A Class 1 Integron Present in a Human Commensal Has a Hybrid Transposition Module Compared to Tn402: Evidence of Interaction with Mobile DNA from Natural Environments

Abstract

In a survey of class 1 integrons from human stools, an unusual class 1 integron from a strain of Enterobacter cloacae was isolated and characterized in detail. Sequence analysis of a fosmid containing the class 1 integron revealed a complex set of transposons which included two Tn402-like transposons. One of these transposons, Tn6007, included a class 1 integron with two non-antibiotic-resistance-type gene cassettes and a complete transposition module. This tni module is a hybrid with a boundary within the res site compared to Tn402, implying that a site-specific recombination event generated either Tn6007 or Tn402. The second Tn402-like transposon, Tn6008, possesses neither a mer operon nor an integron, and most of its tni module has been deleted. Tn6007, Tn6008, and the 2,478 bases between them, collectively designated Tn6006, have transposed into a Tn5036/Tn3926-like transposon as a single unit. Tn6006, Tn6007, and Tn6008 could all transpose as discrete entities. Database analysis also revealed that a version of Tn6008 was present in the genome of Xanthomonas campestris pv. vesicatoria. Overall, the E. cloacae isolate further demonstrated that functional class 1 integrons/transposons are probably common in bacterial communities and have the potential to add substantially to the problem of multidrug-resistant nosocomial infections.

The extensive use of antibiotics in clinical and agricultural settings has been a driving force for the spread of antibiotic resistance genes (8). One significant method for the spread of antibiotic resistance commonly found on resistance plasmids and transposons is facilitated by a genetic element called the class 1 integron (12, 44). The class 1 integron includes a site-specific recombination system capable of integrating and expressing open reading frames contained in structures called mobile gene cassettes. The components of this system include an integrase gene (intI1), whose product is a site-specific integrase, and an attachment site (attI1) at which mobile gene cassettes are inserted. While not part of the site-specific recombination system, class 1 integrons possess a promoter (P_c) that expresses genes within the cassettes. Class 1 integrons, due to their ability to accumulate multiple resistance gene cassettes, play a big role in the dissemination of antibiotic resistance genes in pathogens (46). For the same reason they also contribute greatly to the problem of drug-resistant nosocomial infections (22, 24, 56).

Class 1 integrons are one example of relatively ancient genetic elements that are common in the proteobacteria. While many integron classes show evidence of lateral gene transfer (LGT) over long evolutionary time periods (4, 28), these integrons are not generally immediately mobilizable. All class 1 integrons recovered from clinical isolates, however, are linked to a specific set of transposition functions, the best exemplar of which is Tn402 (40). This family of transposons comprises a suite of four transposition genes (tniR, -Q, -B, and -A; also referred to as the tni module), a resolution site located between tniR and tniQ, and a pair of inverted repeats, designated IRi and IRt, that define the ends of the transposon and are required for transposition (19) (Fig. 1). Transposition of members of this family shows a high level of target site selection. Specifically, the transposition system targets the res sites that are important in the biology of many other transposon types and of many conjugative plasmids. For this reason Tn402 and its relatives have been referred to as res site hunters (29). The ability to target sites characteristic of many transposons and plasmids facilitates the spread of this transposon family by LGT. Thus, members of this family are relatively common in environmental isolates, where they most frequently have a mer operon that confers mercury resistance linked to them (29). When class 1 integrons are recovered from clinical isolates, the integron effectively has replaced the mer operon, and the former is found at the same relative location as the latter (40). Consistent with this res site targeting ability, clinical class 1 integrons have been found to be embedded in a number of different (non-Tn402-like) transposons and/or in a number of different plasmids (25, 34, 37).

FIG. 1.

Alignment of Tn402 and Tn6007. From IRi on, Tn6007 is 99.9% identical to Tn402 until a point between tniR and tniQ, where the level of nucleotide identity drops to 89%. This indicates that there was recombination between two disparate tni modules, with the recombination breakpoint (RBP) occurring at or before the first nucleotide mismatch. The further away from the mismatch, the less likely the recombination breakpoint (indicated by shading in the arrow). The gene cassette entry point into the integron is indicated by filled triangles, and gene cassettes were not included in the alignment. Boxes indicate inverted repeats r1 to r6 in the resolution region, with the recombination point occurring within TT between r1 and r2 (arrow labeled RS).

Tn402 is the best example of a class 1 integron that is also a functional transposon (40) since it has all the transposon functions described above. Most class 1 integrons from clinical isolates were clearly derived from a structure related to Tn402 as they possess many of the features associated with this type of transposon. However, various genetic events, including acquisition of other DNA sequences, deletions, and rearrangements, have resulted in the loss of some transposition functions. Consequently, in a clinical context, a typical class 1 integron is a defective transposon (5). Nonetheless, most of these defective transposons retain IRi and IRt and can still potentially transpose where appropriate transposition proteins are provided in trans (5).

Until recently, it was generally assumed that the Tn402-like class 1 integron arose by a possibly unique event in which a chromosomal integron became integrated into the transposon. With the advent of the antibiotic era, strong selection for antibiotic resistance genes captured by the integron led to very rapid global and promiscuous spread of this element and its descendants. Recent studies have shown that class 1 integrons are very broadly disseminated in bacterial communities (10, 13, 45) and that the lateral spread is not uniquely driven by Tn402-like transposition (45). Notwithstanding this observation, it is still clear that the linkage of a class 1 integron to a Tn402-like transposon was a critical step that contributed greatly to the contemporary problem of multidrug-resistant pathogens and difficult-to-manage nosocomial infections.

Class 1 integrons of the type recovered from clinical isolates are also very common in other environments. Thus, multidrug-resistant class 1 integrons that are defective transposons are routinely recovered from many environments, including wastewater treatment plants (44), meat and produce (7, 51), and even environments relatively remote from the human food chain (26). These observations make it clear that the same class 1 integrons that mediate multidrug-resistant nosocomial infections have become generally widespread, largely as a result of human activities. This broad dispersal of Tn402-like class 1 integrons provides logical routes and mechanisms by which new resistance genes can be recruited from the general environment into the hospital setting (41), although the in situ evidence for this remains largely circumstantial.

In a survey of human stool isolates, a class 1 integron from an Enterobacter cloacae strain was identified. This class 1 integron, designated Tn6007, is one of the few examples of an element that contains a full transposition module and a class 1 integron. However, Tn6007 is a chimera compared to Tn402, meaning that one of these transposons arose from recombination between the other transposon and a so-far-unidentified third member of the res hunter family. The Tn6007 integron contains no antibiotic resistance cassettes but does contain a gene cassette with an unknown function. It also contains a second cassette containing a gene, qacE2b, which encodes resistance to quaternary ammonium compounds. A second transposon (Tn6008) belonging to the Tn402 family with most of the tni module deleted was found adjacent to Tn6007. This element has neither a mer module nor an integron module. We show here that Tn6008 is mobilizable and is present in the genome of the plant pathogen Xanthomonas campestris pv. vesicatoria. Finally, we report that Tn6007 and Tn6008 have transposed into the clinically important transposon Tn5036/Tn3926 or a very close relative as a composite transposon here designated Tn6006. While numerous examples of a class 1 integron inserted into this backbone have been found clinically, the insertion point found here is different than that described previously. These findings provide direct evidence of mixing of genetic information between clinical and environmental contexts and reinforce the point that clinical pathogens continue to have access to a large environmental genetic resource.

MATERIALS AND METHODS

Bacterial strains, plasmids, and growth conditions.

A list of all strains and plasmids used in this study is provided in Table 1. Bacterial isolates containing class 1 integrons were isolated from healthy human fecal samples using a method modified from the method of Barlow et al. (2). Briefly, 5 g of human fecal material was homogenized in 50 ml of buffered peptone water (BPW) and enriched for 6 h at 37°C without agitation. A 100-μl aliquot of an enrichment was reenriched in 100 ml of BPW overnight at 37°C without agitation. Enrichments were screened by PCR for the presence of intI1 as described by Barlow et al. (2). Dilutions of intI1-positive enrichments were plated onto MacConkey agar (Amyl Media, Australia), and between 100 and 200 single-colony boiled lysates were screened for the presence of intI1 by PCR. E. cloacae JKB7 was isolated from a BPW fecal enrichment using no antibiotic selection. All Escherichia coli strains used in the transposition assay and for fosmid library construction were routinely grown at 37°C under aerobic conditions in Luria-Bertani (LB) broth (43).

TABLE 1.

Strains used in this study

Strain or plasmid	Relevant characteristics	Reference or source
Enterobacter cloacae JKB7	Isolated from a healthy human fecal sample	This study
Laboratory strains
Escherichia coli EPI300-T1R	F−mcrA Δ(mrr-hsdRMS-mcrBC) φ80dlacZΔM15 ΔlacX74 recA1 endA1 araD139 Δ(ara leu)7697 galU galK λ−rpsL nupG trfA tonA dhfr Strr Tpr	Epicentre Biotechnologies
Escherichia coli UB5201	F−pro met recA56 gyrA Nxr
Plasmids
pJKB7eC5	pCC1FOS vector containing 29,859-bp insert from E. cloacae JKB7; the insert contains Tn6005 and surrounding sequence	This study
pUB307	Tcr Kmr	3
pRMH236	Spr	6
pSU2056	Apr	27

Antimicrobial agents and antibiotic susceptibility assay.

The following concentrations of antimicrobial agents were used: spectinomycin (Fluka), 50 μg/ml; chloramphenicol (Sigma), 12.5 μg/ml; nalidixic acid (Sigma), 25 μg/ml; kanamycin (Sigma), 50 μg/ml; and HgCl₂ (BDH Chemicals), 4 μg/ml. The antibiotic susceptibility assays for E. cloacae JKB7 were conducted using the BD Phoenix automated susceptibility system.

Conjugation assays.

Conjugation to create the donor strains for transposition assays was performed by concentrating mid-log-phase cells of the donor and the recipient threefold, mixing them at a ratio of 10:1, and incubating the mixed culture at 37°C for 2 h. Transconjugants were selected by plating 150 μl of the mating mixture on an LB agar plate supplemented with appropriate antibiotics. Conjugation experiments to select for transposition events were performed by plating 200-μl aliquots of a 1:1 (vol/vol) mixture of saturated donor and recipient strain cultures onto LB agar and incubating the plates at 37°C for 24 h. Subsequently, the mating mixture was recovered from the plates with 1 ml of LB broth, collected in an Eppendorf tube, and vortexed to resuspend the mating mixture. Serial dilutions of the mating mixture were then plated onto LB agar plates containing the appropriate antibiotic(s).

PCR and sequencing methods.

All primers used in this study are shown in Table 2. The standard PCR was performed using the PCR master mixture (Promega) containing 25 U/ml of Taq DNA polymerase, 200 μM of each deoxynucleoside triphosphate, and 1.5 mM MgCl₂. Primers were used at a final concentration of 0.5 μM. All PCRs were performed by using 30 cycles of denaturation at 94°C for 30 s, annealing at the appropriate temperature for 30 s, and extension at 72°C for 1 min. The intI1 PCR was conducted by using an annealing temperature of 60°C. Mapping of transposition events into pUB307 was conducted by PCR utilizing primers that target the resolution region (Table 2) and ends of the transposon(s) (Table 2) at an annealing temperature of 59°C. The E. cloacae isolate was typed by sequencing a portion of the rpoB gene using primers cm7 and cm31b (31). PCR products were purified using the Wizard SV gel and PCR clean-up system (Promega) prior to sequencing. DNA sequencing reactions were performed at the Macquarie University sequencing facility using dye terminator technology.

TABLE 2.

Primers used in this study

Primer	Sequence (5′-3′)	Target
HS463A	CTGGATTTCGATCACGGCACG	intI1
HS464	ACATGCGTGTAAATCATCGTCG	intI1
FP	GGATGTGCTGCAAGGCGATTAAGTTGG	pCC1FOS sequencing primer
RP	CTCGTATGTTGTGTGGAATTGTGAGC	pCC1FOS sequencing primer
HS902	AATCCTGGCGGATTCACTAC	IRi end of Tn6006/Tn6007
HS903	TCGAGATTGGTGCAGATGAC	IRt end of Tn6007
HS904	TGGACAAGGCTGAGTCAGTG	IRi end of Tn6008
HS905	ACCAGGCGTTACATCCAGTC	IRt end of Tn6006/Tn6008
HS906	TAACGGTCGGTGTCCTTCTC	res region of pUB307
HS907	TGAGCTTGTGGAAGTGTGCT	res region of pUB307
HS908	AGTTCCCGCTCGAATTGC	tniB
HS909	GACGGTCACCAGAAGAACCC	JK008
HS910	GGTGCTGGGTTTGTGACG	tniA (sequencing primer)

Fosmid library construction and screening.

A fosmid library was constructed from E. cloacae JKB7 genomic DNA using a CopyControl fosmid library production kit (Epicentre). Genomic DNA (gDNA) was extracted from overnight cultures using the XS buffer protocol (49), treated with RNase A (Sigma), and further purified by phenol-chloroform extraction. Fosmid library production relies on shearing of the gDNA into approximately 40-kb fragments due to the size restriction of the packaging phage capsids. After gDNA purification, pulsed-field gel electrophoresis was performed using a 1% agarose gel to ensure that the gDNA was sheared sufficiently to obtain approximately 40-kb fragments. Sheared DNA was then end repaired, ligated into the fosmid vector, packaged into phage capsids, and transferred into E. coli EPI300-T1^R as recommended by the CopyControl fosmid library production kit manual. E. coli clones carrying fosmids were plated on LB agar plates containing chloramphenicol. Screening of the fosmid library for intI1 was carried out as previously described (45). Fosmids from positive clones were extracted using a Wizard Plus SV Minipreps kit (Promega), and the terminal ends of the insert were sequenced using the FP and RP vector primers. The insert from one positive fosmid, pJKB7eC5, was fully sequenced by Macrogen (Korea).

Cointegrate construction.

To test the transposition mobility of the integron-containing transposon Tn6007 and composite transposon Tn6006, a spectinomycin-selectable marker was introduced into the Tn6007 module. This was accomplished by creating a cointegrate fusion between the integron in pJKB7eC5 and the plasmid pRMH236, which contains three copies of the aadA2 gene cassette. E. coli EPI300-T1^R carrying pJKB7eC5 was transformed with pSU2056 and pRMH236 to induce cointegrate formation. Total plasmid DNA was extracted and packaged using MaxPlax lambda packaging extracts from Epicentre Biotechnologies. Since pJKB7eC5 exclusively contained the cos phage binding site, only pJKB7eC5 and cointegrates could be packaged. The packaged phage was used to infect E. coli EPI300-T1^R cells according to the manufacturer's instructions, and transformants were plated onto LB agar containing spectinomycin and chloramphenicol for selection of cointegrates.

Transposition assay.

The broad-host-range plasmid pUB307 was conjugated into E. coli EPI300-T1^R bearing the pJKB7eC5/pRMH236 cointegrate. Plasmid pUB307 contains a resolution site that is a known suitable target for transposons with Tn402-like transposition functions (18). Since pJKB7eC5 is nonmobile, transposition into pUB307 was detected by conjugation of pUB307 harboring the transposon derivatives into nalidixic acid-resistant E. coli strain UB5201. Selection for pUB307 carrying the desired transposon insertion(s) was carried out on media containing HgCl₂ (for Tn6005), spectinomycin (for Tn6006/Tn6007, although movement of Tn6005 containing these transposons was also selected for), kanamycin (for pUB307), and nalidixic acid (for the recipient strain). The transposition frequency was calculated by comparing the rate of transposition with the rate of conjugation by selecting the entire population of transconjugants generated in the course of the transposition assay on plates supplemented with only kanamycin and nalidixic acid.

Nucleotide sequence accession numbers.

The rpoB sequence of E. cloacae JKB7 has been deposited in GenBank database under accession number EU591508. The sequence of the insert of the JKB7eC5 fosmid has been deposited in the GenBank database under accession number EU591509.

RESULTS

Class 1 integrons are common in commensal bacteria in the human intestinal tract.

As part of a broad survey of the presence of class 1 integrons in human commensal bacteria, fecal samples were provided by 15 healthy volunteers, enriched in media, and screened for the presence of intI1 by PCR. Fecal enrichments from at least nine individuals were positive as determined by this PCR, suggesting that class 1 integrons are common in human commensal populations. One intI1-positive clone was isolated from seven of the nine enrichments and typed by rpoB sequence analysis. Two isolates had identical sequences, resulting in six distinct intI1-positive clonal types identified. Five of the isolates were identified as E. coli, and one was identified as E. cloacae.

E. cloacae is a cause of many types of nosocomial infections, is often multidrug resistant (23), and has a high incidence of integron carriage (52). Also, E. cloacae-mediated infections can be difficult to manage in a hospital context (33), and they have high rates of mortality (17). For these reasons we examined the E. cloacae isolate (designated JKB7) to determine its integron and surrounding genetic context in detail. This was done by construction of a fosmid library from genomic DNA and complete sequencing of an intI1-positive clone.

The class 1 integron from E. cloacae JKB7 is a Tn402-like transposon.

Sequence analysis of fosmid clone JKB7eC5 revealed that the class 1 integron in E. cloacae JKB7 is located within a structure that is analogous to Tn402 in a structural sense and is designated Tn6007 here (Fig. 1). Tn6007 is bounded by the inverted repeats IRi and IRt, which are characteristic of the Tn402 family. A complete tni module is present; however, a single base pair deletion in tniA was observed, implying that this gene was nonfunctional. This deletion in the fosmid clone was independently confirmed by performing PCR of the relevant region and resequencing. However, the mutation appears to have occurred at a stage during or after fosmid library production since the sequence of the Tn6007 tniA gene amplified by PCR from the original E. cloacae JKB7 genomic DNA did not contain the same base pair deletion. So, while the fosmid clone does not have a functional tniA gene, the original source strain does. In Tn6007, the sequence beginning at IRi, excluding the incorporated gene cassettes and extending to approximately the middle of the res site (that is, including tniR but not tniQ), displays 99.9% identity (1-bp difference) with Tn402. From the middle of the res site to the end of IRt, however, the level of DNA identity with Tn402 is approximately 89% (Fig. 1). When the sequence from tniQ to IRt is considered in isolation, this region is different from the sequences of all other known members of the Tn402 family, although it shows the highest level of similarity to the mer operon containing Tn5718 (96%) (Table 3). In contrast, tniR is 100% identical to the corresponding gene in Tn402 (Table 3). Thus, Tn6007 is a chimera compared to Tn402, with one half essentially identical to Tn402 and the other half derived from an element whose closest known relative is Tn5718.

TABLE 3.

Percent identities of the tniR gene and the tniQBA region for different res hunter-type transposons

Transposon	% Identity
	Tn402		Tn6007		Tn5053		Tn5718
	tniR	tniQBA	tniR	tniQBA	tniR	tniQBA	tniR	tniQBA
Tn6007	100	89
Tn5053	77	83	77	84
Tn5718	77	89	77	96	88	84
Tn5058	76	89	76	92	87	83	98	95

The cassette array in the Tn6007 integron comprises two cassettes. The first of these cassettes, designated JK007, includes a gene whose predicted product has conserved domains related to NADPH-dependent flavin mononucleotide reductases, a family of enzymes involved in the reduction of quinones. The biological significance of JK007 in this context is unclear, although in the case of Helicobacter pylori, an NADPH-dependent flavin mononucleotide reductase (not related to the JK007 predicted protein) does provide protection from oxidative stress (54). This may be significant since there is emerging evidence that accessory proteins that help cells defend against oxidative damage may assist pathogens in colonizing their hosts (9). No cassette of this type has been observed in class 1 integrons previously. The second cassette is 99% identical (it differs at two positions) to another cassette which includes a gene designated qacE2 (accession number DQ462520). Highly similar versions of this cassette have been recovered previously from clinical isolates (16, 21, 48). Here we designate the gene qacE2b. The predicted protein, QacE2b, is 77% identical and 87% similar to QacE encoded by Tn402. The Tn402 qacE gene product has been demonstrated to provide resistance to antiseptics and disinfectants by active transport (39). Since the E. coli strain carrying fosmid clone JKB7eC5 was resistant to 25 μg/ml cetyltrimethylammonium bromide (a detergent) and no growth of the control strain was observed (data not shown), it is highly likely that QacE2b also provides protection from antiseptics and disinfectants. However, the fosmid contains many genes, and thus this conclusion should be validated by further experiments.

Tn6007 is part of a larger complex transposon.

Tn6007 is part of a larger transposon designated Tn6006 here (Fig. 2). Within Tn6006 is a region in addition to Tn6007 that is bounded by second copies of IRi and IRt and is designated Tn6008. Tn6007 and Tn6008 are separated from each other by 2,478 bp. The first 100 bases of Tn6008, beginning at the start of IRi, are identical to the bases of Tn402-like class 1 integrons/transposons. After the first 100 bases, the subsequent 3,933 bases, up to a point 1,966 bases from the end of IRt, contain four genes not related to the Tn402 family of transposons. The remaining 1,966 bases include a complete tniA gene related to other members of the Tn402 family of tniA genes. Over its 1,680-bp length, this gene is 99% identical (3-bp difference) to the tniA gene of the mercury resistance transposon Tn5053 (20). The four intact genes in Tn6008 (Fig. 2) are an araC-like transcriptional regulator gene (JK010), an alkylhydroperoxidase-like gene (ahpD), a gene with an unknown function (JK011), and a lysR-like transcriptional regulator gene (JK012). Homologues of ahpD have been demonstrated to assist in protection from oxidative stress and survivability in macrophages (14, 38).

FIG. 2.

Genetic map describing the insert sequence of fosmid JKB7eC5. Tn6006 is a transposon consisting of two modules bound by the Tn402-like inverted repeats IRi and IRt, one of which is a Tn402-like class 1 integron (Tn6007). The entire Tn6006 module is inserted into a Tn5036/Tn3926 transposon (Tn1696 ancestor) which is inserted into a plasmid. See the text for more information. IR, inverted repeat; DR, direct repeat. Striped arrows indicate tni genes, dark gray arrows indicate mercury resistance genes, and the black arrow indicates intI1.

Tn6006 is presumed to have inserted into its identified location as a single module; that is, Tn6007, Tn6008, and the sequence between these transposons moved as a single unit. This is the inferred outcome since IRi and IRt of Tn6006 are bounded by a 5-bp duplication of the target site (DR2) (Fig. 2). The order of events leading to the creation of Tn6006 is less clear. Perhaps the most likely explanation, however, is that Tn6007 inserted upstream of JK008 first. Subsequent to this, Tn6008 transposed into a sequence downstream of JK009 with transposition proteins provided in trans by the preexisting Tn6007.

Tn6006 is inserted into a Tn5036/Tn3926-like transposon.

In E. cloacae JKB7, Tn6006 is inserted into the res site of a transposon that is 99% identical to a transposon belonging to the Tn3 family designated Tn5036/Tn3926. Here, we define this transposon along with its Tn6006 insert, Tn6005. Like Tn6005, the multidrug resistance transposon Tn1696 contains a class 1 integron inserted into the res region between the tnpR and merE genes (11, 42). Previous studies have determined that Tn1696 was generated from insertion of the integron In4 into Tn5036/Tn3926 or into a close relative (34).

In addition to Tn1696, many other examples of a class 1 integron inserted into a Tn5036/Tn3926 backbone are known. These other examples (Fig. 3) have different (mostly) antibiotic resistance cassette arrays. They are similar, however, in that the associated integron is the clinical type with partial loss of tni genes downstream of the cassette array. In addition, in these examples, the point of insertion of the integron into the backbone is the same (Fig. 3). It is probable that these different isolates are descended from a single integron transposition event since the different isolates are for the most part different only in their cassette arrays. Given the bias of transposition toward res sites, however, the possibility that the site seen here is a preferred insertion site cannot be formally excluded. Tn6005, however, must have been derived from a different integron capture event since the insertion point is six bases away from the insertion point seen previously (Fig. 3). There is one other example of an integron inserting into a different position in a Tn5036/Tn3936 context, and that is the class 1 integron in plasmid t-ST4 from Salmonella enterica serovar Typhimurium (accession no. AJ746361) (53). In this case, the integron is inserted 9 bp from the point of insertion in Tn1696 and 3 bp from the point of insertion in Tn6005 (Fig. 3). Clearly, Tn5036 and Tn3926 and their relatives are continuing to acquire Tn402-like integrons, which presumably then increases the potential for the lateral transfer of class 1 integrons into other mobile elements. In strain JKB7, Tn6005 appears to have inserted into the backbone of a plasmid since terminal sequencing of one other intI1-positive fosmid from the E. cloacae JKB7 library revealed genes involved in conjugative transfer (data not shown). However, further experimental evidence is required to definitively confirm a plasmid location for Tn6005. Insertion of Tn6005 into its current location occurred by transposition since the element is defined by characteristic direct repeats (DR1), as indicated in Fig. 2.

FIG. 3.

Insertion of class 1 integrons into the res region of the Tn1696 ancestor Tn5036. Most of the Tn1696-like clinical class 1 integrons have the same insertion point, indicating that they are all related and descendants of a common ancestor. The class 1 integron from E. cloacae JKB7 is inserted at a position distinct from Tn1696-like integrons, indicating that there was a separate insertion event. There is evidence for a third class 1 integron insertion event in the Tn1696 ancestor in plasmid t-st4. The number in parentheses indicates the length of a sequence not shown. The underlined CAA triplet indicates the TTG start codon (complementary strand) for tnpR. Arrows indicate class 1 integron insertion points.

Transposition activity of transposons in E. cloacae JKB7.

The transposition frequency of Tn6005 was determined using the conjugative plasmid pUB307 as a target and Hg^r as a selective marker (see Materials and Methods). The mean transposition frequency was 1.2 × 10⁻³ (three replicates). Thus, the Tn5036/Tn3926-like transposition system in Tn6005 is functional.

We were also interested in examining the transposition proficiency of the Tn6007 transposition system given its chimeric structure compared to Tn402. While Tn6007 appears to have a fully functional transposition module in E. cloacae JKB7, in pJKB7eC5 tniA contains a single base pair mutation. Unfortunately, pJKB7eC5 was the only intI1-positive fosmid clone that contained the full Tn6007 module. Since Tn6008 contains a complete tniA gene, we hypothesized that this would compensate for the Tn6007 tniA mutation in trans. Assuming that transposition is mediated in a way similar to that of other members of the Tn402 family, res sites are likely transposition targets. Despite the highly selective nature of transposition events, the frequency of movement can be very high when there is a suitable target. One such target is the res site associated with IncP plasmids, and pUB307 is an example of such a plasmid. In the genetic context in which Tn6006/Tn6007 is found, it was not possible to measure rates of transposition since the linkage to Tn6005 made distinguishing between mediating transposition modules difficult in conduction assays. Nonetheless, we were able to demonstrate that the Tn6007 transposition module is mobile. This was done, first, by creating spectinomycin-resistant cointegrates between Tn6007 and the aadA2 gene cassette carrying plasmid pRMH236 (see Materials and Methods). The cointegrate donor then allowed selection for spectinomycin-resistant transposition events in the conduction assay. The mean rate of transposition was found to be 1.3 × 10⁻³ (three replicates), although, as noted above, this could have been derived in part (or in whole) from the Tn5036/Tn3926-like transposition system. To assess the activity of the Tn6007-associated transposition system, at least in a qualitative sense, a number of independent cointegrates were purified, and derived DNA was subjected to PCR with a series of PCR primer pairs designed to detect the presence of DNA inserted at the pUB307 res site (Fig. 4). PCR products consistent with insertion of both a Tn6006 module and a Tn6007 module were readily detected. For one of each of these potential events the PCR products derived from both ends were sequenced to obtain a single example of each type of event. The sequence generated was consistent with a Tn402-like transposition event that included a direct repeat target site duplication when both ends of the same cointegrate were compared. Thus, this hybrid transposition system is functional and can mediate transposition of both the Tn6007 module alone and the larger Tn6006 module.

FIG. 4.

Map of insertion of Tn6006, Tn6007, and Tn6008 into the “multiple resolution site” of pUB307. The boxes with vertical lines indicate the IRi ends of the transposons, the boxes with horizontal lines indicate the IRt ends of the transposons, and the dotted boxes indicate the location of the integrase in the transposons. The relative positions of IRi and IRt in the transposons relative to HS906 and HS907 determine the orientation in which the transposons have inserted into the pUB307. Arrows associated with the primers indicate the direction of amplification. The resolution and site-specific recombination sequences are indicated by bold italics.

In carrying out the PCR experiments with cointegrates, we found at least one example that, despite generating a product across the IRt boundary of Tn6006/Tn6008 using the HS905/HS907 primer pair, did not generate a product with the Tn6006/Tn6007 IRi boundary primers (Fig. 4). We therefore designed an additional primer that targeted the IRi end of the Tn6008 module and carried out a PCR with this primer (HS904) and primer HS906, from which a product was generated. Sequencing of the two products confirmed that the Tn6008 module alone had transposed into the pUB307 res site. A 5-bp duplication of the target site confirmed that a Tn402-like transposition event occurred. This was unexpected since it is the Tn6007 module, not the Tn6008 module, that carries the aadA2 gene and thus encodes spectinomycin resistance. Since the cointegrate was also Hg^r, we presumed that Tn6005 must also be located somewhere within pUB307. If this is so, Tn6008 is likely to have transposed at some point after pUB307 moved to the recipient. Plate mating was carried out overnight (approximately 16 h), so there was ample time for such events to occur. Notwithstanding that we could not determine the frequency of Tn6008 transposition, the fact that we could observe such events at all is interesting. Tn6008 lacks its own functional transposition system and is dependent on Tn6007 to provide the required proteins in trans. Thus, it is likely that any sequences between intact IRi and IRt inverted repeats have the potential to be mobilized in the appropriate circumstances. Further, the Tn6007 tni module is a hybrid compared to Tn402, and this implies that this family of transposons may be relatively promiscuous, potentially allowing transposition in trans even among different members of the family.

A rearranged Tn6008 is present in the X. campestris pv. vesicatoria chromosome.

The sequences comprising Tn6008 are also present in the chromosome of X. campestris pv. vesicatoria; however, they are split into two regions separated by 20,700 bp, and one region is inverted with respect to the other (accession number AM039952; BLASTN hit to nucleotides 2,652,713 to 2,656,474 and 2,677,175 to 2,679,416) (see Fig. S1 in the supplemental material). Both of these regions from X. campestris pv. vesicatoria are 99% identical to Tn6008 (altogether there are differences in 48 bp). This region of the X. campestris pv. vesicatoria chromosome shows evidence of LGT generally, but, since Tn6008 has been identified, the relationship of the corresponding sequences in the chromosome can potentially be explained by a series of transposition events, a deletion event, and an inversion event (see Fig. S1 in the supplemental material). These data further demonstrate that Tn6008 is mobile due to its presence in different genetic contexts.

DISCUSSION

Class 1 integrons are a major medical problem since in a clinical context, they often help bacteria that cause nosocomial infections to manifest a multidrug-resistant phenotype. Most clinical-type class 1 integrons are descendants of a class 1 integron that are associated with a specific suite of transposition functions. These descendants have acquired features that have concomitantly led to the loss of some transposition genes, thereby making them defective transposons (5). The reference integron/transposon that structurally resembles the functional ancestor is Tn402, which was first reported in 1994 (40) and until recently was the only such example. The Tn402 class 1 cassette array includes a dfrB3 cassette that confers trimethoprim resistance (40). Functional transposons/integrons are still quite rare, although several additional examples have been reported recently (47, 50). These examples were obtained from clinical environments. They include the cassettes aacC7, bla_VIM-2, dfrB5, and aacC6-II from an Indian Pseudomonas aeruginosa isolate and aacA7, bla_VIM-2, and dfrB5 from P. aeruginosa isolates in the United States and Russia. In particular, these clinical isolates highlight the fact that multidrug-resistant class 1 integrons that are also functional transposons may be spreading rapidly (50). It remains to be seen whether the recent, apparently rapid appearance of resistance cassettes in functional transposons/class 1 integrons indicates that there is an emerging “new wave” of class 1 integron structures in clinical isolates.

Here we identified Tn6007, another example of a class 1 integron that is also a functional transposon. This element is noteworthy compared to those seen previously for a number of reasons. First, the clone containing Tn6007 was recovered in the absence of antibiotic selection. Consistent with this, the integron does not contain antibiotic resistance cassettes and the isolate is sensitive to all antibiotics tested except some members of the β-lactam family (see Table S1 in the supplemental material). However, the resistance profile is characteristic of the presence of ampC, a gene commonly located in the chromosome of Enterobacter strains (1), of which JKB7 is an example. Second, Tn6007 and its surrounding sequence and cellular context have features that imply sourcing of DNA from both the natural environment and clinical type contexts. These features include the fact that the host was isolated as a human commensal, the fact that Tn6007 is in a species that is actually a common commensal, and the fact that E. cloacae is a known mediator of multidrug-resistant nosocomial infections. Although lacking antibiotic resistance gene cassettes, Tn6007 does contain a cassette with a gene that encodes an antimicrobial membrane Transporter. A highly similar version of this cassette, qacE2b, has also been found inserted into clinical-type class 1 integrons. In contrast, the first of the two cassettes in the array is novel, has no obvious relatives, has not been previously seen in integrons, class 1 or otherwise, and is more consistent with the cassette “type” likely to be recovered from chromosomal integrons or from metagenomic studies (15). In general, clinical-type class 1 integrons totally lacking antibiotic resistance cassettes, where cassettes are present, are highly unusual, if not undescribed. In contrast, cassette arrays with genes that do not encode resistance are the norm in class 1 integrons that are not associated with Tn402-like transposition functions.

The broad family of res hunter-type transposons appears to be widely dispersed in soil bacterial communities (29). Most members of this broader family are generally associated with a mer operon. Since the mer type of transposons has also been recovered from permafrost communities, it has been postulated that the association of mer genes with res hunter-type transposition genes is an association that has been selected for over relatively long periods of time. This hypothesis is supported by the observation that mer-associated transposition modules are relatively divergent (30). In contrast, until now, only two examples of res hunter-type transposition modules for which there is complete sequence information have been reported to be associated with class 1 integrons. These modules are linked to the integrons in Tn402 and plasmid pTB11 (47). Although these two integrons have different resistance cassette arrays, the transposition backbones are identical, implying that they are likely descendants of a very recent common ancestor. In contrast, the tni module in Tn6007 appears to be a hybrid compared to Tn402. The transposon that has the tniQBA component of the transposition module in common with Tn6007 has not been identified yet, but it clusters close to other mer-type members (96% DNA identity in the case of Tn5718). Thus, the so-far-unidentified relative probably also possessed a mer operon. As a consequence, we speculate that the recombination event that led to the formation of the hybrid probably took place in a soil bacterial community, but, irrespective of the actual location, it is clear that different members of the res hunter family are common in bacterial communities. The recombination crossover point that brought the two parts of the hybrid together is located within the res site. It is likely, therefore, that the mediating event was site specific and involved TniR-mediated resolution of the two parents located in the same replicon. Such an event has previously been suggested for evolution of some bla_TEM-containing transposons (36) and for evolution of a transposon in pRMH760 (35). Based on the information available, we concluded only that Tn6007 is a hybrid compared to Tn402. However, it cannot be stated definitively which of these two transposons was one of the two parents and which was the product of recombination.

Tn6007 is closely linked to a second, defective transposon, Tn6008. Tn6008 is of interest as it is linked to Tn6007 such that the two transposons can transpose as a single unit, Tn6006. This is clear from its location in Tn6005, since the signature direct repeats of transposition in the Tn5036/Tn3926 backbone flank the Tn6006 module. The internal structure of Tn6008 is unusual. Tn6008 does not contain any known antibiotic resistance or virulence genes, although it does contain four novel genes. Furthermore, this transposon has never previously been observed in clinical isolates; its only other known location, albeit in rearranged form, is in the genome of the plant pathogen X. campestris pv. vesicatoria. The presence of Tn6008 in two separate locations, however, demonstrates that it is mobilizable when the required transposition proteins can be provided in trans, as transposition assays described here also demonstrated. These observations emphasize that the opportunity to mobilize defective transposons belonging to the Tn402 family is great and helps to explain the broad distribution of the clinical type of class 1 integrons despite the fact that they are defective transposons. Moreover, this type of transposition in trans can probably occur between disparate members of the family since IRi and IRt tend to be relatively conserved. Transposition studies performed here reinforced this hypothesis, since transposition of Tn6007 and Tn6008 was mediated by tni gene products from both modules and, in the case of Tn6008, was detected even in the absence of direct selection.

Tn5036 and Tn3926 and their close relatives contribute significantly to multidrug-resistant nosocomial infections as they have been responsible for the capture and subsequent spread of class 1 integrons and various resistance cassettes on multiple occasions (25, 34). Interestingly, Tn5036 was isolated from an E. cloacae strain obtained from the intestine of a toad, and Tn3926 was isolated from a Yersinia enterocolitica strain found in milk in France (57). One of the major lineages of Tn5036/Tn3926 integron-containing derivatives is that typified by Tn1696 (37), and numerous independent clinical isolates that contain an integron at the same point in the transposon backbone, as observed for Tn1696, have been identified (Fig. 3). Tn6005 represents another example of a class 1 integron in the same Tn5036/Tn3926 backbone; the single important difference is that the insertion point is different, indicating that this transposon must have been derived from a different lineage than Tn1696. This similarly emphasizes the continued evolution of transposition modules occurring outside the clinical context and their ability to make their way into human commensals.

Antibiotic-resistant bacteria that we now know contain class 1 integrons were isolated in the earliest years of the antibiotic era and are still being isolated (32, 55). Largely as a result of this observation, it has been speculated that class 1 integrons may have been widespread in bacteria prior to the antibiotic era (37). It is only recently, however, that more direct evidence for this has been found (45). What is nonetheless still surprising is the possible scale of this spread; the results of quantitative PCR with metagenomic DNA have implied that up to 10% of bacterial cells in typical communities could carry an intI1 gene (13). To date, only a subset of the class 1 integrons, specifically the integrons associated with Tn402-like transposons, have played a key role in the spread of antibiotic resistance. However, evidence that this class 1 subset is also prevalent in general microbial communities continues to emerge, and this may not be solely due to antibiotic era contamination. Complex transposons of the type described here provide more specific evidence that the origin of mobile DNA in clinical isolates, which is an ever-increasing global clinical problem, is the broader environment.