Wastewater treatment plants (WWTPs) are an ideal place to monitor and investigate the development of antibiotic resistance and its dissemination into the environment and thus the biotic ecosystem (Hendriksen et al., 2019). Wastewater treatment plants are also recognized as “hotspots” for spreading antibiotic resistance traits among bacteria through horizontal gene transfer and selective pressure by the residual antibiotics (Kim et al., 2018).

Antibiotic-resistant bacteria are never completely eliminated from sewage via treatment processes at the WWTPs. Thus, it will be released into both the aquatic and terrestrial surrounding environments. In the end, through the food chain, ARBs will enter the human system (Kumar and Pal, 2018). Of greater concern is that the emergence of antibiotic resistance genes (ARGs) and ARBs gradually reduces the therapeutic potential of antibiotics, since a large quantity of antibiotics are released into the environment (Fair and Tor, 2014).

As a Southeast Asian developing country, Bangladesh poses a regional and global threat with a high prevalence of ARBs and ARGs (Ahmed et al., 2019). Several studies have shown that indiscriminate and irrational use of antibiotics in health care, farming and agriculture leads to therapeutic failure in this country (Chowdhury et al., 2021; Hosain et al., 2021). As a consequence, multidrug-resistant (MDR) pathogenic bacteria will continue to emerge in the future, posing a potential risk for combating infectious diseases in developing countries like Bangladesh (Rabbani et al., 2017). Unfortunately, no national antimicrobial resistance (AMR) data are currently available to support this finding, according to the World Health Organization (WHO, 2014). Currently, treatment failure due to drug-resistant infections results in 0.7 million deaths per year, worldwide, and it is anticipated that almost 10 million drug-resistance related deaths will occur globally per year in 2050 (Dadgostar, 2019).

Multidrug-resistant Escherichia coli has been categorized as a priority pathogen by WHO due to emerging AMR (Vaiyapuri et al., 2021). In addition, E. coli is the biggest culprit in enteric infections, with a total of 50,800 children dying from diarrhea in Bangladesh annually (Talukdar et al., 2013).

The data on the prevalence of MDR E. coli and their biofilm-forming abilities in urban WWTP effluents, especially in Dhaka, Bangladesh, is scanty. Therefore, this study is an effort to illustrate the background of the AMR patterns and biofilm formation capacity of E. coli recovered from WWTP effluents in Dhaka, Bangladesh. Escherichia coli, as fecal indicator bacteria, were screened for antibiotic resistance and the plausible public health concern linked to the transfer of ARGs including integrons and plasmids, since WWTPs are known to be an early warning system.

In this study, we characterize E. coli according to their antibiotic resistance profile, plasmid profile, and biofilm formation capacity using various phenotypic methods and gene detection via PCR techniques.

Sample collection

Every month from June 2017 to May 2018 (except January and April 2018), 10 samplings of two-liter representative wastewater grab samples were retrieved from Dhaka City’s WWTP (Pagla WWTP) (N23°41.004"/E90°27.080" with 7 m elevation). The wastewater samples were collected from the main sewage flow after the first filtering step before the WWTP inlets, without effluent processing. After collection, each sample was transferred to the lab within 1 h, stored at 4°C and processed the sample within the day (Hossain et al., 2022).

Isolation and identification of E. coli strains: each sample was studied to detect thermotolerant E. coli by streaking 10 μl of wastewater onto m-TEC (Oxoid) agar plates and incubating them overnight at 45°C. The presumptive colonies were selected for further identification using colony morphology and characteristics. A total of 37 presumptive E. coli isolates were identified on the basis of species-specific uidA genes (Supplementary data Table S1).

Antimicrobial susceptibility pattern for different antibiotic groups: thirty-seven selected E. coli isolates were tested for antimicrobial susceptibility by disk diffusion using the standardized Kirby Bauer method (Bauer et al., 1966) and commercially available disks (Oxoid). The isolates were tested against the following 15 antimicrobials which covered eight different classes (penicillin [β-lactam], cephalosporin [extended spectrum β-lactam], quinolone, tetracyclines, folate pathway inhibitor, aminoglycoside, chloramphenicol, macrolides): ampicillin (AMP/10 μg), amoxicillin (AML/10 μg), nalidixic acid (NA/30 μg), ciprofloxacin (CIP/5 μg), tetracycline (TE/30 μg), trimethoprim/sulfamethoxazole (SXT/25 μg), gentamicin (CN/l0 μg), kanamycin (K/30 μg), streptomycin (S/10 μg), chloramphenicol (C/30 μg), erythromycin (E/15 μg), ceftriaxone (CRO/30 μg), cefixime (CFM/5 μg), cefotaxime (CTX/30 μg), and cefuroxime (CXM/30 μg).

Extended spectrum β-lactamase production detection method: double disk synergy tests (DDSTs) were performed using ceftazidime (CAZ/30 μg), cefotaxime (CTX/30 μg), ceftriaxone (CRO/30 μg) and amoxicillin/clavulanic acid (AUG/30 μg) disks to detect the ability of 37 E. coli isolates to produce ESBL (Jarlier et al., 1988).

Detection of antibiotic resistance gene(s) by the molecular method: genomic DNA was extracted from 37 isolates using the boiled template method (De Medici et al., 2003). These isolates were characterized through the PCR method using a thermal cycler (MJ Research PTC-200). Primer names and sequences, resistance genes, primer position, annealing temperature, amplicon sizes, and references are listed in Supplementary data Table S1.

Plasmid profiling

Isolation of plasmid DNA and plasmid curing: for plasmid extraction, the GeneJET Plasmid Miniprep Kit (Thermo Fisher Scientific Inc.) was used for the 37 E. coli isolates. Following extraction, these plasmid harboring isolates were subjected to curing by using 10% SDS (sodium dodecyl sulphate) adopted the method developed by Zaman et al. (2010).

Antimicrobial susceptibility of post-plasmid curing isolates: the Kirby Bauer disk diffusion method was used for antibiotic susceptibility of post-plasmid curated isolates. The pre- and post-plasmid curing antimicrobial susceptibility pattern was tested against 15 antibiotics from eight different groups, as listed above. The resulting zones of inhibition were compared to the initial zones of inhibition for the plasmid-harboring E. coli.

Statistical analysis

For statistical analysis, IBM SPSS Statistics version 22 software was used, with prime data divided into two categories: pre- and post-plasmid curing. Then, essential recoding was done accordingly. The Chi-square test was used to determine the statistical significance of plasmids in antibiotic resistance.

Determination of biofilm formation ability

Detection of cellulose and curli fimbriae production: cellulose and curli fimbriae production was detected in 37 E. coli isolates cultured on Luria-Bertani (LB) agar plates supplemented with 40 μg/ml Congo red and 20 μg/ml Brilliant Blue R250 without salt, and incubated for 24 h at 37°C and 48 h at 28°C. After incubation, the morphotypes were determined and classified. Three independent tests of each isolate for cellulose and curli production were performed (Römling et al., 2000).

Quantification of biofilm by crystal violet staining assay: the method developed by Christensen et al. (1985) was used for staining. The extent of biofilm formation was interpreted based on the method established by Stepanović et al. (2000), and was classified into three categories: 1) non-biofilm producer: OD ≤ ODc, 2) weak biofilm producer: ODc < OD ≤ 2x ODc, and 3) strong biofilm producer: 4x ODc < OD, based on the mean value of the measured OD595nm.

Biofilm formation in the liquid-air interface: biofilm formation in liquid (i.e., pellicle formation at the liquid-air interface) was observed by using 4.5 ml of LB inoculated with 0.5 ml of freshly grown culture and incubated at 200 rpm and 37°C for seven days.

Detection of biofilm-associated genes: biofilm-associated genes including csgA, papC, and fimH were detected by PCR techniques, as described above.

Multidrug resistance profile of E. coli

All 37 strains of E. coli were tested against 15 antibiotic agents from 8 different antibiotic classes. Results showed that the highest resistance was to erythromycin, at 99% (n = 36), followed by 91.9% (n = 34) to cefotaxime and amoxicillin, 89.2% (n = 33) to streptomycin, 75.7% (n = 28) to ampicillin, 67.6% (n = 25) to cefuroxime, 64.9% (n = 24) to ceftriaxone, 62.2% (n = 23) to cefixime, 59.5% (n = 22) to kanamycin, 54.1% (n = 20) to ciprofloxacin, 51.4% (n = 19) to nalidixic acid and tetracycline, 40.5% (n = 15) to trimethoprim/sulfamethoxazole, 29.7% (n = 11) to gentamicin, and 21.6% (n = 8) to chloramphenicol (Fig. 1).

Fig. 1. Antibiotic resistance profile of E. coli.

Detection of antibiotic resistance markers (ARMs) and ESBL production

Thirty-six isolates of MDR E. coli were found (only one isolate resistant to a single antibiotic was found), among which ESBL producers were identified by phenotypic methods. Therefore, certain ARG markers were detected to correlate the phenotypic data with the genotypic inheritance of ARMs (Supplementary data Table S2).

Through the phenotypic DDST method, 13 of the 37 isolates were found to be ESBL-positive. The presence of ESBL genes was detected using the gene-specific PCR method in those positive isolates. The PCR assay revealed that all but one of the ESBL-positive isolates contained one or more ESBL producing genes (blaTEM and blaCTX-M-15); however, the blaSHV gene was not found. The results showed that 46% (n = 6) of the ESBL-positive isolates contained the blaTEM gene, whereas the blaCTX-M-15 gene was present in 70% (n = 9) of the isolates. Of 37 isolates analyzed, 29% (n = 11) were positive for blaTEM and 24% (n = 9) were positive for blaCTX-M-15 genes.

Of the E. coli strains which showed resistance to quinolone antibiotics, two were positive for the qnrB gene and 12 were positive for the qnrS gene. Results for trimethoprim/sulfamethoxazole (SXT)-specific resistance genes showed that one and eight isolates were positive for the dfrA1 and dfrA17 genes, respectively, and two and three isolates were positive for the sul1and sul2 genes, respectively.

Among the 37 isolates, the most prevalent gene detected was qnrS (32.43%), followed by blaTEM (29.73%), blaCTX-M-15 (24.32%), dfrA17 (21.62%), sul2 (8.10%), qnrB (5.41%), sul1 (5.41%), dfrA1 (2.70%) and blaSHV (0.00%) (Fig. 2). Among 37 isolates, 13 (35.1%) were ESBL positive by DDST.

Fig. 2. Resistance gene(s) detection in E. coli.

Detection of integron classes

The PCR results showed that, out of 37 isolates, 12 harbored Class 1 integron resistance genes but none of them carried Class 2 and 3 integron resistance genes. Among the twelve class 1 integron positive E. coli, seven were ESBL-positive isolates.

Plasmid profile analysis

This analysis revealed that 13 of the isolates (35%) carried plasmids with less than 10 kb in size. Plasmid profile analysis showed that E. coli isolates harbored different size plasmids ranging from > 1 kb to 8 kb. Of these, two isolates (N6EC1 and N7EC2) showed double bands and the rest showed a single band.

Multidrug resistance profile of the plasmid-containing E. coli isolates

Thirteen plasmid-carrying E. coli isolates were found to be resistant to most of the antibiotics tested. All 13 isolates were completely resistant to AMP, AML, E, and third generation cephalosporins: CTX.

Antibiotic susceptibility of the cured E. coli isolates

A significant reduction in resistance, from 71.2% to 44.1%, was also found (from 139 resistant isolates before curing to 86 after curing). For AMP, AML, and E, all isolates remained resistant after plasmid curing. Furthermore, the isolates which were initially sensitive remained sensitive (Fig. 3).

Fig. 3. Antibiotic susceptibility of E. coli before and after plasmid curing.

A contrasting scenario was encountered for isolate N4EC4, which was initially moderately resistant and sensitive to NA and C, respectively, but became resistant after plasmid curing. Plasmid bands were found in isolates N6EC10 and N9EC1 after curing; however, each of them lost resistance to four different antibiotics (Supplementary data Table S3).

Statistical analysis

Data analysis showed 71.4% (139 isolates) resistant (R) and 28.7% (56 isolates) intermediate/sensitive (I/S) isolates prior to plasmid curing. However, after curing, the scenario was changed to 44.1% (86 isolates) R and 55.9% (109 isolates) I/S. Again, among the Rs (139 isolates) in the previous phase, we found that 60.4% (84 isolates) remained Rs and 38.1% (53 isolates) changed to I/Ss in the post phase. But among I/Ss (56 isolates) in the previous phase, 3.5% (2 isolates) shifted to Rs and 96.5% (54 isolates) remained as I/Ss. This figure seems to suggest a positive change with the plasmid curing treatment. The Chi-square test result was statistically significant, as the p-value found (0.000) was less than 0.05 (5% level of significance).

Determination of biofilm formation capability

The brown, rough and dry characteristic morphotype of curli production was found in all 37 isolates. Biofilm was visualized in two isolates following incubation at 37°C for five consecutive days, as a ring of cells adhered to the glass wall at the air-liquid interface. The mean OD595 nm value was found to be 0.00 for the sterile LB medium (negative control) and the cut-off value (ODc) was found to be 0.113. Twenty-seven E. coli isolates were found to be non-biofilm forming (OD595 ranged from 0.046 to 0. 113). Nine (24.32%) were found to be weak biofilm producers (0.113 < OD595 nm ≤ 0.226) and one was a strong biofilm producer (OD 595 > 0.452). The results showed that the papC gene was absent in all the isolates, fimH was present in 72.9% (27 out of 37), and csgA was present in 64.9% (24 out of 37) (Supplementary data Table S4).

The purpose of this study is to identify multidrug resistant E. coli, which WHO has listed as a priority pathogen (Vaiyapuri et al., 2021), and characterize its antimicrobial resistance gene(s) as well as their determinants and biofilm formation, from samples obtained at the WWTP in Dhaka, Bangladesh.

The unrestrained use of antibiotics in health care sectors and animal husbandry creates high selective pressure in favor of developing antibiotic resistance in countries like Bangladesh (Faiz and Basher, 2011). This study provides evidence to support this view, since more than 97.3% (n = 36) of the E. coli isolates are MDR. A comparable finding reports the presence of high levels of MDR E. coli in hospital and domestic WWTPs in South India (Praveenkumarreddy et al., 2020).

Clinically significant Class A β-lactamase producing blaCTX-M-15 positive E. coli are often found in environmental sources (Peirano and Pitout, 2019). Similarly, the presence of the blaTEM gene (30%) and blaCTX-M-15 gene (24%) is identified in those isolates. Qi et al. (2010) found CTX-M-15 genes (64%) in their isolates, consistent with our study showing more CTX-M-15 genes (70%) in ESBL-positive isolates than TEM and SHV genes. Several studies have observed the co-resistance of ESBL-positive E. coli to various antibiotics including non-β-lactam antibiotics (Balkhed et al., 2013; Tacão et al., 2014). This is relevant to our study, as 12 qnrS positive isolates are found, among which seven harbor the ESBL gene.

In this study, almost half of the isolates are resistant to first- and second-generation quinolones, similar to findings from sewage samples reported in Spain (Colomer-Lluch et al., 2013). Our prevalence of qnrB and qnrS genes, indicating quinolone resistance, are similar to the findings from hospital wastewater in Tehran, Iran (Ranjbar and Farahani, 2017). However, half of the qnrS positive isolates has possessed plasmids.

In our study, a higher prevalence of dfrA17 gene is found in SXT resistance. However, some isolates have showed resistance against SXT without having the sul and dfr genes. The paradox might be explained by the fact that more than 30 different SXT resistance genes already exist, and may have conferred resistance in those isolates (Šeputienė et al., 2010).

In our study, Class I integrons is detected in 32.43% of isolates, proportionate to the percentage (31.5%) of an earlier study conducted by Wang et al. (2014) in China. Several previous studies have validated that resistance to a greater number of antibiotics is correlated with the presence of integrons (Grape et al., 2005). All integron 1-positive isolates are MDR E. coli; however, the majority of MDR isolates (25 of 37) are integron-negative. These findings might indicate that integrons represent only one of several factors influencing the generation of an MDR strain.

The ARGs are frequently located on plasmids, conferring resistance to a variety of different antimicrobials (Yang et al., 2020). In our study, plasmid curing therefore resulted in sensitivity to several antibiotics to which there was initial resistance. However, the N4EC4 isolate’s puzzling loss of sensitivity to NA and C after curing can be ascribed to a number of plausible factors. For one thing, the curing methods (Liu et al., 2012) or curing agents (Letchumanan et al., 2015) being used may cause mutations in host chromosomes or increased expression of efflux mechanism might reduce susceptibility to several antimicrobials (Buckner et al., 2018). However, a significant correlation between plasmid curing and antibiotic susceptibility is asserted through statistical analysis.

However, several biofilm-associated genes is found among the isolates. Altogether, 72.90% of the isolates possessed the fimH gene. The major subunit of curli fimbriae encoded by csgA (Uhlich et al., 2006), is present in 64.90% of the isolates. Thus, this study has demonstrated a high prevalence of biofilm-related genes in MDR E. coli isolates, posing a potential threat in the environment.

As best we know, this is the first report on the MDR E. coli isolates recovered from WWTP untreated effluents in Dhaka, Bangladesh. However, there are some limitations in our study. First, there was no exploration for ARB and ARGs in WWTP effluents following the treatment process. However, various studies have reported several beta-lactamase genes in hospital and city wastewater effluents following treatment, raising the potential for ARGs to be disseminated in aquatic environments (Adegoke et al., 2020), since genes may persist for a long time, even after the bacteria are dead (Fouz et al., 2020). Second, our study was limited to Bangladesh, and the study findings might not be applicable to global settings.

This study has revealed a high prevalence of MDR E. coli and high ARG diversity in WWTP untreated effluents in Dhaka, Bangladesh. These biological contaminants may play a pivotal role in an increased frequency of resistance transmission among both pathogenic and non-pathogenic bacteria. Therefore, if proper treatment is not implemented and monitored regularly, resistance might pose a public health threat in the treatment of infections, especially with the development of both regional and global pan-drug resistance.

This study was funded by the Government of the People’s Republic of Bangladesh, Ministry of Education, Bangladesh Bureau of Educational Information & Statistics (BANBEIS) Project ID: LS2018783.

The authors have no relevant financial or non-financial interests to disclose.