In Silico Evaluation of the Antioxidant, Anti-Inﬂammatory, and Dermatocosmetic Activities of Phytoconstituents in Licorice ( Glycyrrhiza glabra L.)

: The global demand for herbal cosmetics is vastly increasing due to their health beneﬁts and relative safety. Glycyrrhiza spp. extracts are used in cosmetic preparations due to their skin-whitening, antisensitizing, and anti-inﬂammatory properties. The aim of this work is to computationally evaluate the bioactive constituents of licorice ( Glycyrrhiza glabra L.) that possess antioxidant, anti-inﬂammatory, and dermatocosmetic activities, and elucidate the dynamics of their molecular targets. The used methods are skin permeability prediction, target prediction, molecular docking, and molecular dynamic simulation (MDS). The results show that, at a skin permeation cut-off value of − 6.0 cm/s, nine phytoconstituents of licorice (furfuraldehyde, glucoliquiritin apioside, glycyrrhizin, isoliquiritin, licopyranocoumarin, licuraside, liquiritigenin, liquiritin, and liquiritin apioside) were workable. Molecular target prediction results indicate probability for tyrosinase, 11-beta -hydroxysteroid dehydrogenase 1 (HSD11B1), monoamine oxidase B, steroid 5-alpha-reductase 1, and cyclo-oxygenase-1. On the basis of molecular docking, glucoliquiritin apioside and glycyrrhizin had the best antioxidant, anti-inﬂammation, and dermatocosmetic activities. MDS results show that the complexes had good stability, and MMGBSA results indicate that the complexes had satisfactory binding energy. Overall, this study demonstrates that licorice extracts are potential antioxidants that could enhance histological dermal and epidermal properties, and reduce the level of inﬂammatory and wrinkling markers.


Introduction
Cosmetics are any bioactive-containing preparation that is intended for use on the external surface area of human or animal bodies with the aim of cleansing, perfuming, protecting, or treating certain diseases [1]. Cosmetic products from natural sources such as plants usually do not pose health risks, but due to exposure to some hazardous agents in the environment such as allergens, toxins, carcinogens, and endocrine disruptors, there is a need for the authentication of plant materials for cosmetic applications [2]. The global demand for herbal cosmetics is vastly increasing due to their health benefits [3]. There is ongoing research in the cosmetic industry to discover new tropically sourced products and ingredients as their raw materials that often have functional properties due to the differential climatic and topographical settings [4].
Phytocosmetics is a segment of cosmetology that utilizes plant species for cosmetic purposes such as beautification and medication for skin diseases, which include abscesses,

In Silico Pharmacokinetics
The SMILESs of each of the ligands were used for in silico absorption, distribution, metabolism, and excretion (ADME) screening on a SwissADME server (http://www. swissadme.ch) (accessed on 7 February 2023) [26] that was performed with the default parameters. Predicted skin permeation log k p from the in silico pharmacokinetics was based on a model by Potts and Guy [27] according to the following equation: log k p (cm/s) = 0.71 * log k ow − 0.0061 * MW − 6.3 where MW is the molecular weight of the compound, and log k ow (or log P o/w ) is the octanol-water partition coefficient, a physicochemical constant used to describe the lipophilicity of the penetrant [28]. Compounds with high skin permeation were noted for further analysis. Hierarchical clustering analysis was also performed on a ChemMine web server (http://chemmine.ucr.edu/) (accessed on 18 April 2023) as previously described by Fatoki et al. [29] using the SMILESs of the ligands.

In Silico Target Prediction
The selected ligands that possessed a high skin-permeability coefficient based on the predicted pharmacokinetics were used for target prediction on a SwissTargetPrediction server (http://www.swisstargetprediction.ch/) (accessed on 13 February 2023), where Homo sapiens was designated as the target organism [30].

Molecular Docking Studies
Molecular docking studies were conducted as described by Fatoki et al. [29]. Briefly, the three-dimensional structures of 10 standard molecular target proteins for antioxidant (superoxide dismutase and glutathione peroxidase), anti-inflammation (11B-hydroxysteroid dehydrogenase 1, lipoxygenase, cyclo-oxygenase and inducible nitric oxide synthase), and dermatocosmetic (tyrosinase, collagenase, hyaluronidase and elastase) activities were obtained from www.rcsb.org/pdb on the basis of literature reports [9,10,31]. The structure of ligands that possessed high skin permeation were subjected to 3D structure optimization using ACDLab/Chemsketch software and saved in mol format. PyMol software was used for ligand file conversion from mol into pdb and for the preparation of protein chain A with the removal of water and existing ligands. Both ligands and proteins were prepared for docking using AutoDock Tools (ADT) v1.5.6 [32] with the default settings, and the output file was saved in pdbqt format. Molecular docking program AutoDock Vina v1.2.3 [33,34] was used for the docking experiment. After docking, close interactions of the binding of the target with the ligands were analyzed and visualized using ezLigPlot on an ezCADD web server (https://dxulab.org/software) (accessed on 18 February 2023) [35].

Protein-Protein Interaction Analysis
To established the relationship between the predicted targets of licorice phytochemicals with high skin permeability, the gene IDs of 10 standard molecular target proteins for antiinflammation, antioxidant, and dermatocosmetic activities were analyzed on the basis of literature reports [9,10,31] in combination with the predicted targets in humans for the protein-protein interaction (PPI) profile on a STRING web server (https://string-db.org/) (accessed on 13 February 2023) [36].

Protein-Ligand Molecular Dynamics Simulation
Molecular dynamics simulations were performed for 100 nanoseconds using Desmond, Schrödinger LLC [37][38][39]. The initial stages of the protein and ligand complexes for molecular dynamics simulation were obtained from the docking studies. Protein-ligand complexes were preprocessed using Maestro's protein preparation wizard, which also included complex optimization and minimization. All systems were prepared with the System Builder tool. A solvent model with an orthorhombic box was selected as the Transferable Intermolecular Interaction Potential 3 Points (TIP3P). The Optimized Potential for Liquid Simulations (OPLS)-2005 force field was used in the simulation [40]. The models were made neutral by adding 0.15 M NaCl counterions to mimic physiological conditions. The NPT ensemble (isothermal-isobaric: moles (N), pressure (P), and temperature (T) were conserved) at 300 K temperature and 1 atm pressure was select for a complete simulation. The models were relaxed before the simulation. The trajectories were saved after every 100 ps during the simulation, and the post-simulation analysis of the trajectories was conducted to determine the root-mean-square deviation (RMSD), root-mean-square fluctuation (RMSF), radius of gyration (Rg), solvent accessibility surface area (SASA), and protein-ligand interaction profile [37,38]. Prime molecular mechanics/generalized Born surface area (MMGBSA) was calculated as follows: where protein* means a protein from the optimized complex; ligand* means "a ligand from the optimized complex; NS means no strain, which is the binding/interaction energy without accounting for conformational receptor and ligand changes needed to form the complex [41,42].
The Desmond simulation package in Schrodinger was used for the MD simulation of two selected results of the docked complex (Figures 6-9). The hyaluronidase protein complex with glucoliquiritin apioside had an RMSD of about 1.6 Å, and the protein was quite stable during 20-100 ns of simulation time, while the RMSD ligand was stable at 25-100 ns ( Figure 6). Overall, the ligand was stable during the simulation. Hyaluronidase had Rg < 0.9 Å, the RMSF was mostly significant at the 190-200 and C-terminal amino acid residues, and the total SASA was about 1800 Å 2 . Figure 7 shows the high interaction of hyaluronidase with glucoliquiritin apioside that occurred on ASN39, GLY63, ILE73, SER76, SER77, GLN78, ASP129, GLU131, TYR202, ASP292 and TRP324 amino acid residues, and the profiles of glucoliquiritin apioside during the simulation.        Figure 8 shows that the RMSD of the inducible nitric oxide synthase (iNOS) complex with glycyrrhizin was about 1.6 Å, and the protein and ligand were both stable during 25-100 ns of the simulation time. In addition, Rg < 0.8 Å, RMSF was mostly significant at the N-terminal, 60-70, 170-190, and 310-330 amino acid residues, and total SASA was about 2000 Å 2 for myeloperoxidase. Figure 9 shows the high interaction of iNOS with glycyrrhizin that occurred on the PRO122, TRP194, ARG199, ILE201, GLY202, ALA262, ARG266, ALA282, ALA351, ASN354, TRP372, ARG381, TRP463 and LEU464 amino acid residues, and the profiles of glycyrrhizin during the simulation. The binding free energies of all complexes were calculated using MMGBSA at 0 and 100 ns. The results indicate a change in the binding energy of the complex of glucoliquiritin apioside and hyaluronidase from −73.732 to −43.085 kcal.mol −1 , while the binding energy of the glycyrrhizin and inducible nitric oxide synthase (iNOS) complex decreased from −91.602 to −74.874 kcal.mol −1 (Tables 4 and 5

Discussion
Therapeutic indication was reported for Glycyrrhiza glabra regarding its antiaging, anti-inflammatory, and antioxidant properties on the basis of its inhibitory extract activity on some enzymes such as elastase and tyrosinase, which led to increased collagen and elastin synthesis [31]. Antiaging activity is due to the free-radical scavenging action and the inhibition of lipoperoxidation by the herbal extracts [43]. Licorice phytochemicals are good anti-inflammatory agents that are useful for treating skin irritations, and in cosmetics for acne and sunburns [44].
Most of the selected licorice constituents for dermatocosmetic effects exerted good ADME properties such as low gastrointestinal absorption, not being BBB permeable and substrates of P-glycoprotein, and having log Kp values that are close to those of kojic acid (−7.62 cm/s) and quercetin (−7.05 cm/s), which were among the standards used in this study; these features ensure high dermal retention and low systemic bioavailability, and thus low side effects. Skin permeability (Kp) describes the rate of chemical permeation through the outermost layer in the stratum corneum of the epidermis [28]. A high log k oct value indicates high lipophilicity and is proportional to a qualitative indicator of penetration [28]. Therefore, substances with high lipophilicity persist in the lipophilic part of the skin, thus being useful for cosmetic purposes. The low GI absorption of cosmetic ingredients is generally desirable because it minimizes the potential for systemic exposure and related health risks.
Moreover, the current findings indicate that glycyrrhizin could modulate HSD11B1, HSD11B2, and HSD17B, while liquiritigenin could modulate COX-1 and possessed high binding affinity for COX-2. Licorice extracts could reduce the activities of HSD enzymes, causing greater amounts of cortisol to be produced in humans and ultimately interacting with mineralocorticoid receptors [24]. In vitro and in vivo experiments showed that licorice extracts possess therapeutic properties against colon cancer by inhibiting HSD11B2 and enhancing the glucocorticoid-mediated suppression of the cyclo-oxygenase 2 (COX-2) signaling pathway [47,48]. Glucoliquiritin apioside and glycyrrhizin have high binding affinities for hyaluronidase and elastase, whereas liquiritin and liquiritin apioside showed high binding affinities for collagenase. Hyaluronidases are a family of enzymes that catalyze hyaluronic acid, and are widely distributed in the body and particularly at the periphery of collagen and elastin fibers, which is an indication of their major role in skin aging [49]. Elastase is a serine protease that preferentially digests elastin, the highly elastic protein that works together with collagen to give the skin its shape and firmness [50]. Collagenases are enzymes that cleave collagen molecules within their helical region and are more generally involved in the degradation of extracellular matrix components, thus leading to skin wrinkles [51].
Tyrosinase is a copper-containing oxidase that plays a key role in melanogenesis, controlling the production of melanin, and it is mainly involved in the hydroxylation of L-tyrosine into L-DOPA (L-3,4-dihydroxyphenylalanine, monophenolase activity) and its further oxidation into dopaquinone (diphenolase activity). This enzyme's tyrosinase inhibitors are greatly concerning in the development of skin whitening agents [52,53]. Recently, a study indicated the synergistic effects of licorice combined with zinc in the treatment of pigmented skin disease, and that the combined preparation decreased tyrosinase, tyrosinase-related protein-1, microphthalmia-associated transcription factor, melanin formation, and cutaneous tissue injury [54]. A computational study also showed that glabridin and its semisynthetic derivatives are potential tyrosinase inhibitors that possess higher binding affinities than that of kojic acid [55].
Both glucoliquiritin apioside and glycyrrhizin had high binding affinities for lipoxygenase, whereas glycyrrhizin and liquiritigenin were predicted with high binding affinity for iNOS. Lipoxygenase, an iron-containing enzyme catalyzing the deoxygenation of polyunsaturated fatty acids into the corresponding hydroperoxides, plays a key role in inflammation [56]. A study reported that licorice extract cream with 10% concentration was more effective in lightening the skin than the concentrations of 20% and 40% were [57].
A molecular dynamics (MD) simulation was performed to determine the variation in the protein-ligand system at the atomic level, and articulate on the stability of the proteinligand complex in a dynamic environment [58]. An RMSD of about 1.6 Å was obtained for both complexes investigated in this study, which indicates that the proteins had undergone relatively small conformational changes and were, thus, stable during the simulation. In addition, Rg < 0.9 Å, which demonstrates the compactness of the protein and the proteinligand complex. Total SASA was in the range of 1800-2000 Å 2 , which indicates the surface area of proteins covered by polar and nonpolar interactions, and declines with an increment in macromolecular compactness. RMSF is useful in characterizing local changes along a protein chain. A study revealed that the tyrosinase-kojic acid complex had an RMSD of about 3.5 Å, an average Rg of nearly 0.5 Å, and SASA of approximately 2500 Å [55]. The binding potential of the ligand was quantitatively estimated using free binding energy calculation analysis with MM-GBSA [38,40,42]. The binding free energy clearly showed that the complexes were stable before and after the simulation with lesser binding energy, which could easily aid the metabolism from dermal compartments; glycyrrhizin and glucoliquiritin apioside were moderately bound to inducible nitric oxide synthase (iNOS) and hyaluronidase, respectively.

Conclusions
This study demonstrated that licorice (Glycyrrhiza glabra) comprises some active phytochemicals (such as glucoliquiritin apioside, glycyrrhizin, isoliquiritin, liquiritin, and liquiritin apioside) that possess high skin-permeability properties. These selected phytochemicals in licorice are potential antioxidants that enhanced dermal and epidermal histological properties, and reduced the level of inflammatory and wrinkling markers. Overall, glucoliquiritin apioside and glycyrrhizin had the best antioxidant, anti-inflammation, and dermatocosmetic activities. Although computational methods are invaluable tools in the development and safety assessment of cosmetic products, they have some limitations, which include the limited availability of data, the inadequate understanding of complex biological systems (e.g., effects on gene expression), limited predictive power (e.g., algorithms and models for prediction), variability between individuals (e.g., skin type, age, and ethnicity), and an incomplete understanding of safety endpoints (e.g., long-term effects of exposure). Therefore, in vitro and in vivo studies, and computational modeling, particularly physiological pharmacokinetics/toxicokinetics (PBPK/PBTK) in the dermal route, are required to validate these molecular pharmacological activities of licorice constituents in terms of their relevance as cosmetics.