Endothelial cell (EC) junctions determine vascular barrier properties and are subject to transient opening to allow liquid flux from blood to tissue. Although EC junctions open in the presence of permeability-enhancing factors, including oxidants, the mechanisms by which they reseal remain inadequately understood. To model opening and resealing of EC junctions in the presence of an oxidant, we quantified changes in H2O2-induced transendothelial resistance (TER) in monolayers of rat lung microvascular EC. During a 30-min exposure, H2O2 (100 μM) decreased TER for an initial ∼10 min, indicating junctional opening. Subsequently, despite continuous presence of H2O2, TER recovered to baseline, indicating the activation of junctional resealing mechanisms. These bimodal TER transients matched the time course of loss and then gain of E-cadherin at EC junctions. The timing of the TER decrease matched the onset of focal adhesion formation, while F-actin increase at the cell periphery occurred with a time course that complemented the recovery of peripheral E-cadherin. In monolayers expressing a focal adhesion kinase (FAK) mutant (del-FAK) that inhibits FAK activity, the initial H2O2-induced junctional opening was present, although the subsequent junctional recovery was blocked. Expression of transfected E-cadherin was evident at the cell periphery of wild-type but not del-FAK-expressing EC. E-cadherin overexpression in del-FAK-expressing EC failed to effect major rescue of the junctional resealing response. These findings indicate that in oxidant-induced EC junction opening, FAK plays a critical role in remodeling the adherens junction to reseal the barrier.
- endothelial cells
- barrier regulation
- transendothelial resistance
endothelial cell (EC) junctions form the microvascular barrier to transvascular cell and fluid transport. The junctions are formed by groups of intercellular proteins, namely the tight junction proteins that include the claudins, zona occludens 1, JAM 1 and occludin, and the adherens junction proteins that include the VE- and E-cadherins (21, 34). Interactions between these proteins and cortical actin filaments underlie actomyosin-based EC contraction that causes junctional opening consequent to the activation of the myosin light chain kinase (MLCK) (9, 18).
Although it is well known that under pathological conditions, opening of EC junctions causes microvascular hyperpermeabilty, tissue edema, and loss of organ function (4, 31), it is increasingly understood that EC are also capable of enhancing barrier properties. This is evident in both cultured and in situ EC exposed to hyperosmolarity (25, 26), to the platelet-derived phospholipid, sphingosine-1-phosphate (30), or to the cAMP analog, 8CPT-2′Ome-cAMP, which activates the guanine-exchange factor, Epac, for the GTPase, Rap (5, 36). Barrier-enhanced EC characteristically display increased actin filament formation in the cell cortex (17, 25, 26) and increased junctional content of E-cadherin (25, 26, 36). However, it is not clear whether this barrier-protective phenotype reflects agonist effects or the phenotype is a defensive EC response against barrier deterioration.
In this regard, endothelial focal complexes and focal adhesions require consideration. Focal complex proteins stabilize nonmigrating EC at sites of cell-matrix contact (16). Translation of the cell membrane on the matrix, for example during cell stretch (2, 33), enlarges focal complexes to form focal adhesions that strengthen cell-matrix stability (24). The focal complex to focal adhesion transition entails activation of kinases, such as the focal adhesion kinase (FAK) that facilitates recruitment of proteins such as paxillin, vinculin, and zyxin to focal adhesions (23, 40). Recently, FAK has been implicated in positive regulation of the EC barrier (19, 25), suggesting that enlargement of focal cell-matrix contacts might impact remodeling of EC junctions.
The studies we report here arose from our unexpected finding that although continuous exposure to a low concentration of H2O2 decreased the EC barrier as expected (7, 15), the decrease was followed by a spontaneous barrier recovery to baseline. The barrier transient, recorded as changes in the transcellular electrical resistance of EC monolayers, provided an opportunity to investigate EC signaling mechanisms that counteract an induced barrier-deteriorating effect. Our findings indicate that in the presence of a permeability-inducing effect, EC reestablish the barrier by FAK-dependent mechanisms.
MATERIALS AND METHODS
Reagents and chemicals.
Chemicals were obtained from Sigma (St. Louis, MO), unless otherwise stated. Cell culture media, M199 medium, Lipofectamine, Geneticin (G418), and Opti-MEM were obtained from Invitrogen (Rockville, MD). All reagents for immunofluorescence studies were obtained from Molecular Probes (Eugene, OR). Anti-phosphotyrosine MAb PY99 (mouse, monoclonal) and protein A/G-agarose beads were obtained from Santa Cruz Biotechnology (Santa Cruz, CA). Anti-paxillin MAb was purchased from Zymed Laboratories (South San Francisco, CA). Anti-E-cadherin MAb was purchased from Transduction Laboratories (Lexington, KY). Anti-vinculin mouse MAb and α-tubulin MAb were obtained from Sigma.
Rat lung microvascular EC (RLMEC) were cultured as described (25), under 5% CO2 in M199 medium supplemented with 5% fetal bovine serum and 5% bovine calf serum (Hyclone, Logan, UT). Cells were plated at a density of 1 × 105 cells/cm2. EC phenotype was confirmed by cell uptake of fluorescein-tagged acetylated low-density lipoprotein (DiI-Ac-LDL) in imaged monolayers.
Transendothelial electrical resistance.
For EC barrier quantification, we determined transendothelial electrical resistance (TER) in RLMEC monolayers grown on sterile polycarbonate inserts held at 37°C (Endohm; World Precision Instruments, Sarasota, FL). After a 30-min baseline period, experimental solutions were added, and TER data were acquired every 5 s (MP100A-CE data acquisition system, World Precision Instruments). Reported TER data are subtracted for insert resistance.
Detection of H2O2.
To detect H2O2 in medium, we grew EC monolayers on coverslips. We loaded the coverslip monolayers with the H2O2-detecting dye DCFH-DA (dichlorofluorescein diacetate, 10 μM for 30 min). Then, we added aliquots of the H2O2-containing medium to coverslip monolayers. Presence of H2O2 was indicated by increase of monolayer fluorescence as detected by fluorescence microscopy (AX-70, Olympus America).
As described previously (25), the FAK mutant (del-FAK) plasmid was generated by deleting sequences between the EaeI sites at 1,176 and 2,793 bp (amino acids 392–931) in the FAK gene. This deletion includes a segment containing tyrosine residues that are critical for FAK activation. These residues include Y-397, at which FAK autophosphorylates (27), Y-576 and Y-577, at which phosphorylation determines the kinase activity of FAK (3), and Y-925, at which phosphorylation leads to activation of Src (28, 29). Plasmid was transfected in RLMEC using nominal procedures (Lipofectamine) and confirmed by RT-PCR. Geneticin (G418, 500 μg/ml; Calbiochem) resistant cells were selectively grown to confluence, ensuring that monolayers consisted only of transfected cells.
The E-cadherin-GFP (green fluorescent protein) plasmid was obtained as a gift from Dr. I. Koyama of Japan Science and Technology Corp., Nagoya, Japan (13). The GFP expression vector pQBI25 (Quantum, Montreal, Canada) was fused to the COOH-terminal of a DNA fragment-encoding full-length E-cadherin, yielding the E-cadherin-GFP expression vector. Plasmid was transfected in RLMEC using LipofectAMINE Plus (Life Technologies) according to the manufacturer's instructions. Cells stably expressing E-cadherin-GFP were selected using Geneticin and positive clones were selectively grown. Presence of E-cadherin-GFP was confirmed by immunoprecipitation and immunofluorescence. In experimental monolayers, 90% of cells expressed E-cadherin-GFP (data not shown).
Detection of cell surface E-cadherin.
RLMEC monolayers grown on 100-mm cell culture plates were placed on ice and exposed to biotin for 1 h (1.5 mg/ml, sulfo-NHS-biotin; Pierce, Rockford, IL), followed by a wash with blocking reagent (50 mM NH4Cl in PBS containing 1 mM MgCl2 and 0.1 mM CaCl2) to remove free biotin. After several further washes in PBS, cells were scraped off and lysed in buffer (300 μl, 25 mM Tris·HCl, pH 7.4, with 150 mM NaCl, 0.1% SDS, 1% Triton X-100, and 1% deoxycholate) containing protease inhibitors (Roche Molecular Biochemicals, Mannheim, Germany). Cell extracts were centrifuged, and the supernatants were incubated with streptavidin beads (Sigma) to bind biotinylated proteins, which were then analyzed by SDS-PAGE and immunoblotting to identify E-cadherin. Staining of transfer membranes with 0.1% Coomassie brilliant blue ensured even protein transfer and protein loading. Proteins detected by immunoblotting were visualized with chemiluminescence and analyzed by densitometry (Image Station 4000MM; Kodak Scientific Imaging Systems, Rochester, NY).
Immunoprecipitation and immunoblotting.
Immunoprecipitation and immunoblotting were performed as described previously (25). Briefly, confluent monolayers of RLMEC cells were exposed to H2O2 at 37°C under 5% CO2 in M199 for indicated periods. Cells were solubilized in cold buffer, containing protease inhibitors, on ice. Soluble cell extracts were obtained by centrifugation, and equal amounts of protein were incubated with E-cadherin MAb (4 μg, 2 h; Transduction Labs) and then washed with protein A Sepharose beads (Santa Cruz Biotechnology) overnight at 4°C. Precipitates were recovered by centrifugation. Nonspecific proteins were removed by washing the agarose beads three times with cell lysate buffer and once with PBS. Bound proteins were eluted in 40 μl of 4× Laemmli loading buffer. The proteins were resolved by SDS-polyacrylamide gel electrophoresis and blotted onto nitrocellulose membrane and analyzed by immunoblotting.
To detect actin, RLMEC monolayers grown on glass coverslips were fixed (4% formaldehyde in PBS, pH 7.4, 20 min, 22°C), rinsed (3× PBS), permeabilized (0.1% Triton X-100), and stained using rhodamine-phalloidin. For other immunofluorescence studies, cells were incubated with blocking solution (1 h, 25°C) and then with primary antibodies (1:50) in blocking solution (1 h, 22°C). After washing (3× PBS), fluorescence-conjugated antibodies were added (1:500, 1 h, 25°C). After washing with PBS (3×), the glass coverslips were mounted upside-down on object slides using fluorescent mounting medium (Dako, Carpinteria, CA).
Confocal microscopy and image analysis.
Confocal images were obtained by laser scanning microscopy (LSM 510; Zeiss, Thornwood, NY). Fluorophores were excited using He-Ne (545 nm) and argon (492 nm) lasers. Image acquisition and analyses were carried out using software provided with the confocal microscope as well as with standard image analysis software (Metamorph, Universal Imaging).
For quantification of actin content at the cell periphery, we determined fluorescence in a zone that included the cell periphery and the cytoplasm but excluded the perinuclear actin band. To determine E-cadherin distribution, we quantified fluorescence in a zone that included only the cell junction using the line feature of the image analysis software to trace the cell junction along its contours. For focal contact analysis in single cells viewed at high magnification, we applied the threshold and area quantification features of the image analysis software to determine size and number of focal contacts.
Fluorescence data obtained reflects analyses on 40 cells per experiment from three experiments. Group data are presented as means ± SE. Differences between groups were tested by paired t-test for two groups and by the Newman-Keuls test for more than two groups. Significance was accepted at P < 0.05.
Wild-type (WT) RLMEC monolayers exposed to medium alone maintained constant TER of 37 ± 3 Ω·cm2 for more than 30 min (n = 5). Addition of medium containing 100 μM H2O2 caused immediate decrease of TER that continued for ∼10 min and then spontaneously reversed, subsequently increasing back to baseline in ∼15 min (Fig. 1, A and B). Since TER changes reflect changes in paracellular permeability, the sequential decrease and increase of TER signified the opening and resealing of endothelial junctions. Using the DCF method for H2O2 detection (see materials and methods), we determined that H2O2-induced fluorescence was similar in aliquots of medium taken at the beginning and end of the TER measurement period (not shown). Hence, the TER recovery occurred despite continuous presence of H2O2.
To consider the role of FAK in these TER responses, we established RLMEC monolayers expressing the mutant, inactive form of FAK, del-FAK. Previously, we showed that del-FAK expression in RLMEC blocks FAK activity (25). In del-FAK-expressing cells, baseline TER was ∼20% lower than WT (n = 6; P < 0.05) (Fig. 1B). H2O2 exposure decreased TER in both WT and del-FAK-expressing monolayers (Fig. 1, A and B). However, in contrast to WT, in del-FAK-expressing cells the TER decrease persisted without recovery to baseline (Fig. 1, A and B). These findings indicated that although these mutant monolayers were successful in establishing junctional opening, the lack of active FAK abrogated junctional resealing.
In WT cells, baseline staining with rhodamine-phalloidin revealed the known distribution of F-actin as bands circumscribing the nucleus centrally and lining the cell cortex peripherally (Fig. 2A). To determine peripheral actin content, viewing single cells at high magnification (Fig. 2A, bottom), we quantified fluorescence in a zone that included the cell periphery and the cytoplasm but excluded the perinuclear actin band (Fig. 2C).
After 10 min of H2O2 exposure, namely at the time point corresponding to the minimum TER decrease (Fig. 1A), increased stress fiber formation was revealed by the appearance of cell-spanning actin filaments (Fig. 2A). After a further 30 min of H2O2 exposure, corresponding to the time point of TER recovery to baseline (Fig. 1A), stress fiber formation persisted and the peripheral actin bands were more prominent than at baseline (Fig. 2A). These changes in the peripheral actin response were evident as a time-dependent fluorescence increase (Fig. 2C). By contrast, in del-FAK-expressing cells, baseline actin distribution was poorly organized (Fig. 2B), and H2O2 exposure failed to induce detectable increases in stress fibers or in peripheral actin (Fig. 2, B and C).
In WT cells, fluorescence spots signified distribution of vinculin- and paxillin-staining at cell-matrix contacts (Fig. 3). We viewed single cells at high magnification (Fig. 3, A and C, bottom), and we quantified the number of fluorescent spots and the sizes of the fluorescence spots by image analysis. We grouped the fluorescence spots in three size classes, namely small (<0.5 μm2), medium (>0.5 to <1 μm2), and large (>1 μm2) (Table 1). For both vinculin and paxillin, the smallest spots were more numerous than the medium or the larger ones (P < 0.05). In the smallest class alone, more spots stained for paxillin than vinculin at baseline and also at the 10- and 30-min time points of H2O2 exposure (Fig. 3E) (P < 0.05). For both vinculin and paxillin, all sizes of fluorescent spots increased in number relative to baseline at the 10-min time point but did not further increase significantly at 30 min (Fig. 3, E–G). However, for all size classes, mean counts for vinculin, but not paxillin, increased more than two times baseline at 30 min (Fig. 3H) (P < 0.05), indicating that the spots progressively recruited vinculin. By contrast, in del-FAK-expressing cells, fluorescent spots were relatively sparse and failed to increase after H2O2 exposure (Fig. 3, B and D) (data from image analysis not shown).
Cell surface E-cadherin.
To determine the role played by E-cadherin in the present TER responses, we immunoblotted for E-cadherin in immunoprecipitates of cell surface proteins recovered by the biotin-streptavidin approach. In WT monolayers, H2O2 exposure decreased cell surface E-cadherin content within 5 min (Fig. 4, A and C). The decrease continued for a further 10–20 min, although by 30 min the content had returned to baseline (Fig. 4, A and C).
Densitometric analyses indicated that in del-FAK-expressing monolayers, E-cadherin contents were lower than WT both for the cell surface fraction as well as the total (P < 0.05) (Fig. 4, B–E). However, contents of paxillin and α-actinin remained unchanged (Fig. 4, D and E), indicating that the del-FAK effect was E-cadherin specific. Moreover, during H2O2 exposure in del-FAK cells, cell surface E-cadherin content progressively decreased with no recovery to baseline (Fig. 4, B and C).
In untreated WT cells, E-cadherin immunofluorescence was distributed as an unbroken line that marked the cell periphery (Fig. 5A, left), indicating that E-cadherin was continuously present at the cell periphery. At baseline, image analysis along the cell periphery revealed peaks of fluorescence (Fig. 5C, arrows), indicating that E-cadherin content was high at specific locations of the cell perimeter. At the 10-min point of H2O2 exposure, corresponding to the point of maximum TER decrease, images revealed a discontinuous and granular fluorescence at the cell periphery (Fig. 5A, middle). Image analysis indicated a downshift of E-cadherin fluorescence and loss of fluorescence peaks (Fig. 5C, red line). In fact, the peripheral fluorescence reached zero at several points along the perimeter (Fig. 5C, red line), indicating that at these locations E-cadherin was completely absent. Subsequently, at 30 min, continuity and intensity of baseline fluorescence, as well as the presence of fluorescence peaks, were reestablished (Fig. 5A, right, and 5C, black line).
By contrast, in del-FAK cells, baseline distribution of peripheral E-cadherin was poorly developed in that the fluorescence line was jagged and discontinuous and reached zero at several points along the perimeter (Fig. 5B). No recovery responses were evident, and the fluorescence diminished after 30 min of H2O2 exposure (Fig. 5, B, D, and E).
We determined the extent to which E-cadherin repletion rescues barrier responses in del-FAK cells. Transfection increased E-cadherin-GFP content of del-FAK cells to levels approaching that of endogenous E-cadherin content in nontransfected WT cells (Fig. 6A; compare lanes 1 and 3). However, although in E-cadherin-GFP-transfected WT cell (WT-Ecad) monolayers E-cadherin-GFP localized to cell junctions (Fig. 6B, left), in del-FAK-Ecad monolayers E-cadherin-GFP distribution was distributed diffusely in the cytosol and lacked definition at cell junctions (Fig. 6B, right). Further, overexpression of E-cadherin-GFP did not increase TER in del-FAK monolayers (Fig. 6C); moreover, there was poor rescue of the TER recovery after 30 min of H2O2 exposure (Fig. 6). These findings indicated that despite E-cadherin enrichment, del-FAK cells were incapable of establishing the WT resealing response.
Our studies address the unanswered question of how open EC junctions reseal. To establish a model of junctional resealing, we exposed endothelial monolayers to a relatively low concentration of H2O2 at which endothelial junctions opened, as evident in the decrease of TER that occurred in ∼10 min. However, TER subsequently increased back to baseline levels in a further 10–15 min, indicating that open endothelial junctions proceeded to reseal despite the continuous presence of H2O2.
The decreasing and increasing phases of the TER response provided a temporal format against which we could determine cellular events corresponding to the opening and resealing of endothelial junctions. Because of likely membrane-matrix interactions induced by junction opening, we considered time-dependent responses of protein structures formed at cell-matrix contacts. The dynamics of these structures are best understood for migrating cells (40, 41). At the leading edge of migrating cells, the structures increase in area, transforming from small (<0.5 μm2) to large (>0.5 μm2) focal complexes and finally to the even larger focal adhesions (1–2 μm2) (6, 8). Here, we detected cell-matrix contacts as small, medium, and large fluorescent spots, among which the small focal complexes were most numerous. Within 10 min of H2O2 exposure, all focal contacts increased in number. Hence as different from migrating cells, these stable cells responded by forming not only new focal adhesions, but also new focal complexes.
A comparison of the mean increases in the numbers of focal contact structures revealed larger increases in vinculin- than paxillin-staining structures, consistent with the notion that vinculin is increasingly recruited to growing focal complexes (8). A causal role for focal adhesions in the resealing response was indicated in cells expressing del-FAK. Previously, we showed that in RLMEC expressing del-FAK, endogenous FAK activity decreases, although FAK expression remains unchanged from WT (25). Here, we show that these focal adhesion-incompetent del-FAK cells failed to establish resealing. We interpret that in the junction-opening phase, focal adhesion formation and FAK activation were essential precursors to junctional resealing.
Studies of the crystal structure of vinculin indicate that concomitant with recruitment to focal complexes, conformational changes allow vinculin to determine cytoskeletal assembly in both focal adhesions and adherens junctions (14). Although present mechanisms remain unclear, it is possible that here vinculin recruitment to focal adhesions was determined by cell contraction-induced FAK activation. In turn, recruited vinculin promoted junctional assembly of E-cadherin.
Our findings are relevant to the increasing interest in the barrier regulatory role of FAK. In lung microvascular EC, FAK is critical in hyperosmolarity-induced barrier enhancement (25). In pulmonary arterial EC, inhibition of FAK by downregulation of FAK expression, or expression of the FAK-related nonkinase (FRNK) that competitively blocks FAK localization to focal adhesions and FAK phosphorylation (24), augmented thrombin-mediated barrier deterioration (12). Although these findings implicate FAK in endothelial barrier enhancement, conflicting findings indicate that FRNK protects the coronary (37) but deteriorates the lung endothelial barrier (12). We believe that these opposite results may be attributable to the nonspecificity of FRNK, since it is shown that FRNK blocks phosphorylation of the proline-rich tyrosine kinase 2 (PYK2) (11) and that PYK2 itself deteriorates the endothelial barrier (35).
To determine the role of cadherins in these barrier responses, we quantified the fluorescent distribution of E-cadherin at the cell periphery. Previously, we reported that E-cadherin is the major cadherin isoform expressed in lung microvascular EC (25), a finding that has also been reported subsequently by others (10, 22). The immunofluorescence data provided an assessment of E-cadherin distribution along the cell perimeter. Our findings indicate that peripheral E-cadherin content is nonuniform, being higher at specific locations marked by fluorescence peaks. Although we cannot be certain, we suggest that at these locations E-cadherin “plaques” (1) form at sites of cell-cell adhesion.
In addition to immunofluorescence, we quantified the total cell surface E-cadherin by immunoprecipitation using the biotin-streptavidin approach. By both approaches, peripheral E-cadherin decreased and the fluorescence peaks along the cell perimeter disappeared in the phase of junction opening. Recovery of peripheral E-cadherin occurred only after focal adhesions were well established. These temporal changes suggest that focal adhesion formation initiated peripheral repletion of E-cadherin.
To determine FAK-E-cadherin interactions, we expressed E-cadherin-GFP in both WT and del-FAK cells. Functionality of the expressed E-cadherin-GFP has been reported (13) in that the expressed protein forms oligomers on the cell surface before being assembled at sites of cell-cell adhesion at which it forms cytoskeletal contacts. In our studies, E-cadherin-GFP was well targeted to cell junctions. Moreover at baseline, TER was not different between WT monolayers and monolayers expressing E-cadherin-GFP, indicating that E-cadherin-GFP expression did not deteriorate barrier properties.
In contrast to WT and consistent with findings in HeLa cells in which FAK knockdown depletes N-cadherin and reduces cell adhesion (38), E-cadherin distribution in del-FAK cells was weak and discontinuous along the cell border. Moreover, H2O2 failed to enrich E-cadherin at cell junctions despite the induction of progressive barrier deterioration. Furthermore, in del-FAK cells enriched with E-cadherin-GFP, fluorescence was present in the cytosol but notably absent at cell junctions. Accordingly, in these cells, despite E-cadherin enrichment the barrier resealing response was poorly developed. These findings implicate FAK in the targeting of E-cadherin to endothelial junctions.
Although stress fiber formation is usually associated with intercellular gap formation (32), an unexpected result was that the cells developed junctional resealing despite the persistence of stress fibers. Consistent with findings from our and many other laboratories, the E-cadherin enhancement corresponding to the recovery of TER was associated with enhancement of peripheral actin. This actin enhancement may provide mechanical stability to the remodeled junction during the process of barrier resealing.
In conclusion, our findings indicate that in EC monolayers, H2O2 exposure induced focal adhesion formation, hence the FAK-induced peripheral E-cadherin and actin enhancements responsible for junctional resealing. In the absence of active FAK, junctional resealing was not evident. The physiological significance of our findings may be that EC decrease E-cadherin to effect partial junction openings, evidently to an extent that permits rapid resealing. Such partial junction openings may be sufficient to accommodate transvascular fluid shifts required for control of extravascular volume or for allowing leukocyte migration. By contrast, the opening and resealing of epithelial junctions occurs over a time course of hours (15, 20, 39). Such an extended time course of junctional resealing would clearly be physiologically counterproductive in the endothelial context, since long-term openings would lead to pathological edema formation. Nevertheless, further understanding is required to define the roles that E-cadherin and other EC proteins might play in resealing more pathological junction openings.
This study was supported by National Heart, Lung, and Blood Institute Grant HL-36024.
We thank Poshala Aluwihare for technical assistance with initial experiments and Dr. Sunita Bhattacharya for reading the manuscript. Drs. Carrie Perlman and Jens Lindert helped with image analyses.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2007 the American Physiological Society