Antisense Suppression of the Chloride Intracellular Channel Family Induces Apoptosis, Enhances Tumor Necrosis Factor α-Induced Apoptosis, and Inhibits Tumor Growth

Chloride intracellular channel (CLIC) family of proteins (p64, CLIC1-5, and parchorin) is frequently localized to intracellular organelles in multiple cell types. The putative chloride ion gating activity of some members of this family suggests that CLIC proteins function to regulate organellar volume, ionic homeostasis, and pH (1). CLIC1 to CLIC5 are similar in size and highly homologous, whereas p64 and parchorin have distinct NH₂-terminal domains but share strong sequence similarity to the other family members in the COOH terminus (CLIC module; ref. 2). Common to all members is a hydrophobic region in the CLIC module consistent with a transmembrane domain, although CLICs are also found in a soluble form in the cytoplasm (2–5). Crystallographic analysis of the structure of soluble CLIC1 indicates homology to the glutathione transferase family of proteins. It is hypothesized that soluble CLICs may become activated as anion channels or channel regulators when “autoinserted” into intracellular membranes (6).

Among the CLIC family proteins, the biological functions of CLIC4 have been most thoroughly studied. CLIC4 is expressed in many cell types. In skin keratinocytes, CLIC4 was first localized to mitochondria and cytoplasm and later was localized specifically to the inner mitochondrial membrane by immunogold electron microscopy (7, 8). Other reports have localized CLIC4 in the trans-Golgi network in pancreatic cells, endoplasmic reticulum in rat hippocampal HT-4 cells, and large dense core vesicles in neurosecretory cells (9–11). CLIC4 has also been associated with the actin cytoskeleton in membrane ruffles and lamellipodia. Electrophysiologic analysis suggests that CLIC4 has Cl⁻ selective channel activity (4, 10, 12, 13). CLIC4 is highly conserved in different species with nearly 95% identity in amino acid sequence indicating an important functional role in cellular physiology (8). CLIC4 associates with dynamin I, actin, tubulin, and 14:3:3 isoforms in neuronal cells, suggesting it may also play a role in cell signaling (14). This is consistent with the recently reported induction of CLIC4 in transforming growth factor-β and serum-activated human breast fibroblasts, where it was associated with transdifferentiation to myofibroblasts (15).

CLIC4 is a direct response gene for p53 transactivation, and the up-regulation of CLIC4 is strongly associated with p53-mediated apoptosis (7). CLIC4 overexpression induces apoptosis characterized by changes in the intrinsic mitochondrial apoptotic pathway such as loss of mitochondrial membrane potential, cytochrome c release, and caspase activation (7). CLIC4 also translocates to the nucleus in cells induced to undergo apoptosis by a variety of stress inducers (16), and nuclear-targeted CLIC4 is strongly proapoptotic even when the mitochondrial death pathway is inhibited by genetic deletion of Apaf1 (16). Exposure to tumor necrosis factor α (TNFα) also increases CLIC4 transcripts and protein and causes CLIC4 to translocate to the nucleus independent of p53 (8, 16).

TNFα induces apoptosis in some cell types and is in clinical trials for the treatment of certain cancers (17). The interaction of TNFα with its receptor can activate a death pathway through caspase-8 and caspase-3 leading to a cytochrome c–independent apoptotic response (18). However, TNFα can simultaneously induce an antiapoptotic response through its activation of the downstream transcription factor NFκB and subsequent induction of inhibitors of apoptosis and other NFκB response genes to blunt the apoptotic response (19). In experimental settings, this can be overcome by inhibiting the TNFα-mediated nuclear translocation of NFκB using the mutant form of the NFκB cytoplasmic binding partner IκB (20, 21). The mutant IκB (IκBsr) cannot be phosphorylated and degraded and thus does not dissociate from NFκB to allow nuclear translocation and DNA binding. Whereas this has been an effective tool to understand the antiapoptotic activity of NFκB, this antiapoptotic pathway could compromise the clinical effectiveness of TNFα as an antitumor agent (22).

The death receptor pathway together with inhibition of NFκB is considered the major route through which TNFα induces apoptosis in experimental settings, but other pathways, such as p53 and mitogen-activated protein kinases, have also been implicated in TNFα-mediated apoptosis (23). These pathways may contribute to cell killing by TNFα independently of NFκB (24, 25). Because expression and nuclear translocation of proapoptotic CLIC4 are induced by TNFα, we embarked on a study to determine where CLIC4 might fit into the TNFα proapoptotic response. We considered this an important undertaking because CLIC4 could be a collateral target in biological approaches to cancer therapy with TNFα.

Cell Culture. Tet-On U2OS cell lines were purchased from Clontech (Palo Alto, CA). Both Tet-On U2OS, p53 Tet-On SaOS cell line (26), and their derivatives were maintained in DMEM/10% fetal bovine serum. SP1 keratinocytes and its derivatives were maintained as described previously (7). Recombinant human and murine TNFα were obtained from Calbiochem (San Diego, CA).

Immunoblot Analysis. Cells were lysed into 100 μL M-Per (Pierce, Rockford, IL), and 30 μg of proteins were resolved by SDS-PAGE and transferred onto polyvinylidene difluoride membranes (Millipore, Bedford, MA). Antibodies against the COOH terminus of CLIC4 (8) were used at a 1:10,000 dilution. The following antibodies were also used: rabbit polyclonal anti-NFκB, Bax, Bcl-2, and anti-IκBα antibodies were from Santa Cruz Biotechnology (Santa Cruz, CA); anti-β-actin mouse monoclonal antibody was from Chemicon International, Inc. (Temecula, CA); peroxidase-conjugated secondary antibodies were obtained from Amersham Biosciences (Piscataway, NJ). Monospecific polyclonal antibodies recognizing CLIC1 and CLIC5 as well as recombinant CLIC1, CLIC4, and CLIC5 proteins were generous gifts from Dr. Mark Berryman (College of Osteopathic Medicine, Ohio University; ref. 27). Immunoblots were developed with SuperSignal chemiluminescent substrates (Pierce). Total Image software (Amersham Pharmacia, Sunnyvale, CA) was used to determine densitometry for protein bands.

Immunofluorescent Microscopy. Cells at the density of 6 × 10⁴ per well were seeded in 60-mm dish or 8-chamber slides and incubated in the presence or absence of TNFα for 30 minutes with null, IκBsr or Tet-Off adenovirus for the specified times. Immunostaining was done as described previously (16), and antibody dilutions were based on suggestions from the manufacturers. The stained cells were detected by Zeiss510 confocal microscope, and LSM browser was used for cropping images.

NFκB-DNA Binding Activity Assay. The activity of p65 binding to oligonucleotides containing a NFκB consensus binding site was measured by using TransAM NFκB p65 Transcription Factor Assay Kits according to the manufacturer's instructions (Active Motif, Inc., Carlsbad, CA).

Adenoviruses. IκBsr (generous gift from Dr. Dennis, Guttridge of the University of North Carolina), Tet-On and Tet-Off adenoviruses (Clontech, Palo Alto, CA) were amplified in the Molecular Biology Core Facility of National Cancer Institute-Frederick Cancer Research and Development Center. Adenoviruses were purified over two CsCl gradients, dialyzed with a TGM buffer [Tris-HCl (pH 7.5), 1 mmol/L MgCl₂, and 10% glycerol], and stored in aliquots at −70°C. Viral titer was determined by plaque assay, and the adenoviral vector without a recombinant insert was used as a viral control (Null virus).

Generation of Bovine Keratin 5–Driven Tet-Off CLIC4-Antisense Squamous Papilloma-1 Cell Line (Bk5/tTA SP1). To generate the pBk5/tTA transactivator plasmid, pBK5 (ref. 28; obtained from David Bol, MD Anderson, TX) was digested with SnaB1 and ligated with BamHI/EcoRI–digested tTA coding sequence from the plasmid pUHD15-1 (29). Recombinant constructs with correct orientation were identified by NotI and SalI digestion. SP1 cells were cotransfected with 10 μg of BK5/tTA DNA and 2 μg of the Hygromycin marker plasmids using lipofectamine (Invitrogen, San Diego, CA) for 6 hours. Hygromycin-resistant colonies were ring cloned and tested for induced regulation of a transfected Tet-On luciferase construct.

Generation of Tetracycline-Inducible CLIC4-Antisense (Tet-On or Tet-Off CLIC4-Antisense) Cell Lines. The cloning of CLIC4 and the construction of the sense CLIC4 plasmids have been described elsewhere (8). The recombinant plasmid was digested using NotI, and the resulting sense and antisense orientations of CLIC4 were subcloned into the NotI site of the expression vector pTRE2pur Vector (Clontech). To allow tetracycline regulation, ∼8 × 10⁵ Tet-On U2OS cells and p53 Tet-On SaOS cells were transfected with 0.4 μg of CLIC4-antisense pTRE2pur plasmid using Effectene Transfection Reagent (Qiagen, Chatsworth, CA). To generate Tet-inducible CLIC4-antisense SP1 cell lines, 1 × 10⁷ Bk5/tTA SP1 cells were transfected with 1 μg of CLIC4-antisense pTRE2pur plasmid. Stable transfectants were selected by limiting dilution or cloning-ring method in a medium containing 1 μg/mL puromycin, and the selected clones were analyzed for inducible suppression of CLIC4 by immunoblotting. These p53 Tet-On SaOS cell, Tet-On U2OS, and SP1 cell derivatives are designated as CLIC4AS-SaOS, CLIC4AS-U2OS, and CLIC4AS-SP1 cells, respectively. The presence of the CLIC4-antisense region was confirmed by genomic PCR analysis. The CLIC4AS-SaOS cells or CLIC4AS-U2OS cells were treated with doxycycline (800 ng/mL) or infected with Tet-Off adenovirus without doxycycline for 17 hours to induce CLIC4-antisense before treatment with test agents. The Tet-Off adenovirus method was used to avoid doxycycline in experiments comparing CLIC4-antisense and the IκBsr adenovirus.

Immunohistochemistry and Tissue BrdU and Terminal Deoxynucleotidyl Transferase (Tdt)–Mediated Nick End Labeling Staining. For tissue BrdU staining, mice were i.p. injected with 20 mg/mL of BrdU (Roche, Nutley, NJ) 1 hour before harvesting tumor tissues, and the proliferative cells from the harvested tissues were detected by 5-bromo-2′-deoxy-uridine Labeling and Detection Kit II (Roche). For tissue terminal deoxynucleotidyl transferase (Tdt)–mediated nick end labeling (TUNEL) staining, Apoptag (Intergen, Burlington, MA) was used as described by the manufacturer except metal-enhanced 3,3′-diaminobenzidine chromogenic substrate was used. Stained slides mounted and analyzed with bright-field microscopy using Leitz-DMRB (Leica, Bannockburn, IL) and OpenLab (Improvision, Lexington, MA) software.

Xenograft and Tumor Development. CLIC4AS-SP1 cells were grafted as a skin graft on the back of nude mice, and tumor size was measured as described previously (30). One group of mice was given doxycycline-containing feed (200 μg/kg w/w, Bioserve, Laurel, MD) starting 1 week before grafting, whereas another group received standard mouse feed (Purina) to allow expression of the CLIC4-antisense. In one experiment, all recipient mice received doxycycline diet 1 week before grafting and subgroups were changed to control diets at 0, 1, and 3 weeks following the application of tumor grafts. In the second experiment, recombinant TNFα (1 μg per injection per mouse) was given by i.p. injection twice per week starting at week 5 after grafting, which then continued for four more weeks before termination.

Construction of Human CLIC4 Promoter Reporter and Luciferase Assays. The cloning of the 3.5-kb human CLIC4 promoter region was described elsewhere (7). Primer sets containing NheI sites (5′NheI-gtgtaccatgagctgtcctctgagccagg-3′ and 5′NheI-ctgtgtttcaggctctgagctagcccttgg-3′) were used to PCR amplify 2.5 kb of the human CLIC4 promoter in pGEM-T Easy (7) excluding the repeated segments. This region was subcloned into pGlow-Topo (Invitrogen), cut with NheI, and subcloned into a NheI-linearized pGL3 (Promega, San Luis Obispo, CA) vector. Osteosarcoma cells were transfected for 17 hours with either the 1 μg/mL of CLIC4 reporter or NFκB reporter plasmid (31; a gift from Dr. Zheng-Gang Liu, National Cancer Institute) using effectene transfection reagent (Qiagen) before treatment with IκBsr and/or TNFα, and luciferase activity was measured by using the luciferase reporter assay kit (Clontech). Results were normalized to the total protein content.

Annexin V and TUNEL Fluorescence-Activated Cell Sorting Analysis. Control or treated cells (n = 2 × 10⁵) were analyzed by allophycocyanin-conjugated recombinant human Annexin V (Caltag, Burlingame, CA) and In situ Cell Death assay (Roche) as described by the manufacturer. Analysis was done after 10,000 counting events. Data acquisition and analysis were done using Cell Quest software. The apoptosis data shown are one of representative results of at least three independent experiments.

Statistical Analysis. Comparisons of experimental data were analyzed by a two-tailed Student's t test. P < 0.05 was considered to indicate a statistically significant difference.

Suppressing NFκB Enhances TNFα-Mediated Apoptosis but Reduces CLIC4. Human osteosarcoma (p53 Tet-On SaOS and Tet-On U2OS) cells were treated with TNFα to activate NFκB and/or with IκBsr (the dominant-negative mutant of IκB) adenovirus to inhibit NFκB activity selectively. NFκB translocated to the nucleus within 30 minutes of TNFα treatment in both osteosarcoma cells that are infected with empty (null) adenovirus (Fig. 1A), and IκB fluorescence was noticeably reduced after TNFα treatment (Fig. 1A). TNFα treatment also caused translocation of CLIC4 to the nucleus as reported previously (16). Expression of CLIC4-antisense (indicated by T in Fig. 1B) did not prevent nuclear translocation of CLIC4 or NFκB in response to TNFα. Introduction of IκBsr (Fig. 1C), or the combination of both CLIC4-antisense and IκBsr (Fig. 1D) did not prevent TNFα-mediated translocation of CLIC4 but did prevent NFκB translocation to the nucleus, suggesting CLIC4 and NFκB nuclear trafficking are mediated by independent mechanisms. IκBsr prevented translocation of NFκB at multiplicity of infection (MOI) of 50 in both cell types (Fig. 1C), indicating that IκBsr adenovirus was effective in blocking NFκB activation in these cells.

For both cell lines, flow cytometry data showed a small increase in Annexin-positive apoptotic cells after treatment with 25 ng/mL TNFα (Fig. 2A and B). This increase in apoptosis is independent of p53 because the p53 Tet-On SaOS cells lack p53 in the absence of doxycycline. TNFα also induced CLIC4 protein (Fig. 2C and D) and caused a modest increase in CLIC4 promoter activity (Fig. 2E and F) in both cell lines. Combined treatment of these cells with TNFα plus IκBsr caused a substantial increase in apoptotic cells but decreased CLIC4 protein levels (Fig. 2A-D). CLIC4 promoter activity was not significantly changed from TNFα alone with the addition of IκBsr (Fig. 2E and F).

Expression of Stably Integrated Conditional CLIC4- Anti-sense Enhances TNFα-Mediated Apoptosis. To determine if reduction of CLIC4 enhanced TNFα-induced apoptosis as seen for inhibition of NFκB activation, CLIC4AS-SaOS and CLIC4AS-U2OS cells were infected with the Tet-Off adenovirus to induce the antisense vector in the presence or absence of TNFα. Treatment with TNFα or Tet-Off adenovirus alone caused apoptosis in CLIC4AS-SaOS cells, and TNFα alone induced apoptosis in CLIC4AS-U2OS cells (Fig. 3A and B). The Annexin V binding was associated with condensed chromatin and nuclear fragmentation as detected by Hoechst staining (data not shown). Immunoblots of the duplicate experiments showed that TNFα alone increased the level of CLIC4 protein in both cell types (Fig. 3C and D). The combination of TNFα and antisense expression enhanced apoptosis in both cell types, and antisense expression prevented the increase in TNFα-mediated CLIC4 protein expression in both cell types (Fig. 3B and D). Apoptosis measured in this setting was dependent on the adenoviral MOI used, suggesting that TNFα and CLIC4-antisense expression synergize to enhance apoptosis. In CLIC4AS-U20S cells, apoptosis induced by TNFα progressed more slowly (48 hours) than in CLIC4AS-SaOS cells, and more adenovirus (5 MOI of Tet-Off adenovirus) was needed to achieve an apoptotic response in 65% of treated cells (Fig. 3B).

CLIC4-antisense induced by the Tet-Off adenovirus was not influencing NFκB nuclear translocation (Fig. 1B), but as an independent approach, NFκB binding activity assays were done after TNFα treatment of CLIC4AS-SaOS and CLIC4AS-U2OS cells (Fig. 4). NFκB in the nucleus as detected by binding to its consensus DNA binding site and NFκB transcriptional function as measured by luciferase reporter activity were not significantly affected by the Tet-Off adenovirus, but were completely inhibited by the IκBsr adenovirus (Fig. 4A-D). In both cell lines, NFκB nuclear activity was enhanced by TNFα.

CLIC4-Antisense Reduces the Expression of Other CLIC Proteins. The extensive sequence homology among the CLIC family proteins (32) prompted us to examine if our reverse orientation full-length CLIC4-antisense could also reduce the expression of other CLIC proteins. Antibodies to CLIC1, CLIC4, and CLIC5 detected single bands on immunoblots of the respective recombinant proteins, although there was slight cross-reactivity of COOH terminus CLIC4 antibody with CLIC5-recombinant protein (Fig. 5A). However, CLIC5 protein is in extremely low abundance in the cell lines used in these studies and requires ultra sensitive reagents for detection (see legend of Fig. 5B); thus, essentially each antibody is monospecific for cell lysates. Upon induction of CLIC4-antisense with addition of doxycycline in the cell culture medium, the levels of all three CLIC proteins decreased indicating that CLIC4-antisense has the broader capability of reducing expression of multiple CLIC family members. To determine if CLIC4 antisense had a more general influence on reducing cellular protein expression, the blot shown in Fig. 5B was reprobed for Bax, another proapoptotic protein associated with mitochondria and Bcl-2, an antiapoptotic protein. In neither case was the protein level reduced by CLIC4-antisense expression (Fig. 5C).

Reduction of CLIC Proteins Is as Effective as Inhibition of NFκB for Enhancing TNFα-Induced Apoptosis in Osteosarcoma Cell Lines. To compare the effects of inhibiting NFκB activity and suppressing CLIC4 expression on TNFα-mediated apoptosis, we treated CLIC4AS-SaOS cells with either IκBsr or Tet-Off adenovirus together with TNFα. After incubation with 25 ng/mL TNFα for 12 or 24 hours in combination with IκBsr, 32% and 55% of the cells underwent apoptosis, whereas a 30% and 66% apoptotic response was measured for Tet-Off adenovirus and TNFα treatment at these time points (Fig. 6A). Thus, reduction of CLIC4 (together with other CLIC family members) is as effective for enhancing TNFα-mediated apoptosis as inhibiting NFκB in this cell line. In TNFα dose-response studies, reduction of CLIC4 was more effective than suppression of NFκB (Fig. 6B) in CLIC4AS-SaOS cells. In CLIC4AS-U2OS cells, both NFκB inhibition and CLIC reduction enhanced TNFα apoptosis at all TNFα doses tested, and suppression of NFκB may be more effective in these cells (Fig. 6C).

CLIC4-Antisense Expression Inhibits Tumor Growth In vivo. The reduction in viability associated with CLIC4-antisense expression in osteosarcoma tumor cell lines prompted us to test the possibility that reduction in CLIC family proteins might be inhibitory to tumor growth in vivo. We created a mouse tumor cell line CLIC4AS-SP1 that expresses abundant CLIC4 in the presence of doxycycline, but substantially reduced the level of CLIC4 when doxycycline is withdrawn or the dose is reduced (Fig. 7A and B). CLIC4AS-SP1 cells were then grafted to nude mice that had been primed with doxycycline diet for 1 week or maintained on a standard diet. Tumor growth was markedly retarded on mice receiving standard diet (Fig. 7C). Withdrawal of doxycycline from the diet at 3 weeks when tumors were well established produced tumors of intermediate size compared with control mice maintained on doxycycline and immediate doxycycline withdrawal, suggesting CLIC4-antisense expression affected growth even in well-established tumors. Three independent grafting experiments showed identical results. All tumors were poorly differentiated squamous cell carcinomas independent of expression of the CLIC4-antisense. Analysis of CLIC4 protein in a series of tumors excised from the experimental groups indicated that tumors expressing the antisense had significantly reduced CLIC4 levels (Fig. 8A and B), whereas the parental tumor and the CLIC4AS-SP1 tumors from mice maintained on doxycycline had CLIC4 levels similar to normal skin keratinocytes. Immunostaining of tumor sections revealed that CLIC4 was relatively abundant in the cytoplasm of tumors not expressing the antisense but substantially reduced in tumors in which doxycycline was withdrawn (Fig. 9A). Furthermore, proliferation in the antisense expressing tumors was reduced by 65% when assayed by BrdU staining (Fig. 9B), although a substantial number of proliferating cells were detected. The number of apoptotic cells detected by TUNEL staining was 3-fold higher in antisense expressing tumors than in tumors in which antisense was suppressed by doxycycline (Fig. 9C). Together, these changes are consistent with a suppressed tumor expansion, reduced proliferation and increased apoptosis when CLIC proteins are reduced.

To determine if the addition of TNFα would further reduce tumor size, mice carrying tumor grafts on doxycycline or standard diets were treated with recombinant mTNFα after appearance of tumors at 5 weeks after grafting (Fig. 10A). Whereas TNFα and CLIC4-antisense both had a substantial antitumor effect as a single agent, combination therapy using TNFα and antisense modestly improved the tumor response for at least 3 weeks after TNFα was given (Fig. 10A). CLIC4-antisense alone had a more powerful tumor inhibitory influence than TNFα as a single therapeutic agent in vivo in this model. In contrast, the apoptotic response of CLIC4AS-SP1 cells to combined antisense expression and TNFα treatment in vitro was synergistic (Fig. 10B), similar to that seen with the human osteosarcoma tumor cell lines.

This study was designed to determine if induction of CLIC4 by TNFα was an essential component of the TNFα-mediated apoptotic response, as we have previously shown for p53-mediated apoptosis (7). Based on the results of the CLIC4 promoter-reporter construct, CLIC4 does seem to be a direct transcriptional target for TNFα independent of NFκB stimulation. However, the up-regulation of CLIC4 is not required for TNFα-mediated apoptosis, and in contrast, reduction of CLIC4 enhances TNFα-mediated apoptosis. This conclusion is tempered somewhat since our antisense approach involved reduction of several CLIC family proteins whereas not affecting other cellular proteins that we have tested. Thus, until more selective reagents for individual CLIC family proteins are available, we conclude that reductions in CLIC proteins together can enhance TNFα-mediated apoptosis independent of p53.

A critical amount of CLIC proteins is needed to maintain cell survival. If the CLIC family proteins are required to maintain ionic balance, pH, and volume in cellular organelles, one would expect precise regulation of protein levels to be required. The inhibition of NFκB in concert with TNFα treatment reduces CLIC4 protein without altering CLIC4 transcription. This suggests that NFκB-regulated factors may contribute to the stability of CLIC proteins, and this could be a component of NFκB antiapoptotic activity. This possibility will require additional studies because the processing of CLIC proteins has not yet been explored.

We considered that suppressing CLIC4 by antisense expression could have an influence on TNFα-induced NFκB activation. However, suppressing CLIC4 did not inhibit TNFα-induced translocation of NFκB to the nucleus, DNA binding activity of NFκB, or NFκB-mediated transcriptional activation, indicating that inhibition of NFκB activation is not one of the mechanisms by which CLIC4 suppression sensitizes cells to TNFα-mediated apoptosis. Reduction in CLIC proteins is as potent as inhibition of NFκB function for enhancing TNFα-mediated apoptosis in human osteosarcoma cells. In CLIC4AS-SP1 mouse cells, the combined influence of TNFα and reduction in CLIC proteins seems to be synergistic for apoptosis.

Whereas our use of antisense methods to study reduction in CLIC4 results in a broader reduction in CLIC family proteins, CLIC4 itself is perhaps the most ubiquitous and abundantly expressed family member and likely to be central to the results reported here. Nevertheless, the development of selective CLIC4 modulating agents will be essential to sorting out specific action of individual family members. A previous report (7) and current results consistently indicate that an increase or reduction of CLIC4 impairs cell viability through an apoptotic pathway. This may be unusual but is not unique. For example, C-MYC up-regulation induces apoptosis in fibroblasts, whereas down-modulation induces apoptosis in hematopoietic cells (33, 34).

The physiologic significance of CLIC4 induction and nuclear translocation in TNFα-treated cells is still unclear. It is possible that induction and translocation of CLIC4 might contribute to another physiologic function of TNFα such as differentiation or growth inhibition. The development of the Caenorhabditis elegans excretory canal requires the function of EXC-4, the worm orthologue of mammalian CLIC proteins (35), suggesting CLIC proteins may also have a role in development or tissue remodeling.

The human CLIC4 gene is mapped to chromosome 1p36.11, a region that is frequently altered in cancers, and previous studies have shown aberrations at this locus in lymphomas, leukemias, invasive ductal and lobular breast carcinomas, metastasizing squamous cell carcinomas, and lung cancer (comparative genomic hybridization and loss of heterozygosity database, National Center for Biotechnology Information). The generation of human tumor cell lines where we can conditionally regulate CLIC expression has provided evidence that CLIC proteins could be useful targets for tumor therapy. This prediction is further supported by the in vivo grafting model of a mouse squamous cell tumor line where conditional expression of CLIC4-antisense reduces CLIC4, inhibits tumor proliferation and expansion of tumor size, and increases tumor cell apoptosis. These results indicate that a pan-CLIC knockdown could serve as a novel anticancer therapy. In vitro studies suggest that lowering CLIC levels could significantly enhance the apoptotic activity of TNFα, independent of NFκB or 53 pathways. The antitumor response to combined treatment was not as potent as the proapoptotic response in vitro; although there was a modest combined effect in vivo. Because only one regimen of TNFα administration was tested, additional modification of the protocol may produce a more pronounced result. This preclinical model suggests that reduction of CLIC levels, physiologically translated to reduced chloride channel activity, could enhance biological or cytotoxic drug–mediated antitumor therapy, but further testing is required to confirm this possibility.

Ion channels and transporters have been among the most successful therapeutic targets (i.e., Ca²⁺ channel blocker for heart disease and Na⁺ transport/exchanger inhibitors for diuretics) for pharmaceuticals. Previous experimental studies indicate that a chloride channel blocker (i.e., tamoxifen), calcium channel inhibitor (i.e., verapamil) or sodium/hydrogen exchanger inhibitor (i.e., amiloride) enhance the effectiveness of cancer chemotherapy by interfering with the activity of multidrug resistance proteins that influence the ability of chemotherapeutic agents to accumulate in cancer cells in sufficient concentrations. From a therapeutic point of view, ion transport inhibitors such as intracellular chloride channel inhibitors can serve as modifiers of drug traffic across the plasma membrane. Recent reports suggest that multiple classes of intracellular chloride channel proteins are central to cell viability and thus should be considered as potential therapeutic targets in cancer. Voltage-gated chloride channels CLC-2, CLC-3, and CLC-5 channels are expressed at high levels in acute patient biopsies from low- and high-grade malignant gliomas (36). Glycine inhibits the growth of B16 melanoma tumors in vivo through a glycine-gated Cl⁻channel in endothelial cells that blocks the effect of vascular endothelial growth factor on blood vessel growth (37), suggesting that chloride channels may influence angiogenesis. CLIC4 expression is increased in myofibroblasts that form the stroma in breast cancers, participating in the regulation of the tumor microenvironment (15). Calcium-activated chloride channels, CLCA1 and CLCA2, are significantly down-regulated in ∼80% of colorectal carcinomas and >90% of highly proliferating tumor cells, suggesting a tumor suppressor activity (38, 39). Reconstitution of CLCA2 expression in highly malignant cell lines reduced Matrigel invasion in vitro and prevented the growth or metastasis of tumor cells transplanted s.c. in nude mice (40, 41). It is anticipated that reduction of CLIC levels could provide an advantage in increasing the sensitivity of tumor cells to drug-mediated anticancer therapy, and further clarification of CLIC functions will be needed for a development of small molecules to modulate CLIC activities. These molecular characteristics of intracellular chloride channels suggest that chloride gradients and flux influence a variety of important cellular controls for proliferation, migration, adhesion, and viability that are involved in cancer pathogenesis, and CLIC proteins should be considered as potential molecular targets for cancer.

Grant support: NIH JSPS Research Fellow in Biomedical and Behavioral Research, 2003 to 2005 (M. Mutoh).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank Drs. Karen Vousden and Kevin Ryan of the National Cancer Institute for supplying the SaOS cell lines; Dr. Zheng-Gang Liu for supplying the NFκB-luciferase plasmid DNA; Dr. Dennis Guttridge of the University of North Carolina for supplying the IκBsr adenovirus; Dr. Narayan Bhat of the Science Applications International Co., Inc., Frederick, National Cancer Institute, Molecular Biology, Gene Expression Laboratory for adenoviral amplification; Barbara Taylor of the CCR FACS Core Facility; Dr. Mark Berryman for the generous contribution of antibodies against CLIC1 and CLIC5 and recombinant CLIC proteins; and Bettie Sugar for the excellent editorial assistance.