Urokinase-type Plasminogen Activator (uPA) Induces Pulmonary Microvascular Endothelial Permeability through Low Density Lipoprotein Receptor-related Protein (LRP)-dependent Activation of Endothelial Nitric-oxide Synthase

Abstract

Urokinase plasminogen activator (uPA) and PA inhibitor type 1 (PAI-1) are elevated in acute lung injury, which is characterized by a loss of endothelial barrier function and the development of pulmonary edema. Two-chain uPA and uPA-PAI-1 complexes (1–20 nm) increased the permeability of monolayers of human pulmonary microvascular endothelial cells (PMVECs) in vitro and lung permeability in vivo. The effects of uPA-PAI-1 were abrogated by the nitric-oxide synthase (NOS) inhibitor l-NAME (N^D-nitro-l-arginine methyl ester). Two-chain uPA (1–20 nm) and uPA-PAI-1 induced phosphorylation of endothelial NOS-Ser¹¹⁷⁷ in PMVECs, which was followed by generation of NO and the nitrosylation and dissociation of β-catenin from VE-cadherin. uPA-induced phosphorylation of eNOS was decreased by anti-low density lipoprotein receptor-related protein-1 (LRP) antibody and an LRP antagonist, receptor-associated protein (RAP), and when binding to the uPA receptor was blocked by the isolated growth factor-like domain of uPA. uPA-induced phosphorylation of eNOS was also inhibited by the protein kinase A (PKA) inhibitor, myristoylated PKI, but was not dependent on PI3K-Akt signaling. LRP blockade and inhibition of PKA prevented uPA- and uPA-PAI-1-induced permeability of PMVEC monolayers in vitro and uPA-induced lung permeability in vivo. These studies identify a novel pathway involved in regulating PMVEC permeability and suggest the utility of uPA-based approaches that attenuate untoward permeability following acute lung injury while preserving its salutary effects on fibrinolysis and airway remodeling.

Introduction

Lung inflammation is accompanied by increased pulmonary microvascular (PMV)³ permeability leading to the exudation of plasma proteins into alveolar spaces, interstitial edema, impaired gas exchange, and increased morbidity and mortality (1). Considerable efforts have been made to understand how endothelial barrier function is lost early in the inflammatory process (2), but this knowledge has yet to translate into specific means to prevent excess permeability from developing and thereby to improve the clinical outcome. Transudation of plasma proteins into airways promotes the formation of a provisional fibrin matrix that can lead to pulmonary fibrosis. Plasminogen activators (PA), primarily urokinase PA (uPA) and tissue-type PA (tPA), convert plasminogen to plasmin, which in turn lyses fibrin matrices and thereby exerts anti-fibrotic effects in the lungs (3). PAs exert a protective role in several experimental models of acute lung injury (ALI) and pulmonary fibrosis (3–8), and in one study, patients with ALI showed significant improvement in oxygenation after treatment with uPA (9).

However, the salutary effects of uPA on lung repair, largely derived from its catalytic activity, are partially offset by deleterious extrafibrinolytic effects on vascular and airway tone (10–14). Moreover, diverse pulmonary insults involved in ALI decrease PA activity (15–19) and markedly increase the concentration of its inhibitor PAI-1 (20). Furthermore, uPA at concentrations measured in lung tissue during inflammation (∼20 nm) impairs pulmonary artery contractility, increases vascular permeability, and enhances airway hyperresponsiveness (10, 11, 14). Our data indicate that these effects are mediated through distinct portions of the uPA molecule and engage different receptor pathways than those involved in fibrinolysis (10, 14).

uPA is secreted as a 50-kDa single-chain proenzyme (scuPA) composed of three structurally identifiable domains: a growth factor-like domain (GFD; amino acids 1–46), which binds a specific receptor (uPAR/CD87) (21, 22), a kringle domain (amino acids 47–135) (23, 24), and a proteolytic domain (amino acids 159–411), which includes a serine protease catalytic site (25). Plasmin and several other serine proteases cleave soluble or uPAR-bound scuPA to generate an enzymatically active, two-chain molecule (tcuPA) (26) that is inhibited primarily by PAI-1 in plasma and in the lung (27, 28).

Enzymatically inactive uPA-PAI-1 complexes bound to the uPAR are internalized by the low density lipoprotein-related receptor protein/α₂-macroglobulin receptor (LRP) (29, 30). Occupation of LRP can also initiate several intracellular signaling cascades, including the activation of cAMP-dependent protein kinase A (PKA) (31–33). uPA can also activate NMDA type 1 receptor (NMDAR-1) (11, 14), which is expressed by pulmonary vascular smooth muscle cells and has been linked to impaired vascular contractility and permeability (34–37).

A number of mediators have been identified that can alter PMV permeability within minutes. Among these is a vascular endothelial growth factor (VEGF), which acts through activation of endothelial NOS (eNOS) (NOS-3) (38, 39). VEGF-induced permeability of retinal microvascular endothelial cells has been shown to require endogenous uPA (40). eNOS converts l-arginine (l-Arg) to l-citrulline (l-Ctr) to generate NO, which causes vasorelaxation and increases microvascular permeability (38, 39, 41–43). However, the role of NO in the pathogenesis of ALI remains controversial. Low levels of endogenous NO help to prevent pulmonary edema from developing in isolated rabbit lungs following ischemia-reperfusion or vagal stimulation (44–46), whereas excessive NO appears detrimental in that it increases vascular permeability (43, 47).

The cellular and mechanistic basis underlying the loss of pulmonary endothelial barrier function has been studied in detail (reviewed in Refs. 2, 48, and 49), but the role of PAs, although clearly implicated in lung injury and repair (27), has not been studied in detail. Here we report that uPA and uPA-PAI-1 complexes increase PMV endothelial permeability by binding to LRP, which activates cAMP-dependent PKA and eNOS, generates NO, and disrupts intercellular contacts in the lung microvasculature.

EXPERIMENTAL PROCEDURES

Materials

PAI-1 and a monoclonal antibody against the α-chain of human LRP were kind gifts from American Diagnostica Inc. (Stamford, CT). The LRP antagonist receptor-associated protein (RAP) was the kind gift of Dr. T. Willnow (Max-Delbrueck-Center for Molecular Medicine, Berlin). Fc-RAP was cloned and characterized as described (50). The Akt inhibitor 1-l-6-hydroxymethyl-chiro-inositol-2-[(R)-2-O-methyl-3-O-octadecylcarbonate] was from Alexis Biochemicals (San Diego, CA); myristoylated PKI (14–22) amide was from Enzo Life Sciences, Inc. (Plymouth Meeting, PA); human PMV endothelial cells (PMVEC), endothelial basal medium-2 (EBM-2), and cell growth supplements were from Lonza Group Ltd. (Conshohocken, PA); l-[¹⁴C]arginine (l-Arg) was from PerkinElmer Life Sciences; N^G-nitro-l-arginine methyl ester (l-NAME) was from Acros Organics (Morris Plains, NJ); N^G-nitro-l-arginine (l-NNA) and the NOS activity assay kit were from Cayman Chemical Inc. (Ann Arbor, MI); rabbit polyclonal antibodies (pAbs) that selectively recognize p-eNOS-Ser¹¹⁷⁷ and total eNOS were from Cell Signaling Technology (Beverly, CA); the PKA assay kit and mouse α-GAPDH monoclonal Abs (mAbs) were from Millipore (Temecula, CA); anti-vascular endothelial (VE)-cadherin goat pAbs were from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA); HRP-conjugated goat anti-mouse IgG was from Jackson ImmunoResearch Inc. (Bar Harbor, ME); Alexa 647-conjugated phalloidin and the WesternBreeze Western blot (WB) kit were from Invitrogen; mouse anti-β-catenin mAbs were from BD Biosciences; goat anti-mouse Alexa 555-conjugated pAbs were from Molecular Probes (Eugene, OR); protease inhibitor mixture for mammalian cells and tissue extracts, phosphatase inhibitor mixture 2, MK-801, FITC-dextran (40 kDa), anti-S-nitroso-cysteine Abs, LY-294002 hydrochloride, and NADPH were from Sigma-Aldrich; biotin-HPDP and streptavidin-agarose were from Pierce; and VEGF165 was from R&D Systems (Minneapolis, MN).

uPA-PAI-1 Complexes

Recombinant human scuPA and catalytically inactive scuPA-S356A were expressed in S2 cells, purified, and characterized as described previously (14, 24). scuPA was converted to tcuPA by limited plasmin digestion (51). Complexes between tcuPA and PAI-1 were prepared as described (52). Complex formation was verified using SDS-PAGE (51).

Permeability in Vivo

Lung permeability was determined by measuring extravasation of intravenously injected Evans blue dye as described (53) with minor modifications (50). Bronchoalveolar lavage (BAL) was performed using 1.5 ml of warmed (30 °C) sterile Hanks' balanced salt solution. BAL fluid was collected and centrifuged at 14,000 × g for 20 min at 4 °C, the supernatant was removed, and the optical density at 620 nm was measured.

Permeability in Vitro

PMVEC from passages 4 to 12 were plated onto culture inserts (3-μm pore size; Falcon, BD Biosciences; 1.0 × 10⁵ cells per 0.3 ml) within 24-well plates and grown until confluent at 37 °C under 5% CO₂. Permeability was measured as described (54). Briefly, PMVEC were incubated with tcuPA, uPA-PAI-1, or PAI-1, with or without inhibitors (RAP, mPKI, and l-NAME), in EBM-2 medium containing 50 μg/ml FITC-dextran (40 kDa). The permeability of untreated cells to FITC-dextran served as a base-line control. Wells without cells served to measure the spontaneous diffusion of FITC-dextran versus transendothelial transport.

Measurement of eNOS and Akt Phosphorylation by WB

PMVEC were grown to 90% confluence and starved in 1% serum for 24 h. Cells were incubated with scuPA, tcuPA, uPA-S356A, or uPA-PAI-1 for the indicated times alone or after preincubation with inhibitors for 60 min. The cells were lysed in radioimmune precipitation assay buffer (20 mm Tris-HCl, pH 7.5, 150 mm NaCl, 0.1% SDS, 1% Nonidet P-40, 0.2% sodium deoxycholate, 2 mm EDTA, 50 mm NaF, and protease and phosphatase inhibitor mixture), and lysates were analyzed by SDS-PAGE and WB using antibodies for p-eNOS-Ser¹¹⁷⁷, total eNOS, p-Ser⁴⁷³-Akt, total Akt, and α-GAPDH.

Nitric-oxide Synthase Activity

eNOS activity was assayed using a NOS assay kit. Briefly, cells were disrupted by 30 strokes in a Dounce homogenizer in cold homogenization buffer. eNOS activity in supernatants was assessed by the conversion of l-[¹⁴C]Arg to l-[¹⁴C]citrulline. Incubations performed in parallel in the presence of the eNOS inhibitor l-NNA (1 mm) served as the negative control. eNOS activity was expressed as pmol of l-Ctr/mg of protein/min. l-Ctr formation was calculated using the formula pmol l-Ctr = (cpm_a – cpm_p)/cpm_s × 290, where cpm_a and cpm_p are cpm in the absence and presence, respectively, of l-NNA and cpm_s represents cpm in the standard. All standards contained 0.1 μCi of l-[¹⁴C]Arg corresponding to 290 pmol.

cAMP-dependent Protein Kinase A Activity

PKA activity was assayed using a PKA assay kit. Briefly, cells were disrupted in cold extraction buffer (25 mm Tris-HCl, pH 7.4, 0.5 mm EDTA, 0.5 mm EGTA, 10 mm β-mercaptoethanol, and protease and phosphatase inhibitor mixtures) by 30 strokes in a Dounce homogenizer. Lysates were clarified by centrifugation, and kinase activity was measured according to the manufacturer's protocol. Radiolabeled peptides were separated from the residual [γ-³²P]ATP using P81 phosphocellulose paper, and radioactivity was quantified.

Confocal Microscopy

PMVEC were grown on chamber slides until confluent, incubated with uPA-PAI-1 (20:40 nm) for the indicated times, fixed in 4% paraformaldehyde/PBS, and permeabilized in 0.1% Triton X-100 as described (55). Mouse anti-β-catenin mAbs and goat anti-mouse Alexa 555-conjugated pAbs or rabbit anti-p-eNOS Ab and goat anti-rabbit Alexa 488-conjugated pAbs were used to detect β-catenin and p-eNOS. Fibrillar actin was detected using Alexa 647-conjugated phalloidin. Nuclei were counterstained with DAPI (0.5 μg/ml). Cells were analyzed using a Zeiss LSM 510 confocal microscope (Carl Zeiss, Heidelberg, Germany).

Fluorescent Detection of S-Nitrosylation and Immunofluorescence

To visualize S-nitrosylated proteins in PMVEC incubated with uPA-PAI-1 (as described above) by immunofluorescence, a biotin-switch assay was performed as described (56). In conjunction, cells were immunostained to detect β-catenin and fibrillar actin as described above.

Immunoprecipitation

PMVEC were plated, serum-starved, and incubated with uPA-PAI-1 (20:40 nm) for 1 h as above. Cells were lysed as described (56), β-Catenin was immunoprecipitated using mouse anti-β-catenin Abs. Co-immunoprecipitated proteins were detected by WB using anti-β-catenin mAb, goat anti-VE-cadherin pAb, or rabbit anti-S-nitroso-cysteine pAb and the corresponding HRP-conjugated secondary Abs.

Statistical Analysis

Differences between groups were compared using the one-way analysis of variance statistical test. Statistical analyses were performed using the EZAnalyse add-in to Microsoft Excel software. Significance was set at a p value of less than 0.05.

RESULTS

uPA Induces Pulmonary Vascular Permeability in Vivo

In view of findings that uPA^−/− mice are protected against LPS-induced pulmonary edema (57) and that uPA increases endothelial permeability in vitro (40), we asked whether uPA regulates pulmonary vascular permeability in vivo. Intravenous injection of scuPA (0.1–2 mg/kg) in mice increased lung permeability in a dose-dependent manner as measured by the extravasation of intravenously administered Evans blue dye into the BAL fluid (Fig. 1A). To elucidate the mechanism, we investigated how uPA affects endothelial permeability in vitro.

FIGURE 1.

A, uPA increases pulmonary vascular permeability in vivo. C57BL/6J mice were injected intravenously with the indicated amounts of scuPA (0.1–2 mg/kg) or saline control followed 1 h later by injection with Evans blue dye. One hour later, extravasation of dye in to the BAL fluid was measured (mean ± S.D., n = 5–7). B, uPA increases PMVEC permeability in vitro. Confluent monolayers of PMVEC were incubated with tcuPA (20 nm), PAI-1 (40 nm), or preformed uPA-PAI-1 (20:40 nm tcuPA:PAI-1, respectively) for the indicated times at 37 °C. FITC-dextran was added to the upper chamber, and diffusion was quantified by measuring fluorescence units in the lower chamber. The data are expressed as -fold stimulation into over untreated cells taken as 1, with passive leakage of FITC-dextran through a noncellular porous support taken as 100 (not shown). Each data point represents the mean ± S.E. of three independent experiments, each performed in triplicate. *, p < 0.05; and **, p < 0.01 versus untreated cells incubated with EBM-2 medium alone. C, uPA-PAI-1 causes VE-cadherin to dissociate from β-catenin. PMVEC were starved in 1% serum/EBM-2 without growth factors or other supplements for 24 h and then incubated with preformed uPA-PAI-1 complexes (20:40 nm) for 1 h at 37 °C alone or in the presence of l-NAME (200 μm). The cells were lysed, β-catenin was immunoprecipitated with an anti-β-catenin mAbs, and the immunoprecipitates (IP) were analyzed by WB for VE-cadherin and β-catenin. D, uPA-PAI-1 induces redistribution of β-catenin. Immunofluorescent staining showed redistribution of immunoreactive β-catenin in PMVEC incubated with uPA-PAI-1. Widening of β-catenin staining at intercellular contacts was evident within 10 min, and the effect was sustained for 60 min. Representative confocal microscopy images are shown from three independent experiments in which 10 fields/condition were observed. E, inhibition of eNOS reduces PMVEC permeability induced by uPA-PAI-1. PMVEC were incubated with uPA-PAI-1 (20:40 nm) in the absence (control (cont), filled bars) or presence of l-NAME (200 μm) for 30 min at 37 °C. PMVEC incubated with EBM-2 medium alone served as the negative control (open bars). Permeability to FITC-dextran was measured over a period of 30 min. Results are expressed as -fold increase in permeability measured in the absence of uPA-PAI-1 and l-NAME. The mean ± S.E. of three independent experiments, each performed in triplicate, is shown. *, p < 0.01.

uPA and uPA-PAI-1 Complexes Induce Permeability of PMVEC Monolayers

We examined the effect of uPA on the permeability of PMVEC monolayers using FITC-dextran. The addition of enzymatically active tcuPA (20 nm) induced a more than 2-fold increase in transendothelial permeability, and the monolayer maintained increased permeability for more than 2 h (Fig. 1B). scuPA increased permeability to ∼30% of that seen in the complete absence of cells and was comparable in magnitude with the effect of VEGF (5 nm) (supplemental Fig. S1A). Both scuPA and tcuPA increased endothelial permeability in vitro in a dose-dependent manner beginning at concentrations as low as 1 nm (supplemental Fig. S1B).

PAI-1 is dramatically elevated in the lungs during ALI to levels well above uPA (5, 16, 19). PAI-1 and uPA are also synthesized and secreted by PMVECs (supplemental Fig. S2). We hypothesized that exogenous tcuPA (or scuPA converted on the cell surface to tcuPA) binds to cell-associated PAI-1, forming stable enzymatically inactive complexes that promote endothelial permeability. To test this hypothesis, we compared the effects of tcuPA (20 nm), PAI-1 (40 nm), and preformed uPA-PAI-1 complexes on the permeability of PMVEC monolayers. PAI-1 was added in molar excess to ensure that all available uPA was complexed, as some recombinant PAI-1 might have undergone conversion to the latent form before exposure to tcuPA. tcuPA (20 nm) and PAI-1 (40 nm) added alone significantly (p < 0.05) increased the permeability of PMVEC to a similar extent (Fig. 1B), suggesting engagement of each with their respective endogenous counterparts. In support of this conclusion, preformed uPA-PAI-1 complex significantly (p < 0.01) increased the permeability of PMVEC monolayers to a greater extent than the same concentration of tcuPA or PAI-1 added individually (Fig. 1B). Therefore, the finding that PAI-1 increased the potency of uPA to induce permeability might be of significance in the setting of lung disease.

uPA-PAI-1 Induces Redistribution of β-Catenin

Endothelial permeability and intercellular contacts are regulated by β-catenin (58). Dissociation of β-catenin from VE-cadherin and redistribution from adherens junctions to the cytoplasm in response to stimuli such as VEGF increases endothelial permeability (40, 56, 58). Likewise, the addition of uPA-PAI-1 caused β-catenin to dissociate from VE-cadherin as evidenced by loss of co-immunoprecipitation (Fig. 1C). This was accompanied by an expansion of the intercellular contact zones that stained for β-catenin within 10 min of exposure to uPA-PAI-1, an effect that was sustained for at least 60 min (Fig. 1D).

uPA-PAI-1-induced PMVEC Permeability Is Mediated through eNOS

eNOS-dependent nitrosylation of β-catenin leads to its redistribution and increased endothelial permeability (56). eNOS is partially localized to intercellular contacts/tight junctions (59). Therefore, we asked whether inhibition of eNOS activity would prevent uPA-PAI-1-induced redistribution of β-catenin. The eNOS inhibitor l-NAME (0.1 mm) attenuated uPA-PAI-1-induced dissociation of β-catenin from VE-cadherin as evidenced by co-immunoprecipitation (Fig. 1C) and by more restricted β-catenin staining at intercellular contacts (Fig. 1D, l-NAME panels). l-NAME also reduced PMVEC permeability evoked by uPA-PAI-1 in parallel (Fig. 1E).

uPA and uPA-PAI-1 Induce eNOS Phosphorylation

Based on this result, we next investigated whether uPA and uPA-PAI-1 augment the activation of eNOS through phosphorylation at the “activation site” on Ser¹¹⁷⁷. scuPA, tcuPA, and uPA-PAI-1 each induced phosphorylation of eNOS-Ser¹¹⁷⁷ in a dose-dependent manner as determined by WB (Fig. 2 and supplemental Fig. S3A) beginning at concentrations as low as 1 nm. PAI-1 itself induced eNOS-Ser¹¹⁷⁷ phosphorylation in a concentration-dependent manner (1–40 nm) as well (data not shown). At each concentration tested, a hierarchy of potency was seen (tcuPA-PAI-1 > tcuPA > scuPA) (Fig. 2 and supplemental Fig. S3A). In contrast, catalytically inactive scuPA-S356A induced NOS-Ser¹¹⁷⁷ phosphorylation only at the highest concentration studied (20 nm) (supplemental Fig. S3A, right panel), consistent with the reported K_d of scuPA for LRP (60) and the effect of scuPA-S356A on PMVEC permeability (supplemental Fig. S1B).

FIGURE 2.

tcuPA and uPA-PAI-1 induce eNOS phosphorylation in PMVEC. PMVEC were incubated with the indicated concentrations of tcuPA or uPA-PAI-1 for 30 min at 37 °C. Cells incubated with EBM-2 medium alone served as the control (cont) and were studied in parallel for each agonist. Cell lysates were prepared, and phosphorylation of eNOS-Ser¹¹⁷⁷ was analyzed by SDS-PAGE and WB using antibodies for p-eNOS-Ser¹¹⁷⁷ or total eNOS. Anti-GAPDH was used to ensure equal protein loading. Representative blots are shown in the upper panel. The lower panel shows the results of pooled data from densitometric analyses. Results are expressed as -fold increase over control. The mean ± S.E. of three independent experiments, each performed in triplicate, is shown. *, p < 0.05; #, p < 0.05; and ##, p < 0.01 versus control.

Phosphorylation of eNOS-Ser¹¹⁷⁷ in response to 2 nm tcuPA was seen at 15 min, peaked at 30 min, and decreased by 60 min (Fig. 3A), whereas the more profound effect of uPA-PAI-1 was evident at 15 min and persisted for at least 60 min, the latest data point studied (supplemental Fig. S3B). In contrast, the phosphorylation induced by 2 nm scuPA was first seen at 30 min, and the signal declined by 60 min (supplemental Fig. S3B). These data suggest that preformed uPA-PAI-1 complexes more rapidly initiate and sustain the signaling cascade, whereas tcuPA must first bind to endogenous PAI-1, and scuPA must be converted to tcuPA on the EC surface (51) prior to binding to PAI-1, which delays eNOS phosphorylation.

FIGURE 3.

A, time course of eNOS phosphorylation in PMVEC in response to tcuPA. PMVEC were incubated with tcuPA (2 nm) for the indicated times. Phosphorylation of eNOS-Ser¹¹⁷⁷ was determined by WB followed by densitometry as described in the legend for Fig. 2. Results are expressed as -fold increase over a control sample incubated in EBM-2 medium alone. *, p < 0.05 versus control (cont), n = 3. B, time course of eNOS activation in PMVEC in response to tcuPA. PMVEC were incubated with tcuPA (2 nm) for the indicated times. eNOS activity in cell lysates was assessed by the conversion of l-Arg to l-Ctr and calculated as described under “Experimental Procedures.” Results are expressed as the rate of conversion of l-Arg to l-Ctr normalized per mg of total cellular protein. The mean ± S.E. of three independent experiments, each performed in triplicate, is shown. *, p < 0.01 versus control.

To ensure that eNOS-Ser¹¹⁷⁷ phosphorylation in response to tcuPA activates the enzyme, we examined whether NO is generated (61). The phosphorylation of eNOS (Fig. 3A) and the time dependence of conversion of l-Arg to l-Ctr (Fig. 3B), indicating the generation of NO, followed a similar time course (42). The generation of NO was abolished by pretreating the cells with the NOS inhibitor l-NNA (1 mm; data not shown), affirming the specificity of the reaction.

uPA-PAI-1 Induces Nitrosylation of β-Catenin and Redistribution of Fibrillar Actin in PMVEC

NO generated by eNOS in response to VEGF promotes S-nitrosylation of β-catenin (56). Therefore, we investigated whether activated p-eNOS-Ser¹¹⁷⁷ co-localizes with β-catenin after exposure of PMVEC to uPA-PAI-1. uPA-PAI-1 increased the appearance of p-eNOS-Ser¹¹⁷⁷-positive staining in PMVEC monolayers, which partially co-localized with β-catenin at intercellular contacts (Fig. 4A). To elucidate whether uPA-PAI-1-induced NO generation causes S-nitrosylation of β-catenin and other proteins in PMVEC, we employed a biotin-switch technique (56) to visualize intracellular S-nitrosothiols. The addition of uPA-PAI-1 increased total cellular levels of S-nitrosothiol and its appearance at intercellular contacts where it co-localized with β-catenin (Fig. 4B, leftmost panel). Using immunoprecipitation of β-catenin and WB detection of S-nitrosothiol, we found that uPA-PAI-1-induced S-nitrosylation of β-catenin in PMVEC was inhibited by l-NAME, indicating the involvement of NO (Fig. 4C). This was confirmed by using a biotin-switch technique and fluorescent detection of S-nitrosothiols (supplemental Fig. S4). uPA-PAI-1 also caused a marked coincident redistribution of actin filaments toward intercellular contacts (Fig. 4B, F-actin panels) that co-localized with β-catenin (Fig. 4B, rightmost panels), a conformation reported to contribute to EC retraction and increased permeability (62).

FIGURE 4.

A, uPA-PAI-1 treatment reveals co-localization of β-catenin and p-eNOS-Ser¹¹⁷⁷ at intercellular contacts in PMVEC. Cells were incubated with uPA-PAI-1 (20:40 nm) for 10 min, fixed, and stained for β-catenin (red) and p-eNOS-Ser¹¹⁷⁷ (green) as described under “Experimental Procedures.” Nuclei were counterstained with DAPI. Arrowheads indicate p-eNOS-Ser¹¹⁷⁷-positive staining co-localized with β-catenin at intercellular contacts. B, uPA-PAI-1 induces S-nitrosylation at intercellular contacts in PMVEC. PMVEC were incubated with uPA-PAI-1 (20:40 nm) for the indicated times. Fixed cells were subjected to a biotin-switch assay, as described under supplemental “Methods.” S-Nitrosylation was visualized using Alexa 488-conjugated avidin (green). β-Catenin immunostaining was performed in parallel (red). Fibrillar actin was detected using Alexa 647-conjugated phalloidin, which is pseudocolored in cyan. Nonspecific staining was determined in control cells incubated with Alexa 488-streptavidin. Arrowheads indicate the presence of S-nitrosylation at intercellular junctions (merge with β-catenin) on the left. C, uPA-PAI-1 induces S-nitrosylation of β-catenin. PMVEC were starved and incubated with uPA-PAI-1 (20:40 nm) for 60 min or preincubated for 30 min with l-NAME (200 μm) prior to uPA-PAI-1 treatment. Cells were lysed, and β-catenin was immunoprecipitated using mouse anti-β-catenin Abs as described under “Experimental Procedures.” S-Nitrosylated β-catenin and total β-catenin were detected by WB using rabbit anti-S-nitroso-cysteine mAbs and anti-β-catenin mAbs, respectively, and the corresponding HRP-conjugated secondary Abs.

Role of PKA and PI3K-Akt Pathways in Activation of eNOS

Phosphorylation of eNOS-Ser¹¹⁷⁷ can be mediated by PKA, phosphatidylinositol-3 kinase (PI3K)-dependent protein kinase B (PKB/Akt), or AMP-dependent kinase (AMPK) (63, 64). PKA regulates permeability through phosphorylation of cytoskeletal proteins (36, 37), and PKB/Akt is implicated in vascular leakage during acute inflammation (65). Therefore, we asked whether PKA or PI3K-dependent PKB/Akt was responsible for uPA-induced eNOS activation by using selective inhibitors of each pathway. As exposure to uPA or uPA-PAI-1 has not been shown to cause metabolic stress in ECs, we did not study AMP-dependent kinase. PMVEC were preincubated with the PI3K inhibitor LY294002, the PKB/Akt inhibitor AktI, or the PKA inhibitor mPKI for 1 h prior to adding tcuPA or uPA-PAI-1 for 30 min. uPA induced phosphorylation of eNOS-Ser¹¹⁷⁷ at all concentrations tested (1–20 nm). Phosphorylation was completely inhibited by the PKA inhibitor mPKI (Fig. 5A). In contrast, LY294002 and AktI had no effect, suggesting that the PI3K/Akt pathway is not essential for uPA-induced phosphorylation of eNOS. The same pattern of inhibition was seen when uPA-PAI-1 was studied (data not shown). Although uPA-induced phosphorylation of PKB/Akt at Ser⁴⁷³ was inhibited in the presence of LY294002, this seemed to be unrelated to uPA-induced eNOS-Ser¹¹⁷⁷ phosphorylation because neither LY294002 nor AktI inhibited uPA-induced permeability (data not shown).

FIGURE 5.

A, effect of kinase inhibitors on uPA-induced phosphorylation of eNOS. PMVEC were preincubated for 60 min with mPKI (1 μm) to inhibit PKA, LY294002 (3 μm) to inhibit PI3K, or 1L-6-hydroxymethyl-chiro-inositol-2-[(R)-2-O-methyl-3-O-octadecylcarbonate] (AktI) (3 μm) to inhibit Akt and then stimulated with tcuPA (2 nm) for 30 min. Phosphorylation of eNOS-Ser¹¹⁷⁷ was determined using WB analysis as in the legend for Fig. 2. Results are expressed as -fold increase over control (cont) incubated with EBM-2 medium alone. Each data point represents the mean ± S.E. of three independent experiments, each performed in triplicate, **, p < 0.01. B, involvement of uPA receptors in uPA-induced eNOS phosphorylation. PMVEC were preincubated for 60 min with an anti-LRP monoclonal antibody (10 μg/ml), GFD (50 nm), RAP (80 nm), or MK-801 (10 μm) and then stimulated with tcuPA (2 nm) for 30 min. Phosphorylation of eNOS-Ser¹¹⁷⁷ was determined using WB analysis as described in the legend for Fig. 2. Results are expressed as -fold increase over control cells incubated with EBM-2 medium alone. Each data point represents the mean ± S.E. of three independent experiments, each performed in triplicate. *, p < 0.05; and **, p < 0.01 versus control or cells incubated with tcuPA only.

LRP Mediates uPA-PAI-1-induced Activation of eNOS

uPA and uPA-PAI-1 can bind to uPAR and LRP simultaneously (24, 30). However, the affinity of uPA-PAI-1 for LRP is almost 100-fold higher than the affinity of either its constituents (60). tcuPA has also been shown to alter cell signaling through the activation of NMDAR-1 in lung airways (14). To elucidate which receptors contribute to uPA and uPA-PAI-1-induced eNOS phosphorylation, we used the following receptor-specific antagonists: isolated GFD, which competes with uPA for uPAR binding (66); RAP (67) and anti-LRP mAb, which block LRP; and an antagonist of NMDAR-1, MK-801 (68). Phosphorylation of eNOS induced by tcuPA was blocked by the anti-LRP Ab and RAP but not by MK-801 as determined by WB analysis (Fig. 5B). GFD alone (2–50 nm) did not induce any response. However, 50 nm GFD blocked the effect of 2 nm tcuPA, but inhibition was overridden at 20 nm tcuPA (supplemental Fig. S5). The same effect was seen when uPA-PAI-1 was studied (data not shown). These data suggest that the binding of uPA or uPA-PAI-1 to LRP is sufficient to induce PKA-mediated eNOS activation. However, high affinity binding to uPAR and the presence of PAI-1 lowers the concentration required for LRP-mediated engagement and activation.

LRP and PKA Mediate uPA-PAI-1-induced Permeability of PMVEC

The binding of uPA to LRP initiates activation of PKA in some cell types (31–33, 69), and phosphorylation of eNOS is inhibited by mPKI (Fig. 5A). Therefore, we next asked whether tcuPA and/or uPA-PAI-1 induce activation of PKA. The addition of tcuPA or uPA-PAI-1 led to the activation of PKA in a time-dependent manner as detected by ³²P phosphorylation of a PKA-specific peptide (Fig. 6A). The activation of PKA by tcuPA (20 nm) was inhibited by RAP (Fig. 6B). RAP and mPKI also significantly suppressed uPA-PAI-1-induced PMVEC permeability, suggesting that binding of the complexes to LRP precedes the activation of PKA (Fig. 7A).

FIGURE 6.

tcuPA and uPA-PAI-1 induce activation of PKA in PMVEC. Cells were incubated with tcuPA (20 nm) or uPA-PAI-1 (20:40 nm) for the indicated times (A) or with tcuPA (20 nm) in the absence or presence of RAP (80 nm) for 30 min (B) at 37 °C. Cell lysates were analyzed for PKA activity as described under “Experimental Procedures.” Results are expressed as -fold increase over a control sample incubated in EBM-2 medium alone studied in parallel for each agonist. The mean ± S.E. of three independent experiments, each performed in triplicate, is shown. *, p < 0.05; **, < 0.01; and #, p < 0.05 versus cells incubated with EBM-2 medium alone.

FIGURE 7.

A, LRP and PKA inhibitors block uPA-PAI-1-induced PMVEC permeability in vitro. PMVEC were incubated with uPA-PAI-1 (20:40 nm) in the absence (control (cont), filled bars) or presence of RAP (80 nm) or mPKI (1 μm) for 30 min at 37 °C. Permeability to FITC-dextran was measured as described in the legend to Fig. 1B. Results are expressed as -fold increase over permeability in control cells incubated with EBM-2 medium alone (control, open bars). The mean ± S.E. of three independent experiments, each performed in triplicate, is shown. *, p < 0.01 versus control; and #, p < 0.01 versus uPA-PAI-1-treated group. B, inhibition of uPA-induced pulmonary vascular permeability in vivo. Lung permeability, assessed by extravasation of intravenously injected Evans blue dye into the BAL fluid was measured after intravenous injection of saline (control, open bars) or uPA (1 mg/kg) in the absence (control, filled bars) or presence of mPKI (6 mg/kg), RAP (1 mg/kg), or l-NAME (5 mg/kg). Shown is the mean ± S.D., n = 5–7; *, p < 0.05 versus uPA-treated group (control, filled bars).

Mechanism of uPA-induced Pulmonary Vascular Permeability in Vivo

To assess the relevance of our in vitro findings that uPA- and uPA-PAI-1-induced lung microvascular permeability involves LRP-, PKA-, and eNOS-dependent mechanisms, we examined the effect of specific inhibitors of each pathway on uPA-induced lung permeability in vivo. Mice were given an intravenous injection of mPKI, Fc-RAP, or l-NAME. One hour later, scuPA (1 mg/kg) and Evans blue dye were injected intravenously. Permeability was accessed 1 h later by measuring the transudation of the dye from its circulation into the BAL fluid. Each of the three inhibitors significantly (p < 0.05) protected the endothelial barrier function in uPA-treated mice (Fig. 7B) consistent with our in vitro findings.

DISCUSSION

The loss of PMV endothelial barrier function as a result of lung injury leads to acute transudation of fluid and plasma proteins into the alveolar space, impairing oxygenation. uPA and PAI-1 are elevated in several lung disorders associated with increased vascular permeability, including ALI, acute respiratory distress syndrome, infectious pneumonia, pulmonary fibrosis, and asthma (3–8, 15–19, 70). uPA plays a salutary role in some forms of ALI through its ability to lyse fibrin deposits in a timely manner and thereby helps prevent fibrosis (8, 19). However, uPA also inhibits pulmonary arterial contraction, which impairs the physiological shunting of blood from underventilated to well ventilated airways (14) and may increase pulmonary vascular permeability (10, 11, 14, 40, 71). These observations led us to investigate the effect of uPA on the PMV endothelium.

VEGF induces vascular permeability in cultured bovine retinal EC through a process that requires uPA and uPAR, and exogenously added uPA increases the permeability of retinal microvascular endothelial monolayers in vitro (40, 56). However, little is known about the mechanisms underlying the effect of uPA on PMV permeability or the involvement of PAI-1.

We focused on PMVEC because the loss of their contact integrity compromises lung function. Our data indicate that uPA-PAI-1 increases the permeability of PMVEC in culture and lung permeability in vivo. In pathological settings where uPA and PAI-1 are elevated (15, 16, 17, 19, 72) disruption of the endothelial barrier by uPA-PAI-1 complexes may permit uPA and other proteins ready access to the underlying vascular smooth muscle cells, which in turn may exacerbate hypoxemia by impairing pulmonary vascular contractility.⁴

The induction of pulmonary vascular permeability requires the binding of uPA or uPA-PAI-1 to LRP. There is a precedent for the involvement of LRP in regulating tissue permeability (73), e.g. permeabilization of the blood-brain barrier induced by tissue-type PA (74), although the mechanism remains uncertain. Some investigations indicate that LRP signal transduction is dependent on the enzymatic activity of the PA, which cleaves the extracellular domain of the receptor (75), whereas other studies implicate uPA-PAI-1-mediated association of domain 3 of uPAR with LRP (76). Whether direct binding of agonists to LRP induces intracellular signaling or whether binding to receptors other than uPAR may be involved in LRP signaling is unknown but both are suggested by our findings.

Several lines of evidence indicate that uPA induces LRP-mediated PMVEC permeability through a non-catalytic mechanism. First, exogenous tcuPA and PAI-1 had comparable effects on permeability, which we posit results from binding to its cellular counterpart, leading to the formation of enzymatically inactive complexes. Second, preformed uPA-PAI-1 complexes, which have much greater affinity for LRP than single- or two-chain uPA (60), mediate LRP-dependent intracellular signaling and induce permeability at lower concentrations and earlier times, and the effect is more sustained than seen with either tcuPA or PAI-1 alone. Our data are also in line with previous findings that PAI-1 sustains a uPA signaling response by engaging an LDL receptor homologue, the very low density lipoprotein receptor VLDLR (77). Third, displacement of tcuPA from uPAR by its GFD fragment raised the concentration required to mediate permeability, likely by impairing the formation of uPA-PAI-1 complexes on the cell surface. Lastly, a catalytically inactive uPA variant induced permeability but only at 20-fold higher concentrations than required for uPA-PAI-1 complexes to do so.

The binding of uPA and uPA-PAI-1 to LRP activates cAMP-dependent PKA, which is required to induce PMVEC permeability. Increased cAMP levels have been reported to stabilize or destabilize endothelial barrier function depending on the origin of the ECs (78). The situation in pulmonary endothelium is complex and might involve competing mechanisms that depend upon the subcellular localization of cAMP due to selective anchoring of PKA and that differ in their effect on permeability (36, 37, 79–81). cAMP, which is generated by plasma membrane-associated adenylate cyclase, activates membrane-associated PKA. This exerts a protective effect on barrier function by phosphorylating/activating filamin and Rac1, thereby stabilizing the cortical actin rim (82). Membrane-associated cAMP-specific phosphodiesterase 4D4 (PDE4D4) hydrolyzes cAMP, which limits its diffusion into the cytosol (83), potentially providing a negative feedback mechanism. In addition, membrane-associated cAMP protects endothelial barrier function by activating the exchange protein directly activated by cAMP (EPAC), which forms a ternary complex with PKA and PDE4D4 and subsequently stabilizes VE-cadherin-based adherens junctions (84). However, if PDE4D4 is inhibited, cAMP diffuses into the cytosol and activates cytosolyic PKA, which initiates a reorganization of the microtubules through phosphorylation of tau-Ser²¹⁴, generating intercellular gaps that increase endothelial permeability (85). Additional studies will be needed to determine whether these or alternative mechanisms are responsible for the uPA-PAI-1-induced PKA-dependent increase in PMV permeability that we observed.

Activation of PKA in response to uPA and uPA-PAI-1 leads to phosphorylation of eNOS-Ser¹¹⁷⁷, activation of the enzyme, and generation of NO, which is known to increase vascular permeability (38, 41). The NOS inhibitor l-NAME reduced PMVEC permeability induced by uPA and uPA-PA-1. This is consistent with the known effects of l-NAME in preventing hyperpermeability during the late phase of allergic responses, smoke-induced lung injury, and endotoxin-induced pulmonary edema; inhalation of l-NAME may be beneficial in the treatment of ALI (46, 47, 86–89).

Recent studies indicate that NO nitrosylates β-catenin, which leads to its dissociation from VE-cadherin and redistribution from tight junctions (56). Destabilization of these junctions contributes to the increased permeability of ECs in response to VEGF (40, 56). Our studies show that uPA and uPA-PAI-1 have a similar effect, i.e. rapidly inducing NO, which leads to nitrosylation of β-catenin concomitant with its dissociation from VE-cadherin. eNOS also has been reported to associate with α-actin (90, 91), which can undergo S-nitrosylation by NO (92) leading to its dissociation from the integrins, which might result in the redistribution of the actin filaments and the impairment of the endothelial barrier function in response to uPA-PAI-1 that we observed. Whether uPA or uPA-PAI-1 promotes nitrosylation of α-actin and how these activities might modulate adherens junctions and alter PMVEC permeability will require additional study.

In summary, the studies presented here show that uPA-PAI-1 at the levels found in ALI induce PMVEC permeability in vitro and increase lung permeability in vivo through uPAR/LRP-mediated activation of PKA, which is followed by activation of eNOS and NO-dependent disruption of adherens junctions. These findings thereby identify several potential sites at which loss of endothelial barrier function in ALI can be interrupted. Understanding the mechanisms that regulate PMV permeability may assist in controlling host defense and repair in diverse common respiratory disorders.