Greater Invasion and Persistence of mcr-1-Bearing Plasmids in Escherichia coli than in Klebsiella pneumoniae

As infections caused by multidrug-resistant “superbugs” are increasing globally, polymyxins are often the only viable therapeutic option. Alarmingly, the wide spread of the plasmid-mediated polymyxin resistance gene mcr-1 is restricting the clinical utility of this last-line treatment option. ABSTRACT The emergence of the plasmid-borne polymyxin resistance gene mcr-1 threatens the clinical utility of last-line polymyxins. Although mcr-1 has disseminated to various Enterobacterales species, the prevalence of mcr-1 is the highest among Escherichia coli isolates while remaining low in Klebsiella pneumoniae. The reason for such a difference in prevalence has not been investigated. In this study, we examined and compared the biological characteristics of various mcr-1 plasmids in these two bacterial species. Although mcr-1-bearing plasmids were stably maintained in both E. coli and K. pneumoniae, the former presented itself to be superior by demonstrating a fitness advantage while carrying the plasmid. The inter- and intraspecies transferability efficiencies were evaluated for common mcr-1-harboring plasmids (IncX4, IncI2, IncHI2, IncP, and IncF types) with native E. coli and K. pneumoniae strains as donors. Here, we found that the conjugation frequencies of mcr-1 plasmids were significantly higher in E. coli than in K. pneumoniae, regardless of the donor species and Inc types of the mcr-1 plasmids. Plasmid invasion experiments revealed that mcr-1 plasmids displayed greater invasiveness and stability in E. coli than in K. pneumoniae. Moreover, K. pneumoniae carrying mcr-1 plasmids showed a competitive disadvantage when cocultured with E. coli. These findings indicate that mcr-1 plasmids could spread more easily among E. coli than among K. pneumoniae isolates and that mcr-1 plasmid-carrying E. coli has a competitive advantage over K. pneumoniae, leading to E. coli being the main mcr-1 reservoir. IMPORTANCE As infections caused by multidrug-resistant “superbugs” are increasing globally, polymyxins are often the only viable therapeutic option. Alarmingly, the wide spread of the plasmid-mediated polymyxin resistance gene mcr-1 is restricting the clinical utility of this last-line treatment option. With this, there is an urgent need to investigate the factors contributing to the spread and persistence of mcr-1-bearing plasmids in the bacterial community. Our research highlights that the higher prevalence of mcr-1 in E. coli than in K. pneumoniae is attributed to the greater transferability and persistence of mcr-1-bearing plasmid in the former species. By gaining these important insights into the persistence of mcr-1 in different bacterial species, we will be able to formulate effective strategies to curb the spread of mcr-1 and prolong the clinical life span of polymyxins.

are considered crucial nosocomial pathogens associated with high rates of antibiotic resistance (2,3), and carbapenem-resistant Enterobacterales (CRE) have been identified by the World Health Organization (WHO) as critical-priority pathogens that urgently need the development of new drugs (4). Due to limited treatment options, polymyxins (i.e., colistin and polymyxin B), belonging to an "old" class of antibiotics, were revived as last-resort agents against infections caused by CRE. Unfortunately, increasing frequencies of polymyxin-resistant bacteria are emerging (5); of note, the global dissemination of a plasmidmediated colistin resistance gene (mcr-1) is threatening the role of polymyxins as a last-line treatment (6,7).
The mcr-1 gene is carried mainly by plasmids such as those of the IncX4, IncI2, IncHI2, IncP, IncFII, and IncFIB plasmid types, and the first three plasmid types represent the most prevalent epidemic vectors (8,9). mcr-1-bearing plasmids have been spreading among various Gram-negative species (Escherichia coli, K. pneumoniae, Klebsiella oxytoca, Salmonella enterica, Cronobacter sakazakii, Aeromonas spp., Moraxella spp., and Enterobacter spp.) in diverse geographical locations (9,10). Moreover, epidemiological data have shown that the prevalence of mcr-1 is significantly higher in E. coli than in K. pneumoniae (11,12); however, the reason for such a difference remains unclear.
The persistence of resistance plasmids in a bacterial population is usually considered to be associated with (i) plasmid maintenance during bacterial replication, (ii) fitness costs to the host, or (iii) plasmid transmission via conjugation (13,14). The expression of mcr-1 has been reported to produce a toxic effect on bacteria and impose fitness costs on both E. coli and K. pneumoniae (15)(16)(17). Notwithstanding, the persistence of mcr-1-bearing plasmids in bacterial populations is promoted by the strict control of the plasmid copy number and effective plasmid conjugation (18,19). Besides, it has been suggested that the plasmid types in different bacterial species vary as a result of plasmid-host interactions (20,21). Therefore, we speculate that the significant differences in the prevalences of mcr-1 between E. coli and K. pneumoniae may be related to differences in the stabilities, conjugation efficiencies, and fitness costs of mcr-1-bearing plasmids in these two bacterial species. We thus compared the biological characteristics, including the fitness, transferability, and stability, of native mcr-1 plasmids in E. coli and K. pneumoniae. Our results reveal that mcr-1-bearing plasmids transfer more readily and are more stably maintained in E. coli than in K. pneumoniae.

RESULTS
Comparison of the stabilities of mcr-1-carrying plasmids in E. coli and K. pneumoniae. As mcr-1 has been commonly identified in IncX4-and IncI2-type plasmids (9, 10), four native plasmids harboring mcr-1, isolated from E. coli (pHNGDE4P170 [IncX4] and pHNSHP45 [IncI2]) and K. pneumoniae (pHNAHM7C25I [IncX4] and pHNBJ7H48 [IncI2]), were chosen for the stability assay in this study. These plasmids were introduced into E. coli C600 and K. pneumoniae ATCC 13883 by electroporation, and the transformants were used for the stability assay. All native plasmids were found to be stably maintained in E. coli C600 and K. pneumoniae ATCC 13883 for 35 days of passage in the absence of selection pressure (see Table S1 in the supplemental material). These results suggest that there is no difference in the stabilities of these native mcr-1-carrying plasmids among the E. coli and K. pneumoniae strains used in this study.
Comparison of the fitness costs of mcr-1-bearing plasmids in E. coli and K. pneumoniae. The biological costs of mcr-1-bearing IncX4-type plasmids (pHNGDE4P170 or pHNAHM7C25I) or IncI2-type plasmids (pHNSHP45 or pHNBJ7H48) in E. coli and K. pneumoniae strains were investigated in a competition assay with E. coli DH5a-GFP as the reference strain. All four E. coli strains carrying different mcr-1 plasmids demonstrated higher selection coefficients than the reference strain, suggesting that mcr-1-positive E. coli showed an increased fitness in comparison to the control strain (C600/pHNGDE4P170, P , 0.0001; C600/pHNAHM7C25I, P , 0.0001; C600/pHNSHP45, P , 0.0001; C600/pBJ7H48, P = 0.0137 [by one-way analysis of variance {ANOVA} with Benjamini-Hochberg correction]) (Fig. 1A). For K. pneumoniae ATCC 13883, a higher selection coefficient was noted for the strain carrying pHNGDE4P170, an IncX4-type mcr-1 plasmid (P = 0.0841 [by ANOVA with Benjamini-Hochberg correction]), while equivalent fitness was displayed by the strain with another  1B). Overall, mcr-1-bearing plasmids conferred a fitness advantage on the E. coli host, whereas a slight fitness cost was imposed on K. pneumoniae.
Comparison of the invasion and persistence of mcr-1-bearing plasmids in E. coli and K. pneumoniae. The invasion and persistence of mcr-1-bearing plasmids in E. coli and K. pneumoniae were examined by assessing the ability of mcr-1-carrying plasmids to invade plasmid-free populations. First, the parental strain (E. coli C600 or K. pneumoniae ATCC 13883) was cocultured with the respective mcr-1 plasmid-carrying strain individually, and the populations with/without the plasmid were detected every 24 h following passaging. For the individual cultures, the population of the plasmid-carrying strain was stably present for a duration of 120 h ( Fig. 3A to D). The competitive cocultures were constructed by coculturing an mcr-1 plasmid-carrying strain with a population of plasmid-free bacteria of the same strain or a strain of another species (Fig. 3E to H). When mcr-1 plasmid-carrying E. coli C600 was cocultured with plasmid-free E. coli C600 and K. pneumoniae ATCC 13883 RIF , the Invasion and Persistence of mcr-1-Bearing Plasmids Microbiology Spectrum population of K. pneumoniae ATCC 13883 RIF carrying an mcr-1 plasmid was detected at a very low count (,10 CFU/mL) at 24 h, which further increased to the highest point of ;10 2 to 10 3 CFU/mL at 48 h, indicating that plasmids carrying mcr-1 are capable of invading K. pneumoniae strains ( Fig. 3E and F). Thereafter, the number of K. pneumoniae bacteria  harboring mcr-1 decreased progressively ( Fig. 3E and F), indicating that K. pneumoniae ATCC 13883 RIF carrying an mcr-1 plasmid has a competitive disadvantage in the coculture environment. With mcr-1 plasmid-carrying K. pneumoniae ATCC 13883 RIF in the cultures containing plasmid-free K. pneumoniae ATCC 13883 RIF and E. coli C600, a rapid increase in the number of E. coli C600 bacteria harboring an mcr-1 plasmid was observed over the initial 48 h, reaching 10 4.5 CFU/mL (pHNGDE4P170) and 10 3.5 CFU/mL ( Fig. 3G and H). Importantly, the population of mcr-1-carrying E. coli C600 was maintained over the whole duration of 120 h ( Fig. 3G and H), even outnumbering mcr-1 plasmid-carrying K. pneumoniae ATCC 13883 RIF (Fig. 3G), indicating that mcr-1-positive E. coli exhibited a long-term competitive advantage in these competition environments. Together, these observations suggest that mcr-1 plasmids were persisting at a higher magnitude in E. coli than in K. pneumoniae, and the latter populations harboring mcr-1 exhibited a long-term competitive disadvantage, causing the population to gradually dwindle.

DISCUSSION
It is widely acknowledged that plasmids are extrachromosomal genetic elements that represent common vehicles for carrying antimicrobial resistance genes (22). Although antibiotic resistance plasmids benefit the host cell in the environment where the antibiotic is present, there could also be fitness costs associated with carrying the plasmid (23,24). In the absence of antibiotics, the fitness cost could outweigh the benefit of carrying the plasmid, leading to plasmid loss during cell division (25,26). To investigate the factors contributing to the different prevalences of mcr-1-bearing plasmids in E. coli and K. pneumoniae, the plasmid stabilities and plasmid-associated fitness costs in these two species were evaluated by using the dominant mcr-1 plasmids (IncX4 and IncI2 types). We found that mcr-1 plasmids of the IncX4 and IncI2 types were stably maintained in both E. coli and K. pneumoniae following 35 days of passage without colistin, suggesting that plasmid stability is not responsible for the different prevalences of mcr-1 in these two bacterial species. A further investigation was conducted on the fitness effect of carrying the IncX4-and IncI2-type mcr-1 plasmids in these two species. The results showed that mcr-1 plasmids of the IncX4 and IncI2 types conferred a fitness advantage on E. coli, while IncI2-type mcr-1 plasmids imposed a slight fitness cost on K. pneumoniae (Fig. 1). Similarly, an increase in fitness with the carriage of mcr-1 plasmids was also observed in a previous study with E. coli DH5a (27), and reduced biological fitness was observed with an mcr-1-carrying K. pneumoniae strain (17). MCR-1 usually imposes a fitness cost on host bacteria (15)(16)(17); however, we observed a slight fitness advantage for E. coli harboring mcr-1 plasmids. We propose that a higher plasmid conjugation transfer efficiency in E. coli competition cultures may overcome the fitness cost of mcr-1-bearing plasmids (19). Overall, these data suggest that the presence of mcr-1 can be beneficial for E. coli while being disadvantageous for K. pneumoniae, indicating that the acquisition of mcr-1 (or mcr-1 plasmids) comes with a fitness cost to the latter strain, reducing the competitive ability of mcr-1-carrying K. pneumoniae.
In addition to the above-mentioned plasmid stability and plasmid-associated fitness costs, conjugation efficiency is another key determinant affecting plasmid persistence in bacterial populations as plasmids are commonly acquired by bacteria via conjugation (13,28,29). With this, the frequency of conjugation of mcr-1 plasmids of the IncX4, IncI2, IncHI2, IncP, or IncF type (natively from E. coli and K. pneumoniae donors) to recipient strains of either E. coli or K. pneumoniae was investigated. We found that the conjugative efficiency of mcr-1 plasmids was dependent on the recipient species. E. coli as a recipient generally demonstrated a higher conjugation efficiency than K. pneumoniae, regardless of the donor species and the Inc types of the mcr-1 plasmids (Fig. 2). This finding suggests that mcr-1bearing plasmids could spread more readily among E. coli than among K. pneumoniae populations. It was previously reported that the conjugation of an extended-spectrumb-lactamase (ESBL)-carrying plasmid was affected by a combination of three factors (plasmid and donor and recipient strains of E. coli) (30). Differences in the genetic backgrounds of E. coli and K. pneumoniae may be responsible for the different conjugation rates, but this remains to be elucidated. Furthermore, in a coculture environment, E. coli also presented itself as an amazing recipient of the mcr-1 plasmid from K. pneumoniae, and the mcr-1 plasmid-carrying E. coli population was able to persist in the mixed environment ( Fig. 3G and H). The population of K. pneumoniae carrying the mcr-1 plasmid increased initially, indicating that K. pneumoniae was able to take up the mcr-1 plasmid from E. coli during the initial coincubation. However, it decreased progressively after 48 h ( Fig. 3E and F). Such a difference likely resulted from the higher frequency of conjugation of mcr-1-bearing plasmids in E. coli than in K. pneumoniae (Fig. 2). This opinion is supported by previous studies in which plasmids with a higher conjugation rate were able to invade bacterial populations more effectively (13,28,29). Also, a recent study by Yi et al. indicated that efficient conjugation was sufficient to overcome the fitness cost of mcr-1 carriage and enhance plasmid persistence in a bacterial community (19). Overall, these observations suggest that mcr-1 plasmids could invade E. coli more easily and be maintained in E. coli with higher stability than in K. pneumoniae (Fig. 3). Together with the fitness benefit mediated by mcr-1-bearing plasmids for E. coli while being disadvantageous for K. pneumoniae (Fig. 1), mcr-1 plasmid-carrying E. coli could have outcompeted K. pneumoniae in the bacterial community, thus contributing to the high prevalence of mcr-1 among E. coli species.
In conclusion, our study revealed that mcr-1-bearing plasmids were associated with a fitness advantage, higher conjugation efficiency, and greater persistence of E. coli. Altogether, these results confirm that effective conjugation and fitness benefits are crucial for the dissemination and maintenance of mcr-1 plasmids in bacteria (19).

MATERIALS AND METHODS
Strains and plasmids. Twenty-six mcr-1-positive Enterobacterales strains (E. coli, n = 19; K. pneumoniae, n = 7) obtained from various sources, including humans, swine, chickens, and fish, in China between 2014 and 2018 were selected for this study (Table 1) (6,27). The mcr-1 gene was carried by plasmids of different types (i.e., IncX4, IncI2, IncHI2, IncP, IncF29:A-:B-, and IncFIB) ( Table 1). mcr-1-negative K. pneumoniae ATCC 13883, E. coli ATCC 25922, and E. coli C600 were employed as recipient strains of mcr-1 plasmids. The pLac-eGFP plasmid, which contains an enhanced green fluorescent protein (eGFP) constitutive expression cassette, was transformed into E. coli DH5a, and the resulting strain was named E. coli DH5a-GFP. E. coli DH5a-GFP was used as a reference strain for the in vitro competition assay. Development of rifampicin resistance by serial passages. The rifampicin-resistant K. pneumoniae ATCC 13883 RIF strain was obtained by serial passaging of rifampicin-sensitive K. pneumoniae ATCC 13883 over a period of 8 days. Briefly, a culture of K. pneumoniae ATCC 13883 grown overnight was diluted to 10 6 CFU/mL, of which 100 mL was plated onto Luria-Bertani (LB) agar containing rifampicin (0.5Â MIC). Following a 24-h incubation at 37°C, viable cells were resuspended in fresh LB broth, and a 100-mL bacterial suspension containing 10 6 CFU/mL was plated onto LB agar containing rifampicin (1Â MIC). The concentration of rifampicin increased exponentially in serial passages. After obtaining rifampicin-resistant K. pneumoniae, the stability of rifampicin resistance in K. pneumoniae was further assessed by daily serial passages of the culture without antibiotics for 6 days. The resistance of K. pneumoniae ATCC 13883 RIF to rifampicin was stable (MIC of rifampicin of .512 mg/L) (see Table S2 in the supplemental material), and we then stored K. pneumoniae ATCC 13883 RIF in 25% glycerol stocks containing 500 mg/L rifampicin.
Antimicrobial susceptibility test. The MICs were determined using the broth microdilution method for streptomycin, rifampicin, and colistin according to Clinical and Laboratory Standards Institute (CLSI) guidelines (10). E. coli ATCC 25922 was used as the reference strain.
The stability of the mcr-1 plasmids in E. coli C600 and K. pneumoniae ATCC 13883 was evaluated as reported previously (31). Briefly, a single colony was selected and grown overnight in 3 mL LB broth at 37°C with shaking at 180 rpm. Serial passages were conducted daily by inoculating 3 mL of a culture grown overnight into 3 mL LB broth for a total duration of 35 days. Every 3 or 7 days, bacterial cultures were collected and plated onto antibiotic-free MacConkey agar. One hundred colonies were randomly selected for detecting the presence of mcr-1 plasmids by PCR. Plasmid retention was calculated as the percentage of mcr-1-carrying colonies among the 100 selected colonies. The assay was conducted in biological triplicates.
Competition experiments in vitro. The fitness of E. coli C600, K. pneumoniae ATCC 13883, and their corresponding transformants carrying the mcr-1 plasmid was measured in a competition assay against a reference strain, E. coli DH5a-GFP (17). Briefly, a single colony was selected and grown overnight in M9 minimal medium supplemented with the appropriate antibiotics. The culture grown overnight was diluted 1:100 in fresh LB broth. The experimental and reference strains were mixed at a ratio of 1:1, followed by a 24-h incubation at 37°C. At 0 and 24 h, the bacterial culture was collected, diluted in phosphate-buffered saline (PBS) at a 1:100 dilution, and analyzed using flow cytometry. The formula RF = [ln(E/R) t 2 ln(E/R) 0 ]/T was used to verify the relative fitness (RF), where E/R is the ratio of the tested experimental strain over the reference strain E. coli DH5a-GFP and T represents the generation. The assay was conducted in biological triplicates.
Bacterial conjugation assays. The plasmid conjugation efficiencies of mcr-1-carrying E. coli and K. pneumoniae strains were investigated as described previously, with minor modifications (32). Briefly, 26 native E. coli and K. pneumoniae strains carrying mcr-1 plasmids were used as the donor strains (Table 1), while E. coli C600 or K. pneumoniae ATCC 13883 RIF was employed as the recipient strain. Transconjugants of E. coli C600 were selected on LB agar supplemented with 2,000 mg/L streptomycin and 4 mg/L colistin, whereas transconjugants of K. pneumoniae ATCC 13883 RIF were selected on LB agar containing 200 mg/L rifampicin and 4 mg/L colistin. The presence of the mcr-1 plasmid was further confirmed by PCR. The conjugation frequency was calculated as the ratio of transconjugants over recipient cells.
Plasmid invasion assays. The plasmid invasion assay was performed as described previously, with some modifications (18). A single colony was inoculated into LB broth and incubated at 37°C overnight with shaking at 180 rpm. The culture of the plasmid-free strain (E. coli C600 or K. pneumoniae ATCC 13883 RIF ) grown overnight was diluted at a 1:100 ratio into 2 mL fresh LB broth and mixed with 1:100,000 dilutions of mcr-1 plasmid-carrying strains (E. coli C600/pHNGDE4P170, E. coli C600/pHNSHP45, K. pneumoniae ATCC 13883 RIF /pHNGDE4P170, or K. pneumoniae ATCC 13883 RIF /pHNSHP45) grown overnight. The mixed culture was incubated at 37°C with a low shaking speed (80 rpm). The mixed culture was passaged in fresh LB broth at a 1:100 dilution every 24 h for a duration of 120 h. Cultures were collected at the end of each 24-h passage (24,48,72,96, and 120 h) for the counting of viable cells using the appropriate antibiotic-containing LB agar (Table S3).
Statistical analyses. Statistical analysis was performed using Prism 9 and the IBM SPSS Statistics 21 program. One-way analysis of variance with Benjamini-Hochberg correction was used to determine the statistical differences in the relative fitnesses of bacteria among the five groups. The nonparametric Mann-Whitney U test was used for comparisons of differences in the conjugation frequencies of plasmids between two groups. The statistical significance threshold was set to an a value of 0.05.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. SUPPLEMENTAL FILE 1, PDF file, 0.2 MB.