Comprehensive Screening and Identification of Phillyrin Metabolites in Rats Based on UHPLC-Q-Exactive Mass Spectrometry Combined with Multi-Channel Data Mining

Phillyrin, a well-known bisepoxylignan, has been shown to have many critical pharmacological activities. In this study, a novel strategy for the extensive acquisition and use of data was established based on UHPLC-Q-Exactive mass spectrometry to analyze and identify the in vivo metabolites of phillyrin and to elucidate the in vivo metabolic pathways of phillyrin. Among them, the generation of data sets was mainly due to multichannel data mining methods, such as high extracted ion chromatogram (HEIC), diagnostic product ion (DPI), and neutral loss filtering (NLF). A total of 60 metabolites (including the prototype compound) were identified in positive and negative ion modes based on intuitive and useful data such as the standard's cleavage rule, accurate molecular mass, and chromatographic retention time. The results showed that a series of biological reactions of phillyrin in vivo mainly included methylation, hydroxylation, hydrogenation, sulfonation, glucuronidation, demethylation, and dehydrogenation and their composite reactions. In summary, this study not only comprehensively explained the in vivo metabolism of phillyrin, but also proposed an effective strategy to quickly analyze and identify the metabolites of natural pharmaceutical ingredients in nature.


Introduction
Phillyrin, as an essential bisepoxylignan, is the main component of Forsythia suspensa ( unb.) Vahl in plants belongs to the family Oleaceae [1][2][3]. Modern pharmacological research has shown that phillyrin not only has potent biological activity, but also plays a huge role that cannot be ignored in resisting diseases and maintaining human health, such as inhibiting inflammatory response, antiviral, antioxidative stress, and anticell apoptosis [4][5][6][7][8]. Phillyrin has been reported to improve insulin resistance in the body [9,10], and it can also reduce the weight of obese mice through specific pathways [11]. Besides, phillyrin can decrease the formation and function of osteoclasts and prevent LPS-induced osteolysis in mice [12]. However, there are still many deficiencies in the comprehensive research on phillyrin metabolism in vivo.
erefore, revealing the metabolites and metabolic pathways of phillyrin is of great significance for its further development and utilization. As we all know, liquid chromatography-mass spectrometry (LC-MS) combines the high separation power of liquid chromatography with the forceful qualitative and quantitative ability of mass spectrometry, which has the advantages of high sensitivity, high selectivity, and rich structural information [13][14][15]. UHPLC-Q-Exactive MS (ultra high-performance liquid chromatography coupled with hybrid quadrupole-Orbitrap mass spectrometry) can thoroughly and extensively obtain the structural information of compounds by using the functions of positive and negative ion scanning mode, full scanning, and automatic triggering of secondary mass spectrometry scanning. e superior resolution based on Orbitrap technology can quickly achieve the measurement of high-accuracy mass. erefore, it can identify and confirm small molecule compounds in mixtures, is a powerful tool for analyzing complex compound systems, and has been extensively used in compound identification and screening [16][17][18]. By obtaining sufficient fragment information and analyzing composition information of a compound online, the molecular weight of the compound can be determined, and the possible molecular structure can be inferred based on the primary, secondary, and even multilevel mass spectral information [19][20][21]. In consequence, it is imperative to form a comprehensive and authoritative data set in the data processing process. Data mining methods reported in recent years have emerged endlessly, mainly including the following categories: mass defect filter, extracted ion chromatogram, diagnostic product ion (DPI), neutral loss filtering (NLF), and isotope pattern filtering [22][23][24][25][26][27].
In this paper, we have established a new strategy that combined UHPLC-Q-Exactive MS with multiple data mining analysis methods. Due to its slight detection limit and narrow deviation range, it is feasible to analyze and identify metabolites in rats for oral administration of phillyrin. Based on this study, we further proposed the in vivo metabolic pathways of phillyrin, which made up for the shortcomings of the current insufficient research on phillyrin metabolism and was conducive to revealing the in vivo mechanism of many pharmacological activities of phillyrin.

Materials and Methods
Based on the previous research results of our research group, we adopted the experimental methods and sample processing procedures of our predecessors [28,29].

Chemicals and Reagents.
e reference standard of phillyrin was purchased from Chengdu Must Biotechnology Co. Ltd. (Sichuan, China). After HPLC-UV analysis, the purity of phillyrin was not less than 98%. Acetonitrile, methanol, and formic acid (HPLC grade) used in the mobile phase were provided by Fisher Scientific (Fair Lawn, NJ, USA). C18-low solid-phase extraction (SPE) cartridges (3 L/ 60 g) for biological sample pretreatment were obtained from Waters (Milford, MA, USA). Ultrapure water was freshly prepared using the Milli-Q GradientÅ 10 water purification system (Millipore, Billerica, MA, USA). Moreover, other reagents and solvents met the requirements of analytical experiments in Beijing Chemical Works (Beijing, China).

Animals and Drug Administration.
Sixteen male SD rats weighing 200-220 g were purchased from Beijing Weitong Lihua Experimental Animals Company (Beijing, China). All animals were kept under specific environmental conditions (temperature 22 ± 1°C, humidity 60 ± 10%, 12-hour day and night change), with free access to food and water for one week. e rats were then randomly divided into Drug Group (for experimental urine, plasma and feces, n � 8) and Control Group (for blank urine, plasma, and feces, n � 8).
e reference standard of phillyrin was suspended in 0.5% sodium carboxymethyl cellulose (CMC-Na) solution. e Drug Group was orally administered phillyrin (300 mg/kg), while Control Group was given an equal amount of 0.5% CMC-Na solution by oral gavage. All animals were fasted for 12 hours before the experiment but had free access to water. e experimental protocol has been approved by the institutional Animal Care and Use Committee in Beijing University of Chinese Medicine. All procedures were carried out according to the Guide for the Care and Use of Laboratory Animals of the US National Institutes of Health.

Plasma Sample Collection.
After oral administration, all rats were placed in metabolic cages. Blood samples (about 0.5 ml) were taken from the infraorbital venous plexus of rats at time points of 0.5, 1, 1.5, 2, and 4 h after administration, and the operation was identical for each rat. Blank and experimental samples were obtained from the Control Group and the Drug Group, respectively. All blood samples were placed in a heparin sodium anticoagulated EP tube for 15 minutes and then centrifuged at 3000 rpm (4°C) for 15 minutes to separate plasma. After that, the plasma from the same group was combined into an aggregate and stored at − 80°C until use.

Urine Sample Collection.
Urine samples (0-24 h) were collected from each rat using a metabolic cage, and each sample was centrifuged at 3000 rpm (4°C) for 15 min to obtain a supernatant. e urine supernatants of each group of rats were mixed and stored at − 80°C until use.

Feces Sample Collection.
Fecal samples (0-24 h) were obtained from each rat using a metabolic cage, then dried, and ground to a powder. e fecal powders of each group of rats were mixed. Firstly, 0.5 g of fecal powder was dissolved in 70% methanol, and then it was extracted for 30 min by ultrasonic. Finally, fecal samples of rats in the Control Group and the Drug Group were obtained.

Biological Sample Preparation.
All biological samples were prepared by precipitating and concentrating proteins and solid residues. Firstly, the SPE cartridges were pretreated by activation with methanol (5 ml) and deionized water (5 ml). Afterward, plasma, urine, and feces samples (1 ml) were added to the SPE cartridges, respectively. en, the SPE cartridges were eluted with deionized water (5 ml) and methanol (3 ml) in that order, and the methanol eluates were collected. Finally, the eluates were dried with nitrogen at room temperature, and the residue was redissolved in 5% acetonitrile solution (100 μl) and centrifuged at 14000 rpm (4°C) for 15 min. e supernatant obtained by the above method was used for instrumental analysis.

Data Processing.
A ermo Xcalibur 2.1 workstation ( ermo Scientific, Bremen, Germany) was utilized to acquire and process HR-ESI-MS 1 and MS n data. Based on the established screening templates of metabolites, by searching the metabolites with specific molecular weight to compare the high-resolution extracted ion chromatograms of the Control Group and the Drug Group, the metabolites related to phillyrin were selected. To obtain as many phillyrin metabolite ions as possible, the peaks detected with intensity over 10,000 for negative ion mode and 50,000 for positive ion mode were selected for further structural characterization. Based on the exact mass of the metabolite and the set elemental composition, the formula predictor could accurately calculate the chemical formula of all parent ions. e type and number of atoms were set as follows: C

e Construction of Analysis Strategy.
is study constructed an effective new strategy based on UHPLC-Q-Exactive MS for data acquisition combined with multipath data mining to analyze the metabolism of phillyrin in vivo. First of all, the ESI-MS n data set of the samples in both positive and negative ion modes were obtained by using the data-dependent scanning (DDS) acquisition method. Secondly, the common metabolites of phillyrin were determined according to literature reports and HREIC search, thereby establishing the screening templates for phillyrin metabolites.
en, by analyzing the mass fragmentation behaviors of the reference standards, the reasonable DPIs and NLFs of phillyrin were summarized. Afterward, based on the chromatographic retention time, the established DPIs and NLFs, and the corresponding calculated ClogP values, the structure of a series of metabolites of phillyrin could be identified. Finally, the metabolic pathway of phillyrin can be inferred from the above metabolic data. e entire research process of this strategy is shown in Figure 1.

Establishment of the Screening Templates for Metabolites.
According to literature reports and HREICs search, three compounds (phillyrin, phillygenin, and enterolactone) were selected as primary metabolites, which are often found in the metabolites of phillyrin. After data sorting, in negative ion mode, the metabolites screening templates were set as follows: (1) phillyrin template (m/z 533.2017); (2) phillygenin template (m/z 371.1489) and its derivative templates (m/z 329.1020 for detrimethyl, m/z 357.1333 for demethylation, m/z 373.1646 for hydrogenation, and m/z 387.1438 for oxygenation); (3) enterolactone template (m/z 297.1121). Based on this effective method, some nondiscoverable metabolites could also be screened out from complex background noise.

Analysis of Mass Fragmentation Behavior of Phillyrin.
To further explore the ESI-MS n fracture behavior of phillyrin, a comprehensive analysis of the standard solution was performed using UHPLC-Q-Exactive MS. For example, phillyrin in negative ion mode could find [M + HCOOH-H] − ion at m/z 579.2072 in the ESI-MS 1 spectrum. Due to the absence of the glucose fragment, its characteristic ion peak was found at m/z 371 of the ESI-MS 2 spectrum. On this basis, several series of characteristic product ions were retrieved at m/z 357, m/z 341, m/z 327, and m/z 311 because a series of fragments such as CH 2 , 2CH 2 , CH 2 + CH 2 O, and 2CH 2 O were successively lost. Compounds with the same parent nucleus will have similar cleavage fragments in the ESI-MS n spectrum, so a comprehensive identification of metabolites can be achieved based on regular DPIs and NLFs. For example, the DPI at m/ z 371 was diagnosed due to the absence of a glucose moiety.
us the presence of DPIs at m/z 371 or m/z 371 + X in the ESI-MS 2 spectrum of the compound provided comparable information for the identification of metabolites. At the same time, the continuous appearance of 14 Da (CH 2 ) and 30 Da (CH 2 O) NLFs in the ESI-MS n spectrum of phillyrin also provided significant help for the identification of metabolites. e cleavage pathway of phillyrin in negative ion mode is presented in Figure 2. Moreover, the mass fragmentation behaviors of phillyrin in positive ion mode are shown in Figure 3.

Identification of Phillyrin Metabolites in Rats.
e total ion chromatograms (TICs) of urine, plasma, and feces samples after oral administration of phillyrin in rats were obtained using UHPLC-Q-Exactive mass spectrometry. A total of 60 metabolites were found in both positive and negative ion modes by processing the data collected from the UHPLC-Q-Exactive instrument. Among them, there were 31 metabolites in positive ion mode and 33 metabolites in negative ion mode. In addition, after the literature search International Journal of Analytical Chemistry and comparison, 21 metabolites have been found and detected by predecessors [30,31], while the remaining 39 metabolites have been screened and identified through the established strategy. All relevant mass spectral data are summarized in Table 1, and high-resolution extracted ion chromatograms (HREICs) of all metabolites of phillyrin are shown in Figure 4.

Identification of Metabolites Based on Phillyrin.
e metabolite M0 producing [M + HCOOH-H]ion at m/z 579.2072 (C 28 H 35 O 13 , 2.32 ppm) was eluted at 6.60 min. In contrast to the elution time and fragmentation behavior of the phillyrin standard, M0 could be accurately inferred as the phillyrin [32,33]. e retention times of the metabolites M17 and M22 were 5.76 and 5.96 min, respectively, and they showed the same molecular ion at m/z 519.1498 (C 25 H 27 O 12 , error ≤ ±2.50 ppm). In the ESI-MS 2 spectrum, the DPI at m/z 343 was generated due to the loss of 2CH 2 by the phillygenin. While m/z 519 was 176 Da larger than the former, it was inferred that m/z 343 was caused by the absence of the glucuronic acid moiety. erefore, it was speculated that M17 (ClogP, − 0.41) and M22 (ClogP, − 0.39) were demethylated and carboxylated metabolites of phillyrin, but the position of glucuronic acid group was different. e metabolite M32 was eluted at 6  spectrum, the DPI at m/z 193 demonstrated that one of the phenyl rings of the phillygenin was introduced into 4OH, and the mother nucleus lost 2CH 2 O. According to the DPI at m/z 124, 137, and 150, the generation of two double bonds could also be proved.
Metabolite    Figure 3: e fragmentation behavior of phillyrin in positive ion mode.  (14), 378 (12)99(9), 69 (7), 359 (6), 172 (2) (17), 153 (14), 308 (14), International Journal of Analytical Chemistry International Journal of Analytical Chemistry     /z 191, 177, 151, 137, and 123, indicated that the two benzene rings underwent substitution reactions with diverse groups (2OCH 3 , OH + OCH 3 ), respectively. At the same time, the NLFs were 30 Da (CH 2 O, from m/z 357 to m/ z 327) and 54 Da (C 4 H 6 , from m/z 177 to m/z 123), it could be inferred that the two benzene rings were mutually connected through a five-membered ring and a double bond. Besides, it was known from the molecular formula that it was difficult to ascertain the substitution site of one OH. In the ESI-MS 2 spectrum of M6, the DPI at m/z 393 was 93 Da higher than m/z 300, presumably due to the neutral loss of hydroxybenzene. At the same time, the DPI at m/z 300 was 60 Da higher than m/z 240 and 28 Da higher than m/z 272, and it was presumed that NLFs were 2CH 2 O (60 Da) and 2CH 2 (28 Da), respectively. Based on the information obtained, we could conclude that the structural formula of M6 contained hydroxybenzene and dimethoxyphenyl groups, and two of the five-membered rings were hydrogenated to open the ring. e DPIs at m/z 216, 246, and 361 further confirmed the above judgment, and the substitution sites of three other hydroxyl groups could not be determined.
M24, M28, and M41 showed the same [M− H] − ion at m/z 437.0901 (C 20 H 21 O 9 S, error ≤ ±3.00 ppm), and they were eluted at 6.14 min, 6.32 min, and 6.88 min, respectively. eir molecular weight was 14 Da lower than that of M42, so they were inferred to be the demethylation products of M42. In the ESI-MS 2 spectrum, the three metabolites had the same DPIs, such as m/z 357, 151, 137, and 80, which proved that their benzene rings had identical substituents (2OCH 3 , 2OH). Because the substituents on the benzene rings and the sulfo (-HSO 3 ) substitution site were different, M24, M28, and M41 were positional isomers [34] e DPIs in the ESI-MS 2 spectrum, such as m/z 179, 137, and 95, clarified the structure of the two benzene rings (OH + OCH 3 , OH). Based on the established molecular formula, it could be inferred that the fabric between two benzene rings was a linear alkane containing four carbon atoms.

Proposed Metabolic Pathways of Phillyrin.
In our study, a total of 60 metabolites (including the prototype compound) were detected and identified in rats that were orally administered phillyrin. e proposed metabolic pathways of forsythiaside are shown in Figure 5.
e main biological reactions of phillyrin in rats included the following types, such as hydrogenation,  glucuronidation, sulfonation, carbonylation, ammoniation, dehydrogenation, demethylation, and ring cleavage and their composite reactions. Also, some extraordinary products had been discovered, such as carbon-nitrogen unsaturated bonds formed in M1 and M44. Moreover, the metabolites M49, M50, and M59, which were generated by hydrogenation or cleavage of benzene, had never been reported before.

Conclusions
After oral administration of phillyrin to rats, metabolites in plasma, urine, and feces of rats were studied comprehensively. Its innovation lay in the use of a new comprehensive strategy to screen and identify 60 in vivo metabolites of phillyrin. Firstly, the combination of UHPLC and Q-Exactive MS overcame many of the shortcomings of traditional triple quadrupole mass spectrometry, and the extremely high resolution could significantly eliminate the interference of the sample matrix. Secondly, the formation of the data set mainly depended on a variety of data mining methods, such as high extracted ion chromatogram (HEIC), diagnostic product ion (DPI), neutral loss filtering (NLF), and isotope pattern filtering (IPF). Finally, accurate qualitative identification of compounds could be achieved based on the precise relative molecular mass of each chromatographic peak in the mass spectrum, the ion fragment information of the secondary mass spectrum, the fragmentation rule of the mass spectrum, and the chromatographic retention time. e results showed that the primary biological reactions of phillyrin in vivo included methylation, hydrogenation, sulfonation, glucuronidation, demethylation, dehydrogenation, ring cleavage, and their composite reactions. Among them, the biological activity of some specific metabolites was unknown. In summary, this study provided vital information to the research field of phillyrin metabolism and had important value significance for studying the mechanism of action of phillyrin and monitoring of drug content.

Data Availability
e data used to support the finding of this study are available from the corresponding author upon request.

Conflicts of Interest
e authors have declared no conflicts of interest.