Ginsenoside Rg1 protects against ischemic/reperfusion-induced neurotoxicity through miR-144/NF-E2-related factor 2/antioxidant response element pathway

Abstract: Ginsenoside Rg1 (Rg1), a purified compound from Panax ginseng, has been well documented to be effective against ischemic/reperfusion (I/R) neurotoxicity. However, the underlying mechanism is still obscure. In this study, we found that the anti-I/R effect of Rg1 is related to its anti-oxidative capacity, which is mainly regulated by the Nrf2/ARE pathway. Further study suggested that Rg1 contributes to the enhancement of Nrf2/ARE pathway, as manifested by increasing the dynamic peak content of Nrf2 prolonging the maintenance stage, and promoting the expression of ARE-target genes after oxygen glucose deprivation/reperfusion (OGD/R) in PC12 cells. Nrf2-siRNA application abolished these changes significantly. Moreover, the enhancement of the Nrf2/ARE pathway by Rg1 is independent of the disassociation with Keap-1, but is a result of post-translational regulations. We found that the Rg1 significantly reduced the expression of miR-144 , which down-regulates Nrf2 production by targeting its 3’-untranlated region, after OGD/R. Knockout of Nrf2 showed no effect on the expression of miR-144, indicating that miR-144 is an upstream regulator of Nrf2. Moreover, we identified that there is a direct binding between Nrf2 and miR-144 in the PC12 cells. Application of anti-miR-144 abolishes Rg1’s anti-OGD/R capacity significantly. Finally, we tested the role of miR-144 in Rg1’s anti-I/R effect by inhibiting miR-144 in the predicted ischemic penumbra when hypoxia inducible factor is activated. The results showed that loss of miR-144 abolished the anti-I/R effect of Rg1, which includes reduced infarct volume, improved neurological scores, attenuated oxidative impairment, as well as the activation of Nrf2/ARE pathway. Collectively, our findings demonstrated that oxidative stress after I/R was alleviated by Rg1 through inhibiting miR-144 activity and subsequently promoting the Nrf2/ARE pathway at the post-translational level. It provides a new target for Rg1 application in potential treatment of ischemic stroke.

Key words: Ginsenoside Rg1; ischemic/reperfusion; Nrf2/ARE; miR-144.

Introduction

Panax Ginseng has been used as a tonic to improve stamina and vitality in traditional Chinese medicine for thousands of years [1,2]. Ginsenoside Rg1 (Rg1), a purified saponin isolated from ginseng, is considered as one of the potent anti-stroke candidates by many recent studies [3-5]. A systematic review and meta-analysis revealed that there is a marked efficacy of Rg1 in experimental acute ischemic stroke, as manifested by its ability to reduce infarct volume and improve neurological score [6,7]. However, the underlying mechanism is still unclear.

Evidences have shown that oxidative stress is closely related to the pathological progression of ischemic/reperfusion (I/R) [8-11], which occurs by excessive production or deficiency in degradation of reactive oxygen species (ROS) [12]. There is a balance between ROS generation and scavenging under physiological conditions. I/R disrupts this balance from the initial stage of I/R as evidenced by increased ROS production, consumption of antioxidants, and finally induction of oxidative damage on proteins, lipids, DNAs and RNAs[13-15]. ROS scavengers, including edalavone, salvianolic acid, etc., have been shown to be beneficial for the treatment of ischemic stroke in clinical trials [15-17].

In addition to scavenging the ROS directly, emerging evidences have validated that activation of Nrf2 and the expression of its downstream genes may have protective effects against I/R [18,19]. Nrf2 is one of the master regulators for oxidative defense which acts through binding to the antioxidant response element (ARE) in the nucleus and promoting a series of anti-oxidative gene expressions[20]. The level of Nrf2 experiences an inverted “U” shape model after ischemic stroke, characterized by the robust increasing from 2 hours of OGD, peaking at 8 hours, and decreasing below baseline from 24 hours in transient MCAO mice[21]. Loss of Nrf2 exacerbates cerebral infarction and neurologic deficits in both transient- and permanent- middle cerebral artery occlusion (tMCAO and pMCAO, respectively) models, as manifested by reduction of oxidative tolerance, increase of TLR4 expression and NFКB activation, breakdown of blood brain barrier, and so on[22-24]. Activation of Nrf2 and its target genes may protect the brain against I/R [25,26]. Hence, Nrf2 is considered as a promising therapeutic target for defending against oxidative stress in stroke.

In addition to Keap1, the Nrf2/ARE pathway is regulated by many other factors, such as DNA methylation, histone acetylation and micro RNAs (miRNAs)[27]. MiRNAs are a group of endogenous small non-coding RNAs that negatively control gene expression by binding to their target sequences at the 3′-UTR of mRNAs[28]. Until now, it has been proven that miR-144, miR-153, miR-27 regulate Nrf2 production directly in neurodegenerative diseases [29,30]. However, whether miRNA-regulated Nrf2 is involved in I/R injury is still unknown.

In the present study, we revealed that the Nrf2/ARE pathway is activated by Rg1 treatment both in vitro and in vivo. Inhibition of Nrf2 abolished Rg1’s effect significantly. Interestingly, miR-144, one of the miRNAs interacts with Nrf2, was involved in this regulating process, rather than Keap1. Furthermore, we found the expression of miRNA-144 is independent from the activation of Nrf2, indicating that miR-144 is an upstream regulator of Nrf2 in I/R injury. Inhibition of miR-144 obliterated the anti-I/R effect of Rg1, as well as the activation of the Nrf2/ARE pathway in vivo. All of these results suggest that oxidative stress after I/R was alleviated by Rg1 by promoting the anti-oxidative defense of the Nrf2/ARE pathway at the post-translational level by inhibiting miR-144 activity, which provide a new drug-development target for Rg1 application in potential treatment of ischemic stroke.

Methods ?

Reagents

Rg1 (HPLC 98%) was provided by Jecui healthy Co., Ltd. (Yunnan, China), and its chemical structure is shown in Fig. 1a. DMEM medium and fetal bovine serum were purchased from Gibco (Grand Island, NY, USA). β-actin primary antibody, ascorbic acid, momotetrazolium (MTT) and dimethylsulfoxide (DMSO) were purchased from Sigma (St. Louis, MO, USA). The test kits for measurement of lactic dehydrogenase (LDH) were purchased from Nanjing Institute of Jiancheng Bioengineering (Nanjing, China). Primary antibodies to Nrf2, NQO-1, HO-1, glutamate cysteine ligase modifier subunit (GCLM), glutamate cysteine ligase catalytic subunit (GCLC) and Lamin A were purchased from Abcam (Cambridge, UK). All other chemicals and solvents used were of analytical grade unless otherwise specified.

PC12 Cell Culture and Oxygen Glucose Deprivation/Reperfusion (OGD/R)

Rat pheochromocytoma PC12 cells were obtained from Peking union medical college. The OGD/R was performed as previously described. Briefly, PC12 cells were cultured in DMEM with 10% FBS, 5% ES, 100 U/ml penicillin, and 100 U/ml streptomycin at 37 °C under a humidified atmosphere containing 5% CO₂. For OGD, cells were cultured in glucose-free Earle’s balanced salt solution supplemented with 20 mM Na₂S₂O₄ for 20 min. After that, they were incubated with conditioned DMEM with or without Rg1 at different concentrations (0.01, 0.10, 1.00 and 10.00µM) for indicated time.

Cell Viability Determination

Cell viability was measured with MTT assay. Briefly, PC12 cells were seeded in 96-well plates (5*10³ cells/well) and incubated for 24 h. The cells then underwent20 min of OGR, followed by incubation with conditioned DMEM with or without Rg1 at different concentrations (0.01, 0.10, 1.00 and 10.00µM) for 24 h. After that, they were incubated at 37℃ for 4 hours in 0.5 mg/ml MTT solution, and the media was carefully removed. 100μl of DMSO was added to each well to dissolve the violet formazan crystals. The absorbance at 570 nm was measured. All values were normalized to the control group.

OGD/R-induced neurotoxicity was also quantified by measuring lactic dehydrogenase (LDH) release from damaged cells according to the manufacture’s instruction (Nanjing Institute of Jiancheng Bioengineering). In brief, 150μl of LDH reaction reagent was added to 50μl medium taken from the cell culture wells. The absorbance was measured at 490 nm by a spectrophotometer plate reader. Data were expressed in the percentage of cell death that was calculated according to the manufacturer’s instruction.

ROS Determination

Intracellular ROS level was detected using the 20, 70-dichlorodihydrofluorescein diacetate (DCFH-DA) fluorescence assay. Briefly, cells were seeded in 96-well plates at the density of 7*10³ per well overnight. Rg1 (0.01, 0.10, 1.00 and 10.00µM) was added into culture for 24 hours followed with OGD. After that, cells were incubated in serum-free medium containing 25μM DCFH-DA at 37℃ for 30 min. ROS levels were measured by the fluorescence intensity of DCF at 525 nm after excitation at 488 nm on a fluorescence plate reader (Thermo Scientific Varioskan Flash).

Nuclei and Cytosol Extraction from Culture Cells

Nuclear and cytosolic protein fractionation was prepared from PC12 cells or brain samples using the nuclear-cytosol extraction kit (Applygen Technologies Inc, Beijing, China) according to the manufacturer’s instructions. Cells were washed with PBS, followed by detachment with trypsinization and centrifugation for 5 min at 800×g to collect cells. 250 μl Cytosol Extraction Buffer A was added to the cells and vortexed for 30 s, then incubated on ice for 10 min. 30 μl of Cytosol Extraction Buffer B was added to the mixture and vortexed for 10 s, followed by incubation on ice for 1 min. The tube was centrifuged for 5 min at 1000×g at 4 °C. The pellet and supernatant were treated respectively. Brain samples were lysed by addinge 10 times volume/quantity of cytosol extraction buffer A, and the subsequent protocol was same as previously described.

Western Blot Analysis
Proteins were separated depending on its size by SDS-PAGE on 15% polyacrylamide gels and transferred to polyvinylidene difluoride membranes (Millipore, USA). The membranes were blocked with 3% bovine serum albumin (BSA) and incubated with primary antibodies, then horseradish peroxidase-conjugated secondary antibodies. Proteins were then detected with the enhanced chemiluminescence plus detection system (PPLYGEN, China). The signal protein bands were scanned and analyzed with Gel-Pro Analyzer software (Media Cybernetics).

RNA Extraction and Real-Time PCR
Total cellular RNA was extracted using Trizol reagent.1 lg RNA was reverse transcribed to cDNA using Quantitect reverse transcription kit according to instructions (Qiagen). For real-time PCR analysis of HO-1, NQO1, GCLC and GCLM, the specific primers are listed below: rHO-1: forward (AGCGAAACAAGCAGAACCCA) and reverse (ACCTCGTGGAGACGCTTTAC); rGCLM: forward (CACAACTCAGGGGCCTTGTA) and reverse: (AAACCACCACATTCACGCCT); rGCLC: forward (GAGCGAGATGCCGTCTTACA) and reverse (AACACGCCTTCCTTCCCATT); rNQO-1: forward (GCCATCATTTGGGCAAGTCC) and reverse (TCCTTGTGGAACAAAGGCGA); rGAPDH: forward (CCGTATTCAGCATTCTATGCTCTC) and reverse (TGGATACACACTCTGGGGCT); The mRNA levels were analyzed by real-time quantitative RT-PCR using a Bio-Red iCycler system (Bio-Rad, Hercules, CA). MiRNA detection by real-time analysis involved reverse transcription of cDNA using a small RNA-specific stem-loop RT primer. Once specific cDNA was generated, individual miRNA was detected using Tagman small RNA assay Real-time PCR analysis. Results were normalized to small nuclear RNA U6 and the data were expressed as Log 2-fold-change in respective miRs/U6 mRNA levels.
Dual Luciferase Reporter Assay
Cells were seeded in 96-well plates and incubated for 24 h, then transfected with various constructs as indicated in the respective figures. 48 h after transfection, cells were lysed with passive lysis buffer for 20 min. Firefly and Renilla luciferase activities were determined from lysates using dual luciferase assay system (Promega, USA) in a Glomax luminometer.

Small Interference RNA and Transfection
Nrf2-siRNA or control-siRNA was purchased from Santa Cruz (Santa Cruz Biotechnology). Transfection was performed using Lipofectamine 2000 according to the manufacturer’s protocol. Briefly, PC12 cells were transfected with Nrf2-siRNA or control siRNA for 48 h, followed by treatment with Rg1 (1.00 μM) for the indicated times.

Animals

Male Sprague-Dawley (SD) rats (6 weeks; 160 to 180 g) were supplied by Experimental Animal Center of Chinese Academy of Medical Sciences (Beijing, China).  They are housed on a 12-hour light/dark schedule room that is temperature- (22 ± 2°C) and humidity- (<40%) controlled with free access to food and water. All experimental procedures were performed according to the standards established in the Guide for the Care and Use of Laboratory Animals published by the Institute of Laboratory Animal Resources of the National Research Council (United States). As well as approved by the Animal Care and Use Committee of the Peking Union Medical College and the Chinese Academy of Medical Sciences. A computer-generated randomization schedule was used to assign the animals into different groups. The assessment of measuring infarct volume and scoring neurobehavioral outcome is blinded.

Conditional Knockout of miR-144 in Predicted Penumbra Zone

According to the report by Ashwal et al.[31,32], the penumbra zone was mainly located in the middle cortex. In order to exclude the effects on miR-144 in non-ischemic regions, miR-144-shRNA expression was initiated by HIF1A promoter using pAAV- HIF1A promoter-EGFP-3FLAG-micro30(miR-144), which was provided by Obio Technology (Shanghai) Corp., Ltd. (China). For viral injections, Male SD rats were anesthetized with 10% chloral hydrate (4ml/kg of body weight, i.p. injection) and injected with 1μl of AAV virus into the cortex (AP, -3.84 mm from bregma; ML, +2.0 mm; DV, -1.2 mm from bregma). The injection sites were examined at the end of all the behavior tests and only data from animals with correct injections were included.

Focal Brain Ischemia

The transient middle cerebral artery occlusion (tMCAO) model was perform same as previously described with modifications. Briefly, the animal was anesthetized with 10% chloral hydrate (4 ml/kg, i.p. injection). Thetip of a 4-0 nylon thread was burned (diameter 0.36 mm) and was inserted into the right common carotid artery. It was advanced until the origin of the right middle cerebral artery was occluded. The thread was removed to allow reperfusion 90 minutes after and the animals were returned to their corresponding cages.

Neurological Function

Longa’s five-point scale was used to evaluate the neurological score. The animals without symptoms of neural damage or death after the surgery were excluded and other rats were recruited.

Infarct Analysis

At indicated time points, brains were removed under anesthesia and cut into 2-mm slices,  six slices per rat. The slices were stained with 1% solution of 2, 3, 5-triphenyltetrazolium chloride (TTC) in PBS at 37°C for 30 min and fixed in 4% phosphate-buffered formalin. Images of the samples were captured with an electronic scanner and computer. The areas of ipsilateral and contralateral corresponding structures and infarct area were calculated with Image J (location). The infarct area was corrected to account for edema post-ischemia and shrinkage due to cutting and staining. The infarction area was corrected using following formula for both edema and shrinkage: Corrected infarct area = measured infarct area + area of contralateral corresponding structure − area of ipsilateral corresponding structure [16].The infarct area in each animal was evaluated using the product of average slice thickness (2 mm) and total of infarct area in all sample slices.

ROS Production Measurement by Dihydroethidium (DHE) Staining

ROS in brain slices were revealed by DHE. Briefly,  a dewaxed and dehydrated paraffin section (4μm) was exposed to DHE (100 μmol/L, Molecular Probes, Sigma, D7008) on the sample surface away from light for 3 h at 37 °C, followed by three washes with PBST. Next, the sections were incubated with Hoechst (1:1000) for 15min at 37℃. The slides were rinsed 3 times with PBST, mounted with 90% glycerin and cover slipped. The tissue sections were then visualized and fluorescence was detected by a laser scanning confocal microscope (Leica TCSSP2, Solms, Germany). The entire quantifying procedure was performed in a blinded manner.

Immunofluorescence Staining for DNA Oxidative Products

DNA oxidative products in brain slices were revealed using immunofluorescence with anti-8-hydroxy-2’deoxyguanosine (8-OHDG).Briefly, dewaxed and dehydrated paraffin sections were incubated for antigen retrieval for 10 min. The slices were then washed three times with PBS at room temperature, 5 min each time. Next they were soaked in 0.5% Triton X-100 for 10 min, then washed three times with PBS, 5 min each time. The sections were blocked with goat serum and incubated with primary antibodies of 8-OHDG at 4 °C overnight. Subsequently, sections were washed with PBST three times and incubated for 2 h at room temperature away from light with secondary fluorescence antibodies. All images were captured under a laser scanning confocal microscope (Leica TCSSP2, Solms, Germany). The entire quantifying procedure was performed in a blinded manner.

Statistical Analysis
All values are presented as mean ± SD. Statistical significance between multiple groups was calculated with one-way ANOVA analysis followed by Newman–Keuls post hoc test. T-test was used for testing between two groups. p<0.05 was considered statistically significant. All statistical analyses were performed using SPSS13.0 software (Chicago, IL)

Results

1. Rg1 protects against OGD/R induced neurotoxicity in PC12 cells

To investigate the effect of Rg1 on OGD/R induced neurotoxicity, PC12 cells were treated with or without Rg1 (structure shown in Fig1a) at indicated concentrations. Neural viability and LDH release were tested by MTT and colorimetry respectively. The results showed that Rg1 improved PC12 cells viability (Fig1b: *p<0.05 vs. model group at 0.01, 0.10 and 1.00 μM) and decreased LDH release (Fig1c: **p<0.01 vs. model group at 0.10 and 1.00 μM) significantly. However, when the concentration reached to 10.00μM, the protective effect showed remarkable decreasing tendency. As shown in Fig 1d, the level of intracellular ROS, determined by DCFH-DA, was significantly increased by OGD/R , and was attenuated by Rg1 treatmentat the concentration of 0.10 and 1.00 μM (*p<0.05, **p<0.01 vs. model group), which is consistent with LDH release. These results suggest that the attenuation of neural viability and LDH release may be related to the changes of ROS level. Moreover, we found that 1.00 μM of Rg1 produced the most favorable effect than any other concentrations in this study. This finding corresponds to the effective concentration in the mechanism studies of Rg1 (citation needed).

2. Rg1 promotes Nrf2 nuclear accumulation without change in Keap1

Next, we investigated whether Nrf2 is involved in the anti-OGD/R effect of Rg1 in neurons. PC12 cells were treated with Rg1 or vehicle for different time length under OGD/R andthe dynamic of Nrf2 is evaluated in the nucleus and cytoplasm respectively. As shown in Fig 2a and b, the nuclear accumulation of Nrf2 is significantly increased in the vehicle group from 2 hours of OGD, and reached its peak at 4 hours, then decreased below baseline from 12 hours. Rg1 treatment enhanced the nuclear accumulation of Nrf2, as manifested by prolonging peaking time to 8 hours (*p<0.05 vs OGD/R + vehicle group at 8 hours) and slowing down the decreasing tendency significantly (**p<0.01 vs OGD/R + vehicle group at 12 hours; ***p<0.001 vs OGD/R + vehicle group at 24 hour).  Keap1 is known to be the major endogenous negative regulator for Nrf2. It is responsible for keeping Nrf2 in the cytoplasm and degradation of Nrf2 by binding to the ubiquitin proteasome system[14].

To find out whether the increased nuclear accumulation caused by Rg1 is regulated by Keap1, its expression was determined at 24 hours of OGD/R. As shown in Fig 2c and d, theexpression of keap1 in both the nucleus and the cytoplasm was significantly decreased after OGD/R, but Rg1 treatment showed no effect on Keap-1 expression compared to the vehicle group. This suggests that Keap1 does not play a role in the enhancement of Nrf2 nuclear accumulation by Rg1.

3. Rg1 increased Nrf2 targeted gene expression

To determine whether the ARE activity is increased by Rg1, ARE-dependent luciferase activity was evaluated. As shown in Fig3a, Rg1 treatment increased ARE luciferase activity in a concentration-dependent manner (**p<0.01, ***p<0.001 vs. vehicle group). We next tested the ability of Rg1 to promote the Nrf2 pathway under OGD/R. As shown in Fig3b, quantitative PCR analysis showed that mRNA levels of NQO-1(**p<0.01, ***p<0.001 vs. model group), HO-1 (*p<0.05, **p<0.01 vs. model group), GCLC (**p<0.01, ***p<0.001 vs. model group), GCLM (*p<0.05, **p<0.01 vs. model group) are significantly increased after Rg1 treatment, which alleviated the reduction in transcription caused by OGD/R. Furthermore, the protein expression of the Nrf2 pathway was also investigated. Fig3c and d show that Rg1 increased the levels of expression of NQO-1, HO-1, GCLC, GCLM significantly at the concentrations of 0.10 μM and 1.00 μM. Collectively, the results suggest that Rg1 stimulates ARE activity in OGD/Rand subsequently up-regulates the expressions of the target mRNAs and proteins.

4. Rg1 improves OGD/R-induced neurotoxicity in a Nrf2-dependent manner in vitro

To investigate whether the neuroprotective effect of Rg1 acts via Nrf2 activation, Nrf2-siRNA was employed to block the expression of Nrf2.  As shown in Fig 4a and b, Nrf2-siRNA abolished the increase in nuclear accumulation of Nrf2 induced by Rg1 after OGD. Correspondingly, the ARE-luciferase activity and Nrf2 pathway were no longer enhanced by Rg1 as shown in Fig4c-e. We also found that Nrf2-siRNA abolished the neuroprotective effect of Rg1 after OGD/R, as manifested by decreased cell viability and increased ROS levels (Fig4f and h). The above results demonstrated that the anti-OGD/R effect of Rg1 is dependent on the activation of Nrf2.

5. MiR-144 mediated the increased Nrf2 transcriptional activity by Rg1 under OGD.

As previously mentioned, Rg1 had no effect on the expression of Keap1, though Nrf2 accumulation was enhanced significantly. It has been well documented that post-transcriptional regulation via microRNAs (miRNAs), such as miR-153, miR-144, and so on, plays an important role in Nrf2/ARE activity in the central nervous system [36,37]. In this study, qPCR evaluation showed that OGD/R raised miR-144 expression, while Rg1 treatment reversed it significantly (**p<0.01 vs model group, Fig5a). However, the expression of miR-144 is unaffected by the inhibition of  Nrf2 expression  by its siRNA, indicating that miR-144 is not regulated by Nrf2. It has been documented that miR-144 directly regulates Nrf2 and modulated oxidative stress response in K562 cell[38]. To find out whether miR-144 could regulate Nrf2 activation directly in neurons, scramble control or miR-144 were transfected into PC12 cellsand the nuclear and cytoplasmic accumulation of Nrf2and ARE-luciferase activity were tested. As shown in Fig5b-e, miR-144 reduced both nuclear and cytoplasmicaccumulation of Nrf2 compared to scramble control transfection (***p<0.001 vs. scramble control-treated cells). Fig5f showed that miR-144 transfection inhibited ARE-luciferase significantly (***p<0.001 vs. scramble control treated cells), which is consistent with findings in K562 cells [38].

According to TargetScan (www.targetscan.org), 3’-untranlated region (3’-UTR) of Nrf2 is targeted by miR-144. To test whether OGD/R-induced increased miR-144 directly targeted Nrf2 by binding to 3’-UTR sequence, two reporters (Nrf2-3’UTR and miR-144 binding site mutated, Nrf2-3’-UTR-MUT) were constructed as shown in Fig5g. Both constructs were transfected into PC12 cells followed by exposure to OGD/R with or without Rg1. MiR-144 was introduced as a positive control. As shown in Fig5h, OGD/R decreased luciferase activity in the Nrf2-3’-UTR group, (**p<0.01 vs. control group), which is similar to that of miR-144 exposure. Rg1 treatment was able to reverse it significantly (**p<0.01 vs OGD/R group). However, OGD/R showed no effect on the luciferase in Nrf2-3’-UTR-MUT group. As a positive control, miR-144 loses its inhibition on luciferase activity. These findings indicate that miR-144 blocks Nrf2 activity by targeting at  the 3’-UTR of Nrf2.  To find out whether OGD/R-induced decreased Nrf2 activity is due to miR-144, anti-miR-144 or control-anti-miR were transfected into PC12 cells along with Nrf2-3’-UTR respectively. The results in Fig5i showed that there is a significantly decreased luciferase activity in control-anti-miR-treated cells after OGD/R. When miR-144 was absent due to the anti-miR-144 transfection, OGD/R was unable to affect the luciferase activity.  These results indicate that the repression of Nrf2 activity by OGD/R was due to the increase of miR-144 specifically. To further investigate the role of miR-144 on ARE luciferase, control-anti-miR or anti-miR-144 were also introduced into PC12 cells respectively. We found that anti-miR-144 abolished the decrease of ARE-luciferase induced by OGD/R (Fig5j). These findings further indicate that miR-144 suppressed Nrf2 transcriptional activity in PC12 cells after OGD/R.

6. Blockage of miR-144 abolished the anti-I/R effect of Rg1 in vivo.

To further investigate a potential mechanistic role for miR-144 in the anti-I/R effect of Rg1 in vivo, we injected miR-144-shRNA in the predicted ischemic penumbra zone [31,32] followed by tMCAO in SD rats to block the increased expression of miR-144 post-I/R. The result showed that Rg1 treatment at the dose of 2-mg/kg improved I/R outcomes in NC animals significantly (Fig6a-d), as manifested by decreased infarct volume (*p<0.05, Fig6a and b) and improved Zea longa score (*p<0.05, Fig6c) compared to the vehicle group. However, both of these improvements were abolished by miR-144-shRNA administration, suggesting that miR-144 plays pivotal role in the anti-I/R effect of Rg1 in vivo.

We then investigated whether the improved outcomes are associated with the changes of oxidative stress status. Based on the ROS evaluation using DCFH-DA, we found that Rg1 administration decreased intracellular levels of ROS significantly (*p<0.05 vs. vehicle-treated animals in NC group), and was abolished by miR-144-shRNA administration (Fig6d and e). Furthermore, the fluorescence intensity of 8-OHDG staining was used to measure the amount of DNA oxidative products. As shown in Fig6f and g, the fluorescence intensity of 8-OHDG was significantly decreased by Rg1 treatment in NC group (**p<0.01 vs. vehicle treated animals), However, application of miR144-shRNA reduced the level of 8-OHDG remarkably when compared to that in NC-treated animals. All of these results suggested that inhibition of miR-144 blocke anti-I/R effect of Rg1 in vivo.

7. Inhibition of miR-144 blocked the enhancement of Nrf2/ARE signaling pathway induced by Rg1 after I/R.

The anti-I/R effect of Rg1 was significantly abolished by the absence of MiR-144, which could interact with the 3’-UTR of Nrf2 directly in vitro. To find out whether miR-144 produces similar biological activities in vivo, the level of Nrf2 and its target genes were determined by qPCR and Western blot 24 hours after tMCAO. As shown in Fig7a and b, there is a significant increase of nuclear accumulation of Nrf2 after Rg1 treatment compared to the vehicle in NC animals. However, the increase induced by Rg1 is abolished by miR-144-shRNA application, suggesting that the inhibition of miR-144 in vivo has similar effect on the nuclear accumulation of Nrf2 in vitro. With the increase of nuclear Nrf2, its target genes were also up-regulated. As shown in Fig7c-g, Western blot result showed that Rg1 treatment increased the expression of NQO-1 (*p<0.05), HO-1 (*p<0.05), GCLC (**p<0.01), and GCLM (**p<0.01) significantly in NC animals. Application of miR-144-shRNA promotes the expression of these proteins  compared to that in NC animals, but abolish the improvement induced by Rg1 treatment. Quantitive PCR determination showed similar changes with proteins correspondingly in Fig7h-k, showing no significant difference in expression after shRNA transfection with or without Rg1 treatment. All of these results suggested that miR-144 is required in the enhancement of the Nrf2/ARE pathway induced by Rg1 in I/R injury.

Discussion

Stroke is one of the leading causes for human death and disability[14] and lacks effective treatment. Rg1 is well documented to have anti-stroke effect [citation]. In this study, we found that Rg1 protects against I/R-induced neurotoxicity by improving nrf2 activity both in vivo and in vitro. We also found that the neuroprotective effect of Rg1 is dependent on the miR-144-Nrf2/ARE pathway, which provides a new clue for the treatment of ischemic stroke.

The neuroprotective actions of Rg1 has been documented to have positive effects in experimental stroke therapy[6], including improved neurological function, reduced infarct volume, attenuated BBB integrity, and so on[citation?]. It has been proven that Rg1 protects the brain against oxidative stress in many pathological conditions, such as Parkinson’s disease[39,40], Alzheimer disease[41], D-galactose-induced brain aging[42], and so on. Our previous study showed that Rg1’s hepatic protective effect was mainly due to its anti-oxidant activity[35,34]. However, whether anti-oxidative stress efficacy of Rg1 is involved in its anti-stroke ability has not been reported until now. To test this hypothesis, we treated PC12 cells with Rg1 or vehicle followed by OGD. We found that Rg1 protects neurons against OGD/R significantly from 0.01 μM to 1.00 μM in a concentration-dependent manner, as manifested by improved neural viability and reduced LDH release shown in Fig1b and c.

It has been widely confirmed that I/R triggers a rapid and excessive ROS production, which promotes oxidative injury by lipid peroxidation, DNA peroxidation, etc [11]. To test whether the neuroprotective effect of Rg1 is related to ROS, DCFH-DA staining was used to determine the level of intracellular ROS. We found that Rg1 treatment reversed the increase of ROS post-I/R significantly at the concentration of 0.10 μM to 1.00 μM in a concentration-dependent manner (Fig1d), which is consistent with its efficacy concentration of LDH release. These results suggest that the neuroprotective effect of Rg1 may relate to its ability to suppress excessive ROS release triggered by oxidative stress.

Oxidative stress occurs when intracellular ROS production overcomes intracellular anti-oxidative defenses [37], including enzymatic and non-enzymatic. Nrf2 is one of the master regulators of the expression of proteins involved in endogenous anti-oxidantive defense[14], which is activated by excessive ROS release following ischemic/reperfusion injury. A number of studies have indicated that the expression of Nrf2 is largely increased in the acute phase of stroke [citation?]. Moreover, the changes of Nrf2 level is mainly in neurons [21]. In our study, we found that Nrf2 nuclear accumulation induced by OGD/R in PC12 cells, begins at 2h of OGD,  peaks at 4h, and decreases from 8h (Fig2a and b). Rg1 treatment enhanced Nrf2 nuclear accumulation significantly, as evidenced by lower decreasing tendency from 8h and lasts for 24h, which provides a solid basis for the promotion of anti-oxidative gene expressions.

It is well known that the distribution and degradation of Nrf2 is regulated by its endogenous inhibitor keap1 [citation needed]. Keap1 is a homodimer with two canonical domains: one binds to Nrf2 and the other binds to E3 ubiqutin ligase complex, which keeps Nrf2 in the cytoplasm and promotes its degradation [citation needed]. To find out the reason of increased nuclear accumulation of Nrf2 after I/R, the changes of keap1 was first investigated. As shown in Fig2c and d, we found that OGD/R decreased  keap1 content in both the cytoplasm and the nucleus. Rg1 application showed no effect on keap1 expression, suggesting that the enhanced Nrf2 nuclear accumulation induced by Rg1 may be resulted from increased Nrf2 expression, rather than keap1 regulation.

Nuclear Nrf2 formsheterodimers with bZIP proteins and recognize the appropriate ARE sequences, then initiate the transcription of anti-oxidative genes harboring ARE in their promoter region, such as NQO-1, HO-1, and so on [citation needed]. As the result of enhanced nuclear accumulation, we found that ARE luciferase activity is enhanced by Rg1 directly (Fig3a), then the activity of its downstream anti-oxidative genes and proteins are increased accordingly (Fig3b-d). All of these results suggest that Nrf2/ARE pathway is activated by Rg1 after OGD/R. Nrf2 siRNA was used to investigate the role of Nrf2 in the anti-OGD/R effect of Rg1. Our results showed that Nrf2 siRNA abolished Rg1’s neuroprotective effect significantly, in terms of ARE luciferase activity, expression of downstream proteins, cell viability, LDH release and ROS levels (Fig4c-h).  These results indicate that the anti-OGD/R effect of Rg1 is in an Nrf2-dependent manner.

As previously mentioned, there is a significantly increased Nrf2 protein level compared to its mRNA contents, indicating that the increased nuclear portion of Nrf2 is due to post-transcriptional regulations, such as miRNAs. A number of miRNAs have been reported for the regulation of Nrf2, such as miR-153 in the PD model [43,37], miR-144 in the AD model [36], and so on. Among them, only miR-144 showed related pathological changes with ischemic impairment [citation needed]. Loss of miR-144 has been shown to impair ischemic precondition-induced cardioprotection, therefore miR-144 is considered as an circulating effector of remote ischemic preconditioning[44]. In this study, we found that the level of miR-144 increased significantly after OGD/R, which exacerbated oxidative stress by inhibiting Nrf2 expression by targeting its 3’-UTR (Fig5a). MiR-144 was able to inhibit both the nuclear and cytoplasmic Nrf2 expression (Fig5b-e) and ARE luciferase activity (Fig5f). Rg1 treatment reversed it significantly by releasing Nrf2 mRNA from miR-144 and promoting Nrf2 expression, shown in Fig5a. Moreover, we found the expression of miR-144 is independent from Nrf2 activation in PC12 cells induced by OGD/R, as Nrf2-siRNA had no effect on the level of miR-144, shown in Fig5a. However, application of anti-miR-144 increased ARE-luciferase activity significantly (Fig5j), suggesting that miR-144 is an upstream regulator of Nrf2 expression. Furthermore, we found that miR-144 regulates Nrf2  by interacting with its 3’-UTR directly in PC12 cells, as shown in Fig5h.MiR-144 inhibited luciferase  activity in the PC12 cells transfected with Nrf2-3’-UTR after OGD/R, but failed to do so in Nrf2-3’-UTR-MUT transfected cells. Rg1 treatment decreased the level of miR-144 significantly after OGD/R, which is also independent of Nrf2. , Rg1’s ability to improve Nrf2 transcriptional activity is lost when Nrf2 3’-UTR was mutated in miR-144 seed sites. This suggests that miR-144 is the mediator of Rg1-induced activation of Nrf2/ARE pathway in vitro because Rg1 was no longer capable of promoting Nrf2 activity when the interaction between miR-144 and Nrf2 was inhibited.

Subsequently, we tested the role of miR-144 in the anti-I/R effect of Rg1 in vivo. Rg1 treatment improved the neural outcomes of tMCAO significantly at the dose of 20mg/kg in NC treated animals, including decreased infarct volume (Fig6a) and improved neurological scores (Fig6b), which are consistent with the previous study[7]. However, the application of miR-144-shRNA abolished these improvement significantly. The improved oxidative stress status by Rg1, including lower intracellular ROS content (Fig6c) and DNA oxidative products (Fig6d-g), was also counteracted in miR-144-shRNA treated animals. Furthermore, we found that Rg1’s positive effect on the Nrf2/ARE pathway in I/R, including higher nuclear accumulation and target gene transcriptions and expressions (Fig7), was also diminished by miR-144-shRNA administration. All of these results suggest that miR-144 plays a crucial role in mediating the Nrf2/ARE pathway in the anti-I/R effect of Rg1 in vivo.

In summary, we identify that the anti-I/R effect of Rg1 involves the Nrf2/ARE pathway and the underlying mechanism was clarified with post-transcriptional regulations of Nrf2 by miR-144. Furthermore, we provide a solid evidence for the direct regulatory effect of miR-144 on the Nrf2/ARE pathway in neurons, and confirmed that the enhanced Nrf2/ARE pathway  by Rg1 in I/R is dependent on miR-144. All of these results provide strong evidence in support of Rg1 as a promising drug for the treatment of ischemic stroke.

Figure Legends

Fig1. Rg1 protects against OGD/R-induced neurotoxicity in PC12 cells. (a) The chemical structure of Rg1. MTT assay (b) and LDH release determination (c) were employed to investigate the role of Rg1 in OGD. The level of intracellular ROS was also revealed by DCFH-DA staining (d). All data were presented as mean ± SD of triplicate independent experiments. One-way ANOVA analysis followed by Newman-Keuls test (*p < 0.05, **p<0.01 vs. model group; n=? for each group)

Fig2. Rg1 promotes Nrf2 nuclear accumulation without change in Keap1. (a) Representative Western blot of nuclear Nrf2 accumulation with different time length of OGD/R.  Lamin A was used as an internal control. (b) Quantitative analysis of Nrf2 levels in Fig2a (Normalized to 0 h for each group). *p<0.05, **p<0.01, ***p<0.001 vs. OGD/R + vehicle group at corresponding time point. (c) Representative Western blot of Keap1 expression in the nucleus and cytoplasm after 8 hours of OGD. Lamin A or β-actin was used as an internal control in the nucleus or cytoplasm, respectively. (d) Quantitative analysis of Nrf2 levels in Fig2c (Normalized to Control for each part). *p<0.05 vs. control group. All data were presented as mean ± SD of triplicate independent experiments. One-way ANOVA analysis followed by Newman-Keuls test. (plz add n = ? / groups for panels “b” and “d”)

Fig3. Rg1 contributes ARE targeted gene expressions. (a) Rg1 induced ARE-dependent luciferase activity. ARE-luciferase reporter was transfected into PC12 cells by Lipofectamine 2000 for 48 hour, then, they were treated with vehicle or Rg1 for 8 hours at indicated concentration. Luciferase activity was determined by an analyzer fluorescent assay (*p<0.05, **p<0.01, ***p<0.001 vs. vehicle group). (b) Rg1 promoted ARE-regulated gene transcription after OGD/R. . The mRNA levels were tested by quantitative RT-PCR, including NQO-1, HO-1, GCLC and GCLM (*p<0.05, **p<0.01, ***p<0.001 vs. model group). (c) Rg1 promoted ARE-regulated protein expression after OGD/R. The levels of NQO-1, HO-1, GCLC and GCLM were tested by western blot. (d) Quantitative analysis of HO-1, GCLC, GCLM and NQO-1 protein levels (*p<0.05, **p<0.01, ***p<0.001 vs. model group). All data were presented as mean ± SD of triplicate independent experiments. One-way ANOVA analysis followed by Newman-Keuls test. (plz add n = ? / groups for panels “b” and “d”)

Fig4. Rg1 improves OGD/R-induced neurotoxicity in an Nrf2-dependent manner in vitro. The neuroprotective effect of Rg1 with Nrf2 inhibition is tested in PC12 cells transfected with control-siRNA or Nrd2-siRNA after OGD/R. (a) Representative Western blot for PC12 cells transfected with control-siRNA or Nrd2-siRNA, with Rg1 or without (vehicle). Nrf2-siRNA abolished the increased nuclear accumulation of Nrf2 induced by Rg1. Lamin A was used as an internal control. (b) Quantitative analysis of Nrf2 level in Fig4a. (***p<0.001 vs. vehicle; ns: no significant difference). (c) Nrf2-siRNA blocked the increased ARE-luciferase activity induced by Rg1 in OGD/R PC12 cells. Luciferase activity was determined by an analyzer fluorescent assay. (***p<0.001 vs. vehicle; ns: no significant difference). (d) Representative Western blot images of ARE-targeted proteins expression, including NQO-1, HO-1, GCLC and GCLM, induced by Rg1 when Nrf2 expression was normal or blocked.  (e) Quantitative analysis of the levels of expression of NQO-1, HO-1, GCLC and GCLM in Fig4d. (**p<0.01 vs. vehicle; ns: no significant difference) (f) Cell viability was determined by MTT assay. (**p<0.01 vs. model; ns: no significant difference). (g) The released LDH from PC12 cells was determined by colorimetry. (*p<0.05 vs. model; ns:  no significant difference). (h) The level of intracellular ROS was also revealed by DCFH-DA staining. (*p<0.05 vs. model; ns: no significant difference). All data were expressed as mean ± SD from three independent experiments. One-way ANOVA analysis followed by Newman-Keuls test. (plz add n = ? / groups for panels “b” to “c”, and “e” to “h”)

Fig5. MiR-144 mediated the increased Nrf2 transcriptional activity by Rg1 under OGD. (a) Real time PCR results of miRNA-144 levels in cells transfected with control- or Nrf2-siRNA, treated with vehicle or Rg1 and after OGD. U6 was used as an internal control. Data were expressed as relative intensity of miR-144 normalized to U6 over control group. (**p<0.01 vs. model). (b andd) Representative Western blot of nuclear and cytoplasmic Nrf2 expression, respectively, after transfected with miR-144 or scramble control.  B-actin or Lamin A was used as an internal control in the cytoplasm or nucleus, respectively. (c) and (e) are the quantitative analyses of (b) and (d) respectivly. (***p<0.001 vs. scramble control). (f) ARE-luciferase activity after transfection with MiR-144 or scramble control. ARE-luciferase was normalized to renilla activity, and the result was normalized to scramble control (***p<0.001 vs. scramble control). (g) The binding site between miR-144 and Nrf2 as predicted by TargetScan at 349-356 nt of 3’-UTR of Nrf2. Nrf2-3’-UTR-MUT was constructed with three mutations in the miR-144 binding site at 352, 353 and 354 as shown in the bottom line of Fig5g.  (h) Luciferase activity results in PC12 cells co-transfected with reporter gene constructs and pRL-renilla constructs with scramble control or miR-144, with vehicle or Rg1 (1.00μM ) treatment, and after OGD. Luciferase activity was normalized to renilla activity. Results were represented as normalized to control for Nrf2-3’-UTR or Nrf2-3’-UTR-MUT respectively. (**p<0.01). (i) Nrf2 3’-UTR construct was co-transfected with control anti-miR or anti-miR-144 into PC12 cells, followed by OGD/R, and then the luciferase activity was measured. (***p<0.001 vs. control, ns: no significant difference). (j) ARE-luciferase activity results after co-transfection with control-anti-miR or anti-miR-144 followed by OGD/R. (**p<0.01 vs. control, ns: significant difference). All data were expressed as mean ± SD from three independent experiments. One-way ANOVA analysis followed by Newman–Keuls test was used for (a), (h), (i), (j). Student’s T test was used for testing between two groups in (c), (e), and (f).  (plz add n = ? / groups for related bargraph panels)

Fig6. Blockage of miR-144 abolished the anti-I/R effect of Rg1 in vivo. (a) Representative TTC staining results of tMCAO rats treated with vehicle or Rg1 (20mg/kg) infected by miR-144 or control virus (NC) in ischemic penumbra. (b) Quantitative analysis of cerebral infarct volume in Fig6a. (*p<0.05 vs. vehicle, ns: no significant difference, n=8-10).  (c) Quantitative analysis of neurological score by Zea longa test. (*p<0.05 vs. vehicle, ns: no significant difference, n=8-10). (d,e) The level of intracellular ROS by DCFH-DA staining in cerebral cortex. (e) Representative images of ROS contents labelled in red,  (d) Quantitative analysis of (e) (*p<0.05 vs. model; ns: no significant difference). (f,g) The level of DNA oxidative product (8-OHDG) by immunofluorescence staining in cerebral cortex. (f) Representative images of 8-OHDG staining shown in green(g) Quantitative analysis of (f) (*p<0.05 vs. model; ns: no significant difference). All data were expressed as mean ± SD from three independent experiments (d-g). One-way ANOVA analysis followed by Newman-Keuls test.  (plz add n = ? / groups for related bargraph panels)

Fig7. Inhibition of miR-144 blocked the enhancement of Nrf2/ARE pathway induced by Rg1 after I/R. (a) Representative Western blot of nuclear Nrf2 isolated from rat predicted penumbra cortex treated with vehicle or Rg1 (20 mg/kg) and transfected with control virus (NC) or miR-144-shRNA. Lamin A was used as an internal control. (b) Quantitative analysis of Nrf2 levels in Fig7a. (*p<0.05 vs. vehicle; ns: no significant difference). (c) Representative Western blot of Nrf2-targeted gene expressions isolated from predicted penumbra cortex treated with vehicle or Rg1 (20 mg/kg) and transfected with control virus (NC) or miR-144-shRNA.. Β-actin was used as an internal control. (d-g) Quantitative analysis of Nrf2-targeted gene expressions in Fig7c, including NQO-1 (d), HO-1 (e), GCLC (f) and GCLM (g), normalized to vehicle treated animals in each viral infection. (*p<0.05, **p<0.01 vs. vehicle; ns: no significant difference). (h-k) Quantitative analysis of Nrf2-targeted mRNA levels determined by qPCR in predicted penumbra cortex, including NQO-1 (h), HO-1 (i), GCLC (j) and GCLM (k),  normalized to vehicle-treated animals in each viral infection. (*p<0.05, **p<0.01 vs. vehicle; ns: no significant difference). All data were expressed as mean ± SD from three independent experiments (d-g). One-way ANOVA analysis followed by Newman-Keuls test. (plz add n = ? / groups for related bargraph panels)

Fig8. Schematic diagram of the anti-I/R effect of Rg1 through miR-144/Nrf2/ARE pathway. I/R induces overproduction of ROS, which triggers oxidative stress and exhaust intracellular antioxidants. Meanwhile, it promotes antioxidants production by promoteing the Nrf2/ARE pathway. I/R also promotes the activity of miR-144, which directly binds to Nrf2-3’UTR in neurons and inhibits the expression of Nrf2. Rg1 inhibits the activity of miR-144 after I/R promotes the Nrf2 / ARE pathway, which enhances the expression of ARE-regulated anti-oxidative genes, reduces oxidative damage and exerts anti-I/R effect.