Regulatory–auxiliary subunits of CLC chloride channel–transport proteins

Abstract

The CLC family of chloride channels and transporters is composed by nine members, but only three of them, ClC-Ka/b, ClC-7 and ClC-2, have been found so far associated with auxiliary subunits. These CLC regulatory subunits are small proteins that present few common characteristics among them, both structurally and functionally, and their effects on the corresponding CLC protein are different. Barttin, a protein with two transmembrane domains, is essential for the membrane localization of ClC-K proteins and their activity in the kidney and inner ear. Ostm1 is a protein with a single transmembrane domain and a highly glycosylated N-terminus. Unlike the other two CLC auxiliary subunits, Ostm1 shows a reciprocal relationship with ClC-7 for their stability. The subcellular localization of Ostm1 depends on ClC-7 and not the other way around. ClC-2 is active on its own, but GlialCAM, a transmembrane cell adhesion molecule with two extracellular immunoglobulin (Ig)-like domains, regulates its subcellular localization and activity in glial cells. The common theme for these three proteins is their requirement for a proper homeostasis, since their malfunction leads to distinct diseases. We will review here their properties and their role in normal chloride physiology and the pathological consequences of their improper function.

Introduction

Chloride is important for many biological functions, such as transepithelial fluid transport, acidification of intracellular organelles, muscle contraction, neuronal membrane potential or cell volume regulation. Chloride flux across membranes is mediated by several classes of proteins (Duran et al. 2010), among them, the CLC family of chloride channels and transporters. The founding member of the family, ClC-0, was identified in the electric organ of the marine ray (Jentsch et al. 1990) and subsequently several members of the family were found in organisms ranging from bacteria to plants and mammals (Stauber et al. 2012).

The first two mammalian CLC channels cloned, ClC-1 and ClC-2, gave rise to substantial Cl⁻ currents when expressed in Xenopus oocytes or in transfected cells (Steinmeyer et al. 1991; Thiemann et al. 1992; Pusch et al. 1994). Two CLC channels were found specifically in the mammalian kidney, ClC-Ka/b in humans and ClC-K1/2 in rodents. Despite an almost 50% sequence identity with ClC-1 or ClC-2, these channels could not be functionally expressed in these systems; only the rat ClC-K1 channel showed some activity, albeit very small currents (Uchida et al. 1993; Adachi et al. 1994; Kieferle et al. 1994). Even in the absence of detectable activity, a relevant physiological role for human ClC-Kb became evident as mutations in the CLCNKB gene lead to classical Bartter syndrome (type III, MIM no. 607364), a condition characterized by renal salt wasting (Simon et al. 1997). The lack of functional expression immediately raised the hypothesis that an additional subunit was required for ClC-K function (Kieferle et al. 1994). The existence of auxiliary subunits was already known for other ion channels such as the voltage-gated Na⁺, K⁺ or Ca²⁺ channels. These subunits do not form the pore of the channel, but are able to regulate their intracellular trafficking and modify their activities (Pongs & Schwarz, 2010; Brackenbury & Isom, 2011; Dolphin, 2012). The missing subunit for ClC-K was discovered in patients with a severe form of antenatal Bartter syndrome associated with sensorineural deafness and renal failure (BSND, Bartter syndrome type IVa, MIM no. 602522) that did not present mutations in the genes previously described as being responsible for Bartter syndrome (Birkenhager et al. 2001). The protein coded by the BSND gene was called Barttin, and it is able to interact with both ClC-K isoforms (Estevez et al. 2001). It is a 320 amino acid membrane protein with two transmembrane segments, a short cytoplasmic N-terminus and a relatively large cytoplasmic C-terminus (Fig.1A). Barttin shows no homology to other human proteins and is exclusively found in vertebrates.

Figure 1.

Proposed topologies for CLC regulatory subunits in the membrane

A, Barttin. B, Ostm1. C, GlialCAM. The approximate position of some pathological mutations is indicated. For GlialCAM, dominant mutations are underlined.

The functional expression in Xenopus oocytes and transfected cells of the remaining CLC proteins identified was puzzling, and it was speculated that some of these proteins may require additional subunits that transform their biophysical properties and contribute to their physiological functions. ClC-3 to ClC-5 gave rise to very outwardly rectifying currents (Steinmeyer et al. 1995) and no currents were detected for ClC-6 and ClC-7 (Brandt & Jentsch, 1995). This lack of functional expression of the latter two proteins was explained by the discovery that they are mostly localized in intracellular compartments (Kornak et al. 2001; Neagoe et al. 2010; Leisle et al. 2011). Additionally, the intracellular CLC proteins were found to be proton–chloride exchangers instead of true channels (Accardi & Miller, 2004; Picollo & Pusch, 2005; Scheel et al. 2005). Analysis of knock-out mouse models combined with the identification of mutations in human patients helped to reveal the physiological role of ClC-7 in osteoclasts for bone remodelling (Lloyd et al. 1996; Piwon et al. 2000; Kornak et al. 2001). As was observed with the ClC-K–Barttin pairing, the similarities in the clinical phenotypes of patients and mice deficient in ClC-7 and the protein Ostm1 suggested that Ostm1 might work as a subunit of ClC-7 (Chalhoub et al. 2003; Lange et al. 2006). Ostm1 is a membrane protein that presents a single transmembrane domain, with a short cytoplasmic tail and a longer luminal domain (Fig.1B).

GlialCAM, the last CLC subunit identified to date, was not identified as such by phenotype comparison. Its partner ClC-2 had actually been ruled out as a candidate gene for the disease involving GlialCAM (Scheper et al. 2010). The ClC-2 channel is almost ubiquitously expressed (Thiemann et al. 1992), but the phenotype of ClC-2-depleted mice suggested a crucial role for this channel in transepithelial fluxes in retina and testes and in the process of potassium siphoning by glial cells in the brain (Bosl et al. 2001; Blanz et al. 2007). The identification of the subunit associated with ClC-2 occurred in a reverse manner: in an unbiased proteomic study aimed at identifying GlialCAM interacting proteins (Lopez-Hernandez et al. 2011a). GLIALCAM is the second gene involved in megalencephalic leukoencephalopathy with subcortical cysts (MLC), a rare type of leukodystrophy characterized by early-onset megalencephaly and white matter oedema and late-onset neurological deterioration (van der Knaap et al. 2012). MLC is mainly caused by mutations in MLC1, a gene that codes for a membrane protein of unknown function (Leegwater et al. 2001). GlialCAM is a cell-adhesion molecule with two Ig-like extracellular domains, a single transmembrane domain and a cytoplasmic tail (Fig.1C), and binds to both MLC1 and ClC-2 (Lopez-Hernandez et al. 2011a; Jeworutzki et al. 2012). As mentioned earlier, although Mlc1, GlialCAM and Clcn2 knock-out mice show similar phenotypes in the central nervous system (Hoegg-Beiler et al. 2014), patients with mutations in CLCN2 present different features from MLC patients (Depienne et al. 2013).

In this review we describe in detail these three subunits of CLC channels and transporters, and how they modify the functional properties of the respective CLCs. A table comparing the three subunits summarizes the known information about these proteins (Table1). We will emphasize how the discovery of these regulatory proteins has been useful to identify the physiological roles of CLC proteins in health and disease.

Table 1.

Comparison of CLC family regulatory–auxiliary subunits

	Barttin	Ostm1	GlialCAM
Endoplasmatic reticulum (ER) exit	Needed for ER exit	Needed for ER exit	Not needed for ER exit
Subcellular localization	Plasma membrane targeting. Basolateral localization?	Not necessary for lysosomal targeting	Localization in cell–cell junctions (Ast–Ast or Ast–Olig)
Required for functional activity	Needed	Needed*	Not needed
Changes in gating or transport activity	Voltage dependence, calcium sensitivity, pharmacology	Modulation of CLC transport activity by C-terminus	Abolish rectification, fast kinetics of activation, favours common gate opening
Specificity in vitro	ClC-K	ClC-7	CLC channels
Other functions independent of CLC proteins	Not studied	Regulation of WNT pathway, haematopoietic differentiation	MLC1 obligate subunit, cell adhesion, tumour suppressor

Ast, astrocyte; Oligo, oligodendrocyte.

^*Ostm1 is not needed when ClC-7 is expressed in plant vacuoles (see text for reference)

Barttin and ClC-K channels

Barttin as a ClC-K regulatory subunit

Soon after its discovery, it was found that Barttin is indeed an essential accessory subunit of ClC-K channels (Estevez et al. 2001). Co-expression of human ClC-Ka or ClC-Kb with Barttin in Xenopus oocytes or mammalian cell lines induced robust Cl⁻ currents, which, however, differed in detail between these two expression systems (Waldegger & Jentsch, 2000; Estevez et al. 2001; Scholl et al. 2006): while ClC-K–Barttin currents in Xenopus oocytes showed time- and voltage-dependent gating relaxations, and currents mediated by ClC-Kb–Barttin were rather small in oocytes, in HEK cells the Cl⁻ currents resulting from ClC-Ka or ClC-Kb co-expression with Barttin were very large and time and voltage independent. The reason for this different behaviour in the two expression systems remains unclear but seems to be independent of differences in membrane cholesterol concentration (Imbrici et al. 2014). The distinct properties shown by ClC-K channels in amphibian oocytes compared to mammalian cell lines are somehow reminiscent of the properties of the ClC-2 channel found in these expression systems (Pusch et al. 1999; Jeworutzki et al. 2012).

The association of Barttin with ClC-K channels (Waldegger et al. 2002; Hayama et al. 2003) dramatically increases their targeting to the plasma membrane (Estevez et al. 2001; Waldegger et al. 2002; Scholl et al. 2006), and the stability of the ClC-K proteins (Waldegger et al. 2002; Hayama et al. 2003; Rickheit et al. 2008). Even though these effects of Barttin on ClC-K channels have been mostly studied in vitro, they are likely to reflect the in vivo condition as shown by the Barttin knock-out mouse model (Rickheit et al. 2008; see below). In the case of rat ClC-K1, which is also expressed functionally without Barttin, ClC-K1–Barttin co-expression not only boosts ClC-K1 currents, but, interestingly, also alters biophysical and pharmacological properties of the channel (Waldegger et al. 2002; Fig.2A). Furthermore, the association of ClC-K channels with Barttin is physiologically essential also in mice, as shown by the fact that Barttin knock-out mice die a few days after birth (Rickheit et al. 2008).

Figure 2.

Barttin and GlialCAM subunits affect the functional expression of ClC-K1 and ClC-2, respectively

Typical current traces of oocytes expressing ClC-K1 or ClC-2 without and with their respective subunit in response to the IV protocols shown above the traces. A, WT ClC-K1 alone (left) and co-expressed with Barttin (right). B, WT ClC-2 alone (left) and co-expressed with GlialCAM (right).

Barttin alters the weak voltage dependence of the Cl⁻ conductance (Waldegger et al. 2002; Scholl et al. 2006; Fischer et al. 2010) of rat, but not mouse, ClC-K1 (L’Hoste et al. 2013). This effect is particularly impressive in the case of the V166E mutation, which introduces a stronger voltage dependence of gating (Waldegger & Jentsch, 2000; Scholl et al. 2006). In addition, Barttin modulates the Ca²⁺ sensitivity of rClC-K1 (Waldegger et al. 2002) and the potentiating effects of niflumic acid (Gradogna et al. 2014). The typical CLC anion selectivity sequence of Cl⁻ ≥ Br⁻ > NO₃⁻ > I⁻ of ClC-K1 is not altered by Barttin (Estevez et al. 2001; Waldegger et al. 2002; L’Hoste et al. 2013). Conflicting results were reported regarding the single channel conductance. L’Hoste et al. found a 40 pS conductance in wild-type mouse ClC-K1 with and without Barttin, reflecting the simultaneous opening of both protopores of the double-barrelled channel (L’Hoste et al. 2013). In the case of the V166E mutation, which introduces the closure of the individual protopores, the single pore conductance was found to be 20 pS, thus corresponding to the opening of individual protopores, but again independent of the presence of Barttin (L’Hoste et al. 2013). In contrast, using non-stationary noise analysis, Scholl et al. reported that Barttin co-expression increases the pore conductance of rat ClC-K1 by a factor of three (Scholl et al. 2006). However, this result should be interpreted with caution, as the analysis method was rather indirect. Overall, it appears that Barttin does not lead to significant changes in pore structure.

Barttin specifically interacts with ClC-K channels and has no functional effect on other CLC proteins (Estevez et al. 2001). This is different from the relatively unspecific interaction observed for GlialCAM, which is able to associate with many different CLC channels, including ClC-K channels, but not with the ClC-5 Cl⁻–H⁺ antiporter (Jeworutzki et al. 2014). Tajima et al. found that short constructs of ClC-K2 containing helix B or helix J are able to biochemically interact with Barttin (Tajima et al. 2007), and most of the C-terminus of Barttin can be deleted without losing the ability to boost surface expression of ClC-K channels (Estevez et al. 2001; Scholl et al. 2006). However, the stoichiometry and the precise regions of the channel–Barttin interaction remain unknown. The C-terminus of Barttin contains a so-called ‘PY’ motif (Fig.1A). Such proline-rich segments have been shown to be important for the regulation by ubiquitination, for example in the epithelial sodium channel (Staub et al. 1996). In fact, eliminating the motif by mutating Y98 to alanine increases by about two-fold the currents of ClC-Ka and ClC-Kb co-expressed with the mutant Barttin in Xenopus oocytes (Estevez et al. 2001). However, results suggesting regulation of ClC-Ka–Barttin channels by Nedd4-2 and serum- and glucocorticoid-dependent kinases (Embark et al. 2004) could not be reproduced (Jentsch, 2008). Thus, in the absence of a convincing ligase or a demonstration of altered ubiquitination, the relevance of the PY motif remains unclear.

Interestingly, the Barttin subunit participates in the pharmacological potentiation of ClC-K channels (Gradogna et al. 2014). Niflumic acid (NFA) is the most potent activator of ClC-K channels so far identified (Picollo et al. 2004; Liantonio et al. 2006; Zifarelli et al. 2010). NFA belongs to the class of fenamates that includes flufenamic acid, which, instead, blocks ClC-Ks. Recently, a particular region of ClC-Ks, the I–J loop, was found to be involved in the modulation of these channels. Residues belonging to I–J loop form an intersubunit Ca²⁺ binding site (Gradogna et al. 2010, 2012). Two adjacent residues of the same loop, F256A and N257A, dramatically change the response to NFA of ClC-Ka. The F256A mutant dramatically increases NFA potentiation of the channel, while the N257A mutant is only blocked by NFA (Gradogna et al. 2014). Moreover, the F256A mutation, when inserted into the background of rat ClC-K1, is able to cause a transient potentiation of the channel by NFA, and this effect is found only when F256A ClC-K1 is co-expressed with Barttin.

Disease-causing mutations in Barttin

BSND is recessively inherited and all mutations in the BSND gene causing Bartter syndrome type IV lead to a loss or large reduction of function. This is evident for early stop codons or mutations that result in the loss of the start methionine (Birkenhager et al. 2001). The loss of function of Barttin leads to a more severe disease phenotype compared to classical Bartter syndrome type III caused by mutations in CLCNKB, because it gives rise to a functional ‘knock-out’ of both ClC-K channels. In fact, simultaneous loss-of-function mutations in the genes coding for ClC-Ka and ClC-Kb also results in a severe Bartter syndrome with sensorineural deafness (Schlingmann et al. 2004; Nozu et al. 2008). Mechanistically, reduced Barttin function could be caused by a Barttin folding defect, the inability of Barttin to be targeted to the correct (i.e. basolateral) membrane, by a reduced affinity for ClC-K proteins, or by the inability to ‘activate’ the ClC-K channel activity, once bound. The reported missense mutations of BSND are R8L, R8W, G10S, and G47R (Birkenhager et al. 2001; Estevez et al. 2001; Miyamura et al. 2003; Shalev et al. 2003; Garcia-Nieto et al. 2006), while the I12T mutation causes non-syndromic deafness but only mild renal symptoms (Riazuddin et al. 2009). The affected amino acids are highlighted in Fig.1A. Mutations R8L, R8W and G47R lead to drastically reduced, but not abolished, boosting of ClC-K channel function in Xenopus oocytes, as well as in transfected cells (Estevez et al. 2001; Janssen et al. 2009), in agreement with the severe Bartter’s syndrome phenotype seen in patients. While Janssen et al. reported that these mutations allowed the insertion of Barttin into the plasma membrane but prevented activation of the channel (Janssen et al. 2009), Hayama et al. found that the R8L mutation caused intracellular retention (Hayama et al. 2003). In agreement with the latter result, analysis of knock-in mice carrying the Barttin R8L mutation confirmed an impaired plasma membrane localization of the mutated protein (Nomura et al. 2011). Thus, results obtained in heterologous expression systems, using rather artificial constructs (i.e. concatamers with other CLC proteins; ClC-K–Barttin or GFP fusion constructs) have to be interpreted with extreme caution (Scholl et al. 2006; Janssen et al. 2009). In fact, it is difficult to clarify parameters like plasma membrane localization and protein stability using heterologous overexpression systems.

Interestingly, the G10S mutant functioned like normal Barttin in oocytes (Estevez et al. 2001) but not in transfected cells (Janssen et al. 2009). It might be that a folding defect induced by the mutation could be reversed at the lower temperature at which amphibian oocytes are incubated (18–19°C) as has been found for other Cl⁻ channels, like CFTR and the most common cystic fibrosis-causing ΔF508 mutation (Denning et al. 1992) or other transporters like the gout-causing Q141K mutation in ABCG2 (Woodward et al. 2013).

The I12T mutation in homozygosity induces a mild renal phenotype but causes sensorineural deafness (Riazuddin et al. 2009). In heterologous expression systems, the mutation reduces the boosting effect on ClC-K channel activity by interfering with the chaperone function of Barttin (Riazuddin et al. 2009). Overall, these observations are consistent with the idea that Cl⁻ transport is rate limiting in the inner ear and simultaneous partial impairment of both isoforms can lead to deafness, while Cl⁻ transport is not rate-limiting in the kidney and a partial reduction of function can be tolerated. However, more realistic animal models (Rickheit et al. 2008; Nomura et al. 2011) are needed to clarify the disease mechanisms in vivo.

Localization and function of Barttin in the kidney

Barttin localization closely follows that of both ClC-K isoforms, suggesting that these proteins form a stable and obligate protein complex (Estevez et al. 2001; Rickheit et al. 2008). The ClC-K1–Barttin complex is found mostly in the thin ascending limb of the loop of Henle (Estevez et al. 2001; Nomura et al. 2011) where it is important for the counter-current system of urine concentration. This role was strongly supported by the phenotype of ClC-K1 knock-out mice, which show a severe defect in urine concentrating ability (Matsumura et al. 1999). Conflicting results have been reported regarding the subcellular localization in the thin limb: Uchida et al. reported apical as well as basolateral expression of ClC-K channels in the thin limb (Uchida et al. 1995), in agreement with functional data on Cl⁻ permeability (Imai & Kokko, 1974, 1976). In contrast, Vandevalle et al. found only basolateral expression of ClC-K channels in this nephron segment (Vandewalle et al. 1997). Barttin was also found exclusively on the basolateral membrane in the thin limb (as in all other nephron segments where Barttin is expressed) (Estevez et al. 2001). Thus, if rodent ClC-K1 (human ClC-Ka) is indeed also localized apically in the thin limb, it would probably need an as-yet unknown apically based accessory subunit.

A predominant function of ClC-Kb–Barttin complexes in the thick ascending limb of the loop of Henle (TAL) and in the distal convoluted tubule (DCT) is to provide a basolateral exit pathway for Cl⁻ ions that have been taken up by the apical NKCC2 Na⁺–K⁺–2Cl⁻ co-transporter (in the TAL) or the Na⁺–Cl⁻ co-transporter in the DCT (Jentsch, 2005, 2008; Jentsch et al. 2005; Uchida & Sasaki, 2005; Sile et al. 2006; Fahlke & Fischer, 2010; Jeck & Seyberth, 2011). The massive salt loss associated with Bartter’s syndrome is due to the defective NaCl reabsorption in these nephron segments. Barttin and ClC-Kb are additionally expressed in the basolateral membrane of acid-secreting alpha intercalated cells and base-secreting beta intercalated cells in the cortical collecting duct (CCD) (Estevez et al. 2001). Tissue specific knock-out studies could help to uncover the physiological role in the CCD.

In MDCK cells, heterologously expressed Barttin localizes mostly to the basolateral membrane, suggesting that Barttin is the major determinant of the basolateral preference of ClC-K–Barttin complexes (Janssen et al. 2009). Interestingly, the disease causing E88X Barttin mutation appeared to affect epithelial sorting (Janssen et al. 2009).

Localization and function of Barttin in the inner ear

Perception of sound is mediated by the movement of stereocilia on inner and outer hair cells coupled to the opening of mechanosensitive cation channels. Quite unusually compared to other excitable cells, depolarization of hair cells is mediated by an influx of K⁺ ions from the K⁺-rich (∼150 mm) endolymph bathing the stereocilia, driven by a ∼100 mV extracellular voltage (Zdebik et al. 2009). The K⁺ and voltage gradients are generated by the stria vascularis, a vascularized epithelial cell layer at some distance from the hair cells. In this way, hair cells do not have to generate the ion gradients necessary for electrical excitation, relieving the mechanically sensitive parts from large direct ATP consumption (via the Na⁺–K⁺-ATPase) and thus from sustained glucose supply via blood vessels that may interfere with the electromechanical properties of the organ of Corti.

ClC-Ka, ClC-Kb and Barttin are co-expressed in the basolateral membrane of marginal cells of the stria vascularis and in dark cells of the vestibular organ (Estevez et al. 2001; Sage & Marcus, 2001; Kobayashi et al. 2002; Rickheit et al. 2008) and are critical components for the generation of the endolymph K⁺ concentration and the positive potential (Estevez et al. 2001; Rickheit et al. 2008; Zdebik et al. 2009), explaining the deafness in BSND patients. Marginal cells import K⁺ through the NKCC1 Na⁺–K⁺–2Cl⁻ co-transporter and through the Na⁺–K⁺-ATPase at the basolateral membrane. Na⁺ is expelled by the Na⁺–K⁺-ATPase, whereas Cl⁻ is basolaterally recycled through ClC-Ka–Barttin and ClC-Kb–Barttin channels. A priori, it might be expected that a loss of function of ClC-K–Barttin channels might completely impair the ability of marginal cells to secrete K⁺ into the endolymph. Detailed information on the physiological role of ClC-K–Barttin for endolymph production was obtained by Rickheit et al. who generated and studied an inner ear specific Barttin knock-out mouse model (Rickheit et al. 2008), which was deaf as expected. Interestingly, in this model the endolymph maintained a high K⁺ concentration, but the extracellular positive potential was drastically reduced (Rickheit et al. 2008). Thus, other Cl⁻ channels might compensate for the lack of Barttin/ClC-K channels and allow for a residual K⁺-secreting activity of marginal cells. It has to be remembered that in the absence of a positive voltage of the endolymph compartment, the driving force for K⁺ entry into hair cells will be reduced, but not completely abolished, as the basolateral pole of hair cells is bathed in extracellular perilymph-like fluid with a low K⁺ concentration. Thus, a small residual Cl⁻ conductance might be sufficient to establish the high endolymph K⁺ concentration.

Ostm1 and ClC-7

Ostm1 as a ClC-7 subunit

ClC-7 is an electrogenic 2Cl⁻–H⁺ exchanger (Leisle et al. 2011) that is located in the late endosome and lysosome compartments of various tissues (Kornak et al. 2001). The loss of ClC-7, both in knock-out mice and human patients, produces osteopetrosis characterized by an abnormal increase of bone mass (Kornak et al. 2001). This phenotype is also present in the long-known grey-lethal (gl) mouse (Gruneberg, 1936), and the identification of the gene responsible for this phenotype led to the discovery of the protein Ostm1 (Osteopetrosis associated transmembrane protein 1) and of the presence of mutations in OSTM1 in a small subset of patients with osteopetrosis (Chalhoub et al. 2003). In contrast to osteopetrosis caused by defects in other genes, mutations in OSTM1 and some mutations found in CLCN7 additionally cause neuronal degeneration, suggesting a possible relationship between the two proteins. Like ClC-7, Ostm1 is found in late endosomes and lysosomes, co-immunoprecipitates with ClC-7, and ClC-7 levels are severely reduced in gl mice, suggesting that Ostm1 is necessary for ClC-7 protein stability, and was hence deduced to be its β-subunit (Lange et al. 2006).

Ostm1 is composed of 338 amino acids (in the mouse), has a single transmembrane domain, a short cytoplasmic tail, and a longer luminal domain with several glycosylation sites and a cleavable signal peptide (Fig.1B; Chalhoub et al. 2003; Lange et al. 2006). The luminal domain of Ostm1 has a weak similarity to a RING finger protein, and based on a wrong assumption of its topology, Ostm1 was initially proposed to be an E3 ubiquitin ligase (Fischer et al. 2003). Interestingly, ClC-7 is the only CLC protein that is not glycosylated, raising the possibility that Ostm1 protects ClC-7 from the action of lysosomal proteases, which would explain the reduction of ClC-7 protein levels in gl mice (Lange et al. 2006). Nevertheless, the absence of ClC-7 also triggers a reduction in Ostm1 protein levels (Lange et al. 2006); consequently the two proteins seem to stabilize each other, a reciprocal relationship not found for the other CLC channels/β-subunits known so far. Moreover, whereas ClC-2 and ClC-K require GlialCAM and Barttin, respectively, for proper subcellular localization (Estevez et al. 2001; Jeworutzki et al. 2012), ClC-7 reaches the lysosomal membrane on its own, and drives Ostm1 to it. When a mutant version of ClC-7 is mistargeted to the plasma membrane, Ostm1 again co-localizes with it (Stauber & Jentsch, 2010). Just such a mutant form of ClC-7 allowed its interaction with Ostm1 to be studied: while the single transmembrane domain of Ostm1 is sufficient to reach the lysosome in the presence of ClC-7, a chimeric protein with the transmembrane domain together with the amino-terminal part is required to activate ClC-7 ion transport (Leisle et al. 2011). However, the cytoplasmic domain of Ostm1 may be needed to modulate ClC-7 activity, since a chimeric form without the domain produced larger ClC-7 currents than the complete Ostm1 (Leisle et al. 2011). It is interesting to note that heterologous expression of ClC-7 in plant cells yielded vacuolar currents independently of Ostm1, which is absent in plant genomes (Costa et al. 2012), raising the possibility that a small fraction of ClC-7 in animal cells may be active even without interacting with Ostm1.

Disease-causing mutations in Ostm1

Four different mutations, all with a recessive inheritance, have been found in OSTM1 (see Fig.1B): a G to A transition close to the donor splice site of intron 5 that leads to the defective splicing of the transcripts, skipping exon 5, and producing truncated forms of the protein after residue 261, which could be potentially secreted (see below; Chalhoub et al. 2003); a two-nucleotide deletion in codon 138, resulting in a frame-shift that produces a truncated protein after eleven more residues (Pangrazio et al. 2006); a T to A change at position 36 of the cDNA that yields a stop codon at position 12 of the protein (Pangrazio et al. 2006); and a G to T mutation that introduces a stop codon at position 86 of the protein (Maranda et al. 2008). Additionally, two different homozygous microdeletions affecting the OSMT1 locus have been described in two unrelated families (Ott et al. 2013).

Localization and function of Ostm1 in the bone and the nervous system

ClC-7 and Ostm1 are significantly abundant in osteoclasts, the bone remodelling cell type of haematopoietic origin. The expression of Clcn7 and Ostm1, together with other osteoclast-specific genes, is activated by the MITF transcription factor (Meadows et al. 2007). In osteoclasts, ClC-7 and Ostm1 concentrate in the ruffled membrane that is formed by lysosomal fusion with the plasma membrane and is in direct contact with the bone matrix (Lange et al. 2006). It was suggested that the coordinated action of the vacuolar H⁺-ATPase and ClC-7–Ostm1 acidifies the resorption lacuna between the osteoclast and the bone matrix, activating bone-degrading enzymes, such as cathepsin K. The loss of ClC-7 or Ostm1 function causes a functionally defective ruffled membrane, producing an osteoclast unable to degrade the bone matrix, and, consequently, it leads to an increase in bone calcification, that even fills the bone marrow cavity. This process also generates blindness and deafness due to the compression of visual and auditory nerves by the thickening bone in the skull (Sobacchi et al. 2013).

The spectrum of phenotypes caused by mutations in CLCN7 ranges from a dominant benign form (autosomal dominant osteopetrosis II, also called Albers-Schönberg disease, MIM no 166600) to a more severe autosomal recessive form, associated with neurological deficits evident early in life and frequently lethal (MIM no. 611490) (Pangrazio et al. 2010). OSTM1 mutations cause a more severe neurological phenotype than the recessive mutations in CLCN7 (MIM no. 259720), which makes bone marrow transplantation to provide healthy osteoclasts unsuitable as a treatment for these patients. Only a few patients with mutations in OSTM1 have been reported and all of them died within the first year of life (Quarello et al. 2004; Pangrazio et al. 2006; Maranda et al. 2008; Ott et al. 2013).

The function of ClC-7–Ostm1 in lysosomes is apparently more critical in neurons, where it underlies the neuronal degeneration, than in other cells like renal proximal tubule cells (Wartosch et al. 2009). Although it was proposed that the role of ClC-7 in lysosomes would be to generate an electrical shunt to allow the acidification of the lumen, lysosomes in cultured neurons from ClC-7 knock-out mice present a normal steady-state pH, as measured by fluorescent methods (Kasper et al. 2005), and the lysosomal function, but not pH, is affected even in the presence of a mutated ClC-7 that works as an uncoupled chloride conductor (Weinert et al. 2010). Furthermore, a knock-in mouse harbouring a mutation that renders ClC-7 completely transport deficient, but properly localized, also showed a normal lysosomal pH, demonstrating that some of ClC-7 cellular functions are independent of ion transport (Weinert et al. 2014). Thus, ClC-7 may have a yet-unknown role in lysosomal physiology. ClC-7 is expressed throughout the nervous system (Kornak et al. 2001), and its loss in mice produces neurodegeneration in hippocampus, cortex, cerebellum and retina, associated with astrogliosis and microglia activation (Kasper et al. 2005). Neuronal death is accompanied by an accumulation of electron dense material and autofluorescent pigments, similarly to the neuronal ceroid lipofuscinoses, a group of lysosomal storage diseases (Kasper et al. 2005). The absence of ClC-7 also reduced the degradation of endocytosed proteins in the lysosomes (Wartosch et al. 2009).

The loss of Ostm1 causes a more severe phenotype: while ClC-7 knock-out mice survive up to 6–7 weeks (Kornak et al. 2001), Ostm1-deficient mice only live 3–4 weeks (Chalhoub et al. 2003). Transgenic rescue of Ostm1 expression in the haematopoietic lineage extended the life expectancy to more than 5 weeks, which allowed a detailed characterization of the neuronal phenotype (Heraud et al. 2014). Massive neuronal death was detected in the hippocampus, cerebral cortex, and retina and among Purkinje cells in the cerebellum, accompanied by thinning of the corpus callosum and enlargement of the lateral ventricles. At the cellular level, the degenerating neurons showed an increased accumulation of glycogen and lipids, which was different from neuronal ceroid lipofuscinosis and, therefore, from ClC-7-deficient mice. The presence of electron-dense ubiquitin-positive inclusions revealed the accumulation of autophagosomes, but not of lysosomes, which appeared normal.

Additional functions for Ostm1

The increased severity of the phenotype caused by the deficiency of Ostm1 opened up the possibility of additional roles for Ostm1 besides stabilizing ClC-7 in lysosomes and ruffled membrane. Like ClC-7, Ostm1 has a broad expression pattern (Chalhoub et al. 2003). Ostm1 is expressed in osteoclasts, bone cells of haematopoietic origin, but it is also expressed in other haematopoietic lineages, such as B, NK and mast cells. gl mice suffer from anaemia, leucopoenia, and lymphopoenia, and furthermore develop a reduced thymus (Pata et al. 2008). Although the osteopetrotic phenotype caused by the loss of ClC-7 can be rescued by transgenic expression of ClC-7 in osteoclasts (Kasper et al. 2005), the phenotype of gl mice cannot be rescued by the expression of Ostm1 in osteoclasts, but by introducing a transgene driving Ostm1 expression in several early haematopoietic progenitors (Pata et al. 2008). This suggests a wider role for Ostm1 in haematopoietic differentiation, independent of ClC-7.

Indeed, Ostm1 has been shown to regulate canonical Wnt signalling (Feigin & Malbon, 2008), which plays a role in bone homeostasis (Baron & Kneissel, 2013). Moreover, a mutation in OSTM1 found in an osteopetrosis patient produces a truncated protein devoid of the transmembrane domain and cytoplasmic tail that could be potentially secreted (Lange et al. 2006), and this mutant Ostm1 suppresses Wnt signalling (Feigin & Malbon, 2008). Also, this secreted Ostm1 binds to osteoclast precursors and inhibits osteoclastogenesis (Shin et al. 2014). Ostm1 has been identified as a target of miR-140 in pluripotent stem cells in response to bone morphogenetic protein-4 (BMP4) treatment, which promotes adipocyte lineage commitment; thus, Ostm1 would work as an anti-adipogenic factor in this system, and it would also potentially link BMP and Wnt signalling pathways (Liu et al. 2013).

GlialCAM and ClC-2

Multiple physiological functions of ClC-2

ClC-2 is almost ubiquitously expressed (Grunder et al. 1992; Thiemann et al. 1992). The distinct physiological roles of the channel have been revealed by Clcn2 knock-out mice (Bosl et al. 2001; Nehrke et al. 2002; Makara et al. 2003; Zdebik et al. 2004; Romanenko et al. 2008; Huang et al. 2009; Cortez et al. 2010; Edwards et al. 2010; Catalan et al. 2012; Nighot & Blikslager, 2012), and from patients with mutations in CLCN2 (Depienne et al. 2013; Di Bella et al. 2014). The first studies showed that ClC-2 protein deficiency caused male germ cell and photoreceptor degeneration in the knock-out mice (Bosl et al. 2001). Only a few patients with CLCN2 mutations have been described, some of them show chorioretinopathy with visual field defects (Depienne et al. 2013) and a recently identified male patient also shows infertility (Di Bella et al. 2014). However, the physiological role of ClC-2 in the retina and testes is unknown, and it is only suggested that ClC-2 may be important for controlling the ionic environment of the cells in these tissues.

The possible physiological role of ClC-2 in the brain has been studied in detail. ClC-2 protein has been detected in neurons, oligodendrocytes and astrocytes (Sik et al. 2000; Blanz et al. 2007; Jeworutzki et al. 2012). Several reports indicated that the ClC-2 channel in neurons may constitute part of the background conductance regulating input resistance and providing an efflux pathway for chloride. This could be a safeguard mechanism to prevent chloride accumulation in active GABAergic synapses (Foldy et al. 2010; Rinke et al. 2010). The role for ClC-2 in chloride efflux is compatible with its electrophysiological activity, since ClC-2 function is activated by hyperpolarization (Clark et al. 1998). However, recent modelling studies (Ratte & Prescott, 2011) together with studies performed on ClC-2 from C. elegans have indicated that the primary role of ClC-2 is to mediate chloride influx and not efflux, regulating egg laying in C. elegans by modulation of motor neuron activity (Branicky et al. 2014).

A putative role of ClC-2-mediated chloride influx has also been suggested to be important in glial cells. Initially, it was indicated that ClC-2 current was reduced in immature astrocytes and reactive astrocytes within a lesion, suggesting a role for ClC-2 in glial cell development (Makara et al. 2003). Later, studies of aged Clcn2^−/− mice showed that they develop a leukoencephalopathy with intramyelinic oedema (Blanz et al. 2007). The observed vacuolization was very similar to that observed in the human disease megalencephalic leukoencephalopathy with subcortical cysts (MLC) (van der Knaap et al. 2012), a rare type of leukodystrophy caused by mutations in MLC1 (Leegwater et al. 2001) or GLIALCAM (Lopez-Hernandez et al. 2011a), suggesting that ClC-2 may have a role in the pathophysiology of MLC. However, mutations in CLCN2 were not found in patients with MLC (Scheper et al. 2010). CLCN2 mutations were found in a few patients suffering from another leukodystrophy of unknown origin characterized by white matter oedema, with clinical or subclinical presentations (Depienne et al. 2013). How might ClC-2 dysfunction lead to leukodystrophy in mice and humans? The exact answer for this question is not known, and it has been suggested that it may alter the ion homeostasis of glial cells (Blanz et al. 2007). Based on indirect evidence, the co-expression of ClC-2 with the potassium channel Kir4.1 and the similarity of vacuolation phenotypes between the knock-out mice deficient in ClC-2, Kir4.1 and with the double ablation of connexins 32 and 47, it was suggested that ClC-2 may be important in the process of potassium siphoning (Blanz et al. 2007), which is needed to transport potassium away from active neurons through the glial syncytium and into blood vessels (Rash, 2010). In this respect, chloride influx into potassium-depolarized glial cells may be needed for thermodynamically favourable electroneutral transport. However, as ClC-2 was assumed to be inactive at depolarized potentials, this suspected physiological role of ClC-2 seemed to contrast with its biophysical properties. The scenario changed after the discovery that GlialCAM increases ClC-2 activity at positive voltages (see below).

GlialCAM as a ClC-2 subunit

GlialCAM (also named HepaCAM; see below) is a type I transmembrane protein comprising two immunoglobulin (Ig)-like domains (one V-set and another one of the C2 type) in the glycosylated extracellular side and an intracellular C-terminus, which contains a low-complexity proline-rich tail that can be phosphorylated. Altogether it is composed of 416 amino acids. Mutations in GLIALCAM were identified in a small subset of patients with MLC (Lopez-Hernandez et al. 2011a). The majority of MLC cases are caused by mutations in MLC1 (MIM no. 604004) (Leegwater et al. 2001), which encodes an oligomeric membrane protein that is mainly expressed in brain astrocytes and whose function is not known (Teijido et al. 2004). GlialCAM interacts and co-localizes with MLC1 in astrocyte–astrocyte junctions (Lopez-Hernandez et al. 2011a). Most mutations found in GLIALCAM abolish GlialCAM localization in cell junctions (Lopez-Hernandez et al. 2011b). Furthermore, astrocytes with a reduced expression of GlialCAM by RNA interference showed that GlialCAM is an obligate MLC1 subunit required for its exit from the endoplasmic reticulum and to target it to cell junctions (Capdevila-Nortes et al. 2013).

GlialCAM is expressed throughout the brain (Favre-Kontula et al. 2008). Interestingly, GlialCAM, but not MLC1, is detected in oligodendrocytes, the myelin-forming cells that are mainly affected in MLC patients (Teijido et al. 2004; Hoegg-Beiler et al. 2014; Dubey et al. 2015). Thus, it was suspected that GlialCAM may have additional protein partners in oligodendrocytes. Mass spectroscopy analysis of GlialCAM-interacting proteins in brain membranes identified peptides corresponding to the ClC-2 chloride channel (Jeworutzki et al. 2012).

Like its interaction with MLC1, GlialCAM targets ClC-2 to cell junctions. In addition, GlialCAM strongly modifies the biophysical properties of ClC-2 (Jeworutzki et al. 2012). In the presence of GlialCAM, ClC-2-mediated currents are increased and are instantaneously activated and almost ohmic, i.e. GlialCAM strongly reduces ClC-2 inward rectification (Fig.2B). Non-stationary noise analysis indicated that GlialCAM does not modify single channel properties of ClC-2. It was observed that GlialCAM diminished the inhibition of ClC-2 by strongly acidic pH, which is thought to be mediated by reducing the open probability of the common gate (Arreola et al. 2002), suggesting that GlialCAM may activate the channel by opening its common gate. In agreement with this initial observation, further studies indicated that GlialCAM was able to interact with the common gate-deficient ClC-2 mutant E211V/H816A and to target it to cell contacts, without altering its function (Jeworutzki et al. 2014).

Unlike the other identified CLC subunits, GlialCAM is able to interact with all CLC channels tested (Jeworutzki et al. 2012, 2014), including ClC-2 from Drosophila (whose genome lacks a GlialCAM orthologue), ClC-0, ClC-1, and even with the heteromer ClC-Ka–Barttin. Similar to its effect on ClC-2, GlialCAM targets all channels to cell junctions and increases, to different degrees, their function by activating the common gate. This was clearly seen in ClC-0 slow gate-deficient mutants, whose function was not activated by GlialCAM (Jeworutzki et al. 2014). In contrast, GlialCAM does not target to junctions, neither does it modify the biophysical properties of the ClC-5 proton–chloride antiporter (Jeworutzki et al. 2012, 2014). It has been suggested that GlialCAM appeared during evolution in chordates (Barrallo-Gimeno & Estevez, 2014) and first developed an interaction surface that is conserved in all CLC channels, but not in CLC transporters.

The closest GlialCAM homologue is Hepacam2 (Klopfleisch et al. 2010), which contains an additional Ig-like domain. No biochemical or functional interaction was observed between HepaCAM2 and ClC-2 (Jeworutzki et al. 2012). In zebrafish, the teleost-specific genome duplication yielded two GlialCAM paralogues: glialcama and glialcamb (Sirisi et al. 2014) and three zebrafish clc-2 channels: clc-2a, clc-2b and clc-2c (Perez-Rius et al. 2015). Glialcama is able to modify the functional properties of ClC-2 and clc-2a, and slightly increases the currents of clc-2b, whereas no clc-2c currents are detected. Furthermore, glialcama is able to target all CLC proteins to cell junctions, whereas glialcamb reduces ClC-2-mediated currents. Chimeric studies between all these proteins may be useful in order to understand how GlialCAM and ClC-2 proteins interact.

Disease-causing mutations in GlialCAM

Mutations in GLIALCAM have been found in approximately 25% of MLC patients. While some patients showed two mutated alleles, others presented only one mutated copy, suggesting dominant inheritance (Lopez-Hernandez et al. 2011a). This difference in genetic behaviour is correlated with the severity of the symptoms: recessive mutations show a classic MLC phenotype (MLC2A, MIM no.613925), and dominant mutations are associated with a more benign, remitting form of MLC (MLC2B, MIM no.613296). All the mutations in GLIALCAM associated with MLC are found in the extracellular part of the protein, and the dominant mutations cluster on the V-set Ig-like domain (Lopez-Hernandez et al. 2011a).

None of the tested MLC-related mutations identified in GlialCAM abolish the interaction with ClC-2 or the ability to modify the functional properties of ClC-2 (Jeworutzki et al. 2012; Arnedo et al. 2014a,b2014). In contrast, some mutations affect the ability of directing ClC-2 (and MLC1) to cell junctions (Jeworutzki et al. 2012; Arnedo et al. 2014a,b2014). These data suggest that the domains of GlialCAM involved in targeting and functional modification are different and independent. Similarly, deletion of the entire C-terminus of GlialCAM or part of the N-terminus of ClC-2 did not abolish the functional interaction between both proteins, indicating that other unidentified regions in the extracellular and/or transmembrane domains are important for functional interaction (Jeworutzki et al. 2012).

A complex protein network formed by ClC-2, GlialCAM and Mlc1

Considering that GlialCAM interacts with MLC1 and ClC-2, and that no functional or biochemical interaction was observed between MLC1 and ClC-2, it was hard to understand how the clinical phenotype of patients with mutations in MLC1 is virtually indistinguishable from those that presented GLIALCAM mutations in both alleles. In order to address this question and to investigate a possible functional network of MLC1, GlialCAM and ClC-2, mice deficient in each protein were compared, together with a Glialcam^dn/dn knock-in mice harbouring a dominant mutation found in patients (Hoegg-Beiler et al. 2014).

All mouse models developed megalencephaly and progressive myelin vacuolization, mostly in the cerebellum, and did not develop neurological signs. There are differences between human and mice in the onset, probably due to the heterochrony of myelination. However, it is clear that vacuolation manifests when myelination is at its highest level in humans and mice. The progressive vacuolation phenotype observed in Mlc1- and GlialCAM-deficient mice is very similar. Thus, it was concluded that both animal models develop MLC only in an early stage of the disease, and thus can be used to study the similarity of phenotypes (Hoegg-Beiler et al. 2014).

As expected, loss of GlialCAM changes the localization and abundance of ClC-2 in astrocytes and oligodendrocytes in the cerebellum. Unexpectedly, loss of MLC1, which is only expressed in astrocytes, also changes the localization and abundance of ClC-2 in both types of cells in the cerebellum. Interestingly, loss of MLC1 also changed the localization of GlialCAM in both cell types, without reducing GlialCAM protein levels (Hoegg-Beiler et al. 2014). Similar results were found in another different Mlc1 knock-out mouse line (Dubey et al. 2015), in a zebrafish mlc1 mutant line that does not express mlc1 and in a brain biopsy from an MLC1 patient (Sirisi et al. 2014). However, impaired localization of GlialCAM was observed in astrocyte cultures from Mlc1-deficient mice only in the presence of elevated potassium levels, which mimics neuronal activity (Capdevila-Nortes et al. 2013; Sirisi et al. 2014). Thus, it was proposed that MLC1 regulates GlialCAM localization in astrocytes in an activity-dependent manner. As GlialCAM molecules form homophilic trans-interactions between different cells, it was hypothesized that GlialCAM may bridge together, in a ternary heterocomplex, Mlc1 from astrocytes and ClC-2 from oligodendrocytes (Hoegg-Beiler et al. 2014), as previously suggested (Maduke & Reimer, 2012).

As expected from the biochemical data that indicate that ClC-2 is associated with GlialCAM in oligodendrocytes, ClC-2 currents recorded in these cells showed neither time dependence nor inward rectification. In contrast, the same measurements in GlialCAM- or Mlc1-deficient mice display typical hyperpolarization-activated ClC-2 currents (Hoegg-Beiler et al. 2014). When the same measurements were performed in astrocytes, ClC-2 currents were as if they were not associated with GlialCAM. In fact, based on recordings of astrocyte cultures in vitro that express ClC-2 and GlialCAM it looks like these two proteins do not interact (Jeworutzki et al. 2012). Two possible explanations are plausible: it could be that the requirements for anchoring or for gating changes have a different stoichiometry or, alternatively, GlialCAM in astrocytes may be sequestered by MLC1, regulating ClC-2 only under special circumstances. In fact, in tissue, GlialCAM and MLC1 share a restricted localization in distal astrocytic processes, whereas ClC-2 has more diffuse membrane localization in astrocytes (Depienne et al. 2013). GlialCAM is clearly an obligate β-subunit of MLC1 required for endoplasmatic reticulum exit (Capdevila-Nortes et al. 2013), but its relationship with ClC-2 is far more complex and still needs to be defined.

Additional roles for GlialCAM

GlialCAM had been also identified as a silenced gene in human hepatocellular carcinoma (hence the reason for the name HepaCAM) (Moh et al. 2005). Transfection of GlialCAM increased cell adherence, motility and invasion properties (Moh et al. 2009a,b2009) and, in some cases, it had an anti-proliferative effect. These data suggest that GlialCAM may act as a cell adhesion molecule but also as a tumour suppressor (Moh et al. 2008). However, it is not clear if this second role has any physiological relevance, as GlialCAM is not expressed in the liver and GlialCAM knock-out mice do not show any evidence of having an increased incidence of tumours (Hoegg-Beiler et al. 2014). It could be that the effects observed in cancer cell lines are an unspecific consequence of the experimental model.

Concluding remarks

The physiological roles of ClC-Ka–Barttin and ClC-Kb–Barttin channels are comparatively well understood in the kidney and in the inner ear. Open questions include their role in acid- and base-secreting intercalated cells in the cortical collecting duct, and their precise localization in the thin ascending limb of the loop of Henle. Although the relationship between ClC-7 and Ostm1 is clear in bone physiology, some questions remain regarding the role of Ostm1 in the brain, possibly independently of ClC-7, and how precisely Ostm1 activates ClC-7 ion transport. Also, it is not well understood if the putative secreted form of Ostm1 has a specific receptor and function in vivo. The physiological role of ClC-2 in the brain still needs to be completely defined. Understanding its relationship with GlialCAM and careful electrophysiological analyses in conditional ClC-2 knock-out mice may provide new suggestions about its role.

Determination of the stoichiometry of the Barttin–ClC-K complex and, more precisely, determination of its atomic 3D structure represents an interesting question for future studies. Such structural insight could also help to develop more specific pharmacological tools to target ClC-K–Barttin channels. Similar studies should also be developed for ClC-2 and GlialCAM.

Specific ClC-Kb–Barttin vs. ClC-Ka–Barttin blockers (or vice versa) might be useful diuretics (Fong, 2004). On the other hand, activators of ClC-K–Barttin channels could, in principle, be useful for Bartter syndrome type III or type IV patients with residual channel function. Thus, further studies regarding ClC-K channel modulation are highly desirable. The recently confirmed role of Barttin in the ClC-K activation by NFA suggests that the knowledge of the Barttin–ClC-K interface might provide new opportunities to obtain more specific drugs. No activators or inhibitors exist for ClC-7–Ostm1 that could be used to treat osteopetrosis or osteoporosis, respectively. Activators of ClC-2 could be used potentially to treat white matter leukoencephalopathies such as MLC, although this may not be the complete answer, as crosses between Clcn2^−/− and Glialcam^−/− mice revealed that the loss of Mlc1 has additional pathogenic effects unrelated to ClC-2 (Hoegg-Beiler et al. 2014).

We envisage that in the near future other regulatory subunits for different CLC proteins required for changing their activity, subcellular localization, or both might be identified. Furthermore, proteins interacting with the already-identified subunits may in turn also regulate CLC protein function. Careful biochemical studies and the power of genetics in humans or other model organisms may provide new CLC regulators. Functional studies of CLC proteins in all kingdoms may provide new conceptual aspects. For example, recent studies in plants suggested that the outwardly rectifying current–voltage relationship was more prominent after AtClC-a phosphorylation by the OST1 kinase, allowing chloride flux release during stomatal closure in response to the stress hormone abscisic acid (Wege et al. 2014). Similar regulation mechanisms may also exist for endosomal CLC proteins or for hitherto unidentified accessory subunits. We believe that new surprises will appear in the near future.

Glossary

Ast

astrocyte

BMP

bone morphogenetic protein

CCD

cortical collecting duct

DCT

distal convoluted tubule

gl

grey-lethal

Ig

immunoglobulin

NFA

niflumic acid

Oligo

oligodendrocyte

TAL

thick ascending limb of the loop of Henle

Biographies

Alejandro Barrallo Gimeno obtained his PhD at the University of Salamanca (Spain) in 1999. After post-doctoral training at the Institute of Developmental Genetics (Helmholtz Zentrum München, Germany) and Institute of Neurosciences (Universidad Miguel Hernández-CSIC, Sant Joan d’Alacant, Spain), in 2012 he became an associate professor of physiology at the University of Barcelona. His research interest is the role of CLC chloride channels in physiology, using zebrafish as amodel for the associated diseases.

AntonellaGradogna received her PhD in “Biosciences and Biotechnology – Biochemistry and Biophysics” from the University of Padua in Italy. She is currently post-doc at the Biophysics Institute of the Italian Research Council in Genoa, Italy. She is interested in the biophysical and pharmacological analysis of ion channels and transporters. Her studies are mainly focused on the family of CLC chloride channels and chloride/proton antiporters.

Ilaria Zanardi studied for her PhD in Biology at the University of York within a Marie Curie European network project. During her first post-doc in Michael Pusch’s lab in Genoa she developed an optical assay of the transport activity of ClC-7. She is also very involved in science communication.

Michael Pusch received his PhD in physics from the Göttingen University, held post-doc positions in Genoa, Italy and Hamburg, Germany. He is currently Director of the Biophysics Institute of the Italian Research Council in Genoa, Italy. His major interest is the experimental study of biophysical mechanisms of ion channels and transporters, many of which are involved in human genetic diseases. The favored object of study is the family of CLC chloride channels and chloride/proton antiporters.

Raúl Estévez studied Biochemistry (1994) at the University of Barcelona, where he also obtained his PhD in Biochemistry (2000, under the supervision of Dr. Manuel Palacín), finishing both studies with extraordinary honours. He also holds an Interuniversitary Master of Business Administration (1999). Following three years of postdoctoral research at the Center for Molecular Neurobiology under the supervision of Dr. Thomas Jenstch and four years as a Ramón y Cajal researcher, he started to work at the University of Barcelona as Professor of Physiology in the Faculty of Medicine in 2007. Dr. Estévez is member of CIBERER. His research is now focused on rare genetic diseases related to movement of chloride across membranes.

Additional information

Competing interests

None declared.

Funding

This work was supported by Telethon Italy (grant GGP 12008 to M.P.), the Compagnia San Paolo (M.P.), SAF 2012-31486 (R.E.), ELA 2012-014C2B (R.E.), AFM Telethon 2012-16305 (R.E. and A.B.G.) and 2014SGR1178 (R.E. and A.B.G.). R.E. was funded by an award within ERA-NET E-RARE-2 framework by ISCIII through CIBERER and by an Icrea Academia Prize. A.B.G. is a Serra-Húnter Fellow.