Characterization of Fosfomycin Resistant Extended-Spectrum β-Lactamase-Producing Escherichia coli Isolates from Human and Pig in Taiwan

To investigate the efficacy of fosfomycin against extended-spectrum β-lactamases (ESBL) producing Escherichia coli in Taiwan and the resistance mechanisms and characterization of human and pig isolates, we analyzed 145 ESBL-producing isolates collected from two hospitals (n = 123) and five farms (n = 22) in Taiwan from February to May, 2013. Antimicrobial susceptibilities were determined. Clonal relatedness was determined by PFGE and multi-locus sequence typing. ESBLs, ampC, and fosfomycin resistant genes were detected by PCR, and their flanking regions were determined by PCR mapping and sequencing. The fosfomycin resistant mechanisms, including modification of the antibiotic target (MurA), functionless transporters (GlpT and UhpT) and their regulating genes such as uhpA, cyaA, and ptsI, and antibiotic inactivation by enzymes (FosA and FosC), were examined. The size and replicon type of plasmids carrying fosfomycin resistant genes were analyzed. Our results revealed the susceptibility rates of fosfomycin were 94% for human ESBL-producing E. coli isolates and 77% for pig isolates. The PFGE analysis revealed 79 pulsotypes. No pulsotype was found existing in both human and pig isolates. Three pulsotypes were distributed among isolates from two hospitals. ISEcp1 carrying bla CTX-M-group 9 was the predominant transposable elements of the ESBL genes. Among the thirteen fosfomycin resistant isolates, functionless transporters were identified in 9 isolates. Three isolates contained novel amino acid substitutions (Asn67Ile, Phe151Ser and Trp164Ser, Val146Ala and His159Tyr, respectively) in MurA (the target of fosfomycin). Four isolates had fosfomycin modified enzyme (fosA3) in their plasmids. The fosA3 gene was harboured in an IncN-type plasmid (101 kbp) in the three pig isolates and an IncB/O-type plasmid (113 kbp) in the human isolate. In conclusion, we identified that 6% and 23% of the ESBL-producing E. coli from human and pigs were resistant to fosfomycin, respectively, in Taiwan. No clonal spread was found between human and pig isolates. Functionless transporters were the major cause of fosfomycin resistance, and the fosA3-transferring plasmid between isolates warrants further monitoring.


Introduction
The emerging problem of extended-spectrum β-lactamases (ESBLs) producing Enterobacteriaceae has been reported worldwide [1,2]. Infections of ESBLs-producing Enterobacteriaceae are difficult to treat because many of the commonly used antibiotics (β-lactams) will not work against these isolates. Falagas et al. reported that fosfomycin could be a solution to this problem; the susceptibility to fosfomycin was 96.8% (1604/1657) in ESBL-producing Escherichia coli and 81.3% (608/748) in Klebsiella pneumoniae in a systemic review [3]. In addition, oral fosfomycin use is effective against uncomplicated urinary tract infections caused by ESBL-producing E. coli, Proteus mirabilis, K. pneumoniae and Staphylococcus saprophyticus [3][4][5][6].
The fosfomycin resistance mechanisms that have been reported include the modification of the antibiotic target (MurA), functionless transporters (GlpT and UhpT transporters and their regulating genes such as uhpA, cyaA, and ptsI) and antibiotic inactivation (FosA and FosC) in E. coli [7]. Due to the fosfomycin inactivation genes (fosA3 and fosC) that were found in the conjugative plasmid or transposon element, transferable fosA3 and fosC genes have become a serious problem in East Asia [8][9][10][11][12][13][14]. Recent studies have focused on the disseminative fosfomycin resistant genes. These reports revealed fosA3 genes were flanked by IS26 and were localized on conjugative plasmids for human isolates in Japan [13,14], Korea [12], and another fosA variant (fosKP96) was found in Hong Kong [8], while some studies have focused on livestock or animal isolates from different regions of China [9][10][11]. In this study, we aimed to identify fosfomycin resistance and its underlying mechanisms and characterization of ESBL-producing E. coli isolates from human and pig sources in Taiwan.

Bacterial isolates
A total of 145 ESBL-producing isolates of E. coli were collected from 2 hospitals and 5 local farms in Taiwan during February-May 2013. Ninety-five isolates were collected from the Kaohsiung Medical University Hospital (KMUH), and the sources included urine (n = 69), blood (n = 7), abscess (n = 8), and others (n = 11). Twenty-eight isolates were collected from the Pingtung Hospital (PTH), and the sources included urine (n = 18), blood (n = 6), pus (n = 3), and sputum (n = 1). Twenty-two pig isolates were all collected from fecal specimens from pigs. The sizes of farms were ranged from 2000 to10000 feeder pigs (Landrace and American Yorkshire pigs). The body weight and age ranged from 7 to 25 Kg and 28 Days to 8 weeks, respectively. All pig isolates were belonging to fecal sample isolated from anal swabs. The distances from hospitals to pig farms were about 20 kilometers.We analysed the bacterial isolates only and the categories of human specimen information or the private information of patient were not included in the study.

Ethics statement
This study was approved by the Institutional Review Board (IRB) of Kaohsiung Medical University Chung-Ho Memorial Hospital, Kaohsiung, Taiwan (KMUH-IRB-860-911). The study subjects were bacterial isolates and the written consent given by the patients was waived by the approving IRB. Before collecting fecal specimens from pigs on this study, we contacted the farm owners and obtained their permission. No specific permits were required for the described field studies and the locations where we sampled are not privately-owned or protected in any way. The field studies did not involve endangered or protected species.

Antimicrobial susceptibility testing
The antimicrobial susceptibility testing was performed by a standard agar dilution method according to the guidelines of the Clinical and Laboratory Standards Institute (CLSI) [15]. Three antimicrobial agents: fosfomycin, cefotaxime, and meropenem were tested in all isolates. The phenotypic detection of ESBLs was conducted by Vitek 2 system (bioMérieux Inc., Durham, NC). The susceptibilities of 4 fosA3-habouring isolates and their transformants were tested for the following antimicrobial agents including ampicillin, amikacin, tetracycline, levofloxacin, chloramphenicol, fosfomycin, cefotaxime, and meropenem.

Bacterial strain typing (PFGE and MLST)
The PFGE typing of XbaI (New England BioLabs, Ipswich, MA)-digested DNA was prepared in accordance with previously described methods [16]. Dice similarity indices were employed to construct the dendrogram of the pulsotype relationships through the unweighted pair group method using arithmetic averages (UPGMA) with BioNumerics software version 6.5 (Applied Maths). The pulsotypes were assigned to the same clusters if they exhibited 80% similarity in the dendrogram. The MLST scheme of E. coli uses internal fragments from seven housekeeping genes: adk (adenylate kinase), fumC (fumarate hydratase), gyrB (DNA gyrase), icd (isocitrate/ isopropylmalate dehydrogenase), mdh (malate dehydrogenase), purA (adenylosuccinate dehydrogenase), and recA (ATP/GTP binding motif). The primers were derived from the E. coli MLST database (http://mlst.ucc.ie/mlst/dbs/Ecoli). The PCR amplification and sequencing were performed following this website's suggested protocols. The four fosA3-habouring isolates were analyzed by MLST.

Detection of antimicrobial resistance genes
The plasmid DNA was extracted using the QIAGEN Plasmid Mini Kit. For the detection of the ESBLs (bla SHV , bla TEM , bla CTX-M-group 1 , bla CTX-M-group 2 , and bla CTX-M-group 9 ) and plasmidmediated AmpC genes bla DHA , and bla CMY and the fosfomycin resistance genes (fosA, fosA3, fosC2), the primer sets from previous reports were used [10,17]. The purified PCR amplicons of bla TEM were sequenced by the dideoxy chain-termination method with an automated DNA Sequencer (Perkin-Elmer ABI3700), and the nucleotide sequences were analyzed with the BLAST sequence alignment database (National Center for Biotechnology Information).

Analysis of the transposable elements upstream of the bla genes
To investigate the different kinds of transposable elements surrounding the blaSHV, blaCTX and blaTEM genes, the linkage of blaSHV with recF and IS26 was determined using specific primers (RECF-F and SHV-F-I; IS26-FCJ and SHV-F-I) [18]. The linkage of blaTEM with tnpA and IS26 was determined using specific primers (RH401 and RH410; RH1270 and RH606) [19]. The linkage of blaCTX-M-group 1 with ISEcp1 and IS26 was determined using specific primers (ISEcpUP and CTX-M1RCJ; IS26-FCJ and CTX-M1RCJ) [18]. The linkage of blaCTX-M-group 9 with ISEcp1 was determined using specific primers (ISEcpUP and CTX-M-9-R1) [18]. All the PCR amplicons were sequenced, and the sequence was compared in the GenBank nucleotide database.

Nucleotide sequencing of fosA3 and the flanking regions
To determine the genetic environment of fosA3, PCR mapping and an LA PCR in vitro cloning kit (Takara Shuzo Co. Ltd., Japan) were used. The flanking regions between IS26 and fosA3 were identified by PCR amplicons and sequencing. The primers used were those from previous studies [9].

Conjugation and transformation assays
The conjugation and electotransformation assays were performed as previous studies [12]. The conjugation experiments were conducted according to the broth mating method using E. coli J53 as the recipient. The transconjugants were selected on LB agar plates supplemented with sodium azide (150 μg/ml), fosfomycin (40 μg/ml), and glucose-6-phosphate (G6P) (25 μg/ml). If the conjugation assay failed, an electotransformation assay using E. coli DH5α as the recipient was conducted. The transconjugants were selected on LB agar plates supplemented with fosfomycin (40 μg/ml) and G6P (25 μg/ml).

Plasmid analysis and Southern blot hybridization
The plasmid incompatibility group analyses were determined according to the PCR-based replicon-typing method [20]. The size estimation of the plasmids was determined by an S1-nuclease PFGE analysis and calculated with the BioNumerics GelCompar software package (Version 5.0, Applied Mathematics, Sint-Martens-Latem, Belgium) as previously described [17]. The Lambda Ladder PFG marker (New England Biolabs, Ipswich, MA, USA) was used as a molecular size marker. The plasmid PFGE gel was then subjected to Southern blotting and hybridized with a DIG (digoxigenin)-labeled fosA3 and CTX-M-group 9 specific probe. The primer sets from previous reports were used [10,17]. The hybridization assay was performed as described previously [17].

Activity of GlpT and UhpT transpoters
The usage of carbohydrates which indicated the activity of GlpT and UhpT transpoters was performed as previous studies [21,22]. Briefly, 0.2% (w/v) G6P or sn-glycerol 3-phosphate (G3P) as the sole carbon source was supplied in M9 minimal medium agar. The overnight bacterial suspension was washed with an equivalent volume of saline and optimal McFarland No. 4. The bacterial suspensions (200 μl) were plated on M9 minimal medium agar supplying different sole carbohydrates at 37°C for 48 h. The negative phenotype was defined as a lack of colonies forming on the plate.

Sequencing analysis of fosfomycin resistance related chromosomal genes
PCR and sequencing were performed to amplify the entire sequences of the 6 genes including murA, glpT, uhpT, uhpA, ptsI, and cyaA. The primers used were those from a previous study [22]. The variations of amino acids were compared to the reference strain E.coli ATCC 25922.

Cloning overexpression of MurA from the different amino acid variations
The murA gene (ATCC25922, A12, A19, and K55) was amplified by PCR using a cloning primer, as previously described [22]. The PCR products were cloned into a T & A (Yeastern Biotech, Taiwan) plasmid and were transformed into E. coli DH5α. The insertion of the murA gene was verified by nucleotide sequence. To determine the fosfomycin susceptibility of the overexpressing murA in E. coli, 25 μg/ml G6P and 1 mM isopropyl β-D-1-thiogalactopyranoside (IPTG) were added to Muller-Hinton agar with various concentrations of fosfomycin (0-1024 μg/ml).
Transposable elements were usually found in the upstream region of ESBLs and could be related to horizontal genetic transferring [18,19]. PCR mapping and sequencing revealed different transposable elements related to ESBLs (S1 Fig). In the upstream region of SHV, two types were found: recF-kdpC-bla SHV (SHV Type A ) and IS26-73 bp spacer sequencebla SHV (SHV Type B ). ISEcp1 was found upstream of the CTX-M-group 1 gene, which was the dominant type (83% of the KMUH isolates, 90% of the PTH isolates, and 50% of the pig isolates) (S1 Table). Two different lengths of spacer regions (42 bp in CTX-M-G9 Type A and 127 bp in CTX-M-G9 Type B ) were found between ISEcp1 and bla CTX-M-group 9 . CTX-M-G9 Type A was the predominant type in the two hospitals and in the pig isolates. 9 isolates that did not grow on G3P, four isolates (A19, K66, K116 and P12) revealed amino acid substitutions in the glpT gene. The A19 and K66 isolates were identified to have single amino acid substitutions at codons 174 (Leu174Val) and 209 (Arg209His), respectively. K66 isolate also carried fosA3 gene that causes fosfomycin resistance. The K116 isolate had 2 amino acid changes (Gly142Cys and Phe176Ser), and the P12 isolate had 3 amino acid changes (Arg50Cys, Ala156Val, and Val149Met). Two isolates (K55 and K101) which did not grow on G3P did not have amino acid changes in the glpT gene but possessed amino acid substitutions in the upstream regulator ptsI (Thr215Ala and Gly227Arg in K55 and Ala104Thr in K101). Three isolates (A10, A11, and A12) did not have amino acid changes in the glpT and ptsI genes; however, these isolates contained the fosA3 gene that results in fosfomycin resistance.

Characterization of 4 fosA3-habouring isolates
For the four fosA3-habouring isolates (A10, A11, A12, and K66), the surrounding region of fosA3 was flanked by two IS26 elements (type A ; Fig 2). Three isolates (A10 from farm A, A12 from farm C, and K66 from KMUH) showed a 100% identity to the GenBank accession number AB522970, which contained fosA3, orf1, orf2 and truncated orf3 (type A ; Fig 2) [14]. The other isolate (A11 from farm A) had an ISAplI inserting a similar structure between IS26 and fosA3 (type B ; Fig 2) (GenBank accession number was KT199757). The PFGE and MLST typing revealed that 3 different pig isolates (pulsotype XXXIV and ST744) were the same genotype, which differed from the human isolate (pulsotype XXX and ST2310) ( Table 3). The plasmid profiles of the pig isolates were similar (146, 101, and 85 kbp) whereas the human isolate had 2 different sized plasmids (184 and 113 kbp). The locations of the fosA3 and bla CTX-M-group 9 genes were analyzed by S1 nuclease-PFGE and hybridization with specific probes for the fosA3 and bla CTX-M-group 9 genes, respectively. The fosA3 gene was hybridized with a 101 kbp IncN-type plasmid in the pig isolates and a 113 kb IncB/O-type plasmid in the human isolate (Fig 3B and  Table 3). In addition, the 101 kbp IncN-type plasmid was hybridized with the bla CTX-M-group 9 Table 1. Antibiotic resistant genes of the ESBL-producing E. coli human and pig isolates. probe in the pig isolates (Fig 3C), indicating the co-localization of the fosA3 and bla CTX-M-group 9 genes in the pig isolates. Although, IncB/O-type plasmid contained fosA3 without bla CTX-M-group 9 gene, bla CMY-2 was found in transformant of IncB/O-type plasmid (K66) by PCR detection. The antibiotic susceptibility tests revealed that these 4 clinical isolates were resistant to ampicillin, amikacin, tetracycline, levofloxacin, chloramphenicol, fosfomycin, and cefotaxime without meropenem, while the transformants were only resistant to ampicillin, cefotaxime and fosfomycin.

Modification of the fosfomycin target MurA
Amino acid substitutions of MurA were found in two pig isolates: A12 (Asn67Ile), A19 (Phe151Ser and Trp164Ser), and one hospital isolate: K55 (Val146Ala and His159Tyr), respectively. To test the correlation between these amino acid substitutions and fosfomycin resistance, wild type and mutant MurA were constructed and expressed in E. coli DH5α ( Table 4). The variations of MurA were shown to cause a 16-fold increase of the fosfomycin MIC compared to the wild-type MurA (Table 4). These results indicate that the amino acid substitutions in MurA were related to fosfomycin resistance.

Discussion
To our knowledge, this study was the first to investigate fosfomycin resistance of both human and pig ESBL-producing E. coli isolates in Taiwan. Thirteen (9%) ESBL-producing E. coli isolates were resistant to fosfomycin in the study. The resistant rate to fosfomycin was lower in human isolates (6% in KMUH and 11% in PTH) than the resistance rate in pig isolates (22%).    Among the only two studies on fosfomycin susceptibility from human isolates in Taiwan, one study revealed that 95.5% ESBL-producing E. coli isolates were susceptible to fosfomycin [23] and the other revealed that 100% E. coli isolates were susceptible to fosfomycin [24]. According to the two reports, fosfomycin is a therapeutic choice for treating ESBL-producing E. coli infection in Taiwan. We identified a relatively high fosfomycin resistant rate for pig ESBL-producing E. coli isolates. Our results suggested antibiotic selection pressure might exist in pig farms. The molecular epidemiology analysis by the PFGE found no clonal associations between human and pig isolates. Clonal spread of three pulsotypes might occur in two hospitals. In this study, four isolates contained the fosA3 gene in a plasmid, and nine isolates displayed the mutation in murA, glpT, uhpT, uhpA, ptsI, and cyaA genes. Nine and eight isolates did not grow on minimal medium agar supplemented with G3P and G6P, respectively. This result indicated that the UhpT and/or GlpT transporter may be defective in these isolates. Although, amino acids substitutions in transporters or regulated genes were not evidenced by in vitro experiments in this study, functionless transporters (GlpT or UhpT) could be detected by minimal medium agar with G3P or G6P, as the previously reported [21,22]. Takahata et al. reported 6 fosfomycin resistant clinical E. coli isolates that harbored different fosfomycin-related gene mutations [22]. The glpT gene mutation (4 isolates) and the loss of the entire uhpT gene (2 isolates) decreased the fosfomycin uptake into the bacterial cells, leading to fosfomycin resistance. The other study by Nilsson et al. investigated fosfomycin resistance mechanisms in 13 clinical E. coli isolates [21]. A mutation in the cyaA (5 isolates) and ptsI (2 isolates) genes impaired both glpT and uhpT expression. In our study, three and one isolates carried mutations in their uhpT and glpT genes, respectively. MurA has been reported to confer clinical fosfomycin resistance at a low fitness cost [25]. The other study described two isolates that contained the murA mutation (Asp369Asn or Leu370Ile), leading to fosfomycin resistance [22]. Our results revealed new variations (Asn67Ile, Phe151Ser and Trp164Ser, Val146Ala and His159Tyr, respectively) in MurA ( Table 4) that might affect fosfomycin binding to the Cys-115 residue (active site) or the three conserved positively charged residues (Lys22, Arg120 and Arg397) in MurA [26].
Research regarding the transposable elements surrounding ESBLs was limited in Taiwan [29,30]. The detection of ESBLs and the linkage of transposable elements were investigated in this study. The most common ESBLs were CTX-M-group 1, CTX-M-group 9 and CMY-2 in E. coli in Taiwan. Our results revealed the same transposable elements surrounding ESBLs in human and pig isolates that indicate that ESBLs could be transferred via a plasmid or transposon between human and pig hosts. SHV was predominant in ESBL-producing K. pneumoniae isolates in Taiwan [30] whereas a low percentage (5.5%) of SHV was found in the ESBL-producing E. coli isolates in this study (Table 2). IS26 was related to disseminated antibiotic resistance genes worldwide [31,32]. In this study, IS26, which was located upstream of blaSHV, and fosA3, was associated with the dissemination of antibiotic resistance genes in Taiwan (S1 Fig and Fig 2). ISEcp1 was mostly found upstream of blaCTX-M in Spanish, French, Indian and Turkish [18,33,34]. Our study identified the role of ISEcp1 in Taiwan that supported its role for global blaCTX-M dissemination. Besides, ISEcp1 was upstream of the blaCTX-M-group 1 and blaCTX-M-group 9 genes with different spacer regions, (S1 Fig). Our previous study demonstrated that blaCTX-M expression was correlated with different spacer sequences (42 and 127 bp), which affected blaCTX-M expression [35]. In this study, the 42 bp spacer region was found predominant (S1 Table).
It should be noted that the number of pig isolates was small in this study. The prevalence of resistance to fosfomycin required further studies with a larger isolate number. Due to the limited sample size, the clonal relationship between human and pig isolates required further studies with a larger isolate number.
In conclusion, the susceptibility rates of fosfomycin were different between human and pig ESBL-producing E. coli isolates. The major resistance mechanism of fosfomycin was amino acid variation in chromosomally encoded fosfomycin-related genes such as the antibiotic target gene (murA), GlpT and UhpT transporters (glpT and uhpT) and regulating genes (uhpA, cyaA, and ptsI). The IncN-type plasmid carrying fosA3 and bla CTX-M-group 9 in pig isolates is of concern. In addition, the horizontal fosA3-transferring via the IS26 mobile element between human and pig isolates might exist in Taiwan and warrant further monitoring.