We bought the Sprague-Dawley (SD) rats (males, N = 40, aged 10 weeks, weighed 240–290 g) from the SiPeiFu Biotech (Beijing, China), and all the rats were fed under the specific-pathogen-free (SPF) circumstances with 12-hour light–dark cycle, standard humidified atmosphere (60 ± 5%) and were freely accessible to water and food. The Puerarin was bought from North China Pharmaceutical Co., Ltd. (Shijiazhuang, China), and was diluted in the physiological PBS buffer at the final concentration of 25 mg/mL for further utilization at 100 mg/kg each rat according to our preliminary data (data not shown) and previous literature (Wang et al., 2018; Zhou et al., 2014). Then, the middle cerebral artery occlusion (MCAO) method was used to establish I/R rat models with the protocols as follows: the rats were intraperitoneally injected with 10% chloral hydrate at the concentration of 350 mg/kg for anesthetization, and a monofilament nylon suture was inserted into the internal carotid to generate MCAO for 90 min, which were then removed and allowed reperfusion for 48 h. Finally, the rat hippocampus tissues were separated and collected for further analysis. We declared that all the animal experiments were approved by the Ethics Committee of Ningbo Rehabilitation Hospital (No. 2018NBR67362).

TTC staining assay was performed to evaluate the infarcted area volume in rat brain in keeping with the standard experimental procedures (Zhang et al., 2019a). Briefly, the rat brains were collected and prepared as sections with 2 mm thickness, which were further stained with 2% TTC (Sigma, MO, USA) for 30 min at 37°C, and the tissues were subsequently fixed with 4% formalin overnight at room temperature. The infarcted area was observed, and the volume was calculated.

Based on the experimental protocols provided by the previous work (Qin et al., 2020), the male newborn rat (N = 5) were used for the isolation and purification of the primary hippocampus neuron (PHN) cells, which were maintained in the Neuro-basal Medium (Invitrogen, USA) containing poly-D-lysine (Invitrogen, USA), 2% B27plus glucose (5 g/L) and 0.25% Glumax, in the incubator with standard culture conditions. At day 4 post-culture, the PHN cells were subjected to hypoxic stimulation to induct H/R cellular models as previously described (Qin et al., 2020), which were further stimulated by Puerarin (80 μM) for 0, 24, and 48 h, respectively. Also, the HEK293T were obtained (ATCC, USA) and cultured in the DMEM medium (Gibco, USA) with 10% FBS (Gibco, USA).

A commercial third-party company GenePharma (Shanghai, China) was employed to design and synthesize the LncRNA DUXAP8 overexpression/downregulation vectors, and the mimic and inhibitor for miR-223-3p were obtained from GenePharma (Shanghai, China), we delivered the above vectors into the PHN cells by using the Lipofectamine 2000 transfection kit (Invitrogen, USA). At 24 h post-transfection, Real-Time qPCR was conducted to examine the transfection efficiency.

Total RNA was extracted from cells and tissues by using the commercial TRIZOL reagent (Beyotime, Shanghai, China), which were further reversely transcribed into cDNA and quantified by performing Real-Time qPCR analysis. The detailed experimental procedures and primers could be found from the previous work (Zhao et al., 2020).

The RIPA lysis buffer (Beyotime Biotech, Shanghai, China) was used for protein extraction and BCA kit (Beyotime Biotech, Shanghai, China) was used for protein quantification. The proteins were separated by 10% SDS-PAGE, and were transferred onto PVDF membranes (Millipore, USA). The membranes were blocked and were sequentially incubated with primary/secondary antibodies. Finally, the Western Blot Hyper HRP Substrate (TAKARA, USA) was used for protein bands visualization.

The primary hippocampus neuron (PHN) cells were cultured in the 96-well plates and treated with 10 μL of MTT (3-[4, 5-dimethylthiazol-2-yl]-2, 5-diphenyltetrazolium bromide) solution (5 mg/mL) for 4 h incubation at 37°C, and the supernatants were carefully discarded and the 96-well plates were added with dimethyl sulfoxide (DMSO). Then, the plates were vortexed, and a microplate reader (ThermoFisher, USA) was used to evaluate cell proliferation. Also, the cells were stained with trypan blue staining buffer (Merck, USA) to determine cell viability, and the cells stained with blue were regarded as dead cells.

In this study, we respectively used TUNEL apoptosis kit (Beyotime, Shanghai, China) and the Annexin V-FITC/PI staining reagent (Invitrogen, USA) followed by Flow Cytometer (Becon Dickinson FACS Calibur, USA) detection to evaluate cell apoptosis in PHN cells, and the protocols were well-documented in the manufacturer’s instructions.

The targeting sites of miR-223-3p with LncRNA DUXAP8 and 3’ untranslated region (3’ UTR) of NLRP3 were predicted, and the targeting sites in LncRNA DUXAP8 and NLRP3 were mutated and named as Mut-LncRNA DUXAP8 and Mut-NLRP3. In addition, their wild-type counterparts were shown as Wt-LncRNA DUXAP8 and Wt-NLRP3. The above sequences were cloned into the p-MIR-luciferase reporter plasmids (GenePharma, Shanghai, China), and were co-transfected with miR-NC, miR-223-3p mimic and inhibitor into the HEK293T cells. At 48 h post-transfection, the relative luciferase activities were measured.

The supernatants of the PHN cells were collected, and the expression levels of IL-1β and IL-18 were examined by using their corresponding ELISA kit (RAPIDBIO, CA, USA) in keeping with the producers’ instruction. The optical density (OD) values were measured by using a microplate reader (ThermoFisher Scientific, USA) with 450 nm wavelength.

SPSS 18.0 software was used to analyze the data, which were shown as mean ± SD. Specifically, the one-way ANOVA analysis was used for the comparisons among different groups. Individual experiment was repeated for at least 3 times, and *P < 0.05 was considered as statistical significance.

Initially, according to the previous publications (An et al., 2019; He et al., 2019; Ma et al., 2019; Qu et al., 2019; Ye et al., 2019), we established the in vivo I/R rat models. In Fig. 1A, the results showed that the white unstained infarcted area was lager in I/R rats but not their normal counterparts (P < 0.01). Next, the hippocampus tissues were collected from the brain of the rats for further analysis. The hippocampus tissues were stained by TUNEL, and the results in Fig. 1B showed that I/R treatment significantly increased the TUNEL-positive apoptotic cell ratio in rat tissues (P < 0.01), and the data in Suppl. Fig. S1A supported that I/R significantly increased neurological deficit score in rats, as determined by using the Longa scoring system. Consistently, the expression levels of cleaved Caspase-3 and Bax were increased (P < 0.05, Fig. 1C), while the Cyclin D1 and CDK2 were increased (P < 0.05, Fig. 1D) in rat tissues, suggesting that I/R suppressed hippocampus viability in vivo. Interestingly, as shown in Fig. 1E, we noticed that I/R also upregulated NLRP3 and ASC to induce cell pyroptosis in rat hippocampus tissues (P < 0.05). In addition, the H/R models in the PHN cells were inducted to simulate the real conditions of I/R injury in vitro. As expected, cell proliferation (Fig. 1F) and viability (Fig. 1G) were significantly decreased in the PHN cells treated with H/R, in contrast with the normal PHN cells (P < 0.05). Also, the levels of NLRP3, ASC, IL-1β and IL-18 in PHN cells (P < 0.05, Fig. 1H) and their supernatants (P < 0.05, Figs. 1I and 1J) were also increased by H/R.

As previously described (Guan et al., 2020; Wang et al., 2020), Puerarin exerted its neuroprotective effects during I/R injury, which were validated by our following experiments in Figs. 2A–2I and Suppl. Fig. S1A and S1B. Specifically, the rats were divided into four groups, including Sham, I/R group, Puerarin (100 mg/kg) group, and I/R + Puerarin (100 mg/kg) group, and the results in Suppl. Figs. S1A and S1B showed that Puerarin decreased neurological deficit score and infarcted area volume in I/R rats. Further Western Blot analysis results showed that Puerarin downregulated cleaved Caspase-3 and Bax (P < 0.05, Fig. 2A), and upregulated Cyclin D1 and CDK2 (P < 0.05, Fig. 2B) in rat hippocampus tissues. Besides, NLRP3 and ASC expressions were also downregulated by Puerarin treatment (P < 0.05, Fig. 2C). In addition, our in vivo results were supported by the following in vitro experiments, and the results suggested that Puerarin rescued cell proliferation (P < 0.05, Fig. 2D) and viability (P < 0.05, Fig. 2E) in PHN cells treated with H/R. Also, we evidenced that Puerarin attenuated cell apoptosis in H/R-treated PHN cells (P < 0.05, Fig. 2F). Finally, the aggravating effects of H/R on NLRP3, ASC, IL-1β and IL-18 in the PHN cells (Fig. 2G) and their supernatants (Figs. 2H and 2I) were abrogated by co-treating cells with Puerarin (P < 0.05), indicating that Puerarin attenuated I/R- and H/R-induced hippocampus injury in vivo and in vitro.

NLRP3 could be negatively modulated by its upstream transcriptional regulators, including miRNAs (Hu et al., 2019; Yu et al., 2018), which were closely associated with I/R injury and could be modulated by Puerarin. Therefore, we speculated that Puerarin might regulate NLRP3-mediated cell pyroptosis through modulating miRNAs. To validate this hypothesis, ten miRNAs that potentially regulated NLRP3 was selected, and we found that only miR-223-3p, but not other types of miRNAs, could be downregulated by H/R, but it was upregulated by Puerarin in PHN cells (P < 0.05, Fig. 3A). Consistently, the results in Fig. 3B showed that I/R treatment also decreased miR-223-3p levels in rat hippocampus tissues, which could be significantly increased by Puerarin (P < 0.05, Fig. 3B). Next, the targeting sites of miR-223-3p with 3’ untranslated regions of NLRP3 mRNA were predicted, and the targeting sites in NLRP3 mRNA were subsequently mutated (Fig. 3C). As shown in Fig. 3D, the dual-luciferase reporter gene system assay validated that miR-223-3p bound to the 3’UTR of NLRP3 mRNA. Subsequently, miR-223-3p was downregulated and upregulated in the H/R PHN cells (Fig. 3E), and the following Real-Time qPCR and Western Blot analysis results evidenced that miR-223-3p negatively regulated NLRP3 expressions in H/R treated PHN cells at both transcriptional (P < 0.05, Fig. 3F) and translational (Fig. 3G) levels. In addition, the inhibiting effects of Puerarin on H/R-induced cell pyroptosis in PHN cells were partly reversed by silencing miR-223-3p (P < 0.05, Figs. 3H and 3I).

Given that there existed LncRNAs-miRNAs-mRNAs ceRNA networks, and LncRNAs had also been reported to participate in the regulation of I/R injury (Wan et al., 2020; Zeng et al., 2019; Zhang et al., 2019b; Zhong et al., 2019), we next screened out that the miR-223-3p/NLRP3 pathway could be modulated by LncRNA DUXAP8 in PHN cells in Figs. 4A–4I. Specifically, seven LncRNAs containing the binding sites with miR-223-3p were selected, and further Real-Time qPCR results screened out that LncRNA DUXAP8 was upregulated by H/R treatment in PHN cells, which could be significantly downregulated by Puerarin (P < 0.05, Fig. 4A). Consistently, we evidenced that LncRNA DUXAP8 was also upregulated by I/R and downregulated by Puerarin in rat hippocampus tissues (P < 0.05, Fig. 4B). Hence, the LncRNA DUXAP8 was chosen for further analysis. The targeting sites of miR-223-3p with LncRNA DUXAP8 (Fig. 4C) were predicted, which were validated by the following experiments (Fig. 4D). Next, the overexpression and silencing vectors for LncRNA DUXAP8 were delivered into the H/R PHN cells (Fig. 4E), by performing Real-Time qPCR and Western Blot, we found that knock-down of LncRNA DUXAP8 decreased the mRNA and protein levels of NLRP3 in H/R-treated PHN cells, which were reversed by silencing miR-223-3p (P < 0.05, Figs. 4F and 4G). In addition, we found that Puerarin decreased the expression levels of NLRP3 in H/R-treated PHN cells, which were increased by upregulating LncRNA DUXAP8 (P < 0.05, Figs. 4H and 4I).

The above data encouraged us to assure whether the LncRNA DUXAP8/miR-223-3p axis involved in regulating the development of I/R-induced injury in vitro. To achieve this, LncRNA DUXAP8 was silenced (P < 0.05, Fig. 4E), while miR-223-3p was overexpressed (P < 0.05, Fig. 3E) in H/R PHN cells, and the results showed that both LncRNA DUXAP8 ablation and miR-223-3p overexpression restored cell proliferation (Fig. 5A) and viability (Fig. 5B) in H/R-treated PHN cells (P < 0.05). In addition, the cell apoptosis examination assay results indicated that H/R treatment significantly promoted cell apoptosis in PHN cells, which were reversed by ablating LncRNA DUXAP8 and overexpressing miR-223-3p (P < 0.05, Figs. 5C and 5D). Further results validated that silencing of LncRNA DUXAP8 and overexpression of miR-223-3p also downregulated NLRP3, ASC, IL-1β and IL-18 to inactivate NLRP3-mediated cell pyroptosis in H/R PHN cells (P < 0.05, Fig. 5E) and the supernatants (P < 0.05, Figs. 5F and 5G).

Finally, we validated that Puerarin regulated the LncRNA DUXAP8/miR-223-3p axis to exert its neuroprotective effects on H/R PHN cells. Specifically, the H/R PHN cells were pre-transfected with LncRNA DUXAP8 overexpression vectors (P < 0.05, Fig. 4E) and miR-223-3p ablation vectors (P < 0.05, Fig. 3E), and were divided into four groups, including Control, Puerarin group, Puerarin + OE- LncRNA DUXAP8 group, and Puerarin + KD-miR-223-3p group. The results showed that Puerarin rescued cell proliferation in H/R PHN cells, which were abrogated by overexpressing LncRNA DUXAP8 and downregulating miR-223-3p (P < 0.05, Fig. 6A). Consistently, further data validated that both LncRNA DUXAP8 overexpression and miR-223-3p ablation abrogated the promoting effects of Puerarin on cell viability in H/R PHN cells (P < 0.05, Fig. 6B). Furthermore, we evidenced that Puerarin also suppress cell apoptosis in H/R PHN cells by targeting the LncRNA DUXAP8/miR-223-3p axis (P < 0.05, Fig. 6C). The above results suggested that Puerarin regulated the LncRNA DUXAP8/miR-223-3p axis to attenuate cell death in H/R PHN cells in vitro.