Pathogen Recognition by Toll-like Receptor 2 Activates Weibel-Palade Body Exocytosis in Human Aortic Endothelial Cells*

The endothelial cell-specific granule Weibel-Palade body releases vasoactive substances capable of modulating vascular inflammation. Although innate recognition of pathogens by Toll-like receptors (TLRs) is thought to play a crucial role in promotion of inflammatory responses, the molecular basis for early-phase responses of endothelial cells to bacterial pathogens has not fully been understood. We here report that human aortic endothelial cells respond to bacterial lipoteichoic acid (LTA) and synthetic bacterial lipopeptides, but not lipopolysaccharide or peptidoglycan, to induce Weibel-Palade body exocytosis, accompanied by release or externalization of the storage components von Willebrand factor and P-selectin. LTA could activate rapid Weibel-Palade body exocytosis through a TLR2- and MyD88-dependent mechanism without de novo protein synthesis. This process was at least mediated through MyD88-dependent phosphorylation and activation of phospholipase Cγ. Moreover, LTA activated interleukin-1 receptor-associated kinase-1-dependent delayed exocytosis with de novo protein synthesis and phospholipase Cγ-dependent activation of the NF-κB pathway. Increased TLR2 expression by transfection or interferon-γ treatment increased TLR2-mediated Weibel-Palade body exocytosis, whereas reduced TLR2 expression under laminar flow decreased the response. Thus, we propose a novel role for TLR2 in induction of a primary proinflammatory event in aortic endothelial cells through Weibel-Palade body exocytosis, which may be an important step for linking innate recognition of bacterial pathogens to vascular inflammation.

The onset of inflammatory responses of vascular endothelial cells plays crucial roles in recruitment of immune cells, thrombus formation, and development of vascular inflammation or atherosclerosis. Early endothelial activation involves dual phases: rapid translocation of P-selectin to the endothelial surface and slower synthesis and expression of adhesion molecules such as ICAM-1 (intercellular adhesion molecule 1). 2 The former process is accompanied by rapid exocytosis of Weibel-Palade bodies, which are endothelial cell-specific storage granules that contain vascular modulators, including von Willebrand factor (VWF), P-selectin, IL-8, eotaxin-3, endothelin-1, CD63/lamp3, osteoprotegerin, and angiopoietin-2 (1,2). During Weibel-Palade body exocytosis, these proteins are transported to the outside of the cell upon stimulation or vascular damage and may control local or systemic pathobiological effects, including thrombosis and atherogenesis. Regulated Weibel-Palade body exocytosis is known to be initiated through an increase of intracellular calcium level after stimulation with various secretagogues, including calcium ionophores, thrombin, histamine, TNF-␣, and extracellular ATP (1,2).
Recently, excess innate immune responses of vessel walls or endothelium to invading pathogens have been suggested to be linked to atherogenesis. Several common bacterial infectious agents or invasive pathogens, such as Chlamydia pneumoniae, Helicobacter pylori, Porphyromonas gingivalis, and oral commensal bacteria, have so far been detected in vessel walls or atheroscrelotic lesions in humans (3,4). However, the linkage between artery endothelial innate recognition of such pathogens and inflammatory responses has not been fully elucidated.
For the detection of invasive bacteria in host defense, several Toll-like receptors (TLRs) are employed to identify molecular motifs that usually compose bacterial bodies (5). Among TLR members in humans, TLR2 detects the widest range of common bacterial constituents, such as lipoteichoic acids (LTA), peptidoglycans (PGN), bacterial di-or triacylated lipoproteins or lipopeptides, lipoarabinomannans, porins, and fimbriae (5)(6)(7)(8). TLR4 and TLR5 contribute to the recognition of only a few bacterial components, i.e. LPS and flagellin (9,10). Because TLR1 and TLR6 participate in the accurate discrimination of molecular structures by TLR2 as coreceptors, several molecules, including CD14, CD36, and LOX-1, further facilitate the interactions of TLR2 with bacterial pathogens (5,11,12). After recognition of cognate agonists, endothelial TLRs activate the classic Toll/IL-1R signaling pathway utilizing MyD88 and IL-1R-associated kinase (IRAK)-1, which ultimately activate a TNFR-associated factor (TRAF) 6 complex and IBs and the release and translocation of active NF-B to the nucleus. The artery endothelial NF-B signaling pathways downstream of TLRs are thought to participate in the development of artery inflammatory diseases or atherogenesis through the promotion of the expression of a large number of proinflammatory mediators and adhesion molecules (13)(14)(15). However, it is still not known whether artery endothelial TLRs are primary initiators or modulators of the diseases.
In this study, we investigated the early-phase proinflammatory responses of human aortic endothelial cells (HAECs) to bacterial cell wall constituents. We found that recognition of bacterial constituents by TLRs, especially by TLR2 but not TLR4, could activate Weibel-Palade body exocytosis. We further investigated the involvement of MyD88 in regulation of the cell response.
DNA Cloning-A human TLR2-encoding plasmid was prepared as described previously (17). The dominant negative TLR2 (P681H) was constructed using a QuikChange II sitedirected mutagenesis kit (Stratagene) according to the manufacturer's instructions.
Cell Culture and Transfection of siRNA-HEK293 cells and human monocytic THP-1 cells were grown as described previously (20). HAECs and HUVECs were grown in endothelial growth medium-2 (Camblex) as described previously (21). These endothelial cells were used for experiments from passages 4 to 8. All of the gene-specific siRNA oligonucleotides for human TLR1, TLR2, TLR6, MyD88, and IRAK-1 and a control oligonucleotide were purchased from Dharmacon. Although the sequences were not provided by the manufacturer, significant suppressive effects on the respective gene expression could be confirmed by reverse transcription-PCR compared with the control transfection (data not shown). For the transfection of siRNA, confluent HAECs or HUVECs seeded on 6-or 24-well plates were prepared and washed once with Opti-MEM I medium (Invitrogen). Transfection of siRNAs (100 nM) was performed with Lipofectin reagent (Invitrogen) as instructed by the manufacturer. Toxi-Blocker transfection supplement (TOYOBO) was used to prevent cytotoxicity of lipofection reagents. After 12 h of incubation, culture media were changed to endothelial growth medium-2 media, and incubation was continued for 24 h.
Luciferase Reporter Gene Assay-HEK293 cells stably transfected with human TLR2 gene (or mock control vector) were plated at 5 ϫ 10 4 cells/well in 24-well plates before DNA transfection. The cells were transiently transfected with 50 ng of an NF-B-driven firefly luciferase reporter plasmid (pNF-B-Luc, Stratagene) and 5 ng of a construct directing expression of Renilla luciferase under the control of a constitutively active thymidine kinase promoter (pRL-TK, Promega). After 12 h of incubation, the cells were transfected with 100 nM siRNA oligonucleotide for MyD88 (or glyceraldehyde-3-phosphate dehydrogenase control). Toxi-Blocker transfection supplement was used to prevent cytotoxicity of lipofection reagents. After a further 24 h of incubation, the cells were stimulated with TLR2 agonists in media containing 1% fetal bovine serum for 6 h. Then the cells were lysed, and luciferase activity was measured as described previously (17,20).
Determination of VWF, IL-8, and TNF-␣ by ELISA-HAECs were grown on 24-well plates, then washed and placed in 200 l of Opti-Mem I (Invitrogen) containing 1% fetal bovine serum without growth factors, and stimulated with various concentrations of TLR2 agonists for 60 min. The amount of VWF released into the medium was measured by a VWF ELISA kit (American Diagnostica) according to the manufacturer's instructions. Adhesion Assay-Confluent HAECs seeded on 24-well plates were treated with 10 g/ml LTA for 60 min. The culture medium was then removed, and monocytic THP-1 cells (2.5 ϫ 10 5 ) prelabeled with Alexa564-conjugated concanavalin A were added to the culture. Cells were then allowed to adhere for 30 min on a rocking platform. After two washes with phosphatebuffered saline, fluorescent images were immediately obtained by a fluorescent microscope IX71 with DP70 image capture (Olympus) and processed using Adobe Photoshop, version 7.0. Adhesion of red fluorescent cells was quantified in three fields per well. Results are representative of three separate experiments and expressed as means Ϯ S.D.
Immunofluorescence of VWF-Confluent HAECs were treated with 10 g/ml LTA or 10 M A23187 for 60 min. The culture media were removed, and the cells were immediately fixed at Ϫ20°C with methanol for 60 min. Immunostaining was carried out using an anti-VWF rabbit polyclonal antibody (Santa Cruz Biotechnology, Santa Cruz, CA) and Alexa488conjugated secondary antibody (Invitrogen). Cell nuclei were also stained with 2.5 g/ml Höechst 33342 for 30 min. Images were obtained by a fluorescent microscope IX71 (magnification: ϫ40) with DP70 image capture (Olympus) in the presence of the Prolong Gold Antifade reagent (Invitrogen) and processed using Adobe Photoshop, version 7.0 (Adobe). Results are representative of three separate experiments.
Immunoblot Analysis-Confluent HAECs seeded on 60-mm plates were transfected with gene-specific siRNA and incubated in Opti-Mem I media containing 5% fetal bovine serum for 4 -6 h. The cells were stimulated with 1 g/ml LTA for 0 -60 min and lysed with a buffer consisting of 20 mM Trishydrochloride (pH 7.2), 150 mM sodium chloride, 5 mM EDTA, and 1% Triton X-100 in the presence of protease inhibitors (Roche Applied Science) at 4°C for 15 min followed by clarification by centrifugation at 12,000 ϫ g for 10 min. SDS-PAGE and immunoblot analyses were performed as described previously (17,20). Results are representative of three separate experiments.
Flow Cytometry-To assess the surface expression of P-selectin, confluent HAECs were treated with 10 g/ml LTA for 30 min. To assess the surface expression of TLR2, confluent HAECs or HUVECs were treated with 10 ng/ml IFN-␥ or they were incubated for 12 h under laminar flow. Cell culture under laminar flow was performed with a cone and plate apparatus as described previously (22). Magnitude of the flow was controlled at ϳ15 dyn/cm 2 . The cells were then removed with phosphatebuffered saline containing 20 mM EDTA and fixed with phosphate-buffered saline containing 4% paraformaldehyde at 4°C for 60 min. The cells were then incubated at 4°C for 60 min with anti-TLR2 monoclonal antibody (IMG-319), anti-P-selectin monoclonal antibody (BD Biosciences), or isotype-matched mouse IgG and then with fluorescein isothiocyanate-conjugated anti-mouse IgG. Fluorescence was measured using a FACSCalibur (BD Biosciences).
Statistics-All values were evaluated by statistical analysis using one-way analysis of variance and Student-Newman-Keul's test. Differences were considered to be statistically significant at the level of p Ͻ 0.05.

Induction of Weibel-Palade Body Exocytosis by Bacterial
Constituents-We first examined whether bacterial LTA activated degranulation of Weibel-Palade bodies, because LTA has been reported to stimulate vascular endothelial cells, leading to induction of production of proinflammatory mediators, dysfunction, or cell death (23)(24)(25). After stimulation of HAECs for 30 min, LTA clearly decreased the amount of Weibel-Palade bodies, stained with an antibody to VWF, in the cells (Fig. 1A). Compared with the calcium ionophore (A23187)-induced response, we found that LTA gradually activated Weibel-Palade body exocytosis, quantification of which was performed by measuring the amount of VWF released into the media (Fig. 1B). VWF release by stimulation with FIGURE 1. MyD88-dependent Weibel-Palade body exocytosis by bacterial constituents. A, HAECs stimulated with 10 g/ml LTA or 1 M A23187 for 60 min were fixed and stained immunofluorescently with anti-VWF antibody (green) and with Hö echst33342 (blue). Left, unstimulated; middle, stimulated with LTA; right, stimulated with A23187. B, HAECs were stimulated with 1 g/ml LTA or 1 M A23187 for the indicated periods. The amounts of VWF released into the media were measured by ELISA. Each value is the mean Ϯ S.D. (n ϭ 3). C, HAECs transfected with MyD88 or IRAK1-specific or control siRNA were prepared. Cells were pretreated with 10 g/ml cycloheximide for 30 min and then washed and stimulated with 10 g/ml LTA for 60 min or 4 h. The amounts of VWF released into the media were measured by ELISA. Each value is the mean Ϯ S.D. (n ϭ 3). *, versus control group, p Ͻ 0.01. D, HAECs transfected with MyD88-specific or control siRNA were stimulated with E. coli LPS O26:B6 (0.01-1 g/ml), LPS from S. minnesota (0.01-1 g/ml), flagellin from S. typhimurium (0.1-10 g/ml), LTA from S. aureus (0.1-10 g/ml), PGN from S. aureus (0.1-10 g/ml), Pam 3 CSK 4 (0.1-10 g/ml), FSL-1 (0.01-1 g/ml), MALP-2 (0.01-1 g/ml), and A23187 (0.1-10 M) for 60 min, and then the amounts of VWF released into the media were measured. Each value is the mean Ϯ S.D. (n ϭ 3).
LTA for 60 min was not suppressed by treatment with the protein synthesis inhibitor cycloheximide, whereas the release by stimulation with LTA for 4 h was significantly suppressed by the treatment (Fig. 1C). We further investigated whether LTA induction of exocytosis was mediated through MyD88 and IRAK-1, common signaling molecules downstream of TLRs, because LTA is known as a TLR2 agonist. Interestingly, VWF release by stimulation with LTA for 60 min was suppressed by knockdown of the expression of MyD88 but not that of IRAK-1, whereas the release by stimulation with LTA for 4 h was significantly suppressed by each knockdown of MyD88 and IRAK-1 (Fig. 1C). Thus, these results suggest that LTAcaninduceWeibel-PaladebodyexocytosisthroughaMyD88dependent rapid mechanism without de novo protein synthesis and an IRAK-1-dependent slower mechanism with de novo protein synthesis.
We also examined whether other bacterial cell wall constituents, as shown in Table 1, activated induction of VWF release after stimulation of HAECs for 60 min. Among the compounds that we tested, the synthetic analogs of bacterial lipoproteins Pam 3 CSK 4 , FSL-1, and MALP-2 and, to a lesser extent, flagellin induced VWF release in a dose-dependent manner (Fig. 1D,  left). Interestingly, LPS from different bacterial species and PGN did not activate Weibel-Palade body exocytosis (Fig. 1D,  left). In addition, we found that induction of exocytosis by bacterial compounds was also mediated by MyD88 as well as that by LTA (Fig. 1D, right). These results suggest that several types of, but not all, bacterial cell wall constituents can activate induction of TLR-MyD88-mediated exocytosis.
Regarding the process of Weibel-Palade body exocytosis, we found that MyD88-dependent externalization of P-selectin was induced after stimulation of HAECs with LTA for 30 min (Fig.  2A). In addition, monocyte adhesion to HAECs was modestly increased in a MyD88-dependent fashion after LTA stimulation for 60 min (Fig. 2B).
Stimulatory Activities of LPS and PGN in HAECs-As stated above, LPS did not activate Weibel-Palade body exocytosis (Fig.  1D). However, LPS potently activated induction of MyD88-dependent IL-8 production in HAECs after stimulation for 4 h (Fig.  3A). Thus, the results shown in Figs. 1D and 3A suggest that endothelial TLR4 lacks the ability to induce rapid Weibel-Palade body exocytosis without de novo protein synthesis. Similarly to LPS, PGN did not activate Weibel-Palade body exocytosis (Fig. 1D). Also, PGN did not induce IL-8 production after stimulation for 4 h in HAECs, whereas LTA did (Fig. 3A). However, our preparation of PGN had activities to induce TNF-␣ production in THP-1 monocytes (Fig. 3B) and TLR2-and MyD88-dependent activation of NF-B in HEK293 cells (Fig. 3C) in a way similar to that in the case of other TLR2 agonists. These results suggest that HAECs lack the ability to respond to PGN.
Induction of Weibel-Palade Body Exocytosis through TLR2-We then focused on LTA-and bacterial lipopeptide-induced Weibel-Palade body exocytosis. It has been reported that LTA and bacterial lipopeptides are TLR2 agonists (Table 1). In HUVECs, the lipopeptide FSL-1 induced VWF release (Fig. 4A). We found that this response was enhanced by increased expression of TLR2 by gene transfection (Fig. 4A). This result suggests that TLR2 recognition of bacterial constituents directly activates Weibel-Palade body exocytosis. Moreover, transfection of mutated TLR2 (P681H), which lacks the ability to interact with MyD88 (26), suppressed the release (Fig. 4B), consistent with the results presented in Figs. 1D and 2A showing that MyD88 was involved in the induction of Weibel-Palade exocytosis. In HAECs, knockdown of TLR2 expression resulted in almost complete suppression of VWF release by Pam 3 CSK 4 , FSL-1, MALP-2, and LTA (Fig. 4B). Moreover, knockdown of TLR6 expression resulted in a decrease in the activities of LTA, FSL-1, and MALP-2 and even that of Pam 3 CSK 4 (Fig. 4B). In contrast to this, TLR1 interference did not affect VWF release (Fig. 4B), consistent with our observation that HAECs express very low levels of TLR1 mRNA compared with the levels of TLR2 mRNA (data not shown). These results suggest that endothelial recognition of pathogens by TLR2, or to a lesser extent by TLR6, contributes to induction of Weibel-Palade body exocytosis.

Involvement of PLC␥ Activation in Weibel-Palade Body Exocytosis-
Recent studies have shown that TLR2 signal transduction results in an increase of intracellular calcium level (27,28). Indeed, we found that the intracellular calcium chelator BAPTA-AM suppressed LTA-induced exocytosis (Fig. 5A). We FIGURE 2. MyD88-dependent P-selectin externalization by LTA. A, HAECs transfected with MyD88-specific or control siRNA were stimulated with 10 g/ml LTA for 60 min, and then surface P-selectin was detected by flow cytometry. Shaded histogram, not stimulated; gray, stimulated with LTA. B, HAECs transfected with MyD88-specific or control siRNA were stimulated with 10 g/ml LTA for 60 min, and monocytes stained with conA-Alexa594 were then allowed to adhere for 20 min. Adhesion of red fluorescent cells was quantified in three fields per well by using an image analysis system. Each value is the mean Ϯ S.D. (n ϭ 3). *, p Ͻ 0.01.  (Fig. 5C). Interestingly, this activity was efficiently suppressed by knockdown of MyD88 expression but not by knockdown of IRAK-1 expression (Fig. 5C). MyD88-dependent activation of PLC␥ was also observed in TLR2-overexpressed 293 cells used as non-endothelial cells (data not shown). These results suggest that TLR2mediated rapid Weibel-Palade body exocytosis is regulated by activation of PLC␥ through MyD88-dependent tyrosine phosphorylation.
We also investigated the role of PLC␥ in TLR2-mediated NF-B signaling. U-73122 treatment clearly suppressed TLR2 agonist-induced production of the NF-B-driven chemokine IL-8 in HAECs (Fig. 5D). U-73122 treatment also suppressed LTA-induced phosphorylation and degradation of IB␣ in HAECs (Fig. 5E). These results suggest that the MyD88-PLC␥ pathway also mediates inflammatory responses through NF-B activation in endothelial cells.
Regulation of TLR2-mediated Weibel-Palade Body Exocytosis-The results shown in Fig. 4 (A and B) raised the possibility that alteration of endothelial TLR2 expression affects the magnitude of Weibel-Palade body exocytosis. We examined TLR2-mediated exocytosis in the presence of vascular modulators, IFN-␥ or laminar flow, which are known to affect TLR2 expression in endothelial cells of human origin. Consistent with the results of a previous study (32), treatment with IFN-␥ increased TLR2 expression level in HAECs (Fig. 6A). Under this condition, the magnitude of TLR2-mediated exocytosis was significantly increased (Fig. 6B). In contrast to this, TLR2 expression slightly decreased in HAECs incubated under laminar flow (Fig. 6C), consistent with the results of a previous study (33). We found that laminar flow decreased the magnitude of TLR2-mediated exocytosis (Fig. 6D).

DISCUSSION
The major finding of this study is that aortic endothelial cells respond to several bacterial constituents that stimulate TLR2, leading to induction of Weibel-Palade body exocytosis through a MyD88-dependent mechanism without de novo protein synthesis. During this process, release of VWF and externalization of P-selectin were induced, by which rolling and adhesion of platelets and leukocytes and thrombus formation in the local vessel walls may be promoted (34,35). The pathological role of this phenomenon in vivo may be support by the observations in mouse experiments, i.e. slight increases of local leukocyte-endothelial interaction after LTA administration (36) and soluble P-selectin level in serum after administration of the synthetic  (n ϭ 3). B, THP-1 cells were stimulated with PGN (0.1-10 g/ml), LTA (0.1-10 g/ml), and FSL-1 (10 -1 g/ml) for 6 h, and then the amounts of TNF-␣ released into the media were measured. Each value is the mean Ϯ S.D. (n ϭ 3). C, HEK293 cells stably expressing TLR2, and control cells were prepared and then transfected with MyD88-specific siRNA and NF-B-driven luciferase gene. The cells were stimulated with Pam 3 CSK 4 (0.1-10 g/ml), FSL-1 (0.01-1 g/ml), MALP-2 (0.01-1 g/ml), LTA (0.1-10 g/ml), PGN (0.1-10 g/ml), and LPS (1-100 ng/ml) for 6 h, and then luciferase activity was measured. Each value is the mean Ϯ S.D. (n ϭ 3). A, HUVECs transfected with WT or P681H mutant of TLR2 or with a control plasmid were stimulated with 1 g/ml FSL-1 for 60 min, and then the amounts of VWF released into the media were measured. Each value is the mean Ϯ S.D. (n ϭ 3). B, HAECs transfected with TLR1-, TLR2-, or TLR6specific or control siRNA were stimulated with Pam 3 CSK 4 (10 g/ml), FSL-1 (1 g/ml), MALP-2 (1 g/ml), and LTA (10 g/ml) for 60 min, and then the amounts of VWF released into the media were measured. Each value is the mean Ϯ S.D. (n ϭ 3). *, p Ͻ 0.05.

TLR2 Mediates Weibel-Palade Body Exocytosis
lipopeptide FSL-1. 3 Sequentially or simultaneously, both PLC␥and IRAK1-mediated signaling pathways activate NF-B, by which production of various proinflammatory cyto-kines, and expression of adhesion molecules such as ICAM-1 are induced to promote adherence and activation of platelets and leukocytes (37). The delayed Weibel-Palade body exocytosis with de novo protein synthesis is further activated in the cells. Therefore, endothelial TLR2 may be able to function as a primary initiator and a modulator of artery inflammation through these early-phase endothelial responses after recognition of cognate agonists.
We investigated the responsiveness of HAECs toward common bacterial constituents. For the TLR2 agonists, we prepared several compounds that have already been proposed to function as TLR2 agonists, because TLR2 forms a complicated recognition system and because human endothelial cells from different vascular beds show different degrees of responsiveness to TLR2 agonists (32,38,39). Unexpectedly, PGN, unlike other TLR2 agonists, could not activate either Weibel-Palade body exocytosis or IL-8 production (Figs. 1D and 3A). The issue of recognition of PGN by TLR2 is still controversial. The existence of an intracellular receptor for PGN (NOD2) further complicates this matter. However, Gupta's group recently concluded that PGN is in fact recognized by TLR2 by showing that muramidase treatment of PGN abolished the TLR2-stimulating activity (8). We showed that recognition of our PGN was at least dependent on TLR2 (Fig. 3A). It has been shown that PGN directly binds TLR2 per se (40), whereas bacterial lipopeptides

TLR2 Mediates Weibel-Palade Body Exocytosis
are thought to directly interact with TLR2-associated molecules such as CD14 and LBP but not with TLR2 per se (7,41,42), suggesting the existence of different ligand-recognition mechanisms by TLR2. Furthermore, a novel family of PGN-binding proteins such as peptidoglycan recognition proteins has been found (43) and might enable discrimination of PGN from other TLR2 agonists. Thus, PGN may be recognized by a TLR2 recognition system different from that for LTA and lipoproteins/ lipopeptides. Collectively, HAECs express functional TLR2 to respond to several TLR2 agonists, including lipopeptides and LTA, but may lack a PGN-recognition system resulting in an inability to respond to PGN. Moreover, aortic endothelial cells may particularly recognize diacylglyceride-containing bacterial lipid derivatives (LTA and bacterial lipopeptides), recognition of which has recently been reported to depend on TLR6 and CD36 (11).
We also showed that the TLR4 agonist LPS did not activate Weibel-Palade body exocytosis (Fig. 1D). Although the reason for this is not clear, several lines of evidence obtained in previous studies may provide an explanation. For example, TLR4 expression has been reported to localize intracellularly in artery endothelial cells (44). This observation suggests that TLR4 in artery endothelial cells may be lacking in induction of phospholipid-dependent signaling events, including PLC␥ activation, which are commonly intrinsic to the signaling receptors spanning the cell membrane. Further investigation is needed to determine the reason.
Several properties of endothelial TLR2 have been proposed to be involved in the development of atherosclerosis. First, endothelial TLR2 expression is enhanced by proinflammatory stimuli, such as TNF-␣, IFN-␥, and LPS (32), and by SP-1-dependent machinery in areas of disturbed blood flow such as lesion predilection within the aortic tree and heart (33). The expression level of TLR2 is indeed increased in an atherosclerotic lesion in humans (45). Furthermore, a recent study has revealed that complete deficiency of TLR2 in atherosclerosisprone LDLR-null mice leads to an apparent reduction in the formation of lesions (46). Proinflammatory signaling pathways downstream of TLR2 have been thought to be activated through TIRAP/Mal, MyD88, IRAK-1, and TRAF6 in endothelial cells. Other pathways involving PI3K and the downstream protein kinase Akt/PKB (47), the Rho family GTPase Rac1 (48), and the redox-activated mitogen-activated protein kinase kinase kinase ASK1 (49) also link TLR2 signaling to the NF-B pathway. In this study, we showed that PLC␥ also mediated the NF-B pathway downstream of TLR2 in HAECs, although involvement of PLC␥ in the TLR2 proinflammatory signaling has been described in several reports (27,50). Because PLC␥ isoforms are thought to be activated by both generation of phosphatidylinositol 3,4,5-triphosphate by PI3K and tyrosine phosphorylation, we found the latter process downstream of TLR2 was dependent on MyD88 but not IRAK-1 (Fig. 5C). Recent studies have suggested a linkage of TLRs and tyrosine kinases, including Syk via MyD88-STAP-2 interaction (51) and Btk via direct interaction with TIR domain (52), both of which have been shown to activate PLC␥ isoforms. Moreover, Btkinduced phosphorylation of TIRAP/Mal has recently been reported to play an important role in TLR signal transduction (53), which may occur at a phosphatidylinositol diphosphaterich membrane compartment after recruitment of MyD88 to membrane-localized TIRAP/Mal (54). A schematic of signaling pathways proposed here is shown in Fig. 7.
Endothelial activation by several proinflammatory agents has been shown to increase endothelial responsiveness toward TLR2 agonists via up-regulation of TLR2 expression (32). Increased endothelial TLR2 expression increased the magnitude of TLR2-mediated exocytosis of Weibel-Palade bodies (Fig. 6B) and endothelial responses (38), suggesting enhanced responsiveness of endothelial cells to pathogens in inflamed lesions. In contrast, fluid shear decreased the magnitude of TLR2 ligand-stimulated Weibel-Palade body exocytosis (Fig.  6D). Physiological fluid shear stress has been suggested to have atheroprotective effects in vivo, because atherosclerosis preferentially occurs in an area of disturbed flow or a low level of shear stress, whereas regions with steady laminar flow and physiological shear stress are protected. Disturbed flow or a low level of shear stress has been reported to regulate expression of various regulatory molecules of endothelial activation, by which atherosclerotic processes may be accelerated in the sites. These observations are consistent with the previous finding that physiological fluid shear stress decreases endothelial TLR2 expression via impaired activity of the transcriptional factor SP1 (33). Thus, our results raise the possibility that bacterial constituentinduced Weibel-Palade body exocytosis can be physiologically or pathologically regulated in particular circumstances of the vessel wall.
In conclusion, our study focused on endothelial exocytosis induced by bacterial pathogens and showed a linkage between endothelial innate recognition of pathogens and early-phase endothelial inflammatory responses. Our results may provide a new insight into the role of endothelial TLR2 in the initiation and modulation of vascular inflammation or atherogenic responses.