The products of arachidonic acid metabolism are key mediators of inflammatory responses in the central nervous system, and yet we do not know the mechanisms of their regulation. The phospholipase A2 enzymes are sources of cellular arachidonic acid, and the enzymes cyclooxygenase-2 (COX-2) and microsomal PGE synthase-1 (mPGES-1) are essential for the synthesis of inflammatory PGE2 in the brain. These studies seek to determine the function of cytosolic phospholipase A2α (cPLA2α) in inflammatory PGE2 production in the brain. We wondered whether cPLA2α functions in inflammation to produce arachidonic acid or to modulate levels of COX-2 or mPGES-1. We investigated these questions in the brains of wild-type mice and mice deficient in cPLA2α (cPLA2α−/−) after systemic administration of LPS. cPLA2α−/− mice had significantly less brain COX-2 mRNA and protein expression in response to LPS than wild-type mice. The reduction in COX-2 was most apparent in the cells of the cerebral blood vessels and the leptomeninges. The brain PGE2 concentration of untreated cPLA2α−/− mice was equal to their wild-type littermates. After LPS treatment, however, the brain concentration of PGE2 was significantly less in cPLA2α−/− than in cPLA2α+/+ mice (24.4 ± 3.8 vs. 49.3 ± 11.6 ng/g). In contrast to COX-2, mPGES-1 RNA levels increased equally in both mouse genotypes, and mPGES-1 protein was unaltered 6 h after LPS. We conclude that cPLA2α regulates COX-2 levels and modulates inflammatory PGE2 levels. These results indicate that cPLA2α inhibition is a novel anti-inflammatory strategy that modulates, but does not completely prevent, eicosanoid responses.
- prostaglandin E synthase
- prostaglandin E2
the inflammatory process is important in the propagation of both acute and chronic neurological injuries such as stroke, Alzheimer disease, and multiple sclerosis (10, 30, 44). Inflammatory stimuli generated by cells within the brain and circulating molecules can alter the expression of inflammatory proteins within the brain. The enzyme cyclooxygenase (COX)-2 is among the earliest and best characterized of these molecules. Furthermore, inhibition of COX-2 activity appears to be neuroprotective in a variety of animal models of neurological disease (19, 25).
The COX enzymes convert arachidonic acid to PGH2, the common precursor of all prostaglandins and thromboxanes. Although there are two COX isoforms, the COX-2 enzyme has been characterized as the inducible form that mediates inflammatory prostanoid generation (36). O’Banion demonstrated that COX-2 is induced by IL-1β in cultured glial cells, and studies have demonstrated in vivo induction of COX-2 after inflammatory stimuli (28, 38).
PGE2 is an important mediator of inflammation in the brain, and PGE2 levels correlate directly with levels of inflammation in a model of stroke (9). Three distinct PGE2 synthase (PGES) enzymes convert PGH2 into PGE2: a glutathione-dependent cytosolic form and two membrane-associated PGES forms (mPGES-1 and -2). Analogous to COXs, PGESs appear to have different cellular distributions and expression patterns (45, 47). LPS triggers the release of proinflammatory cytokines, which induce both COX-2 and mPGES-1 in an identical set of vascular endothelial cells of the central nervous system (CNS) (29, 51). The coordinated actions of these enzymes produce high levels of PGE2 within the brain and cerebral spinal fluid of rodents (11, 21). Studies with COX-2 and mPGES-1 knockout mice have proved that increases in brain PGE2 and the febrile response are dependent on COX-2 and mPGES-1 expression (26, 27). Direct cerebral injection of IL-1β increased levels of PGE2, COX-2, and mPGES-1 in mouse brain tissue (31).
It is believed that phospholipase A2 (PLA2) is the major source of the arachidonic acid that is required for PGE2 and other prostanoid synthesis. PLA2 comprises a large enzyme superfamily that produces free arachidonic acid by hydrolyzing the fatty acid at the sn-2 position of membrane glycerophospholipids (reviewed in Ref. 8). The PLA2 family includes secretory, small-molecular weight PLA2 (sPLA2), calcium-independent PLA2, and the intracellular 85-kDa cytosolic PLA2 (cPLA2α). It has been postulated that cPLA2α is required for the generation of central arachidonic acid after LPS exposure because cPLA2α preferentially liberates arachidonic acid and is subject to diverse mechanisms of regulation (7, 17, 51). In addition, peritoneal macrophages derived from cPLA2α-deficient (cPLA2α−/−) mice are unable to generate any PGE2 in response to LPS (3), and cytokine treatment of bone marrow-derived mast cells harvested from cPLA2α−/− mice fail to induce COX-2 mRNA or protein (15).
The roles of the PLA2s in the coordinated regulation of COX and PG production within the CNS are largely unexplored (8). Rosenberger evaluated lipid metabolism in the brains of cPLA2α−/− mice and found reduced arachidonate turnover (39). Bosetti and Weerasinghe examined cPLA2α−/− mice and determined that basal levels of COX-2 mRNA and protein were reduced compared with wild-type littermates (4).
Thus cell culture data and previous experiments with the cPLA2α−/− mice indicate that cPLA2α has the potential to influence inflammatory PGE2 levels in the brain by three possible mechanisms: 1) cPLA2α may generate metabolically active arachidonic acid; 2) cPLA2α may be necessary for COX-2 induction; or 3) cPLA2α may regulate mPGES-1 induction.
Systemic administration of LPS is a widely used model for the study of COX-2, mPGES-1, and PGE2 in neuroinflammation (51). Here, we studied cPLA2α−/− and wild-type littermate mice in this model and measured COX-2, mPGES-1, and PGE2 levels to determine the dependence of the CNS inflammatory PGE2 response on cPLA2α.
C57BL/6J male and female mice with disruption of the gene encoding for cPLA2α (cPLA2α−/−; 10–16 wk old) were used together with their wild-type littermates (3). Body weights were in the 22–30 g range. Escherichia coli LPS was purchased from Sigma (St. Louis, MO). Goat anti-COX-1 and anti-COX-2 antibodies were obtained from Santa Cruz Biotechnology (Santa Cruz, CA). Cy3-conjugated donkey anti-goat IgG and normal goat serum were purchased from Jackson ImmunoResearch Laboratories (West Grove, PA). Anti-cPLA2α antibody was the gift of André Cybulsky.
All experiments were conducted with the approval of the Institutional Animal Care and Use Committee and in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals (20). Unless otherwise specified, mice were injected intraperitoneally with 40 mg/kg LPS or an equivalent volume of pyrogen-free saline (16 ml/kg) using a 27-gauge needle.
Northern blot analysis.
Brains were rapidly removed from PBS-perfused mice, frozen in liquid nitrogen, and stored at −80°C until use. Total RNA was extracted using the guanidine isothiocyanate-cesium chloride method (6). RNA blots were hybridized with [α-32P]dCTP-labeled cDNA probes.
Preparation of cDNA probes.
The cDNA probes were created by PCR using primers corresponding to the nucleotide sequences published in GeneBank databases: murine COX-1 (sense, 5′-CGGATCCTACTGGCTCTGGAATTTGTGAATGC-3′; antisense 5′-CGAATTCGGTTATGTTCACGAAGCCAGATCGTG-3′), COX-2 (sense, 5′-CGGATCCATCCTTGCTGTTCCAATCCATGTCAA-3′; antisense, 5′-CGAATTCCCAGGTCCTCGCTTATGATCTGTCTT-3′), murine mPGES (sense, 5′-GGTGTCCCCGAGTTGAAGT-3′; antisense, 5′-GGCATTTTGTGAGGTGAAGG-3′), murine cPLA2α (sense, 5′-TGGATCCACATGGTACATGTCAACCTTGTACTC-3′; antisense, 5′-CGAATTCGGATATGATGTGTTGAGATTCAAGCC-3′), murine group V PLA2 (sense, 5′-AAGAGGGTTGTAAGTCCAGAGG-3′; antisense, 5′-CAGGGGGCTTGCTAGAACTCAA-3′), and murine GAPDH (sense, 5′-CTTCATTGACCTCAACTACAT-3′; antisense, 5′-CCAAAGTTGTCATGGATGACC-3′). The sense and antisense primers for COX-1, COX-2, mPGES, and cPLA2α probes had restriction sites for BamHI and EcoRI at their 5′ and 3′ termini, respectively.
Competitive RT-PCR analysis.
Competitive RT-PCR was performed to determine amounts of COX-2 mRNA. The protocol was adopted from a previously published study (37). A deleted DNA construct was generated using the endogenous COX-2 transcript but was missing an internal 132-bp fragment. An amplified COX-2 deletion product (mutCOX-2) of ∼430 bp was constructed and subcloned in plasmid Bluescript. RNA for mutCOX-2 was synthesized with T3 RNA polymerase (Promega) and used for subsequent reverse transcription-coupled PCR. Total RNA of mouse brain (1 μg) was reverse transcribed simultaneously with known amounts of mutCOX-2 RNA and then coamplified with the same set of primers. The relative amount of each PCR product was measured with densitometry (NIH Image) of ethidium bromide-stained agarose gels. In Fig. 4B, the logarithm of the ratio of the optical densities of COX-2 to mutCOX-2 was plotted as a function of the logarithm of the amount of mutCOX-2. The best-fit line was determined by linear regression. When the value of the vertical axis is zero, the amount of competitor equals the amount of native COX-2 mRNA.
In situ hybridization.
Digoxigenin-11-UTP (Boehringer-Mannheim, Germany)-labeled RNA probes were prepared from 1 μg of BamHI (for antisense RNA probe)- or EcoRI (for sense RNA probe)-linearized murine COX-2 cDNA using 2 μg/μl T3 or T7 RNA polymerase (Promega). The RNA probe was hydrolyzed in alkaline solution (in mM: 56 NaHCO3, 84 Na2CO3, 7 dithiothreitol, and 3 EDTA) at 60°C for 10 min and ethanol precipitated. RNA concentrations were estimated by comparison to the digoxigenin-labeled control RNA (Boehringer) using an alkaline phosphatase-conjugated antibody (Boehringer).
For in situ hybridization, perfusion-fixed brains were immediately removed and immersed for 30 min in 4% paraformaldehyde in PBS, then in 30% sucrose in PBS at 4°C for 16 h. Coronal brain sections (18 μm thick) were cut with a cryostat (Leica Frigocat, Wetzler, Germany) at −20°C and thaw-mounted on glass slides. The frozen sections were fixed in 4% paraformaldehyde for 10 min, rinsed with PBS, and immersed in 0.2 N HCl for 10 min. After a brief rinse in PBS, the sections were transferred to 2× SSC for 10 min at 70°C, immersed in 0.1 M triethanolamine (pH 8.0) for 5 min, and then acetylated in 0.25% acetic anhydride/0.1 M triethanolamine at room temperature for 10 min. Sections were incubated overnight at 70°C with hybridization solution (50% formamide, 5× SSC, 2.5× Denhardt’s solution, 0.25 mg/ml yeast tRNA, 0.05 mg/ml sonicated denatured herring sperm DNA) containing 400–1,000 ng/ml of either digoxigenin-11-UTP-labeled antisense RNA or sense RNA for control. After washing, 50 μl per section of sheep anti-digoxigenin-alkaline phosphatase conjugate (1:5,000 dilution in buffer containing 1% normal sheep serum) was applied at 4°C for 18 h. The detection of signal was performed by overnight incubation with 150 μl of chromogen solution (20 μl of nitroblue tetrazolium and 15 μl of 5-bromo-4-chloro-3-indolyl-phosphate in 4 ml of buffer) at room temperature. Color development was checked, and the reaction was terminated by immersing the slides in 10 mM Tris·HCl, pH 7.5, 1 mM EDTA. The slides were mounted and examined by light microscope.
Western blot analysis.
After death, mouse brains were rapidly removed and a single hemisphere from each mouse was homogenized by 20 strokes of a tight-fitting Dounce homogenizer at 4°C in buffer containing protease inhibitors. The crude homogenate was centrifuged at 10,000 g for 20 min at 4°C, and protein concentration was determined by a modified Bradford assay (Bio-Rad). Proteins were separated on 10–12% acrylamide gels using SDS-PAGE and transferred to polyvinylidene difluoride membranes (Immobilon; Millipore, Billerica MA).
Three mice of each genotype were treated with saline or 40 mg/kg LPS and killed at 6 h. Brains were perfusion fixed, as described above, and 5-μm coronal sections were cut and thaw-mounted on glass slides, rinsed briefly in PBS, and incubated with 2% BSA in PBS for 20 min. The sections were incubated with the goat anti-COX-2 antibody (1:200). After three washes in PBS, the sections were incubated with the Cy3-conjugated donkey anti-goat IgG (1:500). In some experiments, 4′,6-diamino-2-phenylindole (final concentration of 2 μg/ml), a nuclear-specific dye, was added to the solution of secondary antibody. The sections were mounted with Vectashield mounting solution (Vector Laboratories, Burlingame CA).
One hemisphere of each brain was weighed and immediately homogenized in 70% ice-cold methanol. The homogenate was centrifuged at 10,000 g for 25 min at 4°C, and the supernatant was evaporated under a nitrogen stream. The lipids were then resuspended in assay buffer, and PGE2 content was measured by ELISA according to instructions (Amersham, Piscataway, NJ).
Data and statistical analysis.
The intensity of Northern blot signals was measured using the National Institutes of Health Image program. The relative signal intensities were normalized to the 18S ribosomal band of the ethidium bromide-stained RNA gel for each lane. Because COX-2 and mPGES mRNAs were barely detected in saline-treated brains of both genotypes, comparisons were made only between LPS-treated mice using paired Student’s t-tests. COX-1 differences were compared using two-factor analysis of variance. The images were analyzed with identical settings and averaged. For in situ RNA hybridization signal measurement, relative threshold densitometry was performed with the Inquiry program (version 3.08; Loats Associates, Westminster, MD). The threshold level was set to eliminate background signal when the sense RNA probe was used. Two mice were analyzed for each condition, and a minimum of three representative images from each mouse were measured. Applicable results are displayed as means ± SE, with P < 0.05 considered significant.
Northern blot analysis.
Based on previous cell culture experiments, we postulated that cPLA2α could effect induced COX-2 expression (15). To evaluate the effects of systemic LPS injection on COX-2 mRNA expression in the cPLA2α+/+ and cPLA2α−/− mice, total RNA was extracted from brain, heart, and lung tissue and examined by Northern blot analysis. Figure 1 shows that COX-2 mRNA is scarcely detectable in saline-treated tissues from both cPLA2α+/+ and cPLA2α−/− mice. LPS injection markedly increased COX-2 mRNA expression in the mouse brains and lungs. Notably, the LPS-induced upregulation of COX-2 mRNA, corrected relative to 18S intensity, was much less in the brain of cPLA2α−/− mice (64%, n = 7, P < 0.01).
In contrast, COX-1 mRNA was constitutively expressed at equivalent levels in the brains of wild-type and cPLA2α−/− mice. Treatment of the cPLA2α−/− mice with LPS resulted in approximately twofold reduction in brain COX-1 mRNA compared with saline-treated mice (n = 3, P < 0.025).
The band for cPLA2α in the wild-type mouse tissues shows the expected size for cPLA2α of 2.8 kbp, as has been reported previously (7). In the saline-treated cPLA2α+/+ mice, cPLA2α mRNA was detected in the lung. After LPS injection, cPLA2α mRNA was increased in the brain, heart, and lung. The increase in levels of brain mRNA is consistent with the finding that seizure activity in rats also induces both COX-2 and cPLA2α (48). As expected, cPLA2α mRNA was not seen in the tissues of the cPLA2α−/− mice. These results demonstrate that induced COX-2 mRNA depends significantly on cPLA2α.
To determine the time of peak levels of COX-2 mRNA in the brain after LPS injection in our model, we harvested total brain RNA at various times. The upregulation of COX-2 reached a maximum between 3 and 6 h after LPS injection and then decreased to basal levels at 12 h (data not shown). A small but reproducible second phase of increase in COX-2 mRNA levels occurred at 24 h after the LPS injection. We conducted subsequent experiments at the 6-h time point to ensure high levels of COX-2.
Levels of COX-2 and mPGES-1 mRNA and protein have been demonstrated to be coordinately upregulated in the vasculature of the rat brain after LPS injection (51). We inquired whether mPGES-1 regulation was also abnormal in the cPLA2α−/− mouse. Northern hybridization with a probe specific for murine mPGES-1 revealed four transcripts, as has been demonstrated in murine peritoneal macrophages (Fig. 2) (49). mPGES-1 was not detected in the saline-treated brains of either genotype (Fig. 2B). LPS treatment strongly induced mPGES-1 in both the cPLA2α+/+ and cPLA2α−/− mice. The mPGES-1 mRNA in the brains of cPLA2α−/− mice exhibited a tendency toward increased levels (1.4-fold of wild type, P = not significant, n = 6; Fig. 2B). This is in contrast to the decreased brain levels of COX-2 mRNA in the cPLA2α−/− mouse after LPS treatment (Fig. 2A) and suggests that cPLA2α-dependent inflammatory gene regulation is specific.
Competitive PCR analysis.
To more precisely determine the magnitude of LPS-induced upregulation of COX-2 message in cPLA2α+/+ and cPLA2α−/− mouse brains, we used a highly accurate and specific competitive PCR technique (37). Figure 3A shows representative ethidium bromide-stained gels of competitive PCR products for COX-2 induced in brain of cPLA2α+/+ and cPLA2α−/− mice after LPS injection. A higher amount of competitor (mutCOX-2) is needed to compete with the COX-2 products in the cPLA2α+/+ brain. After LPS injection, the cPLA2α+/+ brains had 2.3-fold more COX-2 mRNA than did the cPLA2α−/− mice (cPLA2α+/+ mice, 177 ± 11 pg/μg of total RNA, n = 4; cPLA2α−/− mice, 76 ± 7 pg/μg of total RNA, n = 3; P < .05) (Fig. 3C). By this analysis, the LPS-induced brain COX-2 mRNA of the cPLA2α−/− is only 43% of normal, and this compares favorably to the Northern analysis (Fig. 1).
Localization of COX-2 mRNA in brain after LPS injection.
We suspected that COX-2 would be induced at specific sites within the brain and performed in situ RNA hybridization 6 h after LPS injection to define the expression sites. COX-2 signals were undetectable in the leptomeninges of the brain of both groups of saline-injected mice (Fig. 4, A and C), whereas signals were detected in neuronal cells of the cerebral cortex (Fig. 4, B and D). At 6 h after LPS injection, there was high expression of COX-2 mRNA in the leptomeninges of the parietal cortex and the blood vessels located on the surface of the cerebral cortex in cPLA2α+/+ mice (Fig. 4E). In contrast, the intensity of the signal was much less in the corresponding regions of the cPLA2α−/− brain (Fig. 4G). We performed densitometric analysis and found that COX-2 mRNA in these regions for cPLA2α−/− mice was ∼34% of the amount in wild-type mice. The COX-2 of the parenchymal cells was not influenced markedly by LPS injection in the brains of either cPLA2α+/+ or cPLA2α−/− mice (Fig. 4, F and H). No mRNA signal was detectable when the sense probe was used in LPS-treated cPLA2+/+ mice (Fig. 4E, inset). These results demonstrate that cPLA2α-dependent differences in COX-2 mRNA levels are most pronounced in the meninges and within blood vessels after LPS injection.
COX-2 protein in cPLA2α+/+ and cPLA2α−/− brain.
We measured cPLA2α, COX-2, and mPGES-1 protein by the Western blot technique. cPLA2α was detected in the brains of wild-type, but not cPLA2α−/− mice (Fig. 5A). The levels of cPLA2α were not altered 6 h after the treatment with LPS (Fig. 5A). We determined that mPGES-1 protein was expressed equally in the saline-treated wild-type and cPLA2α−/− brains (Fig. 5C). Interestingly, 6 h after LPS, there was no significant increase in the mPGES-1 protein levels in the brains of either genotype (Fig. 5C). In contrast to these findings, COX-2 expression in the brains of the saline-treated cPLA2α−/− mice was only 0.31-fold that of the wild-type mice (P = 0.04; Fig. 5B). After 6 h of LPS treatment, COX-2 protein was significantly increased in the brains of both cPLA2α+/+ and cPLA2α−/− mice. The levels of COX-2 protein 6 h after LPS treatment were significantly less (0.72-fold, P = 0.03) in the cPLA2α−/− compared with the LPS-treated wild-type mice.
We then used immunofluorescence microscopy to compare the expression of COX-2 protein in specific cells of the brains of cPLA2+/+ and cPLA2α−/− mice. As expected, COX-2 immunoreactivity was rarely detected after saline injection in either genotype (Fig. 6, A–C). In cPLA2α+/+ mice, COX-2 immunoreactivity was enhanced 6 h after LPS injection in cells of parenchymal, midsize blood vessels (Fig. 6D), blood vessels penetrating cerebral cortex (Figs. 6E), and large blood vessels of the subarachnoid space (Fig. 6G). COX-2 was present in the leptomeninges and the luminal surface of parenchymal blood vessels. As shown by 4′,6-diamino-2-phenylindole-staining of nuclei, COX-2 immunoreactivity appears specifically around the nuclear envelope. COX-2 was also detected within cells of the parietal cortex, but the signals were much weaker than those in blood vessels and not affected by LPS injection (data not shown). Substitution of normal goat serum for the primary antibody resulted in images that had essentially no background staining (data not shown).
PGE2 content of cPLA2α+/+ and cPLA2α−/− brain.
PGE2 caused by inflammation is the final product of the increased activities of PLA2, COX-2, and mPGES-1. We assessed the functional significance of reduced COX-2 induction and the absence of cPLA2α in the cPLA2α−/− mouse on PGE2 synthesis by measuring the levels of PGE2 in whole brains (Fig. 7). Saline-injected mice of both genotypes had comparable brain levels of PGE2 (7.7 ± 1.6 vs. 6.8 ± 0.12 ng/g, P = not significant). LPS-induced brain PGE2 was significantly less in the cPLA2α−/− mice compared with their wild-type littermates (3.6- and 6.4-fold, respectively; P < 0.05). Thus, although basal PGE2 was equivalent in both genotypes, the reduced PGE2 response to LPS correlated very closely with the reduction in induced COX-2 mRNA.
PLA2s, COX-2, and mPGES-1 are important enzyme mediators in the CNS response to inflammation (5, 14, 31). In this study, we examined the relationship of cPLA2α in the PGE2 response of the CNS to LPS. We found that peak COX-2 mRNA induction after LPS exposure in the brains of cPLA2α−/− mice was only ∼40% of that seen in wild-type mice. Both basal and LPS-stimulated levels of COX-2 protein were significantly decreased in the cPLA2α−/− mouse brains. We used in situ RNA hybridization and protein immunofluorescence to demonstrate that cPLA2α−/− brain perivascular and meningeal cells have less COX-2 induction. In contrast, mPGES-1 mRNA responses in the cPLA2α−/− mouse brain were equal to wild-type responses. PGE2 levels in saline-treated brains were the same in cPLA2α−/− and wild-type mice. After LPS treatment, however, cPLA2α−/− mice had significantly less brain PGE2 than wild-type mice. These results show that cPLA2α plays a significant role in regulating the response of COX-2, but not that of mPGES-1, to inflammatory stimuli in vivo. They also indicate that cPLA2α does not provide arachidonic acid for basal eicosanoid synthesis in the brain.
Results obtained by comparing genetically altered mice to their wild-type littermates must be interpreted with caution. It is possible that other members of the large PLA2 family are able to compensate for the deficiency of cPLA2α in the cPLA2α−/− mouse. Our in vivo results, however, are consistent with studies that used both chemical inhibitors and cells derived from cPLA2α−/− mice and showed cPLA2α-dependent COX-2 induction and inflammatory prostaglandin synthesis (3, 15).
It is possible that cPLA2α has regulatory effects on other eicosanoid synthetic enzymes, such as PGD synthases. We have focused on PGE2 metabolism in this study because of the well-described effects of LPS and other inflammatory agents on PGE2 synthesis (21, 50). Although basal PGD2 levels in the brain are higher than those of PGE2, LPS treatment amplifies PGE2 synthesis to a greater extent than PGD2 (50).
We limited our examination of the PGES enzymes to mPGES-1. Although both mPGES-2 and cytosolic PGES synthesize PGE2, they have been characterized as constitutive enzymes (32, 47). Biochemical analysis has demonstrated enzymatic coupling of COX-2 activity to mPGES-1 (33). Furthermore, in rodents, the rise in central levels of PGE2 after acute systemic injection of LPS appears to be dependent on mPGES-1. Uematsu et al. (49) showed that mPGES-1-deficient mice injected with 1 mg of LPS failed to increase serum levels of PGE2. Engblom et al. (12) found that LPS treatment failed to increase PGE2 in the CSF or PGES activity in the brains of mPGES-1-deficient mice. Moore and colleagues recently demonstrated Il-1β-induced upregulation of cytosolic PGES/p23 in the brain, and it remains to determine whether this PGES contributes to inflammatory PGE2 synthesis in our model (31).
We used a high systemic dose of LPS (40 mg/kg) and examined its acute effects on the cycloxygenase-mPGES-1 axis. This is a well-described model in which both COX-2 and mPGES-1 are induced in vascular endothelial cells (51). We found that COX-2 mRNA levels peak between 3 and 6 h. Other investigators used similar doses of LPS and found that mPGES-1 RNA responses also peak within this time period (49). Given the localization of both the COX-2 mRNA and the immunoreactive protein, the positive cells lining the luminal surface of blood vessels are likely endothelial cells. This finding is consistent with those reported by other groups showing induction of COX-2 in cerebral vascular endothelial cells after LPS treatment (5, 26, 29). It is possible that other cell types, such as astroglia, pericytes, and smooth muscle cells, express COX-2 at the arterioles, a possibility that has important implications for regulation of the neurovascular unit during inflammation (18). The significant reduction of COX-2 responses in the cPLA2α−/− brain may have been different if we had used an isolated CNS model of inflammation. For example, a 6-day infusion of LPS into the fourth cerebral ventricle of the rat did not result in changes in cPLA2α, sPLA2, COX-1, or COX-2 protein levels but did increase cPLA2 activity (40).
Other PLA2s or PLA2-independent mechanisms may be the source of arachidonate for PGE2 synthesis in cPLA2α−/− brain as basal PGE2 concentrations are normal and PGE2 significantly increases after LPS injection (Fig. 7). A model of inflammatory pain in which central COX-2 and PGE2 levels increase dramatically without an increase in central PLA2 activity supports this hypothesis (41). Furthermore, selective overexpression of COX-2 in neurons dramatically increases basal CNS PGE2 levels, indicating that basal arachidonic acid levels are sufficient to produce large increases in PGE2 (50).
It is not surprising that other forms of PLA2 or PLA2-independent mechanisms are responsible for generating metabolically active arachidonate in the brain. The cPLA2α content of rodent brains is low. In rat brain, calcium-independent PLA2 activity is at least 10-fold greater than sPLA2 activity, and cPLA2α activity is even less than that of sPLA2s (52). In wild-type mice, the cPLA2α enzymatic activity of whole brain homogenate was below our ability to detect it using published assays (data not shown) (52). Other laboratories have also been unable to detect differences in total or specific PLA2 activities between cPLA2α+/+ and cPLA2α−/− mice (4). Rosenberger and colleagues performed a detailed analysis of brain lipid metabolism in the cPLA2α−/− mouse and found an increased turnover of arachidonic acid but no difference in free arachidonic acid concentrations (39). They also concluded that other sources of arachidonic acid function in the absence of cPLA2α (39). Furthermore, it is possible that chronic changes or redundancy of PLA2 activities in the brains of the cPLA2α−/− mouse compensate for cPLA2α deficiency. Bosetti and Weerasinghe analyzed in vitro PLA2 activities of brain homogenate and did not find compensation (4). Our findings may also be strain dependent. Group IIA sPLA2 is a major source of PLA2 activity in rat brain (52), but because of a naturally occurring missense mutation, neither the C57BL/6J nor Sv129 mouse strains used to generate the cPLA2α-mouse express group IIA PLA2 (23). Further studies with application of specific inhibitors and with cPLA2α−/− mice of different strains clarify these issues.
Our Northern blot analyses did not reveal differences in basal COX-2 RNA levels, but Western blot analysis showed that constitutive COX-2 protein levels were less in the cPLA2α−/− brain. Bosetti and Weerasinghe (4) found that constitutive levels of both COX-2 protein and RNA are reduced in the cPLA2α−/− mice. It is possible that the more sensitive PCR-based method of those authors detected smaller differences in basal COX-2 RNA. The basal levels of brain PGE2 in both genotypes were the same in our study, suggesting that small differences in basal COX-2 levels have little impact on unstimulated PGE2 synthesis.
cPLA2α and other forms of PLA2 regulate the induction of COX-2 by cytokines in cell culture systems. Cytokine induction of COX-2 in murine bone marrow-derived mast cells is dependent on cPLA2α (15). Balsinde and coworkers (1) used P388D1 murine macrophage-like cells to demonstrate that group V sPLA2 can also modulate COX-2 induction by LPS treatment. Thus it is possible that the ability of the cPLA2α−/− mice to partially induce COX-2 may be the result of the complementary actions of another form of PLA2, such as group V PLA2.
In contrast to COX-2, the mRNA response of mPGES-1 in the cPLA2α−/− mice remains intact. In macrophages, mPGES-1 induction is completely dependent on nuclear factor-IL6 and is essential for the PGE2 response to high-dose LPS (49). The transcriptional responses of COX-2 and mPGES-1 to inflammatory agents share many features. However, in orbital fibroblasts, the increase in mPGES-1 is a purely inductive process, whereas levels of COX-2 mRNA are partially dependent on mRNA stabilization (16). Our results showing no difference between basal mPGES-1 expression are in agreement with those of Bosetti and Weerasinghe (4). Given the apparent increase in mPGES-1 mRNA 6 h after LPS, we were surprised that protein levels did not also increase. In a model of cerebral ventricle injection of IL-1β, Moore and colleagues (31) found that, although brain COX-2 protein levels appear to peak by 6 h, the mPGES-1 levels continued to rise up to 24 h after injection. It is possible that, in our model of systemic LPS injection, the mPGES-1 protein levels respond more slowly than COX-2. A more detailed temporal analysis might resolve this question.
What actions of cPLA2α enhance COX-2 levels? cPLA2α might alter the cytokine milieu and thus transcription of COX-2. We found that cPLA2α−/− peritoneal macrophages treated with LPS secreted significantly more TNF-α than did wild-type macrophages (data not shown). In a mouse multiple sclerosis model, a cPLA2α inhibitor prevented COX-2 and altered cytokine expression (22). The arachidonic acid or other lipid byproducts of cPLA2α might activate intracellular signaling pathways. For example, Serou et al. (43) demonstrated that IL-1β-stimulated induction of COX-2 in cultured rat hippocampal neurons is dependent on platelet-activating factor. Finally, products of cPLA2α activity, such as PGE2, may stabilize COX-2 mRNA (13). After injection of IL-1β, however, COX-2 inhibition does not affect COX-2 levels but does prevent increases in mPGES-1 (31). We postulate that the cytokine response to LPS induces both COX-2 and mPGES-1 but that a cPLA2α-dependent eicosanoid or platelet-activating factor is required for maximal COX-2 response. This is an attractive hypothesis because cPLA2α has a preference for phosphatidylcholine-containing phospholipids that are the precursors of platelet-activating factor (7).
In the future, experiments with specific PLA2 inhibitors will help determine the mechanisms of cPLA2α-dependent COX-2 regulation. There is also a need to further dissect the cPLA2α signaling cascade within the CNS and determine the importance of cPLA2α in other models of neuroinflammation using both in vivo and cell culture techniques. The availability of mice deficient in other forms of PLA2, such as group V or calcium-independent PLA2, will further this work (2, 42). Investigation of such knockouts can provide new insights into the production of inflammatory eicosanoids.
The coordinated regulation of COX-2 by cPLA2α has important implications for potential therapies. Inhibition and gene knockout of cPLA2α have already demonstrated profound improvement in neurological outcomes of mice after inflammatory, oxidative, and immunologic injuries (3, 22, 24, 46). It remains to be determined whether the benefit of eliminating cPLA2α activity in the CNS goes beyond downregulation of the COX-2 response. In the lung, cPLA2α propagates injury by COX-2-independent mechanisms of injury propagation (34, 35). If such effects are also found in the CNS, a specific cPLA2α inhibitor may provide benefits beyond those found in traditional nonsteroidal or selective COX-2 inhibition.
This work was supported by National Institutes of Health Grants DK-39773, DK-38452, NS-10828 (to J. Bonventre), DK-02493 (to A. Sapirstein), and American Heart Association Grant 0150504N (to A. Sapirstein). H. Saito received a Fellowship from the Ministry of Education, Science, Sports, and Culture of Japan for Research Abroad.
We thank Drs. David Borsook and David Linden for helpful comments. We thank Drs. Paolo Ciceri and Peter C. Isakson for advice on PGE2 measurement, Kathleen Blizzard for technical assistance, and Tzipora Sofare for editorial assistance.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2005 the American Physiological Society