Genetic deletion or pharmacological inhibition of soluble epoxide hydrolase reduces brain damage and attenuates neuroinflammation after intracerebral hemorrhage

All animal protocols were carried out according to the Guide for the Care and Use of Laboratory Animals published by the US National Institutes of Health (NIH publication no. 85-23, revised 1996) and were approved by the Animal Research Committee at Cheng Hsin General Hospital (animal permit numbers CHGH-102-01 and CHGH-103-09). Male wild-type (WT) C57BL/6 mice were obtained from BioLASCO (Taipei, Taiwan). Male sEH knockout (KO) mice in the C57BL/6 background were purchased from Jackson Laboratory (Bar Harbor, ME, USA). Animals were housed under conditions of controlled temperature (22–25 °C) and humidity (40–60%) with a 12-h/12-h dark cycle and were allowed free access to water and food.

Eight- to 10-week-old mice were randomized into different treatment groups using computer-generated random numbers. All outcome measurements and analyses were performed in a blinded manner. Sample sizes are determined by power analysis based on our pilot data and previous studies. A total of 409 mice (306 WT and 103 sEH KO) were used. Mice that had neurologic deficit scores greater than 15 or less than 3 at 3 h post-ICH were excluded from the study. Fifty-nine WT and 17 sEH KO mice were excluded due to neurologic deficit standard or death after ICH (WT: 59/260; sEH KO: 17/80). Sixty-five additional sham-operated control mice were used for biochemical assays and Evans blue quantification (44 WT and 21 sEH KO). Four normal mice were used for Western blot analysis (2 WT and 2 sEH KO). Four studies were conducted. The first study examined the specificity of the anti-sEH antibody and the temporal profile and cellular localization of sEH expression after ICH. These assessments included Western blot analysis (n = 5–6/group) and double immunofluorescence labeling. The second study investigated the anti-inflammatory and neuroprotective effects of sEH deletion. Experimental methods employed were as follows: (1) histology staining (n = 5–7/group), (2) cytokine enzyme-linked immunosorbent assays (ELISAs) and matrix metalloproteinase (MMP)-9 zymography (n = 6/group), (3) EET and 14,15-DHET ELISAs (n = 4/group), (4) Evans blue dye extravasation (n = 5–6/group), and (5) hemoglobin assay (n = 5/group). Cytokine ELISAs and MMP-9 analyses were performed on the same experimental group whereas other tests were performed on different experimental groups. The third study evaluated the anti-inflammatory and neuroprotective effects of sEH inhibition by AUDA, a selective sEH inhibitor, which has been widely used to evaluate the biological role of sEH [15, 18]. Experiments were as follows: (1) histology staining (n = 5–6/group), (2) Western blot analysis, cytokine ELISAs, and MMP-9 zymography (n = 6–7/group), (3) EET and 14,15-DHET ELISAs (n = 4/group), (4) Evans blue dye extravasation (n = 6/group), (5) hemoglobin assay (n = 5/group), and (6) behavioral and body weight assessments (n = 14/group). Western blot, cytokine ELISAs, and MMP-9 analyses were performed on the same experimental group.

ICH model was induced by collagenase injection as previously described [25]. Mice were intraperitoneally anesthetized with sodium pentobarbital (65 mg/kg) and injected with bacterial collagenase (0.0375 U of type VII-S in 1 μL of saline; Sigma-Aldrich, St. Louis, MO, USA) into the stratum using stereotactic coordinates: 0.8 mm anterior and 2.5 mm lateral to the bregma, 2.5 mm in depth, and at a rate of 0.1 μL/min over 10 min. The needle was left in place for an additional 20 min to prevent reflux. Sham-operated mice were injected with an equal volume of normal saline in the same manner. Saline instead of heat-inactivated collagenase was infused into sham-operated brains as there were no differences in protein levels of cleaved caspase-3 (cCP-3), Iba1, and IL-1β between the saline- and heat-inactivated collagenase-injected groups (Additional file 2: Figure S2).

AUDA (Cayman; 1 or 10 μM in 0.5 μL of 1% dimethyl sulfoxide, DMSO) or an equal volume of vehicle (1% DMSO) was given by intracerebroventricular (i.c.v.) injection 30 min before ICH as previously described [26]. Briefly, a 30-gauge needle of a Hamilton syringe was inserted into the lateral ventricle (stereotaxic coordinates: 0.5 mm posterior and 1 mm lateral to the bregma, 2 mm in depth). Then, AUDA or vehicle was infused with an infusion pump for 10 min at a rate of 0.05 μL/min. The needle was maintained in the infusion site for 20 min before removal to prevent reflux, and the ICH surgery was performed immediately thereafter.

Mice were sacrificed by transcardial perfusion following terminal anesthesia with sodium pentobarbital (80 mg/kg, i.p.) for histological examinations. Frozen sections (10 μm) were stained with cresyl violet, Fluoro-Jade B (FJB, a marker of degenerating neurons), and terminal deoxynucleotidyl transferase-mediated dUTP-biotin nick end labeling (TUNEL; In situ Cell Death Detection Kit, Roche Molecular Biochemicals) as previously described [5]. For immunostaining, sections were incubated with primary antibodies of rabbit anti-Iba1 (a microglia/macrophage marker; 1:1000; Wako Pure Chemical Industries, Osaka, Japan) or rabbit anti-myeloperoxidase (MPO, a neutrophil marker; 1:1000; Dako, Carpinteria, CA, USA) overnight [5]. Further colorimetric detection was carried out using diaminobenzidine as the peroxidase substrate. The specificity of the staining reaction was assessed by omission of the primary antibody and substitution of the primary antibody with non-immune rabbit serum.

To assess the cellular source of sEH, double immunofluorescence labeling was performed by simultaneous incubation of sections with rabbit anti-sEH (1:1000; Santa Cruz Biotechnology, Santa Cruz, CA, USA) overnight at 4 °C with mouse anti-CD11b (a microglia/macrophage marker; 1:100; Abcam, Cambridge, UK), rat anti-GFAP (an astrocyte marker; 1:200; Invitrogen, Camarillo, CA, USA), mouse anti-neuronal nuclei antigen (NeuN, a neuronal marker; 1:100, Millipore, Billerica, MA, USA), and rat anti-CD31 (an endothelial cell marker; 1:100; BD Biosciences, San Jose, CA, USA). To assess proinflammatory microglia/macrophages, sections were incubated overnight at 4 °C with rabbit anti-Iba1 (1:1000; Wako), together with rat anti-CD16/32 (a classic M1 activation marker; 1:100; BD Biosciences). Sections were washed, then incubated with Alexa Fluor 488- or Alexa Fluor 594-conjugated secondary antibodies (1:500; Molecular Probes, Eugene, OR, USA) for 2 h.

FJB, TUNEL, MPO, or Iba1 staining was quantified in three consecutive sections from the hemorrhagic core at the level of 0.24 mm from the bregma. The number of positive cells was counted in an area of 920 × 860 μm² in 10–12 non-overlapping fields immediately adjacent to the hematoma using a magnification of ×200 as previously described [5]. Iba1-positive resting microglia/macrophages were defined as resting if they contained relatively small cell bodies (< 7.5 μm in diameter) with long slender processes [27]. Microglia/macrophages were defined as activated when a cell body increased in size compared to resting microglia with short, thick processes and intense immunointensity. Activated microglia/macrophages were defined based on a combination of morphological criteria and a cell body diameter cutoff of 7.5 μm. FJB-, MPO-, and Iba1-positive cells were expressed as cells/field. Quantification of TUNEL staining was expressed as (TUNEL-stained nuclei/DAPI-stained nuclei) × 100%. Analyses were performed by two experimenters who were blinded to all animal groups. Inter-rater reliability was within 10%.

Injury volumes, striatum atrophy, and striatum enlargement ratios were quantified using cresyl violet-stained sections at 20 rostral-caudal levels that were spaced 200 μm apart as previously described [5]. Sections were analyzed using ImageJ software (version 1.50i; National Institutes of Health, Bethesda, MD, USA). The volume measurement was computed by summation of the areas multiplied by the interslice distance (200 μm). Striatum atrophy and striatum enlargement to account for brain edema were assessed using the following formula: ([ipsilateral striatum volume − contralateral striatum volume] / contralateral striatum volume) × 100%. Analyses were performed by two experimenters who were blinded to all animal groups. Inter-rater reliability was within 10%.

Colocalization of the microglia/macrophage marker (Iba1) and proinflammatory microglia/macrophage activation marker (CD16/32) was analyzed as previously described [28, 29]. Briefly, colocalization signals (in yellow) in merged images were obtained from the microglia/macrophage marker (in red) and the proinflammatory activation marker (in green) and turned into the gray pixel map by ImageJ software. The degree of colocalization was expressed in arbitrary units.

IL-1β, IL-6, macrophage inflammatory protein-2 (MIP-2), and monocyte chemoattractant protein-1 (MCP-1) (R&D Systems, Minneapolis, MN, USA), along with EETs (MyBiosource, San Diego, CA, USA), and 14,15-DHET (Detroit R&D Inc., Detroit, MI, USA), were measured in brain homogenates or cell lysates using commercially available ELISA kits.

Evans blue dye (2% in normal saline, 4 mL/kg) was injected into the tail vein and allowed to circulate for 1 h. The mice were then transcardially perfused with phosphate-buffered saline (PBS), and the ipsilateral hemispheric samples were homogenized in 1 mL of 60% trichloroacetic acid by sonication. After centrifugation at 4500 rpm for 15 min at 4 °C, the supernatants were diluted with ethanol (1:4). The absorbance of Evans blue in the supernatant was measured at 620 nm. Evans blue concentrations were calculated and expressed as μg/g brain tissue using a standard curve.

Zymography was performed as previously described [25]. Briefly, equal amounts of protein were loaded and separated on a 10% Tris-glycine gel with 0.1% gelatin as the substrate. Then, the gel was washed and renatured with 2.5% Triton X-100 buffer. After incubation with developing buffer at 37 °C for 24 h, the gel was stained with 0.05% Coomassie R-250 dye (Sigma-Aldrich) for 30 min and destained. Gelatinolytic activity (MMP-9: ~ 97 kDa; MMP-2: ~ 72 kDa) was determined as clear bands at the appropriate molecular weights.

The hemoglobin contents of ICH brains were quantified using a spectrophotometric assay according to previously described methods [5]. Distilled water (300 μL) was added to the hemorrhagic hemisphere, followed by homogenization for 30 s and sonication on ice for 1 min. After centrifugation at 13,000 rpm for 30 min, 20 μL of supernatant was reacted with Drabkin reagent (80 μL; Sigma-Aldrich) for 15 min. Optical density was then measured at a wavelength of 545 nm to assess the concentration of cyanmethemoglobin. The hemorrhage volume can then be calculated from the linear relation between the optical density and hemoglobin concentration.

Western blot analyses were performed as previously described [30]. Protein samples obtained from tissue or BV2 microglia homogenates were separated on 8–12% sodium dodecyl sulfate-polyacrylamide gels, transferred to Immobilon-P membranes (Millipore), and probed overnight at 4 °C with primary antibodies including rabbit anti-sEH (1:1000) and rabbit anti-P65 (1:1000) from Santa Cruz; rabbit anti-iNOS (1:1000) and rabbit anti-cyclooxygenase (COX-2 1:1000) from Cayman; rabbit anti-cCP-3 (1:1000), rabbit anti-p-P38 (1:1000), rabbit anti-P38 (1:2000), rabbit anti-p-C-Jun N-terminal kinase (JNK, Thr183/Tyr185, 1:1000), rabbit anti-JNK (1:2000), rabbit anti-p-extracellular signal-regulated kinase p44/42 (ERK p44/42; Thr202/Tyr204, 1:1000), rabbit anti-ERK (1:2000), and rabbit anti-p-P65(1:1000) from Cell Signaling (Danvers, MA, USA); and mouse anti-β-actin (1:10,000, Sigma-Aldrich). Protein band intensities were quantified using ImageJ software and were normalized to the corresponding β-actin intensity.

All the behavioral tests were performed before ICH and at 1, 4, 7, 14, 21, and 28 days after ICH. Mice were pretrained for both rotarod and beam walking tests for 3 days.

Data are presented as the mean and standard error of the mean (mean ± S.E.M.). One-way or two-way analysis of variance (ANOVA) followed by a post hoc Bonferroni test was used for multiple groups to determine significant differences. Student’s t test was used to test the difference between two groups. Statistical significance was set at P < 0.05.

We first examined whether sEH protein expression was altered in brain tissues following collagenase-induced ICH. Western blot results showed that sEH was expressed in sham-operated brains and the liver tissue (as a positive control) of WT mice, but absent in sEH KO brains (Fig. 1a). The sEH protein level in the hemorrhagic hemisphere was significantly increased compared with the sham control by 3 h (P < 0.001) following ICH and remained elevated at 6 h (P < 0.001), 12 h (P < 0.001), 1 day (P = 0.003), and 4 days (P = 0.026, Fig. 1b). The sEH protein level was decreased at 7 days. We next defined the cell types that expressed sEH at 1 day post-ICH. Dual-label immunofluorescence demonstrated that sEH was expressed in microglia/macrophages, astrocytes, neurons, and endothelial cells both in the perihematomal region of the striatum and in the contralateral hemisphere (Fig. 1c).

Since we demonstrated that sEH was upregulated in the hemorrhagic hemisphere following ICH and was colocalized with microglia/macrophages, we hypothesized that sEH inhibition reduces neuroinflammatory responses after ICH. To test this hypothesis, we first used a loss-of-function strategy to evaluate the effect of sEH deletion on microglia/macrophage activation at 1 and 4 days post-ICH. These two time points were chosen because the inflammatory-related signals peak within 4 days after collagenase-induced ICH and decline thereafter [5, 6, 34]. At 1 and 4 days post-ICH, activated Iba1-positive microglia/macrophages accumulated at the hematoma border; however, deletion of sEH significantly reduced the number of activated Iba1-positive microglia/macrophages at 1 day (P = 0.0038, Fig. 2b) and 4 days (P = 0.0049, Fig. 2b). Deletion of sEH also significantly reduced the expression of proinflammatory mediators IL-1β, IL-6, MIP-2, and MCP-1 at 1 day (IL-1β, IL-6, and MIP-2: P < 0.001; MCP-1: P = 0.001; Fig. 2c–f) and 4 days (IL-1β, P = 0.008; IL-6: P = 0.002; MIP-2: P < 0.001; MCP-1: P = 0.038; Fig. 2c–f). As for the direct metabolic consequences of sEH deficiency, ICH caused reductions in EET levels and the EET/14,15-DHET ratio and an increase in 14,15-DHET levels in both WT and sEH KO mice at both 1 and 4 days post-ICH (all P < 0.05; Fig. 2g). Deletion of sEH caused a significant elevation of EET levels and the EET/14,15-DHET ratio and a decrease of 14,15-DHET levels at both time points tested post-ICH (all P < 0.01; Fig. 2g). Taken together, these data demonstrate that sEH deficiency mitigates the proinflammatory response after ICH.

We next evaluated the effect of sEH deletion on BBB integrity, as BBB disruption is a major consequence of post-ICH inflammation and contributes to brain edema and subsequent neuronal damage [35]. The Evans blue content in the hemorrhagic hemispheres of the WT group was significantly attenuated by sEH deletion at both 1 and 4 days (both P < 0.001; Fig. 3a). To explore the possible mechanisms underlying the protective effects of BBB integrity, we also evaluated MMP-9 activity, as it disrupts the BBB by degrading the tight junction proteins and basal lamina proteins [35]. MMP-9 activity in hemorrhagic hemispheres was significantly increased in WT and sEH KO mice at both 1 and 4 days post-ICH (all P < 0.05; Fig. 3b), with only a slight enhancement in the sham-injured brains. Deletion of sEH significantly suppressed the induction of MMP-9 activity at both 1 day (P = 0.009; Fig. 3b) and 4 days post-ICH (P < 0.001; Fig. 3b). Additionally, ICH induces a robust infiltration of neutrophils, which exacerbate BBB breakdown and brain damage by release of cytotoxic mediators or by increasing inflammatory cell recruitment. Deletion of sEH significantly reduced neutrophil accumulation compared with WT mice in the perihematomal area at both 1 day (P = 0.0298; Fig. 3c) and 4 days (P = 0.0167). These data indicate that sEH deficiency sealed the barrier and reduced MMP-9 activity and neutrophil accumulation.

To determine whether the observed changes in inflammatory responses and BBB breakdown were reflected in a reduction of neuronal death and apoptosis, histological outcomes were evaluated. The number of FJB-positive degenerative neurons around the hematoma were significantly reduced in sEH KO brains compared to WT brains at both 1 day (P = 0.0024; Fig. 3d) and 4 days (P = 0.0012; Fig. 3d) post-ICH. Deletion of sEH also significantly attenuated the percentage of TUNEL-positive cells compared with WT mice at both 1 day (P = 0.0359; Fig. 3e) and 4 days (P = 0.0214; Fig. 3e). However, no significant between-group differences were found in brain hemoglobin content, an indicator of hemorrhage size at either 1 or 4 days post-ICH (Fig. 3f). This indicates that the protective mechanisms of sEH deficiency against ICH were independent of hemorrhage volumes.

To explore whether the pharmacological inhibition of sEH also suppressed ICH-mediated neuroinflammation, we treated WT mice with i.c.v. injection of the sEH inhibitor AUDA (1 and 10 μM) 30 min before collagenase-induced ICH (Fig. 4a). We chose the route of i.c.v. administration to focus on the cerebral sEH inhibition effects; as sEH is found in many other organs, systemic administration of the inhibitor could have caused unexpected physiologic results in non-cerebral and vascular systems [36]. We evaluated the effect of AUDA treatment on inflammatory responses at day 1 because both microglia/macrophage activation and proinflammatory chemokine/cytokine expression peaked at this time point and declined at day 4, as shown in Fig. 2. The number of activated Iba1⁺ microglia/macrophages around the hematoma was significantly reduced following 10 μM AUDA treatment (P = 0.007, Fig. 4b) but not 1 μM treatment (P > 0.05, Fig. 4b). Levels of proinflammatory cytokines IL-1β, IL-6, and MIP-2 were significantly attenuated in the 10 μM AUDA-treated hemorrhagic brains compared with the vehicle group at 1 day post-ICH (IL-1β: P = 0.003; IL-6: P = 0.003; MIP-2: P = 0.008; Additional file 3: Figure S3A). However, only the IL-6 level was significantly reduced following 1 μM AUDA treatment (P = 0.035; Additional file 3: Figure S3A). Since 10 μM AUDA provided enhanced protection against microglial activation, the dosage of 10 μM was employed in subsequent studies. We next investigated whether AUDA affected the activation of proinflammatory microglia/macrophages, using double-immunofluorescent staining with the microglia/macrophage marker Iba1 and with the marker of proinflammatory microglia/macrophages (CD16/32). While no CD16/32-positive microglia/macrophages were observed in sham-operated brains, ICH induced an increase of CD16/32-positive microglia/macrophages at the hematoma border at 1 day post-ICH (Fig. 4c). The CD16/32-Iba1 double-positive signal around the hematoma was significantly decreased following 10 μM AUDA treatment (P = 0.0353; Fig. 4c). Moreover, 10 μM AUDA treatment resulted in a significantly elevated level of EET expression (P = 0.001), decreased EET/14,15-DHET ratio (P = 0.009), and reduced 14,15-DHET level (P = 0.006; Additional file 3: Figure S3B). These findings indicate that pharmacological inhibition of sEH using AUDA resulted in a reduction of EET degradation, accompanied by inhibition of proinflammatory microglia/macrophage activation following ICH.

We next evaluate whether sEH inhibition by AUDA attenuated neurodegeneration and apoptosis at the acute stage. Treatment with 10 μM AUDA significantly reduced the number of FJB-positive degenerative neurons around the hematoma and levels of cCP-3, a critical effector in apoptosis, compared with vehicle-treated mice at 1 day post-ICH (both P < 0.001; Fig. 4d, e). Also, AUDA administration resulted in significantly reduced neutrophil accumulation (P = 0.0030; Fig. 4f) and MMP-9 activity (P < 0.001; Fig. 4g) and attenuated levels of extracted Evans Blue, an indicator of BBB breakdown (P = 0.02; Fig. 4h), at 1 day post-ICH. However, no significant between-group differences were found in brain hemoglobin content at 1 day (Fig. 4i).

We further determined whether the suppression of neuroinflammation with AUDA resulted in a reduction in brain tissue damage. Treatment with 10 μM AUDA significantly reduced hemorrhagic injury volumes to 44% of the vehicle group (P = 0.0059; Fig. 5a) at 1 day. Striatal enlargement, an indicator of brain edema, was also significantly smaller in 10 μM AUDA-treated mice than in vehicle-treated mice (P < 0.001; Fig. 5a) at 1 day. At the chronic stage of ICH, 10 μM AUDA significantly reduced the degree of striatal atrophy compared with the vehicle group (P = 0.0012; Fig. 5a) at 28 days. We next assessed the long-term consequences of sEH inhibition on neurological recovery. At 3 h after injury, there was no difference in global neurological deficit as evaluated by mNSS between vehicle-treated and 10 μM AUDA-treated groups, indicating that injury severity was initially similar regardless of treatment (Fig. 5b). Significant improvement in neurological function was observed from 1 to 28 days in the AUDA-treated group compared with the vehicle group (all P < 0.05; Fig. 5b). Treatment with AUDA also significantly increased the rotarod running time from 1 to 28 days post-ICH compared with vehicle-treated mice (all P < 0.05; Fig. 5c). Similarly, AUDA-treated mice exhibited better beam walk performances with significantly reduced latency to cross the beam from 1 to 7 days post-ICH (P < 0.001 on day 1; P < 0.05 on days 4 and 7; Fig. 5d). The hindlimb scores were also significantly higher for AUDA-treated mice than for vehicle-treated mice at 1, 7, 14, and 21 days (all P < 0.05; Fig. 5e). Furthermore, the biased swing percentage was decreased in AUDA-treated mice from 4 to 28 days post-ICH compared with those that received vehicle only (all P < 0.05; Fig. 5f). However, there was no significant difference in body weight changes between the two groups during the 28-day observation period (all P > 0.05; Fig. 5g). Thus, these data demonstrate that when sEH was targeted using the selective inhibitor AUDA, EET degradation was reduced. This resulted in a reduction in proinflammatory microglia/macrophage activation and attenuation of ICH-induced neurological deficits and brain damage.

The in vivo results demonstrated that both genetic deletion and pharmacological inhibition of sEH preserved neuronal function and reduced microglia/macrophage activation after ICH. To directly measure the effects of sEH inhibition on microglia, we next determined whether sEH modulated microglia-mediated neuroinflammation in vitro. We used thrombin, a serine protease activated during coagulation that is present at high levels during ICH, and hemin, a cytotoxic breakdown product of hemoglobin following ICH, to activate microglia. We first assessed whether the lack of sEH would affect microglial activation by comparing thrombin- and hemin-induced NO production using primary microglia derived from WT and sEH KO mice. Deletion of sEH significantly reduced 10 U/mL thrombin-induced NO production to 54% and 10 μM hemin-induced NO production to 65% of that observed for the WT group (both P < 0.001; Fig. 6a) at 24 h post-stimulation, suggesting that sEH deletion suppressed thrombin- and hemin-stimulated microglial activation. Next, we evaluated the effects of pharmacological inhibition by AUDA on microglial activation by using BV2 microglial cells. There was a significant increase in NO release in the culture supernatant 24 h after exposure to thrombin in BV2 microglial cells compared with unstimulated cells, but co-treatment with 1, 5, 10, or 50 μM AUDA for 24 h significantly reduced the thrombin-induced NO production to 77, 71, 63, and 69% of that observed for the vehicle control group, respectively, and 10 μM AUDA provided the highest degree of anti-inflammatory protection (all P < 0.001; Fig. 6b). Therefore, a dosage of 10 μM was employed for subsequent biochemical studies. We further used primary mouse microglia to confirm the results obtained from BV2 microglial cells. Like the results in BV2 cells, co-treatment with 10 μM AUDA for 24 h significantly reduced thrombin-induced NO production to 52% and hemin-induced NO production to 51% of that observed for the vehicle control group (both P < 0.001; Fig. 6c). To confirm whether the protective effect of the sEH inhibition was attributed to a decrease in EET metabolism, thrombin- or hemin-stimulated primary microglia were treated with 10 μM AUDA in the presence or absence of the putative pan-EET receptor antagonist 14,15-EEZE (1 μM). Antagonist 14,15-EEZE did not affect the baseline NO level, but the protective effect of AUDA on microglial activation was completely obliterated by administration of 14,15-EEZE in both thrombin- and hemin-stimulated primary microglia (Fig. 6c), indicating that the beneficial effect observed with sEH inhibition was specifically due to EETs. Treatment with 10 μM AUDA also significantly attenuated thrombin- and hemin-induced production of IL-1β, IL-6, and MIP-2 (all P < 0.01, Fig. 6d, e) in the culture medium of primary microglia. We further assessed the effect of AUDA on the two inducible enzymes, iNOS and COX2, that are expressed in activated microglia [37]. Treatment with 10 μM AUDA significantly attenuated iNOS protein expression in thrombin-stimulated BV2 cells at 6 h (50% of vehicle level, P = 0.015) and 24 h (58% of vehicle level, P = 0.003) and in hemin-stimulated BV2 cells at 6 h (48% of vehicle level, P = 0.011) and 24 h (64% of vehicle level, P = 0.01), respectively (Fig. 6f). Similarly, protein levels of COX2 were significantly reduced following AUDA treatment in thrombin-stimulated BV2 cells at 6 h (56% of vehicle level, P = 0.004) and 24 h (45% of vehicle level, P = 0.029) and in hemin-stimulated BV2 cells at 6 h (53% of vehicle level, P = 0.023) and 24 h (45% of vehicle level, P = 0.001), respectively (Fig. 6g).

We next investigated the effects of AUDA on the activation of the NF-κB, which is an inducible transcription factor that plays a crucial role in inflammatory and immune responses and can be activated by mitogen-activated protein kinases (MAPKs), a family of serine/threonine protein kinases that mediates microglial activation [38]. Co-treatment with 10 μM AUDA significantly attenuated the increased levels of pP65 Ser536, an indicator of NF-κB activation, induced by thrombin (52% of vehicle level, P = 0.011) and hemin (65% of vehicle level, P = 0.016; both in Fig. 7a). We further studied the possible effects of AUDA on the phosphorylation of P38, JNK, and ERK, the major MAPK pathway subfamilies. Co-treatment with 10 μM AUDA significantly reduced phosphorylation of P38 induced by thrombin (49% of vehicle level, P = 0.007) and hemin (57% of vehicle level, P = 0.041) without affecting the level of total P38 (Fig. 7b). However, JNK and ERK phosphorylation was not affected by AUDA treatment (all P > 0.05; Fig. 7c, d). Together, these results suggest that AUDA inhibits thrombin- and hemin-induced microglial activation and limits the production of inflammatory mediators by suppressing the P38 MAPK and NF-κB pathways.

Since our in vivo results showed that deletion and inhibition of sEH reduced microglial/macrophage activation and attenuated neuronal damage, we further hypothesized that inhibition of sEH reduced microglia-derived neurotoxic factors and thus ameliorated neuronal injury. We first assessed the direct effect of thrombin or hemin on cell viability in N2A cultures. While treatment with 10 U/mL thrombin had no direct effect on neuronal viability assessed by the MTT assay, 10 μM hemin significantly reduced viability of N2A cells to 69.2 ± 1.0% of the control level (P < 0.001, Additional file 4: Figure S4). Co-treatment with 10 μM AUDA significantly increased neuronal viability to 75.0 ± 1.0% compared with the vehicle control group (P = 0.002, Additional file 4: Figure S4). In order to mimic clinical settings, we next evaluated the effect of AUDA-treated BV2 microglia-conditioned medium on N2A cell survival in vitro. Microglia cell cultures were treated with 0.05% DMSO (control), 10 μM AUDA, thrombin or hemin, thrombin/hemin and 10 μM AUDA for 24 h (Fig. 7e). Treatment of N2A cells with conditioned medium harvested from the thrombin- or hemin-stimulated BV2 microglia at 48 h markedly decreased cell viability of N2A cells to 69.2 ± 0.6% (P < 0.001) or 58.6 ± 0.6% (P < 0.001) of the control level, respectively (Fig. 7f). When compared with N2A cells treated with conditioned media from thrombin- or hemin-stimulated BV2 microglia, cell viabilities of those treated with 10 μM AUDA were increased significantly to 79.7 ± 0.5% (P < 0.001) and 72.1 ± 0.9% (P < 0.001), respectively. These results show that the neurotoxic effects of condition media from the thrombin- or hemin-stimulated BV2 microglia were reduced by AUDA treatment, indicating that the protective effect of AUDA on neuronal viability is at least partly conferred through the inhibition of microglial-induced injury.