Regulating ENaC’s gate

Abstract

Epithelial Na⁺ channels (ENaCs) are members of a family of cation channels that function as sensors of the extracellular environment. ENaCs are activated by specific proteases in the biosynthetic pathway and at the cell surface and remove embedded inhibitory tracts, which allows channels to transition to higher open-probability states. Resolved structures of ENaC and an acid-sensing ion channel revealed highly organized extracellular regions. Within the periphery of ENaC subunits are unique domains formed by antiparallel β-strands containing the inhibitory tracts and protease cleavage sites. ENaCs are inhibited by Na⁺ binding to specific extracellular site(s), which promotes channel transition to a lower open-probability state. Specific inositol phospholipids and channel modification by Cys-palmitoylation enhance channel open probability. How these regulatory factors interact in a concerted manner to influence channel open probability is an important question that has not been resolved. These various factors are reviewed, and the impact of specific factors on human disorders is discussed.

INTRODUCTION

Epithelial Na⁺ channels (ENaCs) were initially identified as a key pathway for Na⁺ transit across the apical membrane of high-resistance, Na⁺-transporting epithelia (57, 89). These channels were subsequently found to be expressed in a variety of epithelial and nonepithelial tissues (89, 172). In the kidney, ENaCs are expressed in the aldosterone-sensitive distal nephron (ASDN) and have a key role in fine-tuning Na⁺ absorption from the ultrafiltrate, serving as a final pathway for Na⁺ absorption in the nephron, where key hormones involved in the regulation of extracellular fluid volume and blood pressure exert their effects.

In addition to its role in transepithelial Na⁺ transport, electrogenic Na⁺ absorption via ENaC is coupled to K⁺ secretion in the ASDN, which is mediated by several apical K⁺ secretory channels. These include the inwardly rectifying K⁺ channel Kir1.1 [or renal outer medullary K⁺ channel (ROMK)] and the large-conductance Ca²⁺-activated K⁺ (BK) channel (36). Early inhibitors of ENaC, amiloride and triamterene, were developed as K⁺-sparing diuretics (57, 88). In the distal nephron, Na⁺ is absorbed with Cl⁻ via the thiazide-sensitive Na⁺-Cl⁻ cotransporter (NCC) in the distal convoluted tubule, or Na⁺ is absorbed via ENaC in exchange for K⁺ or in parallel with Cl⁻ in later segments of the nephron. Recent studies have highlighted the role of plasma K⁺ concentration ([K⁺]) in regulation of NCC, as well as ENaC and K⁺ secretory channels (60, 160). When plasma [K⁺] is low, NCC is activated and ENaC is suppressed, favoring Na⁺ and Cl⁻ absorption. When plasma [K⁺] is high, NCC is suppressed and ENaC is activated, favoring Na⁺ and K⁺ secretion.

Monogenetic mutations of ENaC in individuals with inherited forms of hypertension associated with hypokalemia (Liddle syndrome) or hypotension associated with hyperkalemia (pseudohypoaldosteronism type 1) underscore the importance of ENaC in controlling blood pressure as well as blood K⁺ (62, 143). Mutations associated with Liddle syndrome disrupt a Pro-Tyr (PY) motif in the COOH terminus of the β- or γ-subunit of ENaC, which impairs interaction of the channel with the ubiquitin ligase Nedd4-2, resulting in an increase in expression of channels at the plasma membrane as well as an increase in channel open probability (1, 49, 90).

It is likely that a subtle dysregulation of ENaC contributes to essential hypertension. A few specific ENaC variants have been associated with hypertension in specific populations, although they have not been associated with a gain of function when examined in heterologous expression systems (7, 16, 120, 151, 155). A gain-of-function ENaC variant was recently described in siblings with a Liddle syndrome phenotype (142). This variant is in the extracellular region of the α-subunit and is one of a growing number of variants in the extracellular regions of ENaC subunits that exhibit a gain-of-function phenotype (38, 136, 137). Whether other gain-of-function ENaC variants are associated with hypertension remains to be shown.

In addition to its role in the kidney in the regulated reabsorption of filtered Na⁺ and facilitation of K⁺ secretion, ENaCs are expressed at other sites that influence blood pressure. For example, the channel is expressed in lingual epithelia, where it has a role in salt taste (37, 96), and in the distal colon, where it absorbs ingested Na⁺ (30, 57, 89, 180). ENaCs are expressed in endothelia and vascular smooth muscle (14, 76, 91, 118, 154, 172). Its role at these sites in regulating vascular tone and blood pressure is still being explored. ENaCs are also expressed in antigen-presenting cells, where increases in extracellular Na⁺ activate an ENaC-dependent signaling cascade, resulting in release of proinflammatory cytokines that increase blood pressure (19, 118, 166).

ENaCs are also expressed in airway and alveolar epithelium, where they have important roles in regulating airway and alveolar surface liquid volumes (72, 106, 158). For example, enhanced ENaC activity in the airway may lead to reduced airway surface liquid volume, impaired mucociliary clearance, and inflammation (102). ENaC activity appears to be enhanced in cystic fibrosis (CF) and has been suggested to contribute to CF pathogenesis (45, 69). However, the role of ENaC in CF pathogenesis has not been settled (40, 74), and ENaC inhibitors have not proved to be beneficial in the treatment of CF (113, 144).

ENaC STRUCTURE

Studies following the initial cloning of ENaC and other members of the ENaC/degenerin family revealed a similar subunit structure: short cytoplasmic NH₂ and COOH termini and two membrane-spanning domains connected by a large extracellular loop (30, 31, 150). The resolved structure of an acid-sensing ion channel (ASIC), a member of the ENaC/degenerin family, provided important insights regarding the structural organization of ASICs and other members of the ENaC/degenerin family (15, 75, 177). These features were confirmed with the recently resolved structure of ENaC (121). ENaCs are heterotrimers composed of structurally related subunits, referred to as α, β, and γ, whereas ASICs are homo- or heterooligomeric trimers. A δ-subunit is expressed in specific tissues, substituting for the α-subunit to form δβγ, with properties that differ from αβγ-subunit channels (58). δ-Subunits are not expressed in rodents (58, 62). Resolved structures revealed highly organized extracellular regions composed of discrete domains resembling a hand holding a ball (Fig. 1). Two of these domains, referred to as the palm and β-ball, are formed by β-strands and are in close proximity to the lipid bilayer. More peripheral domains, referred to as the thumb, finger, and knuckle, are formed primarily by α-helices.

Fig. 1.

Acid-sensing ion channel (ASIC) type 1 and epithelial Na⁺ channel (ENaC) structures. A, left: ribbon illustration of the ASIC1 homotrimer, highlighting the extracellular and transmembrane regions of 1 subunit. Discrete domains within the extracellular region include the proximal palm and β-ball formed by β-strands and the peripheral α-helical thumb, finger, and knuckle domains. Transmembrane helices are indicated in red. A, right: organization of an ASIC1 subunit. Contiguous peripheral domains with α-helices (cylinders) arise from noncontiguous proximal domains with β-strands (arrows). TM, transmembrane. B: ribbon illustration of a human ENaC heterotrimer. Residues in the gating relief of inhibition by proteolysis (GRIP) domains that are unique to ENaC and contain the α- and γ-subunit-embedded inhibitory tracts are shown as surface rendering. α-Subunit (red), β-subunit (blue), and γ-subunit (green) are shown. Side (left) and top (right) views are shown. ASIC1 and ENaC structures are based on Protein Data Bank codes 4NYK and 6BQN (75, 121). [Top right and top left were modified from Kashlan and Kleyman (82), with permission.]

ENaC REGULATION

Most members of the ENaC/degenerin family are silent at baseline and activated by factors in the extracellular environment, including specific ions, peptides, or mechanical forces. On the other hand, ENaC is constitutively active, and its open probability is modified by extracellular factors, including ions, proteases, and mechanical forces (82, 83, 89). There is increasing evidence that these factors interact at specific sites within the extracellular regions of ENaC/degenerin family members, resulting in structural transitions that alter the conformation of the channel gate within the transmembrane pore and, in turn, changes in channel open probability. Within the extracellular region, the least-conserved domains are those containing α-helices, particularly the finger domain (78, 82). We and others have speculated that these poorly conserved regions have key roles in conferring specificity with regard to the factors that regulate distinct members of the ENaC/degenerin family. Perhaps this is best highlighted by the selective activation of ENaC by proteases that target unique regions in the extracellular domains of the α- and γ-subunits (78, 82, 121) (see below).

Functional ENaC expression is largely regulated by altering the number of channels at the plasma membrane and/or by altering open probability. As mentioned above, Liddle syndrome mutations affect both channel density at the cell surface and open probability. Aldosterone increases Na⁺ transport by increasing transcription and translation of specific ENaC subunits and by stabilizing channels at the plasma membrane. Frindt and Palmer (50) showed that this is indeed true in isolated rat tubules, but the increase in subunit density accounts for <25% of the increase in transtubular Na⁺ current, implying that the remaining 75% is likely due to an increase in single-channel open probability. Single-channel recordings in an amphibian principal cell culture model showed that acute application of aldosterone dramatically increases single-channel open probability (84). Aldosterone also enhances expression of proteolytically processed channels at the plasma membrane (see below). There have been extensive reviews on the regulation of ENaC surface density (25, 46, 124, 153). The remainder of our review examines specific factors that regulate ENaC open probability, focusing on αβγ-subunit channels. Several of these factors regulate ENaC open probability by interacting at sites within the extracellular regions of ENaC (see below).

Regulation by Proteases

The observation that Na⁺ transport across toad urinary bladder is reduced by the serine protease inhibitor aprotinin (122) provided the first hint that ENaCs are regulated by proteases. Vallet, Rossier, and colleagues (165) subsequently showed that ENaC is activated by the protease trypsin. This group identified prostasin as a channel-activating serine protease (165). They and others went on to identify a series of serine proteases and metalloproteases that can activate ENaC (63, 167). While it was initially unclear whether ENaC itself was the target of proteases, Masilamani, Knepper, and colleagues provided the first clue that an ENaC subunit (γ) was cleaved (105). Using ENaC subunits with NH₂- and COOH-terminal epitope tags, Hughey, Kleyman, and co-workers showed that the α- and γ-subunits of ENaC were cleaved at defined sites within these subunits by furin, a trans-Golgi resident member of the proprotein family of serine proteases (70, 71). Introduction of mutations at key cleavage sites prevented both ENaC proteolytic processing and channel activation. The use of cell lines that lacked furin expression and selective furin inhibitors provided further evidence regarding the role of furin in cleaving and activating ENaC.

Studies using heterologous expression systems have clearly shown that proteases have a role in activating ENaC. Evidence from studies with in vivo systems is not as clear. These studies have focused on administration of serine protease inhibitors to rodents or have used mice where selected proteases have been knocked out. For example, administration of aprotinin, a nonselective serine protease inhibitor, to mice led to a natriuresis (21) and is consistent with protease-dependent activation of ENaC in vivo. Prostasin knockout mice exhibit abnormal skin development and early mortality (94) and, therefore, are not a useful model to study the role of prostasin in regulating ENaC in specific tissues. Selective knockout of prostasin in alveolae was associated with reduced fluid clearance (129), and a colonic knockout led to a moderate reduction in colonic potential difference (103). However, a kidney tubule-specific prostasin knockout has not yet been described. While kallikrein activates ENaC in heterologous expression systems, kallikrein knockout mice exhibited enhanced Na⁺ absorption in cortical collecting ducts (48). The increased Na⁺ transport may reflect an electroneutral process, as transepithelial voltage was unchanged. One concern is that multiple proteases might activate ENaC in specific settings and in specific cell types. Simply blocking one protease might not be sufficient to reduce ENaC activity in vivo (89).

A related question is whether proteolysis of ENaC subunits correlates with its activation in vivo. Frindt, Palmer, and co-workers recently found a discrepancy between the appearance of cleaved ENaC subunits (early response) and the increase in functional ENaC expression (late response) when rats were fed a low-salt diet (52). While cleaved subunits appeared early after initiation of the low-salt diet, the increase in channel activity was delayed by days. Inasmuch as a variety of factors influence ENaC open probability and/or channel trafficking, this observation is not surprising. Some of the factors that regulate ENaC open probability, in addition to proteolysis, are discussed below.

How do proteases activate the channel? A large body of evidence suggests that proteases activate ENaC by removing embedded inhibitory tracts in the α- and γ-subunits, transitioning channels to higher open-probability states (for reviews see Refs. 82, 87, 89, 141). The α-subunit is cleaved twice by furin. Channels lacking cleavage sites have a very low open probability and have been referred to as near-silent channels (29, 70, 146). Cleavage at both α-subunit furin sites is required to activate the channel, releasing a 26-residue embedded inhibitory tract and transitioning channels to a moderate-activity state (35, 146). Channels with a mutant α-subunit lacking this 26-residue inhibitory tract, as well as the α-subunit furin sites, are functional, although the α-subunit is not cleaved, demonstrating that release of the inhibitory tract, rather than cleavage per se, activates the channel (35). A peptide corresponding to the 26-residue tract inhibits ENaC, and within this 26-residue tract is a key 8-residue (LPHPLQRL) tract that retains inhibitory activity (34).

This paradigm for ENaC activation by proteases is also relevant to the γ-subunit. However, furin cleaves the γ-subunit only once (70). A growing number of proteases cleave the γ-subunit distal to the furin site, releasing an embedded inhibitory tract of >40 residues and transitioning channels to a high open-probability state (22). These proteases include prostasin, transmembrane serine protease 4, matriptase, cathepsins B and S, neutrophil and pancreatic elastase, kallikrein, meprin, urokinase, plasmin, and specific bacterial proteases (2, 5, 22, 26–28, 55, 56, 61, 63, 77, 126–128, 156, 159, 165, 167). Several proteases, including prostasin and kallikrein, appear to target a polybasic RKRK sequence distal to the furin site. In combination with furin, these proteases release a 43-residue tract (22, 126, 128). Within this 43-residue tract is an 11-residue (RFLNLIPLLVF) tract that retains inhibitory activity (125). Other proteases, such as plasmin and neutrophil and pancreatic elastase, cleave the γ-subunit at sites that are just distal to the RKRK tract (2, 127). Channels with a mutant γ-subunit lacking the 43-residue inhibitory tract as well as γ-subunit furin site have a high open probability, although the mutant γ-subunit is not cleaved (22).

Both the α- and γ-subunits have embedded inhibitory tracts that, when released, result in channel activation. Using a combination of wild-type and mutant ENaC subunits where the inhibitory tracts were either retained within a subunit or deleted, Carattino, Hughey, and Kleyman showed that the loss of the γ-subunit inhibitory tract has a dominant role in channel activation (32). They suggested that proteolytic processing and associated channel activation constitute a stepwise process (89) (Fig. 2). Noncleaved channels, in general, have a low open probability. Channels that have been processed by furin, where the α-subunit has lost its inhibitory tract and the γ-subunit has been cleaved once, have, in general, an intermediate open probability. Channels that have been processed by furin and a second protease, where the α- and γ-subunits have lost their inhibitory tracts, have, in general, a high open probability. An important caveat is that other important factors, in addition to proteases, influence channel open probability. Several key factors are discussed below. How these factors interact to modulate channel open probability requires further examination.

Fig. 2.

Epithelial Na⁺ channel (ENaC) activation by proteases. ENaC subunits assemble in the endoplasmic reticulum. As channels transit through the biosynthetic pathway, the α- and γ-subunits are processed by furin, releasing the α-subunit inhibitory tract. This favors transition of channels from a low to an intermediate open probability (P_o). The γ-subunit is cleaved once by furin. Cleavage of this subunit by a second protease results in release of its inhibitory tract and favors transition of channels to a high open probability. There are other important factors that influence channel open probability. [Modified from Ray et al. (139), with permission from Elsevier.]

Several studies have reported COOH-terminal α- or γ-subunit cleavage fragments smaller than fragments generated by proteolytic processing associated with release of inhibitory tracts and channel activation. As discussed above, subunit cleavage is not necessary for channel activation (55, 119, 140, 179). We are not aware of studies demonstrating that ENaC subunit proteolysis resulting in smaller COOH-terminal α- or γ-subunit cleavage fragments modifies channel activity or is associated with the release of an inhibitory tract.

The α- and γ-subunit inhibitory tracts are present in the peripheral finger domains of these subunits and represent inserts that are not present in other members of the ENaC/degenerin family. For example, the α-subunit inhibitory tract is within a unique 72-residue insert that is not present in ASIC1 (75, 78). While the resolved structure of ASIC1 provided important insights into the organization of the extracellular regions of members of this ion channel family, it was unable to inform structural details regarding the inhibitory tracts in ENaC. Kashlan, Kleyman, and co-workers assessed the effects of an α-subunit inhibitory peptide on a large number of α-subunit mutants and, in combination with cross-linking studies, suggested that the inhibitory tract resides at a thumb-and-finger domain interface (78, 79, 81, 82). These investigators hypothesize that this interface is dynamic and that the inhibitory tract reduces channel open probability by stabilizing the interface (79). Subsequent work by Balchak, Thompson, Kashlan, and co-workers suggested that the γ-subunit inhibitory tract reduces channel activity by a similar mechanism (17). Noreng, Baconguis, and co-workers recently resolved the structure of the ENaC αβγ-subunit by cryoelectron microscopy (121). The structure confirmed that the inhibitory tract, which they referred to as the “gating relief of inhibition by proteolysis” (or GRIP) domain, is located at a thumb-and-finger domain interface. The domain is formed by antiparallel β-strands with accessible protease cleavage sites (Fig. 1).

ENaC subunit proteolysis has been observed in states of extracellular volume depletion and decreased effective arterial volume, such as heart failure, and with aldosterone administration in volume-replete states (50, 51, 183). As furin cleaves the α-subunit twice, α-subunit fragments with a size consistent with furin processing likely reflect release of the α-subunit inhibitory tract. However, an increase in γ-subunit cleavage does not necessarily reflect release of the γ-subunit inhibitory tract. Several lines of evidence, including use of an antibody against the γ-subunit inhibitory tract, suggest that this inhibitory tract is released from ENaCs expressed in kidneys in the settings of volume depletion or aldosterone administration (33, 156, 163, 179).

ENaC also appears to be activated by urinary proteases in nephrotic syndrome (21, 127, 138, 139, 156). In this setting, damaged glomeruli allow plasminogen to be filtered, and tubular urokinase converts this filtered plasminogen to activate protease plasmin. Plasmin can directly cleave the γ-subunit, facilitating the release of its inhibitory tract and activating the channel (127, 156). Plasmin may also influence ENaC activity by interacting with other proteases, such as prostasin (157). Other urinary proteases may also have a role in activating ENaC in nephrotic syndrome (77, 92). By cleaving the γ-subunit and activating ENaC, filtered proteases such as plasmin may contribute to urinary Na⁺ retention in nephrotic syndrome. Urinary plasminogen and plasmin have been found in a growing number of disease processes associated with glomerular proteinuria (8, 10, 23, 24, 138). The role of ENaC-mediated Na⁺ absorption in humans with proteinuria is still unclear. Few trials or case reports have examined the efficacy of amiloride in humans with proteinuria, and results have been inconsistent, with hyperkalemia as a complicating factor (9, 68, 123, 164, 176).

Regulation by Extracellular Na⁺

ENaC is a Na⁺- and Li⁺-selective channel. In addition to transporting Na⁺, the channel is inhibited by extracellular Na⁺, a process that has been termed Na⁺ self-inhibition and was first described more than 40 years ago (53). A growing body of evidence suggests that extracellular Na⁺ interacts at site(s) within the extracellular regions of ENaC subunits, driving allosteric changes that are transmitted to the channel gate to reduce channel open probability (20, 80, 81, 146). We have suggested that this regulatory response allows ENaCs in the ASDN to alter rates of Na⁺ influx in response to changes in urinary Na⁺ concentration (89).

The Na⁺ self-inhibition response is tightly linked to the extent of ENaC processing by proteases and is an excellent example of the interplay between different factors that regulate ENaC gating (20, 41, 146). Noncleaved channels exhibit a markedly enhanced Na⁺ self-inhibition response, reflecting the fact that the channels are largely closed when extracellular Na⁺ concentration is high (>100 mM). However, in the presence of low extracellular Na⁺ concentration, the channels are active, with an intermediate open probability reflecting relief from the inhibitory effect of Na⁺ (146). Channels that have been processed by furin exhibit a less-robust Na⁺ self-inhibition response and have an intermediate open probability in the presence of a high Na⁺ concentration, while channels processed by furin and a second protease, such as prostasin, have lost the Na⁺ self-inhibition response and have a notably high open probability in the presence of a high Na⁺ concentration (20, 22, 41, 146).

A large number of sites in the different domains of the extracellular regions of the α- and γ-subunits where amino acid substitutions alter the Na⁺ self-inhibition response and channel activity have been identified (38, 39, 42, 43, 81, 100, 145–149, 175). These mutations are at sites that may directly participate in Na⁺ binding or affect allosteric transitions that occur following Na⁺ binding. It has been challenging to differentiate site(s) involved in Na⁺ binding from other sites that affect Na⁺ self-inhibition. In native channels, this inhibitory response has a clear cation preference of Na⁺ > Li⁺ >> K⁺ (20, 80). Kashlan, Kleyman, and co-workers identified sites on an extracellular loop connecting the β₆- and β₇-strands and in the vicinity of an α-subunit acidic cleft, where mutations affected the cation selectivity of the Na⁺ self-inhibition response, suggesting that residues in this region serve as an effector-binding site for Na⁺ (80).

There are an increasing number of nonsynonymous single-nucleotide variants (nsSNVs, or missense mutations) in the genes encoding the three (α, β, and γ) ENaC subunits expressed in the ASDN that are reported in publicly available databases. We and others have identified rare human ENaC nsSNVs that result in either a gain or a loss of function that was associated with changes in the Na⁺ self-inhibition response. For example, both αW593R and γL511Q are gain-of-function variants that showed an increase in open probability and a loss of Na⁺ self-inhibition (38, 136) (Fig. 3). A recent report described siblings with a Liddle syndrome phenotype and gain-of-function mutation in the α-subunit (C479R) that appeared to reflect an increase in channel open probability (142). It was previously reported that an alanine substitution at the equivalent site in the mouse α-subunit exhibited a reduced Na⁺ self-inhibition response (147). Further studies are needed to answer the following questions. Do humans with ENaC gain-of-function variants that have a loss of the Na⁺ self-inhibition response have an increased risk of hypertension? Do humans with ENaC loss-of-function variants that have an enhanced the Na⁺ self-inhibition response have a reduced risk of hypertension.

Fig. 3.

Epithelial Na⁺ channel (ENaC) Na⁺ self-inhibition response. Current traces illustrate the Na⁺ self-inhibition response assessed in Xenopus oocytes expressing wild-type (WT) human ENaC (left) or the γL511Q mutant (right). Oocytes were initially bathed in a solution containing 1 mM Na and 109 mM N-methyl-d-glucamine. After a rapid transition to a 110 mM Na bath, a rapid increase in inward Na⁺ current (I; downward deflection) was followed by a slow reduction in current, reflecting Na⁺ self-inhibition. The Na⁺ self-inhibition response in oocytes expressing the γL511Q mutant was markedly dampened. [Modified from Chen et al. (38), with permission.]

Regulation by Lipids

ENaC activity requires inositol lipid phosphates.

It has been known for a number of years that ENaC could be activated by application of phosphatidylinositol 4,5-bisphosphate (PIP₂) to the cytosolic surface of channels in excised, inside-out patches (99, 130, 132, 133, 162, 178). Ma, Saxena, and Warnock found that the regulation of ENaC by PIP₂ and phosphatidylinositol 3,4,5-trisphosphate (PIP₃) did not involve a change in surface expression of ENaC, nor did it involve ENaC trafficking (99); i.e., it was a result of a change in open probability. The implication of these electrophysiological experiments is that PIP₂ must interact with one or more of the ENaC subunits. In fact, an anti-PIP₂ antibody coimmunoprecipitates the β- and γ-subunits, but not the α-subunit (178). Other anionic lipids also seemed capable of activating ENaC, but to a lesser extent than PIP₂ (99). Subsequently, two binding sites for PIP₂ were identified at the NH₂ terminus of ENaC β- and γ-subunits (98), and one binding site for PIP₃ was identified at the NH₂ terminus of the ENaC γ-subunit (65). The same work also showed that all the early aldosterone-induced increase in Na⁺ transport was due to increases in inositol lipid phosphate binding. Other investigators described additional PIP₂ and PIP₃ binding sites to sites immediately following second-transmembrane domains (131–133, 152), but the NH₂-terminal sites appear to be the critical sites for PIP₂ regulation of ENaC.

Several other methods have been used to measure PIP₂ binding to ENaC. “PIP strips” are nitrocellulose membranes prespotted with different lipids (3). When the strips are exposed to GST fusion proteins with different β- and γ-subunit domains, the fusion proteins bind to specific lipids. This method avoids the problem of the binding of lipid to nonspecific proteins associated with the antibody or beads in coimmunoprecipitations. It also avoids the problem of vagaries in transfection or expression of constructs regulating lipid production. PIP strip overlay binding assays showed that the NH₂ terminus of β-subunits binds PIP₂ and PIP₃. In excised patches of apical membranes from Na⁺-transporting epithelial cells, both lipids activate ENaC (99, 178). Presumably, the β-subunit preferentially binds to PIP₂ and PIP₃, while the α- and γ-subunits do not because of differences within the NH₂-terminal domains of the three subunits. The β- and γ-subunits also bind strongly to phosphatidic acid, a degradation product of inositol phospholipids that was recently shown to inhibit ENaC activity (182). These results suggest that hydrophobicity and anionic-character membrane phospholipids are critical for binding and that the inositol phosphate head group is necessary for ENaC activation. Investigators have hypothesized that ENaC activity is regulated by interaction of the positive charges within the NH₂ terminus of ENaC with anionic phospholipids of the inner leaflet of the plasma membrane (65, 97, 133, 182).

PIP₂ and PIP₃ are localized in distinct nanoscale regions within the plasma membrane of cultured epithelial cells. Wang and Richards (171) used anti-phospholipid antibodies directly conjugated with Alexa Fluor 647 to show that PIP₂ and PIP₃ cluster in membrane domains of ~60 and ~130 nm, respectively. ENaC appears to be in lipid domains enriched in inositol lipid phosphates. Specifically, these are domains that contain high concentrations of inositol phospholipids and the enzymes that produce the lipids (95, 107, 161). Other investigators, using differential detergent solubility, also showed that some fraction of ENaC is present in PIP₂-rich domains (3, 4, 66, 67). They concluded that ENaC was inserted into the membrane from the Golgi already in fully formed PIP₂-rich domains but that the presence of ENaC in these domains was not necessary for function. However, while their evidence for the presence of ENaC in lipid domains is convincing, the conclusion that ENaC can function just as well in non-PIP₂-rich as in PIP₂-rich domain regions may not be correct. They concluded that if it were necessary for ENaC to be in these domains, then disrupting the domains should rapidly reduce ENaC function. Indeed, when they applied a domain-disrupting agent, methyl-β-cyclodextrin, (MβCD), Na⁺ current did decrease, but at a rate that they concluded was more consistent with disrupting the membrane insertion of rafts (with their embedded ENaC) than requiring the presence of ENaC in lipid domains for ENaC activity.

The problem with applying either MβCD or cholesterol to the apical surface of the cells is that lipids, including cholesterol in the outer membrane leaflet, are more ordered and, therefore, less accessible than in the less-ordered inner leaflet (73). Therefore, removal of cholesterol from the outer leaflet by MβCD may require high concentrations and may act more slowly, while removing cholesterol from the inner leaflet should be faster and occur at lower concentrations. In fact, in excised membrane patches, MβCD applied to the cytosolic surface of the membrane at a concentration 1,000 times lower than had been applied to the luminal surface (10–50 μM, instead of 10–50 mM) reduced ENaC channel activity to nearly zero in <5 min (181). The idea that disrupting the lipid domains eliminates all channel activity implies that only functional channels are in these domains.

MARCKS acts to maintain PIP₂ in lipid domains and to promote ENaC-PIP₂ interaction.

PIP₂ is necessary to open ENaC. However, there is a conceptual problem with a simple model of ENaC and PIP₂ associating by simple lateral diffusion in the membrane. ENaC is a relatively rare protein (only a few functional channels per μm² in the apical membrane). PIP₂ is also a rare molecule, constituting <1 in 1,000 membrane lipid molecules (73). Since PIP₂ association is necessary for ENaC activity, it is possible to estimate the likelihood of random diffusional interaction between ENaC and PIP₂ if we consider that PIP₂ follows a random walk to reach ENaC. PIP₂ moves by filling vacancies in the inner leaflet phospholipid structure; i.e., each step in the random walk is the cross-sectional size of a single PIP₂ molecule, ~0.55 nm² (73). The diffusion constant of PIP₂ measured by fluorescence correlation spectroscopy is ~4 × 10⁻¹² m²/s (101). For a typical principal cell, ENaC density will be ~7 channels/cm². With these values, the mean time between collision of PIP₂ and ENaC would be 6.3 × 10² s, or approximately once in 10 min. In fact, in principal cells, ENaC channels open every 1 or 2 s in a typical patch (18). The implication of these observations is that there must be a mechanism by which the local concentration of PIP₂ could be increased close enough to ENaC to account for the apparently anonymously high opening rate. This could be accomplished by a protein that is associated with apical membrane lipid domains and is capable of binding and sequestering PIP₂ with an affinity that would also allow PIP₂ to also bind and activate ENaC. One such protein is MARCKS (myristoylated alanine-rich C-kinase substrate) or its closely related isoform MLP-1 (MARCKS-like protein 1). Both MARCKS and MLP-1 consist of a myristoylated NH₂-terminal domain and an effector domain containing a large number of positively charged, basic amino acids. The positive charge within this domain electrostatically binds anionic lipids (e.g., PIP₂) and causes MARCKS and MLP-1 to associate with PIP₂-rich lipid domains (Fig. 4) (13, 54, 108, 110, 169, 170). The basic effector domain also contains protein kinase C (PKC) phosphorylation sites, cytoskeletal binding sites, and Ca²⁺/calmodulin-binding sites (64). MARCKS and MLP-1 reversibly associate with the inner leaflet of PIP₂-rich lipid domains through hydrophobic and electrostatic interactions of their myristoyl group and basic effector domain, respectively (86). MARCKS and MLP-1 cross-link to actin, but the binding is regulated by PKC phosphorylation and calmodulin binding. As the name MARCKS implies, PKC phosphorylates three of the serine residues within the basic effector domain of MARCKS or MLP-1, adding anionic charge to the otherwise-positive effector domain. This change in overall charge disrupts the interaction with membrane PIP₂ and leads to the translocation of MARCKS or MLP-1 from the membrane to the cytoplasm. Besides MARCKS and MLP-1, other members of the MARCKS family of proteins (growth-associated protein 43 and cardioactive peptide 23) also sequester PIPs (93), but only MARCKS or MLP-1 is present in Na⁺-transporting epithelial tissue (47, 173, 174). MARCKS has a predicted molecular mass of ~32 kDa but migrates slowly and close to 75 kDa on SDS-PAGE, probably due to its rod-shaped structure, unusual concentration of positive charge, and ability to bind SDS molecules. On SDS-PAGE, 26-kDa MLP-1 runs at 52 kDa.

Fig. 4.

Schematic diagram of myristoylated alanine-rich C-kinase substrate (MARCKS) and MARCKS-like protein 1 (MLP-1). The effector domain has multiple sites for phosphatidylinositol 4,5-bisphosphate (PIP₂) interaction (note positive-charged residues) and sites for phosphorylation (note serine residues). MARCKS and MLP-1 function as a reversible source of PIP₂ at the membrane. MLP-1 is the predominant MARCKS isoform in the mouse kidney. The ability of MARCKS and MLP-1 to function as a PIP₂-sequestering protein at the membrane is dependent on hydrophobic interactions between the myristoylation domain and the membrane, and electrostatic forces between the effector domain and anionic lipids in the membrane.

ENaC is associated with MARCKS.

To be most effective, MARCKS needs to associate with PIP₂-rich lipid domains, which it does through its myristoyl modification and association with PIP₂, as we and others have shown (3, 11–13, 44, 59, 85, 110, 117, 168, 169), but MARCKS should also interact directly with ENaC. In fact, it is possible to coimmunoprecipitate the β-subunit of ENaC with MARCKS antibody (178).

MARCKS/MLP-1 association with PIP₂-rich lipid domains is regulated.

Besides acting as a reversible source of PIP₂ at the cytosolic surface of the apical membrane, MARCKS and MLP-1 can be regulated by controlling their level of association with the apical membrane, since after translocation from the apical membrane to the cytosol, MARCKS and MLP-1 associate with PIP₂ (85, 86, 108, 109). Thus, MARCKS-mediated delivery of PIP₂ to ENaC with consequent ENaC activation can be regulated by controlling MARCKS translocation. This is significant, since it shows an entirely new mechanism for altering ENaC activity. PKC-mediated phosphorylation reduces MARCKS association with the membrane and, thereby, diminishes the ability of MARCKS to present PIP₂ to ENaC, with a concomitant reduction in ENaC activity; therefore, a reduction in PKC activity should increase ENaC activity by increasing open probability. The open probability of ENaC is higher in isolated, split-open tubules from PKCα knockout than wild-type mice. This increase in ENaC activity is associated with a substantial increase in the blood pressure of knockout mice (18), reinforcing the idea that MARCKS and MLP-1 are relevant physiologically.

Ca²⁺ might also neutralize the electrostatic interactions between MARCKS effector domain and PIP₂ and promote phosphorylation and translocation (Fig. 5). Ca²⁺ can also activate calmodulin, which is known to promote MARCKS translocation (11).

Fig. 5.

Interaction of myristoylated alanine-rich C-kinase substrate (MARCKS) with the plasma membrane. Dephosphorylated MARCKS binds to the plasma membrane and cross-links to actin. Phosphorylated MARCKS detaches from the plasma membrane and disrupts actin filaments. PIP₂, phosphatidylinositol 4,5-bisphosphate.

A model for MARCKS/MLP-1 regulation of ENaC and Na⁺ transport.

PIP₂-rich membrane domains are formed in the Golgi in association with another chaperone protein, MAL/VIP17 (6, 104, 111, 112, 134, 135), and transit to the plasma membrane. To be open, ENaC must associate with an inositol lipid phosphate (e.g., PIP₂). This interaction is facilitated by localizing ENaC in a PIP₂-rich, cholesterol-containing apical membrane domain. Both PIP₂ and ENaC can be stabilized in these domains by several chaperone proteins, including MARCKS or its closely related isoform MLP-1, both of which associate with the inner lipid leaflet. Under appropriate conditions (e.g., PKC activation), MARCKS/MLP-1 is phosphorylated and can move from association with the membrane lipid domain into the cytosol and then no longer stabilizes the PIP₂-ENaC complex, allowing PIP₂ to be hydrolyzed and ENaC to leave the specialized lipid domains. In the absence of PIP₂ association, ENaC can be ubiquitinated and internalized.

ENaC is activated by Cys-palmitoylation.

Cys-palmitoylation is a reversible attachment of palmitate to cytoplasmic cysteine residues on proteins. Posttranslational modification by palmitoylation is another mechanism by which lipid molecules activate ENaC (114–116). The β- and γ-subunits of ENaC are palmitoylated at specific cytoplasmic cysteine residues (βCys⁴³, βCys⁵⁵⁷, γCys³³, and γCys⁴¹). Mutating these cysteine residues prevented palmitoylation and dramatically reduced channel open probability, while surface expression and proteolytic processing of channel subunits were unaffected (114, 115). Preventing γ-subunit palmitoylation had a more dominant role in inhibiting ENaC than preventing β-subunit palmitoylation (115). Five of the 23 known palmitoyltransferases (referred to as DHHCs) activate ENaC when coexpressed in Xenopus oocytes (116). The functional roles of palmitoyltransferases in regulating ENaC in vivo are unclear. The specific palmitoyltransferases that regulate ENaCs in principal cells in the kidney in vivo or in other cells are unknown. As multiple palmitoyltransferases could have a role in regulation of ENaC in a specific cell, knocking out a single DHHC may not be sufficient to prevent or reduce ENaC palmitoylation. Furthermore, specific DHHCs could modify other proteins that influence ENaC.

How do subunit palmitoylation and subunit interactions with PIP₂ lead to an increase in channel open probability? We have proposed that palmitoylation or PIP₂ binding facilitates interactions between cytoplasmic domains and the plasma membrane, resulting in conformational changes that are transmitted to the transmembrane domains and channel gating (89, 114, 115). However, the conformational changes associated with ENaC subunit palmitoylation or subunit interactions with PIP₂ have not been defined. As the resolved structure of ENaC did not include cytoplasmic domains and constructs with truncated cytoplasmic domains were used to generate these structures, the detailed structural information needed to address these questions is lacking.

SUMMARY

A variety of extracellular and intracellular factors regulate ENaC open probability. Our overview of several important ENaC regulators provides insights into the complexity of this process. The vast majority of studies have examined these regulatory factors in isolation. How they work in concert to modulate ENaC gating is still unclear, as is an understanding of conformational changes that ultimately impinge on the channel’s gate. Our review has focused on the regulation of αβγ-subunit channels, with studies largely performed using heterologous expression systems and established cell lines. How these factors influence channels with differing subunit compositions and how they affect ENaCs expressed in different tissues are questions that continue to be addressed.

GRANTS

This work was supported by National Institutes of Health Grants HL-147818, DK-038470, and DK-079307 (to T. R. Kleyman) and DK-110409 (to D. C. Eaton).

DISCLOSURES

T. R. Kleyman receives an honorarium from Wiley, Inc., as Editor-in-Chief of Physiological Reports. D. C. Eaton has no conflicts of interest, financial or otherwise, to disclose.

AUTHOR CONTRIBUTIONS

T.R.K. and D.C.E. drafted manuscript; T.R.K. and D.C.E. edited and revised manuscript; T.R.K. and D.C.E. approved final version of manuscript.