Tight junctions are a component of the epithelial junctional complex, and they form the paracellular diffusion barrier that enables epithelial cells to create cellular sheets that separate compartments with different compositions. The assembly and function of tight junctions are intimately linked to the actomyosin cytoskeleton and, hence, are under the control of signalling mechanisms that regulate cytoskeletal dynamics. Tight junctions not only receive signals that guide their assembly and function, but transmit information to the cell interior to regulate cell proliferation, migration and survival. As a crucial component of the epithelial barrier, they are often targeted by pathogenic viruses and bacteria, aiding infection and the development of disease. In this Commentary, we review recent progress in the understanding of the molecular signalling mechanisms that drive junction assembly and function, and the signalling processes by which tight junctions regulate cell behaviour and survival. We also discuss the way in which junctional components are exploited by pathogenic viruses and bacteria, and how this might affect junctional signalling mechanisms.
Epithelia form functional barriers that separate our organs and tissues from the outside world. This requires the formation of intercellular junctions that allow cells to adhere tightly to each other and control the permeability of the paracellular pathway. In vertebrates, tight junctions regulate paracellular diffusion by forming a semipermeable barrier (Anderson and Van Itallie, 2009; Cereijido et al., 2008). Morphologically, tight junctions form a discrete border between the apical and basolateral cell surface domains and help to maintain cell surface polarity by forming a ‘fence’ that prevents lipid diffusion in the exoplasmic leaflet of the plasma membrane. Together with the more basally situated adherens junctions, they form the apical junctional complex (AJC) (Capaldo et al., 2014; Furuse and Tsukita, 2006; Shen et al., 2011). In epithelia, mature adherens junctions and tight junctions are morphologically distinct, whereas in other cell types, such as endothelial cells, the two junctions can be intercalated (Balda and Matter, 2008; Dejana and Orsenigo, 2013; Hirase and Node, 2012). Hence, the relative position of tight junctions and adherens junctions can vary, which might reflect their common origin from a primordial adhesive complex during the formation of cell–cell adhesions. Moreover, some tight junction components (e.g. ZO-1, also known as TJP1) can interact with proteins that are associated with adherens junctions and can be part of the latter complex in tissues that lack tight junctions (Balda and Matter, 2008).
Tight junctions are composed of transmembrane proteins that mediate cell–cell adhesion and cytoplasmic ‘plaque’ proteins that link the junctional membrane to the cytoskeleton (Fig. 1) (Balda and Matter, 2008; Shen et al., 2011). Compared with other intercellular junctions, it took a long time until the first tight junction transmembrane protein, occludin, was discovered (Furuse et al., 1993). Today, tight junctions are the epithelial junction with the greatest number of known transmembrane components, which include the tetraspan proteins of the claudin family, as well as tricellulin (also known as MARVELD2) and MarvelD3. Occludin and the latter two proteins contain a conserved four-transmembrane MARVEL domain that has been proposed to play a role in membrane apposition and/or microdomain organisation (Ikenouchi et al., 2005; Raleigh et al., 2010; Sánchez-Pulido et al., 2002; Steed et al., 2009). Tight junctions also contain BVES (blood vessel epicardial substance), which has three transmembrane domains, and many adhesion proteins of the immunoglobulin (Ig) superfamily, such as the junctional adhesion molecules (JAMs), coxsackievirus and adenovirus receptor (CAR, also known as CXADR) and angulins (Ebnet, 2008; Garrido-Urbani et al., 2014; Higashi et al., 2013; Masuda et al., 2011; Osler et al., 2005). Finally, the apical polarity determinant Crb3 has been shown to associate with tight junctions (Assémat et al., 2008; Lemmers et al., 2004; Pieczynski and Margolis, 2011). These transmembrane proteins are connected to a junctional plaque, a complex protein network formed by adaptor and signalling proteins that links the junction to the actin and microtubule cytoskeleton (Balda and Matter, 2008; Rodgers and Fanning, 2011; Yano et al., 2013). The plaque includes many prominent adaptor proteins, such as the zonula occludens proteins (i.e. ZO-1, ZO-2 and ZO-3), cingulin and JACOP (also known as paracingulin or CGNL1), as well as signalling proteins, including protein kinases, phosphatases, GTP-binding proteins and transcriptional and post-transcriptional regulators (Balda and Matter, 2009; Gonzalez-Mariscal et al., 2012; Guillemot et al., 2008b; Rodgers and Fanning, 2011; Samarin and Nusrat, 2009).
This junctional protein network receives signals from the cell interior to regulate junction assembly and function, and transmits signals to the cell interior that guide cell proliferation, migration, survival and differentiation. In this Commentary, we focus on tight-junction-associated signalling mechanisms that guide junction assembly and epithelial morphogenesis, and we discuss recent findings on how junctions signal to the cell interior to guide cell behaviour and survival. We also consider tight junctions as central components of pathogenic mechanisms and discuss how junctional signalling mechanisms might contribute to the initiation and progression of disease in response to viruses and bacteria that exploit tight junction proteins to infect cells or alter their behaviour.
Regulation of Rho GTPase signalling, actomyosin dynamics and junction formation
Rho GTPases are central components of signalling pathways that guide junction assembly and polarisation, as well as being involved in the mechanisms by which junctions signal to the cell interior. Rho GTPases function as molecular switches, cycling between an active GTP-bound state, which allows association with an effector protein (e.g. during Rho-GTPase-mediated activation of the kinases that activate actomyosin activity), and an inactive GDP-bound state (Hall, 2012; Heasman and Ridley, 2008). Guanine nucleotide exchange factors (GEFs) promote the exchange of GDP for GTP, and GTPase-activating proteins (GAPs) stimulate GTP hydrolysis. Early experiments with a non-hydrolysable analogue of GTP established a link between GTP-binding proteins and the regulation of junction assembly (Balda et al., 1991). Since then, studies from different laboratories have linked different Rho GTPase family members, including RhoA, Rac and Cdc42 to the regulation of junctional signalling mechanisms (McCormack et al., 2013; Samarin and Nusrat, 2009; Terry et al., 2010). Because a specific Rho GTPase can function in multiple, sometimes opposing, signalling mechanisms, and GEFs and GAPs largely outnumber Rho GTPases in mammals, it is thought that GEFs and GAPs are important for the spatial and temporal control of Rho GTPase signalling and that they determine process specificity by forming complexes with specific upstream and downstream components of Rho signalling pathways. Most GEFs and GAPs that are associated with tight junctions regulate RhoA or Cdc42. Rac appears to be primarily involved in adherens junction formation, thereby affecting tight junctions only indirectly, and active Rac is excluded from tight junctions. For example, the Rac GEF TIAM1 is recruited to forming junctions by multiple regulators, such as JACOP/paracingulin and β2-syntrophin (Guillemot et al., 2008a; Mack et al., 2012; Mertens et al., 2005), and mechanisms have evolved that prevent Rac activity at tight junctions, such as the binding of Par3 (also known as PARD3) to TIAM1 and the recruitment of MgcRacGAP (also known as RACGAP1) (Chen and Macara, 2005; Citi et al., 2012; Guillemot et al., 2014; Mack et al., 2012; McCormack et al., 2013). Therefore, we will focus here on the regulation of RhoA and Cdc42.
RhoA and signalling from tight junctions
Early evidence pointed to an involvement of RhoA in epithelial junction formation in different model systems (Braga et al., 1999; Braga et al., 1997; Jou et al., 1998; Nusrat et al., 1995; Takaishi et al., 1997). Both the expression of constitutively active Rho GTPases and their inhibition affect junctional integrity, further supporting the importance of GEFs and GAPs for the spatial and temporal control of Rho signalling. Cell junctions are not only regulated by RhoA, they also guide RhoA signalling, as the formation of cell junctions leads to a decrease in overall RhoA activity (Matter and Balda, 2003; Noren et al., 2001). The first RhoA regulators linked to the AJC have both been identified as components of the molecular mechanisms underlying this phenomenon. p190RhoGAP (also known as ARHGAP35) is recruited to adherens junctions in response to cadherin engagement (Noren et al., 2003; Wildenberg et al., 2006), and GEF-H1 (also known as ARHGEF2) is recruited to tight junctions by binding to cingulin and/or JACOP/paracingulin, two homologous junctional adaptors (Fig. 2) (Aijaz et al., 2005; Benais-Pont et al., 2003; Guillemot et al., 2008a). Binding to cingulin results in the inhibition of GEF-H1 and is promoted by tight junction formation that is induced by the junctional membrane protein BVES; hence, junctional GEF-H1 is thought to be inactive (Aijaz et al., 2005; Russ et al., 2011). If it is not sequestered at tight junctions, GEF-H1 promotes various RhoA-driven processes, including cell spreading and migration, cell cycle progression and gene expression (Birukova et al., 2006; Kakiashvili et al., 2009; Krendel et al., 2002; Nie et al., 2009; Terry et al., 2011; Tsapara et al., 2010). Nevertheless, GEF-H1 regulates paracellular permeability in both epithelial and endothelial cells, and is required for junction dissociation in response to Ca2+ depletion, a process that is influenced by its effect on cell shape and dynamics (Benais-Pont et al., 2003; Birukova et al., 2006; Samarin et al., 2007). GEF-H1 also promotes cell cycle progression by regulating gene expression (Aijaz et al., 2005; Nie et al., 2009). This involves the activation of the ZO-1-associated nucleic-acid-binding protein (ZONAB, the canine homologue of human YBX3, also known as DBPA), a regulator of transcription of cell cycle genes (e.g. cyclin D1, PCNA) (Balda and Matter, 2000; Nie et al., 2009; Sourisseau et al., 2006). This signalling mechanism has also been linked to stress-activated responses that, through the activation of Ras and RalA, promote GEF-H1 activation and cell survival by promoting the function of ZONAB as a regulator of mRNA stability and translation (Frankel et al., 2005; Nie et al., 2012). Hence, GEF-H1 functions in pathways that transmit information from the junction to the cell interior to guide actin reorganisation, gene expression and cell survival.
RhoA signalling and tight junction formation
Although overall RhoA activity is downregulated in many cell types upon induction of cell–cell adhesion, localised RhoA activation at cell–cell contacts is required for junction formation (Yamada and Nelson, 2007). p114RhoGEF (also known as ARHGEF18) is recruited to forming junctions by cingulin and drives junctional RhoA signalling during junction formation (Fig. 2) (Terry et al., 2011). This GEF also interacts with PALS1-associated tight junction protein (PATJ, also known as INADL), a component of the Crumbs polarity complex. The interaction between p114RhoGEF and PATJ is thought to support the recruitment of the former to cell junctions (Nakajima and Tanoue, 2011). Once activated by the induction of junction formation, p114RhoGEF forms a complex with myosin IIA and RockII, thereby forming a module that links p114RhoGEF-driven RhoA activation to junctional actomyosin activity. The role of p114RhoGEF in epithelial differentiation might be pathologically relevant, as it promotes junction formation downstream of the tumour suppressor liver kinase B1 (LKB1, also known as STK11) in lung epithelia (Xu et al., 2013). Unlike GEF-H1, p114RhoGEF thus drives RhoA signalling at junctions, and its downregulation promotes non-junctional RhoA activation and the formation of stress fibres, illustrating how inhibition of RhoA signalling at one subcellular site can promote RhoA signalling elsewhere in the cell.
The complexity of the spatial and temporal regulation of Rho GTPases is further highlighted by recent studies that identified a second tight-junction-associated RhoA GEF. ARHGEF11 was found to localise at primordial adherens junctions and then at tight junctions as epithelial polarity is established, owing to its binding to ZO-1 (Itoh et al., 2012). Similar to p114RhoGEF, ARHGEF11 mediates RhoA-activated actomyosin activation at cell–cell junctions. Whether and how ARHGEF11 and p114RhoGEF cooperate during junction formation to fine-tune RhoA activation or whether they are involved in different mechanisms active in different tissues or under different conditions remains to be determined.
GAPs are required to terminate Rho GTPase signalling and, hence, to control the amplitude, length and spatial confinement of the induced signal. Two related RhoA GAPs, myosin-IXA and myosin-IXB, regulate the formation of functional tight junctions (Abouhamed et al., 2009; Chandhoke and Mooseker, 2012; Omelchenko and Hall, 2012). Both proteins contain an N-terminal actin-dependent motor domain and a C-terminal GAP domain, and their depletion leads to increased levels of active RhoA, indicating that the motor domain ensures proper targeting of the GAP domain (Abouhamed et al., 2009; Chandhoke and Mooseker, 2012; Omelchenko and Hall, 2012). The actual localisation of the proteins is not identical – myosin-IXA localises to cell junctions, with different studies reporting either localisation to tight junctions or a more general association with the AJC, whereas myosin-IXB is found along the entire lateral membrane (Abouhamed et al., 2009; Chandhoke and Mooseker, 2012; Omelchenko and Hall, 2012). Both GAPs are required for collective epithelial cell migration (i.e. the movement of epithelial sheets) (Chandhoke and Mooseker, 2012; Omelchenko and Hall, 2012). RhoA activation by tight-junction-associated p114RhoGEF is also required for collective cell migration and stimulates junctional myosin upon the induction of migration (Terry et al., 2012). Hence, collective cell migration requires a complete junctional RhoA GTPase cycle that controls actomyosin dynamics, possibly enabling junctional remodelling during migration.
Activators of tight-junction-associated RhoA signalling thus associate with specific junctional proteins that regulate junctional recruitment and activation, and form complexes with proteins that mediate downstream events, such as myosin contractility and gene expression. Depending on the GEF, junctional recruitment can lead to either GEF inactivation or activation. The observations that different RhoA GEFs promote junction formation suggests that fine-tuning of RhoA signalling requires multiple factors and might be further influenced by GEFs and GAPs that are associated with adherens junctions (McCormack et al., 2013).
Cdc42 – a regulator of junction assembly and polarisation
In vivo and in vitro evidence indicates that Cdc42 is involved in vertebrate epithelial junction assembly (Bruewer et al., 2004; Du et al., 2009; Kroschewski et al., 1999; Rojas et al., 2001; Wu et al., 2007). Junction formation involves different Cdc42 effectors, of which the evolutionarily conserved Par3–Par6–aPKC polarity complex is the best known (Armenti and Nance, 2012; Yamanaka and Ohno, 2008; Assémat et al., 2008; Pieczynski and Margolis, 2011). The binding of active Cdc42 to Par6 (also known as PARD6A/B) leads to stimulation of the activity of atypical protein kinase C (aPKC). The complex is recruited to the forming tight junctions, and studies using RNA interference (RNAi)-mediated knockdown and expression of dominant-negative mutants indicate that it is required for tight junction formation (Suzuki et al., 2004; Suzuki et al., 2002; Suzuki et al., 2001; Wallace et al., 2010; Yamanaka et al., 2001). Another Cdc42 effector is the protein kinase Pak4, which promotes junctional maturation in a Par6-dependent manner, indicating that there is cooperation between different Cdc42 effectors (Wallace et al., 2010).
Regulation of Cdc42 during junction formation requires multiple regulators that act at different steps (Fig. 3). Two of these regulators are the structurally related GAPs SH3BP1 and RICH1 (also known as ARHGAP17), which appear to function sequentially. In many epithelia, junction assembly is initiated by Cdc42-driven filopodia- mediated induction of cell–cell contacts prior to the formation of junctions (Vasioukhin et al., 2000). The GAP SH3BP1 regulates the transition from filopodia to maturing junction, indicating that it is required to constrain Cdc42 signalling spatially and temporally (Elbediwy et al., 2012). SH3BP1 forms a complex with the junctional adaptor JACOP/paracingulin and the scaffold protein CD2AP, which is required to control Cdc42 activity. The complex also contains the capping protein CapZ, which also suppresses filopodia, indicating that the complex attenuates actin-driven morphogenetic processes by inhibiting Cdc42 signalling, as well as by regulating actin polymerisation directly. Junctional maturation and polarisation then require a second Cdc42 GAP, RICH1, which associates directly with tight junctions by binding to angiomotin (AMOT) and apical polarity determinants (Wells et al., 2006). Hence, SH3BP1 appears to function at an earlier step, whereas RICH1 promotes final differentiation and polarisation.
Similarly, different Cdc42 GEFs have been shown to regulate different aspects of junction formation. The role of one such GEF, Ect2, is controversial. Early studies suggested that it regulates Cdc42 and associates with the Par3–Par6–aPKC complex in Madin-Darby canine kidney (MDCK) cells (Liu et al., 2004; Liu et al., 2006). However, a more recent study in MCF7 cells, a breast cancer cell line, concluded that Ect2 associates with adherens junctions and that it specifically regulates Rho signalling at adherens junctions (Ratheesh et al., 2012). The underlying reason for this discrepancy is not clear but might reflect differences between the epithelial cell lines that were used.
A second tight-junction-associated Cdc42 GEF, Tuba, binds to ZO-1 and regulates junction configuration, but not formation (Otani et al., 2006). Tuba represents another example of a GEF that forms a complex with a GTPase effector – it binds to N-WASP (also known as WASL), which promotes actin polymerisation. Tuba and N-WASP also regulate epithelial morphogenesis and spindle orientation in three-dimensional cultures; however, at least in a renal cell line, this does not appear to rely on a junctional role of Tuba, as the GEF does not localise to tight junctions in this model (Bryant et al., 2010; Kovacs et al., 2011; Qin et al., 2010).
Cdc42 and apical membrane differentiation and domain size
An important feature of epithelial differentiation is the development of distinct apical and basolateral cell surface domains, with the apical membrane often developing organ-specific specialised structures, such as brush border membranes (Cereijido et al., 2008; Mellman and Nelson, 2008). Tight junctions form the apical-lateral border in vertebrates, separating these two domains and defining their relative sizes. In Drosophila, the apical-lateral border is formed by adherens junctions that are flanked apically by a marginal zone where apical polarity factors concentrate and signal. Membrane domain polarisation is the outcome of antagonistic signalling between apical polarity factors (including the Par3–Par6–aPKC module) and the scribble complex, which defines the basolateral domain (Goldstein and Macara, 2007; St Johnston and Sanson, 2011; Yamanaka and Ohno, 2008). Cdc42 activation leads to phosphorylation of Par3 by aPKC and separation of the complex. Par3 remains at the apical-lateral border, whereas the Par6–aPKC complex migrates into the apical membrane (Morais-de-Sá et al., 2010; Suzuki and Ohno, 2006; Walther and Pichaud, 2010). In vertebrates, the Cdc42 GEF that drives apical differentiation is Dbl3, a splice variant of Dbl (also known as MCF2) (Zihni et al., 2014).
Dbl3 is recruited to the differentiating apical membrane owing to an interaction with ezrin, a regulator of apical specification (Fig. 3) (Zihni et al., 2014). Stable association with the apical membrane then requires an N-terminal CRAL-TRIO domain, which differentiates Dbl3 from other Dbl isoforms; how this domain stabilises the GEF at the membrane is not clear. Dbl3 is not required for junction assembly but functions as an activator of the Par3–Par6–aPKC module following junction formation, promoting apical exclusion of Par3 and apical differentiation. Dbl3 thereby regulates apical domain size and positioning of tight junctions, thus controlling the apical-lateral border. Dbl3 localises along the apical membrane and is enriched above tight junctions in a subdomain equivalent to the Drosophila marginal zone along with other apical polarity factors, such as aPKC, ezrin and Crb3 (Pieczynski and Margolis, 2011; Tepass, 2012; Zihni et al., 2014). This indicates that the junctional configuration of an apical signalling domain that forms the interface between the apical-lateral border and the apical membrane is evolutionarily conserved.
Regulation of cell behaviour and survival by transmembrane proteins of the tight junction
Several mechanisms have been identified by which tight junctions signal to the cell interior; these include regulators of signalling cascades and proteins that cycle between the nucleus and junctions to regulate gene expression (Balda and Matter, 2009; Farkas et al., 2012; Gonzalez-Mariscal et al., 2012; Zhao et al., 2011). Our understanding of how junctional membrane proteins regulate these pathways is limited. The expression of several tight junction transmembrane proteins is deregulated in cancer tissues, suggesting that they might be important regulators of cell behaviour; however, it is generally not known whether this is a cause or consequence of transformation (Escudero-Esparza et al., 2012; Jia et al., 2013; Sawada, 2013). Here, we will focus on studies that demonstrate direct molecular links between junctional membrane proteins and tight-junction-associated signalling mechanisms.
The immunoglobulin superfamily adhesion protein JAM-A is a key regulator of junctional signalling mechanisms. During junction assembly, JAM-A recruits the Par3–Par6–aPKC complex to forming junctions (Ebnet et al., 2008). JAM-A also regulates cell–matrix adhesion and cell migration by regulating the expression of β1 integrin (Mandell and Parkos, 2005). This signalling mechanism involves a different cytoplasmic signalling complex that is composed of the Rap GEF PDZGEF2 (also known as RAPGEF6) and the adaptor afadin (also known as AF6), which promotes the activation of Rap1. Active Rap1 is then thought to promote migration by stabilisation of β1 integrin and, hence, enhanced substrate adhesion (Severson et al., 2009). Similarly, JAM-A regulates PDZGEF1-mediated Rap2c activation by binding to ZO-2 in complex with AF6, a mechanism linked to the regulation of tight junctions in the intestine (Monteiro et al., 2013). It is thought that JAM-A dimerisation in cis (i.e. on the same cell membrane) promotes Rap1 activation and migration, whereas dimerisation in trans (i.e. on neighbouring cell membranes) favours Rap2 activation and barrier formation (Monteiro et al., 2014). Through yet another mechanism, JAM-A regulates the proliferation of intestinal epithelial cells in vitro and in vivo; depletion of JAM-A leads to deregulated levels of phosphatidylinositol (3,4,5)-trisphosphate (PIP3), possibly due to reduced expression of PTEN, the phosphatase responsible for the dephosphorylation of PIP3. This results in enhanced activation of Akt, a central regulatory kinase that then promotes the phosphorylation and activation of β-catenin (Nava et al., 2011). Therefore, JAM-A interacts with distinct signalling mechanisms to guide junction assembly and cell behaviour.
BVES is another transmembrane component of tight junctions that is downregulated during the epithelial-to-mesenchymal transition (EMT) (Han et al., 2014; Williams et al., 2011). Re-expression of BVES in colorectal cancer cells is sufficient to repress proliferation and metastasis, indicating that it also has tumour suppressive activities. BVES forms a complex with ZO-1 and suppresses the GEF-H1–ZONAB pathway in corneal epithelial cells. It also regulates junctional integrity through aPKC, indicating that downregulation of BVES results in the modulation of several tight-junction-associated signalling pathways (Russ et al., 2010; Russ et al., 2011; Wu et al., 2012). However, the mechanism(s) underlying the tumour suppressor activity of BVES have not been identified.
Expression of all three tight-junction-associated MARVEL domain proteins – occludin, tricellulin and MarvelD3 – can be deregulated in different types of cancer tissues or cell lines, but only occludin and MarvelD3 have been linked to signalling mechanisms that guide cell behaviour. Strikingly, both interact with mitogen activated protein (MAP) kinase pathways. Occludin plays a role in suppressing the EMT downstream of oncogenic Raf (Li and Mrsny, 2000). Raf1 represses occludin transcription by activating slug (also known as SNAI2), a transcriptional repressor that is activated during EMT, and re-expression of occludin can rescue tight junction assembly in Raf-transformed cells by an unidentified mechanism that might require occludin stabilisation at cell junctions (Wang et al., 2005; Wang et al., 2007). Another role for occludin in cellular proliferation is suggested by its localisation to centrosomes; the phosphorylation of occludin facilitates centrosome separation, promoting mitotic entry and increased proliferation (Runkle et al., 2011). Hence, occludin is not just a factor that stabilises the epithelial phenotype, but its activity is dynamically regulated during epithelial proliferation. However, apart from gastric hyperplasia, occludin knockout in mice did not lead to phenotypes associated with increased proliferation (Saitou et al., 2000). It is currently not clear whether this reflects redundancy or an adaptive response. Alternatively, occludin might be important for proliferation control in response to specific environmental or pathological parameters.
MarvelD3, a more recently identified tight junction component, is, like occludin, not required for the formation of functional tight junctions under standard tissue culture conditions (Raleigh et al., 2010; Steed et al., 2009). In cultured metastatic tumour cell lines, MarvelD3 expression is repressed, suggesting that it also regulates cell behaviour (Kojima et al., 2011; Steed et al., 2014). Manipulation of MarvelD3 expression indeed affects epithelial cell proliferation and migration through a mechanism that involves the recruitment of MEKK1 (also known as mitogen-activated protein kinase kinase kinase 1, MAP3K1) to tight junctions, leading to inhibition of the c-jun N-terminal kinase (JNK) pathway and downregulation of genes that regulate proliferation, such as cyclin D1 (Steed et al., 2014). In contrast to occludin, MarvelD3 thus functions upstream of the MAP kinase cascade. In addition, interplay between MarvelD3 and the JNK pathway is important for the cellular response to osmotic stress, regulating cell survival and barrier integrity.
Finally, the tight-junction-associated Crb3 complex has been linked to cell-density-dependent regulation of proliferation and cancer. In vertebrate cells and Drosophila, the transmembrane protein crumbs regulates proliferation through the Hippo pathway (Chen et al., 2010; Grzeschik et al., 2010; Ling et al., 2010; Robinson et al., 2010; Varelas et al., 2010). In vertebrates, the transmembrane protein Crb3 and its cytoplasmic partners Pals1 (also known as MPP5) and PATJ interact with the tumour suppressors merlin and AMOT (Varelas et al., 2010; Yi et al., 2011). Merlin and AMOT form complexes with the machinery that regulates the activation of the Hippo pathway effectors YAP and TAZ, the Hippo pathway kinases Mst1 and Mst2 (also known as STK4 and STK3, respectively), LATS1, LATS2, and the GAP RICH1, which regulates MAP kinase signalling (Hong and Guan, 2012; Li et al., 2012). Hence, inactivation of Crb3 and its associated apical determinants results in the stimulation of YAP- and TAZ-dependent transcription and proliferation.
The cumulative evidence thus suggests that tight junctions function as multivalent adhesion complexes with several transmembrane proteins that control diverse signalling pathways that promote cell proliferation and migration. These tight-junction-associated signalling components are part of a densely interconnected protein network and multiple tight junction components are thus likely to affect these signalling pathways. For example, YAP and TAZ are not only regulated by the tight-junction-associated Hippo pathway but also interact with the tight junction adaptors ZO-1 and ZO-2, which themselves regulate other transcription factors, such as c-Myc and ZONAB (Balda and Matter, 2000; Gonzalez-Mariscal et al., 2009b; Oka et al., 2012; Remue et al., 2010). Disruption of tight junctions due to cell transformation or tissue damage in disease is thus likely to lead to the activation of multiple pathways that promote disease progression. Indeed, many viruses and bacterial pathogens that induce tissue disruption and/or promote carcinogenesis have been shown to interact with tight-junction-associated proteins and might activate such junctional signalling mechanisms.
Viruses and bacteria that target tight junctions
As a functional component of epithelial barriers, tight junctions are part of the first line of defence that prevents pathogens from entering the body. However, they also represent an Achilles' heel, as some bacteria and viruses exploit tight junction components to either invade cells and/or tissues or to promote signalling responses that facilitate tissue invasion (Fig. 4; Table 1). In this section, we provide an overview of the bacteria and viruses that interact with tight junction proteins.
The majority of bacteria that compromise tight junction integrity are ingested with food and water, and affect the gastrointestinal tract. Common examples are Clostridium, Shigella, Salmonella and enteropathogenic Escherichia coli strains that either disrupt tight junctions by promoting the internalisation or degradation of specific tight junction proteins or that inject factors into host cells, leading to a deregulation of the cytoskeleton and, indirectly, to junction disruption. Consequently, such infections frequently lead to severe diarrhoea and might contribute to the development of chronic inflammatory diseases. The study of Helicobacter pylori, a bacterium that has been linked to gastric cancer, provides a direct link between a bacterium that targets tight junctions and the regulation of proliferation. These bacteria attach to gastric epithelial cells and inject CagA (the protein encoded by cytotoxin-associated gene A) into cells. This protein stimulates the disruption of the AJCs (Amieva et al., 2003), and the expression of CagA in mice is sufficient to induce cancer (Fig. 4A) (Ohnishi et al., 2008). CagA functions as a scaffold that interacts with the tyrosine phosphatase SHP2 (also known as PTPN11), triggering deregulated ERK signalling (Backert et al., 2011). CagA also binds to and inhibits Par1 (also known as MARK), a kinase that inhibits the tight-junction-associated RhoA activator GEF-H1, promoting cytoskeletal reorganisation and proliferation (Nie et al., 2009; Saadat et al., 2007; Yamahashi et al., 2011). Hence, CagA injection results in deregulation of multiple signalling mechanisms that regulate epithelial proliferation and differentiation.
Viruses target many different tight junction proteins, ranging from transmembrane components to scaffolding proteins (Table 1) (Gonzalez-Mariscal et al., 2009a). Viruses that interact with junctional membrane proteins generally exploit them as receptors or co-receptors to invade cells (Fig. 4B). The prime example is the junctional membrane protein CAR (Bergelson et al., 1997; Cohen et al., 2001). Binding of adenoviruses to CAR leads to alterations in Rho GTPase signalling and in the cytoskeleton, thereby promoting internalisation of the virus (Amstutz et al., 2008; Kälin et al., 2010; Meier et al., 2002). At least some coxsackieviruses interact with an additional receptor, decay-accelerating factor (DAF, also known as CD55), a GPI-anchored membrane protein that localises to the apical surface. Viral binding to DAF triggers signalling events that promote cytoskeletal changes. In turn, such changes allow the virus to interact with junctional CAR, which is required for viral internalisation in an occludin- and caveolin-dependent manner (Coyne and Bergelson, 2006; Coyne et al., 2007). The CAR-related protein JAM-A is a receptor for reovirus, and studies performed in JAM-A-knockout mice have revealed that it is required for systemic dissemination, as the viruses exploit endothelial JAM-A to enter the blood stream (Antar et al., 2009; Barton et al., 2001; Schulz et al., 2012). Another prominent example is the hepatitis C virus (HCV), which requires claudin-1 and occludin for infection (Evans et al., 2007; Liu et al., 2009; Ploss et al., 2009; Zeisel et al., 2011). It has been shown that occludin binds to a viral envelope protein, but how claudins interact with the virus is not clear. The role of tight junctions themselves during viral entry is unknown (Burlone and Budkowska, 2009; Yang et al., 2008). Tight junctions might serve as a way to maintain sufficient levels of the junctional membrane proteins at the cell surface to permit infection. If dissociated from tight junctions, junctional membrane proteins can be highly dynamic and enter cells by endocytosis (Steed et al., 2010). Thus, it is likely that viruses might exploit these junctional membrane proteins as vehicles to enter the cells.
Several viruses have been identified that encode proteins that target components of the junctional plaque and, in particular, scaffolding proteins. As these proteins also bind to and regulate signalling proteins, such interactions are likely to affect cell behaviour (Fig. 4B). Several oncogenic viruses belong to this group of pathogens, such as oncogenic adenovirus serotypes that encode a protein called E4-ORF1, which binds to multiple junctional PDZ proteins, including PATJ and ZO-2, leading to defects in junction assembly and polarisation (Glaunsinger et al., 2000; Glaunsinger et al., 2001; Javier, 2008; Latorre et al., 2005; Lee et al., 2000). Similarly, the E6 oncoprotein of high-risk human papillomavirus has the potential to interact with and disrupt several polarity proteins and junctional adaptors that contain a PDZ domain, thus disrupting tight junction assembly. This is thought to confer the transforming capability of this virus (Facciuto et al., 2014; Gardiol et al., 1999; Glaunsinger et al., 2000; Hernández-Monge et al., 2013; Javier, 2008; Latorre et al., 2005; Lee et al., 2000; Nakagawa and Huibregtse, 2000; Pim et al., 2012; Storrs and Silverstein, 2007; Thomas et al., 2008). Another example is the NS1 protein of the avian influenza A virus, which binds to multiple PDZ proteins, including the tight junction adaptors MAGI1, MAGI2 and MAGI3. This binding is thought to contribute to tight junction disruption and altered gene expression (Golebiewski et al., 2011; Kumar et al., 2012).
A third group of viruses deregulate tight junctions indirectly. This group includes human immunodeficiency virus 1 (HIV-1), which stimulates a signalling cascade through one of its co-receptors, the G protein-coupled receptor GPR15. This signalling cascade leads to remodelling of the cytoskeleton and junctional disruption (Nazli et al., 2010). HIV-1 also induces increased expression of the pore-forming claudin-2 and reduced expression of the sealing claudin-1, thereby affecting epithelial barrier properties (Epple et al., 2009). Similarly, in brain endothelial cells, the HIV-1 TAT protein induces the transcriptional downregulation of occludin, as well as its degradation by metalloproteinase-9 (Xu et al., 2012). Another example is rotavirus, which disrupts the intestinal lining by disrupting tight junctions indirectly through its effect on the actin cytoskeleton (Berkova et al., 2007; Gardet et al., 2006; Nava et al., 2004; Obert et al., 2000).
Despite the mounting data regarding junctional signalling mechanisms and the large number of pathogenic bacteria and viruses that affect tight junctions, how these signalling mechanisms contribute to disease development and progression is, in most cases, poorly understood.
Tight junctions have emerged as complex bidirectional signalling centres that host diverse regulatory mechanisms guiding junction assembly and function. Tight junctions signal to the cell interior to guide cell proliferation, migration, survival and differentiation. However, we are only slowly starting to understand the interplay between junctional membrane proteins and these signalling mechanisms. We are slowly beginning to learn precisely how this interplay affects junctional functions on one hand, and how, on the other hand, the junctional adhesion proteins use these mechanisms to signal to the cell interior. Similarly, most of these mechanisms have been studied in isolation and, therefore, it is not clear how distinct signalling mechanisms cooperate and influence one another, and how they are activated in response to different stimuli. It seems likely that the functional relevance of specific membrane proteins depends on the conditions analysed, as in the case of MarvelD3; however, few studies have been performed to determine how specific junctional transmembrane proteins respond to distinct external stimuli, such as different types of stress, and how this affects junctional integrity and signalling output. To understand these processes is likely to be of pathological relevance as junction assembly is affected in many common diseases, including acute and chronic inflammations and different types of cancer. The large number of pathogenic viruses and bacteria that interact with tight junction components are thus of particular interest, as they provide excellent experimental tools to elucidate how the deregulation of junctional signalling mechanisms contributes to disease development.
The authors declare no competing interests.
The work in our laboratories is supported by the Biotechnology and Biological Sciences Research Council and Fight for Sight.
- © 2014. Published by The Company of Biologists Ltd