Integrating metabolic characterization and pharmacokinetics to uncover the active compounds of Salvia miltiorrhiza-Ligusticum chuanxiong for treating cerebral ischemia

2.3.1. Animals

Sprague–Dawley (SD) male rats (220−250 g) were purchased from Jiangsu Qinglongshan Biotechnology Co., Ltd. (Nanjing, China). The Animal Ethics Committee of Yangzhou University gave permission to perform the animal experiment (Approval No. YXYLL-2023-08). These rats were housed for one week under a controlled light-dark cycle. Afterward, the rats were categorized into three groups randomly: the control sample group (n = 3), the brain sample group (n = 15, 3 per time point), and the plasma sample group (n = 3). Rats were administrated via gavage with SM-LC (8.1 g/kg, two-fold of the human clinical dosage) once a day for four continuous days.

2.3.2. Instruments and parameters

The metabolites identification was performed on Waters ACQUITY UHPLC (Waters Corp., Milford, USA) combined with AB SCIEX Triple TOF 4600 mass spectrometer (AB SCIEX Corp., Massachusetts, USA). A Zorbax Eclipse Plus C18 (100 × 2.1 mm, 1.7 µm) column was applied for the chromatographic separation. Column temperature was kept at 35°C. For the characterization of plasma metabolites, a gradient program including solvent A (acetonitrile) and solvent B (aqueous solution containing 0.1% formic acid) was as follows: 0 − 1.5 min, 5% − 6% A; 1.5 − 6.5 mins, 6% − 17% A; 6.5 − 14 mins, 17% − 29% A; 14 − 28 mins, 29% − 95% A; 28 − 30 mins, 95% − 95% A. To identify brain metabolites, the gradient elution program of acetonitrile (solvent A) and water containing 0.1% formic acid (solvent B) was as bellows: 0 − 3.5 mins, 5% − 5% A; 3.5 − 12 mins, 22% − 52% A; 12 − 28 mins, 52% − 95% A; 28 − 30 mins, 95% − 95% A. Injection volume and flow rate were 2 µL and 0.35 mL/min, accordingly. AB SCIEX TripleTOF 4600 was utilized to acquire mass data. The mass parameters included the following: collision energy scope (CES), ±15 V; curtain gas (CUR), 35 psi; mass range 50−1200 Da; collision energy (CE), ±35 V; auxiliary gas (GS2), 50 psi; ion source temperature (TEM), 500°C; nebulizer pressure (GS1), 50 psi; ion spray voltage floating (ISVF), +5000 V/ −4500 V.

2.3.3. Sample collection

Blood samples from ophthalmic venous plexus were collected at 0, 0.25, 0.5, 1, 2, 4, 6, 8, and 12 hrs after the final administration. To prepare plasma samples, blood kept in anticoagulant-treated centrifuge tubes was centrifuged at 4000 rpm (4°C) for 15 mins.

After the final administration of SM-LC, brain samples were obtained at multiple time points (0, 0.5, 1, 2, and 4 hrs). Then, the brain was rapidly rinsed using physiological saline to wash the blood, and absorbent paper was used to dry the samples. These plasma and brain samples were kept at −80°C till the UHPLC-QTOF-MS analysis was done.

2.3.4. Biological sample treatment

To minimize individual variability, plasma samples (50 µL each) collected at various time points were combined into a single sample for the characterization of parent compounds and metabolites. Using three times the volume of methanol, 300 µL of the mixed plasma samples were used to precipitate proteins. The processed samples were vortex-mixed for 3 mins, and centrifuged at 13,000 rpm/min for 15 mins (4°C), the supernatant was collected and dried using nitrogen-blow. The residues were re-dissolved in 120 µL of water−acetonitrile (70:30, v/v). After centrifugation at 13,000 rpm/min for 15 mins (4°C), 100 µL of the supernatant was obtained for the identification of plasma metabolites.

To prepare the brain homogenate, the pre-cooled physiological saline (1:5, w/v) was added to the brain samples. Subsequently, 300 µL of the mixed brain homogenate was mixed with 900 µL of methanol. The mixed brain homogenate was vortex-mixed for 3 mins and centrifugated at 13,000 rpm/min for 15 mins (4°C). The supernatant was evaporated to dryness using nitrogen gas. The residues were redissolved in 200 µL of acetonitrile−water (30:70, v/v). The supernatant was acquired for identification analysis after centrifugation for 15 mins (4°C) at 13,000 rpm/min.

2.3.5. Analytical strategy

The metabolic profiling strategy has been shown in Figure 1. The detailed identification process was composed of four parts. First, according to the set threshold of IDA Explorer (PeakView Software Version, 1.2), the accurate mass and retention time of different samples (blank control plasma and brain, drug-containing plasma as well as drug-containing brain) were exported into the Excel file. To acquire as many ions as possible, the intensities of detected peaks above 1000 were chosen for further characterization. The quality scores of compounds should be higher than 20, and the retention time was set as 1−30 mins. Secondly, a VBA program embedded in Excel was established to eliminate the common ions retrieved from both blank control and drug-containing sample. Thirdly, the potential exogenous ions were further identified by the MMDF technique. The detailed MMDF information based on the structure templates for phenolic acids, phthalides, and tanshinones will be elucidated in the following sections. At last, the comprehensive strategy of VBA-based background removal and MMDF technique was applied for the in vivo metabolic profiling of SM-LC. Compared to the published reports [13,14], the established analytical strategy had the following advantages: (1) the xenobiotic ions could be extracted from the massive mass data using a VBA-based background removal technique; (2) the xenobiotic ions could be further classified into the phenolic acids-, phthalides-, and tanshinones-related metabolites under the assistance of MMDF strategy; (3) the phenolic acids-, phthalides-, and tanshinones-related metabolites could be quickly identified by database querying.

Close

Figure 1.

The schematic diagram of the metabolome profiling strategy.

Export to PPT

2.4.2. Administration and sample collection

Sprague Dawley (SD) rats were randomly categorized into the following groups: LC (n = 8), SM (n = 8), SM-LC (n = 8). LC, SM, and SM-LC groups were administered intragastrically with 2.7 g/kg of LC, 5.4 g/kg of SM, and 8.1 g/kg of SM-LC extract, respectively.

Blood from the venous vein was collected at 0, 0.083, 0.25, 0.33, 0.5, 0.75, 1, 2, 3, 4, 6, 9, 12, and 24 hrs. About 0.35 mL of blood sample was obtained and kept in 1.5 mL centrifuge tubes. The blood samples were centrifuged at 4000 rpm/min (4°C) for 15 mins, and 100 µL of supernatant was added to a new tube. All obtained samples were kept at −80°C until the pharmacokinetic analysis.

2.4.3. Plasma sample preparation

Plasma samples were thawed at room temperature. 50 µL of IS solution containing phenformin (1.02 µg/mL) and ketoprofen (2.06 μg/mL), 100 µL of plasma, and 300 µL of methanol were vortexed for 2 mins. Subsequently, the mixed samples were centrifugated at 13,000 rpm/min for 15 mins (4°C), and the supernatant was dried under nitrogen gas. Residues were dissolved in 100 µL of water−acetonitrile (70:30, v/v), vortex-mixed for 3 mins, and centrifuged for another 10 mins (13,000 rpm/min, 4°C). The samples were then determined by UHPLC-QQQ-MS.

2.4.4. Preparation of standards solutions and QC samples

Precisely weighted phenformin (IS), ketoprofen (IS), salvianolic acid B, chlorogenic acid, danshensu, 3-n-butylphathlide, rosmarinic acid, tanshinone IIA, caffeic acid, salvianolic acid A, senkyunolide A, ferulic acid, lithospermic acid, ligustilide, senkyunolide I, crytotanshinone, butylidenephthalide, and dihydrotanshinone I were dissolved in methanol as standard stock solutions. A series of different concentrations of standard working solutions (except for IS) were mixed and processed by diluting the stock standard solutions with methanol. Certain amounts of the IS stock solutions were prepared using methanol to acquire the IS working solution (9.0 µg/mL of phenformin, 7.2 µg/mL of ketoprofen). Four levels (high, medium, low, and lower limit of quantification (LLOQ)) of quality control (QC) samples were prepared using the same steps as the preparation of standard solutions. All samples were stored at −20°C before pharmacokinetic analysis.

2.4.5. Method validation

2.5.1. Preparation of sample solutions

Sixteen prototypes were accurately weighed and dissolved in dimethyl sulfoxide (DMSO) as the stock solutions. The stock solutions were diluted to various concentrations of working solutions with serum-free medium (containing 0.1% DMSO).

2.5.2. Construction of nerve cell injury model and drug administration

Neurons, microglia, and cerebral vascular endothelial cells are the main types of brain cells. Therefore, HT22, BEND.3, and BV2 cell-injury models were used to evaluate the neuroprotective effects of 16 prototypes. Logarithmic phase HT22 cells (2 × 10⁵ cells/well), BEND.3 cells (3 × 10⁵ cells/well), and BV2 cells (2 × 10⁵ cells/well) were seeded into 96-well plates for 24 hrs. After three washes in Earle’s balanced salt solution, HT22, BEND.3, and BV2 cells were subjected to OGD conditions (94% N₂, 1% O₂, and 5% CO₂) for 6, 7, and 2 hrs, respectively. Then, these cells were reoxygenated with a complete medium containing five concentrations of the chemicals for 24 hrs. Cells cultured in a complete medium without the OGD/R process served as the control group. The cell viability of all groups was measured using a CCK-8 solution according to the manufacturer’s instructions.

3.4.1. Characterization of phenolic acids related metabolites

Our previous work identified 46 phenolic acids from SM-LC in the negative electrospray Ionization (ESI) mode, the phenolic acids could be categorized into polymers (salvianolic acid B and lithospermic acid, etc.) and monomers (danshensu and caffeic acid, etc.) [23]. According to the potential metabolic reactions of the 46 identified phenolic acids [14], these phenolic acids could yield 1001 phenolic acids-related metabolites. The theoretical molecular weight of these metabolites was divided into both decimal and integral parts, which were applied for the establishment of the MMDF screening window of phenolic acids (Y range 3.3 − 276.3 mDa; X range 65 − 950 Da). Diagnostic fragment ions filtering (DFIF) and neutral loss filtering (NLF) can help further characterize candidate ions that fall within the MMDF screening window. For DFIF, the product ions at m/z 179 and 197 could be used as the typical diagnostic fragment ions for phenolic acids [4]. For NLF, due to the carboxyl and hydroxyl groups, the neutral loss of CO (28 Da), H₂O (18 Da), and CO₂ (44 Da) could be found in their MS/MS spectra [24]. As a result, 12 prototypes and 34 related metabolites in plasma, as well as 2 prototypes and 3 related metabolites in the brain were identified.

Prototypes A13, A12, A11, A10, A5, A4, and A1 were characterized as salvianolic acid A, salvianolic acid B, lithospermic acid, rosmarinic acid, ferulic acid, caffeic acid, and danshensu, respectively, by referring to the chemical references. Taken salvianolic acid B (A12, m/z 717.1509) and danshensu (A1, m/z 197.0466) related metabolites as examples, the former was usually metabolized into the latter in vivo. M12-1 and M12-2 showed identical product ions (m/z 493.10, 339.05 and 295.07) as that of A12. Hence, they were salvianolic acid B-related metabolites. M12-1 exhibited [M−H]⁻ ion at m/z 669.1501, and its molecular weight was determined as C₃₆H₃₀O₁₃, which could be generated by losing three molecules of O (48 Da) from A12. M12-1 was then deduced as the deoxidation metabolite of A12. Similarly, M12-2 was tentatively identified as the methylation product of A12. M1-1 (m/z, 373.0788), M1-4 (m/z, 357.0844), and M1-6 (m/z, 267.0886) generated product ions at m/z 197.0, 179.0, and 161.0, respectively, which were representative fragment ions of danshensu, suggesting they were metabolites related to danshensu [19]. Through comparison with the precursor ion of danshensu (m/z 257.0660), the additional molecules C₆H₈O₆ (176 Da), C₆H₈O₅ (160 Da), C₂H₄ (28 Da) and OC₂H₂ (42 Da) units could be detected for M1-1, M1-4, and M1-6. Therefore, M1-1, M1-4, and M1-6 were deduced as the glucuronidation, glucuronidation and deoxidation, and methylation and acetylation products of A1, respectively. M1-2, M1-3, and M1-5 showed the [M−H]⁻ ions at m/z 181.0515 (C₉H₁₀O₄), 165.0561 (C₉H₁₀O₃), and 195.0676 (C₁₀H₁₂O₄), respectively. Their fragmentation ions (135.1, 123.1, and 103.0, respectively) were similar to those of danshensu [22]. Therefore, M1-2, M1-3, and M1-5 were deduced as the deoxidation, di-deoxidation, as well as methylation and deoxidation metabolites of danshensu, respectively. The possible metabolic pathways of phenolic acids in rat plasma and brain have been shown in Figure 4. Glucuronidation, deoxidation, methylation, dehydrogenation, and acetylation were the major metabolic reactions of phenolic acids in MCAO/R rats (Figure 3a).

Close

Figure 4.

The metabolic pathways of phenolic acids in MCAO/R rat biological samples.

Export to PPT

3.4.2. Characterization of phthalides related metabolites

Phthalides were widely found in LC with high concentrations and multiple effects [25]. Fifteen phthalides were identified in the positive ESI mode, and were selected to construct the MMDF window of phthalides, generating 319 possible phthalides-related metabolites [26,27]. The molecular weights of these 319 metabolites were separated into decimal (y) and integer (x) parts, which were plotted on an XY scatter plot, forming the MMDF window (Y range 44.9 – 287.3 mDa; X range 154 – 622 Da). To reduce the false positive results, the consecutive neutral losses (15, 18, 28 and 44 Da) and RT ranges (8.9−25 mins) were applied for the identification of phthalides [27]. The MMDF windows of phthalides were utilized to process the positive MS data. A total of 871 and 267 potential ions from plasma and brain were screened. Then chemical references, combined with DFIF, NLF, and other data processing techniques, assisted in the characterization of 11 phthalides prototypes (11 in plasma and 8 in brain) and 40 metabolites (33 in plasma and 17 in brain).

Compounds A9, A14, A18, A19, A24 were accurately clarified as senkyunolide I, butylidenephthalide, senkyunolide A, butylphthalide, and ligustilide by comparing with chemical standards. Moreover, other five prototypes were presumed to be neocnidilide, senkyunolide P, senkyunolide C, wallichilide, and angelicolide with assistance of the literature [1]. Metabolite M9-3, M13-7, and M17-5 with the retention times at 9.51, 15.95, and 17.46 mins, exerted the same quasi-molecular ions at m/z 191.1059. Their molecular formulae were all deduced as C₁₂H₁₄O₂. The m/z 191.1059 ([M+H]⁺) yielded the fragment ions at m/z 173.10 ([M+H−H₂O]⁺), 145.10 ([M+H−CO−H₂O]⁺), 133.10 ([M+H−OCH₂−CO]⁺), and 115.10 ([M+H−OCH₂−CO−H₂O]⁺). Consulting to the literature [25], M9-3, M13-7, and M17-5 were reasonably identified as the deoxidation and dehydration metabolite of senkyunolide I (8.90 min, m/z 225.1111), hydrogenation of butylidenephthalide (15.75 min, m/z 189.0906) and dehydrogenation of senkyunolide A (18.38 min, m/z 193.1228). M17-1 (m/z 401.1436), M17-2 (m/z 225.1111), and M17-3 (m/z 209.1170) generated the product ions at m/z 193.10, 147.10, and 135.10, respectively, which were the representative product ions of senkyunolide A, indicating they were senkyunolide A related metabolites. Compared with senkyunolide A (m/z 193.1228), the additional C₆H₈O₆ (176 Da) + 2O (32 Da), 2O (32 Da), and O (16 Da) groups could be determined for M17-1, M17-2 and M17-3, accordingly. Hence, M17-1, M17-2 and M17-3 were tentatively identified as the dihydroxylation and glucuronidation, dihydroxylation, hydroxylation products of senkyunolide A [28]. Similarly, other phthalides-related metabolites were identified by comparing their MS data with the relevant literature. The main metabolic process of phthalides in rat brain and plasma included hydrogenation, dehydration, glucuronidation, and deoxidation (Figure 3b and Figure 5).

Close

Figure 5.

The metabolic pathways of phthalides in MCAO/R rat biological samples.

Export to PPT

3.4.3. Characterization of tanshinones related metabolites

A total of 14 tanshinones in SM-LC were characterized in the positive ESI mode. Two MMDF regions of tanshinones (phase II and phase I metabolisms) were used for the data processing of MS spectra, and 395 and 97 candidate ions in plasma and brain were discovered, respectively. Moreover, the typical product ions (m/z 269, 267, 249, 235, and 219 Da) and the typical neutral losses (44 Da, 28 Da, and 15 Da) were also used to discover the tanshinones-related metabolites [29].

By comparison with chemical standards, A31, A27, and A22 were definitely identified as tanshinone IIA, cryptotanshinone, and dihydrotanshinone I, respectively. Other seven tanshinones were presumed through literature review. The 11 tanshinones would contribute to the characterization of 44 metabolites. Among these metabolites, 7 originated from tanshinone IIA. M30-1, M30-5, and M30-6 produced the product ions at m/z 295.10, 277.10, and 249.10, respectively, which were identical to the typical diagnostic ions of tanshinone IIA. These metabolites exerted the [M+H]⁺ ions at m/z 327.1226 (M30-1), 311.1273 (M30-5), and 313.1433 (M30-6), which are 32 Da (+2O), 16 Da (+O), and 18 Da (+H₂O) higher than tanshinone IIA, respectively. M30-1, M30-5, and M30-6 were inferred to be the dihydroxylation, hydroxylation, and internal hydrolysis metabolites of tanshinone IIA, respectively [29]. Similarly, other 4 tanshinone IIA-related metabolites were also identified. In general, a total of 55 tanshinones-related metabolites were identified in the biological samples, including 51 in plasma (11 prototypes and 40 metabolites), 15 in brain (6 prototypes and 9 metabolites). Glucuronidation, hydrolysis, hydroxylation hydrogenation, and decarbonylation were the major metabolic patterns of tanshinones (Figure 3c and Figure 6).

Close

Figure 6.

The metabolic pathways of tanshinones in MCAO/R rat biological samples.

Export to PPT

3.5.1. Specificity and selectivity

Typical MRM chromatograms for the blank control plasma, the blank control plasma added with IS and target analytes, and drug-containing plasma have been shown in Figure 7. No obvious endogenous interference (lower than 20% of the LLOQ) was observed during the pharmacokinetics analysis.

Close

Figure 7.

Typical MRM chromatograms of (I) blank plasma, (II) blank plasma spiked with 16 analytes and (III) drug-containing plasma. A Danshensu; B Caffeic acid; C Phenformin; D Ferulic acid; E Lithospermic acid; F Rosmarinic acid; G Salvianolic acid A; H Salvianolic acid B; I Senkyunolide I; J Senkyunolide A; K: 3-n-butylphathlide; L Dihydrotanshinone I; M Ligustilide; N Ketoprofen; O Butylidenephthalide; P Crytotanshinone; Q Levistilide A; R Tanshinone IIA. MRM: Multiple reaction monitoring.

Export to PPT

3.5.2. Linearity and LLOQ

A weighted least squares regression model using the weight value 1/x² was applied for the development of the calibration curves. The concentrations of calibration standards measured by back calculation were within ±15% of nominal concentrations. As shown in Table S4, the 16 analytes had good linear relationships (r² ≥ 0.9960) within the selected linear ranges, and LLOQs ranged from 1.13 to 6.50 ng/mL. The results suggested that the established quantification method met the sensitivity requirements of pharmacokinetic analysis.

3.5.3. Precision and accuracy

Table S5 indicated the 16 analytes’ precision, intra-day, and inter-day accuracy results at multiple levels. The precision (RSD, %) of all analytes at different concentrations was less than 9.29%, and the accuracy (RE, %) ranged from -10.30% to 8.94%. Therefore, the developed approach exhibited acceptable precision and accuracy (RE < ±15% and RSD < 15%).

3.5.4. Matrix effect and extraction recovery

The matrix effects and extraction recoveries of the 16 analytes at HQC, MQC and LQC have been summarized in Table S6. The average recoveries fell between 84.58% and 109.95% (RSDs ≤ 11.03%), which demonstrated stable extraction recoveries (RSDs ≤ 15%). The matrix effects were in the range of 83.86% − 108.36% (RSDs ≤ 10.45%), and no significant matrix effects (RSDs ≤ 15%) were observed.

3.5.5. Stability

The results of the post-extraction, freeze-thaw, short-term stability and long-term stability at HQC, MQC, and LQC have been summarized in Table S7. All analytes possessed acceptable stability (RSDs ≤ 9.92%, -11.87% ≤ RE ≤ 11.73%).

3.6.1. Pharmacokinetic properties of phenolic acids in MCAO/R rats

Phenolic acids were the major effective compounds in SM-LC with anti-inflammatory and antioxidant activities [4]. Seven phenolic acids (salvianolic acid A, danshensu, ferulic acid, lithospermic acid, salvianolic acid B, caffeic acid, and rosmarinic acid) were quantified in rat plasma. Their T_max ranged from 0.11−0.54 hrs, suggesting the rapid absorption of the phenolic acids in the gastrointestinal tract. Salvianolic acid B exhibited the highest C_max at 2593.46 ng/mL, while lithospermic acid possessed the highest AUC_0−∞ and AUC_0−t, reaching 6226.34 and 6348.79 mg·h·mL⁻¹, respectively. Compared with SM or LC, the combined administration of SM-LC could obviously enhance the C_max of rosmarinic acid, salvianolic acid A, danshensu, lithospermic acid, salvianolic acid B, and ferulic acid (P < 0.05), and increase the MRT_0−t and AUC_0−∞ of danshensu, and lithospermic acid (P < 0.05). However, the T_max of danshensu and ferulic acid was shorter in the SM-LC group than that in the SM or LC groups (P < 0.05). Under the pathological states, the activities of metabolic enzymes might be activated. Salvianolic acid B had the lower AUC in the SM-LC group, probably because salvianolic acid B was quickly metabolized into other phenolic acids (danshensu, caffeic acid, etc.) by metabolic enzymes or gut flora [30]. According to the pharmacokinetic results, the presence of complicated compounds (phthalides, tanshinones, etc.) in SM-LC might affect the elimination and absorption of phenolic acids [14]. Danshensu, salvianolic acid B, and lithospermic acid had significant antioxidant and antithrombotic effects, thereby treating brain ischemia [1].

3.6.2. Pharmacokinetic properties of phthalides in MCAO/R rats

Phthalides are the characteristic compounds in LC known for various activities, such as neuroprotection, anti-platelet aggregation, and anti-apoptotic effects [3]. Ligustilide, senkyunolide A, senkyunolide I, butylidenephthalide, 3-n-butylphathlide, and levistilide A were quantified in rat plasma. They showed similar MRT_0−t (2.68−4.03 h), suggesting slower elimination rates. The AUC and C_max of ligustilide and senkyunolide A were higher than those of other phthalides, owing to the relatively high contents in SM-LC. Compared to the oral administration of SM or LC, the combined administration of SM-LC significantly increased the C_max, t_1/2, AUC_0-t, and AUC_0-∞ of senkyunolide I and butylidenephthalide (P < 0.01). However, it significantly decreased T_max (P < 0.01). Moreover, senkyunolide A and levistilide A showed the higher C_max and lower T_max in the SM-LC group. Overall, phthalides possessed relatively higher C_max, t_1/2 and AUC after the combined administration of SM-LC. These results could be partly explained by the fact that some active compounds in SM (phenolic acids and tanshinones) might expedite the absorption and decelerate the elimination of phthalides through regulating the activities of metabolic enzymes [31]. Ligustilide, senkyunolide I and senkyunolide A exerted the significant roles in treating cerebral ischemia by inhibiting neuronal apoptosis and suppressing platelet aggregation [32].

3.6.3. Pharmacokinetic properties of tanshinones in MCAO/R rats

Tanshinones are the major liposoluble compounds in SM with anti-platelet aggregation, vasodilation, and anti-inflammatory activities [19]. In this study, the contents of tanshinone IIA, crytotanshinone, and dihydrotanshinone I, were measured in rat plasma. Due to the low contents of tanshinones in SM, the C_max and AUC of tanshinones were the lowest compared with other components. Tanshinones showed similar T_max (0.25−0.42 h) and MRT_0-t (2.19−3.60 h), suggesting their rapid absorption in the gastrointestinal tract and quick elimination. The AUC_0−t and AUC_0−∞, and C_max of tanshinone IIA, dihydrotanshinone I, and crytotanshinone were higher in the SM-LC group than those in the SM or LC groups (P < 0.05), while their MRT_0−t was lower after the combined administration of SM and LC (P < 0.01). Some studies have reported that certain coexisting compounds (phenolic acids and phthalides) in SM-LC could increase the in vivo exposure of tanshinones, which was consistent with our results regarding the pharmacokinetic characteristics of tanshinones [33]. Tanshinones have been demonstrated to possess essential roles in the treatment of cerebral ischemia by inhibiting neuro-inflammation and dilating blood vessels [34].