Synaptopodin couples epithelial contractility to α-actinin-4–dependent junction maturation

The epithelial junction experiences mechanical force from an array of cellular processes such as tugging force from cellular contractions (Ganz et al., 2006; Ladoux et al., 2010; Borghi et al., 2012), hydrostatic force from intracellular osmotic pressure (Papakonstanti et al., 2000; Di Ciano et al., 2002; Thirone et al., 2009; Stewart et al., 2011; Jiang and Sun, 2013), and shear stress from cytoplasmic streaming (Iwasaki and Wang, 2008; Keren et al., 2009). Cell–cell adhesion also experiences mechanical stress from extrinsic stimuli such as shear stress from extracellular fluid flow (Tzima et al., 2005; Duan et al., 2008) and hydrostatic pressure from the surrounding tissue (Lorentz et al., 1972; Knight et al., 2006). Fluctuation of intercellular tension can be created by changes of intracellular and extracellular osmotic pressures in disease states such as diabetes (Hsueh and Anderson, 1992; Goel et al., 2007). Moreover, inhibition or stimulation of the cellular contractile system can alter the tension applied to cell–cell adhesions (Smutny et al., 2010; Engl et al., 2014; Hoj et al., 2014). Thus, the levels of tension exerted on epithelial junction vary depending on the physiological states of the body.

The ability of epithelial junction to withstand mechanical stress depends on many factors, including the adhesiveness of cell–cell adhesion proteins (Harrison et al., 2012; Leckband and Sivasankar, 2012; Rikitake et al., 2012; Samanta et al., 2012; Sivasankar, 2013) and the stability of the junctional complex (Sato et al., 2006; Ishiyama et al., 2010; Tang and Brieher, 2013), which is intimately coupled to the attachment of the junctional complex to the actin cytoskeleton (Abe and Takeichi, 2008; Ting et al., 2012; Desai et al., 2013; Hong et al., 2013; Huveneers and de Rooij, 2013; Twiss and de Rooij, 2013; Buckley et al., 2014). In addition, the epithelial junction can respond to changes of mechanical stress and adjust its functions (Gomez et al., 2011; Maître and Heisenberg, 2011; Leerberg et al., 2014). For example, cell–cell adhesion has to become stronger when faced with increased mechanical stress such as elevated blood pressure in hypertension (Preston et al., 2002; Falqui et al., 2007) or during exercise (Goel et al., 2007). Dissecting the complex relationships between mechanical force and cell–cell adhesion is becoming essential for understanding epithelial physiology and the regulation of cell junction in health and disease.

The apical junctional complex in epithelial cells, originally described using EM, consists of morphologically distinct cell–cell contacts (Farquhar and Palade, 1963), including the tight junction, defined morphologically to contain intermittent kisses from the outer leaflets of apposing plasma membranes, and the adherens junction, defined morphologically to contain extracellular spacers of ∼15–40 nm formed between apposing cells (Hirokawa, 1980; Hirokawa and Heuser, 1981; Miyaguchi, 2000; Franke, 2009; Meng and Takeichi, 2009). The tight and adherens junctions are characterized by cytoplasmic membrane-associated electron densities of ∼150 nm and attachment to cytoskeletal structures (Farquhar and Palade, 1963; Hirokawa and Heuser, 1981; Hirokawa and Tilney, 1982; Hirokawa et al., 1982, 1983; Hirokawa, 1986; Madara et al., 1986). In differentiated epithelial cells, the tight and adherens junctions are positioned next to each other but can reorganize in dynamic cellular processes such as disassembly and reassembly of junctions during wound healing, intercellular neighbor exchanges during morphogenetic movement, and lateral mixing of junctional components in cell extrusion (Boyer et al., 1989; Madara, 1990; Collares-Buzato et al., 1998; Tamada et al., 2007; Ebnet, 2008; McGill et al., 2009; Collinet and Lecuit, 2013). The tight and adherens junctions in cells of epithelial lineage such as endothelial cells and kidney podocytes are positioned in close proximity and frequently intermingled (Schnabel et al., 1990; Wolburg et al., 1994; Adamson et al., 1998; Underwood et al., 1999; Reiser et al., 2000; Rüffer et al., 2004; Fukasawa et al., 2009; Tornavaca et al., 2015).

Junctional complexes found in epithelial cells are heterogeneous in composition and size (Hinck et al., 1994; Näthke et al., 1994; McGill et al., 2009; Wu et al., 2015). Proteomics studies have identified hundreds of proteins associating with the tight and adherens junctions (Catimel et al., 2005; Tang, 2006; Guo et al., 2014; Toret et al., 2014; Van Itallie et al., 2014), which can be categorized into functional modalities (Tang, 2006; Guo et al., 2014; Toret et al., 2014; Padmanabhan et al., 2015). Thus, the key to understanding epithelial junctions lies in dissecting the dynamical interactions and regulations of individual proteins within the junctional complexes and among the different functional modalities (Howarth and Stevenson, 1995; Müller et al., 2005; Choi et al., 2009; Garbett and Bretscher, 2014; Pokutta et al., 2014; Viswanatha et al., 2014).

Engagement of cadherin, a key intercellular adhesion molecule at the adherens junction (Sedar and Forte, 1964; Behrens et al., 1985; Vestweber and Kemler, 1985; Volk and Geiger, 1986a; Hirano et al., 1987), initiates biochemical events that lead to structural organization and actin attachment, resulting in strong cell–cell adhesion with a functional epithelial permeability barrier (Gumbiner and Simons, 1986; Madara et al., 1986; Gumbiner et al., 1988; Siliciano and Goodenough, 1988; Capaldo and Macara, 2007). Cadherin organizes the adherens junctional domain through self-association (Yap et al., 1997, 1998; Katz et al., 1998; Strale et al., 2015), as well as by interacting with other cell–cell adhesion complexes such as nectins (Takahashi et al., 1999; Miyahara et al., 2000; Yamada et al., 2004) and Epcam (Litvinov et al., 1997; Winter et al., 2003). Cadherin interacts directly with p120-catenin (Reynolds et al., 1994; Shibamoto et al., 1995), β-catenin (Ozawa et al., 1989), and plakoglobin (Knudsen and Wheelock, 1992), as well as indirectly with α-catenin via β-catenin (Ozawa et al., 1990; Ozawa and Kemler, 1992; Aberle et al., 1994) to form the core cadherin–catenin complex. The catenins interact with many junctional proteins including ZO-1 (Rajasekaran et al., 1996; Itoh et al., 1997; Maiers et al., 2013), afadin (Ikeda et al., 1999), α-actinin (Knudsen et al., 1995; Nieset et al., 1997; Hazan and Norton, 1998), vinculin (Hazan et al., 1997; Watabe-Uchida et al., 1998; Weiss et al., 1998), EPLIN (Abe and Takeichi, 2008; Taguchi et al., 2011), and MAGI (Dobrosotskaya and James, 2000; Nishimura et al., 2000; Stetak and Hajnal, 2011), which can interact with each other (Yamamoto et al., 1997; Ebnet et al., 2000; Ooshio et al., 2010), providing multiple and overlapping interfaces within the apical junctional complex (Takahashi et al., 1999; Tachibana et al., 2000; Yokoyama et al., 2001; Ooshio et al., 2004; Peng et al., 2010).

The cadherin–catenin complex plays an essential role in linking cell–cell adhesions to actin structures (Hirokawa and Heuser, 1981; Volk and Geiger, 1986b; Hirano et al., 1987; Nagafuchi and Takeichi, 1988; Jaffe et al., 1990; Matsuzaki et al., 1990; Balsamo et al., 1991; Miyaguchi, 2000; Zhang et al., 2005; Lambert et al., 2007). Several distinct actin structures are present at the apical junction in epithelial cells, including a centrally positioned actomyosin belt separating two additional actin networks positioned immediately at the level of the tight and adherens junctions (Tamada et al., 2007; Ebrahim et al., 2013). In addition, intercellular junctions between multiple cells form a specialized vertex structure, the tricellular junction (Walker et al., 1985), containing additional junctional proteins (Ikenouchi et al., 2005; Byri et al., 2015) and distinct actin arrangements (Yonemura, 2011), that plays an important role in junctional and epithelial dynamics (Oda et al., 2014). The apical junction also contains a population of stabilized actin associated with the latrunculin-resistant junctional complexes (Yamada et al., 2004; Abe and Takeichi, 2008; Cavey et al., 2008; Tang and Brieher, 2012). Distinct pools of actin are thought to associate with junctional complexes on the lateral membrane (Drenckhahn and Franz, 1986; Yonemura et al., 1995). During junction remodeling, actin reorganizes into bundled cables that can position perpendicular (Vasioukhin et al., 2000; Kishikawa et al., 2008; Taguchi et al., 2011) or parallel (Bement et al., 1993; Mandato and Bement, 2001; Tamada et al., 2007) to the plasma membrane at cell–cell adhesions. Junctional actin structures can be controlled by local and global actin regulation, including actin polymerization (Kobielak et al., 2004; Carramusa et al., 2007; Ryu et al., 2009; Bernadskaya et al., 2011; Verma et al., 2012; Grikscheit et al., 2015), actin depolymerization (Lin et al., 2010; Chu et al., 2012; Slee and Lowe-Krentz, 2013), and actin stabilization (Weber et al., 2007; Cox-Paulson et al., 2012; Tang and Brieher, 2013).

In addition to biochemical inputs, the cadherin–catenin complex provides a platform for biophysical inputs (Miyake et al., 2006; Ladoux et al., 2010; Liu et al., 2010) by integrating actin dynamics and actomyosin activities (Clark et al., 2009; Fernandez-Gonzalez et al., 2009; Saravanan et al., 2013). Ablation of myosin II or disruption of myosin II activities compromises cell–cell adhesion and the development of a mature junction (Conti et al., 2004; Shewan et al., 2005; Lambert et al., 2007; Kishikawa et al., 2008; Franke et al., 2010; Reyes et al., 2014). In turn, cadherin-mediated adhesions are essential for modulating the activities of the cellular actomyosin network and propagating mechanical force to neighboring cells, ultimately determining the overall contractile properties of the epithelial tissue (Harris et al., 2012, 2014; Kuipers et al., 2014). Thus, actin dynamics can directly influence the behavior of the cellular actomyosin network and indirectly through its effect on cell–cell adhesions to control the barrier function and mechanical properties of epithelial cells (Fischer et al., 2014). Therefore, the development of epithelial cell–cell junctions is controlled by both biochemical inputs via protein–protein interactions and assembly of junctional complexes as well as biophysical inputs via actomyosin behavior and actin dynamics.

Mechanistically, how biochemical and biophysical inputs regulate the assembly and maturation of epithelial junctional complexes is unclear but likely to depend on multiple and sequential molecular events. Cadherin has different binding affinity states that can be influenced by its interactions with the catenins (Yap et al., 1998; Imamura et al., 1999; Thoreson et al., 2000; Chu et al., 2004; Bajpai et al., 2009, 2013; Petrova et al., 2012; Shashikanth et al., 2015). Regulating the interaction between cadherin and the catenins provides one level of regulation (Qin et al., 2005). Regulating the interaction between actin filaments and the cadherin–catenin complex provides another level of regulation (Buckley et al., 2014). Regulating the interactions between the cadherin–catenin complex and other junctional actin-binding components such as α-actinin and vinculin provides additional control over complex cellular processes (Hazan and Norton, 1998).

Both α-actinin and vinculin bind to a central region of α-catenin (Knudsen et al., 1995; Nieset et al., 1997; Imamura et al., 1999) that is necessary for promoting force-induced strengthening of cadherin-mediated adhesion (le Duc et al., 2010; Huveneers and de Rooij, 2013; Thomas et al., 2013; Barry et al., 2014), stimulation of junction development (Watabe-Uchida et al., 1998; Twiss et al., 2012), and organization of junctional actin structures (Taguchi et al., 2011; Huveneers and de Rooij, 2013). The vinculin/α-actinin–binding domain (VAABD) of α-catenin is usually buried within the cadherin–catenin complex (Ishiyama et al., 2013) but can be exposed/opened by exogenous mechanical force (Yao et al., 2014; Kim et al., 2015) or endogenous actomyosin contractility (Yonemura et al., 2010). Unfolding of α-catenin can be triggered by a combination of actin filament and a fragment of vinculin, corresponding to a region that is usually buried in the full-length molecule (Rangarajan and Izard, 2012). Exposure of the buried α-catenin–binding site on vinculin requires unfolding of vinculin from an autoinhibited closed configuration (Johnson and Craig, 1995a,b; Bakolitsa et al., 2004), which can be stimulated by a combination of actin filaments and a fragment of α-catenin, corresponding to the central region of α-catenin that is usually buried (Choi et al., 2012; Peng et al., 2012). Thus, in addition to the requirement of actin filament for the activation of α-catenin and vinculin, either α-catenin or vinculin needs to be in a preactivated open configuration before they can activate the other. The complexity of junction regulation is further revealed by the findings that vinculin shows a delayed response to tension-induced conformational changes in α-catenin (Kim et al., 2015) and does not have an obligatory role in cell–cell adhesion and morphogenetic events (Ozawa, 1998; Maiden et al., 2013). Less is known about the role of α-actinin in cell–cell adhesion and junction regulation.α-Actinin colocalizes with the cadherin–catenin complex, vinculin, and actin at the apical junction and lateral membrane of epithelial cells (Drenckhahn and Franz, 1986; Tang and Brieher, 2013). Upon junction maturation, α-actinin preferentially colocalizes with the cadherin–catenin complex at the apical junction and becomes TX-100 insoluble (Tang and Brieher, 2013). Exogenous expression of a chimera E-cadherin–α-catenin protein in CHO cells results in accumulation of α-actinin to sites of cell–cell contacts, indicating that E-cadherin–α-catenin plays an instructive role in α-actinin recruitment (Tang and Brieher, 2012).

α-Actinin contains an actin-binding domain (Otey and Carpen, 2004; Sjöblom et al., 2008), a central spectrin repeat that can bind to α-catenin and vinculin (McGregor et al., 1994; Nieset et al., 1997), and a tail region with EF hands that confers calcium and phosphoinositide regulation (Rosenberg et al., 1981; Noegel et al., 1987; Witke et al., 1993; Tang et al., 2001; Corgan et al., 2004; Franzot et al., 2005). α-Actinin exists as an antiparallel dimer via high-affinity interactions between the spectrin repeats (Flood et al., 1997) and thus acts as an actin filament cross-linker. The mechanical and dynamical properties of actin filament gels is influenced by the affinity of α-actinin to actin (Wachsstock et al., 1993; Broedersz et al., 2010; Yao et al., 2011, 2013), which can be regulated by α-actinin phosphorylation (Zhang et al., 2005; Travers et al., 2013). The actin-binding domain of α-actinin is required for its actin assembly activity (Tang and Brieher, 2012). However, the central spectrin repeats of α-actinin alone are sufficient for junctional targeting (Hijikata et al., 1997), indicating that α-actinin is recruited to the membrane via junctional proteins other than actin filaments. α-Actinin can interact with vinculin (Belkin and Koteliansky, 1987; Wachsstock et al., 1987; McGregor et al., 1994; Kelly et al., 2006) and is capable of activating vinculin in the absence of actin filament (Bois et al., 2005, 2006). The vinculin-binding site of α-actinin is located within the spectrin repeat of α-actinin, which is usually buried in between the dimer interfaces of α-actinin (Bois et al., 2005, 2006). However, spectrin repeats are flexible and can form stable stretched intermediates under force to expose the vinculin-binding site (Rief et al., 1999; Lenne et al., 2000; Ylänne et al., 2001; Ortiz et al., 2005). Binding of α-actinin to vinculin can dramatically shift the energy landscape of vinculin to expose multiple protein-interacting regions, including a binding site for filamentous actin on vinculin (Bois et al., 2005, 2006). Consequently, binding of vinculin to the vinculin-binding site of α-actinin results in conformational changes throughout the rest of α-actinin (Shams et al., 2012), potentially contributing to tension sensing and altering the overall protein–protein interaction landscape at the junction (Altmann et al., 2002; Law et al., 2003; Hampton et al., 2007). Thus, α-actinin may be part of a mechanotransduction mechanism for vinculin recruitment to the cadherin–catenin complex. Besides binding to α-catenin and vinculin, α-actinin also interacts with other junctional proteins, including ADIP (Asada et al., 2003), LMO7 (Ooshio et al., 2004), LPP (Hansen and Beckerle, 2008), MAGI (Patrie et al., 2002), and β-catenin (Catimel et al., 2005; Gujral et al., 2013).

In this study, we provide evidence for a central role of α-actinin in the regulation of vinculin and actin accumulation at the junction. We have identified synaptopodin as an upstream regulator of α-actinin that modulates junction maturation through α-actinin–dependent strengthening of cell–cell adhesion and actin assembly.

Mature epithelial junction in polarized epithelial cells is a membrane specialization that associates with a cytoplasmic density of ∼150–200 nm (Fig. S1 A). Super-resolution microscopy shows that E-cadherin at the apical junction forms clusters of ∼200–250 nm (Wu et al., 2015). Our goal is to understand the assembly and regulation of this junctional complex. We have previously developed a biochemical protocol to study actin assembly using purified junctional membranes (Fig. S1, B and C). Correlating structural information using EM and biochemical information from ex vivo samples is fundamental to the discoveries of both the molecular composition and structure–function relationships in the junction field, including the adherens, tight, and gap junctions (Goodenough and Revel, 1970, 1971; Goodenough, 1975; Kensler and Goodenough, 1980; Colaco and Evans, 1981; Stevenson and Goodenough, 1984; Stevenson et al., 1986; Tsukita and Tsukita, 1989; Tsukita et al., 1989a,b; Itoh et al., 1991; Stauffer et al., 1991). EM approaches can generate information to fill the resolution gap between light microscopy and crystallography. In addition, EM has recently been used to determine structural organization of proteins within a large macromolecular complex (Huang et al., 2014; Schlager et al., 2014; Shukla et al., 2014; Staals et al., 2014; Durand et al., 2015; Peisley and Skiniotis, 2015). Thus, we would like to get some basic understanding of the overall structural organization of the junctional complex to aid the design of our experiments. For example, whether the complexes are stable and how heterogeneous the complexes are will determine the approaches and the conclusions that we can draw. Using immunogold labeling of α-actinin-4, we found that α-actinin-4 is localized to discreet domains on the purified membranes (Fig. S1 A). Exogenous addition of actin monomers to the purified membranes results in actin polymerization and attachment (Fig. S1, E–G) at sites of α-actinin accumulation (Fig. S1 H), indicating that actin assembly is coupled to filament association with the junctional complex.

We have previously shown, using immunofluorescence microscopy, that E-cadherin, α-catenin, β-catenin, and α-actinin-4 colocalize to the junctional complex on TX-100–extracted membranes and in cells after detergent TX-100 extraction (Tang and Brieher, 2012, 2013). Here, we show that the apical junction retains a cytoplasmic density of ∼150 nm under the electron microscope after extraction with TX-100 (Fig. S1 I). Detergent-extracted junctional complexes that contain α-actinin-4 are heterogeneous in size, ∼100–200 nm (Fig. S2, A and B). Detergent-extracted junctional complexes can support actin assembly in the presence of exogenously added actin monomers and show multiple interactions with actin filaments (Fig. S2, C and D). In the absence of detergent, junction-enriched membranes can be found to consist of discreet globular densities aligned in a row where a single actin filament is being held (Fig. S2, E and F). Several actin-binding proteins are associated with cell junctions, including α-actinin, vinculin, α-catenin, eplin, and afadin. Dissecting the structural organization and assembly of the junctional complex is essential for understanding the molecular mechanisms governing actin assembly and epithelial junction development.

The formation of a membrane complex of this size cannot be trivial even for a cell, which can synthesize and fold all of the essential protein components. Multiple molecular and biochemical events are most likely to be required for individual components to come together in the right configurations within the context of the membrane and cell–cell interactions. Our goal is to understand how cells assemble the membrane junctional complex by first characterizing the overall temporal and spatial parameters for cell junction formation.

During normal epithelial maturation, actin and α-actinin-4 accumulated at the cell junction over a period of several days (Fig. 1 A). By the second day post-confluence (2 dpc), most canonical junctional components, E-cadherin, α-catenin, and β-catenin, p120-catenin, and ZO-1, were already present (Fig. 1, A and B). However, vinculin has not been targeted at this early stage of junction development (Fig. 1 B). By 5 dpc, α-actinin-4 and vinculin became localized to the cell junction (Fig. 1 B). During this maturation period, the permeability barrier of the epithelial cell monolayer gradually formed (Fig. 1 C). Knockdown of α-actinin-4 prevented this maturation process and compromised the development of the barrier function (Fig. 1 C), indicating that α-actinin-4 recruitment is part of a normal maturation process during junction development. We have previously built a pressure chamber device that can deliver hydrostatic pressure to the intercellular junction to study the strength of cell–cell adhesion in an epithelial cell monolayer (Tang and Brieher, 2013). Using this setup, we show that intercellular stress induces α-actinin-4 knockdown cells to break away from each other (Fig. 1 D) and eventually detach from the monolayer (Fig. 1 E), suggesting that cell–cell adhesion has been compromised.

Knockdown of α-actinin-4 abolished junctional accumulation of vinculin (Fig. 1 F) without changing the cellular levels of vinculin (Fig. 1 G). Thus, α-actinin-4 is incorporated into the maturing junctional complex before vinculin targeting and behaves as an upstream regulator of vinculin at the cell junction during junction maturation.

We have previously built a pressure chamber device that can deliver hydrostatic pressure to the intercellular junction to study the strength of cell–cell adhesion in an epithelial cell monolayer (Tang and Brieher, 2013). Using this setup, we show that intercellular stress induces α-actinin-4 knockdown cells to break away from each other (Fig. 1 F) and eventually detach from the monolayer (Fig. 1 G), suggesting that cell–cell adhesion has been compromised.

To further characterize the temporal and spatial regulation of α-actinin-4 during junction maturation, we imaged α-actinin live in MDCK cells (Fig. 2). We observed two main characteristics: (1) junction mobility along the length of the junction and (2) junctional deflection perpendicular to the junction. First, the length within the individual junction oscillated during junction maturation (Fig. 2, A, B, D, and E; and Videos 1, 2, and 3). But as the junction matured, these fluctuations subsided (Fig. 2, F and G). Application of blebbistatin abolished the length oscillations (Fig. 2 H), indicating that the behavior is caused by myosin II activities, which is in agreement with the presence of myosin II–dependent tension along the length of epithelial junction (Wu et al., 2014).

More intriguingly, we observed junctional deflection perpendicular to the plasma membrane (Fig. 2, I and J; and Videos 4, 5, and 6), which lead to a transient decrease or increase in the apical diameter of the cell (Fig. 2 K). Fluctuations in apical cell radius have also been described in dynamic cell rearrangements (Blanchard et al., 2010), and variations in apical cell diameters have been shown to correlate with cell shape changes during morphogenetic cell movements (Roh-Johnson et al., 2012; Razzell et al., 2014). In addition, these junctional deflections may be a mechanotransduction mechanism similar to the tugging deflections seen at cell substrate adhesions (Plotnikov et al., 2012; Plotnikov and Waterman, 2013). The force required to induce deflection of cell junction has been experimentally measured to be ∼100 pN, a force that can be supplied by a few tens of myosin II molecules (Bambardekar et al., 2015). Junctional tugging was especially prominent at stable α-actinin–enriched junctional puncta (Videos 7, 8, and 9). Tugging between neighboring cells led to positive and negative junctional deflections, resulting in fluctuations in the apical cell diameter (Fig. 2 L and Videos 7, 8, 9, and 10). The frequency of tugging increased as the epithelial monolayer matured (Fig. 2 M) and was dependent on endogenous myosin II activities (Fig. 2 N). On occasions, actin bundles emanating from epithelial junction could be found to directly connect to the cellular contractile apparatus (Fig. 2 O).

Endogenous contractions acting on cell junctions can create tension at cell–cell adhesions and may be a mechanism to control α-actinin-4 targeting and junction maturation. To further study the role of mechanical force in the assembly of junctional complexes, we induced junctional tension by applying hydraulic pressures to cell–cell adhesions (Fig. S3, B–D). We modified the pressure chamber apparatus to deliver cyclic or pulsatile mechanical force to cell junction in an epithelial monolayer (Fig. S3 E). Using this pressure chamber, a cell monolayer can be subjected to cyclic, pulsatile, or sustained physiological (0–15 mmHg) or pathological (15–30 mmHg) pressures using a push-pull syringe pump.

To guide our experiments, we used previously published measurements of cell–cell adhesive forces of ∼1–4 nN/µm² (Chu et al., 2004; du Roure et al., 2005; Martinez-Rico et al., 2005; Bajpai et al., 2009; Ladoux et al., 2010; Maruthamuthu et al., 2011; Ng et al., 2012; Tseng et al., 2012) to approximate the upper limit of hydraulic pressure that an epithelial cell monolayer could withstand (Fig. S4 A). The breakage force of a cell monolayer is estimated based on the surface area of the lateral cell membrane, thus the contact area and number of cell–cell contacts (Fig. S4 B). From these defined parameters, we calculated that young monolayers that are spread thinner with less cell–cell contact area have an upper limit of <5 mmHg, whereas mature cells that are taller with smaller diameters have an upper limit of >30 mmHg. From these calculations, we predicted that the tolerance level of an epithelial cell monolayer would increase over the course of monolayer maturation process simply because of geometry change. To eliminate the variations in temporal parameters, we synchronized the cell monolayers (Fig. S4 C; see Materials and methods) and showed that cyclic pressure of 10 mmHg would disrupt the barrier function of young monolayers, whereas the same amount of pressure would enhance the barrier function of maturing monolayers (Fig. S4 D). These observations indicate that the maturity of the cell junction determines the strength of cell–cell adhesion as well as the type of response to mechanical force. Thus, an epithelial monolayer can store biophysical information temporally, in addition to biochemical changes that occur during junction maturation.

To assess the effect of mechanical force on α-actinin-4 recruitment, we empirically tested cyclic, pulsatile, sustained pressure parameters, including the level of pressure, the duration of pressure application, the frequency of cyclic pressure, and the duration of pauses between cycles of pulsatile pressures (Fig. 3 A). We found two distinct responses: (1) increased accumulation of α-actinin-4 at the junction and (2) induction of cell extrusion (Fig. 3 A, red arrowheads). Tension-induced α-actinin-4 response depended on the intensity of applied pressure, the number of pressure cycles, and maturity of the cell junction, whereas tension-induced cell extrusion depended on the waveform of the applied pressure and the duration of the pressure experiment. In addition, young and mature monolayers exhibited different sensitivity and response. The induction of cell extrusion by cyclic pressure implicated activation of endogenous cellular contractility. Indeed, inhibition of endogenous myosin II activities with blebbistatin completely blocked tension-induced cell extrusion without affecting tension-induced α-actinin-4 recruitment (Fig. 3 B). Thus, tension-induced α-actinin-4 recruitment is myosin II independent, whereas tension-induced contractility requires myosin II activity. Tension-induced recruitment of α-actinin-4 was not isoform specific because exogenously expressed venus–α-actinin-1 also relocated in a tension-sensitive manner (Fig. 3 C). Tension-induced recruitment of α-actinin-4 was not caused by increased expression of cellular α-actinin-4, which was unchanged after 60 min of pressure treatment (Fig. 3 D). In general, application of sustained pressure had a less pronounced effect on α-actinin-4 recruitment than pulsatile and cyclic pressure cycles that had a comparable force-time product (amount of pressure × duration of pressure applied to the monolayer; Fig. 3 E).

Application of low cyclic pressure of ∼1 mmHg for 15 min induced redistribution of α-actinin-4 and vinculin from the cytoplasm to the junction in maturing 2 dpc monolayers (Fig. 4 A) without altering their total cellular levels (Fig. 4 B). This amount of mechanical force (∼10–30 nN force per cell depending on the size of spread area) was within the range that can be generated by MDCK cells, which has been measured to be ∼100 nN between a cell pair (Maruthamuthu et al., 2011). The redistribution of α-actinin-4 and vinculin was accompanied by a shift in TX-100 solubility of α-actinin-4 and vinculin from mostly TX-100 extractable in 1 dpc monolayers to mostly TX-100 insoluble in pressure-treated and mature monolayers (Fig. 4 B). Consistent with our earlier observation that junction localization of vinculin requires α-actinin-4 (Fig. 2 F), knockdown of α-actinin-4 completely abolished tension-induced vinculin recruitment to the junction and redistribution to the TX-100–insoluble fraction (Fig. 4, C and D), indicating that α-actinin-4 is an upstream regulator of vinculin targeting in MDCK cell monolayers. We have previously shown that α-actinin-4 plays an important role in actin assembly at epithelial junctions (Tang and Brieher, 2012). Here, we show that pretreatment of maturing monolayers with low cyclic pressure for 5 min was sufficient to induce actin accumulation and subsequently protect the cell monolayer from gross mechanical disruption by a high dose of pressure (Fig. 4 E).

Quantitation of deconvolved optical sections showed a sixfold increase in α-actinin-4 and a twofold increase in actin accumulation without dramatically changing the levels of canonical adherens junction proteins (Fig. 5, A and B). In mature cells where α-actinin-4 is already localized, a higher pressure was required to trigger a mechanical-induced response (Fig. 5, C and D). These results show that the sensitivity of junctional complexes to mechanical stimuli depends on the maturity of the junction. Prolonged application of cyclic pressure in maturing junction resulted in reduction in junctional accumulation of β-catenin and p120 (Fig. 5, E and F). However, tension-stimulated α-actinin-4 recruitment remained intact (Fig. 5 G), suggesting that additional factors other than the known canonical junctional proteins are involved. Previously, we used chemical cross-linking to identify interacting proteins of α-actinin-4 in a membrane preparation (Tang and Brieher, 2013). Here, we show that one of the identified proteins, synaptopodin, is localized to cell junction with β-catenin (Fig. 5 H). Synaptopodin levels were unchanged after prolonged mechanical stimulation (Fig. 5, H–J) despite reduction in β-catenin, suggesting that it might play a role in tension-induced α-actinin-4 recruitment.

Synaptopodin is expressed in vertebrates and shares no homology with other proteins (Fig. 6 A). Synaptopodin has been shown to interact with actin and α-actinin-4 in brain (Kremerskothen et al., 2005) and was thought to regulate α-actinin-4 function in kidney podocytes (Asanuma et al., 2005). Synaptopodin colocalizes with CD2AP at cell–cell contacts in podocytes upon formation of cell–cell interactions (Schiwek et al., 2004). In MDCK cells, synaptopodin colocalized with β-catenin on the lateral membrane and was concentrated at the apical junction with α-actinin-4, β-catenin, and E-cadherin (Fig. 6 B). Synaptopodin was localized to apical latrunculin-resistant actin puncta (Fig. 6 C), which are sites of α-actinin-4–dependent actin assembly (Tang and Brieher, 2012). At an early stage of junction development, synaptopodin and α-actinin-4 were poorly localized despite the presence of β-catenin at the junction (Fig. 6 D). Application of cyclic pressure induced accumulation of both synaptopodin and α-actinin-4 (Fig. 6, D and E), indicating that synaptopodin targeting is also tension sensitive. Quantitation of junctional staining showed a strong correlation between α-actinin-4, E-cadherin (Fig. 6 F), and synaptopodin (Fig. 6 G) before and after cyclic pressure. Importantly, both synaptopodin and α-actinin-4 correlated with actin levels before and after cyclic pressure (Fig. 6, H and I), suggesting that they function together in actin assembly at the junction. Upon junction maturation, low cyclic pressure no longer had any effect on synaptopodin and α-actinin-4 recruitment (Fig. 7, A and B), suggesting that the mature junction had reached a homeostatic state. Concomitant to synaptopodin targeting, cyclic pressure also induced an upward shift of synaptopodin in SDS-PAGE (Fig. 7 C), indicating that synaptopodin is modified in a tension-sensitive manner. Thus, tension-induced protein modifications may be one of the mechanisms to regulate the assembly of the junctional complex.

Knockdown of synaptopodin in MDCK cells caused a dramatic decrease in the accumulation of α-actinin-4 and actin at the junction (Fig. 7, D and E) without altering the total cellular level of α-actinin-4 (Fig. 7 F). Synaptopodin knockdown resulted in relocalization of venus–α-actinin from the cell junction to the cytoplasm in mature monolayer (Fig. 7 G), indicating that the absence of α-actinin-4 staining at the junction is not caused by a loss of antibody epitopes. Knockdown of synaptopodin slowed the development of a permeability barrier (Fig. 7 H), and the cells would physically detach from each other when faced with high mechanical stress (Fig. 7 I). In contrast, knockdown of α-actinin-4 had no significant effect on the cellular levels and junctional localization of synaptopodin (Fig. 7, J and K), indicating that synaptopodin targeting does not require α-actinin-4. Both synaptopodin and α-actinin-4 were required for tension-induced barrier enhancement (Fig. 7 L). Moreover, both synaptopodin and α-actinin-4 cells were required for adhesion strengthening (Fig. 7 M). Knockdown of synaptopodin and α-actinin-4 compromised the strength of cell junction (Fig. 7 M), although the levels of E-cadherin, occludin, β-catenin, and p120-catenin were unchanged (Fig. 7 N). Thus, synaptopodin is incorporated into the maturing junctional complex before α-actinin-4 targeting and behaves as an upstream regulator of α-actinin-4 at the cell junction during junction maturation.

Exogenous application of intercellular tension failed to rescue α-actinin-4 recruitment in synaptopodin knockdown monolayers (Fig. 8, A and B). Despite the diminished tension response, the levels of junctional α-actinin-4 remained tightly correlated with junctional synaptopodin (Fig. 8 C), indicating that tension-sensing is intact in synaptopodin knockdown cells. Thus, the defect in α-actinin-4 recruitment is likely to be caused by a downstream mechanism such as mechanotransduction or mechanoresponse rather than mechanosensation by itself.

Vinculin accumulation during normal junction maturation and tension-induced targeting was absent in synaptopodin knockdown cell monolayers (Fig. 8 D). Knockdown of synaptopodin also prevented tension-induced redistribution of α-actinin-4 and vinculin to the TX-100–insoluble fraction (Fig. 8 E), consistent with the role of synaptopodin as an upstream regulator of α-actinin-4.

We have described earlier that application of intercellular tension causes cell extrusion (Fig. 3 A). We are interested in understanding the roles of synaptopodin and α-actinin-4 in tension-induced contractility. Unfortunately, the level of tension that is required to activate endogenous cellular contractility also breaks the cells apart in synaptopodin and α-actinin-4 knockdown monolayers. Therefore, we have to find another way to assess the contribution of synaptopodin and α-acitnin-4 in junction-dependent contractility. In a cell-spreading assay that reports on the balance of forces between cell–cell and cell–substrate tension (Mertz et al., 2013), MDCK cells would form clusters on soft polyacrylamide gel, whereas synaptopodin knockdown cells would spread readily into a monolayer (Fig. 8 F), consistent with a reduction in cell–cell tension in the knockdown cells. However, cell spreading is not a direct readout of junction contractility. Therefore, we investigated wound-induced contraction that requires actomyosin activities coupled to engagement of cell–cell adhesion (Martin et al., 2010; Abreu-Blanco et al., 2011). Wounding MDCK monolayers triggered a contraction response that results in a diameter three times the original wound diameter (Fig. 8, G–I), which is comparable with observations of in vivo wounding measurements (Martin et al., 2010; Abreu-Blanco et al., 2011). In contrast, wound expansion was absent in synaptopodin and α-actinin-4 knockdown monolayers (Fig. 8, G–I), indicating a defect in tissue level contractility. Our results are consistent with other studies showing that the formation of adherens junctions leads to an increase in cell–cell adhesive strength and generation of tissue level tension (Harris et al., 2012, 2014).

To understand the contribution of synaptopodin in the mechanotransduction pathway, we performed phospho-kinase array experiments before and after cyclic pressure (Fig. 9 A). 5 of 43 site-specific phosphorylation sites in the kinase array are activated by mechanical force. Of the 5 tension-induced phosphorylation sites, only Akt S473 was compromised in synaptopodin knockdown monolayers (Fig. 9 B). Another phosphorylation site on Akt, T308 (Fig. 9 A, gray boxes), was not induced by tension or affected by synaptopodin knockdown, indicating that the mechanoresponse is very specific. Application of cyclic pressure to cell monolayers resulted in the accumulation of phosphorylated Akt S473 at cell–cell adhesions (Fig. 9, C andD), suggesting that force-induced mechanotransduction is coupled to reorganization and assembly of epithelial junction.

Previous studies have shown that engagement of cell–cell adhesion activates PI3K signaling and Akt phosphorylation (Laprise et al., 2002, 2004; Brouxhon et al., 2013), which is required for the propagation and assembly of E-cadherin adhesions (Gavard et al., 2004). Our findings concur with the regulation of junctions via PI3K signaling pathways and extend the mechanism to Akt S473 phosphorylation via a synaptopodin-dependent mechanism at the cell junctions. These results show that synaptopodin is central to mechanotransduction at cell–cell adhesion and may play additional roles in tension-induced cellular processes.

To understand the function and regulation of synaptopodin, we performed a series of experiments using synaptopodin truncation mutants. We cut synaptopodin into three pieces, aa 1–600, 245–600, and 600–929. We found that synaptopodin aa 1–600 and aa 245–600 coimmunoprecipitated with endogenous α-actinin-4, β-catenin, and synaptopodin (Fig. 10 A), whereas expression of the C-terminal aa 600–929 disrupted the complex (Fig. 10 A). Interestingly, both aa 1–600 and aa 600–929 were targeted to the junction (Fig. 10, B and C), indicating that different regions of synaptopodin can interact with multiple junctional components. Previous studies have shown that substrate stiffness can regulate tension at cell–cell adhesion (Brevier et al., 2008; Ladoux et al., 2010; Liu et al., 2010; Saez et al., 2010; Maruthamuthu et al., 2011). Therefore, we tested the ability of mutant-expressing cells to respond to substrate stiffness. We found that aa 1–600 would redistribute to the cytoplasm when the cells were grown on soft silicone substrate (Fig. 10 B), suggesting that aa 1–600 contains the tension-responsive region. Consistent with these observations, application of cyclic pressure to maturing cell monolayers induced redistribution of synaptopodin aa 1–600 from the cytoplasm to cell junction (Fig. 10 C). Intriguingly, cyclic pressure wound induced cell rounding in cells expressing aa 600–929 (Fig. 10 C), suggesting that the C-terminal region of synaptopodin might link the cell–cell adhesion complex to the cellular contractile system. To identify the binding partner of synaptopodin aa 600–929, we used affinity chromatography to isolate synaptopodin aa 600–929 tagged with a streptavidin-binding domain. We pulled down one major protein and identified the protein as myosin II by mass spectroscopy analysis (Fig. 10 D). Hence, synaptopodin, through its three functional domains (Fig. 10 E), can (1) respond to mechanical force to accumulate at the cell junction, (2) form a complex with adhesion- and actin-regulating molecules, and (3) interac with myosin II. Thus, synaptopodin can physically link adhesion to the contractile machine, providing a mechanism to regulate cellular contractility by the epithelial junction. These findings provide a plausible molecular basis for the cell–cell adhesion complex to act as both a force-bearing mechanotransducing structure and a force-generating mechanoresponsive structure.

The molecular mechanisms of sensing, transducing, and adapting to mechanical signals are not fully understood, partly because of a lack of cell culture model and methods to mechanically manipulate tension at cell–cell contacts in an epithelial sheet. Here, we have fabricated a pressure chamber that can deliver hydraulic forces to cell–cell adhesions in a monolayer of cells. In this study, we use this new setup to study how mechanical force plays an essential role in junction assembly and maturation. During a maturation process over several days, epithelial cell junctions experience both normal stress from centripetal contractions and shear stress from lateral movements. These mechanical inputs can be transmitted across cell–cell adhesions to induce junctional tension. Indeed, MDCK cells can produce substantial contractile force (Maruthamuthu et al., 2011), indicating that they can use mechanical-driven mechanisms to regulate molecular events. Coupling between cellular contractility and cell–cell adhesion is well established but not well understood at the molecular level (Conti et al., 2004; Ladoux et al., 2010; Wu et al., 2014). We observed that α-actinin-4, a protein that is required for actin assembly at junctional complex (Tang and Brieher, 2012), accumulates at the adherens junction during epithelial maturation. The accumulation of α-actinin-4 coincides with the onset of myosin II–dependent contractile activities, suggesting that mechanical forces may have a role in the maturation process. Exogenous application of intercellular force induces α-actinin-4 targeting in a time- and tension-dependent manner, with concomitant increase in actin and vinculin accumulation, adhesive strength, and permeability barrier function. Thus, endogenous cellular contractility could drive junction maturation via tension-induced recruitment of α-actinin-4. Our results show that α-actinin-4 is an upstream regulator of vinculin. However, it is unclear whether α-actinin directly recruits vinculin or indirectly recruits through an α-catenin–dependent interaction. Future experiments using more sophisticated methods to study the junction as a macromolecular assembly are necessary to dissect the molecular mechanism during tension-induced incorporation of α-actinin and vinculin.

We have identified an essential factor for α-actinin-4 accumulation at the adherens junction as synaptopodin. Synaptopodin and α-actinin-4 have been shown to bind MAGI-1 (Patrie et al., 2002), an adherens junctional protein that also binds β-catenin (Dobrosotskaya and James, 2000; Nishimura et al., 2000). MAGI-1 has also been shown to be necessary for robust cell–cell adhesion (Stetak and Hajnal, 2011) and permeability barrier function through its interaction with β-catenin (Gujral et al., 2013). Therefore, our results are consistent with earlier works and suggest that synaptopodin, MAGI-1, β-catenin, and α-actinin-4 may have functional relationships at the cell junction. Moreover, we found that synaptopodin can interact with myosin II (Fig. 10 D), providing a physical link between the adhesive contacts and the cellular contractile system. Recently, synaptopodin has been shown to play a role in shear stress–induced wound healing (Mun et al., 2014). Consistent with their observations, we show that synaptopodin is mechanoresponsive and would target to the cell junction in a tension-dependent manner. In addition, we found that synaptopodin is required for vinculin recruitment, adhesion strengthening, and epithelial barrier formation. Our results demonstrate a role for synaptopodin at cell–cell adhesion, which could help explain why disruption of synaptopodin function in kidney podocytes results in barrier disruption and proteinuria (Dai et al., 2010; Yaddanapudi et al., 2011; Kimura et al., 2013).

To summarize, this study narrowly describes cell junction development in three stages: (1) synaptopodin- and α-actinin-4–negative junction that contains E-cadherin, α-catenin, β-catenin, and p120-catenin; (2) synaptopodin-positive and α-actinin-4–negative junction; and (3) synaptopodin- and α-actinin-4–positive junction that would support vinculin targeting, actin assembly, and adhesion strengthening (Fig. S5 A). We showed that synaptopodin is incorporated into the maturing junctional complex before α-actinin-4 targeting and behaves as an upstream regulator of α-actinin-4 at the cell junction during junction maturation (Fig. 6 D). Knockdown of synaptopodin prevents α-actinin-4 targeting to the junction (Fig. 7 D). In turn, α-actinin-4 is incorporated into the maturing junctional complex before vinculin targeting and behaves as an upstream regulator of vinculin at the cell junction during junction maturation (Fig. 1 B). Knockdown of α-actinin-4 prevents vinculin targeting to the junction (Fig. 1 D). Thus, we conclude that mechanical force ushers the development of cell junction through these three stages by sequentially stimulating incorporation of tension-responsive components into maturing junctional complexes, resulting in the assembly of the final junctional structure.

Our findings uncover a new function for synaptopodin and a novel molecular mechanism explaining how epithelial contractility plays an important role in adhesion maturation and the integrity of epithelial tissue (Conti et al., 2004; Shewan et al., 2005; Miyake et al., 2006; Vasquez et al., 2014). Our results reveal a mechanism that allows cellular contractility to temporally and spatially control the recruitment of junctional components that are required for adhesion strengthening. By coupling mechanical stimuli to α-actinin-4 recruitment, synaptopodin provides a tunable molecular mechanism that can dictate the outcome of cellular contractility. In the presence of synaptopodin, cellular contractions result in adhesion strengthening, whereas in the absence of synaptopodin, contractions can disengage cell–cell adhesion, resulting in dissolution of the cell junction. Indeed, myosin II–dependent contractile forces have been shown to strengthen as well as weaken cell–cell adhesion. During cell division, myosin II activities are localized at specific membrane domains, which pull on the membranes to create tension and trigger detachment of cell–cell contacts (Founounou et al., 2013; Guillot and Lecuit, 2013; Herszterg et al., 2013). During embryogenesis, myosin II activities are planar polarized to discrete domains of cell–cell interactions (Kasza et al., 2014; Shindo and Wallingford, 2014). These junctional domains are subsequently absorbed, leading to shrinking of selective myosin II–decorated junctions and expansion of nondecorated junctions. Conversely, contractile pulses and tension applied to cell–cell junction have been shown to enhance epithelial barrier function and support tissue integrity (Okano and Yoshida, 1993; Vasquez et al., 2014). Increasing tension at cell–cell adhesion induces accumulation of junctional components and stabilization of E-cadherin (Yoshida et al., 1994; Shewan et al., 2005; Liu et al., 2010), whereas ablation of myosin II function decreases cell–cell adhesion (Conti et al., 2004). However, there remains no molecular explanation to reconcile the duality of myosin II–dependent contractility. Here, we provide a plausible mechanistic explanation to how a cell junction might switch between disengagement and strengthening via synaptopodin.

Depending on whether myosin II activities or α-actinin-4 functions are used for other cellular functions, epithelial cells can choose between using a biophysical input such as mechanical force generated by myosin II–dependent contractions or a biochemical input by regulating synaptopodin levels/activities to control α-actinin-4 recruitment. Thus, a cell can be expressing high levels of synaptopodin and α-actinin-4, but in the absence of tension generated by myosin II, synaptopodin and α-actinin-4 will not incorporate into the junctional complex and strong cell–cell adhesion will not happen. Conversely, a cell can have high contractility, but in the absence of synaptopodin, cell junction will not accumulate α-actinin-4 and strong cell–cell adhesion will not happen. Our findings reveal that mechanical tension through a series of force-dependent biochemical events induces sequential accumulation of junctional proteins, including synaptopodin (Mundel et al., 1997; Patrie et al., 2002; Asanuma et al., 2005), α-actinin-4 (Knudsen et al., 1995; Nieset et al., 1997), vinculin (Hazan et al., 1997; Watabe-Uchida et al., 1998; Imamura et al., 1999), and actin, ultimately leading to an enhancement of epithelial contractility and barrier function. Thus, this study reveals a temporal and spatial bipartite system at the intercellular junction, which is capable of integrating biophysical inputs and biochemical events to control cell–cell interactions.

In this study, we show that mechanical force can induce different biochemical events at the cell junction depending on the composition, i.e., maturity, of the junctional complex. This is manifested as an age dependency of the cell monolayers in their responses to mechanical force. Local, cell-nonautonomous feedback regulation of myosin dynamics has been shown to control the amplitude and spatial propagation of pulsed constrictions through regulation of tension and geometry during epithelial morphogenesis (Saravanan et al., 2013). Therefore, we propose that junction assembly in a monolayer of epithelial cell is the product of a positive feedback loop that requires iterations of multiple rounds of biophysical and biochemical events such as protein modifications, activation of signaling cascades, and contractility until a mature junction is formed (Fig. S5 B).

We discovered that cell–cell contacts have the ability to sense different levels and frequencies of tension. Whereas cyclic and pulsatile pressures could lead to α-actinin-4 accumulation, sustained pressure cannot. Furthermore, prolonged cyclic pressure would induce cell extrusion, whereas prolonged pulsatile pressure would not. These observations implicate the presence of a timer that can distinguish cyclic and pulsatile pressure. Thus, mechanotransduction at the cell junction is not simply an integration of stress over time. The future goal is to understand how cell–cell interactions achieve tension sensing at the molecular level, how mechanical signals are amplified and propagated to trigger junction assembly, and how adhesion receptors are coupled to the regulation of junction-dependent tissue contractility. Dissecting the coupling between actin dynamics, junctional complex formation, and actomyosin activities would require molecular mechanisms of how each protein works when they are simultaneously interacting with multiple partners at the junction. This is further complicated by the difficulties in obtaining biophysical measurement, such as the magnitude and frequency of endogenous forces, which has not been characterized yet. Thus, both biochemical and biophysical inputs have spatial and temporal requirements that would play differential roles in the regulation of junction assembly.

Rabbit polyclonal antibodies to α-actinin-4 were raised in house against a synthetic N-terminal peptide, NQSYQYGPSSAGNGAGC, coupled to KLH. Mouse monoclonal antibodies to E-cadherin RR1 were made from hybridoma and a gift from W.M. Brieher (Department of Cell and Developmental Biology, University of Illinois, Urbana-Champaign). Antibodies to α-actinin-4 (catalog no. 49333, goat), β-catenin (catalog no. 7963, mouse; and 7199, rabbit), synaptopodin (catalog no. 50459, rabbit), α-catenin (catalog no. 1495, goat; and 9988, mouse), p120 (catalog no. 13957, rabbit), and CBP tag (catalog no. 33000, goat) were purchased from Santa Cruz Biotechnology, Inc. Secondary antibodies were purchased from Bio-Rad Laboratories (HRP goat anti–rabbit) and Invitrogen (FITC and Cy3 goat anti–mouse, FITC and Cy3 goat anti–rabbit, Alexa Fluor 488 donkey anti–mouse, Alexa Fluor 568 donkey anti–rabbit, and Alexa Fluor 647 donkey anti–goat). Alexa Fluor 350–phalloidin, Alexa Fluor 647–phalloidin, and N-hydroxysuccinimide–activated rhodamine were purchased from Invitrogen. Leupeptin, Pefabloc, E-64, antipain, aprotinin, bestatin, and calpain inhibitors I and II were purchased from AG Scientific, Inc. Latrunculin B, 3-aminopropyl-trimethoxysilane, hydrazine hydrate, blebbistatin, and BSA were purchased from Sigma-Aldrich. Sephadex S-200 was purchased from GE Healthcare. Protein A agarose and dithiothreitol were purchased form Gold Biotechnology, Inc. Biotin-agarose beads were obtained from Thermo Fisher Scientific. Medical-grade silicone sheeting (catalog no. SH-20002-005) was purchased from BioPlexus. Medical grade visco-gel silicone sheets (catalog no. 3081-S) were purchased from PediFix. Collagen I was purchased from BD.

The coding sequence of human synaptopodin cDNAs (OriGene catalog no. RC228616) corresponding to aa 1–929, aa 1–600, aa 245–600, and aa 600–929 were subcloned into Sal1 and Xho1 sites of the G418-selectable mammalian expression vector pNTAPB (Agilent Technologies) with CBP and SBP tags at the N terminus for expression in MDCK cells. shRNAs for canine α-actinin-4 (5′-AGGTCCTGTTCCTCTGACTCGGTATCTAT-3′) and canine synaptopodin (5′-GAGGTGAGATCCAGCACACTTCTGATTGA-3′) were synthesized and subcloned into blasticidin-selectable pGFP-B-RS and puromycin-selectable pRS vectors, respectively (OriGene). Venus–α-actinin-1 cDNA was provided by F. Meng and F. Sachs (Physiology and Biophysics Department, SUNY at Buffalo, Buffalo, NY).

Purification of junction-enriched membranes was prepared as described previously (Tang and Brieher, 2012). In brief, frozen rat livers (Pel-Freez) were thawed in 2 vol of 10 mM Hepes, pH 8.5, and 10 mM DTT. Protease inhibitors (10 µg/ml Leupeptin, 1 mg/ml Pefabloc, 10 µg/ml E-64, 2 µg/ml antipain, 2 µg/ml aprotinin, 50 µg/ml bestatin, 20 µg/ml calpain inhibitor I, and 10 µg/ml calpain inhibitor II) were added and the livers were briefly blended in a Waring blender (5 × 15 s). The liver slush was filtered through four layers of cheesecloth to obtain the total liver homogenate. Total liver homogenate was centrifuged at 1,000 g for 30 min. The pellet was homogenized in 10 mM Hepes, pH 8.5, and 10 mM DTT in a Dounce homogenizer and centrifuged at 100 g for 30 min. The supernatant was collected and centrifuged at 1,000 g for 30 min. The membrane pellet contains the majority of actin assembly activity and is frozen at −80°C before further purification immediately before use. The day of the experiment, membranes were thawed on ice, diluted 1:1 with 10 mM Hepes, pH 8.5, supplemented with 10 mM DTT, and homogenized through a 25G needle. The homogenates were spun through a 20% sucrose pad for 10 min at 16,000 g. The supernatant was discarded and the pellets were resuspended with 10 mM Hepes, pH 8.5, supplemented with 10 mM DTT. The homogenate was spun through a 20% sucrose pad for 15 min at 1,000 g. The pellet was discarded and the supernatant was spun through a 20% sucrose pad for 15 min at 16,000 g. The tiny membrane pellet contained junction-enriched plasma membrane vesicles.

Actin assembly reactions were performed in actin assembly buffer (50 mM KCl, 2 mM EGTA, 2 mM MgCl₂, and 100 mM Hepes, pH 7.8) supplemented with 2 mM buffered ATP, pH 8.0. In brief, a 20-µl reaction consisting of ∼15 µg of total proteins from the membrane fraction and 0.5 µM fluorescently labeled monomeric actin was allowed to carry out at room temperature for 30 min before incubation with primary antibodies in the presence of 0.1% TX-100. The membranes were spun through a 20% sucrose cushion and resuspended in 0.1% TX-100 in assay buffer. The membranes were incubated with secondary antibodies for 2 h, spun through a 20% sucrose cushion, and resuspended in 0.1% TX-100 in assay buffer. Stained membranes were imaged using an Axio Imager with the Colibri illumination system (both Carl Zeiss) using a 63× objective (NA 1.4) attached to a 1K × 1K ORCA-ER CCD camera (Hamamatsu Photonics).

For visualizing actin assembly, membrane reactions were performed as described in the previous section before processing for EM. For immunogold labeling of junction-enriched membrane, membranes were treated with 0.2% TX-100 in actin assembly buffer on ice for 1 h and spun through a 20% sucrose cushion before incubation with anti–α-actinin-4 antibodies for 2 h on ice. Unbound antibodies were removed by spinning membranes through a 20% sucrose cushion before incubation with 6-nm colloidal gold anti–rabbit antibodies for 2 h. Unbound gold antibodies were removed by spinning membranes through a 20% sucrose cushion before processing for EM. Membrane reactions were put onto glow-discharged carbon-coated grids for 10 min, washed three times with assembly buffer, and stained with 2% uranyl acetate. Images were collected with a microscope (2100EX; JEOL) at 120 kV using a 2K × 2K charge-coupled device camera (UltraScan; Gatan, Inc.). For figure generation, images were cropped, contrasted, and scaled using Photoshop software (Adobe) before importing into Illustrator (Adobe).

Human phosphor-Kinase Array (catalog no. ARY003B; R&D Systems) was performed according to the manufacturer’s instructions using MDCK cells that had been grown on 24-mm Transwells for days with or without cyclic pressure treatment.

MDCK cells were maintained in MEM/Earle’s balanced salt solution supplemented with 25 mM Hepes and 10% FBS. For transfection, cells were incubated in Opti-MEM (Invitrogen) with a 1:1 mixture of DNA/polyethylenimine and selected for 10 d using G418, puromycin, or blasticidin. For latrunculin treatment, cells grown on Transwells for 7 d were treated with 10 µM latrunculin in normal growth media for 2 h and then processed for immunofluorescence.

For soft substrate studies, glass coverslips or medical-grade silicone sheeting (40 Durometer, thickness 0.005 in) were soaked in 100% ethanol for 30 min and sterilized under UV for 60 min. Sterilized coverslips or silicone sheets were coated with 20 µg/ml collagen I in phosphate-buffered saline for 60 min and used immediately for plating of cells. Cells grown on collagen-coated glass coverslips (upside down) or collagen-coated silicone substrates (right-side up) were processed for immunofluorescence staining.

Glass coverslips were soaked in 100% ethanol for 30 min and sterilized under UV for 60 min. Sterilized coverslips were coated with 20 µg/ml collagen I in phosphate-buffered saline for 60 min and used immediately for plating of cells. Cells grown on collagen-coated glass coverslips (upside down) were mounted to the glass slide chamber. Cells were imaged using in-house fabricated glass slide chambers. The chamber was assembled by attaching a 1.5 × 1.5–cm medical-grade silicone gel (PediFix) onto a sterilized glass slide. The center of the silicone gel was removed and used as a sink for culture media during imaging. For blebbistatin perfusion and washout studies, 25 µM blebbistatin (from 5 mM stock solubilized in DMSO) was added to regular growth media and perfused into the sink of the glass slide chamber. For washout, regular media was perfused into the sink from an inlet while blebbistatin-containing media was removed from an outlet of the sink.

Cells grown on Transwell-Clear (Corning) were processed for EM as described previously (Tang, 2006). In brief, MDCK and T84 epithelial cells grown on Transwells were chilled at 4°C for 6 h before fixation with 3.75% glutaraldehyde, 150 mM NaCl, and 20 mM Hepes, pH 7.5, at 4°C for 18 h. The fixation reaction was quenched with 50 mM glycine and 150 mM Hepes, pH 7.5, on ice for 1 h. Cells/Transwells were rinsed in ice-cold distilled water three times, secondarily fixed with 1% osmium tetroxide/1.5% potassium ferrocyanide for 2 h on ice, rinsed four times in ice-cold distilled water, en bloc stained with freshly prepared and filtered 2% uranyl acetate in distilled water on ice for 2 h, and rinsed four times in ice-cold distilled water. Cells/Transwells were dehydrated with sequential 5-min incubations in 50%, 75%, 95%, 100%, 100%, and 100% ethanol at room temperature. epon-Araldite (EMbed 812) was added to Transwells and allowed to polymerize at 60°C for 48 h. Ultrathin sections were cut using an Ultracut-S microtome (Reichert), layered onto carbon-coated copper grids, and stained with freshly made/filtered 2% lead citrate. Grids were examined with Tecnai G2 Spirit BioTwin (FEI) equipped with an AMT 2K CCD camera. Digital images acquired were imported into Photoshop for figure preparation.

Cells grown on Transwell-Clear were used in pressure experiments. Pressure chambers and adaptors were designed by V.W. Tang and fabricated from medical-grade stainless steel by the mechanical engineering machine shop at the University of Illinois, Urbana-Champaign. Transwell filter cups were mounted onto a pressure chamber, held in place to a lid by a screw and a top. Hydraulic pressure was applied to the basal compartment via a syringe hooked up to a syringe pump into the basal chamber that was filled with cell culture media and sealed with an O-ring between the pressure chamber and the Transwell cup. Pressure was monitored through an outlet from the basal pressure chamber to a pressure gauge. An apical adaptor is mounted onto the apical chamber of the Transwell cup that is filled with cell culture media and sealed with an O-ring. An outlet from the apical adaptor is either exposed to ambient pressure or connected via a bifurcation to a pressure gauge and a syringe hooked up to a syringe pump. The pressure chamber was preheated to 37°C and kept at this temperature on a warm plate for the duration of the experiment. Cyclic, pulsatile, or sustained pressures were applied to cell monolayers by programmable infuse/withdrawal syringe pumps (Lagato SPLG270). Cyclic pressure was applied at 30 cycles per min or otherwise specified. Pulsatile pressure was applied at six cycles per minute or otherwise specified.

Total cell lysates were obtained by solubilization of cells directly in SDS-PAGE sample buffer containing 25 mM dithiothreitol, 2% SDS, 50 mM Tris-Cl, 5% glycerol, pH 8.8, and protease inhibitors (see section Purification and immunostaining of junction-enriched membrane). For immunoprecipitation, cells were grown to 100% confluency and maintained at confluency for 5 d. Then the cells were split 1:2 onto new dishes the night before use. The next day, the cells were rinsed and swelled with 10 mM Hepes, pH 8.5, for 3 h at 4°C. Cells were extracted with immunoprecipitation buffer (100 mM NaCl, 10 mM Hepes, pH 7.8, 2 mM EDTA, 0.01% TX-100, and 0.02% azide) containing protease inhibitors and homogenized with a 22G needle. Homogenate was centrifuged for 5 min at 5,000 g. The supernatant was collected and incubated with anti–α-actinin-4 antibodies overnight in the cold room. The next day, the supernatant was centrifuged for 10 min at 5,000 g. The supernatant was added to protein A–agarose beads and incubated overnight with agitation. The next day, the immunoprecipitation mixture was added to an Econo column (Bio-Rad Laboratories), and the beads were washed five times with immunoprecipitation buffer and three times with 10 mM Hepes, pH 8.5. Bound fraction was eluted with boiling 2× sample buffer (4% SDS, 100 mM Tris-Cl, and 10% glycerol, pH 8.8) and stored at −20°C until use. A final concentration of 100 mM dithiothreitol was added to the eluate, and the samples were boiled for 10 min before SDS-PAGE.

For streptavidin-binding protein affinity purification, the cells were split 1:2 onto new dishes the night before use. The next day, the cells were rinsed and swelled with 10 mM Hepes, pH 8.5, for 3 h at 4°C. Cells were extracted with immunoprecipitation buffer (100 mM NaCl, 10 mM Hepes, pH 7.8, 2 mM EDTA, 0.01% TX-100, and 0.02% azide) containing protease inhibitors and homogenized with a 22G needle on ice. Homogenate was centrifuged for 5 min at 5,000 g. The supernatant was collected and incubated with biotin-agarose beads and incubated overnight with agitation at 4°C. The next day, the bead–cell lysate mixture was added to an Econo column, and the beads were washed three times with immunoprecipitation buffer and two times with 10 mM Hepes, pH 8.5. Bound fraction was eluted with boiling 2× sample buffer (4% SDS, 100 mM Tris-Cl, and 10% glycerol, pH 8.8) and stored at −20°C until used. A final concentration of 100 mM dithiothreitol was added to the eluate, and the samples were boiled for 10 min before SDS-PAGE.

For cell TX-100 solubility assays, cells were grown on 24-mm Transwells for 1–8 d. Transwells were used in the pressure experiment before TX-100 extraction. In brief, cells were rinsed three times with ice-cold 10 mM Hepes, pH 8.5, and incubated on ice in 0.01% TX-100, 100 mM NaCl, 10 mM Hepes, pH 7.8, 2 mM EDTA, and 0.02% azide on Transwells for 2 h. Cells were then rinsed twice with ice-cold 10 mM Hepes, pH 8.5, while still attached on Transwells and solubilized immediately in boiling SDS-PAGE sample buffer (2% SDS, 50 mM Tris-Cl, and 5% glycerol, pH 8.8). Samples were stored at −20°C until use. A final concentration of 100 mM dithiothreitol was added to the eluate, and the samples were boiled for 10 min before SDS-PAGE.

Cells were rinsed twice in 150 mM NaCl, 2 mM CaCl₂, 2 mM MgCl₂, and 20 mM Hepes, pH 7.8, and fixed in 1% formaldehyde, 150 mM NaCl, 2 mM CaCl₂, 2 mM MgCl₂, and 20 mM Hepes, pH 7.8, at 4°C for 2 h. The reaction was quenched with 50 mM Tris in staining buffer (0.1% Triton X-100, 100 mM NaCl, and 20 mM Hepes, pH 7.8) for 1 h. After rinsing in staining buffer, the cells were incubated with primary antibodies in staining buffer overnight. After rinsing in staining buffer three times, the cells were incubated in secondary antibodies for 90 min. The cells were rinsed again three times and post-stain fixed with 1% formaldehyde in staining buffer. Finally, the cells were incubated with fluorescently labeled phalloidin or Hoechst for 60 min. Transwell filters were excised and mounted on glass slides using ProLong Gold antifade (Invitrogen).

Live cell and fixed cell fluorescence wide-field images were collected with an Axio Imager using AxioVision Release 4.7 with the Colibri illumination system (Carl Zeiss) using an oil 60× objective (NA 1.40) attached to a 1K × 1K charge-coupled device camera (ORCA-ER; Hamamatsu Photonics). For live cells, cells were imaged in normal growth media, and the images were acquired every 15 s for 120 min at 30°C. Individual cell diameter and junction for each image were measured using ImageJ (National Institutes of Health). Individual images of cropped cells were imported into QuickTime to generate video files. For deconvolution microscopy, optical z slices in 200-nm steps were collected using a 60× objective (NA 1.42) with a 1.6× auxiliary magnification with an inverted microscope (1X71; Olympus) attached to a 1K × 1K charge-coupled device camera (CoolSNAP HQ; Applied Precision). For deconvolution microscopy, fixed cell fluorescence images were obtained at room temperature using SoftWoRx DMS software (Applied Precision) and deconvolved by Enhanced Ratio Deconvolution using 10 iteration cycles (Applied Precision). Z stack projections were generated from deconvolved slices using the maximum intensity criteria. Composite images were generated using ImageJ software. For figure generation, images were cropped, contrasted, and scaled using Photoshop software (Adobe) before importing into Illustrator (Adobe).

Quantitation of immunofluorescence intensity was performed in ImageJ using unprocessed original single optical z slice images taken at the level of adherens junction. A defined junctional area was used to compare the fluorescence intensity of actin (phalloidin), E-cadherin (immunofluorescence using monoclonal RR1 antibodies), α-actinin-4 (immunofluorescence using antibodies against N-terminal peptide), α-catenin (immunofluorescence using antibodies against C-terminal peptide), β-catenin (immunofluorescence using monoclonal antibodies against aa 680–781), p120 (immunofluorescence using antibodies against aa 41–130), and synaptopodin (immunofluorescence using antibodies against aa 781–920). The measured intensities were subtracted from background (cytoplasm) before being used for calculating the intensity ratios. For figure generation, images were cropped, contrasted, and scaled using Photoshop software before importing into Illustrator. Calculation of correlation coefficient R and paired Student’s t tests were performed using KaleidaGraph software (Synergy Software). All curves were fitted using original data points into an exponential function by KaleidaGraph software.

BSA was labeled on lysine residues using N-hydroxysuccinimide–activated rhodamine, purified by gel filtration using Sephadex S-200, and kept frozen until use. Thawed tracers were purified away from residual-free dye using Sephadex 25G spin columns immediately before use. Permeability assays were performed as described in detail previously (Tang and Goodenough, 2003). In brief, 12-mm Transwells were rinsed with 50 ml of prewarmed flux buffer (145 mM NaCl, 2 mM CaCl₂, 2 mM MgCl₂, and 10 mM Hepes, pH 7.4). The apical well was filled with 0.5 ml of flux buffer, and the Transwells were placed in a 12-well plate with 1 ml of flux media containing 3 µM labeled extracellular tracers. Flux was performed at 37°C for 2 h. At the end of the flux experiment, the apical solutions were collected, and fluorescence was measured using a fluorimeter. Permeability is the amount of tracer in the apical bathing solution (µmol) per unit driving force (basal molar concentration of tracer), per unit time (hours), per area of the monolayer (square centimeters).

Cells grown on collagen-coated glass coverslips were allowed to mature for 5 d after confluency. A puncture wound was introduced in the middle of the monolayer using a beveled 25G needle (machine shop, School of Molecular and Cell Biology, University of Illinois, Urbana-Champaign) attached to a vacuum line. The needle was lowered onto the monolayer, and cell debris at the punctured area was removed simultaneously during wounding through the vacuum line. Cells were returned back to the culture incubator for 30 min until fixed. Wounds were captured using the PentaView LCD digital microscope (Celestron) with an attached digital camera. Wound diameters were measured using ImageJ. For figure generation, images were cropped using Photoshop software before importing into Illustrator.

Polyacrylamide substrates were prepared using a modified protocol established by X. Tang and S. Taher in the Mechanical Engineering Department, University of Illinois, Urbana-Champaign, based on published procedures (Damljanović et al., 2005). In brief, glass coverslips were activated with 95% 3-aminopropyl-trimethoxysliane solution and washed with water, followed by treatment with 0.5% glutaraldehyde. Polyacrylamide gel mixtures were allowed to polymerize on the activated coverslips sandwiched with a nonderivatized coverslip. Polyacrylamide gel surface was activated with hydrazine and coated with collagen I before plating of cells. Polyacrylamide gel of different stiffness was prepared according to calibration using atomic force microscopy by X. Tang and S. Taher. Cells were trypsinized from confluent monolayers and plated the night before being fixed for immunofluorescence staining.

Figs. S1 and S2 show the epithelial junction is a large membrane-associated structure ∼150 nm in size that can support α-actinin-4–dependent actin assembly Fig. S3 shows the design of the pressure chamber to apply hydraulic pressure to cell–cell adhesion in a cell monolayer. Fig. S4 shows the strength of epithelial junction is a function of cell geometry and adhesive bonds. Fig. S5 shows that the junctional complex links adhesion molecules to cytoskeletal and contractile elements. Videos 1–3 show MDCK cells expressing venus–α-actinin at 2 dpc, showing mobility of junctions. Videos 4–6 show MDCK cells expressing venus–α-actinin at 2 dpc, showing contractions of cells. Videos 7–9 show MDCK cells expressing venus–α-actinin at 2 dpc, showing tugging of junctions. Video 10 shows MDCK cells expressing venus–α-actinin at 6 dpc, showing lack of mobility of junctions and very fast contraction pulses.

We apologize for not being able to credit all publications that contributed to the foundation of this work. We thank Xin Tang and Saif Taher for the help with polyacrylamide gel preparation. We thank Rahul Kantharia (Bioengineering Department, University of Illinois, Urbana-Champaign) for making the polyacrylamide gel substrates and culturing synaptopodin knockdown cells for cell-spreading assays. We thank Robert Dam (School of Molecular and Cell Biology, University of Illinois, Urbana-Champaign) and Cameron Shahnaz (Chemistry Department, University of Illinois, Urbana-Champaign) for making synaptopodin cDNA constructs. We thank the team at the Mechanical Engineering Shop, University of Illinois, Urbana-Champaign, for fabrication of all parts for the pressure chambers. EM was carried out in part in the University of Illinois Frederick Seitz Materials Research Laboratory central facilities.

Funding was provided by National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health (R01 DK098398) to V.W. Tang.

The authors declare no competing financial interests.