The Phe-Met-Arg-Phe-amide-activated sodium channel is a tetramer.

The Helix aspersa Phe-Met-Arg-Phe-amide (FMRFamide)-gated sodium channel is formed by homomultimerization of several FMRFamide-activated Na+ channel (FaNaCh) proteins. FaNaCh is homologous to the subunits that compose the amiloride-sensitive epithelial sodium channel, to Caenorhabditis elegans degenerins, and to acid-sensing ionic channels. FaNaCh properties were analyzed in stably transfected human embryonic kidney cells (HEK-293). The channel was functional with an EC50 for FMRFamide of 1 microM and an IC50 (25 degreesC) for amiloride of 6.5 microM as assessed by 22Na+ uptake measurements. The channel activity was associated with the presence of a protein at the cell surface with an apparent molecular mass of 82 kDa. The 82-kDa form was derived from an incompletely glycosylated form of 74 kDa found in the endoplasmic reticulum. Formation of covalent bonds between subunits of the same complex were observed either after formation of intersubunit disulfide bonds following cell homogenization and solubilization with Triton X-100 or after use of bifunctional cross-linkers. This resulted in the formation of covalent multimers that contained up to four subunits. Hydrodynamic properties of the solubilized FaNaCh complex also indicated a maximal stoichiometry of four subunits per complex. It is likely that epithelial Na+ channels, acid-sensing ionic channels, degenerins, and the other proteins belonging to the same ion channel superfamily also associate within tetrameric complexes.

The proteins belonging to this family are characterized by a large extracellular domain located between two large hydrophobic zones forming the transmembrane regions. The NH 2and COOH-terminal domains are cytoplasmic (17). Despite increasing information on the structure and function of these channels, the stoichiometry of the functional complexes remains unknown.
Biochemical handling of FaNaCh is facilitated by several properties: (i) unlike ENaC which comprises ␣, ␤, and ␥ subunits, active FMRFamide Na ϩ channels are formed by a homomultimerization of the FaNaCh channels; (ii) unlike ENaC, the FaNaCh channel is totally silent in the absence of FMRFamide, and therefore not toxic for cells that express it. The determination of the oligomeric arrangement of the FaNaCh channel might serve as a model for other channels within this family (ENaC, degenerins, and ASIC). After extensive functional and biochemical characterization of the channel complex, we show that four FaNaCh subunits participate in the formation of the FMRFamide-activated sodium channel.

EXPERIMENTAL PROCEDURES
Constructs and Transfection-FaNaCh cDNA structure has been previously described by Lingueglia et al. (1). The cDNA was inserted either into the plasmid pBSSK Ϫ -SP6-globin (18) for efficient expression in Xenopus oocytes or in pRC/CMV plasmid (Invitrogen) for transfection of eucaryotic cells. A tagged FaNaCh was constructed by polymerase chain reaction, leading to the addition of the FLAG sequence (Asp-Tyr-Lys-Asp-Asp-Asp-Asp-Lys) to the COOH terminus of the protein. This epitope, which does not alter the channel activity (not shown), can be detected by a commercial monoclonal anti-FLAG antibody (Eastman Kodak Co.) with a high specificity and a low background. Stable transfection was carried out on HEK-293 cells with the FaNaCh or FaNaCh-FLAG pRC/CMV plasmid (Invitrogen) by the Ca 2ϩ -phosphate technique (19). 22 Na ϩ Uptake Experiments-HEK-293 cells were seeded in collagencoated 24-well clusters (Falcon). Confluent cells were rinsed twice with the flux buffer (140 mM choline chloride or N-methyl-D-glucamine chloride, 20 mM HEPES/Tris, pH 7.4, 1.8 mM CaCl 2 , 0.8 mM MgSO 4 ) and then incubated with 200 l of flux buffer supplemented with 2 mM 22 NaCl (0.5-1 Ci/ml Amersham Pharmacia Biotech, 12 Ci/mmol) with or without amiloride or FMRFamide. External 22 Na ϩ was washed out by four 1-ml washings with cold flux buffer, and trapped radioactivity was measured using a Packard 1600 CA liquid scintillation counter. Dose-dependent activations and/or inhibitions were measured after a 150-s incubation.
Antibodies-The cytoplasmic COOH-terminal domain of FaNaCh (amino acids 570 -625) was produced in the glutathione sulfhydryl transferase fusion protein expression vector pGEX-3X according to the * This work was supported by CNRS, INSERM, and the Association Française de Lutte contre la Mucoviscidose (AFLM). Thanks are due to Bristol Myers Squibb for an unrestricted award. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Immunoblot Experiments-Cells were rinsed with Hank's balanced salt solution, scraped, and recovered in a lysis buffer containing 50 mM Tris/H 2 SO 4 , pH 7.4, 100 mM K 2 SO 4 , 5 mM EDTA, and protease inhibitors (100 M leupeptin, 1 mM aprotinin, 1 M pepstatin, 1 mM phenylmethylsulfonyl fluoride). After lysis in a Potter and slow centrifugation, the supernatant was centrifuged 1 h at 30,000 ϫ g. Pellets were resuspended at about 2 mg/ml in lysis buffer and stored at Ϫ70°C. 10 -20 g of membrane proteins were heated at 65°C in SDS-PAGE sample buffer (2.5% SDS, 2 M urea, bromphenol blue, with or without 4% ␤-mercaptoethanol) and applied to SDS-PAGE. After migration, proteins were transferred to a nitrocellulose membrane (Hybond C extra, Amersham Pharmacia Biotech). Nonspecific binding was blocked by a 1-2-h incubation in 140 mM NaCl, 20 mM Tris, 5% milk. The filter was incubated overnight in the same buffer with the monoclonal anti-FLAG antibody (dilution 1/1000, Kodak), rinsed 3 times for 10 min with 140 mM NaCl, 20 mM Tris, 0.05% Tween 20, incubated 45 min with a goat anti-rabbit antibody coupled to peroxidase (Jackson Immunotech), rinsed, and revealed by chemiluminescence (SuperSignal, Pierce).
Metabolic Labeling, Cell Surface Iodination, and Immunoprecipitation-Confluent cells were incubated in a methionine-and cysteinedeficient medium for 30 min prior to the addition of 50 Ci/ml [ 35 S]methionine (Trans 35 S-LABEL, ICN, 1077 Ci/mmol). After a pulse of 2-3 h (or 30 min for pulse-chase experiments), the cells were rinsed and incubated in complete medium (plus 10% fetal calf serum) for various lengths of time. Cells were subsequently pelleted and lysed with 50 l of chilled radioimmunoprecipitation assay buffer (50 mM Tris, pH 7.4, 150 mM NaCl, 10 mM EDTA, 1% C 12 E 8 (Nikko Chemicals), protease inhibitors)/35-mm dish. After preclearing with pansorbin (Calbiochem), the lysate was incubated overnight with the polyclonal antibody (1/100), and pelleted using protein A-Sepharose (Sigma). After washing, samples were heated in SDS-PAGE sample buffer, and applied to 6.5 or 8% SDS-PAGE. Gels were treated with 1 M salicylic acid for 30 min, dried, and exposed to a phosphorimager imaging plate.
Cell surface labeling was performed according to Dalemans et al. (20). Confluent cells were rinsed with PBS ϩ (phosphate-buffered saline ϩ 1 mM CaCl 2 ϩ 1 mM MgCl 2 ). 250 l PBS ϩ , 10 units of lactoperoxidase (Sigma), and 500 Ci of 125 I (NEN Life Science Products, 2200 Ci/mmol) were added to each 35-mm dish at 4°C. Reactions were initiated by adding 10 l of H 2 O 2 0.01%; this addition was repeated twice (every 30 s). Cells were rinsed several times in PBS ϩ , and resuspended in the lysis buffer containing 1% Triton X-100, 0.5% deoxycholate. Proteins were immunoprecipitated as described above and analyzed by SDS-PAGE. Gels were dried and exposed to phosphorimager imaging plate.
Sucrose Gradient Centrifugation and Gel Filtration Analysis-Membrane proteins were solubilized and centrifuged at 170,000 ϫ g for 15 min. In some experiments, the solubilized material was also crosslinked using 500 M DSS, or the formation of intersubunit disulfide bonds was prevented by the addition of 10 mM iodoacetamide. The solubilized material was recovered and applied to the top of a linear 5-15% sucrose gradient formed in water or in D 2 O (50 mM Tris/HCl, pH 7.4, 150 mM NaCl, 10 mM EDTA, 0.1% Triton X-100 or CHAPS, protease inhibitors). ␤-galactosidase (100 g), catalase (100 g), alkaline phosphatase (67 units), and cytochrome c (100 g), used as markers, were also applied on top of the same gradient. After a 17 h run at 39,000 rpm (190,000 ϫ g) on a SW41-TI rotor (Beckman), 300-l fractions were recovered. They were assayed for the presence of the standards and either immunoprecipitated overnight with the anti-FaNaCh antibodies (for metabolic and cell surface labelings) or assayed by Dot-blot with the anti-FLAG antibody. Gel filtration was performed on a Superdex 200 column (Amersham Pharmacia Biotech) equilibrated in a buffer con-taining 150 mM NaCl, 10 mM EDTA, 0.1% Triton X-100, at a flow rate of 0.4 ml/min.
The apparent sedimentation coefficients of FaNaCh in H 2 O and D 2 O were determined. From these values and the values of densities and viscosities, it was possible to determine the sedimentation coefficient at 20°C in water (s 20,w ), and the partial specific volume (v , equal to the inverse of the density) of the detergent-protein complex (22). The contribution of the detergent to the complex was deduced from that latter value, knowing the partial specific volume of Triton X-100 (equal to 0.908 cm 3 /g), and the partial specific volume of proteins (using a mean value equal to 0.73 cm 3 /g) (23).
Stokes radius was determined from gel filtration data. According to Siegel and Monty (24), the total mass of the protein-detergent complex was deduced from the Equation 1.
where 20, is the viscosity of water at 20°C (equal to 0.01002 Poise), N A is the Avogadro number (6.02 10 23 ) and 20,w is the density of water at 20°C (0.99823 g/cm 3 ). The molecular weight of the protein complex was deduced from this value, taking into account the contribution of the detergent to the total mass of the complex.

Functional Expression of FMRFamide-activated Sodium Channels in FaNaCh-transfected HEK-293 Cells-FMRF-
amide-activated sodium channel activity was measured after stable transfection of the FaNaCh cDNA into HEK-293 cells. Fig. 1 shows that the properties of the channel observed in FaNaCh-transfected HEK-293 cells were identical to those of the native channel recorded in snail neurons (25) and to those of the channel recorded in Xenopus oocytes injected with the corresponding cRNA (1). 22 Na ϩ uptake was largely increased after stimulation with 30 M FMRFamide in transfected cells (Fig. 1A). The time course of the uptake was biphasic. After a relatively fast rise to a plateau, the 22 Na ϩ uptake declined to a lower intermediate value. The initial event is consistent with the rapid development of a Na ϩ permeability. The decline of the signal observed after 2.5 min is likely due to the inactivation of the channel (25) combined with a sustained Na ϩ /K ϩ -ATPase activity. A Michaelis-Menten relationship was ob- served between the external concentration of sodium and the initial ratio of 22 Na ϩ uptake (Fig. 1B) with a half-maximal activation of 1.4 mM. A similar value (K 0.5 ϭ 5 mM) has been reported for the external sodium dependence of the ENaC activity after expression in Xenopus oocytes (4).
FMRFamide activated the 22 Na ϩ uptake in a dose-dependent manner with a half-maximal activation at a concentration of 1 M (Fig. 1C), nearly identical to the 2 M value measured by electrophysiological techniques after FaNaCh expression in Xenopus oocytes (1). The channel expressed in HEK-293 cells was also blocked by amiloride (Fig. 1D). At 37°C, the IC 50 for amiloride was 42 M. The IC 50 was shifted to 6.5 M at 25°C and to 1.6 M at 8°C. An IC 50 of 10 M was previously determined in snail neurons at 25°C (25).
FaNaCh Is Expressed at the Cell Surface as a 82-kDa Glycoprotein-The protein expressed in transfected cells was first characterized using a polyclonal antibody directed against the COOH-terminal domain. In pulse-chase experiments, the polyclonal antibody immunoprecipitated a 74-kDa protein, which matured after 2-3 h into a 82-kDa protein ( Fig. 2A). After deglycosylation of the immunoprecipitated material with Nglycosidase F (data not shown), or pretreatment of the cells with tunicamycin (a blocker of N-glycosylation), a 67-kDa protein was detected in good agreement with a predicted molecular mass equal to 71 kDa for the polypeptide (Fig. 2B). Fig. 2B also shows that the 74-kDa band was the only product detected after a 3 h pulse-chase made in the presence of brefeldin A, an inhibitor of intracellular transport. Since the 74-kDa form was also sensitive to endoglycosidase H (data not shown), which cleaves high mannose structures characteristic for glycosylations in the endoplasmic reticulum, it probably corresponds to an intermediate form of maturation of FaNaCh. On the contrary, the 82-kDa form, detected after longer times of metabolic labeling, was resistant to endoglycosidase H digestion (data not shown) and is thus considered to correspond to a fully mature terminally glycosylated form. Fig. 2C demonstrates that the 82-kDa protein can indeed be detected at the cell surface since it is immunoprecipitated by anti-FaNaCh antibodies after a cell surface labeling of FaNaCh-transfected cells with Na 125 I and lactoperoxidase. The channel complex is present in large amounts at the surface since the FaNaCh signal represented up to 10% of the total radioactive material.

Solubilization by Triton X-100 Favors the Formation of Intersubunit Disulfide Bonds and Leads to the Formation of Covalent Tetramers-When SDS-PAGE migration was performed
under nonreducing conditions, oligomeric forms of FaNaCh were observed (Fig. 3A). These forms were sensitive to ␤-mercaptoethanol, suggesting the existence of disulfide bonds between subunits. When membranes were prepared in the presence of 10 mM iodoacetamide, the monomer was the only form detected (Fig. 3A). This result shows that formation of intersubunit disulfide bonds is not occurring in the native channel, but rather that it results from a redox potential alteration created by the experimental conditions during cell lysis or the following steps.
Disulfide bond formation was increased after solubilization with Triton X-100. Whereas we detected only the monomeric form (74 -82 kDa) on iodoacetamide-treated membranes after solubilization with 1% Triton X-100, additional bands at 151- 164 kDa (dimer), 205-230 kDa (trimer), and 305-327 kDa (tetramer) were observed in parallel analysis with control membranes prepared in the absence of iodoacetamide (Fig. 3A). The Triton X-100 effect was time-dependent. An overnight incubation with Triton X-100 (instead of 15 min) significantly increased the percentage of tetrameric form (data not shown). In vitro intersubunit disulfide bond rearrangement or formation has been reported for a variety of other oligomeric proteins such as Shaker K ϩ channel (26), synaptophysin (27), and the brain glutamate transporter (28). In the two latter cases, the extent of rearrangement depended on the detergent used for solubilization. It was maximal with SDS for synaptophysin and with CHAPS for brain glutamate transporter. For FaNaCh, formation of tetramers was only observed after solubilization with Triton X-100 and not after solubilization with CHAPS or with deoxycholate.
Reticulation of the Complex by Bifunctional Cross-linkers Generates Up to Tetramers-After a lactoperoxidase labeling of the cell surface proteins with 125 I and covalent reticulation with DSS, the material was immunoprecipitated and analyzed by SDS-PAGE. Cross-linked proteins were detected on a 3.5% Weber-Osborn gel (nonreducing conditions) at apparent molecular masses of 180 (dimer), 260 (trimer), and 345 kDa (tetramer) (Fig. 3C). Most of the cross-linked material appeared as trimers, but the presence of a tetramer was detected in three out of eleven experiments as a diffuse band around 345 kDa.
When membranes were solubilized with Triton X-100 then cross-linked with the heterobifunctional reagent SADP (Fig.  3B), the tetramer was the main form detected by Western blot. Fig. 3B also shows an increase of the tetrameric form when SADP concentration was increased from 100 to 500 M. The formation of covalent tetramers was not restricted to Triton X-100-solubilized material. They were also detected on CHAPS-solubilized material after a reticulation with the heterobifunctional cross-linker NHS-ASA (Fig. 3D).
Hydrodynamic Analysis of Solubilized FaNaCh Demonstrates the Tetrameric Organization of the Complex-The pool of FaNaCh proteins expressed at the cell surface was analyzed by sedimentation experiments through sucrose gradients after a solubilization with 1% Triton X-100 ϩ 0.5% deoxycholate. A unique peak characterized by an apparent sedimentation coefficient of 9.5 S was observed. This peak was not modified after cross-linking of the membrane proteins with DSS (not shown). This proved that a unique FaNaCh complex (formed by a given number of subunits) was expressed at the cell surface and was resistant to solubilization by 1% Triton X-100 ϩ 0.5% deoxycholate.
Hydrodynamic characterization of the complex solubilized in 1% Triton X-100 was then performed. The apparent sedimentation coefficients in a linear 5-15% sucrose gradient formed in water and in D 2 O were determined (Figs. 4, A and B). The sedimentation coefficient at 20°C in water (s 20,w ) of the Triton X-100-protein complex was equal to 9.12 Ϯ 0.02 S, and the partial specific volume v was equal to 0.774 Ϯ 0.004 cm 3 /g (n ϭ 2). A 25% detergent contribution to the total mass of the complex was deduced, assuming a partial specific volume of FaNaCh in the absence of detergent equal to 0.73 cm 3 /g. The Stokes radius was equal to 73 Ϯ 4 Å (n ϭ 3), as determined by a chromatography on a Superdex-200 column (Fig. 4D). Results are summarized in Table I. Using Equation 1, a value of 333 Ϯ 24 kDa was found for the complex formed by the protein and Triton X-100, and 251 Ϯ 18 kDa for the glycopeptidic complex in the absence of detergent. Assuming a mean molecular mass of 78 kDa for a FaNaCh monomer (due to the fact that the 74-and 82-kDa forms contribute equally to the total FaNaCh signal (Fig. 3A), a first estimate of the subunit stoichiometry equal to 3.2 was found.
Since formation of interchains disulfide bonds occurred after solubilization with Triton X-100 (Fig. 3A), two distinct control experiments were performed to ensure that the Triton X-100solubilized material was not denatured. First, a unique peak with similar s 20,w and v (8.8 S and 0.806 cm 3 /g) was detected after membranes had been treated with 10 mM iodoacetamide, which prevented the formation of covalent oligomers. Second, a similar sedimentation profile was obtained after solubilization of the complex with CHAPS (Fig. 4C). The peak was characterized by an apparent sedimentation coefficient equal to 9.7 Ϯ 0.3 S (n ϭ 3).
The unknown value of the partial specific volume of the protein affected the estimation of the stoichiometry (Table I). Small changes can significantly affect the evaluation of the detergent contribution to the complex. A maximal stoichiometry, however, can be obtained assuming that no detergent is bound to the complex. In that case, the total mass of the complex only represents oligomers of FaNaCh. A maximal value equal to 4.3, that is, strictly below 5, was deduced in agreement with the tetrameric organization demonstrated by cross-linking experiments.
The material sedimented through a sucrose gradient was also analyzed by Western blot. Fig. 5 shows that the covalent dimers, trimers, and tetramers formed by oxidation after Triton X-100 solubilization are co-sedimenting. This indicates that formation of disulfide bonds only occurred between subunits of the same complex. Taken together with the other data presented above, this is consistent with a tetrameric organization of the FaNaCh complex.

DISCUSSION
The properties of the FMRFamide-gated sodium channel expressed in HEK-293 cells are quite consistent with the properties of the channel recorded in native tissues or after expression of the cloned cDNA into Xenopus oocytes. Since FaNaCh expression in HEK-293 cells and in Xenopus oocytes is able to generate a large FMRFamide-induced inward current, it is likely that the functional channel corresponds to a homomultimeric structure, and does not need association with endogenous subunits. The transfected cells therefore represent a useful model to study the pharmacological and biophysical properties of the FMRFamide-gated sodium channel. These results also demonstrate that correct assembly of the channel complex is possible in these transfected cells, which can be used to characterize the quaternary structure of the channel.
The main purpose of this study was to analyze the maturation pattern and the quaternary structure of the FMRFamidegated sodium channel. In transfected HEK-293 cells, FaNaCh glycosylation occurs in two steps. An immature precursor coreglycosylated form is produced into the endoplasmic reticulum and is then converted into a fully glycosylated form abundantly expressed at the cell surface, which generates the expected channel activity. Figs. 3 and 5 show that the immature form is already assembled into an oligomeric complex that sediments at the level of a unique peak on the sucrose density gradient. This proves that tetrameric assembly occurs at an early step of maturation in the endoplasmic reticulum. A similar type of assembly has also been reported for the tetrameric potassium channel Shaker (29).
Covalent tetramers were observed after use of three distinct chemical cross-linkers, i.e. DSS, SADP, and NHS-ASA. They were also detected after oxidation of the protein through the formation of interchain disulfide bonds. Triton X-100 increased the latter covalent bonding, but tetramers were also observed after solubilization in the presence of other detergents such as CHAPS, followed by chemical cross-linking (Fig. 3). It is therefore unlikely that tetramers only represent artifactual denaturation due to Triton X-100. Thus, cross-linking experiments provided a minimal stoichiometry of the complex equal to 4, and hydrodynamic properties of the solubilized FaNaCh complex provided a maximal stoichiometry strictly below 5. Taken together, these results led us to the conclusion that the FaNaCh complex is composed of four subunits. Since the FMRFamide-gated sodium channel is a member of the same structural family as ENaC, degenerins, and ASIC, a stoichiometry of 4 can also be expected for these channels.
McNicholas and Canessa (30) analyzed the properties of epithelial Na ϩ channels generated after coexpression of ␣ENaC and ␤ENaC, or ␣ENaC and ␥ENaC. Using different ratios of injected cRNAs, they showed an optimal signal after injection of equal amounts of ␣ENaC ϩ ␤ENaC and ␣ENaC ϩ ␥ENaC. This would also be consistent with a tetrameric organization in which two ␣ subunits would be associated with one ␤ and one ␥ subunit. The same type of conclusion is also suggested by the results of Grü nder et al. (31). These authors have analyzed the residual ENaC current generated after mutation of a highly conserved region located before the first transmembrane region of ␣, ␤, or ␥ENaCs. As assessed by the residual current recorded after expression of a mutant subunit with two other wild-type chains, the role of the three subunits was not strictly equivalent. A stronger effect was observed after mutation of ␣ENaC than after mutation of ␤ or ␥ENaCs. A tetrameric organization of the ENaC complex with two copies of ␣ENaC, one copy of ␤ENaC, and one copy of ␥ENaC might suggest that the dominant negative effect would directly depend upon the number of mutated subunits participating in the complex.
Gain-of-function mutations affecting MEC-4 and MEC-10 degenerins, which are likely subunits of a common complex expressed in the six touch-receptor neurons of C. elegans, cause neurodegeneration of these cells. Modulation of MEC-4 and/or MEC-10-induced degenerations by a supplementary loss-offunction mutation in trans was analyzed for MEC-10 (8) and  5. Western blot analysis of the sucrose gradient fractions. SDS-PAGE was performed on the fractions around the maximal FLAG signal (fractions 9 -16 of the 5-15% sucrose gradient shown in Fig. 4A), under nonreducing conditions. The intensity of dimeric, trimeric, and tetrameric (mature (m) ϩ immature (i)) form in each fraction was quantified using N.I.H. Image 1.60 program. Ⅺ, dimer; f, trimer; E, tetramer. . It was concluded that there are at least two copies of MEC-10 and two copies of MEC-4 in the complex. A tetrameric organization of the MEC-4/MEC-10 complex formed by two copies of MEC-10 and two copies of MEC-4 would then imply that the other mec gene products that build up the complex such as MEC-6 (8) are not degenerins.
This work is the first direct demonstration that one member of the ENaC/FaNaCh/ASIC/degenerins superfamily can form tetramers. It is interesting to note that the family of inward rectifying K ϩ channels, in which each of the subunits is also characterized by the presence of only two transmembrane domains, but which displays no homology with proteins of the ENaC/FaNaCh/ASIC/degenerins family, also has a tetrameric organization (33). It might be tempting to speculate that the P 2x subtype of purinergic receptors, which also displays a structural organization with two transmembrane domains per subunit, (34) also forms tetramers.