The Role of the Carboxyl Terminus in ClC Chloride Channel Function*

The human muscle chloride channel ClC-1 has a 398-amino acid carboxyl-terminal domain that resides in the cytoplasm and contains two CBS (cystathionine-β-synthase) domains. To examine the role of this region, we studied various carboxyl-terminal truncations by heterologous expression in mammalian cells, whole-cell patch clamp recording, and confocal imaging. Channel constructs lacking parts of the distal CBS domain, CBS2, did not produce functional channels, whereas deletion of CBS1 was tolerated. ClC channels are dimeric proteins with two ion conduction pathways (protopores). In heterodimeric channels consisting of one wild type subunit and one subunit in which the carboxyl terminus was completely deleted, only the wild type protopore was functional, indicating that the carboxyl terminus supports the function of the protopore. All carboxyl-terminal-truncated mutant channels fused to yellow fluorescent protein were translated and the majority inserted into the plasma membrane as revealed by confocal microscopy. Fusion proteins of cyan fluorescent protein linked to various fragments of the carboxyl terminus formed soluble proteins that could be redistributed to the surface membrane through binding to certain truncated channel subunits. Stable binding only occurs between carboxyl-terminal fragments of a single subunit, not between carboxyl termini of different subunits and not between carboxyl-terminal and transmembrane domains. However, an interaction with transmembrane domains can modify the binding properties of particular carboxyl-terminal proteins. Our results demonstrate that the carboxyl terminus of ClC-1 is not necessary for intracellular trafficking but is critical for channel function. Carboxyl termini fold independently and modify individual protopores of the double-barreled channel.

ClC channels are found in almost all prokaryotic and in eukaryotic cells. Nine isoforms (ClC-1 to ClC-7, ClC-Ka and ClC-Kb) were shown to be expressed in human tissues and to fulfil a variety of functional roles. ClC-1 is the major muscle chloride channel responsible for the regulation of muscle excit-ability (1,2), and ClC-2 is crucial for neuronal chloride homeostasis (3). ClC-Kb is involved in transepithelial NaCl movement in the thick ascending limb of Henle (4), and ClC-3, -5, and -7 are necessary for the pH adjustment of several cell compartments (5)(6)(7).
Eukaryotic and prokaryotic ClC channels exhibit 18 transmembrane domains (8) followed by cytoplasmic carboxyl termini of variable sequences. The carboxyl-terminal tails of mammalian isoforms contain between 146 and 404 amino acids and exhibit two structurally defined domains, so-called CBS domains (9,10). The functional importance of the carboxyl terminus is illustrated by disease-causing mutations in ClC-1, -2, -5 and ClC-Kb (3,6,11,12). Moreover, truncations removing parts of the distal cystathionine-␤-synthase (CBS) 1 domain of ClC channels were shown to abolish functional expression in heterologous systems (6,(13)(14)(15)(16). Although co-expression of the complementary carboxyl terminus restored the function of these channels, this was not the case for truncations removing carboxyl-terminal fragments containing both CBS domains (13). Although these results demonstrated the importance of this region for the function of the channel in the surface membrane, it is still unclear by which mechanism the carboxyl terminus supports this function. It is not known whether ClC channel carboxyl termini are necessary for channel trafficking or for the proper function of the channel. Because ClC channels exhibit two ion conduction pathways (protopores) per functional channel (8,17,18), the carboxyl terminus of one subunit might modify either the function of an individual protopore or of both protopores of the double-barreled channels.
In the present work, we addressed these questions by studying the functional role of carboxyl-terminal regions of hClC-1 channels and interactions between them and transmembrane domains. We demonstrate that carboxyl-terminal truncations result in a loss of function rather than in an impaired plasma membrane insertion of functional channels. In heterodimeric channels with only one carboxyl terminus, only one protopore is functional. Moreover, we show interactions between carboxylterminal fragments stable enough to redistribute a cytoplasmic channel fragment to the membrane. Such interactions are only possible within one subunit and not between domains belonging to two separate subunits.

EXPERIMENTAL PROCEDURES
Mutagenesis and Channel Expression-Truncations and point mutations were introduced into the plasmid pRc/CMV-hClC-1 containing the full-length WT hClC-1 cDNA (19) by PCR-based strategies. To create YFP or CFP fusion proteins with hClC-1, the YPF/CFP was excised from pEYFP-N1/pECFP-N1 (Clontech) and inserted into pRc/CMV (In-* This work was supported by Deutsche Forschungsgemeinschaft Grants SFB542, TPA2 (to G. M.-N.), FOR450/1, TP1 (to P. H.), and TP3 (to C. F.) and by the Muscular Dystrophy Association (to C. F.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Expression of hClC-1 Channel Fragments-Transient transfections in tsA201 cells were performed with 0.1-10 g of plasmid DNA using a calcium phosphate precipitation method (20). Cells were typically examined 2 days after transient transfection. To identify cells with a high probability of expressing recombinant ion channels, cells were co-transfected with a plasmid encoding the CD8 antigen and incubated for 5 min before use together with polystyrene microbeads precoated with anti-CD8 antibodies (Dynabeads M-450 CD 8; Dynal, Great Neck, NY) (21). Only cells decorated with microbeads were used for electrophysiological recordings.
Data Analysis-Data were analyzed by a combination of PulseTools (HEKA Electronics, Lambrecht, Germany), pClamp (Axon Instruments, Foster City, CA), and SigmaPlot (Jandel Scientific, San Rafael, CA) programs. To determine the unitary current amplitude as well as the absolute open probabilities of mutant hClC-1 channels, non-stationary noise analysis was performed as described (23). To obtain the voltage dependence of activation, instantaneous current amplitudes were determined 200 s after a voltage step to Ϫ125 mV following 0.2-s prepulses to various voltages, normalized by their maximum value, and plotted versus the preceding potential. To determine the voltage dependence of the opening of the slow gate, a short 400-s pulse to ϩ180 mV was inserted before the test step to Ϫ125 mV to fully activate the fast gate (24). A plot of the instantaneous current amplitude at Ϫ125 mV versus the prepulse potential yields the voltage dependence of the slow gate. The relative open probability of the fast gate was calculated by dividing the voltage dependence of the relative open probability by the open probability of the slow gate. Activation curves for the fast and for the slow gate were fit with a sum of a voltage-independent minimum open probability (P min ) and a voltage-dependent term: I(V) ϭ Amp/(1 ϩ e (V Ϫ V 0.5)/kV ) ϩ P min . For the determination of mean peak current amplitudes (Fig. 5), whole-cell and excised outside-out patch currents were averaged. To account for the distinct membrane area in patches compared with a whole cell, the patch current amplitudes were multiplied by the ratios of resting conductance before excision by the corresponding value after excision.
Confocal Microscopy-Confocal imaging was carried out on a Zeiss LSM 510 confocal microscope equipped with an argon-ion laser (Zeiss, Jena, Germany) (25). The argon-ion laser was modulated by an acustooptical modulator. For excitation of CFP the 458-nm line was used. The beampath for CFP contained a 458-nm main dichroic mirror and a 480/20-nm bandpass filter for detection of emitted fluorescence. This narrow filter excludes any fluorescence emerging from YFP that is also excited by the 458-nm laser beam. The beampath for YFP detection contained excitation at 514 nm and a main dichroic mirror of 514 nm. The emitted light was monitored with a 530 -600-nm bandpass filter. CFP is not excited by the 514-nm laser. For cells co-expressing CFP and YFP the multitrack function of the LSM 510 was used. Live cell imaging was carried out with transiently transfected tsA201 cells on a glass coverslip in a perfused recording chamber 48 h after transfection.

Effects of Carboxyl-terminal Truncations and
Deletions of hClC-1-We investigated various truncation mutants of hClC-1 (Fig. 1A) heterologously expressed in tsA201 cells to study the functional role of distinct components of the carboxyl terminus. Experiments were performed with homodimeric proteins obtained through transfection of plasmids encoding hClC-1 monomeric constructs. Truncations distal the second CBS domain did not change macroscopic hClC-1 currents (Fig.  1B). In contrast, deletions of parts of CBS2, even as little as 6 amino acids (E865X), caused a complete disappearance of the ClC-1 current component (Fig. 1B). Although deletion of CBS1 did not change the macroscopic current amplitudes, removal of increasing parts of the interlinker between CBS1 and CBS2 caused a gradual decrease of the current amplitude (Fig. 1B). Substitution of CBS2 with CBS1 in ⌬(D607-Q662)CBS1 hClC-1 resulted in comparable macroscopic current amplitudes, indicating that the two CBS domains can substitute for each other (Fig. 1B).
The reduced macroscopic current amplitude in cells expressing truncated channels might be because of an impaired surface targeting, a reduced unitary current amplitude, or a reduced absolute open probability of functional channels as well as a reduction of functional channels in the surface membrane. To study the subcellular distribution of truncated ClC-1 channels, we generated YFP-tagged channel constructs and examined cells expressing various YFP-ClC-1 del constructs by confocal imaging. Cells expressing WT as well as G750X, A700X, and L590X hClC-1 channels showed a staining of the surface membrane ( Fig. 1, C, E-G). We conclude that these mutant channels do not abolish surface membrane insertion but rather alter the function of inserted channels. In cells expressing Y850X hClC-1 channels, the staining of the surface membrane was decreased, suggesting that this mutation might also interfere with surface membrane insertion. Similar to another ClC isoform, hClC-2 (3), all tested constructs also appeared in internal membrane compartments ( Fig. 1, Functional Alterations in Carboxyl-terminal-deleted Channels-We next investigated gating and ion permeation properties of mutant ClC-1 channels lacking the CBS1 domain plus various fractions of the CBS interlinker (Fig. 2). All tested deletion mutants showed the hallmarks of ClC-1 gating, i.e. depolarization-induced activation, a biexponential deactivation time course, and the presence of two kinetically distinct gating processes (fast and slow gating) (Fig. 2, A and B). Mutant channels only differed in a slightly shifted voltage dependence of the fast and of the slow gate from WT (Fig. 2, A and B). No differences could be observed between distinct truncation mutants (Fig. 2, A and B).
To determine the unitary current amplitude and the absolute open probability, we performed non-stationary noise analysis. Fig. 2 illustrates the time course of the mean current for a voltage step from a holding potential of 0 mV to a test potential of Ϫ160 mV (Fig. 2C) and the corresponding time course of the variance (Fig. 2D) measured at an outside-out patch of a tsA201 cell expressing ⌬(D607-A700) hClC-1 in standard internal and external solution. As shown in Equation 1, to obtain the unitary current amplitude (i) and the absolute open probability of the underlying channels, the variance-mean current plot in Fig. 2E was fitted to where i is the single channel current amplitude, N the number of channels, and I the mean current amplitude (26,27). From these fits, the unitary current amplitude at Ϫ160 mV (0.32 Ϯ 0.03 pA, n ϭ 6, as compared with WT, 0.28 Ϯ 0.02 pA, n ϭ 6) and the number of channels in the cells were obtained. Because hClC-1 exhibits two ion conduction pathways, the so-obtained single channel amplitudes have a value between the half-and the full-conductance current amplitude (23). The results of noise analysis nevertheless demonstrate that deleting parts of the carboxyl terminus does not cause substantial changes of the unitary current amplitude.
By dividing the instantaneous current amplitude by the product of the unitary current amplitude and the number of channels, we then calculated the absolute open probability at the holding potential (⌬(D607-A700) hClC-1, 0.90 Ϯ 0.03, n ϭ 8; WT, 0.89 Ϯ 0.02, n ϭ 6). The results taken together indicate that removal of CBS1 and major components of the CBS inter- linker do not alter ion conduction and only mildly affect gating properties of ClC-1.
Only One CBS2 Domain Is Necessary to Support the Function of Heterodimeric ClC-1 Channels-ClC channels are dimeric proteins with two functional ion conduction pathways. The importance of the carboxyl terminus for the functional expression of ClC-1 (Fig. 1B) raised the question whether each individual protopore or each channel requires the minimum set of carboxyl-terminal fragments for proper function. To address this question we studied heterodimeric channels using a concatameric construct that links two ClC-1 coding regions in a single open reading frame (20). Heterodimeric WT-L590X hClC-1 channels containing only one complete carboxyl terminus and even ⌬(D607-Q662)-L590X hClC-1 channels with only one CBS domain both form functional channels with gating and permeation properties similar to those of WT (Fig. 3, A and B), albeit with decreased macroscopic current amplitudes (Fig.  3C). Obviously, a single CBS domain is sufficient for channel function.
Covalent linkage of two different subunits does not completely dismiss the possibility of the formation of homodimeric channels assembled by two identical subunits belonging to two distinct concatamers. To test whether the expression of heterodimeric concatamers supports the formation of homodimers by such a mechanism, we inserted a naturally occurring, disease-causing point mutation, S132C (28), in one of the two subunits. S132C modifies gating properties of ClC-1 in homodimeric S132C hClC-1 and in heterodimeric WT-S132C hClC-1 channels in a different way (Fig. 3, D-F). Homodimeric S132C hClC-1 channels activate upon membrane hyperpolarization (Fig. 3D), whereas expression of WT-S132C hClC-1 concatamers results in a depolarization-activated anion cur-rent that differs from homodimeric WT channels in voltage dependence and in minimum open probability at negative potentials (Fig. 3, A and E). The absence of a hyperpolarizationactivated current amplitude indicates that homodimeric S132C hClC-1 channels are not formed under these conditions and demonstrates that expression of concatamers results in a homogenous population of heterodimeric anion channels. Expression of a concatameric construct linking S132C hClC-1 with L590X hClC-1 in tsA201 cells gives rise to the expression of an anion current whose gating properties are similar to those of heterodimeric WT-S132C hClC-1 channels (Fig. 3F). Because L590X hClC-1 homodimeric channels are non-functional, currents that deactivate upon hyperpolarizing voltage steps must be conducted by heterodimeric channels with only one carboxyl terminus (Fig. 3F).
Within the double-barreled architecture of ClC-1 channels, an individual carboxyl terminus might support the function of both protopores of an individual channel or the anion transfer through the corresponding protopore. We engineered truncated ClC-1 subunits carrying a point mutation (S537F) rendering mutant hClC-1 channels insensitive to block by 9-anthracene carboxylic acid (9-AC) (29, 30). As 9-AC binds within the ion conduction pathway, a heterodimer consisting of one 9-ACsensitive and one 9-AC-insensitive subunit provides insights into the function of individual protopores. If both protopores were functional in heterodimeric channels with only one carboxyl terminus, one would expect a similar relative current reduction by 9-AC in cells expressing WT-S537F and in cells expressing heterodimeric channels in which one of the two carboxyl termini is deleted, i.e. WT-S537F/L590X and S537F-L590X hClC-1 channels. This was not the case. Fig. 4A shows current recordings of a tsA201 cell expressing a WT-S537F/
L590X hClC-1 heterodimeric construct after application of 0.2 mM 9-AC in the external solution. In the presence of 0.2 mM 9-AC, the current amplitude decreased in a monoexponential fashion to a steady-state value of about 20% of the initial current amplitude (Fig. 4B). The relative block of WT-S537F/ L590X hClC-1 channels (blocked current fraction: 0.80 Ϯ 0.01, n ϭ 15, p Ͻ 0.01) was significantly larger, and the relative block observed for S537F-L590X heterodimers (0.29 Ϯ 0.02, n ϭ 12, p Ͻ 0.01) significantly smaller than the relative block of WT-S537F hClC-1 (0.55 Ϯ 0.03, n ϭ 8) (Fig. 4C). This demonstrates that the protopore without attached carboxyl terminus is non-functional in the WT-S537F/L590X as well as in the S537F-L590X hClC-1 heterodimer. We conclude that the hClC-1 carboxyl terminus supports only the function of the protopore to which it is attached.

Rescue of ClC-1 Function by Co-expression of Truncated Channel Constructs with Isolated Carboxyl Termini-Co-ex-
pressing isolated carboxyl-terminal proteins rescued non-functional truncation mutants of hClC-1 in mammalian cells in agreement with earlier results using Xenopus oocytes or insect cells (13)(14)(15). Channels formed by assembly of such fragments exhibit unaltered gating and conduction properties, but there were marked differences in current amplitudes between separate sets of co-expressed fragments. Fig. 5A shows mean current amplitudes for five distinct carboxyl-terminal truncations that are non-functional when expressed alone, L590X, G650X, A700X, G750X, and E800X. Each of these truncation mutants was co-transfected with expression constructs encoding isolated carboxyl termini of variable length, i.e. C term (L590-L988), C term (A700-L988), C term (G750-L988), C term (E800-L988), as well as C term (A700-L988CBS1) in which CBS2 was substituted with CBS1.
For deletions amino-terminal or within CBS1, none of the co-transfections resulted in the expression of functional channels. A700X ClC-1 and mutant ClC-1 with larger carboxyl termini were rescued by co-expressing various carboxyl-terminal proteins (Fig. 5A). Co-transfection of A700X, G750X, and E800X with C term (E800-L988) did not result in the appearance of a ClC-1 current component, which demonstrates the importance of the linker between amino acids 700 and 800 for this interaction. Moreover, in co-expression experiments CBS1 and CBS2 were not equivalent because co-expressing A700X hClC-1 with C term (A700-L988CBS1) did not result in the formation of functional channels (Fig. 5A).
The mean current amplitudes observed in co-transfected cells depended on the length of the carboxyl-terminal fragment. For all tested transmembrane constructs, expression levels were highest when co-expressed with C term (G750-L988) (Fig.  5A). Comparing expression levels for various transmembrane constructs with a given carboxyl terminus, G750X exhibited the highest expression levels of all combinations (Fig. 5A). When transfected with C term (G750-L988), no significant difference in the current amplitudes for G750X and E800X were observed (Fig. 5A), indicating that a doubling of this particular region is of no functional consequence.
Binding of Carboxyl-terminal Proteins to ClC-1 Truncation Mutants-The co-expression results indicate a specific interaction between carboxyl-terminal proteins and truncated hClC-1 channel constructs. To study interactions between hClC-1 protein fragments more directly, we used co-transfection and confocal imaging of truncated ClC-1 and carboxyl-terminal fragments, both fused to fluorescent proteins. All carboxyl-terminal fusion proteins formed soluble proteins (Fig. 5, B-D). Although fusion proteins containing only parts of the ClC-1 carboxyl terminus (C term (A700-L988)-CFP (Fig. 5C), C term (G750-L988)-CFP, C term (E800-L988)-CFP) (data not shown) were small enough to enter the nucleus, resulting in a fluorescent staining of the whole cell, only the cytoplasm was fluorescently labeled in cells expressing C term (L590-L988)-CFP (Fig. 5B). Co-transfection of carboxyl-terminal proteins together with certain truncation mutants caused a redistribution of the soluble proteins to the membrane (Table I). For example, in cells cotransfected with YFP-A700X hClC-1 and C term (A700-L988)-CFP the majority of CFP fluorescence is overlaid to the YFP fluorescence (visible as orange color in Fig. 5F), indicating a stable binding of YFP-A700X hClC-1 and C term (A700-L988)-CFP. This result is further supported by the finding that no fluorescence is observed in the nucleus of co-transfected cells (Fig. 5F). Fig. 5 and Table I give the results of additional co-transfection experiments of fluorescent carboxyl termini and truncated channels. No redistribution and no overlay could be observed for the two complementary fragments YFP-L590X hClC-1 and C term (L590-L988)-CFP (Fig. 5E), indicating that carboxyl-terminal fusion proteins most likely do not stably bind to the transmembrane domain. In cells co-expressing YFP-⌬(D607-Q662) hClC-1 (containing CBS2) and C term (A700-L988CBS1)-CFP (containing CBS1), the nucleus was still stained and YFP and CFP were not co-localized (Fig. 5G). CBS1 and CBS2 are either not direct binding partners or these domains adopt different conformations when attached to other regions of the carboxyl terminus.
Our results (Table I) cannot be readily explained by assuming two defined binding partners within the carboxyl terminus. Rather, they suggest that certain domains within the carboxyl terminus exhibit distinct conformations in separate constructs. Such a feature can be directly demonstrated by co-transfection experiments involving one complete carboxyl terminus (Fig. 6). YFP-WT hClC-1 channels bind neither the C term (A700-L988)-CFP (Fig. 6A) nor C term (A700-L988CBS1)-CFP (Fig. 6B), indicating that the binding sites that allow truncated channels to translocate these carboxyl-terminal fragments are not accessible in WT. Most likely, these sites are occupied by fragments of the same polypeptide chain, preventing the binding of additional carboxyl-terminal fragments. In contrast, the carboxylterminal fusion protein C term (L590-L988)-CFP is redistributed by YFP-A700X hClC-1 (Fig. 6C). The isolated C term (L590-L988)-CFP obviously exhibits different binding properties than a carboxyl terminus attached to a transmembrane domain.
Carboxyl-terminal Fragments Only Interact within a Single Polypeptide Chain-A dimeric ClC channel possesses two carboxyl termini, and therefore interactions between separate carboxyl-terminal fragments might occur within a single polypeptide chain or between the two subunits. To distinguish between these two possibilities, we performed co-expression experiments with fluorescent protein-tagged channel constructs (Fig.  7). We designed a concatameric construct in which one hClC-1 subunit lacking CBS1 (YFP-⌬(D607-Q662) hClC-1) was linked to E800X hClC-1 (Fig. 7A). This heterodimeric channel is present in the surface membrane and exhibits functional properties similar to homodimeric WT hClC-1 (data not shown). Co-transfection of this concatamer with C term (A700-L988)-CFP caused a redistribution of the carboxyl-terminal fusion protein. YFP and CFP fluorescences overlap, and no fluorescence is visible within the nucleus (Fig. 7A). In contrast, a heterodimer consisting of one YFP-WT hClC-1 and one L590X hClC-1 subunit does not bind C term (A700-L988)-CFP (Fig. 7B). Although both heterodimers exhibit the same set of carboxyl-terminal fragments, they differ in their binding properties. Although separate segments of the WT carboxyl terminus bind to each other, preventing an interaction of the soluble fusion protein with WT-L590X hClC-1 heterodimers, C term (A700-L988)-CFP can bind to YFP-⌬(D607-Q662)-E800X hClC-1 heterodimers, indicating an accessible binding site within the carboxyl termini. We conclude that carboxyl-terminal fragments attached to distinct subunits do not bind to each other. DISCUSSION All mammalian ClC channels exhibit a cytoplasmic carboxyl terminus with two so-called CBS domains (Fig. 1A). The functional importance of this channel region is illustrated by various disease-causing mutations (3, 6, 11, 12) and the experimen-tal finding that certain truncations abolish function of several ClC isoforms (13)(14)(15) (Fig. 1).
The loss of function of truncated ClC channels appears to occur by distinct mechanisms in separate ClC isoforms. Maduke et al. (14) demonstrated that injection of a purified protein fragment corresponding to the last 150 amino acids of ClC-0 restored channel function of a truncated ClC-0; however, this effect was prevented by incubation with brefeldin A, an agent inhibiting plasma membrane trafficking. This result suggested an inability of truncated ClC-0 to insert into the mem- brane and a role of CBS2 in intracellular trafficking. Mutations within CBS2 or deletion of this region caused a mistargeting of Gef1, a yeast ClC channel isoform that is normally present in the Golgi apparatus (16). For ClC-5, a predominantly intracellularly located ClC isoform, truncated channels missing large portions of the carboxyl terminus or even some of the transmembrane domains were reported to insert into the plasma membrane (31).
In the present study, several ClC-1 truncation mutants were inserted into the surface membrane (Fig. 1), demonstrating that the carboxyl terminus of ClC-1 is not necessary for insertion into the plasma membrane. We could show further that carboxyl-terminal deletion does not alter anion permeation or gating properties so that truncated channels are permanently closed, for example by a change of the voltage dependence or of the absolute open probability (Fig. 2). For ClC-1, increasing deletions of CBS1 and the CBS1-CBS2 interlinker do not result in changed gating properties but cause a progressive decrease of the maximum current amplitude (Figs. 1 and 2).
CBS domains are thought to interact with each other (9, 10), and therefore interaction of CBS1 and CBS2 was assumed to be necessary for normal ClC channel function (13). We here demonstrate that heterodimeric channels with only a single CBS domain are functional, refuting this suggestion (Fig. 3). ClC-1 mutants with deleted CBS1 (⌬(D607-Q662) hClC-1) do not bind a carboxyl-terminal fusion protein with CBS1 (C term (A700-L988CBS1)) (Fig. 5G). Moreover, no colocalization of A700X and G750X hClC-1 (containing CBS1) with C term (E800-L988) (containing CBS2) was observed (Table I). These results argue against the notion that CBS1 and CBS2 bind to each other.
All tested carboxyl-terminal fusion proteins are located in the cytoplasm and some, depending on their size, are even able to enter the nucleus (Fig. 5). Co-expression with certain membrane-spanning fragments caused a redistribution, indicating a stable binding of the two components to each other. However, an isolated complete hClC-1 carboxyl terminus (C term (L590-L988)) does not co-localize with a L590X hClC-1, indicating that the carboxyl terminus does not bind directly to a transmembrane domain (Fig. 5). The conformation of C term (L590-L988) depends on the presence of the transmembrane domains of hClC-1. The isolated carboxyl terminus (C term (L590-L988)) is capable of binding to A700X hClC-1 and of relocalization to the membrane (Fig. 6C). When attached to the pore-forming domains in WT channel, it is unable to bind carboxyl-terminal fragments (C term (A700-L988) and C term (A700-L988CBS1)) ( Fig. 6A). Obviously, C term (L590-L988) exhibits a distinct conformation when attached to transmembrane domains, suggesting an interaction but no stable binding of carboxyl terminus and transmembrane domains.
ClC channels exhibit a unique architecture with two subunits each forming an ion conduction pathway. So far, it was unclear whether this particular symmetry also extended to the carboxyl termini, i.e. whether the two carboxyl termini of a dimeric ClC channel interact with each other and whether they affect the function of the individual protopore or both protopores cooperatively. For the Torpedo isoform, ClC-0, it has been postulated that structures at the carboxyl terminus are involved in gating processes that cooperatively open and close both protopores (32), suggesting that some interaction between the two carboxyl termini exists. Our co-expression experiments of heterodimeric channel constructs with carboxyl-terminal proteins (Fig. 7, Table I) support the notion that each carboxyl terminus folds independently and that there are no interactions between them. The carboxyl terminus obviously does not provide a domain that would allow interaction between the two subunits and between the two protopores. The blocking experiments with one 9-AC-sensitive and one -insensitive protopore (Fig. 4) indicate that each carboxyl terminus acts only on the corresponding protopore, thus demonstrating that the role of the carboxyl terminus represents another example of the functional independence of the two channel halves.