A Novel Strategy for Screening Active Components in Cistanche tubulosa Based on Spectrum-Effect Relationship Analysis and Network Pharmacology

Cistanche tubulosa (Schenk) R. Wight is a valuable herbal medicine in China. The study aimed to explore the potential mechanisms of C. tubulosa on antioxidant activity using spectrum-effect relationship and network pharmacology and the possibilities of utilizing herbal dregs. In this work, different extracts of C. tubulosa, including herbal materials, water extracts, and herbal residues, were evaluated using high-performance liquid chromatography (HPLC) technology. In addition, the antioxidant activities were estimated in vitro, including 2, 2-diphenyl-1-picrylhydrazyl; superoxide anion; and hydroxyl radical scavenging assays. The spectrum-effect relationships between the HPLC fingerprints and the biological capabilities were analyzed via partial least squares regression, bivariate correlation analysis, and redundancy analysis. Furthermore, network pharmacology was used to predict potential mechanisms of C. tubulosa in the treatment of antioxidant-related diseases. According to the results, eleven common peaks were shared by different extracts. Geniposidic acid, echinacoside, verbascoside, tubuloside A, and isoacteoside were quantified and compared among different forms of C. tubulosa. The spectrum-effect relationship study indicated that peak A6 might be the most decisive component among the three forms. Based on network pharmacology, there were 159 target genes shared by active components and antioxidant-related diseases. Targets related to antioxidant activity and relevant pathways were discussed. Our results provide a theoretical basis for recycling the herbal residues and the potential mechanisms of C. tubulosa in the treatment of antioxidant-related diseases.


Introduction
Cistanche tubulosa (Schenk) R. Wight, one of the most frequently used herbs in the Cistanche family, is known as the "Ginseng of the Desert" for its various health benefts [1]. Modern pharmacological investigations have discovered that the Cistanche family ofers various pharmaceutical effects, such as antioxidant, anticancer, hypoglycemic, antidepressant, cognitive improvement, and antimicrobial efects [2]. Te primary chemical cluster in Cistancheherb a is phenylethanoid glycosides (PhGs) [1]. PhGs derived from Cistanche species can elevate the activities of superoxide dismutase (SOD) and glutathione peroxidase (GSH-Px) [3]. Echinacoside, one member of the PhG family, reverses efectively in vivo oxidative stress induced by highglucose diets via downregulation of the nitrous oxide system (NOS) activity and phospho-eNOS expression [4]. Te basic skeleton of the PhG consists of phenylethyl alcohol and glycosyl moieties. PhGs are believed to have a signifcant activity due to phenolic hydroxyl groups in their structure. Te antioxidant activity of PhGs increases with the presence of phenolic hydroxyl groups [5].
Reactive oxygen species (ROS) have crucial roles in many physiological processes and essential protective mechanisms. Frequent exposure to high ROS concentrations may contribute to nonspecifc damage to proteins, lipids, and nucleic acids. ROSs can be neutral molecules (e.g., hydrogen peroxide), ions (e.g., superoxide anion), or radicals (e.g., hydroxyl radicals) and exert their efects via regulation of cell signaling cascades [6]. Teir rapid production and removal are infuenced by a variety of mechanisms. Tey are lightweight and difuse easily across short distances [7]. Te superoxide anion (O •− 2 ) is a precursor of most ROSs and an intermediate species in oxidative reactions. Hydroxyl radicals (OH•) are catalyzed by reduced transition metals, which may in turn be reoxidised by O •− 2 . An imbalance between excessive production of ROS and limited antioxidant defenses leads to various deleterious processes also called "oxidative stress" [8]. If such an imbalance can be corrected, the management of several defense mechanisms may be manipulated. Oxidative stress is involved in various pathological conditions, such as cancer, cardiovascular disease, neurological disorders, and diabetes [6]. In addition, the 2,2-diphenyl-1-picrylhydrazyl (DPPH•) radical is colored and remarkably stable, and thus is among the most common radicals considered in numerous studies [9]. A DPPH scavenging assay is an easy-to-implement, accurate method of measuring the total antioxidant capacities of botanical or herbal extracts [10]. Terefore, DPPH, superoxide anion, and hydroxyl radical scavenging assays have been used to evaluate the antioxidant capacity of diferent extracts of C. tubulosa.
Te extract production steps for traditional Chinese medicine (TCM) are complex and involve cutting, processing, extraction, concentration, and drying [11]. It is inevitable that active components in herbal materials (HMs) will be lost during such sophisticated manufacturing stages. PhGs are characterized by at least one glycosyl moiety at their core, which determines their properties. Tese compounds are water soluble and easy to extract via traditional methods. It is also the case that water extraction technology is widely used in the production of TCM [12]. Te resulting water extracts (WEs) are formulated into various dosage forms such as tablets, granules, capsules, and mixtures. Inevitably, the loss of biological ingredients occurs during such a large-scale production process. Tis is why quality assessment is an indispensable step and is required to ensure the quality of semifnished products from processes. Some researchers have already utilized these residual materials via structural modifcation. Astragalus membranaceus residue was purifed to produce a polysaccharide that improved cognitive dysfunction by altering gut microbiota in diabetic mice [13]. A neutral polysaccharide extracted from Codonopsis pilosula residue exhibited a hypoglycemic efect [14]. Large amounts of herbal residues (HRs) are manufactured in China. Te reutilization of HRs has become a new and novel research feld as TCM processes have undergone modernization [15]. Te potential for diferences between the antioxidant properties of various herbs and their WEs has become a focus of our attention. In addition to considering WEs of C. tubulosa, it is hoped to determine whether HRs can prove useful as candidate TCMs and to explore opportunities for transforming discarded C. tubulosa waste material into feasible products.
Spectrum-efect relationships are utilized to determine efective components in complex mixtures and refect the internal quality of herbal medicine. It is indispensable in the process of modernization and internationalization of herbal medicine. Since the spectrum-efect relationship research of herbal medicines is based on the chromatographic fngerprint, a suitable analytical method is required to generate a fngerprint that refects the chemical ingredients of herbal medicines. High-performance liquid chromatography (HPLC) is an important analytical method that has many advantages, such as high separation, good stability, high efciency, and high quantitative precision. PLSR (partial least squares regression) simplifes the data structure and correlation analysis between two sets of variables by using regression modeling [16]. In bivariate correlation analysis (BCA), test scores are correlated with conceptually related constructs in order to establish valid evidence [17]. RDA (redundancy analysis) has been used to identify the primary microbial communities related to special biological capacity, but we applied it to the spectrum-efect relationship [18]. To evaluate the correlation coefcients, PLSR and BCA were used. Te results were then verifed using RDA to determine which models were more appropriate for studying the spectrum-efect relationship with C. tubulosa.
In the methodologies of "multicomponent therapeutics, biological network" in network pharmacology, we try to search for common targets between active molecules and diseases, which may play an indispensable role in providing a reference for the prevention of diseases. Cistanche herba exhibit potential multicomponent and multitarget properties in the previous reports [19,20]. Biomarkers of oxidative damage associated with human diseases are summarized [21]. Few studies clarify the antioxidative mechanism of C. tubulosa. Hence, the network pharmacology technology was adopted to investigate bioactive molecules of C. tubulosa and mechanisms of C. tubulosa against oxidation.
In this study, the chromatographic fngerprints and antioxidant activities of HMs, WEs, and HRs from 11 batches of C. tubulosa were evaluated simultaneously via HPLC and antioxidant assays. A spectrum-efect relationship between the HPLC fngerprints and the antioxidant efects of C. tubulosa was revealed clearly via a series of correlation analyses. Existing studies mainly report on the antioxidative properties of C. tubulosa herbs [22][23][24], but few of them highlight the antioxidant capabilities of the HEs. In addition, the role of HRs might be understated, and this study provides a new opportunity to take advantage of new research in this feld. Te purpose of this research was to identify the major active ingredients within and the antioxidant activities of HMs, Journal of Analytical Methods in Chemistry WEs, and HRs of C. tubulosa. Ten, chemometrics was applied to identify spectrum-efect relationships for the HMs, WEs, and HRs from 11 batches of C. tubulosa. Tis is the frst time that diferences among HMs, WEs, and HRs of C. tubulosa have been studied in this manner. In addition, the network pharmacology analysis was performed to elucidate the underlying mechanisms of C. tubulosa in the treatment of antioxidant-related diseases.

Collection of WEs and
HRs. WEs were prepared as described in the previous article with modifcations [25][26][27]. Herbs (100 g) were extracted with water at a ratio of 1 : 10 (w/v) in a ceramic container and heated for 2 hours. After fltration, the extract and wet material were separated. Te extraction process was repeated using water at a ratio of 1 : 8 (w/v) and the resulting material was heated for 1 hour. Te two separate extracts were combined. Te mixed extracts were concentrated using a rotary evaporator to yield a creamy solution with a density of 1.05 g/cm³. An appropriate amount of maltodextrin was added. Spray dryers have fast drying speeds and good product performance. Spraying atomizes the liquid material into dispersed particles, thereby increasing its surface area. Contacting hot air facilitates the drying process very quickly. A Buchi B-290 spray dryer (Buchi, Switzerland) was used to spray dry the mixture. Te inlet and outlet temperatures were set to 175-205°C and 85-95°C, respectively. Te WEs were collected. During processing of the WEs, the wet residues were collected and dried at 50°C to give the HRs. Figure 1 shows the procedure for preparing WEs and HRs.

Preparation of Standard Solutions.
Appropriate amounts of fve reference compounds were dissolved in 50% aqueous methanol and then fltered through a 0.22-μL microporous membrane to yield a mixed standard solution. Upon adding 50% aqueous methanol to a 10-mL volumetric fask, the mixed standard solution contained geniposidic acid, echinacoside, verbascoside, tubuloside A, and isoacteoside. Pure reference compound solutions were injected into the HPLC system for qualitative analysis, and their retention times were recorded. Comparing retention times allowed reference compounds to be identifed.

Precision, Reproducibility, and Stability.
Te powder of HM1 was prepared as described in Section 2.4. Method precision was evaluated using six successive injections of one sample solution, while reproducibility was estimated by performing six replicates of a sample. Stability tests were performed by replicating injections of one sample solution that had been kept at 15°C for 0, 2, 4, 8, 12, and 24 h.

Linearity.
Te mixed reference solutions with diferent gradient concentrations were injected and analyzed. Te concentration ranges for geniposidic acid, echinacoside, verbascoside, tubuloside A, and isoacteoside were 0.0011-0.3414 mg/mL, 0.0081-2.421 mg/mL, 0.001-0.3132 mg/mL, 0.0005-0.1518 mg/ mL, and 0.0004-0.1116 mg/mL, respectively. Analytical curves for each standard were obtained by considering the correlation between the peak area (y) and concentration (x, mg/mL) using a linear least squares model.

Sample Recovery. Sample recovery was investigated by
adding an accurate amount of standard solution to 0.5 g of HM1 sample powder. Nine samples were prepared in parallel according to Section 2.4. Te mean sample recovery of each component was determined.

Sample Determination.
As described in Section 2.4, HMs, Wes, and HRs were prepared in parallel. Tese sample solutions were injected following the chromatographic conditions described in Section 2.3. Te peak area was recorded and the contents were calculated. Te data were analyzed using GraphPad Prism 8 (GraphPad Software, California, USA). p values were calculated using the one-way analysis of variance followed by the Tukey method. p < 0.05 was considered statistically signifcant.

Fingerprint Establishment and Evaluation.
HPLC chromatographic data were output from Chromeleon 7.2.8 Software (Termo Fisher Scientifc, Massachusetts, USA) in CDF and TXT format. Te HM-WE, WE-HR, and HM-HR similarity values were calculated using a similarity evaluation system designed for chromatographic fngerprints within TCM Software (Version 2004A). HPLC fngerprints were drawn using Origin 2021 Software (OriginLab, Massachusetts, USA).

Antioxidant Activity Evaluation.
Te measurement procedures for the antioxidant level were conducted according to the instructions provided in the various kits. Te absorbance values were measured using a Shimadzu UV-2600i (Shimadzu, Japan). Each sample was run in triplicate, and the average data were recorded. Various sample concentrations and their corresponding absorbance (A) values were recorded. Te 50% inhibiting concentration (IC 50 ) was calculated using GraphPad Prism 8 (GraphPad Software, California, USA).

DPPH Assay.
First, 0.1 g of HM, WE, and HR powders were extracted using 1 mL of an 80% aqueous methanol solution. Te resulting material was sonicated for 30 min and centrifuged at 12, 000 × g for 5 min. A 400 μL sample was mixed with the working fuids (Table S1), and a 1-mL cuvette was prepared. After the reaction in the dark at 25°C for 30 min, the absorbance at 517 nm was recorded and converted to radical scavenging activity (S DPPH ) as follows: Dehydrated ethanol was used to adjust to zero.

O •−
2 Assay. First, 0.1 g of HM, WE, and HR powders were extracted using 1 mL of an 80% aqueous ethanol solution. Te resulting material was sonicated for 30 min and centrifuged at 12, 000 × g for 5 min. A 100 μL sample was added into the working fuids (Table S2). After the reaction was allowed to proceed at 37°C for 10 min, the absorbance at 570 nm was recorded and converted to radical scavenging activity (S O •− 2 ) as follows: Purifed water was used to adjust to zero.  fuids (Table S3). After incubation at 37°C for 20 min, the absorbance at 536 nm was recorded, and the radical scavenging activity (S OH• ) was determined as follows: Purifed water was used to adjust to zero.

Data Analysis
2.8.1. PLSR Analysis. Te main chromatographic peak areas served as the independent variables (X) and the levels of antioxidant activity for the various assays were the dependent variables (Y). PLSR modeling was performed using Unscrambler X 10.4 Software (CAMO Software, Bangalore, India). Te weighted regression coefcients revealed correlations between the peak areas and antioxidant activity levels, and the raw regression coefcients defned the model equation.

BCA Analysis.
Peak areas were the independent variables (X), and the antioxidant levels for the various assays were treated as the dependent variables (Y). Ten, the BCA between X and Y was analyzed using a Pearson model.

Network Pharmacology Analysis
2.9.1. Screening for Active Ingredients of C. tubulosa. All the chemical constituents of C. tubulosa were obtained using traditional Chinese medicine systems pharmacology (TCMSP, https://www.tcmsp-e.com/). Te screening thresholds of each chemical component were set as oral bioavailability (OB) ≥ 30% and drug-likeness (DL) ≥ 0.18, respectively. Te InChIKey of bioactive ingredients was collected through the PubChem database (https://pubchem.ncbi. nlm.nih.gov/). Te protein targets of the active compounds were screened out through the SwissTargetPrediction database (https://www.swisstargetprediction.ch/). Te target names were converted into gene names using the UniPort protein database (https://www.uniprot.org/).

Construction of a Component-Target Network.
Te keyword "antioxidant" was used to search for disease-related targets on the GeneCards database (https://www.genecards. org/) and OMIM database (https://omim.org/). Te intersections of genes between active components and diseaserelated targets were visualized using a Venn diagram online (https://bioinformatics.psb.ugent.be/webtools/Venn/). Te bioactive ingredient targets of C. tubulosa were mapped to the target genes using Cytoscape 3.9.1 software (https:// cytoscape.org/) for constructing the component-target (C-T) network.

Gene Ontology and Kyoto Encyclopedia of Genes and
Genomes Enrichment Analyses. Gene ontology (GO) enrichment in biological processes (BP), cellular component (CC), and molecular function (MF), and kyoto encyclopedia of genes and genomes (KEGG) pathway enrichment were analyzed online using the Metascape database (https://www. metascape.org/) with the "Homo sapiens" setting. Te visualization bubble chart and GO histogram were formed online (https://www.bioinformatics.com.cn/).

Establishment of Protein-Protein Interaction and
Component-Target-Pathway Networks. Te overlapping antioxidation-related and predicted targets from active components were used to construct a protein-protein interaction (PPI) using the STRING database (https://stringdb. org/). Te conditions were set as described by Xin et al. [29]. Te PPI network was visualized using the Cytoscape software. Degree centrality (DC), betweenness centrality (BC), and closeness centrality (CC) were calculated through the "network analysis" function. DC, BC, and CC were set as >100, >0.03, and >0.3, respectively. According to KEGG pathways and target genes, Cytoscape software was used to construct a component-target-pathway (C-T-P) network.

Method Validation.
Te validation for the HPLC method showed that the relative standard deviation (RSD) for method precision, reproducibility, and stability was less than 2.85% for the relative peak area (n � 11) and 0.77% for the relative retention time (n � 11). Te precision of the same sample solution appeared within the range of 0.05-0.77% for relative time and 0.28-2.70% for the relative area of the common peaks. Te reproducibility of the experiment was within the range of 0.03-0.20% for the relative time and 0.23-2.59% for the relative area of the common peaks. Te sample stability was 0.09-0.24% for relative retention time and 0.75-2.85% for the relative area of the common peaks. Tese results indicated that the established fngerprint was satisfed. Te linear relationships for geniposidic acid, echinacoside, acteoside, tubuloside A, and isoacteoside are shown in Table S4. Te value of R square was 1.0000, indicating good linearity. Te results of sample recovery showed that the average recoveries of geniposidic acid, echinacoside, acteoside, tubuloside A, and isoacteoside were 100.37%, 103.59%, 98.46%, 100.81%, and 101.19%, and the RSD for sample recoveries was less than 2.68%.

Peak Area (PA) and Relative Retention Time (RRT).
Te reference fngerprints and fngerprints of HMs, WEs, and HRs from 11 batches of C. tubulosa are presented in Figure 2. Eleven peaks, which exhibited good separation and resolution, were identifed as common peaks among HMs, WEs, and HRs. Te fve standard compounds were identifed as geniposidic acid (A 2 ), echinacoside (A 8 ), acteoside (A 9 ), tubuloside A (A 10 ), and isoacteoside (A 11 ). Te standard compound, echinacoside, which was present in all chromatograms (average retention time 12.86 min) with a suitable peak area and good stability, was selected as the reference peak and used to calculate the relative retention times (RRTs) of the other ten common peaks. Te RRTs of these diferent forms are in the 0.16-1.51 range. Te PA and coefcient of variance (CV%) of these common peaks are listed in Tables S5-S7. From the data, the CV% values for PA in various forms are 25.78%-142.02%, 23.36%-150.38%, and 28.91%-112.78% for HMs, WEs, and HRs, respectively. Tese results reveal signifcant diferences in the concentration of each Cistanche tubulosa compound among the diferent forms. Te fngerprints of HMs, WEs, and HRs are shown in Figure 3.

Contents of HMs, WEs, and HRs.
Five standard constituents of C. tubulosa were measured. Te contents of the main components are shown in Table 1. Te comparison between HMs, WEs, and HRs is shown in Table 2 and Figure 4. PhGs in C. tubulosa are biologically active but thermosensitive. Heat-sensitive components dissolving in water can be efciently extracted using a reasonable method. Generally, Cistanche herba is extracted with water and then evaporated into a concentrated solution for the following chemical analysis [26,27,30]. After extraction and concentration, spray-drying technology was used and the procedure was modifed from the previous article. Water was quickly removed from the liquid steam, and then dry extracts of raw materials from plants were obtained. In this step, the addition of maltodextrin is considered as a common carrier to enhance the dispersion and extend the storage time. Trough a series of manufacturing processes, herbal plants were then pressed into formula granules with additives. Tis step of adding excipients was not included in the experiment. Generally speaking, our production process includes extraction, concentration, and spray drying, as described in Section 2.2, in parallel with a formula granule production process. In order to produce these semifnished products, the above three steps must be followed. Te procedure of forming WEs involves concentration and spray-drying, which easily cause the loss of thermosensitive components, but HRs are obtained after extraction and drying of HMs. We wonder whether it is possible that active components remain in HRs. According to our results, the content of verbascoside reduced signifcantly from HMs to WEs and HRs (p < 0.05 and p < 0.01, respectively). Te thermal stability of verbascoside is investigated by monitoring the changes in the peak area through HPLC during the heating process. After heating for 4 h, 41.6% of verbascoside is left. It indicates that verbascoside is thermosensitive [31]. Isoacteoside, tubuloside A, and echinacoside in WEs remained stable after complex processing procedures. During the long-term drying process, the accumulation of PhGs showed a signifcant decrease, which might be attributed to the thermal degradation of these thermosensitive components [32]. In terms of the other target components, HRs and WEs did not difer signifcantly except for verbascoside. Our understanding of this diference will enable us to develop better quality standards for herbal dregs in the future and advance them into products.

Antioxidant Activity Test Results.
Te antioxidant activities of the various forms of C. tubulosa were determined using the DPPH, O •− 2 , and OH• scavenging capacity assays, and the relevant results are presented in Figure 5. In Table S8, the ranges for the DPPH, O •− 2 and OH• scavenging capacity assay results were 0.04-37.80, 0.98-843.90, and 0.32-27.65 mg/mL for the three diferent forms among the 11 batches of C. tubulosa. In three antioxidant activity tests, HMs and WEs exhibited close inhibition activity, whereas HRs showed the weakest inhibition.
Te spray-dried WEs were found to exhibit signifcant activities even at low concentrations. A previous report indicated that a spray-driedVernonia amygdalina WE achieved 50% scavenging inhibition at 0.17 mg/mL [33]. Te application of long extraction times and high temperatures is a double-edged sword. On the one hand, increasing the extraction time and spray drying inlet temperature improves the yield and efciency. Moreover, the extracts achieve strong antioxidant activity and higher concentrations of biological components than those plants [34]. On the other hand, excessively hot inlet air degrades the bioactive compounds. Such elevated air inlet temperatures led to losses of antioxidant Bidens pilosa extract activity and were attributed to decreases in phenolic compounds [35]. Present results are consistent with the aforementioned report. For instance, the WE in S6 exhibited weaker radical inhibitory abilities than both HM and HR. Furthermore, HR in S5 exhibited stronger DPPH and superoxide anion scavenging abilities than HM and WE. Te structure of PhGs consists of glycosidic bonds and acetyl groups that are hydrolyzed easily under enzymatic action or decomposed at high temperatures. Tese reactions may account for decreases in some main components during large-scale production. However, the hydrolysis or isomerization of certain components might accelerate the synthesis of other components. Such transformations are common when processing Cistanches herbs [36][37][38]   being water-soluble implies that most biological components can be utilized via water extraction. Te contention that the majority of the active components remain in WEs has persisted for decades, so it seems reasonable to assume that the wet residual materials can be discarded after extraction. However, it is incorrect to regard HRs of C. tubulosa as waste. Researchers point out that PhGs are unstable, and they are susceptible to enzymatic or hydrolytic degradation [39]. Hydrolysis or isomerization reactions that contribute to decreases in biological ingredients within phytomedicines during processing might at the same time present new opportunities for exploiting HRs. By converting traditional extraction methods, medicinal residues can be developed and utilized more efectively. Enzymatic hydrolysis was performed to convert the Panax ginseng residue into monosugars. Yields of polysaccharides and ginsenosides increased, such as sugar, succinic acid, ginseng polysaccharides, and ginsenosides [40]. Sophora favescens residues are reextracted by ultrasonic waves with ethyl acetate [41]. Te updated technologies for utilizing herbal residues are summarized by Huang et al. [42].

Spectrum-Efect Relationship.
Te spectrum-efect relationships between chromatographic peaks and antioxidant abilities were revealed using PLSR (regression equations obtained using the PLSR model can be seen in Supplementary Materials) and BCA models. Te heatmap diagram was drawn to visualize the relationship ( Figure 6). Te relationship values and ranks are listed in Table 4.
Based on the PLSR and BCA results, the top fve peaks of diferent forms were screened using the DPPH, superoxide anion, and hydroxyl radical scavenging assays to identify the most important peaks. Te results are illustrated in the Venn chart (Figure 7). A 2 , A 6 , A 8 , and A 10 are the common peaks that are shared by HM, WE, and HR (Figures 7(a) and 7(c)) in the superoxide anion and hydroxyl radical scavenging assays, whereas HM, WE, and HR share no DPPH assay peaks. Meanwhile, the BCA models show that A 1 , A 2 , A 3 , and A 6 are the common peaks shared by HM, WE, and HR (Figures 7(d)-7(f )). Notably, the overlaps in the Venn diagram indicate that the BCA model appears more suitable than the PLSR model, the former exhibiting more repetition. Te BCA model coefcients and antioxidant ability IC 50 values were analyzed via RDA. As the RDA shown in Figure 8, A 1 , A 3 , and A 6 from HM and HR are related positively to the antioxidant indexes, except that A 3 is related negatively to the hydroxyl radical scavenging capacity. A 1 and A 6 from WE have strong correlations with DPPH and the superoxide anion. Te A 6 peaks noted from the various forms exhibit the strongest connection to the DPPH, superoxide anion, and hydroxyl radicals. A 1 and A 3 also exhibit a similar connection.

Construction of C-T Network.
A total of 4359 targets related to the antioxidant activity were obtained from the GeneCards database and the OMIM database. At the same time, active components were screened from the TCMSP database and the SwissTargetPrediction database. Ten, 198 targets were collected and standardized through the UniPort database. Tere were 159 target genes shared by active components and antioxidant-related diseases (see Figure S1). Te C-T network was constructed to illustrate the correlation between the compounds and the key gene targets (Figure 9).

Construction of the PPI Network and Screening of Key
Targets. PPI was visualized using the STRING database ( Figure 10). Te network included 159 nodes and 2528 edges. In the entire interaction network, the connecting components or the nodes with more target points may be the key component or target gene that plays an antioxidant role in C. tubulosa. Te results were downloaded and introduced into Cytoscape for visualization. Te higher the DC value,      Figure 4: Content determination of fve components from diferent forms (n � 11). * p < 0.05, * * * p < 0.001, ns: not signifcant. the darker the color, and the larger the combined score value, the thicker the edge. We found that RAC-alpha serine/ threonine-protein kinase (AKT1), interlukin-6 (IL6), tumor necrosis factor (TNF), and vascular endothelial growth factor A (VEGFA) were centrally located (Figure 11), indicating that they were key targets when active components exerted an antioxidant efect. It is reported that echinacoside reduces mitochondrial dysfunction via regulation of mitogen-activated protein kinases (MAPK) and AKT and their phosphorylated forms [43]. Researchers speculated that the antidiabetic efect of glycosides of C. tubulosa might be due to the antioxidant activity of PhGs by downregulating proinfammatory cytokines, such as IL-6 and TNF-α [44]. In addition, echinacoside could impair ovarian cancer cell growth by downregulating the expression of VEGFA to inhibit angiogenesis [45], which is closely correlated to the ROS system for ROS induces the expression of VEGF signaling [46].

Enrichment Analysis and C-T-P Network
Establishment. Te potential antioxidant compounds acted on numerous biological functions, including BP, CC, and MF. In Figure 12(a), the top 10 pathways are shown. Te predicted targets from the PPI network mainly responded to many biological processes, such as organic cyclic compounds, xenobiotic stimulus, inorganic substances, oxygen levels, and positive regulation of the cellular component    movement. Te cellular component analysis showed that the genes were mainly related to the membrane raft, extracellular matrix, secretory granule lumen, transcription regulator complex, and apical part of the cell. Tese targets are also involved in many molecular functions, including DNAbinding transcription factor binding, protein homodimerization activity, protein domain-specifc binding, and cytokine receptor binding.
To investigate the biological functions of these major hubs, a pathway enrichment analysis was conducted. From KEGG enrichment results, a bubble diagram was drawn to show top 20 pathways. Te larger the spot was, the more genes were included in the pathway. As shown in Figure 12(b), the key pathways of C. tubulosa were related to pathways in cancer, lipid and atherosclerosis, AGE-RAGE signaling pathway in diabetic complications, chemical  carcinogenesis-receptor activation, and MAPK signaling pathway. Efects of C. tubulosa on apoptosis and cellular redox homeostasis were investigated. Te data suggest that C. tubulosa can be a promising candidate for anti-coloncancer therapy [47]. C. deserticola extract is found in aged people [48]. Figure 13 illustrates the correlation between the pathways and their related targets and the relationship between the overlapping target genes and biologically active components of C. tubulosa. A global view of the C-T-P network was generated, which consisted of 12 ingredients, 159 targets, and 20 pathways. Most of the targets were shared by the candidate active compounds. Tese candidate active ingredients with high interconnection degrees were responsible for the high interconnectedness of the C-T-P network, especially quercetin (degree � 131). Te majority of the targets, such as AKT1, IL6, TNF, and VEGFA, were mapped to KEGG pathways associated with pathways in cancer.

Conclusions
In this study, we primarily probed complex situations when considering the spectrum-efect relationships among HM, WE, and HR of C. tubulosa. Te HPLC fngerprints and antioxidant assays were used to identify the diferences between Hs, WEs, and HRs of C. tubulosa. According to the HPLC fngerprints, 11 peaks were common among the 11 batches of Hs, WEs, and HRs. Geniposidic acid, echinacoside, verbascoside, tubuloside A, and isoacteoside were identifed among these peaks. Te contents of these fve components were determined. In addition, the antioxidant efects of the C. tubulosa Hs, WEs, and HRs varied due to the alterations in the chemical compositions caused by complex manufacturing conditions. Based on diversifed statistical models, the spectrum-efect relationship study indicated that peak A 6 might be the most decisive component among the three forms of C. tubulosa. Te study was based on network pharmacology to explore potential mechanisms of C. tubulosa on antioxidation through screening of compounds, prediction of key targets, construction of networks, and conduction of enrichment analysis. Our results provide a theoretical basis for recycling the herbal residues and the potential of C. tubulosa in the treatment of antioxidantrelated diseases.