Filtration of Active Components with Antioxidant Activity Based on the Differing Antioxidant Abilities of Schisandrae Sphenantherae Fructus and Schisandrae Chinensis Fructus through UPLC/MS Coupling with Network Pharmacology

This study attempted to filter active components with antioxidant activities based on the differing antioxidant abilities of Schisandrae Sphenantherae Fructus (SSF) and Schisandrae Chinensis Fructus (SCF). First, the antioxidant activity of SSF and SCF was evaluated through the DPPH free radical scavenging method and compared with the half maximal inhibitory concentration (IC50) value. Next, components of SSF and SCF were detected by employing ultrahigh-performance liquid chromatography-Q-Exactive Orbitrap mass spectrometry (UPLC-QEO/MS) technology, and differential compounds were screened out as potential antioxidant compounds by using Compound Discover 3.1 Software. After that step, in order to verify the antioxidant compounds, the network method was applied. Biological targets were searched in the GeneCards database, and that related to antioxidant ability were selected in the Comparative Toxicogenomics Database (CTD). Finally, the pharmacology network was constructed. Results showed that SSF and SCF possessed different compounds and antioxidant abilities. A total of 14 differential compounds such as γ-schizandrin, schisandrin B, schisandrin, and tigloylgomisin H between them were screened out and identified. Twenty targets associated with antioxidant activity contained MAP2K1, MAPK8, RPS6KB1, PRKCB, HIF1A, and so on were investigated. Thirty-six pathways contained HIF-1 signaling pathways, choline metabolism in cancer, serotonergic synapse, Fc epsilon RI signaling pathway, GnRH signaling pathway, and so on related to the above twenty targets were identified. The pharmacology network analysis indicated that the differential components may be helpful in treating various diseases, especially cancer, by exerting antioxidant activity. In conclusion, this study provided a novel method for identifying active components with antioxidant activity in SSF and SCF, and this method may be applicable for the filtration of bioactive components in other herbs.


Introduction
Antioxidation is a process that can effectively inhibit the oxidation reaction caused by free radicals. e mechanism of antioxidation involves either an action on free radicals or the consumption of substances that can easily generate free radicals.
Free radicals are constantly produced in the body due to exposure to the environment through processes such as respiration (oxidation reaction), air pollution, and radiation. Free hydroxyl radicals are associated with aging stress [1], autoimmunity [2], bone loss, and cancer development [3]. However, substances that possess antioxidant activity can terminate or inhibit the oxidation process by scavenging free

Sample Preparation. SCF
and SSF were prepared using the reflux extraction process [11]; the extraction method for them was optimized based on the DPPH radical clearance rate by utilizing an L9 (3 4 ) orthogonal design. During orthogonal experiments, factors concerning extract times, reflux time, and material-liquid ratio were all inspected with four levels. e optimal extraction process was as follows: first, SCF or SSF was weighed 2.5 g. Next, the herbal medicine was mixed with 20 volumes of water and then boiled for 2 h, 5 times. e extract was then filtered, mixed, and concentrated to 50 mL. After that, 2 mL of the extract was measured and dried. Finally, the residue was dissolved using 2 mL anhydrous ethanol before detection.

DPPH Radical Clearance
Test. DPPH was dissolved with 95% ethanol aqueous solution to prepare 0.1 mmol/L of DPPH solution and 2 mL of it was mixed with 2 mL sample. e mixture was then detected at 517 nm using a UV spectrometer after 30 min.

Condition of Detection.
e absorbance of the SCF and SSF extracts was measured on an ultraviolet-visible spectrophotometer TU1901 (Persee Co., China) in the spectral scan mode in the range of 510∼516 nm.
Compounds in SCF and SSF were detected on UPLC-MS equipment consisting of a Dionex UltiMate 3000 UHPLC and Q-Exactive Orbitrap MS with an electrospray source in positive-ion mode ( ermo Co., USA).
UPLC separation was performed on a TM C18 column (2.1 × 50 mm, 2.6 μm, ermo Co., USA). Formic acid was added to the mobile phase, which consisted of methanol (A) and 0.1% formate water (B), to improve ionization efficacy and acquire a better peak shape. e flow rate was set at 0.3 mL·min −1 . UPLC resolution was optimized with gradient elution as follows: 0∼2 min, 40%∼60% A; 2∼6 min, 60%∼ 100% A; and 6∼8 min, 100% A. e column temperature was 25°C, and the injection volume was 0.5 µL. e optimal mass spectrometry signal was obtained as follows: capillary (+4.0 kV), desolvation temperature (350°C), S-lens voltage (50 V), shield gas (35 arb), aux gas (10 arb), scan type (full scan), and scan range (100∼1200 Da). e collision energy for MS 2 was set at 10, 30, and 40 eV to acquire abundant fragment ions. e UV method was confirmed by inspecting the accuracy and stability of the absorbance of the same SCF sample six times. e stability was determined by testing six times in 30 min. A relative standard deviation percentage (RSD%) of absorbance below 1% was considered to indicate a good detection method. e UPLC-MS method was validated by inspecting the accuracy and stability through the detection of six compounds in both SCF and SSF. e accuracy was determined by injecting the same SCF sample continuously six times. e stability was implemented by injecting the same SCF sample at 0, 12, 24, 32, 40, and 48 h. e RSD% of retention time and peak area below 5% was considered to indicate a good detection method.

Data Analysis.
e UV absorbance data of SSF and SCF were used to calculate the DPPH radical clearance, as well as the IC 50 value. First, the DPPH radical clearance was calculated as (Abs blank −Abs sample )/Abs blank × 100%. Second, the IC 50 value was acquired through curve fitting, with extract concentrations of 0.2, 0.4, 0.6, 0.8, and 1.0 mg/mL serving as the X-axis and the DPPH radical clearance value serving as the Y-axis, by GraphPad software. ird, the IC 50 values of the medicines were compared with that of vitamin C to estimate their antioxidant ability.
All UPLC/MS raw data were acquired by Q-Exactive Tune ( ermo Co., USA) and processed with Compound Discoverer 3.1 ( ermo Co., USA). Next, a data list containing data such as molecular weight, retention time, peak intensity, and p value of SCF versus SSF was acquired. Compounds with a p value <0.05 were considered to be significantly different between SCF and SSF. Principal component analysis (PCA) was used to observe the overall classification of the data. Next, the differential compounds were identified by searching the mzCloud, ChemSpider, PubChem, NIST Chemistry WebBook databases, and references based on their MS 2 spectra. e targets of differential compounds were screened out in the Swiss Target Prediction database after the canonical SMILES number was obtained from PubChem and compounds were standardized by the UniProt database. e CTD and GeneCards databases were searched for the targets of antioxidant function. After that step, the common targets of differential compounds and antioxidant function were imported into the STRING database, and "multiple proteins" and "human sources" were checked. Next, proteinprotein interactions (PPIs) were depicted using Cytoscape. Next, Gene Ontology (GO) and Encyclopedia of Genes and Genomes (KEGG) enrichment analyses were performed using the Metascape database. Finally, a network pharmacology diagram was drawn with the help of Cytoscape software.

Filtration of the Optimal Extraction Process for Antioxidant Activity.
e results of the orthogonal test are shown in Table 1. It can be seen that, through the optimization of the extract method, sample prepared by the extraction process with the material-liquid ratio of 1 : 20, extracted for 5 times and 2.0 h/time showed the optimal DPPH radical clearance rate, which is used for the extraction of SSF and SCF.

Estimation of Antioxidant Activity between SSF and SCF.
e DPPH radical clearance ability of SSF and SCF was estimated by comparing the IC 50 value with vitamin C (VC); the lower the IC 50 value was, the better the antioxidant activity was. e results showed that the IC 50 values of SCF, SSF, and VC were 0.4518 ± 0.0008 mg/mL, 0.4928 ± 0.0015 mg/mL, and 0.2225 ± 0.0037 mg/mL, respectively, which indicated that SCF exhibited better antioxidant ability than SSF.

Filteration and Identification of the Potential Antioxidant Active Components.
To determine the reason for the antioxidant ability of SSF being better than that of SCF, it was surmised that there were some compounds that showed a higher intensity in SCF than in SSF; therefore, the differential compounds between SSF and SCF were filtered.
First, samples of SSF and SCF were detected by UPLC/ MS, and total chromatograms were obtained using Q-Exactive Tune ( ermo Co.). e clear difference in the peak amount and intensity between SSF and SCF is depicted in Figure 1.
Second, the overall difference between SSF and SCF was observed through unsupervised principal component analysis (PCA) (Figure 2), and variables making significant contributions to discrimination between them were selected based on a p value <0.05 by using Compound Discoverer 3.1 software. Next, the Traditional Chinese Medicine Systems Pharmacology Database (TCMSP, https://tcmspw.com/ tcmsp.php), mzCloud (https://www.mzcloud.org/), Chem-Spider (https://www.chemspider.com/), PubChem (https:// pubchem.ncbi.nlm.nih.gov/), and the National Institute of Standards and Technology (NIST) Chemistry WebBook (https://webbook.nist.gov/chemistry/) were searched to determine the molecular weight and the molecular formula of different variables, and their MS 2 spectra were compared with those in references to identify the chemical structures. A total of 14 compounds were identified, and all their intensity ratios in SCF versus SSF were greater than 1; therefore, they were initially considered antioxidant compounds, as shown in Table 2.
Among these compounds, espatulenol was proposed by comparing the MS 2 spectrum with that in the NIST Chemistry WebBook database, c-schizandrin, schisantherin B, schisantherin C, tigloylgomisin H, gomisin J, angeloylgomisin P, angeloylgomisin H, gomisin M1, gomisin E, Evidence-Based Complementary and Alternative Medicine and gomisin K1 were proposed by comparing the MS 2 spectrum with that in references [12][13][14][15][16][17], and schilancifolignan A was speculated based on its MS 2 fragments [18]. Schisandrin and schisandrin B were verified by their standards. e ion of m/z 455.2040, which was one significantly different compound between SSF and SCF, was selected as an example to illustrate the compound identification process. First, the extract ion spectrum of m/z 455.2040 was acquired at t R 3.71 min. en, its MS 2 spectrum was acquired under 40 eV, which is shown in Figure 3.  3 ] + based on high-resolution mass spectrometry. Hence, it was deduced to schisandrin initially. Further, the schisandrin standard was detected by the same UPLC/MS condition for verifying the deduction, which showed that their retention time and MS 2 spectrum were matched completely. erefore, the ion of 455.2040 was identified as schisandrin.
e proposed fragmentation pathways of schisandrin are shown in Figure 4.

Prediction of the Targets of Antioxidant Compounds.
A total of 441 targets on which the abovementioned potential antioxidant compounds acted were obtained. At the same time, 4151 targets associated with antioxidant activity were obtained by searching the GeneCards database, and 3946 of them were retained after UniProt processing. Furthermore, 4531 targets related to antioxidant activity were obtained by searching the CTD database, and 150 of them were retained after UniProt processing. ere were 147 common targets in the GeneCards database and CTD database. Furthermore, the 147 targets were compared to the abovementioned 441 targets, and it was found that 20 targets on which the antioxidant compounds acted were related to antioxidant activity; these targets are listed in Table 3.

Protein-Protein Network Construction.
e proteinprotein interaction (PPI) network diagram was constructed using Cytoscape software ( Figure 5). Figure 5 shows that the network graph contained targets and 24 edges. Each target connected several functions. e larger the degree value was, the larger the node was, and the larger the combined score value was, the thicker the edge was. Among these genes, the top 3 genes with larger nodes included MAPK8, CCND1, and RPS6KB1, which indicated that they were important targets on which the potential antioxidant components acted.

Gene Ontology Enrichment
Analysis for Targets. GO (gene ontology) enrichment analysis of potential targets was performed using the Metascape database, and the functions were enriched under the conditions of p value <0.01 and enrichment factor >1.5. Figure 6 shows the top 20 most significant functions. In Figure 7, the point size represents the number of genes with the same function; the larger the point size is, the more genes there are. e results showed that potential antioxidant compounds affected numerous biological functions involving biological processes, cellular components, and molecular functions, such as the regulation of aging, the active regulation of programmed cell death, the positive regulation of reactive oxygen metabolism, the regulation of neuronal apoptosis, positive regulation of the cell cycle, and regulation of the collagen metabolism process. ese targets might slow the occurrence and development of aging and hypoxia caused by the oxidation process by participating in the process of antioxidant reactions.

Pathway Analysis for Targets.
e top 36 enriched pathways of the above 20 targets were obtained from the DAVID (Database for Annotation, Visualization, and Integrated Discovery) database, which is shown in Figure 7. e length of the column represents the number of genes participating in the pathway; the longer the column is, the more genes there are. Among these pathways, pathways in cancer, the HIF-1 signaling pathway, insulin resistance, the insulin signaling pathway, hepatitis B, the AMPK signaling pathway, the MAPK signaling pathway, the TNF signaling pathway, and the FoxO signaling pathway were reported to be related to antioxidant activity [19][20][21][22][23][24][25][26][27].

Construction of Compound-Target-Pathway Network.
e compound-target-pathway network was constructed by connecting 14 differential compounds, 20 targets, and 36 pathways. Figure 8 shows that there were a total of 70 nodes and 230 edges. e larger the node was, the more nodes were connected to it. erefore, angeloylgomisin H, aigloylgomisin H, angeloylgomisin P, schisantherin C, and schisantherin B were deemed to have stronger antioxidant activity than other compounds. MAP2K1, MAPK8, PRKCB, and CCND1 were considered to be the primary antioxidant effect

Discussion
SSF and SCF, which serve as typical "medicine-food homology" herbal medicines, showed similar pharmacological effects, such as antioxidant activity. Nevertheless, SCF was generally more popular than SSF because it was said that SCF possessed higher quality than SSF. Regarding SSF and SCF, because both of these medicines had multipharmacological effects, the main effects that were essential for quality differences were not clear. e antioxidant activity or components of medicines have long been compared by medicinal scholars [28,29], but to the best of our knowledge, studies on the relationship of antioxidant ability and differential components have not been performed to date. erefore, this study described their relationship and predicted the functional targets of antioxidant activities using a network pharmacology strategy. e compound-target-pathway network diagram is shown in Figure 8, in which there were a total of 70 nodes and 230 edges. e larger the degree value was, the larger the nodes were, and the stronger the Evidence-Based Complementary and Alternative Medicine antioxidant activity of the compounds was. Among the 14 differential compounds, seven compounds, namely, c-schizandrin, espatulenol, angeloylgomisin H, tigloylgomisin H, schisantherin B, gomisin K1, and gomisin M1, which were colored yellow, were unique components in SCF that were not observed in SSF. In addition, five compounds, namely, angeloylgomisin H, tigloylgomisin H, angeloylgomisin P, schisantherin C, and schisantherin, showed larger nodes than others, which indicated that they were the main antioxidant components. Among the 20 targets, four, namely, MAP2K1, MAPK8, PRKCB, and CCND1, showed larger nodes than the others, which indicated that they were the main targets of antioxidant functions. Among the 36 pathways, pathways in cancer, choline metabolism in cancer, and proteoglycans in cancer showed larger nodes than others, which indicated that they were the main pathways through which the abovementioned bioactive compounds exerted antioxidant function. As a dual-specific kinase, mitogen-activated protein kinase kinase 1 (MAP2K1) participates in the extracellular  regulated protein kinase (ERK) pathway by activating ERK1 and ERK2. rough the identification of transforming genes for refractory diseases in the sclerosing subtypes of highly invasive tumors of gastric cancer, it was determined that cancer cells are dependent on the increased proliferation activity of MAP2K1, and MAP2K1 could be utilized as a cancer inhibitor target [30]. MAPK8 (mitogen-activated protein kinase 8), which is a protein-coding gene, was observed to participate in the abovementioned 16 pathways and to connect to 9 targets in the PPI network diagram, which indicated its importance. MAPK8-associated diseases include fatty liver disease and hepatitis C. e pathways related to MAPK8 include the ATM (serine/threonine kinase) pathway and the link between physicochemical characteristics and toxicity-related pathways. Previous studies showed that MAPK8 is involved in the gene expression process of anterior papillary hypertension caused by hypoxia [31].
PRKCB (protein kinase C) has been identified as a drug treatment target for specific diseases involved in various cellular processes, such as the regulation of B-cell receptor (BCR) signaling bodies, oxidative stress-induced apoptosis, male hormone receptor-dependent transcriptional regulation, insulin signaling, and endothelial cell proliferation.    indicator of tumorigenesis and tumor progression. Increased levels of MAPK8 and HIF1A could induce the expression of related genes and subsequently affect the cell growth and proliferation [33]. erefore, it could be deduced that the differential compounds might help to treat cancer by acting on MAPK8 and HIF1A.      G1/S checkpoint of the cell cycle. Takano et al. [34] hypothesized that the synergistic effect of CCND1 and cdk4 with ER may cause breast cancer. rough pathway analysis, it was observed that the antioxidant activities of SSF and SCF were closely related to the HIF-1 (hypoxia-inducible factor 1) signaling pathway, pathways in cancer, and choline metabolism in cancer. HIF-1 is a transcription factor and a major regulator of oxygen homeostasis. is protein consists of an inducible HIF-1alpha subunit and a constitutively expressed HIF-1beta subunit. Under normoxia and hypoxia, the conversion of the two subunits regulates their transcriptional activity. Under hypoxia, HIF-1 is the main regulator of many hypoxia-induced genes. HIF-1-related factors may cause such diseases as diabetic retinopathy, glucocorticoid-induced osteonecrosis, and malignant paraganglioma [35].
Antioxidants were frequently used for cancer treatment in previous studies [36], and angeloylgomisin H was reported to be cytotoxic to human cancer cell lines [37], which suggested that antioxidant active compounds might also possess anticancer activity. Among the 36 pathways obtained in this study, 9 related to pathways in cancer accounted for 25%, which indicated that SCF and SSF might serve as cancer treatments by exerting antioxidant activity.

Conclusion
In this study, the antioxidant activity of SSF and SCF was assessed and compared by DPPH free radical scavenging experiments, and the differential compounds, as well as their associated biological functions, were obtained by coupling UPLC/Q-Exactive Orbitrap MS technology with network pharmacological analysis. e results showed that the antioxidant ability of SCF was stronger than that of SSF, and 14 differential compounds were identified between SCF and SSF, which indicated that these compounds were potential active components with antioxidant activity. Further analysis showed that there were a total of 20 predicted targets and 36 pathways related to these active components with antioxidant activity, and most of these targets and pathways were associated with cancer regulation. e pharmacology network predicted that the differential components might exert antioxidant effects on the 20 targets by regulating the 36 pathways. ese medicines may be helpful for treating various diseases, especially cancer, by exerting antioxidant activity. erefore, this study provided a novel method for identifying active components with antioxidant activity, and this technique may be applicable for the filtration of bioactive components in other herbs.