Biological Role of Connexin Intercellular Channels and Hemichannels

Heart

GJ in heart mainly contains three different connexins, Cx40, Cx43 and Cx45; however, Cx43 is the predominant one. Cx43 is expressed in characteristic combination with the other two connexins in a chamber-related and myocyte-type manner (29;30). The ventricular myocytes mainly contain Cx43-formed GJ, whereas the GJ in the atrial myocytes contains both Cx43 and Cx40 (31). In comparison to Cx43 and Cx40, Cx45 is expressed in low quantities with higher expression in the atria compared to the ventricle (30). The role of GJ in cardiac morphogenesis is revealed by the cardiac malformations in the connexin knockout mice (32–36). Mutations in the carboxyl terminal phosphorylation sites in Cx43 are also shown to cause complex cardiac malformations visceroatria heterotaxia (37) and hypoplastic left heart syndrome (38). Further studies fail to find Cx43 gene mutations in patients with human autosomal recessive lateralization defects (39) and with sporadic and familial heterotaxy (40). GJ is implicated in the spread of molecular signals of ischemia-reperfusion injury among myocytes resulting in rigor contracture and cell death (41;42). Ischemic preconditioning is shown to result in trafficking of Cx43 to the mitochondria (43) and the benefit of preconditioning on infarct size is shown to be abolished in Cx43 heterozygous knockout mice (44), suggesting critical contribution of this connexin to cardiac remodeling following preconditioning. In contrast another report suggests that these mice have smaller infarct size following coronary occlusion (45). These contradicting results could have been due to measurement of infarct size at different time points. Though ischemic preconditioning prevents ischemia induced Cx43 dephosphorylation, which leads to increased open probability of these channels, the underlying mechanism is still not known. Totzeck et al reported that interaction of phosphatases with Cx43 is not affected during ischemic preconditioning, thus ruling out the contribution of phosphatases in preventing ischemia induced dephosphorylation of this connexin (46). In the cardiomyocytes, Cx43 is shown to be present only in the subsarcolemmal mitochondria but not in the interfibrillar mitochondria (47). Since mitochondrial Cx43 plays a critical role in mediating cardioprotective function of ischemic preconditioning, subsarcolemmal and interfibrillar mitochondia likely differ in transmitting the signals of ischemic preconditioning. Mass spectrometric approach of purified mitochondrial preparation confirms the presence of Cx43 in the mouse myocardium (48). This study further reveals that Cx43 hexamers are present on mitochondrial membranes and this connexin contributes to the mitochondrial K⁺ uptake. Rottlaender et al (2010) (47) demonstrate that in mouse cardiomyocytes, Cx43 in the mitochondria provides cytoprotection by stimulating mitochondrial K_ATP channels, which play a critical role in conferring protection against ischemic cell injury (49;50). Further support for role of GJ in cardiac function comes from findings showing increased susceptibility of mice to lethal ventricular arrhythmias when level of Cx43 drops to 18% of control levels (51). Another study reports decreased Cx43 expression and increased Cx45 expression in the failing human ventricle (52), which could potentially alter the ratio of the two connexins and cause significant effect on channel properties and overall cardiac function. Support for this comes from the increased susceptibility of transgenic mice overexpressing Cx45 to ventricular tachycardia (53). Cx40 expression is also shown to increase in failing human ventricle but this is confined to the end-stage ischemic heart disease (54).

With regard to the cellular mechanism of regulation of GJ in cardiac disease, it is shown that microRNA1 expression is increased in diseased human ventricle, and it downregulates Cx43 expression in ischaemic heart (55). Furthermore, this study also shows that blocking of microRNA1 expression by antisense prevents arrhythmia in an experimental animal model. Recently, much more attention has been given to identify interacting partners for GJ proteins which may modulate cardiac function. In this regard, it is reported that ZO-1 (zonula occludens-1) restricts the recruitment of connexons to the GJ periphery, thereby reduces the GJ size and decreased GJ in the diseased heart (56). Furthermore, plakophilin 2 (PKP2), a component of the desmosome is shown to regulate function and distribution of Cx43 (57). These authors show that decreasing PKP2 expression by siRNA in neonatal rat ventricular myocytes causes decreased Cx43 expression, significant disruption of Cx43 gap junction plaques and reduced dye coupling between cells. All these reports firmly establish the role of GJ in cardiac function; however, future studies should be aimed at understanding the regulatory mechanisms of GJ remodeling, the role of GJ in atrial fibrillation and the nature of signaling molecules that pass through GJ to modulate cardiac remodeling.

Although most studies have focused on understanding the role of GJ in heart, some studies have characterized Cx43 HC in heart cells. Functional HC in isolated adult rabbit ventricular myocytes have been detected (58;59), giving way for the possibility of HC inside this tissue. Kondo et al (2000) (59) show electrophysiologically that removal of extracellular [Ca²⁺] and metabolic inhibition allows activation of non-selective current which may have a role in alteration of ionic fluxes thereby promoting arrhythmias and myocardial infarction. During myocardial ischemia, Cx43 is dephosphorylated and activity of several protein kinases and phosphatases gets affected (60–62). The initiation of ischemia which accounts for the first few minutes does not affect the partially phosphorylated status of the physiological sarcolemmal Cx43 (63). However, prolonged ischemia corresponds with dephosphorylation of Cx43 usually associated with opened HC (60). Studies on severe pathophysiological conditions including ischemia, hypoxia and diverse forms of metabolic inhibition suggest the role of HC in the destabilization of the ionic homeostasis and eventual cell death (58;64). During ischemia and hypoxia, the concentration of ATP is elevated in the interstitial spaces within the heart due to its release from various cell types including the cardiomyoctes. However, the mechanism for the release of ATP has been understood only recently by a study which showed a negative feedback regulation of ATP release involved HC in cardiomyocytes (65). The maxi anion channels that released the ATP suppressed the release of ATP from HC in the cardiomyocytes. In other studies signaling pathways involving mitogen-activated protein kinase (MAPK), protein kinase C (PKC) and protein tyrosine kinases (PTK) have been shown to be involved in cardiac protection during ischemia. HC phosphorylated at the MAPK signaling sites is kept closed under normal extracellular [Ca²⁺]. However, dephosphorylation of those sites by phosphatases activated by biological stress like hyperosmolarity and metabolic stress opens the HC allowing the influx of harmful ions critical to heart physiology and pathophysiology (66).

In order to understand the contribution of HC to ischemia-reperfusion injury in rat neonatal cardiomyocytes, changes in HC activity have been investigated during in vitro simulated ischemia under oxygen-glucose deprivation (OGD). Transient Cx43 HC opening and increase in intracellular [Ca²⁺] due to OGD are inhibited in the presence of a Cx43 mimetic peptide, Gap26 and lanthanum chloride as is the reduction in cell viability (67). These findings suggest that transient HC opening during OGD is responsible for cell injury. Cx43 HC is present on the sarcolemma of cardiomyocytes and remains closed under physiological conditions. However, ischemic stress opens these channels and leads to irreversible tissue damage and cell death. This study is also the first to confirm a reduction in the infarct size in the presence of Gap26, which can efficiently block HC opening thereby providing protection to intact heart against ischemia-reperfusion injury whether administered before or after the occurrence of ischemia (68). Opening of Cx43 HC appears to be an important mechanism for ischemia-reperfusion injury in the heart since Gap26 significantly reduces the infarct/risk ratio. However, pannexin-forming HC is not involved (69). Based on the studies mentioned above, it can be speculated that HC may play an important part in the ischemic preconditioning of the heart intercepted with brief periods of ischemia and timely closure of the HC. Phosphorylation by specific kinases like MAPK, PKC and CaMKII may delay the HC opening and associated deleterious consequences.

Inner ear

Several different connexins, Cx26, 30, 31, 32 and 43 are expressed in the cochlea and GJIC is observed in both the cochlea and organ of Corti (reviewed in (70)). The functional relevance of GJ in cochlear function receives much attention due to the syndromic and non-syndromic deafness caused by mutations in connexin genes. Mutations in Cx26 contribute to a large proportion of deafness caused in human population; however, other four connexins including Cx30, Cx31, Cx32 and Cx43 also have been shown to be involved in either syndromic or non-syndromic hearing loss (18;71;72). Deafness-associated connexin mutations either affect HC trafficking or produce non-functional channels or channels with defective gating and permeability properties (73). Several functions of GJ have been speculated to contribute to the normal cochlea function. In this regard, GJIC is proposed to participate in K⁺ recirculation in the cochlea and in the passage of critical metabolites like IP3 (74;75). However, the exact function of GJIC in regulating cochlear physiology is yet to be investigated.

Conductance of fluorescent dye uptake assays which cannot penetrate the membrane on dissociated cochlear cells and in acute or cultured preparations of the cochlear epithelium have successfully revealed the presence of functional HCs in cochlea (76;77). These studies indicate that cochlear HC may exert a control on hearing sensitivity by releasing ATP to the exterior that could modulate the electromobility of outer hair cells. Gain-of-function mutation resulting in abnormal HC opening at rest period has been responsible for certain deafness-related diseases (78). One deafness-linked Cx26 mutation, G45E, has been associated with a fatal form of keratitis-ichthyosis-deafness syndrome (79;80). G45E changes the charge of the amino acid side chain from neutral to negative and likely affects the Ca²⁺ binding site present on the first extracellular loop of Cx26 HC (78). Findings from Stong et al (2006) suggests that Cx26 G45E mutation causes the HC to become leaky in presence of normal [Ca²⁺] leading to cell death (81). Cell death can however be rescued by increasing the extracellular [Ca²⁺]. In the cochlea, Panxs 1 and 2 have been detected in the spiral ganglion and Scarpa’s ganglion neurons (82). A more widespread cochlear expression of Panx1 and Panx2 has been reported and the expression of Panx3 has also been detected in the cochlear bone by a recent study (83). Molecular mechanisms of the effect of mutations impairing HC function in the ear are beginning to emerge. Development of animal model for testing the deafness-associated mutations resulting in HC dysfunction requires further investigation.

Kidney

A series of earlier studies have provided evidence for the expression of mRNAs of ten different connexins in the kidney including Cx26, Cx30, Cx30.1, Cx31, Cx32, Cx37, Cx40, Cx43, Cx45 and Cx46 (reviewed in Hanner et al (2010) (84). Cx37, Cx40, Cx43, and Cx45 are expressed in the renal vasculature (85–88) with Cx40 being predominantly expressed in the endothelial cells and Cx45 being abundantly expressed in the vascular smooth muscle cells (84). Morphological evidence for the presence of GJ plaques in the tubules exists only in the proximal tubules (89;90). Cx30, Cx30.3 and Cx37 are expressed in the distal nephron; however, it is not clear whether they form GJ or just primarily act as HC, as the presence of this connexin protein is observed only in the unopposed apical membrane of the cells. Furthermore, ultrastructural studies have shown the presence of GJ between all cells of the glomerulus (91;92) with strong Cx40 expression in the entire intraglomerular mesangium and Cx37 only in the mesanglial cells at the vascular pole of the glomerulus (87;88) and Cx43 formed GJ between the podocytes (93).

Renal connexins facilitate vascular conduction between the cells of the vascular wall (94). The propagation of vasodilation induced by acetylcholine is significantly decreased in Cx40 knockout mice (95) and by connexin mimetic peptides (96–98), suggesting participation of GJ in propagation of vasodilation. In contrast to the effect of Cx40 knockout on vasodilation, vasoconstriction induced by local application of KCl is not inhibited in the absence of Cx40 (95). The myoendothelial GJ between the endothelial cells and the vascular smooth muscle cells is suggested to mediate part of the endothelial-derived hyperpolarizing factor response, by allowing agonist-induced hyperpolarization of endothelial cell membrane to spread to the vascular smooth muscle cells (94;99).

Connexins, in particular, Cx40 have been shown to regulate renin secretion. In Cx40 knockout mice the renin producing cells disappear from the media of the afferent arteriole and appear in the glomerular tuft, the extraglomerular mesangium and the periglomerular interstitium (100). These mice are hypertensive and have increase in plasma renin concentration (101). Selective knockout of Cx45 in nestin-expressing cells results in a phenotype showing increased expression and secretion of renin, and increased blood pressure (102). Though data from Cx43KI32 mice, where Cx43 is replaced with Cx32, points to role of Cx43 in renin secretion (103), Conditional knockout of Cx43 in endothelial cells has no effect on renin secretion (104;105), suggesting that more studies are needed to clearly establish a role for Cx43 in renin secretion.

In spite of the expression of connexins in the tubules, not much is known about their physiological role in the tubules. Some studies however, indicate that the level of Cx30 and Cx37 is modulated by dietary salt levels (106;107). Even though most kidney cells have GJ, knockout of individual connexins have only minor effect on development of kidney (108–110). Connexin mimetic peptides against Cx37, Cx40 or Cx43 increase blood pressure, which is prevented by co-treatment with losartan, the angiotensin receptor blocker (111), suggesting that connexin-mediated modulation of blood pressure is through the renin-angiotensin system. Data from Cx40 knockout mice supports these observations (112).

Activation of Cx43 HC has a role in the pathogenesis of renal ischaemic lesions. Human renal proximal tubule cells (hPT cells) express Cx43 protein in the plasma membrane. Moderate ATP depletion activates HC in hPTcell monolayers and increases cell death. HC activation may dissipate solute fluxes across the cell membrane and amplify other deleterious effects of ATP depletion. These fluxes, if not appropriately replaced, will alter intracellular concentrations causing cell damage and impairing the recovery from the metabolic insult when the ischaemia is relieved (113;114). Expression of Cx30 has been observed throughout the luminal membrane of select cells in the distal nephron with the highest level of staining observed in the distal convoluted tubule. Indeed, high salt upregulates Cx30 expression which may function as a HC involved in the regulation of salt reabsorption in the distal nephron. More detailed studies on HC in the kidney are required to understand their role and regulatory mechanism.

Lens

Lens expresses three different connexins including Cx43, 46 and Cx50. The lens with an anterior epithelium and posterior fibers continues to grow in volume throughout the life of an organism, differentiating new fiber cells from the equator region. The older fiber cells do not turnover and remain in the interior of the lens. The lens differentiating fiber cells synthesize high amounts of the soluble protein crystallin to achieve transparency and high refractive index, and then lose their nuclei and the light scattering organelles. Thus the lens fiber cells are dependent on the anterior epithelial cells for their metabolic needs. The lens fiber cells communicate with each other and the epithelial cells through large number of GJ (115). The lens epithelium has abundant expression of Cx43 and Cx50 (116;117), and Cx46 and Cx50 form the GJ between the lens fiber cells (118;119). Mutations in Cx46 and Cx50 have been linked to development of lens congenital cataracts in humans and rodents (reviewed in Jiang (2010) (22)). Some of these are loss-of-function mutations, whereas others are dominant negative in nature.

Even though Cx46 and Cx50 colocalize in all junctional plaques in the fiber cells, targeted deletion of these connexins results in unique cataracts with distinct difference in the timing of onset and morphology (120;121). Cx46 knockout mice develop nuclear cataracts and the opacity of the lens is possibly due to accumulation of cleavage products of γ-crystallin (120). Furthermore, loss of Cx46 causes accumulation of calcium and subsequent activation of Lp82, a calcium-dependent protease (122). The deletion of Cx50, but not that of Cx46, results in delayed postnatal growth, decreased lens size and micropthalmia (121). Stimulation of epithelial proliferation and fiber cell differentiation has been proposed as the underlying mechanism of Cx50-mediated lens growth (123;124). Our group reports that chick Cx50, but not Cx46 and Cx43 stimulates epithelial to fiber cell differentiation and this stimulatory effect appears to be independent of epithelial cell proliferation (123). Furthermore, our group shows that the C-terminus of Cx50 is required for Cx50-mediated lens cell differentiation (125). Another report suggests that in Cx50 knockout mice epithelial cell proliferation is compromised (124). Lenses without Cx50 lose their sensitivity to pH in response to acidification in the differentiating lens fiber region (126). Replacing the coding region of Cx50 with that of Cx46 restores lens transparency but fails to rescue the normal growth (124;127).

Cx43 knockout neonatal mice lenses have loose apposition of epithelial cells and fiber cells with dilated extracellular spaces and intracellular vacuoles (128). In contrast to this finding, another study shows that Cx43 knockout lenses have normal development through embryonic day 19 (129). The Cx43 and Cx50-double knockout mice lenses are structurally normal; however, there is great reduction of GJ between the epithelial cells and the epithelial-fiber cells with no change in degree of communication between fiber cells (129).

Although an extensive network of GJ spanning the vertebrate lens is essential for transparency and homeostasis, several reports point to other roles of Cx50 which are not related to GJ function. Overexpression of chick Cx50 promotes lens epithelial-fiber differentiation, but fails to increase intercellular coupling (123). GJ-deficient mutants stimulate lens epithelial-fiber differentiation to similar level as wild-type Cx50 (125). Several studies demonstrate the roles of Cx46 and Cx50 HC in lens development. Cx46-forming HC is gated by calcium and voltage, and is also mechanosensitive (130;131). Due to their mechanosensitive nature, Cx46 HC is proposed to possibly assist in accommodation of the lens by providing transit path for volume flow as the lens changes shape (130). HC formed by Cx50 is sensitive to cations like Na⁺ and K⁺ (132). The density of Cx50 HC is also directly proportional to the magnitude of the Cx50 Ca²⁺-sensitive current. Structural studies indicate that there are far more Cx50-forming GJ channels compared to Cx50-HC in the lens (133). A study from our laboratory has shown that PKA activation enhances both Cx50 GJ and HC functions (134). Cx50 in the chick lens is phosphorylated in vivo by PKA at residue Ser-395, which is a highly conserved amino acid residue across different animal species.

Liver

This is one of the first organs where gap junction channels were detected. In isolated liver, plasma membranes, Revel and Karnovsky in 1967 showed the presence of hexagonal arrays of subunits in the intercellular junctions (135). In 1974, Goodenough characterized the gap junction proteins from mouse hepatocytes and named them as connexins (136). Gap junctions are abundant in hepatocytes and mostly are composed of Cx32 and Cx26 (137;138). Cx37 and Cx40 are mostly expressed in the liver vascular cells (139), whereas Cx43 is expressed in the nonparenchymal liver cells (140). Apart from a few conflicting reports, most studies show that in the hepatocyte life cycle, GJIC increases during the G1 phase but decreases dramatically upon initiation of the S phase (141–146). Similar pattern is observed for the expression of Cx32 and to a great extent for Cx26, however, Cx43 expression remains unchanged (141;144;146). Though an alteration in GJIC and connexin expression is observed during hepatocyte cell cycle, the relevance of such changes is not clearly understood. In the Cx32 knockout mice, the proliferative activity of the hepatocytes of regenerating rat liver is not increased, however, the extent of synchronous initiation and DNA synthesis termination is decreased, suggesting decrease GJIC likely permits cell cycle progression but does not provide any direct signal for cell division (146) (147). In contrast to this others believe that gap junctions play a very critical role in regulating cell proliferation rather than just assisting in cell cycle progression. In line with this, overexpression of Cx32 and Cx26 in liver epithelial cells and hepatoma cells increases expression of p27 and E-cadherin respectively leading to decrease cell proliferation (148;149). Apart from changes in connexin expression during proliferation, several groups have shown changes in expression of these gap junction proteins during differentiation of early rat hepatic progenitor cells into adult liver parenchymal cells (150;151). GJIC regulates several liver specific processes including albumin secretion, ammonia detoxification (152), glycogenolysis (153), bile secretion (154) and xenobiotic phase I transformation (155–158). Since gap junctions play critical role in regulating hepatocyte homeostasis, these structures are often affected during liver toxicity. Xenobiotics induced hepatotoxicity is associated with decreased connexin expression (159;160). Furthermore, Cx32 dominant-negative mutant transgenic rats are more resistant to chemically induced hepatic injury (161). In PLC/PRF/5 hepatoma cells, overexpression of Cx26 accelerates apoptosis (162), whereas in Hep3B hepatoma cells and WB-F344 liver epithelial cells Cx43 GJIC decreases during acute choline deficiency apoptosis (163). Wilson et al reported that GJIC increases during early phases of apoptosis whereas declines during late stages in serum deprived rat WB-F344 liver epithelial cells (164). This increase of GJIC during the initial stages of apoptosis could be to spread the cell death signal from cell to cell.

Cx32 HC have been shown to contribute to the apoptotic to necrotic transition during Fas mediated hepatocyte cell death. Fas ligands (FasL) binds to its receptors at the cell surface and triggers apoptosis in the liver (165;166). In this study, siRNA and mimetic peptides were used to inhibit Cx32 expression and the authors show that Cx32 increases the transformation of apoptotic cells to necrotic state, partially due to the ability of Cx32 to form functional HC (167). In another recent study by Vinken et al (168), Cx43 mediated signaling was shown to be responsible for induction of spontaneous apoptosis in primary hepatocytes. Cx43 expression, localization, gap junction and HC activity were monitored in freshly isolated adult rat hepatocytes. Inhibitors to Cx43 resulted in downregulation of both HC and GJ activity which paralleled with decreased expression and activity of caspase 3 as well as reduced expression of Bid.

Lungs

Gap junctions play critical roles in the lung. GJIC in the airways regulate calcium signaling between ciliated epithelial cells which thus regulates ciliary beating (169). Gap junctions also play a role in regulating secretion of pulmonary surfractants by type II alveolar epithelial cells, in passing calcium transients from the type I alveolar cells to the type II cells (170) and in propagating intercellular signals from one alveolus to the other (171). The endothelial cells in the normal lung predominantly express Cx37, Cx40 and Cx43 (172–174). Cx26, Cx32, Cx43 and Cx46 are the major connexins expressed by alveolar epithelial cells and Cx30.3 and Cx40 are expressed at a very low level (175).

Cx43 in the pulmonary endothelium is required for propagation of calcium waves along pulmonary vessels as these waves are absent in the Cx43-deficient mouse model (176). These calcium waves induced by mechanical stimulation are shown to increase pulmonary endothelial P-selection expression at the cell surface, suggesting a role of Cx43 in spreading proinflammatory signals. This effect of Cx43 is counterbalanced by Cx37, which is found in circulating monocytes, macrophages and macrophage foam cells in early and late atherosclerotic plaques (177). Cx37 HC can release ATP into the extracellular space (178) and thereby prevent leukocyte adhesion in primary monocytes and macrophages and inhibit inflammation (179). In addition, a recent study indicated to the role of pannexin 1 for the release of ATP in the airway epithelial cells (180).

Connexins in the lung also contribute to improved barrier function. Though lung barrier function is mainly regulated by tight junctions, gap junctions are often seen adjacent to tight junction strands (181). Role of gap junctions in improving barrier function comes from studies where inhibition of GJIC by glycyrrhetinic acid or oleamide decreases barrier function by 50 to 75% (182) and expression of Cx32 increases barrier function of immortalized hepatocytes from Cx32-deficient mice by ~25% (183). In addition to the role of GJIC in improving barrier function, this form of intercellular communication also plays important role in preventing lung injury. Mice lacking Cx40 and endothelial Cx43 develop symptoms similar to pulmonary fibrosis spontaneously (184). However, Cx40 alone or endothelial Cx43 alone deficient mice did not show any obvious pulmonary phenotype. However, these mice showed decrease lung barrier function, disorganized alveoli and extracellular matrix deposition after 8 weeks of birth (104;185;186). Although different connexins are expressed in the lung, Cx43 seems to play important role in alveolar function. Interestingly, newborn Cx43 knockout mice are cyanotic (32), however, this could have been caused due to defect in cardiac outflow. In this regard, lung targeted Cx43 knockout would provide valuable insight into the role of Cx43 in overall functioning of lung.

Mammary Gland

The expression and function of connexins are dynamically regulated throughout mammary gland development and differentiation. In normal human mammary epithelial cells, expression of Cx26 and Cx43 messengers and proteins is observed by Northern blot and immunocytochemistry, respectively, and functional GJIC is revealed by dye transfer experiments (187). Normal human mammary fibroblasts are also shown to express Cx43 and heterocellular GJIC is observed between the fibroblasts and epithelial cells (188). Cx43 was shown to be mostly present in the myoepithelial cells (189–191), with sparse expression in some luminal cells of the mammary gland in situ (192) and Cx26 expression is found predominantly between the luminal cells of ducts (190). From these expression patterns, it seems likely that Cx43 is required for myoepithelial differentiation, whereas Cx26 has more variable role in luminal cell function. Cx26 protein levels reach highest expression at the onset of lactation and decreases during involution (193;194). Cx26 is shown to be present between luminal cells of the ducts at the basolateral borders and in all stages of development (189;194). Though this expression profile of Cx26 suggests its role in milk production and secretion, its expression in virgin mouse mammary glands and nonpregnant human breast tissue suggests that this connexin possibly also has a role in the differentiation and homeostasis of this gland. Conditional suppression of Cx26 in the mammary epithelium before puberty leads to the impairment of development of lobular structure of the gland and lactation; however, conditional knockout during pregnancy has no developmental outcome (195).

Apart from Cx43 and Cx26, mouse mammary gland also expresses Cx32 and Cx30. Cx30 mRNA expression is identified in pregnant and lactating mouse mammary gland (194;196) and the protein expression is first detected at postnatal day 15 and followed a peak onset at lactation and decline in involution similar to Cx26 (194). Cx32 mRNA is first detected at parturition and lactation (189;197). Though studies from several groups report its expression at the basolateral regions of the luminal cells in the lactating gland (189;193;197), Talhouk et al (2005) (194) show the expression of this protein throughout all developmental stages. In the Cx32-null mice, the mammary glands develop and function normally suggesting that this connexin is dispensable for the development of the mammary gland (195). Cx26 and Cx32 co-oligomerize in the luminal epithelial cells of the mouse mammary gland and at the onset of parturition Cx32 is only found in heteromeric channels having Cx26 and during lactation the ratio of Cx32 increases to form the homomeric Cx32 connexons, probably to allow the passage of larger molecules including cAMP and cGMP (198).

In contrast to the expression pattern of Cx26 and Cx30, Northern and quantitative PCR analysis reveal that Cx43 transcript decreases during mid-pregnancy, disappears during lactation and reappears during involution (194;199). During last days of pregnancy and at the onset of lactation, a shift to the more highly phosphorylated species of Cx43 is evident without any change in the total Cx43 protein level (194). Cx43 is shown to be expressed at the cell surfaces between the basal myoepithelial cells (189) and at the myoepithelial-epithelial junction (194). Cx43 knockout mice die at birth (32) and conditional mammary gland-specific Cx43 knockout mice is not available. However, heterozygous knockin mice in which Cx43 gene is replaced by Cx32 show defects in milk ejection, though they have normal mammary gland development and milk production (200). Inhibition of GJIC in mammary epithelial cells in vitro inhibits secretion of the milk protein -casein (201). These studies implicate role of connexins and GJIC in mammary gland development, milk secretion and ejection.

Nervous System

The central nervous system (CNS) has 11 different connexins and is highly coupled through GJ network (202;203). Cx26, 30, 32, 36, 37, 40, 43 and 45 are well expressed in the brain with a unique expression pattern for each one during differentiation (204). Intercellular coupling occurs between neurons, astrocytes, oligodendrocytes, microglia, ependymal cells and also between astrocytes-oligodendrocytes or neurons-astrocytes (reviewed in Talhouk et al (2008)(205)). Connexin function is well established in neurogenesis, cell migration, neuronal development, differentiation and morphogenesis (206;207). Inhibition of GJ function in postnatal day 19 mouse embryonic carcinoma cells in a model of neuronal and glial differentiation prevents the appearance of astrocytic and neuronal phenotype (208), suggesting role of GJIC in glial cell function. Role of GJ in neuronal migration is elucidated by the delayed neocortical neuronal migration in the Cx43-null mutant mice (207). Knockdown of Cx26 and Cx43 in embryonic day-16 cortex by shRNA prevents migration of neurons along radial glia in the intermediate zone and loss of cells that arrive in the lower and upper cortical plates (209). Interestingly, these migration defects are rescued by a channel-dead mutant whereas mutants with loss of connexon pairing (not HC activity) and C-terminal truncations fail to rescue the defects, suggesting that channel-independent function of connexins contributed to proper neuronal migration. Mutations in Cx32 underlie the X-linked Charcot-Marie-Tooth syndrome leading to failure of myelination of the Schwann cells (210).

GJIC plays a critical role in determining susceptibility of brain to hypoxia-ischemia injury. In an in vitro model of ischemia, inhibition of GJ function by carbenoxolone and by Cx26 and 32 knockdown significantly reduces cell death (211). Similar neuroprotection is also observed upon inhibition of GJ function by chemical inhibitors or in organotypic slices from Cx43 knockout mice in a traumatic brain injury model (212). Further evidence for role of GJIC in neuronal ischemic injury comes from studies, where administration of carbenoxolone to ischemic pups after intrauterine hypoxia-ischemia decreases neuronal damage (213). Several in vivo studies also confirm contribution of GJ to ischemia. Pretreatment of rats with octanol, the gap junction blocker during ischemia induced by occlusion of the middle cerebral artery reduces the infarct volume (214). Administration of octanol in a transient forebrain ischemia model decreases neuronal cell death (215). All these experimental evidences clearly point to the role of GJ in contributing to the cell death after hypoxia and ischemia; however, GJ themselves undergo significant changes following ischemia. Thus future studies should be directed towards the underlying mechanism of GJ modulation of cell death during ischemia including the nonchannel- dependent function of connexins in regulating susceptibility of cells to hypoxia-ischemia.

Both connexin and pannexin HC form an integrated network of channels present in the astrocytes and neurons. The majority of the cell types in the CNS have the potential to express HC activity but the physiological relevance of HC has just begun to be uncovered. One of the major GJ forming proteins of astrocytes is the Cx43. Cx43 HC in astrocytes and C6 glioma cells may mediate the release of gliotransmitters like glutamate and ATP. Interestingly, several reports demonstrate that Cx43 expression and zero extracellular [Ca²⁺] potentiate ATP release from C6 gliomas which is inhibited by GJ/HC blockers (14;216;217). However, the role of Cx43 HC in ATP release is not a widely accepted concept. It has also been suggested that ATP release from astrocytes is mediated by the purinergic receptor, P2X7 and not by the Cx43 HC since ATP release was absent in P2X7−/− spinal cord astrocytes but not those from Cx43−/− mice (218). Several other connexin-forming HC have been observed in the different cells of the nervous system with associated pathological implications. GABAnergic interneurons in the neocortex and hippocampus as well as neurons in the olivary nucleus and cerebellum express Cx36 (219;220). Although, formation of Cx36 GJ is known, formation of Cx36 HC is as yet questionable. There are contrasting views about the presence of HC formed by Cx36. One study suggests that Cx36 does not form HC because oocytes injected with Cx36 mRNA showed no HC activity even in the absence of extracellular [Ca²⁺] (221). However, another study reported Cx36 HC opening in cortical neurons in the presence of very high K⁺ (100 mM) or KCN with the release of ATP and induction of ischemic tolerance (222). Moreover, Cx36 siRNA-treated neurons show reduction in cell death induced by combined KCl/KCN. Cx45 is also expressed in axons. Although biophysical properties of Cx45 are extensively characterized, functional role as a GJ and HC is not quite clear. Under normal conditions of negative membrane potential and external 2 mM [Ca²⁺], the open probability of the Cx45 HC is low suggesting a strong influence of membrane potential on the HC activity (223).

Cx30 is also expressed in astrocytes (224) and cochlea (225). It forms GJ and mutations in Cx30 cause deafness (226). However, to this date there is no strong evidence to support the role of Cx30 HC in astrocytes but two mutations in Cx30 (A88V and G11R) form functional HC at the cell surface but generate a leakage of ATP when expressed in HeLa cells (227). Currently, evidence for connexin-based HC in neurons and astrocytes is weak and needs further validation in the in vivo system to support studies done in cultured cells. Several reports have provided center stage to the possible role of pannexins as HC in the brain. Among the three known isoforms of pannexins, Panx1 and Panx2 are expressed in the brain in the hippocampus, neocortex, cerebellum, thalamus and hypothalamus whereas the Panx3 appears solely in skin and osteoblasts (228–230). Panx1 expression has been reported in cultured astrocytes, oligodendrocytes and neurons (231), but is absent in C6 glioma cells (26). Neurons show increased expression of Panx2 but astrocytes express Panx2 only upon ischemia/reperfusion (232). In a recent study (233), FGF1 released during spinal cord injury induces activation of spinal actrocytes in culture to release ATP during early time period of 2 hours by Panx1 HC activated by P2X7 receptor, and 7 hours later, both Cx43 HC and Panx1 HC are involved in the ATP release. Interesting speculations also indicate that signaling molecules released from one cell type (neuron) may affect the HC on a nearby cell type (other neurons or astrocytes). Appreciably, the current status of the connexin and pannexin HC in the CNS clearly suggest of cooperative gating mechanisms or features of the conduction pathway that differentially regulate the ionic and metabolic fluxes.

Skeletal Tissues

GJ regulates development, differentiation and remodeling of the bone. Though three different connexins including Cx43, Cx45 and Cx46 are expressed in the bone cells, Cx43 is the most abundant isoform. Direct evidence for a role of Cx43 in skeletal development comes from the malformation of limbs and truncation of limb buds in chick embryo after incubation with Cx43 antisense oligonucleotide (234). Similar observation is reported in dissociated embryonic chick mandibular mesenchyme treated with Cx43 antisense, which is further confirmed by the delayed ossification of the cranial vault, vertebrae, clavicle, ribs and limbs in Cx43-null embryos (235). The osteoblasts isolated from these mice also have decreased mineralization potential as compared to wild-type control cells. ODDD in human patients due to Cx43 mutations causes craniofacial abnormalities, syndactyly of the hands and foot and hypoplasia (23;236;237). Mutagenesis screening in mice identifies a G60S substitution in Cx43 which results in bone defects similar to that of human ODDD (238). The role of Cx43 in adult skeleton has been studied using conditional knockout of Cx43 in osteoprogenitors, osteoblasts or osteocytes. Osteoblast-specific deletion of Cx43 in mice causes low bone mass, reduced bone formation and attenuated anabolic response to PTH treatment (239). Deletion of Cx43 using Dermo1 (Twist2)-Cre system, which drives expression of Cre in mesenchymal condensations, the cells that give rise to osteoblast and chondrocyte lineages, develops age-related osteopenia (240). In contrast to these findings, knockout of Cx43 using Cre recombinase driven by osteocalcin promoter, which drives Cre expression in the mature osteoblasts, does not have any effect on overall bone mass (241). However, bone mineral density (BMD) of these mice is measured between 2 and 4.5 months of age, thus, effect of Cx43 knockout on bone density after peak bone mass accrual cannot be ruled out. Further evidence for regulation of bone development by Cx43 comes from transgenic mice overexpressing microRNA 206 (miR206) (242). These mice have a low bone phenotype. Cx43 is found to be a target of this microRNA and overexpression of Cx43 in the miR206 expressing osteoblasts restore their differentiation potential. In 3D cultures overexpression of Cx43 in bone marrow stromal cells is shown to enhance osteogenic differentiation (243). GJIC is shown to regulate chondrocyte differentiation as inhibition of GJ function decreases production of proteoglycans and type II collagen (244). Similar to chondrocytes, GJIC is also shown to positively regulate osteoblast differentiation (245). Apart from the role of connexins in overall bone development, GJ and HC have also been suggested to regulate mechanotransduction in the musculoskeletal system (246;247). Fluid flow and mechanical stimulation are shown to increase GJIC and open Cx43 channels leading to the release of important bone anabolic molecules (4). HC has been demonstrated to exist both in osteoblasts and osteocytes (248–250). Clearly, HC in osteocytes appears to have essential, distinctive roles in transmitting the signals elicited by mechanical stimulation to other bone cells and the extracellular matrix, and further promote bone formation and remodeling (246). Some years ago, we found that in addition to GJ, Cx43 forming HC also mediates the biological responses like PGE₂ release elicited by fluid flow (16). Fluid flow-mediated release of PGE₂ increases the surface expression of Cx43 to open HC.

Besides the release of bone modulators like PGE₂, Cx43 HC may allow the transfer of other key molecules to and from osteocytes; however, the identity of these factors remains to be defined. Bisphosphonates have also been shown to open the Cx43 HC present on the osteocytic cells by triggering the activation of Src and ERK kinases for promoting cell survival (249). Study by Genetos et al (2007) (251) demonstrates that fluid flow activates HC in MLO-Y4 osteocytic cells through a mechanism involving PKC, which induces ATP and PGE₂ release. To understand how Cx43 HC are responsive to mechanical stimulation, we study the mRNA expression of the proteins upregulated upon mechanical stimulation and affected by blocking activity of Cx43 HC using the antibody against the extracellular loop of Cx43 (252). We recently show that integrin α5β1 associates with Cx43 forming HC in bone osteocytic cells and facilitate channel opening upon shear stress (253). Integrins are known mechanosensors on cell surface as mechanical stimulation applied on bone can be directly sensed by integrin molecules and transmit the signals across cell membrane (254–256). Although, osteocytes are the known mechanosensory cells in the bone but which part of the osteocytes is the real sensor of mechanical stimulation which opens the Cx43 HC has been a mystery. Our recent study shows that Cx43 HC opening is only observed in the cell body part of the osteocytes and not the dendrites suggesting the presence of HC on the cell body (257).

HC formed by pannexins has also been reported in bone osteoblasts and osteocytes (258). Panx3 expression has been demonstrated in osteoblasts (83). In a very recent finding, Panx3 promotes osteoblast differentiation by functioning as an ER Ca²⁺ channel and a HC, and by forming GJ (259). HC formed from Panx3 releases ATP and activates the purinergic receptors which further activates PI3K/AKT signaling and promotes osteoblast differentiation. Clearly, HC formed in the bone has the ability to control the survival and death of the different bone cells and thereby regulate the bone remodeling process apart from the GJ.

Testis

Expression of 11 different connexin isoforms, including Cx26, Cx31, Cx31.1, Cx32, Cx33, Cx37, Cx40, Cx43, Cx45, Cx46 and Cx50, has been reported in the testis among which Cx43 is the predominant connexin (260). Direct evidence for a role of Cx43 in testis comes from a 50% depletion of primordial germ cells (GCs) in fetal male mice with Cx43 knockout (261) and the decreased number of GCs is demonstrated to be a result of increased apoptosis of these Cx43 knockout cells (262). This same observation is also confirmed by an in vitro study, which shows that Cx43 GJ function regulates GC number by affecting survival of spermatogonia (263). The decrease in Cx43 expression in the knockout mice not only affects the number of GCs, but also the expression patterns of other testicular connexins. The knockout mice have detectable mRNA expression of four different connexins as opposed to wild-type expressing eight members of connexin family (261). The role of Cx43 in spermatogenesis is also confirmed by knockin studies in which Cx43 gene is substituted by either Cx26 (264), Cx32 or Cx40 (200) coding sequences. These mice also show defective spermatogonial amplification further emphasizing critical role played by Cx43 channels in regulating testis function. The sertoli cell-specific Cx43 conditional knockout mice have reduced testis size, weight, an arrest of spermatogenesis at the level of spermatogonia or sertoli cell-only syndrome. Furthermore, these mice show reduced number of spermatogonia and increased number of sertoli cells per seminiferous tubules (265;266). These knockout mice and in vitro studies provide evidence for negative regulation of sertoli cell proliferation by Cx43 (263;267). Taken together these studies point to critical role played by connexin 43 and GJIC in regulating testicular function. However, the exact mechanism by which GJ or HC regulates the development of GC is still unknown. Until now, not much is known about the possibility of HC in the testis.

Ovary

The ovarian follicles of rodents express Cx32, Cx37, Cx43 and Cx45 (Reviewed in Kidder and Mhawi, 2002 (268)). Since Cx43 and Cx37 are the most predominant connexins in the ovary of many species, their role in ovarian function has been studied the most. GJIC is reported to regulate folliculogenesis, meiotic arrest and apoptosis of female gonad. Communication between oocytes and the somatic cells is established from primordial stage of follicle development and continues throughout the follicular development period (269), GJ has been suggested to transfer amino acids and nucleotides from the somatic cells to the oocyte (270;271). In the Cx37 knockout mice folliculogenesis is arrested at the early antral stage, the oocytes do not reach their full size and also fail to have full meiotic potential (272;273). Since Cx43 knockout mice die soon after birth, ovaries from prenatal mice are grown either in vitro in organ culture or in vivo under the kidney capsule of wild type mice. Under both in vitro and in vivo condition, folliculogenesis is arrested at the primary follicle stage and oocyte growth is retarded (261;274). These oocytes fail to undergo meiotic maturation and fertilization. GJ has been shown to play a role in the transfer of cAMP, which maintains the fully grown oocytes in a meiotically arrested state, from the somatic cells to the oocyte (275). Decreased ovarian Cx43 expression in acute hyperglycemia and chronic diabetes results in increased number of apoptotic follicles (276). In consistence with this, ovaries of Cx43 knockout females have reduced number oocytes (261) due to increased primordial germ cell apoptosis (262). Decreased expression of Cx43 in apoptotic follicle is also observed in bovine and porcine ovary (276;277). In contrast to decreased expression of Cx43 in the apoptotic follicles, Cx32 expression is increased in the bovine apoptotic follicles (277).

Cx43 is strongly expressed in the granulosa cells of the ovary where it forms both GJ and HC. It is also required for the ovarian follicle development in the mouse. Indeed, a study shows that application of mechanical stimulation or reduction of extracellular divalent cations in the granulosa cells trigger Cx43 HC opening (278). ATP release is also detected, and can be abolished by connexin-channel blockers. Cx43-deficient granulosa cells do not exhibit any of the HC-mediated activities suggesting that HC account for the essential role of Cx43 in folliculogenesis.