Dynamics of metal(loid) resistance genes driven by succession of bacterial community during manure composting

Environmental Pollution 255 (2019) 113276

Contents lists available at ScienceDirect

Environmental Pollution journal homepage: www.elsevier.com/locate/envpol

Dynamics of metal(loid) resistance genes driven by succession of bacterial community during manure composting* Wan-Ying Xie, Xi Zou, Dong-Yang Liu, Qian Li, Qirong Shen, Fang-Jie Zhao* Jiangsu Key Laboratory for Organic Solid Waste Utilization, Jiangsu Collaborative Innovation Center for Solid Organic Waste Resource Utilization, College of Resources and Environmental Sciences, Nanjing Agricultural University, Nanjing 210095, China

a r t i c l e i n f o

a b s t r a c t

Article history: Received 17 July 2019 Received in revised form 16 September 2019 Accepted 17 September 2019 Available online 20 September 2019

Metal(loid) resistance genes (MRGs) play important roles in conferring resistance to metal(loid)s in bacterial communities. How MRGs respond to bacterial succession during manure composting remains largely unknown. Metagenomics was used in the present study to investigate the compositional changes of MRGs, their candidate hosts and association with integrons during thermophilic composting of chicken manures. MRGs conferring resistance to 20 metal(loid)s were detected, and their diversity and abundance (normalized to the abundance of 16S rRNA genes) were significantly reduced during composting. MRGs associated with integron were exclusively observed in proteobacterial species. Class 1 integron likely played an important role in maintaining mercury-resistance mer operon genes in composts. Escherichia coli harbored the most abundant MRGs in the original composting material, whereas species of Actinobacteria and Bacilli became more important in carrying MRGs during the late phases. There were significant linear relationships between the relative abundance of some specific bacterial species (E. coli, Actinobacteria species and Enterococcus faecium) and the abundance of MRGs they potentially harbored. The succession of these bacteria contributed to an overall linear regression between the relative abundance of all predicted candidate hosts and the abundance of total MRGs. Our results suggest that the succession of bacterial community was the main driver of MRG dynamics during thermophilic composting. © 2019 Elsevier Ltd. All rights reserved.

Keywords: Metal(loid) resistance genes Succession Bacterial community Animal manure Thermophilic composting Integron

1. Introduction Animal wastes often contain elevated levels of heavy metals and metalloids. Metals such as copper (Cu), zinc (Zn), cadmium (Cd), chromium (Cr), vanadium (V), nickel (Ni) and molybdenum (Mo) and metalloids such as arsenic (As) have long been utilized as nutritional supplements in animal feeds (Pal et al., 2014; Yu et al., 2017). In order to ensure their nutrition efficacies, these metals and metalloids are often supplemented with concentrations much higher than the actual requirements by the animals (Ao and Pierce, 2013; Medardus et al., 2014; Yu et al., 2017). In addition, some of the metals, especially Cu and Zn, are used at high concentrations as alternatives of antibiotics for the purposes of growth promotion and feed efficiency improvement (Rensing et al., 2018; Thacker, 2013). Large proportions of these elements are excreted into manures (Medardus et al., 2014). The metals or metalloids in the * This paper has been recommended for acceptance by Prof. Dr. Klaus Kümmerer. * Corresponding author. E-mail address: [email protected] (F.-J. Zhao).

https://doi.org/10.1016/j.envpol.2019.113276 0269-7491/© 2019 Elsevier Ltd. All rights reserved.

animal intestines can exert a significant selective pressure on the microbial community and generate diverse profiles of metal(loid) resistance genes (MRGs), which are transferred to the manures (Medardus et al., 2014). Direct land applications of animal manures introduce large amounts of metal(loid)s as well as MRGs into soil environments, which can exert adverse impacts on soil microbial communities and potentially affect biogeochemical processes mediated by microbes (Berg et al., 2010; Luo et al., 2009). In addition, due to the tendency of MRGs to associate with antibiotic resistance genes (ARGs) (Pal et al., 2015), the transmission of MRGs could increase the survival of antibiotic resistant bacteria in soils, particularly those contaminated with metals and metalloids. Composting of animal manures prior to land application is an efficient and economic pathway to reduce manure volume, pathogens, antibiotics, organic contaminations, and many other hazardous factors (Bernal et al., 2009; Mitchell et al., 2015; Qian et al., 2016; Wang et al., 2015). During composting, changes in temperature, nutrient concentrations and water content are coupled with a succession of microbial community (Koyama et al., 2018; Meng et al., 2019; Qiao et al., 2019; Su et al., 2015). These successions of

2

W.-Y. Xie et al. / Environmental Pollution 255 (2019) 113276

microbial communities have been suggested to cause compositional changes of antibiotic resistance genes (ARGs) during composting of waste materials (Liao et al., 2018; Su et al., 2015). In contrast to antibiotics, which were substantially degraded during composting (Wang et al., 2015), metal(loid)s are persistent and may exert a prolonged selective effect on MRGs. However, to our knowledge, there have been few studies investigating the dynamics of MRG profile and the relationship with the succession of microbial community during composting. Of particular interest is the association of MRGs with integrons, the gene acquisition systems in bacteria, which can facilitate the assembly of multiple resistance genes on the same genetic elements. The objective of the present study was to investigate the dynamics of MRGs and their association with integrons during the process of thermophilic composting. A bioinformatic approach based on metagenomic sequencing was employed to reveal microbial community succession, the dynamics of MRG abundances and their association with integrons. Candidate hosts of MRGs were predicted by annotating the genes to bacterial species. Abundances of MRGs were compared with the relative abundances of the corresponding candidate hosts at different taxonomic levels to identify hosts with high capacities to carry MRGs. 2. Methods and materials 2.1. Thermophilic composting and sampling Thermophilic composting was conducted with 5 tons of fresh chicken manures and 1.5 tons of rice chaff at a composting plant. The initial materials were mixed by a mechanical mixer and piled into a windrow shape of approximately 5.0 m
1.5 m
1.0 m. The pile was longitudinally divided into three sections (replicate C1, C2 and C3; separated by two unsampled sections, each around 1-m long). The day of mixing was designated as Day 0, and the composting lasted for 55 days, during which 23 days (from Day 5 to Day 27) had a temperature above 55  C (Fig. S1). Samples were taken on Day 0, 4, 21 and 28 in triplicate to represent the initial material, mesophilic phase, thermophilic phase and cooling phase of the composting, respectively. The samples were transported in ice boxes to the laboratory as soon as possible and stored at 80  C until further analysis. 2.2. Determination of metals and metalloids Fresh samples were freeze-dried, ground and sieved (0.15-mm) before analysis. The concentrations of metals and metalloids were determined by inductively coupled plasma mass spectrometry (PerkinElmer NexION 300X, USA) after the samples were digested with HNO3 and HClO4 (85:15, V: V) (Yang et al., 2017). The concentration of Hg was determined by an atomic fluorescence spectrometer (Haiguang 230E, Beijing, China) after a digestion process with aqua regia (HCl:HNO3 ¼ 4:1, V:V) (Yang et al., 2017). A certificated standard material of soil (GBW07428, Institute of Geophysical and Geochemical Exploration, China) and reagent blanks were digested and determined along with the samples for quality assurance. 2.3. DNA extraction and purification DNA from 5.0 g fresh samples was extracted using PowerMax Soil DNA Isolation Kits (MoBio Laboratories, Carlsbad, CA) with the default instructions. DNA extracts were purified with phenol/ chloroform/isoamyl alcohol and concentrated with sodium acetate as described previously (Xie et al., 2018). Quality of DNA (A260/A280 and A260/A230) was measured by a NanoDrop 2000C

spectrophotometer (Thermo Scientific, Wilmington, USA). A260/ A280 and A260/A230 of the purified DNA were all above 1.8 and 1.7, respectively. A Quant-iT PicoGreen double-stranded DNA assay kit (Invitrogen, USA) was employed to determine the DNA concentration. The purified DNA was stored at 20  C until sequencing. 2.4. Metagenomic sequencing and bioinformatics analysis The DNA samples were sequenced at Biozeron (Shanghai, China) using an Illumina Hiseq X system. The sequencing generated 2
150 bp pair-end reads. On average, each of the 12 samples contained 6.9 Gb metagenomic data. The raw data are available at the NCBI Sequence Read Archive (SRP132747). Clean reads were obtained after a process of quality control of the reads by removing those containing more than three ambiguous nucleotides, with sequence length shorter than 50 bp or with an average quality value lower than 20 (Xiao et al., 2016). The clean reads were assembled into contigs using SOAPdenova2 (Version 2.01, Kmer 3947). MetaGene (http://metagene.cb.k.u-tokyo.ac.jp/) was used to predict the open reading frame (ORFs) on contigs longer than 500 bp. A non-redundant gene catalog was generated with the longest representative genes from the gene clustering of all samples (cdhit, 95% identity, 90% coverage). Taxonomy of each gene was annotated by BLASTp (E value  1e5) searching the gene catalog against the NCBI non-redundant protein (NR) database with MEGAN (Community Edition, Version 6.15.0) with an implementation of the Lowest Common Ancestor (LCA) algorithm (Huson et al., 2016). MEGAN has been widely used in taxonomic annotation of microbial communities as well as in host-tracking of specific genes, such as ARGs (Cernava et al., 2019; Franzosa et al., 2018; Hu et al., 2013; Ma et al., 2017; Mahnert et al., 2019). Bacterial taxonomic composition in each sample and the candidate hosts of MRGs were obtained based on the above annotation. MRGs were identified by searching the gene catalog against the manually curated database of BacMet (version 2.0), which contained MRGs with experimentally confirmed functions (753 individual genes), using the following criteria: E value  1e5, hit length  25 amino acids and similarity  90% (Luo et al., 2017; Pal et al., 2014). Counts of MRGs were calculated by blasting the clean reads in each sample against the gene catalog (SOAPaligner, 95% identity). The counts of 16S rRNA genes were calculated by BLASTn (E value  1e5) searching the clean reads against SILVA database. The abundances of MRGs were normalized to 16S rRNA genes with their corresponding counts according to a previous study (Yang et al., 2016). Integron marker genes were identified through BLASTn (E value  1e5, hit length  50 bp and similarity  90%) searching the gene catalog against the database of INTEGRALL. The contigs containing integron marker genes were then searched for MRGs as above to identify integron-associated MRGs. 2.5. Statistical analysis Heatmaps were constructed in R (version 3.3.1) with package of “pheatmap”. Regression analyses were conducted in SigmaPlot 12.0 with positive data that had been log10-transformed to achieve normality distribution of the data (Shapiro-Wilk test, p > 0.05) in at least ten samples. Comparisons and significance tests were performed based on log10-transformed data using ANOVA (Duncan test) or t-test at the probability level of 0.05 in SPSS. 3. Results 3.1. Changes in the concentrations of elements during composting The concentrations of Fe, Pb, Cr, Co, Ni, As, Cd and Hg on a dry

W.-Y. Xie et al. / Environmental Pollution 255 (2019) 113276

3

weight basis showed a significant (P < 0.05) increasing trend during the thermophilic composting (Fig. 1, Table S1), and this could be attributed to the decomposition of organic matter. The concentrations of P, K, Zn and Cu showed no such trend, but rather a slight decrease after the composting, possibly due to the high heterogeneity of these elements in the samples. The variation of elemental concentrations in the original material was much higher than that in the composts, especially for P, K, Cu and Co (Fig. 1, Table S1). The pH in the sample significantly increased during composting (Table S1). 3.2. Diversity and abundance of MRGs during composting

Fig. 1. Concentrations of elements during thermophilic composting. Units for the concentrations of P, K, Zn, Cu are g/kg (DW) and Fe are g/kg dry weight (DW). Units for the concentrations of the other elements are mg/kg DW.

The total abundance and diversity (Shannon index) of MRGs exhibited a sharp decrease during the thermophilic phase (Fig. 2a). During the cooling phase, the diversity remained at a low level, whereas the total abundance increased significantly (P < 0.05) to show some extent of abundance recovery. Accordingly, there was significant dissimilarity between samples at the early phases (original material and mesophilic phase) and those at the late phases (thermophilic phase and cooling phase) in terms of MRG composition (Fig. 2b). MRGs conferring resistance to 17 metals and 3 metalloids (As, Se and Te) were detected in the original material and composting samples (Fig. 2c). Genes conferring resistance to Zn, Te, Cu, Se, Cr and Cd were the most abundant in the original material (Fig. 2c). ruvB and merR that confer resistance to Cr and Hg, respectively, were the two most prominent individual MRGs in the original material (Table 1). A majority of MRGs exhibited significant decreases in the abundance between Day 4 and Day 21, and remained

Fig. 2. Changes in the diversity and abundance of metal(loid) resistance genes (MRGs) during thermophilic composting. (a) Shannon index of the MRG profile and the total abundance during composting. Letters above the bars and under the scatters indicate results from multiple comparisons for Shannon index and total abundance, respectively. (b) Principal coordinates analysis based on the Bray-Curtis dissimilarity between the compositions of MRGs at different composting time. (c) Abundance of MRGs classified according to the related metal(loid)s during the composting. Comparisons were based on log10-transformed abundances.

4

W.-Y. Xie et al. / Environmental Pollution 255 (2019) 113276

Table 1 Information on the most abundant MRGs in samples of Day 0. Gene_name

Resistance

Average abundance-Day 0

Candidate host

BacMet ID

yhcN corC mgtA ruvB cusC/ylcB fetA/ybbL fetB/ybbM merR corA mntP/yebN modE nikB tehA tehB zntR/yhdM znuA/yebL zur/yjbK zraS/hydG

Cd Co Mg Co Mg Cr Cu Ag Fe Fe Hg Mg Co Ni Mn Mn Mg Mo W Ni Te Te Zn Zn Zn Zn Pb

6.42E-04 5.98E-04 6.00E-04 1.37E-02 7.65E-04 7.53E-04 7.27E-04 1.61E-03 6.37E-04 6.15E-04 6.99E-04 6.68E-04 6.35E-04 6.65E-04 6.21E-04 6.26E-04 6.18E-04 6.75E-04

Escherichia coli Escherichia coli Escherichia coli Pseudomonas stutzeri Escherichia coli Escherichia coli Escherichia coli Escherichia coli Escherichia coli Escherichia coli Escherichia coli Escherichia coli Escherichia coli Escherichia coli Escherichia coli Escherichia coli Escherichia coli Escherichia coli

BAC0446 BAC0088 BAC0087 BAC0293 BAC0109 BAC0165 BAC0166 BAC0232 BAC0086 BAC0252 BAC0608 BAC0271 BAC0384 BAC0385 BAC0462 BAC0463 BAC0470 BAC0468

undetectable on Day 28 (Fig. 2c). Abundances of some MRGs recovered on Day 28, especially those conferring resistance to Hg (merA, merR and merT), Cu (dnaK) and Fe (ideR) (Fig. 2c, Table 2). No significant correlation was observed between the concentrations of metal(loid)s and the abundance or diversity of their respective MRGs (Data not shown). 3.3. Candidate hosts of MRGs during composting The bacterial composition displayed a substantial succession during the thermophilic composting (Table 3). High temperature during thermophilic phase eliminated most of the bacterial members in the phyla of Bacteroidetes and Proteobacteria present in the original material, whereas the members in the phyla of Firmicutes and Actinobacteria increased significantly in the composts of Day 21 and Day 28 (Table 3). Candidate hosts of MRGs were predicated to bacterial species using MEGAN. Based on their annotation in bacterial species, the distribution of MRGs in taxonomic composition from the levels of genus to phylum were summarized (Fig. 3). The main candidate hosts of MRGs in terms of abundance shifted significantly during the composting process. Gammaproteobacteria harboured the most abundant MRGs in the samples from the early phases (Day 0 and 4), while Actinobacteria (class level), Bacilli and Gammaproteobacteria were the dominant candidate hosts in the samples from the late phases (Day 21 and 28) (Fig. 3b). Proteobacteria, especially Gammaproteobacteria, harbored the most diverse MRGs, including 89 different types of MRGs that confer resistance to 19 metal(loid)s (Fig. 3a, Table S2). Enterobacteriaceae and Pseudomonadaceae were the major families in Gammaproteobacteria that hosted the most abundant MRGs (Fig. 3c). At the genus level, Escherichia (mainly E. coli) and Pseudomonas (mainly Pseudomonas stutzeri) were the main contributors in the above two families, respectively (Fig. 3d). The majority of Proteobacteria-harboring MRGs were predicted in E. coli, encompassing 83 genes that encode resistance to 19 metal(loid)s (Fig. 3e, Table S2), even though their relative abundances were only 0.02%e 0.87% during the composting. Species in Pseudomonas were predicted to contain only genes involved in resistance towards Cr (ruvB, the most abundant gene in the original material) and Hg (merA and merP) (Table S2). Four species in Actinobacteria that have been predicted as candidate hosts contained only two MRGs, dnaK and ideR (members of the most abundant genes on Day 28) that confer resistance to Cu and Fe, respectively (Fig. 3e, Table 2,

Association with integron marker genes integron (In323) integron (In323) integron (In1067)

integron (In717) integron (In323) integron (In323)

integron (In323) integron (In323) integron (In323)

Table S2). Five MRGs were detected in Firmicutes, including cadC (conferring resistance to Cd, Bi, Zn and Pb; potentially carried by Staphylococcus sciuri), tcrB (conferring resistance to Cu), merA, merR and merT (all three conferring resistance to Hg; potentially carried by Staphylococcus aureus). Among the five genes, tcrB was prevalent as the sole MRG across five Firmicute species, including Alkalibacterium olivapovliticus, Dolosigranulum pigrum, Trichococcus ilyis, Enterococcus faecium and Lachnoanaerobaculum saburreum (Table S2). All the annotated candidate hosts at the species level, except E. coli, contained only one to three MRGs (Fig. 3e). Although Bacteroidetes accounted for half of the bacterial community in the original material (Table 3), no MRG was annotated to the species in this phylum. The abundances of MRGs predicted in Actinobacteria and Proteobacteria generally changed in accordance with the relative abundances of their corresponding phyla. However, MRGs predicted in Firmicutes showed distinct dynamic patterns with their corresponding candidate hosts at all taxonomic levels. These results led to different abundance correlations between MRGs and their candidate hosts in different bacteria taxonomies (Fig. 3k). There were significant linear relationships between the abundances of Actinobacteria-harbored MRGs (R2 ¼ 0.68, P ¼ 0.006) and Proteobacteria-harbored MRGs (R2 ¼ 0.92, P < 0.0001) and their corresponding candidate hosts. One prominent example was E. colicarrying mer operon genes (merA, merT and merR) showing a significant linear regression with the relative abundance of this species (R2 ¼ 0.69, P ¼ 0.002). Firmicutes-harbored MRGs showed no such significant regression with their candidate hosts at the phylum level (Fig. 3k) However, one species in Firmicutes, Enterococcus faecium, showed significant linear regression with tcrB that was predicted to be carried by the species (R2 ¼ 0.73, P ¼ 0.002). The dominance of Proteobacteria-harbored MRGs in the whole profile contributed to a similar regression pattern on the basis of all MRGharboring taxonomies (R2 ¼ 0.92, P < 0.0001). As major members of Proteobacteria, E. coli, Enterobacteriaceae and Gammaproteobacteria all exhibited regression patterns similar to that of the phylum. The slope of the regression is indicative of the capacity of gene hosting at the taxonomic level. Based on this criteria, E. coli (slope ¼ 1.32), which was dominant in MRG-harboring Enterobacteriaceae (slope ¼ 1.34), Gammaproteobacteria (slope ¼ 1.40) and Proteobacteria (slope ¼ 1.40), showed a much higher capacity to contain MRGs than the species of Actinobacteria (slope ¼ 0.61).

W.-Y. Xie et al. / Environmental Pollution 255 (2019) 113276

5

Table 2 Information on the most abundant MRGs in samples of Day 28. Gene_name

Resistance

Average abundance-Day 28

Species/Family of candidate host

BacMet ID

Association with integron marker genes

dnaK dnaK dnaK ideR merA merR merT merA merR merT

Cu Cu Cu Fe Hg Hg Hg Hg Hg Hg

5.62E-04 3.26E-04 6.12E-05 4.04E-04 4.00E-04 4.26E-04 4.19E-04 4.37E-04 5.35E-04 6.81E-04

Mycobacterium duvalii/Mycobacteriaceae Dietzia cinnamea/Dietziaceae Lechevalieria aerocolonigenes/Pseudonocardiaceae Mycobacterium sp./Mycobacteriaceae Staphylococcus aureus/Staphylococcaceae Staphylococcus aureus/Staphylococcaceae Staphylococcus aureus/Staphylococcaceae Escherichia coli/Enterobacteriaceae Klebsiella pneumoniae/Enterobacteriaceae Escherichia coli/Enterobacteriaceae

BAC0133 BAC0133 BAC0133 BAC0498 BAC0651 BAC0680 BAC0696 BAC0652 BAC0232 BAC0693

integron (In715) integron (In715) integron (In715)

Table 3 Relative abundances (%) of the main phyla during thermophilic composting. Phylum

Composting time (d) 0

4

21

28

Firmicutes Bacteroidetes Proteobacteria Actinobacteria Other

20.7 ± 1.2 c 51.3 ± 0.5 a 24.6 ± 1.7 a 1.8 ± 0.1 c 1.6 ± 0.2 b

65.1 ± 5.8 b 11.0 ± 2.0 b 19.5 ± 3.8 b 2.8 ± 0.3 c 1.7 ± 0.3 b

83.2 ± 5.4 a 0.6 ± 0.1 c 2.7 ± 0.3 d 12.6 ± 5.4 b 1.0 ± 0.1 b

60.4 ± 5.4 b 2.1 ± 0.4 c 8.8 ± 1.4 c 25.0 ± 3.3 a 3.7 ± 0.9 a

Data are compared among sampling points using ANOVA (Duncan test) at the probability level of 0.05 in SPSS 18. Significant differences are represented alphabetically where “a” stands for the highest value in each comparison.

3.4. MRGs and integron marker genes during composting More than half of MRGs (57.7% of the total abundance) were associated with integron marker genes in the samples on Day 0. The percentage decreased to 29.5% on Day 21 before a small recovery to 31.5% on Day 28 (Fig. 4). All the detected integron marker genes that associated with MRGs were carried by class 1 integrons. Members in Gammaproteobacteria, except for Escherichia fergusonii and Stenotrophomonas sp., were all observed to carry integronassociated MRGs (Fig. 3e). The most abundant genes ruvB and merR in the original material were associated with In1067 and In717, respectively. Other integron-associated MRGs with a high abundance in the original material were associated with In323 (Table 1). Integron-associated MRGs in the samples of Day 28 were those conferring resistance to Hg (merA, merR and merT), showing a close relation to In715 (Table 2). No MRGs annotated to Actinobacteria or Firmicutes were observed to be associated with integron marker genes. The aforementioned mer operon genes (merA, merR and merT) were all suggested by BacMet to locate on the plasmids.

4. Discussion In this study, the concentrations of metals and arsenic in the original composting materials fell within the median ranges found in recent surveys of animal manures and composts in China (Wang et al., 2013; Yang et al., 2017). Therefore, the levels of metals and metalloids in our study can be considered to be representative of the manure composts in China. How the presence of metals and metalloids in the composting materials impacts MRG diversity and abundance has not received much attention in the past. Our study reveals diverse MRGs conferring resistance to tens of metal(loid)s in the original composting materials, indicating that fresh animal manure is an important reservoir of MRGs (Fig. 2c). Direct application of fresh animal manure might introduce a large amount of MRGs into soils and potentially affect the resistome of metal(loid)s in them. Due to the immobilization of metal(loid)s by composting

(García et al., 1990; Greenway and Song, 2002), the selective pressure from metal(loid)s on MRGs might have been limited during the composting process. The pH increased significantly as the composting progressed (Table S1), which could have decreased the availability of cationic metals. This would explain the lack of a significant correlation between the abundance or diversity of MRGs and the concentrations of metal(loid)s in the present study. The thermophilic composting in the present study generally resulted in increased concentrations of metal(loid)s due to decomposition of easily degradable organic matter (García et al., 1990). Despite general increases in the concentrations of metal(loid)s, the diversity and abundance of MRGs and their association with integrons were found to be significantly decreased by the composting process. This result suggests that thermophilic composting is an effective pathway for mitigation of MRGs from animal origin that could pose a potential risk of transmission to the general environments. Bacterial species belonging to different phyla showed distinct abilities to host MRGs (Fig. 3k). The high ability of Proteobacteria to harbor MRGs can be mainly attributed to E. coli. Other candidate hosts showed no significant difference in the ability to carry MRGs, with only one to three genes in each species (Fig. 3). Species in Proteobacteria, especially those of clinical importance, such as Escherichia, Shigella, Klebsiella and Enterobacter, have been found to contain highly abundant MRGs in their genomes based on a survey of fully sequenced bacterial genomes and plasmids (Pal et al., 2015). The broad range of MRGs carried by E. coli might result from the diversity and plasticity of this species' genome, which appears to be constantly expanding through horizontal gene transfer (Blount, 2015). This was further supported by the finding that a considerable portion of the most abundant E. coli-harbored MRGs were associated with the same class 1 integrons that could effectively gather multiple resistance genes into nearby locations (Table 1, Table 2). All the integron-associated MRGs were exclusively distributed in the species of Proteobacteria (Fig. 3e). Consistent with our findings, a preferential distribution of class 1 integrons in Proteobacteria, particularly in Gammaproteobacteria, was uncovered by assessing the collections of bacterial genomes and plasmids (Zhang et al., 2018). However, it should be pointed out that only a few entries of MRGs from Bacteroidetes were included in the experimentally confirmed database of BacMet, which might have missed out some MRGs in this phylum. In addition, the assembly process could have excluded some of MRG-harboring reads of different bacterial origins, including some Bacteroidetes. More studies are needed to expand the database and to increase the assembly efficiency in order to provide relatively unbiased interpretation in the future. Using MEGAN, it was possible to annotate MRGs into specific bacterial species (Ma et al., 2017). The results clearly showed that the succession of bacterial community was the main factor driving the dynamics of the overall profile of MRGs (Fig. 3). During the

6

W.-Y. Xie et al. / Environmental Pollution 255 (2019) 113276

Fig. 3. Abundances of MRGs and their corresponding candidate hosts. (a)e(e) MRGs with taxonomic annotation at different level. The numbers at the right side of (e) indicate the gene number annotated to each species. (f)e(j) MRG-harboring taxa at different level. Taxonomic composition at different levels were summarized based on the species annotation by MEGAN. The black lines indicate borders between different phyla. The red stars indicate association of MRGs with integrons. (k) Correlation analysis between relative abundance of taxonomy and abundance of MRGs. The symbol for Proteobacteria is totally shaded by that of Gammaproteobacteria. The lines indicate significant (P ¼ 0.006 for Actinobacteria, P < 0.0001 for other taxa) linear regressions. Data points for Actinobacteria on Day 0 that contained no annotated MRGs were not included in the analysis. The “Total” is a summary of Actinobacteria, Firmicutes and Proteobacteria. For a better presentation, abundance of MRGs and relative abundance of MRG-harboring taxa were multiplied by 108 and log 10 transformed. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

thermophilic phase, the relative abundances of host species in the Proteobacteria phyla decreased significantly (Fig. 3j), resulting in decreased abundance of MRGs harbored by Proteobacteria (Fig. 3e). This trend was also observed for the integron-associated MRGs (Fig. 4). During the cooling phase, the abundance of MRGs was partially maintained by four species (Dietzia cinnamea, Mycobacterium sp, Mycobacterium duvalii and Lechevalieria aerocolonigenes; harboring dnaK and ideR) in Actinobacteria and Staphylococcus aureus (harboring merA, merT and merR) in Firmicutes, which all exhibited high relative abundance on Day 28 or displayed an increase during composting (Fig. 3e and j, Table 2). MRGs predicted to be harbored by Escherichia coli and Klebsiella pneumoniae

significantly increased from Day 21 to Day 28 (Fig. 3e), even though the relative abundances of these two species did not increase significantly (Fig. 3j). These increasing MRGs are some mer operon genes encoding resistance to mercury via reduction of Hg2þ to Hg (Table 2), with merA, merT and merR encoding a mercuric reductase, a membrane-bound protein and a regulator protein, respectively (Matsui and Endo, 2018). Mercury resistance genes are often associated with integrons and carried by plasmids (Pal et al., 2015), as was also observed in the present study (Table 2). Such association could help maintain the abundance of mer operon genes in Escherichia coli and Klebsiella pneumoniae when the compost cooled down. Escherichia coli and Klebsiella pneumoniae grow best at the

W.-Y. Xie et al. / Environmental Pollution 255 (2019) 113276

7

References

Fig. 4. Abundance and percentage of MRGs associated with integron marker genes during thermophilic composting. Integron: MRGs associated with integrons (E value  1e5, hit length  50 bp and similarity  90%). Other: MRGs not associated with integrons (E value > 1e5, hit length < 50 bp or similarity < 90%).

temperature range between 18  C and 46  C (Tsuji et al., 1982). The metabolic activity of integron integrase and/or conjugative plasmids could have revived and facilitated the replication of the mer operon genes in Escherichia coli and Klebsiella pneumoniae when the fitness of the two species recovered to some extent at a temperature under 50  C (Fig. S1). Compared to other MRGs, the mer operon genes are more likely to co-occur with integron marker genes and ARGs on bacterial plasmids (Pal et al., 2015). How this association affects their transferring ability to bacteria in the general environment remains to be investigated.

5. Conclusions Distinct bacterial succession was observed during thermophilic composting of animal manures. This served as an ideal model for investigating the relationships among MRGs, integrons and the bacterial community. The succession of bacterial community was the main driver of the dynamics of MRGs during the composting process. Association of MRGs with Class 1 integron likely played an important role in maintaining MRGs, especially mer operon genes, in the compost during the cooling phase. E. coli was observed as the dominant candidate host of MRGs in the compost during the early phases. Thermophilic composting significantly reduced the relative abundance of E. coli and other species in Proteobacteria, resulting in decreased abundances of MRGs, including those associated with integrons.

Acknowledgement This research was supported by “the Fundamental Research Funds for the Central Universities” (KJQN201669, KYZ201518), National Natural Science Foundation of China (41501260), and the Innovative Research Team Development Plan of the Ministry of Education of China (Grant No. IRT_17R56).

Appendix A. Supplementary data Supplementary data to this article can be found online at https://doi.org/10.1016/j.envpol.2019.113276.

Ao, T., Pierce, J., 2013. The replacement of inorganic mineral salts with mineral proteinates in poultry diets. World Poult. Sci. J. 69, 5e16. Berg, J., Thorsen, M.K., Holm, P.E., Jensen, J., Nybroe, O., Brandt, K.K., 2010. Cu exposure under field conditions coselects for antibiotic resistance as determined by a novel cultivation-independent bacterial community tolerance assay. Environ. Sci. Technol. 44, 8824e8728. Bernal, M.P., Alburquerque, J.A., Moral, R., 2009. Composting of animal manures and chemical criteria for compost maturity assessment: a review. Bioresour. Technol. 100, 5444e5453. Blount, Z.D., 2015. The unexhausted potential of E. Coli. eLife 4, e05826. Cernava, T., Erlacher, A., Soh, J., Sensen, C.W., Grube, M., Berg, G., 2019. Enterobacteriaceae dominate the core microbiome and contribute to the resistome of arugula (Eruca sativa Mill.). Microbiome 7, 13. Franzosa, E.A., McIver, L.J., Rahnavard, G., Thompson, L.R., Schirmer, M., Weingart, G., Lipson, K.S., Knight, R., Caporaso, J.G., Segata, N., Huttenhower, C., 2018. Species-level functional profiling of metagenomes and metatranscriptomes. Nat. Methods 15, 962e968. García, C., Hern andez, T., Costa, F., 1990. The influence of composting and maturation processes on the heavy-metal extractability from some organic wastes. Biol. Wastes 31, 291e301. Greenway, G.M., Song, Q.J., 2002. Heavy metal speciation in the composting process. J. Environ. Monit. 4, 300e305. Hu, Y., Yang, X., Qin, J., Lu, N., Cheng, G., Wu, N., Pan, Y., Li, J., Zhu, L., Wang, X., Meng, Z., Zhao, F., Liu, D., Ma, J., Qin, N., Xiang, C., Xiao, Y., Li, L., Yang, H., Wang, J., Yang, R., Gao, G.F., Wang, J., Zhu, B., 2013. Metagenome-wide analysis of antibiotic resistance genes in a large cohort of human gut microbiota. Nat. Commun. 4, 2151. Huson, D.H., Beier, S., Flade, I., Gorska, A., El-Hadidi, M., Mitra, S., Ruscheweyh, H.J., Tappu, R., 2016. MEGAN community edition - interactive exploration and analysis of large-scale microbiome sequencing data. PLoS Comput. Biol. 12, e1004957. Koyama, M., Nagao, N., Syukri, F., Rahim, A.A., Kamarudin, M.S., Toda, T., Mitsuhashi, T., Nakasaki, K., 2018. Effect of temperature on thermophilic composting of aquaculture sludge: NH3 recovery, nitrogen mass balance, and microbial community dynamics. Bioresour. Technol. 265, 207e213. Liao, H., Lu, X., Rensing, C., Friman, V.P., Geisen, S., Chen, Z., Yu, Z., Wei, Z., Zhou, S., Zhu, Y.G., 2018. Hyperthermophilic composting accelerates the removal of antibiotic resistance genes and mobile genetic elements in sewage sludge. Environ. Sci. Technol. 52, 266e276. Luo, G., Li, B., Li, L.G., Zhang, T., Angelidaki, I., 2017. Antibiotic resistance genes and correlations with microbial community and metal resistance genes in full-scale biogas reactors as revealed by metagenomic analysis. Environ. Sci. Technol. 51, 4069e4080. Luo, L., Ma, Y., Zhang, S., Wei, D., Zhu, Y.G., 2009. An inventory of trace element inputs to agricultural soils in China. J. Environ. Manag. 90, 2524e2530. Ma, L., Li, B., Jiang, X.T., Wang, Y.L., Xia, Y., Li, A.D., Zhang, T., 2017. Catalogue of antibiotic resistome and host-tracking in drinking water deciphered by a large scale survey. Microbiome 5, 154. Mahnert, A., Moissl-Eichinger, C., Zojer, M., Bogumil, D., Mizrahi, I., Rattei, T., Martinez, J.L., Berg, G., 2019. Man-made microbial resistances in built environments. Nat. Commun. 10, 1e12. Matsui, K., Endo, G., 2018. Mercury bioremediation by mercury resistance transposon-mediated in situ molecular breeding. Appl. Microbiol. Biotechnol. 102, 3037e3048. Medardus, J.J., Molla, B.Z., Nicol, M., Morrow, W.M., Rajala-Schultz, P.J., Kazwala, R., Gebreyes, W.A., 2014. In-feed use of heavy metal micronutrients in U.S. swine production systems and its role in persistence of multidrug-resistant salmonellae. Appl. Environ. Microbiol. 80, 2317e2325. Meng, Q., Xu, X., Zhang, W., Men, M., Xu, B., Deng, L., Bello, A., Jiang, X., Sheng, S., Wu, X., 2019. Bacterial community succession in dairy manure composting with a static composting technique. Can. J. Microbiol. 65, 436e449. Mitchell, S.M., Ullman, J.L., Bary, A., Cogger, C.G., Teel, A.L., Watts, R.J., 2015. Antibiotic degradation during thermophilic composting. Water Air Soil Pollut. 226, 13. Pal, C., Bengtsson-Palme, J., Kristiansson, E., Larsson, D.G.J., 2015. Co-occurrence of resistance genes to antibiotics, biocides and metals reveals novel insights into their co-selection potential. BMC Genomics 16. Pal, C., Bengtsson-Palme, J., Rensing, C., Kristiansson, E., Larsson, D.G., 2014. BacMet: antibacterial biocide and metal resistance genes database. Nucleic Acids Res. 42, D737eD743. Qian, X., Sun, W., Gu, J., Wang, X.J., Zhang, Y.J., Duan, M.L., Li, H.C., Zhang, R.R., 2016. Reducing antibiotic resistance genes, integrons, and pathogens in dairy manure by continuous thermophilic composting. Bioresour. Technol. 220, 425e432. Qiao, C., Ryan Penton, C., Liu, C., Shen, Z., Ou, Y., Liu, Z., Xu, X., Li, R., Shen, Q., 2019. Key extracellular enzymes triggered high-efficiency composting associated with bacterial community succession. Bioresour. Technol. 288, 121576. Rensing, C., Moodley, A., Cavaco, L.M., Mcdevitt, S.F., 2018. Resistance to metals used in agricultural production. Microbiol. Spectr. 6. ARBA-0025-2017. Su, J.Q., Wei, B., Ou-Yang, W.-Y., Huang, F.Y., Zhao, Y., Xu, H.-J., Zhu, Y.G., 2015. Antibiotic resistome and its association with bacterial communities during sewage sludge composting. Environ. Sci. Technol. 49, 7356e7363. Thacker, P.A., 2013. Alternatives to antibiotics as growth promoters for use in swine

8

W.-Y. Xie et al. / Environmental Pollution 255 (2019) 113276

production: a review. J. Anim. Sci. Biotechnol. 4, 35. Tsuji, A., Kaneko, Y., Takahashi, K., Ogawa, M., Goto, S., 1982. The effects of temperature and pH on the growth of eight enteric and nine glucose non-fermenting species of gram-negative rods. Microbiol. Immunol. 26, 15e24. Wang, H., Dong, Y., Yang, Y., Toor, G.S., Zhang, X., 2013. Changes in heavy metal contents in animal feeds and manures in an intensive animal production region of China. J. Environ. Sci. 25, 2435e2442. Wang, J., Ben, W., Zhang, Y., Yang, M., Qiang, Z., 2015. Effects of thermophilic composting on oxytetracycline, sulfamethazine, and their corresponding resistance genes in swine manure. Environ. Sci. Proc. Impacts 17, 1654e1660. Xiao, K.Q., Li, B., Ma, L., Bao, P., Zhou, X., Zhang, T., Zhu, Y.G., 2016. Metagenomic profiles of antibiotic resistance genes in paddy soils from south China. FEMS Microbiol. Ecol. 92, fiw023. Xie, W.-Y., Yuan, S.-T., Xu, M.-G., Yang, X.-P., Shen, Q.-R., Zhang, W.-W., Su, J.-Q., Zhao, F.-J., 2018. Long-term effects of manure and chemical fertilizers on soil

antibiotic resistome. Soil Biol. Biochem. 122, 111e119. Yang, X., Li, Q., Tang, Z., Zhang, W., Yu, G., Shen, Q., Zhao, F.J., 2017. Heavy metal concentrations and arsenic speciation in animal manure composts in China. Waste Manag. 64, 333e339. Yang, Y., Jiang, X., Chai, B., Ma, L., Li, B., Zhang, A., Cole, J.R., Tiedje, J.M., Zhang, T., 2016. ARGs-OAP: online analysis pipeline for antibiotic resistance genes detection from metagenomic data using an integrated structured ARG-database. Bioinformatics 32, 2346e2351. Yu, Z., Gunn, L., Wall, P., Fanning, S., 2017. Antimicrobial resistance and its association with tolerance to heavy metals in agriculture production. Food Microbiol. 64, 23e32. Zhang, A.N., Li, L.G., Ma, L., Gillings, M.R., Tiedje, J.M., Zhang, T., 2018. Conserved phylogenetic distribution and limited antibiotic resistance of class 1 integrons revealed by assessing the bacterial genome and plasmid collection. Microbiome 6, 130.