Phenotypic and Molecular Characterization of Extended-Spectrum Beta-Lactamase-Producing Escherichia coli in Bangladesh

Background Resistance to cephalosporins in Enterobacteriaceae is mainly due to the production of extended-spectrum beta-lactamase (ESBL). Little is known about ESBL-producing bacteria in Bangladesh. Therefore, the study presents results of phenotypic and molecular characterization of ESBL-producing Escherichia coli from hospitals in Bangladesh. Methods A total of 339 E. coli isolated from patients with urinary tract and wound infections attending three different medical hospitals in urban and rural areas of Bangladesh between 2003–2007 were screened for ESBL-production by the double disk diffusion test. Isolates with ESBL-phenotype were further characterized by antibiotic susceptibility testing, PCR and sequencing of different β-lactamase and virulence genes, serotyping, and XbaI-macrorestriction followed by pulsed-field gel electrophoresis (PFGE). Results We identified 40 E. coli with ESBL phenotype. These isolates were resistant to ceftriaxone, ceftazidime, cefotaxime, aztreonam, cefepime, and nalidixic acid but remained susceptible to imipenem. All but one isolate were additionally resistant to ciprofloxacin, and 3 isolates were resistant to cefoxitin. ESBL genes of blaCTX-M-1-group were detected in all isolates; blaTEM-type and blaOXA-1-type genes were detected in 33 (82.5%) and 19 (47.5%) isolates, respectively. Virulence genes that are present in diarrhoeagenic E. coli were not found. Class-1 integron was present in 20 (50%) isolates. All the ESBL-producing E. coli isolates harbored plasmids ranging between 1.1 and 120 MDa. PFGE-typing revealed 26 different pulsotypes, but identical pulsotype showed 6 isolates of serotype O25:H4. Conclusion The prevalence of multidrug-resistant ESBL-producing E. coli isolates appears to be high and the majority of the isolates were positive for bla CTX-M. Although there was genetic heterogeneity among isolates, presence of a cluster of isolates belonging to serotype O25:H4 indicates dissemination of the pandemic uropathogenic E. coli clone in Bangladesh.


Introduction
In Gram-negative bacteria, production of beta-lactamases is one of the most common mechanisms resulting in resistance to betalactam antibiotics. Recently, extended-spectrum beta-lactamases (ESBLs) have contributed to the dramatic increase in resistance to new generation beta-lactam agents throughout the world [1][2][3][4]. These enzymes are usually plasmid-encoded and have the capacity to hydrolyze many antibiotics including penicillins, cephalosporins, and aztreonam and are inhibited by clavulanic acid (CA) [2,3]. These phenotypic characteristics facilitate the identification of ESBLs-producing organisms using routine laboratory tests such as double disk diffusion test or E-test. However, screening of ESBL-producing bacteria by antibiotic susceptibility testing is challenging [5]. Thus detection of specific genes by PCR and sequencing are commonly used for final confirmation of ESBL producers. The association of ESBLs and the presence of TEM, SHV, OXA and CTX-M-type enzymes have been investigated in many studies [6,7]. Previous studies have shown that CTX-M-type enzymes were the most prevalent ESBLs [7,8]. According to Lahey database in Burlington (http://www.lahey.org/Studies/) that controls numbering of new beta-lactamases, more than 130 CTX-M-type enzymes are registered, which are classified into at least five different subgroups [9]. Since ESBLs are frequently encoded by genes located on different transferable genetic elements, a variety of epidemiological situations have been documented, ranging from sporadic cases to large outbreaks [10]. Moreover, ESBL-producing strains are often resistant to antibiotics of other classes (sulfonamides, aminoglycosides, quinolones) which complicates the treatment strategies in many hospitalized patients [11].
Although ESBLs have been detected in a many different of Gram-negative bacteria, Klebsiella pneumoniae and Escherichia coli remain the major ESBL-producing organisms worldwide [6,12]. In addition, ESBL-producing organisms are frequently the cause of urinary tract infections (UTI) and also in surgical wound infections [13,14]. There is one report from Bangladesh where they found 43.2% of E. coli and 39.5% of K. pneumoniae isolated from a hospital were ESBL producers [15]. However, there is a lack of information on molecular characterization of ESBLproducing organisms isolated in Bangladesh. Therefore, the objectives of the present study were the examination of presence of beta-lactamase genes in E. coli from different hospitals in Bangladesh. These E. coli were isolated from either urine samples or surgical wound swabs from 339 patients with symptoms of UTI (n = 274) and wound infection (n = 64). All urine samples were obtained from outpatient departments and all surgical wound swab samples were from inpatient department of respected hospitals. E. coli were isolated and identified by using standard microbiological and biochemical methods [16]. These isolates were grown in trypticase soy broth containing 0.3% yeast extract (TSBY) and stored at 2 70uC after addition of 15% glycerol.

Bacterial isolates
Since all isolates were collected anonymously no ethical approval was necessary for this study.
All 339 E. coli isolates were screened for the production of ESBLs by using the double disk diffusion test as described previously [18], with some modifications [4]. After overnight incubation at 37uC, any enhancement of the zone of inhibition between a beta-lactam disk and that containing the beta-lactamase inhibitor was indicative of the presence of an ESBL.

Amplification of qnrA, qnrB, and qnrS
Multiplex PCR of 40 ESBL isolates were performed to detect the qnrA, qnrB, and qnrS according to the procedure described earlier [25].

DNA sequence analysis
Chromosomal DNA from representative strains was prepared and purified by procedures described previously. Sequencing was performed to identify specific TEM and OXA type ESBL genes. Sequencing of gyrA and parC genes were performed in the qnr positive ESBL isolates according the procedure described elsewhere [25][26][27][28]. After PCR, the amplicons were purified with the GFX PCR DNA and gel band purification kit (Amersham Pharmacia, USA), and sequenced using the dideoxy-nucleotide chain termination method with an ABI PRISM BigDye Terminator Cycle Sequencing Reaction kit (Perkin-Elmer Applied Biosystems, Foster City, California) on an automated sequencer (ABI PRISM 310). The chromatogram sequencing files were inspected using Chromas 2.23 (Technelysium, Queensland, Australia), and contigs were prepared using SeqMan II (DNAS-TAR, Madison, WI). Nucleotide and protein sequence similarity searches were performed using the National Center for Biotechnology Information (NCBI, National Institutes of Health, Bethesda, MD) BLAST (Basic Local Alignment Search Tool) server on GenBank database, release 138.0 [29]. Multiple sequence alignments were developed using CLUSTALX 1.81 [30]. Sequences were manually edited in the GeneDoc version 2.6.002 alignment editor. Sequences of OXA and TEM genes reported in this paper were submitted to the Genbank using the National Center for Biotechnology Information (NCBI, Bethesda, MD) Sequin, version 7.77 under the accession number EU752482-86 and EU752487-91, respectively.

Bacterial strain typing
Serotyping. All E. coli with ESBL phenotype were serotyped by agglutination assays using 96-well microtitre plates with rabbit serum (SERUNAM) obtained against 187 somatic antigens and 53 flagellar antigens for E. coli, and against 45 somatic antigens for Shigella species described earlier [31].
Isolation of plasmid DNA. Plasmid DNA was prepared according to the alkaline lysis method of Kado and Liu [32] with some modifications as described previously [33]. The molecular mass of the unknown plasmid DNA was assessed by comparing with the mobility of the known molecular weight plasmids. Plasmids present in strains E. coli PDK-9, R1, RP 4 and V517 described previously were used as molecular mass standards [33].
Pulsed-field gel electrophoresis (PFGE). Intact agaroseembedded chromosomal DNA from clinical isolates of E. coli was prepared and PFGE was performed using the contour-clamped homogeneous electric field (CHEF-Mapper) apparatus from Bio-  Rad Laboratories (Richmond, CA, USA) according to the procedures described elsewhere [33]. Genomic DNA was digested with XbaI restriction enzyme (Gibco-BRL). The restriction fragments were separated by using CHEF-mapper system apparatus in 1% pulsed-field certified agarose in 0.56 TBE buffer. The DNA size standards used was the Salmonella enterica serovar Braenderup (H9812) ranging from 20.5 to 1,135 kb [34]. Banding patterns were analyzed according to the established criteria reported elsewhere [35].

Results
Of the 339 E. coli isolates, 11.8% (n = 40) were suspected to be ESBL producers using double disk diffusion test, of which 8.4% (23/274) were isolated from urine and 26.5% (17/64) from surgical wound samples. All E. coli with ESBL phenotype (n = 40) isolates were resistant to 3 rd generation cephalosporins. These isolates were also resistant to monobactams, such as, aztreonam and cefepime (4 th generation cephalosporin). Only 3 (7.5%) isolates were resistant to cefoxitin (2 nd generation cephalosporins). In addition, all isolates except for one were resistant to ciprofloxacin and all were resistant to nalidixic acid, whereas none of the isolates were resistant to imipenem. PCR for beta lactamase specific genes showed that all the 40 isolates contained the bla CTX-M-1 group and bla CTX-M-15 genes. In addition, bla OXA-1 type and bla TEM -type, genes were present in 33 (82.5%) and 19 (47.5%) isolates, respectively (Table 1). Class 1 integron gene was detected in 20 (50%) isolates. Furthermore, we also examined for the presence of different virulence genes that are usually present in diarrhoeagenic E. coli isolates. None of these isolates were positive for any virulence genes such as, lt, st, eae, ial, stx 1 , stx 2 and e Agg .
Of 40 ESBL isolates, 5 were found to carry the qnrS gene. In order to understand the fluoroquinolone resistance mechanism of 5 qnrS positive ESBL isolates, we sequenced the gyrA (gyrase A, a subunit of topoisomerase II) and parC (subunit of topoisomerase IV) genes and identified double mutations in gyrA (Ser 83 R Leu, Asp 87 R Asn) and a single mutation in parC (Ser 80 R Ile) genes. Sequencing of bla OXA -1 type and bla TEM -type genes from 5 representative isolates identified these genes as beta-lactamase genes bla OXA-1 and bla TEM-1 respectively.
All 40 isolates yielded heterogeneous plasmid pattern ranging in size from 1.1 to 120 MDa. Middle ranged plasmid (plasmids with a size between 30 MDa and 90 MDa) was found to be present in 65% (n = 26) isolates, while 45% (n = 18) isolates harbored 120 MDa plasmid. Plasmids ranged in size from 53 to 120 MDa were detected in all the ESBL-producing E. coli (Table S1).

Discussion
The emergence of ESBL-producing organisms has been reported in many parts of the world [1][2][3][4]. In 2004, a study in Bangladesh showed that high prevalence (43.2%) of ESBLproducing E. coli was found in an urban Hospital in Dhaka [15]. The present study showed that around 12% of E. coli isolates obtained from patients with UTI and wound infections were ESBL-producers. The difference in prevalence between these two studies would be due to the sampling from different hospitals and from different geographic locations, which is the case for this study (Table 1) whereas in the previous study, samples were collected only from single hospital setting. However, occurrence of ESBLproducing bacteria in clinical specimens at present might be different than the rate that we found in this study as we analyzed the isolates collected during the period 2003-2007.
In present study, among 40 ESBL-producing E. coli strains, O25:H4 was identified as the predominant serotype (n = 8, 20%). Importantly, a particular clone detected by multilocus sequence typing (MLST) in this serogroup (O25:H4-ST131) of ESBLproducing E. coli has disseminated in many countries in Europe and in Asia [36,37]. O25:H4 is mainly associated with production of CTX-M-15. Ceftazidime resistance indicated presence of CTX-M-15 in our O25:H4 isolates and we confirmed the presence of CTX-M-1group and CTX-M-15 in all isolates. Based on our results it is highly probable that the ESBL-producing clone E. coli O25:H4-ST131 is present in our study collection from different hospitals. Thus, further analyses especially MLST are needed to investigate the dissemination of this clone in future.
The antibiotic susceptibility data indicated that ESBL-producing E. coli isolates acquired resistance mechanisms against different classes of antibiotics that are frequently used for empirical treatment of UTIs and other extra intestinal infections. Interestingly, all 40 ESBL-producing isolates except for one were resistant to ciprofloxacin (n = 39, 97.5%), of which 5 isolates were positive for plasmid mediated quinolone resistance gene qnrS along with mutations in gyrA/parC genes demonstrating underlying causes of resistance to fluoroquinolones. In our previous study, we found ciprofloxacin resistance ESBL-producing E. coli were positive for qnrS gene [38] and similar findings were found in a study from Spain in 2008 [27]. This may reflect the indiscriminate and inappropriate use of antibiotics in Bangladesh where these drugs are often sold over the counter without a physician's prescription [15]. All ESBL-producing E. coli isolates were susceptible to carbapenems, the current drug of choice for the treatment of patients infected with multidrug-resistant ESBL-producing bacteria.
Similar to other studies within or outside of Bangladesh, we found that all clinical isolates were positive for CTX-M-1 group and CTX-M-15 specific genes [7,39]. Moreover, previous reports from Bangladesh showed a high prevalence of CTX-M-15 group in ESBL producing E. coli isolated from household tap water and pigeons [37,38,40]. The CTX-M type beta-lactamases represent a rapidly emerging group worldwide, which have been found predominantly in Enterobacteriaceae, particularly in E. coli, Klebsiella. pneumoniae, Proteus mirabilis and Salmonella Typhimurium [2,8,[41][42][43].
Sequence analysis of PCR product of TEM-1 and OXA-1 genes from selected isolates showed that all strains were identical to TEM-1 and OXA-1 which correlates with other studies [44,45]. ESBL genes located on integrons-like structures are being increasingly reported worldwide [10,41]. Hence, we examined the presence of class 1 and class 2 integrons. In our study, 50% (n = 20) of E. coli isolates harbored class 1 integron while none of the isolates contained class 2 integron as found in previous studies [46,47]. However, further study is needed to investigate the association between class 1 integron production and dissemination of ESBL type genes in E. coli from Bangladesh.
Interestingly, we identified no virulence genes (ipaH, ial, lt, st, eae, stx 1 , stx 2 , and e Agg genes). These virulence genes are normally present in diarrhoeagenic E. coli which cause gastrointestinal  infections [48]. Therefore, it can be explained that E. coli causing extra-intestinal infections such as UTI and surgical wound infection might have a different set of virulence factors than that of E. coli causing intestinal infection.
Apart from serotyping, plasmid profiles of all 40 isolates were compared between each other to demonstrate variation and epidemiological linkage among ESBL producing isolates. All isolates contained multiple plasmids ranging in size from 1.1 to 120 MDa and plasmid patterns of these isolates were heterogeneous (Table S1), which correlates with previous findings [38]. Plasmids ranged in size either 53 or.120 MDa or both were detected in all the ESBL-producing E. coli strains supporting previous findings that plasmids carrying beta-lactamase genes are ranged in size from 53 to 200 MDa [49]. Due to possible occurrence of intrinsic bla CTX-M genes further analyses are needed to determine the exact location of the bla CTX-M genes.
PFGE profile analysis showed heterogeneity among majority of isolates except for a few that could be clustered into a single PFGE type (9 isolates in type A, 4 isolates in type B, 3 isolates in type C and 2 isolates type D) ( Table 1). It is interesting to note that isolates grouped into PFGE pattern A were mostly belonged to serotype O25:H4 (6/8) of which 4 were from patients with UTI at outpatient department and the remaining 2 were from patients with wound infections attending hospitals. These isolates were isolated in 2006 from three different hospitals which are geographically distant from each other. However, PFGE analysis of the majority of isolates demonstrated a low clonal relationship (26 clones/40 isolates), similar findings reported in previous studies [6,50]. These findings can be explained by the fact that ESBL production might be a consequence of horizontal gene transfer between bacteria rather than the spread of specific bacterial clones.
In conclusion, high prevalence (11.8%) and presence of variety of beta-lactamase genes in ESBL-producing E. coli possibly reflects the overuse and misuse of antibiotics in Bangladesh and severely limits the therapeutic options in Bangladesh. Surveillance of multidrug-resistant Gram negative bacteria causing intestinal and extra-intestinal infections is essential to guide the empirical treatment strategies for these infections in Bangladesh. Nevertheless, this is the first effort to characterize the ESBL-producing clinical isolates in Bangladesh using molecular techniques.

Ethical statement
The Enteric and Food Microbiology laboratory of icddr,b is a research laboratory where all anonymous samples were tested for respective pathogens, and for this present analysis anonymous samples were collected from different hospitals. So no ethical approval was needed.