An Integrated Strategy of Chemical Fingerprint and Network Pharmacology for the Discovery of Efficacy-Related Q-Markers of Pheretima

Pheretima, one of the animal-derived traditional Chinese medicines, has been wildly used in various cardiovascular and cerebrovascular diseases, including stroke, coronary heart disease, hyperlipidemia, and hyperglycemia. However, it was still a big challenge to select the quality markers for Pheretima quality control. The fingerprint and network pharmacology-based strategy was proposed to screen the efficiency related quality markers (Q-Markers) of Pheretima. The ratio of sample to liquid, ultrasonic-extraction time, temperature, and power were optimized by orthogonal design, respectively. The chemical fingerprint of forty batches of Pheretima was established, and six common peaks were screened. The network pharmacology was used to construct the Pheretima-Components-Targets-Pathways-Stroke network. It was found that six potential efficacy Q-markers in Pheretima could exert the relaxing meridians effect to treat stroke through acting on multiple targets and regulating various pathways. A simple HPLC-DAD method was developed and validated to determine the efficacy Q-markers. Grey relational analysis was used to further verify the relation of potential efficiency related quality markers with the anticoagulation activity of Pheretima, which indicated that the contents of these markers exhibited high relationship with the anticoagulation activity. It was concluded that hypoxanthine, uridine, phenylalanine, inosine, guanosine, and tryptophan were selected as quality markers related to relaxing meridians to evaluate the quality of Pheretima. The fingerprint and network pharmacology-based strategy was proved to be a powerful strategy for the discovery of efficiency related Q-markers of Pheretima.


Introduction
Stroke remained the foremost cause of disability and the third leading cause of mortality all over the world. e demand for stroke prevention and recovery services was growing [1]. Stroke was usually caused by various risk factors. One was direct factors, including blockage or bleeding of the anterior cerebral artery or cerebral artery, and the other was indirect causes such as smoking, drug abuse, hypertension, obesity, and so on [2,3]. Stroke could induce cerebral ischemia. ereafter, the vascular endothelial cells release generous inflammatory factors and reactive oxygen species (ROS). en neurocyte damage would occur rapidly and result in facioplegia and paralysis [4,5]. Aspirin and clopidogrel were often used to treat stroke [6,7]. However, long-term oral administration might cause gastrointestinal discomfort, dizziness, nausea, and many other side effects [8,9]. Compared with chemical medicines, traditional Chinese medicines (TCMs) have been broadly used in many countries for centuries due to the significant therapeutic effect and fewer side effects. ere were many types of TCMs used for stroke treatment, such as Pheretima, Chuanxiong rhizome, Rhei radix et rhizome, and so on [10].
Forty batches of Pheretimasamples were purchased from medicine markets and pharmacies in different provinces in China.
e Pheretima samples were dried at 60°C under reduced pressure before being pulverized into homogeneous powder (through a 65-mesh sieve).
Although the detailed information on Pheretima species was not identified because they have been processed into decoction pieces. e samples were authenticated as Pheretima according to Chinese Pharmacopeia by Prof. Yanxu Chang (Tianjin University of Traditional Chinese Medicine).

Preparation of Standard Stock Solutions, rombin Solution, and Fibrinogen Solution.
Eight reference substances were accurately weighed and dissolved with 0.05 M sodium hydroxide (hypoxanthine at the concentration of 1 mg/mL), 0.05 M hydrochloric acid (adenine and guanosine at the concentration of 1 mg/mL), and water (uridine, tryptophan, and adenosine at the concentration of 1 mg/mL, inosine and phenylalanine at the concentration of 2 mg/mL), respectively. e standard stock solutions were mixed and diluted to different concentrations by water to establish standard curves. All standard stock solutions were stored at 4°C before being used.
rombin and fibrinogen were dissolved in normal saline (1000 U/mL) and PBS (5 mg/mL), respectively. e thrombin solution was stepwise diluted to 40 U/mL by normal saline. All thrombin and fibrinogen solutions were stored in ice bath before being used.

HPLC-DAD Analysis.
ermo Scientific Ultimate 3000 HPLC system consisting of a ternary pump, autosampler, thermostated column compartment, and diode array detector (DAD) was employed throughout the analysis. All separation was performed on a Welch AQ-C18 HPLC column (4.6 mm × 250 mm, 5 μm). e flow rate was set at 0.8 mL/min, and the column temperature was fixed at 35°C.

Optimization of Ultrasonic-Assisted Extraction.
Single-factor experiments were firstly employed to define the rang of each specific parameter in the ultrasonic-assisted extraction. Pheretima powder (20 mg/mL) was initially used as the object for the optimization of related conditions. Four single-factor experiments were sequential implemented and repeated three times. e ratio of sample to liquid, extraction temperature, power, and time was optimized, respectively. e relevant parameters of the extraction program were initially established and employed in the following study.
Considering the interaction of different factors, a fourfactor and three-level (L 9 3 4 ) orthogonal design was proposed correspondingly for ultrasonic-assisted extraction. Taking total content of analytes as index, a ratio of sample to solvent (A, mg/mL), extraction time (B, min), temperature (C,°C) and power (D, W) was tested in the orthogonal design. Based on the water as the extraction solvent, the experiments were performed. e experimental scheme and orthogonal header design are shown in Table 1.

Establishment of Pheretima HPLC-DAD Fingerprint .
e optimized method was used to extract forty batches of Pheretima samples. As a result, 100 mg samples were weighted into 10 mL volumetric flask and then soaked 10 min after adding water to the scale mark. Ultrasonic-assisted extraction (40 kHz, 220W) was performed in a water bath at 40°C for 40 min. Each extract was centrifuged at 14000 rpm for 10 min and through a 0.22 μm membrane. e subsequent filtrates were injected into the HPLC-DAD system to establish Pheretima fingerprint. e recognized common peaks were used as potential active components for the following study.

Establishment of Protein-Protein Interaction (PPI).
e common targets between potential active components and stroke were imported into STRING (Version 11.5, https://string-db.org/) and Cytoscape software (version 3.7.2) for visualization and statistical analysis.

Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) Enrichment Analysis.
For preliminarily exploring the relevant mechanism of key nodes, GO, and KEGG enrichment analyses were performed to gather the corresponding biological functions and processes via the DAVID Bioinformatics Resources (version 6.8, https:// david.ncifcrf.gov/summary.jsp). GO enrichment analysis was mainly involved in GO biological process (BP), GO molecular function (MF), and GO cellular component (CC). e result of KEGGen richment was plotted by WeiShengxin, an online platform for visualization (http://www. bioinformatics.com.cn/).

Establishment of Pheretima-Components-Targets-
Pathways-Stroke Network. All related targets and pathways International Journal of Analytical Chemistry 3 were summarized, and the network was constructed using Cytoscape software (version 3.7.2).

Anticoagulation Activity Assay.
e thrombin titration method reported in Chinese Pharmacopeia was used to evaluate the anticoagulation activity of Pheretima [11,36,37]. Briefly, 200 μL fibrinogen (5 mg/mL in PBS) and 100 μL Pheretima extract (diluted or concentrated from 200 mg/ mL) were added to 2 mL centrifuge tube. ey were shaken gently and incubated for 5 min at 37°C. ereafter, 5 microliters of thrombin (40 U/mL in normal saline) were added per minute until fibrous coagulation occurred. e consumed volume of thrombin was used to evaluate the anticoagulation activity of Pheretima. e samples were grouped according to anticoagulation activity by SPSS software (version 25). e formula of anticoagulation activity was as follows: U (U/g) represented the active unit of thrombin per Gram. C 1 (U/mL) and V 1 (μL), respectively, expressed the concentration and consumed volume of thrombin. C 2 (g/mL) and V 2 (μL) showed the concentration and dosage of samples.

Optimization of Chromatographic Conditions.
To obtain good chromatographic separation, the HPLC conditions include a concentration of additive (0, 5, and 10 mM KH 2 PO 4 ), a flow rate of mobile phase (0.6, 0.8, and 1.0 mL/ min), and detective wavelength (210, 254, and 280 nm) were comprehensively investigated. Considering the peak resolutions, shapes and responses, the best separation was achieved when 10 mM H 2 KPO 4 was selected as an additive to the water phase. e flow rate and detective wavelengths were 0.8 mL/min and 210 nm, respectively.

Optimization of Sample Extraction.
e optimized extraction program was obtained by the orthogonal experimental design. Although the optimized level of extraction time and extraction temperature was 60 min and 60°C, respectively (Table 1).
ere was no significant difference between each level. In order to protect the environment and save time and energy, 100 mg samples were weighted into 10 mL volumetric flask and soaked 10 min after water was added to the scale mark (Table 1). Ultrasonic-assisted extraction (40 kHz, 220 W) was performed in a water bath at 40°C for 40 min.

Analysis of PPI Network.
ere were 135 related targets of potential active components by PharmMapper database with Norm Fit ≥0.6. Among them, 51 targets were the same as disease targets (Figure 2(a)). To reveal the function of 51 targets against stroke, a PPI network was successfully visualized by the STRING database and analyzed by Cytoscape software. e network contained 51 nodes and 297 edges.
In view of the fact that node degree value could not only reflect the number of nodes connected to it, but also incarnated its importance. As shown in Figure2(b), the biggest node was ALB, which was serum albumin that can regulate the colloidal osmotic pressure of blood. ALB was a valuable diagnostic marker of many diseases, such as rheumatoid  arthritis, ischemia, and diseases that need monitoring of glycemic control. Clinically, ALB has been extensively used for disease treatment, including shock, hemorrhage, and hypovolemia [38]. AKT1 was one of three closely related serine/threonine-protein kinases (AKT1, AKT2, and AKT3).
It played an important role in diverse biological progress, for instance, proliferation, migration, cell growth, angiogenesis, and vasorelaxation [39]. NOS3 was the most crucial isoform of nitric oxide synthase in endothelial cells. Nitric oxide synthase (NOS) could produce nitric oxide (NO), which was an important vasodilator with a well-established role in cardiovascular homeostasis [40]. It has been confirmed that NOS-knocked mice had more severe brain damage after cerebral stroke. e mechanism might be that NO can protect brain tissue by regulating hemodynamics [41].

GO Enrichment Analysis.
After filtering by both P value and FDR (false discovery rate) less than 0.05, GO enrichment has enriched 45, 9, and 8 terms in BP, MF, and CC, respectively (Figure2(c) and Table S2). According to the rank of "count" value, BP terms were mainly related to response to the drug (GO:0042493), cellular response to lipopolysaccharide (
As numerous experimental data showed, inflammatory injury and oxidative stress were the pivotal pathological reaction of stroke. VEGF (vascular endothelial growth factors) signaling pathway was a crucial reducer in both physiologic and pathologic angiogenesis. It possessed a combined mechanism that relieved the injury on brain tissue after ischemia to regulate angiogenesis, neuroprotection, and the migration of neuronal stem cells into the ischemic area. It was noteworthy that the PI3K-Akt pathway was also enriched. PI3K-Akt signaling pathway was one of the downstream pathways of the VEGF signaling pathway. VEGF signaling pathway could promote phosphorylation of Akt, which was able to activate nitric oxide synthase (NOS). e function of NOS has already been mentioned above [42]. MAPK (mitogen-activated protein kinase) signaling pathway regulated a wide range of biological progress, such as cell growth, inflammation, and stress responses. It has been reported that Pheretima extracts could inhibit the MAPK signaling pathway through ERK 1/2 and p38 [43].

Pheretima-Components-Targets-Pathways-Stroke Network
Analysis. APheretima-Components-Targets-Pathways-Stroke network has been constructed ( Figure 3). It was composed of 79 nodes (containing 6 compounds, 51 targets and 20 pathways) and 281 edges. According to the network, six components of Pheretima could exert the relaxing meridians effect to treat stroke through acting on multiple targets and regulating various pathways.
Atherosclerosis was one of the risk triggers of stroke, and it was fatal for stroke patients as it induced hypertension and venous thrombosis. High-level of hypoxanthine and xanthine in hyperlipidemia patients were used to maintain fluid circulation homeostasis. It has been confirmed that organisms would act on the immune system and upgrade the concentrations of hypoxanthine and xanthine when a disturbance of blood circulation occurs [44,45]. Furthermore, purine nucleosides, such as hypoxanthine, xanthine, and so on, have been considered as diagnostic biomarkers of tissue ischemia [46]. In vivo, tryptophan could be metabolized to several kynurenines with antioxidant activity. It has been reported that kynurenines could scavenge reactive oxygen species (ROS) and ease oxidative stress in ischemic tissue. Kynurenic acid, one of the kynurenines, was often considered a neuroprotection agent in reducing brain damage [47]. Overall, a total of six compounds were selected as efficacy-related Q-markers of Pheretima. en, these six efficacy-related Q-markers of Pheretima were simultaneously determined by HPLC-DAD according to the chromatography conditions of a fingerprint.

Linearity and Sensitivity.
e standard curves of all analytes were contracted by plotting the peak area against concentration. eir correlation coefficients (r 2 ) were more than 0.9994. e limit of detection (LOD) and limit of quantitation (LOQ) of analytes correspond to the concentration level at signal-to-noise ratios (S/N around 3 and 10), respectively. e LOD and LOQ ranged from 0.04 to 0.44 μg/ mL and 0.13 to 1.21 μg/mL, respectively (Table 2).

Precision, Stability, and Repeatability.
e precision assay was expressed as relative standard deviation (RSD) by detecting eight compounds in Pheretima solution with six replicates. e stability was assessed with RSD by repeating the determination six times. Pheretima solution was deposited in an autosampler and analyzed at 0, 2, 4, 8, 12, and 24 h. e repeatability could be obtained by determining the target ingredients of six replicates using the configured system. As the results (Table 3), the RSDs of precision, stability, and repeatability were in the range of 0.36% to 2.85%, 1.35% to 4.35%, and 1.21% to 4.95%, respectively.

Accuracy.
e accuracy was assessed by recoveries of spiking samples. A 50 mg aliquot of Pheretima was extracted by pooling known amounts of standard solutions with the well-defined method. e formula of recovery was presented as follows: recovery(%) � spiked sample amount − sample amount adding amount × 100%.
(2) e range of eight components recoveries was from 98.1% to 102%, and RSD ranged from 1.97% to 3.28%. e detailed data is displayed in Table 3.

Quantitative Analysis.
Comparing with the reference standards, eight peaks were identified as hypoxanthine, uridine, phenylalanine, adenine, inosine, guanosine, tryptophan, and adenosine ( Figure 4 and Figure S1). e optimized ultrasonic-extraction procedure and HPLC-DAD method were triplicately performed for extracting and determining eight targets content in 40 batches of Pheretima. e concentrations of adenine and adenosine did not reach LOQ in some batches (Figure 5(a) and Table S3). It made sense that adenine and adenosine could not be recognized in common peaks of the chemical fingerprint of Pheretima. It was obvious that phenylalanine was the highest-content chemical component in most of the samples. e content of phenylalanine ranged from 508.42 ± 20.51 to 2995.55 ± 51.71 μg/g. us, phenylalanine might play an important role in Pheretima quality control.

Comparison with Other Methods.
e difference between the established and reported methods is shown in Table 4. Multiple kinds of chemical components, including anomic acid, nucleoside, and nucleobase, were simultaneously detected in one run. What's more, the derivative reagents were not used, which simplified the operating steps. e established HPLC-DAD method was an alternative method for establishing fingerprint and quantification of efficiency Q-markers of Pheretima.

Anticoagulation Activity.
rombin and fibrinogen played key roles in the coagulation cascade. e former could induce platelet aggregation and fibrin formation [53]. However, the therapeutic basis of the Pheretima anticoagulation effect was ambiguous. e ability to inhibit thrombin-induced fibrinogen activation was determined. Quality control (QC) sample replacing the extract with water was used to avoid false positive results. After adding three drops of thrombin, fibrin appears in QC. erefore, the sample was considered to have no anticoagulation effect if the number of drops was no more than three. Finally, the anticoagulation potency unit of all of the samples was determined (Table S3). e anticoagulation activity of Pheretima ranged from 0 to 20800 U/g. e anticoagulation activity of each sample was greatly various. It indicated that the anticoagulation activity-based quality control method of Pheretima was urgent to be established.

Screen Q-Markers with Anticoagulation by Grey
Relational Analysis. Considering 6 components recognized in the HPLC fingerprint chromatogram, grey relational analysis was used to clarify the relation between component contents and total anticoagulation activity. e results showed that the correlations of hypoxanthine, uridine, phenylalanine, inosine, guanosine, and tryptophan are 0.87, 0.90, 0.71, 0.72, 0.92, and 0.91, respectively. It indicated that the contents of uridine, guanosine, and tryptophan possess a high correlation with the anticoagulation International Journal of Analytical Chemistry effect. According to anticoagulation activities, samples were divided into two groups by cluster analysis. e quantitative matrix was subjected to WeiShengxin (http:// www.bioinformatics.com.cn/) to afford heatmap for adding direct comparison. As shown in Figure 5(b), comparable distribution patterns appeared between "A" and "B" major appearances were found in four components, namely uridine, phenylalanine, guanosine, and tryptophan, which indicated that they had a positive correlation between content and activity.
In order to explore the anticoagulation contribution of these screened six compounds, the mixture reference    1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39 S25  S27  S28  S29  S1  S2  S3  S4  S5  S6  S7  S8 S9 S10 S11 S12 S13 S14 S15 S16 S18 S20 S21  S22  S23  S24  S26  S30  S31  S32  S33  S34  S35  S36  S37  S38  S39  standard (10-fold concentration of sample S27) was used to detect anticoagulation activity. It was found that the mixture reference standards solution showed no anticoagulation activity and no contribution to the total anticoagulation activity of Pheretima. Some compounds with direct inhibitory fibrinogen activity in Pheretima extract solution needed to be further clarified. It was reported that lumbrokinase was the representative and mainly anticoagulation enzyme. e contents of lumbrokinase in Pheretima could be determined by a thrombin-induced fibrinogen activation assay, which is similar to the method used in this paper [54,55]. Based on the facts, there were no quality markers that were related to efficacy. e results of grey relational analysis and heatmap indicate that these six compounds had a positive relationship with the anticoagulation activity of Pheretima. us, these six compounds were selected as indirect markers of anticoagulation activity.

Conclusion
In the present work, an efficient and simple method, chemical fingerprint coupled with network pharmacology, has been established for screening potential efficacy-related quality markers of Pheretima. Six ingredients (hypoxanthine, uridine, phenylalanine, inosine, guanosine, and tryptophan) were highlighted as Pheretima efficacy-related Q-marker owing to the high positive correlation with anticoagulation activity. Furthermore, it could be concluded that Pheretima works against stroke presumably by regulating various targets and pathways, which mainly involved anti-inflammatory and antioxidation. To sum up, the integrated strategy of chemical fingerprint and network pharmacology was meaningful to discover the efficacy related Q-markers and improve the quality control of Pheretima.

Data Availability
All the datasets presented in this study are included in the article.

Conflicts of Interest
e authors declare that they have no conflicts of interest.  Figure S1. e extract DAD spectra of eight compounds in mixed reference standards (A) and Pheretima (B). Table S1. e similarity indexes between each sample and control chromatogram. Table S2. e annotation of GO enrichment. Table S3.