Epithelial protein lost in neoplasm (EPLIN) interacts with α-catenin and actin filaments in endothelial cells and stabilizes vascular capillary network in vitro

Abstract

Adherens junctions are required for vascular endothelium integrity. These structures are formed by the clustering of the homophilic adhesive protein VE-cadherin, which recruits intracellular partners, such as β- and α-catenins, vinculin, and actin filaments. The dogma according to which α-catenin bridges cadherin·β-catenin complexes to the actin cytoskeleton has been challenged during the past few years, and the link between the VE-cadherin·catenin complex and the actin cytoskeleton remains unclear. Recently, epithelial protein lost in neoplasm (EPLIN) has been proposed as a possible bond between the E-cadherin·catenin complex and actin in epithelial cells. Herein, we show that EPLIN is expressed at similar levels in endothelial and epithelial cells and is located at interendothelial junctions in confluent cells. Co-immunoprecipitation and GST pulldown experiments provided evidence that EPLIN interacts directly with α-catenin and tethers the VE-cadherin·catenin complex to the actin cytoskeleton. In the absence of EPLIN, vinculin was delocalized from the junctions. Furthermore, suppression of actomyosin tension using blebbistatin triggered a similar vinculin delocalization from the junctions. In a Matrigel assay, EPLIN-depleted endothelial cells exhibited a reduced capacity to form pseudocapillary networks because of numerous breakage events. In conclusion, we propose a model in which EPLIN establishes a link between the cadherin·catenin complex and actin that is independent of actomyosin tension. This link acts as a mechanotransmitter, allowing vinculin binding to α-catenin and formation of a secondary molecular bond between the adherens complex and the cytoskeleton through vinculin. In addition, we provide evidence that the EPLIN clutch is necessary for stabilization of capillary structures in an angiogenesis model.

FIGURE 1.

EPLIN expression in epithelial and endothelial cells. A, RT-PCR analysis of EPLIN and VE-cadherin mRNAs in HUVECs. EPLIN (lane 1), HPRT (lane 2), HPRT and VE-cad (lane 3), and HPRT and EPLIN (lane 4) mRNAs were converted into cDNAs prior to being used as templates for subsequent PCR amplification. VE-cad was used as a positive control, and HPRT was used as an internal standard. A negative control was performed with no template (lane 5). B, Western blot analysis of EPLIN expression in Caco2, MDCK, Eahy926 (EAHY), hCMEC/D3 (HCMEC), and HUVEC lysates. The membrane was reprobed with anti-β-tubulin (β-tub) antibody as a protein loading control. C, Western blot analysis of the subcellular localization of EPLIN and α-catenin proteins in confluent HUVECs. The membrane (Mb) and cytoplasmic (Cyto) fractions were separated from HUVEC lysates by ultracentrifugation and then analyzed by Western blot for their EPLIN and α-catenin contents. Molecular mass markers (kDa) are given at the left margin of each blot.

FIGURE 2.

Differential distribution of EPLIN in sparse and confluent HUVECs. HUVECs were plated on glass coverslips at two different densities and stained with anti-EPLIN antibody, FITC-phalloidin, and Hoechst. Bar, 20 μm.

FIGURE 3.

Subcellular localization of EPLIN and α-catenin in endothelial and epithelial cells. A, Caco2, MDCK, Eahy926 (EAHY), hCMEC/D3 (HCMEC), and human umbilical vascular endothelial cells were immunolabeled for EPLIN and α-catenin (α-Cat). Scale bar, 20 μm. B, confocal analysis of α-catenin, EPLIN, and F-actin on Caco2 and HUVEC monolayers. White lines reflect the plane for Z acquisition. Scale bar, 20 μm.

FIGURE 4.

Association of EPLIN with VE-cadherin-based complex in HUVECs. Co-immunoprecipitations of EPLIN with VE-cad and α- and β-catenins in HUVECs are shown. Anti-EPLIN (B), α-catenin (α-cat) (C), β-catenin (β-cat) (D), and VE-cad (E) immunoprecipitations (IP) were probed successively for EPLIN, α-catenin, β-catenin, and VE-cad. As controls, the whole cell lysate (Lys) was run in parallel (A), and the immunoprecipitations performed using rabbit (negative control for anti-EPLIN, anti-α-catenin, and anti-VE-cad immunoprecipitations) or mouse (negative control for anti-β-catenin immunoprecipitation) non-immune IgG (NI) were analyzed simultaneously. Note that in B immunoprecipitations performed using the rabbit anti-EPLIN antibody were probed with the mouse anti-EPLIN antibody. *, nonspecific band revealed by mouse anti-EPLIN antibody. VE-Cad tr corresponds to a truncated form of VE-Cad cleaved at its C terminus and deleted of the catenin-binding domains (18). F, Interaction of α-catenin with EPLIN. GST or GST-α-catenin immobilized on glutathione beads was incubated with HUVEC lysates. Pulldown assays and whole HUVEC lysate were revealed with the mouse anti-EPLIN antibody. G, direct interaction between purified EPLIN and α-catenin. Purified His-EPLIN was incubated with purified GST or GST-α-catenin immobilized on glutathione beads as indicated. Pulldown assays and whole HUVEC lysate were revealed with the mouse anti-EPLIN antibody. Each panel is illustrative of three experiments. WB, Western blot.

FIGURE 5.

Differential recruitment of EPLIN, VE-cadherin, and α-catenin to newly formed cell-cell contacts in HUVECs. HUVECs grown in EGTA-containing medium were switched from low Ca²⁺ to standard Ca²⁺ conditions and fixed in normal Ca²⁺ conditions at 0, 10, and 60 min post-Ca²⁺ addition. Cells were stained for EPLIN (D, G, J), and α-catenin (α-cat; B, E, H, K). Bars, 20 μm.

FIGURE 6.

Transient depletion of EPLIN in HUVECs. A, Western blot analysis of EPLIN protein expression in siRNA-treated HUVECs. Lysates of HUVECs transfected with control (Ctrl) or EPLIN siRNA were analyzed by Western blot to determine EPLINα and -β protein levels. Immunoblots were reprobed with β-tubulin (β-tub) as a protein loading control. B, efficiency of EPLIN extinction at 24, 48, and 72 h post-siRNA transfection. Evaluation of the residual amounts of EPLIN was performed using the quantification of the immunoreactive bands after normalization to β-tubulin using ImageJ software. C, Western blot analysis of VE-cadherin, α-catenin, and tubulin protein expression levels in HUVECs transfected with control (siCtrl) and EPLIN (siEPLIN) siRNA. Lysates of siRNA-treated HUVECs were probed successively for EPLIN, VE-cad, α-catenin (α-cat), and β-tubulin (β-tub). β-Tubulin expression was monitored for normalization of protein loading. tr, truncated.

FIGURE 7.

Immunofluorescence analysis of EPLIN, α-catenin, VE-cad, and actin distributions in siRNA-treated confluent HUVECs. 24 h after transfection with control (Ctrl) or EPLIN siRNA, HUVECs were stained for EPLIN, α-catenin, and nuclei (Hoechst) (A); for EPLIN, VE-Cad, and nuclei (B); and for EPLIN, actin, and nuclei (C). In C, arrows point to cortical F-actin (above), and arrowheads point to gaps within the cortical F-actin network (below). Confocal images are representative of at least three independent experiments. Scale bars, 10 (A) and 20 μm (B and C).

FIGURE 8.

Effect of EPLIN depletion on in vitro proliferation, adhesion, and wound healing assays. A and B, in vitro proliferation assay. siRNA-treated HUVECs were imaged after 24 and 48 h of proliferation after prestaining with crystal violet (A). Following lysis, optical density (OD) of the resulting solutions was measured at 560 nm. (B). □, control (Ctrl) siRNA-treated HUVECs; ■, siRNA-treated HUVECs. C and E, in vitro adhesion assay. 15 min after cell seeding, adherent siRNA-treated cells, stained with crystal violet, were imaged (C). After lysis, optical density of the resulting solutions was measured at 560 nm (E). D and F, wound healing assays. Confluent monolayers of siRNA-treated cells were scratched (time = 0) and imaged over time to track wound reclosure (D). The distances covered by the cell fronts were measured over time (F). Each panel is illustrative of three experiments. Fn, fibronectin.

FIGURE 9.

Dynamics of VE-cad-GFP at endothelial cell-cell contacts. A, representative examples of pre- and postphotobleached images of VE-Cad-GFP at cell-cell contacts. VE-cad-GFP-overexpressing HUVECs transfected with either control or EPLIN siRNA were grown to confluence prior to performing FRAP experiments. Photobleached spots are circled. B, mean fluorescence recovery curves for control and EPLIN siRNA-treated HUVECs. The curves were generated from 50 independent experiments for control (top) and EPLIN (bottom) siRNA-treated cells.

FIGURE 10.

Subcellular distributions of EPLIN and vinculin in HUVECs. A, HUVECs were stained with polyclonal anti-EPLIN, mAb anti-vinculin, Alexa Fluor 488-labeled goat anti-rabbit, and Alexa Fluor 568-labeled goat anti-mouse antibodies and Hoechst. Bar, 20 μm. Arrows and arrowheads point to mature and immature cell-cell junctions, respectively. B, immunofluorescence analysis of EPLIN and vinculin (Vin) distributions in siRNA-treated HUVECs. 24 h after transfection with control (Ctrl) or EPLIN siRNA, HUVECs were stained for EPLIN, vinculin, and nuclei (Hoechst) as described in A. Scale bar, 20 μm. Boxed regions are magnified in the side panels. Scale bar, 10 μm.

FIGURE 11.

Effect of blebbistatin treatment on subcellular localization of α-catenin, EPLIN, and vinculin revealed by confocal microscopy analysis. HUVECs treated with 0, 5, 10, and 50 μm blebbistatin (Blebb) were fixed before double labeling for α-catenin and EPLIN (A) and for α-catenin and vinculin (B). Scale bars, 20 μm.

FIGURE 12.

Effects of blebbistatin on junctional recruitment of vinculin and EPLIN. After pretreatment with 0, 5, or 10 μm blebbistatin, HUVECs were double stained for EPLIN and α-catenin and for vinculin and α-catenin. Fluorescence intensities were measured along white lines perpendicularly drawn to cell-cell junctions as illustrated in A. After background subtraction, fluorescence intensity profiles were generated, and the maximal intensity of fluorescence was used as a measure of junctional protein level. For normalization, EPLIN and vinculin maximal intensities of fluorescence were divided by the corresponding α-catenin maximal intensity of fluorescence to generate EPLIN/α-catenin (B) and vinculin/α-catenin (C) ratios. The mean values for the ratios were obtained from 20 measures for each condition. Statistical analysis using the Tukey method indicated that the junctional level of vinculin decreased, whereas the junctional level of EPLIN remained roughly constant as blebbistatin concentration increased.

FIGURE 13.

Localization of EPLIN and VE-cadherin in pseudovascular structures elaborated in Matrigel matrix. Pseudovascular tubules were formed using untreated HUVECs (A) or siRNA-treated HUVECs (B). 24 h after cell seeding, cells were labeled for EPLIN, VE-cadherin, and nuclei as performed in Fig. 7B. Ctrl, control.

FIGURE 14.

Localization of EPLIN and α-catenin in pseudovascular structures elaborated in Matrigel matrix. Pseudovascular tubules were formed using untreated HUVECs (A) or siRNA (si)-treated HUVECs (B). 24 h after cell seeding, cells were labeled for EPLIN and α-catenin (α-cat) as performed in Fig. 7A. Ctrl, control.

FIGURE 15.

Effect of EPLIN depletion on in vitro tubule formation assay. A, phase-contrast videomicroscopy images comparing the vascular network formation process for HUVECs pretreated with control (siCtrl) or EPLIN (siEPLIN) siRNAs. 24 h after siRNA transfection, HUVECs were plated onto Matrigel layers for 2 h before time lapse microscopy of the temporal evolution of precapillary structures. Postseeding time is mentioned on each image. Arrows point to capillary network breakage. Note that some endothelial cells aggregated into bunches (arrowheads). B, quantification of the vascular network full length. Phase-contrast videomicroscopy images, such as those presented in A, were analyzed using ImageJ software to calculate the full length of the capillary network. Mean values reported in this panel were calculated from three different observation fields. ■, control siRNA-treated HUVECs; □, EPLIN siRNA-treated HUVECs. * indicates a statistical difference with a p value <0.05. C, quantification of the average number of breakage events per image over time. Breakage events were counted 26 h after cell seeding. Mean values reported in this panel were calculated from three different observation fields. Each panel is illustrative of four experiments.

FIGURE 16.

EPLIN acts as a tension transducer at the basis of the conformational modification in α-catenin (adapted from Ref. 9). A, in the absence of EPLIN, actomyosin contractility is not transmitted to α-catenin (α-cat). Consequently, α-catenin keeps its folded conformation and cannot interact with vinculin as shown in Fig. 10B. B, in the presence of blebbistatin, actomyosin contractility is blocked, and α-catenin is maintained under a folded conformation, preventing vinculin recruitment as illustrated in Fig. 11B. C, when myosin II exerts its contractile activity, α-catenin adopts a stretched conformation, allowing vinculin recruitment and thus F-actin anchoring to cell-cell junctions. β-cat, β-catenin.