Fate of antibiotic resistance genes and resistant bacteria under various operating temperatures of sludge anaerobic digestion Open Access

The discovery of antibiotics in the early 20th century revolutionized medicine by providing effective treatments for previously life-threatening bacterial infections. However, the widespread use of antibiotics in human healthcare and animal husbandry has led to significant unintended consequences. Research indicates that a large portion of antibiotics administered to humans and animals is excreted either unmetabolized or as secondary metabolites, which contributes to environmental contamination (Ma et al. 2022). This accumulation puts selective pressure on bacteria, enabling them to develop resistance mechanisms and facilitating the survival and proliferation of antibiotic-resistant bacteria (ARB) and antibiotic resistance genes (ARGs). Antibiotic resistance is responsible for at least 1.27 million deaths each year (Murray et al. 2022), thereby prompting the World Health Organization to identify antibiotic resistance as one of the critical public health challenges of the 21st century that poses a significant threat to human, animal, and environmental health (Ping et al. 2022).

Antibiotic resistance arises primarily through genetic changes in bacteria, driven by two key mechanisms: mutation and horizontal gene transfer (HGT). Spontaneous mutations during bacterial replication can result in resistance, while HGT enables bacteria to acquire resistance genes from others through conjugation, transformation, or transduction. These mechanisms allow resistance to spread rapidly within bacterial populations, particularly under selective pressures from antibiotics present in the environment. Such pressures are amplified by extensive antibiotic use in clinical settings, and agricultural and animal farming practices, creating hotspots for the dissemination of resistance (Munita & Arias 2016; Zhang et al. 2022).

Antibiotics and ARB enter the environment through multiple pathways, including the improper disposal of pharmaceuticals, wastewater discharge, and agricultural practices. For instance, antibiotics used in livestock production can lead to resistant bacteria in animals, which spread to humans through direct contact or consumption of contaminated food (Manyi-Loh et al. 2018). Rainwater runoff exacerbates this issue by carrying ARGs and ARB into rivers and streams, further spreading resistance across ecosystems. Also, environmental contamination plays a critical role in the spread of ARGs, especially the use of biosolids from wastewater treatment plants (WWTPs) in agriculture (Leiva et al. 2021). This dissemination creates a feedback loop with serious implications for public health and the environment.

In WWTPs, wastewater is collected from various sources, including hospitals, households, pharmaceutical industries, and agricultural runoff. This results in a complex mixture of antibiotics, resistant bacteria, and ARGs. Such a mixture creates an environment that promotes the proliferation and exchange of resistance genes, presenting significant challenges for treatment systems. While WWTPs play a crucial role in mitigating environmental contamination, they often struggle to fully eliminate all antibiotics, ARB, and ARGs, especially in wastewater sludge, where these contaminants are highly concentrated (Harrison et al. 2024).

In WWTPs, anaerobic digestion (AD) is a widely utilized technology for the treatment of sludge. This biological process facilitates the breakdown of organic matter by microorganisms in the absence of oxygen, resulting in the production of biogas and the reduction of pathogens in the sludge. However, the effectiveness of AD in reducing ARGs varies considerably. Research has identified several key factors that influence ARG reduction during AD, including retention time, the presence of other pollutants, and, importantly, temperature (Zhao & Liu 2019; Zou et al. 2020; Mortezaei et al. 2023). The relationship between temperature and ARG reduction is complex and can yield inconsistent results in some cases. Certain studies have indicated that specific ARGs may persist or even increase under thermophilic conditions, and some antibiotics may survive the digestion process entirely (Huang et al. 2019; Xiao et al. 2021). Nonetheless, thermophilic digestion (around 55 °C) tends to be more effective at reducing ARB and ARGs than mesophilic digestion (around 37 °C) (Syafiuddin & Boopathy 2021).

These findings emphasize the intricate interplay between digestion temperature, microbial communities, and the proliferation of ARGs. The varied results highlight the challenges of addressing antibiotic resistance in WWTPs and underscore the significant role that temperature plays in the elimination of ARGs. Our study aimed to investigate and compare the efficacy of temperatures under various conditions (mesophilic and thermophilic), including those just above thermophilic levels, since these temperatures appear promising for the elimination of ARGs. By focusing on the effects of elevated temperatures, we were able to gain a deeper understanding of the varying behaviors and selectivity in ARG reduction across different temperature regimes, ultimately providing clearer insights into the potential benefits of AD.

Three bench-scale continuously stirred anaerobic digesters, each with a total volume of 11 L and an operating volume of 9 L, were set up to evaluate the effects of temperature on AD. The reactors were maintained at three different temperatures: 37, 55, and 65 °C, using heating coils monitored by temperature probes connected to cRIO LabVIEW modules (National Instruments, USA). The reactors were inoculated with sludge sourced from anaerobic digesters of two municipal WWTPs in Prague (Czech Republic). The first reactor (37 °C) was inoculated with mesophilic sludge from a mid-size WWTP, while the second and third reactors (55 and 65 °C) were inoculated with thermophilic sludge obtained from a large WWTP. To ensure anaerobic conditions, all reactors were flushed with nitrogen gas prior to startup. After inoculation, the reactors underwent a stabilization period to allow microbial communities to adapt to the specific operating conditions. Stabilization was considered complete when steady biogas production and stable pH values were achieved, ensuring reliable data collection during the experimental phase (Figure 1).

Once the stable state of operation was achieved, the digesters were fed with the feedstock. The feedstock used for all reactors was mixed raw sludge (MRS) from the Prague municipal WWTP, where the anaerobic digesters are operated at thermophilic conditions of 55 °C (±1 °C). The MRS was composed of a 1:1 volume ratio of primary and activated sludge. Before feeding, the MRS was homogenized to ensure uniform distribution of solids and consistent composition. The sludge was stored at 4 °C to minimize microbial activity prior to feeding and continuously mixed to maintain homogeneity. Each reactor was fed 0.45 L of sludge per day using a peristaltic pump, achieving a consistent sludge retention time of 20 days. This feeding rate ensured stable operation. Biogas production was continuously measured using a Ritter gas meter (Germany), allowing for the monitoring of biogas output. Total COD (T_COD) and soluble COD (S_COD) were measured in both the influent (feedstock) and the effluent (digestate) from all three digesters to assess the organic matter reduction during digestion. COD removal was then calculated as a performance indicator for each reactor. In addition to COD, key parameters influencing AD, like pH, total solids (TS), and volatile solids (VS), were monitored regularly. All these parameters were measured following standard methods (APHA 2017). Additionally, biogas composition (methane, carbon dioxide, and other gases) was analyzed periodically using gas chromatography to evaluate the efficiency of AD under different temperature regimes (Table 1).

Upon achieving steady-state conditions, digestate samples were systematically collected from all three anaerobic digesters. To ensure homogeneity, the digestate was thoroughly stirred using a magnetic stirrer for 5 min at a controlled speed. Aliquots weighing between 0.25 and 0.31 g were transferred into sterile 2 mL Eppendorf tubes, which were then used to extract the DNA, and the extracted DNA was stored at −20 °C until further analysis. Biweekly sampling was conducted over a 1-year period. DNA extraction was performed using the DNeasy PowerSoil Kit (Qiagen, Hilden, Germany), following the manufacturer's protocol. DNA concentration was quantified using a Qubit fluorometer (Thermo Fisher Scientific), confirming a final concentration of 10 ng/μL with sample volumes of 100 μL. Purity was assessed via spectrophotometric analysis, achieving an A260/A280 ratio of 1.8 or above, thereby validating suitability for downstream applications.

Quantitative PCR (qPCR) analysis was outsourced to Resistomap (Finland) and conducted using a high-throughput qPCR array system on the SmartChip PCR platform (Takara Bio). Target ARGs were selected from a panel of 128 genes based on a preliminary survey of wastewater samples from seven WWTPs in Czechia. Selection criteria included gene abundance, prevalence, and representation across major ARG groups (e.g., beta-lactams, tetracyclines, macrolides). Primers used in the Resistomap system were validated for specificity, taxonomic coverage, and efficiency using the UniPriVal tool (Gorecki et al. 2019). Absolute quantification adhered to established protocols (Rocha et al. 2020; Muurinen et al. 2022) and included calibration curves prepared using known standards. Each sample was analyzed in triplicate for reliability, and ARG abundance was normalized to 16S rRNA gene copies to account for microbial biomass differences.

The DNA was sequenced commercially using the platform NovaSeq 6000. The preparation of libraries involved a two-step PCR employing the Nextera technology targeting the 16S rRNA gene's V4 and V5 hypervariable regions using the primers 515F (5′-GTGYCAGCMGCCGCGGTAA-3′) and 926R (5′-CCGYCAATTYMTTTRAGTTT-3′). A mock community consisting of 10 bacterial strains was included as a positive control. The data obtained in this work were deposited in the NCBI Short Read Archive under BioProject accession number PRJNA1244292 (Table 2).

The amplicon sequence variants (ASVs) for each sample and replicate were analyzed using the DADA2 pipeline (Callahan et al. 2016). The ASVs were trimmed and filtered by their quality [truncLen = c(225, 220), maxN = 0, maxEE = c (1, 1), truncQ = 2], and error rates were learned by weights and span, and monotonicity enforcement. As part of the DADA2 pipeline, sequencing errors were removed, the denoised forward and reverse reads were merged, and chimeric sequences were removed. The taxonomy of ASVs was assigned using the Silva ribosomal RNA gene database (Quast et al. 2013). A maximum likelihood phylogenetic tree was generated using the Phangorn package (Schliep 2011) after aligning the ASVs with the DECIPHER package (Wright 2016). The best-fitting model was determined using the Phangorn function modelTest. The data were further manipulated using the Phyloseq package (McMurdie & Holmes 2013). Sequences coming from mitochondria, eukaryotes, and chloroplasts, as well as uncharacterized phyla, were removed from the dataset. Low-abundance data were removed by filtering out ASVs with abundances less than 20. Less than 4% of the data were discarded by this low-abundance filtering. The data were ordinated using the ordinate function in Phyloseq, using the principal coordinate analysis method with the UNIFRAC distance obtained from the phylogenetic tree.

Statistical analysis was performed to evaluate trends in ARG abundance and microbial diversity across the reactors. For qPCR results, ARG abundance data were log-transformed and normalized. Analysis of variance (ANOVA) was conducted to identify differences between reactors, followed by post hoc Tukey-HSD tests with corrections for multiple comparisons. To find the relationship between ASV data and the ARG abundance, a sparse canonical correlation analysis using the penalized matrix decomposition was done in R using the package PMA using the canonical correspondence analysis (CCA) function (Witten et al. 2009). ASV data were agglomerated to the genus level, and taxa without any count were filtered out of the analysis. Both ASV abundances and ARG abundances were normalized to the 16s rRNA gene quantification resulting from the resistomap analysis. These two abundances, which constitute two different matrices, were analyzed with the CCA function using a penalty value of 0.2. It was required that both arguments vpos and upos were positive since only a positive correlation between ARG abundance and ASV abundance was of interest. The data from this analysis was manipulated and visualized using the R packages tidyverse (Wickham et al. 2019), ggplot (Wickham 2016), Cowplot (Wilke 2024), and ComplexHeatmap (Gu et al. 2016; Gu 2024).

The performance of the three anaerobic digesters (R1, R2, and R3) was assessed throughout the experiment by monitoring various operational parameters, including temperature, pH, COD removal, biogas quality, and solids characteristics. The temperatures within each reactor remained stable during the entire run, with R1 operating at 37 °C (±0.8), R2 at 55 °C (±0.6), and R3 at 65 °C (±0.9), which correspond to the respective mesophilic, thermophilic, and higher thermophilic conditions. The pH values recorded were within the optimal range for AD, with R1 maintaining a pH of 7.1 (±0.2), R2 at 7.3 (±0.3), and R3 at 7.6 (±0.2). These pH levels supported microbial activity, ensuring stable reactor performance. The slight increase in pH observed at higher temperatures can be attributed to the increased production of alkaline byproducts like ammonia (Lin et al. 2016).

In terms of COD removal efficiency, the thermophilic reactor R2 achieved the highest removal rate at 58.6%, followed by R1 at 52.2%, and R3 at 47.7%. This trend indicates that while thermophilic conditions (55 °C) significantly enhance the degradation of organic matter, the performance at 65 °C was comparatively less effective. The reduced COD removal in R3 may be linked to factors such as increased thermal stress on microbial communities, which is a recognized limitation at elevated temperatures. These findings support the hypothesis that diminishing returns in COD removal occur when operating beyond an optimal temperature threshold, as increased temperatures may place stress on microbial populations and impair the efficiency of the digestion process (Finore et al. 2023). A similar trend is observed when it comes to solids measurements; the lower TS, VS, and VS/TS values of R1 and R2 are indicative of the optimal digestion of the available organic matter, and hence stable functioning of these reactors. On the other hand, the high values of these parameters for R3 are indicative of the reduced microbial activity due to the increased thermal stress (Table 3).

It is clearly demonstrated in Figure 2 that AD alone is effective to some degree in reducing ARGs. However, an increase in temperature enhances the reduction rates of these ARGs. This trend is distinctly evident for all studied genes, with the exception of ermB. Notably, the treatment at R3, which operates at the highest temperature of 65 °C, consistently achieves a reduction of ≥1.5 log units across all gene categories. This indicated that temperature does indeed affect ARGs (Zhang et al. 2015; Wang et al. 2021).

Furthermore, the distribution of data across each temperature condition offers additional insights into the variability and consistency of ARG reduction. For R1, the box plots exhibit a larger interquartile range, indicating greater variability in reduction percentages at this temperature. This suggests that at 37 °C, environmental or experimental factors may have contributed to less predictable outcomes, with some samples achieving high reductions while others displayed significantly lower reductions. In contrast, at higher temperatures, particularly R2 and R3, the interquartile ranges are narrower, reflecting more consistent and reliable reductions in ARG concentrations. This indicates that higher temperatures not only enhance the reduction of ARGs but also do so in a more uniform and predictable manner.

The marked reduction of ARGs at 65 °C, achieving ≥1.5log units (∼97% removal), demonstrates the unique potential of higher thermophilic conditions as a frontline strategy for mitigating microbial risks in sludge. This reduction surpasses conventional thermophilic digestion (55 °C), where ARG removal typically ranges between 0.1 and 0.72 log units (Wu et al. 2016), and aligns with hyperthermophilic composting systems that achieve 89% ARG elimination through similar temperature-driven suppression of bacterial hosts (Liao et al. 2018). At 65 °C, the drastic decline in microbial diversity and viability disrupts ARG reservoirs, while extracellular DNA degradation further destabilizes HGT pathways. However, the tradeoff lies in organic degradation efficiency: hydrolytic and methanogenic activity is significantly curtailed at 65 °C, limiting biogas yields and sludge stabilization. This underscores the need to contextualize hyperthermophilic treatment within broader systems, such as temperature-phased anaerobic digestion, where 65 °C could serve as a pretreatment phase to maximize ARG suppression before transitioning to lower temperatures for organic recovery.

In terms of ARG-specific trends, the reduction of different ARGs was not uniform across all temperatures. Some genes, such as aadA (aminoglycoside resistance) and strB (streptomycin resistance), exhibited relatively high and consistent reductions across all temperature conditions. This consistent reduction suggests that these genes are more susceptible to temperature-induced degradation or inactivation. This trend may be because both aadA and strB are often plasmid-borne, as plasmid DNA tends to be more susceptible to heat stress and instability, leading to a loss or degradation of the plasmid that carries the gene (Mingeot-Leclercq et al. 1999; Ludvigsen et al. 2018). The cmlA gene, which provides resistance to chloramphenicol, shows a significant and consistent reduction in expression at elevated temperatures. Interestingly, this gene encodes an efflux pump, which can be disrupted by heat. Thus, the efficiency of antibiotics transport from the bacterial cell is compromised at elevated temperature (Bischoff et al. 2005). Most probably, these efflux pumps are not dominantly present in the organisms adapted to higher temperatures, which would explain the lower presence of the cmlA gene in the R2 and R3.

There are also other genes that exhibit consistent and significant reductions in ARG concentrations only at higher temperatures (55 and 65 °C). These genes are particularly sensitive to thermal stress, with their resistance mechanisms disrupted by heat, leading to a significant reduction in resistance levels. For example, blaVIM and blaOXA, which encode beta-lactamases that hydrolyze beta-lactam antibiotics, likely lose enzymatic activity at elevated temperatures due to protein denaturation (Lauretti et al. 1999; Zhou et al. 2024). Similarly, intl1, which is involved in integron-mediated gene transfer, likely experiences a reduction in its activity as the integrase enzyme becomes destabilized under heat (Sawa et al. 2020). The sul1 gene, which confers resistance to sulfonamides, also shows significant reductions at higher temperatures, as the dihydropteroate synthase enzyme becomes thermally unstable (Venkatesan et al. 2023). These genes exhibit high variability at 37 °C, but their response becomes more consistent and significant at higher temperatures, suggesting that heat disrupts their functions and reduces resistance.

In contrast, other efflux-related genes like mexE and mefB display greater variability in their responses to temperature changes. While these genes also encode pumps that expel antibiotics, bacteria can often compensate by activating alternative resistance mechanisms, resulting in inconsistent reductions (Webber & Piddock 2003; Li et al. 2015). The gene tetM, which provides resistance to tetracyclines by protecting the bacterial ribosome, appears more structurally stable at higher temperatures, contributing to its lower and more variable reduction across all temperature conditions (Trzcinski et al. 2000). These genes are more resilient or adaptable to heat-induced stress, as their resistance mechanisms may involve compensatory processes or more stable structural elements.

This study demonstrates the efficacy of higher thermophilic digestion in reducing key ARGs such as sul1, ermB, and intI1. However, not all clinically critical genes were included. ARGs like sul2, blaTEM, and qnr were not selected due to their low abundance in our sludge samples. These targets – prioritized in standardized risk panels, should be incorporated in future analyses to fully assess treatment efficacy against high-priority AMR threats (Guo et al. 2017). Still, observed reductions in intI1 and plasmid-borne genes such as aadA suggest that 65 °C treatment broadly disrupts mobile genetic elements (MGEs), key vectors of resistance dissemination (Guo et al. 2017; Jang et al. 2018).

Although pharmaceuticals and metals were not quantified in this study, prior research shows that higher thermophilic conditions can degrade antibiotics like sulfamethoxazole by destabilizing target enzymes (e.g., dihydropteroate synthase in sul1) (Liao et al. 2018) and reduce metal bioavailability via precipitation (Tian et al. 2016; Jang et al. 2018). These effects likely lower selective pressures compared to mesophilic processes. Nonetheless, interactions between residual chemicals and ARGs merit further investigation, especially for sludge intended for agricultural use (Ma et al. 2011). Given that even low ARG levels pose risks when MGEs persist (Urra et al. 2019), we recommend limiting high-risk applications – such as raw land use – unless paired with additional treatments to degrade extracellular DNA (Urra et al. 2019; Hong et al. 2020). For lower-risk scenarios, 65 °C treatment offers a practical option by reducing both pathogens and ARGs. However, long-term field trials are essential to monitor HGT in receiving environments (Sorinolu et al. 2021).

Our findings suggest that while AD inherently reduces ARGs, the effectiveness of this reduction is significantly enhanced at elevated temperatures. In our study, AD at elevated temperatures (65 °C) achieved substantial reductions in ARGs, with ≥1.5 log reduction (∼97% elimination) for key targets such as sul1, aadA, and intI1, correlating with a marked decline in gram-negative bacterial hosts (e.g., Alicycliphilus, Macellibacteroides, Dokdonella, Ahniella, Thauera) critical to ARG dissemination. However, this enhanced ARG reduction comes at the cost of reduced COD removal efficiency by 5%–10% compared to digestion at 37 and 55 °C. The energy demands of maintaining higher thermophilic conditions further underscore the need to balance operational costs with ARG mitigation. Future work should prioritize cost–benefit analyses of such hybrid configurations while expanding ARG profiling to include high-risk clinical targets (e.g., sul2, blaTEM) to align with standardized risk frameworks. By integrating microbial ecology insights with process engineering, AD can evolve into a dual-purpose solution – curbing environmental AMR propagation while maintaining feasibility for large-scale wastewater treatment.

Graphical Abstract and Figure 1 were created in BioRender. Can be accessed at: https://BioRender.com/f41a849; https://BioRender.com/s19t687.

This work was supported by the Technological Agency of the Czech Republic [Project ARG Tech, No. SS01020112].

J.M.B. experimentation; sample collection and chemical analysis; data gathering; statistical data analysis; writing manuscript; S.P. contributed in qPCR data analysis; writing molecular methods; M.A.L.M. sample sequencing; sequencing data analysis; L.A., J.B. conceptualized the work; supervised the whole project; wrote and reviewed and edited the article.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.