The chloride intracellular channel protein CLIC4 inhibits filopodium formation induced by constitutively active mutants of formin mDia2

Chloride intracellular channel 4 (CLIC4) functions in diverse actin‐dependent processes. Upon Rho activation, CLIC4 reversibly translocates from the cytosol to the plasma membrane to regulate cell adhesion and migration. At the plasma membrane, CLIC4 counters the formation of filopodia, which requires actin assembly by the formin mammalian Diaphanous (mDia)2. To this end, mDia2 must be activated through conversion from the closed to the open conformation. Thus, CLIC4 could harness the activation or the open conformation of mDia2 to inhibit filopodium formation. Here, we find that CLIC4 silencing enhances the filopodia induced by two constitutively active mDia2 mutants. Furthermore, we report that CLIC4 binds the actin‐regulatory region of mDia2 in vitro. These results suggest that CLIC4 modulates the activity of the open conformation of mDia2, shedding new light into how cells may control filopodia.

Contrary to the well-characterized mDia2 activation mechanism involving the conversion from the closed to the open conformation, the regulation of active mDia2 remains poorly understood. In this regard, it is known that profilin 1 increases the processivity and the actin-filament elongation rate of mDia2 in vitro [24]. Furthermore, recent data show that CLIC4 restrains mDia2driven filopodium formation by restricting filopodial tip protrusion [10]. This raises the possibility that CLIC4 could counter the activation or the open conformation of mDia2. Here, we have tested this hypothesis by leveraging point mutations and deletions that relieve mDia2 autoinhibition. We show that CLIC4 restricts the formation of filopodia induced by two constitutively active mDia2 mutants and also binds to the actin-regulatory FH1-FH2 region of mDia2 in vitro. These results jointly suggest that CLIC4 acts on the open conformation of mDia2 to inhibit filopodium formation.

Cell Culture, knockdowns, and transfections
HeLa cells were grown in Dulbecco's modified Eagle's medium (DMEM; Thermo Fisher Scientific, Waltham, MA, USA) supplemented with 10% FBS under 5% CO 2 at 37°C. Stable CLIC4 knockdown HeLa cells were previously described and characterized [8,10]: In brief, shRNA TRCN0000044360 (#3) and TRCN0000044362 (#5) (TRC human shRNA library; Sigma) targeting human CLIC4, or the corresponding empty lentiviral vector (control knockdown) was used, along packaging vectors, to generate lentiviral particles upon transient transfection in HEK293T cells. Viruses were collected 48 h after transfection. CLIC4 stable knockdown and control cells were selected and maintained in medium with 2 µgÁmL À1 puromycin. Plasmid transfections were performed with X-tremeGene 9 (Roche) reagent according to the manufacturer's instructions.

Pull-down assays
Purified recombinant proteins containing the GST moiety were immobilized on Glutathione Sepharose Ò 4B beads (GE Healthcare) and used as baits. Purified recombinant proteins without GST tag were used as preys. The pulldown assays were performed in NET buffer [16] supplemented with 1 mM DTT and EDTA-free protease inhibitor cocktail as previously described [22,26,27].

Western blotting
The pulled-down material was solubilized in Laemmli sample buffer (2% SDS, glycerol 10%, 62.5 mM Tris/HCl, pH 6.8, 5% 2-mercaptoethanol, and 0.002% bromophenol blue), boiled at 100°C for 5 min, separated by SDS/PAGE, and then transferred on a nitrocellulose membrane (0.2 µm pore size; Whatman, GE Healthcare Europe). After transfer, membranes were blocked in nonfat dry milk and incubated with primary antibodies according to the manufacturer's

Imaging and analysis of filopodia
Control KD and CLIC4 KD cells were plated on collagencoated coverslips and, one day later, starved overnight in DMEM supplemented with 0.1% FBS. Cells were fixed in 4% PFA dissolved in PEM buffer [28,29], permeabilized, and stained as previously described [17]. A CLSM Leica TCS SP5 operated with Leica Confocal Software (LAS-AF; Leica Microsystems, Wetzlar, Germany) and set on the sequential mode was used to acquire all images.
Filopodium length was measured manually using IMAGEJ (https://imagej.nih.gov) as explained previously [10]. For quantification of length, we considered genuine filopodiaonly finger-like structures of at least 1.5 µm [22]. Filopodium density (ratio between the number of filopodia and the total cell perimeter in a given image) was determined using FiloQuant ImageJ plugin as explained previously [10]. For quantification of density, no cutoff was set.

Densitometry
To estimate band intensities, nonsaturated exposures of western blots were subjected to densitometric analyses using IMAGEJ (https://imagej.nih.gov). Normalized CLIC4 expression was determined as follows: The intensities of CLIC4 and of a housekeeping protein (e.g., actin) were determined by densitometry, and the CLIC4/housekeeping protein ratio was calculated for each lane. These values were then normalized taking that of the control knockdown lane as a reference. Normalized CLIC4 binding was determined as follows: The intensities of CLIC4 were determined by densitometry, and the obtained values were then normalized taking that of the input lane as a reference.

Statistical analyses
For determination of statistical significance, unpaired twotailed Student's t-test and one-way ANOVA were performed using GRAPHPAD PRISM 7 software San Diego, CA, USA. Significance values are compared either with control conditions or with each other.

CLIC4 restricts filopodium formation induced by constitutively active mDia2 mutants
Mutation of methionine 1041 within the DAD (M1041-to-A mutation, hereafter referred to as MA) is known to prevent the DID-DAD interaction and mDia2 autoinhibition [30] (Fig. 1). Consistently, expression of mDia2 MA has been shown to induce the formation of filopodia in a variety of cell lines [15][16]30,31] (Fig. 1). The same phenotype is elicited also by the isolated FH1-FH2 of mDia2, which is constitutively active in vitro and cannot bind to Rho-GTPases [10,18,22] (Fig. 1).
mDia2 MA and the FH1-FH2 of mDia2 were transfected in control and CLIC4 knockdown HeLa cells ( [8,10] and Fig. S1), the latter forming longer and  . Ex post sample size calculations performed using the measured mean densities and SDs, and setting P < 0.05, gives a power > 95%. One-way ANOVA with Tukey's multiple comparisons test (*P < 0.05, **P < 0.01, and ****P < 0.0001, ns, nonsignificant).  [10]. To carry out reliable single-cell morphometric analyses, we compared only cells that expressed these constructs at similar low anti-Flag intensity levels. We observed that mDia2 MA stimulated the formation of filopodia in control knockdown HeLa cells and was enriched at their tip ( Fig. 2A), as previously reported [10]. Filopodial density was higher in mDia2 MA-expressing cells compared with mock-transfected control cells, whereas filopodia were only slightly longer (Fig. S2A-B). mDia2 MA-dependent filopodia could be observed also upon silencing of CLIC4 ( Fig. 2A), with their length and density being significantly increased compared with those measured in control knockdown cells (Fig. 2B). These observations suggest that CLIC4 regulates the open conformation of mDia2 and confirm that CLIC4 restrains the formation of mDia2-driven filopodia [10].
We sought to strengthen this notion and used the FH1-FH2 of mDia2, which is both constitutively active and irresponsive to Rho-GTPases (Fig. 1). This eliminates the potential effects that the increased basal activity of Rho and Cdc42 in CLIC4 knockdown cells may exert on both mDia2 MA and filopodium formation [10]. Expression of FH1-FH2 in control knockdown cells increased both filopodial density and length with respect to mock-transfected cells (Fig. S2A,C). These results are in line with the ability of the FH1-FH2 to stimulate actin nucleation and elongation in vitro, and filopodium formation in cells [10,18,22]. The FH1-FH2 showed enrichment at the tip of filopodia like mDia2 MA, although most of the protein was in the nucleus (Fig. 3A), as previously reported [22]. More importantly, the filopodia triggered by the FH1-FH2 of mDia2 were longer in CLIC4 knockdown than in control cells (Fig. 3A,B). At variance with mDia2 MA, CLIC4 did not affect the density of FH1-FH2-induced filopodia in the CLIC4 knockdown cells (Fig. 3B). This and the observation that the filopodia induced by the FH1-FH2 are denser than those induced by mDia2 MA suggest that CLIC4 regulates mDia2 at multiple levels. In any case, these data show that CLIC4 can harness the activity of mDia2 after conversion to the open conformation. Moreover, they rule out that the previously reported filopodial phenotypes of the CLIC4 knockdown cells were merely due to the increased basal activity of Rho and Cdc42 [10].

CLIC4 binds to the FH1-FH2 of mDia2 in vitro
In line with this view, we found that purified recombinant CLIC4 and the FH1-FH2 region of mDia2 formed a complex in vitro (Fig. 4A). Parallel experiments conducted with CLIC4 C35A showed that this interaction does not require cysteine 35 (Fig. 4B), which is instead critical for profilin 1 binding and CLIC4 translocation to the plasma membrane [10]. However, we could not observe co-immunoprecipitation between endogenous mDia2 and CLIC4 in HeLa cells (data not shown) and previous studies failed to detect an association between CLIC4 and mDia2 in cell lysates [15,16]. Nevertheless, this interaction can potentially contribute to the activity of CLIC4 in filopodia as further explained below.

Discussion
Here, we have found that CLIC4 acts on the open conformation of mDia2 to inhibit filopodium formation. To this end, we have exploited two mDia2 mutants that are not autoinhibited, mDia2 MA and the isolated FH1-FH2 region. We showed that both mutants induce longer filopodia in cells silenced for CLIC4. While mDia2 MA-dependent filopodia are denser in the CLIC4 knockdown cells compared with the control ones, this is not the case for those induced by the FH1-FH2. This and the observation that the density of the FH1-FH2-induced filopodia is higher than that of the mDia2 MA-induced filopodia raise the enticing possibility that the disruption of the DID-DAD interaction is insufficient for full activation of mDia2. Detailed structure-function analyses of full-length mDia2 are Interestingly, we also found that CLIC4 binds physically to the FH1-FH2 of mDia2 and this interaction does not require cysteine 35. The lack of co-immunoprecipitation between endogenous CLIC4 and mDia2 along with proteomics studies listing profilins, but not CLIC4, in the mDia2 interactome [15,16] suggests that the CLIC4-mDia2 association occurs only at concentrations that can be achieved on intact membranes and/or is regulated by post-translational modifications in mammalian cells. Given that weak interactions control the activities of CLIC4 and mDia2 [3,10,18], we speculate that the CLIC4-mDia2 association will be functionally important at membrane sites of high local concentration of both proteins, such as the tip of filopodia. It is also intriguing that ubiquitination of the FH1 and FH2 domains of mDia2 has been reported in vivo [15,32] and future work should explore the role of mDia2 ubiquitination in CLIC4 binding and filopodium formation.
In summary, these results suggest that CLIC4 modulates the function of the open conformation of mDia2 and pave the way for future studies that may shed new light into how cells control filopodia.

Supporting information
Additional supporting information may be found online in the Supporting Information section at the end of the article. Fig. S1. Validation of CLIC4 knockdown.