Role of Novel Multidrug Efflux Pump Involved in Drug Resistance in Klebsiella pneumoniae

Background Multidrug resistant Klebsiella pneumoniae have caused major therapeutic problems worldwide due to the emergence of the extended-spectrum β-lactamase producing strains. Although there are >10 major facilitator super family (MFS) efflux pumps annotated in the genome sequence of the K. pneumoniae bacillus, apparently less is known about their physiological relevance. Principal Findings Insertional inactivation of kpnGH resulting in increased susceptibility to antibiotics such as azithromycin, ceftazidime, ciprofloxacin, ertapenem, erythromycin, gentamicin, imipenem, ticarcillin, norfloxacin, polymyxin-B, piperacillin, spectinomycin, tobramycin and streptomycin, including dyes and detergents such as ethidium bromide, acriflavine, deoxycholate, sodium dodecyl sulphate, and disinfectants benzalkonium chloride, chlorhexidine and triclosan signifies the wide substrate specificity of the transporter in K. pneumoniae. Growth inactivation and direct fluorimetric efflux assays provide evidence that kpnGH mediates antimicrobial resistance by active extrusion in K. pneumoniae. The kpnGH isogenic mutant displayed decreased tolerance to cell envelope stressors emphasizing its added role in K. pneumoniae physiology. Conclusions and Significance The MFS efflux pump KpnGH involves in crucial physiological functions besides being an intrinsic resistance determinant in K. pneumoniae.


Introduction
Drug efflux, in particular multidrug efflux, is a serious problem to drug therapy as they efflux broad range of substrates including clinically relevant antimicrobial agents [1,2,3,4,5,6]. Nosocomial pathogen Klebsiella pneumoniae is a Gram-negative bacillus associated with healthcare infections, known to cause variety of diseases in humans with significant morbidity and mortality [7]. K. pneumoniae causes various serious illnesses and the problem is aggravated due to its propensity to acquire multidrug resistance determinants [8,9]. Recently hypervirulent, hyperviscous K. pneumoniae NTUH-K2044, that belongs to the K1 serotype, known to cause pyogenic liver abscess, sometimes complicated by endophthalmitis or meningitis, has emerged in Taiwan, Singapore, Korea and other Asian countries [10]. Analysis of the available genome sequence of different K. pneumoniae isolates from NCBI database www.ncbi.nlm.nih.gov revealed that more than 10% of total genes were annotated as transporter proteins or efflux pumps, of which, to date only few have been studied and demonstrated to have a role in drug resistance. The 5.2-Mb genome of K. pneumoniae strain, NTUH-K2044 (encoding 4,992 proteins, GC content: 57.7%) is reported to harbor .15 open reading frames encoding for putative efflux pumps from different families (accession number NC_012731) [10]. The efflux systems functionally characterized in K. pneumoniae so far include acrAB and kexD from resistance/ nodulation/cell division (RND), family [11,12,13]; the kdeA efflux gene from multi drug and toxic compound extrusion (MATE) family [14]; the kmrA gene from major facilitator super family (MFS) family [15]; and kpnEF from small multidrug resistance (SMR) family [16]. In Gram-negative bacteria, a subset of inner membrane proteins in the MFS act as efflux pumps to decrease the intracellular concentrations of multiple toxic substrates and confer multidrug resistance [17,18]. The MFS type of transporters perhaps the most largest and diverse among all the efflux super families are found in all kingdoms of life. Well-studied examples such as QacA and NorA of Staphylococcus aureus and SmvA of Salmonella enterica serovar Typhimurium belong to the latter family [19,20,21]. Analysis of the K. pneumoniae NTUH-K2044 genome reveals the presence of a novel two component efflux pump operon, an emrAB homolog that belongs to MFS super family; whose functions have remained completely unexplored so far [10]. The objective of the present study was to investigate the role of putative MFS efflux system, an emrAB homolog (denoted kpnGH) in K. pneumoniae with respect to cellular physiology and broad spectrum antimicrobial resistance.

Bacterial Strains, Plasmids and Media
K. pneumoniae NTUH-K2044 (a strain that resulted in pyogenic liver abscess in a 66 year old patient) was kindly provided by Dr. Jin Town Wang of the National Taiwan University Hospital, Taipei, Taiwan [10]. E. coli SM10 lambda pir and pUT-Km was used to create isogenic mutants. Genomic DNA, plasmid, restriction digestion, DNA elution (Qiagen), ligation, transformation, conjugation, DNA sequencing (Applied Biosystems) were performed as previously described [22,23,24]. Primers used in the present study were custom-synthesized (Eurofins MWG operons, Germany).

Cloning of kpnGH in Hyper Susceptible E. coli Strain KAM32
The putative efflux genes kpnG and kpnH were amplified by standard PCR protocol using primers kpnG-F; kpnG-R and kpnH-F; kpnH-R (Table 1) respectively and cloned into EcoRI and PstI (New England Biolabs, MA, USA) site of pUC18. The resulting recombinant plasmid pkpnG, pkpnH and pkpnGH were transformed into E. coli KAM32 (DacrAB and DydhE) for functional characterization.

Construction of the kpnGH Mutant in K. pneumoniae
The putative emrAB homolog, KP1_4279/KP1_4280 (designated kpnGH) is located starting from nucleotides 4108112 to 4109284 bps (kpnG: 1173 bp, 390 aa, 42.59 Kda and kpnH: 1536 bp, 511 aa, 55.589 Kda) in K. pneumoniae [10]. A 720 bp internal fragment was amplified by PCR using DkpnGH-F and DkpnGH-R primers and cloned in EcoRI digested pUT-Km and transformed into E. coli SM10 lambda pir strain (thi thr leu tonA lacY supE recA::RP4-2-Tc::Mu Km lpir). The recombinant plasmid pUT-kpnGH was mobilized into recipient K. pneumoniae from donor E. coli SM10 lambda pir as described previously to create inactivation (insertional inactivation means the operon is present on the chromosome but is disrupted, therefore the kpnGH efflux pump is non-functional) in kpnGH [22,23]. The intact kpnGH operon along with its promoter was amplified with primer R-1 and primer R-2 and cloned into pCRIITOPO-CAT. The resulting construct was electroporated into DkpnGH and selected on LB agar plates supplemented with 50 mg/mL kanamycin and 100 mg/mL chloramphenicol to get the transcomplemented strain DkpnGHVkpnGH.

Motility Assay
The motility assays were done as reported before [23], where LB grown K. pneumoniae cultures (OD 600nm = 1.0) were inoculated with a toothpick on LB plates with 0.25%, 0.45% and 0.7% agar and incubated for 10 hrs at 37uC. In this growth medium bacteria can swim through the soft agar and produce a halo. The diameter of the halo is a measure of the ability to swim.

Crystal Violet Binding Assay
The crystal violet binding assay was done as described before [23].

Fluorimetric Efflux Studies
Accumulation of EtBr was monitored as described previously [26]. Briefly, bacterial cells were grown to mid-log phase, harvested, and suspended in 16 phosphate buffered saline (0.136 M NaCl, 0.0026 M KCl, 0.01 M Na2HPO4, 0.00176 M KH2PO4; pH 7.0) to an optical density at OD600 nm = 0.2. EtBr was added at a concentration of 10 mg/ml. Aliquots were taken at different time intervals, harvested, suspended in 1 ml of 0.1 M glycine HCl buffer (pH 2.3), and incubated for 6 h at 37uC, and the fluorescence of the supernatant was measured with excitation 530 nm and emission 600 nm. Where indicated, the proton motive force uncoupler CCCP was added to the assay mixture at a final concentration; 25 mg/ml.

OMP Preparation
OMPs were purified by the method as described previously [23].

Bioinformatic Analysis and Statistical Analysis
All data are presented as means 6 the standard error of the mean. Plotting and calculation of the standard deviation was performed in Microsoft Excel. Statistical analysis was performed on crude data by using a paired Student t test. P values of ,0.05 were considered significant.

Insilico Analysis of kpnGH MFS-type Efflux Pump
The deduced KpnG and KpnH proteins consist of 390 and 511 amino acid residues, respectively, similar to the size of EmrA and EmrB in E. coli. The nucleotide sequence deduced from the 2709 bp DNA fragment obtained from K. pneumoniae shares .90% identity with EmrAB efflux system in E. coli, S. dysenteriae and S. Typhimurium. Thus, KpnG and KpnH are both members of the MFS family of drug transporters. In addition, significant sequence similarities were detected between KpnGH and EmrAB of E. coli, Proteus vulgaris, and Citrobacter freundii. The multiple alignment of KpnGH with EmrAB of E. coli has been shown in Figure S1. The KpnG has two hydrophobic, presumably transmembrane regions spanning from amino acids 23-45; (

Contributions of MFS-type Efflux kpnGH in K. pneumoniae Growth
Experimentally the growth characteristics of WT, DkpnGH (inactivation confirmed by southern and RT-PCR) and DkpnGHVkpnGH (confirmed by RT-PCR) were determined over a period of ,18 h in LB medium with different pH and analysis revealed unique patterns. We tested the growth kinetics at pH 5.

Role of Membrane Transporter in Gastrointestinal (GI) Stress Challenges
To determine the role of kpnGH in intestinal colonization, WT, DkpnGH and DkpnGHVkpnGH underwent specific GI stress associated with bile and osmotic challenges. In the bile resistance assay, different strains were exposed to different concentrations (physiological concentration is 0.2% to 2%, [27]) of bile. When mid-log phase cultures were exposed to different concentrations of bile it was observed that the total CFU count of control (measure of surviving capacity) was higher as compared to DkpnGH. The ability of control to grow in the presence of 0.5% bile was 1.15 fold (60.042), 0.75% bile was 1.3 fold (60.024), 1% bile was 1.3 fold (60.002) and 2% was 2.45 fold (60.045) higher when compared to DkpnGH, while transcomplemented DkpnGHVkpnGH strain restored the ability to tolerate stress (Figure 2 The ability of control to grow in the presence of NaCl (physiological concentration being 150 mM, [28]) at 0.25M was  To deduce the role of kpnGH in temperature tolerance, we performed the heat shock assay. The temperature dependent assay showed that the kpnGH mutant displayed 15% reduced survival than the wild type at 42uC, thereby demonstrating that the response of K. pneumoniae kpnGH mutant in temperature stress (Figure 2-C). Overall these results imply that kpnGH had a contributory role towards varied stress tolerance in K. pneumoniae.

Functions of KpnGH in Antibiotic Resistance
To evaluate the role of kpnGH in drug resistance, antibiotic susceptibilities of WT and DkpnGH was monitored. The results of disc diffusion assay displayed that upon inactivation of the efflux pump, the bacterial cells displayed significantly altered susceptibility to azithromycin, ceftazidime, ciprofloxacin, ertapenem, erythromycin, gentamicin, imipenem, ticarcillin, norfloxacin, polymyxin-B, piperacillin, spectinomycin, tobramycin and streptomycin ( Figure 3).

Mutation in kpnGH Increases Sensitivity to Structurally Unrelated Compounds
The WT, DkpnGH and DkpnGHVkpnGH cultures in this study were tested for their ability to withstand high concentration of different substrates that are structurally unrelated to antibiotics. The ability of DkpnGH to withstand different concentrations of EtBr (figure S3-A) and acriflavine (figure S4-B) was reduced by 1.3 and 2.0 fold when compared to WT respectively. Results were similar when done with other dyes such as rhodamine, safranine and acridine orange (data not shown).
Upon exposing the cells to different concentrations of SDS it was observed that the total CFU count of WT at 16384 mg/ml was 1.5 fold higher than DkpnGH [P = /0.009962331] (figure S3-C). Analysis indicates the wide substrate specificity for kpnGH in K. pneumoniae.
The growth rate of DkpnGH in the presence of 2.0 mg/ml EtBr, was .12 fold lesser when compared to that of WT (P = 0.000477). To distinguish whether this is due to the loss of efflux pump activity in kpnGH mutant, the growth profile was monitored with CCCP, an uncoupler known to collapse the membrane energy and block the energy-dependent efflux pump. As expected a substantial decrease in growth was observed in kpnGH mutant (15.18-fold; P = 0.000697), (1.39-fold; P = 0.000775), (6.32-fold; P = 0.000681) and (15.06-fold; P = 0.000174) in the presence of CCCP, reserpine, verampamil and 2, 4 DNP respectively ( Figure S3, D I-IV).
The growth rate of DkpnGH in the presence of 2.0 mg/ml rhodamine, was .17.5 fold lesser when compared to that of WT (P = 0.000153). Akin a substantial decrease in growth was observed in kpnGH mutant (

Tolerance to Different Classes of Antibiotics
Upon exposing the cells to different concentrations of ticarcillin it was observed that the total CFU count of WT at 0.5 mg/ml was 1 fold, 4 mg/ml was 1.13 fold, 16 mg/ml was 1.11 fold, 64 mg/ml was 1.09 fold, 256 mg/ml was 1.09 fold, 512 mg/ml was 1.13 fold increased than DkpnGH [P = 0.027] (Figure 4-A).
Upon exposing the cells to different concentrations of neomycin it was observed that the total CFU count of WT at 0.5 mg/ml was 1.14 fold, 4 mg/ml was 0.97 fold, 16 mg/ml was 0.34 fold increased than DkpnGH [P = 0.042] (Figure 4-B).
Exposing the cells to different concentrations of ciprofloxacin it was observed that the total CFU count of WT at 0.5 mg/ml was 1.5 fold, 4 mg/ml was 7.3 fold increased than DkpnGH [P = 0.0188] (Figure 4-C).
Exposing the cells to different concentrations of chloramphenicol it was observed that the total CFU count of WT at 0.5 mg/ml   The EtBr accumulation data indicate that the efflux was most efficient in the control, whereas it was less efficient in DkpnGH. The addition of CCCP increased the EtBr levels which eventually reached a plateau in both control and DkpnGH (Figure 4-H). Overall, these preliminary results suggest that mutant of kpnGH possibly affects active efflux in K. pneumoniae.

MFS Efflux Pump kpnGH Influences Tolerance to Disinfectants
We tested the susceptibilities of control and DkpnGH towards different concentrations of popularly used hospital based disinfectants such as chlorhexidine and benzalkonium chloride [29,30]. Upon exposing the cells to different concentrations of benzalkonium chloride it was observed that the total CFU count of control was 5 to 12 fold higher than DkpnGH [P = 0.047277226] ( figure  S4-A).
When cells were exposed to different concentrations of chlorhexidine it was observed that the total CFU count of control was 1.3 to 1.7 fold higher than DkpnGH [P = /0.080120359] ( figure  S4-B).
Upon exposing the cells to different concentrations of triclosan it was observed that the total CFU count of control was 1.3 to 1.4 fold higher than DkpnGH [P = /0.065323347] ( figure S4-C). Decisively, the results imply the broad spectrum antimicrobial resistance property of kpnGH, a MFS-type efflux pump for the first time in K. pneumoniae.

Alterations in Outer Membrane Profile of the kpnGH Inactivated Mutant in K. pneumoniae
A reduction in the permeation of antibiotics is generally related to a decrease in porin expression or an alteration in the porin structure [2,5,6]. Thus, we compared the OMP profiles of DkpnGH with WT to find out whether a kpnGH inactivated mutant expresses alternative porins/OMPs to maintain normal cellular functions. It was interesting to note that there was a marked difference in the OMP profiles of mutant compared to wild type ( Figure S5), and deciphering the identity and function of these proteins is highly warranted.

Discussion
Klebsiella pneumoniae, a member of Enterobacteriaceae family, is an important pathogen in both the community and the clinical setting. K. pneumoniae is largely responsible for causing diseases such as nosocomial pneumonia (7 to 14% of all cases), septicaemia (4 to 15% of all cases), wound infections (2 to 4% of all cases), and neonatal septicaemia (3 to 30% of all cases) [1]. Few reports have assessed the masked role of efflux pumps in multidrug resistant phenotype in K. pneumoniae clinical isolates. Analysis of K. pneumoniae genome sequences namely K. pneumoniae strain 1084 (NC_018522.1); K. pneumoniae strain HS11286 (NC_016845.1); K. pneumoniae strain MGH 78578 (NC_009648.1); as well as K. pneumoniae strain NTUH-K2044 (NC_012731.1) indicates the presence of several putative multidrug efflux pumps [10]. In this study, we report the biological functions of MFS-type efflux pump kpnGH in modulating the cellular physiology and antimicrobial resistance in K. pneumoniae for the first time.
The E. coli KAM32 cells expressing kpnGH mounted significant growth difference than the vector control on a broad range of increased than DkpnGH [P = 0.027]. The data is the means of measurements made in triplicate and performed three times. B) Sensitivity of WT, DkpnGH and DkpnGHVkpnGH towards neomycin. Upon exposing the cells to different concentrations of neomycin it was observed that the total CFU count of WT at 0.5 mg/ml was 1.14 fold, 4 mg/ml was 0.97 fold, 16 mg/ml was 0.34 fold increased than DkpnGH [P = 0.042]. The data is the means of measurements made in triplicate and performed three times. C) Sensitivity of WT, DkpnGH and DkpnGHVkpnGH towards ciprofloxacin. Exposing the cells to different concentrations of ciprofloxacin it was observed that the total CFU count of WT at 0.5 mg/ml was 1.5 fold, 4 mg/ml was 7.3 fold increased than DkpnGH [P = 0.0188]. The data is the means of measurements made in triplicate and performed three times. D) Sensitivity of WT, DkpnGH and DkpnGHVkpnGH towards chloramphenicol. Exposing the cells to different concentrations of chloramphenicol it was observed that the total CFU count of WT at 0.5 mg/ml was 1.39 fold, 4 mg/ml was 1.08 fold increased than DkpnGH [P = 0.0266]. The data is the means of measurements made in triplicate and performed three times. E) Sensitivity of WT, DkpnGH and DkpnGHVkpnGH towards tetracycline. Exposing the cells to different concentrations of tetracycline it was observed that the total CFU count of WT at 0.5 mg/ml was 1.18 fold, 4 mg/ml was 1.45 fold increased than DkpnGH [P = 0.0288]. The data is the means of measurements made in triplicate and performed three times. F) Growth inactivation assay using wild type, DkpnGH and DkpnGHVkpnGH in the presence of cefepime. Growth pattern in absence of any drug or inhibitor is included as control. The growth rate of DkpnGH in the presence of 0.1 mg/ml cefepime, was 15.373 fold lesser when compared to that of WT (P = 0.000574). A decrease in growth was observed in kpnGH mutant ( antimicrobial compounds, including representative antimicrobial agents, antiseptics, cationic dyes and detergents confirming that K. pneumoniae kpnGH, like the emrAB homolog in E. coli, S. Typhimurium, is required for resistance to various antimicrobials [21,31,32]. While inactivation of kpnEF resulted in increased susceptibility towards cefepime, ceftriaxone, imipenem, ertapenem, enrofloxacin, norfloxacin and ciprofloxacin, however non functional kpnGH results in increased susceptibility to cefepime, ceftazidime, ceftriaxone, ciprofloxacin, erythromycin, spectinomycin, streptomycin, tetracycline and tobramycin and structurally unrelated compounds such as SDS, deoxycholate, EtBr, and acriflavine. KpnGH displayed sensitivity to benzalkonium chloride, chlorhexidine and triclosan; however KpnEF displayed greater sensitivity to these compounds. Results presented here provide evidence that inactivation of kpnGH diminishes drug efflux capacity in K. pneumoniae. Where the previously reported efflux pump kpnEF displayed higher sensitivity to high bile (,4.0 fold) concentration, interestingly, the kpnGH mutant is only 2.5 fold sensitive to varied bile challenges indicating that K. pneumoniae possibly utilizes kpnGH to extrude bile salts from the cytoplasm out of the cell to successfully thrive in the GI tract of human host [23]. Evidence indicating the role of KpnGH in surviving conditions that mimic the upper parts of the GI, where they encounter hyper osmotic condition in a microaerobic environment adds to the multifaceted, unprecedented functions displayed by MFS-type efflux pump in K. pneumoniae.
In conclusion, using various molecular approaches, we have demonstrated the functions of kpnGH, a MFS-type efflux pump in cellular physiology and antimicrobial resistance in Klebsiella spp for the first time.  Figure S3 Susceptibility and Growth inactivation assays towards different dyes and detergent. A) Sensitivity of wild type, DkpnGH and DkpnGHVkpnGH towards EtBr when cells were exposed to different concentrations of dye (2 mg/ml, 4 mg/ml, 64 mg/ml, 128 mg/ml, 256 mg/ml and 512 mg/ml). The data is the means of measurements made in triplicate and performed three times. B) Sensitivity of wild type, DkpnGH and DkpnGHVkpnGH towards acriflavine when cells were exposed to different concentrations of dye (2 mg/ml, 4 mg/ml, 64 mg/ml, 128 mg/ml, 256 mg/ml and 512 mg/ml). The data is the means of measurements made in triplicate and performed three times. C) Sensitivity of wild type, DkpnGH and DkpnGHVkpnGH towards SDS when cells were exposed to different concentrations of detergent (1024 mg/ml, 2048 mg/ml, 4096 mg/ml, 8192 mg/ml, 16834 mg/ml). The data is the means of measurements made in triplicate and performed three times. D) Growth inactivation assay using wild type, DkpnGH and DkpnGHVkpnGH in the presence of EtBr. Growth pattern in absence of any EtBr or inhibitor is included as control. The growth rate of DkpnGH in the presence of 2.0 mg/ml EtBr, was .12 fold lesser when compared to that of WT (P = 0.000477). A decrease in growth was observed in kpnGH mutant (15.18-fold; P = 0.000697), (1.39-fold; P = 0.000775), (6.32-fold; P = 0.000681) and (15.06fold; P = 0.000174) in the presence of CCCP (I), reserpine (II), verampamil (III) and 2, 4 DNP (IV) respectively. The mean values of three independent experiments have been used for plotting the graph. E) Growth inactivation assay using wild type, DkpnGH and DkpnGHVkpnGH in the presence of rhodamine. Growth pattern in absence of any rhodamine or inhibitor is included as control. The growth rate of DkpnGH in the presence of 2.0 mg/ml rhodamine, was .17.5 fold lesser when compared to that of WT (P = 0.000153). A decrease in growth was observed in kpnGH mutant (10.476-fold; P = 0.000234), (1.39102-fold; P = 0.000796), (8.13-fold; P = 0.000101) and (23.1-fold; P = 0.000196) in the presence of CCCP (I), reserpine (II), verampamil (III) and 2, 4 DNP (IV) respectively. The mean values of three independent experiments have been used for plotting the graph. F) Growth inactivation assay using wild type, DkpnGH and DkpnGHVkpnGH in the presence of safranine. Growth pattern in absence of any safranine or inhibitor is included as control. The growth rate of DkpnGH in the presence of 2.0 mg/ml safranine, was .15.94 fold lesser when compared to that of WT (P = 0.000221). A decrease in growth was observed in kpnGH mutant (5. 16