Overexpression of target enzyme gene fabI and efflux pump decrease triclosan susceptibility in Escherichia coli

Background: Escherichia coli isolates, the most opportunistic pathogen in the gut, are responsible for the most acquired infections. Triclosan is an effective disinfectant for inhibits microorganisms, but its widespread use causes its residue in urine, resulting in long-term exposure of E. coli in the intestine to triclosan environment and increasing triclosan resistance. We aim to provide the mechanism of action of E. coli isolates against triclosan and the molecular epidemiological analysis of triclosan-resistant strains. Results: Five triclosan-resistant isolates were screened out from 200 E. coli isolates by agar dilution method by to further study, interestingly, multidrug-resistant and cross-resistance phenotypes were observed in triclosan-resistant strains, but not in susceptible strains, and all except one exhibited an inhibition of efflux pump activity by efflux pump inhibition testing. Furthermore, compared with susceptible E. coli strain ATCC 25922, except fabI , increased expression were also found in efflux pump encoding genes ydcV , ydcU , ydcS , ydcT , cysP , yihV , acrB , acrD and mdfA in studied strains which had different PFGE patterns and STs types. Conclusions: These findings indicated that triclosan resistance in E. coli were mainly involved by overexpression of fabI gene, and there was a close association between overexpression of efflux pump with reducing susceptibility to triclosan. Besides, we described cross-resistance between triclosan and antibiotics may be related to the exposure time of triclosan.

urinary tract infection, as well as serious infections in the immunocompromised patients [2,3]. Over the past few decades, self-medication and antibiotics misuse led to the increasing resistance in clinical practice. Even worse, some of E. coli isolates have been reported as multidrug-resistant (MDR) pathogens (generally considered non-susceptible to ≥ 3 antibiotic classes) [4,5], which makes the clinical treatments of infections caused by E. coli face a great challenge.
Triclosan, a broad-spectrum and highly effective antibacterial agent, can inhibit various microorganisms at low concentrations, and be bactericidal at high concentrations [6]. In fact, it is not only used in disinfections, but also in medical equipment to prevent infections. According to a recent research report, there has a clinical effectiveness of triclosan-coated sutures compared to uncoated sutures for surgical site infections (SSIs) prevention [7]. Hence, triclosan plays a key role in reducing the dissemination and spread of pathogenic bacteria in hospital and community environments. Unfortunately, owing to increased clinical use, obvious levels of triclosan in various natural and engineered environment, such as soil and water, even in body fluids, such as urine, have been reported [8,9]. In China, Yin et al. estimated the average concentration of triclosan reached 0.36 μg/l in 80% urine samples [10]. Furthermore, triclosan was known as a "new environmental endocrine disruptor" due to its potential endocrine disrupting effects, which took an adverse effect on human health [11]. Therefore, widespread use or sustained exposure of triclosan has given rise to concern regarding its impact on accelerating the emergence of MDR E. coli, which poses great threats to public health as well as put the people at higher risk [12].
Although triclosan has a benefic effect on sterilizing for most pathogens in hospital settings, as previously mentioned, long-term exposure to triclosan have promoted a reduced sensitivity of triclosan in E. coli through extensive resistance mechanisms [13]. Of these resistance mechanisms, active efflux, reducing the drugs concentration in the bacteria whereby efflux pump systems to pump the intracellular antibacterial drugs out of the cell, conferred bacteria the ability to against a wide range of antimicrobials and biocides, including triclosan [14,15]. Indeed, many drug efflux pumps that were known to mediate resistance to traditional antibiotics and biocides include the efflux systems the resistance nodulation division (RND) family, the major facilitator superfamily (MFS), the staphylococcal multiresistance (SMR) and multidrug and toxic compound extrusion (MATE) families [16]. Additionally, triclosan was an inhibitor of enoyl-acyl carrier protein (ACP) reductase (FabI) enzyme, which involved in lipid biosynthesis and encoded by fabI gene.
Mutations within fabI contribute to E. coli resistance to triclosan [17,18]. However, the role of many efflux pumps in triclosan resistance in E. coli is not well understood. In order to provide a better scientific theoretical basis for the clinical rational use and nosocomial infections control in the future, investigations are urgently warranted.
In this study, we focus on gene expression and molecular typing to establish relation between the expression levels of efflux pump and triclosan resistance, and to investigate transmission of drug resistance in E. coli.

Susceptibility of triclosan and antibiotics before and after exposure to triclosan
MICs of triclosan among 200 isolates were ranging from 0.03125µg/ml to 8 µg/ml, five triclosan-resistant isolates were screened out (2.5%, 1/200), as shown in Table 1.
Interestingly, triclosan-resistant E. coli isolates tend to be resistant to multiple antibacterial agents, including ampicillin, cefepime, ceftazidime and gentamycin, but randomly selected five triclosan-sensitive strains did not exhibit multidrug-resistant phenotypes. Whether triclosan affects antibiotics sensitivity, further serial passage experiments were needed and performed.

Efflux pump phenotype testing
We sought to further investigate the basis for elevated triclosan MICs among the resistant isolates. The triclosan MICs of DC8358, DC8419, DC8424 and DC8724 reduced by 8,4,4 and 16 times in the presence of the concentration of 10 μg/ml of CCCP compared to the absence of CCCP, respectively. The result indicated that efflux pump systems were extremely active among above mentioned four isolates. Inversely, DC8603, unlike other resistant strains, showed a negative phenotype in efflux pump testing, which the MICs value of triclosan had not changed whether with the presence or the absence of CCCP ( Table 2).

Analysis of fabI gene mutation
PCR revealed that fabI and 14 efflux pump encoding genes were present in all tested strains except acrF gene. A variety of different mutations were detected in both resistant and susceptible strains. In DC8419 and DC8424, the Gly79Ala and Ala69Thr mutations were found, respectively. However, these two mutations were also observed in susceptible strains. Besides, we discovered other mutations of fabI gene in susceptible strains, such as Met2Arg, Ser5Leu, Val4Ser and Asp235Glu. Nonsense mutations were found in other resistant strains (Table 1).

Expression levels of target encoding gene and efflux pump encoding genes
The expression levels of fabI gene were evaluated in this study. Increased expression (> 2-fold) of fabI gene was observed in all triclosan-resistant strains. Compared with the triclosan-susceptible control strain ATCC 25922, the fold-changes of fabI gene with notable changes in the expression levels were between 5.69 to 41.85 times (Fig. 1).
In addition, to gain the better understanding of the relationship between triclosan resistance and gene expression level of efflux pump, expression of different efflux pump types was also examined for resistant strains as shown in Fig. 2 3.89 ± 0.2; yihV, 2.00 ± 0.03, respectively) was observed in DC8724. The result was consistent with efflux pump inhibition testing, which indicated that overactivity of efflux pump induced overexpression of efflux pump gene, which in turn mediates the susceptibility of triclosan.

Molecular epidemiological analysis
PFGE analysis revealed that the similarity of these isolates was low (< 0.85) due to the large differences in PFGE patterns. Similarly, the results of MLST comfirmed that they were categorised into multiple and scattered STs, including ST3, ST833, ST567, ST471 and ST1, respectively (Fig. 3). In short, the results above illustrated that triclosan-resistant strains had extremely low clonal relatedness in this study.

Discussion
In our study, 200 E. coli isolates were collected from patients with UTIs, and only five triclosan-resistant strains (2.5%, 5/200) were selected, the rate of which was lower than the previous report [19]. These triclosan-resistant strains, rather than susceptible strains, were characterized by a MDR profiles, we supposed that there may be a cross-resistance between triclosan and antibiotics in E. coli based on previous studies [20,21].
To test our hypothesis, DC8361, DC8363, DC8400, DC8413 and DC8510 parent strains that were sensitive to almost all antibiotics and triclosan, have been randomly selected for serial passage experiments. But it was not consistent with our supposition that changes in antimicrobial susceptibility were not observed before and after these strains adaptation, except triclosan. Actually, cross-resistance has been reported under the long-term selective pressure of triclosan, rather than in the short term according to Kampf et al. [20,22], that's why cross-resistance has not been observed during a continuous induction period of four days. Based on the obtained phenomena, there maybe had a close relationship between cross-resistance and exposure time of triclosan, which provided a sufficient evidence for clinical practice to avoid long-term use of triclosan to prevent cross-resistance. Moreover, it is worth noting that significantly reduced triclosan sensitivity was found by serial passage experiments, which provided a beneficial guidance for reasonable use of triclosan, including concentration, dosage and exposure time, in order to prevent the increase resistance of pathogens to triclosan.
Nevertheless, these mutations were not identified in our study. Notably, it was the first time that two different mutations in fabI , Gly79Ala and Ala69Thr were characterized.
However, these mutations were detected both in resistant strains and in susceptible strains, in other words, they may have no effect on the susceptibility of triclosan. Besides, as described before [23], overproduction of target gene fabI was also observed, which was the most commonly described resistant mechanisms of triclosan in E. coli.
Previously study suggested effectively upregulated efflux pump genes played an important role in triclosan non-susceptibility E. coli in a hostile environment [24]. In this respect, we tested the activity of four different types of Tolc and their relative expression of the corresponding coding genes, including ABC transporters system, Arac-regulator genes, the MdfA efflux Tolc, and the NorE efflux pump. Under triclosan stimulation, increased efflux pump activity of the isolates were found in our study, and the expression levels of ABC transporters system encoding genes ydcT, ydcU, ydcV, ydcS, and cysU and RND-type tolc encoding genes acrB, acrD and yihV, as well as mdfA gene which belonging to MFS family had a significant increase compared with ATCC 25922, which were consistent with a previous study in China [23]. In contrast to previous studies we did not observed any increase in the expression levels of other efflux pump encoding genes as evidenced by Sonbol and Curiao et al. [15,21], maybe it can be understood that ABC transport efflux pump, RND-type tolc and MFS family activity a stronger advantage than others during making adaptive changes of the studied isolates to triclosan [15,21]. All in all, we found that there was a strong relationship between gene overexpression and the increased tolerance of E. coli against triclosan, which suggested multiple efflux pumps may synergistically mediate triclosan resistance.
In addition, PFGE and MLST were useful and helpful methods for phylogenetic relationships of a large collection of bacterial lineages. Different pulse types and STs types were observed in the studied isolates by PFGE and MLST, suggesting that there was no transmission and a clonal dissemination among these triclosan-resistant strains.
However, our study also has some limitations. Although, cross-resistance between triclosan and antibiotics was noticed here, we did not illuminate the possible triclosan cross-resistance mechanisms, which is the focus of our future research.

Conclusions
Our study fills the gap in the field that various efflux pumps mediated triclosan resistance by analyzing genes expression which indicated that triclosan resistance was mainly mediated by fabI overexpression, and corresponding overexpression of efflux pump genes may contribute to enhanced triclosan tolerance in E. coli. Besides, we found crossresistance between triclosan and antibiotics may be related to the exposure time of triclosan, which will provide a better scientific theoretical basis for rational use of triclosan in clinic and nosocomial infections control. Hence, the necessary monitoring methods, such as MICs, β-lactamase and resistance genes mutation, are still important for clinical control of triclosan resistance.

Antimicrobial susceptibility testing
Agar dilution method was used to detect the minimum inhibitory concentrations (MICs) of clinical conventional antibiotics, including ampicillin, ciprofloxacin, levofloxacin, cefepime, ceftazidime, ertapenem, imipenem, gentamycin, nitrofurantoin and tobramycin, and the results were interpreted in accordance with the guidelines of the Clinical and Laboratory Standards Institute (CLSI) 2019 [25]. Furthermore, antimicrobial susceptibility testing of triclosan was performed by agar dilution method and according to previously study [19], isolates with an MICs value of > MIC 90 (Minimum drug concentrations at which 90% of strains were inhibited; MIC 90 = 0.5 µg/ml) were classified as triclosan-resistant. E. coli ATCC 25922 served as the quality control strain for susceptibility testing.

Serial passage experiments
In order to determine whether there was cross-resistance between triclosan and other antibiotics in E. coli, serial passage experiments were conducted in vitro for triclosansusceptible isolates DC8361, DC8363, DC8400, DC8413 and DC8510 as previously described [26]. Specifically, the isolates were cultivated on Maconkey ager plate and cultured overnight at 37 ºC to obtain a single isogenic strain, which was then inoculated

Mutant Stability testing
The stability of triclosan resistance was confirmed by continuous passage in vitrofor triclosan-mutant strains. Briefly, the triclosan-mutant strains were cultured in 3 ml fresh LB broth without triclosan at 37 ºC for 24 hours. Every 24 hours, 30 µl of overnight culture supernatants were transferred to another 5 ml tube containing 2.97 ml fresh LB broth without triclosan. After six cycles, the MICs of triclosan as well as ten antibiotics were tested in triplicate respectively using the same method described previously.

Efflux pump inhibition testing
To test efflux pump activity of triclosan-resistant E. coli, carbonyl cyanide 3chlorophenylhydrazone (CCCP) was trialed. The resistant strains were tested on agar plates with the presence or absence of efflux pump inhibitor CCCP by the agar dilution method as described [25]. Test concentrations of CCCP were ensured with consideration to determine the optimal subminimal inhibitory concentrations (sub-MICs) that could inhibit the overexpression of efflux pump without affecting the growth of bacteria. Compared with triclosan alone, MICs value of triclosan decreased by four times or more was confirmed having an inhibitory effect when triclosan was used in combination with the efflux pump inhibitors (CCCP, 10 μg/ml) [27].

Polymerase chain reaction (PCR) detection of mutations in fabI gene
Genome DNA of triclosan-resistant E. coli strains as well as randomly selected equal numbers of triclosan-susceptible strains, were extracted using the Biospin Bacterial

Quantitative real-time polymerase chain reaction (qRT-PCR)
Analysis of the transcriptional levels was undertook using qRT-PCR on six different ABC transporters system encoding genes ydcT, ydcU, ydcV, ydcS, cysP and cysU; two Aracregulator genes marA and soxS; four RND efflux pump encoding genes of the yhiv and

Genotyping by Multilocus sequence typing (MLST)
All the triclosan non-susceptible isolates were typed by using MLST method. In short, the sequences of eight housekeeping gene (trpB, uidA, dinB, icdA, pabB, polB, put and trpA) of E. coli were amplified with specific primers available at the MLST database (https://bigsdb.pasteur.fr/index.html), and sequence types (STs) were evaluated by comparing the allelic profiles to the MLST database.

Strain typing by pulse field gel electrophoresis (PFGE)
To confirm and analyze the clonal relatedness of the triclosan-resistant isolates, PFGE was also used for analysis the clonal relatedness of the triclosan-resistant isolates according to the PulseNet protocols published by the US Centers for Disease Control and Prevention (CDC) with minor modifications. The cell suspensions treated with protease K were incubated with XbaI restriction enzyme at least for 2 hours at 37 ºC to digest the DNA fragments. Then PFGE was performed using a CHEF-MAPPER XA PFG system (Bio-Rad, USA) for 18 hours. The detailed running condition were as follows: initial switch time value of 2.16 sec, final switch time of 54.17 sec at a gradient of 6 V/cm at a 120° included angle [28]. Next, the electrophoretic banding patterns were visualized by GelDoc XR gel imaging system (Bio-Rad, USA) and further analyzed by Quantity One (Bio-Rad Laboratories, USA).
The Unweighted Pair Group Method with Arithmatic Mean (UPGMA) with optimization set at 1.5% to create the dendrogram, cut off line at 85% was considered to analyze genetic relatedness [29]. Salmonella standard strain H9812 was taken as the positive control.

Consent for publication
Not applicable.

Availability of data and materials
All data generated or analyzed during this study are included in this published article.

Competing interests
The authors declare that they have no competing interests

Funding
This work was financially supported by Planned Science and Technology Project of Wenzhou (no. Y20180193). The funder had no role in the design of the study and collection, analysis, and interpretation of data and writing of the manuscript.