Ginsenoside Rg3 stereoisomers differentially inhibit vascular smooth muscle cell proliferation and migration in diabetic atherosclerosis

Abstract Ginsenoside 20(R/S)‐Rg3, as a natural peroxisome proliferator‐activated receptor gamma (PPARγ) ligand, has been reported to exhibit differential biological effects. It is of great interest to understand the stereochemical selectivity of 20(R/S)‐Rg3 and explore whether differential PPARγ activation by Rg3 stereoisomers, if it exists, could lead to differential physiological outcome and therapeutic effects in diabetic atherosclerosis. Here, we investigated the binding modes of 20(R/S)‐Rg3 stereoisomers in the PPARγ ligand‐binding domain (PPARγ‐LBD) using molecular modelling and their effects on smooth muscle cell proliferation and migration induced by advanced glycation end products (AGEs). The results revealed that 20(S)‐Rg3 exhibited stronger antiproliferative and antimigratory effects due to stronger PPARγ activation. To validate the in vitro results, we used a mice model with diabetic atherosclerosis and obtained that 20(S)‐Rg3 markedly reduced the plaque size secondary to reducing the proliferation and migration of VSMCs, while the plaques were more stable due to improvements in other plaque compositions. The results shed light on the structural difference between Rg3 stereoisomers that can lead to significant differential physiological outcome, and the (S)‐isomer seems to be the more potent isomer to be developed as a promising drug for diabetic atherosclerosis.

Recent experiments have demonstrated that ginsenoside Rg3, which contains 2 neighbouring hydroxyl groups near and on the chiral centre C-20, can act as a natural ligand of PPARc. 10 Angiogenesis assay found that both Rg3 stereoisomers can induce differential angiogenesis effects via PPARc, and the PPARc agonist activity of 20(S)-Rg3 is 10 times stronger than that of 20(R)-Rg3. 11 A fluorescence polarization and total internal reflection fluorescence (FP-TIRF) binding study also confirmed that only 20(S)-Rg3 can quantitatively bind to the PPARc ligand-binding domain (PPARc-LBD). 12 To further understand the stereochemical selectivity of Rg3 enantiomers, it is timely and of great interest to model the binding modes of 20(R/S)-Rg3 in the PPARc-LBD.
PPARc is a member of the nuclear receptor superfamily of ligand-inducible transcription factors and regulates multiple pathways involved in the development of diabetes and CVDs. 13,14 Recent studies have implied the role of PPARc in regulating vascular smooth muscle cell (VSMC) proliferation and migration, an essential event in the development of diabetic atherosclerosis. 15,16 Under diabetic conditions, the accumulation of hyperglycaemia-induced AGEs and activation of the receptor for AGEs (RAGE) are key factors mediating these events. [17][18][19] The objective of this study was to investigate stereo-selective binding of Rg3 enantiomers to PPARc based on the stereochemical structures and to explore whether differential PPARc activation by Rg3 stereoisomers could lead to differential effects on AGEsstimulated proliferation and migration of VSMCs and diabetic atherosclerosis formation.

| Simulation and calculation
The initial co-ordinates of the protein were taken from the crystal structure of the PPARc-LBD complexed with the full-agonist LT160 Essential Dynamics Analysis (EDA) was based on the last 10-ns equilibrated trajectory for each system. The analysis followed the "Interactive Essential Dynamics" method closely. 24

| Cell culture and treatment
Mouse vascular smooth muscle cells (MOVAS cells, ATCC) were seeded into 6-well plates in routine Dulbecco's modified Eagle's medium with 10% foetal bovine serum for 24 hours, and then the culture medium was changed to 0.1% serum medium. VSMCs were pre-treated with 20(R/S)-Rg3 (25 lmol/L, dissolved in 0.1% DMSO; Felton, China) in the presence or absence of GW9662

| Animal model
All experimental procedures were performed in accordance with guidelines for Institutional Animal Care and were approved by the Animal Ethics Committee of Shandong University. Eight-week-old male ApoEÀ/À mice (n = 60) were randomly divided into the following 6 groups (n = 10 per group): non-diabetic control group, DM (diabetes mellitus) + placebo group, DM + 20(R)-Rg3 group, DM + 20 (R)-Rg3 + GW9662 group, DM + 20(S)-Rg3 group and DM + 20(S)-Rg3 + GW9662 group. Diabetes was initiated by the administration of 5 daily intraperitoneal injections of 50 mg/kg streptozotocin (STZ) in citrate buffer (0.05 mol/L; pH 4.5). Mice with blood glucose levels of >300 mg/dL at 2 weeks after the initial STZ administration were considered diabetic and were included in the DM cohorts. Mice received a normal chow for the remaining 10 weeks.
During the 6th-10th weeks, mice were given Rg3 at a dose of 10 mg/kg i.p. once 2 days, 25 with oral gavage of GW9662 at 3 mg/kg per day. 26

| Biochemical measurements
Serum lipid profiles, including total cholesterol, triglyceride levels and glucose concentration, were measured by enzymatic assay with the use of an automatic biochemical analyser (Roche Cobas Integra 800, Switzerland).

| Histology and immunohistochemistry
To assess overall burden and distribution of atherosclerosis, en face lesion staining with Oil Red O was performed as previously described. 27 The remaining staining involved cross sections of the aortic roots (predilection site of atherosclerosis). The sections were stained with haematoxylin and eosin (H&E) following a standard protocol of our laboratory. Haematoxylin was applied for 4 minutes followed by a 20-second differentiation in ammonia, after which eosin was applied for 20 seconds. The content of lipids and collagen of aortic plaques was detected by Oil Red O staining and Picrosirius Red staining, respectively. 28 The immunohistochemical staining was performed as previously described. 27  Olympus BX53 fluorescence microscope. The results were analysed using the Image-Pro Plus 6.0 software.

| Statistical analysis
All experiments were repeated at least 3 times, and data were presented as the mean AE SEM. Statistical analysis was carried out using ANOVA followed by Tukey's post hoc test (GraphPad Software, USA). P < .05 was considered significant.

| Characteristics of the binding modes of 20 (R/S)-Rg3 stereoisomers
The initial reasonable poses of 20(R/S)-Rg3 enantiomers in the LBP were obtained from computational docking via regarding the orientation of the full-agonist LT160 as template. Totally 6 candidates (S1, S2, S3 for 20(S)-Rg3 and R1, R2, R3 for 20(R)-Rg3) were submitted to MD simulations (Table S1). The average charge-clamp distance and MM/PB-SA estimated binding energy (Table S2), as well as the root-mean-squared fluctuation (RMSF) for each system ( Figure S1), all indicate that models S1 and R2 are the most probable binding modes of 20(R/S)-Rg3 enantiomers in PPARc. The orientations of the enantiomers in the 2 models were superimposed in Figure 1A.
EDA was then employed to differentiate these 2 models, and the backbone atom traces of the models are compiled in Table 1. The smaller backbone atom trace of model S1 implies that 20(S)-Rg3 stabilizes PPARc more than 20(R)-Rg3. The EDA projections of the 2 models are both split into 2 separated regions with respect to the first eigenvector and different to those of full agonist and antagonist. While the projection shape of model S1 seems more like that of the partial agonist LT127-bound system ( Figure 1B).

| Differential effects of Rg3 stereoisomers on
AGEs-induced VSMC proliferation  I G U R E 1 Simulation and calculation of the binding modes of 20(R/S)-Rg3 in models R2 and S1. A, Superposition of 20(R/S)-Rg3 in models R2 and S1 after simulation. The LBD in model S1 is shown as a green ribbon and that in model R2 is shown as a blue ribbon. 20(S)-Rg3 in model S1 is shown as green ball-and stick models, and 20(R)-Rg3 in model R2 is shown as blue sticks. Tyr473, Ser342, Glu343 and His449 in model S1 are shown as green wires and those in model R2 are shown as blue wires. The hydrogen bonds in model S1 are shown as orange dotted lines, and those in model R2 are shown as black dotted lines. B, EDA projections onto the first 2 eigenvectors of models S1 and R2 involved in those of the apo-bound, GW9662-bound, LT160-bound and LT127-bound systems. The calculation was based on the last 10-ns equilibrated trajectories. C, D, PPARc transcriptional activity in transfected 293T cells with and without AGEs incubation (n = 3). Values are presented as the mean AE SEM. *P < .05, **P < .01, ***P < .001 vs control; ##P < .01, ###P < .001 vs the 20(R)-Rg3 groups did not show significant cytotoxicity to non-stimulated VSMCs

| Biochemical parameters after Rg3 treatment in vivo
The serum levels of total cholesterol, triglyceride and fasting glucose of each group are depicted in Table S3. Total cholesterol and triglycerides levels did not differ among the diabetic apoE-/-groups. However, there was a significant reduction in fasting glucose levels in the 20(S)-Rg3-treated group compared to the placebo-treated diabetic group (20.35 AE 2.06 mmol/L vs 24.65 AE 0.82 mmol/L, P < .05). In addition, treatment with 20(R/S)-Rg3 was well tolerated and did not impair the apparent health or survival of the mice.

| Effects of Rg3 stereoisomers on atherosclerotic plaque size and frequency of intraplaque VSMCs
The relative en face lesion area of the entire aorta ( Figure 5A

| Effects of Rg3 stereoisomers on the proliferation and migration of VSMCs within atherosclerotic plaques
Co-immunofluorescence staining indicated a marked reduction in the frequency of proliferating VSMCs in the 20(S)-Rg3-treated group compared to that in the placebo-treated diabetic mice ( Figure 6A).
Immunohistochemical analysis revealed that 20(S)-Rg3 markedly decreased the MMP2 and MMP9 intensity within atherosclerotic lesions, more potent than 20(R)-Rg3 treatment ( Figure 7A). In addition, Rg3 treatment, especially 20(S)-Rg3, significantly reduced the protein levels of these markers as compared to the placebo-treated group (Figures 6B and 7B). For all above, GW9662 co-administration mostly reversed the effects of Rg3, indicating that Rg3 stereoisomers may differentially regulate the proliferation and migration of VSMCs via differential PPARc activation.
T A B L E 1 Backbone atom trace of S1 and R2 models bond, similar to that of rosiglitazone, 33,34 while the sterically strained binding pocket prevented the optimal interaction of 20(R)-Rg3 with Tyr473. 11 In this report, we further performed MD simulation and EDA to investigate the binding modes of 20(R/S)-Rg3 stereoisomers in PPARc-LBD. Here, we predicted that both of the 20(R/S)-Rg3 stereoisomers might act as PPARc partial agonists, but the binding of 20(S)-Rg3 to PPARc is more probable. This speculation was further confirmed by the results of the reporter gene assay, indicating that the differential binding modes of Rg3 stereoisomers in the PPARc-LBD contribute to differential PPARc activation at the cell level.
F I G U R E 7 Differential effects of the 20(R/S)-Rg3 stereoisomers on MMP expression within plaques. A, Representative immunohistochemical MMP2 and MMP9 staining and quantification analysis (n = 3, respectively). Scale bar: 50 lm. B, Western blot analysis of MMP2 and MMP9 protein expression within plaques. The bands are quantified by densitometric analysis, and protein expression was normalized to GADPH (n = 3, respectively). The results are expressed as the mean values AE SEM. *P < .05, **P < .01, ***P < .001 Ginsenosides such as Rb1, Rg1 and compound K have been reported to inhibit VSMC proliferation and neointimal hyperplasia 35 There is a concern that whether reduced smooth muscle after 20 (S)-Rg3 treatment may lead to a much less stable plaque and one that is more likely to rupture and cause further damage and possibly death, we further assessed the plaque compositions. To our great relief, 20(S)-Rg3 significantly improved other compositions in the plaque (increased the intraplaque content of collagen and decreased that of lipids and macrophages), and the plaque stability was actually increased. It has been reported that PPARc agonist could block classical activation of macrophages and attenuates inflammation in diabetic atherosclerosis, thus may promote a more favourable plaque morphology. 43,44 In addition, atherosclerosis is also affected by lipid profiles and glucose metabolism. 45,46 There was a trend towards a decrease in blood glucose levels in the 20(S)-Rg3-treated diabetic mice, which might also contribute to its anti-atherosclerotic effects. 47 In summary, this study provides new insights into the differential pharmacological effects of the 20(R/S)-Rg3 stereoisomers, in which the differential binding modes in the PPARc-LBD account for differential PPARc activation, and stronger PPARc activation elicits more effective downstream activity, leading to more significant suppres-