Rho-ROCK controls de novo tight junction assembly between newly adjacent cells during apical extrusion

(A) Schematic showing the elimination of an apoptotic cell by apical extrusion and de novo tight junction (TJ) formation. Tight junctions are established between newly adjacent cells (shown in green) concurrent with extrusion of the apoptotic cell. The TJs between the apoptotic cell and the cells that effect extrusion (shown in purple) gradually diminish as the new TJs mature, ensuring that the epithelial barrier is never disrupted.

(B) Live cell images of EpH4 cells expressing GFP-claudin-3 in cells neighboring laser-wounded cell. Cell marked by an asterisk was wounded at time zero. The schematics represent the cells before laser irradiation and after apical extrusion. Purple arrowheads indicate the old TJ with the apoptotic cell and green arrowheads indicate the newly assembled TJ. The upper images show higher magnification corresponding to each time point. Scale bar = 20 μm. (See also Movie 1)

(C) Immunofluorescence images showing apical extrusion at 60 min post-laser irradiation. Cells were stained with anti-claudin-3 pAb (green) and either anti-ZO-1 mAb (magenta, left) or anti-occludin mAb (magenta, right). Purple arrowheads indicate old TJs with the apoptotic cell and green arrowheads indicate newly assembled TJs. Scale bar = 10 μm.

(D) Live cell images of cells surrounding a laser-wounded cell expressing GFP-AnillinC as a probe for active RhoA. Cell marked by an asterisk was wounded at 0 min. The arrowheads indicate accumulation of active RhoA at the junction between an apoptotic cell and a neighboring cell (white) or between newly adjacent cells (yellow). Scale bar = 20 μm. (See also Movie 2)

(E) Immunofluorescence images showing apical extrusion at 60 min post-laser irradiation. Cells were stained with phosphorylated myosin light chain (ppMLC) mAb (magenta) and anti-NMIIB mAb (green). The asterisk in the X-Z image indicates the apoptotic cell. Scale bar = 10 μm.

(F) Live cell images of cells surrounding a laser-wounded cell co-expressing GFP-claudin-3 and mScarlet-PLCδPH (membrane marker). Cells were pre-treated with H1152 and the cell marked by an asterisk was wounded at 0 min. The filled purple arrowhead indicates a TJ with an apoptotic cell and the unfilled purple arrowheads indicate the gradual dissolution of said TJ. Filled purple arrowheads indicate TJ between an apoptotic cell and neighboring cells, while unfilled purple arrowheads indicate the disappearance of TJ. Images of scarlet-PLCδ PH are projections of basal confocal slices and the extending lamellipodia are indicated by the white arrowheads. The inset enlarged at right—and illustrated below—show the absence of new TJs among the now adjacent cells A-E. Scale bar = 20 µm. (See also Movie 3).

(G) Bar graph showing the paracellular flux of 70kDa FITC-dextran tracer molecule at 9 h post-doxorubicin treatment (4 µM) (control: N=3, H1152: N=5; error bar: ±SD; Mann–Whitney U test).

(H) Schematic illustrating the formation of new TJs between neighbor cells concurrent with extrusion of the apoptotic cell. The formation of these TJ requires the activation of RhoA.

Claudins in epidermal cells are present at the plasma membrane in an unpolymerized state before differentiation in the granular layer

(A) Schematic showing the differentiation of human keratinocyte (HaCaT cells) into granular-like cells in high-calcium (9.8 mM) medium supplemented with a JNK inhibitor (40 μM) (Ca+JNK inh medium) for 24 h.

(B) HaCaT cells were cultured in normal medium (control) or Ca+JNK inh medium for 24 h, fixed, and then stained with anti-claudin-1 pAb (green), anti-E-cadherin mAb (magenta) and phalloidin (grayscale; left) or anti-ZO-1 mAb (green), anti-alpha-catenin (magenta) and phalloidin (grayscale; right). X-Z images are shown below. Yellow arrowheads indicate TJs. Scale bar = 20 µm.

(C) Whole-cell lysates of cells cultured in normal medium or Ca+JNK inh medium for 24h were immunoblotted with the indicated antibodies. Molecular weight measurements are in kDa.

(D) Mouse ear whole-mount immunofluorescence analysis for claudin-1, ZO-1 and E-cadherin. The asterisk indicates the nucleus of cells belonging to that layer. Scale bar = 20 µm.

De novo TJ formation in the SG2 layer requires activation of Rho-ROCK

(A and B) RhoA activities in undifferentiated and differentiated HaCaT cells. Cells were lysed and subjected to a GST-Rhotekin pulldown assay. Total cell lysates and precipitates were analyzed by immunoblotting with anti-RhoA mAb. Molecular weight measurements are in kDa. B is a graph illustrating the rate of RhoA activities (N=6; error bar: ±SD; Student’s t test).

(C and D) HaCaT cells were cultured in normal medium or Ca+JNK inh medium for 24 h, fixed, and then stained with anti-ppMLC mAb (green) and anti–E-cadherin mAb (magenta). Scale bar: 20 μm. Graph showing the quantification of junctional enrichment of ppMLC signals

(D). The quantification methods are illustrated in Figure S1 and details are described in Methods (control: N=5, Ca+JNK inh: N=8; error bar: ±SD; Student’s t test).

(E) Mouse ear whole-mount immunofluorescence analysis for ppMLC, ZO-1 and E-cadherin. The asterisk indicates the nucleus of cells belonging to the indicated layer.

(F and G) HaCaT cells expressing GFP-claudin-3 were cultured in normal medium or Ca+JNK inh medium supplemented with DMSO (control) or Y27632 (ROCK inhibitor) for 24 h, fixed, and then stained with anti–E-cadherin mAb (magenta). Scale bar: 20 μm. G is a bar graph illustrating the junctional enrichment of claudin-3 after differentiation (control: N=5, Ca+JNK inh: N=6; error bar: ±SD; Student’s t test).

(H) Schematic illustrating TJ formation in the epidermal granular layer. In the epidermis, all keratinocytes of basal, spinous and granular layer express claudin-1. However, functional TJs are only formed in the SG2 layer. The activation of the Rho-ROCK pathway is crucial for the formation of TJs in the SG2 layer.

Rho-ROCK activation induces ectopic TJ formation in undifferentiated keratinocytes

(A) Undifferentiated HaCaT cells expressing constitutive active RhoA (RhoA CA) were stained with anti-claudin-3 pAb (green) and phalloidin (grayscale). The arrowheads indicate ectopic tight junction. Scale bar: 20 μm.

(B) Undifferentiated HaCaT cells expressing RhoA CA were stained with anti-claudin-3 pAb (magenta) and either anti-ZO-1 mAb (green; upper) or anti-occludin mAb (green; lower). The arrowheads indicate ectopic TJs. Scale bar: 20 μm.

(C and D) Confluent undifferentiated HaCaT cells were cultured in serum-free medium for 24h, treated with a recombinant bacterial cytotoxic necrotizing factor (CNF) toxin (CN-03), also known as Rho activator II (1 μg/ml) for 2 h, fixed, and then stained with anti-claudin-3 pAb (green), anti-E-cadherin mAb (magenta) and phalloidin (grayscale). Scale bar: 20 μm. (D) shows an alternate field of view corresponding to the experiment in (C).

(E and F) Confluent undifferentiated HaCaT cells were treated with DMSO (control) or Narciclasine (100 nM; ROCK activator) for 4 h, fixed, and then stained with anti-claudin-3 pAb (green), anti-E-cadherin mAb (magenta) and phalloidin (grayscale). Scale bar: 20 μm. (F) shows an alternate field of view of the experiment in (E).

(G) Junctional enrichment of claudin-3 was quantified based on the method described in Figure S1 (N=6; error bar: ±SD; Tukey-Kramer One-way Anova).

(H) Western blot of Triton X-100–soluble fractions and -insoluble fractions from HaCaT cells treated with Rho activator (CN-03) or Narciclasine.

Rho-ROCK pathway acts upstream of matriptase activation in de novo TJ formation

(A) Whole-cell lysates of WT treated with Rho activator (CN-03), Narciclasine or acid medium (pH 6.0) were immunoblotted with the indicated antibodies. Molecular weight measurements are in kDa.

(B and C) Confluent undifferentiated HaCaT cells were cultured in normal (control) or acid medium (pH 6.0) for 20 min, fixed, and then stained with anti-claudin-3 pAb (green), anti-E-cadherin mAb (magenta) and phalloidin (grayscale). Scale bar = 20 μm. C is a bar graph illustrating the junctional enrichment of claudin-3 (N=6; error bar: ±SD; Student’s t test).

(D) Western blot of Triton X-100–soluble fractions and -insoluble fractions from confluent undifferentiated HaCaT cells cultured in acid medium.

(E) Undifferentiated HaCaT cells treated with DMSO (control) or camostat (serine protease inhibitor) expressing constitutive active RhoA (Rho CA) were fixed and stained with anti-claudin-3 pAb (green) and phalloidin (grayscale). Scale bar: 20 μm.

(F) Bar graph illustrating the extent to which the matriptase inhibitor camostat negated the Narciclasine-induced claudin enrichment at cell-cell contacts. After inhibition of serine protease with camostat confluent undifferentiated HaCaT cells were treated with Narciclasine (100 nM; ROCK activator) for 4 h and stained to quantify the junctional enrichment of claudin-3 (N=5; error bar: ±SD; Student’s t test).

(G) Undifferentiated HaCaT cells expressing constitutive active RhoA (Rho CA) were stained with anti-TROP2 mAb (green) and claudin-4 mAb (magenta). Insets are enlarged images. Scale bar: 20 μm.

TROP2 is required for intracellular transport of claudin to the plasma membrane in keratinocytes

(A) Whole-cell lysates of WT and TROP2 KO HaCaT cells were immunoblotted with the indicated antibodies. Molecular weight measurements are in kDa.

(B and C) Representative western blots of surface and total amount of claudin-1. Molecular weight measurements are in kDa. The graph of (C) illustrates the quantification of claudin-1 present on the cell surface relative to the total amount (N=3; error bar: ±SD; Student’s t test).

(D) Undifferentiated TROP2 KO HaCaT cells were stained with anti-claudin-1 pAb (green) and anti-GM130 mAb (magenta; Golgi marker). Insets are enlarged images. Scale bar: 20 μm.

(E) Undifferentiated WT or TROP2 KO HaCaT cells expressing constitutive active RhoA (Rho CA) were stained with anti-claudin-3 pAb (green) and phalloidin (grayscale). Scale bar: 20 μm.

(F) Undifferentiated WT and TROP2 KO HaCaT cells were co-cultured, incubated in acid buffer (pH 6.0) for 20min, fixed, and then stained with anti-claudin-4 mAb (green) and anti-TROP2 mAb (magenta). Dotted line overlays the border between WT and TROP2 KO cells. Scale bar: 20μm.

(G) Bar graph illustrating the impairment of either Narciclasine- or acid medium-induced claudin-3 accumulation at cell-cell contacts by TROP2 depletion. Confluent undifferentiated WT or TROP2 KO HaCaT cells were treated with Narciclasine (ROCK activator) or acid medium (pH 6.0) and stained to quantify junctional enrichment of claudin-3 (N=5; error bar: ±SD; Tukey-Kramer One-way Anova).

(H) Schematic illustrating the mechanism of ectopic TJ formation through RhoA activation. In wild-type keratinocytes, claudin and TROP2 form a complex in the Golgi apparatus and are subsequently transported to the plasma membrane. In contrast, in cells lacking TROP2, claudin, which cannot form a complex with TROP2, accumulates in the Golgi apparatus, leading to a significant decrease in the amount of claudin at the plasma membrane. In wild-type cells, TROP2 is cleaved by matriptase activated via the Rho-ROCK pathway. This cleavage results in the breakdown of the TROP2-claudin complex, allowing released claudins to form de novo TJs.

Matriptase cleaves TROP2 In the SG2 layer of epidermis.

(A and B) WT HaCaT cells were cultured in Ca+JNK inh medium for indicated time points and time-course change of indicated proteins were examined by western blotting. Molecular weight measurements are in kDa. The amount of matriptase relative to the control α-tubulin (control) was quantified in (B) (N=3; error bar: ±SD; Dunn’s multiple comparison test).

(C) HaCaT cells expressing GFP-claudin-3 were cultured in normal medium or Ca+JNK inh medium supplemented with DMSO (control) or camostat (serine protease inhibitor) for 24 h, fixed, and then stained with anti–E-cadherin mAb (magenta). Scale bar: 20 μm.

(D) WT HaCaT cells and cells over-expressing Hai1-GFP were co-cultured in Ca+JNK inh medium for 24 h, fixed, and then stained with anti-claudin-1 pAb. Scale bar: 20 μm.

(E) Bar graph illustrating the effects of either camostat treatment or over-expression of Hai1 on accumulation of claudin-3 at cell-cell contacts in HaCaT cells cultured in Ca+JNK inh medium (N=6; error bar: ±SD; Tukey-Kramer One-way Anova).

(F) Mouse ear whole-mount immunofluorescence analysis for TROP2 and ZO-1. The asterisk indicates the nucleus of cells belonging to the indicated layer.

(G) Schematic illustrating the mechanism of TJ formation during differentiation into the granular layer in the epidermis. In undifferentiated cells, claudin is expressed and bound to TROP2 on the plasma membrane. As cells differentiate into the granular layer, RhoA becomes activated, followed by the activation of matriptase, leading to the cleavage of TROP2. The cleavage of TROP2 results in the dissociation of claudin, and the dissociated claudins then polymerized into de novo TJs.

Signaling through RhoA-Matriptase-EpCAM is critical for the formation of de novo TJ following apical extrusion

(A) Whole-cell lysates of WT and EpCAM KO EpH4 cells were immunoblotted with the indicated antibodies. Molecular weight measurements are in kDa.

(B) WT and EpCAM KO EpH4 cells were co-cultured, fixed, and then stained with anti-EpCAM pAb (green) and anti-claudin-4 mAb (magenta). Dotted line overlays the border between WT and EpCAM KO cells. Scale bar: 20 μm.

(C) WT or EpCAM KO EpH4 cells were cultured, fixed, permeabilized with digitonin, and then stained with anti-claudin-7 pAb (green) and anti-ZO-1 mAb (magenta). Yellow arrowheads indicate claudin-7 at the lateral membrane, while white arrowheads indicate the intercellular accumulation of claudin-7. Scale bar: 20 μm.

(D) EpCAM KO EpH4 cells permeabilized with digitonin were stained with anti-claudin-7 pAb (green) and anti-GM130 mAb (magenta; Golgi marker). Scale bar: 20 μm.

(E) WT and EpH4 KO EpH4 cells were co-cultured, incubated in acid buffer (pH 6.0) for 20min, fixed, and then stained with anti-claudin-4 mAb (green) and anti-EpCAM pAb (magenta). Yellow arrowheads indicate the formation of ectopic TJ strands on the basolateral membrane. Scale bar: 20 μm.

(F, G and H) Live cell imagings of EpCAM KO EpH4 cells expressing GFP-claudin-3 (F), EpH4 cells expressing GFP-claudin-3 treated with camostat (G), or EpH4 expressing both Hai1-mScarlet and GFP-claudin-3 (H) after laser irradiation. Cell marked by an asterisk was wounded at time zero. The right diagram depicts the condition of TJs 60 min after laser irradiation. The navy color represents pre-existing TJs, and the dashed lines indicate the absence of newly formed TJs between neighboring cells. Scale bar = 20 μm. (See also Movies 4-6).

(I) Bar graph showing the paracellular flux of 70kDa FITC-dextran tracer molecule at 9 h post-doxorubicin treatment (4 µM) (control: N=4, EpCAM KO: N=6, camostat: N=4; error bar: ±SD; Dunn’s multiple comparison test).

Activation of Rho-ROCK pathway triggers de novo TJ formation via disruption of the EpCAM/TROP2-claudin complex

De novo TJ formation associated with apoptotic cell elimination in simple epithelia and with keratinocyte differentiation in stratified epithelia, is regulated by a common molecular mechanism. In both types of epithelia, EpCAM or TROP2 is necessary for the intracellular transport of claudin from the Golgi apparatus to the plasma membrane. Claudin, when in complex with EpCAM or TROP2, cannot spontaneously polymerize into TJs at the plasma membrane. However, upon activation of the Rho-ROCK pathway, which activates matriptase, EpCAM or TROP2 is cleaved by matriptase. This cleavage breaks down the EpCAM/TROP-2-claudin complex, enabling claudin to polymerize into TJs. Consequently, rapid TJ formation occurs without de novo transcription or translation.