Clustered PI(4,5)P₂ accumulation and ezrin phosphorylation in response to CLIC5A

The chloride intracellular channel (CLIC) 5A belongs to a family of highly conserved redox-sensitive metamorphic proteins distantly related to glutathione-S-transferase (Cromer et al., 2002; Littler et al., 2004). CLICs exist as soluble, membrane-bound or cytoskeleton-associated forms and exhibit ion-conducting properties in artificial phospholipid bilayers (Ashley, 2003; Berryman et al., 2004). Whether CLICs function as specific Cl⁻ channels in cells is uncertain, given that antagonists of classic Cl⁻ channels, 4,4′-diisothiocyanatostilbene-2,2′-disulfonate (DIDS) and 4-acetamido-4′-isothiocyanato-2,2′-stilbenedisulfonic acid (SITS) have no effect on CLIC channel activity. Nonetheless, the inhibitor indanyloxyacetic acid 94 (IAA-94), originally used to isolate p64, the bovine CLIC5B isoform, reduces ion conductance of CLICs (Ashley, 2003; Landry et al., 1993; Redhead et al., 1992; Tulk et al., 2000). CLICs are found in diverse subcellular locations, for instance at the base of actin-based microvilli and stereocilia (Berryman and Bretscher, 2000; Gagnon et al., 2006; Salles et al., 2013), the Golgi, mitochondria and secretory vesicles (Berry et al., 2003; Chuang et al., 1999; Fernández-Salas et al., 1999; Redhead et al., 1992; Shanks et al., 2002). Their mechanism(s) of action are as yet incompletely understood.

Two alternative first exons (A and B) of the CLIC5 gene produce the 32-kDa isoform CLIC5A (251 amino acids) and the 49-kDa CLIC5B (410 amino acids), respectively (Berryman and Bretscher, 2000; Shanks et al., 2002). CLIC5A was first discovered in a protein complex isolated from human placental microvilli, together with ezrin and several other actin-associated proteins (Berryman and Bretscher, 2000). When CLIC5A was overexpressed in cells, it localized to microvilli and a substantial portion became resistant to detergent extraction and remained coupled to the cytoskeleton (Berryman et al., 2004). In the inner ear, CLIC5A colocalizes with actin at the base of cochlear and vestibular hair cell stereocilia, with a similar distribution to that of the ERM (ezrin, radixin and moesin) protein radixin (Gagnon et al., 2006; Salles et al., 2013). In CLIC5-deficient mice (CLIC5^jbg/jbg), CLIC5A is absent from the cochlear and vestibular hair cells, radixin expression is reduced and stereocilia degenerate after birth, leading to vestibular dysfunction and complete deafness by 7 months of age (Gagnon et al., 2006; Salles et al., 2013), a phenocopy of radixin deficiency (Kitajiri et al., 2004).

We previously reported that CLIC5A mRNA is highly expressed and enriched in human renal glomeruli compared with its expression in other tissues and cells (Nyström et al., 2009). In glomeruli, CLIC5A colocalizes with ezrin and podocalyxin in a highly polarized fashion at the apical plasma membrane of podocyte foot processes. In mice lacking CLIC5A, transmission electron microscopy shows broadening and patchy fusion of podocyte foot processes, and ezrin as well as phospho-ezrin levels are reduced relative to those of wild-type mice (Pierchala et al., 2010; Wegner et al., 2010). These findings suggest that CLIC5A is required for the development and maintenance of the ezrin-dependent ultrastructural organization of glomerular podocyte foot processes, akin to the dependence of stereocilia integrity on CLIC5A in cochlear and vestibular hair cells.

ERM proteins function to link integral plasma membrane proteins to cortical F-actin (Bretscher, 1999; Tsukita and Yonemura, 1999). They play a crucial role in defining cell morphology and processes like adhesion, migration and endocytosis (Bretscher et al., 2002; Denker and Barber, 2002; Denker et al., 2000; Ivetic and Ridley, 2004; Takeuchi et al., 1994). ERM proteins play a crucial role in the formation of actin-based cellular extensions like microvilli, filopodia, microspikes, lamellipodia and podocyte foot processes (Baumgartner et al., 2006; Lamb et al., 1997; Schwartz-Albiez et al., 1995; Takeuchi et al., 1994). They are involved in signal transduction pathways (Tsukita and Yonemura, 1997) and in cancer cell metastasis (Khanna et al., 2004; Machesky, 2008). In their inactive state, intramolecular associations mask the N- and C-ERM association domains (ERMAD), rendering ERM proteins unable to bind to integral membrane proteins or F-actin. Inactive ERM proteins remain cytosolic and are unstable (Grune et al., 2002). Interaction of the ERM protein N-terminus with phosphatidylinositol 4,5-bisphosphate [PI(4,5)P₂] in lipid bilayers releases the inhibitory interaction between N- and C-termini and supports subsequent phosphorylation of a conserved C-terminal threonine (Thr) residue (Thr567 in ezrin, Thr564 in radixin and Thr558 in moesin) (Fievet et al., 2004). Activation of ERM proteins increases their stability and enables interaction of the N-ERMAD with integral membrane proteins and the C-ERMAD with F-actin (Bretscher et al., 2002; Ivetic and Ridley, 2004; Yonemura et al., 2002). Several kinases, including protein kinase C (PKC), MRCK and AKT can directly phosphorylate PI(4,5)P₂-anchored ERM proteins (Ivetic and Ridley, 2004; Shiue et al., 2005). Rho kinase (ROCK) can also phosphorylate ERM proteins in vitro (Hirao et al., 1996), but in cells ERM phosphorylation is not blocked by ROCK inhibition (Matsui et al., 1999).

In this study, we used the simple model of ectopically expressed CLIC5A in COS-7 cells, which do not express CLIC5A at baseline, to determine whether CLIC5A regulates ezrin function. Ectopically expressed CLIC5A increases PI(4,5)P₂ production in dorsal plasma membrane clusters through a functional interaction with PI(4,5)P₂-generating kinases, in turn facilitating ezrin phosphorylation, its association with the actin cytoskeleton and the formation of cell surface projections. These data suggest that CLIC5A promotes localized PI(4,5)P₂ generation, thus stimulating ezrin protein activation.

Given that CLIC5A deficiency results in reduced ezrin and phosphorylated (phospho)-ERM abundance in glomerular podocytes of CLIC5^jbg/jbg mice (Wegner et al., 2010), and dephosphorylated ERM proteins are unstable (Grune et al., 2002), we hypothesized that CLIC5A regulates ezrin phosphorylation. To investigate this, COS-7 cells, which lack detectable CLIC5A, were transiently transfected with pcDNA3.1-CLIC5A or with empty vector and evaluated for ERM phosphorylation. Although total ezrin levels were similar in CLIC5A- and vector-transfected cells, ERM protein phosphorylation was greater in SDS lysates of cells expressing CLIC5A than in those of vector-transfected cells (Fig. 1A, Lysate). Two phospho-ERM protein bands corresponding to phospho-ezrin and phospho-moesin, based on their respective molecular mass, were observed in COS-7 cells and both were hyperphosphorylated in the presence of CLIC5A. Densitometric quantification showed that the ratio of phospho-ERM to total ezrin increased approximately threefold when COS-7 cells expressed ectopic CLIC5A, relative to vector-transfected cells (Fig. 1B, Lysate).

Given that phosphorylated ERM proteins associate with F-actin (Bretscher et al., 2002; Ivetic and Ridley, 2004), we determined whether ectopic CLIC5A expression alters the association of ezrin with the cytoskeletal fraction. Cytoskeletons were prepared by lysing adherent cells with Triton X-100 and then harvesting the detergent-insoluble material that remained attached to the plates (Berryman et al., 2004). The amount of total ezrin and phospho-ERM associated with such cytoskeletal preparations was much greater in CLIC5A-transfected than in vector-transfected cells (Fig. 1A, Cytoskeleton). Densitometric quantification showed that at least sixfold more phospho-ERM was associated with the cytoskeleton of cells expressing CLIC5A compared with that of cells transfected with vector alone (Fig. 1B, Cytoskeleton). Overexposure might have led to underestimation of phospho-ERM abundance in the cytoskeletons. There also was approximately fourfold more total ezrin and phospho-ERM in Triton-X-100-insoluble fractions prepared from whole CLIC5A-transfected COS-7 cells than in similar fractions prepared from cells transfected with vector alone (data not shown). Because total ezrin abundance in whole SDS lysates was unchanged when CLIC5A was expressed (Fig. 1A), the results indicate that a greater fraction of ezrin is phosphorylated and therefore associates with the cytoskeleton when CLIC5A is expressed.

Confocal immunofluorescence microscopy also showed that phospho-ERM immunofluorescence was greater in COS-7 cells transfected with CLIC5A or GFP–CLIC5A compared with that in cells that remained untransfected (Fig. 1C). Hence, ectopic CLIC5A expression in COS-7 cells results in enhanced ERM phosphorylation.

It is well established that ezrin phosphorylation can induce the formation of actin-rich apical protrusions (Yonemura et al., 1999; Zwaenepoel et al., 2012). We therefore determined whether ectopic CLIC5A expression in COS-7 cells altered actin polymerization and cell surface architecture. COS-7 cells were co-transfected with CLIC5A and YFP–actin or with vector and YFP–actin, both at a 10∶1 (CLIC5A or vector∶YFP–actin) ratio to maximize the probability of co-transfection of CLIC5A and vector with the reporter, and evaluated 48 h later by confocal microscopy. Substantially more YFP–actin organized in filaments was observed in cells that were (presumably) transfected with CLIC5A than in vector-transfected cells (Fig. 2A). By using scanning electron microscopy (SEM), we observed that many cells displayed cell surface projections and ruffles in cultures transfected with CLIC5A. Such projections were not observed in cultures of vector-transfected cells (Fig. 2B). Rhodamine–phalloidin staining also revealed an increase in polymerized actin in CLIC5A-transfected compared with vector-transfected cells (Fig. 2C). These data indicate that CLIC5A increases actin polymerization and cell surface remodeling, consistent with previous reports for CLIC5A (Berryman et al., 2004) and CLIC4 (Viswanatha et al., 2013).

We next examined whether the increase in ERM phosphorylation in response to CLIC5A could reflect the inhibition of ERM phosphatases. It was reported previously that both PKC and ROCK can phosphorylate ERM proteins on the conserved C-terminal Thr residue (Ivetic and Ridley, 2004). Untransfected COS-7 cells were therefore treated with the PKC inhibitor staurosporine (SSP), the ROCK inhibitor Y27632 and the protein phosphatase 1 inhibitor calyculin A (Cal-A). In the presence of Cal-A, phospho-ERM abundance increased substantially in SDS-soluble COS-7 cell lysates, indicating that Cal-A-sensitive phosphatases actively dephosphorylate ERM proteins in COS-7 cells. Conversely, SSP treatment significantly reduced ERM phosphorylation, whereas Y27632 was essentially without effect (Fig. 3A, Lysate). Thus, PKC is the principal kinase phosphorylating ERM proteins in COS-7 cells. In keeping with the findings in cell lysates, Cal-A enhanced and SSP strongly reduced the amount of ezrin and phospho-ERM in Triton-X-100-resistant cytoskeletal fractions (Fig. 3A).

To determine whether CLIC5A blocks ERM phosphatases, we evaluated the rate of ERM dephosphorylation during PKC inhibition. Although the amount of phospho-ERM was again much greater in CLIC5A-transfected compared with vector-transfected cells, the abundance of phospho-ERM declined rapidly upon addition of SSP in both CLIC5A- and vector-transfected cells, without any change in total ezrin abundance (Fig. 3B). Nonetheless, the rate at which the phospho-ERM∶total ezrin ratio declined did not differ between COS-7 cells expressing CLIC5A and those transfected with the vector alone (Fig. 3C). These findings suggest that the increase in ERM phosphorylation in cells expressing CLIC5A cannot be attributed to inhibition of ERM phosphatase(s). Interestingly, the abundance of the CLIC5A immunoreactive band also declined rapidly upon SSP addition. This finding implies either that CLIC5A is rapidly degraded or that it undergoes a conformational change and reduced antibody reactivity upon PKC inhibition.

We next determined whether ectopic expression of CLIC5A alters the negative surface potential of the plasma membrane inner leaflet in COS-7 cells by co-transfecting them with the surface potential reporter GFP–Kras (Yeung et al., 2006) and CLIC5A or empty vector. Live-cell imaging revealed that the surface potential biosensor GFP–Kras was evenly distributed along the plasma membrane in the absence of CLIC5A. There was a dramatic redistribution of GFP–Kras into plasma membrane clusters at the cell surface in the presence of CLIC5A (Fig. 4A). The head of PI(4,5)P₂ is negatively charged, and PI(4,5)P₂ is required as a membrane anchor for ERM protein activation and phosphorylation (Fievet et al., 2004). We therefore investigated whether CLIC5A alters the distribution of PI(4,5)P₂ in COS-7 cells, by co-transfecting them with the PI(4,5)P₂ biosensor GFP–PH-PLC and CLIC5A or vector alone. To maximize co-transfection, the ratio of GFP–PH-PLC reporter∶CLIC5A or vector cDNA was 1∶10. Live-cell imaging showed that the PI(4,5)P₂ reporter localized to plasma membrane clusters that extended vertically from the COS-7 cell surface (Fig. 4B), and that these sites of PI(4,5)P₂ accumulation were more abundant in CLIC5A-transfected than in vector-transfected cells. The dorsal membrane∶cytoplasm GFP–PH-PLC ratio was significantly greater in CLIC5A-transfected than in vector-transfected cells, indicating an increase in dorsal plasma membrane PI(4,5)P₂ abundance in CLIC5A-expressing cells (Fig. 4C).

When COS-7 cells were co-transfected with RFP–PH-PLC and GFP–CLIC5A (ratio 1∶10), there was substantial colocalization of GFP–CLIC5A with the PI(4,5)P₂ reporter at the dorsal plasma membrane (Fig. 4D, X-Z). In control cells, GFP remained cytoplasmic and did not colocalize with the PI(4,5)P₂ reporter. Examination of Z-stacks revealed that the PI(4,5)P₂ reporter was organized in surface ruffles whose density was much greater in cells transfected with CLIC5A compared with that of vector-transfected cells (Fig. 4D). These experiments suggest strongly that the reorganization of the dorsal plasma membrane in COS-7 cells transfected with CLIC5A is accompanied by a polarized accumulation of PI(4,5)P₂ in cell surface clusters containing CLIC5A.

PI(4,5)P₂ is a substrate for PLC, and substantial PLC activation depletes plasma membrane PI(4,5)P₂. We therefore determined whether activation of PLC would alter CLIC5A-induced ERM phosphorylation. COS-7 cells transfected with CLIC5A or vector alone were treated with 800 µM of the PLC activator m-3M3FBS. As early as 5 min after initiation of m-3M3FBS treatment, ERM phosphorylation decreased significantly in both CLIC5A- and vector-transfected cells (Fig. 5A), and the effect of CLIC5A on ERM phosphorylation was lost. In HeLa cells transfected with CLIC5A or vector and then treated with m-3M3FBS, CLIC5A-dependent ERM phosphorylation was similarly lost (data not shown). The abundance of CLIC5A also declined in response to m-3M3FBS (Fig. 5A). Imaging revealed that dorsal plasma membrane clusters bearing the RFP–PH-PLC reporter were reduced in a time-dependent fashion after m-3M3FBS treatment, and CLIC5A was redistributed into the cytosol (Fig. 5B). These findings suggest that CLIC5A-dependent ERM phosphorylation is abrogated in response to PI(4,5)P₂ depletion, and are consistent with the possibility that the alteration in the distribution of GFP–Kras (Fig. 4A) and GFP–PH-PLC or RFP–PH-PLC (Fig. 4B,D) in CLIC5A-transfected cells results from changes in PI(4,5)P₂ abundance in the dorsal plasma membrane. The finding that m-3M3FBS also reduced CLIC5A association with the dorsal plasma membrane and resulted in a reduction in CLIC5A abundance suggests that the association of CLIC5A with the dorsal plasma membrane of cells is itself dependent on PI(4,5)P₂.

Given that the ion conductance of CLIC5A is inhibited by IAA-94 (Berryman et al., 2004), we next determined whether IAA-94 alters ERM phosphorylation in COS-7 cells expressing CLIC5A. IAA-94 did not inhibit the increase in phospho-ERM abundance in cell lysates or in cytoskeletal fractions of COS-7 cells expressing CLIC5A (Fig. 5C). This observation suggests that the CLIC5A-dependent effect on ERM phosphorylation is independent of its putative ion channel activity.

The PI(4,5)P₂ in phospholipid bilayers is produced from phosphatidylinositol 4-phosphate (PI4P) by PI4P5 kinases, and to a lesser extent from phosphatidylinositol 5-phosphate (PI5P) by PI5P4 kinases (Divecha, 2010). All isoforms of these PI(4,5)P₂-generating kinases contain surface cationic residues conferring a positive charge that allows them to bind to negatively charged membrane phospholipids (Fairn et al., 2009). We therefore determined whether the presence of CLIC5A alters the distribution of PI4P5Kα in COS-7 cells. To this end, COS-7 cells were co-transfected with HA–PI4P5Kα and GFP–CLIC5A or GFP. There was considerable colocalization of GFP–CLIC5A with HA–PI4P5Kα at the dorsal plasma membrane (Fig. 6A, X-Z) but not with GFP in control cells.

To further probe whether CLIC5A might interact with PI(4,5)P₂-generating kinases, COS-7 cells were transfected with HA-tagged PI4P5Kα, PI4P5Kβ, PI5P4Kα and PI5P4Kβ cDNA. Cell lysates were then incubated with recombinant GST–CLIC5A or GST immobilized on glutathione–Sepharose beads. Western blot analysis of the eluted proteins showed that HA–PI4P5Kα (Fig. 6B) and HA–PI5P4Kα (Fig. 6C) in cell lysates bound to GST-CLIC5A but not to GST. Identical results were obtained with HA–PI4P5Kβ and HA–PI5P4Kβ (data not shown).

Occasionally, overexpression of CLIC5A or the combination of CLIC5A and HA–PI4P5Kα resulted in the formation of very large vacuoles in COS-7 cells. Such vacuoles were not observed in vector-transfected cells. CLIC5A and HA–PI4P5Kα were found to colocalize to the membrane of such vacuoles (Fig. 6D). In vacuolated cells transfected only with CLIC5A, the PI(4,5)P₂ reporter GFP–PH-PLC also localized to the vacuolar membranes (Fig. 6E). These data also support the concept that CLIC5A colocalizes with HA–PI4P5Kα at sites of PI(4,5)P₂ production.

To further probe the functional interaction between CLIC5A and PI4P5Kα, endogenous PI4P5Kα was depleted using specific small interfering (si)RNA. In both COS-7 and HeLa cells, PI4P5Kα-specific but not nonspecific siRNA inhibited PI4P5Kα protein expression (Fig. 7A). Silencing of PI4P5Kα abrogated the effect of CLIC5A overexpression on ERM phosphorylation in both COS-7 and HeLa cells (Fig. 7A), consistent with an essential role for PI4P5Kα in CLIC5A-dependent ERM phosphorylation.

We next determined whether CLIC5A N- or C-terminal deletions would alter the functional effect on ERM phosphorylation. CLIC5A in which the first β-pleated sheet alone (CLIC5A^22–251), the first β-pleated sheet and the putative membrane-spanning domain (CLIC5A^55–251) or only the N-terminal ninth α-helix (CLIC5A^1–232) were deleted were therefore expressed in COS-7 cells. In contrast to wild-type CLIC5A, none of the deletion mutants supported enhanced ERM phosphorylation in COS-7 cells (Fig. 7B). The mutant CLIC5A proteins also failed to associate with the plasma membrane, did not colocalize with HA–PI4P5Kα (Fig. 7C) and failed to induce plasma membrane PI(4,5)P₂ accumulation (Fig. 7D). The mutant CLIC5A proteins did not achieve the level of expression obtained for full-length CLIC5A (Fig. 7B), leaving open the possibility that the functional defect is related, in part, to low levels of expression. Nonetheless, the findings suggest that both N- and C-termini are required for the membrane association of CLIC5A, and potentially also for the stabilization of CLIC5A in the plasma membrane. Although the dorsal plasma membrane localization of PI4P5Kα appears to be independent of CLIC5A, increased PI(4,5)P₂ accumulation was nonetheless dependent on the presence of full-length CLIC5A. This finding implies that CLIC5A is not required for the recruitment of PI4P5Kα to the plasma membrane, but that it is required for localized PI4P5 and/or PI5P4 kinase activation.

CLIC5A is expressed at extraordinarily high levels in renal glomeruli (Nyström et al., 2009), where it colocalizes with ezrin in podocytes (Pierchala et al., 2010; Wegner et al., 2010). In turn, ezrin interacts with the transmembrane protein podocalyxin (Orlando et al., 2001), with NHERF2 as the intermediary (Takeda, 2003). We now observe that much less ezrin and phospho-ERM are associated with the detergent-insoluble fraction of glomeruli from CLIC5-deficient mice compared with wild-type mice (Fig. 8A). In glomerular podocytes, phospho-ezrin binds to NHERF2, coupling it to the cytoskeleton. We therefore determined whether deletion of CLIC5A would also alter the association of NHERF2 with the cytoskeletal fraction. Because NHERF2 is expressed predominantly in glomeruli (Wade et al., 2001), kidney cortex was used for these experiments. Substantial NHERF2 was observed in the detergent-insoluble fraction of kidney cortex lysate from wild-type mice, but in CLIC5-deficient mice most of the NHERF2 remained associated with the detergent-soluble fraction (Fig. 8B). Furthermore, the colocalization of phospho-ERM with podocalyxin observed in wild-type mice was markedly reduced in the glomeruli of CLIC5-deficient mice (Fig. 8C). Thus, CLIC5A also regulates ERM phosphorylation and function in vivo, in keeping with the increased susceptibility to glomerular injury in these mice (Wegner et al., 2010).

CLIC and ERM proteins are commonly associated (Jiang et al., 2014), and both CLIC5A and CLIC4 have been shown previously to associate with ezrin and to induce cell surface microvillus formation in epithelial cells (Berryman et al., 2004; Viswanatha et al., 2013). However, the mechanism(s) governing the functional interactions between CLICs and ERM proteins are not understood. Here, we focused on CLIC5A, which is associated with radixin in sensory stereocilia of inner ear hair cells (Gagnon et al., 2006; Salles et al., 2013) and with ezrin in renal glomerular podocytes (Wegner et al., 2010). We observe that ectopic CLIC5A expression in COS-7 cells, which are null for CLIC5A at baseline, strongly stimulates ERM phosphorylation, ezrin association with the cytoskeleton and remodeling of the dorsal membrane architecture. CLIC5A induces highly localized PI(4,5)P₂ accumulation in discrete clusters at the dorsal plasma membrane and colocalizes with overexpressed HA–PI4P5Kα in these clusters. Furthermore, GST–CLIC5A can pull down several PI(4,5)P₂-generating kinases from cell lysates, and PLC activation, which depletes PI(4,5)P₂, brings CLIC5A-dependent ezrin phosphorylation back to control levels as does silencing of endogenous PI4P5Kα expression. CLIC5A deletion mutants that fail to localize to the plasma membrane also fail to support clustered PI(4,5)P₂ accumulation and ERM phosphorylation. The data suggest that CLIC5A serves to increase PI(4,5)P₂ formation in discrete dorsal plasma membrane clusters to promote localized ERM phosphorylation, which in turn supports the formation of cell surface projections. Because the effect of CLIC5A is crucially dependent on its association with the plasma membrane, we postulate that CLIC5A might be essential for the formation of a signaling complex that leads to the activation of PI(4,5)P₂-generating kinases specifically targeting localized ERM activation. The findings in renal glomeruli, where deletion of CLIC5A leads to reduced ERM phosphorylation and function are consistent with a similar role for CLIC5A in ERM phosphorylation in vivo.

The hypothesis that CLIC5A regulates ERM phosphorylation was based on previous findings that CLIC5A is associated with ERM proteins in sensory stereocilia and podocyte foot processes, and that ERM protein abundance is reduced in these locations in CLIC5-deficient mice (Gagnon et al., 2006; Wegner et al., 2010). Ezrin mRNA expression in the glomeruli of CLIC5-deficient mice is not reduced (A.A.-M., data not shown) and unphosphorylated ERM proteins are more readily degraded, which raised the possibility that CLIC5A might regulate ezrin phosphorylation. We found that ectopic expression of CLIC5A in COS-7 cells leads to a significant and highly reproducible increase in ERM phosphorylation and its association with the cytoskeletal fraction (Fig. 1). Nonetheless, in the COS-7 cells, overall ezrin abundance did not appear to change in the presence of CLIC5A. Therefore, whether the reduced ERM protein abundance in sensory stereocilia and podocytes of CLIC5-deficient mice results from ezrin destabilization due to reduced phosphorylation remains unclear.

The activity of ERM proteins involves phosphorylation-dephosphorylation cycles. Phosphorylation is stimulated by PKC, ROCK (Ivetic and Ridley, 2004) and AKT2 (Shiue et al., 2005), and dephosphorylation by myosin phosphatase and protein phosphatase 2C (PP2C) (Ivetic and Ridley, 2004). In COS-7 cells, inhibition of PKC but not ROCK inhibited ERM phosphorylation (Fig. 3), and we found no change in Thr308 and Ser473 phosphorylation of AKT2 in response to CLIC5A (data not shown). These results are consistent with the previous report showing no effect of the potent RhoA kinase inhibitor C3 transferase on ERM phosphorylation in various kidney-derived epithelial cells (Yonemura et al., 2002). Although ERM protein phosphorylation was much greater in the presence than in the absence of CLIC5A, the rate of ERM dephosphorylation during PKC inhibition was not altered by CLIC5A, suggesting that the effect of CLIC5A on ERM phosphorylation is not explained by inhibition of ERM phosphatases. Interestingly, we observed a consistent rapid decline in CLIC5A abundance upon treatment of COS-7 cells with SSP (Fig. 3B). The cause for the reduction in CLIC5A abundance is not yet clear, but given that SSP rapidly induces apoptosis in cells, it is conceivable that CLIC5A is subject to caspase-mediated cleavage. Alternatively, it is also possible that CLIC5A itself is phosphorylated by PKC and that inhibition of PKC induces a conformational change altering antibody reactivity. Further work will be needed to determine the mechanism whereby SSP causes the CLIC5A abundance to decline.

The CLIC proteins are redox sensitive and partition into lipid bilayers upon oxidation (Goodchild et al., 2009; Littler et al., 2005; Littler et al., 2004; Valenzuela et al., 2013), where they have ion-conducting properties that are blocked by IAA-94. Indeed, bovine CLIC5B (p64) was first isolated from kidney by affinity purification with IAA-94 (Landry et al., 1993; Redhead et al., 1992). It has been argued that CLICs might not function as Cl⁻ channels because the better characterized inhibitors SIDS or DIDS are without effect on CLIC ion conductance (Jentsch et al., 2002). However, in keeping with CLIC-mediated Cl⁻ ion conductance, a role in vacuolar and/or vesicular acidification has been reported for CLIC5B in osteoclasts (Edwards et al., 2006), for CLIC1 in macrophages (Jiang et al., 2012) and for CLIC4 in endothelial cells (Ulmasov et al., 2009). Here, we observed that the effect of CLIC5A on ERM phosphorylation was not inhibited by extracellular IAA-94 (Fig. 5C). It is possible that we did not achieve a sufficiently high intracellular concentration of IAA-94 to inhibit Cl⁻ conductance, but others have observed functional effects of IAA-94 in whole cells using the same concentration and timecourse (Orlando, 2002). It is also of note that CLIC5A expression leads to actin polymerization (Fig. 2A) and that F-actin inhibits CLIC-mediated ion conductance (Singh et al., 2007). Although our experiment does not exclude the possibility that CLIC5A forms ion-conducting pores when expressed in COS-7 cells, the data suggest that the effect of CLIC5A on ERM phosphorylation is not due an IAA-94-sensitive ion conductance.

Activation of ERM proteins requires the initial binding of the ERM N-terminus to PI(4,5)P₂, prompting a conformational change that then facilitates C-terminal actin binding and phosphorylation of a highly conserved threonine residue at the C-terminus (Bretscher et al., 2002; Ivetic and Ridley, 2004). PI(4,5)P₂ therefore plays an essential role in ERM phosphorylation (Barret et al., 2000; Niggli et al., 2001; Niggli et al., 1995). Because phospholipids are negatively charged, we first determined whether the presence of CLIC5A alters the plasma membrane negative charge distribution using the GFP–Kras reporter (Yeung et al., 2006). The dramatic clustering of the reporter at the dorsal plasma membrane of CLIC5A-transfected cells (Fig. 4A) could represent the remodeling of the dorsal plasma membrane (Fig. 2B), a redistribution of negatively charged lipids and/or binding of the reporter to the acidic foot loop (Cromer et al., 2007; Littler et al., 2010) of membrane-associated CLIC5A. Ectopic expression of CLIC5A was also associated with clustered accumulation of the PI(4,5)P₂ reporter GFP–PH-PLC at the dorsal plasma membrane of COS-7 cells, where it colocalized with CLIC5A (Fig. 4C). Redistribution of the PI(4,5)P₂ reporter from cytosol to plasma membrane (Fig. 4B) suggests enhanced PI(4,5)P₂ production at the plasma membrane in the presence of CLIC5A. It is conceivable that CLIC5A could functionally mimic PI(4,5)P₂, binding to the PI(4,5)P₂ reporter and directly supporting the conformational change in ezrin that then leads to its phosphorylation. Previous reports have shown that the PLC activator m-3M3FBS accelerates the hydrolysis of PI(4,5)P₂ and reduces the level of ERM phosphorylation in HeLa (Canals et al., 2012) and Jurkat T cells (Hao et al., 2009). In our studies, m-3M3FBS largely abolished increased ERM phosphorylation in the presence of CLIC5A. In addition, m-3M3FBS rapidly reduced the association of CLIC5A with the plasma membrane and immunoreactive CLIC5A in cell lysates. Hence, the finding that PLC activation abolished the effect of CLIC5A on ERM phosphorylation suggests that PI(4,5)P₂ mediates the effect of CLIC5A on ERM phosphorylation. The fact that CLIC5A redistributes to the cytosol upon PLC activation suggests that its association with the plasma membrane is also dependent on the plasma membrane phospholipid composition.

The PI4P5 kinases (PI4P5Kα, β and γ) phosphorylate the fifth position on the inositol ring of PI4P, producing the majority of PI(4,5)P₂ at the plasma membrane. The PI5P4 kinases also exist as α, β and γ isoforms and catalyze phosphorylation of the fourth position on the inositol ring of PI5P to produce PI(4,5)P₂ (Rameh et al., 1997). It is already known that overexpression of PI4P5Kα induces actin polymerization (Shibasaki et al., 1997) and that all isoforms of PI4P5K contain a positively charged domain necessary for their recruitment to and association with negatively charged regions of the plasma membrane (Fairn et al., 2009). Indeed, recruitment of PI4P5 and/or PI5P4 kinases to the plasma membrane is necessary for PI(4,5)P₂ production. We observed that overexpressed HA–PI4P5Kα colocalizes with CLIC5A at the dorsal plasma membrane of COS-7 cells and that HA-tagged PI4P5K and PI5P4K α and β isoforms were pulled from cell lysates by GST–CLIC5A (Fig. 6), suggesting that CLIC5A and PI(4,5)P₂-generating kinases can associate in cells. Some cells in cultures transfected with CLIC5A developed large intracellular vacuoles (Fig. 6D) similar those described previously in cells overexpressing PI4P5Kα (Yonemura et al., 2002). These vacuoles were observed whether CLIC5A was expressed with or without exogenous HA–PI4P5Kα, but never in vector-transfected cells lacking CLIC5A. We observed that CLIC5A and HA–PI4P5Kα colocalized at the membranes of these large vacuoles, and that PI(4,5)P₂ accumulated at that location (Fig. 6E). Therefore, there appears to be a close association of CLIC5A with PI4P5Kα at locations of enhanced PI(4,5)P₂ generation in CLIC5A-transfected cells. These findings are consistent with a yeast two-hybrid screen that has suggested a possible direct interaction between the C-terminal region of PI4P5Kβ and both CLIC1 and CLIC4 (http://www.signaling-gateway.org). Furthermore, Ikenouchi et al. (Ikenouchi et al., 2013) have shown that PI4P5Kβ associates with NHERF2 in a podocalyxin–NHERF2 complex. Our studies suggest that CLIC5A can interact with all isoforms of PI4P5K and PI5P5K, but because we did not identify the specific endogenous kinase that interacts with CLIC5A in podocytes, it is possible that the effect of CLIC5A on ERM phosphorylation in vivo is restricted to PI4P5Kβ. Although proof of a direct interaction between CLIC5A and PI(4,5)P₂-generating kinases is still lacking, the finding that silencing of endogenous PI4P5Kα abrogates CLIC5A-induced ERM phosphorylation (Fig. 7A) strongly suggests that, in COS-7 cells, CLIC5A directs PI4P5Kα-dependent PI(4,5)P₂ accumulation to sites where ERM proteins can then dock.

When exogenous HA–PIP5Kα was expressed in COS-7 cells, it localized to the dorsal plasma membrane of COS-7 cells even in the absence of CLIC5A. Furthermore, HA–PIP5Kα was also found at the dorsal plasma membrane in the presence of CLIC5A truncation mutants (22–251, 55–251 and 1–232) that remained cytosolic. Therefore, it seems unlikely that CLIC5A is required for the recruitment of PIP5Kα to a plasma membrane location. Nonetheless, even though overexpressed PIP5Kα was observed at the plasma membrane in the absence of CLIC5A, and in the presence of the CLIC5A mutants, we only observed enhanced clustered PI(4,5)P₂ accumulation (Fig. 7D) and ERM phosphorylation (Fig. 7B) in the presence of full-length CLIC5A. Hence, PIP5Kα membrane association did not depend on CLIC5A, but the generation of PI(4,5)P₂ and its downstream effect on ERM phosphorylation did. This finding is most consistent with a model in which CLIC5A increases the localized activity of PI(4,5)P₂-generating enzymes, rather than simply recruiting them to the plasma membrane. Because the abundance of endogenous PI(4,5)P₂-generating enzymes was too low for detection by co-immunoprecipitation or colocalization in this study, it remains unclear whether CLIC5A associates with and activates a specific isoform. Taken together with our evidence of increased clustered accumulation of PI(4,5)P₂ at the dorsal plasma membrane, increased PI(4,5)P₂-dependent ERM phosphorylation in the presence of CLIC5A and the formation of surface ruffles and projections in CLIC5A-expressing cells, it is attractive to postulate that the mechanism of CLIC5A action involves activation of PI4P5K and/or PI5P4K with consequent PI(4,5)P₂ generation that, in turn, facilitates ERM phosphorylation. It is also possible that CLIC5A forms a scaffold that helps assemble proteins involved in activating PI4P5 and/or PI5P4 kinases in specific compartments of the phospholipid bilayer involved in the formation of ERM-dependent projections.

Finally, we evaluated the functional effect of CLIC5A on phosphorylation of ERM protein in renal glomeruli in vivo. In glomerular podocytes, ezrin serves to couple the single transmembrane sialoglycoprotein podocalyxin to cortical actin, through the intermediate NHERF2 (Takeda, 2003; Takeda et al., 2001). There was a striking reduction of phospho-ERM and ezrin in the detergent-insoluble fraction prepared from glomeruli of CLIC5-deficient compared with wild-type mice, in keeping with our findings in the COS-7 cells, and our previous finding of reduced phospho-ERM by immunofluorescence microscopy (Wegner et al., 2010). Furthermore, the association of NHERF-2 with the detergent-insoluble fraction was markedly reduced in the CLIC5-deficient mice. The work in glomeruli of CLIC5-deficient mice is consistent with the findings in COS-7 cells and suggests that CLIC5A is a functionally important component of the podocalyxin–NHERF2–ezrin complex that is required for ezrin activation and coupling of the podocalyxin–NHERF2 complex to the actin cytoskeleton, in turn, supporting the normal foot process architecture.

Taken together, the findings in cultured cells and in vivo are consistent with a model (Fig. 8D) whereby CLIC5A associates with the apical plasma membrane of podocyte foot processes where it supports clustered PI(4,5)P₂ generation in a highly localized fashion. PI(4,5)P₂ in turn engages the ezrin N-terminus, resulting in a conformational change in ezrin that then allows the ezrin N-terminus to bind to NHERF2, and its C-terminus to bind to actin. The conformational change of ezrin induced by CLIC5A-dependent PI(4,5)P₂ generation would also promote PKC-mediated phosphorylation of the ezrin at Thr567 (Fig. 8D). Further work is required to determine whether CLIC5A directly binds to and activates PI(4,5)P₂-generating kinase(s), or whether it serves as a scaffolding platform that assembles the appropriate components of the PI(4,5)P₂ signaling complex. It is also still unclear which of the PI4P5 or PI5P4 kinases interacts with CLIC5A in podocytes in vivo.

All chemicals were of reagent grade and purchased from Sigma (Oakville, ON, Canada) unless otherwise noted. Rabbit anti-phospho-Thr567 ezrin antibodies were from Signalway Antibody LLC (SAB; College Park, MD), and mouse anti-phospho-Thr567 ezrin antibodies were from BD Biosciences (Franklin Lakes, NJ). The anti-phospho-Thr567 ezrin antibodies crossreact with phosphorylated radixin and moesin. Signals obtained with these antibodies are therefore referred to as phosphorylated ERM (phospho-ERM). Rabbit anti-ezrin and anti-PI4P5Kα antibodies were from Cell Signaling (Danvers, MA), rabbit anti-CLIC5 antibodies were from Aviva System Biology (San Diego, CA), mouse anti-β-tubulin and anti-β-actin antibodies were from Millipore (Billerica, MA) and Sigma (Oakville, ON, Canada) respectively. Calyculin A (Cal-A) and m-3M3FBS and were from Millipore (Billerica, MA). Goat anti-podocalyxin antibodies were from R&D Systems (Minneapolis, MN), rat anti-HA antibodies were from Roche (Indianapolis, IN) and goat anti-NHERF2 antibodies were from Santa Cruz Biotechnology Inc. (Dallas, TX).

COS-7 and HeLa cells were maintained in Dulbecco's Modified Eagles Medium (DMEM) containing 10% fetal bovine serum (FBS) (Life Technologies, Burlington, ON) in a 37°C and 5% CO₂ humidified incubator. COS-7 cells were derived from CV-1 cells (Gluzman, 1981) that, in turn, were originally derived from green monkey kidney epithelial cells (Jensen et al., 1964). However, because COS-7 cells are not well polarized, we refer to the cell surface facing the medium as ‘dorsal’ throughout. Cells in six-well or P35 dishes were transfected with 4 µg of plasmid construct using Lipofectamine 2000 (Life Technologies) for COS-7 or Lipofectamine LTX (Life Technologies) for HeLa cells according to the manufacturer's instruction. To maximize the probability of CLIC5A expression with reporter constructs (YFP–actin, GFP–PH-PLC and GFP–Kras), cells were transfected with a mixture of 4 µg CLIC5A and 0.4 µg reporter construct cDNA (ratio of CLIC5A∶reporter of 10∶1). To prepare cell lysates, cells in P35 plates were washed twice with ice-cold PBS 2 days after transfection and then scraped into 500 µl Triton X-100 lysis buffer [0.5% Triton X-100, 10 mM HEPES pH 7.0, 100 mM NaCl, 2.5 mM EGTA, 5 mM MgCl₂, 100 nM Cal-A and 1× PhosStop (Pierce)]. To prepare total SDS-soluble cell lysates, the material in the Triton X-100 buffer was evenly suspended and 100 µl was placed into 2× Laemmli buffer containing 14 mM 2-mercaptoethanol and boiled for 5 min. To harvest the Triton-X-100-insoluble proteins, the remaining 400 µl of the material in Triton X-100 lysis buffer was centrifuged at 18,000 g for 30 min at 4°C. The supernatant was removed and the Triton-X-100-insoluble fraction was resuspended in 100 µl of Laemmli buffer and boiled for 5 min. For some experiments, COS-7 cells were plated in six-well plates on the day prior to transfection, transfected with either CLIC5A or vector and treated 48 h later with 100 nM staurosporine (SSP), 15 µM Y-27632 or 100 nM Cal-A. After the appropriate period of incubation, cells were washed twice with ice-cold PBS and harvested as described.

Constructs encoding the PI(4,5)P₂ reporters (GFP–PH-PLC and RFP–PH-PLC) and negative charge biosensor (GFP–Kras) were obtained from Sergio Grinstein (University of Toronto, Canada). Constructs encoding the HA–PI4,5P kinases were provided by Greg Longmore (Washington University School of Medicine, St Louis, MO) and the YFP–actin construct was provided by Allan Murray (University of Alberta, Canada). A cDNA for CLIC5A encoding the complete open reading frame (ORF) (GenBank™ number DQ679794) was PCR amplified from human kidney cDNA with the primers 5′-CGCACTCGAGACCATGGGGCATCATCATCATCATCATACAGACTCGGCGACAGCTAAC-3′ (forward) and 5′-CCGGGATCCTCAGGATCGGCTGAGGCGTTTGGC-3′ (reverse). In the forward primer, one Kozak consensus sequence (bold) was added to enhance expression and a 6×His tag sequence (italic) was integrated for detection. The PCR product was directly cloned into pCDNA3.1/V5-his-TOPO vector (Life Technologies). This construct is designated pcDNA3.1-CLIC5A (or CLIC5A in the figure legends). The GFP–CLIC5A construct was generated by PCR amplification of human CLIC5A full coding region from pcDNA3.1-CLIC5A using the primers 5′-CGCACTCGAGCCATGACAGACTCGGCGACAGCTAAC-3′ (forward) and 5′-CCGGGATCCTCAGGATCGGCTGAGGCGTTTGGC-3′ (reverse), and cloning into the XhoI-BamHI site of the pEGFP-C1 vector (Clontech, Mountain View, CA). Xpress-tagged wild-type CLIC5A (1–251) and truncation deletion cDNAs (22–251, first β-pleated sheet deletion; 55–251, first β-pleated sheet and putative transmembrane domain deletion; and 1–232, number 9 helix domain deletion) were PCR amplified from pCDNA3.1-CLIC5A using the primers 5′-AGCTTCATGGGGGATCTGTACGACGATGACGATAAGACAGACTCGGCGACAGCTAACGGG-3′ (forward for 1–251 and 1–232), 5′-GCTTCATGGGGGATCTGTACGACGATGACGATAAGGCTGGAATCGATGGAGAAAGCATC-3′ (forward for 22–251), 5′-AGCTTCATGGGGGATCTGTACGACGATGACGATAAGGATCTGAAAAGAAAGCCAGCTGAC-3′ (forward for 55–251), 5′-CCGGGATCCTCAGGATCGGCTGAGGCGTTTGGC-3′ (reverse for 1–251, 22–251 and 55–251) and 5′-CCGGGATCCTCATGCACAGGTGTTGGTGAACTCATC-3′ (reverse for 1–232), followed by cloning into pTARGET vector (Promega, Madison, WI). Restriction enzyme digestion and full-insert sequencing verified the DNA sequence orientation and fidelity.

The full coding region of human CLIC5A was amplified from the pCDNA3.1-CLIC5A construct using PCR with the primers 5′-CGTGGGATCCCCATGACAGACTCGGCGACAGCTAAC-3′ (forward) and 5′-CCGGAATTCTCAGGATCGGCTGAGGCGTTTGGC-3′ (reverse). The PCR-derived CLIC5A was cloned into pGEX-3 (GE Healthcare, Piscataway, NJ). The DNA sequences were verified by sequencing. Escherichia coli BL21 Gold (DE3) (Agilent Technologies, Santa Clara, CA) were transformed with pGEX-3 or pGEX-CLIC5A constructs, induced with 0.2 mM IPTG and grown at 37°C for 4 h. Cells were harvested in PBS containing 1× proteinase inhibitor cocktail (Roche Diagnostics, Indianapolis, IN), followed by sonication and addition of Triton X-100 to a final concentration of 1%. GST and GST–CLIC5A proteins were purified by affinity chromatography on glutathione−Sepharose 4B (GE Healthcare) according the manufacturer's protocol.

To silence PI4P5Kα expression, control siRNA (sc-37007) or PI4P5Kα-specific siRNA (-sc-36232) were purchased from Santa Cruz Biotechnology. The PI4P5Kα siRNA mix consisted of equal concentrations of three distinct siRNAs targeting 5′-GAAGAACCGGAUUGAAAGA-3′, 5′-CAACCUCAUCAGCCUUGAA-3′ and 5′-CCUUCCUCUUCCUCAUGAA-3′ of the human PI4P5Kα mRNA sequence. COS-7 or HeLa cells were plated in 35-mm plates at approximately 60% confluence the day prior to transfection. The medium was replaced with 1 ml of medium without antibiotics immediately before transfection. A total of 2–3 µg of CLIC5A or vector plasmid with 6 pmol control or PI4P5Kα siRNA were combined with 4 µl of Lipofectamine 2000 in 100 µl of Opti-MEM I medium (Life Technologies), co-incubated at room temperature for 45 min and then added to the cells. At 6 h later, 1 ml of growth medium containing 20% FBS was added to the cells. The medium was replaced with serum-free medium (6–10 h) prior to lysis at 48 h after initiation of the transfection.

Cell cytoskeletal fractions were prepared as described previously (Berryman et al., 2004). Briefly, cells in P35 plates were washed twice on ice with cold PBS containing 5 nM Cal-A, and were then incubated at room temperature in 500 µl of PBS containing 0.5% Triton X-100, 1× complete proteinase inhibitor, 1× PhosStop (Roche) and 100 nM Cal-A. The supernatant was removed and the cells were washed gently twice with cold PBS to remove solubilized material. The resulting cytoskeletons were collected in 200 µl of Laemmli buffer and boiled for 5 min prior to western blotting.

Protein samples were boiled for 5 min in Laemmli buffer containing β-mercaptoethanol. They were then separated by electrophoresis on a 12% SDS-PAGE and transferred to polyvinylidene fluoride (PVDF) membranes. Membranes were blocked, incubated with the appropriate primary and HRP-conjugated secondary antibodies and developed with enhanced chemiluminescence (ECL; GE Amersham, Baie d'Urfe, QC, Canada). The membranes were then exposed to X-ray film (Fuji Medical X-Ray Film Super Rx, Fujifilm) for several different time periods. Band density was evaluated using Quantity One software (Biorad Mississauga, ON, Canada). Protein loading was controlled by probing with anti-tubulin or anti-β-actin antibodies.

Cells grown on glass coverslips were washed with cold PBS on ice and fixed with 10% trichloroacetic acid (TCA) to preserve ERM phosphorylation (Hayashi et al., 1999) for 15 min at room temperature. Cells were permeabilized with 0.05% Triton X-100 in PBS for 15 min, blocked with PBS containing 5% goat serum and incubated with primary antibodies overnight at 4°C. The cells were washed four to six times with PBS containing 30 mM glycine (G-PBS) and incubated with secondary antibodies for 1 h at room temperature, followed by three G-PBS washes and mounting in Prolong Antifade (Life Technologies). For actin staining, cells were fixed with 4% paraformaldehyde in PBS for 15 min at room temperature, washed three times with PBS and blocked with PBS containing 5% goat serum for 1 h at room temperature. Cells were then incubated with Rhodamine–phalloidin for 30 min at room temperature, washed with PBS and mounted in Prolong Antifade. For live-cell imaging, COS-7 cells grown on 25-mm glass coverslips were placed into the imaging chamber (37°C, 5% CO₂) of an Olympus spinning-disk fluorescence microscope (Olympus IX-81), and image acquisition was performed with Volocity (Improvision) software. Quantification of the plasma membrane∶cytosol ratio of GFP–PH-PLC (Fig. 4B) was performed using ImageJ (National Institutes of Health). For each cell, the green pixel density was determined in the X-Z optical sections using three uniform squares overlaid on cytosol, dorsal plasma membrane (cell apex and immediately left and right) and outside the cell (background). After subtraction of background, the mean density was determined for cytosol and plasma membrane and the ratio was determined. Quantification was performed for ten cells in each of three independent experiments.

Cells were fixed with 2% glutaraldehyde for 15 min, washed twice with PBS and dehydrated in a graded series of ethanol (30%, 50%, 70%, 95% and 100%). After dehydration, cells were passed through a serial dilution of hexamethyldisilazane (HMDS) in ethanol (75% ethanol, 25% HMDS; 50% ethanol, 50% HMDS; 25% ethanol, 75% HMDS) and 100% HMDS, dried, mounted on SEM stubs and sputter-coated with Au/Pd. Samples were imaged with a Philips/FEI (XL30) scanning electron microscope.

The GST-pull down assay was performed as described previously (Berryman and Bretscher, 2000) with some modifications. Briefly, COS-7 cells expressing HA–PI4P5 or HA–PI5P4 kinases were lysed for 20 min in buffer containing 1% Triton X-100, 20 mM HEPES pH 7.4, 0.6 M KCl and 1 mM EDTA, followed by centrifugation for 30 min at 20,800 g. The clarified supernatant was diluted six times in buffer containing 20 mM HEPES pH 7.4, 10 mM MgCl₂, 1.0 mM ATP, 1× proteinase inhibitor complex and PhosStop. A total of 100 µl was taken as input and the rest was diluted and incubated with GST or GST–CLIC5A immobilized on glutathione beads on a rotator for 3 h at room temperature. The beads were washed five times (for 3 min each) with buffer containing 20 mM HEPES pH 7.4, 100 mM NaCl and 0.1% Triton X-100. Bound proteins were eluted in 2× Laemmli buffer followed by western blotting using anti-HA antibodies.

The CLIC5-deficient C57/BL6 mice were derived from the original CLIC5^jbg/jbg strain on the C3H/HeJ background (Gagnon et al., 2006; Wegner et al., 2010), by crossing them for more than ten generations into the C57/BL6 background. All procedures in mice were approved by the University of Alberta Animal Care and Use Committee (protocol number 545). Genotyping and immunofluorescence studies were performed as described previously (Wegner et al., 2010). Renal cortex was isolated, minced with a razor blade and incubated for in RPMI 1640 medium containing 10 nM calyculin A and 1 mg/ml collagenase IV (Worthington, Lakewood, NJ) for 1 h at 37°C. Glomeruli were then isolated with the sieving technique as described previously (Skorecki et al., 1983). Contaminating tubules were removed by differential adhesion to cell culture plastic during two successive 10-min periods of incubation in RPMI 1640 medium containing 0.5% FBS and 10 nM calyculin A. The final glomerular preparations were >99% pure. Glomeruli were then sedimented by centrifugation at 1000 g for 5 min and resuspended in lysis buffer (50 mM Tris-HCl pH 7.5, 150 mM NaCl, 1% Nonidet P40, 0.5% sodium deoxycholate, 1× proteinase inhibitor mix, 1× PhosStop and 100 nM Calyculin A). After incubation on ice for 15 min, glomeruli were gently homogenized by passing them through a 28-gauge needle three times. The detergent-insoluble fraction was pelleted by centrifugation at 18,000 g for 10 min. Detergent soluble and insoluble fractions were resuspended in Laemmli buffer, boiled for 5 min and processed for western blot analysis. For NHERF2 studies, renal cortex was isolated from the renal medulla, finely minced with a razor blade and suspended in calyculin-A-containing lysis buffer as for glomeruli. The cortex was homogenized (PowerGen, Fisher Scientific). The material was then subjected to centrifugation at 14,000 g for 30 min at 4°C. The detergent-soluble and -insoluble fractions were then suspended in Laemmli buffer, boiled for 5 min and processed for western blot analysis.

The HA–PI4P5 and HA–PI5P4 kinase constructs were kindly provided by Greg Longmore (Washington University School of Medicine, St Louis, MO).