Gene ontology enrichment analysis of α-amylase inhibitors from Duranta repens in diabetes mellitus

Abstract

Background

Although α-amylase is the choice of target to manage postprandial hyperglycemia, inhibitors of this enzyme may get absorbed into the systemic circulation and modulate proteins involved in the pathogenesis of diabetes mellitus. Hence, the present study aimed to identify α-amylase inhibitors from Duranta repens via in silico and in vitro and predict their role in the modulation of multiple pathways involved in diabetes mellitus.

Methods

α-amylase inhibitory activity of hydroalcoholic extract/fractions (s) and pure compounds from D. repens was performed using in vitro enzyme inhibitory assay. Multiple open-source databases and published literature were used to retrieve reported phytoconstituents present in D. repens and their targets. The network was constructed between α-amylase inhibitors, modulated proteins, and expressed pathways. Further, hit molecules were also confirmed for their potency to inhibit α-amylase using in silico molecular docking and in vitro enzyme inhibitory assay. The glucose uptake assay was performed to assess the effect of hydrolcoholic extract/fraction(s) using rat hemidiaphragm.

Results

Fraction rich in flavonoids showed the highest α-amylase inhibitory activity with a IC₅₀ of 644.29 ± 4.36 µg/ml compared to other fractions. PI3K-Akt signaling pathway and p53 signaling pathway were predicted to be primarily modulated in the compound-protein-pathway network. Similarly, scutellarein was predicted as lead hit based on α-amylase inhibitory action, binding affinity, and regulated pathways. Further, α-amylase inhibitors were also predicted to modulate the pathways involved in diabetes complications like AGE-RAGE and FoxO signaling pathway. Fraction rich in flavonoids showed the highest glucose uptake in rat hemidiaphragm with an effective concentration of 534.73 ± 0.79 µg/ml.

Conclusions

The α-amylase inhibitors from D. repens may not be limited within the gastrointestinal tract to inhibit α-amylase but may get absorbed into the systemic circulation and modulate multiple pathways involved in the pathogenesis of diabetes mellitus to produce synergistic/additive effect.

Introduction

Pancreatic α-amylase catalyzes the hydrolysis of starch into smaller oligosaccharides (maltose, sucrose, etc.) and multiple α-(l-6 and 1–4) oligoglucans [1]; further hydrolyzed by α-glucosidase to convert into glucose for its systemic absorption; contributes in post-prandial hyperglycemia [2]. Hence, deceleration of starch digestion by inhibiting α-amylase can be considered as a key step to manage postprandial hyperglycemia [3]. Although molecules like acarbose, miglitol, and voglibose are clinically approved molecules to inhibit α-amylase and α-glucosidase, these agents are limited due to multiple side effects i.e. bloating, flatulence abdominal discomfort, and diarrhea; not recommended in diabetic complications [4]. Further, the current pharmacotherapy of Type 2 diabetes mellitus (T2DM) utilizes synthetic oral hypoglycaemic agents which are single target molecules [5]; however, T2DM is a polygenic condition in which multiple proteins are involved to shape complex network among them by regulating multiple pathways [6]. Hence, a modified approach is needed to manage T2DM; molecules can be designed to inhibit specific enzymes like α-glucosidase, α-amylase, DPP4, and PTPN1B but also regulate other proteins that contribute to diabetes pathogenesis. However, this approach could be more convenient if these groups of molecules are from traditional folk medicines as these agents exert a synergistic/additive effect by targeting multiple proteins rather than single [7].

D. repens (Verbenaceae), commonly identified as “Golden Dewdrop” composes the plethora amount of multiple phytoconstituents like flavonoids, alkaloids, polyphenols, and saponins [8] which have been reported to contribute in lowering the elevated blood glucose level in multiple experimental diabetic models [9–12]. Further, these groups of phytoconstituents are also α-amylase inhibitors [13] that may contribute to manage postprandial hyperglycemia. Similarly, previous reports have also identified few molecules from D. repens to manage postprandial hyperglycemia [14, 15]. Additionally, D. repens is also reported for its free radical scavenging capacity [8] which could be beneficial in plummeting oxidative stress in diabetes mellitus (DM). However, the present literature lacks the potency of D. repens to inhibit α-amylase followed by probable modulation of multiple pathways if they are absorbed from the intestine.

Hence, the study aimed to investigate the hydroalcoholic extract/fraction(s) of D. repens to inhibit α-amylase followed by gene-set enrichment analysis and network pharmacology in respect of multiple pathways involved in diabetes mellitus.

Materials and methods

Software, database, and chemicals used

Database: ChEBI (https://www.ebi.ac.uk/chebi/), PCIDB (https://www.genome.jp/db/pcidb), PubMed (https://pubmed.ncbi.nlm.nih.gov/), DIGEP-Pred (http://www.way2drug.com/GE/), STRING (https://string-db.org/), PubChem (https://pubchem.ncbi.nlm.nih.gov/), PASS (http://www.pharmaexpert.ru/passonline/), KEGG (https://www.genome.jp/kegg/), RCSB (https://www.rcsb.org/); Software: Openbabel, autodock4, Discovery Studio 2019, chemsketch (https://www.acdlabs.com/resources/freeware/chemsketch/) Cytoscape 3.5.1; Chemicals: α-amylase, acarbose, naringenin, scutellarein, kusaginin (Sigma-Aldrich). Rest of the chemicals was of analytical grade.

Plant collection, authentication, and extract/fraction preparation

The whole plant of wild-grown D. repens was collected from local areas of Belagavi and authenticated at ICMR-NITM; herbarium was deposited (accession number: 1406) for the same for future reference. The plant was washed, dried, and the coarse powder was extracted and fractionated as explained by Cos et al [16] with minor modifications using multiple solvents i.e. methanol, acetone, dichloromethane, petroleum ether, and 5% citric acid (Fig. 1).

Fig. 1.

Extraction and fractionation of hydroalcoholic extract of Duranta repens

In vitro α-amylase inhibitory activity

In vitro α-amylase inhibitory activity was performed as explained by Telagari & Hullatti [2]. Multiple concentrations of hydroalcoholic extract/fractions and pure compounds i.e. naringenin, scutellarein, and kusaginin were tested to evaluate α-amylase inhibitory activity; compared with acarbose using 1% soluble starch as a substrate. All experiments were performed in triplicates and percentage inhibition was calculated using the following formula

$$\text{inhibitory}\text{activity}\left(\%\right)=\left(1-\frac{\mathit{As}}{\mathit{Ac}}\right)\times 100$$

where “Ac” and “As” are the absorbance of control and sample respectively.

Mining of α-amylase inhibitors

Reported bioactives from D. repens were mined using open-source databases i.e. PCIDB and ChEBI and published literature (PubMed). Canonical SMILES for each compound was retrieved from PubChem database or drawn in chemsketch and applicable files were created using Discovery Studio 2019 [17] as required. Inhibitors of α-amylase were identified using regression models using the Prediction of Activity Spectra for Substance (PASS) [18] at a probable activity(Pa) > probable inactivity (Pi).

Gene expression and enrichment analysis

Compounds were queried for regulated proteins using DIGEP-Pred; includes proteins-based training set (consists of 1357 up- and 1204 down-regulated genes) [19] at Pa > 0.5 by each compound and their interaction was assessed using STRING [20]. Finally, the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway database was used to identify pathways related to diabetes/diabetic complications.

Network construction and analysis

Network representing the interaction of phytoconstituents-proteins-pathways was constructed using Cytoscape [21] and the network was analyzed using the “network analyzer” tool based on node size and count representing the edge count as “low values to small size” and “low values to bright colors” respectively.

In silico molecular docking

Ligand molecules as α-amylase inhibitors; predicted from PASS were minimized using mmff94 forcefield [22] and converted using .pdbqt format. The α-amylase crystallographic moiety was retrieved from RCSB protein data bank (PDB: 5VA9) which was in a complex with water molecules; removed using Discovery Studio 2019 to avoid docking interference and saved in .pdb format. Autodock4.0 [23] was used to dock ligand molecules with α-amylase to obtain ten different poses; pose scoring minimum binding energy was chosen to visualize ligand-protein interaction using Discovery Studio 2019.

Glucose uptake assay using the rat hemidiaphragm

Efficacy of glucose uptake in rat hemidiaphragm by hydroalcoholic extract/fraction(s) was estimated in the presence of insulin as explained by the previous method [24] with minor modifications after receiving approval from Institutional Animal Ethics Committee (IAEC) at KLE College of Pharmacy, Belagavi; resolution no. KLECOP/CPCSEA-Reg, No.221/Po/Re/S/2000/CPCSEA,Res.28 − 12/10/2019. Briefly, albino Wistar rats were fasted overnight and sacrificed with an overdose of anesthetic ether. Diaphragms were dissected out quickly with minimal trauma and divided it into two halves. The hemidiaphragm were placed in test tubes and incubated for 30 min at 37 °C in an atmosphere of 100% oxygen with shaking at 140 cycles/min; multiple concentrations of hydroalcoholic extract/fraction(s) were used to evaluate the efficacy for glucose uptake in the presence of insulin (1 mL of 0.25 IU/mL solution). Experiments were performed in triplicates and effective concentration (EC₅₀) was calculated using the following formula.

$$\text{efficacy for glucose uptake}\left(\%\right)=\left(1-\frac{\mathit{As}}{\mathit{Ac}}\right)\times 100$$

where “Ac” and “As” are the absorbance of the control (sampled at 0 min) and test (sampled at 30 min) respectively.

Statistical analysis

Data obtained from wet-lab experiments were analyzed using a linear regression curve and inhibitory concentrations were compared using one-way ANOVA followed by post hoc test wherever applicable. Similarly, curve fitting in the network was performed using linear and logarithmic values wherever applicable.

Results

In vitroα-amylase inhibitory activity

Among five different fractions, fraction rich in flavonoids showed the highest α-amylase inhibitory activity (IC₅₀:644.29 ± 4.36 µg/ml). Table 1; Fig. 2 summarize the inhibitory concentration and percentage inhibitory activity of hydroalcoholic extract/fraction(s)/hit molecules respectively.

Table 1.

α-amylase inhibitory activity of hydroalcoholic extract/fractions and hit molecules

Test agents	Phytoconstituents	IC50 (µg/ml)
Hydroalcoholic extract	Flavonoids, Saponins, Alkaloids, Polyphenols, Triterpenes, Steroids	704.99 ± 8.96
Fraction 1	Fraction rich in Flavonoids	644.29 ± 4.36
Fraction 2	Fraction rich in Saponins	709.61 ± 5.93
Fraction 3	Fraction rich in Alkaloids	703.85 ± 8.92
Fraction 4	Fraction rich in Polyphenols	658.77 ± 3.13
Fraction 5	Fraction rich in Steroids	726.55 ± 6.70
Kusaginin	Hit molecule	250.90 ± 4.26
Scutellarein	Hit molecule	326.80 ± 1.29
Naringenin	Hit molecule	256.30 ± 3.25
Acarbose	Gold standard	71.28 ± 0.45

All data are represented in Mean ± SD, IC₅₀: Inhibitory Concentration 50

Fig. 2.

α-amylase inhibitory activity of hydroalcoholic extract/fractions, kusaginin, scutellarin, naringenin, and acarbose

Mining of bioactive and identification of α-amylase inhibitors

A total of 51 different phytoconstituents were identified from the whole plant of D. repens; 13 were predicted to inhibit α-amylase with multiple phytochemistry i.e. flavonoids, coumarins, glycosides, steroids, and organic compounds. The Pa/Pi score of each compound to inhibit α-amylase along with their type is summarized in Table 2.

Table 2.

Hit molecules as α-amylase inhibitors from D. repens

Phytoconstituents	Compound type	Probability score to inhibit α-amylase		Count of regulated proteins
		Pa	Pi	Downregulated	Upregulated
Kusaginin	Hydroxycinnamic derivative	0.589	0.003	1	8
Scoparone	Coumarins	0.180	0.067	8	9
3,7,4’-Trihydroxy-3’-(8’’-acetoxy-7’’-methyloctyl)-5,6-dimethoxyflavone	Flavonoid	0.142	0.096	1	4
7-O-alpha-D-glucopyranosyl-3,5-dihydroxy-3’-(4’’-acetoxyl-3’’-methylbutyl)-6,4’-dimethoxyflavone	Flavonoid	0.175	0.070	2	4
Scutellarein	Flavone	0.154	0.085	13	16
Naringenin	Flavonoid	0.147	0.092	8	12
Durantoside III	Iridoid glycosides	0.142	0.096	1	3
Durantoside I	Iridoid glycosides	0.191	0.061	1	3
Durantoside II	Iridoid glycosides	0.129	0.111	1	3
(E)-cinnamic acid	Organic compound	0.441	0.011	10	17
(E)-p-methoxycinnamic acid	Organic compound	0.294	0.031	8	14
β-sitosterol glucoside	Phytosterol	0.123	0.119	3	8
Myristoleic acid	Tetradecenoic acid	0.513	0.005	11	15

Pa :Probable activity, Pi: Probable inactivity

Gene expression profiles, enrichment, and network analysis

Among 13 bioactive molecules, scutellarein was predicted to regulate the highest number of proteins; 16 were downregulated and 13 were upregulated. The count of regulated proteins by each molecule is summarized in Table 2. However, enrichment analysis identified only 7 molecules i.e. (E)-cinnamic acid, (E)-p-methoxycinnamic acid, β-sitosterol glucoside, myristoleic acid, naringenin, scoparone, and scutellarein to regulate multiple pathways involved in diabetes and its complications. Among them, (E)-cinnamic acid, (E)-p-methoxycinnamic acid, β-sitosterol glucoside, myristoleic acid was chiefly involved in the regulation of the PI3K-Akt signaling pathway. Similarly, naringenin, scoparone, and scutellarein were predicted to regulate the p53 signaling pathway (Table 3). However, the combined effect of 7 α-amylase inhibitors reflected the HIF-1 signaling pathway to be primarily modulated; represented in Fig. 3. Further, gene GO analysis of modulated proteins for biological process, cellular component, and molecular function is represented in Fig. 4. Similarly, the line was fitted for the topological coefficient (y = 0.549-0.027x, R-square 0.545), node degree distribution (y = 8.706-0.662x, R-square 0.269), between centralities (y = 0.002x^2.103, R-square 0.753), closeness centralities (y = 0.271x^0.164, R-square 0.667) and neighborhood connectivity distribution (y = 10.399-0.532x, R-square 0.746) for compounds-proteins-pathways network; represented in Fig. 5. Similarly, the relation between shared neighbors vs. frequency and frequency vs. path length is represented in Fig. 6.

Table 3.

Enrichment analysis of α-amylase inhibitors from D. repens

α-amylase inhibitors	#term ID	term description	Observed/ background gene count	false discovery rate	matching proteins in network (labels)
(E)-cinnamic acid	hsa04115	p53 signaling pathway	3/68	0.00220	CASP8, CCND2, MDM2
	hsa04933*	AGE-RAGE signaling pathway in diabetic complications	3/98	0.00370	CCL2, COL1A1, MMP2
	hsa04151	PI3K-Akt signaling pathway	4/348	0.01040	CCND2, COL1A1, FLT1, MDM2
	hsa04066	HIF-1 signaling pathway	2/98	0.03340	FLT1, HMOX1
	hsa04110	Cell cycle	2/123	0.04280	CCND2, MDM2
	hsa04068*	FoxO signaling pathway	2/130	0.04600	CCND2, MDM2
(E)-p-methoxycinnamic acid	hsa04115	p53 signaling pathway	3/68	0.00220	CASP8, CCND2, MDM2
	hsa04933*	AGE-RAGE signaling pathway in diabetic complications	3/98	0.00370	CCL2, COL1A1, MMP2
	hsa04151	PI3K-Akt signaling pathway	4/348	0.01040	CCND2,COL1A1, FLT1, MDM2
	hsa04066	HIF-1 signaling pathway	2/98	0.03340	FLT1, HMOX1
	hsa04110	Cell cycle	2/123	0.04280	CCND2, MDM2
	hsa04068*	FoxO signaling pathway	2/130	0.04600	CCND2, MDM2
β-sitosterol glucoside	hsa04115	p53 signaling pathway	2/68	0.02770	CHEK1, MDM2
	hsa04068*	FoxO signaling pathway	2/130	0.03070	CAT, MDM2
	hsa04110	Cell cycle	2/123	0.03070	CHEK1, MDM2
	hsa04114	Oocyte meiosis	2/116	0.03070	AR, PGR
	hsa04060	Cytokine-cytokine receptor interaction	2/263	0.03810	CCL2, TNFRSF1A
Myristoleic acid	hsa04933*	AGE-RAGE signaling pathway in diabetic complications	5/98	3.40E-05	CCL2, COL1A1, MMP2, PRKCA, RAC1
	hsa04151	PI3K-Akt signaling pathway	5/348	0.00300	COL1A1, GH1, MDM2, PRKCA,RAC1
	hsa04066	HIF-1 signaling pathway	3/98	0.00620	GAPDH, PRKCA, TIMP1
	hsa03320	PPAR signaling pathway	2/72	0.02360	ADIPOQ, PPARA
	hsa04211*	Longevity regulating pathway	2/88	0.02970	ADIPOQ, CAT
	hsa04972	Pancreatic secretion	2/95	0.03060	PRKCA, RAC1
	hsa04068*	FoxO signaling pathway	2/130	0.04640	CAT, MDM2
Naringenin	hsa04066*	HIF-1 signaling pathway	3/98	0.00230	HMOX1, NOS2, TIMP1
	hsa04115	p53 signaling pathway	2/68	0.01620	CASP8, MDM2
	hsa04114	Oocyte meiosis	2/116	0.02530	AR, PGR
	hsa04068*	FoxO signaling pathway	2/130	0.02920	CAT, MDM2
	hsa04310	Wnt signaling pathway	2/143	0.02980	CTNNB1, MMP7
Scoparone	hsa04066	HIF-1 signaling pathway	3/98	0.00170	HMOX1, NOS2, TIMP1
	hsa04115	p53 signaling pathway	2/68	0.01430	CASP8, CHEK1
	hsa04310	Wnt signaling pathway	2/143	0.03360	CTNNB1, MMP7
Scutellarein	hsa04115	p53 signaling pathway	4/68	9.07E-05	CASP8, MDM2, TP53I3, TP73
	hsa04066	HIF-1 signaling pathway	4/98	0.00029	EGLN1, HMOX1, NOS2, TIMP1
	hsa04068*	FoxO signaling pathway	3/130	0.01110	CAT, MDM2, SIRT1
	hsa04213	Longevity regulating pathway - multiple species	2/61	0.02570	CAT, SIRT1
	hsa04211*	Longevity regulating pathway	2/88	0.03930	CAT, SIRT1

*pathways involved in diabetic complications

Fig. 3.

Interaction of phytoconstituents between phytoconstituents, their targets, and modulated pathways

Fig. 4.

Gene GO analysis of modulated proteins for biological process, cellular component and molecular function

Fig. 5.

Relation of a between centralities vs. number of neighbors, b closeness centralities vs. number of neighbours c average network centralities vs. number of neighbours d number of nodes vs. degree, e number of nodes vs. stress centrality, and f topological coefficient vs. number of neighbours in the constructed network

Fig. 6.

Relation of a number of shared neighbours with frequency and b frequency with path length respectively

In silico molecular docking

Molecular docking identified top 5 hit bioactives as α-amylase inhibitors i.e. β-sitosterol glucoside (-9.4 kcal/mol), naringenin (-8.8 kcal/mol), 7-O-α-D-glucopyranosyl-3,5-dihydroxy-3’-(4’’-acetoxyl-3’’-methylbutyl)-6,4’-dimethoxyflavone (-8.9 kcal/mol), kusaginin (-8.7 kcal/mol), and scutellarein (-8.7 kcal/mol) from D. repens. Among them, kusaginin was identified to possess the highest number of hydrogen bond interactions with α-amylase via ASN298, GLU233, ASP300, TRP59, ASP356, and GLU63. The binding energy, number of hydrogen bond interactions, and residues of each phytoconstituent with α-amylase are summarized in Table 4. Similarly, the interaction of naringenin, kusaginin, and scutellarein with α-amylase is represented in Fig. 7.

Table 4.

Binding affinity and number of hydrogen bonds/residues of each ligand with α-amylase

Ligand	Binding energy (kcal/mol)	Number of hydrogen bonds	Hydrogen bond residues
(E)-cinnamic acid	-6.3	2	TRP59, GLN63
(E)-p-methoxycinnamic acid	-6.6	1	GLN63
3, 7, 4’-Trihydroxy-3’-(8’’-acetoxy-7’’-methyloctyl)-5, 6 dimethoxyflavone	-8.3	4	ARG10, ASP402, ARG421, SER289
7-O-α-D-glucopyranosyl-3, 5-dihydroxy-3’-(4’’-acetoxyl-3’’-methylbutyl)-6, 4’-dimethoxyflavone	-8.9	2	HIS305, ASP300
Durantoside III	-8.3	6	ASP356, HIS305, ASP300, THR163
Durantoside II	-8.3	5	GLY403, SER289, GLY334, ARG252
Durantoside I	-8.5	4	ARG195, ASP300, GLU233, TYR151
Myristoleic acid	-5.3	2	GLN63
Naringenin	-8.8	2	GLN63, HIS299
Scoparone	-6.3	-	-
Scutellarein	-8.7	2	HIS299, GLN63
β-sitosterol glucoside	-9.4	2	ASP356, ASP353
Kusaginin	-8.7	6	ASN298, GLU233, ASP300, TRP59, ASP356, GLU63

Fig. 7.

Interaction of a kusaginin b naringenin and c scutellarein with α-amylase

Glucose uptake assay using rat hemidiaphragm

Fraction rich in flavonoids from D. repens showed the highest efficacy for uptake glucose in the presence of insulin (EC₅₀: 534.73 ± 0.79 µg/ml). Table 5 summarizes the EC₅₀ of each fraction and hydroalcoholic extract for glucose uptake in rat hemidiaphragm.

Table 5.

EC₅₀ of hydroalcoholic extract/fraction(s) of D. repens for glucose uptake in rat hemidiaphragm

Test agents	Phytoconstituents	EC50 (µg/ml)
Hydroalcoholic extract	Flavonoids, Saponins, Alkaloids, Polyphenols, Triterpenes, Steroids	723.46 ± 6.49
Fraction 1	Fraction rich in Flavonoids	534.73 ± 0.79
Fraction 2	Fraction rich in Saponins	603.99 ± 2.92
Fraction 3	Fraction rich in Alkaloids	642.74 ± 5.14
Fraction 4	Fraction rich in Polyphenols	559.16 ± 8.12
Fraction 5	Fraction rich in Steroids	765.37 ± 13.44
Metformin	Gold standard	169.38 ± 2.44

All data are represented in Mean ± SD, EC₅₀: Effective concentration 50

Discussion

The conventional drug invention/identification utilizes the concept of “lock and key” and “one gene-one disease drug track” by targeting the single protein [25] which has failed many times in multiple polygenic conditions like DM. Consequently, the practice of identifying new drug molecules for pharmacotherapy of multiple diseases, chiefly polygenic conditions like T2DM needs to be modified in such a way that the compound could target multiple proteins involved in disease pathogenesis in a synergistic mechanism. For this purpose, phytoconstituents from folk medicines can be utilized as they possess minimal side effects; can be evaluated by constructing “in house libraries” and matching with drug-like substances. Previous studies also identified few lead molecules to regulate multiple proteins and pathways via gene-set enrichment analysis [26] and network pharmacology [25, 27]. Further, Gene Ontology enrichment analysis (GO enrichment analysis) helps to perform the enrichment analysis of gene sets to identify GO terms that are under- or over-represented by using annotations of testable gene sets which may have an association with disease phenotypes. The analysis of gene set in enrichment analysis depends upon the background frequency; represents the gene annotated in the background set and sample frequency; represents genes annotated in the input list [28]. Hence, the present study also investigated hydroalcoholic extract and multiple fractions(s) rich in polyphenols, alkaloids, saponins, steroids, and flavonoids for α-amylase inhibitory activity followed by identification of hit molecules and their gene-set enrichment analysis. Further, hit molecules i.e. kusaginin, scutellarein, and naringenin were also investigated individually to assess their α-amylase inhibitory activity.

Fraction rich in flavonoids from D. repens was found to possess higher α-amylase inhibitory activity compared to other fractions conceivably due to higher binding affinity with an enzyme. Although polyphenols, alkaloids, saponins, and steroids rich fractions were identified to inhibit α-amylase, our cheminformatic investigation failed to spot lead enzyme inhibitors from these groups of phytoconstituents which needs to be further investigated. Further based on molecular docking and gene-set enrichment analysis we attempted to identify hits as α-amylase inhibitors. This approach was investigated based on probable activity to inhibit α-amylase, the binding affinity of phytoconstituents with an enzyme and modulated proteins and pathways. Based on the above parameters, scutellarein was predicted as a lead molecule to inhibit α-amylase which was confirmed via in vitro methods.

PI3K/Akt signaling pathway acts as a stabilizer to maintain the normal homeostatic condition and its abnormalities may lead to an eruption of a polygenic condition like diabetes and obesity [29]; was regulated by three phytoconstituents i.e. myristoleic acid, (E)-p-methoxycinnamic acid, and (E)-cinnamic acid. Since T2DM is a polygenic condition [6] and its pathogenesis includes insulin resistance and gradual damage in pancreatic β-cells, there is altered signaling in adipose tissue and skeletal muscle (inhibition of glucose uptake), pancreas (damage to β-cells), liver (increase in gluconeogenesis) and brain (progression of Type III DM) [30]. Hence, modulation of this pathway could be beneficial in managing T2DM as it regulates the insulin signaling pathway, enhances glucose uptake, maintains glucose homeostasis, and promotes insulin action [31]. However, the p53 signaling pathway was predicted to be commonly modulated by all phytoconstituents. The p53 signaling pathway is well recorded to repress GLUT1 and GLUT4 (decreases glucose influx), downregulates peroxisome proliferator-activated receptor-γ coactivator 1α, and as insulin receptor promoter. Further, it also inhibits glycolysis by inhibiting phosphoglycerate mutase, hexokinase 1 and 2, and glucose-6-phosphate isomerase; also concerned with an undemonstrative rate for glycogen synthesis [32]. Additionally, the p53 signaling pathway also maintains pancreatic β-cell mass. Further, this pathway also regulates mitochondrial metabolism of pyruvate; also inhibits embryonic malformation in diabetic conditions [26].

A single compound can modulate multiple proteins and pathways [25, 26, 33, 34], and it is to be understood that the mode of activity represented by a compound may not be similar compared to a group of alike compounds. For example, in the present study, PI3K/Akt signaling pathway was primarily modulated by three major constituents i.e. myristoleic acid, (E)-p-methoxycinnamic acid, (E)-cinnamic acid and also possess the binding affinity with α-amylase. However, combined network interaction of all α-amylase inhibitors, regulated, and pathways reflect the expression of the p53 signaling pathway. This suggests the choice of suitable molecules to target the specific protein followed by the regulation of multiple proteins involved in particular pathogenesis rather than a single hit molecule.

Management of diabetes complications is one of the challenging aspects of diabetic patients [35]. Additionally, some clinically approved anti-diabetic agents are also suspected as hepatotoxic, nephrotoxic, and are contraindicated in pregnancy [35–37]. Hence, along with the pharmacotherapy of T2DM, focus over diabetic complications could be a smarter approach. The identified lead hits and their combination may fulfill this aspect by modulating the AGE-RAGE signaling pathway (diabetic neuropathy, nephropathy, and retinopathy) [38], FoxO signaling pathway (diabetic nephropathy) [39], and Longevity regulating pathway (diabetic nephropathy) [40].

Hit molecules obtained from docking analysis, gene set enrichment analysis, and network pharmacology approaches were further confirmed for their α-amylase inhibitory activity. This helped to identify kusaginin as a potent α-amylase inhibitor as predicted by PASS and docking hit compared to scutellarein and naringenin. Unfortunately, this inhibitor failed to modulate pathways involved in DM. Although scutellarein was predicted for higher Pa to inhibit α-amylase and equivalent binding energy to naringenin, the IC₅₀ was significantly higher (p < 0.001) compared to kusaginin and naringenin. These observations reflect the combination of phytoconstituents should be carried for further investigation instead of a single lead hit.

In the present network, multiple parameters have been utilized to assess the constructed phytoconstituent-protein-pathway network. The closeness of each node to other nodes within the whole network is explained by “closeness centralities” to assess the superior broadcaster in the whole network [41]. Network analysis identified the closeness centralities of nodes were dependent on neighbors by 83.4%. Similarly, “between centralities” identifies the node to act as a bridge between other nodes i.e. reflects the shortest path to connect one node to another node by counting the fall of one node over other and helps to identify the flow around the system [41] which was found to be 75.3%. The involvement of one node to another node is explained by the “topological coefficient”; assigns zero to the nodes having no or one neighbor [42]. The topological coefficient of the constructed network was explained by neighbors at 54.5%. Similarly, “neighborhood connectivity” distribution explains the average connectivity of all nodes with their neighbors [42] which was defined by neighbors at 74.6%. “Node degree distribution” explains the relationship between in-degree and out-degree node distribution [42]; was explained by their degrees at 26.9%. Similarly, “stress centrality” measures the relation of “n” number of nodes with stress “s” for multiple values by grouping the values into bins as {{0}}; [1, 10); [10, 100);.... and assessing stress values to grow exponentially which was explained by stress centralities at 8.9%.

The high expression of the p53 signaling pathway could play a prime role in the long term progression of T2DM pathogenesis due to the downregulation of GLUT1 and GLUT4 gene expression. [43] Since GLUT4 plays an important role in the glucose uptake to the skeletal muscle [44], the involvement of the p53 signaling pathway may alter this mechanism and could be the outcome of insulin-resistant in T2DM. Further, GLUT1 has an association with TGF-β to contribute to diabetic nephropathy [45]. Likewise, the p53 pathway also acts as a negative regulator of glycolysis and a positive regulator of gluconeogenesis [46]. Glucose uptake assay by rat hemidiaphragm helps to understand the peripheral glucose uptake [24] which is also disturbed by the p53 signaling pathway. In the present study, we identified the fraction rich in flavonoids to possess α-amylase inhibitory activity as well as efficacy to enhance the glucose uptake to the rat hemidiaphragm which could be due to the combined action of modulated proteins i.e. MDM2, CASP8, CCND2, CHEK1, TP53I3, and TP73 of p53 signaling pathway by naringenin, scoparone, scutellarein, (E)-cinnamic acid, (E)-p-methoxycinnamic acid, and β-sitosterol glucoside. Further, among three hit α-amylase inhibitors i.e. kusaginin, scutellarein, and naringenin, two were flavonoids that could have contributed a major role in this process.

Conclusions

The present study dealt to identify the potential α-amylase inhibitors from D. repens and predict the probably modulated proteins/pathways by them. Likewise, the study also reflected the management of a polygenic condition like DM could be possible if multiple proteins are targeted to produce a synergistic/additive effect. To assess the potency of the modulation of the p53 signaling pathway we utilized ex vivo experiment for glucose uptake assay which identified the flavonoids to have a major contribution. Further, these agents need to be tested to assess their role in maintaining pancreatic β-cell mass in experimental models which is the future scope of the present study. Limiting to the present study, the enrichment analysis of modulated proteins is based on the database query and computer simulations which need to be further validated by experimental procedures.

Abbreviations

ChEBI

Chemical Entities of Biological Interest

DM

Diabetes Mellitus

DPP4

Dipeptidyl peptidase-4

GLUT

Glucose transporter

IC₅₀

Inhibitory Concentration 50

ICMR-NITM

Indian Council of Medical Research -National Institute of Traditional Medicine

KEGG

Kyoto Encyclopedia of Genes and Genomes

Pa

Probable activity

PASS

Prediction of Activity Spectra for Substances

PCIDB

PhytoChemical Interactions DB

PDB

Protein Data Bank

Pi

Probable inactivity

PTPN1B

Protein tyrosine phosphatase 1B

RCSB

Research Collaboratory for Structural Bioinformatics

STRING

Search Tool for the Retrieval of Interacting Genes/Proteins

T2DM

Type 2 Diabetes Mellitus

Funding information

This work has not received any funds from national and international agencies.

Data availability

Data will be provided in case of a request.

Compliance with ethical standards

Conflict of interest

There are no conflicts of interest to declare.

Consent for publication

Not Applicable.