Mutations in the pore region modify epithelial sodium channel gating by shear stress.

Previous studies have shown that epithelial Na+ channels (ENaCs) are activated by laminar shear stress (LSS). ENaCs with a high intrinsic open probability because of a mutation (betaS518K) or covalent modification of an introduced Cys residue (alphaS580C) in the pre-second transmembrane domain (pre-M2) were not activated by LSS, suggesting that the pre-M2 region participates in conformational rearrangements during channel activation. We examined the role of the pore region of the alpha-subunit in channel gating by studying the kinetics of activation by LSS of wild-type ENaC and channels with Cys mutations in the tract Ser576-Ser592. Whole cell Na+ currents were monitored in oocytes expressing wild-type or mutant ENaCs prior to and following application of LSS. Following a 2.2-s delay, a monoexponential increase in Na+ currents was observed with a time constant (tau) of 8.1 s in oocytes expressing wild-type ENaC. Cys substitutions within the alpha-subunit in the tract Ser580-Ser589 resulted in: (i) a reduction (Ser580-Trp585, Gly587) or increase (Ser589) in delay times preceding channel activation by LSS, (ii) an increase (Gln581, Leu584, Trp585, Phe586, Ser588) or decrease (Ser589) in the rate of channel activation, or (iii) a decrease in the magnitude of the response (Ser583, Gly587, Leu584). Cys substitutions at a putative amiloride-binding site (alphaSer583 or betaGly525) or within the selectivity filter (alphaGly587) resulted in a reduction in the LSS response, and exhibited a multiexponential time course of activation. The corresponding gamma-subunit mutant (alphabetagammaG542C) had a minimal response to LSS and exhibited a high intrinsic open probability. These data suggest that residues in the pore region participate in the sensing and/or transduction of the mechanical stimulus that results in channel activation and are consistent with the hypothesis that the ENaC pore region has a key role in modulating channel gating.

The amiloride-sensitive epithelial sodium channel (ENaC) 1 is a member of the degenerin/ENaC family of ion channels that encodes a group of related proteins implicated in the control of Na ϩ homeostasis, salt taste, nociception, and mechanotransduction (1,2). Degenerin/ENaC family members form homo-or heteromeric channels composed of subunits that are similar in structure, with two membrane-spanning domains linked by a large extracellular region, and intracellular N and C termini. ENaCs are composed of three subunits ␣, ␤, and ␥, arranged with a presumed subunit stoichiometry of 2␣:1␤:1␥ (3,4) although an alternative 3␣:3␤:3␥ stoichiometry has been proposed (5). ENaCs mediate Na ϩ transport across apical membranes of epithelial cells that line the distal nephron, airway and alveoli, and distal colon. In addition, ENaC subunits are expressed in mechanosensory structures, including the Merkel cell-neurite complex and baroreceptor terminal endings (6), although their functional roles have not been elucidated.
Micropuncture and microperfusion studies of distal nephron segments have revealed that rates of K ϩ secretion and Na ϩ reabsorption are flow-dependent (7)(8)(9)(10)(11). Similar to cells lining the vasculature, cortical-collecting duct cells experience two kinds of forces in response to changes in rates of flow, circumferential stretch, and fluid flow-induced shear (12). We recently demonstrated that laminar fluid perfusion of oocytes expressing ENaCs resulted in a flow-dependent increase in whole cell Na ϩ currents. Furthermore, mutant channels (␣␤S518K␥ or ␣S580C␤␥ following modification by a specific sulfhydryl reagent) that are known to exhibit a high intrinsic open probability did not respond to laminar shear stress (LSS), suggesting that LSS activates ENaC by increasing channel open probability (13).
Degenerin/ENaC channels with mutations within either (i) a region preceding the second membrane-spanning domain (pre-M2), (ii) the extracellular loop, (iii) the N-or (iv) C-terminal domains alter channel openings and closures (1,(13)(14)(15)(16)(17), indicating that multiple regions have a role in modulating channel gating. For example, the combination of site-specific introduction of Cys residues and covalent modification by thiol-reactive reagents has shown that the pre-M2 region of ENaC subunits has a role in the regulation of channel gating (16,18,19). Channels with introduced Cys residues at selected sites in the ␣-subunit preceding a putative amiloride-binding site (Val 572 , Ser 576 , Asn 577 , Ser 580 , Gln 581 , and Trp 582 ) responded to thiolreactive reagents with increases in activity that are likely caused by increases in channel open probability (16). In contrast, an amiloride binding site mutant (␣S583C) responded to thiol-reactive reagents with a large decrease of channel activity.
We have examined whether the introduction of Cys residues in the pre-M2 region of the ␣-subunit, at sites adjacent to or within the selectivity filter region, alters the response of ENaC to LSS. We observed that Cys substitutions within the tract ␣Ser 580 -␣Ser 589 produced significant changes in the time course of channel activation by LSS, suggesting that conforma-tional changes in this area occur during channel gating in response to LSS.

MATERIALS AND METHODS
DNA Constructs-Point mutations of mouse ENaC subunits were previously generated by site-directed mutagenesis with the sequential polymerase chain reaction method using Pfu DNA polymerase (Stratagene, La Jolla, CA) (16). Mutations were confirmed by DNA sequencing.
Two-electrode Voltage Clamp-Two-electrode voltage clamp (TEV) was performed at 22-25°C using a GeneClamp 500B amplifier (Axon Instruments, Union City, CA). Temperatures were determined with a probe localized in close proximity to the oocyte with a TC-10 temperature controller (Dagan, Minneapolis, MN). Data were acquired through Clampex 8.0 using a DigiData 1200 interface and stored on the hard disk of a 233 MHz Pentium II computer. Pipettes filled with 3 M KCl had resistances of 0.5-5 M⍀. The extracellular solution (TEV solution) was (in mM): 110 NaCl, 2 KCl, 1.54 CaCl 2 , 10 HEPES, pH 7.4. The recording chamber was perfused at a rate of 3.5 ml/min. Laminar shear stress was applied by perfusing TEV solution through a vertical pipette localized above the oocyte surface at a rate of 1.6 ml/min, corresponding to 0.137 dynes/cm 2 of shear stress as described previously (13). Bath perfusion was maintained during application of LSS.
ENaC-mediated whole cell Na ϩ currents were defined as the benzamil-sensitive component of the current. To determine the time course of activation by LSS, oocytes were clamped at a holding potential of Ϫ60 mV and stimulated by perfusion through a vertical pipette with a computer-controlled protocol generated in pClamp 8.0. Following the stimulation process, the ENaC-mediated component of the whole cell Na ϩ current was determined by bath perfusion with TEV solution supplemented with 5 M benzamil, unless otherwise indicated. Na ϩ Self-inhibition-Two-electrode voltage clamp was performed 20 -48 h after injection of oocytes with cRNAs for ␣␤␥ or ␣␤␥G542C mENaCs. Oocytes were bathed in a low or high [Na ϩ ] solution and continuously clamped at Ϫ60 mV. The bath solutions contain either 1 mM (low) or 110 mM (high) NaCl, 2 mM KCl, 2 mM CaCl 2 , and 10 mM HEPES with a pH of 7.4. In the low [Na ϩ ] solution, Na ϩ was replaced by N-methyl-D-glucamine. Low and high [Na ϩ ] bath solutions were rapidly exchanged with a computer-controlled perfusion system (Auto-Mate Scientific Inc., San Francisco, CA). The inward current decay following a fast switching from low [Na ϩ ] to high [Na ϩ ] bath solution represents the response of Na ϩ self-inhibition as described previously (20,21). The decay was fitted with an exponential equation to obtain a time constant (). The magnitude of the inhibition is indicated by the ratio of the steady state (I ss ) to the peak current (I peak ).
Single Channel Studies-Bath and pipette solutions were identical containing 110 mM LiCl, 2 mM CaCl 2 , 2 mM KCl, and 10 mM Hepes, pH 7.4. Oocytes were placed in a hypertonic solution (bath solution supplemented with 200 mM sucrose) for 5-10 min, and the vitelline membranes were removed manually. The oocytes were then transferred to the bath solution at room temperature (22-25°C). Patch pipettes with a tip resistance of 5-12 megohms were used. Currents were recorded in the cell-attached mode using an Axopatch 200B Amplifier (Axon Instruments) and a DigiData 1322A interface (Axon Instruments) connected to a Pentium 4 PC (Gateway). Single channel recordings were acquired using pClamp 9.0 (Axon) at 20 kHz, filtered at 1000 Hz by a 4-pole low pass Bessel Filter built in the amplifier, and stored on the hard disk. Single channel currents were further filtered at 100 Hz with a Gaussian filter for display and analysis. Recordings were performed at ϩ60 mV (pipette potential). Single channel currents were estimated from amplitude histograms generated from recordings. Open probabilities were determined by fitting amplitude histograms of at least 4 min in duration with Gaussian functions by the Levenberg-Marquardt method using pClamp 6 (Axon Instruments).
Data and Statistical Analyses-Data were expressed as the mean Ϯ S.E. (n), where n equals the number of independent experiments analyzed. Data were analyzed with Clampfit 9.0 (Axon Instruments) or where I is the macroscopic current, 1 and 2 are time constants, a 1 , a 2 , and c are constants determined by curve fitting. To determine the parameters of the exponential function, currents were fit excluding data within 5-10% of the beginning of the exponential phase, which produced the best fit of the exponential increase. For mutants carrying mutations in the amiloride binding site (␣S583C␤␥ and ␣␤G525C␥), the response to LSS was not monoexponential. For oocytes expressing ␣S583C␤␥ channels only the late phase of current increase was adequately estimated by fitting with an exponential function. For the ␣␤G525C␥ mutant, recordings were visually inspected and currents were fit by a single exponential function including data from each of the two exponential steps. The delay time (␦) was estimated as the time between the delivery of the stimulus and the intersection of the exponential phase with the baseline current, using Equation 3, where I 0 is the current at time 0 before stimulation (see Fig. 2). For the mutants ␣S583C␤␥, ␣␤G525C␥, and ␣G587C␤␥, delay times were determined as the time at which the initial change in current occurred by inspection of the recordings. Statistical comparisons were performed using GraphPad Instant (GraphPad Software, San Diego, CA).

RESULTS
Whereas the location and nature of the ENaC gate and mechanisms by which signals are transmitted to alter channel openings and closures have not been defined, analyses of ENaC mutants have shown that multiple regions within ENaC influence channel gating (1,(13)(14)(15)(16)(17). Previous studies have shown that selected mutations within the pre-M2 tract have dramatic effects on channel gating (18,22). We examined the kinetics of ENaC activation by LSS, and the effects of Cys substitutions of residues in the pre-M2 region of the ␣-subunit on this process. Fig. 1 shows the pre-M2 region of mouse ENaC (mENaC) subunits. Sites associated with alterations of ENaC activity following covalent modification of introduced Cys residues with thiol-reactive reagents are illustrated. A putative amiloridebinding site and a key three-residue selectivity filter are also shown.
Oocytes expressing wild-type channels or ENaCs with individual Cys substitutions in the tract ␣Ser 576 -␣Ser 592 of mENaC were positioned in a chamber and stimulated by delivery of a fluid jet of TEV solution as described under "Materials and Methods." Whole cell Na ϩ currents were monitored at a holding potential of Ϫ60 mV. Following the stimulation period, oocytes were perfused with TEV solution containing 5 M benzamil, unless otherwise indicated.
Previous analyses of the kinetics of the voltage-dependent activation of Shaker potassium channels have shown a discrete delay time between a voltage pulse and observable increase in channel open probability (23,24). In a similar manner, analyses of ENaC activation by LSS show a delay time between the delivery of the LSS stimulus and the monoexponential increase in whole cell Na ϩ current (Fig. 2). The delay time for oocytes expressing wild-type channels was 2.18 Ϯ 0.11 s (n ϭ 57, see Fig. 2). We previously reported a half-time for activation (time at which the current is 50% of maximum) by LSS of ϳ5 s (13). In agreement with our previously published findings, the time constant () for ENaC activation was 8.06 Ϯ 0.33 s (n ϭ 57). Fig.  2 illustrates a representative response to LSS, as well as the methods employed to determine the delay time (␦) and the time constant () for the monoexponential increase in whole cell Na ϩ current.
Analyses of delay times between the delivery of the LSS stimulus and a monoexponential increase in whole cell Na ϩ current ( Fig. 3) revealed that oocytes expressing Cys mutations in the tract ␣Ser 580 -␣Trp 585 exhibited a significantly reduced ␦, compared with wild-type channels (Fig. 3). Analyses of oocytes expressing mutations in the amiloride binding site (␣S583C␤␥, ␣␤G525C␥, ␣␤␥G542C) and the selectivity filter (␣S587C and ␣S589C) are shown in separate figures. The delay time for channel activation by LSS likely represents electrically silent steps that occur prior to increases in macroscopic currents.
Analyses of the time course of the exponential activation by LSS showed that Cys substitutions at ␣Gln 581 , ␣Leu 584 , ␣Trp 585 , ␣Phe 586 , and ␣Ser 588 led to a significant reduction of the time constant for channel activation (Fig. 4). Fig. 5 illustrates representative recordings of whole cell current activation by LSS of wild-type channel and selected ENaC mutants that exhibited an altered ␦ or . We previously presented a structural model of the ENaC pore region, based on MTS reagent accessibility and secondary structural predictions (16). Residues between ␣Val 572 and ␣Ser 580 were proposed to form a ␣-helix. Furthermore, we proposed that a transition region (residues ␣Ser 580 -␣Ser 583 ) was followed by an extended region (␣Leu 584 -␣Ser 592 ) that includes the selectivity filter (␣Ser 587 -␣Ser 589 ). Our results indicate that mutations in the predicted transition and extended regions from ␣Ser 580 to ␣Ser 589 are associated with changes in the time course of activation during LSS stimulation.
We also examined whether Cys substitutions in the tract ␣Ser 576 -␣Ser 592 affects the magnitude of the LSS response. Whole cell Na ϩ currents were measured prior to and following stimulation by LSS, and after bath perfusion with 5 M benzamil. Ratios of the benzamil-sensitive peak current, relative to the benzamil-sensitive basal current (I peak /I basal ) are shown in Fig. 6. Data are presented relative to the response of wild-type Oocytes expressing wild-type mENaC were perfused with TEV solution, and a computer-controlled perfusion system was used to activate fluid-jet perfusion via a vertical pipette at time 0. Experimental data were fit as described under "Materials and Methods" using Equation 1, where I is the whole cell current, and is the time constant. The delay time (time between dashed lines) was estimated from the parameters obtained from the exponential fitting using the Equation 3, where I 0 is the current at time 0; c, a, and were obtained from the exponential fit.
FIG. 3. Delay times of wild-type and Cys-substituted mENaCs following the initiation of LSS. Oocytes expressing mENaCs were perfused with TEV solution, and a computer-controlled perfusion system was used to activate fluid-jet perfusion via a vertical pipette. Delay times of activation by LSS were estimated as described under "Materials and Methods." A gray bar indicates statistically significant differences in delay times between wild-type control versus ␣Q581C␤␥, ␣W582C␤␥, or ␣W585C␤␥ (p Ͻ 0.01) and wild-type control versus ␣S580␤␥ or ␣L584C␤␥ (p Ͻ 0.001) (Kruskal Wallis test (nonparametric ANOVA) followed by Dunn's multiple comparisons post-test). Experiments were performed with 9 -15 oocytes for mutant channels and 57 oocytes for wild-type channels. Asterisks indicate mutants that are presented in Fig. 8 and Table III. FIG. 4. Time constants of activation of wild-type and Cys-substituted mENaCs by LSS. Oocytes expressing mENaCs were perfused with TEV solution, and a computer-controlled perfusion system was used to activate fluid-jet perfusion via a vertical pipette. Time constants of the exponential activation by LSS of wild-type and Cyssubstituted mENaCs were estimated as described under "Materials and Methods." A gray bar indicate statistically significant differences in between wild-type control versus ␣W585C␤␥, ␣F586C␤␥, or ␣S588C␤␥ (p Ͻ 0.01) and wild-type control versus ␣Q581C␤␥ or ␣L584C␤␥ (p Ͻ 0.001) (Kruskal Wallis test (nonparametric ANOVA) followed by Dunn's multiple comparisons post-test). Experiments were performed with 9 -15 oocytes for mutated channels and 57 oocytes for wild-type channels. Asterisks indicate mutants that are presented in Fig. 8 and Table III. ENaC. The I peak /I basal of wild-type ENaC was 1.37 Ϯ 0.04 (n ϭ 62), in agreement with our previously published findings (13). Cys substitutions at multiple sites between ␣Ser 576 and ␣Ser 592 modified the magnitude of the response to LSS. However, a statistically significant difference was only observed between oocytes expressing wild-type and ␣L584C␤␥ channels when data were compared by nonparametric ANOVA (Kruskal Wallis test) followed by Dunn's multiple comparisons post-test.
Previous studies have identified a homologous site in the ␤and ␥-subunits (␤Gly 525 and ␥Gly 542 ) where Cys substitutions result in a dramatic decrease in the apparent affinity of the channel for amiloride (3,25). Cys substitution at the homologous position in the ␣-subunit (␣Ser 583 ) was associated with only a modest change in amiloride affinity (3,25). Furthermore, covalent modification of channels with Cys substitutions at these sites (␣Ser 583 , ␤Gly 525 , or ␥Gly 542 ) by sulfhydryl-reactive reagents led to irreversible channel block (3, 16, 26, 27).
Oocytes expressing the amiloride-binding site mutants ␣S583C␤␥ or ␣␤G525␥ responded to LSS with a biphasic activation of whole cell Na ϩ currents (Fig. 7). The delay times for ␣S583C␤␥ and ␣␤G525C␥ were also significantly shorter than FIG. 6. Relative response of whole cell Na ؉ currents to LSS. Oocytes expressing wild-type or Cys-substituted mENaCs were perfused with TEV solution, and a computer-controlled perfusion system was used to activate fluid-jet perfusion via a vertical pipette. The peak response of the whole cell benzamil-sensitive Na ϩ current following the initiation of LSS was normalized to the basal current. The I peak /I basal values were normalized to the I peak /I basal values observed for wild-type ENaC. The I peak /I basal of wild-type ENaC was 1.37 Ϯ 0.04 (n ϭ 62). Significant changes in the magnitude of the response were observed between oocytes expressing wild-type channels versus ␣L584C␤␥ channels (p Ͻ 0.05, Kruskal Wallis test (nonparametric ANOVA) followed by Dunn's multiple comparisons post-test). Experiments were performed with 8 -15 oocytes for mutated channels and 62 oocytes for wild-type channels. Asterisks indicate mutants that are presented in Fig. 8 and Table III. Cys-substituted mENaC-expressing oocytes (␣S583C␤␥, ␣␤G525C␥, or ␣␤␥G542C) were perfused with TEV solution, and a computer-controlled perfusion system was used to activate fluid-jet perfusion via a vertical pipette as is indicated by the vertical dash line. Currents were monitored at a holding potential of Ϫ60 mV. Data for ␣S583C␤␥ and ␣␤G525C␥ were fit by exponential functions as described under "Materials and Methods." that observed for wild-type channels (Fig. 8A and Table I). Current activation of oocytes expressing ␣S583C␤␥ and ␣␤G525C␥ channels stimulated by LSS was fit by visual inspection of the recordings (Fig. 8B). For oocytes expressing ␣S583C␤␥ channels, only the late phase of current increase was adequately estimated by a single exponential function. For ␣␤G525C␥ channels, currents were fit by single exponential functions including data from each of the two exponential steps. The results show that the activation process involved multiple steps as indicated by delay times and time constants of activation (Table I). The extent of activation of mutated channels at the amiloride-binding site in response to LSS was significantly reduced, when compared with wild-type channels (Fig. 8C). Surprisingly, for ␣␤␥G542C the response was too small to determine delay times and time constants.
We recently reported that Na ϩ self-inhibition was eliminated by covalent modification of an introduced Cys residue within ENaC pore (␣S580C), suggesting that channels with a high intrinsic open probability do not exhibit Na ϩ self-inhibition (20,21). To gain more information regarding the role of ␥G542C on ENaC gating, we examined Na ϩ self-inhibition in oocytes expressing wild-type or ␣␤␥G542C channels. Oocytes expressing the ␣␤␥ or ␣␤␥G542C channels were clamped at Ϫ60 mV. A change from a low Na ϩ to a high Na ϩ solution lead to a rapid increase in currents followed by an exponentially decrease to a steady-state level (Fig. 9). As indicated in Table II, the ratio of steady state current (I ss ) to the peak current (I peak ), for wildtype channels was 0.62 Ϯ 0.01, and the inhibitory time constant was 7.14 Ϯ 0.47 s. In contrast, the ratio (I ss /I peak ) for ␣␤␥G542C channels was 0.85 Ϯ 0.02, and the inhibitory time constant was 14.52 Ϯ 2.14 s. These results indicate that the ␥G542C mutation greatly reduces the magnitude of the selfinhibition response. Additionally, the ␥G542C mutation signif- FIG. 9. Na ؉ self-inhibition of wild-type and ␣␤␥G542C mEN-aCs. Oocytes expressing wild-type or ␣␤␥G542C channels were clamped at Ϫ60 mV. The bath Na ϩ concentrations are indicated by open (1 mM) and gray (110 mM) bars. An increase in the bath concentration of Na ϩ of wild-type channels (black) induced a rapid increase in the current that was follow by an exponential decrease (Na ϩ self-inhibition). Oocytes expressing ␣␤␥G542C channels (gray) showed a relatively small Na ϩ self-inhibition response. The traces are representative of at least eight observations for each type of channel.

FIG. 8. Delay times, time constants, and relative responses to LSS of oocytes expressing wild-type ENaCs or channels with mutations in the putative amiloride-binding site.
Oocytes expressing mENaCs were perfused with TEV solution, and a computercontrolled perfusion system was used to activate fluid-jet perfusion via a vertical pipette. A, delay times of activation by LSS were estimated as described under "Materials and Methods." Gray bars indicate statistically significant differences between wild-type control versus ␣S583C␤␥ or ␣␤G525C␥ (p Ͻ 0.001) (Kruskal Wallis test (nonparametric ANOVA) followed by Dunn's multiple comparisons post-test). B, time constants of the exponential activation by LSS of wild-type and Cys-substituted mENaCs were estimated as described under "Materials and Methods." ␣S583C␤␥ and ␣␤G525C␥ channels exhibited biphasic increases in current in response to LSS. For ␣S583C␤␥, was determined for the late phase of channel activation in response to LSS. For ␣␤G525C␥, was determined for the early and late phase of channel activation. C, relative response to LSS. The peak response of the whole cell Na ϩ current following the initiation of LSS was normalized to the basal current immediately preceding LSS. The I peak /I basal values were normalized to the I peak /I basal values observed for wild-type ENaC. Gray bars indicate statistically significant differences between wild-type control versus ␣S583C␤␥ (p Ͻ 0.05) or wild-type control versus ␣␤G525C␥ or ␣␤␥G542C (p Ͻ 0.001) (Kruskal Wallis test (nonparametric ANOVA) followed by Dunn's multiple comparisons post-test). Experiments were performed with 6 -26 oocytes for each group. amiloride binding site mutated channels The peak response of the whole cell Na ϩ current following the initiation of LSS was normalized to the basal current immediately preceding LSS. The I peak /I basal values were normalized to the I peak /I basal values observed for wild-type ENaC. The I peak /I basal for ␣␤␥ was 1.35 Ϯ 0.04 (n ϭ 24). icantly affects the kinetic of inhibition by Na ϩ . Shear stress activation and Na ϩ self-inhibition reflect change in ENaC gating, and our previous studies suggest that channels with a high open probability do not respond to LSS or exhibit Na ϩ selfinhibition (13,20,21). We performed single channel recordings of ␣␤␥G542C channels to examine whether these channels had a high open probability. Fig. 10 shows single channel recordings of wild-type or ␣␤␥G542C mutant channels performed with a cell attached configuration. The open probability of ␣␤␥G542C channels was high (0.79 Ϯ 0.11) compared with wild-type channel (0.23 Ϯ 0.09) (see Figs. 10 and 11A). In agreement with previous findings (3), the single channel conductance of ␣␤␥G542C channels (4.5 Ϯ 0.2 pS) was significantly lower than that of wild-type channels (6.9 Ϯ 0.5 pS). These results demonstrate that the ␥G542C mutation alters the gating of ENaC.
To determine the parameters of activation by LSS, the mutant subunits (␣G587C and ␣S589C) were co-injected with subunits carrying deletions in the C-terminal domains (␤R564X (␤T) and ␥R583X (␥T)) that significantly increase whole cell currents (31). Fig. 12 shows representative tracings of oocytes expressing channels with mutations in the selectivity filter (␣G587C␤T␥T and ␣S589C␤T␥T) before and following stimulation by LSS. The time course of activation of oocytes expressing ␣G587C␤T␥T channels was best fit by a bi-exponential function (see "Materials and Methods"). The time course of activation of ␣S589C␤T␥T channels was monoexponential and was significantly slower than that observed for oocytes expressing ␣␤T␥T channels. The delay times for ␣G587C␤T␥T were significantly shorter and for ␣S589C␤T␥T were longer than that observed for ␣␤T␥T channels. Table III shows the values of the delay times, time constants, and relative responses for ␣␤T␥T, ␣G587C␤T␥T, and ␣S589C␤T␥T channels. Interestingly, ␣S589C␤␥ is the only mutant that has an increased delay time and an increased time constant of activation by LSS, when compared with controls (i.e. ␣␤T␥T). These results suggest that the tract ␣Gly 587 -␣Ser 589 contributes to the regulation of gating in addition to serving as a selectivity filter. DISCUSSION Cells are potentially exposed to a variety of mechanical forces including indentations, circumferential stretch, high fre-

TABLE II
Fitting parameters for the Na ϩ self-inhibition response of ␣␤␥ and ␣␤␥G542C mENaCs The time constants were obtained by fitting of the current decay following a sudden increase in bath ͓Na ϩ ͔ from 1 to 110 mM with a single exponential equation. The I peak and I ss represent the peak current and the steady state current, respectively.  quency vibrations, osmotic pressure gradients, hydrostatic pressure, and fluid shear stress (32). In the kidney, frictional forces of ultrafiltrate flowing through tubules regulates the vectorial transport of glucose (33), chloride (34), magnesium (34), potassium (7-10), sodium (8,11), the organization of the cytoskeleton, the synthesis of matrix proteases, and the activity of specific transcription factors (35). Mechanosensitive ion channels are transducers that can respond to mechanical stimuli changing the permeability of the membrane to specific ions. Modifications in channel kinetics could result from changes in membrane tension, thickness, local curvature, or by direct tugging on the protein by cytoskeletal or extracellular tethers (32). The free energy generated by movement of mechanosensitive structures directly associated with the channel is used to generate conformational changes in the protein and finally gate the channel. In addition, for some channels the stimulus maybe force acting directly on the channel.
Genetic analyses of Caenorhabditis elegans touch-insensitive animals permitted the development of a model of the touchtransducing complex in which the degenerins MEC-4 and MEC-10 constitute the channel core of the mechanosensory transduction complex (2). The extracellular domains of the degenerin ion channel subunits are thought to be associated with a specialized extracellular matrix called the mantle, whereas the intracellular domains interact with the cytoskeleton. The channel is thought to be tethered within the extracellular and intracellular domains, allowing for mechanical forces to be transmitted to the channel gating machinery (2). The introduction of amino acid residues with bulky side chains at a key site in the pre-M2 region of MEC-4 and MEC-10, termed the degenerin site, are thought to generate hyperactive channels that result in neurodegeneration (2). Previous studies from our group and others suggest that ENaC pre-M2 regions have an important role in modulating channel gating. For example, introduction of amino acid residues with bulky side chains at the equivalent degenerin site within of the ␤-ENaC subunit (S518K), or chemical modification at the same position of ␣or ␤-subunits (␣S576C or ␤S518C), appears to dramatically enhance channel open probability (16,18,19). In addition, oocytes expressing ENaC with Cys substitutions of key residues in the ␣-subunit within the region immediately preceding ␣Ser 583 responded to sulfhydryl-reactive reagents with rapid increases in whole cell Na ϩ currents that likely reflect increases in channel open probability (16).
We have previously shown that the increase in whole cell Na ϩ current in response to LSS in oocytes expressing ENaC reflects an increase in channel open probability (13). ENaCs with a high intrinsic open probability because of a mutation (␤S518K) or covalent modification of an introduced Cys residue (␣S580C) in the pre-M2 region were not activated by LSS, suggesting that the pre-M2 domain participates in conformational rearrangements during channel activation (13). Cys substitutions at specific sites in the tract extending from ␣Ser 580 to ␣Ser 589 altered the response to LSS. Specifically, the changes in response to LSS included (i) a reduction (Ser 580 -Trp 585 , Gly 587 ) or an increase (Ser 589 ) in delay times preceding channel activation by LSS, (ii) an increase (Gln 581 , Leu 584 , Trp 585 , Phe 586 , Ser 588 ) or a decrease (S589C) in the rate of channel activation, (iii) a decrease in the magnitude of the response (Ser 583 , Gly 587 , Leu 584 ). The tract ␣Ser 580 to ␣Ser 589 is within a region, extending from ␣Ser 580 to ␣Ser 592 , that we proposed transitions from an ␣-helix (preceding ␣Ser 580 ) and includes an extended region that contains the selectivity filter (␣Gly 587 -␣Ser 589 ) (16,30). Interestingly, mutations within a putative ␣-helical region preceding ␣Ser 580 or following the selectivity filter did not affect the channel response to LSS. We propose that mutations that are associated with modifications in the time course of activation are located at regions where conformational changes occur in response to LSS. Our data support the idea that during channel gating by LSS, residues located in the tract ␣Ser 580 -␣Ser 589 experience conformational rearrangements.
A putative amiloride binding site is present within the tract encompassing ␣Ser 580 to ␣Ser 589 . We observed that Cys substitutions at the putative amiloride binding site of each ENaC subunit led to significant changes in the response to LSS. The ␣S583C and ␤G525C mutants exhibited a decrease in the delay time, as well as a biphasic increase in Na ϩ current in response LSS. The magnitude of the LSS response was significantly lower in the amiloride site mutants. The ␥Gly 542 mutant was largely unresponsive to LSS. These channels have a reduced sodium self-inhibition and a high intrinsic open probability. As channel activation by LSS reflect an increase in the open probability, it is not surprising that ␣␤␥G542C channels did not respond to LSS. The mutation ␥G542C was associated with an over 1000-fold increase in the amiloride K i , a 40% decrease in single channel conductance, and Zn 2ϩ -dependent block (K i , 1.9 mM) (25), suggesting that ␥Gly 542 contributes directly (or indirectly) to the conduction pore of ␣␤␥ ENaC. In the context of previous observations, and given the structural roles of Gly residues in proteins (i.e. flexibility contributed by Gly residues), our current results support the possibility that a significant alteration in the pore structure is induced by the ␥G542C mutation.
Model of Channel Activation by LSS-While the nature of the mechanical sensor associated with ENaC is unknown, it is likely that spatial and temporal variations of flow rate in renal tubules produce a deformation of a mechanical sensor that is transduced to the channel gate. Work performed as result of shear distortion of mechanosensitive structures facilitate conformational changes that eventually gate the channel. We assume that the process of activation of ENaC by flow involves transitions between discrete closed and open states separated by large energy barriers. Single channel studies of wild-type ENaCs performed under basal conditions in epithelial cells have described the presence of at least 2 open states and 2 close states, with mean open and close times in the order of either tenth of seconds or seconds (36 -39). For a sequential gating scheme, the activation process of ENaC by LSS can be described by Reaction 1, The delay in channel activation following initiation of LSS can be modeled with numerous steps with variable rates preceding the increase in the open probability of the channel (40). A single slow step may have a dominant effect on the duration of the delay preceding channel activation, while the contribution of steps that are relatively rapid may be minimal (40). For ENaC activation in response to LSS, the delay time may reflect several components, including: (i) the time necessary for the mechanosensor to sense the mechanical stimulus, (ii) transmission of the stimulus from the mechanosensor to the channel gate, and (iii) changes in conformation of the gate that allow for ions to cross the channel. Our data suggest that ENaCs experience transitions through electrically silent steps during the activation process prior to increases in channel open probability, as reflected by the changes in delay times observed. If the pre-M2 region adjacent to and including the selectivity filter functions as ENaC primary gate, mutations in this region might affect rates of transitions between closed and open states and result in the changes in delay times and time constants that we observed with mutants in the tract ␣Ser 580 -␣Ser 589 . The modifications observed in delay times and time constants suggest that a significant number of steps occur before channel activation, which likely involves conformational rearrangements in the pore region. Although in the classical description of voltage-dependent channels the mechanisms of gating and ionic conduction were independent, subsequent experiments indicated that the pore structure influences the gating process of Na ϩ , K ϩ , CNG, and Ca 2ϩ channels (41)(42)(43)(44)(45)(46)(47)(48)(49). For example, using electron paramagnetic resonance spectroscopy (EPR) conformational rearrangements of residues preceding the selectivity filter of the KscA channel were observed in association with openings (50). The structure of the ENaC pore region is unknown. However, based on our results we predict that during channel gating, the tract extending from ␣Ser 580 to ␣Ser 589 experiences conformational rearrangements influencing channel openings and closures. Cys substitutions in the selectivity filter of mENaCs result in substantial changes in the time course of activation, suggesting that selectivity and gating are not independent processes. Two molecular models have been proposed to describe the gating of mechanosensitive channels in response to mechanical stimuli, the bilayer model and the tethered model. The bilayer model proposed that tension developed in the bilayer alone in response to a mechanical force directly gates the channel, whereas the tethered model proposed that mechanical forces are transmitted directly to the channel through the cytoskeleton or extracellular tethers (32). In prokaryotes, mechanosensitive channels protect the bacteria from lysis upon osmotic shock. The large conductance mechanosensitive channel (MscL) of the Escherichia coli inner membrane forms a penta-TABLE III Delay times, time constants, and relative response of ␣␤T␥T and selectivity filter-mutated channels The peak response of the whole cell Na ϩ current following the initiation of LSS was normalized to the basal current immediately preceding LSS. The I peak /I basal values were normalized to the I peak /I basal values observed for controls (␣␤T␥T). The I peak /I basal for ␣␤T␥T was 1.41 Ϯ 0.08 (n ϭ 11). mer that is activated by membrane stretch (51). Following purification and reconstitution of the channel in liposomes, it was demonstrated that MscL opens directly by tension applied to membrane lipids (52,53). The threshold pressure at which channels are opened could be decreased by substitution of Gly 22 with hydrophilic residues, or increased by substitution with hydrophobic residues (54). Based on the close proximity of residues in the pre-M2 and M2 domains of ENaC with the lipid bilayer, it is possible that ENaC responds to LSS by sensing the deformation of the lipid bilayer produced during flow stimulation. Alternatively, ENaC might be tethered by intracellular and extracellular elements. Members of the degenerin/ENaC family include putative mechanosensitive ion channels found in C. elegans and in mammalian neurons (2,(55)(56)(57). Genetic analyses performed in touch-insensitive mutants in C. elegans led to the identification of several proteins that could potentially interact with the intracellular and extracellular domains of degenerins and serve as tethers (2,57). Whereas interactions have been demonstrated between ENaC and components of the cytoskeleton (58 -62), interactions between ENaC extracellular domains and components of the extracellular matrix have not been described. Although the mechanism by which ENaC senses mechanical forces, such as LSS, have not been defined, our data suggest that residues in the tract ␣Ser 580 -␣Ser 589 experience conformational rearrangements during channel gating by LSS. Additionally our results suggest that the selectivity filter influences or affects the gating of the channel during activation by LSS.