Transition of antimicrobial resistome in wastewater treatment plants: impact of process configuration, geographical location and season

Antimicrobial resistance is a major global health threat and wastewater is a key environmental pathway for dissemination. Wastewater treatment plants are critical barriers but can also serve as reservoirs and sources of antimicrobial-resistant bacteria and genes. Prior studies indicate that influent and effluent resistomes differ, yet generalizable determinants governing resistome transitions across treatment configurations, locations, and seasons remain unclear. Additionally, previous shotgun metagenomics often relied on raw read counts, potentially biasing results toward longer genes. This study investigates how antimicrobial resistomes transition from influent to sludge to effluent across multiple WWTPs, seasons, and process configurations, using gene-length-normalized metagenomic data to accurately characterize abundance and composition and to identify factors (e.g., seasonality, process configuration) that shape these transitions.

Past work shows WWTPs can be hotspots and reservoirs for antibiotic resistance genes and mobile elements, and that effluents still carry ARB/ARGs. Metagenomic and qPCR studies have reported reductions in total ARGs from influent to sludge, while culture-based studies sometimes show limited change due to targeting specific fecal bacteria. Case studies comparing influent and effluent resistomes exist but typically focus on single plants, limiting generalizability. Moreover, many analyses quantified ARGs from raw read counts without correcting for gene length, which can overestimate long ARGs (e.g., mexB, tetQ) relative to shorter genes (e.g., mphB, qnrS), leading to biased composition estimates. The present study addresses these gaps by surveying diverse WWTPs and seasons and by normalizing read counts to reads per kilobase to better reflect gene copy proportions.

Study design and sampling: Five municipal WWTPs in different Japanese municipalities were selected to represent diverse biological configurations: A (conventional activated sludge, CAS), B (parallel CAS and membrane bioreactor, MBR, treating identical wastewater), C (enhanced biological phosphorus removal, EBPR), D (A2O: anaerobic-anoxic-oxic), and E (oxidation ditch, OD). Influent wastewater, activated sludge (aeration/oxic tank), and treated effluent (post-chlorination) were sampled during dry weather three times in winter (Dec 2018–Feb 2019) and three times in summer (Jun–Aug 2019), at least two weeks apart. At WWTP B, samples from each treatment stage were also collected in February and August: influent (WW), primary sedimentation effluent (PTE), equalization tank (EQ), anoxic (Ano, MBR), oxic/aeration (Ox, CAS and MBR), membrane tank (Mem, MBR), secondary treated effluents (STE: CAS final clarifier effluent and MBR permeate), and final chlorinated effluent (EFF). For WWTP E, influent was sampled at the aeration tank inlet due to no primary sedimentation. Sample processing and DNA extraction: For influent and sludge, 50 mL samples were centrifuged (10,000×g, 15 min); DNA was extracted from pellets using the FastDNA Spin Kit for Soil. For treated effluent, 200 mL (or 2 L for MBR effluent at WWTP B) was filtered through 0.2-µm polycarbonate membranes; DNA was extracted from filters with the same kit. DNA concentrations were measured spectrophotometrically. Seasonal composite DNA extracts were prepared by mixing equal DNA masses from the three sampling dates per matrix (wastewater, sludge, effluent) per WWTP to reduce daily fluctuation, yielding 43 DNA extracts. 16S rRNA gene analysis: Microbial communities were profiled from V3–V4 16S rRNA regions using a two-step tailed PCR (primers and cycling in Supplementary Table 3), library cleanup, quality validation, and Illumina MiSeq 2×300 bp sequencing. Reads were trimmed/filtered (Fastx toolkit, sickle), merged (FLASH), and analyzed in QIIME2 (v2020.2). Chimeras were removed (dada2), and OTUs assigned using Greengenes (13_8, 97% identity). Data deposited to DDBJ (accessions provided). Total bacterial population was quantified by qPCR of 16S rRNA gene using primers 341F/805R and SYBR Green chemistry; cycling: 95°C 3 min; 40 cycles of 55°C 45 s and 72°C 1 min. Shotgun metagenomics: DNA was fragmented to ~500 bp (Covaris S220), libraries prepared (KAPA HyperPlus with FastGene adapters), quality-checked, and sequenced (Illumina HiSeq X, 2×151 bp). Reads were filtered/trimmed (Enveomics pipeline). ARG-containing reads were identified by BLASTn against CARD v3.0.7 with E-value cutoff 1e-5, producing raw ARG profiles (read counts per ARG). To correct length bias, read counts were normalized to reads per kilobase (RPK) using subject sequence lengths from CARD. 16S rRNA gene reads were counted (Parallel-META 3) and normalized to RPK using 1541 bp as 16S length. ARG composition was the relative proportion of each ARG’s RPK to total ARG RPK. Total ARG abundance was the sum of ARG RPK divided by 16S rRNA RPK. Multivariate analyses: Principal component analysis (PCA) compared ARG composition profiles and genus-level 16S community profiles (R v4.0.0), with scaling to emphasize relative changes. Correlations between ARGs and phylogenetic classes (>1% in any composite) were computed to associate microbial groups with ARG classes. Sequencing data are available in DDBJ; correlation tables are available via Mendeley Data.

• Total ARG abundance per 16S population in influent wastewater was approximately 32–50% across WWTPs and seasons. It consistently decreased in activated sludge to about 5–19% across plants and seasons.
• Along the treatment train at WWTP B, total ARG abundance showed little change from influent through primary sedimentation and equalization, then decreased during biological treatment to 15–21% in CAS and 10–12% in MBR. Chlorination markedly reduced bacterial 16S counts (by ~1.0–1.7 log) but had limited impact on total ARG abundance.
• ARG composition shifted at two points: (1) at the start of biological treatment, from influent traits to sludge traits; and (2) at sludge separation in CAS (final sedimentation), but not in MBR (membrane filtration).
• Only 45–50% of influent ARGs were retained in activated sludge; of those present in sludge, 84–91% were carried over to treated effluent.
• Influent wastewater was enriched in ARGs conferring resistance to clinically important drugs (tetracyclines, fluoroquinolones, macrolides, aminoglycosides, cephalosporins); sludge was enriched in multidrug resistance (MAR) genes, particularly efflux pump families (e.g., mex, Mux). The proportion of ARGs with broad resistance to more than six drug classes notably increased in sludge.
• Seasonality was the primary factor affecting the wastewater resistome: winter samples showed higher loads of beta-lactam inactivation genes (e.g., OXA, MOX, GES, SHV, CMY, ADC) aligned with increased clinical use of cephalosporins and macrolides in winter in Japan. Seasonal effects were most pronounced in influent and partially propagated to sludge and effluent.
• Treated effluent ARG compositions were intermediate between influent and sludge in CAS, indicating carryover of wastewater-origin ARGs through final sedimentation. In MBR, effluent ARG composition closely matched sludge (positive PC1 scores), indicating effective exclusion of wastewater-origin ARGs by membrane separation.
• Chlorination reduced bacterial populations but did not markedly reduce ARG abundance, consistent with prior reports under typical chlorination intensities.
• Microbial community analyses paralleled ARG patterns: influent enriched in gut-associated anaerobes (e.g., Bacteroidetes, Firmicutes, Gammaproteobacteria), correlated with macrolide, quinolone, tetracycline, and vancomycin ARGs; sludge enriched in environmental/aerobic taxa (e.g., Alphaproteobacteria, Deltaproteobacteria, Planctomycetes, Chloroflexi), correlated with sulfonamide ARGs and multidrug efflux genes.
• Process configuration notably influenced the effluent resistome via sludge separation: CAS showed a second compositional shift due to overflow of wastewater-origin ARGs at final sedimentation; MBR did not, implying ARGs of wastewater origin are abundant in the supernatant and are better removed by membrane filtration (including extracellular DNA).

The study addressed how antimicrobial resistomes transition through WWTPs and which factors determine these changes. It demonstrated consistent decreases in total ARG abundance during biological treatment and highlighted distinct resistome signatures: influent dominated by ARGs to clinically important antimicrobials, sludge dominated by multidrug efflux-related ARGs, and effluent reflecting a mixture governed by sludge separation performance. Seasonality in clinical antibiotic usage primarily shaped the wastewater resistome and partially propagated downstream. PCA showed two key transition points, underscoring the importance of the biological treatment onset and the sludge separation step in determining final effluent resistomes. The microbial community structure was aligned with ARG patterns, suggesting probable hosts for different ARG classes: gut-associated taxa for clinically relevant ARGs and sludge environmental taxa for multidrug efflux genes. Process configuration did not markedly alter the sludge resistome but had a strong effect on effluent resistome via sludge separation, with MBR better limiting wastewater-origin ARGs. These findings are significant for AMR mitigation policies: improving sludge separation and targeting extracellular ARGs in supernatants can reduce discharge of clinically relevant ARGs; recognizing seasonal influences can aid surveillance and management strategies.

This work provides a multi-plant, multi-season, gene-length-normalized metagenomic assessment of resistome transitions in WWTPs. It identifies two critical transition points shaping resistome composition, reveals that influent, sludge, and effluent harbor distinct resistomes, and shows that seasonality governs wastewater resistomes while sludge resistomes consistently enrich multidrug efflux genes. The effluent resistome is strongly influenced by sludge separation configuration: CAS effluents retain wastewater-origin ARGs, whereas MBR effluents more closely reflect sludge resistomes, indicating superior removal of wastewater-origin (including extracellular) ARGs by membrane filtration. These insights suggest that optimizing sludge separation (e.g., adopting MBR or enhancing removal of supernatant-associated ARGs) and accounting for seasonal antibiotic use can help reduce ARG discharge to aquatic environments. Future research should capture temporal fluctuations (hourly/daily), evaluate cross-country uniformity or differences in WWTP resistomes, and develop and test process designs and operating strategies specifically targeting extracellular and supernatant-associated ARG removal.

The study used seasonal composite samples, precluding statistical comparisons across individual sampling days and limiting resolution of short-term (hourly/daily) fluctuations in resistomes. Some sampling constraints (e.g., influent sampling point at WWTP E and potential sludge carryover in certain influent samples) may have introduced minor biases. Findings reflect Japanese WWTPs and seasons in 2018–2019; broader generalization warrants cross-regional studies. Chlorination conditions reflected typical practice and may not represent higher-intensity disinfection scenarios.