Molecular typing and virulence analysis of multidrug resistant Klebsiella pneumoniae clinical isolates recovered from Egyptian hospitals

Klebsiella pneumonia infection rates have increased dramatically. Molecular typing and virulence analysis are powerful tools that can shed light on Klebsiella pneumonia infections. Whereas 77.7% (28/36) of clinical isolates indicated multidrug resistant (MDR) patterns, 50% (18/36) indicated carpabenem resistance. Gene prevalence for the AcrAB efflux pump (82.14%) was more than that of the mdtK efflux pump (32.14%) in the MDR isolates. FimH-1 and mrkD genes were prevalent in wound and blood isolates. FimH-1 gene was prevalent in sputum while mrkD gene was prevalent in urine. Serum resistance associated with outer membrane protein coding gene (traT) was found in all blood isolates. IucC, entB, and Irp-1 were detected in 32.14%, 78.5% and 10.7% of MDR isolates, respectively. We used two Polymerase Chain Reaction (PCR) analyses: Enterobacterial Repetitive Intergenic Consensus (ERIC) and Random Amplified Polymorphic DNA (RAPD). ERIC-PCR revealed 21 and RAPD-PCR revealed 18 distinct patterns of isolates with similarity ≥80%. ERIC genotyping significantly correlated with resistance patterns and virulence determinants. RAPD genotyping significantly correlated with resistance patterns but not with virulence determinants. Both RAPD and ERIC genotyping methods had no correlation with the capsule types. These findings can help up better predict MDR Klebsiella pneumoniae outbreaks associated with specific genotyping patterns.

Antimicrobial susceptibility pattern and detection of genes coding for MDR efflux pumps and outer membrane porins. As determined by disc diffusion antimicrobial susceptibility testing method, a percentage 77.7% (28/36) of isolates showed multidrug resistance (MDR) patterns, but all these MDR isolates were sensitive to colisitin (10 μ g). All MDR isolates were resistant to beta lactam antibiotics and 64.28%, 82.15%, and 85.7% showed resistance to carbapenem, quinolone, and aminoglycosides, respectively. Tetracycline and chloramphenicol were effective against 61.1% of carbapenem-resistant isolates. The tested isolates were distributed into 24 antimicrobial resistance patterns ( Table 2). Most patterns showed resistance to cephalosporin and beta lactam/beta lactamase inhibitors. The most predominant pattern was A6 and A8.
Gene prevalence for the AcrAB efflux pump system (82.14%) was more than that of the mdtK efflux pump (32.14%) in the MDR isolates. Incomplete AcrAB efflux pump system was detected in the remaining five isolates. The genes coding for porin protein (ompK35) and (ompK36) were not detected in six and four MDR isolates, respectively. Genes coding for the porins (ompK 35 and ompK36) were detected in all isolates recovered from wound and CSF specimens. The presence of these two porin-coding genes was variable in blood, sputum, and urine samples.
Detection of virulence genes. The prevalence and distribution of virulence factors are shown in Table 3.
The fimH-1 and mrkD genes, encoding type 1 and type 3 fimbrial adhesins, were present in all wound and blood isolates. The fimH-1 gene was prevalent in sputum isolates whereas mrkD gene was prevalent in all urine samples. Serum resistance associated with the outer membrane protein coding gene (traT) was detected in all blood isolates.
The prevalence of capsule K genotypes in the 28 MDR isolates revealed that K1 (n = 8), K2 (n = 2) and the remaining isolates were non-typable as K1 or K2 genotypes. Isolates showing K1 genotypes were obtained from urine, blood, and sputum specimens, while K2 isolates was recovered from urine and wound samples. There were 16 virulence profiles according to detected virulence genes. It is noteworthy that virulence genetic profiles indicated that virulence determinants were variable among K. pneumoniae strains that possess the same capsule genotype ( Table 3).

Genotyping of Klebsiella pneumoniae isolates by RAPD and ERIC analyses.
According to the dendrograms, Enterobacterial Repetitive Intergenic Consensus (ERIC) and Random Amplified Polymorphic DNA (RAPD) analyses revealed 21 and 18 distinct patterns of K. pneumoniae isolates with similarity > 80%, respectively (Figs 1 and 2). The 21 ERIC genotypes were designated E1 to E21 while the RAPD genotypes were designated R1 to R18 and each of their variant subtypes were indicated by a letter suffix. Dendrogram analysis of ERIC genotyping showed three clusters (A-C): clusters A, B, and C contained 12/28, 9/28, and 7/28 of the MDR isolates, respectively. The isolates (18 and 21) and (28, 35, 51, and 56) showed high similarity which may suggest that those isolates constitute a clonal lineage (Fig. 1). On the other hand, the RAPD genotyping revealed different pattern with 6 clusters (A-F). The isolates (15 and 58), (20 and 21), (9 and 18), and (36 and 56) showed high similarity (Fig. 2). Based on Simpson's index of diversity, the discriminatory potential of different typing techniques used with K. pneumoniae isolates varied from 0.519 to 0.984 (Table 4). The high Simpson's index of diversity for the antibiotyping, virulence, RAPD, and ERIC typing indicates greater diversity. Kendall's tau-b correlation coefficient was calculated between RAPD and ERIC genotyping methods versus resistance patterns, virulence determinants, and capsule types of K. pneumoniae. Based on the statistical correlation tests ( genotyping significantly correlates with resistance patterns (p < 0.01) and virulence determinants (p < 0.05). On the other hand, the RAPD genotyping significantly correlates with resistance patterns (p < 0.05) but not with virulence determinants (p > 0.05). Both RAPD and ERIC genotyping methods have no correlation with the capsule types (p > 0.05).

Discussion
K. pneumoniae is the causative agent of several different healthcare-associated infections, such as bloodstream infections, wound infections, pneumonia, and meningitis. The extensive use of antimicrobials led to high incidence of resistance in K. pneumoniae 24 . In our study, K. pneumoniae isolates showing multidrug resistance comprised 71.1% of total samples. Rates as high as 66.7% of MDR K. pneumoniae isolates were also detected in other studies 25 . The high rates of antimicrobial resistance detected in our study can be attributed to the lack of strict policies that govern the use of antibiotics in Egypt 3 .
Antibiotic efflux pumps represent one of the most important antimicrobial resistance mechanisms used by K. pneumoniae clinical isolates 26,27 . The increased efflux of the antimicrobial agent leads to the reduction of its intracellular concentration, which can enhance bacterial survival 28 . The AcrAB efflux pump was more common than mdtK. The presence of the multidrug efflux pump system (AcrAB-TolC) was significantly correlated with the MDR pattern. On the other hand, five MDR isolates was missing either the AcrAB efflux pump or the TolC outer membrane protein or both.
Gram negative bacterial outer membranes are poorly permeable to both hydrophobic and hydrophilic molecules. Thus, most antimicrobial agents other than β -lactam must cross the membrane in order to reach their intracellular drug targets and so require the presence of porin to bypass the asymmetric bilayer of phospholipid and lipopolysaccharide membrane 29 . Consequently, it has been reported that loss of porins ompK 35 and ompK 36 led to an increase in carbapenem, ciprofloxacin, and chloramphenicol resistance 30 . Surprisingly, in our research, porin loss was not significantly correlated to the MDR pattern (P > 0.05). This could be attributed to the presence of point mutations, disruption in the protein coding sequence, or promoter region mutations 31 .
In the current study, about fifty percent of the total isolates showed resistance to both imipenem and ertapenem. In this context, there has been a significant increase in carbapenem resistance among K. pneumoniae isolates in Egypt during the last few years (from 13.9% to 44.4%) 3,32,33 . K. pneumoniae isolates possess several mechanisms to evade the activity of carbapenems. These include AmpC production or ESBL production together with porin loss, carbapenemase production, and production of acquired Metallobetalactamase (MBL) 34 .
Contrary to the New Delhi metallo-β -lactamase, which is a broad spectrum carbapenemase with ability to inactivate β -lactams except aztreonam 35 , all carabapenem resistant isolates in our study were also resistant to aztreonem. This may be due to the development of a new antimicrobial resistance pattern in Egyptian hospitals. In all tested carbapenem-resistant isolates (n = 18), there was no simultaneous porin loss with AmpC or ESBL production. In another study 36  were less permeable than those of a susceptible control strain. This suggested that the possession of these two porins could lead to higher resistance due to an associated pump system 36 . This is relevant to our study given the fact that OmpF genes in E. coli are homologues to OmpK35 genes in K. pneumoniae 29 .
Carbapenemase-producing enterobacteriaceae, such as K. pneumonia, can cause deadly infections 20,21 . Colistin was used to treat Gram-negative infections but was abandoned because of its toxicity. Recently, it has been revived again as a treatment for life-threatening infections caused by some resistant Gram-negative bacteria, such as Pseudomonas aeruginosa and Acinetobacter baumannii 37 . Interestingly, all the tested K. pneumoniae isolates in this study were sensitive to colistin.
Both OmpK35 and OmpK36 play a role in K. pneumoniae virulence and infection. Deletion of OmpK36 or OmpK35/OmpK36 can lead to the reduction in virulence of highly virulent strains and can increase their susceptibility to neutrophil phagocytosis 38,39 . In our investigation, both Ompk 35 and OmpK36 porin-coding genes were simultaneously detected in all K. pneumoniae isolates recovered from wound and CSF samples. Their presence was variable though in sputum, blood, and urine samples. A direct correlation between efflux pumps and virulence of pathogenic bacteria was reported by Padilla et al 40 . Several genes essential for intracellular invasion and survival were downregulated in mutant strains lacking acrAB-tolC efflux pumps 41 .
Type 1 fimbriae are the most common adhesive organelles in enterobacteriaceae and can lead to urinary tract infections 42 . Type 3 fimbrial adhesin can mediate the binding of K. pneumoniae to endothelial cells and to epithelial cells of the respiratory and urinary tracts. MrkD protein is a crucial factor in binding bacteria to the collagen molecules of the mammalian cells 43 . Many K. pneumoniae clinical isolated normally express both type 1 and type 3 fimbrial adhesins 44 . In the current study, the two genes coding for these adhesive structures (Type 3 fimbrial adhesin and MrkD) were detected in all wound, blood, and CSF isolates and in about 80% of sputum and urine isolates. The plasmidic traT gene encodes an outer membrane protein involved in bacterial conjugation and blocks the complement-mediated cascade, and act as an invasin 45 . We detected the traT gene in twenty two K. pneumoniae isolates (78.5%). The prevalence of traT gene in our isolates was relatively high as it was frequently associated with the K1 capsule serotype. Most enterobacteriaceae strains contain genes encoding iron uptake systems, such as enterochelin or aerobactin 46 . These siderophores have dual roles as they can also inhibit T cell proliferation in addition to their role in enhancing iron uptake 47 . The iron siderophores aerobactin synthase gene (IucC), enterobactin biosynthesis gene (entB), and yersinibactin biosynthesis gene (Irp-1) were detected in 32.14%, 78.5%, and 10.7% of MDR K. pneumoniae isolates, respectively. Highly pathogenic Yersinia strains have high-pathogenicity island (HPI) that contain the gene Irp-1. This HPI is also prevalent in Klebsiella and other enterbocateria 48 , such as E. coli, K. oxytoca, K. pneumonia, Citrobacter species, and Enterobacter species 46 .
The capsular serotypes K1 and K2 are associated with the predominant virulent strains of K. pneumoniae 49 . Feizabadi et al. has shown that K1 and K2 serotypes represented 11.2% and 14.6%, respectively, of the total K. pneumoniae isolates 50 . In our study, K1 and K2 serotypes represented 28.5% and 7.14% of the MDR K. pneumoniae isolates.
Molecular typing is a potent tool for the study of nosocomial infections 51 . RAPD is a widely used genotyping tool for K. pneumoniae strains with ESBLs production (Gori et al., 1996). Out of a total of 28 MDR K. pneumoniae isolates in the current investigation, ERIC-PCR revealed 21 and RAPD-PCR revealed 18 distinct patterns of K. pneumoniae isolates. This may be attributed to the genetic variation in pathogenic these K. pneumoniae strains. Our data confirms the observations of Lai et al. 52 that pathogenic K. pneumoniae is highly heterogeneous, due to differences in nucleotide sequences. The large number of serotypes in this species could also explain this genetic diversity highlighted by the RAPD-PCR genotypic analysis 53 .
Correlations between RAPD-PCR genotyping and antibiotic resistance patterns of K. pneumonia were observed by Ashayeri-Panah et al. 54 and Espinar et al. 55 . In the current study, both RAPD-PCR and ERIC-PCR genotypic analyses revealed correlations with resistance patterns of K. pneumonia. The highest correlation coefficients were observed with ERIC genotyping, indicating that the latter may be more valuable in prediction of resistance patterns of K. pneumoniae as compared to the RAPD-PCR genotyping method. Moreover, ERIC, but not RAPD-PCR, revealed statistically significant correlations with virulence determinants of K. pneumoniae.  Finally, both RAPD-PCR and ERIC-PCR showed no statistically significant correlation with the detected capsule types of K. pneumoniae isolates. Results included in this study can help up better predict MDR Klebsiella pneumoniae outbreaks associated with specific genotyping patterns in the future.

Materials and Methods
Bacterial strains. Thirty six K. pneumonia clinical isolates were recovered from patients at Kasr El Aini Hospitals, Cairo, Egypt. Approvals from the institutional review board of the hospitals and the Research Ethics committee of the October University for Modern Sciences and Arts, Giza, Egypt were obtained prior to conducting the study. All methods were performed in accordance with the required guidelines and regulations. For experiments involving human samples, informed consent was obtained from all subjects. Strains were isolated from sputum, urine, blood, wound, and cerebrospinal fluid (CSF) specimens. Specimens were collected in the period from August 2015 through December 2015. Isolates were identified by conventional and biochemical tests as described previously 56 and then were stored at − 20 °C in brain heart infusion broth with 15% v/v glycerol. Antimicrobial susceptibility. Antibiotic susceptibility testing of Klebsiella sp., was performed according to the Kirby-Bauer disk diffusion method 57 . The antimicrobial sensitivity assays to nineteen antibacterial drugs were done using commercially available antibiotic discs (OXOID, UK) including Ampicillin (AMP, 10 μ g), Amoxicillin/Clavulanic acid (AMC, 20/10 μ g), Piperacillin/Tazobactam (TZP, 110/10 μ g), Cefoxitin (FOX, 30 μ g), Ceftazidime (CAZ, 30 μ g), Cefuroxime (CXM, 30 μ g), Aztreonam (ATM, 30 μ g), Ertapenem (ETP, 10 μ g), Impinem (IMP, 10 μ g), Gentamicin (CN, 10 μ g), Tobramycin (TOB, 10 μ g), Amikacin (AK, 30 μ g), Tetracycline (TE, 30 μ g), Doxycycline (DO, 30 μ g), Ciprofloxacin (CIP, 5 μ g), Nalidixic acid (NA, 30 g), Co-trimoxazole (SXT, 30 μ g), Colistin (CT, 10 μ g) and Chloramphenicol (C, 30 μ g). For all tested antimicrobials except colistin (10 μ g), the plates were then incubated at 37 °C for 24 hours, the diameters of the inhibition zones were measured in millimeter and interpretation of results was done according to CLSI standards 57 . For Colistin, breakpoints were used for interpretation 58 . Multidrug resistant (MDR) isolates were selected according to their non-susceptibility to at least one agent in three or more antimicrobial categories 59 .

Detection of multidrug resistance and virulence determinants using PCR. DNA extraction.
Genomic DNA was extracted from overnight culture using ZYMO Quick-gDNA ™ MiniPrep (ZYMO Research, CA, USA). Concentration of the DNA extract and purity was determined by measuring absorbance at wavelengths 260 and 280 nm. The integrity of genomic DNA was tested by resolving DNA extracts on a 0.8% w/v agarose gel by electrophoresis. These crude DNA extracts were frozen at − 20 °C. Primers used for amplification are listed in Table 1 and were prepared by Invitrogen ® (Thermo Fisher scientific Inc., MA, USA). Primers were designed using the complete genome sequence of K. pneumoniae MGH 78578 (accession no. CP000647) and the internet based software Basic Local Alignment Tool (BLAST) in NCBI and Multiple sequence alignment using the CLUSTAL Omega in EMBL-EBI. PCR detection of multidrug resistance genes. Isolates that showed multidrug resistance phenotypes were tested for genes coding for the multidrug efflux pump system AcrAB-TolC and MdtK, in addition to porin coding genes (OmpK35 and OmpK36). The amplifications of these genes were performed in cycles with initial denaturing at 94 °C for 5 min followed by 35 cycles, each cycle consisting of 30 seconds at 94 °C for denaturation, 30 seconds for primer annealing (Table 1), and 1.5 min at 72 °C for elongation. After these cycles, the final elongation step was carried out at 72 °C for 10 min 45 .
The PCR conditions were similar to those used for detection of multidrug resistance genes with annealing temperatures included in Table 1.  randomly amplified polymorphic DNA (RAPD) analysis was performed according to the protocol published by Deschaght et al. 60 using the primer RAPD4 (5′ -AAGACGCCGT-3′ ). Briefly, two microliters of the DNA template were added to 12.5 μ L multiplex mastermix (MyTaq ™ HM Mix, Bioline ® , MA, USA), 1 μ L primer (10 pmol), and 9.5 μ L H 2 O. PCR cycles of initial incubation at 94 °C for 15 min followed by cycling for 40 times at 94 °C for 1 min, 37 °C for 1 min, and a final elongation at 72 °C for 2 min was performed.

Molecular typing of
ERIC typing was carried out using the primer ERIC2 (5′ -AAGTAAGTGACTGGGGTGAGCG-3′ ) using a similar PCR program to that of the RAPD method except for an extension time of 8 min 59 .
RAPD and ERIC fragments were visualized by 1.5% w/v agarose gel electrophoresis and results were analyzed using GelCompar II software (Version 6.6.11, Applied Maths, Kortrijk, Belgium). The patterns were normalized with bands of the marker and bands that were consistently present in all patterns. Computer-assisted analyses implemented in this study were performed according to the manufacturer's instructions.
Comparing different typing methods and calculation of discriminatory index. Simpsons index of diversity [discriminatory index (D)], based on the probability that two unrelated isolate samples from the test population are located in different typing groups, was calculated according to the following equation:   where N is the total number of isolates in the sample population, s is the total number of types described, and nj is the number of strains belonging to the jth type. Simpsons index of diversity ranges from 0.0 to 1.0, where 1.0 indicates that a typing method is able to distinguish each member of a population from all other members of that population and, conversely, 0.0 indicates that all members of a strain population are of an identical type 61 .
Statistical analysis. All statistical analyses were performed using SPSS, version 18.0 (SPSS Inc., NY, USA).
Chi-square tests were used to compare categorical measures between groups (Fisher's exact test where appropriate). Statistical correlation tests, including Kendall's tau-b nonparametric correlation coefficients, were determined at the two-tailed significance level for correlation of genotyping methods with virulence determinants, antimicrobial resistance, and capsule types. Data output of correlation analyses with p values less than 0.05 were considered statistically significant.