miR-26a Improves Microglial Activation and Neuronal Apoptosis in a Rat Model of Cerebral Infarction by Regulating the TREM1-TLR4/MyD88/NF-κB Axis

Cerebral infarction is the primary factor causing stroke, disability, and death worldwide, and its prevalence is increasing yearly [1]. Following ischemic stroke, metabolic dysregulation, and immune disruptions are closely associated with perioperative complications [2]. Mediating microglial activation and subsequent neuroinflammation after cerebral ischemia is a potential therapeutic strategy for ischemic stroke-induced neurological injuries [3, 4].

MicroRNAs (miRNAs) are highly conserved small endogenous noncoding RNAs (approximately 22 nt in length). They complete the cleavage or translation inhibition through combination with the 3′-UTR of the target gene [5]. It has been reported that miRNAs are aberrantly expressed in diverse neurological diseases and can serve as essential targets for disease diagnosis and treatment [6]. Recent studies have suggested that miR-16, which regulates gene expression after ischemic stroke, is a prognostic marker for hyperacute cerebral infarction [7]. In addition, low expression of miR-29b and miR-424 in patients suffering from acute cerebral infarction is negatively correlated with IL-6, IL-4, and TNFα serum levels, indicating poor prognosis [8]. miR-26a is a miRNA that contributes to tumor progression as an oncogene or tumor suppressor gene [9, 10]. Furthermore, we found that lentivirus-mediated miR-26a-modified neural stem cells alleviate brain injury and inhibit brain cell apoptosis and astrocyte activation in cerebral palsy rats [11]. However, whether miR-26a regulates cerebral infarction-induced neuronal apoptosis remains unknown.

Triggering receptors expressed on myeloid cells (TREM) are the cell surface receptors that regulate myeloid cells [12]. TRE-1 stimulates microglial apoptosis by activating the NLRP3 inflammasome in subarachnoid hemorrhage (SAH) mice, while intranasal administration of the TREM1 antagonist in SAH mice alleviates microglial apoptosis by reducing the levels of the N-terminal fragment of GSDMD (GSDMD-N) and IL-1β [13]. Additionally, TREM1 is implicated in SAH-induced early brain injury by facilitating p38MAPK/MMP-9 activation and ZO-1 degradation, while the inhibitory effect of TREM1 eases early brain injury [14]. Thus, TREM1 aggravates inflammation in the cerebral hemorrhage model, while its function in cerebral infarction remains elusive.

Toll-like receptor 4 (TLR4), an immune surveillance receptor, exerts an essential function in innate immunity [15]. MyD88 combines IL-1 receptors (IL-1R) or TLR family members with IL-1R-associated kinase (IRAK) family kinases through homotypic protein-protein interactions [16]. The activation of kinases in the IRAK family culminates in multifarious functional outputs, such as the activation of nuclear factor-κB (NF-κB), mitogen-activated protein kinase, and activator protein 1, making MyD88 a central node in the inflammatory pathway. Several studies have shown that TLR4/MyD88/NF-κB contributes to inflammation [17]. For instance, Su Q et al. found that TLR4/MyD88/NF-κB signaling participates in CME-induced myocardial inflammation. It initiates the NLRP3 inflammasome, promotes the inflammatory cascade, and exacerbates myocardial damage [18]. In addition, dexmedetomidine dampens inflammation and pulmonary edema in lung ischemia-reperfusion injury by attenuating TLR4/MyD88/NF-κB expression [19]. On the other hand, studies have shown that miR-27a reduces inflammation and apoptosis by blocking TLR4/MyD88/NF-κB activation, thereby relieving LPS-induced acute lung injury [20]. However, the specific mechanisms of miR-26a and TLR4/MyD88/NF-κB in cerebral infarction are unknown.

Here, we aim to probe the impact of miR-26a on the middle cerebral artery occlusion (MCAO) rat model, the oxygen-glucose deprivation (OGD)-induced hippocampal neuron cell line HT22 injury model and the OGD-induced microglia cell line BV2 injury model and the relevant mechanisms. It was found that overexpressing miR-26a decreased TREM1-mediated microglial inflammation and neuronal apoptosis. Thus, we speculated that miR-26a inactivated TLR4/MyD88/NF-κB pathway by inhibiting TREM1 expression, thus alleviating cerebral infarction-induced neuronal apoptosis.

The Chengdu Dashuo Experimental Animal Co., Ltd. supplied 100 male Sprague-Dawley (SD) rats (220–250 g in weight). The rats were reared in cages and given food and water with a 12-h light-dark cycle under the conditions of 50–52% humidity at 20–25°C. All experiments received imprimatur from the Ethics Committee of Bazhong Central Hospital and abided by the Guidelines on Animal Care and Use of the National Institutes of Health.

The rat model of ischemic injury was established through MCAO [21]. In brief, 5% isoflurane was applied for anesthesia (the anesthesia machine was adjusted to 1.0 L/min O₂ and 1.0 L/min N₂O). Then, the isoflurane dose was reduced to 1–2% to maintain anesthesia. A laser Doppler flowmeter was used to measure cerebral blood flow in the parietal bone (2 mm posterior and 3 mm lateral to the bregma) prior and subsequent to ischemia. The resting CBF value of each rat was taken as the baseline, and the cerebral ischemia-induced alterations in blood flow were presented as the percentage of resting value. A midline neck incision was made to expose the right common carotid and right external carotid arteries. The external carotid artery was dissected to the distal end. Iris excision scissors were adopted to make a minimal incision on the external carotid artery stump. Subsequently, a silicon-coated 6-0 nylon monofilament (Doccol Corp; 0.23 mm) was employed to construct MCAO. During right MCAO, an approximately 60% reduction in cerebral blood flow was detected via laser Doppler flowmeter. The blood vessels of the sham animals were exposed without blocking the middle cerebral artery. After the surgery, the rats were rested in cages at 34°C for 1 h.

Agomir-26a and agomir-NC (0.8 nmol dissolved in 4 μL PBS) were intracerebroventricularly injected 2 h before MCAO. After anesthesia, the rats were prone to the stereotactic instrument (RWD Life Science, China). The rat scalp was cut along the midline, and a burr hole was drilled on the right side of the skull (0.5 mm posterior, 1.0 mm lateral). Agomir-26a and agomir-NC (4 μL) were injected into the right ventricle using a Hamilton syringe (2.5 mm vertical) driven by a syringe pump (KDS 310, KD Scientific) at a speed of 0.2 L/min. After the injection, the syringe needle was maintained in situ for another 5 min to guard against any leakage. Next, the needle was pulled out slowly within 4 min. Next, the pores were filled with bone wax, and the incisions were sutured with suture lines to make the mice return to normal.

Behavioral changes in the rats were monitored through the Morris water maze assay [22]. A round stainless-steel pool (diameter: 100 cm; height: 50 cm) was divided into four quadrants. In addition, a hidden round platform (diameter: 9 cm; height: 27 cm) was placed at the center of the target quadrant. The temperature of the pool was kept at 22 ± 1°C, and the water level was 1 cm higher than the platform. Then, an appropriate amount of milk powder was given to the pool to hide the platform. The motion track of rats was recorded by a camera above the pool and evaluated with the assistance of EthoVision XT. The experiment included a navigation test and a spatial probe test.

For the navigation test, we placed the rat in the water farthest from the platform with its back against the wall. The rats were expected to find the platform if they climbed up the platform within 60 s and stayed for more than 3 s. In the meantime, the latency time was recorded. If the rats failed to reach the platform within 60 s, the latency time was set as 60 s. Then, the animals were moved to the platform to rest for 15–20 s for the next experiment. The experiment was carried out for 5 days in total, once in the morning and once in the afternoon.

In the spatial probe test, the hidden platform was removed the next day after the navigation test, and the rats were placed into the water with a random entry point for 60 s. The number of times the rats crossed the platform quadrant and their residence time in this quadrant were documented. Noldus EthoVision XT was used for analysis.

As described in a previous study, mNSS was harnessed to evaluate the sensorimotor function of the rats [23]. Five rats were chosen from each group 24 h after cerebral ischemia. Later, well-trained researchers blinded to the experiment scored the walking, sensation, tail-lift, balance, and reflex loss of the rats according to mNSS. The lowest total score was 0, suggesting that the rat was completely normal without any neurological impairment. The highest total score was 18, indicating that the rat lost consciousness or died. Rats with scores ranging from 3 to 14 were selected for subsequent experiments.

As shown in a previous study [24], rats were killed, and their brains were removed on the third day after modeling. Then, the fresh cerebral cortex approximately 2 mm away from the edge of the bone window was taken, and the wet weight (W) was determined on the electronic analytical balance. After that, the cerebral cortex was coated with numbered tin foil with known weight and baked in an oven (100°C) for 24 h. Next, we ascertained the dry weight (D). $Thecontentofwater%=W−DW×100%$⁠.

Twenty-four hours after the modeling, the rats (n = 5) were executed and infused with 10% formalin to take the brains. The brains were paraffin-embedded and sectioned (5 μm thick). Then, gradient ethanol hydration, hematoxylin and eosin staining, dehydration, and mounting were performed. The results were observed under an optical microscope (×200). Six sections were taken from each damaged area of rats. Morphological changes in neurons from each group were monitored with the use of Motic 6.0 (n = 5). The number of viable neurons in the high-power field of vision was counted, and the average value was calculated [25].

The ischemic brain tissue was collected, immersed in optimal cutting temperature compound for 10 min, frozen, and sectioned (15 μm thick), and then the sections were stained with Iba1 immunofluorescence staining. After that, the sections were removed and blocked at 37°C for 2 h with 10% sheep serum. Then, the sections were incubated overnight (4°C) with the primary anti-Iba1 antibody (1:300, ab178846, Abcam). Next, PBS was used to flush the slices 3 times. They were incubated with the secondary antibody for 2 h at room temperature (RT). Subsequently, they were cleared three times using PBS, mounted with DAPI, and imaged with confocal microscopy (×200). Three randomly chosen sections in each rat’s brain tissue were dyed, and 5 fields were randomly taken for photography in each section.

The method was described previously [26]. Twenty-four hours after cerebral ischemia, 5 rats in each group were anesthetized and perfused with 200 mL heparinized saline and 4% paraformaldehyde. After that, the rats were killed, and their brains were removed and immobilized in 4% paraformaldehyde for 24 h. Next, the tissues were dehydrated, transparentized, embedded, and sliced up (5 μm thick) to make paraffin sections. Following deparaffinization, rehydration, antigen retrieval, removal of endogenous peroxidases, and blocking, the sections were incubated with anti-NeuN antibody (cat. no. ab104224, Abcam, 1:300) at 4°C for 15 h. After washing, HRP-labeled secondary antibody (cat. No. 8125, Cell Signaling Technology, Inc.) was used for incubation with the sections for 1 h at RT; the Dako Envision kit HRP (cat. No. K4006; Dako; Agilent Technologies, Inc.) was used for visualizing the immunoreactivity. Hematoxylin was used for counterstaining at 22°C for 5 min. A total of five fields of view were randomly selected under a light microscope, and the percentage of cells positive for the neuron-specific protein was analyzed using ImageJ, and the average value was calculated.

Hippocampal neurons HT22 and microglia BV2 were obtained from ATCC (Rockville, USA). A DMEM (Gibco, MA, USA) filled with 10% fetal bovine serum (Thermo Fisher Scientific, MA, USA) and 1% penicillin/streptomycin (Invitrogen, CA, USA) was utilized for the culture of cells, which was implemented in an incubator under the conditions of 5% CO₂ and 37°C. The cells amid logarithmic growth were trypsinized by employing 0.25% trypsin (Thermo Fisher HyClone, Utah, USA) and subcultured.

GenePharma Co., Ltd. (Shanghai, China) provided us with pcDNA-TREM1 (TREM1, 5 μg/mL), the miRNA control (miR-NC, 50 nm), and miR-26a mimics (miR-26a, 50 nm). Prior to the transfection of cells, HT22 and BV2 cells were seeded onto 24-well plates (3 × 10⁵ cells/well) and maintained for 24 h (37°C, 5% CO₂). Then, Lipofectamine^® 3000 (Invitrogen; Thermo Fisher Scientific, Inc.) transfected BV2 and HT22 cells. Quantitative real-time polymerase chain reaction (qRT-PCR) was used to evaluate the validity of the transfection. The cells were incubated (37°C, 5% CO₂) for 24 h for further analysis.

Briefly, DMEM flushed neurons and microglia 3 times subsequent to 6 days of culture [27]. Glucose-free DMEM was prebalanced with 1% O₂, 5% CO₂, and 94% N₂ in an incubator at 37°C. Subsequently, the neurobasal medium was substituted with DMEM without glucose (Gibco, Carlsbad, CA, USA), and the cells were relocated to an incubator supplemented with 1% O₂, 5% CO₂, and 94% N₂ and kept at 37°C for 1.5 h. The control group was cultured for the same time in neurobasal medium (pH 7.2) in an atmospheric incubator with 5% CO₂. The relevant indexes were determined after 24 h. The LDH Cytotoxicity Assay Kit (cat. no. C0017, Beyotime, Shanghai, China) was used for evaluating the cytotoxicity of HT22 cells.

For evaluating HT22 cell viability, the CCK8 assay (cat. no. C0037, Beyotime, Shanghai, China) was used for cell viability detection. HT22 cells were seeded in 96-well plates, and each well contained 5,000 cells. Following miR-26a mimic transfection and OGD treatment, 10 μL CCK8 solution was added to each well, and the plates were put in an incubator at 37°C for 1 h. After that, the OD value was tested using a microplate reader at 450 nm.

Using TRIzol Reagent (Invitrogen, Carlsbad, CA, USA), we extracted total RNA from the brain tissues of MCAO, BV2, and HT22 rats. The PrimeScript^™ RT Reagent Kit (Invitrogen, Shanghai, China) was used to reverse transcribe the RNA into cDNA. qRT-PCR was performed with the assistance of the Bio-Rad CFX96 quantitative PCR system and SYBR Green qPCR Master Mix (MedChemExpress, NJ, USA) in conformity with the supplier’s regulations, with 5 min of predenaturation (95°C), 15 s of denaturation (95°C), and 30 s of annealing (60°C). U6 was the endogenous control of miR-26a. The 2 ^(−ΔΔCt) approach was harnessed to analyze the data. Each experiment was conducted 3 times. The sequences of specific primers are as follows:

OGD-treated BV2 microglia were seeded onto 6-well plates. Each group contained 4 replicate wells. The cell supernatant was harvested and centrifuged (1,000 rpm) for 10 min at 4°C subsequent to 48 h of culture. Then, the supernatant was collected. Additionally, the rat brain tissues on the injured side were extracted and weighed. Afterward, nine equal volumes of cold saline (incorporating the protease inhibitor) were added for homogenate. Next, the tissues were centrifuged for 20 min (3,000 rpm, 4°C), and the supernatant was retained. IL-1β, IL-6, and TNFα contents were tested by utilizing enzyme-linked immunosorbent assay (ELISA) kits (Invitrogen, USA).

TUNEL staining was used to detect HT22 neuron apoptosis. HT22 cells were inoculated in 6-well plates (2 × 10⁵/well) and transfected with NC and miR-26a mimics, followed by OGD stimulation. Cell apoptosis was monitored by the TUNEL staining cell apoptosis detection kit (cat. no. C1086, Beyotime, Shanghai, China). The nuclei were stained by DAPI solution (cat. no. C1005, Beyotime, Shanghai, China). The trial was performed 3 times.

Promega (Madison, WI, USA) built all luciferase reporting vectors (IGF-1-WT and IGF-1-MUT). Microglia (4.5 × 10⁴) were inoculated onto 48-well plates and cultivated to 70% confluence. Then, IGF-1-WT, IGF-1-MUT, and miR-186-5p mimics or negative controls were cotransfected into microglia with liposome 2000. Forty-eight hours after transfection, we tested luciferase activity. All tests were carried out in triplicate.

After the brain tissues and cells of MCAO rats were treated, we removed the medium. Then, the total proteins were isolated with protein lysates (Beyotime Biotechnology, Shanghai, China). Next, 50 μg of total protein was loaded onto a 12% polyacrylamide gel for 100 V electrophoresis for 2 h and transferred to polyvinylidene fluoride (Millipore, Bedford, MA, USA) membranes. At RT, 5% skim milk was used to block the membranes for an hour. TBST was used to wash them 3 times (10 min each time). Then, the membranes were incubated overnight (4°C) with anti-iNOS (1:1,000, ab178945, Abcam), anti-COX2 (1:1,000, ab179800, Abcam), anti-Bad (1:1,000, ab32445, Abcam), anti-Bax (1:1,000, ab32503, Abcam), anti-Caspase3 (1:1,000, ab13847, Abcam), anti-TREM1 (1:1,000, ab104413, Abcam), anti-TLR4 (1:1,000, ab13556, Abcam), anti-MyD88 (1:1,000, ab133739, Abcam), anti-NF-κB (1:1,000, ab207297, Abcam), anti-p-NF-κB (1:1,000, ab222494, Abcam), and anti-β-actin (1:1,000, ab115777, Abcam). Subsequently, the membranes were rinsed with TBST. The anti-rabbit secondary antibody labeled by horseradish peroxidase (HRP) (concentration: 1:300) was applied for a 1-h incubation at RT. TBST was used to wash the membranes 3 times (10 min/wash). Ultimately, Western blotting (WB) reagent (Invitrogen) was utilized for color imaging. ImageJ 1.44 software was employed for density detection.

SPSS 24.0 statistical software was used to analyze the data. Measurement statistics are represented as the mean ± standard deviation (x ± s). Student’s t test contrasted two groups, whereas paired t test was harnessed for the data comparison of the paired groups. The Tukey-Kramer test was used to conduct one-way ANOVA for comparing multiple groups. GraphPad Prism software (version 8.0) was used for drawing. If p < 0.05, statistical significance was identified.

The miR-26a level in mouse microglia cells (BV2) and hippocampal neuronal cell line HT22 cells was evaluated. Our data showed that BV2 cells had higher miR-26a level than that in HT22 cells (online suppl. Fig. 1A; for all online suppl. material, see https://doi.org/10.1159/000533813). Followed by OGD stimulation, miR-26a level in BV2 cells and HT22 cells was both downregulated (online suppl. Fig. 1B, C). Moreover, we analyzed miR-26a expression on Washington University mir-RNA database (https://mirna.wustl.edu/search/?name=miR-26a-5p#mrna_3). We found that miR-26a-5p has expression in all neurons, motor neurons, astrocytes, and microglia in the brainstem and spinal cord (online suppl. Fig. 2A, B). We also noticed that the miR-26a-5p level shows enhancement in astrocytes and microglia (compared with all neurons) (online suppl. Fig. 2A). To check the effect of miR-26a, miR-26a mimics were transfected into OGD-activated BV2 microglia. qRT-PCR verified that miR-26a expression was increased after transfecting miR-26a mimics compared with the OGD group (Fig. 1a). The profiles of IL-1β, IL-6, and TNFα (which are all proinflammatory cytokines) in BV2 cells were ascertained by ELISA. Consequently, IL-1β, IL-6, and TNFα levels were elevated in the OGD group vis-à-vis the control group, while their levels were hampered after transfecting miR-26a mimics (Fig. 1b, d). WB was used to assess the proinflammatory proteins iNOS and COX2 in BV2 cells. Interestingly, miR-26a mimics inhibited iNOS and COX2 levels, which were heightened in the OGD group. Their profiles were decreased after transfecting miR-26a mimics (Fig. 1e). The TREM1-TLR4/MyD88/NF-κB level was checked by employing WB. The levels of TREM1, TLR4, MyD88, and NF-κB phosphorylation were markedly suppressed by miR-26a mimics. Instead, OGD stimulation increased their profiles. Compared with the OGD group, miR-26a mimics attenuated the elevated TREM1-TLR4/MyD88/NF-κB level caused by OGD (Fig. 1f). All the above findings indicated that overexpressing miR-26a attenuated OGD-activated microglial inflammation.

To probe the impact of miR-26a on TREM1, we cotransfected miR-26a mimics and TREM1 overexpression plasmids into OGD-activated BV2 cells. qRT-PCR data suggested that TREM1 overexpression attenuated miR-26a level, which was further reduced in the OGD+TREM1 group. miR-26a mimics rescued miR-26a level (Fig. 2a). The profiles of inflammatory cytokines in each group were tested by ELISA. Consequently, IL-1β, IL-6, and TNFα levels were elevated in the TREM1 group compared to the control group. In addition, their profiles were distinctly increased after OGD induction on the basis of TREM1 treatment, while they were lowered after overexpressing miR-26a subsequent to the application of TREM1+OGD (Fig. 2b–d). iNOS and COX2 profiles in each group were assessed by WB. Notably, iNOS and COX2 levels in the TREM1 group were increased compared to those in the control group. Moreover, their levels were further enhanced in the TREM1+OGD group, while they were hampered after transfection with miR-26a mimics (Fig. 2e). WB was utilized to detect TREM1-TLR4/MyD88/NF-κB expression. As evidenced by the experiment, TREM1 elevated the TREM1-TLR4/MyD88 level and NF-κB activity (vs. the control group). Furthermore, OGD strengthened the function of TREM1, while miR-26a mimics attenuated the effects of TREM1+OGD (Fig. 2f). These findings revealed that TREM1 enhanced OGD-induced microglial inflammation, while overexpressing miR-26a attenuated this effect.

OGD was used to induce neurons, which were then transfected with miR-26a mimics to probe the effect of miR-26a. Given the outcomes of qRT-PCR, miR-26a was overexpressed in neurons following miR-26a mimic transfection and inhibited by OGD treatment (Fig. 3a). CCK8 assay showed that HT22 cell viability was not affected after miR-26a mimic transfection but markedly inhibited by OGD treatment. However, compared with the OGD group, miR-26a partly rescued cell viability (Fig. 3b). The LDH Cytotoxicity Detection Kit and TUNEL staining showed that compared to the control group, miR-26a mimics had no significant effects on neuronal cytotoxicity and apoptosis. However, LDH release and TUNEL-positive cell rate were both strengthened by OGD. miR-26a mimics dampened LDH release and TUNEL-positive cell rate (Fig. 3c–e). Then, the profiles of Bax, Caspase3, and Bad (which are all proapoptotic proteins) were monitored by WB. The statistics showed that, versus the control group, their expression was facilitated in the OGD group, while they were attenuated after miR-26a overexpression (Fig. 3f, g). The TREM1-TLR4/MyD88/NF-κB level in neurons was measured by WB. As a consequence, when contrasted to the control group, TREM1, TLR4, MyD88, and NF-κB profiles were enhanced in the OGD group, while their levels were reduced after transfecting miR-26a mimics (Fig. 3h). These phenomena demonstrated that overexpressing miR-26a hindered OGD-induced neuronal apoptosis and TREM1-TLR4/MyD88/NF-κB expression.

The above studies explored the effect of miR-26a on OGD-induced neurons, while the association between miR-26a and TREM1 remained unclear. Therefore, we transfected TREM1 overexpression plasmids into neurons and cotransfected OGD-induced neurons with TREM1 overexpression plasmids and miR-26a mimics. qRT-PCR data suggested that TREM1 overexpression attenuated miR-26a level, which was further reduced in the OGD+TREM1 group. miR-26a mimics rescued miR-26a level (Fig. 4a). Cell viability was reduced in the TREM1+OGD group, and enhanced after miR-26a overexpression (Fig. 4b). LDH release and TUNEL-positive cell rate in the TREM1 group were greater than that in the control group and were further upregulated in the TREM1+OGD group, while it was abated after miR-26a overexpression (Fig. 4c–e). The levels of Bax, Caspase3, Bad, and TREM1-TLR4/MyD88 were determined by utilizing WB. Consequently, Bax, Caspase3, Bad, TREM1-TLR4/MyD88 and NF-κB were upregulated in the TREM1 group (vis-à-vis the control group), and they were further upregulated in the TREM1+OGD group. In contrast, their levels were dampened by miR-26a mimics (Fig. 4f–h). The findings showed that overexpressing miR-26a repressed TREM1-mediated neuronal apoptosis.

StarBase suggested that TREM1 was an underlying target for miR-26a (Fig. 5a). Then, we carried out a dual-luciferase reporter assay to clarify whether miR-26a targeted the 3′-UTR of TREM1. The results showed that overexpressing miR-26a attenuated TREM1-3′-UTR-WT luciferase activity in BV2 cells but exerted no effect on that of mutant TREM1 (Fig. 5b). Moreover, we transfected a TREM1 overexpression plasmid into BV2 cells to explore the regulatory relationship between TREM1 and miR-26a. In addition, qRT-PCR was performed to examine the miR-26a level. As expected, miR-26a was downregulated in microglia transfected with the TREM1 overexpression plasmids compared with that of the vector group (Fig. 5c), illustrating that miR-26a negatively regulated TREM1. However, the molecular mechanism remains unclear. Therefore, we activated TLR4/MyD88/NF-κB with LPS, treated BV2 cells with Bay 11-7082, and conducted qRT-PCR to test miR-26a expression. Compared with the control group, the miR-26a profile was downregulated after TLR4/MyD88/NF-κB activation by LPS (Fig. 5d). However, it was upregulated after Bay 11-7082 administration (Fig. 5e). Finally, the profiles of TREM1, TLR4, MyD88, and NF-κB under different conditions were detected via WB. As a result, their levels were enhanced after transfection with TREM1 overexpression plasmids (by contrast to the control group) (Fig. 5f) and further enhanced after LPS treatment (Fig. 5g). In contrast, Bay 11-7082 administration hampered the above effects (Fig. 5h). These findings confirmed that miR-26a targeted TREM1 and regulated TREM1 through TREM1/TLR4/MyD88/NF-κB.

We constructed an MCAO rat model and injected agomir-26a into the ventricle 2 h before modeling to ascertain the influence of miR-26a on cerebral infarction in vivo. qRT-PCR data reflected that miR-26a was overexpressed following agomir-26a transfection, and agomir-26a rescued miR-26a reduction caused by MCAO (Fig. 6a). First, mNSS was adopted to verify the neural function of rats. Compared with the MCAO+Agomir-NC group, the mNSS of the rats decreased, and the latency time was reduced (Fig. 6b). The water content of the rat brain was gauged through the W/D method on day 3 subsequent to modeling. These outcomes suggested that the water content in the MCAO+Agomir-26a group was lower than that in the MCAO+Agomir-NC group (Fig. 6c). Histological changes in the rat brain were monitored by HE staining. As a result, the morphology of the majority of neurons was clear and showed no morphological abnormalities. Conversely, most neurons in the MCAO group showed shrinkage. However, these pathological abnormalities were improved after intracerebroventricular injection of agomir-26a (Fig. 6d). Morris water maze tests showed that the duration of the rats in the platform quadrant and the number of platform quadrant crossings increased in the MCAO+Agomir-26a group (Fig. 6e–g). These studies demonstrated that agomir-26a reduced brain edema and improved neurological severity in MCAO rats.

Next, we evaluated the influence of overexpressing miR-26a on neurons in MCAO rats and investigated the specific mechanism. First, NeuN staining and TUNEL staining were used to check the number of neurons. Interestingly, the rat neurons in the MCAO+Agomir-NC group had shriveled shapes, reduced NeuN expression, and enhanced TUNEL-positive cell rate. The neuronal structure was relatively intact after overexpressing miR-26a (Fig. 7a, b). Then, an immunofluorescence assay was utilized to determine the microglial phenotype in ischemic areas. The number of Ibal cells in the MCAO+Agomir-26a group was abated (against the MCAO+Agomir-NC group) (Fig. 7c, d). Furthermore, IL-1β, IL-6, and TNFα levels in ischemic brain tissues were verified by ELISA. The profiles of IL-1β, IL-6, and TNFα in the MCAO+Agomir-26a group were lower than those in the MCAO+Agomir-NC group (Fig. 7e–g). Finally, the TREM1-TLR4/MyD88/NF-κB level was detected by WB, and the results indicated that their profiles were lower in the MCAO+Agomir-26s group than in the MCAO+Agomir-NC group (Fig. 7h). These findings revealed that overexpressing miR-26a weakened neuronal apoptosis and inflammation in MCAO rats. 8c.

Cerebral infarction-induced release of dangerous/damage-associated molecular patterns is closely related to neuroinflammation, which may cause secondary damage, causing neuronal death and perennial glial activation, which might lead to poor prognosis and high disability rates in patients with cerebral infarction [28]. Microglia are considered the main factor causing inflammatory damage after cerebral infarction because they secrete multiple inflammatory cytokines and cytotoxic factors [29, 30]. This study intends to explore the role of miR-26a in inflammatory damage and alleviate cerebral infarction. Our experiments showed that the miR-26a profile was attenuated in MCAO rats and OGD-treated BV2 and HT22 cells. Overexpressing miR-26a improved motor function, cognitive and learning abilities, and cerebral edema in MCAO rats. Similarly, overexpressing miR-26a reduced inflammation and apoptosis in OGD-treated HT22 cells. Moreover, the role of miR-26a was partly mediated by TREM1, which further confirmed that miR-26a was an underlying therapeutic target for cerebral infarction.

In terms of the biological pathogenicity and pathology of ischemic stroke, miRNAs have become essential mediators of posttranscriptional gene silencing [31]. Many miRNAs were found altered following cerebral stroke. It has been reported that miR-634 [32], miR-9 [33], and miR-335 [34] regulate apoptosis in cerebral infarction. In terms of the functions of miRNAs, miR-3473b [35], miR-199a-5p [36], miR-210 [37], miR-579-3p [38], and miR-20b [39] modulate neuroinflammation in cerebral infarction. miR-497 [40], miR-21, and miR-24 [41] serve as biological indicators for the prognosis and diagnosis of patients with acute cerebral infarction. As a miRNA, miR-26a can be detected in the dentate gyrus of the hippocampus [42], mesenchymal stem cell-derived small extracellular vesicles [43], mouse dorsal root ganglia neurons [44], and normal human glial cells (HEB) and glioma cell lines (SHG-44, BT325, T98G, A172, and U251) [45]. In addition, overexpressing miR-26a inhibits proinflammatory cytokines by targeting HMGA2 in neuroinflammation, thus improving the neurological behavior and rotation test performance of mice with ICH [46]. However, the effect of miR-26a on cerebral infarction remains elusive. Here, miR-26a was knocked down in MCAO rats and OGD-treated BV2 and HT22 cells. Moreover, miR-26a overexpression ameliorated the neurological behaviors of MCAO rats and abated microglial inflammation and neuronal apoptosis. As supported by these outcomes, overexpressing miR-26a exerted an active impact on cerebral infarction treatment.

TREM1 is responsible for the activation of the innate immune response and mediates inflammation. Xu et al. [47] found that microglial TREM1 receptors induce poststroke neuroinflammatory damage by binding to SYK and promoting microglial apoptosis. Liang et al. [48] stated that inhibiting TREM1 prevents cerebral ischemia-induced neuronal injury and alleviates microglia-mediated neuroinflammation by reducing oxidative stress. Thus, TREM1 facilitates inflammatory factors in cerebral infarction, which is consistent with our study. Here, overexpressing TREM1 was found to aggravate microglial inflammation and neuronal apoptosis. Additionally, miR-26a targeted TREM1, and overexpressing miR-26a weakened the above effects of TREM1, suggesting that miR-26a exerted a positive role in cerebral infarction by repressing TREM1.

Several studies have confirmed that the TLR and NF-κB pathways are implicated in ischemic inflammation. For instance, upregulating miR-451 abates inflammatory cytokines in cerebral ischemia-reperfusion injury by reversely modulating TLR4/MyD88/NF-κB [49]. Diosgenin abates ischemic stroke-induced inflammation in rats by inhibiting TLR4/MyD88/NF-κB [50]. Furthermore, a few studies have demonstrated that inhibiting TLR4/MyD88/NF-κB protects microglia from OGD damage [51, 52]. Thus, attenuating TLR4/MyD88/NF-κB contributes to reducing inflammatory cytokines in cerebral ischemia. On the other hand, some miRNAs, such as miR-182-5p [53] and miR-27a-3p [54], have been shown to play an anti-inflammatory role in cerebral ischemia by attenuating TLR4/MyD88/NF-κB.

Additionally, some studies have shown that NF-κB reversely modulates miRNAs [55]. For example, ANXA1-NF-κB negatively regulates miR-26 to facilitate non-small cell lung cancer cell migration and invasion [56]. TNFα is a non-glycosylated cytokine protein with potent inflammatory effects by activating NF-κB pathway activation and subsequently dysregulates the expression of miRNAs. Therefore, TNFα gets involved in the pathophysiology of endometriotic cells [57]. In this study, TLR4/MyD88/NF-κB was activated by LPS, and NF-κB phosphorylation was abated by Bay 11-7082 to discover the expression of miR-26a and TREM1 in microglia. Interestingly, miR-26a was downregulated, while TREM1 was upregulated in BV2 cells after LPS induction. In contrast, Bay 11-7082 blocked NF-κB phosphorylation, elevated miR-26a expression, and downregulated TREM. These findings indicated that NF-κB negatively regulated miR-26a (Fig. 8).

To summarize, this research confirmed that the miR-26a profile was abated in the rat MCAO model and OGD-treated BV2 and HT22 cells. Upregulating miR-26a reduced neuroinflammation and neuronal apoptosis via TREM1-TLR4/MyD88/NF-κB in a cerebral ischemia model (Fig. 8). However, whether miR-26a can serve as a new therapeutic target for alleviating human cerebral ischemia needs more study.

This study protocol was reviewed and approved by the Ethics Committee of Bazhong Central Hospital (approval number: BZH2019-054).

The authors declare no conflicts of interest.

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Conceived and designed the experiments, performed the experiments, conducted statistical analysis, and wrote the manuscript: Daxiong Xu and Qi’an Guo. All authors read and approved the final manuscript.

All data generated or analyzed during this study are included in this article and its online supplementary material files. Further inquiries can be directed to the corresponding author.