The Prevalence of Plasmid-Mediated Colistin Resistance Gene mcr-1 and Different Transferability and Fitness of mcr-1-Bearing IncX4 Plasmids in Escherichia coli from Pigeons

The emergence of plasmid-mediated colistin resistance gene mcr-1 incurs great concerns. Since the close proximity of pigeons with humans, it is significant to understand the prevalence and molecular characterization of mcr-1-positive isolates in pigeons, to provide a rationale for controlling its spread. ABSTRACT The prevalence of colistin-resistant bacteria limited the usage of colistin in the treatment of clinical multidrug-resistant Gram-negative bacterial infections. Here, we aimed to investigate the prevalence and molecular characterization of mcr-1-carrying isolates from pigeons close to humans following the ban on the use of colistin as an animal feed additive in China. Methods, including PCR, antimicrobial susceptibility testing, conjugation experiments, plasmid replicon typing, genome sequencing, bioinformatics analysis, measurement of growth curves, competition experiments, and plasmid stability assays were used to identify and characterize mcr-1-positive isolates. In total, 45 mcr-1-positive E. coli isolates were acquired from 100 fecal samples, and MICs of colistin ranged from 4 to 8 mg/L. The prevalence of mcr-1-positive E. coli isolates from pigeons was mainly mediated by IncX4 plasmids (39/45), including transferable mcr-1-bearing IncX4 plasmids with fitness advantage in 21 isolates, and nontransferable mcr-1-bearing IncX4 plasmids with fitness disadvantage in 18 isolates. There is a similar structure among the 6 mcr-1-bearing nontransferable IncX4 plasmids and 10 mcr-1-bearing transferable IncX4 plasmids in 16 E. coli isolates that have been sequenced. Plasmid transferability evaluation indicated that the same IncX4 plasmid has different transferability in different E. coli isolates. In conclusion, this study demonstrates that pigeons could act as potential reservoirs for the spread of mcr-1-positive E. coli in China. Transferability of IncX4 plasmids may be influenced by host chromosome in the same bacterial species. Additional research on the factors influencing the transferability of IncX4 plasmids in different bacterial hosts is required to help combat antimicrobial resistance. IMPORTANCE The emergence of plasmid-mediated colistin resistance gene mcr-1 incurs great concerns. Since the close proximity of pigeons with humans, it is significant to understand the prevalence and molecular characterization of mcr-1-positive isolates in pigeons, to provide a rationale for controlling its spread. Here, we found that the prevalence of mcr-1-positive E. coli from pigeons was mainly mediated by IncX4 plasmids. However, different transferability and fitness of mcr-1-bearing IncX4 plasmids in E. coli were observed, which demonstrated that transferability of IncX4 plasmids could be affected not only by genes on plasmids, but also by chromosomal factors in the same bacterial species. Our finding provided a new insight on studying the factors influencing the transferability of plasmids.

between humans and animals (3,4), the Chinese government banned colistin as an animal growth promoter on 1 May 2017 (5). Withdrawal of colistin from animal feed may contribute to the decline of mcr-1-positive isolates, but they are still prevalent in humans, animals, and the environment, posing a serious threat to public health (6).
It is known that a large number of pigeons are living in close proximity to humans and animals all around the world. Numerous studies have indicated that pigeon feces is a possible reservoir spreading antibiotic-resistant bacteria (7 to 9). The prevalence of mcr-1-carrying isolates from pigeons likely contributes to its prevalence in the environment, humans, and animals. Therefore, it is significant to understand the prevalence characteristics of mcr-1positive isolates in pigeons to provide a rationale for controlling its spread. Pigeon, harboring high nutritious value, is one important food animal in China (10). The prevalence of mcr-1positive isolates from pigeons before 2017 in China has been reported (11), but the prevalence and genomic features of the mcr-1-carrying isolates were not fully explored after the ban of colistin as a feed additive. In this study, we aim to investigate the prevalence and molecular characteristics of mcr-1-bearing isolates from pigeons close to humans after China banned the use of colistin as an animal feed additive.
Transferability of mcr-1 and mcr-1-associated plasmid types. To investigate the transmissibility of the mcr-1 gene, all mcr-1-harboring isolates were subjected to conjugation experiments. The genetic structures carrying the mcr-1 gene of 23 E. coli isolates with colistin resistance phenotypes were successfully transferred into recipient E. coli C600 or J53. To learn the location of mcr-1, PCR-based replicon typing (PBRT) was performed for all 45 mcr-1-positive isolates and 23 transconjugants obtained by conjugation assays. Replicon types of 45 mcr-1-positive isolates and 23 transconjugants are shown in Table 2. We confirmed that mcr-1 was located on IncX4-type plasmids in 39 isolates (Table 2) using specific primers IncX4-F and MCR1-RC-F targeting the IncX4type plasmid replicon and the mcr-1 gene with a product length of 2,854 bp (13). This indicated that mcr-1-bearing IncX4-type plasmid is the main vector for prevalence of mcr-1 in pigeons from Jiangsu. It is well known that mcr-1-harboring IncX4-type plasmid is one of the most prevalent conjugative plasmids worldwide (13 to 15). However, mcr-1-bearing IncX4-type plasmids in 18 out of 39 isolates were nontransferable in this study. In the remaining 6 isolates carrying no mcr-1-harboring IncX4-type plasmid, we found that mcr-1 is located on the IncI2 plasmids in two isolates, LP64-1 and LP71-1, by comparing the replicon types of parent isolates and their corresponding transconjugants ( Table 2). The mcr-1 in the other four isolates were nontransferable; therefore, we could not confirm the location of the mcr-1 by PBRT (Table 2).
Characterization of nontransferable mcr-1-bearing IncX4 plasmids. In total, the mcr-1-bearing IncX4 plasmids in 18 out of 39 isolates were nontransferable ( Table 2). Six isolates carrying nontransferable IncX4 plasmids were sequenced with the Illumina HiSeq 2500 platform, and they belonged to ST646 (n = 5) and ST6164 (n = 1). However, nontransferable IncX4 plasmids had a highly similar structure to transferable IncX4 plasmids, including type IV secretion systems that are responsible for plasmid conjugative transfer. Therefore, we speculated that it may be factors other than the mcr-1-bearing IncX4 plasmids that cause this type of plasmid to fail to transfer horizontally. According to the results of PBRT, we found that all isolates carrying nontransferable IncX4 plasmids had the same replicons (IncX4, IncFIB, and IncF), except the ST6164 E. coli LP87-1. It is possible that all the isolates except LP87-1, carrying nontransferable IncX4 plasmids, were the ST646 E. coli (Fig. 1a and Table 2). This indicated that the ST646 E. coli isolates could spread mcr-1-bearing IncX4 plasmids by clonal dissemination rather than horizontal transfer. In fact, the phenomenon that commonly conjugative plasmids cannot be transferred into the recipient appeared in other studies as well (14,19,20).
Fitness effects of mcr-1-bearing IncX4 plasmids. Isolates LP53-1 and LP63-1 carrying nontransferable IncX4 plasmids and isolates LP39-1 and LP51-1 carrying transferable IncX4 plasmids were selected to study the fitness of IncX4 plasmids. We first acquired the mcr-1-positive transconjugant CLP39-1 of LP39-1, and eliminated mcr-1-bearing IncX4 plasmids by the CRISPR-Cas9 system in four isolates to acquire strains LP53-1DIncX4, LP63-1DIncX4, LP39-1DIncX4, and LP51-1DIncX4. Then we assessed plasmid fitness effects by measurement of growth curves and competition assays. Growth curves indicated that the growth rate of LP39-1DIncX4 was slower than that of LP39-1 (Fig. 3a). No significant differences in growth rates were observed between LP51-1 and LP51-1DIncX4 (Fig. 3b), and between C600 and CLP39-1 (Fig. 3c). However, the growth rate of LP53-1 was slower than that of LP53-1DIncX4 (Fig. 3d). The growth rate of LP63-1 was slower than that of LP53-1DIncX4 in the logarithmic growth phase, and there was no obvious difference between the two after the logarithmic growth phase (Fig. 3e). Competition assays showed that LP39-1, LP51-1, and CLP39-1 were more competitive than LP39-1DIncX4, LP51-1DIncX4, and C600, but LP53-1 and LP63-1 were less competitive than LP53-1DIncX4 and LP63-1DIncX4 (Fig. 3f). The above results suggested that carriage of transferable mcr-1-bearing IncX4 plasmids improves host fitness, but carriage of nontransferable mcr-1-bearing IncX4 plasmids imposed a burden on the host. A previous report indicated that the mcr-1-carrying IncX4 plasmid increased fitness of E. coli DH5a (19), which was consistent with our findings on strains in which mcr-1-carrying IncX4 plasmid was transferable. In addition, the transferability may contribute to increased fitness of IncX4 plasmids on host bacteria (21). The transferable mcr-1-bearing IncX4 plasmids make a significant contribution to the prevalence of mcr-1 from the fitness perspective. We electroporated the nontransferable mcr-1-positive IncX4 plasmid of LP63-1 into LP39-1DIncX4 to acquire LP39-1DIncX4::IncX4, and found that the mcr-1-positive IncX4 plasmid could be transferred from LP39-1DIncX4::IncX4 into C600 again. We speculated that the transferability of these plasmids was related to genes of chromosome. Further study on the inhibitor to the transferability of IncX4 plasmids is required to contribute to the control of antimicrobial resistance. Eight isolates, including four isolates carrying transferable IncX4 plasmids and four isolates carrying nontransferable IncX4 The Prevalence of mcr-1 and Transferability of Plasmids in E. coli Microbiology Spectrum plasmids, were randomly selected to assess stability of mcr-1-bearing IncX4 plasmids. The results showed that the transferable IncX4 plasmids and nontransferable IncX4 plasmids remained relatively stable after long-term passages (Fig. 3g). It is possible that the fitness advantage of transferable mcr-1-bearing IncX4 plasmids and the stability of mcr-1-bearing IncX4 plasmids allow mcr-1 to remain prevalent without selection of colistin after colistin was banned as an animal feed additive.
Conclusions. To conclude, this study clearly illustrates that pigeons could act as reservoirs of mcr-1-positive E. coli in China. The prevalence of mcr-1-positive E. coli isolates from pigeons was mainly mediated by IncX4 plasmids, including transferable mcr-1-bearing IncX4 plasmids with fitness advantage in some E. coli isolates, and nontransferable mcr-1-bearing IncX4 plasmids with fitness disadvantage in some E. coli isolates. Plasmid transferability evaluation indicated that the same IncX4 plasmid has different transferability in different E. coli isolates. The transferability of these plasmids may be influenced by host chromosome in the same bacterial species. Additional research on the factors influencing the transferability of IncX4 plasmids in different bacterial hosts is required to help combat antimicrobial resistance. Continuous monitoring of mcr-1-positive E. coli from pigeons is necessary to understand its prevalence trends, and effective strategies to prevent such prevalence are urgently needed.
follows. First, we replaced the kanamycin resistance gene in pSGKP-km (Addgene plasmid ID: 117233) (39) with the rifampicin resistance gene arr-2 from the clinical isolate PK8215 (CP080122) to generate plasmid pSGKP-rif. Then the gene encoding the Cas9 nuclease was amplified from plasmid pCasKP-hph (Addgene plasmid ID: 117232) (39) and cloned into pSGKp-rif, creating the plasmid pSGKp-Cas9-rif. Finally, the araC gene and L-arabiinducible promoter araBAD were amplified from pCasKP-hph and subsequently inserted into the plasmid pSGKp-Cas9-rif as the promoter for Cas9, resulting in pCure-rif. The 20-nt base-pairing region of an sgRNA targeting the mcr-1 gene was designed through the online tool CHOPCHOP (http://chopchop.cbu.uib.no/). The paired oligonucleotides were randomly selected (mcr-spacer-F: tagtAAAGCTGTTTGATGTCACCG and mcrspacer-R: aaacCGGTGACATCAAACAGCTTT), then annealed and ligated to the BsaI-digested pCure-rif using T4 DNA ligase (NEB, USA), resulting in pCure-rif-mcr, and then electroporated into competent cells of isolates carrying mcr-1-bearing IncX4 plasmids, followed by selection on agar plates supplemented with 100 mg/L rifampicin. Overnight cultures of strains carrying pCure-rif-mcr were diluted in 1 mL LB broth containing 0.1% L-arabinose in combination with 100 mg/L rifampicin and incubated for 12 h at 37°C. Then, the culture was plated on LB agar plates, and colonies lacking mcr-1-bearing IncX4 plasmid were selected and confirmed by PCR. For curing plasmid pCure-rif-mcr, colonies lacking mcr-1-bearing IncX4 plasmid were streaked onto LB agar plates containing 5% sucrose at 37°C for 12 h.
Measurement of growth curves. Overnight cultures from single colonies were adjusted to the 0.5 McFarland standard. Then, 5 mL adjusted cultures were diluted into 5 mL LB broth and incubated at 37°C, 200 rpm for 12 h. Bacterial growth was monitored by measuring the OD620 every 1 h using Multuskan FC (Thermo Fisher Scientific) (40).
Competitive fitness. The relative competitive fitness of plasmid-carrying clones was determined in pairwise serial competition experiments with the isogenic plasmid-free strain. Briefly, overnight cultures of each competitor were adjusted to the 0.5 McFarland standard and mixed in a 1:1 ratio, and 0.05 mL mixed competitors were transferred into 5 mL fresh LB broth (day 0). After 24 h of growth at 37°C, 0.05 mL cultures were transferred into 5 mL fresh LB broth (day 1), and then two transfers were performed (days 2 and 3). The colony-forming unit (CFU) of each competitor were determined by plating serial dilutions on antibiotic-free LB plates and selective plates containing 2 mg/L colistin. log e ratio was calculated as follow: log e ratio = log e R(t)-log e R(0), where "R" represents the ratio of the CFU of plasmidbearing and plasmid-free cells in the competing cultures and "t" represents the time in days. If there is no difference in fitness between competing strains, the log e ratio will be considered 0. If plasmid carriage improves host fitness relative to that of plasmid-free strains, log e Ratio is positive, and it is negative if plasmid carriage reduces host fitness (40,41).
Plasmid stability. To learn the stability of mcr-1-bearing IncX4 plasmids, amounts of 5 mL overnight cultures from single colonies were transferred into 5 mL antibiotic-free LB broth on the first day. Then, serial transfers of 5 mL cultures to 5 mL fresh LB broth were performed every 12 h and passaged for 30 days (60 passages; ;600 generations). Cultures from 60 passages were serially diluted in 0.9% saline and plated onto LB plates without antibiotics and LB plates supplemented with colistin (2 mg/L). The stability frequency was calculated as follow: (CFU on LB plate containing colistin/CFU on antibiotic-free LB plate) Â 100% (40).
Data availability statement. The nucleotide sequences acquired in this study have been deposited under the NCBI BioProject with the accession number PRJNA861424.