ABSTRACT
Non-culture-based procedures were used to investigate plasmids showing ampicillin resistance properties in two different environments: remote mountain soil (Mt. Jeombong) and sludge (Tancheon wastewater treatment plant). Total DNA extracted from the environmental samples was directly transformed into Escherichia coli TOP10, and a single and three different plasmids were obtained from the mountain soil and sludge samples, respectively. Interestingly, the restriction fragment length polymorphism pattern of the plasmid from the mountain soil sample, designated pEMB1, was identical to the pattern of one of the three plasmids from the sludge sample. Complete DNA sequencing of plasmid pEMB1 (8,744 bp) showed the presence of six open reading frames, including a β-lactamase gene. Using BLASTX, the orf5 and orf6 genes were suggested to encode a CopG family transcriptional regulator and a plasmid stabilization system, respectively. Functional characterization of these genes using a knockout orf5 plasmid (pEMB1ΔparD) and the cloning and expression of orf6 (pET21bparE) indicated that these genes were antitoxin (parD) and toxin (parE) genes. Plasmid stability tests using pEMB1 and pEMB1ΔparDE in E. coli revealed that the orf5 and orf6 genes enhanced plasmid maintenance in the absence of ampicillin. Using a PCR-based survey, pEMB1-like plasmids were additionally detected in samples from other human-impacted sites (sludge samples) and two other remote mountain soil samples, suggesting that plasmids harboring a β-lactamase gene with a ParD-ParE toxin-antitoxin system occurs broadly in the environment. This study extends knowledge about the dissemination and persistence of antibiotic resistance genes in naturally occurring microbial populations.
INTRODUCTION
Antibiotic treatments have been one of the most effective ways to control infectious diseases, but the frequency of detecting bacteria that are resistant to antibiotics from environmental samples has recently increased (1). It is generally thought that the use of antibiotics as human and veterinary medicine or as animal feed additives is a major driving force in the development of antibiotic-resistant bacteria and the dissemination of antibiotic resistance genes (2–4). Indeed, high levels of antibiotic resistance genes have been identified in a variety of human-impacted milieus such as wastewater sludge, soils fertilized with manure, and river waters that have been frequently exposed to antibiotics (5–9). Surprisingly, however, many other studies have shown that the widespread occurrence of antibiotic resistance genes are sometimes unrelated to human activities. For example, antibiotic resistance genes have been documented in pristine habitats such as Alaskan soil, Antarctic marine waters, ancient sediment samples, glacier ice cores, and nonagricultural soil (2, 10–15), often in abundances well above trace levels. These findings raise questions about the stable maintenance of antibiotic resistance genes in the absence of human-mediated selective pressures. It should also be noted that some antibiotic resistance genes may have functions other than neutralizing antibiotics (16, 17).
Antibiotic resistance traits in bacteria can be acquired via lateral gene transfer (4, 13). In particular, plasmids may play important roles as mobile genetic elements that can disseminate antibiotic resistance genes to clinically relevant pathogenic bacteria (18, 19). Surprisingly, some recent studies have shown that antibiotic resistance genes encoded by plasmids were stably maintained in the absence of selective pressures (20–22); however, in these studies, the mechanisms of plasmid maintenance were not suggested.
Bacterial toxin-antitoxin (TA) systems have been identified in many pathogenic bacteria; TA systems can stably maintain plasmids in their hosts via postsegregational killing of daughter cells that fail to inherit the TA systems because antitoxin degradation occurs faster than toxin degradation (23–27). In the present study, we discovered and analyzed a putative TA system linked to a β-lactamase gene borne by a plasmid that was isolated from both wastewater sludge and pristine mountain soil samples. Although β-lactamase and ampicillin resistance plasmids have been recovered from environmental samples (28–31), none of these plasmids, to our knowledge, has been fully sequenced, and none has been associated with TA systems. We functionally characterized here a putative TA system and showed that it can play an important role in maintaining the plasmid in its host in the absence of antibiotic selective pressure. This study examines in detail the dissemination and persistence of antibiotic resistance genes in naturally occurring microbial populations.
MATERIALS AND METHODS
Bacterial strains, plasmids, and PCR primers.The E. coli strains, plasmids, vectors, and PCR primers used in the present study are listed in Table 1. All E. coli strains were grown at 37°C in Luria-Bertani (LB) medium in a rotary incubator (180 rpm).
Bacterial strains, plasmids, and PCR primers used in this study
Isolation of plasmids harboring ampicillin resistance genes.To isolate plasmids harboring ampicillin resistance genes from environmental samples, activated sludge and pristine mountain soil samples were collected from the Tancheon municipal wastewater treatment plant (WWTP; Seoul) and a remote area of Mt. Jeombong (38°02′17″N, 128°27′40″E, altitude approximately 900 to 950 m, Gangwon-do) in South Korea, respectively. Approximately 200 g of soil was sampled from the upper 1 to 5 cm of the mountain soil using a sterile spatula and aseptic plastic bags (Fisher Scientific, USA) and transported immediately to the laboratory on ice. Total DNA from sludge and soil samples was extracted by using a FastDNA spin kit (MP Biomedicals, USA) according to the manufacturer's instructions with a modification in the homogenization step: homogenization for the cell lysis was performed by using a Mini-Beadbeater 3110BX (BioSpec, USA) for 30 s at 4,800 rpm. Approximately 250 ng of DNA was directly transformed into 50 μl of competent E. coli TOP10 cells (Invitrogen, USA) by the electroporation method using a Micropulser system (Bio-Rad, USA). The transformed suspension was spread onto LB agar containing ampicillin (50 μg/ml) and incubated at 37°C overnight. Colonies were randomly picked and cultivated overnight in 5 ml of LB broth containing ampicillin. Plasmid DNA from each culture was extracted using a plasmid mini-extraction kit (Bioneer, South Korea) according to the manufacturer's instructions. To evaluate the diversity of the plasmids, a restriction fragment length polymorphism (RFLP) technique with Sau3AI was used. The digested plasmids were analyzed on a 1.5% (wt/vol) agarose gel, and their fragment patterns were compared. Plasmids showing the same RFLP patterns were considered to be the same plasmids, and this was verified by partial sequencing of the plasmids using BlaParD-OF and BlaParD-OR primers (Table 1): the primers were designed after the complete sequencing of plasmid pEMB1.
Sequencing of plasmid DNA.To obtain complete sequences of the three plasmids, designated pEMB1, pEMB2, and pEMB3, showing unique RFLP patterns, 25 μg of each purified plasmid DNA in 50 μl of TE buffer (10 mM Tris-Cl, 1 mM EDTA [pH 8.0]) was mixed with 750 μl of shearing buffer (10% glycerol in TE buffer) and fragmented by mechanical shearing using a nebulizer (Invitrogen) at 13 lb/in2 (nitrogen) for 2 min on ice to obtain a mean fragment size of 1.5 to 2.0 kb. Each of the fragmented plasmids was precipitated by centrifugation after the addition of 1 volume of ice-cold isopropanol and end repaired using DNA end repair mix (Invitrogen) according to the manufacturer's instructions. The DNA fragments of approximately 1.5 to 2.0 kb in size were extracted by using a gel purification kit (Bioneer, South Korea) following agarose gel electrophoresis and ligated into the EcoRV site of pCR2.1-TOPO (Invitrogen). This was followed by electrotransformation into E. coli TOP10 to construct clone libraries for each of the three plasmids. Plasmid DNA from 25 colonies of each clone library were extracted and then sequenced using the M13 reverse and T7 primers and a 3730xl DNA Analyzer. The resulting sequences were assembled using SeqMan in DNAStar Lasergene (DNASTAR, USA) and the assembled contigs were gap-closed using a primer-walking approach. Regions with poor sequencing qualities in each plasmid were resequenced and, finally, the complete sequences for plasmids pEMB1, pEMB2, and pEMB3 were obtained.
Sequence analysis of plasmids.The origin of replication in plasmid pEMB1 was predicted by Ori-Finder (http://tubic.tju.edu.cn/Ori-Finder/) (32). Putative open reading frames (ORFs) were predicted using the Web-based ORF Finder program at the National Center for Biotechnology Information (NCBI) and potential protein-coding sequences were subsequently analyzed using BLASTX searches. Conserved domains and motifs of the putative proteins were searched using the Conserved Domain search service at the NCBI (33). Multiple amino acid sequences were aligned using ClustalX in DNAStar Lasergene. Restriction enzyme sites on plasmid pEMB1 were identified using the NEBcutter version 2.0 program (http://tools.neb.com/NEBcutter2/index.php). Transcriptional promoters and termination sequences in plasmid pEMB1 were predicted by using a Web-based program (Softberry [http://www.softberry.com/berry.phtml?topic=index&group=programs&subgroup=gfindb]).
Determination of MIC.MICs of some β-lactam antibiotics, including ampicillin, penicillin G, cefotaxime, meropenem, and aztreonam, against E. coli TOP10 harboring pEMB1 were determined using the serial 2-fold dilution method of antibiotics based on the Clinical and Laboratory Standards Institute (CLSI) guidelines (34). Briefly, overnight cultures of E. coli TOP10 harboring pEMB1 in LB broth containing ampicillin (50 μg/ml) at 37°C were diluted to be 106 cells/ml in LB broth. The diluted cells were dispensed in 100-μl portions into 96-well plates with 100 μl of LB broth containing 2-fold-diluted antibiotics (2 to 2,048 μg/ml) and incubated at 37°C for 18 h. The growth was assayed by using a microtiter enzyme-linked immunosorbent assay reader (SynergyMx; BioTek, USA) at 660 nm, and the MICs were determined as the lowest concentrations of antibiotics to inhibit the growth of the test strain. E. coli TOP10 strain without pEMB1 was used as a negative control.
Construction of knockout plasmids of orf5 and orf6 and cloning or expression of orf6.orf5 and orf6, which were tentatively annotated as putative antitoxin (parD) and toxin (parE) genes in plasmid pEMB1, respectively, were functionally characterized. To construct knockout plasmids of orf5 and orf6, pEMB1ΔparD and pEMB1ΔparE, plasmid pEMB1 was separately digested with MfeI and NcoI, single cutting enzymes that cleave orf5 and orf6 genes, respectively, and subsequently treated with mung bean nuclease (New England BioLabs, United Kingdom) to generate blunt ends. Each blunt-ended plasmid DNA was self-ligated using T4 DNA ligase and transformed into competent E. coli TOP10 cells by electroporation. The construction of pEMB1ΔparD and pEMB1ΔparE with 4-bp deletion in orf5 and orf6, respectively, in transformed E. coli TOP10 cells was confirmed via PCR using the primer pair OparD-F and OparE-R (Table 1) and sequencing.
To characterize the functional properties of orf6 as a putative toxin gene (parE), orf6 was cloned into expression vector pET21b (Novagen, USA) as described previously (35). Briefly, orf6 was PCR amplified from plasmid pEMB1 using Pfu DNA polymerase (Solgent, South Korea) and the primer pair ParE-F and ParE-R, which contained NdeI and BamHI restriction sites, respectively (Table 1). The amplified PCR product was excised with NdeI and BamHI and ligated into the NdeI and BamHI sites of pET21b. The resulting recombinant plasmid, designated pET21bparE, was transformed into E. coli BL21(DE3), and the cloning of orf6 in transformed E. coli BL21(DE3) cells was confirmed by PCR and sequencing. The function of orf6 as a toxin protein (ParE)-encoded gene was evaluated by the growth of E. coli BL21(DE3) cells containing pET21bparE on LB agar with or without induction using IPTG (isopropyl-β-d-thiogalactopyranoside; 1 mM) at 37°C after 2 days. In addition, the growth of E. coli BL21(DE3) harboring pET21bparE was also tested in LB broth with or without IPTG. Portions (1 ml) of overnight cultures of E. coli BL21(DE3) harboring pET21bparE in LB broth containing ampicillin (50 μg/ml) at 37°C were inoculated into two flasks containing 100 ml of LB broth with ampicillin, and the flasks were incubated with shaking at 37°C. When the optical density at 660 nm (OD660) of the cultures reached ∼0.6, IPTG (1 mM) was added to one flask, and the cells were continually cultivated. The growth of E. coli BL21(DE3) without pET21bparE was tested in parallel in LB broth without ampicillin. The growth and viability of E. coli cells were monitored by measuring the OD660 and CFU.
Construction of a pEMB1 plasmid with orf5 and orf6 deleted and plasmid stability assays.For the construction of pEMB1 with the putative antitoxin (orf5) and toxin (orf6) genes deleted, plasmid pEMB1 was digested with BsrGI and PvuII cutting upstream and downstream of orf5 and orf6, and the plasmid DNA ends were repaired using the Klenow treatment to generate blunt ends. The blunt-ended plasmid DNA was ligated using T4 DNA ligase and transformed into E. coli TOP10 by electroporation, resulting in pEMB1ΔparDE. The deletion of orf5 and orf6 in plasmid pEMB1 in transformed E. coli TOP10 cells was verified by PCR using the outer primer pair OparD-F/OparE-R, and by sequencing (Table 1).
The stability of pEMB1 and pEMB1ΔparDE was investigated in triplicate as previously described, with some modifications (24). E. coli TOP10 cells containing pEMB1 and pEMB1ΔparDE were cultivated in LB broth containing ampicillin (50 μg/ml) overnight at 37°C, and 50-μl portions of the overnight-incubated cultures were used to inoculate 5 ml of fresh LB broth with or without ampicillin. After 8 h of cultivation at 37°C, 1% of the cultures was used to inoculate 5 ml of fresh LB broth with or without ampicillin. This process was repeated until approximately 500 generations of growth were achieved. Periodically, culture broths were serially diluted in 0.9% saline and plated onto LB agar without ampicillin for viable cell counts. Viable-count plates containing approximately 100 colonies were replica plated onto LB agar plates with ampicillin. The proportion of E. coli cells containing pEMB1 or pEMB1ΔparDE plasmids was calculated by comparing E. coli cell numbers grown on LB agar with or without ampicillin. The maintenance or loss of pEMB1 or pEMB1ΔparDE in E. coli after the stability assays was confirmed by plasmid extraction from selected colonies.
PCR-based monitoring of pEMB1-like plasmids in additional habitats.A nested touchdown-PCR approach using PCR primer pairs targeting plasmid pEMB1 was used to monitor the presence of pEMB1-like plasmids in samples aseptically gathered from various environments described below. Two nested PCR primer pairs, BlaParD-OF/BlaParD-OR and BlaParD-IF/BlaParD-IR, were designed to target orf3 (bla), encoding β-lactamase, and orf5, encoding a putative ParD in plasmid pEMB1 (Table 2). In addition to the Tancheon WWTP sludge and Mt. Jeombong soil samples, 20 additional environmental samples were obtained from various habitats (four WWTPs, 10 remote mountain soils, three tidal flats, and three rice paddies), and the total DNA was extracted using a FastDNA spin kit (MP Biomedicals, USA) according to the manufacturer's instructions. The first round of touchdown PCR was carried out using the outer primer pair (BlaParD-OF/BlaParD-OR) with a cycling regime of 94°C for 5 min (1 cycle), followed by 94°C for 45 s, 60°C for 45 s, and 72°C for 2 min and 30 s, followed by 20 cycles at decreasing annealing temperatures in decrements of 0.5°C per cycle, followed by 94°C for 45 s, 50°C for 45 s, and 72°C for 2 min and 30 s (15 cycles), and 72°C for 10 min (1 cycle). The second round of touchdown PCR was performed using the inner primer pair (BlaParD-IF/BlaParD-IR) and 1 μl of first-round PCR products as with the first round of touchdown PCR. The PCR products were verified by direct sequencing of the second-round PCR products.
Open reading frames found in plasmid pEMB1 and their putative protein functionsa
RESULTS
Isolation of plasmids harboring ampicillin resistance genes.To find environmental plasmids conferring ampicillin resistance, total DNA extracted from activated sludge (Tancheon WWTP) and remote mountain soil (Mt. Jeombong) samples was directly transformed into E. coli TOP10 cells. RFLP analysis of plasmids recovered from transformed ampicillin-resistant E. coli colonies revealed that only a single plasmid type was obtained from the remote mountain soil sample (Fig. 1A), while plasmids with three different RFLP patterns were obtained from the sludge sample (Fig. 1B; plasmid 1, lanes 4 and 6; plasmid 2, lanes 1, 2, 5, and 8; and plasmid 3, lanes 3 and 7). Interestingly, the RFLP pattern of the plasmids from the remote mountain soil was identical to that of plasmid 1 derived from the activated sludge sample, and partial plasmid sequencing of ∼2 kb revealed no sequence differences between representatives of each (data not shown). We selected one plasmid representative of plasmid RFLP patterns 1, 2, and 3 (Fig. 1) and designated the three identified plasmids as pEMB1, pEMB2, and pEMB3, respectively.
RFLP analysis of plasmids from environmental samples conferring antibiotic resistance to E. coli. Plasmids derived from remote mountain soil (Mt. Jeombong [A]) and sludge (Tancheon WWTP [B]) samples were digested with Sau3AI. The RFLP analysis showed that the fragment patterns of plasmids in panel A were identical to those of plasmids B4 and B6. Plasmids A1, B1, and B3 were designated pEMB1, pEMB2, and pEMB3, respectively. M, 100-bp ladder (Bioneer, South Korea).
Plasmid sequencing and sequence analysis.After sequencing, plasmids pEMB1, pEMB2, and pEMB3 were determined to be 8,744, 9,042, and 9,344 bp in size, respectively. Sequence analysis of the plasmids showed that all three plasmids contained a putative β-lactamase gene that might mediate ampicillin resistance. Plasmids pEMB1 and pEMB3 were novel, but plasmid pEMB2 was very similar to plasmid pQ7 of E. coli TB7 isolated from the urine sample of a Swiss patient, with only 7 bp differences (36). Interestingly, the analysis of potential protein-coding sequences using BLASTX searches showed that plasmid pEMB1, identified from both wastewater sludge and remote mountain soil samples, harbored a putative TA system with the β-lactamase gene (Table 2). Therefore, in the present study, we focused only on the characterization of plasmid pEMB1.
The average G+C content of pEMB1 was 55.41%. Analysis of the plasmid nucleotide sequences revealed six putative ORFs, all more than 78 amino acid residues in length (Table 2). A putative oriC region with five possible E. coli DnaA box motifs was identified in the pEMB1 plasmid, using the Ori-Finder program (Fig. 2). orf1, orf2, and orf3 in plasmid pEMB1were predicted to encode transposase (tnpA), DNA invertase (tnpR), and β-lactamase (bla), respectively, which are three components of the Tn3 transposon. Moreover, a Tn3 transposase DDE domain (Asp689-Asp765-Glu895; pfam01526) was found in the predicted protein product of orf1 (tnpA), suggesting that plasmid pEMB1 might be a chimeric plasmid derived from the easily transferable Tn3 transposon. The predicted protein product of orf3 (bla) shared 82% sequence identity with the class A β-lactamase from Pseudomonas putida HB3267 (YP_007232230.1). The protein product of orf3 (bla) contained four specific elements of class A β-lactamases, an STHK tetrad (positions 78 to 81), an SDN triad (positions 139 to 141), an EPELN pentad (positions 175 to 179), and a KTG triad (positions 243 to 245) (37), similar to other reported class A β-lactamases with carbenicillin-hydrolyzing activity (see Fig. S1 in the supplemental material). The translated orf4 encoded an 80-amino acid protein that was most closely related to the plasmid replication proteins (RepB) from Burkholderia caribensis MBA4 (ETY75818.1) and Aeromonas salmonicida (WP_005321398.1), with 100 and 96% amino acid sequence identities, respectively (Table 2). The predicted protein product of orf4 was therefore designated RepB.
Genetic map of plasmid pEMB1 showing ORFs, putative functions (in parentheses), and oriC as described in the text. Restriction enzyme sites used for genetic manipulation of the present study are also indicated on the map.
The predicted protein product of orf5 showed 92 and 79% identities to a hypothetical protein PAAH01 (predicted as a transcriptional regulator) from Aeromonas hydrophila (YP_005230994.1) and to the CopG family transcriptional regulator from Shewanella putrefaciens 200 (YP_006010550.1), respectively. The translated orf5 encoded a 79-amino-acid protein that included a ribbon-helix-helix (RHH) structure from the RHH family of proteins, which includes an antitoxin ParD and transcriptional repressors CopG, Arc, and Mnt (38), although their sequence identities were very low (Fig. 3). The Orf5 protein also contained specifically positioned hydrophobic residues found in the hydrophobic cores of folding motifs and a highly conserved turn with a typical GXT/S pattern between two helices, as seen in the RHH family of proteins.
Multiple alignment of the predicted protein product of orf5 in pEMB1 with other ribbon-helix-helix proteins. Conserved hydrophobic residues are shaded in light gray, and the highly conserved turns between helices A and B are shaded in dark gray. ParD, Pseudomonas aeruginosa (YP_758694.1); CopG, Streptococcus agalactiae (2CPG_A); Arc, Enterobacteria phage P22 (1PAR_A); Mnt, Enterobacteria phage P22 (1MNT_A).
The predicted protein of orf5 had a C-terminal tail region, which is a distinct feature of ParD and Mnt among the RHH family proteins (Fig. 3) (39, 40). The predicted protein product of orf6 was most closely related to the plasmid stabilization system protein (ParE) of Shewanella putrefaciens 200 and Pectobacterium wasabiae WPP163, although their amino acid sequences were evolutionarily distant from known plasmid stabilization system proteins (70 and 62% identities, respectively) (Table 2). Moreover, Web-based conserved domain analysis based on the Conserved Domain Database showed that the translated protein of orf6 contained a conserved domain of ParE (COG3668), suggesting that orf6 in pEMB1 might be a ParE toxin gene. It is known that toxin genes of bacterial TA systems are expressed with an antitoxin gene in a single gene cluster. It was predicted that a transcription promoter and a termination sequence were present upstream and downstream of the orf5 and orf6 genes, respectively, suggesting that the orf5 and orf6 genes in pEMB1 might be transcribed as a single operon (data not shown). These gene analyses therefore suggested that orf5 that formed a single gene cluster with a putative toxin gene (parE) may be an antitoxin gene (parD), rather than a CopG family transcriptional regulator (Table 2).
Determination of MIC.MICs of five representative β-lactam antibiotics belonging to four β-lactam classes, penicillins (ampicillin and penicillin G), cephalosporins (cefotaxime), carbapenems (meropenem), and monobactams (aztreonam), against E. coli TOP10 cells harboring pEMB1 were determined to investigate the antibiotic resistance range of β-lactamase in pEMB1. As shown in Table 3, the MICs of ampicillin and penicillin G belonging to the penicillin class were >512 and >1,024 μg/ml, respectively (Table 3), meaning that β-lactamase in pEMB1 can readily hydrolyze the β-lactam rings of ampicillin and penicillin G. However, the MIC test showed that E. coli harboring pEMB1 was very sensitive to cefotaxime, meropenem, and aztreonam (<1 μg/ml), suggesting that the β-lactam rings of antibiotics belonging to other β-lactam classes other than penicillins are resistant to hydrolysis by β-lactamase in pEMB1.
MICs of various β-lactam antibiotics against E. coli TOP10 with or without pEMB1
Characterization of orf5 and orf6 as a TA system.To verify the function of orf5 and orf6 as antitoxin and toxin genes in plasmid pEMB1, two knockout plasmids of orf5 and orf6, pEMB1ΔparD and pEMB1ΔparE, were constructed, respectively. pEMB1ΔparE was successfully transformed into competent E. coli TOP10 cells by electroporation, while no colonies were obtained in spite of repeated electroporation of pEMB1ΔparD, which is consistent with the hypothesis that these constitute a TA system (orf6 may be lethal in the absence of orf5). In search of confirmation that orf6 functions as a toxin gene, the putative toxin (orf6) gene was cloned into pET21b (pET21bparE). When the growth of recombinant E. coli BL21(DE3) cells harboring pET21bparE was tested on LB agar with or without IPTG, they grew well on LB agar without IPTG, while they did not grow on LB agar with IPTG (data not shown). The growth of E. coli cells harboring pET21bparE was also tested in LB broth (Fig. 4). Recombinant E. coli BL21(DE3) cells harboring pET21bparE grew well in LB broth without IPTG as host E. coli BL21(DE3) cells did, but their cell counts decreased rapidly by the addition of IPTG (Fig. 4B). These results indicated that E. coli cells were killed by the expression of orf6 gene and is fully consistent with its role as a toxin gene in a TA system.
Overexpression of the toxin (parE) gene of pEMB1 in E. coli. Growth and viability of recombinant E. coli BL21(DE3) harboring pET21bparE in LB broth with or without IPTG were monitored by determining the OD660 every 40 min (A) and the CFU every 80 min (B). Host E. coli BL21(DE3) without pET21bparE was also tested in parallel. Arrows indicate the time IPTG (1 mM) was added.
Stability assay of plasmids pEMB1 and pEMB1ΔparDE.To investigate the stability enhancement of plasmid pEMB1 by orf5 and orf6 as a TA system, a pEMB1 plasmid with orf5 and orf6 deleted (pEMB1ΔparDE) was constructed, and the stabilities of plasmids pEMB1 and pEMB1ΔparDE were compared by repeated cultivation of E. coli cells harboring pEMB1 and pEMB1ΔparDE in LB with or without ampicillin. Plasmid pEMB1 in E. coli was found to be very stable even in the absence of ampicillin, surviving for more than 500 bacterial generations In contrast, plasmid pEMB1ΔparDE in E. coli in the absence of ampicillin was rapidly lost after 250 bacterial generations (Fig. 5).
Plasmid stability assay. Plasmid stability was determined by replica plating onto LB agar with or without ampicillin (amp) after repeated cultivation of E. coli cells harboring pEMB1 and pEMB1ΔparDE in LB broth with or without ampicillin. Error bars represent standard deviations from the mean values.
Monitoring of pEMB1-like plasmids in the additional habitats.Besides Mt. Jeombong soil and Tancheon wastewater treatment plant (WWTP) sludge, the presence of pEMB1-like plasmids was investigated in samples from 20 other environments. Among these, pEMB1-like plasmids harboring a β-lactamase gene and a toxin (ParD)-antitoxin (ParE) system were identified in two other human-influenced sites (sludge samples) and two other remote mountain soil samples (see Fig. S2 in the supplemental material).
DISCUSSION
The global dissemination of antibiotic resistance genes is related to the selective pressure generated by the use of antibiotics in clinical and veterinary practices, and in animal feeds. This is supported by many studies showing correlations between the dissemination and concentrations of antibiotic resistance genes and the use of antibiotics (2, 8, 9). It is generally accepted that bacteria readily lose their antibiotic resistance genes in the absence of selective pressures because, in most cases, antibiotic resistance is associated with reduced bacterial fitness due to a decrease in growth rate (41, 42). However, some studies have shown that antibiotic resistance genes do not disappear from bacterial populations even after the antibiotic is no longer used (3, 43–46). To date, it is not known why antibiotic resistance genes in bacteria are so stable in the absence of antibiotics (47). It has been proposed that bacteria harboring antibiotic resistance genes may be more competitive in microbial communities because bacteria can be exposed to antibiotics intermittently in nature or that antibiotic resistance genes may play roles other than in antibiotic resistance in natural ecosystems (3, 41, 48). This suggests that bacteria may harbor maintenance mechanisms for antibiotic resistance genes in the absence of antibiotic selective pressure.
In the present study, a non-culture-based strategy was used to recover a variety of plasmids harboring ampicillin resistance genes in two quite distinct habitats, the Tancheon WWTP and Mt. Jeombong. The Tancheon WWTP receives sewage from private households, hospitals, and animal facilities, and previous research detected antibiotic residues at relatively high concentrations (49), which suggests that antibiotic resistance genes may be abundant in the Tancheon WWTP sludge. Congruent with this, three different plasmids harboring ampicillin resistance genes were identified from the Tancheon WWTP sludge sample (Fig. 1B), whereas only one kind of plasmid showing ampicillin resistance was identified from the remote Mt. Jeombong soil, where medical antibiotic exposure has been likely absent or minimal. Interestingly, the plasmid (pEMB1) from the remote Mt. Jeombong soil was indistinguishable from one of the three plasmids retrieved from the Tancheon WWTP sludge sample; pEMB1 harbored an ampicillin resistance gene, as well as a TA system. Typically, TA systems are identified by the primary sequence of toxin genes because toxin genes are relatively well conserved, whereas antitoxin genes are phylogenetically diverse (50). In plasmid pEMB1, the predicted protein product of orf6 was annotated as a plasmid stabilization system protein (ParE), while the translated protein of orf5 was annotated as a putative CopG family transcriptional regulator from its sequence (Table 2). Based on the annotation of orf6 as a toxin gene, we hypothesized that orf5 may function as an antitoxin (ParD) gene within a TA system. The impact of orf5 (Fig. 3) on plasmid retention was experimentally characterized (Fig. 5). It is known that ParE inhibits DNA gyrase and thereby blocks chromosomal DNA replication, whereas ParD exhibits inhibition or the reversal of the toxic activity of ParE and regulation of the parD-parE cluster (51, 52). In the future, genetic and biochemical investigations must be completed to verify that the protein products of orf5 and orf6 on pEMB1 function as a ParD-ParE TA system.
ParD-ParE toxin-antitoxin systems are commonly found in bacterial plasmids and chromosomes of Gammaproteobacteria and Alphaproteobacteria (53–55), and plasmids harboring both TA systems and antibiotic resistance genes are widely distributed in enterococci (24, 56). However, to the best of our knowledge, pEMB1 is the first plasmid harboring a ParD-ParE TA system and an ampicillin resistance gene identified from human-influenced (wastewater sludge) and pristine (remote mountain soil) environments. Moreover, PCR-based monitoring showed that pEMB1-like plasmids harboring a ParD-ParE TA system occurred more broadly in human-influenced, as well as remote, environmental samples (see Fig. S2 in the supplemental material). As shown in Fig. 5, even in the absence of ampicillin, there was no detectable plasmid loss from transformed E. coli cells over 500 generations of bacterial growth, which suggests that the ParD-ParE TA system may play an important role in maintaining plasmids stably in hosts in habitats without antibiotic selective pressures. The present study, therefore, adds to our understanding of the development of antibiotic-resistant bacteria and of the dissemination and persistence of antibiotic resistance genes in the environment.
ACKNOWLEDGMENTS
These efforts were supported by the National Research Foundation of Korea (grant 2013R1A2A2A07068946), funded by the Korean Government (MEST), Republic of Korea. E.L.M. was supported by NSF grant DEB-0841999.
FOOTNOTES
- Received 19 August 2014.
- Accepted 3 October 2014.
- Accepted manuscript posted online 10 October 2014.
Supplemental material for this article may be found at http://dx.doi.org/10.1128/AEM.02691-14.
- Copyright © 2015, American Society for Microbiology. All Rights Reserved.
