|Home | About | Journals | Submit | Contact Us | Français|
We describe a modification of the most probable number (MPN) method for rapid enumeration of antimicrobial-resistant Escherichia coli bacteria in aqueous environmental samples. E. coli (total and antimicrobial-resistant) bacteria were enumerated in effluent samples from a hospital (n = 17) and municipal sewers upstream (n = 5) and downstream (n = 5) from the hospital, effluent samples from throughout the treatment process (n = 4), and treated effluent samples (n = 13). Effluent downstream from the hospital contained a higher proportion of antimicrobial-resistant E. coli than that upstream from the hospital. Wastewater treatment reduced the numbers of E. coli bacteria (total and antimicrobial resistant); however, antimicrobial-resistant E. coli was not eliminated, and E. coli resistant to cefotaxime (including extended-spectrum beta-lactamase [ESBL] producers), ciprofloxacin, and cefoxitin was present in treated effluent samples.
The emergence and dissemination of antimicrobial resistance are well established as clinical problems that affect human and animal health. Escherichia coli is an important element of the flora of the human and animal intestine and a significant pathogen associated with gastrointestinal infection, urinary tract infections, and a variety of other extraintestinal infections (4). E. coli shed into the environment can survive for significant periods (7, 14, 23). Detection of E. coli in water and food is widely used as a microbiological indication of fecal contamination.
Data on the significance of environmental contamination with antimicrobial-resistant E. coli for human health are limited. Previous reports have shown that antimicrobial-resistant strains of bacteria are present in various effluents, such as hospital effluent discharge (8, 10, 16, 21), inflow effluent to a wastewater treatment plant (WWTP) (15), and outflow-treated effluent from a wastewater treatment plant (2, 12, 13, 18, 27). A wastewater treatment plant treating effluent from hospitals may be associated with discharge of relatively high levels of antimicrobial-resistant E. coli compared with those of a plant treating municipal effluent that does not include hospital effluent discharge (22). There are few reports of quantitative data on antimicrobial-resistant E. coli bacteria in effluent, reflecting the lack of a convenient method for their enumeration (12, 15, 22). Previous methods available for the detection of antimicrobial-resistant E. coli in a water sample have generally involved the isolation of E. coli and the selection of some isolates for susceptibility testing. In such cases, the proportions of antimicrobial-resistant organisms are based only on those isolates selected and are therefore not representative of the entire population. By adding the antimicrobial agent of interest to the water sample before testing, we have adapted a commercial most probable number (MPN) method (the Colilert system) for enumerating the total number of E. coli isolates resistant to that agent in a sample.
For validation of the method, suspensions of E. coli ATCC 25922 (susceptible to all tested agents) and of antimicrobial-resistant clinical E. coli strains (Table (Table1)1) were prepared to equal a 0.5 McFarland standard (~1.5 × 108 CFU/ml). Dilutions were prepared in sterile distilled water to give suspensions of E. coli within the enumeration range of the Colilert Quanti-Tray/2000 (IDEXX, Technopath, Limerick, Ireland) (<2,419.6 MPN/100 ml). Suspensions of E. coli ATCC 25922 alone and mixed with defined proportions of antimicrobial-resistant clinical strains were prepared. Total E. coli bacteria enumeration was performed on a 100-ml aliquot of suspension using Colilert Quanti-Trays (IDEXX, Technopath, Limerick, Ireland), according to the manufacturer's instructions.
Stock solutions of ampicillin, streptomycin, sulfamethoxazole, tetracycline, cefoxitin, cefotaxime, and ciprofloxacin (Sigma, Dublin, Ireland) were prepared. A volume of stock antimicrobial solution was added to a 100-ml aliquot of the aqueous sample to achieve the required final concentrations of ampicillin (32 μg/ml), streptomycin (32 μg/ml), sulfamethoxazole (256 μg/ml), tetracycline (4 μg/ml), cefoxitin (32 μg/ml), cefotaxime (2 μg/ml), and ciprofloxacin (4 μg/ml). The concentration of cefotaxime was selected to facilitate detection of extended spectrum beta-lactamase (ESBL)-producing E. coli, for which elevation of the cefotaxime MIC may be modest (3). Following addition of the antimicrobial agent, the specimen was processed in Colilert Quanti-Trays, according to the manufacturer's instructions. Following incubation at 37°C (+/−1°C) for 18 to 24 h, trays were read, according to the manufacturer's instructions. The backs of the trays were disinfected with 70% ethanol, and 1-ml aliquots were taken using a sterile syringe and needle (Sarstedt, Nümbrecht, Germany). E. coli was isolated from these aliquots and identified using API 20E (bioMérieux, Inc., Marcy L'Etoile, France). Antimicrobial susceptibility testing was performed by using Clinical Laboratory Standards Institute (CLSI) disc diffusion methods to confirm that positive wells reflected growth of targeted antimicrobial-resistant E. coli (3).
The wastewater treatment plant studied is located 1 km off the west coast of Ireland and has been in operation since May 2004. The plant is a secondary treatment facility with an average hydraulic loading in 2006 of 49,000 m3 per day. The population served by the plant is approximately 72,000 people. The effluent treated by the plant includes effluent from four hospitals.
Effluent samples were collected directly from the foul pump house sewer located on the grounds of hospital A. Hospital A is a tertiary referral hospital with 639 beds. Effluent from this source contains only effluent from hospital A. From this point, effluent joins the municipal city effluent before treatment at the municipal wastewater treatment facility. Municipal effluent was collected from points in the sewerage system upstream and downstream from the hospital A effluent discharge.
In total, 44 effluent samples were collected from hospital A (n = 17), upstream (n = 5) and downstream (n = 5) municipal effluent, secondary-treated effluent (n = 13), and effluent from throughout the treatment process (n = 4). The 17 hospital samples were collected between August 2006 and September 2008. Municipal effluent samples were collected between November 2006 and June 2008. Treated effluent samples were collected as follows: six samples were collected over a 10-day period in August/September 2006, six samples over a 3-day period (both AM and PM) in August 2007, and five samples on one sampling occasion from various stages of treatment in 2008. Samples were collected in 1-liter sterile bottles (Lennox, Dublin, Ireland) from either the direct outflow pipe of effluent from the plant before being emitted out to sea or as a composite sample, which consists of a continuous hourly collection of the final treated effluent over 24 h.
Samples were stored at 4°C until analysis. It was empirically determined that a 1:100,000 dilution for the municipal and hospital effluent discharge and a 1:1,000 dilution for the WWTP outflow yielded levels of E. coli that were within the range for enumeration in the Colilert Quanti-Tray. A 100-μl aliquot of effluent was added to 99.9 ml of sterile water (1:1,000 dilution) in sterile 100-ml pots (LIP, Galway, Ireland). Further dilutions were made as appropriate. Antimicrobial stock solutions were added, and trays were incubated and read as described above. Aliquots were removed from selected positive wells at random for isolation of E. coli, and susceptibility testing was carried out on isolates, as per CLSI guidelines (3). Testing with the following antimicrobial agents was performed: ampicillin (10 μg), chloramphenicol (30 μg), streptomycin (10 μg), sulfamethoxazole (300 μg), tetracycline (30 μg), trimethoprim (5 μg), nalidixic acid (30 μg), kanamycin (30 μg), ciprofloxacin (5 μg), cefpodoxime (5 μg), ceftazidime (30 μg), cefoxitin (30 μg), gentamicin (10 μg), minocycline (30 μg), and cefotaxime (30 μg) (Oxoid, Basingstoke, United Kingdom). E. coli ATCC 25922 was used as a control. ESBL production was confirmed by the combination disk method of the CLSI using cefpodoxime (10 μg), cefpodoxime-clavulanic acid (10 μg/1 μg), and the following Etests: cefotaxime, cefotaxime-clavulanic acid, cefepime, cefepime-clavulanic acid, ceftazidime, and ceftazidime-clavulanic acid, in accordance with the manufacturer's instructions (bioMérieux Inc., Marcy L'Etoile, France) (3). Klebsiella pneumoniae ATCC 700603 was used as a positive control.
Bacterial DNA was extracted using an extraction kit (Qiagen, West Sussex, United Kingdom). Confirmed ESBL-producing E. coli isolates were tested for blaTEM, blaSHV, and blaCTX-M by PCR, using specific primers as described by Leung et al. (11a), Stapleton et al. (26), and Woodford et al. (28). Amplicons of blaCTX-M were sequenced (Sequiserve, Germany), analyzed, and aligned using basic local sequence alignment software (BLAST). ESBL-producing E. coli were typed by pulsed-field gel electrophoresis (PFGE) with XbaI (11a), in accordance with PulseNet protocols (24). Statistical analysis was carried out using SPSS 15.0 software, with α set at 0.05. Due to the low numbers, all data were analyzed using nonparametric tests. Correlation was calculated using Spearman's correlation coefficient, and proportions were compared using the chi-square test (Mantel-Haenszel correction). Significant results should be interpreted as indicative only, as multiple comparisons were made without applying (Bonferroni) corrections.
The numbers of antimicrobial-susceptible and antimicrobial-resistant E. coli strains in each artificially contaminated sample are summarized in Table Table1.1. The results confirm that only the antimicrobial-resistant population of E. coli is enumerated in the presence of the relevant antimicrobial agent. Testing on isolates from the antimicrobial-containing trays confirmed that the E. coli isolates growing in the tray were the antimicrobial-resistant isolates. In artificially contaminated samples, ciprofloxacin-resistant E. coli strain 228 was detected in suspensions, in which it represented as little as 0.0001% of the total E. coli population (approximately 15 CFU of E. coli 228 mixed with 1.5 ×107 E. coli ATCC 25922).
(a) Hospital effluent discharge. Data are summarized in Table Table2.2. Ampicillin-resistant E. coli was detected in every sample of hospital A effluent and was generally present in higher proportions than other antimicrobial-resistant E. coli, accounting for 100% of the total population in one sample (HA19). E. coli isolates resistant to streptomycin, sulfamethoxazole, and tetracycline were also present at relatively high levels in most samples. E. coli isolates resistant to the cephalosporins and ciprofloxacin were present less consistently and generally at lower levels. A higher proportion of E. coli resistant to ampicillin is correlated to a high proportion of E. coli resistant to streptomycin (r = 0.8; P = 0.001). No other correlation between the proportions could be observed.
(b) Municipal effluent.The number of E. coli isolates in samples of effluent downstream from hospital A was significantly reduced in comparison with the number of those isolates in samples of effluent taken upstream (P = 0.047). In municipal effluent downstream from the hospital effluent discharge, the proportion of E. coli resistant to all tested antimicrobial agents was considerably higher than the proportion of that in the upstream samples, but this reached borderline statistical significance for only one antimicrobial agent (sulfamethoxazole) (P = 0.047).
(c) Effluent from wastewater treatment plant. The number of E. coli isolates (MPN/100 ml) in the treated effluent sample ranged from 5.76 × 103 to 1.55 × 105, representing an approximately 1,000 (3-log10) decrease from those in municipal and hospital effluent discharge samples; however, a large proportion of E. coli resistant to each antimicrobial agent was detected in the treated effluent sample.
Analysis shows that total E. coli isolate counts at consecutive sampling points through the treatment process show a significant decline (linear R2 = 80.6%; P = 0.039). This decline was observed in the following order: raw intake effluent (stage 0), post-return effluent (stage 1), primary-treated effluent (stage 2), aeration effluent (stage 3), and final treated effluent (stage 4). The percentage of antimicrobial-resistant E. coli in effluent at sampling points throughout the treatment process showed no consistent pattern of increase or decrease using linear regression analysis.
As the numbers of effluent samples taken from throughout wastewater treatment were limited, these results are indicative rather than conclusive.
After enumeration of total E. coli bacteria in effluent samples, 254 isolates were collected from effluent samples from individual Colilert Quanti-Tray wells (Table (Table3).3). All isolates were resistant to the antimicrobial agent used in the corresponding Colilert Quanti-Tray, thus confirming the specificity of the method for isolation of the targeted antimicrobial-resistant E. coli isolates. Isolates were obtained from hospital A effluent samples (n = 106), municipal effluent samples upstream (n = 36) and downstream (n = 36) from hospital A, treated effluent samples (2006 and 2007) (n = 47), and effluent samples from throughout the treatment process (n = 29). These isolates were obtained from trays selecting for resistance to ampicillin (n = 60; 24%), streptomycin (n = 44; 17%), sulfamethoxazole (n = 44; 17%), tetracycline (n = 40; 16%) cefotaxime (n = 19; 7%), cefoxitin (n = 12; 5%), and ciprofloxacin (n = 35; 14%). As expected, a high correlation was observed among resistance to antimicrobials of the same class, e.g., nalidixic acid and ciprofloxacin. In some instances, correlations were also observed among resistances to antimicrobial agents of different classes, in particular,, for ampicillin, streptomycin and sulfonamide. Screening with cefotaxime and ciprofloxacin selected for E. coli isolates with overall higher levels of resistance to all antimicrobials tested than screening with older antimicrobial agents (Table (Table33).
Twenty-three (22) extended-spectrum beta-lactamase (ESBL)-producing E. coli isolates were identified from the 254 total isolates studied (Table (Table4).4). These isolates were obtained from hospital A effluent samples (n = 5/106) and municipal effluent samples upstream (n = 1/36) and downstream (n = 1/36) from hospital A. From the 47 E. coli isolates obtained from treated effluent samples, one ESBL-positive isolate was obtained from the intake effluent of a treatment plant, seven were obtained from secondary-treated effluent (2006 samples), five were obtained from the secondary-treated effluent (2007 samples), one was obtained from primary-treated effluent (2008 sample), and two were obtained from secondary-treated effluent (2008 sample). The number of ESBL-producing E. coli isolates obtained from treated effluent (n = 15; 6%) was higher than those obtained from nontreated effluent (n = 8; 3%). However, as the total number of ESBL-producing isolates was not enumerated, and the isolates for characterization were selected in an unstructured way, statistical analysis to assess the apparent difference was not carried out.
The blaCTX-M gene was present in 22 of 23 ESBL-producing isolates, with all belonging to group 1 (n = 19; 86%) or group 9 (n = 3; 14%) blaCTX-M. Sequencing of blaCTX-M genes from samples of treated effluent discharged into the environment identified group 1 genes as blaCTX-M-28 (n = 1), blaCTX-M-3 (n = 2), blaCTX-M-61 (n = 3), and blaCTX-M-15 (n = 2) and the group 9 gene as blaCTX-M-14 (n = 3). The blaTEM gene was detected in 11 isolates in the following samples: hospital A effluent (n = 1), raw intake effluent (n = 1), primary-treated effluent (n = 1), and secondary-treated effluent (n = 8). The blaSHV gene was detected in one hospital A effluent sample isolate (Table (Table44).
PFGE analysis showed considerable diversity of genotypes among the isolates examined (Fig. (Fig.1).1). Indistinguishable pulsed-field profiles were obtained for isolates 0108 and 0121, obtained from final treated effluent samples on different dates; however, the isolates differed in antimicrobial resistance pattern (Table (Table4).4). Both isolates harbored the blaTEM and blaCTX-M-14 genes.
PFGE-indistinguishable isolates 0895 and 0900, both obtained from secondary-treated effluent samples collected at different times on the same day in 2007, were resistant to ampicillin, streptomycin, sulfonamide, trimethoprim, nalidixic acid, cefpodoxime, and cefotaxime. Isolate s-0895 was also resistant to ciprofloxacin, with a MIC of >32 μg/ml, while isolate s-0900 was intermediate to ciprofloxacin, with a MIC of 2 μg/ml. Both isolates harbored blaCTX-M-61 genes. A third isolate, 0891 (from the same sample as 0895), also obtained from treated effluent samples, was very similar (97.5%) to isolates 0895 and 0900 by PFGE but with additional resistance to minocycline. All isolates were negative for qnrA, qnrB, and qnrS.
Isolates 0982 and 1016 were obtained from samples of municipal effluent near hospital A (one upstream and one downstream). These isolates showed 97.5% similarity of PFGE pattern by BioNumerics analysis.
A comparison of pulsed-field profiles from effluent isolates and ESBL-producing E. coli clinical isolates (as previously described ) was performed. All clinical isolates were obtained from samples received in the same hospital from which effluent samples were collected. Overall, the resulting dendrogram shows that most clinical isolates cluster together with more than 75% similarity, with only a small number of effluent isolates within this group. Most effluent isolates are relatively distinct from clinical isolates and from each other, although a small number of clinical isolates are grouped among the predominantly environmental clusters (Fig. (Fig.1).1). The largest cluster of isolates (n = 27) all harboring group 1 blaCTX-M genes comprises 25 clinical isolates and 2 isolates from hospital A effluent samples (0586 and 0977). By PFGE, effluent isolate 0586 was 97.5% similar to clinical isolate c-838 obtained from a urine sample. Isolate 0977 showed 97.5% similarity to four clinical isolates (c-918, c-900, c-690, and c-901) obtained from urine samples. It is of note that the two environmental isolates (0586 and 0977) most similar to clinical isolates were directly from samples of hospital effluent discharge rather than those taken from other sites. However, there are limitations to the typing methods applied, and conclusions regarding relationships between isolates are tentative.
The MPN method is an established approach to enumeration of E. coli bacteria in water. Colilert products are widely used for this purpose because they are simple to use and interpret. We have demonstrated that the Colilert system can be readily adapted for enumeration of E. coli bacteria resistant to specific antimicrobial agents in aqueous samples. As a result, the proportion of antimicrobial-resistant E. coli bacteria in an aqueous environmental sample can be calculated. Previous methods have generally involved studying selected isolates obtained from filtration of aqueous samples. The method described here can be used for detection of extended-spectrum beta-lactamase (ESBL)-producing organisms, in addition to other specific resistance phenotypes. The method can also be applied to enumeration of E. coli bacteria with specific multiple antimicrobial resistance phenotypes by addition of multiple antimicrobial agents (data not shown).
Antimicrobial-resistant bacteria may be discharged into the environment from human sources (hospital and municipal effluent) and agricultural sources (15, 16). There is considerable potential for dissemination of antimicrobial-resistant organisms and resistant determinants from such sources through contamination of food and water (5, 11). The risk may be greatest when contaminated wastewater is discharged directly into the environment. However, secondary wastewater treatment does not remove all fecal organisms. The Environmental Protection Agency (EPA) in Ireland estimates that approximately 70% of wastewater in Ireland is subject to secondary treatment (19). Outflow from secondary wastewater treatment facilities has much reduced E. coli levels compared with those of untreated effluent; however, there are limited data to indicate the extent to which antimicrobial-resistant E. coli bacteria are removed through the treatment process and the levels present in the discharge. Our data indicate that a high proportion of E. coli discharged from secondary wastewater treatment facilities is resistant to ampicillin and that isolates resistant to newer agents such as extended-spectrum cephalosporins (including the ESBL phenotype) and fluoroquinolones are also present. It appears that antimicrobial-resistant E. coli is not at any significant survival disadvantage in the environment.
This is the first report confirming that ESBL-producing E. coli survives the wastewater treatment process of a modern secondary treatment facility. ESBL-producing Escherichia coli has emerged as a significant human health issue in hospitals and communities in the past decade. It is a significant local and global health concern (9, 17, 25, 29). We have previously documented an outbreak of ESBL-producing E. coli in a nursing home in this region, and infection with ESBL-producing E. coli is now a common clinical problem (17, 20). The ESBL variants (CTX-M groups 1 and 9) detected in the outflow of the wastewater treatment plant generally correspond to the most common variant/variants detected in association with clinical infection in our population (17). Pulsed-field gel electrophoresis shows considerable heterogeneity among the environmental ESBL-producing E. coli bacteria isolated, and they are frequently quite different from isolates associated with human infection by PFGE. However, three closely related strains harboring blaCTX-M-61 (isolates 891, 895, and 900) were isolated from two effluent samples taken on the same day (15 August 2007). This CTX-M variant has not previously been reported in Ireland (17). The diversity in CTX-M variants detected in the treated effluent samples reflects the spread of these important genotypes in the environment.
The comparison between clinical and environmental ESBL-producing E. coli shows that environmental isolates similar to clinical isolates were detected in direct hospital effluent discharge samples on two occasions. One might speculate that PFGE diversity of ESBL-producing E. coli in the wider environment may be related to plasmid transfer between E. coli strains in the environment and or because those isolates associated with clinical infection are a virulent subset of ESBL-producing E. coli associated with gastrointestinal colonization.
The extent to which discharge of antimicrobial-resistant E. coli and other bacteria into the environment contributes to the dissemination of antimicrobial resistance is uncertain. To date, studies of antimicrobial resistance in bacteria in the environment have tended to address the issue in terms of the presence or absence of antimicrobial-resistant bacteria (1, 6, 21). Quantitative data are likely to be important in efforts to assess the potential risk and to assess the impact of specific elements of the effluent treatment process in removing antimicrobial-resistant E. coli, as quantitative microbial risk assessment is increasingly being used. The method we report is a convenient method for enumeration of antimicrobial-resistant E. coli bacteria in aqueous samples and will be valuable in facilitating more detailed studies of the potential impact of municipal and agricultural effluent on antimicrobial resistant E. coli in the environment.
This research is funded by the Environmental Protection Agency (EPA) of Ireland, through the national development plan, as part of the Enhancing Human Health Through Improved Water Quality (EHHTIWQ) project (http://www.nuigalway.ie/ehh/).
Published ahead of print on 4 June 2010.