Inducible Prostaglandin E Synthase as a Pharmacological Target for Ischemic Stroke

Abstract

As the inducible terminal enzyme for prostaglandin E2 (PGE₂) synthesis, microsomal PGE synthase-1 (mPGES-1) contributes to neuroinflammation and secondary brain injury after cerebral ischemia via producing excessive PGE₂. However, a proof of concept that mPGES-1 is a therapeutic target for ischemic stroke has not been established by a pharmacological strategy mainly due to the lack of drug-like mPGES-1 inhibitors that can be used in relevant rodent models. To this end, we recently developed a series of novel small-molecule compounds that can inhibit both human and rodent mPGES-1. In this study, blockade of mPGES-1 by our several novel compounds abolished the lipopolysaccharide (LPS)-induced PGE₂ and pro-inflammatory cytokines interleukin 1β (IL-1β), IL-6, and tumor necrosis factor α (TNF-α) in mouse primary brain microglia. Inhibition of mPGES-1 also decreased PGE₂ produced by neuronal cells under oxygen–glucose deprivation (OGD) stress. Among the five enzymes for PGE₂ biosynthesis, mPGES-1 was the most induced one in cerebral ischemic lesions. Systemic treatment with our lead compound MPO-0063 (5 or 10 mg/kg, i.p.) in mice after transient middle cerebral artery occlusion (MCAO) improved post-stroke well-being, decreased infarction and edema, suppressed induction of brain cytokines (IL-1β, IL-6, and TNF-α), alleviated locomotor dysfunction and anxiety-like behavior, and reduced the long-term cognitive impairments. The therapeutic effects of MPO-0063 in this proof-of-concept study provide the first pharmacological evidence that mPGES-1 represents a feasible target for delayed, adjunct treatment — along with reperfusion therapies — for acute brain ischemia.

Supplementary Information

The online version contains supplementary material available at 10.1007/s13311-022-01191-1.

Introduction

Despite recent marked advances in disease diagnosis and prevention, stroke remains a leading cause of death and adult disability globally. Ischemic stroke, the most common type of stroke, accounts for about 87% of all stroke cases. Currently, the intravenous thrombolysis utilizing recombinant tissue-type plasminogen activator is the only medicinal therapy approved by the US FDA for acute ischemic stroke. However, it must be administered in about 4.5 h after stroke symptom onset and has potential risks in peripheral bleeding and hemorrhagic transformation [1]. The emerging endovascular therapy by intra-arterial mechanical thrombectomy has some advantage owing to a slightly more extended intervention window — up to 6 h; however, its current restriction to occlusions in large arteries dramatically reduces the patient eligibility [2, 3]. Due to overall narrow therapeutic windows and limited patient eligibility, a large proportion of patients suffering from acute brain ischemia are unable to access these two therapies. Development of alternative, safe, and effective medications for delayed treatment of ischemic stroke is in urgent unmet demand.

Ischemic injuries are often followed by excessive immune activation and inflammation involving immunity systems in both the central nervous system (CNS) and peripheral, leading to delayed but profoundly escalated injuries to the ischemic brain [4–6]. It has been widely proposed that targeting these inflammatory and immune pathways would be able to confer neuroprotective effects following brain ischemia–reperfusion injuries with extended therapeutic windows [7]. As the inducible isoform of cyclooxygenase (COX) and a well-known target for anti-inflammatory therapeutics, COX-2 in the brain can be rapidly and robustly upregulated by ischemic injuries [8, 9]. The sudden increase in COX-2 activity in turn leads to the synthesis and quick accumulation of prostaglandin E2 (PGE₂) at the lesion sites, which is considered a chief executor of COX-2 cascade-mediated aggravation of the inflammation-associated secondary tissue damages. This has been supported by extensive evidence that the genetic ablation or pharmacological inhibition of COX-2 consistently mitigates the ischemic damages in animal models [10–12]. However, the therapeutic strategy of targeting COX-2 for ischemic injuries and other inflammatory conditions has been dampened by the life-threatening adverse effects of COX-2-targeting drugs within the cerebrovascular and cardiovascular systems. Long-term COX-2 inhibition is believed to cause the deficiency of prostacyclin, which is another prostanoid product of COX in the circulatory system and functions as an essential vasodilator to counteract blood clot formation and vasoconstriction mediated by thromboxane, leading to a prothrombotic status [13–15]. The Hyde and Jekyll nature of COX/prostanoid cascade and the importance of balanced prostacyclin and thromboxane to the vascular systems inspired us and many others that targeting the downstream PGE₂ synthases or receptors might represent alternative anti-inflammatory and neuroprotective strategies for the delayed treatment of brain ischemia with more specificity than the general blockade of the entire COX cascade [16–19]

Coupled with COX-2, the microsomal prostaglandin E2 synthase-1 (mPGES-1) is an inducible form of the terminal enzyme for PGE₂ biosynthesis — PGES, which also has two other constitutive isozymes: microsomal PGES-2 (mPGES-2) and cytosolic PGES (cPGES). To synthesize PGE₂, COXs (COX-1 and COX-2) first convert arachidonic acid to an intermediate molecule prostaglandin H2 (PGH₂), which is then further catalyzed to PGE₂ by PGES (mPGES-1, mPGES-2, and cPGES) [20, 21]. In line with COX-2, mPGES-1 is also widely upregulated in neurons, microglia, and endothelial cells at the lesion sites of ischemic brain and directly converts COX-2-derived PGH₂ to PGE₂ [22, 23]. Genetic deficiency of mPGES-1 in mice led to the depletion of cortical PGE₂ following middle cerebral artery occlusion (MCAO) and diminished brain ischemia–induced infarction, edema, apoptosis, and neurological deficits [22]. These interesting findings from genetic strategies suggest that the mPGES-1 might represent an alternative to COX-2 as an anti-inflammatory target for the treatment of ischemic stroke [23]. However, the feasibility of mPGES-1 inhibition as a therapeutic strategy has not been pharmacologically validated in animal models of brain ischemia because none of the traditional potent inhibitors of human mPGES-1 has shown adequate blockade of its mouse or rat counterpart due to the interspecies difference in the structure between human and rodent mPGES-1 [24, 25].

We recently developed a series of phenylsulfonyl hydrazide derivatives that selectively and potently inhibit both human and murine mPGES-1 without affecting COX activities [26]. Among these, MPO-0063 (N-phenyl-N′-(4-benzyloxyphenoxycarbonyl)-4-chlorophenylsulfonyl hydrazide, PBCH) showed promising drug-like properties and broad anti-inflammatory effects in several in vitro and in vivo models of inflammation-associated conditions such as peripheral edema and arthritis [27]. Taking advantage of these newly developed mPGES-1 inhibitors, we herein determined the feasibility of pharmacological inhibition of mPGES-1 as a delayed treatment for ischemic stroke in a mouse model of transient focal brain ischemia. We first evaluated the effects of our mPGES-1 inhibitors on lipopolysaccharide (LPS)-activated mouse primary microglia as well as mouse neuronal cell lines under oxygen–glucose deprivation (OGD) stress. We then assessed the post-stroke effects of our current lead mPGES-1 inhibitor MPO-0063 on neurological deficits and cerebral infarction following transient MCAO. Lastly, the long-term effects of mPGES-1 inhibition on brain ischemia-associated locomotor dysfunction, anxiety-like behavior, and cognitive deficits were determined.

Materials and Methods

Chemicals and Drugs

The selective mPGES-1 inhibitors MPO-0057, MPO-0063, and MPO-0112 were synthesized in our laboratories as we previously described [26–28]. Selective mPGES-1 inhibitor compound 3 (C3) was purchased from Tocris Bioscience (cat. no. 5957). COX-2 inhibitor celecoxib was purchased from Cayman Chemical (cat. no. 10008672). Lipopolysaccharide (LPS) was purchased from Sigma-Aldrich (cat. no. L4391).

Mouse Primary Microglial Cultures

The mouse primary microglial cultures were generated using a modified protocol for rat primary microglial cultures that we previously described [29–31]. In brief, primary microglia were isolated from the cortices of newborn C57BL/6 mouse pups (P1). Cortical tissues were dissected in icy Hanks’ Balanced Salt Solution (HBSS, Corning) with meninges and blood vessels being carefully removed. The tissues were minced and triturated a few times to separate cortical cells, which were then rinsed once in HBSS. The cells were initially cultured in Minimum Essential Medium (MEM) supplemented with 10% fetal bovine serum (FBS), 100 U/ml penicillin, and 100 µg/ml streptomycin in 0.001% poly-l-ornithine-coated flasks. After 4–5 days, the culture medium was replaced by fresh complete MEM plus 2 ng/ml granulocyte–macrophage colony-stimulating factor (GM-CSF) (R&D Systems, cat. no. 518-GM-005/CF). After another 4–5 days, microglial cells were separated from the underlying astrocytic monolayer by gentle agitation at room temperature. The detached cells were collected and cultured in complete MEM with 0.2 ng/ml GM-CSF in 24-well plates. Such mouse primary cultures typically consisted of > 95% Iba1-positive cells and were ready for experimental treatment.

Microglia-Mediated Inflammation

Mouse primary microglial cells were first pre-treated by tested compounds (mPGES-1 and COX-2 inhibitors) at concentrations indicated for 15 min and then were co-incubated with LPS (100 ng/ml) overnight. ELISA kits were used to measure the levels of PGE₂ (Arbor Assay, cat. no. K051), interleukin 1β (IL-1β) (R&D Systems, cat. no. MLB00C), IL-6 (R&D Systems, cat. no. M6000B0), and tumor necrosis factor α (TNF-α) (R&D Systems, cat. no. MTA00B) that were released into the culture medium by microglia following the manufacturers’ protocols [30].

OGD in Neuronal Cells

In vitro oxygen–glucose deprivation (OGD) in mouse neuronal cell lines HT-22 (EMD Millipore, cat. no. SCC129) and Neuro-2a (ATCC, cat. no. CCL-131) was performed using a modified hypoxic cell culture method as previously described [32, 33]. Briefly, mouse neuronal cells were seeded in 24-well plates (HT-22: 5 × 10⁴ per well; Neuro-2a: 8 × 10⁴ per well) with normal complete DMEM plus 10% FBS and cultured overnight. Cells were then cultured in glucose/glutamine-free DMEM (Gibco, cat. no. A1443001) supplemented with 1% FBS, and the culture plates were sealed in a vacuum bag. To achieve the hypoxic condition, air in the bag was continually aspirated using a vacuum pump. This approach dramatically and consistently decreased air pressure in the vacuum bag to as low as 0.079 standard atmosphere, which is equivalent to about 1.66% oxygen level.

Experimental Animals

All animal experiments and procedures were approved by the Institutional Animal Care and Use Committee (IACUC) of the University of Tennessee Health Science Center and carried out in accordance with the Guide for the Care and Use of Laboratory Animals (the Guide) from the National Institutes of Health (NIH). Adult male C57BL/6 mice (10–12 weeks old) from Charles River Laboratories were housed in standard humidity (~45–50%) at room temperature (~21–25 °C) and under a 12 h light/dark cycle with free access to food and water.

Complete Blood Count

The whole blood samples were drawn from the facial veins of mice and collected into EDTA-coated blood collection tubes (BD Microtainer, cat. no. 365974). Complete blood count (CBC) was performed using an Xpedite HEM Vet hematology analyzer in the Regional Biocontainment Laboratory (RBL) at the University of Tennessee Health Science Center.

Brain Ischemia Model

Acute transient focal brain ischemia was introduced by middle cerebral artery occlusion (MCAO) for 30 min (mild) or 1 h (severe) as we previously described [34]. The anesthesia of mice was induced and maintained via the inhalation of vaporized isoflurane (Henry Schein) at 5% and 1.5–2%, respectively. The rectal temperature was monitored by a digital thermometer, and the body temperature was virtually maintained at 37 °C using a heating pad throughout the surgery. Once the anesthetic status was reached, a skin incision was made along the middle line on the neck. The subcutaneous soft tissues were gently separated, and the common carotid artery (CCA) on the right side was carefully exposed. With the adjacent vagal nerve being protected from surgical damage, the superior thyroid artery was ablated before the ligation of right external carotid artery (ECA). The ECA was cut off using a cauterizer (low-temperature cautery kit, Bovie Medical Corporation). The occipital artery was then carefully ablated above the bifurcation of CCA-ECA/ICA (internal carotid artery). The pterygopalatine artery (PPA) was thus visible. The CCA and ICA were clamped with microvascular clips. A 5–0 suture knot (LOOK Silk Suture) was loosely prepared at the root of ECA. A tiny nick was made along the ECA stump for the insertion of filament into the ECA lumen.

MCAO was achieved through the delivery of a 2-cm-long 6–0 silicon-coated Doccol filament into exactly 1 cm to block the blood supply at the origin of MCA. The filament was then fixed by tightening the loose knot at the root of ECA, and it is important to ensure no hemorrhage at the nick site on ECA. During the maintenance of occlusion, animals were placed above a heating pad before their restoration of consciousness from anesthesia. The reperfusion was introduced by withdrawing the Doccol filament. Analgesia was achieved by administering a dose of Buprenorphine SR-LAB (1.0 mg/kg, s.c.) 60 min before the surgery. Buprenorphine releases over 72 h and provides blood levels greater than 1.0 ng/ml in mice for sufficient postoperative analgesia. The researcher who performed the MCAO surgery had no knowledge about the plan of treatment for these mice.

Animal Treatment and Mortality After MCAO

Mice were randomized and blindly administered with either vehicle (10% DMSO, 50% PEG 400, 40% ddH₂O) or mPGES-1 selective inhibitor MPO-0063 (5 or 10 mg/kg, i.p.) at 4.5, 12, and 24 h after MCAO. Mice were monitored daily for weights and neurological deficits, provided with moistened soft chow, and treated with Ringer’s lactate solution supplemented with 5% dextrose (Baxter) (0.5–1 ml, s.c.). The endpoints were 72 h after 1 h MCAO for short-term biochemical and histological examinations and 30 days after 30 min MCAO for behavioral tests.

In the short-term study (1 h MCAO), there were 39, 28, and 17 mice in the 0, 5, and 10 mg/kg MPO-0063 treatment groups, respectively, and 13 mice in each group survived 72 h after 1 h MCAO. In the long-term behavioral study (30 min MCAO), there were 13 mice in the vehicle group, among which 10 and 8 animals survived 7 and 30 days after MCAO for 30 min, respectively; twelve MCAO mice were treated by 10 mg/kg MPO-0063, and 8 and 7 animals survived 7 and 30 days after 30 min MCAO, respectively. It appears that the post-stroke mortality rate was highly correlated with the duration of MCAO because when vehicle-treated mice were compared, 1 h MCAO led to a much higher mortality rate than 30 min MCAO in the first week (66.7% vs. 23.1%).

Neurological Assessment

The assessment of post-MCAO neurological deficits was performed following a protocol modified from Bederson and Longa scoring systems in a blinded manner (Table 1) [34–36].

Table 1.

Modified Bederson scale for post-stroke neurological assessment

Score	Description
0	No deficit: when suspended by tail, the animal fully extends bilateral forelimbs forward
1	Forelimb flexion: the animal flexes the forelimb on its contralateral side; forelimb flexion involves the wrist, elbow, and shoulder
2	Reduced resistance to lateral push: a normal or mildly injured animal shows equivalent resistance to unilateral push–induced passive movement; mouse with more severe ischemic damage typically shows weaker resistance on the contralateral side than that on the ipsilateral side
3	Unidirectional circling: circling and tail-chasing usually suggest that large ischemic damage is obtained
4	“Barrel rolling/spinning” along the longitudinal (rolling) axis, a sign of seizure attack with the loss of consciousness
5	No movement: the animal completely loses its spontaneous locomotion and body movement

Quantification of Infarct and Edema

At the endpoint of study, mice were euthanized under deep isoflurane anesthesia and transcardial perfusion with ice-cold phosphate-buffered saline (PBS). A mouse coronal brain matrix (Alto) was used to prepare 1-mm coronal sections. The brain sections were stained with 0.2% TTC solution (2,3,5-triphenyltetrazolium chloride, Santa Cruz Biotechnology) for 20 min and imaged digitally for analyses. The infarct size of each hemisphere was measured using Fiji software (NIH) in a blinded manner. The calculation of global infarct size follows the Cavalieri’s principle in three-dimensional case (or “trapezoidal rule”). The reported infarction was adjusted according to the ipsilateral edema size, which is the ipsilateral hemispheric size subtracted by that of the contralateral counterpart. This procedure is referred to as ‘Swanson’s correction’ [37]. The infarct volume and edema size in the ipsilateral hemisphere were normalized to its contralateral counterpart to avoid any artifactual effects (extension or shrinkage) that might be introduced during the tissue processing. Brains with excessive hemorrhagic transformation and/or infarct that extended outside of the MCA territory were excluded from the study. Consequently, four mice in the vehicle group, 3 mice in the 5 mg/kg MPO-0063 group, and 2 mice in the 10 mg/kg MPO-0063 group were identified and excluded from the analyses.

Quantitative PCR

The total RNA from mouse brain tissues was isolated using TRIzol (Invitrogen) with the PureLink RNA Mini Kit (Invitrogen). RNA purity and concentration were measured by A260/A280 ratio and A260 value, respectively, using a NanoDrop One microvolume spectrophotometer (Thermo Fisher). The first-strand complementary DNA (cDNA) was synthesized using the SuperScript III First-Strand Synthesis SuperMix (Invitrogen) following the product manual. The quantitative PCR (qPCR) was performed using 8 µl of 10 × diluted cDNA, 0.4 µM of primers (Table 2), and 2 × SYBR Green Supermix (Bio-Rad Laboratories) with a final volume of 20 µl in a CFX96 Touch Real-Time PCR Detection System (Bio-Rad Laboratories). Cycling conditions were as follows: 95 °C for 2 min followed by 40 cycles of 95 °C for 15 s and then 60 °C for 1 min. Melting curve analysis was used to verify single-species PCR product. Fluorescent data were acquired at the 60 °C step. The cycle of quantification for GAPDH gene was subtracted from the cycle of quantification measured for each gene of interest to yield ∆Cq [30, 38]. Samples without cDNA template served as negative controls.

Table 2.

Primers for quantitative PCR (qPCR)

Gene	Forward primer (5′ → 3′)	Reverse primer (5′ → 3′)	Amplicon size (bp)	NCBI reference sequence
Ptgs1 (COX-1)	GGGAATTTGTGAATGCCACC	GGGATAAGGTTGGACCGCA	76	NM_008969.4
Ptgs2 (COX-2)	CTCCACCGCCACCACTAC	TGGATTGGAACAGCAAGGAT	118	NM_011198.4
Ptges (mPGES-1)	ATCAAGATGTACGCGGTGGC	GAGGAAATGTATCCAGGCGA	240	NM_022415.3
Ptges2 (mPGES-2)	CCGTGAGAAGGACTGAGATC	GTAGGTCTTGAGGGCACTAATG	125	NM_133783.2
Ptges3 (cPGES)	GTTTGCGAAAAGGAGAATCCG	CTCAGAGAAACGGTCAAAATTCG	149	NM_019766.4
Il1b (IL-1β)	TGAGCACCTTCTTTTCCTTCA	TTGTCTAATGGGAACGTCACAC	101	NM_008361.4
Il6 (IL-6)	TCTAATTCATATCTTCAACCAAGAGG	TGGTCCTTAGCCACTCCTTC	119	NM_031168.2
Tnf (TNF-α)	TCTTCTGTCTACTGAACTTCGG	AAGATGATCTGAGTGTGAGGG	111	NM_013693.3
Gapdh (GAPDH)	TGTCCGTCGTGGATCTGAC	CCTGCTTCACCACCTTCTTG	75	NM_008084.3

ELISA for Brain Tissue Homogenates

Fresh brain tissues from ipsilateral hemispheres were homogenized with RIPA buffer (Pierce, cat. no. 89901), and the total protein levels were quantified using BCA assays (Pierce, cat. no. 23225). ELISA kits were used to measure the levels of PGE₂ (Arbor Assay, cat. no. K051), PGI₂ (BIOMATIK, cat. no. EKC37639), PGF_2α (Cayman chemical, cat. no. 516011), IL-1β (R&D Systems, cat. no. MLB00C), IL-6 (R&D Systems, cat. no. M6000B0), and TNF-α (R&D Systems, cat. no. MTA00B) in the brain tissue homogenates following the manufacturers’ protocols.

Open Field Test

The open field test was utilized to assess locomotor dysfunction and anxiety-like behavior of mice after MCAO. In brief, a white opaque plastic chamber (L44 × W44 × H40 cm, Ugo Basile, cat. no. 47445) was used as an open arena. The chamber was illuminated with a lamp at 65 Lux. Each mouse was allowed a completely spontaneous and uninterrupted locomotion inside the arena within a testing period of 10 min. A camera above the arena was set to record the animal behavior (8 frames per second). The video clips were exported, processed, and analyzed using Fiji software (NIH). The animals’ maximal locomotion speed and presence in the outer zone were used to assess the locomotor dysfunction and anxiety-like behavior, respectively.

Light/Dark Box Test

The light/dark box test was used to assess the unconditioned anxiety of mice after MCAO. A smaller plastic dark-colored chamber (L44 × W22.5 × H38.5 cm) was placed inside the open-field chamber, occupying half of the entire arena. A small hole allowed the mice to cross between light/dark areas without restrictions. The light zone was illuminated with a lamp at 65 Lux. A camera (8 frames per second) above the arena was set to record the animal behavior in the illuminated area during a testing period of 15 min. The video clips were analyzed for the animal presence in the dark side and average latency in the light zone.

Novel Object Recognition Test

The novel object recognition (NOR) test was developed based on the fact that mice have innate preference for novelty. If animals can recognize a familiar object, they will spend more time in exploring the unfamiliar one. The memory retention interval in this study was 24 h. The NOR test consists of three 10 min phases: day 1 for habituation of the arena, day 2 for training with two identical objects (X1 and X2), and day 3 for testing with one familiar object (X1) and one novel object (Y). The novel object recognition test was performed using a white opaque plastic chamber (L44 × W44 × H40 cm) that was illuminated with a lamp at 20 Lux. Two objects were placed along the diagonal and secured using double-sided foam tape. For this study, familiar objects were two identical toy cones, whereas the novel object was a water-filled glass bottle. A camera was set above the arena to record the animal performance during a testing period of 10 min. The video clips were processed and analyzed using Fiji software (NIH). The percentage of time that the animal spent in exploring the object on each position was calculated. It should be noted that the persons who performed all these behavioral tests and analyzed the data were blinded to treatment groups.

Statistical Analysis

All statistical analyses were performed using Prism (GraphPad Software). Data were first subjected to the Shapiro–Wilk test of normality and Levene’s test of homogeneity of variance. The Mann–Whitney U test (two groups) and Kruskal–Wallis test (> two groups) were used for nonparametric tests. Paired/unpaired t-test (two groups) and one/two-way ANOVA (> two groups) were utilized for parametric tests. The outliers were determined by Grubbs’ test. p < 0.05 was considered statistically significant. Data are presented as mean + / ± SEM.

Results

Reactive Microglia and Injured Neurons Mediate Inflammation via mPGES-1

We previously identified a series of phenylsulfonyl hydrazide derivatives as novel selective mPGES-1 inhibitors with moderate potency utilizing high-throughput screening (HTS) [39]. A preliminary hit-to-lead optimization of the best compound from HTS led to the discovery of compound 8n (or MPO-0057: Fig. 1a) as an mPGES-1 inhibitor with nanomolar IC₅₀ values in mouse macrophages [28], which also showed marked inhibition on 6-hydroxydopamine-induced PGE₂ in both mouse and human neuronal cells [40]. Continual efforts in the structure–activity relationship study identified a novel potent phenylsulfonyl hydrazide analog — N-phenyl-N′-(4-benzyloxyphenoxycarbonyl)-4-chlorophenylsulfonyl hydrazide (PBCH, 7d, or MPO-0063) (Fig. 1a) with improved selectivity for mPGES-1 against COX-1 and COX-2 [26]. Discovery of these novel compounds that potently inhibit both human and mouse mPGES-1 allowed us to study the effects of blockade of PGE₂ synthesis in in vitro and in vivo models of inflammation without affecting other COX-derived and physiologically important prostanoids including prostacyclin and thromboxane [27].

Fig. 1.

mPGES-1 is involved in acute neuroinflammation engaging both microglia and neurons. a Chemical structures of our selective mPGES-1 inhibitors MPO-0057, MPO-0063, and MPO-0112. b Mouse primary microglia were treated by selective mPGES-1 inhibitors MPO-0057, MPO-0063, MPO-0112, compound 3 (C3), or COX-2 inhibitor celecoxib (all at 10 µM). About 15 min later, LPS (100 ng/ml) was added to the culture medium. With LPS stimulation overnight, the levels of PGE₂ in the culture medium were measured by ELISA. 100 ng/ml LPS dramatically induced microglia to secrete PGE₂ (****p < 0.0001, Mann–Whitney U test), which was decreased by co-treatment with these tested compounds (****p < 0.0001, Kruskal–Wallis test with post hoc Dunn’s multiple comparisons). c Our current lead mPGES-1 inhibitor MPO-0063 decreased LPS-induced PGE₂ in mouse primary microglia in a concentration-dependent manner with an IC₅₀ value of 27.8 nM, calculated by GraphPad Prism software. d LPS-induced microglia to secrete prototypical inflammatory cytokines including IL-1β, IL-6, and TNF-α (****p < 0.0001, Mann–Whitney U test). Co-treatment with MPO-0063 concentration-dependently blocked the LPS-induced IL-1β (**p = 0.0028; ****p < 0.0001, Kruskal–Wallis test with post hoc Dunn’s multiple comparisons), IL-6 (**p = 0.0073; **p = 0.0012, Kruskal–Wallis test with post hoc Dunn’s multiple comparisons), and TNF-α produced by these mouse microglial cells. e Mouse neuronal cell lines HT-22 and Neuro-2a were subjected to oxygen–glucose deprivation (OGD) for 4.5 h, and the levels of PGE₂ in the culture medium were measured by ELISA. Exposure to OGD increased the PGE₂ produced by both cell lines (HT-22: *p = 0.0188; Neuro-2a: **p = 0.0087, Mann–Whitney U test), which was decreased by co-treatment with either MPO-0063 (HT-22: ***p = 0.0003; Neuro-2a: **p = 0.0043, Mann–Whitney U test) or celecoxib (HT-22: ***p = 0.0003; Neuro-2a: *p = 0.0260, Mann–Whitney U test). All data are presented as mean + / ± SEM (n = 4–8)

To validate their inhibition on mPGES-1, we first treated mouse primary brain microglia with our novel selective mPGES-1 inhibitors MPO-0057, MPO-0063, and MPO-0112 (Fig. 1a). Fifteen minutes later, microglial cells were treated with 100 ng/ml lipopolysaccharide (LPS) for inflammatory induction and microglial activation. After overnight incubation, the levels of PGE₂ and prototypical pro-inflammatory cytokines such as IL-1β, IL-6, and TNF-α in the culture medium were measured by ELISA. Selective COX-2 inhibitor celecoxib and mPGES-1 inhibitor with a different chemical scaffold compound 3 (C3) were also included in the experiment as positive controls for comparisons [41]. Stimulation with 100 ng/ml LPS powerfully induced mouse microglial activation, evidenced by a dramatic increase in PGE₂ in the culture medium, which however was largely prevented by co-treatment with these COX-2 or mPGES-1 inhibitors at 10 µM (Fig. 1b). Among all the tested compounds, MPO-0063 showed the highest potency, as treatment with this mPGES-1 inhibitor nearly completely prevented the LPS-induced PGE₂ in these mouse microglial cells (Fig. 1b). It appeared that MPO-0063 inhibited PGE₂ biosynthesis in a concentration-dependent manner with a calculated IC₅₀ value of 27.8 nM (Fig. 1c), which is similar to what was previously reported in RAW 264.7 monocyte/macrophage-like cells [26] and validated its high potency on mouse mPGES-1.

The mPGES-1-derived PGE₂ functions as an important lipid messenger molecule that can regulate the synthesis and release of pro-inflammatory cytokines such as IL-1β [42], IL-6 [43, 44], and TNF-α [45]. In line, the induction of PGE₂ by LPS treatment in mouse primary brain microglia was accompanied by the elevated secretion of all these three prototypical inflammatory cytokines. We further found that the LPS-induced elevation of pro-inflammatory cytokines was largely blocked by co-treatment with MPO-0063 in a concentration-dependent manner (Fig. 1d). These findings together suggest that mPGES-1 is the major PGES for the biosynthesis of PGE₂ in mouse microglia responding to inflammatory stimuli, and our novel mPGES-1 inhibitors are sufficiently potent to block microglia-mediated induction of pro-inflammatory cytokines.

Though widely used as a stimulatory agent to provoke inflammatory reactions in vitro and in vivo, LPS via activating microglia alone cannot recapitulate the complexity of neuroinflammation-related conditions in which other types of brain cells, particularly neurons and astrocytes, are also engaged. To further evaluate the inhibitory efficacy of our newly developed mPGES-1 inhibitors on PGE₂ biosynthesis in an in vitro condition that might better mimic brain ischemia, we cultured mouse neuronal cell lines HT-22 and Neuro-2a under oxygen–glucose deprivation (OGD) stress. We found that an episode of 4.5 h OGD significantly increased the PGE₂ produced by injured neuronal cells (Fig. 1e). Interestingly, the OGD-induced neuron-derived PGE₂ was markedly prevented by treatment with MPO-0063 or celecoxib (Fig. 1e). These results indicate that both microglia and neurons can be the cellular resources of PGE₂ under adverse conditions, and mPGES-1 inhibition by our compound is able to suppress the induction of PGE₂ derived from either origin.

Transient Focal Brain Ischemia Robustly Induces mPGES-1

The powerful in vitro anti-inflammatory effects from mPGES-1 inhibition encouraged us to examine the potential therapeutic effects of our novel mPGES-1 inhibitors in animal models of inflammation-associated CNS conditions. We chose to study a mouse model of brain ischemia because PGE₂ plays essential roles in ischemic injury–promoted neuroinflammation that can aggravate the secondary damages to the brain [18, 19, 22, 23]. In our model, adult C57BL/6 mice (10–12 weeks old) were subjected to a 60 min episode of MCAO, followed by a 72 h reperfusion. To validate the induction of COX/PGE₂ cascade in this model, we first examined the expression levels of all five enzymes in ischemic mouse brain that are responsible for the two-step biosynthesis of PGE₂: COX-1 and COX-2 for the first step (conversion from arachidonic acid to PGH₂) and mPGES-1, mPGES-2, and cPGES for the second step (from PGH₂ to PGE₂). We found that the mRNA levels of both COX-1 and COX-2 were elevated in the ipsilateral ischemic hemisphere, measured 72 h after acute focal cerebral ischemia by qPCR (Fig. 2a). Among the three terminal enzymes for PGE₂ synthesis, on the other hand, only mPGES-1 but not mPGES-2 or cPGES, showed robust induction by ischemic injuries (Fig. 2b). Very impressively, the induction fold of mPGES-1 in the ischemic brain tissues is about three times higher than that of COXs. These findings in mice align with a recent study in rats showing a very similar high induction of mPGES-1 in the ipsilateral hemisphere following transient focal MCAO for 2 h [46], raising the possibility that mPGES-1 might provide a pharmacological target for anti-inflammatory therapeutics of acute brain ischemia.

Fig. 2.

Transient focal brain ischemia induces enzymes for PGE₂ biosynthesis. Adult male C57BL/6 mice (10–12 weeks old) underwent a 60 min episode of transient middle cerebral artery occlusion (MCAO). Three days later, the expression of enzymes for PGE₂ biosynthesis — two COXs (a) and three PGESs (b) — in the ipsilateral hemisphere was examined by qPCR. Note that the mRNA expression of COX-1 (encoded by Ptgs1), COX-2 (Ptgs2), and mPGES-1 (Ptges), but not that of mPGES-2 (Ptges2) and cPGES (Ptges3), was induced by MCAO (n = 5 for sham group and 11 for MCAO group, **p = 0.0011 for COX-1; **p = 0.0032 for mPGES-1, Mann–Whitney U test). Data are shown as mean ± SEM

Pharmacological Inhibition of mPGES-1 Improves Post-stroke Well-being

We next wanted to determine the feasibility of pharmacological inhibition of mPGES-1 by our novel MPO compounds in this mouse model of transient cerebral ischemia. We elected to use compound MPO-0063 for this in vivo study because, as our current lead mPGES-1 inhibitor, MPO-0063 showed the highest inhibition on PGE₂ production in activated mouse brain microglia among peer compounds (Fig. 1b). Its broad therapeutic benefits observed in rat models of peripheral edema and arthritis demonstrated its powerful anti-inflammatory actions and sufficient pharmacokinetic properties for preclinical studies in rodents [27]. In this proof-of-concept study, adult C57BL/6 mice (10–12 weeks old) were subjected to MCAO for 60 min, followed by 72 h reperfusion, during which mice were intraperitoneally treated by MPO-0063 with increasing doses (0, 5, or 10 mg/kg) at 4.5, 12, and 24 h following MCAO (Fig. 3a). Three days after MCAO, all mice were then euthanized under deep anesthesia for histological and biochemical analyses. This treatment paradigm aimed to block the majority of mPGES-1 activities during the early inflammatory phase of MCAO-induced neuronal injuries and to detect any potential dose-dependent effects of our lead mPGES-1 inhibitor MPO-0063.

Fig. 3.

Post-stroke inhibition of mPGES-1 improves well-being after transient MCAO. a Adult male C57BL/6 mice (10–12 weeks old) were subjected to transient MCAO for 60 min, followed by 72 h reperfusion. The animals were treated by vehicle or selective mPGES-1 inhibitor MPO-0063 (5 or 10 mg/kg, i.p.) at 4.5, 12, and 24 h after MCAO. Three days after MCAO, all mice were euthanized for histological and biochemical analyses. b Treatment of naïve male mice with MPO-0063 (10 mg/kg, i.p.) for 3 times within 24 h did not alter the counts of immune cells. c The real-time post-stroke survival rates were plotted and compared between vehicle and mPGES-1 inhibitor-treated animals using Kaplan–Meier survival analysis with post hoc log-rank test (*p = 0.0492). d The percentages of mice that survived at 12, 24, 36, 48, 60, and 72 h after MCAO were compared between the vehicle only (0 mg/kg) and MPO-0063 (10 mg/kg)-treated mice (**p = 0.0031, paired t-test). (e) The post-stroke weight loss of animals was measured 24 h after MCAO (*p = 0.0123 for 10 mg/kg MPO-0063 group compared to vehicle only group, Kruskal–Wallis test with post hoc Dunn’s multiple comparisons). Data are visualized by the violin plot. f The modified Bederson scale (0–5) was performed at 24, 48, and 72 h after MCAO to assess progressive neurological deficits in mice. The overall outcome for each animal was defined by its averaged score over the 72 h observation period. Results are visualized by the violin plot. Treatment with MPO-0063 at 10 mg/kg alleviated the neurological deficits (*p = 0.0303 compared to the vehicle group, Kruskal–Wallis test with post hoc Dunn’s multiple comparisons)

Strokes are well known to cause immunosuppression [47], raising a concern that mPGES-1 inhibition might increase post-stroke infection in this model given the essential roles of PGE₂ in the regulation of immune responses [48]. Thus, it is important to examine the effects of our lead compound MPO-0063 on immune cells in the mouse whole blood. We found that treatment with MPO-0063 (10 mg/kg, i.p.) for 3 doses within 24 h in mice had no effect on the numbers of total white blood cells, lymphocytes, neutrophils, and other immune cells (Fig. 3b). Thus, it is unlikely that mPGES-1 inhibition by our lead compound with the proposed dosing regimen in Fig. 3a would increase the risk of post-stroke infection and lead to peripheral complications. Indeed, we did not observe any signs of infection throughout all animal experiments in this study.

We further found that the 60 min MCAO in mice led to marked mortality that escalated during the 72 h period of reperfusion (Fig. 3c). About one-third of vehicle-treated mice deceased by 24 h after MCAO, and only one-third of these animals eventually survived the 72 h reperfusion (Fig. 3d). Conversely, treatment with mPGES-1 for only three doses overall improved the post-stroke survival rates (Fig. 3c). Particularly, the survival rates were substantially increased by treatment with the higher systemic dose of MPO-0063 (10 mg/kg, i.p.), when compared to treatment with vehicle only (75.0% vs. 33.3% at day 3 post-MCAO) (Fig. 3d). In the first 24 h after MCAO, mice lost up to about 15% of normal body weight, which was largely prevented by post-MCAO treatment with our mPGES-1 inhibitor MPO-0063 in a dose-dependent manner (Fig. 3e). To assess the neurological impairments caused by MCAO-induced ischemic injuries in these mice, we performed a 72 h global neurological assessment using a modified Bederson scale (Table 1). A 60 min episode of MCAO followed by reperfusion was sufficient to cause substantial neurological deficits, demonstrated by typical score-4 deficits such as seizure-like abrupt body spinning (the “barrel rolling” behavior) and the transient loss of consciousness in vehicle-treated mice (Fig. 3f). However, the MCAO-triggered neurological impairments were dose-dependently reduced by treatment with mPGES-1 inhibitor, and such seizure-like behaviors rarely occurred in mice treated by 10 mg/kg MPO-0063 (Fig. 3f). Given these interesting findings and the involvement of mPGES-1 in experimental epilepsy [49–51], it is likely that mPGES-1 might contribute to cerebral ischemia-triggered seizures. These reductions in post-stroke mortality, weight loss, and neurological impairments by treatment with our lead compound MPO-0063 demonstrate that mPGES-1 inhibition provides a strategy to improve the overall well-being of mice after transient focal brain ischemia.

Post-stroke Inhibition of mPGES-1 is Neuroprotective

We next examined the ischemia/reperfusion-triggered brain infarction in mice using triphenyl tetrazolium chloride (TTC) staining. Three days after MCAO, mice were euthanized, brain tissues were harvested, and 1-mm coronal sections were prepared for TTC staining. The infarct size and brain edema volume in the ipsilateral hemisphere were quantified and adjusted according to the contralateral hemisphere. We found that the transient MCAO for 60 min caused marked brain lesions that were mainly located in the striatum and associated cortex (Fig. 4a), which on average involved nearly 55% area of the ipsilateral hemisphere (Fig. 4b) and led to up to 15% increase in brain swelling (Fig. 4c). Post-stroke treatment with MPO-0063 for only three doses (at 4.5, 12, and 24 h after MCAO, respectively) considerably decreased both transient focal brain ischemia-induced acute infarction and cerebral edema in a dose-dependent manner (Fig. 4b, c). These findings strongly suggest that mPGES-1-derived PGE₂ contributes to the secondary damages to the ischemic brain and that the post-stroke mPGES-1 inhibition by our lead compound MPO-0063 provides a novel neuroprotective strategy against ischemic injuries.

Fig. 4.

Post-stroke mPGES-1 inhibition is neuroprotective after transient MCAO. a Triphenyltetrazolium chloride (TTC) staining was performed to detect the brain infarct in mice 72 h after MCAO. Representative images of each group are displayed. Note that the viable brain parenchyma looked reddish, whereas the infarcted areas appeared pale. b Treatment with MPO-0063 reduced the adjusted infarct size, measured 3 days after transient MCAO (*p = 0.0288 for 5 mg/kg group and *p = 0.016 for 10 mg/kg group compared to the vehicle group, Kruskal–Wallis test with post hoc Dunn’s multiple comparisons). c Treatment with MPO-0063 attenuated the transient ischemia-induced acute brain swelling in the ipsilateral hemisphere (*p = 0.0215 for 5 mg/kg group and **p = 0.0024 for 10 mg/kg group compared to the vehicle group, Kruskal–Wallis test with post hoc Dunn’s multiple comparisons). All data are shown as mean ± SEM

Elevated mPGES-1 Contributes to Cytokine Surge in Ischemic Brain

Cytokines typically are small pleiotropic polypeptides that are mostly undetectable in normal naïve brain with their receptors constitutively expressed at relatively low levels. However, upon various brain injuries, the expression of inflammatory cytokines is quickly upregulated in damaged neurons and activated glial cells [5, 52]. Among the major pro-inflammatory cytokines that are commonly induced by ischemic stroke and in turn can aggravate inflammatory responses are IL-1β, IL-6, and TNF-α [53, 54]. Given their significance as the major inflammatory markers and mediators as well as their leading roles in the secondary neuronal injuries, we next investigated the effects of post-stroke inhibition of the mPGES-1 by MPO-0063 (10 mg/kg, i.p.) on the levels of these three prototypical pro-inflammatory in the injured brain areas. We focused on the ischemic tissues located at 1.5–0.5 mm anterior to bregma point for cytokine expression because these brain regions are mainly supplied by the microvasculature derived from MCA and, thus, severe infarction is typically developed in these brain areas by transient MCAO. The mRNA expression of cytokines and mPGES-1 was measured by qPCR; protein levels of cytokines were quantified by ELISA. PGE₂ levels in the ischemic tissues were also determined by ELISA with samples being processed as previously described [55]. Our results showed that the expression of all these three cytokines was dramatically increased by ischemic injuries at both mRNA levels (Fig. 5a) and protein levels (Fig. 5b). However, this brain cytokine surge triggered by acute cerebral ischemia was largely prevented by systemic treatment with our mPGES-1 inhibitor MPO-0063 (Fig. 5a, b). The adequate inhibition of mPGES-1 by MPO-0063 was validated by a significant reduction in PGE₂ levels in ischemic brain areas (Fig. 5b). Interestingly, mPGES-1 inhibition also led to a substantial decrease in mPGES-1 mRNA expression in the same lesion areas (Fig. 5a), indicative of a positive feedback mechanism whereby PGE₂ signaling augments the expression of mPGES-1 itself.

Fig. 5.

Inhibition of mPGES-1 diminishes cytokine surge and PGE₂ induction in ischemic brain. Three days after MCAO in mice, ipsilateral brain tissues were collected from infarcted brain areas (1.5–0.5 mm anterior to bregma) and analyzed by qPCR and ELISA. Mann–Whitney U test was used for statistical comparisons (n = 4–6 for sham + vehicle group, n = 6–9 for MCAO + vehicle group, and n = 6–10 for MCAO + 10 mg/kg MPO-0063 group). a The mRNA expression levels of pro-inflammatory cytokines (IL-1β, IL-6, and TNF-α) and mPGES-1 were consistently induced in the infarcted brain areas (**p = 0.0025 for IL-1β; *p = 0.0121 for IL-6; **p = 0.0025 for TNF-α; *p = 0.0177 for mPGES-1 compared to sham), which were decreased by treatment with the mPGES-1 inhibitor (*p = 0.0373 for mPGES-1). b The levels of cytokine proteins (IL-1β, IL-6, and TNF-α) and PGE₂ were significantly increased in the infarcted brain tissues (*p = 0.0426 for IL-1β; *p = 0.012 for IL-6; *p = 0.0256 for TNF-α; *p = 0.0152 for PGE₂ compared to sham), which were decreased by treatment with MPO-0063 (**p = 0.0037 for IL-1β; *p = 0.0496 for PGE₂). Prostacyclin PGI₂ was increased, but prostaglandin PGF_2α was decreased in ischemic brain tissues 3 days after MCAO in mice (*p = 0.0221 for PGI₂; ***p = 0.0007 for PGF_2α). Treatment with MPO-0063 did not alter either PGI₂ or PGF_2α levels in these brain tissues when compared to vehicle treatment. Data are presented by boxplot with whiskers ranging from minimum to maximum

We previously reported that long-term treatment with our mPGES-1 inhibitor MPO-0063 (PBCH) in rats reduced the plasma levels of PGE₂ but had no effects on other COX-derived products, such as prostaglandin I2 (PGI₂ or prostacyclin) and thromboxane A2 (TXA₂) [27]. Conversely, in the same study, treatment with COX inhibitor indomethacin decreased all examined prostanoids including PGE₂, PGI₂, and TXA₂ [27], demonstrating broad, non-selective effects. To examine the on-target selectivity of our lead compound MPO-0063 in mice after MCAO, we measured the ipsilateral levels of PGI₂ and PGF_2α using ELISA. Intriguingly, the brain ischemia caused a marked increase in PGI₂ but a considerable reduction of PGF_2α in the same ischemic brain areas (Fig. 5b), revealing dichotomous effects of ischemic injury on different types of prostaglandins. It appears that the treatment with MPO-0063 had no effects on either PGI₂ or PGF_2α in ischemic brain (Fig. 5b). These findings together reveal that our lead compound MPO-0063 selectively acts on PGE₂ biosynthesis after ischemic stroke. The lack of effects on other prostanoids suggests that mPGES-1 inhibition unlikely causes an imbalance between PGI₂ and TXA₂, which is thought to underlie the pathophysiological mechanisms for the side effects of COX inhibition in the cerebrovascular and cardiovascular systems [13, 15].

Post-stroke mPGES-1 Inhibition Improves Locomotor Function and is Anxiolytic

Depending on the locations of lesions within the brain, ischemic stroke can cause a variety of behavioral morbidities, such as locomotor dysfunctions, fear, anxiety, cognitive deficits, etc., in both human patients and experimental rodents [56–59]. Therefore, we next wanted to know whether the neuroprotective and anti-inflammatory benefits of mPGES-1 inhibition observed in the ischemic brain can be translated to any improvements in these behavioral impairments. To determine this possibility, adult C57BL/6 mice were subjected to a 30 min episode of MCAO, which was followed by reperfusion and then intraperitoneal treatment of mPGES-1 inhibitor MPO-0063 (10 mg/kg) or vehicle at 4.5, 12, and 24 h after MCAO. One week after MCAO, mice were subjected to a panel of behavioral tests, i.e., the open field test, light/dark box test, and novel object recognition test, for behavioral assessments. About 30 days after MCAO, these behavioral tests were performed again to evaluate the long-term behavioral alterations in these mice (Fig. 6a). After transient focal brain ischemia, mice suffered a continual loss of body weight and did not begin to recover until 5 days after MCAO (Fig. 6b). In alignment with the finding that mPGES-1 inhibition largely prevented the post-stroke weight loss (Fig. 3e), treatment with MPO-0063 facilitated the post-stroke recovery, demonstrated by the animals’ full regain of their normal body weights in 5 days after MCAO (Fig. 6b).

Fig. 6.

Inhibition of mPGES-1 reduces functional deficits after transient brain ischemia. a Adult male C57BL/6 mice (10–12 weeks old) were subjected to a 30 min episode of transient MCAO followed by reperfusion. The animals were treated by vehicle or mPGES-1 inhibitor MPO-0063 (10 mg/kg, i.p.) at 4.5, 12, and 24 h after MCAO (n = 12–13). A battery of behavioral tests including open field, light/dark box, and novel object recognition were performed 7 and 30 days after MCAO. b Body weights of animals were measured daily within the first week and at 30 days after MCAO. Note that the treatment with MPO-0063 facilitated the animals to regain their normal body weights (two-way ANOVA: ***p = 0.0001, post hoc Šidák’s multiple comparisons test: *p = 0.0471 on day 2; *p = 0.0343 on day 5; *p = 0.0290 on day 6). c The maximal locomotion speed during a 10 min open field test was quantified by analyzing the video clips. The locomotion velocity significantly declined on day 7 post-MCAO (two-way ANOVA and post hoc Šidák’s test: *p = 0.0271) and returned to the baseline by 1 month. The early administration of MPO-0063 prevented the decline of top speed observed on day 7 (two-way ANOVA and post hoc Tukey’s test: *p = 0.0259). d The animal presence along the side walls of open field during a 10 min open field test was quantified to indicate anxiety-like behavior. Transient brain ischemia increased the presence of mice along the side walls over a 30-day period of testing after MCAO (two-way ANOVA and post hoc Šidák’s test: ***p = 0.001 compared to pre-MCAO), which was reduced by systemic treatment with MPO-0063 (two-way ANOVA: **p = 0.0025; post hoc Tukey’s test: *p = 0.0289 compared to vehicle-treated mice). e The light/dark preference of animals was quantified by the percentage of time spent in the dark side of the box during a 15 min test. Note that a 30 min episode of MCAO did not alter the light/dark preference for the vehicle-treated mice. However, MPO-0063-treated mice spent less time in the dark side than in the light area on day 7 post-MCAO (two-way ANOVA and post hoc Šidák’s test: ***p = 0.0004 compared to pre-MCAO; two-way ANOVA and post hoc Tukey’s test: ***p = 0.0011 compared to vehicle-treated mice). f The average latencies of mice in the light side before they switched to the dark area. Note that a 30 min episode of MCAO did not alter the light latency for the vehicle-treated mice. However, treatment with MPO-0063 prolonged the light latency on day 7 post-MCAO (two-way ANOVA and post hoc Šidák’s test: ***p = 0.0001 compared to pre-MCAO; two-way ANOVA and post hoc Tukey’s test: *p = 0.0399 compared to vehicle-treated mice). All data are presented as mean ± SEM

The open field test has been commonly used to measure general locomotor function and anxiety-like behavior in experimental rodents [60]. We first focused on the maximum movement speed of mice during a 10 min test as an indicator of their locomotor functions. As anticipated, the maximum movement speed of vehicle-treated mice significantly declined during the first week after MCAO and gradually returned to baseline by day 30 after MCAO (Fig. 6c), implicating a locomotor dysfunction caused by ischemic brain injuries. On the contrary, mice treated by mPGES-1 inhibitor MPO-0063 did not experience visible locomotor dysfunction during the same testing period, demonstrated by their consistent maximum movement speed before and after MCAO (Fig. 6d). We further found that the transient brain ischemia markedly increased the presence of mice along the walls of open field maze, indicative of an anxiety-like behavior that lasted over 30 days after MCAO (Fig. 6d). However, inhibition of the mPGES-1 lessened the brain ischemia-associated anxiety, evidenced by the decreased presence of MPO-0063-treated mice in the margin areas of the open field, particularly on day 7 post-MCAO, when compared to their vehicle-treated peers (Fig. 6d).

In the 15 min light/dark box test, the brain ischemia did not obviously alter the presence of mice in the dark side of the box (Fig. 6e). Nor did it palpably change the latencies of mice in the light side before they switched to the dark area (Fig. 6f). These results suggest that moderate to mild focal brain ischemia by an episode of 30 min MCAO was not sufficient to cause an alternation in anxiety-like behavior that can be detected by the light/dark box test. However, the post-stroke administration of MPO-0063 rendered the mice to spend less time in the dark side of the box, measured at day 7 post-MCAO (Fig. 6e). In line, mPGES-1 inhibitor-treated mice showed increased latencies in the light side when compared to the vehicle-treated cohorts at day 7 after the focal brain ischemia (Fig. 6f). This decreased preference of animals in the dark areas together with the less time spent in the margin areas of the open field observed in MPO-0063-treated mice at day 7 after MCAO, though transient, indicates an anxiolytic effect of mPGES-1 inhibition following brain ischemia.

Inhibiting mPGES-1 Diminishes Cognitive Deficits Triggered by Ischemic Injuries

Current evidence suggests that up to 30% of human patients with ischemic stroke can develop long-term cognitive impairments [61], which can be well recapitulated in rodent models of ischemic stroke such as transient MCAO in mice [56, 62]. The novel object recognition (NOR) is a commonly used cognition test to evaluate cognitive functions such as learning and memory in rodent models of stroke [63, 64]. In this study, the NOR test was performed on mice before MCAO and 30 days after MCAO in an open field arena with objects that are only different in shape and appearance (Fig. 7a). The task consisted of three 10 min sessions: day 1 for the animals to habituate the arena, day 2 for the training of mice with two identical objects (X1 and X2), and day 3 for testing with a familiar object (X1) and a novel object (Y) (Fig. 7a). In the training sessions, none of the mice on average showed any preference toward the two identical objects (X1 and X2) (Fig. 7b, c). On the testing days, naïve mice showed significantly more interest in the novel substitute object (Fig. 7b, c). However, with brain ischemia induced by MCAO, mice were unable to identify the novel object (Y) at day 30 post-MCAO. Conversely, treatment with mPGES-1 inhibitor MPO-0063 restored the animals’ capability of differentiating the novel object (Y) from the familiar objects (X2) (Fig. 7b, c). Likewise, naïve mice and MPO-0063-treated MCAO mice, but not vehicle-treated MCAO cohorts, demonstrated considerable preference toward the novel object (Y) over the familiar object (X1) on the testing days (Fig. 7d). These findings reveal that the post-stroke treatment with our lead compound MPO-0063 could restore the animals’ cognition of learning and memory after ischemic stroke. Therefore, mPGES-1 inhibition might represent a potential strategy to relieve ischemic injuries-associated cognitive deficits.

Fig. 7.

Inhibition of mPGES-1 alleviates transient brain ischemia-induced cognitive decline. a Novel object recognition (NOR) test was performed before and 30 days after transient MCAO for 30 min. The NOR task consisted of three 10 min sessions: on day 1, mice were allowed to habituate the arena; on day 2, mice were trained with two identical objects (X1 and X2); on day 3, mice were tested with a familiar object (X1) and a novel object (Y). The specific time of animals spent on the novel object (Y) or the familiar objects (X1 and X2) during the 10 min training/testing was used to indicate the cognition, particularly recognition memory. b The 2D spatial distribution frequencies of representative mice were visualized on the heatmaps that were merged with live scenes. c Percentages of time that animals spent on a familiar object (X2) during training sessions and a novel object (Y) during testing sessions were quantified and compared (t-test: p = 0.0078, 0.1484, and 0.0485 for naïve mice, vehicle-treated MCAO mice, and MPO-0063-treated MCAO mice). d Percentages of time that mice spent on a familiar object (X1) and a novel object (Y) during 10 min testing sessions were quantified and compared (t-test: p = 0.0039, 0.0781, and 0.0313 for naïve mice, vehicle-treated MCAO mice, and MPO-0063-treated MCAO mice). Data are presented as mean + SEM (n = 12–13)

Discussion

As the inducible form of terminal enzyme for PGE₂ synthesis, mPGES-1 has broad implications in acute neuroinflammation-associated conditions such as ischemic stroke. Genetic ablation of mPGES-1 in mice reduces infarction, edema, neuronal apoptosis, and behavioral neurological dysfunctions after transient brain ischemia [22]. However, up to date, mPGES-1 as a feasible therapeutic target for ischemic stroke has not been demonstrated by a proof-of-concept study using a pharmacological strategy due to the lack of drug-like compounds that can inhibit both human and rodent mPGES-1 [24, 25]. In the present study, we evaluated the in vitro and in vivo effects of our novel selective mPGES-1 inhibitors that we recently developed from HTS and follow-up hit-to-lead optimization [26, 28, 39]. These new series of compounds showed powerful counteractions against the reactive microglia-mediated neuroinflammation. As our current lead mPGES-1 inhibitor, MPO-0063, when systemically administered several hours after ischemia–reperfusion in mice, substantially improved post-stroke well-being and reduced infarction, edema, cytokine storm, locomotor dysfunction, and anxiety-like behavior. Importantly, these immediate benefits were followed by the longer-term improvement in cognitive functions. These findings, for the first time, provide direct pharmacological evidence that mPGES-1 represents a feasible anti-inflammatory therapeutic target for transient brain ischemia. Given that the large vessel occlusion is often permanent in most patients with ischemic stroke [65], it is important to evaluate the effects of mPGES-1 inhibition by our compounds in a mouse model of permanent MCAO, which is recommended by the initial Stroke Therapy Academic Industry Roundtable (STAIR) as the primary model for small and large animals [66].

As a rate-limiting enzyme for the biosynthesis of PGE₂ and other prostanoids, COX-2 was once considered a popular anti-inflammatory target for ischemic stroke and many other brain inflammation-associated conditions including epilepsy, brain cancer, and neurodegeneration [16, 19, 67–70]. Dynamically regulated by neuronal activities, COX-2 can be quickly and vigorously induced within the brain by cerebral ischemia [8, 9], and in turn contributes to the delayed secondary neuronal injury via producing excessive PGE₂ [8–12]. However, mounting clinical evidence in the past two decades points to a strong positive connection between chronic consumption of COX-2 inhibitor drugs, such as valdecoxib, rofecoxib, and celecoxib, and severe side effects including heart attack and stroke. The adversity of COX-2 inhibition indicates that some of its downstream prostanoid pathways (e.g., the one mediated by prostacyclin) may play protective roles in the microvascular systems [13–15]. Therefore, targeting PGE₂ synthases or receptors might provide alternative anti-inflammatory strategies for the delayed treatment of brain ischemia with more specificity than the general blockade of the entire COX cascade.

COX-2 and mPGES-1 are co-localized within the brain and can be co-induced in the infarct regions by acute cerebral ischemia [23]. These two enzymes are thought to act together to aggravate ischemic injuries by generating excessive PGE₂. COX-2 inhibitor NS-398 reduced ischemic injuries in wild-type mice, but not in mice lacking mPGES-1 [23], suggesting that the COX-2-mediated detrimental effects in ischemic brain should mainly be attributed to PGE₂ that is directly synthesized by mPGES-1. The results from this early study using a genetic strategy were completely validated by pharmacological outcomes in the present study. These findings together strongly raise a notion that mPGES-1 represents an alternative to COX-2 as a feasible anti-inflammatory and neuroprotective target for new treatment of brain ischemia without affecting other prostanoids that may benefit the post-stroke recovery [19].

As a major product of the COX-2/mPGES-1 enzymatic complex in diseased brains, PGE₂ plays central roles in mediating the deleterious effects of co-induced COX-2 and mPGES-1 in various neurological conditions including ischemic stroke [16, 18]. PGE₂ exerts pathophysiological effects via signaling through a group of four G protein-coupled EP receptors (EP1, EP2, PE3, and EP4) that have divergent engagement in phosphoinositol and cAMP turnover [17, 71–74]. These cell membrane-bound receptors are differentially expressed on neuronal and glial cells throughout the brain to mediate a variety of physiological and pathological functions. Given that targeting PGE₂-mediated signaling pathways might provide more therapeutic specificity than targeting COX-2 itself, these four EP receptors have been under intensive investigation as potential new targets for ischemic stroke in several preclinical models [18]. Intriguingly, all these PGE₂ receptors are involved in the regulation of post-stroke brain inflammation and injuries. Overall, the activation of EP1, EP2, and EP3 receptors by PGE₂ has been found to exacerbate ischemic injuries [34, 75–80], whereas the EP4-mediated PGE₂ signaling exerts neuronal and vascular protection following cerebral ischemia [81–83]. Therefore, modulating one specific EP receptor likely will leave therapeutic space for another. Selectively targeting mPGES-1 by drug-like small-molecule inhibitors such as MPO-0063 would prevent detrimental effects mediated by all these EP receptors without affecting other COX-2-derived prostanoids, such as prostacyclin and thromboxane that play normal physiological roles essential to the healthy vascular systems. Therefore, the inducibility and unique position of mPGES-1 in the COX cascade entitles the enzyme druggability with an ideal therapeutic balance between efficacy and specificity.

The human mPGES-1 was first reported as p53-induced gene 12 (PIG12) in 1997 [84], then cloned and characterized in 1999 [85]. The rat and mouse forms of mPGES-1 were described 1 year later [86]. Since then, mPGES-1 has widely been considered as a promising target for anti-inflammatory therapeutics, leading to the development of a myriad of small-molecule inhibitors targeting this enzyme [24, 25]. Unfortunately, most of these compounds inhibit human but not murine mPGES-1 due to the interspecies discrepancy of the enzyme. Particularly, three amino acids (T131, L135, and A138) in human mPGES-1 that play an essential role as gatekeepers in the active site of the enzyme are replaced in rodent mPGES-1 [87, 88]. Among potent and well-characterized mPGES-1 inhibitors that show cross-species activity are C3 and our former lead compound MPO-0057 (8n) [25]. Compared to these two, our current lead mPGES-1 inhibitor MPO-0063 showed higher inhibition on PGE₂ production in activated mouse primary brain microglia, rendering it a favorable pharmacological tool to study mPGES-1 in rodent models of CNS conditions, where neuroinflammation engaging activated microglia is a disease-driving force. As such, post-stroke inhibition of mPGES-1 by MPO-0063 showed broad promising therapeutic effects in a mouse model of transient brain ischemia, ranging from immediate neuroprotection, survival, and body weight recovery to long-term cognitive improvement.

In conclusion, the current study strongly suggests, for the first time, that the pharmacological inhibition of the inducible terminal enzyme for PGE₂ biosynthesis mitigates ischemic brain injuries and the consequent behavioral impairments. The powerful anti-inflammatory and neuroprotective actions of drug-like compound MPO-0063 reinforce a notion that mPGES-1 represents an alternative to COX-2 as a feasible and more specific target for delayed adjunct treatment — along with reperfusion therapies — for acute brain ischemia. Its lack of effects on immune cells suggests the unlikelihood that inhibition of mPGES-1 would lead to immunodepression or post-stroke infection. Unlike the COX inhibition, mPGES-1 blockade appears not to alter the levels of prostacyclins and thromboxanes, reducing the potential adverse effects on microvascular systems. Therefore, our proof-of-concept study well positions our current lead mPGES-1 inhibitor, MPO-0063, as a prime candidate for progression to more intensive preclinical and clinical studies.

Supplementary Information

Below is the link to the electronic supplementary material.

Funding

This work was supported by the National Institutes of Health (NIH)/National Institute of Neurological Disorders and Stroke (NINDS) grants R00NS082379 (to J.J.), R01NS100947 (to J.J.), R21NS109687 (to J.J.), R61NS124923 (to J.J.), R01NS105787 (to J.H.), and the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIT) (No. 2020R1A5A2019413 to J.Y.L.).