A Study Based on Network Pharmacology Decoding the Multi-Target Mechanism of Duhuo Jisheng Decoction for the Treatment of Intervertebral Disc Degeneration

Intervertebral disc degeneration (IDD) poses a grim public health impact. Duhuo Jisheng Decoction (DJD), a traditional Chinese medicine formula, has recently received significant attention for its efficacy and safety in treating IDD. However, the pathological processes of IDD in which DJD interferes and molecular mechanism involved are poorly understood, which brings difficulties to the clinical practice of DJD for the treatment of IDD. This study systematically investigated the underlying mechanism of DJD treatment of IDD. Network pharmacology approaches were employed, integrating molecular docking and random walk with restart (RWR) algorithm, to identify key compounds and targets for DJD in the treatment of IDD. Bioinformatics approaches were used to further explore the biological insights in DJD treatment of IDD. The analysis identifies AKT1, PIK3R1, CHUK, ALB, TP53, MYC, NR3C1, IL1B, ERBB2, CAV1, CTNNB1, AR, IGF2, and ESR1 as key targets. Responses to mechanical stress, oxidative stress, cellular inflammatory responses, autophagy, and apoptosis are identified as the critical biological processes involved in DJD treatment of IDD. The regulation of DJD targets in extracellular matrix components, ion channel regulation, transcriptional regulation, synthesis and metabolic regulation of reactive oxygen products in the respiratory chain and mitochondria, fatty acid oxidation, the metabolism of Arachidonic acid, and regulation of Rho and Ras protein activation are found to be potential mechanisms in disc tissue response to mechanical stress and oxidative stress. MAPK, PI3K/AKT, and NF-κB signaling pathways are identified as vital signaling pathways for DJD to treat IDD. Quercetin and Kaempferol are assigned a central position in the treatment of IDD. This study contributes to a more comprehensive understanding of the mechanism of DJD in treating IDD. It provides a reference for applying natural products to delay the pathological process of IDD.


Introduction
Low back pain (LBP) is a signifcant cause of chronic pain worldwide. Intervertebral disc degeneration (IDD), considered to be the leading cause of LBP, is the pathological basis of multiple disc degenerative diseases (DDD) such as intervertebral disc herniation and spinal stenosis, which poses a massive burden to health and the economy [1,2]. If conservative treatment fails, spinal fusion is considered the current "gold standard" for DDD [3]. Although signifcant progress such as gene therapy, stem cell therapy, and bioengineering treatment was developed in the treatment of IDD, the potency of these new technologies is limited by the unique anatomical features of the intervertebral disc, the harsh microenvironment of the degenerative intervertebral disc (such as high glucose and oxidative stress), and the technical limitations of the technology itself [4][5][6]. More exploration is needed before the new technology can be translated into practical clinical applications, at which time drugs play an irreplaceable role in the treatment of IDD [6][7][8].
Natural products have often been used in Chinese medicine for centuries to treat many diseases. Duhuo Jisheng Decoction (DJD), a traditional Chinese formula, was considered to have the functions of nourishing the liver and kidney, activating qi, and promoting blood circulation according to traditional Chinese medicine theory [9]. It has been used to treat osteoarthritis in the past due to its antiautophagy and anti-infammatory efects [10]. Recently, several systematic reviews have stated the safety and efcacy of DJD in slowing the progression of IDD and alleviating LBP, demonstrating the value of DJD in the treatment of IDD [11,12]. A study confrmed in human degenerative nucleus pulposus cells (NPCs) in vitro that DJD inhibited the infammation of NPCs and the reduction of extracellular matrix. Also, this way of inhibiting the infammatory response may be by inhibiting the NF-κB pathway [13]. Furthermore, another study using compression-induced aging of the intervertebral disc in a rat model found that DJD can activate autophagy and signifcantly reduce apoptosis of NPCs and matrix degeneration. Further research found that DJD may lead to the corresponding biological behavior of NPCs by inhibiting the MAPK pathway [14]. Previous studies have reported that the MAPK pathway and NF-κB pathway may play a role in the treatment of IDD by DJD [13,14], but the molecular mechanism by which DJD targets the above pathways is unclear. Moreover, there are many bioactive compounds in DJD, which may involve more pathological processes and molecular mechanism, but there is currently a lack of systematic understanding of the mechanism. Network pharmacology mining the associations between drug and disease targets is a novel and promising strategy to reveal the complex mechanism of disease and identify new therapeutics [10]. Molecular docking is an important technique in the feld of computer-aided drug research, which is used to predict the afnity and binding properties of drugs to specifc targets, and has become a mature technology in pharmacological research [15]. Te present study employed network pharmacology and molecular docking techniques to investigate the specifc molecular mechanism of DJD regulating MAPK pathway and NF-κB pathway to treat IDD. It systematically explored the underlying mechanisms of DJD in the treatment of IDD, aiming to enhance a more comprehensive understanding of the mechanism of DJD in the treatment of IDD and provide valuable insights for the application of natural products in delaying IDD.

DJD Target Screening and Toxicity Prediction.
Te Traditional Chinese Medicine Systems Pharmacology Database and Analysis Platform (TCMSP, https://tcmspw.com/tcmsp. php) was adapted to search the compounds of DJD and obtain the structural parameters of these compounds. Based on the structural parameters, Lipinski's rule of fve [16] (RO5, details are provided in Table S1) and oral bioavailability (OB) were used to construct a flter to assess the oral potential of compounds: OB > 30% or meet the conditions of RO5. Moreover, other three parameters which were previously recommended were employed to screen out the compounds with higher bioactivity and drug-likeness: Caco-2 permeability > 0.4, druglike (DL) properties > 0.18, and half-life > three hours [15,17]. DrugBank (https://go.drugbank.com/), Swiss Target Prediction (https://www.swisstargetprediction.ch/), and TargetNet web server (https://targetnet.scbdd.com) were adapted to collect the targets of the qualifed compounds, and only targets with a probability degree greater than 0.8 were included [18]. Te Protox II web server (https://tox-new.charite.de/protox_II/) was then used to make toxicity predictions for these qualifed compounds [19].

Collection of Targets Related to IDD.
We collected IDDrelated targets from DisGeNET (https://www.disgenet.org/) and GeneCards (https://www.genecards.org/) databases [20]. It should be noted that GeneCards employs a parameter "Score" to indicate the relevance of the retrieval results to the subject terms used for search. In order to include targets that are more relevant to IDD, we sorted the targets retrieved on GeneCards in descending order of Score and only included the top 50 percent of the targets. In addition, we also collected genes enriched in MAPK, P38-MAPK, PI3K/AKT, Wnt/β-catenin, ATM-p53-P21-Rb, and mTOR signaling pathways from PathCards (https://pathcards. genecards.org/) [16]. Since the pathways mentioned above are reported to be closely related to IDD [21][22][23], the genes in these pathways and IDD-related targets were intersected and labeled as potentially important targets.

Compound-Target Network Construction and Analysis.
We used the Venn method to obtain overlapping targets of DJD and IDD as common targets for drugs and diseases. However, a single compound may be shared by multiple botanical drugs, which complicates understanding the relationship among the botanical drugs, compounds, and targets. To show this complicated relationship more intuitively, we renamed the compounds contained in the various botanical drugs in DJD according to the following rules: the compounds numbered "A" to "H" represented compounds shared by two or more botanical drugs; the name of a compound consisting of an abbreviation plus an Arabic numeral sufx indicates that the herb represented by the abbreviation uniquely occupies the compound. Te renaming results of compounds in DJD are provided in Table S2. Ten, a compounds-targets network was generated using Cytoscape (v3.8.2). Te degree of nodes representing each compound in the network was analyzed using CytoNCA (a Cytoscape plugin). We then analyzed the protein functions encoded by these common target genes and the network of transcription factors that regulate them. Te above information was retrieved through the Panther classifcation system (https:// pantherdb.org/) and Transcriptional Regulatory Relationships Unraveled by Sentence-based Text mining (TRRUST, https://www.grnpedia.org/trrust/), respectively [24]. A bar graph summarizing the transcriptional regulatory network of common target genes was generated from Metascape (https:// metascape.org/gp/index.html/). Targets   2 Computational Intelligence and Neuroscience and Protein Complexes or Functional Modules. Te common targets obtained through the Venn method were uploaded to the STRING database (https://string-db.org/) with the species limited to "9606" (human), to collect protein-protein interaction (PPI) information [16]. We then screened for interactions with confdence scores greater than 0.4 and built the PPI network on Cytoscape (v3.8.2). Te topological analysis of the resulting PPI network was performed using CytoNCA (a Cytoscape plugin). Te parameters obtained from the analysis were imported into R software 4.1.2 to screen the nodes with degree and betweenness greater than 2 times the median in the PPI network as the core targets. In addition, in living phenomena, proteins usually form complexes or functional modules to function. So, we further use the "Molecular Complex Detection" (MCODE) algorithm to investigate the underlying protein complexes or functional modules in the PPI network [25].

Analysis of Biological Insights. Gene Ontology (GO) and
Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analyses were conducted by R package "cluster-Profler" on these common targets, p value <0.005 was used as the cutof to be considered signifcantly enriched, and R package "ggplot2" was used to visualize the enrichment results [26]. Te restart random walk algorithm with restart was employed to evaluate the infuence of DJD targets on a PPI network to screen genes with specifc biological signifcance [27]. Ten, biological function analysis of genes based on co-expression correlations was performed using the R package "Correlation AnalyzeR" [28], defning a predesigned gene correlation matrix as a "cartilaginous" tissue source. Two similar groups in the PPI network can exhibit similar biological efects. Here we used the proximity index proposed by Menche et al. [29] to calculate the proximity index to understand the role of DJD in the biological process of interest. Te calculation formula of the network proximity index of group A genes and group B genes is as follows: Te value sAB < 0 indicates that the targets of the two groups are located in the same neighborhood, suggesting similar efects. We collected genes related to cellular infammatory response, autophagy, and apoptosis on Gene-Cards. We then screened the same number of genes as DJD targets from the above three gene sets based on higher correlation with biological processes and calculated their network proximity index to DJD targets. Ten, we randomly selected 50 groups of genes with the same number of DJD targets from the human protein interaction network containing 10,995 genes constructed on STRING and calculated the average value of their network proximity index to DJD targets as control.
2.6. Molecular Docking. Before carrying out molecular docking, we identifed binding site of each receptor protein by reviewing the literature as well as referring to the coligand binding site of the receptor protein. For receptor proteins lacking evidence to identify binding sites, POCASA (https://altair.sci.hokudai.ac.jp/g6/service/pocasa/) [30] was used to predict their active pockets as binding sites. We then downloaded the 3D structure of the receptor proteins containing the binding sites from the RCSB-PDB database (https://www.rcsb.org/) [15]. Te SMILE format fles of the small molecule ligands were obtained from PubChem (https://pubchem.ncbi.nlm.nih.gov/) [27]. UCSF Chimera (1.16) was used to generate 3D structures through the SMILE format of small molecule ligands [31]. Information on the receptor proteins and small molecule ligands employed by our docking is provided in Table S3. Ten, we used UCSF Chimera (1.16) to optimize receptor proteins and small molecule ligands (minimization routines are provided by MMTK, which is included with Chimera) [31]. We then imported the optimized receptor proteins and small molecule ligands into AutoDockTools for predocking preparation. We performed removal of water molecules, addition of hydrogen atoms, calculation of charge, and addition of atom type to the receptor proteins. Operations performed on small molecule ligands include adding hydrogen atoms, adjusting charges, determining roots, and detecting and setting torsional bonds. After completing the predocking preparation, we set the parameters of the docking box according to the binding sites identifed previously. Te parameters of the docking box are recorded in Table S3. We then performed molecular docking using AutoDock Vina with docking set to be based on AutoDock4 force feld and exhaustiveness set to 32 (when the exhaustiveness is greater than 25, more resource consumption can only bring little beneft to the scoring function; we set exhaustiveness to 32 to ensure accuracy when performing molecular docking) [32]. After molecular docking, the PLIP web tool (https://pliptool.biotec.tu-dresden.de/plip-web/plip/index) was used to analyze the protein-ligand interaction [33]. UCSF Chimera (1.16) was used to visualize the docking results [31].

Compounds of DJD and Common
Targets. 67 compounds of DJD are screened out by employing the flter, ensuring the compounds have oral bioavailability and therapeutic potential (Table S4). Te toxicity parameters of these compounds were predicted through the Protox II web server to assess their toxicity. Figure 1(a) describes the potential of these compounds in terms of hepatotoxicity, carcinogenicity, immunotoxicity, mutagenicity, cytotoxicity, and acute oral toxicity (LD50, mg/kg). As shown in Figure 1(d), only 2 compounds (Dianthramine and Mairin) are predicted to be hepatotoxic. In addition, the acute oral toxicity of Dianthramine and Mairin is predicted to be at a high level of 300 mg/kg and 1190 mg/kg, respectively. Te above results suggest that DJD has a comparatively lower risk of hepatotoxicity. However, the toxicity risks of DJD in terms of cytotoxicity, immunotoxicity, and cardiotoxicity need to be considered. Many of the compounds are predicted to exhibit cytotoxicity, immunotoxicity, and cardiotoxicity, and the acute oral toxicity of some compounds is at a low level (such as Deoxyharringtonine, 3-O-Methylviolanone, and

Computational Intelligence and Neuroscience
Wallichilide). Although none of the compounds with mutagenic toxicity in DJD are at the low level of acute oral toxicity, there are still many compounds predicted to have mutagenic toxicity. Further evaluation of the risk of mutagenic toxicity of DJD is needed. Te above provides information for balancing the efcacy and safety of DJD. Ten, according to the constraints in the method to ensure high confdence of targets, a total of 375 DJD-related targets and 1193 IDD-related targets are fnally included. 68 common targets of DJD and IDD are identifed through the Venn method ( Figure 1(a)).

Construction of a Compound-Target Network of DJD and Gaining Insight into Key Compounds from a Network
Perspective. A network of botanical drugs, compounds, and common targets was constructed to help understand the complicated interactions between them (Figure 1(b)). Compounds represented by the letters "A" through "H" indicate their presence in two or more botanical drugs of DJD. Tey are Mairin, β-Sitosterol, Sitosterol, Kaempferol, Mannitol, Stigmasterol, Quercetin, and Wogonin, respectively. Among them, Beta-Sitosterol, Sitosterol, and Kaempferol are the three most widely distributed compounds, which exist in 8, 6, and 5 botanical drugs, respectively (Table S2). Te abovementioned compounds are widely distributed in botanical drugs, suggesting that they are an important part of the DJD compound library, which to some extent refects that they may play an important role in the treatment of DJD. More importantly, Figure 1

Protein Function Classifcation and Transcriptional
Regulatory Network. Function classifcation of the proteins encoded by these common targets reveals that metabolite interconversion enzyme, transmembrane signal receptor, protein-modifying enzyme, and gene-specifc transcriptional regulator are the most distributed groups, with 17, 10, 10, and 10 target enrichment (Figure 2(a)). Interestingly, oxidoreductase is the primary type in metabolite interconversion enzymes, including SOD1, ACADM, ACOX1, ADHA1, ALDH9A1, ETFA, MAOA, SRD5A2, and TYR, suggesting that DJD has the potential to regulate cellular oxidative stress (Figure 2(a)). Oxidative stress is considered to be one of the initial factors inducing nucleus pulposus cell senescence, so DJD may regulate oxidative stress as one of the mechanisms of its treatment of IDD [5]. Proteinmodifying enzymes include AKT1, ADAM29, CASP8, CHEK2, HPR, LCK, MMP11, MMP1, PRSS1, and STSD ( Figure 2(a)). AKT1, as a non-receptor serine protein kinase, Cinchonan-9-al, 6'-methoxy-, (9R)- Spinasterol   10  28  80  100  125  155  159  200  200  200  210  263  300  322  522  562  600  650  777  832  856  890  890  940  940  1000  1190  1213  1350  1500  1500  1510  1679  2000  2000  2000  2000  2000  2000  2000  2340  2340  2500  2500  2500  2573  3000  3000  3919  3919  3919  4000  5000  5000  5000  8000  8000  10000  20000  20000 Acute oral toxicity (LD50, mg/Kg) Hepatotoxicity Carcinogenicity Immunotoxicity Mutagenicity Cytotoxicity regulates the conversion of activated forms of proteins and is essential for signal transduction. AKT1 plays a vital role in various signaling pathways such as the MAPK and PI3K/ AKT pathways and also plays an essential role in gene transcription mediated by NF-κB pathway [34,35]. Moreover, these pathways were previously reported to be related to IDD [13,14,21,22]; therefore, DJD may target AKT1 to regulate the above pathways and thus treat IDD. MMP1 and MMP11 are members of matrix metalloproteinases (MMPs), which are key enzymes in the degradation of extracellular matrix and afect the balance of synthesis and catabolism of extracellular matrix in nucleus pulposus [36]. Te imbalance of extracellular matrix metabolism directly leads to the morphological changes of intervertebral disc and accelerates the degeneration process [37]. Terefore, the efect of DJD on extracellular matrix metabolism by targeting MMPs may serve as a potential mechanism for its treatment of IDD. As for the cell surface transmembrane signal receptors binding ligands, DJD includes transmembrane signal receptor and G-protein coupled receptor, and the former is the primary (Figure 2(a)). C4 zinc fnger nuclear receptors are major gene-specifc transcriptional regulators in DJD, including AR, NR2E3, NR3C1, PPARG, and THRB (Figure 2(a)). Te transcriptional regulatory network of DJD target genes was analyzed. Figure 2(b) shows the number of genes regulated by all genes playing transcriptional regulatory roles and their enrichment ranking in the network. SP1, as a core member of the transcriptional regulatory network, regulates the transcription of 94 target genes, followed by YY1, which regulates 27 genes, and GATA1, which regulates 20 target genes. It is worth noting that HIF1A, PARP1, RELA, and AR are not only transcriptional regulators but also target genes of DJD. Figure 2(c) shows the regulatory relationship between the above four transcriptional regulators and target genes of DJD. Among them, RELA has an extensive regulatory relationship with other target genes, which suggests that it is an essential transcriptional regulator for DJD treatment of IDD. Table 1 details the biological processes associated with IDD regulated by these transcription factors, mainly involving the regulation of cellular senescence, apoptosis, infammation, mechanical stress, and hypoxia. Mechanical stress, infammation, and hypoxia afect the senescence and apoptosis of NPCs, which are closely related to IDD [38]. Tese results suggest that DJD may regulate specifc transcription factors to afect degeneration of disc at the transcriptional level.

Identifcation of Key Targets and Results of Molecular
Docking. A PPI network was constructed with 68 common targets (Figure 3(b)). Te PPI network has 402 edges, the average node degree is 11.8, and the average local clustering coefcient is 0.666. Among the common targets, CTNNB1, MYC, PDGFRA, CACNA1S, FLT3, PI3KR1, INSR, COL1A1, IGF2, RASA1, TGFB1, AKT1, ERBB2, MET, TP53, CHUK, IL1B, and RPS6KA3 are marked as IDD-related pathway members (Figure 3(b)), 15 in MAPK pathway, 13 in PI3K/AKT pathway, 2 in Wnt/β-catenin pathway, and 1 in ATM-p53-P21-Rb pathway. Most of them were of high degree (Figure 3(a)) and interacted extensively with other common targets in the PPI network, suggesting that those targets may play an important role in the treatment of IDD by DJD. Te network topology analysis identifed 13 key targets, including AKT1, PIK3R1, ALB, TP53, MYC, NR3C1, IL1B, ERBB2, CAV1, CTNNB1, AR, IGF2, and ESR1. Tese key targets are at the core of this PPI network and closely interact with other common targets, suggesting their central role in DJD treatment of IDD, especially AKT1, as it exhibits the highest degree (degree � 48, Figure 3(a)). Te results of molecular docking show that the binding free energy ranged from −5.12 to −12.61 kcal/mol and the inhibition constant (Ki) ranged from 0.57 × 10 −3 to 175.26 μmol/ml (Figure 3(c), Table 2). Figure 4 visualizes the binding of receptor proteins and their small molecule ligands. Te results of protein-ligand interaction are shown in Table S5. Te protein-ligand interaction analysis shows that there are mainly hydrophobic interactions and hydrogen bonds between receptor proteins and small molecule ligands. In addition, there is a π-stacking interaction between TP53 and Quercetin. Te above results suggest that the core targets as receptor proteins can form solid binding with the corresponding small molecule ligands in DJD. Moreover, the MCODE algorithm investigated the PPI network's protein complexes or functional modules. Te analysis results show that one module is detected, while all the key targets are distributed in this module ( Figure S1), further indicating that these key targets play an essential role in the treatment of IDD by DJD.

DJD Regulates Cellular Mechanical Stress Response and
Reactive Oxygen Species Processing. 1240 enrichment results of biological processes (BPs) were identifed through the GO analysis. Figure 5(a) shows the top 10 enriched BPs. Tree BPs caught our attention, namely, responses to mechanical stimuli (p �1.46e −10 ), reactive oxygen species metabolic process (p � 3.28e −10 ), and reactive oxygen species biosynthetic process (p � 5.65e −10 ), as mechanical stress and oxidative stress are considered to promote cellular senescence and serve as risk factors for the development of IDD [38]. We merged the reactive oxygen species metabolic process and reactive oxygen species biosynthetic process as the reactive oxygen species synthesis and metabolism process. Figures 5(b) and 5(c) show the interaction between the proteins enriched in the above two BPs. Both BPs were enriched in 11 targets, respectively. To further explore how DJD regulates the cellular response to mechanical stress, we used the target genes enriched in this BP as seed genes and the random walk algorithm with restart to calculate its difusion score in the PPI network composed of 216 mechanical stress-related genes. Ten, the top 10 genes with scores were selected as candidate genes for subsequent analysis: FOS, RETN, TNF, PTGS2, EDN1, MMP2, TLR4, JUN, NRXN1, and MAPK3 ( Figure 6(a)). Notably, PPARG is the highest scoring gene, indicating its essential role in DJD regulating cellular mechanical stress responses. Cluster analysis based on the co-expression correlation of genes in cartilage tissue divides candidate genes into 4 clusters (Figure 6(b)), each of which may synergistically play a specifc role in the cellular response to mechanical stress. Te genes in cluster 1 include MMP2, AKT1, CASP8, and CTNNB1, and analysis found that they are all signifcantly associated with "laminin interaction" (Figure 6(c)). Te second cluster genes include COL1A1, GJA1, TLR4, DRD2, EDN1, MAPK3, and HR2A. Although COL1A1 and HR2A are associated with extracellular matrix composition, no typical biological process is signifcantly associated with them. Te third cluster genes are related to the regulation of ion channel activity ( Figure 6(c)). COL3A1 and MPO are related to the regulation of potassium ion channel activity, while RETN and NRXN1 are related to the regulation of extracellular ligandgated channel activity. Te fourth cluster genes are signifcantly associated with transcriptional regulation, including PPARG, JUN, PTGS2, and FOS ( Figure 6(c)).

KEGG Pathway Analysis.
Te KEGG pathway analysis was conducted, and a pathway with a p value <0.005 was considered signifcantly enriched. Finally, 76 signifcantly enriched pathways were screened out. Figure 8(a) shows the top 20 pathways, and more information is detailed in Table S6. Te most signifcantly enriched pathway is the MAPK pathway (Figure 8(b), Figure S2), and the PI3K/AKT pathway is also signifcantly enriched (Figure 8(c), Figure S3). Interestingly, many other signifcantly enriched pathways intercommunicate with PI3K/AKT pathway (Table S6), including MAPK pathway, cell cycle, apoptosis, FoxO pathway, and toll-like receptor pathway [39]. Te above suggests that the MAPK and PI3K/AKTpathways may play a vital role in treating IDD by DJD. In addition, it is worth noting that CHUK (IKK-A), one of the targets of DJD, acts as part of the canonical IKK complex which is a hub for PI3K/AKT pathway to connect with the NF-κB pathway [40].

Discussion
IDD was believed to be DDD's pathological basis [2]. A series of factors, such as mechanical stress, oxidative stress, and infammation, promote the senescence of NPCs [38]. DJD is a traditional Chinese medicine formula with a history of thousands of years, and it has been used to treat osteoarthritis in the past due to its anti-autophagy and antiinfammatory efects [10]. Recent studies have also found the potential of DJD to treat IDD, such as activating autophagy and signifcantly reducing apoptosis and matrix degeneration in nucleus pulposus cells and inhibiting   Computational Intelligence and Neuroscience infammation [11][12][13][14]. Tere are many botanical drugs in DJD, forming a compound library that synergistically exerts biological functions. Among the compounds of DJD, Quercetin and Kaempferol are assigned a central position in the treatment of IDD because they form the densest associations with IDD-related targets and target crucial targets (such as AKT1, PIK3R1, TP53, CHUK, IGF-1, ERBB2, MYC, IL1B, and CAV1). AKT1 was reported to involve in the regulation of autophagy of degenerated NPCs and extracellular matrix metabolism. Moreover, cellular senescence is the primary pathological process of IDD, and Akt can phosphorylate and inhibit p27 and p21, which are closely related to cellular senescence. ERBB2 is involved in regulating the extracellular matrix metabolism of disc, while IL1BR, IGF-1, and CTNNB1 are mainly related to the regulation of the infammatory response in degenerative discs [41][42][43][44][45]. TP53 is closely related to cell senescence. p53-p21 pathway and p16-Rb pathway are the most important signaling pathways that mediate most cellular senescence phenomena [46]. Te above results indicate that these key targets play an important role in DJD treatment of IDD. Based on the predicted toxicity parameters, DJD has a relatively low risk in terms of hepatotoxicity, which was confrmed in a previous study. A clinical trial that evaluated the possible liver and kidney damage of DJD revealed that no signifcant changes in liver or kidney functions and the  Table S5. 10 Computational Intelligence and Neuroscience severe incidence of adverse events were observed during the 4 weeks of administration of DJD [47]. However, the risks of DJD in terms of cytotoxicity, immunotoxicity, and cardiotoxicity require more consideration. While there is currently insufcient evidence to suggest that DJD may pose a safety risk, understanding the efects of the various botanical drugs in DJD can help reduce the risk and provide further insight into its treatment mechanism.
A previous study has identifed the MAPK pathway as a possible mechanism for DJD to treat IDD [14]. Interestingly, this study also revealed that the MAPK pathway has a prominent performance in DJD treatment of IDD ( Figure 8). Furthermore, this study reveals the mechanism by which DJD can treat IDD by regulating the MAPK pathway: Quercetin targets AKT1, TP53, ERBB2, MYC, IL1B, CHUK, IGF2, RASA1, MET, and RPS6KA3; Kaempferol targets AKT1, INSR, and FLT3; Beta-Sitosterol targets TGFB1; Yangambin targets CACNA1S; and Methylicosa-11,19-dienoate targets PDGFRA. Te above targets are mainly involved in the classical MAP kinase signaling pathway (IGF2, CACNA1S, INSR, FLT3, ERBB2, MET, PDGFRA, RASA1, CHUK, RPS6KA3, and MYC) represented by Ras and the JNK and p38 kinase pathway (IL1B, TGFB1, AKT1, and TP53) ( Figure S2). Te physiological efects of the two MAP kinase pathways are mainly related to infammation and apoptosis [48,49]. In addition, the analysis of network proximity and transcription factor regulatory network also indicates that the targets of DJD have roles in regulating infammation and autophagy. Tis suggests that the MAPK signaling pathway may be an important approach for DJD to achieve the regulation of infammation and apoptosis to treat IDD. Quercetin and Kaempferol are key compounds that DJD depends on to regulate the MAPK pathway because they have more targets on the pathway and their targets cover some important members, such as AKT1, TP53, MYC, and CHUK. Te above results further confrmed and supplemented the specifc molecular mechanism by which DJD regulates the MAPK pathway in the treatment of IDD. Te NF-κB pathway has also been shown to play a role in the DJD treatment of IDD in a previous study [13]. CHUK (IKK-A) acts as part of the canonical IKK complex in the conventional pathway of NFkappa-B activation and phosphorylates inhibitors of NFkappa-B on serine residues [40]. Moreover, CHUK is the  Figure 6: DJD modulates the response to mechanical stimulus in the intervertebral disc. (a) Te network spread score of gene set related to mechanical stress response in common target genes in the PPI network of response to mechanical stimulation (GO: 0009612) was calculated using the random walk with restart algorithm, and top 10 were selected as candidate genes for subsequent analysis. (b) In the candidate gene set, gene groups were identifed by the "Correlation AnalyzeR" R package as the members of the set who share correlations in common that are not shared with other members, thereby classifying genes with common biological functions. (c) Te correlation-based gene set enrichment analysis results of genes in each cluster.

Random Walk with Restart
Spread the infuence   Figure 7: DJD modulates the response to mechanical stimulus in the intervertebral disc. (a) Te network spread score of gene set related to mechanical stimulation in common target genes in the PPI network of response to mechanical stimulation (GO: 0009612) was calculated using the random walk with restart algorithm, and top 10 were selected as candidate genes for subsequent analysis. (b) In the candidate gene set, gene groups were identifed by the "Correlation AnalyzeR" R package as the members of the set who share correlations in common that are not shared with other members, thereby classifying genes with common biological functions. (c) Te correlation-based gene set enrichment analysis results of genes in each cluster.  target of Quercetin, which provides a potential mechanism by which DJD regulates the NF-κB pathway. In addition, AKT1 is involved in the phosphorylation of CHUK and has an important role in NF-κB-dependent regulation of gene transcription ( Figure S3) [40]. Terefore, AKT1/CHUK is also a possible pathway for DJD to regulate NF-κB pathway. NF-κB pathway is involved in the pathological process of a variety of infammatory diseases. Activation of NF-κB pathway targets downstream infammatory cytokines and promotes intervertebral disc degeneration [50]. Terefore, the NF-κB pathway may play a role in the regulation of infammation in the treatment of IDD by DJD. Te PI3K/ AKT pathway has attracted our great attention. Both core members of this pathway are targets of DJD (Kaempferol, Beta-Carotene, Quercetin, Baicalein, and Wogonin target AKT1; Quercetin targets PIK3R1). Possible biological efects induced by the PI3K/AKT signaling pathway in IDD include increasing extracellular matrix content, anti-apoptosis, induction or inhibition of autophagy to prevent IDD, and antioxidative stress [51]. Besides, more importantly, PI3K/AKT pathway has broad associations with other pathways that were signifcantly enriched for common target genes [39]. Interestingly, many related pathways (cell cycle, apoptosis, and FoxO signaling pathway) play important roles in cellular senescence, which was considered the pathological basis of IDD, by regulating cell cycle, apoptosis, autophagy, and oxidative stress 6, 55. Te above results suggest that DJD may target genes located in a network consisting of the PI3K/ AKT pathway and its related pathways, especially the MAPK pathway, which contains the most common target genes, to regulate a series of biological processes (such as cellular senescence, infammatory response, and oxidative stress) to afect the aging process of the intervertebral disc [48,49,51,52].
Functional classifcation analysis of common target genes revealed that DJD might be involved in regulating oxidative stress and kinases and transcription factors (Figure 2(a)). Te regulatory role of DJD in the cellular oxidative stress response is also verifed by GO analysis ( Figure 5). Furthermore, GO analysis also suggests that DJD can modulate cellular responses to mechanical stress ( Figure 5). Moreover, further analysis shows that DJD may act on extracellular matrix components, especially laminin ( Figure 6(c)), which not only plays an important role in mechanical stress signal transduction but also regulates the synthesis and metabolism of extracellular matrix [43]. In addition, MMP1 and MMP11, the targets of DJD, are key enzymes in extracellular matrix degradation, afecting the balance of extracellular matrix synthesis and catabolism in nucleus pulposus [36]. Imbalances in extracellular matrix synthesis and metabolism in IDD lead to its reduction, which further develops with severe consequences, including rupture of the annulus fbrosus and destruction of NPCs [37]. Terefore, the possible regulatory role of DJD on the extracellular matrix is of great signifcance for delaying IDD. In addition, regulating ion channels is a key process in cellular mechanical stress signal transduction [53], and DJD may be involved in regulating ion channels, especially potassium channels. Moreover, DJD may also play a role in the cellular response to mechanical stress by regulating transcription (Figure 6(c)). As for the regulation of oxidative stress by DJD, this study shows that DJD may play a role in the synthesis and metabolism of reactive oxygen species in the respiratory chain and mitochondria and fatty acid oxidation and may also be involved in the cyclooxygenase P450 pathway and regulation of Rho and Ras protein activation. Te cytochrome P450 cyclooxygenase pathway mediates the metabolism of Arachidonic acid and is involved in BPs such as oxidative stress, infammation, immunity, apoptosis, and proliferation [54]. Rho protein and Ras protein are GTPases. Rho kinase (ROCK) is an efector of Rho. Its upregulation induces oxidative stress [55]. Besides the regulatory role of DJD in cellular responses to mechanical and oxidative stress, the network proximity index analysis also indicates that DJD plays a role in cellular infammatory responses, apoptosis, and autophagy (Table 3). Furthermore, HIF1A, PARP1, RELA, and AR function as transcriptional regulators and targets of DJD and play roles in apoptosis, infammatory responses, hypoxia, cellular senescence, and mechanical stress stimulation (Table 1). Overall, our fndings above support the role of DJD in several pathological processes in IDD, including oxidative stress, mechanical stress, infammation, extracellular matrix synthesis and metabolism, apoptosis, and autophagy.

Conclusion
AKT1, PIK3R1, TP53, MYC, CTNNB1, ALB, NR3C1, IL1B, ERBB2, CAV1, AR, IGF2, and ESR1 are crucial targets of DJD in the treatment of IDD. DJD is involved in multiple physiological and pathological processes of IDD, mainly including the regulation of mechanical stress, oxidative stress, infammation, and autophagy. MAPK pathway, PI3K/ AKT pathway, and NF-κB pathway play a pivotal role in DJD treatment of IDD. Quercetin and Kaempferol are the key compounds of DJD in the treatment of IDD.

Data Availability
Te data generated or analyzed in this study are included in the article and its supplementary material; further inquiries can be directed to the corresponding author.

Disclosure
Hao Liu and Yumin Li are co-frst authors.

Conflicts of Interest
Te authors declare that they have no conficts of interest.
proofreading. Haopeng Li reviewed and edited the manuscript. All authors have read and agreed to the published version of the manuscript. Hao Liu and Yumin Li contributed equally to this work.

Acknowledgments
Tis work was supported by the Social Development Science and Technology Project in Shaanxi Province (grant no. 2021SF-172).

Supplementary Materials
Table S1 records Lipinski's rule of fve (RO5). Table S2 details the renaming results of the compounds in DJD. Table S3 shows the information on receptor proteins and docking sites. Table S4 lists the 67 DJD compounds that we screened for oral bioavailability and therapeutic potential. Table S5 details the protein-ligand interaction information. Table S6 exhibits the results of KEGG pathway analysis of common targets. Figure S1 shows the protein complex or functional module in the PPI network of common targets. Figure S2 shows the target genes of DJD involved in MAPK signaling pathway. Figure S3 shows the target genes of DJD involved in PI3K/AKT signaling pathway. (Supplementary Materials)