Inhibition of Aldose Reductase by Ginsenoside Derivatives via a Specific Structure Activity Relationship with Kinetics Mechanism and Molecular Docking Study

This present work is designed to evaluate the anti-diabetic potential of 22 ginsenosides via the inhibition against rat lens aldose reductase (RLAR), and human recombinant aldose reductase (HRAR), using DL-glyceraldehyde as a substrate. Among the ginsenosides tested, ginsenoside Rh2, (20S) ginsenoside Rg3, (20R) ginsenoside Rg3, and ginsenoside Rh1 inhibited RLAR significantly, with IC50 values of 0.67, 1.25, 4.28, and 7.28 µM, respectively. Moreover, protopanaxadiol, protopanaxatriol, compound K, and ginsenoside Rh1 were potent inhibitors of HRAR, with IC50 values of 0.36, 1.43, 2.23, and 4.66 µM, respectively. The relationship of structure–activity exposed that the existence of hydroxyl groups, linkages, and their stereo-structure, as well as the sugar moieties of the ginsenoside skeleton, represented a significant role in the inhibition of HRAR and RLAR. Additional, various modes of ginsenoside inhibition and molecular docking simulation indicated negative binding energies. It was also indicated that it has a strong capacity and high affinity to bind the active sites of enzymes. Further, active ginsenosides suppressed sorbitol accumulation in rat lenses under high-glucose conditions, demonstrating their potential to prevent sorbitol accumulation ex vivo. The findings of the present study suggest the potential of ginsenoside derivatives for use in the development of therapeutic or preventive agents for diabetic complications.


Introduction
According to the World Health Organization, approximately 200 million people-5.9% of the adult population worldwide-suffer from diabetes, a very common disease that has assumed epidemic proportions. As a result, it is expected that by 2025, the disease will have a significant impact on human health [1,2]. Diabetes mellitus (DM) is a complex and chronic metabolic disease characterized by hyperglycemia, a condition in which a significant amount of glucose enters in the polyol pathway, which is considered as the main cause of the

RLAR Inhibitory Activities of Ginsenoside Derivatives and Analysis of Structure-Activity Relation
To evaluate the potential activity of the ginsenosides against diabetic complications, we examined whether they inhibited RLAR, and as Table 1 shows, they did so significantly. The most potent RLAR inhibitors were ginsenoside Rh2, (20S) ginsenoside Rg3, (20R) ginsenoside Rg3, and ginsenoside Rh1, with IC 50 values of 0.67 ± 0.01, 1.25 ± 0.28, 4.28 ± 0.31, and 7.28 ± 0.27 µM, respectively, relative to the positive control quercetin (4.88 ± 0.71 µM). In addition, ginsenoside Rf, (20S) ginsenoside Rg2, protopanaxadiol, protopanaxatriol, (20R) ginsenoside Rg2, compound K, ginsenoside Rg1, and ginsenoside Rg5 demonstrated significant inhibitory activities against RLAR, with IC 50 values of 11.29 ± 1. 49 Structurally, the ginsenosides are characterized within the protopanaxatriol (PPT) or protopanaxadiol (PPD) groups ( Figure 1). The inhibitory effects on RLAR were examined between the ginsenosides and the target enzyme inhibition, which are listed as in Table 1. The ginsenosides' RLAR inhibitory activity is associated to multiple factors, such as the position of the hydroxyl groups, the number of sugar molecules, and the stereoselectivity. The different positions of the ginsenoside hydroxyl groups influenced the RLAR inhibition. The number of hydroxyl groups at the C-20 position without or with sugar fractions increased the ginsenosides' RLAR inhibition. The ginsenoside RLAR inhibitory activity of Rh1, Rh2, Rf, Rg2, (20R) Rg2, (20R) Rg3, (20S) Rg3, (20S), protopanaxatriol, and protopanaxadiol were likened scientifically. Ginsenoside Rh1 and Rh2 comprise a single hydroxyl group at the C-20 location with a single sugar moiety at the C-3 or C-6 location. This result exhibited the utmost effective RLAR inhibitory action, whereas two or more sugar moieties at the C-6 or C-3 location decreased the RLAR inhibitory action of ginsenoside Rf, (20S) Rg2, (20R) Rg3, (20S) Rg3, and (20R) Rg2. Similarly, protopanaxatriol and aglycone protopanaxadiol comprise a hydroxyl group at the C-20 location that are engaged in RLAR inhibitory action. Later, the presence of a hydroxyl group at the C-20 location with a single sugar moiety on the ginsenoside molecule plays a dynamic role in inhibiting RLAR and declines the action of two or more sugar moieties. Previously, Fatmawati et al. [35] reported that the hydroxyl group at the C-20 position of ginsenosides plays an important role in aldose reductase inhibition, which is similar to our finding.
The RLAR inhibitory action enhanced the number of sugar moieties, whereas ginsenoside molecule reduced the number. Ginsenosides with three or more sugar molecules, such as Rb1, Rc, Rb2, Ra1, Rb3, Ra2, Rs2, Rs1, Re, and Rd, exhibited weak or no RLAR inhibitory action as compared to ginsenosides Rf, Rh2, Rh1, (20S) Rg3, (20R) Rg3, (20S) Rg2, (20R) Rg2. Ginsenosides Rf, Rh1, Rh2, (20R) Rg3, (20S) Rg2, (20S) Rg3, (20R) Rg2, PPT, PPD, and compound K displayed the solid RLAR inhibition, demonstrating that a partial number of sugar moieties at the C-3, C-6, or C-20 location possibly plays a significant role in ginsenoside action, and the exchange of more sugar residues enhances the decreased action.   Recently, Zhang et al. [36] reported that a traditional Chinese herbal extract (referred to as the Shenqi Jiangtang granule), which contains ginseng stem and leaf extract, demonstrated aldose reductase inhibitory activity. The authors also identified several ginsenoside derivatives and tested them against aldose reductase using rabbit crystalline lenses. According to the authors' reports, the ginsenoside derivatives demonstrated significant inhibitory activity against aldose reductase, which differs from our data slightly, because we used a different source of lenses (rat lenses) and Zhang et al. [36] used rabbit crystalline lenses. The pure rat lens homogenate extract was used to test the AR activity of ginsenoside derivatives in the current study. However, there are some drawbacks to using rat lens crude homogenate because it contains other NADPH-dependent reductase enzymes that could affect specific AR activity [10]. As a result, for testing the AR activity of ginsenoside derivatives, we used a more selective pure grade HRAR enzyme.

HRAR Inhibitory Activities of Ginsenoside Derivatives and Analysis of Structure-Activity Relation
To evaluate the anti-diabetic activity of 22 ginsenoside derivatives, the inhibitory potential of human recombinant aldose reductase (HRAR) was evaluated using DL -glyceraldehyde as the substrate, and the results are expressed as IC 50  , and Rs2 were found to be inactive at the concentrations tested. To explain the relationship between structure and action, we examined the various capacity of ginsenosides to inhibit HRAR. This inhibitory action enhanced the number of sugar moieties but reduced the ginsenoside molecule. Ginsenosides with three or more sugar molecules, such as Rc, Ra1, Ra2, Rb1, Rb2, Rb3, Rs1, Rs2, Re, and Rd exhibited weak HRAR inhibitory action as compared with the ginsenoside compound K (one sugar residue), Rh1, Rh2, PPT (no sugar moiety), and aglycone PPD. Remarkably, PPT and aglycone PPD exposed solid HRAR inhibition, demonstrating that the absence of a sugar moiety at the C-6, C-3, or C-20 location on the aglycone possibly plays a significant role in ginsenoside action, and the exchange of any sugar residues enhanced the decreasing action. We also detected that the sugar linkage cloud impacted the inhibition of HRAR. Compound K and ginsenosides Rh1 and Rh2 have a sugar linkage at C-20, C-6, and C-3, respectively. It has analogous chemical structures, but HRAR is highly inhibited by compound K, with a glucose linkage at location C-20, then Rh1 or Rh2.
Furthermore, the number and position of ginsenoside hydroxyl groups influenced HRAR inhibition. On the ginsenosides, a hydroxyl group at C-3 or C-20 with a single sugar moiety increased HRAR inhibition. Compound K, ginsenosides Rh2, Rh1, and Rf, (20S) ginsenoside Rg3, (20S) ginsenoside Rg3, (20R) ginsenoside Rg3, (20S) ginsenoside Rg2, and (20R) ginsenoside Rg2, which contain a single hydroxyl group at the C-3 or C-20 position, were more effective in inhibiting HRAR than those devoid of any hydroxyl group. As a result, it is possible that the presence of a hydroxyl group at the C-3 or C-20 position on the ginsenoside molecule plays a critical role in inhibiting HRAR, whereas the activity decreased when other groups were substituted. Further, we observed the stereoselectivity of the 20(S) and 20(R) forms of ginsenosides, which are stereoisomers that depend upon the orientation of the C-20 hydroxyl in ginsenosides. As we saw in (20R) Rg2, (20R) Rg3, (20S) Rg2, and (20S) Rg3, the stereochemistry difference creates various HRAR inhibitory effects. Both (20R) Rg2 and (20R) Rg3 inhibited HRAR more effectively than (20S) Rg2 and (20S) Rg3. These findings suggest that the stereo-structure of the C-20 hydroxyl may alter ginsenosides' HRAR inhibitory action.
There were significant differences in the inhibitory potencies of ginsenosides against HRAR and RLAR, according to our findings. These results demonstrate clearly that the differences in potency are attributable to the ARIs' chemical structures, as well as different AR sources, such as the species, organ, and tissue [10]. We found that AR's susceptibility to various ARIs can exhibit striking differences, depending upon human and animal sources, which may occur, at least in part, because of different degrees of bulk tolerance for the various enzymes [37,38].

Enzyme Kinetics Analysis of RLAR and HRAR Inhibition with Ginsenoside Derivatives
To explain the mode of enzymatic inhibition of ginsenoside derivatives, we performed a kinetics analysis of RLAR and HRAR to determine the representative inhibitors ( Table 1). The types of inhibition and the inhibition constants (K ic ) of eight active ginsenosides were investigated using Dixon (Figure 2a Figure 3i,j,m,n, and Table 1, protopanaxadiol, (20S) ginsenoside Rg3, compound K, and protopanaxatriol exhibited a competitive mode of HRAR inhibition with an increased K m and an unchanged V max with the increase in concentration, and their respective K ic values were 0.72, 8.06, 1.59, and 2.61 µM. (20R) ginsenoside Rg3, ginsenoside Rh2, and ginsenoside Rh1 demonstrated a mixed type of HRAR inhibition, as the inhibitor concentration increased, K m increased and V max decreased, with K ic values of 11.08, 3.63, and 3.40 µM, respectively (Figure 3k,l,p), whereas K iu had values of 17.99, 9.22, and 8.31 µM, respectively. For the (20R) ginsenoside Rg2, the reduced effect of the V max with unaffected K m indicates the non-competitive inhibition mode, with a K ic value of 11.67 µM ( Figure 3o). As the K i value represents the concentration needed to form an enzyme-inhibitor complex, a lower K i value may manifest more effective inhibitors against RLAR and HRAR in the development of preventive and therapeutic agents.

Molecular Docking Analysis
Molecular docking studies were performed to predict the binding modes of the compounds investigated. The comparison of docked compounds was carried out with the cognate ligands, which were re-docked in their respective proteins for validation purposes.

Molecular Docking Analysis
Molecular docking studies were performed to predict the binding modes of the compounds investigated. The comparison of docked compounds was carried out with the cognate ligands, which were re-docked in their respective proteins for validation purposes. The crystal (X-ray) structures of desired receptors were downloaded from the Protein Data Bank [39,40], and docking studies were performed by LeadIT software (version 2.3.2; Sankt Augustin, Germany) using FlexX features with default settings [41]. The cognate ligands for the human aldose reductases were zenarestat and sulindac with PDB IDs: 1IEI and 3RX2, respectively. To justify the experimental results, binding interactions were explored to gain an in-depth insight and the key interactions of ginsenosides and cognate ligands were demonstrated.

Docking Interactions of Ginsenosides and Cognate Ligand with RLAR
As shown in Figure 4a−j, 3D interactions in the docking simulation took place between ginsenosides and within the RLAR active pocket. The results of the docking analysis of the cognate ligand (sulindac) of RLAR demonstrated that the main amino acids involved in the interactions are Phe122, Cys298, His110, Trp111, Tyr48, Trp20, Trp219, and Val47, which form a substrate binding cavity toward the nicotinamide adenine dinucleotide phosphate (NAP316). These amino acids are responsible for hydrogen bonds, charge-charge interactions, π-alkyl linkages, and π-π T-shaped interactions. The probable binding mode of ginsenosides within the active pocket of aldose reductase revealed similar interactions within the entrance of the hydrophobic cavity and hydrophilic head, thus stabilizing the ginsenosides. The key interactions and amino acid residues are reported briefly in Table 2. The ginsenosides exhibited conventional hydrogen bonds with Tyr48, Phe121, Phe122, Ser302, Glu120, Val47, Trp20, and Tyr48, in addition to the amino acids reported earlier. Moreover, His110 and Tyr209 demonstrated π-π stacked interactions. The other contacts ginsenosides demonstrated were π-sulfur with Cys298, π-lone pair by Trp219, π-sigma by Ile260, and π-alkyl interaction with Trp20, Trp79, Trp111, Trp219, Phe122, His46, and Phe121. Moreover, the binding modes of ginsenosides manifested a pose similar to that of sulindac, and noticeable interactions with the proteins were monitored. The detailed analysis of the binding modes revealed that the network of H-bonds with the di-and trihydroxy groups, hydrophobic interactions with the alkyl chain, and π-π interactions with the hexadecahydrocyclopenta[α]phenanthrene nucleus of the ginsenosides extended to the entrance cavity of the hydrophobic domain and thus stabilized the binding. Most of the ginsenosides were inclined toward the hydrophilic head and exhibited the most conspicuous interaction with His110 and Trp111. The R1 and R2 substituents occupied the hydrophobic cavity and were oriented toward the catalytic cavity of 3RX2, which produced the best-oriented conformation. These interactions can be instrumental factors in the binding of ginsenosides with protein in the presence of cofactor NAP316. In addition to strong hydrogen bonds, some of the compounds exhibited hydrophobic interactions, as illustrated in the 2D interaction diagrams ( Figure S1, Supplementary Materials). Hence, our results revealed that the putative binding modes of ginsenosides are extended toward the entrance pocket and within the catalytic domain of the receptors. (e) (f)  , yellow (πsulfur interactions), tea pink (π-π T-shaped and π-π stacked interactions), cyan (fluorine) and yellow (π-sulfur). The interactions are shown by green (conventional hydrogen bonding), yellow (π-sulfur interactions), tea pink (π-π T-shaped and π-π stacked interactions), cyan (fluorine) and yellow (π-sulfur).
The 3D interaction in Figure 5a-j revealed the binding of ginsenosides with the hydrophilic cavity with multiple H-bonding, alkyl linkages, π-alkyl interactions, and π-π stacking near the NAP350. As shown in the figures, the ginsenosides adopted the conformations at the entrance site, active pocket, and catalytic cleft of 1IEI well by exhibiting multiple bonds of several types attributable to many hydroxyl groups (H-bonding), the main hexadecahydrocyclopenta[α]phenanthrene nucleus (π-π T-shaped, π-alkyl and π-π stacked interactions), and the alkene chain (alkyl linkages). Therefore, the binding of ginsenosides within the aldose reductase (1IEI) was stabilized by the presence of several types of interactions and demonstrated 3D-binding interactions similar to those zenarestat exhibited. In addition to strong hydrogen bonds, some of the ginsenosides exhibited hydrophobic interactions, as shown in the 2D interaction diagrams ( Figure S2, Supplementary Materials).

Inhibitory Activities of Active Ginsenosides on Sorbitol Accumulation
We studied the effects of the 10 active ginsenosides on sorbitol accumulation in rat lenses to demonstrate their efficacy in ex vivo. Sorbitol can be created faster than fructose can be processed in a hyperglycemic state, resulting in sorbitol buildup. The polarity of the sugar alcohol promotes intracellular accumulation because it prevents easy membrane penetration and subsequent removal via diffusion. Thus, the intracellular accumulation of a polar sugar alcohol can cause a hyperosmotic effect, resulting in a fluid infusion to counteract the osmotic gradient. Fluid influx has been demonstrated to cause alterations in membrane permeability and the start of cellular disease [42,43]. Rat lenses treated in 25 mM glucose for 6 days exhibited enhanced intracellular sorbitol accumulation. As shown in Table 4, the ginsenosides protopanaxadiol, (20S) ginsenoside Rg3, (20R) ginsenoside Rg3, ginsenoside Rh2, compound K, protopanaxatriol, ginsenoside Rf, (20S) ginsenoside Rg2, (20R) ginsenoside Rg2, and ginsenoside Rh1 inhibited sorbitol accumulation effectively by 96.51%, 78.48%, 48.25%, 97.09%, 73.25%, 94.76%, 83.72%, 57.55%, 51.74%, and 94.18%, respectively, at a concentration of 5 µg/mL. The positive control (quercetin) inhibited sorbitol accumulation in rat lenses by 87.20% and reduced sorbitol levels in a culture medium containing a high glucose concentration. Aldose reductase inhibitors have already been demonstrated to prevent sorbitol accumulation and reduce diabetic consequences [44]; therefore, edible ginseng extracts and their major active ingredients (ginsenosides) are a promising therapeutic aid in the treatment of diabetes complications, as ginsenosides may exert comprehensive inhibitory effects against diabetic complications via the ARpolyol pathway and the insulin resistant and AGE formation systems, which are implicated highly in oxidative stress. In our previous study, ginsenoside derivatives exhibited AGE formation inhibitory activity [33], and antioxidant activities in our antioxidant assays, including the reactive oxygen species (ROS) and ONOO − systems, and improved glucose uptake in insulin-resistant HepG2 cells [34]. Thus, our previous and current results suggest that ginsenosides may possess good inhibitory activities against diabetic complications, as suggested by their significant antioxidant activity. Table 4. Inhibitory effect of rat lens aldose reductase-active ginsenoside derivatives on sorbitol accumulation in rat lens.

Compounds Sorbitol Content (mg)/Lens Wet Weight (g) Inhibition (%)
Blank (glucose-free) -- Results are presented as mean ± SD (n = 3). Values within a column marked with different letters are significantly different from each other (p < 0.05). Samples' concentration was used at 5 µg/mL on sorbitol accumulation in rat lens. a Quercetin was used as the positive control.

Assay for RLAR Inhibitory Activity
A rat lens homogenate was prepared with minor modifications according to the method reported by Hayman et al. [45]. Briefly, the lenses were removed from the eyes of Sprague-Dawley rats weighing 250-280 g. They were homogenized in sodium phosphate buffer (pH 6.2) that was prepared with dibasic sodium phosphate (Na 2 HPO 4 ·H 2 O, 0.66 g) and monobasic sodium phosphate (NaH 2 PO 4 ·H 2 O, 1.27 g) in 100 mL of double-distilled H 2 O. The homogenate was clarified by centrifugation at 10,000 rpm at 4 • C for 20 min and the resultant supernatant was frozen until use. A crude AR homogenate with a specific activity of 6.5 U/mg was used in all enzyme inhibition evaluations. The reaction solution consisted of 620 µL of 100 mM sodium phosphate buffer (pH 6.2), 90 µL of AR homogenate, 90 µL of 1.6 mM NADPH, and 9 µL of sample, while the substrate included 90 µL of 50 mM DL -glyceraldehyde. Quercetin, a well-known AR inhibitor, was used as a control. AR activity was determined by measuring the decrease in NADPH absorption at 340 nm over a 4 min period on an Ultrospec ® 2100pro UV/Visible spectrophotometer. SWIFT II Applications software (Amersham Biosciences, NJ, USA) was used for all data analyses. The inhibition percentage (%) was calculated as follows in Equation (1): (1) in which ∆Asample/min represents the test sample and substrate's reduction in absorbance over 4 min, and ∆A control/min represents the same reduction in absorbance, but with 100% DMSO (dimethyl sulfoxide) rather than the test sample. The 50% inhibition concentration is expressed as the mean ± SEM.

HRAR Inhibition Assay
HRAR inhibition was studied according to Nishimura et al. [38]. AR action was studied by measuring the reduction in NADPH absorption at 340 nm over a period of 1 min on a UV/visible spectrophotometer (Ultrospec ® 2100pro, Biochrom, Holliston, MA, USA). For HRAR inhibition, the HRAR enzyme (0.4 units), was used in all enzyme inhibition evaluations. ARIs (aldose reductase inhibitors) were applied as positive controls (quercetin and zenaresta). The inhibition percentage (%) was measured as designated in the RLAR assay. The DA sample/min value characterizes the decrease in absorbance for 1 min with the test samples and substrate. The ability of each sample to inhibit HRAR was observed as an IC 50 value (µM) and measured from the log-dose inhibition curve.

Lens Culture and Intracellular Sorbitol Measurement
Lenses were isolated from Sprague-Dawley rats through cultures with sterile conditions and an atmosphere (95% air and 5% CO 2 at 37 • C for 6 d) in TC-199 medium comprising 100 U/mL penicillin, 15% fetal bovine serum, and 0.1 mg/mL streptomycin. The ginsenosides were liquified in DMSO. The lenses were separated into three groups with a medium containing RLAR-active ginsenosides and 25 mM glucose. Each lens was placed in a medium (2.0 mL). The sorbitol was calculated by HPLC [49].

Molecular Docking Simulation in RLAR and HRAR Inhibition
To perform the molecular docking analysis, structures of the target proteins HRAR (PDB: 1IEI) [39], and RLAR (PDB: 3RX2) [40] were obtained from the Protein Data Bank. The binding pocket and binding poses of docked compounds were investigated together with cognate ligands, as reported previously [7,33]. The active pocket of each receptor was identified using the SiteFinder feature within the Molecular Operating Environment (version 2019. 0201, Montreal, QC, Canada) [50]. The structures of the target proteins/receptors were protonated by the AMBER99 force field followed by the energy minimization at RMSD gradient of 0.05 kcal/mol [40]. Three-dimensional (3D) structures of the ginsenosides were prepared using the builder utility of MOE builder default parameters, followed by energy minimization; subsequently, the interactions of ginsenosides and the proteins' cognate ligands were examined within the active pocket, and the results were exhibited in the form of 3D poses [50]. Energy minimizations for these structures were performed using the MMFF94x force field at the RMSD of 0.01 kcal/mol Å in MOE [51], and cognate ligands were downloaded from the RCSB Protein Data Bank. The molecular docking analysis was performed using LeadIT software (version 2.3.2, Sankt Augustin, Germany) by setting default parameters [41]. Before the analysis, a validation of the methodology was performed by docking the cognate ligands within their respective proteins followed by docking the ginsenosides within the active site of the target protein. The top 50 binding poses were selected and inspected by the HYDE visual affinity. The HYDE assessment is helpful in selecting the final poses based upon favorable interactions for each ligand [52]. The poses with the lowest binding score and the most favorable affinity were selected and visualized with the Discovery Studio Visualizer (version 21.1.0.20298; San Diego, CA, USA) [53].

Statistical Analysis
All experimental results were expressed as the mean ± SEM of triplicate experiments. Statistically significant values were analyzed using analysis of variance (ANOVA) and Duncan's test (Systat Inc., Evanston, IL, USA). A p-value < 0.05 was considered statistically significant.

Conclusions
In summary, this study investigated the RLAR and HRAR inhibitory activities of ginsenoside derivatives. Ginsenosides demonstrated high inhibitory activities in two in vitro assays, and their potentially beneficial effects on diabetic complications make ginsenosides an excellent ingredient for functional foods. Additionally, the eight most active ginsenosides demonstrated potent inhibitory activities on sorbitol accumulation in isolated rat lenses. Our data indicated as well that aglycone PPD or PPT, and a hydroxyl group with a single sugar moiety containing ginsenosides, inhibit aldose reductase activity in vitro, and the mode of action depends upon the ginsenosides' structure. These structure−function relations could be useful in designing new aldose reductase inhibitors based upon ginsenosides. Collectively, this study together with our previous finding on the AGE formation inhibitory activity, suggests that ginsenoside derivatives can be potent functional food ingredients as RLAR and AGE inhibitors and can be used as a naturotherapy for diabetic complications. Further investigation to clarify their beneficial/harmful effects in vivo are underway in our laboratory.