The interaction of ion channels with the actin-based cytoskeleton in epithelial cells not only maintains the polarized expression of ion channels within specific membrane domains, it also functions in the intracellular trafficking and regulation of channel activity. Initial evidence supporting an interaction between epithelial ion channels and the actin-based cytoskeleton came from patch-clamp studies examining the effects of cytochalasins on channel activity. Cytochalasins were shown to either activate or inactivate epithelial ion channels. An interaction between the actin-based cytoskeleton and epithelial ion channels was further supported by the fact that the addition of monomeric or filamentous actin to excised patches had an effect on channel activity comparable to that of cytochalasins. Through the recent application of molecular and proteomic approaches, we now know that the interactions between epithelial ion channels and actin can either be direct or indirect, the latter being mediated through scaffolding or actin-binding proteins that serve as links between the channels and the actin-based cytoskeleton. This review discusses recent advances in our understanding of the interactions between epithelial ion channels and the actin-based cytoskeleton, and the roles these interactions play in regulating the cell surface expression, activity, and intracellular trafficking of epithelial ion channels.
- membrane transport
- actin-binding protein
- scaffolding protein
ion channels are essential for a variety of the physiological processes performed by epithelial cells, including extracellular ion and volume homeostasis and signal transduction. During differentiation, epithelial cells polarize, leading to the formation of two distinct plasma membrane domains, the apical and basolateral domains. Establishment and maintenance of ion channels within a specific plasma membrane domain play an important role in determining the physiological function of an epithelial cell. The clustering and retention of ion channels within a particular plasma membrane domain are, in part, mediated through interactions with actin, actin-binding proteins, or scaffolding proteins. In addition, interactions between epithelial ion channels and actin or actin-binding/scaffolding proteins play a role in the regulation of channel activity and in their intracellular trafficking.
Initial evidence supporting an interaction between epithelial ion channels and the actin-based cytoskeleton came from patch-clamp studies examining the effects of cytochalasins on channel activity. Cytochalasins, agents that depolymerize actin filaments, were shown to either activate or inactivate epithelial Na+ (15, 16), Cl− (44, 89, 94, 104), and K+ channels (44, 117, 118). An interaction between actin and epithelial ions channels was further supported by the fact that the addition of either monomeric or filamentous actin to excised patches had an effect on channel activity comparable to that of cytochalasins (15, 16, 87–89, 117). However, it was unclear from these studies whether the effects of actin on ion channel activity were related to actin directly binding to the channel or a channel-associated actin-binding protein or whether the observed effects were mediated through the activation or inactivation of cytoskeleton-associated second messengers.
Through the application of molecular and proteomic approaches, we now know that actin can interact directly or indirectly with epithelial ion channels. The interactions between epithelial ion channels and actin that have been elucidated to date are predominantly indirect, being mediated through scaffolding proteins [e.g., Na+/H+ exchanger regulatory factors (NHERFs)] or actin-binding proteins (e.g., α-actinin, spectrin) that serve as a link between the channel and the actin-based cytoskeleton (Table 1). Scaffolding proteins and actin-binding proteins typically interact with ion channels through conserved, modular protein domains. Such conserved motifs include the Src homology 3 (SH3) domain that binds to proline-rich amino acid sequences (Pro-X-X-Pro) (62, 82) and the PDZ domain, a 80- to 90-amino acid modular protein interaction domain, named after the three proteins in which this domain was first identified (PSD-95/Discs large protein/ZO-1) (45, 96). The PDZ domains of scaffolding proteins typically bind to short consensus sequences situated within the extreme COOH termini of epithelial ion channels (see Refs. 45 and 96 for a review). PDZ domain-containing proteins possess multiple protein-protein interaction domains that allow them to serve as scaffolds for the assembly of ion channels and signaling molecules into macromolecular complexes (45, 96).
In this brief review, we present an overview of our current understanding of the interactions between epithelial ion channels and the actin-based cytoskeleton, and the roles these interactions play in regulating the cell surface expression, activity, and intracellular trafficking of epithelial ion channels. We first summarize recent advances in our understanding of the interactions between epithelial ion channels and the actin-based cytoskeleton that are mediated through scaffolding and actin-binding proteins. We then conclude with a discussion of recent studies that support a direct interaction between the actin-based cytoskeleton and epithelial ion channels.
Indirect Association of Epithelial Ion Channels with the Actin-Based Cytoskeleton
In epithelial cells, the NHERF family of scaffolding proteins has been identified as a link between ion channels and the actin-based cytoskeleton (97) (Fig. 1). The NHERF family is composed of four members that possess either two or four PDZ domains. NHERF1 (EBP50) and NHERF2 (E3KARP) have two PDZ domains, whereas NHERF3 (PDZK1, CAP70, NaPi CAP-1) and NHERF4 (IKEPP, NaPi CAP-2) have four PDZ domains (22, 109, 119). The PDZ domains of the NHERF family members interact with the type I class of PDZ domain-binding motifs (S/T-x-I/V/L/M). In addition to the PDZ domains, NHERF1 and NHERF2 have an ezrin-radixin-moesin (ERM)-binding domain (22, 109, 119). This domain enables NHERF1 and NHERF2 to bind to members of the ERM family of proteins, which in turn links the complex to the actin-based cytoskeleton (83, 111, 124) (Fig. 1). The indirect linkage of epithelial ion channels to the actin-based cytoskeleton by members of the NHERF family has emerged as a common mechanism whereby ion channels interact with the actin-based cytoskeleton in epithelial cells. NHERF1 and NHERF2 were initially identified as binding partners for the apical membrane Na+/H+ exchanger type 3 (NHE3) (120, 126). Subsequent studies have demonstrated that CFTR (36, 98, 103, 116), TRPV5/6 epithelial Ca2+ channels (24, 81, 112), TRPC4/5 nonselective cation channels (108), ClC-3 Cl− channels (29, 78), and ROMK K+ channels (125) interact with members of the NHERF family in epithelial cells.
Our current understanding of the interaction between epithelial ion channels and members of the NHERF family stems from the extensive studies on the interactions between CFTR and NHERF1 and NHERF2 (1, 32, 35, 36, 52–54, 65, 70, 72, 90, 91, 98, 103, 116). The COOH terminus of CFTR possesses a consensus PDZ-binding motif [(D/E)-T-(R/K)-L] that is conserved across species (32, 52). This motif has been shown to bind to NHERF1 and NHERF2, thereby linking CFTR to ezrin and F-actin (36, 98, 103, 116) (Fig. 1). Direct binding studies have demonstrated that the COOH terminus of CFTR binds to PDZ domain 1 of NHERF1 and NHERF2 with high affinity and to PDZ domain 2 with lesser affinity (36, 98, 103, 116); however, the binding affinities of both PDZ domains for CFTR are within the range of affinities reported for other PDZ domain-protein interactions (91). Foskett and co-workers (91) have presented data indicating that binding of CFTR to PDZ domains 1 and 2 of NHERF1 promotes CFTR dimerization. Addition of a recombinant NHERF fragment containing both PDZ domains to excised patches from the human Calu-3 airway submucosal cell line increased the open probability (Po) of single CFTR channels, whereas addition of fusion constructs containing either PDZ domain 1 or 2 alone was without effect (91). Addition of fusion proteins containing either PDZ domain 1 or 2 was able to competitively inhibit the stimulation of CFTR by the fusion protein containing both PDZ domains. The concentration dependence of the PDZ domain 1 and 2 fusion construct on CFTR Po was biphasic, with lower concentrations activating CFTR and higher concentrations inhibiting CFTR (91).
Recently, ezrin has been shown to play a role in promoting the formation of CFTR dimers by NHERF1 (54). Binding of the ERM domain of ezrin to NHERF1 regulates the cooperative binding of NHERF1 by bringing the two COOH-terminal tails of CFTR into spatial proximity with each other to form a CFTR dimer. Ezrin binding changes the confirmation of NHERF1, resulting in the activation of PDZ domain 2 for binding to the COOH termini of CFTR, forming a 2:1:1 CFTR-NHERF1-ezrin complex (54). Without ezrin binding, the COOH terminus of CFTR interacts only with PDZ domain 1 of NHERF1 to form a 1:1 CFTR-NHERF1 complex (54). It should be noted that ezrin is both an actin-binding protein and a PKA anchoring protein (AKAP) (23). Although it is tempting to speculate that ezrin may link CFTR to both actin and PKA through NHERF1 or NHERF2 (102), there is no direct evidence supporting the role of ezrin as an AKAP linking PKA to CFTR (43).
The interaction between CFTR and NHERF1 is dynamically regulated by phosphorylation. PKC-mediated phosphorylation of serine residue S162 in PDZ domain 2 of NHERF1 decreases the binding of CFTR to NHERF1 and inhibits the stimulatory effect of NHERF1 on CFTR channel Po (90). Based on the identification of consensus phosphorylation sites in the PDZ domains of several PDZ domain-containing proteins, Foskett et al. (90) have proposed that phosphorylation of PDZ domains may represent a mechanism whereby the interaction between ion channels and PDZ domain-containing proteins are dynamically regulated.
NHERF1 and NHERF2 assemble CFTR into apical signaling complexes that sequester CFTR in close proximity to ion channels (125) and transport proteins (59), kinases, and other regulatory molecules (2) (Fig. 1). NHERF2 interacts with both CFTR and Yes-associated protein 65 (YAP65) in the apical membrane of airway epithelia (67). The non-receptor-mediated tyrosine kinase c-Yes is sequestered in this complex through its association with YAP65 (67) (Fig. 1B). Naren and colleagues (72) have revealed that the β2-adrenergic receptor, which possesses a COOH-terminal PDZ binding motif that interacts with NHERF1 and NHERF2, is sequestered with CFTR into a macromolecular complex within the apical membrane of Calu-3 cells through their interaction with NHERF2 (Fig. 1A). In addition to being physically associated in this complex with the β2-adrenergic receptor, CFTR is functionally coupled to the β2-adrenergic receptor, as activation of the receptor stimulates CFTR activity (72). Deletion of the PDZ-binding motif from CFTR abolishes both the physical and functional interactions between CFTR and the β2-adrenergic receptor. Together, these data indicate that modulation of the interaction between CFTR with NHERF may be a mechanism for regulating the signaling complexes formed between CFTR and specific receptors and kinases.
In addition to incorporating CFTR into macromolecular signaling complexes, the interaction of CFTR with NHERF1 and NHERF2 sequesters and/or stabilizes CFTR in the apical plasma membrane domain. Stanton and colleagues (65, 70, 71) have demonstrated that deletion of the PDZ-binding motif (ΔTRL) of human CFTR eliminates the polarized expression of CFTR to the apical membrane in Madin-Darby canine kidney (MDCK) and 16HBE14o− airway epithelial cells and it also eliminates CFTR-mediated transepithelial Cl− secretion in MDCK cells. Recently, an elegant study from this group (35) measured the diffusional mobility of green fluorescent protein (GFP)-CFTR chimeras in the apical membrane of MDCK cells to determine whether PDZ domain interactions tether CFTR to the actin-based cytoskeleton. Photobleaching of GFP-CFTR chimeras at the plasma membrane was performed to determine both the diffusion coefficients and lateral mobility of the channels. Following photobleaching, wild-type GFP-CFTR fluorescence in the bleached area recovered to 90% of its initial level, indicating that nearly all of the channels were mobile in the plane of the membrane, with a diffusion coefficient (D) of 0.99 ± 0.09 × 10−10 cm2/s (35). Activation of neither PKA nor PKC altered the diffusion of the wild-type channels, indicating that although association or dissociation of proteins with CFTR may occur during stimulation, they do not alter the diffusion of CFTR (35). GFP-CFTR ΔTRL chimeras were also mobile, with D increased by 60% compared with wild-type GFP-CFTR chimeras. However, both chimeras had similar mobilities in the endoplasmic reticulum (ER) of brefeldin A-treated cells (∼12 × 10−10 cm2/s) (35). Interestingly, treatment with cytochalasin D, an actin-depolymerizing agent, or jasplakinolide, an actin-stabilizing agent, increased the mobility of wild-type GFP-CFTR chimeras but did not alter the mobility of GFP-CFTR ΔTRL chimeras (35). These data indicate that wild-type CFTR is not tethered statically at the plasma membrane but its diffusion is dependent on PDZ-domain mediated interactions and interactions with an intact actin-based cytoskeleton (35). Moreover, these data demonstrate for the first time that the PDZ domain-mediated interactions of CFTR with scaffolding proteins are dynamic and occur on a time scale of seconds or faster. The dynamic nature of the interaction between CFTR and NHERF explains the findings of Naren and co-workers (53) that NHERF constitutes only ∼2% of the protein in macromolecular complexes isolated by chemical cross-linking of endogenous CFTR in the apical membrane of Calu-3 cells.
Patch-clamp studies by Prat et al. (89) have indicated that F-actin also plays a role in regulating the activity of CFTR. When mouse mammary adenocarcinoma cells stably expressing CFTR were treated with cytochalasin D, an increase in the whole-cell Cl− current was observed. This increase in whole cell Cl− current resembled the cAMP-mediated whole-cell Cl− current attributed to the expression of CFTR in this cell line (92). An increase in the whole-cell Cl− current was not observed when the parental (nontransfected) cells were treated with cytochalasin D. When actin was added to the cytoplasmic side of excised inside-out patches from the CFTR-expressing cell line, CFTR was activated (89). In contrast, the same treatment on excised inside-out patches from the parental cell line did not induce Cl− channel activation (89). In addition, when cells were exposed to cytochalasin D for >6 h, protein kinase A (PKA) activation of CFTR in inside-out patches did not occur. However, when actin was added to these patches, CFTR channel activity was observed. When DNase I, which prevents actin filament formation, and filamin, an actin filament cross-linking protein, were added to the cytoplasmic side of inside-out patches, actin-induced CFTR channel activity was completely inhibited (89). The authors interpreted these findings to indicate that “short” F-actin filaments were responsible for CFTR activation (89). To gain further understanding into the role of the F-actin network in the cAMP/PKA activation of CFTR, Prat and co-workers (88) used a human melanoma cell line lacking the gene for an actin cross-linking protein, ABP-280. A cross-linked F-actin network was found to be necessary for cAMP/PKA-mediated activation of CFTR. In cells that were deficient in either ABP-280 or both ABP-280 and CFTR, there was no increase in whole-cell Cl− current when they were treated with a cAMP-stimulatory cocktail (88). However, in cells that expressed both ABP-280 and CFTR, there was a dramatic increase in whole-cell Cl− current in response to the cAMP cocktail (88). A similar response was observed in cells that were dialyzed with exogenous ABP-280. When PKA was added to the cytoplasmic side of inside-out patches, single-channel Cl− currents were detected only in patches from cells expressing both CFTR and ABP-280 (88). These data were interpreted to mean that an organized actin-based cytoskeleton is necessary for cAMP/PKA-mediated CFTR channel activation. Recently, Chasan et al. (17) have used atomic force microscopy (AFM) to examine whether there is a direct interaction between of F-actin and CFTR. Functional CFTR that had been reconstituted into phospholipid vesicles was placed on freshly cleaved mica, and actin was added to the sample before AFM imaging (17). Actin filaments were observed to be associated with the CFTR molecules. The authors interpreted these data as supporting a direct interaction between actin and CFTR (17). Although the AFM data are highly suggestive of an interaction between CFTR and actin, biochemical data corroborating a direct interaction between the two proteins are lacking.
Association of epithelial ion channels with actin-binding proteins.
Recent studies have revealed that epithelial ion channels are also indirectly linked to the actin-based cytoskeleton through actin-binding proteins, such as α-actinin and spectrin. Li and co-workers (57) have demonstrated that polycystin-2 (PC2, TRPP2) associates with α-actinin, an actin-binding and -bundling protein. PC2, a member of the TRP superfamily of ion channels (68), is a Ca2+-permeable cation channel that is encoded by the PKD2 gene (30, 37, 50, 60, 66, 80, 113). Mutations in PKD2 are associated with 10–15% of the patients having autosomal dominant polycystic kidney disease (ADPKD) (80). Although the subcellular distribution of PC2 is a subject of debate, PC2 has been localized to the plasma membrane, cilia, cell junctions, and ER of epithelial cells (80, 122). Li et al. (57) identified the spectrin-repeat domains of α-actinin-1 and α-actinin-2 as interacting with the NH2 and COOH termini of PC2. Nonmuscle α-actinin was found to markedly stimulate the single-channel activity of PC2 reconstituted into planar lipid bilayers (57). These authors interpreted the data to indicate that the interaction between PC2 and α-actinin connects PC2 to the actin-based cytoskeleton for both structural and regulatory purposes (57). In light of the role of α-actinin and actin filaments in cellular morphogenesis, migration, and proliferation, Li et al. postulated that the functional interaction between PC2 and α-actinin may be an important link between cyst formation and the altered cell adhesion, proliferation, and migration observed in ADPKD (80).
In addition to α-actinin, this group has presented data supporting an interaction between PC-2 and tropomyosin 1 (55). Tropomyosins are ubiquitously expressed actin-binding proteins that participate in a variety of cellular processes including cell-cell adhesion, cellular morphogenesis, and suppression of neoplastic growth (33, 34, 61, 85, 86, 95). The interaction between PC2 and tropomyosin 1 was identified in a yeast two-hybrid screen of human kidney and heart cDNA libraries using the COOH terminus of PC2 as the bait (55). Evidence of a direct interaction between tropomyosin and the COOH terminus of PC-2 was further supported by in vitro pull-down and overlay assays and by coimmunoprecipitation of both heterologously expressed and endogenous PC2 and tropomyosin (55). Elucidation of the interaction between the COOH terminus of PC2 and tropomyosin 1 reveals an additional mechanism through which PC2 can indirectly interact with the actin-based cytoskeleton. Because tropomyosin-1 functions as a suppressor of neoplastic growth in transformed cells (61, 86, 95), it has been proposed that tropomyosin contributes to cyst formation and growth when the anchorage of PC2 to the actin-based cytoskeleton via tropomyosin 1 is altered in ADPKD patients with mutations in PC2 (55).
Witzgall and co-workers (26) identified Hax-1, a cortactin-binding protein, as a protein interacting with a region between transmembrane domains 5 and 6, which forms the pore region of PC2. Both PC2 and Hax-1 were found to coimmunoprecipitate when heterologously expressed in mammalian cells. The interaction between Hax-1 and PC2 is highly specific because Hax-1 does not bind to the comparable region of PC2L, the PC2-related protein (26). Immunofluorescence microscopy revealed colocalization of PC2 and Hax-1, as well as colocalization of Hax-1 with cortactin (26). These data indicate an indirect interaction between PC2 and actin via the proteins Hax-1 and cortactin. Two models were proposed to explain the functional significance of the interaction among PC2, Hax-1, and cortactin (26). In the first model, integrins, as integral membrane proteins mediating binding to the extracellular matrix, signal through non-receptor tyrosine kinases to cortactin. Cortactin connects to Hax-1, which in turns binds to PC2. Because the region between transmembrane domains 5 and 6 forms the pore region of PC2, Hax-1 may regulate ion fluxes from the ER mediated through PC2 situated within the plasma membrane or ER (26). A noted above, PC2 has been localized to both the ER and the plasma membrane (80, 122). In the second model, PC2 as a plasma membrane protein is associated with the extracellular matrix through its interaction with polycystin-1 (PC1). Signaling occurs through the interaction between PC2 and Hax-1, which in turn would be mediated by cortactin and its interaction with actin (26). In ADPKD patients with mutations in PC2, signaling through cell-matrix contacts would be altered, leading to cyst formation and enlargement.
In a similar fashion, Rotin and co-workers (93) have demonstrated a direct interaction between the α-subunit of the epithelial sodium channel (ENaC) and α-spectrin. α-Spectrin exists as a heterodimer with β-spectrin, an actin-binding protein (4). ENaC is situated in the apical membrane of Na+-reabsorbing epithelia, and it is rate limiting for net Na+ reabsorption because it mediates the cellular entry of Na+ from the luminal fluid during the first stage of electrogenic transepithelial Na+ transport (3, 5, 6, 28). The COOH-terminal domain of α-ENaC contains two proline-rich regions that resemble SH3 domain-binding motifs (3, 7, 28). Rotin et al. (93) have presented four lines of evidence demonstrating that the second proline-rich domain of α-ENaC mediates binding to the SH3 domain of α-spectrin: 1) endogenous α-spectrin coimmunoprecipitated with α-ENaC from MDCK cells overexpressing α-ENaC; 2) a fusion protein of the COOH terminus of α-ENaC bound to both native α-spectrin and a fusion protein of the SH3 domain in vitro assays; 3) a fusion protein of the second proline-rich region bound to the SH3 domain of α-spectrin and binding could be inhibited by mutagenesis of the proline-rich region; and 4) microinjection of a COOH-terminal α-ENaC fusion protein into rat alveolar epithelial cells, which express apically located α-spectrin, resulted in the apical localization of the fusion protein, whereas a microinjected NH2-terminal fusion protein remained diffuse within the cytoplasm. Based on these data, Rotin and colleagues concluded that the interaction of α-ENaC with α-spectrin is involved in maintaining the polarized distribution of ENaC within the apical membrane. Zuckerman et al. (128) have presented data demonstrating an in vivo interaction between endogenous α-ENaC and α-spectrin in Xenopus laevis A6 renal epithelial cells, a model Na+-transporting epithelium.
Direct Interaction of Epithelial Ion Channels with the Actin-Based Cytoskeleton
There is increasing evidence that ion channels can directly interact with actin in epithelial cells. Bear and co-workers (1) have demonstrated that actin directly interacts with and modulates the activity of the ClC-2 chloride channel. ClC-2 has been implicated in both cell-volume regulation and fluid secretion by airway and gastric mucosa epithelia (1). When heterologously expressed in X. laevis oocytes, ClC-2 is activated by hyposmotic shock and hyperpolarization (1). Treatment of X. laevis oocytes expressing ClC-2 with the actin-disrupting agents cytochalasin D or latrunculin markedly enhanced ClC-2-mediated Cl− channel activity, whereas jasplakinolide, an actin-stabilizing peptide, was without significant effect (1). These data suggest that ClC-2 is functionally linked to the actin cytoskeleton and that this linkage normally has an inhibitory effect on ClC-2 channel activity. Using a combination of actin overlay and cosedimentation assays, actin was shown to bind directly to the NH2 terminus of ClC-2 via electrostatic interactions (1).
In addition to being linked to the actin-based cytoskeleton through spectrin, ENaC interacts directly with actin. Biophysical data suggesting an interaction of ENaC with actin were first presented by Cantiello et al. (16). Using patch-clamp analysis, they observed an increase in amiloride-sensitive sodium channel activity in cell-attached patches from X. laevis A6 cells within 2–5 min of application of cytochalasin D. In addition, they demonstrated that the addition of short F-actin filaments to excised patches activated an amiloride-sensitive 9-pS sodium channel (16). These effects were reversed by the addition of the G-actin-binding protein DNase I. Subsequent to the cloning of the α-, β-, and γ-subunits of ENaC, Berdiev et al. (9) demonstrated that short actin filaments caused a twofold decrease in unitary conductance and a twofold increase in Po of the cloned rat (r) αβγ-ENaC subunits reconstituted into planar lipid bilayers. αβγ-rENaC was found to be activated by PKA in the presence, but not in the absence, of actin, and DNase I prevented the activation of αβγ-rENaC by PKA and actin. Interestingly, gelsolin, a protein that stabilizes the polymerization of actin into short oligomers, evoked a sustained activation of rENaC at actin/gelsolin ratios of 32:1, with a maximal effect at an actin/gelsolin ratio of 2:1. These data indicate that ENaC activity is regulated through an interaction with short actin filaments (8, 15, 87). Because phosphorylated actin is a poor substrate for polymerization, PKA-mediated phosphorylation of actin augments channel activation by decreasing the rate of actin filament elongation (9). Figure 2 presents a hypothetical model to explain the role of short actin filaments and PKA in the activation of ENaC.
Jovov et al. (46) demonstrated that actin did not activate a COOH terminally truncated α-rENaC subunit (αR613X) construct when it was coexpressed with wild-type βγ-subunits in X. laevis oocytes. Single-channel ENaC activity of the αR613Xβγ-rENaC construct exposed to F-actin did not differ from wild-type αβγ-rENaC that was not exposed to actin, thereby suggesting that the COOH terminus of α-ENaC participates in the actin-mediated regulation of ENaC activity (46). To further define the region of the COOH terminus participating in actin-mediated regulation of α-ENaC, Copeland et al. (19) made a series of truncation and deletion mutants from a chimeric rat-bovine α-ENaC (α-rbENaC) subunit. Although there was no effect of actin on an E631X truncation mutant of α-rbENaC when reconstituted into planar lipid bilayers, a C645X α-rbENaC truncation mutant was activated by the addition of short F-actin filaments (19). Deletion of amino acid residues E631-F644 from the α-rbENaC chimera (ΔE631-F644) nullified the effect of actin normally seen on the activity of the chimeric α-rbENaC construct (19). These data suggest that a 14-amino acid residue region between E631 and F644 represent an actin-binding domain in the COOH terminus of α-ENaC. Recently, we (63) have provided biochemical evidence for a direct interaction between F-actin and the COOH terminus of α-ENaC, thereby corroborating the biophysical data. Using an MDCK cell line stably expressing αβγ-rENaC (69), we were able to demonstrate an association between α-ENaC and F-actin via coimmunoprecipitation and by colocalization (63). As illustrated in Fig. 3, F-actin and ENaC were observed to colocalize at both the apical membrane domain and within the subapical cytoplasm. In addition, we were able to demonstrate that short F-actin filaments bind directly and specifically to the COOH terminus of α-ENaC in overlay and cosedimentation assays (63).
The colocalization of α-ENaC and F-actin in the subapical cytoplasm of MDCK cells stably expressing αβγ-rENaC (63) suggests that in addition to regulating ENaC activity, a direct interaction between F-actin and ENaC may function in the intracellular trafficking of ENaC from a subapical pool to the plasma membrane in response to stimulation by hormones such as vasopressin and insulin (11, 13, 14, 69, 99, 101, 121). The subapical ENaC is presumably composed of both recycled (13) and newly synthesized ENaC (100).
Verrey and co-workers (114) have demonstrated that treatment of A6 renal epithelial cells with cytochalasin D decreased both vasopressin-stimulated protein secretion (exocytotic movement) and vasopressin amiloride-sensitive Na+ transport, suggesting a role for actin in the apical vesicular trafficking of Na+ channels in response to vasopressin. Subsequently, Butterworth et al. (13) have shown that treatment of a mouse cortical collecting duct cell line (MPKCCD14), with latrunculin A, an actin-depolymerizing agent, inhibited the forskolin-mediated increase in amiloride-sensitive transepithelial Na+ transport, further demonstrating that actin is required for the trafficking of ENaC to the plasma membrane. However, it is unclear which step(s) in the intracellular trafficking of ENaC to the plasma membrane is affected by the actin-depolymerizing agents.
Trafficking of Aquaporin 2: A Paradigm for Our Understanding of the Role of Actin and Actin-Associated Proteins in the Intracellular Trafficking of Epithelial Ion Channels
The cAMP/vasopressin-induced trafficking of the aquaporin-2 (AQP2) water channel from an intracellular, recycling pool to the apical plasma membrane of the renal collecting ducts (12, 76, 77) serves as a paradigm for our understanding of the role actin and actin-associated proteins play in the intracellular trafficking of ENaC and other epithelial ion channels that undergo regulated insertion into the apical plasma membrane. Before the discovery of AQP2, the studies of Hays and colleagues (21, 25, 27, 40, 41), using the toad urinary bladder as a model system, demonstrated that vasopressin induces a transient depolymerization of the subapical membrane pool of actin that functions as a barrier to water channel fusion. This transient depolymerization of the subapical actin pool was proposed to be essential for facilitating the trafficking and fusion of water channel containing vesicles with the apical plasma membrane (21, 25, 27, 40, 41). Because cytochalasin D treatment of the toad urinary bladder did not increase water permeability across the epithelium, Hays and co-workers (40, 41) proposed that there are two pools of actin participating in the intracellular trafficking and fusion of water channels: the subapical pool, which functions as an apical barrier to vesicle fusion, and a second pool involved in the intracellular trafficking of water channels.
Recent studies have shown that actin depolymerization is essential for cAMP/vasopressin-mediated trafficking of AQP2 vesicles to the apical membrane and that inhibition of Rho GTPase promotes actin depolymerization. Klussmann et al. (49) have revealed that the vasopressin-induced translocation of AQP2 to the apical plasma membrane of inner medullary collecting duct cells is stimulated by the inhibition of RhoA, whereas it is inhibited by the expression of a constitutively active RhoA that induces the formation of actin filaments. Tamma et al. (107) have also shown that PKA-mediated phosphorylation of RhoA, which stabilizes RhoA in the inactive form, is essential for actin depolymerization and apical fusion of the AQP2 vesicles. Tajika and colleagues (106) have shown in MDCK cells stably expressing AQP2 that recycling AQP2 moves from early endosomes to a subapical Rab11-positive storage compartment. Treatment with either cytochalasin D or latrunculin B induced the accumulation of AQP2 in early endosomes, revealing that actin filaments are required for the transfer of AQP2 from early endosomes to Rab11-positive subapical storage vesicles (106).
The signal-induced proliferation-associated gene 1 (SPA1) binds to the COOH terminus of AQP2 (73). SPA1 interacts with α-actinin (38) and is a specific GTPase-activating protein (GAP) for the Ras-related small GTPase Rap1 (51, 110). Rap1, whose expression is upregulated in the rat IMCD in response to vasopressin (84), is involved in cytoskeletal reorganization via cross talk with other small GTPases, including Ral and Rho (39). The active form of Rap1 (Rap1GTP) binds to Ral guanine nucleotide exchange factors (RalGEFs) that activate RalGTPase. RalBP1, one of the Ral GTP-binding proteins (76, 123), is a GAP for Rho family GTPases. Noda et al. (73) have shown that expression of either a SPA-1 mutant lacking GAP activity or a constitutively active Rap1 mutant in a MDCK cell line stably expressing AQP2 blocks the apical delivery of AQP2, whereas a dominant negative Rap1 mutant that suppresses endogenous Rap1 activity increases AQP2 trafficking to the apical plasma membrane. It has been hypothesized that Rap1 modulates the assembly of the subapical F-actin cytoskeleton that functions as a barrier for AQP2 trafficking to the apical plasma membrane through cross talk with Rho GTPases (76). However, as noted by Noda and Sasaki (76), manipulation of the cytoskeleton by pharmacological modulating agents or though the overexpression of dominant negative constructs inevitably alters the overall cell structure and makes experiments examining the role of actin in the intracellular trafficking of AQP2 or ion channels difficult to interpret. Moreover, physiological stimuli may alter actin dynamics in a restricted area around the intracellular vesicles.
Noda et al. (75) have demonstrated that the COOH terminus of AQP2 binds directly to actin. They were able to identify the β- and γ-isoforms of actin as proteins interacting with AQP2 by matrix-assisted laser desorption-ionization time-of-flight (MALDI-TOF) mass spectrometric analysis of eluates from an AQP2 immunoaffinity chromatography column to which rat renal papillary lysate had been applied. Through the use of surface plasmon resonance (SPR), they were able to demonstrate that the COOH terminus of AQP2 binds directly to actin (75). In addition to actin, Noda and co-workers (74) have recently identified a multiprotein “motor” complex associated with AQP2 by MALDI-TOF analysis. This “motor protein” complex consists of ionized calcium-binding adapter molecule 2, myosin-regulatory light chain smooth muscle isoforms II-A and II-B, α-tropomyosin 5b, annexin A2 and A6, scinderin, gelsolin, α-actinin, α-II spectrin, and myosin heavy chain nonmuscle type A. Phosphorylation of myosin-regulatory light chain by myosin light chain kinase is increased in response to vasopressin (18). Phosphorylation of myosin light chain results in an enhancement of myosin-actin filament interactions and the formation of actin filaments (76). Because all of the proteins identified within the motor protein complex have the ability to bind directly to actin, Noda et al. (74) proposed that these proteins may indirectly associate with AQP2 through actin and drive the anterograde movement of the AQP2 vesicles. Furthermore, it was postulated that the interactions between these proteins are dynamic and serve key points in the regulation of AQP2 trafficking (74).
The proteins forming the motor protein/actin-regulatory protein complex that facilitate in the intracellular trafficking of ENaC from a subapical compartment to the plasma membrane in response to hormonal stimulation have not been identified. However, the fact that both AQP2 and ENaC are directly linked to actin and actin links the motor protein/actin-regulatory protein complex to AQP2 raises the possibility that ENaC, and perhaps other epithelial ions that interact with actin and undergo regulated insertion into the plasma membrane, has a similar motor protein/actin-regulatory protein complex associated with it. Based on our current understanding of the vasopressin-mediated intracellular trafficking of AQP2, we present a hypothetical model in Fig. 4 to explain the role of the actin-based cytoskeleton in the intracellular trafficking of ENaC in response to hormonal stimuli.
In this review, we have attempted to present an overview of our current understanding of the interactions between epithelial ion channels and the actin-based cytoskeleton and the roles these interactions play in regulating channel localization, activity, and intracellular trafficking. Although significant progress has been made in elucidating the molecular interactions between epithelial ion channels and actin or actin-binding proteins, our understanding of the mechanisms regulating these interactions and their physiological roles is only beginning to emerge. Clearly, one of the challenges that lies ahead is to establish the functional relationship between epithelial ion channels, their associated actin or actin-binding/scaffolding proteins, and the signaling pathways that are involved in the physiological regulation of these channels.
This work was supported by National Institute of Diabetes and Digestive and Kidney Diseases Grant DK-37206 to D. J. Benos and by American Heart Association Grant-in-Aid 0555333B to P. R. Smith.
Present address of C. Mazzochi: Dept. of Physiology, David Geffen School of Medicine at UCLA, 53–231 CHS, Box 951751, Los Angeles, CA 90095-1751.
- Copyright © 2006 the American Physiological Society