Analysis of In Vivo Existence Forms of Nardosinone in Mice by UHPLC-Q-TOF-MS Technique

Nardosinone, a sesquiterpene peroxide, is one of the main active constituents of the ethnomedicine Nardostachyos Radix et Rhizoma, and it has many bioactivities, such as antiarrhythmia and cardioprotection. To elucidate its in vivo existence forms, its metabolism is first studied using mice. All urine and feces are collected during the six days of oral dosing of nardosinone, and blood is collected at one hour after the last dose. Besides, to validate some metabolites, a fast experiment is performed, in which nardosinone was orally administered and the subsequent one-hour urine is collected and immediately analyzed by UHPLC-Q-TOF-MS. In total, 76 new metabolites are identified in this study, including 39, 51, and 12 metabolites in urine, plasma, and feces, respectively. Nardosinone can be converted into nardosinone acid or its isomers. The metabolic reactions of nardosinone included hydroxylation, hydrogenation, dehydration, glucuronidation, sulfation, demethylation, and carboxylation. There are 56 and 20 metabolites with the structural skeleton of nardosinone and nardosinone acid, respectively. In total, 77 in vivo existence forms of nardosinone are found in mice. Nardosinone is mainly excreted in urine and is not detected in the feces. These findings will lay the foundation for further research of the in vivo effective forms of nardosinone and Nardostachyos Radix et Rhizoma.


Introduction
Nardosinone is a sesquiterpene peroxide and exists in a variety of medicinal plants, such as Nardostachys jatamansi DC.
Modern pharmacological studies have shown that nardosinone could exert antiinflammatory, neuroprotective, cardioprotective, and more pharmacological actions [1].   To conveniently describe the fragment characteristics of nardosinone and its metabolites, the three rings were named A, B, and C, respectively. The C-C bonds and carbon atoms in the skeleton were designated by the letters a-n and numbers 1-15 ( Figure 2).
In this study, the usual neutral losses in mass spectrometry were 58.   The proposed fragmentation pathway of nardosinone. The pink circle represents nardosinone, and the C-C bonds and carbon atoms in the skeleton were designated by the letters a-n and numbers 1-15; A, B and C represent the three rings, respectively.

Mass Spectral Fragmentation Features of Nardosinone
The proposed fragmentation pathway of nardosinone is shown in Figure 2. Nardosinone (C 15 14.02 Da (CH 2 ) from the ion at m/z 207.10, which could imply that hydroxylation did not occur at C-15. C-3 was a secondary carbon with parahydrogen, and C-4 was a tertiary carbon with tertiary hydrogen, which was more active than the parahydrogen of C-3, so it was inferred that the hydroxylation was more likely to occur at C-4. Thus, M1 was preliminarily identified as 4-hydroxyl nardosinone. The possible structure and fragmentation pathways of M1 are shown in Figure 3. spectra, which inferred that the hydroxylation occurred at C-3, C-4, or C-15. The ion at m/z 177.06 (C 10 H 9 O 3 ) was formed by the consecutive losses of two 14.02 Da (CH 2 ) from the ion at m/z 207.10, which could imply that hydroxylation did not occur at C-15. As described in the identification of M1, C-3 was a secondary carbon with parahydrogen, and C-4 was a tertiary carbon with tertiary hydrogen, which was more active than the parahydrogen of C-3, so it was inferred that the hydroxylation was more likely to occur at C-4. Therefore, M3 was preliminarily identified as 4-hydroxyl nardosinone. The possible structure and fragmentation pathways of M3 are shown in Figure 5. were not observed in MS 2 spectra, implying that the hydroxylation did not occur at C-2, C-3, C-4, C-6, C-7, C-12, C-13, or C-15. The ion at m/z 165.09 (C 10 H 13 O 2 ) consecutively lost 16.03 Da (CH 4 ) and 43.99 Da (CO 2 ) to generate m/z 149.06 (C 9 H 9 O 2 ) and m/z 105.07 (C 8 H 9 ), suggesting that the hydroxylation did not occur at C-1, C-2, and C-14. Thus, it was speculated that the hydroxylation occurred at C-8, and M4 was preliminarily identified as 8-hydroxyl nardosinone. The possible structure and fragmentation pathways of M4 are shown in Figure 6.  M3: The ions at m/z 207.10 (C12H15O3) and m/z 149.06 (C9H9O2) were formed by the consecutive losses of two C3H6O from [M−H] − in the MS 2 spectra, which inferred that the hydroxylation occurred at C-3, C-4, or C-15. The ion at m/z 177.06 (C10H9O3) was formed by the consecutive losses of two 14.02 Da (CH2) from the ion at m/z 207.10, which could imply that hydroxylation did not occur at C-15. As described in the identification of M1, C-3 was a secondary carbon with parahydrogen, and C-4 was a tertiary carbon with tertiary hydrogen, which was more active than the parahydrogen of C-3, so it was inferred that the hydroxylation was more likely to occur at C-4. Therefore, M3 was preliminarily identified as 4-hydroxyl nardosinone. The possible structure and fragmentation pathways of M3 are shown in Figure 5.  , which implied that the hydroxylation did not occur at C-8. It also lost 40.03 Da (C 3 H 4 ) to generate m/z 137.03 (C 7 H 5 O 3 ), suggesting that the hydroxylation did not occur at C-2. Therefore, it was speculated that the hydroxylation should occur at C-1, and M6 was preliminarily identified as 1-hydroxyl nardosinone. The possible structure and fragmentation pathways of M6 are shown in Figure 8.   √ " represents the possible structure. "×" represents the impossible structure. The orange rim represents the possible structures. The red rim represents the most possible fragmentation pathway. occur at C-15. If hydroxylation occurred at C-3, C-4, or C-15, the ion at m/z 207.10 (C12H15O3) would directly lose 70.04 Da (C4H6O) to form the ion at m/z 137.06 (C8H9O2). In addition, the ions at m/z 165.09 (C10H13O2) and m/z 137.06 (C8H9O2) formed by the consecutive losses of 42.01 Da (C2H2O) and 28.03 Da (C2H4) from the ion at m/z 207.10 (C12H15O3) were observed in the MS 2 spectra. Therefore, M5 was preliminarily identified as 3-hydroxyl nardosinone or 4-hydroxyl nardosinone. The possible structure and fragmentation pathways of M5 are shown in Figure 7.  , suggesting that the hydroxylation did not occur at C-8 and C-14. Therefore, the hydroxylation should occur at C-1 or C-2. C-2 has α-H of alkene, which is more active. Consequently, it was speculated that the hydroxylation more likely occurred at C-2, and M7 was identified as 2-hydroxyl nardosinone. The possible structure and fragmentation pathways of M7 are shown in Figure 9.  (1-hydroxyl nardosinone). " √ " represents the possible structure. "×" represents the impossible structure. The olive rim represents the possible structures; C represents the hexatomic ring. The red rim represents the most possible fragmentation pathway.  Figure 9. The proposed fragmentation pathways of M7 (2-hydroxyl nardosinone). "√" represents the possible structure. "×" represents the impossible structure. The atrovirens rim represents the possible structures; A and C represent the hexatomic rings. The red rim represents the most possible fragmentation pathway.  It was speculated that the hydroxylation more likely occurred at C-3, C-4, or C-15 and a carbon-carbon double bond existed among C-3, C-4, and C-15. Therefore, the neutral loss of 56.03 Da (C 3 H 4 O) was speculated to be generated by the hydroxylation that occurred at C-15 and the dehydrogenation that occurred at C-3 and C-4, or the hydroxylation that occurred at C-15 and the dehydrogenation that occurred at C-15 and C-4, or the hydroxylation and dehydrogenation that both occurred at C-15, or the hydroxylation that occurred at C-3 and the dehydrogenation that occurred at C-3 and C-4. Therefore, M8 was preliminarily identified as 3-hydroxyl nardosinone or 15-hydroxyl nardosinone. The possible structure and fragmentation pathways of M8 are shown in Figure 10.  The molecular formula of M41 was predicted to be C 15  suggesting that the glucuronidation did not occur at the hydroxyl, which was generated by the cleavage of the peroxide bridge, but probably occurred at the hydroxyl of C-7 or C-9. While the neutral loss of 58.04 Da (C 3 H 6 O) in MS 2 spectra of M49, M52, and M53 was not observed, suggesting that the glucuronidation probably occurred at the hydroxyl, which was generated by the cleavage of the peroxide bridge.
The molecular formulae of M54−M56 were verified to be C 21 O 8 ). These implied that it was impossible that hydroxylation occurred at C-1, C-2, C-6, C-8, and C-14 and glucuronidation occurred at the hydroxyl of C-7, and hydroxylation, hydrogenation, and glucuronidation simultaneously occurred at C-7 of M55. In addition, m/z 193.03 (C 6 H 9 O 7 , glucuronic acid ion) was also observed in MS 2 of M55, and it was speculated that the compound easily lost glucuronic acid and formed conjugated systems in the aglycone, and the possibility that hydroxylation and glucuronidation simultaneously occurred at C-4, C-14, and C-15 was basically excluded because it could not form conjugated systems after losing glucuronic acid of M55. Moreover, the possibility that hydroxylation and glucuronidation simultaneously occurred at C-1 was also excluded, owing to the glycosidic bond of enol form hydroxyl at C-1 being difficult to break. In conclusion, it was speculated that M55 was generated by the hydrogenation of the peroxide bridge and simultaneous hydroxylation and glucuronidation at C-2, C-3, C-6, or C-8. . Therefore, they were identified as nardosinone acid glucuronide isomers. The ion at m/z 193.04 (C 6 H 9 O 7 ) was observed in the MS 2 spectra of M63−M64, implying that the glycosidic bond broke easily and there were no conjugated systems near the glycosidic bond. Because enol form hydroxyl of C-7 and carbonyl of C-9 both had a stable conjugated system, glucuronidation should occur at hydroxyl of C-6 (the rearrangement of C-7 hydroxyl to C-6) for M63−M64. The ion at m/z 193.04 was not observed in the MS 2 spectra of M65. Considering the energy difference required for bond cleavage, it was speculated that the glucuronidation occurred at the enol form hydroxyl of C-7 or the carbonyl of C-9, but the carbonyl of C-9 has already formed a stable conjugated system with the carbon-carbon double bond between C-10 and C-1, so the possibility that glucuronidation occurred at the carbonyl of C-9 was less. Therefore, it was speculated that the glucuronidation occurred at the enol form hydroxyl of C-7 in M65.

Identification of
M66 were not observed. Considering the energy difference required for bond cleavage, it was speculated that the sulfation occurred at enol form hydroxyl. However, the carbonyl of C-9 already formed a stable conjugated system with the carbon-carbon double bond between C-10 and C-1, so the possibility that sulfation occurred at the carbonyl of C-9 was less. Therefore, it was speculated that the sulfation occurred at the enol hydroxyl of C-7 in M66. In the MS 2 spectra of M67, its ) all indicated that the sulfate ester and the vicinal H were simultaneously lost, suggesting that the hydroxyl, which was sulfated was not enol form hydroxyl and did not in the conjugated system. Thus, it was speculated that the sulfation should occur at the hydroxyl of C-6. Therefore, M66−M70 were identified as nardosinone acid sulfates and the possible structures and the fragmentation pathways are shown in Figure S4.
The molecular formulae of M71−M76 were verified to be C 12 H 18 O 5 S based on their [M−H] − at m/z 273.08. The characteristic fragment ions at m/z 191.11 (C 12 H 15 O 2 ) and m/z 80.96 (HSO 3 ) were observed in the MS 2 spectra of M73, M74, and M76, and the characteristic fragment ion at m/z 80.96 (HSO 3 ) was observed in the MS 2 spectra of M71, M72, and M75. As mentioned above, it was speculated that the sulfation did not occur at the enol form hydroxyl of M71−M76, and the possible sulfation sites were deduced as follows: (1) if sulfation occurred at hydroxyl of C-6 and the hydrogenation occurred at carbon-carbon double bond between C-7 and C-8, the conjugated system could not be formed after losing H 2 SO 3 , thus, it was less likely to occur; (2) if sulfation occurred at hydroxyl of C-6 and the hydrogenation occurred at carbon-carbon double bond between C-10 and C-1, the conjugated system could be formed after losing H 2 SO 3 , thus, it was more likely to occur; (3) if sulfation occurred at hydroxyl of C-6 and the hydrogenation occurred at carbonyl of C-9, the conjugated system could not be formed after losing H 2 SO 3 , thus, it was less likely to occur; (4) if sulfation occurred at hydroxyl of C-7 (reduction of carbonyl), the conjugated system could be formed after losing H 2 SO 3 , thus, it was more likely to occur; (5) if the sulfation occurred at hydroxyl of C-9 (reduction of carbonyl), the conjugated system could be formed with the enol hydroxyl of C-7 after losing H 2 SO 3 , but the carbonyl of C-9 was less likely to be hydrogenated because it already formed a conjugated system, therefore, the possibility that sulfation and hydrogenation simultaneously occurred at carbonyl of C-9 was less. Since the sulfation could not occur at the enol form hydroxyl (because no conjugated system could be formed after losing H 2 SO 3 ) for M71−M76, when the hydrogenation occurred at the carbon-carbon double bond between C-10 and C-1, the sulfation should not occur at the enol form hydroxyl of C-7 and C-9. In conclusion, it was speculated that the sulfation should occur at the hydroxyl of C-6 and the hydrogenation occur at the carbon-carbon double bond between C-10 and C-1, or that the sulfation and hydrogenation simultaneously occur at the enol form hydroxyl of C-7. Compared to M66−M70, M71−M76 had two additional H atoms in their molecular formulae, therefore, M71−M76 were identified as hydrogenated nardosinone acid sulfates, and their possible structures and fragmentation pathways are shown in Figure S5.

Identification of Several Metabolites of Nardosinone with the Skeleton of Nardosinone Acid in the Fast Validation Experiment
We detected and identified three metabolites (F1-F3, the extracted ion chromatogram of them is shown in Figure S6). The molecular formulae of F1-F3 were predicted to be C 12 H 18 O 5 S according to their [M−H] − at m/z 273.08. The ion at m/z 79.96 (SO 3 ) was observed in the MS 2 spectra of F1, implying that the sulfation occurred at the enol form hydroxyl (C-7 or C-9); the ion at m/z 80.96 Da (HSO 3 ) was observed in the MS 2 spectra of F2 and F3, indicating that the sulfation occurred at the hydroxyl of C-6 or C-7 (reduction of carbonyl). F1-F3 showed [Aglycon−H] − (C 12 H 17 O 2 ) at m/z 193.12, which had two more H atoms than the molecular formula of nardosinone acid, thus F1-F3 were speculated as hydrogenated nardosinone acid sulfates. The larger CLogP value means a longer retention time in reversed-phase UHPLC, thus, the possible structures of F2 (ClogP = 1.52 and t R = 12.89 min) and F3 (ClogP = 1.78 and t R = 15.82 min) are shown in Figure 11. sulfation occurred at hydroxyl of C-6 and the hydrogenation occurred at carbonyl of C-9, the conjugated system could not be formed after losing H2SO3, thus, it was less likely to occur; (4) if sulfation occurred at hydroxyl of C-7 (reduction of carbonyl), the conjugated system could be formed after losing H2SO3, thus, it was more likely to occur; (5) if the sulfation occurred at hydroxyl of C-9 (reduction of carbonyl), the conjugated system could be formed with the enol hydroxyl of C-7 after losing H2SO3, but the carbonyl of C-9 was less likely to be hydrogenated because it already formed a conjugated system, therefore, the possibility that sulfation and hydrogenation simultaneously occurred at carbonyl of C-9 was less. Since the sulfation could not occur at the enol form hydroxyl (because no conjugated system could be formed after losing H2SO3) for M71−M76, when the hydrogenation occurred at the carbon-carbon double bond between C-10 and C-1, the sulfation should not occur at the enol form hydroxyl of C-7 and C-9. In conclusion, it was speculated that the sulfation should occur at the hydroxyl of C-6 and the hydrogenation occur at the carbon-carbon double bond between C-10 and C-1, or that the sulfation and hydrogenation simultaneously occur at the enol form hydroxyl of C-7. Compared to M66−M70, M71−M76 had two additional H atoms in their molecular formulae, therefore, M71−M76 were identified as hydrogenated nardosinone acid sulfates, and their possible structures and fragmentation pathways are shown in Figure S5.

Identification of Several Metabolites of Nardosinone with the Skeleton of Nardosinone Acid in the Fast Validation Experiment
We detected and identified three metabolites (F1-F3, the extracted ion chromatogram of them is shown in Figure S6). The molecular formulae of F1-F3 were predicted to be C12H18O5S according to their [M−H] − at m/z 273.08. The ion at m/z 79.96 (SO3) was observed in the MS 2 spectra of F1, implying that the sulfation occurred at the enol form hydroxyl (C-7 or C-9); the ion at m/z 80.96 Da (HSO3) was observed in the MS 2 spectra of F2 and F3, indicating that the sulfation occurred at the hydroxyl of C-6 or C-7 (reduction of carbonyl).

Comparative Analysis of the Metabolic Characteristics of Nardosinone and Other Sesquiterpene Peroxides
Artemisinin compounds are typical representatives of sesquiterpene peroxides, and many researchers have carried out the metabolism studies of them owing to their unique structures and antimalarial activity. It was inferred that the peroxide bridge was broken into a mono-oxygen bridge based on the metabolic researches of dihydroartemisinininin derived-dimer and dihydroartemisinin. Moreover, the main metabolic reactions included hydroxylation, dehydration, glucuronidation, carbonylation, dehydrogenation, etc. It was worth noting that the metabolic reactions of mono-hydroxylation, dihydroxylation, trihydroxylation, and quahydroxylation were not found in the metabolism of dihydroartemisinin but dihydroartemisinininin derived-dimer. Furthermore, carboxylated metabolites were found in the in vitro metabolism of artemether [31,32]. The in vivo and in vitro metabolism of artemisinin and dihydroartemisinin manifested that the peroxide bridge could be broken into a mono-oxygen bridge and the metabolic reactions mainly included dihydroxylation, deoxidation, hydroxylation, and glucuronidation, and 25 metabolites of artemisinin and 16 metabolites of dihydroartemisinin were identified, respectively [25]. The metabolites of hydroxylated dihydroartemisinin (DHA+O), dehydro-hydroxylated dihydroartemisinin (DHA−H2+O), and dehydration-hydroxylated dihydroartemisinin (DHA−H 2 O+O) were identified in the in vivo metabolism of dihydroartemisinin [33], which resembled the metabolites of hydroxylated nardosinone, dehydro-hydroxylated nardosinone, and dehydra-hydroxylated nardosinone in this study. Compared with the metabolism of other sesquiterpene peroxides, combined with the metabolic pathway and metabolites of nardosinone, we speculated that the peroxide bridge cleavage (loss of C 3 H 6 O) of nardosinone was more likely to generate enol hydroxyl or carbonyl other than the mono-oxygen bridge, which was inconsistent with the cleavage of the peroxide bridge and the formation of a mono-oxygen bridge in the metabolism of artemisinin. Furthermore, the common metabolic reactions reported in other sesquiterpene peroxides were also discovered in the metabolism of nardosinone, e.g., hydroxylation, carboxylation, sulfation, glucuronidation, etc.

Discussion on the Origin of Metabolites with the Nardosinone Acid Skeleton
It was reasonable that the hydroxy-isopropyl (C 3 H 6 O) could be lost in the fragmentation pathway of nardosinone, while we found many metabolites with the nardosinone acid skeleton (e.g., M57−M76) in biological samples. The loss of hydroxy-isopropyl involved the cleavage of the C-C bond, which is a rare metabolic reaction. Therefore, we designed a fast validation experiment to confirm it. If metabolites with a nardosinone acid skeleton are detected in this experiment, it means that they are indeed generated by in vivo metabolism.
According to the literature, nardosinone is stable under the conditions of an alkaline, low temperature, and away from light [18]. Therefore, the environmental factors should be strictly controlled to inhibit the degradation of nardosinone during the experimental processes.
In our fast validation experiment, nardosinone was suspended in a 0.5% carboxymethyl cellulose sodium (CMC-Na) solution (alkaline) and stored at −20 • C in a dark place. Then the mice were orally administered with nardosinone once, and in the first-hour urine samples of mice were collected and immediately filtered through a 0.22-µm membrane and analyzed by using UHPLC-Q-TOF-MS as quickly as possible. However, the metabolites with a nardosinone acid skeleton were still detected, which indicated that nardosinone acid and its metabolites were not generated by experimental operations such as ultrasonic extraction but generated by the in vivo metabolism of nardosinone. Furthermore, the results of the liver metabolism of the Cistanche deserticola total glycosides showed that the Cistanche deserticola total glycosides could lose the C 3 H 5 O group (m/z 57.06) to generate the metabolite methylated laricin [34]. Therefore, we consider that it is reasonable that nardosinone can lose hydroxy-isopropyl (C 3 H 6 O) to generate nardosinone acid in the in vivo metabolism.
But we still do not know where nardosinone loses the hydroxy-isopropyl (C 3 H 6 O) to generate nardosinone acid in vivo, and it could be the stomach, the liver, the intestinal tract, or somewhere else. We speculate that nardosinone is transformed into nardosinone acid by the intestinal microflora. Therefore, it deserves further research to explore the in vivo formation mechanism of nardosinone acid.

Reagent
The purity of nardosinone was greater than 98% (UHPLC, 254 nm) and it was pur-

Animal Experiments
Nine ICR mice (male, 30 ± 2 g) were purchased from the Department of Laboratory Animal Sciences at the Peking University Health Science Center. They were randomized into three groups (test group I, test group II, and blank group) with three mice in each group. The experiment lasted for 10 d for the test group I and blank group, and 4 d for group II. All mice were housed in mouse metabolism cages with water and food ad libitum for the first three days, then the mice were dosed by gavage once per day for the following 7 d for the test group I and blank group, and 1 d for the test group II. The mice of test groups were orally administered with a dose of 80 mg/kg mouse-weight nardosinone suspended in 2.0 mL 0.5% carboxymethyl cellulose sodium (CMC-Na) solution, and the mice of blank group were orally administered with the same volume of 0.5% carboxymethyl cellulose sodium (CMC-Na) solution. The mice were allowed to eat and drink ad libitum. All animal experiments were approved by the Animal Ethics Committee of Peking University Health Science Center (approval number: LA2019117).

Collection of Samples
In total, 8 mL ethanol was added into each urine collection tube for bacteriostasis before each urine collection. All the urine and feces were collected during the six days of dosing for the test group I and blank group, and one hour after the last oral administration, the blood was collected into 1.5 mL heparin sodium-containing tubes by excising the eyeballs. All samples were kept at −80 • C.
The first-hour urine of mice after orally administered with nardosinone was collected for the test group II, and was immediately filtered through 0.22-µm membrane, and analyzed by UHPLC-Q-TOF-MS directly. It took 2.5 h from drug administration to obtain mass spectra data.

Preparation of Samples
All urine samples from the test group I and blank group were merged into two samples (test group I and blank group), respectively. Each sample was centrifuged at 8000 rpm at 4 • C for 15 min, and the supernatant was harvested, concentrated, and dried at 55 • C. Then, the urine residue was ultrasonically extracted with 10 times volume of methanol for 30 min and filtered to get the supernatant. The supernatant was evaporated to dryness at 55 • C. Finally, 0.5 g residue was dissolved in 1.0 mL methanol, and was filtered through 0.22-µm membrane, and stored at −80 • C before further analysis.
All feces samples from the test group I and blank group were combined into two samples (test group I and blank group) and were dried at 50 • C for 48 h and mashed. Each feces sample was ultrasonically extracted 30 min with 10 times volume of methanol for three times. The three extracts were filtered and merged, and dried at 55 • C. Next, the residue was ultrasonically extracted by 10 times volume of methanol for 30 min and centrifuged at 8000 rpm at 4 • C for 15 min to get the supernatant, which was further evaporated to dryness at 55 • C to obtain the residue. Finally, 0.5 g residue was dissolved in 1.5 mL methanol, and was filtered through 0.22-µm membrane, and stored at −80 • C before further analysis.
All plasma samples from the test group I and blank group were mixed into two samples (test group I and blank group) and centrifuged at 5000 rpm at 4 • C for 15 min to collect 1.0 mL supernatant, and it was mixed with 5 mL methanol and centrifuged at 5000 rpm at 4 • C for 15 min. The supernatant was separated and dried by nitrogen blow at 40 • C. Finally, 0.2 mg residue was dissolved in 0.3 mL methanol, and filtered through 0.22-µm membrane, and stored at −80 • C before further analysis.

Identification of the Existence Forms of Nardosinone In Vivo (Original Constituents and Metabolites)
The base peak chromatograms (BPCs) of the drug-containing group and blank group samples were compared to find the distinguishing peaks and tentatively determine the in vivo existence forms of nardosinone. Afterward, the distinguishing peaks were confirmed by comparing the corresponding extracted ion chromatograms (EICs) of the drugcontaining group and blank group. If a compound could be detected by the extracted ion chromatogram (EIC) of the drug-containing group but did not exist in blank group, it was preliminarily verified as a metabolite. Furthermore, the MS data of reference substances, the MS fragmentation information reported in the literature, and the information obtained by searching the SciFinder database could all be used to identify the in vivo forms [35,36].

Conclusions
The in vivo metabolism of nardosinone was studied for the first time in mice. A total of 76 new metabolites were identified by UHPLC-Q-TOF-MS technology, and the metabolic reactions mainly included hydroxylation, dehydration, hydrogenation, sulfation, glucuronidation, demethylation, carboxylation, etc. There were 56 and 20 metabolites with the skeletons of nardosinone and nardosinone acid, respectively. It was confirmed that nardosinone could be biotransformed into nardosinone acid or isomer in vivo based on the results of the fast validation experiment. These results will be conducive to an in-depth study of the in vivo effective forms of nardosinone and NRR and will have certain reference values for the in vivo metabolism studies of other sesquiterpene peroxides.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/molecules27217267/s1, Figure S1: Extracted ion chromatograms (EICs) of 39 metabolites of nardosinone in mice urine. A, B, C, and D represent the EICs of 39 metabolites of nardosinone group, the magnified EICs of 39 metabolites of nardosinone group, the EICs of 39 metabolites of blank group, the magnified EICs of 39 metabolites of blank group, respectively. Figure S2: Extracted ion chromatograms (EICs) of 51 metabolites of nardosinone in mice plasma. A, B, C, and D represent the EICs of 51 metabolites of nardosinone group, the magnified EICs of 51 metabolites of nardosinone group, the EICs of 51 metabolites of blank group, the magnified EICs of 51 metabolites of blank group, respectively. Figure S3: Extracted ion chromatograms (EICs) of 12 metabolites of nardosinone in mice feces. A and B represent the EICs of 12 metabolites of nardosinone group and blank group, respectively. Figure S4: The possible structures and proposed fragmentation pathways of M66-M70. The red rim represents the proposed fragmentation pathways of M67-M70 and the blue rim represents the proposed fragmentation pathway of M66. Figure S5