ClC-7 is a slowly voltage-gated 2Cl/1H+-exchanger and requires Ostm1 for transport activity

Authors

  • Lilia Leisle,

    1. Leibniz-Institut für Molekulare Pharmakologie (FMP) and Max-Delbrück-Centrum für Molekulare Medizin (MDC), Berlin, Germany
    Search for more papers by this author
    • These authors contributed equally to this work
  • Carmen F Ludwig,

    1. Leibniz-Institut für Molekulare Pharmakologie (FMP) and Max-Delbrück-Centrum für Molekulare Medizin (MDC), Berlin, Germany
    Search for more papers by this author
    • These authors contributed equally to this work
  • Florian A Wagner,

    1. Leibniz-Institut für Molekulare Pharmakologie (FMP) and Max-Delbrück-Centrum für Molekulare Medizin (MDC), Berlin, Germany
    Search for more papers by this author
  • Thomas J Jentsch,

    Corresponding author
    1. Leibniz-Institut für Molekulare Pharmakologie (FMP) and Max-Delbrück-Centrum für Molekulare Medizin (MDC), Berlin, Germany
    • Corresponding author. Physiology and Pathology of Ion Transport, Leibniz-Institut für Molekulare Pharmakologie (FMP), Max-Delbrück-Centrum für Molekulare Medizin (MDC), Robert-Roessle-Str. 10, 13125 Berlin, Germany. Tel.: +49 309 406 2961; Fax: +49 309 406 2960; E-mail: Jentsch@fmp-berlin.de

    Search for more papers by this author
  • Tobias Stauber

    1. Leibniz-Institut für Molekulare Pharmakologie (FMP) and Max-Delbrück-Centrum für Molekulare Medizin (MDC), Berlin, Germany
    Search for more papers by this author

Abstract

Mutations in the ClC-7/Ostm1 ion transporter lead to osteopetrosis and lysosomal storage disease. Its lysosomal localization hitherto precluded detailed functional characterization. Using a mutated ClC-7 that reaches the plasma membrane, we now show that both the aminoterminus and transmembrane span of the Ostm1 β-subunit are required for ClC-7 Cl/H+-exchange, whereas the Ostm1 transmembrane domain suffices for its ClC-7-dependent trafficking to lysosomes. ClC-7/Ostm1 currents were strongly outwardly rectifying owing to slow gating of ion exchange, which itself displays an intrinsically almost linear voltage dependence. Reversal potentials of tail currents revealed a 2Cl/1H+-exchange stoichiometry. Several disease-causing CLCN7 mutations accelerated gating. Such mutations cluster to the second cytosolic cystathionine-β-synthase domain and potential contact sites at the transmembrane segment. Our work suggests that gating underlies the rectification of all endosomal/lysosomal CLCs and extends the concept of voltage gating beyond channels to ion exchangers.

Introduction

CLC anion transport proteins (Jentsch, 2008), first identified by the cloning of the Cl-channel ClC-0 from Torpedo (Jentsch et al, 1990), associate with dimers of identical or closely related subunits. Each CLC subunit contains an ion translocation pathway that is largely independent from the other subunit (Lorenz et al, 1996; Ludewig et al, 1996; Middleton et al, 1996; Weinreich and Jentsch, 2001; Dutzler et al, 2002; Robertson et al, 2010). Some CLC channels, however, display ‘common gating’ of both pores (Miller and White, 1984; Bauer et al, 1991; Accardi and Pusch, 2000). Eukaryotic CLC proteins have large cytosolic carboxyterminal domains comprising two CBS (cystathionine-β-synthase) domains that in some cases can bind nucleotides like ATP (Meyer et al, 2007) and have a poorly understood role in gating (Fong et al, 1998; Estévez et al, 2004; Bykova et al, 2006; Zhang et al, 2008; Zifarelli and Pusch, 2009b). Crystal structures of bacterial (Dutzler et al, 2002) and algal (Feng et al, 2010) CLC proteins, and of CBS domains from vertebrate CLCs (Meyer and Dutzler, 2006; Markovic and Dutzler, 2007; Meyer et al, 2007), have yielded important insights on how their structure relates to their biophysical properties.

Intriguingly, the CLC gene family comprises both Cl channels and electrogenic Cl/H+-exchangers (Jentsch, 2008). The border between these different transport classes, however, is blurred, as gating of the ClC-0 Cl channel may involve the transport of a proton (Lisal and Maduke, 2008) and because certain anions can uncouple anion flux from proton countertransport (Nguitragool and Miller, 2006; Zdebik et al, 2008; Bergsdorf et al, 2009; Zifarelli and Pusch, 2009a). Moreover, transport activity of mammalian ClC-3 to ClC-6 Cl/H+-exchangers is strongly voltage dependent (Steinmeyer et al, 1995; Friedrich et al, 1999; Li et al, 2002; Neagoe et al, 2010). Their almost instantaneous deactivation at negative voltages precludes measurements of tail currents and it remains unresolved whether their voltage dependence results from a voltage sensitivity of the exchange process per se or from turning the transporter ‘on’ and ‘off’ (‘gating’) (Hebeisen et al, 2003; Zdebik et al, 2008; Picollo et al, 2010; Smith and Lippiat, 2010).

Mammalian endosomal/lysosomal Cl/H+-exchangers (ClC-3 to ClC-7) regulate vesicular H+ and Cl concentration (Jentsch, 2007; Novarino et al, 2010; Weinert et al, 2010). Disruption of endosomal ClC-5 impairs renal endocytosis (Piwon et al, 2000) in Dent's disease (Lloyd et al, 1996), whereas mutations in lysosomal ClC-7 entail osteopetrosis and lysosomal storage disease (Kornak et al, 2001; Kasper et al, 2005). Similar phenotypes were observed when uncoupling point mutations converted these exchangers into pure anion conductors (Novarino et al, 2010; Weinert et al, 2010). ClC-7 needs Ostm1 as β-subunit for protein stability (Lange et al, 2006). Hence, disruption of Ostm1 results in osteopetrosis (Chalhoub et al, 2003) and lysosomal pathology (Lange et al, 2006; Pressey et al, 2010) just like a loss of ClC-7. It has remained unclear which parts of Ostm1 interact with ClC-7 and whether Ostm1 not only stabilizes ClC-7 but also modulates its ion transport activity.

Apart from the acid-secreting membrane of osteoclasts (Kornak et al, 2001; Lange et al, 2006), ClC-7/Ostm1 is absent from the plasma membrane, severely limiting its biophysical characterization. Transport studies of native lysosomes (Graves et al, 2008; Weinert et al, 2010) suggest that ClC-7 mediates Cl/H+-exchange. However, no currents could be measured, essential properties like voltage dependence, kinetics and substrate specificity have remained unknown, and no structure–function analysis could be performed. Here, we exploit the partial plasma membrane expression of recently described ClC-7 mutants which disrupted cytosolic sorting motifs (Stauber and Jentsch, 2010) to characterize the biophysical properties of ClC-7 and its functional interaction with Ostm1. The slow deactivation of ClC-7/Ostm1 resulted in tail currents that revealed functional features that could not be studied with other CLC anion/proton exchangers.

Results

Basic characterization of ClC-7/Ostm1 Cl/H+-exchange

Disrupting two dileucine lysosomal sorting motifs in the cytosolic N-terminus of rat ClC-7 (rClC-7) partially redirects the mutant protein (rClC-7LL23/24AA,LL36/37AA, in short rClC-7PM) to the plasma membrane (Stauber and Jentsch, 2010). Likewise human ClC-7 (hClC-7) carrying the mutations LL23/24AA and LL68/69AA (hClC-7PM) partially traffics to the plasma membrane, as ascertained in a chemiluminescence assay for an added extracytosolic HA tag (Figure 1A). This assay failed to detect hClC-7PM when co-expressed with Ostm1, possibly owing to a shielding of the epitope by the highly glycosylated N-terminus of Ostm1 (Lange et al, 2006). Indeed, both rClC-7PM/Ostm1 and hClC-7PM/Ostm1 gave robust plasma membrane currents (Figure 1B and C for rClC-7PM; Supplementary Figure S1A for hClC-7PM). Since currents of human and rat ClC-7PM were indistinguishable in both Xenopus oocytes and transfected mammalian cells, we refer to both as ClC-7PM in the following.

Figure 1.

Basic characterization of ClC-7PM/Ostm1 in Xenopus oocytes. (A) Chemiluminescence assay for surface detection of hClC-7 and hClC-7PM with an extracytosolic HA tag. hClC-7PM–exHA, but not hClC-7–exHA is detected at the surface of Xenopus oocytes. Co-expression with Ostm1 suppresses the luminescence signal although ClC-7PM/Ostm1 yields plasma membrane currents (in B, C). Mean luminescence intensity (error bars, s.e.m.) normalized to hClC-7PM—exHA from four independent experiments. (B, C) Two-electrode voltage-clamp analysis in Xenopus oocytes. Representative voltage-clamp traces (C) of rClC-7PM (‘WT’), rClC-7PM(E245A) and rClC-7PM(E312A) co-expressed with Ostm1. Arrow indicates tail currents. Voltage was clamped from −80 to +80 mV in 2 s steps of 20 mV (inset). Mean±s.e.m. of currents reached after 2 s plotted (B) as function of voltage (rClC-7PM, n=20; rClC-7PM(E245A), n=13; rClC-7PM(E312A), n=11; uninjected, n=16 oocytes from at least three batches). Virtually identical results were obtained with hClC-7PM (Supplementary Figure S1A). (D, E) Intracellular pH changes of Xenopus oocytes co-expressing rClC-7PM (‘WT’ or glutamate mutants) with Ostm1 in response to a 10-s depolarization. Top traces, clamp currents; bottom traces, pH-dependent BCECF fluorescence measured with the Fluorocyte method (Zdebik et al, 2008). Increased fluorescence means alkalinization. Unless indicated otherwise, extracellular solution contained 96 mM Cl at pH 7.4. For 0 Cl, gluconate replaced Cl. Left traces in (D) and (E) are from the same oocyte and centre and right recordings in (E) are from one oocyte as well. Similar results were obtained with at least five oocytes from three batches.

Expression of ClC-7PM/Ostm1 in Xenopus oocytes (Figure 1B and C; Supplementary Figure S1A), tsA201 or HeLa cells (Supplementary Figure S1B) yielded strongly outwardly rectifying currents that activated slowly at voltages more positive than ∼+20 mV. In stark contrast to ClC-3 through ClC-6 (Friedrich et al, 1999; Li et al, 2002; Matsuda et al, 2008; Neagoe et al, 2010), full activation was not even observed after several seconds and slow deactivation resulted in tail currents at negative voltages (Figure 1C (arrow); Supplementary Figure S1A and B). Whole-cell patch-clamp experiments in HeLa cells showed that ClC-7PM/Ostm1 currents do not require intracellular ATP (Supplementary Figure S1C). We neither observed significant changes in current amplitudes like described for ClC-5 (Zifarelli and Pusch, 2009b) which is known to bind ATP by its CBS domains (Meyer et al, 2007), nor changes in voltage dependence.

As typical for CLC antiporters (Friedrich et al, 1999; Li et al, 2002; Dutzler et al, 2003; Picollo and Pusch, 2005; Scheel et al, 2005; Zdebik et al, 2008; Bergsdorf et al, 2009; Neagoe et al, 2010), mutating the ‘gating glutamate’ (E245 in rat) of ClC-7PM to alanine resulted in almost ohmic, time-independent currents, and changing the ‘proton glutamate’ (E312 in rat) to alanine reduced currents to background levels (Figure 1B and C; Supplementary Figure S1A). ClC-7PM/Ostm1 mediated Cl/H+-exchange as evident from depolarization-induced intracellular alkalinization of Xenopus oocytes expressing these proteins (Figure 1D). In these ‘Fluorocyte’ experiments, the pH-dependent fluorescence of BCECF previously injected into oocytes provides a semi-quantitative measure of cytosolic pH changes in response to depolarizing voltage steps. Depolarization not only activates ClC-7PM/Ostm1 but also provides a driving force for coupled H+-exit/Cl-entry. Outward transport of protons required extracellular Cl (Figure 1E), could occur against its electrochemical gradient (pHo=5.5, Figure 1E), and was abolished by either the E245A or the E312A mutation (Figure 1D). ClC-7PM/Ostm1 currents decreased upon replacing extracellular Cl by I, but unlike ClC-4 and ClC-5 (Friedrich et al, 1999), currents were not larger with NO3 (Figure 2A). Replacing a Cl-coordinating serine by proline (rClC-7 (S202P)) increased the nitrate/chloride conductance ratio as with other CLC antiporters (Bergsdorf et al, 2009; Zifarelli and Pusch, 2009a; Neagoe et al, 2010) and with ClC-0 (Bergsdorf et al, 2009; Picollo et al, 2009). Akin to ClC-4, ClC-5 and ClC-6 (Friedrich et al, 1999; Neagoe et al, 2010; Picollo et al, 2010), currents were decreased by acidic extracellular pH (Figure 2B). In addition to a diminished driving force for Cl/H+-exchange with increased extracellular [H+], faster activation kinetics at more alkaline pHo contributes to the pH dependence of ClC-7PM/Ostm1 outward currents (Figure 2C). Voltage-dependent current activation was also strongly dependent on temperature (Figure 2D). Mono-exponential fits yielded activation rate constants of 2.8±0.2 s−1 at 21°C and 16.6±1.9 s−1 at 37°C, giving an estimate of Q10≈3.

Figure 2.

Modulation of ClC-7/Ostm1 by anions, protons and temperature. (A) Relative anion conductance of oocyte-expressed rClC-7PM/Ostm1 in the presence of different extracellular anions (96 mM). Clamp protocol as in Figure 1C. Mean±s.e.m. of currents reached after 2 s at +80 mV was normalized to the current in Cl for each oocyte (white bars) (Cl, n=32 oocytes; Br, n=7; NO3, n=6; I, n=5; gluconate (gluc), n=5). Grey bar, NO3 conductance of rClC-7PM(S202P)/Ostm1 mutant measured and normalized as above (n=9). (B) Dependence of rClC-7PM/Ostm1 currents on pHo. I/V curves were obtained as in Figure 1B with currents normalized to those at pHo=7.4 and 80 mV; ⩾6 oocytes per data point. (C) Left, typical voltage-clamp traces (top right, protocol) obtained at different pHo values. Note different current scales that were chosen to normalize current amplitudes to the end of +80 mV pulse for better visualizing changes in activation kinetics. Right, τ was determined by single-exponential fit of the 80-mV traces for ⩾6 oocytes per pH value. Mean±s.e.m. as function of pHo. (D) Typical voltage-clamp traces of rClC-7PM/Ostm1 (protocol as in (C)) at different temperatures, representative for 11 oocytes in which temperature was changed between 21, 29°C (n=8) and/or 37°C (n=9).

Slow ClC-7/Ostm1 gating allows characterization of an ‘open exchanger’

The slow deactivation of ClC-7PM/Ostm1 currents provides a unique opportunity to study mammalian Cl/H+-exchange at negative membrane voltages. Using protocols developed for ion channels, we activated (‘opened’) ClC-7PM/Ostm1 by positive prepulses and measured tail currents from transfected HeLa cells at different test voltages (Figure 3A). We increased tail current amplitudes by including 121 mM Cl in the patch pipette. Tail currents were extrapolated to the beginning of the test pulse to obtain the voltage dependence of ‘open exchanger’ currents (Figure 3A). Contrasting with the strong voltage dependence of pseudo-steady-state currents, ‘open exchanger’ currents displayed only very slight outward rectification.

Figure 3.

Tail current analysis of ClC-7PM/Ostm1. (A) After activating HeLa cell-expressed rClC-7PM/Ostm1 by pulses to +80 mV in whole-cell patch-clamp experiments, tail currents were measured at test voltages between −100 and +100 mV. Left, representative current traces (inset, clamp protocol). Right, I/V curve of ‘open exchanger’ obtained by extrapolation to the beginning of test pulses, shown together with ‘pseudo-steady-state’ currents measured after 2 s without preceding activation. Mean values±s.e.m. normalized to the current at +80 mV of 8 (‘open exchanger’) and 19 (‘pseudo-steady-state’) cells. Error bars are mostly smaller than symbols. (B) Determination of nCl/H+-exchange stoichiometry from reversal potentials of tail currents. HeLa cell-expressed rClC-7PM/Ostm1 was clamped using a protocol as in (A), but tail currents were measured at only three voltages close to reversal potentials (−20 to +20 mV or 0 to +40 mV). The contribution of endogenous currents was estimated by short pulses from −80 to 0 mV before activating ClC-7PM/Ostm1 (see Materials and methods). [Cl]o was shifted from 139 to 39 and/or 19 mM Cl (top, representative traces from one cell) and pHo from 7.4 to 6.4 and/or 8.4. Bottom, reversal potentials corrected for background currents and liquid-junction potentials. Crosses, individual measurements. Filled circles and error bars, mean±s.d. Lines, predictions for an nCl/H+-exchanger with n=1, 2 and 3, and for a Cl-channel (1:0) under our experimental conditions. Dashed lines in (A) and (B), I=0. (C) Tail current analysis of hClC-7PM(R762Q)/Ostm1 expressed in HeLa cells to determine Popen(V). Clamp protocol at bottom. (D) Apparent open probability Popen as function of prepulse voltage, determined from tail currents as shown in (C). The line shows the fit by the Boltzman function Popen=1/(1+exp(zn × e0(V½V)/kT)), which yielded zn=1.32 and V½=82 mV. Values are mean of five experiments. Error bars represent s.e.m.

Whereas the strong rectification and near-instantaneous deactivation of ClC-4 and ClC-5 precludes measurements of reversal potentials (Steinmeyer et al, 1995; Friedrich et al, 1999), ClC-7PM/Ostm1 tail currents allowed us to determine Cl/H+-coupling ratios from Cl and H+-dependent shifts in reversal potentials (Figure 3B). Our results were best fitted by a 2Cl:1H+ stoichiometry. The apparent deviation from this stoichiometry at nominal pHo of 8.4 might be explained by depolarization-induced outward transport of protons through the exchanger (Zifarelli and Pusch, 2009a). This process is expected to cause a larger deviation of actual from nominal pH at the lower H+ concentrations of more alkaline pH.

Whereas the tail current analysis of instantaneous ‘open exchanger’ currents requires the same open probability Popen at the beginning of test pulses (as indicated above by identical macroscopic currents), Popen must have reached steady state at the respective voltage when Popen is determined as a function of voltage by tail currents. The slow voltage-dependent activation of ClC-7PM/Ostm1, however, precluded reliable measurements of steady-state currents that are needed for this analysis. We, therefore, resorted to a ClC-7 point mutant (R762Q; described below) that drastically accelerates activation. ClC-7PM(R762Q)/Ostm1 currents reached steady state already ∼400 ms after the beginning of voltage steps (Figure 3C). At t=500 ms, tail currents were measured at a constant test voltage (+80 mV) as function of the voltage of the preceding pulse (between −40 and +140 mV). After correcting for endogenous HeLa cell currents, apparent Popen(V) was obtained by extrapolating tail currents to the time of the voltage step (see Materials and methods). Boltzmann fits (Figure 3D) revealed a voltage of half-maximal activation V½≈82 mV and an apparent gating charge of zn≈1.32. Although we performed this tail analysis study with a mutant, we expect the values of ‘WT’ ClC-7/Ostm1 to be similar.

Structural basis and functional consequences of Ostm1–ClC-7 interactions

So far, all experiments on ClC-7PM were performed in co-expression with Ostm1. When we expressed ClC-7PM with or without Ostm1 in HeLa cells (Supplementary Figure S1D) or Xenopus oocytes (not shown), ClC-7PM yielded currents only together with Ostm1. Since ClC-7PM clearly reaches the plasma membrane also without Ostm1 (Figure 1A; Stauber and Jentsch, 2010), these results indicate that Ostm1 is needed to activate ClC-7 ion transport.

We next asked which parts of Ostm1 interact with ClC-7. We constructed chimeras with CD4, a protein that shares the type I transmembrane topology of Ostm1 (Lange et al, 2006) but traffics to the plasma membrane by default. The extracellular, transmembrane and intracellular domains of Ostm1 were replaced by those of CD4 either individually or in combination. Without ClC-7, Ostm1 stays in the endoplasmic reticulum (ER), whereas a portion of Ostm1 reaches lysosomes upon co-expression with ClC-7 (Lange et al, 2006). We first ascertained that Ostm1 and Ostm1/CD4 chimeras carrying C-terminal green fluorescent protein (GFP) tags were confined to the ER and/or plasma membrane of transfected HeLa cells (Figure 4A). We then co-transfected GFP-tagged Ostm1/CD4 chimeras with ClC-7 and assayed the co-localization of GFP fluorescence with the lysosomal marker LAMP-1 as read-out for Ostm1–ClC-7 interaction (Figure 4B and C; Supplementary Figure S2). The transmembrane domain (TMD) of Ostm1 was necessary and sufficient for Ostm1 constructs being carried to lysosomes by ClC-7 (Figure 4C).

Figure 4.

Domains of Ostm1 that interact with ClC-7. (A) When transfected into HeLa cells, Ostm1–GFP localizes to the ER, CD4–GFP mostly to the plasma membrane and a GFP-tagged CD4 chimera containing the TMD of Ostm1 (COC–GFP) to the ER and plasma membrane. (B) When co-transfected with HA-tagged rClC-7, Ostm1–GFP and COC–GFP, but not CD4–GFP, co-localized with rClC-7–HA (immunolabelled for the HA epitope) to late endosomes/lysosomes (marked by immunolabelling for LAMP-1). (C) Statistical analysis of lysosomal targeting of Ostm1, CD4 or chimeras thereof expressed either without (−) or with (+) rClC-7–HA assayed as in (A) and (B). In the 3-letter abbreviations, C means CD4, O means Ostm1, in the sequence extracytosolic N-terminal part, TMD, and cytoplasmic C-terminus. Means of 3–4 independent experiments with >100 cells each evaluated. Error bars represent s.e.m. Constructs containing the TMD of Ostm1 localized to lysosomes upon co-expression with rClC-7. (D) Typical current traces of Xenopus oocytes co-expressing rClC-7PM with Ostm1, CD4 or CD4/Ostm1 chimeras. Expression of significant currents required the presence of both the N-terminus and TMD of Ostm1. Similar results were obtained with at least 10 oocytes of at least 3 batches.

While these experiments suggested that the TMD of Ostm1 binds ClC-7, other parts of Ostm1 may modulate ClC-7 transport activity. We, therefore, assayed currents of Xenopus oocytes co-expressing ClC-7PM and Ostm1/CD4 chimeras (Figure 4D). As expected from our localization assay, currents were not detectable when the TMD of Ostm1 was replaced by that of CD4. Even larger currents were observed when ClC-7PM was co-expressed with a chimera (OOC), in which the Ostm1 C-terminus was replaced by that of CD4. By contrast, currents were indistinguishable from background with chimeras lacking the Ostm1 N-terminus (Figure 4D), even though ClC-7PM was still able to reach the plasma membrane (Supplementary Figure S3). Hence, both the N-terminus and TMD of Ostm1 are required for ClC-7 transport activity.

Functional effects of human CLCN7 mutations underlying osteopetrosis

The plasma membrane expression of ClC-7PM/Ostm1 allowed us for the first time to study functional consequences of disease-causing CLCN7 mutations (Cleiren et al, 2001; Kornak et al, 2001; Frattini et al, 2003; Waguespack et al, 2003; Letizia et al, 2004; Pangrazio et al, 2010; Phadke et al, 2010), which we selected from different categories based on the mode of inheritance (recessive versus dominant) and on their location within the protein (transmembrane region or cytoplasmic CBS domain) (Figure 5A; Supplementary Figure S4).

Figure 5.

Characterization of osteopetrosis-causing mutations in human ClC-7. (A) Position of analysed dominant and recessive osteopetrosis-causing mutations (solid and open stars, respectively) in a CLC topology model (Dutzler et al, 2002). Mutations yielding no currents shown in purple (when retained with Ostm1–GFP in the ER of HeLa cells) and red (when exported from the ER), those with apparently normal currents in green, and those with accelerated activation in blue (see (C, D) and Supplementary Figure S5). (B) Close-up of X-ray structure of CmClC (Feng et al, 2010) displaying the location of ClC-7 residues that accelerate gating when mutated. Except for L213 (corresponding to L174 in CmClC), the depicted ClC-7 residues are not identical to those of CmClC at these positions (R286, P619, R762 and R767 of hClC-7 correspond to L241, R532, V680 and S685, respectively, in CmClC) (Feng et al, 2010). The transmembrane part of one subunit is shown in grey, CBS2 and the linker to CBS1 of that subunit in red and yellow, respectively. Green helices at left are from the second subunit of the homodimer. (C) Representative current traces of hClC-7PM (‘WT’ or selected osteopetrosis-causing mutants) upon expression with Ostm1 in Xenopus oocytes. Mutants yielded either no or very low currents (R526W and L490F), apparently normal currents (S744F), or displayed accelerated activation (R762Q). (D) Subcellular localization of hClC-7 and selected mutants co-expressed with Ostm1–GFP in HeLa cells. Cells were immunolabelled for hClC-7 and LAMP-2. In most cases, hClC-7 co-localized with Ostm1–GFP to LAMP-2-positive structures in addition to localizing to ER-like structures. However, hClC-7(R526W) remains with Ostm1–GFP in the ER (highlighted by nuclear envelope staining). In cells with a clear excess of Ostm1–GFP (asterisk), it predominantly localizes to the ER.

Within all CLCN7 mutation categories mentioned above, we found mutations that abolished or strongly reduced currents (Figure 5C; Supplementary Figure S5A). Surprisingly, other mutations either left ClC-7PM currents virtually unchanged (e.g., S744F; Frattini et al, 2003) or accelerated their activation kinetics between moderately (e.g., the frameshift mutation G796fs; Cleiren et al, 2001) and dramatically (e.g., R762Q; Kornak et al, 2001) (Figure 5C; Supplementary Figure S5A). Mutants that yielded plasma membrane currents also transported H+ as revealed by Fluorocyte (Zdebik et al, 2008) experiments (not shown). To test whether changed subcellular targeting of ClC-7/Ostm1 might explain the disease-causing effect of those mutants, we inserted them into hClC-7 instead of hClC-7PM and co-expressed them with GFP-tagged Ostm1 in HeLa cells. However, in addition to a pronounced ER-like labelling, all mutants that gave currents (in hClC-7PM) reached their normal destination (lysosomes) where they co-localized with Ostm1 (Figure 5D; Supplementary Figure S5B). Only three of the mutants with reduced currents were retained in the ER, whereas the other six partially localized with Ostm1 to late endosomes/lysosomes (Figure 5D; Supplementary Figure S5B). In the absence of lysosomal targeting sequences, ClC-7 reaches the plasma membrane by default once it has left the ER (Stauber and Jentsch, 2010). Hence, normal lysosomal targeting (in hClC-7) of the V297M, F318L, L490F, L651P, R767P and R767W mutants, all of which reduce or abolish plasma membrane currents in hClC-7PM, suggests that these mutations may interfere directly with the ion transport of ClC-7 or with the mechanism by which Ostm1 activates ClC-7. One should note, however, that all these mutants were able to carry Ostm1 to lysosomes. A reduction in the expression level due to limited stability may also contribute to reduced currents. Western blot analysis showed that this was not the case for the V297M and F318L mutants, whereas protein levels were markedly reduced with the L490F mutant (Supplementary Figure S6).

Discussion

Despite the medical importance of ClC-7/Ostm1 and its crucial role in lysosomal function (Kornak et al, 2001; Kasper et al, 2005; Lange et al, 2006; Wartosch et al, 2009; Weinert et al, 2010), the only available information concerning its biophysical properties has remained its ability to perform Cl/H+-exchange (Graves et al, 2008; Weinert et al, 2010). Acid-activated currents previously ascribed to ClC-7 (Diewald et al, 2002) most likely represent currents endogenous to the expression systems (Jentsch, 2008). Exploiting the partial plasma membrane localization of ClC-7 mutants which we have recently described (Stauber and Jentsch, 2010), we could now study for the first time important details of ion transport properties, investigate effects of human disease-causing mutations, and show that ClC-7 needs specific domains of the Ostm1 β-subunit not only for protein stability (Lange et al, 2006), but also for ion transport activity.

Voltage gating of intrinsically linear voltage-dependent 2Cl/H+-exchange

Several properties of ClC-7/Ostm1 described here have recently emerged as being typical for mammalian CLC exchangers (Jentsch, 2008). This includes the preference of Cl over I, decreased transport activity with acidic external pH, strong outward rectification and effects of neutralizing ‘gating’ and ‘proton’ glutamates which result in an uncoupling of anion movement from protons and an apparent abrogation of all ion transport, respectively.

While ClC-7/Ostm1 differs from ClC-3 to ClC-6 with respect to NO3 selectivity, the most important biophysical difference to those transporters is the slow voltage-dependent activation and deactivation of ClC-7/Ostm1. Whereas ClC-4 and ClC-5 show a similarly steep voltage dependence with significant currents being observable only at positive voltages, a major component of the current activated upon depolarization appears instantaneous, with the remaining <30% reaching steady state in less than ∼100 ms (Steinmeyer et al, 1995; Friedrich et al, 1999). Despite the drastic difference in gating kinetics, the V½ of ClC-7PM(R762Q)/Ostm1 (∼82 mV) agrees well with that of ClC-4 measured in the presence of uncoupling anions (Orhan et al, 2011). Most importantly, currents of ClC-4 and ClC-5 deactivated almost instantaneously at negative voltages, with no tail currents being detectable (Steinmeyer et al, 1995; Friedrich et al, 1999). Therefore, it was impossible to determine whether their Cl/H+-exchange can function, in principle, also at negative voltages (i.e., lumen positive for vesicles).

ClC-7/Ostm1 currents, by contrast, did not reach steady state even after >2 s. We have avoided using longer pulses to minimize confounding local changes in Cl and H+ concentrations. Although deactivation upon stepping back to negative potentials was faster than activation, it resulted in respectable tail currents, in particular when [Cl]i was kept high in whole-cell patch-clamp measurements. Currents at negative voltages clearly demonstrated that net Cl/H+-exchange can occur in both directions, which has not been demonstrated previously for any mammalian CLC. Instantaneous tail currents displayed an almost linear dependence on voltage. As their slight outward rectification might be owed to the difference in [H+] and [Cl] on both sides of the membrane, we propose that the intrinsic Cl/H+-exchange rate is nearly linearly related to the driving force. This hypothesis needs to be tested in future studies under a broader range of conditions. The almost ohmic tail currents also indicate that the model of Zdebik et al (2008), in which the voltage dependence of ClC-5 may be related to a voltage-driven transport of cytosolic H+ to the central exchange site, does not apply for ClC-7/Ostm1. It has also been questioned for ClC-5 (Picollo et al, 2010).

We conclude that voltage dependence of ClC-7/Ostm1 is almost exclusively caused by a slow activation/deactivation process that we like to refer to as ‘gating’ in analogy to the terminology used for ion channels. CLC Cl channels have two kinds of gates, one for each pore of the (homo)dimeric channel (the ‘protopore gate’, which relates to the ‘gating glutamate’) and a less well-understood ‘common gate’ that acts on both pores simultaneously (Jentsch, 2008). In the Torpedo channel ClC-0, the protopore gate is fast and has a Q10 of ∼2.2, whereas the common gate is slow with a Q10 of ∼40 (Pusch et al, 1997). However, protopore and common gates of the mammalian muscle Cl channel ClC-1 display Q10 values of ∼3 and ∼4, respectively (Bennetts et al, 2001). Therefore, the Q10 of ClC-7/Ostm1 activation cannot serve as criterion for deciding whether the underlying process is analogous to protopore or common gating.

The majority of those human CLCN7 mutations found to accelerate the ‘gating’ of ClC-7/Ostm1 affect residues in the ClC-7 C-terminus. Two of these residues (R762 and R767) are located at the surface of the second CBS domain in close proximity to the transmembrane part as revealed by the X-ray structure of algal CmClC (Feng et al, 2010). Intriguingly, two other ‘accelerating’ mutations (L213F and R286Q) affect residues in the membrane part which are in close proximity to CBS2 (Figure 5B; Supplementary Video 1).

Hence, the slow gating of ClC-7/Ostm1 may involve interactions of CBS2 with the transmembrane part of ClC-7. As the cytoplasmic C-termini were implicated in ‘common gating’ of ClC-0 (Fong et al, 1998; Estévez et al, 2004; Bykova et al, 2006), these observations indirectly suggest that ClC-7/Ostm1 activity is regulated by a ‘common gate’.

Voltage dependence was abolished by the uncoupling mutation E245A as in other CLC transporters (Friedrich et al, 1999; Picollo and Pusch, 2005; Scheel et al, 2005; Bergsdorf et al, 2009; Neagoe et al, 2010). This result is compatible with the notion that E245, which may change its position during 2Cl/H+-exchange cycles (Feng et al, 2010), acts as a ‘gate’. The short distance, which its negative side chain may move in the electric field, however, seems to exclude a function as voltage sensor as it would result in an apparent ‘gating charge’ much smaller than 1. The apparent gating valence determined here for ClC-7PM(R762Q)/Ostm1 (zn≈1.32) agrees well with that of ClC-4 in the presence of uncoupling anions (Orhan et al, 2011) and with gating charges determined for ClC-0 (Hanke and Miller, 1983; Bauer et al, 1991; Pusch et al, 1995), ClC-1 (Pusch et al, 1994) and ClC-2 (de Santiago et al, 2005) Cl channels. We envisage a conformational change, possibly involving cytosolic CBS domains, that fixes the position of E245 ‘gating glutamate’, thereby resulting in a ‘closed state’ of the transporter. Whether the ‘gating charge’ is supplied by permeant ions (Pusch et al, 1995; Lisal and Maduke, 2008), intrinsic charges of the protein or a combination thereof (Smith and Lippiat, 2010), remains an open question.

The tail currents of ClC-7/Ostm1 provided the unique possibility to determine the Cl/H+-exchange stoichiometry from reversal potentials. These cannot be measured reliably with other endosomal/lysosomal CLC exchangers. Our data were best fit by a 2Cl:1H+ stoichiometry as originally described for the prokaryotic EcClC-1 (Accardi and Miller, 2004) and which might be rationalized in terms of a switch in the position of the ‘gating glutamate’ (Feng et al, 2010).

It is intriguing that ClC-7/Ostm1, just like the other vesicular CLC Cl/H+-exchangers (Steinmeyer et al, 1995; Friedrich et al, 1999; Li et al, 2002; Neagoe et al, 2010), displays such strong voltage dependence. This rectification is not an inevitable consequence of the Cl/H+-exchange per se, as our measurements of ‘open transporter’ currents have shown. This suggests that the rectification might have a physiological importance, which, however, remains obscure. The strong outward rectification of ClC-4 and ClC-5 has puzzled the field for a long time because it implies that these transporters are almost inactive in inside-positive endosomes (Jentsch et al, 2002; Jentsch, 2007). However, recent model calculations showed that vesicles may attain an inside-negative voltage through the activity of 2Cl/H+-exchangers (Weinert et al, 2010).

TMD and exoplasmic N-terminus of Ostm1 are needed for ClC-7 ion transport activity

ClC-7 is the only CLC exchanger known to need an accessory subunit (Lange et al, 2006). This β-subunit, Ostm1, is a small type I transmembrane protein with a highly glycosylated N-terminus (Lange et al, 2006) that was identified as being truncated in the osteopetrotic grey lethal mouse (Chalhoub et al, 2003). The pathology resulting from a loss of Ostm1 was explained by the concomitant loss of ClC-7 which is unstable without its β-subunit (Lange et al, 2006). Our work now shows that Ostm1 is also needed for the ion transport activity of ClC-7. Hence, the low levels of ClC-7 (∼5% of WT protein) remaining in tissues of Ostm1−/− mice (Lange et al, 2006) are non-functional. Disruption of Ostm1 leads to a complete loss of ClC-7 ion transport.

To elucidate which parts of Ostm1 interact with ClC-7, we assayed the ClC-7-dependent trafficking of Ostm1 to lysosomes, an assay that turned out to be more reliable than co-immunoprecipitation (Lange et al, 2006). The single TMD of Ostm1 was necessary and sufficient for correct trafficking, suggesting that Ostm1–ClC-7 binding involves interactions between transmembrane helices of both ClC-7 and Ostm1. Consistent with the lack of interaction in the trafficking assay, a chimera in which the Ostm1 TMD was replaced by that of CD4 failed to activate ClC-7PM ion transport. Surprisingly, ClC-7PM transport activity also required the highly glycosylated extracytosolic aminoterminus of Ostm1. Normally, Ostm1 is cleaved proteolytically in (or on its way to) lysosomes, but the cleavage products are still bound together by disulphide bonds (Lange et al, 2006). Since it is unlikely that ClC-7PM/Ostm1 reaches the plasma membrane through a prelysosomal/lysosomal compartment, our results provide circumstantial evidence that cleavage of Ostm1 is not needed for its ability to stimulate ion transport of ClC-7. The cytosolic C-terminus of Ostm1 was not needed, although we cannot exclude that it modulates ClC-7 transport as hinted at by the larger currents induced by the OOC chimera. Barttin, a protein that only interacts with ClC-K Cl channels (Estévez et al, 2001), is the only other β-subunit of CLC proteins known to date. Like Ostm1, barttin is thought to bind to ClC-K α-subunits through its TMDs (Scholl et al, 2006; Tajima et al, 2007), but activation of ion transport requires its intracellular carboxyterminus (Scholl et al, 2006).

Effects of disease-causing CLCN7 mutations

Roughly 40 different mutations in CLCN7 have been found in humans suffering from osteopetrosis, from which we have selected 18 for the present analysis. No consistent differences concerning their effects on ion transport or trafficking were found when comparing recessive and dominant, or membrane- and CBS-domain localized mutations. As expected, several mutants yielded reduced currents, and in others voltage-activated currents were not detectable. G240R, G521R and R526W, all of which alter charges in a TMD, led to retention in the ER irrespective of co-expression with Ostm1. Like previously reported for the G215R mutant (Schulz et al, 2010), however, some mutants overcame ER retention when co-expressed with Ostm1 (data not shown). Several mutations reduced or abolished currents although normal lysosomal targeting (when inserted into hClC-7/Ostm1) indirectly indicated that they reached the plasma membrane in hClC-7PM/Ostm1.

Surprisingly, half of the disease-causing mutations gave currents when studied in ClC-7PM/Ostm1. In many cases, these currents showed accelerated kinetics of activation, such as L213F, R286Q, R762Q, R762L and R767Q. When expressed in HeLa cells together with Ostm1, these mutants (in hClC-7) could reach lysosomes. This raises the question whether the slow activation kinetics of ClC-7/Ostm1 is needed for its physiological function. However, trafficking and protein stability might be different in vivo as compared with heterologous overexpression. Indeed, western blots and immunofluorescence previously failed to detect ClC-7 protein in fibroblasts from a patient heterozygous for an early stop codon in CLCN7 and the ‘accelerating’ R762Q mutation (Kornak et al, 2001).

Conclusions and outlook

While ClC-7 shares many biophysical properties with the other mammalian CLC Cl/H+-exchangers, it is unique in that it needs a β-subunit (Ostm1) and that its activation and inactivation by voltage is much slower.

Whereas an interaction through TMDs, as observed here for ClC-7/Ostm1, is not without precedent (Scholl et al, 2006; Tajima et al, 2007), the apparently strict dependence of ion transport activity on the highly glycosylated Ostm1 N-terminus (Lange et al, 2006) is surprising. It raises the question whether sugar moieties on Ostm1 interact with the rather limited extracytosolic part of the ClC-7 protein and how this activates ion transport. This activation might involve the same gating mechanism that ‘opens’ and ‘closes’ ClC-7/Ostm1 like an ion channel—the difference being that it is not a purely diffusive pore, but ion exchange which is ‘gated’. It is intriguing that ClC-7/Ostm1 is the only endosomal/lysosomal CLC, which is gated so slowly and that several disease-causing missense mutations in the ClC-7 CBS2 domain accelerate its activation. Interestingly, there is not even one missense mutation in a CBS domain among the many CLCN5 mutations identified so far in Dent's disease (Lloyd et al, 1996; Jentsch, 2008). Together with the fact that ClC-5 activates and deactivates much faster, these observations suggest that the sluggish voltage dependence of ClC-7/Ostm1 might have a physiological role. This seems surprising in view of its localization to the non-excitable membranes of lysosomes and the osteoclast ruffled border, but abrupt voltage changes that would not significantly ‘open’ ClC-7/Ostm1 seem possible—for instance, in second messenger-induced release of Ca2+ from lysosomes.

Materials and methods

Expression constructs and antibodies

For expression of untagged proteins in cell culture, constructs were subcloned into pcDNA3 or pcDNA3.1(+) vector (both from Invitrogen). Constructs for rat ClC-7 in this vector have been described previously (Stauber and Jentsch, 2010). The construct for expression of fluorescently tagged Ostm1 in the pEGFP-N3 vector (Clontech) linking Ostm1 at the C-terminus with GFP by the sequence VDGTAGPGSIAT has been described (Stauber and Jentsch, 2010). To generate chimeric constructs between Ostm1 and CD4, the DNA sequences encoding the aminoterminal part (aa 1–286 for Ostm1 and aa 1–397 for CD4), the transmembrane region (estimated by hydrophobicity; aa 287–309 for Ostm1 and aa 398–420 for CD4) and the carboxyterminal region (aa 310–338 for Ostm1 and aa 421–458 for CD4) of the respective protein were combined by recombinant PCR and cloned into pEGFP-N3 vector.

Constructs for heterologous expression in Xenopus oocytes were cloned into pTLN vector (Lorenz et al, 1996). For the surface expression assay (below), a HA epitope was inserted between residues E168 and K169 (extracytosolic loop between helices B and C) of hClC-7. Point mutations were introduced by PCR. For the G796fs mutation of hClC-7, the 3′UTR until the new stop codon was added. All constructs were confirmed by sequencing the complete ORF.

Primary antibodies were mouse anti-LAMP-1, anti-LAMP-2 (both DSHB; clones H4A3 and H4B4, respectively) and anti-HA epitope (Covance, clone HA.11), rat anti-HA epitope (Roche, clone 3F10) and rabbit anti-ClC-7 (7N4B; Kornak et al, 2001). Secondary antibodies conjugated to AlexaFluor 488, 546 or 633 were from Molecular Probes, HRP-conjugated antibodies were from Jackson ImmunoResearch.

Voltage-clamp, qualitative pHi determinations and surface expression assay in Xenopus laevis oocytes

Xenopus laevis oocytes were injected with cRNA (23 ng for ClC-7 constructs and additional 23 ng for Ostm1 or Ostm1/CD4 chimeras) which was transcribed with the mMessage Machine kit (Ambion) from pTLN vector (Lorenz et al, 1996). After 3 days incubation at 17°C, currents were measured using standard two-electrode voltage clamp (TEVC) employing TurboTEC amplifiers (npi electronic) and pClamp10.2 software (Molecular Devices). Measurements were performed at room temperature or, when specified, the temperature was regulated by a TC-344 Heater Controller (Warner).

Oocytes were superfused with modified ND96 saline (96 mM NaCl, 2 mM K-gluconate, 1.8 mM Ca-gluconate and 1 mM Mg-gluconate). pH was buffered with 5 mM HEPES, Tris, MES or 10 mM phosphate as appropriate. Ion substitutions were done by replacing NaCl with equimolar amounts of NaNO3, NaI, NaBr or Na-gluconate. Proton transport activity was measured qualitatively by recording intracellular pH (pHi) changes exploiting the pH-sensitive fluorescence of BCECF (injected 10–30 min before the experiment) in the Fluorocyte (Zdebik et al, 2008). ClC-7/Ostm1 was activated by an extended depolarization to +80 mV for 10 s followed by 3 s hyperpolarization to −80 mV, with a holding potential of −30 mV using TEVC. Currents and fluorescence changes were recorded simultaneously with the pClamp10.2 software.

To determine the temperature dependence of activation rate constants R=1/τ of ClC-7PM/Ostm1, only the first 250 ms of depolarization (+80 mV) were fitted to an exponential function, because we observed a slow component at 37°C that significantly adds to the activation at later times. This was also observed with uninjected oocytes. Q10 values were determined by Q10=(R2/R1)10/(T2T1), with R1 and R2 being the activation rate constants at temperatures T1 (37°C) and T2 (21°C), respectively.

Surface expression of HA-tagged ClC-7 protein was determined 3 days after cRNA injection by an immunochemiluminescence assay as described (Zerangue et al, 1999) with up to 20 oocytes per construct and experiment.

Patch-clamp experiments

Whole-cell patch-clamp measurements on cultured cells used patch pipettes of 3–5 MΩ resistance filled with (in mM) 110 CsCl, 10 NaCl, 0.5 CaCl2, 1 EGTA, 2 MgATP, 40 HEPES, pH 7.2. The calculated free Ca2+ concentration was 180 nM. The bath solution contained (in mM) 130 NaCl, 5 KCl, 1 MgCl2, 1 CaCl2, 10 glucose, 20 HEPES, pH 7.4 with NaOH. Osmolarity was adjusted with sucrose to 280–290 mOsmol/l for the pipette solution and to 300 mOsmol/l for the extracellular solution. To alter Cl and pH gradients, NaCl was replaced partially by Na-gluconate and pH was buffered with 20 mM Tris (pH 8.4) or 20 mM MES (pH 6.4). Data were acquired with an EPC-10 double amplifier and Pulse software (HEKA).

Tail current analysis

To measure ‘open exchanger’ currents, activation pulses of 1 s to +80 mV were applied followed by 500 ms test voltage steps of 20 mV from −100 to +100 mV. To determine reversal potentials, the length and voltage of the activation pulse were adjusted to yield similarly strong current amplitudes at different [Cl]o and pHo values. Activation pulses were preceded by ‘control pulses’ of 50 ms to −80, −40 and 0 mV to estimate background currents and their reversal potentials. Test pulses were applied in 20 mV steps between 0 and +40 or −20 and +20 mV. Reversal potentials of background currents (Erev,bg) and of tail currents (Erev,meas) were determined by intrapolation. Reversal potentials of ClC-7PM/Ostm1 (Erev,ClC-7) were then calculated by

display math

(where γbg is the background slope conductance and γtot is the total slope conductance at the test pulses) and corrected for liquid-junction potentials. The equation

display math

(where R, T and F have the usual meaning) was used for reversal potential calculation for an mCl/nH+-exchanger with different coupling ratios of m:n.

To determine the voltage dependence of gating, we transfected HeLa cells with human ClC-7PM(R762Q) and Ostm1–GFP. Adding GFP to the C-terminus of Ostm1 did not change currents, but selecting for fluorescent cells increased the success rate of patch clamping. Fits of mono-exponential functions to tail currents measured as in Figure 3C gave currents at the time of the voltage step. These values were corrected by subtracting mean currents of non-transfected cells at +80 mV (the test voltage). These corrected currents were normalized to currents at +140 mV, and normalized currents were averaged from five experiments. These data were fitted to a Boltzmann function of the form:

display math

with e0 being the elementary charge, V the voltage, k the Boltzmann constant and T the temperature, to yield V½ (the voltage for half-maximal activation) and zn (the gating charge (or valence)). Apparent Popen (Figure 3D) was obtained by normalization to obtain Popen=1 for V → ∞.

Cell culture and immunofluorescence microscopy

Plasmid DNA encoding the respective construct(s) was transfected into HeLa or tsA201 cells using FuGENE6 (Roche). Cells were grown at 37°C for 24–48 h before patch-clamp experiments or fixation with 4% PFA in PBS for 15 min. For immunolabelling, cells were incubated with 30 mM glycine in PBS for 5 min and permeabilized with 0.1% saponin in PBS for 10 min. Both primary and AlexaFluor-coupled secondary antibodies were applied in PBS/0.05% saponin supplemented with 3% BSA. Images were acquired with an LSM510 laser scanning confocal microscope equipped with a × 63 1.4 NA oil-immersion lens (Zeiss).

To evaluate ClC-7-dependent trafficking of Ostm1/CD4 chimeras to late endosomes/lysosomes, GFP-tagged chimeras were assayed by co-localization of GFP fluorescence with LAMP-1 immunolabelling. For double-transfected cells, only those cells in which rClC-7–HA co-localized with LAMP-1 were considered. More than 100 cells were counted per construct and experiment. Three to four independent experiments were performed per construct, with the investigator blinded to the kind of chimera.

Supplementary data

Supplementary data are available at The EMBO Journal Online (http://www.embojournal.org).

Acknowledgements

We thank Janet Liebold, Nicole Krönke, Patrick Seidler and Silke Zillmann for technical assistance, Anselm A Zdebik for discussions, Andrew J Plested for critical reading of the paper, Philipp F Lange for preliminary studies on the ClC-7/Ostm1 interaction and Pawel Fidzinski for advice on patch clamping. This work was supported, in part, by the Deutsche Forschungsgemeinschaft (JE 164/7).

Author contributions: The study was conceived and designed by TJJ and TS, with contributions of LL and CFL. TS generated constructs, and investigated expression, trafficking and localization of mutants. LL and CFL performed electrophysiological experiments and FAW trafficking studies to determine ClC-7/Ostm1 interactions. LL, CFL, TJJ and TS analysed the data. TJJ and TS wrote the paper, with contributions of LL and CFL.

Conflict of Interest

The authors declare that they have no conflict of interest.

Ancillary