VE-PTP stabilizes VE-cadherin junctions and the endothelial barrier via a phosphatase-independent mechanism

Abstract

Vascular endothelial (VE) protein tyrosine phosphatase (PTP) is an endothelial-specific phosphatase that stabilizes VE-cadherin junctions. Although studies have focused on the role of VE-PTP in dephosphorylating VE-cadherin in the activated endothelium, little is known of VE-PTP's role in the quiescent endothelial monolayer. Here, we used the photoconvertible fluorescent protein VE-cadherin-Dendra2 to monitor VE-cadherin dynamics at adherens junctions (AJs) in confluent endothelial monolayers. We discovered that VE-PTP stabilizes VE-cadherin junctions by reducing the rate of VE-cadherin internalization independently of its phosphatase activity. VE-PTP serves as an adaptor protein that through binding and inhibiting the RhoGEF GEF-H1 modulates RhoA activity and tension across VE-cadherin junctions. Overexpression of the VE-PTP cytosolic domain mutant interacting with GEF-H1 in VE-PTP-depleted endothelial cells reduced GEF-H1 activity and restored VE-cadherin dynamics at AJs. Thus, VE-PTP stabilizes VE-cadherin junctions and restricts endothelial permeability by inhibiting GEF-H1, thereby limiting RhoA signaling at AJs and reducing the VE-cadherin internalization rate.

Figure 1.

VE-PTP stabilizes the endothelial barrier by decreasing the VE-cadherin internalization rate. (A) Permeability of HPAEC monolayers to FITC-conjugated albumin tracer after treatment with NT (control) siRNA or VE-PTP siRNA; mean ± SEM, n = 3–4 independent experiments; *, P < 0.05, unpaired t test. (B) Endothelial permeability rate constants of 0.48 ± 0.06 min⁻¹ and 0.88 ± 0.05 min⁻¹ in cells from A treated with NT siRNA or VE-PTP siRNA, respectively; mean ± SEM; n = 3–4; **, P < 0.001, unpaired t test. (C) Time-lapse images of VE-cad-Dendra2 emitting green fluorescence before photoconversion and red fluorescence after photoconversion within a selected region (indicated by circle) in HPAECs treated with NT siRNA or VE-PTP siRNA. Scale bars, 5 µm. (D) VE-cadherin internalization rate (decay in red fluorescence within photoconversion zone in C) in NT siRNA and VE-PTP siRNA-treated HPAECs; mean ± SEM; n = 9–12 junctions from four independent experiments. (E) Internalization rate constants of 0.15 ± 0.01 min⁻¹ and 0.23 ± 0.01 min⁻¹ from data in D in cells treated with NT siRNA or VE-PTP siRNA, respectively; mean ± SEM; n = 9–12 junctions from four independent experiments; ***, P < 0.0001, unpaired t test. (F) Schematic representation of VE-PTP mutants used in G–I; mCyan (control), full-length (WT) VE-PTP, Δ16FN VE-PTP mutant (lacking FN1-16 but capable of binding to VE-cadherin via intact 17th FN domain), or ΔN VE-PTP mutant (lacking entire extracellular VE-PTP domain). (G) Time-lapse images of VE-cad-Dendra2 in HPAECs overexpressing constructs in F. Scale bar, 5 µm. (H) VE-cadherin internalization rates from AJs in HPAECs transfected with constructs in F; mean ± SEM; n = 7–12 junctions from four independent experiments. (I) Internalization rate constants from data in H in cells overexpressing mCyan (0.16 ± 0.012 min⁻¹), WT VE-PTP (0.09 ± 0.01 min⁻¹), Δ16FN (0.10 ± 0.01 min⁻¹), or ΔN (0.16 ± 0.01 min⁻¹); mean ± SEM; n = 7–12 junctions from four independent experiments; *, P < 0.05; **, P < 0.001, one-way ANOVA.

Figure 2.

VE-PTP phosphatase activity is not required for stabilization of VE-cadherin junctions in the quiescent endothelium. (A) Schematic representation of VE-PTP mutants overexpressed in HPAECs; mCyan, WT VE-PTP, VE-PTP PI, and VE-PTP ΔC (lacking cytoplasmic domain) mutants. (B) Time-lapse images of VE-cad-Dendra2 in HPAECs overexpressing constructs in A. Scale bars, 5 µm. (C) VE-cadherin internalization from AJs in HPAECs overexpressing constructs in A; mean ± SEM; n = 8–11 junctions from three to four independent experiments. (D) Internalization rate constants calculated from C in cells overexpressing mCyan (0.17 ± 0.01 min⁻¹), WT (0.11 ± 0.01 min⁻¹), VE-PTP PI (0.11 ± 0.01 min⁻¹), and ΔC VE-PTP (0.18 ± 0.02 min⁻¹); mean ± SEM; n = 8–11 from three to four independent experiments; *, P < 0.05, one-way ANOVA.

Figure 3.

VE-PTP interacts with C terminus of GEF-H1. (A) Reverse immunoprecipitation (IP) of endogenous VE-PTP or GEF-H1 proteins from HPAEC lysates. Blots were probed for GEF-H1 and VE-PTP. (B) Schematic representation of indicated His-tagged C terminus of VE-PTP and GST-tagged GEF-H1 deletion mutants. (C and D) Domain interaction of GEF-H1 tested in pull-down experiments. Gel electrophoresis stained with Coomassie blue of bacteria purified proteins indicated in B (left). Direct interactions between cytosolic domain of His-VE-PTP (aa 1,651–1,998) and various GEF-H1 deletion mutants (right) detected by Western blot analysis.

Figure 4.

VE-PTP reduces GEF-H1 binding to RhoA and inhibits RhoA activity at VE-cadherin junctions. (A and B) Interaction of GEF-H1 with GST-RhoA (G17A) in HPAECs treated with NT siRNA or VE-PTP siRNA. The resulting precipitates were probed for GEF-H1 (A) and quantification of data (B); mean ± SEM; n = 3; *, P < 0.05; **, P < 0.001; ****, P < 0.0001; one-way ANOVA. (C) Immunofluorescent images of VE-cadherin (red) and GEF-H1 (green) in confluent HPAEC monolayers treated with NT siRNA or VE-PTP siRNA. Scale bars, 5 µm. (D) Analysis of GEF-H1 expression at VE-cadherin junctions from data in C; mean ± SEM, n = 18 images per group from two independent experiments. (E) Differential interference contrast (DIC) and confocal images of biosensor (YFP) and RhoA activity (FRET/CFP) in HPAECs treated with NT siRNA or siRNA against VE-PTP, GEF-H1, or both proteins. The ratiometric images were scaled from 1 to 3.5 and color-coded as indicated on right. Warmer colors denote higher RhoA activity. Scale bars, 5 µm. (F and G) Relative RhoA activity at the AJs (F) or in cytosol (G) of cells in E; mean ± SEM; n = 10–19 junctions from three independent experiments; *, P < 0.05; **, P < 0.001; one-way ANOVA. (H and I) Interaction of GEF-H1 with GST-RhoA (G17A) in HPAECs overexpressing CFP or CFP-VE-PTP. The resulting precipitates were probed for GEF-H1 using Western blot analysis (H) and quantification of data (I). GST precipitates from CFP-expressing cells used as a control; n = 2; *, P < 0.05; **, P < 0.001; one-way ANOVA. (J) DIC and confocal images of biosensor (YFP) and RhoA activity (FRET/CFP) in HPAECs expressing mPlum (control), mPlum-VE-PTP (WT), or mPlum-VE-PTP PI (PI). The ratiometric images were scaled from 3 to 10.5 and color-coded as indicated on the right. Scale bars, 5 µm. (K) Relative RhoA activity at AJs of cells shown in H; mean ± SEM; n = 14–17 junctions from three independent experiments; **, P < 0.001; one-way ANOVA. KD, knockdown.

Figure 5.

VE-PTP relieves tension across VE-cadherin junctions in the quiescent endothelium. (A) DIC and confocal images of VE-cadherin biosensor (YFP) and VE-cadherin tension (FRET/CFP) in HPAECs depleted of VE-PTP, GEF-H1, or both. The ratiometric images scaled from 1 to 3.5 and color coded as indicated on right. Warmer colors denote low tension. Scale bars, 5 µm. (B) Relative tension at AJs for groups in A. Higher values denote lower tension; mean ± SEM; n = 10–15 junctions from three independent experiments; *, P < 0.05; **, P < 0.001; one-way ANOVA. (C) DIC and confocal images of VE-cadherin biosensor (YFP) and VE-cadherin tension (FRET/CFP) in HPAECs overexpressing mPlum (control), mPlum-VE-PTP (WT), or mPlum-VE-PTP PI (PI). The ratiometric images are scaled as in A. Scale bars, 5 µm. (D) Relative tension at AJs for groups in C; mean ± SEM; n = 8–16 junctions from three independent experiments; *, P < 0.05; **, P < 0.001; one-way ANOVA. KD, knockdown.

Figure 6.

GEF-H1 knockdown restores VE-cadherin internalization rate in VE-PTP–depleted endothelial monolayers. (A) VE-cad-Dendra2 before (green) and after (red) photoconversion in HPAECs depleted of VE-PTP, GEF-H1, or VE-PTP and GEF-H1 simultaneously. Scale bars, 5 µm. (B) VE-cadherin internalization from AJs from data in A; mean ± SEM; n = 9–13 junctions from three independent experiments. (C) Internalization rate constants calculated from B were 0.17 ± 0.02 min⁻¹ in NT siRNA–treated cells, 0.29 ± 0.04 min⁻¹ and 0.10 ± 0.01 min⁻¹ in VE-PTP– and GEF-H1–depleted cells, or 0.19 ± 0.01 min⁻¹ after simultaneous depletion of VE-PTP and GEF-H1; mean ± SEM; n = 9–13 junctions from three independent experiments; *, P < 0.05; **, P < 0.001; one-way ANOVA. (D) Permeability of HPAEC monolayers to FITC-conjugated albumin in HPAECs depleted of VE-PTP, GEF-H1, or VE-PTP and GEF-H1 simultaneously; n = 3–4. *, P < 0.05; one-way ANOVA. (E) Permeability rate constants from D were 0.54 ± 0.06 min⁻¹ in NT siRNA–treated cells, 0.83 ± 0.06 min⁻¹ and 0.31 ± 0.02 min⁻¹ after VE-PTP and GEF-H1 depletion, or 0.49 ± 0.06 min⁻¹ after simultaneous depletion of VE-PTP and GEF-H1; mean ± SEM; n = 3–4; *, P < 0.05; one-way ANOVA. KD, knockdown.

Figure 7.

The VE-PTP cytosolic domain restores the VE-cadherin internalization rate and GEF-H1 activity in VE-PTP–depleted endothelial monolayers. (A) Immunoprecipitation (IP) of the CFP-tagged VE-PTP C domain from HPAEC lysates. Blots were probed for GEF-H1 and CFP. (B) Interaction of GEF-H1 with GST-RhoA (G17A) in HPAECs treated with NT siRNA or depleted of VE-PTP with and without overexpression of the VE-PTP cytosolic (C) domain. The resulting precipitates were probed for GEF-H1. (C) Analysis of interaction from data in B; mean ± SEM; n = 3; *, P < 0.05; one-way ANOVA. (D) VE-cad-Dendra2 before (green) and after (red) photoconversion in HPAECs treated with of NT siRNA or VE-PTP siRNA and overexpressing mCyan or HPAECs treated with VE-PTP siRNA and expressing the VE-PTP cytosolic (C) domain. Scale bars, 5 µm. (E) VE-cadherin internalization rate curves for groups in D; mean ± SEM; n = 10–13 junctions from three independent experiments. (F) Internalization rate constants calculated from E were 0.12 ± 0.01 min⁻¹ in NT siRNA–treated cells, 0.17 ± 0.01 min⁻¹ in VE-PTP–depleted cells, or 0.11 ± 0.01 min⁻¹ in VE-PTP–depleted cells overexpressing the VE-PTP C domain; mean ± SEM; n = 10–13 junctions from three independent experiments; *, P < 0.05; one-way ANOVA. (G) Permeability of HPAEC monolayers to FITC-conjugated albumin in HPAECs treated with of NT siRNA or VE-PTP siRNA and overexpressing mCyan or HPAECs treated with VE-PTP siRNA and expressing the VE-PTP cytosolic (C) domain; n = 3–4; *, P < 0.05; one-way ANOVA. (H) Permeability rate constants from G were 0.80 ± 0.12 min⁻¹ in NT siRNA-treated cells, 1.22 ± 0.05 min⁻¹ in VE-PTP–depleted cells, or 1.00 ± 0.07 min⁻¹ in cells overexpressing VE-PTP domain and on the background of VE-PTP depletion; mean ± SEM; n = 3–4; *, P < 0.05; one-way ANOVA. KD, knockdown.