Introduction

Timosaponin B-II (TB-II), (25S)-26-O-β-D-glucopyranosyl-22-hydroxy-5β-furostane-3β,26-diol-3-O-β-D-glucopyranosyl-(1 → 2)-β-D-galactopyranoside [1, 2] (Fig. 1) was reported as the major furostanol saponin in Rhizoma anemarrhenae. In our previous studies [3, 4], TB-II could be found in rat serum and urine after oral administration of Huangbai-Zhimu herb-pair extract and a Rhizoma anemarrhenae decoction, respectively. According to the theories of serum pharmacochemistry, TB-II was the key efficacy component of Rhizoma anemarrhenae. Recent pharmacological research showed that TB-II could significantly inhibit platelet aggregation and thrombus formation [5, 6], the expression of proinflammatory cytokines [7] and the proliferation of vascular smooth muscle cells [8]. Moreover, TB-II could improve memory and learning deficits and have an anti-dementia effect [9], protect primary neurons against Aβ25–35 induced toxicity [10], potentiate the brain 5-HT and DA nerve systems and act as an antidepressant [11]. However, it was poorly absorbed via the gastrointestinal segment or may have undergone acid-induced degradation following oral administration to rats, with the absolute oral bioavailability being only 1.1 ± 0.3 % [12, 13]. It could be inferred that the metabolites of TB-II might be the authentic effective factors rather than the parent compound.

Fig. 1
figure 1

The chemical structure of TB-II

Full size image

Against this background, an investigation of the metabolism of TB-II is of great significance for the elucidation of its pharmacological mechanisms and even the development of novel drugs from its metabolites. Liu et al. [14, 15] compared the pharmacokinetics of timosaponin B-II in rat plasma after oral administration of Zhimu-Baihe herb-pair, Zhimu extract and free timosaponin B-II; moreover, they studied the metabolites of Anemarrhena asphodeloides saponin extract. Feng et al. [16] also performed the determination of TB-II in rat plasma and carried out a pharmacokinetic study. However, there have been no reports on a comprehensive in-vivo and in-vitro metabolism study of TB-II.

In the present study, a sensitive and specific HPLC-ESI-MSn method was used for systematic analysis and identification of TB-II and its 19 metabolites in biological samples of in-vivo rat experiments and in-vitro incubation experiments. In addition, the structures of its seven metabolites were unambiguously identified by using the prepared standards together with the previously isolated standards of timosaponin B-III (TB-III) and timosaponin A-III (TA-III), while the others were tentatively identified by comparing their HPLC behavior and MSn spectra profile with those of the parent drug (TB-II). The results of this study will be helpful for better understanding the metabolic processes of TB-II and the potential pharmacological application of these metabolites.

Experimental

Materials and Reagents

TB-II, TB-III and TA-III (purities >98 %) were isolated and purified from Rhizoma anemarrhenae. Five artificial steroidal saponins (timosaponin BII-b, timosaponin BII-d, timosaponin BIII-a, timosaponin BIII-b and timosaponin BIII-c) were obtained previously by our laboratory [17, 18]. Their structures (Fig. 2) were confirmed by 1H-NMR, 13C-NMR, 2D NMR and MS and compared with the literature. Pepsin (1:15,000, CAT: 88918B) and pancreatin (USP, CAT: 84278A) were purchased from Adamas Reagent Co., Ltd. (Shanghai, China). Acetonitrile and formic acid were HPLC grade from Fisher Scientific (Tustin, CA, USA). Water was triple distilled using a Milli-Q system (Millipore, Billerica, MA, USA). All other analytical grade reagents were purchased from Sinopharm Chemical Reagent Co., Ltd. (Shanghai, China). Pretreatment of samples by SPE was performed with Waters Oasis HLB cartridges (30 mg, 1 mL) obtained from Waters (Milford, MA, USA).

Fig. 2
figure 2

Structures of eight reference compounds

Full size image

Chromatographic System and Mass Spectrometry Conditions

The chromatographic separations were performed using a reversed-phase column (Inertsil ODS-3, 4.6 mm × 250 mm i.d., 5 μm, GL Science, Tokyo, Japan) connected to an Easy-Guard Kit C18 (4 mm × 2 mm, Grace, USA) guard column, with the column temperature set at 25 °C. The mobile phase consisted of linear gradients of 0.1 % (v/v) ammonia (A) and acetonitrile (B): 0–10 min, 10–27 % B (v/v); 10–18 min, 27–29 % B; 18–40 min, 29–95 % B. The composition was then held at 95 % B for 5 min. Then the gradients were returned to 10 % B and maintained for 10 min for equilibration. The mobile phase flow rate was 1.0 mL·min−1, and the whole analysis time was 55 min.

The LC-MSn analysis was performed on a 1200 Series HPLC instrument (Agilent, Waldbronn, Germany) coupled with a 6300 Series ion trap (IT) (Agilent Technology, Palo Alto, CA, USA) via an electrospray ionization (ESI) interface. The hardware inclued an Agilent 1200 Series LC and the 6300 series ion trap mass spectrometer. The software (version 6.1) included the Trap Control Program for trap control, data acquisition, data analysis, quantitative analysis and the LC control programs. MSn analyses were conducted in negative-ion mode, and the instrument was operated under the following optimized conditions: collision gas, ultra-high-purity helium (He); nebulizing gas, high-purity nitrogen (N2); capillary voltage, 3.5 kV; end plate offset, 500 V; nebulizer, 30 psi; drying gas flow rate, 10 L·mL−1; drying gas temperature, 350 °C. For full-scan MS analysis, the spectra were recorded in the range of m/z 50–1500.

Animals and Drug Administration

Twenty-six male Sprague-Dawley rats (200–220 g) were supplied by Shanghai SLAC Laboratory Animal Co., Ltd. (Shanghai, China). All animal experiments adhered to protocols approved by the Review Committee of Animal Care and Use at the Shanghai Institute of Materia Medica (Shanghai, China). The animals were divided into three groups at random: group A for dosed urine and feces collection (n = 2); group B for dosed blood and tissue withdrawal (n = 20); group C for blank urine and feces, blood and tissue collection (n = 4). Animals were fasted overnight, but with access to water, and administered TB-II by oral gavage at a single dose of 300 mg kg−1.

Sample Preparation

In-vivo Samples

Urine samples were collected from 0 to 24 h after administration and centrifuged at 10,000×g for 10 min. Feces samples were collected from 0 to 24 h after the dose, dried for 48 h at 37 °C and crushed. Blood samples were withdrawn from the abdominal aorta in heparinized tubes at 0.5, 1.0, 1.5, 2.0, 3.0, 4.0, 6.0, 8.0, 10.0 and 12.0 h after administration of TB-II (with two rats at each time point) and centrifuged at 4000×g for 10 min to obtain the plasma. Tissues, including brain, heart, liver, spleen, lung and kidney, were obtained at 2.0, 4.0, 6.0, 8.0, 10.0 and 12.0 h after the treatment of TB-II (with two rats at each time point), then weighed, homogenized with 0.1 M PBS buffer (pH 7.4) at a ratio of 1 g/5 mL and centrifuged at 10,000×g for 20 min to obtain the supernatants. Then 400 μL urine, plasma and tissue homogenates obtained from each time point were combined as the drug-containing plasma, urine and tissue homogenate samples, respectively. All samples were stored at −80 °C until they were analyzed.

The drug-containing sample (600 μL) was loaded onto a SPE cartridge that was preconditioned with 1 mL methanol and 1 mL water subsequently. The cartridge was then washed with 1 mL water and the analytes were eluted with 1 mL methanol, respectively. The 200-mg crushed feces sample was extracted by ultrasonication three times with 5 mL acetonitrile-water (1:1, v/v) for 30 min each time. The extracted solutions, with the different pretreatment methods described above, were all dried under nitrogen gas over a 37 °C water bath. The residues were then dissolved with 200 μL acetonitrile-water (1:1, v/v) and centrifuged at 10,000×g for 10 min at 4 °C. The supernatant was analyzed by LC–MS-MS. Blank samples as controls were prepared with the same method as for the drug-containing samples.

In-vitro Samples

The AGJ and AIJ were prepared as described in the literature [19]. Samples of 30 μL TB-II (20 mg mL−1) were added separately to 3-mL volumes of AGJ and AIJ and incubated in a water-bathing vibrator at 37 °C. After 0, 0.5, 1.0, 2.0, 4.0, 8.0 and 12.0 h in AGJ, and 0, 1.0, 2.0, 4.0, 8.0, 12.0 and 24.0 h in AIJ (with two samples at each time point), reactions were terminated by adding 3 mL acetonitrile. Blank samples were terminated after 12.0 h in AGJ and 24.0 h in AIJ. All sample preparations were in accordance with the procedures described for in-vivo samples.

Results and Discussion

Optimization of MS Ion Modes and LC Mobile Phase Additives

Choosing an appropriate ion mode would be more conducive to detecting and characterizing the analytes from biological samples; both the positive and negative ion modes were thus tested using TB-II. For the positive ion mode, the ion at m/z 903.6 ([M-H2O + H]+) was obtained as the quasi-molecular ion peak in the primary mass spectra (Fig. 3c), which was possibly produced by loss of the hydroxyl at the C-22 position. It was similar to the quasi-molecular ion peak of timosaponin B-III and would be prone to cause confusion if this quasi-molecular ion were used to speculate the molecular formula. Therefore, the negative ion mode was selected as the mass scan ion mode.

Fig. 3
figure 3

ESI(-) MS spectra of TB-II during optimization: a with 0.2 % formic acid as additive in negative ion mode; b with 0.1 % ammonia as additive in negative ion mode; c with 0.2 % formic acid as additive in positive ion mode

Full size image

Furthermore, the additives in mobile phases also needed to be examined. A suitable additive added to the mobile phases could alleviate the peak tailing and increase the ionization effect. As can be seen in Fig. 3a, b, the ions at m/z 955.4 ([M + HCOOH-H]−) and m/z 919.5 ([M-H]−) were simultaneously obtained when 0.2 % formic acid was used as the additive, while only one ion at m/z 919.5 ([M-H]−) was acquired for 0.1 % ammonia. Moreover, the latter abundance of the quasi-molecular ion peak was higher because the stronger ability of ammonia to achieve protons which had a benefit for the formation of [M-H]−. The formation of [M + HCOO]− would impede the potential of the multiple-stage fragmentation technique. Therefore, 0.1 % ammonia was used as the additive.

Metabolism of TB-II In Vivo

In order to identify the metabolites of TB-II, probable structures were first speculated according to the metabolism rule of drugs in vivo. The full-scan mass spectra of the free fractions obtained from dosed rat plasma, urine, feces and six selected tissues were compared with those of blank rat biological samples and the drug standard solution to explore the possible metabolites. Identification and structural elucidation of these metabolites were performed by comparing their changes in molecular masses (ΔM), retention times and mass spectral patterns of product ions with those of the parent drug. Using the negative ion mode, the parent compound and 18 metabolites were detected in rat plasma, urine, feces and six selected tissues. Extracted ion chromatographic peaks of these metabolites are shown in Fig. 4. The location and MSn data along with the retention times of TB-II and its metabolites are summarized in Table 1. Our results indicated that dehydration, deglycosylation, oxidation and E-ring cleavage were the major metabolic pathways of TB-II in vivo.

Fig. 4
figure 4

Extracted ion chromatograms (EIC) of TB-II and its metabolites

Full size image
Table 1 Chromatographic retention time, location and MSn data of timosaponin B-II and its metabolites
Full size table

Characterization of Metabolites in Rat Plasma and Feces

Parent Compound (M0)

Since it has been well proven by LC–MS analysis that the parent compounds and their metabolites always have very similar chemical structures, the structures of metabolites can be rapidly characterized by comparing their product ions with those of the parent drug, even if no standard substance is available for each metabolite [20]. Therefore, in order to identify metabolites of TB-II, it is first necessary to analyze the regular fragmentation patterns of TB-II. The standard TB-II was detected at 11.9 min and gave the precursor ion of [M-H]− at m/z 919, which was the base peak in the ESI mass spectrum (negative ion mode). Additionally, the MS2 spectrum of the ion at m/z 919 gave the product ion at m/z 757 [M-C6H10O5-H]−, the MS3 spectrum of the ion at m/z 757 from the m/z 919 ion yielded the m/z 595 [M-2 × C6H10O5-H]− ion, and the MS4 spectrum of the ion at m/z 595 from the m/z 757 ion produced the m/z 433 [M-3 × C6H10O5-H]− ion (Fig. 5a–d). These characteristic product ions of TB-II were the most important information for the identification of its metabolites [3, 4]. The retention time and MSn spectra of the molecular ion at m/z 919 (M0) were consistent with those of the authentic standard. Therefore, M0 could be confirmed as the unchanged parent drug.

Fig. 5
figure 5

ESI(-) MSn spectra of TB-II: a MS spectrum of the [M-H]- ion at m/z 919; b MS2 spectrum of the ion at m/z 919; c MS3 spectrum of the ion at m/z 757 from the m/z 919 ion; d MS4 spectrum of the ion at m/z 595 from the m/z 757 ion

Full size image

Dehydration Metabolite (M1)

M1 with the retention time of 15.6 min gave the [M-H]− ion at m/z 901, which was 18 Da less than that of the parent drug. There would be one dehydration reaction existing at TB-II. In addition, the characteristic fragment ions of M1 in the MS2, MS3 and MS4 spectrum were at m/z 739 [M-C6H10O5-H]−, 577 [M-2 × C6H10O5-H]− and 415 [M-3 × C6H10O5-H]−, respectively. Compared with the standard compound of TB-III, M1 could be confirmed as TB-III.

Deglycosylated Metabolites (M4, M5, M17 and M19)

M4 and M5 showed the [M-H]− ion at m/z 757 in the MS2 spectrum, which was 162 Da less than that of M0. The product ion at m/z 595 [M-C6H10O5-H]− could be observed in the MS2 spectrum, which consecutively caused the deglycosylated fragment ions at m/z 433 in the MS3 spectrum. According to the fragmentation pathways obtained above, M4 and M5 were tentatively identified as the deglycosylataed metabolites of TB-II. The sites of metabolism might be the branched chain glucose at C-3 or C-26. As for the other two deglycosylated metabolites (M17 and M19), which were consistent with those of the authentic standards, they could be directly identified as TB-II-b and TB-III-b.

Deglycosylated and Dehydration Metabolite (M9)

M9, which was detected at 26.8 min, afforded the [M-H]− ion at m/z 739, which was decreased by 180 Da compared to that of M0 and sequentially fragmented in the same pathway as M0. With the aid of the standard, M9 was identified as timosaponin AIII. It could be inferred that M9 was produced from the C-26 deglycosylated metabolite (M4/M5) by the loss of a molecule of H2O and further formed spirostan saponin through the cyclization of the F ring.

Hydroxylation Metabolites (M2, M3, M6–M8 and M10–M12)

M2 and M3 were eluted at 10.0 min and 10.9 min, respectively, which were both shorter than that of TB-II. The [M-H]− ion at m/z 935 successively yielded the product ions at m/z 773 [M-C6H10O5-H]−, 611 [M-2 × C6H10O5-H]− and 449 [M-3 × C6H10O5-H]− in the MSn spectra of M2 and M3, which were all 16 Da higher than those of the parent drug. The above fragments indicated that M2 and M3 should be the hydroxylation metabolites of timosaponin B-II.

As for the metabolites of M6, M7 and M8, they coexisted in the extracted ion chromatograms of m/z 917, which were eluted at 11.3 min, 13.2 min and 14.1 min, respectively. Series of characteristic product ions [M-C6H10O5-H]− at m/z 755, [M-2 × C6H10O5-H]− at m/z 593 and [M-3 × C6H10O5-H]− at m/z 431 could be found in the MSn spectra. Moreover, these three metabolites were eluted earlier than M1 (TB-III) on chromatographic columns. The above observations provided clear evidence that M6, M7 and M8 might be the hydroxylation metabolites of timosaponin B-III. Similarly, the other three isomers of M10, M11 and M12 (t R = 19.4 min, 20.5 min, 22.5 min, respectively), with the same parent ions at m/z 755, could be temporarily identified as the hydroxylation metabolites of timosaponin A-III.

However, because of the limitation of mass spectra, we can only speculate about the possible structure of the metabolites or the relationship between metabolites and parent compounds. On this occasion, the position of hydroxy groups of M2, M3, M6–M8 and M10–M12 could not be assigned exactly. As reported in the previous papers, C2 and C15 may be prone to hydroxylation, and they were the most possible metabolism sites [14, 21, 22]. However, their further structural determination required other techniques such as 1H and 13C NMR.

Oxidation Metabolite (M18)

Previously, our group obtained a derivative of TB-II by acid hydrolysis and acetylation and determined its structure by NMR [18]. The metabolites of M18 possessed the same chromatographic behaviors and mass spectra characters as those of TB-II-d. Thus, M18 could be identified as TB-II-d, an oxidation metabolite of M19.

E-ring Cleavage Metabolites (M14)

M14 (t R = 16.5 min) presented the [M-H]− ion at m/z 639, which was 262 Da less than that of M1 (TB-III). The product ions at m/z 477 and 459 originated from the successive loss of glucose molecules (162 Da) and a neutral H2O (18 Da). Based on the above data and literature [21, 23], M14 was tentatively identified as the glycosylated isopregnanolone, formed via the cleavage of the E-ring during the metabolism process.

Distribution of TB-II and its Metabolites

After oral administration, samples of rat heart, liver, spleen, lung, kidney and brain were collected and analyzed by the LC-ESI-MSn method. According to the results, the tissue distribution of TB-II and its metabolites showed great differences after oral administration to rats. The parent compound (M0) and its metabolites (M9) were identified in kidney and heart. M0 and its seven metabolites (M4, M5, M9, M10, M11, M12, M12 and M13) were screened in liver. No metabolites were observed in rat lung, spleen and brain in this study. The results indicated that liver, heart and kidney were the major distribution organs of TB-II in rats.

As for M13 (t R = 23.7 min), it gave a [M-H]− ion at m/z 737 and product ions [M-C6H10O5-H]− at m/z 575 and [M-2 × C6H10O5-H]− at m/z 413, which were all 18 Da less than that of M10, M11 and M12 (hydroxylated TA-III). Thus, M13 was tentatively identified as the dehydration metabolite of hydroxylated TA-III, and it was only detected in liver.

Metabolism of TB-II In Vitro

After incubation with AGJ and AIJ, ten compounds (M0, M1, M5, M7, M9–M12, M14, M17) were detected in the AGJ incubation samples, while seven metabolites (M1, M9, M14–M17, M19) were observed in the AIJ incubation samples. M15 and M16 were identified as follows. M15 (t R = 17.6 min) and M16 (t R = 20.0 min) showed the same [M-H]- ions at m/z 595, revealing that these two compounds were isomers. Moreover, the major product ions at 433 [M-H-Glu]−, 415 [M-H-H2O-Glu]− and 289 [M-H-H2O-Glu-C8H16O]− could be detected in MS2 and MS3 spectra. The above fragmentations and retention behavior were consistent with the known TB-III-a and TB-III-c; thus, M15 and M16 were unambiguously identified as TB-III-a and TB-III-c, respectively.

Combining these results, we came to the conclusion that the gastrointestinal tract is the major site of TB-II metabolism. The hydrolysis of TB-II was considered an initial step in the biotransformation, allowing for easier diffusion and absorption in the intestines.

Conclusion

The results described in this work illustrated a specific and sensitive HPLC-ESI-MSn method for the identification of TB-II and its in-vivo and in-vitro metabolites for the first time. Except for M1–M3 and M9, all other compounds, especially the five artificial steroidal saponins, were first reported as the metabolites of TB-II. The proposed metabolic pathways of TB-II in rats were thought to be dehydration, deglycosylation, oxidation and E-ring cleavage (shown in Fig. 6). This study provided systematic and valuable information on the in-vivo and in-vitro metabolism of TB-II, which could be the basis for explaining the pharmacological actions of TB-II and used for further application in preclinical studies. The pharmacological activities of its metabolites will be examined in future investigations.

Fig. 6
figure 6

Structures of TB-II metabolites in rats and possible metabolic pathways for their production: a deglycosylation; b dehydration; c hydroxylation; d E-ring cleavage; e oxidation

Full size image