Edinburgh Research Explorer Effects of micron-scale zero valent iron on behaviors of antibiotic resistance genes and pathogens in thermophilic anaerobic digestion of waste activated sludge

: 9 This work investigated the metagenomics-based behavior and risk of antibiotic resistance 10 genes (ARGs), and their potential hosts during thermophilic anaerobic digestion (TAD) 11 of waste activated sludge, enhanced by micron-scale zero valent iron (mZVI). Tests were 12 conducted with 0, 25, 100, and 250 mg mZVI/g total solids (TS). Results showed that up 13 to 7.3% and 4.8% decrease in ARGs’ abundance and diversity, respectively, were 14 achieved with 100 mg mZVI/g TS. At these conditions, ARGs with health risk in 15 abundance and human pathogenic bacteria (HPB) diversity were also decreased by 8.3% 16 and 3.6%, respectively. Additionally, mZVI reduced abundance of 72 potential 17 pathogenic supercarriers for ARGs at high health risk by 2.5%, 5.0%, and 6.1%, as its 18 dosage increased. Overall, mZVI especially at 100 mg/g TS can mitigate antibiotic 19 resistance risk in TAD. These findings are important for better understanding risks of 20 ARGs and their pathogenic hosts in ZVI-enhanced TAD of solid wastes. 21


Introduction
Antibiotic resistance encoded in antibiotic resistance genes (ARGs) is becoming a daunting threat to global "One Health".It is generally accepted that waste activated sludge (WAS) is an important hotspot of multiple and abundant ARGs (Zhang et al., 2022b).Up to 181 ARG subtypes, mainly resistant to multidrug, macrolide-lincosamidestreptogramin (MLS), bacitracin, sulfonamides, and tetracycline, with the highest abundance being 1.57 × 10 -1 copies per 16S rRNA gene have been detected in WAS in sewage treatment plants (Yoo et al., 2020).Alarmingly, some of them, even at low abundance, could pose great clinical significance and transmission to the environment via WAS treatment and disposal (Zhang et al., 2022c), which can eventually put human health at risk (Bondarczuk et al., 2016).Moreover, human pathogenic bacteria (HPB), such as Escherichia coli, Vibrio cholerae, Streptococcus pneumoniae, have also been revealed in WAS (Cai & Zhang, 2013;Zhang et al., 2022a).Some of these bacteria can develop multiple antibiotic resistances and are recognized as ARGs' supercarriers (Jang et al., 2019), thus acting as another factor aggravating the antibiotic resistance crisis (Lin et al., 2022).These thereby indicate the pressing need to comprehensively understand the health risk of ARGs in the WAS treatment process in addition to ARGs' abundance and types, if the current prevailed antibiotic resistance crisis is to be mitigated.Thermophilic anaerobic digestion (TAD) has been proven to be a promising technology for WAS reduction, energization, and pollution control (Gao et al., 2017).Meanwhile, as previously reported, TAD can create a favorable environment to eliminate the abundance of ARGs (Diehl & Lapara, 2010), but there is no consistent conclusion, due to the different WAS properties and surveillance approaches that have been studied (Jang et al., 2019;Tian et al., 2016;Zhang et al., 2015).Moreover, WAS with high microbial and chemical densities can facilitate the spread of ARGs in TAD processes (Zhang et al., 2011).Many reports have also demonstrated that the microbial community and physicochemical properties of WAS can vary and shift the behavior of ARGs and their hosts during anaerobic digestion (AD).These shifts are often triggered by the changes in AD conditions, such as temperature (Xu et al., 2020), or the presence of exogenous substances (Jang & Kan, 2022;Pang et al., 2022), which are commonly applied to enhance the performance of TAD.Therefore, there is a risk that ARGs and their hosts would be affected by these TAD enhancers, but this has been barely studied until now.
Recently, micron zero valent iron (mZVI) has attracted widespread interest as a costeffective agent in improving methane production in AD of food waste (Jing et al., 2022), swine manure (Yang et al., 2018b), and glucose-substrate (Zhong et al., 2022).Also, it has been observed that mZVI performed well in eliminating ARGs in various AD applications, such as wastewater treatment (Xu et al., 2021b), co-digestion of waste sludge and kitchen waste (Gao et al., 2017), and swine manure treatment (Zhang et al., 2018).However, current investigations on the effect of ZVI on the overall abundance and diversity of ARGs in sludge AD process have mainly focused on nano-sized ZVI (nZVI).Yet, considering the different effects on physiochemical properties and microbial composition induced by ZVI with various particle sizes (Xu et al., 2021a;Zhong et al., 2022), the role of mZVI in the TAD process regarding the entire antibiotic resistome in WAS has not been fully explored.
This work aims to fill this knowledge gap by following a metagenomic approach.Firstly, a full picture of the ARGs profile and microbial community in WAS before and after TAD with different mZVI dosages will be provided.Second, the behavior and risk of ARGs will be elucidated with regards to abundance, diversity, and human health-based impacts using current assessment frameworks (Zhang et al., 2022c), as well as prioritizing HPB with or without mZVI.Finally, the potential pathogenic hosts for high risk ARGs will be identified to explore the antibiotic resistance risk in WAS in the presence of mZVI.
Findings of this work will be fundamentally important to promote an understanding of the potential effect of mZVI on TAD treatment of solid wastes containing ARGs and opportunistic pathogens.

Substrate and additive
WAS collected from a municipal sewage treatment plant in Chengdu, China, was settled at 4°C before use.Its physicochemical properties are listed in Table 1.mZVI (150 μm, 99% metals basis) was purchased from Macklin Reagent Co. Ltd., China.

Experimental design
Four batch experimental groups spiked with 0, 25, 100, and 250 mg mZVI/g TS and marked as Z0, Z1, Z2, and Z3, respectively, were digested for 32-day in serum bottles with working volume of 200 mL.WAS with 8% (w/v) TS were added in each serum bottle, then sealed with rubber stoppers and incubated at 55(±1)°C in a water bath after being flushed with high-purity nitrogen for 2 min.WAS samples collected at the end of TAD experiments in Z0, Z1, Z2, and Z3 groups were centrifuged at 8,000 rpm for 10 min.From these, the supernatant was filtered through a 0.45 μm membrane for measuring the physicochemical properties, and sediments along with raw WAS (marked as Raw) were stored at -80°C for metagenomic analysis.Gas samples were collected using air bags (300 mL) for measuring daily and cumulative biogas and methane productions.Duplicate experiments (n=2) were also carried out in parallel.

DNA extraction and metagenomic sequencing
DNA from raw WAS and sediment samples from the experimental groups were extracted using Fast DNA Spin Kit for Soil (MP Biomedicals, USA).DNA from the same operational conditions were mixed for one sequencing sample.Genomic DNA was then detected by 1% agarose gel electrophoresis.

Antibiotic resistance genes annotation, normalization, and risk assessment
ARGs were annotated against the comprehensive antibiotic resistance database (CARD, v 3.0.9)using Diamond (http://www.diamondsearch.org/index.php,v 0.8.35).The abundance (coverage, times per Giga base, ×/Gb) of ARGs was defined using Eq. ( 1) (Xiong et al., 2018): where n is the number of the annotated ARG-like ORFs belonging to that ARG type or subtype; Ni (reads) is the number of the reads mapped to the ARG-like ORFs; Li (reads) is the length of the Illumina sequencing reads (bp); L (ARG-like ORFs) is the length of the target ARG-like ORFs (bp); S is the size of the sequencing data after quality control (Gb).
The health risk of annotated ARGs was evaluated based on the database provided by Zhang et al. (2022c), which includes four ranked risks for 2561 ARG subtypes considering their human accessibility, mobility, pathogenicity, and clinical availability.
Thereinto, the ARGs at risk were classified into 4 levels, with Q1 ranked as the highest risk, followed by Q2, Q3, and Q4.

Microbial characterization and human pathogenic bacteria identification
Microbial composition was characterized by NR database (v 20200604) using BLASTP via Diamond software (http://ab.inf.uni-tuebingen.de/software/diamond/).The prioritized HPB were identified according to A-to-Z database from the National Infection Prevention and Control Manual.

Statistical analysis
SPSS 21.0 and Origin 2022b were used for statistical analysis and processing.Significant differences (p < 0.05) between groups were assessed by Fisher's exact test.The correlation was analyzed by Spearman linear correlation (R>0.85,p<0.05).Network visualization was conducted on the platform of Gephi (v 0.9.7).

Availability of data
The metagenomics data was deposited into the National Center for Biotechnology Information (NCBI) sequence read archive database (SRA) (Accession Number: PRJNA914103).

Performance of thermophilic anaerobic digestion (TAD)
Biogas and methane production, and characteristics of WAS before (Raw) and after TAD treatment with different dosages of mZVI are listed in Table 1.These results indicated that TAD performances varied significantly between groups without (Z0) and with mZVI (Z1, Z2, and Z3) in terms of methane (biogas) production, SCOD, SC, NH4 + -N, and VFAs content.For example, cumulative methane yield in groups with mZVI (Z1, Z2, and Z3) was 1.7, 12.0, and 12.0 times higher than in Z0, respectively.Moreover, in groups with mZVI, concentrations of iso-butyric acid and pH were positively correlated with the dosage of mZVI.In addition, Z2 and Z3 performed similarly but better than Z1 in methane and biogas production, and this could be confirmed by the observed decrease in concentrations of SCOD, SC, acetic and butyric acids, as well as substrates for methane and biogas yield, in Z2 and Z3.Overall, an up to 12-fold increase of methane yield enhanced by mZVI was obtained in this study.This improvement was higher than those (35.9%-44.5% promotions) observed previously in mesophilic AD.This might be attributed to the fact that although mZVI could enhance the release of substrates for methane production, and the activity of key enzymes involved in the AD process (Jing et al., 2022;Liang et al., 2021), the substrate properties and operational temperature would also affect the AD performance (Xu et al., 2020).
ARG types of multidrug, glycopeptide, tetracycline, and MLS were prevalent in TAD treated sludge with and without mZVI.This is consistent with previous findings in sewage treatment plants (Li et al., 2015).This might be attributed to the fact that these antibiotics were widely used in human activities (Singh et al., 2019).Multidrug resistance genes were the most abundant in this study possibly due to the dominant resistance mechanism of antibiotic efflux, which could provide defense against several inhibitory constituents, including multidrug resistance (Christgen et al., 2015).
The total abundance of ARGs were 12329, 14962, 14592, 13949, and 14465 ×/Gb in raw WAS and the four TAD experimental groups (Fig. 1b).It can be observed that TAD could enrich ARG abundance no matter whether mZVI is present or not, and the highest enrichment occurred in Z0.Although many studies demonstrated that TAD performs better in driving a sharp decline in the total abundance of ARGs ( The addition of mZVI could mitigate the enrichment by TAD, except ARGs in types of nucleoside and fusidic acid.Notably, mZVI at 100 mg/g TS proved best in mitigating the risk of ARGs by reducing their abundance, since the abundances of 15 ARG types were decreased to the lowest level in Z2.Furthermore, Venn analysis (Fig. 1c) showed that 621 ARG subtypes were shared among raw and the four treated WAS, with the number of respective unique ARG subtypes being 52, 9, 7, 3, and 4. The diversity of ARG subtypes in raw WAS could be reduced by TAD treatment from 800 to 761 in Z0, and this reduction was further improved to 747, 726, and 728 when the mZVI dosages increased and so were the unique ARG subtypes.
[Figure 1] The presence of mZVI, especially at 100 mg/g TS (Z2), was found effective in improving methane production and in mitigating ARGs risk by total abundance and diversity of ARGs in the TAD process (Z0), and this was dosage independent.Such performance was consistent with previous findings that have demonstrated a correlation between a decrease in ARGs' abundance and an increase in methane production (Lu et al., 2022a).Besides, current studies on the response behavior of ARGs to mZVI in TAD are mainly focused on certain types of antibiotic resistance.For example, regarding the widely studied tetracycline-resistant genes, qPCR results showed that ZVI (40 μm) at 60 g/L were beneficial for eliminating tetA, tetG, tetM, tetO, tetC, and tetX but without great differences from those at 5 g/L in TAD (50°C) of WAS and kitchen waste (Gao et al., 2017).In this study, similar reductions, compared with Z0, in the abundance of 14 tetracycline ARGs including tetX were also achieved by mZVI (Z1-Z3).Meanwhile, the total abundance of other 46 genes resistant to tetracycline was reduced by 0.5% and 3.3% in Z1 and Z2, while being slightly enriched in Z3.This could be explained by the release of ferrous iron at certain concentrations, like 5.5 μg/mL in Z2 in this study.This effect can significantly alter the dynamics of functional genes, including ARGs, and exert various selection pressures on them, thereby shifting their abundance and composition

Health risk assessment of antibiotic resistance genes
The public health concern was prioritized in this study by focusing on clinically relevant ARGs and their potential to transfer to humans, thereby driving the evolution of antibiotic-resistant pathogens (Zhang et al., 2022c).350 ARG subtypes were identified at health risk (see supplementary material), and the ARG subtypes in Q1 (highest risk), Q2, Q3, and Q4 were 128, 68, 62, and 92, occupying 50.9%, 12.8%, 13.5%, and 22.8%, respectively, in total abundance of 43311 ×/Gb (Fig. 2).In consistence with the behavior of all ARGs, the abundance of these ARGs at risk was also increased across the TAD process from 7751 ×/Gb (raw WAS) to 9238 ×/Gb (Z0), which dropped to 8971, 8533, and 8819 ×/Gb, respectively in Z1, Z2, and Z3.While, the diversity of risky ARGs in raw WAS (328) decreased in TAD (Z0, 324), and was further reduced in the presence of mZVI (319 in Z1, 321 in Z2, and 309 in Z3).This suggested that the abundance and diversity of all ARGs at risk increased in Z0, while it was reduced in mZVI-enhanced TAD (Z1-Z3) with the best performance observed at 100 mg mZVI/g TS (Z2).
Regarding the 128 ARG subtypes at high risk (Q1), they could be classified mainly in genes resistant to multidrug (74), tetracycline (16), and aminoglycoside (13).Their abundance in raw WAS (4042 ×/Gb, 123 subtypes) was also enriched in Z0 (4677 ×/Gb, 123 subtypes), and mZVI could decrease this to 4530 (120 subtypes), 4319 (125 subtypes), and 4496 ×/Gb (125 subtypes), respectively.This indicated that mZVI at all three concentrations could further mitigate the risk of ARGs in Q1 by abundance.However, only the addition of 25 mg mZVI/g TS improved their diversity reduction.These results suggest that mZVI exerts an overall positive effect in relieving the abundance and diversity of ARGs at risk, but for those at high risk it depends on the dosage of mZVI (Xu et al., 2021a).
It should be noted that 350 ARGs at risk were revealed including 134 multidrug resistance genes, 74 of which were found in 128 high risk ARGs (Q1), accounting for the majority of ARGs at (high) risk.Such high health risk induced by ARGs for multidrug could be due to the fact that they can render multiple antibiotics, even those with clinical importance, inefficient and/or invalid, while further triggering severe mortality (Christgen et al., 2015).
[Figure 3] The top three dominant phyla (relative abundance > 1%) in the four experimental groups were the same but with different relative abundances (Fig. 3g).It can be seen that, compared to raw WAS, the TAD treatment (Z0) increased these three dominant phyla significantly, and reduced the abundances of phyla Bacteroidetes (from 13.9% to 1.6%), Nitrospirae (from 4.6% to 1.2%), and Acidobacteria (from 3.7% to 1.1%).This reduction was further enhanced by mZVI (Z1-Z3).However, the presence of mZVI showed the potential to enrich phylum Chloroflexi compared with raw WAS and Z0, and its dosage was found to be positively correlated with its relative abundances.
The relative abundances in total of 27 classified genera (with total relative abundance > 0.3%) increased after the TAD treatment (from 15.8% in Raw to 29.3% in Z0), but fluctuated slightly with mZVI (27.8%, 31.2%, and 29.4% in Z1, Z2, and Z3, respectively) (Fig. 3h).Among these 27 genera, 4 of them showed a decrease in relative abundance after the TAD process, while Nitrospira and Candidatus_Competibacter decreased further as the dosage of mZVI increased.23 genera were enriched in Z0 compared to raw WAS.Interestingly, only 4 genera (Candidatus_Promineofilum, Nocardioides, Marmoricola, and Microbacterium) were further increased with mZVI, while 9 genera like Dechloromonas and Candidatus_Accumulibacter were reduced by mZVI, especially at 100 mg /g TS.
Their relative abundances were enriched in TAD (Z0), but only the first genus could be further enriched in the presence of mZVI regardless of its dosage.As previously reported, Dechloromonas has the ability to degrade organic matters via electron transfer (Yang et al., 2015), and Coprothermobacter can hydrolyze protein and take part in thermophilic syntrophic interaction with hydrogenotrophic methanogenic archaea (Gagliano et al., 2014).Also, Candidatus_Promineofilum is capable of generating energy by respiration or by carbohydrate fermentation (like sugar) in AD (Tandishabo et al., 2012).Additionally, Methanothermobacter, as the dominant thermophilic hydrogenotrophic methanogen, and Methanosarcina, as the acetoclastic methanogen (Barros et al., 2017), were enhanced by the application of mZVI.These indicated that mZVI could enhance the hydrogenotrophic and acetoclastic methanogenic pathway probably via promoting hydrolysis and fermentation of carbohydrates.
[Figure 4] Interestingly, mZVI (Z1-Z3) could decrease the relative abundance of superbugs, such as Pseudomonas aeruginosa, Klebsiella pneumoniae MDR, and Acinetobacter baumannii in TAD (Z0), while not being effective in eliminating Enterococcus faecalis.Such positive inhibitory effects on superbugs, like Pseudomonas aeruginosa, were also found in nZVI application (Anbouhi et al., 2019), probably due to its particle size and its role on the cell wall of Gram-negative bacteria.Furthermore, the observed increase in the relative abundance of Enterococcus faecalis in this study could be attributed to the hydrogen-producing capacity of the genus Enterococcus (Yang & Wang, 2022), which have been enhanced by mZVI (Jing et al., 2022).

Relationships between human pathogenic bacteria and antibiotic resistance genes
In addition to the health risk induced by ARGs themselves, their bacterial hosts, especially the pathogenic ones, can trigger the clinical inefficacy or invalidity of antibiotics, thereby posing a threat to human health.Potential hosts at the genus level with relative abundance in total > 0.3%, as predominant genera, were explored for 128 high risk ARGs (Q1) (Fig. 4b).The co-occurrence network consisted of 131 nodes and 588 edges, including 27 genera and 104 ARGs in Q1.All 27 genera were significantly positively correlated with 90 high risk ARGs, with 4 genera of Tetrasphaera, Coprothermobacter, Nitrospira, and Candidatus_Competibacter, potentially carrying at least 20 ARGs in Q1.Furthermore, genus Pseudomonas, a known HPB species, was found as the potential host for 18 ARGs in Q1, like kdpE, efrA, bacA, YojI, mgrA, and aadA5.Moreover, special attention was paid to how HPB can host potentially high risk ARGs.
The co-occurrence pattern between 98 HPB detected and ARGs in Q1, focusing on their positive correlations, was analyzed to determine possible twin risk from pathogens and antibiotic resistance (Fig. 4c).The entire network consisting of 221 nodes and 687 edges, showed that 95 HPB species had significant positive correlations with 126 ARGs at high risk.This indicates serious risks induced by the existence of either ARGs or HPB or both in raw WAS and TAD treated sludge with and without mZVI.Such ARG-carrying HPB have also been uncovered in AD systems of sludge (Ju et al., 2022), the mixture of livestock manure and fruit wastes (Lin et al., 2022), and dairy manure (Jang & Kan, 2022).
Of the 95 potential pathogenic hosts, the majority belonged to 35 genera within the phylum Proteobacteria (49 species) and 6 genera within the phylum Firmicutes (20 species), both of which were also reported as potential hosts for ARGs (Pang et al., 2022), and acted as opportunistic pathogens (Jang et al., 2019;Lin et al., 2022).Besides, these two phyla play a vital role in hydrolytic acidification as the core community (Zhao et al., 2022).This may be the reason why pathogens were enriched in the TAD process.The abundance of HPB affiliated with these two phyla increased from 72.2% in raw WAS to 78.5% in Z0, but was mitigated by mZVI treatment.In addition, this alleviation was enhanced as dosages of mZVI increased (77.1%, 76.0%, 74.6% in Z1, Z2, Z3).These suggest that mZVI can relieve the propagation risk of ARGs, and this effect was dosage dependent.
Thereinto, 15 ARGs had significant positive correlations with at least 10 HPB, and 10 of these 15 ARGs were resistant to multidrug.Remarkably, 72 species were recognized as potential supercarriers for at least 3 ARGs, and their relative abundance was increased in TAD treatment (from 73.0% in Raw to 79.2% in Z0), and reduced with mZVI (77.2%, 75.2%, and 74.4% in Z1-Z3, respectively).This implies the dosage-dependent mitigation effect of mZVI on the risk of possible simultaneous pathogenicity and antibiotic resistance in TAD.Besides, Acinetobacter baumannii, Chlamydia trachomatis, Francisella tularensis, and Providencia stuartii co-occurred with at least 20 ARGs at high risk, respectively.
These genes were mainly resistant to multidrug (29), tetracycline (9), aminoglycoside (3), and MLS (3).Similar superbug hosts were also revealed in the report through metagenome-assembled genomes (Zhang et al., 2022c).For example, Klebsiella pneumoniae MDR was recognized as the host for acrB, adeF, which is consistent with the findings of this study.Enterococcus faecalis could acquire IsaA, whereas it was found to co-occur with ErmA.The inconsistency may be because mZVI/TAD changed the ARGs' hosts.Therefore, an in-depth investigation should be carried out to explore how ARGs at high risk can be transferred among HPB, especially superbugs, and even to humans during the TAD treatment of WAS regarding mitigation of their environmental and health risks.

Conclusion
The comprehensive profiles of ARGs in terms of abundance, diversity, health risk, and potential pathogenic hosts in WAS before and after the TAD treatment with different mZVI dosages were investigated.Results showed that mZVI, especially at 100 mg/g TS could mitigate the risk of ARGs by abundance and diversity in TAD.Such effect was also observed in abundance of ARGs at (high) health risk and diversity of HPB.Finally, the increased abundance in TAD of 72 HPB, as supercarriers for ARGs at high health risk,    predominant genera in the top 5 phyla, and relative abundance distribution of (g) all phyla and (h) genera in raw sludge (Raw) and sludge treated by thermophilic anaerobic digestion with mZVI at 0 (Z0), 25 (Z1), 100 (Z2), 250 (Z3) mg/g TS.Others: (a-g) phylum and genus (in its corresponding phylum) with relative abundance  1.0% and (h) genus with relative abundance  0.3%.Unclassified: (a-g) unclassified phylum and genus (in its corresponding phylum) with relative abundance > 1.0% and (h) unclassified genus with relative abundance > 0.3%.N.A.: not available; TS: total solids; COD: chemical oxygen demand.
Diehl & Lapara, 2010;Jang et al., 2019;Tian et al., 2016), this was not the case in this study, since only a decrease in diversity was obtained.This phenomenon might be attributed to the fact that the fate and profile of ARGs could vary due to the different AD conditions, substrates properties, and applied surveillance approaches, like qPCR, which also limited the understanding of the change of ARGs' diversity in AD process(Xu et al., 2020; Zhang et     al., 2011).

(
Gao et al., 2017;Lu et al., 2022a).Notably, some ARGs rebounded with mZVI in the TAD treatment compared to those in raw WAS, such as APH(3')-IIb, OXA-1, tet(E), and PER-2, and this was probably due to the fact that mZVI can perform well in directly affecting ARGs themselves instead of inactivating their hosts(Zhang et al., 2020).

Figure Captions Figure 1 (
Figure Captions

Figure 3
Figure 3 Overview profiles of microbial (a) composition at the phylum level with (b-f)

Figure 4
Figure 4 (a) Composition of human pathogenic bacteria (top 30 in average relative