Epithelial Sodium Channel/Degenerin Family of Ion Channels: A Variety of Functions for a Shared Structure

Stephan Kellenberger, Laurent Schild


The recently discovered epithelial sodium channel (ENaC)/degenerin (DEG) gene family encodes sodium channels involved in various cell functions in metazoans. Subfamilies found in invertebrates or mammals are functionally distinct. The degenerins in Caenorhabditis elegansparticipate in mechanotransduction in neuronal cells, FaNaC in snails is a ligand-gated channel activated by neuropeptides, and theDrosophila subfamily is expressed in gonads and neurons. In mammals, ENaC mediates Na+ transport in epithelia and is essential for sodium homeostasis. The ASIC genes encode proton-gated cation channels in both the central and peripheral nervous system that could be involved in pain transduction. This review summarizes the physiological roles of the different channels belonging to this family, their biophysical and pharmacological characteristics, and the emerging knowledge of their molecular structure. Although functionally different, the ENaC/DEG family members share functional domains that are involved in the control of channel activity and in the formation of the pore. The functional heterogeneity among the members of the ENaC/DEG channel family provides a unique opportunity to address the molecular basis of basic channel functions such as activation by ligands, mechanotransduction, ionic selectivity, or block by pharmacological ligands.


The epithelial sodium channel (ENaC)/degenerin gene family represents a new class of ion channels that was discovered at the beginning of the 1990s. The degenerin genes deg-1 andmec-4 were first identified from a genetic screen of the mechanosensory pathway of Caenorhabditis elegans. The name degenerin (DEG) comes from the cellular phenotype induced by mutations of the deg-1 gene and other related genes that result in selective degeneration of sensory neurons involved in touch sensitivity. At the time of the identification of the DEG gene family in C. elegans, a functional cloning strategy inXenopus laevis oocytes allowed the isolation and sequencing of a cDNA encoding the α-subunit of the amiloride-sensitive epithelial Na+ channel ENaC (found in databanks under SCA, “sodium channel, amiloride sensitive,” and SCNN1, “sodium channel, nonneuronal”). This channel was already known to play a crucial role in Na+ absorption in the distal part of the kidney tubule and to be the target of aldosterone action. ENaC and degenerins were found to have substantial sequence homology. Additional members that formed a new subfamily within this emerging ion channel family were subsequently identified by sequence homology and characterized by functional expression. These related genes were found to be expressed mainly in the central and peripheral nervous system and were called mammalian degenerins (MDEG) or brain Na+channels (BNaC, BNC). After the discovery of their activation by extracellular protons these channels were named acid-sensing ion channels (ASICs). At about the same time the FMRF-amide-gated ion channel (FaNaC) was cloned from the mollusk Helix aspersa. This channel forms its own subfamily within the ENaC/DEG family of ion channels.

In contrast to channels that appeared at an early stage of evolution such as potassium, chloride, or water channels, ENaC/DEG genes are present only in animals (metazoa) with specialized organ functions for reproduction, digestion, and coordination. The members of the ENaC/DEG gene family show a high degree of functional heterogeneity that is unusual among the known gene families of ion channels. Their wide tissue distribution that includes transporting epithelia as well as neuronal excitable tissues best reflects the functional heterogeneity of the ENaC/DEG family members. Depending on their function in the cell, these channels are either constitutively active like ENaC or activated by mechanical stimuli as postulated for C. elegansdegenerins, or by ligands such as peptides or protons in the case of FaNaC and ASICs. The evolution of the ENaC/DEG gene family certainly followed quite divergent paths to finally achieve a variety of different functions in the cell.

This review summarizes our present knowledge of the fundamental roles of the ENaC/DEG proteins and the relationship between their biophysical properties and structural characteristics. Despite their different functions in the cell, the members of the ENaC/DEG family retain common functional characteristics. Other recent reviews cover relevant aspects of ENaC/DEG family members in more detail than this review: amiloride-sensitive channels in epithelia (87), molecular mechanisms of human hypertension (153,209), taste reception (155,156), nociception (131, 175,269), touch sensation in C. elegans(82, 243), and mechanotransduction (70, 88, 102).


Figure 1 shows a phylogenetic tree of the most relevant ENaC/DEG sequences available to date. Seven different branches can presently be distinguished in this gene family: the three main subfamilies comprise the SCNN1 genes encoding the ENaC α-, β-, γ-, and δ-subunits; the C. elegansdegenerins (UNC, MEC, DEG, DEL); and the ASICs. Smaller subfamilies found in invertebrates include the Drosophila proteins RPK/dGNaC1 and PPK/dmdNaC1, the peptide-gated Na+channel FaNaC of mollusks, and FLR-1 in C. elegans that is clearly distinct from the degenerins (171). The mammalian BLINaC (brain-liver-intestine amiloride-sensitive Na+channel) and hINaC (human intestine Na+channel) genes encode Na+ channels that are clearly distinct from ENaC. The amino acid sequence identity between the different ENaC/DEG subfamilies is ∼15–20%, whereas the identity within subfamilies is ∼30% for the different ENaC subunits; ∼30% for degenerins; 45–60% between the four ASIC genes ASIC1, ASIC2, ASIC3, and ASIC4; 38% for the two characterized Drosophilamembers; and ∼65% for the three FaNaC orthologs. Within the ENaC subfamily two branches can be distinguished, one leading to the α- and δ-subunits and the other to β- and γ-subunits. The ENaC genes have been cloned from various species such as rat, human, cow, mouse, and Xenopus laevis, and for the sake of clarity, only human and rat sequences were used for the phylogenetic tree in Figure 1. The homology between human and rat orthologs of ENaC subunits is ∼85%, and it is close to 100% between human and rat orthologs of ASIC1, ASIC2, and ASIC4 and ∼83% for ASIC3. The hINaC and BLINaC genes share 79% sequence identity, and hINaC is probably the ortholog of the rodent BLINaC.

Fig. 1.

Phylogenetic tree of the epithelial sodium channel (ENaC)/degenerin (DEG) family showing the organization into subfamilies of related sequences. Sequences were aligned by using the ClustalW algorithm. The channels from vertebrates are divided into three groups: ENaC, acid-sensing ion channels (ASICs), and brain-liver-intestine sodium channel (BLINaC)/human intestine sodium channel (hINaC). ENaC/DEG proteins of invertebrates can be divided into four groups:1) the degenerins from Caenorhabditis elegans;2) the Drosophila channels RPK/dGNaC1 and PPK/dmdNaC1; 3) FMRFamide-gated sodium channel (FaNaC), which is expressed in mollusks; and 4) FLR-1, which is the only characterized member of a group of C. elegans ENaC/DEG family members that are different from the degenerins.

ENaC (47, 54, 142,151, 247) and ASIC genes (see below) have different splice variants. For ASIC channels the nomenclature remains somewhat confusing in the literature. Initially ASICs were considered as the homologs of the degenerins because of their distribution in the central nervous system (CNS), and they were named mammalian degenerins (MDEG). The phylogenic tree however shows that the DEG and ENaC subfamilies are equally distant from the ASIC subfamily (Fig. 1); therefore, the name mammalian DEG for ASIC is phylogenetically incorrect. We use here a unified nomenclature adapted from Waldmann and Lazdunski (267). The ASIC subfamily comprises the following members: ASIC1a (also known as BNaC2, Refs. 83,264), ASIC1b (or ASICβ, Refs. 18, 40), and additional splice variants of ASIC1 (40, 252), ASIC2a (or MDEG, MDEG1, BNC1, BNaC1, Refs. 83, 201, 266, 268), ASIC2b (MDEG2, Ref. 158), ASIC3 (or DRASIC, hTNaC1, Refs. 13, 61, 116, 263), and ASIC4 (or SPASIC, Refs. 7, 97).

The degenerin subfamily comprises only 6 members out of the 23 predicted ENaC/DEG proteins identified in the C. elegansgenome (171, 246). FLR-1 is the only characterized member of a subfamily of eight members that can be defined by the presence of a conserved region in the extracellular domain related to, but distinct from, degenerin CRDII (171). FLR-1 is expressed in the C. elegansintestine and appears to be also functionally distinct from the degenerins (126, 134, 241).

Sequencing of the Drosophila genome identified 24 putative proteins related to the ENaC/DEG family (162), but so far only the two gene products Ripped Pocket (RPK/dGNaC1) (3,55) and Pickpocket (PPK/dmdNaC1) (3,56) have been characterized with regard to their function and/or tissue distribution.

The first draft sequence of the human genome reveals the presence of 11 proteins that can be assigned to the ENaC/DEG family: α-, β-, γ-, and δ-ENaC, ASIC1, -2, -3, and -4, and two or three proteins related to hINaC (256). Thus, at the present state of completion of the human genome, it seems unlikely that the number of mammalian genes encoding ENaC/ASIC proteins will greatly expand. It is interesting that no mammalian orthologs of the C. elegans DEG or FLR-1 genes, the DrosophilaRPK/dGNaC1 and PPK/dmdNaC1 genes or of FaNaC have yet been identified in the human and the mouse genomes. This might indicate a divergence of the different ENaC/DEG subfamilies early in evolution.


A.  ENaC

The ENaC is located in the apical membrane of polarized epithelial cells where it mediates Na+ transport across tight epithelia. In contrast to other Na+-selective channels involved in the generation of electrical signals in excitable cells, the basic function of ENaC in polarized epithelial cells is to allow vectorial transcellular transport of Na+. This transepithelial Na+ transport through a cell basically involves two steps as illustrated in Figure2. The large electrochemical gradient for Na+ existing across the apical membrane provides the driving force for the entry of Na+ into the cell. Active Na+ transport across the basolateral membrane is accomplished by the Na+-K+-ATPase.

Fig. 2.

Transepithelial ion transport in a principal cell of the cortical collecting duct (CCD). ENaC mediates Na+ entry from the tubule lumen at the apical membrane, and the Na+-K+-ATPase extrudes Na+ at the basolateral side. K+ channels are present on the basolateral and apical membranes. K+ channels at the apical membrane mediate K+ secretion into the tubular lumen. The diagram also illustrates the action of aldosterone (Aldo) which binds to intracellular receptors that are translocated to the nucleus and affect the expression and subcellular localization of ENaC and the Na+-K+-ATPase as well as other target proteins via aldosterone-induced transcripts (AITs) and aldosterone-repressed transcripts (ARTs).

The apical entry of Na+ is blocked by submicromolar concentrations of amiloride. The ENaC-mediated electrogenic Na+ absorption in the distal nephron creates a favorable electrochemical driving force for K+ secretion into the tubule lumen. This active transepithelial transport of Na+is important for maintaining the composition and the volume of the fluid on either side of the epithelium. In the kidney and the colon, which are target tissues for aldosterone action, the transepithelial sodium transport is crucial for the maintenance of blood Na+ and K+ levels and their homeostasis. In the lung or in salivary glands, Na+ transport is certainly not important for the whole body Na+ homeostasis, but rather for keeping the composition and the volume of the luminal fluid, i.e., the saliva or the alveolar fluid constant.

1.  Kidney

In the kidney, ENaC is expressed in the distal nephron where sodium reabsorption is controlled by the mineralocorticoid hormone aldosterone. In the distal nephron, Na+ reabsorption can be measured in in vitro microperfused tubules as a large amiloride-sensitive Na+ flux in sodium-deprived or aldosterone-treated animals (41, 204,221). The first descriptions of the amiloride-sensitive Na+ current at the single-channel level using the patch-clamp technique were obtained from principal cells of microdissected cortical collecting ducts (CCDs) (190) and from a cell line derived from the distal nephron of amphibians that responds to aldosterone (104). The presence of the amiloride-sensitive Na+ current correlates with the expression of ENaC both at the mRNA and protein levels in the distal convoluted tubule, the connecting tubule, the CCD, and the outer medullary collecting duct. There is a clear axial heterogeneity of ENaC expression along the distal nephron; expression of ENaC is higher in superficial cortical regions of the distal nephron such as the connecting tubule or the CCD than in the deeper medullary regions such as the inner medullary collecting duct (IMCD) (69, 167,168, 173, 260). In the CCD ENaC colocalizes with the water channel aquaporin-2, both being stimulated by vasopressin in this nephron segment (166). An electrogenic Na+ transport sensitive to aldosterone is important in amphibian bladder and has also been found in the bladder of some mammals (150).

Under physiological conditions of hydration and salt balance, Na+ reabsorbed in the distal nephron represents only a small fraction (<5%) of the filtered load. However, Na+reabsorption in the distal nephron can increase considerably in response to aldosterone or vasopressin secretion that is stimulated by dehydration and/or salt deprivation. This hormonal control is essential for the fine-tuning of Na+ reabsorption in the distal nephron and the maintenance of sodium and fluid balance (208). The physiological and pathophysiological role of ENaC in Na+ and K+ homeostasis has been clearly demonstrated in human genetic studies and later confirmed by disruption of ENaC genes in mouse models by homologous recombination. Mutations in the β- and γ-ENaC genes causing hyperactive channels have been found in patients with Liddle's syndrome (105,106, 114, 152,218). Liddle's syndrome is a rare hereditary form of hypertension characterized by low plasma aldosterone levels and a low renin activity often associated with hypokalemia and/or metabolic alkalosis. These clinical features reflect an abnormally high Na+ reabsorption in the distal nephron leading to expansion of the extracellular fluid volume and to the development of arterial hypertension. The insertion of a mutation causing Liddle's syndrome in the mouse β-ENaC gene locus by homologous recombination generated a mouse with a phenotype milder than the Liddle's syndrome in humans with extracellular volume expansion, high blood pressure associated with metabolic alkalosis, and hypokalemia that become evident only when animals are challenged with a high-salt diet (199). The elucidation of the genetic basis of Liddle's syndrome demonstrated the critical role of ENaC in maintaining a balance between the Na+ intake and Na+ excretion by the kidney (see Ref. 153 for review). The role of ENaC in Na+ homeostasis was further evidenced by the identification of mutations in ENaC causing reduced channel activity or complete loss of channel function associated with pseudohypoaldosteronism type 1 (PHA-I) (39). The renal symptoms of this heterogeneous syndrome include hyponatremia, hypotension, and hyperkalemia and are associated with elevated plasma aldosterone and renin levels. Mouse models for PHA-I were obtained by different gene targeting strategies leading to decreased expression of α- or β-ENaC genes (112,198) or complete knockout of the α-, β-, or γ-ENaC gene (15, 111, 176). The renal symptoms of these PHA-I mice were similar to those of the PHA-I patients, with a renal loss of sodium associated with hyperkalemia despite elevated plasma aldosterone levels. The physiological consequences of ENaC gene targeting in these animal models confirmed the critical importance of the three genes α-, β-, and γ-ENaC for channel function and Na+ absorption in the distal nephron.

2.  Lungs

The airway epithelia absorb Na+ via an amiloride-sensitive electrogenic transport. This active Na+ absorption is important for the maintenance of the composition of the airway surface liquid. The expression of the ENaC subunits along the respiratory epithelium is complex and varies between species. In adult rats and humans, the α-, β-, and γ-ENaC subunits are highly expressed in small and medium-sized airways (30, 74, 242). The α- and γ-subunits but not the β-subunit are expressed more distally in the lung, which may well correspond to a localization in the type II alveolar cells. This heterogeneity of the expression of ENaC subunits along the airways suggests differential regulation of liquid absorption by channels of various subunit compositions.

At birth the amiloride-sensitive electrogenic Na+transport is important to clear the liquid that fills the alveoli and the airways of the fetal mouse lung. mRNAs for α-, β-, and γ-ENaC can be detected in the fetal lung around days 15–17 of gestation, and expression of ENaC subunits (mainly α- and γ-ENaC) sharply increases in the late fetal and early postnatal life when the lung turns from a secretory to an absorptive organ (242). The physiological role of ENaC in lung liquid balance was clearly demonstrated in mice in which the α-ENaC gene was inactivated by homologous recombination (111). These α-ENaC knock-out mice die soon after birth from respiratory failure due to a severe defect in the clearance of the fetal liquid that fills the lungs. These studies suggest that at birth α-ENaC in the mouse fetal lung is essential for Na+ absorption. The disruption of the β- and γ-ENaC gene loci results in a slower clearance of the fetal lung liquid at birth but does not severely affect the blood gas parameters. The β- or γ-ENaC knockout mice die slightly later than the α-knockout from severe electrolyte imbalance, namely, hyperkalemia due to deficient renal K+ secretion (15, 176). Thus, in contrast to the kidney, the Na+ transport in the lung can be maintained efficiently by only two functional ENaC genes, i.e., the pairs α-β or α-γ.

In humans the contribution of α-ENaC to the clearance of fetal lung liquid at birth is still unclear. Very premature infants with respiratory distress syndrome have a reduced sodium absorption across the respiratory epithelia, as demonstrated by a reduced nasal transepithelial potential difference, likely contributing to the pathogenesis of this syndrome (14). However, PHA-I patients with severe disruption of the α-ENaC gene leading to near-complete channel loss of function have no report of respiratory distress syndrome at birth but show a more than twofold higher liquid volume in airway epithelia than normal individuals (139). Thus ENaC function in humans does not seem to be limiting at birth for the liquid clearance in the mature fetal lung. Differences between species in maturation of the lung, in mucociliary clearance, or in ENaC subunit expression in the respiratory epithelium may account for the phenotypic differences between human and mice.

3.  Gastrointestinal tract and skin

In salivary glands, α-, β-, and γ-ENaC are detected at the RNA and protein levels in the apical membrane of the striated and interlobular ducts (69). An amiloride-sensitive Na+ conductance has been described in mandibular duct cells using the whole cell patch-clamp technique (63). The IC50 of amiloride inhibition was ∼5 μM, thus ∼50 times higher than the IC50 of ENaC. The functional characteristics of this amiloride-sensitive current still await more detailed characterization to ascertain the functional contribution of ENaC in mediating Na+ reabsorption in salivary glands.

The colon is a tight epithelium and an important site for Na+ absorption. Part of this Na+ transport is electrogenic, sensitive to amiloride and stimulated by aldosterone (71). Initially, the α-, β-, and γ-ENaC cDNAs were isolated from a rat distal colon cDNA library because of the high level of channel expression in this tissue (31,33, 159).

The classical model of transepithelial Na+ transport in tight epithelia (144) was originally described by Koefoed-Johnson and Ussing in the skin of amphibians and postulated the presence of an apical Na+ channel and a basolateral Na+-K+-ATPase in series. The Na+transport in the amphibian skin responds to aldosterone with simultaneous morphological changes (261). In mammals, the three ENaC subunits are expressed in keratinocytes of all epidermal layers as well as in epithelial cells of hair follicles and sweat glands (210). ENaC transcripts can also be found in the pluristratified epithelium of the esophagus where ENaC's role is unknown (74). In sweat glands ENaC likely controls Na+ excretion as in the kidney, but in the epidermis and hair follicles, the physiological role of ENaC is less clear. There is still no clear evidence for an ENaC-mediated transcellular Na+ transport in mammalian epidermal cells, despite the presence of a benzamil-sensitive current in cultured human keratinocytes consistent with ENaC activity (29). Interestingly, expression of the α- and β-ENaC subunits is upregulated in cultured keratinocytes during differentiation, suggesting that ENaC may play a role in epidermal differentiation and skin development. Alternatively, this upregulation might be indirectly linked to the regulation of cell volume during this process (29, 183).

Finally, salt taste is transduced by direct amiloride-sensitive influx of Na+ in the taste cells of the fungiform papillae of the anterior part of the tongue, suggesting the presence of an amiloride-sensitive Na+ channel (for review, see Refs.155, 156). The three α-, β-, and γ-ENaC subunits are expressed in the taste receptor cells of the fungiform papillae (147,154). The specific role of ENaC in salty taste transduction still remains to be clearly demonstrated, since the amiloride-sensitive inward currents recorded from whole taste receptor cells show lower affinity for amiloride and a lower single-channel conductance than expected from a typical ENaC current as estimated from noise analysis (9,10).

4.  Other tissues

The δ-ENaC subunit, closer to α- than to β- and γ-ENaC in its amino acid sequence, has been identified by sequence homology (Fig.1). As expected from its close homology to α-ENaC, the δ-ENaC subunit can substitute for α-ENaC to form functional amiloride-sensitive Na+ channels. The δ-ENaC subunit is expressed in testis, ovary, pancreas, and to a lesser extent in brain and heart (265). In the pancreas as in the ducts of salivary glands, an amiloride-sensitive channel might be responsible for Na+ reabsorption. In the other tissues expressing δ-ENaC, amiloride-sensitive Na+ currents have not yet been detected, and therefore, the physiological role of δ-ENaC in these tissues remains unknown.

The membranous labyrinth of the cochlea is a complex sensorineural epithelium. The basolateral side is bathed with the perilymph of similar ionic composition as the plasma or as cerebrospinal fluid. The endolymph at the apical surface of the membranous labyrinth is a K+- rich, hyperosmotic fluid almost devoid of Na+. The endolymph bathes the hair bundles of the sensory cells, and its ionic composition is critical for mechanotransduction. The α-, β-, and γ-ENaC subunit mRNAs are detected in the rat cochlea in both epithelial and nonepithelial structures (53). Interestingly, expression of the three ENaC subunits was strongest in epithelial cells (Claudius cells in particular) lining the scala media. The Claudius cells might be responsible for Na+ reabsorption from the endolymph; the unusually high potential difference existing across the apical membrane of these cells (+80 mV in the endolymph and −100 mV intracellular) allows absorption of Na+ from the endolymph at lower than millimolar concentrations (99). Furthermore, the mRNA abundance of the α-, β-, and γ-ENaC subunits in the cochlea increases during the first 10 days of life when the high-K+, low-Na+ composition of the endolymph is established. This observation suggests that ENaC in the epithelial cells of the scala media is critical for Na+ extraction from the endolymph. However, no severe hearing problems have been reported for PHA-I patients with loss-of-function mutations of ENaC.

ENaC transcripts and proteins were detected in retina photoreceptors, but ENaC's role in phototransduction remains to be established (89). Finally, the γ-ENaC subunit has been localized in baroreceptor nerve termini innervating the carotid sinus (67). Structures specialized in mechanosensation in nerve endings of the rat footpad also express β- and γ-ENaC, and it was proposed that ENaC subunits represent components of mechanoreceptors for touch or blood pressure sensing (66). An involvement of ENaC in mechanotransduction has not yet been directly demonstrated.


ASICs were cloned based on their homology to ENaC/DEG channels (for a recent review, see Ref. 267). Four ASIC genes (ASIC1–4) have been identified, and three of them exist in different splice variants (Fig. 1, see sect. ii). The expression of ASIC mRNAs has been analyzed by Northern blotting and in situ hybridization (26, 267) and shows a high expression in the nervous system. ASIC1a, ASIC2a, and ASIC2b have a similar widespread distribution pattern in the brain, with the highest expression levels in the hippocampus, the cerebellum, the neo- and allocortical regions, the main olfactory bulb, the habenula, and the basolateral amygdaloid nuclei. Coexpression with ASIC4 is found in many areas of the brain (7, 97). In addition, strong expression of ASIC4 has been found in the pituitary gland (97).

In the peripheral nervous system, the ASIC channels are found predominantly in small-diameter sensory neurons involved in pain sensation (40). Rat ASIC3 is expressed almost exclusively in sensory neurons (12, 13, 40,263). ASIC3 is expressed in sympathetic cardiac afferents that innervate the heart where they may play a role as mediators of cardiac pain (23, 240). Expression of ASIC2a in sensory neurons is lower compared with other ASICs (158, 200). ASIC2a is also found in the taste buds of the circumvallate papillae where it may have a function in sour taste perception (250). ASIC2b has to date only been found in the rat. As for ENaC subunits, ASIC4 expression was demonstrated in the inner ear, but its role in hearing function remains to be established (97).

ASIC channels are activated by a drop of the extracellular pH. Expression of homomultimers or heteromers of ASIC1–3 in heterologous cell systems generates proton-gated cation currents with functional properties resembling the native currents found in sensory and CNS neurons and in oligodendrocytes (25, 92,148, 164, 255,267). The magnitude of these depolarizing H+-gated currents is sufficient to initiate action potentials in neurons (72, 255). So far, no channel function has been found for ASIC4 (7,97).

The physiological role of ASICs remains uncertain. Their expression in sensory neurons suggests a role in pain perception following tissue acidosis (203, 263, 267). For instance, good electrophysiological evidence supports the involvement of ASIC3 in cardiac ischemic pain sensation (23,240). Myocardial extracellular pH drops to 6.7 during severe cardiac ischemia, a pH value that can evoke a high-amplitude current in the sympathetic cardiac afferents expressing ASIC3 (240). This current shows the same functional characteristics as the H+-gated current in COS-7 cells expressing ASIC3 (240). Amiloride that blocks ASICs with low affinity has recently been shown to have analgesic effects in a variety of animal pain models (75). This finding however has to be interpreted with some caution because at concentrations at which amiloride blocks ASICs, it also blocks other channels and proteins in the brain, as, e.g., voltage-gated Ca2+channels (143). Finally, nonsteroid anti-inflammatory drugs inhibit ASIC currents (259). The potential involvement of ASICs and other ion channels in pain sensation has been reviewed (131, 175, 269).

In the CNS, tissue acidosis is a well-established feature of cerebral ischemia which aggravates cell damage (reviewed in Ref. 161). Transient global ischemia induces expression of ASIC2a protein in neurons that survive ischemia (129). ASICs may also be involved in controlling neuronal activity associated with external pH fluctuations (44). A significant acidification of the extracellular space has been found to be associated with repetitive stimulation and epileptiform activity (43,44, 233, 253, 270). Recently, it has been shown that expression levels of ASIC1a and ASIC2b (but not ASIC2a) decrease in specific brain regions in the pilocarpine model of epilepsy (26). The question remains if in vivo extracellular pH changes are large enough to activate ASIC channels. It is possible that under physiological conditions endogenous ligands of ASICs such as Zn2+ or FMRFamide-related peptides may facilitate activation (8, 16).

ASIC2 knockout mice (200) had normal appearance, growth, size, temperature, fertility, and life span, with no obvious defects in pain sensation. The only phenotype that could be detected in an in vitro skin-nerve preparation from ASIC2−/− mice was a reduced sensitivity of a specific component of mechanosensation, involving the low-threshold rapidly adapting mechanoreceptors. In rodent hairy skin, these mechanoreceptors are excited by hair movement. Consistent with this function, ASIC2a/b was found in the lanceolate nerve endings that lie adjacent to and surround the hair follicle. In a recent immunocytochemical study, expression of ASIC2a was found in specialized cutaneous mechanosensory nerve terminals (86). The relatively discrete phenotype due to ASIC2 loss of function could be due to a redundancy of ASIC proteins in the formation of heteromultimeric channels.

C.  Degenerins

Mechanosensory transduction refers to processes that convert mechanical forces into bioelectrical signals. Mechanotransduction is important for different physiological functions such as touch, hearing, or proprioception. The extensive genetic dissection of mechanosensory behavior in C. elegans led to the identification of a number of genes involved in the development, survival, function, and regulation of touch receptor neurons (37). Among these genes, the degenerins that form a subgroup of the ENaC/DEG family play a critical role in touch sensation and proprioception.

Specific mutations in MEC-4, MEC-10, UNC-8, UNC-105, or DEG-1 cause swelling and subsequent death of the cells in which the mutant proteins are expressed. These proteins were called degenerins because of the cell degenerative phenotype they cause when mutated (reviewed in Refs.171, 243). Typical morphological features involved in this phenomenon include the production of membrane infoldings, membrane whorls, and vacuoles that resemble those found in excitotoxic cell death after ischemia, hypoxia, or epilepsy. The initial response of the cell is an enhanced membrane cycling that may reflect attempts to dilute or to sequester harmful components localized to various membranes and to translocate them to degradative vacuoles (101). On the basis of their homology with ENaC, it has been proposed that these degenerins function as mechanosensory channels which when mutated become hyperactive leading to increased cation influx into the cell and to cell death.

Mechanotransduction in C. elegans has been described in detail in several recent reviews (82, 171,243). When a C. elegans moving over an agar plate is gently touched with an eyelash hair on the posterior part of the body, the animal will move forward, and when touched on the anterior body, it will move backward. This gentle touch is sensed by six touch receptor neurons. Their processes run longitudinally along the body wall on the ventral and both lateral sides and are embedded in the hypodermis (the “worm skin”), to which they appear to be glued by an extracellular material called the mantle (reviewed in Refs. 82,243). The position of the processes along the body axis correlates with the sensory field of the touch cells.

Genetic analysis has identified 16 genes which, when mutated, specifically disrupt gentle body touch sensation; they represent, therefore, candidate mediators of touch sensitivity. These genes were named mec genes because when defective the animals become “mechanosensory abnormal.” Many of the mec genes have been molecularly identified, and the encoded proteins were postulated to form a touch-transducing complex that converts the mechanical stimuli into an electrical signal. Only some of themec genes belong to the ENaC/DEG family such as MEC-4 (243) or MEC-10 expressed in the touch receptor neurons (110).

The transduction of mechanical stimuli into bioelectrical signals requires activation of an ion channel to generate an ionic current. So far, functional expression experiments have failed to demonstrate mechanically inducible channel activity. Constitutively active channels were obtained after coexpression of MEC-4 and MEC-10 with the stomatin-related protein MEC-2 (Fig. 4) or upon expression of UNC-105 containing activating mutations (84,90). However, the mechanosensitivity of these degenerins still remains to be demonstrated in heterologous expression systems.

UNC-8 (unc stands for uncoordinated) is expressed in several sensory neurons, interneurons, and the motorneurons VA and VB (245). This subclass of motorneurons shows the peculiar neuroanatomical feature of having their synapses with the muscle or the interneurons restricted to a limited region close to the cell body. Similar anatomical features are found in corresponding motorneurons in the nematode Ascaris suum that are known to be stretch responsive, suggesting that VA and VB motorneurons in C. elegans have similar functions (57). The degenerins UNC-8 and DEL-1 are coexpressed in the ventral cord VA and VB motorneurons. Functional evidence for involvement of UNC-8 in modulation of locomotion comes from observation of wild-type and UNC-8 mutant animals. When moving on agar plates, wild-type worms inscribe a sinusoidal wave. In unc-8 null mutants, the amplitude is reduced by ∼3.5-fold (245). It has been proposed that the contribution of UNC-8 to modulation of nematode locomotion relies on feedback information on body posture, thus on an activity that is related to proprioception (243,245). Alternatively UNC-8, which is also expressed in interneurons that regulate locomotion, might act within these neurons to modulate coordinated movement. The only evidence for the involvement of DEL-1 in modulation of C. elegans locomotion lies in its coexpression with UNC-8 in the VA and VB motorneurons and its homology to other ENaC/DEG family members (243, 245). Like MEC-4 and MEC-10, UNC-8 and DEL-1 are thought to form the ion channel core of the stretch-sensitive complex mediating modulation of C. elegans locomotion.

Yet another member of the C. elegans degenerin subfamily contributes to proprioception by monitoring muscle stretch. The UNC-105 gene is not expressed in neurons but in muscles. Mutations of the UNC-105 ENaC/DEG gene have been identified that create constitutively activated channels that cause muscle hypercontraction (163, 194), presumably because muscle cells are depolarized by an inappropriate cation influx through the UNC-105 channels. As mentioned above, expression of UNC-105 containing activating mutations in Xenopus oocytes results in ion channel activity, but mechanosensitivity of the UNC-105 channel has not been demonstrated (84).

D.  Drosophila ENaC/DEG Members

Among the 24 predicted proteins of Drosophila related to the ENaC/DEG family (162), only Ripped Pocket (RPK/dGNaC1) (3, 55) and Pickpocket (PPK/dmdNaC1) (3, 56) have so far been characterized with regard to their function and/or tissue distribution. RPK/dGNaC1 transcripts are present exclusively in fly ovary and testis; they are maternally deposited into the embryo, where they persist until late gastrulation. The expression of RPK/dGNaC1 transcripts inDrosophila is restricted to oocytes in late vitellogenic stages and to early embryos, as well as to nurse and follicular cells. This suggests that RPK/dGNaC1 is involved in early development. When expressed in Xenopus oocytes RPK/dGNaC1 forms functional Na+-selective channels with low apparent affinity for amiloride (see Table 1). It was not possible to record any RPK/dGNaC1 activity in transfected mammalian COS cells. This observation was interpreted as RPK/dGNaC1 activity being modulated by specific factors that are present only during early development and might therefore be present in oocytes but not in COS cells (55). Amiloride-sensitive Na+channels in mammalian embryos play an important role in fluid transport across the trophectoderm and in the formation of the blastocyst (3, 55). Like RPK/dGNaC1, these channels appear to have a low sensitivity to amiloride (K i = 12 μM, Ref. 207). RPK/dGNaC1 and related members in Drosophila may play roles in fluid distribution and cell volume regulation during gametogenesis and early development.

View this table:
Table 1.

Biophysiological and pharmacological properties of functionally expressed ENaC/DEG channels

The PPK/dmdNaC1 gene product appears at later developmental stages ofDrosophila. It is expressed in sensory dendrites of a subset of peripheral neurons in late-stage embryos and early larvae. The late PPK/dmdNaC1 expression coincides with the end of elaboration of the nervous system, after the complete determination and differentiation of neuroprecursor cells. This led to the hypothesis that PPK/dmdNaC1 is not involved in the differentiation of the peripheral nervous system but rather in the function of PPK/dmdNaC1-expressing neurons (56). In insects, multiple dendritic neurons in which PPK/dmdNaC1 is expressed are thought to play a role in touch sensation and proprioception. For this reason, it has been proposed that in analogy to degenerins, PPK/dmdNaC1 may function as a mechanotransduction channel. PPK/dmdNaC1 does not produce functional ion channels when expressed in Xenopus oocytes (3, 56).

E.  FaNaC

The FMRFamide (Phe-Met-Arg-Phe amide)-gated channel, or FaNaC, was originally cloned from a cDNA library from the snail Helix aspersa nervous system, based on its homology to ENaC/DEG family members. FaNaC mRNA is detected in the nervous system and in pedal muscle of this snail (157). The biophysical and pharmacological properties of the cloned FaNaC expressed inXenopus oocytes are very similar to those of the native FMRFamide-gated current in Helix neurons (51, 93, 94). FMRFamide induces a fast excitatory depolarizing response due to direct activation of the channel in neurons of the snail Helix aspersa as well as inAplysia motor neurons (19, 94,211). This fast response is readily distinguished from other, slower neuronal FMRFamide responses in Helix aspersa(reviewed in Ref. 51) that depend on G proteins. Evidence that the peptide directly gates a channel was obtained using isolated membrane patches from the Helix C2 neuron and from Xenopusoocytes expressing the cloned FaNaC. Inward unitary currents could be generated by external application of FMRFamide to excised outside-out patches when G protein-mediated responses were inhibited (94, 157). This finding represented the first functional description of a peptide-gated ion channel. Recently, the cDNAs encoding FMRFamide-gated channels from two other mollusks, Helisoma trivolvis and Lymnaea stagnalis, were cloned (127, 196). These different clones share 60–65% identity of the amino acid sequence. So far, no mammalian homologs of FaNaC have been identified (8, 256), but FMRFamide and the mammalian neuropeptide FF as well as related peptides can modulate the activity of ASICs in heterologous expression systems (8).


A.  Primary Structure and Membrane Topology

The size of the ENaC/DEG proteins ranges from ∼530 to ∼740 amino acids. Regions of conserved amino acid sequences in ENaC/DEG family members are likely to represent structural elements important for channel function. These conserved domains shown in Fig.3, A and B, are located essentially in the transmembrane segments and their proximity and in the extracellular loop.

Fig. 3.

Conserved domains and their localization in ENaC/DEG family members.A: linear representation of the primary structure shows the conserved regions found in each of the main subfamilies. The length of the domains in the diagram reflects their relative size in the protein. M1, M2, transmembrane segments; HG, conserved His-Gly motif within a conserved NH2-terminal domain; pM1, post-M1, conserved domain directly downstream of M1; CRD, cysteine-rich domain; pre-M2, hydrophobic domain directly preceding M2; ERD, extracellular regulatory domain unique to C. elegans degenerins; NTD, neurotoxin domain (244); PY, PPPxY domain of α-, β-, and γ-ENaC. B: membrane topological organization of an ENaC/DEG subunit. The part that is unique to degenerins (containing CRDI and ERD) is shown on a gray background encircled by an interrupted line. Abbreviations are as in A. The residue in the pre-M2 domain whose mutation induces degeneration of the cells expressing these mutant degenerins is indicated as “deg.”

A few sequences are completely conserved among the ENaC/DEG family members; they include a HG motif (His-Gly) located in the NH2-terminal cytoplasmic domain in close proximity to the M1 segment, a FPxxTxC sequence following M1 (post-M1) and completely conserved residues in the M2 region. The extracellular loop contains cysteine-rich domains (CRD) II and III. The conserved Cys residues might be involved in the formation of disulfide bonds to maintain the tertiary structure of the large extracellular loop. Candidate cysteine residues in the extracellular loop of α-ENaC involved in disulfide bridges have been identified by tandem Cys mutations and functional analysis (77).

Other domains are conserved only within particular ENaC/DEG subfamilies. Two domains in the extracellular loop are conserved exclusively within the degenerin subfamily, the cysteine-rich domain CRDI and the extracellular regulatory domain ERD (85). ENaC subunits contain conserved proline-rich motifs in the cytoplasmic COOH terminus that are unique to the ENaC subfamily. These motifs have the consensus sequence PPPXYXXL and are involved in protein-protein interactions.

The membrane topology has been investigated essentially for ENaC, but we can assume from the overall sequence homology that this topology is shared by the other ENaC/DEG family members. Analysis of the membrane topology of ENaC was performed by protease digestion analysis of α-ENaC translated in vitro and reconstituted in microsomes (205). This analysis was consistent with M1 and M2 being separated by a large extracellular loop of ∼50 kDa, the NH2 and COOH termini being cytoplasmic. This large extracellular loop makes up more than one-half of the channel protein and represents a structural feature unique to the ENaC/DEG family members that is not found in other ion channel families. Several putative glycosylation sites are present in the extracellular loop of ENaC subunits (6 for α, 12 for β, and 5 for γ). Site-directed mutagenesis in α-ENaC showed that the six potentialN-glycosylation sites between M1 and M2 but not the two putative N-glycosylation sites preceding M1 are used, demonstrating the extracellular topology of the loop between M1 and M2 (32, 230). In addition, with the use of truncated α-ENaC fused with the Na+-K+-ATPase β-subunit that is known to be glycosylated, it was demonstrated that M1 and M2 serve as start-transfer and stop-transfer signals, respectively, consistent with both M1 and M2 crossing the membrane and leaving the COOH terminus cytoplasmic (32).

The hydrophobic profile of ENaC subunits reveals a stretch of hydrophobic amino acid residues preceding M1 and M2. The protease digestion pattern of α-ENaC is consistent with the presence of a short segment preceding M2 that is accessible for the protease from the cytoplasmic side (205). In all known Na+- or K+-selective cation channels, the short hydrophobic segment preceding the most COOH-terminal transmembrane domain forms the outer pore of the channel with the receptor sites for external blockers. This analogy led to the suggestion that the pre-M2 segment of ENaC subunits may contribute to the formation of the channel pore.

B.  Multimeric Channels and Subunit Stoichiometry

1.  ENaC/FaNaC

The initial cloning experiments demonstrated that the three homologous ENaC subunits α, β, and γ are required for maximal expression of ENaC activity (33). Thus it became clear that the structure of ENaC is a heteromultimeric channel similar to the nicotinic acetylcholine receptor. There is presently good evidence that the three ENaC subunits contribute to the formation of the channel pore and line a unique ion permeation pathway (137,138, 220, 223, 231) (see sect. v). It should be mentioned that other models for the subunit arrangement around the channel pore have been proposed based on functional experiments with channels reconstituted in lipid bilayers. According to these models, ENaC shows a triple barrel organization to account for the different subconductance states of the channel reconstituted in lipid bilayers (22). Such models for ENaC have not gained any experimental support from functional studies in cellular expression systems such as the Xenopusoocyte.

The relative abundance of the three ENaC subunits at the cell surface in Xenopus oocytes expressing amiloride-sensitive Na+ current was determined by quantitative analysis of the binding of radiolabeled antibodies directed against a FLAG epitope introduced into the extracellular domain of α-, β-, and γ-subunits (78). The low current expression after expression of α-ENaC alone is correlated with a low level of α-ENaC subunit expression at the cell surface, suggesting a retention of α-ENaC in the endoplasmic reticulum due to the absence of a molecular signal essential for targeting the channel to the cell surface. In oocytes expressing the three ENaC subunits α, β, and γ, the α-subunit is approximately two times more abundant in active ENaC channels at the cell surface compared with β- or γ-subunits (76). This 2:1:1 ratio of channel subunits remains constant even under conditions where one subunit is overexpressed relative to the two other homologous subunits (76), consistent with a preferred fixed subunit stoichiometry when the three ENaC subunits are available for channel assembly.

Despite the preferred subunit stoichiometry of ENaC, the α-, β-, and γ-subunits are to some extent interchangeable. Expression of α- or δ-ENaC alone in Xenopus oocytes has been shown to generate small amiloride-sensitive currents, suggesting that homomeric α-ENaC and δ-ENaC channels are functional. Expression of the ENaC subunit combinations α-β or α-γ generates a similar level of amiloride-sensitive current that is higher than after expression of α-ENaC alone, suggesting that β- and γ-subunits can replace each other in the channel complex (33,78, 177). Early expression experiments inXenopus oocytes indicated that β-γ channels were not expressed at the cell surface (33, 78). However, it has been later shown that 6 days after injection of cRNA encoding β- and γ-ENaC subunits a small but significant amiloride-sensitive current could be detected, suggesting that functional channels can be formed with β- and/or γ-subunits (27).

In the absence of high-resolution images, the determination of the number of subunits in the functional channel relies on a biochemical analysis of the channel complex at the cell surface. This approach was first performed on FaNaC because this channel is functional as a homomultimer. In addition, the mature FaNaC at the cell surface is fully glycosylated and thus biochemically identifiable. Following the maturation process and the oligomerization of the channel complex, Coscoy et al. (50) showed that the tetrameric assembly of the channel occurs already in the endoplasmic reticulum where this immature oligomeric complex is core glycosylated. Chemical cross-linking experiments were consistent with a minimal stoichiometry of four subunits, and the hydrodynamic characterization of the channel complex revealed a maximal number of five subunits (50).

Alternatively, the number of homologous subunits forming a channel can be assessed by the quantitative analysis of the contribution of each subunit to channel function. The identification of mutations in α-, β-, and γ-subunits that confer to the channel a resistance to block by amiloride or a sensitivity to block by Zn2+ allowed the design of experiments to assess for each of the three ENaC subunits the number of subunits per channel complex that contribute to a particular channel function (76, 220). These experiments followed the basic principles originally described to determine the subunit stoichiometry of the Shaker K+ channel by studying the interaction of a toxin with channels made of different fractions of wild-type and toxin-resistant mutant subunits (170). The Zn2+ sensitivity of ENaC was conferred to the channel when the channel complex contained two α-subunits carrying a cysteine substitution at the amiloride binding site (76). Inversely, block by the sulfhydryl reagent 2-aminoethyl methanethiosulfonate (MTSEA) was conferred to the channel when at least one of two α-subunits carried the same cysteine substitution (146). These experiments were consistent with the presence of two α-ENaC subunits in the functional channel. The same approach using point mutations on β- or γ-ENaC that confer amiloride resistance did not provide any information about the absolute number of β- and γ-subunits, since the magnitude of the amiloride-resistant current was directly proportional to the fraction of mutant β- or γ-subunit expressed (76,146).

In contrast to FaNaC, ENaC at the cell surface of Xenopusoocytes is not fully glycosylated, making a biochemical approach difficult for assessing the channel subunit stoichiometry. Subunit stoichiometry of ENaC can be assessed by expression of trimeric or tetrameric constructs made of α-, β-, and γ-ENaC subunits linked in a head-to-tail fashion. The rationale of this approach relies on the assumption that the correct concatameric construct does not need to be complemented by monomeric subunits for channel function (145). The analysis revealed that the α-β-α-γ construct is sufficient for the expression of functional channels and does not require complementation by additional monomeric subunits (76). Thus, taken together, the following observations support a tetrameric subunit organization around the channel pore, as illustrated in Figure 4 A:1) the surface expression of α-ENaC per active channel is approximately two times higher than for β, 2) there are consistently two α-subunits per active channel, 3) the number of β- and γ-subunits are equal, and 4) α-β-α-γ is the minimal concatameric construct for functional channel expression. This tetrameric subunit composition of ENaC is consistent with the subunit stoichiometry of other Na+- or K+-selective ion channels and in particular with the stoichiometry of channels made of subunits with two transmembrane domains.

Fig. 4.

Model of ENaC/DEG channel complexes. A: model of the tetrameric assembly of an ENaC channel. The model shows the tetrameric assembly of ENaC subunits around the central channel pore. The model is based on functional analyses that showed that all ENaC subunits participate in the formation of the channel pore and established the subunit stoichiometry (76). B: model of a touch-transducing complex in C. elegans. This model is based on genetic analysis. Touch transduction is thought to be mediated by a degenerin ion channel made of at least two MEC-4 (light gray) and two MEC-10 (dark gray) subunits. The large extracellular domains of the subunits project into a specialized extracellular matrix called the mantle, which includes the MEC-5 collagen and the MEC-9 protein. Inside the touch receptor neuron are special microtubules made up of the MEC-12 α-tubulin and the MEC-7 β-tubulin. The microtubules are proposed to be linked to the channel by means of stomatin-related MEC-2, a peripheral membrane protein. Proteins tethered to the extracellular and intracellular sides are thought to exert pressure on the channel, conferring gating tension when the channel is stretched between these two contact points. [Adapted from Mano and Driscoll (171).]

With a similar approach looking at interactions of sulfhydryl reagents with ENaC wild-type and mutants together with the biochemical determination of the molecular mass of ENaC and FaNaCh, Snyder et al. (229) reached the conclusion that ENaC as well as other members of the ENaC/DEG gene family are made of nine subunits: three α-, three β-, and three γ-subunits in the case of ENaC. This nonameric channel stoichiometry was further supported by freeze-fracture electron microscopy in oocytes expressing ENaC (73). Such high-magnification images revealed particles of 8 nm in diameter compatible with ENaC channels. Based on the assumption that the cross-sectional area of membrane proteins in freeze-fracture microscopy is proportional to the number of the membrane-spanning domains, ENaC was found to have 17 membrane-spanning segments corresponding to 8 or 9 subunits. These conclusions should be taken with caution because the relationship between the cross-sectional area and the number of transmembrane segments is certainly not constant for all membrane proteins, and the functional ENaC channel complex may well include additional helical segments embedded in the membrane or associated proteins.

2.  Touch- and stretch-transducing complexes inC. elegans degenerins

C. elegans degenerins are implicated in touch sensation and proprioception. Based on the genetic analysis of C. elegans mutants with impaired touch and stretch sensation, a model of the touch-transducing complex was proposed in which the degenerins MEC-4 and MEC-10 constitute the channel core of the mechanosensory transduction complex (see Fig. 4 B) (70). MEC-4 and MEC-10 subunits are coexpressed in touch receptor neurons, and genetic interactions could be demonstrated between mec-4 and mec-10 (100,110, 243). In Xenopus oocytes expressing MEC-4 and MEC-10, both proteins could be immunoprecipitated, suggesting the presence of functional heteromeric channels (90). The number of MEC subunits that form the channel pore has not yet been established, but genetic evidence is compatible with at least two MEC-4 and two MEC-10 subunits per channel (108, 110).

Because mechanosensation could not be reconstituted in heterologous expression systems with MEC-4 and MEC-10 alone, additional associated proteins likely participate in the functional complex at the cell surface. Degeneration caused by mutations of a number of degenerins (deg-1, mec-4, mec-10, andunc-8) requires an additional gene, MEC-6, essential for activation of the mechanosensitive channel (reviewed in Ref. 243). MEC-6 has not been cloned yet and might encode a protein that mediates localization or posttranslational modification of the putative channel-forming subunits. In addition, MEC-2, a stomatin-related protein essential for touch sensation, is associated with the channel complex and increases channel activity in oocytes expressing MEC-4 and MEC-10 (90).

To sense mechanical stimuli, the large extracellular loops of MEC-4 /MEC-10 channels are postulated to be linked to the touch cell-specific extracellular matrix, by interaction with the MEC-5 collagen and/or MEC-9, a large protein rich in interaction domains such as epidermal growth factor (EGF)-like repeats (2 of them of the Ca2+-binding type) and Kunitz protease inhibitor domains (68) (see Fig. 4 B). The intracellular domains of MEC-4 and MEC-10 are hypothesized to be tied to the unique 15-protofilament microtubules (made of the MEC-12 α-tubulin and the MEC-7 β-tubulin) by the linker protein MEC-2, which is related to stomatin (reviewed in Ref. 243). By anchoring the channel on both sides of the membrane, displacement of the extracellular matrix with respect to the cytoskeleton is supposed to open the channel in response to mechanical stimuli. A model for the stretch-sensitive complex mediating proprioception similar to that of the touch-transducing complex has been proposed, containing UNC-8 and DEL-1 as subunits of the channel core in the mechanotransducing complex (243).

In summary, there is good evidence that C. elegans DEG subfamily members represent the channel subunits of a large mechanotransduction complex in the plasma membrane. The channel core of this complex is heteromultimeric and made of at least four DEG channel subunits belonging to the ENaC/DEG family. Other components might be involved in this mechanotransduction complex to anchor the channel to cell constituents on both sides of the membrane.

C.  Genomic Organization

The human α-ENaC gene (SCNN1A) covers 17 kb on chromosome 12p13. The human β- and γ-ENaC genes (SCNN1B and SCNN1G) are located on chromosome 16p in very close proximity, suggesting that the two genes arise from gene duplication. The three ENaC genes α, β, and γ share a remarkable degree of conservation in their genomic organization (45, 169, 213,249). The human ENaC genes are divided into 13 exons; the 2 transmembrane regions of ENaC proteins are encoded by parts of exon 2 and exon 13. The δ-ENaC gene is located on the chromosome 1.

The promoters of ENaC genes remain to be precisely identified. In human and rat, the region within −75 bp upstream of the transcription start site of the γ-ENaC gene contains two GC-rich boxes that are sufficient for promoter activity (248). Analysis of the nucleotide sequence in the region farther upstream between −289 and −142 shows two imperfect glucocorticoid response elements (GRE) that represent potential transcriptional regulatory elements necessary for ENaC regulation by glucocorticoids (214). Sequential deletions in this region showed that the downstream GRE is sufficient to confer glucocorticoid stimulation and is also able to bind glucocorticoids specifically. An additional third GRE motif further upstream (between −300 and −2,400) has been identified in the 5′-flanking region of human α-ENaC. Reporter constructs containing this GRE motif also exhibit glucocorticoid-inducible expression (45). So far no specific mineralocorticoid response element has been identified, and the elements that determine the tissue specificity of the glucocorticoid versus mineralocorticoid regulation of ENaC genes remain to be elucidated.

ASIC1, ASIC2, ASIC3, and ASIC4 are located on human chromosomes 12, 17, 7, and 2, respectively (61, 256).


A.  Functional Characteristics

The functional properties of the ENaC/DEG channels have been investigated in native tissues and in cell expression systems using electrophysiological techniques. Long before its cloning by functional expression, the highly selective amiloride-sensitive ENaC was functionally characterized in tight epithelia such as frog skin and toad urinary bladder. More recently, ENaC function at the single-channel level could be resolved in the rat cortical collecting tubule using the patch-clamp technique (87). Some members of the ENaC/DEG family exhibit macroscopic ionic currents when expressed in Xenopusoocytes, but the single-channel characteristics have not yet been determined (Table 1). Electrophysiological investigations of nativeC. elegans cells are possible but remain difficult because of their small size (91, 165). In cell expression systems, hINaC, BLINaC, and UNC-105 channels simply do not express ionic currents unless an activating mutation is introduced in their sequence. Thus much more functional data on ion permeation are available for the amiloride-sensitive ENaC channel compared with other family members.

1.  ENaC

The macroscopic amiloride-sensitive conductance measured in toad bladder or frog skin was found to be selective for the small inorganic Na+ and Li+. In these tissues, the Na+/K+ selectivity ratio was estimated to be >500 (20, 185). This high selectivity of ENaC was confirmed at the single-channel level in the mammalian CCD (190) and later in oocytes expressing the α-, β-, and γ-ENaC subunits (138). In the toad bladder an amiloride-sensitive current carried by protons could be measured when the mucosal milieu was acidified below pH 5, indicating that protons can permeate the epithelial Na+ channel (185, 186). The proton permeability of ENaC expressed in Xenopus oocytes has not been addressed yet. Thus ENaC seems to discriminate among cations based on their size, small cations such as Na+, Li+, or protons being permeant whereas larger cations such as K+ or organic NH 4+ cannot pass through the channel. This simple rule for the ionic permeability of ENaC applies only to monovalent cations, because ENaC is not permeable to divalent cations. In the CCD and inXenopus oocytes expressing α-, β-, and γ-ENaC, the unitary conductance of the amiloride-sensitive epithelial Na+ channel measured in the presence of 140 mM Na+ at room temperature is 5 pS (33,138, 190). A typical trace of ENaC single-channel activity is shown in Figure5 A. The single-channel conductance of ENaC saturates at external Na+concentrations above 100 mM with a concentration for half-maximal conductance (K m) around 20–50 mM extracellular Na+ (138, 184, 192). In the presence of 100–150 mM external Li+, the single-channel conductance measured at room temperature is 9–10 pS in the cortical collecting tubule and in Xenopus oocytes expressing α-, β-, and γ-ENaC. Similar values were measured in a canine kidney cell line (Madin-Darby canine kidney cells) stably transfected with α-, β-, and γ-ENaC (117). The apparent affinity for Li+ is lower than that for Na+, with a K m for Li+around 90 mM in cortical collecting tubule and 120 mM inXenopus oocytes (138, 192). A higher dissociation rate of Li+ from its binding site in the channel pore toward the cytoplasmic side can account for the lower affinity for Li+ compared with Na+ and for the faster movement of Li+ through the channel, i.e., the higher channel conductance for Li+. These data obtained in the native tissue and in heterologous expression systems set unambiguously the functional conductance signature of ENaC: 4–5 pS and 9–10 pS conductance with Na+ and Li+ at concentrations on the order of 100–150 mM and a high Na+/K+ selectivity.

Fig. 5.

Functional fingerprints of some ENaC/DEG proteins. A: ENaC single-channel trace from an outside-out patch of an oocyte expressing rat αβγ-ENaC in 120 mM Na+extracellular solution at −100 mV. C, closed; O, open channel.B: H+-induced currents at two different voltages from COS cells expressing ASIC2 (left) or ASIC3 (right). [Adapted from Lingueglia et al. (158).] C: trace of a current induced by continuous application of 30 μM FMRFamide to an oocyte expressing FaNaC. [Adapted from Lingueglia et al. (157).] Note the difference in time scale between B and C.

The single-channel conductance slightly differs according to the subunit composition of ENaC. In oocytes expressing αβ- or αγ-ENaC, single-channel conductance is 4–5 pS at 150 mM Na+. The Li+ conductance in the presence of 150 mM Li is 9–10 pS in αγ-ENaC (thus similar to αβγ-ENaC) but similar to the Na+ conductance in αβ-ENaC (177). The physiological relevance of dimeric αβ- or αγ-ENaC in native tissues remains uncertain. Several other amiloride-sensitive channels have been described in kidney epithelia (for review, see Ref. 87) which significantly differ from ENaC regarding their biophysical properties. In general, these channels are characterized by a higher conductance in the presence of Na+, a lower Na+ over K+selectivity ratio, and a lower affinity for amiloride. None of these channels has yet been cloned, and therefore, the relationship between these nonselective epithelial Na+ channels and ENaC remains to be established.

A controversial issue concerns the relationship between epithelial Na+ channels reconstituted in lipid bilayers and ENaC. Amiloride-sensitive Na+ channels have been incorporated in lipid bilayers after immunopurification from renal bovine papilla, after in vitro translation of ENaC subunits, or from membrane vesicles of ENaC-expressing Xenopus oocytes (121). The unitary currents of these reconstituted channels showed a 40-pS conductance in the presence of 100 mM Na+ and a low selectivity for Na+ over K+ (11,21, 24, 28, 81,118-123). To account for this difference, it has been proposed that the lipid environment of ENaC in artificial bilayers, in particular the absence of interactions with the cytoskeleton, could modify the biophysical properties of ENaC reconstituted in lipid bilayers (24, 124, 130,274). However, recently it has been shown that ENaC reconstituted in planar lipid bilayers retains its functional characteristics: a 6-pS single-channel Na+ conductance, a slow gating, and a high sensitivity to amiloride (202). Thus it becomes likely that the large-conductance (40 pS) cation channel reconstituted in lipid bilayers by Benos and co-workers (11, 21, 24, 28,81, 118-123) is not ENaC, and the interpretation of the data obtained from these lipid bilayer reconstitution experiments does not apply to ENaC.

2.  ASICs

ASICs are activated by H+ and inactivate rapidly after activation. Typical ASIC current traces are shown in Figure5 B. This transient current carried by Li+ and Na+ displays a Na+/K+ selectivity ratio ≤10. The ASIC single-channel conductance has been determined for several homomeric and heteromeric ASIC channels and ranges from 10 to 15 pS (see Table 1). In contrast to ENaC, ASICs show some permeability to Ca2+. The relative Ca2+permeability is highest in ASIC1a, for whichP Na/P Ca permeability ratios of 2.5 (264) and ∼17 (18,240) have been determined. TheP Na/P Ca ratio for ASIC3 and ASIC1b is >100 (18, 240). The Ca2+ permeability of ASICs remains quite difficult to determine precisely because Ca2+ can either block or potentiate ASIC currents depending on its extracellular concentration and the external pH (18, 40, 60,240). It is possible that the Ca2+permeability of certain ASIC channels might be important for their cellular function.

3.  Other channels

UNC-105 is the first degenerin that has been functionally expressed in a heterologous cell system (84). UNC-105 expressed in Xenopus oocytes is not spontaneously active, but channel activity could be obtained after introduction of activating mutations (mutations P134S or A692V) that cause muscle hypercontraction and degeneration in C. elegans (163,194). The UNC-105 A692V mutant appears to be nonselective for monovalent cations but shows a selectivity for monovalent over divalent cations with a Na+/Ca2+ current ratio of 20 (84). In Xenopus oocytes expressing either one of the two different UNC-105 mutants (P134S or A692V), a channel with a unitary conductance of 4.6 pS for Na+, thus similar to ENaC, was detected (Table 1). However, conductances ranging from 2–30 pS were also recorded that were attributed to the activity of UNC-105 (84). The MEC-4/MEC-10 channel is selective for Na+, but no information is available yet on its single-channel properties.

A current trace of the FMRF-gated channel HaFaNaC is shown in Figure 5 C. HaFaNaC has a low Na+/K+ selectivity with a Na+/K+ permeability ratio ≤10 (157, 272). As for other neurotransmitter-gated channels such as the nicotinic acetylcholine receptor, FaNaC does not need a high cation selectivity to generate an electrical signal in response to the release of neurotransmitters. The unitary conductance of Ha- and HtFaNaC with Na+ as charge carrier is of the order of 7–13 pS (Table 1) (127, 157, 272).

B.  Structure-Function Relationship: The Ion Permeation Pathway

According to the way ENaC discriminates among cations, its selectivity filter likely constitutes the narrowest part of the pore, allowing only small cations to pass through the channel (187). Recent mutagenesis studies support the notion that the selectivity filter of ENaC involves a stretch of three conserved amino acids, G/SxS preceding the putative M2 segment in α-, β-, and γ-subunits (135, 137, 138,223, 231) (Fig.6). Mutations of residue αS589 in this G/SxS sequence allow larger ions such as K+, Rb+, Cs+, NH 4+ , and divalent cations to pass through the channel consistent with an enlargement of the pore at the selectivity filter (135,138). Surprisingly, it was found in these mutants that the larger the substituting amino acid was at position α589, the higher was the channel permeability to larger cations. This finding suggests that the side chain of the residue αS589 points away from the pore lumen, which is likely lined by the main chain of the amino acid residues G/SxS (135). An additional argument for such an orientation of the amino acid lining the selectivity filter is the high NH 4+ permeability in αS589 mutants compared with the smaller cation K+. The high NH 4+ permeability is best explained by the formation of hydrogen bonds between NH 4+ and the carbonyl oxygens of the peptide backbone lining the pore lumen (107, 135). Such a picture of the ENaC narrow selectivity filter lined by the carbonyl oxygens of three amphiphilic residues resembles the basic structure of the K+ channel selectivity filter, suggesting that similar mechanisms are involved in accommodating the permeant cation.

Fig. 6.

Alignment of the pore region. The pre-M2 and the NH2-terminal part of M2 are shown, which make up the outer pore entry and the selectivity filter in ENaC. The location of the degeneration-inducing mutation, the amiloride-binding site, and the selectivity filter are indicated on top. Numbering indicates the rat α-ENaC amino acid position. The shading of the residues in the sequence alignment emphasizes the degree of conservation in each column in the alignment. Black shading indicates 100% conserved, and gray shading indicates 80% or greater conserved. Conservative substitutions [based on the similarity group list of equivalent residues in the multiple sequence alignment editor “genedoc” (http://www.psc.edu/biomed/genedoc)] are considered conserved.

As predicted by a channel pore surrounded by the three homologous ENaC subunits, mutations in the G/SxS selectivity filter sequence in the β- and the γ-ENaC subunit also reduce the Na+/K+ selectivity (138,223, 231). The role of the residues homologous to αS589 for the ionic selectivity of other ENaC/DEG channels has not yet been carefully addressed, but the findings that mutations of these residues disrupt function of DEG-1 (85), MEC-4 (108), and ASIC2a (266) point to their critical roles in channel function.

The contribution of residues other than the conserved G/SxS sequence in the channel ionic selectivity remains to be precisely determined. A mutation of a conserved acidic residue (αD602) in M2 to Cys, Ala, or the basic Lys led to a slight increase in the permeability to K+ (222). The truncation of the cytoplasmic COOH terminus of α-ENaC or mutations of basic residues at the start of the intracellular COOH terminus end have been claimed to increase channel permeability to K+ (128,130), but this remains controversial (135). If real, the changes in ENaC selectivity induced by these mutations in the M2 and cytoplasmic COOH terminus remain quite small compared with the changes obtained with mutations of the αS589 and homologous residues (135, 137, 223,231).

Finally, slight differences have been measured in the Na+/K+ permeability ratio for different ASIC channels (Table 1). The generation of ASIC2a-ASIC2b chimeras and subsequent site-directed mutagenesis identified residues of the region preceding the first transmembrane segment that when mutated affect Na+/K+ selectivity (Fig.7) (49). Recently, an ASIC1a/1b chimera approach provided evidence that this region may play a role in the selectivity toward divalent cations (18).

Fig. 7.

Alignment of the conserved region in the cytoplasmic NH2terminus. Open arrows indicate residues important for ionic selectivity in ASICs (49), black arrows indicate residues important for gating in ENaC (96, 98), and residues marked by a cross are important for MEC-4 function (109). Numbering indicates the rat α-ENaC amino acid position. The shading of the residues in the sequence alignment emphasizes the degree of conservation, as detailed in the legend to Fig. 6.

The crystal structure of the KcsA showed that the narrow channel selectivity filter is the site for intimate interactions between the permeating ion and the residues lining the channel pore. In K+ channels, the oxygens of the carbonyl groups of the polypeptide backbone at the selectivity filter act as surrogate water molecules for the dehydrated ion in the pore (64,178, 273). The interaction between the ion and these pore residues is essential for ion dehydration and critically depends on the distances between the permeating ion and the oxygen atoms. Similarly, changes in the ENaC pore geometry by mutations at position αS589 are expected to perturb these interactions between Na+ or Li+ and the pore with functional consequences on the ion conduction. Indeed, the αS589 mutations that cause the highest increase in channel permeability to large cations decrease at the same time the Na+ and Li+single-channel conductance making the channel pore less hospitable for these smaller ions (135, 137). Other mutations within the G/SxS selectivity filter decreased ion conductance. The mutations αG587S, βG529S, and αS589D (Figs. 6and 8 B) decreased both Na+ and Li+ conductance, mutations γS541A and βG529A decreased selectively the channel conductance for Li+, whereas the αS588I mutation increased the single-channel conductance for Na+ (138). These alterations in single-channel conductance were associated with changes in channel affinity for the permeant ion (138). For instance, the low permeability of the γS541A ENaC mutant to Li+ is associated with a greater than fourfold increase in the channel affinity for Li+essentially due to a lower rate of dissociation from its binding site. In this mutant, Li+ appear to stick to the pore, blocking the Na+ current if both Na+ and Li+are present (138). The effects of the mutations in the conserved G/SxS sequence on channel conductance and ionic selectivity indicate that this region of the channel pore is essential for the ion recognition and for its permeation through the channel.

Fig. 8.

Model of the pore region of the KcsA K+ channel and of ENaC. A: the model is based on the crystal structure of the KcsA K+ channel and shows the pore region (64). Two of the four subunits are shown. The second transmembrane domains, the “inner helices,” are shown in black; the pore helices are in gray; and the selectivity filter GYG is indicated. B: the model of the ENaC pore is based on functional data of wild-type and mutant ENaC. The pre-M2 segments and the second putative transmembrane α-helices (M2) of two of the four ENaC subunits are shown. Amino acid numbering of the α-subunit is indicated. In contrast to KcsA subunits which contain only ∼30 residues between the two transmembrane segments M1 and M2, the extracellular portion of ENaC subunits represents >50% of the subunit mass and extends much farther outward than shown in the figure. The outer vestibule is lined by the pre-M2 segments with the site where amiloride binds to αS583 and the corresponding Glyresidues in β and γ (220). The vestibule narrows down to the selectivity filter formed by αG587, βG529, γS541, and the ring of Ser residues (αS589 and homologous residues in β and γ) (137, 138, 223,231). In this model the M2 domain forms the intracellular part of the pore, in analogy to K+ channels.

Other residues located outside the selectivity filter, αW582, as well as the residues βG525 and γG537 that form the amiloride binding site have been reported to change channel conductance for Na+ and Li+ (220,223). These observations are consistent with close interactions between the amino acid residues lining the outer channel pore and the permeant Na+ or Li+. These interactions near the amiloride binding site might be involved in ion dehydration before entering the narrow selectivity filter.

C.  Pore Blockers and Putative Binding Sites

1.  Pore blockers

Amiloride is a widely used K+-sparing diuretic that was identified during an extensive screening of some 25,000 compounds susceptible to enhance renal Na+ excretion and diminish K+ excretion. Amiloride was found to block the electrogenic Na+ transport stimulated by aldosterone in the distal nephron by decreasing the apical Na+ conductance in tight epithelia. The resolution of the apical Na+ currents at the single-channel level in cortical collecting duct cells directly demonstrated the inhibitory effect of amiloride on ENaC activity by a substantial reduction of the mean open time without changes in the single-channel conductance (104, 190).

Amiloride block has been extensively analyzed in the native epithelial Na+ channel (reviewed in Refs. 87, 143). ENaC is blocked by submicromolar concentrations of amiloride. Amiloride is a weak base, and the pH dependence of the channel block suggested that the ionized form of amiloride is effective in blocking the channel (see references in Ref. 87). Consistent with this notion, ENaC block by amiloride depends on the transmembrane potential difference in such a way that hyperpolarization of the membrane increases channel block (see references in Ref. 87 and see Ref. 177). This voltage dependence of channel block is usually interpreted as the effect of the transmembrane potential on the blocker that pulls the charged molecule, i.e., the cationic form of amiloride, toward its binding site located within the transmembrane electric field. This voltage dependence of channel inhibition is typically observed for pore blockers that act by a simple physical occlusion of the pore. Quantitative analysis indicates that the charged guanidinium moiety experiences 15–30% of the transmembrane electrical field when the molecule is bound to its blocking site. As expected for a pore blocker, amiloride interacts with permeating Na+ and Li+. Raising the extracellular concentration of Na+ reduces the channel affinity for amiloride. Experiments in the toad bladder convincingly showed that the interaction between the permeant Na+ and amiloride was competitive, suggesting overlapping binding sites within the external ion permeation pathway (189). In other epithelia, experiments addressing this issue did not allow a clear demonstration of a strict competitive interaction between the permeating Na+ and amiloride (reviewed in Ref. 87).

Both the charged guanidinium group and the pyrazine ring of amiloride are necessary for high-affinity binding. Guanidinium itself and pyrazine analogs such as 6-chloro-3,5-diaminopyrazine-2-carboxamide (CDPC) can block the channel, with apparent affinities of ∼100 mM for guanidinium and 100 μM for CDPC. Addition of a benzyl or a phenyl group to the guanidinium side chain of amiloride (e.g., benzamil or phenamil) increases the affinity of block by a factor of ∼10 for the epithelial Na+ channel (143). This suggests that hydrophobic interactions with a part of the channel or the membrane very close to the amiloride binding site can stabilize the bound state of the drug.

A variety of organic and inorganic cations block ENaC with low affinity from the extracellular side in a voltage-dependent manner (187). Their ability to block ENaC critically depends on their size, namely, a molecular diameter lower than 5 Å. The voltage dependence of the block by inorganic cations is steeper than the block by amiloride or other organic cations, suggesting that the smaller inorganic cations penetrate deeper into the channel pore. From the analysis of the external channel block by cations it was proposed that the pore progressively narrows down to a binding site that accepts smaller inorganic cations and finally to the narrowest part of the pore, the selectivity filter, that accommodates exclusively the permeating ions H+, Li+, and Na+(187).

All ENaC/DEG channels other than ENaC require about 100-fold higher amiloride concentrations for channel block, except for Mec-4/MEC-10 channels whose sensitivity to amiloride block is between that of ENaC and that of other ENaC/DEG family members (Table 1). It is not clear whether the mechanism of channel block by amiloride is common to all ENaC/DEG channels. It should be remembered that at concentrations in the higher micromolar range amiloride blocks many other ion channels and transporters that are unrelated to ENaC, likely by different molecular mechanisms (143).

No specific pharmacological blockers of ASICs are available so far except for a toxin from tarantula venom that has been found to specifically block homomultimeric ASIC1a with high affinity (IC50 = 0.9 nM) (72). Psalmotoxin 1 (PcTX1) is a 40-amino acid toxin from the venom of the South American tarantula Psalmopoeus cambridgei. PcTX1 has a limited overall homology to other spider venom toxins identified to date. PcTX1 blocks rapidly and reversibly recombinant homomeric ASIC1a only, but not the homomultimers ASIC1b, ASIC2a, ASIC3a, or the heteromers ASIC1a/ASIC2a or ASIC1a/ASIC3a, nor αβγ-ENaC or K+channels.

ASIC currents are inhibited by divalent and trivalent cations. Ca2+ inhibit Na+ currents in several different ASIC multimers with an apparent IC50 in the millimolar range (38, 40, 60,240, 264). As mentioned earlier, the interactions of ASICs with Ca2+ that can also permeate ASICs are complex and not completely understood. Recently, it was shown that lactic acid, which is produced by anaerobic metabolism during ischemia, enhances ASIC function by chelating divalent cations that in the absence of lactate partially inhibit ASIC function (113). This mechanism may be of physiological importance making ASICs extra sensitive to lactic acidosis of local ischemia.

In mutant UNC-105(A692V) channels, Na+ currents are blocked by Ca2+ and Mg2+ with an IC50 of 25 mM. Gd3+ is a blocker of mechanosensitive channels with an IC50 of the order of 10–100 μM (103). Currents through RPK/dGNaC1 channels are blocked by Gd3+with an IC50 of 100 μM (3). ASIC3 homomultimers and ASIC2a/ASIC3 heteromers are also sensitive to Gd3+ with an IC50 ∼40 μM compared with >1 mM for ASIC2a homomeric channels (12). The sensitivity of ASICs and RPK channels to Gd3+ does by no means indicate that these channels are mechanosensitive.

2.  The amiloride binding site

A mutagenesis screen of the amino acid residues preceding the M2 transmembrane segment of α-, β-, and γ-ENaC identified residues in the β (βG525) and at the homologous position of γ-ENaC (γG537), each of which when mutated reduces the apparent channel affinity to amiloride by ∼1,000-fold (see Figs. 6 and 8 B) (220). The homologous αS583C mutation reduces the affinity to amiloride by ∼20-fold. Because amiloride acts as a pore blocker, these observations demonstrated for the first time that the pre-M2 segments of ENaC subunits form the outer pore and that all three subunits line the ENaC pore like barrel staves (188). Introduction of Cys residues at position S583 of α-ENaC generates a high-affinity binding site for zinc ions that also block the channel in a way very similar to amiloride (220). According to the α-β-α-γ subunit arrangement around the channel pore in which the two α-subunits face each other (76, Fig.4 A) and given the optimal distance of two sulfhydryl groups for coordinated ligation of Zn2+ (149), the diameter of the pore at position αS583 can be estimated to be ∼5 Å. Amiloride, which has a diameter of 4–5 Å, would fit snugly into the outer channel pore.

The 1,000-fold change in channel affinity for amiloride caused by the mutations βG525 or γG537 suggests that these residues are directly involved in binding interactions with amiloride. The observation that sulfhydryl reagents such as MTSEA or Zn2+ are able to reproduce channel block when binding to Cys residues introduced in the amiloride binding site at corresponding αS583, βG525, or γG537 positions in the ENaC subunits strongly supports that these residues are part of a blocking site in the channel pore with affinity for amiloride (146, 220, 229). The ENaC/DEG channels exhibit a wide range of affinities for amiloride (Table 1), and comparison of their primary sequences indicate that the Gly residues β525 and γ537 that are crucial for the high affinity of ENaC for amiloride are completely conserved except for the α- and δ-ENaC subunits which contain at the corresponding position a Ser (αS583). The δ-ENaC subunit can substitute the α-subunit to make functional channels (265). This combination of the Ser residue on α (αS583) and the Gly residues on β and γ (β525 and γ537) within the amiloride binding site likely represents the structural basis for the unique high affinity of ENaC for amiloride block. Consistent with this hypothesis it was found that mutation of αS583 to Gly results in a ∼20-fold decrease of the apparent amiloride affinity (S. Kellenberger, I. Gautschi, and L. Schild, unpublished data).

The short amino acid segment WYRFHY NH2-terminal of CRDII in the α-ENaC extracellular loop was proposed to participate in amiloride binding interactions with ENaC and therefore with channel block. This putative amiloride binding site was identified based on sequence homology with an antibody that recognizes limited regions of amiloride. The homology between ENaC and the antibody is rather low within the WYRFHY segment and involves only the two Tyr and the His residues (125, 141). The WYRFHY deletion mutant (αΔ278–283) coexpressed in Xenopus oocytes with wild-type β- and γ-ENaC subunits did not express enough current to determine the apparent amiloride affinity (125,138). A His to Asp substitution within the WYRFHY sequence was found to decrease amiloride affinity in ENaC reconstituted in lipid bilayers (125), but the mutant channel coexpressing αH282D with wild-type β- and γ-ENaC in Xenopusoocytes has the same affinity for amiloride as ENaC wild-type (138). Thus the αWYRFHY sequence in the context of αβγ-ENaC does not represent an amiloride binding site involved in channel block. It should be mentioned that in all other cases of amino acid substitutions reported to cause channel resistance to amiloride that were based on experiments in lipid bilayers, not a single one was confirmed on ENaC expressed in Xenopus oocytes (81, 121, 125). Thus the available experimental data indicate that amiloride occludes the external channel pore by binding to a high-affinity site constituted by αS583 and the conserved Gly residues of β- and γ-ENaC subunits (βG525, γG537) located in the pore domain preceding the putative M2 segment (Fig. 8 B).

In two of the cloned FaNaC orthologs amiloride has opposite effects. While it blocks HaFaNaC (of Helix aspersa), it potentiates FMRFamide responses in HtFaNaC (ofHexisoma trivolvis). With the use of a chimeric approach, a region of ∼120 amino acids in the NH2-terminal part of the extracellular loop was identified which determines whether amiloride acts as agonist or antagonist on FaNaC channels (52). This region is also critical for the concentration dependence of the channel activation by FMRFamide. It is interesting to note in this context that the guanidinium compound benzimidazolyl-guanidine (BIG) can potentiate ENaC in the frog skin by abolishing the channel self-inhibition by external ions (271).

PcTX1 specifically inhibits ASIC1a channels, but the mechanism of channel inhibition remains unclear. The splice variant ASIC1b is not sensitive to PcTX1 block (72) and differs from the ASIC1a essentially in the intracellular NH2 terminus, the transmembrane segment M1, and the first 110 residues of the extracellular loop. Considering that the toxin acts from the extracellular side it was proposed that the PcTX1 binding site in ASIC1a is located in the NH2-terminal 110 amino acids of the extracellular loop.

D.  A Model of the Channel Pore

The only available high-resolution structure of a highly selective cation channel is that of the KcsA K+ channel. Certain comparisons with the possible structure of ENaC can be made, although there is no homology between the two proteins. A schematic view of the KcsA pore region is shown in Figure 8 A. As ENaC, KcsA contains two transmembrane regions. These are linked on their extracellular ends by the ∼30 amino acids forming the pore region, whereas the NH2 and COOH termini are intracellular. The KcsA is a tetramer, like all other known K+ channels, and the subunits are related by a fourfold axis of molecular symmetry around the central ion-conducting pore. The architecture of the channel pore is primarily defined by four inner helices formed by the second transmembrane region M2 (black helices in Fig. 8 A). The inner helices are tilted with respect to the membrane normal by ∼25° and are slightly kinked so that the subunits open like the petals of a flower facing the outside of the cell. The open petals house the structure formed by the pore region near the extracellular surface of the membrane. The narrow selectivity filter that is located in this region is only 12 Å long, whereas the remainder of the pore is wider and lined with hydrophobic amino acids of M2. The pore entrance and selectivity filter are formed by the ∼30 amino acids between the two transmembrane regions. For each subunit, in going from the NH2 to the COOH terminus, this comprises a small extracellular structure called the “turret” and the pore helix (gray in Fig. 8 A) which slots between the inner helices and points to the intracellular end of the selectivity filter. The pore helix is followed by the selectivity filter GYG that is separated from M2 by a linker of approximately seven residues. K+permeation is thought to depend critically on a tight match between the size of the ion and the geometry of the pore at the selectivity filter.

Comparison of ENaC with KcsA is presently limited to the outer pore region, since the structures that constitute the internal pore beyond the selectivity filter have not been clearly identified so far in ENaC. Similarities between ENaC and the KcsA channel include the location of the narrow selectivity filter toward the extracellular solution, the narrow selectivity filter consisting of three conserved residues in each subunit, and the apparent orientation of the side chains of these residues away from the pore lumen. However, clear differences exist between ENaC and KcsA in particular regarding the topology of M2 relative to the selectivity filter and the external binding sites for blockers. In KcsA the sequence preceding the GYG selectivity filter dips into the membrane, and the GYG sequence is oriented with the COOH-terminal end toward the extracellular side, then a seven-amino acid linker separates the selectivity filter from the M2 transmembrane segment helix (Figs. 6 and 8) (64). In ENaC there is no linker between the selectivity filter sequence αG587-S589 and the start of M2 segment (138,223). The sequence preceding the ENaC selectivity filter is the site for binding of large blocking molecules such as amiloride that enter the pore from the extracellular solution. A model based on the KcsA channel structure would locate the amiloride binding site deep into the channel pore beyond the narrow selectivity filter. More likely, the ENaC pre-M2 segment extends from the NH2terminus of M2 toward the extracellular side, which is supported by the observation that a Cys introduced at position G536 in human γ-ENaC within the amiloride binding site (corresponding to αS583) is accessible to the hydrophilic sulfhydryl reagent MTSET from the extracellular side (231). Thus the six amino acid residues of the pre-M2 and the very first residue of the putative M2 segment (Figs. 6 and 8 B) likely represent the outer ENaC pore that narrows from the amiloride binding site down to the ion selectivity filter.

Within families of ion channels such as the voltage-gated or the ligand-gated ion channels, there is a remarkable homology in the structure of the ion permeation pathway (160). Furthermore, within the large family of K+ channels, the pore structure is not only conserved, but basic physicochemical principles of K+ permeation are also similar. So far, the pore of ENaC has been the best characterized among the members of the ENaC/DEG channel family, and it is likely that the basic structural features of the channel pore are conserved within this family. However, a possible involvement of a region in the pre-M1 domain in ion selectivity has been described in ASICs but not ENaC (18,49, 98) (see sect.v B). In addition, involvement of the M1 segment of FaNaC was proposed in a recent study that applied the substituted cysteine accessibility method (SCAM) (197). Cysteine residues introduced at seven different positions in M1 were accessible to modification by externally applied sulfhydryl reagents that inhibited 50–60% of the FaNaC current. Inhibition was prevented if the sulfhydryl reagents were applied in the presence of amiloride. These findings were interpreted as the demonstration that M1 of FaNaC lines the channel pore. However, channel inhibition by hydrophilic sulfhydryl reagents can occur by different mechanisms, as, e.g., effects on gating or on ion permeation. Thus the present evidence for a M1 segment lining the external channel pore remains quite weak, and more detailed functional and structural studies are necessary to draw a clearer picture of the role of the intracellular NH2terminus in ASICs and the M1 in FaNaC in ion conduction.


A.  Channel Activation

Channel gating is probably the most variable function among the members of the ENaC/DEG family: some channels open in response to mechanical stimuli, others in response to external ligands, and some are constitutively active. Mutations in both intracellular and extracellular domains affect channel gating.

1.  ENaC

ENaC is constitutively active in the apical membrane of tight epithelia. Single ENaC recordings in the cortical collecting tubule show slow kinetics with open and closed dwell times in the range of seconds at room temperature (190). Similar gating properties are observed in oocytes expressing αβγ-ENaC shown in Figure 3 A. The systematic analysis of ENaC single-channel gating kinetics reveals that the open probability (P o) of the channel varies from <0.1 to 0.9 and that active ENaC channels can be divided basically into two groups: one with a low P o and long closed times and the second with high P o and long open times. These variations in the ENaC P o likely reflect different channel gating modes (193). Presently it is not clear how these gating modes are influenced by extracellular and intracellular factors. For instance, external Na+ decreases channel activity by a mechanism termed “self-inhibition.” The physiological role of this Na+-induced channel inhibition is not clear, but it could play a role in maintaining a homogeneous Na+ reabsorption along the nephron (87,184). Intracellular pH, Ca2+ (35,191), or Na+ exert a feedback inhibition on the channel (1, 62, 80,136, 191, 227). Finally, enzymes such as serine proteases or channel activating protein (CAP)-1 that are expressed in tight epithelia stimulate channel activity when coexpressed with ENaC in Xenopus oocytes (254,262). This activation of ENaC is also observed inXenopus oocytes after addition of a low concentration (2 μg/ml) of trypsin to the extracellular side (46). The physiological relevance of the serine proteases in ENaC regulation is still unknown.

2.  ASICs

In general, ASICs display a transient peak current activated by protons that lasts hundreds of milliseconds to seconds, followed by a channel desensitization despite the continued presence of a low external pH solution (Fig. 5 B). The pH of half-maximal activation of ASIC currents depends on the channel type. Transient currents carried by homomeric channels formed by ASIC1a, -1b, or -3 show pH values of half-maximal activation (pH50 activation) of the range of 5.9–6.5. ASIC2 is less sensitive to pH changes and displays a pH50 activation of 4.4. It has been shown for ASIC1a (264) and ASIC2a (38) that only extracellular and not intracellular protons activate the channel.

Desensitization varies among the different ASIC channels: for instance, ASIC1 and ASIC3 desensitize faster than ASIC2 (Fig. 5 B). Under certain conditions, inactivation of the current carried by some ASIC multimers is incomplete, leaving a sustained residual inward current that follows the fast transient current. This sustained current is found during prolonged acidification in cells expressing ASIC3 homomultimers or the heteromers ASIC2a/ASIC2b, ASIC2a/ASIC3, and ASIC2b/ASIC3 (12, 13, 61,158, 263, 267). Surprisingly, the sustained currents in ASIC-expressing cells exhibit biophysical and pharmacological characteristics different from the transient currents (25, 263). The reasons for these functional differences are not clear. Transient and sustained currents might involve distinct ASIC channel types, or the sustained currents could represent leakage currents due to ASIC overexpression in the cell. Finally, it remains possible that ASICs change their biophysical and pharmacological characteristics during activity as documented for the purinergic P2X7 (239) and for neuronal P2X receptors (140, 258) and for K+ channels undergoing C-type inactivation (234).

From the rather nonphysiological pH for half-maximal activation of some ASICs, it has been hypothesized that endogenous ligands might shift channel activation toward more physiological pH values. Zn2+ at physiological concentrations potentiates the acid activation of homomeric or heteromeric ASIC2a channels by shifting the pH dependence of activation to higher values (16). Although ASICs are not the mammalian homologs of FaNaC, FMRFamide and related neuropeptides have been found to modulate ASIC currents (8). FMRFamide is a neurotransmitter in invertebrates only, but the related neuropeptides FF or AF are expressed in the mammalian CNS (8). Administration of FMRFamide to mammals induces a variety of physiological effects, including alterations in blood pressure, respiratory rate, glucose-stimulated insulin release, pain perception, and behavior, some of which are mediated through opioid receptors (reviewed in Ref.8). Coapplication of FMRFamide at micromolar concentrations with protons slows inactivation of the proton-induced current and induces a noninactivating inward current. This sustained current at the end of a 30-s application is on the order of 10% of the peak current. A similar although considerably smaller effect could be induced by application of the mammalian peptide neuropeptide FF. This modulation of ASIC activity by neuropeptides might be of physiological relevance in pain sensation (25, 267). Recently, it was shown that FMRFamide or neuropeptide FF potentiate the acid-induced peak current and slow the inactivation of ASIC2A/3 heteromers expressed in Xenopus oocytes (34).

3.  FaNaC

Extracellular application of the peptide FMRFamide to cells expressing FaNaC induces a fast inward current that displays a slow and partial desensitization in the presence of the agonist (see Fig.5 C). The EC50 for FMRFamide-induced activation depends on the FaNaC ortholog and is 2–70 μM (157, 272). The agonist specificity for channel activation was recently investigated inHaFaNaC expressed in Xenopus oocytes, and several agonists have been identified (51, see Table 1 therein and Ref. 157). HaFaNaC was the first FaNaC to be cloned and is the best studied so far. The single-channel kinetics in the presence of the activating ligands FMRFamide or FLRFamide are complex and involve transitions between several open and closed states. The dwell times are of the order of hundreds of microseconds to several milliseconds (272). At higher concentrations of the activating peptide the number of long openings increases. Based on this kinetic analysis and a Hill coefficient of activation of >1, it was suggested thatHaFaNaC binds two agonist molecules for full activation of the channel (157, 272). At concentrations ≥30 μM, the agonists FMRFamide and FLRFamide block the channel (93).

4.  Degenerins

The molecular mechanisms underlying channel opening in response to mechanical stimuli remain poorly understood for the degenerins and likely involve complex interactions between the different proteins that form the mechanotransducing complex (Fig. 4 B). Mechanosensitivity of UNC-105 alone or MEC-4/MEC-10 together with MEC-2 expressed in heterologous cell systems could not be demonstrated, suggesting that other associated or interacting proteins are necessary for this function. While wild-type MEC-4/MEC-10/MEC-2 expressed inXenopus oocytes displays amiloride-sensitive current, UNC-105 requires activating mutations to induce channel openings (84, 90).

B.  Gating Domains

The generation of ENaC/DEG channel mutations of conserved amino acid residues changing the channel openings and closures identified specific regions in the extracellular loop and in the cytoplasmic NH2 terminus that control channel gating.

1.  Extracellular gating domains

The first evidence for the contribution of a site in the extracellular loop to channel gating came from the identification of mutations in the pre-M2 region of the degenerin MEC-4 that cause degeneration of touch receptor neurons in C. elegans (Fig.3 B). The morphological features of this degenerative phenotype are consistent with an abnormal leak of cations into the neuronal cell due to a serious channel dysfunction consisting presumably in abnormal openings and/or inefficient closures of the touch receptor channel complex. The mec-4 alleles causing the dominant degenerative phenotype encode substitutions of the conserved residue A713 in the pre-M2 region (Fig. 6) by large side chain amino acids, whereas A713 substitutions by small side chain amino acids (Ala, Ser, Cys) are without effects on the cell (65). It was hypothesized that the presence of a bulky side chain at position A713 within an extracellular gate may prevent the channel from closing efficiently.

Subsequently, it was shown for most ENaC/DEG family members that this conserved residue plays an important role in channel gating. Substitution of the residue homologous to MEC-4 A713 in ASIC2a, G430, by large residues results in constitutively active channels (4, 266). In addition to the constitutive currents in the ASIC2aG430 mutants, H+-gated currents can still be activated by lowering of the pH (38). The H+-gated current in G430 mutants differs from the wild-type current in two aspects. Inactivation of the H+-activated ASIC2a current is slowed in Ser or Cys G430 substitution mutants (which do not display a constitutive current) and is completely disrupted by the larger substitutions Thr and Phe (38). These mutations also shift the pH dependence of activation to less acidic pH values. The size of the pH shift and the degree of disruption of inactivation correlate with the ability of the substitution to induce a constitutive current, indicating that these different aspects of channel function related to H+ gating are linked. Thus the G430 mutation seems to favor the active conformation of the channel and prevent the channel from relaxing back to an inactivated or resting state. The role of G430 in ASIC2a in gating is further supported by the observation that this residue undergoes conformational changes during channel activation (4, 6). Covalent modification of aCys residue at position G430 in the ASIC2a G430C mutant by hydrophilic MTS sulfhydryl reagents is possible only when the channel is activated by protons. Thus, in the activated state, the C430 is accessible to extracellular MTS reagents leading to a noninactivating current that persists even after switching the pH back to resting pH (4). This G430C mutant can also be activated by Zn2+ at pH values at which the channel is normally closed, suggesting that the engineered Cys at the degenerin site may be accessible to small ligands such as Zn2+ independently of the conformational state (6). Alternatively, Zn2+ which is a known coactivator of ASICs may react with endogenous His residues and induce a conformational change that exposes the Cys to the extracellular solution (16). The mutation of the homologous residue in ASIC2b, G481 to Phe was characterized in ASIC2a/2b heteromers, because ASIC2b does not form functional channels by itself (251). This mutation completely disrupted inactivation but did not change any other functional parameters. Introduction of a cysteine residue at the position corresponding to G430 in α- or β-ENaC subunits and its covalent modification by sulfhydryl reagents induces an increase in the channelP o (228). As for ASIC, the cysteine modification leading to channel openings is observed when the channel is in the open conformation. Thus the conserved extracellular “degenerin” residue is tightly involved in opening and closing of ENaC/DEG channels.

Another extracellular regulatory domain named “ERD” was found between the cysteine-rich domains CRDI and CRDII in a region unique to the degenerins (85). Activating mutations in the ERD domain were found in DEG-1, MEC-4, and UNC-8. The mec-4alleles harboring a missense mutation (A404T) or a small deletion (Δ399–407) in the ERD induced neuronal degeneration in C. elegans (85). Similarly to the degenerin site at the pore entrance, touch cell death also resulted from the substitution of large-side-chain amino acids (Thr, Leu, or Lys) but not small-side-chain amino acids (Gly or Cys) for Ala in MEC-4.

The post-M1 (pM1) domain is highly conserved among all the ENaC/DEG family members (Fig. 3) (84). A mutation that causes muscle hypercontraction consistent with a channel gain-of-function mutation has been identified in this region of the degenerin UNC-105P134S (163). Expression of the UNC-105 P134S mutant in Xenopus oocytes confirmed changes in the channelP o and gating kinetics (84). There is currently no evidence indicating that the post-M1 region also contributes to gating of ASIC or ENaC channels. A loss-of-function mutation causing PHA-I was found in this post-M1 region of α-ENaC. So far, there is no evidence for an effect of this Cys substitution on channel gating (see sect. vii) (77, 95). Finally, inactivating mutations were found in CRDIII and in the pre-M2/M2 domain (85,108, 109). It is unclear how these mutations can be strictly assigned to defects in channel gating, or whether they impair channel function by affecting the biosynthesis, maturation, or targeting of the channel complex to the cell surface.

Zinc is a coactivator of ASICs. Two His residues have been identified that are essential for the activating effect of Zn2+ in ASIC: one in the CRDII and one directly COOH terminal of CRDIII (Fig.3) (16). In the course of this analysis, a His residue was identified that abolishes the pH sensitivity of ASIC2a when mutated. This residue is located directly COOH terminal of M1 and seems to be an essential component of the extracellular pH sensor.

2.  An intracellular gating domain

A loss-of-function mutation in the highly conserved HG motif (Figs. 3 and 7) preceding the first transmembrane segment of β-ENaC was found to be associated with the renal salt-losing syndrome PHA-I (see sect. iii A) (39). When expressed in Xenopus oocytes, the βG37S mutation causing PHA-I as well as mutations of the corresponding Gly residues in α- or γ-ENaC subunits result in an important reduction in channel P o (39, 96,98). Surface expression of the channel is not affected by these Gly to Ser mutants. Thus these mutations point to a role of the intracellular pre-M1 segment in channel gating. Mutations of the conserved neighboring residues of the homologous Gly residue in α-ENaC (G95), H94 and R98, resulted in a similar channel inhibition (Fig. 7) (98). It remains to be elucidated how these residues control channel gating. Interactions between subunits, with intracellular proteins or with ligands, and phosphorylation events are possibilities that remain to be experimentally addressed. Biochemical evidence suggests that an NH2-terminal fragment of the γ-ENaC subunit may interact with the α-subunit and modulate channel activity (5).

Recently it was demonstrated that mutations in the homologous region of MEC-4, T91I, S92F, or G95E (Fig. 7) disrupt touch sensation in C. elegans (109). In addition, it was shown that overexpression of a peptide corresponding to the intracellular NH2 terminus of MEC-4 dramatically increased the fraction of touch-insensitive animals in the wild-type background and that it caused a significant reduction in the number of degenerating touch neurons in transgenic animals expressing the A713V degenerin mutation. This experiment thus suggests that the NH2terminus of MEC-4 is involved in protein-protein interactions. If the NH2-terminal peptide contained either of the mutations T91I, S92F, or G95E, the interference with MEC-4 was almost completely abolished, indicating that this short conserved domain is critical for the interaction (109).


The regulation of channel activity at the cell surface by intracellular messengers or by hormone-induced cascades of cellular events is well established for ENaC, but not for ASICs or other ENaC/DEG family members, and it is unknown whether at all and how the activity at the cell surface of these other ENaC/DEG channels might be regulated.

A.  Transcriptional and Posttranscriptional Regulation

The excretion of sodium by the kidney has to balance the daily salt intake. The site of fine regulation in the kidney is the distal nephron where Na+ absorption from the urine via ENaC is under the control of hormones. In response to a decrease in extracellular fluid volume or to hyponatremia the renin-angiotensin-aldosterone cascade is activated leading to an increase in Na+ reabsorption in the distal nephron (257). Aldosterone binds to a cytosolic receptor, and this hormone-receptor complex is translocated to the nucleus where transcriptional events occur leading to an increase in ENaC mRNAs and other aldosterone-induced transcripts (AITs) together with a decrease of the level of aldosterone-repressed transcripts (ARTs) (Fig. 2) (167, 168, 172,174, 206, 237). The AITs and ARTs are key elements for the upregulation of ENaC at the apical membrane and of the Na+-K+-ATPase at the basolateral membrane by aldosterone allowing the stimulation of a transepithelial Na+ flux. Recently, it was shown that aldosterone rapidly increases the mRNA level of the serine/threonine kinase “serum- and glucocorticoid-regulated kinase” (SGK), and it was demonstrated that SGK stimulates ENaC when coexpressed inXenopus oocytes (42, 59,180, 195). It has been clearly demonstrated that the early effect of aldosterone consists in a marked redistribution of ENaC from a cytoplasmic pool to the apical membrane (166-168, 172). This increase in ENaC channel density at the cell surface is thought to contribute essentially to the early response to aldosterone which increases Na+ transport within 1–3 h of stimulation by the hormone.

Vasopressin (AVP) increases both water and Na+ absorption in the distal nephron (79, 115,181, 182, 217). Binding of AVP to AVP receptors (V2 receptors) at the basolateral membrane of principal cells causes a redistribution of aquaporins (AQP-2) from an intracellular compartment to the apical membrane leading to an increase in cell membrane water permeability. In parallel, the stimulation of the V2 receptors increases ENaC activity at the apical membrane that accounts for the enhanced Na+ absorption. Both AVP effects on AQP-2 and ENaC are mediated by cAMP and can be mimicked by cAMP analogs or forskolin (115).

B.  Cell Surface Regulation

The specific targets on ENaC for the regulation of channel activity by aldosterone or AVP likely reside in the intracellular NH2 and/or COOH termini. The COOH termini of αβγ-ENaC subunits have a low homology except for a conserved proline-rich motif that includes three adjacent Pro residues separated by one residue from a downstream Tyr, the PPPxY or “PY” motif. This PY motif is found exclusively on the ENaC subfamily members. This motif on β- and γ-subunits is mutated or deleted in a rare form of monogenic hypertension, Liddle's syndrome (105, 106,114, 152, 218). Missense or deletion mutations within the PY motif of β- or γ-ENaC result in a retention of hyperactive channels at the cell surface after expression in Xenopus oocytes, suggesting that the conserved Pro and Tyr are involved in the internalization of ENaC at the cell surface (78, 136, 226).

The amino acid sequence encompassing the PY motif in α-, β-, and γ-ENaC shows two adjacent endocytic motifs, NPxY and YxxL, that are recognized by the clathrin:Ap-2 complex (226,232). These endocytic motifs could play a role in removing ENaC from the plasma membrane via clathrin-coated pits. Dynamin is required for clathrin-mediated endocytosis. Expression of catalytically inactive dynamin mutants with ENaC results in an increase in ENaC activity at the cell surface of similar magnitude as that obtained by mutations in the ENaC PY motif (226). Other investigations followed the hypothesis that the PY motif is involved in specific protein-protein interactions involving binding of the cytoplasmic PY motif to conserved Trp-Trp (WW) domains. The WW domains are protein-protein interaction modules that bind to proline-rich sequences. With the use of a yeast two-hybrid screen with a PY motif as bait for a binding partner of ENaC, neuronal precursor developmentally downregulated 4 (Nedd4) was shown to bind to PY motifs of β- and γ-ENaC and with a lower affinity to the PY motif of α-ENaC (235). Recently, the structure of the binding complex between a Nedd4 WW domain and the ENaC PY motif has been solved by NMR spectroscopy (133).

Nedd4 is a ubiquitin protein ligase with conserved WW domains that specifically interact with PY motifs. Ubiquitin ligase is responsible for attachment of ubiquitin chains to Lys residues of the target protein as a tag for rapid protein degradation. Evidence that Nedd4 regulates channel activity comes from functional coexpression of ENaC together with Nedd4 in Xenopus oocytes (2). The ENaC-Nedd4 coexpression results in an almost complete inhibition of channel activity at the cell surface due to a corresponding decrease in the number of channel molecules at the plasma membrane of oocytes. Conversely, channel activity was increased and active channels were retained at the cell surface when ENaC was coexpressed with a catalytically inactive Nedd4 mutant. This Nedd4 mutant is still able to bind PY motifs of ENaC but unable to ubiquitinate the channel, and it therefore competes with the endogenous Nedd4 of the Xenopusoocyte (2). These findings are consistent with a regulation of ENaC at the cell surface that involves binding of Nedd4 ubiquitin ligase to the PY motifs of β- and γ-ENaC leading to channel ubiquitination, endocytosis, and degradation via the proteasome and/or the endosomal pathway. It should be pointed out that clathrin-mediated endocytosis and ubiquitination are not mutually exclusive mechanisms and could both coregulate ENaC activity at the cell surface.

The specificity of the interaction between the PY motifs in α-, β-, and γ-ENaC subunits and protein partners in different tissues where ENaC is expressed needs further investigation. Different homologs of Nedd4 have been identified with different ability to interact with ENaC and to modulate its activity in Xenopus oocytes (132). Finally, the contribution of the PY motifs in the control of ENaC activity at the cell surface by aldosterone or AVP remains to be established. When coexpressed in Xenopusoocytes, the aldosterone-induced kinase Sgk can phosphorylate and inactivate Nedd4 leading to a retention of ENaC at the cell surface (58). The relevance of this signaling pathway in aldosterone-responding cells remains to be demonstrated.

Further support for a role of Nedd4 in the regulation of ENaC at the cell surface comes from the fact that ENaC is ubiquitinated inXenopus oocytes. The ubiquitination sites involve the Lys residues in the NH2 terminus of α- and γ-ENaC. Expression of ubiquitination-deficient ENaC mutants by substitution of Lys to Arg in the NH2 terminus of α- and γ-ENaC considerably decreased channel ubiquitination and increased the number of active channels at the cell surface (235,236). Finally, the PY motif on ENaC might be involved in feedback regulation of the channel activity by intracellular Na+. Wild-type ENaC is downregulated by intracellular Na+. Liddle mutants have a considerably lower sensitivity to inhibition by intracellular Na+ at least when expressed in Xenopus oocytes (136). How intracellular Na+ may activate the ENaC-Nedd4 interaction is not known.

Chalfant et al. (36) have systematically approached the functional role of the intracellular NH2 terminus in ENaC. NH2-terminal truncation up to residue 46 did not affect currents through ENaC formed by the truncated α- and wild-type β- and γ-subunits. Different truncations of the first 51 up to the first 67 residues increased the current by about fourfold, and truncations of the first 79 or more residues abolished the amiloride-sensitive current. This indicates that a motif in the region 47–50 has a negative control on the channel activity and that residues in the region 68–79 are crucial for channel function. The deletion of residues 47–50 increases the channel current by increasing the number of channels at the cell surface (36).

Aldosterone is the main hormone controlling ENaC activity at the cell surface. The final step of regulation by aldosterone that leads to an increase in the number of active ENaCs at the cell surface remains unknown but might involve phosphorylation of the ENaC subunits. This question was addressed in Madin-Darby canine kidney cells transfected with αβγ-ENaC subunits, and it was found that aldosterone, insulin, and protein kinases A and C increase the phosphorylation state of β- and γ- but not α-ENaC subunits (225). After aldosterone application, the increase in βγ-ENaC phosphorylation was already observed after 15 min and lasted over a 24-h period, indicating that this ENaC modification covers the early and late response of aldosterone. In this case, ENaC phosphorylation is associated with an increased ENaC activity at the cell surface. Recently phosphorylation of a conserved threonine residue in the close vicinity of the PY motif in β- (T613) and γ-ENaC (T623) subunits could be demonstrated in vivo in Xenopusoocytes (224). The functional relevance of these Thr residues is underlined by a three- to fourfold increase in channel activity after mutation of both residues to Ala or Glu. These results suggest that phosphorylation of the βT613 and γT623 likely by the extracellular regulated kinase (ERK kinase) downregulates ENaC at the cell surface by facilitating the interaction between the PY motif and Nedd4. Thus, depending on the target residue, phosphorylation of the COOH terminus of β- or γ-ENaC could either promote or decrease ENaC activity at the cell surface. The fact that phosphorylation inactivates Nedd4 as mentioned previously further adds to the complex role of the COOH terminus in the regulation of ENaC activity at the cell surface.


PHA-I and Liddle's syndrome are inherited disorders of Na+ balance due to ENaC mutations (described in sects.iii A and vii). Homozygous mutations on the three αβγ-ENaC subunits causing PHA-I have been identified that include premature stop codons [(numbering of human α ENaC) αR56stop, αR508stop], frameshift mutations (αI68fr, αT168fr, αF453fr, βT216fr, βD305fr), an abnormal γ-ENaC splice product at nucleotide 318 (238), and missense mutations (αC133Y and βG37S in human ENaC) (39,95). Recently, the mutation of αS562, homologous to the selectivity filter residue (rat) αS589, to Leu has been found in a PHA-I patient (215). It is expected that truncation of α-ENaC at I68 which precedes M1 leads to a nonfunctional channel. In contrast, truncation of α-ENaC at αR508 leaves a protein with M1 and the major portion of the extracellular loop intact, that is able to assemble with β- and γ-subunits to be targeted to the cell membrane and to generate a small but detectable amiloride-sensitive current (27). This suggests that the αR508stop truncated protein retains sufficient information to promote subunit assembly and trafficking to the membrane. The residue βG37 is part of an intracellular domain that controls channel gating and has been discussed in section vi B of this review. Mutation of a conserved Cys in the post-M1 domain of αENaC to Tyr (human αENaC C133Y, corresponding to rat αENaC C158Y) also causes PHA-I. Functional analysis of this mutation introduced in rat αENaC (C158Y) coexpressed with βγ wild-type ENaC inXenopus oocytes showed that the mutant channel has very similar expression levels as wild-type when the oocytes are incubated at 19°C. If the oocytes are incubated at 30°C however, the Na+ current of ENaC wild-type increases, whereas the current expression of the mutant decreases (95). The homologous C→S mutation in α-ENaC clearly decreases the current and cell surface expression in parallel even at 19°C, and this effect increases at higher temperature (77). Similarly, the analogous point mutation in β- and γ-ENaC decreases current expression. Because neither the assembly nor the degradation rate was affected by these mutants, it was suggested that the mutations may have introduced a Golgi retention signal as in the case for AQP-2 mutants causing nephrogenic diabetes insipidus (179).

The ENaC mutations causing Liddle's syndrome were very informative for the understanding of important structural and functional aspects of ENaC regulation (discussed in sect. vii B). There was a remarkable correlation between the channel gain-of-function mutations identified by a systematic alanine scanning approach of the COOH terminus of β- and γ-ENaC (219) and the mutations found in Liddle's patients (105, 106,114, 152, 218). Both the molecular screening and the genetic studies identified the PPPxY motif (PY motif) in the β- and γ-ENaC as the target sequence for increased channel activity at the cell surface.

Because of the role of ENaC in Na+ reabsorption and in the control of blood pressure, studies were undertaken to screen large numbers of hypertensive patients for mutations in the α-, β-, and γ-ENaC subunits. Particular attention was brought to exon 13 of ENaC genes where mutations in the PY motif causing Liddle's syndrome have been identified (see for review Ref. 48). In the COOH terminus, the polymorphisms αT663A, βG589S, βR597S, βR624C, βE623G, βT594M, γ594insP, γS613A, and γR631H (numbering of the human ENaC subunits) have been identified. ENaC variants in the extracellular loop include αA334, βA336P, βV434M, βG442V, γW177R, and γP501A. At present there is no evidence that any of these mutations affects ENaC function. Except for the αW493R variant, all these polymorphisms were found in regions of poorly conserved sequence. The αW493 residue is located ∼50 residues NH2-terminal of M2 in a highly conserved WPS motif found in all members of the ENaC/DEG family except for ASICs and FaNaC where W is replaced by I. The functional role of this WPS motif remains unknown. It remains possible that the Xenopus oocyte expression system is not consistent enough to detect small changes in channel activity for these variants, which may have smaller functional effects than the Liddle's variants which cause a greater than two- to threefold increase in channel activity. Other ENaC expression systems need to be developed to address the consequences of ENaC polymorphism on channel function to assess the role of ENaC in essential hypertension.


The molecular identification of ENaC, ASIC, and degenerin channels by functional cloning or genetic screening had a major impact on our understanding of transepithelial sodium transport, acid sensing in neurons, and mechanotransduction. Our knowledge regarding the structure and function of these channels is still emerging and needs to be improved. Important structural domains that form the channel core remain to be identified, in particular the structures participating in the inner channel pore. Solving the structure and determining the role of the large extracellular loop that constitutes more than one-half of the protein mass is of primary importance for our understanding of channel function. The functional heterogeneity among the ENaC/DEG family and the conservation of important functional domains provide at the same time a unique opportunity to address the molecular basis of channel functions such as activation by various stimuli or block by pharmacological ligands. We hope that in the near future the structure-function analysis as described in this review will be complemented by the determination of the three-dimensional structure of functional components of the channel protein.

There is increasing evidence that ENaC interacts with proteins that control its synthesis, membrane insertion, degradation, and function. It is possible that the functional ENaC complex found at the cell surface comprises in addition to the ENaC subunits other associated proteins similarly to the mechanotransduction complex in C. elegans. The analysis of regulatory pathways and of protein-protein interactions is in progress in the case of ENaC and will undoubtedly also help in understanding the function of other ENaC/DEG family members as, e.g., ASICs. In addition, strategies need to be developed to obtain a precise biochemical characterization of the functional channel complex at the cell surface and to identify possible associated proteins.

Inactivation of ASIC genes in animal models will hopefully clarify the physiological role of these acid-sensing channels in the central and peripheral nervous system. These models will certainly address the question whether a mechanotransducing system as found in C. elegans also exists in mammals or whether it remains confined to invertebrates. Inherited human diseases caused by the ENaC/DEG family members hINAC and ASICs have not been identified to date, but identification of such diseases would greatly help our understanding of the pathophysiological roles of these channels.


We thank J.-D. Horisberger for invaluable comments and discussions and B. C. Rossier, M. Driscoll, E. Lingueglia, L. Palmer, and P. Greasley for critical reading of the manuscript.

The work in the authors' laboratories was supported by Swiss National Science Foundation Grants 3100–059217.99 (to L. Schild) and 31–65233.01 (to S. Kellenberger) and Human Frontier Science Program Organization Grant RG0261 (to L. Schild). S. Kellenberger was supported by a National Institutes of Health Specialized Center of Research grant for hypertension.


  • Address for reprint requests and other correspondence: L. Schild, Institut de Pharmacologie et de Toxicologie, Bugnon 27, Université de Lausanne, CH-1005 Lausanne, Switzerland (E-mail: Laurent.Schild{at}ipharm.unil.ch).