Introduction
Proper cardiac function requires the synchronous mechanical and electrical activity of individual cardiomyocytes to ensure the coordinated excitation and contractile performance of the heart, as an organ. The intercalated disc (ID), a unique membrane structure forming at the edges of mammalian cardiomyocytes (Li and Radice, 2010), fulfills this role by allowing the transmission of mechanical and electrical activity between neighboring cells; (reviewed in Delmar and McKenna, 2010; Noorman et al., 2009).
A brief history of the ID
A decade later, Engelman described the heart as a continuous syncytium, while a century later Weidmann suggested the presence of "membrane areas of synchronicity", characterized by low resistance that allows the transmission of electrical potential (Engelmann, 1875; Weidmann, 1952).
The idea of a continuous region connecting two cells was challenged in the mid 1950′s by several groups who used electron microscopy to show that cardiac cells are separated from one another by a specialized extension of the sarcoplasm oriented transversely with respect to the cell’s boundaries (Sjostrand and Andersson, 1954; Van Breemen, 1953). Since the middle of the 20th century, we have significantly advanced our understanding of the structure and composition of the ID. Accordingly, the ID was found to be a highly organized structure composed of three main junctional complexes; the gap junctions, which enable the propagation of electrical stimuli throughout heart cells, the adherens junctions, and the desmosomes, which provide mechanical coupling and stability to cardiomyocytes, respectively. The advent of electron microscopy in the 1950′s to 1970′s further provided detailed visualizations of the regions connecting two cardiomyocytes (Fawcett and McNutt, 1969; McNutt et al., 1970; Muir, 1957; Rayns et al., 1969; Sjostrand and Andersson, 1954; Van Breemen, 1953). Recently, novel cellular isolation techniques combined with scanning or transmission electron microscopy (SEM and TEM, respectively) have yielded three-dimensional images of the ID (Hoyt et al., 1989; Shimada et al., 2004; Tandler et al., 2006), showing that in the mammalian ventricular heart, IDs are arranged both transversely and longitudinally in a stairwell like fashion with steps and risers. Transverse or plicate segments, resembling the steps, run in a zigzag arrangement with finger-like micro-projections, and contain mainly adherens junctions and desmosomes with smaller regions of gap junction plaques. Longitudinal or interplicate segments resemble the risers and contain mainly desmosomes and larger areas of gap junction plaques. The many folds and projections found within this region, increase the surface area of the ID, providing the cardiac cells with superior intercellular communication.
Spatiotemporal distribution of ID components
During cardiomyocyte development and maturation, major changes occur in structures associated with the ID. Studies using human myocardium showed that during embryonic development adherens junction and desmosomal organization follows that of gap junctions (Pieperhoff and Franke, 2007). However, during postnatal development proteins of the adherens and gap junctions appear to orient themselves at IDs simultaneously (Peters et al., 1994). Moreover, in vivo studies of lower mammals (including rodent, bovine and canine) have shown that at embryonic stages and postnatal day 1, components of gap junctions, desmosomes and adherens junctions are uniformly distributed throughout the sarcolemma, mutually exclusive from one another (Angst et al., 1997; Hirschy et al., 2006). However, at later postnatal stages (days 6-20), proteins of the adherens junctions and desmosomes begin to concentrate towards the termini of cardiomyocytes, leaving proteins of gap junctions uniformly distributed at the plasma membrane. By postnatal day 90, all components of the three junctions are segregated and organized at IDs. These findings were also supported by in vitro studies using primary cultures of rat and mouse cardiomyocytes (Geisler et al., 2010; Kostin et al., 1999). Interestingly, the latter further demonstrated that when individual cardiocytes are allowed to make contact in culture, proteins of the adherens junctions are the first to assemble and "mark" the location of the developing ID, closely followed by desmosomal proteins and finally proteins of gap junctions (Geisler et al., 2010; Kostin et al., 1999). Supporting this, the organization of adherens junctions and desmosomes is independent of gap junctions; however, gap junction organization requires that of adherens junctions and desmosomes (Gutstein et al., 2003; Wei et al., 2005). Taken together, these observations suggest that proteins necessary for mechanical coupling, i.e. components of adherens junctions and desmosomes, create the appropriate environment for proteins mediating electro-chemical coupling, i.e. those associated with gap junctions.
Organization of the ID
Gap junctions mediate the direct communication between neighboring cells by forming a low resistance pathway for the transmission of signals and electrical current (Rohr, 2004). A gap junction is composed of twelve connexin proteins, with connexin-43 being the most prominent in mammalian cardiomyocytes, along with low amounts of connexin-45 and 40 (Beyer et al., 1987; Vozzi et al., 1999). Each cardiomyocyte contributes six connexin molecules to form a hemi-channel, or a connexon; two connexons join to form a pore or gap junction channel, which is isolated from the extracellular space and connects the cytosol of two neighboring cells (Sohl and Willecke, 2004). These channels are responsible for the occurrence of synchronous contractions throughout the heart (Sohl and Willecke, 2004).
Consequently, in the absence of connexin-43 channels, normal propagation of contraction is disrupted, and lethal arrhythmias develop (Gutstein et al., 2001a; Gutstein et al., 2001b). Adherens junctions facilitate the transmission of contractile force from one cell to the next and are crucial in maintaining mechanical strength uniformly across the heart (Tepass et al., 2000). They are mainly composed of transmembrane cadherins and cytosolic catenins (Niessen, 2007). N-cadherin, the main cardiac isoform, is a transmembrane protein, with extracellular and intracellular components (Niessen, 2007). Its extracellular portion forms homodimers bringing together the membranes of two opposing cells, while its intracellular segment forms a complex with various members of the catenin family (a-, P-, y- and p120) present in the cytosol, which in turn are linked to the actin cytoskeleton (Bass-Zubek et al.,2009). Consequently, adherens junctions serve as anchors between the extracellular space and the actin cytoskeleton (Noorman et al., 2009).
Desmosomes provide structural support to cardiomyocytes, which are subjected to strong contractile stress (Delmar, 2004). Desmosomes, similar to adherens junctions, are composed of intercellular and intracellular components (Rayns et al., 1969). The intercellular component consists of desmosomal cadherins, desmocollin and desmoglein, which form a hetero-complex within the extracellular space joining together two bordering cells (Green and Simpson, 2007), while the intracellular component consists of proteins of the armadillo/catenin (plakoglobin and plakophilin) and plakin (desmoplakin) families (Bass-Zubek et al., 2009). Desmoplakin directly interacts with intermediate filaments to stabilize the desmosomal structure. Importantly, a high incidence of mutations within genes encoding desmosomal proteins has been linked to the development of arrhythmogenic right ventricular cardiomyopathy (ARVC).
Although, the ID has been traditionally described to contain three distinct structures (i.e. gap junctions, adherens junctions and desmosomes), recent technological advancements indicate that they are more interwined than originally proposed (Delmar and McKenna,2010). Consistent with this, adherens junctions and desmosomes are intimately associated in the "area composita" where proteins from both structures are present (Borrmann et al., 2006; Franke et al., 2006). Similarly, proteins of the adherens and gap junctions have been shown to interact directly (Delmar, 2004). Taken together, these observations suggest that the ID is actually a single functional unit where macromolecular complexes interact to maintain structural integrity and synchronous contraction throughout the heart.
Bridging the gap between the ID and the sarcomeric cytoskeleton is a newly defined region termed the transitional junction. This area is rich in structural proteins, including spectrin, ankyrin-G, a-actinin and the NH2-terminal region of titin, which typically localizes to the Z-disc (Bennett et al., 2006). The transitional junction is suggested to connect the ID with the contractile apparatus, mediating the transmission of force between adjacent cardiocytes. The high degree of complexity and organization of junctions at the ID suggests a tight interplay between mechanical and electrical activities. Disruption of either mechanical or electrical coupling leads to irregular conduction of electrical impulses and deterioration of cardiac function, subsequently resulting in the development of cardiac arrhythmias. Various mutations in genes encoding for ID proteins have been causatively linked to these complex disorders, many of which manifest themselves as ARVC; (recently reviewed in Protonotarios et al., 2011).
There are ~200 known proteins that are associated with the ID (Dowling et al., 2008; Estigoy et al., 2009; Geisler et al., 2007; Lin et al.; Kargacin et al., 2006; Satomi-Kobayashi et al., 2009;Schroen et al., 2007; Seeger et al., 2010). Herein, we provide a summation of the current knowledge on the junctional structures present in the ID, focusing on their most prominent and influential components, and how these relate to each other and the sarcomeric cytoskeleton in normal and disease states.
Gap junctions
Gap junctions were first described by Revel and Karnovsky in 1967, as "hexagonal arrays" that localize to the ID and mediate the electrical and metabolic coupling of adjacent cardiomyocytes by allowing the diffusion of small molecules (<1000 Da) (Elfgang et al., 1995; Ravel and Karnovsky, 1967). At gap junctions, the distance between opposing membranes is ~3 nm (Fig. 1; Perkins et al., 1997). Gap junction plaques can contain from a few up to 200,000 connexon channels (Evans et al., 2006).
Fig. 1. Gap junctions in ventricular cardiomyocytes are composed of two homo-hexameric hemi-channels. forming a channel or a connexon. Each hemi-channel consists of six connexin-43 monomers (shown in dark purple), allowing the transmission of electrical current and small signalling molecules from adjoining cardiomyocytes. Zona Occludens-1 (ZO-1) (depicted in light purple) interacts directly with connexin-43. In addition, the connexin-43 complex interacts with members of the caveolin family (shown in light grey) that target gap junctions to lipid rafts, and cytosolic a/P tubulin heterodimers (shown in dark grey) that link gap junctions to the microtubular network.
Structural organization of connexons: Connexin-43
Connexin-43: The human connexin super-family is composed of at least twenty-one members. Connexin-43 is the predominant form expressed in the human heart, while connexins 40 and 45 are present in lower amounts (reviewed in Sohl and Willecke, 2004). Connexin-43 is a four-pass transmembrane protein that contains a cytoplasmic loop and two extracellular loops (Fig. 2A). Notably, both its NH2- and COOH- termini are located in the cytosol (reviewed in Sohl and Willecke, 2004). Three conserved cysteine residues, located in the extracellular loops, have been implicated in disulfide bond formation between neighboring connexins of adjacent cells, and contribute to the development of a tight seal that prevents the exchange of materials with the extracellular matrix (Unger et al., 1999). Consistent with this, a constitutive connexin-43 null murine model is embryonic lethal (Reaume et al., 1995), while a cardiac-specific knock-out model exhibits sudden cardiac death by 2 months of age (Gutstein et al., 2001a; Gutstein et al., 2001b).
Fig. 2. Schematic representation of the domain structure of major ID proteins. Grey ovals denote protein specific domains.
NMR studies have demonstrated the presence of short, flexible a-helical segments in the cytoplasmic loop and the COOH-terminus of connexin-43, which provide binding sites for several proteins and mediate gating of the connexon (Duffy et al., 2002; reviewed in Gonzalez et al. , 2007). Consequently, connexons can exist in a closed or open conformation; at high Ca2+ concentrations (i.e. 1.8 mM), they tend to adapt a closed conformation, however, in the absence of Ca2+ they exist in an open state (Thimm et al., 2005). Importantly, the gating of connexons is regulated by additional factors, including pH, levels of Mg2+, voltage as well as the phosphorylation status of connexins (please see below; Bukauskas and Verselis, 2004; Delmar, 2004; Ek et al., 1994; Ek-Vitorin et al., 1996; Gonzalez et al., 2007; Matsuda et al., 2010).
Propagation of electrical stimulation throughout the heart
The propagation of electrical stimulation is the driving force for heart contraction. It originates at the sinoatrial (SA) node, traverses through the atria, crosses the atrioventricular (AV) node and propagates through the bundle of His and the Purkinje fibers before it activates the ventricles. The coordinated contraction of the atria and ventricles is achieved by conduction of the electrical impulse at variable speeds, mediated by the different forms of connexins, which confer to gap junction plaques distinct electrophysiological properties. As such, connexin-45 is preferentially expressed in the SA and AV nodes, but co-expressed with connexin-43 in the bundle of His and the Purkinje fibers. Conversely, connexin-43 is primarily present in the ventricles, but also co-expressed with connexin-40 in the atria; (reviewed in Severs et al., 2008).
Conferring low conductance in a single homotypic channel, connexin-45 is distributed at SA node in a sparse and scattered pattern that ensures poor coupling between adjacent cardiocytes. Similarly at the AV node, connexin-45 contributes to the sequential activation of the atria and ventricles reducing the occurrence of arrhythmias; (reviewed in Severs et al., 2008). On the contrary, the rapid propagation of electrical signals through the Purkinje fibers is mediated by gap junctions mainly consisting of connexins 43 and 40, which confer relatively large conductance, and to a lesser extent connexin-45, thus maintaining the regular contractions of the heart (Gonzalez et al., 2007; Kirchhoff et al., 1998).
Phosphorylation regulates the permeability of connexons
Several kinases modulate the function of connexons. Although a complete listing of all identified kinases is beyond the scope of this topic, we will refer to major ones, highlighting their roles during normalcy and stress. Connexin-43 is a substrate of Src tyrosine kinase, which phosphorylates Tyr-265 to disrupt its interaction with ZO-1 (discussed below, Toyfuku et al., 2001), and suppress gap junction communication in the failing heart (Giepmans et al., 2001a; Toyofuku et al., 2001). Similarly, mitogen-activated protein kinase (MAPK) phosphorylates connexin-43 at Ser-255, Ser-279 and Ser-282 to repress gap junction communication (Warn-Cramer et al., 1998; Warn-Cramer et al., 1996). Conversely, phosphorylation of Ser-365 by protein kinase A (PKA) promotes gap junction assembly and communication (Burghardt et al., 1995; Solan et al., 2007; TenBroek et al., 2001). Although several isozymes of protein kinase C (PKC) phosphorylate connexin-43 in diverse cell types and tissues, PKCe is the only isoform that phosphorylates it at the ID (Bowling et al., 2001; Doble et al., 2000; Lampe et al., 2000; Lin et al., 2003; Saez et al., 1997). Consistent with this, PKCe suppresses gap junction communication in the ischemic heart through phosphorylation of connexin-43 at Ser-368 (Ek-Vitorin et al., 2006; Hund et al., 2007; Hund et al., 2008).
Several phosphorylation sites (i.e. Ser-306 and Ser-325/Ser-328/Ser-330) on connexin-43 are non- or de-phosphorylated in ischemic and hypertrophic hearts (Lampe et al., 2006; Procida et al., 2009). The absence of phosphorylation at these sites has been suggested to correlate with reduced cardiac conductance (Lampe et al., 2006; Procida et al., 2009). In agreement with this, transgenic mice in which Ser-325/Ser-328/Ser-300 were substituted by glutamic acid (phosphomimetic residue) were less susceptible to arrhythmia. Yet, the kinase(s) that is responsible for these phosphorylation events remain(s) to be identified. Importantly, Ser-325/Ser-328/Ser-330 are substrates of casein kinase 1 (CK1) in normal rat kidney cells (Cooper and Lampe, 2002), however, its role in cardiac muscle remains to be defined. Moreover, Ca2+/Calmodulin-dependent protein kinase II (CaMKII) is capable of phosphorylating many Ser residues on connexin-43 in vitro, including Ser-306, Ser-325, Ser-328 and Ser-330 (Huang et al., 2011), however the physiological significance of these results requires further investigation.
The phosphatases acting upon and regulating the activities of connexin-43 have been also long sought after. Receptor protein tyrosine phosphatase | (RPTP|) has been suggested to dephosphorylate Tyr residues present in connexin-43 in lung cells (Giepmans et al., 2003), however, its physiological relevance in the myocardium remains to be established. Moreover, serine/threonine phosphatase type 1 and type 2A (PP1 and PP2A, respectively) have been implicated in the dephosphorylation of connexin-43 (Ai and Pogwizd, 2005; Duthe et al., 2001; Jeyaraman et al., 2003). For instance, PP1, but not PP2A, modulates the phosphorylation status of Ser-368 (Jeyaraman et al., 2003). Conversely, PP2A exists in a complex with connexin-43 in homogenates prepared from patients suffering from dilated cardiomyopathy (DCM) or idiopathic dilated cardiomyopathy (IDCM), as well as from a non-ischemic heart failure rabbit model (Ai and Pogwizd, 2005; Ai et al., 2011). Consistent with this, application of specific PP2A inhibitors prevented uncoupling of cardiocytes in the rabbit failing heart (Ai and Pogwizd, 2005).
Connexin-43 interacts with ZO-1, caveolins and microtubules at the ID
Zona occludens-1: Zona occludens-1 (ZO-1) interacts with connexin-43 in cardiac myocytes via its PDZ2 domain that directly binds to the last five residues present in the COOH-terminus of connexin-43 (Giepmans and Moolenaar, 1998; Giepmans et al., 2001a; Toyofuku et al., 1998). Interestingly though, their interaction is not abolished in a transgenic murine model that expresses a truncated form of connexin-43 that is missing the last 124 amino acid residues (Maass et al., 2007), suggesting that additional domains contribute to binding.
The interaction of ZO-1 and connexin-43 mainly takes place at the periphery of the gap junctional plaque (Hunter et al., 2005; Zhu et al., 2005). Notably, their binding is suppressed in the presence of Src (Sorgen et al., 2004; Toyofuku et al., 2001). A number of early studies suggested that ZO-1 targets or retains connexin-43 to the ID, while others proposed that it regulates the size of gap junctions or the internalization of connexin-43 (Barker et al., 2002; Hunter et al., 2005; Rhett et al., 2011; Toyofuku et al., 1998). Intriguingly, recent studies from failing human hearts have provided conflicting results. Bruce et al. reported that in hearts of DCM and IDCM patients, ZO-1 interacts more extensively with connexin-43 compared to healthy ones (Bruce et al., 2008), whereas Laing et al. and Kostin, described diminished colocalization of connexin-43 and ZO-1 in hearts from patients with DCM, ischaemic cardiomyopathy and end-stage heart failure (Kostin, 2007; Laing et al., 2007). In support of this, transgenic mice lacking ZO-1 are embryonic lethal, exhibiting cardiac developmental challanges (Katsuno et al., 2008; Xu et al., 2008). Recently, Rhett et al. proposed a model, whereby ZO-1 interacts with connexin-43 to inhibit the incorporation of additional connexons into gap junctional plaques (Rhett et al., 2011).
Caveolin-1: Caveolins are the main scaffolding components of caveolae in lipid rafts, and have been found to interact with connexin-43 in different cell types (Langlois et al., 2008; Liu et al., 2010; Schubert et al., 2002). While the caveolin scaffolding domain along with the COOH-terminus of caveolin-1 are sufficient to support binding to connexin-43, the respective interacting region of the latter has yet to be defined (Schubert et al., 2002). Contrary to epithelial cells, the interaction between caveolins and connexin-43 in the myocardium is less understood. Along these lines, caveolin-3, which is specifically expressed in heart and skeletal muscle (Tang et al., 1996), has been shown to interact with connexin-43 in a yeast-two-hybrid study and confirmed by co-immunoprecipitation assays using heart homogenates (Liu et al., 2010). As caveolin-3 is present at the sarcolemma, but not the ID, the physiological relevance of this interaction remains to be examined (Abi-Char et al., 2007; Yarbrough et al., 2002). Moreover, a murine model lacking caveolin develop DCM at early stages (Zhao, 2002).
Microtubules: First characterized as binding partners of connexin-43 in RAT-1 cells and other fibroblast and epithelial cell lines, a/P-tubulins have been shown to specifically interact with the tubulin-binding motif present in the COOH-terminus of connexin-43 (Giepmans et al., 2001a; Giepmans et al., 2001b). Immunofluorescence studies and live-cell imaging further demonstrated that connexin-43 co-localizes with tubulins along microtubule tracks, as it traverses from the Golgi apparatus to other membranes (Giepmans et al., 2001b; Lauf et al., 2002; Shaw et al., 2007). To date, only a handful of studies have described the interaction between tubulins and connexin-43 in the heart. Accordingly, a recent study by Smyth et al. proposed that EB1, a microtubule plus-end tracking protein, is required to deliver connexin-43 to the ID (Smyth et al., 2010). Consistent with this, disruption of the interaction between EB1 and microtubules (e.g. during ischemia) significantly decreases the surface expression of connexin-43 at the ID (Smyth et al., 2010).
