Global Profiling of Urinary Mercapturic Acids Using Integrated Library-Guided Analysis

Abstract

Urinary mercapturic acids (MAs) are often used as biomarkers for monitoring human exposures to occupational and environmental xenobiotics. In this study, we developed an integrated library-guided analysis workflow using ultraperformance liquid chromatography−quadrupole time-of-flight mass spectrometry. This method includes expanded assignment criteria and a curated library of 220 MAs and addresses the shortcomings of previous untargeted approaches. We employed this workflow to profile MAs in the urine of 70 participants—40 nonsmokers and 30 smokers. We found approximately 500 MA candidates in each urine sample, and 116 MAs from 63 precursors were putatively annotated. These include 25 previously unreported MAs derived mostly from alkenals and hydroxyalkenals. Levels of 68 MAs were comparable in nonsmokers and smokers, 2 MAs were higher in nonsmokers, and 46 MAs were elevated in smokers. These included MAs of polycyclic aromatic hydrocarbons and hydroxyalkenals and those derived from toxicants present in cigarette smoke (e.g., acrolein, 1,3-butadiene, isoprene, acrylamide, benzene, and toluene). Our workflow allowed profiling of known and unreported MAs from endogenous and environmental sources, and the levels of several MAs were increased in smokers. Our method can also be expanded and applied to other exposure-wide association studies.

Graphical Abstract

INTRODUCTION

Humans are likely to be exposed to 1−3 million chemicals in their lifetime.¹ Prominent sources of xenobiotic exposure include industrial chemicals, petroleum products, fossil fuels, household chemicals, tobacco smoke, plant products, pesticides, and pharmaceutical reagents.^2,3 Several of these chemicals are highly reactive and injurious to health.^4,5 Highly reactive chemicals are also generated endogenously, especially during the conditions of oxidative stress.^6,7 Due to their high electrophilicity, unsaturated carbonyl compounds such as acrolein (abundant in cigarette smoke and automobile exhaust) and 4-hydroxy-trans-2-nonenal (generated by lipid peroxidation) can react with cellular nucleophiles of proteins and DNA.^6,8−10 Assessment of biomarkers of exposure to these toxicants will assist in developing appropriate remediation or detoxification mechanisms to limit the toxicity.

A common way to measure exposures to electrophilic xenobiotics is by analyzing their downstream N-acetyl-l-cysteine S-conjugate metabolites or mercapturic acids (MAs) in the urine.^9,11 MAs are formed from electrophile-glutathione S-conjugates through sequential transformation, including hydrolysis followed by N-acetylation in the kidney to form MAs, which are excreted in the urine.¹² The measurement of MAs is noninvasive and has been typically performed using liquid chromatography−mass spectrometry (LC-MS) targeted¹³⁻¹⁵ and untargeted methods.¹⁶⁻¹⁸

The untargeted approaches allow for the discovery, identification, and relative quantification of the detected species. Therefore, they have gained popularity for biomarker discovery and metabolite identification.¹⁶⁻¹⁹ Several proof-of-concept studies relying on untargeted approaches have been previously conducted by screening a common neutral loss (CNL) of 129 Da (a feature of N-acetyl-l-cysteine moiety) for MA detections.^16,18−20 Newer methodologies that combined untargeted metabolomics, CNL scanning, and online database searching allowed for the identification of some MAs.^17,21 In a previous study that analyzed urine samples from smokers and nonsmokers via an untargeted metabolomic approach, out of 91 potential metabolites annotated, 3 MAs were confidently identified through comparing retention time (RT) and MS/MS mass spectrum against authentic standards.²² However, CNL-based screening may miss MA ions that do not undergo a loss of 129 Da. Similarly, CNL alone may lead to false-positive assignments. Moreover, the small number of MAs listed in existing databases limits the range of annotated and identified species in previous mercapturomics studies. Due to a relatively small amount of untargeted studies and insufficient MA coverage, the field of untargeted MA analysis is still in its infancy.

We developed an integrated library-guided analysis (ILGA) workflow using ultraperformance liquid chromatography−quadrupole time-of-flight mass spectrometry (UPLC-TOF/MS) with MS^E data-independent acquisition (DIA) for MA profiling in human urine. This method includes expanded annotation criteria, which yielded ∼500 MA candidates in a typical urine sample—a sizable increase over previously reported workflows. In addition, we curated a library of 220 structures, which markedly enhanced the range of annotated MA species—116 MAs in the urine of study participants.

MATERIALS AND METHODS

Chemicals and Reagents.

Thirteen authentic MA standards, including N-acetyl-S-(2,4-dimethylbenzene)-l-cysteine (MPhMA), N-acetyl-S-(3-hydroxypropyl)-l-cysteine (3HPMA), N-acetyl-S-(2,3-dihydroxypropyl)-l-cysteine (23HPMA), N-acetyl-S-(2-carbamoylethyl)-l-cysteine (2CaEMA), N-acetyl-S-(N-methylcarbamoyl)cysteine (MCaMA), N-Acetyl-S-benzyl-l-cysteine (BzMA), N-acetyl-S-(n-propyl)-l-cysteine (1PMA), N-acetyl-S-(2-carboxyethyl)-l-cysteine (2CoEMA), N-acetyl-S-(2-cyanoethyl)-l-cysteine (2CyEMA), N-acetyl-S-(3,4-dihydroxybutyl)-l-cysteine (34HBMA), N-acetyl-S-(2-carbamoyl-2-hydroxyethy)-l-cysteine (2CaHEMA), N-acetyl-S-(3-hydroxy-1-methylpropyl)-l-cysteine (3HMPMA), and N-acetyl-S-[(1R)-2-hydroxy-1phenylethyl]-l-cysteine (2HPhEMA), were purchased from Toronto Research Chemicals, Canada. UHPLC-MS grade water, UHPLC-MS grade acetonitrile, LC-MS grade formic acid, and Infinity Creatinine Liquid Stable Reagent were purchased from Thermo Fisher Scientific, Inc., Waltham MA.

Study Population and Sample Collection.

Spot urine specimens were collected throughout the day (outside first morning void) from 70 participants, and stored at −80 °C until UPLC-QTOF/MS analysis. Figure 1 shows the general workflow developed for the analysis and assignment of MAs. Urinary cotinine levels were measured to determine the current smoking status of the study subjects. Participants with urinary cotinine level >40 μg/g creatinine were regarded as smokers.^23,24 Demographic information (Table 1) including age, sex, race, cotinine level, creatinine level, tobacco product usage, and medical information was obtained through the questionnaires. The study was approved by the University of Louisville Institutional Review Board (IRB 15.1260).

Figure 1.

Workflow displaying general steps of mercapturic acid (MA) analysis and profiling.

Table 1.

Demographic Characteristics, Medical Information, Urinary Cotinine Level, Urinary Creatinine Level, and Tobacco Product Usage of Study Participants by Nonsmokers and Smokers^a,b

	nonsmokers	smokers
variable	(n = 40)	(n = 30)	p value
sex, male	14 (35%)	8 (27%)	0.457
race			0.241
white	33 (83%)	25 (83%)
black	4 (10%)	5 (17%)
other	3 (8%)	0 (0%)
age (years)	49.2 ± 13.2	51 ± 11.1	0.543
body mass index, BMI (kg/m2)	28.8 ± 7.2	28.8 ± 6.5	1.000
hypertension	27 (68%)	17 (57%)	0.353
diabetes	7 (18%)	5 (17%)	0.927
urinary creatinine (mg/dL)	80.3 ± 83.3	112.7 ± 76.8	0.132
urinary cotinine (μg/g creatinine)	10.9 ± 10.9	2624 ± 2764.8	<0.001
cigarettes smoked per day in the past 30 days	0 ± 0	15.4 ± 9.7	<0.001
E-cig use			0.198
never	30 (97%)	25 (83%)
rarely	1 (3%)	4 (13%)
sometimes	0 (0%)	1 (3%)
chewing tobacco >20 times in lifetime			0.352
yes	1 (2%)	0 (0%)
no	38 (96%)	28 (93%)
unreported	1 (2%)	2 (7%)

^aValues represent mean ± standard deviation (SD) for continuous variables, and frequency (%) for categorical variables.

^bStatistical tests for the comparison between nonsmokers and smokers: the Student’s t-test for age and BMI; the Chi-square test for sex, race, hypertension, and diabetes.

UPLC-QTOF/MS Analysis.

The urine samples (50 μL) were thawed on ice and mixed with 0.1% formic acid (450 μL) in water. An aliquot of 7.5 μL of the mixture was analysed using an Acquity I-Class UPLC system (Waters, MA). The separation was performed using a 2.1 mm×150 mm Acquity Premier HSS T3 1.8 μm UPLC column (Waters, MA) maintained at 45 °C at a flow rate of 0.45 mL/min. The column was eluted with a gradient composed of 0.1% formic acid in water (solvent A) and 0.1% formic acid in acetonitrile (solvent B). The gradient profile started at 0% of B, increased to 23% B over 11 min, and then increased to 95% B over 3.6 min. The gradient was held at 95% B for 2.4 min before returning to the initial conditions over 0.05 min, and reequilibrated for 2.95 min before the next injection. To monitor the performance of LC-MS method (i.e., retention time shifts (ΔRT) and mass errors (Δm/z)), one injection of QC sample prepared by pooling equal amounts of each urine sample was added for every 10 sample injections. The acceptable ΔRT was ±0.15 min, and Δm/z was ±5 mDa. The list of ΔRT and Δm/z of MAs in QC samples can be found in Table S1.

QTOF-MS data were collected using a Synapt XS HDMS (Waters, MA) with Masslynx 4.2 software and an electrospray ion source operated in negative mode. The capillary voltage was 2.25 kV, the source temperature was 120 °C, the desolvation gas flow was 700 L/h at a temperature of 650 °C, and the cone gas flow was 150 L/h. The MS^E DIA acquisition was performed over the m/z range of 40−930 Da. MS^E is a method of data acquisition that collects mass spectral information of precursors and fragment ions in a single analysis by alternating two energy channels. The low-energy channel (function 1) generates precursor ion spectra, while the high-energy channel (function 2) generates fragment ion spectra. The precursor ions are then linked to their corresponding fragments that are co-eluted, creating spectra that resemble MS/MS. A collision energy of 2 V was applied in function 1 and a voltage ramp from 10 to 40 V was applied in function 2. Each of these functions employed a scan time of 0.2 s. Sodium formate was used for the mass calibration before the sample run, and leucine enkephalin with m/z ∼ 554.2620 was used as the lock mass solution during the acquisition. The raw files were analyzed using UNIFI 1.9 software package (Waters, MA).

Mercapturic Acid Library Building.

An ILGA workflow was established to profile MAs. This method started with curating an in-house MA library in UNIFI, which consists of molecular structures of known and deduced MAs. Structures of 159 reported MAs were collected by searching structures and keywords (e.g., “mercapturic acid” and “l-cysteine, N-acetyl-S”) in the SciFinderⁿ database, downloaded as .mol files, and imported into the library. Additional information, including retention times and MS/MS spectra, was added to the library after analyzing available MA standards. Furthermore, we deduced 61 prospective structures for MAs of harmful and potentially harmful constituents (HPHCs) found in tobacco smoke.²⁵ Proposed structures were deduced based on metabolic pathways that are known for major groups of reactive xenobiotics such as alkenals (e.g., acrolein),^11,26 aromatic hydrocarbons (e.g., benzene),²⁷ and polycyclic aromatic hydrocarbons (PAHs; e.g., phenanthrene).²⁸ Examples of known and proposed formation pathways of MAs from acrolein¹¹ and benzene²⁷ are shown in Figure S1. These pathways were explored to propose MA structures of analogous aliphatic and aromatic compounds. Together, our library comprised 220 MA structures. A specific acronym for each MA was created according to the guidelines outlined by Tevis et al.²⁹ (Table S2). Chemical formula, CAS number, and IUPAC name of each MA are available in Table S3. In addition, Supplementary Materials containing .mol structures and a combined .sdf file of all library items and SMILES for 116 annotated MAs are available.

Annotation and Assignment Criteria.

To be considered as a potential MA, selected MS features had to satisfy one of two selection criteria: (1) a common neutral loss (CNL) of m/z 129.043 Da (-C₅H₇NO₃) and at least one of the common ion fragments (CIFs) specific to N-acetyl-l-cysteine moiety at m/z 74.024 (C₂H₄NO₂), 84.045 (C₄H₆NO), m/z 128.035 (C₅H₆NO₃), and m/z 162.023 (C₅H₈NO₃S) in MS/MS spectra; (2) at least two CIFs. Figure 2 shows representative MS and MS/MS spectra of N-acetyl-S-[1-(hydroxymethyl)-2-propenyl]-l-cysteine (1HMPeMA) that includes all specified features.

Figure 2.

Characteristic fragmentation features of a MA using N-acetyl-S-[1-(hydroxymethyl)-2-propenyl]-l-cysteine (1HMPeMA) as an example (a). Representative MS and MS/MS spectra of 1HMPeMA (b).

Specific criteria were established for the annotation and assignment of MS features in four confidence levels (Table 2) modeled on the Compound Identification workgroup of the Metabolomics Society (2017).³⁰ Level 4 features were potential MAs with undetermined structure and no library match. Level 3 putative annotations matched the proposed MA structure(s) in curated library with ±5 mDa. In addition to the above criteria, Level 2 annotations were confirmed with published databases. If multiple level 3 or level 2 features matched the same annotation (different RT), a sequential number was added at the end of an acronym for the assumed isomer. Finally, Level 1 assignments were matched to RT and MS/MS of corresponding reference standards. Figure 3 shows an example of such confirmation using the standard compound 3HPMA.

Table 2.

Confidence Levels of Compound Annotations Modeled from Blazenovic et al., 2018

confidence level	description	data requirements
Level 1	MA with confident 2D structure. matches a reference standard or full structure elucidation.	at least MS/MS and RT match.
Level 2	MA with probable structure. matches an entry in published databases and the literature.	at least MS/MS match with in silico fragmentation based on the reported structure (.mol file) and external database elucidation (ChemSpider, HMDB, etc.).
Level 3	MA with the proposed structure. matches proposed structure in the curated library.	one or more library entries with matching m/z and elemental composition based on the formula.
Level 4a	potential MA with undetermined structure.	no library match.
	matches one of two selection criteria:   common neutral loss (CNL) of m/z 129.043 Da (-C5H7NO3), and at least one of the common ion fragments (CIF) specific to N-acetyl-l-cysteine moiety at m/z 74.024 (C2H4NO2), 84.045 (C4H6NO), m/z 128.035 (C5H6NO3), and m/z 162.023 (C5H8NO3S) in MS/MS spectra (function 2);or   at least two CIF

^aUsed for initial screening. Level 4 annotations are not discussed in this manuscript.

Figure 3.

Validation of a Level 1 MA assignment (i.e., 3HPMA) using an authentic standard. Top panels show the chromatograms, MS, and MS/ MS spectra of 3HPMA standard. Bottom panels show the corresponding results for the assigned 3HPMA in a urine sample.

Relative Quantification and Statistical Analysis.

The characteristics of the total selected study participants are expressed as mean ± standard deviation (SD) for continuous variables and frequency (%) for categorical variables. To examine the different characteristics between nonsmokers and smokers, Student’s t-test was performed for age and BMI, and the Chi-square test was performed for sex, race, hypertension, and diabetes (Table 1). The demographics (age, sex, race, etc.) of the smokers and nonsmokers were comparable.

For the relative quantification, the instrumentation responses (defined as 3-dimensional chromatographic peak volume with a cutoff value of 75) were used to measure the relative abundance of MAs in study subjects after normalization to urinary creatinine to adjust for dilution. Response values for undetected MAs were imputed by dividing the cutoff value (75 counts) by the square root of 2. Since the distributions of these normalized urinary MAs were rightskewed, they were log-transformed to improve the normality. The fold changes (FC) of the mean MAs were presented in the volcano plot (Figure S2), and 95% confidence limits (CL) between smokers and nonsmokers were presented in the forest plots (Figure 4).

Figure 4.

Forest plots showing significant differences in levels of 48 MAs derived from aliphatic (a) and aromatic (b) precursors between smokers and nonsmokers. The mean differences (round points) and 95% confidence limits (horizontal bars) were presented for each MA. The dotted lines represent no change. The instrumentation responses (defined as 3-dimensional chromatographic peak volume) normalized to urinary creatinine and log-transformed were used for comparison.

Linear regression models were constructed to examine the associations between urinary cotinine and MAs. Both independent variables (cotinine) and dependent variables were log-transformed to improve the normality in the models. Since demographic and clinical characteristics were not significantly different between nonsmokers and smokers (Table 1), these potential confounders were not adjusted in the regression models. The statistical significance was set at the p-value <0.05.

The correlation heatmap of selected MAs and cotinine, and principal component analysis (PCA) of urinary mercapturome were conducted using the MetaboAnalyst 5.0 platform (https://www.metaboanalyst.ca/). Other statistical analyses were performed using SAS, version 9.4 (SAS Institute, Inc., NC), and the forest plots were produced in GraphPad Prism, version 9.1 (GraphPad Software, CA).

RESULTS

Characteristics of the Study Population.

The demographic and clinical characteristics of the study participants are provided in Table 1. This study population comprised 69% females and 31% males, 83% White and 13% Black. The mean age was 50 ± 12 years, and the mean BMI was 28.8 ± 6.9 kg/m². Forty-three percent of the participants were smokers (urinary cotinine level >40 μg/g creatinine),²⁴ 63% were hypertensive, and 17% were diabetics. No significant differences were observed in the demographics and clinical characteristics of smokers and nonsmokers. The smokers in the study used 15.4 ± 9.7 cigarettes/day for the last 30 days. None of the participants were active oral tobacco product user or consumed chewing tobacco >20 times in lifetime. None of the participants was active e-cigarette user. One nonsmoker and 5 smokers occasionally used e-cigarettes in the past.

Mercapturic Acid Annotation Levels in the Study Group.

UPLC-MS analysis of the urine samples of study participants (n = 70) revealed that out of 10,000 features, 500 met Level 4 or higher confidence criteria (Table 2). Interestingly, the presence of CNL and at least 2 CIFs were found in over 85% of the Level 4 features. One hundred sixteen MAs matched the curated library and were assigned or putatively annotated (Level 1−3) as metabolites of 63 parent xenobiotics. These xenobiotics account for 37 aliphatic (alkenals, hydroxyalkenals, halogenated aliphatics, etc.) and 26 aromatic compounds (benzene and monocyclic substituted aromatic compounds, aromatic aldehydes, PAHs, halogenated aromatics, pharmaceutical agents, etc.). Full names, acronyms, structures, confidence levels of annotations, and CAS numbers of 116 MAs are provided in Table S3. Notably, to our knowledge, 25 MAs (with no CAS number available) are reported here for the first time. As noted in Table S3, these MAs were derived from 9 aliphatic precursors (2-pentenal, 2-hexenal, 2-heptenal, 2-octenal, 4-hydroxy-2-pentenal, 4-hydroxy-2-hexenal, 4-hydroxy-2-heptenal, 4-hydroxy-2-octenal, N-tert-butyl-acetamide).

Stratification of data between nonsmokers and smokers showed that 107 out of 116 MAs were detected in nonsmokers. However, as shown in Table 3, 53 MAs were above the instrument’s detection limit in >25% samples, 25 MAs in >50% samples, and 9 MAs (primarily derived from aliphatic precursors) in >75% samples. Smokers’ urine contained all 107 MAs found in nonsmokers and 9 additional MAs derived from aromatic (benzene, aniline, styrene, trimethylbenzene, naphthalene, coumarin) and aliphatic (4hydroxy-2-hexenal, sulforaphane) compounds. Ninety-six MAs were found in >25%, 55 MAs in >50%, and 37 MAs in >75% smokers’ samples (Table 3). The levels of 46 MAs were significantly increased in smokers (Table S3).

Table 3.

Number of Putative Mercapturic Acids Detected in Urine Samples Categorized by Their Parent Compound Subgroups^a

main group of the parent compound	subgroup of the parent compound	nonsmokers				smokers
		at least in one sample	>25% samples	>50% samples	>75% samples	at least in one sample	>25% samples	>50% samples	>75% samples
aliphatics	alkenals	13	11	8	3	13	14	11	11
	hydroxyalkenals	21	8	3	1	22	19	8	5
	halogenated aliphatics	8	4	1	N.D.	8	8	N.D.	N.D.
	other aliphatics	28	12	4	3	29	22	15	11
	aliphatics total	70	35	16	7	72	63	34	27
aromatics	benzene and monocyclic substituted aromatics	7	3	3	1	11	10	6	3
	polycyclic aromatic hydrocarbons	8	5	1	1	8	9	5	3
	aromatic aldehydes	6	5	4	N.D.	6	8	5	3
	halogenated aromatics	2	1	1	N.D.	2	3	2	1
	pharmaceuticals	11	2	N.D.	N.D.	13	2	2	N.D.
	other aromatics	3	2	N.D.	N.D.	4	1	1	N.D.
	aromatics total	37	18	9	2	44	33	21	10
grand total		107	53	25	9	116	96	55	37

^aN.D.: Not detected.

Forest plots illustrate significant changes in MA levels in each subgroup of parent compounds between smokers and nonsmokers (Figure 4). As shown in Figure 4a, 31 MAs derived from aliphatic precursors were significantly different between smokers and nonsmokers. The levels of 2 MAs, N-acetyl-S-methyl-l-cysteine (MMA) and N-acetyl-S-ethyl-l-cysteine (EMA), were significantly higher in nonsmokers. Twenty-nine MAs were significantly higher in smokers. These comprised 9 MAs from alkenals, 6 MAs from hydroxyalkenals, 1 MA from halogenated aliphatic chemicals, and 13 MAs from other aliphatics. Levels of all 17 MAs derived from aromatic precursors were significantly higher in smokers (Figure 4b). The levels of 68 MAs derived from alkenals (acrolein, crotonaldehyde, 2-heptenal), hydroxyalkenals (4-hydroxy-2pentenal, 4-hydroxy-2-hexenal, 4-hydroxy-2-heptenal, 4-hydroxy-2-octenal, 4-hydroxy-trans-2-nonenal), halogenated aliphatics (halogenated propane, pentane, heptane, 2,3-dichloropropene, ethyl chloroacetate), other aliphatics (ethylene oxide, allyl halide, 1,3-butadiene, acrylamide, acrylonitrile, methacrylonitrile, methylacrylate, butyl acrylate, 2-ethylhexyl acrylate, N-tert-butyl-acetamide, 4-methylthiobutyl isothiocyanate, allyl isothiocyanate, sulforaphane), and finally aromatic compounds (benzene, aniline, styrene, xylene, trimethylbenzene, hydroquinone, trihydroxybenzene, orthocetamol, cinnamaldehyde, 2phenylpropenal, α-phenylacrylic acid, naphthalene, phenanthrene, bromocyclohexane, halogenated dinitrobenzene, acetaminophen, 1,4-benzoquinone, phenethyl isothiocyanate, coumarin) were comparable between nonsmokers and smokers.

Because the reliability of statistical analyses deteriorates with increasing number of missing values, 55 MAs that were detected in >50% of smokers were used for the bivariate and multivariate analysis. A correlation heatmap was generated with Pearson correlation coefficients (r) between 55 MAs and cotinine to examine their relationship (Figure 5). Metabolites of 1,3-butadiene (2HBeMA), isoprene (4HMBeMA), and acrylonitrile (2CyEMA) were highly correlated with cotinine (r > 0.66; Figure 5). Interestingly, metabolites of acrolein (2CoEMA, 3HPMA), crotonaldehyde (3HMPMA), 2-pentenal (N-acetyl-S-(1-ethyl-3-hydroxypropyl)-l-cysteine,1EHPMA), and ethyl acrylate (N-acetyl-S-(3-ethoxy-3-oxopropyl)-l-cysteine, 3EoOxPMA) were highly correlated with each other (r > 0.67), suggesting co-exposure to the parent compounds of these MAs. We also observed that MAs from 2-octenal (CoMHxMA) and 4-hydroxy-2-octenal (2OxEHHxMA) were highly correlated to crotonaldehyde metabolite (3HMPMA), indicating the possible relationship between these parent compounds. Moreover, 3OxPMA, a proposed acrolein metabolite, showed a relatively poor correlation with the other two acrolein metabolites, 3HPMA and 2CoEMA. Notably, among the MAs correlated with cotinine, metabolites of acrolein (2CoEMA, 3HPMA), crotonaldehyde (3HMPMA), 1,3-butadiene (2HBeMA), acrylonitrile (2CyEMA), and isoprene (2HMBeMA, 4HMBeMA) are well-known biomarkers of exposure to volatile organic compounds (VOCs) and HPHCs in cigarette smoke.^13,23,31

Figure 5.

Correlation heatmap of urinary cotinine and 55 MAs in smokers and nonsmokers. The dendrogram was generated by hierarchical cluster analysis based on Pearson correlation coefficient (r). The MAs and cotinine are in both columns and rows with their r values repre sented by color (blue is low and red is high).

The PCA performed using the same 55 MAs found in >50% of smokers (Figure 6) shows that this unsupervised technique produced only partial separation between the two groups. In the PCA score plot (Figure 6a), smokers were located at the high end of the axis of principal component 1 (PC1). The PC1, which accounts for the highest share of the total variance among all PCs, represents MAs abundant in smokers’ urine, such as the ones from acrolein (2CoEMA, 3HPMA), crotonaldehyde (3HMPMA), 1,3-butadiene (2HBeMA), isoprene (4HMBeMA), and acrylonitrile (2CyEMA), which drove high PC1 scores (Figure 6b). However, instead of clustering at the low end of the PC1, nonsmokers distributed widely throughout the PC1 range (Figure 6a) with some MAs overlapping with smokers.

Figure 6.

PCA of 55 MAs in smokers and nonsmokers. PCA score plot of PC1 and PC2 with symbols denoting nonsmoker (blue circles) and smoker (red triangles) samples (a). Loading plot of PC1 and PC2 with each dot representing individual MA (b).

DISCUSSION

We developed a UPLC-QTOF/MS with MS^E DIA workflow that utilizes the ILGA method to profile MAs in the urine of 70 participants. Our workflow expands on and adds to the previous untargeted profiling approaches.^16,18−20 To increase confidence, in addition to utilizing the CNL 129 Da, we expanded assignment criteria to include one or more common MS/MS fragments (162.0229, 128.0349, 84.0454, or 74.0244) characteristic to N-acetyl-l-cysteine moiety. Furthermore, the ILGA approach establishes the connection between MAs and their precursors through the generation of an inclusive and expandable library that combines known and proposed structures of 220 MAs. The structural information of the metabolites and the metabolite−parent connection greatly facilitated the interpretation of the results and improved assignment confidence.

Our approach allowed for uncovering nearly 500 metabolite candidates and putatively assigning 116 MAs derived from 63 parent xenobiotics. The ILGA workflow uncovered 25 previously unreported MAs derived from 9 xenobiotic precursors mostly from alkenals and hydroxyalkenals. Interestingly, two MAs, MMA and EMA, from methyl and ethyl halides, respectively, were significantly higher in nonsmokers. Ethyl chloride and halogenated methane are industrial chemicals, used as alkylating agents, and are classified as potential carcinogens and mutagens.³² Our studies also showed that the levels of 70 MAs were comparable in nonsmokers and smokers, suggesting that these MAs are derived from organic pollutants present in the ambient air.

Endogenous lipid peroxidation and dietary sources are other potential sources of MAs. Hydroxyalkenals are predominantly derived from the oxidation of ω−6 polyunsaturated fatty acids. Due to their high electrophilicity, these unsaturated aldehydes are highly reactive and can crosslink with DNA and proteins. Increased accumulation of α,β-unsaturated aldehydes has been observed in several pathological conditions including atherosclerosis, Alzheimer’s, and Parkinson’s diseases.³³⁻³⁵ Lipid peroxidation also generates high concentrations of other alkanal and alkenals. Myeloperoxidase-derived reactions generate alkenals, such as acrolein during inflammatory conditions and metabolism of drugs such as cyclophosphamide and industrial chemicals such as allylamine.³³⁻³⁶ Longer-chain alkenals—2-hexenal and 2-octenal—are natural food constituents used widely as flavoring agents.^37,38

Increased abundance of MAs of several hydroxalkenals and alkenals in smokers is consistent with increased oxidative stress.^39,40 Our analyses are also in agreement with previous studies demonstrating that metabolites of acrolein (2CoEMA, 3HPMA), crotonaldehyde (3HMPMA), 1,3-butadiene (2HBeMA), N,N-dimethylformamide (N-acetyl-S-(N-methyl-carbamoyl)-l-cysteine, MCaMA), and acrylamide (N-acetyl-S-(2-carbamoylethyl)-l-cysteine, 2CaEMA; N-acetyl-S-(2-carbamoyl-2-hydroxyethyl)-l-cysteine, 2CaHEMA) have been reported to elevate in smokers and routinely used as biomarkers of exposure to tobacco smoke.^13,15,41 Also, metabolites of glycidol (N-acetyl-S-(2,3-dihydroxypropyl)-l-cysteine, 23HPMA) and isoprene (4MBeMA, 2HMBeMA, 4HMBeMA) were previously found to be elevated in the urine of smokers,^31,42,43 but are not specific biomarkers of cigarette use. Moreover, well-established biomarkers of cigarette exposure, aromatics-derived MAs, were also elevated in smokers. These included MAs from benzene (N-acetyl-S-(6-hydroxy-2,4-cyclohexadien-1-yl)-l-cysteine, 6HCycHeMA),⁴⁴ toluene (BzMA),¹³ and xylene (N-acetyl-S-[(2-methylphenyl)-methyl]-l-cysteine, 2MPhMMA).⁴⁵

Similar to several aliphatic compounds, we also observed significantly higher levels of MAs derived from aromatics including naphthalene (N-acetyl-S-(3,4-dihydroxy-1-naphthalenyl)-l-cysteine, 34HNeMA) and phenanthrene (N-acetyl-S-(2-phenanthrenyl)-l-cysteine, 2PaMA; N-acetyl-S-(4-phenanthrenyl)-l-cysteine, 4PaMA) in the urine of smokers. These PAHs have been found in combustible tobacco products,⁴⁶ but humans could be exposed to these PAHs through a variety of sources.^47,48 Notably, to the best of our knowledge, this is the first time to report mercapturic acid metabolites of these PAHs in human urine.

Exposures to halogenated alkanes, ethyl acrylate, isobutyl acrylates, trimethylbenzene, 2-fluorobenzaldehyde, 4-tert-butylbenzaldehyde, and 4-chloronitrobenzene are primarily industrial and occupational.^32,49−52 Detection of MAs derived from 4-methoxybenzaldehyde, acetaminophen, and 6-chloropurine indicates exposures to biogenic and pharmaceutical parent compounds.⁵³⁻⁵⁵ Interestingly, the source of 3-(2furanyl)-3-hydroxy-1-(4-methylphenyl)-1-propanone (FHMPP), the precursor of N-acetyl-S-[1-(2-furanyl)-3-(4methylphenyl)-3-oxopropyl]-l-cysteine (2FuMPhOxPMA), is not well known.

Some of the MAs derived from the same parent compounds were not necessarily correlated with each other. For example, among the 3 MAs derived from acrolein, 3OxPMA was not correlated with 3HPMA and 2CoEMA. This could possibly be due to the differences in the metabolism and pharmacokinetics of these metabolites.

In summary, we developed a UPLC-QTOF/MS with MS^E DIA workflow that utilizes the ILGA approach and applied it for profiling MAs in urine samples collected from 70 participants. Our studies identified 25 new MAs and detected 107 out of 116 MAs in both nonsmokers and smokers, suggesting that these MAs are derived, at least in parts, from environmental exposure. Significantly higher levels of 46 MAs in smokers reveal that smoking increases the levels of these metabolites. The ILGA method used in this study could be expanded to profile a wider set of metabolites for biomonitoring human exposure to VOCs and xenobiotics.

ABBREVIATIONS

MA

mercapturic acid

LC-MS

liquid chromatography−mass spectrometry

CNL

common neutral loss

ILGA

integrated library-guided analysis

UPLC-QTOF/MS

ultraperformance liquid chromatography−quadrupole time-of-flight mass spectrometry

DIA

data-independent acquisition

HPHC

harmful and potentially harmful constituent

RT

retention time

PAH

polycyclic aromatic hydrocarbon

CIF

common ion fragment

SD

standard deviation

BMI

body mass index

FC

fold change

CL

confidence limit

PCA

principal component analysis

VOC

volatile organic compound