2.1 Preparation of Diabetic Rats

Male Sprague Dawley (SD) rats, with a weight range between 230 and 260 g, were attained from Hunan Slaike Jingda Laboratory Animal Co. Ltd. The Xiangya Hospital of Central South University's Ethics Committee authorized experimental procedures. Random assignment was conducted for rats, who received intraperitoneal injection of streptozotocin (STZ) attained from Sigma-Aldrich that was headquartered in St. Louis (USA) with dissolving in 0.1 mol/L citrate buffer and administered at a dose of 50 mg/kg to trigger DM [29]. After a period of 72 h following STZ injection, it was attempted to quantify fasting blood glucose (FBG) level utilizing an Accu-Chek Compact Plus system attained from Roche Diagnostics that was headquartered in IN (USA). Only rats exhibiting FBG levels exceeding 250 mg/dL were regarded as DM rats and were consequently involved.

2.2 Establishment of MCAO Models

It was attempted to trigger an ischemic stroke model through a precise 90-min transient MCAO procedure [30]. Briefly, male SD rats (230–260 g) were anesthetized and maintained with 5% isoflurane during surgery. The skin of the neck of rats was sterilized with iodine disinfectant and bluntly separated after skin incision. Subsequently, it was attempted to isolate the common, external, and internal carotid arteries. Ligations were applied to the common carotid artery (CCA) and external carotid artery (ECA). Precise clamping of the internal carotid artery (ICA) was accomplished through an arterial clip, followed by a precise incision made at the CCA utilizing ophthalmic scissors. A monofilament attained from Beijing Cinontech Co. Ltd. was inserted to block the middle cerebral artery (18 ~ 20 mm). 90 min after the onset of the MCAO, monofilaments were removed to restore perfusion.

2.3 Culture of RBMVECs and OGD Combined With Hyperglycemia Treatment

SD rats weighing approximately 200 ± 20 g were humanely euthanized using 2% sodium pentobarbital (30 mg/kg), and it was then attempted to immerse them in a bowl containing 75% ethanol for 3–5 min for disinfection. Following ethanol exposure, heads of rats were swiftly decapitated, and their fur was precisely removed to expose the brain. With the aid of tweezers, the resilient meningeal layer was delicately peeled away from the brain's surface, and the left and right brain hemispheres were accurately segregated and transferred into dishes encompassing sterile PBS solution. It was attempted to wash brain tissue three times with 5–10 mL of sterile PBS to eliminate any extraneous debris. Subsequently, the pia mater was precisely detached, and prominent blood vessels were excised from the brain tissue's surface. The cerebral cortex was meticulously harvested and transferred into clean dishes filled with sterile PBS. The harvested cerebral cortex tissue was promptly diced into 1 mm³ fragments, followed by their transfer into 50 mL centrifuge tubes and being homogenized until reaching a homogeneous liquid state. The homogenized tissue was sequentially filtered through 80- and 200-mesh screens, and the cells that retained on the screens were precisely rinsed off with PBS into new centrifuge tubes. The tubes were centrifuged at 1000 g, 4°C, particularly for 5 min, and it was attempted to collect the resulting sediment. The sediment was treated with 2 mL of prepared collagenase solution, followed by a 2-h incubation, particularly at 37°C. After digestion, the tubes were recentrifuged at 1000 g, 4°C especially for 5 min, and the supernatant was thereafter discarded. Resuspending the sediment in 10 mL of sterile PBS was undertaken, followed by centrifuging at 1000 g, 4°C especially for 5 min. It was then attempted to transfer the resuspended sediment to a new centrifuge tube and recentrifuge at 1000 g, 4°C, particularly for 5 min. The resulting sediment was meticulously mixed with 2 mL of prepared mixed media and transferred to 60 mm culture dishes for incubation at 37°C, 5% CO₂ for 24 h. Thereafter, careful observations were made regarding cell attachment and morphology. The mixed media were replaced with fresh EC media to continue cultivation, and upon sufficient coverage of the culture dish bottom by the cells, subculturing was undertaken.

It was attempted to seed RBMVECs into the pore plate at a density of 1.0 × 10⁵ cells/ml and cultured for 24 h to adhere to the wall, followed with a culture medium containing 30 mM glucose for 24 h. These cells were placed in an anaerobic incubator for 6 h with a continuous flux of gas (95% N₂, 5%CO₂) and treated with a glucose-free medium without FBS after being washed with PBS, and then re-oxygenated with a normal medium containing 30 mM glucose under 95% air and 5% CO₂ at 37°C to establish the OGD + HG model. For OGD/R cells, a conventional RPMI-1640 medium was utilized before hypoxia and after re-oxygenation.

2.4 TTC Staining

The brains of MCAO rats were collected and removed quickly and frozen at −20°C for 15 min, and 2-mm-thick brain slices were thereafter cut and put into 2% TTC solution (Solarbio, G3005) in a 37°C water bath and protected from light for 30 min. It was attempted to rinse these slices with PBS for 3–5 min, followed by fixing with 10% neutral formaldehyde for 24 h. The infarct areas of rats in each group were photographed and calculated.

2.5 Nissl Staining

Fresh brain tissue underwent an initial fixation step with paraformaldehyde, followed by a precise dehydration process utilizing a sucrose gradient (15% and 30%). Thereafter, the tissue was embedded into optimal cutting temperature compound (OCT) attained from SAKURA, which was headquartered in the USA, and subjected to freezing for sectioning. It was attempted to immerse resulting sections in a gradient of alcohol ranging from 100% to 75% for a duration of 10 min, followed by thorough rinsing with distilled water. Staining was undertaken through a 1% toluidine blue solution for a period of 15 min, succeeded by additional rinsing until the sections became colorless. Dehydration was precisely carried out utilizing anhydrous alcohol, with transparency achieved through treatment with xylene. Ultimately, the sealing of sections was conducted with neutral gum and subjected to observation under a light microscope.

2.6 TUNEL Staining

The tissue sections underwent a meticulous process involving deparaffinization and permeabilization, which was achieved by treating them with 100 μL of proteinase K working solution for 30 min, particularly at 37°C. Thereafter, the sections were subjected to 20 min of incubation in a 0.3% hydrogen peroxide solution prepared in PBS (0.3% H₂O₂ in PBS) at room temperature to effectively neutralize the endogenous peroxidase activity. An appropriate volume of biotin labeling solution, precisely prepared on the basis of instructions outlined by the manufacturer (Beyotime, C1098), was applied to the samples, ensuring thorough mixing, and subsequently incubated, particularly at 37°C when no light was provided for a period of 60 min. Finally, each sample underwent 15 min of incubation, particularly at room temperature utilizing 0.3 mL of DAB chromoplast and was precisely visualized under a microscope. The enumeration of TUNEL-positive cells in three distinct, non-overlapping fields of brain tissue was carried out for subsequent assay.

2.7 Extraction of Mitochondrial Proteins and Nuclear Proteins

It was attempted to extract mitochondrial proteins and nuclear proteins employing the Cell Mitochondria Isolation kit (C3601) and the Nuclear and Cytoplasmic Protein Extraction kit (P0027), respectively, which were both attained from Beyotime, on the basis of instructions outlined by the manufacturer. The corresponding cytoplasmic plasma proteins were also isolated. In brief, cells were collected by scraping the cells with a cell scraper, followed by centrifuging and discarding the supernatant. Thereafter, the mitochondria were extracted by adding mitochondrial extraction reagent to the cell sediment, which was homogenized and centrifuged to obtain the cytoplasmic protein supernatant and mitochondrial precipitate, which was then lysed with mitochondrial lysate supplemented with PMSF to attain mitochondrial proteins. However, for the extraction of nuclear proteins, it was attempted to add PMSF-containing cytoplasmic protein extraction reagent to the cell precipitate, and the supernatant was aspirated into a pre-cooled plastic tube immediately after centrifugation to obtain the cytoplasmic proteins. Thereafter, addition of 50 μL of cytosolic protein extraction reagent to the precipitate was undertaken, and the supernatant was aspirated by vortexing and centrifugation to obtain the cytosolic proteins.

2.8 Western Blotting

It was attempted to quantify protein concentrations through the BCA assay, followed by protein gel electrophoresis and membrane transfer. Thereafter, overnight co-incubating the protein (20 μg in total)-containing membranes was undertaken, particularly at 4°C, with the primary antibodies outlined in the following: cleaved caspase-3 p17 (1:1000, ZENBIO, R23727), cleaved caspase-9 (1:1000, Proteintech, 10,380-1-AP), Bcl-2 (1:5000, Abcam, ab196495), Bax (1:1000, ZENBIO, 250412), Cytochrome C (1:1000, ZENBIO, R22867), capase-8 (1:1000, HUABIO, RT1099), caspase-12 (1:1000, HUABIO, HA500144), Annexin A2 (1:5000, Abcam, ab189473), p-Annexin A2 (1:1000, Santa Cruz Biotechnology, SC1924), β-tubulin (1:1000, Engibody, AT0003), VDAC1 (1:1000, Proteintech, 55,259-1-AP), p-ERK1/2 (1:1000, Servicebio, GB110041), p-AKT (1:1000, ZENBIO, R22961), p-STAT3 (1:1000, CST, 9145S), p-Ser⁴⁷³ (1:400, SAB4504331, Sigma-Aldrich), and proliferating cell nuclear antigen (PCNA, 1:1000, Servicebio, K109134P). After the primary antibody was eluted, it was attempted to 1-h incubate with the secondary antibody (goat anti-mouse IgG/goat anti-rabbit IgG, 1:5000, CST) particularly at room temperature. The secondary antibody was eluted by PBST, and immunoblot results were visualized on the ImageQuant 800 system.

2.9 Vector Construction and Cell Transfection

To construct lncRNA-MEG3 and Anxa2 vectors, the wild-type (WT) open reading frame (ORF) of lncRNA-MEG3 and Anxa2 was cloned into the pcDNA 3.1 vector (TaKaRa Biotechnology, Dalian, China). Transfection of small interfering RNA (siRNA) or expression vector into RBMVECs was performed using Lipofectamine 2000 (Thermo Fisher Scientific). Besides, the sequences of siRNAs utilized in this study were listed as follows: lncRNA-MEG3 siRNA: 5′-GGAAGAAACUCUGAAGUAA-3′; Plectin siRNA: 5′-GCACUCAUCUUGCGUGACA-3′; Anxa2 siRNA: 5′-GGTCTGAATTCAAGAGAAA-3′; NC siRNA: 5′-UUCUCCGAACGUGUCACGU-3′. All siRNAs were synthesized by Hechuang Biotechnology Co. Ltd. (Guangzhou, China).

2.10 Quantitative PCR (qPCR)

Following the completion of the RIP assay, RNA extraction was precisely undertaken through TRIzol (#15596026) attained from Invitrogen that was headquartered in Carlsbad (CA, USA). Thereafter, it was attempted to synthesize cDNA via reverse transcription, employing the PrimeScript II 1st Strand cDNA Synthesis kit (#6210A) attained from TaKaRa Biotechnology that was headquartered in Dalian (China). Moving forward, qPCR was carefully conducted utilizing the sophisticated CFX96 Touch Real-Time PCR system attained from Bio-Rad that was headquartered in Hercules (CA, USA), in tandem with the highly efficient TB Green Premix Ex Taq (Tli RNaseH Plus) (#RR420A) attained from Takara Biotechnology. The quantification of the target RNA was precisely standardized against the internal control (GAPDH), and it was attempted to indicate the fold change in target RNA expression through the rigorous application of the 2^‒△△Ct method, relative to the control group. Comprehensive primer sequences utilized for qPCR were outlined below:

MEG3: forward: 5′-TTGCAACCCTCCTGGAATAG-3′

Reverse: 5’-ACATGAGACTGTGAGTTGGGTGAC-3′

Anxa2: forward: 5′-GGTCTGAATTCAAGAGAAA-3′

Reverse: 5′-GCCAAAGAAATGAACATTC-3′

GAPDH: forward: 5′-TGCCACTCAGAAGACTGTGG-3′

Reverse: 5′-TTCAGCTCTGGGATGACCTT-3′

2.11 CCK-8 and LDH Assays

It was attempted to inoculate cells into 96-well plates, particularly at a density of 1.0 × 10⁵ cells/mL, and upon completion of the modeling phase, the cell counting kit-8 detection reagent (DOJINDO, GW770) was supplied at a concentration of 10%. Subsequent to 1-h incubation, particularly at 37°C, shielded from light, the absorbance of the test sample at 450 nm was measured via a microplate reader. Similarly, take the cell supernatant after modeling and add it into another new 96-well plate, add 60 μL of lactate dehydrogenase (LDH, Beyotime, C0016) assay working solution respectively, mix well, incubate for 30 minutes, particularly at room temperature in the absence of light, and determine its absorbance at 490 nm. The LDH activity in the sample (mU/mL) was computed by subtracting the absorbance of the background blank control wells from that of the sample wells, followed by dividing this difference by the absorbance difference between the standard tube and the standard blank tube. The resulting quotient was subsequently multiplied by the standard concentration (mU/mL).

2.12 ROS Detection

Adhere to the guidelines of the Reactive Oxygen Species Assay kit (S0033S) that was outlined by the manufacturer (Beyotime). Specifically, mix DCFH-DA with serum-free culture medium in a 1:1000 ratio to create a solution (final concentration: 10 μmol/L). Thereafter, gather cells and immerse them in the diluted DCFH-DA solution at a concentration ranging from 1 to 20 million cells/mL. Subsequently, for a duration of 20 min, subject the suspension to an incubation process at 37°C in a cell culture incubator, carefully inverting the mixture every 3–5 min to promote thorough interaction between the probe and the cells. Afterward, precisely conduct rinsing the cells three times utilizing serum-free cell culture medium for the purpose of eliminating any residual DCFH-DA that might remain unabsorbed in the cellular environment. Finally, measure the fluorescence intensity of ROS utilizing the FITC detector in flow cytometry.

2.13 Flow Detection of Apoptosis

The cell culture solution was aspirated into a centrifuge tube and used to terminate EDTA-free tryptic digestion. After centrifugation and discarding the supernatant, the cells were reconstituted in 400 μL of 1× binding buffer, followed by the sequential addition of 5 μL of Annexin V-FITC and 10 μL of PI staining solution. Subsequently, the cells were subjected to 15 min of incubation, particularly at room temperature, ensuring that they were shielded from light. Thereafter, it was attempted to transfer them to an ice bath, maintaining light protection, for an additional 5 min. Within a comprehensive timeframe of 30 min, it was attempted to analyze the cells utilizing a flow cytometer. During analysis, visualization of the green fluorescence emitted by Annexin V-FITC was undertaken through the FITC channel, while the red fluorescence emitted by PI was recognized via the PI channel.

2.14 COIP Assay

Cells transfected with the FLAG-tag vector were collected and lysed in 500 μL of cell lysate for 30 min to collect the supernatant. The protein A/G beads were divided into IP and IgG groups, and incubated with Flag and IgG antibodies, respectively, at 4°C for 2 h for pretreatment. Subsequently, the cell lysate was added to the IP and IgG groups in the ratio of 2:1 and incubated at 4°C overnight. Prior to conducting immunoblotting on the supernatant, enrich the IP, IgG, and input groups with SDS-PAGE protein sampling buffer (5×), and subject them to 5-min heat denaturation, particularly at 99°C, to achieve effective separation of the magnetic beads.

2.15 Immunofluorescence Analysis

Cells, cultivated on Biosharp coverslips (BS-24-RC), underwent a precise experimental procedure. Initially, it was attempted to fix them for 30 min utilizing 4% paraformaldehyde (BL539A) attained from Biosharp, ensuring optimal preservation of cellular morphology. Thereafter, permeabilization was carefully undertaken with 0.1% Triton X-100 (Solarbio, T8200), allowing for thorough penetration of subsequent reagents. Afterward, the cells were subjected to overnight incubation, particularly at 4°C, with the primary antibody. Post-incubation, it was attempted to wash the cells three times with PBS to exclude any unbound primary antibody. Subsequently, they were exposed to the appropriate fluorescent secondary antibody, which was precisely diluted in 10% FBS-PBS, for a duration of 1 h. Eventually, sealing of coverslips was accurately undertaken through an anti-fluorescence quenching sealer containing DAPI (P36971) attained from Invitrogen. Image acquisition was implemented with the assistance of a microscope attained from Nikon. It was attempted to quantitatively analyze images through ImageJ software.

2.16 Constructing Plasmids With Anxa2 and Its Mutations

Utilizing PCR site-directed mutagenesis technology, the generation of a plasmid harboring mutations at the tyrosine residue 23 of Anxa2 was carried out, substituting it with both aspartic acid and alanine. This process was on the basis of the pEGFP-N3-Anxa2^WT vector. Primer Express 2.0 software was run to design two sets of primers, each containing the desired mutation sequences:

Tyr-Ala-F: 5′-AGTGCAGCTGGGTCTGTCAAAGCCTATACTAAC-3′;

Tyr-Ala-R: 5′-CAGCTGCACTTGGGGGTAGAGTGATC-3′;

Tyr-Asp-F: 5′-AGTGCAGATGGGTCTGTCAAAGCCTArACTAAC-3′;

Tyr-Asp-R: 5′-CATCTGCACTTGGGGGTGTAGAGTGATC-3’.

Underlined parts represent the mutation sites, and primers were synthesized by Beijing Liuhe Genomics Co. Ltd. Upstream and downstream primers were incorporated to the pEGFP-N3-Anxa2^WT plasmid using Phusion DNA polymerase, and PCR was undertaken on a PCR machine. The PCR methodology followed a comprehensive regimen, commencing with an initial denaturation phase at 98°C particularly for 45 s, followed by successive denaturation cycles at 98°C particularly for 10 s, annealing at 70°C particularly for 3 min, and repeated for a total of 25 cycles. Subsequently, a final extension step ensued at 72°C particularly for 10 min, culminating in a gradual cooling process down to 40°C. The PCR product was digested by DpnI enzyme, and the DNA was transformed into E. coli DH5a competent cells. Single clones were picked, and the plasmids were sequenced. The pEGFP-N3-Anxa2^WT plasmid was cut by Notl and Bag II enzymes, and it was attempted to separate the products on a 0.8% agarose gel, recovered, and then cut by BamH 1 enzymes. The separation of reaction products was undertaken on a 0.8% agarose gel, and the gel was imaged using a portable UV analyzer. The desired genes were recovered from the gel utilizing a gel extraction kit. The plasmid was cut twice by Notl and Bag II enzymes, and the products were separated on a 0.8% agarose gel and sequenced.

2.17 Bioinformatics Analysis

It was attempted to utilize catRAPID omics v2.0 to analyze the potential binding sites of Annexin A2 protein on lncRNA-MEG3 and the binding domains of lncRNA-MEG3 on the Annexin A2 protein. Besides, potential RNA-binding domains on Annexin A2 were also analyzed using catRAPID omics v2.0.

2.18 Vector Construction

The WT open reading frame (ORF) of Annexin A2, along with three mutant constructs—a deletion of the 156-bp sequence encoding amino acids 26–77 (MUT1), a deletion encoding amino acids 126–177 (MUT2), and a deletion encoding amino acids 176–227 (MUT3)—was inserted into the pCMV-T7-MCS-3 × FLAG-WPRE-Neo vector. This process resulted in the establishment of FLAG-WT Annexin A2, FLAG-MUT1 Annexin A2, FLAG-MUT2 Annexin A2, and FLAG-MUT3 Annexin A2 expression vectors, respectively.

2.19 Cell Transfection

Transfection of vectors into RBMVECs was undertaken through Lipofectamine 2000 (#11668019) attained from Invitrogen.

2.20 RNA Immunoprecipitation (RIP) Assay

Briefly, 2 × 10⁷ RBMVECs were collected and lysed, followed by ultrasonic treatment to fragment the nucleic acid components. After being centrifuged, cell lysates were incubated with protein A/G beads and FLAG antibody (#66008-4-lg; Proteintech, Rosemont, IL, USA) or IgG antibody (negative control) with gentle shaking at 4°C, particularly for 24 h. Subsequently, A/G beads were collected by a magnetic separation device and cleaned. Then, immunoprecipitated RNAs were eluted and analyzed by PCR and RT-qPCR.

2.21 Statistical Analysis

It was attempted to statistically analyze data through SPSS 22.0 software developed by IBM that was headquartered in Armonk (NY, USA). The expression of continuous variables was in the form of mean ± standard deviation (SD). All data were tested for normality by Shapiro–Wilk. The analysis of normally distributed data was implemented through unpaired Student's t-test for comparisons between two principal groups. One-way ANOVA with Tukey's post hoc test was used for comparison among multiple groups. Data that do not exhibit a normal/Gaussian distribution were analyzed via the Mann–Whitney U test employed as an alternative to the t-test and the Kruskal-Wallis test selected as a replacement for ANOVA. Pearson correlation analysis was conducted to elucidate the strength of association between pairs of variables. To figure out the diagnostic utility of the measured parameters, receiver operating characteristic (ROC) curve analysis was undertaken, accompanied by computation of the area under the curve (AUC) plus 95% confidence interval (CI). A p threshold below 0.05 was suggestive of statistical significance.

3.1 Clinical Characteristics and lncRNA-MEG3 Expression in AIS Cases With or Without DM

Firstly, DM + AIS and DM groups had significantly elevated FBG and hemoglobin A1C (HbA1C) levels than AIS and healthy control groups (p < 0.01). However, the absence of significant differences in both FBG and HbA1C levels, particularly between DM + AIS and the DM groups was noteworthy. Secondly, cases in the DM + AIS and DM groups, had markedly greater serum cholesterol and triglyceride (TG) levels than those in the AIS and healthy control groups (p < 0.05). Similarly, the absence of significant differences particularly in cholesterol and TG levels, was noteworthy between DM + AIS and the DM groups, as illuminated in Table 1. Next, cases in the four groups had similar general features, involving mean age, sex distribution, and body mass index (BMI) (p > 0.05). Finally, cases in DM + AIS and AIS groups had substantially higher C-reactive protein (CRP) and tumor necrosis factor-alpha (TNF-α) levels than cases in DM and healthy control groups (p < 0.01, Figure 1A). The absence of significant difference in their levels, particularly between the DM + AIS and AIS groups, was noteworthy (p > 0.05).

It was attempted to collect the plasma samples from participants, and lncRNA-MEG3 level was quantified via RT-qPCR. As displayed in Figure 1B, DM + AIS cases had a greater level of lncRNA-MEG3 than those in the other three groups (p < 0.01). Moreover, the level of lncRNA-MEG3 exhibited to be higher in AIS group than in healthy control group (p < 0.01). Additionally, the greater level of lncRNA-MEG3 was found in the DM group relative to the healthy control group. Importantly, the NIHSS score in the DM + AIS group was 12.5 ± 3.8, and it was 11.7 ± 2.2 in the AIS group, with a nonsignificant difference. Meanwhile, in AIS cases with DM, elevated lncRNA-MEG3 level was positively correlated with NIHSS score (p < 0.01, Figure 1C). Furthermore, the lncRNA-MEG3 displayed a notable capacity in prognosticating the risk associated with AIS (AUC = 0.7406, 95% CI: 0.6137–0.8674) and AIS risk in diabetics (AUC = 0.8911, 95% CI: 0.8052–0.9770) (Figure 1D,E). Therefore, lncRNA-MEG3 may function as a novel potential prognostic marker for AIS cases, especially for diabetics with AIS.

3.2 Downregulation of lncRNA-MEG3 Alleviated Neurological Impairments Following CIRI in Diabetic Rats Accompanied by Activation of Annexin A2

Through RT-qPCR, it was attempted to quantify lncRNA-MEG3 expression in cerebral tissues of diabetic rats following MCAO. Notably, lncRNA-MEG3 was remarkably elevated in diabetic rats without ischemic stroke when compared with that in normal control (NC). Compared with Sham and DM groups, the MCAO group exhibited markedly elevated expression of lncRNA-MEG3 in ischemic penumbra regions. Moreover, the expression of lncRNA-MEG3 in brain tissues was remarkably escalated following MCAO in diabetic rats, which suggested the relation of lncRNA-MEG3 with CIRI in diabetic rats (p < 0.01, Figure 2A). To determine the notable function of lncRNA-MEG3 in ischemic stroke, it was attempted to conduct lncRNA-MEG3 knockdown in the rat brains utilizing a lentivirus vector. Diabetic rats administered with sh-lncRNA-MEG3 expressed a low level of lncRNA-MEG3 (0.22-fold), relative to the control group. Through TTC staining, it was attempted to evaluate the infarct areas of diabetic rats. Minimal brain infarction was noteworthy in the Sham group, whereas the estimated volumes of infarction in the MCAO and NC-MEG3 groups were remarkably elevated. However, rats in the Sh-MEG3 group had smaller infarct areas than those in the MCAO and NC-MEG3 groups (p < 0.01, Figure 2B,C). Neurological outcomes achieved through the Longa scoring system are illuminated in Figure 2D. No notable neurological deficit score (NDS) was identified in the sham group, whereas a significant reduction was noted in NDS among rats infected with lentivirus containing small hairpin RNA of lncRNA-MEG3, as compared to those infected with the control vehicle at 24 h following reperfusion. Then, quantification of water content particularly in rat brain tissues, was undertaken to evaluate cerebral edema. In ischemic hemisphere, relative to Sham group, the water content, was elevated in the MCAO and NC-MEG3 groups, reflecting the production of cerebral edema, especially in the ischemic hemisphere following CIRI (p < 0.01). Sh-MEG3 group exhibited reduced water content in the brain tissues (p < 0.01, Figure 2E). Hence, lncRNA-MEG3 inhibition exhibited a significant functional enhancement.

In Nissl-stained sections analyzed 24 h post-MCAO, a notable presence of atrophic neurons exhibiting shrunken cytoplasm and damaged nuclei was recognized in the hippocampus and penumbral cortex of MCAO-treated diabetic rats. Conversely, the corresponding regions in the sham group displayed no discernible morphological alterations, serving as the control. Furthermore, the inhibition of lncRNA-MEG3 yielded significant protection against neuronal loss relative to both the MCAO and NC-MEG3 groups (p < 0.01, Figure 3A,B). Subsequently, TUNEL staining was employed to evaluate the impact of lncRNA-MEG3 on neuronal apoptosis within the ischemic hemisphere of diabetic rats with CIRI. Minimal apoptotic neurons were detected in the ischemic hemisphere in the sham group. In contrast, following ischemic stroke, both the MCAO and NC-MEG3 groups exhibited a notable increase in the number of apoptotic neurons in the ischemic hemisphere (p < 0.05). Remarkably, compared to the MCAO and NC-MEG3 groups, the Sh-MEG3 group demonstrated a substantial reduction in the number of apoptotic neurons within the brain tissues (p < 0.05, Figure 3C,D).

The involvement of mitochondria-related apoptosis in modulating CIRI under diabetic conditions has been emphasized. To figure out the mechanisms through which lncRNA-MEG3 could adjust mitochondria-related apoptosis, it was attempted to evaluate alterations in apoptotic signaling in ischemic brains. WB revealed that diabetic ischemic stroke enhanced triggering caspase-3 and -9, while diminishing the Bcl-2/Bax ratio (Figure 3E–G), accompanied by an elevation in the cytoplasmic release of Cyt-C (Figure 3H,I). Similarly, the suppression of lncRNA-MEG3 expression resulted in a remarkable reduction in the levels of active caspase-3 and -9 an elevation in the Bcl-2/Bax ratio, and a prevention of the release of Cyt-C from mitochondria to the cytoplasm. Notably, despite triggering an upregulation in the expression levels of caspase-12 and -8 (p < 0.05) following focal brain ischemic injury, the absence of discernible alterations in ER-associated caspase-12 or Fas-involved caspase-8 upon suppression of lncRNA-MEG3 was noteworthy (Figure 3J,K). These outcomes highlight that lncRNA-MEG3 plays a notable protective function against diabetic CIRI-triggered damage via diminishing mitochondria-derived apoptosis.

Additionally, our previous study had indicated that lncRNA-MEG3-associated mitochondrial protein was Anxa2 [26]. To attain more solid evidence, the impact of lncRNA-MEG3 on Anxa2 expression was detected, as well as activation of phosphorylated Anxa2. Ischemic stroke notably resulted in a substantial reduction in Anxa2 expression. Nevertheless, phosphorylated A2 (activated form of Anxa2) expression was remarkably elevated in MCAO-treated diabetic rats, while lncRNA-MEG3 consumption reserved Anxa2 expression and further promoted the level of phosphorylated Anxa2, suggesting that lncRNA-MEG3 modulated Anxa2 expression and its phosphorylation activation in focal cerebral ischemia injury of diabetic rats subjected to MCAO (p < 0.05, Figure 3L,M).

3.3 Anxa2 Overexpression Inhibited MCAO-Induced Mitochondria-Derived Apoptosis in Diabetic Rats

To more directly assess the notable function of Anxa2 in ischemic stroke of diabetic rats subjected to MCAO, lentivirus vectors cloned with Anxa2 cDNA were injected into the right lateral ventricle to induce Anxa2 overexpression. Firstly, WB indicated that Anxa2 expression was remarkably raised in control diabetic rats (data not shown). TTC staining unveiled that the cerebral infarct volume in the vehicle and MCAO groups was higher relative to the MCAO+Anxa2-OE group in diabetic rats (p < 0.05, Figure 4A,B).

The results of Nissl staining are displayed in Figure 4C,D. Staining of rat brain tissue from the sham group revealed arranged cells with discernible Nissl bodies. Conversely, in the MCAO group, a notable reduction in Nissl bodies in the cytoplasm was identified, accompanied by subdued staining intensity relative to the sham group. Conversely, while the Anxa2-OE group exhibited no significant alteration in cell morphology relative to the MCAO group, a substantial elevation in the number of Nissl bodies was evident. For evaluating neuronal apoptosis in the ischemic penumbra, TUNEL staining was undertaken 24 h, post-CIRI. Abundant TUNEL-positive neurons were identified in the ischemic penumbra of rats in the MCAO and vehicle groups, contrasting with fewer TUNEL-positive neurons recognized in the sham group. Furthermore, the Anxa2-OE group displayed a diminished number of TUNEL-positive neurons relative to the vehicle group (p < 0.05, Figure 4E,F).

As illuminated in Figure 4G,H, the release of Cyt-C from mitochondria to cytoplasm in the MCAO and vehicle groups markedly increased relative to the sham group. Anxa2 overexpression emerged to effectively restore Cyt-C level. Moreover, the correlation of Bcl-2, an anti-apoptotic protein, with Bax, a pro-apoptotic protein, plays a notable function in controlling apoptotic responses and the release of Cyt-C. The outcomes unveiled a remarkable elevation in Bcl-2 protein expression, particularly in the Anxa2 overexpression group relative to the MCAO group, while Bax expression exhibited an inverse trend. Furthermore, in line with expectations, triggering caspase-9 and -3, indicative of apoptotic cell death, displayed a notable reduction in the Anxa2-OE group (p < 0.05, Figure 4I–K). Lastly, the expression level of lncRNA-MEG3 was quantified through RT-qPCR, revealing that Anxa2 overexpression had no discernible impact on its expression in diabetic rats subjected to MCAO (p > 0.05, Figure 4L).

3.4 Knockdown of lncRNA-MEG3 Alleviated OGD + HG-Induced Mitochondria-Dependent Apoptosis

To trigger diabetic brain ischemic reperfusion injury in vitro, RBMVECs underwent a rigorous experimental protocol. Initially, it was attempted to cultivate RBMVECs in the medium containing 30 mM glucose for 24 h, and they were thereafter subjected to 6 h OGD, particularly in an anaerobic incubator, mimicking the ischemic insult. Subsequently, reperfusion was undertaken through reintroducing the RBMVECs to culture in media encompassing 30 mM glucose for a duration of 24 h. As illuminated in Figure 5A, exposure to OGD/R alone resulted in a remarkable elevation of lncRNA-MEG3 expression relative to the control group (p < 0.05). Moreover, the lncRNA-MEG3 expression in the OGD + HG group exceeded that of the OGD/R group, reflecting that sustained exposure to a moderately high concentration of glucose (30 mM) notably elevated lncRNA-MEG3 expression in RBMVECs subjected to OGD.

In RBMVECs, OGD + HG significantly suppressed cellular viability and augmented the cytosolic LDH leakage. In addition, OGD + HG remarkably elevated ROS production and apoptotic cells, impaired the mitochondrial electrochemical gradient (ΔΨm), and escalated mitochondrial permeability transition pore (MPTP) opening. However, knockdown of lncRNA-MEG3 attenuated OGD + HG-induced injury as evidenced by improved cellular viability (p < 0.05, Figure 5B), reduced LDH release (p < 0.05, Figure 5C), and formation of ROS (p < 0.05, Figure 5D,E) and inhibited apoptotic cells (p < 0.05, Figure 5F,G). Meanwhile, lncRNA-MEG3 silencing normalized ΔΨm (p < 0.05, Figure 5H) and recovered the stability of MPTP (p < 0.05, Figure 5I). However, lncRNA-MEG3 overexpression exhibited inverse influences (p < 0.05).

Moreover, the Cyt-C release was enhanced in RBMVECs treated with OGD + HG, while inhibition of lncRNA-MEG3 significantly reduced this release (p < 0.05, Figure 5J,K). The outcomes unveiled that lncRNA-MEG3 inhibition remarkably reduced active caspase-9 and -3 and escalated the Bcl-2/Bax ratio (p < 0.05, Figure 5L–N).

3.5 Anxa2, Identified as a Downstream of lncRNA-MEG3, Appeared Protective for RBMVECs Against OGD + HG-Induced Apoptosis

Although the mRNA expression of Anxa2 was remarkably reduced under OGD + HG conditions, depletion of lncRNA-MEG3 was significantly elevated, whereas overexpression of LncRNA-MEG3 further decreased mRNA expression of Anxa2 (p < 0.05, Figure 6A). Consequently, both overexpression and depletion of lncRNA-MEG3 could modulate mRNA and protein expression of Anxa2, reflecting that lncRNA-MEG3 could modulate Anxa2 expression at the transcriptional level and Anxa2 might serve as a downstream target of lncRNA-MEG3. Furthermore, lncRNA-MEG3 knockdown markedly elevated the phosphorylation of Anxa2, whereas lncRNA-MEG3 overexpression had an opposite effect (p < 0.05, Figure 6B,C).

To further confirm the relationship of lncRNA-MEG3 with Anxa2, rescue experiments were implemented. Firstly, Anxa2 mRNA and protein expression levels were diminished in RBMVECs via transfection of Anxa2-specific small interfering RNA (siRNA) (data not shown). Then, it was noted that silencing of Anxa2 counteracted the effect of lncRNA-MEG3 knockdown on the cellular viability and LDH leakage in OGD + HG-stimulated RBMVECs (p < 0.05, Figure 6D,E). Next, the reduction in ROS production (p < 0.05, Figure 6F,G), apoptotic rate (p < 0.05, Figure 6H,I), normalized ΔΨm (p < 0.05, Figure 6J), and recovered stability of MPTP (p < 0.05, Figure 6K) originating from lncRNA-MEG3 depletion was also counteracted by silencing of Anxa2. Moreover, the increase in the Bcl-2/Bax ratio and reduction in the activation of caspase-9 and -3 (p < 0.05, Figure 6L–N), as well as the inhibition of Cyt-C leakage from mitochondria (p < 0.05, Figure 6O,P) induced by lncRNA-MEG3 depletion, could be reversed by Anxa2 knockdown. Consequently, lncRNA-MEG3 might be characterized as an Anxa2-dependent mitochondria-related apoptosis suppressor.

3.6 LncRNA-MEG3 Knockdown Elevated the Mitochondrial Translocation of Anxa2 Through Promoting Anxa2 Phosphorylation at Tyr23 in OGD + HG-Treated RBMVECs

Bioinformatics analysis by catRAPID omics v2.0 unveiled that 519-594 bp and 559-634 bp of lncRNA-MEG3 appeared as the most potential binding sites of Anxa2. Besides, it was indicated that corresponding binding domains for 519-594 bp and 559-634 bp of lncRNA-MEG3 on Anxa2 included 1-23 aa, 23-77 aa, 139-190 aa, 151-202 aa, and 176-227 aa of Anxa2 (Figure 7A). Further analysis by catRAPID omics v2.0 also revealed that 1-77 aa, 109-174 aa, and 193-259aa of Anxa2 should be RNA binding domains (RBDs). Overall, these outcomes unveiled that 1-23 aa, 23-77 aa, and 176-227 aa of Anxa2 might be candidate binding domains of lncRNA-MEG3.

To further identify the binding domains of lncRNA-MEG3 on Anxa2, FLAG-WT Anxa2, FLAG-MUT1 Anxa2, FLAG-MUT2 Anxa2, and FLAG-MUT3 Anxa2 expression vectors were constructed. FLAG-WT Anxa2 expression vector was inserted with WT Anxa2 ORF, while FLAG-MUT1 Anxa2, FLAG-MUT2 Anxa2, and FLAG-MUT3 Anxa2 expression vectors were inserted with Anxa2 ORF without sequence encoding 1-23aa, 23-77aa, and 176-227aa of Anxa2, respectively.

Next, the above expression vectors were transfected into RBMVECs, followed by the RIP assay conducted by FLAG antibody. Results of the RIP assay demonstrated that FLAG-tagged Anxa2 bound with lncRNA-MEG3. Besides, FLAG-tagged Anxa2 without 176-227aa could still bind with lncRNA-MEG3, whereas FLAG-tagged Annexin A2 without 1-23aa and 23-77aa could not bind with lncRNA-MEG3. Additionally, transfection of a blank FLAG-tagged expression vector had no effect on the interaction of Annexin A2 with lncRNA-MEG3 (Figure 7B,C). Hence, these outcomes together suggest that 1-23aa and 23-77aa of Anxa2 could be lncRNA-MEG3 binding domains.

Previous studies have documented that the phosphorylation of Anxa2 at tyrosine 23 is critical for its activation and subcellular localization [31]. Therefore, it was attempted to identify the influences of lncRNA-MEG3 inhibition on mitochondrial or nuclear translocation of Anxa2 in RBMVECs under OGD + HG exposure. As displayed in (Figure 7D,E), lncRNA-MEG3 knockdown remarkably promoted mitochondrial translocation of p-Anxa2, whereas there was no effect on its nuclear translocation. To attain more solid evidence, it was attempted to transfer the Anxa2 gene into RBMVECs of two types, wild-type and two mutations. One type of mutation type involves the conversion of tyrosine 23 in Anxa2 to alanine (Y23A), making the Anxa2 protein unable to be activated at position 23 tyrosine. The other mutation type is the conversion of tyrosine 23 in Anxa2 to aspartic acid (Y23D), which imitates the continuous phosphorylation of Anxa2 protein at position 23 tyrosine, thereby keeping Anxa2 protein in an activated state. The outcomes unveiled that Anxa2 with high expression and continuous phosphorylation could significantly alleviate the injury induced by OGD + HG, while the Anxa2 protein with continuous dephosphorylation of tyrosine at position 23 lost this function (p < 0.05, Figure 7F,G). Moreover, the anti-apoptotic effects of lncRNA-MEG3 depletion were neutralized in Anxa2^Y23A-treated RBMVECs under OGD + HG conditions (p < 0.05, Figure 7H–L). Contrarily, neurotoxicity of lncRNA-MEG3 overexpression was abolished in RBMVECs following the introduction of Anxa2^Y23D (p < 0.05, Figure 7M–Q). Taken together, it was unveiled that the neuroprotective effect of lncRNA-MEG3 depletion was closely linked to Anxa2 phosphorylation at tyrosine 23.

3.7 Downregulation of lncRNA-MEG3 Promoted Interaction of p-Anxa2 With p-Akt in Mitochondria

Overexpression and activation of tyrosine kinase receptors, such as epidermal growth factor receptor (EGFR), have protective effects on cerebral ischemic injury [32]. Several studies highlighted the noticeable function of Anxa2 phosphorylation at Tyr23 in regulating cancer metastasis through its interaction with EGFR [20]. To obtain further evidence, analysis of EGFR downstream signaling pathways in OGD + HG-stimulated RBMVECs revealed a remarkable elevation, particularly in the phosphorylation levels of STAT3, ERK1/2, and Akt (Figure 8A,B). These outcomes signify a robust induction of these signaling pathways in the presence of the OGD + HG condition. Subsequently, the impact of Anxa2 knockdown on triggering EGFR signaling was figured out. Noteworthy, while the depletion of Anxa2 did not remarkably influence the activation of ERK1/2 and STAT3, it distinctly hindered the phosphorylation of Akt (p-Akt). An augmentation in the expression of p-Akt was discerned in the high Anxa2 expression group (p < 0.05, Figure 8C,D). Additionally, Akt phosphorylation at serine 473 (pSer473-Akt) emerged as critical for its activation and thus the subcellular translocation [33]. Then, the variations in the subcellular location of pSer473-Akt were explored, and it was found that Anxa2 overexpression did not influence subcellular location of pSer473-Akt in the nucleus while markedly promoting pSer473-Akt in the mitochondrial fraction (p < 0.05, Figure 8E,F). Next, the changes in the mitochondrial localization of p-Anxa2 and pSer473-Akt following OGD + HG stimulation in lncRNA-MEG3-silenced cells were quantified. Expectedly, lncRNA-MEG3 knockdown significantly elevated mitochondrial translocation of p-Anxa2 in RBMVECs, and pSer473-Akt was also remarkably elevated in the mitochondria in parallel (p < 0.05, Figure 8G,H). However, the expression of Ser473-phosphorylated Akt in mitochondria was notably diminished in RBMVECs with attenuated Anxa2 levels, while it was significantly restored in Anxa2-overexpressing RBMVECs (p < 0.05, Figure 8I,J). In addition, to test whether the phosphorylation of Anxa2 at Tyr23 is notable for the mitochondrial translocation of pSer473-Akt, WB (Figure 8K,L) and immunofluorescence staining analysis (Figure 8M) unveiled that pSer473-Akt was notably localized in the mitochondria of Anxa2-Y23D cells under OGD + HG conditions, compared with that of Anxa2-WT, whereas less pSer473-Akt was detectable in Anxa2-Y23A cells. Hence, Anxa2 phosphorylation at Tyr23 escalated pSer473-Akt expression in mitochondria. Noteworthy, lncRNA-MEG3 depletion elevated the interaction between p-Anxa2 and p-Akt in mitochondria as evidenced through co-immunoprecipitation (COIP) assay (p < 0.05, Figure 8N,O) and immunofluorescence staining (Figure 8P,Q). Altogether, the data unveiled that Anxa2 is essential for Akt phosphorylation activation and mitochondrial translocation triggered by OGD + HG, and the interaction between phosphorylated Anxa2 and Akt in mitochondria may play a role in the neuroprotection conferred by lncRNA-MEG3 depletion under OGD + HG conditions.

Additionally, compared with the OGD/R group, OGD combined with HG markedly promoted the mitochondrial translocation of lncRNA-MEG3. In addition, the findings also demonstrated that Plectin silencing significantly inhibited the mitochondrial localization of lncRNA-MEG3 in RBMVECs under OGD + HG exposure (p < 0.05, Figure 9A). Subsequently, COIP analysis further revealed that p-Anxa2 strongly interacted with p-Akt in Plectin-siRNA-treated RBMVECs (p < 0.05, Figure 9B,C), suggesting that Plectin impeded the interaction between p-Anxa2 and p-Akt in mitochondria. Hence, lncRNA-MEG3 translocated into mitochondria in a Plectin-dependent manner and subsequently impeded the interaction of p-Anxa2 with p-Akt in mitochondria. Finally, the major outcomes could be confirmed using the in vivo model. Consequently, compared with the sham group, the MCAO group exhibited markedly elevated lncRNA-MEG3 expression in mitochondria of ischemic penumbra regions. Moreover, lncRNA-MEG3 expression was remarkably escalated following MCAO in diabetic rats, which suggested that diabetic CIRI promoted mitochondrial translocation of lncRNA-MEG3 (p < 0.01, Figure 9D). Moreover, mitochondrial translocation of p-Anxa2 and p-Akt (p < 0.05, Figure 9E,F), and the interactions between p-Anxa2 and p-Akt in the mitochondria (p < 0.05, Figure 9G,H) were elevated in diabetic rats subjected to ischemic stroke following lncRNA-MEG3 silencing, as detected by WB and COIP assays, respectively.