Characterization of extended spectrum β-lactamase (ESBL)-producing Escherichia coli in Asi (Orontes) River in Turkey

During the past few decades, antimicrobial resistance has been one of the most important global crises which significantly threaten human health and also add economic burden to health care systems worldwide. Unnecessary and inappropriate uses of antibiotics are considered to be the major contributing factors for the emergence of resistant strains (Berendonk et al. 2015). The use of antibiotics in veterinary medicine to prevent infections and to promote growth of livestock adds to the existing problem with the emergence of resistant strains (Smith et al. 2002). Apart from clinical settings, antibiotic resistant bacteria have been also shown to be prevalent in aquatic environments including sewage effluents of domestic, municipal and hospitals, groundwater and surface runoffs (Czekalski et al. 2012). Animal and human excreta as well as agricultural use are significant sources of pharmaceutical compounds sinking into wastewater systems, which consequently lead to the widespread distribution of antibiotic resistant bacteria in the aquatic environment (Halling-Sørensen et al. 1998). In addition, aquatic environments create ‘hotspots’ for exchange of antibiotic resistance genes among bacterial species with horizontal gene transfer through mobile genetic elements such as integrons, transposons, and plasmids (Moura et al. 2010; Amos et al. 2014). It has also been suggested that many human pathogens enter into the food chain due to the agricultural utilization of contaminated water sources (Benjamin et al. 2013).

The global dissemination of multidrug resistance among Gram-negative pathogens is becoming increasingly common via clinical specimens, foods of animal origin as well as aquatic environments which pose a dire threat for the community. Among Gram-negative pathogens, extended spectrum beta-lactamase-producing (ESBL) Escherichia coli have attracted a lot of attention (Shaikh et al. 2015). These strains encode enzymes conferring resistance to a wide variety of β-lactam antibiotics including penicillin, first, second and third-generation cephalosporins, and aztreonam (Malloy & Campos 2011; Gao et al. 2015). In the last decade, CTX-M has appeared to be the most dominant ESBL type among clinical isolates where more than 80 different CTX-M enzymes classified into five groups have been described so far (Bonnet 2004). ESBL-producing E. coli strains frequently contain resistance genes to other types of antibiotics (e.g. fluoroquinolones, aminoglycosides and tetracycline), which minimizes the availability of effective therapeutic options (Jacoby 2009; Pitout 2010). Over the past few years, ESBL-producing strains of E. coli have been frequently isolated from a variety of sources including clinical sources and a wide range of foods as well as occasionally from drinking water sources (De Boeck et al. 2012; Altınkum et al. 2013; Pehlivanlar-Önen et al. 2015). There are numerous studies showing the presence of genes (blaTEM, blaCTX-M and blaSHV) conferring resistance to β-lactam antibiotics in wastewater effluents and rivers around the world (Lu et al. 2010; Zurfluh et al. 2013).

Also known as Asi, Orontes River is one of the most important rivers in southern Turkey. It originates from Labweh, Lebanon and flows through for about 380 km within three countries and finally pours into the Mediterranean Sea in Hatay province (Göksu et al. 2005). The river has been egregiously polluted due to various forms of pollutions, particularly from industrial and agricultural discharges and most importantly untreated effluents from urban wastewater treatment plants (WWTP). Orontes River is the main source for agricultural irrigation as well as fishing for the communities in the area. There are reports of ESBL-producing E. coli strains originating from human and animal clinical samples and foods of animal origin in Turkey (Pehlivanlar-Önen et al. 2015; Kürekci et al. 2016). However, to the best of our knowledge, rivers in Turkey have received little attention in this respect and there no study has been carried out to investigate ESBL-producing strains and the carriage of ESBL types in rivers in Turkey.

The objective of the present study was to characterize the cefotaxime resistant E. coli strains isolated from the Orontes River and WWTP in Hatay, Turkey. The presence of ESBL gene variants (bla_SHV, bla_TEM, bla_OXA, and bla_CTX-M) and the prevalence of plasmid-encoded sulphonamide resistance genes in these isolates were also examined.

Sewage water samples were collected on three occasions during July, September and December 2014. Samples were collected from the influent and the effluent of a WWTP located in Hatay province with more than 500,000 inhabitants. The WWTP receives sewage (approximately 200–300 L/second) from domestic, hospitals and slaughterhouses in the province and discharges treated sewage (200–300 L/second) directly into Orontes River without disinfection treatment. The WWTP processes influent water through physical treatment including screening, grit removal, gravity sedimentation and trickling filter, followed by secondary sedimentation and discharge treated effluents into Orontes River without chlorination. Water samples of Orontes River were also collected from two locations, i.e. upstream of the Hatay city center (approximately 5 km away from the center) and after passing the city center which is 500 m before the discharge point of the effluent of the WWTP. Water samples were collected in sterile glass bottles, transported to the laboratory in an icebox within 2 h of collection and processed immediately for microbiological analysis.

Membrane fecal coliform (mFC) agar with cefotaxime (2 μg/mL) was used to isolate ESBL-producing E. coli. One hundred μL of each sample was inoculated onto mFC agar and incubated at 37 °C for 3 h, then incubated overnight at 44 °C. Up to five suspected E. coli colonies (the ones with black color) on mFC agar were chosen and inoculated on blood agar (5% defibrinated horse blood) and incubated at 37 °C for 24 h. The cefotaxime resistant strains were identified as E. coli by the Phoenix system and subsequently confirmed by polymerase chain reaction (PCR) amplification of the universal stress protein (uspA) gene as described by Chen & Griffiths (1998). Genomic DNA of these isolates was extracted by using a standard boiling method as described by Anastasi et al. (2012). The primer sequences used in the current study and reaction conditions are given in Table 1.

A double disc diffusion method was used for the confirmation of ESBL production according to the guidelines described by the Clinical Laboratory Standards Institute (CLSI 2012a). Klebsiella pneumonia (ATCC 700603) was used as the standard strain.

Antimicrobial susceptibility testing of confirmed ESBL positive E. coli strains was performed using Phoenix system to determine the minimum inhibitory concentrations (MICs). The following panel of 20 antimicrobials (Phoenix) were used in the current study: amikacin (AM), ampicillin-sulbactam (AMP-SUL), aztreonam (AZT), cefazolin (CFZ), cefepime (FEP), cefoperazone-sulbactam (CEF-SUL), cefoxitin (FOX), ceftazidime (CAZ), ceftriaxone (CTR), ciprofloxacin (CIP), levofloxacin (LEV), colistin (COL), ertapenem (ERT), imipenem (IMI), meropenem (MER), gentamicin (GEN), piperacillin-tazobactam (PIP-TAZ), ticarcillin-clavulanate (TIC-CLA), tigecycline (TIG), and trimethoprim-sulfamethoxazole (STX). In this study, according to MIC values, antimicrobial drug susceptibilities were classified as susceptible, intermediate resistant, and resistant according to the guidelines published by CLSI (2012b). Any intermediate resistant results were counted as susceptible throughout this study.

In order to exclude homologous isolates, one representative isolate among those with the same resistance pattern and originating from the same source and time points was selected. A total of 54 selected ESBL-producing E. coli isolates were examined for their phylogenetic groups (PGGs) using the new quadruplex PCR, as previously reported by Clermont et al. (2013).

All E. coli strains (n = 54) were typed using PhP fingerprinting method as described by Gündoğdu et al. (2011). In this study high-resolution biochemical-fingerprinting PhP-RE plates, which are specifically developed for typing of E. coli, were used (Landgren et al. 2005). The biochemical fingerprinting values obtained after typing of the isolates were compared pairwise and the similarity between strains was calculated as the similarity coefficient and clustered using UPGMA clustering methods (Sneath & Sokal 1973; Saeedi et al. 2005). Isolates having similarity coefficients with each other greater than the default identity level of the software (0.975) were regarded as identical and assigned to the same PhP type. All data handling, including calculations of correlations coefficients, as well as clustering, was performed using the PhPlate software v. 4002 (PhPlate AB). Strains were regarded as belonging to a common clonal group if they had the same PhP type and PGG; otherwise, they were referred to as a single clone.

The presence of the bla_TEM, bla_OXA, bla_SHV, and bla_CTX genes was tested by PCR as previously described (Ahmed et al. 2007). Further, the subtypes of β-lactamase genes were determined by sequencing of the PCR amplicons. The obtained sequencing data were compared to those in GenBank data library using BLASTn program.

The presence of sul genes (sul1, sul2 and sul3) were determined using a multiplex PCR method as described previously by Kerrn et al. (2002) and a multiplex PCR was performed to determine the presence of different classes of integron-associated integrase genes (int1, int2, and int3) as described by Dillon et al. (2005).

During the initial phase of isolation, a total of 65 cefotaxime resistant E. coli strains were recovered from river water (n = 30) and WWTP (n = 35) samples and were identified as ESBL producers by phenotypic test according to the CLSI-recommended double-disk diffusion method. Of these, 54 ESBL-producing E. coli were included for further studies and the remaining 11 isolates were excluded due to having the same antimicrobial resistance patterns as well as coming from the same samples.

In river water samples, the most common PGG was group A (52%) followed by group C (16%) and E (16%), and D (8%) and B (8%). Almost a similar pattern was found among E. coli strains isolated from WWTP samples. The most common PGG in these samples was group A (48.3%) followed by group C (27.6%), B (10.4%), E (6.9%) and finally D (6.9%) (Table 3). Based on the combination of PhP-typing and phylogenetic grouping, 14 isolates were grouped under four common clonal types (CTs) comprising of between two and five isolates each. The remaining 40 isolates belonged to single clonal types (STs) (Table 2).

According to the PCR and sequencing analysis, among those 54 cefotaxime resistant E. coli strains with ESBL phenotype, 20 (37%) possessed bla_CTX-M-15 and 16 (29.6%) isolates had bla_CTX-M-15 + bla_TEM-1, while only one (1.9%) isolate was found to have bla_CTX-M-1 + bla_TEM-1 (Table 3). In addition, 14 (25.9%) isolates had the bla_TEM-1β-lactamase gene alone. Nevertheless, none of the isolates were found to be SHV or OXA positive.

At least one sul gene was found in 37 (68.5%) isolates of which 31 (57.4%) isolates contained both sul1 and sul2 genes. Additionally, sul2 gene alone was found in five (9.3%) isolates and sul1 gene alone was found in one isolate. None of the isolates were found to be carrying the sul3 gene (Table 3). Thirty-three isolates (61.1%) were found to be positive for the presence of integrase genes. Of these, 31 isolates (57.4%) carried int1 and two isolates (3.7%) carried class 2 integron. None of the isolates carried both int1 and int2 or int3 genes (Table 3).

Since their discovery in Western Europe in 1984, ESBL-producing strains have been continually isolated worldwide with increasing frequency among clinical strains as well as isolates from healthy humans and foods of animal origin (Hu et al. 2013). In addition, ESBL-producing bacteria have been shown to be widely distributed in the environment, and in particular in soil, plants and water bodies such as rivers, lakes, wastewater effluents and occasionally in drinking water (Hu et al. 2013; Zurfluh et al. 2013). However, there are not many reports on the persistence of antibiotic resistance of pathogenic bacteria and the molecular basis for resistance phenotype in aquatic environments in Turkey (Toroğlu & Toroğlu 2009; Ozgumus et al. 2009). To the best of our knowledge, there is no report related to the ESBL-producing E. coli in the rivers and WWTP in this country so far. Hence, the current study for the first time highlights the presence and the importance of ESBL-producing E. coli in the Orontes River with the nearby WWTP being the most probable contributing source.

The presence of ESBL-producing E. coli strains in surface waters in the current study is in agreement with previous studies focused on the aquatic environments in European countries (Amos et al. 2014; Blaak et al. 2014), Asian countries (Lu et al. 2010; Su et al. 2012; Hu et al. 2013), Tunisia (Ben Said et al. 2016) and Australia (Gündoğdu et al. 2013). Interestingly, despite the presence of few CTs of these bacteria in both sources sampled, the majority of ESBL-producing strains belonged to diverse clonal groups based on their PhP-PGG patterns. The PhP system used in this study has been shown to be a highly discriminatory typing method and as powerful as molecular typing methods such as RAPD-PCR (Ramos et al. 2010) and ERIC-PCR (Ansaruzzaman et al. 2000) for typing of E. coli strains. Amos et al. (2014) also reported a high genetic diversity among E. coli strains carrying bla_CTX-M-15 genes in river sediment samples in the UK and suggested that they were of WWTP origin. According to the phylogenetic classification by Clermont et al. (2013), the majority of isolates belonged to commensal PGG A and C whereas only a few strains (n = 4; 7.4%) belonged to phylogroup D, indicating low prevalence of extraintestinal pathogenic strains in water samples. This is consistent with the results of a recent study on ESBL-producing E. coli strains obtained from rivers and lakes in Switzerland (Zurfluh et al. 2013). It has to be mentioned however, that the most common phylogroup in our river samples, i.e. PGG A, has been determined as the least abundant group in foods of animal origin in Turkey (Pehlivanlar-Önen et al. 2015).

Examining the antibiotic resistance profiles of 54 ESBL-producing E. coli strains, it was found that more than 65% of the ESBL-producing E. coli isolates were resistant to one of the three antibiotics, i.e. CIP, LEV, and STX. This is not surprising as ESBL-producing E. coli have been shown to be commonly resistant to different antibiotic classes, particularly to quinolone and aminogylcoside and sulphonamides due to the carriage of resistance genes on the same conjugative plasmids (Wang et al. 2013). A recent study carried out in the Netherlands has reported that ESBL-producing E. coli isolated from recreational water were resistant to nalidixic acid (60%), CIP (42%) and cefoxitin (3%) (Blaak et al. 2014). As shown in another study (Lu et al. 2010), PIP-TAZ and CEF-SUL have been the most potent ESBL enzyme inhibitors with 24.1 and 31.2% resistance, respectively, while AMP-SUL was the least efficient with 100% resistance. All ESBL-producing E. coli strains were sensitive to carbapenems, which is of great importance as these molecules are one of the last resort antibiotics for ESBL-producing bacteria for humans. Similar data have been reported for the high susceptibility of E. coli to carbapenems (Diallo et al. 2013; Blaak et al. 2014).

The most common ESBLs genes among our E. coli strains was the CTX-M group found in 68% of the isolates. This was not surprising as these enzymes are the most predominant ESBLs among E. coli strains isolated from human clinical samples (Altınkum et al. 2013) as well as foods of animal origin (Pehlivanlar-Önen et al. 2015; Kürekci et al. 2016). Widespread dissemination of E. coli carrying CTX-M enzyme has also been reported in wastewater, surface water and river sediments worldwide. In a detailed study, Zurfluh et al. (2013) investigated the occurrence of ESBL-producing Enterobacteriaceae in rivers and lakes in Switzerland and showed a strong correlation between the strains isolated from clinical samples, healthy human and food producing animal isolates with the environmental isolates in this country. In our study CTX-M-15 was found to be the most common ESBL type, which is consistent with the findings reported from England and the Netherlands (Amos et al. 2014; Blaak et al. 2014). For example, Amos et al. (2014) found CTX-M-15 type as the predominant ESBLs detected in E. coli from a river in England, which was linked to the high abundance of CTX-M-15 type in the effluent of WWTP. Blaak et al. (2014) also reported a similar finding on widespread dissemination of CTX-M enzyme, with CTX-M-15 being the most dominant type in recreational water in the Netherlands. The predominance of CTX-M-15 type in the current study could also be explained by selective enrichment with cefotaxime. In fact it has been shown that isolates carrying CTX-M-1 group ESBLs were significantly more resistant to CAZ and cefepime when compared to other CTX-M groups (Hu et al. 2013). On the other hand, Korzeniewska & Harnisz (2013) only found OXA-1 and TEM-1 types in their study in Poland, but did not detect CTX-M types.

A combination of trimethoprim and sulphamethoxazole is an important synthetic antimicrobial agent used for the treatment of urinary tract infections caused by E. coli (Grape et al. 2003). Sulfonamides, alone or in combination with other antimicrobial compounds, have been widely used to treat various infectious diseases caused by bacteria, toxoplasma, and protozoa in animas as well as growth promoting agents in swine production (Prescott 2013). Resistance to sulfonamides among E. coli strains is mediated either through the mutations in the chromosomal dihydropteroate synthase gene (folP) or through the acquisition of an alternative DHPS gene (sul) (Sköld 2000) for which three sul genes (sul1, sul2 and sul3) have been described so far (Prescott 2013). We previously found that 90% of the E. coli strains carrying sul genes in surface waters are similar to those obtained from uropathogenic E. coli strains (Gündoğdu et al. 2011). In the present study, 68.5% of ESBL positive isolates were found to have sul genes (sul2; 66.6% and sul1; 59.2%). These results are not surprising since DHPS gene mediated sulphonamide resistance has become common among E. coli isolates from clinical samples, food producing animals and foods of animal origins (Sköld 2000; Soufi et al. 2011). We found that 61% of isolates carried integrase genes, with the majority carrying int1. Detection of class I integrons varies in different studies and ranges between 64% (Ben Said et al. 2016) and 41% in surface waters (Chen et al. 2011). The presence of E. coli strains carrying integrin classes 1 and 2 has also been reported in rivers in the northern region of Turkey (Ozgumus et al. 2009).

In conclusion, we found a high level of CTX-M-15 type ESBL enzyme together with sul genes among our isolates, indicating the importance of the Orontes River as a potential reservoir of antimicrobial resistance genes. We also found a high diversity of ESBL-producing E. coli strains in both the river and WWTP samples which point at the diversity of sources of contamination of the Orontes River in this country.

This study was supported by a Mustafa Kemal University Scientific Research Fund (Project Number: BAP-14621).

The authors declare that there is no conflict of interest with the organization that sponsored this research and publications arising from this research.