FEBS Letters 584 (2010) 2093–2101

journal homepage: www.FEBSLetters.org

Review The enigma of the CLIC : Ion channels, redox proteins, enzymes, scaffolding proteins?

Dene R. Littler a,1, Stephen J. Harrop a, Sophia C. Goodchild b, Juanita M. Phang a, Andrew V. Mynott a, Lele Jiang c, Stella M. Valenzuela d, Michele Mazzanti e, Louise J. Brown b, Samuel N. Breit c, Paul M.G. Curmi a,c,* a School of Physics, University of New South Wales, Sydney, NSW 2052, Australia b Department of Chemistry and Biomolecular Sciences, Macquarie University, Sydney, NSW 2109, Australia c St. Vincent’s Centre for Applied Medical Research, St. Vincent’s Hospital and University of New South Wales, Sydney, NSW 2010, Australia d Department of Medical and Molecular Biosciences, University of Technology Sydney, Sydney, NSW 2007, Australia e Dipartimento di Scienze Biomolecolari e Biotecnologie, Universita’ degli Studi di Milano, Via Celoria 26, 20133 Milano, Italy article info abstract

Article history: Chloride intracellular channel proteins (CLICs) are distinct from most ion channels in that they have Received 18 December 2009 both soluble and integral membrane forms. CLICs are highly conserved in chordates, with six verte- Revised 13 January 2010 brate paralogues. CLIC-like proteins are found in other metazoans. CLICs form channels in artificial Accepted 13 January 2010 bilayers in a process favoured by oxidising conditions and low pH. They are structurally plastic, with Available online 19 January 2010 CLIC1 adopting two distinct soluble conformations. Phylogenetic and structural data indicate that Edited by Wilhelm Just CLICs are likely to have enzymatic function. The physiological role of CLICs appears to be mainte- nance of intracellular membranes, which is associated with tubulogenesis but may involve other substructures. Keywords: Chloride intracellular channel Ó 2010 Federation of European Biochemical Societies. Published by Elsevier B.V. All rights reserved. Redox Membrane insertion Tubulogenesis CLIC1

1. Introduction function, however, our understanding of their function is still very incomplete. The chloride intracellular channel (CLIC) proteins form a class of The history of the discovery of CLIC proteins has skewed the sci- intracellular anion channels that do not fit the paradigm set by entific community’s view of their behaviour and has directed our classical proteins. Their properties are enigmatic as collective experiments towards characterising ion channel activity. they can exist as both soluble globular proteins and integral mem- This is important as the CLIC proteins themselves differ in many brane proteins with ion channel function. To compound this struc- ways from ‘normal’ ion channels and their soluble structure does tural gymnastics, human CLIC1 adopts at least two stable soluble not immediately suggest such a function. Thus, as we learn more structures, with redox status controlling the transition between about CLIC proteins, we also learn more about the boundaries of them thus making CLIC1 a member of the rare category of ‘‘meta- what is possible in protein science and cell biology. morphic proteins” [1]. The high level of conservation of CLIC pro- teins in vertebrates argues for a significant, conserved biological 2. What are CLIC proteins?

Abbreviations: CLIC, chloride intracellular channel; GST, glutathione S-transfer- ase; GSH, glutathione; GSTO, GST omega; TM, transmembrane; Grx, glutaredoxin; The CLIC proteins are characterised by the presence of a 240 ERM, ezrin–moesin–radixin; DHAR, dehydroascorbate reductase; ER, endoplasmic residue CLIC module whose structure belongs to the glutathione reticulum S-transferase (GST) fold superfamily (Fig. 1A). The classical GSTs * Corresponding author. Address: School of Physics, University of New South are cytosolic proteins catalysing the conjugation of the tripeptide Wales, Sydney, NSW 2052, Australia. Fax: +61 2 9385 6060. glutathione (GSH) to electrophilic centres in unwanted endoge- E-mail address: [email protected] (P.M.G. Curmi). 1 Present address: Biota Structural Biology Laboratory, St. Vincent’s Institute, nous or xenobiotic molecules [2]. The first GSTs to be studied were Princes Street, Fitzroy, Victoria 3065, Australia. promiscuous enzymes discovered due to their role in xenobiotic

0014-5793/$36.00 Ó 2010 Federation of European Biochemical Societies. Published by Elsevier B.V. All rights reserved. doi:10.1016/j.febslet.2010.01.027 2094 D.R. Littler et al. / FEBS Letters 584 (2010) 2093–2101

Fig. 1. Structure of CLIC proteins. (A) The structure of CLIC1 in the soluble, reduced monomeric form coloured by secondary structure (helices in red, strands in yellow, loops in green). The putative TM region is shown in cyan (residues 25–46). The N-domain is on the left (b-sheet plus helices h1, h2 and h3) and the all-helical C-domain is on the right (helices h4–h9). The three cysteines (Cys24, Cys178 and Cys223) that are conserved in all vertebrate CLIC proteins are shown as CPK models. (B) The crystal structure of the oxidised CLIC1 dimer. The left hand side subunit is coloured yellow, while the right hand side subunit is coloured and oriented as per the CLIC1 monomer in panel (A). (C) A proposed model for the transition between the soluble form of CLIC1 and the integral membrane ion channel form. (D) Glutathione (stick model) covalently bound to CLIC1, where the molecular surface of CLIC1 in the vicinity of the binding site is shown. Orientation is similar to panel (A).

metabolism, however, the genomic era has recently expanded the The degree of conservation between the six vertebrate CLIC par- number of proteins known to be part of the GST fold superfamily alogues suggests that they arose from rounds of duplication from a yielding a number of non-classical members whose functions are single ancestral protein in an ancient chordate. This hypothesis is not yet clear. supported by the fact that the urochordate Ciona intestinalis has a The CLICs differ from the classical enzymatic GSTs in that they single CLIC protein that is highly homologous to the vertebrate par- have an active site cysteine residue (Cys24 in human CLIC1), which alogues (45% sequence identity), with which it also shares the is rendered reactive by the protein itself. In contrast, the classical three conserved cysteine residues. GSTs activate the thiol group of the GSH, which binds non-cova- CLIC-like proteins are also present in invertebrates (Fig. 2A). The lently within their active sites with very high affinity. The location sequences of these proteins are quite divergent from the vertebrate of the GSH thiol in classical GSTs is approximately equivalent to CLICs (35% sequence identity), although they still usually con- the location of the active site cysteine in CLIC proteins. All CLICs serve the three characteristic cysteines. In some nematodes, the ac- have two additional cysteines that are conserved (Cys178 and tive site cysteine has been replaced by aspartate (Caenorhabditis Cys223 in CLIC1, Fig. 1A). elegans CLIC-like proteins EXC-4 and EXL1). The CLIC proteins are highly conserved in vertebrates, which The molecular, cellular and physiological function of the CLIC usually possess six distinct paralogues (CLIC1–CLIC6) (Fig. 2A). proteins is still being unravelled. Different experiments have at- There is some variation in the number of paralogues with the tel- tempted to address each of these levels of CLIC function. At the eost (bony ray-finned) fish having a second copy of CLIC5 (CLIC5L) molecular level extensive biochemical, biophysical and structural while birds and lizards lack CLIC1. Moreover, several of the CLICs work has been performed to characterise the soluble form of the also have splice variants (CLIC2, CLIC5 and CLIC6). CLIC5B and individual CLIC proteins. The membrane inserted form of the CLICs CLIC6 have an additional domain N-terminal to the CLIC module. has largely been probed indirectly through electrophysiological These domains often include repetitive sequences that are poorly experiments. These have included the electrophysiological conserved in both sequence and size. response of purified CLIC proteins incorporated into artificial D.R. Littler et al. / FEBS Letters 584 (2010) 2093–2101 2095

Fig. 2. Phylogenetic analysis of CLIC proteins. (A) A neighbour joining phylogenetic tree of vertebrate CLIC and in vertebrate CLIC-like proteins. Clades are colour coded as per the Key. (B) A neighbour joining phylogenetic tree of proteins having a GST fold and an active site cysteine. Lineages are colour coded by domain or kingdom (Key on the top right). The colour of the circle arcs indicates whether the state of the purified proteins is monomeric (red), dimeric (green) or unknown (violet; Key bottom right). Consensus sequence patterns that structurally align with the putative CLIC active site are given for each clade with cysteines coloured magenta.

membranes, patch clamping and whole cell current experiments a number of different knock-out organisms. A C. elegans mutant performed on endogenous channels as well as those produced deficient in one of its two CLIC-like proteins, EXC-4, has proved upon overexpression of CLIC paralogues in mammalian cells. Final- particularly informative in this regard due to the clarity of its phe- ly, the physiological characterisation of the CLICs relies heavily on notype [3,4]. Knock out mice with more subtle phenotypes have 2096 D.R. Littler et al. / FEBS Letters 584 (2010) 2093–2101 also been generated for CLIC1 [5,6], CLIC4 [5,7] and CLIC5 [8]. How- sequence analysis indicating a putative transmembrane (TM) do- ever, a clear picture of CLIC function in vertebrates is clouded by main near the N-terminus (a region shared with p64 and CLIC1). the likely redundancy arising from having six paralogues. Proteolysis studies showed that the protein made a single passage In the following sections, we wish to examine the current status across the ER membrane with a larger, C-terminal portion remain- of our understanding of CLIC function. The history and order of dis- ing in the cytoplasm. The model resulting from the proteolytic covery has been important in shaping our view of this function and studies was consistent with the sequence analysis in locating the hence the sections are in approximate historical order. When look- TM domain. ing at this evolution of discovery, one is reminded of the Parable of The CLIC2 was discovered in the same period, by an anal- the Blind Men and the Elephant: ysis of transcripts in chromosomal region Xq28 [17] and with it, the acronym ‘‘CLIC” for chloride intracellular channel was coined ‘‘Once upon a time there was a certain raja who called to his and all proteins renamed as CLICs. This nomenclature strengthened servant and said, ‘Come, good fellow, go and gather together the view that CLIC proteins function as intracellular chloride chan- in one place all the men of Savatthi who were born blind... nels even though the data at the time was rather sparse. and show them an elephant.’” (Buddhist version from the Udana 68–69). 4. The conundrum of the CLIC structure: is it a channel or a channel modulator? 3. Are CLIC proteins ion channels? The sequences of the CLIC modules do not fit the patterns of Discovery of ion channel function of CLIC proteins is intimately other, well-characterised ion channel proteins that show clear linked to the original discovery of the CLICs themselves. If it were and often multiple TM domains. A weak sequence identity not for the historical sequence of events, the CLIC channel function (15%) was detected between CLIC1 and the then, newly discov- would likely have been missed for a long time. The Al-Awqati ered, GST omega (GSTO) protein [18]. Despite the low sequence group was searching for a chloride ion channel that may have been identity, a plausible homology model was constructed which the then illusive cystic fibrosis channel. They functionally charac- showed significant sequence conservation around the GST GSH terised a chloride ion channel in bovine kidney in part by designing binding site. This structural conservation included a Cys-Pro- a series of small molecule inhibitors based on the ion channel mod- Phe motif in the GSH site, which was found in GSTO but not ulator ethacrynic acid [9]. One of these compounds, indanyloxy- in other GST proteins. In addition, this work showed that GSTO acetic acid 94 (IAA94), proved to be a moderate inhibitor of a could modulate the calcium release channel activity of the ryan- in bovine kidney cortex membrane vesicles and odine receptor [18], a property which was later shown to be it was used to identify and purify a protein, p64, as a bovine chlo- shared with CLIC2 [19]. This raised the question as to whether ride channel [10]. This channel was localised to apical and intracel- the CLIC proteins were actually GST family members that regu- lular membranes [11]. The cloning of p64 revealed a protein lated ion channel proteins rather than forming ion channels sequence that did not resemble an integral , themselves. let alone any known ion channel family, yet it appeared able to The crystal structure of CLIC1 followed shortly thereafter and reconstitute chloride channel activity [11,12]. This p64 protein indeed showed that CLIC proteins are part of the GST fold super- would later be called bovine CLIC5B. family (Fig. 1A) [20]. However, CLIC1 did not appear to bind GSH The next major advance in analysing the ion channel function of non-covalently. A covalent complex with GSH forming a mixed CLIC proteins was the discovery of proteins that comprised only a disulphide bond with Cys24 of CLIC1 was crystallised, but even single CLIC module with no N-terminal extension as was seen in in this covalent complex, the GSH was poorly ordered and did p64. CLIC1 [13] and CLIC4 [14,15] were the first such minimal CLIC not make extensive non-covalent interactions with CLIC1 proteins to be cloned and their function characterised. Had these (Fig. 1D) [20], although it did bind in the same site as does GSH proteins been discovered before the p64 work, it is unlikely that to classic GST proteins [2]. they would have been immediately linked to ion channel function. The GST fold comprises two domains: a thioredoxin domain and This is particularly true given that their sequences predicted a an all-helical C-terminal domain (Fig. 1A). The origin of the C-ter- globular protein with no clear signature membrane spanning do- minal domain is not known, however, the thioredoxin fold of the mains. Despite this, the channel function of both CLIC1 and CLIC4 N-terminal domain is shared by many large families of redox-ac- was explored immediately because of their homology to p64 [12]. tive proteins. The active site of many thioredoxin-domain proteins The initial work on CLIC1 suggested localisation to the cell nu- is centred on a redox-active cysteine, which is often the first resi- cleus and cytoplasmic organelles, hence it was originally called due in a Cys-X-X-Cys motif at the N-terminus of an alpha helix. The NCC27 (nuclear chloride channel-27) [13]. Electrophysiological vertebrate CLIC proteins have a similar motif: Cys-Pro-(Phe/Ser)- experiments were performed on transfected CHO-K1 cells express- (Ser/Cys) at the N-terminus of helix h1, which resembles the glut- ing CLIC1 showing that the protein was associated with chloride aredoxin (Grx) motif Cys-Pro-(Phe/Tyr)-(Cys/Ser) [20]. This Grx- ion channel activity. The channel activity was further characterised like active site is absent in all classic GST proteins [2], with the by stably transfecting cell lines with CLIC1 bearing either an N- or notable exception of the omega GSTs [21]. C-terminal FLAG epitope tag. Electrophysiological studies with an The crystal structure of CLIC1 firmly raised the question as to antibody to the FLAG epitope indicated that CLIC1 spanned the whether the CLIC proteins per se can form integral membrane plasma membrane an odd number of times with the C-terminus ion channels. First, the structure shows a stable, globular protein in the cytoplasm of the stably transfected cells [16]. This work con- that is a member of a large superfamily of stable, globular proteins firmed that CLIC1 was indeed an integral membrane protein asso- that act as enzymes. Second, the putative TM region of the CLIC ciated with the channel activity. channel (as identified by sequence analysis and proteolytic exper- At about the same time as the work on CLIC1, CLIC4 (initially iments [15]) forms an integral part of the N-domain thioredoxin called p64H1) was also identified and characterised [14] and ion fold (helix h1 through to b-strand s2, coloured cyan in Fig. 1A). channel activity was measured in microsomes from the endoplas- Hence, it would require a large-scale structural transition for a CLIC mic reticulum (ER) obtained from HEK-293 cells expressing CLIC4 to enter the membrane and utilise the putative TM region to [15]. The protein was shown to be a membrane protein with anchor it in the lipid bilayer. This would involve the complete D.R. Littler et al. / FEBS Letters 584 (2010) 2093–2101 2097 disruption of the N-domain thioredoxin fold, which at the time was 5. Are CLIC proteins redox controlled? thought to be relatively stable. To directly demonstrate that CLIC1 proteins are channels per The initial structure of CLIC1 clearly showed that CLIC proteins se rather than modulators of some other, real integral membrane are related to the thioredoxin fold family, in particular, they have a ion channel protein, it was essential to show that pure recombi- Grx-like active site centred on a cysteine residue (Cys24 in human nant, soluble globular CLIC proteins could generate the chloride CLIC1) [20]. This begged the question as to whether GSH or redox channel activity in an artificial bilayer made of synthetic lipids. in general, was involved in CLIC function. To date, the exploration To this end, several groups have shown that when soluble, recom- of GSH interactions has not been illuminating however, several binant CLIC1, CLIC2, CLIC4 or CLIC5 proteins are added to artifi- CLIC proteins have been shown to be redox activated. cial, synthetic lipid bilayers, they integrate into the bilayer and The channel activity of all CLICs with a cysteine residue at the electrophysiological measurements detect reproducible ion chan- centre of the Grx-like site has been shown to be redox dependent nel activity [20,22–30]. In the case of CLIC1, it has been shown [25–28,31]. To distinguish between regulation of the channel ver- that adding purified protein to a buffer bath results in its move- sus the transition from the soluble form CLIC to the integral mem- ment to and integration into the lipid bilayer. Further a single brane form, surface plasmon resonance experiments were used to point mutation in CLIC1 of the active site cysteine residue monitor binding of CLIC proteins to lipid bilayers. This process has (C24A) in the putative TM region have resulted in alterations in been shown to be dependent on oxidising conditions for all CLIC the electrophysiological characteristics of the channels [27]. The proteins, which possess a cysteine at the reactive site CLIC1 channel conductance for this mutant was reduced and no [25,26,28,31,33]. The C. elegans CLIC-like protein EXC-4 has an longer trans-redox sensitive. Finally, CLIC-like proteins from the aspartic acid residue at this site [3,4] and hence binding to a lipid invertebrates C. elegans and Drosophila melanogaster, EXC-4 and bilayer is not redox dependent [31]. Mutation of CLIC1 Cys24 to DmCLIC also show the ability to form ion channels in artificial alanine abrogated redox sensitivity of the channel [27]. Thus, oxi- bilayers [31]. dising conditions and the presence of the reactive site cysteine ap- The electrophysiological characterisation of the individual CLIC pear to favour the transition from the soluble, globular form to a channels is far from complete. The best studied CLIC channel is hu- membrane bound form for most CLICs, however, redox is not likely man CLIC1 which has been characterised independently by three to be the only mechanism underpinning this transition. research groups [13,16,20,22,23,25,27,32]. CLIC1 forms a poorly selective anion channel where the conductance is dependent on 6. Redox controls dramatic structural transitions in CLIC1 – the anion concentration. The consensus anion permeability sequence acrobat uncovered À À À 2À À is: SCN > F > Cl > NO3 > HCO3 > acetate at 110–140 mM an- ion concentration [13,27] although this is concentration dependent Exploration of the effect of oxidation on CLIC1 led to the discov- À and the two reports differ in their positioning of I . With a sym- ery of a reversible transition between a reduced, soluble mono- metric solution of 140 mM KCl, the main single channel conduc- meric state and an oxidised, soluble, non-covalent dimeric state tance is approximately 30 pS (values ranging from 22 to 38 pS) [25]. The crystal structure of this dimeric state shows that CLIC1 [13,20,23,25,27]. These channels show the presence of substates has undergone a dramatic structural change, where nearly the with conductances of approximately 8, 14 and 21 pS, indicating whole N-domain (with a thioredoxin fold) has altered its second- that the channel consists of four coupled pores [23,27]. When re- ary and tertiary structure (Fig. 1B). The four-stranded mixed b- combinant CLIC1 is added to artificial bilayers, 7–8 pS conduc- sheet that is characteristic of the thioredoxin fold has been com- tances with long mean open times (250 ± 34 ms) appear prior to pletely transformed into helices and loops. The discovery of the the mature 30 pS conductance with a 30 ms mean open time and CLIC1 transition, along with five other proteins that can switch be- these may represent primitive pores that assemble to form a ma- tween two dramatically different three-dimensional structures, led ture, tetrameric assembly [23]. Channels with multiples of the to the coining of the term ‘‘metamorphic proteins” to describe pro- 30 pS conductance are also observed in vitro [22,23]. teins with more than one stable tertiary structure [1]. This class is Recent electrophysiological characterisations of channels likely to grow in numbers now that its existence has been formed by recombinant CLIC4 and CLIC5 in artificial bilayers have established. questioned the notion that CLICs are specific chloride or indeed an- The physiological relevance of the oxidised dimeric state of ion specific channels [29,30]. Each of these studies shows that the CLIC1 is unclear, particularly since this state has only been ob- channels formed by these CLICs are almost equally permeable by served for CLIC1. The non-covalent dimer interface is formed by + À K and Cl . Thus, CLIC4 and CLIC5 form poorly-selective ion chan- residues from the N-domain and it comprises a flat, hydrophobic nels which do not discriminate between cations and anions. surface. It has been proposed that in vivo, such a surface would in- Although this is incompatible with the etymology of the word sert into a lipid bilayer rather than form a protein dimer, thus, the ‘‘CLIC”, the nomenclature is unlikely to be revised. half dimer structure may represent the membrane docking form of Given the sum of the experimental data, it appears beyond CLIC1 [25]. In any case, the observed structural transition includes doubt that the individual CLIC proteins can form integral mem- the putative TM domain (coloured cyan in Fig. 1A and B) and hence brane ion channels in vitro. The properties of these in vitro chan- it shows that this region is capable of dramatic structural rear- nels match those observed in vivo when CLIC proteins are rangements, lending support to its role as the TM domain in the overexpressed. Controversy still remains regarding what part this channel form of CLIC proteins (Fig. 1C). If the half dimer structure ion channel activity plays in the in vivo physiological function of represents the membrane docking form, an additional structural the CLIC proteins. change is required to integrate the TM domain into the membrane. The observation that CLIC proteins can adopt both a soluble, This is likely to be followed by oligomerisation to form the active globular fold and an integral membrane form puts them into a ion channel, as shown in the model for channel formation (Fig. 1C). wider class of proteins that include bacterial toxins and intracellu- lar proteins such as the Bcl family proteins. All of these proteins undergo dramatic structural changes so as to transit between the 7. Is the transition to the channel state regulated by pH? soluble and membrane forms. In each case, this transition appears to be tightly regulated. The question is: what controls the struc- Many studies of CLIC protein chloride ion channel activity have tural transition in CLIC proteins? noted a strong pH dependence [22,23,26,28,31]. These include 2098 D.R. Littler et al. / FEBS Letters 584 (2010) 2093–2101 human CLIC1, CLIC2, CLIC4, DmCLIC and EXC-4. In each case, the theme for some of the characterised clades of GST fold proteins channel activity is minimal around neutral pH and rises rapidly with a Cys-Pro-[aromatic]-[Ser/Cys/Ala/Val]-X-Arg active site mo- as the pH is reduced. Some experiments have also shown an in- tif is that they are enzymes that couple GSH to a redox reaction. crease in activity at higher pH for CLIC1 [22] and a smaller increase The secondary target substrate may be a small molecule or a bio- for CLIC4 [26]. Unfolding experiments [34] and hydrogen deute- logical macromolecule. Based on this, we speculate that the CLIC rium exchange [35] on the soluble form of CLIC1 indicate that proteins may also have an enzymatic component to their activity low pH primes the N-domain for membrane insertion. The physio- and it is likely to involve a GSH-related cofactor. Indeed it has been logical significance of the link between channel activity and proton shown in CLIC4 that mutating residues in either the GSH-like bind- concentration, most prominent at low pH is not clear, but it may be ing site or the secondary substrate binding site abrogates the abil- linked with conditions within cytoplasmic vesicles (such as endo- ity of the protein to localise towards the plasma membrane somes) or it may imply a coupling of CLIC channel activity with following LPA stimulation [41]. vesicle acidification by a proton pump. Adjacent to the putative GSH site in the CLIC protein structures, is a long groove or slot (Fig. 1D) [20]. The crystals structure of the 8. Are CLIC proteins also enzymes? covalent complex between CLIC1 and GSH shows that the GSH group is poorly ordered and makes few contacts with CLIC1, this The Grx-like active site motif Cys-Pro-Phe-(Ser/Cys) located at is because it lies along one edge of this open slot (Fig. 1D) [20]. the N-terminus of helix h1 in the N-domain with a thioredoxin fold We speculate that this slot is the binding site for an extended mac- looks suspiciously like the active site of an enzyme that possibly romolecular chain, probably a polypeptide. Binding of such a poly- utilises a cofactor related to GSH for its activity. The CLIC structures peptide could complete the GSH site and hence convert it to a show the position of the cysteine sulfhydryl is in a basic environ- strong binding site. In the crystal structures of CLIC4 [26,36] and ment, forming hydrogen bonds to the backbone amides of helix CLIC2 [28,37] internal peptide loops have been observed in the h1 making it likely to be a reactive group [20,26,28,31,36,37].A vicinity of this slot demonstrating in theory that it is able to incor- similar site in the GSTO proteins [21] suggests that this ‘‘active porate such molecules. site” may be a more general feature of GST fold proteins. Extensive BLAST searches were performed using the CLIC se- quences as queries. The resultant sequences were aligned with 9. Are the invertebrate CLIC-like proteins enzymes too? CLUSTAL W and filtered so as only to include sequences with cys- teine in the vicinity of the putative active site (although this only Although the invertebrate CLIC-like proteins cluster in a removed the most distantly related sequences). A neighbour join- neighbouring clade to the vertebrate CLICs, they appear to have ing phylogenetic tree was constructed (Fig. 2B). a distinct active site. First, the consensus sequence for the The CLIC proteins form a well-defined clade that subdivides into invertebrate CLIC-like proteins is: [Cys/Asp]-Leu-Phe-Cys-Gln- the vertebrate CLIC proteins and the invertebrate CLIC-like proteins Glu. Second, the crystal structures of both EXC-4 and DmCLIC with the urochordate, C. intestinalis CLIC separating these groups. show that there is a metal ion bound to the protein in what The clade most closely related to the CLICs is that for plant dehy- would be part of the GSH site [31]. Although initially assigned 2+ droascorbate reductase (DHAR) proteins [38], where one of the as a Ca ion, it has subsequently been identified as a potas- Arabidopsis thaliana DHARs, AtDHAR1, has been shown to have sium ion [Mynott, unpublished data]. These two features indi- CLIC-like properties [39]. Beyond this, there appear many other cate that the invertebrate CLIC-like proteins may have evolved well-defined clades of proteins with a GST fold and a cysteine in away from the putative enzymatic function of the vertebrate the active site region. CLICs. Using representative crystal structures from all characterised clades, a precise location of the putative active site cysteine was obtained by structural superposition with the CLIC structures. 10. Are CLICs scaffolding proteins coupling the membrane to The structure-based alignment of the motif corresponding to the cytoskeleton? Cys-Pro-Phe-Ser in CLIC1 divided the clades into two classes. Those with cysteines not coinciding with the CLIC site include: GSTB CLIC proteins are localised to membranes and as such may (bacterial GSTs), GSTZ (maleylacetoacetate isomerase), GSTD (in- play a role in maintaining the structure of intracellular organ- sect GSTs) and GSTF (plant GSTs) (Fig. 2B). elles via interactions between the membrane and the cytoskel- The remaining clades, including the CLICs and omega class eton. The disruption of EXC-4, the CLIC-like protein in C. GSTs (GSTO), usually share a consensus site that fits the generic elegans, gives a poignant example in that the phenotype shows motif: Cys-Pro-[Phe/Tyr/Trp]-[Ser/Cys/Ala/Val]-X-Arg (although the tubular, intracellular excretory membrane system trans- we note that the invertebrate CLIC-like proteins deviate from this formed into large cystic vesicles [3]. Indeed, many CLICs are motif). Most of the clades are made up of proteins that have not thought to interact directly with the actin cytoskeleton been functionally characterised. Those with some characterisation [8,24,30,41–45] or via scaffolding proteins such as the ezrin– include: bacterial Grx 2, plant DHARs [38,39], GSTOs [21], bacte- moesin–radixin (ERM) proteins ezrin [24,42] and radixin [8,46] rial SspA, plant GST lambda (In2) [38] and yeast Gto [40]. Most of and AKAP350 [44,47]. CLIC proteins have also been reported these are monomeric (indicated by a red coloured circle arc in to bind to the cytoplasmic C-terminal tail regions of the seven Fig. 2B), whereas GSTs (including GSTO) are usually dimeric with TM helix GPCRs including: CLIC4 to human rhodopsin [48] and the GSH site formed at the dimer interface [2]. Although some of the histamine H3 receptor [49] and CLIC6 to the dopamine these protein families have been labelled ‘‘GST” because of struc- D(2)-like receptors [46]. Unravelling these protein–protein inter- tural or sequence similarity, many of them have not yet been actions has proved problematic in that the pull-down experi- shown to possess GST activity and may not be bona fide GST en- ments inevitably trap actin and hence a plethora of actin- zymes. This is also an example of where the historical order of binding and actin associated proteins. Higher resolution meth- discovery prejudices our research. ods will be required to map the proteins that interact directly So what do these proteins do, in particular, what does it say with the CLICs and hence determine how these are related to about the function of vertebrate CLIC proteins? The common membranes and the cytoskeleton. D.R. Littler et al. / FEBS Letters 584 (2010) 2093–2101 2099

11. What have we learnt about CLIC physiological function from defect is in the inner ear hair cell stereocilia, where the lack of gene disruption experiments? CLIC5 in the basal region of the hair bundle causes stereocilia to de- grade, which results in deafness and loss of balance and coordina- The gene disruption and subsequent rescue of the CLIC-like exc- tion. The exact role of CLIC5 in maintaining these structures is 4 in C. elegans was ground breaking. The initial study linked EXC-4 unclear. One possibility is that it interacts with the ERM protein with the formation and maintenance of the intracellular, tubular radixin, which co-localises to the same region of the cell and is esti- excretory vesicle [3]. It showed that the N-terminal 66 residues mated to be in equimolar stoichiometry [8]. Radixin is concen- of EXC-4, a region that includes the putative TM region, were trated at the base of the stereocilia to the region where they join essential for localisation to the lumenal intracellular apical mem- the body of the hair cell and it is thought to link the base of the ac- brane. Subsequent work showed that other invertebrate CLIC-like tin bundle of the stereocilium to the surrounding plasma proteins, EXL1 (C. elegans) and DmCLIC (D. melanogaster) could also membrane. rescue the phenotype and restore function, while vertebrate CLICs ‘‘When the blind men had each felt a part of the elephant, the could not [4]. Chimeras of GST fold proteins with the first 66 resi- raja went to each of them and said to each: ‘Well, blind man, dues of EXC-4 also restored function, indicating that this was the have you seen the elephant? Tell me, what sort of thing is an critical region for maintenance of the integrity of the intracellular elephant’” (Buddhist version from the Udana 68–69). vesicle. The link between EXC-4 and tubulogenesis inspired several studies linking vertebrate CLIC proteins with tubulogenesis, specif- 12. So where are we at in understanding CLIC function? ically CLIC4 with angiogenesis. The current model for vertebrate tubulogenesis is called cord hollowing and it involves formation In terms of defining the CLIC protein family (i.e. determining and fusion of intracellular vesicles/vacuoles in endothelial cells which proteins are really CLICs), the vertebrate and chordate CLIC which then fuse across cell boundaries resulting in vascular lumen proteins are easily identified because there is a fixed set of para- formation. The first link between angiogenesis and CLIC4 resulted logue lineages. This restricted divergence to produce a fixed set from proteomic work followed by antisense RNA to transiently of paralogues and subsequent conservative evolution of each para- knockdown CLIC4 in immortalised microvascular endothelial cells logue lineage suggests that CLIC function is central and needs to be [45]. Subsequent RNAi studies in stably transfected primary endo- tightly controlled. When we move to invertebrates, it is not com- thelial cells show that CLIC4 knockdowns are defective in angio- pletely clear that the CLIC-like proteins are bona fide CLICs (i.e. genesis, in particular, in network formation, capillary sprouting equivalent to the vertebrate CLICs). They may have distinct or addi- and lumen formation using in vitro assays [50]. tional functions. This confusion arises because of the similarity of Initial reports of CLIC1À/À [5,6] and CLIC4À/À [5,7] mice indicate CLICs and other members of the large GST fold superfamily. Out- that they have minimal apparent phenotype in unstressed labora- side the metazoa, there are many clades of GST fold proteins with tory conditions. CLIC1À/À mice show a mild platelet dysfunction a Grx-like active site containing a cysteine and these may or may characterised by prolonged bleeding times and decreased platelet not be functionally related to CLICs. activation in response to adenosine diphosphate stimulation linked In terms of their molecular function in vitro, it is clear that all À/À to P2Y12 receptor signalling [6]. CLIC4 mice have a higher rate of CLIC proteins and the invertebrate CLIC-like proteins that have still births than wild type [7]. been studied to date exist as both soluble globular proteins and Defects in angiogenesis were discovered in CLIC4À/À mice using integral membrane proteins that possess ion channel activity. the matrigel assay and a model for oxygen-induced retinopathy The transition between these two states is influenced by pH and [7]. Cell culture experiments show that cells from CLIC4À/À are redox conditions in most instances. We speculate here that the defective in endothelial cell tubulogenesis. Measurements of vesic- CLICs are also likely to possess an enzymatic function that involves ular pH show that there is less acidification in large intracellular redox with the participation of a GSH analogue. This activity may vesicles in endothelial cells derived from CLIC4À/À mice compared or may not be linked to membrane binding and the ion channel with wild type, while small intracellular vesicles show the same function. pH for both CLIC4À/À and wild type. The large intracellular vesicles In terms of cellular function, the CLICs and CLIC-like proteins are on the tubulogenic pathway, while the small vesicles are lyso- appear to be associated with intracellular vesicular membranes. somes/endosomes. These results suggest that CLIC4 (possibly via Vertebrate CLICs seem to show both membrane bound and soluble its channel function) may assist the vacuolar proton ATPase in forms whereas in C. elegans, EXC-4 appears to be solely bound to acidification of large vesicles during tubulogenesis. Clearly, angio- the intracellular lumenal membrane. The current view is that the genesis still occurs in these CLIC4À/À mice. This may be due to CLICs are necessary for the formation and maintenance of intracel- redundancy in the CLICs, in particular the ubiquitous presence of lular membrane-enclosed vesicles. It is still unclear as to how this CLIC1. cellular function is achieved, whether the ion channel function CLIC4À/À mice differ from wild type in collateral circulation maintains the osmotic balance between the vesicle and the cyto- with CLIC4À/À but not CLIC1À/À mice having fewer arterial collater- plasm or whether CLICs maintain the correct interactions between als in skeletal muscle and in brain [5]. Collateral remodelling in the the membrane and the cytoskeleton. Low pH inside some of the period after arterial ligation appears to proceed normally in the vesicles may also be important, where CLICs may balance the ac- CLIC4À/À mice, although the level of CLIC1 expression increases sig- tion of proton pumps. The pH dependence of CLIC channel activity nificantly. This may indicate that CLIC1, which is ubiquitously ex- may be important in this function. pressed, compensates for the lack of CLIC4. It is speculated that In terms of physiological function in whole organisms, we are CLIC4 may be involved in tubulogenesis events during collateral just beginning to understand the role of the CLIC proteins. The formation [5]. EXC-4 phenotype indicated that they may be important in tubulo- Study of the CLIC5 knockout mouse was not initially prejudiced genesis. In line with this, CLIC4 may be important for angiogenesis by preconceptions of CLIC function, as the knockout occurred due and the development and maintenance of the arterial collateral to a spontaneous, recessive mutation in and had a clear phe- system. CLIC5 appears to play a special role in the stabilisation of notype [8]. Known as ‘‘jitterbug”, these mice lack coordination and cochlear stereocilia where may work with the ERM protein radixin progressively become deaf. They also show lung damage. The main to couple the actin bundle in the stereocilium to the surrounding 2100 D.R. Littler et al. / FEBS Letters 584 (2010) 2093–2101 plasma membrane. The ubiquitous tissue localisation of CLIC1 may [2] Wilce, M.C. and Parker, M.W. (1994) Structure and function of glutathione S- indicate that it plays a more general physiological role. transferases. Biochim. Biophys. Acta 1205, 1–18. [3] Berry, K.L., Bulow, H.E., Hall, D.H. and Hobert, O. (2003) A C. elegans CLIC-like protein required for intracellular tube formation and maintenance. Science 13. Future? 302, 2134–2137. [4] Berry, K.L. and Hobert, O. (2006) Mapping functional domains of chloride intracellular channel (CLIC) proteins in vivo. J. Mol. Biol. 359, 1316–1333. On the molecular front, it is essential that any putative enzy- [5] Chalothorn, D., Zhang, H., Smith, J.E., Edwards, J.C. and Faber, J.E. (2009) matic function of the CLIC proteins is revealed and characterised. Chloride intracellular channel-4 is a determinant of native collateral formation The conservation of the active site sequence and structure argues in skeletal muscle and brain. Circ. Res. 105, 89–98. [6] Qiu, M.R. et al. (2009) Generation and characterization of mice with null that such a function exists. There simply remains the question of mutation of the chloride intracellular channel 1 gene. Genesis 1, 1–10. finding out what the potential substrate may be and how it relates [7] Ulmasov, B., Bruno, J., Gordon, N., Hartnett, M.E. and Edwards, J.C. (2009) to the workings of the cell. Determining how any identified enzy- Chloride intracellular channel protein-4 functions in angiogenesis by supporting acidification of vacuoles along the intracellular tubulogenic matic function relates to ion channel formation in cells will also be pathway. Am. J. Pathol. 174, 1084–1096. critical for understanding CLIC function. [8] Gagnon, L.H., Longo-Guess, C.M., Berryman, M., Shin, J.B., Saylor, K.W., Yu, H., In cells, the first question that needs to be addressed is in what Gillespie, P.G. and Johnson, K.R. (2006) The chloride intracellular channel protein CLIC5 is expressed at high levels in hair cell stereocilia and is essential way the ion channel activity of CLICs is related to their cellular for normal inner ear function. J. Neurosci. 26, 10188–10198. function. Although there is no doubt that CLICs can form channels [9] Landry, D.W., Reitman, M., Cragoe Jr., E.J. and Al-Awqati, Q. (1987) Epithelial in vitro and there is clear evidence that these same channels form chloride channel. Development of inhibitory ligands. J. Gen. Physiol. 90, 779– 798. in cells when CLICs are overexpressed, it is still not proven that, in [10] Landry, D.W., Akabas, M.H., Redhead, C., Edelman, A., Cragoe Jr., E.J. and Al- native cells, the endogenous CLIC proteins are forming functional Awqati, Q. (1989) Purification and reconstitution of chloride channels from channels that are utilised by the cell. The advent of the CLIC knock- kidney and trachea. Science 244, 1469–1472. [11] Redhead, C.R., Edelman, A.E., Brown, D., Landry, D.W. and Al-Awqati, Q. (1992) out mice along with more sophisticated transgenic techniques A ubiquitous 64-kDa protein is a component of a chloride channel of plasma should be able to answer this central question. and intracellular membranes. Proc. Natl. Acad. Sci. USA 89, 3716–3720. It is also essential that we get a map of the proteins that interact [12] Landry, D., Sullivan, S., Nicolaides, M., Redhead, C., Edelman, A., Field, M., Al- directly with the CLICs in their soluble and membrane bound Awqati, Q. and Edwards, J. (1993) Molecular cloning and characterization of p64, a chloride channel protein from kidney microsomes. J. Biol. Chem. 268, forms. This will require careful experimental design with emphasis 14948–14955. on validation through different techniques, the interaction be- [13] Valenzuela, S.M. et al. (1997) Molecular cloning and expression of a chloride tween the CLIC proteins and the actin cytoskeleton results in a ion channel of cell nuclei. J. Biol. Chem. 272, 12575–12582. [14] Howell, S., Duncan, R.R. and Ashley, R.H. (1996) Identification and large number of indirect interactions. Such studies must be cou- characterisation of a homologue of p64 in rat tissues. FEBS Lett. 390, 207– pled with in vitro studies using purified recombinant proteins to 210. show that the partner proteins do indeed interact directly with [15] Duncan, R.R., Westwood, P.K., Boyd, A. and Ashley, R.H. (1997) Rat brain p64H1, expression of a new member of the p64 chloride channel protein the CLICs. High resolution crystal structures of such complexes family in endoplasmic reticulum. J. Biol. Chem. 272, 23880–23886. would also advance the field significantly. [16] Tonini, R., Ferroni, A., Valenzuela, S.M., Warton, K., Campbell, T.J., Breit, S.N. In terms of whole organisms, understanding physiological func- and Mazzanti, M. (2000) Functional characterization of the NCC27 nuclear protein in stable transfected CHO-K1 cells. FASEB J. 14, 1171–1178. tion will be clouded by the existence of the six CLIC paralogues in [17] Heiss, N.S. and Poustka, A. (1997) Genomic structure of a novel chloride vertebrates and the resultant degeneracy and redundancy. Double channel gene, CLIC2, in Xq28. Genomics 45, 224–228. CLIC knockout mice will help unravel this problem. One avenue [18] Dulhunty, A., Gage, P., Curtis, S., Chelvanayagam, G. and Board, P. (2001) The glutathione transferase structural family includes a nuclear chloride channel that could prove fruitful is the use of morpholinos in the uro- and a calcium release channel modulator. J. Biol. Chem. chordate C. intestinalis. This tunicate has a single CLIC gene that 276, 3319–3323. is highly homologous to all vertebrate CLICs. Following the effects [19] Board, P.G., Coggan, M., Watson, S., Gage, P.W. and Dulhunty, A.F. (2004) CLIC- 2+ release channels. Int. J. Biochem. of knockdown experiments should be at least as valuable as the C. 2 modulates cardiac ryanodine receptor Ca Cell Biol. 36, 1599–1612. elegans work in dissecting CLIC function. [20] Harrop, S.J. et al. (2001) Crystal structure of a soluble form of the intracellular Finally, as structural biologists, it is essential that we character- chloride ion channel CLIC1 (NCC27) at 1.4-A resolution. J. Biol. Chem. 276, ise the integral membrane form of at least one of the CLIC proteins. 44993–45000. [21] Board, P.G. et al. (2000) Identification, characterization, and crystal structure To date, efforts to get a crystal structure of a CLIC protein have been of the Omega class glutathione transferases. J. Biol. Chem. 275, 24798–24806. hampered by our inability to stabilise the membrane form. Other [22] Tulk, B.M., Kapadia, S. and Edwards, J.C. (2002) CLIC1 inserts from the aqueous biophysical techniques must be explored in parallel to understand phase into phospholipid membranes where it functions as an anion channel. AJP-Cell. the nature and transition of the CLIC to its integral membrane form [23] Warton, K. et al. (2002) Recombinant CLIC1 (NCC27) assembles in lipid in order to pursue a high resolution X-ray crystallography struc- bilayers via a pH-dependent two-state process to form chloride ion channels ture and the molecular understanding this would bring. The key with identical characteristics to those observed in Chinese hamster ovary cells expressing CLIC1. J. Biol. Chem. 277, 26003–26011. to discovering the secret that keeps CLICs in the membrane may [24] Berryman, M., Bruno, J., Price, J. and Edwards, J.C. (2004) CLIC-5A functions as a well come from the above studies of CLIC proteins in vitro, in cells chloride channel in vitro and associates with the cortical actin cytoskeleton and in vivo. in vitro and in vivo. J. Biol. Chem. 279, 34794–34801. [25] Littler, D.R. et al. (2004) The intracellular chloride ion channel protein CLIC1 undergoes a redox-controlled structural transition. J. Biol. Chem. 279, 9298– 9305. Acknowledgements [26] Littler, D.R. et al. (2005) Crystal structure of the soluble form of the redox- regulated chloride ion channel protein CLIC4. FEBS J. 272, 4996–5007. This work was supported by grants from the Australian Na- [27] Singh, H. and Ashley, R.H. (2006) Redox regulation of CLIC1 by cysteine residues associated with the putative channel pore. Biophys. J. 90, 1628–1638. tional Health and Medical Research Council (P.M.G.C. and S.N.B.), [28] Cromer, B.A. et al. (2007) Structure of the Janus protein human CLIC2. J. Mol. Australian Research Council (ARC) grants (L.J.B. and S.M.V.) and Biol. 374, 719–731. an ARC APD fellowship (L.J.B.); and the Italian Ministry of Univer- [29] Singh, H. and Ashley, R.H. (2007) CLIC4 (p64H1) and its putative sity and Scientific Research COFIN Funds (M.M.). transmembrane domain form poorly selective, redox-regulated ion channels. Mol. Membr. Biol. 24, 41–52. [30] Singh, H., Cousin, M.A. and Ashley, R.H. (2007) Functional reconstitution of mammalian ‘chloride intracellular channels’ CLIC1, CLIC4 and CLIC5 reveals References differential regulation by cytoskeletal actin. FEBS J. 274, 6306–6316. [31] Littler, D.R. et al. (2008) Comparison of vertebrate and invertebrate CLIC [1] Murzin, A.G. (2008) Biochemistry. Metamorphic proteins. Science 320, 1725– proteins: the crystal structures of Caenorhabditis elegans EXC-4 and Drosophila 1726. melanogaster DmCLIC. Proteins 71, 364–378. D.R. Littler et al. / FEBS Letters 584 (2010) 2093–2101 2101

[32] Valenzuela, S.M., Mazzanti, M., Tonini, R., Qiu, M.R., Warton, K., Musgrove, E.A., [42] Berryman, M. and Bretscher, A. (2000) Identification of a novel member of Campbell, T.J. and Breit, S.N. (2000) The nuclear chloride ion channel NCC27 is the chloride intracellular channel gene family (CLIC5) that associates with involved in regulation of the cell cycle. J. Physiol. 529 (Pt. 3), 541–552. the actin cytoskeleton of placental microvilli. Mol. Biol. Cell 11, 1509– [33] Goodchild, S.C., Howell, M.W., Cordina, N.M., Littler, D.R., Breit, S.N., Curmi, 1521. P.M. and Brown, L.J. (2009) Oxidation promotes insertion of the CLIC1 chloride [43] Suginta, W., Karoulias, N., Aitken, A. and Ashley, R.H. (2001) Chloride intracellular channel into the membrane. Eur. Biophys. J. 39, 129–138. intracellular channel protein CLIC4 (p64H1) binds directly to brain dynamin [34] Fanucchi, S., Adamson, R.J. and Dirr, H.W. (2008) Formation of an unfolding I in a complex containing actin, tubulin and 14-3-3 isoforms. Biochem. J. 359, intermediate state of soluble chloride intracellular channel protein CLIC1 at 55–64. acidic pH. Biochemistry 47, 11674–11681. [44] Berryman, M.A. and Goldenring, J.R. (2003) CLIC4 is enriched at cell–cell [35] Stoychev, S.H., Nathaniel, C., Fanucchi, S., Brock, M., Li, S., Asmus, K., Woods Jr., junctions and colocalizes with AKAP350 at the centrosome and midbody of V.L. and Dirr, H.W. (2009) Structural dynamics of soluble chloride intracellular cultured mammalian cells. Cell Motil. Cytoskeleton 56, 159–172. channel protein CLIC1 examined by amide hydrogen-deuterium exchange [45] Bohman, S., Matsumoto, T., Suh, K., Dimberg, A., Jakobsson, L., Yuspa, S. and mass spectrometry. Biochemistry 48, 8413–8421. Claesson-Welsh, L. (2005) Proteomic analysis of vascular endothelial growth [36] Li, Y., Li, D., Zeng, Z. and Wang, D. (2006) Trimeric structure of the wild soluble factor-induced endothelial cell differentiation reveals a role for chloride chloride intracellular ion channel CLIC4 observed in crystals. Biochem. intracellular channel 4 (CLIC4) in tubular morphogenesis. J. Biol. Chem. 280, Biophys. Res. Commun. 343, 1272–1278. 42397–42404. [37] Mi, W., Liang, Y.H., Li, L. and Su, X.D. (2008) The crystal structure of human [46] Griffon, N., Jeanneteau, F., Prieur, F., Diaz, J. and Sokoloff, P. (2003) CLIC6, a chloride intracellular channel protein 2: a disulfide bond with functional member of the intracellular chloride channel family, interacts with dopamine implications. Proteins 71, 509–513. D(2)-like receptors. Brain Res. Mol. Brain Res. 117, 47–57. [38] Dixon, D.P., Davis, B.G. and Edwards, R. (2002) Functional divergence in the [47] Shanks, R.A., Larocca, M.C., Berryman, M., Edwards, J.C., Urushidani, T., glutathione transferase superfamily in plants. Identification of two classes Navarre, J. and Goldenring, J.R. (2002) AKAP350 at the Golgi apparatus. II. with putative functions in redox homeostasis in Arabidopsis thaliana. J. Biol. Association of AKAP350 with a novel chloride intracellular channel (CLIC) Chem. 277, 30859–30869. family member. J. Biol. Chem. 277, 40973–40980. [39] Elter, A., Hartel, A., Sieben, C., Hertel, B., Fischer-Schliebs, E., Luttge, U., Moroni, [48] Chuang, J.Z., Milner, T.A., Zhu, M. and Sung, C.H. (1999) A 29 kDa intracellular A. and Thiel, G. (2007) A plant homolog of animal chloride intracellular chloride channel p64H1 is associated with large dense-core vesicles in rat channels (CLICs) generates an ion conductance in heterologous systems. J. hippocampal neurons. J. Neurosci. 19, 2919–2928. Biol. Chem. 282, 8786–8792. [49] Maeda, K. et al. (2008) CLIC4 interacts with histamine H3 receptor and [40] Garcera, A., Barreto, L., Piedrafita, L., Tamarit, J. and Herrero, E. (2006) enhances the receptor cell surface expression. Biochem. Biophys. Res. Saccharomyces cerevisiae cells have three Omega class glutathione S- Commun. 369, 603–608. transferases acting as 1-Cys thiol transferases. Biochem. J. 398, 187–196. [50] Tung, J.J., Hobert, O., Berryman, M. and Kitajewski, J. (2009) Chloride [41] Ponsioen, B., van Zeijl, L., Langeslag, M., Berryman, M., Littler, D., Jalink, K. and intracellular channel 4 is involved in endothelial proliferation and Moolenaar, W.H. (2009) Spatiotemporal regulation of chloride intracellular morphogenesis in vitro. Angiogenesis 12, 209–220. channel protein CLIC4 by RhoA. Mol. Biol. Cell 20, 4664–4672.