The characterization of ESBL genes in Escherichia coli and Klebsiella pneumoniae causing nosocomial infections in Vietnam

Background: Extended-spectrum β-lactamases (ESBLs) are enzymes capable of hydrolyzing oxyimino-β-lactams and inducing resistance to third generation cephalosporins. The genes encoding ESBLs are widespread and generally located on highly transmissible resistance plasmids. We aimed to investigate the complement of ESBL genes in E. coli and Klebsiella pneumoniae causing nosocomial infections in hospitals in Ho Chi Minh City, Vietnam. Methodology: Thirty-two non-duplicate isolates of E. coli and Klebsiella pneumoniae causing nosocomial infections, isolated between March and June 2010, were subjected to antimicrobial susceptibility testing. All isolates were PCR-amplified to detect the blaSHV, blaTEM and blaCTX-M ESBL genes and subjected to plasmid analysis. Results: We found that co-resistance to multiple antimicrobials was highly prevalent, and we report the predominance of the blaCTX-M-15 and blaCTX-M-27 genes, located on highly transmissible plasmids ranging from 50 to 170 kb in size. Conclusions: Our study represents a snap shot of ESBL-producing enteric bacteria causing nosocomial infections in this setting. We suggest that antimicrobial resistance in nosocomial E. coli and Klebsiella pneumoniae is rampant in Vietnam and ESBL organisms are widespread. In view of these data and the dramatic levels of antimicrobial resistance reported in Vietnam we advocate an urgent review of antimicrobial use in the Vietnamese healthcare system.


Introduction
The production of β-lactamases is the most common mechanism of bacterial resistance to the βlactam antimicrobials. β-lactamase genes are widespread and mutate in response to continuous antimicrobial exposure. This prolonged exposure has led to the emergence of extended-spectrum βlactamases (ESBLs) [1]. ESBLs are enzymes capable of hydrolyzing oxyimino-β-lactams, such as third generation cephalosporins, which include the commonly used antimicrobials, ceftriaxone and cefixime. The dissemination of ESBLs is a global problem, particularly in sentinel members of the Enterobacteriaceae [2]. Among the known ESBL enzymes, the CTX-M-type β-lactamases, which preferentially hydrolyze cefotaxime, were first reported in the late 1980s and have undergone a rapid, global spread. The spread of CTX-M-type βlactamases has been dramatic and greater than the impact of the TEM-and SHV-type ESBLs [3][4][5][6].
In Vietnam, the presence of pathogens expressing ESBLs has been increasingly reported over the past ten years. A study conducted in 2001 in seven hospitals across Ho Chi Minh City in the south of Vietnam, found that 5.6 % of all Gram negative bacterial isolates were ESBL positive, with the rate of positivity in Escherichia coli and Klebsiella pneumonia being 58 % and 23.6 %, respectively [7]. A further study, also conducted in Ho Chi Minh City, between February 2002 and May 2005, found that 33 % of all Gram-negative bacterial isolates were ESBL positive. From these ESBL positive isolates, E. coli and K. pneumoniae accounted for 74.1 % of all the organisms isolated [8]. A pan Asia-Pacific study regarding Gram-negative bacilli from intra-abdominal infections in 2007, found that the ESBL positivity rate in Vietnam was 35.6 % (34.4 % and 39.1 % of the ESBL positive strains were E. coli and K. pneumoniae, respectively) [9]. As such, the prevalence of ESBL producing strains in Vietnam and the Asia-Pacific region is now higher than those observed in Europe, suggesting differing geographical pressures and exposures to antimicrobials [9,10].
ESBL producers can often transfer resistance to multiple bacterial species through plasmid-mediated conjugation [11]. The widespread use of antimicrobials, coupled with the transmissibility of resistance determinants mediated by plasmids, transposons, and integrons, contribute to increasing the prevalence of antimicrobial resistance in pathogenic members of the Enterobacteriaceae [12]. These elements pose serious problems in hospital settings worldwide. Therefore, surveillance of ESBLs producing Enterobacteriaceae is necessary to add insight into ESBL transmission, the emergence of predominant ESBL groups and the mobile elements inducing the dissemination of ESBL determinants. In Vietnam, limited studies have been performed investigating the molecular characteristics of ESBL genes and their corresponding mobile elements. Here, we aimed to define the characteristics of common ESBL genes and their encoding plasmid profiles in members of the Enterobacteriaceae causing nosocomial infections in hospitalized patients in Ho Chi Minh City, Vietnam.

Ethics statement
This study was conducted according to the principles expressed in the Declaration of Helsinki and was approved by the institutional ethical review boards of the participating hospitals. Samples were collected as part of "standard of care" for treatment and diagnosis; therefore, the institutional ethical review boards did not require us to collect informed consent.

Clinical isolates, antimicrobial susceptibility testing and ESBL phenotyping
The microbiology laboratories at Cho Ray and Thong Nhat hospitals in Ho Chi Minh City isolated 72 bacterial isolates (E. coli or K. pneumoniae) causing nosocomial infections demonstrating resistance to ceftriazone and ceftazidime between March and June, 2010. Thirty-two of these isolates, comprising 23 E. coli and 9 K. pneumonia, were latterly confirmed to be ESBL producing at the laboratories of Oxford University Clinical Research Unit by the double-disc synergy test (the remainder were ESBL negative) [13]. The double-disc synergy method utilizes discs containing cefotaxime (CTX) (30 µg) and ceftazidime (CAZ) only (30 µg) and both antimicrobials in combination with clavulanic acid (CLA) (10 µg). ESBL producing strains were identified as those with a greater than 5 mm increase in zone with the single antimicrobial compared to the combined antimicrobials. All 32 bacterial isolates were additionally subjected to susceptibility testing by assessing the minimum inhibitory concentrations (MICs) against amoxicillin/ clavulanic acid (AMC), cefepime (FEP), ceftriaxone (CRO), imipenem (IPM), ciprofloxacin (CIP), nalidixic acid (NAL), trimethoprim/ sulfamethoxazole (SXT) and chloramphenicol (CHL) by E-test (AB Biodisk, Solna, Sweden). All antimicrobial susceptibility tests were performed on Mueller-Hinton agar and the resulting data were interpreted according to the Clinical and Laboratory Standards Institute guidelines [13].

Nucleic acid amplification and sequencing
Genomic DNA was isolated from all bacterial isolates from 1 mL of an overnight bacterial culture using the Wizard Genomic DNA Extraction Kit (Promega, Fitchburg, USA), according to manufacturer's recommendations. All isolates were screened for the presence of bla SHV , bla TEM , bla CTX-M ESBL genes using previously published primers [14,15]. Further characterization of the bla CTX-M was performed using the primers specific for CTX-M-1, CTX-M-2, CTX-M-9 subgroups [11]. PCR amplifications were performed using 30 cycles, of 30s at 95 o C, 30s at 57 o C, and 90s at 72 o C. All PCR amplifications were performed using Taq DNA polymerase and appropriate recommended concentrations of reagents (Bioline, London, UK). All PCR amplicons were sequenced using big dye terminators in a forward and reverse orientation on an ABI3130XL sequencing machine (ABI, Advanced Biotechnology Inc, Columbia, USA), according to the manufacturer's recommendations. Resulting DNA sequences were verified and aligned using BioEdit and Vector NTI Suite 7 software. BLASTn at NCBI was used to compare all resulting ESBL gene sequences against additional ESBL sequences.

Plasmid extraction and visualization
Plasmid DNA was isolated from all ESBL bacterial isolates using an adapted methodology originally described by Kado and Liu [11]. Briefly, plasmid DNA was separated by agarose gel electrophoresis in 0.7 % agarose gels with 1X TBE buffer. Agarose gels were subjected to 100V for 4 hours, stained with ethidium bromide and photographed. E.coli 39R861 containing plasmids of 7, 36, 63 and 147 kb was used for sizing plasmid extractions on agarose gels [16]. Plasmid DNA was size-separated and analyzed using Bionumerics software (Applied Maths, Sint-Martens-Latem, Belgium).

Southern blotting and hybridization
Plasmid DNA was electrophoresed and transferred to a Hybond N + membrane (Amersham Biosciences, Little Chalfont, UK). The PCR amplicons of bla TEM , bla CTX-M-1 , bla CTX-M-9 were labeled with horseradish peroxidase using the ECL direct nucleic acid labeling and detection systems kit (Amersham Biosciences, Little Chalfont, UK), and were used as hybridization probes. Hybridization and detection were performed according to the manufacturer's instructions.

Bacterial conjugation
Conjugation was performed by combining equal volumes (500 µL) of overnight cultures grown in Luria-Bertani (LB) media of donor and recipient strains in 4 mL of sterile LB media. The donor strains were ESBL-producing isolates (E. coli and K. pneumoniae) and the recipient was a sodium azide resistant E. coli (strain J53 resistant). Bacteria were mixed in a 1:1 ratio at 37 o C and incubated without agitation overnight. Transconjugants were selected on LB media containing sodium azide (100 µg/mL) and ceftriaxone (6 µg/mL) and were verified by plasmid extraction and visualization, as before. Conjugation frequency was calculated by dividing the mean number of transconjugants by the mean number of recipient cells.

Characterization of bla genes
PCR amplifications were performed to detect the bla TEM , bla SHV and bla CTX-M genes. The resulting amplifications demonstrated that all 32 of the ESBLproducing isolates carried a bla CTX-M gene, 24/32 isolates harbored an additional bla TEM gene and no isolates carried a bla SHV gene. All strains were additionally amplified with primers specific for the three major CTX-M clusters, bla CTX-M-1 , bla CTX-M-2 and bla CTX-M-9 . With these specific CTX-M cluster primers, one E.coli isolate, produced an amplicon with both bla CTX-M-1 and bla CTX-M-9 primers, the remaining strains produced single amplicons with either the bla CTX-M-1 primers or the bla CTX-M-9 primers and none tested positive with the bla CTX-M-2 primers ( Table 2). All 33 PCR amplicons were DNA-sequenced to identify the specific bla CTX-M variants. DNA sequence analysis showed that multiple bla CTX-M loci were circulating in the E. coli and K. pneumoniae isolates. We

Characterization of ESBL encoding plasmids
Plasmid profiling of 32 ESBL-positive isolates demonstrated that all strains harbored at least one large plasmid, ranging from 50 to 171 kb in size (Table 2). Furthermore, the majority of the strains also harbored multiple other plasmids, ranging from 1 to 50 kb in size. The number of plasmids in these isolates ranged from one to nine and after gel sizing analysis, using a binary scoring system with a Pearson correlation, we found that plasmid profiles were not specific to E. coli or K. pneumoniae (data not shown).
Plasmid DNA hybridization demonstrated that the majority of the ESBL-producing strains (31/32 strains) carried the bla genes on a large plasmid (ranging from 53.8 to 157 kb in size) ( Table 2). These large plasmids encoded a bla CTX-M only or both a bla CTX-M and a bla TEM gene. Among the 24 strains carrying both a bla CTX-M and bla TEM genes, 18 strains carried these genes on the same plasmid and four strains carried these genes on different plasmids; we were unable to confirm the PCR result for two strains as a presumed consequence of plasmid instability after in vitro passage (this was, however, latterly confirmed by PCR) ( Table 2). ESBL-producing strains containing genes belonging to the bla CTX-M-9 gene cluster exhibited less activity against ceftazidime in comparison to strains carrying gene belonging to the bla CTX-M-1 gene cluster. Susceptibility testing against ceftazidime with ESBL strains showed two distinct zone clearance areas with the bla CTX-M-9 cluster (median; 18.3 mm) and the bla CTX-M-1 cluster (median; 12.2 mm) ( Table 2).
We performed bacterial conjugation experiments on all 32 ESBL-producing strains (donors) using E. coli J53 as a recipient. Results demonstrated that plasmids harboring ESBL genes of twenty-two isolates (69%) were transmissible via conjugation, with conjugation frequencies ranging from 6.25 x 10 -8 to 1 x 10 -3 per recipient cell ( Table 2). Of the 10 isolates carrying non-transmissible plasmids, eight carried ESBL plasmids with sizes greater than 100 kb and we were unable to confirm the presence of ESBL genes on plasmids by Southern Blotting in two (Table 2). We further confirmed the transmission of ESBL plasmids between both species (E. coli to E. coli) and genus (K. pneumoniae to E. coli).

Discussion
Our work shows that the CTX-M-type ESBLs are the most common ESBL found amongst E. coli and K.  [3,10,17,18]. This particular bla CTX-M gene is generally found on large conjugative plasmids and is located downstream of an ISEcp1 insertion sequence which explains its remarkable transmission success [7,11]. The CTX-M-15 type enzyme differs from that of CTX-M-3 type by an asparagine to glycine substitution at codon 240, which leads to increased activity against ceftazidime. These CTX-M-15 type enzymes may have been selected by the increasing use of ceftazidime in clinical practice [19][20][21].
We can additionally show that the ESBLproducing organisms additionally exhibited coresistance against multiple antimicrobials from other classes. Many studies have also reported co-resistance to tetracycline, aminoglycosides, and fluoroquinolones in ESBL producers [7,8,11]. It has also been demonstrated that CTX-M-15 ESBL hydrolyzes cefepime with higher efficiency than other ESBL variation [5]. Here, ESBL producers also demonstrated a high level of resistance against ciprofloxacin, trimethoprim-sulfamethoxazole, nalidixic acid and chloramphenicol. Our work shows that ESBL producing strains carrying bla CTX-M-15 exhibit complete resistance and intermediate resistance to cefepime with significantly higher MICs than other bla CTX-M alleles. This complexity in antimicrobials resistance combinations limits suitable drug of choice for antimicrobial therapy, leaving cabapenems and aminoglycosides the last options for treatment in some cases. The emergence of NDM-1 clearly compounds potential treatment options, and more recent data additionally suggests an association between resistance to beta-lactams and aminoglycosides in ESBL-producing bacteria [22]. Furthermore, organisms carrying ESBLs are highly efficient at transferring their resistance to other organisms within the same or different species through conjugation, increasing the rate of antimicrobial resistance transmission. Selective pressure, from heavy use of extended-spectrum beta lactam will presumably maintain the presence of these ESBL-producing pathogens resulting in the persistence and transmission of ESBL resistance determinants among Gramnegative bacteria. Therefore, further characterization of other antimicrobial resistance mechanisms will be important to understand the co-transmission of a range of antimicrobial determinants.
Our study represents a snap shot of ESBL producing enteric bacteria causing nosocomial infections in our setting. We report that antimicrobial resistance in hospital isolates is common in Vietnam and ESBL organisms are widespread. CTX-M-type ESBLs were the most common enzymes found in both E. coli and K. pneumoniae. Furthermore, the ESBL genes were consistently located on highly transmissible plasmids ranging from 50 to 170 kb in size. We suggest that the rampant use of extendedspectrum cephalosporins in the hospital is driving the on-going selection, maintenance and dissemination of these ESBL genes across a spectrum of Gram-negative organisms and recommend a stringent review of antimicrobial use in the Vietnamese healthcare system.