Chemical Isotope Labeling Exposome (CIL-EXPOSOME): One High

Subscriber access provided by UNIV OF LOUISIANA

Ecotoxicology and Human Environmental Health

Chemical Isotope Labeling Exposome (CIL-EXPOSOME): One Highthroughput Platform for Human Urinary Global Exposome Characterization Shenglan Jia, Tengfei Xu, Tao Huan, Maria Chong, Min Liu, Wenjuan Fang, and Mingliang Fang Environ. Sci. Technol., Just Accepted Manuscript • DOI: 10.1021/acs.est.9b00285 • Publication Date (Web): 03 Apr 2019 Downloaded from http://pubs.acs.org on April 3, 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 27

Environmental Science & Technology

1

Chemical Isotope Labeling Exposome (CIL-EXPOSOME): One High-throughput

2

Platform for Human Urinary Global Exposome Characterization

3

Shenglan Jia1, 2, 3; Tengfei Xu1, 3; Tao Huan4; Maria Chong1; Min Liu1, 2; Wenjuan Fang2;

4

Mingliang Fang1, 3, 5*

5 6

1

7

Nanyang Avenue, Singapore 639798

8

2

9

Institute, Nanyang Technological University, 1 Cleantech Loop, CleanTech One, Singapore

School of Civil and Environmental Engineering, Nanyang Technological University, 50

Residues and Resource Reclamation Centre, Nanyang Environment and Water Research

10

637141

11

3

12

Institute, Nanyang Technological University, Singapore 637141

13

4

14

Mall, Vancouver, BC Canada V6T 1Z1

15

5

16

Technological University, 1 Cleantech Loop, CleanTech One, Singapore 637141

Advanced Environmental Biotechnology Centre, Nanyang Environment & Water Research

Department of Chemistry, University of British Columbia, Vancouver Campus, 2036 Main

Analytics Cluster, Nanyang Environment and Water Research Institute, Nanyang

17 18

Address for Correspondence

19

Dr. Mingliang FANG, School of Civil and Environmental Engineering, Nanyang

20

Technological University, 50 Nanyang Avenue, Singapore 639798.

21

TEL: +65 6790 5331;

22

Email: [email protected]

1 ACS Paragon Plus Environment


23

TOC

24


Page 2 of 27

Page 3 of 27


25

ABSTRACT: Human exposure to hundreds of chemicals, a primary component of the

26

exposome, has been associated with many diseases. Urinary biomarkers of these chemicals are

27

commonly monitored to quantify their exposure. However, due to low concentrations and the

28

great variability in physicochemical properties of exposure biomarkers, exposome research has

29

been limited by low-throughput and costly methods. Here, we developed a sensitive and high-

30

throughput exposome analytical platform (CIL-EXPOSOME) by isotopically-labeling urinary

31

biomarkers with common functional groups (hydroxyl/carboxyl/primary amine). After a

32

simple cleanup, we used mass spectrometry to perform a screening for both targeted and

33

untargeted biomarkers, which was further processed by an automatic computational pipeline

34

method for qualification and quantification. This platform has effectively introduced an isotope

35

tag for the absolute quantification of biomarkers and has improved sensitivity of 2-1,184 fold

36

compared to existing methods. For putative identification, we built a database of 818 urinary

37

biomarkers with MS/MS fragmentation information from either standards or in silico

38

predictions. Using this platform, we have found 671 urinary exposure biomarker candidates

39

from a 2 mL pooled urine sample. The exposome data acquisition and analysis time has also

40

been greatly shortened. The results showed that CIL-EXPOSOME is a useful tool for global

41

exposome analysis of complex samples.



42

INTRODUCTION

43

The concept of the “exposome”, which is a measure of the effects of life-long environmental

44

exposure on health, has been a hot topic since its introduction in 2005.1 Recent initiatives in

45

environmental exposome research have attempted to map the complex relationships between

46

biomarkers and disease risks.2-4 Therefore, measurable and characteristic changes of all

47

biomarkers taken together form the key foundation of exposome research.5 Exposure

48

biomarkers possess enormous structural diversity and have a broad concentration ranges (from

49

10-7 μM to 1 mM for environmental pollutants).6, 7 Meanwhile, we are constantly exposed to

50

an array of ever-increasing chemicals, both endogenous and exogenous, from our diet, the

51

environment and our behavior.8 As such, profiling a large number of exposure biomarkers

52

requires powerful analytical techniques with broad coverage and high sensitivity.

53

Current exposure analytical methods include both traditional targeted measurement and

54

untargeted discovery of chemicals with biological importance (mainly endogenous

55

metabolites).9-11 For targeted appraoches, high sensitivity, wide dynamic range, and good

56

reproducibility are achieved with complex cleanup steps and expensive chemical isotope-

57

labeled (CIL) analogues.12 Moreover, the targeted method can only measure several biomarkers

58

at one time and requires large volumes of sample. Untargeted approaches have primarily

59

evolved from global metabolomics methods, by combining advanced analytical and

60

computational tools.13 Untargeted metabolomics retain a broad coverage of analytes by

61

sacrificing sensitivity, making the approach inefficient at measuring low concentration

62

environmental chemicals.5,

63

throughput platform, capable of capturing both endogenous and low level exogenous exposure

64

biomarkers.

14

Therefore, there is a great demand for a sensitive and high-

65

Among all biological samples, urine samples have a great potential to be used in long-term

66

exposome characterization due to their non-invasive and easy collection.15 Most environmental 4 ACS Paragon Plus Environment

Page 4 of 27

Page 5 of 27


67

chemical exposure biomarkers are excreted and can be found in the urine following oxidation,

68

reduction, or hydrolysis and further phase II conjugations.16 However, high concentrations of

69

salts and the diversity of chemical compounds present, make the analysis of urinary exposure

70

biomarkers across a wide range of concentrations challenging.

71

To address this analytical challenge and establish a sensitive, high-throughput exposome

72

platform, we applied a chemical-derivatization method for exposome profiling in urinary

73

samples. Chemical isotope labeling combined with liquid chromatography-mass spectrometry

74

(CIL-LC-MS) analysis was used to enhance the detection of both endogenous and exogenous

75

compounds, simplify both data acquisition and processing during exposome analysis. CIL-LC-

76

MS could significantly improve chemical detection and quantification of endogenous

77

metabolites, with sensitivity increases of 10-1,000-fold observed in previous studies.17-19

78

Furthermore, isotope forms of the derivatization reagents can be used as an economical internal

79

standard for absolute quantification.20,

80

subgroups based on their functional moieties (i.e., carboxyl, carbonyl, and amine etc.), different

81

labelling chemistries can be used to profile different sub-metabolomes in order to provide better

82

coverage of the entire compound range of interest.22, 23 For most exogenous biomarkers in the

83

urinary exposome, either the parent compound or metabolites commonly contain hydroxyl or

84

carboxylic acid functional groups to enhance their aqueous solubility, making the CIL method

85

promising for this application.24

21

After categorizing the metabolites into several

86

Here, we present a urinary exposome profiling workflow that employs CIL and automatic

87

data processing to examine hydroxyl/primary amine and carboxyl biomarkers originating from

88

both endogenous and exogenous chemicals. This approach provides a solution to several

89

challenges currently limiting exposure biomarker profiling. First, CIL methods were used to

90

increase sensitivity and greatly improve the coverage of exposure biomarkers. Second, an

91

automatic data processing was built up to confirm potential exposure biomarkers, enabling fast



Page 6 of 27

92

screening and compound identification. Finally, the method’s performance was validated in a

93

pooled human urine sample, demonstrating that the CIL-EXPOSOME workflow is a promising

94

tool for future global exposome analysis.

95

MATERIALS AND METHODS

96

Chemicals and Stock Preparation. All chemicals and reagents were purchased from

97

Sigma-Aldrich Singapore (Singapore) except those otherwise noted. The labelling reagent, p-

98

dimethylaminophenacyl bromide (DmPA), was purchased from Apollo Scientific Ltd

99

(Stockport, UK). The isotope reagents,

13C

2-danysl

chloride (DNS) and

13C -p2 13C

100

dimethylaminophenacyl bromide, were purchased from TMIC (Edmonton, Canada).

101

BPA,

102

Isotope Laboratories (Andover, MA, USA). HPLC grade water, ethyl acetate, and acetonitrile

103

(ACN) were purchased from Thermo Fisher Scientific (Singapore).

13C

6-monobenzyl

phthalate and

13C

4-methyl

12-

paraben were purchased from Cambridge

104

Standard Solution Preparation. Twenty-two hydroxyl- or carboxyl- containing standards

105

were chosen to optimize the analytical method. Additional information is provided in Table

106

S1. Each compound dissolved in ACN at a concentration of 10 mg/mL. Seven-level mixed

107

standard working solutions (10, 1, 0.1, 0.01, 0.001, 0.0001, 0.00001 µg/mL), also serving as

108

calibration standards, were prepared with ACN through serial dilution of stock solutions for

109

method validation. In addition, three-level quality control (QC) working solutions were

110

prepared in a similar manner to provide low (0.0001 µg/mL), medium (0.01 µg/mL), and high

111

(1 µg/mL) level controls. All stock and working solutions were stocked at −40°C, and diluted

112

with ACN to the desired level before use.

113

Urine Sample Collection and Preparation. A pooled human urine sample was prepared

114

by mixing first morning voids of 35 volunteers (healthy, age: 22 - 37) which were immediately

115

frozen at −20°C after sampling. Informed written consent was obtained from all participants.

116

Institutional Review Board approval for human specimen analysis were obtained from 6 ACS Paragon Plus Environment

Page 7 of 27


117

Singapore (IRB-2017-02-023). The pooled urine sample was buffered to pH 5.5 and treated

118

with an enzyme mixture (β-glucuronidase/sulfatase) for 12 h at 37°C. After centrifugation (10

119

min, 16000 g, 4°C), the urine was processed with labelling reagents for method development.

120

Spiking experiments were performed by adding a mixed working solution of the selected

121

compounds or surrogate standards to the pooled samples before sample preparation.

122

CIL-EXPOSOME Procedure.Work flow. In this study, two CIL methods were used to profile

123

hydroxyl/primary amine/ carboxyl endogenous and exogenous biomarkers. Figure 1 illustrates

124

the overall workflow of the method and data processing design with following steps: (1)

125

dansylation or p-dimethylaminophenacyl bromide (DmPA) labeling of the de-conjugated urine

126

sample; (2) equal amounts of 12C-/13C- dansylation/DmPA were mixed with labeled samples;

127

(3) liquid-liquid extraction was performed to enrich the labeled biomarkers from above

128

complex mixtures; (4) raw LC-qTOF analysis data were imported into IsoMS (Test S1) to get

129

12C-/13C-

130

database under mass accuracy tolerance (e.g., < 20 ppm); (6) the resulting information

131

(biomarker candidate name, mass tolerance, and m/z) were exported; and (7) the exported

132

biomarker candidates were processed and putatively identified with standard or in silico

133

predicted MS/MS fragmentation pattern.

peak pairs; (5) the exported paired suspect biomarkers were mass matched to the

134



135 136

Figure 1. Workflow for the determination and quantitation of hydroxyl/primary amine and

137

carboxyl exposure biomarkers in human urine by the CIL-EXPOSOME platform. Following

138

derivatization and sample processing, the LC-qTOF raw data were imported into the IsoMS,

139

and suspect exposure biomarkers were searched against the compounds in the CIL-

140

EXPOSOME database under different mass accuracy tolerances. The resulting information

141

(name, mass tolerance, accurate mass and m/z light) was exported. The exposure biomarker

142

candidates were further confirmed with MS/MS (i.e., MS2) fragmentation.

143

Reaction method optimization. To optimize the derivatization conditions, three-level quality

144

control (QC) mixed standard solutions reflecting human relevant levels were utilized. The

145

mixtures were allowed to react under different concentrations of derivatization reagent,

146

preparation times of derivatization reagent, ACN/H2O percentages, reaction temperatures and

147

incubation times individually. A detailed description of the labelling reaction can be found in

148

Text S2.

149

LC-qTOF Profiling and MS/MS fragmentation acquisition. Standard and urine analyses

150

were performed using a high-performance liquid chromatography (HPLC) system (1200 series,


Page 8 of 27

Page 9 of 27


151

Agilent Technologies) coupled to an Agilent 6550 Quadrupole Time-of-flight (qTOF) mass

152

spectrometer (Agilent, Singapore). Samples were injected for reversed-phase analysis in

153

electrospray ionization (ESI) positive and negative modes, as detailed in Text S3. The

154

instrument was set to acquire over a mass to charge ratio (m/z) range from 50 to 1700 with the

155

MS acquisition rate of 2 Hz. For the acquisition of MS/MS spectra of selected precursors, the

156

default isolation width was set to 1.3 Da with MS and MS/MS acquisition rates of 4 Hz. The

157

collision energy was set to 20 - 40 eV. Data were processed by using the CIL-EXPOSOME

158

Data Processing approach as described in the results.

159

CIL-EXPOSOME Method Validation.

160

Linearity and Sensitivity. Calibration was conducted based on the analysis of derivatized 7-

161

level labelled mix standard working solutions. Calibration curves were established by using

162

linear regression. The LOD and LOQ were estimated as 3 and 10 times of the standard

163

deviation calculated from 6 replicates at the lowest spiking level, respectively.

164

Precision and Accuracy. Three-level derivatization standard quality control (DQC) working

165

solutions were prepared to provide low, medium, and high scenarios of human biomarker

166

concentration. Intra- and inter-day precision and accuracy were assessed by using relative

167

standard deviation (RSD) within a day (n = 6) and on three consecutive days (n = 9);

168

respectively.

169

Stability and Cleanup Efficiency. Sample preparation stability of all derivatives was

170

evaluated by analysis of the DQC solutions at 48 h (28°C) and −40°C at 30-day intervals.

171

Pooled urine samples spiked with DQC solutions were prepared to evaluate the cleanup

172

efficiency, which was defined as the response ratio of extracted concentrations from a spiked

173

sample to the same concentrations of DQC. Each measurement was performed in triplicate.

174

Matrix Effect. In the absence of a biomarker-free urine matrix, a pooled urine sample was

175

processed as described in Urine Sample Collection and Preparation, and divided into two parts 9 ACS Paragon Plus Environment


Page 10 of 27

176

(A and B). Part A was dried to prepare the urine matrix, and an equal volume of DQC solutions

177

were added to estimate the influence of the matrix effect. Part B was analyzed directly to

178

determine the background concentration of each analyte. The matrix factor of each compound

179

was calculated by difference between Part A and DQC relative to the response of DQC in pure

180

solvent. To further distinguish the matrix effect from derivatization efficiency,

181

monobenzyl phthalate and 13C4-methyl paraben were added before derivatization to examine

182

the derivatization efficiencies on the two types of labeling. Labelled 13C12-BPA was dansylated

183

and added before injection as a recovery standard to assess the matrix effect.

13C

6-

184

Statistical Analyses and Quality Assurance and Quality Control (QA/QC). One-way

185

ANOVA and Tukey's post-hoc test were used to compare the response difference during

186

optimization. Procedure MiliQ-water blanks were prepared to evaluate contamination arising

187

from laboratory materials and solvents. An ACN blank was injected after 6 samples, and a

188

calibration check standard was injected after every 12 samples.

189

RESULTS AND DISCUSSION

190

CIL-EXPOSOME Chemical Derivatization Method Development

191

Derivatization reagent selection. One of the important considerations in developing labeling

192

chemistry is how to tag the urinary biomarkers with differential isotopic groups in robust way.

193

In this study, we evaluated the functional groups of urinary biomarkers from a recent published

194

comprehensive database (Exposome Explorer).25 Together, biomarkers with hydroxyl, primary

195

amine and carboxylic functional groups consist of > 70% of the total database. Hydroxyl

196

compounds and primary amines are preferentially labeled with dansylation methods.18, 26, 27

197

The derivatization of carboxylic acids can be conducted with a variety of chemical reactions

198

for analytical applications and phenacyl bromide has been used to label the acids for improved

199

HPLC and UV detection.28 However, a new reagent, DmPA, was designed that allows for the

200

introduction of a mass tag and concurrent improvement during LC-MS analysis.19 Here, we 10 ACS Paragon Plus Environment

Page 11 of 27


201

used dansyl chloride (DnsCl) and DmPA to label hydroxyl/primary amine and carboxyl urinary

202

biomarkers, respectively.

203

Reaction Condition Optimization. Although the stable isotope labeling method has been

204

previously optimized for the endogenous metabolites,29-31 the derivatization efficiency of high

205

concentration endogenous (µM to mM level) and low concentration xenobiotic biomarkers

206

(nM or pM level) may vary from each other. Furthermore, compared to endogenous metabolites,

207

the chemical structure of xenobiotics and their biotransformation products are more diverse.

208

To optimize the labeling conditions for urinary biomarkers, MiliQ water and urine spiked with

209

three-level QC working solutions were labeled under different reaction conditions. Figure 2 (a,

210

b) shows the effects of reaction solvent, time, temperature, DnsCl/compound mole ratio, and

211

DnsCl preparation time (from left to right) on the efficiency of standard dansylation. For the

212

pure chemical optimization, results indicate that the effect of water in the incubation solvent

213

on the labeling efficiency is compound dependent. After the optimization of all the tested

214

compounds, ACN/H2O (1:1, v/v) was selected as the reaction solvent for dansylation. The

215

reaction temperature and time also had an effect on the labeling efficiency. We found that the

216

optimal conditions were 40 min at 40°C, which differs from the previous method for

217

endogenous metabolites.18 The optimal mole ratio of compound to DnsCl was 1:105. The

218

optimization results for urine were the same as those for the MiliQ water, except that the mole

219

ratio of compound to DnsCl increased to 1:106.

220

Under both MiliQ water and urine background, 80% ACN/H2O was selected as the reaction

221

solvent for DmPA reaction, and the optimized reaction temperature and time was found to be

222

90°C for 60 min (Figure S1 (a, b)). The mole ratio of compound to DmPA was optimal at 1:104

223

for MiliQ water and 1:105 for urine.



224 225

Figure 2. Comparison of labeling efficiencies for 15 hydroxyl standards (a) spiked in MiliQ

226

water and (b) spiked in urine under different labeling reaction conditions. From left to right:

227

reaction time, reaction temperature, incubation solvent, DnsCl preparation time, and

228

compound:DnsCl mole ratio; (c) Comparison of cleanup method optimization for standards

229

under different cleanup procedures. All data in (a) (b) (c) are presented as the mean from three

230

separate experiments (n = 3) of three levels of QC working solution (0.1, 10, and 1000 ppb);

231

Extracted ion chromatograms of labelled standard mixture (d) after derivatization and (e)

232

before derivatization (10 ppb each). The compounds are shown as follows: 1, Methylparaben;

233

2, Ethylparaben; 3, 4-Bromophenol; 4, 1-Naphthol; 5, Benzylparaben; 6, 2-Hydroxyfluorene; 12 ACS Paragon Plus Environment

Page 12 of 27

Page 13 of 27


234

7, Butylparaben; 8, 9-Phenanthrol; 9, Bisphenol S; 10, Benzoresorcinol; 11, 1-Hydroxypyrene;

235

12, Triclosan; 13, Bisphenol A; 14, Genistein; 15, 3,5-DHPPA.

236

Cleanup Method Optimization. A proper cleanup method is crucial to minimizing matrix

237

effects. In the present study, we compared the recoveries of a few pre-treatment methods: (1)

238

extraction of the free biomarkers from MiliQ water followed by derivatization (LLE-Label);

239

(2) biomarker derivatization in MiliQ water followed by extraction (Label-LLE, Label-SPE);

240

(3) extraction of free biomarkers from urine followed by derivatization (urine LLE-Label) and

241

(4) biomarker derivatization in urine followed by extraction (urine Label-LLE). As shown in

242

Figure 2 (c), derivatized standards were more efficiently extracted by the liquid-liquid

243

extraction (LLE) strategy than free standards in both MiliQ water and urine. This is probably

244

because the derivatization step increases the hydrophobicity of hydroxyl, amines and

245

carboxylic acids in the urine sample. As a result, the chemicals are more likely to partition to

246

the organic phase.17 Sample processing was also streamlined by combining DnsCl- and DmPA-

247

derivatized samples prior to extraction. Finally, the optimized cleanup in this study was

248

simplified with a single liquid-liquid solvent extraction, which was enabled by the use of the

249

non-polar derivatization reagent. Details are as follows:

250

DnsCl/DmPA labeled urine samples were combined to profile hydroxyl/primary amine and

251

carboxyl biomarkers; (2) saturated NaCl solution (v/v, 1:5) was added to the mixture, then

252

extracted with ethyl acetate twice (v/v, 1:3); (3) the extracted solutions were combined and

253

dried, followed by reconstitution in ACN prior to LC-qTOF injection.

(1) equal amounts

12C-/13C

2-

254

Method Performance Validation.

255

Sensitivity and Separation Improvement. After derivatization, both the sensitivity and

256

retention of all the tested compounds were greatly improved compared with their non-

257

derivatized counterparts (Figure 2 d, e). A similar finding was observed for the urine samples

258

(Figure S2). The results showed that the detection sensitivities of these standard compounds 13 ACS Paragon Plus Environment


259

generally increased by 2-1,184 fold upon chemical labeling (Table S2). Before derivatization,

260

only three of the free form hydroxyl compounds (Bisphenol S, Benzoresorcinol and Genistein)

261

were detected at the concentration of 10 ppb. Chemical labeling will not only efficiently

262

improve the detection sensitivities, but will also lead to the discovery of more low-abundance

263

biomarkers. The chromatographic separation also showed significant improvement after

264

derivatization. Due to the diverse chemical structures and physicochemical properties among

265

biomarkers, separation of these compounds in a complex urine sample is not readily

266

accomplished with a single LC method. Derivatization can overcome this difficulty by

267

introducing a hydrophobic functional group to the analytes. Following derivatization, sufficient

268

separation can be accomplished by using simple reverse phase chromatography. By

269

chromatographic optimization, the derivatization methods greatly reduced the analysis time in

270

the traditional targeted analysis, especially for polar biomarkers that are not retained by regular

271

reverse-phase chromatography.

272

Analytical figures of merit. We estimated a series of attributes of the proposed quantification

273

method, such as calibration range, linearity, sensitivity, precision, accuracy, stability,

274

extraction efficiency, and matrix effect. The linearities of calibration curves for all analytes

275

were excellent among the validated concentration range of 100 ppt to 1 ppm, with

276

determination coefficients (R2) greater than 0.99. Due to the diversity of biomarker

277

concentrations, we considered two injection volumes (1 and 10 µL) to avoid response

278

saturation. As shown in Table S2, all analytes had an LOQ lower than 1 ppb. The analytical

279

method was shown to be highly reproducible, with all of the relative standard deviations (RSDs)

280

less than 20% for both intra-day and inter-day variations (Table S3). The stability of all

281

derivatized standards had coefficients of variation (CVs) that were ≤ 18% and ≤ 16% for

282

storage at 28°C for 48 h or at −20°C for 30 days, respectively (Table S4). Extraction

283

efficiencies of three-level DQC using the above optimized cleanup method were approximately


Page 14 of 27

Page 15 of 27


284

73 - 92% (Table S5). The interference of the matrix with the analytes was within 40% (Table

285

S5).

286

CIL-EXPOSOME Biomarker Database. To further facilitate the identification of urinary

287

biomarkers, we established a urinary CIL-EXPOSOME Biomarker Database. Each entry in the

288

exposure database included the chemical name, accurate molecular weight, exposure source,

289

chemical class, predicted exact m/z after being labelled, MS/MS fragmentation, and a detailed

290

information link (HMDB, KEGG, or PUBCHEM). To create a broad exposome database, the

291

CIL-EXPOSOME Biomarker Database was complied from a few existing data sources:

292

Exposome-Explorer database,25 Human Metabolome Database (HMDB),32 the U.S. National

293

Health and Nutrition Examination Survey (NHANES),33 and MyCompoundID database.34 The

294

MS/MS fragmentation of urinary biomarkers was generated either using authentic standards or

295

predicted using CFM-ID (see below). As a result, the CIL-EXPOSOME Biomarker Database

296

that was established (Table S6) included 818 reported hydroxyl/primary amine and carboxyl

297

biomarkers from food, drugs, endogenous metabolites, plasticizers, personal care products

298

(PPCPs), polybrominated diphenyl ethers (PBDEs), polycyclic aromatic hydrocarbons (PAHs),

299

polychlorinated biphenyls (PCBs), phthalates, pesticides, and flame retardants (excluding

300

PBDEs) (Figure 3).



301 302

Figure 3. Composition of CIL-EXPOSOME Biomarker Database: number of compounds (n =

303

818) by chemical classes.

304

CIL-EXPOSOME MS/MS fragmentation pattern and database development. MS/MS

305

fragmentation provides an important criteria for compound identification. In the present study,

306

we created a MS/MS fragmentation library of ~100 xenobiotics and their biotransformation

307

products by conducting dissociation at different collision energies (0, 10, 20, 30 eV). Together

308

with ~ 250 MS/MS spectra of endogenous metabolites from mycompound ID,35 we have

309

generated a MS/MS database comprised of ~350 authentic standards. For the rest of compounds

310

in the database, in silico MS/MS information was predicted using a machine-learning method

311

(CFM-ID).36 The database is available to the public upon request via Google Drive.

312

We further investigated the MS/MS fragmentation behavior of compounds labeled by two

313

reagents (DnsCl and DmPA). As shown in Figure 4, the expected precursor ions were m/z

314

378.1/380.1 for DNS/13C2-DNS-labeled naphthol and m/z 418.1/420.1 for DmPA/13C2-DmPA-

315

labeled mono-benzyl phthalate. The DNS/13C2-DNS-labeled hydroxyl compounds generated

316

two characteristic product ions, 171.1/173.1 and 235.0/237.0, by neutral loss from the precursor

317

ion, [M + H]+. For naphthol, another characteristic product ion was formed, [M + H]+ 207.0,


Page 16 of 27

Page 17 of 27


318

which is the parent ion of [M + H]+ 171.1/173.1 (Figure 4a and 4b). The DmPA/13C2-DmPA-

319

labeled carboxylic acids typically lose the common derivatization group (m/z 149.0/151.0).

320

The resulting product ion [M + H]+ −270.1 for mono-benzyl phthalate was formed (Figure 4c

321

and 4d). Overall, we have observed m/z 171.1 and 235.0 as two typical fragments for

322

dansylation, and m/z 149.0 as a typical fragment for the DmPA label. The molecular MS

323

fragment feature of the targeted compounds can also be observed in other MS/MS

324

fragmentation patterns. The use of the selected paired-peaks after chemical labeling as the

325

targeted precursor ions facilitates subsequent MS/MS analysis. In addition, the characteristic

326

fragmentation of metabolites after chemical labeling make identification feasible. This strategy

327

can narrow down the candidate compounds from fragmentation ions. [M+H]+ - 207.0

（ a（ 100

100 N

O O S O

m/z 171.1

75

171.1 [M+H]+

Intensity abundance (%)


[M+H]+ - 207.0

（ b（

N

m/z 378.1

50 O O S O

25

207.0 0

O S

N

378.1 [M+H]+

O

m/z 235.0 235.0 [M+H]+

200

C13 N N C13

O S O

75

m/z 173.1

50 O O S O

25

207.0 0

300

O O S

173.1 [M+H]+ C13 N 13 C

O S

380.1 [M+H]+

m/z 237.0 237.0 [M+H]+ 300

m/z [M+H]+ - 150.0

[M+H]+ - 148.0

（ c（

（ d（

O

100

O

O

O O

O O

100

N O

50 O

O

m/z 149.0

O

310.1

149.0 [M+H]+ 100

200

300

O

O O

25

O O O

N

N

0

328

270.1 [M+H]+

75



O

m/z 270.1

91.0

m/z 380.1

O

200

m/z

C13 N N 13 C

O O S O

m/z 418.1 418.1 [M+H]+

270.1 [M+H]+

75

C13 N 13 C O

O

O

25

O O O

C13 N 13 C

50

0

400

m/z 270.1

91.0

m/z 151.0

m/z 420.1

151.0 [M+H]+

420.1 [M+H]+

100

m/z

200

300

400

m/z

329

Figure 4. Fragmentation pattern of chemically labeled products. (a) DNS-labeled hydroxyl

330

compound; (b) 13C2-DNS-labeled hydroxyl compound; (c) DmPA-labeled carboxylic acid; (d)



331

13C

332

labelled m/z, respectively.

2-DmPA-labeled

carboxylic acid. Highlighted in blue and green are the theoretical and

333

CIL-EXPOSOME Automatic Profiling Workflow. In addition to working as an internal

334

standard, 13C reagent labelled biomarkers can play a unique role in untargeted profiling and

335

identification. Here, the obtained full scan (MS) raw data was processed by IsoMS to extract

336

12C/13C

337

difference (2.0067 Da for one light/heavy labeled metabolites), tolerance window (e.g., < 20

338

ppm), and intensities ratios (0.80 to 1.20) were extracted from the spectra. Information on

339

detected compounds, including accurate m/z, retention time (t), and peak intensity, was

340

generated for each exposure biomarker candidate (Table S7).

peak pairs from mass spectra. The paired-peaks that were with a defined mass

341

We further designed a MATLAB program to automatically search the exposure biomarker

342

candidate peak pairs against the CIL-EXPOSOME Biomarker Database; the detailed manual

343

can be found in Google Drive. Based on the IsoMS preselected mass list, the prospective

344

candidates of each exposure biomarker were output with mass tolerance, name, database ID,

345

accurate mass and labeled accurate mass (Figure S3). Users can select different values to use

346

for mass tolerance. For the identification, targeted MS/MS fragmentation of the potential

347

biomarker was generated with a qTOF mass spectrometer. QTOF parameters included and

348

width of 1.3 m/z units and a collision energy of 20 V. The obtained MS/MS fragmentation

349

spectra was then further matched with prospective candidates in the established MS/MS library.

350

Additionally, compounds were excluded that did not meet the following specifications: (1)

351

targeted MS/MS of the potential biomarker included 171.1/173.1 or 149.0/151.0 in the native

352

and isotope labelled compounds; and/or (2) the MS/MS fragment matched with the precursor

353

compounds.


Page 18 of 27

Page 19 of 27


7

8.0x105

8 9

（ a（ 12C

75

Intensity

4.0x10

3

labeled

2

1

6

5

4

7

0.0

10

7

-4.0x105

13C

12

11

13

15 14

labeled

7

-8.0x105

0

5

10

15

Retention time (min) 4-Ethylphenol (peak 11)

7575

（ b（ 12C

13C

labeled

labeled

2525

75

0

340

360

m/z 469.2

m/z 465.2

5050

12C

13C

labeled

m/z

00

380

440

460

m/z

m/z 13

N

171.1

0 100 100 3

173.1

100

6

25

Intensity (%)

235.0

050 50

100

2525

200 356.1

122.1

00

100

25

200

300

237.0

300 122.1

m/z

0 400

500 100

N

C

400

149.0

N

358.1

m/z

700 300

0 400

m/z

270.1

100500

1.2x104 1.0x104

6.0x107 （ d（

C

C N O

O

100

200600

O

m/z

O75

50

465.2

300700

O

O

O

25

600 200

500 13

O

75

50

480

13

O

9 O S O O

50

13

N

100

time (min) O S Retention O 75 O

7575

C

Intensity (%)

50

labeled

2525

m/z

1x1060 320

（ c（

7575

Intensity (%)

Intensity (%)

m/z 358.1

m/z 356.1

6 50 2x1050

Intensity (%)

O O

O

151.0

N

270.1

469.2

25

0 500

100 600

200 700

300 800

500

600

700

m/z

（ e（

2.0x103

Con. (ng/mL)

2.0x107

0.0 0.0

0.0

4.0x103

8.0x103

1.2x104

1.6x104

A or B P M es o S et hy rcin o lp ar l ab e n 4- Tri Br c l o om s a op n he N nol ap h 3, 5- tho D H l PP G en A is te in

4.0x10

7

BP

Intensity abundance (%) Intensity

100

Abundance (CIL-EXPOSOME)

Trans, trans-Muconic Acid (peak 10) 100 100

Intensity (%)

100 100

nz

Abundance (LC-QQQ)

Be

354 355

Figure 5. Chemical isotope-labeled urine sample analyzed by the CIL-EXPOSOME platform.

356

Extracted ion chromatograms of (a) exposure biomarker examples from ~671 potential

357

hydroxyl/primary amine and carboxylic acid metabolites from combined DNS/13C2-DNS and


800


358

DmPA/13C2-DmPA labeled urine sample. Fifteen exposure biomarkers have been confirmed

359

with authentic standards, namely, 1, Methylparaben; 2, 4-Bromophenol; 3, Naphthol; 4, BPS;

360

5 Benzoresorcinol; 6, Triclosan; 7, BPA; 8, Genistein; 9, 3,5-DHPPA; 10, Trans, trans-

361

Muconic Acid; 11, 4-Ethylphenol; 12, 2,4-Di-tert-butylphenol; 13, 2-Methylphenol; 14, 2-

362

Isopropoxyphenol; and 15, 4-Propoxyphenol; an example of MS/MS from the newly identified

363

(b) hydroxyl compound 4-ethylphenol and (c) carboxylic acid Trans, trans-muconic acid; (d)

364

correlation of BPA abundance between LC-QQQ and the CIL-EXPOSOME platform (R2=0.98,

365

p

Chemical Isotope Labeling Exposome (CIL-EXPOSOME): One High

Recommend Documents