key: cord-0025691-tdxbhro9
authors: Shi, Danyang; Hao, Han; Wei, Zilin; Yang, Dong; Yin, Jing; Li, Haibei; Chen, Zhengshan; Yang, Zhongwei; Chen, Tianjiao; Zhou, Shuqing; Wu, Haiyan; Li, Junwen; Jin, Min
title: Combined exposure to non-antibiotic pharmaceutics and antibiotics in the gut synergistically promote the development of multi-drug-resistance in Escherichia coli
date: 2022-01-11
journal: nan
DOI: 10.1080/19490976.2021.2018901
sha: 5038f1e86f9e562c820e9ae4493cc70e4e8fb2f0
doc_id: 25691
cord_uid: tdxbhro9

The gut microbiota represents an important reservoir of antibiotic-resistant bacteria (ARB), which poses a significant threat to public health. However, little is known about the emergence of ARB in the gut after the combined exposure to antibiotics and non-antibiotic pharmaceutics. Here, Escherichia coli, a common opportunistic pathogen in the gut microbiota, was exposed to the antidepressant duloxetine (2.5 µg/L–25 mg/L) and/or chloramphenicol (6 µg/L–4 mg/L). The resistant strains were isolated to determine the minimum inhibition concentration (MIC) of 29 antibiotics. Then, genome-wide DNA sequencing, global transcriptomic sequencing, and real-time quantitative polymerase chain reaction were performed to quantify the synergy between duloxetine and chloramphenicol. Combined exposure synergistically increased the mutation frequency of chloramphenicol resistance by 2.45–9.01 fold compared with the independent exposure. A combination index reaching 187.7 indicated strong duloxetine and chloramphenicol synergy. The resultant mutants presented heritable enhanced resistance to 12 antibiotics and became ARB to eight antibiotics. Furthermore, combined exposure significantly increased the transcriptomic expression of acrA, acrB, and marA in E. coli, and generated a more robust oxidative stress response. Together with the occurrence of DNA mutations in marR in the mutants, stronger triggers to the AcrAB-TolC transport system and the MlaFEDB ABC transporter via reactive oxygen species (ROS)-induced mutagenesis, verified by gene knockout, contributed to the synergistic enhancement of antibiotic resistance in the combined exposure group. Regardless of whether their formation was induced by duloxetine, chloramphenicol, or their combination, the E. coli mutants showed 1.1–1.7-fold increases in the expression levels of acrA, acrB, acrZ, mdtE, and mdtF. This pattern indicated that the mutants shared the same resistance mechanisms against chloramphenicol, involving the improved efflux pumps AcrAB-TolC and mdtEF. Our findings demonstrated that antibiotics and non-antibiotic pharmaceutics synergistically accelerate the evolution of ARB and may enhance their spread.

The appearance of antibiotic-resistant bacteria (ARB) has resulted in the failure to treat serious infectious diseases in clinics. In the EU, 25,000 people die each year from infections caused by ARB. 1 It is alarming that if adequate measures are not taken to control the spread of ARB, an additional 10 million people may be killed globally each year. 2 Consequently, ARB have become a severe worldwide threat to public health.

The gut microbiota is an important reservoir of ARB; this could be related to the overuse or misuse of antibiotics in hospitals and the contamination of food and drinking water antibiotics. [3] [4] [5] [6] A clinical trial found that antibiotic concentrations in the guts of healthy individuals reached 26.51 ± 33.48 ng/g, which could put severe antibiotic pressure on gut microbiota and result in a global increase in the dissemination of ARB. 7, 8 Above all, with the discovery of the strengthened resistance of Escherichia coli to multiple antibiotics, e.g., tetracycline, chloramphenicol, cefalexin, levofloxacin, and norfloxacin, after exposure to the antidepressant fluoxetine, 9 the contribution of non-antibiotic pharmaceuticals, particularly antidepressant drugs, toward the dissemination of ARB has become an urgent concern due to their widespread use.

Non-antibiotic pharmaceuticals typically have a prolonged usage. As a result, patients could consume non-antibiotic pharmaceuticals and antibiotics together, resulting in the gut microbiota being exposed to both. The combined persistence of these drugs may pose a cumulative or magnified pressure on microorganisms. Therefore, the significant risk of ARB dissemination posed by the coexistence of non-antibiotic pharmaceuticals and antibiotics makes it imperative that we better understand the emergence of ARB under the combined exposure to nonantibiotic pharmaceuticals and antibiotics. Previous studies have reported a synergistic effect between the antidepressant sertraline with the antibiotic tetracycline against tetracyclineresistant E. coli, 10 as well as the synergistic interactions between the antibiotic polymyxin B and non-antibiotics, which provide alternative strategies for difficult-to-treat infections. 11 However, whether the combined exposure synergistically promotes the evolution of ARB remains unclear.

Depression has become one of the most prevalent disorders worldwide, impacting >16% of the world's population and leading to the massive use of antidepressant drugs. [12] [13] [14] In this study, duloxetine, the leading antidepressant pharmaceutical, and chloramphenicol were selected as a representative non-antibiotic pharmaceutical and antibiotic, respectively, to investigate their combined effects on antibiotic resistance in E. coli, an opportunistic pathogen in the gut microbiota. The mechanisms underlying their collaborative enhancement of resistance to antibiotics were revealed by genomic sequencing, together with transcriptomic analysis, and were further verified by real-time quantitative polymerase chain reaction (RT-qPCR) and clustered regularly interspaced palindromic repeat and Cas9 (CRISPR/Cas9) gene editing. 15 To the best of our knowledge, this work, for the first time, shows the synergistic induction of the resistance to multiple antibiotics by nonantibiotic pharmaceuticals and antibiotics, specifically at residual levels. Our findings will broaden our understanding of the emergence of antibiotic resistance and provide a novel theoretical basis for the dissemination of ARB.

To reveal the effects of duloxetine and/or chloramphenicol exposure on antibiotic resistance, E. coli was exposed to duloxetine at concentrations ranging from 2.5 µg/L to 25 mg/L and/or chloramphenicol at a sublethal concentration (6 µg/L-4 mg/L).

The results showed that chloramphenicol resistance occurred in E. coli exposed solely to chloramphenicol (Figure 1(a) ). Furthermore, they presented a dose-response pattern, and the mutation frequency declined sharply with the decrease in chloramphenicol exposure levels from 16 to 0.5 mg/L (p < .01). At the exposure level of 16 mg/L chloramphenicol for five days, the mutation frequencies of chloramphenicol resistance increased by at least 7.0 × 10 7 -fold to 0.41 (Table S1 ), compared with the spontaneous mutation frequency of chloramphenicol, which was less than 5.9 × 10 −9 (no spontaneous chloramphenicol-resistant mutants were observed on the plates containing 16 mg/L chloramphenicol). However, when the chloramphenicol dose was lowered to 1 mg/L, the mutation frequencies dramatically decreased to 2.6 × 10 −6 over the same period. Furthermore, the bacterial mutation frequency of chloramphenicol also showed a time-dependent pattern. While no increase in the mutation frequency was found on day 1 in case of 4 mg/L chloramphenicol, the mutation frequency of chloramphenicol increased to 1.3 × 10 -4 on day 3 (clones named C 4 -3d) and to 3.4 × 10 −3 on day 5. Evidently, the higher the exposure concentration, the faster was the increase in the mutation frequency over time. The resistant E. coli began to grow on chloramphenicol selecting-plates on day 1, day 3, and day 5 for for 8 mg/L, 4 mg/L, and 1 mg/L chloramphenicol exposure, respectively. No significant increase in the mutation frequency occurred even if E. coli was exposed to 6 μg/L chloramphenicol for 50 days.

For the first time, the antidepressant duloxetine was also found to induce chloramphenicol resistance in E. coli, showing dose-and timedependent patterns similar to those observed with chloramphenicol exposure (Figure 1(b) , Table S2 ). After one-day exposure to 100 mg/L duloxetine, E. coli obtained resistance to chloramphenicol with a mutation frequency increase of at least 3.0 × 10 3 -fold to 1.7 × 10 −5 (clones named D 100 -1d), while it reached at least a -1.3 × 10 5 -fold increase to 7.4 × 10 −4 over a fiveday exposure period. However, when the concentrations of duloxetine decreased to 50 mg/L, significant decreases in mutation frequency could be found compared with those exposed to 100 mg/L for the same duration (p < .05). Overall, the resistant E. coli started growing on chloramphenicol plates on day 1 in the presence of 100 mg/L duloxetine and day 3 in the presence of 80 mg/L duloxetine. However, when the duloxetine concentration was 2.5 μg/ L, no colonies were found on chloramphenicol plates even if the exposure time was prolonged to 50 days.

Considering that both the concentrations of antidepressants and antibiotics in the gut environment were lowered to the level of μg/L to mg/L, 7,16 the effects of combined exposure to duloxetine and chloramphenicol at gut environmental concentration on the mutation frequency of chloramphenicol-resistance was further tested. Surprisingly, after 30 days of combined exposure to 25 μg/L duloxetine and 60 μg/L chloramphenicol and 50 days of combined exposure to 2.5 μg/L duloxetine and 6 μg/ L chloramphenicol, resistant E. coli (clones named D 0.025 -C 0.06 -30d and D 0.0025 -C 0.006 -50d) started growing on chloramphenicol-resistance plates and the frequency of chloramphenicol resistance mutations increased by an average of at least 2.5-fold to 1.4 × 10 −8 , indicating that the combined exposure to duloxetine and chloramphenicol can synergistically promote bacterial resistance to chloramphenicol, compared with the exposure to duloxetine or chloramphenicol independently. To determine the combination index between duloxetine and chloramphenicol with regard to the mutation frequency of chloramphenicol, the effects of combined exposures to 1 or 4 mg/L chloramphenicol and duloxetine in the range of 0.025-25 mg/L were observed. Figure 1 significant difference occurred between the independent exposure to 25 mg/L duloxetine or 4 mg/L chloramphenicol with the spontaneous mutation frequency of chloramphenicol. The combination index of duloxetine and chloramphenicol was 4.5 ( Figure 1(d) ). Figure 1 (e) shows that the mutation frequency increased by at least 375-fold to 2.2 × 10 −6 on day 3 for E. coli treated with combined exposure to 25 mg/L duloxetine and 1 mg/L chloramphenicol, and the combination index reached 187.7 (Figure 1 (f)), which clearly shows that duloxetine and chloramphenicol could synergistically induce resistance to multiple antibiotics in E. coli.

The MICs of 29 types of representative antibiotics were tested for 5-8 isolates obtained from the above chloramphenicol plates that were treated with independent or combined exposure (Tables S4, S5-1 and S5-2). The result shows that the MIC of 12 antibiotics for the clone D 25 -C 4 -1d significantly increased by up to 32-fold, compared with that of wild-type E. coli (p < .05). Above all, the MICs of 6 antibiotics (ampicillin, cefotaxime, ceftazidime, cefazolin, cefoxitin, and doxycycline) were significantly higher for the clones D 25 -C 4 -1d than for the clones C 4 -3d and D 100 -1d (p < .05). The clones D 25 -C 4 -1d developed multiple antibiotic resistance to tetracycline, cefazolin, cefoxitin, ampicillin, and chloramphenicol. The clones C 4 -3d developed resistance to tetracycline, ampicillin, and chloramphenicol. The clones D 100 -1d developed resistance to cefoxitin and chloramphenicol. However, the clones D 0.025 -C 0.06 -30d and D 0.0025 -C 0.006 -50d only gained significantly higher MICs of 6 antibiotics (tetracycline, cefoxitin, doxycycline, chloramphenicol, minocycline and nalidixic acid) compared with the wild-type E. coli K12 (p < .05). The clones D 0.025 -C 0.06 -30d developed resistance to tetracycline and chloramphenicol. The clones D 0.0025 -C 0.006 -50d only developed resistance to chloramphenicol ( Figure 2 ). The MICs of antibiotics for all the isolates remained unchanged even after 10 passages through the lysogeny broth (LB) medium (Table S4 ), suggesting that antibioticresistant isolates displayed the genetic stability of antibiotic resistance, which would result in the vertical transfer of antibiotic-resistant genes (ARGs).

Global transcriptomic sequencing analyses were performed to study the mechanism underlying the synergistic enhancement of resistance to chloramphenicol by duloxetine and chloramphenicol in E. coli (Figure 3 and S1). For E. coli treated with combined exposure to 25 mg/L duloxetine and 4 mg/L chloramphenicol, multiple genes were significantly upregulated by more than 2-fold, including at least 26 genes related to bacterial antibiotic resistance, 23 genes related to DNA repair/replication, and nine genes related to oxidative stress, in comparison to E. coli treated independently with 25 mg/L duloxetine or 4 mg/L chloramphenicol. Each category of pathways are analyzed individually as follows to further reveal the mechanisms underlying the synergistic effects of duloxetine and chloramphenicol through their combined exposure.

According to the transcriptomic analysis, the genes related to bacterial antibiotic resistance that were upregulated after the combined exposure include the efflux pump system/transporter (e.g., acrA and acrB) and their regulatory genes (e.g., marR and marA), membrane pore protein genes (e.g., ompA and ompF), mobile element-associated genes (e.g., intD, intZ, and intJ), penicillin-resistance genes (prc and mrcA) etc. (Table S6 , Figure. S1). Notably, none of the genes showed significant changes in expression after the independent exposure to8 25 mg/L duloxetine, but 13 genes showed notably upregulated expression after the independent exposure to 4 mg/L chloramphenicol Among the efflux pump system and its regulatory genes, the expression levels of acrA and acrB increased by around 4-fold after the combined exposure to 25 mg/L duloxetine and 4 mg/L chloramphenicol. The periplasmic fusion protein AcrA and drug proton transporter AcrB, together with the outer membrane channel protein TolC, forms the active external pump AcrAB-TolC transport system of E. coli. AcrAB-TolC belongs to the resistancenodulation-cell division (RND) superfamily of the E. coli efflux pump system, which helps bacteria resist various antibiotics, such as fluoroquinolones, tetracyclines, and cephalosporins. [17] [18] [19] [20] Above all, the expression of the gene marA, which modulates the transcription of the acrAB operon, increased by 2.8-fold in the combined exposure group. In contrast, in the independent exposure group, the expressions levels of marA were not significantly increased. The genes mlaD, mlaE, and mlaF belonging to the MlaFEDB ABC transporter, which drives phospholipid trafficking across the bacterial envelope to maintain the outer membrane integrity, 21, 22 were also found to be significantly upregulated by 3.7-4.6-fold in the combined exposure group, according to the global transcriptomic sequencing analyses.

To verify the contribution of the efflux pump system to the synergy between duloxetine and chloramphenicol, the expression of key genes in the transcriptome (acrA, acrB, bamA, bamC, emrA, mlaD, mlaE, mlaF, marA, marR, mdtK, msbA, tsx, yadG, and ydhP) were tested using qPCR. The results demonstrated that the expression levels of acrA, acrB, and marA in E. coli treated with the combined exposure to 25 mg/L duloxetine and 4 mg/L chloramphenicol were upregulated 1.7-2.2-fold; these were significantly higher than those observed after the exposure to both agents independently (p < .05, Figure 4 , Table S7 ). Meanwhile, after validation with qPCR, the expression level of mlaF in the combined exposure group was found to be 1.3-fold higher (p < .05, Table S8 ); this was significantly higher than that observed after the exposure to both agents independently. Therefore, the enhanced expression of the AcrAB-TolC transport system-and MlaFEDB-associated genes in the combined exposure group may contribute to the synergistic promotion of antibiotic resistance in E. coli. To further validate the role of the AcrAB-TolC and MlaFEDB systems in the synergy between duloxetine and chloramphenicol, the genes belonging to these systems (marA, acrA, acrB, mlaD, malE, and mlaF) were knocked out in wild-type E. coli K12 using the CRISPR/Cas system, followed by the exposure to duloxetine and/or chloramphenicol (Table S8 ). The knockout of the marA gene rendered E. coli completely devoid of drug resistance mutations after exposure to duloxetine and/or chloramphenicol. However, E. coli still mutates under drug exposure after knocking out the genes acrA, acrB, mlaD, mlaE, or mlaF, suggesting that the AcrAB-TolC transport system and the MlaFEDB ABC transporter act together in the synergistic promotion of antibiotic resistance in E. coli.

The expression of antioxidant genes in response to oxidative stress was also analyzed by transcriptomic sequencing (Table S9, Figure. S1 ). For E. coli, the oxyR regulon (primarily associated with response to hydrogen peroxide (H 2 O 2 )) and soxRS regulon (primarily associated with response to superoxide (O 2 − )) defends cells against the damage caused by reactive oxygen species (ROS). 23, 24 The results show that not only did the expression of oxyR increase by 2.1-fold in the combined exposure group, but the expression levels of the downstream genes upregulated by the OxyR redox response system were also elevated: grxA (Glutaredoxin A; by 27.9-fold), ahpF (alkyl hydroperoxide reductase; by 8.6-fold), fur (ferric iron uptake regulon transcriptional repressor; by 2.3-fold), and trxB (thioredoxin reductase; by 2.8-fold). Overexpression of these genes conferred the abilities of peroxide metabolism, redox balance, peroxide protection, glutathione synthesis and reduction, DNA protection under oxidative stress, etc., onto the strains. 25 Therefore, the OxyR redox response system may play an important protective role in E. coli exposed to duloxetine and chloramphenicol. Additionally, the expression levels of the components of the rsxABCDGE gene cluster, which encodes a SoxR-reducing system to turn off the SoxRmediated induction of the activation of the transcription factor SoxS in the absence of oxidizing agents, 26 were elevated by 2.3-4.3-fold, indicating that soxS transcription may not be constitutively activated here. Taken together, there was no significant increase (n = 3) . The baseline conditions were: 100 µL of fresh overnight cultured wild-type E. coli K12 was inoculated into 10 mL LB medium containing 25 mg/L duloxetine or/and 4 mg/L chloramphenicol and cultured for 10 h at 37°C. Then, the relative concentration of interesting genes was assayed by RT-qPCR. As controls, bacteria without any exposure were also assayed and the fold difference of the expression levels of interesting genes between them were calculated.

in the expression levels of the aforementioned genes after the independent exposure to 25 mg/L duloxetine, and only the expression level of ahpF increased by 4-fold after the independent exposure to 4 mg/L chloramphenicol; correspondingly, combined exposure resulted in a more robust oxidative stress response in E. coli than the independent exposure.

To validate the oxidative stress response in E. coli treated with 25 mg/L duloxetine and/or 4 mg/L chloramphenicol, the ROS production and activities of the cellular antioxidant systems were further analyzed ( Figure 5 , Table S10 ). After the combined exposure to 25 mg/L duloxetine and 4 mg/L chloramphenicol, the ROS content increased by 2-fold, compared with that in the control group (p < .05). In comparison, after 25 mg/L duloxetine or 4 mg/L chloramphenicol treatment, there was no significant difference in the ROS content (p > .05). Meanwhile, to protect the bacteria from ROS-induced destruction, cellular antioxidant systems, including superoxide dismutase (SOD), catalase (CAT), glutathione peroxidase (GSH-PX), and total antioxidant capacity (T-AOC) were activated, correspondingly. After the combined exposure to 25 mg/L duloxetine and 4 mg/L chloramphenicol, the GSH-PX activity increased by 29.6-fold, the SOD activity, by 2.8-fold, and the T-AOC, by 7.7-fold, compared with those in the control group (p < .05, Table S10 ). Above all, ROS production, SOD activity, GSX-PH activity, CAT activity, and T-AOC of the combined exposure group were significantly higher than those of the two independent-exposure groups (p < .05). 25 , C 4 ) for 10 h at 37°C was centrifuged at 10,000 × g, 4°C for 1 min and then the collected pellets were suspended in 0.2 mol/L phosphate-buffered saline (PBS) and sonicated for three minutes under an ice-water bath. After centrifugation at 5,000 × g, 4°C for 3 min, the contents of superoxide dismutase (SOD), catalase (CAT), glutathione peroxidase (GSH-PX), and total antioxidant capacity (T-AOC) in the supernatant were assayed (a-d). The reactive oxygen species (ROS) was measured (e) using the DCF-DA/H 2 DCFDA-cellular ROS detection assay kit (Abcam, USA). As controls, bacteria without any exposure were also assayed. e2018901-8 transcriptomic sequencing (Figure 3 and S1 and Table S11), the exposed E. coli were revealed to show a dramatic increase in the lexA and recA expression levels by about 2.6-fold, indicating that combined exposure damaged E. coli DNA and activated its SOS repair system to repair DNA for survival. Meanwhile, the expression levels of genes related to mismatch repair proteins, e.g., mutL and mutS, and exonucleases, e.g., recX, recC, recJ, recF, xseA, and xthA, were upregulated by 2.3-2.6-fold and 2.1-2.8-fold, respectively; this may account for the increased mutation rate in the bacteria exposed to both drugs. Otherwise, the expression levels of none of these genes changed significantly in E. coli treated with 25 mg/L duloxetine exposure. The genes rne, obgE, ybfE, and ydjM were significantly upregulated by 2.6-4.6-fold in E. coli treated with 4 mg/L chloramphenicol exposure; the expression levels of these genes were lower than those in E. coli treated with the combined exposure.

To observe if the mutations occurred during the exposure, genome sequencing targeting the resistant isolates was performed. No chloramphenicol-resistant clones occurred in case of E. coli exposed to 25 mg/L duloxetine or 4 mg/L chloramphenicol for one day, 25 µg/L duloxetine or 60 µg/L chloramphenicol for 30 days, or 2.5 µg/L duloxetine or 6 µg/L chloramphenicol for 50 days; therefore, we sequenced the genome of the resistant isolates induced by their combined exposure. The findings revealed that all the isolates presented a mutation type in marR, which is a repressor of the marORAB operon (Table 1) . This includes base insertion or deletion leading to an early stop codon in marR or a substitution of a single nucleotide polymorphism in marR. Otherwise, no mutation occurred in the genes associated with DNA replication/repair or oxidative stress systems. E. coli K12 with marA knock out showed no occurrence of mutations on the chloramphenicol-containing plates after the combined exposure to 25 mg/L duloxetine and 4 mg/L chloramphenicol; this suggests that in this study, the mutation in marR enabled the development of antibiotic resistance.

To investigate the mechanisms underlying the chloramphenicol resistance of E. coli mutants isolated from combined or independent exposure, the E. coli mutants (D 100 -1d, C 4 -3d, and D 25 -C 4 -1d) were cultured in the LB broth containing 4 mg/L chloramphenicol for 10 h at 37°C and were then subjected to Figure 7 . Differential expression of genes in E. coli mutants when exposed to 4 mg/L Chloramphenicol. 100 µL of fresh overnight cultured wild-type E. coli (controls) or E. coli mutants D 25 -C 4 -1d, D 100 -1d, and C 4 -3d isolated from the combined exposure (25 mg/L duloxetine and 4 mg/L Chloramphenicol) and sole exposure to either duloxetine (100 mg/L) or chloramphenicol (4 mg/L) groups were inoculated into 10 mL LB containing 4 mg/L chloramphenicol for 10 h at 37°C. Then, the relative concentrations of interesting genes were assayed by RT-qPCR. The fold differences in the expression levels of interesting genes between various mutants and wild-type strains were calculated. transcriptomic analyses. According to the transcriptional profile analysis, all mutants presented differential expression of genes associated with the drug efflux system and stress defense when subjected to the 4 mg/ L chloramphenicol treatment (Tables S12 and S13 and Figure 6 ), but no significant differences were found between the expression levels of these genes in the E. coli mutants. This observation indicates that the mechanism underlying chloramphenicol resistance was the same for all the mutants, regardless of whether they were induced by duloxetine, chloramphenicol, or their combination.

Among the 31 ARGs that may contribute to the enhanced chloramphenicol resistance for mutants induced by the combined exposure, the genes associated with the multi-drug efflux system transporter MdtEF, which confers resistance to compounds such as erythromycin, doxorubicin, ethidium bromide, deoxycholate, and benzalkonium 27,28 were upregulated (mdtE by 8.6-fold and mdtF by 2.0-fold). The gene mdtM, which encodes a 410-amino acid residue protein that belongs to the large and ubiquitous major facilitator superfamily (MFS), 29 was upregulated by 12.1-fold. Above all, rob, marA, and crp, which transcriptionally activate the acrAB operon, 30 were upregulated by 2.1-13.0-fold. The gene acrZ, which encodes the small protein AcrZ in E. coli to interact with the transmembrane portion of the multi-drug efflux pump AcrB and increases the bacterium's resistance to a subset of the antibiotic substrates of that transporter, 31, 32 was upregulated by 7.5-fold. Figure 7 further validated (by qPCR) that the expression levels of acrA, acrB, acrZ, mdtE, and mdtF increased by 1.2-1.7-fold in the mutants D 25 -C 4 -1d. More importantly, the aforementioned mutants completely lost chloramphenicol resistance after the genes acrA, acrB, acrZ, mdtE, or mdtF were knocked out (Table S14) . Therefore, this study showed that the upregulation of the efflux pump-associated genes AcrAB-TolC and mdtEF play key roles in helping E. coli mutants resist chloramphenicol.

The overuse or misuse of antibiotics was thought to be the main driver inducing the development of antibiotic resistance in pathogenic bacteria. 1 However, with the findings that antidepressant fluoxetine can promote the resistance of E. coli to multiple antibiotics, 33 ranking potential risk factors that may enable the development of antibiotic resistance in the gut microbiota, which is becoming a worldwide concern, is urgently required. In this study we describe that the combined exposure to duloxetine and chloramphenicol synergistically promote the development of multi-antibiotic resistance in E. coli. In a sense, this suggests that the combined exposure to antibiotics and non-antibiotic pharmaceutics in the gut may synergistically accelerate the evolution of ARB, resulting in the exacerbation of the antibiotic resistance crisis in humans.

Generally, gut micribiota may be subjected to short-term high-dose exposure at the level of mg/ L due to clinical drug use or long-term low-dose exposure to antibiotics and depressants at the level of μg/L due to residual drugs in the food chain or drinking water. 5, 8, 16, 34 Maier et al. 35 deduced the concentrations of drugs in the colon on the basis of the drug excretion patterns from published work and those in the small intestine on the basis of the daily doses of individual drugs consumed. Their findings showed that the media estimated in the small intestine and colon concentrations were higher than 20 μM for both active and inactive compounds, indicating that the simultaneous use of duloxetine and chloramphenicol in clinical trials would lead to a short-term combined exposure to duloxetine and chloramphenicol, at the level of mg/ L, to the gut microbiota. In addition, findings from previous studies, which showed that the antibiotic concentration in the meconium reached up to over 150 ng/g (including chloramphenicol), 36 suggested that the cumulative exposure to low-dose antibiotics was common during pregnancy. A clinical trial found that the antibiotic concentrations in the guts of healthy individuals were 26.51 ± 33.48 ng/g. 7 Therefore, gut microbiota may have been exposed to both residual non-antibiotic pharmaceuticals and antibiotics, at the level of μg/L, for a long time. Base on these studies we believe that E. coli can develop multiple antibiotic resistance after being to duloxetine and chloramphenicol not only in vitro, but also in vivo.

Previous studies found that the antidepressant sertraline has antimicrobial abilities and exhibits synergistic effects in enhancing the activity of multiple antibiotics through the inhibition of efflux pumps. 37, 38 However, in this study, we found that the enhanced expression of the efflux pumps, including the AcrAB-TolC transport system and the MlaFEDB ABC transporter together via the ROSinduced mutagenesis of marR, contribute to synergistic antibiotic resistance. In the multi-drug resistance operon (Mar), which is involved in regulating intrinsic susceptibility to structurally unrelated antibiotics, MarR binds to the marRAB promoter and negatively regulates gene expression. 19 The overexpression of marA results in the multiple antibiotic resistance phenotype. Transcription of the acrAB operon, whose expression is modulated by global stress signals, 30 is elevated in strains containing marR mutations. 39 Therefore, the mutation in marR here may result in the loss of its DNA-binding ability, 40 leading to the increased transcription of the marRAB and acrAB operons via MarA activation, resulting in bacterial resistance to multiple antibiotics. In addition, combined exposure to duloxetine and chloramphenicol resulted in a more robust oxidative stress response in E. coli than in the independent study. Above all, the combined group showed a two-fold increase in the ROS content, while the ROS content in the independent exposure group showed no significant difference compared with that in the control group. Considering that oxidative stress in bacteria may cause serious oxidative damage in DNA through nonspecific rapid reactions, 41 marR mutations herein can be induced by ROS produced in the combined exposure group. Previously, Jin et al. 33 demonstrated that the exposure to the antidepressant fluoxetine induces multiple antibiotic resistance in E. coli via the ROS-mediated mutagenesis of DNA-binding transcriptional regulators, which enhances the antibiotic efflux. MlaF is also required for Acinetobacter baumannii survival in the presence of sub-inhibitory concentrations of colistin, 42 and deleting mlaF makes A. baumannii more susceptible to antibiotics. 43 The dominant mechanisms underlying the resistance to chloramphenicol in bacteria are enzymatic inactivation by acetylation, clearance via efflux pumps, and ribosome protection. [44] [45] [46] [47] However, in this study, no transcriptional enchancement of chloramphenicol acetyltransferase or ribosome protection were found in the mutants. Interestingly, in all mutants, regardless of whether chloramphenicol resistance was induced by duloxetine, chloramphenicol, or their combination, the mechanism underlying the resistance against chloramphenicol was the same, i.e., the upregulation of the efflux pumps AcrAB-TolC and mdtEF. Therefore, herein, the enhanced antibiotic efflux pumps, and not chloramphenicol acetyltransferase or ribosome protection, contributes to the resistance of E. coli against chloramphenicol. Multidrug efflux pumps play important roles in the multiple antibiotic resistance of E. coli, indicating that it may serve an efficient target to control infections caused by ARB using drug efflux inhibitors.

In conclusion, our findings demonstrated that the combined exposure to duloxetine and chloramphenicol can synergistically promote the development of multi-antibiotic resistance in E. coli, indicating that antibiotics and non-antibiotic pharmaceutics can synergistically accelerate the evolution of resistance to multiple antibiotics in bacteria in vitro. Moreover, the enhancement of the activities of the AcrAB-TolC transport system and the MlaFEDB ABC transporter via ROS-induced mutagenesis contribute to the synergistic increase in antibiotic resistance in the combined exposure group; further, the improved resistance of the resulting mutants against chloramphenicol involved the improved activities of the efflux e2018901-12 pump AcrAB-TolC and mdtEF. These results broaden our understanding of the emergence of antibiotic resistance in the gut and provides a theoretical basis for ARB dissemination in nature.

E. coli strain K12 (MG 1655) purchased from the American Type Culture Collection (ATCC 15597) was grown in lysogeny broth (LB) (10 g/L tryptone (Difco), 5 g/L yeast extract (Difco), 10 g/L NaCl) at 37°C. E. coli DH5α competent cells purchased from TIANGEN BIOTECH (Beijing, China) were cultured in LB at 37°C. E. coli with pRedCas9 plasmid was cultured in LB at 30°C. Table S15 lists the bacteria strains and plasmids used in this study. Table S16 lists the primers used in this study.

After E. coli strain K12 (MG 1655) was incubated at 37°C for 12 h, 30 µL of fresh cell suspension was added to 2.97 mL LB broth containing duloxetine (2.5 µg/L-100 mg/L) and/or chloramphenicol (6 µg/L-16 mg/L) to reach the final cell density of 10 6 -10 7 CFU/mL. The E. coli cultures were incubated at 37°C for 24 h. Then 30 µL of the cultures were transferred into 2.97 mL fresh LB containing the corresponding concentration of duloxetine and/or chloramphenicol. This subculture process was repeated for up to 50 days. During this process, 100 µL of E. coli cultures were spread on M-endo agar containing 16 mg/L chloramphenicol to screen for chloramphenicol-resistant bacteria, and antibiotic-free LB agar, which was used for the total bacterial count. This procedure was performed on Days 1, 3, 5, 10, 20, 30, 40, and 50 and the plates were incubated for 24 h at 37°C. Then, 5-8 chloramphenicol-resistant mutants from each batch were selected at random and stored at −20°C. As controls, E. coli cultured in the LB broth without duloxetine or chloramphenicol were also transferred for subculture and then were spread on M-endo agar containing 16 mg/L chloramphenicol to observe spontaneous mutation frequency. All experiments were performed in biological triplicates.

The mutation frequency against chloramphenicol was calculated by dividing the number of chloramphenicol-resistant colonies by the total bacterial count using Equation 1 .

Where N 0 refers to the number of chloramphenicolresistant bacteria and N refers to total bacteria count.

To evaluate the combined effects of duloxetine and chloramphenicol, the combination index was calculated using the modified Burgi formula (i.e., Jin equation) 48 here written as Equation 2, which is universally accepted in pharmacology.

where f C , f D , and f C+D , refer to the mutation frequency of E. coli independently exposed to chloramphenicol, duloxetine, and their combined exposure, respectively. The expected value q represents the combination index, 0.85 < q < 1.15 would mean addition, q > 1.15 means synergism, and q < 0.85 means antagonism.

The MICs of the resistant mutants E. coli against 29 different antibiotics were determined using the Thermo Scientific TM Sensititre TM Susceptibility Testing System (Thermo Fisher Scientific, USA) according to the manufacturer's instructions. Supplementary Text 1 introduced the detailed procedures.

To test the hereditary stability of the resistant mutant E. coli, the resistant mutants were cultured overnight in LB then inoculated into fresh 1% (V/V) LB for continuous overnight culture. After ten days of repeated subculture, the antibiotic susceptibility of the cells was determined as described above. 

RNA sequencing and differential expression analysis were performed in accordance with previous methods. 49 Gene expression level was quantified using HTSeq (Version 0.6.1). Differential expression analysis of the groups was performed using the DESeq R package (1.18.0). The resultant P-values were adjusted using Benjamini and Hochberg's approach for controlling the false discovery rate. Genes with significant fold differences (log 2 > 1 or log 2 < −1) and adjusted P-value <0.05 found by DESeq were assigned as differentially expressed. The data visualization tool Rstudio (Version 1.2.1335) was used for comprehensive comparison and graphical drawing.

Genomic alignment and mutation analysis were performed according to the previous method 50 with reference to the genome of E. coli K12 strain (Genbank accession number, U00096.3). Brimmy Trimmomatic (version 0.36) was used to trim Illumina's paired-end raw data, leaving only the correctly paired readings for further downstream analysis. To study small variations, these high read depth data sets were aligned to the reference sequence using BWA (Version: 0.7.8) 51 and SAMTOOLS (Version: 0.1.18), and then the highly divergent loci were further analyzed. The data visualization tools Rstudio (Version 1.2.1335) and Circos (Version 0.64) were used for comprehensive comparison and graphical drawing.

Relative gene expression in bacteria was quantified by RT-qPCR using One-step TB green RT-PCR kits (Takara, USA). Briefly, 2 μL of RNA samples from bacteria and 0.4 μL of each primer (0.2 μM) were added to a 17.2 μL reaction mixture containing 10 μL 2 × One-step TB Green RT-PCR BufferI III, 0.4 μL PrimeScript RT enzyme Mix II, 0.4 μLTaKaRa Ex TaqHS (5 U/ μL), 0.4 μL ROX Reference Dye II (50×), and 6 μL RNase-free water. The reaction conditions were 42°C for 5 min, 95°C for 10 s, followed by 40 cycles of 95°C for 5 s and 60°C for 34 s. The reactions were performed in an ABI Viia7 sequence detection system (Applied Biosystems, USA) in triplicate. The E. coli 16s RNA genes were used as a housekeeping gene. The fold difference of the genes was calculated using the comparative ∆CT method (Equation 3). 52 Table  S16 lists the primers used in this study. Where F represents the fold difference, CT (GOI.sample) and CT (16s RNA.sample) represent the CT value of the gene of interest and of the 16s RNA genes in the sample group, respectively, and CT (GOI.control) and CT (16s RNA.control) represent the CT value of the gene of interest and 16s RNA genes in the control group, respectively.

Genes marR, marA, AcrA, acrA, acrB, mlaD, mlaE were knocked out from wild-type E. coli K12 to verify their contribution to antibiotic resistance development during exposure to duloxetine and/or chloramphenicol. Genes acrA, acrB, mdtE, mdtF, acrZ were knocked out form E. coli mutants selected form drug exposure to verify their resistance mechanisms against chloramphenicol. Supplementary Text 5 describes the process of gene knockout in detail.

Bacterial ROS were quantified by flow cytometry (BD FACS Calibur, USA) using the DCF-DA/H 2 DCFDAcellular ROS detection assay kit (Abcam, USA). The activity of superoxide dismutase (SOD), catalase (CAT), glutathione peroxidase (GSH-PX), and total antioxidant capacity (T-AOC) in bacteria were quantified in SpectraMax M5 (Molecular Devices, USA) using the corresponding detection kits from Nanjing Jiancheng Lnc, China. They were all performed following the manufacturer's instructions; Supplementary Texts 2 and 3 introduce detailed procedures. All measurements were performed in biological triplicate.

The Bacterial Challange: time to React

Antimicrobial resistance: tackling a crisis for the health and wealth of nations. London: The Review on Antimicrobial Resistance

Antibiotic resistome in a large-scale healthy human gut microbiota deciphered by metagenomic and network analyses

Antibiotics and the gut microbiota

Gut bacteria of animals/pests living in polluted environments are a potential source of antibacterials

Gut bacteria of animals living in polluted environments exhibit broad-spectrum antibacterial activities

Gut resistomes, microbiota and antibiotic residues in Chinese patients undergoing antibiotic administration and healthy individuals

Occurrence and distribution of clinical and veterinary antibiotics in the faeces of a Chinese population

Antidepressant fluoxetine induces multiple antibiotics resistance in Escherichia coli via ROS-mediated mutagenesis

Insight into synergetic mechanisms of tetracycline and the selective serotonin reuptake inhibitor, sertraline, in a tetracycline-resistant strain of Escherichia coli

An alternative strategy for combination therapy: interactions between polymyxin B and non-antibiotics

The role of gut mucins in the etiology of depression

The neurobiology of depression: an integrated view

Global prevalence and burden of depressive and anxiety disorders in 204 countries and territories in 2020 due to the COVID-19 pandemic

CRISPR/Cas9 in genome editing and beyond

Environmental concentrations of the selective serotonin reuptake inhibitor fluoxetine impact specific behaviors involved in reproduction, feeding and predator avoidance in the fish Pimephales promelas

Cross-resistance to fluoroquinolones in multiple-antibiotic-resistant (Mar) Escherichia coli selected by tetracycline or chloramphenicol: decreased drug accumulation associated with membrane changes in addition to OmpF reduction

Multiple antibiotic resistance (mar) locus protects Escherichia coli from rapid cell killing by fluoroquinolones

Genetic and functional analysis of the multiple antibiotic resistance (mar) locus in Escherichia coli

The challenge of efflux-mediated antibiotic resistance in Gram-negative bacteria

Structural insights into outer membrane asymmetry maintenance in Gram-negative bacteria by MlaFEDB

Structural mechanism of phospholipids translocation by MlaFEDB complex

Transcription factors that defend bacteria against reactive oxygen species

The molecular mechanisms and physiological consequences of oxidative stress: lessons from a model bacterium

Regulators of oxidative stress response genes in Escherichia coli and their functional conservation in bacteria

A reducing system of the superoxide sensor SoxR in Escherichia coli

Analysis of a Complete Library of Putative Drug Transporter Genes in Escherichia coli

Exploring the multi-drug resistance in Escherichia coli O157:H7 by gene interaction network: a systems biology approach

Functional and biochemical characterisation of the Escherichia coli major facilitator superfamily multidrug transporter MdtM

Role of the acrAB locus in organic solvent tolerance mediated by expression of marA, soxS, or robA in Escherichia coli

Interactions of a bacterial RND transporter with a transmembrane small protein in a lipid environment

Conserved small protein associates with the multidrug efflux pump AcrB and differentially affects antibiotic resistance

Antidepressant fluoxetine induces multiple antibiotics resistance in Escherichia coli via ROS-mediated mutagenesis

Antibiotic pollution in surface fresh waters: occurrence and effects

Extensive impact of non-antibiotic drugs on human gut bacteria

Determination of antibiotic concentration in meconium and its association with fetal growth and development

Efflux inhibition by selective serotonin reuptake inhibitors in Escherichia coli

A review of the antimicrobial side of antidepressants and its putative implications on the gut microbiome

AcrAB efflux pump plays a major role in the antibiotic resistance phenotype of Escherichia coli multiple-antibiotic-resistance (Mar) mutants

Mutational analysis of the multiple-antibiotic resistance regulator MarR reveals a ligand binding pocket at the interface between the dimerization and DNA binding domains

Role of reactive oxygen species (ROS) in therapeutics and drug resistance in cancer and bacteria

Clinical and laboratory-induced colistin-resistance mechanisms in Acinetobacter baumannii

Modulating isoprenoid biosynthesis increases lipooligosaccharides and restores Acinetobacter baumannii resistance to host and antibiotic stress

Whole transcriptome analysis and gene deletion to understand the chloramphenicol resistance mechanism and develop a screening method for homologous recombination in Myxococcus xanthus

A novel family of intrinsic chloramphenicol acetyltransferase CATC in Vibrio parahaemolyticus: naturally occurring variants reveal diverse resistance levels against chloramphenicol

Responses of Wild-Type and Resistant Strains of the Hyperthermophilic Bacterium Thermotoga maritima to Chloramphenicol Challenge

Chloramphenicol derivatives as antibacterial and anticancer agents: historic problems and current solutions

About the evaluation of drug combination

Copper Oxide Nanoparticles Induce Lysogenic Bacteriophage and Metal-Resistance Genes in Pseudomonas aeruginosa PAO1

Water disinfection byproducts induce antibiotic resistance-role of environmental pollutants in resistance phenomena

Fast and accurate short read alignment with Burrows-Wheeler transform

Analysis of relative ge ne expression data using real-time quantitative PCR and the 2(-Delta Delta C(T)) Method

This work is supported by the grant from the National Natural Science Foundation of China (Grant No. 81972994). The CRISPR/Cas9 gene edit system was kindly donated by Prof. Tao Chen of Tianjin University.

The authors declare no competing interests

No potential conflict of interest was reported by the author(s).

This work was supported by the National Natural Science Foundation of China [81972994].

DNA and RNA sequencing data are accessible through the Sequence Read Archive (SRA accession number PRJNA698263, PRJNA725495).

Statistical analyses were performed using OriginPro 2018C (OriginLab Corporation, USA), SPSS 20.0 (IBM Corporation, USA), and Rstudio (Version 1.2.1335). Student's T-test was used to compare the difference in mutation frequency, ROS, antioxidant levels, MICs, and mutant gene levels between treatment groups. For all tests, only data resulting in P values < .05 were regarded as statistically significant.