Antibiotic Resistome Alteration by Different Disinfection Strategies in a

Subscriber access provided by Gothenburg University Library

Ecotoxicology and Human Environmental Health

Antibiotic resistome alteration by different disinfection strategies in a fullscale drinking water treatment plant deciphered by metagenomic assembly Huaicheng Zhang, Fangyu Chang, PENG SHI, Lin Ye, Qing Zhou, Yang Pan, and Ai-Min Li Environ. Sci. Technol., Just Accepted Manuscript • DOI: 10.1021/acs.est.8b05907 • Publication Date (Web): 23 Jan 2019 Downloaded from http://pubs.acs.org on January 24, 2019

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 43

Environmental Science & Technology

1

Title: Antibiotic resistome alteration by different disinfection strategies in a full-scale

2

drinking water treatment plant deciphered by metagenomic assembly

3

Authors:

4

Huaicheng Zhang, Fangyu Chang, Peng Shi,* Lin Ye, Qing Zhou, Yang Pan, Aimin Li

5

Affiliations of authors:

6

State Key Laboratory of Pollution Control and Resource Reuse, School of the

7

Environment, Nanjing University, Nanjing 210023, China.

8

*Corresponding

9

Addresses:

author

10

State Key Laboratory of Pollution Control and Resource Reuse, School of the

11

Environment, Nanjing University, 163 Xianlin Avenue, Nanjing 210023, China.

12

Phone: +86-25-89680507

13

Fax: +86-25-89680507

14

Email: [email protected]

15 16 17 18 19 20 21

1

ACS Paragon Plus Environment


22

Abstract

23

Disinfection regimes are considered the most solid strategy to reduce microbial risks

24

in drinking water, but their roles in shaping the antibiotic resistome are poorly

25

understood. This study revealed the alteration of antibiotic resistance genes (ARGs)

26

profiles, their co-occurrence with mobile genetic elements (MGEs) and potential hosts

27

during drinking water disinfection based on metagenomic assembly. We found the

28

ozone/chlorine (O3/Cl2) coupled disinfection significantly increased the relative

29

abundance of ARGs and MGE–carrying antibiotic resistance contigs (ARCs) through

30

the enrichment of ARGs within the resistance–nodulation–cell division and

31

ATP-binding cassette antibiotic efflux families that are primarily carried by

32

Pseudomonas, Acinetobacter, Mycobacterium and Methylocystis, whereas the

33

antimicrobial resin/chlorine coupled disinfection posed unremarkable changes to the

34

ARG and MGE abundances. Moreover, the co-occurrence patterns of antibiotic efflux

35

and beta-lactam ARGs and MGEs were widely identified, ARCs carrying the recR

36

and mexH genes were detected in all the samples, with the highest abundance of 2.25

37

× 10-2 copies per cell after O3/Cl2 disinfection. Sequence-independent binning

38

analysis successfully retrieved two draft ARG-carrying genomes of Acidovorax sp.

39

MR-S7 and Hydrogenophaga sp. IBVHS2, further revealing the host-ARG

40

relationship during O3/Cl2 disinfection. Overall, this study provides novel insights into

41

the antibiotic resistome alteration during drinking water disinfection.

2


Page 2 of 43

Page 3 of 43


42

Keywords: antibiotic resistance genes, mobile genetic elements, drinking water,

43

disinfection strategy, high-throughput sequencing

44 45

INTRODUCTION

46

Antibiotic resistance has become a challenging problem in clinical medicine

47

worldwide, whereas new antibiotic development and usage is always far behind the

48

evolution of antibiotic-resistant bacteria (ARB) and the rapid spread of antibiotic

49

resistance genes (ARGs).1 Drinking water has been regarded as an important medium

50

to propagate ARB and ARGs between the environment and humans,2-4 demonstrating

51

the necessity and urgency of blocking their transmission. Disinfection is the most

52

solid barrier that provides reliable physico-chemical removal of microorganisms in

53

drinking water.5 However, increasing work has recently documented the unintended

54

effects of many disinfection approaches on the drinking water microbiome.6,7

55

Disinfection processes for drinking water mainly include chlorination, ozonation,

56

ultraviolet radiation, antimicrobial resin (AR) treatment and their combinations, and

57

usually act as the last barrier for reducing potential pathogens in drinking water

58

treatment plants (DWTPs). Among these disinfection treatment, chlorination is a

59

widely used disinfection approach in DWTPs worldwide, but mounting evidence has

60

demonstrated its enrichment effects on several types of ARGs among drinking water

61

microbiomes.8-10 The safety of ultraviolet (UV) disinfection has been well

62

acknowledged due to fewer disinfection byproducts,11 but there are still doubts about

3



63

its bacterial inactivation capacity in drinking water.12 Several studies have suggested

64

that to damage ARGs requires much greater UV doses than to inactivate ARB,13,14

65

and total ARG abundances in wastewater were elevated after UV disinfection.15 To

66

date, the fates of ARGs after UV disinfection are rarely reported in full-scale DWTPs.

67

Moreover, previous studies also observed increased levels of some ARGs after ozone

68

disinfection in DWTPs,16 and that higher doses of ozone required for ARG

69

elimination.17 AR with quaternary ammonium salt moieties such as trimethylamine

70

hydrochloride is able to interact with bacterial membranes through electrostatic

71

attractions and then physically damages the bacterial morphology and thus induces

72

cell death.18 However, its effects on the antibiotic resistome in drinking water have

73

not yet been deeply explored. Overall, there is growing evidence to suggest that the

74

antibiotic resistome alteration occurs and is closely associated with survival of

75

bacteria during disinfection treatments,19,20 which this study further explores.

76

Disinfectants often share the same mechanism of resistance with many

77

antibiotics,21,22 and different ARGs within the same resistance mechanism may

78

consistently respond to disinfection, so the dynamic variation of whole ARG

79

responses to disinfection cannot be completely revealed if they are categorized

80

according to antibiotic type. Furthermore, mobile genetic elements (MGEs) are

81

hypothesized to play vital roles in shaping the antibiotic resistome during drinking

82

water disinfection.23 Host and co-occurring MGEs have been proven to influence the

83

antibiotic resistome,24 and this probably results in different fates of the same ARGs

4


Page 4 of 43

Page 5 of 43


84

during drinking water disinfection. Moreover, Resfams has been specifically

85

developed for ARG identification according to antibiotic resistance mechanism and

86

ontology, which can provide higher annotation sensitivity and resolution than

87

previous ARG databases.25 Metagenomic assembly combined with Resfams

88

annotation has the advantage of improving the annotation accuracy of ARGs, the

89

resolution of host identification and the determination of ARG co-occurrence, which

90

have been successfully applied in antibiotic resistome analysis of the gut microbes,26

91

animal manure27 and sediment.28

92

In the present study, we selected a steady-operation DWTP that simultaneously

93

operates disinfection units of chlorine, ozone, UV and AR in order to investigate the

94

antibiotic resistome alteration under different disinfection strategies. Metagenomic

95

assembly and Resfams annotation were jointly utilized to uncover the co-occurrence

96

patterns of ARGs and MGEs, as well as their bacterial hosts. To our knowledge, this

97

study is the first to comprehensively explore the effects of multiple disinfection

98

strategies on the antibiotic resistome in a full-scale DWTP based on same source

99

water and may provide novel insights leading to the further reduction of human health

100

risks induced by the antibiotic resistome in drinking water.

101 102

MATERIALS AND METHODS

103

Sampling Locations in DWTP. The full-scale DWTP is located at Yancheng City,

104

Jiangsu Province, China, which is simultaneously equipped with three sets of

5



Page 6 of 43

105

disinfection strategies. As shown in Figure S1, the first strategy includes ozone and

106

chlorine (O3/Cl2) disinfection units, and another two strategies couple AR with

107

chlorine (AR/Cl2) and UV (AR/UV) disinfection units. Specifically, the AR used in

108

this study is a typical type of magnetic polymer with polyacrylic acid matrixes and

109

quaternary ammonium salt groups, and its physiochemical parameters and application

110

were reported in a previous study.29 In brief, the AR treatment in drinking water has

111

realized engineering application in China and other countries. Detailed information

112

and operational parameters of the DWTP are provided in Table S1. In detail, 800 L of

113

source

114

chlorine-disinfected water (O-CW), AR-disinfected water (ARW), AR coupled with

115

chlorine-disinfected water (AR-CW) and AR coupled with UV-disinfected water

116

(AR-UVW) were collected from the different disinfection strategies in August 2016

117

according to our previously established method.8 Sampling campaigns were

118

performed in triplicate in each location using water filters and lasted for 24 h to

119

control for the temporal variation of influent water. A total of 18 water filters were

120

transported to the laboratory at 4 °C within 4 h for further experiments. The water

121

quality of each sample is presented in Table S2, which were measured following the

122

standard methods.30

123

DNA Extraction and Illumina High-throughput Sequencing (HTS). Bacterial

124

suspension of each water sample was isolated from the water filter according to our

125

previous method,8 and then subjected to DNA extraction using the FastDNA SPIN Kit

water

(SW),

ozone-disinfected

water

(OW),

6


ozone

coupled

with

Page 7 of 43


126

for Soil (MP Biomedicals, CA) according to the manufacturer’s instructions. The

127

concentration and purity of DNA were quantified using microspectrophotometry

128

(NanoDrop ND−2000, NanoDrop Technologies, Wilmington, DE), and the quality

129

was determined by 1% agarose gel electrophoresis. The quantified DNA

130

(approximately 6 μg) for each sample was subjected to the 350-bp library construction

131

(Nextera DNA library Preparation Kit), and Illumina HTS with paired-end sequencing

132

strategy (2 × 150 bp) was performed on an Illumina HiSeq 4000 platform (Illumina

133

Inc., San Diego, CA) in the Novogene Sequencing Company (Beijing, China). The

134

metagenomic data were deposited into the Sequence Read Archive (SRA) database of

135

the National Center for Biotechnology Information (NCBI) under accession number

136

SUB2036733. Low-quality reads were filtered to achieve better results in the

137

downstream analysis by using Trimmomatic (Version 0.36) following a previous

138

study.24 Finally, approximately 8.70−12.50 Gb high-quality paired-end reads were

139

obtained for each sample with an average length of 150 bp (Table S3).

140

Metagenomic Assembly and Gene Prediction. Clean reads of each sample were de

141

novo assembled into contigs using the CLC Genomics Workbench (Version 9.0.1,

142

CLC Bio., Aarhus, Denmark) with the k−mer size of 63.24 A total of 151,624–219,429

143

contigs for each sample were generated with N50 values of 1,063–2,033 (Table S4),

144

and then open reading frames (ORFs) on the assembled contigs of each sample were

145

predicted using metagenome version of Prodigal (version 2.6.3). Finally, 271,619–

146

410,328 ORFs were obtained for each sample (Table S4).

7



Page 8 of 43

147

ARGs and MGEs Identification and Quantification. The protein sequences of

148

predicted ORFs were aligned against the Resfams core genes database

149

(http://dantaslab.wustl.edu/resfams) using HMMSCAN (version 3.0) with the

150

recommended parameters to identify the ARG-like ORFs,25 which are defined as the

151

ARGs in this study. A contig was annotated as an antibiotic resistance contig (ARC)

152

if at least one ARG was located on it. Then, the protein sequences of all ORFs on

153

each ARC were searched against the nonredundant (NR) protein sequence database

154

(retrieved on December, 2016) of NCBI using BLASTP with e-value cutoff ≤ 10-5.24

155

One ORF was regarded as an MGE if one of the following keywords was in its best

156

BLAST hit description: transposase, transposon, conjugative, conjugal, integrase,

157

integron,

158

co-occurrence of ARGs and MGEs was identified if they were both located on the

159

same contig.

160 161

plasmid,

recombinase,

or

mobilization.31,32

The

The abundance was defined as the copies of ARG/MGE/ARC per cell (hereafter abbreviated as cpc) and calculated by the following equation:33 𝑛

162

recombination

Abundance =

∑ 𝑖=1

𝑁𝑖mapped reads × 𝐿𝑖read/𝐿𝑖ORF 𝑁16S sequence × 𝐿𝑖read/𝐿16S sequence

× 𝑁16𝑆 𝑐𝑜𝑝𝑦 𝑛𝑢𝑚𝑏𝑒𝑟

163

where n is the number of ARG- or MGE-like ORFs or ARCs belonging to the same

164

category; Nimapped reads is the number of reads mapped to ARG- or MGE-like ORFs or

165

ARCs; 𝐿𝑖ORF is the sequence length of the corresponding target ARG- or MGE-like

166

ORF or ARC sequences; Nimapped

167

Genomics Workbench; 𝐿𝑖read is the read length (150 bp); 𝑁16S sequence is the total

reads

and 𝐿𝑖ORF are both calculated by the CLC

8


Page 9 of 43


168

number of the 16S rRNA sequences in each metagenomic dataset compared to the

169

Greengenes database;34 L16S sequence is 1432 bp in this study and stands for the average

170

length of 16S rRNA gene in Greengenes;35 N16S copy number is the average copy number

171

of 16S rRNA genes per cell number in each metagenomic dataset, which is calculated

172

by DIAMOND (v1.09) with 30 sets of universal essential single copy marker genes

173

according to a previous study.33 Moreover, the abundance of each ARG category was

174

the sum of the ARG abundance within the same resistance mechanism. The

175

percentage (%) of each ARG/MGE/ARC category was calculated as the ratio of the

176

abundance of the target ARG/MGE/ARC family in the total ARG/MGE/ARC

177

abundance in each sample. Core ARGs are defined as being detected with the

178

percentage of over 20% in all the samples.

179

Taxonomic Classification of the Bacterial Hosts of Core ARGs. To identify

180

potential hosts of ARGs, all the ORFs on each ARC were compared against the NCBI

181

NR database using BLASTP with e-value cutoff ≤ 10−5.24 The output results of the

182

BLASTP comparison were comprehensively parsed and taxonomically classified

183

using MEGAN (Version 6, MEtaGenome Analyzer).24 The potential host of an ARC

184

was confirmed based on a majority vote method meaning that if more than half of the

185

ORFs on each ARC were assigned to the same taxonomy rank (the genus level), then

186

the ARC was assigned to that taxon.36

187

Sequence-independent

188

composition-independent genome binning was performed to explore the fate and

Binning

of

Metagenome.

9


Sequence


189

distribution of ARGs and to retrieve genomes carrying ARGs from samples in the

190

O3/Cl2 disinfection strategy. Thus, the trimmed reads from paired samples (SW/OW

191

and OW/O-CW) were merged to form new datasets for metagenomic assembly, and

192

then ORF prediction and antibiotic resistance annotation of the coassembled contigs

193

were performed as described above. Next, two coverage values were obtained by

194

mapping the trimmed reads from each paired sample to the coassembled contigs by

195

using CLC Genomics Workbench and plotting against each other using R.37 The

196

contigs were assigned to different colors based on taxonomic classification and

197

guanine-cytosine (GC) content. The draft genomes were retrieved by using

198

differential coverage binning of multiple metagenomes recommended by Albertsen et

199

al. (2013),37 and the quality of the genome bins (completeness and potential

200

contamination level) was determined by searching a set of 107 HMMs of essential

201

single-copy genes, which represented 107 proteins conserved in 95% of all sequenced

202

bacteria,38 against the predicted ORFs using HMMER3.37

203

Statistical Analysis. The ARG/MGE/ARC abundance was expressed as the mean ±

204

standard deviation, and statistical significance between different groups was assessed

205

by one-way analysis of variance (ANOVA) with Tukey post hoc tests using IBM

206

SPSS Statistics 22.0 (IBM Inc., USA) with a significance cutoff of 0.05. To evaluate

207

the significant difference in the antibiotic resistomes of different samples, principal

208

coordinate analysis (PCoA) and Adonis test were simultaneously conducted based on

209

the Bray−Curtis distance of ARG abundances by R software (version 3.3.2) with the

10


Page 10 of 43

Page 11 of 43


210

Vegan package. Additionally, cluster analysis using UPGMA algorithm based on

211

Bray–Curtis similarity index was also performed to group samples with different

212

co-occurrence patterns of ARGs and MGEs using PAST (PAleontological STatistics)

213

software (University of Oslo, Norway).

214 215

RESULTS

216

Antibiotic Resistome Characterization and Alteration. In this study, 0.04–0.06%

217

of the total predicted ORFs were identified as ARGs, and 0.08–0.12% of the total

218

contigs were considered ARCs in different samples (Figure S2). In total, 0.65–1.25

219

cpc ARGs belonging to the resistance–nodulation–cell division (RND) antibiotic

220

efflux, ATP–binding cassette (ABC) antibiotic efflux, major facilitator superfamily

221

(MFS) antibiotic efflux, beta-lactamase, antibiotic inactivation, phosphotransferase,

222

acetyltransferase,

223

modulating resistance families were detected in all the drinking water samples (Figure

224

1). In detail, ARGs within RND, ABC and MFS antibiotic efflux families are

225

affiliated to the major function of antibiotic efflux; ARGs within beta-lactamase and

226

antibiotic inactivation families belong to the functional group of antibiotic

227

degradation; the major function of antibiotic modification contains ARGs within

228

phosphotransferase, acetyltransferase, nucleotidyltransferase families; ARGs within

229

rRNA methyltransferase and gene modulating resistance families pertain to function

230

of target protection and others, respectively (Figure 1). Notably, the abundance of

nucleotidyltransferase,

rRNA

methyltransferase

11


and

gene


231

ARGs within the RND and ABC antibiotic efflux and beta-lactamase families

232

accounted for 85.34–95.54% of the abundance of total ARGs (Figure S3), and these

233

were the dominant ARGs in all of the samples and were identified as the core

234

antibiotic resistome in this study. Moreover, 2.45–6.08% of the total identified ARCs

235

were found to carry multiple ARGs (≥ 2 ARGs) (Figure S4A), and 86.07–93.93% of

236

the ARCs carried ARGs within the RND, ABC antibiotic efflux and beta-lactamase

237

families in different samples (Figure S4B).

238

The relative abundance of total ARGs was remarkably increased from 0.82 ± 0.05

239

cpc in the SW sample to 1.24 ± 0.04 cpc after ozone disinfection (p < 0.05), and

240

further increased to 1.25 ± 0.08 cpc after the subsequent chlorination (Figure 1). In

241

detail, ozone disinfection mainly increased the abundances of RND antibiotic efflux,

242

ABC antibiotic efflux and beta-lactam ARGs (p < 0.05), but the following

243

chlorination decreased the abundance of beta-lactam ARGs (p < 0.05) (Figure 1).

244

However, the relative abundance of total ARGs was significantly reduced to 0.65 ±

245

0.01 cpc after AR disinfection mainly through the reduction of the RND antibiotic

246

efflux ARGs (p < 0.05) (Figure 1). The subsequent chlorination showed no significant

247

effect on the relative abundance of total ARGs but remarkably increased the

248

abundance of RND antibiotic efflux ARGs from 0.13 ± 0.01 cpc in the ARW sample

249

to 0.26 ± 0.01 cpc in the AR-CW sample (p < 0.05) (Figure 1). Conversely, the

250

alternative UV disinfection after AR treatment significantly enhanced the relative

251

abundance of total ARGs (1.01 ± 0.17 cpc), and the abundances of all detected types

12


Page 12 of 43

Page 13 of 43


252

of ARGs except for those in the MFS antibiotic efflux, antibiotic inactivation and

253

rRNA methyltransferase families (p < 0.05) (Figure 1). Compared to the SW, the

254

relative abundances of total ARGs in the final effluents were significantly increased

255

after the O3/Cl2 and AR/UV coupling disinfection strategies, respectively (p < 0.05)

256

(Figure 1). However, no significant change in the relative abundance of total ARGs

257

was observed after the AR/Cl2 coupling disinfection (Figure 1). PCoA based on the

258

Bray–Curtis distance showed the significant shift of antibiotic resistome after

259

different drinking water disinfection strategies (Figure S5, Adonis test, R2 = 0.91, p

Antibiotic Resistome Alteration by Different Disinfection Strategies in a

Recommend Documents