Propagation of antibiotic resistance genes during anaerobic digestion of thermally hydrolyzed sludge and their correlation with extracellular polymeric substances

The positive impact of the thermal hydrolysis process (THP) of sewage sludge on antibiotic resistance genes (ARGs) removal during anaerobic digestion (AD) has been reported in the literature. However, little information is available on how changes in different extracellular polymeric substances (EPS) due to THP can influence ARG propagation during AD. This study focused on systematically correlating EPS components and ARG abundance in AD of sewage sludge pretreated with THP (80 °C, 110 °C, 140 °C, 170 °C). THP under different conditions improved sludge solubilization followed by improved methane yields in the biochemical methane potential (BMP) test. The highest methane yield of 275 ± 11.5 ml CH4/g COD was observed for THP-140 °C, which was 40.5 ± 2.5% higher than the control. Increasing THP operating temperatures showed a non-linear response of ARG propagation in AD due to the rebound effect. The highest ARGs removal in AD was achieved with THP at 140 °C. The multivariate analysis showed that EPS polysaccharides positively correlated with most ARGs and integrons, except for macrolides resistance genes. In contrast, EPS protein was only strongly correlated with β-lactam resistance genes. These results suggest that manipulating THP operating conditions targeting specific EPS components will be critical to effectively mitigating the dissemination of particular ARG types in AD.


Material and methods
Sludge and inoculum. Primary sludge, waste activated sludge, and anaerobically digested sludge were collected from the Gold Bar Wastewater Treatment Plant (Edmonton, Alberta, Canada) and stored at 4 °C before use. The primary sludge (i.e., settled solids from primary clarifier) was mixed with waste activated sludge at a volume ratio of 1:1 and used for the experiment. The detailed characteristics of sludge and inoculum are provided in Table 1.
Thermal hydrolysis and biochemical methane potential (BMP) test. 2 L bench-scale hydrothermal reactor (Parr 4848, Parr Instrument Company, Moline, IL, USA) was used for thermal hydrolysis of sludge at four different temperatures (80 °C, 110 °C, 140 °C, 170 °C) for 60 min exposure time. This exposure time is within the range reported in the literature 7,30 . For each experimental condition, 500 mL of feedstock (mixture of primary sludge and waste activated sludge) was fed to the hydrothermal reactor. The detailed operating protocol has been described elsewhere 27 .
The biomethane potential of raw and pretreated sludge was appraised with the BMP test. The BMP tests were performed with glass anaerobic bioreactors (working volume of 300 mL) equipped with mechanical agitators and electric motors (ISES-Canada, Vaughan, ON, Canada). The feedstock and inoculum volumes were used based on the food to microorganism ratio (F/M) of 2 (g of total chemical oxygen demand (TCOD) of sludge/g of volatile suspended solids (VSS) of inoculum). Furthermore, a blank test (inoculum + deionized water) was performed to evaluate methane production from the inoculum. Before start-up, the reactors were purged with nitrogen gas for 3 min and then placed in water baths at 37 ± 2 °C. The liquid was continuously mixed at 300 rpm. www.nature.com/scientificreports/ All tests were conducted in triplicate. The BMP tests were operated in a batch mode for 38 days, and the samples were taken before (day 0) and after the BMP tests (day 38) for analyses. Methane production was monitored daily using gas bags connected to sequestration bottles for capturing acidic gases (e.g., CO 2 , H 2 S). These bottles contained 3 M NaOH solution and a thymolphthalein indicator 31 . The methane gas volume was measured by a frictionless glass syringe.
Analytical methods. TCOD (Table S1). For microbial diversity evaluation, the nucleotides sequence reads were stored out by using a data analysis pipeline. A denoising and chimera detection steps were carried out to remove short sequences, chimeric sequences, and noisy reads. After that, each sample was run using the analysis pipeline to determine the taxonomic information for each component in the sample. Quantitative Insights Into Microbial Ecology (QIIME) pipeline (QIIME2, Version 2021.2) was used to analyze microbial communities' taxonomy according to Zakaria et al. 34 .

Quantification of ARGs.
Quantitative polymerase chain reaction (qPCR) was used for quantifying thirteen frequently detected ARGs including 7 tetracycline resistance genes (tetA, tetB, tetC, tetW, tetM, tetQ, tetX), 2 sulfonamide resistance genes (sul1, sul2), 2 macrolide resistance genes (ermB, ermC) and 2 ß-lactam resistance genes (bla AOX , bla TEM ). In addition, integrons (intl1, intl2) and 16S rRNA were also quantified. The primers of the selected genes are provided in Table S1. QuantiFast SYBR Green PCR Kit (Qiagen, CA) was used for the preparation of qPCR mixtures in 25 µL reactions as following: 2 µL of the DNA template, 12.5 µL 2 × master mix, 2.5 µL forward and reverse specific primer, and 5.5 µL nuclease-free water. Then, the CFX 96 real-time PCR system with a C1000 Thermal Cycler (Bio-Rad, USA) was used for the quantification process according to the Quanti-Fast SYBR Green PCR Kit's protocol. The PCR initial heat activation cycle at 95 °C for 5 min, 35 cycles at 95 °C for 10 s and 60 °C for 30 s, and one cycle at 40 °C for 30 s. All samples were run in triplicate.
EPS characterization. The EPS extraction was carried out by the heating method due to its high performance reported in the literature 35 . The biomass samples were centrifuged at 3000 × g for 15 min at 21 °C. Then, the supernatant was removed, and the pellet was washed with 0.1 M PBS (pH 7.4) three times. After washing, pellets were collected for cell lysis rate examination by Glucose-6-Phosphate Dehydrogenase kit (Sigma-Aldrich, USA). The details of EPS extraction and analytical methods were performed as previously described in the literature 34 . Carbohydrates were measured using the phenol-sulfuric acid method using glucose as a standard; details could be found in the literature 34
Compared to the raw sludge, TAN concentrations increased in all pretreated samples (Fig. 1d). Notably, TAN concentration reached up to 436.3 ± 8.9 mg/L (p = 0.008) at 170 °C, while TAN concentration in raw sludge was 139.15 ± 1.2 mg/L. A significant increment of TAN concentrations is typically attributed to the hydrolysis of nitrogenous compounds, such as proteins 27 . Despite remarkable increments in TAN after THP, TAN levels in all the samples were lower than 440 mg/L, which was much lower than inhibitory TAN concentrations (4.2 g/L) previously reported for AD 40 . Due to the further hydrolysis during AD, TAN concentrations increased > 1,000 mg/L in digestate samples after the BMP test (Fig. S2b), with the highest concentration of 1446.05 ± 0.95 mg/L (p = 0.001) was observed for the digested THP sample at 170 °C. However, the digester operating conditions, methanogenic communities can be inhibited if FAN concentration is around 215-1450 mg/L 41 . In both raw and pretreated samples, FAN concentrations were < 1 mg/L (Fig. S2a). Although FAN concentrations increased in all digestate samples, the pretreated digestate samples showed considerably lower FAN levels than the control. Notably, the highest FAN of 209 ± 0.77 mg/L was observed for the digested control sample, while the lowest concentration of 169 ± 0.13 mg/L (p = 0.004) was observed for the digested THP 140 °C. As the operating temperature was the same for all conditions, estimated FAN concentrations (Fig. S2c) were correlated with TAN and pH values (Fig. S2b).

Changes in EPS and macromolecules.
The changes in EPS were characterized in terms of polysaccharides and proteins as they are considered the most dominant EPS components in sludge 42 . As shown in Fig. 1e, polysaccharides and proteins contents in sludge decreased after the THP. Polysaccharide and protein contents in the THP-80ºC sample were 173.5 ± 3.5 and 306 ± 6 mg/g sludge, which is 7 and 37%, respectively, lower than the control (i.e., raw sludge). The highest reduction of polysaccharide and protein (45 and 64%, respectively) was observed for THP-170 °C. Noticeably, EPS contents decreased gradually with increasing THP temperature. Previous studies also reported that THP could disrupt the EPS network, releasing intra-and extracellular organics in the aqueous phase 20,43 . Furthermore, FTIR analysis of solids was carried out to identify the effects of THP on functional groups associated with macromolecular compounds (Fig. S3). FTIR results also confirmed solubilization of macromolecular organics after the THP. Moreover, the gradual decrease of the absorption peaks with increasing the THP operating temperature accentuated the relationship between THP operating temperatures and solubilization efficiencies.
Both protein and polysaccharide contents in the extracted EPS from the digestate (after BMP) from THP-110 °C and THP-140ºC were less than those in control, THP-80 °C, and THP-170 °C samples (Fig. S4). Interestingly, THP-110 °C and THP-140 °C also showed higher methane production than other conditions (discussed later). Notably, a dramatic shift in the EPS composition was observed for the digestate THP-170 °C sample. That might be attributed to the microorganisms' protection mechanism that involves polysaccharides secretion to form a protective layer against the recalcitrant or inhibitory compounds commonly formed at high temperatures 22,44 . Moreover, for the THP-170ºC sample, the increased EPS contents during AD and decreased methane generation (discussed later) suggest that THP at 170ºC might form some recalcitrant/inhibitory compounds. Increasing EPS in the form of proteins can have a positive impact, as proteins can act as electron shuttles due to the exoenzyme's existence, enhancing the extracellular electron transfer and improving AD performance 44 . However, there is no evidence of such a positive impact of EPS polysaccharides 44 . Fig. 2, THP under different temperatures significantly improved the total cumulative methane yields than the control. For control, THP-80 °C, and THP-110 °C, methane production started immediately without any noticeable lag phases. In contrast, minor lag phases appeared for THP-140 °C and THP-170 °C. The estimated lag phases with the modified Gompertz model were also consistent with these experimental observations (Table S2). These results may attribute to the period that microorganisms need to adapt to the thermally pretreated sludge 37 . Particularly, high-temperature THP may release some refractory and inhibitory compounds that can extend the lag phases during AD 45 . Nonetheless, ultimately, all THP-treated samples led to higher total cumulative methane yields than the control. Despite higher lag phases than the control, THP-140 °C and THP-170 °C ultimately led to higher methane yields than the control, attributed to the higher maximum methane production rate than the control (see Table S2). The accumulated methane production increased by 20.6 ± 1.9%, 32.3 ± 1.7%, 40.5 ± 2.5% and 19.3 ± 0.2% for THP-80 °C, THP-110 °C, THP-140 °C, and THP-170 °C, respectively, compared to the control. Among the THP samples, the maximum methane yield (p = 0.03) of 275 ± 11.5 ml CH 4 /g COD was obtained for THP-140 °C, while the least methane yield (p = 0.07) of 203 ± 6.9 ml CH 4 /g COD was observed for THP-170 °C. Thus, the cumulative methane yields increased linearly with temperature increment except for THP-170 °C. The negative effect of THP at 170 °C on methanogenesis kinetics might attribute to recalcitrant compounds or toxic intermediates (e.g., melanoidins) formation 46  www.nature.com/scientificreports/ Fate of ARGs. All targeted ARG subtypes were found in raw and pretreated sludge, as well as in the digestate after AD (Fig. 3). The absolute copy number of ARGs in the initial sludge sample was 3.31 × 10 5 copies/g sludge (Fig. 3a). Compared to the raw sludge, a significant reduction (p = 0.000002-0.000009) of total ARGs was observed after THP. The ARGs copy number after THP at 80 °C, 110 °C, 140 °C, 170 °C were 1.54 × 10 5 , 9.96 × 10 4 , 1.09 × 10 5 , 1.06 × 10 5 , respectively (Fig. 3a). Thus, THP could remarkably reduce ARGs abundance prior to AD, while THP-110 °C provided the highest total ARGs removal (70%). The total ARGs also decreased in the subsequent AD except for the THP-110 °C sample (Fig. 3a). After AD, digestate from THP-140 °C and THP-170 °C showed lower ARG abundances than the digestate from control, while THP-140 °C was the most effective for overall ARG removal (79%) in the final digestate. Despite THP-110 °C being the most effective for ARG removal from before AD, the corresponding digestate sample showed an increase in ARG abundance (Fig. 3a). The increase in total ARGs for THP-110 °C sample indicates that rebounding of ARGs occurred during AD. Previous studies also reported similar ARG rebounding for thermally hydrolyzed sludge 8 . As shown in Fig. 3b, c, various ARG subtypes (e.g., tetW, tetM, tetB, tetA, and sul1) might rebound during AD. Some potential ARG host microbes (e.g., Treponema 47 , Pseudomonas 47 , Desulfotomaculum 48 ) can resist extreme environmental conditions (e.g., high temperature and pressure up) up to a certain limit during THP by forming endospores to cope with stressful conditions 49 . Thus, some host microbes might also exist in the pretreated sludge. Such a survival mechanism may happen under moderate temperature (110 °C) than high temperature (140 °C and 170 °C). When favorable conditions return, these endospores sprouts and the active bacterium is released to proliferate 50 . Thus, host microbes may proliferate in subsequent AD. Moreover, the possibility of residual DNA and horizontal gene transfer (HGT) may be a reason behind such rebounding 12 . Noteworthy, the HGT is mediated by the MGEs, such as integrons (e.g., intl1, intl2) that control the DNA movement by encoding specific proteins 51 .

Methane production. As shown in
Microbial quantity, diversity, and richness. The quantitative PCR analysis was performed for the initial (inoculum + sludge) and final digestate. Due to the pretreatment, 16S rRNA gene copies remarkably decreased from 8.71 × 10 9 gene copies/g sludge in control to as low as 3.84 × 10 6 gene copies/g sludge for THP-170 °C (Fig. 4). However, 16S rRNA gene copies increased after AD. Notably, 16S rRNA gene copies gradually increased in digestate samples from THP-80 °C to THP-140 °C. The solubilization of organics via THP led to the proliferation of microbial communities. In contrast, the formation of recalcitrant or toxic compounds 46 at 170 °C might decrease microbial propagation.
The estimated alpha diversity indices were provided in Table S3. Compared to the control, all the indices were decreased after the THP except for the Chao1 and OTUs for THP-80 °C. The highest reduction in the microbial alpha diversity was observed for THP-170 °C. For instance, Chao1, Shannon, Pielou, and observed OTUSs were reduced from 171, 6.4, 0.86, and 170 to 95, 5.1, 0.79, and 95, respectively. Thus, THP could mostly reduce microbial community diversity and richness. This result agrees with previous studies that reported that temperature is the major factor affecting microbial alpha diversity 52 . Compared to the control, digestate for THP samples exhibited higher microbial diversity and richness. This might attribute to the enhanced sludge solubilization due to THP, which subsequently enhanced microbial diversity and richness during AD 53 .
Bacterial community. Among the pretreated digested samples, dominant bacterial phyla were WWE1, Firmicutes, Chloroflexi, Bacteroidetes, and Proteobacteria (Fig. S5). Notably, members of WWE1 were the most dominant in all samples. They are known for the fermentation of sugars in AD 54 . Firmicutes are syntrophic bacteria involved in VFAs degradation 55 . Compared to the control, their relative abundance increased in digested www.nature.com/scientificreports/ THP samples. Chloroflexi species are known to hydrolyze carbohydrates 55 . Their relative abundance in control (21%) was higher than all digested THP samples. Bacteroidetes and Proteobacteria can degrade various organics, including cellulose and proteins 56 . Their relative abundances were higher in the digested THP samples than the control. Notably, their highest abundance was observed for the digested THP-170 °C sample (12 and 13%, respectively). Bacteroidetes and Proteobacteria are known as potential carriers of tetracycline resistance genes www.nature.com/scientificreports/ and other ARGs in general [57][58][59] . Thus, the highest abundance of tetracycline resistance genes observed in digestate of THP-170 °C sample was consistent with their high abundance. At the genus level (Fig. 5a), the most dominant genera were W22 (family Cloacamonaceae) and T78 (family Anaerolinaceae) in all samples. Their highest abundances were observed in the control (47% and 20%, respectively), while both showed a remarkable decrease in the digested THP samples. The members belonging to W22 (family Cloacamonaceae) were reported as syntrophic VFAs oxidizers 60 . Moreover, their potential roles in hydrolysis and acidogenesis have also been suggested 61 . The genus T78 can contribute to hydrolysis/acidification, including carbohydrates and oil organics degradation 55,62 .
Like control, W22 was still the most dominant genus (20-35%) in digested THP samples. However, their relative abundances noticeably decreased for higher THP operating temperatures (140-170 °C). Other dominant bacterial genera in digested THP samples include Bacteroides (6-8%), T78 (9-12%), Clostridium (5-9%), and Treponema (4-6%). Members belonging to the genus Clostridium and Bacteroides are obligate anaerobes and can contribute to the fermentation of organics in AD 63,64 . Among all digested THP samples, the lowest relative abundances of these bacterial genera were observed for THP-170 °C. However, the abundance of Syntrophomonas and Acidovorax increased for this condition. Thus, these results indicate that increasing temperature led to distinct differences in bacterial communities.
Archaeal community. Figure 5b shows the relative abundances of archaeal communities. The digested control sample was dominated by the genus Methanosaeta (41%), the family Methanospirillaceae (29%), followed by genera Methanoculleus (16%), and Methanobacterium (14%). The relative abundance of acetoclastic Methanosaeta 65 was reduced in all digested THP samples. In contrast, various known hydrogenotrophic methanogens (Methanoculleus, Methanospirillaceae, and Methanobacterium) were dominant in these samples. Methanoculleus was the most prevalent in the digested THP-140 °C sample (41%). Also, hydrogenotrophic Methanospirillaceae (19%) and metabolically versatile Methanosarcina (20%) were dominant in this sample. The digested THP-170 °C sample was dominated by Methanosarcina (28%), and Methanobacterium (38%). Among the digested THP samples, the relative abundances of acetoclastic Methanosaeta species were higher in the digested THP-80 °C and THP-110 °C samples. Hydrogenotrophic methanogens usually have a higher ability to resist environmental changes than acetoclastic methanogens 66 . Thus, it appeared that high-temperature THP (140 and 170 °C) might have more pronounced effects on the archaeal community distribution and methanogenesis pathways.
Multivariate analysis. The multivariate PCA was performed to evaluate the correlation between ARG abundance and bacterial communities in digested samples (Fig. 6a). For THP-110 °C and THP-80 °C, Clostridium, Bacteroides, Thermovirgaceae, and Syntrophus were closely associated with sulfonamide resistance genes. Cloacamonaceae_W22, Cloacamonaceae_W5, Treponema, and Anaerolinaceae were clustered in a different quadrant close to the control and associated with β-lactam and macrolide resistance genes. The tetracycline resistance genes, strongly correlated with integrons, were close to THP-170 °C, where Acidovorax, Paludibacter, and Syntrophomonas genera were dominant. Based on previous reports, Clostridium, Bacteroides, Treponema, Paludibacter, Syntrophomonas, Acidovorax species are potential ARG hosts [67][68][69][70] . For instance, the prevalence of macrolide resistance genes in Treponema species has already been widely reported 68 . Thus, THP played a critical role in shaping the bacterial communities and consequently changed the ARG profiles.
Interestingly, digested THP-80 °C, THP-110 °C, and THP-140 °C samples showed a similar decreasing trend for EPS (discussed earlier) and total ARGs, indicating a possible positive correlation between them. Therefore, the relationships between ARG abundances and EPS composition were further analyzed (Fig. 6b). EPS proteins, intl1, sul2, bla TEM , bla OXA , and ermB were located close to the control and THP-80 °C samples. On the other hand,   www.nature.com/scientificreports/ To further understand the relationship between ARG and EPS, correlation analysis was performed. Figure 6c shows the correlation coefficients of THP-AD, EPS components, and ARG abundances in digested sludge. Obviously, there was a positive correlation between the EPS protein component and all variables except for the THP-AD (r = − 0.76). In contrast, all the process variables showed a high negative correlation with THP-AD except for the tetracycline resistance genes and integrons (r = 0.29 and 0.30), respectively. For instance, a strong positive correlation between the EPS protein component exhibited a strong positive correlation with β-lactam resistance genes (r = 0.75), while the correlation with other ARGs, such as sulfonamides and macrolides (r = 0.17 and 0.33, respectively) was fairly weak. On the other hand, EPS polysaccharides showed a positive correlation with all ARGs, especially with tetracycline resistance genes (r = 0.87), while macrolides resistance genes were the only exception (r = − 0.31). Also, the integrons, which are considered biomarkers for ARG spread 71 , exhibited a www.nature.com/scientificreports/ very strong correlation with tetracycline resistance genes (r = 0.99). In contrast, the integrons showed a relatively weak correlation with other ARGs.

Implications.
Our results suggest that different EPS components (proteins and polysaccharides) correlate with different ARGs and MGEs. As an important component of sludge, EPS may provide ample adsorption sites for ARGs and play a critical role in their propagation 23,45,72 . Different sludge EPS components have different functional groups, such as carboxyl, phenolic, hydroxyl, etc. 73 . Thus, the adsorption of different ARGs and MGEs onto various EPS components can be different. Interestingly, the digestate THP-140 °C sample had the lowest level of proteins and polysaccharides among all digestate samples, exhibiting the lowest ARG abundance. Overall, these results infer a functional link between EPS (proteins and polysaccharides) composition and ARGs in AD of thermally hydrolyzed sludge under different temperatures. Although a recent report suggested that EPS-associated ARGs would present a most significant portion of ARGs in sludge 23 , the relationships observed in the multivariate analysis in this study remain correlational. Thus, further research should focus on the detailed characterization and changes of EPS-associated, intracellular, and cell-free ARGs under THP. Also, different layers of EPS might affect the fate and abundance of ARGs 72 , which require further investigations.

Data availability
The raw 16S rRNA sequencing data were deposited in the Sequence Read Archive (SRA) of National