Tightening of Endothelial Cell Contacts: a Physiologic Response to Cocultures with Smooth-Muscle-Like 10T1/2 Cells

View metadata, citation and similar papers at core.ac.uk brought to you by CORE

provided by Elsevier - Publisher Connector

Tightening of Endothelial Cell Contacts: A Physiologic Response to Cocultures with Smooth-Muscle-Like 10T1/2 Cells

Hjalmar Kurzen,* Sabine Manns,* Gudrun Dandekar, Tim Schmidt, Silke PraÈtzel, and Birgit Maria KraÈling German Cancer Research Center, Division of Cell Biology/A0100, Heidelberg, and *Department of Dermatology, University of Heidelberg, Germany

Tightening of endothelial cell-to-cell contacts is an ive monocultures. Immuno¯uorescence on cells and important event at the end of angiogenesis in order Western blot analyses were performed for marker to achieve controlled transfer of solutes between the proteins of adherens and tight junctions. Functional blood stream and solid tissues. We found that tight- permeability assays were performed for the tracer ening of endothelial cell-to-cell contacts and the for- molecule biotin-dextran. The results indicated that mation of a permeability barrier can be induced 10T1/2 cells induced the tightening of endothelial in vitro by dibutyryl cAMP and hydrocortisone. This cell-to-cell contacts. Plakoglobin, ZO-1, ZO-2, and process is accompanied by increased junctional local- occludin showed increased junctional localization ization and cytoskeletal association of the adherens when 10T1/2 cells were present. Cocultures also dis- junctional plakoglobin and the tight junction associ- played a signi®cantly higher permeability barrier for ated proteins ZO-1, ZO-2, and occludin. Based on the tracer molecule biotin-dextran. In conclusion, these ®ndings, we proceeded to investigate whether mural cells such as smooth muscle cells and pericytes smooth-muscle-like mesenchymal cells would in¯u- may be important for stabilizing endothelial cell-to- ence endothelial junctional differentiation. For this cell contacts and may in¯uence vessel-type speci®c purpose, human umbilical chord vein endothelial differences of the endothelial phenotype. Key words: cells and murine smooth-muscle-like 10T1/2 cells adherens junction/angiogenesis/cell culture/differentiation/ were cocultivated and compared with their respect- tight junction. J Invest Dermatol 119:143±153, 2002

ngiogenesis is the formation of new capillaries from glycoproteins of the cadherin family connect via speci®c cytoplas- pre-existing blood vessels (Risau, 1995; Wilting and mic plaque proteins of the armadillo family to the cytoskeleton Christ, 1996). For angiogenesis to proceed, the (Farquhar and Palade, 1963; Schmidt et al, 1994; Takeichi, 1995). endothelial cells ®rst have to loosen their cell-to-cell By their junctional composition and the type of ®lament anchored, Acontacts, degrade the underlying basement membrane, two basic types of adhering junctions can be distinguished in migrate into the surrounding tissue, and proliferate, after which the epithelia: (i) adherens junctions in which the classical type I endothelial cells realign and form a new capillary sprout. In order cadherins (E-, N-, P-, R-cadherin) link via a-catenin, b-catenin, for the endothelium to form a selective barrier between the blood and plakoglobin to the actin micro®laments; and (ii) desmosomes in stream and solid tissues, vascular maturation requires the re- which the desmosomal cadherins (desmogleins and desmocollins) formation of tight endothelial cell-to-cell contacts, the down- connect through desmoplakin, plakophilins, and plakoglobin to the regulation of endothelial proliferation, the deposition of a basal intermediate ®laments (Schmidt et al, 1994; Takeichi, 1995). lamina to which the endothelium tightly adheres, and the Adherens junctions are known to form lateral clusters that seem to recruitment of supporting cells to the vessel walls such as pericytes be mediated by p120ctn, another member of the armadillo family. and smooth muscle cells (Grant et al, 1991; Lampugnani and The initial cell-to-cell contacts are mediated by the classical Dejana, 1997; Risau, 1997; Balda and Matter, 1998; Hirschi et al, cadherins and adherens junctions, thus leading to further tightening 1998). Endothelial barrier formation is achieved by the establish- through desmosomes and tight junctions (Adams and Nelson, 1998; ment of speci®c, but functionally still less well understood, adhering Watabe-Uchida et al, 1998). These tight junctions are characterized and tight junctions (Schnittler, 1998; Vestweber, 2000). by the transmembrane proteins occludin and the growing family of Epithelial adhering junctions are well characterized by an claudins (Balda and Matter, 1998; Furuse et al, 1998). The tight electron-dense cytoplasmic plaque, in which transmembrane junctional cytoplasmic complexes include the proteins ZO-1, ZO- 2, ZO-3, cingulin, 7H6, and symplekin (Keon et al, 1996), which Manuscript received May 10, 2001; revised November 2, 2001; accepted mediate the linkage to the actin cytoskeleton. for publication December 5, 2001. In contrast to the epithelium, endothelial junctional complexes Reprint requests to: Dr. Birgit KraÈling, Heidelberg Innovation, Im do not show a clear spatial organization and morphologic Neuenheimer Feld 581, 69120 Heidelberg, Germany. Email: birgitkra- differentiation into adhering junctions and apical tight junctions [email protected] Abbreviations: HUVEC, human umbilical cord vein endothelial cells; (Franke et al, 1988; Schnittler, 1998; Vestweber, 2000). Instead, HDMEC, human dermal microvascular endothelial cells; Bt2, N6,2¢-O- both junctional structures seem to be present along the complete dibutyryladenosine 3¢:5¢-cyclic monophosphate; HC, hydrocortisone. lateral aspect of the endothelial contacting zones. Furthermore,

0022-202X/02/$15.00 ´ Copyright # 2002 by The Society for Investigative Dermatology, Inc. 143 144 KURZEN ET AL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY despite the reported presence of desmosomal plaque proteins fetal bovine serum (FBS; Atlanta Biologicals, Atlanta, GA), 2 mM desmoplakin and plakoglobin (Schmelz et al, 1994; Kowalczyk et al, glutamine (Gibco BRL, Paisley, U.K.) 100 U per ml penicillin (Gibco 1998), endothelial cells do not have structures that resemble BRL), 100 mg per ml streptomycin (Gibco BRL), 25 mg per ml classical desmosomes. In the endothelium, ``adherens junctions'' endothelial cell growth supplement (Sigma), and 10 mg per ml heparin (Sigma). For studies on endothelial barrier function and the respective seem to connect to both the actin micro®laments and the analysis of endothelial junctional complexes, the basal medium was intermediate ®laments of the vimentin type (Kowalczyk et al, supplemented with 0.5 mM N6,2¢-O-dibutyryladenosine 3¢:5¢-cyclic 1998). The plaque proteins a-catenin, b-catenin, plakoglobin, and monophosphate (Bt2) (Sigma) and 1 mg per ml HC (Sigma). After 3 d, desmoplakin link the cytoskeleton to VE-cadherin, the only endothelial cells in basal conditions [O] were compared with cells treated classical cadherin so far reliably identi®ed in endothelial cell with Bt2 and HC [+]. junctions. In contrast, N-cadherin, although present in endothelial Electron microscopy Con¯uent endothelial cells were treated for 3 d cells, is thought to be distributed evenly over the plasma membrane with Bt2 and HC whereas control cells were left untreated. In the without restriction to junctional complexes (Salomon et al, 1992). following, cells were ®xed for 20 min in 2.5% glutaraldehyde in 50 mM To gain insight into the functional organization of the sodium cacodylate buffer (pH 7.2) containing 50 mM KCl, 2.5 mM endothelial cell-to-cell contacts, we utilized an in vitro model for MgCl2, and 1.25 mM CaCl2 (4°C), rinsed twice in cacodylate buffer the study of endothelial maturation. It has been shown previously (4°C), post®xed for 2 h in 2% osmium tetroxide (4°C), block-stained on the basis of this model that the cultivation of endothelial cells in overnight with 0.5% uranyl acetate in H2O, and further processed for the presence of dibutyryl cAMP (Bt2) and hydrocortisone (HC) Epon ¯at embedding as described by Troyanovsky et al (1993). induced a phenotypically differentiated endothelial cell. The Immuno¯uorescence staining of cell cultures and human characteristics of this differentiated endothelial phenotype were tissues Five micron cryostat sections and cultured cells were ®xed the deposition of a basement membrane type of matrix, reduced either in ice-cold acetone or in freshly prepared 4% buffered cell proliferation, tightening of the endothelial adherens junctions, formaldehyde. In addition, cultured cells were permeabilized with and an increased association of the plaque proteins plakoglobin and methanol. Immuno¯uorescence reactions were performed according to p120ctn with the actin micro®lament system (KraÈling et al, 1999; standard procedures (KraÈling and Bischoff, 1998). After blocking with Koch et al, 2000). In this study, we analyzed the organization of the 2% nonfat dry milk in Tris-buffered saline containing 0.1% Triton X- 100 (TBST), the primary antibodies were applied to the specimens for endothelial tight junctions and the establishment of an endothelial 1 h and secondary antibodies for 30 min at 37°C. Washes (3 3 5 min) permeability barrier. To ®nd evidence for a physiologic role of the were performed after each step in phosphate-buffered saline (PBS) or endothelial differentiation in the presence of Bt2 and HC, we TBST. The staining was either observed with a Zeiss Axiophot furthermore analyzed cocultures between endothelial cells and microscope equipped for epi¯uorescence or analyzed by laser scanning 10T1/2 cells with a smooth-muscle-cell-like phenotype. The confocal microscope (Type LSM 510 UV, Zeiss, Germany). purpose of this part of the study was to investigate whether the Western blot analysis of detergent soluble and insoluble presence of 10T1/2 cells would induce tightening of endothelial cytoplasmic fractions Con¯uent endothelial cell cultures in 100 mm cell-to-cell contacts, thus leading to the formation of a permeability plates were treated for 3 d with Bt2 and HC whereas controls were left barrier. in basal conditions. The monolayers were lyzed with 1 ml lysis buffer [10 mM Tris±HCl, pH 7.4, containing 50 mg per ml digitonin and 1 MATERIALS AND METHODS tablet Complete Mini protease inhibitors (Boehringer Mannheim, Germany)] to release all soluble cytoplasmic components. The remaining Antibodies used in this study The following primary antibodies were pellets were scraped into 500 ml of either lysis buffer A (25 mM Tris± used in this study: mouse antihuman VE-cadherin (clone TEA1.31, HCl, pH 7.4, 2% Nonidet P-40) or lysis buffer B [PBS with 1% Triton Immunotech, Marseille, France; and clone BV9, a kind gift from Dr. X-100, 0.5% DOC, 0.005% sodium dodecyl sulfate (SDS), and 1 mM Elisabetta Dejana, Mario Negri Institute, Mailand, Italy) (Leach et al, ethylenediamine tetraacetic acid (EDTA)]. The detergent-insoluble 1993); mouse antihuman N-cadherin (clone 32, Transduction pellets with the remaining cytoskeleton associated constituents were Laboratories, Lexington, MA) and mouse anti-N-cadherin (clone 3B9, directly boiled in 500 ml 2 3 reducing SDS sample buffer [250 mM Zymed, San Francisco, CA); mouse antihuman a-catenin (clone 5, Tris±HCl, pH 6.8, 40% glycerol, 4% SDS (Bio-Rad Laboratories, Transduction Laboratories) and rabbit antiserum against a-catenin Richmond, CA), 4% b-mercaptoethanol (Bio-Rad), 0.004% (Sigma, St. Louis, MO); mouse antihuman b-catenin (clone 14, bromophenol blue (Sigma), 0.2 mM phenylmethylsulfonyl ¯uoride Transduction Laboratories) and polyclonal rabbit anti-b-catenin (Sigma); (Sigma)] containing 40 mM dithiothreitol (Bio-Rad). All these fractions mouse antihuman plakoglobin (clone 11E4, a kind gift from Margarete were subjected to 8% SDS polyacrylamide gel electrophoresis (PAGE), Wheelock, Toronto, Canada); mouse antihuman p120 (clone 98, subsequent Western blot analysis, and detection by ECL Reagent (NEN, Transduction Laboratories); mouse antipan cadherin (clone CH-19, Boston, MA). The speci®c gel loading was normalized to relative cell Sigma); rabbit anti-ZO-1, rabbit anti-ZO-2, rabbit antioccludin, rabbit densities as determined by methylene blue assays (Oliver et al, 1989) and antihuman claudin-1 antisera (all from Zymed); rabbit antihuman protein content as determined by Coomassie blue staining or vimentin claudin-5/-6 serum, a kind gift from Dr. Shoichiro Tsukita, Kyoto, levels. All experiments were repeated at least twice. Japan (Morita et al, 1999); mouse antihuman vimentin (clone 3B4, Coculture between HUVEC and mesenchymal 10T1/2 Progen; Heid et al 1988), rabbit antihuman von Willebrand factor (Dako cells 10T1/2 ®broblasts of mouse origin were kindly provided by Dr. Patts, Denmark). Patricia d'Amore at Children's Hospital, Boston, MA (Hirschi et al, The following secondary antibodies were used: Cy-2 and Cy-3 1998). 10T1/2 cells were cultivated in Dulbecco's modi®ed Eagle's conjugated goat antimouse, rabbit, and guinea pig antibodies obtained medium (DMEM) with high glucose (Gibco BRL), 10% FBS, 2 mM from Dianova; ¯uorescein-isothiocyanate- and Texas-red-conjugated glutamine, 100 U per ml penicillin, and 100 mg per ml streptomycin. goat antirabbit antibodies (Dianova); Alexa 488 and Alexa 594 conju- Before coculturing 10T1/2 with endothelial cells, 10T1/2 cells were gated goat antimouse, rabbit, and guinea pig antibodies (Mobitec, adjusted to the basal endothelial culture medium (DMEM:EBM = 1:1 Germany); horseradish peroxidase (HRP) conjugated horse antimouse for at least 24 h; then transferred into endothelial basal medium). Each IgG (Vector Laboratories, Burlingame, CA); HRP-conjugated goat experiment was performed three times. antirabbit IgG (Vector Laboratories). Mixed cocultures The ratio of HUVEC to 10T1/2 cells in cocultures was Culture and treatment of endothelial cells Human dermal 2:1 as the 10T1/2 cells would otherwise quickly overgrow the microvascular endothelial cells (HDMEC) were isolated from neonatal endothelial cells. The monocultures of HUVEC or 10T1/2 cells were foreskin and cultured as described by KraÈling and Bischoff (1998). established in parallel with cell numbers that were equal to the total cell Human umbilical vein endothelial cells (HUVEC) were isolated from number of the cocultures. Endothelial cells (3±4.5 3 105 cells) and umbilical chords of newborns according to the method of Jaffe et al 10T1/2 cells (1.5±2.3 3 105 cells) were mixed and plated in 5 ml (1972). HUVEC and HDMEC were cultivated on gelatinized Primaria endothelial growth medium onto 60 mm gelatinized cell culture dishes. cell culture dishes (Becton Dickinson, Franklin Lakes, NJ) and assayed For immuno¯uorescence analysis, glass coverslips (100 mm circles; between passages 3 and 15. Passage numbers between HDMEC and Langenbrink, Germany) were placed into the cell culture dishes prior to HUVEC were matched for comparison studies. The basal endothelial gelatin coating. Monocultures with either 4.5±6.8 3 105 endothelial cells cell culture medium was EBM 131 (Clonetics, Walkersville, MD), 20% or 4.5±6.8 3 105 10T1/2 cells were established in parallel as controls. VOL. 119, NO. 1 JULY 2002 TIGHTENING OF ENDOTHELIAL CELL CONTACTS 145

Figure 1. Increased interdigitations between neighboring HDMEC after treatment with Bt2 and HC. Electron microscopy on representative cross-sections of con¯uent HDMEC after treatment for Figure 2. Differential localization of constituents of adherens 3 d with basal conditions or Bt2 and HC revealed that Bt2- and HC- junctions during endothelial maturation in vitro. treated endothelial cells showed stronger interdigitations (*) between Immuno¯uorescence staining was performed on con¯uent HUVEC after neighboring endothelial cells (A¢) than those in basal conditions (A). Scale 3 d in basal conditions (A±C) or treatment with Bt2 and HC (A¢±C¢) for bar: 600 nm. N-cadherin (B, B¢) and VE-cadherin (C, C¢). Panels A and A¢ show negative immuno¯uorescence staining in the absence of primary antibodies. In basal conditions, N-cadherin (B) and VE-cadherin (C) Cultures were maintained at 5% CO2 and 37°C in a humidi®ed incubator and analyzed after 3 d. Cells were washed with HEPES wash were present at endothelial cell-to-cell contacts. The respective pattern buffer (10 mM HEPES, pH 7.4, 135 mM NaCl, 5 mM KCl, 1 mM of these components appeared focal and discontinuous. After treatment with Bt2 and HC, the junctional localization of N-cadherin almost MgSO4 and 1.8 mM CaCl2), directly scraped into 0.5 ml 2 3 reducing SDS sample buffer, and analyzed by SDS-PAGE and Western blotting as disappeared (B¢) whereas VE-cadherin (C¢) changed to continuous lines described above. For microscopic analysis, the glass coverslips were with prominent signals. Scale bars: 50 mm. removed and processed as described for immuno¯uorescence staining. All experiments were repeated at least twice. used as a substrate for HRP and the color reaction was stopped with 2 Permeability assay on endothelial cells and cocultures NH2SO4 after 15 min. Color intensities were measured at 450 nm using Endothelial cell monolayers Cells were plated on top of gelatinized a Titertek Multiscan plate reader. The background was calculated as the Transwell chambers for 24-well plates (0.4 mm pores) (Becton mean value from measuring pure endothelial growth medium, pure FBS, and pure biotin (200 mg per ml) in endothelial growth medium. The Dickinson) at densities of 1 3 104 cells per well. When they had mean background was subtracted from the raw data and the mean value reached con¯uence at day 5, one set of cells was treated with Bt2 and and standard deviation for each data point were calculated. Statistical HC and one set was maintained in basal medium for an additional 4 d at analysis on the data sets was performed by Student's t test. which point the permeability assays were performed. All experiments were repeated at least twice. RESULTS Cocultures of 10T1/2 cells with endothelial cells Gelatinized ®lters of Transwell chambers for 24-multiwell plates (Becton Dickinson) were Adherens junctions in the differentiated endothelial seeded with 10T1/2 and endothelial cells as described above except that phenotype Previously, the induction of a differentiated 5 3 103 10T1/2 cells in endothelial growth medium were plated onto 4 endothelial phenotype was described after treatment of the bottom side of the ®lter membranes and 1 3 10 endothelial cells in endothelial cell cultures with Bt2 and HC (KraÈling et al, 1999; endothelial growth medium were plated onto the upper side of the ®lter Koch et al, 2000). One hallmark of this differentiated phenotype membranes. Control cultures were established with either endothelial cells alone or 10T1/2 cells alone in comparable cell numbers (5 3 103 was the tightening of endothelial cell-to-cell contacts after addition cells onto the bottom side and 1 3 104 cells onto the top side of the of Bt2 and HC. In contrast, endothelial cells in basal conditions showed small gaps in the contacting zones between individual cells ®lter membrane). The cultures were maintained at 5% CO2 and 37°C in a humidi®ed incubator with fresh medium every third day for a total of even after reaching con¯uence (Koch et al, 2000). The mechanisms 8 d when the permeability assays were performed. for closing the endothelial cell-to-cell contacts could be the Permeability assay The tracer molecule biotin-dextran (Sigma) (»70 kDa) elongation of interdigitations (not shown) or the formation of was added to the top compartment of the Transwells at a concentration increased numbers of interdigitations between neighboring of 200 mg per ml. The permeability of the cell layers for this tracer endothelial cells that could be observed after treatment with Bt2 molecule was measured by collecting aliquots from the bottom and HC (Fig 1A¢) in comparison to endothelial cells in basal compartments. For this purpose, at each indicated time point, the conditions (Fig 1A). We observed examples of both mechanisms solution from the bottom compartment was aspirated and replaced by after electron microscope analysis on cross-sections of endothelial 0.5 ml HEPES-buffered saline (10 mM HEPES, pH 7.4, 135 mM NaCl, monolayers before and after BT2 and HC treatment. Furthermore, 5 mM KCl, 1 mM MgSO4, and 1.8 mM CaCl2). Duplicate aliquots (2 3 35 ml) for sample detection were removed after 10 min. Sample as we could already show (Koch et al, 2000), VE-cadherin and detection occurred by regular enzyme-linked immunosorbent assay plakoglobin displayed an increased localization at the sites of (ELISA) methods (KraÈling and Bischoff, 1998) on streptavidin-coated 96- endothelial cell-to-cell contact after Bt2 and HC treatment (shown well microtiter plates (MicroCoat, Germany) with HRP-conjugated for VE-cadherin in Fig 2C¢). A novel ®nding of this study, mouse antibiotin antibody (Sigma). Tetramethylbenzidin (Sigma) was however, was that N-cadherin, even though showing strong signals 146 KURZEN ET AL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY after immuno¯uorescence staining at the endothelial cell-to-cell prominent staining for occludin, ZO-1, and ZO-2, which were contacts in basal growth conditions (Fig 2B), became barely visible as continuous lines (Fig 4B¢±E¢). Claudin-1 and claudin-5 detectable after treatment with Bt2 and HC (Fig 2B¢). The were increasingly found at the sites of cell contacts with a observed downregulation of N-cadherin after treatment with Bt2 heterogeneous distribution within the monolayer (Fig 4A¢). and HC was further con®rmed by Western blot analysis (Fig 3a). The increased localization of the tight junctional proteins at the Treatment with Bt2 and HC led to decreased protein levels of N- endothelial cell-to-cell contacts after Bt2 and HC treatment was cadherin and lessened the association of this cell adhesion molecule accompanied by an increased association of these components with with the cytoskeleton by 2-fold as shown by densitometry (data not the cytoskeleton, which could be clearly demonstrated after shown). differential extraction of endothelial monolayers with digitonin and detergents (Fig 3a). First, the soluble components of the Localization and cytoskeletal association of tight junction cytoplasm were released with 50 mg per ml digitonin. The associated constituents after treatment with Bt2 and remaining pellets were extracted with buffer A (25 mM Tris± HC Con¯uent HDMEC and HUVEC were treated with Bt2 HCl, pH 7.4, 2% Nonidet P-40) or buffer B (PBS with 1% Triton and HC for 3 d whereas controls were maintained in basal X-100, 0.5% DOC, 0.005% SDS, and 1 mM EDTA) to release medium. Claudin-1 and claudin-5 were present in scattered those components that were weakly associated with the cytoske- endothelial cells (Fig 4A). Junctional labeling for ZO-1 and ZO- leton. The tightly associated constituents remained in the residual 2 was focal and discontinuous (Fig 4B, D, E). Occludin was barely pellets. The increased association of ZO-1, ZO-2, and occludin present at the cell-to-cell contacts in basal conditions (Fig 4C). In with the cytoskeleton was most obvious when comparing the contrast, Bt2- and HC-treated endothelial cell cultures displayed residual pellets after the basal condition (0) with those after Bt2 and regularly delineated cell-to-cell contacts. The junctions showed HC treatment (+) (Fig 3a, compare lane 7 with lane 8 or lane 9 with lane 10). The increased association of ZO-1 with the cytoskeleton was observed after extraction with buffer A whereas that of ZO-2 and occludin was demonstrated after extraction with buffers A and B. Densitometry revealed that the levels of ZO-1 and ZO-2 in the insoluble fraction increased by a factor of 2 and occludin by a factor of 3 (data not shown). Additionally, the observed tightening of the endothelial cell-to- cell contacts had a functional effect on HDMEC and HUVEC as the formation of a signi®cantly increased permeability barrier to the tracer molecule biotin-dextran could be observed after Bt2 and HC treatment (Fig 3b) (p < 0.0001). The permeability assays furthermore revealed that HUVEC had an intrinsically higher capability for barrier formation than HDMEC. This may re¯ect vessel-type- speci®c differences between these endothelial cells due to their origin from dermal microvessels (HDMEC) in comparison to large veins (HUVEC). Endothelial junctional differentiation in the presence of smooth-muscle-cell-like 10T1/2 cells We established cocultures between HUVEC and 10T1/2 cells in order to

Figure 3. Cytoskeletal association of tight junctional constituents in endothelial cells during junctional differentiation and the formation of an increased permeability barrier. (a) Detergent soluble and insoluble fractions of HUVEC were separated by 8% SDS- PAGE followed by Western blot analysis. Lanes 1, 2, 3, 7, and 9 represent basal conditions; lanes 4, 5, 6, 8, and 10 represent Bt2 and HC treatment. Lanes 1 and 4 show cell lysates with 50 mg per ml digitonin, lanes 2 and 5 extracts with buffer A (25 mM Tris±HCl, pH 7.4, 2% Nonidet P-40), and lanes 3 and 6 extracts with buffer B (PBS with 1% Triton X-100, 0.5% DOC, 0.005% SDS, and 1 mM EDTA). The insoluble pellets after extraction with buffer A are shown in lanes 7 and 8; the insoluble pellets after extraction with buffer B in lanes 9 and 10. Treatment with Bt2 and HC resulted in decreased levels of N-cadherin in the detergent soluble fractions (compare lane 2 with lane 5 and lane 3 with lane 6) as well as in the detergent insoluble fractions (compare lane 7 with 8 and lane 9 with 10). In contrast, ZO-1, ZO-2, and occludin became more closely associated with the cytoskeleton after Bt2 and HC treatment, which was most obvious after extraction with buffer A (compare lane 7 with lane 8). Gel loading was controlled by vimentin levels. (b) Permeability assays for the tracer molecule biotin-dextran were performed on con¯uent monolayers of HUVEC and HDMEC in Transwell chambers 3 d after treatment with basal conditions or Bt2 and HC. Biotin-dextran (200 mg per ml) was added to the top compartment and aliquots were taken from the bottom compartment after 40 min to be analyzed by regular ELISA on streptavidin-coated microtiter plates. Bt2 and HC signi®cantly increased the permeability barrier in both endothelial cell lines (gray bars) compared to basal conditions (white bars) (p < 0.0001). HUVEC showed an intrinsically higher ability for barrier formation than HDMEC in basal conditions as well as after in vitro differentiation. VOL. 119, NO. 1 JULY 2002 TIGHTENING OF ENDOTHELIAL CELL CONTACTS 147

Figure 4. Increased localization of tight junctional constituents at sites of cell-to-cell contact after in vitro maturation of HUVEC. Con¯uent HUVEC after 3 d in basal conditions (A±E) and after 3 d of treatment with Bt2 and HC (A¢±E¢) were ®xed in methanol. Immuno¯uorescence labeling was performed for claudin-1 (A, A¢), claudin-5 (B, B¢), occludin (C, C¢), ZO-1 (D, D¢), and ZO-2 (E, E¢). In basal conditions, claudin-1 (A) and claudin-5 (B) were barely detected at the cell-to-cell contacts whereas occludin (C), ZO-1 (D), and ZO-2 (E) were clearly present in a punctate and discontinuous pattern. Treatment with Bt2 and HC led to an increased junctional localization of claudin-1 (A¢) and claudin-5 (B¢). In addition, junctional localization of occludin (C¢), ZO-1 (D¢), and ZO-2 (E¢) became even more prominent and changed to a continuous appearance that was most noticeable for ZO-1 (D¢). Scale bars: 50 mm. investigate whether the presence of smooth-muscle-like cells of endothelial and 10T1/2 cells in comparison to their mixed would in¯uence endothelial junctional differentiation. First, we coculture by Western blot analysis (Fig 5a, b). Cocultures were analyzed the production of junctional components in monocultures established of HUVEC to 10T1/2 cells at a ratio of 2:1. 148 KURZEN ET AL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

Figure 5. Adherens and tight junctional proteins in monocultures of HUVEC and 10T1/2 cells and their mixed coculture. (a) Adherens junctional proteins were analyzed in total cell lysates of HUVEC in monoculture (1, 1¢), 10T1/2 cells in monoculture (2, 2¢), and mixed cocultures of HUVEC and 10T1/2 cells (3, 3¢). Samples were separated by 8% SDS-PAGE and blotted onto polyvinylidine ¯uoride (PVDF) membranes (for Coomassie blue staining see b). Endothelial cell monocultures (1) and mixed cocultures (3) were positive for the endothelial-cell-speci®c VE-cadherin and for plakoglobin whereas 10T1/2 cell monocultures (2) were negative for these proteins. Conversely, 10T1/2 cell monocultures and mixed cocultures were positive for the smooth muscle marker protein a-SMC-actin, which was not detected in HUVEC monocultures. N-cadherin seemed to be more abundant in 10T1/2 cells than in HUVEC whereas the junctional proteins p120, a-catenin, and b-catenin were present in all monocultures and cocultures. (b) Tight junction associated proteins were analyzed in total cell lysates as described for (a). Staining of the PVDF membrane after blotting with 0.1% Coomassie Brilliant Blue in methanol is shown as a control for equal gel loading (1¢, 2¢, 3¢). Endothelial monocultures (1) and mixed cocultures (3) were positive for the tight junction associated proteins ZO-1, ZO-2, and occludin whereas 10T1/2 cell monocultures (2) were weakly positive for ZO-1 and negative for ZO-2 and occludin.

Monocultures of either HUVEC and 10T1/2 cells had cell surrounded by 10T1/2 cells (arrows in Fig 6; asterisks in Fig 7). The numbers that were equal to the total cell count of the cocultures. islands of HUVEC were identi®ed by labeling for VE-cadherin Gel loading was normalized to total cell numbers and was (Fig 6A¢) and were characterized by compacted endothelial cells. controlled by Coomassie blue staining of PVDF membranes after The endothelial cell-to-cell contacts showed prominent and blotting or of polyacrylamide gels after SDS-PAGE. continuous labeling for VE-cadherin (Fig 6A¢), plakoglobin HUVEC were positive for plakoglobin and the endothelial (Fig 6B¢), ZO-1 (Fig 6C¢), and ZO-2 (Fig 6D¢). In addition, marker protein VE-cadherin, neither of which were detected in occludin was almost absent from HUVEC in monoculture 10T1/2 monocultures (Fig 5a). In contrast, 10T1/2 cells were (Fig 7A) but was detected at cell-to-cell contacts of HUVEC in positive for a-SMC-actin, which veri®ed their smooth muscle cell cocultures (arrows in Fig 7B) where it codistributed with VE- phenotype (Fig 5a). In addition, 10T1/2 cell monocultures cadherin (arrows in Fig 7C). produced high levels of N-cadherin (Fig 5a). Other components Endothelial monocultures displayed junctional codistribution of of adherens junctional complexes, a-catenin, b-catenin, and p120, VE- and N-cadherin (Fig 8A±C; arrowheads in Fig 8A). Focally, were produced in both HUVEC and 10T1/2 monocultures. With N-cadherin appeared to be even more prominent than the VE- respect to tight junctional proteins in monocultures, ZO-1, ZO-2, cadherin (arrows in Fig 8A). After cocultivating HUVEC with and occludin were almost absent from 10T1/2 cells but clearly 10T1/2 cells, however, VE-cadherin could be detected at the sites present in HUVEC (Fig 5b). The reduction in protein levels of endothelial cell-to-cell contacts whereas N-cadherin was absent observed in cocultures may result from the reduced cell numbers of (arrowheads in Fig 8D). In these cultures, N-cadherin was merely HUVEC or 10T1/2 cells in coculture in comparison to associated with contacting 10T1/2 cells (arrows in Fig 8D). These monocultures. As plakoglobin, ZO-1, and occludin were basically results indicate that N-cadherin became downregulated at the absent from 10T1/2 cells, these proteins were utilized to further endothelial cell-to-cell contacts after cocultivation with mesench- analyze the junctional differentiation of HUVEC in the presence of ymal 10T1/2 cells. the smooth-muscle-like 10T1/2 cells. To functionally analyze the effect of smooth-muscle-like Immuno¯uorescence studies were performed in parallel on mesenchymal cells on the tightness of the endothelium, we HUVEC in monocultures and in cocultures with 10T1/2 cells measured the transendothelial ¯ux of the small tracer molecule (Figs 6±8). Both cultures were ®xed and analyzed at subcon¯uence biotin-dextran in HUVEC monocultures and cocultures with 3 d after establishing the cultures. Otherwise, 10T1/2 cells would 10T1/2 cells. We found that these cocultures showed a signi®cantly have overgrown the endothelial cells in cocultures. HUVEC increased permeability barrier compared with HUVEC in mono- monocultures showed large and extended endothelial cells cultures (p < 0.01) (Fig 9). (Fig 6A±D). VE-cadherin (Fig 6A), ZO-1 (Fig 6C), and ZO-2 In summary, the presence of smooth-muscle-like 10T1/2 cells (Fig 6D) could be detected at sites of endothelial cell-to-cell induced junctional differentiation in endothelial cells, which was contact. The pattern of the staining was focal and discontinuous. phenotypically and functionally similar to the in vitro differentiation Plakoglobin was rarely detected in HUVEC during monoculture after Bt2 and HC treatment. Treatment with Bt2 and HC or (Fig 6B). In contrast, in the presence of smooth-muscle-like coculture with smooth-muscle-like cells induced a differentiated 10T1/2 cells, HUVEC were found in compacted islands that were endothelial phenotype that was characterized by enhanced junc- VOL. 119, NO. 1 JULY 2002 TIGHTENING OF ENDOTHELIAL CELL CONTACTS 149

Figure 6. Compacted endothelial cell islands in cocultures with 10T1/2 cells. Corresponding monocultures of HUVEC (A±D) and cocultures between HUVEC and 10T1/2 cells (A¢±D¢) were ®xed in methanol and labeled for VE-cadherin (A, A¢), plakoglobin (B, B¢), ZO- 1 (C, C¢), and ZO-2 (D, D¢). In endothelial monocultures, the endothelial cell bodies were large and spread out, and gaps were apparent between the endothelial cells. At the sites of endothelial cell-to-cell contacts, focal signals for VE-cadherin (A), ZO-1 (C), and ZO-2 (D) could be observed whereas plakoglobin (B) could barely be detected. In cocultures with 10T1/2 cells, the HUVEC became visible after staining for VE-cadherin as islands of compacted cells with dense and uninterrupted cell-to-cell contacts (A¢). In contrast to the monocultures, the contacting zones between endothelial cell islands in coculture displayed prominent and continuous signals for plakoglobin (B¢), ZO-1 (C¢), and ZO-2 (D¢). The arrows point at 10T1/2 cells in cocultures. Scale bar: 50 mm. tional localization and tighter cytoskeletal association of plakoglobin, ZO-1, and occludin, thus functionally leading to an increased Figure 7. Occludin displayed a junctional localization in compacted endothelial cell islands after coculture with 10T1/2 barrier formation. cells. Immuno¯uorescence labeling of HUVEC in monocultures showed a diffuse staining for occludin (A), which was not detectable at sites of DISCUSSION cell-to-cell contact. In cocultures of HUVEC and 10T1/2 cells, however, occludin could be observed at endothelial cell-to-cell junctions We have established an in vitro model for endothelial differenti- (arrows in B) where it codistributed with VE-cadherin (arrows in ation, which we have exploited to examine the junctional C).10T1/2 cells are designated by asterisks. Scale bar: 50 mm. composition of endothelial cells in the last stages of angiogenesis. The purpose of this study was to analyze the effect of different growth conditions on the tightening of endothelial cell-to-cell contacts and barrier formation. First, we investigated endothelial vated with bovine aortic endothelial cells, changed to a smooth- cell-to-cell contacts after treatment with Bt2 and HC by which we muscle-like phenotype expressing marker proteins of smooth observed an increased recruitment of tight junction molecules to muscle cells such as calponin, a-smooth muscle cell actin (a- sites of endothelial cell-to-cell contact and an increased association SMC-actin) and SMC-myosin (Hirschi et al, 1998). Therefore, we of these molecules with the cytoskeleton. Physiologic relevance for obtained this cell line for establishing cocultures between HUVEC these observations could be found after coculture of human and 10T1/2 cells in order to investigate whether the presence of endothelial cells with mouse mesenchymal 10T1/2 cells. It was smooth-muscle-like cells would in¯uence endothelial junctional previously shown that mesenchymal 10T1/2 cells, when coculti- differentiation. We observed that coculture between endothelial 150 KURZEN ET AL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

Figure 8. Downregulation of N-cadherin at endothelial cell-to-cell contacts in cocultures with mesenchymal 10T1/2 cells. Double immuno¯uorescence labeling for VE-cadherin (A, B, D, E: ``Texas Red'' red ¯uorescence) and for N-cadherin (A, C, D, F: ``FITC'' green ¯uorescence) was performed on HUVEC grown in monoculture (A±C) and after cocultivation with 10T1/2 cells (D±F). Image analysis occurred by laser scanning confocal microscopy. Superimposed ¯uorescent images are shown in (A) for HUVEC monocultures and in (D) for cocultures. VE-cadherin was observed at sites of endothelial cell-to-cell contact between HUVEC in monocultures as well as in cocultures, where it was most prominent (arrowheads in D). Co- distribution of VE- and N-cadherin at the endothelial cell contacts could be detected in monocultures (arrowheads in A); however, yellow staining could not be demonstrated, which would be indicative of true colocalization. Occasionally, strong labeling for N-cadherin was observed in HUVEC monocultures (arrows in A). In contrast, N-cadherin was no longer localized at endothelial cell contacts after cocultivation (arrowheads in D) when it became detectable on contacting 10T1/2 cells instead (arrows in D).

cells and 10T1/2 cells led to a tightening of the endothelial cell-to- cellular calcium increase endothelial permeability (Alexander et al, cell contacts, which was also con®rmed on a functional level by 1993; Koch et al, 2000). comparing the permeability of cocultures to monocultures for the VE-cadherin is the only cadherin reliably localized to endothelial tracer molecule biotin-dextran. Thus, treatment with Bt2 and HC junctions. It is required for vascular morphogenesis, as can be caused similar changes to the endothelial phenotype as were concluded from VE-cadherin-de®cient mouse embryos, which observed after coculture with 10T1/2 cells. show severe defects of the developing vasculature (Carmeliet et al, The majority of transendothelial solute exchange occurs 1999; Gory-Faure et al, 1999). paracellularly at the interendothelial junctions (Shasby et al, 1982; In the early phase of contact formation, VE-cadherin assembles Franke et al, 1988; Alexander et al, 1993; Schnittler, 1998; cytoplasmic b-catenin at the junction and induces contact to the Vestweber, 2000). Our results indicate that the tightness of the actin cytoskeleton (Halama et al, 1999); in later stages plakoglobin endothelial barrier is mediated through the degree of interdigitation becomes predominantly associated with the adherens junctional between endothelial cells and the stability of the endothelial complexes (Dejana, 1996; Schnittler et al, 1997; Schnittler, 1998). junctions. Previous studies have shown that a tight association of Another member of the cadherin family, N-cadherin, was cloned junctional molecules with the cytoskeleton is required for the from endothelial cells (Salomon et al, 1992). Although N-cadherin- stability of interendothelial adhesion (Lampugnani et al, 1995; de®cient mouse embryos fail to form blood islands (Radice et al, Dejana et al, 1999). Many physiologic agents have been shown to 1997), a functional role for this cadherin in the endothelium is still modulate the permeability of the endothelial barrier. Treatments missing. N-cadherin is at least transiently required for the that increase endothelial cAMP (isoproterenol, Bt2) produce a reformation of endothelial junctions after cell culture in low decrease in permeability, whereas treatments that increase intra- calcium (Alexander et al, 1993). VOL. 119, NO. 1 JULY 2002 TIGHTENING OF ENDOTHELIAL CELL CONTACTS 151

support previous ®ndings by Kevil et al (1998) who could demonstrate vessel-type-speci®c differences in vitro for venous and arterial endothelial cells, the latter being much less permeable. Here we show that microvascular endothelial cells are more permeable than endothelial cells derived from large veins. These ®ndings indicate that the intrinsic capability of endothelial cells for the formation of a permeability barrier depends on the vessel of origin. The possible mechanisms by which these differences are maintained in cell culture are part of future investigations. Coculture between endothelial cells and smooth-muscle-like 10T1/2 cells induced a similar differentiated endothelial phenotype to that observed after treatment with Bt2 and HC, namely tightening of the endothelial cell-to-cell contacts and an increased cytoskeletal association of the endothelial junctional proteins plakoglobin, occludin, and ZO-1. Most strikingly, the presence of 10T1/2 cells signi®cantly increased the permeability barrier of Figure 9. Increased barrier formation of HUVEC in the presence the endothelium. We used HUVEC for these cocultures as they are of 10T1/2 cells. Permeability assays of HUVEC and 10T1/2 cells in naturally surrounded by smooth muscle cells within the organism. monoculture and coculture for biotin-dextran in Transwell chambers. Both the endothelium and the mural cells are important for the HUVEC and 10T1/2 cells were cultured on opposite sides of the differentiation of mature and patent blood vessels (Hirschi et al, Transwell ®lter membrane. Controls were set up with HUVEC only or 1999; Nicosia and Villaschi, 1999; Carmeliet, 2000). In our study, 10T1/2 cells only. After 8 d, the permeability of the con¯uent cell layers we show that smooth-muscle-like cells induce a differentiated was analyzed. Biotin-dextran was added to the top compartment and aliquots were taken from the bottom compartment after 3 h to be endothelial phenotype in vitro. It is known from previous studies analyzed by regular ELISA methods (compare Fig 3b). Cocultures of that pericytes and smooth muscle cells inhibit endothelial cell HUVEC with 10T1/2 cells (black bar) had a signi®cantly reduced migration and proliferation in cell culture (Orlidge and d'Amore, permeability for biotin-dextran compared to HUVEC monocultures 1987; Sato and Rifkin, 1989). This inhibitory activity is mediated (light gray bar) (p < 0.01). Monocultures of 10T1/2 cells (white bar) had by transforming growth factor b (TGF-b), which becomes the highest permeability for biotin-dextran. activated when smooth muscle cells or pericytes make contact to endothelial cells (Sato and Rifkin, 1989; Sato et al, 1990). Endothelial cells in turn release platelet derived growth factor BB Support for an early involvement of N-cadherin in the process of (PDGF-BB), which induces pericytes and smooth muscle cells to angiogenesis was also observed during blood±brain barrier matur- migrate towards the endothelium and to make contact (Hirschi et ation. Initially, b-catenin, but not plakoglobin, is codistributed al, 1998, 1999). Furthermore, mesenchymal ®broblastoid 10T1/2 with N-cadherin at the abluminal endothelial membrane at contact cells differentiate into a smooth muscle cell phenotype when they sites to perivascular cells. In contrast, plakoglobin is most prom- make contact to endothelial cells, a process that is inhibited by inent at the interendothelial junctions where only small amounts of blocking antibodies to TGF-b (Hirschi et al, 1998). Thus, cells of b-catenin are present. In contrast to b-catenin and plakoglobin, N- the vessel wall and endothelial cells reciprocally activate mechan- cadherin is completely lost during later stages of development isms that induce vascular differentiation. This is further supported (Liebner et al, 2000). The presence of VE-cadherin has been shown by our ®ndings in this study, which suggest that smooth muscle like cells exert a functionally stabilizing effect on endothelial barrier to exclude N-cadherin from the junctions (Navarro et al, 1998). formation. This is well in agreement with our observation that the induction Further insight into blood vessel maturation and the import- of a differentiated endothelial phenotype seems to increase the ance of the interaction between endothelial and mural cells for junctional association of VE-cadherin whereas lowering that of N- blood vessel maturation stems from studies on gene-de®cient cadherin. This leads to the assumption that VE-cadherin is mice. Mice lacking endoglin, which is a TGF-b binding protein responsible for homotypic cell-to-cell contact, whereas N-cadherin on the surface of endothelial cells, are characterized by severe may account for adhesion to different cell types, like smooth muscle vascular malformations that are embryonic lethal. In these mice, cells or pericytes (Gilbertson-Beadling and Fisher, 1993). In the capillary plexus does not show proper recruitment of a contrast to this assumption, we did not observe any staining for supporting layer of smooth muscle cells (Li et al, 1999). Severe N-cadherin at the sites of cell contact between HUVEC and 10T1/ defects in the establishment of a differentiated vascular network 2 cells. This might be due to the inability of N-cadherins from are also observed in mice lacking either PDGF-B or the different species to form ``homophilic'' interactions. Another PDGF-receptor-b (Lindahl et al, 1997; Hellstrom et al, 1999). explanation could be that the N±cadherin-mediated interaction Again, the capillary plexus lacks supporting mural cells. All between endothelial and mural cells is limited to a very short time these studies furthermore present evidence that the lack of window. supporting cells causes an arrest in endothelial remodeling. The endothelial adherens junction is thought to play a role in A primitive capillary network is developed in mice with a tight junction formation, which restricts and regulates solute homozygous de®ciency in the tie-2 tyrosine kinase receptor (Sato et exchange (Lum and Malik, 1994; Anderson and Van Italie, 1995). al, 1995), in mice with a homozygous de®ciency in the tie-2 ligand Therefore, we analyzed tight junction associated proteins in our angiopoietin-1 (ang-1) (Suri et al; 1996), and in transgenic mice model of endothelial differentiation. We found that tight junction overexpressing its negative ligand ang-2 (Maisonpierre et al, 1997). associated transmembrane proteins, most prominently occludin, The primitive capillary network, however, does not become and associated plaque proteins (ZO-1, ZO-2) were signi®cantly remodeled and blood vessel differentiation does not proceed. The upregulated after treatment with Bt2 and HC. endothelium fails both to adhere tightly to the underlying basement Additionally, we observed that the tightening of the endothelial membrane and to recruit mural cells, thus rendering these mice cell-to-cell contacts had a functional effect on HDMEC and embryonic lethal (for review, see also Hanahan, 1997). HUVEC. The formation of a signi®cantly increased permeability Interestingly, mice with a homozygous disruption of the ephrin- barrier to the tracer molecule biotin-dextran was detected after B2 gene show a similar phenotype to tie-2 or ang-1 null mice treatment with Bt2 and HC (p < 0.0001). The permeability assays (Wang et al, 1998; Adams et al, 1999, 2001). Ephrins are cell surface furthermore revealed that HUVEC had an intrinsically higher bound ligands to Eph receptor tyrosine kinases, which are capability for barrier formation than HDMEC. These results implicated in repulsive axon guidance, cell migration, and 152 KURZEN ET AL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY angiogenesis (Klein, 2001). These mice fail to remodel the primary Hellstrom M, Kal NM, Lindahl P, Abramsson A, Betsholtz C: Role of PDGF-B and capillary plexus, resulting in defective developmental angiogenesis. PDGFR-beta in recruitment of vascular smooth muscle cells and pericytes during embryonic blood vessel formation in the mouse. Development 126:3047± Adams et al (1999) presented evidence that ephrin±Eph interactions 3055, 1999 could be involved in signaling processes between the endothelium Hirschi KK, Rohovsky SA, D'Amore PA: PDGF, TGF-beta, and heterotypic cell± and the adjacent mesenchyme. Furthermore, mice with a cell interactions mediate endothelial cell-induced recruitment of 10T1/2 cells homozygous de®ciency of the tie-1 orphan receptor die shortly and their differentiation to a smooth muscle fate. J Cell Biol 141:805±814, 1998 Hirschi KK, Rohovsky SA, Beck LH, Smith SR, D'Amore PA: Endothelial cells after birth due to edema and severe hemorrhaging, which are also modulate the proliferation of mural cell precursors via platelet-derived growth caused by aberrant vessel integrity (Sato et al, 1995). factor-BB and heterotypic cell contact. Circ Res 84:298±305, 1999 All these ®ndings in gene-de®cient mice describe severe and Jaffe EA, Nachmann RL, Becker CG, Minick RC: Culture of endothelial cells lethal vascular defects due to malfunctioning cell-to-cell signaling derived from human umbilical cord veins. Circulation 46:211±253, 1972 Keon BH, Schafer S, Kuhn C, Grund C, Franke WW: Symplekin, a novel type of between the endothelium and the supporting mural cells. In our tight junction plaque protein. J Cell Biol 134:1003±1018, 1996 study, we describe how smooth-muscle-like cells can induce the Kevil CG, Okayama N, Trocha SD, et al: Expression of zonula occludens and phenotypic differentiation of endothelial cells and the formation of adherens junctional proteins in human venous and arterial endothelial cells: role an increased endothelial barrier. Our data underscore the import- of occludin in endothelial solute barriers. Microcirculations 5:197±210, 1998 Klein R: Excitatory Eph receptors and adhesive ephrin ligands. Curr Opin Cell Biol ance of mural cells for stabilizing the vascular endothelium. 13:196±203, 2001 Therefore, this cell culture system may represent a valuable Koch G, Pratzel S, Rode M, Kraling BM: Induction of endothelial barrier function model to analyze mediators of vascular differentiation that are in vitro. Ann NY Acad Sci 915:123±128, 2000 involved in heterotypic signaling between endothelial and mural Kowalczyk AP, Navarro P, Dejana E, Bornslaeger EA, Green KJ, Kopp DS, Borgwardt JE: VE-cadherin and desmoplakin are assembled into dermal cells within a controlled cell culture environment, which will be microvascular endothelial intercellular junctions: a pivotal role for plakoglobin part of our future studies. in the recruitment of desmoplakin to intercellular junctions. J Cell Sci 111:3045±3057, 1998 KraÈling BM, Bischoff J: A simpli®ed method for growth of human dermal microvascular endothelial cells results in decreased senescence and continued This work was supported by the Deutsche Forschungsgemeinschaft (grant Kr 1985/ responsiveness to cytokines and growth factors. In Vitro Cell Dev Biol Anim 1-1). We thank Professor W.W. Franke, German Cancer Research Center, for 33:308±315, 1998 support and valuable discussions. KraÈling BM, Wiederschain DG, Boehm T, Rehn M, Mulliken JB, Moses MA: The role of matrix metalloproteinase activity in the maturation of human capillary endothelial cells in vitro. J Cell Sci 112:1599±1609, 1999 Lampugnani MG, Dejana E: Interendothelial junctions: structure, signaling and REFERENCES functional roles. Curr Opin Cell Biol 9:674±682, 1997 Lampugnani MG, Corada M, Caveda L, Breviario F, Ayalon O, Geiger B, Dejana E: Adams CL, Nelson WJ: Cytomechanics of cadherin-mediated cell±cell adhesion. The molecular organization of endothelial cell to cell junctions: differential Curr Opin Cell Biol 10:572±577, 1998 association of plakoglobin, b-catenin, and a-catenin with vascular endothelial Adams RH, Wilkinson GA, Weiss C, et al: Roles of ephrinB ligands and EphB cadherin (VE-cadherin). J Cell Biol 129:203±217, 1995 receptors in cardiovascular development: demarcation of arterial/venous Leach LP, Clark MG, Lampugnani AG, Arroyo E, Dejana E, Firth JA: domains, vascular morphogenesis, and sprouting angiogenesis. Genes Dev Immunoelectron characterization of the inter-endothelial junctions of the 13:295±306, 1999 human term placenta. J Cell Sci 104:1073±1081, 1993 Adams RH, Diella F, Hennig S, Helmbacher F, Deutsch U, Klein R: The Li DY, Sorensen LK, Brooke BS, et al: Defective angiogenesis in mice lacking cytoplasmic domain of the ligand ephrinB2 is required for vascular endoglin. Science 284:1534±1537, 1999 morphogenesis but not cranial neural crest migration. Cell 104:57±69, 2001 Liebner S, Gerhardt H, Wolburg H: Differential expression of endothelial b-catenin Alexander JS, Blaschuk OW, Haselton FR: An N-cadherin-like protein contributes and plakoglobin during development and maturation of the blood±brain and to solute barrier maintenance in cultured endothelium. J Cell Phys 156:610± blood±retina barrier in the chicken. Dev Dyn 217:86±98, 2000 618, 1993 Lindahl P, Johansson BR, Leveen P, Betsholtz C: Pericyte loss and microaneurysm Anderson JM, Van Italie CM: Tight junctions and the molecular basis for regulation formation in PDGF-B-de®cient mice. Science 277:242±245, 1997 of paracellular permeability. Am J Physiol 269:G467±G475, 1995 Lum H, Malik AB: Regulation of vascular endothelial barrier function. Am J Physiol Balda MS, Matter K: Tight junctions. J Cell Sci 111:541±547, 1998 11:L223±L241, 1994 Carmeliet P: Mechanisms of angiogenesis and arteriogenesis. Nat Med 6:389±395, Maisonpierre PC, Suri C, Jones PF, et al: Angiopoietin-2, a natural antagonist for 2000 Tie2 that disrupts in vivo angiogenesis. Science 277:55±60, 1997 Carmeliet P, Lampugnani MG, Moons L, et al: Targeted de®ciency or cytosolic Morita K, Sasaki H, Furuse M, Tsukita S: Endothelial claudin: claudin-5/TMVCF truncation of the VE-cadherin gene in mice impairs VEGF-mediated constitutes tight junction strands in endothelial cells. J Cell Biol 147:185±194, endothelial survival and angiogenesis. Cell 98:147±157, 1999 1999 Dejana E: Endothelial adherens junctions: implications in the control of vascular Navarro P, Ruco L, Dejana E: Differential localization of VE- and N-cadherins in permeability and angiogenesis. J Clin Invest 98:1949±1953, 1996 human endothelial cells: VE-cadherin competes with N-cadherin for junctional Dejana E, Bazzoni G, Lampugnani MG: Vascular endothelial (VE) cadherin: only an localization. J Cell Biol 140:1475±1484, 1998 intercellular glue? Exp Cell Res 252:13±19, 1999 Nicosia RF, Villaschi S: Autoregulation by cells of the vessel wall. Int Rev Cyt 185:1± Farquhar MG, Palade GE: Junctional complexes in various epithelia. J Cell Biol 43, 1999 17:375±411, 1963 Oliver MH, Harison NK, Bishop JE, Cole PJ, Laurent GJ: A rapid and convenient Franke WW, Cowin P, Grund C, Kuhn C, Kapprell HP: The endothelial junction. assay for counting cells cultured in microwell plates: application for assessment The plaque and its components. In: Simionescu N, Simionescu M, eds. of growth factors. J Cell Sci 92:513±518, 1989 Endothelial Cell Biology in Health and Disease. New York: Plenum Press, 1988:pp Orlidge A, D'Amore PA: Inhibition of capillary endothelial cell growth by pericytes 147±166 and smooth muscle cells. J Cell Biol 105:1455±1462, 1987 Furuse M, Sasaki H, Fujimoto K, Tsukita S: A single gene product, claudin-1 or -2, Radice GL, Rayburn H, Matsunami H, Knudsen KA, Takeichi M, Hynes RO: reconstitutes tight junction strands and recruits occludin in ®broblasts. J Cell Developmental defects in mouse embryos lacking N-cadherin. Dev Biol Biol 143:391±401, 1998 181:64±78, 1997 Gilbertson-Beadling SK, Fisher C: A potential role for N-cadherin in mediating Risau W: Differentiation of endothelium. FASEB J 9:926±933, 1995 endothelial cell±smooth muscle cell interactions in the rat vasculature. Lab Risau W: Mechanisms of angiogenesis. Nature 386:671±674, 1997 Invest 69:203±209, 1993 Salomon D, Ayalon O, Patel-King R, Hynes RO, Geiger B: Extrajunctional Gory-Faure S, Prandini MH, Pointu H, Roullot V, Pignot-Paintrand I, Vernet M, distribution of N-cadherin in cultured human endothelial cells. J Cell Sci Huber P: Role of vascular endothelial-cadherin in vascular morphogenesis. 102:7±17, 1992 Development 126:2093±2102, 1999 Sato TN, Tozawa Y, Deutsch U, et al: Distinct roles of the receptor tyrosine kinases Grant DS, Lelkes PI, Fukuda K, Kleinman HK: Intracellular mechanisms involved in Tie-1 and Tie-2 in blood vessel formation. Nature 376:70±74, 1995 basement membrane induced blood vessel differentiation in vitro. In Vitro Cell Sato Y, Rifkin DB: Inhibition of endothelial cell movement by pericytes and smooth Dev Biol 27A:327±336, 1991 muscle cells: activation of latent transforming growth factor-b1-like molecule Halama T, Staf¯er G, Hoch S, Stockinger H, Wolff K, Petzelbauer P: Vascular- by plasmin during co-culture. J Cell Biol 109:309±315, 1989 endothelial cadherin (CD144)- but not PECAM-1 (CD31)-based cell-to-cell Sato Y, Tsuboi R, Lyons R, Moses H, Rifkin RB: Characterization of the activation contacts convey the maintenance of a quiescent endothelial monolayer. Int Arch of latent TGF-b by co-cultures of endothelial cells and pericytes of smooth Allergy Immunol 120:237±244, 1999 muscle cells: a self regulating system. J Cell Biol 111:757±763, 1990 Hanahan D: Signaling vascular morphogenesis and maintenance. Science 277:48±50, Schmelz M, Moll R, Kuhn C, Franke WW: Complexus adhaerentes, a new group of 1997 desmoplakin-containing junctions in endothelial cells. II. Different types of Heid HW, Moll I, Franke WW: Patterns of expression of trichocytic and epithelial lymphatic vessels. Differentiation 57:97±117, 1994 cytokeratins in mammalian tissues. I. Human and bovine hair follicles. Schmidt A, Heid HW, Schafer S, Nuber UA, Zimbelmann R, Franke WW: Differentiation 37:137±157, 1988 Desmosomes and cytoskeletal architecture in epithelial differentiation: cell VOL. 119, NO. 1 JULY 2002 TIGHTENING OF ENDOTHELIAL CELL CONTACTS 153

type-speci®c plaque components and intermediate ®lament anchorage. Eur J Troyanovsky SM, Eshkind LG, Troyanovsky RB, Leube RE, Franke WW: Cell Biol 65:229±245, 1994 Contribution of cytoplasmic domains of desmosomal cadherins to Schnittler HJ, Puschel B, Drenckhahn D: Role of cadherins and plakoglobin in desmosome assembly and intermediate ®lament anchorage. Cell 72:561±574, interendothelial adhesion under resting conditions and shear stress. Am J Physiol 1993 273:H2396±H2405, 1997 Vestweber D: Molecular mechanisms that control endothelial contacts. J Pathol Schnittler HJ: Structural and functional aspects of intercellular junctions in vascular 190:281±291, 2000 endothelium. Basic Res Cardiol 93:30±39, 1998 Wang HU, Chen ZF, Anderson DJ: Molecular distinction and angiogenic interaction Shasby DM, Shasby SS, Sullivan JM, Peach MJ: The role of the endothelial between embryonic arteries and veins revealed by ephrin-B2 and its receptor cytoskeleton in the control of endothelial permeability. Circ Res 51:657±661, Eph-B4. Cell 93:741±753, 1998 1982 Watabe-Uchida M, Uchida N, Imamura Y, et al: Alpha±Catenin±vinculin Suri C, Jones PF, Patan S, et al: Requisite role of angiopoietin-1, a ligand for the interaction functions to organize the apical junctional complex in epithelial TIE2 receptor, during embryonic angiogenesis. Cell 87:1171±1180, 1996 cells. J Cell Biol 142:847±857, 1998 Takeichi M: Morphogenetic roles of classic cadherins. Curr Opin Cell Biol 7:619±627, Wilting J, Christ B: Embryonic angiogenesis: a review. Naturwissenschaften 83:153± 1995 164, 1996