Predication of the Underlying Protective Effect of Saffron on Parkinson's Disease via Network Pharmacology and Molecular Docking

Reports indicate that saffron originated from Iran then introduced into mainland China through the earlier established trans-Tibet trade routes in ancient China. Based to the theory of traditional Chinese medicine, saffron functions by promoting blood circulation and removing blood stasis, cooling blood and detoxication, as well as relieving depression for tranquilization. Modern medical research has demonstrated several properties of saffron including antitumor, antidepressant, enhancement of immunity, cardioprotection. Notably, recent studies introduce that saffron has a neuroprotective effect, specically in the treatment of Parkinson's disease in recent years. Nevertheless, the underlying molecular mechanism remains elusive so far. As such, this study aims to predict the underlying mechanism of saffron on Parkinson's disease through network pharmacology and molecular docking. effect via multiple targets and pathways in the treatment of Parkinson's disease and our systemic pharmacological analysis provides a basis for the clinical application and in-depth study of saffron.


Introduction
Parkinson's disease (PD) is a neurological disorder characterized by degeneration of dopaminergic neurons in the substantia nigra pars compacta, with dyskinesia as the prominent clinical symptom which includes resting tremor, bradykinesia, rigidity, and postural instability [1]. The pathological mechanism of PD is complex, and multiple biological mechanisms are related to PD, including oxidative stress, immune in ammation, mitochondrial dysfunction, and accumulation of toxic proteins [2][3][4][5][6]. Additionally, epidemiological studies have suggested that several factors play key roles in the development of PD, including, race, age, environment, lifestyle, and heredity among others [7][8][9][10]. Notebly, a retrospective analysis of PD in 2016 revealed that the number of patients nearly tripled from 2.5 million in 1990 to 6.1 million across the globe [11]. In China, the current situation remains unoptimistic due to the surge of the aging population. Besides, a statistical analysis estimates that the Chinese PD patients will reach 4.94 million, more than half the global incidence by 2030 [12]. Dopamine-related drugs including levodopa are presently used as rst-line therapy in clinical treatment based on the authoritative guideline [13,14]. As a consequence, although it can alleviate symptoms in the short term, there is still lack of evidence-based medical evidence that the drug can reverse the pathological process of PD. The problem that has to be paid attention to in the process of clinical treatment is that patients receiving long-term treatment will develop tolerance, so it is necessary to continuously increase the dose to achieve the therapeutic effect [15][16][17]. The use of such drugs has caused many side effects, including dyskinesias and behavioral inhibition and control damage. Therefore, continuing to carry out in-depth research on the mechanism of PD and discovering better therapeutic drugs is the direction of future efforts [18].
Crocus sativus L., commonly known as saffron, is a perennial stemless herb of the Iridaceae. Notably, commercial saffron comprises the dried red stigma. Saffron is widely cultivated in Europe, Turkey, Iran, Central Asia, India, China, and Algeria, of which Iran has been the world's major producer and exporter since time immemorial (85% of the global saffron production) [19]. A lot of Islamic traditional medicine literature introduced saffron as an astringent, resolvent, and concoctive drug with therapeutic effects on gastro-hepatoprotective, urogenital disorders, antidepressants, ocular disorders, etc. [20]. The accumulating pharmacological studies of saffron indicated that it can elicit anti-tumor, anti-in ammatory, antioxidant, immunomodulatory effects, and neuroprotective action [21][22][23][24][25]. Notably, the pharmacological mechanism of anti-oxidation, anti-in ammatory and inhibition of toxic α-Syn protein aggregation of saffron extract may be related to ameliorating the symptoms of PD patients [26][27][28]. The use of new research methods to clarify the precise pharmacological mechanism of saffron has important clinical signi cance for further promoting the use of the drug in the treatment of PD patients. Scientists believe that the active substance groups of traditional Chinese medicine (TCM) play a therapeutic role via multitarget and multi-pathway. Of note, network pharmacology is an effective method to understand human body-drug interaction from a macro perspective, and to further explore the scienti c problem on mechanism by which the complex components of TCM interfere with disease targets [29]. Given that the precise mechanism of saffron in the treatment of PD remains unclear, we sought to predict the underlying mechanism of saffron on PD through network pharmacology and molecular docking. First, we attempted to screen the related active component of saffron in therapeutic target on PD and further conducted interrelation analysis using multiple databases. Moreover, to clarify the mechanism of saffron in the treatment of PD, the bond strength between components and disease targets were evaluated by molecular docking. The owchart of the experimental procedures is shown in Fig. 1.

Methods
Screening the candidate compounds and targets of saffron Conventionally, oral administration of Saffron was usually the selected route, hence absorption, distribution, metabolism, and excretion (ADME) regulated the process of drug exerting its action. Oral bioavailability (OB) and drug-likeness (DL) have been introduced to characterize the pharmacokinetic characteristics of active components, and to further provide criteria for screening active components from TCM [30,31]. OB is a vital index to describe the relative amount of drug absorption into the blood circulation, whereas DL refers to the similarities between the active components and most currently known drugs in physical and chemical properties. Generally, compounds that meet these basic pharmacokinetic parameters might accelerate new drug discovery and improve the success rate of drug development. Based on the candidate compounds screening thresholds, this study set OB≥30% and DL≥0.18 to screen the active components of saffron using the traditional Chinese medicine system pharmacology database and analysis platform (TCMSP, http://lsp.nwu.edu.cn/tcmsp.php) [32,33]. As supplementary, a literature search was conducted in PubMed (https://pubmed.ncbi.nlm.nih.gov ), using the items "Parkinson", "Crocus sativus L", and "Saffron" as keywords. Based on this, we added the following active components: Crocin I, Crocin II, Safranal, and Picrocrocin. Despite these 4 components not meeting the above requirements, it was clearly mentioned highlighted in relevant literature for neuroprotective and the therapeutic effect of PD [34][35][36][37]. Further, the biological targets of the candidate compounds in Saffron were gathered from TCMSP. Moreover, α-Synuclein protein (α-Syn, protein-coding gene known as SNCA) was supplemented into the nal result because of it being the main components in the Lewy body, the most important pathological markers of PD. It has been convincingly demonstrated that Crocin I and Crocin II could interact with α-Syn and inhibit the formation of aggregates in vitro and computer simulation studies [28,38]. The last step was to delete duplicate targets and further ensure data quality as well as standardization using the Uniprot database (https://www.uniprot.org/).
Collection of the con rmed and potential therapeutic targets of PD Notably, disease therapeutic targets usually include receptors, enzymes, ion channels, transporters etc. The interaction between drugs and targets triggered changes in physiological and biochemical functions of the body, to realize the regulation of drugs on the pathological and physiological state of the body and achieve the purpose of disease treatment [39,40]. In light of all of this, the con rmation of disease targets is an important basic work in drug development and clinical diagnosis and treatment. Search terms including"Parkinson's disease" as keywords was used to conduct a systematic data search for disease targets from databases with no date restrictions, including, Online Mendelian Inheritance in Man Databases (OMIM, http://www.omim.org/), Genecards Databases (http://www.genecards.org/, cut off value larger than 20), DrugBank Databases (https://www.drugbank.ca/), Comparative Toxicogenomics Database (CTD, http://ctdbase.org/, Score greater than 40). This selection of databases was based on susceptibility genes, con rmed drug targets, and environmental exposure factors hinged on the focus of different databases. Further, the search results were comprehensively analyzed and converted to the corresponding o cial gene symbol by Uniprot Database to nally obtain the de nitive targets of PD.
Construction of the drug-candidate compounds -therapeutic targets-disease network Further interrelation analysis was performed to obtain the intersection of candidate compounds targets and therapeutic targets all the above collected, namely common targets, these candidate compounds from saffron might act directly on the therapeutic targets of PD. Then, the network of drug-candidate compounds-therapeutic targets-disease was constructed using Cytoscape software version3.7.2. Nodes with different shapes and colors represented different properties of the nodes in the visual network diagram. Speci cally, the connecting line represented the subordination relationship of drug-compounds or disease-targets and the compound-targets action relationship between nodes [41].

Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis
In this work, we took a step further transforming the o cial gene symbol of common targets into entrezID using the "org.Hs.eg.db" R package of R software version 3.6.2 (x64). Then, Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses were performed using the "clusterPro ler" R package, meanwhile, associated signaling pathways were rendered using the" Pathview" R package. The results were presented in the form of a bar chart and a bubble chart. P-value < 0.05 or Q value < 0.05 were considered statistically signi cant enrichment.

Construction of the protein-protein interaction (PPI) network for the therapeutic targets of PD
To predict all the potential interactions of common targets between candidate compounds targets and therapeutic targets, all the common targets were inputted into the String database (https://stringdb.org/cgi/input.pl), with the species limited to "Homo sapiens" and "minimum required interaction score>0.4" as the basis of con dence for the interaction between proteins. Subsequently, the basic topological parameters of protein-protein interaction network including the degree value of the node (Degree), closeness centrality (CC), and betweenness centrality (BC) were calculated using the Network Analyzer tool function of String database, all the results were presented as a bar chart and 3D scatter plot.

Molecular docking simulation
The concept of molecular docking refers to placing small molecular ligands in the binding region of macromolecular receptors through computer simulation then calculating physical and chemical parameters to predict the binding a nity between them. Brie y, the X-ray crystal structure of the target protein was downloaded from the protein data bank database (PDB, http://www.rcsb.org/), and the molecular docking was performed using MOE 2014.09 software. The Triangular Matching docking method was used to dock the molecules, then, 30 conformations of each ligand-protein complex were generated based on docking score. The docking poses between the active components and the binding pocket residues of the target protein were evaluated via visual inspection relying on the visualization function of the software. Finally, the clear images of ligand-protein interactions were retained and outputted

Results
Candidate compounds and targets of Saffron A total of 9 candidate compounds were collected, 4 of which were collected from the primary literature as mentioned above, whereas the remaining 5 including Quercetin, Kaempferol, Isorhamnetin, Crocetin, n-Heptanal were screened from the TCMSP database. Also, the TCMSP database was used for target analysis of candidate compounds, on this basis, the repeated targets were consolidated and the abovementioned con rmed target SNCA was added. A total of 92 targets of candidate compounds were collected and the basic pieces of information, including OB, DL, and numbers of corresponding targets are listed in Table 1. Results from four databases are merged and de-duplicated to provide a list of all con rmed and potential therapeutic targets of PD. A total of 38 therapeutic targets were screened out from Drugbank. The results were more convincing because the resources provided by Drugbank came from the con rmation information of listed drugs. Up to 104 PD related targets with cut off value larger than 20 were collected from the Genecards database while 29 therapeutic targets were collected from the OMIM database.
Notably, the Genecards database provides genes related to common human diseases, while the OMIM database focuses on the relationship between disease phenotypes and their pathogenic genes. CTD database considered the genes affected by PD related environmental toxicant exposure, and a total of 466 therapeutic targets were collected when the screening score was set as greater than 40. A total of 556 PD therapeutic targets were collected from different databases, as shown in Fig. 2. There was not much overlap between different databases, indicating that the therapeutic targets collected in this study were quite complete.
Common targets of between compounds targets of saffron and therapeutic targets of PD Further analysis revealed that 52 common targets were observed in both candidate compounds targets of saffron and therapeutic targets of PD (Fig. 3) and their basic information regarding Uniprot ID, o cial gene symbol of target, and protein full name of target (Table 2). Then, this result was classi ed based on the main molecular function of the target proteins, and these molecular functions concentrated on apoptosis, oxidative stress, in ammation, cholinergic system disorder, and toxic protein damage. This was consistent with the recent studies on the pathological mechanism of PD, (Fig. 4). The network of the drug, candidate compounds, therapeutic targets, and PD was built to analyze the complex interactions between saffron and PD. As shown in Fig.5, the network comprised 153 edges and 63 nodes including 52 therapeutic targets of PD represented by green circular nodes, and 9 compounds of saffron represented by orange triangle nodes. The node size and depth of the color determined the degree value of the node, for further explanation, the larger the size and the darker the color, the greater the degree value of the node. The current ndings con rm that the relationship between candidate components and therapeutic targets is not a simple rule like one-to-one correspondence, but complex, for instance, one-to-many or many-to-one. Many candidate components were related to multiple therapeutic targets of PD, particularly quercetin and kaempferol, which might be the important compounds in the treatment of PD.

GO and KEGG enrichment analyses
The purpose of GO enrichment was to clarify the potential mechanism of saffron in the treatment of PD from the biological process (BP), cellular component (CC), molecular function (MF), respectively. The results of GO enrichment of 52 common targets were depicted as a representative bar chart, the longer the bar, the smaller the p-value is, namely, the more signi cant the enrichment is. As shown in Fig. 6, the top 6 biological processes implicated in the treatment of PD were cellular response to chemical stress, response to the xenobiotic stimulus, cellular response to the xenobiotic stimulus, response to the metal ion, response to oxidative stress, and cellular response to oxidative stress. In terms of cellular components, the top 6 were membrane raft, membrane microdomain, membrane region, transcription regulator complex, an integral component of postsynaptic membrane, and axon terminus. Additionally, DNA-bind transcription factor binding, RNA polymerase II-speci c DNA-binding transcription factor binding, activating transcription factor binding, ubiquitin-like protein ligase binding, and glutathione binding were the top 6 molecular functions implicated in the treatment of PD. Fig. 7 is a simpli ed bubble chart that describes the results of KEGG enrichment, as its purpose was to clarify the top 16 signal pathways for the treatment of PD using saffron. In general, PD-related pathways including PI3K/Akt signaling pathway, apoptosis signaling pathway, and TNF signaling pathway showed signi cant enrichment. For clearer presentation, the speci c position and function of common targets are colored red in the signaling pathway in Fig .8.

Protein-protein interaction (PPI) network for the therapeutic targets of PD
In this work, the protein-protein interaction (PPI) complexity of 52 common targets is clearly shown in Fig.   9 (A), Further, the degree values of the top 30 proteins in interaction relationship are shown in Fig. 9 (B).
Here, the degree value of protein was a critical parameter, a direct indication of the importance in the network, and the protein with high a degree value was more likely to be the core target in the treatment of PD. To describe the topological properties of the nodes and comprehensively evaluate the status of proteins in the PPI network, a three-dimensional scatter plot was used to show the degree value, closeness centrality (CC), and betweenness centrality (BC) on each node (Fig. 9) (C). There was a signi cant positive correlation between them. Generally, the closeness centrality and betweenness centrality of protein with high degree value was also greater, indicating that the protein directly interacts with many proteins and act as a "bridge" to mediate the indirect interaction between other proteins. This conclusion implies that CASP3, IL6, MAPK8 were the three most important genes in the PPI network, which might be the core proteins of PD.

Molecular docking
Multiple studies have reported that the active components of saffron, including quercetin, kaempferol, isorhamnetin, crocetin, crocin I, crocin II, safranal, and picrocrocin exhibit anti-apoptotic and antiin ammatory properties [42][43][44][45][46][47][48]. The above results suggest that CASP3, IL6, and MAPK8 are at the core of the PPI network, consistent with the literature, there were closely related to neuronal apoptosis and neuroin ammation of PD [49][50][51]. Therefore, additional extensive research is essential to explore the effect of saffron on the therapeutic targets of PD. The molecular docking technology used in this work can be presented as macro, micro, and local state of the interaction between compounds and core proteins. The crystal structure of CASP3 (PDB: 5IC4), IL6 (PDB: 1N26), and MAPK8 (PDB: 4QTD) were retrieved from the PDB database and executed protonated and energy minimized before molecular docking. The docking score of the 8 compounds and 3 proteins was shown by the heatmap (Fig. 10). The scores were used to describe the binding free energy between the compounds and the proteins, i.e, the smaller the value, the lower the binding free energy, indicating a more stable binding between compounds and the proteins. Furthermore, it was observed that the binding free energy of crocin I and crocin II was less than -8.0 kJ/mol. It was also suggested that crocin I and crocin II selectively act on the targets associated with apoptosis and in ammation, which then functioned as the limitation of in ammation and apoptosis to play a neuroprotective role.

Discussion
Being a complex disease affecting the central nervous system, PD has become a major threat to human health, and, researchers indicate that the situation is getting worse in the years ahead. The apparent pathological mechanism is a decrease of dopaminergic neurons in the substantia nigra pars compacta (SNpc), causing the depletion of dopaminergic neurotransmitters in the striatal projection neurons, and eventually leading to the functional imbalance between the dopaminergic system of substantia nigra and the cholinergic system of the striatum [52]. The death of dopaminergic neurons involves complex pathogenic factors and pathological mechanisms, suggesting that it is di cult for drugs with a single target to achieve a satisfactory therapeutic effect in the treatment of PD. Traditional TCM has a long application history in China and other Asian countries, besides, TCM is considered the pioneer of "multicomponent and multi-target pharmacology" [53]. Based on the theory of TCM, PD belongs to the category of tremble syndrome which considers de ciency of body caused by old age, liver and kidney de ciency, and wind evil stirring internally caused by tendons dystrophy as its etiology. According to the theory of syndrome differentiation and treatment ("bian zheng shi zhi" in Chinese) in traditional Chinese medicine, the treatment principle of PD is to tonify liver and kidney, balance the liver and extinguish wind, nourish blood and soften tendons. Notably, this work was intrigued by a classical famous formula, Five Dragon Tremor Decoction ("wu long zhen chan tang" in Chinese) that was rst documented in the classical Chinese medical book Discussion on Febrile Disease ("wen bing Tiao bian" in Chinese) in the Qing Dynasty (18th century). Saffron is an important adjuvant in this formula and plays a signi cant role in promoting blood circulation and removing blood stasis. Further, it has been suggested that saffron might be a potential drug for the treatment of PD, a scenario that warrants in-depth investigation. Therefore, TMC has its understanding and characteristics on the pathological mechanism and treatment of PD.
Summarily, it executes a multi-component, multi-target, and multi-signaling pathways to achieve the purpose of disease treatment. Here, we used the network pharmacology method to clarify how the components of saffron interfere with targets and related pathways of PD disease.
Recent research ndings reveal that saffron extract harbors an antidepressant effect by inhibiting central in ammation, ameliorating oxidative stress, increasing the expression of synaptic plasticity-related proteins in hippocampal neurons, and regulating intestinal micro ora [54][55][56]. Additionally, our ndings show that crocin reduced the structural damage in soma volume and axon length of neurons and inhibited their spontaneous discharge in dopaminergic neurons in the ventral tegmental area (VTA) [57]. Moreover, evidence from clinical trials con rms the antidepressant e cacy of saffron extract comparable to Citalopram [58]. The above research results inspire the exploration of whether saffron extract protects the substantia nigra pars compacta, which is close to the space of VTA and dominated by dopaminergic neurons. So far, no clinical trial data supports the idea that saffron is a potential drug in preventing and treating PD, however, the biological activity of neuroprotective is gradually being clari ed in vitro studies. In an animal model of PD induced by rotenone, an inhibitor of the Mitochondrial Electron Transport Chain (ETC), crocin activated the PI3K/Akt pathway, then, phosphorylated Akt inhibited the phosphorylation of GSK-3β to reduce the phosphorylation of α-Syn and eventually inhibited the aggregation of α-Syn [59].
Importantly, phosphorylated α-Syn is more aggressive and more toxic compared to its prototypes [60]. Also, the activation of mTOR by p-Akt promoted the expression of downstream effector p70S6K and increased the level of Bcl-2 in preventing cell apoptosis. Also, p-Akt inhibited the expression of Caspase-9 by inhibiting the transcription of FoxO3a to prevent the initiation of the apoptosis cascade reaction [59]. Therefore, the activation of the PI3K/Akt pathway is an effective intervention approach in reducing the abnormal aggregation of toxic proteins and preventing cell apoptosis. PI3K/Akt signal transduction pathway was signi cantly enriched in our study, and this was consistent with a growing body of evidence in the literature.
Additional studies have recently suggested central in ammation as one of the key factors causing the death of dopaminergic neurons. In recent years, there has been an increasing focus on PD-related neuroin ammation caused by the aggregation of α-Syn and microglial activation caused by intestinal ora imbalance [61]. Also, crocin and crocetin might reduce the production of TNF-α and IL-1β induced by LPS in the rat brain, as well as simultaneously inhibit the activation of NF-κB signal pathway induced by LPS, with the ultimate goal of inhibiting the in ammatory factors released by activated microglia [62].
Another study also con rmed that crocin signi cantly reduced the levels of TNF-α and lipid peroxidation in the brain of PD model animals induced by 6-OHDA, and improved motor ability and cognitive function [63]. Noteworthy, reports indicate that the crocus formula ameliorates the blood glucose level of diabetic rats by regulating intestinal ora [64]. In summary, the anti PD effect of saffron might involve the in ammatory reaction mediated by microglia, inhibit the aggregation of α-Syn and regulate the intestinal ora in exerting a regulatory effect on central in ammation.
In our study, a series of network pharmacology methods including the OB and DL evaluation, target prediction, pathway identi cation, PPI network establishment, and molecular docking technology were applied to clarify the underlying mechanism of saffron in the treatment of PD. Our analysis provides convincing evidence that CASP3, IL6, and MAPK8 are the key targets to expend in the treatment of PD with saffron. Notably, crocin I and crocin II demonstrated strong binding to CASP3, IL6, and MAPK8, the binding free energy close to or even less than-8.0 kJ/mol, signi cantly better compared to others. Besides, the auxiliary neuroprotective effects potentially originated from quercetin, kaempferol, isorhamnetin, crocetin, safranal, and microglial due to their binding capacity to CASP3, IL6, and MAPK8. Our previous studies revealed that Saffron comprised avonoids and carotenoids, especially rich in crocin I and crocin II [65]. Moreover, pharmacokinetic studies indicated that crocins were absorbed by passive transcellular diffusion to a high extent within a short time interval over the intestinal barrier, then, penetrated in a quite slow process in the blood-brain barrier to reach the central nervous system [66]. All the evidence provided in these studies con rms the conclusion that saffron exhibits a satisfactory development prospect because of the potential neuroprotective effects in the treatment of PD. It is worth noting that, molecular docking is only a method to simulate the binding mode and a nity of active components and biological macromolecules by computer software from some algorithms. In fact, the interaction mode of the two molecules in vivo is very complex and affected by many physiological factors. Future research needs to verify the current results by designing in vitro and in vivo experiments, and provide important information for PD drug development.

Conclusion
In summary, we used a network pharmacology strategy to predicate the compounds with biological activity from saffron and the possible targets and pathways related to PD. Consequently, there was su cient evidence that crocin I and crocin II play a neuroprotective role by interfering with the target of apoptosis and in ammation pathway, and the good drug-target relationship implies that saffron extract may have potential research and development value in the treatment of PD, namely MAPK8, CASP3 and IL6, which have been predicted as the hopeful target of saffron on PD. The network pharmacology was an effective tool to explore the material basis of TCM and comprehensively reveal the effects of multicomponent, multi-target, and multi-path on the treatment of diseases. However, further experimental veri cation is essential, an important next step to consider.  Availability of data and materials The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.
Ethics approval and consent to participate Not applicable.

Consent for publication
Not applicable.   The drug-candidate compounds targets-therapeutic targets-disease network. The network comprised 153 edges and 63 nodes including 52 therapeutic targets of PD represented by green circular nodes, and 9 compounds of saffron represented by orange triangle nodes.
Page 24/30    The heatmap of docking score. The redder the heatmap color is, the smaller the docking score is, the stronger the binding ability between compounds and the proteins.