Pharmacokinetic and Lipidomic Assessment of the In Vivo Effects of Parishin A-Isorhynchophylline in Rat Migraine Models

Migraine is a chronic brain disease that leads to periodic neurological attacks. Parishin A and isorhynchophylline (PI) is the active monomer component extracted from the traditional antimigraine Chinese medicinal combination of Gastrodia and Uncaria, respectively. In this study, using high-performance liquid chromatography coupled with tandem mass spectrometry (HPLC-MS/MS) technology, we performed pharmacokinetic and lipidomic study on migraine model rats after administration of PI. For the detection of the compounds in plasma, AB Sciex Triple Quad™ 4500 was applied for quantitative analysis, and the COSMOSIL C18 column (2.1 × 100 mm, 2.6 μm) was used for separation. Isorhynchophylline (ISO: m/z 384.8–241.2) and its main metabolite rhynchophylline (RHY: m/z 384.8–160.2) were simultaneously detected under positive ion modes. Besides, parishin A (PA: m/z 995.1–726.9) and its main metabolite gastrodin (GAS: m/z 331.1–123.0) were simultaneously detected with negative ion modes. For the analysis of endogenous lipid components, Dionex Ultimate 3000 (UHPLC) Thermo Orbitrap Elite was applied for the detection, and the Waters UPLCRBEH C18 column (1.7 μm 100 ∗ 2.1 mm) was used for separation. Chloroform/methanol (2 : 1, v : v) was used for extraction. The results demonstrated that PI exists significant difference in metabolism between single- and coadministration and can regulate lipid levels associated with migraine.


Introduction
Migraine causes severe head pain and other disabling neurological symptoms, approximately 14.7% of the global population suffer from migraines, and curative therapies are limited [1,2]. Although western medicines had achieved certain therapeutic effects, high recurrence rates, toxic side effects, and relatively high costs limit their effectiveness in China [3,4]. e use of traditional Chinese medicine for the treatment of migraines has been intensely explored. Traditional Chinese medicines provide systemic regulation, have limited side effects, and have multiple targets that minimize drug resistance [5,6].
Gastrodia-Uncaria (known as Tian Ma and Gou Teng in Chinese) is a commonly used medicinal combination for the treatment of vascular migraines and liver-yang-hyperactivity migraines [7,8]. In our previous studies, PA and ISO were identified as the major active chemical constituents of Gastrodia and Uncaria and can be absorbed into blood [9]. Besides, further pharmacodynamic assessment of the combination of PA and ISO revealed their antimigraine properties. Because PA and ISO are monomeric components derived from traditional Chinese medicine, the in vivo studies of PI will lay a foundation for the formation of a new antimigraine drug with high efficiency, low toxicity, and controllability.
However, the metabolic process of PA and ISO remains unclear, and the advantages of the combination of PI have not been clarified; besides, the efficacy in animal migraine models has not been clarified. Whilst the pathology of migraines remains largely incomplete, a close relationship to lipid metabolism has been demonstrated as altered lipid levels are frequently reported in migraine patients [10][11][12][13][14].
Pharmacokinetic studies provide important references for drug safety and efficacy [15]. Preexperimental data showed that both PA and ISO are unstable in rats and are readily converted to GAS and RHY, respectively. In this study, we employed pharmacokinetic and lipidomic methods to characterize the in vivo mechanism of PI from the perspective of drug metabolism and endogenous lipid regulation. Negative ion modes were employed to assess PA and GAS, whilst positive ion modes were used for the assessment of ISO and RHY. Lipidomics were employed to systematically analyze lipid metabolism under PI treatment conditions [16][17][18].

Animal
Handling. Healthy Sprague-Dawley (SD) female rats weighing 180 g ± 20 were provided by Hunan SJA Laboratory Animal Co., Ltd. (animal license number: SCXK (Xiang) 2016-0002). Rats underwent adaptive feeding for one week and were fasted for 12 h. Rats had open access to water throughout the study. For pharmacokinetic analysis, 18 SD rats were randomly divided into 3 groups and 0.4 mL of nitroglycerin (dosage: 10 mg/kg) was injected into the forehead to establish the migraine model [19]. After 1 h, rats were intragastrically administered PA alone (372.4 mg/kg), ISO alone (40 mg/kg), and PI coadministered (ISO: 40 mg/ kg, PA: 372.4 mg/kg). For blood sampling, 0.3 mL of blood was taken from the rat orbit prior to drug administration and 0, 5, 15, 30, and 45 min and 1, 2, 4, 6, 8, 10, 12, and 24 h after administration. ese blood samples were collected into EP tubes coated with heparin sodium. After centrifugation at 4000 rmp for 10 min, supernatants were collected and stored at −80°C.
For lipidomics, 24 SD rats were randomly divided into three groups: (1) control; (2) model; and (3) administration groups. For the drug-administration groups, 0.4 mL of nitroglycerin was injected into the rat foreheads, and 1 h later, rats were intragastrically administered PI (ISO: 40 mg/kg, PA: 372.4 mg/kg). Blood was collected from the rat eyelids 1 h and 24 h after administration.
e collected blood samples were centrifuged at 4000 rmp for 10 min, and supernatants were stored at −80°C.

Preparation of Calibration Standards and Quality Control
Samples. For the biological analysis method of simultaneously detecting ISO and RHY, the concentrations of the standard curve was prepared by gradient dilution of the 1 mg/mL mixed storage solution with appropriate amount of acetonitrile solution, followed by adding blank plasma and IS solution to the final concentration of 0.5, 2, 10, 25, 50, 100, and 200 ng/mL. e QC samples included LLOQ (0.5 ng/ mL), LOQ (1 ng/mL), MQC (15 ng/mL), and HQC (150 ng/ mL); for detecting PA and GAS, the final concentrations (5, 10, 50, 100, 500, 1000, and 2000 ng/mL) of the standard curve were prepared by gradient dilution of the 1.2 mg/mL mixed storage solution with appropriate methanol solution, followed by adding blank plasma and IS solution. e QC samples included LLOQ (5 ng/mL), LOQ (15 ng/mL), MQC (150 ng/mL), and HQC (1500 ng/mL).
For lipidomics, QC samples were prepared by mixing 10 μL of each sample, using PC (17 : 0) at a concentration of 30 μg/mL as the IS.

Sample Processing.
For pharmacokinetic studies on RHY and ISO, 50 μL of plasma was transferred into 1.5 mL EP tubes to which 50 μL of pioglitazone hydrochloride (IS: 175 ng/mL) and 250 μL of acetonitrile were added. e mixture was vortexed for 3 min and centrifuged for 10 min at 13000 rpm. 200 μL of supernatants was collected for LC/MS analysis. For GAS and PA assessments, 50 μL of geniposide (IS: 350 ng/mL) and 250 μL of methanol were added to 50 μL of the plasma sample. e mixture was vortexed for 3 min and centrifuged for 10 min at 13000 rpm. 200 μL of supernatants was collected into centrifuge tubes, dried in nitrogen, and dissolved in 200 μL of 5% acetonitrile solution (95 : For lipidomic assessment, 100 μL of serum was accurately absorbed and 300 μL of chloroform/methanol (2 : 1, v : v, containing IS) was added to each sample. Samples were vortexed for 5 min and subjected to ultrasound for 10 min. en, the samples were centrifuged at 12000 rpm at 4°C for 10 min. Supernatants were collected and precipitates were extracted in 2 mL of chloroform/methanol (2 : 1, v : v) twice. All supernatants were dried in nitrogen and dissolved in chloroform/methanol (2 : 1, v : v). 200 μL of supernatants was transferred to sample vials for detection.

Validation Criteria of Bioanalytical Method.
Validation of the pharmacokinetic method was according to the FDA guidelines, including selectivity, linearity, precision, accuracy, matrix effects, extraction recovery, and stability [20].
For the selectivity, the signal-to-noise ratio of the analyte was at least 10. For the linear relationship, the correlation coefficient of the standard curves for each compound was greater than 0.99. For inter-and intraday accuracy (RE) and precision (RSD) of QC, the samples did not exceed 15% and the LLOQ did not exceed 20%. e RSD of mean extraction recoveries, matrix effect, and stability should did not exceed 15%.

Data
Analysis. DAS 3.2.8 software was used to analyze the plasma concentrations in the different groups across all experimental time points [21]. Time-concentration curves were drawn using Sigma Plot 10.0. For lipidomics, raw data were converted to the common (mzData) format using Agilent MassHunter Qualitative Analysis B.08.00 software (Agilent Technologies, USA). In the R software platform, the XCMS program was used to assess peak identification times, retention time correction, and automatic integration pretreatments [22]. Data were subjected to IS and weight normalization. Visualization matrices containing the sample name, m/z-RT pair, and peak area were obtained. After removing the ions with a signal-to-noise ratio greater than 500, the ions with the missing values greater than 80% in each group, and the internal and isotopic ions, a total of 1654 features were acquired in the positive mode and 1194 were obtained in the negative mode. After editing, data matrices were imported into SIMCA-P 13.0 (Umetrics, Umea, Sweden), mean-centered, and scaled to Pareto variance. Multivariate analysis was then conducted. Combining the data in positive and negative ion modes, an overview was created using unsupervised principal component analysis (PCA) and a DModX was applied to remove outliers. Supervised orthogonal partial least squares-discriminant analysis (OPLS-DA) was applied to distinguish the contribution of the detected variables to the discrimination between model and control groups [23]. e R2 and Q2 values were applied to assess the model. Desirable conditions were obtained when the R2 was close to 1. Similarly, a Q2 value larger than 0.5 indicated good predictability. In addition, the OPLS-DA models were validated by CV-ANOVA. Metabolic pathways were enriched on the MetaboAnalyst website (http://www.metabo-analyst.ca/). IBM SPSS statistics 21 software was used and two-tailed independent Student's ttests were performed to identify markers with a VIP greater than 1. Filtered metabolites (VIP > 1 and P < 0.05) were identified as potential biomarkers and matched with the HMDB (http://www.hmdb.ca/), METLIN (https://isometlin. scripps.edu/), MassBank (http://www.massbank.jp/), and ChemSpider (http://www.chemspider.com/). e receiveroperating characteristic (ROC) curve was applied to judge the potential biomarkers in which important biomarkers of migraines had AUC values greater than 0.7 [18].

Optimization of LC-MS/MS Conditions. RHY and ISO
are isomers and so different ion pairs were selected to facilitate both the integration and calculation of the sample peak areas. Liquid chromatography conditions were optimized to meet the baseline separation requirements. As both are alkaloids, favorable intensities for mass spectrometry and improved peak conditions were obtained on the C 18 column. e addition of 0.1% formic acid in the aqueous phase improved the symmetry of the peaks. Acetonitrile promoted sample separation and reduced the column pressure, protecting the ultra high-pressure pumps. Formic acid (0.1%) and acetonitrile were thus used as the mobile phases. PA and GAS have different structures, so appropriate IS was required. GEN has similar physical, chemical, and structural properties and was selected as the IS. GAS has a relatively large polarity and short peak time, meaning that it was necessary to select a suitable chromatographic column and prolong the analysis time. During sample processing, the double peaks of GAS were eliminated using 5% acetonitrile water (5 : 95/v : v) to redissolve the GAS and PA. Under acidic conditions, GAS easily formed additive ions with HCOO − . When formic acid was used as the mobile phase, 331.1-123.0 was selected for ion pair detection. Figure S1 shows the representative multiple reaction monitoring (MRM) chromatograms of the determined compounds at 15 minutes after coadministration in migraine model rats. And Figure S2 displays the representative total ion chromatogram (TIC) of plasma samples in positive and negative ion modes. Both of them show the good results of conditional optimization.

Validation of Bioanalytical Method.
e standard curve ( Figure S3) shows the detected compounds and the chromatogram.
e selective study of the method validation indicated good linear relationship and selectivity. Table S1 shows the intra-and interday precision and accuracy of PA, GAS, ISO, and RHY. Table S2 shows the extraction recovery, matrix effect, and stability (including short-term, long-term, and repeated freeze-thaw conditions) of each compound.
e results indicated that the intra-and interday accuracy of each compound was between 0.08% and 7.08%, and the precision ranged from 1.50% to 8.39%, indicating high accuracy and precision. e extraction recovery was between 86.18% and 104.12%, and the matrix effect was between 85.82% and 106.83%, indicating good recovery and no effects of the matrix. e RSD of the stability was between 1.04% and 11.36%, which satisfied the analytical requirements.
us, the conditions met the requirements for PI analysis in the rat plasma. Figure 1 shows the plasma concentration-time curves following single-and coadministration of each compound. e coadministration of ISO, RHY, and GSA was above those of the single-administration groups, whilst PA was almost undetected in either group. Both PA and ISO are unstable in plasma, PA is easily metabolized to GAS, and ISO is easily converted to RHY, resulting in a larger error in the concentration-time curve of the compounds. Table 1 shows the pharmacokinetic parameters of each compound. Following coadministration, the AUC of ISO was 1.31-fold higher than the administration-alone group, whilst RHY was 1.44-fold higher. In addition, the C-max and T-max of both ISO and RHY were larger than those of the single-administration group, suggesting that coadministration promoted the absorption of ISO and RHY into the blood. e half-life (t1/2z) of ISO through coadministration was 1.99-fold higher than the administration-alone group, whilst RHY was 3.46-fold higher. e clearance rate (CL z/F) of ISO in the administration-alone group was 3.15-fold higher than the coadministration group, whilst RHY was 3.16-fold higher. e mean residence time (MRT) of both groups was higher than those of the single-administration group. is indicated that coadministration prolongs the t1/2z of ISO and RHY, reduces clearance rates, and increases the average residence time, prolonging drug activity in vivo. GAS is a slowly transforming metabolite of PA in the blood [24]. When the pharmacokinetic parameters of GAS were optimized via coadministration, the AUC was 1.31-fold higher than the single-administration group, whilst the t1/2z was 2.79-fold higher and the MRT was 1.84-fold higher. is suggested that coadministration promotes drug efficacy in vivo. To date, traditional Chinese medicinal pharmacokinetic studies have been based on healthy experimental animals, discounting the physiological and pathological changes that occur during disease. e pharmacokinetic analyses of drugs used in the treatment of diabetes, cerebral ischemia, hepatobiliary, intestinal disorders, and kidney disease significantly differ from the normal physiological state [25][26][27][28].

Pharmacokinetic Study.
e disease process can lead to pathological states that influence the activity of related enzymes. As such, animal disease models provide more credible guidance for clinical drug use and enhance practical significance. Figure S4 shows a representative total ion chromatogram (TIC) of plasma samples in positive and negative ion modes. Figure 2 shows the multivariate analysis of potential migraine biomarkers. PCA data are shown in Figure 2

(d) (the 24 h group). ese included the control (green) administration groups at 1 h and 24 h (blue) and model groups at 1 h and 24 h (red).
e PCA figures showed that the QC samples displayed good aggregation which reflected the stability of the LC-MS system. e quality of the sample was thus reliable [29]. e PCA analysis was validated no outliers. To further separate the samples and better reflect the differences between control and model groups, samples were analyzed using the supervised mode OPLS-DA, which possesses an improved clustering effect. e results of OPLS-DA showed that the model group and the control group were divided well from each other. Besides, relevant R 2 y and Q 2 y values were applied to evaluate the quality of the OPLS-DA model. In the rat lipid metabolism curves, the R 2 y was 0.966, R 2 x was 0.861, and Q2 was 0.897. In the 24 h group, the R 2 y was 0.967, R 2 x was 0.867, and Q2 was 0.889. ese results indicate that the model displayed improved predictive parameters indicating significant differences in metabolism between the migraine and control groups [30].

Identification of Important Lipid Biomarkers. For the assessment of lipid biomarkers, samples were divided into 1 h analysis (including the 1 h control and 1 h model groups) and 24 h analysis groups (including the 24 h control and 24 h model groups).
e OPLS-DA model was established to screen contributing variables. A total of 77 variables with a VIP greater than 1 were observed in the 1 h analysis group, and 170 were identified in the 24 h analysis group. e metabolites with a VIP > 1 and P < 0.05 were identified as candidates. e accurate mass and its MS/MS fragments of the biomarkers were then identified by matching with the online databases such as HMDB (http://www.hmdb.ca/), METLIN (https://isometlin.scripps.edu/), MassBank (http:// www.mass-bank.jp/), and ChemSpider (http://www. chemspider.com/). e error between extraction mass value and experimental mass value was less than 5 ppm. During this stage, 44 and 90 potential lipid biomarkers were identified in the 1 h and 24 h groups, respectively. To assess the importance of each lipid biomarker, the ROC was assessed. A total of 37 biomarkers were obtained according to the AUC values (Figure 3), 8 of which were in the 1 h analysis group, 16 were in the 1 h and 24 h analysis groups, and 13 were in the 24 h analysis group. In the model group, the biomarkers (Table 2) in both 1 h and 24 h groups were considered important. ese included phosphatidylcholines (PCs), lysophosphatides (lysoPC and lysoPE), and sphingolipids (Cer and SM). Box plot analysis (Figure 4) of the 16 biomarkers showed a significantly higher content in the model group compared to the control group, followed by a downward trend following PI administration.

Analysis of Metabolic Pathways and Important Lipid
Biomarkers. Metabolic pathway analysis is shown in Figure 5. Figure 5(a) provides the metabolic network of migraines of the 1 h analysis group. ese include glycerophospholipid, glycerolipid, and glycosylphosphatidylinositol (GPI)-anchor biosynthesis metabolism. Figure 5(b) shows the metabolic network of migraines of the 24 h analysis group. ese include sphingolipids, linoleic acid, and glycerolipid metabolism. According to the impact value, glycerophospholipid and sphingolipid metabolism were the major pathways of migraine regulation. PC, LPE, and LPC were involved in glycerophospholipid metabolism. In addition, SM and ceramides participated in sphingolipid metabolism. Glycerophosphatidyl is the most abundant type of phospholipids in the body with the basic structures of phosphatidic acid and substituents, which can be divided into phosphatidylcholine, phosphatidylethanolamine, phosphatidylserine, phosphatidylglycerol, phosphatidylinositol, and so on, according to different substituents [31,32]. e main function is to form biofilm and participate in protein recognition and signal transduction in cell membranes [33]. Lysophospholipids are produced by the hydrolysis of glycerophospholipids or sphingomyelins to remove one fatty acid side chain [34]. Among which, lysoPC and lysoPE have high abundance and possess important physiological functions and activities [35]. In this study, PC components, LPC (16 : 0), and LPE (16 : 0) were identified as important lipid biomarkers of migraine, which involved in the metabolic pathway of glycerophospholipids. Compared with the blank group, the content in migraine model rats increased significantly. PC is the precursor of acetylcholine and can be hydrolyzed into choline in the body [36]. Choline enters the brain along with the blood circulatory system and combines with acetic acid to convert to acetylcholine. Acetylcholine is an important neurotransmitter that promotes communication between brain nerve cells, especially brain neurons [37,38]. When the content is increased, the information transmission speed between the brain neurons is accelerated, and the brain nerve function is enhanced. In migraine, promoting the production of PC may regulate the activity of brain tissue. Besides, studies have shown that most migraine sufferers suffer from anxiety and insomnia with or without other stimuli, and these causes of migraine can lead to oxidative stress; when oxidative stress occurs, the production of free radicals activates phospholipase       Journal of Analytical Methods in Chemistry A2 (a biological enzyme that promotes the production of lysophospholipids), thereby increasing the content of lysophospholipids [39][40][41][42]. Sphingolipids are the main components of eukaryotic cell membranes, which play an important role in regulating the fluidity of lipid bilayer membranes and in cell signal transduction [43]. Sphingolipids mainly exist in the nervous system, including sphingosine, ceramide (Cer), and sphingomyelin (SM). Human research demonstrated that changes in the balance of sphingolipid metabolism were closely related to neurological diseases [44,45]. Another basic scientific study suggested that sphingolipids may be involved in pain-related neurological and signaling pathways [46][47][48]. In this study, we found SM (d17 : 0/24 : 2), SM (d16 : 0/26 : 2), SM (d25 : 0/16 : 1), and CerG2 (d13 : 0/19 : 1) were important markers for migraine.

Conclusions
In this study, the metabolic processes of PI in migraine models were assessed using a combination of pharmacokinetic and lipidomic methods in vivo (endogenous lipid components) and in vitro (PI components). rough comparison of the pharmacokinetic parameters following co-or single-administration, the advantages of coadministration were highlighted. In addition, important lipid biomarkers of migraines were revealed through lipidomic analysis, and the regulatory effects of PI on migraines were confirmed. ese studies provide practical reference values for the clinical application of PI.

Data Availability
e data used to support the findings of this study are included within the article.

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

Authors' Contributions
Chaoqun Zhou and Mingzhen He contributed equally to this work.