Characterization of qnrB-Like Genes in Citrobacter Species of the American Type Culture Collection

Abstract

Among five American Type Culture Collection (ATCC) Citrobacter strains, qnrB60 in Citrobacter freundii ATCC 6879, an isolate from the preantibiotic era, and qnrB61 in Citrobacter braakii ATCC 51113^T, a type strain belonging to the C. freundii complex, were identified. Meanwhile, a truncated qnrB-like pseudogene was identified in C. freundii ATCC 8090^T and ATCC 43864. No qnrB-like sequence was found in Citrobacter koseri ATCC 27028^T. These findings underscore the close relationship between this species and qnrB.

TEXT

An increasing number of plasmid-mediated quinolone resistance (PMQR) determinants have been reported (1, 2). In most studies, PMQR determinants have been found among multiple-drug-resistant bacteria, especially among extended-spectrum β-lactamase producers. Despite its only recent discovery, qnrB is one of the most prevalent PMQR determinants among clinical isolates of Enterobacteriaceae (3–7). Interestingly, the distribution of qnrB is uneven among species; most new variants of qnrB have been found in Citrobacter spp., and the prevalence of qnrB seems high among certain populations of Citrobacter spp. strains, a substantial part of which were not multidrug resistant (8). One possible explanation is that qnrB was originally present among some populations of Citrobacter bacteria irrespective of quinolone exposure. More recently, the association between qnrB and clinical isolates of Citrobacter spp. has been well documented by Jacoby et al. (8). In the present study, to clarify this linkage more generally and robustly, reference Citrobacter strains of the American Type Culture Collection (ATCC), including type strains, nonclinical isolates, and a strain from the preantibiotic era, were subjected to detection and characterization of the qnrB-like gene (9–11).

(This work was presented in part at the 49th Interscience Conference on Antimicrobial Agents and Chemotherapy, San Francisco, CA, 2009.)

Five Citrobacter strains of the ATCC, including three type strains, were subjected to qnr detection by PCR, i.e., Citrobacter freundii ATCC 8090^T, C. freundii ATCC 43864, C. freundii ATCC 6879, Citrobacter braakii ATCC 51113^T, and Citrobacter koseri ATCC 27028^T (see Table S1 in the supplemental material). Two strains, namely, C. freundii ATCC 6879 and C. braakii ATCC 51113^T, were positive for qnrB with two different primer pairs. qnrA and qnrS were not amplified in any strain. In order to perform cloning of qnrB, a BamHI-digested genomic DNA fragment containing the qnrB-like gene and BamHI-digested, bacterial alkaline phosphatase (BAP)-treated pUC119 (TaKaRa Bio Inc., Otsu, Japan) were ligated with TaKaRa Ligation Mighty Mix (TaKaRa Bio Inc.) and chemically transformed into Escherichia coli TOP10 (Invitrogen, Carlsbad, CA). Transformants were selected on LB-agarose plates containing 0.016 μg/ml of levofloxacin to screen for strains with elevated quinolone resistance and thereafter on plates containing 100 μg/ml of ampicillin (Sigma-Aldrich, St. Louis, MO) to confirm that the resultant clones were truly transformants. We successfully cloned a 5,513-bp DNA fragment containing qnrB61 of C. braakii ATCC 51113^T. qnrB61 was 645 bp, and its upstream and downstream regions were the psp gene cluster and the sap gene cluster, respectively (Fig. 1). QnrB61 differed from QnrB10 by one amino acid (G17S). By I-CeuI digestion, followed by pulsed-field gel electrophoresis and Southern hybridization (12), qnrB61 was shown to be located on the chromosome, but not on the plasmids, of C. braakii ATCC 51113^T (data not shown). Our attempt to clone the qnrB-like gene of C. freundii ATCC 6879 was unsuccessful.

Fig 1.

Genetic map of the region between the psp and sap gene clusters of Citrobacter ATCC strains. Each underlined value is the length (bp) of the qnr-like gene/pseudogene, and its upstream/downstream environments are separated by vertical black bars. In C. koseri ATCC 27028^T, truncated putative open reading frame CKO_01383, but no qnr-like gene/pseudogene, was identified between pspF and sapA.

PCR amplification and sequencing of the region between pspF and sapA confirmed that C. freundii ATCC 6879 harbors a full-length qnrB gene (qnrB60) and that C. braakii ATCC 51113^T harbors qnrB61. QnrB60 in C. freundii ATCC 6879 was found to differ by one amino acid from QnrB38 (A202V). The promoter region and putative SOS box of the upstream region of qnrB60 and qnrB61 were intact. In contrast, the truncated qnrB-like pseudogenes in C. freundii ATCC 8090^T and C. freundii ATCC 43864 (99.6% identity) were found between pspF and sapA; these pseudogenes contained the last 285 bp of the qnrB-like nucleotide sequence (97% identical to qnrB19 and qnrB36) but lacked the first 360 bp of the putative open reading frame in addition to the upstream region, including the promoter and SOS box (Fig. 1). In contrast, there was no nucleotide sequence homologous to qnrB in C. koseri ATCC 27028^T.

To estimate the contribution of the qnrB-like gene to reduction of quinolone susceptibility in E. coli, we constructed vectors expressing each qnrB-like gene under the control of the same promoter and compared the drug susceptibility of transformed E. coli TOP10. A T vector was constructed by PCR with primers pTV-T-EcoRV-F and pTV-T-EcoRV-R (see Table S1 in the supplemental material) and pTV118N (TaKaRa Bio Inc.) as the template, the resultant PCR product was digested with EcoRV, and the DNA fragment was self-ligated and chemically transformed into E. coli. Just before use, EcoRV-digested plasmid DNA with a thymidine base at the 3′ end was added with TaKaRa Ex-taq and dTTP (TaKaRa Bio Inc.) according to a protocol described previously (13). With this T vector, TA cloning was performed for the following PCR products: (i) the region between the psp (phage shock protein) and sap (sensitivity to antimicrobial peptides) gene clusters; (ii) the full-length qnrB-like gene, namely, qnrB60, qnrB61, and qnrB2, or the qnrB-pseudogene (ΔqnrB_{ATCC 8090T}) with an artificial start codon; and (iii) truncated qnrB61 with artificial start and stop codons (see Table S1 in the supplemental material). As expected, the MICs of fluoroquinolones and nalidixic acid were elevated 8- to 32-fold and 2- to 4-fold, respectively, in E. coli expressing a full-length qnrB-like gene (Table 1). The extent of the susceptibility change due to qnrB60 and qnrB61 was comparable to that due to qnrB2. In contrast, the truncated qnrB-like nucleotide sequence of C. freundii ATCC 8090^T, even when the start codon was added, did not affect quinolone susceptibility in E. coli. Consistently, the MICs of quinolones for transformed E. coli were increased only when the genetic region between the psp and sap gene clusters containing the full-length qnrB-like gene was introduced into E. coli.

Table 1.

Drug susceptibilities of the transformants used in this study

Straina	qnrB-like sequencec	MIC (μg/ml)b
		LVX	CIP	NAL
E. coli TOP10		0.008	0.004	2
E. coli TOP10/pTV		0.008	0.004	2
E. coli TOP10/pTV-psp-sap region of C. freundii ATCC 8090T	ΔqnrBATCC 8090T	0.008	0.004	2
E. coli TOP10/pTV-psp-sap region of C. freundii ATCC 43864	ΔqnrBATCC 43864	0.008	0.004	2
E. coli TOP10/pTV-psp-sap region of C. freundii ATCC 6879	qnrB60	0.06 (8)	0.06 (16)	4 (2)
E. coli TOP10/pTV-psp-sap region of C. braakii ATCC 51113T	qnrB61	0.13 (16)	0.13 (32)	4 (2)
E. coli TOP10/pTV-psp-sap region of C. koseri ATCC 27028T	None	0.008	0.004	2
E. coli TOP10/pTV-qnrB2	Full-length qnrB2	0.13 (16)	0.13 (32)	8 (4)
E. coli TOP10/pTV-ATG-ΔqnrBATCC 8090T	ΔqnrBATCC 8090T with artificial start codon	0.008	0.004	2
E. coli TOP10/pTV-qnrB60	Full-length qnrB60	0.06 (8)	0.06 (16)	4 (2)
E. coli TOP10/pTV-qnrB61	Full-length qnrB61	0.13 (16)	0.13 (32)	8 (4)
E. coli TOP10/pTV-qnrB61 (N-0, C-20)	qnrB61 lacking 20 aa in C terminus	0.008	0.004	2
E. coli TOP10/pTV-qnrB61 (N-0, C-40)	qnrB61 lacking 40 aa in C terminus	0.008	0.004	2
E. coli TOP10/pTV-qnrB61 (N-20, C-0)	qnrB61 lacking 20 aa in N terminus	0.008	0.004	2
E. coli TOP10/pTV-qnrB61 (N-20, C-20)	qnrB61 lacking 20 aa in N terminus and 20 aa in C terminus	0.008	0.004	2
E. coli TOP10/pTV-qnrB61 (N-20, C-40)	qnrB61 lacking 20 aa in N terminus and 40 aa in C terminus	0.008	0.004	2
E. coli TOP10/pTV-qnrB61 (N-40, C-0)	qnrB61 lacking 40 aa in N terminus	0.008	0.004	2
E. coli TOP10/pTV-qnrB61 (N-40, C-20)	qnrB61 lacking 40 aa in N terminus and 20 aa in C terminus	0.008	0.004	2
E. coli TOP10/pTV-qnrB61 (N-40, C-40)	qnrB61 lacking 40 aa in N terminus and 40 aa in C terminus	0.008	0.004	2

^aThe pTV vector was constructed on the basis of pTV118N. “N-XX, C-YY” means lacking XX amino acids in the N terminus and YY amino acids in the C terminus.

^bThe n-fold change relative to the MIC of the control transformed strain is shown in parentheses. LVX, levofloxacin; CIP, ciprofloxacin; NAL, nalidixic acid.

^caa, amino acids.

Acquired qnrB expression is known to be regulated by the SOS response in E. coli (14, 15). Therefore, regulation of chromosomal qnrB in Citrobacter spp. was examined. C. braakii ATCC 51113^T was incubated at 35°C with shaking in LB broth. Full-growth medium was diluted by adding a 40-fold volume of fresh medium. After 1 h of incubation, levofloxacin at 1× MIC (0.06 μg/ml), 0.5× MIC (0.03 μg/ml), or 0.25× MIC (0.015 μg/ml) was added, and after an additional 4 h of incubation, 0.5 ml of broth was harvested. Reverse transcriptase PCR was performed with a Qiagen RNeasy kit, RNAprotect bacterial reagents, a DNase kit (Qiagen, Valencia, CA), a cDNA High reverse transcriptase kit (Applied Biosystems, Foster City, CA), and SYBR green ER quantitative PCR Supermix for ABI Prism (Invitrogen), and the product was analyzed with the ABI Prism 7000 sequence detection system (Applied Biosystems). As expected, expression of qnrB61 was increased in parallel with the upregulation of recA and lexA, which are key effectors of the SOS response (Fig. 2). This result was consistent with the presence of an intact SOS box upstream of qnrB61 in C. braakii ATCC 51113^T.

Fig 2.

The change in qnrB61, recA, and lexA transcription due to fluoroquinolone exposure in C. braakii ATCC 51113^T. After 1 h of incubation, levofloxacin was added at a final concentration of 1× MIC (0.06 μg/ml), 0.5× MIC (0.03 μg/ml), or 0.25× MIC (0.015 μg/ml) and the bacteria were incubated for an additional 4 h. The relative expression of each gene was estimated by the ΔΔC_T method with comparison to the drug-free condition. Values are means ± standard errors of the means.

As quinolone is a synthetic antimicrobial agent that was introduced into the clinical setting in the early 1960s, the quinolone resistance determinant was not expected to be prevalent in nature before the development and practical use of quinolones. Our findings in the present study, however, indicated that there should be a revision of this framework; a “functional” qnrB61 gene was identified on the chromosome of the type strain of C. braakii, ATCC 51113^T, which is a member of the C. freundii complex and was isolated from a snake (http://www.atcc.org/) (9). qnrB60 was also found in C. freundii ATCC 6879, which was an isolate from milk set for cottage cheese before 1932 in the preantibiotic era, long before the synthesis of quinolones (11). Moreover, ATCC 8090^T, a type strain of C. freundii isolated before 1931, harbored a truncated qnrB pseudogene (10). Although the psp and sap gene clusters have been identified in a wide variety of chromosomes of Enterobacteriaceae, the presence of a qnrB-like sequence between these clusters seems unusual (16).

Although our results are not conclusive in themselves, others have also suggested that there might be a relationship between the C. freundii complex and quinolone resistance: for example, Jacoby et al. demonstrated the presence of qnrB or its pseudogene in clinical C. freundii complex strains (8). The integrative hypothesis is that a progenitor of the C. freundii complex, but not of other members of the family Enterobacteriaceae, including C. koseri, might have acquired the qnrB-like gene in its chromosomal region between the psp and sap gene clusters in the past. Thereafter, although some C. freundii complex strains have preserved a functional qnrB-like gene to date, others have presumably lost it by deletion, leaving the pseudogene as a trace. Although the transcription of qnrB61 in the chromosome of C. braakii ATCC 51113^T was increased, presumably via the SOS response, by exposure to fluoroquinolone (and this agent might potentially facilitate the selection of qnr-positive strains), the physiological role of qnr would be essentially independent of the presence of quinolone agents, as suggested by a bioinformatic approach (17).

Another finding regarding the function of Qnr in the present study was that both termini of qnrB61 seem to be necessary to confer quinolone resistance on E. coli. This result seems consistent with the study of Guo et al. in which the deletion of only around 10 amino acids from each terminus of QnrA1 abolished quinolone resistance (18).

In conclusion, our results with reference ATCC strains strengthen the relationship between qnrB and Citrobacter spp. more generally and robustly. The physiological role of Qnr in bacteria, as well as the actual role of the development of quinolone resistance, remains to be elucidated by further studies.

Nucleotide sequence accession numbers.

The nucleotide sequences determined in this study have been deposited in the DNA databases and assigned accession numbers AB734052 to AB734056.