Mechanism of Inverted Activation of ClC-1 Channels Caused by a Novel Myotonia Congenita Mutation*

The voltage-gated chloride channel ClC-1 is the major contributor of membrane conductance in skeletal muscle and has been associated with the inherited muscular disorder myotonia congenita. Here, we report a novel mutation identified in a recessive myotonia congenita family. This mutation, Gly-499 to Arg (G499R) is located in the putative transmembrane domain 10 of the ClC-1 protein. In contrast to normal ClC-1 channels that deactivate upon hyperpolarization, functional expression of G499R ClC-1 yielded a hyperpolarization-activated chloride current when measured in the presence of a high (134 mm) intracellular chloride concentration. Current was abolished when measured with a physiological chloride transmembrane gradient. Electrophysiological analysis of other Gly-499 mutants (G499K, G499Q, and G499E) suggests that the positive charge introduced by the G499R mutation may be responsible for this unique gating behavior. To further explore the function of domain 10, we mutated two charged residues near Gly-499 of ClC-1. Functional analyses of R496Q, R496Q/G499R, R496K, and E500Q mutant channels suggest that the charged residues in domain 10 are important for normal channel function. Study of these mutants may shed further light on the structure and voltage-gating of this channel.

The voltage-gated chloride channel ClC-1 is the major contributor of membrane conductance in skeletal muscle and has been associated with the inherited muscular disorder myotonia congenita. Here, we report a novel mutation identified in a recessive myotonia congenita family. This mutation, Gly-499 to Arg (G499R) is located in the putative transmembrane domain 10 of the ClC-1 protein. In contrast to normal ClC-1 channels that deactivate upon hyperpolarization, functional expression of G499R ClC-1 yielded a hyperpolarization-activated chloride current when measured in the presence of a high (134 mM) intracellular chloride concentration. Current was abolished when measured with a physiological chloride transmembrane gradient. Electrophysiological analysis of other Gly-499 mutants (G499K, G499Q, and G499E) suggests that the positive charge introduced by the G499R mutation may be responsible for this unique gating behavior. To further explore the function of domain 10, we mutated two charged residues near Gly-499 of ClC-1. Functional analyses of R496Q, R496Q/G499R, R496K, and E500Q mutant channels suggest that the charged residues in domain 10 are important for normal channel function. Study of these mutants may shed further light on the structure and voltage-gating of this channel.
The skeletal muscle chloride channel is a member of the ClC family of voltage-gated chloride channels. Each ClC-1 subunit is a 110-kDa protein composed of 13 putative transmembrane domains (26 -28). Evidence supports at least two different transmembrane topologies for a single ClC-1 subunit (28,29) that co-assemble as a homomultimeric channel (18,27,30,31). Unlike voltage-gated sodium and potassium channels, in which the voltage dependence is coupled to the movement of several charged residues located on a transmembrane segment of the protein (32), no similar domain is present in the ClC-1 subunit. There are two models to account for the voltage-dependent gating of ClC channels. According to one model, the voltagedependent opening of the ClC channel depends on the binding of chloride ions to the channel and the movement of these ions through the conduction pathway (33)(34)(35). The second model proposed the existence of an intrinsic voltage sensor (36), in which bound anions interact with a voltage-dependent gate. However, in either model, the domains involved with voltage sensing or anion binding remain unclear. In contrast to WT ClC-1 channels that normally deactivate in response to hyperpolarization under physiologic conditions, one ClC-1 mutant (D136G) has been shown to activate upon hyperpolarization (36,37). A similar behavior of activation with hyperpolarization was observed for WT channels when studied at low extracellular pH (38). The mechanism for this behavior is unknown.
In this study, a novel Gly-499 3 Arg (G499R) mutation was identified in a recessive MC (Becker's disease) patient. Functional analysis revealed that G499R ClC-1 channels activate upon hyperpolarization, similar to D136G ClC-1 channels. Sitedirected mutagenesis at several sites in putative transmembrane domain 10 (D10) revealed the importance of electrostatic effects in this unique gating behavior and suggested an important role for D10 in the voltage-dependent gating of the ClC-1 channel.

EXPERIMENTAL PROCEDURES
Identification of Patients-A Polish kindred ( Fig. 1A; father, mother, and son) was evaluated, and the son was diagnosed with Becker's MC. The proband and his parents underwent a complete neurological exam, with particular attention directed toward the neuromuscular system, including a standard electromyographic examination. A blood sample was drawn from each member of the family for DNA extraction.
PCR Amplification of DNA from Study Subjects-Intronic sequence was used to design primers (19) that allowed amplification of CLCN1 exons from genomic DNA. The 10-l PCR mixture contained 50 ng of genomic DNA, 6.7 l of dH 2 O, 0.6 l of dNTP (125 M of each deoxynucleoside triphosphate), 0.3 l (25 pmol/l) of forward and reverse primer, 1 l of reaction buffer (500 mM KCl, 100 mM Tris-HCl, 1.5 mM MgCl 2 , 0.01% gelatin), 0.25 units of Taq DNA polymerase (5 units/l, Perkin-Elmer Corp.), and 0.1 l of [␣-32 P]dCTP (Amersham Pharmacia Biotech). The mixtures were initially denatured at 94°C for 3 min followed by 30 cycles each of 30 s for denaturing at 94°C, 30 s for annealing at temperature, and 30 s for extension at 72°C.
Single Strand Conformation Polymorphism (SSCP) Analysis and Sequencing-SSCP was carried out using previously described methods (23). Briefly, the PCR products were diluted and denatured in 50 l of 0.1% SDS/10 mM EDTA. After the addition of loading dye, the mixtures were heated at 94°C for 3 min. The denatured mixture (2 l) was separated by electrophoresis through a 5% nondenaturing polyacrylamide gel at 40 W for 3-7 h. The gels were run under the following two conditions: room temperature with glycerol, and 4°C without glycerol. Gels were transferred to filter paper, dried on a vacuum slab dryer for 1 h at 85°C, and then exposed to x-ray film at Ϫ20°C for 12-24 h. Eluted DNA (10 l) from the aberrant bands was reamplified using both the original PCR primers pair with additional M13 sequencing primer tails. Samples were purified by a centrifugation wash with a Centricon-100 column (Amicon, Millipore Corp., Bedford, MA) and then sequenced on an Applied BioSystems model 373A DNA sequencer using the dideoxy termination method.
Site-directed Mutagenesis-Two missense mutations, G499R and G482R (21), and seven point mutations (see under "Results") were introduced into the pRc/CMV-ClC-1 vector by standard two-step PCRbased site-directed mutagenesis. All PCRs were conducted using Pfu DNA polymerase (Stratagene, La Jolla, CA) for high fidelity amplifications. Two fragments were amplified in the first step, using either primer pair A/B or pair C/D, which contained the desired mutations, and pRc/CMV-ClC-1 as a template. In the second step, the two partial overlapping fragments were joined by annealing and extended by polymerization. These full-length templates were amplified using primer pair A/D. The final products were digested with the appropriate restriction enzymes (HindIII (Promega, Madison, WI) and Eco47 III (Stratagene)) and ligated into the pRc/CMV-ClC-1 vector. All PCRs were carried out in a DNA Engine Tetrad (M.J. Research, Waltham, MA). The inserts were sequenced to exclude any polymerase errors.
Cell Culture and Transfection-Human embryonic kidney (HEK)-293 cells (American Type Culture Collection, CRL 1573) were grown on Dulbecco's modified Eagle's medium; supplemented with penicillin, streptomycin, and 10% fetal calf serum (Life Technologies, Inc.); and maintained at 37°C with 5% CO 2 . One day after the cultures were split, the calcium/phosphate precipitation technique (39) was used for cell transfection. About 0.4 g of the appropriate pRc/CMV-ClC-1 plasmid DNA was used for each transfection in 35-mm culture dishes. To allow identification of transfected cells during patch clamp experiments, cells were co-transfected with green fluorescent protein plasmid cDNA (CLONTECH Inc., Palo Alto, CA). Approximately 15 h after transfection, cells were split again, and the medium was changed.
Electrophysiology-Currents were recorded from 36 to 48 h after transfection by the whole-cell configuration of the patch clamp technique (40) using an Axopatch 200 B amplifier (Axon Instruments Inc., Foster City, CA). Data were acquired on-line by a personal computer with pClamp6 software and a Digidata 1200 A/D interface (Axon Instruments Inc.). Currents were low pass filtered at 2 kHz and digitized at 10 kHz. The external solution contained 140 mM NaCl, 4 mM KCl, 1 mM MgCl 2 , 2 mM CaCl 2 , and 5 mM HEPES at pH 7.4. The standard internal pipette solution contained 130 mM NaCl, 2 mM MgCl 2 , 5 mM EGTA, and 10 mM HEPES at pH 7.4. A lower internal Cl Ϫ concentration was achieved by equimolar substitution with sodium glutamic acid. Recording pipettes were pulled from borosilicate glass (Kimax, Kimble Glass Inc. Vineland, NJ) by a Sutter P-87 puller (Novato, CA) and were heat-polished with an MF-83 microforge (Narishige Scientific Instruments, Japan). Pipettes had a resistance of between 0.8 and 1.5 M⍀ when filled with standard internal solution. The junction potential was adjusted to zero for each experiment, and series resistance was compensated 80 -90% by analog circuitry.
Pulse Protocols and Data Analysis-The current-voltage (I-V) relationship for ClC-1 channels was determined from a holding potential of Ϫ35 mV. Currents were elicited by 180-ms test pulses ranging from Ϫ190 to ϩ170 mV. Instantaneous currents were determined by extrapolation of fitted current traces to the beginning of each test potential. The isochronal I-V relationship was obtained by measuring current at the end of each test potential.
The voltage dependence of relative open probability (P o ), was obtained by measuring the instantaneous current, or the peak of tail current (I V ) at Ϫ135 mV after 1.4-s prepulses ranging from Ϫ130 to ϩ170 mV. Tail currents were fitted by a two-exponential function, and the peaks of tail currents obtained by extrapolation were plotted as a function of test voltage and then fitted to a Boltzmann function, where I max is the maximal current, I o is a constant offset, V1 ⁄2 is the half-maximal activation voltage, and dx is the slope factor. The relative P o was obtained by the normalization P o ϭ I V /I max . Current deactivation was determined using a 2-pulse protocol. Cells were held at Ϫ35 mV, and then the voltage was stepped to ϩ45 mV for 500 ms. Tail currents were measured during a subsequent 180-ms test pulse to potentials ranging from Ϫ135 to ϩ45 mV, applied in 20-mV incremental steps. The time courses of current deactivation and activation were determined by fitting current traces with a two-exponential function, All data are shown as mean Ϯ S.E. Statistical evaluation was performed using Student's t test and two-way analysis of variance.

RESULTS
A Novel G499R Mutation-Two different aberrant conformers were observed in SSCP analysis from DNA samples in this MC kindred (Fig. 1A). DNA sequencing revealed both a novel C-1495-T (antisense strand) transition in exon 14 (Fig. 1B) and a previously identified nonsense mutation (C-2680-T) in exon 23 of the CLCN1 gene. When read in-frame, these mutations predict a Gly to Arg change at amino acid position 499 and an Arg to Stop change at amino acid position 894 in the ClC-1 protein, respectively. Additionally, both mutations were found in the son diagnosed with Becker's MC (Fig. 1A). Two hundred unrelated normal alleles have been screened for this C-1495-T conformer by SSCP analysis, and no similar aberrant pattern has been observed (data not shown). According to the putative channel topology, this G499R mutation is located at the end of the putative transmembrane D10 segment ( Fig. 2A). Sequence alignment shows that Gly-499 is highly conserved among the different ClC homologues (Fig. 2B).
Functional Expression of G499R in HEK-293 Cells-In order to understand the functional consequence of this novel ClC-1 mutation and its role in MC, we characterized the biophysical properties of the G499R ClC-1 channels expressed in HEK-293 cells. Currents recorded from HEK-293 cells expressing either WT ClC-1 or G499R ClC-1 are shown in Fig. 3. Membrane hyperpolarization (from ϩ45 mV) of cells expressing WT ClC-1 channels elicited a large inward current that rapidly deactivated. At steady state, the maximum inward current saturated near Ϫ100 mV (Fig. 3A, left). Upon depolarization (from Ϫ135 mV), WT channels activated and reached saturation near ϩ100 mV (Fig. 3A, right). In contrast, the G499R mutant channels activated upon hyperpolarization did not deactivate nor saturate, even at Ϫ220 mV (Fig. 3B, left). Unlike WT ClC-1 channels, mutant channels closed upon depolarization (Fig. 3B,  right).
The amplitude of G499R ClC-1 current was smaller than that of WT ClC-1 channel (Fig. 4, A and B). For example, at Ϫ130 mV, the mean instantaneous current amplitude for WT was 3.0 Ϯ 0.7 nA, and for G499R was 0.4 Ϯ 0.03 nA. G499R ClC-1 also exhibited more pronounced inward rectification than WT channels, and effects were notable in both instantaneous and isochronal I-V relationships (Fig. 4, C and D). In addition to the decrease in current amplitude, the voltage dependence of the relative open probability (P o ) of G499R ClC-1 channels was dramatically altered. In the voltage range from Ϫ200 to ϩ100 mV, the P o -voltage curve for WT ClC-1 could be described by a single Boltzmann function (V1 ⁄2 ϭ Ϫ64.1 Ϯ 3.0 mV; slope factor, 24.3 Ϯ 0.8 mV). In contrast to WT, the voltage dependence of P o for G499R ClC-1 channels was completely reversed (Fig. 4E). In the voltage range from Ϫ200 to ϩ50 mV, the V1 ⁄2 was Ϫ56.6 Ϯ 2.4 mV, with a slope factor of Ϫ22.2 Ϯ 2.2 mV.
To further investigate the pathogenic role of the G499R mutant channel in MC, we also studied its function using a physiological transmembrane gradient of chloride concentration. Current traces recorded with 5 mM intracellular (pipette) chloride from both the WT ClC-1 and the G499R ClC-1 channels are shown in Fig. 5, A and B. The G499R mutation caused a dramatic reduction in chloride conductance (Fig. 5, C and D). For example, at Ϫ105 mV, the average instantaneous current amplitude for WT ClC-1 was 0.7 Ϯ 0.1 nA, but it was only 0.01 Ϯ 0.004 nA for the G499R mutant. The dramatic reduction of G499R current at a physiological chloride concentration fully explains its pathogenic role in MC.
Functional Analysis of G499K, G499Q, and G499E Mutations-In order to understand how the G499R mutation causes such a dramatic change in the gating of the ClC-1 channels, we constructed several missense mutations of Gly-499 in the ClC-1 protein. First, to investigate whether or not Arg is necessary for this unique voltage-dependent gating, we replaced the Gly at position 499 with Lys, which is similar to Arg in both size and charge. If the unusual gating of G499R ClC-1 channels was only charge-related, then the properties of G499K should be similar to G499R. Fig. 6, A and B, shows current traces recorded from HEK-293 cells transfected with G499K ClC-1 cDNA. As predicted, these currents were similar to G499R ClC-1 (Fig. 3, A and B). The P o -voltage curve for G499K ClC-1 was also similar to G499R ClC-1 (Fig. 6C, V1 ⁄2 ϭ Ϫ56.9 Ϯ 4.6 mV; slope factor, Ϫ17.4 Ϯ 2.3 mV; for G499K). In the voltage range of Ϫ130 to ϩ50 mV, the V1 ⁄2 of the P o -voltage curve was the same as that for G499R (p Ͼ 0.05). This hyperpolarizationactivated gating of the G499R and G499K mutants could result from either electrostatic or structure changes (or both) at this residue.
To distinguish among these possibilities, we then introduced a Gln at position 499. This residue is similar in size to Arg and Lys but neutral in charge. The gating of G499Q ClC-1 channels (Fig. 7, A and B) was similar to WT ClC-1 channels (Fig. 3A), suggesting that the hyperpolarization activation is an electrostatic effect of the G499R mutant. However, unlike WT ClC-1 channels, the P o -voltage curve of G499Q ClC-1 channel currents could not be simply fitted by single Boltzmann function (Fig. 7C).
To further evaluate the importance of the electostatic forces at this position, we mutated this residue to Glu, which is similar in size to Arg but negatively charged. The G499E channels did not exhibit the inverted-gating property of G499(R/K). Instead, G499E ClC-1 channels deactivated slowly upon hyperpolarization from a holding potential of ϩ45 mV and exhibited much slower deactivation kinetics (Fig. 8, A and B). In contrast to the WT deactivation requiring a fit with two exponential functions, only a single time constant was needed to describe the deactivation of G499E ClC-1 channels (Fig. 8C). Furthermore, the P o -voltage relationship of G499E ClC-1 channels was very different from that of the WT ClC-1 channels (Fig. 8D). In response to voltage steps from negative prepulse potentials (Ϫ130 to Ϫ10 mV) to the test potential (Ϫ135mV), the tail current remained constant (P o ϭ 0.3) even though the conductance (g Cl ) decreased throughout this range. That is, conductance decreased with decreasing driving force, but channel gating was not voltage-dependent in this range of potentials (Fig.  8D). The P o of G499E increased over the range of positive potentials and could be described by a single Boltzmann func-tion with a V1 ⁄2 of 89.7 Ϯ 4.9 mV and a slope factor of 48.2 Ϯ 3.2 mV (Fig. 8D). These data indicate that substituting a negative residue for Gly-499 not only abolished the inverted voltage-dependent gating property of G499R but also impaired the closing of ClC-1 channels at negative potentials.
Functional Consequences of R496Q, R496Q/G499R, R496K, E500Q, and G482R Mutations in ClC-1 Channels-Due to the importance of charged residues at position 499 in D10, we also examined Arg-496, the only charged residue in D10 of the WT channel (Fig. 2). Expression of channels with Arg-499 changed to a neutral residue (Gln) did not yield any detectable chloride current (data not shown). Because Arg-496 and Gly-499 are in close proximity in the putative ␣-helical structure for D10, we hypothesized that G499R might compensate for the charge loss of the R496Q mutation. However, the double mutant (R496Q/ G499R) did not functionally express (data not shown). We replaced Arg-496 with Lys (R496K) and showed that channel current was similar to that of WT (Fig. 9, A and B). However, the P o -voltage curve was shifted to more negative potentials when compared with WT (Fig. 9C, V1 ⁄2 ϭ Ϫ97.8 Ϯ 5.2 mV; p Ͻ 0.05). Taken together, these results suggest that a positive charge at position 496 is essential for normal channel function.
Another naturally occurring MC mutation is known to introduce an additional charged residue in D10. The G482R mutation was originally identified in a Becker's MC family (21) (Fig.  2). As with the R496Q mutant, expression of G482R ClC-1 channels did not yield detectable current (data not shown).
To further investigate the importance of charge in this region of ClC-1, we also mutated Glu-500, the first charged residue in the putative loop between D10 and D11 (Fig. 2), to Gln (E500Q). Expression of E500Q ClC-1 channels yielded a current similar to WT (Fig. 10, A and B) without having significant effects on the voltage dependence of P o (p Ͼ 0.05; Fig. 10C). These data indicate that electrostatic effects due to charge alterations at position 500 are not as important for ClC-1 function as changes at position 482, 496, or 499. G499R-MC has been recognized for over a century (41). The muscle stiffness and delayed relaxation observed in MC are the result of repetitive electrical discharges and hyperexcitability of muscles that is known to result from decreased Cl Ϫ conductance of sarcolemma (1-6). However, elucidation of the molecular basis of this disease was not possible until the cloning of CLCN1 gene. Although mutational analysis and functional characterization of ClC-1 channels have provided some insights into MC, the mechanisms of disease need to be further defined and may shed light on the function of WT channels.

Mechanism of MC Caused by
In the present study, we identified a novel G499R mutation and a previously known nonsense R894X mutation (14,23) in a Becker's MC family. Our genetic analysis indicates that G499R and R894X are transmitted as recessive alleles. Under physiological ionic conditions, the conductance of the G499R ClC-1 channel was undetectable. Thus, G499R ClC-1 channels are unlikely to contribute any chloride conductance in muscle membranes. A similar phenomenon has also been described for the D136G mutation (36). Despite the fact that R894X mutation produces a truncated protein, expression of mutant channels was reported to cause a reduction but not a complete abolition of chloride conductance (21). Therefore, it is likely that any muscle membrane chloride conductance in our patient results from channels encoded by the R894X allele.
Implication of G499(R/K) for a Model of Inverted ClC-1 Gating-Hyperpolarization-activated gating behavior of ClC-1 observed in the G499(R/K) mutants has been noted previously in a naturally occurring Becker's mutation, D136G (36), and in WT ClC-1 channels when external pH was lowered to 5.5 (38). In neither case is the factor responsible for this unique voltagedependent gating known. We have demonstrated that the introduction of a positive charge at position 499 (G499R or G499K) of D10 could also evoke this unique gating of ClC-1. It has been proposed that D136G, located in D1, could be involved in voltage-dependent gating of ClC-1 channels (36). Similar gating defects caused by both G499R and D136G mutations suggest that charge changes at these positions of ClC-1 could affect voltage-dependent gating in a similar way in these mutant channels.
The mechanism underlying the voltage-dependent gating of ClC-1 channels remains unclear. However, it has been proposed that Cl Ϫ may act as a gating charge in the depolarization-activated gating of ClC channels (33)(34)(35). According to this model, WT channels have Cl Ϫ binding sites that are sensitive to external Cl Ϫ concentrations. The voltage dependence of gating is therefore determined by the binding of Cl Ϫ to these sites and the movement of Cl Ϫ inside the conducting pore. Alternatively, Cl Ϫ also may interact with an intrinsic voltage sensor, in which the gating of the ClC-1 channel could be affected by anion binding to the sites lying along the pore pathway (36,42,43). In these models, the voltage dependence of opening in ClC-1 is dependent upon the binding of Cl Ϫ to these sites. Although evidence supports the role of Cl Ϫ in voltage-dependent gating, more studies are needed to identify the putative Cl Ϫ binding site. It is possible that the charge reversal in D10 (or D1) affects the efficiency of anion binding, which in turn alters open probability of the channel.
Mutation of Gly-499 to a positively charged residue (Arg or Lys) inverted the voltage dependence of gating of ClC-1 such that P o was maximal at very negative and positive transmembrane potentials, with a minimum value near 0 mV. These two phases of the P o -voltage curve suggests the existence of two  5). D, the voltage dependence of relative P o for G499E ClC-1 channels (filled squares) (n ϭ 10). The V1 ⁄2 for G499E ClC-1 channels was 89.7 Ϯ 4.9 mV (slope factor, 48.2 Ϯ 3.2 mV). The P o curve for wild-type is shown by the dashed line (from Fig. 4E). different gating mechanisms in these mutant channels. To explain this, we propose a model based on the chloride-dependent mechanisms of gating described above. Because the inverted gating behavior was only observed for the G499R and G499K mutants, we propose that an additional positively charged residue in D10 may introduce a new chloride binding site near the conducting pore of the ClC-1 channel. This new binding site could be occupied by intracellular Cl Ϫ in a voltagedependent manner. Thus, hyperpolarization to a negative transmembrane potential favors the binding of intracellular Cl Ϫ to this new site due to the electrostatic forces, which leads to an increase of channel open probability (Figs. 4E and 6C). Upon depolarization to positive transmembrane potentials, the chloride binding at the original (extracellular) sites will be favored from the outside, and the P o would start to increase (Figs. 4E and 6C). Previous findings indicated that the P o and kinetics of ClC-1 channels depend on external but not internal Cl Ϫ concentration (38). In contrast, the putative new binding site makes the G499(R/K) ClC-1 channels more sensitive to intracellular Cl Ϫ concentration than WT channels. A low intracellular Cl Ϫ concentration (e.g. 5 mM) will not favor internal Cl Ϫ binding and therefore will cause a decrease of inward current (Fig. 5). According to this model, the descending phase of the P o -voltage relationship for G499(R/K) ClC-1 channels at negative membrane potentials is due to a gating mode modulated by the new Cl Ϫ binding site. In contrast, the ascending phase of P o in these mutant channels at positive membrane potential likely represents normal channel gating involving the normal anion binding (sites) mechanism as that described in WT ClC-1 channels. Both G499R and D136G mutations result in the addition of positive charge or reduction of negative charge. Very likely, as in G499R ClC-1 channels, the charge change plays a similar role in the hyperpolarization-activated D136G ClC-1 channels.
The anion-dependent gating mechanism could also be applied to the G499E mutant channels. Concerning the importance of the electrical charge on D10, the additional negative charge at position 499 could disrupt the normal chloride-binding sites of WT ClC-1 channels, or at least electrostatically repulse the presumed gating charge and disrupt Cl Ϫ binding to the extracellular sites. In this case, the channel opening is not voltage-dependent at negative potentials (Fig. 8D). However, this disruption of Cl Ϫ binding may be compensated by positive transmembrane potentials in a voltage-dependent manner. As the membrane becomes more depolarized, binding of Cl Ϫ to the external binding site increases, and the open probability of G499E channels increases (Fig. 8D). Falhke et al. (44) have proposed the existence of a negatively charged cytoplasmic gate to explain the deactivation of ClC-1 channels. The electrostatic repulsion between Glu-499 and this putative gate also may affect the kinetics of ClC-1 channels, in which markedly slower kinetics were observed (Fig. 8C). Our observations on G499R, G499K, and G499E ClC-1 channels further support the role of chloride ions in the voltage-dependent gating of this channel.
One may argue an alternative explanation for the biphasic P o -voltage relationship for G499(R/K) channels based on the double-barreled model of ClC-1 channels. For example, a G499R homodimer containing one normal pore and a second pore that is affected by the mutation could have a descending P o -voltage relationship (due to the mutant pore) and an ascending P o -voltage relationship (due to the WT pore). The sum of two such processes could give the biphasic relationship that we observed. However, recent experimental data on ClC-0 (27,45) and ClC-1 (45) suggest that functional voltage-gated chloride channels are formed as homodimers, in which each pore is very likely formed by a single subunit instead of different portions of both subunits. These data suggest that different contributions FIG. 9. R496K ClC-1 channel currents. Currents were recorded from HEK-293 cells transfected with R496K ClC-1 cDNA, using deactivating (A) and activating (B) protocols. C, voltage dependence of relative P o for R496K ClC-1 channels (filled squares) (n ϭ 5). The V1 ⁄2 for the R496K mutant was Ϫ97.8 Ϯ 5.2 mV (slope factor, 32.0 Ϯ 1.6 mV). The P o curve for wild-type is shown by the dashed line (from Fig. 4E).
FIG. 10. E500Q ClC-1 channel currents. Currents were recorded from HEK-293 cells transfected with E500Q ClC-1 cDNA, using deactivating (A) and activating (B) protocols. C, voltage dependence of relative P o for E500Q ClC-1 channels (filled squares) (n ϭ 5). The V1 ⁄2 for E500Q ClC-1 channels was Ϫ49.8 Ϯ 5.6 mV (slope factor, 31.3 Ϯ 2.0 mV). The P o curve for wild-type is shown by the dashed line (from Fig.  4E). by two barrels is an unlikely explanation for the biphasic P o -voltage relationship observed in vitro for homodimeric mutant channels. However, controversy still exists regarding the single-versus double-barrel structure of ClC-1 channels.
Implication of D10 for ClC-1 Gating-The Arg at position 496 is the only charged residue in D10 and is highly conserved among all ClC channels (Fig. 2). Similar to R496Q, the naturally occurring mutation R496S also did not yield any detectable current (17). Furthermore, the G499R mutation failed to compensate for the R496Q charge loss, although they are thought to be in close proximity based on the putative ␣-helical structure of D10. This suggests that the Arg at position 496 is critical for normal channel function. We cannot exclude the possibility that the positively charged residue in D10 (R496) interacts with normal anion-binding sites or alters gating by an allosteric mechanism via interaction with another domain in the channel. The neutralized Gln might impair or weaken this binding site and, therefore, produce a nonconducting channel. Alternatively, we also cannot exclude the possibility that Arg-496 is important as a part of an intrinsic voltage-sensor in WT ClC-1 channels, similar to the mechanism of D136G proposed by Fahlke et al. (36,44). In this scenario, the replacement with Gln at this position would disrupt the voltage-sensor, resulting in a nonfunctional channel. Interestingly, the replacement of Arg-496 with Lys favors channel opening (Fig. 9C). The parallel leftward shift of the P o -voltage relationship further suggests that this charged residue in D10 is involved in the voltage-dependent gating of ClC-1 channels. Surprisingly, we did not observe any significant effect when the highly conserved negatively charged Glu at position 500 was replaced with Gln. This implies that this charged residue, located in the loop between D10 and D11, is not as critical as those in D10 for the voltagedependent gating of ClC-1 channels. Finally, the G482R mutation, located at the amino-terminal end of D10, completely abolishes chloride conductance. In contrast to the G499R mutation, in which the channels are still conducting, this observation suggests that the G482R may also affect folding, trafficking, or degradation of the ClC-1 channels.