Activated protein C alters cytosolic calcium flux in human brain endothelium via binding to endothelial protein C receptor and activation of protease activated receptor-1

Activated protein C (APC) exerts neuroprotective effects in a brain focal ischemia model that implies a potential therapeutic role for APC in the central nervous system (CNS) vascular and neurodegenerative disorders.¹  APC reduces organ damage in animal models of sepsis and of ischemic injury, including stroke, and significantly reduces mortality in patients with severe sepsis.^2-4  Prospective epidemiologic data suggest that plasma protein C levels are inversely related to incidence of ischemic stroke.⁵  APC exerts direct effects on human umbilical vein endothelial cells (HUVECs) via modulation of gene expression that may influence inflammation and apoptosis,⁶  and it can trigger intracellular signaling via activation of protease activated receptor-1 (PAR-1) and binding to endothelial protein C receptor (EPCR).⁷  It has been suggested that APC down-regulates nuclear factor κB–dependent expression of adhesion molecules in HUVECs.⁶ 

PAR-1 and PAR-2 are present on systemic vascular endothelial cells^8-10  and on brain microvascular endothelium.¹¹  EPCR is also present on brain endothelial cells¹²  where it may augment activation of protein C by the thrombin-thrombomodulin complex,^13,14  by providing a binding site for protein C on target cell membranes.^7,12,15  Recently, it has been shown that both PAR-1 and EPCR are essential for the ability of APC to protect brain endothelium from ischemic injury via down-regulation of p53-dependent proapoptotic pathway.¹⁶  Cleavage and activation of PARs by extracellular proteases, however, are also capable of causing intracellular calcium concentration ([Ca²⁺]_i) signaling.^17,18  In this study we show that APC induces a potent [Ca²⁺]_i signal in the brain endothelium through PAR-1 and that APC binding to EPCR is a prerequisite for this [Ca²⁺]_i signal.

Purified human plasma–derived APC, protein C zymogen, APC inactivated by heat, and recombinant human mutant Ser360Ala-APC lacking the active serine were prepared as described.¹⁹  No residual thrombin activity was detected in any of the APC preparations.

All [Ca²⁺]_i experiments were performed in normal HEPES (N-2-hydroxyethylpiperazine-N′-2-ethanesulfonic acid) buffer containing 135 mM NaCl, 5.4 mM KCl, 1.8 mM CaCl_2, 1.2 mM MgCl_2, 10 mM glucose, and 20 mM HEPES. RPMI 1640 was purchased from Cellgro (Herndon, VA). For calcium imaging we used the following antibodies: anti-EPCR RCR-292 and RCR-92,¹⁴  monoclonal antihuman APC-immunoglobulin G (IgG; C3),²⁰  antihuman PAR-1 (N-19, H-111, and ATAP-2), antihuman PAR-2 (SAM-11), and antihuman PAR-3 (H-103) (Santa Cruz Biotechnology, Santa Cruz, CA). Leech-derived hirudin was obtained from Sigma (St Louis, MO).

Human brain capillaries were isolated from brain microvascular fragments from autopsies of neurologically healthy young individuals after trauma, and primary human BECs were characterized and cultured as described previously.²¹  After fluorescence-activated cell sorter (FACS) analysis using fluorescence-labeled acetylated low-density lipoprotein (Di-ac-LDL, ligand for endothelial scavenger receptor), cells were more than 98% positive for the endothelial markers, factor VIII–related antigen and CD105, and were negative for glial fibrillary acidic protein (GFAP; astrocytes), CD11b (macrophages/microglia), and α-actin (smooth muscle cells). Early passage cells were plated on double-coated (polylysine, gelatin) glass coverslips for [Ca²⁺]_i measurements.

BEC cultures were kept in RPMI 1640–based medium with HEPES and L-glutamine, containing 20% fetal bovine serum, supplemented with 1% minimal essential medium (MEM) nonessential amino acids, 1% MEM vitamins, 5 U/mL heparin, 30 μg/mL endothelial cell growth supplement, 1 mM sodium pyruvate, and 100 U/mL penicillin/streptomycin. Microarray analysis using U95A Affymetrix (Santa Clara, CA) DNA chips on human BECs confirmed the presence of PAR-1, PAR-2, and PAR-3 mRNA as well as EPCR mRNA (Zlokovic et al, unpublished observation, July 2002).

Primary HUVECs were obtained from ScienCell Research Laboratories (San Diego, CA). Cells were cultured with 17% fetal bovine serum-containing RPMI 1640 medium. Early passage cultures were used for experiments.

Cell cultures on polylysine-gelatin–coated glass coverslips were used on the second to third day after seeding when they reached 70% to 90% confluency. The intracellular calcium level of endothelial cells was imaged using a calcium-sensitive fluorescent dye, Fura-2 AM (Teflabs, Austin, TX). BECs and HUVECs were incubated with 4 μM Fura-2 AM in RPMI 1640 medium for 40 or 20 minutes, respectively. The coverslips were transferred to a perfusion chamber fitted to a stage of an inverted Nikon Diaphot 300 microscope and superfused with normal HEPES buffer for 15 minutes prior to experiments. All reagents were infused via a multitube perfusion system. [Ca²⁺]_i was measured by digital image fluorescence microscopy (objective, Fluor 40/1.3; Nikon) using the Vision 4.0 Software from T.I.L.L. Photonics (Grafelfing, Germany). Excitation wavelengths were 340 and 380 nm, generated by a polychromator illumination system (T.I.L.L. Photonics). Fluorescence emission was monitored at 510 nm. The fluorescent images were collected with a charge-coupled device (CCD) camera (T.I.L.L. Photonics). A fluorescence ratio image (340/380 nm) was acquired in every 2 seconds. The experiments were carried out in HEPES buffer at room temperature (21°C). Fluorescence ratios were converted to free [Ca²⁺]_i using the equation described by Grynkiewicz et al.²²  The maximum (R_max) and the minimum (R_min) ratio value and the ratio of fluorescence for Ca²⁺-bound/Ca²⁺-free dye measured at 380 nm (S_f2/S_b2) were determined using an in vitro calibration method and were corrected for viscosity.²³  Experimental results are shown as [Ca²⁺]_i or increase of [Ca²⁺]_i from basal as indicated in the figures.

Data analysis was performed with Sigmaplot 2000 (SPSS, Chicago, IL). Calibrated data were pooled and plotted as means ± SEMs of [Ca²⁺]_i. In statistical calculations for comparison of groups Student paired t test was used. For the calculation of agonists potency, dose-response data were fitted to a single sigmoidal curve (Hill equation, 3 parameters).

APC administration (50 nM) to BECs for 60 seconds elevated [Ca²⁺]_i from 83.1 nM (± 3.8) to 381.7 ± 26.8 nM (Figure 1Ai,C). Subsequent challenge with APC at 4 minutes after the first APC exposure did not result in any further [Ca²⁺]_i rise, indicating desensitization. Stimulation with APC (10 nM) in the presence of hirudin (40 nM) did not alter the APC-induced [Ca²⁺]_i transients (Figure 1Aii,C). To determine whether the APC-mediated [Ca²⁺]_i signal was dependent on the presence of APC's active site Ser, the active site mutant Ser360Ala-APC was used. Neither recombinant mutant Ser360Ala-APC nor protein C zymogen produced a [Ca²⁺]_i response (Figure 1Aiii-iv,C). Neutralizing anti-APC IgG (C3) antibody abolished APC's effect, and heat-inactivated APC was without effect (Figure 1Av-vi,C). Thus, the change in [Ca²⁺]_i signal in BECs was caused by APC and required APC's active site Ser. The APC-evoked rise in [Ca²⁺]_i was dose dependent (Figure 1B); the threshold APC concentration resulting in a detectable [Ca²⁺]_i signal was between 0.02 and 0.10 nM, whereas the half-maximal response to APC (EC₅₀) was at 0.23 ± 0.02 nM.

To study the role of PARs, APC-induced [Ca²⁺]_i responses in BECs were measured in the presence of different antihuman PAR antibodies at 25 μg/mL that were incubated with BECs for 15 minutes at 37°C prior to APC administration. A control antibody (N-19) against an N-terminal epitope of PAR-1 did not alter the APC-induced [Ca²⁺]_i response (Figure 2Ai-ii,B) whereas H-111, a cleavage site-blocking anti–PAR-1 antibody, completely inhibited the APC-induced [Ca²⁺]_i rise (Figure 2Aiii,B). SAM-11, an antibody raised against the PAR-2 cleavage site, did not block or diminish the APC-evoked [Ca²⁺]_i signal (Figure 2Aiv,B). In studies using HUVECs, a different cleavage-blocking anti–PAR-1 antibody (ATAP-2) also abolished the APC-induced [Ca²⁺], whereas the anti–PAR-3 antibody (H-103) did not inhibit the APC signal (Figure 2C). HUVECs challenged with 150 nM APC responded with an elevation in [Ca²⁺]_i of 161.2 ± 46.2 nM, comparable to BEC (Figure 2C).

Two antihuman EPCR antibodies, RCR-252 and RCR-92, were used to study the effect of blocking EPCR on APC-evoked [Ca²⁺]_i response. RCR-252 (10 μg/mL), an antibody that blocks APC binding to human EPCR,¹⁴  abolished the APC-evoked [Ca²⁺]_i response in both BECs and HUVECs (Figure 2Av,B-C). RCR-92 (10 μg/mL), an antibody raised against an epitope on human EPCR distant from the APC binding site,¹⁴  did not block the APC-induced [Ca²⁺]_i signal in either BECs or HUVECs (Figure 2Avi,B-C).

The amplitude of the APC-induced [Ca²⁺]_i signal was assessed in BECs perfused with media containing either normal extracellular [Ca²⁺] (1.8 mM) or low extracellular [Ca²⁺] (50 nM). Perfusing cells with medium containing low Ca²⁺ for 30 seconds did not alter basal [Ca²⁺]_i. APC treatment of BECs exposed to this condition resulted in a [Ca²⁺]_i signal whose maximum amplitude was within the range of that obtained with BECs perfused with medium containing 1.8 mM Ca²⁺ (Figure 3A), suggesting that, for the initiation of [Ca²⁺]_i signal in response to APC, normal extracellular Ca²⁺ concentrations are not required. To confirm the role of intracellular Ca²⁺ stores in the APC-induced [Ca²⁺]_i signal, thapsigargin (1 μM), a specific blocker of the endoplasmic reticulum Ca²⁺ adenosine triphosphatase (ATPase),²⁴  was used to deplete endoplasmic reticulum (ER) stores. Addition of thapsigargin for 1 minute gradually elevated [Ca²⁺]_i, which subsequently decreased to basal values in 2 minutes (Figure 3B). Subsequent exposure of cells to APC did not result in a [Ca²⁺]_i signal, confirming an essential role of endoplasmic reticulum [Ca²⁺]_i stores for the APC-evoked [Ca²⁺]_i signals (Figure 3B).

The ability of APC to induce direct cellular responses in endothelial cells⁶  may involve distinct signal transduction pathways, such as phosphorylation of mitogen-activated protein kinases (MAPKs).⁷  However, little is known about the wide variety of signaling mechanisms that could mediate APC's direct cellular effects. In this study we demonstrate that APC induces an intracellular [Ca²⁺]_i signal in human BECs and in HUVECs. [Ca²⁺]_i signaling in brain endothelium is of high importance and may lead to changes in cell morphology and blood-brain barrier (BBB) permeability through different mechanisms that include activation of myosin light-chain kinase, and/or conformational and biochemical changes in cytoskeletal proteins comprising the adherens junctions and tight junctions of the BBB.^25-33  Brown and Davis noted³³  that the exact mechanisms by which calcium flux in brain endothelial cells alters signaling cascades or regulates BBB function are as yet undefined. APC is neuroprotective in vivo¹  and appears to protect stressed brain endothelial cells from hypoxic/ischemic damage in vitro and in vivo,¹⁸  but the implications of the intracellular [Ca²⁺]_i signal elicited by APC under physiologic and pathophysiologic conditions remains to be elucidated.

Because APC is generated from protein C by thrombin, we exposed endothelial cells to APC in the presence of hirudin, a highly specific blocker of thrombin. Hirudin did not abolish or reduce the APC-mediated [Ca²⁺]_i signals, excluding the possibility that the APC-evoked [Ca²⁺]_i was caused by contamination with thrombin.³⁴  The proteolytic activity of APC was required for its effect on [Ca²⁺]_i signaling because replacement of the active site Ser360 by Ala was without effect on [Ca²⁺]_i. APC signaling in BECs and HUVECs was inhibited by PAR-1 cleavage-blocking antibodies (H-111 or ATAP-2), whereas pretreatment with a control anti–PAR-1 antibody (N-19) or with PAR-2 cleavage-blocking antibody (SAM-11) or anti–PAR-3 antibody (H-103) did not prevent the APC-induced [Ca²⁺]_i rise. These results support the hypothesis that APC cleaves and activates PAR-1 that in turn mediates [Ca²⁺]_i signaling in both human BECs and HUVECs, whereas PAR-2 or PAR-3 is not implicated in the APC-mediated [Ca²⁺]_i response. These findings are consistent with the APC-induced effects on PAR-1 signaling in endothelial cells.^7,16 

Binding of APC to the endothelial receptor, EPCR is essential for the APC-induced MAPK phosphorylation⁷  and for the activation of APC's antiapoptotic pathway in stressed brain endothelial cells via down-regulation of proapoptotic transcription factor p53. To investigate the necessity of EPCR for APC-induced [Ca²⁺]_i responses, BECs and HUVECs were pretreated with different antibodies against EPCR. An antibody that blocks APC binding to EPCR (RCR-252) blocked the APC-mediated [Ca²⁺]_i signals, implicating binding of APC to EPCR for signaling (Figure 2). Although EPCR expression is relatively low in the brain endothelium,¹²  the data presented here show that binding of APC to EPCR is required for the APC-mediated [Ca²⁺]_i signal in BECs.

The source of intracellular calcium ions responsible for the flux in [Ca²⁺]_i induced by APC was also investigated. When extracellular Ca²⁺ was lowered to 50 nM, APC induced a [Ca²⁺]_i signal similar in amplitude to that evoked in the presence of 1.8 mM extracellular Ca²⁺, suggesting that normal extracellular Ca²⁺ concentrations are not required for the initiation of the APC-mediated [Ca²⁺]_i signal.³⁵  BEC cultures failed to respond to APC when normal endoplasmic reticulum Ca²⁺ stores had been depleted by thapsigargin, implying that the APC-induced [Ca²⁺]_i flux is mobilized initially from the ER Ca²⁺ stores (Figure 3B). Because thapsigargin-induced Ca²⁺ release activates the 44-kDa and 42-kDa MAP kinases in human fibroblasts and epidermal cells,^26,36  and, because both MAP kinases are rapidly phosphorylated by APC in HUVECs⁷  and human BECs,³⁷  it is possible that intracellular [Ca²⁺] mobilization by APC from the ER results in rapid activation of MAP kinases by a calcium-dependent pathway.^26,36  Because of the complexity and multiplicity of the signal transduction mechanisms that potentially contribute to cytoprotection by EPCR-dependent action of APC on PAR-1,^6,7,16  future studies on [Ca²⁺]_i regulation by APC in various cells are warranted.

In summary, this study shows that human brain microvessel endothelium as well as human umbilical vein endothelium responds to APC with [Ca²⁺]_i signaling, where PAR-1 but not PAR-2 and PAR-3 is involved and where APC binding to EPCR is required for this [Ca²⁺]_i signaling process.

We thank Drs. Andrew Gale and José A. Fernández for the gifts of purified APC.