Predicting the levels of orcinol glucoside during the treatment of osteoporosis by network pharmacology and molecular docking

Introduction: As a traditional Chinese medicine (TCM), Curculigo orchioides Gaertn. (Xianmao) has been widely used to treat bone-related diseases. However, the active components of this TCM, and the specic mechanisms by which it exerts effect, have yet to be elucidated. To identify potential targets for orcinol glucoside (OG), an active constituent of C. orchioides, during the treatment of osteoporosis (OP) by adopting a network pharmacology approach. Methods: First, we mined the Similarity ensemble approach (SEA), SwissTargetPrediction, DisGeNET, and Genecards databases were mined for data related to the prediction of OG- and OP-related targets. Next, we identied the common targets for OG and OP, and then used STRING software to create a protein-protein interaction (PPI) network. Then, we used topological analysis to identify which of the common targets were most signicant. Then, we used the common signicant targets and g:proler to perform gene ontology (GO) term and Kyoto Encyclopedia of Genes and Genomes ( KEGG) pathway enrichment analysis. Finally, we used molecular docking to predict the targets of OG that were most relevant to the treatment of OP and investigated the potential pharmacological mechanisms that might be involved. Results: In total, 130 potential targets of OG, and 4582 targets relevant to OP, were subjected to network analysis. There were 73 common targets; these identied the principal pathways linked to OP. In addition, topological analysis identied 14 key targets. Most of the predicted targets played crucial roles in the PI3K-AKT signaling pathway. Molecular docking identied ten core targets (VEGFA, IL6, EGFR, MAPK1, HRAS, CCND1, FGF2, IL2, MCL1 and CDK4), thus indicating that OG may promote osteoblast proliferation and differentiation by accelerating progression of the cell cycle. Conclusions: This research provides a theoretical base for identifying the specic potential mechanisms of OG in treatment of OP.


Introduction
Osteoporosis (OP) is a common metabolic bone disease that is characterized by low bone mineral density and microarchitectural changes in the bone, thus resulting in debilitating fragility and fractures [1]. OP is a global health issue that shows a close association with the aging population. Current estimates show that the population of patients with OP in China will increase by 60 million to over 120 million by the year 2050 [2]. The pathogenesis of OP is very complicated, including physical variables, nutritional status, endocrine factors, genetic factors, and a range of immune factors [3].
Numerous clinical therapies have been applied to treat OP, including calcium and vitamin D, bisphosphonates, anti-RANKL, selective estrogen receptor modulators, anabolic agents, and sclerostin inhibitors [4]. However, these drugs can only partially ease bone loss; their lasting clinical use is limited by high costs and low tolerability [5].
Traditional Chinese medicine (TCMs), such as Epimedium (Yinyanghuo), Curculigo orchioides Gaertn. (Xianmao), have been used for many years to treat OP [6]. The effects of water extract from epimedium treatment on osteoporosis can be mitigated through a mechanism associated with several neuropeptides that regulate the brain/spinal cord/bone axis [7]. C. orchioides ethanol extract can inhibit bone absorption in rats underoing oophorectomy, increase serum phosphorus and calcium levels, and have a certain protective effect on osteoporosis; however, this product does not affect bone formation [8].
Phytochemical studies previously demonstrated that C. orchioides contains and abundance of phenols and phenolic glycosides, triterpenes and triterpenoid glycosides, lignans, lignan glycosides, and many other types of compounds [9]. OG is one of the major bioactive phenolic glycosides of C. orchioides and is reported to have a wide range of pharmacological actions in mouse models, including antiosteoporosis [10], anxiolytic [11] and antidepressive [12] effects. The speci c mechanisms underlying the therapeutic effects of OG in patients with OP, however, has yet to be elucidated. The role of OG in the treatment of OP can be better understood by detailed studies on molecular targets and related signal pathways.
With the rapid progression of bioinformatics technology, the development of network pharmacology has proven to be an innovative method for investigating the effects of TCM [13][14]; this new discipline involves network analysis, molecular docking, experimental methods, and integrates multiple information sources [15]. Therefore, network-based methods are expected to signi cantly enhance our understanding of drug effect by considering multiple sources of information [16].
In this research, we systematically explored the potential targets and molecular mechanisms of OG in the treatment of OP. First, we identi ed the molecular targets of OG and then predicted targets of disease using different types of bioinformatics platforms. Second, we selected the most signi cant common targets for OG and OP used STRING software to create a protein-protein interaction (PPI) network; Cytoscape was used to visualize the results of PPI. Third, we carried out enrichment analysis of common targets according to gene ontology (GO) terms and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways. Finally, the therapeutic effect of OG, and several key targets identi ed by topological analysis, was con rmed by molecular docking experiments.

Methods
The prediction of targets for OG PubChem (https://pubchem.ncbi.nlm.nih.gov/) is the world's largest free chemical database and offers information related to compound structures and biological activities. First, we used OG to search the PubChem database to acquire canonical SMILES strings. These strings were then sent to SwissTargetPrediction (http://new.swisstargetprediction.ch/?) and Similarity ensemble approach (SEA) (http://sea.bkslab.org/) to identify targets. In this part of the analysis, the species was set to "human" and the prediction results were collated and classi ed Screening of targets related to OP Disease-related gene screening was carried out by screening two free public databases: the DisGeNET database v6.0 (https://www.disgenet.org/) [17] and Genecards database (https://www.genecards.org/) [18] with the keyword "osteoporosis". We also used UniProtKB (https://www.uniprot.org/) [19] to acquire the names of standard targets with "homo sapiens" as the selected organism.
After collating the targets and removing any duplicates, the nal list of predicted targets were estimated to represent common targets. Next, the common targets of OG that showed relevance with regards to its effects on OP were identi ed. We were particularly interested in targets that may play an active role in the proliferation and differentiation of osteoblasts by accelerating cell cycle progression.

Construction of a visualization network
The identi ed targets were then inputted into the STRING database v11.0 (https://string-db.org/) [20] for PPI analysis, including their physical and functional associations. During this analysis, we set the score > 0.4 medium con dence, and the results were exported in tab-separated value (TSV) format. PPI results were subsequently imported into Cytoscape 3.7.2 [21] to visualize the results.
Next, we used Network Analyzer to calculate the network topological parameters by treating the network as undirected [22]. We also used the CytoNCA plugin [23], and the "without weight" method to determine the three centralities: degree centrality (DC), betweenness centrality (BC), and closeness centrality (CC).
In this study, the nodes with a DC and BC score that were greater than the median by two-fold were taken as key targets in the network. We assumed that these were the key targets for anti-osteoporosis. In this experiment, the median DC was 8, and the median BC was 15.14135.
GO and KEGG pathway enrichment analysis GO and KEGG pathway enrichment analyses was carried out for OG targets in the treatment of OP. To facilitate this analysis, we used a web server for functional enrichment analysis known as g:Pro ler (https://biit.cs.ut.ee/gpro ler/gost) [24]. This analysis was performed and visualized by OmicShare tools, a free online platform for data analysis (https://www.omicshare.com/tools/).

Molecular docking
Three-dimensional (3D) structures of the ligand was obtained from PubChem. The crystal structures of key targets were then downloaded from the RCSB Protein Data Bank (https://www.wwpdb.org/) and modi ed using PyMOL Institute) was used to analyze the binding characteristics of the ligand for each protein [26]. First, we used Autodock Tools to save the proteins and ligand in PDBQT formats [27]; the grid box parameters are shown in Table 1. Exhaustiveness was set to the value of 60 for all docking analysis. Other settings were all set as default values. After docking, we selected the lowest binding a nity score for each protein for further analysis. Ligplus [28] and PyMOL were used to visualize the interactions between OG and the key targets as two-dimensional (2D) and three-dimensional (3D) graphs.

Screening of potential targets
After the elimination of duplicates, SEA and SwissTargetPrediction databases identi ed a total of 130 predicted targets for OG (Table S1). DisGeNET and Genecards databases further identi ed a total of 4582 targets associated with OP (Table S2). The intersection between these two lists of targets identi ed 73 common targets of OG that were associated with OP.

OG-OP target network
Based on the 73 common targets, we created a PPI network by importing the genes of the common targets into the STRING database. Then, Cytoscape 3.7.2 software was used to visualize the PPI network, which consisted of 73 nodes and 461 edges (Fig. 1). According to strict criteria (twice the median of DC and BC), 14 critical nodes ( "key targets") were further identi ed, as shown in Table 2. These 14 key targets may represent valuable targets for OG in the treatment of OP.

GO terms and KEGG pathway enrichment analyses
Next, the 73 common targets were imported to the g: Pro ler for GO and KEGG analysis. GO terms and KEGG pathways with a p-value < 0.05 were considered to be notably enriched. The top 20 elements were then visualized using OmicShare tools (Fig. 2).
As shown in Fig. 2, the top ve items of biological processes (each with a p value < 0.05) included organonitrogen compound metabolic process, cell population proliferation, protein metabolic process, regulation of response to stress, and response to chemicals. The top ve items for the cellular components category included cyclin-dependent protein kinase holoenzyme complex, serine/threonine protein kinase complex, protein kinase complex, extracellular region, and extracellular space. The top ve items for the molecular functions category included catalytic activity, catalytic activity, acting on a protein, protein kinase activity, metalloendopeptidase activity, and serine-type endopeptidase activity.
KEGG analysis further showed that 54 of the 73 common targets (74.0%) were notably enriched in 51 pathways (Table S3). Among the enriched pathways, we identi ed abnormalities in terms of the PI3K-AKT signaling pathway in OP. The predicted targets that refer to the PI3K-AKT signaling pathway are indicated in red in Fig. 3.

Discussion
With the increasing incidence of OP and the poor therapeutic e cacy of clinical drugs, there is an urgent ned to develop new treatment strategies. Therefore, the search of complementary and alternative medicine has become our top priority. As a famous kidney-tonifying traditional medicine, C. orchioides has been widely used against OP. The major active constituent of OG is C. orchioides; previous research has shown that C. orchioides exerts signi cant effects against OP [10]. Here, we used network pharmacology and molecular docking approaches to investigate existing literature and con rm potential mechanisms underlying the role of OG.
We hypothesized that OG can be used as a therapeutic for the treatment of OP and that OG promotes the proliferation and differentiation of osteoblasts by accelerating cell cycle progression. We then carried out a series of investigations to test this hypothesis. Another active molecule, Curculigoside, also isolated from C. orchioides has been reported to have a similar therapeutic mechanism and exhibits anti-OP effects by inducing the proliferation and differentiation of osteoblasts [29] and by reducing the in ammatory response [30].
Enrichment analysis of KEGG pathways showed that OG targets were enriched in the PIK3-AKT signaling pathway. Previous research also showed that the activation of the PI3K-AKT signaling pathway exhibits a strong correlation with the occurrence and development of OP [31]. During the pathogenesis of OP, the PIK3-AKT pathway contributes to the progression of disease through cell survival, proliferation, differentiation, and apoptosis [32]. Therefore, targeting the PI3K/Akt signaling pathway may be a potential treatment for osteoporosis.
In this study, we rst identi ed 73 common targets for drugs and diseases that might also represent targets for OG during the treatment of OP . Based on topological analysis, we further identi ed 14 key  targets, including GAPDH, VEGFA, IL6, EGFR, MAPK1, HRAS, MMP9, CCND1, ESR1, FGF2, IL2, MCL1, CDK4,  and F2. Among the top 20 enriched KEGG pathways, the PI3K-AKT signaling pathway was shown to be particularly important as abnormalities were clearly evident in OP. It was evident that the key target genes were not exactly the same as those in the PI3K-AKT signaling pathway. Proteins that were con rmed as key targets and involving the PI3K-AKT signaling pathway were thus selected for molecular docking.
In total, ten targets were selected for the molecular docking studies: VEGFA, IL6, EGFR, MAPK1, HRAS, CCND1, FGF2, IL2, MCL1 and CDK4. Cell proliferation and differentiation can be adjusted via alterations of the cell cycle phase, thus causing indirect effects on bone formation [33]. Our molecular docking studies suggest that OG can affect the osteoblast cell cycle and has strong a nity for CCND1 (-7.0 kcal·moL − 1 ), and CDK4 (-7.8 kcal·moL − 1 ); these are key players in cell cycle progression. Consistent with this hypothesis, pharmacological studies have consistently shown that related compounds can downgrade CCND1 or CDK4 to repress osteogenic proliferation and differentiation [34][35]. Studies have also shown that VEGFA, EGFR, MAPK1, and FGF2 play a critical role in the proliferation and differentiation of osteoblast cells [36][37][38][39]. In this study, we found that OG may activate VEGFA, EGFR, and MAPK1, as their binding a nities for interaction were all <-7.0. Furthermore, IL6 and IL2 are involved not only in in ammatory responses but also in the regulation of bone mineral density, osteoclast differentiation and activation [40][41]. However, our molecular docking results for IL6 and IL2 showed binding a nities of -6.7 and − 6.1 kcal·moL − 1 ; these are higher than − 7.0 kcal·moL − 1 and therefore indicate that OG exerts only weak inhibition on in ammatory responses.

Conclusion
By combining network pharmacology and molecular docking, we showed that OG is a promising treatment for OP and acts via several key targets (VEGFA, IL6, EGFR, MAPK1, HRAS, CCND1, FGF2, IL2, MCL1 and CDK4). Via these targets, OG may promote the proliferation of osteoblast by altering cell cycle progression. However, further experimental testing is now required to investigate the speci c mechanisms underlying the action of OG on OP.

Declarations
Ethics approval and consent to paricipate Not applicable.

Consent for publication
Not applicable.
Availability of data and materials The data and materials used or analyzed during this current study are available from the corresponding author on reasonable request.

Competing interests
The authors declare that they have no competing interests.

Funding
The authors are grateful to National Natural Science Foundation of China (Grant number: 81774202), Chongqing Municipal Natural Science Foundation (Grant number: cstc2018jcyjAX0388) and the Chongqing Municipal Performance Incentive Foundation (Grant number: cstc2018jxj1130032) for nancial support.
Authors' contributions MZ designed the study. XL, MH, CY and QW carried out the experimental work., MZ, XL and MH analyzed the experimental data and wrote the paper. All authors reviewed the manuscript and approved its submission.

Figure 1
The orcinol glucoside-osteoporosis target network. The color and size of each circle re ects the node degree for the common targets.

Figure 3
Orcinol glucoside is expressed as a target in the PI3K-AKT signaling pathway. The red rectangles indicates the identi ed targets