Paclitaxel, docetaxel, doxycycline and 18-α-glycyrrhetinic acid (18-α-GA) were purchased from Sigma-Aldrich (St. Louis, MO, USA). Calcein acetoxymethyl ester (Calcein-AM) and cell culture reagents were purchased from Invitrogen Life Technologies (Carlsbad, CA, USA). Primary and secondary antibodies for western blot analysis were obtained from Sigma-Aldrich. Secondary antibodies for immunofluorescence were purchased from Invitrogen Life Technologies. All other reagents were purchased from Sigma-Aldrich unless otherwise stated.

The HeLa cell line used in this study expressed the Cx32 gene under the control of a bidirectional tetracycline-inducible promoter and was previously described (17). The Cx32 coding sequence was followed by an influenza hemagglutinin (HA) epitope tag at the C-terminus and Cx32 expression was induced by doxycycline (1 μg/ml) for 48 h. Cells were grown in DMEM supplemented with 10% (v/v) fetal bovine serum, 200 μg/ml hygromycin B and 100 μg/ml G418 sulfate.

Paclitaxel/docetaxel and 18-α-GA were dissolved in dimethylsulfoxide (DMSO) and diluted in cell culture medium where required. The final DMSO concentration did not exceed 0.1% (v/v). 18-α-GA (10 μM) was added to the cells 1 h prior to treatment. 18-α-GA was not removed from the cells during paclitaxel/docetaxel treatment.

The toxicity of paclitaxel/docetaxel was evaluated using a standard colony-forming assay (10). Briefly, cells were seeded at high density (~3×10⁴ cells/cm²) to ensure cells were 70–100% confluent at the time of drug treatment. Following exposure to paclitaxel/docetaxel for 6 h at 37ºC, cells were rinsed twice with PBS, harvested by trypsinization, counted, diluted and reseeded into 6-well plates at a density of 500 cells/well. The cells were cultured for 5–8 days and then stained with 4% crystal violet in ethanol. Colonies containing ≥50 cells were counted. For the low cell density condition, cells were seeded at ~100 cells/cm². GJs are not formed under these conditions as the cells were not able to form contacts with each other. Cells were incubated with paclitaxel/docetaxel for 6 h following attachment and processed as described. Plating efficiency was normalized against the number of colonies formed by vehicle-treated cells. The colony forming efficiency of the untreated cells grown under the low- and high-density conditions were not found to be significantly different from each other (data not shown).

A ‘parachute’ dye-coupling assay was used to test the functional GJs, as described previously (17,18). Donor cells were dual-labeled in culture medium supplemented with 5 μM CM-Dil and 5 μM Calcein-AM. CM-Dil is a red membrane dye that does not permeate through GJs to spread to adjacent cells. Calcein-AM is transformed intracellularly into the GJ-permeable dye calcein. The donor cells were washed with PBS, trypsinized and seeded on a monolayer of receiver cells at a ratio of 1:150 (donor/receiver) for 4 h at 37ºC. Receiver cells containing calcein were monitored by fluorescence microscopy and the average number of receiver cells containing calcein/donor cell was scored and normalized to that of the control.

Western blot analysis was performed as described previously (11). In brief, whole-cell lysates (20 μg) were fractionated by 10% SDS-PAGE and transferred onto a nitrocellulose membrane. Cx32 protein was detected by exposure to mouse anti-HA clone HA-7 IgG in Tris-buffered saline and Tween-20 (dilution, 1:1,000), followed by the addition of alkaline phosphatase-conjugated goat anti-mouse IgG (secondary; dilution, 1:2,000). The anti-β-actin primary and secondary detection antibodies were diluted to 1:10,000. All western blot exposures were within the linear range of detection and the intensities of immunopositive bands were quantified using Quantity One software (Bio-Rad, Hercules, CA, USA).

Following treatment with paclitaxel/docetaxel for 1 h, cells were fixed with 4% paraformaldehyde and then blocked with 2% BSA. Mouse anti-HA (dilution, 1:200) was applied for 4 h at room temperature, followed by an FITC-conjugated goat anti-mouse secondary antibody (dilution, 1:400) for 1 h in the dark. Hoechst 33258 was applied for 5 min to stain the nuclei of the cells. Following washing three times with PBS, cells were visualized and images were captured using an Olympus IX71 fluorescence microscope.

A dye uptake assay was used for evaluating the hemichannel activity. Cx32 HeLa cells were plated at low densities such that the majority of the cells were not in physical contact with each other. The cells were exposed to paclitaxel/docetaxel in Ca²⁺-free DMEM for 1 h, followed by incubation with 1% lucifer yellow (LY) and 1% rhodamine dextran (RD) in PBS for 10 min. LY is a GJ-permeable dye and serves as a tracer for hemichannel activity. RD is too large to spread via GJs and was therefore used as a negative control. Following treatment, the cells were rinsed with culture medium containing normal calcium and observed under a fluorescence microscope. Dye uptake was measured using the percentage of fluorescent cells containing LY normalized to that of the solvent control.

Data were analyzed with the Sigma Plot software (Jandel Scientific, San Rafael, CA, USA) using the unpaired Student's t test and presented as the mean ± SEM. P<0.05 was considered to indicate a statistically significant difference.

HeLa cells expressing Cx32 were cultured under low-density conditions, in which GJs did not form as the cells were not in physical contact and high-density conditions, which permitted GJ formation. Following exposure to paclitaxel/docetaxel for 6 h, cell survival was assessed using a standard colony formation assay. At the two densities, paclitaxel and docetaxel reduced the clonogenic survival in a dose-dependent manner (Fig. 2). However, the survival of cells plated at a high density was significantly less than that of cells plated at a low density at concentrations of paclitaxel up to 0.1 μM (Fig. 2A) or at concentrations of docetaxel up to 0.05 μM (Fig. 2B). These results indicate that the cytotoxicity of these agents is greater when cells are plated at high densities. In addition, Fig. 2C and D demonstrate that the reduction ratios of the cell survival between high- and low-density cultures were 18.5±3.5, 17.7±2.7, 16.8±4.9 and 11.11±2.5% at concentrations of paclitaxel between 0.01 and 0.5 μM, respectively and 33.8±9.0, 44.2±7.7, 41.0±6.4 and 48.3±2.0% at concentrations of docetaxel between 0.001 and 0.05 μM, respectively. These results indicate that the enhanced sensitivity of docetaxel attributable to the high-density culture is higher than that of paclitaxel at the tested concentrations. Specifically, cell death at low density was ~20% at 0.05 μM paclitaxel and 0.001 μM docetaxel. However, cell death at high density was 34 and 49% at the same concentrations of paclitaxel and docetaxel, which represent increases by factors of 1.7 and 2.5, respectively. Thus, the toxic effect of docetaxel was higher than that of paclitaxel when cells were grown at high-cell density.

Together, these results demonstrate that the cytotoxicities of paclitaxel and docetaxel are dependent on cell density and indicate that docetaxel induces a greater increase in toxicity to high-density culture than paclitaxel.

The difference between the cytotoxicity of paclitaxel and docetaxel in Cx32 HeLa cells at low- and high-density conditions indicates a possible role for intercellular communication. GJIC is a major pathway for such intercellular communication. To investigate the effect of GJIC on paclitaxel/docetaxel sensitivity, two methods were adopted to modulate Cx expression and GJ function, including the induction of Cx32 expression with doxycycline and the inhibition of GJIC using a chemical inhibitor. The induction of Cx32 expression with doxycycline and its localization on the cell membrane were confirmed by western blot analysis and immunofluorescence (Fig. 3A and B). The emergence of GJIC in Cx32 HeLa cells and the inhibitory effect of 18-α-GA, a known GJ blocker (19), were detected using a ‘parachute’ dye-coupling assay (Fig. 3C).

At high densities, the survival of doxycycline-induced (Cx32 expression) cells was substantially decreased at 0.05 μM paclitaxel or 0.001 μM docetaxel, compared with the survival of uninduced (no Cx32 expression) cells. Specifically, cell survival induced by paclitaxel or docetaxel was significantly reduced by 20.3 and 39.3%, respectively (Fig. 4A). Incubation of Cx32-expressing cells with 10 μM 18-α-GA under high cell-density conditions increased survival from 65.7 to 79.0% in the presence of 0.05 μM paclitaxel (P<0.05) and from 51.0 to 81.2% in the presence of 0.001 μM docetaxel (P<0.05; Fig. 4B). These results indicate that the toxicities of paclitaxel and docetaxel are significantly increased when cells are plated at high density and Cx32 is expressed, or when junctional channels are not blocked. These results indicate that enhanced paclitaxel and docetaxel cytotoxicities at high cell densities are mediated by GJIC.

As described, paclitaxel/docetaxel toxicity is regulated by GJIC at high cell densities. Cell death is likely to markedly reduce the valid cell density for forming GJIC. Therefore, if paclitaxel or docetaxel affected channel function by influencing exclusion from cell death, the toxicity of the agents would be altered. To verify this hypothesis, a ‘parachute’ dye-coupling assay was performed to determine the effects of paclitaxel and docetaxel on Cx32 channels. Following treatment of Cx32 HeLa cells with paclitaxel/docetaxel for 1 h, which did not lead to cell death, paclitaxel markedly reduced dye coupling (Fig. 4C), while docetaxel had no effect on the spread of dye among cells (Fig. 4D). As demonstrated in Fig. 2, at a concentration range over which the cell density affected paclitaxel/docetaxel toxicity, the sensitivity of docetaxel was higher than that of paclitaxel in the high-density culture (with GJIC) compared with the low-density culture (without GJIC). These observations indicate that the effect of paclitaxel and docetaxel on GJIC may affect their own toxicities.

Changes in the number of GJs by affecting Cx32 expression or its membrane localization is one of the mechanisms by which paclitaxel and docetaxel have been hypothesized to alter GJ function. The expression of Cx32 was determined by western blot analysis. Treatment of Cx32 HeLa cells with a range of paclitaxel/docetaxel concentrations for 1 h did not alter Cx32 expression levels compared with that of the vehicle control (Fig. 5A).

Following this, Cx32 localization on the cell membrane was analyzed using immunofluorescence analysis. In control cells (treated with DMSO), Cx32-specific immunoreactivity was predominantly localized to the plasma membrane at cell-cell junctions (Fig. 5B). Cells treated with various concentrations of paclitaxel/docetaxel for 1 h revealed an almost invariant level of Cx32 immunoreactive foci at the cell membrane, even at 10 μM paclitaxel and 1 μM docetaxel (Fig. 5B).

These observations indicate that, following short-term (1 h) treatment, the effects of paclitaxel/docetaxel on junctional function are not a result of affecting Cx32 expression or its membrane localization.

Treatment with paclitaxel/docetaxel for 1 h did not alter the expression and localization of Cx32, indicating that these agents may inhibit dye coupling by affecting channel gating. To test this hypothesis, the permeability of the hemichannel was examined by a dye uptake assay. LY uptake was observed in control Cx32 HeLa cells (Fig. 6A). Dye uptake activity was altered in cells treated with paclitaxel for 1 h. As demonstrated in Fig. 6B, docetaxel did not affect Cx32 hemichannel activity; however, paclitaxel had an inhibitory effect. The results are consistent with the effect of these agents on junctional function (Fig. 4C and D); however, specific values exhibited small deviations. These deviations may be due to differences in the two experimental approaches. Together, the observations demonstrate that paclitaxel-induced closure of ‘gating’ is largely responsible for its GJIC inhibition.