The Analytical Strategy of “Ion Induction and Deduction Based on Net-Hubs” for the Comprehensive Characterization of Naringenin Metabolites In Vivo and In Vitro Using a UHPLC-Q-Exactive Orbitrap Mass Spectrometer

Naringenin (5,7,4′-trihydroxyflavanone), belonging to the flavanone subclass, is associated with beneficial effects such as anti-oxidation, anticancer, anti-inflammatory, and anti-diabetic effects. Drug metabolism plays an essential role in drug discovery and clinical safety. However, due to the interference of numerous endogenous substances in metabolic samples, the identification and efficient characterization of drug metabolites are difficult. Here, ultra-high-performance liquid chromatography (UHPLC) coupled with high-resolution mass spectrometry was used to obtain mass spectral information of plasma (processed by three methods), urine, feces, liver tissue, and liver microsome samples. Moreover, a novel analytical strategy named “ion induction and deduction” was proposed to systematically screen and identify naringenin metabolites in vivo and in vitro. The analysis strategy was accomplished by the establishment of multiple “net-hubs” and the induction and deduction of fragmentation behavior. Finally, 78 naringenin metabolites were detected and identified from samples of rat plasma, urine, feces, liver tissue, and liver microsomes, of which 67 were detected in vivo and 13 were detected in vitro. Naringenin primarily underwent glucuronidation, sulfation, oxidation, methylation, ring fission, and conversion into phenolic acid and their composite reactions. The current study provides significant help in extracting target information from complex samples and sets the foundation for other pharmacology and toxicology research.


Introduction
Flavonoids, consisting of two aromatic rings with six carbon atoms interconnected by a heterocycle with three carbon atoms (rings A, B, and C), are essential components in the human diet, even though they are not considered nutrients [1]. With the chemical name of 5,7,4 -trihydroxyflavanone, naringenin is widespread in some edible fruits and vegetables, such as Citrus species, tomatoes, and figs. More importantly, evidence from epidemiological, clinical, and preclinical studies has demonstrated that naringenin is responsible for various pharmacological activities such as anti-oxidation [2,3], anticancer [4][5][6], antiinflammatory [7,8], and anti-diabetic [9] effects, cardiovascular protection [10,11], the regulation of immunity [12], the protection of the nervous system [13], and the prevention of diet-induced weight gain [14]. Naringenin may exert therapeutic effects against COVID-19 by inhibiting the main protease of COVID-19, 3CLpro, and by reducing the activity of ACE2 receptors [15].
The drug disposition in the body involves absorption, distribution, metabolism, and excretion (ADME). Drug metabolism is a complex biotransformation process in which drugs are structurally modified to different molecules by diverse enzyme families, such as CYP450s, dehydrogenases, and flavin-containing monooxygenases [16,17]. The primary purpose of metabolism is to clear endogenous and/or exogenous molecules from the body by converting lipophilic chemicals into hydrophilic products to facilitate elimination. The products of metabolism are known as metabolites, either pharmacologically active or inactive. Revealing the metabolic properties of drugs could contribute to deeper understanding of the effectiveness of drugs, changing the in vivo half-life and risk-benefit ratio of a drug, and playing a significant role in drug design. In general, metabolic reactions are classified into 'phase I' and 'phase II' metabolic reactions. As for the former, drugs expose or introduce groups such as -OH, -COOH, and -NH 2 through oxidation, reduction, and hydrolysis reactions. During 'phase II' drug metabolism, which mainly includes sulfated, glucuronidated, and methylated biotransformation reactions, the drugs usually conjugate with a hydrophilic endogenous molecule to facilitate the elimination of drugs from the body smoothly. The liver is the primary organ responsible for drug metabolic processes, while liver microsomes, subcellular fractions derived from the endoplasmic reticulum of hepatic cells, are the dominant liver models in vitro and the default tools for drug discovery and development [18,19]. Some preliminary studies have been carried out on the metabolic process of naringenin and interrelated components. The pharmacokinetic parameters of naringin and naringenin were calculated by single-dose studies and multiple-dose studies, and 12 metabolites detected in liver microsomes were also analyzed [20]. Naringenin chalcone metabolites in rat plasma and urine were identified by LC-MS and NMR [21]. Moreover, naringenin-7-O-glucuronide and narigenin-4 -O-glucuronide, phase II metabolites of naringenin, were reported to perturb macrophage gene expression [22]. However, information obtained on the metabolic process of naringenin in the body is insufficient, suggesting that a comprehensive analysis of the transformation mechanism of naringenin is still needed.
The detection instrumentation is the critical factor in determining the results of metabolite identification. Liquid chromatography coupled with mass spectrometry has played a fundamental role as the predominant platform for metabolism studies ever since the introduction of atmospheric ionization techniques [23]. Subsequent information filtering and structural identification are the most significant procedures in the process of metabolite identification, playing a pivotal role in drug discovery and development [24]. However, partly due to the significant excess of endogenous material, detecting and characterizing drug metabolites in the complex biological matrices are difficult.
In this study, a UHPLC-Q-Exactive Orbitrap mass spectrometer, characterized by a high resolution, high-quality accuracy, wide quality range, and wide dynamic range, was adopted to collect naringenin metabolite data from rat plasma, urine, feces, liver tissue, and liver microsomes. It is worth mentioning that three methods (solid phase extraction (SPE), methanol precipitation, and acetonitrile precipitation) were used to process the plasma samples to obtain more comprehensive metabolite information. Moreover, a novel analytical strategy of "ion induction and deduction" based on the establishment of multiple "net-hubs", an improvement of the previous analysis strategy [25], was proposed to improve the efficiency of metabolite identification. In the analytical strategy, the reaction pathways of metabolism were divided step by step due to the differences in naringenin biochemical modification in the metabolic process. Two major types of reactions ("reaction I" and "reaction II") were first proposed, and then a series of representative "net-hubs" were also established, which further refined the possible metabolic pathway and identified the attribution of fragment ions. Based on this analytical strategy, we distinguished a series of diagnostic product ion (DPI) groups, which were used to screen possible naringenin metabolites by rapidly matching the cleavage behavior of compounds in the samples. Ultimately, this facilitated the efficient capture of target metabolites from numerous en-dogenous metabolites. Furthermore, this strategy could also be deduced for metabolite identification of other similar compounds to increase work efficiency.

Results
In total, 78 metabolites (naringenin included) were detected and characterized, i.e., 31 in plasma, 43 in urine, 11 in feces, 8 in liver tissue, and 13 in liver microsomes. The detected metabolites are listed in Table 1.

The Establishment of an Analytical Strategy
Regarding the complexity of metabolites and the interference from endogenous material, establishing an accurate and efficient data analytical strategy to identify and characterize drug metabolites scientifically in complex organisms is of great importance. Therefore, an analytical strategy of "ion induction and deduction" based on metabolic "net-hubs" was proposed to systematically screen and identify naringenin metabolites in this work ( Figure 1). In the analytical strategy, we comprehensively analyzed and mined the data in three steps. The induction of "DPI group I" of naringenin based on the mass fragmentation behaviors of naringenin in standard samples and metabolic samples was the first step. The second step was searching for the metabolites of naringenin based on the original nucleus of naringenin with the "DPI group I", setting up related "net-hubs", and inducing the "DPI group II". When naringenin was metabolized by "reaction I" (phase II drug metabolism reactions and some phase I drug metabolism reactions such as a part of oxidation reaction), the mass spectral information of such metabolites matched well with that of naringenin. In contrast, the mass spectral information of metabolites produced by "reaction II" (other metabolic reactions, mainly including reduction, hydrolysis reactions, and so on) was not better matched with that of naringenin because their structures of naringenin were destroyed. Therefore, the third step was to construct "net-hubs" of such metabolic reactions and induce the "DPI group III" according to the occurrence characteristics of these reactions and related literature. Finally, all DPI groups were deduced, and various metabolites were screened by comparison with the DPI groups.    their structures of naringenin were destroyed. Therefore, the third step was to construct "net-hubs" of such metabolic reactions and induce the "DPI group III" according to the occurrence characteristics of these reactions and related literature. Finally, all DPI groups were deduced, and various metabolites were screened by comparison with the DPI groups.

The Induction of "DPI Group I"
Naringenin belongs to the class of dihydro flavonoids, so Retro-Diels-Alder (RDA) cleavage and the neutral loss of H 2 O, CO, and CO 2 occur easily in the process of MS fragmentation [26]. Other diagnostic product ions were generated due to the cleavage at different positions of the C-ring. In the ESI-MS/MS spectrum, naringenin gave rise to the and combining relevant references [27,28]. In negative ion mode, through cleavage pathway A, product ions with high abundance at m/z 151 and m/z 119 were observed, which were associated with RDA cleavage. Similarly, according to the other cleavage pathways, product ions at m/z 145, m/z 125, m/z 177, m/z 93, m/z 165, m/z 107, and m/z 161 also appeared. Meanwhile, in positive ion mode, naringenin yielded a series of corresponding fragment ions such as m/z 153, m/z 121, m/z 179, m/z 96, m/z 147, m/z 163, and m/z 111 according to the cleavage pathways A, B, C, and E. The product ion at m/z 231 was also detected due to the cleavage pathway F in positive ion mode. In addition, the neutral loss of H2O (18 Da) and CO (28 Da

The Induction of "DPI Group II" and the Establishment of Related "Net-Hubs"
Four main "net-hubs" were established according to the general laws of metabolic reactions. They were mono-oxidized, glucuronidated, sulfated, and methylated metabolites of naringenin. We found that these metabolites generated some other diagnostic production ions depending on the bound groups.

The Induction of "DPI Group II" and the Establishment of Related "Net-Hubs"
Four main "net-hubs" were established according to the general laws of metabolic reactions. They were mono-oxidized, glucuronidated, sulfated, and methylated metabolites of naringenin. We found that these metabolites generated some other diagnostic production ions depending on the bound groups.

The Induction of "DPI Group III" and the Establishment of Other "Net-Hubs"
"Reaction II" may change the original nucleus structure of naringenin. By combining the literature, we found that apigenin and naringenin chalcone were important metabolites of naringenin and were set as the "net hub". Based on the occurrence characteristics of these reactions and the relevant references, "DPI group III" was determined as m/z 151 ± x (m/z 153 ± x), m/z 177 ± x (m/z 179 ± x), m/z 125 ± x (m/z 127 ± x), m/z 119 ± x (m/z 121 ± x), m/z 145 ± x (m/z 147 ± x), m/z 93 ± x (m/z 95 ± x), m/z 161 ± x (m/z 163 ± x), and m/z 111 ± x (x = molecular mass difference due to the structural change of naringenin). These metabolites were further subject to other metabolic reactions; thus, their fragment ions could be further changed.
Furthermore, research has shown that 70% of ingested flavanones enter the colon where they are degraded by microbiota, principally producing small phenolic acid metabolites, and then they are absorbed into the circulatory system [29][30][31]. Thus, phenolic acid might be another meaningful metabolite of naringenin; thus, phenolic acid metabolites were also "net-hubs". Further, they were identified individually by the existence of carboxyl groups, benzene rings, and so on. The other metabolites were directly classified in the metabolite part of naringenin due to their small amount, so no "net-hubs" could be established. Through the screening and matching the mass spectrometry information of potential metabolites in various biosamples using "DPI group I" and "DPI group II", 40 metabolites of naringenin were unambiguously or tentatively identified.   ] + was generated by a series of neutral losses of H 2 O, CO 2 , CO, and GluA. All of the above-mentioned fragment ions indicated that M22, M23, and M24 were glucuronidated and glucosylated metabolites of naringenin. Furthermore, the discovery of product ions belonging to "DPI group II" at m/z 313 (m/z 151+Glc) and m/z 489 (m/z 151+Glc+ GluA) in negative ion mode and at m/z 315 (m/z 151+Glc) in positive ion mode suggested that the transformation reaction took place on the A-ring. Similarly, according to the relative molecular mass change, the fragment cleavage behavior, and the identification process, M12, M13, and M14 and the other five isomeric metabolites, M25, M26, M27, M28, and M29, were deduced to be sulfated and glucuronidated metabolites of naringenin. In addition, M30 and M31 were identified as the hydroxylated sulfated and glucuronidated metabolites of naringenin in the same way.
The fragment ions belonging to "DPI group I" of M32 and M33 at m/z 151, m/z 177, and m/z 271 in negative ion mode and at m/z 153, m/z 121, and m/z 179 in positive ion mode proved the existence of the nucleus of naringenin. In negative ion mode, the product ion at m/z 301 was attributed to the neutral loss of GluA. According to the difference in the relative molecular mass and molecular formula between M32, M33, and naringenin, M32 and M33 were presumed to be the hydroxylated glucuronidated and methylated metabolites of naringenin. The fragment ion at m/z 303 generated in positive ion mode also corroborated the inference. Therefore, M34 and M35 were believed to be the glucuronidated and methylated metabolites of naringenin. However, the distinction was that the glucuronidation in M34 and M35 might occur on their B-rings based on the fragment ions belonging to "DPI group II" at m/z 337 (m/z 161 + GluA), while the metabolic reaction of M34 and M35 could not be determined due to the limited information available.    ±5 ppm), meaning that they might be isomers of naringenin. In their ESI-MS/MS spectra, the product ions at m/z 227, m/z 151, and m/z 119 in negative ion mode and at m/z 153, m/z 147, and m/z 179 in positive ion mode supported our initial conjecture. However, the information obtained from mass spectrometry was limited. Thus, N22, N23, and N24 were presumed to be isomers of naringenin, which may be present in naringenin chalcone [21].

Identification of Phenolic Acid Metabolites
Combined with relevant literature [29,30] and fragment ions produced by carboxyl groups and benzene rings, 13 phenolic acid metabolites were finally identified.
In the ESI-MS/MS spectra of H1, H2, H3, and H4, the fragment ions at m/z 137 and m/z 121 were generated on account of the neutral loss CO and CO 2 H1, H2, H3, and H4 could be deduced as 3-(4-Hydroxyphenyl) propionic acid or its isomers. Likewise, it was possible to presume the structures of H11 and H12, which were eventually identified as p-hydroxybenzoic acid or its isomers.
Five isomeric metabolites, H6, H7, H8, H9, and H10 were 2 Da less smaller than  H1, H2, H3, and H4. Thus, they were preliminarily deduced to be the dehydrogenated products of H1, H2, H3, and H4 Similar to the identification process of H11 and H12, the fragment ions of H5 at m/z 134 and m/z 160 in negative ion mode and at m/z 136 and m/z 162 in positive ion mode suggested the presence of a hydroxyl group and carboxyl group. Therefore, according to the accurate measurement, chromatographic behavior, cleavage fragmentation, and characteristic fragment ions, H5 was characterized as hippuric acid or its isomer.

Possible Biotransformation Pathways of Naringenin
A total of 78 metabolites were finally identified in the study, 41 were detected by "reaction I" (naringenin included), and 37 were detected by "reaction II", illustrated in Figure 5. The specific information of these metabolites is illustrated in Table 1. In the process of naringenin metabolism, conjugation reactions could occur on naringenin to produce its conjugation products such as glucuronidated metabolites, sulfated metabolites, methylated metabolites, and glucosylated metabolites. Naringenin could also be converted to naringenin chalcone, phloretin, O-desmethylangolensin, and multiple phenolic acid metabolites such as p-hydroxybenzoic acid, hippuric acid, and p-hydroxycinnamic acid by ring fission. In addition, naringenin was found to produce apigenin and other redox metabolites during the metabolic process. Then, naringenin formed multiple metabolic reaction chains through further multi-level metabolic reactions, and eventually, a complex biotransformation network was formed.
converted to naringenin chalcone, phloretin, O-desmethylangolensin, and multiple phenolic acid metabolites such as p-hydroxybenzoic acid, hippuric acid, and p-hydroxycinnamic acid by ring fission. In addition, naringenin was found to produce apigenin and other redox metabolites during the metabolic process. Then, naringenin formed multiple metabolic reaction chains through further multi-level metabolic reactions, and eventually, a complex biotransformation network was formed.

Naringenin Metabolites In Vivo and In Vitro
In the present study, naringenin metabolites in vivo and in vitro were thoroughly investigated using a UHPLC-Q-Exactive mass spectrometer combining a novel analytical strategy of "ion induction and deduction" and multiple sample preparation methods (SPE, methanol precipitation, and acetonitrile precipitation). In total, 78 metabolites were

Naringenin Metabolites In Vivo and In Vitro
In the present study, naringenin metabolites in vivo and in vitro were thoroughly investigated using a UHPLC-Q-Exactive mass spectrometer combining a novel analytical strategy of "ion induction and deduction" and multiple sample preparation methods (SPE, methanol precipitation, and acetonitrile precipitation). In total, 78 metabolites were finally identified; 67 metabolites were detected in vivo, and 13 metabolites were detected in vitro. Their distribution in each sample is illustrated in Figure 6A,B.
At the overall metabolite level, the main biotransformation pathways observed in vivo and in vitro were glucuronidation, sulfation, oxidation, methylation, ring fission, conversion into phenolic acid, and their secondary metabolic metabolism, which promoted the changes of the structure, variation of polarity, and biological properties of naringenin.
In the study of in vivo metabolites, we collected biological samples of plasma, urine, feces, and liver tissue from rats. In Figure 6A,B, it can be seen that most naringenin metabolites could be excreted by urine, implying that a urine sample might be a powerful sample for metabolite identification of naringenin. As shown in Figure 6A, the glucuronidated metabolites of naringenin, sulfated metabolites of naringenin, and phenolic acid metabo-lites were the major metabolites in the plasma and urine samples, while in feces, phenolic acids were mainly detected, meaning that phenolic acids might be the final metabolites of naringenin. It is worth mentioning that although the liver was the most important organ for metabolism in vivo, our study did not obtain large amount of information of metabolites from the liver tissue, and the metabolites obtained were all phenolic acids. It was preliminarily speculated that naringenin was metabolized rapidly in vivo (within 24 h), so other metabolites were not detected in the liver tissue after 24 h of administration.
Molecules 2022, 27, x FOR PEER REVIEW 20 of 26 finally identified; 67 metabolites were detected in vivo, and 13 metabolites were detected in vitro. Their distribution in each sample is illustrated in Figure 6A,B. At the overall metabolite level, the main biotransformation pathways observed in vivo and in vitro were glucuronidation, sulfation, oxidation, methylation, ring fission, conversion into phenolic acid, and their secondary metabolic metabolism, which promoted the changes of the structure, variation of polarity, and biological properties of naringenin.
In the study of in vivo metabolites, we collected biological samples of plasma, urine, feces, and liver tissue from rats. In Figure 6A,B, it can be seen that most naringenin metabolites could be excreted by urine, implying that a urine sample might be a powerful sample for metabolite identification of naringenin. As shown in Figure 6A, the glucuronidated metabolites of naringenin, sulfated metabolites of naringenin, and phenolic acid metabolites were the major metabolites in the plasma and urine samples, while in feces, phenolic acids were mainly detected, meaning that phenolic acids might be the final metabolites of naringenin. It is worth mentioning that although the liver was the most important organ for metabolism in vivo, our study did not obtain large amount of information of metabolites from the liver tissue, and the metabolites obtained were all phenolic acids. It was preliminarily speculated that naringenin was metabolized rapidly in vivo (within 24 h), so other metabolites were not detected in the liver tissue after 24 h of administration.
Liver microsomes were used to conduct metabolic studies in vitro. In liver microsomes, we identified only 14 metabolites. However, apigenin (M18), an important metabolite of naringenin, was identified in liver microsome samples but not in plasma, urine, feces, or liver tissue samples, which also suggested that apigenin quickly underwent other metabolic reactions and then participated in the bodily physiological functions after being transformed.
The prototype of naringenin was only detected in plasma and liver microsomes, meaning that naringenin would not be excreted in prototype form. Apigenin, an edible Liver microsomes were used to conduct metabolic studies in vitro. In liver microsomes, we identified only 14 metabolites. However, apigenin (M18), an important metabolite of naringenin, was identified in liver microsome samples but not in plasma, urine, feces, or liver tissue samples, which also suggested that apigenin quickly underwent other metabolic reactions and then participated in the bodily physiological functions after being transformed.
The prototype of naringenin was only detected in plasma and liver microsomes, meaning that naringenin would not be excreted in prototype form. Apigenin, an edible plant-derived flavonoid, was proved to be one of the important metabolites of naringenin in this study. Certainly, apigenin is widely known for its anti-oxidant, anticancer, anti-inflammatory, anti-apoptotic, and anti-hyperglycemic effects [32,33], similar to the pharmacological activity of naringenin. However, apigenin also has biological properties that distinguish it from naringenin. For example, naringenin exerts its effect through a post-transcriptional mechanism to inhibit cytokine production, while apigenin mainly regulates cytokine production at the transcriptional level [8]. Furthermore, although both apigenin and naringenin alleviated hyperglycemia and hyperlipemia and insulin resistance, apigenin was more effective than naringenin at an equivalent dose [34]. It was also claimed that the inhibitory activity of apigenin against α-glucosidase was higher than that of naringenin [35]. Therefore, the metabolism of naringenin in vivo to produce apigenin may enhance pharmacological activity and synergize with naringenin to have stronger medicinal effects. Simultaneously, other compounds such as phloretin (N1) and O-desmethylangolensin (N5) with superior activity were also found in the metabolic process [36,37]. Furthermore, hippuric acid (M35) was detected in all biological samples in the metabolic pathway of naringenin. Naringenin undergoes ring fission in vivo to produce a series of organic acids, such as p-hydroxybenzoic acid and 3-(4-hydroxyphenyl) propionic acid, and is ultimately converted into hippuric acid and 4 -hydroxyhippuric acid [38].

Comparison of the Different Biological Treatment Methods
The main disadvantage of chromatographic methods is the need for sample preparation, which determines whether the sample can be used in the chromatograph in its original form. According to the accurate mass measurements, fragmentation patterns, diagnostic product ions, and literature reports, 31 metabolites were screened and identified in the plasma samples by using the UHPLC-HRMS method, i.e., eight in samples treated with the SPE solid-phase extraction method, 16 in samples treated with methanol precipitation, and 15 in samples treated with acetonitrile precipitation ( Figure 6C). Among them, six identical metabolites were discovered in samples subjected to methanol precipitation and acetonitrile precipitation, and two identical metabolites were identified in samples subjected to SPE and methanol precipitation. In contrast, only one identical metabolite was detected in samples subjected to SPE and acetonitrile precipitation.
Although SPE can be used to extract and pre-concentrate a wide range of compound classes and is easy to automate to increase the reproducibility of extractions, this sample preparation method is complex and requires extensive organic solvent consumption. It also has some technical problems [39,40]. In this study, there were lower amounts of samples prepared by SPE than samples prepared by methanol precipitation and acetonitrile precipitation, meaning that SPE may not be suitable for in vivo metabolism studies of such substances. Furthermore, the samples obtained by methanol precipitation and acetonitrile precipitation were similar, so one of them could be selected during sample preparation. During the in vivo metabolic analysis of naringenin, the difference between SPE and the other two preparation methods was mainly reflected in glucuronidated metabolites and organic acid metabolites. More phenolic acid metabolites could be detected in samples prepared by SPE, while more glucuronidated metabolites could be detected in samples prepared by methanol precipitation and acetonitrile precipitation. Differences in metabolite quantity might be due to the different separation selectivity of organic solvents, leading to signal peaks with diverse intensity and quantity concerning naringenin metabolites after UHPLC-HRMS detection.

Animals and Drug Administration
Six male SD rats weighing 220 ± 20 g were obtained from Jinan Pengyue Experimental Animals Company (Jinan, China). The rats were randomly divided into two groups: the drug group (n = 3) for test plasma, urine, feces, and liver, and the control group (n = 3) for blank plasma, urine, feces, and liver. The rats in the drug group were given a dose of 255 mg·kg −1 body weight orally, administered for 3 d. A standard saline solution (2 mL) was administered to the rats in the control group every day. Other culture conditions were consistent with our previous report [27]. The animal facilities and protocols complied with the Guide for the Care and Use of Laboratory Animals (USA National Research Council, 1996).

Sample Collection and Preparation
Plasma, urine, and feces samples were taken from rats and treated with SPE cartridges by the previous processing method [27]. After 24 h of the last drug administration, rat liver tissues were removed from dissected rats and quenched in liquid nitrogen, and then they were stored at −80 • C. During further processing, liver tissue (0.5 g) was ground with methanol and centrifuged to get the supernatant. The supernatant was added to the pretreated SPE cartridges [27], and then the same process described above was conducted. Furthermore, plasma samples were processed by two other methods so as to explore the impact of different processes of biological samples on metabolite detection: methanol and acetonitrile were added separately to reach a final concentration of 75%. These samples were precipitated for 30 min and then centrifuged at 3500 rpm for 15 min to obtain the solutions after treatment. Finally, all samples were dried under N 2 at room temperature. The residue was then redissolved in 300 µL of methanol and centrifuged for 15 min (14,000 rpm, 4 • C). The obtained supernatant was used for further instrumental analysis.

Experiment In Vitro
The in vitro metabolism of naringenin was carried out in rat liver microsomes. A reaction mixture containing phosphate buffer (pH 7.4), MgCl 2 (3 mM, final concentration), rat liver microsomes (1 mg·mL −1 , final protein concentration), and naringenin (0.11 mg·mL −1 , final concentration) was prepared and called the dosing solution. The drug-negative solution contained phosphate buffer (pH 7.4), MgCl 2 (3 mM, final concentration), and rat liver microsomes (1 mg·mL −1 , final protein concentration). In the 6-well plate, the dosing solution was added to the drug group, while the control group was given a drug-negative solution. After that, the drug group and the control group were pre-incubated in a water bath at 37 • C for 5 min. The reaction was started by adding NADPH (2.55 mg·mL −1 , final concentration) dissolved in buffer. The incubation continued at 37 • C, and 100 µL of supernatant was taken out after 5, 10, 15, 30, 45, 60, 120, and 240 min. Subsequently, 200 µL cold acetonitrile was added to stop the reaction, followed by centrifugation at 3500 rpm for 15 min. Finally, the supernatant was dried under N 2 at room temperature. The residue was then redissolved in 300 µL of methanol and centrifuged for 15 min (14,000 rpm, 4 • C). The obtained supernatant was used for further LC-MS analysis. via a heated electrospray ionization (HESI) source. Mass spectrometric detection was performed in both positive and negative ion modes. The ion source parameters were as follows: nitrogen (purity ≥ 99.99%) served as the sheath gas and auxiliary gas at a flow rate of 45 and 10 (arbitrary units), respectively; a capillary temperature of 320 • C and spray voltage of 3800/3500 V (+/−) were used. HRMS was acquired at full scan in a mass range of m/z 80-1200 at a resolution of 140,000, while the resolution of dd-MS 2 was 17,500.

Conclusions
In this study, three methods of biological sample preparation were applied to analyze the in vivo and in vitro metabolism of naringenin. It is worth noting that more metabolites could be retained in samples treated with methanol precipitation and acetonitrile precipitation. Although SPE may not apply to all compounds, it could still be a crucial complement to sample preparation. Our study has demonstrated that compounds obtained by SPE were different from those obtained by other methods. In addition, a UHPLC-Q-Exactive Orbitrap mass spectrometer, which has high selectivity, specificity, and sensitivity [41], was used to investigate the in vitro and in vivo metabolic profiles of naringenin, and an analytical strategy of "ion induction and deduction" based on metabolic "net-hubs" was performed. Due to the difference in metabolic reactions, we divided the metabolic reactions into two categories: in the first, the mass spectral information of such metabolites matched well with that of naringenin, but this did not occur in the second. Therefore, we constructed multiple "net-hubs" and three DPI groups according to the different characteristics of these reactions. Then, the mass spectral information of samples was quickly compared with that of the DPI groups to rapidly screen possible naringenin metabolites.
Finally, 78 naringenin metabolites were identified from plasma, urine, feces, liver tissue, and liver microsome samples. The main biotransformation pathways observed were glucuronidation, sulfation, oxidation, methylation, and so on. Furthermore, we found that naringenin could undergo ring fission to generate naringenin chalcone, phloretin, and various aromatic acids such as 3-(4-hydroxyphenyl) propionic acid and hippuric acid. Metabolites such as apigenin can play a synergistic effect in the body, helping to exert stronger biological activity. Our results supply valuable data for a better understanding of naringenin and provide ideas for the analysis of the metabolites of other natural compounds.  Institutional Review Board Statement: The animal study protocol was approved by the Institutional Review Board (or Ethics Committee) of Binzhou Medical University (2021-083).

Informed Consent Statement: Not applicable.
Data Availability Statement: Most of the data used during the preparation of the manuscript are included in the Results and Discussion sections. However, for any additional details of the procedures and the original raw files, please contact the corresponding authors.