UPLC-QTOF-MS Based Comparison of Rotundic Acid Metabolic Profiles in Normal and NAFLD Rats

Rotundic acid, the principal bioactive constituent of the herbal remedy “Jiubiying”, has been considered as a candidate compound for treating non-alcoholic fatty liver disease (NAFLD). However, the in vivo and in vitro metabolism of rotundic acid has remained unclear. With the aim of elucidating its metabolic profile, a reliable approach that used ultra-high performance liquid chromatography combined with quadrupole time-of-flight mass spectrometry (UPLC-QTOF-MS) was applied for screening and identifying rotundic acid in vivo (plasma, feces, urine, and liver tissue of normal and NAFLD model rats) and in vitro (rat liver microsomes) metabolites. Herein, 26 metabolites of rotundic acid were identified, including 22 metabolites in normal rats, 20 metabolites in NAFLD model rats, and eight metabolites in rat liver microsomes. Among them, 17 metabolites were identified for the first time. These data illustrate that the pathological status of NAFLD affects the metabolism of rotundic acid. Furthermore, the major pathways of metabolism included phase Ⅰ (demethylation, desaturation, etc.) and phase Ⅱ (sulfation and glucuronidation) reactions, as well as a combined multiple-step metabolism. This work provides important information on the metabolism of rotundic acid and lays the foundation for its future clinical application.


Introduction
Non-alcoholic fatty liver disease (NAFLD), featured by liver macrovesicular steatosis caused by factors other than excessive alcohol use, is a chronic liver disease that affects people worldwide [1,2]. NAFLD is highly correlated with obesity, cardiovascular disease, and insulin resistance, and is a vital predisposing element for cirrhosis and hepatocellular carcinoma pathogenesis [3]. The latest data shows that over 25% of the population suffers from NAFLD globally, and the incidence is gradually growing year after year [4]. Due to its uncertain pathogenesis, no effective drugs have been approved by the Food and Drug Administration (FDA) for treating NAFLD [5]. Hence, it is critical to develop safe and effective drugs to treat NAFLD.
Natural products are an excellent source modern drug development. Between 1981 and 2019, almost 70% of drugs approved by the FDA were natural products or corresponding derivatives [6]. "Jiubiying" is the dry leaf and bark of Ilex rotunda Thumb, which is often used to treat diarrhea, metaphysitis, bruises, colds, fever, and rheumatism [7,8]. Furthermore, it is currently included in the Chinese Pharmacopoeia. Rotundic acid (RA) is the main bioactive component in "Jiubiying" and belongs to the pentacyclic triterpenoids [9]. Increasing reports have demonstrated that Rotundic acid (RA) possesses many pharmacological functions, including, but not limited to, anti-cancer and anti-inflammatory activities [10][11][12][13]. Additionally, it has been revealed that RA can prevent and alleviate hepatic disorders [12,14,15]. Yuan-Man Hsu and co-workers [12] have found that RA has a significant lipid-lowering effect, with mild anti-inflammatory activity in diabetic mice, while reducing liver lipid droplets. Our previous study, which aimed to evaluate • Plasma The plasma samples, which were obtained from individual rats, were mixed in equal amounts into 200 µL, and then treated to precipitate proteins with 600 µL acetonitrile. The procedure included centrifugation (12,000 rpm, 10 min), the collection of supernatants, drying at room temperature by a vacuum concentrator (Eppendorf, Hamburg, Germany), and the reconstitution of the residue in 100 µL diluent containing methanol and water (1:1, v/v).

• Urine
A 200 µL urine sample was precipitated with 400 µL methanol and mixed thoroughly. The other procedure was the same as that of plasma samples.

• Feces
The freeze-dried feces were ground into powder. Next, 4 mL methanol was used to immerse the fecal powder (0.4 g), and the metabolites in the solution underwent a 30-min extraction by ultrasound. After centrifugation (8000 rpm, 10 min), the collected upper layer was dried by evaporation at room temperature. A 200 µL diluent containing a mixture (1:1, v/v) of methanol and water was then employed to reconstitute the residue.

•
Liver tissues One gram of liver sample and 1 mL of 0.9% sodium chloride solution were mixed and thoroughly ground. In order to precipitate proteins in the liver sample mixture, the acetonitrile was added to the liver sample at a ratio of 4:1, then vortexed for 3 min. The rest of the procedure was similar to the method used for feces, except that the volume of reconstituted solution was 300 µL.

RA Metabolism In Vitro
For in vitro metabolism, according to a previous report [24], RA (200 ng/mL, dissolved in methanol) was incubated in rat liver microsomal incubation solution (200 µL) consisting of hepatic microsomes [1.0 mg (protein)/mL], phosphate-buffered saline (0.1 M PBS, pH 7.4) and nicotinamide adenine dinucleotide phosphate (NADPH, 1 mM). The mixture of rat hepatic microsomes, phosphate-buffered saline, and RA was preincubated at 37 • C for 5 min before the addition of 1 mM NADPH. After that, the system was incubated at 37 • C for 30 min to address the reaction. Next, 1 mL ice-cold ethyl acetate solution was added to bring the reaction to an end. The collection and drying of the upper organic layer were conducted by centrifugation (12,000 rpm, 10 min) and evaporation, respectively. Thereafter, 50 µL acetonitrile-water solution (1:1, v/v) was applied to dissolve the residue. The blank control samples were prepared without RA. After centrifugation (12,000 rpm, 10 min), 5 µL supernatant was subjected to analysis. The incubation was conducted in triplicate.
A Waters Q-TOF SYNAPT G2 Spectrometer equipped with ESI source under negative ion mode was utilized to conduct mass spectrometry detection, for which the full scan mode was set at the mass range of 100-1200 Da. The superlative parameters of MS for maximum sensitivity were set: the cone voltage was 30 V and capillary voltage was 3.0 kV; the desolvation temperature was 550 • C and source temperature was 120 • C; the desolvation gas (N 2 ) flow rate was 700 L/h and the cone gas flow rate was 50 L/h. In MS E centroid mode, the MS data were acquired with the low energy function in the trap collision energy (6 eV), and the tandem mass data were acquired with the high energy function in the ramp trap collision energy (20-50 eV). During MS analysis, in order to acquire accurate mass, the Leucine enkephalin was used as the lock mass of m/z 554.2615 ([M-H] − ). The instrument operation and data acquisition were monitored by Masslynx NT 4.1 (Waters, Milford, MA, USA).

Data Analysis
The Metabolynx XS software (Waters, MA, USA) with the mass defect filter (MDF) was utilized to process the metabolism data. The MDF window was ±0.1 Da, 5 ppm was used as the maximum tolerance for mass error, and the spectrum was 2% higher than the relative intensity. The data of pathological and biochemical changes are presented as mean ± standard deviation (SD), and were analyzed by Student's t-test. p < 0.05 was considered as statistically significant.

Establishment of NAFLD Model
The liver pathological sections were evaluated by H&E stain, a "gold standard" for the diagnosis of NAFLD, which provided reliable evidence on the establishment of the NAFLD model. As illustrated in Figure 1A, significant microvesicular steatosis and the inflammatory changes of hepatic lobules were observed in the NAFLD group when compared to the normal group, which is the typical NAFLD feature. This suggested the success of the establishment of the NAFLD rat model in the current study. Similarly, in the oil red staining ( Figure 1B,C), high-fat diet feeding resulted in an elevated area of lipid accumula-tion and aggravated steatosis in the model group, while there was no significant change in the control group. nificant change in the control group.
Matching the pathological outcomes, the concentrations of the hepatic CHOL and TG ( Figure 1D) in the model groups was obviously higher than those in the control groups. Malondialdehyde (MDA) is the main product of lipid peroxidation in the body, while superoxide dismutase (SOD) exerts its function to improve the oxidative stress state of the body and inhibit lipid peroxidation. As indicated in Figures 1E,F, the level of MDA in the NAFLD group was substantially increased, whereas the content of SOD reduced, indicating that lipid metabolism in the body was disordered.  Matching the pathological outcomes, the concentrations of the hepatic CHOL and TG ( Figure 1D) in the model groups was obviously higher than those in the control groups. Malondialdehyde (MDA) is the main product of lipid peroxidation in the body, while superoxide dismutase (SOD) exerts its function to improve the oxidative stress state of the body and inhibit lipid peroxidation. As indicated in Figure 1E,F, the level of MDA in the NAFLD group was substantially increased, whereas the content of SOD reduced, indicating that lipid metabolism in the body was disordered.

The Characteristic Fragmentation of RA
It is confirmed that metabolites and parent compounds share the same splitting properties. Thus, the analysis of the fragmentation characteristics of RA is helpful and crucial to deduce and recognize RA (M0) and its metabolites [17,25]. In this study, the RA standard was assessed under both positive and negative modes of the ESI source. RA gave a higher signal intensity under the negative mode. In addition, RA was eluted at 8.04 min under the analysis conditions, and the deprotonated mass [M-H] − was 487.3420 (C 30 H 47 O 5 − ). The incorporation of the MS 2 fragment information with the previous reports [26][27][28] and the fragment ions of RA observed were mostly constituted by the continuing losses of neutral molecules, involving CO 2, which is 44 Da, CH 4 O, which is 32 Da, H 2 O, which is 18 Da, and HCOOH, which is 46 Da. In its MS/MS fragmentation pattern (Figure 2A properties. Thus, the analysis of the fragmentation characteristics of RA is helpful and crucial to deduce and recognize RA (M0) and its metabolites [17,25]. In this study, the RA standard was assessed under both positive and negative modes of the ESI source. RA gave a higher signal intensity under the negative mode. In addition, RA was eluted at 8.04 min under the analysis conditions, and the deprotonated mass [M-H] − was 487.3420 (C30H47O5 − ).
The incorporation of the MS 2 fragment information with the previous reports [26][27][28] and the fragment ions of RA observed were mostly constituted by the continuing losses of neutral molecules, involving CO2, which is 44 Da, CH4O, which is 32 Da, H2O, which is 18 Da, and HCOOH, which is 46 Da. In its MS/MS fragmentation pattern ( Figure  2A), RA provided ample fragment ions at m/z 469.3310 produced by a reduction of H2O (18 Da) and at m/z 437.3031 constituted through the successive neutral cleavages of H2O and CH4O (32 Da). Furthermore, the daughter product at m/z 455.2479 was formed by the fragmentation of CH4O, m/z 423.3235 was generated by the consecutive losses of H2O and HCOOH, m/z 405.3140 was via the successive eliminations of H2O, HCOOH, and H2O, m/z 393.3111 was obtained by the consecutive cleavages of H2O, CH4O, and CO2, and m/z 391.3871 was formed through the successive losses of H2O, HCOOH, and CH4O. The detailed formation pathway of RA fragmentation ions, which is based on the structural properties and MS/MS fragment ions, is proposed in Figure 2.

Identification of the Metabolites of RA
In this study, compared with blank samples, parent compound RA (M0) and its 26 metabolites were determined both in vivo and in vitro, and were identified by accurate mass, elemental compositions, MS/MS fragment information, and reference literature information. An overview of the characteristics of all metabolites is listed ( Table 1). The total ion chromatograms are shown in Figure S1. The extracted ion chromatograms of the metabolites are presented (Figure 3), and the MS 2 spectra of the metabolites are displayed in Figure S2.   [19,27], M5 could be confirmed as rotundanonic acid. The daughter ion at m/z 405.3729, discovered in the MS/MS spectrum of M6, was consistent with [M-H-2H 2 O-CO 2 ] − . Finally, according to the above results and the literature [19], M5 and M6 were speculated to be the isomers of the dehydrogenation metabolites of RA.

Identification of the Metabolites of RA
In this study, compared with blank samples, parent compound RA (M0) and i metabolites were determined both in vivo and in vitro, and were identified by accu mass, elemental compositions, MS/MS fragment information, and reference litera information. An overview of the characteristics of all metabolites is listed (Table 1) total ion chromatograms are shown in Figure S1. The extracted ion chromatograms o metabolites are presented (Figure 3), and the MS 2 spectra of the metabolites are displ in Figure S2.    [25,28,29], suggesting that sulfate conjugation presented at C-3. Thus, M19 and M20 were tentatively elucidated as the sulfate conjugate product of RA. M21 (t R = 5.94 min) and M22 (t R = 6.21 min) shared the same molecular formula of C 29  , it was speculated that the sulfate conjugation occurred at C-3 based on the previous report [25,28]. Therefore, M23 was proposed as the sulfate conjugate product of . Therefore, M26 was diagnosed as the C-3 glucuronide conjugate product of RA, which is consistent with a previous study [19].
In summary, 26 metabolites were detected and identified, including eight metabolites in vitro, and 26 metabolites in vivo. Among them, compared with previous reports [19], 17 metabolites (except for M1, M5, M6, M10, M11, M15, M16, M20, M26) were detected for the first time. According to the above analyses, the possible metabolic profiles for RA in vivo (normal and NAFLD rats) and in vitro rat liver microsome incubation were proposed ( Figure 4). As shown in Figure 4, RA underwent extensive metabolism including phase I reactions (desaturation, demethylation, reduction, and hydroxylation), phase II reactions (methylation, sulfation, and glucuronidation), as well as multiple-step metabolism, which may explain the low bioavailability of RA. However, the exact structure of the metabolites needs to be confirmed by further study due to the lack of standards. 023, 12, x FOR PEER REVIEW 12 of 15

Discussion
The in vitro metabolism of RA was performed in a rat liver microsome incubation system. Rat liver microsomes are an excellent model in vitro for drug metabolism due to the low cost, high-throughput and high efficiency [29]. The cytochrome P450 enzymes

Discussion
The in vitro metabolism of RA was performed in a rat liver microsome incubation system. Rat liver microsomes are an excellent model in vitro for drug metabolism due to the low cost, high-throughput and high efficiency [29]. The cytochrome P450 enzymes (Phase I reactions) markedly expressed in rat liver microsome offer predictive value for in vivo drug metabolism. In the current study, eight metabolites of RA (M1, M6, M10-M13, M15, M17) were detected in rat liver microsomes, all of which were also found in the in vivo metabolism of RA in normal rats. However, this conclusion is limited by the lack of phase II reactions in the liver microsome. This requires further detailed study.
Furthermore, after the oral administration of RA, 22 metabolites were detected in normal rat samples, including 22 in feces, 12 in plasma, 14 in urine, and 13 in the liver. To better understand the influence of NAFLD's pathological status on in vivo RA metabolism, the metabolic profile in NAFLD rats was conducted and compared with that in normal rats. In total, 20 metabolites were detected in NAFLD rat samples, comprising 16 in feces, 9 in plasma, 12 in urine, and 12 in the liver. From these results, obvious differences were observed in normal and NAFLD model rats. Six metabolites (M7, M11, M17-M19, M25) were only detected in the normal rats, while four metabolites (M2-M4, M16) were only detected in NAFLD model rat samples. There were fewer classes of metabolites determined in NAFLD rats, and the reduction of RA only occurred in normal rats. It was reported that NAFLD's physiological status could affect the quantity and function of hepatic drug metabolism enzymes [30]. In this study, the liver suffered some damage in NAFLD rats. Therefore, the divergent metabolism between normal and NAFLD rats was possibly owing to changes in drug metabolism enzymes under the pathological condition. In this study, after the oral administration of RA, most metabolites were detected in feces, which was consistent with Li's report [19], suggesting that feces are the main metabolic clearance way of RA and its metabolites. In addition, according to previous reports [31][32][33], the gut microbiome may change dramatically in NAFLD rats. Thus, the alteration of the gut microbiome in NAFLD rats was proposed as another reason for the metabolic differences. The speculation that the altered liver function and gut microbiome led to the changed metabolic profile of RA was supported by the fact that in the current study, the feces of normal rats contained all metabolites, while the feces of NAFLD rats lacked some metabolites.

Conclusions
In conclusion, a comprehensive metabolic profile of RA in vivo and in vitro was elucidated utilizing UPLC-Q/TOF-MS. Taken together, 26 metabolites were determined, including 22 metabolites in normal rats, 20 metabolites in NAFLD rats, and eight metabolites in vitro. Among them, 17 metabolites were identified for the first time. The major metabolic reactions of RA included demethylation, desaturation, hydroxylation, reduction, sulfation, and glucuronidation. There are differences regarding the metabolite types between the normal and NAFLD model rats, which suggested that the pathological status of NAFLD may affect the RA metabolism. This study offers reliable scientific evidence for a comprehensive understanding of the mechanism of RA regarding efficacy and side effects, which will eventually benefit the clinical application of RA.

Conflicts of Interest:
The authors declare that they have no conflicts of interest.