Generation and characterization of VEC-αC-ΔVBD and VEC-αC-swapVBD knock-in mouse lines

To test the relevance of constitutive binding of vinculin to VEC-αC for endothelial junction stabilization in vivo, we generated two equivalent knock-in mouse lines in which the VBD of VEC-αC was either deleted or modified, thereby preventing the association of VEC-αC with vinculin (Fig. 2A). In one case, most of the M1 domain (amino acid 301–401) containing the VBD was deleted (VEC-αC-ΔVBD). To avoid too drastic structural changes of the original VEC-αC chimera, an alternative construct was generated by swapping most of the M1 domain for the corresponding domain of vinculin (VEC-αC-swapVBD). Importantly, this vinculin domain is unable to bind to vinculin (Huveneers et al., 2012). The two mutated VEC-αC knock-in mouse lines were successfully established using recombinase-mediated cassette exchange (RMCE) technology (Schulte et al., 2011), replacing endogenous VE-cadherin by the VEC-αC mutants (Fig. 2B). Homozygous mutant pups were borne for each of the two new mouse lines at sub-Mendelian levels, pointing to partial embryonic lethality (Table S1), similar to what was found for the original VEC-αC mice (Schulte et al., 2011).

Open in a separate window

Fig. 2.

Characterization of the VEC-αC-ΔVBD and VEC-αC-swapVBD constructs. (A) Schematic illustration of VE-cadherin, β-catenin, α-catenin, and VEC-αC, VEC-αC-ΔVBD and VEC-αC-swapVBD fusion proteins. VEC-αC consists of a truncated form of VE-cadherin lacking the β-catenin-binding site (C-terminal 75 aa), which was fused to the C-terminal part (aa 301–906) of α-catenin. VEC-αC-ΔVBD and the VEC-αC-swapVBD are both unable to associate with vinculin and were constructed by modifying VEC–αC. The vinculin binding domain of α-catenin (aa 301–401) was deleted in VEC-αC-ΔVBD or was replaced by a homologous domain from mouse vinculin (green) in VEC-αC-swapVBD. Numbers refer to amino acid positions. (B) Targeting strategy for the generation of VEC–αC mutant knock-in mice by recombinase-mediated cassette exchange (RMCE). (C) MDMVECs were isolated from the four knock-in mouse lines (as indicated) and lysates were immunoblotted for VE-cadherin and α-tubulin (representative of n=2 independent experiments).

To characterize the modified VEC-αC mutants in detail, primary ECs were isolated from the skin of both mouse lines and lysates of these cultured cells were subjected to western blotting for VE-cadherin. As shown in Fig. 2C, both mutants completely replaced endogenous VE-cadherin and showed the expected size on SDS-PAGE gels. Specifically, VEC-αC-swapVBD displayed the same apparent molecular mass as VEC-αC, whereas VEC-αC-ΔVBD was slightly smaller.

Next, we compared the junctional expression of all three VEC-αC versions and their ability to recruit vinculin. Immunostaining of confluent MDMVECs isolated from WT mice and the three knock-in mouse lines revealed, that the expression level of each VEC-αC chimeric protein at junctions was similar to that of VE-cadherin, and that each of the three chimeras caused some linearization of cell contacts in comparison to what was seen with VE-cadherin (Fig. 3A). Vinculin was much more strongly recruited to junctions in VEC-αC MDMVECs when compared to VE-cadherin, whereas no junctional vinculin recruitment was observed in VEC-αC-ΔVBD or VEC-αC-swapVBD MDMVECs (Fig. 3A,B). In line with this, we found in immunoblots that neither VEC-αC-ΔVBD nor VEC-αC-swapVBD could be efficiently co-immunoprecipitated with anti-vinculin antibodies from these cultured MDMVECs, whereas VE-cadherin was weakly and VEC-αC was strongly co-precipitated (Fig. 3C,D). Thus, the VEC-αC-ΔVBD and the VEC-αC-swapVBD chimeras are properly expressed at junctions, but lose the ability to recruit vinculin.

Open in a separate window

Fig. 3.

Impaired vinculin recruitment by VEC-αC-ΔVBD and VEC-αC-swapVBD. (A) MDMVECs from the indicated mouse lines were fixed and stained for the indicated antigens. Scale bars: 25 µm. (B) Quantification of vinculin signal intensities at cell contacts relative to VE-cadherin signal intensities as shown in A (n=3 independent experiments). (C) Anti-vinculin and control (IgG) immunoprecipitates of MDMVECs from the four indicated knock-in mouse lines (IP) or complete cell lysates (lysates; 2% of input) were immunoblotted for VE-cadherin, vinculin or α-tubulin. Positions of molecular mass markers are indicated on the right. (D) Quantification of VE-cadherin signal relative to vinculin signal intensities as shown in C (n=3 independent experiments). Results are shown as means±s.e.m. *P≤0.05; **P≤0.01; ***P≤0.001 (one-way ANOVA with Tukey's multiple comparisons test).

Constitutive binding of VEC-αC to vinculin is irrelevant for junction stabilization in VEC-αC mice

Next, we compared the ability of the different VEC-αC constructs to block the induction of vascular permeability in vivo. To this end, we performed Miles assays with each of the four mouse lines (VEC-WT, VEC-αC, VEC-αC-ΔVBD and VEC-αC-swapVBD) by injecting Evans Blue intravenously and stimulating intradermally with histamine. After 30 min, permeability was measured by extraction of the leaked dye from the excised skin area with formamide. In line with our previous study (Schulte et al., 2011), the strong histamine-induced increase of permeability that we observed in VEC-WT mice was completely blocked in VEC-αC mice (Fig. 4). Interestingly, VEC-αC-ΔVBD blocked the induction of vascular permeability only partially, reducing the increase observed in VEC-WT mice by 63% (Fig. 4, left panel). In contrast, VEC-αC-swapVBD blocked the induction of vascular permeability as completely as did VEC-αC (Fig. 4, right panel). Thus, deleting the VBD of VEC-αC partially impaired the junction stabilizing effect of the chimera, whereas replacing this domain by a homologous, non-vinculin-binding domain left the junction-stabilizing effect intact. This excludes a role for vinculin in the junction stabilizing effect of VEC-αC in vivo.

Open in a separate window

Fig. 4.

Induction of vascular permeability is completely blocked in VEC-αC and in VEC-αC-swapVBD, but only partially blocked in VEC-αC-ΔVBD knock-in mice. VEC-WT, VEC-αC, and either VEC-αC-ΔVBD mice (left graph) or VEC-αC-swapVBD mice (right graph) were intravenously injected with Evans Blue, followed by intradermal injection of histamine or PBS after 15 min. At 30 min after injection, skin areas were excised and the dye was extracted and quantified. Data are pooled from at least two independent experiments with seven to ten mice per group. Results are shown as means±s.e.m. *P≤0.05; **P≤0.01; ****P≤0.0001 (two-way ANOVA with Tukey's multiple comparisons test). OD620, optical density at 620 nm.

The unexpected differences between the effects of VEC-αC-ΔVBD and VEC-αC-swapVBD on junction stabilization prompted the idea that the complete deletion of aa 301–401 in the VEC-αC-ΔVBD may have caused allosteric structural changes that reduced actin binding in a vinculin-independent way, whereas the more subtle replacement of this domain by a non-vinculin-binding domain left the strong constitutive binding to actin intact. To test this, we compared detergent extractability of the different constructs from primary MDMVECs. Upon extraction of MDMVEC monolayers with a mild detergent buffer (soluble fraction), insoluble residual material was re-extracted with an SDS-containing buffer (insoluble fraction) and the two fractions were analyzed by immunoblotting. As shown in Fig. 5, 31% of VE-cadherin, but 59% of VEC-αC, was found in the insoluble fraction. In contrast, only 41% of VEC-αC-ΔVBD was present in the insoluble fraction versus 53% of the VEC-αC-swapVBD chimera. Thus, VEC-αC-ΔVBD is less resistant to detergent extraction than VEC-αC, in line with its less efficient junction stabilization. In contrast, VEC-αC-swapVBD shows similar resistance to detergent extractability to that of VEC-αC and is as efficient in junction stabilization. We therefore conclude that both junction stabilization and resistance to detergent extractability of VEC-αC do not require vinculin anchorage.

Open in a separate window

Fig. 5.

Deleting the VBD in VEC-αC, but not replacing it enhances detergent extractability. (A) To separate the Triton-soluble (sol) from the Triton-insoluble (is) fraction of the various forms of VE-cadherin, confluent MDMVECs were incubated with Triton buffer for 10 min (soluble fraction), followed by lysing the residual monolayers in a harsh SDS-containing lysis buffer (insoluble fraction). Both fractions were subjected to western blotting for the indicated antigens. (B) Quantification of soluble/insoluble fractions from four independent experiments. Results are shown as mean±s.e.m. *P≤0.05; **P≤0.01; ***P≤0.001 (one-way ANOVA with Tukey's multiple comparisons test).

Changes in the conformation of the ABD of α-catenin in VEC-αC

Next, we wanted to determine whether conformational changes in the ABD of VEC-αC could be found, which might be relevant for its enhanced interaction with the actin cytoskeleton in endothelial cells. Based on SMD simulations of αN-catenin, it was recently suggested that a short N-terminal α1-helix of the ABD (in αE-catenin residues 669–675) unfolds upon simulating pulling forces (Ishiyama et al., 2018). Furthermore, inserting a mutation in this helix that impaired its folding, resulted in a recombinant form of the ABD, which showed enhanced binding to F-actin (Ishiyama et al., 2018). These results suggested that catch bonding of α-catenin to F-actin may be accompanied by unfolding of the α1-helix. Based on this concept, we speculated that the α1-helix might be constitutively unfolded in VEC-αC, explaining enhanced F-actin binding of the chimera in cells. To test this, we generated a rabbit polyclonal antibody (VD7) against a peptide covering amino acids 665 to 678 of human and mouse αE-catenin, which contained the 7 amino acids of the short α1-helix (Fig. 6A). Antibodies from this serum specifically recognize α-catenin in western blots (Fig. S1).

Open in a separate window

Fig. 6.

Constitutive unfolding of the α1-helix of α-catenin in junctional VEC-αC. (A) Peptide sequence of α-catenin used to generate the VD7 antibodies. The part covering the α1-helix is indicated. (B) Confluent MDMVECs were fixed, permeabilized, and stained with VD7, total anti-α-catenin or anti-VE-cadherin antibodies. Scale bars: 30 μm. (C) Quantification of VD7 signal intensities at cell contacts relative to total α-catenin signal intensities as shown in B (n=3 independent experiments). Results are shown as means±s.e.m. *P≤0.05 (unpaired two-tailed Student's t-test).

Using these antibodies, we stained cultured primary MDMVECs isolated from VEC-WT or VEC-αC mice. The same cells were co-stained with antibodies against VE-cadherin and antibodies that recognized total α-catenin independent of its conformation. As shown in Fig. 6B, the staining intensity for VE-cadherin and α-catenin was similar in both types of cells. In contrast, staining with VD7 was much stronger for VEC-αC-expressing cells than for VEC-WT-expressing cells (Fig. 6B). Quantification of VD7 signal intensities at cell contacts relative to total α-catenin signal intensities showed a significant increase in the VD7 staining in VEC-αC cells compared to VEC-WT cells (Fig. 6C). These results reveal that the epitopes recognized by the VD7 antibodies are much more accessible in VEC-αC than in normal α-catenin recruited to endothelial junctions. To rule out that staining might have been based on potential epitopes adjacent to the 7-amino-acid helix, we affinity purified antibodies from the serum with a peptide lacking the amino acids adjacent to this helix (Fig. S2A). With these antibodies, we again found strongly enhanced staining at junctions of cells expressing VEC-αC compared to that in those expressing VEC-WT (Fig. S2B,C). This indicates, that the α1-helix is unfolded in VEC-αC, which is in line with the concept that the α-catenin ABD in VEC-αC is conformationally modified and resembles the high-affinity conformation of ABD, which is formed when α-catenin is under tension.

The antibody VD7 specifically recognizes α-catenin in a force-dependent manner

Next, we asked whether the VD7 antibody does indeed recognize α-catenin in a force-dependent manner. To this end, HUVECs were stimulated with thrombin in the presence or absence of the myosin II inhibitor Blebbistatin, and then fixed, permeabilized and stained with VD7 and control antibodies against total α-catenin. As shown in Fig. 7A and quantified in Fig. 7B, there was a significant increase in relative signal intensity of VD7 to total α-catenin staining in thrombin-treated HUVECs compared to that in control cells. Furthermore, pre-treatment with Blebbistatin abolished the increase in VD7 staining upon thrombin-treatment, showing that force created by the contractile actomyosin machinery is needed to unfold the α1-helix and allow staining with VD7 antibody.

Open in a separate window

Fig. 7.

The antibody VD7 recognizes α-catenin in a force-dependent manner. (A) Confluent HUVEC monolayers were pre-treated with DMSO only or Blebbistatin for 30 min before treatment with medium only or thrombin for 10 min (as indicated). Cells were fixed, permeabilized and stained with VD7 and total α-catenin antibodies. (B) Quantification of VD7 signal intensities at cell contacts relative to total α-catenin signal intensities as shown in A (n=3 independent experiments). Results are shown as means±s.e.m. *P≤0.05; **P≤0.01 (one-way ANOVA with Tukey's multiple comparisons test). (C) Schematic representation of VE-cadherin tension sensor. (D) Confluent HUVEC monolayers expressing a VE-cadherin tension sensor were treated for 10 min with medium only or thrombin. Cells were fixed, permeabilized, and stained with VD7 and total α-catenin antibodies. The lifetime of YPet in VE-cadherin and the immunofluorescence intensity of VD7 and total α-catenin were acquired. (E) Scatter plot of the average fluorescence ratio (VD7/total α-catenin) and YPet lifetime of different cell–cell junctions. The points presented correspond to regions of interest segmented from a mixture of cells from a single representative experiment of four. Regression line with confidence interval of 95%. Spearman correlation (r) and P-value were calculated by a Spearman correlation test (n=48 values). Scale bars: 20 μm (A); 10 µm (D).

To directly test whether tension across the VE-cadherin–catenin complex in cells coincides  with the unfolding of the α1-helix of α-catenin, we applied fluorescence lifetime imaging (FLIM) and made use of a recently generated VE-cadherin tension sensor (Arif et al., 2021). In this construct, a ferredoxin-like (FL) linker-based Förster resonance energy transfer (FRET) module was inserted between the p120-catenin (encoded by CTNND1) and the β-catenin-binding site of VE-cadherin. The FRET module contains YPet (donor) and mCherry (acceptor) separated by the elastic FL linker (Fig. 7C). Tension induces separation of YPet and mCherry, thereby increasing YPet lifetime. As the increase of VD7 staining in thrombin-treated HUVECs did not equally distribute to all junction areas, we wanted to test whether there is a pixel-by-pixel correlation between tension on VE-cadherin (YPet lifetime) and VD7 staining. To address this, HUVECs expressing the VE-cadherin tension sensor were treated with thrombin and stained with VD7 and for comparison with antibodies against total α-catenin. As shown in Fig. 7D and quantified in Fig. 7E, the ratio of pixel intensity for VD7/total α-catenin staining had a strong correlation with pixel lifetime of the YPet molecule. Thus, unfolding of the α1-helix, as monitored by an increase in VD7 epitope accessibility, coincided with enhanced tension across the VE-cadherin tension sensor.

Collectively, these results demonstrate that the VD7 antibody recognizes epitopes that are buried in the α1-helix of α-catenin and become accessible upon exposure to force.

Mice

VEC-WT and VEC-αC knock-in mice were described previously (Schulte et al., 2011). VEC-αC-ΔVBD or VEC-αC-swapVBD knock-in mice were generated by inserting the appropriate cDNAs (see above) into the VE-cadherin locus via RMCE (Schulte et al., 2011), thereby replacing expression of endogenous VE-cadherin. All mice used in this study were on the mixed 129SV×C57Bl/6 genetic background. All animal studies were approved by Landesamt fuer Natur, Umwelt und Verbraucherschutz Nordrhein-Westfalen, Germany.

Cell culture

Primary ECs from skin (MDMVECs) of VEC-WT, VEC-αC, VEC-αC-ΔVBD or VEC-αC-swapVBD mice were isolated and cultured as previously described (Frye et al., 2015). Endothelioma cells established from VEC-WT or VEC-α-C mice were cultured as previously described (Schulte et al., 2011). Human umbilical vein endothelial cells (HUVECs; Lonza, MD, USA) were cultured in EBM-2 medium supplemented with EGM-2 SingleQuots (Lonza, MD, USA) at 37°C and 5% CO₂. Prior to experiments, cells were authenticated and confirmed to be free of contamination.

Antibodies and reagents

The following antibodies were used (IF, immunofluorescence; WB, western blotting; IP, immunoprecipitation).

VD7 directed against the entire α1-helix sequence of α-catenin and its flanking regions (AGQSARAIMAQLPQ) was obtained by immunization of rabbits and affinity purification using a described method (including an additional N-terminal cysteine for coupling) (Ebnet et al., 2000; IF, 1 µg/ml; WB, 0.5 µg/ml; IP, 5 µg/reaction). To obtain antibodies that only recognized the α1-helix part, antibodies were affinity purified from this serum on the peptide GGGSARAIMAQ coupled with a N-terminal cysteine to SulfoLink resin (Thermo Fisher Scientific, MA, USA) (IF, 1 µg/ml). Rabbit polyclonal antibodies VE-42 (rab pAb) against mouse VE-cadherin have been described previously (Broermann et al., 2011) (WB, 0.5 µg/ml).

The following antibodies were commercially obtained: goat pAb AF1002 (R&D Systems, MN, USA) against mouse VE-cadherin (IF, 3 µg/ml), mouse monoclonal antibodies (mAb) V284 (Merck, Darmstadt, Germany) (IP, 5 µg/reaction), mouse mAb hVIN-1 (Sigma-Aldrich, MO, USA) (WB, 2 µg/ml) and rabbit mAb EPR8185 (Abcam, Cambridge, UK) (IF, 1:100) against vinculin, mouse mAb G11 (Santa Cruz Biotechnology, CA, USA) (IF, 5 µg/ml; WB, 1 µg/ml) and mAb alpha-CAT-7A4 (WB, 0.5 µg/ml; IP, 5 µg/reaction) (Thermo Fisher Scientific, MA, USA) against α-catenin, isotype control Ab (Rat IgG1, Thermo Fisher Scientific, MA, USA), (IP: 5 µg/reaction), mAb against α-tubulin (B-5-1-2, Sigma-Aldrich, MO, USA) (WB, 1:5000). Alexa Fluor 405-, Alexa-Fluor 488-, Alexa Fluor 568-, Alexa Fluor 594- and Alexa Fluor 647-coupled secondary antibodies were purchased from Invitrogen CA, USA) (IF, 1:700). IRDye 680RD- and IRDye 800CW-coupled secondary antibodies were purchased from LI-COR Biosciences (Cambridge, UK) (WB, 1:10,000). All other secondary antibodies were purchased from Dianova (Hamburg, Germany) (WB, 1:10,000).

HUVECs were stimulated with 1 U/ml thrombin (Calbiochem, CA, USA) for 10 min at 37°C and 5% CO₂. To inhibit myosin II, HUVECs were treated with 50 μM Blebbistatin (Sigma-Aldrich, MO, USA) for 30 min. DMSO was used as vehicle control.

Immunoprecipitation and immunoblotting

MDMVECs or eEND cells were lysed at 4°C for 10 min in IP-lysis buffer [1% Nonidet P-40, 25 mM Tris-HCl pH 7.4, 100 mM NaCl, 10 mM MgCl₂, 10% glycerol, and Complete EDTA-free proteinase inhibitor cocktail (Roche, Penzberg, Germany)]. Lysates were cleared by centrifugation at 20,817 g for 2 min before aliquots for direct blot analysis were set aside and aliquots for immunoprecipitation were incubated for 1 h at 4°C with 4.5 μm Dynabeads (Thermo Fisher Scientific, MA, USA) loaded with the respective antibodies. Immunocomplexes were washed five times with lysis buffer and dissolved in Laemmli sample buffer for standard western blot analysis.

To analyze binding of the VD7 antibody to α-catenin forms, ~200 ng of purified α-catenin dimers or monomers were diluted in 1 ml IP buffer (0.2% Nonidet P-40, 25 mM Tris-HCl pH 7.4, 100 mM NaCl, 10 mM MgCl₂, 10% glycerol, and 1 mM DTT). Aliquots for direct blot analysis were set aside and α-catenin IP was performed as described above.

Generation of polyacrylamide hydrogels

Polyacrylamide hydrogels attached to 12 mm glass coverslips were fabricated following an established protocol (Tse and Engler, 2010) with slight modifications. In brief, aqueous solutions with two different concentrations of acrylamide/bis-acrylamide (3%/0.025% and 8%/0.264%; Bio-Rad) were prepared to yield hydrogels of varying stiffness (Young's moduli of <0.2 kPa and 20 kPa, respectively). To enable cell adhesion, collagen I was conjugated to the hydrogel surface using the heterobifunctional linker sulfo-SANPAH. A 1 mg/ml aqueous solution of sulfo-SANPAH (Sigma, MO, USA) was pipetted on the hydrogel surface, followed by irradiation with 365 nm UV light (intensity of 10 mW/cm²) for 1 min. Substrates were washed with PBS and incubated with a 50 μg/ml rat-tail collagen I (Corning, NY, USA) solution in PBS for 3 h at 37°C. Samples were washed with PBS prior to cell seeding. Young's moduli of the hydrogels were characterized using a nanoindenter (Piuma, Optics 11, The Netherlands).

Separation of cellular protein pools according to detergent solubility

Cells were rinsed twice with ice-cold PBS before extraction buffer I (20 mM Tris-HCl pH 7.4, 2 mM CaCl₂, 2 mM MgCl₂, 150 mM NaCl, 0.5% Triton-X-100, Complete EDTA-free proteinase inhibitors) was added and incubated for 10 min on ice. Subsequently, the buffer was collected and centrifuged at 20,817 g for 10 min at 4°C and the supernatant was defined as the high-Triton-X soluble fraction. Remaining cellular material on the dish was gently washed twice with cold PBS before addition of extraction buffer II [50 mM Tris-HCl pH 7.4,150 mM NaCl, 1% NP- 40, 2 mM EDTA, 1% (w/v) sodium deoxycholate, 0.5% (w/v) SDS, 0.01 M NaPi, Complete–EDTA–free proteinase inhibitors] and scraping all residual material off the dish. Lysis was performed for 30 min at 4°C in a head-over-end shaker and lysates were centrifuged for 30 min at 20,817 g and 4°C. Supernatants with proteins of high and low detergent solubility were subjected to SDS-PAGE and western blot analysis.

Immunofluorescence staining

MDMVECs, eEND cells or HUVECs were seeded on fibronectin-coated chamber slides (ibidi, Martinsried, Germany, μ-Slide 8 Well) and grown to confluence. Cells were washed with PBS, fixed with 4% PFA in PBS for 7 min at room temperature, and permeabilized using 0.5% Triton X-100 in PBS for 10 min at room temperature, followed by blocking with 3% BSA, 2% donkey-serum for 2 h and incubation with primary antibodies. Primary antibodies were detected with Alexa Fluor-coupled secondary antibodies. Fluorescence signals were detected using a confocal laser-scanning microscope (LSM 780/LSM 880) with online fingerprinting mode [ZEN 2.3 (Zeiss, Oberkochen, Germany)]. ImageJ/Fiji or Imaris (Bitplane, Zurich, Switzerland) software were used for image processing. To quantify VD7 intensity at cell–cell junctions, a colocalization channel between VD7 and total α-catenin was created, and fluorescence intensity of VD7 was measured over the colocalization channel and normalized to fluorescence intensity of total α-catenin.

RNA-mediated interference

For interference with human α-catenin expression, the siRNA 5′-ACAUGGCAGAUGUCUACAAtt-3′ produced by Life Technologies (CA, USA) was used. For interference with human VE-cadherin expression, CDH5 siRNA by Life Technologies was used (5′-GGGUUUUUGCAUAAUAAGCTT-3′). AllStars negative control siRNA (Qiagen, Hilden, Germany, sequence not provided), which does not target any known mammalian gene, was used as control. Routinely, HUVECs were transfected with 40–80 nM siRNAs for 48–72 h using INTERFERin^® (Polyplus, Illkirch, France) according to manufacturer's guidelines.

FLIM-FRET and fluorescence intensity analysis via the VE-cadherin tension sensor during thrombin stimulation

VE-cadherin-siRNA treated HUVECs were seeded on fibronectin coated ibidi 8-well µ-chamber slides with 3×10⁴ cells per chamber, followed by re-expression of the VE-cadherin tension sensor (VEC-FL) (Arif et al., 2021) for 24 h. Cells were stimulated with thrombin for 10 min, washed with DPBS and fixed with 4% PFA in PBS for 5 min at 37°C. For IF analysis, cell monolayers were permeabilized using 0.5% Triton X-100 in PBS for 10 min at room temperature and blocked with 3% BSA, 2% donkey serum for 2 h, followed by incubation with VD7 and total α-catenin antibodies. Donkey anti-mouse-IgG conjugated to Alexa Fluor 405 and donkey anti-rabbit-IgG conjugated to Alexa Fluor 647 secondary antibodies were used to detect primary antibodies. FLIM images were acquired with a 485 nm pulsed laser at 40 MHz using ZEN 2.3 (Zeiss, Oberkochen, Germany) with SymPhoTime 64 2.4 (PicoQuant, Berlin, Germany). Signal was detected using a MultiHarp 150 TCSPC unit filtered by a 520/35 bandpass filter. Prior to each measurement, a z-stack of the fluorescence signal of VD7 and total α-catenin was recorded using 405 and 633 nm constant lasers within the same field of view as the FLIM acquisition.

For the analysis of FLIM-FRET data, pixel-wise data fitting (SymPhoTime, Picoquant, Berlin, Germany) was used with subsequent bi-exponential tail-fitting. For fluorescence intensity images, ImageJ/Fiji was used to create a mask based on total α-catenin intensity signal. This mask was further segmented with the Watershed segmentation algorithm (ImageJ/Fiji) to create different regions of interest (ROIs). Next, the plugin ‘Ratio Plus’ was used to create a pixel-by-pixel ratio image of VD7 to total α-catenin intensity. Average ratio and VEC-FL lifetime of the automatically generated ROIs were used for correlation analysis in R software (Version 1.3.1093).

In vivo vascular permeability assay in the skin

A modified Miles assay for the induction of vascular permeability in the skin was performed as described previously (Mamluk et al., 2005). 8–12-week-old female mice were used for this assay. Evans Blue dye (Sigma-Aldrich) was injected into the tail vein (100 μl of a 1% solution in PBS), and after 15 min, 50 μl PBS or 225 ng histamine in 50 μl PBS was injected intradermally into the shaved back skin. 30 min later, skin areas were excised and extracted with formamide for 5 days, and the concentration of the dye was measured at 620 nm with a spectrophotometer (Shimadzu, Kyoto, Japan).

Recombinant protein expression and purification

cDNA encoding the mouse cytoplasmic tail of VEC-αC (cyto VEC-αC) [aa 623–784 of VE-cadherin fused to the C-terminal two-thirds of mouse α-catenin (aa 301–906)] was generated by PCR using the original VEC-αC cDNA as a template. The cDNA encoding cyto VEC-αC or cDNA encoding full-length mouse α-catenin were inserted into the pET21a vector (Sigma-Aldrich) to generate an in-frame fusion between mouse α-catenin and a C-terminal His₆ tag. Recombinant fusion proteins were expressed in BL21 (DE3) Codon Plus E. coli cells, purified on HisTrap HP columns (Cytiva, MA, USA), and eluted with imidazole. Fusion proteins were further purified at 4°C using Superdex 200 gel filtration chromatography (ÄKTA, GE HealthCare, MA, USA) in 20 mM Tris-HCl pH 8.0, 150 mM NaCl and 2.5 mM DTT. Recombinant α-catenin containing a C-terminal His₆ tag was expressed and purified similarly to cyto VEC-αC, and the monomer was separated from the dimer by two consecutive rounds of gel filtration chromatography on Superdex 200.

F-actin co-sedimentation assay

The assay was performed using the Actin Binding Protein Spin-Down Biochem Kit (Cytoskeleton, CO, USA) according to the manufacturer's instructions with minor modifications. Rabbit muscle G-actin was incubated in 1× actin polymerization buffer (Cytoskeleton, CO, USA) for 1 h at room temperature to polymerize filaments. Gel filtered cytoplasmic tail of VEC-αC, as well as dimers and monomers of α-catenin were diluted to a final concentration of 1 µM in 1× reaction buffer (Cytoskeleton, CO, USA), with and without F-actin and incubated for 30 min at room temperature. Actin cushion buffer (Cytoskeleton, CO, USA) was pipetted into centrifuge tubes, then the reaction mixture was carefully loaded on top of the cushion buffer and centrifuged at 186,000 g for 30 min at 4°C in a TLA 55 rotor (Beckman Coulter, IN, USA). Pellets were resuspended in 1× Laemmli sample buffer, separated by SDS-PAGE, and stained with Coomassie Brilliant Blue. Gels were imaged on an Epson scanner (CA, USA) and band intensities were quantified using the Image Studio software (LI-COR Biosciences, Cambridge, UK).

We thank Milos Galic (University of Münster, Germany) for his help and scientific advice.