ClC-3 expression enhances etoposide resistance by increasing acidification of the late endocytic compartment

Resistance to anticancer drugs and consequent failure of chemotherapy is a complex problem severely limiting therapeutic options in metastatic cancer. Many studies have shown a role for drug efflux pumps of the ATP-binding cassette transporters family in the development of drug resistance. ClC-3, a member of the CLC family of chloride channels and transporters, is expressed in intracellular compartments of neuronal cells and involved in vesicular acidification. It has previously been suggested that acidification of intracellular organelles can promote drug resistance by increasing drug sequestration. Therefore, we hypothesized a role for ClC-3 in drug resistance. Here, we show that ClC-3 is expressed in neuroendocrine tumor cell lines, such as BON, LCC-18, and QGP-1, and localized in intracellular vesicles colabeled with the late endosomal/lysosomal marker LAMP-1. ClC-3 overexpression increased the acidity of intracellular vesicles, as assessed by acridine orange staining, and enhanced resistance to the chemotherapeutic drug etoposide by almost doubling the IC₅₀ in either BON or HEK293 cell lines. Prevention of organellar acidification, by inhibition of the vacuolar H⁺-ATPase, reduced etoposide resistance. No expression of common multidrug resistance transporters, such as P-glycoprotein or multidrug-related protein-1, was detected in either the BON parental cell line or the derivative clone overexpressing ClC-3. The probable mechanism of enhanced etoposide resistance can be attributed to the increase of vesicular acidification as consequence of ClC-3 overexpression. This study therefore provides first evidence for a role of intracellular CLC proteins in the modulation of cancer drug resistance. [Mol Cancer Ther 2007;6(3):979–86]

Metastatic cancers are usually treated using chemotherapy regimens to target tumor cells disseminated throughout the organism. Yet, although the initial response to treatment might be good, multidrug resistance eventually emerges, leading to treatment failure and death. Several mechanisms are responsible for the development of drug resistance at a cellular level. Although it has been shown that target modification can lead to drug resistance, it is frequently the emergence of expression of ATP-binding cassette transporters, responsible for drug efflux from cancer cells, that actually leads to resistance (1). Proteins of this family have also been shown to sequester drugs in intracellular membrane compartments, thereby decreasing effective drug concentrations and inducing resistance (2).

Intracellular acidic compartments can sequester basic anticancer drugs, which accumulate passively in response to the pH gradient, and thus contribute to drug resistance. Several studies have implied a link between drug resistance and intravesicular acidity either by comparison of drug-sensitive and drug-resistant cell lines (3, 4) or through treatment with agents disrupting organellar pH (5). A recent study showed that omeprazole, a proton pump inhibitor used in antiacid treatment of peptic disease, and the vacuolar proton pump (v-H⁺-ATPase) inhibitor bafilomycin A increased cytotoxicity of the basic chemotherapeutic drugs doxorubicin and mitoxantrone (6). Similarly, in vivo studies of human melanoma xenografts in severe combined immunodeficient mice showed that proton pump inhibitors affected sensitivity to cisplatin (7). Moreover, a study with a novel v-H⁺-ATPase inhibitor showed an increased effectiveness of topotecan chemotherapy when combined with this inhibitor, supporting the hypothesis of a role for intracellular acidic compartments in mediating chemotherapy resistance (8). Vesicular proton pump inhibition might therefore be a future target option for increasing the effectiveness of cancer chemotherapy.

Although these studies have focused on the role of intracellular proton pumps and transporters for chemotherapy resistance, the presence also of an anion shunt current is a requisite for the accumulation of protons in intracellular membrane compartments. This current is necessary to dissipate the charge gradient generated by the electrogenic activity of the v-H⁺-ATPase during the H⁺ transport. Chloride channels/transporters of the CLC family have been suggested to provide the anion shunt current (9). We therefore tested the hypothesis of an involvement of intracellular CLC channels/transporters in drug resistance.

ClC-3, a member of the CLC family of chloride channels and transporters, is expressed predominantly in the cells of the nervous system and is located in acidic intracellular compartments (10, 11). ClC-3 has been suggested to generate a shunt current of chloride for v-H⁺-ATPases, thereby aiding the acidification of endosomes and synaptic vesicles (10) as well as lysosomes (12). This hypothesis was confirmed in experiments directly showing a role for ClC-3 in endosomal acidification (13). The homologous CLC proteins ClC-5 and ClC-7 play also a role in modulating the pH of intracellular compartments and vesicle trafficking in the cell (9, 14). Because ClC-3 is expressed in pheochromocytoma cells (15), we hypothesized the presence of ClC-3 also in gastrointestinal neuroendocrine tumor cells.

Etoposide, in combination with a platin substance, is the mainstay of chemotherapy treatment of undifferentiated, high-grade neuroendocrine tumors (16–19) and small cell lung cancers (20). At the time of diagnosis, these tumors are mostly disseminated so that systemic chemotherapy remains as the only applicable treatment option. Although initial responses to chemotherapy treatment are usually good with significant reduction of tumor size, most of these tumors rapidly develop chemotherapy resistance. As a consequence, nearly all patients with these cancers eventually succumb to progressive disease.

As for most other anticancer drugs, etoposide has a basic pKa (9.8; ref. 21), predisposing it to trapping in acidic compartments. We therefore investigated the hypothesis that expression of ClC-3 in a neuroendocrine tumor cell line increases the acidity of intracellular compartments and thereby etoposide drug resistance.

BON is a neuroendocrine cell line established from a human pancreatic neuroendocrine carcinoma (22, 23). BON cells were cultured at 37°C in a 5% CO₂ and water-saturated atmosphere and grown in DMEM and F12 (1:1), 25 mmol/L HEPES, 10% fetal bovine serum (FBS), and 10 μg/mL ciprofloxacin (Ciprobay). LCC-18 is a cell line established from a human neuroendocrine-differentiated colonic carcinoma (24). LCC-18 cells were cultured in RPMI 1640, 10% FBS, insulin, transferrin, and selenium liquid medium supplement (Sigma, St. Louis, MO), and 10 μg/mL ciprofloxacin (Ciprobay). QGP-1 cells, a human pancreatic carcinoma cell line of islet origin (25), were cultured in DMEM and 10% FBS. PC12 cells are an established line derivative of a rat pheochromocytoma and were cultured in high-glucose DMEM, 10% FBS, and 5% horse serum. CaCo-2 is a cell line established from an adenocarcinoma of the colon and was cultured in MEM plus 10% FCS. The BON cell clone permanently overexpressing ClC-3-green fluorescent protein (GFP) was generated by transfection with the pCIneo plasmid coding for the fusion protein and selection in G418 following a similar procedure used to generate the HEK293 clone expressing ClC-3-GFP (11). HEK293 clone expressing ClC-3-GFP was cultured in DMEM, 10% FBS, and 500 μg/mL G418. NIH3T3-MDR1 cell line (26), a derivative of the mouse fibroblast NIH3T3 line, permanently transfected with the human MDR1 gene coding for P-glycoprotein, was grown in DMEM, 10% FBS, and 1 μg/mL colchicine. SW-620 cell line, derived from a human colon adenocarcinoma (27), was grown in RPMI 1640 and 10% FBS.

Cells, cultured as described above, were homogenized and crude membrane preparation was obtained (see Supplementary Data for a detailed description).³

Membrane proteins were resolved by SDS-PAGE and transferred onto polyvinylidene difluoride membranes (Immobilon-P, Millipore, Billiberia, MA). ClC-3 proteins were detected with the D1-specific polyclonal antibody against ClC-3 (1:100 overnight at 5°C; ref. 11), P-glycoprotein was detected with the monoclonal antibody C219 (1:1,000 overnight; Dako, Carpinteria, CA), and the multidrug-related protein-1 (MRP-1) was detected with the monoclonal antibody MRPm5 (1:50 overnight; Abcam, Cambridge, United Kingdom) followed by appropriate horseradish peroxidase secondary antibodies (1:1,000; Dako). Signal was visualized by chemiluminescence (Amersham ECL system; Amersham, Little Chalfont, United Kingdom). A detailed description of the Western blotting procedure is available in the Supplementary Data.³

For immunocytochemistry on BON cells permanently transfected with ClC-3-GFP, cells were plated on poly-l-lysine–coated (Sigma) glass coverslips and cultured in six-well plates for 24 to 48 h before fixation with a solution of 4% formaldehyde/4% sucrose in PBS buffer or methanol at −20°C. Cells were then permeabilized with 0.1% Triton X-100 for 4 min and costained with antibodies against chromogranin A (DAK-A3; Dako), synaptophysin (gift from C. Groetzinger, Charité University Medicine, Campus Virchow-Klinikum, Berlin, Germany), EEA-1 (1:50; Santa Cruz Biotechnology, Santa Cruz, CA), and LAMP-1 (1:50; Santa Cruz Biotechnology) for 12 h at 4°C after blocking with 0.2% fish skin gelatin in PBS. All primary antibodies were detected by appropriate secondary antibody conjugated to the fluorophore Alexa Fluor 568 (1:600; Molecular Probes/Invitrogen, Carlsbad, CA). Cells were also stained with 4′,6-diamidino-2-phenylindole (DAPI; 15 mg/mL; 1:10,000 dilution; Molecular Probes) for nuclear DNA. Coverslips were mounted in Vectashield (Vector Laboratories, Burlingame, CA) for examination and cells were imaged using a Leica SP confocal microscope equipped with a 100×/1.4 NA PlanApoChromat oil immersion objective lens (Leica, Wetzlar, Germany) as described in the Supplementary Data.³ Analysis of LAMP-1 signal at the fluorescence-activated cell sorting was conducted by fixing the resuspended cells in 70% ice-cold methanol and staining for LAMP-1 with the above described antibody (1:50) for 2 h at room temperature in 0.2% fish skin gelatin, and then, the primary antibody was recognized by goat anti-mouse Cy5-conjugated antibody (1:200; 30 min at room temperature; Abcam; see Supplementary Data for details of fluorescence-activated cell sorting settings).³ Control was obtained by exposure of the preparation to the secondary antibody only. To identify acidic compartments, live BON cells expressing ClC-3-GFP were stained for 30 min at 37°C with 50 nmol/L DND-99/LysoTracker Red, a low pH-specific fluorophore (Molecular Probes). Cells were imaged in a perfusion chamber with a Leica SP confocal microscope equipped with 63× 1.32 NA PlanApoChromat oil immersion objective, and to avoid bleed through, the fluorophores were excited sequentially as described in the Supplementary Data.³ Similarly, identification of acidic compartments was achieved by staining the cells in the perfusion chamber for 5 min with 10 μmol/L acridine orange (Molecular Probes) until organelle steady-state accumulation was reached (see Supplementary Data and the next paragraph for further details).³ All images were analyzed with MetaMorph 5.0v1 software (Universal Imaging, Downingtown, PA). Figures were assembled for publication with Adobe Photoshop 6.0 software (Adobe Systems, San Jose, CA).

Acridine orange (AO) was used to measure acidic compartments. The fluorophore accumulates in acidic compartments of cells where it remains trapped following protonation. On increasing concentrations of AO, the spontaneous formation of dimeric AO molecules leads to a shift in its fluorescence emission maximum from green to far red, thereby labeling acidic compartments in red (28). Accumulation in the acidic compartment is dependent on extracellular acridine orange concentration [AO]_ext and proportional to the magnitude of the ΔpH across the compartment membrane. When the external pH is kept constant, the [AO]_int of the acidic compartment is linearly proportional to the [AO]_ext. This was exploited to quantify the acidic compartment in large numbers of cells by using flow cytometry (see Supplementary Data³ and ref. 29 for further details). Data so acquired were analyzed off-line by FlowJo (TriStar, Inc., San Carlos, CA). In brief, the mean average intensity fluorescence signal in the FL-3 channel from a sample population of 10,000 cells was assessed for each [AO]_ext to which cells were exposed. The mean FL-3 intensity values were plotted against the [AO]_ext and fitted with a linear regression of the first order by using the Marquardt-Levenberg algorithm based on the least-square method (SigmaPlot version 8). The slopes of the relationship obtained for different treatments were statistically compared by the parallelism test (F test; significantly different for P ≤ 0.05). For each treatment, at least three independent measurements were done.

Drug cytotoxicity was assessed with the Cell Proliferation Kit II as described by the manufacturer (Roche Diagnostics, Indianapolis, IN). Cells were seeded in 96-well tissue culture plates with 100 μL of appropriate tissue culture medium. After 24 h, culture medium was exchanged with fresh medium without phenol red (to avoid interference with the absorption readings) and containing appropriate concentration of etoposide (stock solution in DMSO). Percentage of cell survival was expressed as normalized average absorption values from four replicates per each drug concentration ± SD. The concentration of drug required to decrease cell proliferation by 50% (IC₅₀) was determined for each treatment from the concentration-response curve of the 2,3-bis[2-methoxy-4-nitro-5-sulfophenyl]-2H-tetrazolium-5-carboxanilide inner salt assay and statistically compared by Student's t test. The concentration-response curves were obtained by fitting the experimental data by nonlinear regression with a logistic curve (SigmaPlot version 8.0).

Western blot analysis (Fig. 1A), using a polyclonal antibody specific against ClC-3 (11), showed endogenous expression of ClC-3 in the BON cell line derived from a human pancreatic neuroendocrine tumor and well established as a model for neuroendocrine cells (23). ClC-3 was also expressed in other human neuroendocrine cell lines derived from tumors of the gastrointestinal tract, such as LCC-18 and QGP-1. ClC-3 expression was confirmed in PC12 cells, a cell line derived from rat pheochromocytoma and previously shown to express ClC-3 (15). In contrast, the gastrointestinal tumor cell line CaCo-2, established from a human colon adenocarcinoma, showed significantly lower levels of ClC-3 expression.

To study the intracellular localization of ClC-3 in neuroendocrine cells, a clone of BON cells constitutively expressing a C-terminally GFP-tagged ClC-3 protein was generated (see Materials and Methods). ClC-3-GFP was primarily expressed in intracellular vesicles colabeled by the late endosomal/lysosomal marker LAMP-1 (Fig. 1B). No colocalization of ClC-3 was observed with markers of neuroendocrine secretory vesicles, such as chromogranin A or synaptophysin (data not shown). The late endosomal/lysosomal compartments are acidic and play a critical role in cellular metabolism as the site of protein degradation and of specialized autolytic processes. A v-type H⁺-ATPase in the vesicular membrane is responsible for establishment of the acidic intracellular pH. Because this protein is an electrogenic pump, it is generally assumed that efficient acidification also requires an anionic shunt current to abolish the charge buildup generated by the transport of H⁺ (9). Live BON cells expressing ClC-3-GFP were stained for intracellular acidic compartments using either the weak amine base acridine orange (Fig. 2A) or the red fluorescent cyanine dye DND-99/LysoTracker Red (Fig. 2B). The colocalization with fluorescent ClC-3-GFP showed that ClC-3 resides in the membrane of acidic vesicles.

To assess the acidity of vesicles overexpressing ClC-3, we used a titration approach using acridine orange as a fluorophore and flow cytometry analysis (ref. 29; see also Materials and Methods and Supplementary Data).³ Briefly, changes in acidity of intracellular organelles can be followed by measuring the change in slope of the relationship between the intensity of AO fluorescence emission in the far red (FL-3 channel) and the concentration of [AO]_ext to which the cells are exposed. To maintain ClC-3-GFP expression, cells were grown in selective medium containing G418. This has the drawback that any detectable functional difference between the overexpressing clone and the parental cell line could be either associated to the expression of ClC-3 or to unavoidable and unpredictable changes due to the selection process. To avoid this, we exploited the naturally occurring difference in ClC-3-GFP level expression within cells of the same clone. Clonal BON cells expressing ClC-3-GFP were fluorescence-sorted for populations with high and low ClC-3-GFP expression and compared for acidity of intracellular organelles. Higher ClC-3-GFP expression conferred an increase of the acidity of intracellular organelles as shown by a significantly steeper slope, in comparison with BON cells expressing low ClC-3-GFP (Fig. 2C). Previous studies have suggested that intracellular acidic compartments may sequester basic drugs, thereby decreasing effective concentrations of drug in the cell nucleus and cytoplasm and, consequently, cytotoxicity. BON cells sorted for high ClC-3-GFP expression showed, in comparison with BON cells sorted for low ClC-3-GFP expression, a significantly increased resistance to etoposide, a chemotherapeutic drug with a basic pKa (Fig. 2D). The cell growth rate was unaffected by ClC-3 expression (see Supplementary Fig. S1).³ To validate these results in a different cell model, a clone of HEK293 cells permanently expressing ClC-3-GFP was used (11). As for the neuroendocrine BON cells, HEK293 cell populations sorted for high expression of ClC-3-GFP showed a significant increase in intravesicular acidity and etoposide resistance (Fig. 3A and B, respectively).

The intensity of the red fluorescent acridine orange signal, as determined by flow cytometry, is dependent on both the acidity and number of intracellular organelles. We assessed whether ClC-3 overexpression leads to an up-regulation of the formation of late endosomes and lysosomes, which could potentially account for the differences observed. However, BON cells overexpressing ClC-3 were indistinguishable from untransfected BON cells in expression levels of the lysosomal marker LAMP-1 (Fig. 4A, right hand side peaks). Similar results were obtained in HEK293 cells overexpressing ClC-3-GFP (data not shown).

To confirm the role of acidification of vesicular compartments in etoposide resistance, BON cells overexpressing ClC-3-GFP were exposed to concanamycin A, a specific inhibitor of v-H⁺-ATPase, and then cultured in the presence of etoposide. Although concanamycin A treatment alone did not affect cell survival (see Supplementary Fig. S2),³ pretreatment with concanamycin A significantly increased cells sensitivity to etoposide (Fig. 4B). We also excluded that either the parental cell line BON cells or its derivative clone overexpressing ClC-3-GFP were expressing the multidrug resistance transporter P-glycoprotein (Fig. 5A) or MRP-1 (Fig. 5B).

Many solid tumors have an acidic extracellular environment probably due to glycolytic and anaerobic metabolism of tumor cells. This leads to intracellular acidification and increased extrusion of protons from intracellular compartments to prevent cytoplasmic acidosis (30).

Several studies have implied a link between endosomal and lysosomal acidity and chemotherapy resistance, hypothesizing that weakly basic drugs can be sequestered to acidic intracellular membrane compartments (31). However, all these studies have focused their view on the role of the vesicular proton pump, establishing strong evidence for the vesicular proton pumps as mediators of chemotherapy resistance due to sequestration and extrusion of chemotherapeutic drugs. Modulation of pH-dependent cellular resistance mechanisms by targeting of the vacuolar H⁺-ATPase may thus be a novel approach to increase the therapeutic efficacy of antitumor agents (7, 8, 32–34).

The results presented here show for the first time that increased intracompartmental acidification as a result of overexpression of an intracellular chloride channel/transporter of the CLC protein family enhances drug resistance.

ClC-3 is expressed in the neuroendocrine gastrointestinal tumor cell line BON as well as in other gastrointestinal neuroendocrine cell lines. Overexpressed ClC-3-GFP localized to the late endocytic compartment in this tumor cell line. This is consistent with previous data showing expression of ClC-3 in intracellular membrane compartments of other cell types (10, 12, 13). Previous studies have also implicated ClC-3 expression in the acidification of intracellular membrane compartments, a function similar to that described for ClC-5 and ClC-7 (14, 35–37). ClC-7 knockout mouse shows a decreased bone resorption and osteopetrosis. This is similar to the effect of SB242784, a potent and selective inhibitor of the osteoclast vesicular proton pump that inhibits bone resorption in vitro (38) and in vivo (39). These findings support further our hypothesis of intracellular chloride channels/transporters as alternative targets in the modulation of intracellular and intracompartmental pH regulation and function.

We have shown that the observed increase of resistance to etoposide following overexpression of ClC-3 in BON cells is probably due to a mechanism of sequestration of the drug in an acidic compartment because exposure to an inhibitor of v-H⁺-ATPase, such as concanamycin A, reduces the resistance. The observed effect, approximate doubling of the IC₅₀ by overexpression of ClC-3 (Figs. 2D and 3B), is comparable with the sensitizing effect of concanamycin A, which was able to reduce IC₅₀ to half (Fig. 4B). A similar effect has been reported for a resistant cell line, derived from proximal tubule cells, exposed to daunomycin (33), supporting biological relevance of these results. The magnitude of the effect obtained by either exposure to concanamycin A or overexpression of ClC-3 is probably limited by the extent the pH of intracellular vesicles can be modified. This is supported by the fact that exposure to 100 nmol/L concanamycin A did not modify the pH of the vesicles any more than exposure to 10 nmol/L (see Supplementary Fig. S3).³

For the studies presented here, a stable overexpression system of ClC-3-GFP was used. As the selection procedure for stably expressing ClC-3 involved selection in G418 for extended times, it was necessary to control for the fact that this selection process itself might have changed resistance to etoposide. This was done by exploiting different levels of ClC-3 expression in a single ClC-3-GFP–expressing clone, comparing populations with low ClC-3-GFP expression (as determined and sorted by the GFP signal) with those with high ClC-3 GFP expression from the same clone. Interclonal heterogeneity could thus be minimized, adding validity to our observations.

The activity of multidrug resistance transporters, such as ATP-binding cassette transporters, is unlikely to play a role in the observed enhancement of etoposide resistance because the parental BON cell line and its derivative overexpressing ClC-3 were found not to express P-glycoprotein, a common culprit for multidrug resistance (1). Furthermore, there is no expression of MRP-1, an ATP-binding cassette multidrug resistance transporter expressed also in the lysosomal compartment and implicated in the establishment of resistance by intralysosomal drug sequestration (2).

Our results may be of relevance to the pharmacokinetics of chemotherapy regimens in neuroendocrine and neuronal tumors because these tissues endogenously express high amounts of ClC-3 protein as shown here and elsewhere (10, 11).

However, the data presented here cannot provide a correlation between level of ClC-3 expression and etoposide resistance, as ClC-3 presence might not be the rate-limiting step of intravesicular acidification, and a direct relationship between intravesicular pH and ClC-3 expression is unlikely because pH cannot be lowered beyond a certain threshold.

In spite of these limitations, our study provides proof of principle of a role for intracellular chloride channels in establishing intracompartmental acidity and subsequent increased resistance to a basic chemodrug, etoposide.

Other proteins of the CLC family could confer a similar effect. Further studies will be necessary to evaluate these findings with regard to other proteins of the CLC family as well as in vivo models of cancer drug resistance and to determine whether intracellular CLC proteins might be a target to modulate drug sensitivity in tumors.