Cerebroprotection by the neuronal PGE₂ receptor EP2 after intracerebral hemorrhage in middle-aged mice

Abstract

Inflammatory responses mediated by prostaglandins such as PGE₂ may contribute to secondary brain injury after intracerebral hemorrhage (ICH). However, the cell-specific signaling by PGE₂ receptor EP2 differs depending on whether the neuropathic insult is acute or chronic. Using genetic and pharmacologic approaches, we investigated the role of EP2 receptor in two mouse models of ICH induced by intrastriatal injection of collagenase or autologous arterial whole blood. We used middle-aged male mice to enhance the clinical relevance of the study. EP2 receptor was expressed in neurons but not in astrocytes or microglia after collagenase-induced ICH. Brain injury after collagenase-induced ICH was associated with enhanced cellular and molecular inflammatory responses, oxidative stress, and matrix metalloproteinase (MMP)-2/9 activity. EP2 receptor deletion exacerbated brain injury, brain swelling/edema, neuronal death, and neurobehavioral deficits, whereas EP2 receptor activation by the highly selective agonist AE1-259-01 reversed these outcomes. EP2 receptor deletion also exacerbated brain edema and neurologic deficits in the blood ICH model. These findings support the premise that neuronal EP2 receptor activation by PGE₂ protects brain against ICH injury in middle-aged mice through its anti-inflammatory and anti-oxidant effects and anti-MMP-2/9 activity. PGE₂/EP2 signaling warrants further investigation for potential use in ICH treatment.

Introduction

Intracerebral hemorrhage (ICH) causes high morbidity and mortality, but effective therapies are lacking. Inflammatory responses, including those mediated by prostaglandin (PG) E₂, contribute to ICH-induced secondary brain injury.¹ PGE₂ is a major proinflammatory prostaglandin in the brain. Its synthesis is catalyzed by cyclooxygenases and PGE₂ synthases, the expression of which is increased after ICH.^2,3 As a result, PGE₂ accumulates predominately in the perihematomal region after ICH.⁴ Notably, celecoxib, a selective inhibitor of cyclooxygenase-2, is able to reduce ICH-induced brain injury and improve functional outcomes.⁴ A recent clinical trial showed that patients treated with celecoxib in the acute stage of ICH had a smaller expansion of perihematomal edema than did controls.⁵

PGE₂ exerts its biologic function through four divergent G-protein-coupled receptor subtypes known as EP1–EP4. We and others have shown that EP1 and EP3 receptor signaling has toxic effects in in vitro and in vivo models of ICH^6,7 and in thrombin-induced brain injury.⁸ In contrast, EP2 receptor signaling protects cultured neurons under various conditions through the cAMP/PKA or cAMP/Epac pathway^9–12 and regulates microglial activation and function in vitro.^13–16 These findings reflect cell-specific differences in EP2 signaling. In vivo work has shown that the EP2 receptor has beneficial effects in models of ischemic stroke^12,17,18 but detrimental effects in models of Alzheimer’s disease,^19,20 Parkinson’s disease,²¹ and amyotrophic lateral sclerosis.²² Thus, the EP2 receptor may function differently under acute and chronic neuropathologic conditions, and cell-specific EP2 signaling may explain this dichotomy of action. To our knowledge, the role of EP2 in ICH pathology has not been clearly defined.

Using established collagenase and blood ICH models, we recently tested the efficacy of misoprostol, a relatively weak EP2/EP4 agonist, and found that it confers cerebroprotective effects via mechanisms that may involve the high-mobility group box 1 (HMGB1), Src kinase, and matrix metalloproteinase (MMP)-2/9 pathways.²³ In the present study, we used genetic and pharmacologic approaches to selectively target the EP2 receptor in order to determine its contribution to ICH pathology in these same two mouse models. We used middle-aged mice to enhance the clinical relevance of the study, as ICH occurs more frequently in the middle-aged population than in younger individuals. We identified marked cerebroprotective effects of neuronal EP2 signaling in the ICH models.

Materials and methods

Animals

All animal studies were conducted in accordance with National Institutes of Health guidelines and were approved by the Johns Hopkins University Animal Care and Use Committee. EP2 knockout (^−/−) mice on a C57BL/6 background were maintained in the Johns Hopkins animal facility and were used at 10–12 months of age (25–35 g). Age-matched C57BL/6 wild-type mice were obtained from Charles River Laboratories (Frederick, MD, USA). Cerebrovascular effects of aging are well developed in middle-aged mice.²⁴ Only male mice were used in this study. All efforts were made to minimize the numbers of animals used and ensure minimal suffering. Animal experiments were reported in accordance with the ARRIVE guidelines.

ICH models

The procedures for modeling ICH by intrastriatal injection of collagenase VII-S (sterile-filtered, 0.075 U in 0.5 µL sterile saline, Sigma, St. Louis, MO, USA) and autologous whole blood without anticoagulant (10 µL collected from the central tail artery) were adapted from our established protocols.^7,23,25,26 Mice positioned on a stereotaxic frame (Stoelting Co., Wood Dale, IL, USA) were anesthetized by isoflurane (3.0% for induction and 1.0% for maintenance) and ventilated with oxygen-enriched air (20%:80%) via a nose cone. We injected collagenase into the left striatum of mice at the following stereotactic coordinates: 0.8 mm anterior and 2.0 mm lateral of the bregma, and 2.9 mm in depth. For the blood model, we infused 4 µL of blood over 20 min at the first site 2.9 mm below the surface of the brain, advanced the needle 0.8 mm ventrally, paused for 6 min, and then infused the remaining 6 µL of blood over 30 min. We withdrew the needle slowly (at a rate of 1 mm/min) 10 minutes after the injection of collagenase or blood to minimize backflow of the infused substance along the needle track. In the two ICH models, the burr hole was sealed with bone wax, and mice were allowed to recover under observation. Rectal temperature of the animals was maintained at 37.0 ± 0.5℃ with a heating pad throughout the experimental and recovery periods.

Experimental groups

In this study, two sets of experiments were performed. In the first set, EP2^−/− and C57BL/6 wild-type mice (n = 46/group) were subjected to one of the two ICH models. Except where stated otherwise, ICH was induced by collagenase in this study. One subgroup of C57BL/6 mice (n = 3) was used to define the cell type that expresses EP2 receptor in the ICH brain. To corroborate results from the EP2^−/− mice, in the second set of experiments, we subjected C57BL/6 mice (n = 47/group) to collagenase-induced ICH and randomly assigned them to receive highly selective EP2 receptor agonist ONO-AE1-259-01 (0.2 µL, 2.0 nM; ki values: 3 nM for EP2 and>10 µM for the other EP receptors; ONO Pharmaceutical Co. Ltd., Tokyo, Japan)²⁷ or vehicle (saline). AE1-259-01 or vehicle was injected into the left striatum 10 min before collagenase injection at the same stereotaxic coordinates (0.8 mm anterior, 2.9 mm ventral, and 2.0 mm lateral to bregma). The selectivity of AE1-259-01 for EP2 receptor has been well established.^27–29 We chose dosing and treatment regimens for AE1-259-01 based on previous work in ischemic stroke models¹⁷ and our preliminary studies. For randomized allocation of animals, we used computer-generated random numbers. Sham control mice (n = 5/group) were subjected to needle insertion only. Endpoints were assessed by investigators blinded to experimental groups.

Brain section preparation and immunofluorescence

On day 3 after ICH, mice were anesthetized with isoflurane, euthanized, and perfused transcardially with phosphate-buffered saline (PBS) followed by 4% paraformaldehyde. Brains were dissected, post-fixed by 4% paraformaldehyde overnight at 4℃, and then transferred to 30% sucrose in PBS. Brains were cut into 20-µm coronal cryosections for use in immunofluorescence based on our established protocol.² Sections were coded and labeled in a manner that protected the blinding. Primary antibodies used were rabbit anti-EP2 receptor (1:200, Cayman Chemical, Ann Arbor, MI, USA); mouse anti-NeuN (neuron marker; 1:1000; Chemicon, Temecula, CA, USA); rat anti-glial fibrillary acidic protein (GFAP, astrocyte marker; 1:500; Life Technologies, Grand Island, NY, USA); rat anti-CD11b (microglia/macrophage marker; 1:1000; Serotec, Raleigh, NC, USA); rabbit anti-Iba 1 (microglia/macrophage marker; 1:1000; Wako Chemicals, Richmond, VA, USA); mouse anti-GFAP (1:500, Sigma); and rabbit anti-myeloperoxidase (MPO, neutrophil marker; 1:500; Dako, Carpinteria, CA, USA). Free-floating sections were then incubated with Alexa 488-conjugated goat anti-rabbit secondary antibody (1:1000; Molecular Probes, Eugene, OR) and/or Cy3-conjugated secondary antibody (1:1000; Jackson Labs, West Grove, PA, USA) for 90 min. Stained brain sections were examined with a fluorescence microscope (ECLIPSE TE2000-E, Nikon, Japan). Control sections were processed without primary antibodies. The specificity of the anti-EP2 receptor antibody was confirmed by preincubation of the antibody with EP2 receptor blocking peptide (Cayman Chemical).³⁰ In 12 locations per mouse (four fields per section × three sections per mouse), we used Image J software (NIH, Image J 1.47 t) to count Iba1-, GFAP-, and MPO-immunoreactive cells in the lateral edge of the hematoma along the rostral-caudal axis under a 40 x objective.⁷ Quantifications were averaged and expressed as positive cells per square millimeter (n = 8 mice/group). To target similar regions of interest, we selected sections with similar lesion area. We also used a combination of morphologic criteria to define microglia and macrophages as either resting or activated.³¹

Neurologic deficit score

We assessed neurologic deficit score on days 1 and 3 after ICH (n = 8 mice/group).²⁶ An investigator blinded to the experimental cohort scored all mice on six neurologic tests, including body symmetry, gait, climbing, circling behavior, front limb symmetry, and compulsory circling. Each test was graded from 0 to 4. The maximum deficit score was 24.

Brain lesion volume, swelling, and water content

Mice were euthanized after the neurologic examination on day 3 after ICH (n = 6/group). The entire brain of each mouse was cut with a cryostat into 50-µm sections that were spaced 360 µm apart. Brain sections were stained with Luxol fast blue (for myelin) and Cresyl Violet (for neurons) before being quantified for brain lesion volume and swelling with SigmaScan Pro software (version 5.0.0 for Windows; Systat, San Jose, CA, USA). The lesion volume in cubic millimeters was calculated by multiplying the thickness by the sum of the damaged areas of each section.⁷ On the same sections, brain swelling was quantified by calculating the percentage of hemispheric enlargement, which was expressed as [(ipsilateral hemisphere volume – contralateral hemisphere volume)/contralateral hemisphere volume] × 100%.³²

On day 3 after ICH induced by arterial whole blood, we evaluated brain edema by measuring brain water content. The brains were dissected into ipsilateral and contralateral striatum and cerebellum, which served as an internal control. The percentage of brain water content was calculated as (wet weight − dry weight)/wet weight × 100%.³³

Brain tissue hemoglobin content

Drabkin’s reagent (Sigma) was used to quantify the hemoglobin content in striatal tissue on day 1 after collagenase-induced ICH (n = 5 mice/group).³⁴ The concentration of cyanomethemoglobin in the homogenates of striatal tissue was measured spectrophotometrically at 540 nm.

Histology

On day 3 after ICH, we cut coronal brain sections through the entire striatum and stained them with Fluoro-Jade B (FJB) to quantify degenerating neurons (n = 8 mice/group).^25,32 FJB-positive cells were quantified with the same procedure as detailed under Brain section preparation and immunofluorescence.

Western blotting

Brain tissue from the hemorrhagic hemisphere was obtained at 24 h after ICH (n = 5 mice/group) based on established protocols.^7,35 Twenty microgram protein samples were separated by 4–12% sodium dodecyl sulfate polyacrylamide gel electrophoresis and transferred onto polyvinylidene fluoride membranes. The membranes were probed with the following primary antibodies: rabbit anti-HMGB1 (1:500, Abcam, Cambridge, MA, USA), rabbit anti-interleukin (IL)-1β (1:200, Abcam), and rabbit anti-β-actin (1:3000, Santa Cruz Biotechnology, Dallas, TX, USA). β-Actin was used as a loading control. Bound antibodies were visualized by using a chemiluminescence detection system (LAS-4000, GE Healthcare, Piscataway, NJ, USA). Resulting protein bands were scanned and analyzed with Image J software.

In situ detection of reactive oxygen species

On day 1 after ICH, production of superoxide was measured by in situ detection of oxidized hydroethidine (n = 5 mice/group).^7,23,25 Hydroethidine is a cell-permeable oxidative fluorescent dye that can be oxidized by superoxide to ethidium, which intercalates within the DNA and emits a red fluorescent signal. To compare fluorescent intensity of ethidium in the peri-ICH region between groups, we captured all images using the same contrast settings, intensity, and exposure times. We then calculated fluorescence intensity of ethidium in predefined areas of the hemorrhagic striatum (at the injection site and at 360 µm on each side) after subtracting the color density on the contralateral striatum.

Protein oxidation assay

On day 1 after ICH, protein carbonylation was evaluated with an OxyBlot protein oxidation detection kit (Millipore, Billerica, MA, USA) for protein carbonyl groups, as described previously (n = 5 mice/group).^7,23 Relative values of protein carbonylation were obtained by a Luminescent Image Analyzer LAS-4000 system as for Western blotting.

Gelatin in situ and gel zymography

Both zymography assays were performed based on our established protocols.^7,36 Brains were removed quickly, snap frozen in liquid nitrogen, and stored at −80℃ until further analysis. In situ gelatinolytic activity was analyzed on fresh-frozen, unfixed brain sections by using an EnzChek Gelatinase Assay kit (Life Technologies) on day 1 after ICH (n = 5 mice/group). Cleavage of DQ gelatin-FITC by gelatinases results in a green fluorescent product (excitation, 495 nm; emission, 515 nm). Stained sections were examined with a fluorescence microscope (ECLIPSE TE2000-E, Nikon, Japan). We quantified gelatinolytic activity–positive cells in the perihematomal region of the striatum by using a strategy similar to that described for quantification of immunofluorescence staining.

For gelatin gel zymography, protein samples were prepared from mouse brains containing the striatum at 24 h after ICH (n = 5 mice/group). Equal amounts of protein (500 µg) were purified with gelatin-Sepharose 4B (GE Healthcare Bio-Sciences, Piscataway, NJ, USA) and separated on a 10% Tris-glycine gel with 0.1% gelatin as substrate. A mixture of mouse pro-MMP-9 (98 kDa) and pro-MMP-2 (72 kDa) (R&D Systems, Minneapolis, MN, USA) was used as the gelatinase standard. Gels that were stained with 0.5% Coomassie blue R-250 and destained appropriately were analyzed densitometrically (ImageJ) on randomized coded samples. Gelatinolytic activity was quantified by optical density and expressed as fold increase over that of sham controls.

Statistics

Parametric data are expressed as means ± standard deviations. Differences between two groups were determined by two-tailed Student’s t-test. One-way ANOVA followed by Bonferroni correction was used to compare differences between multiple groups. Statistical significance was set at P < 0.05.

Results

EP2 receptor is expressed in neurons after collagenase-induced ICH

To define the predominant cell type that expresses EP2 receptor on day 3 after ICH, we performed double-immunofluorescence staining with cell-type specific markers. After collagenase-induced ICH, EP2 receptor immunoreactivity colocalized primarily with NeuN⁺ neurons in the perihematomal region of the striatum and in the cortex, but not with GFAP⁺ astrocytes or CD11b⁺ microglia/macrophages (Figure 1).

Figure 1.

EP2 expression in the striatum of middle-aged mice subjected to collagenase-induced intracerebral hemorrhage (ICH). Double labeling immunofluorescence showed that EP2 receptor immunoreactivity (green) colocalized primarily with NeuN⁺ neurons (red), but not with GFAP⁺ astrocytes (red) or CD11b⁺ microglia/macrophages (red), in the peri-hematomal region on day 3 after collagenase-induced ICH. Three sections were analyzed per animal. Scale bar = 40 µm, n = 3 mice.

EP2 receptor deletion and activation modulate brain injury, brain swelling, and neurologic deficits after ICH

To understand the role of EP2 receptor after ICH, we first subjected middle-aged EP2^−/− mice and C57BL/6 wild-type mice to collagenase-induced ICH. Compared with the wild-type mice, EP2^−/− mice showed a twofold increase in lesion volume, a marked increase in brain swelling on day 3, and increased neurologic deficit scores on days 1 and 3 after ICH (n = 8 mice/group, all P < 0.05; Figure 2(a), (c), (e) and (g)); EP2^−/− mice also exhibited elevated brain water content in the ipsilateral striatum (n = 6 mice/group, P < 0.05; Supplementary Figure 1(a)). None of the EP2^−/− or wild-type control mice died before the end of this study.

Figure 2.

Effect of EP2 receptor deletion or activation on brain lesion volume, brain swelling, and neurologic deficits in middle-aged male mice subjected to collagenase-induced intracerebral hemorrhage (ICH). (a, b) Representative Luxol fast blue/Cresyl Violet-stained brain sections on day 3 after ICH; injured areas lack staining. (c, e, g) Bar graphs show that EP2^−/− mice had larger brain lesion volume (c), a greater degree of brain swelling (e), and more severe neurologic deficits (g) than did C57BL/6 wild-type mice. n = 8 mice/group. (d, f, h) Bar graphs show that mice treated with EP2 receptor agonist AE1-259-01 had smaller lesion volume (d), less brain swelling (f), and less severe neurologic deficits (h) than did mice treated with vehicle. Values are mean ± SD; n = 8 mice per group. *P < 0.05, **P < 0.01. NDS, neurologic deficit score.

To exclude the possibility that the increased brain injury in EP2^−/− mice resulted from greater cerebral bleeding, we measured hemoglobin content in the injured striatum at 24 h after collagenase injection, when hematoma reaches its maximum in this model.³⁴ We did not observe any difference in hemoglobin content between C57BL/6 wild-type mice and EP2^−/− mice (n = 8 mice/group, P > 0.05; Supplementary Figure 1(b)), indicating that EP2 receptor deletion does not increase collagenase-induced bleeding.

We corroborated the deleterious effects of EP2 deletion in the blood ICH model by showing that in this model also, EP2^−/− mice had greater brain water content and higher neurologic deficit scores than did wild-type mice (n = 6 mice/group, both P < 0.05; Supplementary Figure 1(C) and (D)).

Using pharmacologic tools, we further confirmed that, compared with outcomes in the vehicle-treated group, EP2 receptor activation by AE1-259-01 lessened lesion volume (by 54%), brain swelling, and neurologic deficit score on day 3 after collagenase-induced ICH (n = 8 mice/group, all P < 0.05; Figure 2(b), (d), (f) and (h)). The mortality rate was 2.1% (1 of 47) in both mice treated with AE1-259-01 and mice treated with vehicle. Together, these data support the protective effects of EP2 receptor after ICH at the tissue level.

Neuronal degeneration is increased by EP2 receptor deletion but decreased by EP2 receptor activation after collagenase-induced ICH

To determine whether the beneficial effect of EP2 receptor is reflected on the cellular level, we used FJB histologic staining to quantify neuronal death in free-floating brain sections collected three days after ICH. EP2^−/− mice had more FJB-positive neurons in the perihematomal region than did C57BL/6 wild-type mice (n = 8 mice/group, P < 0.05; Figure 3(a) and (b)), whereas mice treated with EP2 receptor agonist AE1-259-01 had fewer FJB-positive neurons (n = 8 mice/group, P < 0.01; Figure 3(c) and (d)).

Figure 3.

Effect of EP2 receptor deletion or activation on neuronal death in middle-aged male mice subjected to collagenase-induced intracerebral hemorrhage (ICH). (a, c) Representative Fluoro-Jade B (FJB)-stained brain sections on day 3 after ICH showing intensely labeled neurons and processes in the perihematomal region. Scale bar = 30 µm. (b, d) Quantification analysis shows that EP2^−/− mice had more FJB-positive neurons in the perihematomal region than did C57BL/6 wild-type mice (b), whereas mice treated with AE1-259-01 had fewer FJB-positive neurons in the perihematomal region than did vehicle-treated mice (d). Insets in (a, c) represent FJB-positive degenerating neurons at higher magnification. Values are means ± SD. n = 8 mice per group. *P < 0.05, **P < 0.01.

Cellular inflammatory responses are increased by EP2 receptor deletion and decreased by EP2 receptor activation after collagenase-induced ICH

In the early stage after ICH, microglia and astrocytes are activated in the perihematomal region, and neutrophils and macrophages infiltrate the area.^1,37,38 Using cell-type-specific markers, we showed that EP2^−/− mice had more activated microglia/macrophages (Iba1), astrocytes (GFAP), and infiltrating neutrophils (MPO) in the perihematomal region than did C57BL/6 wild-type mice on day 3 after ICH (n = 8 mice/group, all P < 0.05; Figure 4(a)). Conversely, mice treated with EP2 receptor agonist AE1-259-01 showed a decrease in the number of these inflammatory cells (n = 8 mice/group, all P < 0.05; Figure 4(b)). Similarly, on day 1 after ICH, EP2^−/− mice exhibited increased production of cytokines IL-1β (Figure 4(c) and (d)) and HMGB1 (Figure 5(a) and (b)) compared with that in the C57BL/6 wild-type mice (n = 5 mice/group, both P < 0.05), and AE1-259-01-treated mice exhibited a decrease in production of these proinflammatory markers compared with that in vehicle-treated controls (n = 5 mice/group, both P < 0.05).

Figure 4.

Effects of EP2 receptor deletion or activation on cellular inflammatory responses and proinflammatory cytokine IL-1β production in middle-aged male mice subjected to collagenase-induced intracerebral hemorrhage (ICH). (a, b) In representative images, Iba1-, GFAP-, and MPO-immunopositive cells are evident around the hematoma on day 3 after ICH. Scale bar = 40 µm. Bar graphs show that EP2^−/− mice had more activated microglia/macrophages, activated astrocytes, and infiltrating neutrophils than did C57BL/6 wild-type mice (a). n = 8 mice/group. In contrast, mice treated with AE1-259-01 had fewer activated microglia/macrophages, activated astrocytes, and infiltrating neutrophils than did mice treated with vehicle (b). n = 8 mice/group. (c, d) Western blot analysis shows that the level of IL-1β was increased in EP2^−/− mice compared with that in C57BL/6 wild-type mice ((c), n = 5 mice/group) and was decreased in mice treated with AE1-259-01 compared with that in mice treated with vehicle ((d), n = 5 mice/group) on day 1 after ICH. β-actin was used as loading control. Values are means ± SD. *P < 0.05, **P < 0.01 versus C57BL/6 wild-type mice or vehicle-treated mice; ^#P < 0.05 versus sham mice.

Figure 5.

Effects of EP2 receptor deletion or activation on proinflammatory cytokine HMGB1 expression in middle-aged male mice subjected to collagenase-induced intracerebral hemorrhage (ICH). (a, b) Representative Western blots show the expression of HMGB1 protein on day 1 after ICH. Densitometric analysis shows that the level of HMGB1 protein increased in EP2^−/− mice compared with that in C57BL/6 wild-type mice (a) and decreased in AE1-259-01-treated mice compared with that in vehicle-treated mice on day 1 after ICH. β-actin was used as a loading control. Values are means ± SD. n = 5 mice/group; *P < 0.05 versus C57BL/6 wild-type mice or vehicle-treated mice; ^#P < 0.05 versus sham mice.

Oxidative stress is increased by EP2 receptor deletion and decreased by EP2 receptor activation after collagenase-induced ICH

Oxidative stress in the hemorrhagic brain causes neuronal death.^1,35,36 We used the fluorescent indicator hydroethidine to evaluate in situ production of superoxide. We also assessed protein carbonylation. The basal levels of superoxide production (small red particles) in C57BL/6 wild-type and EP2^−/− mice were very low. On day 1 after ICH, EP2^−/− mice had elevated superoxide production in the perihematomal region (n = 8 mice/group, Figure 6(a)) and an increased level of carbonylated proteins (n = 5 mice/group, Figure 6(c)) compared with that in the C57BL/6 wild-type mice (both P < 0.05); conversely, mice treated with EP2 receptor agonist AE1-259-01 had lower levels of superoxide (n = 8 mice/group, Figure 6(b)) and carbonylated proteins (n = 5 mice/group, Figure 6(d); both P < 0.01).

Figure 6.

Effects of EP2 receptor deletion or activation on superoxide production and protein oxidation in middle-aged male mice subjected to collagenase-induced intracerebral hemorrhage (ICH). (a, b) In representative images, ethidium fluorescence (small red particles), a marker for superoxide production, was evident in the perihematomal region on day 1 after ICH. Scale bar: 30 µm. Bar graphs show the quantification analysis of ethidium fluorescence intensity in the perihematomal region. EP2^−/− mice had greater signal intensity than did C57BL/6 wild-type mice (a), and mice treated with AE1-259-01 had less signal intensity than did mice treated with vehicle (b). (c, d) Representative immunoblots of hemorrhagic brain tissue. Bar graphs show that EP2^−/− mice had a higher level of protein carbonylation than C57BL/6 wild-type mice (c) and that mice treated with AE1-259-01 had less protein carbonylation than mice treated with vehicle (d) on day 1 after ICH. Optical density was integrated over multiple protein bands for carbonyls. β-actin was used as loading control. Values are means ± SD. n = 5–8 mice/group; *P < 0.05, **P < 0.01.

Gelatinolytic activity is increased by EP2 receptor deletion and decreased by EP2 receptor activation after collagenase-induced ICH

The inflammatory response and resulting increase in oxidative stress lead to an increase in gelatinolytic activity (MMP-2/9), which is known to contribute to blood–brain barrier disruption and brain edema after ICH.^1,35,36 We reported previoulsy that gelatinolytic activity is associated mostly with neurons and vascular structures after ICH.³⁶ In this study, we examined gelatinolytic activity in fresh-frozen brain sections by in situ zymography and in brain tissue by gelatin gel zymography. The basal levels of the in situ gelatinolytic activity in C57BL/6 wild-type and EP2^−/− mice were barely detectable. On day 1 after ICH, EP2^−/− mice had increased gelatinolytic activity in the perihematomal region (n = 8 mice/group, Figure 7(a)) and increased intensity of pro-MMP-2 and pro-MMP-9 bands (n = 5 mice/group, Figure 7(c)) compared with that in the C57BL/6 wild-type mice (both P < 0.01), whereas mice treated with AE1-259-01 had lower levels of gelatinolytic activity and lower intensity pro-MMP-2 and pro-MMP-9 bands (n = 5 mice/group, both P < 0.05; Figure 7(b) and (d)). The active MMP-2/9 bands were not reliably detectable, consistent with previous reports.^39,40

Figure 7.

Effects of EP2 receptor deletion or activation on gelatinolytic activity in middle-aged male mice subjected to collagenase-induced intracerebral hemorrhage (ICH). (a, b) Representative gelatin in situ zymography fluorescent images of the perihematomal region from EP2^−/− and C57BL/6 wild-type mice (a) and from vehicle- and AE1-259-01-treated mice (b) on day 1 after ICH. Scale bar: 30 µm. Bar graphs show that the number of gelatinolytic-positive cells was increased in EP2^−/− mice compared with that in C57BL/6 wild-type mice (a) and decreased in AE1-259-01–treated mice compared with that in vehicle-treated mice (b). (c, d) Representative gelatin gel zymographs of MMP-2 and MMP-9 activity, visible as white bands on gels where gelatin was degraded. Bar graphs show that the gelatinolytic activity of pro-MMP-2 and pro-MMP-9 was greater in EP2^−/− mice than in C57BL/6 wild-type mice (c) and less in AE1-259-01-treated mice than in vehicle-treated mice (d) on day 1 after ICH. Std, mouse gelatinase standards. Values are means ± SD. n = 5–8 mice/group; *P < 0.05, **P < 0.01.

Discussion

Previously, we and others have shown that EP1 receptor confers toxicity in in vitro and in vivo models of ICH.^6,7 The current study shows that in contrast to EP1, the EP2 receptor confers cerebroprotection after ICH in middle-aged male mice. Our study presents several novel findings: (1) EP2 receptor is expressed primarily in neurons after ICH, not in astrocytes or microglia/macrophages; (2) in mice subjected to the collagenase and blood ICH models, EP2 deletion exacerbates brain injury, swelling or edema, and neurologic deficits, whereas EP2 receptor activation reverses these negative effects; (3) in mice subjected to the collagenase ICH model, EP2 receptor deletion increases neuronal death, cellular inflammatory responses, production of proinflammatory cytokines IL-1β and HMGB1, oxidative stress, and MMP-2/9 activity, whereas EP2 activation decreases those factors. Together, these observations indicate that activation of neuronal EP2 receptor by PGE₂ protects brain against ICH injury through its anti-inflammatory, antioxidative, and anti-MMP-2/9 properties. To minimize the concern that increases or decreases in neuronal death, cellular inflammatory responses, reactive oxygen species (ROS) production, and in situ gelatinolytic activity are due to differences in lesion volume, we performed profile-based cell counting on brain sections with similar lesion areas from each mouse group.

In the brain, EP2 receptor is expressed primarily in striatal neurons.^12,41 Neuronal EP2 signaling has been shown to be protective in in vitro and in vivo models of cerebral ischemia.^11,16,17 Interestingly, in vitro studies have shown that activation of the EP2 receptor also regulates microglial activation and function,^13–16 which implies a detrimental role in chronic neuropathologic conditions such as Alzheimer’s disease,^19,20 Parkinson’s disease,²¹ and amyotrophic lateral sclerosis.²² In mice that underwent collagenase-induced ICH, EP2 receptors were expressed primarily in striatal neurons. Moreover, we found that EP2 deletion increased acute ICH injury, whereas its activation decreased injury. These results are consistent with those from ischemic stroke models.^12,41 A neuroprotective role of EP2 signaling was also recently reported in an in vitro model of ICH in which mouse primary cortical neurons were treated with toxic levels of hemin.¹¹ That in vitro data, together with our in vivo findings, indicate that neuronal, but not microglial, EP2 signaling promotes cerebroprotection in acute ICH pathology. Interestingly, contrary to their in vitro data and our in vivo data, the same group later reported that EP2 deletion decreases ICH-induced brain damage and improves functional recovery in young adult mice.⁴² Such contradictory in vivo data may be attributable to the differences in the age of the animals studied (middle-aged mice versus young adult mice) and the lesion size of the wild-type control mice (4.3 versus 17.8 mm³). This discrepancy supports a previous hypothesis that the effect of ICH on activation of certain pathways depends on hematoma size.⁴³ Collectively, the data would suggest that EP2 receptor activation is protective in the setting of small hematomas but deleterious after a large ICH.

Preclinical and clinical evidence indicates that inflammatory responses contribute to ICH-induced secondary brain injury.^37,44 The cellular inflammatory responses involve activation of microglia and astrocytes, infiltration of leukocytes, and consequent release or activation of inflammatory mediators, including proinflammatory cytokines, ROS, and MMPs.³⁷ We have reported that PGE₂/EP1 signaling exacerbates neuroinflammation.⁷ Here, we addressed whether the EP2 receptor also modulates these inflammatory responses. Our data indicate that deletion of EP2 receptor exacerbates cellular inflammatory responses, including microglial and astrocyte activation, neutrophil infiltration, and proinflammatory cytokine IL-1β and HMGB1 production. Importantly, EP2 activation mitigates these inflammatory responses. These data suggest that the EP2 receptor has anti-inflammatory properties after ICH. Microglia/macrophages are known to be able to differentiate into an M1 (classically activated) or M2 (alternatively activated) phenotype. Therefore, it would be valuable to investigate whether neuronal EP2 signaling modulates microglia/macrophage phenotype and function.

The proinflammatory cytokine HMGB1 is released from dying neurons and represents a paracrine inflammatory mediator within the neurovascular unit.⁴⁵ Anti-HMGB1 neutralizing antibody was able to protect the blood–brain barrier and decrease ischemic stroke injury.⁴⁶ Additionally, serum level of HMGB1 was shown to increase within 12 h after onset of ICH in patients.⁴⁷ We recently reported that HMGB1 is expressed mostly in neurons in the perihematomal region at 5 and 24 h after ICH and that the nonselective EP2/EP4 receptor agonist misoprostol decreases HMGB1 expression and confers neuroprotection after ICH.²³ The toxic role of HMGB1 in the acute stage of ICH was further confirmed by using the nonspecific HMGB1 inhibitor glycyrrhizin.²³ In this study, the marked increase or decrease in IL-1β and HMGB1 conferred by EP2 deletion and activation, respectively, suggest that neuronal EP2 signaling can also modulate molecular inflammatory responses.

Microglia/macrophages and leukocytes are major sources of ROS, which can activate MMP-2/9 and damage the blood–brain barrier and neurovascular unit.^37,38,44 In ICH models, microglial activation occurs earlier than neutrophil infiltration.³⁷ We and others have confirmed that inhibition of early microglial activation,^48,49 depletion of neutrophils,⁵⁰ and inhibition of MMP-2/9 activity^36,51 are able to reduce ROS production and provide neuroprotection after ICH. Moreover, PGE₂ is able to induce MMP-9 expression in dendritic cells.⁵² After cerebral ischemia, HMGB1 can upregulate MMP-9 expression/activity in neurons and astrocytes, signaling neighboring cells to increase blood–brain barrier permeability and recruit immune cells.⁵³ We recently reported that HMGB1 inhibition decreases MMP-9 activity, thereby decreasing ICH-induced early brain injury.²³ We showed here that the increase or decrease in cellular and molecular inflammatory responses conferred by EP2 deletion or activation, respectively, was associated with a corresponding increase or decrease in ROS production and MMP-2/9 activity. These data indicate that neuronal EP2 signaling has antioxidant effects and anti-MMP2/9 activity.

Our study provides proof of concept that selective activation of EP2 receptor after ICH has protective effects. In line with this finding and in a more clinically relevant setting, we have shown recently that post-treatment with the EP2/EP4 receptor agonist misoprostol protects brain against ICH injury.²³ However, PGE₂ acts through four receptors (EP1–4), and we recently reported that the toxicity of EP1 activation after ICH is mediated through mechanisms that involve the Src kinases and the MMP-9 signaling pathway.⁷ Therefore, the synergistic or antagonistic effects of EP2 receptor with other EP receptors need to be determined.

In conclusion, using genetic and pharmacologic strategies, we have identified an unrecognized cerebroprotective effect of PGE₂ EP2 receptor in two mouse ICH models. This protection by neuronal EP2 receptor signaling could be mediated through anti-inflammatory, antioxidant, and anti-MMP-2/9 effects. PGE₂/EP2 signaling warrants further investigation as a potential approach to ICH treatment.

Funding

This work was supported by The Natural Science Foundation of Heilongjiang Province of China LC2013C30 (to HW), and the National Natural Science Foundation of China 81200885 (to HW); American Heart Association grant 13GRNT15730001 (to JW); National Institutes of Health grants K01AG031926, R01NS078026, and R01AT007317 (to JW); and an American Heart Association postdoctoral fellowship award (14POST20140003, to XH).

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Authors’ contributions

JW conceived, designed, and coordinated the research; HW and TW participated in the research design; HW, TW, and WC performed the research; HW, TW, XH, JW, CJ, WC, HL, QY, and JW analyzed data and wrote the paper; JW obtained funding. All authors read and approved the final draft. HW and TW contributed equally to this work.

Supplementary material

Supplementary material for this paper can be found at http://jcbfm.sagepub.com/content/by/supplemental-data