Comprehensive Metabolic Profiling of Euphorbiasteroid in Rats by Integrating UPLC-Q/TOF-MS and NMR as Well as Microbial Biotransformation

Euphorbiasteroid, a lathyrane-type diterpene from Euphorbiae semen (the seeds of Euphorbia lathyris L.), has been shown to have a variety of pharmacological effects such as anti-tumor and anti-obesity. This study aims to investigate the metabolic profiles of euphorbiasteroid in rats and rat liver microsomes (RLMs) and Cunninghamella elegans bio-110930 by integrating ultra-performance liquid chromatography-quadrupole time-of-flight-mass spectrometry (UPLC-Q/TOF-MS), UNIFI software, and NMR techniques. A total of 31 metabolites were identified in rats. Twelve metabolites (M1–M5, M8, M12–M13, M16, M24–M25, and M29) were matched to the metabolites obtained by RLMs incubation and the microbial transformation of C. elegans bio-110930 and their structures were exactly determined through analysis of NMR spectroscopic data. In addition, the metabolic pathways of euphorbiasteroid were then clarified, mainly including hydroxylation, hydrolysis, oxygenation, sulfonation, and glycosylation. Finally, three metabolites, M3 (20-hydroxyl euphorbiasteroid), M24 (epoxylathyrol) and M25 (15-deacetyl euphorbiasteroid), showed significant cytotoxicity against four human cell lines with IC50 values from 3.60 μM to 40.74 μM. This is the first systematic investigation into the in vivo metabolic pathways of euphorbiasteroid and the cytotoxicity of its metabolites, which will be beneficial for better predicting the metabolism profile of euphorbiasteroid in humans and understanding its possible toxic material basis.


Introduction
Euphorbiae semen, known as "Qian-Jin-Zi" in China, is the dried and ripe seeds of Euphorbia lathyris L. According to the 2020 edition of the Chinese Pharmacopoeia, Euphorbiae semen could be clinically used either alone or in combination with other herbal medicines as remedies for reducing water and phlegm retention, promoting blood circulation, removing blood stasis, curing tinea and scabies, and treating amenorrhea, snakebites, terminal schistosomiasis, anuria, and constipation. It is worth noting that Euphorbiae semen was listed as one of the 28 toxic Chinese herbal medicines in the Chinese Pharmacopoeia [1]. It has been proven to exhibit a strong stimulating effect on the gastrointestinal tract, and its intensity is three times that of croton oil. The main toxic components of Euphorbiae semen are diterpenoids, identified by means of fractionation. Euphorbiasteroid (M0, Figure 1), a natural lathyrane-type diterpenoid, was deemed as one of the intestinal tract stimulant constituents to induce diarrhea [2]. Moreover, euphorbiasteroid has the potential to reverse resistance to anticancer drugs in MES-SA/Dx5 cells [3], and also exhibits anti-obesity [4], The in-depth study of drug metabolism can clarify the metabolic pathways as wel as the toxic and active mechanisms of drugs so as to ensure their safety and lay a theo retical foundation for clinical application and toxic and side effect detection [7]. There fore, it is of great significance and value to establish a sensitive and reliable analytica method for the identification of drug metabolites. Liquid chromatography (LC) is one o the main methods used to analyze active pharmaceutical ingredients (APIs) in American and European Pharmacopoeia, and has been combined with a variety of detectors, such as fluorescence detectors, electrochemical detectors or flame photometric detectors However, these detectors cannot clearly identify compounds separated by LC. In con trast, high performance liquid chromatography-mass spectrometry (HPLC-MS), with liquid chromatography as the separation system and mass spectrometry as the detection system, combines the high separation capability of HPLC for complex samples with the high selectivity and sensitivity of mass spectrometry and its ability to provide relative molecular mass and structural information, perfectly compensating for this deficiency [8,9].
Especially, ultra-performance liquid chromatography-quadrupole time-of-flight-mass spectrometry (UPLC-Q/TOF-MS) combined with the computer-aided identification platform Waters UNIFI, has become a powerful analytical tool, which ha the advantages of accurate mass measurement, efficient separation technology, and rapid identification of metabolites, and has been widely applied in the metabolism research o traditional Chinese medicine in recent years [10]. However, the main disadvantage o mass spectrometry analysis is that the molecular weight of structural characterization depends on the collision-induced dissociation of protonated molecular ions of targe compounds, so it cannot provide accurate metabolite structure in most cases [11]. Nu clear magnetic resonance (NMR) is a complementary analytical method, which can characterize metabolites in more detail. The advantages of the combination o UPLC-Q/TOF-MS and NMR have been verified in several reports in recent years [12][13][14] However, the sensitivity of NMR is not as good as that of mass spectrometry, and the presence of impurities in the sample has great influence on NMR signal. Therefore, it i The in-depth study of drug metabolism can clarify the metabolic pathways as well as the toxic and active mechanisms of drugs so as to ensure their safety and lay a theoretical foundation for clinical application and toxic and side effect detection [7]. Therefore, it is of great significance and value to establish a sensitive and reliable analytical method for the identification of drug metabolites. Liquid chromatography (LC) is one of the main methods used to analyze active pharmaceutical ingredients (APIs) in American and European Pharmacopoeia, and has been combined with a variety of detectors, such as fluorescence detectors, electrochemical detectors or flame photometric detectors. However, these detectors cannot clearly identify compounds separated by LC. In contrast, high performance liquid chromatography-mass spectrometry (HPLC-MS), with liquid chromatography as the separation system and mass spectrometry as the detection system, combines the high separation capability of HPLC for complex samples with the high selectivity and sensitivity of mass spectrometry and its ability to provide relative molecular mass and structural information, perfectly compensating for this deficiency [8,9]. Especially, ultra-performance liquid chromatography-quadrupole time-of-flight-mass spectrometry (UPLC-Q/TOF-MS) combined with the computer-aided identification platform Waters UNIFI, has become a powerful analytical tool, which has the advantages of accurate mass measurement, efficient separation technology, and rapid identification of metabolites, and has been widely applied in the metabolism research of traditional Chinese medicine in recent years [10]. However, the main disadvantage of mass spectrometry analysis is that the molecular weight of structural characterization depends on the collision-induced dissociation of protonated molecular ions of target compounds, so it cannot provide accurate metabolite structure in most cases [11]. Nuclear magnetic resonance (NMR) is a complementary analytical method, which can characterize metabolites in more detail. The advantages of the combination of UPLC-Q/TOF-MS and NMR have been verified in several reports in recent years [12][13][14]. However, the sensitivity of NMR is not as good as that of mass spectrometry, and the presence of impurities in the sample has great influence on NMR signal. Therefore, it is necessary to separate the analyte from the impurity well and obtain enough quantity for structural identification by NMR. Generally, supplementary methods can be used to obtain a sufficient sample amount of metabolites, including chemical methods [11] and small experimental animal models [15], microsomal preparations [16,17], enzyme-catalyzed reactions [18,19], microbial transformation [11,20] and so on. Compared with other methods, microbial transformation is more convenient and economical, especially with the advantages of in vitro large-scale preparation [20][21][22][23][24]. In particular, the filamentous zygote fungus C. elegans has been shown to possess a human-like cytochrome P450 monooxygenase system, including the CYP509A1 isoenzyme that is similar to the CYP51 family, thus producing similar metabolic profiles to mammalian animals [25,26].Sufficient samples for structural characterization based on NMR techniques as well as the evaluation of bioactivity and toxicity can thus be acquired. Therefore, a combination of in vivo animal experiments and in vitro microbial transformation will contribute to accurately elucidating the structures of metabolites.
The current work aimed to identify the metabolites of euphorbiasteroid in rats and in vitro models (RLMs and C. elegans bio-110930), by using UPLC-Q/TOF-MS and the UNIFI platform as well as NMR technique, and to rationalize the elimination mechanism of euphorbiasteroid, in which the biotransformation based on the fungus C. elegans bio-110930 and chemical hydrolysis were applied to prepare the samples of potential metabolites. This method has been used in the previous research of our research group and is reasonable [27]. Finally, a total of 31 metabolites in vivo, including four phase II metabolites, were detected and identified. Additionally, the structures of 12 metabolites were accurately characterized and confirmed by structural elucidation based on NMR technique and by comparing the chromatography retention times and mass spectra with those of standard compounds from the biotransformation products of C. elegans bio-110930. Then, the metabolic pathway of euphorbiasteroid in rats was rationally elucidated on the basis of the study of metabolism in vivo and in vitro. The cytotoxic assay showed that three metabolites of euphorbiasteroid (M3, M24-M25) have cytotoxicity on four strains of human cells (SH-SY5Y, LO2, AC-16, and HK-2) with the IC 50 values from 3.60 µM to 40.74 µM, while euphorbiasteroid did not show cytotoxicity on the same cell lines (IC 50 > 50 µM). Therefore, our research findings will provide new insights into the metabolism mechanism of euphorbiasteroid and the possible toxicity of the metabolites, be beneficial for understanding the in vivo elimination process of euphorbiasteroid, and provide data support and reference for safe, reasonable and controllable clinical application of Euphorbiae semen.
The mass spectrometry detection was performed on the SYNAPT G2-Si HDMS system, equipped with an electrospray ionization (ESI) source (Waters Corp., Manchester, UK). A positive ion mode was conducted in this analysis. Mass spectrometry conditions were finally set as follows: capillary voltage of 3.0 kV, cone voltage of 40 V, source temperature of 120 • C, and de-solvation temperature of 400 • C. Nitrogen was used as the desolvation and cone gas with a flow rate of 800 and 50 L/h, respectively, and the full-scan mass range was set as m/z 50-1500 Da. In the auto mass spectrometry mode, the collision-induced dissociation energies were set at 0 eV for the precursor ion at the low-energy mode, and the collision-induced dissociation energies were set from 2 to 10 eV for the high-energy mode. Real-time data were calibrated using an external reference (LockSpray™) at a concentration of 0.2 ng/mL with an infusion flow rate of 5 µL/min, generating a reference ion for the positive ion mode (m/z 556.2771) during the UPLC-MS analysis. Data were acquired and processed using MassLynx ™ NT 4.1 software (Waters, Milford, MA, USA).
Accurate molecular weights of some metabolites were acquired using an Agilent 6520 Accurate Mass quadrupole time-of-flight mass spectrometer (Q-TOF MS; Agilent Technologies, Santa Clara, CA, USA). The capillary voltage of the ion source was set at 3.0 kV in positive ion mode. Nitrogen was used as the de-solvation and nebulizing gas at a constant temperature of 350 • C. The scan range was set at m/z 100-1500 Da.
The isolation and purification of metabolites were achieved using an Agilent 1200 series Semi-preparative High Performance Liquid Chromatography (HPLC) system (Palo Alto, CA, USA) consisting of a G1311A quat pump solvent delivery system, a G1379A degasser unit, a G1313A autosampler, and a G1315B DAD detector. The preparation was performed with a Zorbax SB-C18 (5 µm, 9.4 mm × 25 cm) column (Agilent Technologies, Santa Clara, CA, USA). The wavelength was set at 280 nm.
Nuclear magnetic resonance (NMR) spectra of euphorbiasteroid and metabolites were measured on Bruker AV-500 spectrometers (Faellanden, Switzerland) using tetramethylsilane as an internal standard.

Animal and Drug Administration
Male Sprague-Dawley rats (200-220 g) were commercially supplied by Shanghai Sippr-BK Laboratory Animal Co., Ltd. (Shanghai, China) and were housed in a humidityand temperature-controlled room (50 ± 10% and 22-24 • C) with a 12-h light/dark cycle. The experimental rats were allowed to access food and water ad libitum and acclimatized to the conditions mentioned above for a week, then fasted overnight but with free access to water before the experiments. Euphorbiasteroid was dissolved in soybean oil solution (containing 0.5% ethanol) to form a concentration of 10 mg/mL. A single dose of 100 mg/kg euphorbiasteroid was orally administered to rats and the same concentration of soybean oil solution (containing 0.5% ethanol) was administered as a blank control. All animal procedures were performed in accordance with the Guidelines for Care and Use of Laboratory Animals of Naval Medical University and approved by the Animal Ethics Committee of Naval Medical University.

Sample Collection of Blood, Urine, and Feces
Blood samples (0.5 mL) were collected from six rats through the orbital sinus before administration (blank sample) and 0.25, 0.5, 1, 2, 4, 6, and 12 h after administration. Plasma samples were prepared by centrifugation at 4000 rpm for 10 min. For urine and feces sampling, 12 rats were divided into an administration group and a blank group, and were placed separately in stainless steel metabolic cages. Urine and feces samples were collected in containers surrounded by ice over 0-6, 6-12, and 12-24 h after drug administration. The mixed urine samples were centrifuged at 4000 rpm for 10 min at 4 • C to obtain the supernatants, and fecal samples were left in a cool and dry place until dry. All the biological samples were frozen at −80 • C before analysis.

Preparation of Blood, Urine, and Feces Samples
An aliquot of 200 µL of plasma and urine samples was put in a 1.5 mL tube, respectively. 800 µL of acetonitrile was added and vortexed for 5 min to extract metabolites. Feces samples (1.0 g) were crushed and then ultrasonically extracted with acetonitrile (10 mL) for 30 min. All the above-mentioned mixtures were centrifuged at 13,000 rpm at 4 • C for 10 min. The supernatants were then transferred and evaporated to dryness under a nitrogen stream at 30 • C. The residues were dissolved in 100 µL of methanol and then centrifuged at 13,000 rpm at 4 • C for 10 min. All supernatants were injected into the UPLC-Q/TOF-MS system for analysis.

In Vitro Incubation of Euphorbiasteroid with Rat Liver Microsomes
The microsomal incubation approach was based on previous metabolism studies published by Wintermeyer et al. [28] and Franziska et al. [17]. A 200 µL incubation system containing 10 µL of solution A, 2 µL of solution B, 5 µL of rat liver microsomes (20 mg/mL) and 182 µL 0.1 M PBS buffer (pH = 7.4) was constructed. The above solution was heated in a 37 • C water bath, then 1 µL of euphorbiasteroid (dissolved in DMSO solution, 10 mM) was used to start the reaction, and the mixture was then incubated at 37 • C for 1 h. The reactions were terminated by the addition of 200 µL of ice-cold acetonitrile. The mixture was then centrifuged at 13,000 rpm for 10 min, and a 2-µL aliquot of the supernatant was directly injected into the UPLC-Q-TOF-MS system.

Microbial Transformation of Euphorbiasteroid
The biotransformation process was conducted at two scales: preliminary screening and preparative. Preliminary screening scale biotransformation of euphorbiasteroid was carried out in 250 mL Erlenmeyer flasks containing 100 mL of liquid medium. The flasks were placed on a rotary shaker (160 rpm, 28 • C). A standard two-stage fermentation protocol was employed in all experiments [29,30]. After 2 days of pre-culture, the substrates of 5 mg (dissolved in 0.5 mL of acetone) were added into each flask. Taking 1 mL samples on days 0, 2, 4, 7, 10, and 14, samples were centrifuged and the degree of transformation was compared to controls on TLC and HPLC, and a 2-µL aliquot of the supernatant was directly injected into the UPLC-Q-TOF-MS system. Culture controls consisted of sterile medium, in which microorganisms were grown under identical conditions without substrate. Substrate controls were composed of sterile medium and the same amount of substrate incubated under the same conditions without microorganisms.

Preparation of the Transformation Products of Euphorbiasteroid
The preparative scale biotransformation of euphorbiasteroid was carried out in 50 1000 mL Erlenmeyer flasks, each containing 400 mL of sterilized potato medium. The flasks were placed on a rotary shaker operating at 160 rpm at 28 • C. After 48 h of pre-culture, 20 mg of substrates in 2 mL of acetone were added to each flask. After 12 days of incubation, the culture was pooled and filtered. The filtrate was extracted three times with an equal volume of EtOAc and concentrated under reduced pressure to dryness. Compounds 10 and 11 were obtained by the hydrolysis of euphorbiasteroid. 210 mg of euphorbiasteroid was dissolved in 45 mL of MeOH, and 3 mL of 1.54 mol/L KHCO 3 aqueous solution was added dropwise to this solvent with stirring. The mixture was stirred and hydrolyzed at 30 • C. After 5 days, methanol was removed by evaporation under vacuum, and the remaining solution was then extracted three times with 200 mL of solution (ethyl acetate: water = 1:1). The combined organic layer was evaporated under a vacuum. The ethyl acetate extract (184.5 mg) was re-dissolved in 10 mL of methanol, followed by Sephadex LH-20 CC (3 × 150 cm) eluting with MeOH/H 2 O (50%) to afford four fractions (Fr. 1-Fr. 5). Fr. 2 was further purified using semi-preparative HPLC (50% MeOH in water, v/v, 2.0 mL/min) to afford compounds 10 (44.6 mg, t R = 5.5 min) and 11 (45.7 mg, t R = 19.5 min).

Cell Culture and Cell Cytotoxicity Assay
Cell cytotoxicity was determined by the CCK-8 assay. Four strains of human cells (SH-SY5Y) were seeded in 96-well plates at a density of 3 × 10 3 cells/well under 37 • C and 5% CO 2 for 12 h and subsequently treated with the test sample solution (euphorbiasteroid and its metabolites, 10 µL) for 72 h. After treatment, each well with 10 µL CCK-8 reagent was incubated for 1-2 h in the incubator. Afterwards, the optical OD-value was measured at 450 nm through a microplate reader. Three multiple wells were set as parallel experimental groups.

Mass Fragmentation Behavior Analyses of Euphorbiasteroid
In order to obtain the overall fragmentation profile of euphorbiasteroid, the standard solution of euphorbiasteroid was analyzed by UPLC-Q/TOF-MS, which is helpful to better understand the MS/MS spectrum of its metabolites. The parent drug euphorbiasteroid had a protonated molecular ion [M + H] + at m/z 553.2809 with a retention time of 12.67 min. In the MS/MS spectrum, it had the characteristic and most abundant fragment ion at m/z 297.1850, derived from the loss of two CH 3 COOH and one C 6 H 5 CH 2 COOH neutral molecules, which was further fragmented to form ion peaks at m/z 279.1746, m/z COOH] + ). In addition, the fragment ion at m/z 315.1953 resulted from the ions at m/z 433.2371 and m/z 357.2062 by loss of C 6 H 4 CH 2 CO and CH 2 CO, respectively, which further lost a water to form ion at m/z 297.1850. Therefore, CH 3 COOH (m/z 493.2587), C 6 H 5 CH 2 COOH (m/z 417.2276), CH 3 COOH + C 6 H 5 CH 2 COOH (m/z 433.2371) and 2 CH 3 COOH + C 6 H 5 CH 2 COOH (m/z 279.1746) were the characteristic product ions of euphorbiasteroid. Mass spectra and the fragmentation scheme for euphorbiasteroid were shown in Figure 2. The 1 H NMR and 13 C NMR spectral data of euphorbiasteroid are listed in Tables 1-3, with the carbon position labeled as shown in Figure 1.
ard solution of euphorbiasteroid was analyzed by UPLC-Q/TOF-MS, which is helpfu better understand the MS/MS spectrum of its metabolites. The parent drug euphorb teroid had a protonated molecular ion [M + H] + at m/z 553.2809 with a retention tim 12.67 min. In the MS/MS spectrum, it had the characteristic and most abundant fragm ion at m/z 297.1850, derived from the loss of two CH3COOH and one C6H5CH2CO neutral molecules, which was further fragmented to form ion peaks at m/z 279.

Identification of Metabolites of Euphorbiasteroid In Vitro and In Vivo
First, the metabolites of euphorbiasteroid in rats (plasma, urine, and feces), RLMs, and C. elegans culture medium were predicted by setting the prototype components and potential biological metabolic reactions in UNIFI 4.1 software. Then, the predicted metabolites in each sample were further compared according to the characteristic mass spectrum behaviors (including parent ions, internal cleavage in the ion source, and characteristic fragment ions of each metabolite) and retention times. A total of 31 metabolites identified in vitro and in vivo are listed in Table 4. The retention times, precursor molecular ions, and key fragments of euphorbiasteroid and its metabolites are listed in Table 4. The extracted ion chromatograms and product ion spectra of metabolites are shown in Figures 3 and 4.       molecular ions, and key fragments of euphorbiasteroid and its metabolites are listed in Table 4. The extracted ion chromatograms and product ion spectra of metabolites are shown in Figures 3 and 4.    The compounds might be produced by the oxidation of M14 or M15, which was consistent with our initial speculation. Thus, it is provisionally interpreted that M16 and M17 were produced by oxidizing the aldehyde of M14 or M15 into carboxyl groups.
Metabolites (M18-M20) were detected between 5.96 and 6.75 min. Taking M19 as an example (Figure 4 1898, and m/z 251.1802 were produced, which were consistent with fragmentation behaviors of euphorbiasteroid. Therefore, the metabolite M24 was probably formed via the loss of two CH 2 CO and a C 6 H 5 CHCO from the prototype compound euphorbiasteroid. Similar findings have been reported before [32]. The metabolite M25 provided its protonated ion at m/z 511.2694 with a retention time of 12.49 min (Figure 4). It was 42 Da less than that of euphorbiasteroid, meaning that it might be a hydrolysis product of euphorbiasteroid. In the MS/MS spectrum, a battery of fragment ions at m/z 451.2482 [M + H-CH 3  The metabolites M26 and M27, which showed their respective positive ions at m/z 527.2661 and 527.2647 with the retention times of 7.66 and 9.14 min, were plausibly assigned as monohydroxylated products of metabolite M25 due to their protonated ions being 16 Da more than that of M25 (Figure 4) O] + proved the existence of hydroxylation. According to the reliable analysis above, they were provisionally characterized as monohydroxylated products of M25.
The metabolites M28 and M29 gave rise to protonated ions at m/z 731.3280 and 731.3286, which were eluted at 5.23 and 6.97 min, respectively. They were 162 Da more than metabolites M1-M6. Taking M28 as an example (Figure 4), the fragment ion at m/z 569.2768 provided reliable evidence for identifying the metabolites as glycosylation products of euphorbiasteroid due to the loss of 162 mass units. The fragment ions at m/z 509. 2549, 449.2335, 357.2074, 315.1971, 297.1868, 279.1754, and 269.1917 demonstrated that a hydroxyl group in the metabolites M7-M11 was replaced by C 6 H 10 O 5 . Therefore, they were tentatively identified as glycosylation products of euphorbiasteroid.
The metabolites M30 and M31 were observed with their protonated ions at m/z 665.2271 and 665.2267 with retention times of 2.87 and 3.19 min, which were deductively assigned as sulfonated products of the metabolites M7-M11 because their protonated ions were 80 Da larger than the metabolites M7-M11 (Figure 4) Finally, in order to explore whether RLMs and the microbial model could simulate the metabolism of euphorbiasteroid in rats, the incubation of RLMs and C. elegans bio-110930 with euphorbiasteroid were studied, respectively. The results showed that 18 and 14 metabolites identified in the above two models could match those of rats in vivo, respectively, indicating that they could simulate the metabolism of euphorbiasteroid in vivo to a certain extent. Considering the economic benefits and transformation efficiency, the large-scale microbial transformation experiment was used for the subsequent preparation of metabolite standards and the accurate characterization of the metabolite structures [33][34][35].
Compound 1 was isolated as a white powder with an optical rotation of [α] 20 D +116.10 • (c 0.118, MeOH). The molecular formula of C 32 H 40 O 9 was deduced from the [M + H] + ion at m/z 569.2746, and the molecular weight of compound 1 is 16 Da more than that of the precursor compound euphorbiasteroid. Analysis of the 1 H and 13 C NMR spectroscopic data of 1 revealed its structural similarity to euphorbiasteroid (M0 , Tables 1 and 3). The only difference between these two compounds was the occurrence of an extra hydroxymethyl [δ C 71.2, δ H 3.52 (d, J = 11.25 Hz, H-18a)] and δ H 3.44 (d, J = 11.25 Hz, H-18b)) in compound 1, taking the place of a methyl of euphorbiasteroid. In the HMBC spectrum of compound 1, the two protons of hydroxymethyl at δ H 3.52 and 3.44 exhibited long-range HMBC correlations with C-9 (δ C 30.2) and C-11 (δ C 25.2) (Figure 6), suggesting that a hydroxyl might be substituted at C-18 or C-19 position. In the NOESY spectrum (Figure 7), the key NOE correlations of two hydroxymethyl protons (δ H 3.52 and 3.44) with H-9 (δ H 1.21) and H-11 (δ H 1.65) proved that the C-18 of compound 1 was substituted by a hydroxyl, and assigned the relative configuration of 18-hydroxymethyl to be α-oriented. Unambiguous complete assignments for the 1 H and 13 C NMR signals were made by combination of DEPT, 1 H-1 H COSY, HSQC, HMBC, and NOESY spectra (Figures S1-S10). On the basis of the above evidence, the structure of compound 1 was thus established as 18α-hydroxyl euphorbiasteroid.
Compound 2 was isolated as a white powder with [α] 20 D +74.89 • (c 0.150, MeOH). The molecular formula was assigned as C 32 H 40 O 9 based on its HR-ESI-MS data that displayed an [M + H] + ion at m/z 569.2747. Similar to compound 1, the molecular weight of 2 was 16 mass units more than that of euphorbiasteroid and shared the same molecular formula.
The 1 H and 13 C NMR data of compound 2 were quite close to those of euphorbiasteroid (Tables 1 and 3), except for the absence of an aromatic proton. In the 1 H NMR spectrum of 1, two groups of aromatic protons (each two protons) supported the presence of a p-substituted benzyl ring. Two groups of carbon resonances (each two carbons) at δ C 130.6 and 115.4 as well as the 13 C chemical shift of C-4 proved the above deduction, and attributed a hydroxyl substitution at C-4 ( Figures S11-S20). On the basis of the above evidence, the structure of compound 2 was established as 4 -hydroxyl euphorbiasteroid.

The Structure Elucidation of Transformation Products
Through large-scale microbial transformation and chemical hydrolysis experimen twelve transformation products (1)(2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12) were prepared, including eleven new compoun (1-9, 11 and 12) and one known compound (10). Compound 10 was characterized epoxylathyrol by comparison of the NMR and HR-ESI-MS data with the literature [3 The structures of eleven new transformation products were characterized by ESI-M HR-ESI-MS, 1D-, and 2D-NMR data ( Figure 5).  in HRESIMS. Compound 3 was proposed to be a hydroxylated product of euphorbiasteroid due to its molecular weight being 16 mass units more than that of euphorbiasteroid. When comparing its 1 H and 13 C NMR data with those of euphorbiasteroid (Tables 1 and 3), it was revealed that the two compounds shared a great similarity. In the NMR spectra, the appearance of an extra oxygenated methylene unit [δ H 4.28 (d, J = 12.38 Hz, H-20a) and δ H 4.42 (d, J = 12.38 Hz, H-20b); δ C 58.1], in place of the methyl signals for C-20 (δ H 1.82; δ C 12.3) in euphorbiasteroid, suggested that one hydroxyl was incorporated to 3 at C-20 position. Key HMBC correlations of two protons of the hydroxymethyl (δ H 4.28 and 4.42) with C-14 (δ C 198.4) and C-12 (δ C 147.6) attached a hydroxyl to C-20 ( Figure 6). Assignments of the 1 H and 13 C NMR signals were achieved by a combination of DEPT, 1 H-1 H COSY, HSQC, and HMBC experiments (Figures S21-S30). Thus, the structure of compound 3 was identified as 20-hydroxyl euphorbiasteroid.
R PEER REVIEW 21 of 32 ( Figure 7), the key NOE correlations of two hydroxymethyl protons (δH 3.52 and 3.44) with H-9 (δH 1.21) and H-11 (δH 1.65) proved that the C-18 of compound 1 was substituted by a hydroxyl, and assigned the relative configuration of 18-hydroxymethyl to be α-oriented. Unambiguous complete assignments for the 1 H and 13 C NMR signals were made by combination of DEPT, 1 H-1 H COSY, HSQC, HMBC, and NOESY spectra ( Figure  S1-S10). On the basis of the above evidence, the structure of compound 1 was thus established as 18α-hydroxyl euphorbiasteroid.   (Tables 1 and 3) The only difference was the appearance of the signals for an oxygenated methine (δ H 5.10, δ C 72.8) in 4, rather than the C-7 methylene signals (δ H 3.58, δ C 41.5). Thus, it was speculated that compound 4 was a C-7 hydroxylated product of euphorbiasteroid. In addition, relative to euphorbiasteroid, the chemical shifts of C-1 and C-8 of compound 4 were downfield shifted significantly from δ C 133.8 to δ C 138.4, and from δ C 170.9 to δ C 173.7, respectively, while the resonances of C-2 and C-6 was upfield shifted from δ C 129.4 to δ C 126.5. The key HMBC correlations from H-7 at δ H 5.10 to C-1 , C-2 , C-6 and C-8 ( Figure 6) further proved that a hydroxyl was substituted at C-7 position. The stereo configuration of 7 -OH failed to be determined due to a lack of substantial NOE correlation (Figures S31-S40). Therefore, compound 4 was assigned as 7 -hydroxyl euphorbiasteroid.  Compound 2 was isolated as a white powder with [ ] +74.89° (c 0.150, MeOH The molecular formula was assigned as C32H40O9 based on its HR-ESI-MS data that d played an [M + H] + ion at m/z 569.2747. Similar to compound 1, the molecular weight o was 16 mass units more than that of euphorbiasteroid and shared the same molecu formula. The 1 H and 13 C NMR data of compound 2 were quite close to those of euph biasteroid (Tables 1 and 3), except for the absence of an aromatic proton. In the 1 H NM spectrum of 1, two groups of aromatic protons (each two protons) supported the pr ence of a p-substituted benzyl ring. Two groups of carbon resonances (each two carbo at δC 130.6 and 115.4 as well as the 13 C chemical shift of C-4′ proved the above deductio   (Tables 1 and 3), compared to euphorbiasteroid, it was observed to have one additional oxygenated quaternary carbon at δ C 154.7, replacing an aromatic methine of the benzyl ring. Additionally, the 1 H NMR spectrum exhibited four aromatic proton signals for an AA BB coupling system at δ H 6.90, 7.18, 6.88, and 7.11. The above information, in combination with key HMBC correlations from H 2 -7 (δ H 3.70 and 3.55) to C-2 (δ C 154.7) and C-6 (δ C 131.1), supported that a hydroxyl was substituted at the C-2 of the benzyl ring. Unambiguously complete assignments for the 1 H and 13 C NMR signals were made by a combination of DEPT, 1 H-1 H COSY, HSQC, and HMBC spectra (Figures S41-S50). Thus, compound 5 was identified as 2 -hydroxyl euphorbiasteroid.
The molecular formula of compound 6 was deduced to be C 32 H 40 O 10 from the HR-ESI-MS [M + H] + ion m/z 585.2704 (calcd for C 32 H 41 O 10 , 585.2694). Its molecular weight was 16 mass units more than that of compound 5, implying that compound 6 might be a dihydroxylated product of euphorbiasteroid. The 1 H and 13 C NMR data closely resembled those of 5 ( Figures S51-S60). The only difference was that compound 6 had an extra hydroxymethyl (δ H 3.48, 3.38; δ C 70.9), instead of a methyl (δ H 1.20; δ C 28.9) of 5. In the HMBC spectrum ( Figure 6), key long-range correlations from two hydroxymethylene protons at δ H 3.48 and 3.38 to C-9 (δ C 31.1) and C-11 (δ C 26.2) revealed that a hydroxyl was linked to C-18 or C-19 position. The observation of NOE correlations (Figure 7) from the hydroxymethylene protons at δ H 3.48 and 3.38 to H-9 (δ H 1.33) and H-11 (δ H 1.78) unambiguously attributed the hydroxymethylene as C-18, and determined the relative configuration of 18-CH 2 OH to be α-oriented. Therefore, the chemical structure of compound 6 was identified as 18α, 2 -dihydroxyl euphorbiasteroid.
Compound 7 was obtained as a white powder with an optical rotation [α] 20 D +60.00 • (c 0.170, MeOH). Its molecular formula was deduced as C 32 H 42 O 10 due to the [M + H] + ion at m/z 587.2851 (calcd for C 32 H 43 O 10 , 587.2878) in HR-ESI-MS. Interestingly, the molecular weight of 7 was 34 mass units more than that of euphorbiasteroid, and had one less degree of unsaturation, suggesting that there was a possible missing double bond or ring as well as two hydroxyl substituents. The 1 H and 13 C NMR spectroscopic data (Tables 2 and 3) showed that compound 7 shared most of its featured functionalities and had the same diterpene skeleton as euphorbiasteroid, but had a great difference in the phenylacetyl moiety. Only three aromatic proton signals occurred at δ H 5.54, 7.54, and 6.07 in the 1 H NMR spectrum, and four aromatic carbon resonances were observed at δ C 150. the key HMBC long-range correlations from H 2 -7 (δ H 2.67, 2.54) to C-2 (δ C 115.5) and C-6 (δ C 72.9) ( Figure 6) evidenced that the C-5 and C-6 double bond of phenyl ring was hydrogenated and subsequently substituted by two hydroxyls, which was in agreement with the molecular weight and degrees of unsaturation of 7 as mentioned above. The stereo configurations of 5 -OH and 6 -OH failed to be determined due to a lack of substantial NOE correlations in the NOESY spectrum of compound 7 (Figures S61-S70). Based on the above analysis, the chemical structure of compound 7 was identified as 5 ,6 -dihydroxyl dihydroeuphorbiasteroid.
Compound  6.15) implied that the C-3 and C-4 double bond of phenyl ring was hydrogenated and subsequently substituted by two hydroxyls, respectively. The above deduction was further proved by key HMBC correlations of H 2 -7 (δ H 3.69, 2.55) with the aromatic C-2 (δ C 117.6) and C-6 (δ C 130.7), and of H-4 [δ H 5.66 (brd., J = 5.56 Hz) with C-6 ( Figures S71-S80). Thus, the structure of compound 8 was identified as 3 ,4 -dihydroxyl dihydroeuphorbiasteroid. The stereo configurations of 3 -OH and 4 -OH failed to be determined due to a lack of substantial NOE correlations.  10 . The molecular weight of 9 was 30 mass units more than that of the substrate euphorbiasteroid, implying that compound 9 might be a carboxylated product. Compared to the substrate euphorbiasteroid, the 1 H and 13 C NMR spectra (Tables 2 and 3) exhibited an additional carboxyl signal (δ C 179.8), but an absence of the signals for CH 3 -18 (δ H 1.20; δ C 28.9). Additionally, the chemical shift of C-19 was significantly shifted to high field by 6.4 ppm. These data implied that compound 9 might be a C-18 carboxylated product of euhporbiasteroid. This deduction was substantially proved by key HMBC correlations ( Figure 6) (Tables 2 and 3). The downfield shifted carbon signals at C-1 (∆δ +2.0)/C-4 (∆δ +1.7)/C-5 (∆δ +0.8)/C-14 (∆δ +2.4) and the upfield shifted C-15 (∆δ -3.0) suggested that the acetoxyl at C-15 was replaced by one hydroxyl. Furthermore, the key HMBC correlations from the protons at δ H 5.92 (H-5) and 5.46 (H-3) to C-15 (δ C 88.7) further supported the above deduction ( Figure 6). Interpretation of the NOESY spectrum revealed that compound 11 shared the same relative configurations as those of M0 (Figure 7). Therefore, the structure of compound 11 was identified as 15-deacetyl euphorbiasteroid ( Figures S94-S103).
The HR-ESI-MS spectrum of compound 12 gave a hydrogen adduct ion [M + H] + m/z 731.3280 (calcd for C 38 H 50 O 14 , 731.3273), assigning the molecular formula as C 38 H 50 O 14 . Its molecular weight was 162 Da more than that of the substrate euphorbiasteroid, implying that compound 12 might be a hexosylated product of euphorbiasteroid. The 1 H and 13 C NMR spectroscopic data of compound 12 were very similar to those of the substrate euphorbiasteroid, and the major difference was the presence of an additional glucosyl unit [δ H 4.81 and 3.49-3.89 (6H); δ C 101.4, 75.9, 75.9, 73.5, 69.6, 61.8] in the structure of 12 (Tables 2 and 3). The substitution position of glucosyl was unambiguously determined to be the C-2 of the phenyl ring due to key HMBC correlations from the anomeric proton (δ H 4.81) of glucosyl and H 2 -7 (δ H 3.67, 3.64) to C-2 (δ C 155.3) (Figure 6). The 13 C chemical shifts of glucosyl and the large coupling constant of the anomeric proton (J H-1 /H-2 = 6.77 Hz) attributed the glucosyl to be β-D-glucose. Assignment of the 1 H and 13 C NMR signals was achieved by a combination of DEPT, 1 H-1 H COSY, and HSQC experiments ( Figure S104-S113). Based on this evidence, the structure of compound 12 was determined to be euphorbiasteroid 2 -O-β-D-glucopyranoside.

Comparison of Metabolite Formation In Vitro and In Vivo
Various results of metabolite formation were discovered in the three approaches used in this study. Rat plasma, urine, and feces samples were generally rich in metabolites, with 27, 20, and 29 metabolites, respectively. In vitro incubation of RLMs produced the same 18 phase I metabolites as the metabolites in rats, just less in number and amount. In vitro co-incubation of C. elegans bio-110930 with euphorbiasteroid yielded 14 metabolites, 12 of which were further prepared by large scale microbial transformation and confirmed to be the same metabolites as those in rats by comparison of their HPLC retention times and MS/MS fragments (Figure 8

Proposed Metabolic Pathways of Euphorbiasteroid In Vivo
Based on the metabolites identified in rats (plasma, urine, and feces), RLMs and fungus mycelium (Table 4), metabolic pathways of euphorbiasteroid can be proposed.
According to the above discoveries in the metabolism of euphorbiasteroid, the hy-

Proposed Metabolic Pathways of Euphorbiasteroid In Vivo
Based on the metabolites identified in rats (plasma, urine, and feces), RLMs and fungus mycelium (Table 4), metabolic pathways of euphorbiasteroid can be proposed.
Finally, phase II metabolism was the predominant pathway in rat feces samples, where several metabolites were formed by glycosylation and sulfonation ( Figure 12). Overall, hydroxylation, oxidation, hydrolysis, sulfonation, and glycosylation are the main metabolic pathways of euphorbiasteroid in rats.

Cytotoxicity of Euphorbiasteroid and Its Metabolites
Euphorbiasteroid and its metabolites were tested for their cytotoxicity on human cell lines SH-SY5Y, LO2, AC-16, and HK-2 by the CCK-8 assay. The results (Table 5) indicated that euphorbiasteroid showed no cytotoxicity against four human cell lines with IC50 values of more than 50 μM. Among metabolites, the C-20 hydroxylated product M3 (20-hydroxyl euphorbiasteroid) and two hydrolysis products M24 (epoxylathyrol) and M25 (15-deacetyl euphorbiasteroid) showed significant cytotoxicity against four human cell lines with IC50 values from 3.60 μM to 40.74 μM. Therefore, considering the high content of euphorbiasteroid in Euphorbiae semen, it was speculated that the metabolites from hydroxylation and hydrolysis might be the potential toxic constituents of Euphorbiae semen. Finally, phase II metabolism was the predominant pathway in rat feces samples, where several metabolites were formed by glycosylation and sulfonation ( Figure 12). Overall, hydroxylation, oxidation, hydrolysis, sulfonation, and glycosylation are the main metabolic pathways of euphorbiasteroid in rats.
Finally, phase II metabolism was the predominant pathway in rat feces samples, where several metabolites were formed by glycosylation and sulfonation ( Figure 12). Overall, hydroxylation, oxidation, hydrolysis, sulfonation, and glycosylation are the main metabolic pathways of euphorbiasteroid in rats.

Cytotoxicity of Euphorbiasteroid and Its Metabolites
Euphorbiasteroid and its metabolites were tested for their cytotoxicity on human cell lines SH-SY5Y, LO2, AC-16, and HK-2 by the CCK-8 assay. The results (Table 5) indicated that euphorbiasteroid showed no cytotoxicity against four human cell lines with IC50 values of more than 50 μM. Among metabolites, the C-20 hydroxylated product M3 (20-hydroxyl euphorbiasteroid) and two hydrolysis products M24 (epoxylathyrol) and M25 (15-deacetyl euphorbiasteroid) showed significant cytotoxicity against four human cell lines with IC50 values from 3.60 μM to 40.74 μM. Therefore, considering the high content of euphorbiasteroid in Euphorbiae semen, it was speculated that the metabolites from hydroxylation and hydrolysis might be the potential toxic constituents of Euphorbiae semen.

Cytotoxicity of Euphorbiasteroid and Its Metabolites
Euphorbiasteroid and its metabolites were tested for their cytotoxicity on human cell lines SH-SY5Y, LO2, AC-16, and HK-2 by the CCK-8 assay. The results (Table 5) indicated that euphorbiasteroid showed no cytotoxicity against four human cell lines with IC 50 values of more than 50 µM. Among metabolites, the C-20 hydroxylated product M3 (20-hydroxyl euphorbiasteroid) and two hydrolysis products M24 (epoxylathyrol) and M25 (15-deacetyl euphorbiasteroid) showed significant cytotoxicity against four human cell lines with IC 50 values from 3.60 µM to 40.74 µM. Therefore, considering the high content of euphorbiasteroid in Euphorbiae semen, it was speculated that the metabolites from hydroxylation and hydrolysis might be the potential toxic constituents of Euphorbiae semen.

Conclusions
In the present study, euphorbiasteroid metabolites generated in vivo (rat plasma, urine and feces) and in vitro (RLMs and C. elegans bio-110930 model) were characterized through UPLC-Q/TOF-MS. According to the molecular ions and the MS/MS fragments, a total of 31 metabolites were identified, including 27 phase I and 4 phase II metabolites. Additionally, the structures of twelve metabolites were exactly confirmed by comparing their HPLC retention times and MS/MS fragments with those of the prepared reference standards, whose structures were exactly determined based on 1D and 2D NMR analysis. The twelve identified metabolites were 18α-hydroxyl euphorbiasteroid (M1), 4 -hydroxyl euphorbiasteroid (M2), 20-hydroxyl euphorbiasteroid (M3), 7 -hydroxyl euphorbiasteroid (M4), 2 -hydroxyl euphorbiasteroid (M5), 18α,2 -dihydroxyl euphorbiasteroid (M8), 5 ,6 -dihydroxyl dihydroeuphorbiasteroid (M12), 3 ,4 -dihydroxyl dihydroeuphorbiasteroid (M13), 18-carboxyl euphorbiasteroid (M16), epoxylathyrol (M24), 15-deacetyl euphorbiasteroid (M25), and euphorbiasteroid 2 -O-β-D-glucopyranoside (M29). These results showed that the majority of phase I metabolites were generated by hydroxylation and hydrolysis, followed by oxidation and hydroxylation. Glycosylation and sulfonation played significant roles in the formation of phase II metabolites. Moreover, RLMs and C. elegans bio-110930 could be suitable models to simulate and prepare phase I metabolites of euphorbiasteroid. Thus, an overall description of metabolites of euphorbiasteroid from rats, RLMs and C. elegans bio-110930 has been provided. Three metabolites M3, M24, and M25 exhibited potent cytotoxicity against four human cell lines. Furthermore, our study provides valuable information for predicting in vivo human metabolites and important clues for further clarifying the mechanism of drug toxicity of euphorbiasteroid and its metabolites. The method can also be applied to the study of other herbal components, providing new ideas in the field of metabolic studies of traditional Chinese medicine.
Author Contributions: X.Z. and Y.S. designed the study, aided in funding acquisition and reviewed the manuscript. S.X., X.X. and X.W. conducted the study and analyzed the data. S.X. and and X.W. drafted the original manuscript. S.L., Y.L., W.C. and B.X. aided in animal experiments. J.X. and W.Y. aided in the design of the illustrations. All authors have read and agreed to the published version of the manuscript.