A Network Pharmacology-Based Approach to Explore Therapeutic Mechanism of Indian Herbal Formulation Nisha Amalaki in Treating Type 2 Diabetes Mellitus

Nisha Amalaki (NA), an Indian herbal formulation consisting of two herbs, Curcuma longa and Emblica o�cinalis, has been commonly used to treat Type 2 diabetes mellitus (T2DM). However, the pharmacological mechanism of NA remains unknown. In this study, a network pharmacology-based approach was used to explore its underlying mechanism. NA phytochemicals were collected from PubChem, KNApSAcK, IMPPAT, and ChEBI databases, and their potential targets were investigated using similarity ensemble approach (Tanimoto coe�cient ≥ 0.6). A protein-protein interaction network was constructed to study the interactions among the targets and clustered into separate modules using NetworkAnalyst 3.0. A signi�cant module (P ≤ .01) was identi�ed, and DAVID web tool was utilized for the enrichment analysis. A total of 201 phytochemicals and 262 targets of NA were selected. Forty-ve nodes of the signi�cant module were identi�ed as potential targets of NA. The enrichment analysis exhibited 27 biological processes and 78 pathways (P ≤ .01). Out of 45, 18 nodes were associated with T2DM as probable targets of NA. The metabolite-target-pathway network revealed that anti-diabetic effect of NA is a synergy of multi-target and multi-pathway efforts via regulation of glucose, lipid metabolism, insulin resistance, β -cell survival and proliferation, in�ammation, apoptosis, and cell cycle.


Introduction
Diabetes mellitus (DM) is a chronic, complex metabolic disorder, and the most common form is Type 2 diabetes mellitus (T2DM) 1 .It is estimated to affect approximately 422 million people worldwide, resulting in 1.6 million deaths annually 2 .Conventional therapies used to treat diabetes may be promising in glycemic control but are reported to have potential side effects like hypoglycemia, vitamin B12 de ciency, pancreatitis, upper RTI infection, lipoatrophy, weight gain, and gastrointestinal dysfunction 3,4 .
Consequently, people are turning their attention to traditional herbal medicine or diet-based therapy as a safer and more cost-e cient alternative medicine for T2DM [5][6][7] .Indian traditional medicinal system is a rich resource that describes various Indian medicinal plants used to prevent and treat T2DM 8,9 .Nisha Amalaki (NA), an Ayurvedic herbal formulation, has been used in the traditional Indian medicinal system to treat T2DM 10,11 .It consists of a ne powder of turmeric rhizomes (Curcuma longa L.; Nisha, Haridra in Sanskrit; Family: Zingiberaceae; CL) and Indian gooseberry (Emblica o cinalis L.; Amalaki in Sanskrit; Family: Euphorbeaceae; EO), prepared as a 1:1 (w/w) mixture 12 .Both CL and EO are known for diverse medicinal properties.C. longa is a common Indian spice traditionally used to treat several ailments such as diabetes, rheumatism, cancer, urinary disease, liver disorders, in ammation, cough, wound, and bruise healing 13 .Curcumin and its derivatives, such as bisdemethoxycurcumin and desmethoxycurcumin, are major phytochemicals in C. longa.They have been reported to possess signi cant antioxidant, anti-in ammatory, anti-infective, anti-carcinogenic, anticoagulant, and anti-diabetic activity [14][15][16][17] .On the other hand, E. o cinalis contains phytochemicals such as pedunculagin, gallic acid, emblicanin, quercetin, chebulinic acid, and corilagin, which has been shown to have antioxidant, anticancer, anti-in ammatory, anti-diabetic, antimicrobial, adaptogenic, nootropic, and immunomodulatory potential 18,19 .It has also been reported to prevent hyperlipidemia, osteoporosis, and several other ailments 20 .Although both the herbs possess anti-diabetic activity, the pharmacological action of the NA formulation remains to be explored.Both herbs and their formulation, NA, have been implicated in the treatment of diabetes, but their underlying mechanism of action is still not clear.
Network pharmacology (NP) approach has been a promising for understanding traditional herbal formulas 21 , identifying probable new drugs or targets [22][23][24] , and provide novel insights into drug actions.
In addition, it explores potential target spaces by allowing an unbiased examination of current drug molecules used in various therapeutic conditions 25 .It elucidates the probable mechanism of action of phytochemicals/bio-active compounds through huge dataset analysis and determines their synergistic effects in treating complex diseases 26 .
Therefore, this study was designed to develop an NP-based method to identify possible therapeutic targets and explore the underlying mechanism of this herbal formulation.First, the protein-protein interaction (PPI) network was generated using putative targets of phytochemicals from NA. Next, the network was clustered into various modules containing targets sharing a functional similarity.Finally, the modules with signi cant P-value were identi ed and enriched to pathways to generate the metabolitetarget-pathway interaction network.Also, the gene-disease association network was created to explore the use of NA in other diseases.The work ow of the NP-based method for NA herbal formulation has been shown in Figure 1.

Results
NA phytochemicals and target prediction.A total of 201 phytochemicals identi ed in NA (108 in CL and 93 in EO) were collected with CAS ID (Chemical Abstracts Service registry number) and PubChem CID (Supplementary le 1).The possible targets of the NA phytochemicals were determined using similarity ensemble approach (SEA).The scope of potential targets of NA was narrowed from 5187 to 1052 based on the Tanimoto Coe cient (Tc max ≥ 0.6) (Supplementary le 2).Further duplicate entries and genes not found in humans were removed, and the number of targets for analysis gradually decreased from 1052 to 262.PPI network analysis and module identi cation.The PPI network was created using NetworkAnalyst 3.0 as an undirected network, i.e., edges having no direction.The target genes/proteins were represented as 'nodes,' and the interaction between any two genes/proteins was represented by 'edge.'The network analysis revealed the interaction of 163 nodes via 604 edges (Figure 2).In the network, 42 nodes showed a degree of one, while 121 nodes showed a degree more than one.Out of 121 nodes, 39 nodes had ≥ ten connections to other nodes.We also found "betweenness" ranging from 2.5 to 2617.62 for 94 nodes in the constructed network.The results indicate that the constructed network was abundant in the hub proteins (high degree, i.e., number of connections with other nodes) and bottleneck proteins (high betweenness, i.e., number of shortest routes passing through a node), which suggests that they may be important proteins 27,28 .Based on the results, proteins having the high degree in the PPI network showed the high betweenness.As hub proteins contribute to many interactions and hold the network together 29 , they play a crucial role in regulating signaling pathways as well as transcription.Therefore, hub proteins may serve as potential therapeutic targets or biomarkers.
The constructed PPI network was further clustered into modules, which contain proteins with similar functions.A network module is a subnetwork in which nodes are more closely linked to each other than rest of the network.Identifying the modules within the network is important as it might help in detecting the hidden structural information.Seven highly connected independent modules were observed, out of which only Module 1 showed a signi cant P-value (P ≤ .001)(Table 1).Thus, the PPI network of Module 1 was extracted for further analysis (Figure 3).The particulars of topological parameters, i.e., closeness centrality, betweenness centrality, eccentricity, and degree, have been shown in Table 2, highlighting the importance of each target in the network.Gene ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis.GO enrichment analysis was done on 45 target genes of Module 1, and the GO terms with P ≤ .01 were selected and represented on the graph as − log P values (Figure 4).The results showed that these 45 target genes are involved in various biological processes like negative regulation of apoptotic process, aging, regulation of signal transduction by p53 class mediator, histone H3 deacetylation, positive regulation of transcription from RNA polymerase II promoter, negative regulation of cell growth, etc. (Figure 4a).In addition, these processes are associated with molecular functions such as transcription factor binding, NF-kappa B binding, protein kinase activity, DNA binding, protein homodimerization activity, etc. (Figure 4b).These processes occur in different cellular components like nucleoplasm, nucleus, cytosol, spindle microtubule, nuclear chromosome, etc. (Figure 4c).
KEGG pathway enrichment analysis was also done to explore the target's role (Supplementary le 3).The top 30 enriched pathways have been shown in Figure 5.The results showed that the targets were highly enriched in Pathways in cancer, Endocrine resistance, IL-17 signaling pathway, Apoptosis, Cell cycle, Wnt signaling pathway, Longevity regulating pathway-multiple species, etc.In addition, pathways related to the T2DM and its complications were also observed, including, PI3K-Akt signaling pathway, Insulin resistance, TNF signaling pathway, AGE-RAGE signaling pathway in diabetic complications, FoxO signaling pathway, NF-kappa B signaling pathway, Jak-STAT signaling pathway, MAPK signaling pathway, HIF-1 signaling pathway, Non-alcoholic fatty liver disease (NAFLD), etc.These results suggest that NA herbal formulation may exert therapeutic effects by regulating these pathways.
Gene-disease association network.A gene-disease association network constructed for the 45 target genes of Module 1 showed 424 nodes and 611 edges (Figure 6).The degree and betweenness of the resultant diseases ranged from 11 to 1 and 9700.68 to 0, respectively.The diseases with betweenness ≥ 50 were considered signi cant (Supplementary le 4).The results showed that besides diabetic conditions, NA could be explored in other disease conditions like neoplasms, leukemia, carcinoma, obesity, hypertensive disease, atherosclerosis, osteoporosis, liver cirrhosis, fatigue, heatstroke, depressive and anxiety disorders.
Identi cation of T2DM genes and corresponding NA phytochemicals.A list of 579 genes related to T2DM was identi ed using various databases as described in methodology (Supplementary le 5).Out of 45 genes, 18 were common among Module 1 and the T2DM related gene list (Table 3).The NA phytochemicals targeting these 18 gene targets were identi ed as curcumin, quercetin, (2S)-Eriodictyol 7-O-beta-D-glucopyranoside, arachidic acid, bis-(4-hydroxycinnamoyl)methane, bisdemethoxycurcumin, calebin A, demethoxycurcumin, dihydrocurcumin, letestuianin B, corilagin, indole-3-acetic acid, chebulinic acid, tauroursodeoxycholic acid, Go-Y022, epigallocatechin gallate, eriodictyol, glycocholic acid, naringenin, naringenin 7-O-beta-D-glucoside, beta-carotene, and quercetin-3-O-glucoside.The results also showed that AKT1, BCL2, CYP19A1, ESR1, IL2, MCL1, NR4A1, and RXRA are the targets of EO, while EP300, HDAC1, JUN, NFKB1, NR3C1, PPARD, and PPARG are the targets of CL.However, GSK3B, MMP2, and MMP9 are the common targets of both CL and EO.Next, the M-T-P network using the T2DM related metabolites, potential targets, and associated pathways was constructed using Cytoscape v3.8.2 (Figure 7).The network showed 63 nodes and 197 edges with an average clustering coe cient of 0.088 and network density of 0.084.In this network, phytochemicals like bisdemethoxycurcumin, bis-(4hydroxycinnamoyl) methane, and demethoxycurcumin showed the highest degree, each having ten targets suggesting that these compounds may be the signi cant phytochemicals of NA in treating T2DM.It was followed by curcumin, 1,7-bis(4-hydroxy-3-methoxyphenyl)-1,4,6-heptatrien-3-one, calebin A, and quercetin which had a degree equal to 8. The network analysis showed that one metabolite could correspond to multiple targets, and one target could correspond to multiple metabolites and pathways.Thus, the network re ected the features of the synergetic relationships between the multiple metabolites, targets, and pathways of NA.Based on the M-T-P network, a proposed schematic diagram was drawn outlining the target proteins and pathways involved in T2DM (Figure 8).

Discussion
Traditional medicinal plants have been used for centuries to treat complex diseases such as cancer and diabetes 30 .Traditional medicinal systems generally use herbal formulations comprising multiple compatible herbs to improve therapeutic effect through synergism 31 .Moreover, it implements a comprehensive approach that focuses on supporting complete functional recovery and eradicating underlying cause of the disease.The concept of NP is comparable to the theory of the Traditional medicinal system.Hence, it is appropriate to explore the components and mechanism of action of complex Traditional herbal formulations using various databases and available software.The present work has explored the mechanism of action of NA, traditionally used in India to treat T2DM.The network module approach and widely used enrichment analysis methods have been utilized to uncover the concealed information within the target PPI network.In this study, 201 phytochemicals in NA were predicted by network analysis, of which 20 have been found to have anti-diabetic effects.Subsequently, we found that these metabolites have therapeutic effects through regulating various T2DM related target proteins of different metabolic pathways.
The results are consistent with the earlier studies suggesting that these proteins and pathways play a crucial role in the pathophysiology of T2DM.PI3K/Akt signaling pathway activation induces insulin secretion from pancreatic β-cells 34,35 .Also, activation of AKT and its downstream signaling intermediates, viz., GSK3, mTOR1, and FoxO1, leads to increased proliferation, mass, and cell size of pancreatic β-cells 36 .It regulates various signaling pathways such as NF-κB, MAPK, and FoxO.These pathways play a crucial role in regulating protein synthesis, cell differentiation, proliferation, cell survival, and apoptosis 37 .NR4A1 protein is elevated in response to glucose and saturated fatty acids in pancreatic β-cells 38,39 , further regulating cell proliferation and insulin secretion 40 .Knockout of NR4A1 has been shown to reduce β-cell density in the islets 41 .Another protein, p300 (EP300), is a transcriptional coactivator, and many β-cell transcription factors require p300 along with CBP protein.Studies have shown that p300 is a limiting cofactor for islet development, making it vital for β-cell function and health in vivo 42 .
In insulin resistance, glucose tolerance is decreased in response to β-cell dysfunction 43 .The β-cells dysfunction is caused by numerous factors, such as oxidative stress and in ammation, and the FoxO pathway is highly linked to these risk factors 44 .Also, the AMPK protein activity is reduced in skeletal muscles and liver reduces in insulin resistance, leading to reduced free fatty acid oxidation and a lesser glucose intake, which deteriorates glycemic control 45 .Peroxisome proliferator-activated receptor (PPAR-γ) is a nuclear hormone receptor expressed primarily in the adipose tissues 46 .Apart from increasing insulin sensitivity in peripheral tissues, PPAR-γ has also been shown to enhance the glucose-sensing ability of pancreatic β-cells.It has been also shown to improve glucose homeostasis by directly affecting the liver and pancreatic β-cells 47 .Furthermore, GSK3B is postulated to be a potential kinase that induces insulin resistance.It can directly phosphorylate the IR and IRS-1 at Ser residues, thereby attenuating the insulinstimulated phosphorylation of their Tyr residues 48 .
NA phytochemicals could also regulate the TNF signaling pathway, which induces many cascade reactions, such as stimulating the transcription factor NF-κB, in ammatory response 49,50 , and apoptosis 51 .Research has shown that the TNF signaling pathway, a negative feedback mechanism, inhibits cell death by activating NF-κB 52 .TNF-α induces in ammation in pancreatic islets, leading to apoptosis in pancreatic β-cells [53][54][55] .In addition, TNF-α down-regulates PI3K/Akt signaling pathway and activates transcriptional factor NF-κB, an essential modulator of pancreatic cell death [56][57][58] .The activated signal transduction eventually initiates pancreatic β-cell apoptosis by regulating several proteins such as Bcl-2 and Mcl-1 [59][60][61] .The Bcl-2 regulates the mitochondrial-mediated β-cell apoptosis triggered by proin ammatory cytokines.Few Bcl-2 family proteins also play important role in regulation of β-cell function and glucose metabolism 62 .T2DM is associated with impaired wound healing, resulting from complex pathophysiology involving vascular, immune, neuropathic, and biochemical components 13 .The network analysis showed that NA regulates MMP-9, which exhibited a protective role in diabetic mice by improving wound healing 63,64 .It suggests that NA could play a vital role in improving healing diabetic ulcers.
Our network analysis is supported by pre-clinical studies using phytochemicals of NA.Curcumin has been reported to inhibit TNF-α 65,66 , caspase-3 67 , and JNK phosphorylation 68,69 and induces Bcl-2 activity 67 .In addition, it has also been shown to upregulate PPAR-γ via AMPK activation 70 .Quercetin upregulates AKT expression and follows the AMPK-P38 MAPK pathway to induce glucose uptake, which may contribute to correcting insulin resistance via bypassing the GLUT4 translocation via insulin-regulated system 71 .
Ellagic acid exerts anti-diabetic activity by inducing insulin secretion and reducing glucose intolerance in pancreatic β-cells.Also, increased β-cell size and number in diabetic rats 72 .Also, epigallocatechin gallate has been reported to reduce oxidative stress, pro-in ammatory cytokines (TNF-α and IL-6), p53, and caspase levels, and upregulate Bcl-2 in diabetic rats suggesting its anti-in ammatory and anti-apoptotic action 73 .
Network analysis has revealed that NA may also be explored in other diseases.In our study, phytochemicals of NA were putatively associated with pathways involved in leukemia, anemia, infertility, renal failure, hepatitis, fatigue, dermatitis, hyperhidrosis, etc.Interestingly, the description of CL and EO in Ayurvedic classical texts also supports their use in tvak dosa (skin disorders), rasayana (rejuvenator), shotha (in ammatory disorders), sveda (excessive sweating), pandu (hematological disorders), etc. 74,75 .
The experimental studies further support these facts.CL and EO have been reported to treat tumor 76,77 , Alzheimer 78,79 , obesity 80,81 anxiety disorders 82,83 , infertility 84,85 , and anemia 86,87 .Indications of different NA phytochemicals correspond to synergistic effects of polyherbal formulas used in traditional medicine.Accordingly, NP seems to be an appropriate approach to study the complex traditional herbal formulations.

Materials And Methods
Phytochemical compounds of Nisha Amalaki.Identi cation of NA target genes.The target genes of phytochemical compounds from NA herbal formulation were identi ed using the similarity ensemble approach (SEA; http://sea.bkslab.org/) 92.It is a chemical similarity search-based prediction tool known worldwide for its accuracy 93,94 .Although the SEA approach account only for ~2,800 potentially active proteins as alternate binding targets, the method is coherent with the already identi ed druggable genome (~3000) [95][96][97] .
Target PPI network construction and module identi cation.The target genes selected above were used to build a PPI network using NetworkAnalyst 3.0 tool (http://www.networkanalyst.ca/) 98,99.The network construction was constrained to contain only the original seed proteins by choosing the zero-order interactions in order to avoid the "hairball effect."NetworkAnalyst 3.0 incorporates extensive PPI data from already published literature with experimental evidence accessible across various PPI related databases such as IntAct 100 , BIND 101 , MINT 102 , BioGRID 103 , and DIP 104 , integrated into InnateDB 105 .The tightly associated group of target proteins, also referred to as modules in the PPI network, was identi ed using the "module explorer" tool of NetworkAnalyst 3.0 that uses a random walk-based method for detecting modules.Wilcoxon rank-sum test was used to calculate the P-value of the modules 106 , and the modules with signi cant P-value (P ≤ .001)were selected.The selected module was analyzed using the NetworkAnalyzer tool v4.4.8 within Cytoscape v3.8.2 107 .
GO and KEGG pathway enrichment analysis.GO enrichment analysis, and KEGG pathway annotation were carried out to elucidate the role of target genes that interact with the phytochemicals of NA using the Database for Annotation, Visualization and Integrated Discovery (DAVID, https://david.ncifcrf.gov/)v6.8 108 and NetworkAnalyst 3.0 tool, respectively.The GO analysis provides a curated and predicted annotation of genes with standardized terms relating to cellular components, biological processes, and molecular functions.The GO term was restricted to P ≤ .01,which is based on the false discovery rate (FDR; Benjamini-Hochberg).The enriched KEGG pathways with adjusted P-value, i.e., FDR ≤ .01 were used for the subsequent analysis.
Gene-Disease network construction.To identify the diseases associated with the target genes in the signi cant module gene-disease network was constructed using the 'gene-disease associations' network mapping tool available on NetworkAnalyst 3.0 platform.This tool uses the literature curated gene-disease association data gathered using DisGeNET database (https://www.disgenet.org/).The DisGeNET database contains most comprehensive collections of genes and variants associated with human diseases 109 .
Construction of the M-T-P network.The NA metabolites, target genes, and the related KEGG pathway were all imported into Cytoscape v3.8.2 to establish a M-T-P network.The nodes denote metabolites, targets, and pathways in the network, while the edges denote the interaction between the nodes.
Gene-disease association network.The degree sorted network was constructed using Cytoscape v3.8.2.
The red diamonds and blue circles represent the target genes of Nisha Amalaki in Module 1 and the signi cant diseases with betweenness ≥ 50, respectively.Edges represent the interaction between genes and diseases.The orange arrows and green circles denote the T2DM related target genes and pathways, respectively.Edges denote the interaction between metabolites, targets, and pathways.

Figures
Figures

Figure 1 Work
Figure 1

Figure 2 Protein
Figure 2

Figure 3 Protein
Figure 3

Figure 4 GO
Figure 4

Table 3
Nisha Amalaki gene targets related to Type 2 diabetes mellitus