Highly Selective Cyclooxygenase-1 Inhibitors P6 and Mofezolac Counteract Inflammatory State both In Vitro and In Vivo Models of Neuroinflammation.

Activated microglia secrete an array of pro-inflammatory factors, such as prostaglandins, whose accumulation contributes to neuronal damages. Prostaglandin endoperoxide synthases or cyclooxygenases (COX-1 and COX-2), which play a critical role in the inflammation, are the pharmacological targets of non-steroidal anti-inflammatory drugs, used to treat pain and inflammation. Since it was reported that COX-1 is the major player in mediating the brain inflammatory response, the aim of this study was to evaluate the effects of highly selective COX-1 inhibitors, such as P6 and mofezolac, in neuroinflammation models. Lipopolysaccharide (LPS)-activated mouse BV-2 microglial cells and LPS intracerebroventricular-injected mice as in vitro and in vivo neuroinflammation models, respectively, were used to probe the antiinflammatory efficacy of P6 and mofezolac. Both P6 and mofezolac reduce COX-1 expression in LPS-activated BV-2 cells. This reduction was accompanied with PGE2 release reduction and NF-kB activation downregulation. Coextensively, in the in vivo model, both glial fibrillary acidic protein and ionized calcium-binding adapter molecule-1 expression, two markers of inflammation, were reduced by mofezolac to a rank depending on the encephalon area analyzed. The increase of COX-1 expression observed in all the brain sections of LPS-treated mice was selectively downregulated by the in vivo treatment with mofezolac as well as PGE2 release and Ikβα phosphorylation amount assayed in the brain areas tested. These results indicate the capability of P6 and mofezolac to modulate the NF-kB signaling pathway, emphasizing the neuroprotective effect and therapeutic potential of COX-1 inhibitors in the control of neuroinflammatory diseases.

Microglia are recognized as the innate immune cells of the CNS, where they mediate a number of tissue homeostatic functions, including immune surveillance (2), synaptic regulation (3,4), and neurogenesis (5,6). Microglial activation and chronic inflammation thereafter is the starting point for elevated levels of a wide array of potentially neurotoxic molecules including proinflammatory cytokines, proteinases, and reactive oxygen species (ROS), which are believed to contribute to neurodegenerative processes (7)(8)(9).
Epidemiological data-based link between neuroinflammation and neurodegenerative diseases increased the worldwide scientific interest aimed to determine whether reducing inflammation would reverse neurodegeneration. Such data also indicate an inverse relationship between the use of traditional non-steroidal anti-inflammatory drugs (tNSAIDs) and Alzheimer's disease risk. tNSAIDs pharmacological action is due to their ability to inhibit the cyclooxygenase (COX) and, hence, the biosynthesis of the prostaglandins (PG) involved in neuroinflammation (10).
Two COXs are known, COX-1 and COX-2. Upon inflammatory stimuli, COX-1, being constitutive in microglia, is responsible of the primary inflammatory response by inducing the production of PG, mainly PGE2. COX-2 is responsible, upon its induction, of a later and secondary response, with the exception of conditions in which the neurons are directly challenged (excitotoxicity and ischemia) (11). In the latter case, the primary response is COX-2 mediated, and microglia derived COX-1 succeeds as a secondary response upon perpetuating stimuli. In the first circumstances, selective COX-1 catalytic activity inhibition may result beneficial. tNSAIDs inhibit COXs not selectively. Then, drugs useful to treat the different neuroinflammatory conditions are needed (12,13).
Preliminarily, we accomplished an in vitro study using some selective COX-1 inhibitors (P6, P10, and SC-560) compared to aspirin and two selective COX-2 inhibitors (celecoxib and etoricoxib) in a in vitro model of neuroinflammation, represented by lipopolysaccharide (LPS)-stimulated N13 microglial cell line (14). In such a context, the treatment of LPS-stimulated microglial cells by selective COX-1 inhibitors P6 and P10 were able to downregulate COX-1 protein expression without affecting COX-2 levels; interfering with NF-kB activation (14).
In continuation of our investigations, herein, we report the validation of the study above mentioned by using both an in vitro model of neuroinflammation, represented by LPS-activated BV2 cell line, and an in vivo animal model, constituted by LPS-treated mice.
In particular, in this study, COX-1 role in neuroinflammation was explored by using P6 and mofezolac (Figure 1). Both compounds belong to the diarylisoxazole class, and mofezolac is the only clinically used diarylisoxazole approved in Japan as Disopain ® .
All the other reagents and solvents were purchased from Sigma-Aldrich (Milan, Italy) and used without any further purification.
Then, microglial cells were plated, at a density of 25 × 10 4 / well in 6-well plates (Falcon) and treated with the chosen COX inhibitors (P6 and mofezolac) when they reached 80% confluence. Preliminary experiments were conducted to establish the optimal concentration and exposure times necessary for LPS (1 µg/mL) treatment, which were found to be in accordance with other reports (17,18), as well as to establish the optimal dose of the COX inhibitors and exposure times to detect their effects on LPS-stimulated BV2 microglial cell function. In this regard, a set of experiments was carried out in which microglial cells were 1 h pretreated with the selective COX-1 inhibitors P6 (0.5 and 1.0 µM) or mofezolac (0.1 and 0.5 µM). Cells were then incubated for different times at 37°C with LPS, as pro-inflammatory stimulus. Experiments included cells grown in medium alone (control).

cell Viability Test
Cell viability of microglial cells was quantified using the MTT reduction assay. The cells (8 × 10 3 /well) were grown in 96-well plates (Becton Dickinson Labware) in complete medium and treated with different concentration of COX inhibitors, in presence or absence of LPS. Untreated cells were used as a control. A PBS 1× solution of MTT (5 mg/mL) was prepared and added to the cell medium at a final concentration of 0.5 mg/mL. Cells were incubated for 4 h at 37°C and 5% CO2 to allow the MTT metabolism. The formazan crystals formed (from MTT) into the cells were solubilized with DMSO (Sigma-Aldrich). The levels of MTT formazan were determined measuring the optical density at λ = 560 nm and subtracting the background (λ = 670 nm) with a Victor Multiplate Reader (Wallac). Optical density was directly correlated to cell quantity.
After electrophoresis, the resolved proteins were transferred from the gel to nitrocellulose membranes. A blotting buffer [20 mM Tris/150 mM glycine, pH = 8.0, and 20% (v/v) methanol] was used for gel and membrane saturation and blotting. Then, membranes were incubated in the dark with the following specific primary Abs reported in reagents section for 60 min at room temperature. The membranes were washed with 0.1% Tween 20-PBS (for 20 min, three times) and then incubated with the secondary Ab diluted 1:2,000 for 60 min. Bands were visualized by chemiluminescence detection (Invitrogen, Milan, Italy). The β-actin level was used as a protein loading control. For tissue analysis, obtained bands were normalized to the level of β-actin performed for each cerebral area tested. For cell cultures, the bands were normalized to the β-actin level of each experimental condition. The bands obtained after immunoblotting were submitted to densitometric analysis using 1D Image Analysis Software (Kodak Digital Science). Results were expressed as arbitrary units.

animals and Treatment Protocols
This study was carried out in strict accordance with the European Council Directive 86/609/EEC and the Italian animal welfare legislation (art. 4 and 5 of D.L. 116/92). Seventy adult male 129S2/Sv mice (22-24 g body mass, 8-10 weeks of age) were purchased from Harlan-Italy and were kept under environmentally controlled conditions (20 ± 2°C, 50-80% humidity, 12 h light/dark cycle, food and water ad libitum) and divided into 7 groups of 10 animals (5 for western blotting and 5 for immunohistochemical procedures).
Mice received either the selective COX-1 inhibitor mofezolac (

immunohistochemical staining
Mice were transcardially perfused with tris-buffered saline (pH = 7.6) followed by 4% paraformaldehyde in PBS pH = 7.4 at 4°C. Brains were subsequently postfixed in the same fixative, paraffin embedded, and 10 µm slices were obtained with a rotative microtome (Leica, Milan, Italy). Immunohistochemistry was performed following a standard avidin-biotin complex procedure. Briefly, specimens were incubated with mouse primary mAb anti-GFAP at a ratio of 1:1,000 (Merck Millipore, Milan, Italy), or a mouse mAb anti-Iba-1 at a ratio of 1:500 (Merck Millipore) overnight at 4°C and then with an anti-mouse biotinylated secondary Ab (Dako, Milan, Italy), at a 1:1,000 dilution for 1 h at room temperature. The antigen-Ab complexes were visualized by sections incubation for 1 h with extravidin peroxidase (Sigma-Aldrich) diluted 1:1,500 and 3,3′-diaminobenzidine oxidation in the presence of H2O2.

Pge 2 assay
Microglial cells were cultured in 6-well plates at a density of 3 × 10 6 cells/well. Then, the cells were pretreated with selective COX-1 inhibitors P6 or mofezolac for 1 h and, subsequently stimulated with LPS (1 µg/mL). The cultures were maintained at 37°C for 24 and 48 h in a humidified air containing a 5% CO2. PGE2 levels were determined in the supernatant using a competitive binding immunoassay (Cayman Chemical, Ann Arbor, MI, USA) following the manufacturer's instructions. Unstimulated cells were included as a control. PGE2 amount determination in the brain was performed in the tissue extracts, according to the manufacturer's instructions. The optical density was measured at λ = 405-420 nm with precision microplate reader and the amount of PGE2 (ng/mL) was calculated using a PGE2 standard curve.

statistical analysis
Student's t-test and analysis of variance (one-way ANOVA) on the results of at least five independent biological replicates were performed. Values of p < 0.05 were considered statistically significant.

BV2 Microglial cell Viability assay
MTT assay was used to quantitatively evaluate cell viability. This was performed to verify whether the tested selective COX-1 inhibitors (P6 and mofezolac) caused toxicity in LPS-treated BV2 cell line. Preliminarily, the effect of two different concentrations of P6 (0.5 and 1 µM) and mofezolac (0.1 and 0.5 µM) on BV2 microglial cell viability was evaluated. No cell toxicity was exerted by either P6, mofezolac, and LPS alone or a combination of LPS and each of the two inhibitors at 24 h. The two concentrations of P6 and mofezolac were chosen based on the basis of previous studies and their COXs IC50 values (14).
Cell viability was found to be significantly (p ≤ 0.001) lower, after 48 h of LPS treatment (percentage cell viability equal to 78.4 ± 0.15 vs 99.8 ± 0.07 of the control), than untreated cells (control). None of the inhibitors used at the above indicated doses were toxic to BV2 microglial cells, when used in the absence of LPS (percentage cell viability ranging between 95.5 ± 0.22 and 98.2 ± 0.16 vs control). In addition, inhibitors used at the same doses, resulted protective toward BV2 microglial cells after 48 h LPS treatment (1 µg/mL), being the percentage of cell viability comparable to that observed in control.

evaluation of Mice glial activation
Astroglial activation was characterized by immunoreactivity and immunoblotting analysis of the GFAP expression, a marker used to distinguish astrocytes from other glial cells of the CNS. LPS treatment determined an increase of immunoreactive cell bodies in comparison to untreated mice suggesting astrocyte activation in different brain regions. In particular, the caudate, frontal lobe, hippocampus, and substantia nigra were selected to evaluate the astrocyte activation after different animal treatment (Figures 2A-D). Images report GFAP reactivity in samples of mice control group (CTR), mice treated with LPS vehicle (V-LPS), LPS alone (LPS) or in combination with mofezolac (M + LPS). In CTR sections, there is a physiological astroglial distribution with few immunoreactive elements surrounding the blood vessels as part of the brain-blood barrier, which means that there are no sign of reactive astrogliosis. In V-LPS-treated mice, the expression is similar to CTR, with perivascular immunoreactivity and the presence of a few astrocytes in the tissue, due to inflammation probably caused from the injection. In LPS-treated mice, images show a marked increase of the immunoreactive elements, whereas in vivo administration of mofezolac reduced the presence of GFAP immunoreactive cells in all the tested brain areas (Figures 2A-D).
Ionized calcium-binding adapter molecule-1 immunoreactiv ity, a marker of activated microglia, was also evaluated (Figure 3). LPStreated mice Iba-1 positive cells were more numerous, showing a more intense immunoreactivity, as well as a ramified phenotype, in comparison to untreated mice (Figures 3A-D). Mofezolac administration determined the reduction of Iba-1 immunoreactivity in LPS-injected mice in all the tested brain areas (Figures 3A-D), suggesting that mofezolac reduces, at least in part, the microglial activation induced by the neurotoxic insult.
Immunoblotting analysis was also performed to semiquantitatively evaluate both astrocyte and microglia activation in samples derived from mice groups previously described. In LPS-treated mice, a significant increase of GFAP expression was detected in all the brain regions tested when compared to controls or vehicle-LPS (Figure 4). Similar results were obtained for Iba-1 expression analysis (Figure 4).
Conversely, in mofezolac-treated animals previously injected with LPS, the GFAP as well as Iba-1 levels resulted significantly reduced in comparison to the animals that received LPS alone (Figure 4).

effect of the cOX-1 inhibitors-P6 and Mofezolac-on cOXs expression in BV2 Microglial cell line and Mice Brains
The effect of mofezolac and P6 on COX-1 expression in LPStreated microglial cells was evaluated by western blotting analysis. After 24 h, no significant difference between LPS-stimulated cells in the presence or absence of both COX-1 inhibitors was observed (data not shown). Interestingly, LPS-stimulated BV2 cells exhibited, at 48 h, increased levels of COX-1 expression in comparison to untreated cells ( Figure 5). The COX-1 expression levels evaluated at 48 h of incubation with LPS was significantly reduced in 1 h pretreated cells with either selective COX-1 inhibitor P6 or mofezolac ( Figure 5). Therefore, COX-1 expression was found to be time dependently reduced by both selective inhibitors tested.
The expression of COX-2 in the same cell lysates was also evaluated. Interestingly, COX-2 protein levels were unaffected by the presence of P6 or mofezolac. In these conditions, COX-2 expression resulted comparable to the level observed in cells stimulated with LPS alone (Figure 5).
Therefore, in LPS-treated BV2 microglial cell line, P6 and mofezolac were able to reduce COX-1 expression without affecting COX-2.
Cyclooxygenases expression was also in vivo evaluated, assaying different brain regions (Figure 4). LPS-challenged mice exhibited increased levels of COX-1 both in comparison to con trols and vehicle of LPS (vLPS) administered animals in all the different areas tested (Figure 4). Interestingly, LPS-challenged mice that received mofezolac treatment exhibited a significantly reduction of COX-1 expression in all brain areas tested. Protein levels comparable to controls were detected in mice treated with vLPS + mofezolac, whereas in mice treated with the vehicle of mofezolac + LPS, COX-1 levels resulted comparable to those detected in LPS-treated mice (Figure 4).
Immunoblotting assay on tissues of the same brain areas was performed to test the in vivo effects of mofezolac administration on COX-2 expression extent. COX-2 resulted overexpressed in all tested brain regions in LPS-administrated mice. Interestingly, mofezolac exhibited no effect on COX-2 modulation in LPStreated mice, being the protein levels almost comparable to LPS-treated mice (Figure 4).

effect of cOX-1 inhibitors on Pge 2 Biosynthesis in BV2 Microglial cells and Mouse Brain
The PGE2 biosynthesis extent was evaluated in supernatants of cell cultures at 48 h incubation time (Table 1A). PGE2 production in LPS-stimulated cells was significantly higher than its basal level present in the untreated cells (control). Interestingly, both COX-1 inhibitors were able to reduce in a dose-dependent manner PGE2 release in LPS-treated cells (Table 1A). In the tested brain regions of LPS-treated mice, PGE2 levels were significantly increased with respect to control animals or mice receiving the vLPS (Table 1B). Interestingly, a significant reduction of PGE2 levels was detected in mofezolac-treated mice before LPS challenging ( Table 1B).

effect of cOX-1 inhibitors on nF-kB activation in lPs-Treated BV2 Microglial cells and Mouse Brain
Since the phosphorylation and degradation of IkBα, the inhibitory complex of NF-kB, is an essential step for NF-kB activation, the expression levels of the phosphorylated form of IkBα (p-IkBα) were evaluated. In this context, BV2 microglial cells exposed to LPS exhibited a significant increase of p-IkBα in comparison with no stimulated cells (control) (Figure 5). Densitometric analysis revealed little p-IkBα in the untreated cells, whereas pretreatment with all tested COX-1 inhibitors, significantly reduced IkBα phosphorylation in LPS-stimulated cells (Figure 5).
Similar results were obtained in in vivo model. In fact, in all tested brain areas of LPS-treated mice, an increase of p-IkBα, in comparison to control animals as well as to vLPS administered mice, was detected. Conversely, in mice receiving mofezolac (M), a signifi cant reduction of p-IkBα in all tested areas was detected (Figure 4). p-IkBα levels comparable to controls were detected in mice treated with vLPS + mofezolac, whereas in mice treated with vehicle of mofezolac + LPS, p-IkBα levels were comparable to those detected in LPS-treated mice (Figure 4).

DiscUssiOn
Clinical data and basic research outcomes showed a strict correlation between neurodegeneration and neuroinflammation (1,8). COXs (both COX-1 and COX-2) play a central role in the inflammatory cascade by converting arachidonic acid (AA) into bioactive prostanoids (19). Both COX isoforms catalyze the same reactions: bis-oxygenation of AA to yield prostaglandin G2 (PGG2), and a peroxidase reaction, which converts PGG2 into prostaglandin H2 (PGH2). PGH2 is then transformed into PGE2, PGF2α, PGD2, PGI2, and thromboxane (TX) by specific terminal synthases. In the brain, both COX-1 and COX-2 are constitutively expressed (20), and there are regional differences in the regulation of COXs, probably as a consequence of the different distribution of COX-1 and COX-2 in the cerebral areas (21).    In physiological conditions, COX-1 is mainly expressed in microglia and perivascular cells, whereas COX-2 is found in postsynaptic dendrites and excitatory terminals, particularly in the cortex, hippocampus, and amygdala, with both neuronal and vascular associations. COX-1 recently has been recognized as a pivotal player in neuroinflammation (11,22). Elevated levels of TXA2 and increased COX-1 expression in the rat hippocampus seem to be correlated to increased brain susceptibility to inflammation and neurodegenerative diseases (23). Moreover, in several models of neuroinflammation, COX-1 has been shown to support inflammatory processes facilitating pro-inflammatory PG upregulation, mainly PGE2 (22).
In this regard, we previously reported that the COX-1 selective inhibitor P6 was able to control the inflammatory response in LPS-treated N13 microglial cells, an in vitro model of neuroinflammation (14). In this context, P6 was able to selectively downregulate COX-1 protein expression, without affecting COX-2 levels, as well as reduce IkBα phosphorylation. In this previous study, we proposed a new possible mechanism by which P6 negatively regulates COX-1 expression thereby NF-kB mediated signal-pathway modulation, suggesting the necessity of further studies to validate the efficacy of P6 as well as its analogs in the treatment of neuroinflammatory-based neurodegenerative disease (14).
Since the potential beneficial effect of COX-1 inhibition in the treatment of neuroinflammation has been considered, in the present study, we determined the modulatory effect of two selective COX-1 inhibitors-P6 and mofezolac-both in in vitro and in vivo models of neuroinflammation. For this purpose, we employed the mouse BV-2 microglial cells activated by LPS instead of N13 cell line to exclude a possible bias linked to the in vitro model used. Data of the present study clearly demonstrated that P6 was able to reduce COX-1 expression and negatively regulate the NF-kB activation, thus showing that these effects are independent from cell line used. Besides, P6 action on LPS-treated BV2 microglial cell responses was compared with the effects of mofezolac, an analgesic drug already approved to be used in humans, exerting its pharmacological action through a highly selective inhibition of COX- 1 (24, 25). In this regard, mofezolac was able to reduce COX-1 expression in LPS-activated BV-2 cells accompanied to PGE2 release reduction and NF-kB activation downregulation.
More interestingly, these results were confirmed with those derived from the preclinical in vivo model of neuroinflammation, where mice intra-cerebroventricularly received LPS. Concerning in vivo investigation, GFAP and Iba-1 expression, two markers of neuroinflammation, were reduced by mofezolac to an extent depending on the encephalon area analyzed (see Results). Moreover, the increase of COX-1 expression observed in all the brain sections of LPS-treated mice was selectively downregulated by the in vivo treatment with mofezolac, as well as PGE2 release and IkBα phosphorylation extent assayed in the tested brain areas.
In a previous work of Bosetti et al. carried out on a murine model, it was reported that evident neurodegeneration mediated by activated microglia and increased pro-inflammatory cytokines appeared in the hippocampus after LPS injection (11). Although, previous reports have shown sustained inflammation at 24 h after LPS injection (11,26), in our in vivo model, we prolonged up to 72 h the time after LPS intraventricular injection to determine the optimal degree of inflammatory response. In fact, at 72 h after LPS injection, we clearly detected in all the brain regions tested, represented not only by hippocampus but also caudate, substantia nigra, and frontal lobe, the signs of neuroinflammation in terms of astrocytes and microglia activation.
Hippocampus is a part of the encephalon, located at the temporal lobe, where it plays a crucial role in processing information selected for the long-term memory; therefore, it is important to check what happens at level of this brain area the neuroinflammation modulation (27). Moreover, substantia nigra and caudateputamen, both involved in neurodegenerative processes typically of the Parkinson's disease, as well as frontal lobe, involved in superior cognitive pathways, were investigated to evaluate the effects of COX-1 inhibitors (28,29).
Results from our experiments demonstrated that in vivo treatment with mofezolac was able to reduce glia activation acting selectively on COX-1 expression, thus supporting that COX-1 modulation represents a viable target useful to attenuate the neuroinflammatory response. In this regard, it was reported how COX-1 genetic depletion was able to attenuate microglial and astrocyte activation, reducing pro-inflammatory cytokine expression and preventing the neuronal cells loss in the hippocampus (11).
Recently, it was reported that the potent PET probe highly selective for COX-1, [ 11 C]-radiolabeled ketoprofen methyl ester was able to detect activated microglia associated with amyloid plaque progression, suggesting the involvement of COX-1 in the neuroinflammatory process in AD (30). The role of COX-1 in neuroinflammation is also supported by the fact that COX-1 inhibition through drug or gene deletion reduces neuronal damage and inflammatory responses after injection of LPS or Aβ in mouse brain (31,32). The LPS model is particularly relevant to examine activation of brain innate immunity, since it specifically and directly targets microglia, the immune resident cells in the brain, through the CD14 protein binding and subsequent toll-like receptor 4 mediated pro-inflammatory signaling pathway activation. These events cascade culminates to NF-kB activation leading to the release of cytokines, chemokines, ROS, PG, and TX (33,34).

cOnclUsiOn
Our results demonstrated that both P6 and mofezolac were able to reduce COX-1-derived PGE2 release complemented with an antiinflammatory activity at the level of the NF-kB, thus providing mechanistic insights into the suppressive effect of these COX-1 inhibitors on LPS-induced neuroinflammatory response by microglia.
In conclusion, this work consolidated the hypothesis that selective COX-1 inhibitors can positively modify the inflammatory response in LPS-induced neuroinflammatory models. Overall these results, from in vitro and in vivo experiments, indicate the capability of two highly selective COX-1 inhibitors P6 and mofezolac to modulate the NF-kB signaling pathway. These findings emphasize the neuroprotective effect and therapeutic potential of COX-1 inhibitors useful in the control of neuroinflammatory diseases.

eThics sTaTeMenT
This study was carried out in accordance with the recommendations of Ministero della Salute Decreto Ministeriale, Dott.