A Comprehensive Screening and Identification of Genistin Metabolites in Rats Based on Multiple Metabolite Templates Combined with UHPLC-HRMS Analysis

Genistin, an isoflavone belonging to the phytoestrogen family, has been reported to possess various therapeutic effects. In the present study, the genistin metabolites in rats were investigated by UHPLC-LTQ-Orbitrap mass spectrometer in both positive and negative ion modes. Firstly, the data sets were obtained based on data-dependent acquisition method and then 10 metabolite templates were established based on the previous reports. Then diagnostic product ions (DPIs) and neutral loss fragments (NLFs) were proposed to efficiently screen and ascertain the major-to-trace genistin metabolites. Meanwhile, the calculated Clog P values were used to identify the positional isomers with different retention times. Consequently, a total of 64 metabolites, including prototype drug, were positively or putatively characterized. Among them, 40 metabolites were found according to the templates of genistin and genistein, which was the same as the previous research method. After using other metabolite templates, 24 metabolites were added. The results demonstrated that genistin mainly underwent methylation, hydrogenation, hydroxylation, glucosylation, glucuronidation, sulfonation, acetylation, ring-cleavage and their composite reactions in vivo biotransformation. In conclusion, the research not only revealed the genistein metabolites and metabolic pathways in vivo comprehensively, but also proposed a method based on multiple metabolite templates to screen and identify metabolites of other natural compounds.


Introduction
Recently, isoflavones have received much attention due to their nutritional and potential health benefits. They are non-steroidal phytoestrogenic and antioxidative polyphenolic molecules, which have the possibilities to protect against hormone-dependent diseases, such as prostate cancer, breast cancer, menopausal symptoms, cardiovascular disease and osteoporosis [1][2][3]. Genistin, as one of the common isoflavones, has a similar structure with phytoestrogens and belongs to the annual plant of Fabaceae family especially Glycine max (L.) MERR. It possesses various therapeutic effects, including anti-inflammatory, anticancer activities, cardioprotective effects, antioxygenation, etc., through pharmacological studies [4][5][6][7]. For example, it was found that genistin could enhance the expression of acetylcholinesterase (AChE) related enzymes and related proteins in cultured rat osteoblasts. Therefore, it was speculated that genistin could improve osteoporosis [8]. Compared with other drugs, genistin has many advantages, such as abundant sources, stable nature, low toxicity and side effects. Although there is no clear evidence that the ingestion of genistin is harmful, genistin (i.e., genistein-7-glucoside) can metabolize to produce genistein after entering the body. The genotoxicity and potential adverse effects (cell growth inhibition, apoptosis, topoisomerase inhibition, and DNA damages) of genistein were reported in vitro as well as in experimental animals (e.g., rats) [9][10][11][12][13]. Genistin produces metabolic products of different biological effects through metabolic process in vivo, and the metabolic level of different organisms is different. However, it is still ambiguous that what the material basis is responsible for the effects of genistin. But it is certain that metabolites play an important role. Therefore, it is necessary to study the metabolism of genistin, which helps to explore its effects on cardiovascular and cerebrovascular diseases, tumor and other diseases. And it is of great significance to understand the biological activity of isoflavones and explore its development and utilization.
Contemporarily, ultra-high performance liquid chromatography (UHPLC) coupled with mass spectrometry (MS) has been widely used in drug metabolism analysis [14][15][16]. UHPLC has a powerful separation capability, which is currently one of the most effective tools for the complex components separation. MS owns the properties of high speed, sensitivity and selectivity, during which the high resolution (HR) and multistage mass spectrometry technology have been comprehensively used for online structural analysis and quantitative detection of known and unknown components. In recent years, with the continuous development of Orbitrap technology, Orbitrap mass spectrometer is widely used in pharmaceutical analysis due to the HR detection through combining with the linear ion trap [17][18][19]. Nowadays, UHPLC-LTQ-Orbitrap mass spectrometer has been widely applied in metabolism studies [20][21][22][23][24].
To date, there are dozens of reports concerned with the techniques of dealing the dataset to extract the metabolite information. These techniques include the adoption of mass defect filters, isotope pattern filters, and background subtraction and so on [25]. Generally, the parent drug is selected as the template to predict metabolites according to the common metabolic pathways. The obvious limitation is that many drug metabolites are yielded through the composite reactions. In the previous studies of the metabolites of genistin and the corresponding aglycone (genistein), the number of the detected metabolites was small [26][27][28][29]. This greatly hinders the discussion of the material basis for the efficacy of genistin after entering the body. The characteristic of our study is that according to the literature reports and the results of pre experiments, the multiple metabolite templates are established to assist with data processing. In combination of diagnostic product ions (DPIs), neutral loss fragments (NLFs) and calculated Clog P values, the potential metabolites of genistin were proposed in this study. As far as we know, it was the first time to comprehensively investigate the metabolism of genistin in vivo.

The Establishment of Analytical Strategy
In this study, we established a comprehensive and effective strategy to discover and identify genistin metabolites by using UHPLC-HRMS. At the beginning, HR-ESI-MS 1 analysis was performed on the HRMS instrument with a resolution of 30,000. Meanwhile ESI-MS n data sets were got both in negative and positive ion modes based on data-dependent acquisition method. Secondly, the metabolite templates were established according to the literature reports and the pre experiment results. Then DPIs and NLFs, which were proposed by the mass fragmentation behaviors of reference standards, were applied to efficiently confirm and identify the genistin metabolites. After that, according to different retention times, positional isomers were distinguished by using the corresponding Clog P values calculated by ChemBioDraw Ultra 14.0 (PerkinElmer, Waltham, MA, USA). Depending on the information obtained above, all metabolites were identified positively or putatively. Finally, grounded on metabolites data and references, the metabolic pathway of genistin was proposed. Through the whole experiment, we had summed up a schematic diagram, as shown in Figure 1, which was an analytical strategy for detecting and identifying genistin metabolites.
Molecules 2018, 23, x 3 of 26 was proposed. Through the whole experiment, we had summed up a schematic diagram, as shown in Figure 1, which was an analytical strategy for detecting and identifying genistin metabolites.

The Establishment of Multiple Metabolite Templates Screening Method
Ten metabolite templates, genistin, genistein, daidzin, daidzein, puerarin, dihydrogenistein, tetrahydrogenistein, dihydrodaidzein, equol, and O-desmethylangolensin, were selected as the screening templates for genistin metabolites. Of course, based on this method, the metabolites selected from different templates would be the same. Combined with the retention time, DPIs, and NLFs, the compounds with the same structure need to be deleted. Based on this efficient method, some uncommon compound metabolites could be much more comprehensively detected.

Establishment of Diagnostic Product Ions Basing on Fragmentation Behaviors of Genistin and Its Homologues
In order to obtain an extensive fragmentation behavior of genistin and the other four homologues (genistein, daidzin, daidzein, and puerarin) (

The Establishment of Multiple Metabolite Templates Screening Method
Ten metabolite templates, genistin, genistein, daidzin, daidzein, puerarin, dihydrogenistein, tetrahydrogenistein, dihydrodaidzein, equol, and O-desmethylangolensin, were selected as the screening templates for genistin metabolites. Of course, based on this method, the metabolites selected from different templates would be the same. Combined with the retention time, DPIs, and NLFs, the compounds with the same structure need to be deleted. Based on this efficient method, some uncommon compound metabolites could be much more comprehensively detected.

Establishment of Diagnostic Product Ions Basing on Fragmentation Behaviors of Genistin and Its Homologues
In order to obtain an extensive fragmentation behavior of genistin and the other four homologues (genistein, daidzin, daidzein, and puerarin) (Figure 2), the mixed standard solution was comprehensively analyzed by using UHPLC-LTQ-Orbitrap MS. It can provide the useful information, such as DPIs and characteristic fragmentation patterns, which were utilized for deducting the structures of related metabolites. Taking genistin as an example, it showed a protonated [M + H] + ion at m/z 433.1110 (C 21  were the same as the ESI-MS 3 ions of genistin, which could be attributed to the fact that genistin is the glucoside of genistein. Meanwhile, daidzin and daidzein were eluted at 3.86 and 6.28 min, respectively. In the MS 3 spectrum of daidzin and MS 2 spectrum of daidzein, the most abundant fragment ion at m/z 199 was produced due to the loss of 2CO. Moreover, several major fragment ions at m/z 137, m/z 227, m/z 237, and m/z 145 were also observed due to the respective loss of C 8 H 6 O, CO, H 2 O, and C 6 H 6 O 2 , which indicated that they owned the similar cracking behavior with genistin [26]. Molecules 2018, 23, x 4 of 26 fragment ion at m/z 199 was produced due to the loss of 2CO. Moreover, several major fragment ions at m/z 137, m/z 227, m/z 237, and m/z 145 were also observed due to the respective loss of C8H6O, CO, H2O, and C6H6O2, which indicated that they owned the similar cracking behavior with genistin [26].  In addition, it was necessary to note that m/z 269 and m/z 268 were produced in the ESI-MS 2 of genistin in negative ion mode under the lower collision energies. It showed the loss of a C6H10O5 (162 Da) moiety and the corresponding pure scission process by losing C6H11O5 (163 Da) moiety, which resulted in the formation of the aglycone and free radical aglycone product ions [30]. Some of the main characteristic ions were generated by rearrangement and scission. For example, m/z 267 (100%), 239 (81%), and 223 (56%) were occurred in the ESI-MS 3 spectrum. The residues in the ESI-MS 3 of genistin were similar to those in the ESI-MS 2 spectrum of genistein as showed in Figure 4, which could facilitate the structural elucidation of the genistin metabolites in vivo. Besides, the fragmentation pathways of daidzin and daidzein were also similar to those of genistin [31].
Unlike the above four reference standards, puerarin, attributed to C-glycoside, possessed its own unique cleavage mode. Taking puerarin in positive ion mode as an example, it yielded the protonated molecule [M + H] + ion at m/z 417.1160 (C21H21O9, −4.89 ppm) in the ESI-MS 1 spectrum. In fragment ion at m/z 199 was produced due to the loss of 2CO. Moreover, several major fragment ions at m/z 137, m/z 227, m/z 237, and m/z 145 were also observed due to the respective loss of C8H6O, CO, H2O, and C6H6O2, which indicated that they owned the similar cracking behavior with genistin [26].  In addition, it was necessary to note that m/z 269 and m/z 268 were produced in the ESI-MS 2 of genistin in negative ion mode under the lower collision energies. It showed the loss of a C6H10O5 (162 Da) moiety and the corresponding pure scission process by losing C6H11O5 (163 Da) moiety, which resulted in the formation of the aglycone and free radical aglycone product ions [30]. Some of the main characteristic ions were generated by rearrangement and scission. For example, m/z 267 (100%), 239 (81%), and 223 (56%) were occurred in the ESI-MS 3 spectrum. The residues in the ESI-MS 3 of genistin were similar to those in the ESI-MS 2 spectrum of genistein as showed in Figure 4, which could facilitate the structural elucidation of the genistin metabolites in vivo. Besides, the fragmentation pathways of daidzin and daidzein were also similar to those of genistin [31].
Unlike the above four reference standards, puerarin, attributed to C-glycoside, possessed its own unique cleavage mode. Taking puerarin in positive ion mode as an example, it yielded the protonated molecule [M + H] + ion at m/z 417.1160 (C21H21O9, −4.89 ppm) in the ESI-MS 1 spectrum. In In addition, it was necessary to note that m/z 269 and m/z 268 were produced in the ESI-MS 2 of genistin in negative ion mode under the lower collision energies. It showed the loss of a C 6 H 10 O 5 (162 Da) moiety and the corresponding pure scission process by losing C 6 H 11 O 5 (163 Da) moiety, which resulted in the formation of the aglycone and free radical aglycone product ions [30]. Some of the main characteristic ions were generated by rearrangement and scission. For example, m/z 267 (100%), 239 (81%), and 223 (56%) were occurred in the ESI-MS 3 spectrum. The residues in the ESI-MS 3 of genistin were similar to those in the ESI-MS 2 spectrum of genistein as showed in Figure 4, which could facilitate the structural elucidation of the genistin metabolites in vivo. Besides, the fragmentation pathways of daidzin and daidzein were also similar to those of genistin [31].  Figure 5 [32,33].   Unlike the above four reference standards, puerarin, attributed to C-glycoside, possessed its own unique cleavage mode. Taking puerarin in positive ion mode as an example, it yielded the protonated molecule [M + H] + ion at m/z 417.1160 (C 21 Figure 5 [32,33]. To facilitate the structural illustration of genistin metabolites in the complex matrix, the DPIs were applied to distinguish genistin metabolites in this study [25]. The homologues possessed the similar fragmentation behaviors, which meant that there would be similar DPIs and regular NLFs for providing adequate evidences for elucidating the metabolites. With the biological reactions in vivo, the DPIs might change accordingly. For instance, in positive ion mode, the aglycone moiety of genistin was easily lost in ESI-MS analysis, and the DPI at m/z 271 [M + H − C 6 H 10 O 5 ] + would be produced in ESI-MS 2 spectrum. If some different kinds of biological reactions occurred on the original drug (not on the aglycone moiety), new DPIs resulted from newly generated metabolites at m/z 271 + X (X: mass weight of substituent groups, such as 14, 16, 80, 162, etc.) giving information about the type of the substituent groups. Besides, the DPI at m/z 153 was generated due to RDA rearrangement which occurred on the 1,4-position of C-ring of genistin in positive ion mode. It would give the information whether bio-reactions occurred on the A-ring or not. To point out, NLFs also provided tremendous help for efficiently elucidating the structures of unknown metabolites. For instance, the successive losses of 28 Da (CO) eliminated in the ESI-MS n spectra of genistin and the NLF of 44 Da (CO 2 ) was only found in negative ion mode of genistin, which could also offer much information to identify these metabolites. Although the metabolites were difficult to identify because of the absence of corresponding reference standards, the DPIs coupled with NLFs were calculated to clarify these complex metabolites.

Identification of Genistin Metabolites in Rats
The total ion chromatograms (TICs) of urine and plasma samples from the rats after oral administration of genistin were obtained by using UHPLC-LTQ-Orbitrap mass spectrometry. After comparing the drug-obtained samples with the corresponding control ones, 64 metabolites were detected in both positive and negative ion modes, among which 39 metabolites in positive ion mode and 43 metabolites in negative ion mode were identified in urine, while six metabolites in positive ion mode and eight metabolites in negative ion mode were identified in plasma. The high-resolution extracted ion chromatograms (HREICs) of related metabolites were showed in Figure 6 and the chromatographic retention times and some other concerned data were summarized in Table 1 , were eluted at 4.64 and 5.01 min, respectively. By comparing the full-scan MS/MS n spectra and retention time with the reference standard, M0 was unequivocally deduced to be the parent drug. The accurate mass weight and major product ions (m/z 268, m/z 269, m/z 223, m/z 224, and m/z 267, etc.) of M12 were coincident with those of M0, indicating that it could be deduced as a genistin isomer.      comparing the drug-obtained samples with the corresponding control ones, 64 metabolites were detected in both positive and negative ion modes, among which 39 metabolites in positive ion mode and 43 metabolites in negative ion mode were identified in urine, while six metabolites in positive ion mode and eight metabolites in negative ion mode were identified in plasma. The high-resolution extracted ion chromatograms (HREICs) of related metabolites were showed in Figure 6 and the chromatographic retention times and some other concerned data were summarized in Table 1.     The DPI at m/z 242 generated by the NLF of CO and the DPI at m/z 152 yielded by RDA rearrangement occurred on positions 1 and 3 further confirmed our conjecture. Due to a compound with a larger Clog P value would have a longer retention time in a reverse phase (RP) chromatographic system [34]. Therefore M1, M24, and M39 were putatively thought as 5-O-methylgenistein (Clog P, 2.09), 4 -O-methylgenistein (Clog P, 2.99), and 7-O-methyl-genistein (Clog P, 2.99).
There was a NLF of 15 Da (CH 3 ) from M13, indicating that a methyl group existed in M13. The subsequent MS 3 of m/z 268 generated the fragment ion at m/z 240 by the NLF of 28 Da. According the limited information, M13 was guessed as the O-methyl product of an isomer of genistein. Different from the above methylated products, the ion of losing 15 Da was not found in the mass spectrum of M3 and M7. As a result, we considered they might be the C- Metabolites M2, M19, and M22, whose elution times were 3.96, 5.83, and 6.38 min, were 80 Da higher than that of genistein in negative ion mode, which indicated the occurrence of sulfonation reaction. In its MS/MS 2 spectrum, the DPI at m/z 269 was detected due to the NLF of 80 Da from the parent ion at m/z 349 and in its MS/MS 3 spectrum, the similar characteristic fragments with genistein were observed, which all indicated that the three metabolites were the sulfonation products of genistein. According to their Clog P values, they were briefly inferred to be 5-O-sulfate-genistein, 4 -O-sulfate-genistein, and 7-O-sulfate-genistein, respectively.
Metabolites M18 and M23 were eluted at 5.82 and 6.53 min, respectively, and produced the same theoretical [M − H] − ion at m/z 285.0393 (C 15 H 9 O 6 , error within ±2.50 ppm) in negative ion mode, which were 16 Da higher than that of genistein. Thus they were supposed to be hydroxylation products of genistein due to the similar fragmentation behaviors. The DPI at m/z 153 (RDA rearrangement occurred on positions 1 and 3 of C-ring) of M23 in positive ion mode, indicating that the hydroxyl group was added to the B-ring, which was deduced as 3 -hydroxygenistein [35]. M18 was provisionally characterized as 8-hydroxygenistein on the basis of the calculated Clog P values [36]. M9, possessed the same fragment ions (m/z 285, 257, 229, and 217) with M23 in negative ion mode, was 80 Da more than that of M23, manifesting the presence of sulfuric group.
M5, M10, and M14, whose retention times are 4.47, 4.85, and 5.11 min, were 162 Da more than that of hydroxylation of genistein, indicating they might be glucosylation products of hydroxylated genistein. The DPI at m/z 285 ([M − H − Glc] − ) was yielded by a NLF of glucose moiety, which identified our hypothesis. Therefore, they were putatively identified as hydroxygenistein-O-glucoside. In the MS 2 , the loss of CH 3 formed the characteristic ion at m/z 272. Then the MS 3 of m/z 272 presented the specific fragment ion at m/z 151 via RDA rearrangement occurred on positions 1 and 3. According to the possible metabolic path, the methyl group was attached to 4 -OH to get the hydroxylation and methylation product of M62.

Proposed Metabolic Pathways of Genistin−
In our study, a total of 64 metabolites (parent drug included) were found in rats after oral administration of genistin. The proposed metabolic pathways of genistin are illustrated in Figure 7. There was a series of metabolic reactions including methylation, hydrogenation, hydroxylation, glucosylation, glucuronidation, sulfonation, acetylation, ring cleavage and their composite reactions in vivo biotransformation. In addition, it should be noted that some cracked ring and rearrangement products were produced, such as M26, M27, M29-M32, and M49. Furthermore some special products were detected which some methyl groups were introduced to carbon rings, such as M3 and M7. in vivo biotransformation. In addition, it should be noted that some cracked ring and rearrangement products were produced, such as M26, M27, M29-M32, and M49. Furthermore some special products were detected which some methyl groups were introduced to carbon rings, such as M3 and M7.

Chemicals and Reagents
Genistin, genistein, daidzin, daidzein, and puerain were commercially provided by Chengdu Must Biotechnology Co., Ltd. (Chengdu, Sichuan, China). These five reference standards with the purity higher than 98% were applicable to UHPLC-LTQ-Orbitrap analysis. Acetonitrile, methanol, and formic acid (HPLC grade) used in the mobile phase were obtained from Fisher Scientific (Fair Lawn, NJ, USA). In addition, the other reagents and solvents met the requirements of analytical experiments, which were from Beijing Chemical Works (Beijing, China). The ultrapure water was

Chemicals and Reagents
Genistin, genistein, daidzin, daidzein, and puerain were commercially provided by Chengdu Must Biotechnology Co., Ltd. (Chengdu, Sichuan, China). These five reference standards with the purity higher than 98% were applicable to UHPLC-LTQ-Orbitrap analysis. Acetonitrile, methanol, and formic acid (HPLC grade) used in the mobile phase were obtained from Fisher Scientific (Fair Lawn, NJ, USA). In addition, the other reagents and solvents met the requirements of analytical experiments, which were from Beijing Chemical Works (Beijing, China). The ultrapure water was derived from Milli-Q Gradient Å 10 water purification system (Millipore, Billerica, MA, USA). Grace Pure TM SPE C18-Low solid-phase extraction (SPE) cartridges (200 mg/3 mL, 59 µm, 70 Å) for solid phase pretreatment of biological samples were supplied by Grace Davison Discovery Science™ (Deerfield, IL, USA).

Animals and Drug Administration
Eight SD rats (male, 200-240 g) were purchased from Beijing Weitong Lihua Experimental Animals Company (Beijing, China) and kept under controlled environmental conditions (temperature: 22-26 • C, relative humidity: 65-75%, day and night alternation time: 12-h light/dark cycle) with free food intake and water consumption. After one week of acclimatization, rats were randomly divided into two groups (n = 4 each): Drug Group and Control Group. The standard of genistin was suspended in 0.5% sodium carboxymethyl cellulose (CMC-Na) solution. The Drug Group was orally administered genistin (350 mg/kg), while Control Group was given equivalent 0.5% CMC-Na solution by oral gavage. Before the experiment, all the rats were fasted for 12 h but free drinking water. All procedures were conducted according to the guidelines of Animal Care and Use Committee of Beijing University of Chinese Medicine and Guide for the Care and Use of Laboratory Animals of the US National Institutes of Health.

Plasma Sample Collection
After oral administration, the blood samples were taken from the suborbital venous plexus of the rats in Drug Group at 0.5, 1, 1.5, 2, and 4 h with blood volume of about 0.5 mL. The obtained blood samples were placed in the anticoagulant EP tubes of heparin sodium. After resting for 10 min, each blood sample was centrifuged for 15 min (3500 rpm, 4 • C) and all the blood supernatants in Drug Group were merged into a collective one to get the test plasma containing the drug. The blank plasma sample was from Control Group and the method of collection was the same as the test samples. The above plasma samples were stored at −80 • C.

Urine Sample Collection
Urine samples from 0 to 24 h of each rat in Drug Group were collected by using separate metabolic cages after administration. Each urine sample was centrifuged for 15 min (3500 rpm, 4 • C) and all the urine supernatants in Drug Group were mixed to obtain urine test sample. The blank urine was from Control Group and the method of collection was the same as the test samples. The above urine samples were stored at −80 • C.
Finally we used the SPE method to pretreat all biological samples, which was a method for precipitation and concentration of protein and solid residues. The SPE cartridges were pre-activated with 5 mL of methanol and 5 mL of deionized water and then 1 mL of plasma and urine samples were added. At last the cartridges were eluted by 5 mL of deionized water and 3 mL of methanol orderly. The methanol eluent was collected and dried under N 2 at room temperature. The residue was then redissolved in 80 µL 10% acetonitrile solution and centrifuged for 30 min (13,500 rpm, 4 • C). The supernatant was used for subsequent analysis.

Instruments and Analytical Conditions
An UHPLC-LTQ-Orbitrap mass spectrometer (Thermo Scientific, Bremen, Germany) coupled with an ESI source was used to identify metabolites. The separation of samples was performed on a Waters ACQUITY BEH C18 column (2.1 mm × 100 mm i.d., 1.7 µm; Waters Corporation, Milford, MA, USA). The mobile phase consisted of 0.1% formic acid aqueous solution (A) and acetonitrile (B), and the linear gradient procedure was described as follows: 0-2 min, 5-20% B; 2-27 min, 20-85% B; 27-30 min, 85% B. The column temperature was set at 30 • C, and the flow rate was 0.3 mL/min with 2 µL of the injection volume.
The optimized operating parameters were set as follows: capillary temperature, 350 • C; electrospray voltage, 3.5 kV; sheath gas, 35 arb; auxiliary gas, 10 arb; and probe heater temperature, 300 • C. The metabolites were detected by full-scan mass analysis from m/z 100-1000 with a resolution of 30,000 under positive ion and negative ion modes. The MS 2 and MS 3 analysis were based on the data dependent scan, and the three most abundant ions from the precursor list were selected for collision induced dissociation (CID). The collision energy was adjusted to 40% of maximum to minimize the total analysis time. The dynamic exclusion (DE) was used to prevent duplication. The repeat count was set at 5 and the dynamic repeat time was 30 s with the dynamic exclusion duration at 60 s. Furthermore, MS n stages of the obtained data sets were obtained by the parent ion list (PIL)-DE dependent acquisition mode.

Data Processing
The collected data sets were recorded and processed by Thermo Xcalibur 2.1 workstation (Thermo Scientific, Bremen, Germany). In order to acquire as many fragment ions as possible, we selected the peaks with intensity over 10,000 under negative ion mode and 40,000 under positive ion mode to identify the metabolites. Based on the accurate mass and considering the potential element compositions and the occurrence of possible reactions, the types and number of the predicted atoms were set as follows:

Conclusions
In the present study, a single oral administration of genistin to SD rats was used to study the metabolic profile of genistin in urine and plasma both in positive and negative ion modes. Different from the previous researches, the multiple metabolite templates were set up to assist in data processing combining with DPIs and calculated Clog P values. First, based on the ten metabolite templates, 101 metabolites including the prototype drug were found and confirmed positively or ambiguously according to the fragmentation patterns, accurate mass measurements, chromatographic retention times and relevant drug biotransformation knowledge. Among them, 47 metabolites were found according to the templates of genistin and genistein, which was the same as the previous research method. After using the other metabolite templates, many more metabolites were added, among which 24 metabolites were detected with the templates of daidzin and daidzein. Taking puerarin as a template, only one metabolite was found in vivo. There were also six metabolites on the basis of the template of dihydrodaidzein. Moreover, 16 metabolites with the template of hydrogenated genistein (including dihydrogenistein and tetrahydrogenistein), three metabolites based on the template of equol and four metabolites with O-desmethylangolensin as the template were identified respectively. After deleting the same metabolites, a total of 64 metabolites (including the parent drug) were left. The metabolite templates and the number of corresponding metabolites were as follows: genistin and genistein (40 metabolites), daidzin and daidzein (11 metabolites), puerarin (one metabolite), dihydrodaidzein (one metabolite), hydrogenated genistein (six metabolites), equol (three metabolites), and O-desmethylangolensin (two metabolites). The results demonstrated that genistin mainly underwent methylation, hydrogenation, hydroxylation, glucuronide conjugation, sulfate conjugation, acetylation, ring-cleavage and their composite reactions in vivo biotransformation. In conclusion, our research not only revealed the metabolites of genistein in vivo roundly, but also proposed an integrated strategy based on multiple metabolite templates to screen and identify the metabolites of natural compounds roundly. Meanwhile, an effective method based on the multiple metabolite templates combined with UHPLC-LTQ-Orbitrap mass spectrometer was established for screening metabolites.