|Home | About | Journals | Submit | Contact Us | Français|
Tetracycline antibiotics are widely used in livestock, and tetracycline resistance genes (TRG) are frequently reported in the manure of farmed animals. However, the diversity of TRG-carrying transposons in manure has still been rarely investigated. Using a culture-free functional metagenomic procedure, combined with large-insert library construction and sequencing, bioinformatic analyses, and functional experiments, we identified 17 distinct TRGs in a single pig manure sample, including two new tet genes: tet(59), encoding a tetracycline efflux pump, and tet(W/N/W), encoding mosaic ribosomal protection. Our study also revealed six new TRG-carrying putative nonconjugative transposons: Tn5706-like transposon Tn6298, IS200/605-related transposon Tn6303, Tn3 family transposon Tn6299, and three ISCR2-related transposons, Tn62300, Tn62301, and Tn62302.
IMPORTANCE Fertilization of agricultural fields with animal manure is believed to play a major role in antibiotic resistance dissemination in the environment. There is growing concern for the possible spread of antibiotic resistance from the environment to humans since genetic resistance determinants may be located in transposons and other mobile genetic elements potentially transferable to pathogens. Among the various antibiotic resistance genes found in manure, tetracycline resistance genes (TRGs) are some of the most common. The present study provides a detailed snapshot of the tetracycline mobilome in a single pig manure sample, revealing an unappreciated diversity of TRGs and potential TRG mobility vectors. Our precise identification of the TRG-carrying units will enable us to investigate in more details their mobility effectiveness.
Upon their discovery in the late 1940s, the broad-spectrum efficiency, ease of use, and relatively low price of tetracycline derivatives placed them among the antibiotics of choice for the treatment of infections inside and outside clinics (1, 2). Their extensive use soon also brought the development of high-level resistance in most human pathogens, and tetracycline use progressively decayed in hospitals, especially after new broad-spectrum antibiotics became available (1). More than 40 distinct tetracycline resistance genes (TRGs) have been identified, classified according to the mechanism of action of their encoded proteins: energy-dependent efflux of the drug out of the cell, ribosomal protection (since tetracyclines bind to the 30S ribosomal subunit to inhibit protein translation), and drug inactivation through enzymatic activity (3). The classical nomenclature has also been challenged by the discovery of mosaic resistance genes resulting from fragment exchange between different genes encoding ribosomal protection (4, 5). Like many other antibiotic resistance genes, TRGs are widely and efficiently disseminated among bacteria through mobile genetic elements (MGEs). Most of them can be found on small nonconjugative transposons that are themselves embedded in conjugative plasmids or transposons (6,–10). However, despite the fairly good understanding of the genetic basis driving tetracycline resistance accumulated over the last 40 years, new TRGs and new TRG-carrying transposons are still continuously discovered (see http://faculty.washington.edu/marilynr and http://www.ucl.ac.uk/eastman/research/departments/microbial-diseases/tn for the most recent updates).
Apart from their use in human medicine, tetracycline antibiotics were and are still broadly used in livestock farms. As expected, the prevalence of tetracycline resistance in animals' gut bacterial communities is very high, leading to the dissemination of TRGs and TRG-carrying MGEs to the environment when manure is used as fertilizer to amend arable soil (11, 12). In turn, contaminated field soil was proposed to act as a reservoir from which resistance determinants can be transferred to animal or human pathogens (11, 13). The TRG pools in manure from various animals have been extensively studied (14,–17), and the presence of TRGs on sequenced plasmids isolated from animals' gut bacteria or manure samples is frequently reported (18,–22). But surprisingly, the diversity of TRG-carrying MGEs that can be found in manure is still largely overlooked. To our knowledge, only one study is available focusing on the tetracycline mobilome of a given manure sample, i.e., the diversity of TRGs and the MGEs that carry them (23).
The present study provides a comprehensive view of the diversity of TRG-carrying MGEs found in a single sample of fresh pig manure, focusing on nonconjugative transposons. By coupling fosmid tetracycline resistance library construction and next-generation sequencing, we confidently identified 17 distinct TRGs, including 2 described here for the first time, along with their genomic context. For 12 of them, location in a complete mobile transposon has been determined, most of them representing as-yet-unknown TRG-transposon associations or even completely new transposons.
The manure sample originated from a single pen containing 6 to 10 healthy pigs from a medium-scale swine farm (200 to 400 animals) in Beijing, China, sampled in June 2014. The manure was collected by scraping the pen's ground with a clean shovel within the day of excretion, immediately sealed in sterile plastic bags, and then stored at −70°C. The amount and variety of antibiotics administered to the pigs were not provided by the farm, but tetracyclines are part of the standard therapeutic cocktail at most Chinese swine farms (2). DNA was extracted from 5 g of frozen manure by a standard phenol-chloroform extraction procedure and randomly sheared. Fragments of ~40 kbp were purified from an electrophoretic gel and inserted into pCCFOS2 vectors carried by the tetracycline-sensitive Escherichia coli EPI-300 host with the CopyControl HTP kit (Epicentre, Madison, WI, USA). The library was then plated onto Luria-Bertani (LB) medium containing tetracycline (100 μg/ml) to select for clones expressing tetracycline resistance. Ninety-one tetracycline-resistant clones were isolated and stored in 30% glycerol for further analysis.
Vector DNA was extracted from each of the 91 clones independently and pooled in a single solution before being sent for sequencing. The pooled solution was submitted to Beijing Aibijingnuo Gene Technology, Beijing, China, and Shanghai Hanyubio Technology, Shanghai, for Illumina HiSeq 125-bp pair-end sequencing and Pacific Biosciences SMRT sequencing, respectively. Around 3.2 million Illumina read pairs and 150,292 SMRT reads were produced, for a total sequencing effort of ~1.35 Gbp. Assuming a size of 30 to 40 kb for each insert added to the ~8.2 kbp of the pCCFOS2 vector, this data set represents a per clone average coverage of 310 to 390×. Four assembly strategies were then concurrently conducted with both sequence data sets. The first assembly consisted of 313 scaffolds produced with the VelvetOptimizer v.2.2.5 wrapper for the Velvet assembly software (24) on pair-end reads only and optimized for N50/length of longest contig. After filling of the residual gaps with GapFiller v.1.11 (25) and correction of the assembled scaffolds, this yielded 19 complete inserts, inferred from the presence of vector sequence at each boundary. The second assembly was built from the PacBio sequences only with the RS_HGAP_Assembly.2 protocol (26) of the SMRT Portal v.2.3.0 software (Pacific Biosciences, Menlo Park, CA, USA) by using standard parameters. Around 120 contigs were produced, from which 10 additional complete inserts were recovered. For the third assembly, we ran the PBcR algorithm implemented in the CA assembler (27), which first corrects SMRT reads with the paired-end read data set and then uses the corrected reads to build an assembly. With a partition option of 50 and a minimal output length option of 500 bp, this strategy yielded one additional complete insert out of the 74 contigs produced. The last strategy consisted of a homemade algorithm with the 21,424 corrected reads produced by the PBcR algorithm and specifically designed to assemble pooled fosmid inserts. Briefly, reads carrying the vector (pCCFOS2) sequence adjacent to the insertion site were used as seeds and sequentially enlarged in the direction of the insert by using other reads, from the longest to the shortest. New reads completely overlapping one of these combined contigs were discarded. When all of the reads were screened, contigs seeded from the 5′ and 3′ boundaries of the insertion site were assembled to recover full inserts. This strategy successfully returned 11 additional complete insert sequences, including those carrying genes highly repeated in our pooled library (plasmid, transposon, and insertion sequence [IS] mobility genes, as well as resistance genes). Illumina pair-end reads were finally mapped back on each complete insert sequence independently to check for and correct possible misassemblies. Unclear mapped regions were also corrected by PCR and sequencing directly from the vector DNA extracted from clones before pooling.
Complete insert sequences were first annotated with the PROKKA pipeline v1.10 (28) with an E value of 0.0001 and the bacterial genetic code and without tRNA and rRNA detection. Antibiotic resistance gene annotations were then recovered with the Resistance Gene Identifier v2.0 online interface (29). Inserts carrying a putative TRG annotation were then selected for a more in-depth annotation. This consisted of systematic BLASTP comparisons of each coding sequence-encoded protein sequence against the RefSeq database and the ACLAME database (30) to recover better functional annotation and against the ISFINDER database (31) in the case of proteins initially annotated as transposases, integrases, recombinases, or resolvases. IS and transposon terminal inverted repeats (TIRs), direct repeats (DRs), and other features were tentatively identified by comparisons of insert sequences or comparisons with known published elements. An insert's genomic origin (chromosome, plasmid, or phage/prophage) was inferred from the annotation of genes found outside detected transposons or transposon fragments and from sequence comparison with the GenBank NR database. Inserts carrying plasmid-related mobility genes and/or with high sequence similarity to known plasmids were considered to be of plasmid origin. Inserts carrying phage-related mobility genes and/or with high sequence similarity to known phages were considered to be of phage/prophage origin. Inserts carrying no obvious plasmid or phage gene and carrying typical chromosomal genes were considered to be of chromosomal origin. Finally, inserts carrying both chromosomal and plasmid or phage genes were considered to be of unclear origin. An insert's taxonomic origin was inferred from comparison of the protein sequence of each gene found outside detected transposons with the GenBank NR database. The dominant taxonomic family over the 20 best hits for most genes was selected as the probable taxonomic origin. When the probable taxonomic origin could not be inferred at the family level without ambiguity, the lowest dominant taxonomic level was chosen.
Each extracted vector DNA was subjected to digestion analysis to assess the insert diversity of our library. Sixteen microliters of DNA was incubated for 45 min at 37°C with 1 μl of EcoRI, 1 μl of SphI, and 2 μl of buffer H and then allowed to migrate for 40 min at 110 V in a 1% agarose gel. Both the enzyme solutions and the buffer were bought from TaKaRa China. The EcoRI and SphI enzymes were selected because they cut the pCCFOS2 vector sequence close to the insertion site only, leading to an easily identifiable 8-kbp band in all clone migrations, and because they produced an exploitable number of bands for most of the clones.
tet(59) and tet(W/N/W) were amplified by PCR from IN-02 and IN-13 clones with the EasyPfu DNA polymerase (TransGen, Beijing, China). The primers used are listed in Table 1. Each clone DNA sample was heat denatured (5 min at 94°C) prior to amplification, and the PCR program was set up as follows: 30 s at 94°C, 30 s at 55°C, and 1 min 20 s [tet(59)] or 4 min 30 s [tet(W/N/W)] at 72°C for 30 cycles, followed by further incubation at 72°C for 10 min. The correct PCR product size was confirmed by electrophoresis through a 1.5% (wt/vol) agarose gel. PCR products were then cloned into the pUC18 vector according to the insertion sites provided in Table 1. E. coli K-12 JW0451-2 and E. coli DH5α were transformed with these recombinant plasmids, and clones were selected by blue-white screening on LB agar plates supplemented with 100 μg/ml ampicillin. The presence of the correct inserted fragment into transformed hosts was confirmed by sequencing.
E. coli K-12 JW0451-2, a derivative of E. coli K-12 susceptible to erythromycin because of inactivation of the acrB pump (32), was used as the expression host in erythromycin resistance tests. E. coli DH5α was used as the expression host in tests of resistance to other antibiotics. These E. coli strains were grown on LB agar plates at 37°C for 16 to 18 h. The resistance phenotype was determined by the microdilution method for oxytetracycline (0.25 to 512 μg/ml), chlortetracycline (0.125 to 256 μg/ml), tetracycline (0.5 to 1024 μg/ml), minocycline (0.125 to 256 μg/ml), tigecycline (0.125 to 256 μg/ml), chloramphenicol (0.5 to 1024 μg/ml), kanamycin (0.5 to 1024 μg/ml), nalidixic acid (0.125 to 256 μg/ml), and erythromycin (0.125 to 256 μg/ml); gene expression was induced with 1 mM isopropyl-β-d-thiogalactopyranoside (IPTG), as recommended by CLSI (http://www.clsi.org/).
Genomic sequences of the tet-carrying inserts analyzed in this study were deposited in GenBank under accession numbers KU736866 to KU736879.
Forty-one complete insert sequences were successfully assembled from the 91 clones expressing a tetracycline resistance phenotype. Digestion patterns of vector DNA extracted from the clones revealed abundant redundancy and suggested that the 41 recovered insert sequences represented the whole library insert diversity. Fifteen of them, ranging from 11 to 41 kb, carried a recognizable tet gene and were thoroughly annotated (Table 2). Two inserts, IN-09 and IN-15, overlap completely and are considered a single insert here to avoid redundancy. An insert's genomic origin was estimated by gene annotation and protein BLAST comparison against the GenBank database. Five of them could confidently be assigned a chromosomal origin, four a plasmid origin, and two a phage origin; there were three whose original location was unclear (Table 2). Three inserts most probably originated from Alcaligenaceae taxa (Betaproteobacteria), three others from Pseudomonadaceae taxa (Gammaproteobacteria), and two from unidentified Gammaproteobacteria and Proteobacteria taxa, respectively (Table 2). Some inserts also probably originated from Firmicutes, with three of them most closely related to Clostridiales sequences and two others to Lactobacilliales sequences. Finally a last insert showed the best sequence homology with the plasmid of an Eggerthellaceae taxon (Actinobacteria). The wide range of inferred taxonomic origins, with many only distantly related to cultivable taxa, and the high number of inserts suspected to be of chromosomal origin indicate that our methodology is suitable for the discovery of new transposons carrying antibiotic resistance genes, which could not have been captured with culture-based or plasmid extraction procedures only.
A total of 14 known tet genes were retrieved, with 9 genes coding for drug efflux pumps and 5 genes coding for ribosomal protection (Table 2). Inserts generally carried a single tet gene, except three of them in which two tet genes were placed one after the other, with a ribosomal protection type followed by an efflux pump type. In addition to these 14 already described tet genes, we identified three genes with substantial similarity to known tet genes. The first two, found on IN-13 and IN-14, are identical and share 65% amino acid identity with tet(D), the closest described gene in GenBank. This new tet gene, named tet(59), provided specific resistance to most tetracycline antibiotic types except minocycline and tigecycline (Table 3). In both inserts, tet(59) is preceded by a homolog of the tetracycline repressor tetR typically found upstream of tet genes encoding efflux pumps and include the two palindromic operator sequences present in all regulatory regions of the tet(A)-tet(R) family (33), suggesting that tet(59) probably belongs to the efflux pump family as well.
The second new tet gene is found in insert IN-02 and shares 96% amino acid identity with tet(W). However, a closer look at the sequence alignment revealed that the differences are not evenly distributed, with most of the protein sequence being more than 97% identical between the two proteins while a small region of 69 amino acid in the middle shows only 74% identity. This region does not show better sequence identity with any of the other known tet genes and thus probably represents a portion of a still unknown tet determinant, combined with tet(W) to form a new mosaic tet gene. According to the nomenclature currently in use for mosaic tet genes (4, 5), the name tet(W/N/W) was assigned to this gene. tet(W/N/W) provides tetracycline-specific resistance when cloned into an E. coli vector (Table 3). Like all other tet(W)-based mosaic genes, tet(W/N/W) probably belongs to the ribosomal protection class of tet genes. Interestingly, 35 sequences from GenBank matched the 69-amino-acid tet(N) central region with more similarity than they matched the same region of tet(W). All of these sequences are currently annotated as tet(W), since amino acid sequences they encode were more than 80% similar to those encoded by tet(W) in the whole protein alignment. Therefore, all of these hits actually represent previously sequenced tet(W/N/W) mosaic genes misannotated as tet(W). Such results suggest that since most TRGs in public databases are annotated according to the sequence similarity along the whole protein, some other mosaic genes have probably already been sequenced but not yet described.
The mobility potential of the detected tet genes was then investigated by analyzing their genomic context in detail. Twelve of them are located in putative complete transposons, with only two for which the association has already been described (or with related transposons) (Table 2). No complete transposon could be identified from the genomic context of the five remaining tet genes, but comparisons with available sequences from GenBank provided some clues about their mobility potential. Only one of them showed an association with a known transposon. In the following, we will first describe the complete and incomplete transposons already known to carry tet genes and then we will focus on tet genes with an unclear mobility pattern and finish with TRG-carrying complete transposons that have not been reported yet.
The first structure similar to a known transposon is the novel tet(H)-carrying transposon Tn6298, a derivative of transposon Tn5706 (34), found on insert IN-01. Unlike Tn5706, Tn6298 includes only a single copy of IS1596 and the tetR repressor gene is truncated (Fig. 1A). Tn6298 shows hallmarks of mobility; it is bounded by 7-pb DRs also detected at the boundary of reference transposon Tn5706 (34), and its upstream and downstream regions in IN-01 are 55 to 85% identical at the amino acid level to a contiguous region in the Pelistega sp. strain HM-7 draft genome sequence (Fig. 1A). The genomic organization in IN-01 thus probably resulted from direct insertion of Tn6298 into the chromosome of an unknown member of the family Alcaligenaceae.
The mosaic tet(W/N/W) gene characterized above was also initially reported in a mobilizable element called ATE-1 (35). A comparison of the original ATE-1 element of Trueperella (Arcanobacterium) pyogenes with insert IN-02 revealed extensive sequence homology upstream of tet(W/N/W), including the mob gene and the oriT region known to be essential for ATE-1 mobility (35), as well as a putative integrase (Fig. 1B). Interestingly, a BLASTN search for insert IN-02 in the NCBI Whole Genome Shotgun nucleotide sequence database returned a region of the draft genome of Mitsuokella jalaludinii with 98% similarity to the first 8 kbp of previously described ATE-1 (Fig. 1B). No other part of IN-02 showed any homology with the M. jalaludinii draft genome, suggesting that this 8-kbp region can be a variation of the mobile ATE-1 element described in T. pyogenes.
The last TRG that shows a known association with a transposon is tet(A) in IN-03. The gene and its associated tetR regulator are included in a 3.7-kbp region 99% similar to Tn1721, a Tn3-like transposon involved in tet(A) mobility in Gammaproteobacteria (9) (Fig. 1C). However, the Tn1721 mobility module (the transposase tnpA and the resolvase tnpR) have been here replaced with those of Tn6162, another Tn3-like transposon only 90% similar to Tn1721 in this region at the nucleotide level. Another part of Tn6162, the orfABCD operon, is also found downstream of the tetR-tet(A) block in IN-03, but the organization is not consistent with those of the complete sequence of Tn6162. This suggests that the structure observed in IN-03 is the result of several transposition/rearrangement events shuffling the various transposon modules and may not be mobile per se.
Among the tet genes with no known associated transposons, four [the two tet(W) genes and the tandem tet(O)+tet(40) genes] are located close to insert boundaries (Fig. 1D and andE),E), preventing any definitive conclusion about their mobility potential. However, each of them is included in genomic regions potentially belonging to as-yet-uncharacterized mobile elements. For instance, the tet(W) gene in insert IN-04 is followed by a gene encoding an S-adenosylmethionine-dependent methyltransferase. This association is a common feature of most of the tet(W) genes sequenced to date (36) and seems to represent the 5′ end of an 11.3-kbp element found in various members of the orders Clostridiales and Burkholderiales (Fig. 1D). Of the 11 or 12 open reading frames located in this element, 4 encode putative plasmid mobility proteins. Although tet(W) in IN-05 is not followed by the characteristic methyltransferase found in the other sequences, the 5′ boundary of the 11.3-kbp element can be identified upstream of the gene (Fig. 1D). The region directly upstream of the 11.3-kbp element boundary in IN-04 and IN-05 did not provided any additional clue about the mobility of this element, with no relevant nucleotide sequence similarity to the GenBank database. Altogether, these observations suggest that tet(W) genes located in IN-04 and IN-05 are or were both part of a small, integrative, nonconjugative plasmid that was disseminated in a wide range of host bacteria.
The tet(O)+tet(40) tandem sequence is included in insert IN-06 in a 5-kbp region 98% identical to a region of the Streptococcus suis SC070731 chromosome (Fig. 1E). This region starts with tnpV, a gene found in Clostridiales transposons encoding a small protein of unknown function (37) and ends with sigB, a gene encoding an RNA polymerase sigma factor. The same sequence, lacking only the tet(40) gene, is also present in an unrelated region of the Clostridium cellulovorans 743B chromosome. This structure may thus represent the minimal element triggering tet(O)+tet(40) mobility. However, comparison with the sequence in which tet(40) was originally described, originating from a human gut bacterial community clone library (38), is not consistent with this hypothesis. In this sequence, tet(O/32/O) replaces the tet(O), tnpV is lacking, and the homology of the downstream sequence with the S. suis chromosome runs for an additional 1 kb (Fig. 1E). It is not clear whether these discrepancies result from inconsistent transposition events or from postinsertion rearrangements, and experimental assays may be required to clarify the exact boundaries of the mobile unit.
The tet(O/W/32/O)+tet(L) tandem arrangement is detected in two inserts, IN-07 and IN-08, ~70% of whose nucleotide sequences are >99% similar but organized differently (Fig. 2A). In IN-08, the central region of IN-07, which includes tet(O/W/32/O)+tet(L), was replaced with a sequence carrying the lincosamide-streptogramin resistance gene lsaA. The replacement took place between two IS1216 elements, suggesting that tet(O/W/32/O)+tet(L) in IN-07 and lsaA in IN-08 are part of two independent IS1216 composite transposons inserted in the same original location. IS1216 elements are related to IS26 elements, which can be inserted into previously existing IS26 though site-specific recombination (39, 40). The two IS1216 composite transposons observed here may thus have been inserted independently into the same original IS1216 element by a similar process. In IN-08, the tet(O/W/32/O)+tet(L) block is located at the insert's 3′ end and is preceded by a region almost identical to those found in IN-07. The similarity ends again at an IS1216 boundary (Fig. 2A), suggesting that this structure is probably a derivative of the composite transposon observed in IN-07.
In insert IN-09/15, tet(A) is included in a new transposon, Tn6303, composed of an IS of the IS200/IS605 family (26% amino acid sequence identity with the IS605 reference protein) and a 3-kbp region including the tetR-tet(A) block and a truncated resolvase-encoding gene, Δres (Fig. 2B). This 3-kbp region shows 87% nucleotide similarity to a sequence found in an Aeromonas salmonicida plasmid and other various Gammaproteobacteria and was probably captured by the IS element. The ~25-kbp sequence surrounding Tn6303 in IN-09/15 is 96% identical at the nucleotide level to a contiguous chromosomal region in Pseudomonadaceae bacterium C6819, providing evidence of insertion (Fig. 2B). IS elements from the IS200/IS605 family transpose through an unusual single-stranded-DNA-mediated mechanism described only recently (41) and are not known to capture additional DNA. Since this IS family is common in some pathogens present in the digestive tracts of humans, like Salmonella species and Helicobacter pylori (42, 43), its potential involvement in antibiotic resistance mobilization deserves further attention.
Another new TRG-carrying transposon in our data set is represented by Tn6299 in insert IN-10. This element, which includes the tet(Y) gene, carries complete Tn3-like transposase tnpA and resolvase tnpR genes and is precisely inserted in a genomic context 60 to 85% homologous at the nucleotide level to a chromosomal region of an uncharacterized Pseudomonadaceae species (Fig. 2C). Characteristic 33/34-bp IRs are present at each transposon's boundary, and 5-bp DRs can be observed at the insertion site, strongly supporting the completeness and mobility potential of this element. Tn6299 is a multiresistance transposon with florfenicol (floR), aminoglycoside (strA and strB), and sulfonamide (sul1) antibiotic resistance genes in addition to tet(Y). A full-length copy of the class I integron integrase intI1 is also present on the transposon but is probably not functional since the attI sequence necessary for cassette insertion is disrupted.
The tet genes in the last four inserts are located in a new family of transposons, related to ISCR2 transposases (44). All of these transposons are composed of the ISCR2 transposase tnpA on one side and of the sulfonamide resistance gene sul2 on the other side (Fig. 2D). Other than this, the inner regions of these elements are completely unrelated. Tn6300, on insert IN-11, carries the TRG gene tet(31), while Tn6301 on insert IN-12 carries tet(Y) and Tn6302 in IN-13 and IN-14 carries the novel tet(59) gene described above. Tn6300 and Tn6302 also have the truncated phosphoglucosamine mutase gene ΔglmM frequently reported in the vicinity of ISCR2 transposases (44). For its part, Tn6301 is a multiresistance transposon with the streptomycin resistance genes strA and strB and the florfenicol resistance gene floR in addition to tet(Y) and sul2. A BLASTN search in GenBank revealed the presence of Tn6300 in the Pseudomonadaceae bacterium E5571 chromosome and the presence of Tn6301 in plasmid pAb559 of Aeromonas bestiarum. In both cases, the sequence similarity extends from 119 bp downstream of tnpA to 304 bp upstream of sul2, with more than 99% nucleotide sequence identity (Fig. 2D). No additional copy of Tn6302 was found in the databases, but when the transposon is compared to Tn6300 or Tn6301 of A. bestiarum, the sequence similarity starts and ends at the exact same 119th and 304th nucleotides downstream and upstream of the tnpA and sul2 genes, respectively (Fig. 2D). Further upstream and downstream genomic contexts are, in all cases, completely unrelated, suggesting that these conserved regions are the actual boundaries of this family of transposons. ISCR (IS common region) elements are ISs of the IS91 family and probably transpose through rolling-circle replication, although their mobility has never been demonstrated (44). Rolling-circle transposition usually involves replication of the element from an oriIS site downstream of the transposase to a tetIS site upstream of the transposase and integration of the synthesized molecule at a new genomic location (45). The terIS site previously suspected for ISCR2 (44) matches the end of the downstream 119-bp conserved region found in our inserts, but we could not detect the ISCR2-inferred oriIS sequence (44), or any sequence sharing some homology, inside the upstream 304-bp conserved region. This suggests that the ISCR2 transposase may somehow recognize a completely unrelated oriIS sequence or that the hypothesis of transposition by rolling-circle replication may not be relevant for this transposon.
The large number of new TRG-carrying transposons and novel tet genes described in this study clearly shows that our knowledge of the antibiotic resistance mobilome is still far from complete, even for a very common class of antibiotics and despite several decades of intense experimental and sequencing efforts. The detailed characterization reported here for the newly identified elements also paves the way for future research. First, although most of them show a clear pattern of transposition, their mobility effectiveness has not been demonstrated. In vitro transposition experiments are required to assess their TRG dissemination potential. Next, investigating the prevalence of these elements in various habitats, such as other manure samples or other host bacterial communities, will enlighten us about their potential role in the spread of tetracycline resistance in the environment.
This work was supported by grants from the National Science and Technology Support Program of China (2013BAD21B02-04) and the National Natural Science Foundation of China (31450110073) and a fellowship for young international scientists from the Chinese Academy of Sciences (2013Y2SB0004).