Functionally, the dimeric human skeletal muscle chloride channel hClC-1 is characterized by two distinctive gating processes, fast (protopore) gating and slow (common) gating. Of these, common gating is poorly understood, but extensive conformational rearrangement is suspected. To examine this possibility, we used FRET (fluorescence resonance energy transfer) and assessed the effects of manipulating the common-gating process. Closure of the common gate was accompanied by a separation of the C-termini, whereas, with opening, the C-termini approached each other more closely. These movements were considerably smaller than those seen in ClC-0. To estimate the C-terminus depth within the cytoplasm we constructed a pair of split hClC-1 fragments tagged extracellularly and intracellularly respectively. These not only combined appropriately to rescue channel function, but we detected positive FRET between them. This restricts the C-termini of hClC-1 to a position close to its membrane-resident domain. From mutants in which fast or common gating were affected, FRET revealed a close linkage between the two gating processes with the carboxyl group of Glu232 apparently acting as the final effector for both.
- common gating
- conformation change
- cytoplasmic domain
- human skeletal muscle chloride channel (hClC-1)
- fluorescence resonance energy transfer (FRET)
Like other members of the CLC family of ion channels and transporters, the human skeletal muscle Cl− channel hClC-1 occurs as a homodimer. Each subunit in these CLC proteins has its own individual Cl−-permeation pathway with complex exchanging and/or gating mechanisms to catalyse ion flow. For hClC-1 and the Torpedo electroplaque Cl− channel ClC-0, two distinct gating processes have been identified. One is fast (protopore) gating which regulates each individual pore independently, and the other is slow (common) gating, which operates on both pores simultaneously. It is generally agreed that the basis for fast gating is the intermittent movement of a glutamate side-chain carboxyl group into or out of the permeation pathway in competition with Cl− [1–3]. In contrast, the nature of common gating seems to be much more complicated and the picture is, at present, still far from clear [4–6].
One major hypothesis, supported by a variety of experimental observations, suggests that common gating involves alterations in subunit–subunit interactions and large conformational changes. Deactivation kinetics, especially for common gating in ClC-0, is unusually temperature-dependent, pointing to a complex rearrangement of the proteins . Again indicative of a large structural reorganization, extracellular Zn2+ inhibition of both ClC-0 and hClC-1 is highly temperature-dependent, with common gating being affected, but not fast gating [5,8]. In hClC-1, electrophysiological studies of naturally occurring dominant myotonic mutations and similar site-directed mutations reinforce the view that common gating involves helices at the dimer interface . Also, overall apparent channel-open probability and voltage-dependence, which are known to be dependent on common gating, are affected by mutations or truncations in the cytoplasmic domain [9–14], probably through non-covalent interaction between the cytoplasmic domain and the membrane-resident pore region . We have taken a particular interest in common gating in hClC-1 because it is typically aberrant in dominant myotonia congenita (Thomsen's disease) [5,15,16], and because its characteristics in hClC-1 are quite different from those in the more-frequently studied ClC-0. It has a temperature sensitivity that is considerably lower, an opposite voltage-dependence to that of ClC-0, being activated by depolarization, and its kinetics are orders of magnitude faster . Progress has been made towards determining the nature of the conformational changes associated with common gating in ClC-0 , with large displacements of the carboxyl termini (C-termini) of the channel being observed through the use of FRET (fluorescence resonance energy transfer).
In the present study, we have also used FRET to investigate conformational changes during common gating in hClC-1. Our analyses were confined to spectra FRET measurements from the plasma membrane, thereby focusing on correctly trafficked and assembled channel proteins. Open and closed states of the common gate were then found to correspond to differing separations of the channel's C-termini, under the assumption that FRET efficiency reflects distance changes between tagged Cerulean and eYFP (enhanced yellow fluorescent protein) fluorophores driven by host protein conformation changes [19,20]. As well, lack of knowledge about the configuration of the cytoplasmic domain of hClC-1, in conjunction with its extraordinary length, led us to try to estimate the C-terminus depth within the cytoplasm. From a pair of functionally reconstituted split-channel constructs, one with an extracellular fluorophore and the other tagged intracellularly, we have used FRET to place a limit on the transmembrane distance by which the fluorophores are separated and, thereby, to gauge the approximate cytoplasmic position of the C-terminus. Our findings are supported by the recent X-ray crystallographic determination of the close conjunction between the cytoplasmic and membrane-resident domains of a eukaryotic CLC protein, CmCLC from Cyanidioschyzon merolae . Finally, we have made additional FRET measurements on hClC-1 channels carrying point mutations that affect gating in specific ways. From these we conclude that fast and common gating are closely coupled, thereby challenging the generally held view that these two gating processes are independent.
MATERIALS AND METHODS
Constructs of hClC-1 and TrpV2
From insertion of full-length hClC-1 cDNA into vector pCerulean-N3 or pEYFP-N3, we were able to generate proteins labelled at their C-terminus (Figure 1) with either donor fluorescent protein Cerulean (hClC-1–Cerulean) or fluorescence acceptor eYFP (hClC-1–eYFP) respectively, including a 2-residue linker (glycine-serine) between the channel protein and fluorescently labelled sequences. For our truncation and split-channel studies, fragments of hClC-1 cDNA, were amplified by PCR using the plasmid pCI-neo-hClC-1 as the template . As shown in Table 1, we termed our hClC-1 N-terminal region (N-region) peptide fragments N887 (for amino acids 1–887; i.e. Met1 to Phe887) and N451 (for amino acids 1–451; i.e. Met1 to His451) and our C-terminal region (C-region) peptide fragment 380C (for amino acid 380–988; i.e. Lys380 to Leu988). Truncation and mutation sites are depicted in relation to the secondary structure of hClC-1 in Figure 1. Fluorescent tags were attached as for full-length hClC-1. Mutations were introduced into our hClC-1 constructs using the QuikChange® site-directed mutagenesis kit (Stratagene) according to the manufacturer's instructions. A TRP (transient receptor potential) channel, TrpV2, was also labelled with eYFP at its C-terminus (TrpV2–eYFP) . All of the mutants were subsequently verified by sequencing.
Cell culture and transfection
HEK-293T cells [HEK (human embryonic kidney)-293 cells expressing the large T-antigen of SV40 (simian virus 40); A.T.C.C.] were maintained as described previously . Plasmid DNA was transfected into HEK-293T cells using Lipofectamine™ 2000 (Invitrogen) 24 h before electrophysiological measurement and 48 h before FRET recording. Just before imaging, the culture medium was substituted with a control bath solution containing 130 mM NaCl, 5 mM MgCl2, 5 mM Hepes and 1 mM EGTA (pH 7.4). When NaI was used, 65 mM NaCl was replaced by the same concentration of NaI. For the low [Cl−]e condition, 130 mM NaCl was replaced by 4 mM NaCl and 126 mM sodium glutamate. All of the Zn2+ solutions were freshly made just before the imaging experiments.
Patch-clamp experiments were conducted at room temperature (22–24 °C) in the whole-cell configuration using an Axopatch 200A patch-clamp amplifier (Axon Instruments) and associated standard equipment. Cells were continuously superfused with a bath solution containing 140 mM NaCl, 4 mM CsCl, 2 mM CaCl2, 2 mM MgCl2 and 10 mM Hepes (pH 7.4). Patch pipettes were pulled from borosilicate glass and typically had a resistance of 2–3 MΩ when filled with the standard pipette solution containing 40 mM CsCl, 85 mM caesium glutamate, 10 mM sodium EGTA and 10 mM Hepes (pH 7.2). Series resistance did not exceed 3 MΩ and was 75–85% compensated. Currents obtained were filtered at 3 kHz and recordings were made from at least four cells in each case. Voltage protocols and software used in the data analysis have been described previously .
Specifically, apparent overall open probability (Po) was obtained from normalized peak tail currents  using test steps to −100 mV for 40 ms after 160 ms conditioning pulses at increasingly hyperpolarizing (−20 mV) steps in the range from +100 mV to −140 mV, from a holding potential of −30 mV. Open probability for common gating (Poc) was obtained by using a similar protocol except that a 400 μs activation pulse to +180 mV was inserted before stepping to −100 mV. This pulse fully activates the fast gates of the channel, and then tail currents at −100 mV reflect only common gating . Since overall apparent open probability equals the product of its fast and common components, open probability of the fast gates (Pof) was then calculated for a given test voltage by dividing the relevant Po by its corresponding Poc [4,23]. Apparent Po curves were obtained by fitting the data points obtained in this way for Po, Poc and Pof with Boltzmann distributions of the form (eqn 1): (1) where Pmin is an offset or minimum Po at very negative potentials, V is the membrane potential, V1/2 is the potential at which Po=(1+Pmin)/2 (half maximal activation potential) and k is the slope factor.
Spectra FRET and FRET efficiency quantification
The spectra FRET system was set up in conjunction with an inverted fluorescence microscope (Olympus IX-81) equipped with a 100 W mercury lamp and a 60× oil-immersion objective [NA (numerical aperture) 1.42]. For spectroscopic imaging, a spectrograph (Acton SpectraPro 2150i) was used in conjunction with a Hamamatsu HQ CCD (charge-coupled device) camera. Fluorescence signals from the region of the plasma membrane were selected for analysis from spectral images. Owing to the limitation of optical resolution, cytosolic fluorescence adjacent to the plasma membrane would be included , but this type of contamination was minimized by choosing cells with strong surface fluorescence. Two filter cubes (Chroma) were used to excite Cerulean (excitation filter D436/20, dichroic mirror 455dclp) and eYFP (excitation filter HQ500/20, dichroic mirror Q515lp) respectively. No emission filter was used in these cubes. Under these experimental conditions, autofluorescence from untransfected cells was negligible.
To eliminate the direct effects of extracellularly applied NaI on fluorescence proteins, exposure of each batch of cells in NaI was not longer than 10 min. By contrast, hClC-1 channels are locked in a closed state by Zn2+ with a half time of approximately 6 min under experimental conditions, similar to those employed in the present study . This action of Zn2+ is irreversible which, combined with taking measurements out to between 120 min and 180 min in two separate experiments, ensured that most channels would be in the closed state for the FRET analysis.
Fluorescence imaging and analysis were carried out using MetaMorph software (Universal Imaging) . Emission spectra were constructed as illustrated in Figure 2. Spectra were corrected for background light, which was estimated from the blank region of the same image. To eliminate the bleed-through and cross-talk contaminations between Cerulean and eYFP due to donor emission and direct excitation of the acceptor , standard emission spectra were also collected from cells expressing only Cerulean-tagged protein or cells expressing only eYFP-tagged protein. Following scaling and subtraction, the FRET efficiency was calculated from the increase in eYFP emission due to energy transfer .
FRET efficiency was estimated by fitting a FRET model (see below) using a minimal sum-of-squares criterion, and also statistically analysed, by GraphPad Prism 5 software (GraphPad Software) using non-linear regression fitting of the data points. All the estimated FRET efficiencies were expressed as means±S.E.M. and most statistical analyses were done using the extra-sum-of-squares F test. Since the data sets in Figure 8 could not be fitted by our model, they were compared by an unpaired Student's t test.
Modelling of FRET efficiency followed our previously published procedures  which are based on the assumption that Cerulean-containing subunits and eYFP-containing subunits assemble randomly. In this case, the distribution of channels containing two Ceruleans, one Cerulean and one eYFP, and two eYFPs can be described by a binomial function which is in turn determined by the expression ratio, r, between the Cerulean-containing subunit and the eYFP-containing subunit. Measured FRET efficiency, Eapp, should then be related to the true FRET efficiency, E, of the Cerulean–eYFP pair in hClC-1 by the following equation (eqn 2): (2) Although the expression ratio r is unknown for individual cells, it is related to the measurable fluorescence intensities of Cerulean (FC) and eYFP (FY) by the equation (eqn 3): (3) in which ϵC/ϵY is the ratio of molar absorption coefficients of Cerulean and eYFP respectively, which were derived from their excitation spectra, and SC/SY is a constant related to the recording system and the fluorophores that can be determined experimentally in one of two ways: (i) it can be measured from a FRET pair with a known FRET efficiency, or (ii) it can be estimated from a global fit of multiple data sets collected with the same recording system and the same fluorophores. In the present study, the following values were used: ϵC/ϵY=12.05 and SC/SY=0.53. Combination of eqns (2) and (3) eliminates the unknown factor r allowing the true FRET efficiency to be determined by fitting of the relationship between Eapp and FC/FY.
Anisotropy was measured from channel-attached eYFP as described previously . Briefly, two linear excitation polarizers were placed in a filter wheel in the excitation light path in a parallel (I∥) and a horizontal (I⊥) position; one polarizer was placed adjacent to the emission filter in a parallel position. Steady-state anisotropy, A, was calculated using the equation (eqn 4): (4) Fluorescein maleimide dissolved in glycerol, whose anisotropy value was assumed to be 0.38, was used to determine the intrinsic polarization properties of the recording system, as described previously .
Validation of function and FRET in C-terminus-tagged hClC-1 channels
We first tested the function of fluorescently tagged channels. Deactivation currents from whole-cell patch-clamp recordings in HEK-293T cells expressing hClC-1–eYFP (Figure 3B) were not substantially different from those in cells expressing WT (wild-type) hClC-1 (Figure 3A). Apparent open probability curves for overall gating (Po) and for common gating (Poc) did, however, differ to a small, but significant, extent from WT (Figures 3C and 3D), V1/2 values for hClC-1–eYFP being shifted by 10 to 20 mV in the hyperpolarizing direction (Table 2). Nevertheless, and importantly for the purposes of the present study, although V1/2 of Poc was −63 mV for WT and −73 mV for hClC-1–eYFP, at voltages between −50 and +50 mV, the Poc data points almost completely overlapped (Figure 3D). Since our HEK-293T cells typically have a resting membrane potential of −30 to −40 mV, this gave us confidence that, for our FRET experiments, common gate activity in our tagged constructs would closely parallel that of WT hClC-1.
We next employed spectra FRET in cells expressing fluorescently tagged ClC-0 as our positive control for FRET. Apparent FRET efficiency was calculated from individual cells co-expressing ClC-0–Cerulean and ClC-0–eYFP, and plotted as a function of the fluorescence intensity ratio between Cerulean and eYFP (FC/FY). Data points from the 74 cells analysed were then fitted (solid curve) using our model for dimeric proteins which yielded an estimated true FRET efficiency of 23.7±0.5% (indicated by the upper dotted line in Figure 4A). As a negative control, we co-expressed hClC-1–Cerulean with TrpV2–eYFP. In this case, as shown in Figure 4(B), no correlation could be seen between FC/FY and apparent FRET.
When we applied the same approach to WT hClC-1 channels formed by co-expressing WT hClC-1–Cerulean and WT hClC-1–eYFP, a clear correlation between FC/FY and the apparent FRET efficiency was again observed (Figure 4C, n=187). Model fitting now yielded a true FRET efficiency estimate of 16.9±0.2%. In control solution with [Cl−]e of 140 mM, this probably represents an equilibrium between the closed and open states of the common gate . We therefore measured FRET under conditions which were designed to hold the common gate either open or closed. As shown in Figure 4(D), when the common gate was closed by [Zn2+]e (5 mM) , the FRET efficiency was significantly reduced to 12.9±0.3% (n=127) (P<0.0001 compared with control; Figure 4C). Similarly, inhibition of common gating by a low [Cl−]e (14 mM) , caused the FRET efficiency to decrease to 13.4±0.2% (n=102) (P<0.0001 compared with control; Figure 4E). By contrast, when the common gate was opened from control levels towards its maximum value (increased Poc) by replacing [NaCl]e (65 mM) with the equivalent amount of NaI (Supplementary Figure S1 at http://www.BiochemJ.org/bj/436/bj4360415add.htm), FRET efficiency was increased to 21.7±0.4% (n=56) (P<0.0001 compared with control; Figure 4F). Attempts to determine a FRET time course associated with channel closure in single cells, as had been obtained for ClC-0 , were unsuccessful, probably because the decrease in FRET efficiency between the equilibrium condition and the closed state in hClC-1 is much smaller than in ClC-0. Although there was a tendency for FRET to decrease along the anticipated time course in individual cells, it did not reach significance.
FRET in truncated hClC-1, N887, resembles that in ClC-0
In our experiments, under normal conditions, FRET efficiency of hClC-1 (16.9±0.2%) is lower than that of ClC-0 (23.7±0.5%). Could this be due to the considerably longer post-CBS2 sequence in the cytoplasmic domain of hClC-1 which causes its two C-termini to be further apart? A previous study of rat ClC-1 (rClC-1) has shown that the final 100 amino acids (fs895X, equivalent to N889 in hClC-1), but not the final 125 amino acids (L869X, equivalent to N863 in hClC-1), could be removed with only insubstantial changes to the channel function . We assessed FRET in a similarly shortened channel, N887, which, based on the observations of Hryciw et al. , as well as those of others, e.g. , was predicted to have characteristics like WT hClC-1. Also, its length beyond CBS2 is more like that of ClC-0 (WT hClC-1=117 amino acids, N887=16 amino acids and ClC-0=36 amino acids). Although truncation of approximately 80 amino acids in hClC-1 would have resulted in a closer approximation to ClC-0, these channels had been found to have very low conductance when expressed in Xenopus oocytes  and were, therefore, not chosen for the present study. Electrophysiological analysis of N887–Cerulean showed that it retained all of the basic channel characteristics of WT hClC-1 (Figure 5 and Table 2).
In our control solution, FRET efficiency for N887 was estimated to be 20.9±0.3% (Figure 6A, n=110), and, indeed, closer to that of ClC-0. Similar to WT hClC-1, when the common gate of N887 was stabilized in a closed state by [Zn2+]e (5 mM) or by low [Cl−]e (14 mM), FRET efficiency decreased to 16.2±0.3% (Figure 6B, n=53) and to 16.1±0.3% (Figure 6C, n=61) respectively (P<0.0001 compared with the control condition for both). By contrast, and again similar to WT, FRET efficiency for N887 increased to 25.5±0.4% (n=83) (P<0.0001 compared with control) when the common gate was activated by NaI (Figure 6D). These observations suggest that conformational rearrangements in the cytoplasmic domains of hClC-1, similar to those of ClC-0, occur during common gating.
Anisotropy measurements from hClC-1-attached fluorescence protein
A potential cause of changes in the FRET value is that the fluorescent proteins, covalently linked to channel subunits, might be fixed in specific orientations when the common gate was in its closed and open states. Although in this scenario FRET changes would still indicate conformational changes in the cytoplasmic domains of hClC-1, an increase or decrease in the FRET value would not be necessarily correlated with changes in proximity between the C-terminus fluorophores. We thus carried out anisotropy measurements to assess the mobility of channel-attached fluorescent proteins. We found that eYFP in several constructs used in the present study exhibited very similar anisotropy values between completely fixed (A=1) and highly mobile (A=0) (Table 3), suggesting that channel-attached fluorescent proteins retained significant mobility. Furthermore, under conditions favouring common gate closing or opening, eYFP in WT hClC-1 also exhibited very similar anisotropy (Table 3). These observations are consistent with the idea that changes in FRET reflected movement of hClC-1 C-termini that altered their proximity.
FRET study of mutants indicates a strong link between fast gating and common gating in hClC-1
Based on experiments described above and those from ClC-0, FRET changes appear to be able to track conformational rearrangements associated with common gating. Since FRET efficiency is altered by an effect of Zn2+, low [Cl−]e or [NaI]e, on common gating in WT hClC-1, we proposed that site-directed mutations, which do not show common gating in electrophysiological experiments, should also change FRET efficiency. Among such mutations, C277S (equivalent to C212S in ClC-0) is known to lock the common gate in an open and Zn2+-independent state, while leaving fast gating intact . As we expected, C277S channels showed a FRET efficiency (22.8±0.5%, n=46; Figure 7A) notably higher than that of WT hClC-1 (16.9±0.2%, n=187; Figure 4C) (P<0.0001) under normal conditions and similar to that measured under conditions promoting common gate opening (the action of NaI, above). In addition, FRET was unchanged by Zn2+ or low [Cl−]e (Figures 7B and 7C) (P>0.05 compared with control). Thus, like C212S in ClC-0, C277S appeared to hold the C-termini in a conformation normally associated with the common-gate open state.
In another point mutant, E232Q (equivalent to E166Q in ClC-0), the glutamate side-chain carboxyl group which forms the putative fast gate  has been deleted, but both fast and common gating are eliminated [27,28]. As distinct from C277S, FRET efficiency measured in E232Q (18.0±0.3%, n=87; Figure 7D) was not substantially different from that of WT hClC-1. In addition, when E232Q was introduced into the truncated channel N887 background, it showed a FRET efficiency (20.0±0.4%, n=97; Figure 7E) that was also quite similar to the WT N887 value (20.9±0.3%, n=110; Figure 6A) (P>0.05). If the conformational changes associated with the common gate actually occur in E232Q, but these are prevented from coupling to the fast gates (because the latter have been deleted), then this could be the reason why functional common gating is eliminated. In this case, insertion of C277S should modify E232Q in the same way that it modifies WT hClC-1. As shown in Figure 7(F), FRET efficiency from the double-mutant E232Q/C277S was estimated to be 22.5±1.0% (n=36), which is indeed close to that of the single C277S mutant (22.8±0.5%; Figure 7A) (P>0.05), but significantly different from that of the single mutant E232Q (18.0±0.3%; Figure 7D) (P<0.0001).
FRET study of split hClC-1 channel N451–Cerulean+380C–eYFP places constraints on the position of the hClC-1 C-terminus perpendicular to the plane of the membrane
Our previous studies  have shown that the whole cytoplasmic domain of hClC-1 (598C, containing both CBS1 and CBS2) is able to interact with, and restore function to, a non-functional membrane-resident construct (N720Δ, containing no CBS domain). In the present study, we constructed another split hClC-1 channel, this time divided within the membrane-resident region (Figure 8A). Its first (N-terminal) component was tagged with Cerulean at its extracellularly located C-terminus (N451–Cerulean). Its second component, extending to the normal cytoplasmic C-terminus of the protein, was tagged at its end with eYFP (380C–eYFP). Co-expression of these fragments allowed typical hClC-1 Cl− currents to be elicited (Figures 8B–8E and Table 2) demonstrating functional complementation of the two otherwise non-functional components. Data from our FRET studies on this split channel could not be fitted by the same equation used for our other experiments (Figure 8F) because each functional complex is expected to contain two Ceruleans and two eYFPs. The average FRET value from 59 cells co-expressing N451–Cerulean and 380C–eYFP, at 3.82±0.17%, was significantly higher than the average value (0.80±0.18%) from 64 cells co-expressing unrelated proteins, hClC-1–Cerulean and TrpV2–eYFP (negative control) (P<0.0001, Student's t test), indicating the existence of positive FRET signals between extracellularly located Cerulean and intracellularly located eYFP. Owing to the existence of two FRET donors and two acceptors per complex, there will be multiple FRET coupling pathways. Since measured FRET efficiency reflects the contribution of all fluorophores and pathways, its value should not be directly compared with those from a single FRET pair between subunits. Nonetheless, the positive FRET signal confirmed that the fluorescent proteins were within FRET distance across the cell membrane and provided an upper limit for the vertical separation between those fluorophore pairs located nearest to each other.
Underscoring a critical role for the cytoplasmic domains of eukaryotic CLC proteins, internal deletions, truncations or mutations at various sites in their extensive cytoplasmic domains can result in poor expression, non-functional channels or dysfunction where it is common gating that is specifically affected [7,9–13]. Crystal structures for the cytoplasmic domains of ClC-0, ClC-5 and ClC-Ka, as well as one for almost the entire CmCLC protein (both its cytoplasmic and membrane-resident domains) have been reported and compared [3,29–31]. Although no equivalent high-resolution structural information is yet available for hClC-1, the basic arrangement and location of its cytoplasmic domain probably closely resembles that of these other family members. Interestingly, some crystallographically unresolved sequences within the cytoplasmic domains of muscle-type CLCs, the CBS1–CBS2 linker of ClC-0  and the post-CBS2 sequences of both hClC-1  and ClC-0 [26,32], have been predicted to contain regions of organization as a result of NMR studies. For example, when the entire soluble cytoplasmic domain of ClC-0 was analysed , there was evidence for the existence of an α-helix in the CBS1–CBS2 linker. A helix in this linker had previously been seen in the crystal structure of the cytoplasmic domain of ClC-0 . In addition, the NMR analysis located two short β-strands between 20 and 30 residues downstream from CBS2. It would, however, be imprudent to anticipate the presence of analogous β-strands in hClC-1 because there is no conservation of the sequences involved. Nevertheless, these results confirm that the cytoplasmic domain of ClC-0 contains extended sequences of considerable structural flexibility , at least when that domain is not closely juxtaposed to the membrane-resident component of the complete channel protein. In the other study of the apparently disordered region just beyond CBS2 , NMR and corroborating functional evidence was used to predict the presence of a short poly-proline helix (helix S in Figure 1) in both ClC-0 and hClC-1. However, only synthetic peptides that mimicked a proline-rich stretch some ten residues downstream from CBS2 were analysed by NMR and, in the absence of flanking sequence and adjacent structures, these peptides might not assume their normal in situ organization. Since the two NMR studies do not concur with respect to the putative organized regions beyond CBS2 and they are not confirmed from any of the crystal structures of CLC cytoplasmic domains, their existence remains unproven.
Because of significant amino acid sequence similarity, common structural features would be expected in the cytoplasmic domains of hClC-1 and ClC-0 up to and including the immediate post-CBS2 region (the putative helix S). By contrast, the linear sequence in hClC-1 extends to its C-terminus some 80 residues beyond that of ClC-0. This difference might well be responsible for the lower, but comparable, FRET efficiency (~17% in hClC-1 compared with ~24% in ClC-0) that we found in our present study. Although these measured FRET efficiency differences might seem small, it is important to note that FRET efficiency is inversely dependent on the sixth power of distance. From the Förster equation E=1/[1+(R/R0)6], and assuming R0=4.9 nm, our FRET efficiency values for hClC-1 and ClC-0 translate into C-terminal fluorophore spatial separations of 6.4 nm and 6.0 nm respectively. When the post-CBS2 sequence of the cytoplasmic domain of hClC-1 was truncated to a length more like that of ClC-0, FRET efficiency indeed increased (from ~17% to ~21%, corresponding to a decrease in separation to 6.1 nm). These results suggest that, under normal control conditions, the initial post-CBS2 regions may assume defined positions similar to each other in these two channels. This view is supported by X-ray crystallography of the cytoplasmic domains of ClC-5, ClC-Ka and CmCLC [3,30,31]. In CmCLC, CBS2 and its C-terminus lie closely adjacent to the intracellular surface of the membrane-resident domain. This might explain why little structure was apparent in the NMR study of the post-CBS2 region of the soluble cytoplasmic domain of ClC-0 , an intimate relationship with the membrane-resident domain probably being required to establish a definite conformation. It might also explain the small Po differences between WT hClC-1 and our fluorescently tagged constructs (Figures 3, 5 and 8, and Table 2), our fluorescent protein tags perhaps slightly disturbing the arrangement. Indeed, there is good evidence that an ordered relationship is both structurally and functionally essential. An otherwise non-functional, severely truncated version of hClC-1 (N720) was able to be functionally reconstituted by co-expressing it with a 26 amino acid peptide spanning just the final residues of the α-2 helix of CBS2 and the beginning of the post-CBS2 region from Leu863 to Arg888 . Mutations and truncations in exactly this region of ClC-0 and hClC-1 , as well as deletion of the last 36 amino acids (all of those after CBS2) in ClC-0 , dramatically affect common gating properties.
Conformational changes throughout helix R are associated with the ion-transport process in the prokaryotic CLC, ClC-ec1 . Juxtaposition of helix R, the helix R to CBS1 linker, CBS2, the immediate post-CBS2 sequence and helix D, as has been found in CmCLC , supports functional involvement of this complex in ion permeation. A connection from the cytoplasmic domain through helix R to the membrane-resident region in eukaryotic CLCs might, therefore, appear to have relevance to common gating. At least in hClC-1, however, direct connectivity through helix R does not appear to be necessary, or likely, because we have shown in previous work that approximately WT function is maintained in a split channel reconstitution where there is no connection from the cytoplasmic domain via the R helix . We, therefore, investigated potential alternative interaction routes between the cytoplasmic domain of hClC-1 and the cytoplasmic face of its membrane-resident domain . The consequences of alanine exchange of entire cytoplasmic helix–helix loops or parts of these loops in split channel pull-down experiments were correlated with the effects of similar alanine exchange on the electrophysiological function of the channel. Alanine exchange of some loops profoundly affected function. It was apparent, however, that no single cytoplasmic helix–helix loop could be responsible for the observed strong binding between the cytoplasmic domain and membrane-resident domain, that more than one loop at a time might be involved, each being individually dispensable with respect to binding, or that the cytoplasmic domain was binding to some other site, or sites, accessible from the cytoplasm. This agrees well with the newly discovered, extensive and highly complementary interaction interface in CmCLC between the cytoplasmic surface of its membrane-resident and CBS components .
To extend our understanding of the structural relationship between the cytoplasmic domain and the membrane-resident region of hClC-1, especially with respect to the position of the C-termini perpendicular to the plane of the membrane, we split the channel into two fragments, N451+380C, and labelled their C-termini extracellularly with Cerulean in the first fragment and intracellularly with eYFP in the second. Splitting hClC-1into these particular components was chosen because Schmidt-Rose and Jentsch  had previously found that a very similar complementary pair, N451+369C, could be functionally reconstituted. Furthermore, their results from the co-expression of several different complementary pairs suggested to us that preservation of the glycosylation sites in the helix L–helix M linkers of both fragments might enhance Cl− current amplitude. But their N451+369C split channel duplicated part of helix J (from His369 to Lys379) along with helices K and L, and we decided to avoid the complication of a partial helix J overlap by beginning our second component, instead, at the helix J–helix K linker (from His380). Since Helix K is at the apex of the triangular CLC protein monomer and opposite its base [1,2], it is at the greatest possible distance from the functionally important dimer interface and relatively distant from the channel pore and fast gate. By contrast, although it might have seemed more logical to split the membrane-resident domain approximately in half within the helix I–helix J linker, this loop is located at the dimer interface and a split there, especially with an extracellular tag attached to the first fragment, might well have interfered with dimerization, prevented reconstitution and destroyed function. Taking both the knowledge of channel structure [1,2] and the prior functional reconstitution experiments  into account, we, therefore, considered that helix K and the subsequent short helix L might be able to be duplicated without much perturbation of either the monomeric or dimeric configuration of the channel and without eliminating function or substantially modifying gating.
When co-expressed, our N451+380C fragments, like those of Schmidt-Rose and Jentsch , recombined to form functional channels with typical hClC-1 currents, and we found evidence of positive FRET between their respective fluorophore tags. Although each of the split-channel components might separately be trafficked into the plasma membrane, only a single form of functional reconstitution is possible between them, where Cerulean is located on the extracellular side of the membrane (at the C-terminus of N451) and eYFP on the intracellular side (at the C-terminus of 380C). Energy transfer then has to occur vertically through the cell membrane (which is transparent to electromagnetic coupling). Positive FRET (albeit only ~4%) obtained from co-expression of these constructs suggests, according to the Förster equation, that the closest donor and acceptor fluorophores probably have a transmembrane separation of not more than ~8 nm perpendicular to the membrane. If the membrane-resident region of hClC-1 is structurally similar to that of CmCLC, it should have a transmembane thickness of ~3.5 nm  and this would constitute a minimum fluorophore separation. Further spatial separation due to effective fluorophore radius, and at least some lateral displacement of the closest extracellular donor Cerulean fluorophore with respect to its intracellular acceptor eYFP, could make a substantial addition to this minimum. It is, therefore, likely that the C-terminus of hClC-1 is located at quite a shallow depth within the cytoplasm and close to the cytoplasmic surface of the membrane-resident region. According to its crystal structure, the C-terminus of CmCLC is situated in just this position .
Since the lateral distance between the two C-termini in hClC-1 under normal conditions is approximately 6.4 nm, the structural space able to be occupied by each C-terminus is quite constrained. Although the putative poly-proline helix in the immediate post-CBS2 sequence  might act as a site for interaction with other (regulatory) cytoplasmic proteins , it could, instead, bind on to the cytoplasmic loop region of the membrane-resident domain. An interaction of this kind would help to keep the C-terminus close to the membrane-resident region and explain how A885P can be a dominant myotonic mutation .
In previous work  on ClC-0, FRET measurements showed that common-gate opening and closing was associated with large displacements of the C-termini from the positions they assume under resting conditions. The present study demonstrates that similar conformational changes occur in hClC-1. In both ClC-0 and hClC-1 channels, shifting the equilibrium of the common gate towards its open state (using high [Cl−]e and I− substitution respectively) increases FRET efficiency, which corresponds to closer proximity of their two C-termini. On the other hand, closure of the common gate (with Zn2+ or low [Cl−]e) reduces FRET efficiency which corresponds to increased separation. Despite these similarities, the magnitude of FRET changes in the two channels is quite different. Common-gate closure from its maximum open position in ClC-0 is associated with a large FRET decrease from 23% to 5% , whereas, in hClC-1, the decrease is modest, from 22% to 13% (the present study). Displacement of the C-termini associated with closure must, therefore, be considerably greater in ClC-0 than in hClC-1. According to the Förster equation, an increase in separation by more than 2 nm would be required to reduce FRET from 23% (slow gate open state) to less than 5% (closed state) in ClC-0 , whereas to reduce FRET from 22% to 13% in hClC-1, the increase in separation could not be more than approximately 0.7 nm (approximately one-third of that in ClC-0).
Although qualitatively similar conformational changes in the two channels are consistent with their predicted structural similarity, a close correspondence might not necessarily have been anticipated since the common gate of ClC-0 has an opposite voltage-dependence to that of hClC-1, the ClC-0 common gate closing with depolarization. Conversely, some differences might have been expected because of the significant disparity in temperature-dependence of their gating relaxations, the Q10 values for the common gating of ClC-0 and hClC-1 being approximately 40 and 4 respectively [7,17]. Conformational rearrangements during common gating in ClC-0, therefore, are likely to be considerably greater than in hClC-1. Our FRET measurements are consistent with this view. Indeed, time constants for common gating are of the order of seconds to minutes for ClC-0 , and just tens of milliseconds for hClC-1 , consistent with a smaller movement in hClC-1.
Combining FRET with mutations that have specific effects on common gating allowed us to examine the relationship between gating and C-terminus movement more closely. In the C277S mutant, fast gating is maintained, although its common gate is effectively locked open . In the present study, an elevated FRET efficiency above that of WT hClC-1 (Figure 4C) suggests that the C-termini are indeed in close proximity, as is the case for the common gate open state of WT channels. By contrast, in the E232Q mutant, common-gate closure is eliminated along with fast gating, as can be discerned from raw current records  and from analysis of the apparent open probabilities of the two gating processes . Notwithstanding the total absence of common gating, FRET is unchanged by E232Q in either the full-length hClC-1 or N887 truncated channel (Figures 7D and 7E), suggesting that its cytoplasmic domain might, nevertheless, be capable of undergoing the type of conformational change found during common gating of WT channels. This possibility is supported by the observation that, when these two mutations are combined in a double-mutant construct (E232Q/C277S), FRET efficiency is enhanced (Figure 7F; see also the Supplementary Online Data and Supplementary Figure S2 at http://www.BiochemJ.org/bj/436/bj4360415add.htm). In other words, C277S is capable of locking the common gate in its open conformational state even in the E232Q background that otherwise exhibits the common-gate conformation of WT hClC-1.
Although fast and common gating have been described as ‘not necessarily independent’ , independence has generally been assumed. Our present conclusions depart significantly from this common presumption, which is: (i) that fast-gate transitions occur independently of each other within each pore and independently of the common gate, (ii) that common-gate transitions occur independently of the fast gates and (iii) that the common gate is a separate structural entity that physically obstructs the two pores, simultaneously. Based on the findings of the present study, where deletion of the fast gates also functionally abolishes common gating but permits the conformational rearrangements underlying common gating, we propose instead that fast and common gating must be closely coupled in the following way: (i) that fast-gate transitions occur independently of each other within each pore, depending on the common gate being in an open (permissive) state, (ii) that common-gate transitions occur independently of the fast gates and (iii) that the common gate is not a structural entity that causes concurrent obstruction of both pores, but rather a conformational change that simultaneously permits or prevents the operation of both fast gates, alternately allowing their freely independent action or locking both closed (Figures 9A and 9B). So common gating is manifested exclusively through synchronized switching (coupling) that affects both fast gates at once, the associated conformational change being reflected at the C-termini. There would be every opportunity for dynamic structural communication between the fast gate latching sites in the channel pores and the C-termini through the intricate bond between the membrane-resident and cytoplasmic domains. We also propose that the rearrangements normally associated with common gating can still occur in the E232Q mutant but, because the fast gates are absent, there can no longer be any coupling of the gating to the conformation (Figures 9C and 9D). Common gating is eliminated by C277S because this mutation prevents the common gate from switching out of its open, permissive configuration where the channel's two fast gates retain their individual operation (Figures 9E and 9F).
It is particularly pertinent that the FRET changes we have found to be associated with common gating in hClC-1 reinforce the validity of those obtained from ClC-0 under similar conditions. They are, in all cases, congruent with the expectations arising from our electrophysiological observations and these, in turn, are consistent with experimental data from many other laboratories. Our new FRET and electrophysiological results from site-directed hClC-1 mutants can, therefore, be expected to provide a sufficient basis for the novel interpretation of common gating that we derive from them. Finally, many of our conclusions regarding cytoplasmic domain conformation and its involvement in common gating are now supported by X-ray crystallographic structural evidence from several CLC proteins [30,31].
This present study was conceived by Allan Bretag and Grigori Rychkov, and they also supervised the molecular biology and electrophysiology. Most of the individual experiments were designed and performed by Linlin Ma, who was assisted by Ekaterina Bykova in some of the FRET studies. FRET experiments were supervised by Jie Zheng, who also undertook the anisotropy measurements. All authors were involved in interpretation of the results. Linlin Ma wrote the manuscript with input from Allan Bretag, Grigori Rychkov and Jie Zheng.
This work was supported by the Muscular Dystrophy Association of South Australia (to A. H. B.); the Research Committee of the University of South Australia (to A. H. B.); the National Institutes of Health [grant number REY016754A (to J. Z.)]; a University of California Davis Health System Research Award (to J. Z.); a National Health and Medical Research Council Senior Research Fellowship [number 453580 (to G. Y. R.)].
We thank Professor Thomas J. Jentsch, then of the Center for Molecular Neurobiology at the University of Hamburg, Germany, for supplying our original WT hClC-1 cDNA clone. Laboratory members, staff and colleagues at the University of South Australia, The University of Adelaide and the University of California at Davis are gratefully acknowledged for help and helpful discussions.
Abbreviations: CCD, charge-coupled device; ClC, Cl− channel; eYFP, enhanced yellow fluorescent protein; FRET, fluorescence resonance energy transfer; hClC, human skeletal muscle ClC; HEK-293T, cells, HEK (human embryonic kidney)-293 cells expressing the large T-antigen of SV40 (simian virus 40); TRP, transient receptor potential; WT, wild-type
- © The Authors Journal compilation © 2011 Biochemical Society