Phospholipase Cɛ links G protein-coupled receptor activation to inflammatory astrocytic responses

Abstract

Neuroinflammation plays a major role in the pathophysiology of diseases of the central nervous system, and the role of astroglial cells in this process is increasingly recognized. Thrombin and the lysophospholipids lysophosphatidic acid and sphingosine 1-phosphate (S1P) are generated during injury and can activate G protein-coupled receptors (GPCRs) on astrocytes. We postulated that GPCRs that couple to Ras homolog gene family, member A (RhoA) induce inflammatory gene expression in astrocytes through the small GTPase responsive phospholipase Cɛ (PLCɛ). Using primary astrocytes from wild-type and PLCɛ knockout mice, we demonstrate that 1-h treatment with thrombin or S1P increases cyclooxygenase 2 (COX-2) mRNA levels ∼10-fold and that this requires PLCɛ. Interleukin-6 and interleukin-1β mRNA levels are also increased in a PLCɛ-dependent manner. Thrombin, lysophosphatidic acid, and S1P increase COX-2 protein expression through a mechanism involving RhoA, catalytically active PLCɛ, sustained activation of protein kinase D (PKD), and nuclear translocation of NF-κB. Endogenous ligands that are released from astrocytes in an in vitro wounding assay also induce COX-2 expression through a PLCɛ- and NF-κB–dependent pathway. Additionally, in vivo stab wound injury activates PKD and induces COX-2 and other inflammatory genes in WT but not in PLCɛ knockout mouse brain. Thus, PLCɛ links GPCRs to sustained PKD activation, providing a means for GPCR ligands that couple to RhoA to induce NF-κB signaling and promote neuroinflammation.

Many diseases of the central nervous system (CNS) are initiated or exacerbated by the activation of inflammatory signaling cascades. Microglial and astroglial cells participate in this process by both producing and responding to proinflammatory cytokines and mediators including IL-1β, IL-6, and inducible nitric oxide synthase (1–5). Astrocytes also respond to ligands for G protein-coupled receptors (GPCRs), including thrombin and the lysophospholipids lysophosphatidic acid (LPA) and sphingosine 1-phosphate (S1P) (5). These ligands can be generated within or supplied to the injured brain as a result of breakdown of the blood–brain barrier (6, 7). Signaling through these GPCRs has been implicated in CNS injury and disease. For example, genetic deletion of the protease-activated receptor 1 (PAR1) thrombin receptor inhibits astrocyte proliferation and neuronal cell death induced by transient cerebral ischemia or stab wound injury (8, 9). Ligand-mediated activation of LPA and S1P receptors has also been shown to stimulate astrocyte proliferation (5). Additionally, recent studies implicate S1P receptors on astrocytes in the astrogliosis and inflammation observed in a model of multiple sclerosis (10). Despite considerable evidence that signaling via PAR1, LPA, and S1P receptors stimulates astrogliosis, the molecular mechanism by which GPCRs activated by these ligands contribute to CNS pathophysiology is largely unknown.

We previously reported that phospholipase Cɛ (PLCɛ) is the PLC isoform responsible for mediating phosphoinositide hydrolysis in astrocytes stimulated with thrombin, LPA, or S1P (11). PLCɛ was discovered a decade ago and differs from other members of the PLC family in that it is predominantly activated by small GTPases including Ras family members and Ras homolog gene family, member A (RhoA) (12–19). The more widely studied PLCβ is regulated by GPCRs that activate Gα_q and transduce physiological responses via rapid inositol (1,4,5)trisphosphate-mediated calcium mobilization (20, 21). In contrast, PLCɛ appears to mediate more-sustained phosphoinositide hydrolysis (22) and does so through its regulation by a different set of GPCRs, particularly those that couple to Gα_12/13 to activate RhoA (23–30). A physiological role of this more-sustained signaling is suggested by studies from our laboratory demonstrating that activation of PLCɛ by thrombin leads to prolonged Ras-related protein 1 (Rap1) activation, ERK signaling, and DNA synthesis in astrocytes (11). In addition, there is a role for PLCɛ in GPCR-mediated hypertrophic growth of cardiomyocytes (31, 32). PLCɛ has also been implicated in the induction of the inflammatory mediators TGF-β and cyclooxygenase 2 (COX-2) in response to chemical carcinogens or UV irradiation (33–36). There is, however, no available information on whether GPCRs use PLCɛ signaling to regulate inflammatory gene expression.

We hypothesized that GPCR agonists including thrombin, LPA, and S1P contribute to the pathophysiological role of astrocytes by activating PLCɛ and hence inflammatory gene expression. We examined this hypothesis using astrocytes derived from brains of WT and PLCɛ KO mice. The studies presented here demonstrate that PLCɛ is required to transduce astrocyte activation by GPCRs to induction of COX-2 gene expression. The molecular mechanisms elucidated in this work reveal that the diacylglycerol (DAG)-regulated kinase, protein kinase D (PKD), is activated in a sustained manner through PLCɛ and that both PLCɛ and PKD are required to activate NF-κB and regulate proinflammatory gene expression. PLCɛ is also shown to contribute to inflammatory signaling in astrocytes in response to in vitro wounding and in vivo brain injury.

Results

PLCɛ Is Required for Induction of COX-2 in Response to Thrombin, LPA, and S1P.

Activation of PAR1, LPA, and S1P receptors has been linked to astrogliosis (4, 5, 9, 10). Because we have shown that these receptors stimulate phosphoinositide hydrolysis in astrocytes through PLCɛ (11), we asked whether PLCɛ could be the link between these GPCRs and downstream inflammatory processes associated with astrogliosis. COX-2 and other cytokines were used as markers of inflammation. In WT astrocytes treated with 5 nM thrombin for 1 h, COX-2 mRNA increased nearly 10-fold. This response was markedly attenuated in astrocytes from PLCɛ KO mice (Fig. 1A). The increase in COX-2 mRNA was accompanied by a threefold increase in COX-2 protein in thrombin-treated WT astrocytes, which was also significantly diminished in PLCɛ KO astrocytes (Fig. 1B). S1P also increased mRNA levels for COX-2, as well as for IL-1β and IL-6, and these responses were significantly attenuated by deletion of PLCɛ (Fig. S1). The ability of S1P and LPA to increase COX-2 protein was also attenuated in PLCɛ KO astrocytes (Fig. 1B). In contrast, carbachol, which activates muscarinic cholinergic receptors that do not signal through PLCɛ (11), did not induce COX-2.

Fig. 1.

PLCɛ is required for induction of COX-2 in response to thrombin, LPA, and S1P. (A) COX-2 mRNA levels in primary WT and PLCɛ KO astrocytes treated with thrombin (5 nM) or vehicle (control) for 1 h were assessed by quantitative PCR. Fold increase is expressed relative to the WT or KO averaged controls. Data shown are the mean ± SEM of values (n = 6) from three independent experiments. (B) Western blot and quantification of COX-2 protein levels following 6-h vehicle (control), thrombin (5 nM), LPA (10 μM), S1P (5 μM), or carbachol (500 μM) treatment in WT and PLCɛ KO astrocytes. Western blots of triplicates from one representative experiment are shown. The bar graph is quantitated data representing the mean ± SEM of values (n = 9) from three independent experiments. COX-2 protein levels were normalized to GAPDH and expressed relative to the WT or KO averaged controls. *P < 0.01 between control and agonist treatment; ^#P < 0.05 and ^##P < 0.01 between agonist-treated groups, one-way ANOVA.

NF-κB Is Activated Through PLCɛ and Required for COX-2 Expression.

NF-κB is known to regulate a number of inflammatory genes including COX-2. We examined translocation of the p65 subunit of NF-κB from the cytosol to the nucleus as a measure of thrombin-stimulated NF-κB activation. Thrombin induced a significant increase in nuclear p65 in WT astrocytes, but this response was absent in PLCɛ KO cells (Fig. 2A). Carbachol, which did not induce COX-2 expression, also failed to induce NF-κB activation (Fig. 2A). To demonstrate that NF-κB activation mediates thrombin-induced COX-2 expression, we examined the effects of BMS-345541, an inhibitor of the upstream IκB kinase (IKK). BMS-345541 has been shown to selectively inhibit IKK relative to a panel of other serine/threonine and tyrosine kinases and prevent activation of NFκB (37). The ability of thrombin to increase COX-2 expression was inhibited by 80% following treatment with BMS-345541 (Fig. 2B). COX-2 expression in response to LPA and S1P was likewise abolished when NF-κB activation was inhibited with BMS-345541 (Fig. 2B).

Fig. 2.

NF-κB is activated through PLCɛ and required for COX-2 expression. (A) NF-κB activation was assessed by measuring increases in p65 in the nuclear fraction of WT and PLCɛ KO astrocytes following 1-h vehicle, thrombin (5 nM), or carbachol (500 μM) treatment. Nuclear p65 protein levels were normalized to lamin A/C and expressed relative to the WT or KO averaged controls. Values from three independent experiments were quantitated as the mean ± SEM. (B) WT astrocytes were pretreated with BMS-345541 (5 μM) for 1 h before 6 h of treatment with vehicle, thrombin (5 nM), LPA (10 μM), or S1P (5 μM). COX-2 protein levels were normalized to GAPDH and expressed relative to the averaged ± inhibitor controls. Mean ± SEM of values (n = 9) from three independent experiments are shown. *P < 0.05 and **P < 0.01 between control and agonist treatment; ^#P < 0.05 and ^##P < 0.01 between agonist- ± drug-treated groups, one-way ANOVA.

Scratch Wounding Induces COX-2 Through PLCɛ and NF-κB.

To explore the functional implications of this pathway in the astrocytic response to injury in vitro, we applied multidirectional scratch to monolayers of WT and PLCɛ KO astrocytes. Expression of COX-2 protein was increased nearly 2.5-fold at 8 h after scratch in WT astrocytes but not in PLCɛ KO astrocytes (Fig. 3A). Blocking NF-κB activation with BMS-345541 fully prevented COX-2 induction, implicating NF-κB activation in this in vitro wounding response (Fig. 3B). Because no exogenous agonist was added to activate PLCɛ, we surmised that scratch elicits release of GPCR ligands or other soluble activators of PLCɛ into the cell medium. This was tested by applying media collected from scratched cells to naïve cells. Conditioned medium from scratched cells significantly increased COX-2 protein expression in naïve astrocytes from WT mice but not in those from PLCɛ KO mice (Fig. 3C). These data indicate that factors released during scratch injury also signal through PLCɛ to induce COX-2.

Fig. 3.

Scratch wounding induces COX-2 through PLCɛ and NF-κB. (A) Confluent monolayers of WT and PLCɛ KO astrocytes were scratched for 8 h and COX-2 protein was assessed via Western blot. (B) WT astrocytes were pretreated with BMS-345541 (5 μM) before wounding and COX-2 detection. (C) Media from scratched WT astrocytes were applied to naïve WT and KO astrocytes and COX-2 expression was detected. COX-2 protein levels were normalized to GAPDH and expressed relative to the WT or KO averaged controls (A and C) or to the averaged ± inhibitor controls (B). Representative Western blots are shown and data were quantitated as the mean ± SEM of values (n = 9) from three independent experiments. *P < 0.01 between control and scratch wounding (A and B) or conditioned media (C); ^#P < 0.01 between scratch wounding (B) or conditioned media (C), one-way ANOVA.

PKD Is Activated Through PLCɛ and Mediates NF-κB Activation and COX-2 Expression.

Phosphoinositide hydrolysis leads to generation of DAG, which in turn regulates DAG-sensitive protein kinases including protein kinase C (PKC) and its downstream target PKD. We examined the involvement of PLCɛ in the activation of PKD by thrombin, LPA, and S1P. PKD activation was assessed by measuring its phosphorylation at the S916 autophosphorylation site. Thrombin increased PKD phosphorylation in WT astrocytes 2.5-fold at 1 h, and phosphorylation was sustained for up to 6 h after thrombin treatment (Fig. 4A). PKD activation was increased to an even greater extent in WT astrocytes treated for 1 h with LPA and S1P (5.5- and 7.5-fold, respectively) and remained elevated above basal at 6 h (Fig. 4A). The activation of PKD by all three ligands was dramatically reduced in astrocytes from PLCɛ KO mice. Carbachol also induced PKD activation at early times (15 min), but the response was more modest and neither sustained nor PLCɛ-dependent (Fig. 4A).

Fig. 4.

PKD is activated through PLCɛ and mediates NF-κB activation and COX-2 expression. (A) Time course of phosphorylation of PKD (p-PKDS916) in WT (solid line) and PLCɛ KO (dotted line) astrocytes treated with vehicle (control, plotted as t = 0) and thrombin (5 nM), LPA (10 μM), S1P (5 μM), or carbachol (500 μM) for 1 and 6 h. Representative Western blots are shown for the 1-h time point for thrombin, LPA, and S1P and for the time course for carbachol. The p-PKDS916 protein levels were normalized to total PKD and expressed relative to the WT or KO averaged controls of values (n = 9) from three independent experiments quantitated as the mean ± SEM. (B) NF-κB activation was measured by detecting increases in p65 in the nuclear fraction of WT and PLCɛ KO astrocytes treated with thrombin (5 nM) for 1 h following knockdown of PKD with siRNA (2 μM). p65 protein levels were normalized to lamin A/C and expressed relative to the averaged siRNA control. Representative Western blots are shown and data were quantitated as the mean ± SEM of values from three independent experiments. (C) COX-2 levels were assessed by Western blotting after thrombin treatment (5 nM, 6 h) following knockdown of PKD with siRNA (2 μM). COX-2 protein levels were normalized to GAPDH and expressed relative to the averaged siRNA control. Representative Western blots are shown and data were quantitated as the mean ± SEM of values (n = 9) from three independent experiments. (D) PLCɛ KO astrocytes were infected with WT FLAG-tagged PLCɛ adenovirus or catalytically dead FLAG-tagged PLCɛ adenovirus before vehicle, thrombin (5 nM), or LPA (10 μM) treatment for 6 h. COX-2 protein levels were assessed by Western blotting. COX-2 protein levels were normalized to GAPDH and expressed relative to the EYFP control. Representative Western blots are shown and data were quantitated as the mean ± SEM of values (n = 6) from three independent experiments. *P < 0.05 and **P < 0.01 between control and agonist treatment; ^#P < 0.05 and ^##P < 0.01 between agonist-treated groups, one-way ANOVA.

We previously showed that RhoA is required for PLCɛ activation by thrombin (11). To determine whether PKD activation and subsequent COX-2 expression are also downstream effectors of RhoA activation, we treated WT astrocytes with C3 exoenzyme, an agent that ribosylates and prevents RhoA signaling. Thrombin was unable to increase PKD activation or COX-2 expression when RhoA function was blocked (Fig. S2 A and B). Inhibition of RhoA also prevented COX-2 induction by scratch wounding (Fig. S2C).

To determine whether PKD mediates NF-κB activation and subsequent COX-2 induction, we used siRNA to knock down PKD in WT astrocytes. Treatment with siRNA significantly decreased PKD expression (Fig. S3). In astrocytes transfected with control scrambled siRNA, thrombin induced a marked increase in nuclear p65; this response was abolished in cells transfected with PKD siRNA (Fig. 4B). The ability of thrombin to increase COX-2 expression was likewise abolished by PKD knockdown (Fig. 4C).

Last, to demonstrate that PLCɛ and its catalytic activity are required for COX-2 expression, we performed a rescue experiment. In PLCɛ KO astrocytes infected with WT PLCɛ adenovirus, the induction of COX-2 expression by thrombin and LPA treatment was restored (Fig. 4D). In contrast, PLCɛ KO astrocytes infected with catalytically dead PLCɛ adenovirus still failed to respond to thrombin or LPA with significant increases in COX-2 (Fig. 4D).

In Vivo Stab Wound Injury Induces COX-2 Through PLCɛ.

Penetrating head injury destroys neurons and leads to the generation of reactive astrocytes. Our findings suggest that GPCR agonists present at sites of injury could act through PLCɛ to drive astrogliosis. To extend the observation that PLCɛ mediates inflammatory signaling from GPCRs to an in vivo setting, cortical stab wound was applied to WT and PLCɛ KO mice as described (38). Mice were killed at day 7 after wounding and the cortex was removed for Western blotting. We observed significant increases in COX-2 expression and PKD activation in the cortical region of injured WT mice (Fig. 5 A, B, and E). Monocyte chemotactic protein-1 (MCP-1), another marker of inflammation, and GFAP, a marker of astrogliosis, were also significantly increased in the WT mouse cortex following stab injury (Fig. 5 A, C, and D). Neither of these markers was increased following stab wound injury in brains of PLCɛ KO mice (Fig. 5).

Fig. 5.

In vivo stab wound injury induces COX-2 through PLCɛ. WT and PLCɛ KO mice at 8 wk of age were subjected to stab wound injury as described in Methods. Seven days following injury, brains were removed and lysates were prepared for Western blotting. (A) Representative Western blots of COX-2, MCP-1 (another inflammatory mediator), GFAP (a marker of astrogliosis), and phosphorylated PKD are shown for two animals per group. (B–E) Quantification of Western blot data from n = 8–10 mice per group. COX-2, GFAP, and MCP-1 protein levels were normalized to GAPDH, and p-PKDS916 was normalized to total PKD. *P < 0.05 and **P < 0.01 between control and stab wound injury; ^#P < 0.01 between stab wound injuries, one-way ANOVA.

Discussion

PAR1, LPA, and S1P receptor activation contribute to astrogliosis, a process associated with inflammation (4, 5, 9, 10). The mechanism by which signaling through these GPCRs induces neuroinflammation and participates in pathophysiological response to ischemia or brain injury is not well-understood. PAR1, LPA, and S1P receptors are among the most efficacious of the GPCRs in their ability to couple to Gα_12/13 and activate the low-molecular weight G protein RhoA (23–30). RhoA signaling has been implicated in the regulation of inflammatory genes, including COX-2, in endothelial and NIH 3T3 cells (39–43). However, the molecular mechanisms by which GPCRs that activate RhoA lead to NF-κB and inflammatory gene expression have not been fully elucidated. The studies presented here demonstrate that stimulation of PLCɛ and the subsequent sustained activation of PKD transduce GPCR activation to inflammatory gene expression (Fig. 6).

Fig. 6.

Schematic of pathway by which PLCɛ mediates GPCR activation and inflammatory responses in astrocytes. CDC25 is the guanine nucleotide exchange factor domain, PH is the Pleckstrin Homology domain, X and Y comprise the catalytic domain, and the RA2 is the Ras association domain.

PKD, a downstream target for regulation by PKC, also requires DAG for its activation (44–46). Because DAG generation and PKC activation occur downstream of phosphoinositide hydrolysis, it is generally assumed that PKD is activated through the canonical Gα_q/PLCβ signaling cascade (47, 48). We show here, however, that carbachol, which activates muscarinic receptors that couple through Gα_q/PLCβ to regulate phosphoinositide hydrolysis, induces only transient PKD activation. In contrast, thrombin, LPA, and S1P, which activate PLCɛ (11), lead to sustained PKD activation. This is consistent with previous data showing that PLCβ knockdown significantly reduces short-term inositol phosphate production, whereas knockdown of PLCɛ affects the longer-term agonist-induced generation of inositol phosphates (22). PLCɛ is unique relative to other isoforms of PLC in that it has an extended carboxyl terminus that contains a CDC25 domain that functions as a Rap guanine nucleotide exchange factor (11, 15, 49). We have suggested that Rap1, activated by PLCɛ, feeds forward through the RA2 domain on PLCɛ to further activate and prolong generation of second messengers such as DAG (12). The ability of PLCɛ to mediate sustained signaling could be of particular importance with regard to its proposed role in pathophysiology.

The data presented here support the hypothesis that activation of PLCɛ and sustained PKD signaling are required for NF-κB activation and COX-2 expression. In this regard, we show that transient PKD activation, as observed in response to carbachol, fails to increase NF-κB activation and subsequent COX-2 expression. In addition, we demonstrate that in the absence of sustained PKD activation (i.e., in the PLCɛ KO or when PKD is down-regulated), thrombin treatment fails to activate NF-κB or increase COX-2 expression. Previous studies have shown that Gα_12/13 and Rho signaling can lead to PKD or NF-κB activation and COX-2 expression (43, 48, 50, 51), but the molecular mechanism by which Rho signaling is transduced into these responses has not been delineated. These studies show that PLCɛ, and its catalytic activity, are required for induction of COX-2 expression, and that this occurs through PKD and NF-κB activation, defining a critical and previously unknown link between GPCR-mediated Rho activation and inflammatory processes. Our data further argue that the ability of PLCɛ to support sustained signals defines its unique role in regulation of astrocyte function.

The physiological significance of our observations is supported by results from an in vitro wounding assay and an in vivo cortical stab wound model. These studies confirm that PLCɛ is a mediator of inflammation and in addition provide indirect evidence that PLCɛ is activated by endogenous mediators released in response to stress. The scratch wounding response could be recapitulated by the addition of conditioned medium to unscratched cells. Similarly, during in vivo stab wound, breakdown of the blood–brain barrier or generation of endogenous ligands would appear to signal via PLCɛ to elicit PKD activation and induce COX-2, GFAP, and MCP-1 expression. Future studies, beyond the scope of this manuscript, will be needed to determine what agonists are released to activate PLCɛ in vitro and in vivo. Thrombin, LPA, and S1P are possible candidates because injection of these agonists into the striatum of mice induces astrogliosis (5, 9). Of additional interest, ligands for orphan GPCRs could signal through receptor coupling to Gα_12/13 to converge on RhoA and PLCɛ and induce inflammatory gene expression.

Up-regulation of IL-1β, IL-6, and COX-2 has important disease implications: COX-2 plays a proinflammatory role in neurodegenerative disease and is increased in cerebral ischemia and traumatic brain injury (52–54), whereas IL-1β and IL-6 are up-regulated and contribute to damage of the myelin sheath during multiple sclerosis (10). Delineating a signaling pathway that links GPCRs through RhoA to PLCɛ and astrogliosis has the potential to reveal new therapeutic targets to limit neuroinflammation.

Methods

Agonists and Inhibitors.

Sources were as follows: thrombin (Enzyme Research Laboratories), LPA (Avanti Polar Lipids), S1P (Avanti Polar Lipids), BMS-345541 (Sigma-Aldrich), and C3 Rho inhibitor cell-permeable (Cytoskeleton).

Animals.

All procedures were performed in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals and were approved by the Institutional Animal Care and Use Committee at the University of California at San Diego. Generation of homozygous C57BL6/Sv129 PLCɛ KO mice has been described previously (31). PLCɛ heterozygous mice were bred to generate homozygous animals.

Primary Culture of Astrocytes.

Astrocytes were isolated from postnatal days 1–3 WT and KO mice as previously described (11). Purity of astrocytes was determined to be ∼95% based on GFAP staining. In all experiments, WT and PLCɛ KO astrocytes were used at passage 2.

PKD siRNA Transfections.

Predesigned mouse siRNA for PKD and control siRNA were purchased from Bioneer. WT astrocytes were transfected with 2 μM siRNA using DharmaFECT 3 transfection reagent (Thermo Scientific). siRNA and DharmaFECT 3, in a 1:3 ratio, respectively, were individually incubated in Opti-MEM media (Gibco) at room temperature for 10 min, mixed, and incubated further for 20 min. The siRNA/DharmaFECT 3 mixture was added to plates containing fresh media. Following overnight incubation, astrocytes were serum-starved for 18–24 h before agonist treatment and analysis by Western blot.

Transduction of Astrocytes with Adenovirus.

PLCɛ KO astrocytes were plated and infected the next day for 4–6 h in complete media with 200 multiplicity of infection of adenovirus expressing FLAG-tagged WT PLCɛ, FLAG-tagged catalytically dead PLCɛ, or enhanced yellow fluorescent protein (EYFP) as previously described (11, 32, 55). Following 4–6 h of infection, astrocytes were washed and serum-starved for 18–24 h before agonist treatment.

Quantitative PCR.

Total RNA was extracted from agonist-treated WT and PLCɛ KO astrocytes using an RNeasy Kit (Invitrogen) as previously described (11). cDNA was amplified using TaqMan Universal Master Mix in the presence of gene-specific primers for COX-2, IL-1β, and IL-6, with GAPDH used as an internal control (Applied Biosystems). Data were normalized to internal GAPDH, and fold change was determined according to a published protocol (56).

Nuclear Fractionation Experiments.

WT and PLCɛ KO astrocytes were grown to confluence in 10-cm dishes. Before agonist treatment, cells were serum-starved for 18–24 h. Cells were fractionated according to a previously reported protocol (57). p65 expression in the nuclear fraction was assessed by Western blotting, and purity was determined with the cytosolic marker Rho GDP dissociation inhibitor (RhoGDI) and the nuclear marker lamin A/C.

Western Blotting.

Freshly removed brain cortices were snap-frozen in liquid nitrogen and lysates were prepared with RIPA buffer. Astrocyte lysates were prepared with the same lysis buffer. Western blot analysis was performed according to the previous described protocol (58). The antibodies used for immunoblotting were as follows: NF-κB p65 (H-286) from Santa Cruz Biotechnology; RhoGDI, p-PKD (Ser916), PKD, MCP-1, FLAG tag (DYKDDDDK), and lamin A/C from Cell Signaling Technology; and COX-2 from Cayman.

Wounding Assays.

WT and PLCɛ KO astrocytes were plated in six-well plates, grown to confluence, and serum-starved 18–24 h before wounding as previously described (59). The wound was made by scraping astrocytes with a 200-μL pipette tip six to eight times bidirectionally across the dish. Eight or 24 h after wounding, cells were harvested for analysis by Western blotting or conditioned media were collected.

Cortical Stab Wound.

Bilateral cortical stab wound was performed in WT and PLCɛ KO (8 wk of age) as previously described (38). Briefly, beginning at the level of bregma, a 26-gauge needle was inserted −2.0 mm, and 1.5 mm lateral of the midline, to a depth of 1.4 mm and extended three times in the cerebral cortex. Following 7 d of injury, animals were killed and the brain was removed. The brain hemisphere was removed and snap-frozen in liquid nitrogen and lysates were prepared with RIPA buffer. Sham-operated and noninjured mice were determined to have equivalent basal levels of COX-2 expression, and thus the latter group was used as the control in most experiments.

Statistical Analysis.

Statistical differences were determined using Tukey’s multicomparison analysis after one-way ANOVA with Prism software (GraphPad). P < 0.05 was considered significant.