Diagnostic Fragmentation Pathways for Identification of Phthalate Metabolites in Non-Targeted Analysis Studies

Abstract

Phthalates are widely used as plasticizers in polyvinyl chloride (PVC) plastics and solvents in many consumer products, resulting in the constant exposure of humans to these chemicals. Although many phthalate metabolites have been targeted in quantitative analysis for exposure assessment, there still remain many which are unknown and thus unmonitored. Due to its discovery based nature, non-targeted analysis using high-resolution mass spectrometry (HRMS) is increasingly used to screen and identify previously unknown metabolites as exposure biomarkers. In this study, we developed a new non-targeted analysis approach for identification of phthalate metabolites based on a comprehensive study of the fragmentation pathways in electrospray ionization (ESI) quadrupole-time-of-flight mass spectrometry (QTOF-MS). The method uses three structurally specific fragment ions as diagnostic filters to identify precursor ions associated with phthalate metabolites as well as distinguish between the types of phthalate metabolites identified. It was found that the fragmentation pathway of phthalate oxidative metabolites is significantly different from non-oxidative metabolites. The fragment ions generated at various collision energies (CE) from 0 to 40 V were investigated to understand their cleavage mechanisms. All phthalate metabolites including oxidative and non-oxidative metabolites produce a specific ion at m/z 121.0295, representing the deprotonated benzoate ion [C₆H₅COO]^-. Most phthalate metabolites can produce a specific ion at m/z 147.0088, the deprotonated o-phthalic anhydride ion. However, phthalate carboxylate metabolites can only produce the [M-H-R]^- ion at m/z 165.0193 and do not produce the fragment at m/z 147.0088. Based on these findings, four fragmentation pathways of phthalate metabolites have also been proposed in this study and applied to improve the identification of new metabolites in non-targeted analysis. With this workflow, eight unknown precursor ions were identified in the pooled urine sample, but only one followed the fragmentation pathway. It was interpreted as unreported new phthalate metabolite by the MS/MS spectrum and its predicted retention time supported the interpretation. The developed method can identify new phthalate metabolites in human urine with a detection limit of less than 50 ppb.

1. Introduction

Phthalates are primarily used in polyvinyl chloride (PVC) plastics and childcare products and toys to increase product durability and flexibility.¹ They are also widely applied as solvents and additives in many consumer products, including clothes, medical supplies, toys, childcare products, food packaging products, and personal care products such as soaps, shampoos, and cosmetics.^2–4 Widespread use of phthalates makes them ubiquitously present in the environment and one of the synthetic organic chemical classes to which humans are most frequently exposed.^5–7 Human exposure to phthalates is associated with altered hormone levels and some chronic diseases such as asthma,⁸ allergies,⁹ reproductive and developmental effects,^10–12 negative birth outcomes,¹³ learning and behavioral problems in children¹⁴ and even some cancers.^15,16 Therefore, the use of phthalates has been regulated in all childcare articles and toys by many governments including Canada,¹⁷ European Union,¹⁸ and USA.¹⁹

Phthalates undergo hydrolysis in the human body producing their primary monoesters within 24 hours and are subsequently excreted in urine as either conjugated-form or free-form.^20,21 Some primary phthalate monoesters can be further metabolized to secondary oxidative metabolites before urinary excretion.²² Liquid chromatography - mass spectrometry (LC-MS) systems are commonly used to provide quantitative information for targeted surveillance of well-known phthalate metabolites in urine and exposure assessment.^23,24 However, well-known targeted phthalate metabolites with commercially available standards represent only a portion of phthalate metabolites in real samples.²⁵ New phthalate compounds are constantly being generated by industry,²⁶ and the transformation and metabolism of both new and old phthalates has not been completely characterized. Therefore, there remain many phthalate metabolites which have yet to be monitored for assessment of human exposure. This highlights the need for non-targeted analysis approaches for the identification of previously unknown biomarkers for exposure assessment, made possible due to the high resolving power and high mass accuracy of high-resolution mass spectrometers (HRMS).^27–29

The workflow of non-targeted analysis using HRMS starts from data acquisition to data analysis and structural identification. However, it is still challenging to prioritize the precursor ions and assign the formulas using mass spectrometry without any prior information about the unknown compounds.²⁹ Interpreting non-targeted mass spectra and assigning confident identifications of unknown phthalate metabolites remains challenging because of the lack of reference MS/MS fragment spectra and standards, which impedes unambiguous assignment of unknown metabolites in human samples. As a result, it is necessary to develop more filters to narrow down the candidate ions and remove the interferences in data analysis in order to prioritize precursor ions and improve the confidence of structure interpretation using MS/MS spectra. The fragment ions generated in MS/MS scans from chemicals with similar structural skeletons can yield insight into common fragmentation mechanisms, which can be useful for prioritizing candidate precursor ions, assigning chemical formulas and elucidating structures in non-targeted analyses of unknown metabolites.

In this study, the fragmentation pathways and patterns of phthalate metabolites have been investigated using 36 metabolites and 14 isotope-labeled metabolites with linear, branched, and oxidized (oxo, hydroxyl, and carbonyl) alkyl chains. A non-targeted analysis method has been developed for identification of new phthalate metabolites in human urine. In this method, the three structurally specific ions were used as a filter to quickly prioritize candidate precursor ions from the data collected in data independent acquisition (DIA) mode (all ion fragmentation). The fragmentation pathway was applied to structure interpretation, and the predicted retention time (RT) of the proposed structure was used to support the interpretation via comparing it with the observed RT.

2. Materials and Methods

2.1. Chemicals

Methanol (≥99.9%), acetonitrile (≥99.9%), and formic acid (≥99.0%) were purchased from Fisher Scientific (Ottawa, ON, Canada). Ammonium acetate (≥98%), glacial acetic acid (≥99.7%), β-glucuronidase from Escherichia coli (Type VII-A, lyophilized powder, 5292 U/mg), sodium phosphate monobasic (≥98%), and phosphoric acid (85% wt in H₂O, ≥99.999%) were purchased from Sigma-Aldrich (St. Louis, MO, USA). MilliQ deionized water (DIW) (resistivity 18.2 MΩ.cm, TOC <5ppb, 0.22 μm final filter) was made in-house with a MilliQ water system (Oakville, ON, Canada). The 36 phthalate metabolite standards and phthalic acid were purchased from different suppliers (CanSyn Chem. Corp. (neat compound), Santa Cruz Biotechnology (neat compound), Toronto Research chemicals (neat compound) or Cambridge Isotope Laboratories (100 μg/mL)), and their information is provided in Table S1. The stock solutions of each of the 14 isotope labelled standards (100 μg/mL, Table S1) were obtained from Cambridge Isotope Laboratories. Those neat compounds were prepared into 100 μg/mL stock solution by weighing 0.50 mg of each neat compound and dissolving it in 5 mL of methanol. The working mixture solution (2 μg/mL) was prepared by combining each reference solution (100 μg/mL) followed by evaporating the solvents with a gentle flow of nitrogen and reconstituting the residues in 20% acetonitrile in water.

2.2. Sample preparation

The study was approved by the Research Ethics Board of Health Canada (REB 2017–0023). The method validation samples were prepared from the pooled human urine that were collected from five volunteers by spiking phthalate metabolite standards in the urine. The spiked standards are indicated in Table S1, and the spiking levels in the pooled urine were Level-0 (0 ppb), Level-I (50 ppb), and Level-II (100 ppb). The sample preparation was consistent with the method reported previously.³⁰ Briefly, one millilitre of the trial urine sample was transferred to a 15-mL polypropylene tube and 250 μL of 1 M ammonium acetate buffer (pH = 6.5), and 100 μL of 2800 U mL⁻¹ β-glucuronidase solution were added to the sample. The mixture was incubated at 37 °C for 120 min in the Eppendorf Thermomixer 5430R (Eppendorf, Hamburg, Germany) to hydrolyze the conjugated metabolites. After incubation, samples were cooled to room temperature and cleaned up using the SPE method developed previously.³⁰ The analytes on the 60 mg/3 mL Oasis-HLB cartridge were eluted with 2 mL of acetonitrile followed by 2 mL of ethyl acetate, and the eluent was evaporated and reconstituted in 125 μL of DIW containing 30% acetonitrile (v/v) for sample analysis using the Agilent ultra high performance liquid chromatography quadrupole time-of-flight (UHPLC-QTOF) MS system. Blanks were conducted following the same procedure of urine sample preparation by using 1 mL of DIW water instead of 1 mL of urine.

2.3. UHPLC-MS conditions

All mass data acquisition was performed using an Agilent 6530 Accurate Mass Q-TOF mass spectrometer with an electrospray ionization source in negative mode (Santa Clara, USA), coupled with an Agilent 1290 infinity II UHPLC system (Santa Clara, CA, USA). The MassHunter version B.06.01software was used to control the Agilent LC-QTOF system in the data acquisition. A Kinetex biphenyl column (2.1 × 100 mm, 2.6 μm) (Phenomenex, Torrance, CA, USA) with a SecurityGuard ULTRA cartridge (2.1 × 2 mm) (Phenomenex was used for separation of compounds. Mobile phases were A (0.1% formic acid in DIW) and B (0.1 % formic acid in methanol). The separation was performed at a flow rate of 0.35 mL/min and the solvent gradient comprised the following linear steps: start at 15% B and hold for 1 minute; increase to 58% B over 2 minutes, and hold for 4.5 minutes; increase to 75% B over 1 minute and hold for 3.5 minutes; increase to 80% B over 1 minute and hold for 7 minutes; increase to 95% B over 1 minute and hold for 5 minutes; decrease back to 15% B over 2 minutes. The equilibrium time was 15 minutes before next injection. The injection volume was 10 μL. The column temperature was set at 40 °C. Different data acquisition modes were used for different sections of this work, details of which are described below.

2.4. Targeted MS/MS fragmentation data acquisition for phthalate metabolite standards

The data for fragmentation pattern/behaviour and pathway investigation were acquired in targeted MS/MS using the same spray settings for sample analysis mentioned above. The acquisition rate for MS scan and MS/MS scan were 10 spectra per second and 7 spectra per second, respectively. The scan range was set to 50 – 600 m/z for both MS and MS/MS scans. The temperature and flow rate of drying gas (nitrogen) were set at 325 °C and 10 L/min, respectively. The pressure of the nebulizer gas (nitrogen) was 35 psi, and the voltage of the capillary (Vcap) was set at 3500 V. The voltages of the fragmentor and the skimmer were 110 V and 65 V, respectively. Targeted MS/MS spectra were collected using a CE of 0, 10, 20 and 40 V with a resolving power of 2 GHz. The delta retention time for all targeted compounds was set to 0.5 minutes and the isolation width was set to “narrow” (~1.3 m/z). A mixture of standards that can be separated chromatographically was used to generate targeted MS/MS analysis data. Prior to each injection, the mass spectrometer was calibrated with Agilent ESI-L Low Concentration Tuning Mix as calibration solution to keep mass error below 0.8 ppm. Before running the samples, the mix standard solution of phthalate metabolites was used as the reference compounds to check the mass shifting.

2.5. Data acquisition for non-targeted analysis of urine samples

Non-targeted sample analyses were conducted using DIA mode to acquire the full scan fragmentation data in the mass range of m/z 50 to m/z 600. Specifically, for each urine sample, four injections were conducted to perform DIA under four different CE, i.e., 0 V, 10 V, 20 V, and 40 V with resolving power of 2 GHz. The acquisition rate was 3 spectra per second. All other conditions were the same as the conditions applied in target MS/MS fragmentation data acquisition.

2.6. Data analysis for non-targeted screening of unknown metabolites

Agilent MassHunter Qualitative Analysis software (version B.08) was used to perform the data analysis. Three specific diagnostic fragment ions (ChemBioDraw calculation), m/z 121.0295 C₇H₅O₂^-, m/z 147.0088 C₈H₃O₃^-, and m/z 165.0193 C₈H₅O₄^-, were used to generate an extracted ion chromatogram (EIC) and prioritize candidate precursor ions. Molecules that do not produce these three specific ions are not shown in the EIC and are therefore filtered out for further prioritization. The peak retention times and accurate masses observed were first manually compared with the retention times and the exact masses of 36 standards in the in-house library (Table S1) for the first step identification. Peaks that only contained the m/z 121 fragment were manually excluded from further consideration as this ion is non-specific to this group of chemicals. An ion intensity threshold of 10000 counts in the total ion chromatogram (TIC) was also manually applied. Following three steps were used to prioritize unknown candidate precursor ions. In the first step, the RTs of the peaks on the EICs were used to determine the RT of candidate precursor ions in the TIC at CE = 0. The mass spectrum of the CE=0 file at the RT where a peak appears on the EIC of another file was used to identify candidate precursor ions. Ions with the intensity higher than the threshold were considered as the candidate precursor ions. The second step was to compare the intensity of the potential precursor ions at different levels of CE at the same RT, based on the idea that as CE increases, the intensity of the precursor ion should decrease (i.e., precursor ion intensity in 0V > 10V > 20V > 40V). Concomitantly, the intensity of the phthalate indicator ions should increase, including m/z 121.0295, m/z 147.0088, m/z 165.0193, or/and other specific ions for different groups of phthalate metabolites listed in Table S1, such as m/z 134.0373 and m/z 77.0397 (ChemBioDraw calculation). This trend of decreasing intensity of potential candidate precursor ions and corresponding increasing intensity of associated fragment ions was used to identify phthalate precursor ions. The third step is to incorporate elemental composition restrictions for calculation of candidate precursor formulas with molecular features of phthalate metabolites to further filter out ions that do not follow the restrictions (see below). Finally, these identified precursor ions from formula calculation were acquired for targeted MS/MS spectra. Those spectra that do not contain two out of the three diagnostic fragment ions were further filtered out for their structure interpretation. When the structure of the candidate precursor ion was interpreted and proposed from the targeted MS/MS spectrum, the proposed structure was applied to the in-house RT prediction model³⁰ to predict their RTs. The predicted RT was compared with the observed retention time as additional complementary supporting evidence to increase the identification confidence of structures interpreted from the MS/MS spectra. The elemental composition restrictions for the molecular feature of monophthalates used in formula calculation include carbon: 8–30, hydrogen: 6–50, oxygen: 4–10, other elements: 0, negative ion: H, charge: −1, MS ion electron state: allow both even and odd, minimum overall score per charge carrier: 35, maximum MS mass error: 10 ppm, double bond equivalent (DBE): 6–12, mass score: 100, isotope abundance score: 60, isotope spacing score: 50, retention time score: 100, expected data variation: MS 2.0 mDa + 10 ppm, and match score: ≥ 85.

3. Results and discussion

The identification of product ions whose peak intensities increased in correlation with CE allowed the phthalate metabolites to be classified into four groups with distinct fragmentation patterns: pure alkyl chains (linear and branched), oxidized alkyl chains with a carbonyl group, oxidized alkyl chains with an oxo group, and oxidized alkyl chains with a hydroxyl group. Table 1 lists these fragments that are presented in reference to the CE used.

Table 1:

Summary of common fragments (m/z) shared by phthalate metabolites in the same class and observed trends in their fragmentation patterns

Phthalate monoester class	Common fragment mass (m/z)	Formula of fragment	Fragmentation pattern trend
pure alkyl chains (linear and branched) (Group I)	77.0397 121.0295 134.0373 147.0088	C6H5- C7H5O2- C8H6O2- C8H3O3-	• m/z 77 is the most abundant fragment at CE = 20 and 40. • there is an increasing ratio of [M-H]-/[C6H5]- at CE=10 with the increase in alkyl chain length • there is an decreasing ratio of [M-H]-/[C7H5O2]- and [M-H]-/[C8H5O3]-at CE=10 with the increase in alkyl chain length • The ratio of [C8H6O2]-/[C7H5O2]- and/or [C8H6O2]-/[C8H5O3]- at CE=10 for phthalate metabolites with branched alkyl chain are > 1. • fragmentation pattern is characterized at low collision energy by [M-H]-, [M-H-CO2]-, followed by [M-H-CO2-CO]-
oxidized alkyl chains with a carbonyl group? (Group II)	77.0397 121.0295 165.0193	C6H5- C7H5O2- C8H5O4-	• there is an increasing ratio of [M-H]-/[C7H5O2]-, and [M-H]-/[C6H5]- at CE=10 with the increase in chain length • fragmentation pattern is characterized by [M-H]-, [M-H-C8H4O3]-, followed by [C8H5O4]- • fragmentation pattern is characterized by [OR-X]-, where X = 46 (CO2+H2 or H2O+CO)
oxidized alkyl chains with an oxo group (Group III)	77.0397 121.0295 147.0088	C6H5- C7H5O2- C8H3O3-	• m/z 121 is the most abundant fragment at CE = 20. • there is an increasing ratio of [M-H]-/[C7H5O2]- and [M-H]-/[OR]- at CE=10 with the increase in chain length • fragmentation pattern is characterized by [M-H]-, [M-H-CO2]-, followed by [M-H-C8H4O3]- • fragmentation pattern is characterized by [OR-X]-, where X = 30, 44, 58, 72, 86…
oxidized alkyl chains with a hydroxyl group (Group IV)	77.0397 121.0295 147.0088	C6H5- C7H5O2- C8H3O3-	• m/z 121 is the most abundant fragment at CE = 20. • there is an increasing ratio of [M-H]-/[C7H5O2]- and [M-H]-/[OR]- at CE=10 with the increase in chain length • fragmentation pattern is characterized by [M-H]-, [M-H-C8H4O3]- • fragmentation pattern is characterized by [OR-X]-, where X = 18, 32, 46, 60, 74…

3.1. The MS/MS fragmentation patterns for phthalate metabolites with pure alkyl chains (linear and branched).

The majority of pure alkyl chain phthalate metabolites produced fragment ions at m/z 147.0088 and m/z 121.0295, but did not produce m/z 165.0193 as shown in Table 1. Monomethyl isophthalate did not produce a fragment ion at m/z 147.0088 (deprotonated o-phthalic anhydride), likely due to the carboxylic acid group in the meta position relative to the ester group inducing in the formation of a stable o-phthalic anhydride in the gas phase. Table S1 shows that for the most part, phthalates with a pure alkyl chain are susceptible to fragment easily leading to a neutral loss of CO₂ ([M-H-44]) even at CE = 0. While higher CE is required to fragment the heavy phthalate metabolites with longer alkyl chains, phthalate metabolites with a short alkyl chain can produce more fragments even with zero CE. For example, at CE = 0 monomethyl phthalate (MMP) fragments to produce a product ion at m/z 135 with further loss of CO and CH₃ to form a product ion at m/z 107. For monoethyl phthalate (MEP), one fragment is generated by the loss of CO₂ and another by the loss of CH₂ to form the deprotonated o-phthalic anhydride ion at m147. Overall, the majority of pure alkyl chain phthalate metabolites fragment to produce an ion at m/z [M-H-44] through a loss of carbon dioxide and an ion at m/z [OR-2] through losses of a phthalic anhydride and a hydrogen; the “OR” refers to the deprotonated O–alkyl chain (alkoxy group) of phthalate metabolites. Mono-8-methylnonyl phthalate (C₁₈H₂₆O₄) and mono-3,7-dimethyl-1-octyl phthalate (C₁₈H₂₆O₄), both larger compounds with masses over 300 Da, were the only metabolites within this group where fragmentation was not observed at CE = 0. Figure 1 is a typical example using the fragmentation behaviours of mono-n-octyl phthalate (C₁₆H₂₂O₄) (Figure 1A) and mono-3,7-dimethyl-1-octyl phthalate (C₁₈H₂₆O₄) (Figure 1B) to explain the fragmentation pathway of this group of chemicals. Figure 1C is the proposed fragmentation pathway for all phthalate metabolites with a pure alkyl chain that are summarized in Table S1. The pathway suggests that all phthalate metabolites with a pure alkyl chain follow the fragmentation pathway to lose the O–alkyl chain to form the deprotonated o-phthalic anhydride ion at m/z 147. A previous report proposed that the deprotonated molecular ions at m/z [M-H]^- first lose the R group (alkyl/aryl chain) to form the [M-H-R]^- ion at m/z 165 and the carbonyl oxygen attacks the ortho-carbonyl carbon to produce the deprotonated o-phthalic anhydride ion at m/z 147 through the loss of OH.²⁸ However, the results in this study demonstrate that the formation of the deprotonated o-phthalic anhydride ion ([C₈H₃O₃]^-) was not through this pathway, which was also confirmed by the fragmentation of phthalic acid and ¹³C labelled standards that are listed in Table S1 and shown in Figure 1A and 1B. The deprotonated o-phthalic anhydride ion ([C₈H₃O₃]^-) can further lose a molecule of CO to form the deprotonated benzoate ion [C₆H₅COO]^- at m/z 121, which further loses a molecule of CO₂ to form the deprotonated C₆H₆ ion at m/z 77. As shown in Table S1, all phthalate metabolites with pure alkyl chains can produce the deprotonated radical 2-methylbenzoic acid [C₆H₅COO]^- m/z 134 by losing a molecule of carbon dioxide from the deprotonated benzoate ester ion at m/z [M-H-44]^-, except light chain phthalate metabolites. In brief, the ortho radical can also attack the carbon of the deprotonated benzoate ester, which makes the deprotonated benzoate ester ion lose either a molecule of CO to form the ion at m/z [M-H-72], or the alkyl chain R’ to form the deprotonated radical 2-methylbenzoic acid ion at m/z 134. The ion at m/z [M-H-72] can further lose the side alkyl chain R’ to form the deprotonated benzene alcohol ion ([C₇H₇O]^-) at m/z 107. The deprotonated benzene alcohol ion ([C₇H₇O]^-) at m/z 107 can further lose H₂ to form the ion ([C₇H₅O]^-) at m/z 105 and can further process a loss of CO to form the deprotonated C₆H₆ ion at m/z 77. The deprotonated benzoate ester ion can also further lose H₂ through rearrangement to form the ion at m/z [M-H-46]^-. The further loss of CO from the ion at m/z [M-H-46]^- forms the ion at m/z [M-H-74]^- as shown in Figure 1C. The fragmentation behaviours of this group chemicals also show that the alkyl/aryl oxygen attacks the ortho-carbonyl carbon of the carboxylic acid resulting in the loss of C₈H₅O₃ (m/z 147) to form the ion [OR-2] and the ion [OR-4] by loss of H₂ (Figure 1C).

Figure 1.

The fragmentation behaviours and pathway of isotopically labeled phthalate metabolites with a non-oxidised alkyl chain. (A), The fragmentation behaviours of mono-n-octyl phthalate at CE = 10 V; (B), the fragmentation behaviours of mono-3,7-dimethyl-1-octyl phthalate at CE = 20 V; (C) the proposed fragmentation pathway for all phthalate metabolites with pure alkyl chains (linear and branched).

3.2. The MS/MS fragmentation patterns for phthalate metabolites with oxidized alkyl chains with a carbonyl group

In contrast to phthalate metabolites with a pure alkyl chain, phthalate metabolites with a carboxyl alkyl chain do not produce the deprotonated o-phthalic anhydride ion ([C₈H₃O₃]^-) at m/z 147, but produce the [M-H-R]^- ion at m/z 165 and the deprotonated benzoate ion [C₆H₅COO]^- at m/z 121 instead as shown in Table S1. This was also confirmed by the fragmentation of the ¹³C labelled compounds (Table S1) and a typical example using the fragmentation behaviors of mono-3-carboxypropyl phthalate (C₁₂H₁₂O₆) in Figure 2A and mono-2-(carboxymethyl) hexyl phthalate (C₁₆H₂₀O₆) in Figure 2B. At CE = 0, an ion at m/z [OR], which refers to the deprotonated O–alkyl chain (carboxylate alkoxy group), was observed for all chemicals in this group, and an ion at m/z 165 was observed only for a few metabolites in this group. This suggests that the ortho-carbon-oxygen bond is easily broken by attacking the carbonyl oxygen to form the deprotonated alkoxy group. However, some CE is required to break the alkyl-oxygen bond of the alkoxy group to form a deprotonated phthalic acid ion at m/z 165; only a few phthalate carboxylate metabolites fragmented to form this fragment ion at CE = 0, including mono-3-carboxypropyl phthalate (C₁₂H₁₂O₆) and mono-(2-(carboxymethyl)) hexyl phthalate (C₁₆H₂₀O₆). This typical fragmentation pattern can be useful in identification of unknown compounds in this group. In summary, the fragmentation pathway of this group of phthalate metabolites is proposed as Figure 2C. All metabolites of this group can form the deprotonated O–alkyl chain ion at m/z [OR] by attacking of the alkyl/aryl oxygen to the ortho-carbonyl carbon of the carboxylic acid. The deprotonated O–alkyl chain ion further loses a molecule of hydrogen to form the ion [OR-2] that further loses CO₂ to form the ion [OR-46]. The deprotonated O–alkyl chain ion can also lose CO₂ to form the ion [OR-44] followed by losing a molecule of hydrogen to form the ion [OR-46]. The fragmentation behaviour of this group of phthalate metabolites indicate that the deprotonated O–alkyl chain ion can also lose a water molecule to form the ion at m/z [OR-18] which further loses a CO to the ion at m/z [OR-46]. For this group of phthalate metabolites, the deprotonated benzoate ion [C₆H₅COO]^- at m/z 121 is likely produced by the loss of the OR group and a CO₂, which is different from the pathway for the phthalate metabolites with a pure alkyl side chain. Similarly, the ion at m/z 77 forms from the loss of a carbon dioxide molecule. The formation of the deprotonated phthalic acid ion [M-H-R]^- at m/z 165 through the loss of the carbonyl alkyl chain is unique to this group of phthalate metabolites. This fragmentation pathway was also supported by the fragmentation behaviours of ¹³C labelled standards in this group metabolites by observing the corresponding ¹³C labelled ions such as ions m/z 169 and m/z 124 listed in Table S1.

Figure 2.

(A) The fragmentation behaviours of mono (3-carboxypropyl) phthalate at CE = 10 V; (B) The fragmentation behaviours of mono-2-(carboxymethyl) hexyl phthalate at CE = 20 V; (C) the proposed fragmentation pathway of phthalate metabolites with oxidized alkyl chains with a carbonyl group.

3.3. The MS/MS fragmentation patterns for phthalate metabolites with oxidized alkyl chains with an oxo group.

The MS/MS fragmentation behaviours of phthalate metabolites with an oxo alkyl chain show similar primary fragments to the non-oxidative and carboxylate metabolites, but unique secondary fragments were observed in both the native and labelled compounds (Figure 3 and Table S1). We use the spectra of mono-(2-ethyl-5-oxohexy) phthalate (C₁₆H₂₀O₅) in Figure 3A and mono-(2-propyl-6-oxo-heptyl) phthalate (C₁₈H₂₄O₅) in Figure 3B as an example to analyze the fragmentation behaviors of this group compounds. Specifically, at CE = 0, the attaching of carbonyl oxygen to the ortho-carbonyl carbon forms the deprotonated O–alkyl chain ion at m/z [OR] due to the carbonyl oxygen attacking the ortho-carbonyl carbon. The carboxyl oxygen can also attack and break the C-O bond of the ortho-carbonyl carbon at CE = 0 and result in a neutral loss of COOR (or C₂H₃OOR″ in Figure 3C) to form the deprotonated benzoate ion [C₆H₅COO]⁻, at m/z 121. This indicates that ions [C₆H₅COO]^- and [OR] are primary fragments at CE = 0. This is further supported by the observation of ions m/z 143 and m/z 124 from the spectrum of ¹³C labelled mono-(2-ethyl-5-oxohexy) phthalate (C₁₆H₂₀O₅) at CE = 0 as listed in Table S1. When the CE was increased from zero to 10, the carboxyl oxygen attacking the ortho-carbonyl carbon causes the loss of CO₂ to form the ion at m/z [M-H-44]. At higher CE, the deprotonated benzoate ion [C₆H₅COO]^- then loses CO₂ to form the deprotonated benzene ion ([C₆H₅]^-) at m/z 77. The deprotonated O–alkyl chain ion at m/z [OR] undergoes a neutral loss of CH₂O to form the ion at m/z [OR-30], which is unique to this group of chemicals. The ion at m/z [OR-30] further loses CH₂ to form the ion at m/z [OR-44], which then loses CH₂ to form the ion at m/z [OR-58] and continue to lose CH₂ to form the ion at m/z [OR-72]. This ion further loses CH₂ to form the ion at m/z [OR-86] and keep losing of CH₂ to form ions up to [OR-114]. Interestingly, when the CE was increased up to 20, the deprotonated o-phthalic anhydride ion ([C₈H₃O₃]^-) at m/z 147 was observed, indicating that it requires more energy to lose the O–alkyl chain to form the deprotonated o-phthalic anhydride ion compared to phthalate metabolites with a pure alkyl chain.

Figure 3.

(A) The fragmentation behaviour of mono-(2-ethyl-5-oxohexyl) phthalate at CE = 10 V; (B) The fragmentation behaviour of mono-(2-propyl-6-oxo-heptyl) phthalate at CE = 20 V; (C) The fragmentation pathway of phthalate metabolites with oxidized alkyl chains with an oxo group.

3.4. The MS/MS fragmentation patterns for phthalate metabolites with oxidized alkyl chains with a hydroxyl group.

Similar to phthalate metabolites with an oxo alkyl chain, phthalate metabolites with oxidized alkyl chains with a hydroxyl group also generated the deprotonated benzoate ion, [C₆H₅COO]^- at m/z 121 and the deprotonated O–alkyl chain ion at m/z [OR] when CE = 0. This pathway was also observed by the presence of the two ions at m/z 145 and m/z 121, respectively in the spectrum of mono-(2-ethyl-5-hydroxyhexyl) phthalate (C₁₆H₂₂O₅) in Figure 4A and the presence of the two ions at m/z 173 and m/z 121 in the spectrum of mono-(2-propyl-6-hydroxy-heptyl) phthalate in Figure 4B, respectively. This result was also confirmed by the presence of m/z 124 in the spectra of the ¹³C labelled standards (Table S1). However, upon increasing the CE, this group of chemicals fragment via a similar pathway to the carboxylate phthalate metabolites (not the oxo phthalate metabolites) to form the ion at m/z [OR-18] through the loss of H₂O from the deprotonated O–alkyl chain ion, and the ion at m/z [OR-2] from the deprotonated O–alkyl chain ion by a loss of H₂. The only difference is that the ion [OR-18] can further lose CH₂ to form the ion at m/z [OR-32] as shown in Figure 4C. With high CE, the deprotonated o-phthalic anhydride ion at m/z 147 can be observed, indicating that oxidized alkyl chain can hinder the loss of the O–alkyl chain to form the deprotonated o-phthalic anhydride ion. Compared to phthalate metabolites with an oxo alkyl chain, in addition to the specific fragment ions mentioned above, the spectra of this group of metabolites perform a continuous loss of CH₂ after to the loss of 46 from the deprotonated O–alkyl chain ion m/z [OR] to form a pattern of fragment ions m/z [OR-46], m/z [OR-60], m/z [OR-74] and m/z [OR-86] as shown in Table S1. This specific fragmentation pattern can be used to identify and classify this group in non-targeted analysis. However, we noticed that not all of this series ions can be clearly observed in the spectra of every compound in this group.

Figure 4.

(A) The fragmentation behaviour of mono-(2-ethyl-5-hydroxyhexyl) phthalate at CE = 20 V; (B) The fragmentation behaviour of mono-(2-propyl-6-hydroxy-heptyl) phthalate at CE = 10 V; (C) The fragmentation pathway of phthalate metabolites with oxidized alkyl chains with a hydroxyl group.

3.5. Determination of MS fragmentation pattern

The above findings on fragmentation behaviours of four groups of phthalate metabolites suggest that varying CEs produced unique fragmentation patterns for each class of phthalate metabolites, with the three fragment ions at m/z 121.0295 C₇H₅O₂^-, m/z 147.0088 C₈H₃O₃^-, and m/z 165.0193 C₈H₅O₄^- being diagnostic for specific classes as shown in Table 1. Understanding of these common fragments and fragmentation patterns is valuable in determining the structure of unknown phthalate metabolites in non-targeted screening analyses. Moreover, these characteristic fragments can also be applied to prioritize/determine the number of phthalate metabolites present in the sample. In this study, fragment ions at m/z 121.0295 (corresponding to the deprotonated benzoate ion [C₆H₅COO]^-), a common fragment shared by all compounds with a phthalic acid structure, and m/z 147.0088 (corresponding to the deprotonated o-phthalic anhydride ion [C₆H₃C₂O₃]^-), the common fragment ion shared by all 1,2-benzenedicarboxylic monoesters, were extracted to determine how many compounds with the basic structure of phthalate metabolites were present in the sample. A fragment ion at m/z 165.0193 (corresponding to the deprotonated phthalic acid [C₆H₄COOHCOO]^-) was extracted to determine how many phthalate metabolites containing a carbonyl alkyl chain were present within the sample. After extraction of the specified fragment ions, the number of peaks observed in the chromatogram is correlated with the number of phthalate metabolites in the sample sharing these common ions. Further investigation of the fragmentation patterns can reveal the structure of the unknown phthalate metabolites, as described in Table S2, S3 and S4. Importantly, fragmentation patterns for phthalate metabolite class reported in this study (Table 1) can be used to aid in future identification of legacy phthalate metabolites as well as new emerging phthalates that are being introduced in the environment.

3.6. Application in non-targeted analysis

Pooled human urine samples with and without 24 spiked phthalate metabolites (listed in Table S1) were used to evaluate and demonstrate the application of fragmentation pathway and patterns in non-targeted prioritization and classification of unknown phthalate compounds in the sample. The DIA mode was used to collect all-ion fragmentation under four collision energies, CE=0, CE=10, CE=20, and CE=40, respectively. Figure S1 shows EICs of the three fragment ions common and diagnostic of phthalate metabolites (m/z 121.0295, 147.0088, 165.0193) from the pooled urine sample spiked with 100 ppb of 24 metabolite standards in the urine at CE = 10. Each EIC shows distinct peaks observed throughout the chromatogram, each corresponding to a specific observed phthalate metabolite, demonstrating that the proposed diagnostic ions can filter out all chemicals that do not belong to this group of chemicals at this CE. Figure 5 shows the intensity change of EIC peaks under different CEs, which can also be used to filter or prioritize candidate precursor ions. Following the filtering criteria described in the method section, candidate precursors in each sample were identified and listed in Table S2, S3, and S4. Among those candidate precursors, 3 out of 26, 22 out of 45, and 22 out of 45 were confirmed at a Level 1 identification³¹ as monophthalates in Level-0. Level-I, Level-II, respectively by matching the retention time, exact masses, and fragment ions with the targeted analysis results of phthalate metabolite standards. When mass error restriction is set as 20 ppm, MMP was observed in Level-II but not Level I, suggesting the large mass error for small m/z molecules for the QTOF used in this study. Monomethyl isophthalate was not observed in this study but it is not surprising as it does not follow the proposed fragmentation pathways. The results indicate that putative phthalate metabolites in urine present at concentration levels around 50 ppb, except MMP and monomethyl isophthalate (MMiP), can be detected using this approach. Among those ions that were not confirmed in Tables S2, S3, and S4, those ions that can only be observed in the EIC peaks of m/z 121 were excluded for further structure elucidation.

Figure 5.

A typical chromatogram for TIC and EIC of m/z 121.0295 at different CEs. Sample: pooled urine sample spiked with 100 ppb standard metabolites. A: TIC Scan (CE=0V); B: EIC at m/z 121.0295, CE=10V; C: EIC at m/z 121.0295, CE=20V; D: EIC at m/z 121.0295, CE=40V.

In this study, only those ions that were observed in all three urine samples (Level-0, Level-I, and Level-II) were applied for targeted MS/MS analysis for structure interpretation to increase the NTA detection repeatability. The major fragment ions of eight candidate precursor ions in the targeted MS/MS spectra were listed Table S2. The eight candidate precursor ions included m/z 223.0601 (Peak #2, retention time (RT) = 4.12 min), m/z 271.0594 (Peak #6, RT = 4.95 min), m/z 263.1273, m/z 297.1118, and m/z 343.1183 (Peak #8, RT = 5.52 min), m/z 249.1117 (Peak #9, RT = 5.84 min), and m/z 307.1178 and m/z 263.1281 (Peak #20, RT = 8.97 min). According to the method developed in this study, only suspect ions whose MS/MS spectrum contain at least two out of the three diagnostic ions, m/z 121.0295, m/z 147.0088, and m/z 165.0193, belong to the group of monophthalates and hold a base structure of monophthalates. Otherwise, they do not belong to this group of chemicals and should be removed from further structure interpretation (i.e., Peaks 1, 3, 4, 7, 10, 14, 17, 23, and 25 in Table S2). The targeted MS/MS spectra showed that the precursor ion m/z 263 at 5.52 min and m/z 307 at 8.97 min contain only one diagnostic ion (m/z 165) while the precursor ions m/z 223 at 4.12 min, m/z 249 at 5.84 min, and m/z 263 at 8.97 min produce no diagnostic ions. This might indicate that the intensity of the actual precursor ions is lower than the set threshold intensity value when picking precursor ions from the TIC. This also suggests that these unknown features do not possess structural features of phthalate metabolites although they were passed through the filters applied in this study, and could result in false positives without targeted experiments. Interestingly, the MS/MS spectra of the precursor ions m/z 297.1118 and m/z 343.1183 contain two false diagnostic ions of m/z 165 and m/z 121; their accurate mass were m/z 165.0541 and m/z 121.0661, m/z 165.0518 and m/z 121.0649, respectively. However, they are not the true diagnostic ions; m/z 165.0193 and m/z 121.0295 are in the range of mass error 10 ppm. This suggests that these two unknown precursor ions m/z 297.1118 and m/z 343.1183 do not possess the structural features of monophthalates, and therefore, they were not considered for further structure interpretation either. As a result, only the precursor ion m/z 271.0584 at 4.95 was considered to possess the structural features of phthalate metabolites. The targeted MS/MS analysis of the ion m/z 271.0594 at 4.95 min produced ions at m/z 253, 227, 165, 151, 137, and 121 as shown in Figure 6 and Table S2. A loss of H₂O (18) from precursor ion m/z 271 formed the ion m/z 253, indicating OH on the alkyl chain. The loss of CO₂ formed the ion m/z 227. The DBE for this compound is 10, indicating that the alkyl chain contains three double bonds. Although the precursor ion produced the diagnostic ion m/z 165, which is the specific ion of monophthalates that have oxidized alkyl chains with a carbonyl group, the high unsaturated chain of this compound altered the fragmentation behavior. Therefore, we proposed this compound as mono-2-((2-cyclopenta-2,4-dien)-1-hydroxyethyoxy) phthalate (C₁₅H₁₂O₅). With the RT prediction model,³⁰ the predicted RT is 4.924 min, which is close to the observed RT of 4.95 min for this proposed structure, supporting the likelihood of the proposed structure. This is a new phthalate metabolite that was not previously reported.

Figure 6.

The MS/MS spectrum of unknown precursor ion m/z 271.0594 at 4.95 min, collision cell energy: CE=20V.

4. Conclusion

To reduce false positives and increase the identification confidence in non-targeted analyses, prioritizing precursor ions from the large amount of HRMS data is a key step. Understanding fragmentation pathways of a certain group of chemicals is helpful in order to prioritize candidate precursor ions. The findings of this study indicate that the structural similarity of phthalate metabolites gives three typical common fragment ions, m/z 121.0295, 147.0088, 165.0193 in MS spectra at certain CEs, which can be used as a filter to prioritize candidate precursor ions as phthalate metabolites. Further analysis of the fragmentation behaviours of prioritized precursor ions can be useful for further classifying and identifying unknown phthalate metabolites. The fragmentation pathway workflow proposed in this study can quickly and reliably prioritize unknown precursor ions in suspect analyses and minimize the number of possible structures for compound identification in non-targeted analysis of new phthalate metabolites in human urine, as well as potentially other biological matrices. The assay can detect unknown phthalate metabolites at concentrations around 50 ppb in urine. While eight unknown precursor ions were putatively identified in the pooled urine sample using the first step of the proposed prioritization scheme, only one ion follows the fragmentation pathway observed for known phthalate metabolites. While the number is low, the method demonstrates how the combination of accurate mass and knowledge of the fragmentation pattern prevented us from misannotating unknown features that may at first appear to be phthalate metabolites. Utilization of RT prediction and MS/MS data for structure interpretation resulted in the tentative identification of a previously unreported metabolite in human urine. Further study on its toxicity and exposure scenario is recommended.