RamA, a transcriptional regulator conferring florfenicol resistance in Leclercia adecarboxylata R25

Due to the inappropriate use of florfenicol in agricultural practice, florfenicol resistance has become increasingly serious. In this work, we studied the novel florfenicol resistance mechanism of an animal-derived Leclercia adecarboxylata strain R25 with high-level florfenicol resistance. A random genomic DNA library was constructed to screen the novel florfenicol resistance gene. Gene cloning, gene knockout, and complementation combined with the minimum inhibitory concentration (MIC) detection were conducted to determine the function of the resistance-related gene. Sequencing and bioinformatics methods were applied to analyze the structure of the resistance gene-related sequences. Finally, we obtained a regulatory gene of an RND (resistance-nodulation-cell division) system, ramA, that confers resistance to florfenicol and other antibiotics. The ramA-deleted variant (LA-R25ΔramA) decreased the level of resistance against florfenicol and several other antibiotics, while a ramA-complemented strain (pUCP24-prom-ramA/LA-R25ΔramA) restored the drug resistance. The whole-genome sequencing revealed that there were five RND efflux pump genes (mdtABC, acrAB, acrD, acrEF, and acrAB-like) encoded over the chromosome, and ramA located upstream of the acrAB-like genes. The results of this work suggest that ramA confers resistance to florfenicol and other structurally unrelated antibiotics, presumably by regulating the RND efflux pump genes in L. adecarboxylata R25. Electronic supplementary material The online version of this article (10.1007/s12223-020-00816-2) contains supplementary material, which is available to authorized users.


Introduction
Florfenicol, a derivative of chloramphenicol with better antibacterial activity and fewer adverse effects, has been widely used in veterinary medicine (Schwarz and Chaslus-Dancla 2001;Schwarz et al. 2004). However, the resistance levels and the number of resistant bacteria to florfenicol have increased due to the ever-increasing use of florfenicol in agricultural practice (Chang et al. 2015;Geng et al. 2012;Sun et al. 2016). To date, more than ten florfenicol resistance genes have been reported. These genes belong to four molecular categories: the major facilitator superfamily (MFS, including floR, floRv, flost, fexA, fexB, and pexA) (Alessiani et al. 2014;Braibant et al. 2005;He et al. 2015;Kehrenberg and Schwarz 2004;Lang et al. 2010;Liu et al. 2012); the rRNA methyltransferase family [cfr, cfr(B), and cfr(C)] (Hansen and Vester 2015;Schwarz et al. 2000;Tang et al. 2017); the ATP-binding cassette (ABC) family (optrA) (Wang et al. 2015b); and a chloramphenicol acetate esterase-encoding gene, estDL136 (Tao et al. 2012). However, no efflux pump from RND family-related genes has been reported to be associated with florfenicol resistance. Cong Cheng and Yuanyuan Ying contributed equally to this work.
The ramA global regulator belongs to the AraC/XylS family. Transcriptional regulators of the AraC/XylS family have been associated with multidrug resistance, organic solvent tolerance, oxidative stress, and virulence in Enterobacteriaceae (Gallegos et al. 1997). The ramA gene was first described in a multidrug-resistant (MDR) Klebsiella pneumoniae (George et al. 1995). Later, it was also discovered in Enterobacter aerogenes, Enterobacter cloacae, and Salmonella enterica (Bailey et al. 2008;Chollet et al. 2004;Keeney et al. 2007). Surprisingly, no ramA was identified in Escherichia coli. Previous studies showed that overexpression of the ramA gene in E. coli conferred decreased susceptibility to diverse antibiotics such as chloramphenicol, tetracycline, tigecycline, fluoroquinolones, and trimethoprim (Chollet et al. 2004;George et al. 1995). Concrete resistance mechanisms have been documented in E. cloacae and S. enterica serovar Typhimurium, where ramA-mediated overexpression of efflux pumps, primarily the AcrAB-TolC efflux pump, leads to increased tigecycline and ciprofloxacin resistance (Keeney et al. 2007;Sun et al. 2011). However, there are no available data regarding the role of the regulator gene ramA, which may potentially regulate the expression of other RND efflux pumps for florfenicol resistance.
Leclercia adecarboxylata, a motile, facultative anaerobic, Gram-negative bacillus of the Enterobacteriaceae family, is an opportunistic human pathogen that is normally present in environmental or animal sources (Stock et al. 2004;Yao et al. 2011). In this work, we used the random cloning approach to investigate the unidentified florfenicol resistance mechanism of a L. adecarboxylata strain R25 with the high florfenicol resistance level isolated from a rabbit anal feces sample.

Material and methods
The bacteria and plasmids L. adecarboxylata R25 was isolated from an anal fecal sample of a rabbit during a survey of florfenicol-resistant bacteria from animal farms in 2015 in Wenzhou, Zhejiang Province, China, and it has been deposited in China Center for Type Culture Collection (CCTCC), Wuhan, China (CCTCC AB 2020046). The anal fecal samples were directly streaked on Luria-Bertani (LB) agar supplemented with 16 μg/mL florfenicol and cultured overnight at 37°C. Bacterial identification was performed using a Vitek-60 microorganism autoanalysis system (BioMerieux Corporate, Craponne, France). The strain was further confirmed by the analysis of its 16S rRNA gene and a whole-genome sequence comparison of L. adecarboxylata R25 against the nucleotide database at NCBI (https://blast.ncbi.nlm.nih.gov/Blast.cgi). The bacterial strains and plasmids used in this study are listed in Table 1.

Whole-genome sequencing
Bacterial genomic DNA was extracted using an AxyPrep Bacterial Genomic DNA Miniprep Kit (Axygen Scientific, Union City, CA, USA). A 20-kb library was sequenced by PacBio RS II platform (Pacific Biosciences, Menlo Park, CA) and an Illumina library with 300-bp insert sizes sequenced from both ends was obtained with a HiSeq 2500 platform (both PacBio RS II and HiSeq 2500 sequencing were finished at Annoroad Gene Technology Co., Ltd, Beijing, China). The PacBio long reads were assembled using the Canu software (Koren et al. 2017) and the assembly quality was corrected with the Illumina short reads. Potential ORFs were predicted using Glimmer software (http://ccb.jhu.edu/software/ glimmer) and annotated against a nonredundant protein database using the BLASTX program (https://blast.ncbi. nlm.nih.gov). The 16S rRNA sequences were annotated by the online tool RNAmmer (http://www.cbs.dtu.dk/ services/RNAmmer/).

Random genomic DNA library construction and screening for potential florfenicol resistance genes
The genomic DNA of L. adecarboxylata R25 was extracted and partially digested with the Sau3AI enzyme. Fragments approximately 2 to 5 kb in size were retrieved from the gel and ligated into a pUC118 vector digested with BamHI. The ligated sample was transformed into competent E. coli DH5α cells, and the transformants containing the cloned fragments were then selected on LB agar plates containing ampicillin (100 μg/mL) and florfenicol (32 μg/mL). Plasmid DNA from the positive transformant was purified, and the recombinant plasmid was subjected to PCR for the detection of the known florfenicol resistance gene (floR) of L. adecarboxylata R25 (floR-F: 5′-ATGACCACCACACGCCCCGC-3′ and floR-R: 5′-TTAGACGACTGGCGACTTCT-3′). The inserted fragments of recombinants without a known florfenicol resistance gene were sequenced, and the ORF of the potential resistance gene was obtained.

Cloning of candidate resistance genes
Primers with restriction endonuclease adapters at both ends were used to clone the candidate gene with its potential upstream promoter region (Table 2). PrimeStarHS DNA Polymerase (TaKaRa, Dalian, China) was used to amplify the potential resistance genes according to the manufacturer's instructions. The purified PCR product (prom-ORF) was digested with its corresponding restriction endonucleases and cloned into a pUCP24 vector that had been treated with the same restriction endonucleases. The resulting recombinant plasmid (pUCP24-prom-ORF) was transformed into E. coli DH5α using the calcium chloride method. The transformant was selected on LB agar plates containing 20 μg/mL gentamicin. The cloned PCR product was further confirmed by Sanger sequencing (ABI 3730 Analyzer, Foster City, CA, USA).

Construction of ramA-knockout and ramA-complemented strains
The inactivation of ramA in the wild-type strain L. adecarboxylata R25 was performed according to the  (Datsenko and Wanner 2000). Briefly, a kanamycin resistance gene (aph) flanked by FLP recognition target (FRT) sites in pKD4 was amplified by PCR using ramA-knockout-F/R primers and the template plasmid pKD4 under standard conditions. The ramA-knockout-F/R primers (Table 2) consisted of 20 nucleotides (nt) of the helper plasmid pKD4 and 50 nt of the 5′ and 3′ ends of the corresponding inactivated gene (ramA). The purified PCR fragment was digested with DpnI, purified, and transformed into L. adecarboxylata R25 by electroporation in the presence of pKD46 (carrying the Red recombinase gene). The mutant strain (LA-R25ΔramA-aph) was verified by PCR using the ramA-inner-F/R primers. The aph gene was further excised from LA-R25ΔramA-aph by the plasmid pCP20, which encodes FLP nuclease, introduced via transformation. Finally, the ramA-deleted variant was obtained and named LA-R25ΔramA. The recombinant plasmid with ramA and its upstream predicted promoter region (pUCP24-prom-ramA) was transformed into the ramA-deleted strain (LA-R25ΔramA), and the transformant (pUCP24-prom-ramA/LA-R25ΔramA) was selected on LB agar containing 20 μg/mL gentamicin. The plasmid (pUCP24-prom-ramA) in the transformant (pUCP24-prom-ramA/LA-R25ΔramA) was confirmed by PCR with primers targeting ramA and further sequenced by Sanger sequencing.

Antibiotic susceptibility assay
The MICs of the antimicrobial agents against L. adecarboxylata R25 and other strains, including LA-R25ΔramA, pUCP24-prom-ramA/LA-R25ΔramA, and pUCP24-prom-ramA/E. coli DH5α, were determined by the standard agar dilution method recommended by CLSI-2017 (the Clinical and Laboratory Standards Institute in 2017). The measurements of MICs in the presence of the carbonyl cyanide m-chlorophenylhydrazone (CCCP) were also carried out by standard agar dilution method containing serial dilutions of six antibiotics (florfenicol, tetracycline, erythromycin, nalidixic acid, clarithromycin, and levofloxacin) in 18 mL of Mueller-Hinton broth (MHB), followed by the addition of 2 ml of CCCP (100 μg/mL in dimethyl sulfoxide [DMSO]) to give a final concentration of 10 μg/mL. The MIC was recognized as the lowest antibiotic concentration showing no colony growth. Each of the tests was carried out in triplicate. E. coli ATCC 25922 was used as a quality control strain.
qRT-PCR analysis of ramA mRNA concentration in L. adecarboxylata R25 and the recombinant strains  Total RNAs were extracted from the wild, mutant, and recombinant strains cultured in LB with or without florfenicol using the Trizol. cDNA was obtained by reverse transcription using the PrimeScript RT-PCR Kit (TaKaRa, Dalian, China). After qPCR analysis, relative quantification of the target in each sample was calculated using rpsL as the internal control (Mikhail et al. 2019).

Bioinformatic analysis of the genetic environment of ramA
Sequences containing the ramA gene were obtained from the NCBI nucleotide database by the BLAST program using an approximately 10-kb fragment (including the sequences of the upstream 1-kb region and the downstream 9-kb region of the ramA gene) of the L. adecarboxylata R25 genome sequence as the query. A total of 100 complete bacterial genome sequences sharing the greatest sequence identity with the 10-kb fragment of the L. adecarboxylata R25 genome were retrieved (until March 15th, 2018). The sequences were filtered to accept only those encoding at least one of the four genes (tetR, ramA, and acrA-, and acrB-like) with an amino acid identity ≥ 50%. The sequences were clustered into different groups according to the number and order of the four homologous genes. A representative sequence in each group was chosen as the candidate for structural comparative analysis.
A randomly cloned ramA gene in E. coli conferring florfenicol resistance A transformant with florfenicol and chloramphenicol resistance and free of the known florfenicol resistance gene was obtained from random cloning of R25 genomic DNA. The 2.8-kb insert  fragment encoded three ORFs, including an MBL-fold metallohydrolase (mbl), a RamA family antibiotic efflux transcriptional regulator (ramA), and an outer membrane protease (omp), which shared 96, 100, and 54% amino acid identity with the top hits in the UniProtKB database, A0A2S4X220, A0A2T3CIN1, and A0A078LS26, respectively, of which the gene of RamA (A0A2T3CIN1) was from Enterobacter sp. FS01 isolated from parkland soil, Aarhus, Denmark. The minimum inhibitory concentration (MIC) results for the cloned genes with the predicted promotor regions (pUCP24-prom-ramA/ DH5α, pUCP24-prom-omp/DH5α, pUCP24-prom-mbl/ DH5α) showed that only the recombinant with ramA (pUCP24-prom-ramA/DH5α) was responsible for multidrug resistance, including resistance to florfenicol and chloramphenicol, and had the same MIC levels as the original transformant with the 2.8-kb insert fragment. The other two recombinants (pUCP24-prom-omp/DH5α and pUCP24-prom-mbl/DH5α) did not show any resistance to the antibiotics (Table 3).
Prominent functions of ramA in mediating antimicrobial resistance in L. adecarboxylata R25 and its derivatives The recombinant strain pUCP24-prom-ramA/DH5α exhibited increased MIC levels against florfenicol (MIC increased from 64 to 1024 μg/mL, 8-fold), chloramphenicol (8-fold), linezolid (> 4-fold), tetracycline (8-fold), erythromycin (4-fold), tigecycline (8-fold), nalidixic acid (4-fold), and lincomycin (> 2-fold) relative to recipient E. coli DH5α or E. coli DH5α carrying the vector pUCP24 (pUCP24/DH5α) (Table 3). No differences in rifampicin, ampicillin, streptomycin, amikacin, and kanamycin resistance were observed among these strains. LA-R25ΔramA showed 2-fold decreased resistance levels against florfenicol and chloramphenicol relative to the parental strain L. adecarboxylata R25. After complementation with ramA (pUCP24-prom-ramA), the resistance levels of the recombinant (pUCP24-pro-ramA/LA-R25ΔramA) to florfenicol and chloramphenicol were fully recovered, with even higher resistance levels than the wild-type L. adecarboxylata R25 (MIC increased from 64 to 1024 μg/mL, 16-fold). Similar results were also found for the other antibiotics (Table 3). To further evaluate the role of ramA in the resistance of the strains to florfenicol, the transcription levels of ramA and its potential targets in L. adecarboxylata R25 were determined. It revealed that the wild strain L. adecarboxylata R25 cultured with florfenicol showed significantly higher transcription level of ramA than the one cultured without it. The ramAcomplemented LA-R25ΔramA (pUCP24-pro-ramA/LA-R25ΔramA) which showed much higher MIC level to florfenicol (1024 μg/mL) than that (128 μg/mL) of the wild strain L. adecarboxylata R25 also showed significantly higher transcription level of ramA than that of the wild strain L. adecarboxylata R25 (Table 4). The expression levels of the potential target genes of ramA lowered in the ramA-deleted L. adecarboxylata R25 (LA-R25ΔramA) and recovered in ramA-complemented strain (pUCP24-pro-ramA/LA-R25ΔramA). They had (much) higher expression levels when the wild strain L. adecarboxylata R25 was treated with florfenicol (Table 4). It seemed that they were all regulated by ramA even though at different degrees. To confirm if the florfenicol resistance of L. adecarboxylata R25 was related with the efflux pumps, CCCP was used as an inhibitory agent for the efflux pumps. CCCP itself exhibited an antimicrobial effect on L. adecarboxylata R25 and its derivatives with MICs of 20 μg/mL. We tested the MIC levels of the a n t i b i o t i c s c o m b i n e d w i t h C C C P a t a l o w e r concentration (10 μg/mL). It showed that CCCP increased the sensitivity of L. adecarboxylata R25 and LA-R25ΔramA against florfenicol, tetracycline, and levofloxacin to 4-to 8-fold, but it did not show any effect on erythromycin, clarithromycin, and nalidixic acid (Table 5).  (2) GC skew (G − C/G + C), with a positive GC skew toward the outside and a negative GC skew toward the inside.
(3) GC content, with an average of 50%, and a G + C content of more than 50% is shown toward the outside; otherwise, inward. (4) Genes encoded in the leading strand (outwards) or the lagging strand (inwards). Genes with different functions are shown in different colors, and five groups of RND efflux pump-related elements are in red. Structural comparison of the genetic environment of ramA and five RND systems encoded on the L. adecarboxylata R25 chromosome Based on their similarity to the four ORFs (encoded by ramA and the other three genes in its neighboring region) in L. adecarboxylata R25, 96 sequences were chosen for further homology analysis. All were from chromosomes of species belonging to the family Enterobacteriaceae and were clustered into four groups. Four representative sequences from each group are illustrated in Fig. 1 and Table S1. Structural analysis of the four representatives showed that three of them contained a TetR family transcriptional regulator, which was absent in CP011662. Interestingly, the four representatives all contained two RND family efflux transporter genes, acrA-like and acrB-like. Moreover, a set of genes, including asmA, hp, kdgK, yhjJ, dctA, and yhjK were inserted between tetR and ramA in CP024834. The chromosome of L. adecarboxylata R25 encoded four clusters of antibiotic resistance-related IMP-MFP genes and an IMP gene independent of the RND systems (Table 6 and Fig. 2). These genes included mdtABC, acrEF, acrAB, acrABlike, and acrD. In the neighboring regions of the four clusters of IMP-MFP genes, the predicted regulators of RND systems were identified. These regulators were ramA, baeSR, ttgR, and acrR, respectively, of which only two (ramA and baeSR) were positive regulators (activating the transcription of the RNDtype efflux pumps) (Fig. 2). long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.