Background

Tetracycline antibiotics have been extensively used in prophylaxis and treatment of human and animal infections, as well as at subtherapeutic levels as growth-promoters in animal feed [1, 2]. The third-generation tetracycline antibiotic, tigecycline, is regarded as one of the last-resort antibiotics to treat clinical multidrug-resistant (MDR) bacterial infections. It exhibits an expanded spectrum of activities against both Gram-negative and Gram-positive bacteria, including carbapenem-resistant Enterobacteriaceae (CRE) and Acinetobacter baumannii (CRAB), methicillin-resistant Staphylococcus aureus (MRSA), and vancomycin-resistant Enterococcus (VRE) strains [3, 4], and is classified as a critically important antimicrobial by the World Health Organization.

However, the recent emergence of plasmid-mediated high-level tigecycline resistance mechanisms, Tet(X3), Tet(X4), and Tet(X5), raises the concern that this last-resort antibiotic may be ineffective, further limiting clinical treatment choices [5,6,7]. Tet(X), a flavin-dependent monooxygenase, is capable of degrading all tetracycline antibiotics by hydroxylation, representing a unique enzymatic tetracycline inactivation mechanism [8, 9]. Thus far, tet(X) genes [especially tet(X3)–tet(X5)] have been detected in over 16 different bacterial species from various ecological niches of humans, migratory birds, food-producing animals, and their neighboring environments, with Acinetobacter spp. and Escherichia coli the most predominate species [5, 7, 10, 11]. Nevertheless, the original source and evolutionary history of tet(X) genes remain poorly understood. In addition, our previous surveillance study on E. coli revealed a low prevalence of tet(X)-positive E. coli isolates from various sources in China, with tet(X4) the only variant detected [6]. Conversely, it remains unclear how much Acinetobacter spp. strains contribute to the dissemination of tet(X).

The genus Acinetobacter is a heterogeneous group and comprised of more than 60 bacterial species [12], which are ubiquitous in the nature but can also cause serious infections in hospital settings [13, 14]. Extensively drug-resistant (XDR) Acinetobacter species, especially CRAB, have emerged as a clinically troublesome pathogen [15]. In this study, we investigated the prevalence of tet(X) genes in Acinetobacter spp. isolates from human, migratory bird, pig, and surrounding environmental samples in China, and explored the genetic diversity and characteristics of tet(X) genes, plasmids, and strains.

Methods

Sample collection and bacterial isolation

We conducted a multiregional study to investigate the prevalence of tet(X) genes in Acinetobacter isolates from human, pig, migratory bird, and surrounding environmental samples between May 2015 and May 2018. There is no self-selection bias that may be present during the sample collection. Briefly, the animal stool samples were randomly collected, with approximate 50 samples per pig farm (n = 2083) or 150 samples per migratory bird habitat (n = 863). If possible, the soil (n = 182), dust (n = 170), sewage (n = 136), water (n = 54), and vegetable (n = 59) samples were also collected at least three per site. In addition, the human specimens (urine, n = 175; nasal swabs, n = 64; rectal swabs, n = 60) were collected from Guangdong province for clinical investigation. Subsequently, the feces of pigs and migratory birds, soils, dusts, and chopped vegetables were suspended in sterile saline at a weight/volume ratio of 1:5, respectively. Meanwhile, the nasal and rectal swabs of inpatients were directly discharged into 1.5 mL of sterile saline. For the treated samples, together with sewage, water, and urine of physical examination people, 100 μL of them was used for the next bacterial isolation.

A total of 3846 none-duplicate samples were collected from one tertiary-care hospital (n = 299), five migratory bird habitats (n = 972), and 33 intensive pig farms (n = 2575) in 14 provinces and municipalities in China. These samples were then selected by CHROMagar™ Acinetobacter plates (CHROMagar, France) containing tigecycline (2 μg/mL), and the isolates were screened for tet(X3) and tet(X4) by PCR amplification with primers listed in the supplementary information (Additional file 1: Table S1). In addition, a random collection of 402 A. baumannii clinical isolates from two hospitals in Guangdong and Jiangsu provinces was also screened as described above. All these tet(X)-positive Acinetobacter spp. strains were further characterized by whole-genome sequencing (described as below).

Antimicrobial susceptibility testing

Minimum inhibitory concentrations (MICs) of ten antimicrobials for tet(X)-positive Acinetobacter spp. isolates were determined by twofold agar dilution method and interpreted according to the Clinical and Laboratory Standards Institute (CLSI) guideline [16]. In brief, the antibiotic to be tested was diluted by 1 mL of sterile ddH2O to make a series of concentrations, followed by mixing with 19 mL of fresh Mueller-Hinton (MH) agar to produce plates in which the final antibiotic concentrations represented a 2-fold dilution series. The bacterial suspension (~ 105 cfu) was then spotted on MH plates and incubated at 35 °C for 20 h. The lowest concentration of antibiotics that prevented bacterial growth was considered to be the MIC. Additionally, MICs of tigecycline, eravacycline, and omadacycline were determined by broth microdilution method. In particular, the breakpoint of tigecycline was interpreted according to the United States Food and Drug Administration (USFDA) criteria for Enterobacteriaceae bacteria as previously reported [17]. E. coli strain ATCC 25922 was used as the quality control strain.

Whole-genome sequencing (WGS) and bioinformatics analysis

The genomic DNA of 193 tet(X)-positive Acinetobacter spp. strains was extracted and sequenced using the Illumina HiSeq platform (Novogene, China). The raw sequence data were then assembled by SPAdes version 3.12.0 [18]. To obtain complete sequences, five representative Acinetobacter spp. strains, namely Acinetobacter Clade_U6 10FS3-1 [tet(X3)-positive], Acinetobacter Clade_U1 YH12138 [tet(X3)- and tet(X5.2)-positive], A. piscicola YH12207 [tet(X3)- and tet(X5.3)-positive], A. indicus Q186-3 [tet(X4)- and tet(X5.2)-positive], and A. indicus C15 [tet(X3)- and blaNDM-1-positive], were further subjected to Oxford Nanopore sequencing (Nextomics, China), followed by assembling with Unicycler version 0.4.8 [19]. To determine the bacterial species, pair-wised average nucleotide identities (ANIs) were calculated using FastANI and Mash [20, 21], and compared with Acinetobacter genomes from the National Center for Biotechnology Information (NCBI) RefSeq database [22]. A cutoff of > 95% and < 83% ANI values was used to determine intra-species and inter-species boundaries, respectively [20]. A Mash distance tree was generated using mashtree [23].

Gene prediction and annotation were performed according to the NCBI Prokaryotic Genome Annotation Pipeline [24]. The heatmap of antibiotic resistance genes was created using the R package pheatmap version 1.0.12 [25]. The visual representation of tet(X)-carrying plasmids was generated with DNAPlotter version 1.11 [26]. A further BLASTn/BLASTp analysis of tet(X5.3) against the NCBI database identified over 200 homologous sequences with high-scoring hits (E value < 2e−65) [27]. Subsequently, the amino acid sequences of 54 non-duplicated Tet(X) proteins and monooxygenases were used to estimate the substitution rate among 404 sites and produce a divergence time tree by BEAST version 2.6.0, with tip-dates defined as the years of isolation or submission [28]. In brief, we analyzed the data under a Bayesian framework using the strict clock+Gamma site or relaxed clock+Gamma site model, and the latter model was a better one because it had higher effective sample size values (ESS, > 200). To test whether population growth rates differed between different lineages, a Markov Chain Monte Carlo method was used for posterior probability distributions. Three independent runs, with chain lengths 100,000,000 and 20% burn-in, were conducted to confirm the convergence of Bayesian time-calibrated phylogenetic analyses. Parameter convergence was visualized by Tracer version 1.7 [29]. The collected trees were then annotated into a maximum clade credibility tree using TreeAnnotator version 2.6.0 in the BEAST package.

Cloning experiments

Firstly, the original tet(X5) (accession number: CP040912) and putative ancestral flavin-dependent monooxygenase gene fmo (CP014021) were artificially synthesized according to the reference sequences from NCBI. The tet(X3), tet(X4), tet(X5.2), and tet(X5.3) variants were PCR amplified from isolates Clade_U6 10FS3-1, A. indicus Q278-1, Clade_U1 YH12138, and A. piscicola YH12207, respectively. These genes were then cloned into the plasmid vector pBAD24 under an arabinose inducible PBAD promoter as our previous description [6]. In addition, the recombinant plasmids co-harboring tet(X3)/tet(X5.2) or tet(X3)/tet(X5.3) were further constructed, using pBAD24+tet(X3) as the plasmid skeleton. All these constructs were selected on Luria-Bertani (LB) agar plates containing 100 μg/mL ampicillin (resistance encoded by amp on the pBAD24 backbone), and confirmed by PCR detection and Sanger sequencing for tet(X) genes (Additional file 1: Table S1). Susceptibility testing for six tetracyclines (namely tetracycline, doxycycline, minocycline, tigecycline, eravacycline, and omadacycline) was performed by broth microdilution with the addition of 0.1% l-arabinose.

Eravacycline degradation assay

The eravacycline degradation ability of five tet(X) variant constructs and their parental strains was initially determined by agar well diffusion assay in triplicate [6]. In brief, the MH agar plate surface was inoculated by spreading 100 μL of overnight culture of eravacycline-susceptible Bacillus stearothermophilus ATCC 7953 (0.5 McFarland), and the well with a diameter of 6 to 8 mm was punched aseptically with a sterile cork borer. The tet(X) constructs were cultured in 5 μg/mL eravacycline and 0.1% l-arabinose at 37 °C for 8 h. Supernatant (20 μL) was then transferred into the agar well and incubated at 60 °C for 16 h. The media only containing eravacycline and the samples treated with E. coli JM109 carrying an empty plasmid pBAD24 were used as the control groups. The parental strains were examined similarly, but without the addition of 0.1% l-arabinose during incubation.

In addition, the degradation levels of eravacycline by tet(X5.2) and tet(X5.3) were also quantified by liquid chromatography-tandem mass spectrometry (LC-MS/MS) in quadruplicate as previously described [6]. Briefly, the tet(X) clones were incubated in 4 mL of optimized M9 media with eravacycline (2 μg/mL) and l-arabinose (0.1%) at 200 rpm for 16 h. Following centrifugation at 10,000 rpm for 2 min, the supernatant was then passed through a 0.22-μm filter and subjected to LC-MS/MS quantification. Meanwhile, the previously reported tet(X3), tet(X4), and tet(X5) genes served as the positive controls for comparative analyses, while E. coli JM109 carrying pBAD24 was used as a negative control. The statistical analysis was conducted using an unpaired, two-sided t test. The linear range of the standard curve for eravacycline was from 10 to 500 ppb, with r2 = 0.994.

Quantitative reverse transcription-PCR (qRT-PCR)

The transcript expression levels of different tet(X) variants in a tandem structure were determined by qRT-PCR in triplicate. Total bacterial RNA of Acinetobacter Clade_U1 YH12138 and A. piscicola YH12207 was obtained using the E.Z.N.A. Bacterial RNA Kit (OMEGA, China), respectively, and then reversely transcribed into cDNA using the M-MLV First Strand cDNA Synthesis Kit (OMEGA, China). Quantitative real-time PCR was performed with SYBR Premix Ex Taq (Takara, China) on a LightCycler 96 instrument (Roche, Switzerland) as previously reported [30]. 16S rRNA was used as the endogenous control (Additional file 1: Table S1). Relative expression levels between tet(X3) and tet(X5.2) or tet(X5.3) were calculated using the 2-ΔΔCT method [31].

Mobilization of tet(X)-mediated tigecycline resistance

The transferability of tigecycline resistance mediated by tet(X3), tet(X4), tet(X5.2), and tet(X5.3) was determined by filter mating using representative Acinetobacter spp. isolates as the donor strains and A. baylyi ADP1 (rifampicin-resistant), E. coli C600 (streptomycin-resistant), and clinical E. coli 1314 (meropenem-resistant) strains as the recipient strains. The transconjugants were selected on LB agar plates containing tigecycline (2 μg/mL) in combination with rifampin (100 μg/mL), streptomycin (1500 μg/mL), or meropenem (2 μg/mL), respectively.

To confirm the ISCR2-mediated tet(X) transfer, a transposition assay of tet(X) was conducted as previously reported for mcr-1 [32]. The putative transposition unit of tet(X3) in Clade_U6 10FS3-1, namely ΔtpnF-tet(X3)-hp-hp-ISCR2, was cloned in a suicide plasmid pSV03 using the ClonExpress Ultra One Step Cloning Kit (Vazyme, China) (Additional file 1: Table S1) and then transformed into competent E. coli WM3064. Subsequently, the transformant was treated by filter mating with the recipient E. coli K-12 strain MG1655 (recA∷Km) and A. baylyi ADP1, and the transposon-insertion mutants were selected by LB agar plates containing tigecycline (2 μg/mL) as well as kanamycin (30 μg/mL) or rifampin (100 μg/mL).

All the putative transconjugants and transposon mutant strains were screened for tet(X) genes by PCR and Sanger sequencing (Additional file 1: Table S1). The recipient strain backgrounds were confirmed by PCR-fingerprints for A. baylyi and enterobacterial repetitive intergenic consensus PCR (ERIC-PCR) for E. coli [33, 34]. Transfer efficiency was calculated based on colony counts of the transconjugant, transposon mutant, and recipient cells in triplicate [35].

Results

Prevalence of tet(X)-positive Acinetobacter spp. strains in China

In this study, a total of 3846 samples were collected from 14 provinces and municipalities in China from pig farms, migratory bird habitats, and human specimens. PCR assays identified a total of 193 tet(X3)- or tet(X4)-positive Acinetobacter spp. isolates (5.0%, 193/3846) from pig farms (7.3%, 187/2575), migratory birds (0.5%, 5/972), and human samples (0.3%; 1/299) in nine provinces and municipalities (Fig. 1; Additional file 1: Table S2). Among them, 188 isolates were positive for tet(X3) and five for tet(X4). The tet(X3) gene was widely distributed in southern and eastern China, including Hainan (25.3%, 19/75), Shanghai (18.8%, 16/85), Jiangsu (13.2%, 27/204), Hunan (12.6%, 14/111), Zhejiang (7.6%, 31/406), Jiangxi (4.7%, 31/666), Guangdong (4.2%, 38/914), and Fujian (3.1%, 12/390), while tet(X4) was only found in Qinghai (2.3%, 5/214) (Fig. 1). An A. lwoffii strain (YH18001) recovered from a urine sample in a healthy individual in Huizhou, Guangdong, carried tet(X3). The screening of 402 A. baumannii clinical isolates from two hospitals in Guangdong and Jiangsu provinces did not identify tet(X)-positive isolates.

Fig. 1
figure 1

Distribution of tet(X)-positive Acinetobacter spp. strains in China. The 14 sampling provinces and municipalities are marked. The bacterial number per region and corresponding proportion are also indicated

Full size image

The 193 tet(X)-positive strains were widely distributed in 12 different Acinetobacter species, including six putative novel species. An ANI analysis showed that 101 genomes could not be assigned into any known Acinetobacter species, but they demonstrated 84–90% FastANI and Mash-based ANI values against known Acinetobacter species from the NCBI RefSeq database, suggesting that they likely belong to novel species in the Acinetobacter genus. The 101 genomes were clustered in six groups, with > 95% ANI values in each group. In this study, we tentatively named the six novel species as Acinetobacter Clade_U1 to Clade_U6 (Fig. 2a). Interestingly, the Acinetobacter Clade_U1 was the most predominant tet(X)-harboring species (33.7%, 65/193), followed by A. indicus (16.6%, 32/193), A. towneri (16.6%, 32/193), Clade_U2 (13.0%, 25/193), and A. lwoffii (7.8%, 15/193) (Fig. 2b).

Fig. 2
figure 2

Genomic characteristics of tet(X)-positive Acinetobacter species. a Mash distance relationship of 193 tet(X)-positive Acinetobacter spp. isolates. The branch lines are colored based on the bacterial species as listed on the right. The bacterial sources of representative isolates are illustrated by silhouettes. The closely related isolates from different sources are highlighted by shading. b Distribution of tet(X)-positive strains in 12 Acinetobacter species. c Heatmap of antibiotic resistance genes in 12 Acinetobacter species. The color of each box represents the percentage of resistance genes in corresponding species, ranging from 0.0 (cyan) to 100.0% (red)

Full size image

It should be noted that these tet(X3)-positive strains were detected in various sources in pig farms, including soil (9.4%, 12/127), pig (7.6%, 158/2083), sewage (6.6%, 9/136), vegetable (5.1%, 3/59), and dust (2.9%, 5/170) samples, suggesting tet(X3) is widely spread in the intensive pig farms in China (Additional file 1: Table S2). The genome of aforementioned human A. lwoffii isolate YH18001 was closely related to some A. lwoffii isolates collected from pig stool samples (e.g., YH12105 from a pig in Shanghai) (Fig. 2a). In addition, the tet(X4) gene was detected in five A. indicus strains from migratory birds [bar-headed goose (Anser indicus)] in Qinghai province (Fig. 1; Additional file 1: Table S2). Notably, three out of them were clustered together with some A. indicus strains (e.g., YH12090 and YH12091) isolated from pig sources in Guangdong province (Fig. 2a).

Identification of tet(X5.2) and tet(X5.3) variants in Acinetobacter spp.

Genomic sequencing of tet(X3)- or tet(X4)-positive Acinetobacter spp. strains identified two additional tet(X)-like genes. Both of them were 1137 bp open reading frames (ORFs), encoding 378 amino acid proteins (Additional file 1: Figure S1), which showed 95% [designated as Tet(X5.2) herein] and 96% [Tet(X5.3)] amino acid sequence identities to the first reported Tet(X5) (CP040912) in a clinical A. baumannii strain, respectively. Homology modeling of Tet(X5), Tet(X5.2), and Tet(X5.3) illustrated an overall analogous architecture that consisted of the substrate-binding domain, FAD-binding domain, and C-terminal α-helix (Additional file 1: Figure S2). Similar to previously reported tet(X3), tet(X4), and tet(X5), the tigecycline MICs of tet(X5.2) (8 μg/mL) and tet(X5.3) (8 μg/mL) constructs increased 64-fold when compared with that of the E. coli JM109 control carrying the empty pBAD24 vector (0.13 μg/mL) (Additional file 1: Table S3). The same tigecycline MIC (16 μg/mL) was observed between tet(X3) and tet(X3)/tet(X5.2) or tet(X3)/tet(X5.3) constructs. In addition, these constructs exhibited at least 64-fold increases of MICs to the other tetracyclines, including eravacycline and omadacycline, while the putative ancestral gene fmo was inactive.

Microbiological degradation assays revealed that these five tet(X) clones as well as their parental strains could effectively inactivate eravacycline (Fig. 3a). Notably, qRT-PCR revealed that the transcript expression levels of chromosomal tet(X5.2) in Acinetobacter clade_U1 YH12138 and plasmid-mediated tet(X5.3) in A. piscicola YH12207 were (17.6 ± 3.2)% and (83.3 ± 4.9)%, respectively, when compared with that of the coexisting tet(X3). The eravacycline inactivation effects were further confirmed by LC-MS/MS. The results showed the eravacycline concentrations had (70.8 ± 2.2)% reduction in Tet(X5.2) and (74.8 ± 3.4)% reduction in Tet(X5.3) constructs, respectively (Fig. 3b). Similar eravacycline degradation efficiencies were observed in the control Tet(X3) [(87.5 ± 2.3)%], Tet(X4) [(89.4 ± 1.1)%], and Tet(X5) [(87.9 ± 1.0)%] constructs (Fig. 3b). These results indicated that tet(X5.2) and tet(X5.3) genes were also able to degrade tetracycline antibiotics, thereby conferring to high-level resistance.

Fig. 3
figure 3

Comparative analysis of tet(X)-mediated eravacycline degradation. a Microbiological degradation for eravacycline. Five pBAD24-tet(X) constructs as well as their parental strains, including Clade_U6 10FS3-1 [tet(X3)-positive], A. indicus Q278-1 [tet(X4)-positive], Clade_U1 YH12138 [tet(X3)- and tet(X5.2)-positive], and A. piscicola YH12207 [tet(X3)- and tet(X5.3)-positive], are used. The presence of Tet(X) degrades eravacycline, and consequently, the degraded eravacycline loses its antimicrobial activity against a susceptible indicator strain. By contrast, the absence of Tet(X) yields a clear inhibition zone, with a diameter of > 18 mm. The groups with the addition of untreated eravacycline or with the addition of eravacycline treated with E. coli JM109 carrying the empty vector pBAD24 serve as negative controls. b LC-MS/MS quantification of eravacycline degradation by tet(X) clones. Individual values of four biological replicates are shown as dots, while the means and standard deviations are displayed as error bars. Vector (−), the control group with the addition of eravacycline treated with pBAD24-carrying E. coli JM109

Full size image

In total, 20.7% (40/193) of the tet(X)-positive Acinetobacter spp. isolates were found to carry two different tet(X) variants. Specifically, 18.7% (36/193) of the tet(X3)-positive (n = 34) and tet(X4)-positive (n = 2) Acinetobacter spp. strains were found to co-harbor tet(X5.2), which were collected from various sample sources in nine provinces (Fig. 1; Additional file 1: Table S2). In addition, 2.1% (4/193) of isolates were found to co-harbor tet(X5.3) and tet(X3). They were all from A. piscicola and isolated from pig (n = 2) and soil (n = 2) samples in Zhejiang province. None of the isolates was found to co-harbor other or more than two tet(X) variants.

Antimicrobial resistance profile of tet(X)-carrying Acinetobacter spp. isolates

Susceptibility testing showed that all tet(X)-carrying Acinetobacter spp. strains were resistant to tigecycline and tetracycline, and exhibited high MIC levels to eravacycline (2–8 μg/mL) and omadacycline (8–16 μg/mL). These isolates were commonly resistant to trimethoprim/sulfamethoxazole (91.2%; 176/193), florfenicol (88.1%; 170/193), and ciprofloxacin (69.9%; 135/193), but less frequently resistant to other antibiotics (namely gentamicin, ampicillin, amikacin, colistin, meropenem, and cefotaxime), with the resistance rates ranging from 0.5 to 24.9% (Additional file 1: Figure S3). Among the major species, Clade_U1 showed higher ciprofloxacin (83.1%) and ampicillin (40.0%) resistance rates, while Clade_U2 exhibited 100% florfenicol resistance. Additionally, 53.3% of A. lwoffii isolates showed gentamicin resistance. In silico resistance gene mining from their whole-genome sequences showed that the tet(X) genes were commonly associated with sul2 and floR (Fig. 2c), which correlated with the trimethoprim/sulfamethoxazole and florfenicol resistance phenotypes described above. Notably, one A. indicus isolate (0.5%; 1/193) was resistant to meropenem and cefotaxime. A further WGS analysis identified the presence of carbapenem-resistant gene blaNDM-1 and tet(X3) on the same 3.2-Mb chromosome.

tet(X)-harboring plasmid and chromosome sequences

The combination of Nanopore and HiSeq genomic assembly obtained the complete tet(X)-carrying plasmid and chromosome sequences from several representative Acinetobacter spp. strains (Fig. 4a). However, these tet(X)-positive plasmid sequences showed low nucleotide sequence identities (< 70%) between each other, suggesting the spread of tet(X) is not due to the horizontal transfer of a predominant plasmid.

Fig. 4
figure 4

Genetic structures of tet(X)-positive plasmids and chromosomes. a Sketch maps of the tet(X3)-, tet(X4)-, tet(X5.2)-, and tet(X5.3)-positive Acinetobacter spp. strains. Bacterial species, isolation sources, and tet(X) variants are illustrated. b, c Structures of the tet(X3)-harboring plasmid p10FS3-1-3 (b) and the tet(X3)- and tet(X5.3)-co-harboring plasmid pYH12207-2 (c). GC skew and GC content are indicated from the inside out. The arrows represent the positions and transcriptional directions of the ORFs. Δ symbol indicates the truncated gene. d, e Putative integration processes of tet(X)-carrying composite transposons in Clade_U1 YH12138 (d), A. piscicola YH12207 (d), and A. indicus Q186-3 (e). The replication initiation site oriIS and replication termination site terIS of the ISCR2 element are also indicated

Full size image

For the Acinetobacter Clade_U6 strain 10FS3-1, the tet(X3) gene was located on a 73,803-bp untypable plasmid, p10FS3-1-3. p10FS3-1-3 had an average GC content of 42.5% and harbored 61 putative ORFs, including genes encoding resistance to sulfonamide (sul2), aminoglycosides [aph(3″)-Ib, aph(6)-Id], and heavy metals (czcA-czcD) (Fig. 4b). For the A. piscicola strain YH12207, the tet(X3) and tet(X5.3) genes formed a composite transposon-like structure of ΔISCR2-hp-tet(X5.3)-hp-hp-ΔISCR2-IS26-hp-aph(3′)-Ia-IS26-ΔtpnF-tet(X3)-hp-hp-ISCR2 and were localized on an 148,815-bp untypable plasmid pYH12207-2, possessing 128 ORFs with a GC content of 39% (Fig. 4c). Similarly, the tet(X3) and tet(X5.2) genes were located on the chromosome of the Acinetobacter clade_U1 strain YH12138, carrying a similar composite transposon-like structure of ΔISCR2-hp-tet(X5.2)-hp-hp-ΔISCR2-IS26-hp-aph(3′)-Ia-IS26-ΔtpnF-tet(X3)-hp-hp-ISCR2. Furthermore, tet(X4) and tet(X5.2) were found on the chromosome of A. indicus Q186-3 at different regions.

Conjugation experiments showed that both p10FS3-1-3 [tet(X3)-positive] and pYH12207-2 [tet(X3)- and tet(X5.3)-positive] could be successfully transferred into the recipient strain A. baylyi ADP1 at a low frequency of ~ 10−9, confirming the conjugability of tet(X)-harboring plasmids. There was no significant additive effect on MICs of tetracycline antibiotics between pYH12207-2 and p10FS3-1-3 transconjugants (Additional file 1: Table S3). Moreover, the conjugation experiments failed despite multiple attempts when Enterobacteriaceae bacteria were used as recipient strains, suggesting these plasmids could only replicate in Acinetobacter species.

Genetic environments of tet(X) variants

Previous studies showed that the tet(X) variants were usually associated with an IS91-like element ISCR2, and some typical transposon structures ISCR2-tpnF-tet(X3)-hp-hp-ISCR2 for tet(X3), ISCR2-catD-tet(X4)-ISCR2 for tet(X4), and ISCR2-tpnF-tet(X5)-hp-hp-ISCR2 for tet(X5) were described [5,6,7]. Similarly, the genetic analyses of tet(X3), tet(X4), tet(X5.2), and tet(X5.3) in this study revealed the same close association between ISCR2 and different tet(X) variants. Intact or partial ISCR2 sequences were identified in the four tet(X) variants among most isolates (Additional file 1: Figure S4).

Two major tet(X3) genetic structures were identified in 187 tet(X3)-positive Acinetobacter spp. strains [one with a truncated tet(X3) was not included; Additional file 1: Figure S4a]. The previously described ISCR2-tpnF-tet(X3)-hp-hp-ISCR2 gene cassette was found in 38.0% (71/187) of Acinetobacter isolates, including two isolates with partial sequence deletion between tpnF and tet(X3) (types II and III). In the rest 115 (61.5%) tet(X3)-harboring isolates, the recombinase gene tpnF that located upstream of tet(X3) was truncated by an IS26 insertion (type I).

Similarly, the same ISCR2 element was found downstream of tet(X4), tet(X5.2), and tet(X5.3), respectively, but additional genetic rearrangements and IS insertion were identified (Additional file 1: Figure S4b-4c). It was noteworthy that the tet(X4)-harboring contigs in four Acinetobacter spp. strains showed 100% sequence identities to the gene cassette of ISCR2-catD-tet(X4)-ISCR2 that we previously described in Enterobacteriaceae and Aeromonas caviae strains (Additional file 1: Figure S4b). For tet(X5) variants, tet(X5.2) and tet(X5.3), 80.0% (32/40) of them shared a similar cassette as the structure in E. coli 912, Proteus cibarius ZF2, and Chryseobacterium spp. BGARF1 strains (Additional file 1: Figure S4c). The close association between ISCR2 and different tet(X) variants, and the high sequence identities of tet(X) cassette from difference species further highlighted the critical role of ISCR2 in the dissemination of tet(X).

The ISCR2-mediated tet(X) transposition was further experimentally confirmed using the tet(X3) cassette [ΔtpnF-tet(X3)-hp-hp-ISCR2] cloned from Clade_U6 10FS3-1. The tet(X3) cassette was successfully transferred to the recipient strains A. baylyi ADP1 and E. coli MG1655 at a frequency of (9.6 ± 2.7) × 10−6 and (5.4 ± 1.4) × 10−8, respectively, which was consistent with our previous report for tet(X4) in A. caviae [11].

Molecular evolutionary analysis of Tet(X)

The phylogenetic reconstruction showed that Tet(X) variants, Tet(X0)–Tet(X5), and their closely related sequences formed a separate subclade (Fig. 5). Among them, Tet(X0) [first described as Tet(X) in Bacteroides fragilis [36], and tentatively designated as Tet(X0) in this study to differentiate from other Tet(X) variants], Tet(X1), and Tet(X2) were the earliest identified Tet(X) enzymes, but Tet(X1) was nonfunctional due to its N-terminal truncation [8]. The sequences in Tet(X3), Tet(X4), and Tet(X5) subclades showed approximate 85%, 95%, and 87% amino acid sequence identities to those in the Tet(X0)/Tet(X2) subclade, respectively. Notably, the Tet(X0)/Tet(X2) subclade mainly consisted of anaerobic Bacteroides spp. and Riemerella anatipestifer, while the Tet(X3), Tet(X4), and Tet(X5) subclades consisted of aerobic Acinetobacter spp. and Enterobacteriaceae bacteria.

Fig. 5
figure 5

Bayesian phylogenetic inference of Tet(X) proteins by relaxed clock model. The reported Tet(X) variants are underlined, while different Tet(X) subclades are shown by shading in different colors. The divergence time as well as its 95% HPD between Tet(X) and monooxygenase subclades is shown around the node. The tip labels are annotated by bacterial species and their corresponding GenBank accession numbers

Full size image

By contrast, the chromosomal monooxygenases from several environmental Flavobacteriaceae bacteria, including Chryseobacterium spp. (e.g., CP015199), Flavobacterium spp. (e.g., CP000685), and Elizabethkingia spp. (e.g., CP016372), shared an approximate 55.8% sequence identity to the first reported Tet(X0). A Bayesian evolutionary analysis indicated the divergence time of Tet(X) clades [except the truncated Tet(X1)] from the chromosomal monooxygenases likely occurred ~ 9900 years ago [7887 BC; 95% highest probability density (HPD), 26,517 BC–AD 1099].

The average GC content of Tet(X) variants (approximate 37%) was similar to the average genomic GC content of Flavobacteriaceae bacteria (33–38%), but much lower than that of Acinetobacter spp. and Enterobacteriaceae bacteria (> 45%). These results suggested that the environmental Flavobacteriaceae-related bacteria constitute a putative ancestral subclade, leading to the recent emergence of tet(X)-like genes in Acinetobacter spp. and Enterobacteriaceae (Additional file 1: Figure S5).

Discussion

One of the most significant findings in this study was the high prevalence of tet(X) genes among Acinetobacter spp. strains from different sources, especially in pig farms (7.3%, 187/2575), which was considerably higher than what we previously reported for E. coli from samples in pigs and their surrounding environments (1.1%, 34/2970) [6]. The results suggested that tet(X)-harboring Acinetobacter spp. strains have been widely spread in animal and environment sources from different regions in China. Importantly, these tet(X)-harboring Acinetobacter spp. strains were broadly dispensed into 12 species, including six putative novel species that were firstly described here. The six novel species covered over 50% of the tet(X)-positive Acinetobacter spp. isolates identified in this study. The results further indicated that our understanding about the bacterial hosts for tet(X) was limited, and previous studies might significantly underestimate the paramount role of Acinetobacter species in the dissemination of tet(X)-mediated tigecycline resistance.

The Acinetobacter spp. strains are ubiquitous in the natural environment [37], and we suspect that these tet(X)-positive Acinetobacter strains, including isolates from the novel species (Clade_U1 to U6), likely originate from environmental sources. Although tigecycline is only approved for clinical settings, the other tetracycline antibiotics, including tetracycline, oxytetracycline, and chlortetracycline, were heavily utilized in animal and agricultural production [38, 39]. The selection pressure exacted by them might promote the horizontal transfer of tet(X) into Acinetobacter spp. strains. Similarly, some other antimicrobials (e.g., sulfonamides, phenicols, and quinolones) were also frequently used in animal feed in China, which were consistent with our > 65% resistance rates for trimethoprim/sulfamethoxazole, florfenicol, and ciprofloxacin in these tet(X)-positive Acinetobacter spp. isolates (Additional file 1: Figure S3).

Although tet(X)-positive Acinetobacter spp. strains are universal in animal and environmental sources, some of the Acinetobacter species (e.g., A. lwoffii and A. schindleri) are opportunistic pathogens and have been identified as sources of nosocomial infections, including septicemia, pneumonia, meningitis, urinary tract infections, and skin and wound infections [40, 41]. In this study, we detected a colonized tet(X3)-positive A. lwoffii isolate (YH18001) from the urine sample in a healthy person undergoing physical examination. Similarly, previous studies also reported the identification of tet(X)-harboring A. baumannii isolates from clinical specimens among inpatients [5, 7]. These results suggested tet(X) genes have disseminated into pathogenic Acinetobacter species, which might render tigecycline ineffective and severely limited therapeutic options. Worrisomely, some tet(X)-positive Acinetobacter strains also harbored additional resistance mechanisms, such as carbapenemase gene blaNDM-1 (e.g., A. indicus C15), conferring resistance to clinically important antimicrobials (e.g., cephalosporins and carbapenems) [42, 43].

The high diversity of Acinetobacter species harboring tet(X) variants across China further suggested that the wide distribution of tet(X) is mediated by horizontal genetic transfer rather than a single predominant bacterial species or clone, which was also supported by the finding of low sequence identities between different tet(X)-harboring plasmids (e.g., p10FS3-1-3 and pYH12207-2). In addition to tet(X3) and tet(X4), two tet(X) variants tet(X5.2) and tet(X5.3) were detected in this study, with conservative genetic environments across 12 Acinetobacter species in human, pig, migratory bird, soil, dust, sewage, and vegetable samples. Our gene construct and phenotypic work also confirmed the high-level enzymatic degradation ability and tigecycline resistance that conferred by them.

Interestingly, 40 (20.7%) isolates harbored two different tet(X) variants. The molecular mechanism underlying the genetic redundancy of tet(X) remained poorly understood, which might be in part explained by the repeated acquisitions of tet(X) genes through ISCR2-mediated transposition. Thus far, all tet(X) variants conferring high-level tigecycline resistance have been linked to ISCR2, which shows the ability to mobilize various resistance genes (e.g., rmtH and blaVEB) by rolling-circle transposition [44, 45]. Our current and previous studies also confirmed that ISCR2 could transfer tet(X) in Acinetobacter spp. and A. caviae strains via transposition [11]. Accordingly, the composite structures of tet(X) genes in Acinetobacter strains might be generated randomly by the ISCR2-mediated transposition and recombination (Fig. 4d, e), as previously reported for the tandem structure of tet(X4) [5]. It should be noted that these genetic structures were found on both plasmid and chromosome in Acinetobacter species, highlighting that ISCR2 was highly active in mobilizing tet(X)-mediated tigecycline resistance. However, the presence of more than one tet(X) variants did not seem to confer additive resistance to tetracyclines (Additional file 1: Table S3). We suspected that the relative stability of tet(X)-mediated tigecycline resistance may result from the balance between tet(X) expression and bacterial survival, which warranted further studies [46].

Previous studies indicated that ISCR2 was highly dispersed among Acinetobacter spp. strains [47]. Genomic examination of 5433 Acinetobacter spp. draft genome sequences in the NCBI database revealed that 54.5% (n = 2960) of Acinetobacter genomes contained ISCR2 [48]. Similarly, 17.7% (n = 3333) of 18,815 draft E. coli genomes were found to harbor ISCR2. However, the frequency of ISCR2 in Flavobacteriaceae bacteria, including Chryseobacterium spp., Flavobacterium spp., and Elizabethkingia spp., was much lower (0.6%, 5/830). The results suggested that, in comparison to Flavobacteriaceae species, ISCR2 (and its neighboring sequences) has a higher propensity to be integrated in the Acinetobacter and E. coli genomes, which might partially explain the higher detection rates of tet(X) among these species.

Our phylogenetic reconstruction indicated that the tet(X) variants shared the same most recent common ancestor as the chromosome-borne monooxygenase genes in Flavobacteriaceae bacteria, and the divergence between them occurred at ~ 7887 BC, although a potential polyphyletic relationship could not be completely ruled out. We hypothesized that the molecular evolution of tet(X) started with the mobilization of tet(X)-like monooxygenase sequences (or along with their neighboring genes) by ISCR2-like transposase gene from the chromosome of Flavobacteriaceae species over ~ 9900 years ago, followed by the integration into some environmental bacterial genomes (e.g., Acinetobacter spp. or E. coli). ISCR2 further mobilized the tet(X) cassette into different plasmid or chromosome locations through transposition due to its active nature among these new hosts. The divergence of tet(X) variants pre-dated the discovery and clinical use of tetracycline antibiotics. Under the antibiotic selection pressure, the tet(X) variants were selected and subsequently transferred through conjugation or other transfer mechanisms. Our conjugation and transposition experiments also confirmed the transferability of tet(X)-harboring plasmids and transposon. The similar mechanism has been proposed to interpret the molecular evolution of mcr-1 [49].

Conclusions

Taken together, our study reported a significantly high prevalence of tet(X) genes in Acinetobacter spp. strains across China from various sources. In addition to tet(X3) and tet(X4), tet(X5.2) and tet(X5.3) were also confirmed to have high-level degradation activities on tetracyclines but without additive effects between different variants. These tet(X) genes might be further spread by plasmids and the ISCR2 element. Worrisomely, the tet(X)-mediated tigecycline resistance has been detected in carbapenem- and colistin-resistant Acinetobacter spp. strains. Future efforts are needed to improve the surveillance of tet(X) genes from all related sectors, and to monitor the occurrence of tet(X) in clinical pathogens and evaluate the clinical impacts.