Genetic diversity and characteristics of high-level tigecycline resistance Tet(X) in Acinetobacter species

Background The recent emergence and dissemination of high-level mobile tigecycline resistance Tet(X) challenge the clinical effectiveness of tigecycline, one of the last-resort therapeutic options for complicated infections caused by multidrug-resistant Gram-negative and Gram-positive pathogens. Although tet(X) has been found in various bacterial species, less is known about phylogeographic distribution and phenotypic variance of different genetic variants. Methods Herein, we conducted a multiregional whole-genome sequencing study of tet(X)-positive Acinetobacter isolates from human, animal, and their surrounding environmental sources in China. The molecular and enzymatic features of tet(X) variants were characterized by clonal expression, microbial degradation, reverse transcription, and gene transfer experiments, while the tet(X) genetic diversity and molecular evolution were explored by comparative genomic and Bayesian evolutionary analyses. Results We identified 193 tet(X)-positive isolates from 3846 samples, with the prevalence ranging from 2.3 to 25.3% in nine provinces in China. The tet(X) was broadly distributed in 12 Acinetobacter species, including six novel species firstly described here. Besides tet(X3) (n = 188) and tet(X4) (n = 5), two tet(X5) variants, tet(X5.2) (n = 36) and tet(X5.3) (n = 4), were also found together with tet(X3) or tet(X4) but without additive effects on tetracyclines. These tet(X)-positive Acinetobacter spp. isolates exhibited 100% resistance rates to tigecycline and tetracycline, as well as high minimum inhibitory concentrations to eravacycline (2–8 μg/mL) and omadacycline (8–16 μg/mL). Genetic analysis revealed that different tet(X) variants shared an analogous ISCR2-mediated transposon structure. The molecular evolutionary analysis indicated that Tet(X) variants likely shared the same common ancestor with the chromosomal monooxygenases that are found in environmental Flavobacteriaceae bacteria, but sequence divergence suggested separation ~ 9900 years ago (7887 BC), presumably associated with the mobilization of tet(X)-like genes through horizontal transfer. Conclusions Four tet(X) variants were identified in this study, and they were widely distributed in multiple Acinetobacter spp. strains from various ecological niches across China. Our research also highlighted the crucial role of ISCR2 in mobilizing tet(X)-like genes between different Acinetobacter species and explored the evolutionary history of Tet(X)-like monooxygenases. Further studies are needed to evaluate the clinical impact of these mobile tigecycline resistance genes. Supplementary information The online version contains supplementary material available at 10.1186/s13073-020-00807-5.

(Continued from previous page) separation~9900 years ago (7887 BC), presumably associated with the mobilization of tet(X)-like genes through horizontal transfer. Conclusions: Four tet(X) variants were identified in this study, and they were widely distributed in multiple Acinetobacter spp. strains from various ecological niches across China. Our research also highlighted the crucial role of ISCR2 in mobilizing tet(X)-like genes between different Acinetobacter species and explored the evolutionary history of Tet(X)-like monooxygenases. Further studies are needed to evaluate the clinical impact of these mobile tigecycline resistance genes.
Keywords: tet(X), bla NDM-1 , Tigecycline resistance, ISCR2, Acinetobacter species, Flavobacteriaceae bacteria, Ecological niches Background Tetracycline antibiotics have been extensively used in prophylaxis and treatment of human and animal infections, as well as at subtherapeutic levels as growthpromoters in animal feed [1,2]. The third-generation tetracycline antibiotic, tigecycline, is regarded as one of the last-resort antibiotics to treat clinical multidrugresistant (MDR) bacterial infections. It exhibits an expanded spectrum of activities against both Gramnegative and Gram-positive bacteria, including carbapenem-resistant Enterobacteriaceae (CRE) and Acinetobacter baumannii (CRAB), methicillin-resistant Staphylococcus aureus (MRSA), and vancomycinresistant 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 lastresort 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, foodproducing 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.

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 ddH 2 O 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 (~10 5 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 indicus C15 [tet(X3)-and bla NDM-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 nonduplicated 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 timecalibrated 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.

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 eravacyclinesusceptible 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 r 2 = 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].
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].

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.
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 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 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 bla NDM-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 highlevel 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 bla VEB ) 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.
Additional file 1: Figure S1. Sequence alignment of Tet(X) variants by ESPript version 3.0. Figure S2. Structural characteristics of Tet(X5)-like proteins. Figure S3. Resistance rates of the tet(X)-positive Acinetobacter spp. strains against 11 antibiotics. Figure S4. Comparative analysis of the tet(X)-carrying structures. Figure S5. Hypothesized evolution model of tet(X) genes. Table S1. Primers used in this study. Table S2. Prevalence of tet(X) genes in Acinetobacter spp. strains by origin. Table S3. MICs of tetracyclines for the studied strains.