Irak3 KO mice on C57BL/6 background were obtained from Jackson Laboratories (Bar Harbor, ME, United States, RRID:IMSR_JAX:007016) and WT littermates were generated through heterozygous mating. Mice were housed in a SPF environment with a 12 h light/dark cycle and a temperature of (22 ± 1°C) and were offered free access to food and water. Irak3 deficient mice developed severe osteoporosis (Li et al., 2005), increased inflammatory responses to bacterial infection (Kobayashi et al., 2002), promoted adverse remodeling and worse systolic dysfunction in myocardial infarction (Chen et al., 2012), and enhanced the development of type 1 diabetes in NOD mice (Tan et al., 2014). DNA of WT and KO mice was extracted from mouse tails with a genomic DNA kit (CoWin Biotech, Beijing, China). To confirm the genotype of mice, extracted DNA was amplified by DNA polymerase (Takara, Kusatsu, Japan) and then separated by agarose gel. Male WT and KO mice (age 7–10 weeks; weight 22–26 g) were used for the experiments. Randomization was performed by a third person unrelated to the study using a randomization table. Animal experimental protocols were approved by the Animal Care and Use Committee of the Nanfang Hospital, Southern Medical University (NFYY-2017-42) and followed the National Guidelines for Animal Experimentation. All efforts were made to minimize the number of animals used and their suffering.

Mouse tMCAO was produced by filament occlusion of the right MCA to induce transient focal brain ischemia-reperfusion. In the pre-experiment, we found that 1 h of tMCAO caused a very high mortality in Irak3 KO mice. Therefore, we chose 45 min tMCAO to conduct following experiments. In brief, mice were anesthetized with isoflurane (induction with 4%; maintenance with 1.5%) in oxygen-enriched air by face mask, and murine temperature was maintained at 37 ± 0.5°C throughout the experiment with heating pad. MCAO was performed by inserting a 7-0 nylon monofilament into the internal carotid artery via an external carotid artery stump and then by positioning the filament tip for occlusion at a distance of 8–9 mm beyond the internal carotid/pterygopalatine artery bifurcation. MCA was occluded for 45 min followed by reperfusion. Regional CBF (rCBF) was measured by the laser Doppler blood flow assessment (Moor Instruments, Wilmington, DE, United States) and presented as percentage of basal level (Figure 1A). The blunt needle probe of the laser Doppler blood flow assessment was positioned on skull surface in the MCA territory (1 mm posterior to the Bregma, and 5 mm lateral to the midline on the right parietal skull). Regional CBF was detected at four time points: before and after tMCAO; before and after reperfusion. Mice that did not show a >75% rCBF reduction or a <60% rCBF reperfusion over baseline levels or died after ischemia induction were excluded.

In order to study whether HIF-1α is associated with the expression of IRAK-M, WT mice were injected with 2-methoxyestradiol (2ME2) to suppress the expression of HIF-1α. Animals were randomly divided into different groups. The inhibitor of HIF-1α, 2-methoxyestradiol (2ME2, MedChemExpress, NJ, United States), was dissolved in dimethyl sulfoxide (DMSO) and further diluted in saline. 2ME2 solution was administered (5 mg/kg, i.p.) at 10 min before tMCAO. The WT mice in vehicle groups were intraperitoneally injected with same volume of DMSO diluted in saline.

At 24 h after reperfusion, the global neurological and motor function of the animals were assessed using a modified Bederson Score (Table 1) and the grip test (string test) (Denorme et al., 2016) by a researcher blinded to group allocation. For the grip test, a length of string (50 cm) was pulled tight between two vertical supports and elevated 40 cm from a flat surface. The mouse was placed midway on the string and rated as shown in Table 1.

After 45 min of MCAO and 24 h of reperfusion, brain tissues at 1.0–8.0 mm posterior to the frontal pole were sliced coronally at a thickness of 1 mm and stained with 1% TTC (Sigma-Aldrich, St Louis, MO, United States). Infarct volume was then measured with using ImageJ software (NIH, Bethesda, MD, United States) and presented as percentage of the total brain volume by an experimenter blinded to group allocation (Wang et al., 2016). Hemispheric brain swelling was quantified according to the brain sections with the equation: hemispheric brain swelling = (volume of ipsilateral hemisphere/volume of contralateral hemisphere – 1) × 100%.

To evaluate the integrity of the BBB, EB leakage and IgG immunohistochemistry were performed at 24 h after reperfusion. EB was injected to mice via femoral vein (2% in normal saline at 4 ml/kg). At 24 h after reperfusion, animals were anesthetized and perfused with PBS. Then, cerebral hemispheres were incubated in 10 ml of methanamide (Macklin, Shanghai, China) at 60°C for 24 h and centrifuged. The supernatant was taken for spectrophotometric quantification of extravasated EB dye at by spectrophotometer (BMG Labtech, Offenburg, Germany).

For water content assay brain tissues were weighed on an electronic analytical balance to get the wet weight, and were dried in an oven at 60°C for 7 days to obtain the dry weight. Brain tissue water content was measured by the weighing the brain at wet and dry state and was calculated as (wet weight – dry weight)/wet weight × 100%.

Brain tissues from the same region (2.0–4.0mm posterior to the frontal pole) were obtained with coronal cut, were divided into ipsilateral and contralateral hemispheres, and were homogenized in RIPA lysis buffer (Leagene, Beijing, China) containing protease and phosphatase inhibitors. After denatured in loading buffer, proteins were separated by SDS-PAGE (8–10% acrylamide) and then transferred to polyvinylidene difluoride membranes (Millipore, Billerica, MA, United States). After blocked by 5% non-fat milk, the membranes were incubated overnight at 4°C with primary antibodies including anti-COX-2 (1:1000; CST, Beverly, MA, United States, RRID:AB_2571729), anti-TNF-α (1:500; Santa Cruz; CA, United States, RRID:AB_1567355), anti-IL-10 (1:500; Santa Cruz, RRID:AB_10859554), anti-NLRP-3 (1:1000; NOVUS, CO, Cat# NBP2-12446), anti-iNOS (1:100; Santa Cruz, RRID:AB_627810), anti-p65 (1:1000; Proteintech, Chicago, IL, United States, RRID:AB_2178878), anti-β-actin (1:1000; Proteintech, Chicago, IL, United States, RRID:AB_2289225), and anti-TBP (1:500; Proteintech, RRID:AB_10951514) followed by incubation with anti-rabbit or anti-mouse horseradish peroxidase-conjugated secondary antibodies (1:5000; Santa Cruz). Bands were detected by chemiluminescence technology with using enhanced chemiluminescence advance Western blotting detection reagents (FDbio, Hangzhou, China), quantified and normalized to β-actin using ImageJ by an experimenter blinded to group allocation. The pictures of protein bands were opened by ImageJ and maximized to an appropriate zoom level. A frame was drawn around each band to make the band in the center of frame. The densities of protein blots were measured by program in ImageJ. The final relative quantification values are the ratios of net target bands to net loading controls.

Mice were perfused with saline and 4% paraformaldehyde at 24 h after reperfusion. Brains were post-fixed and subsequently immersed in 15% sucrose and 30% sucrose. Based on TTC stained and EB images, we selected the region at 4.0–5.0 mm posterior to the frontal pole. Coronal brain sections were obtained by slicing with Cryostat Microtome (Leica CM1800, Heidelberg, Germany). To detect the BBB permeability, immunohistochemistry was conducted with antibodies against mouse IgG (CoWin Biotech, Beijing, China) in accordance with the protocol suggested by the supplier (Lin et al., 2017). One picture of immunohistochemistry was opened by ImageJ. Region of interest (ROI) was selected by H-DAB in select model mode. Then the IgG relative intensity was measured. The antibody against Iba1 was used to detect microglia. Cryosections were permeabilized with Triton X-100, and were blocked with donkey serum. Sections were incubated with primary antibody against Iba1 (1:100; Abcam, Cambridge, United Kingdom, RRID:AB_1141557) overnight at 4°C and then with donkey anti-rabbit secondary antibody-conjugated to Alexa488 (1:500; Abcam) for 1 h at 37°C. Four serial sections of each brain sample were observed, three similar fields were selected in the brain region surrounded by three black boxes (Figure 8C) under an Olympus Fluorview laser scanning confocal microscope (Olympus, Tokyo, Japan). The number of Iba1-positive cells was counted by using ImageJ by an independent investigator who did not know the animal grouping (Li et al., 2018).

Brain tissues of the same region (2.0–4.0mm posterior to the frontal pole) were obtained with coronal cut and were divided into ipsilateral and contralateral hemispheres. Total RNA of ipsilateral and contralateral hemispheres was isolated by using Trizol (Takara), and 1 μg of total RNA was reverse-transcribed to cDNA using PrimeScript^TM RT Master Mix Kit (Takara). RT-qPCR was performed with Roche LightCycler480 System (Roche, Basel, Switzerland) by using SYBR Green master mixes (Takara). Quantification was performed by the delta cycle time method, with mouse β-actin used for normalization. The mouse-specific primers (Sangon Biotech, Shanghai, China) are listed in Table 2.

Statistical analyses were conducted by using SPSS 20.0 (IBM, Armonk, NY, United States) and GraphPad Prism 5.0 (GraphPad, La Jolla, CA, United States). During experiments and analysis, the investigators were blinded to genotype and experimental group. Data points outside the 95% confidence interval were excluded from the data analysis. Comparison between two groups was performed by using a two-tailed t-test. Differences within multiple groups were examined by two-way ANOVA followed by the Bonferroni post hoc correction. The incidence of hemorrhagic transformation in two groups was compared by using a Chi-square test. Data were expressed as mean ± SD. P < 0.05 was considered as statistically significant.

To explore the response of IRAK-M expression to cerebral ischemia-reperfusion, we firstly conducted mouse model of tMCAO with different reperfusion time followed by determination of IRAK-M mRNA expression with real-time fluorescence quantitative PCR. Regional CBF (rCBF) of each mouse was measured and shown in Figure 1A. After MCA was occluded, CBF decreased from 190.10 ± 17.02 to 28.03 ± 10.25 IU. When the nylon monofilament was removed, CBF returned to 163.70 ± 17.92 from 29.96 ± 8.33 IU. Results showed that IRAK-M mRNA was significantly upregulated in the infarcted brain tissue. Moreover, its expression peaked at 1h after reperfusion, 23.53 ± 13.27 folds higher than sham operative mice, and decreased to 18.79 ± 7.33, 5.70 ± 3.00 and 3.23 ± 1.70 folds at 8, 12, 24 h after reperfusion (Figure 1B), suggesting the time-dependent response of IRAK-M expression in ischemia-reperfused mice.

Hypoxia-inducible factor 1α is a key transcription factor induced by cerebral ischemia (Zhu et al., 2014) and is possibly involved in the IRAK-M expression (Shalova et al., 2015). To investigate whether HIF1α was responsible for inducing IRAK-M upregulation after reperfusion, we inhibited HIF1α mRNA expression before tMCAO onset. The results displayed that 2ME2 markedly suppressed the expression of HIF-1α and IRAK-M at 1 h after brain ischemia compared with the vehicle treated group (Figures 1C,D), indicating that the increase of HIF-1α correlates with the upregulation of IRAK-M after cerebrovascular obstruction and reperfusion.

Next, IRAK-M KO mice were used to observe the role IRAK-M in stroke mice. We assessed the neurological deficits and infarct volume of the mice via neurological function score and TTC staining, respectively, at 24 h after cerebral ischemia. Eleven mice were employed in both KO and WT groups to carefully observe the neurological outcome. Results showed that the deficiency of the IRAK-M led to a larger brain infarct volume (Figures 2A,D, P < 0.05 vs. WT) and deteriorated motor function deficit, including higher Bederson score and lower grip test score (Figures 2B,C, P < 0.05 vs. WT) compared with WT controls. We also observed hemispheric swelling on the basis of Figure 2A, and the data showed that the genetic deletion of IRAK-M significantly increased hemispheric swelling (Figure 2E, P < 0.05 vs. WT). These results demonstrate that IRAK-M deletion increases the infarct volume, and deteriorates the motor function of stroke mice.

Based on the increased swelling index and the observed severe edema (Figure 3A) in KO mice, we further examined the effect of IRAK-M KO on water content in mouse brain subjected to tMCAO. Data revealed that, after 24 h of reperfusion, brain water content from ischemic KO mice was remarkably more than that in WT littermates (78.98 ± 0.87 vs. 77.22 ± 0.59, P < 0.05, Figures 3A,B).

During the process of brain isolation, the obvious hemorrhagic foci were found in KO mice (Figure 3C). Therefore, the incidence of visible hemorrhagic transformation was calculated among all included stroke mice. Twelve of the 31 KO mice developed visible hemorrhagic transformation after cerebral ischemia. In contrast, only two out of the 25 WT mice had intracerebral hemorrhage, much less than that in KO mice. In other word, IRAK-M deficiency remarkably increased the incidence of hemorrhagic transformation after cerebral ischemia-reperfusion (37.05% vs. 8.70%, P < 0.05, Figure 3D).

More remarkable brain edema and visible hemorrhagic foci in IRAK-M KO mice indicate more severe BBB leakage. Thus, temporal changes in BBB integrity were assessed by EB extravasation and IgG leakage after focal ischemic insult. As expected, EB leakage assay showed that the dye content in ipsilateral ischemic hemispheres was obviously higher in KO group than that in WT group (P < 0.05, Figures 4A,B), whereas no difference was found in contralateral hemispheres.

Moreover, endogenous IgG immunohistochemistry showing its extravasation in the brain was conducted as another approach to examine the location of BBB disruption. As shown in Figures 4C,D, ischemia reperfusion markedly increased IgG staining both in cortex and hippocampus of the ipsilateral hemispheres. Consistently, IRAK-M deficiency strikingly increased relative intensity of IgG staining in cortex and hippocampus (Figures 4C–F).

We next studied the potential mechanisms underlying the deteriorated BBB disruption after the deletion of IRAK-M. According to the Figures 5A–D, mRNA expression levels of MMP-2 and MMP-9 were robustly elevated in the injured brain tissue of KO group compared with WT group at 4 h after brain ischemia. After 24 h reperfusion, MMP-9, but not MMP-2, mRNA level increased in the ischemic cortex of KO mice compared with those of WT mice. Together, genetic deletion of IRAK-M accentuated BBB disruption via the upregulated expression of MMP-2 and MMP-9 in the mouse model of ischemic stroke.

We also examined the expression of inflammatory cytokines in the ischemic hemisphere of the WT and IRAK-M KO mice. At the transcriptional level, the expression levels of IL-1β, TNF-α, iNOS, COX-2, IL-10, and IL-6 were dramatically elevated in IRAK-M KO mice at 4 and 24 h after reperfusion (Figures 6A–L). Moreover, the protein expressions of COX-2, TNF-α, NLRP-3, and iNOS were also upregulated at both time points (Figures 7A,B,D,E). However, the protein expression of IL-10, an anti-inflammatory mediator, was markedly decreased contrary to its mRNA expression (Figure 7C).

Those inflammatory mediators are either directly or indirectly triggered by NF-κB activation and inhibition of NF-κB activation is the critical step for the anti-inflammatory function of IRAK-M. We extracted proteins in nucleus to assess the localization of transcription factor P65 in the ischemic ipsilateral and contralateral hemisphere from different groups. Results in Figure 7F showed that the targeted deletion of IRAK-M induced more P65 nuclear translocation. In conclusion, these data suggest that, probably through inhibiting NF-κB signaling to alleviate inflammation, IRAK-M offers better neurovascular outcome in stroke mice.

Microglia is usually activated in the resident inflammatory environment and microglia leads to injury to BBB and exacerbates brain edema. To elucidate the effect of IRAK-M on microglia after stroke, we examined microglia activation in the cerebral cortex in the WT and Irak3 KO mice at 24 h after reperfusion. As illustrated in Figure 8, the number of microglia, visualized by the staining of Iba1 antibody that specially binds to microglia common antigen, was significantly elevated at 24 h after reperfusion in KO mice compared with WT littermates (66.00 ± 15.28 vs. 47.80 ± 8.17), suggesting the increased microglia also participate in the exaggeration of neurovascular injury under the background of IRAK-M deficiency.