Store-operated Ca2+ Entry (SOCE) Induced by Protease-activated Receptor-1 Mediates STIM1 Protein Phosphorylation to Inhibit SOCE in Endothelial Cells through AMP-activated Protein Kinase and p38β Mitogen-activated Protein Kinase*

Background: STIM1 is essential for store-operated Ca2+ entry (SOCE) in endothelial cells. Results: SOCE-activated AMPKα1-p38β signaling phosphorylates STIM1, which in turn inhibits SOCE in endothelial cells. Conclusion: SOCE-activated signaling pathway completes a negative feedback loop to regulate SOCE in endothelial cells. Significance: Selective p38β agonists may represent potential therapeutic agents to reverse the vascular leak syndrome. The Ca2+ sensor STIM1 is crucial for activation of store-operated Ca2+ entry (SOCE) through transient receptor potential canonical and Orai channels. STIM1 phosphorylation serves as an “off switch” for SOCE. However, the signaling pathway for STIM1 phosphorylation is unknown. Here, we show that SOCE activates AMP-activated protein kinase (AMPK); its effector p38β mitogen-activated protein kinase (p38β MAPK) phosphorylates STIM1, thus inhibiting SOCE in human lung microvascular endothelial cells. Activation of AMPK using 5-aminoimidazole-4-carboxamide-1-β-d-ribofuranoside (AICAR) resulted in STIM1 phosphorylation on serine residues and prevented protease-activated receptor-1 (PAR-1)-induced Ca2+ entry. Furthermore, AICAR pretreatment blocked PAR-1-induced increase in the permeability of mouse lung microvessels. Activation of SOCE with thrombin caused phosphorylation of isoform α1 but not α2 of the AMPK catalytic subunit. Moreover, knockdown of AMPKα1 augmented SOCE induced by thrombin. Interestingly, SB203580, a selective inhibitor of p38 MAPK, blocked STIM1 phosphorylation and led to sustained STIM1-puncta formation and Ca2+ entry. Of the three p38 MAPK isoforms expressed in endothelial cells, p38β knockdown prevented PAR-1-mediated STIM1 phosphorylation and potentiated SOCE. In addition, inhibition of the SOCE downstream target CaM kinase kinase β (CaMKKβ) or knockdown of AMPKα1 suppressed PAR-1-mediated phosphorylation of p38β and hence STIM1. Thus, our findings demonstrate that SOCE activates CaMKKβ-AMPKα1-p38β MAPK signaling to phosphorylate STIM1, thereby suppressing endothelial SOCE and permeability responses.

Previous studies from our laboratory have demonstrated that an increase in intracellular Ca 2ϩ signaling is critical for protease-activated receptor-1 (PAR-1) 2 -mediated endothelial hyper-permeability (1). Thrombin-induced increase in intracellular Ca 2ϩ concentration in endothelial cells is dependent on both inositol 1,4,5-triphosphate-induced release of stored Ca 2ϩ and Ca 2ϩ store depletion-mediated Ca 2ϩ entry, termed storeoperated Ca 2ϩ entry (SOCE) (1). The channel responsible for mediating Ca 2ϩ entry secondary to ER-stored Ca 2ϩ depletion is termed store-operated Ca 2ϩ entry channels (SOCs) (1,2). In recent studies, we have shown that transient receptor potential canonical (TRPC) 1 and 4 channels function as SOCs in endothelial cells (3). Other studies have shown that a Ca 2ϩ -selective channel (I CRAC ), Orai1 channel also contributes to SOCE in endothelial cells (4,5).
Recent studies have elucidated the mechanism of the ER-localized Ca 2ϩ sensor protein, stromal interacting molecule-1 (STIM1), in activating SOCE through TRPC and Orai1 channels (6 -10). ER-store Ca 2ϩ depletion induces clustering of STIM1 at "puncta" on the ER/plasma membrane interface, which in turn binds to and activates SOCs (TRPC and Orai1 channels) (6 -10). Many of the molecular details of STIM1mediated Ca 2ϩ entry (i.e. SOCE) are well understood (6). STIM1 is a multidomain protein containing an EF hand domain at the N terminus projecting into the ER lumen and at the C-terminal ezrin-radixin-moesin (ERM), serine/proline, and lysine-rich cytosolic domains. The ERM domain contains a coiled-coil domain and a highly conserved SOAR (STIM1 Orai activating region) domain (6). The SOAR domain binds to both TRPC and Orai1. STIM1 SOAR domain binding to Orai1 is sufficient to gate Orai1 (6,7). In the case of TRPC channels, electrostatic interaction between the STIM1 C-terminal Lys domain and TRPC C-terminal acidic residues is required to activate Ca 2ϩ entry through TRPC channels (6,11). STIM1 is critical for thrombin-induced SOCE by its interaction with TRPC1 and TRPC4 in endothelial cells (3). Studies from another laboratory have shown that STIM1-Orai1 association also mediates SOCE in endothelial cells (4,5). Regulation of SOCE activity is not as well understood in general and has not been investigated in endothelial cells.
STIM1 was originally identified as a phosphoprotein with multiple serine (Ser) phosphorylation sites (12). Recently, Smyth et al. (13) showed that STIM1-mediated Ca 2ϩ entry was "turned off" by phosphorylation of Ser-486 and Ser-668 residues at the C terminus during mitosis in HeLa cells. Furthermore, they have shown that STIM1 phosphorylation prevented store depletion-induced STIM1 punta at ER-plasma membrane junctions, an event essential for SOCE activation. Another study showed that ERK1/2-mediated phosphorylation of STIM1 at Ser-519 and Ser-575 modulated SOCE in HEK293 cells (14). Thus, we investigated the underlying signaling pathway downstream of PAR-1 in inducing STIM1 phosphorylation at its Ser residues to "turn off" SOCE in endothelial cells.
Sequence analysis for human STIM1, using Group-based prediction system, version 2.1.1 software, revealed the presence of 10 consensus phosphorylation sites (Ser-486, Ser-492, Ser-575, Ser-600, Ser-608, Ser-618, Ser-621, Thr-626, Ser-628, and Ser-668) for p38 MAPK indicating the possibility that p38 MAPK-mediated STIM1 phosphorylation may modulate SOCE in endothelial cells. In recent studies, we have shown that SOCE induced by thrombin resulted in activation of AMPK and its downstream target p38 MAPK in endothelial cells (15). Thus, we addressed the possibility that SOCE-activated AMPK-p38 MAPK signaling axis is involved in inhibiting SOCE in endothelial cells. Our results show that SOCE signal activates AMPK␣1 and its downstream target p38␤ MAPK, which in turn phosphorylates STIM1 to turn off SOCE in endothelial cells.
Cytosolic Ca 2ϩ ([Ca 2ϩ ] i ) Measurement-The cytoplasmic Ca 2ϩ concentration ([Ca 2ϩ ] i ) in ECs was measured using the Ca 2ϩ -sensitive fluorescent dye Fura-2/AM (3). Cells were grown to confluence on gelatin-coated glass coverslips and then washed two times with serum-free medium and incubated for 2 h at 37°C in culture medium containing 1% FBS. Cells were washed once and loaded with 3 M Fura-2/AM for 30 min. After loading, cells were washed with HBSS, and the coverslips were transferred on a perfusion chamber at 37°C and imaged using a semi-motorized microscope (Axio Observer D1; Carl Zeiss GmbH, Jena, Germany) equipped with an AxioCam HSm camera (Carl Zeiss) and a Fluar ϫ40 oil immersion objective. Light was provided by the DG-4 wavelength switcher (Princeton Scientific Instruments, Monmouth Junction, NJ). A dual excitation at 340 and 380 nm was used, and emission was col-lected at 520 nm. The AxioVision physiology software module was used to acquire the images at 1-s intervals, and the data were analyzed off-line. In each experiment, 20 -30 cells were selected to measure change in [Ca 2ϩ ] i .
Transendothelial Electrical Resistance-The real time change in endothelial monolayer resistance (TER) was measured to assess endothelial barrier function. Before the experiment, confluent endothelial monolayer was kept in 1% FBS containing medium for 2 h and then a thrombin-induced real time change in TER was measured. Data are presented in resistance normalized to its value at time 0 (18).
Assessment of Lung Microvessel Permeability in Mice-C57BL6J mice obtained from Charles Rivers Laboratories (Wilmington, MA) were housed in the University of Illinois Animal Care Facility and used according to approved animal protocols. Mice (22-25 g) were anesthetized with (2.5% sevoflurane in room air) for insertion of an indwelling jugular catheter and then were allowed to recover for 30 min. Mice were then injected with AICAR (500 mg/kg, intraperitoneal) or saline. At 195 min after AICAR or saline administration, mice received 100 l of Evans blue dye conjugated with albumin (EBA) (20 mg/kg) through the jugular vein. At the end of 4 h, the mice were sacrificed and lungs harvested. Thirty min before sacrificing, mice received either saline (100 l) or PAR-1 peptide (100 l (1 mg/kg)) through the jugular vein. The EBA presence in lung tissue was measured as described previously (19).
siRNA Transfection-ECs grown to ϳ70% confluence on gelatin-coated culture dishes were transfected with target siRNAs or sc-siRNA using DharmaFECT transfection reagent as per the manufacturer's instructions. At 72 h after transfection, cells were used for Ca 2ϩ measurements or harvested for Western blot analysis.
Immunoprecipitation-ECs grown to confluence challenged with agonists were washed three times with phosphate-buffered saline at 4°C and lysed in lysis buffer as in Ref. 17. Lysate protein (300 g) was subjected to immunoprecipitation. Insoluble material was removed by centrifugation (13,000 ϫ g for 15 min) before overnight immunoprecipitation with 1 g/ml antibody at 4°C. Protein A/G-agarose beads were added to each sample and incubated for 1 h at 4°C. Immunoprecipitates were gently washed three times with wash buffer (Tris-buffered saline containing 0.05% Triton X-100, 1 mM Na 3 VO 4 , 1 mM NaF, 2 g/ml leupeptin, 2 g/ml pepstatin A, 2 g/ml aprotinin, and 44 g/ml phenylmethylsulfonyl fluoride). Immunoprecipitated proteins were resolved on SDS-PAGE and immunoblotted with appropriate antibodies.
Immunoblotting-EC lysates or immunoprecipitates were resolved by SDS-PAGE on a 10% separating gel under reducing conditions and transferred to Duralose membrane. Membranes were blocked with 5% dry milk in 10 mM Tris-HCl, pH 7.5, 150 mM NaCl, and 0.05% Tween 20 for 1 h. Membranes were incubated with the indicated primary antibody (diluted in blocking buffer) overnight. After three washes, membranes were incubated with horseradish peroxidase-conjugated secondary antibody. Protein bands were detected by enhanced chemiluminescence.
Confocal Imaging-ECs were transfected with the YFP-WT-STIM1 (2 g/ml) construct as described previously (3). 48 h after transfection, cells were washed and placed in HBSS, and then confocal live cell images of the YFP-tagged fluorescent protein were acquired near the surface of the cell using a 514-nm laser excitation/530-nm LP emission filter with the pinhole set to achieve 1 Airy unit (ϳ0.5-m optical sections). Image configurations acquired before and after thrombin stimulation were not changed, and the cells were maintained at 37°C.
Statistical Analysis-Comparisons were made with a twotailed Student's t test. Experimental values were reported as mean Ϯ S.E. Differences in mean values between two or more groups were determined by one-way analysis of variance. A p value Ͻ0.05 was considered statistically significant.

STIM1 Phosphorylation Inhibits SOCE in Endothelial Cells-
To study the relationship between STIM1 phosphorylation and SOCE in endothelial cells, we treated HLMVECs with thrombin for different time periods, and then cells were used to examine STIM1 phosphorylation at its serine residues. After thrombin treatment, cell lysates were immunoprecipitated with anti-STIM1 mAb, and the precipitate was immunoblotted with anti-phospho-Ser pAb. Here, we observed that thrombin stimulation caused a time-dependent phosphorylation of STIM1, which reached a maximum of ϳ8-fold over basal within 10 min (Fig. 1A). At 30 and 60 min after thrombin treatment, STIM1 phosphorylation was significantly reduced (Fig.  1A). These results indicate that thrombin stimulation caused STIM1 phosphorylation in endothelial cells.
Next, we investigated whether STIM1 phosphorylation inhibits SOCE in HLMVECs. We challenged HLMVECs with thrombin or PAR-1 peptide for different time intervals. The cells were then washed, loaded with Fura-2AM for 30 min, and then thapsigargin (TG)-induced ER-stored Ca 2ϩ release and Ca 2ϩ release-activated Ca 2ϩ entry (SOCE) were measured. In control cells (not pretreated with thrombin or PAR-1 peptide), we observed a normal TG-induced store release and SOCE (Fig.  1B). In cells pretreated with thrombin or PAR-1 peptide for 10 min, TG failed to induce either store release or SOCE (Fig. 1B). Interestingly, TG response was partially rescued in 30-min thrombin-or -PAR-1 peptide-pretreated cells (Fig. 1B), whereas in TG response was largely rescued in 60-min thrombin or -PAR-1 peptide-pretreated cells (Fig. 1B). These results are in agreement with the time course of thrombin-induced STIM1 phosphorylation. Thus, STIM1 phosphorylation in its Ser residues may inhibit SOCE in HLMVECs.

Pharmacological Activation of AMPK Induces STIM1 Phosphorylation and Prevents PAR-1-induced Ca 2ϩ Entry and Lung
Microvessel Permeability-Studies from our laboratory (15) and others (22) showed that Ca 2ϩ entry signal activates AMPK in endothelial cells. Therefore, we tested whether AMPK signaling is involved in modulating SOCE in endothelial cells. AMPK is a serine/threonine (Ser/Thr) protein kinase composed of a catalytic ␣-subunit and regulatory ␤and ␥-subunits (23,24). Thr-172 phosphorylation in the ␣-subunit is essential for catalytic function of AMPK (23,24). We pretreated HLM-VECs with the AMPK activator AICAR (23) and measured Thr-172 phosphorylation of AMPK ␣-subunit (AMPK␣). We observed a dose-dependent increase in the phosphorylation of AMPK␣ ( Fig. 2A), with an optimal increase (ϳ2.5-fold) in cells treated with 1 or 2 mM ( Fig. 2A). To study the effect of AMPK activation on STIM1 phosphorylation, we measured STIM1 phosphorylation in its Ser residues in control and AICARtreated HLMVECs. In this experiment, cells were lysed; lysates were immunoprecipitated with anti-phospho-Ser pAb, and the immunoprecipitate was immunoblotted with anti-STIM1 mAb. We observed that STIM1 phosphorylation was increased significantly in AICAR-pretreated cells compared with control cells (Fig. 2B).
Next, we measured thrombin-induced Ca 2ϩ entry (i.e. SOCE) in cells treated with or without AICAR. AICAR pretreatment had no significant effect on thrombin-induced store Ca 2ϩ release (Fig. 2C), whereas thrombin-induced Ca 2ϩ entry was blocked in AICAR-treated cells (Fig. 2C). These results indicate that the AMPK signal may play a role in the reversal of SOCE in ECs. We also investigated the in vivo relevance of AMPK signaling in regulating lung microvessel permeability. In this study, we injected mice (C57BL6J) with AICAR (500 mg/kg i.p.), and the control mice received saline. At 4 h after AICAR or saline injection, mice were used to assess PAR-1 agonist peptide-induced lung microvessel permeability by measuring EBA uptake in lungs (see details under "Experimental Procedures"). PAR-1 peptide induced an ϳ2-fold increase in EBA uptake over control (Fig. 2D). AICAR alone had no effect on basal EBA uptake in lungs (Fig. 2D), whereas AICAR pretreatment markedly reduced PAR-1-induced EBA uptake in lungs (Fig. 2D). These results collectively suggest that AMPK-mediated STIM1 phosphorylation may play a role in reversing lung vascular permeability responses.
p38 MAPK Inhibition Augments SOCE and Prevents SOCEmediated Phosphorylation of STIM1 in HLMVECs-Because AMPK lies upstream of p38 MAPK (15), we investigated whether AMPK modulates SOCE by activating its downstream target p38 MAPK. In this study, we initially performed an in vitro kinase assay to determine whether p38 MAPK directly phosphorylates STIM1 using recombinant active p38 MAPK (20,21). In this assay, we determined STIM1 phosphorylation by Western using the anti-phospho-Ser pAb. We observed that active p38␤2 caused phosphorylation of STIM1 at its serine residues (Fig. 4, A and B). The phosphorylation observed was prevented by SB203580, a p38 MAPK-selective inhibitor (Fig. 4,  A and B) (29 -31) indicating that STIM1 is a direct target of p38 MAPK. Next, we studied the role of p38 MAPK in regulating SOCE in HLMVECs. In this experiment, HLMVECs were treated with vehicle (0.01% DMSO) or SB203580, and then thrombin-induced STIM1 phosphorylation was measured. Thrombin stimulation increased a time-dependent phosphorylation of STIM1 with a maximum level at 10 min and a return to basal level 60 min after thrombin stimulation (Fig. 4C). In HLMVECs grown to 80% confluence were pretreated with AICAR (0, 0.1, 1, and 2 mM) for 2 h in 1% serum-containing medium. Cells were then lysed and immunoblotted with anti-phospho-AMPK␣ mAb (top), anti-AMPK␣ pAb (middle), and anti-␤-actin mAb. A representative blot is shown from four independent experiments. The protein bands were quantified by densitometry relative to ␤-actin (right panel). *, significantly different from cells not treated with AICAR. B, AICAR induces STIM1 phosphorylation. HLMVECs exposed to the indicated concentrations of AICAR for 2 h as above. Cells were lysed; lysates were immunoprecipitated (IP) with anti-phospho-Ser pAb, and then the precipitate was immunoblotted (IB) with anti-STIM1 mAb. Results shown are from representative of four experiments. The quantified results are shown in the right panel. *, significantly different from control. C, AICAR inhibits thrombin-induced Ca 2ϩ entry. HLMVECs grown on coverslips and pretreated with AICAR (1 or 2 mM) for 2 h were used to measure Ca 2ϩ entry. Fura-2-loaded cells placed in Ca 2ϩ -and Mg 2ϩ -free HBSS were stimulated with thrombin (50 nM). After return of [Ca 2ϩ ] i to base-line levels, CaCl 2 (1.5 mM) was added to extracellular medium to induce Ca 2ϩ entry. Arrow indicates time at which cells were stimulated with thrombin (Thr). Results shown are mean Ϯ S.E. of four experiments. D, AICAR pretreatment abrogates PAR-1-induced lung vascular permeability increase. C57BL/6 mice either injected with AICAR (500 mg/kg, intraperitoneally) or saline were used to measure PAR-1-peptideinduced EBA uptake in lungs. Top panel shows experimental design. Results are mean Ϯ S.E. of changes in lung EBA after PAR-1 agonist peptide administration (n ϭ 6; in each group). * indicates the significance between the treatment groups and the respective control groups (p Ͻ 0.05).
cells preincubated with SB203580, thrombin-induced STIM1 phosphorylation was suppressed (Fig. 4C). To understand the functional relevance, we determined thrombin-induced Ca 2ϩ entry in control (vehicle-treated) cells, in SB203580-treated cells or in SB202474 (a negative control compound)-treated cells. We observed normal thrombin-induced Ca 2ϩ entry in vehicle-or SB202474-treated cells (Fig. 4D), whereas in cells preincubated with the p38 MAPK inhibitor (SB203580), we observed a sustained increase in Ca 2ϩ entry (Fig. 4D), indicating that p38 MAPK signaling is required for the reversal of SOCE.
Studies have demonstrated that ER-stored Ca 2ϩ release-mediated assembly of STIM1 into puncta at the ER/plasma membrane interface is required for Ca 2ϩ entry (6). Also, studies have shown that STIM1 phosphorylation prevents store depletioninduced STIM1 puncta formation and SOCE (13). To address whether inhibition of p38 MAPK may influence the STIM1 puncta formation, we expressed YFP-WT-STIM1 in HLM-VECs and used a confocal microscope to observe STIM1 puncta after PAR-1-agonist peptide stimulation. In cells treated with 0.01% DMSO (vehicle), we noted that PAR-1 peptide stimulation caused STIM1 puncta formation in a time-dependent  (Fig. 4E, top panel). STIM1 puncta appeared within 20 s of PAR-1 peptide addition and were seen for 400 s. By contrast, in cells incubated with SB203580 for 30 min, we observed a sustained increase in puncta formation over a period of 500 s (Fig. 4E, bottom panel). These results indicate that p38 MAPKmediated STIM1 phosphorylation may regulate SOCE in endothelial cells.
p38␤ MAPK Signaling Is Required to Reverse PAR-1-induced SOCE-Next, we investigated which isoform of p38 MAPK activation downstream of AMPK is involved in regulating SOCE in endothelial cells. Because p38 MAPK has four isoforms (␣, ␤, ␥, and ␦) (29 -31), we first determined the expression profiles of p38 MAPK isoforms in HLMVECs and mouse (m) LECs. RT-PCR confirmed that the mRNA for p38␣, p38␤, and p38␥ were present in HLMVECs, but mLECs express the mRNA for only ␣ and ␤ isoforms (Fig. 5A).
In the next set of experiments, we knocked down individual p38 isoforms using siRNA to address the specific role of p38 isoforms in regulating SOCE in HLMVECs. In these experiments, cells were transfected with control siRNA (Sc-siRNA), si-RNA specific to human p38␣, human p38␤, or human p38␥. At 72 h after transfection, we examined the protein expression by Western blot. p38 MAPK levels were markedly suppressed in target siRNA-transfected cells compared with Sc-siRNA-transfected cells or control cells (Fig. 5B). We then determined Ca 2ϩ entry secondary to thrombin-induced Ca 2ϩ store depletion. In p38␣ knockdown cells, both Ca 2ϩ entry (Fig.  5C) and STIM1 expression (Fig. 5F) were blocked, whereas p38␥ knockdown had no significant effect on thrombin-induced Ca 2ϩ entry (Fig. 5E). Interestingly, p38␤ knockdown had no effect on STIM1 expression (Fig. 5F), but thrombin-induced Ca 2ϩ entry was sustained (Fig. 5D). These results suggest that p38␤ may regulate SOCE presumably by phosphorylating STIM1, whereas the p38␣ signal may be required for the expression of STIM1 in endothelial cells.
Ca 2ϩ Entry-CaMKK␤-AMPK␣1 Axis Signaling Is Required to Activate p38␤ MAPK and STIM1 Phosphorylation-We have shown that Ca 2ϩ entry through TRPC channels activates the CaMKK␤-AMPK-p38 signaling pathway in human pulmonary artery endothelial cells (15). In this study, we exposed FIGURE 4. p38 MAPK downstream of AMPK signaling controls SOCE via phosphorylation of STIM1. A and B, phosphorylation of STIM1 by active p38␤2 was determined using in vitro kinase assay (see details under "Experimental Procedures"). A, Myc-STIM1 was ectopically expressed in HEK293 cells and immunoprecipitated (IP) using anti-Myc; mAb was used as substrate for active p38␤2. The assay was performed in the presence (ϩ) and absence (Ϫ) of p38 inhibitor, SB203580 (10 M). Lane 1, active p38␤2 was not included in the kinase mixture; lane 7, control A/G beads were incubated with Myc-STIM1 expressing HEK cell lysates included in the kinase assay mixture loaded. Equal volume of assay mixture was immunoblotted (IB) with anti-phospho-Ser pAb, anti-Myc pAb, anti-STIM1 mAb, or anti-phospho-p38 pAb (left panel). Phosphoprotein bands were quantified by densitometry and expressed as relative to Myc-STIM1 (right panel). Results shown are mean Ϯ S.E. of three independent experiments. *, significantly different from SB203580 treatment. B, unstimulated HLMVECs were immunoprecipitated using anti-STIM1 pAb, and the immunoprecipitate was used as substrate for active p38␤2. The assay was performed as above. Equal volume of assay mixture was immunoblotted with anti-phospho-Ser pAb, anti-STIM1 mAb, or anti-phospho-p38 pAb (left panel). Results shown are mean Ϯ S.E. of four independent experiments (right panel).
*, significantly different from SB203580 treatment. Note that active p38␤2mediated STIM1 phosphorylation was detectable by anti-phospho-Ser pAb. C, HLMVECs pretreated with vehicle (DMSO, 0.01%) or SB203580 (10 M) for 30 min were used to measure thrombin-induced phosphorylation of STIM1 as above in Fig. 1A. Phosphoprotein bands were quantified by densitometry and expressed as relative to control (right panel). Results shown are mean Ϯ S.E. of three experiments. *, significantly different from cells not stimulated with thrombin or significant difference between control and SB203580-treated cells. D, HLMVECs pretreated with SB203580 (10 M) or SB202474 (10 M) were used to measure thrombin-induced Ca 2ϩ entry as described above. HLMVECs to the CaMKK␤-selective inhibitor STO-609 and then measured thrombin-induced phosphorylation of AMPK␣1. We observed that STO-609 treatment blocked thrombin-induced phosphorylation of AMPK␣1 (Fig. 6A) indicating that CaMKK␤ is essential to activate AMPK␣1 in HLMVECs. Next, we investigated whether SOCE signaling activates p38␤ in HLMVECs. In this experiment, we measured thrombin-induced phosphorylation of p38␤ in the presence and absence of SOCE blocker Gd 3ϩ (3,4). We observed a timedependent increase in phosphorylation of p38␤ in control HLMVECs challenged with thrombin, whereas a marked reduction in thrombin-induced p38␤ phosphorylation was observed in cells treated with Gd 3ϩ (Fig. 6B). These results show an obligatory role for SOCE in signaling thrombin-induced p38␤ activation in HLMVECs.
To address the functional relevance of p38␤ activity downstream of AMPK␣1 and thereby regulating vascular endothelial barrier function, we determined thrombin-mediated changes in TER (18). In control HLMVECs, thrombin addition caused an ϳ70% maximum decrease in TER and the TER return to basal 2 h after thrombin challenge (Fig. 6E). In cells incubated with the p38 inhibitor (SB203580), thrombin caused a similar FIGURE 5. p38␤ MAPK regulates SOCE in HLMVECs. A, RT-PCR analysis of mRNA expression for p38 MAPK isoforms in HLMVECs and mLECs. Total RNA from HLMVECs and mLECs was isolated, and RT-PCR was performed to determine the expression of transcripts for p38 MAPK (␣, ␤, ␥, and ␦) and GAPDH. B, HLMVECs were transfected with Sc-siRNA or siRNA specific to p38 MAPK isoforms (p38␣, p38␤, and p38␥). At 72 h after transfection, cells were used to determine expression of p38␣, p38␤, and p38␥ by immunoblot. C-E, HLMVECs were transfected with Sc-siRNA or siRNA specific to p38␣ (C), ␤ (D), or ␥ (E). At 72 h after transfection, cells were used to determine thrombin-induced Ca 2ϩ entry as described above. Note the sustained Ca 2ϩ entry in p38␤-siRNA transfected cells (D), whereas in p38␣-siRNA transfected cells the Ca 2ϩ entry was blocked (C). Experiments were repeated at least three times, and the results shown are mean Ϯ S.E. F, HLMVECs transfected with Sc-siRNA, p38␣-siRNA, or p38␤-siRNA were immunoblotted with anti-STIM1 mAb. Note that in p38␣-siRNA-transfected cells STIM1 expression was reduced. decrease in TER, but TER return to basal did not occur (Fig. 6E). In the control compound (SB202474), the incubated cells thrombin-induced response was similar to control cells. In another set of experiments, we knocked down p38␤ in HLMVECs and measured TER. In control cells or in cells transfected with Sc-siRNA, thrombin produced an ϳ60% decrease in TER, and the TER recovered to basal levels within 2 h of thrombin addition (Fig. 6F), whereas in p38␤ siRNA-transfected cells, thrombin produced a similar decrease in TER, but TER recovery to basal levels did not occur (Fig. 6F). Thus, these results collectively support the conclusion that SOCE-mediated CaMKK␤-AMPK␣1-p38␤ signaling serves as a turn off switch for SOCE, reversing the permeability responses.

DISCUSSION
SOCE in nonexcitable cells regulates many cellular processes, including cell migration, apoptosis, and induction of inflammatory genes. We have shown that SOCE induced by thrombin ligation of PAR-1 mediates vascular barrier dysfunction and amplifies the expression of inflammatory genes in endothelial cells (3,15,16,32,33). Recent studies have shown that STIM1 is crucial for the activation of SOCE in endothelial cells (3)(4)(5). However, the downstream signaling pathways involved in terminating SOCE are unknown. Further evidence suggests that SOC function may be inhibited by phosphorylation of STIM1 on its C-terminal serine residues (13). The deduced STIM1 sequence revealed the presence of 10 putative phosphorylation sites for p38 MAPK. In recent studies, we showed that SOCE activates the CaMKK␤-AMPK-p38 MAPK signaling pathway in endothelial cells (15). Thus, we tested the postulate that PAR-1-induced SOCE signaling is essential for Ser phosphorylation of STIM1 and subsequent inhibition of SOCE in endothelial cells. Using an antibody that reacts with FIGURE 6. Ca 2؉ entry-CaMKK␤-AMPK␣1-p38␤ MAPK axis signaling mediates STIM1 phosphorylation to inhibit SOCE and endothelial permeability. A, HLMVECs were pretreated with or without CaMKK␤ inhibitor STO-609 (1 M) for 30 min, and then thrombin-induced AMPK␣1 phosphorylation was measured as described in Fig. 3A. The experiment was repeated four times, and the results shown are mean Ϯ S.E. (right panel). *, significantly different from cells not stimulated with thrombin. B, HLMVECs, thrombin-induced p38␤ phosphorylation was measured in the absence and presence of Gd 3ϩ (10 M). After thrombin treatment, cell lysates were immunoprecipitated (IP) with anti-phospho-p38 mAb, and the precipitate was immunoblotted (IB) with anti-p38␤ pAb to determine p38␤ phosphorylation (top panel). Total cell lysates were blotted with anti-p38␤ mAb (bottom panel). A representative blot is shown from four independent experiments. Phosphoprotein bands were quantified by densitometry and are expressed in arbitrary units. *, sig-nificantly different from cells not stimulated with thrombin. Note that impaired thrombin-induced p38␤ phosphorylation in cells treated with Gd 3ϩ to inhibit Ca 2ϩ entry. C, HLMVECs pretreated with AICAR (0, 1, and 2 mM) were lysed and immunoprecipitated with anti-phospho-p38 mAb. The precipitated proteins were immunoblotted with anti-p38␤ MAPK pAb. Total cell lysates were immunoblotted with anti-p38␤ MAPK pAb. D, HLMVECs transfected with either Sc-siRNA or AMPK␣1siRNA (200 nM) as described in Fig. 3D were stimulated with thrombin (50 nM) for different time intervals at 37°C. After thrombin treatment, cells were lysed and immunoprecipitated (IP) with anti-phospho-p38 mAb. The precipitated proteins were immunoblotted (IB) with anti-p38␤ pAb (top row) or anti-p38␣ pAb (2nd row). Total cell lysates were immunoprecipitated with anti-phospho-Ser pAb, and the precipitate was immunoblotted with STIM1 mAb (3rd row). Total cell lysates were immunoblotted with anti-STIM1 mAb (4th row) and anti-AMPK␣1 pAb (bottom row). Results shown are the mean Ϯ S.E. of four independent experiments (right panels). *, significantly different from control cells. E, HLMVECs were grown to confluence on gold electrodes (see details under "Experimental Procedures"). Cells were washed and incubated with 1% serum-containing medium for 2 h and then incubated 30 min with or without the indicated concentrations of 10 M SB203580 or SB202474 before the addition of 50 nM thrombin (Thr). Note that in SB203580-treated cells, thrombin produced a marked decrease in TER, but the TER recovery to basal level was delayed compared with control cells treated with thrombin or cells pretreated with SB202474 followed by thrombin addition. The arrow indicates the time at which the cells were challenged with thrombin (Thr) or medium. F, HLMVECs transfected with 200 nM Sc-siRNA or p38 ␤-siRNA were used to measure thrombin-induced TER changes. Note the delay in thrombin-induced decrease in TER recovery to basal level in p38␤-siRNA transfected cells indicating hyper-permeability associated with prolonged SOCE. The arrow indicates the time at which the cells were challenged with thrombin (Thr) or medium. Results shown are the mean Ϯ S.E. of four independent experiments. *, significantly different from control cells treated with thrombin.

STIM1 Phosphorylation by SOCE, AMPK␣1, and p38␤
phospho-Ser, we observed phosphorylation of STIM1 in response to PAR-1-induced SOCE. Interestingly, we observed a negative correlation between the time course of PAR-1-induced STIM1 phosphorylation and SOCE activity indicating that STIM1 phosphorylation serves as an "off switch" for SOCE in endothelial cells.
AMPK is an energy sensor activated by an increase in AMP levels in cells (23). AMPK activation requires phosphorylation of Thr-172 in the activation loop of the ␣-subunit (23,24). The upstream-activating enzyme AMPK kinase or the closely related tumor suppressor kinase LKB1 can phosphorylate the catalytic ␣-subunit of AMPK in an AMP-dependent manner (23,34). AMPK can also be activated in an AMP-independent manner, which involves CaMKK␤ (22,23). An increase in intracellular Ca 2ϩ is required for CaMKK␤ activation (22). The pharmacological AMPK activator, AICAR, has been shown to phosphorylate AMPK subunit ␣ and activate AMPK both in vitro and in vivo (23,34). AICAR is metabolized to 5-aminoimidazole-4-carboxamide ribonucleoside, which mimics all effects of AMP on AMPK systems (23,34). Zhao et al., (35) have shown that AICAR-mediated AMPK activation reduced endotoxininduced acute lung injury in mice. Another study showed that administration of AMPK activator metformin (anti-diabetic drug) increased the survival rate of endotoxemic mice (36). Creighton et al. (27) have demonstrated that AMPK␣1 signaling promotes the endothelial barrier repair process. More importantly, studies have shown that AICAR-induced AMPK activation targets its downstream effector p38 MAPK (37). Based on the existing evidence, we tested the hypothesis that pharmacological activation of AMPK would block PAR-1-mediated SOCE by phosphorylating STIM1 in endothelial cells. In support of this hypothesis, we observed that AICAR pretreatment induced AMPK subunit ␣ phosphorylation in HLMVECs. Also, AICAR pretreatment induced STIM1 phosphorylation and blocked thrombin-induced Ca 2ϩ entry in HLMVECs. To address the in vivo physiological relevance of AMPK signaling, we pretreated mice with AICAR and then determined PAR-1induced lung vascular permeability by measuring EBA uptake in lungs. We observed that AICAR pretreatment markedly reduced PAR-1-induced increase in lung vascular permeability indicating that AICAR-induced STIM1 phosphorylation may contribute to the inhibition of PAR-1-mediated lung vascular permeability.
Because endothelial cells express both the catalytic ␣1and ␣2-subunits of AMPK (25-28), we compared the extent of phosphorylation of the ␣1and ␣2-subunits in response to thrombin. We observed that thrombin induced phosphorylation of the ␣1-subunit but not the ␣2-subunit of AMPK in HLMVECs. To determine the functional role of SOCE-mediated AMPK␣1 phosphorylation, we transfected HLMVECs with siRNA specific to AMPK␣1 and then measured SOCE in response to thrombin. We observed that in AMPK␣1-depleted cells, thrombin-induced SOCE was augmented, raising the possibility that AMPK␣1 activation is required for STIM1 phosphorylation and hence inhibition of SOCE in endothelial cells.
Li. et al. (38) have shown that activated AMPK interacts with the scaffold protein TAB1, and the resulting complex associates with the p38 MAPK, which in turn promotes p38 MAPK auto-phosphorylation in ischemic hearts. We have shown that inhibition of AMPK in endothelial cells prevented thrombin-induced p38 MAPK activation, demonstrating that AMPK lies upstream of p38 MAPK (15). Thus, we exposed HLMVECs to a specific inhibitor of p38 MAPK (SB203580, which inhibits both p38␣ and p38␤ isoforms (29 -31)) and measured thrombininduced STIM1 phosphorylation and Ca 2ϩ entry. We observed that p38 inhibition suppressed thrombin-induced STIM1 phosphorylation and enhanced thrombin-induced Ca 2ϩ entry in HLMVECs. Because ER-store Ca 2ϩ release-mediated assembly of STIM1 into puncta at ER/plasma membrane activates Ca 2ϩ entry (6, 13), we studied the effect of p38 MAPK inhibition on PAR-1-induced STIM1 puncta formation. We observed that p38 MAPK inhibition prolonged the STIM1 puncta formation upon PAR-1 activation indicating the possibility that STIM1 phosphorylation may regulate SOCE in endothelial cells. It is also possible that STIM1 phosphorylation may induce the dissociation of SOC components or undetermined STIM1binding proteins from STIM1 to limit SOCE in endothelial cells.
There are four p38 MAPK isoforms (MAPK14 (p38␣), MAPK11 (p38␤), MAPK12 (p38␥), and MAPK13 (p38␦)) expressed in mammalian cells (29 -31). Gene knock-out and pharmacological studies suggest that p38␣ signaling is essential for development, transcriptional regulation of genes, and inflammatory responses (29 -31, 39 -41). Recent emerging studies suggest that p38␤ signaling may play a critical role in cell survival and the reversal of inflammatory responses (42). p38␦ and p38␥ isoforms have been shown to regulate transcription of genes (43). Because we observed endothelial expression of ␣, ␤, and ␥ isoforms of p38 MAPK, we attempted to identify the isoform(s) activated downstream of AMPK in response to thrombin-induced SOCE in HLMVECs. We suppressed the expression of each p38 isoform by gene silencing. In this study, we observed that knockdown of p38␣ suppressed STIM1 expression and thrombin-induced SOCE in HLMVECs indicating that p38␣ signaling may regulate the expression of STIM1 rather than SOCE function in HLMVECs. Knockdown of p38␥ had no significant effect on thrombin-induced SOCE in HLM-VECs. Interestingly, knockdown of p38␤ had no effect on STIM1 expression but enhanced thrombin-induced SOCE in HLMVECs. Consistent with the enhanced SOCE, the thrombin-induced increase in permeability was also augmented in p38␤-depleted HLMVECs.
It is known that Ca 2ϩ entry signaling activates CaMKK␤ to induce AMPK␣ phosphorylation in endothelial cells (15,22). In this study, we observed that pretreatment with CaMKK␤ inhibitor STO-609 suppressed thrombin-induced AMPK␣1 phosphorylation indicating that CaMKK␤ is essential to activate AMPK␣1 in endothelial cells. To determine whether AMPK␣1 is essential for SOCE-mediated p38␤ activation and subsequent STIM1 phosphorylation, we silenced AMPK␣1 expression by transfection of siRNA specific to AMPK␣1 and measured phosphorylation of p38␤ and STIM1 in response to thrombin in HLMVECs. We observed that AMPK␣1 knockdown markedly reduced thrombin-induced phosphorylation of p38␤ and STIM1. In another experiment, we observed that AICAR treatment also resulted in p38␤ phosphorylation in HLMVECs.
Moreover, we measured phosphorylation of p38␤ in the presence of SOCE inhibitor Gd 3ϩ (3,4) and observed markedly reduced phosphorylation of p38␤ in response to thrombin. These findings demonstrate that the SOCE-induced CaMKK␤-AMPK␣1-p38␤ signaling pathway is vital in the mechanism of reversal of SOCE and permeability responses.
In summary, we have shown that PAR-1-mediated SOCE (i.e. Ca 2ϩ entry through SOC) results in activation of the CaMKK␤-AMPK␣1-p38␤ signaling axis (Fig. 7), which is essential for phosphorylation of ER-localized STIM1 to turn off SOCE in endothelial cells. Thus, Ca 2ϩ entry-dependent phosphorylation of STIM1 via the CaMKK␤-AMPK␣1-p38␤ axis provides an important negative feedback signal to terminate SOCE and thereby regulates vascular permeability responses.