Presence of antibiotic resistance genes in the receiving environment of Iqaluit's wastewater treatment plant in water, sediment, and clams sampled from Frobisher Bay, Nunavut: a preliminary study in the Canadian Arctic

Iqaluit is the capital city of the Nunavut territory and is situated on Baffin Island, Canada (2016 population, 7740). The city is located within Koojesse Inlet, near the top of Frobisher Bay, which is macrotidal (12 m) with 17 km³ of seawater flushing the Bay during a single tide (Hsiao 1992). At the time of sampling, the wastewater system in Iqaluit was comprised of mechanical screening (Salsnes filter) prior to release, which was discharged continuously year-round into an open channel, i.e., directly into the marine environment (see Fig. 1). The associated lagoon was for overflow or when the plant was not functional and was released monthly (Neudorf et al. 2017). Approximately 7.2 × 10⁸ L of wastewater is released annually (Neudorf et al. 2017).

A total of 28 pharmaceuticals were sampled in this work using the organic diffusive gradients in thin film (o-DGT) passive sampler as reported previously (Stroski et al. 2020), including 17-estradiol, 17-ethynylestradiol, atenolol, atrazine, carbamazepine, clarithromycin, clofibric acid, diclofenac, enrofloxacin, erythromycin, estrone, fenoprofen, fluoxetine, gemfibrozil, ibuprofen, ketoprofen, metoprolol, naproxen, paroxetine, propranolol, roxithromycin, sulfamethazine, sulfamethoxazole, sulfapyridine, sulfisoxazole, sulfachloropyridazine, sulfadimethoxine, and trimethoprim. These PPCPs were monitored as they are commonly found in wastewater (Ying et al. 2009; Gagnon and Lajeunesse 2012; Challis et al. 2016), including in arctic regions (Stroski et al. 2020), and existing laboratory methods and capacity were in place. The assembly, extraction, and calculation of time-weighted average (TWA) water concentrations o-DGT are detailed elsewhere (Challis et al. 2016) and reported for Frobisher Bay, NU (Stroski et al. 2020). Laboratory and field blanks were extracted with each set of samples and had negligible levels of all analytes measured. Analyte concentrations were determined by liquid chromatography–tandem mass spectrometry (LC-MS/MS) using an Agilent 1200 Series LC pump and Agilent 6410B MS/MS (Agilent Technologies, Mississauga, ON) in electrospray ionization positive and negative mode. Limits of detections (LOD) and quantifications (LOQ) are found in the Supplementary material, while chromatographic and MS/MS method details are found elsewhere (Challis et al. 2016).

Samples were collected 8–10 August 2019 (Table 1 and Fig. 1). From the WWTP, we sampled the wastewater influent, the retention lagoon, and discharged effluent and sediment as it flowed along the tidal flats to Frobisher Bay. At varying distances, we sampled the water in Frobisher Bay; additionally, the clam tissues, sampled for another project during the same time period, were graciously provided to us (Schaefer et al. 2022; see Table 1). Additional water samples from Lake Geraldine and the Sylvia Grinnell River, representing pristine water sources, were also examined; however, their shoreline conditions were too rocky to collect any sediment. Finally, a soil sample near the WTTP, but not exposed to wastewater, was also collected as an additional control. All sampling was done with the support of the Amaruq Hunters and Trappers Association (Nunavut Scientific Research Licence No. 01 031 18 N-M) and under a license approved by Fisheries and Oceans Canada (Licence No. S-19/20–1040-NU).

Waters were aseptically collected in 1L-polypropylene bottles after a triple on-site rinse; sediments were obtained in 50 mL polypropylene centrifuge tubes; both were collected by hand. Cells from wastewaters (50–150 mL) and natural waters (900 mL) were filtered (Whatman 0.2 µm membrane filters). All samples were kept cool (<10 °C) during transport and frozen (−15 °C) during storage. DNA from sediment and soil were extracted directly.

Sediment, soil (0.5 g), and water filters (cells harvested from water) were homogenized by a FastPrep24 cell disruptor (MP Biomedicals; 6.0 speed 2 × 20 s) and extracted using DNeasy PowerSoil kit (Qiagen; Venlo, The Netherlands). The clam tissues (10–20 g) in 50 mL solution of 10 mM PBS (pH 7.4) and protease K were digested in a Seward Stomacher (Seward, Worthing, UK); the DNA was extracted by DNeasy Blood and Tissue kit (Qiagen). Both extraction methods followed the manufacturer's recommended protocols. The purity and quantity of extracted DNA were 1.7–2.0 (A260/A280) and 10–150 ng/μL, respectively, as determined by the micro-UV-spectrophotometer. The extraction kits were selected based on these purities and yields for each sample type.

Sediment and wastewater samples were diluted 1:20 with molecular-grade water to minimize co-eluted polymerase chain reaction (PCR) inhibitors; the DNA from the clams were diluted 1:200. This was determined by serially diluting the extractions (i.e., from soils and clams) pre-spiked with 10⁹ genes of E. coli 16S rRNA. The resultant PCR efficiencies and expected threshold abundances were compared against a “neat” standard curve.

To distinguish between ARGs from the wastewater and those that may naturally occur, extracted DNA (1 µg) from sediment exposed to wastewater effluent and environmental soil samples (nearby, not in contact with wastewater) were analyzed using the Open Array platform (Chinese Academy of Sciences—Xiamen; e.g., Zhu et al. (2013)). We selected 11 gene targets from the results (Table S1): sul1, ermB, acrA, qacH, dfrA, tet39, qnrA, cphA, ampC, bla_TEM, and aphA for qPCR (see below). Table S2 summarizes gene information. In addition, a 16S-rRNA gene target was used as a surrogate measure of “total bacteria,” to which all relative abundances were calculated.

Quantitative PCR was used to target 11 ARGs in the samples. Each 10 µL reaction consisted of 2 µL of diluted DNA template, 5 µL of ssoFAST-EvaGreen qPCR reagent (BioRad; Hercules, CA, USA), 0.2 µm primers and was made to volume with molecular-grade water. Reaction conditions included initial denaturation (94 ˚C, 3 min), 30–50 cycles of primer annealing (5 sec, 60 ˚C), and denaturation (94˚C, 3 sec); all were conducted on a BioRad iCycler5 (BioRad instrument) in triplicate. Blanks and standards were routinely run with samples. In addition, a post-analytical melt curve (Δ0.2 C/s) was run to verify amplification quality and specificity. Detections were valid when at least two replicates were within one cycle of each other without aberrant fluorescent signals.

Standards comprised of spiked DNA (10²–10⁸ copies/μL) in previously UV-irradiated sample matrix (i.e., extracts from sediment, marine water, or clam tissue); the 10 min exposure under UV sterilizing conditions prevents existing DNA from becoming PCR-able. Therefore, DNA standards were prepared from PCR amplicons, purified with a Qiagen PCR Purification Kit, and quantified by UV-micro-spectrophotometry; sequences were verified by Sanger Sequencing (GATC-Eurofins Genomics).

The abundances of genes were log-transformed prior to statistical analysis to improve sample distribution (normality). Absolute abundances were presented for the “total bacteria” (as measured by 16S-rRNA gene targets) and represented genes detected per mL (filtered water) or gram (clam tissue, sediment or soil). In contrast, relative abundances have been normalized to the 16S-rRNA gene counts and provide a sense of whether the selection of ARGs in the system has been enriched. Due to data distributions, non-parametric tests (e.g, Mann Whitney and Kruskal Wallis) were used for statistical comparisons.

Repeat sample events were cancelled due to the 2019 SARS-CoV-2 pandemic. However, for data comparisons, we grouped like samples: wastewater effluent (n = 3), the sediment on which the wastewater effluent flowed (n = 3), freshwater samples (Lake Geraldine and Sylvia Grinnell River), Frobisher Bay water (n = 4), and clams (5–8 clams per location; see Table 1).

Compounds detected were atenolol, carbamazepine, metoprolol, naproxen, sulfapyridine, and trimethoprim, but only in the sewage lagoon (Fig. 2). The compounds were found at levels similar (in the ng/L range) to other wastewater systems in Nunavut (Chaves Barquero et al. 2016; Stroski et al. 2020), including work in Iqaluit previously (Stroski et al. 2020). They were also not detected outside of the lagoon itself, which is consistent with previous work at this location (Stroski et al. 2020). The lack of detection suggests that compounds are being (a) degraded through photolytic or biological means within the lagoon or before discharge or (b) the large body of water the compounds enter (i.e., Frobisher Bay) act to dilute so much as to make the concentrations negligible. There was no evidence of these compounds in the drinking water source (Lake Geraldine) and the upstream river reference (all were non-detects).

The pre-screening assay provided relative abundances of 308 ARGs, 53 genetic elements, and 11 critical taxonomies associated with resistance genes (Tables S1 and S3). Among the taxonomy results, differences in gene frequency helped select ARG targets most relevant to the wastewater. For example, the influent comprised 33% Bacteroidetes, 26% Firmicutes, 9.7% Acinetobacter sp., and 8.2% Pseudomonas, representing human-gut microbiota (Thomas et al. 2011). In contrast, the soil had half the percentages of Bacteroidetes and Firmicutes and <1% of the latter two genera.

Based on the ARG pre-screen, the wastewater had a higher richness of resistance genes (positive gene detections) of each antibiotic “type” (see Table S4 for the complete list). The wastewater and soil had the following %positive detections of resistance genes (respectively; selected gene targets for the qPCR are also mentioned) per antibiotic class: aminoglycosides (61%, 22%; aphA3), beta-lactamases (54%, 19%; ampC, cphA, bla_TEM), fluoroquinolones (80%, 50%; qnrA), multidrug resistance genes (83%, 55%; qacH, acrA), macrolide–lincosamide–streptogramin B (59%, 39%; ermB), phenicols (40%, 20%), sulfonamides (71%, 57%; sul1, dfrA), tetracyclines (50%, 36%; tet39), trimethoprim (29%, 18%; dfrA) vancomycin (38% 13%), other resistance genes (53%, 24%), and mobile genetic elements (79%, 55%).

Wastewater received only primary treatment at the time of sampling (August 2019). The WWTP was able to reduce the bacterial gene concentrations from 10^{8.6 (±0.1)} in the influent to 10^{7.6 (±0.5)} genes/mL (90% reduction). When examining ARG distribution between wastewater influent (10^−0.7 genes/16S-rRNA) and effluent (10^−1.2), there appears to be a slight shift in their relative abundances. Neudorf et al. (2017) found similar removal rates. However, here, for most genes, there were no significant differences (t test, p > 0.05; see Supplementary Table S5); exceptions included lower effluent concentrations for qacH and dfrA and higher concentrations of sul1. Primary treatment often remains ineffective in removing ARGs (Graham et al. 2019a); any removal would be attributed to bacteria physically removed with the solids during primary treatment.

Following primary treatment, the discharged wastewater flowed >200 m along the upper-tidal flat to Frobisher Bay. Most ARG values representing wastewater discharges and sediment were not statistically different (Table 2). However, lower relative abundances of qnrA, dfrA, and bla_TEM genes were found in the sediment than in the flowing waters; the differences could be influenced by indigenous bacteria on the tidal flat surface on which the discharged effluent flowed. We do not anticipate selective pressures from discharged pharmaceuticals on the bacteria (PPCP concentrations are low); instead, we were detecting the presence and fate of faecal bacteria and the ARG they contained.

Relative concentrations of ARGs became diluted once they entered Frobisher Bay but remained detectable (Table 2). To determine whether they impacted the Bay, we compared the concentrations in Frobisher Bay with those of inland freshwater sources (i.e., Sylvia Grinnell River and Lake Geraldine). The two inland sites represent “pristine” (minimally impacted) sites.

Although bacteria and fungus levels were orders of magnitude lower in Frobisher Bay and freshwater samples than in wastewater (Table 2), the relative abundances helped discern whether selective pressures remained. From the results, the marine and freshwater samples had comparable concentrations for all but three genes: qnrA, cphA, and ampC, which were similar in all locations. However, non-wastewater samples had lower relative abundances for the other genes. The decreased total bacteria and decline in relative abundances further reduce ARG risks to Frobisher Bay.

However, some concerns become highlighted when one examines the ARG more closely in Frobisher Bay by comparing relative gene abundances from the discharge point. A clear inverse trend was observed between gene abundances and distance (Table 3), which suggests that wastewater discharges may impact water conditions in Frobisher Bay. This analysis remains rudimentary as actual travel distance would not be direct but would be influenced by the complex hydrological dynamics of circulation and tidal fluctuations. However, sample collection began at high tide, and the influx of marine waters could have influenced the results. Kituriaqannigituq (Bay No. 4) is located at a different inlet and unlikely to be influenced by Iqaluit's wastewater; as a “control” site, it provides a context of expected gene concentrations in the Bay.

Similarly, we detected resistance genes in the tissues of truncate softshell clams sampled at multiple sites (Table 4). The Kruskal–Wallis test showed no significant differences among the ARG at the six sampling locations and no clear trends or patterns (Table S6). However, higher bacteria levels were found in Koojesse Inlet (near the point of wastewater discharge (Clam No. 1), Apex (Clam No. 3), and Monument Island (Clam No. 4); H₅ = 12.8, p = 0.03; Table 4).

It is hypothesized that the reason for no significant differences in ARG levels in M. truncata was related to their storage following harvest. As previously mentioned, clams were harvested by Schaefer et al. (2022) for their biometrics; as part of their study, clams were held in artificial seawater (2–4 ˚C) before dissection. M. truncata filters 2.5 L/h (Petersen et al. 2003; Bernard and Noakes 1990) and can rapidly digest the ARG-containing wastewater bacteria. As such, it was likely that the bacteria would have been flushed from the clams. This “depuration” method has been utilized to reduce potential health hazards of bacteria (Metcalf et al. 1979) and viruses (Polo et al. 2014), and the same process could have happened here. Further investigation is required to scrutinize depuration impacts. Nevertheless, an environmental risk would remain for harvesters in contact with potentially contaminated seawater. Therefore, improving wastewater treatment would confer the greatest anthropogenic and environmental benefits.