INTRODUCTION

Top Abstract Introduction Methods Results Discussion References

THE VASCULAR ENDOTHELIUM forms a continuous cellular barrier that controls the exchange of macromolecules and solutes across the vascular wall. Endothelial cells are attached to each other by a circumferential junctional zone, consisting of a beltlike adherens junction into which tight junctions and gap junctions are morphologically inserted (10). Interendothelial junctions not only control paracellular permeability but probably also contribute to the mechanical cohesiveness of the endothelium and assist the monolayer to resist the mechanical forces of blood flow (shear stress) and blood pressure (wall distension).

Endothelial cells are held together by various types of intercellular adhesion molecules: 1) Ca²⁺-dependent transmembrane cadherins comprising vascular endothelial (VE) cadherin, P-cadherin, and N-cadherin; 2) platelet endothelial cell adhesion molecule 1 (PECAM-1), a transmembrane adhesion molecule belonging to the Ca²⁺-independent immunoglobulin superfamily; and, finally, 3) the integrins ₂/₁ and ₅/₁ (for review, see Ref. 6). The precise function of these different intercellular adhesion molecules and their contribution to the adhesive strength of the interendothelial junctions are largely unknown.

In the present study, we focused on endothelial cadherins and their associated cytoplasmic protein, plakoglobin (4, 11). Cadherins have been shown previously in various cellular systems to play an important role in intercellular adhesion and differentiation (for reviews, see Refs. 14, 30). The known endothelial cadherins are transmembrane proteins with a large NH₃-terminal extracellular domain and a short COOH-terminal cytoplasmic tail. The functional unit of these classical cadherins appears to be a side-to-side dimer (parallel "strand dimer") formed between the extracellular domains (28). The extracellular domains of the dimers provide Ca²⁺-dependent homophilic binding to strand dimers of the opposite membrane of adhering cells (antiparallel adhesion dimers), whereas the cytoplasmic tail of the dimers is anchored to the actin filament cytoskeleton. This linkage is mediated by plakoglobin and -catenin, which bind to the cytoplasmic domain of cadherins and associate with -catenin, an actin-associated protein that binds to actin and -actinin (12, 13, 15, 21, 22).

Plakoglobin and -catenin can be coimmunoprecipitated in separate complexes with cadherin and -catenin in epithelial (3) and endothelioma cells (U. Gotsch and D. Vestweber, personal communication). In epithelial cells, plakoglobin and -catenin appear to fulfil, at least in part, different functions as indicated by the exclusive association of desmosomes with plakoglobin but not with -catenin (4, 25). First hints to different functions between plakoglobin and -catenin in endothelial cells were obtained by the observation that during formation of a confluent monolayer -catenin appears first (together with VE-cadherin) at the intercellular junctions, followed by plakoglobin (16).

This observation, in conjunction with the exclusive desmosomal location of plakoglobin, suggests a more general role of plakoglobin in certain aspects of the assembly of adhesive junctions. To address this question, we designed two experimental approaches. In a first series of experiments, we wanted to know whether cadherins are absolutely essential for intercellular adhesion or whether the remaining adhesion molecules, such as PECAM-1 (20), are sufficient for adhesion under both resting and flow conditions. In a second experimental design, the expression of plakoglobin was suppressed by an antisense approach in the hope of obtaining further clues to the functional role of plakoglobin in endothelial cells.

	MATERIAL AND METHODS

Top Abstract Introduction Methods Results Discussion References

Antibodies

Cell Culture

Ethylene glycol-bis(-aminoethyl ether)-N,N,N',N'-tetraacetic acid treatment. Cell cultures were treated for various periods with 3 mM ethylene glycol-bis(-aminoethyl ether)-N,N,N',N'-tetraacetic acid (EGTA; Sigma) dissolved in M199, M199-serum, or phosphate-buffered saline [PBS; consisting of (in mM) 137 NaCl, 2.7 KCl, 8.1 Na₂HPO₄, and 1.5 KH₂PO₄, pH 7.4]. The stock solution of EGTA was 0.3 M adjusted to pH 7.4 with 5 M NaOH. The free Ca²⁺ concentration was <10⁷ M in all solutions. Ca²⁺ concentrations were calculated according to Föhr et al. (8) using a computer program kindly provided by M. Gratzl (Munich, Germany).

Immunofluorescence and Silver Nitrate Staining

For silver staining of cell borders, cultures were rinsed briefly with distilled water and then exposed for 10 min to 0.5% AgNO₃ in distilled water. Subsequently, cultures were rinsed three times (1 min each) with distilled water and then (still covered with a thin layer of water) exposed to ultraviolet light for 5 min.

Cell Extraction

Cell fractionation was performed at 4°C. All buffers contained 1 mM phenylmethylsulfonyl fluoride, 25 µg/ml leupeptin, 25 µg/ml aprotinin, and 25 µg/ml pepstatin (Sigma). Cell cultures were briefly rinsed with M199, followed by a further brief rinse with a hypotonic buffer containing 10 mM NaH₂PO₄, 2 mM MgCl₂, pH 7.4 (buffer H). Different extraction buffers were used: buffer H, buffer H containing 2.5% Nonidet P-40, and buffer H containing a mixture of 1% Triton X-100 and 1% Nonidet P-40. For cell extraction, 5 ml of buffer were used for one 75-cm² culture flask. After exposure to one of these extraction buffers for 15 min, cells were scraped off the substratum with a rubber policeman. Subsequently, samples were homogenized by 50 strokes in a 7.5-ml dounce homogenizer and then centrifuged for 5 min at 14,000 g. The 100,000-g (45 min) pellets of the supernatant were defined as the crude membrane fraction (CMF), and the supernatants of this pellet were defined as the cytosolic fraction (CF). Proteins of supernatants were precipitated by 10% trichloroacetic acid (TCA) for 12 h, pelleted (10,000 g, 5 min), and washed three times with 1% TCA in water. Samples were dissolved in sample buffer (boiled for 5 min) and further processed for Western blot analysis.

Microinjection of Oligonucleotides

Microinjection was performed using an Eppendorf (Hamburg, Germany) micromanipulator 5170 equipped with a microinjector (Sauer, Würzburg, Germany). Capillaries were manufactured with a capillary puller (Bachhofer, Reutlingen, Germany) that had an inner diameter of ~0.5 µm. Holding pressure was 30 hPa, the injection pressure was 80 hPa, the injection time was ~200 ms, and the injection volume was 5-15 fl. If microinjection caused cellular damage of only one cell, experiments were discarded. Microinjection data are based on five independent experiments; in each experiment, groups of 10-20 adjacent cells were injected with control and antisense oligonucleotides without any cellular damage. Microinjected cells were identified by a numbered grid etched into the backside of the cover slides on which the endothelial cells were grown (Eppendorf). All experiments were repeated five times.

Exposure to Fluid Shear Stress and Quantitative Estimation of Gaps

	RESULTS

Top Abstract Introduction Methods Results Discussion References

Adhesive Properties of Interendothelial Junctions in Low Extracellular Ca²⁺

View larger version (108K): [in this window] [in a new window]   Fig. 1.   Effect of extracellular Ca2+ depletion [3 mM EGTA in M199 supplemented with 20% pooled human serum (M199-serum)] on distribution of vascular endothelial (VE) cadherin (a and d), plakoglobin (b and e), and platelet endothelial cell adhesion molecule 1 (PECAM-1; c and f) in confluent endothelial monolayers (indirect immunofluorescence). In low Ca2+, VE-cadherin and plakoglobin disappear from junctions, whereas junctional staining for PECAM-1 remains undisturbed for several hours. Bar, 40 µm.

The disappearance of cadherins and plakoglobin from interendothelial junctions at low Ca²⁺ (EGTA in M199-serum) was fully reversible after recalcification (medium exchange), even in the presence of cycloheximide. Within 5 min, VE-cadherin and -catenin became detectable at the junctions, followed by plakoglobin that began to accumulate at the junctions during the following 20 min (Fig. 2). -Catenin behaved like VE-cadherin and -catenin (not shown). Western blot analysis showed that the bulk of cadherins and plakoglobin remained associated with the membrane fraction before and after EGTA treatment. Membrane-bound cadherins and plakoglobin could be largely released from the membrane by 1% Triton X-100 (Fig. 3), suggesting that plakoglobin remained attached to VE-cadherin under these conditions.

View larger version (75K): [in this window] [in a new window]   Fig. 2.   Time course for reappearance of VE-cadherin, -catenin, and plakoglobin at intercellular junctions after Ca2+ depletion for 2 h and subsequent recalcification of culture medium (M199-serum; indirect immunofluorescence). VE-cadherin (a and d) and -catenin (b and e) are reappearing within a few minutes at junctions followed by plakoglobin (c and f) that becomes detectable 20-30 min later. b, c, e, and f were taken from monolayers double labeled with rabbit polyclonal anti--catenin and mouse monoclonal antiplakoglobin. Bar, 40 µm.

View larger version (27K): [in this window] [in a new window]   Fig. 3.   Subcellular compartmentalization of cadherins and plakoglobin in endothelial cells grown in M199-serum (untreated) in low Ca2+ (3 mM EGTA for 3 h). Cell homogenates were cleared at 14,000 g (5 min), and supernatant was centrifuged at 100,000 g (45 min). Resulting 100,000-g supernatant (cytosolic fraction, cf) and pellet (crude membrane fraction, cmf) were probed with antibodies directed to cadherins (pan-cadherin; A) and plakoglobin (B), respectively. Fractionation was conducted in absence and presence of 1% Triton X-100. In EGTA-treated and untreated cultures, bulk of cadherins and plakoglobin was associated with membrane fraction. Triton X-100 caused a release of cadherin and plakoglobin from membranes, and only a small fraction of both proteins (20-30%) remained in a Triton-resistant pool.

As stated above, EGTA treatment (in M199-serum) of endothelial monolayers for 15 min caused complete disappearance of junctional cadherins, whereas the amount of PECAM-1 appeared unchanged at the junctions. Intercellular gaps were not observed by silver staining, although the intensity of silver precipitations filling the intercellular space was slightly reduced in cultures exposed to low Ca²⁺ (Fig. 4). This observation is compatible with our previous findings (27) showing no increase in hydraulic conductivity in EGTA-treated endothelial monolayers. When monolayers were exposed to rheological shear stress (20 dyn/cm²) 5-30 min after extracellular Ca²⁺ depletion, disruption of intercellular junctions occurred within a few minutes, as revealed by the formation of numerous intercellular gaps that were increasing in number and size during the following 15 min of shear stress exposure (Fig. 4d). Treatment for 5 min with EGTA and subsequent exposure for 15 min to shear stress caused numerous gaps ranging from 11.9 to 19.8% (n = 3) of the total cell area. Shear stress-induced gaps were fully reversible under subsequent resting conditions even in the presence of EGTA (not shown). Exposure to lower shear stress levels (6 dyn/cm²) resulted also in the formation of intercellular gaps, but gap formation was clearly delayed and took 30-60 min to become visible (not shown).

View larger version (128K): [in this window] [in a new window]   Fig. 4.   Effect of extracellular Ca2+ on integrity of confluent endothelial monolayers under resting conditions (a and b) and mechanical loads of fluid shear stress (20 dyn/cm2 for 15 min; c-f). Cell nuclei and cell borders were visualized by methylene blue and silver staining. Low extracellular Ca2+ (3 mM EGTA in M199-serum) reduced silver staining intensity along cell borders but did not affect monolayer integrity (b). However, shear stress exposure caused disruption (gap formation) of intercellular junctions (d), whereas, in control monolayers cultured in presence of 1.8 mM Ca2+, no gaps were seen (c). Shear stress resistance of junctions was gradually retained after recalcification, as seen by reduction of gaps 5 min later (e; arrows indicate gaps) and full resistance of junctions (no gaps, reappearence of silver lines) 30 min after readjustment of extracellular Ca2+ (f). Bar, 40 µm.

When EGTA-treated cultures were exposed to shear stress (20 dyn/cm²) 5 min after recalcification (when VE-cadherin and -catenin were becoming detectable at the junctions), gap formation was markedly reduced (decreased to 2.5-5.5% of total cell area) but not completely prevented (Fig. 4e). However, complete shear stress resistance of junctions was retained only 30 min after recalcification (Fig. 4f), when junctional VE-cadherin and -catenin had reached control levels and plakoglobin was reappearing. Under these conditions, gap formation was reduced to control levels of ~1%. Thus full shear stress stability of the junctions is retained at a stage when plakoglobin becomes visible at the junctions. This conclusion is compatible with our observation that exposure of recalcified (5 min) cultures to lower levels of shear stress (6 dyn/cm²) did not cause significant gap formation. During the relatively long interval needed for gap formation (at least 30 min in EGTA, see above), plakoglobin had time enough to reappear and stabilize the junctions.

Effect of Plakoglobin Depletion on Junctional Integrity Under Resting and Flow Conditions

View larger version (101K): [in this window] [in a new window]   Fig. 5.   Simultaneous immunocytochemical visualization of plakoglobin (a and c), VE-cadherin (b), and -catenin (d) in an endothelial monolayer 24 h after microinjection of plakoglobin-specific antisense oligonucleotides. Border between injected and noninjected cells is indicated by dashed lines. In b, VE-cadherin stain is superimposed on plakoglobin stain shown in a (sequential staining; star indicates an identical cell). Note absence of junctional immunoreactivity for plakoglobin in microinjected cells (arrows) that, however, still show immunostaining for VE-cadherin (b) and -catenin (d). Bar, 40 µm.

Loss of plakoglobin also did not affect the junctional location of -catenin, as revealed by double immunostaining using monoclonal antiplakoglobin (Fig. 5c) and polyclonal anti--catenin (Fig. 5d), which remained associated with the cell borders of microinjected cells in amounts (immunostaining intensity) comparable to the neighboring noninjected cells. The junctional location of PECAM-1 also remained undisturbed in cells microinjected with antisense oligonucleotides (Fig. 6c). No gap formation was observed between microinjected cells, although we cannot exclude the possibility that ultrasmall gaps might have been formed below the resolution of the light microscope and the gray-value range used for image analysis. Cells microinjected with control oligonucleotides behaved like noninjected control cells; i.e., no changes in the staining pattern for plakoglobin (Fig. 6a), VE-cadherin (Fig. 6b), and PECAM-1 (Fig. 6d) were detectable.

View larger version (98K): [in this window] [in a new window]   Fig. 6.   Control experiments in which groups of endothelial cells of confluent monolayers were microinjected with control oligonucleotides (a, b, and d) or antisense oligonucleotides to plakoglobin (c) and then (after 24 h) immunostained for plakoglobin (a), VE-cadherin (b), and PECAM-1 (c and d). Dashed lines (a and b) indicate border between injected and noninjected cells. Microinjection of control oligonucleotides does not significantly affect immunostaining for all 3 proteins (a, b, and d). Note that plakoglobin depletion (c) leaves immunostaining for PECAM-1 unchanged compared with cells injected with control probes (d). Bar, 40 µm.

Shear stress exposure (6 dyn/cm²) of monolayers 24 h after microinjection of antisense oligonucleotides resulted in significant gap formation between microinjected cells, whereas intercellular adhesion was not altered between the neighboring noninjected cells and cells injected with control oligonucleotides. Gaps became visible between the antisense-injected cells 10 min after the onset of shear stress exposure and then continuously increased in number and size (Fig. 7a). Image analysis of the gaps formed between plakoglobin-deficient cells shear stressed for 1 h revealed 7.0-13.0% (n = 3) of the total cell area covered by gaps (Fig. 7b). Control injected or noninjected cells did not show significant gap formation (<1.5%).

View larger version (80K): [in this window] [in a new window]   Fig. 7.   Mechanical stability of intercellular junctions in cultured endothelial monolayers in which groups of cells were microinjected with plakoglobin-specific antisense oligonucleotides 24 h before shear stress exposure (6 dyn/cm2) for 1 h. a: Phase contrast image. b: Same monolayer area as in a processed by gray-value image analysis for visualization and quantification of gaps. Dashed lines indicate border between injected and noninjected cells. Exposure of monolayers to fluid shear stress results in conspicuous gap formation between cells injected with antisense oligonucleotides (arrows) but not between noninjected cells. Bar, 40 µm.

In addition to these experiments performed with cultured endothelial cells of the human umbilical cord vein, we also used cultured arterial endothelial cells of the porcine pulmonary trunk for microinjection. Cells were isolated and cultured as described in detail elsewhere (26). Shear stress exposure (6 dyn/cm²) 24 h after microinjection of antisense oligonucleotides caused numerous gaps between the microinjected cells within 10 min but not between noninjected cells and cells injected with control oligonucleotides. Immunostaining confirmed complete absence of plakoglobin immunoreactivity of the microinjected cells. In Fig. 8, cells were stained with TRITC-labeled phalloidin (26) to allow discrimination between unstained intercellular gaps and the stained endothelial cytoplasm with its many stress fibers and the F-actin-rich cortical cytoplasm.

View larger version (144K): [in this window] [in a new window]   Fig. 8.   Cultured arterial endothelial cells of porcine pulmonary trunk treated in same way as described in Fig. 7. Stars and dotted line indicate borders between injected and noninjected cells. Cells were stained after shear stress exposure with fluorescent phalloidin to facilitate visualization of gaps (unstained structures indicated by arrows). Arrowheads point to cell borders of adjacent noninjected cells. Bar, 10 µm.

	DISCUSSION

Top Abstract Introduction Methods Results Discussion References

In the present study, we obtained new insights into the functional role of endothelial cadherins and of the cadherin-binding cytoplasmic protein plakoglobin.

Vascular Cadherins Are Essential for Cohesiveness Under Flow But Not Under Resting Condition

Whereas, at low extracellular Ca²⁺ and resting conditions, adhesion between endothelial cells was not disturbed in our experiments, adhesiveness between endothelial cells was too weak under these conditions to withstand arterial levels of fluid shear stress. This indicates that Ca²⁺-dependent adhesion molecules are essential for holding the cells together under the physiological loads of blood flow. The shear stress-induced gaps were fully reversible in low extracellular Ca²⁺, and the closure of the gaps occurred in the absence of any detectable amounts of cadherins at the junctions. This observation is surprising, because intercellular adhesion between various types of cultured epithelial cells has been shown to critically depend on extracellular Ca²⁺ even under resting conditions (for review, see Ref. 3). This obvious difference between endothelial and epithelial cells may be related to the fact that adhesion between epithelial cells is more complex (3 different types of Ca²⁺-dependent adhesion junctions) than adhesion between endothelial cells (6, 7, 25).

Role of Plakoglobin for Junctional Integrity Under Flow Conditions

A similar cooperative transmembrane event would also explain the loss of cadherins and catenins from the junctional membrane in low extracellular Ca²⁺. Low Ca²⁺ may primarily lead to an exoplasmic dissociation of the adhesion dimers, whereas the strand dimers might not dissociate into monomers under these conditions. This increase of the entropy of the dimers might result in a cytoskeletal disconnection and subsequent dissociation of the junctional cadherin clusters.