Contribution of target alteration, protection and efflux pump in achieving high ciprofloxacin resistance in Enterobacteriaceae

The study aims at revealing the comprehensive contribution of target alteration, target protection and efflux pump to the development of high level of ciprofloxacin (CIP) resistance in Enterobacteriaceae bacteria of environmental, clinical and poultry origins. Antibiotic susceptibility test was used to detect CIP resistant (CIPR) isolates and MICCIP was determined by broth microdilution method. The presence of qnrS gene was identified by PCR and Southern blot hybridization (SBH) confirmed their location. Checkerboard titration demonstrated the effect of NMP on CIP action. PCR followed by sequencing and in silico analysis revealed the contribution of mutations in acrR, marR and gyrA to CIPR development. Out of 152 isolates, 101 were detected as CIPR. Randomly selected 53 isolates (MICCIP 4–512 µg/mL) were identified as Escherichia spp. (26), Enterobacter spp. (7), Klebsiella spp. (5) and Salmonella spp. (15) and of them 31 isolates carried qnrS. qnrS harboring 18 highly CIPR isolates (MICCIP: 256–512 µg/mL) were selected for further study. SBH confirmed 7 isolates harbored qnrS gene in plasmids. The acrA, acrB and tolC were present in all 18 isolates and NMP had an additive (12-isolates) or synergistic (6-isolates) effect on CIP action. Most isolates contained double amino acid (aa) substitutions (S83L and D87N) in QRDR of GyrA resulting in an altered conformation of putative CIP binding pocket. However, some isolates contained single (S83L or S83Y) or no aa substitution but showed high CIPR implicating that the concerted action of three mechanisms is responsible for high CIPR with the most significant role of efflux pump. Electronic supplementary material The online version of this article (doi:10.1186/s13568-016-0294-9) contains supplementary material, which is available to authorized users.


Introduction
Ciprofloxacin (CIP) is a second generation fluoroquinolone and extensively used in the treatment of a wide range of infections caused by Enterobacteriaceae, and Pseudomonas aeruginosa (Kaplan et al. 2013;Oliphant and Green 2002). CIP usually exerts its effect by binding with targets such as DNA gyrase and DNA topoisomerase IV. However, frequent reports about the emergence of CIP resistance have created a conundrum regarding its use (Boyd et al. 2008;Lautenbach et al. 2004). So far, the emergence of resistance to CIP can be attributed to three known mechanisms such as protection of targets with Qnr protein, enhanced efflux pump expression and alteration in the quinolone-resistance determiningregion (QRDR) of target enzymes (Alekshun and Levy 2007;Hooper 2001). Among these mechanisms, target alteration has been reported to be responsible for a high level of resistance to CIP whereas efflux pump and Qnr protein mediated mechanisms attributed to a low level of resistance (Jacoby 2005;Strahilevitz et al. 2009). Most of the previous studies focused on either single mechanism in many organisms or all three mechanisms in a single type (Kuo et al. 2009;Li et al. 2011;Lindgren et al. 2003;Tran and Jacoby 2002;Tran et al. 2005;Vanni et al. 2014). However, high resistance to this drug is emerging swiftly among Enterobacteriaceae leaving this drug ineffective against many infections and increasing the cost of treatment. Furthermore, insufficient comprehensive studies on CIP resistance mechanisms within highly resistant isolates would impede the attempts to increase the potency and decrease the resistance emergence by modifying the existing current drug or designing new one. Therefore, this investigation focused on addressing this fundamental gap in our knowledge by unveiling the contribution of different prevailing mechanisms to the development of high level of CIP resistance among multidrug resistant Enterobacteriaceae bacteria isolated from clinical waste water (CWW), urinary tract infection (UTI) and cloacal swabs of poultry (CSP) origins in Bangladesh for public health interest.

Screening and selection of ciprofloxacin resistant Enterobacteriaceae isolates
A total of 152 presumptively identified MDR Enterobacteriaceae bacteria previously isolated from samples of 3 different origins such as CWW (24 isolates), UTI (61 isolates) and CSP (67 isolates) (Additional file 1: Table S1) were selected for initial screening of CIP resistance by the modified Kirby-Bauer disc diffusion method (Barry et al. 1985) and an organism was reported as susceptible, intermediate or resistant to CIP based on the diameter of zones of inhibition (Cockerill 2011).

Identification of the Enterobacteriaceae isolates
All the isolates were identified on the basis of their growth phenotypes, Gram staining and biochemical properties according to the methods described in the "Manual of Methods for General Bacteriology (American Society for Microbiology (ASM) 1981)". The biochemical results were used to identify the isolates presumptively using the tool BioCluster (Abdullah et al. 2015). The identification of the isolates was further verified by ARDRA (Amplified ribosomal DNA restriction analysis) grouping of 16S rRNA gene amplicons amplified using 27F and 1492R primers (Additional file 1: Table S2). The digestion was done using the AluI (Promega, USA) restriction enzyme. The resulting digestion products were resolved by agarose gel electrophoresis using 1.5% agarose (w/v) gel running for 90 min at 70 V and the gel was viewed using Alpha Imager HP Gel-documentation system (Cell bioscience, USA). The restriction patterns were analyzed to cluster the genetically related isolates using the tool Phoretix 1D (Totallab, UK). The experimental controls used were uncut experimental DNA, digestion of commercially supplied control DNA and no-enzyme "mock" digestion. Two different size markers, 1 kb (Promega, USA) and 100 bp (Promega, USA) DNA ladders were used to analyze different restriction fragments. 16S rRNA gene amplicons of selected isolates representative of each genotype were sequenced followed by phylogenetic analysis to find out their close relatives . The 16S rRNA gene sequences of the selected isolates have been deposited in the Gen-Bank database (accession no. KT825916-KT825923). The GenBank accession numbers of previously identified isolates such as 26N, 28N, E36, E40, G3, G4 and 77 are KC542889.1, KC542890.1, KJ544200.1, KJ544201.1, KJ544205.1, KJ544206.1 and KF188422.1 respectively.

Determination of MIC
The MIC of CIP (Wako Pure Chemical Industries Ltd, Japan) for the selected CIP-resistant Enterobacteriaceae isolates was determined by broth microdilution assay according to the Clinical and Laboratory Standards Institute (CLSI) guidelines (Wikler 2009). Microtiter plates were prepared by double dilution method so that each well of a 96 well microtiter plate contains 95 µL Mueller-Hinton Broth (MHB) and the concentration of the CIP ranges from 512 to 2 µg/mL. In each plate, two negative controls were used; one column contained MHB + 2 µg/ mL ciprofloxacin (blank for the microtiter plate scanner) and another column contained MHB only (sterility control). All the wells in each row were inoculated with 5 µL (McFarland equivalent) of a particular organism except the negative controls. For each isolate, the MIC-CIP was determined in triplicate and the median MIC CIP was recorded. The plate was incubated at 37 °C overnight at 300 rpm in a shaking incubator (WiseCube, Germany). When satisfactory growth was obtained (after 24-36 h) the plate was scanned with a microplate reader (Poweam Medical Systems Co., Limited, China) and the background OD was subtracted from the OD of each well. The bacterial cultures from the wells of microtiter plate were streaked on MHA containing 2 µg/mL ciprofloxacin to check the purity of the isolates.

Screening of qnr gene within the Enterobacteriaceae isolates
Quinolone resistance encoding gene (qnrS) was investigated in selected isolates by PCR with a specific set of primers-qnrF and qnrR (Additional file 1: Table S2) and qnrS positive 18 isolates covering all genotypes with high MIC CIP value (256-512 µg/mL) were selected for determining the location of qnrS gene by Southern blot hybridization. qnrS probe was prepared by PCR amplification and labeled with a PCR DIG-labeling kit (Roche Diagnostics GmbH, Mannheim, Germany) according to the instructions of the manufacturer. Plasmid DNA from the bacterial isolates and marker plasmids of E. coli V 517 were extracted using Wizard Plus SV Minipreps plasmid DNA Purification kit (Promega, USA) and was separated in 0.8% agarose gel at 70 volts for 4 h. After depurination, denaturation, and neutralization of the gel, DNAs were transferred onto a Hybond N + nylone membrane (Nycomed Amershamplc, Buckinghamshire, UK) with a vacuum blotting system for 3-4 h and fixed by UV exposure. The membrane with blotted DNA was sequentially subjected to pre-hybridization and hybridization with a labeled probe. After hybridization, the DIG-High Prime DNA labeling and detection system (Digoxigenin Labeling and Detection Kit; Roche Diagnostics, Mannheim, Germany) was used for signal detection according to the manufacturer's instruction.

Efflux pump mediated ciprofloxacin resistance
The chromosomal DNAs extracted from the selected Enterobacteriaceae isolates were subjected to PCR using primers specific for acrA, acrB and tolC genes encoding AcrAB-TolC efflux pump complex and primers specific for efflux pump regulatory region genes acrR and marR (Additional file 1: Table S2). Mutations within the regulatory proteins were studied after sequencing by bioinformatics analysis of the deduced amino acid sequences (Akter et al. 2012). PCR positive isolates were subjected to MIC CIP determination by microdilution broth checkerboard technique in the presence and absence of an efflux pump inhibitor 1-(1-naphthylmethyl) piperazine (NMP) (SIGMA-ALDRICH, USA) in 96-well microtiter plates (Li et al. 2011). The checkerboard plates were inoculated with 10 5 -10 6 CFU/mL each of bacterial culture and the final concentrations of NMP and CIP ranged from 512-4 µg/mL and 512-2 µg/mL respectively and bacterial growth was monitored after 24-36 h. The OD 600 nm of the plate was taken and the background OD was subtracted from the OD of each well. The interaction between the antibiotic and the inhibitor was interpreted on the basis of fractional inhibitory concentration (FIC) index where FIC indices of <0.5, 0.5 to <4.0 and >4.0 usually refer to synergism, additive and antagonism respectively (Braga et al. 2005;Li et al. 2011;Odds 2003).

Analysis of mutation in gyrA gene
A 648 bp fragment of gyrA gene covering QRDR region (nucleotide position 199-318) of selected Enterobacteriaceae isolates (screened for Qnr and efflux pump) were amplified by PCR using primers gyrAF and gyrAR (Additional file 1: Table S2). The PCR amplicons were purified, sequenced and analyzed to find out amino acid substitutions. Reference amino acid sequences downloaded from NCBI (http://www.ncbi.nlm.nih.gov) (accession no. NP_416734.1, WP_047361088.1, WP_023280374.1, NP_461214.1) were compared with that of test isolates (accession no. KT825924-KT825939) and in silico site directed mutagenesis in a reference sequence (accession no. NP_416734.1) was carried out. Three dimensional (3D) homology models for both the reference and mutated sequences were built using SWISS-MODEL workspace (Arnold et al. 2006;Biasini et al. 2014). The best models determined by GMQE value and QMEAN values were obtained using the template 3lpx.1B which covered 56% of the query sequences with a sequence identity of 77.19 and 76.99% respectively for reference and mutated sequences. The energy minimization in YASARA (http://www.yasara.org/) refined this model and the Ramachandran plot was developed using Accelrys software package Discovery Studio Visual-

Ciprofloxacin resistance in Enterobacteriaceae isolates
Kirby-Bauer disk diffusion susceptibility test revealed that 101 out of 152 Enterobacteriaceae isolates were resistant to CIP in the order-CWW (~96%) > UTI (~72%) > CSP (51%) (Additional file 1: Table S1). Fiftythree Enterobacteriaceae isolates (23, CWW; 15, UTI; and 15 CSP) were selected for further study based on the growth and the biochemical properties, ARDRA grouping, 16S rRNA gene sequencing and phylogenetic analysis. All the analyses corroborated the results and revealed that the isolates representing ARDRA Group I, Group II, Group III and Group IV were closely related to Escherichia spp., Enterobacter spp., Klebsiella spp. and Salmonella spp. (Table 1; Fig. 1a, b).

Contribution of efflux pump on resistance
All 18 isolates displayed positive PCR results for acrA, acrB and tolC genes encoding AcrAB-TolC efflux pump complex of Resistance-Nodulation-Division (RND) family (Table 3; Additional file 1: Figure S1) In all other isolates, NMP had an additive effect (0.5 < FICI ≤ 1) on the action of ciprofloxacin (Additional file 1: Table S3). Inhibition of AcrAB-TolC efflux pump significantly reduced the resistance to ciprofloxacin in selected 18 isolates. Furthermore, to detect the specific mutations in efflux pump regulatory genes, 8 Escherichia spp. (CR1, E34, G4, CR2, 28N, E23, 26N and NCX9) in some of which NMP had synergistic effect (e.g. CR1 and CR2) and in some of which NMP had additive effect (E34, G4, 28N, E23, 26N and NCX9) on the action of CIP were selected for amplification of acrR and marR genes by PCR and sequencing (Table 3). Comparative analysis of amino acid sequences of acrR (accession no. KT825940-KT825947) and marR (accession no. KT825948-KT825955) with that of references (accession no. NP_414997.1; accession no. NP_416047.4) revealed that in acrR gene, Escherichia spp. G4 and 28N contained the same double amino acid substitutions-T213I and N214T; and Escherichia spp. E34, E23 and 26 N contained the same single amino acid substitution-H115Y and in marR gene, all isolates contained the same double amino acid substitutions (G103S and Y137H) ( Table 3). In addition, Escherichia spp. 26N, E23 and E34 contained another amino acid substitution-A53E and Escherichia spp. G4 and 26N also contained another amino acid substitution-K62R in marR gene (Table 3).

Mutations in QRDR of gyrA
A 648 bp fragment of gyrA covering QRDR was targeted to amplify by PCR in 18 selected isolates. However, for  (Table 3).

Discussion
Here, we report a very high and alarming level of CIP resistance in Enterobacteriaceae family of microorganisms, especially in opportunistic pathogens-Escherichia spp., emerging pathogens-Enterobacter spp., and well documented pathogens-Klebsiella spp. and Salmonella spp., belonging to clinical and poultry origin. Furthermore, this investigation conclusively demonstrated that all the 3-types of CIP resistance mechanisms-alteration of target enzyme, protection of target and efflux of the drug, were operative in Enterobacteriaceae isolates to attain the higher resistance. However, in contrast to our current knowledge that efflux pump is usually responsible for low level of CIP resistance (Hooper 2001;Jacoby 2005), this investigation demonstrated that efflux pump can contribute to a high resistance phenotype even in the absence of any mutation in the DNA gyrase subunit A.

Abundance of ciprofloxacin resistance in MDR Enterobacteriaceae isolates
High level of resistance to CIP was found in isolates of CWW and UTI origins which seems cogent, because CIP has been widely used in the treatment of infections caused by both Gram-negative and Gram-positive microorganisms in the hospitals (Adnan et al. 2013;Kaplan et al. 2013). The presence of residual active fluoroquinolones in CWW exerts a selective pressure for the emergence, maintenance and horizontal transfer of resistant genes among microorganisms resulting in a complex resistant situation. The bacteria isolated from CSP also showed higher occurrence of CIP resistance, but MIC CIP value was much lower than CWW and UTI isolates. This is probably due to low dosages of fluoroquinolone antibiotics used in poultry compared to human infection treatment. Salmonella spp. 74 and 77 of CSP origin were exceptional and could withstand very high concentration of CIP (MIC CIP 256 and 512 µg/ mL respectively) which insinuates a threat of the emergence of zoonotic infections. So far we know, there is no well-documented report of very high level of resistance (MIC CIP 256-512 µg/mL) in Enterobacter spp., Klebsiella spp. and Salmonella spp. There are few reports available for Escherichia spp. with high MIC CIP (128-256 µg/mL) (Sato et al. 2013a, b, c) but the molecular mechanisms underlying the ciprofloxacin resistance in them have not been explored in detail (Azmi et al. 2014;Lautenbach et al. 2010;Sato et al. 2013a, b, c;Weigel et al. 1998).  (Kuo et al. 2009). Southern blot hybridization revealed that 7 out of 13 isolates harbored qnrS gene in the plasmids. Three Escherichia spp. isolates (CR1, CR2 and MCX14) along with Klebsiella sp. MCX10 carried qnrS gene in large plasmids of approximately same size (>54.2 kb) which is corroborated by the findings of other researchers (Kuo et al. 2009) but Enterobacter spp. NCX14 harbored the gene in small plasmids (3.9 and 7.0 kb) which could be two different plasmids or the same plasmids with different conformations. Although the presence of qnrS in small plasmid was unusual but not novel. Similar plasmids harboring qnrS was isolated from Salmonella enterica and Aeromonas hydrophila by other researchers (Hammerl et al. 2010;Han et al. 2012). However, Escherichia sp. 26N and Enterobacter sp. MCX6 were shown to carry qnrS in very small plasmids (<2.0 and 2.2 kb respectively) which was not reported earlier. In Klebsiella sp. MCX10, qnrS gene was carried in a large plasmid (>54.2 kb) along with two small plasmids (2.5 and 2.7 kb) which could also be the fragments of the large plasmid or could be acquired through vertical or horizontal transfer. Isolates from which no plasmid could be isolated or the isolates from which plasmids were isolated but did not hybridize with qnrS probe indicated that qnrS might be chromosome-borne. Alternatively, qnrS in these isolates could be borne by episomes, plasmids that can integrate with the chromosome which was reported by other researchers also (Kuo et al. 2009;Strahilevitz et al. 2009).

Effects of efflux pump and its inhibitor on resistance
The selected qnrS positive and highly CIPR 18 isolates were equipped with active AcrAB-TolC efflux pump complex. Checkerboard titration revealed the synergistic effect of NMP on the action of CIP on three Escherichia spp. (CR1, CR2 and CR4, all of CWW origin), two Klebsiella spp., (NCX6 and MCX10) and Enterobacter sp. MCX5 (FICIs were <0.5) that means, the combined effect of NMP and CIP was higher than the sum of the individual effect, i.e. the efflux pump contributed more to the development of resistance to CIP than other two mechanisms. In remaining 13 isolates, NMP had an additive effect on the action of ciprofloxacin (FICIs were 0.5 < FICI ≤ 1.0) which means that efflux pump and mutations in QRDR and/or target protection by Qnr protein, all have significant role in the development of CIPR. Moreover, nonsynonymous mutations in efflux pump regulatory regions (acrR and marR) of different Escherichia spp. indicate that the efflux pump expression might have been increased due to these mutations.

Alteration of the QRDR of GyrA to the development of ciprofloxacin resistance
In this study, it was found that Escherichia spp. E23, E34, 26N, 28N, CR2, CR4, MCX14 and G4; Enterobacter spp. MCX5 and NCX14 and both Salmonella spp. 74 and 77 contained same double amino acid substitutions (S83L and D87N) in the QRDR of GyrA. The nucleotide change within a genus was also the same but within different genera, nucleotide change was different. S83L and D87N amino acid substitutions are reported to be associated with high level of CIPR (MIC CIP ≥ 16 µg/mL) but not with such elevated level (MIC CIP was 256-512 µg/ mL) (Kaplan et al. 2013;Kocsis and Szabó 2013). However, highly CIPR Escherichia sp. CR1 (MIC CIP 512 µg/ mL), and Klebsiella spp. MCX10 (MIC CIP 512 µg/mL) contained same single (S83L) amino acid substitution in QRDR of GyrA; and NMP had synergistic effect on the action of ciprofloxacin for them. In addition, highly CIPR Enterobacter sp. MCX6 contained single amino acid substitution (S83Y) whereas Escherichia sp. NCX 9 had no amino acid substitution in QRDR of GyrA but both isolates showed similar levels of resistance (CIP MIC 256 µg/ mL) and NMP had additive effect on the action of CIP for them. This means that even in the absence of mutation in QRDR, active efflux pump along with qnrS could contribute to high level of CIPR which differs from previous hypotheses that efflux pump and Qnr protein are only responsible for very low level of resistance (Hooper 2001;Jacoby 2005). According to our knowledge, there have been only two reports of fluoroquinolone resistant isolates (one was clinical isolate and another was laboratory derived strain) without any mutations in QRDR, however, they were just exceeding the breakpoint MIC of CIP (whereas MIC CIP for Escherichia sp. NCX9. was 256 µg/ mL) (Chopra and Galande 2011;Sato et al. 2013a, b, c). It was also observed in Escherichia spp. CR1, CR2 and CR4; Enterobacter sp. MCX5, Klebsiella spp. NCX6 and MCX10 that NMP had synergistic effect on the activity of CIP and exhibited the highest level of resistance (MIC CIP , 512 µg/mL) irrespective of single or double amino acid substitutions in QRDR of GyrA; although Escherichia sp. CR1, Klebsiella sp. MCX10 and Enterobacter sp. MCX6 could contain additional amino acid substitution in GyrB subunit or ParC subunit. However, isolates with double amino acid substitutions (S83L and D87N) and with additive effect of NMP on CIP action showed very high level of resistance (MIC CIP 256-512 µg/mL). So it can be inferred that in case of selected Enterobacteriaceae isolates high level of CIPR resulted from the cumulative action of all three mechanisms of resistance to CIP with the mandatory requirement of active efflux pump.
Although the nucleotide variation in QRDR between different species was up to 14.1%, but in comparison to the E. coli str. K-12 substr. MG1665 (NC_000913.3), the variation was mostly 4.4% (Additional file 1: Table S4) and therefore structural analysis was performed based on the amino acid sequence of GyrA of this organism (accession no. NP_416734.1) which is sensitive to CIP and an amino acid sequence derived from this sequence by in silico site directed mutagenesis with two amino acid substitutions-S83L and D87N (most common type of amino acid substitutions found in this study). Based on homology modelling and protein-ligand docking to produce ternary ciprofloxacin-GyrA-DNA complex, it was found that QRDR of GyrA constitutes the quinolone binding pocket and amino acids alteration can diminish the affinity of quinolone binding. It was elucidated that D87N mutation disrupt the salt-bridge formation between D87 and R91 resulting the change in drug binding pocket conformation. But the role of mutation at 83 position which occurred in almost all CIP Enterobacteriaceae isolates and also abundantly reported in literature has not been clear from in silico analysis. Therefore, further analysis to elucidate the role of 83-position mutation is needed for understanding the fluoroquinolone resistance mechanism (Fig. 3).
The present study conclusively demonstrated that Enterobacteriaceae isolates of different sources is being resistant to a very high and clinically significant concentration of ciprofloxacin (MIC ~ 512 µg/mL) by acquiring multiple resistance mechanisms in Bangladesh which has not been previously reported. Furthermore, in contrast to earlier reports, it was observed that efflux pump played a major role in introducing high level of ciprofloxacin resistance in the Enterobacteriaceae isolates, although concerted activity of all three reported mechanisms of