Two WWTW located in Cape Town, SA were selected for this study (Supplementary Figure 1). Both WWTW made use of conventional activated sludge treatment; however, after screening and grit removal, WWTW1 made use of primary sedimentation and a mixed flow bioreactor setup, followed by secondary sedimentation and chlorination, and finally discharged into maturation ponds. In contrast, WWTW2 made use of plug-flow treatment followed by the discharge of clarified water into a maturation pond prior to chlorination. The two WWTW serve approximately 248,000 people and 1,102,000 people respectively as calculated by Archer et al. (2021).

Wastewater grab samples were collected in 100 ml sterile Schott bottles from the influent (post grit screens), return activated sludge (RAS), and treated effluent (post-chlorination in WWTW1; post-chlorination and post-maturation pond in WWTW2). All samples were stored on ice during transport to the laboratory and stored at 4°C until processed, within 2 h of arriving at the laboratory.

Whole community and pure culture MICs were determined to identify the effect of antibiotic exposure in sub-MIC concentrations on the copy number of ARG present in isolates obtained from WWTW.

Data obtained from culturing was analyzed in GraphPad Prism (v. 6) using a multiple t-test to determine significance between the two WWTW in influent, RAS, and effluent samples. Student's t-tests for independent means were used to determine significant differences in resistance gene copy numbers between WWTW and in single colonies before and after exposure to sub-MIC concentrations of antibiotics. Significance was determined at a confidence interval of 95% (p < 0.05).

Multiple authors state that WWTW are hotspots for the transmission of ARG and ARB (Berendonk et al., 2015; Guo et al., 2017; Dafale et al., 2020). This study shows that while these environments do indeed have waters that are contaminated with ARG and ARB, all tested genes, except bla_KPC in WWTW2, were less prominent in the effluent than in the influent (Figure 1). The increase in K. pneumoniae carbapenemase (KPC) is a concern as it confers resistance to carbapenems, which, along with the last resort antibiotics, is of greater concern than AMX (Feil, 2016). However, all other ARG were removed by the WWTW processes. As total DNA was extracted from the WWTW samples, the ARGs detected in this part of the study are not necessarily only from bacterial origin but also potentially from other sources including fungi, algae, protozoa, and viruses as suggested by the DNA extraction kit manufacturers (ZymoResearch, n.d.). Similarly, Figures 2, 3 mostly show no net increase in ARB between the influent and effluent of the WWTW and while not statistically significant, there were notable decreases in ARB abundance. Similar results have been found in other studies (Duong et al., 2008; Lupan et al., 2017). This does not suggest that the ARG and ARB released into the environment from the WWTW effluent do not play a role in the transmission of AMR downstream but demonstrates that the WWTW included in this study led to reduced numbers of ARG and ARB that entered the environment. In this case, without wastewater treatment, ARG and ARB would be more prominent in environmental waters.

The antibiotic concentrations used for ARB selection in this study were selected based on resistance breakpoints (μg/ml range) and, as a result, were much higher than concentrations observed in WWTW. Antibiotic levels in WWTW vary depending on the location, type, and size of the facility, as well as operational efficiency. Antibiotic concentrations in local WWTW range from 1,344.8 ng/L−3,612 ng/L for SMX in WWTW effluent (Suzuki et al., 2015; Archer et al., 2017); 1,400 ng/L for AMX in influent (Watkinson et al., 2009); and 0.019–7.6 ng/L for GM (Löffler and Ternes, 2003; Tahrani et al., 2016), while, to our knowledge, no quantification of CST in WWTW has been published. The concentration of a particular antibiotic that does not select for antibiotic resistance in bacteria is termed the predicted no-effect concentration (PNEC). Some PNECs have been identified by Bengtsson-Palme and Larsson (2016) with respective values for AMX, SMX, GM, and CST of 250, 16,000, 1,000, and 2,000 ng/L. Therefore, it is reasonable to question whether concentrations of antibiotics in WWTW would need to be higher than these PNEC values to drive bacteria to develop resistance mechanisms and select for resistant communities in WWTW – a point that requires further analyses. Figures 2C,D shows that the positive controls (without antibiotics) also had increased abundance of bacteria after enrichment in antibiotic-free media, similar to trends observed for abundance of ARB enriched in antibiotic media. This could be expanded to a WWTW setting where the highly labile nutrient source present is the driving factor for bacterial proliferation that selects for organisms with higher growth rates and opportunists, which multiply rapidly and produce many progenies that are less adapted for their particular environment opposed to extensive proliferation of organisms with ARG when exposed to sub-inhibitory concentrations of antibiotics (Vadstein et al., 2018). A suppressive pressure on whole-community growth rate (OD₆₀₀) for AMX, SMX, and GM antibiotic enrichments in comparison to the POS MHB control was observed (Figure 2). In contrast, the OD₆₀₀ for POS MAC in Figure 2C indicated that exposure to CST resulted in a higher growth rate in influent and RAS microbial communities from WWTW1 compared to the control. This may indicate a highly adaptive community, with lower competitive pressure in these matrices. However, wavelength interference of the acidity indicator in MacConkey media may also play a role in the observed CST results. The complex interactions in wastewater microbial metabolism could influence wavelength transmission in all media types. Thus, culturing was selected as a quantification metric for subsequent analyses.

The percentage of ARGs removed from influent to effluent was significantly higher in WWTW2 for mcr3, sul1, and sul2 (Figure 1). It was surprising that no bla_OXA−48 genes were identified despite successful amplification of the lab-synthesized bla_OXA−48 control gene, as it has been previously identified in various countries, in multiple plasmid types, and in SA environments (Bonomo, 2017; Ebomah and Okoh, 2020; Gibbon et al., 2021). Differences in capacity, sources of influent, and type of effluent treatment strategies between the two WWTW may contribute to the variation in ARB and ARG abundance and the dominant species observed between WWTW. The first WWTW uses a mixed flow approach for tertiary treatment, has a 200 Ml/day capacity, and was sampled upstream of the maturation pond. Due to the layout of this WWTW, a sample point downstream of the maturation pond was inaccessible. In contrast, WWTW2 uses a plug flow system for tertiary treatment, has a 38 Ml/day capacity, and was sampled downstream of the maturation pond. In a mixed flow system, contaminants such as antibiotics, ARG, and ARB, are immediately distributed throughout the system. This reduces the concentrations of these contaminants and promotes persistence and proliferation. In contrast, in a plug flow there is less mixing, and contaminants in treated water are thus “pushed out” of WWTW within its hydraulic retention time as new contaminants enter (Rumbaugh, 2014). Due to the fact that samples were collected after the maturation pond in WWTW2, aggregates of suspended solids may have settled in the maturation ponds, and may not be present in the collected sample, further reducing the ARG and ARB seen in the effluent of WWTW2 (Figures 1, 3). This study did not sample other parts of wastewater treatment such as solid waste prior to grit screening or waste sludge. Thus, ARG and ARB may still enter the environment including rivers, the ocean, and sediments through different avenues despite seeming reduced in liquid wastewater effluent, as suggested by Sabri et al. (2020). Through this dissemination, other organisms such as bacteriophages and other vectors could acquire these genes, increasing the spread of resistance (Czekalski et al., 2012; Fouz et al., 2020; Wang et al., 2021; Zare et al., 2021). However, the amount of ARB in sludge has also been found to reduce significantly after anaerobic digestion due to the heat generated during the process that kills bacteria (Zhao and Liu, 2019).

The most prevalent species identified from a randomly isolated footprint of colonies, M. morganii (Figure 4), is intrinsically (innately) resistant to AMX and CST (Liu et al., 2016), both of which were used as a selective pressure in this study and could explain why this organism was the most prominent in the samples. Several other identified genera (Aeromonas, Escherichia, Citrobacter, and Providencia, etc.) in Figure 4 are commensal bacteria in humans and/or animals and are frequently found in WWTW, regardless of the treatment strategy (Lee et al., 2009; Janda and Abbott, 2010; Abbott, 2011; Liu et al., 2016, 2017; Bassetti et al., 2018). This is evident in Figure 5 where these commensal genera are present in both WWTW1 and WWTW2. These organisms can be present in different environments and are not exclusive to the human gut, however, WWTW are receivers of human (and other) waste, and the fact that antibiotic resistant commensal organisms are dominant in these environments suggests that humans and animals might be a source of entry of ARB into WWTW and subsequently play a major role in the emergence of AMR in the environment. This is further supported by the increased abundance of ARB in WWTW samples after antibiotic exposure (Figures 2, 3C,D). Higher concentrations of antibiotics used in treating a patient or animal in a clinical or agricultural setting (or using them as supplements in animal feed) applies a higher selection pressure within the microbiome which has been shown to increase antibiotic resistance at a faster rate than lower antibiotic concentrations (Oz et al., 2014; Dafale et al., 2020). As a result, selection of intrinsically resistant pathogens and resistant microbial populations are more likely to occur in these microbial communities which may be excreted and enter WWTW in addition to surface waters. These organisms that persist in the WWTW effluent may have detrimental implications downstream of the WWTW. For example, the sprinkler systems used for irrigation have been found to play a role in the spread of aerosolized pathogens (Adegoke et al., 2018) with several bacterial and viral outbreaks being reported due to crops being contaminated after irrigation with treated wastewater (Figueras Salvat and Ashbolt, 2019). These incidences reinforce the potential of antibiotic resistant organisms to disseminate into the environment and through the food chain if they are present in WWTW discharge.

The mode MICs for pure cultures for CST, AMX, SMX, and GM (2,048, 16,384, 4,096, and 16 μg/ml, respectively) (Figure 7) are 256, 2,048, 8 times, and equal to the respective EUCAST MIC resistant breakpoint values, indicating that the microbes in these WWTW, as a collective, are resistant to all the antibiotics tested in this study (European Committee on Antimicrobial Susceptibility Testing, 2017; Clinical Laboratory Standards Institute., 2018). High MICs for bacteria isolated from wastewater have also been reported in numerous studies (Ovejero et al., 2017; Obayiuwana et al., 2018). Mode MICs for individual isolates (Figure 7) were found to be higher than the MICs for the whole community wastewater samples for AMX and CST (Figure 6). MICs for SMX were similar between whole community and pure cultures, and MICs for GM were higher in whole communities. Increased prevalence of β-lactamase-producing bacteria has led to AMX having poor activity against Enterobacteriaceae and Pseudomonas spp. when used without a β-lactamase inhibitor (European Committee on Antimicrobial Susceptibility Testing., 2010; González-Bello et al., 2020). However, referring to the previous example of resistance emergence occurring in the gut microbiome of animals and humans, individual members or a sub-group of these microbial communities that enter the environment through excretion may develop higher MICs after being exposed to higher concentrations in the gut, compared to organisms in the environment exposed to trace antibiotic concentrations that are lower than expected PNEC values (Rolain, 2013; Bengtsson-Palme and Larsson, 2016; Booth et al., 2020) that could be diluted further by rainfall and stormwater before entering WWTW. The lack of a significant change in resistant gene copy number in various isolates from WWTW, after sub-inhibitory antibiotic exposure for 18–24 h, emphasizes this as well; however, it needs to be acknowledged that these results are indicative of the ARG copy numbers in bacterial isolates, and the presence of some ARGs in bacteriophages as a result of HGT are not represented here (Figure 8). It has been shown previously that exposure of E. coli isolates to sub-MIC concentrations of various antibiotics result in increased MICs within 21 days of exposure through serial passage and that organisms exposed to higher concentrations (0.5 × MIC) had a higher occurrence of cross-resistance compared to isolates exposed to lower concentrations (1/8 × MIC). As a result, whole communities that dominate regions of WWTW for biological treatment may have varying proportions of species intrinsically resistant to certain antibiotics. The abundance of these species may be lower and as a result the MICs may appear to be lower than single isolates that have the necessary resistance mechanisms. This may explain the trends seen for the whole community and pure culture MICs in Figures 6, 7. Thus, curbing AMR in the environment cannot be addressed by confronting WWTW alone and needs to be stewarded carefully from the point of usage. The wide range of antibiotic reactivities, resistance genes, and mechanisms that would respond differently to various treatments lead to a range of AMR issues. Potential outbreaks in humans or animals with organisms identified in this study would be untreatable with the target antibiotics due to the MIC values seen in Figure 7.

While the problem areas within the wastewater treatment process in terms of AMR should be identified, and alternative technologies that promote more effective removal of all harmful contaminants should be implemented, it is important to recall that WWTW are the receiving bodies rather than the original source of antibiotics and ARB that carry ARG. In addition to the entry of ARB, other industrial chemicals may contribute to AMR. Chlorine treatment has been found to increase the permeability of bacterial cells, allowing increased horizontal gene transfer (HGT) of ARGs between donor and recipient cells (Guo et al., 2015; Wu et al., 2022). The entry of chlorine and other disinfectants used for cleaning hospitals, abattoirs, swimming pools, etc., into WWTW may increase HGT in WWTW with a dense microbial load, such as activated sludge, or promote gene transfer at the original source of contamination. In addition, extracellular ARGs have been found to adsorb to microplastics which are ubiquitous environmental contaminants and would play a role in AMR dissemination in the environment outside of a WWTW (Ross et al., 2021; Cheng et al., 2022).