Mobilizable IncQ-Related Plasmid Carrying a New Quinolone Resistance Gene, qnrS2, Isolated from the Bacterial Community of a Wastewater Treatment Plant

Abstract

Plasmid-encoded quinolone resistance was previously reported for different bacteria isolated from patients not only in the United States and Asia but also in Europe. Here we describe the isolation, by applying a new selection strategy, of the quinolone resistance plasmid pGNB2 from an activated sludge bacterial community of a wastewater treatment plant in Germany. The hypersensitive Escherichia coli strain KAM3 carrying a mutation in the multidrug efflux system genes acrAB was transformed with total plasmid DNA preparations isolated from activated sludge bacteria and subsequently selected on medium containing the fluoroquinolone norfloxacin. This approach resulted in the isolation of plasmid pGNB2 conferring decreased susceptibility to nalidixic acid and to different fluoroquinolones. Analysis of the pGNB2 nucleotide sequence revealed that it is 8,469 bp in size and has a G+C content of 58.2%. The plasmid backbone is composed of a replication initiation module (repA-repC) belonging to the IncQ-family and a two-component mobilization module that confers horizontal mobility to the plasmid. In addition, plasmid pGNB2 carries an accessory module consisting of a transposon Tn1721 remnant and the quinolone resistance gene, qnrS2, that is 92% identical to the qnrS gene located on plasmid pAH0376 from Shigella flexneri 2b. QnrS2 belongs to the pentapeptide repeat protein family and is predicted to protect DNA-gyrase activity against quinolones. This is not only the first report on a completely sequenced plasmid mediating quinolone resistance isolated from an environmental sample but also on the first qnrS-like gene detected in Europe.

Quinolones and especially fluoroquinolones are among the most often prescribed antimicrobial drugs worldwide (1). As a consequence, high resistance rates have developed due to the persisting selection pressure. Resistances are mainly attributed to chromosomal mutations in the genes gyrA and parC, which encode, respectively, subunit A of DNA-gyrase (GyrA) and topoisomerase IV (ParC), representing the target enzymes for quinolones (25). Nevertheless, plasmid-encoded quinolone resistance is of great concern since these resistance determinants potentially can be disseminated among bacteria due to plasmid mobility. The isolation of the multiresistance plasmid pMG252 from a clinical strain of Klebsiella pneumoniae in Birmingham, Alabama, in 1994 was the first documented discovery of a plasmid-encoded resistance to quinolones (19). The quinolone resistance gene qnrA, located on pMG252, encodes a 218-amino-acid protein of the pentapeptide repeat family. Members of this protein family are characterized by the repetition of the consensus sequence (C/A)-(D/N)-(L/F)-X-X (3). It was shown that Qnr protects DNA-gyrase and topoisomerase IV activity against inhibition by quinolones (38-40). Recently, other qnr genes, namely, qnrS and qnrB, were identified on plasmids originating from clinical isolates (12, 16).

Several multiresistance plasmids were previously isolated from the activated sludge bacterial communities of wastewater treatment plants by applying, respectively, the exogenous plasmid isolation method (7, 13, 28, 29, 34, 35, 37) and a transformation-based approach (32). Among these, the erythromycin resistance plasmid pRSB101 also confers resistance to the quinolones nalidixic acid and norfloxacin, which could be attributed to the presence of a plasmid-borne multidrug efflux system (33). This finding stimulated the search for further plasmids mediating quinolone resistance in wastewater habitats.

Here we report on an isolation strategy for plasmid-encoded quinolone resistance determinants, even for those that confer low-level resistance. The quinolone resistance plasmid pGNB2 was isolated from the activated sludge bacteria of a municipal wastewater treatment plant. Plasmid pGNB2, conferring quinolone resistance, was further characterized at the genomic level.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

The Escherichia coli strains KAM3 (kindly provided by T. Tsuchiya, Okayama, Japan) containing a deletion in the multidrug efflux gene region acrAB (22), DH5α mcr (11), S17-1 (31), and XL1-Blue (5) were grown at 37°C in Luria broth (LB) medium. Sinorhizobium meliloti 2011 (fdxN::Tc) was grown in TY medium (20). Selection for plasmids that mediate fluoroquinolone-resistance was done on agar plates containing 0.04 μg of norfloxacin per ml.

Isolation of plasmids from activated sludge bacteria.

Activated sludge samples were taken from an activated sludge basin of the wastewater treatment plant Bielefeld-Heepen (Germany) in October 2004. Isolation of plasmids from activated sludge bacteria was done as described previously (36). Briefly, activated sludge bacteria were cultivated on selective LB medium. Growing bacteria were collected, suspended in 70 ml of LB medium, and subjected to total plasmid-DNA preparation using the Nucleobond PC100 kit and AX100 columns according to the protocol supplied by the manufacturer (Macherey-Nagel, Düren, Germany). Plasmid-DNA preparations were subsequently used to transform CaCl₂-competent E. coli KAM3 cells.

Standard DNA techniques.

Determination of the plasmid content of E. coli KAM3 was done by Eckhardt-gel analysis as described by Hynes et al. (14). Plasmid-DNAs from E. coli KAM3(pGNB2) and DH5α strains carrying recombinant plasmids were isolated by using the QIAprep Spin Miniprep Kit (QIAGEN, Hilden, Germany) according to the protocol supplied by the manufacturer.

Antibiotic resistance tests.

Antimicrobial susceptibility testing was done according to the guidelines of the Clinical and Laboratory Standards Institute (15). For disk diffusion assays, the following disks purchased from Oxoid GmbH (Wesel, Germany) were used: amikacin (AK30), ampicillin (AMP10), cefaclor (CEC30), cefuroxime (CXM30), chloramphenicol (C10), ciprofloxacin (CIP5), colistinsulfat (CT10), erythromycin (E10), gentamicin (CN10), kanamycin (K30), nalidixic acid (NA30), neomycin (N30), norfloxacin (NOR5), novobiocin (NV30), rifampin (RD30), spectinomycin (SH10), streptomycin (S10), tetracycline (TE10), tobramycin (TOB10), and trimethoprim (W5). MIC tests were done as described by Dröge et al. (7).

Subcloning and sequencing.

Based on restriction enzyme mapping, restriction fragments of plasmid pGNB2 were subcloned into the sequencing vector pUC19. Sequencing of pGNB2 was done by the IIT-Biotech GmbH (Bielefeld, Germany). For this purpose the pUC19 restriction fragment bank of plasmid pGNB2 was sequenced by using standard sequencing primers and newly designed walking primers. Computer-assisted assembly of sequencing reads was carried out with the CONSED/AUTOFINISH software tool (9, 10). Gap closure and polishing of the sequence was done by applying a primer walking strategy on appropriate restriction fragment clones and on pGNB2 plasmid-DNA. We decided to rely on PHRED 40 quality in the finished pGNB2 nucleotide sequence, which means that less than one mistake is expected in 10,000 sequenced base pairs. Annotation of the finished pGNB2 nucleotide sequence was carried out with the GenDB 2.2 Annotation Tool (21). Repeat regions within the pGNB2 nucleotide sequence were identified and analyzed by using the REPuter software (17).

Tagging of pGNB2 with the Tn5 neomycin resistance gene.

The kanamycin/neomycin resistance gene of transposon Tn5 (4) was cloned on a 3.4-kb HindIII fragment into the single HindIII restriction site located 554 bp downstream of the pGNB2 orfI start codon. The resulting construct, pGNB2-aphII, was used for transfer tests to facilitate selection of putative plasmid-containing transconjugants.

Mobilization of pGNB2 by the mobilisator strain E. coli S17-1.

The pGNB2 derivative pGNB2-aphII was transferred into the mobilisator strain E. coli S17-1 containing an integrated derivative of the IncP-1α plasmid RP4, providing transfer functions in trans by transformation (31). E. coli S17-1(pGNB2-aphII) was mated with the tetracycline-resistant E. coli strain XL1-Blue (5) and the tetracycline-resistant α-proteobacterium Sinorhizobium meliloti 2011 (fdxN::Tc) (20) under conditions described previously (33). Plasmid pGNB2-aphII-containing transconjugants were selected on media containing, respectively, 5 μg of tetracycline ml⁻¹ and 120 μg of neomycin ml⁻¹ for S. meliloti 2011 and 5 μg of tetracycline ml⁻¹ and 50 μg of kanamycin ml⁻¹ for E. coli XL1-Blue. The Tn5 aphII gene encoding aminoglycoside-3′-O-phosphotransferase confers resistance to the aminoglycosides kanamycin and neomycin (4).

Incompatibility testing.

E. coli JM108 carrying RSF1010 (30) was transformed with pGNB2::aphII (selection on LB-agar containing 50 μg of kanamycin ml⁻¹). A total of 24 randomly selected transformants were tested for their plasmid content in Eckhardt gels (14).

Nucleotide sequence accession number.

The pGNB2 nucleotide sequence was submitted to the GenBank database and is available under accession number DQ460733.

RESULTS AND DISCUSSION

Isolation of quinolone resistance plasmids from activated sludge bacteria by applying a new selection strategy.

Plasmids conferring quinolone resistance were isolated from bacteria residing in the activated sludge of a wastewater treatment plant by a transformation-based approach (36) using the E. coli strain KAM3 (22), which carries a deletion in the multidrug efflux system genes acrAB as a recipient. This strain differs from the corresponding parental strain by its hypersensitivity to different substances including certain dyes, quinolones, and fluoroquinolones (22).

Transformation of total plasmid-DNAs from activated sludge bacteria resulted in E. coli KAM3 transformants that grew on LB-agar plates containing 0.04 μg of norfloxacin ml⁻¹. Eckhardt gel analysis of the resulting transformants revealed that some of them contained a small plasmid of the same size (ca. 8 to 10 kb). Three plasmid-containing KAM3 derivatives were chosen for plasmid-DNA isolation and subsequent restriction of the plasmid-DNA with the restriction enzymes BamHI and HindIII. It appeared that all three plasmids possessed identical restriction patterns for the tested enzymes, and therefore only one of them, designated pGNB2, was chosen for further analysis.

A disk diffusion assay revealed that plasmid pGNB2 confers decreased susceptibility to norfloxacin and nalidixic acid but not to amikacin, ampicillin, cefaclor, cefuroxime, chloramphenicol, colistinsulfat, erythromycin, gentamicin, kanamycin, neomycin, novobiocin, rifampin, spectinomycin, streptomycin, tetracycline, tobramycin, and trimethoprim. The MIC for the pGNB2-containing strain was ∼15 times higher for norfloxacin and ∼6.6 times higher for nalidixic acid compared to the recipient strain E. coli KAM3. In addition, pGNB2 confers decreased susceptibility to the fluoroquinolones ciprofloxacin, levofloxacin, ofloxacin, and sparfloxacin (see Table 1).

TABLE 1.

MIC for E. coli KAM3 and KAM3(pGNB2) of nalidixic acid and different fluoroquinolones

Antibiotic	MIC (μg/ml) for E. coli:
	KAM3	KAM3 (pGNB2)
Nalidixic acid	4.25	28
Norfloxacin	<0.03	0.45
Ciprofloxacin	<0.02	0.15
Levofloxacin	<0.02	0.10
Ofloxacin	<0.05	0.20
Sparfloxacin	<0.02	0.15

Complete nucleotide sequence analysis of the quinolone resistance plasmid pGNB2.

The complete nucleotide sequence of plasmid pGNB2 was determined, and it appeared to be 8,469 bp in length with a mean G+C content of 58.2 mol%. Annotation of the pGNB2 nucleotide sequence revealed that the plasmid backbone is composed of two modules for replication initiation and mobilization, respectively. In addition, an accessory element containing a quinolone resistance gene was inserted downstream of the replication gene repC. The genetic map of pGNB2 is shown in Fig. 1, whereas the annotation results are given in Table 2.

FIG. 1.

Genetic map of the quinolone resistance plasmid pGNB2. Coding regions are indicated by arrows giving the direction of transcription. The plasmid is composed of the replication initiation genes repA and repC, the putative mobilization module genes orf1 and mobC, a gene of unknown function (orf2), and the accessory module genes orfI and qnrS2 conferring quinolone resistance. The origins of vegetative (oriV) and transfer replication (oriT) are marked by black circles. The next circle closer to the center represents the G+C plot, where a G+C content of <50% is shown in gray and a G+C content of >50% is shown in black. The G+C plot was generated by using the GenDB (version 2.2) tool. The inner circle gives the scale of the plasmid in base pairs. The HindIII site used for insertion of the Tn5 kanamycin/neomycin resistance gene cassette aphII is shown.

TABLE 2.

Predicted genes on the 8,469-bp quinolone resistance plasmid pGNB2

Gene	Position	Gene product	Function	Pfama	COGb	% Identityc	Reference
repC	545-1426	Putative replication protein C	Replication initiation	06504		91 (RepC of plasmid p22K9 from Klebsiella pneumoniae)	6
repA	1413-2276	Replicative helicase, NTPase	Replication initiation		3598	84 (RepA of plasmid pRAS3.1 from Aeromonas salmonicida)	18
orf1	2347-4788	Relaxase, topoisomerase/primase	Mobilization	03432, 01751		28 (BmgA from Bacteroides fragilis)	2
mobC	4772-5203	Mobilization protein C	Mobilization	05713		32 (MobC of plasmid pLJ42 from Lactobacillus plantarum)	AAZ13605
orf2	5661-6128	Conserved hypothetical protein	Unknown			35 (Lfe203p1 from Leptospirillum ferrooxidans)	23
orfI	6680-7519	Methyl-accepting chemotaxis protein	Putative function in signal transduction		2202	95 (OrfI of Tn1721 on plasmid pFBAOT6 from Aeromonas punctata)	27
qnrS2	7814-8469	Pentapeptide repeat family protein	Quinolone resistance	00805	1357	92 (QnrS of plasmid pAH0376 from Shigella flexneri)	12

^aPfam is a large collection of multiple sequence alignments and hidden Markov models covering many common protein domains and families.

^bThe database of Clusters of Orthologous Groups of proteins (COG) phylogenetically classifies proteins.

^cThe object of the percent identity comparison is given in parentheses.

Plasmid pGNB2 carries a replication module belonging to the IncQ family.

The pGNB2 replication module consists of the IncQ-type replication genes repA and repC. The replicative helicase RepA is a RecA-like NTPase (NCBI Conserved Domains Database accession no. cd01120) that is necessary for plasmid replication. RepA of pGNB2 is closely related (84% identity) to corresponding enzymes encoded on a plasmid-derived genomic island of the pig pathogen Chlamydia suis (8) and on the mobilizable R plasmid pRAS3 from Aeromonas salmonicida subsp. salmonicida (18). Phylogenetic analysis shows that pGNB2 RepA clusters within the tree for RepA proteins of IncQ-family plasmids (not shown). The replication initiation protein RepC of pGNB2 is 91% identical to the RSF1010-type RepC protein of the cephalosporin resistance plasmid p22K9 from a clinical isolate of Klebsiella pneumoniae (6). Replication modules consisting of the genes repA and repC were also identified on the prototype IncQ-like plasmid pTF-FC2 and pTC-F14 originating from, respectively, Acidithiobacillus ferrooxidans and Acidithiobacillus caldus (26).

The pGNB2 origin of vegetative replication (oriV) is located in the intergenic region between orf2 and orfI and consists of a G+C-rich and an A+T-rich region flanked on the one hand side by three direct repeats and on the other hand side by two palindromic motifs. A 36-bp stretch of DNA containing part of the G+C/A+T-rich segment is nearly identical to the corresponding part of the oriV region of the IncQ-like plasmid pTF-FC2 (26). The pGNB2 repeated motifs differ considerably compared to other plasmids belonging to the IncQ family. The incompatibility of pGNB2 with the prototype IncQ plasmid RSF1010 (30) was tested, and it appeared that both plasmids are able to coexist in the same cell (see Fig. 2), suggesting that pGNB2 and RSF1010 belong to different incompatibility subgroups within the IncQ family. The compatibility of different plasmids belonging to the IncQ family is not uncommon. For example, the IncQ plasmids pTF-FC2 and pTC-F14 are compatible with each other (26).

FIG. 2.

Compatibility of pGNB2 and the prototype IncQ plasmid RSF1010. E. coli JM108(RSF1010) (lane 1), E. coli DH5α(pGNB2-aphII) (lane 2), and JM108 transformants carrying both plasmids (lanes 3 to 7) were analyzed in Eckhardt gels. Plasmids RSF1010 and pGNB2::aphII are, respectively, 8,684 and 11,893 bp in size. Coexistence of RSF1010 and pGNB2::aphII in one cell could be shown for 24 randomly selected JM108 transformants.

Plasmid pGNB2 contains a putative mobilization module.

Plasmid pGNB2 harbors a thus-far-unknown putative mobilization module consisting of a mobC-like gene and orf1. The deduced gene product of mobC has a predicted auxiliary function in the relaxosome complex and shows limited similarity (47%) to MobC of the cryptic plasmid pLJ42 from Lactobacillus plantarum (AAZ13605). The deduced gene product of orf1 is 813 amino acids long and possesses an N-terminal relaxase/mobilization nuclease domain (pfam03432) that might function in nicking duplex DNA in the relaxosome complex. The C-terminal part of the gene product contains a topoisomerase/primase (nucleotidyl transferase) domain (Toprim, pfam01751). The N-terminal part of the pGNB2 mobilization protein is homologous (28% identity) to Bacteroides fragilis BmgA, which is encoded downstream of the mobilization gene bmgB on the chromosomal transfer factor cLV25 (2). An IncQ-like origin of transfer (oriT) motif could not be identified upstream of mobC, which is not surprising since the pGNB2 mob genes only show very limited similarity to corresponding genes encoded on IncQ plasmids. In contrast, a putative nicking (nic) site homologous to the nic site of the chromosomal transfer factor cLV25 of Bacteroides fragilis (2) was identified 113 bp upstream of mobC.

The functionality of the pGNB2 mobilization module was tested in mating experiments with the mobilisator strain E. coli S17-1 carrying a chromosomally integrated derivative of the IncP-1α plasmid RP4 that encodes a functional type IV secretion apparatus and conjugative DNA-transfer gene products facilitating mobilization of plasmids possessing an appropriate mobilization site. Plasmid pGNB2 was tagged with the Tn5 kanamycin/neomycin resistance gene aphII and transferred to E. coli S17-1 by transformation. Mating experiments with S17-1(pGNB2-aphII) as a donor revealed that the plasmid can be mobilized to the γ-proteobacterium E. coli XL1-Blue (at a rate of nearly one transconjugant per recipient cell) and to the α-proteobacterium Sinorhizobium meliloti 2011 (rate of ca. 5 × 10⁻² transconjugants per recipient cell). Plasmid pGNB2 is thus mobilizable and can replicate in two different classes of the proteobacteria and therefore was categorized as highly mobile broad-host-range plasmid.

Plasmid pGNB2 contains an accessory module composed of a Tn1721 remnant and the quinolone resistance gene qnrS2.

In the vicinity of the pGNB2 origin of replication (oriV), a remnant of the tetracycline resistance transposon Tn1721 was identified. An 871-bp segment encoding the 5′ half of orfI is 99% identical to the corresponding part of Tn1721. OrfI contains a PAS domain (NCBI Conserved Domains Database accession no. cd00130) that might bind an unknown ligand, thereby acting as a sensor in a signal transduction pathway. Tn1721 OrfI was annotated as a methyl-accepting chemotaxis protein, but its functional context is unknown. The orfI gene product encoded on pGNB2 is truncated. Likewise, both terminal ends of Tn1721 are missing on pGNB2.

Downstream of pGNB2 orfI a gene product that is composed of pentapeptide repeats with the consensus sequence (A/C)-(D/N)-(L/F)-X-X (where X can be any amino acid) is encoded (see Fig. 3). It shows the highest degree of identity (92%) to the quinolone resistance determinant QnrS of the transferable Shigella flexneri plasmid pAH0376 (12) and therefore was designated QnrS2. Plasmid pAH0376 derives from a clinical Shigella flexneri isolate that was discovered after an outbreak of food poisoning in the prefecture Aichi in Japan in the year 2003. The pAH0376 qnrS gene is located close to a Tn3-like transposon, which includes the β-lactamase gene bla_TEM-1. In contrast to most qnrA genes, qnrS of plasmid pAH0376 is not associated with class 1 integrons (41). Likewise, the pGNB2 qnrS2 gene is not part of an integron but is linked to a remnant of a Tn1721-like transposon. The pAH0376 qnrS upstream region, including a putative promoter structure, shows some similarity to the corresponding region upstream of qnrS2 on pGNB2 (64% identity over a length of 191 bp). Interestingly, the G+C content of the orfI-qnrS2 gene region differs considerably from the rest of the plasmid, suggesting that pGNB2 acquired qnrS2 together with Tn1721 orfI. Most probably, qnrS2 had entered pGNB2 as part of a Tn1721-like transposon that was partly truncated after transposition.

FIG. 3.

Sequence comparison of the plasmid-encoded QnrS2 protein of pGNB2 with other Qnr proteins. QnrA1 proteins (AAL60061, AAY46799, AAY46800, AAX18268, AAX18278, AAP20926, AAP20910, and AAW31096) are encoded on plasmids from Klebsiella pneumoniae or Escherichia coli. QnrB2 is encoded on the Citrobacter koseri plasmid pMG301 (DQ351242). Although the host of the plasmid that encodes for QnrS2 is unknown, QnrS1 is encoded on a plasmid from Shigella flexneri 2b (AB187515). Dots indicate identical amino acids compared to QnrA1. Pentapeptide repeats with the consensus sequence (A/C)-(D/N)-(L/F)-X-X that occur at the same position in all Qnr-proteins are marked with asterisks. The glycine residue (G) separating the two domains of Qnr proteins is marked by a vertical arrow.

The origin of plasmid pGNB2 is unknown since it was isolated by a transformation-based approach involving a total plasmid-DNA preparation from activated sludge bacteria. The ancestry of pGNB2 qnrS2 also remains unknown. Recently, Poirel et al. (24) identified the salt and freshwater bacterium Shewanella algae as a reservoir for qnrA-like genes. The deduced amino acid sequence of QnrS2 shows 61% identity to QnrA4 from Shewanella algae.

In summary, plasmid pGNB2 was identified by complementation of the hypersensitive phenotype of E. coli KAM3 carrying a deletion in the multidrug efflux system genes acrAB and represents the first quinolone resistance plasmid that derives from an environmental sample and not from a clinical bacterium. Plasmid pGNB2 is a mobilizable broad-host-range plasmid of the IncQ family that, in contrast to most other quinolone resistance plasmids, carries no other resistance determinants. Isolation of pGNB2 from a wastewater treatment plant bacterial population indicates that dissemination of quinolone resistance plasmids already occurs in wastewater habitats, which in the case of pGNB2 is especially worrisome since the qnrS2 gene is located on a highly mobile plasmid that is able to replicate in very different host bacteria. Thus, the transfer of qnr-like genes to pathogens residing in sewage is conceivable.