Serum Pharmacochemistry Analysis Combined with Network Pharmacology Approach to Investigate the Antiosteoporosis Effect of Xianlinggubao Capsule in vivo

Abstract Xianlinggubao capsule (XLGB) is a traditional Chinese medicine multi-component herbal prescription and has been widely used in osteoporosis (OP) treatment. However, the underlying anti-OP mechanisms of XLGB have not been fully studied. In this study, an ovariectomized rat model of OP was established. The OP rats were orally administrated with XLGB, and then the main absorbed components in serum sample were assessed based on liquid chromatography-tandem mass spectrometry (LC-MS/MS). Subsequently, the potential anti-OP markers in XLGB were screened based on a network pharmacology strategy. Molecular docking analysis was used for confirmation. LC-MS showed 22 absorbed components in the serum sample of OP rat with XLGB treatment. Network pharmacology and pathway analysis suggested 19 potential anti-OP markers in XLGB. According to molecular docking process, most of the potential markers displayed strong interactions with the 22 absorbed components mentioned above. Besides, an absorbed component–potential marker–pathway network was further established. In conclusion, our data suggested the possible mechanisms for XLGB in OP treatment, in which the “multicomponents, multitargets, and multipathways” participated. Our article provided possible direction for drug discovery in OP and could help for exploring novel application of XLGB in clinical setting.


Introduction
Osteoporosis (OP) is a complex bone disorder with a genetic background, and is characterized by progressive deterioration in bone tissue, which results in a decrease in bone mass, quality, and strength, and an increase in bone fracture risk. 1,2 It has been estimated that the number of annual OPrelated fractures in the United States will increase from 2 to 3 million in the period from 2005 to 2025, and the associated acute and long-term medical care cost will increase from $17 billion to $25 billion. 3 The prevalence and high medical costs of OP also remain true in developing countries including China. According to epidemiological investigation, China's OP incidence rate in 2008 was 14.94%; however, during 2012 to 2015, it increased up to 27.96%, 4 with, the direct medical costs of hip fracture in 2006 reaching approximately US$ 1.5 billion. 5 It is obvious that with the rapid increase in aging population, OP incidence in developing countries will increase more and more significantly in the future. Evidence suggests that by 2050, the number of hip fractured in Asia will be increase by 50%. 6,7 Given above, OP is a growing global public health problem with substantial medical, social, and economic burden.
Chinese herbal medicines play an important role in the preclinical and clinical studies of OP. 8 Xianlinggubao formula was designed based on modification of the empirical "Miao minority" medicine, which was commonly used to tone the "kidney system" and nourish bones. Xianlinggubao capsule (XLGB) was officially approved and sold as the over-the-counter (OTC) drug in China (CFDA, Z20025337). It is the only traditional Chinese medicine (TCM) prescription that was used for the prevention and treatment of OP on the National Health Insurance List in China. Its safety and efficacy have been proven by the principle of evidence-based medicine. 9 Preclinical studies showed that XLGB could improve bone mineral density and mechanical strength in ovariectomized (OVX)-induced OP in old rats. 10 However, the active compounds and the underlying mechanism of XLGB have not been fully understood.
As a TCM Fufang (multi-component) prescription, XLGB consists of six commonly used Chinese herbs with their percentages in weight as follows: Herba Epimedii (Epimedium brevicornu Maxim, Yinyanghuo) (70%), Radix Dipsaci (root of Dipsacus asper Wall ex Henry, Xuduan) (10%), Rhizoma Anemarrhenae (rhizome of Anemarrhena asphodeloides Bunge, Zhimu) (5%), Radix et Rhizoma Salviae (root and rhizome of Salvia miltiorrhiza Bunge, Danshen) (5%), Fructus Psoraleae (fruit of Psoralea corylifolia L., Buguzhi) (5%), and Radix Rehmanniae (root of Rehmannia glutinosa Libosch, Dihuang) (5%). 11 Chemical analysis showed that there are a large number of compounds with different structures in this formula, 12 which makes the study of the anti-OP mechanism of XLGB a great challenge. In 2020, Bao et al used the network pharmacology, a popular method for investigating the underlying molecular mechanisms for TCM in recent years, to systematically study the possible therapeutic mechanisms of XLGB in OP. 13,14 It is well known that XLGB, an oral medicine, plays roles only if it is absorbed into the blood, thus, investigating the anti-OP mechanisms of XLGB at the molecular and system levels in vivo remains urgent.
In this study, we identified absorbed compounds of XLGB in the serum sample of OVX rats through a pharmacochemistry test. Potential markers were screened by network pharmacology analysis based on protein-protein interaction (PPI) data and KEGG (Kyoto Encyclopedia of Genes and Genomes) enrichment. Molecular docking was employed to investigate the strength of interaction between absorbed components and potential markers. Furthermore, an absorbed componentpotential marker-pathway network was established. The detailed flowchart of this work is depicted in ►Fig. 1. Our data first and systematically demonstrated the therapeutic mechanisms XLGB in an in vivo study.

Establishment and Treatment of OVX-Induced OP in Rats
Female Sprague-Dawley rats (200-220 g) were anesthetized with pentobarbital sodium (40 mg/kg, intraperitoneal), and their ovaries were removed bilaterally. Thirty days later, the indexes of bone density were measured by X-ray absorptiometry. Serum concentrations of osteocalcin, N-terminal propeptide of type I procollagen (PINP), alkaline phosphatase (ALP), and tartrate resistant acid phosphatase (TRAP) were estimated using an enzyme-linked immunosorbent assay kit following the reported study to assess the successful establishment of OVX-induced OP in rat models. [15][16][17] OVX rats were divided into two groups: XLGB and control groups. Rats in the XLGB group (n ¼ 5) were intragastrically treated with XLGB (300 mg/kg daily, Lot No. 1611016, GuizhouTongji Tang Pharmaceutical Co., Ltd.) for 50 consecutive days. Rats in the control group (n ¼ 5) were treated with physiological saline in the same way as that in the XLGB group. All experiments were approved by the Laboratory Animal

Identification of the Absorbed Components
On the 50th day, all rats were anesthetized by intraperitoneal injection of 10% aqueous chloral hydrate. After 2 hours, the blood samples from the hepatic portal vein were collected, centrifuged at 14,000 rpm for 15 minutes at 4°C, and then mixed to produce the pooled plasma.
Serum pharmacochemistry analysis was performed on a Waters UPLC (ultra-performance liquid chromatography) system using an ACQUITY UPLCHSS T3 column (2.1 Â 100 mm, 1.7 μm) with a security guard column of the same material at a flow rate of 0.4 mL/minute. The mobile phases A and B were 0.1% formic acid in water and acetonitrile, respectively. The gradient program was as follows: 0-0. 15  Mass spectra were acquired in both positive and negative ion modes by using a Waters definition accurate mass quadrupole time-of-flight (Q-TOF) Xevo G2-XS mass spectrometer (Waters MS Technologies, United Kingdom) equipped with an electrospray ionization source. The optimized operating parameters were set as follows: mass range, m/z: 50-1,500; the flow rate of drying gas (N 2 ): 800 L/h; drying gas temperature: 400°C; cone gas flow: 100 L/h; source temperature: 120°C; capillary voltage: 2.5 kV; cone voltage: 40 V; in MSE mode, the energies for collision induced dissociation were 6 V for the precursor ion at low energy mode and 30 to 60 V for fragmentation information at high energy mode. An external reference (Lock-Spray) consisting of a 0.

Potential Marker Screening
Composite Compounds and ADME (Absorption, Distribution, Metabolism, Excretion) Screening of XLGB The chemical composition of all the six herbs of XLGB was obtained from the Traditional Chinese Medicine System Pharmacology Database (TCMSP, http://lsp.nwu.edu.cn/ tcmsp.php) and TCM database@Taiwan (http://tcm.cmu. edu.tw/index.php). Pharmacokinetic properties of each compound, including the prediction of oral bioavailability (OB), intestinal epithelial permeability (Caco-2 cells), drug likeness (DL), blood-brain barrier (BBB), drug half-life, and Lipinski's rule (LR) of five 18 were assessed for screening and evaluation of the target compounds. The selected candidate molecules should satisfy the criteria: OB ! 30%, DL ! 0.18, Caco-2 > À 0.40, suggested by the TCMSP database 19 and LR of five. Besides, to discover the active ingredients as much as possible, we searched a large number of texts and selected some main ingredients of these herbs or compounds with pharmacological activity in our prescreening process to supplement the compound library.

Compound Targets for XLGB
The potential targets of the predicted active compounds of XLGB were explored from TCMSP. Besides, SWISS database (http://www.swisstargetprediction.ch), STITCH database (http://stitch.embl.de), and target prediction system (https://prediction.charite.de/subpages/target_prediction. php) were further used for prediction for those that failed to be collected in TCMSP. UniProt (http://www.uniprot.org/) serves as a central hub for the collection of functional information on proteins, and has the advantages of accurate, consistent, and rich annotation. In this study, UniProt was applied to resolve nonstandard naming.

Kernel Targets for XLGB against OP
System biology studies show that genes and proteins are interconnected and the PPI networks are usually used to understand the role of various proteins in complex diseases. PPI data can be obtained using the Cytoscape plugin Bisogenet. In this study, the intersection of compound target PPI data and OP target PPI data was established to study the anti-OP effect of compounds, and the kernel targets were screened with node degree and Cytoscape plugin CytoNCA.

Enrichment Analysis and Potential Target Screening
The Database for Annotation, Visualization and Integrated Discovery (DAVID; https://david-d.ncifcrf.gov, ver. 6.8) was applied. 20 Pathway enrichment was performed through inputting the kernel genes into DAVID.

Molecular Docking
The structure of the absorbed compounds was downloaded from PubChem (https://pubchem.ncbi.nlm.nih.gov/). In this study, three-dimensional structures of the 19 target peptides were derived from the RCSB Protein Data Bank (http://www.rcsb.org/pdb/home/home.do). The docking calculation was performed using the software DOCK (6.7) on Yinfo Cloud Computing Platform, a 6 server for biomedicinal, material, and statistical researches (http://cloud. yinfotek.com). 21 The best ranked docking pose of the peptide in the active site of each compound was determined according to the grid scores.

Network Construction
All the networks can be created via utilizing the network visualization software Cytoscape (http://cytoscape.org/, ver. 3.4.0). 22 This is a software that applies to visualizing biological pathways, intermolecular interaction networks, etc. Meanwhile, it also supplies a basic set of features for data integration, analysis, and visualization for complicated network analysis.

Results
Pharmacochemistry Analysis of XLGB in Serum Sample of OP Rats

Successful Establishment of OVX Model of OP in Rats
In this article, the OVX-induced OP model in rat was established. ►Fig. 2 shows that bone density and osteocalcin in model rats decreased significantly (p < 0.05 and p < 0.01 vs. normal, respectively), while serum levels of PINP, ALP, and TRAP increased significantly (p < 0.01 vs. normal), demonstrating the successful establishment of the OVX model of OP in rats.

Identification of the Absorbed Components
According to the UPLC-QTOF MS analysis, 22 prototype components were identified from OVX rat plasma after oral administration of XLGB (►Fig. 3 and ►Table 1).

Candidate Compound Screening for XLGB
A total of 520 ingredients were obtained from TCMSP, and 83 candidate compounds were retained after LR and ADME screening; 28 compounds of Fructus Psoraleae were gained from TCM database@Taiwan. In addition, 15 compounds, either with extensive pharmacological activities or were the typical components of six herbs but failed to meet LR or the ADME criteria, have been supplemented. In total, 126 candidate compounds were determined and are listed in ►Supplementary Table S1 (online only).

Identification of Common Targets for XLGB against OP
A total of 328 putative targets for 124 candidate compounds were collected from the aforementioned database, while 285 OP-related targets were collected, of which 94, 215, 10, and 24 from OMIM, GAD, pharm-GKB, and TTD, respectively.

Pathway Enrichment for Kernel Targets and Potential Marker Screening
First, an intersected network was constructed between a compound putative target PPI network and an OP-related target PPI network to unravel the pharmacological mechanisms of XLGB against OP. Further, based on the plugin named CytoNCA, 23 343 kernel targets were identified. We then input all kernel targets into the DAVID to enrich OP-associated pathways, and 31 pathways with p 0.05 were obtained. They were performed using the OmicShare tools (http:// www.omicshare.com/tools) and are shown in ►Fig. 4.
The 31 pathways containing 104 of the 343 targets might be the key pathways for the anti-OP mechanism of XLGB. Comparing the kernel targets of the pathway with the targets of 124 candidate compounds, we found that 19 kernel targets were overlapped (►Table 2), which might be considered as potential markers and will function in the anti-OP process of XLGB.
Then, a compound-target-pathway network was constructed as shown in ►Fig. 5, in which 105 compounds from six herbs of XLGB (►Supplementary Table S1, online only) acted on 28 pathways (►Table 3) through 19 potential markers.

Docking Results
Molecular docking analysis was performed to investigate the interaction between the 22 absorbed compounds and 19 potential markers. Proteins without detected crystal structures were excluded from the analysis. Grid scores of the docking results are showed in ►Supplementary Table S2 (online only). Grid scores less than À80 kcal/mol were considered to be strong interaction. Our article suggested 14 potential markers with 13 absorbed components, among which 9 absorbed components were from Herba Epimedii, 3 absorbed components were from Radix Dipsaci, and 1 absorbed component was from Rhizoma Anemarrhenae.

Absorbed Component-Potential Marker-Pathway Network
A network was constructed to show the interaction among main absorbed components, potential markers, and pathways. ►Fig. 6 demonstrates 13 absorbed components acted on 14 potential targets that connected to 28 pathways.

Discussion
It is well known that compounds will function on the premise that they are absorbed into the body. Thus, in this study, we investigated the interaction between screened potential markers and absorbed components to illuminate the anti-OP effect of XLGB in an in vivo study.
The clinical specification of XLGB is 0.5 g/piece following the instructions described. Adults take three pieces every time, twice a day. We first suggested an oral dose of 300 mg/kg/day of XLGB for an OVX rat model of OP, which was equivalent to the corresponding clinical prescription dose for a 60 kg human subject and would make sure the results of our research study are as close as possible to clinical use. Then, we identified 22 main prototype components of XLGB in the serum sample of an OVX rat model of OP. This number was smaller than that given in Geng et al's report based on a healthy mice 24 (57 prototype components). This may be resulted from animals in different conditions (normal and pathological). The TCM theory supports that the monarch and minister drugs may play a major role in a fufang prescription. In this article, we found that 11 of the 22 compounds are from Herba Epimedii (a monarch drug in the prescription), 7 from Radix Dipsaci, and 3 from Fructus Psoraleae, both of which are the minister drugs of XLGB.
The PPI networks were applied to investigate the relationship between compound targets and disease genes. We screened our potential markers based on network pharmacology analysis with PPI data and the enrichment analysis of KEGG pathways. We suggested a total of 124 candidate active components of XLGB with 328 putative targets, among which 286 target genes were related to the development of OP. Enrichment analysis of KEGG pathway discovered 343 kernel targets and 31 potential pathways (p < 0.05), and 19 kernel targets from 28 pathways overlapped with the target compound, confirming their roles in anti-OP effect of XLGB. Then a network with 105 compounds, 19 targets, and 28 pathways was obtained, in which 105 compounds covered all six herbs of XLGB. However, in the compound-target-pathway network, only three compounds were identified in the plasma sample of OVX-OP rats, and this may be attributed to the very low dose of some compounds that could not be detected in this prescription. 12 Moreover, molecular docking analysis showed 13 compounds (epimedin A, epimedin B, epimedin C, diphylloside B, diphylloside A, icariin, timosaponin B-II, asperosaponin VI, akebiasaponin V, hederagenin-β-sophorosyl ester, olivil-4''-O-β-D-glucopyranoside, 2''-O-rhamnosylicariside II, and 3′'-O-Xylopyranosyl epimedoside A) exhibited strong interactions with 14 potential markers (ABL1, CDK2, EGFR, HSP90AA1, HSP90AB1, HSP90B1, MAPK1, NFKB1, NFKB2, NOS2, RB1, SRC, TP53, and VCAM1). Nine of these compounds are from the monarch herb Herba Epimedii, three are from the assistant herb Radix Dipsaci, and one is from the minister herb Rhizoma Anemarrhenae. Our data suggested that the main active compounds in the monarch herbs are flavonoids and saponins, which make up the most important therapeutic material basis in XLGB. As shown in the absorbed component-potential marker-pathway network, most of the compounds could regulate multiple signaling pathways via multiple targets. For example, epimedin A, epimedin B, and epimedin C (from Herba Epimedii) acted on NOS2, CDK2, EGFR, and MAPK1, among which MAPK1 and EGFR are kernel targets in this network, and connect with 22 and 9 signaling pathways, respectively. Timosaponin B-II (from Rhizoma Anemarrhenae) was linked to 10 targets and finally acted on 23 pathways. Icariin (from Herba Epimedii) connected with 21 pathways through four targets, and has been reported to promote osteogenesis through different pathways including EGFR, 25 MAPK, 26 and PI3K/ATP 27 signaling pathways.

Conclusions
Our study was in accordance with traditional Chinese medicine theory and the multicomponent, multitarget, and multipathway characteristics of TCM. We investigated the possible active compounds of XLGB and the underlying therapeutic mechanisms from a systemic perspective. Our data provided a strong link between the major active components and potential targets, helped to delineate the anti-OP effect of XLGB and guide of the deconvolution of therapeutic effects of multi-component herbal treatment.

Conflicts of Interest
The authors declare no conflict of interest. Fig. 6 Absorbed component-potential marker-pathway network (pink, purple, and orange circles stands for compounds of Yinyanghuo, Zhimu, Xuduan, respectively). Green V stands for potential targets. Blue triangle stands for pathways. Gray lines stand for interaction between targets and pathways. Green lines stand for interaction between targets and compounds.