Plasmid-encoded tet(X) genes that confer high-level tigecycline resistance in Escherichia coli.

Tigecycline is one of the last-resort antibiotics to treat complicated infections caused by both multidrug-resistant Gram-negative and Gram-positive bacteria1. Tigecycline resistance has sporadically occurred in recent years, primarily due to chromosome-encoding mechanisms, such as overexpression of efflux pumps and ribosome protection2,3. Here, we report the emergence of the plasmid-mediated mobile tigecycline resistance mechanism Tet(X4) in Escherichia coli isolates from China, which is capable of degrading all tetracyclines, including tigecycline and the US FDA newly approved eravacycline. The tet(X4)-harbouring IncQ1 plasmid is highly transferable, and can be successfully mobilized and stabilized in recipient clinical and laboratory strains of Enterobacteriaceae bacteria. It is noteworthy that tet(X4)-positive E. coli strains, including isolates co-harbouring mcr-1, have been widely detected in pigs, chickens, soil and dust samples in China. In vivo murine models demonstrated that the presence of Tet(X4) led to tigecycline treatment failure. Consequently, the emergence of plasmid-mediated Tet(X4) challenges the clinical efficacy of the entire family of tetracycline antibiotics. Importantly, our study raises concern that the plasmid-mediated tigecycline resistance may further spread into various ecological niches and into clinical high-risk pathogens. Collective efforts are in urgent need to preserve the potency of these essential antibiotics.

The emergence and spread of antimicrobial resistance in Enterobacteriaceae bacteria pose a serious threat to human and animal health. Of special concern is the emerging resistance to carbapenems, since these agents are often regarded as the last line of effective therapy for the treatment of infections caused by MDR Gram-negative bacteria[4]. Antibiotic treatment options for carbapenem-resistant Enterobacteriaceae (CRE) bacteria are becoming limited, and colistin and tigecycline have been regarded as the final armamentarium[5]. However, the recent discovery of a plasmid-mediated colistin resistance gene, mcr-1, threatens the clinical utility of colistin to treat these infections, leaving tigecycline as one of the last options[6].

Tigecycline resistance has inevitably emerged over the recent years, mostly identified among extensively drug- and carbapenem-resistant isolates[7,8]. Decreased susceptibility to tigecycline is commonly associated with overexpression of efflux pumps of the resistance-nodulation-cell division protein family, such as AcrAB-TolC, OqxAB, and AdeABC[9-11]. Meanwhile, mutations in plsC, rpsJ, trm, tet(A), and tet(M) have also been found to decrease tigecycline susceptibility[12-16]. In contrast, tetracycline destructases, such as Tet(X), represent a unique enzymatic tetracycline inactivation mechanism[17]. Tet(X), the flagship tetracycline-inactivating enzyme[18], has been confirmed for in vitro activity in degrading all tetracyclines including tigecycline[19]. Since first described in obligate anaerobes Bacteroides fragilis, tet(X) and its orthologous genes [namely tet(X1) and tet(X2)] have now been detected in clinical Enterobacteriaceae and Acinetobacter baumannii isolates, but their distribution, genetic structure, and clinical significance remain to be explored[20,21]. In this study, we described a plasmid-mediated mobile tigecycline resistance gene tet(X4), and explored the prevalence of tet(X4)-positive E. coli strains in humans, food-producing animals, and surrounding environment in China (Fig. 1).

Figure 1 |

Map of tet(X4) sampling areas in China.

The tet(X4) sampling areas of pig farms, chicken houses, and hospitals (A, B, C, and D) are denoted with pink, green, and blue circles, respectively. The distribution of tet(X4)-positive E. coli strains in China is shaded in light blue.

Initially, a tigecycline-resistant E. coli strain LHM10–1 (8 mg/L; Supplementary Table 1) was isolated from the stool sample in an intensive pig farm (Jiangxi, China) in July 2017. The tigecycline resistance could be successfully transferred into the recipient E. coli C600 (8 mg/L; Supplementary Table 1), suggesting that the gene encoding tigecycline resistance might be plasmid-borne. WGS of E. coli LHM10–1 showed that it belonged to sequence type (ST) 515, and contained one 4.81-Mb chromosome and six plasmids with sizes of 108,366 bp (pLHM10–1-p1, IncF), 98,254 bp (pLHM10–1-p2, untypeable), 33,301 bp (pLHM10–1-p3, IncX4), 28,942 bp (pLHM10–1-p4, untypeable), 28,519 bp (pLHM10–1-p5, untypeable), and 12,783 bp (pLHM10–1-p6, IncQ1).

A tet(X)-like gene [designated as tet(X4)], with 1,158 bp in length, was found to be located on the IncQ1 plasmid pLHM10–1-p6. The tet(X4) gene encoded a 385 amino acid protein (Fig. 2a; Supplementary Fig. 1a), which showed 94.5% amino acid sequence identity to Tet(X), initially found in the obligate anaerobe B. fragilis (GenBank accession number M37699). Meanwhile, Tet(X4) displayed 63.9%, 95.1%, and 85.7% identities to the other named Tet(X) orthologs Tet(X1) (AJ311171), Tet(X2) (AJ311171), and Tet(X3) (AB097942), respectively (Supplementary Fig. 1a).

Figure 2 |

The activity of Tet(X4) on tetracyclines in vitro and in vivo.

a, Homology modelling of the tetracyclines-inactivating protein Tet(X4). The modelling of Tet(X4) is based on published tigecycline/Tet(X) complex (PDB accession number 4A6N[40]) using the online server Phyre2 (ref. 41) and AutoDock version 4.2.6 (ref. 42). The substrate binding domain (light green), FAD binding domain (pink), C-terminal helix (light blue), and tigecycline (orange) are displayed. The C-terminal and N-terminal are marked in black characters. b, Microbiological degradation assays. The activity of tet(X4) on tetracycline antibiotics degradation is evaluated by measuring the changes of inhibition zones after the addition of supernatant from different cocultures. The experiments are performed in triplicate and repeated three times with similar results. Blank, the supernatant without treatment (only containing tetracyclines) is added; X+, the supernatant from the coculture incubated with the tet(X4) construct, namely E. coli JM109+pBAD24-tet(X4), is added; X-, the supernatant from the coculture incubated with E. coli JM109 containing the empty vector pBAD24 alone is added; TC, tetracycline; CTC, chlortetracycline; OTC, oxytetracycline; DOX, doxycycline; MIN, minocycline; TGC, tigecycline; ERA, eravacycline. c-d, The levels of tetracycline (c) and eravacycline (d) degradation by Tet(X4). Statistical analysis is conducted using unpaired and two sided t test. Individual values of biological replicates (n=6) are shown as dots, while the means (middle lines) and standard deviations are displayed as error bars. Theoretical max indicates the initial tetracycline or eravacycline concentration in the medium prior to inoculation. e, In vivo effects of tigecycline treatment (50 mg/kg per 24 hours) in a murine thigh model. P value is calculated by a two-way ANOVA test. Individual values of animals (n=6) are shown as dots, while the means (middle lines) and standard deviations are displayed as error bars.

Gene cloning experiment showed that the tet(X4) construct, namely E. coli JM109+pBAD24-tet(X4), demonstrated a 64-fold increase in the minimum inhibitory concentration (MIC) of tigecycline when compared with the host E. coli JM109 strain carrying pBAD24 (0.25 mg/L; Supplementary Table 1). Similarly, MICs of the first- and second-generation tetracyclines were increased at least 32-fold. In addition, MIC of the FDA newly approved eravacycline, which was structurally similar to tigecycline with two modifications at C-7 and C-9, increased 512-fold (4 mg/L; Supplementary Table 1). In addition, the agar well diffusion assay showed that the addition of supernatant from E. coli JM109+pBAD24-tet(X4) resulted in the elimination of tetracyclines inhibition zone, while there were no significant changes between the negative and blank controls, suggestive of inactivation of tetracyclines by Tet(X4) (Fig. 2b). Furthermore, LC-MS/MS showed that the level of tetracycline remained (16.1 ± 1.2)% after incubation with E. coli JM109+pBAD24-tet(X4) for 16 hours [verse (80.2 ± 1.2)% in the negative control] (Fig. 2c; P <0.0001), and eravacycline remained only (7.8 ± 0.5)% after incubation for the same time [verse (79.0 ± 3.8)% in the negative control] (Fig. 2d; P <0.0001). Our above results demonstrated that Tet(X4) was able to confer resistance to the entire family of tetracyclines.

In E. coli LHM10–1, the tet(X4)-harboring plasmid pLHM10–1-p6 was 12,783 bp in size with an average G+C content of 55.4% and contained 15 predicted ORFs, of which tet(X4) was adjacent to two copies of ISCR2 (one is truncated; Fig. 3a). Plasmid replicon analysis showed that pLHM10–1-p6 belonged to the broad host-range IncQ1 group, possessing the IncQ1 plasmid backbone genes for plasmid replication (repA, repB, and repC) and mobilization (mobA, mobB, and mobC). A comparison of several IncQ1 plasmids from the public National Center for Biotechnology Information (NCBI) database (https://www.ncbi.nlm.nih.gov/) demonstrated that they shared similar replication and mobilization regions, but diverse acquired regions (between repC and mobC) (Fig. 3b). The results suggested that pLHM10–1-p6 originated by acquisition of tet(X4) into an IncQ1 plasmid, likely associated with the translocation of ISCR2.

Figure 3 |

Characteristics of the IncQ1 tet(X4)-harboring plasmid pLHM10–1-p6.

a, Structure of the index plasmid pLHM10–1-p6. GC skew and GC content are indicated from the inside out. Positions and transcriptional directions of the predicted ORFs are denoted with arrows. Genes associated with the plasmid replication, antimicrobial resistance, heavy metal resistance, mobile element, and conjugative transfer are highlighted in green, pink, dark yellow, blue, and cyan, respectively. Other genes are marked as grey arrows. b, Linear comparison of the representative IncQ1 plasmid sequences. Results of sequence alignment are generated with Easyfig version 2.1 (ref. 43). The arrows represent the position and transcriptional direction of the ORFs. Regions of homology between 73% and 100% are marked by grey shading. c, Conjugation transfer efficiencies of the index plasmid pLHM10–1-p6 into E. coli, K. pneumoniae, and S. Typhimurium strains. Transfer efficiency is calculated based on colony counts of the transconjugant and recipient cells in triplicate, and all data points are displayed, along with mean and standard deviation, respectively. d, Plasmid stability experiment results. All experiments are conducted in triplicate. Error bars denote the means (middle lines) and standard deviations.

Interestingly, a mcr-1-carrying IncX4 plasmid pLHM10–1-p3 (33,301 bp) was identified in the same E. coli LHM10–1, which shared >99% sequence identity and 100% query coverage to several other mcr-1-harboring plasmids (e.g., KX084392 and CP018773) from animal and human sources in China, USA, and some other countries. In addition, an extended-spectrum β-lactamase (ESBL) gene blaCTX-M-14 was identified on the chromosome of E. coli LHM10–1.

Further conjugation assays showed that tigecycline resistance could be successfully transferred from E. coli LHM10–1 into laboratory strains E. coli C600, S. Typhimurium ATCC 14028, and K. pneumoniae ATCC 700603, at a frequency of (1.2 ± 0.5)×10−2, (1.2 ± 0.3)×10−4, and (6.2 ± 1.3)×10−2 cells per recipient cell, respectively (Fig. 3c). Notably, pLHM10–1-p6 also exhibited good transferability into clinical CRE strains, including NDM-5-producing ST167 E. coli 1314 and KPC-2-producing ST11 K. pneumoniae 1332, at the frequency of (2.2 ± 0.5)×10−4 and (8.0 ± 0.5)×10−6, respectively (Fig. 3c). MICs of tetracyclines, including tigecycline and eravacycline, among these transconjugants were increased at least 8-fold (Supplementary Table 1). After 22 days (approximately 220 generations) of serial passage without antibiotic treatment, pLHM10–1-p6 was still stably maintained (75%−100% fraction of plasmid-containing lineages) in the parental E. coli LHM10–1 and transconjugants hosts, including the clinical CRE strains (Fig. 3d). Given the high transferability and stability of tet(X4) in clinical CRE isolates harboring blaKPC or blaNDM, further acquisition of carbapenem resistance genes by tet(X4)-harboring strains, or alternatively, the spread of tet(X4)-carrying plasmids into clinical MDR strains, had the potential to create truly pan-drug resistant strains, resulting in untreatable infections.

In this study, the tet(X4) gene was totally detected in 42 E. coli strains [sharing 100% nucleotide sequence identity of tet(X4) with each other] from 4,189 samples, consisting of 30 (1.3%) from 2,337 pigs, 8 (0.8%) from 1,061 chickens, 2 (0.8%) from 256 soil samples, and 2 (0.9%) from 232 dust samples (Supplementary Table 2). These tet(X4)-positive E. coli isolates were found in five provinces in Southern and Eastern China (Fig. 1), including Guangdong (n=12), Fujian (n=11), Jiangsu (n=8), Jiangxi (n=6), and Guangxi (n=5). However, the tet(X4)-positive E. coli strains weren’t detected in others.

Although we have not identified tet(X4)-positive E. coli strains from human sources, a series of bacterial genomes carrying tet(X4)-like genes from human (e.g., CP023968), animal (e.g., CP002562), water (e.g., CP023049), and soil (e.g., LT906465) sources were identified in NCBI (Supplementary Fig. 1b). Furthermore, the sequence analysis also indicated that the tet(X4)-like genes had spread outside China and into other countries, including USA (e.g., NZ_JYQO01000042), UK (e.g., CP000685), and Thailand (e.g., NZ_JRQZ01000088) (Supplementary Fig. 1b).

The susceptibility testing demonstrated that the 42 tet(X4)-positive E. coli strains were all resistant to tigecycline, tetracycline, sulfamethoxazole/trimethoprim, and florfenicol, while 15 (35.7%), 12 (28.6%), 10 (23.8%), and 9 (21.4%) isolates showed resistance to ciprofloxacin, cefotaxime, gentamicin, and colistin, respectively (Fig. 4). In addition, all isolates exhibited an eravacycline MIC of 4–8 mg/L, significantly higher than the MIC90 (0.5 mg/L) for E. coli strains from the routine surveillance[22]. Notably, the 12 cefotaxime-resistant isolates were all found to carry ESBL genes, including blaCTX-M-14 (n=7), blaCTX-M-65 (n=4), and blaCTX-M-27 (n=1). The nine colistin-resistant E. coli strains all carried colistin resistance gene mcr-1 and ESBL gene blaCTX-M (blaCTX-M-14 or blaCTX-M-65), belonging to two STs (ST515 [n=5] and ST1196 [n=4]).

Figure 4 |

XbaI-PFGE dendrogram and details about tet(X4)-positive E. coli isolates.

The PFGE assay is conducted successfully for once according to the standard protocol, and then used for the following analysis. The full gel images have also been provided in the supplementary material. PFGE patterns with a cutoff at 85% similarity (the dotted line) are considered to the same PFGE cluster, and indicated as groups A-K, respectively. E. coli strains carrying pLHM10–1-p6-like plasmid are underlined. aThese strains are isolated from environmental samples in pig farms. bTGC, tigecycline; TC, tetracycline; SXT, sulfamethoxazole/trimethoprim; FFC, florfenicol; CTX, cefotaxime; CS, colistin; CIP, ciprofloxacin; GEN, gentamicin.

MLST showed that the 42 tet(X4)-positive E. coli strains belonged to nine distinct STs, which were in general consistent with the pulsotypes (A-K) from the PFGE analysis (Fig. 4). Strains from the same ST or PFGE pulsotype usually shared similar antimicrobial resistance profiles. Two STs (ST8302 and ST8338) were first identified in the current study, with ST8302 the most predominant ST (n=16; 38.1%) overall. Notably, the same STs of strains (ST8302, ST515, ST761, and ST10) in this study were identified in different geological regions and sources, suggesting the possibility of clonal spread of the same tet(X4)-positive strains across different regions.

The detection of IncQ1 pLHM10–1-p6-like plasmid revealed that 57.1% (24/42) of the tet(X4)-positive E. coli isolates harbored the same pLHM10–1-p6-like plasmid (Fig. 4). These strains belonged to four distinct STs (ST8302 [n=16], ST515 [n=6], ST761 [n=1], and ST871 [n=1]), and were isolated from four different provinces, including Guangdong (n=7), Jiangsu (n=7), Jiangxi (n=6), and Fujian (n=4). Most of them were collected from pigs (n=14) and chickens (n=7), followed by dusts (n=2) and soils (n=1).

Notably, IncQ-type plasmids are a group of relatively small (approximately 5 to 15 kb), non-conjugative but mobilizable plasmids that can be transferred into a broad range of bacterial hosts in the presence of self-transmissible helper plasmids[23]. Our mating experiment results showed that pLHM10–1-p6 could be successfully transferred in the presence of mcr-1-carrying IncX4 plasmid pZ6–2 (transfer frequency, (5.4±1.8)×10−2), IncI2 plasmid pHLJ179 ((7.9±1.5)×10–5) and IncFII plasmid pH12–4 ((4.3±1.0)×10−5) (Supplementary Table 3), but the transfer was failed in the absence of helper plasmids (pLHM10–1-p6 alone). Co-transfer of the helper plasmids along with pLHM10–1-p6 into the recipient E. coli C600 wasn’t observed under the selection by tigecycline. Our results indicated that a helper plasmid is needed for the mobilization of tet(X4)-harboring pLHM10–1-p6-like plasmids, and importantly, common mcr-1-bearing plasmids may serve as helper plasmids to further facilitate the spread of plasmid-mediated tigecycline resistance.

Moreover, WGS of another colistin-resistant ST1196 strain E. coli G3X16–2 showed that it consisted of one 4.78-Mb chromosome and two plasmids in the sizes of 265,575 bp (pG3X16–2-p1) and 138,950 bp (pG3X16–2-p2), respectively. Plasmid pG3X16–2-p1 belonged to IncHI2 group, and harbored mcr-1 and blaCTX-M-65. The tet(X4) was located on an IncF plasmid (F-:A18:B-), pG3X16–2-p2, co-harboring aadA1, aadA2, aph(3”)-Ib, aph(6)-Id, blaTEM-1B, erm(42), cmlA1, floR, sul3, tet(A), tet(M), and dfrA12 (Supplementary Fig. 2), which suggested that the tigecycline resistance may be co-selected by other antimicrobial resistances. Intriguingly, pG3X16–2-p2 harbored four tandem repeated ISCR2-catD-tet(X4)-ISCR2 gene cassettes on the same plasmid, suggesting the tet(X4)-harboring genetic elements are highly active and may further transfer to other plasmids or chromosome.

Results of in vivo experiment showed that the mice infected with plasmid-cured isogeneic strain E. coli STB20–1S [tet(X4)-negative] were highly susceptible to tigecycline treatment, with more than two log10(CFU/g) reduction in E. coli counts, while tigecycline had no significant effect on animals infected by E. coli STB20–1 after 24 hours of treatment (Fig. 2e; P <0.0001). The results indicated that the presence of tet(X4) may contribute to tigecycline treatment failure.

In this study, the tet(X4)-positive E. coli strains were identified in pigs, chickens, and their surrounding environmental samples from different geographical regions, indicating that Tet(X4)-encoding tigecycline resistance might have already spread in China. We suspected that the usage of “older” generations of tetracyclines had largely contributed to the emergence of newer mobile tigecycline resistance. In China, tigecycline is only approved for the treatment of clinical infections in human whereas its usage in food-producing animals is not authorized. However, the first- and second-generation tetracyclines were widely used in the treatment of infections in food-producing animals and for growth promotion at sub-therapeutic doses in China, likely providing a selective pressure for the emergence of tigecycline resistance[24]. As one of the countries with the largest amount of antibiotics usage in the world, 12,000 tons (7.4%) of tetracyclines were consumed by China yearly (data from in 2013), and most of them eventually entered into environment[25]. Similarly in Europe, although the European Union (EU) had phased out the use of antibiotics as growth promoters, the latest ECDC/EFSA/EMA joint report (2017) on the integrated analysis of antibiotic consumption and antimicrobial resistance in bacteria revealed that tetracyclines were still the most sold antibiotics in Europe, and a statistically significant positive association between tetracycline consumption and resistance in E. coli strains was observed in food-producing animals[26]. Of concern, the selective pressure imposed by the continuous application of tetracyclines in veterinary practice could serve to maintain and spread the tet(X4)-like genes into pathogenic microorganisms.

The mobile antimicrobial resistance mechanisms, including Tet(X4), MCR, and different families of carbapenemase (e.g., NDM), represent a plasmid-mediated antibiotic crisis. But the reality is that only limited antibiotics on the market or in the pipeline are effective against these highly resistant organisms. Continuous selection pressure from the environment, veterinary usage, and clinical practice, will likely accelerate the spread of tet(X4) and other resistance genes. The horizontal transfer ability of the resistance genes and plasmids, such as the tet(X4)-harboring broad-host range IncQ1 plasmid, can be a challenge for hospital infection control and public health surveillance. Our study underscores the need for a “one-health” strategy (i.e., considering the span of humans, animals, and their environment) for antimicrobial resistance, as infectious disease surveillance and containment across different sectors is paramount. To make this last-resort drug available, the prevalence of tet(X4) in both human and veterinary fields should be considered by surveillance programs, along with judicious administration of tetracyclines in animals and the environments.

Methods

Sample collection and bacterial strains

This study was approved by the institutional review board (IRB) of hospital A (Huizhou Municipal Central Hospital) and South China Agricultural University (SCAU). The clinical isolates from hospitals B, C, and D were retrospectively collected, and patient data were not included in this study, therefore ethical approval wasn’t sought and obtained. The fecal samples were randomly collected from pigs (3 to 8 months old) and chickens (7 to 20 weeks old), with approximately 50 samples per farm. If possible, the soil, dust, sewage, and vegetable samples were also collected at least three samples per farm.

In order to examine the spread of mobile tigecycline resistance gene tet(X4), samples and clinical isolates were screened from four tertiary care hospitals and 63 intensive animal farms (pig farms, n=41; chicken houses, n=22) in 16 provinces or municipalities in China (Fig. 1). The four hospitals are located in Guangdong (hospital A and B), Jiangsu (hospital C), and Zhejiang (hospital D) provinces, respectively. Pig farms and chicken houses are distributed in four geographic regions of China (northeast, southeast, west, and central China) and cover most livestock breeding areas in China. Briefly, a total of 4189 samples from the animal sector were collected from May 2015 to May 2018 (Supplementary Table 2) and subjected to selection on MacConkey agar (MAC) plates containing tigecycline (2 mg/L). The selected tigecycline-nonsusceptible isolates were then subjected to tet(X4) screening. In addition, 299 human specimens from hospital A collected in the same period and a random collection of 720 E. coli clinical isolates from hospitals B, C, and D were also screened as described above (Supplementary Table 2).

Porcine E. coli LHM10–1 was selected as the donor strain while laboratory strains (E. coli C600 [streptomycin-resistance], Salmonella Typhimurium ATCC 14028 [rifampin-resistance], and K. pneumoniae ATCC 700603 [rifampin-resistance]), and clinical MDR strains (E. coli 1314 carrying blaNDM-5 [ST167] and K. pneumoniae 1332 carrying blaKPC-2 [ST11]) were chosen as recipient strains for conjugation and plasmid stability tests. Porcine E. coli STB20–1 carrying a single plasmid [tet(X4)-positive] was selected for the in vivo infection model.

Whole genome sequencing (WGS)

Genomic DNA was extracted from E. coli LHM10–1 using a TIANamp Bacteria DNA kit (Tiangen, Beijing, China), followed by genomic DNA sequencing with a MiSeq platform (Huayin Health, Guangzhou, China). To obtain the complete plasmid sequences, E. coli LHM10–1 was also subjected to single-molecule real-time (SMRT) sequencing with a PacBio RSII system (Nextomics, Wuhan, China). In addition, another tet(X4)-positive E. coli strain G3X16–2 was completely sequenced using a combined Illumina HiSeq and Nanopore sequencing approach (Nextomics, Wuhan, China). Genome assembly was performed with Unicycler version 0.4.1 (ref. 27) using a combination of short and long reads, followed by error correction by Pilon version 1.12 (ref. 28). The functional genes were annotated and classified using the RAST server[29]. The sequence types and plasmid replicon types were analyzed by the CGE server (https://cge.cbs.dtu.dk/services/). Visual representation of the genetic structure of tet(X4)-carrying plasmids was generated with DNAPlotter version 1.11 (ref. 30).

Subcloning experiments

To confirm the role of Tet(X4) in tigecycline resistance, the full-length open reading frame (ORF) of tet(X4) was amplified from genomic DNA of E. coli LHM10–1 with addition of an EcoRI restriction site at the 5′-end and a SalI restriction site at the 3′-end (Supplementary Table 4). Both the purified PCR product and the plasmid vector pBAD24 containing the arabinose pBAD promoter were digested with restriction endonucleases EcoRI/SalI and then ligated at 16 °C overnight following the manufacturer’s instructions (New England Biolabs, Ipswich, MA). Subsequently, the recombinant plasmid pBAD24-tet(X4) was transformed into competent E. coli JM109 by heat shock and then subjected to susceptibility testing against different tetracyclines using a broth microdilution method.

Microbiological degradation assays

The activities of Tet(X4) on tetracycline antibiotics degradation were initially assessed by an agar well diffusion method[31]. Firstly, a suspension of the E. coli construct JM109+pBAD24-tet(X4) was made by suspending the culture from a Mueller-Hinton (MH) agar plate in 500 μl of MH broth containing 0.1% L-arabinose (OD600 adjusted as 2.0). The suspension was then inoculated with tetracycline (10 mg/L), oxytetracycline (20 mg/L), chlortetracycline (20 mg/L), doxycycline (10 mg/L), minocycline (10 mg/L), tigecycline (2.5 mg/L), and eravacycline (2.5 mg/L), respectively, and cultured at 37 °C for 8 hours. Meanwhile, 100 μl of the overnight culture (0.5 McFarland) of Bacillus stearothermophilus 7953 (tetracyclines-susceptible) was spread on a MH agar plate surface. A hole with 6 mm diameter was punched aseptically with a disposable hole puncher, and its bottom was sealed by a drop of heated liquid MH agar. Following centrifugation and filtration with a 0.22 μm filter, 20 μl of the supernatant from the JM109+pBAD24-tet(X4) coculture with different tetracycline antibiotics was added into the prepared agar hole and the plate was incubated at 60 °C for 16 hours to examine the zone of inhibition. The E. coli JM109+pBAD24 strain was used as a negative control and the medium containing only tetracyclines was acted as the blank control.

Quantifying the levels of tetracyclines via liquid chromatography-tandem mass spectrometry (LC-MS/MS)

To further examine if Tet(X4) was able to inactivate tetracyclines, the levels of tetracycline and eravacycline (tigecycline wasn’t used because of its poor stability in growth media, even with the protection of 0.6 g/L ascorbic acid as previously described[32]) were quantified in growth media of E. coli JM109+pBAD24-tet(X4) by LC-MS/MS in sextuplicate, respectively[17]. For tetracycline quantification, E. coli JM109+pBAD24-tet(X4) was grown in 4 ml of M9 minimal media (1×M9 minimal salts, 2 mM MgSO4, and 100 μM CaCl2) containing 0.1% L-arabinose supplemented with 9 g/L glucose, 100 mg/L thiamine, 100 mg/L leucine, and 8 mg/L tetracycline. The strain was incubated at 37 °C with shaking at 200 rpm and protected from light for 16 hours. Subsequently, cells were pelleted, and the supernatant was passed through a 0.22 μm filter. The filtered supernatant was then diluted 10-fold and subjected to LC-MS/MS quantification. Similarly, the level of eravacycline degradation was also determined by incubation with eravacycline (2 mg/L) for 16 hours. The E. coli JM109+pBAD24 strain was used as a negative control. The linear ranges of standard curves for tetracycline and eravacycline quantification were 25 to 800 ppb and 10 to 200 ppb, respectively, with r2 values >0.995.

Mating experiments and plasmid stability testing

The transferability of tet(X4) in E. coli LHM10–1 was determined by filter mating using E. coli C600, S. Typhimurium ATCC 14028, K. pneumoniae ATCC 700603, E. coli 1314, and K. pneumoniae 1332 as recipient strains. Briefly, the donor and recipient strains were grown at log-phase in Luria-Bertani (LB) broth, and were then mixed at the donor to recipient ratio of 1:3 and applied to a 0.22 μm filter, followed by culture at 37 °C for 16 hours. The putative transconjugants were selected as follows. For E. coli C600 and S. Typhimurium ATCC 14028, they were selected by MAC plates containing tigecycline (4 mg/L) in combined with streptomycin (1000 mg/L) or rifampin (100 mg/L). 8 mg/L of tigecycline together with rifampin (100 mg/L) was used for the selection of the putative transconjugants in K. pneumoniae ATCC 700603, which has an MIC 4 mg/L for tigecycline. For E. coli 1314 and 1332, they were selected by MAC plates containing tigecycline (4 mg/L) and meropenem (4 mg/L). Transconjugants were further confirmed by PCR screening for tet(X4), Enterobacterial Repetitive Intergenic Consensus PCR (ERIC-PCR), and 16S rRNA gene sequencing with the primers in appendix (Supplementary Table 4). Transfer efficiency was calculated based on colony counts of the transconjugant and recipient cells in triplicate as previously reported[33].

Subsequently, the host E. coli LHM10–1 strain and its transconjugants were subjected to plasmid stability testing as previously described[34]. Briefly, three independent lineages of them were cultured overnight at 37 °C in 4 ml of antibiotic-free LB broth, respectively. Serial passaging of 4 μl of the overnight culture to 4 ml of LB broth was performed daily, yielding approximately 10 generations of growth per passage. For every 20 generations, samples were diluted and plated on LB agar plates. Then, 50 colonies from each lineage were screened on LB agar plates with or without tigecycline (4 mg/L) to determine the fraction of plasmid-containing cells. 8 μg/ml of tigecycline was used for the transconjugant generated with K. pneumoniae ATCC 700603. The loss of plasmids was confirmed by the tet(X4) target PCR and ERIC-PCR as mentioned above.

In addition, the mobilizing efficiency of different helper plasmids on IncQ1 plasmid pLHM10–1-p6 was determined by filter mating in triplicate. Firstly, donor strains containing pLHM10–1-p6 and a single putative helper plasmid were constructed. In brief, the IncQ1 plasmid pLHM10–1-p6 was transferred into E. coli DH5α by heat shock and the putative transformant (namely E. coli DH5α+ pLHM10–1-p6) was selected by LB agar plates containing 4 mg/L of tigecycline. Then, three laboratory E. coli strains Z6–2, H12–4, and HLJ179, which have been confirmed to carry mcr-1-positive lncX4, lncFII, and I2 plasmids, respectively, were used for filter mating with the recipient strain E. coli DH5α+ pLHM10–1-p6. The putative constructs were selected by LB agar plates containing tigecycline (4 mg/L) and colistin (2 mg/L). PCR-based replicon typing (PBRT)[35-37], S1-digested pulsed field gel electrophoresis (S1-PFGE)[4], and ERIC-PCR were conducted to confirm that only a single mcr-1-harboring helper plasmid was transferred into E. coli DH5α+ pLHM10–1-p6. Secondly, the pLHM10–1-p6-harboring E. coli DH5α strains with a single putative helper plasmid (namely pZ6–2, pH12–4, or pHLJ179) were used as donors for conjugation, with E. coli C600 as the recipient strain, followed by selection of transconjugants with the LB agar plates containing tigecycline (4 mg/L) and streptomycin (1000 mg/L). The putative transconjugants and mobilizing efficiency were further determined as described above, respectively.

Screening for the presence of tet(X4) and tet(X4)-harboring plasmid like pLHM10–1-p6

All tigecycline-nonsusceptible isolates were screened for tet(X4) by PCR, followed by Sanger sequencing. The pLHM10–1-p6-like plasmid was also screened by overlapping PCRs targeting four overlapping regions (IncQ-1 [ΔczcD-orf1-ΔISCR2-Δtet(X4)], IncQ-2 [Δtet(X4)-catD-ISCR2-Δorf2], IncQ-3 [Δorf2-mobC-mobA-mobB-repB-ΔrepF], and IncQ-4 [repF-repA-repC-Δsul2-czcD-Δorf1]) on pLHM10–1-p6 (Supplementary Fig. 3). All primers used here were listed in appendix (Supplementary Table 4).

Antimicrobial susceptibility testing

MICs of ten antibiotics (tetracycline, amikacin, gentamicin, meropenem, cefoxitin, cefotaxime, fosfomycin, ciprofloxacin, sulfamethoxazole/trimethoprim, and florfenicol) for all tet(X4)-harboring strains were determined by agar dilution and interpreted according to the Clinical & Laboratory Standards Institute (CLSI) guidelines. MICs of tigecycline, eravacycline, and colistin were determined by broth microdilution. Especially, the breakpoint of tigecycline for E. coli strains was interpreted according to the FDA criteria (susceptible, ≤2 mg/L; intermediate, 4 mg/L; resistant, ≥8 mg/L), while the eravacycline was uninterpreted with no breakpoint. Colistin was interpreted in accordance with the European Committee on Antimicrobial Susceptibility Testing (EUCAST) breakpoint (susceptible, ≤2 mg/L; resistant, >2 mg/L). E. coli ATCC 25922 was served as a quality control strain for susceptibility testing.

Molecular typing

All tet(X4)-positive E. coli strains were classified by XbaI-digested (Takara, Dalian, China) PFGE according to the PulseNet protocol (http://www.pulsenetinternational.org/protocols/) using a CHEF Mapper® System (Bio-Rad, Hercules, CA, USA). PFGE patterns were compared using BioNumerics version 6.6 (Applied Maths Software) under appropriate optimization (1.5%) and tolerance (1.5%) settings, with a cutoff at 85% similarity to delineate PFGE clusters. Multilocus sequence typing (MLST) was performed by the primers and protocol specified in E. coli MLST database website (http://enterobase.warwick.ac.uk).

Murine thigh infection model

Female ICR mice (6–10 weeks old; Guangdong Medical Lab Animal Center, Guangzhou, China) were used in this experiment. Mice were maintained in accordance with the National Standards for Laboratory Animals in China (GB 14925–2010). All animal studies were conducted in accordance with SCAU Institutional Animal Welfare and Ethics guidelines. The animal use procedures were approved by the Animal Research Committees of SCAU. No specific statistical consideration was taken in determining the sample size for the animal experiments. Investigators were blinded as for group allocation when determining colony forming unit (CFU). Meanwhile, we were able to cure the tet(X4)-carrying plasmid from wild-type E. coli STB20–1 [tet(X4)-positive] using sodium dodecyl sulfate (SDS) as previously described[10], to create STB20–1S [tet(X4)-negative], and thereby any difference caused by treatment will be most likely due to the presence of the plasmid only.

Prior to infection, mice were rendered neutropenic by injecting two doses of cyclophosphamide intraperitoneally on 4 days (150 mg/kg) and 1 day (100 mg/kg). In order to study the impact of the tet(X4)-carrying plasmid on efficacy of tigecycline, thigh infections were produced by injecting 0.1 ml of suspensions of E. coli STB20–1 or its tet(X4)-cured STB20–1S strain at 106 CFU/ml. At 2 hours post-infection, the mice were administered subcutaneously with a single dose of tigecycline at 50 mg/kg per 24 hours, which was approximately equivalent to the dosing regimen in humans given as 50 mg per 12 hours for 6 days[38,39]. Mice were humanely euthanized at 24 hours after treatment and thigh muscles were aseptically homogenized in 3 ml of saline. After serial dilutions, suspensions were quantitatively cultured on antibiotic-free LB agar plates for CFU enumeration.

Statistical analysis

Descriptive analyses on percentage (together with standard deviation) and prevalence were performed using functions provided in Excel 2007 (Microsoft Software). All statistical analyses (unpaired t test, two tailed; two-way ANOVA) were performed with GraphPad Prism 5 (GraphPad Software).

Data availability

The whole genome sequence data of E. coli LHM10–1 and G3X16–2 strains have been submitted to NCBI under the BioSample accession number SAMN11087649 and SAMN11180601, respectively. Extra data supporting the findings of this study are available from the corresponding authors upon reasonable request.