Metabolism of Paeoniae Radix Rubra and its 14 constituents in mice

Objective: Paeoniae Radix Rubra (PRR) is a commonly used traditional Chinese medicine with the effects of clearing away heat, cooling the blood, and relieving blood stasis. To 1) elucidate the metabolites and metabolic pathways of PRR and its 14 main constituents in mice and 2) reveal the possible origins of the known effective forms of PRR and their isomers, the metabolism of PRR in mice was systematically studied for the first time. Methods: PRR and its 14 constituents were administered to mice by gavage once a day for seven consecutive days, respectively. All urine and feces were collected during the 7 days of dosing, and blood was collected at 1 h after the last dose. Metabolites were detected and identified using high performance liquid chromatography with diode array detector and combined with electrospray ionization ion trap time-of-flight multistage mass spectrometry (HPLC-DAD-ESI-IT-TOF-MSn). Results: In total, 23, 16, 24, 17, 18, 30, 27, 17, 22, 17, 33, 3, 8, 24, and 31 metabolites of paeoniflorin, albiflorin, oxypaeoniflorin, benzoylpaeoniflorin, hydroxybenzoylpaeoniflorin, benzoyloxypaeoniflorin, galloylpaeoniflorin, lactiflorin, epicatechin gallate, catechin gallate, catechin, ellagic acid, 3,3′-di-O-methylellagic acid, methylgallate, and PRR were respectively identified in mice; after eliminating identical metabolites, a total of 195 metabolites remained, including 8, 11, 25, 17, 18, 30, 27, 17, 21, 17, 1, 2, 8, 20, and 20 newly identified metabolites, respectively. The metabolic reactions of PRR and its 14 main constituents in mice were primarily methylation, hydrogenation, hydrolysis, hydroxylation, glucuronidation, and sulfation. Conclusion: We elucidated the metabolites and metabolic pathways of PRR and its 14 constituents (e.g., paeoniflorin, catechin, ellagic acid, and gallic acid) in mice and revealed the possible origins of the 10 known effective forms of PRR and their isomers. The findings are of great significance to studying the mechanism of action and quality control of PRR.


Introduction
The effective forms of traditional Chinese medicines can be the original constituents or the active metabolites produced in vivo (Xu et al., 2022). The metabolism of traditional Chinese medicines is the key link between their phytochemistry in vitro and their pharmacological activity in vivo. Therefore, studying the metabolism of traditional Chinese medicines is crucial to understanding the forms of the medicine that exist and are active in vivo along with the mechanisms of action of traditional Chinese medicines.
Paeoniae Radix Rubra (PRR), obtained from the dried roots of Paeonia lactiflora Pall. or Paeonia veitchii Lynch, is a commonly used traditional Chinese medicine with the effects of clearing away heat, cooling the blood, and relieving blood stasis (Zhao et al., 2021). PRR has many pharmacological effects, such as preventing liver fibrosis, curing jaundice, improving cholestasis in rats, relieving inflammation, and improving myocardial infarction, hypertrophy, and fibrosis (Tan et al., 2020).
Among the 14 constituents mentioned above, 11 (all but hydroxybenzoylpaeoniflorin, benzoyloxypaeoniflorin, and lactiflorin) exhibit various biological activities. For example, paeoniflorin (the main active constituent of PRR) shows good anti-inflammatory, immunomodulatory, and anti-tumor effects (Zhao et al., 2021). Albiflorin has the function of soothing the liver and relieving depression (Zhao et al., 2018). Oxypaeoniflorin can prevent acute lung injury caused by lipopolysaccharides in mice (Fan et al., 2021). Benzoylpaeoniflorin exhibits anti-allergic activity, making it a potential candidate drug for the treatment of allergic diseases (Zhong et al., 2021). Galloylpaeoniflorin can relieve osteoporosis following oophorectomy (Liu et al., 2021b). Ellagic acid reduces the toxicity of diclofenac in rat hepatocytes by enhancing the activity of antioxidant enzymes such as catalase (Hatefi-Hesari et al., 2021). 3,3′-Di-O-methylellagic acid significantly reduces retinal vasodilation caused by high glucose levels in juvenile zebrafish (Lee et al., 2018). Methylgallate improves potassium oxazinate-induced kidney damage in mice with hyperuricemia nephropathy by inhibiting the NOD-like receptor thermal protein domain associated protein 3 (NLRP3) pathway, thereby producing a renal protective effect (Liu et al., 2021a). Catechin protects rat cardiomyocytes against hypoxic damage  and also exerts an anti-inflammatory effect (Syed Hussein et al., 2015). Both catechin gallate and epicatechin gallate have significant anti-inflammatory and anti-proliferative activities (Kurbitz et al., 2011). Catechin gallate inhibits catechol methylation in rat hepatocyte cytoplasm and hepatocyte cultures by inhibiting the activity of catechol oxymethyltransferase (Kadowaki et al., 2005).
In this study, we systematically explored the metabolism of PRR and its 14 constituents with four structure types (paeoniflorins, catechins, gallic acids, and ellagic acids) in mice. The objectives of the study were to 1) elucidate the metabolites and metabolic pathways of PRR and its 14 main constituents in mice, 2) determine the possible origins of the metabolites of PRR, and 3) clarify whether these compounds can be converted into the recognized effective forms in vivo. The findings are helpful for further elucidating the forms of PRR that exist in vivo and identifying the effective forms of PRR. The findings are also of great significance for studying the mechanisms of action and quality control of PRR.

Preparation and qualitative analysis of the lyophilized powder of PRR decoction
The lyophilized powder of PRR decoction was prepared as described previously . Each Gram of the lyophilized powder was equivalent to 2.63 g of the crude drug. The constituents in the lyophilized powder of the PRR decoction were identified by high performance liquid chromatography with diode array detector and combined with electrospray ionization ion trap time-of-flight multistage mass spectrometry (HPLC-DAD-ESI-IT-TOF-MS n ) with reference to the mass spectral data of the reference substances or the relevant literature (Supplementary Figure S1).

Animal and metabolic studies
Forty-eight ICR mice (male, 30 ± 2 g) were purchased from the Department of Laboratory Animal Sciences at the Peking University Health Science Center and randomized into one control group and 15 dosing groups (14 compound-dosed groups and one PRR decoction-dosed group) with three mice in each group. All mice were housed in mouse metabolism cages. The experiment lasted for 10 days. After adapting for the first 3 days, the mice were dosed by gavage once per day for the following 7 days. In the 14 compound-dosed groups, the dose was 40 mg/kg mouse weight, while the dose in the decoctiondosed group was 70 mg of lyophilized powder of PRR decoction per kg mouse weight (equivalent to 200 mg/kg of PRR crude drug). All compounds and the lyophilized powder of PRR decoction were suspended in 0.5% sodium carboxymethyl cellulose solution, and the dose volume was approximately 0.2 ml for each mouse. The control group was given the same volume of 0.5% sodium carboxymethyl cellulose solution. The mice were maintained in an environment at 22 ± 2°C (relative humidity 50 ± 5%) and 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 and preparation of samples 2.4.1 Collection of samples
All urine and feces were collected during the 7 days of dosing. One hour after the last dosing by gavage, the blood was collected into 1.5-ml heparin sodium-containing tubes by excising the eyeballs. All samples were stored in a refrigerator at −20°C prior to use.

Preparation of urine, feces, and plasma samples
All urine, feces, and plasma samples from the same group were pooled, resulting in 16 urine, 16 feces, and 16 plasma samples. Each urine sample was centrifuged at 8,000 rpm at 4°C for 15 min. The supernatant was harvested, concentrated, and dried at 55°C followed by the addition of 10 ml of methanol and ultrasonic extraction for 30 min. The extract was filtered, concentrated, and dried at 55°C. Each feces sample was dried at 50°C for 48 h and smashed followed by the addition of 30 ml of methanol and ultrasonic extraction for 30 min per round (three rounds of extraction). The extract was filtered, and the filtrates obtained from the three rounds of extraction were pooled, concentrated, and dried at 55°C Frontiers in Pharmacology frontiersin.org 03 followed by the addition of 10 ml of methanol, ultrasonic extraction for 30 min, and centrifugation at 8,000 rpm and 4°C for 15 min. The supernatant was concentrated and dried at 55°C. The urine and feces samples (0.5 g each) were separately dissolved in 1 ml methanol. Each plasma sample was centrifuged at 5,000 rpm and 4°C for 15 min, and approximately 0.9 ml supernatant was collected for each group. After adding 4.5 ml methanol, the mixture was centrifuged at 5,000 rpm and 4°C for 15 min. The supernatant was blown with nitrogen and dried at 40°C followed by the addition of 0.5 ml methanol for reconstitution. All samples were filtered through a 0.22-μm membrane and stored at −20°C prior to further analysis.

Instruments and Conditions
HPLC-DAD-ESI-IT-TOF-MS n analysis was performed using an HPLC instrument and an IT-TOF mass spectrometer connected to two LC-20AD pumps, an SIL-20AC autosampler, a CTO-20A column heater, an SPD-M20A photo-diode array (PDA) detector, and a CBM-20A system controller (Shimadzu, Kyoto, Japan). Data analysis was performed using LCMS Solution v.3.60, Formula Predictor v.1.2, and Accurate Mass Calculator (Shimadzu, Kyoto, Japan).

Identification of the forms of PRR present in vivo (original constituents and metabolites)
The in vivo existence forms of PRR were identified as previously described . First, the base peak chromatograms (BPCs) of the samples from the dosing and control groups were compared to find the distinguishing peaks and tentatively determine the in vivo existence forms of PRR. Second, the extracted ion chromatograms (EICs) of the compounds in the dosing and control groups were compared to confirm the distinguishing peaks. The chromatographic peaks that appeared in the dosing groups but not in the control group were considered to represent the in vivo existence forms of PRR. Finally, the forms of PRR existing in vivo were analyzed structurally based on the obtained liquid chromatographyhigh resolution multi-stage mass spectrometry data combined with 1) the mass spectrometry data of reference substances, 2) the mass spectrometry fragmentation patterns, 3) the mass spectrometry fragmentation information reported in the literature, and 4) information obtained by searching the SciFinder database.
In this study, the usual neutral losses in mass spectrometry were 30.01 Da (CH 2 85.97 C 10 H 18 O 6 S 265.0733 −6.79 61.87 C 10 H 20 O 6 S 267.0936 0.00 37.17 C 9 H 9 NO 3 178.0516 3.37                   Therefore, based on the mass spectral features and the literature Liang et al., 2013)  were predicted as C 23 H 28 O 11 and C 23 H 28 O 12 , respectively. Their retention time and mass spectral features were consistent with those of the reference substances paeoniflorin and oxypaeoniflorin (see Table 1 and Supplementary Table S1). Thus, M1 and M2 were respectively identified as paeoniflorin and oxypaeoniflorin. As reported previously , M11-M13 are isomers of desbenzoylpaeoniflorin (C 16 H 24 O 10 ) derived from albiflorin. Therefore, we identified M11-M13 as desbenzoyl albiflorin isomers.

FIGURE 1
The 30 metabolites (all new) and proposed metabolic pathways of benzoyloxypaeoniflorin in mice.

Identification of the metabolites of three catechins (CG, ECG, C)
The metabolic pathway of catechin gallate is shown in Figure 2, and the metabolic pathways of other catechins are shown in Supplementary Figure S4 Table 1 and Supplementary Table S1). Therefore, M79 was identified as catechin.
M80-M85 showed [M−H] − at m/z 369.02, and their molecular formula was predicted as C 15 H 14 O 9 S. The fragment ion at m/z 289.07 (C 15 H 13 O 6 ) was formed by a neutral loss of 79.96 Da in the MS 2 spectra of M80-M85. Therefore, based on the mass spectral features and the literature (Gonzalez-Manzano et al., 2009;Rodriguez-Mateos et al., 2014), M80-M85 were predicted to be (epi)catechin sulfate isomers. In addition, in the MS 2 spectra of M81 and M84, the fragment ion at m/z 216.97 (C 7 H 5 O 6 S) of the sulfate conjugate of the characteristic fragment ion at m/z 137.02 (C 7 H 5 O 3 ) formed after the RDA cleavage of the C-ring of catechin. Therefore, we presumed that the sulfate group bound to the hydroxyl group at C-5 or C-7. In the MS 2 spectra of M85, the ions at m/z 289.06 (C 15 H 13 O 6 ) and m/z 179.03 (C 9 H 7 O 4 ) were formed by the sequential losses of 79.96 Da (SO 3 ) and 110.04 Da (C 6 H 6 O 2 ) from the ion at m/z 369.02; however, no neutral loss of 82 Da (H 2 SO 3 ) was observed. We concluded that the sulfate group did not bind to the hydroxyl group at C-3 due to the energy barrier required for bond breakage. No fragment ion was observed at m/z 216.97 (C 7 H 6 O 6 S), and we concluded that the sulfate group did not bind to the hydroxyl group at C-5 or C-7. Hence, we assumed that the sulfate group bound to the hydroxyl group at C-3′ or C-4'.

Identification of methyl catechin gallate (M112)
M112 showed [M−H] − at m/z 455.09, and its molecular formula was predicted as C 23 H 20 O 10 . The fragment ion at m/z 303.08 (C 16 H 15 O 6 ) was formed by a neutral loss of 152.01 Da (C 7 H 4 O 4 ,galloyl) in the MS 2 spectrum, and the signal at m/z 303.08 indicated an additional methyl group compared to catechin (C 15 H 13 O 6 ). Thus, M112 was predicted to having a methylcatechin skeleton. In addition, the fragment ion at m/z 169.01 (C 7 H 5 O 5 , gallic acid group) was observed; thus, its structure was presumed to contain a gallic acid group. Comparison of the molecular formula of M112 (C 23 H 20 O 10 ) with that of the original constituent (catechin gallate, C 22 H 18 O 10 ) suggested that M112 was a methylated product of catechin gallate. According to a previous report, M112 was predicted to be methyl catechin gallate .  Da (C 6 H 8 O 6 ) in the MS 2 spectra. The fragment ion at m/z 313.0573 (C 13 H 13 O 9 ) was observed in the MS 2 spectrum of M129, indicating that the glucuronide group binds to the hydroxyl group at C-5 or C-7, and the methylation reaction occurs at the hydroxyl group at C-3, C-3′, or C-4′. According to our previous study , M127-M129 were predicted to be methyl catechin glucuronide isomers.

Metabolites of two ellagic acid compounds (EA and DEA)
The metabolic pathway of 3,3′-di-O-methylellagic acid is shown in Figure 3, and the metabolic pathway of ellagic acid is shown in Supplementary Figure S5 (Espin et al., 2013), while M139-M141 were predicted as urolithin B sulfate and isomers (Wang et al., 2017a).

Identification of the metabolites of methyl gallate
The metabolic pathway of methyl gallate is shown in Figure 4 Da (SO 3 ) in the MS 2 spectra. Due to its chemical structure, methyl gallate is unlikely to produce the metabolites of these nine sulfate conjugates. Thus, we speculated that the aglycones of M147-M155 may be methyl gallate and methylgallic acid, which are isomers. These two isomers show different relative abundances of the characteristic ions at m/z 168.00 and m/z 124.01; for methyl gallate, the relative abundance of the characteristic ion at m/z 124.01 is higher than that at m/z 168.00, whereas the opposite is true for methylgallic acid. The MS 2 spectra of both M147 and M151 showed higher relative abundances of the characteristic ion at m/z 124.01 compared to that at m/z 168.00; thus, they were predicted to be methyl gallate sulfates. The other metabolites (M148-M150 and M152-M155) were predicted to be methyl gallate sulfate isomers and potentially originated from the metabolic reactions of methyl gallate (e.g., hydroxylation, dehydroxylation, hydrolysis, methylation, and sulfation).

Identification of the metabolites of PRR
A total of 31 PRR metabolites were identified (Supplementary Figure S6), the analyses of 30 metabolites were consistent with the above (except M170).
M170 showed [M−H] − at m/z 187.00, and its molecular formula was predicted as C 7 H 8 O 4 S. The characteristic fragment ion at m/z 107.05 (C 7 H 7 O) was formed by a neutral loss of 79.96 Da (SO 3 ) in its MS 2 spectrum. Based on a previous report (Liu et al., 2020), M170 was tentatively identified as benzyl alcohol sulfate.
In rats, our previous research  identified 27, 27, six, 25, and five PRR metabolites derived from catechins,

FIGURE 4
The 24 metabolites and proposed metabolic pathways of methyl gallate in mice. The red numbers denote new metabolites.  , the administered dosage in male Sprague-Dawley rats was 9.96 g PRR crude drug/kg rat body weight (equivalent to 18.99 g PRR crude drug/kg of mouse body weight). Pre-experiments revealed that the administration of a high dosage of PRR to ICR mice may lead to diarrhea. Therefore, the administered dosage of PRR in this study was 200 mg crude drug/kg mouse body weight, a much lower dosage than that administered in rats. Due to the lower PRR dose used in this study, we did not expect to discover more metabolites of PRR than that previously reported in rats. In addition, significant species differences have been reported in phase I and phase II metabolism (Qin et al., 2021); hence, the 20 new metabolites of PRR identified in mice might be explained by species differences. PRR produces 11 identical metabolites in rats and mice: paeonimetabolin II and its isomers (M4 and M5); C 10 H 18 O 4 glucuronide (M34); methyl dibenzoylpaeoniflorin isomer (M17); 3/4-hydroxy phenylacetic acid sulfate isomers (M175-M176); 3,4-dihydroxy phenylpropionic acid sulfate isomers (M185-M186); 3,4-dihydroxy phenylacetic acid sulfate (M191); 3/4-hydroxy benzoic acid sulfate isomer (M135); and hippuric acid (M65).

Origins of the effective forms of PRR
In a previous study, we found 21 effective forms of PRR that account for its effects of clearing away heat, cooling the blood, and dissipating blood stasis (Xu et al., 2022). In this study, we elucidated some possible origins of the 10 known effective forms of PRR and their isomers: • Desbenzoylpaeoniflorin isomer (C3) can be derived from eight original constituents: paeoniflorin, albiflorin, oxypaeoniflorin, benzoylpaeoniflorin, hydroxybenz oylpaeoniflorin, benzoyloxypaeoniflorin, galloylpae oniflorin, and lactiflorin.
According to our previous study , C 10 H 18 O 2 glucuronides (C9-C15) and C 10 H 14 O 3 glucuronide (C17), two effective forms of PRR, can be derived from paeoniflorin in rats; however, their origins could not be determined in this study. Other effective forms of PRR including 3-hydroxy-4-methoxyphenylpropionic acid sulfate (C7), 3-methoxy-4-hydroxyphenylpropionic acid sulfate (C18), and benzoyl glucuronide (C21) can be derived from catechins in rats , but their specific origins need further investigation.

Insights for the quality control of PRR
Seventeen of the effective forms of PRR are metabolites that are not present in PRR and cannot be used for quality control. This study identified some of the precursors of the 10 effective forms of PRR, which can be used as indicators for the quality control of PRR.

Conclusion
This was the first study on the in vivo metabolism of PRR and its 14 constituents in mice. The metabolites were identified by the HPLC-DAD-ESI-IT-TOF-MS n . In total, we identified 23, 16,24,17,18,30,27,17,22,17,33,3,8,24, and 31 metabolites of paeoniflorin, albiflorin, oxypaeoniflorin, benzoylpaeoniflorin, hydroxybenzoylpaeoniflorin, benzoyloxypaeoniflorin, galloylpaeoniflorin, lactiflorin, epicatechin gallate, catechin gallate, catechin, ellagic acid, 3,3′-di-Omethylellagic acid, methylgallate, and PRR, respectively, in mice. The main metabolic reactions included methylation, hydrogenation, hydrolysis, hydroxylation, glucuronidation, and sulfation. We elucidated the metabolites and metabolic pathways of the 14 constituents of PRR, including paeoniflorins, catechins, ellagic acids, and gallic acids, in mice, and clarified the possible origins of the 10 known effective forms of PRR and their isomers. The findings will facilitate further studies on the effective forms of PRR in vivo and are of great significance for exploring the mechanisms of action and quality control of PRR.

Data availability statement
The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding authors.

Ethics statement
The animal study was reviewed and approved by the Biomedical Ethical Committee of Peking University (approval no. LA2019117).