Suspect Screening of Exogenous Compounds Using Multiple Reaction Screening (MRM) Profiling in Human Urine Samples

Abstract

Thousands of chemical compounds produced by industry are dispersed in the human environment widely enough to reach the world population, and the introduction of new chemicals constantly occurs. As new synthetic molecules emerge, rapid analytical workflows for screening possible presence of exogenous compounds in biofluids can be useful as a first pass analysis to detect chemical exposure and guide the development and application of more elaborate LC-MS/MS methods for quantification. In this study, a suspect screening workflow using the multiple reaction monitoring (MRM) profiling method is proposed as a first pass exploratory technique to survey selected exogenous molecules in human urine samples. The workflow was applied to investigate 12 human urine samples using 310 MRMs related to the chemical functionalities of 87 exogenous compounds present in the METLIN database and reported in the literature. A total of 11 MRMs associated with five different compounds were detected in the samples. Product ion scans for the precursor ions of the selected MRMs were acquired as a further identification step for these chemicals. The suspect screening results suggested the presence of five exogenous compounds in the human urine samples analyzed, namely metformin, metoprolol, acetaminophen, paraxanthine and acrylamide. LC-MS/MS was applied as a last step to confirm these results, and the presence of four out of the five targets selected by MRM profiling were corroborated, indicating that this workflow can support the selection of suspect compounds to screen complex samples and guide more time-consuming and specific quantification analyses.

1. INTRODUCTION

The exposome is defined as the totality of environmental exposures of organisms and it is estimated to include over 100,000 chemicals currently produced at industrial scale. Out of them, around 5,000 are currently dispersed in the environment widely enough to reach the human population and potentially impact the metabolism and health of communities [1,2]. Chemical exposure of living organisms includes pesticides and biocides, over the counter as well as prescription pharmaceuticals, personal care products, and food chemicals. Therefore, the main objective of exposome research is to assess and identify exposure to chemicals, since environmental chemical exposure frequently results in unfavorable biological responses [2].

Exposome research is conventionally focused on the use of liquid chromatography with tandem mass spectrometry (LC-MS/MS), which allows simultaneous detection, structural confirmation, and quantification of a wide range of chemical compounds. Depending on the level of knowledge about the exogenous compounds of interest, three LC-MS methodological approaches are usually employed to screen or identify exogenous compounds, namely (i) targeted methods, (ii) non-targeted and (iii) suspect screening for unknown compounds [3,4].

Targeted methods are based on specific compounds with known chemical structures and properties, and these methods are frequently used for quantification of specific compounds present in a defined sample [4,5]. The detection is usually performed by a low-resolution mass spectrometer, such as a triple quadrupole, and operated by selected reaction monitoring (SRM), to obtain high selectivity and sensitivity. In this type of LC-MS methodological approach, the identification of compounds is supported by comparison with reference data acquired from certified standards, used during the method optimization and validation process performed before the sample analysis [4,5].

Non-targeted LC-MS screening aims to detect unknown compounds without using any a priori criteria. For data interpretation of non-targeted screening, previous knowledge of chemistry is essential to interpret the MS data and propose a structural elucidation of the compounds. Therefore, the most popular technique for non-targeted screening includes chromatographic separation and full mass scan analysis with high resolution MS (HRMS) [4,5]. Lastly, the suspect screening (“known unknowns”), which are “suspect” compounds in relation to chemical structure, but the chemical name is known [4,5]. In this methodology, the objective is to generate semi-quantitative data that contribute to the elucidation of complex mixtures. Molecular characteristics such as neutral accurate mass, MS and retention time are compared to a chemical database to identify possible matches and identify chemical compounds.

Multiple reaction monitoring (MRM) profiling is a non-chromatographic mass spectrometry method performed using low-resolution MS for small molecule screening based on surveying of chemical functionalities at a speed and with sensitivity sufficient to be considered as a potential exploratory technique for exposome analysis [6]. MRM profiling applications include identification of drug resistance in bacteria [7], epidermal lipids markers in a mouse model of dermatitis [8,9], assessment of human coronary artery disease [10], as well as characterization of microscopic samples, such as single preimplantation bovine blastocysts and extracellular vesicles [11,12]. The MRM profiling strategy workflow uses around 1/10 of the instrument time needed for conventional LC-MS/MS methods due to the direct sample introduction and higher sensitivity when compared to full mass scan analyses. It has been applied so far to the screening of metabolites and lipids, and the method’s workflow includes discovery and screening steps. The discovery step surveys for chemical functionalities and it is performed on a restricted number of samples (usually pooled samples). These chemical functionalities are usually common to related drug classes, such as fentanyls, and can be used to profile the entire class, including new or still not fully characterized targets [13]. The information from the discovery step focused on profiling chemical functionalities is next organized in MRM scan lists and then used to profile individual samples in few minutes (usually at the rated of 100 MRMs/minute) still using direct sample injection. In this way, the method streamlines the use of instrument time and takes advantage of the high sensitivity of MRM scans. Chromatographic separation is reserved for a final structural validation step, if necessary. Sample analysis using MS without chromatographic separation is increasingly being recognized to have potential as a clinical tool for objectives including therapeutic drug monitoring and on-site drug testing [14,15].

The aim of this study was to propose a suspect screening workflow to the exploratory analysis of exogenous compounds an some of their metabolites in human urine of hospitalized patients and health controls taking advantage of the MRM profiling method. Even though the MRM profiling discovery step is usually performed using neutral loss and precursor ion scans, in this study we started with 310 MRMs (235 in positive mode and 75 in negative mode) related to the chemical functionalities of 87 exposome compounds listed in the METLIN database [16] and in the literature [17–40]. The discovery step of suspect compounds was performed on one pooled urine sample (i.e., an aliquot of each of the 12 samples) and was followed by the screening step carried out on individual samples. The screening step included only the MRMs detected in the discovery step, and these were associated with metformin, metoprolol, acetaminophen, acrylamide, and paraxanthine. We followed the discovery and screening steps by a compound identification step with product ion scans of each precursor in the successful MRM data. Subsequently, LC-MS/MS confirmed the presence and the identity of most of exogenous compounds indicated to be present in the samples by MRM profiling.

2. EXPERIMENTAL SECTION

2.1. Sample preparation

Human urine samples (n = 12) were obtained following informed consent from adult patients treated in the Ohio State University (OSU) Wexner Medical Center Emergency Department (ED). These patient samples were from subjects presenting urinary tract infection (UTI) and healthy controls selected from samples available of a larger study. Institutional review board approval was obtained (protocol code 2016H033, date of approval 06/01/2016). The samples were preserved in Assay Assure (Thermo Scientific) a universal urine collection and transport processing medium that preserves analytes for up to 26 days at room temperature according to the manufacturer. The samples were processed within two weeks which is the time-period that we have independently validated as the period of urine analyte stability using this preservative [41]. Samples were prepared according to the Bligh & Dyer protocol [42] and the experiments were performed using the polar phase. After drying in a vacuum centrifuge (Savant, SPD 2010, Thermo Fisher Scientific, Waltham, MA) during 12 hours, the extracts of the urine samples were redissolved in 200 µL of H₂O:ACN 95:5 (v/v) with 0.1 % of formic acid (stock solution). For sample injection, the stock solution was diluted 200 times into injection solvent ACN/MeOH/NH₄Ac 300 mmol L⁻¹ 3:6.65:0.35 (v/v/v) for all experiments.

2.2. MRM profiling mass spectrometry

For the discovery step, a list of 310 MRMs related to 87 exposome compounds listed at the METLIN database [16] and reported in the literature [17–40] (Table S1) was used to screen a urine pool sample from 12 human patients. Next, in the screening step, the MRMs of ion intensity equal or higher than 30 % the blank sample signal were selected to screen the individual urine samples. Usually, for profiling analyzes, only background subtraction is performed, i.e., all ions presenting intensity higher than the background are included in the profile. Here, we took a more conservative approach and applied a threshold of 30 % based on our experience with observing MRMs that will respond linearly with sample concentration [6]. In the identification step of the exploratory analysis, the precursor ion of the ion pair representing the MRM was selected and product ion scans were acquired on a pooled sample to gain further evidence of compound identity. The workflow of the study can be visualized in Figure 1. Note that the final LC-MS/MS validation step is not listed in the Scheme.

Figure 1.

Overview of the experimental design performed for the screening for endogenous compounds exposure in human urine samples using MRM Profiling. 310 MRMs related to the functional groups of 87 compounds were selected for the discovery step, 11 MRMs were detected in a pooled sample. The samples were investigated individually in the screening step for the selected MRMs. In the identification step, product ion scans were performed for these MRMs. The 11 MRMs and the product ion scans were related to 5 compounds and this information was used for validation by LC-MS/MS.

Experiments were carried out using a triple quadrupole mass spectrometer, (Agilent QQQ 6410, Santa Clara, CA), equipped with Agilent G1367A 1100 series autosampler (Santa Clara, CA). Samples were introduced into the mass spectrometer ESI ion source by direct injection, i.e., without chromatographic separation. The ESI voltage was 4.0 kV and dwell time was 25 ms. Collision energies (CE) were optimized in the discovery phase of the method using a range of collision energy (CE) settings (15, 20, 25, 30, 35, 40, and 45 V) and CE of 25 was set as optimal. An 8 μL volume of sample was injected at a flow rate of 10 μL min⁻¹. Sample dilution solvents and concentrations were based on an optimization study (data not shown). Quality control samples (Equisplash Mix, Avanti, 1.2 ng of each internal standard in the mix) were used to monitor instrument performance. An amount of 8 µL mixed solvent of methanol:water 50:50 (v/v) was used to flush the system between samples.

The identification stage of the proposed suspect screening method used product ion scans to evaluate if one or more expected product ions were present as additional supporting data for the presence of these chemicals in the samples. For this step, we selected the precursor ions for the most intense MRMs related to the five compounds, namely metformin, metoprolol, acetaminophen, acrylamide, and paraxanthine for product ion scan.

2.3. Data analysis for the MRM profiling data

Ion counts of the MRMs monitored were compared to the ion counts of a blank (injection solvent) and transitions 30 % higher than the blank were considered for the screening and discovery steps. Ratios of sample/blank ion signal were used for the analysis of the results in the discovery and screening steps. For the identification step, the precursor ions for the MRMs detected in the screening step were selected for product ion scan experiments, where one or more product ions could be observed as an additional step of structural characterization.

2.4. Structural Confirmation by LC-MS/MS

2.4.1. Chemicals and reagents

Analytical standards of metformin, metoprolol, acetaminophen, and acrylamide were purchased from Sigma-Aldrich^® (St. Louis, MO, USA). Paraxanthine standard was purchased from Cayman Chemical^® (Ann Arbor, MI, USA).

Acetonitrile (ACN) HPLC grade, and formic acid (88 %; J.T. Baker^® Phillipsburg, NJ, USA) were used as an eluent additives. Water was purified using Simplicity UV from Merck^® (Darmstadt, Hessen, DEU). Stock solutions of 1 mg mL⁻¹ were prepared in methanol for acetaminophen and paraxanthine, and in water for acrylamide, metoprolol and metformin. The solutions were stored under refrigeration (−20°C) and further dilutions were prepared in water as needed.

2.4.2. LC-MS/MS conditions

LC-MS/MS experiments were targeted to confirm suspect compound identity and not to quantification. Data acquisition was performed using a Shimadzu^® Nexera X2 HPLC/UHPLC system (Kyoto, JPN) equipped with two LC-30AD pumps (binary flow pumping mode), integrated with a degasser unit DGU-20A3R, autosampler SIL-30AC with injection capacity of 0.1 to 50 μL, and a column oven compartment CTO-20A. To complete the modular system, a HPLC system controller CBM-20A acts as an interface for connecting instrumentation to LC workstation.

The LC system was coupled an AB SCIEX 3200 QTRAP (Foster City, CA, USA) mass spectrometer equipped with a Turbo V Ion Source electrospray ionization (ESI) source. A Phenomenex Kinetex^® PFP 100 Å (150 × 4.6 mm, 5 μm) was used as analytical column in the chromatographic separation. The analytical column was kept at 40 °C in the oven compartment. The eluents were ultrapure water with 0.1 % formic acid (A), and acetonitrile with 0.1 % formic acid (B) in a flow rate of 0.6 mL min⁻¹. An elution in gradient mode was employed according to Figure S1.

The mass spectrometer valve (diverter valve) was kept open until 1.9 min in order to direct eluent to waste. From 2 min to 10.9 min the system flow was directed to the mass spectrometer. The diverter valve returned to waste in 11 min, remaining in this position until the end of the analysis. After each analysis batch, the system was washed with eluents of the mobile phase and an injection of acetonitrile to flush and clean the lines and avoid carryover in the system. Parameters dependent on the electrospray ionization source were optimized by flow injection analysis (FIA): curtain gas (CUR) 15 V, nitrogen medium collision gas (CAD), source temperature (TEM) 650 °C, ion spray voltage 5000 V, nebulizing gas (GS1) 40 psi and drying gas (GS2) 40 psi. The mobile phase used was water and acetonitrile (50:50 v/v) with 0.1 % formic acid at a flow rate of 0.6 mL min⁻¹, with injections of 10 μL of a 1 mg L⁻¹ solution containing all the analytes. Declustering potential (DP), collision energy (CE), entrance potential (EP) and collision cell exit potential (CXP) were optimized by direct infusion of a standard solution from each compound at a concentration of 1 mg L⁻¹ with addition of 0.1 % of formic acid to improve the ionization.

The mass spectrometer was operated in positive ionization mode (ESI+) with selected reaction monitoring (SRM). For quantitative purposes, three different SRM transitions were monitored for each compound. SRM transitions and MS parameters for detection are shown in Table S2. The resolution in the first and third quadrupole (Q1 and Q3) was defined as unitary and the delay between scans was 5 ms. All data were collected and processed using the Analyst^® 1.5.2 software. Quality control samples, i.e., standards at 300 µg L⁻¹ were prepared in purified water and analyzed regularly before and after each analysis batch of samples. This procedure was adopted to ensure the method performance and reliability. In addition, after the QC samples, an acetonitrile was injected to clean the system and avoid carryover before the start of a new analysis batch of samples.

3. RESULTS AND DISCUSSION

In this research, the MRM profiling was applied in a three-step workflow. For the discovery step, a list of MRMs related to the chemical functionalities of exogenous compounds that could be potentially present in the patient’s urine samples based on METLIN database and literature [17–40] (Table S1) was applied to survey a single representative sample produced by pooling the same volume of each individual urine sample (n =12) included in the study. For studies including a larger number of samples or diverse experimental groups, diverse pools could be used for the discovery step. Next, as a screening step, MRMs presenting higher ion signals than those of the blank sample were selected to screen samples individually. The MRMs to be used for the screening step could have been obtained (as an alternative procedure) by performing neutral loss and precursor ion scans in the discovery step. As a third step of the MRM profiling analysis, the precursor ions of the MRMs detected in the individual samples were used for product ion scan experiments to gain further evidence of the compounds’ molecular identity. LC-MS/MS analysis targeted to the five tentatively identified compounds (metformin, metoprolol, acetaminophen, acrylamide, and paraxanthine) by MRM-profiling was performed and four of them were confirmed. The analytical workflow is shown at Figure 2.

Figure 2.

Analytical workflow of the validation of the MRM profiling strategy for suspect compound screening in urine samples. The sample was processed using liquid-liquid extraction and the polar phase of one pooled sample was used for the selecting the ion transitions of interest (discovery step). Next, each sample was interrogated for the selected MRMs (screening step) and product ion scan was acquired from the MRMs detected in each sample (confirmation step). Validation of the presence of the compounds indicated by the MRM profiling was performed using LC-MS/MS.

The described workflow does not include the use of internal standards until LC-MS/MS confirmation. Instead, a range of collision energies was used at the discovery step, and the signal for the MRMs was assayed for each sample at the screening step using the collision energy that provided the highest ion signals. A confirmatory step included acquiring product ion scan was performed before the extra effort of adding standards and chromatographic separation.

At first, the development of an analytical method for set of compounds without having their authentic standards and including minimal optimization can raise questions regarding its analytical quality. Nonetheless, to establish a compromise between speed, cost, and appropriate validation, the use of standards can be reserved for the compound confirmation through LC-MS/MS or LC-HRMS. Therefore, the proposed workflow results in a fast-screening allied to an extra confirmatory step which permits to obtain a gold-standard analytical method. Some compounds of interest may not be detected using this suspect screening method based on MRM profiling due to levels below the threshold of the instrument and this fact does constitute a limitation of the proposed method.

In the discovery step, 11 MRMs were detected higher than the blank with 25 V which was set as the optimized collision energy. Each individual sample was investigated, and these 11 MRMs proved to be higher than the blank at least in one of the 12 samples (Table S3). Product ion scans for the parent ions of the MRMs observed in the screening phase suggested the presence of metformin (130^®71; 130^®46; 130^®60), metoprolol (268^®56; 268^®74; 268^®116), acetaminophen (152^®110; 152^®65), acrylamide (72^®55; 72^®44), and paraxanthine (181^®124).

From the 12 samples analyzed in the study, half of the patients had positive urine cultures characteristic of UTI, and these samples presented the highest signal/noise ratio for the MRMs detected for exogenous compounds (10 MRMs) (Table S4). Most of the exogenous compounds indicated by MRM profiling in the human urine samples in this research could be related to drugs prescribed or accessible as over the counter products for common health conditions or for pain. Metformin is a drug used to treat type II diabetes; metoprolol is a beta-blocker commonly used to control arterial hypertension and ventricular arrhythmia. Acetaminophen is a pain reliever and fever reducer with widespread use. Acrylamide found in human fluids is believed to originate from heat-processed foods such as potato chips and tobacco smoke [43,44]. Paraxanthine is a caffeine metabolite that can be used as a marker of caffeine consumption in epidemiological studies [45].

For the MRM confirmatory stage, the product ion scan experiment was performed for the precursor ions of the 11 MRMs detected in the samples. These MRMs corresponded to five individual exogenous compounds. Results showed that the fragments detected for metformin, metoprolol, acetaminophen, acrylamide, and paraxanthine in the product ion scans, matched previous reports and available databases [16–40] as demonstrated in Figures S2, S3, S4, S5 and S6.

Additional confirmation by LC-MS/MS was performed. For this MS/MS experiment the [M+H]⁺ ion was used as the precursor ion and two product ions were selected. In this way, two ion transitions with the highest intensity were used as qualifier ions. The LC-MS/MS method included 617 scans with a dwell time ranging from 150 to 200 ms according to the number of ion transitions monitored. Around 10 data points per peak were recorded for all compounds allowing the simultaneous determination of the selected compounds in the same injection (shown in Figure S7). Chromatographic conditions were also optimized to obtain a suitable elution of the target compounds in a short analysis time. The selected flow rate of 0.6 mL min⁻¹, which showed be appropriate to elute the target compounds and an injection volume of 10 µL. Formic acid at 0.1 % was chosen as an eluent additive, since higher concentrations did not increase the signal response. No carryover effect for any compound was observed. The total LC MS-MS analysis time was 19 min, including column washing and re-equilibration step.

All the samples contained at least one of the target compounds, except for paraxanthine that was not found in any of the analyzed samples (Table S5). Therefore, this was a false-positive result, probably due to an isomeric compound. An example of the target compounds chromatograms found in the samples can be visualized in Figure 3.

Figure 3.

Extracted ion chromatogram (XIC) of detected compounds found in some of the human urine samples: metformin METF (130 ^®71); acrylamide ACRL (72 ^®55); acetaminophen ACTM (152 ^®110) and metoprolol METP (268 ^®116). Specific samples where compounds were found are listed at Table S5.

LC-MS/MS analysis confirmed the presence of four out of five compounds indicated by MRM profiling in the samples, suggesting the suitability and practicality of this method as a first step screening of exogenous compounds in biological samples. We applied the method to the polar phase recovered through the Bligh & Dyer method, but the non-polar phase could also be used for exogenous compound screening or for evaluating lipid changes that could be related to the exposure to exogenous compounds. For a class-based suspect screening, the proposed workflow would have a discovery step based on a collection of precursor ion and neutral loss scans to identify parent ions containing expected fragments and obtain the MRMs for the screening step. Overall, the proposed MRM profiling workflow allowed the screening of endogenous compounds in complex samples at 50 compounds (using 2 MRMs for each compound) per min acquisition rate.

4. CONCLUSIONS

In this research we demonstrate the workflow of suspect screening by MRM profiling in three steps (discovery, screening, and identification) as an analytical tool for surveying exogenous exposure candidates present in biological samples. The selected 310 MRMs included fragments of functional groups for 87 exogenous compounds and the screening step data acquisition was completed in 2 minutes/sample. Results for the 11 highest intensity MRMs indicated the presence metformin, metoprolol, acetaminophen, paraxanthine and acrylamide in urine samples. The presence of all compounds, except paraxanthine, was confirmed by LC-MS/MS. Even though this research has been performed in a small set of samples, the results are promising since urine is a biofluid that can be readily obtained noninvasively. The use of MRM profiling has also the potential to rapidly on-site analysis if miniature instruments are used for on-site environmental exposure evaluation in remote places or in the field.

Highlights.

• The suspect screening workflow using MRM Profiling to survey the presence of exogenous compounds in human urine;

• LC-MS/MS as a last step confirmed the presence of the targets selected by MRM profiling;

• A minimal sample preparation and direct sample injection resulted in a rapid analytical screening method.