Molecular Identification of Aminoglycoside-Modifying Enzymes and Plasmid-Mediated Quinolone Resistance Genes among Klebsiella pneumoniae Clinical Isolates Recovered from Egyptian Patients

Inappropriate use of antibiotics in clinical settings is thought to have led to the global emergence and spread of multidrug-resistant pathogens. The goal of this study was to investigate the prevalence of genes encoding aminoglycoside resistance and plasmid-mediated quinolone resistance among clinical isolates of Klebsiella pneumoniae. All K. pneumoniae isolates were phenotypically identified using API 20E and then confirmed genotypically through amplification of the specific K. pneumoniae phoE gene. All isolates were genotyped by the enterobacterial repetitive intergenic consensus polymerase chain reaction technique (ERIC-PCR). Antibiotic susceptibility testing was done by a modified Kirby-Bauer method and broth microdilution. All resistant or intermediate-resistant isolates to either gentamicin or amikacin were screened for 7 different genes encoding aminoglycoside-modifying enzymes (AMEs). In addition, all resistant or intermediate-resistant isolates to either ciprofloxacin or levofloxacin were screened for 5 genes encoding the quinolone resistance protein (Qnr), 1 gene encoding quinolone-modifying enzyme, and 3 genes encoding quinolone efflux pumps. Biotyping using API 20E revealed 13 different biotypes. Genotyping demonstrated that all isolates were related to 2 main phylogenetic groups. Susceptibility testing revealed that carbapenems and tigecycline were the most effective agents. Investigation of genes encoding AMEs revealed that acc(6′)-Ib was the most prevalent, followed by acc(3′)-II, aph(3′)-IV, and ant(3′′)-I. Examination of genes encoding Qnr proteins demonstrated that qnrB was the most prevalent, followed by qnrS, qnrD, and qnrC. It was found that 61%, 26%, and 12% of quinolone-resistant K. pneumoniae isolates harbored acc(6′)-Ib-cr, oqxAB, and qebA, respectively. The current study demonstrated a high prevalence of aminoglycoside and quinolone resistance genes among clinical isolates of K. pneumoniae.


Introduction
Few studies have been performed in Egypt concerning the coexistence of genes encoding aminoglycoside-modifying enzymes (AMEs) and plasmid-mediated quinolone resistance (PMQR) among isolates of Klebsiella pneumoniae. The present study investigated the prevalence and coexistence of seven genes encoding AMEs and nine genes encoding PMQR. Also, clonal relatedness between K. pneumoniae isolates was determined by the enterobacterial repetitive intergenic consensus polymerase chain reaction technique (ERIC-PCR). We found a high prevalence and coexistence 2 International Journal of Microbiology of genes encoding quinolone and aminoglycosides resistance that were heterogenous and mostly clonally unrelated.
The most common mechanism of aminoglycoside resistance arises from enzymatic modification rendering aminoglycosides unable to bind with the aminoacyl site of 16S rRNA with a subsequent failure to inhibit protein synthesis [1]. Modification of aminoglycosides is mediated by AMEs, which catalyze the modification at -OH or -NH 2 groups of the 2-deoxystreptamine nucleus or of the sugar moieties of aminoglycoside molecules [2,3], resulting in reduced or abolished binding of the aminoglycoside molecule to the ribosome. AMEs can be divided into three families: (i) aminoglycoside N-acetyltransferases (AACs), (ii) aminoglycoside O-phosphotransferases (APHs), and (iii) aminoglycoside O-nucleotidyltransferases (ANTs) [4]. Many of the AMEs result in clinically relevant resistance, but only APHs produce high-level resistance [2].
Regarding the quinolones, there are four known mechanisms of resistance that work separately or in combination, resulting in varying degrees of resistance that range from reduced susceptibility to clinically relevant resistance. These mechanisms may be chromosomal or plasmid-mediated [5]. The term "resistance" in the setting of PMQR refers to any increase in minimum inhibitory concentration (MIC) rather than to an increase above a susceptibility breakpoint [6]. Three mechanisms are responsible for PMQR: (i) target alteration by Qnr, (ii) drug modification by the aminoglycoside acetyltransferase AAC(6 )-Ib-cr, which can reduce ciprofloxacin activity, and (iii) efflux pump activation by two quinolone efflux pumps, which are known as OqxAB and QepA [6,7].
Qnr proteins protect DNA gyrase and topoisomerase IV from the inhibitory activity of quinolones [8]. Currently, there are six different qnr genes: qnrA, qnrB, qnrC, qnrD, qnrS, and the most recently reported, qnrVC [6]. The sequences of qnr genes generally differ from each other and from qnrA by 35% [9]. Enzymatic inactivation of quinolones arises from aminoglycoside acetyltransferase [AAC (6 )-Ib-cr], which is a bifunctional variant of a common AAC(6 )-Ib. AAC(6 )-Ib-cr acetylates fluoroquinolones, such as ciprofloxacin and norfloxacin, that have an amino nitrogen on the C7 of piperazinyl ring [10]. Finally, we consider PMQR attributed to efflux pumps that specifically extrude quinolones from bacterial cells. Plasmid-mediated quinolone efflux involves two types of pumps, the quinolone efflux pump (QepA) and the olaquindox (OqxAB) efflux pump. QepA belongs to the major facilitator (MFS) family that decreases susceptibility to hydrophilic fluoroquinolones, especially ciprofloxacin and norfloxacin [7]. The qepA gene is often located on plasmids that encode aminoglycoside ribosomal methylase (rmtB) [6]. The OqxAB pump belongs to the resistance-nodulationdivision (RND) family. The OqxAB pump was first detected on a conjugative plasmid (pOLA52) that was harbored by Escherichia coli strains isolated from swine manure [7,11,12]. The QqxAB efflux pump has wide substrate specificity that includes chloramphenicol, trimethoprim, and quinolones (ciprofloxacin, norfloxacin, and nalidixic acid [13]). oqxAB genes are located on plasmids in clinical isolates of E. coli and on both chromosomes and plasmids of Salmonella spp. and K. pneumoniae. We found that oqxAB genes are commonly located on the chromosome of K. pneumoniae [6,14].

Isolation and Identification of K. pneumoniae Isolates.
All isolates of K. pneumoniae were initially isolated on Mac-Conkey's agar (Oxoid, UK) and then subcultured on eosin methylene blue (EMB) agar (Scharlau, Spain). The isolated strains were identified phenotypically using API 20E (Biomerieux, France) and then confirmed genotypically through amplification of the specific phoE gene using primers and cycling conditions listed in Table 1.

Genotyping of Clinical
Isolates. Clonal relatedness between clinical isolates of K. pneumoniae was determined by ERIC-PCR. The primer was obtained from Macrogen (Korea, Geumcheon-gu, Seoul). Gene amplification was carried out according to cycling conditions as described in Table 1 using Mastercycler5 personal (Eppendorf, California, USA).

Fingerprint Pattern Analysis.
The banding pattern generated by ERIC-PCR was analyzed using BioNumerics 7.5 software (Applied Maths, Kortrijk, Belgium). The PCR fingerprint profile was analyzed using Dice (similarity) coefficient. Cluster analysis was performed based on the unweighted pair group method with arithmetic averages (UPGMA) at position tolerance at 0.15, as previously described [15].
2.5. Antimicrobial Susceptibility Testing. All K. pneumoniae isolates were tested for susceptibility to 23 different antibiotics of several classes. Antimicrobial susceptibility testing was by Kirby-Bauer disc diffusion using Mueller-Hinton agar (MHA) (Oxoid, UK) [16]. Broth microdilution [17] was performed using cation-modified Mueller-Hinton broth (Oxoid, UK) to determine the MIC for the tested antibiotics by the Kirby-Bauer disc-diffusion method. Results were interpreted according to guidelines of the Clinical Laboratory Standards Institute (CLSI) [18]. Both E. coli ATCC 25922 and K. pneumoniae ATCC 700603 were used as quality-control strains.  were resistant to amikacin and/or gentamicin were screened for 7 genes encoding AMEs, namely, aac(3)-II, aac(6 )-Ib, aac(6 )-II, ant(3 )-I, aph(3 )-VI, armA, and rmtB, using primers and cycling conditions listed in Table 1.

Preparation of DNA
Templates. DNA was extracted as previously described by Englen and Kelley [19]. Briefly, three to six colonies of bacterial isolates (depending on colony size) were picked from a nutrient agar plate and suspended in 100 l of DNase-free water in a sterile 1.5 ml microfuge tube to obtain a bacterial suspension equivalent to 1-2 × 10 9 CFU/ml. The bacterial suspension was placed in a boiling water bath for 10 min to lyse the bacterial cells. The lysed bacterial suspension was centrifuged at maximum speed (13,000 ×g) for 3 min. The supernatant, which contains total genomic DNA, was transferred to a new sterile tube using DNase-free tips. DNA was stored in −20 ∘ C.

Fingerprint Pattern Analysis.
A UPGMA dendrogram generated according to Dice (similarity) coefficient revealed that the 85 fingerprint profiles were related to 67 different profiles, including 67 isolates with 18 different combined profiles that included 42 isolates. All genotyped K. pneumoniae were classified into 2 major phylogenetic groups (group A and group B), as shown in Figure 2. Phylogenetic group A included 4 isolates (K184, K109, K162, and K161). Two isolates (K161 and K162) within phylogenetic group A had the same fingerprint pattern. Phylogenetic group B contained the remaining 105 isolates.
Apart from imipenem and meropenem, the MIC 50 and MIC 90 values for other tested -lactams ranged between 32 to 512 g/ml and 256 to >1024 g/ml, respectively. On the other hand, the MIC 50 and MIC 90 for gentamicin and amikacin ranged from ≤0.5 to 8 g/ml and >1024 g/ml, respectively, while MIC 50 and MIC 90 of ciprofloxacin and levofloxacin ranged from ≤0.5 to 1 g/ml and 64 to 128 g/ml, respectively.

Detection of Genes Encoding AMEs. Genotypic results
for AMEs among the aminoglycoside-resistant isolates, as shown in Figure 3, revealed that acetyltransferases were the most prevalent type of AME. The acc(6 )-Ib and acc(3 )-II genes were detected among 88% (58/66) and 58% (38/66) of the investigated isolates (Table 4). In contrast, the acc(6 )-II variant was not detected.
The second most common types of AME were phosphotransferases, followed by nucleotidyltransferases in which aph (3 )   was detected among 14% (9/66) of tested isolates. A rmtB variant was not detected.
The Qnr proteins are considered as one of the three reported mechanisms of PMQR. The qnr genes encode proteins that protect DNA gyrase and topoisomerase IV from inhibition by quinolones and have recently been found worldwide [32]. The current study examined the prevalence of Qnr proteins among K. pneumoniae isolates that showed full or intermediate resistance to quinolones. The qnrA gene was not detected, which was consistent with Yang et al. [32] but differed from a Portuguese study (19% (4/21) of MDR K. pneumoniae isolates [33] and another Egyptian report (12% (14/121) of ESBL-producing K. pneumoniae [34]). Regarding qnrB, we found that 74% (42/57) of the isolates tested positive, which was slightly more than the 50% (11/22) seen in a Korean study [32]. In contrast, a relatively low prevalence rate was reported by Tunisian study (13% (21/165) [35]). We found       that qnrC was represented by only a single isolate, which was consistent with findings from the recent Turkish and Tunisian studies that failed to detect qnrC among quinoloneresistant K. pneumoniae isolates [35,36]. The current study detected qnrD at a prevalence of 40% (23/57); previous work failed to detect this gene [35,37]. The qnrS gene was seen in 49% (28/57) of the investigated K. pneumoniae isolates. Lower incidences (9% (2/22) and 2% (3/165)) had been reported in Korean [37] and Tunisian [35] studies, respectively. A much higher incidence (64%, 28/44) was reported in China [38]. We conclude that plasmids carrying qnr genes were highly spread in Egypt and China, probably due to misuse of quinolones in clinical settings. The current work is the first Egyptian study to investigate the QepA and OqxAB efflux pumps among K. pneumoniae isolates. We found that 12% (7/57) tested positive for qepA, which is far higher than the 2% (5/247) reported previously [39] or the absence of qebA among K. pneumoniae isolates [40]. The prevalence of oqxA and oqxB was higher, 88% (50/57) and 30% (17/57), respectively. Previously Rodríguez-Martínez et al. reported values of 76% (87/114) and 75% (86/114), respectively [41]. Only 26% (15/57) of quinoloneresistant K. pneumoniae isolates were positive for both oqxA and oqxB, double that reported earlier (11% (11/102) [32]). Interestingly, Yuan et al. reported that 100% (154/154) of their K. pneumoniae isolates tested positive for both oqxA and oqxB, suggesting that in that case the genes encoding the OqxAB protein were located on the chromosome of K. pneumoniae, perhaps as a reservoir for these genes [42]. Thus, high resistance rates to quinolones may be expected among K. pneumoniae isolates recovered from clinical settings that frequently prescribe quinolones, since the chromosomal genes coding for OqxAB efflux pump proteins will be overexpressed.
Genotypic identification of K. pneumoniae isolates via amplification of phoE identified 81% (92/114) of the K. pneumoniae isolates; this finding contradicted Sun et al., who reported 100% [21]. This difference may be due to a mutation in the phoE gene of our isolates. Genotyping of K. pneumoniae isolates using the ERIC-PCR technique revealed that the majority of isolates had different origins; 32 isolates were related to 18 different single origins, indicating that the spread of K. pneumoniae among different hospital departments was due to poor infection control.

Conclusion
The current study demonstrated that ACCs were the most prevalent AMEs, followed by APHs and then ANTs. Screening of qnr genes revealed that qnrB was the most prevalent, followed by qnrS. This is the first Egyptian study to detect qnrC and acc(6 )-Ib-cr among quinolone-resistant K. pneumoniae isolates. Genotypic identification of K. pneumoniae through amplification of the phoE gene was not 100%. Most K. pneumoniae isolates included in this study displayed different genetic and phenotypic profiles, indicating different origins of dissemination.

Conflicts of Interest
The authors declare that there are no conflicts of interest regarding the publication of this paper.