We purchased PNS from Chengdu Manster Biotechnology Co., Ltd. (DST200706-054; Chengdu, China). The major effective constituents of PNS compounds included 6.96% (v/v) Notoginsenoside R1, 34.29% (v/v) Ginsenoside Rg1, 2.65% (v/v) Ginsenoside Re, 41.34% (v/v) Ginsenoside Rb1, and 10.19% (v/v) Ginsenoside Rd [21]. We also purchased an autophagy inhibitor (3-methyladenine [3-MA]; M9281; Sigma-Aldrich, St. Louis, MO, USA), Caspase-1 inhibitor (AC-TYR-VAL-ALA-ASP-C MK [Ac-YVAD-CMK]; SML0429; Sigma-Aldrich), NLRP3 inhibitor (MCC950; S7809; Selleck Chemicals, Houston, TX, USA), mitophagy inhibitor (mitochondrial-division inhibitor 1 [Mdivi-1]; S7162; Selleck), goat anti-rabbit secondary antibody (AP132P; Merck Millipore, Darmstadt, Germany), and goat anti-mouse secondary antibody (AP124P; Merck Millipore, Darmstadt, Germany).

We purchased healthy adult male Sprague Dawley rats weighing 200–220 g (specific-pathogen-free [SPF] grade) from the Experimental Animal Center of HUCM (Animal Certificate No. SYXK, Xiang, 2018-0002). The rats were fed under SPF conditions for 1 week before experiments, with tap water and normal chow supplied ad libitum. The breeding room was provided with natural light and maintained at a temperature of 20–25 °C. All protocols followed the ARRIVE guidelines in terms of study design, sample size, randomization, outcome measures, data analysis, experimental procedures, and reporting of results. This study was approved by the Animal Ethics Committee of Hunan University of Chinese Medicine (HUCM; Changsha, China; Approval No. LLBH-202103290002), in compliance with the Guide for the Care and Use of Laboratory Animals published by the US National Institutes of Health (NIH publication no. 85-23, revised 1996). We declare that all methods were performed in accordance with the relevant guidelines and regulations.

We generated an MCAO model as described previously [21]. Rats were anesthetized with isoflurane (1.5%) in 30%/70% oxygen/nitrous oxide. Their body temperature was maintained at 37 °C $\pm$ 0.5 °C using a heating pad throughout the surgical procedure and afterward, until the rats recovered from the anesthesia. The right common carotid artery (CCA), external carotid artery (ECA), and internal carotid artery (ICA) were exposed via a neck incision and then isolated. We subsequently ligated the distal portion of the ECA using surgical thread and then disconnected it and its branches proximal to the ligation site using electrocoagulation. We then made a slipknot to ligate the distal part of the ICA and clamped the distal ICA and CCA with arterial clamps. A small orifice was carefully made in the ECA stump. Then, occlusion wire (2636A2; Beijing Cinontech Co., Ltd., Beijing, China) was introduced into the ECA and directed into the ICA via the bifurcation of the CCA. Next, we opened the artery clip that clamped the distal ICA and inserted the occlusion wire into the intracranial portion of the ICA.

The length of the line embolism inserted into the artery was 18 $\pm$ 2 mm distal from the CCA bifurcation. When the filament was inserted into the anterior cerebral artery (ACA), its tip reached the origin of the middle cerebral artery; the insertion was stopped when resistance increased. We then fastened the ICA to prevent bleeding and movement and sutured the neck skin. After 2 h of blood flow blockage and followed by 24 h of reperfusion. The sham group underwent operations involving isolation of the CCA, ECA, and ICA without occlusion wire insertion. We then carried out subsequent experiments.

We randomly divided the rats into the following treatment subgroups: sham, model, low-dose PNS (PNS-L), medium-dose PNS (PNS-M), high-dose PNS (PNS-H), PNS + 3-MA, PNS + Mdivi-1, Ac-YVAD-CMK, and MCC950 group. The rats were given microinjections of 3-MA (40 nmol) at a concentration of 5 $\mu$L per lateral ventricle [22, 23]. Mdivi-1 (0.15 mg/mL) was administered by i.p. injection at 10 mL/kg [24], and Ac-YVAD-CMK (0.5 mg/mL) and MCC950 (2.5 mg/mL) were administered at doses of 10 mL/kg [25]. The PNS dose was calculated based on a previous study [26]: low dose, 40 mg/kg; medium dose, 80 mg/kg; and high dose, 160 mg/kg. PNS was administered three times via gavage at the abovementioned concentrations for the respective PNS groups. Optimal administration times were 50, 26, and 2 h before the model establishment [27]. The sham and model groups were given equal volumes of physiological saline, whereas the remaining groups were injected using a microliter syringe (Hamilton Co., Reno, NV, USA) at the onset of reperfusion.

Following reperfusion for 24 h, we evaluated rats in each group to determine the NDS via the Longa method [21]: 0 = no neurological deficit; 1 = the left forelimb could not be extended when the rat lifted its tail; 2 = circling to the contralateral side when walking; 3 = tilting to the contralateral side when walking; 4 = no spontaneous motor activity, or loss of consciousness.

After reperfusion for 24 h, rats were euthanized by intracardial injection of 500 mg/kg 10% potassium chloride solution; their brains were removed, placed on ice, and frozen for 15 min at –20 °C. We evenly cut each brain into five 2-mm serial coronal sections. Then, the slices were incubated in 2% 2,3,5-triphenyltetrazolium chloride (TTC) phosphate buffer (BCBP3272V; Sigma-Aldrich, Saint Louis, MO, USA), immersed in a dark, constant-temperature bath of 37 °C for 15 min, and stirred gently every 5 min. After staining, we fixed the brain sections in 4% paraformaldehyde for a minimum of 24 h. The infarct regions eventually turned white, and the unaffected regions turned red. The brain slices were then photographed and analyzed using Image-Pro Plus software (Media Cybernetics, Silver Spring, MD, USA). Additionally, brains were divided into the ipsilateral hemisphere (ischemic side) and contralateral hemisphere. The following formula was used to calculate cerebral infarction volume, as described previously [28]: percentage of infarct volume = (total contralateral hemisphere volume-non-infarcted ipsilateral hemisphere volume)/(total contralateral hemisphere volume) $\times$ 100%.

Following reperfusion for 24 h, rats (n = 8) were anesthetized and euthanized via intracardial injection of 500 mg/kg 10% potassium chloride solution and then killed using rapid decapitation. We then fixed the brains in 4% paraformaldehyde and sliced them into sections with 3-mm thickness, cutting coronally through the optic cross-plane. The slices were dehydrated using gradient alcohol, made transparent with xylene, embedded in paraffin, and finally cut into 3-$\mu$m-thin sections, all of which were fixed on glass slides for hematoxylin and eosin (H&E) staining. Then, we observed the H&E-stained sections under a Zeiss Axioscope 5 microscope (Zeiss, Oberkochen, Germany) and calculated the rate of cell damage. Additionally, five non-overlapping visual fields in the ischemic cortex were randomly selected for observation of damaged cells. Such cells were primarily characterized by vacuolar degeneration, eosinophilic degeneration, nuclear contraction, and nuclear dissolution. Finally, we counted the total number of cells and number of damaged cells in each high-magnification field. The rate of cell damage was calculated as described previously: [28] (number of damaged cells/total number of cells) $\times$ 100%.

We used WB to detect NLRP3 inflammasome levels and mitophagy-related proteins. After reperfusion at 24 h, in accordance with the literature [29], we weighed and homogenized the cerebral cortex of the ischemic penumbra and then added it to 10 volumes of pyrolysis liquid (P0013; Shanghai Biyuntian Biotechnology Co., Ltd., Shanghai, China), a protease inhibitor cocktail (B14002; Bimake, Houston, TX, USA), and phosphatase inhibitors (B15002; Bimake, Houston, TX, USA). The lysate was then centrifuged at 12,000 $\times$ g for 10 min at 4 °C, and the supernatant was removed as total protein. According to the manufacturer’s instructions, we used an isolation kit (89801; Thermo Fisher Scientific, Waltham, MA, USA) to extract cytosolic and mitochondrial proteins, which were then separated via gradient electrophoresis on 8–15% sodium dodecyl sulfate gels and transferred to polyvinylidene difluoride (PVDF) membranes (ISEQ00010; Merck Millipore, Darmstadt, Germany). The PVDF membranes were then washed and incubated with the following diluted primary antibodies at 4 °C overnight: NLRP3 (1:1000; NBP2-12446; Novus Biologicals, Littleton, CO, USA), ASC (1:500; NBP1-45453; Novus), Caspase-1 (1:500; NBP1-45433; Novus Biologicals, Littleton, CO, USA), IL-1$\beta$ (1:1000; ab9722; Abcam, Cambridge, UK), IL-18 (1:1000; ab191860; Abcam, Cambridge, UK), pro-Caspase-1 (1:500; 22915-1-AP; Proteintech, Chicago, IL, USA), pro-IL-1$\beta$ (1:200; 16806-1-AP; Proteintech, Chicago, IL, USA), pro-IL-18 (1:500; 10663-1-AP; Proteintech, Chicago, IL, USA), light chain 3B (LC3B; 1:1000; 2775S; Cell Signaling Technology [CST], Danvers, MA, USA), P62 (1:1000; 23214s; CST, Danvers, MA, USA), cytochrome C oxidase subunit 4I1 (COX IV 1, 1:1000; 4850s; CST, Danvers, MA, USA), translocase of outer mitochondrial-membrane 40 homolog (TOMM20; 1:1000; 42406s; CST, Danvers, MA, USA), PINK1 (1:2000; sc-517353; Santa Cruz Biotechnology, Dallas, TX, USA), Parkin (1:1000, 4211s, CST, Danvers, MA, USA), and $\beta$-actin (1:5000; A2228; Sigma-Aldrich, Shanghai, China). After washing, we added horseradish peroxidase-conjugated goat anti-rabbit (1:10,000; AP132P; EMD Millipore, Billerica, USA) or goat anti-mouse immunoglobulin G (1:10,000; AP124P; EMD Millipore, Billerica, USA) secondary antibodies, diluted at 1:10,000, and incubated the PVDF membranes for 1 h at 20–25 °C. Finally, the membranes were washed, developed, and fixed, and images were observed and analyzed using Quantity One software (Bio-Rad Laboratories, Hercules, CA, USA). We used $\beta$ and COX4I1 as internal references for total-brain and mitochondrial proteins, respectively.

We analyzed the collected data using IBM SPSS version 22.0 (IBM Corp., Armonk, NY, USA). The results are expressed as mean $\pm$ standard deviation. For comparisons among multiple groups, a one-way analysis of variance was applied. Pairwise comparisons were conducted using the least-significant-difference method. Differences were considered statistically significant with p 0.05.

We detected NLRP3 inflammasome–related protein expression using WB. The results showed that proteins related to NLRP3 inflammasome were significantly elevated in the model group, as compared with the sham group. Expression of NLRP3, Caspase-1, pro-IL-1$\beta$, IL-1$\beta$, pro-IL-18, and IL-18 reached the highest levels after 24 h of reperfusion (Fig. 1a,b,e–i) whereas the expression of ASC and pro-Caspase-1 reached such levels after 12 h (Fig. 1a,c,d). Therefore, we selected 24 h after reperfusion as the optimal time point for the subsequent experiments.

Fig. 1.

NLRP3 inflammasome was activated, and suppressing it reduced injury during cerebral I/R in rats. (a) WB bands of NLRP3 inflammasome-related proteins in different groups. (b–i) Comparisons of relative protein expression levels among groups (n = 5). (j) TTC-stained brain slices; pale areas are infarcted (n = 8). * p 0.05, ** p 0.01 vs. the sham group; ${}^{\mathrm{\#}}$ p 0.05, ${}^{\mathrm{\#}\mathrm{\#}}$ p 0.01 vs. the 0-h reperfusion group. (k) Statistical bar chart of determined cerebral-infarct volumes, created using Image-Pro Plus software (n = 8). (l) Statistical bar chart of evaluated NDS (n = 11). ** p 0.01 vs. sham group; ${}^{\mathrm{\#}\mathrm{\#}}$ p 0.01 vs. model group.

To determine the role of NLRP3 inflammasome activation in cerebral I/R, MCAO rats received i.p. injections of Ac-YVAD-CMK and MCC950, after which we ascertained their infarction volumes and NDS. The results showed that infarction volumes were larger (Fig. 1j,k), and NDS was higher (Fig. 1l), in the model group than in the sham group. Conversely, infarction volumes and NDS were significantly decreased in both Ac-YVAD-CMK and MCC950 groups versus the model group (Fig. 1j,k,l). These findings showed that cerebral I/R could activate both the NLRP3 inflammasome and Caspase-1 and that inhibiting these could reduce cerebral I/R injury in rats.

To reveal the protective effects of PNS on I/R injury, we recorded the infarction size, NDS, and cell damage rate after PNS treatment. The results of TTC staining showed that the infarct volume was larger in the model group than in the sham group whereas the administration of PNS reduced infarct volume in MCAO rats in a dose-dependent manner (Fig. 2a,b). Neurological function in the sham group was normal (NDS = 0) whereas the model group displayed clear neurological deficits and had a noticeably higher NDS than the sham group (Fig. 2c). Additionally, the NDS and rate of cell damage in the PNS groups were significantly reduced compared with the model group (Fig. 2c,d). From H&E staining, we also found a reduced degree of cell damage that was directly proportional to PNS dose (Fig. 2d). Thus, the conclusion can be drawn that PNS could effectively reduce cerebral I/R injury in rats.

Fig. 2.

PNS reduced cerebral I/R damage in rats. (a) TTC-stained brain slices from the sham, model, and PNS (PNS-L, -M, and -H) groups; pale areas are areas of infarction (n = 8). (b) Statistical chart of determined cerebral-infarct volumes in each group (n = 8), created using Image-Pro Plus software. (c) Statistical bar chart of evaluated NDS in each group (n = 13). (d) Pathological changes in the cerebral cortex and statistical map of the cell injury rate in each group (n = 8). * p 0.05 and ** p 0.01 vs. sham group; ${}^{\mathrm{\#}}$ p 0.05 and ${}^{\mathrm{\#}\mathrm{\#}}$ p 0.01 vs. model group; p 0.05 and p 0.01 vs. PNS-L group; ${}^{\mathrm{\Delta }}$p 0.05 vs. PNS-M group.

We used WB to assess changes in the NLRP3 inflammasome and its downstream proteins in rats treated with or without PNS to explore the function of PNS on NLRP3 inflammasome activation. Compared with the sham group, the expressions of NLRP3, ASC, pro-Caspase, Caspase, pro-IL-1$\beta$, pro-IL-18, IL-1$\beta$, and IL-18 were significantly increased in the model group (Fig. 3a–i). Additionally, there were significant reductions in NLRP3, Caspase-1, IL-1$\beta$, and IL-18 expression levels but no significant difference in ASC, pro-Caspase-1, pro-IL-1$\beta$, or pro-IL-18 levels after PNS treatment, with IL-18 in the PNS-M group and IL-1$\beta$ and IL-18 in the PNS-H group showing the most obvious changes (Fig. 3a–i). This suggested that PNS could inhibit activation of the NLRP3 inflammasome and its downstream molecules.

Fig. 3.

PNS inhibited NLRP3 inflammasome activation and activated mitophagy via the PINK1/Parkin pathway in rats. (a,j) WB analysis of relative proteins in the sham, model, and PNS (PNS-L, -M, and -H) groups (n = 5). (b–i, k–r). Densitometry scanning of band densities was used to quantify expression of proteins in each group using Quantity One software (n = 5). * p 0.05 and ** p 0.01 vs. sham group; ${}^{\mathrm{\#}}$ p 0.05 and ${}^{\mathrm{\#}\mathrm{\#}}$ p 0.01 vs. model group; p 0.05 and p 0.01 vs. PNS-L group; ${}^{\mathrm{\Delta }}$p 0.05 and ${}^{\mathrm{\Delta }\mathrm{\Delta }}$p 0.01 vs. PNS-M group.

To discern the role of mitochondrial autophagy in PNS-mediated neuroprotection, we evaluated changes in protein markers of mitochondrial autophagy via WB. The results indicated that mitochondrial p62 protein, as well as LC3-II/LC3-I ratio of both total-brain and mitochondrial proteins, increased significantly in MCAO rat brain tissues; expression of total-brain P62 and mitochondrial COX4I1 and TOMM20 exhibited the opposite changes (Fig. 3j–p). Furthermore, there was a significant reduction in cerebral P62 level, with insignificant changes in the ratio of cerebral LC3-II/LC3-I after PNS administration. Moreover, in the PNS-M and PNS-H groups, expression levels of COX4I1 and TOMM20 proteins were significantly decreased compared with the vehicle group, and the LC3-II/LC3-I ratio and p62 protein level in mitochondria were also increased (Fig. 3j–p). Additionally, with decreasing expression of total-brain P62, levels of mitochondrial P62, COX4I1, and TOMM20 were concurrently increased in the PNS-H group compared with the PNS-L group (Fig. 3j,m,n,p).

We then examined the effect of PNS on PINK1 and Parkin expression levels. The WB results showed significantly increased mitochondrial PINK1 and Parkin in both the model and PNS-H groups compared with the sham group (Fig. 3j,q,r). These results showed that PNS inhibited activation of NLRP3 inflammasome and activated PINK1/Parkin–mediated mitophagy during cerebral I/R.

We administered autophagy inhibitor 3-MA or the mitochondrial-autophagy inhibitor Mdivi-1 to rats just after MCAO to confirm the effect of PNS-mediated mitophagy on NLRP3 inflammasome activation. As shown in Fig. 4, expression levels of cerebral NLRP3, Caspase-1, IL-1$\beta$, and IL-18 proteins (Fig. 4a–e) along with P62 and the LC3-II/LC3-I ratio of mitochondrial proteins were significantly higher (Fig. 4a,h,i) and TOMM20 and COX4I1 were lower in the model group than in the sham group (Fig. 4f,g). However, only NLRP3, Caspase-1, IL-1$\beta$, and IL-18 levels were reversed; the other proteins mentioned above exhibited further significant changes over their original levels after medium-dose PNS treatment (Fig. 4a–i). Moreover, 3-MA or Mdivi-1 treatment significantly reversed the above protein changes in the PNS + 3-MA and PNS + Mdivi-1 groups (Fig. 4a–i), suggesting that suppression of mitophagy could reverse the inhibitory effect of PNS on activation of the NLRP3 inflammasome.

Fig. 4.

PNS-mediated mitophagy suppressed NLRP3 inflammasome activation and alleviated cerebral I/R injury in rats. (a) WB analysis of relative protein levels in the sham, model, PNS-M, PNS + 3-MA, and PNS + Mdivi-1 groups (n = 5). (b–i) Densitometry scanning of band densities was used to quantify the expression of proteins in each group (n = 5). (j) TTC-stained brain slices from the sham, model, PNS-M, PNS + 3-MA, and PNS + Mdivi-1 groups; pale areas are areas of infarction. (k) Statistical bar chart of determined cerebral-infarct volumes in each group (n = 8), created using Image-Pro Plus software. (l) Statistical bar chart of NDS evaluated according to the Longa method in each group (n = 13). (m) Pathological changes in the cerebral cortex and statistical map of the cell injury rate in each group (n = 8). * p 0.05 and ** p 0.01 vs. sham group; ${}^{\mathrm{\#}}$ p 0.05 and ${}^{\mathrm{\#}\mathrm{\#}}$ p 0.01 vs. model group; p 0.05 and p 0.01 vs. PNS-M group.

To further identify the involvement of mitophagy in the protective effect of PNS, we detected cerebral I/R injury after PNS co-administration with 3-MA or Mdivi-1 in each group. The results showed that cerebral-infarct volume and NDS were significantly elevated in the vehicle group compared with the sham group (Fig. 4j–l). In addition, these two indicators were significantly decreased by medium-dose PNS but increased again after the subsequent addition of 3-MA or Mdivi-1 (Fig. 4j–l). Finally, we also observed similar results in cell damage rates via H&E staining (Fig. 4m). These results suggested that mitophagy mediated the inhibitory effect of PNS on activation of the NLRP3 inflammasome and had anti-brain injury effects in rats with cerebral I/R.