All eukaryotic ClCs (and some prokaryotic ClCs) have a large cytoplasmic domain involved in modulating the trafficking and function of ClC proteins (Estévez et al., 2004; Hebeisen et al., 2004). Mutations in the cytoplasmic domains cause severe defects in slow gating, and are also associated with human genetic diseases (Fong et al., 1998; Estévez et al., 2004; Puljak and Kilic, 2006; Planells-Cases and Jentsch, 2009; Kim, 2014; Stölting et al., 2014b). The crystal structures of cytoplasmic domains from ClC-0, ClC-Ka, and ClC-5 have been resolved; cytoplasmic domains of each subunit contain two CBS domains that interact with one another via an extensive interface. The CBS domains also interact with the transmembrane part of the same subunit, and with CBS domains of the other subunit. Additionally, the cytoplasmic domains display a dimeric organization resembling the transmembranal architecture (Meyer and Dutzler, 2006; Markovic and Dutzler, 2007; Meyer et al., 2007; Feng et al., 2010; Park et al., 2016).

The cytoplasmic domains connect with the α-helix R, which contains Tyr_cen that participates directly in Cl^- coordination during transport. As mutations in the cytoplasmic domains are involved in genetic diseases, several studies have addressed the influence of alterations in the cytoplasmic domains in channel gating behavior. A point mutation downstream of the second CBS domain (A885P) in ClC-1 results in a dramatic reduction in channel open probability at voltages near the optimal membrane potential for ClC-1 to function (Beck et al., 1996). Two truncated ClC-1 mutants (R875X and K894X), the first removing the whole region downstream of CBS2 and the second mimicking a naturally occurring mutation in myotonic patients, display changes in anion binding affinity, resulting in changes in the voltage dependence for both fast and slow gates (Hebeisen and Fahlke, 2005). He et al. (2006) analyzed two splice variants of Caenorhabditis elegans ClC channel, CLH3a and CLH3b, which display marked gating differences. CHL3a has a N-terminal splice insertion that when deleted do not alter gating properties. CHL3b has two splice insertions at the cytoplasmic domains, one between the two CBS domains and the second distal to CBS2. Deletion of either the insertion distal to CBS2 or the last 11 amino acids of CBS1 gives rise to channels with gating properties similar to CHL3a (He et al., 2006). Those studies demonstrate that alterations at the cytoplasmic domains modify the conformation of the pore affecting channel gating.

Cytoplasmic domains of some ClC proteins interact with adenosine nucleotides. In ClC-1, binding of intracellular ATP inhibits the channel by stabilizing it in its closed state (Bennetts et al., 2005). ATP binding has the opposite effect in ClC-5 exchangers, activating the transporter. Binding of ATP to ClC-2 slows down the rate of activation and deactivation, but does not affect the maximal open probability of the channel (Stölting et al., 2013). Nucleotides bind at the interface between the two CBS domains, as revealed by the crystal structure of ClC-5. The nucleotide binding site has no catalytic properties, and to it AMP, ADP, and ATP bind with equal affinity (Meyer et al., 2007). There are no apparent nucleotide binding sites in the cytoplasmic domains of ClC-0 or ClC-Ka (Meyer and Dutzler, 2006; Markovic and Dutzler, 2007). Interestingly, the cytoplasmic domains of the plant ClC AtClC-a also interacts with adenosine nucleotides. At this exchanger, ATP reduces transport activity by a maximum of 60%. Unlike ClC-5, only ATP produces this effect, with AMP working only as a competitor limiting ATP inhibition when present in solution (De Angeli et al., 2009).

The antiparallel dimerization observed with the CBS domains of ClC proteins following ATP binding is a feature also seen in the CFTR chloride channel, in which ATP binds at two conserved motifs at the interface of two intracellular nucleotide binding domains. These domains dimerize in a head-to-tail conformation leading to channel gating and chloride movement following conformational changes. By analogy, one may speculate that adenosine nucleotide binding to the CBS domains could cause protein rearrangements that affect channel behavior in some ClCs.

In contrast to the well-studied fast gating mechanism of ClC proteins, the molecular mechanism of the slow (common) gate is still obscure. Evidence suggests that extensive conformational rearrangements in the protein could contribute to the slow gate (Pusch et al., 1997; Duffield et al., 2005; Bykova et al., 2006; Ma et al., 2011). Two facts support the idea that conformational changes promoted by the cytoplasmic domains may lead to the movement of critical transmembrane helices and play an important role in the common gating mechanism. First, point mutations at helices localized at the dimer interface cause changes in common gating (Duffield et al., 2003). Second, the crystal structure of the eukaryotic ClC transporter show relevant connections between the CBS domains and helices H and I, localized at the dimer interface (Feng et al., 2010).

Another interesting hypothesis suggests that ClC channels likely behave as ‘broken’ exchangers in which proton transport is involved in the common gating, suggesting that the conformational changes of channels’ common gating and coupled Cl^-/H⁺ transport have an evolutionary linkage (Lísal and Maduke, 2008). Since gating of ClC-0 channels is not in a thermodynamic equilibrium (Richard and Miller, 1990), the authors demonstrated that proton transport is involved in ClC-0 gating and is, in fact, the source of energy that keeps ClC channels in this asymmetric gating state (Lísal and Maduke, 2008). Further support for this position comes from a study in which a small but reproducible H⁺ transport demonstrated in ClC-1 channels was no longer identified in ClC-0 channels carrying the C212S mutation that abolish common gating (Picollo and Pusch, 2005).

There is evidence for a critical role for the Glu_ext residue in this mechanism (Dutzler et al., 2003; Cederholm et al., 2010), which would make this residue an essential part of both gating processes in ClC channels and also in the Cl^-/H⁺ ion translocation in ClC exchangers. Feng et al. (2010) proposed a hypothesis for the mechanism of coupled Cl^-/H⁺ transport, in which Glu_ext cycles between S_ext, S_cen, and the extracellular environment. While occupying S_cen, Glu_ext interacts with Tyr_cen and accepts a proton from the intracellular H⁺ pathway. Then, following a conformational change after protonation, it would deliver the H⁺ to the extracellular solution. Presupposing that common gating and Cl^-/H⁺ translocation are evolutionarily linked, and using the Cl^-/H⁺ transport mechanism described above as a model, Bennetts and Parker (2013) suggested that Glu_ext and Tyr_cen play an important role for ClC-0 and ClC-1 common gating as they do for Cl^-/H⁺ translocation. Additionally, they proposed that conformational changes for closure of the common gating involve helices G, F, H, I, and the CBS2 domain of the adjacent subunit, resulting in an arrangement that places Glu_ext (helix F) in position for hydrogen bonding with Tyr_cen (helix R), locking the channel closed. In this model, helix G would function as the coordinator between protopore and subunit interface, integrating both subunits for the common gating.

In the same work, the authors reported the involvement of Tyr_cen in Zn⁺² inhibition and NAD⁺ modulation of the common gate (Bennetts and Parker, 2013). This research sheds some light on the molecular determinants of the common gating of ClC channels, but much remains unclear. The pathway for the H⁺ transport—proposed to be involved in the common gating—is not yet defined, as the suggested intracellular coordinator (Glu_in) is changed by a valine residue in ClC channels. Also, in the eukaryotic CmClC Cl^-/H⁺ exchanger, Glu_in is replaced by a threonine residue that either perform this transport or this exchanger would use an alternative H⁺ pathway (Feng et al., 2010). A neighboring conserved Glu residue (E291 in ClC-1), however, was proposed as a substitute to execute this function (Lísal and Maduke, 2009). Mutation to a protonable aspartate (E291D) shifted voltage dependence to more positive values but preserved the pH dependence, whereas mutation to a neutral glutamine (E291Q) remarkably reduced voltage and pH dependence, suggesting the participation of this residue in the H⁺ transport (Lísal and Maduke, 2009). This assumption, however, cannot be confirmed based solely on mutagenesis experiments. The exact molecular rearrangement necessary for the common gating is another puzzle, with many parts still missing.

One recent study analyzing ClC-1/ClC-2 heterodimeric channels revealed channels with original gating properties. The common gating was abolished, with each subunit displaying individual slow gates as well as independent fast gates (Stölting et al., 2014a). These findings suggest that conformational changes underlying common gating mechanisms may originate within each protopore gate, and that fast and slow gating may in fact be linked mechanisms (Bennetts and Parker, 2013; Stölting et al., 2014a). Homodimeric channels are able to coordinate both slow gates, resulting in a single common gating, whereas heterodimeric channels lack this coordination and display individual slow gating for each subunit (Figure 1).

Skeletal muscle has a uniquely high resting Cl^- conductance that is more than four times greater than the K⁺ conductance (Bretag, 1987). ClC-1 is the predominant mediator of the Cl^- conductance in skeletal muscle. The first insight into the physiological role of ClC-1 came from studies using myotonic goats (Lipicky and Bryant, 1966) and myotonic adr mice (Steinmeyer et al., 1991a). Skeletal muscle fibers from these animals failed to repolarize following repeated action potentials, resulting in the so-called ‘myotonic after-discharge’ condition, characterized by muscle stiffness (Adrian and Bryant, 1974). After an action potential, Na⁺ channels close and K⁺ channels open to allow the ion efflux necessary for repolarization. In the T-tubules this K⁺ accumulation (increasing of [K⁺]_ext) may generate small depolarizations even after inputs from the nervous system have ceased. ClC-1 mediates Cl^- conductance that prevents the K⁺-mediated depolarization from propagating along the sarcolemma (Figure 2). In myotonic fibers, the lack of ClC-1 conductance leads to autonomous fiber action potentials that keep the muscle active, delaying relaxation (Steinmeyer et al., 1991a; Stölting et al., 2014b). Mutations in the ClC-1 gene were found in families with myotonia congenita. These mutations lead to partial or complete loss of function of ClC-1, affecting channel function in different ways depending on the mutation. A group of mutations cause a reverted voltage dependency, i.e., D136G (Fahlke et al., 1995), G499R (Zhang et al., 2000), C277Y (Weinberger et al., 2012), G523D (Ha et al., 2014). These mutations cause the channels to activate upon hyperpolarization rather than deactivate like wild-type ClC-1, rendering channels with dramatically reduced or abolished currents at physiological chloride gradients. Mutation G230E (Fahlke et al., 1997a) and the aforementioned C277Y alters the ion selectivity of the channel pore. The A531V has normal gating properties but has reduced expression at the plasma membrane due to an increased degradation rate (Lee et al., 2013). It was later shown by Chen et al. (2015) that a ubiquitin ligase complex (CUL4A/B-DDB1-CRBN) ubiquinates the A531V mutant leading to its subsequent degradation. To date, more than 130 mutations have been identified in the gene encoding ClC-1, and heterologous expression of mutated channels has played a valuable role in helping scientists to understand channel structure and function and disease pathogenesis (Matthews et al., 2010; Imbrici et al., 2015).

Myotonia congenita is the most common skeletal muscle hereditary channelopathy in humans, characterized by an atypical delay in muscle relaxation after voluntary contractions, called muscle stiffness. The myotonic stiffness is worse after rest, and improves after repetitive movements, referred as the warm-up phenomenon (Thomsen, 1876; Bryant, 1973). In humans, myotonia can be inherited as a dominant (Thomsen disease) or a recessive (Becker disease) trait, with more severe symptoms found in the latter form (Saviane et al., 1999). In the dominant form, mutant subunits exert a dominant negative effect on wild-type subunits; that is, the mutant impairs (at variable levels) the function of the wild-type subunit.

Using the crystal structure of cmClC (an eukaryotic ClC exchanger) (Feng et al., 2010) as a model for human ClC-1 allowed the identification of residues found mutated in myotonic patients in the dimer interface and in the ion conduction protopore (Skálová et al., 2013). Moreover, mutations causing the dominant-negative effect were located in or proximal to the dimer interface region. Meanwhile, mutations affecting the channel protopore do not exert the dominant-negative effect (Skálová et al., 2013). As the slow gating involves subunit interactions at the dimer interface, mutations affecting this area in only one subunit prevent the coordination necessary for the common gating and explain the impairment of the adjacent wild-type subunit. In the recessive form, both subunits are affected and ClC-1 currents may be abolished completely, leading to the more severe symptoms reported (Saviane et al., 1999; Imbrici et al., 2015).

To date, there is no specific treatment for patients with myotonia congenita. To surpass ClC-1 defect, the ideal drug should specifically enhance its Cl^- currents; unfortunately, this objective seems to be far from completion (Imbrici et al., 2015). One early study showed that the R-isomer of CPP, a clofibric acid derivative, was able to increase Cl^- conductance in voltage clamp recordings of muscle fibers (De Luca et al., 1992). This activity was not recognized in heterologously expressed channels, suggesting that the drug does not interact directly with ClC-1 and probably uses a muscle-specific component to exert its effect (Pusch et al., 2000). Acetazolamide (a carbonic anhydrase inhibitor) was also reported to be able to shift the voltage-dependence of ClC-1 channel opening to more negative voltages, possibly through changes in intracellular pH, consequently enhancing Cl^- conductance. However, this potentially anti-myotonic effect was not effective in some mutant channels (Eguchi et al., 2006; Desaphy et al., 2013).

In the testes, tight junctions between Sertoli cells isolate the adluminal compartment of the seminiferous tubules from the blood irrigation (blood–testis barrier). Because of this barrier, maturation and differentiation of spermatogonia into sperm cells require a close physical contact with Sertoli cells that are also responsible for the nourishment of the germ cells during this process. Disruption of ClC-2 function results in transepithelial transport defect in Sertoli cells and subsequent degeneration of male germ cells (azoospermia), as observed in ClC-2 knock-out (KO) mice (Bösl et al., 2001; Bi et al., 2013).

ClC-2 KO mice also develop leukodystrophy (the general term for diseases affecting the growth or maintenance of the white matter), which culminates with gradual development of vacuoles in the myelin sheath of the central nervous system, worsening with age (Blanz et al., 2007). Human patients carrying mutations that disrupt ClC-2 function develop similar leukodystrophy symptoms (Depienne et al., 2013); in one patient, azoospermia was found together with a subclinical leukodystrophy (Di Bella et al., 2014). This was the first case report demonstrating azoospermia and leukodystrophy in a patient with ClC-2 mutation.

ClC-2 is also expressed in epithelial cells of the gastrointestinal tract and lungs. In the past, ClC-2 was proposed to play a role in Cl^- efflux at the apical membrane of epithelial cells of these tissues, working as an alternative pathway to CFTR-dependent Cl^- secretion. However, the intestinal phenotype observed in CFTR-KO mice was not aggravated in double KO mice, in the absence of both CFTR and ClC-2. Instead, double KO mice survived better than CFTR-KO mice (Zdebik et al., 2004). Later on, it was demonstrated that ClC-2 localizes at the basolateral membrane of enterocytes, facilitating water and salt absorption (Figure 3) (Catalán et al., 2004). In the basolateral membrane, ClC-2 is proposed to move Cl^- in the opposite direction of CFTR, e.g., moving Cl^- from the cell to the interstitium. Loss of ClC-2 in CFTR-KO mice would then increase Cl^- concentration inside the cell, facilitating Cl^- efflux in the apical compartment by an alternative pathway and compensating for the loss of CFTR from the apical membrane. These and other reports (Catalán et al., 2004; Peña-Münzenmayer et al., 2005) provide convincing data for the basolateral localization of ClC-2 in intestinal epithelia. ClC-2 could play the same role in the lung epithelium, although its precise localization is still not conclusive.

ClC-2 is also expressed in neurons and glial cells, where it is proposed to lower the intracellular concentration of Cl^-. ClC-2 would be activated after a Cl^- influx mediated by hyperpolarizing GABA currents. ClC-2, then, would extrude the excess of intracellular Cl^- down to its electrochemical equilibrium helping in the maintenance of a Cl^- gradient favorable to cell hyperpolarization by GABA currents (Staley et al., 1996; Földy et al., 2010; Rinke et al., 2010). This theory, however, was questioned by a study using a computational model—based on ClC-2 parameters previously characterized in CA1 pyramidal cells—simulating physiological conditions which showed ClC-2 actually mediating chloride influx, directly reducing cell excitability (Ratté and Prescott, 2011).

The retinal pigment epithelia (RPE) are responsible for forming the blood–organ barrier in the eye, creating the optimal microenvironment for photoreceptor function. Loss of retinal photoreceptors induces retinal degeneration. Loss of ClC-2 function has been proposed to affect transepithelial transport in the RPE by disrupting microenvironment ion homeostasis, resulting in photoreceptor degeneration (Bösl et al., 2001; Bi et al., 2013). Studies on ClC-2 KO mice revealed retinal degeneration, indicating an important role for this channel in RPE. This degenerative phenotype suggests the disruption of ion homeostasis in this tissue (Bösl et al., 2001).

Previously, several other functions were thought to be assigned to ClC-2. Suggested roles in gastric acid secretion (Sherry et al., 2001) and lung development (Murray et al., 1995) were supported neither by experimental data nor by ClC-2 KO mice phenotype. A role in epilepsy was also considered, but after the retraction of a widely cited paper correlating ClC-2 mutations to idiopathic generalized epilepsy, there is no credible evidence for a ClC-2 role in human epilepsy. This is consistent with the lack of seizures observed in ClC-2 KO mice (Bösl et al., 2001; Blanz et al., 2007).

ClC-Kb/barttin is mainly expressed in basolateral membranes of the thick ascending limb of Henle’s loop, where it is involved in the reabsorption of salt and, consequently, water (Uchida et al., 1995). In this part of the nephron, the Na⁺ electrochemical gradient (created by basolateral Na⁺/K⁺ pump) drives the secondary active transport of NKCC2 (present in the apical membrane), accumulating Na⁺, Cl^-, and K⁺ into the cell. K⁺ is extruded back to the lumen through ROMK K⁺ channels (also present in the apical membrane), whereas Na⁺ and Cl^- are reabsorbed by the interstitial fluid through the Na⁺/K⁺ pump and ClC-Kb channels, respectively. Thus, the end product of this system is the reabsorption of NaCl into the blood stream (Figure 4A).

In the inner ear, both ClC-K isomers are expressed in the basolateral membrane of marginal cells of the stria vascularis. This multilayered epithelium is responsible for both the high concentration of K⁺ and the positive potential (about 100 mV higher than normal extracellular fluids) of the endolymph of the scala media, both of which are important properties for hearing. In marginal cells—the more apical layer in the stria vascularis—Na⁺/K⁺ pumps and NKCC1 transporters build up K⁺ and Cl^- inside the cells. ClC-K/barttin channels recycle Cl^- back to the interstitial fluid, while apical KCNQ1/KCNE1 K⁺ channels secrete the excess of potassium ions into the endolymph (Figure 4B) (Rickheit et al., 2008).

In agreement with the transport models involving ClC-K/barttin channels, mutations in the gene encoding ClC-Kb cause salt-losing Bartter syndrome type III (Simon et al., 1997), characterized by hypokalemia, metabolic alkalosis and secondary hyperaldosteronism with normal or low blood pressure (Andrini et al., 2015). Mutations in the gene encoding barttin cause Bartter syndrome type IV that combines the salt waste with congenital deafness, since both ClC-K proteins are non-functional in the absence of barttin (Birkenhäger et al., 2001). When disruption occurs in only one of the ClC-K channels, as it does in ClC-Kb mutations in Bartter type III, hearing is preserved; the other isomer channel still provides the necessary Cl^- recycling. Deafness occurs only on disruption of both ClC-K channels or upon disruption of barttin (Birkenhäger et al., 2001; Schlingmann et al., 2004).

Although there are no reports of patients presenting mutations only in ClC-Ka, two patients presenting symptoms similar to those accompanying Bartter syndrome type IV—severe renal salt loss and sensorineural deafness—were described with loss-of-function mutations in both ClC-K isomers instead of barttin (Schlingmann et al., 2004; Nozu et al., 2008).

Polymorphisms in ClC-Ka and -Kb genes were described, and their relationship with cardiovascular diseases was analyzed. ClC-Kb gene polymorphism T481S increases currents in heterologously expressed channels by approximately 20-fold (Jeck et al., 2004). This may lead to increased salt reabsorption in the thick ascending limb of Henle’s loop, suggesting a possible connection with hypertension. However, several cohort studies found discrepant results, and a link between this activating polymorphism and hypertension is still lacking (Jeck et al., 2004; Speirs et al., 2005; Fava et al., 2007; Sile et al., 2009). One frequent polymorphism in the ClC-Ka gene (R83G) was linked to heart failure. R83G was reported to reduce ClC-Ka currents by about 50%, and was statistically associated with heart failure in three independent Caucasian cohorts (Cappola et al., 2011). However, a functional link between this half-loss-of-function polymorphism and heart failure is still not established.

ClC-K/barttin channels are promising candidates for therapeutic drugs. As ClC-Kb is involved in salt and water reabsorption in the thick ascending limb of Henle’s loop, drugs blocking ClC-Kb/barttin function could reduce renal salt and water reabsorption, which would decrease blood volume and consequently reduce blood pressure. In the inner ear, hearing depends on the depolarization of mechanosensitive hair cells. Different from other excitable cells that use Na⁺ currents for depolarization, depolarization of hair cells is mediated by K⁺ influx.

Drugs capable of increasing ClC-K/barttin function in the stria vascularis would increase endolymph K⁺ concentration, and therefore could be used to treat hearing disorders. However, due to expression of ClC-K/barttin channels in both the kidney and inner ear, it will be difficult to develop specific drugs without undesirable side effects. Recently, while testing new benzofuran derivatives designed to block ClC-K function, Liantonio et al. (2016) described the most potent and selective ClC-K blocker discovered to date (SRA-36). This compound is able to inhibit not only wild-type channels, but also the Cl^- currents of polymorphic ClC-K channels associated with hypertension (Liantonio et al., 2016). Although several studies have made significant progress on the identification of compounds modulating ClC-K channel function (Liantonio et al., 2004, 2006, 2016; Picollo et al., 2004), there are not yet therapeutic drugs available.

Dent’s disease is a rare X-linked kidney disorder associated with low molecular weight proteinuria, hyperphosphaturia, hypercalciuria, kidney stones, and nephrocalcinosis (Wrong et al., 1994). After the identification of ClC-5 mutations as the cause of Dent’s disease (Lloyd et al., 1997), more than 100 such mutations were described (Pusch and Zifarelli, 2015). Most mutations in ClC-5 are missense and non-sense mutations, with many of them located at or near the subunit’s interface, resulting in non-functional truncated proteins (Wu et al., 2003; Stauber et al., 2012). Two missense mutations (G212A and E267A) were analyzed regarding their functional consequences. The particular interest in these mutations is explained by their close proximity to Glu_ext (E211) and Glu_int (E268). Both mutations result in impaired endosomal acidification, however, the causes are distinct. For the G212A mutant, a shift to more depolarizing potentials is the cause of reduced transport, whereas in E267A mutant the inability to transport intracellular protons results in an incomplete transport cycle (Alekov, 2015).

Proximal tubule cells are the main site for re-uptake of low molecular weight proteins from the primary urine filtrated at the glomeruli. The co-localization of ClC-5 and H⁺-ATPase in PTCs, and the loss of low molecular weight proteins in Dent’s disease patients, suggests that ClC-5 might be involved in early tubular endocytosis in nephrons. To better investigate this hypothesis, two ClC-5 KO mice models were independently generated. Both models showed loss of low molecular weight proteins and high levels of retinal- and Vitamin D-binding proteins in the urine—proteins also elevated in the urine of Dent’s disease patients (Wrong et al., 1994), and defective endocytosis in the proximal tubule (Piwon et al., 2000; Wang et al., 2000). Interestingly, only one of the mouse models displayed hypercalciuria and interstitial calcium accumulation (Wang et al., 2000). In vivo endocytosis experiments showed that both fluid-phase and receptor-mediated endocytosis were severely reduced, and that the apical expression of NHE3 (Na⁺-H⁺ exchanger) and NaPi-2a (coupled Na⁺-Pi co-transporter) were also reduced, all in a cell-autonomous effect of ClC-5 disruption (Piwon et al., 2000).

Moreover, the protein megalin—an endocytotic receptor responsible for the endocytosis of several proteins and other substances—and its co-receptor cubilin were also decreased in the brush border membrane of PTCs from ClC-5 KO mice (Piwon et al., 2000; Christensen et al., 2003). Although almost all Dent’s disease patients present low molecular weight proteinuria, with values ranging from 100- to over 1000-fold of the normal excretion values, the other clinical features show great variability (Scheinman, 1998; Claverie-Martín et al., 2011).

ClC proteins are involved in the acidification of both early and late endosomes. Endosomal acidification is a process mediated by active proton influx carried by the H⁺-ATPase. The inward H⁺ movement requires a charge balance, which can be achieved both by outward movement of cations such as K⁺, and by inward movement of anions such as Cl^-. Extensive experimental data suggest that Cl^- is the principal ion providing the electrical shunt for luminal acidification of endosomes (Bae and Verkman, 1990; Al-Awqati, 1995; Grabe and Oster, 2001). After ClC-5 was determined to be a Cl^-/H⁺ exchanger and not a Cl^- channel (Picollo and Pusch, 2005; Scheel et al., 2005), its role as an electrical shunt for proton pumps was questioned; such an exchanger would provide a counter-current of H⁺, opposing the ATP-driving accumulation of protons.

To assess the consequences of this new CIC-5 feature, knock-in mice were generated carrying a point mutation in the ‘gating glutamate’ (E211A) that uncouples Cl^- and H⁺ transport, converting ClC-5 to a pure passive Cl^- conductor (called ClC-5^unc; for uncoupled) (Novarino et al., 2010). Surprisingly, these mice presented normal renal endosomal acidification, but also an impaired proximal tubular endocytosis similar to that found in ClC-5 KO mice. Two facts suggest that other parameters such as Cl^- concentration may play a critical role in endocytosis. First, ClC-5^unc mice presented phenotypes similar to the ClC-5 KO group, including hypercalciuria and hyperphosphaturia. Second, PTC endosomes from ClC-5^unc showed normal acidification but impaired endocytosis. Recently, a patient with Dent’s disease was identified as carrying a similar mutation in the ‘gating glutamate’ (E211Q) (Sekine et al., 2014). Further support for the role of Cl^- concentration in endocytosis comes from mathematical models of simplified vesicles (containing a proton pump, a proton leak and either a Cl^-/H⁺ exchanger or a Cl^- channel) predicting that coupled transport would provide a higher endosomal Cl^- concentration than a pure Cl^- current (Weinert et al., 2010). Unfortunately, the exact role of ClC-5 coupled Cl^-/H⁺ transport in early endosomes is still not fully understood.

To study the physiological roles of ClC-7/Ostm1, knock-out mouse models were generated and analyzed. ClC-7 KO mice present short life spans, severe osteopetrosis, retinal degeneration, lysosomal storage disease, and neurodegeneration (Kornak et al., 2001; Kasper et al., 2005). Gray-lethal mice (Ostm1 KO) display a very similar phenotype (Chalhoub et al., 2003; Lange et al., 2006), as expected for these two closely functionally related proteins. Interestingly, both ClC-7 KO and Ostm1 KO mice have gray fur in an agouti background (in which wild-type mice have brown fur), suggesting a possible role of ClC-7/Ostm1 in melanosomes (Kornak et al., 2001).

Loss of function of ClC-7 in osteoclasts results in osteopetrosis, a disease characterized by increased bone radiodensity because of ineffective osteoclast-mediated bone resorption (Shapiro, 1993). The ruffled border of osteoclasts—a membrane domain responsible for acidic digestion of bone tissue—is formed by lysosomal membrane insertion and exocytosis of their content. Acidification of the resorption lacuna—the space between the ruffled border and the bone tissue—is carried by V-type H⁺-ATPase that, similarly to compartments of the endosomal/lysosomal pathway, requires an electrical shunt thought to be performed by ClC-7/Ostm1 (Planells-Cases and Jentsch, 2009; Stauber et al., 2012). In the resorption lacuna, ClC-7/Ostm1 is responsible for the Cl^- influx necessary for neutralization (shunting) of protons, which keeps the proton pump functional and maintains the lacuna’s highly acidic environment (Figure 7). Indeed, osteoclasts from ClC-7 KO mice displayed underdeveloped ruffled borders and impaired acidification of the resorption lacuna, which causes the osteopetrotic phenotype (Kornak et al., 2001).

After establishment of the link between ClC-7/Ostm1 mutations and osteopetrosis, approximately 50 different human ClC-7 mutations were identified in osteopetrotic patients (for details, see Stauber et al., 2012). Most of these are missense mutations, some of which cause an autosomal dominant form of osteopetrosis that presents less severe symptoms and does not implicate the nervous system (Cleiren et al., 2001; Frattini et al., 2003). Others yield ClC-7/Ostm1 complexes, which are retained in the ER or strongly reduce their ion transport activity (Leisle et al., 2011). However, other missense mutations produce transporters that reach the normal subcellular localization, remaining functional when expressed heterologously in the plasma membrane, and carrying the sorting motif point mutation mentioned above. Curiously, these functional mutants yielded currents with accelerated kinetics of activation and deactivation. Given that patients carrying these mutations present symptoms similar to others with non-functional transporters, the slow gating of ClC-7/Ostm1 is likely physiologically important (Leisle et al., 2011).

In addition to osteopetrosis, ClC-7 KO and gray-lethal mice also display retinal and neurodegeneration associated with lysosomal storage (Kornak et al., 2001; Kasper et al., 2005; Lange et al., 2006). Retinal degeneration leads to blindness about 4 weeks after birth (Kornak et al., 2001; Lange et al., 2006). Although previously the blindness was believed to be the result of osteopetrotic narrowing of the optic nerve canal (Steward, 2003), retinal degeneration is a direct effect of disruption of ClC-7 or Ostm1 at retinal neurons (Kasper et al., 2005). Neurodegeneration is the probable cause of death of ClC-7 KO mice at approximately 6 weeks. They present neuronal cell loss, particularly in the hippocampus and cerebral cortex, as well as lysosomal storage material scattered throughout the neuronal somata (Kasper et al., 2005). Neurodegeneration and lysosomal storage disease are cell-autonomous effects of disruption of ClC-7 as demonstrated by tissue-specific ClC-7 KO mice. In this study, PTCs and neurons lacking or expressing ClC-7 were compared within the same environment; only cells devoid of ClC-7 displayed lysosomal disease (Wartosch et al., 2009).

In correlation with ClC-7 function in osteoclasts and ClC-3 to ClC-5 roles in acidification of their respective compartments (Mohammad-Panah et al., 2003; Hara-Chikuma et al., 2005a,b), the lysosomal storage phenotype was first proposed to result from impaired acidification. However, ratiometric measurements showed that ClC-7/Ostm1 does not play a role in lysosomal acidification; that is, lysosomes from ClC-7 KO mice display normal steady-state pH (Kasper et al., 2005; Lange et al., 2006; Weinert et al., 2010). The counter-charge conductance suggested to neutralize H⁺-ATPase currents in this case seems to be provided by a lysosomal cation efflux (Steinberg et al., 2010).

These intriguing results concerning the function of ClC-7 in lysosomes led to the generation of two other ClC-7 mice models, the ClC-7^unc mice (Weinert et al., 2010)—generated similarly to the ClC-5^unc mice, by a ‘gating glutamate’ mutation that yields uncoupled Cl^- pure conductance—and the ClC-7^td mice (Weinert et al., 2014), generated by a mutation in the ‘proton glutamate’ that abolishes transport (td: transport deficient). Much like ClC-7 KO mice, both new models displayed normal lysosomal pH and reduced Cl^- concentration. However, although ClC-7^unc mice presented neuronal cell loss and lysosomal storage disease similar to the ClC-KO mice, the osteopetrotic phenotype was partially rescued by the presence of pure Cl^- currents (Weinert et al., 2010). Meanwhile, ClC-7^td mice displayed similar osteopetrotic phenotype to ClC-7 KO mice, but a delayed and less severe neurodegeneration (Weinert et al., 2014).

Surprisingly, unlike ClC-7 KO and Ostm1 KO mice, the two new models presented normal agouti brown fur color (Weinert et al., 2014). Taken together, these results suggest that the physical presence of the non-functioning ClC-7^td alone was sufficient to rescue the normal fur pigmentation and alleviate neurodegeneration, whereas ClC-7^unc Cl^- currents had a positive effect in partially rescued osteopetrosis and normal fur pigmentation, but a negative effect in neurodegeneration (Weinert et al., 2014). Therefore, normal fur pigmentation seems to require only the presence of ClC-7/Ostm1 complex, whereas lysosomes and osteoclasts require fully functional Cl^-/H⁺ exchangers in order to function properly. The precise role of ClC-7 in lysosomes is still obscure, as is the previously discussed role of ClC-5 in endosomes. More work is necessary to better understand the role of coupled Cl^-/H⁺ transport in the endosomal/lysosomal pathway.

ClC-3 is a broadly expressed intracellular ClC protein with controversial biophysical and physiological characteristics. Several mutually conflicting Cl^- currents have been attributed to ClC-3. First, a slightly outwardly rectifying Cl^- current inhibited by PKC (Kawasaki et al., 1994); second, Cl^- currents inhibited by intracellular Ca⁺² (Kawasaki et al., 1995), followed by swelling-activated Cl^- currents, also known as VRACs (Duan et al., 1997). The role of VRACs was not confirmed by data from three independently generated ClC-3 KO mice (Stobrawa et al., 2001; Dickerson et al., 2002; Yoshikawa et al., 2002), as swelling-activated Cl^- currents were unaffected in the hepatocytes, salivary acinar cells and cardiomyocytes (Arreola et al., 2002; Gong et al., 2004). Moreover, a recent report identified LRRC8 proteins as essential components of VRACs, as their disruption abolishes VRAC currents (Voss et al., 2014).

ClC-3 presents very low cell surface expression, even when heterologously overexpressed, which has hindered a thorough analysis of its biophysical properties. Nevertheless, in some studies, small but strongly outwardly rectifying Cl^- currents were found (Li et al., 2002; Matsuda et al., 2008; Guzman et al., 2013), similar to those reported for ClC-4 and ClC-5 (Steinmeyer et al., 1995; Friedrich et al., 1999). Moreover, because of its substantial sequence identity—approximately 80% with ClC-4 and ClC-5, which are well established ClC exchangers—and the presence of the conserved ‘gating glutamate,’ which mutation abolish rectification as in other ClCs, ClC-3 is most likely a voltage dependent Cl^-/H⁺ exchanger (Li et al., 2002; Guzman et al., 2013).

ClC-3 is expressed in most tissues, including the brain, retina, adrenal gland, pancreas, intestines, epididymis, kidney, liver, skeletal muscle, and heart (Kawasaki et al., 1994; Stobrawa et al., 2001; Suzuki et al., 2006; Maritzen et al., 2008). It mainly resides in endosomes, where it was found co-localized with ClC-4 and ClC-5, and also with both early and late endosomal markers (Figure 5) (Suzuki et al., 2006). CLC-3 is also found in synaptic vesicles (Stobrawa et al., 2001; Salazar et al., 2004) and synaptic-like micro vesicles (Salazar et al., 2004; Maritzen et al., 2008). Three different ClC-3 KO mouse lines displayed similar phenotypes of severe degeneration of the retina and brain, with prominent effects in the hippocampus (Stobrawa et al., 2001; Dickerson et al., 2002; Yoshikawa et al., 2002). In one model, signs of lysosomal storage disease were observed (Yoshikawa et al., 2002), but these effects were much weaker than those found in ClC-6 and ClC-7 KO mice (Poët et al., 2006; Kasper et al., 2005). The mechanism by which ClC-3 causes neurodegeneration is still unclear.

Although ClC-3 was thought to provide the electrical shunt for acidification of intracellular compartments like the other ClC exchangers, its role in endosomes and synaptic vesicles is still controversial. In ClC-3 KO mice, acidification and Cl^- accumulation were reduced in early and late endosomes (Hara-Chikuma et al., 2005b) and synaptic vesicles showed less efficient acidification in vitro (Stobrawa et al., 2001). Synaptic vesicles from ClC-3 KO mice exhibit reduced glutamate uptake, but this feature has been ascribed to diminished levels of the vesicular glutamate transporter VGLUT1 (Stobrawa et al., 2001). In another study, evidence against ClC-3’s role in acidifying synaptic vesicles was reported (Schenck et al., 2009). The fact that VGLUT1 KO mice display no chloride-dependent acidification of synaptic vesicles, and very little expression of CLC-3 in synapse vesicles, led the authors to suggest that VGLUT1 represents the major Cl^- conductance pathway in synaptic vesicles (Schenck et al., 2009). Therefore, the reduced acidification of synaptic vesicle reported in ClC-3 KO mice (Stobrawa et al., 2001) was attributed to reduced levels of VGLUT1, which itself most likely resulted from severe neurodegeneration.

To avoid the effects of neurodegeneration on VGLUT1 levels, Guzman et al. (2014) used cultured hippocampal neurons to analyze ClC-3 disruption in synaptic vesicles. The authors observed enlargement of synaptic vesicles and increased glutamate content in cells lacking ClC-3. The probability of vesicle fusion and release of its content was also increased in those cells, indicating that exaggerated release of glutamate in the synaptic cleft contributes to neurodegeneration in ClC-3 KO mice (Guzman et al., 2014).

Recently, reduced levels of ClC-3 were found in patients with inflammatory bowel disease (IBD) and from mice treated with dextran sulfate sodium (DSS) to induce colitis and mimic IBD. ClC-3 KO mice were more susceptible to DSS-induced colitis with no signs of recovery after treatment. Lack of ClC-3 provoked apoptosis of intestinal epithelial cells, causing disruption of the epithelial barrier and bacterial invasion. Thus, the authors defend the involvement of ClC-3 in IBDs pathogenesis (Huang et al., 2014). Another study points to ClC-3’s involvement atherosclerosis. In an ApoE null mice background, further disruption of ClC-3 reduced the size of atherosclerotic lesions present in the aorta. The authors suggest that ClC-3 insufficiency disrupts scavenger receptor SR-A expression (via JNK/p38 MAPK) and foam cell formation, leading to reduction/inhibition of atherosclerotic lesions (Tao et al., 2015). The protective effect of ClC-3 deficiency was again addressed in endothelial progenitor cells (EPCs). Angiotensin II-induced apoptosis of EPCs was remarkably reduced in ClC-3 KO mice. This inhibition was attributed to suppressed levels of reactive oxygen species and NADPH oxidase activity—direct effects of angiotensin-II—that are suppressed by ClC-3 disruption (Liu et al., 2013).

Despite ClC-3 diversity in different cell types, phenotypes present in ClC-3 KO mice strongly suggest that ClC-3 is important for neurotransmission in the CNS. Three different splicing variants of ClC-3 (ClC-3a, ClC-3b, and ClC-3c) were described in the brain with different subcellular localization but similar transport function (Guzman et al., 2015). Another splicing variant, ClC-3d, was described in mouse livers as displaying different localizations but identical transport properties (Okada et al., 2014). The number of splicing variants with different subcellular localizations might explain the diversity of functions ascribed to ClC-3; the study of these isoforms could be a promising direction for further study of the precise function and localization of ClC-3 proteins, which could provide an explanation for the phenotypes described in KO mice.

ClC-4 is a broadly expressed ClC exchanger found in various tissues which differ between species. It is found mostly in the muscles, brain, and heart of humans (van Slegtenhorst et al., 1994); in the liver and brain, heart, muscles, spleen, and kidneys of rats (Jentsch et al., 1995); and in the brain, intestines, and kidneys of mice (Mohammad-Panah et al., 2003). Interestingly, the gene encoding ClC-4 is localized on chromosome 7 in inbred laboratory mice, but in humans and rats, the gene resides on the X chromosome (Rugarli et al., 1995). This may partially explain the variety and species-specificity of expression patterns.

ClC-4 localizes mainly at endosomes’ membranes; upon heterologous overexpression, a small portion is also found within the plasma membrane (Mohammad-Panah et al., 2003; Suzuki et al., 2006). Like other members of this family, the cell surface localization allows for a better analysis of its biophysical properties. ClC-4 yields a strongly outwardly rectifying current that is inhibited by low extracellular pH; to date, the physiological relevance of pH regulation remains unclear. ClC-4 is a Cl^-/H⁺ exchanger with anion conductance sequence of Cl^- > Br^- > I^-, similar to other intracellular ClCs. Mutation of the ‘gating glutamate’ strongly changes rectification and converts the exchanger into a passive Cl^- conductor, highlighting the importance of this residue in coupling proton to Cl^- transport (Friedrich et al., 1999; Picollo and Pusch, 2005; Scheel et al., 2005).

Alekov and Fahlke (2009) have shown that ClC-4 proteins, when exposed to different types of anions in the extracellular buffer, can display a phenomenon called ‘slippage,’ where the transporter behaves as a channel rather than an obligatory exchanger. High extracellular SCN^- cause increased current amplitudes and uncouple H⁺ transport rendering ClC-4 channel-like transport with reduced H⁺ currents and biophysical properties similar to other well-characterized ClC channels. Restoring extracellular Cl^- rescues Cl^-/H⁺ exchange. In ClC-5, extracellular SCN^- uncouples transport but does not affect proton transport (Grieschat and Alekov, 2012), an apparent isoform-specific effect. Thus, ClC-4 is suggested to function as a channel or exchanger depending on the extracellular anion (Alekov and Fahlke, 2009).

Currently, there is little scientific consensus regarding the sub-cellular localization of ClC-4 (Figure 5). On those studies, ClC-4 was found in sub-apical vesicles of proximal tubule epithelium (Mohammad-Panah et al., 2003); in intracellular compartments of HEK 293 cells co-localizing with ClC-3 and ClC-5 (Suzuki et al., 2006); and in the endoplasmic reticulum (Okkenhaug et al., 2006). However, none of the immunohistochemistry studies performed thus far have used cells from ClC-4 KO mice as a negative control to confirm their data.

ClC-4 was suggested to facilitate endosomal acidification by working as the electrical shunt for proton accumulation mediated by the proton pump. However, ClC-4 KO mice do not display any obvious abnormal phenotypes (Rickheit et al., 2010). Although ClC-4 trafficking is similar to ClC-5, they do not appear to perform similar physiological functions (Mohammad-Panah et al., 2003). The additional disruption of ClC-4 in ClC-5 KO mice did not aggravate the impaired endocytosis phenotype in PTCs (Rickheit et al., 2010).

One naturally occurring mutation (G544R), found in a patient with severe epilepsy and delayed development, nearly abolished ClC-4 currents when expressed heterologously (Veeramah et al., 2013). Hu et al. (2015), analyzing X-linked intellectual disabilities, identified five different mutations in the ClC-4 gene in five families. Currents of ClC-4 proteins carrying each of these mutations were much smaller or even absent compared to wild-type ClC-4. Moreover, C1C-4 depletion in cultured hippocampal neurons, affected neuronal differentiation; the cells displayed a 30% reduction of neuritic outgrowth and branching (Hu et al., 2015). Additional studies using specific antibodies and appropriate KO controls are necessary to further understand ClC-4 physiological function in specific cell compartments, determine precise sub-cellular localization, and investigate possible roles in human diseases.

ClC-6 shares approximately 45% of its sequence identity with ClC-7; together, they form the third branch of the ClC protein family. Like other ClC exchangers, ClC-6 localizes at membranes of the endosomal/lysosomal pathway (Figure 5) (Poët et al., 2006). ClC-6 mRNA was found in several tissues (Brandt and Jentsch, 1995), but the expressed ClC-6 protein is found almost exclusively in the nervous system (Poët et al., 2006). First attempts to record ClC-6 currents by heterologous expression were frustrated by its late endosomal localization (Brandt and Jentsch, 1995; Buyse et al., 1998), and biophysical characterization only became possible when GFP-tagged ClC-6 proteins were expressed in the plasma membrane (Neagoe et al., 2010). ClC-6 mediates outwardly rectifying currents that are reduced by extracellular acidification, as in other ClC-exchangers. Mutation in the ‘gating glutamate’ also disrupts rectification, turning ClC-6 into a passive Cl^- conduit (Neagoe et al., 2010).

Knock-out controlled immunohistochemistry studies have shown that native ClC-6 localizes predominantly at late endosomes of neurons in situ (Poët et al., 2006) and in cultured cells (Ignoul et al., 2007), whereas in heterologous expression it is also found co-localized with early and late endosomal markers (Suzuki et al., 2006; Stauber and Jentsch, 2010).

ClC-6 KO mice present no apparent abnormal phenotypes, with normal life span and weight. However, late in life (>3 months old), the mice display a peculiar form of lysosomal storage disease, with deposits found in central and peripheral neurons (Poët et al., 2006). Different from ClC-7 KO mice, in which such deposits are localized all over the neuronal soma and the disease progression is much more aggressive, deposits in ClC-6 KO neurons are mainly localized at initial axon segments and the disease progresses very slowly (Poët et al., 2006; Pressey et al., 2010). Moreover, the absence of ClC-6 in hippocampal neurons does not affect lysosomal steady-state pH (Poët et al., 2006). Deposits found in ClC-6 KO mice tested positive for markers typically found in neuronal ceroid lipofuscinosis (NCL), a lysosomal storage disease. The authors therefore proposed ClC-6 gene as a candidate for mild forms of NCL, but did not find convincing association upon analysis of 75 NCL patients (Poët et al., 2006).

In general, neuropathology in ClC-6 KO mice is much milder than in ClC-3 and ClC-7 KO mice. They show no vision impairment, and little neuronal cell loss and microglial activation (Poët et al., 2006; Pressey et al., 2010). ClC-6 KO mice also demonstrate reduced pain sensitivity, correlated with an impairment of dorsal root ganglion neuronal function due to dramatic lysosomal storage accumulation (Poët et al., 2006). After all, like ClC-3 and ClC-4, ClC-6 is another ClC exchanger whose physiological role is poorly understood at present.