Thromboxane A2 exacerbates acute lung injury via promoting edema formation

Thromboxane A2 (TXA2) is produced in the lungs of patients suffering from acute lung injury (ALI). We assessed its contribution in disease progression using three different ALI mouse models. The administration of hydrochloric acid (HCl) or oleic acid (OA)+ lipopolysaccharide (LPS) caused tissue edema and neutrophil infiltration with TXA2 production in the lungs of the experimental mice. The administration of LPS induced only neutrophil accumulation without TXA2 production. Pretreatment with T prostanoid receptor (TP) antagonist attenuated the tissue edema but not neutrophil infiltration in these models. Intravital imaging and immunostaining demonstrated that administration of TP agonist caused vascular hyper-permeability by disrupting the endothelial barrier formation in the mouse ear. In vitro experiments showed that TP-stimulation disrupted the endothelial adherens junction, and it was inhibited by Ca2+ channel blockade or Rho kinase inhibition. Thus endogenous TXA2 exacerbates ALI, and its blockade attenuates it by modulating the extent of lung edema. This can be explained by the endothelial hyper-permeability caused by the activation of TXA2-TP axis, via Ca2+- and Rho kinase-dependent signaling.

Inflammatory lipid mediators, prostanoids (PGs), are produced enzymatically by the activation of cyclooxygenase (COX) and prostaglandin/thromboxane synthases from fatty acids. There are five major PGs; prostaglandin D 2 (PGD 2 ), PGE 2 , PGF 2α , PGI 2 and thromboxane A 2 (TXA 2 ) which are strongly involved in the inflammatory response. Previous studies have focused on the contribution of COX and PGs in ALI/ARDS [8][9][10][11] . Elevated levels of PGs have been reported in bronchoalveolar lavage (BAL) fluid obtained from patients with ARDS 12 . Hinshaw et al. originally found that COX inhibition prevented the development of sepsis and improved survival rates in dogs 8 . Another group reported that in mouse transfusion-related ALI/ARDS model, treatment with a COX inhibitor, aspirin, ameliorated lung edema and increased the survival rate 9 . These findings suggest that COX-mediated production of PGs is crucial for the initiation and progression of lung inflammation. However, in a clinical study, treatment with a nonselective COX inhibitor ibuprofen did not reduce the incidence of ARDS in patients with sepsis 10 . These observations suggest that PGs play a multifaceted role, being both pro-inflammatory and/or anti-inflammatory mediators in the pathophysiology of airway inflammation. To reveal this complexity, detailed evaluation of the role of each PG using multiple models is indispensable.
Previous studies revealed that major PGs including PGE 2 and PGI 2 act mostly as anti-inflammatory PGs in ALI/ARDS. Treatment with the PGE 2 or PGI 2 analog, beraprost, reduces the increase in BAL cell count and the extravasation of BAL proteins in the ventilator-related ALI mouse model 13 . Anti-inflammatory roles of lipoxin A 4 or 15-deoxy Δ12,14 prostaglandin J 2 have been reported in mouse HCl-or carrageenin-induced ALI/ARDS models, respectively 14,15 . We also reported that deficiency of hematopoietic PGD synthase accelerates, and stimulation of PGD 2 receptor inhibits induction of edema and neutrophil accumulation in the lungs of LPS-induced ALI/ ARDS model mice 16 . Given that the pathophysiological action of each PG varies with its target cells, context of activation, and pathogenesis, a detailed evaluation of the role of each PG using multiple models is indispensable to reveal this complexity and to overcome ALI/ARDS. TXA 2 is one of the prostanoids generated in several cell types such as platelets, monocytes, macrophages, and epithelial cells 17 . TXA 2 is produced by the action of COX, followed by thromboxane synthase (TXS). TXA 2 binds to a GPCR, T prostanoid (TP) receptor, that couples to the G q or G 12/13 signaling molecules 18 . TP stimulation is known to cause a broad range of cellular responses such as platelet aggregation and vasoconstriction 18 . In the inflamed lungs of ALI/ARDS patients, TXA 2 as well as the other PGs mentioned before, were detected 19 . Experimental studies have suggested that TXA 2 has a pro-inflammatory role in lung inflammation [20][21][22] . The treatment with a TXA 2 synthase inhibitor, ozagrel, restored the impaired respiratory function and decreased arterial O 2 pressure in guinea-pig ALI induced by OA 21 . The treatment with a TP antagonist, SQ29548, improved the symptoms of HCl-induced mouse ALI/ARDS 22 . Thus, TXA 2 has been suggested to be involved in the progression of ALI/ARDS. However, it remains unclear how TXA 2 -signaling exacerbates ALI/ARDS.
We investigated the role of TXA 2 -signaling in three different models of ALI/ARDS. We found that production of TXA 2 varies according to the pathogenesis of ALI and its production positively correlates with induction of edema, but not with neutrophil accumulation in inflamed lungs. Mechanistically, TXA 2 stimulates the endothelial TP receptor and disrupts endothelial barrier via Ca 2+ /Rho kinase-signaling. These findings reveal pathophysiological implications of TXA 2 and might provide novel therapeutic targets for treating ALI.

Results
TXA 2 exacerbates HCl-induced ALI. We first investigated the role of TXA 2 in the acid aspiration-induced ALI model. The administration of HCl (intra-nasally, 0.1 M, 2.5 μ l/g, 6 h) caused severe hemorrhage in the lungs of the mice (Fig. 1A). In these mice, infiltrative shadow indicating tissue inflammation was observed, especially around bronchi in computed tomography (CT) imaging (Fig. 1B, indicated by arrowhead). Treatment with a TP receptor antagonist (SQ29548, i.p., 2 mg/kg, 0, 2 and 4 h after HCl administration) significantly inhibited these features. The HCl administration also impaired respiratory function indexed as the value of saturation of peripheral oxygen (SpO 2 , Fig. 1C). The treatment with SQ29548 significantly recovered the HCl-induced respiratory dysfunction.
We next measured the amount of a stable TXA 2 metabolite TXB 2 in BAL fluids. In the vehicle-administered mice, no TXB 2 was detected, while 284 ± 118 pg/ml of TXB 2 was detected in the HCl-administered mice (Fig. 1D). The HCl-administration also increased the production of PGE 2 ( Fig. 1E) but it did not change the amount of a PGI 2 metabolite, 6-keto PGF1α in the BAL fluid (Fig. S1).
These results suggested that the HCl-administration stimulates TXA 2 -TP signaling which leads to severe inflammation and impaired respiratory function.

TXA 2 -TP signaling activation causes lung edema in HCl-induced ALI. Morphological studies
showed that the HCl administration caused pulmonary hemorrhage and neutrophil accumulation, especially around the bronchi ( Fig. 2A, neutrophils are indicated by black arrow heads in lower panels). These phenomena were accompanied with the leakage of protein rich fluid into tissue interstitium (indicated by white arrowheads). The treatment with SQ29548 reduced hemorrhage and leakage of protein rich fluid. Interestingly, neutrophil accumulation was observed even in the SQ29548-treated group.
We evaluated the effects of TXA 2 -TP signaling on two major pathological features of ALI; lung edema and neutrophil accumulation. The administration of HCl significantly increased water content, which is an index of tissue edema, and the activity of myeloperoxidase (MPO) which is an index of neutrophil infiltration (Fig. 2B). The treatment with SQ29548 almost completely suppressed edema formation. In contrast, this regimen slightly, but not significantly, reduced the HCl-induced neutrophil infiltration into the lung tissue (Fig. 2C).
We further examined whether TP agonism itself affects lung edema and neutrophil accumulation. Intraperitoneal treatment with a TP agonist (U46619, 25 μ g/kg, 1 h) induced lung edema ( Fig. 2D) but did not affect MPO activity in lung (Fig. 2E). These results showed that TXA 2 -TP signaling exacerbates the HCl-induced ALI via induction of edema.
Scientific RepoRts | 6:32109 | DOI: 10.1038/srep32109 Epithelial cells produces TXA 2 in HCl-induced ALI. We examined what type of cells produced TXA 2 in the HCl-induced ALI. Immunohistochemistry showed that the HCl-administration induced TXS expression along alveolar walls (Fig. 2F). Consistently, immunofluorescence study showed that E-cadherin positive epithelial cells strongly express TXS in the HCl-treated mouse lung (Fig. 2G,H), suggesting that epithelial cells are the major source of TXA 2 in the HCl-induced ALI.
The role of TXA 2 -TP signaling in oleic acid-induced ALI. Oleic acid (OA) causes severe histopathological changes, similar to those observed in ARDS. Administration of OA (150 μ l/kg) in addition to LPS (1.5 mg/kg) caused severe alveolar damage. At 6 h after the LPS administration, damage to the lung was characterized by extensive pulmonary hemorrhage (Fig. S2A) and broad infiltrative shadow in CT imaging (Fig. S2B). However, this stimulation did not cause respiratory dysfunction (Fig. S2C). As shown in Fig. 3A, morphological studies showed that the LPS + OA-administration caused severe hemorrhage, protein leakage (middle panel, indicated by while arrowheads), and neutrophil accumulation (indicated by black arrow heads). Consistently, tissue water   TXA 2 production was abolished by the pretreatment with a TXS inhibitor ozagrel (i.p., 50 mg/kg, 15 min before OA administration). Pretreatment with either SQ29548 (i.p., 2 mg/kg, 0, 2 and 4 h after LPS administration) or ozagrel effectively diminished the injuries in this model (Fig. 3A, right panels and Fig. 3B). As observed in the HCl-induced ALI model, these regimens did not rescue the neutrophil accumulation in the LPS + OA-treated lungs (Fig. 3C).
The role of TXA 2 -TP signaling in LPS-induced ALI. We next utilized the LPS-induced lung inflammation model, which reproduces the sepsis-associated ALI/ARDS. The administration of LPS (intra-nasally, 3.75 mg/kg, 6 h) caused lung inflammation accompanied with intensive neutrophil accumulation (Fig. 3E, middle panel, indicated by black arrowheads), as well as an increase in MPO activity in the lungs (Fig. 3G). This stimulation did not cause severe hemorrhage (Fig. S3A), lung dysfunction (Fig. S3B), and pulmonary edema (Fig. 3F), which were observed in the HCl-or LPS + OA-induced ALI. The treatment with SQ29548 (i.p., 2 mg/kg, 0, 2 and 4 h after LPS administration) did not have any beneficial action against the neutrophil accumulation induced by LPS (Fig. 3E, right panel). Although the LPS-administration increased the amounts of TXB 2 and PGE 2 in the BAL fluid (Figs 3H and S2C), its amount (36 ± 24 and 301 ± 34 pg/ml, respectively) was much smaller than that in the HCl-or LPS + OA-treated lungs (Figs 1D,E, 3D and S2D). Thus, the production and contribution of TXA 2 seems to be relatively lower in this model.

TXA 2 increases vascular permeability in vivo.
Sustained vascular hyper-permeability can cause tissue edema. Therefore, we investigated the effect of TXA 2 -TP signaling on vascular permeability in vivo. A TP agonist U46619 was applied to the right ear of mice followed by the injection of Evans Blue for 30 min. As shown in Fig. 4A (representative pictures) and 4B (summary), treatment with 20-200 μ g/ear U46619 extravasated blue dye in mouse ear in a dose-dependent manner.
Ear vasculature is mainly composed of proximal vessels, distal vessels, and capillaries (Fig. S4A). In proximal vessels or distal vessels, endothelial cells are covered by vascular mural cells such as smooth muscle cells or pericytes. Capillaries are composed only of endothelial cells. Contraction of mural cells reduces blood flow which decreases the intravascular hydrostatic pressure and vascular permeability. Given that TXA 2 is known as a vasoconstrictor, we examined the effect of TXA 2 on vascular leakage and contraction in vivo. Intravital microscopy showed that the application of U46619 (20 μ g/ear, 15 min) caused the FITC-dextran leakage in capillaries (Fig. 4C). However, this regimen induced contraction in veins but not significantly in arteries ( Fig. 4D: representative picture, 4E and 4F: summaries). Laser Doppler blood-flow imaging showed that the application of U46619 (20 μ g/ear, 20 min) significantly decreased the blood flow in mouse pinna (representative picture: Fig. 4G, summary: 4H). Thus, TP-stimulation may limit blood flow by inducing vascular contraction in mouse ear.
We next assessed the effect of U46619 on endothelial barrier by observing intercellular adherens junctions in the ear. Whole-mount immunostaining showed that an intracellular adhesion molecule, VE-cadherin, was located at cell-cell contact areas under non-stimulated conditions in venulae (Fig. 4I, left panel). Changes in VE-cadherin localization, internalization or disassembly, induces endothelial barrier disruption. The administration of U46619 (20 μ g/ear, 15 min) caused partial disassembly of VE-cadherin, as indicated by the arrowheads in Fig. 4I (middle panel). Exposure to histamine (400 μ g/ear, 15 min), a well-characterized endothelial barrier disruptor, also triggered VE-cadherin disassembly (Fig. 4I, right panel). Figure S4B shows that similar results were obtained by en face immunostaining of mouse pulmonary artery. VE-cadherin was located at cell-cell contact areas in the absence of stimulation (left panel). The treatment with U46619 (1 μ M) disassembled VE-cadherin (middle panel. indicated by black arrows heads). The treatment with histamine (10 μ M) also caused VE-cadherin disassembly (right panel).
Using immunostaining, we evaluated the effect of TXA 2 on the endothelial barrier in vitro. In the basal condition, VE-cadherin and F-actin were aligned along cell-cell adhesion area of HUVECs indicating adherens junction formation (Fig. 5D, left panels). The administration of U46619 (1 μ M, 20 min) destroyed VE-cadherin alignment (indicated by arrowheads) and induced actin stress fiber formation (indicated by arrows), which are characteristic phenomena of adherens junction disassembly. Pretreatment with Y27632 (10 μ M, 30 min pretreatment) rescued the U46619-induced disruption of adherence junction (right panels).

Discussion
Although several pharmacological reagents including corticosteroids and adrenaline β 2 agonists are known to have a beneficial therapeutic effect on ALI and ARDS, they do not improve the mortality in large scale clinical trials 4,23 . As of now, only supportive treatment is available and the mortality rate of ALI is still high 24 . ALI and ARDS are complex syndromes with severe and rapidly progressing respiratory dysfunction caused by various pathogenic factors. Therefore, there is an urgent need to understand the mechanism of disease progression and develop effective and appropriate therapeutic strategies that considers the pathogenic mechanism.
Using three different ALI models, we showed that the production of a lipid mediator TXA 2 resulted in different pathogenicity, and that TXA 2 also contributed to lung edema by disrupting the endothelial barrier. Comparing these three models, the amounts of TXA 2 (indexed as TXB 2 ) positively correlated with the edema formation (indexed as water content, Fig. S6, left panel) but not neutrophil infiltration (indexed as MPO activity, Fig. S6, right panel). These observations highlight our conclusion that TXA 2 is the critical factor of edema formation in ALI progression.
HCl aspiration directly injures alveolar epithelial cells and causes lung edema neutrophil accumulation and respiratory dysfunction 25 . We propose in this study, that epithelial cells express TXS and are the major source of TXA 2 in the HCl-induced ALI. Alveolar damage by chemical and/or physical stimulation may be attributed to the TXS induction and TXA 2 production, and subsequent edema formation in ALIs. Previous studies showed that the administration of HCl activated platelets, which in turn produced TXA 2 resulting in lung edema 22 . However, we could neither detect significant increase in the number of platelets nor in TXS expression in platelets in our current model (Fig. S7). Further investigations are required to clarify this point.
The administration of HCl induced lung edema and neutrophil accumulation accompanied with TXA 2 production. This stimulation impaired respiratory function. The LPS + OA challenge induced more severe manifestation of the disease with TXA 2 production than the HCl did, while it did not cause respiratory dysfunction. This discrepancy seen between lung dysfunction and other features may be due to the difference in the pathogenic mechanisms of these two models. Aspirated HCl, but not OA directly injures and then induces necrosis of the epithelial cells which play a crucial role in gas exchange 26 . Although tissue edema, i.e. accumulation of fluid in the pulmonary interstitium, also influences gas exchange, its effects on the pathophysiology of ALI seem to be relatively small at least in the present models. This difference in the pathogenic mechanism probably reflects the responsiveness to TP antagonism. The effect of TP antagonism on lung dysfunction was limited in the HCl-induced ALI (Fig. 1C). Inhibition of induction of edema by TP antagonism may be insufficient to restore the impaired gas exchange induced by the direct epithelial damage.
The LPS instillation did not induce lung edema. Supporting our hypothesis that TXA 2 production may be responsible for edema in ALI, PGE 2 but not TXB 2 was detected in these lungs. In contrast to our observations, previous reports showed that LPS challenge induces edema in mouse lung [27][28][29] . We also previously showed that the LPS instillation caused lung hyper-permeability 16 . This discrepancy may be because of the differences in the time points selected for observation. In the present study, we examined symptoms at 6 h after the LPS treatment while the other studies made the observations 24 h or later after the treatment. We selected the 6 h-LPS instillation which is enough to cause neutrophil accumulation but not for induction of edema and lung dysfunction. There is a possibility that the longer period of stimulation with LPS may cause lung edema accompanied by TXA 2 production.
The current study demonstrates that TP antagonism abolished the hyper-permeability and induction of edema, but it did not affect neutrophil migration in the lungs. We also confirmed that a strong migratory factor leukotriene B 4 , but not TXA 2, stimulates neutrophil migration activity in vitro (Fig. S8). These observations support the idea that TXA 2 exacerbates ALI by disrupting endothelial barrier without influencing neutrophil activity. However, the vascular hyper-permeability is an initial step in neutrophil infiltration into the inflamed tissue. Infiltrated neutrophils produce proteases and reactive oxygen species, which in turn destroy endothelial barrier and induce further edema. Thus, it is difficult to clearly dissociate these two phenomena in vivo.
Although we here focused on the role of TXA 2 , the HCl administration elevated the amount of PGE 2 in the BAL fluid (Fig. 1E). Previous studies showed that PGE 2 administration enhanced endothelial barrier resulting in relief of symptoms in the LPS-induced ALI 30 . PGE 2 produced may counteract excessive inflammatory response by reducing endothelial permeability and inflammatory cell infiltration in ALI.
Endothelial barrier disruption increases the leakiness of the vascular wall. In addition, we and others showed that tissue blood flow is the important factor of vascular permeability. The increase in tissue blood flow resulting from vasodilation elevates intravascular hydrostatic pressure, which intensifies outward transfer of serum components. Several physiological substances differently and uniquely modulate these two factors and regulate vascular permeability in vivo. Histamine induces vascular hyper-permeability through both blood flow increase and endothelial barrier disruption 31 . PGE 2 increases tissue blood flow and induces vascular hyper-permeability while it enhances endothelial barrier integrity 32 . We have shown herein that TP-stimulation caused hyper-permeability in the skin ear capillary vessel. TXA 2 analog U46619 reduced ear blood flow by inducing vasocontraction, while it disrupted endothelial barrier of microvessels (Fig. 4G,I). Lungs are composed mainly of endothelial cell-rich capillary vessels, which seem to be more sensitive to TXA 2 . Excessive TXA 2 production can cause significant hyper-permeability resulting in severe edema in the ALI. TXA 2 is known to increase endothelial monolayer permeability in bovine aortic endothelial cells and HUVECs 33,34 . Consistently, we showed that endothelial TP agonism disrupted endothelial barrier in HUVECs. TP receptor is G q -and G 12/13 -protein coupled receptor and its activation elevates intracellular Ca 2+ concentration and Rho activity 18 . Ca 2+ increase and Rho activation induces actin cytoskeletal rearrangement such as stress fiber formation and myosin phosphorylation resulting in endothelial barrier disruption 35 . As expected, blockade of these signaling pathways impaired TP-mediated stress fiber formation and endothelial hyper-permeability (Fig. 5). These results allowed us to conclude that TP agonism increased endothelial permeability through Ca 2+ / Rho-mediated actin rearrangement.
In summary, the present study shows that TXA 2 exacerbates ALI mainly by causing lung edema and that TXA 2 -TP signaling induces endothelial hyper-permeability through Ca 2+ /Rho signaling. These findings might help in understanding the pathological process of ALI, and provide new insight for therapeutic application of TP antagonists. We have utilized three different murine ALI models; HCl-, LPS + OA-, and LPS-induced ALI, which mimic acid aspiration-, lipid embolism-, or sepsis-induced human ALI. There are several other causative factors of ALI including mechanical ventilation and blood transfusion. Further detailed examination of each model and each disease stage will be indispensable to develop better therapy against ALI/ARDS.

ALI Induction.
All the experiments were approved by the Institutional Animal Care Use Committee of the University of Tokyo. All experimental methods were performed in accordance with the approved guidelines. Eight-week old male C57BL/6J mice or BALB/c mice were purchased from CLEA Japan, Inc. (Tokyo, Japan). Mice were anesthetized with 3% isoflurane and maintained with 2% isoflurane. To make the HCl-or LPS-induced ALI model, HCl (0.1 M, 2.5 μ l/g) or LPS (3.75 mg/kg) was intra-nasally administrated to C57BL/6J mice. To make LPS + OA-induced ALI model, LPS (1.5 mg/kg) was intra-nasally administrated, and 3 min later, OA (0.15 ml/kg) was intravenously injected into C57BL/6J mice. Six hours after the HCl or LPS instillation, mice were anesthetized Scientific RepoRts | 6:32109 | DOI: 10.1038/srep32109 and utilized for saturation of peripheral oxygen (SpO 2 ) measurement or computed tomography analysis. For some experiments, the mice were euthanized and lungs were immediately perfused with normal physiological saline solution (33 mmHg).
Computed Tomography Analysis. Mice were put into the chamber of the micro-CT system (LaTheta LCT-200, Aloka, Tokyo, Japan). X-ray voltage was 50 kV, the field of view was 48 × 48 mm, and the interval of scanning was 192 μ m. The data were analyzed by the software provided by LaTheta.

Saturation of peripheral oxygen (SpO 2 ) Measurement. A mouse pulse oximeter (STARR Life
Sciences, Oakmont, PA, USA) was used to monitor SpO 2 . The MouseOX collar clip was put on the neck of the anesthetized mice. After 3-4 min, the readings of pulse oximetry were recorded. The rate of sampling was once a second.
Morphological Analysis. The lungs from the experimental mice were fixed in 4% paraformaldehyde (PFA) for 24 h. Four μ m-thick paraffin sections were stained with hematoxylin and eosin (H&E). For immunostaining, frozen 4 μ m-thick sections were incubated with 0.3% Triton X-100 and 3% normal goat serum for 30 min, the sections were incubated with rat anti-mouse CD324 (1:200) and rabbit anti-thromboxane synthase antibody (1:100) overnight at 4 °C. After labeling with secondary antibodies, the samples were observed under a fluorescence microscope (Eclipse E800, Nikon, Tokyo, Japan). Some sections were used for chromogenic reaction with 3, 3′-diaminobenzidine (DAB) and observed under a light microscope (Optiphot-2, Nikon, Tokyo, Japan).
For whole-mount immunostaining of ear, mice were euthanized and perfused with 4% PFA. The ears were excised and ventral part of the skin was removed. The samples were permeabilized with 0.3% Triton X-100 for 30 min and blocked with 3% bovine serum albumin for 30 min. For cell staining, human umbilical vein endothelial cells (HUVECs, Lonza, Basel, Switzerland, passage 4-8) were seeded on glass coverslips. After the fixation with 4% PFA for 5 min, the cells were incubated with 0.1% Triton X-100 and 3% bovine serum albumin for 30 min. The ear or cell samples were incubated with goat anti-VE-cadherin antibody (1:200) overnight at 4 °C, and then incubated with rhodamine-phalloidin (3:100) at room temperature for 1 h. After labeling with secondary antibodies, the samples were observed under the fluorescence microscope.
Water Content Measurement. Excised lungs were weighed to obtain the wet weight. The lungs were dried at 50 °C in an oven for 48 h, and weighed to get the dry weight. The lung water content was calculated by dividing the wet lung weight by the dry lung weight.
Modified Miles Assay. U46619 (20 or 200 μ g diluted in methyl acetate) was applied onto the right auricle of BALB/c mice. Evans Blue (50 mg/kg) was immediately administered intravenously. After 30 min, mice were euthanized and ears were excised, dried and weighed. The samples were then incubated with formamide in the 50 °C oven. Evans Blue extracted in formamide was quantified as absorbance at 610 nm, using the spectrophotometer.
Blood flow measurement. U46619 was applied onto the auricle of BALB/c mice. Changes in ear blood flow were monitored for 1 h by using an Omegazone laser Doppler blood-flow imaging system (Omegawave, Inc., Tokyo, Japan). Blood flow was quantified for each experimental mouse after 20 min of stimulation, and expressed as the difference in intensity between the right and left ear.
Transwell Permeability Assay. HUVECs (8 × 10 4 ) were seeded onto the upper inserts of transwell (1 μ m pore size, BD Biosciences, Bedford, MA, USA) and grown to confluent. After 4 h starvation with 2% fetal bovine serum, U46619 (0.1-10 μ M) was administrated into the upper inserts, and then 70,000 Da FITC-dextran (24 μ g per 300 μ l of medium) was added to the upper inserts. FITC-dextran passed through the cell monolayer was collected every 10 min. The amount of FITC-dextran was measured by the fluorescence spectrophotometer using excitation/emission wavelengths of 485/535 nm. Transendothelial Electrical Resistance (TER) Measurement. TER was measured using an xCELLigence Real-time Cell Analyzer DP system (Roche, Indianapolis, IN, USA). HUVECs were seeded onto electrodes and grown until confluent. After the serum starvation, U46619 (0.1-10 μ M) was administrated and TER was measured every 1 min. For the normalization, TER values are expressed as the ratio to the initial value at 30 min before the U46619 administration.