrC and Rru_A1640), and the polyhydroxybutyrate synthase (phaS) were also included.

Sequencing of the enoyl-ACP reductase gene (fabI1).

The gene encoding the putative FabI1 protein in R. rubrum S1H was sequenced for each of the 10 Tcs^r mutants and the parent strain. Sequencing with the forward and the reverse primers (see Table S1 in the supplemental material) yielded fragments between 196 and 272 bp; in all cases, the codons encoding both the 98th and the 158th amino acids were found. The fabI1 sequence remained 100% identical to the R. rubrum S1 sequence for the parent strain S1H and all of the mutants, except for mutants M36 and M36⁺, which appeared to have undergone a point mutation in the region of the 98th amino acid, with a TCS-selected sequence alteration from GCC to GTC, corresponding to a FabI1(G98V) mutation. At the Pro¹⁵⁸ site, no differences were observed for any of the mutants. It is believed that the modification of Gly⁹⁸ to Val⁹⁸ conferred high-level specific resistance to TCS in mutants M36 and M36⁺ (Table 1).

Identification of resistance mechanisms through transcriptional analyses.

In total, 22 genes were screened for constitutive expression in the 10 Tcs^r variants to assess their potential involvement in antimicrobial resistance. Of those, only 15 genes were differentially regulated in at least one mutant compared to the level for the parent strain S1H (Table 2). The remaining 7 genes (ompC, ompF, fabI1, fabI2, acrC, phaS, and Rru_A1640) were not constitutively upregulated in any mutant. The most strongly constitutively overexpressed genes were the micropollutant-upregulated factors mufA1 and mufM (Table 2). Both mufA1 and mufM were found to be important in 5 of the 10 mutants (M26, M13, M27, M28, and M34), with expression ratios between 141.6- and 351.9-fold for mufA1 and 8.1- to 25.9-fold for mufM. The extremely high fold induction values are probably indicative of a “derepression” of the muf genes in the respective Tcs^r mutants. In addition, the overexpression of muf genes appeared to be limited to mutants with low-level TCS resistance, as these genes were not found upregulated in mutants resistant to high TCS concentrations (M33, M36, or M36⁺) (Table 2). Interestingly, the mutants which overexpressed both muf genes also upregulated the emrAB operon (Table 2).

TABLE 2.

Genes constitutively upregulated in the Tcs^r mutants compared to the level for the parent strain R. rubrum S1H^a

Strain	TCS MIC (mg/liter)	Fold induction ratio for:
		Micropollutant-upregulated factors		emrA	emrB	Multidrug efflux pump genes				Rru_A1250	EPS synthesis genes		gpo	Phage shock protein genes		atoE
		mufA1	mufM			mexA	mexB	mexD	mexF		capD	exoD		pspA	pspC
M26	0.075	351.9 ± 154.9	25.9 ± 5.7	9.1 ± 1.5	2.6 ± 0.2	11.7 ± 2.0	5.2 ± 1.6	−	−	−	−	−	−	−	−	−
M17	0.100	2.3 ± 1.0	−	3.5 ± 0.3	2.5 ± 0.8	−	−	−	14.5 ± 2.3	4.3 ± 1.5	3.2 ± 0.2	−	−	−	−	−
M3	0.100	−	−	3.0 ± 1.9	−	2.3 ± 0.5	3.8 ± 0.9	−	−	2.6 ± 1.5	−	−	−	−	−	2.5 ± 0.4
M13	0.100	305.8 ± 56.6	23.4 ± 3.3	13.5 ± 5.6	2.6 ± 1.8	3.4 ± 0.4	4.1 ± 0.9	−	2.7 ± 1.5	3.5 ± 2.8	−	−	2.0 ± 1.1	−	2.6 ± 0.5	−
M27	0.100	141.6 ± 45.0	15.7 ± 4.3	11.7 ± 2.4	2.9 ± 0.8	−	2.4 ± 0.5	−	−	−	−	−	−	−	−	−
M21	0.250	2.8 ± 0.3	−	11.7 ± 7.4	2.2 ± 0.2	−	−	2.3 ± 2.2	19.4 ± 2.5	4.0 ± 0.2	2.5 ± 0.3	−	2.4 ± 0.7	−	−	3.4 ± 1.3
M28	0.250	143.4 ± 20.3	8.1 ± 1.8	8.3 ± 2.9	3.0 ± 0.8	−	2.0 ± 0.6	−	−	2.6 ± 1.0	−	−	−	−	−	−
M34	0.500	188.6 ± 51.7	17.8 ± 3.0	13.4 ± 2.5	2.9 ± 0.8	−	2.9 ± 0.3	−	−	−	−	−	−	−	−	−
M33	1.000	3.4 ± 0.7	−	−	−	−	−	−	−	2.4 ± 0.1	−	−	−	−	2.5 ± 0.3	−
M36	8.000	−	−	−	−	−	−	−	−	−	−	−	−	−	−	−
M36+	>16.000	−	−	−	−	−	2.6 ± 0.5	−	1.9 ± 0.4	−	−	−	−	−	−	−

^aValues are shown for genes with relative fold induction ratios of ≥2.0. The standard deviations for the biological triplicates are given. The genes fabI1, fabI2, ompC, ompF, acrC, phaS, and Rru_A1640 are not included, since they were not found significantly upregulated in any mutant. A minus sign indicates that the gene was not significantly (≥2.0-fold) upregulated. Bold indicates a P value of ≤0.05.

One mutant (M33) exhibited a relatively elevated level of resistance for TCS without overexpressing any of the 22 screened genes, except the presumed β-lactamase gene Rru_A1250. This gene was also significantly upregulated in two other mutants (M17 and M21), which also overexpressed mexF and capD (Table 2). Mutant M36, which had a MIC of 8 mg/liter TCS (Table 1) and which carried a point mutation in the FabI1 active site, was exposed to 16 mg/liter TCS on SIS-agar. Hence, the derivate mutant M36⁺ was yielded. Next to carrying the FabI1(G98V) mutation, mutant M36⁺ constitutively upregulated mexB and mexF (Table 2).

Determination of antibiotic MICs of the Tcs^r mutants.

It was hypothesized that different TCS resistance mechanisms could yield different antibiotic cross-resistance profiles among the Tcs^r mutants, despite similar Tcs^r degrees. Therefore, the Tcs^r mutants and the parent strain were exposed to a range of antibiotics to determine the respective MICs for tetracycline, ampicillin, carbenicillin, nalidixic acid, chloramphenicol, kanamycin, streptomycin, and gentamicin (Table 1). It was seen that some isolates with identical Tcs^r degrees had vast differences in their antibiotic cross-resistance profiles (Table 1). It was mainly the mutants with low Tcs^r degrees (with TCS MICs [MIC_TCS] between 0.1 and 0.5 mg/liter) that showed significantly increased or decreased antibiotic resistance compared to the intrinsic antibiotic resistance of the parent strain. For 3 mutants (M26, M17, and M21), the acquired Tcs^r phenotype caused those strains to become up to four times more susceptible to tetracycline, while the latter two also became up to eight times more susceptible to ampicillin. Two other mutants (M28 and M33) were also more susceptible than the parent strain, but for ampicillin alone. For mutants M17, M21, and M33, this increased Amp susceptibility correlated with a decrease in carbenicillin resistance. The latter observations are remarkable given the fact that the parent strain possesses seven genes currently annotated as β-lactamase(-like) genes and can be considered Amp^r, with a MIC of 400 mg/liter Amp. Mutants M17 and M21 were unique in that they were able to grow in the presence of 8 mg/liter chloramphenicol (Table 1).

One mutant (M13) showed broad-range antimicrobial resistance. Although its MIC for TCS had increased only 4-fold, this mutant had also become more resilient to tetracycline, ampicillin, carbenicillin, and chloramphenicol. Two other mutants, M3 and M27, with the same MIC_TCS (0.1 mg/liter), had become either more resilient to tetracycline and chloramphenicol or more resistant to the β-lactams (Amp and Car) and chloramphenicol, respectively.

The mutants with high-level TCS resistance (M33, M36, and M36⁺), with MIC_TCS of 1 to 8 mg/liter, were often more sensitive to antibiotics than the parent strain (e.g., mutant M33) or showed no antibiotic cross-resistance at all (e.g., mutant M36) (Table 1). The latter indicated that this Tcs^r mutant possessed a resistance mechanism that is highly selective for TCS.

Some mutants (M33, M17, and M21) were more sensitive to the β-lactam antibiotics than the parent strain, and all upregulated the one β-lactamase-annotated gene (Table 2). This observation would suggest that Rru_A1250 is not a β-lactamase but more likely another type of hydrolase with a β-lactamase-like protein domain or fold.

DISCUSSION

Factors determining the intrinsic and conferred TCS resistance degrees.

With the exception of P. aeruginosa, which possesses strong innate resistance against TCS (10), most Bacteria have a MIC in the low μg/liter range (1, 14, 41). Nevertheless, R. rubrum S1H was shown to be relatively sensitive to TCS-mediated inhibition for a Gram-negative bacterium, since its MIC is in the range that is reported for Gram-positive bacteria and the most-TCS-sensitive Gram-negative strains (14, 41). The present study also showed that photoheterotrophic R. rubrum S1H (grown under LAN conditions) is slightly more resistant to TCS than the faster-growing chemoorganotrophic R. rubrum S1H (grown under DAE conditions). This observation is consistent with previous reports, where bacteria with lower growth metabolisms (i.e., under oxygen-limited, anaerobic, or biofilm conditions) were found to be more resistant to various stressors, including triclosan (2, 4, 19, 42). Under both LAN and DAE conditions, a biphasic growth inhibition profile that was at least partially the result of adaptive mutagenesis was observed. It is possible that the inhibition curve is biphasic because of a regulatory threshold around 50 μg/liter TCS, especially since the bend in the inhibition curve was observed under both DAE and LAN conditions and since the existence of such a regulatory threshold concentration was shown to exist in our previous study (35a). It was previously shown that in fast-growing model bacteria, such as E. coli, a low innate resistance degree can be increased readily by way of four different approaches, i.e., target alteration (i.e., mutation), target protein overexpression, increased efflux, or decreased influx. However, a screen of 11 Tcs^r mutants showed that besides metabolism, target mutation and constitutive efflux upregulation are the most frequently occurring triclosan resistance mechanisms in the slow-growing environmental bacterium R. rubrum S1H.

FabI mutations confer high-level triclosan resistance.

The enoyl-acyl carrier protein reductase (FabI) catalyzes the final step in each fatty acid elongation cycle and is an important regulatory point in the fatty acid biosynthesis pathway. Some bacteria, like the facultative photoheterotroph Rhodobacter sphaeroides, possess two enoyl-ACP protein reductases, FabI1 and FabI2. Where the former displays most of the cellular enzyme activity, the latter seems to be weakly expressed and dispensable but is unaffected by TCS (24). The genes encoding FabI1 and FabI2 were also found in R. rubrum, but they are not overexpressed in TCS-exposed wild-type cultures (35a), nor were they found constitutively upregulated in any Tcs^r mutant in the present study. This stands in contrast with what was found for Staphylococcus aureus, where fabI upregulation was observed and led to a 2.5-fold-increased MIC compared to the level for TCS-susceptible strains (41).

Besides target upregulation, a mutation in the active site of FabI could alter the activity of the FabI protein to prevent TCS inhibition (30). In the present study, TCS was shown to induce a FabI1(G98V) point mutation in R. rubrum S1H. This adaptation, which is equivalent to the FabI(G93V/S) mutation in E. coli (20), appeared to be the most potent mutation in R. rubrum S1H for conferring TCS resistance. The G98V mutation probably caused sterical hindrance, preventing TCS from optimally interacting with NAD⁺ in the FabI1 active site (36). In S. aureus, a FabI(F204C) mutant that overexpressed the FabI protein was isolated (41). This S. aureus mutant has a MIC of 1 to 2 mg/liter, while in R. rubrum, a FabI1(G98V) mutation led to a MIC of 8 mg/liter. Moreover, in combination with efflux pump overexpression, this fabI1 mutation led to a MIC of >16 mg/liter. Hence, our results confirmed the hypothesis of Webber and colleagues (43), who found that a point mutation in the FabI protein of Salmonella enterica led to a minimal bactericidal concentration (MBC) between 4 and 8 mg/liter and presumed that another (unidentified) resistance system would cause the MIC to increase significantly. In the present study, it was shown that upregulation of the mexB and mexF efflux systems in R. rubrum could be at least one of these stated unidentified resistance mechanisms, which even at very low fold induction values were able to increase TCS tolerance significantly. This demonstrated that a FabI active site mutation is very efficient when combined with efflux pump overexpression, since the target mutation prevents the high-affinity complex formation while the efflux pumps can avert TCS-mediated cell lysis at high levels of TCS (as observed in the flow cytometric membrane permeability assay). Moreover, these results suggested that efflux pumps can remove the TCS fraction which accumulates in the cellular envelope and can thus prevent TCS-induced cell lysis. While R. rubrum S1H can be regarded as relatively sensitive to the bacteriostatic effects of TCS (compared to other bacterial species) (14), it is more resilient to the bactericidal effects than S. enterica and some strains of E. coli (29, 43). However, E. coli AGT11 appears more resistant than R. rubrum S1H when insensitized through a fabI mutation, with an MBC of >32 mg/liter TCS (29), while the Pasteurella multocida parent strain remains the most sensitive Gram-negative bacterium, with an MBC of 2 mg/liter TCS (14).

Efflux mechanisms alone confer low-level resistance in R. rubrum.

To date, five families of efflux transporter systems have been described (26). Efflux of TCS in Gram-negative bacteria has mostly been limited to members of the RND family (9-11, 29, 38). These efflux pumps can mediate efflux of TCS but also of a wide range of structurally unrelated antibiotics (39). Consequently, intrinsic and acquired Tcs^r via efflux confers a certain degree of antibiotic cross-resistance (9-11, 41). All efflux systems previously described to be of significance for TCS resistance in P. aeruginosa and E. coli were shown to be upregulated in one or more Tcs^r R. rubrum mutants. In addition, the present study was the first to show the involvement of the EmrAB homolog in acquired TCS tolerance. This confirms the importance of this efflux system (and of the others), previously reported to be upregulated in R. rubrum S1H upon exposure to low-levels of TCS (35a).

In P. aeruginosa, the MexAB-OprM system alone mediates efflux of a wide range of compounds, i.e., tetracycline, chloramphenicol, fluoroquinolones, β-lactams, and others (38). In R. rubrum, the MexAB-OprM homologs appeared to be less efficient for the efflux of antibiotics, as upregulation of the mexAB genes did not correlate directly with increased resistance to the tested antibiotics. This is consistent with the hypothesis that TCS resistance is species specific rather than caused by universal resistance mechanisms that occur in all Bacteria (6).

In R. rubrum S1H, the most-efficient (TCS- and antibiotic-eliminating) efflux pumps are most likely EmrAB and MexF, as their upregulation correlated well with increased β-lactam and chloramphenicol resistance, respectively. Cross-resistance to the latter antibiotic has often been described to occur in TCS-adapted cultures of E. coli and S. enterica (5, 6). This observation is most likely due to the fact that chloramphenicol is also hydrophobic, like triclosan, a property which is also important for the trans-envelope migration (14).

The loss of innate Amp^r in some of the Tcs^r mutants was most likely the result of a specialization event. The R. rubrum S1H MexD, AcrC, and Rru_A1640 efflux system components appeared not to be related to the Tcs^r phenotype.

In the TCS-resistant mutants, the low degree of resistance for antibiotics was most probably the result of the poor affinity of the efflux pumps (e.g., the AcrAB-TolC system) for the selected antibiotics (18, 29, 30, 32). Therefore, the only acquired form of Tcs^r that could be of environmental significance is the cross-resistance with antimicrobials affecting the same target as TCS, such as isoniazid, as was reported for Mycobacterium smegmatis (28). Thus, the FabI1 mutation that was discovered in the present study appears to be the most important resistance mechanism, from the environmental perspective.

Origin of efflux system upregulation.

Most efflux pump operons are regulated by adjacent transcriptional activators. However, in most of the isolated Tcs^r P. aeruginosa, E. coli, and S. enterica strains, a point mutation in the general regulatory genes (i.e., nfxB, marA, soxS, and ramA), instead of local ones, is at the origin of the efflux-mediated Tcs^r phenotype (6, 9, 23, 25, 39, 43). During the present study, more than one efflux pump could be upregulated in the same Tcs^r mutant of R. rubrum. This observation could be indicative of a hitherto unknown general stress regulator in R. rubrum, which regulated the expression of several defensive systems, as was observed in E. coli with marA and soxS (29). However, homologs of neither gene were found in the R. rubrum S1 genome. Potentially, the Muf proteins could fit this role in the Tcs^r variants, as their upregulation correlated with emrAB upregulation. Nevertheless, further experimentation is required to confirm this hypothesis.

Porins and conferred TCS resistance.

The fourth path of acquired Tcs^r that was screened in the present study is associated with a decrease in influx/membrane permeability. Even though the exact uptake mechanisms for hydrophobic compounds, such as TCS, across the cellular envelope are poorly understood, differences in permeability can be due to differential porin expression, porin modification, or exopolysaccharide overproduction. Despite the fact that porins have been shown to play an important role in antibiotic resistance in S. enterica, E. coli, Mycobacterium bovis, and Enterobacter aerogenes (16, 27, 43), in R. rubrum, the OmpC and OmpF proteins do not appear to play a role in the conferred TCS resistance. Neither of the corresponding genes was upregulated in the TCS-exposed parent strain (35a) or in the TCS-resistant mutants.

Environmental implications.

From our findings, we could conclude that the direct environmental implications of triclosan contamination seem to be limited, when using R. rubrum S1H as a model strain. On the one hand, the global environmental concentrations of triclosan are currently too low to cause an acute impact on bacterial proliferation. On the other hand, the risk of antibiotic cross-resistance (with any kind of significance) will be limited because (i) the concentrations required for mutant selection are too low in the environment, (ii) the pumps' efflux efficiencies were shown to be very low for a wide range of antibiotics, and (iii) the risk of antibiotic resistance dissemination from R. rubrum S1H is very low because the pumps are not locally regulated and the resistance mechanisms themselves are chromosomally located. Nevertheless, the implications of long-term exposure to environmentally significant triclosan concentrations would still need to be investigated.

Conclusions.

Replying to the need for surveys that demonstrate the impact of TCS in the natural environment (41), the present study assessed the resistance generation in the slow-growing environmental bacterium Rhodospirillum rubrum S1H. In the low μg/liter range, the growth inhibition profiles were found to be biphasic and shown to be at least partially the result of rapid adaptive mutagenesis. Conferred TCS resistance was often multifactorial, i.e., the result of a number of resistance mechanisms acting in synergy. Hence, it was evident that more than one “biological solution” for overcoming the TCS-mediated effects exists. However, the hazard of TCS resistance formation in the environment will most likely remain limited.