Thrombin Promotes Sustained Signaling and Inflammatory Gene Expression through the CDC25 and Ras-associating Domains of Phospholipase Cϵ*

Background: PLCϵ activation is sustained but the underlying regulatory mechanisms are unknown. Results: PLCϵ gene deletion and rescue demonstrate that Rap1 activation by the CDC25 domain and regulation via the RA2 domain sustain thrombin-mediated PLCϵ and PKD activation and inflammatory gene expression. Conclusion: Unique domains and compartmentalization of PLCϵ allow for sustained GPCR signaling. Significance: Targeting these PLCϵ domains could ameliorate pathophysiological inflammation. Phospholipase C-epsilon (PLCϵ) plays a critical role in G-protein-coupled receptor-mediated inflammation. In addition to its ability to generate the second messengers inositol 1,4,5-trisphosphate and diacylglycerol, PLCϵ, unlike the other phospholipase C family members, is activated in a sustained manner. We hypothesized that the ability of PLCϵ to function as a guanine nucleotide exchange factor (GEF) for Rap1 supports sustained downstream signaling via feedback of Rap1 to the enzyme Ras-associating (RA2) domain. Using gene deletion and adenoviral rescue, we demonstrate that both the GEF (CDC25 homology domain) and RA2 domains of PLCϵ are required for long term protein kinase D (PKD) activation and subsequent induction of inflammatory genes. PLCϵ localization is largely intracellular and its compartmentalization could contribute to its sustained activation. Here we show that localization of PLCϵ to the Golgi is required for activation of PKD in this compartment as well as for subsequent induction of inflammatory genes. These data provide a molecular mechanism by which PLCϵ mediates sustained signaling and by which astrocytes mediate pathophysiological inflammatory responses.

Phospholipase C-epsilon (PLC⑀) plays a critical role in G-protein-coupled receptor-mediated inflammation. In addition to its ability to generate the second messengers inositol 1,4,5-trisphosphate and diacylglycerol, PLC⑀, unlike the other phospholipase C family members, is activated in a sustained manner. We hypothesized that the ability of PLC⑀ to function as a guanine nucleotide exchange factor (GEF) for Rap1 supports sustained downstream signaling via feedback of Rap1 to the enzyme Rasassociating (RA2) domain. Using gene deletion and adenoviral rescue, we demonstrate that both the GEF (CDC25 homology domain) and RA2 domains of PLC⑀ are required for long term protein kinase D (PKD) activation and subsequent induction of inflammatory genes. PLC⑀ localization is largely intracellular and its compartmentalization could contribute to its sustained activation. Here we show that localization of PLC⑀ to the Golgi is required for activation of PKD in this compartment as well as for subsequent induction of inflammatory genes. These data provide a molecular mechanism by which PLC⑀ mediates sustained signaling and by which astrocytes mediate pathophysiological inflammatory responses.
First identified in Caenorhabditis elegans as a Let-60 Rasbinding protein (24), PLC⑀ was demonstrated to contain the X, Y, and C2 domains characteristic of enzymes in the phospholipase C (PLC) family (24). Like the other PLC family members, PLC⑀ was found to function to hydrolyze phosphatidylinositol 4,5-bisphosphate to generate the second messengers inositol 1,4,5-trisphosphate (IP 3 )and diacylglycerol (DAG) (25). Additionally, PLC⑀ has an extended N-terminal, which contains a CDC25 domain not found in the other PLCs and which functions as a GEF for the low molecular weight G-protein Rap1 (26 -28). This domain has been shown to be important for PLC⑀ localization to the Golgi, and its deletion leads to more transient PLC⑀ localization to this compartment (26). Moreover, PLC⑀ was found to be uniquely regulated by the small G-protein RhoA through a 65 amino acid sequence within the Y domain (4 -8), as well as by other Ras family members through their interactions with the RA2 domain (5,29).
Of particular interest, while both PLC-beta (PLC␤) and PLC⑀ are regulated in response to endothelin-1 (ET-1), lysophosphatidic acid (LPA), and thrombin, knockdown of PLC␤ inhibits inositol phosphate generation at short times (1-3 min) whereas knockdown of PLC⑀ is required to inhibit inositol phosphate generation at longer times (10 -60 min) (9). Using primary astrocytes from PLC⑀ knock-out (KO) mice, we demonstrated that PLC⑀ is needed for the sustained activation of protein kinase D (PKD) which occurs in response to ligands that activate G␣ 12/13 /Rho whereas ligands that stimulate G␣ q /PLC␤ lead to a more transient activation of PKD (2). Our data also revealed that the sustained activation of PKD is necessary for induction of inflammatory gene expression (2).
We postulate and demonstrate here that the non-catalytic CDC25 and RA2 domains of PLC⑀ are essential components required for sustained PKD activation and inflammatory gene expression. This conclusion is supported by studies using astrocytes from PLC⑀ KO mice and rescue by adenoviral expression of wild-type (WT) and mutant PLC⑀. A role for compartmentalized PLC⑀ signaling at the Golgi is also established. We conclude that PLC⑀ signaling, initiated by GPCR stimulation and RhoA binding, is sustained by a feedback mechanism involving the CDC25 domain as a generator of active Rap1 and the RA2 domain of PLC⑀ as its effector.

Experimental Procedures
Animals-All procedures were performed in accordance with NIH Guide and Care and Use of Laboratory Animals and approved by the Institutional Animal Care and Use Committee at the University of California San Diego. Generation of homozygous C57BL6/Sv129 PLC⑀ KO mice has been described previously (19). PLC⑀ heterozygous KO mice were bred to generate homozygous KO animals.
Primary Culture of Astrocytes-Astrocytes were isolated from P1-P3 postnatal WT and KO mice as previously described (1). Purity of astrocytes was determined to be ϳ95% based on GFAP staining. In all experiments, WT and PLC⑀ KO astrocytes were used at passage 2.
Fluorescence Resonance Energy Transfer-Astrocytes were plated onto glass coverslips in 35-mm dishes and Golgi or plasma membrane-targeted DKAR constructs were transfected using Dharma-FECT 3 transfection reagent at a 1:3 DNA: Dharma-FECT3 ratio (Thermo Scientific). Cells were serum starved the next day for 18 -24 h and then washed with HBSS (Gibco) prior to collecting DKAR images as described previously (30) on a Zeiss Axiovert microscope (Carl Zeiss Micro-Imaging, Inc.) with a cooled charge-couple device camera (Photometric) controlled by MetaFluor software (Universal Imaging Corp.). Images were collected at baseline for 4 min followed by treatment with thrombin for up to twelve minutes.
Immunofluorescence-Astrocytes were grown on glass slides and infected with 150 moi of mCherry-PLC⑀. Following serum starvation for 18 -24 h, cells were fixed using 4% paraformaldehyde in PBS and then permeabilized with 0.1% Triton in PBS for 5 min before blocking with 5% BSA in PBS and 10% normal goat serum. Antibody GM-130 (BD Biosciences) was diluted in the blocking solution before addition of Alexa 488 mouse. Cells were then washed and mounted with coverslips using Vectashield with DAPI (Vector Labs). Pictures were acquired using the Olympus FV-1000 confocal microscope.
Quantitative-PCR (q-PCR)-Total RNA was extracted from agonist treated WT and PLC⑀ KO astrocytes using an RNeasy kit (Invitrogen) as previously described (2). cDNA was ampli-fied using the TaqMan Universal Master Mix in the presence of gene-specific primers for IL-6 and COX-2 with GAPDH used as an internal control (Applied Biosystems). Data were normalized to internal GAPDH and fold change determined according to published protocol (31).
Western Blotting-Astrocyte lysates were prepared in RIPA buffer. Western blot analysis was performed according to the previous described protocol (32). The antibodies used for immunoblotting were as follows: p-PKD (Ser-916), PKD, and GAPDH from Cell Signaling Technology, COX-2 from Cayman, and Rap1 from Santa Cruz. Immunoblots shown represent a single gel; images are split in cases where unnecessary lanes were removed.
Rap Pull-down-Serum-starved astrocytes were treated with thrombin for the indicated times and then lysed with buffer containing 50 mM Tris, 150 mM NaCl, 5 mM MgCl 2 , 0.1% Triton X-100, and 10% glycerol. Lysates were incubated with 20 g of RalGDS RBD conjugated to glutathione beads for 45 min at 4°C and then centrifuged to pellet the agarose beads. Agarose beads were washed with lysis buffer, and the pellets were resuspended in 2.5ϫ Laemmli sample buffer containing DTT, boiled for 10 min, and centrifuged. Western blot analysis was then performed.
Statistical Analysis-Statistical differences were determined using Tukey's multicomparison analysis after one-way ANOVA with Prism software (GraphPad). p Ͻ 0.05 was considered significant.

Rap1 Activation Is Sustained and Requires the CDC25
Domain of PLC⑀-Our earlier work demonstrated that endogenous PLC⑀ functions as a Rap1GEF that is activated in response to GPCR stimulation (1,16). Here we examined the kinetics of Rap1 activation in primary mouse astrocytes stimulated with thrombin, measuring activated Rap1 using a pulldown assay. Rap1 activation was significantly elevated at 15 min, further increased at 1 h, and sustained for up to 6 h (Fig.  1A). This response was absent in astrocytes from PLC⑀ KO mice (Fig. 1A). To demonstrate that it is the CDC25 domain of PLC⑀ that functions as the thrombin-regulated Rap1GEF, we used a previously generated mutant PLC⑀ construct in which the CDC25 domain was deleted (CDC25⌬) (16). Adenoviral expression of WT PLC⑀ in PLC⑀ KO astrocytes lead to significant recovery of Rap1 activation (Fig. 1B). In contrast Rap1 activation was not recovered in the KO cells expressing the CDC25⌬ mutant PLC⑀ (Fig. 1B).
PKD Activation and COX-2 Expression Require the CDC25 Domain of PLC⑀-We previously demonstrated that sustained activation of PKD and subsequent COX-2 expression in astrocytes require PLC⑀ (2). To determine whether Rap1 activation plays a role in these responses we compared the ability of the CDC25 domain mutant and WT PLC⑀ to support activation of PKD and induction of COX-2 in PLC⑀ KO astrocytes. The activation of PKD was assessed using an antibody directed at the PKD autophosphorylation site (Ser-916) and COX-2 expression was assessed by Western blotting. Thrombin activation of PKD was significantly attenuated in the CDC25⌬ mutant compared with those expressing WT PLC⑀ ( Fig. 2A). COX-2 expression was also restored by WT PLC⑀, but not the CDC25⌬ mutant (Fig. 2B).
PKD Activation and COX-2 Expression Require the RA2 Domain of PLC⑀-The data presented above provide evidence that the CDC25 domain of PLC⑀ is important not only for sustained Rap1 activation but also for PKD activation and induction of COX-2 expression. PKD activation is mediated through second messengers generated by phosphoinositide hydrolysis, specifically DAG and its target protein kinase C. Accordingly, we hypothesized that continued PLC⑀-mediated phosphoinositide hydrolysis and DAG generation are facilitated by generation of activated Rap1 via the CDC25 domain. A mechanism through which this feedback could occur would be via Rap1 interaction with the RA2 domain. This domain was previously shown to regulate PLC⑀ activity in response to heterologously expressed or recombinant Rap1 (5,26). We utilized a RA2 mutant of PLC⑀ (PLC⑀ K2150E), shown to lack PLC activation by Ras family proteins (29), to test this hypothesis. In contrast to WT PLC⑀ which supported thrombin stimulated PKD and COX-2 expression in PLC⑀ KO cells (Fig. 3, A and B), the PLC⑀ RA2 mutant was ineffective at supporting thrombin induced PKD activation and induction of COX-2 (Fig. 3, A and B).
The CDC25 and RA2 Domains Are Required for IL-6 mRNA Expression-To extend our findings on COX-2 regulation to another inflammatory gene, we assessed the induction of interleukin-6 (IL-6). Thrombin treatment increased IL-6 mRNA expression, and this response was also dependent on PLC⑀ (Fig.  4A). To demonstrate the importance of the CDC25 and RA2 domains in thrombin stimulated IL-6 induction, we expressed WT PLC⑀, the CDC25⌬ mutant, or the RA2 domain mutant in KO astrocytes. In contrast to what was observed in cells expressing WT PLC⑀, thrombin failed to induce IL-6 mRNA in cells expressing either the CDC25⌬ or RA2 mutant PLC⑀ (Fig. 4B).
PLC⑀ Is Localized to the Golgi-The observation that heterologously expressed PLC⑀ localizes to an intracellular perinuclear compartment (26,28,33), confirmed by recent studies in cardiomyocytes (10), has important signaling implications. We hypothesized that this unique localization is critical for the feedback mechanism and sustained signaling proposed above. To examine PLC⑀ localization in primary astrocytes, we expressed an adenoviral mCherry-PLC⑀ construct in KO PLC⑀ astrocytes. The mCherry fluorescence was most intense in an area surrounding the DAPI stained nucleus and was co-localized with the Golgi marker GM-130 (Fig. 5).
PKD Is Activated at the Golgi in a PLC⑀-dependent Manner-To determine whether PKD is activated through PLC⑀ signaling at the Golgi, we expressed a FRET reporter for PKD (Golgi-DKAR) that is targeted to this compartment (34,35). Thrombin treatment significantly increased the FRET signal in astrocytes from WT mice (Fig. 6A). In contrast PKD activation was not evidenced by an increase in the FRET signal at the Golgi in cells lacking PLC⑀ (Fig. 6B). Parallel studies using a plasma membrane targeted PKD activity reporter revealed minimal thrombin induced activation of PKD at the plasma membrane (Fig.  6C). Thus PKD activation in response to thrombin occurs through PLC⑀ localized at the Golgi.
Intact Golgi Is Necessary for Rap1 Activation, PKD Activation, and COX-2 Expression-We demonstrated in studies above that PLC⑀ activation is required for thrombin stimulated activation of Rap1 and PKD and for induction of COX-2. To demonstrate the importance of compartmentalization at the Golgi, we disrupted the Golgi with brefeldin A (BFA) (Fig. 7A). Astrocytes pretreated with BFA showed significantly reduced Rap1 activation by thrombin (Fig. 7B). PKD activation (Fig. 7C) and COX-2 expression (Fig. 7D) were also markedly disrupted in BFA-treated cells compared with vehicle treated cells. A, Rap1 activation was assessed by measuring levels of Rap1-GTP after Rap1 pull-down and Western blot in WT and PLC⑀ KO astrocytes treated with vehicle or thrombin (5 nM) for 15 min, 1 h, and 6 h. Rap1-GTP was normalized to total Rap1 and data normalized to its own control. Representative Western blots are shown and data quantitated as the mean Ϯ S.E. (n ϭ 6) of three independent experiments. B, Rap1 activation was assessed in KO astrocytes that were infected with either WT PLC⑀ adenovirus or mutant PLC⑀ adenovirus that lacks the CDC25 domain (CDC25⌬) followed by vehicle or thrombin (5 nM) treatment for 1 h and 6 h. Rap1-GTP was normalized to total Rap1 and data normalized to each of its own control. Representative Western blots are shown and data quantitated as the mean Ϯ S.E. (n ϭ 9) of three independent experiments. *, p Ͻ .05 and **, p Ͻ .01 between control and thrombin treatment; #, p Ͻ .05 and ##, p Ͻ .01 between thrombin treatments, one-way ANOVA.

FIGURE 2. PKD activation and COX-2 expression require the CDC25 domain of PLC⑀.
A, PKD phosphorylation (p-PKDS916) was measured in PLC⑀ KO astrocytes that were infected with WT PLC⑀ or CDC25⌬ adenovirus followed by vehicle or thrombin (5 nM) treatment for 1 h. The p-PKDS916 protein levels were normalized to total PKD and expressed relative to its own averaged control. Representative Western blots are shown and data quantitated as the mean Ϯ S.E. (n ϭ 9) of four independent experiments (control error bars are small but present). B, COX-2 protein levels were measured in PLC⑀ KO astrocytes infected with WT PLC⑀ or CDC25⌬ adenovirus followed by thrombin (5 nM) treatment for 1 h and 6 h. COX-2 protein levels were normalized to GAPDH and expressed relative to its own control. Representative Western blots are shown and data quantitated at the 6 h time point as the mean Ϯ S.E. (n ϭ 6) of three independent experiments. *, p Ͻ .01 between control and thrombin treatment; #, p Ͻ .01 between thrombin treatments, one-way ANOVA. FIGURE 3. PKD activation and COX-2 expression require the RA2 domain of PLC⑀. A, PKD phosphorylation (p-PKDS916) was measured in PLC⑀ KO astrocytes infected with WT PLC⑀ or RA2 mutant (K2150E) adenovirus followed by thrombin (5 nM) treatment for 1 h. The p-PKDS916 protein levels were normalized to total PKD and expressed relative to their own averaged control. Representative Western blots and data quantitated as the mean Ϯ S.E. (n ϭ 6) of three independent experiments. B, COX-2 protein levels were measured in PLC⑀ KO astrocytes infected with WT PLC⑀ or RA2 mutant adenovirus followed by treatment with thrombin (5 nM) for 1 h and 6 h. COX-2 protein levels were normalized to GAPDH and expressed relative to its own control. Representative Western blots and data quantitated at the 6 h time point as the mean Ϯ S.E. (n ϭ 6) of three independent experiments. *, p Ͻ .05 and **, p Ͻ .01 between control and thrombin treatment; #, p Ͻ .05 and ##, p Ͻ .01 between thrombin treatments, one-way ANOVA.

Discussion
PLC⑀ has been shown to be a critical mediator in a range of disorders including cancer, CNS injury, cardiac hypertrophy, and cardiac ischemia/reperfusion injury (1-3, 10 -23). Our hypothesis is that the ability of PLC⑀ to integrate signals from GPCRs to downstream changes in gene expression contributes to the pathophysiology of these diseases. PLC⑀ has a unique structure and is compartmentalized within the cell. Data described here provide evidence that these features are central to its role in sustained inflammatory signaling in astrocytes.
Compared with other PLC family members, PLC⑀ is unique in containing a CDC25 domain that functions as a GEF for Rap1 (26,27). Previous work using heterologous expression of PLC⑀ or a PLC⑀ mutant lacking the CDC25 domain, demonstrated that the CDC25 domain is required for the Rap1 activation observed for up to 30 min following EGF stimulation (26). We previously reported that PLC⑀ is required for the Rap1 activation observed 15 min following thrombin addition (1). The data presented here are the first to show that PLC⑀ signaling is required for GPCR ligands to elicit long term activation of  Rap1, persisting for at least 6 h. We demonstrate that the unique CDC25 domain is needed for sustained agonist-induced Rap1 activation, and further show that it is necessary for subsequent activation of chronic signaling resulting in PKD activation and induction of COX-2.
Ras family members including Ras, Rap1, Rap2, and TC21 have been shown to directly interact with PLC⑀ to regulate its activity (5,29). The findings presented here establish that the RA2 domain of the enzyme is responsible for sustained signaling and suggest that this occurs through feedback by active Rap1 generated through the CDC25 domain. Specifically studies using a mutant PLC⑀, which is unable to bind Ras or Rap (29), demonstrate the requirement for the RA2 domain in long term activation of PKD and subsequent COX-2 expression, responses that require continued generation of DAG.
The CDC25 domain of PLC⑀ has been shown to be important for its cellular localization. The Kataoka group demonstrated that Rap1 is required for PLC⑀ localization to the Golgi and that this requires the CDC25 domain (26). Thus Rap1 activation occurs through the CDC25 domain and interaction of Rap1 with this domain contributes to PLC⑀ localization. Not only does Rap1 retain PLC⑀ at the Golgi but this organelle is also a rich source of PI4P that can serve as a substrate for PLC⑀ and a source of DAG (10,34,36). Work shown here, as well as recent studies in cardiomyocytes (10), demonstrate that PLC⑀ localized to the Golgi is important for PKD activation and subsequent induction of genes involved in inflammation and hypertrophy. Furthermore, localization of PLC⑀ to the Golgi has a biological consequence. Disruption of the Golgi with BFA affects downstream PLC⑀ signaling including Rap1 and PKD activation and subsequent induction of COX-2.
The protease activate receptor 1 (PAR1) has been implicated in CNS injury and disease and has been shown to induce astrocyte activation and proliferation in response to thrombin both in vitro and in vivo (37,38). PAR1 is one of the most efficacious GPCRs in coupling to G␣ 12/13 and Rho (1, 39 -44). We have previously demonstrated that thrombin stimulates phosphoinositide hydrolysis in astrocytes exclusively through PLC⑀ and FIGURE 7. Intact Golgi is necessary for Rap1 activation, PKD activation, and COX-2 expression. A, BFA (5.0 g/ml) was used to disrupt the Golgi. B, following pretreatment with BFA (5.0 g/ml), WT astrocytes were treated with vehicle and thrombin (5 nM) for 1 h and 6 h. Rap1 activation was assessed by measuring Rap1-GTP after Rap1 pull-down and Western blot. Rap1-GTP was normalized to total Rap1 and expressed relative to its own averaged control. Representative Western blots are shown and data quantitated as mean Ϯ S.E. (n ϭ 8) of four independent experiments. C, following pretreatment with BFA (5.0 g/ml), WT astrocytes were treated with thrombin (5 nM) for 1 h, and PKD phosphorylation (p-PKDS916) was assessed via Western blot. The p-PKDS916 protein levels were normalized to total PKD and expressed relative to its own averaged control. Representative Western blots are shown and data quantitated as mean Ϯ S.E. (n ϭ 8) of four independent experiments. D, COX-2 mRNA levels were measured after pretreatment with BFA (5.0 g/ml) and subsequent treatment with thrombin (5 nM) for 1 h. Fold increase is expressed relative to the averaged Ϯ inhibitor controls. *, p Ͻ .05 and **, p Ͻ .01 between control and thrombin treatment; #, p Ͻ .05 and ##, p Ͻ .01 between thrombin treatments, one-way ANOVA. that thrombin signals through Rho and PLC⑀ to mediate inflammatory gene expression in these cells (1,2). Here we demonstrate that the molecular and cellular mechanism by which PLC⑀ regulates astrocytic inflammatory gene expression requires its ability to function as a Rap1GEF, to generate activated Rap1, to further activate its RA2 domain, and to sustain PKD activation at the Golgi. As such, the domains that allow PLC⑀ to function as a critical signaling node could represent possible therapeutic targets in diseases characterized by chronic inflammation.
Author Contributions-S. S. D. designed, performed, analyzed the experiments for the figures of the paper, and wrote the paper. M. T. K. provided technical assistance in particular for Fig. 6. A. V. S. provided technical assistance for Figs. 1, 2, 3, 4, and 5. J. H. B. conceived and coordinated the study and wrote the paper. All authors reviewed the results and approved the final version of the manuscript.