Tight junction membrane proteins regulate the mechanical resistance of the apical junctional complex

Epithelia cover the body and act as barriers to segregate the internal body from the external environment. Epithelial cells adhere to one another via intercellular junctions that consist of tight junctions (TJs), adherens junctions (AJs), and desmosomes, collectively known as the apical junctional complex (Farquhar and Palade, 1963). TJs are located at the most apical region of the intercellular junctions where adjacent plasma membranes are closely apposed and restrict free diffusion of solutes across the paracellular space (Farquhar and Palade, 1963; Anderson and van Itallie, 2009; Shen et al., 2011; Zihni et al., 2016; Otani and Furuse, 2020).

TJs are composed of integral membrane proteins, including claudins, TJ-associated MARVEL domain-containing proteins (occludin, tricellulin, and MarvelD3), and immunoglobulin superfamily proteins (e.g., junctional adhesion molecules [JAMs] and coxsackie and adenovirus receptor [CAR]) (Furuse et al., 1993; Furuse et al., 1998; Martìn-Padura et al., 1998; Cohen et al., 2001; Ikenouchi et al., 2005; Steed et al., 2009; Raleigh et al., 2010). These membrane proteins interact with ZO (zonula occludens) family scaffolding proteins through their cytoplasmic region. ZO family proteins are multidomain scaffolding proteins required for TJ formation (Umeda et al., 2006; Phua et al., 2014; Otani et al., 2019; Rouaud et al., 2020), and have N-terminal PDZ (PSD-95/Dlg/ZO-1) domains that interact with claudins and JAMs (Itoh et al., 1999, 2001; Bazzoni et al., 2000; Ebnet et al., 2000) and a C-terminal actin-binding region involved in the epithelial barrier function (Fanning et al., 1998, 2002; Belardi et al., 2020).

Epithelia are subjected to various mechanical stresses, including morphogenetic movements, visceral muscle contractions, cytokinesis, and cell death within the epithelial sheet. The intercellular junctions are involved in maintaining the epithelial integrity under these types of stress. Loss of AJs results in disruption of tissue integrity (Matsunaga et al., 1988; Takeichi, 1991; Kintner, 1992; Harris et al., 2012), while dysfunction of desmosomes leads to pemphigus, which is characterized by skin blistering (Amagai et al., 1991). Recent findings have suggested that mechanosensor molecules are localized at the intercellular junctions and participate in mechanotransduction (Charras and Yap, 2018; Angulo-Urarte et al., 2020). For example, α-catenin, which acts as a linker between the cadherin–catenin complex and actin filaments at AJs, adopts a closed conformation that unfolds in response to tension to unmask a cryptic binding site for vinculin (Yonemura et al., 2010). Similarly, ZO-1 adopts a folded conformation that unfolds in response to application of mechanical force, allowing recruitment of the transcription factor DbpA to TJs (Spadaro et al., 2017). However, unlike AJs and desmosomes, the roles of TJs in regulating the mechanical resistance of epithelia have remained unclear (Citi, 2019), partly due to the lack of methods to specifically and completely perturb TJs in epithelial cells.

We have recently performed systematic genome editing studies on TJ proteins and demonstrated that MDCK II cells deficient in claudins (claudin-2/4/3/7/1) and JAM-A (claudin/JAM-A knockout [KO] cells) exhibit specific and complete loss of the TJ structure and function (Otani et al., 2019). Although AJs were present in the claudin/JAM-A KO cells, a characteristic focal junction breakage phenotype was observed (Otani et al., 2019). However, the underlying mechanism for the phenotype remained unclear.

In the present study, we utilized time-lapse imaging to examine how the junction breakage phenotype arises in claudin/JAM-A KO cells. We found that the apical junctions in the cells showed weak resistance to mechanical stress. Moreover, TJ membrane proteins were required for nanometer-scale ordering of ZO-1 molecules, which appeared to be critical for the resilience of apical junctions against mechanical stress. The present findings establish a role for TJs in the mechanical resistance of epithelial cell junctions.

To characterize the junction breakage phenotype in claudin/JAM-A KO cells, we performed immunofluorescence staining of various cell–cell junction markers. Control MDCK II cells showed continuous ZO-1 staining along the apical junctions (Fig. 1 A). In contrast, sporadic discontinuity in ZO-1 staining was observed in claudin/JAM-A KO cells (Fig. 1 B′, arrow), and in extreme cases, large gaps in ZO-1 staining were observed (Fig. 1 B″, asterisk). Apical cell shape appeared to be more variable in claudin/JAM-A KO cells, confirmed by the quantification of the apical area variance (Fig. S1 A). Quantification of the number of cell junction endpoints per unit area confirmed the presence of junction breakage in claudin/JAM-A KO cells, but not in MDCK II, JAM-A KO, or claudin KO cells (Fig. 1 C). ZO-2 and occludin staining showed similar defects to ZO-1 staining (Fig. 1, D and E). Moreover, AJ and desmosome proteins, including afadin, E-cadherin, and desmoplakin, exhibited discontinuous staining at the apical junctions in claudin/JAM-A KO cells (Fig. 1, F–L), while continuous staining of E-cadherin and desmoplakin was maintained at lateral cell–cell contacts (Fig. 1, J and L). Desmoplakin retained its linear staining where the ZO-1 discontinuity was small, suggesting that desmosome breakage occurred secondary to the breakage of the apical junctions (Fig. 1, M and N). These results suggest that claudins and JAM-A are required for the integrity of the apical junctional complex.

To understand how the junction breakage phenotype arises in claudin/JAM-A KO cells, we stably expressed ZO-1-GFP in control MDCK II cells and claudin/JAM-A KO cells and performed time-lapse imaging. ZO-1-GFP was functional, as it was able to rescue the continuous apical junction localization of occludin in ZO-1/ZO-2 double KO cells (Fig. S5 A). In control MDCK II cells, the continuity of ZO-1-GFP was maintained during cell stretching (Fig. 2 A and Video 1) and cytokinesis (Fig. 2 B and Video 1), consistent with previous reports (Jinguji and Ishikawa, 1992; Baker and Garrod, 1993; Higashi et al., 2016). These results indicate that the cell junctions in control cells were robust against mechanical stress. Meanwhile, although the chicken-wire pattern of ZO-1-GFP was established in claudin/JAM-A KO cells during monolayer formation, the junctions underwent breakage upon spontaneous stretching of the cells (0.0 ± 0.0 × 10⁻³ breakage/μm²·h in MDCK II cells, 4.0 ± 0.5 × 10⁻³ breakage/μm²·h in claudin/JAM-A KO cells) (Fig. 2 C and Video 2). The broken junctions were reassembled within a few hours, implying that the junction repair pathways (Stephenson et al., 2019; Higashi et al., 2023) were not severely compromised in the claudin/JAM-A KO cells. Similar junction breakage was observed when claudin/JAM-A KO cells underwent cytokinesis (Fig. 2 D and Video 2).

Quantification of the cell circumference and the fluorescence intensity of ZO-1-GFP confirmed that the junction breakage occurred upon acute spontaneous cell stretching (Fig. 2, E and F). Measurements from multiple spontaneous cell stretching events revealed that junction breakage tends to occur in claudin/JAM-A KO cells when they are stretched above a certain threshold (∼25% stretching), while the cell junctions in control MDCK II cells withstood a similar degree of stretching (Fig. 2 G). The extent of cell stretching and the duration of individual spontaneous stretching events were greater in MDCK II cells, probably because MDCK II cells could withstand a greater extent of cell stretching (Fig. S1, B and C). In contrast, cell stretching speed was not significantly different between control and claudin/JAM-A KO cells (Fig. S1 D), and junction breakage was observed among a wide range of cell stretching speeds (Fig. S1 E). Some cells were found to undergo junction breakage although the degree of stretching was small (Fig. 2 G). Close inspection of these outliers revealed that these cells underwent repetitive sequential cell stretching events (Fig. S1, F and G). Analyses of the spatial distribution of strain showed that local junctional strain correlated with the junction breakage (correlation coefficient between junction length change and ZO-1-GFP intensity change: r_ΔL,ΔI = −0.39) (Fig. S1 H), suggesting that the junction breakage upon spontaneous cell stretching in claudin/JAM-A KO cells involves a mechanical fracture of cell junctions. In contrast, junction breakage during cell division temporally correlated with the cell cycle progression but did not correlate with the local junctional strain (r_ΔL,ΔI = −0.04) (Fig. S1 I). Hereafter, we focus on the junction breakage associated with spontaneous cell stretching in this paper.

To test the idea that claudin/JAM-A regulates the mechanical resistance of cell junctions, we treated MDCK II and claudin/JAM-A KO cells with a hypertonic solution to induce acute cell shrinking. Although hypertonic shock induced junction breakage in both MDCK II and claudin/JAM-A KO cells (Fig. 3, A–F and Video 3), quantification of the junction breakage showed that claudin/JAM-A KO cells junctions were more vulnerable to hypertonic shock (Fig. 3 G). These results indicate that claudins and JAM-A regulate the mechanical resistance of apical junctions.

Because the actin cytoskeleton plays a central role in regulating the apical junction integrity, we focused on the organization of the actin cytoskeleton during junction breakage. In control MDCK II cells, actin filaments visualized by phalloidin and myosin IIA and IIB were weakly localized at cell–cell junctions with additional staining in the apical cytoplasm (Fig. S2, A–E). In contrast, F-actin and myosin IIA and IIB were strongly localized in the extensively developed circumferential actin bundles underlying the apical junctional complex in claudin/JAM-A KO cells (Fig. S2, A′–E′). This localization resulted in the application of strong tension to AJs, as revealed by increased staining of the α-18 antibody, which recognizes an epitope exposed in α-catenin in its extended conformation (Fig. S2 F) (Yonemura et al., 2010), although the amount of total α-catenin at the cell–cell junctions was not significantly changed (Fig. S2 G). Vinculin, which is recruited to the extended α-catenin (Yonemura et al., 2010), was also strongly accumulated in claudin/JAM-A KO cells (Fig. S2 I), indicating that the cell–cell junctions were subjected to increased tension. Both α-18 and vinculin were reduced in the large gaps of ZO-1, confirming the disorganization of AJs at the junction breakage sites (Fig. S2, F–J).

To test whether junctional tension is elevated in claudin/JAM-A KO cells, laser ablation of the cell junctions in MDCK II and claudin/JAM-A KO cells expressing ZO-1-GFP was performed (Fig. 4, A and B). The recoil velocity of cell junctions upon laser ablation was greater in claudin/JAM-A KO cells (0.70 ± 0.08 μm/s in MDCK II cells, 1.99 ± 0.53 μm/s in claudin/JAM-A KO cells) (Fig. 4, C–E), demonstrating that claudin/JAM-A KO cells junctions were under higher tension.

Super-resolution imaging using stimulated-emission depletion (STED) revealed that the circumferential actin bundles were disorganized at the junction breakage sites in claudin/JAM-A KO cells. Although the circumferential actin bundles were coalesced and tightly associated with the intact cell–cell junctions in claudin/JAM-A KO cells (Fig. 5, A and B), the bundles became loosened at the junction breakage sites (Fig. 5, C–E, arrows). The actin bundles remained tightly attached to the cell–cell junctions where focal ZO-1 localization persisted (Fig. 5, D and E, arrowheads). Similar results were observed for myosin IIA (Fig. 5, F–J) and myosin IIB (Fig. 5, K–O). Furthermore, the actomyosin bundles from the adjacent cells were often separated from one another in the large gaps of ZO-1 staining (Fig. 5, C, H, and M, arrows). These results demonstrate that the junction breakages in claudin/JAM-A KO cells are accompanied by disorganization of the actin cytoskeleton.

We performed two-color time-lapse imaging to examine the spatiotemporal relationship between the actin organization and junction breakage by stably expressing ZO-1-GFP and LifeAct-mCherry in claudin/JAM-A KO cells. The time-lapse imaging revealed that the loosening of the junction-associated actin bundles occurred simultaneously with the breakage of the junctions, suggesting that the junction breakage was tightly correlated with the disorganization of the actin cytoskeleton (Fig. 6, A and B; and Video 4). Interestingly, the reassembly of the actin bundles preceded the relocalization of ZO-1-GFP during the junction repair process (Fig. 6, C and D; and Video 5). These results suggest that the actin cytoskeleton plays an important role in regulating the junction integrity.

To investigate whether the disorganization of the actin cytoskeleton has a causal role in regulating the junction integrity, we pharmacologically perturbed the actomyosin cytoskeleton. The RhoA-ROCK signaling pathway plays a central role in regulating the organization of the actomyosin cytoskeleton. The ROCK inhibitor Y-27632 did not affect the continuity of ZO-1 staining in control MDCK II cells (Fig. 7, A and B) but induced an exaggerated junction breakage phenotype in claudin/JAM-A KO cells (Fig. 7, B′ and F). ROCK was reported to stimulate myosin II–dependent contraction through phosphorylation of the myosin II regulatory light chain (Amano et al., 1996; Kimura et al., 1996) and to promote actin polymerization by inhibiting cofilin via LIMK (Maekawa et al., 1999; Yang et al., 1998). These observations prompted us to dissect the downstream signaling pathways of ROCK. The myosin II inhibitor blebbistatin did not modify the junction breakage phenotype in claudin/JAM-A KO cells (Fig. 7, C′ and F). However, inhibition of LIMK by BMS-5 induced discontinuity of the ZO-1 staining in claudin/JAM-A KO cells (Fig. 7, D′ and F), suggesting that ROCK-LIMK-cofilin dependent regulation of actin polymerization was important for the cell junction integrity. Consistent with this idea, two-color time-lapse imaging showed that the junction breakage in Y-27632 or BMS-5–treated cells was accompanied by disorganization of the circumferential actin bundles (Fig. 7, G and H; and Videos 6 and 7). Furthermore, latrunculin A, which inhibits actin polymerization, enhanced the junction breakage phenotype in claudin/JAM-A KO cells (Fig. 7, E′ and F) at concentrations that did not severely affect the junction integrity in control MDCK II cells (Fig. 5, E and F). The apical junction localization of RhoA strikingly increased in claudin/JAM-A KO cells (Fig. S3, A and B), and cofilin became concentrated along the cell junctions in claudin/JAM-A KO cells (Fig. S3, E and F), suggesting that the Rho-ROCK-LIMK-cofilin pathway is dysregulated in the absence of TJs. These results suggest that actin polymerization is required for the apical junctional complex integrity in claudin/JAM-A KO cells.

Previous reports suggested that cell junction defects could be repaired by local transient activation of Rho, termed Rho flares (Stephenson et al., 2019). To examine whether the repair of junction defects in claudin/JAM-A KO cells also involves this pathway, we examined the localization of RhoA. As noted above, RhoA strongly localized to cell junctions in claudin/JAM-A KO cells (Fig. S3A, B). Interestingly, RhoA was often localized at the edge of the broken junctions (Fig. S3, C and D). We further examined the temporal activation pattern of Rho by utilizing tdTomato-2×rGBD, which has been previously validated as a Rho activation reporter (Mahlandt et al., 2021). We confirmed that this probe could detect Rho activation during cytokinesis in both MDCK II and claudin/JAM-A KO cells (Fig. S3, G and H). We did not observe Rho flares in MDCK II cells, although focal activation of Rho beneath the remodeling tricellular junctions was occasionally observed (Fig. S3 I and Video 8). Weak and transient activation of Rho activation was observed around the junction breakage site during most junction repair events in claudin/JAM-A KO cells (Fig. S3 J and Video 9), although focal prominent activation of Rho was occasionally detected at the edge of the broken junctions during some junction repair events in claudin/JAM-A KO cells (Fig. S3 K and Video 10). These results suggest that Rho flares may account for at least part of the junction repair events in claudin/JAM-A KO cells.

The above results suggested that claudins and JAM-A together with the actin cytoskeleton regulate the mechanical resistance of the apical junctional complex. To further gain insight into how claudins and JAM-A regulate the junction integrity, we performed structure–function analyses of claudins and JAM-A using a mutant that inhibits the trans-interaction and strand formation of claudins (claudin-1[F147A]) (Piontek et al., 2008; Suzuki et al., 2014), a mutant that inhibits the cis-dimerization and trans-interaction of JAM-A (JAM-A[ΔDL1]) (Monteiro et al., 2014), and mutants that uncouple ZO-1 from claudins (claudin-1[ΔYV]) and JAM-A (JAM-A[ΔLV]) (Fig. 8 A). The inability of claudin-1[F147A] and JAM-A[ΔDL1] to trans-interact (Fig. S4, A–F), and the inability of claudin-1[ΔYV] and JAM-A[ΔLV] to recruit ZO-1 to cell–cell contacts (Fig. S4, G–R) were confirmed by expressing these molecules in L fibroblasts. Full-length claudin-1 and JAM-A and their mutants were stably expressed in claudin/JAM-A KO cells, and their ability to rescue the junction breakage phenotype was assessed. Full-length claudin-1 and JAM-A completely rescued the junction breakage phenotype in claudin/JAM-A KO cells (Fig. 8, B, E, and H). In contrast, neither claudin-1[F147A] nor claudin-1[ΔYV] rescued the junction breakage phenotype, and large gaps in ZO-1 staining were observed (Fig. 8, C, D, and H). JAM-A[ΔLV] also failed to rescue the junction breakage phenotype (Fig. 8, G and H). JAM-A[ΔDL1] suppressed the formation of large gaps but focal discontinuity in ZO-1 staining was frequently observed (Fig. 8, F and H). These results suggest that both the trans-interaction and the ability to interact with ZO-1 are important for the mechanical resistance of apical cell junctions.

The structure–function analyses of claudins and JAM-A suggested that the transmembrane linkage of neighboring cells via claudin and JAM-A and the actin cytoskeleton through ZO-1 was important for the mechanical resistance of cell junctions. Recent reports have indicated that ZO-1 may act as a mechanosensor and undergoes a conformational change in response to tension (Spadaro et al., 2017). These findings raise the possibility that claudins and JAM-A may regulate the mechanical resistance of the apical cell junctions by regulating the conformational status of ZO-1.

To investigate this possibility, we examined the molecular conformation of ZO-1 by adding distinct epitope tags to its N- and C-termini and measuring the distance between the termini using super-resolution STED microscopy (Fig. 9 A). The HA-ZO-1-Flag protein was functional as it was able to rescue the continuous apical localization of occludin when expressed in ZO-1/ZO-2 double KO cells (Fig. S5 B). MDCK II cells or claudin/JAM-A KO cells stably expressing ZO-1 with an N-terminal HA tag and a C-terminal Flag tag were cocultured with non-transfected cells, and the junctions between the transfected and non-transfected cells were observed (Fig. 9 A). In control MDCK II cells, the STED imaging revealed that the N-terminal HA tag was localized proximal to the membrane, while the C-terminal Flag tag was located more distally on the cytoplasmic side (Fig. 9 B). Measurement of the distance between the N- and C-termini showed that 73% of the ZO-1 molecules had >60 nm distance between the termini (Fig. 9 D), suggesting that ZO-1 had adopted an open conformation as reported previously (Spadaro et al., 2017). In claudin/JAM-A KO cells, the oriented localization of ZO-1 was maintained, with the N-terminus localized at the membrane-proximal region and the C-terminus located on the cytoplasmic side (Fig. 9 C). However, quantitative analysis revealed that the distance between the N- and C-termini of ZO-1 was shorter in claudin/JAM-A KO cells and only 53% of the ZO-1 molecules in claudin/JAM-A KO cells had >60 nm distance between the termini (Fig. 9 D), indicating that ZO-1 preferred a closed conformation in claudin/JAM-A KO cells. The line-scan curves of HA and Flag intensity were normalized and the overlapping area between the curves was measured to quantitatively evaluate the proximity of the two epitopes (Fig. 9 E). The overlap between the two curves was significantly greater in claudin/JAM-A KO cells (Fig. 9 F), indicating that the N-terminus and C-terminus were more proximal in claudin/JAM-A KO cells. These results suggest that claudins and JAM-A regulate the conformation of ZO-1 molecules.

Although the above results demonstrated that claudins and JAM-A regulate the conformation of ZO-1, the N-terminus of ZO-1 was still located at the membrane-proximal region in claudin/JAM-A KO cells, suggesting that other membrane proteins with PDZ-binding motifs may be involved in anchoring the N-terminal PDZ domains of ZO-1 to the membrane-proximal region in these cells. This prompted us to examine the contributions of occludin and CAR, as integral membrane proteins that are localized at TJs and interact with ZO-1 (Furuse et al., 1993, 1994; Cohen et al., 2001). For this, we generated claudin/JAM-A/occludin KO cells and claudin/JAM-A/CAR KO cells by CRISPR/Cas9-mediated genome editing and analyzed their phenotypes (Fig. S5 C).

ZO-1 with an N-terminal HA tag and a C-terminal Flag tag was stably expressed in claudin/JAM-A/occludin KO cells and claudin/JAM-A/CAR cells, and the conformation of ZO-1 was examined. STED imaging revealed that the molecular orientation of ZO-1 remained unaltered in claudin/JAM-A/occludin KO cells, with the N-terminus localized at the membrane-proximal region and the C-terminus located on the cytoplasmic side (Fig. 10, A and B). In contrast, the parallel orientation of ZO-1 molecules was severely perturbed in claudin/JAM-A/CAR KO cells, and the N- and C- termini of ZO-1 were randomly localized within the cell junctions (Fig. 10 C). The disordering of ZO-1 molecules in claudin/JAM-A/CAR KO cells was confirmed by quantification of the order parameter (Fig. 10 D). These results demonstrate that claudins, JAM-A, and CAR anchor the ZO-1 N-terminus to the membrane-proximal region and regulate the nanometer-scale ordering of ZO-1 molecules.

Finally, we examined the junction integrity in claudin/JAM-A/occludin KO and claudin/JAM-A/CAR KO cells. MDCK II, CAR KO, JAM-A KO, and claudin quintuple KO cells did not have the junction breakage phenotype (Fig. 10, E–H and M). Claudin/CAR KO cells showed spontaneous discontinuity of ZO-1 staining (Fig. 10, J and M, arrow), while claudin/JAM-A/occludin KO cells exhibited a slight enhancement of the junction breakage phenotype compared with claudin/JAM-A KO cells (Fig. 10, I, K, and M). In contrast, numerous large gaps in ZO-1 staining were observed in claudin/JAM-A/CAR KO cells (Fig. 10, L and M, asterisks). Taken together, these results suggest that TJ membrane proteins are required for the nanometer-scale ordering of ZO-1 and the apical cell junction integrity.

Epithelia must be able to resist mechanical stress to maintain the tissue integrity. Although AJs and desmosomes are well known to play important roles in the mechanical resistance of epithelia (Matsunaga et al., 1988; Amagai et al., 1991; Takeichi, 1991; Kintner, 1992; Harris et al., 2012), the roles of TJs remain unclear (Citi, 2019). In the present study, we utilized claudin/JAM-A KO cells, which exhibit specific and complete disruption of the TJ structure and function, and revealed that TJs are required for the mechanical resistance of the apical junctional complex. Quantitative analyses showed that loss of TJs increased the breakage frequency of cell junctions. Potential defects in junction repair cannot be ruled out due to the extremely low frequency of junction breakage in control cells.

AJs and the cadherin–catenin complex have been considered to play central roles in organizing the apical junctional complex because the perturbation of cadherin-mediated adhesion can severely impair apical junctional complex assembly (Gumbiner and Simons, 1986; Gumbiner et al., 1988; Watabe et al., 1994). However, the roles of TJs in apical junctional complex organization have remained unclear. ZO-1 and ZO-2 have been shown to regulate the organization of AJs and TJs (Umeda et al., 2006; Phua et al., 2014; Otani et al., 2019). It should be noted that although ZO-1 is specifically localized at TJs in mature epithelia, it can also be localized at AJs during junction assembly (Rajasekaran et al., 1996; Ando-Akatsuka et al., 1999), and it remains to be clarified whether the apical junctional complex disorganization in ZO protein-deficient cells reflects the functions of ZO proteins in TJs or AJs. Here, we examined claudin/JAM-A KO cells, in which the TJ structure and function are specifically perturbed, and demonstrated that TJ membrane proteins are essential for the apical junctional complex integrity, despite the presence of AJs and cadherin-mediated adhesion. These findings suggest that, in addition to AJs and the cadherin-catenin complex, TJs play an important role in regulating the apical junctional complex integrity. Because AJs and TJs have distinct mechanosensor molecules (Yonemura et al., 2010; Spadaro et al., 2017), the mechanism for how the AJs and TJs coordinately regulate the mechanical resistance of the apical junctional complex remains an interesting issue to address in future studies.

The present data revealed that TJ membrane proteins regulate the nanometer-scale ordering of ZO-1 molecules. Taken together with previous studies showing essential roles for ZO-1/2 in TJ organization (Umeda et al., 2006; Phua et al., 2014; Otani et al., 2019), the present findings suggest that reciprocal interactions between TJ membrane proteins and ZO proteins are important for the apical junctional complex integrity. It should be noted that ZO proteins still became localized at the apical junctions in the absence of TJ membrane proteins. AJs are present in claudin/JAM-A KO cells (Otani et al., 2019) and ZO-1 can localize to AJs in non-epithelial cells (Itoh et al., 1991, 1993), suggesting that ZO proteins could be localized to AJs in TJ membrane protein–deficient cells. Other ZO-interacting proteins, including α-catenin and afadin, may play important roles in the recruitment of ZO proteins to the apical junctions in the TJ membrane protein–deficient cells (Rajasekaran et al., 1996; Itoh et al., 1997; Yamamoto et al., 1997; Ooshio et al., 2010). Alternatively, ZO proteins may be recruited to the apical junctions by interacting with F-actin (Itoh et al., 1997).

Another intriguing finding of the present study is the importance of CAR for the apical junctional complex integrity. Although CAR was originally identified as an adenovirus and coxsackievirus receptor (Bergelson et al., 1997), subsequent studies demonstrated that it can also interact with ZO-1 and become localized at TJs (Cohen et al., 2001). It was shown that adenoviruses and coxsackie viruses infect epithelial cells through TJs using CAR as their receptors and that the epithelial barrier function is transiently disrupted during infection (Walters et al., 2002; Coyne and Bergelson, 2006). Cell culture studies further suggested a role for CAR in the epithelial barrier function (Cohen et al., 2001; Excoffon et al., 2004), and CAR KO mice exhibit complex phenotypes in their epithelial tissues (Pazirandeh et al., 2011). CAR is also expressed in the intercalated discs of the heart and its loss results in cardiac defects (Asher et al., 2005; Dorner et al., 2005; Chen et al., 2006; Lim et al., 2008). These results suggest that CAR functions in parallel with other junctional proteins to coordinately regulate the cell junction organization.

The present results demonstrated that TJ membrane proteins regulate the conformation of ZO-1. Importantly, the nanometer-scale organization of ZO-1 molecules was strongly correlated with the apical junctional complex integrity under mechanical stress, suggesting that the mechanosensitive conformational changes of ZO-1 may play important roles in regulating the mechanical resistance of the apical junctional complex. Because claudins, JAM-A, and CAR interact with the N-terminal PDZ domains in ZO-1 (Itoh et al., 1999, 2001; Bazzoni et al., 2000; Ebnet et al., 2000), TJ membrane proteins in conjunction with the actomyosin-dependent tensile force applied through the C-terminal actin-binding region of ZO-1 (Fanning et al., 1998, 2002; Belardi et al., 2020) may regulate the nanometer-scale ordering of ZO-1 molecules at cell junctions. In the TJ membrane protein–deficient cells, the ordering of ZO-1 molecules at cell junctions is compromised due to the absence of its N-terminal binding partners, resulting in a failure to maintain the apical junctional complex integrity under mechanical stress.

The question arises as to how the conformation of ZO-1 can regulate the mechanical resistance of apical junctions. At least three models can be postulated. First, it was reported that ZO-1 undergoes liquid–liquid phase separation and that this ability of ZO-1 to undergo phase separation is important for TJ formation (Beutel et al., 2019; Schwayer et al., 2019). It was also suggested that the extended conformation of ZO-1 could be a prerequisite for its phase separation (Beutel et al., 2019). These observations suggest that the elongated conformation of ZO-1 may promote ZO-1 condensate formation, which in turn promotes the apical junctional complex integrity. The nanometer-scale alignment of ZO-1 molecules may stabilize the elongated conformation of ZO-1 in a similar manner to polymer brushes, and in the absence of TJ membrane proteins, the extended conformation may become unstable, resulting in a catastrophic collapse of the structure upon application of excessive mechanical stress.

Second, the mechanosensitive conformational change of ZO-1 may act to dissipate excessive mechanical energy, thereby serving as a shock absorber for cell–cell junctions. It was shown that ZO-1 forms multiple intramolecular interactions, and that application of ∼2–4 pN force can disrupt these intramolecular interactions and lead to unfolding and extension of ZO-1 (Spadaro et al., 2017). These findings suggest that when mechanical stress is applied to the apical junctional complex, part of the mechanical energy is used to disrupt the intramolecular interactions of ZO-1, thus dissipating the excessive mechanical energy and protecting the cell junctions. In TJ membrane protein–deficient cells, the force-dependent conformational change of ZO-1 may be abrogated, and when the load exceeds a certain threshold, the apical junctions may undergo mechanical fracture.

Third, ZO-1 was shown to recruit the transcription factor DbpA to TJs in response to its conformational change (Spadaro et al., 2017). DbpA was also shown to regulate epithelial cell proliferation, lumen morphogenesis, and the monolayer integrity of retinal pigment epithelial cells (Balda et al., 2003; Sourisseau et al., 2006; Georgiadis et al., 2010). Although it remains unclear whether DbpA regulates the apical junctional complex integrity, the mechanosensitive recruitment of DbpA may modify its transcriptional response and control the apical junctional complex integrity in a homeostatic manner.

The present findings suggest that actin polymerization plays a critical role in maintaining the junction integrity when TJs are perturbed. We further demonstrated that actin reassembly preceded junction reformation during the junction repair process, suggesting that the junction repair occurs in an actin filament–guided manner, at least in claudin/JAM-A KO cells. Taken together with its well-established roles in junction assembly (Yonemura et al., 1995; Krendel and Bonder, 1999; Vasioukhin et al., 2000; Verma et al., 2004), these results suggest that the actin cytoskeleton plays pivotal roles in cell–cell junction repair and homeostasis. Consistent with this notion, it was reported that small breakages of TJs can be repaired by transient localized activations of Rho, termed Rho flares, that promote junction repair through actin polymerization and myosin II–dependent contraction (Stephenson et al., 2019). Rho often accumulated at the edges of broken junctions, and Rho flare–like transient focal activation of Rho adjacent to the junction breakage site was observed, suggesting that Rho flares may be involved in repairing the junctions in claudin/JAM-A KO cells. On the other hand, Rho flare was weak and transient during most junction repair events, although the junction repair was always preceded by the reassembly of circumferential actin bundles, suggesting that Rho-independent pathways may also exist to regulate the junction repair. These observations suggest that actin polymerization plays a pivotal role in regulating epithelial junction homeostasis.

The circumferential actin bundles were extremely well developed in claudin/JAM-A KO cells, suggesting that the loss of TJs promotes the circumferential actin bundle development in a compensatory manner to maintain the apical junction integrity. On the other hand, it is also possible that the increased tension results in a reduction of ductility of the cell junctions, increasing the risk of junction fracturing. Although the detailed signaling mechanism involved in the regulation of circumferential actin bundle formation in response to the loss of TJs remains to be fully clarified, RhoA was strongly localized to the apical junctions in claudin/JAM-A KO cells, suggesting that it may be involved in the circumferential actin bundle organization. Further dissection of the signaling pathways, including the molecules downstream of the TJ membrane proteins (Zihni et al., 2016) such as the mechanosensitive calcium channels, p114RhoGEF, p115RhoGEF, and Shroom3 (Choi et al., 2016; Haas et al., 2020; Varadarajan et al., 2022; Chumki et al., 2022), is required to clarify the molecular mechanisms that regulate the circumferential actin bundle development in claudin/JAM-A KO cells.

The present findings suggest that the ROCK-LIMK-cofilin pathway may regulate the mechanical resistance of apical junctions by promoting actin polymerization. Consistently, the RhoA-ROCK-LIMK-cofilin pathway was reported to promote actin assembly, resulting in stabilization of the apical junctional complex in colon carcinoma cells (Ito et al., 2017). Meanwhile, cofilin-dependent actin turnover was required to stabilize the newly formed junctions during the tissue tension-dependent junction remodeling in the Drosophila pupal wing (Ikawa and Sugimura, 2018), suggesting that the roles of cofilin may differ when the junctions are undergoing dynamic remodeling. ROCK-LIMK-cofilin–dependent turnover of the actin cytoskeleton may be involved in reorganization of the actin network to protect the junctions from catastrophic collapse when the junction integrity is challenged during morphogenesis.

In addition to the junction breakage triggered by spontaneous cell stretching, we also observed junction breakage during cell division. Interestingly, the cell division–associated junction breakage was not correlated with the local junctional strain, suggesting that junction breakage could be associated with the cell-cycle progression, although we cannot completely exclude the involvement of mechanical stress. As the junctional complex organization is regulated by phosphorylation (González-Mariscal et al., 2008), it is possible that the mitosis-associated protein phosphorylation may contribute to the junction breakage during cell division. Regardless of the underlying molecular mechanisms, the junction breakage during cell division in claudin/JAM-A KO cells suggests that TJs play important roles in maintaining the cell junctions during cell division. In addition to the junction breakage, apical cell area variance was increased in claudin/JAM-A KO cells, suggesting that TJs regulate epithelial cell packing. Heterogeneity in junctional tension or adhesive force may account for the heterogenous cell shape. On the other hand, the increase in junctional tension may lead to heterogenous cell packing as previously proposed (Staple et al., 2010).

In conclusion, we have demonstrated that TJ membrane proteins and the actin cytoskeleton coordinately regulate the apical junctional complex integrity when epithelial cells are mechanically challenged. Further analyses of the molecular mechanisms that regulate the coordinated response of cell junctions and the actin cytoskeleton to mechanical stress, as well as the quantitative measurements of the responses of cell junctions and the actin cytoskeleton to mechanical load, should deepen our understanding of epithelial homeostasis in the future.

MDCK II cells (Richardson et al., 1981) derived from the canine kidney were provided by Masayuki Murata (Tokyo Institute of Technology, Yokohama, Japan), and mouse L fibroblasts (L tk^–) (Murayama-Okabayashi et al., 1971; Nagafuchi et al., 1987) were provided by Masatoshi Takeichi (RIKEN Center for Biosystems Dynamics Research, Kobe, Japan). The establishment of claudin/JAM-A KO cells and ZO-1/ZO-2 double KO cells was described previously (Otani et al., 2019). All cells were maintained in low-glucose DMEM (#05919; Nissui) supplemented with 2 mM L-glutamine (#16948; Nacalai Tesque) and 10% FBS (#S1820-500; Bio-West) at 37°C under 5% CO₂. The medium was changed every 2–3 days. ROCK inhibitor Y-27632 (#030-24021; Wako) (Uehata et al., 1997) was used at 10 μM, myosin II inhibitor (−)-blebbistatin (#021-17041; Wako) (Straight et al., 2003) was used at 100 μM, LIMK inhibitor BMS-5 (#S0185; Selleck) (Ross-Macdonald et al., 2008) was used at 10 μM, and actin polymerization inhibitor latrunculin A (#428021; Calbiochem) (Coué et al., 1987) was used at 0.3 μM. The drugs were administered to confluent monolayers, and the cells were incubated for 1 h (latrunculin A) or 3 h (other inhibitors) before further analyses. The hypertonic shock was applied by adding D(−)-mannitol (#130-00855; WAKO) at a final concentration of 300 mM to the culture medium.

Canine ZO-1 was cloned from MDCK II cells using the following primers: 5′-TTG​CTA​GCA​TGT​CCG​CCA​GAG​CTG​CGG​C-3′, 5′-CCG​TCG​ACA​AAG​TGG​TCA​ATC​AGG​ACA​GAA​ACA​CAG​TT-3′. Superscript IV reverse transcriptase (#18090010; Invitrogen) was used for cDNA synthesis, and cloning was performed by using KOD-Plus-Neo (#KOD-401; TOYOBO). Amplified cDNA was cloned into pEGFP-N3, pCANw-Sal-EGFP (Ichii and Takeichi, 2007), or pCAH-Sal-EGFP, wherein the neomycin-resistance gene of pCANw-Sal-EGFP was replaced with a hygromycin resistance gene using an In-Fusion HD Cloning Kit (#Z9649N; Takara). pCANw-HA-cZO1-Flag was created by inserting an HA tag into the N-terminus of cZO-1 by PCR using the following primers: HA-forward: 5′-ATG​TAT​CCG​TAT​GAT​GTT​CCG​GAT​TAT​GCA​AAG​CTT​ATG​TCC​GCC​AGA​GCT​GCG​GCC​G-3′; and Flag-reverse: 5′-TCG​TCG​TCC​TTG​TAG​TCG​ACA​AAG​TGG​TCA​ATC​AGG​ACA​GAA​ACA​CAG​TTT​GCT​CCA​ACA​AGG-3′. The PCR product was inserted into pCANw-Sal-Flag, wherein the GFP tag in pCANw-Sal-EGFP was replaced with a Flag tag. LifeAct7-mCherry was a gift from Michael Davidson (Florida State University, Tallahassee, FL, USA) (plasmid #54491; Addgene) (Riedl et al., 2008), and was cloned to pCAB-Sal, wherein the neomycin-resistance gene in pCANw was replaced with a blasticidin S-resistance gene. tdTomato-2×rGBD was a gift from Dorus Gadella (University of Amsterdam, Amsterdam, Netherlands) (plasmid #176098; Addgene) (Mahlandt et al., 2021) and was cloned to pCANw-Sal-Flag, wherein a stop codon was inserted before the Flag tag.

Mouse claudin-1 (Furuse et al., 1998) and human JAM-A (Itoh et al., 2001) were cloned into pCANw. The mutant constructs were generated by PCR using KOD Plus ver.2 (#KOD-211; TOYOBO) and the following primer sets: Cl1-F147A-forward: 5′-TCA​AGA​AGC​CTA​TGA​CCC​CTT​GAC​C-3′ and Cl1-F147A-reverse: 5′-GGT​CAT​AGG​CTT​CTT​GAA​CAA​TTC​TGT​TTC​C-3′; Cl1-ΔYV-forward: 5′-GGG​AAA​GAC​TGA​CTC​GAG​TAC​AAG​G-3′ and Cl1ΔYV-reverse: 5′-CTC​GAG​TCA​GTC​TTT​CCC​ACT​AGA​AGG-3′, JAMA-ΔDL1-forward: 5′-GGC​ATT​GGG​CCT​CAT​CGT​GCT​TGT​GCC​TCC​ATC​CAA​GCC​TAC​AGT​TAA​C-3′ and JAMA-ΔDL1-reverse: 5′-GCA​CGA​TGA​GGC​CCA​ATG​CCA​GGG​AGC​ACA​ACA​GGA​TC-3′; JAMA-ΔLV-forward: 5′-ACC​TCA​AGC​TTC​TGA​GTC​GAC​TAC​AAG​GAC​GAC​G-3′ and JAMA-ΔLV-reverse: 5′-CGA​CTC​AGA​AGC​TTG​AGG​TCT​GTT​TGA​ATT​CTC​C-3′.

Claudin/JAM-A/occludin KO cells and claudin/JAM-A/CAR KO cells were generated by introducing Cas9-gRNA RNP complexes into claudin/JAM-A KO cells. CRISPR RNA and trans-activating CRISPR RNA were synthesized by IDT and annealed, and subsequently complexed with Cas9 (IDT) to form the Cas9-gRNA RNP complex. Next, 100 pmol of Cas9 and 120 pmol of gRNA duplex were introduced into 1 × 10⁵–1 × 10⁶ cells using a CUY21 Pro-Vitro electroporator (Nepagene) under the following conditions: prepulse, 150 V for 10 ms; postpulse, 10 × 20 V, 50 ms pulses at 50-ms intervals. The target sequences of the gRNAs were as follows (protospacer adjacent motif sequences are underlined): occludin: 5′-GCA​CCG​AGC​AAT​GAT​GTG​TACGG-3′; and CAR: 5′-ACC​CTT​AGT​CCA​GAA​GAC​CAGGG-3′. Electroporated cells were screened by immunofluorescence, and multiple KO cell clones were isolated and cultured. KO was verified by immunofluorescence, western blotting, and genomic sequencing, and all clones showed consistent phenotypes as for the phenotypes described in this work.

Stable transfectants were isolated after transfection of the expression vectors using Lipofectamine LTX reagent with PLUS reagent (#A12621; Thermo Fisher Scientific) in accordance with the manufacturer’s instructions. Transfected cells were selected using 500 μg/ml G418 (#16512-26; Nacalai Tesque), 150 μg/ml hygromycin B (#09287-84; Nacalai Tesque), or 2 μg/ml blasticidin S (#029-18701; Wako). Multiple surviving clones were isolated, and the expression of each transgene was confirmed by immunofluorescence and western blotting. At least three clones were isolated for each construct.

The following primary antibodies were used for immunohistochemistry and western blotting analyses: mouse monoclonal anti-ZO-1 (clone T-8754) (Itoh et al., 1991); rabbit polyclonal anti-ZO-1 (#61-7300; Thermo Fisher Scientific); rabbit polyclonal anti-ZO-2 (#38-9100; Thermo Fisher Scientific); rabbit polyclonal anti-claudin-1 (#51-9000; Thermo Fisher Scientific); rabbit polyclonal anti-human JAM-A (#PA5-120157; Invitrogen); rabbit polyclonal anti-CAR (kindly provided by Jeffrey M. Bergelson, University of Pennsylvania, Philadelphia, PA, USA) (Cohen et al., 2001); rat monoclonal anti-occludin (clone MOc37) (Saitou et al., 1997); rabbit polyclonal anti-occludin (Saitou et al., 1997); rat monoclonal anti-E-cadherin (clone ECCD-2; kindly provided by Masatoshi Takeichi) (Shirayoshi et al., 1986); rabbit polyclonal anti-1/s-afadin (#A0224; Sigma-Aldrich); rabbit polyclonal anti-α-catenin (#C2081; Sigma-Aldrich); mouse monoclonal anti-desmoplakin 1&2 (#65146; Progen); rat monoclonal α-18, which recognizes the extended conformation of α-catenin (kindly provided by Akira Nagafuchi, Nara Medical University, Kashihara, Japan) (Nagafuchi and Tsukita, 1994; Yonemura et al., 2010); rabbit polyclonal anti-MHCIIA (#M8064; Sigma-Aldrich); rabbit polyclonal anti-MHCIIB (#909901; BioLegend); rat monoclonal anti-RhoA (clone Lulu51; kindly provided by Shigenobu Yonemura, Tokushima University, Tokushima, Japan) (Yonemura et al., 2004); mouse monoclonal anti-cofilin (clone MAB-22; kind gift of Takashi Obinata, Chiba University, Chiba, Japan, or Kazumasa Ohashi, Tohoku University, Sendai, Japan) (Abe et al., 1989); mouse monoclonal anti-α-tubulin (clone DM1A; #14-4502-82; Invitrogen) (Blose et al., 1984); rat monoclonal anti-HA (clone 3F10; Roche); and mouse monoclonal anti-FLAG (#014-22383; Wako).

The following secondary antibodies were used: Alexa Fluor 488-conjugated donkey anti-mouse IgG (#A21202; Molecular Probes); Alexa Fluor 488-conjugated donkey anti-rabbit IgG (#A21206; Molecular Probes); Alexa Fluor 488-conjugated donkey anti-rat IgG (#A21208; Molecular Probes), Alexa Fluor 555-conjugated donkey anti-mouse IgG (#A31570; Invitrogen); Alexa Fluor 555-conjugated donkey anti-rabbit IgG (#A31572; Invitrogen); Alexa Fluor 555-conjugated goat anti-rat IgG (#A21434; Invitrogen); Alexa Fluor 647-conjugated donkey anti-mouse IgG (#A31571; Invitrogen); Alexa Fluor 647-conjugated phalloidin (#A22287; Invitrogen); sheep anti-mouse IgG HRP-conjugated whole antibody (#NA931V; Cytiva); donkey anti-rabbit HRP-conjugated F(a’b’)2 fragment (#NA9340V; Cytiva); and anti-rat IgG HRP-conjugated antibody (#NA935V; Cytiva).

For immunofluorescence of MDCK II cells and their derivatives, 1 × 10⁵ cells were cultured on Transwell polycarbonate filters (0.4 µm pore size; #3401; Corning) for 5–7 days. For immunofluorescence of L cells, 1.5 × 10⁵ cells were plated on coverslips in 35-mm dishes and cultured for 2 days. Cells were fixed with 1–2% PFA (RT, 15 min), 10% TCA (4°C, 15 min), or 100% methanol (−20°C, 10 min). Specifically, 1% PFA fixation was used for anti-α-catenin, α-18, anti-vinculin, anti-E-cadherin, anti-afadin, anti-HA, and anti-Flag; 2% PFA fixation was used for anti-ZO-1, anti-JAM-A, phalloidin, anti-myosin IIA, and anti-myosin IIB; TCA fixation was used for anti-RhoA and anti-ZO-2; and methanol fixation was used for anti-occludin. After fixation, the cells were washed three times with PBS, permeabilized with 0.1% TritonX-100 in PBS (RT, 15 min), rinsed once with PBS, and blocked with 10% FBS (RT, 30 min). For cofilin staining, cells were fixed with 4% PFA (RT, 20 min), washed three times with PBS, permeabilized with 100% methanol (−20°C, 5 min), rinsed three times with PBS, and blocked with 10% FBS (RT, 30 min). For cells cultured on Transwell filters, the filters were excised using scalpels. All cells were incubated with the primary antibodies diluted in blocking solution (RT, 1 h), washed three times with PBS, and incubated with secondary antibodies diluted in blocking solution (RT, 1 h). Finally, the cells were washed three times with PBS and mounted using FluoroSave Reagent (#345789; Calbiochem). For STED imaging, samples were mounted using ProLong Diamond Antifade Mountant (#P36961; Thermo Fisher Scientific) and covered with coverslips (#No.1S HT; 0.17 ± 0.005 mm; Matsunami).

Confocal images were obtained using an AX R confocal laser scanning microscope system mounted on an Eclipse Ti2 inverted microscope with a CFI PLAN Apochromat lambda D 60× oil (NA 1.42) objective and diode lasers (488/561/640 nm) accompanied by the NIS-Elements C Imaging software (all from Nikon Solutions).

For STED imaging, cells were imaged using a TCS SP8 STED confocal system mounted on a DMI8 inverted microscope with an HC PL APO CS2 100× oil (NA 1.4) objective and diode lasers (confocal: 488/555 nm; STED: 592/660 nm) accompanied by Application Suite X acquisition software (all from Leica Microsystems).

For time-lapse imaging, 1 × 10⁵ cells were plated at the center of glass-based dishes (#3910-035; Iwaki). After attachment, the cells were cultured in low glucose, phenol red-free DMEM (#08490-05; Nacalai Tesque), supplemented with 10% FBS and penicillin–streptomycin–glutamine (#10378-016; Gibco). Time-lapse imaging was performed using a CellVoyager CV1000 spinning disc confocal imaging system (Yokogawa) using 40× dry (NA 1.3) or 60× silicone (NA 1.35) objective lenses with diode lasers (488/561 nm). The cells were imaged at 1-min intervals for 10 h (for ZO-1-GFP/LifeAct7-mCherry two-color imaging and hypertonic shock imaging) with z-sectioning to 22 and 0.6-µm step size, or at 12 min intervals for 5 days (ZO-1-GFP movies) with z-sectioning to 22 and 1-µm step size. Movie acquisition was performed using CV1000 software (Yokogawa).

Laser ablation was performed using a spinning disc confocal microscope CSU-X1 (Yokogawa) mounted on an IX81 inverted microscope (Evident) with UPLSAPO 60×W/NA 1.20 water immersion objective (Evident) and a stage-top incubator (Tokai Hit; # INUBSF-FI-RK). MDCK II or claudin/JAM-A KO cells expressing ZO-1-GFP were imaged, and laser ablation was performed by N2 Micropoint laser (16 Hz, 365 nm, 3.4 μW; Photonic Instruments), and images were acquired by an EMCCD camera (#DU-897; Andor) at 400-ms intervals, starting from 4 s before the laser ablation to 30 s after the laser ablation. Laser ablation took 1 s and was not included in the movie.

All images were analyzed and processed using Fiji/ImageJ 1.53t software (National Institutes of Health). Median filter or brightness and contrast adjustments without any nonlinear adjustments were applied for qualitative presentation of images when necessary.

Junction breakage was quantified as follows. First, noise reduction was performed by application of band-pass and median filters, and the junction outlines were extracted by thresholding and binarization. After removal of outliers and manual removal of misannotated junctions, the image was skeletonized. Next, an averaging filter was applied to the skeletonized image, and the endpoints were extracted by thresholding and subsequent application of a band-pass filter to remove the edges of the bicellular junctions. The number of endpoints was counted using the “analyze particles” function, and the area of the region of interest was measured. Finally, the number of endpoints was divided by the area to yield the number of endpoints per unit area.

Cell stretching was quantified by measuring the outline of the cell of interest. The intensity of ZO-1-GFP was measured by measuring the intensity along the outline of the cell. The beginning of a stretching event was defined as the starting point, yielding L₀ and I₀, with ΔL and ΔI defined as the maximum relative length change and the maximum relative intensity change during a single stretching event, respectively. The degree of cell stretching was defined as 100 × ΔL/L₀ (%), and the change in ZO1-GFP intensity was defined as 100 × ΔI/I₀ (%). Cell stretching speed was defined as the maximum stretching speed (%/10 min) within a single stretching event, and cell stretching duration was defined as the length of a single stretching event (min).

The local strain was quantified by measuring the change in the length of the individual junctions (%) along the two consecutive frames of a time-lapse movie. Individual junctions were segmented and pseudo-colored according to the magnitude of strain.

Kymographs for the laser ablation experiments were generated using the KymoResliceWide plugin. The initial recoil velocity was quantified by measuring the changes in the distance between the vertices of the ablated junctions after laser ablation.

The distance between the N- and C-termini of ZO-1 was measured by performing a line scan across pairs of HA- and Flag-tag puncta. Only pairs that appeared as doublets were measured. The fluorescent intensity profile along the line scans was normalized and the overlapping area between the line scans was quantified. The distance between the fluorescence peaks of the HA and Flag tags was measured. The order parameter was quantified as follows. First, the angle between the N- and C-termini of ZO-1 was measured. Next, the mean angle of the particles within a single image was defined, and the deviation of individual vectors from the average vector was measured to determine the relative angle θ. Finally, the order parameter S = (3cos²θ − 1)/2 was calculated for each vector.

Graph generation and statistical tests (unpaired two-tailed t test or one-way ANOVA with Tukey post hoc test or Bonferroni’s post hoc test) were performed using Excel software (Microsoft) or R (R foundation).

Cells grown on 35-mm dishes were rinsed once with ice-cold PBS, lysed with 200 µl of Laemmli sample buffer supplemented with 100 mM DTT, sonicated, and boiled at 97°C for 5 min. Samples were separated by SDS-PAGE using standard methods with 20 µl of sample loaded per lane. The separated proteins were transferred to 0.45-μm pore Protran nitrocellulose transfer membranes (#10-401-196; Whatman). The membranes were blocked with 5% skim milk in TBS/0.1% Tween-20 for 1 h at RT and incubated with primary antibodies diluted in the blocking solution overnight at 4°C. The membranes were rinsed three times with TBS/0.1% Tween-20 and incubated with secondary antibodies diluted in blocking solution for 1 h at RT. After three washes with TBS/0.1% Tween-20, signals were detected by chemiluminescence using ECL Prime Western Blotting Detection Reagents (#RPN2232; Cytiva). Images were obtained using an LAS3000 mini system (Fujifilm), and data were processed and analyzed using Fiji/ImageJ 1.53t software.

Fig. S1 shows the junction morphology and dynamics in claudin/JAM-A KO cells. Fig. S2 shows the actomyosin organization in claudin/JAM-A KO cells. Fig. S3 shows RhoA and cofilin localization and Rho activation in MDCK II and claudin/JAM-A KO cells. Fig. S4 shows the characterization of claudin-1 and JAM-A mutants. Fig. S5 shows the characterization of ZO-1 constructs and claudin/JAM-A/occludin and claudin/JAM-A/CAR KO cells. Video 1 shows the dynamics of ZO-1-GFP in MDCK II cells. Video 2 shows the junction breakage events in claudin/JAM-A KO cells expressing ZO-1-GFP. Video 3 shows the response to hypertonic shock in MDCK II and claudin/JAM-A KO cells. Video 4 shows the actin dynamics during junction breakage observed by two-color time-lapse imaging of claudin/JAM-A KO cells expressing ZO-1-GFP and LifeAct-mCherry. Video 5 shows the actin dynamics during junction repair observed by two-color time-lapse imaging of claudin/JAM-A KO cells expressing ZO-1-GFP and LifeAct-mCherry. Video 6 shows the actin dynamics during junction breakage in Y-27632–treated claudin/JAM-A KO cells expressing ZO-1-GFP and LifeAct-mCherry. Video 7 shows the actin dynamics during junction breakage observed by two-color time-lapse imaging of BMS-5–treated claudin/JAM-A KO cells expressing ZO-1-GFP and LifeAct-mCherry. Video 8 shows Rho activation during junction remodeling in MDCK II cells expressing ZO-1-GFP and tdTomato-2×rGBD. Video 9 shows transient focal Rho activation during junction repair in claudin/JAM-A KO cells expressing ZO-1-GFP and tdTomato-2×rGBD. Video 10 shows Rho activation at the edge of a broken junction in claudin/JAM-A KO cells expressing ZO-1-GFP and tdTomato-2×rGBD.

The data underlying this study are available from the corresponding authors upon reasonable request.

We thank Akira Nagafuchi (Nara Medical University, Kashihara, Japan), Masayuki Murata (Tokyo Institute of Technology, Yokohama, Japan), Jeffrey M. Bergelson (University of Pennsylvania, Philadelphia, PA, USA), Masatoshi Takeichi (RIKEN Center for Biosystems Dynamics Research, Kobe, Japan), Takashi Obinata (Chiba University, Chiba, Japan), Kazumasa Ohashi (Tohoku University, Sendai, Japan), Michael Davidson (Florida State University, Tallahassee, FL, USA), Dorus Gadella (University of Amsterdam, Amsterdam, Netherlands), Shigenobu Yonemura (Tokushima University, Tokushima, Japan), and the Developmental Studies Hybridoma Bank for kindly providing reagents; Mika Watanabe and Yuichiro Kano for technical assistance; the Bioimaging Facility of National Institute for Basic Biology and Misako Saida for their support in laser ablation experiments; Kaoru Sugimura for advice on laser ablation experiments; Tomohito Higashi for advice on Rho flare experiments; and all members of the Furuse laboratory for discussions and comments. We also thank Alison Sherwin from Edanz for editing a draft of this manuscript.

This work was supported by a Japan Society for the Promotion of Science (JSPS) Grant-in-Aid for Challenging Exploratory Research (16K15226 to M. Furuse), JSPS Grants-in-Aid for Scientific Research (B) (26291043, 18H02440, 21H02523 to M. Furuse), JSPS Grant-in-Aid for Scientific Research (C) (18K06234 T. Otani), JSPS Grants-in-Aid for Young Scientists (B) (16K18544 to T. Otani; 21K15095 to S. Fujiwara), Ministry of Education, Culture, Sports, Science and Technology (MEXT)/JSPS Grant-in-Aid for Scientific Research on Innovative Areas (17H05627 to T. Otani), MEXT/JSPS Grant-in-Aid for Transformative Research Areas (21H05286, to T. Otani), Japan Science and Technology Corporation, Precursory Research for Embryonic Science and Technology (JPMJPR204, to T. Otani); MEXT/JSPS Grant-in-Aid for Scientific Research on Innovative Areas—Platforms for Advanced Technologies and Research Resources “Advanced Bioimaging Support” (JP16H06280, JP22H04926), the Inamori Foundation (to T. Otani), the Japan Spina Bifida and Hydrocephalus Research Foundation (to T. Otani), and the Takeda Science Foundation (to M. Furuse and T. Otani).

Author contributions: Conceptualization, Funding acquisition, Supervision, Project administration: T. Otani and M. Furuse; Resources: T. Otani, T.P. Nguyen, M. Tsutsumi, N. Kinoshita, S. Fujiwara, T. Nemoto, T. Fujimori, and M. Furuse; Formal analysis: T. Otani and T.P. Nguyen; Investigation, Visualization: T.P. Nguyen and T. Otani; Writing—original draft: T. Otani, T.P. Nguyen, and M. Furuse; Writing—review & editing: all authors.