Comprehensive analysis of resveratrol metabolites in rats using ultra high performance liquid chromatography coupled with high resolution mass spectrometry

2.1 Chemicals and materials

Resveratrol and polydatin were purchased from Chengdu Must Biotechnology Co. Ltd (Sichuan, China). These reference standards with purities higher than 98% were applicable to UHPLC-LTQ-Orbitrap analysis.

HPLC grade acetonitrile, methanol and formic acid were purchased from Thermo Fisher Scientifific (Fair Lawn, NJ, USA). All the other chemicals of analytical grade were available at the work station, Beijing Chemical Works (Beijing, China). Deionized water used throughoutthe experiment was purifified by a Milli-Q Gradient Å 10 System (Millipore, Billerica, MA, USA). Grace Pure™ SPE C₁₈-Low solid phase extraction cartridges (200 mg/3 mL, 59 μm, 70 Å) were purchased from Grace Davison Discovery Science (Deerfifield, IL, USA).

2.2 Animals

Eight male SD rats weighted 200 ± 20 g were obtained from Beijing Weitong Lihua Experimental Animals Company (Beijing, China). The rats were housed in a controlled room at standard temperature (24 ± 2 °C) and humidity (70 ± 5%), and kept on a 12 h light/12 h dark regime. After a week acclimation, rats were randomly divided into two groups: Drug Group (n = 4) for test plasma, urine and feces; Control Group (n = 4) for blank plasma, urine and feces. They were fasted for 12 h with free access to water prior to the experiment. The animal protocols were approved by the institutional Animal Care and Use Committee at Beijing University of Chinese Medicine. The animal facilities and protocols were complied with the Guide for the Care and Use of Laboratory Animals.

2.3 Drug administration and biological sample preparation

Resveratrol was suspended in 0.5% carboxymethylcellulose sodium (CMC-Na) solution and given at a dose of 300 mg/kg body weight to rats in the drug group. A 0.5% CMC-Na aqueous solution (2 mL) was parallelly administered to rats in the control group. Blood samples (0.5 mL) were taken from the suborbital venous plexus of rats at 0.5, 1, 2 and 4 h post-administration. Each sample was centrifuged at 3 500 rpm for 10 min to obtain plasma samples. Urine and faecal samples were collected 0–24 h after intragastric administration. All homogeneous biological samples from the same group were finally merged into a collective sample.

Different processing methods were used to pretreat the collected blood, urine and faecal samples. The first method (method I) was performed to prepare biological samples by solid phase extraction (SPE). Plasma and urine samples (1 mL) were added to an SPE cartridge pretreated with methanol (5 mL) and deionized water (5 mL). Then, the SPE cartridges were successively washed with deionized water (5 mL) and methanol (3 mL). The methanol eluate was collected and evaporated by nitrogen at room temperature. The residue was redissolved in 100 μL of acetonitrile/water (10:90, v/v) and then centrifuged at 14,000 rpm for 15 min. Faecal samples (1.0 g) were ultrasonically extracted with deionized water (5.0 mL) for 15 min and then filtered. The supernatant (1 mL) was added to the pretreated SPE cartridge, and then the same process described above was conducted. The second method (method II) used methanol to precipitate the supernatants of the faeces, blood and urine samples. The proportion of methanol to these samples was 1:3. These samples were precipitated for 30 min and then centrifuged at 14,000 rpm for 15 min to obtain the solutions after treatment. The third method (method III) used acetonitrile to precipitate the above biological samples. The ratio of acetonitrile to sample was also one to three, and then the same process described in method II was carried out. All the supernatants were used for further instrumental analysis.

2.4 Instrument and conditions

LC analysis was performed on a DIONEX Ultimate 3000 UHPLC system (Thermo Electron, Bremen, Germany), equipped with a binary pump and an auto-sampler. The chromatographic separation was carried out using Waters ACQUITY BEH C18 column (2.1 × 100 mm i.d., 1.7 μm; Waters Corporation, Milford, MA, USA) at 35 °C. Methanol (solvent B) and 0.1% formic acid aqueous solution (solvent A) were used as mobile phase. The flow rate was set to 0.3 mL/min with a linear gradient as follows: 0–2 min, 25% B; 2–11 min, 25–55% B; 11–17 min, 55–95% B. The injection volume was 2 μL.

High resolution ESI-MS and MS/MS spectra were obtained using LTQ-Orbitrap MS (Thermo Electron, Bremen, Germany) connected to the UHPLC instrument via ESI interface. Samples were analyzed in negative ion mode with the tune method set as follows: sheath gas (nitrogen) flow rate of 40 arb, auxiliary gas (nitrogen) flow rate of 20 arb, capillary temperature of 350 °C, spray voltage of 4.0 kV, capillary voltage of 25 V, tube lens voltage of 110 V. Metabolites were detected using full-scan MS analysis from m/z 50-800 at a resolving power of 30,000 FWHM. The collision energy for collision induced dissociation (CID) was adjusted to 40% of maximum. In addition, the PIL-dependent acquisition mode was employed as a complementary tool to obtain the ESI-MS/MS dataset of potential metabolites.

2.5 Peak selections and data processing

Thermo Xcalibur 2.1 workstation was used for data acquisition and processing. In order to obtain as many ESI-MS/MS fragment ions of resveratrol metabolites as possible, the peaks detected with intensity over 10,000 were selected for identification. The accurate mass of chemical formulas attributed to all parent ions of the selected peaks were calculated using a formula predictor by setting the parameters as follows: C [0–35], H [0–40], O [0–16], S [0–1] and ring double bond (RDB) equivalent value [0–14]. Accurate mass measurements were set within mass error of ±5 ppm.

3.1 The establishment of analytical strategy

In this study, all the collected biological samples (urine, plasma and faeces) were prepared using the three abovementioned methods, which included solid phase extraction, methanol precipitation and acetonitrile precipitation. Then the samples were injected into UHPLC-Q-Exactive Orbitrap MS to gain the HRMS data with full MS scanning acquisition. Then, data mining was processing based on common biotransformation reactions as well as the reported metabolites in literature. And subsequently, the ions which we were interested in were put into PIL to obtain more comprehensive MS2 information for structural identification. Among them, polydatin, dihydroresveratrol and pterostilbene were also regarded as pivotal candidate metabolites and filled into PIL. Finally, an efficient analysis based on UHPLC-LTQ-Orbitrap MS coupled with FS-PIL-DE was proposed to characterize and screen the metabolic products. Combining the chromatographic retention behaviours, accurate mass measurements, mass fragmentation patterns and previous relevant literature, was conducive to the comprehensive analysis of the metabolites of the resveratrol representative chemical components. In addition, a diagram that summarizes the presently developed analytical strategy and methodology for the detection and identification of resveratrol metabolites is shown in Fig. 1.

Close

Fig. 1

The summary diagram of analytical strategy and methodology.

Export to PPT

3.2 Fragmentation patterns analysis of resveratrol

In order to better characterize the in vivo metabolites, DPIs were essential for structural identification. Resveratrol was selected as the subject to determine its DPIs based on the comprehensive ESI-MSⁿ information of the resveratrol standard via UHPLC-HRMS analysis. For example, resveratrol gave rise to deprotonated [M−H]⁻ ions at m/z 227.07057, as shown in Fig. 2. Then, it would further afford a series of DPIs at m/z 185, 159 and 143 owing to the loss of C₂H₂O (42 Da), C₂H₂O + C₂H₂ (68 Da) and 2C₂H₂O (84 Da) (Wang et al., 2005; Mei et al., 2019). Therefore, the product ions of resveratrol metabolites at m/z 185 + X, m/z 159 + X and m/z 143 + X (X = molecular weight of the substituent groups), such as 14 (CH₂), 80 (SO₃), and 176 (GluA), could also be observed in the ESI-MS/MS spectra.

Close

Fig. 2

The proposed fragmentation pathways of resveratrol.

Export to PPT

3.3 Identification of resveratrol metabolites

A total of 56 resveratrol metabolites (resveratrol included) were detected and characterized from rat urine, plasma and faecal samples by means of the UHPLC-LTQ-Orbitrap method coupled with the established strategy. Among them, 39 metabolites were found in rat urine, 6 metabolites were detected in rat plasma, and 16 metabolites were characterized from rat faeces. The correlative data in negative ion mode are summarized in Table 1.

Metabolites N1, N2, N25, N29 and N34 possessed deprotonated molecular ions at m/z 243.06580, m/z 243.06572, m/z 243.06593, m/z 243.06587, and m/z 243.06572 (C₁₄H₁₁O₄) with mass errors of 2.529, 2.200, 3.064, 2.817 and 3.269 ppm, respectively. Fragment ions at m/z 201 [M−H−C₂H₂O]⁻, m/z 175 [M−H−C₂H₂O−C₂H₂]⁻ and m/z 159 [M−H−2C₂H₂O]⁻ were all observed in the ESI-MS² spectra. In addition, their [M−H]⁻ ions at m/z 243 were 16 Da more than that of resveratrol, which implied the occurrence of a hydroxylation reaction. Thus, N1, N2, N25, N29 and N34 were plausibly assigned as hydroxylated resveratrol.

N3 and N4 afforded [M−H]⁻ ions at m/z 137.02394 and m/z 137.02321 (C₇H₅O₃, mass errors 4.521 ppm and −0.807 ppm), respectively. They generated ESI-MS² DPIs at m/z 93 and m/z 109 due to the neutral loss of CO₂ and CO, which indicated the presence of carboxyl groups. Therefore, N3 and N4 could be deduced as isomeric 4-hydroxybenzoic acid.

N5 was 194 Da greater than that of dihydroresveratrol, which displayed an [M−H]⁻ ion at m/z 421.11301 with a mass error of 0.206 ppm. In its ESI-MS² spectrum, the deprotonated molecular [M−H]⁻ ion at m/z 421 further generated several product ions at m/z 245 [M−H−GluA]⁻ and m/z 175 [GluA-H]⁻. Based on this, N5 was identified as glucuronidation and hydroxylation product of dihydroresveratrol.

Six isomeric constituents, N6, N12, N13, N17, N22 and N48, gave rise to [M−H]⁻ ions at m/z 323.02252, m/z 323.02237, m/z 323.02264, m/z 323.02255, m/z 323.02261 and m/z 323.02136 (C₁₄H₁₁O₇S) with mass errors of 1.610, 1.146, 1.982, 1.703, 1.889 and −1.981 ppm, respectively. In the ESI-MS² spectra, the characteristic product ion at m/z 243 was generated by the loss of 80 (SO₃) from the deprotonated molecular [M−H]⁻ ion at m/z 323. Moreover, N6, N12, N13, N17, N22 and N48 were 96 Da more than that of resveratrol. Therefore, N6, N12, N13, N17, N22 and N48 were isomeric sulfated and hydroxylated metabolites of resveratrol.

N7, N8 and N19 showed [M−H]⁻ ions at m/z 565.15515, m/z 565.15533 and m/z 565.15515 (C₂₆H₂₉O₁₄, mass error −0.056, 0.262 and −0.056 ppm), respectively.

These peaks were 338 Da more than that of resveratrol, indicating that they could be deduced as glucosylated and glucuronidated metabolites of resveratrol. In addition, the fragment ions at m/z 547 [M−H−H₂O]⁻, DPIs at m/z 403 [M−H−Glc]⁻ and m/z 227 [M−H−Glc−GluA]⁻ were all observed in the ESI-MS² spectra. Based on this, N7, N8 and N19 were tentatively identified as isomeric resveratrol glucosylated and glucuronidated metabolites.

N9 produced the theoretical [M−H]⁻ ion at m/z 579.13428 (C₂₆H₂₇O₁₅, mass error −0.287 ppm) with the mass being 352 Da more than that of resveratrol and was formed by the bis-glucuronidation of resveratrol in vivo. Owing to the successive loss of GluA, the [M−H]⁻ ion generated an ion at m/z 403 in the ESI-MS² spectrum. Therefore, N9 could be deduced as a resveratrol bis-glucuronidated metabolite. Additionally, N10 generated an [M−H]⁻ ion at m/z 581.15100 (C₂₆H₂₉O₁₅, mass error 1.555 ppm). The ESI-MS² characteristic product ion [GluA-H]⁻ at m/z 175 showed the loss of 176 (GluA). Moreover, its [M−H]⁻ ions were 2 Da more than that of N9, revealing that the bis-glucuronidation reaction most likely occurred. Thus, N10 was speculated to be a dihydroresveratrol bis-glucuronidation metabolite.

In negative ion mode, N11, N14 and N18 produced [M−H]⁻ ions at m/z 469.08044, m/z 469.07977 and m/z 469.08044 with mass errors of 1.133, −0.295 and 1.133 ppm (C₂₀H₂₁O₁₁S), respectively. In the ESI-MS² spectra, the fragment ion at m/z 389 [M−H−SO₃]⁻ indicated that a sulfation reaction occurred. In addition, the neutral loss of 162 Da (m/z 389 → m/z 227) demonstrated the existence of a glucosyl group. Therefore, N11, N14 and N18 were tentatively characterized as isomeric glucosylated and sulfated metabolites of resveratrol.

Eight isomers, N15, N21, N28, N33, N35, N45, N47 and N53, were 176 Da more than that of resveratrol, which respectively afforded [M−H]⁻ ions at m/z 403.10275, m/z 403.10269, m/z 403.10364, m/z 403.10233, m/z 403.10248, m/z 403.10272, m/z 403.10287, and m/z 403.10260 (C₂₀H₁₉O₉, mass errors 0.971, 0.822, 3.179, −0.071, 0.301, 0.897, 1.269 and 0.599 ppm), respectively. In their ESI-MS² spectra, the characteristic product ions at m/z 227 [M−H−GluA]⁻ and m/z 175 [GluA-H]⁻ were detected. Therefore, N15, N21, N28, N33, N35, N45, N47 and N53 were tentatively identified as glucuronidated resveratrol or its isomers (Zhou et al., 2009; Zhang et al., 2008).

N16 possessed theoretical [M−H]⁻ ions at m/z 419.09790 (C₂₀H₁₉O₁₀, mass error 1.496 ppm). In the ESI-MS² spectra, a battery of fragment ions at m/z 243 [M−H−GluA]⁻, m/z 175 [GluA-H]⁻ and m/z 401 [M−H−H₂O]⁻ provided substantial evidence for this deduction. From the abovementioned analysis, N16 was tentatively judged as isomeric hydroxylation and glucuronidation metabolite of resveratrol.

N20, N27 and N37 displayed significant [M−H]⁻ ions at m/z 389.12402, m/z 389.12427 and m/z 389.12375 (C₂₀H₂₁O₈, mass errors 2.380, 3.022 and 1.686 ppm), respectively. They generated a DPI ion at m/z 227 in their ESI-MS² spectra. In addition, their [M−H]⁻ ions were 162 Da more than that of resveratrol, which indicated the occurrence of a glucosylation reaction. Based upon comparison of their ESI-MS/MS spectra and retention time with the corresponding reference standard, N20 was characterized as polydatin, while N27 and N37 were identified as polydatin isomers (Fu et al., 2018).

N23, N24, N26, N30, N31, N39 and N43 showed [M−H]⁻ ions at m/z 307.02789, m/z 307.02713, m/z 307.02695, m/z 307.02771, m/z 307.02800, m/z 307.02747, and m/z 307.02747 with mass errors of 2.621, 0.146, −0.440, 2.035, 2.979, 1.253 and 1.253 ppm, respectively. They were 80 Da more than that of resveratrol, which illustrated the existence of a sulfation reaction. Based upon the obtained high-resolution mass spectrometry data, they yielded DPIs at m/z 227 due to the loss of SO₃. Thus, N23, N24, N26, N30, N31, N39 and N43 were tentatively interpreted as isomeric sulfated metabolites of resveratrol.

N32, N41, N42, N51 and N52 yielded significant [M−H]⁻ ions at m/z 257.08115, m/z 257.08112, m/z 257.08136, m/z 257.08142 and m/z 257.08072 (C₁₅H₁₃O₄, mass error 1.224, 1.107, 2.041, 2.274 and −0.449 ppm), respectively. They were 30 Da more than that of resveratrol, indicating that they could be methylated and hydroxylated metabolites of resveratrol. Moreover, they afforded DPIs at m/z 242 [M−H−CH₃]⁻ and 147 [M−H−2C₂H₂O−C₂H₂]⁻. According to the above deduction, N32, N41, N42, N51 and N52 were predicted to be methylated and hydroxylated metabolites of resveratrol.

Nine isomers, N36, N38, N44, N46, N49, N50, N54, N55 and N56, gave rise to theoretical [M−H]⁻ ions at m/z 227.07103, m/z 227.07065, m/z 227.07057, m/z 227.07083, m/z 227.07089, m/z 227.07088, m/z 227.07092, m/z 227.07086, m/z 227.07088 with mass errors of 3.344, 1.670, 1.318, 2.463, 2.727, 2.683, 2.859, 2.595 and 2.683 ppm, respectively. In the ESI-MS² spectra, there were many observed characteristic product ions at m/z 185, m/z 159 and m/z 143. Among them, the ion at m/z 185 [M−H−C₂H₂O]⁻ was further fragmented to generate m/z 159 [M−H−C₂H₂O−C₂H₂]⁻ and m/z 143 [M−H−2C₂H₂O]⁻ after the loss of 26 Da and 42 Da, which were consistent with those documented in a previous report. Compared with the standard substance, N44 was positively identified as resveratrol. From the above data, N36, N38, N46, N49, N50, N54, N55 and N56 were tentatively characterized as resveratrol isomers.

N40 generated a deprotonated molecular [M−H]⁻ ion at m/z 241.05030 (C₁₄H₉O₄, mass error 3.172 ppm). It was 14 Da more than that of resveratrol, revealing that hydroxylation and dehydrogenation in vivo might have occurred. In its ESI-MS² spectra, DPIs at m/z 199 and m/z 173 were observed from the loss of C₂H₂O and C₂H₂O + C₂H₂. Therefore, N40 was tentatively identified as an isomeric hydroxylated and dehydrogenated metabolite of resveratrol. In addition, the ESI-MS² spectra of N8, N11 and N16 are illustrated in Fig. 3.

Close

Fig. 3

The ESI-MS² spectra and chemical structures of N8, N11 and N16.

Export to PPT

3.4 Comparison of the different biological treatment methods

It should be noted that three methods were used to prepare the biological samples, and then the same analytical method was applied to detect the signals. According to the accurate mass measurements, fragmentation patterns, diagnostic product ions and literature reports, a total of 56 resveratrol phase I and phase II metabolites were screened and identified in rats by using the UHPLC-LTQ-Orbitrap method. Among them, 56 metabolic products were obtained by solid phase extraction, which was also named method I. Twelve and 15 metabolites were screened by methanol precipitation (method II) and acetonitrile precipitation (method III), respectively, as shown in Fig. 4.

Close

Fig. 4

The metabolic prodycts distribution of three biological treatment methods.

Export to PPT

As far as we know, the above three methods which bear a great significance in the pretreatment of biological materials are widely applied in various laboratories at present. On the basis of the results, it could be inferred that the majority of the metabolites were obtained by method I after the same detection analysis. It showed that the SPE cartridge was the most suitable method with the functions of enrichment and purification. In the preparation process, the target substance could be enriched, and then the impurities could be removed before instrumental analysis to improve analytical sensitivity and reduce damage to the instrument (Jordan et al., 2009; Nema et al., 2010). It might be deduced that SPE cartridge could provide much more effective and reliable support for removing matrix effect. After the sample was concentrated by SPE cartridge, more response signals could be detected in mass spectrometry. However, methanol and acetonitrile as organic solvents played the role of extraction and dilution in the process of sample precipitation, which might also indirectly lead to the concentration difference and undetection phenomenon. Furthermore, products of metabolism afforded by acetonitrile precipitation were slightly more abundant than those obtained by methanol precipitation, which indicated that acetonitrile might be more adaptive for resveratrol as a precipitation solvent than methanol. These differences in metabolites quantity might be due to the different separation selectivities of organic solvents, which lead to signal peaks with diverse intensity and quantity concerning resveratrol metabolites after UHPLC-HRMS detection.

3.5 Possible biotransformation pathways of resveratrol in rats

As“metabolite clusters” mentioned by Mei (Mei et al., 2019), resveratrol was regarded as the parent drug. When the parent drug was taken into the rat body, a series of metabolic reactions could occur to form a metabolic network similar to a triangular pyramid model (as shown in Figs. 5 and 6). These reactions could be roughly summarized as follows: the parent drug acted as the metabolic centre to generate secondary metabolic centres such as polydatin, dihydroresveratrol and hydroxylated resveratrol, and a great deal of final metabolites were derived from the secondary metabolic centres step by step. Among them, resveratrol mainly underwent glucuronidation, glucosylation, sulfation, hydroxylation, dehydrogenation, hydrogenation, methylation and their composite reactions. Phase II metabolism and its complicated reactions occupied the main status of these reactions. Of course, there were differences between the metabolic products and metabolic rates. More water-soluble metabolites, such as glucuronidated resveratrol and sulfated resveratrol, are generally produced through phase II reactions in order to prepare the drug for excretion (Wilkinson, 2005; Xu et al., 2005). It is also worth noting that the metabolic centres or the final metabolism products were not isolated. There might be some transformation relationships among these products. With the development of the reactions in vivo, drug metabolites were also enriched step by step. This “metabolite clusters” theory could better guide the screening of metabolites and reveal the metabolic process. However, these metabolites were also difficult to detect due to their low content, and many isomers were easily yielded in the biotransformation process (Wang et al., 2019). Therefore, it is urgent to establish high performance testing instruments and comprehensive and accurate analysis methods to help solve such problems.

Close

Fig. 5

The possible resveratrol metabolic patterns in rats in vivo.

Export to PPT

Close

Fig. 6

The metabolite cluster model of resveratrol in rats. (Note: The color that these compounds represent are corresponding between Fig. 5 and Fig. 6).

Export to PPT