If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
The Centenary Institute, Newtown, NSW, AustraliaDiscipline of Dermatology, University of Sydney, Sydney, NSW, AustraliaDepartment of Dermatology, Royal Prince Alfred Hospital, Camperdown, NSW, Australia
Correspondence: Nikolas K. Haass, The University of Queensland Diamantina Institute, Translational Research Institute, 37 Kent St, Woolloongabba, Queensland 4102, Australia.
The Centenary Institute, Newtown, NSW, AustraliaThe University of Queensland, The University of Queensland Diamantina Institute, Translational Research Institute, Brisbane, Queensland, AustraliaDiscipline of Dermatology, University of Sydney, Sydney, NSW, Australia
The tumor microenvironment is characterized by cancer cell subpopulations with heterogeneous cell cycle profiles. For example, hypoxic tumor zones contain clusters of cancer cells that arrest in G1 phase. It is conceivable that neoplastic cells exhibit differential drug sensitivity based on their residence in specific cell cycle phases. In this study, we used two-dimensional and organotypic melanoma culture models in combination with fluorescent cell cycle indicators to investigate the effects of cell cycle phases on clinically used drugs. We demonstrate that G1-arrested melanoma cells, irrespective of the underlying cause mediating G1 arrest, are resistant to apoptosis induced by the proteasome inhibitor bortezomib or the alkylating agent temozolomide. In contrast, G1-arrested cells were more sensitive to mitogen-activated protein kinase pathway inhibitor-induced cell death. Of clinical relevance, pretreatment of melanoma cells with a mitogen-activated protein kinase pathway inhibitor, which induced G1 arrest, resulted in resistance to temozolomide or bortezomib. On the other hand, pretreatment with temozolomide, which induced G2 arrest, did not result in resistance to mitogen-activated protein kinase pathway inhibitors. In summary, we established a model to study the effects of the cell cycle on drug sensitivity. Cell cycle phase-specific drug resistance is an escape mechanism of melanoma cells that has implications on the choice and timing of drug combination therapies.
Small molecule inhibitors that selectively target mutant BRAF or its downstream effector MEK have provided unprecedented responses in a subset of BRAF mutant melanoma patients (
). Therefore, an improved understanding of the underlying biological mechanisms mediating drug efficacy and the emergence of resistance is required to develop more effective therapeutic strategies for patients with metastatic melanoma.
Mitogen-activated protein kinase (MAPK) pathway inhibitors (MAPKis) effectively induce G1-phase cell cycle arrest (G1 arrest) and apoptosis (
The mitogen-activated protein/extracellular signal-regulated kinase kinase inhibitor AZD6244 (ARRY-142886) induces growth arrest in melanoma cells and tumor regression when combined with docetaxel.
). However, whereas bortezomib induces robust cytotoxicity of proliferating melanoma cells in vitro, the response of melanoma cells to bortezomib is greatly reduced in vivo (
). The DNA-alkylating agent temozolomide, a derivative of dacarbazine, induces DNA damage and G2/M arrest leading to apoptosis in melanoma cell lines in vitro (
Efficacy and side effects of dacarbazine in comparison with temozolomide in the treatment of malignant melanoma: a meta-analysis consisting of 1314 patients.
) and cell cycle status on melanoma response to bortezomib, temozolomide, MAPKi, or combinations of these drugs, we used the fluorescent ubiquitination-based cell cycle indicator (FUCCI) (
) in three melanoma cell lines to track the cell cycle in two-dimensional (2D)-cultured cells or within three-dimensional (3D) collagen-embedded spheroids (
). This model allowed us to study the effect of the cell cycle on drug sensitivity in real time. We found that both pharmacologically and environmentally G1-arrested melanoma cells are resistant to bortezomib- and temozolomide-induced cytotoxicity but are sensitized to MAPK inhibition, a finding that has implications on the choice and timing of drug combination therapies.
Results
Bortezomib induces dose-dependent G2 arrest of melanoma cells
To track the cell cycle status in melanoma cell lines, we used FUCCI in which red fluorescence indicates G1, yellow early S, and green S/G2/M phases, with a short loss of fluorescence just after division (
). Consistently, flow cytometry and image analysis of 4′,6-diamidino-2-phenylindole (DAPI)-stained FUCCI-C8161, -WM164, and -1205Lu cells demonstrated dose-dependent G2-phase accumulation after 24 hours of bortezomib treatment in 2D culture (Figure 1a and b, and Supplementary Figure S1a and b online). G2-arrested cells appeared yellow (rather than green), which is likely due to inhibition of proteasomal degradation of the fluorescent reporters by bortezomib (
). There was a significant dose-dependent inhibition of cell viability/proliferation after 48 hours of bortezomib treatment (Figure 1c and Supplementary Figure S1c). These data confirm that bortezomib causes dose-dependent G2 arrest of melanoma cells.
Figure 1Bortezomib induces dose-dependent G2 arrest of melanoma cells. (a) Flow cytometric cell cycle profile of FUCCI-C8161 cells cultured in two dimensions treated with vehicle (control) or 10 nM bortezomib for 24 hours. DNA content histograms are overlaid with the FUCCI distribution. Graphs are representative of n = 3 independent experiments. (b) Fluorescence microscopic image analysis of % G1, early S, and S/G2/M FUCCI-C8161 cells cultured in two dimensions after 24 hours of bortezomib treatment. Representative of n = 2 independent experiments. (c) 3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium (MTS) proliferation assay of C8161 cells treated with bortezomib for 48 hours at the indicated concentrations. n = 3. Values are given as mean ± SEM. All samples were compared to the control sample to determine statistical differences. ***P < 0.001; ****P < 0.0001. DAPI, 4′,6-diamidino-2-phenylindole; FUCCI, fluorescent ubiquitination-based cell cycle indicator; ns, not significant; SEM, standard error of the mean.
Bortezomib induces G2 and G1 arrest, but preferential apoptosis of G2-phase cells
The increase in G2-phase cells after bortezomib treatment for 24 hours was replicated in 3D spheroids, in which FUCCI-C8161 and -WM164 cells accumulated as a yellow/green population (Figure 2 and Supplementary Figure S2 online). However, the G1-arrested spheroid core, a result of hypoxia and nutrient deprivation (
), persisted (Figure 2a). Consistently, flow cytometric quantification showed an increase in G2-phase cells after 24 hours of bortezomib treatment in both 2D- and 3D-cultured cells; this increase was more substantial in 2D culture (Figure 2c and Supplementary Figure 2d and e).
Figure 2Bortezomib induces G2 and G1-arrest. (a) Confocal extended focus images of collagen-embedded FUCCI-C8161 spheroids treated with 10 nM bortezomib for indicated times. Top slices of the z-stack were removed to reveal the red, G1-arrested spheroid core. Images are representative of n = 3 experiments. Gray line indicates spheroid edge as seen in bright field. Insets shows untreated control. Bar = 200 μm. (b) Flow cytometric cell cycle profiles of live cells of FUCCI-C8161 spheroids treated with vehicle (control) or 10 nM bortezomib for indicated periods. DNA content histogram is overlaid with the FUCCI distribution. Graphs are representative of n = 3 independent experiments. (c) Flow cytometric analysis of live FUCCI-C8161 spheroids (left) or two-dimensional cultured FUCCI cells (right) after 24-, 48-, and 72-hour treatment with 10 nM bortezomib. n = 2–4 experiments. Values are given as mean ± SEM. BTZ, bortezomib; DAPI, 4′,6-diamidino-2-phenylindole; FUCCI, fluorescent ubiquitination-based cell cycle indicator; SEM, standard error of the mean.
Spheroid culture in the presence of 10 nM bortezomib beyond 24 hours resulted in a gradual reduction in the yellow/green population and an emergence of a primarily red population by 72 hours, indicating a loss of S/G2/M phase cells and a relative increase in G0/G1-phase cells (Figure 2 and Supplementary Figure S2). Flow cytometric cell cycle analysis of DAPI-stained cells after 72 hours of treatment with 10 nM bortezomib confirmed that most of the remaining live cells were red and resided in G0/G1 phase (Figure 2b and Supplementary Figure S2c and d). A similar increase in FUCCI-red G1-cell percentage after 48 and 72 hours of bortezomib treatment was observed in 2D culture (Figure 2c and Supplementary Figure S2e). To exclude the possibility that degradation of bortezomib resulted in synchronization of the cell cycle and reentry into G1 phase, cell cycle progression of C8161 spheroids was tracked after replenishment of bortezomib treatment every 24 hours. Similar to previous experiments in which the drug was only added once, there was complete loss of S/G2/M-phase cells by 72 hours (data not shown), indicating the remaining G0/G1-phase cells did not come from S/G2/M-phase cells reentering the cell cycle due to degradation of the drug. After 11 days of bortezomib treatment of C8161 and WM164 spheroids (fresh drug added every 3 days), most cells were dead, and the remaining live cells were FUCCI-red (data not shown). This suggests that bortezomib induces both G2 and G1 arrest in 2D and 3D culture, and that G1-arrested cells survive longer than G2-arrested cells.
Flow cytometry of live/dead-stained C8161, WM164, and 1205Lu melanoma cells confirmed that bortezomib induced time-dependent cytotoxicity in 2D culture (Figure 3a and Supplementary Figure S3a online). Bortezomib also induced time-dependent cytotoxicity in C8161 in 3D spheroids (Figure 3a), albeit less potently compared to 2D culture, consistent with the observation that spheroids were more resistant to bortezomib-induced G2 arrest.
Figure 3Bortezomib induces apoptosis preferentially of G2-phase cells. (a) Flow cytometric quantification of % cell death in C8161 cells cultured in two (2D) or three dimensions (3D) and treated with vehicle (control) or 10 nM bortezomib for the indicated time. n = 2–4. Values are given as mean ± SEM. (b) Percent cells (2D) in G1 versus S/G2/M phase at death assessed by time-lapse image analysis. Error bars = mean ± SEM. n = 2–3. Cell death was determined by cell morphology. (c) G1 versus S/G2/M phase at time of bortezomib-induced death (2D) assessed by time-lapse image analysis. Cell death was ascertained by cell morphology. Error bars = mean ± SEM. Data pooled from n = 2–3 experiments. (d) The 2D-cultured FUCCI-C8161 cells were treated with 10 nM bortezomib for the indicated times. Percentage of annexin V-positive cells in G1 or S/G2/M phase was quantified by flow cytometry. Error bars = mean ± SEM. n = 3. (e) FUCCI-1205Lu cells treated for 24 hours with 10 nM bortezomib were sorted into G1-phase or S/G2/M-phase live cell fractions, or were left unsorted before immunoblotting. Blots are representative of n = 3 experiments. Black line indicates separate blots. *P < 0.05; ****P < 0.0001. SEM, standard error of the mean.
Time-lapse microscopy of C8161 and 1205Lu FUCCI-expressing melanoma cells in 2D culture confirmed that bortezomib treatment induced primarily S/G2/M arrest (FUCCI-yellow/green) within 24 hours (see Supplementary Figure S3b), with few cells completing mitosis after addition of the drug (mean 18% for C8161 and 3.5% for 1205Lu, compared to 100% in the controls). Of the bortezomib-treated cells that stayed alive, 13% (C8161) or 84% (1205Lu) remained in G1 (FUCCI-red) for the entire observation period (40 hours) compared to 0.5% (C8161) and 3% (1205Lu) in the controls, confirming that a subset of cells undergo G0/G1 arrest in response to bortezomib. Single-cell tracking showed that although cells in both G1 and G2 phase died, more green/yellow S/G2/M-phase cells died than red G1 cells (Figure 3b) and that S/G2/M-phase cells died significantly earlier than G0/G1-phase cells (Figure 3c).
Annexin V staining of FUCCI-C8161 and -1205Lu cells after bortezomib treatment indicated the mode of cytotoxicity was primarily apoptosis (Figure 3d and Supplementary Figure S3c). The majority of apoptotic cells were in S/G2/M phase, and the proportion of apoptotic cells increased over time. Apoptotic G1-phase cells were also observed, although the proportion was significantly lower and apoptosis was delayed compared to S/G2/M-phase cells. These results, together with the imaging data, indicate that bortezomib causes G2 and G1 arrest, and that melanoma cells in G1 phase are less sensitive to bortezomib-induced apoptosis.
As the proapoptotic protein NOXA promotes bortezomib-mediated apoptosis in melanoma (
), we chose to investigate whether NOXA was also involved in cell cycle-specific bortezomib-mediated apoptosis. FUCCI-1205Lu cells were treated for 24 hours with 10 nM bortezomib and then sorted by cell cycle phase (MoFlo Astrios Cell Sorter, Beckman Coulter; see Supplementary Materials online). Immunoblotting of sorted and unsorted samples revealed that NOXA levels increased with bortezomib treatment; this increase was higher in S/G2/M- compared to G0/1-phase cells (Figure 3e), consistent with the higher level of apoptosis in G2 phase. NOXA expression was also lower in untreated G0/1- compared to S/G2/M-phase cells. Consistent with the NOXA levels and annexin V staining, cleaved caspase 3 was detected in bortezomib-treated S/G2/M- but not G1-phase cells (Figure 3e).
), we sought to determine whether bortezomib-induced G1 arrest is also reversible. To test whether bortezomib-treated G1-arrested cells can reenter the cell cycle after drug removal, we treated 2D-cultured FUCCI-C8161 and -1205Lu cells for 3 days, then changed to normal medium. Time-lapse imaging indicated that G1-arrested cells reentered the cell cycle after bortezomib removal (see Supplementary Figure S3d–g and Supplementary Movies 1 and 2 online). Reexposure to bortezomib resulted in an increase in S/G2/M-phase cells and subsequent cell death, indicating retained drug sensitivity (see Supplementary Figure S3d–g). Whereas drug washout after 3 days led to recovery of the treated cells, continuous exposure to bortezomib (fresh drug replacement after 3 days) resulted in continuing G1 arrest after day 3 and eventually cell death of most cells at day 6 (see Supplementary Figure S3h).
Pharmacologically induced G1 arrest inhibits bortezomib- and temozolomide-induced cytotoxicity
Resistance of G1-arrested cells in the spheroid core to bortezomib-induced G2-arrest, as well as the longer survival of G1-arrested bortezomib-treated cells, indicates that G1-arrested cells are less sensitive to bortezomib-induced apoptosis. To investigate this theory further, melanoma cells were pretreated with G1 arrest-inducing drugs before exposure to bortezomib (Figure 4a). We demonstrated previously that MEK or BRAF inhibition induces G1 arrest (
The mitogen-activated protein/extracellular signal-regulated kinase kinase inhibitor AZD6244 (ARRY-142886) induces growth arrest in melanoma cells and tumor regression when combined with docetaxel.
). Therefore, C8161, WM164, and 1205Lu cells in either 2D or 3D culture were treated for 24 hours with the MEK inhibitor U0126 or the selective BRAF inhibitor PLX4032 at a concentration that was optimized for each cell line to induce G1 arrest without significant cell death (see Supplementary Figure S4a and b online). Cells were then treated with bortezomib for another 48 hours in the continued presence of U0126 or PLX4032. Imaging of FUCCI-C8161, -WM164, and -1205Lu cells grown in 2D confirmed that U0126 prompted G1 arrest in all three cells lines after 24 hours, and PLX4032 in WM164 and 1205Lu cells (see Supplementary Figure S4c). C8161 is BRAFWT at codon 600 (
Figure 4Pharmacologically induced G1 arrest inhibits bortezomib and temozolomide-induced cytotoxicity. (a) Time line demonstrating the drug treatment schedule. Flow cytometric quantification of % cell death in (b) WM164, (c) 1205Lu, and (d) C8161 cells cultured in two dimensions (2D) and treated with vehicle (DMSO), 10 nM BTZ, 10 μM U0126 (1μM for WM164), 1 μM PLX4032 (0.1 μM for WM164), or the indicated combinations. BTZ was added after 24 hours, for a total of 48 hours, whereas the other treatments were applied for the total 72 hours. Error bars = mean ± SEM. n = 3–5. Samples were compared to DMSO+BTZ to determine statistical differences. (e) Confocal extended focus images of FUCCI-C8161 spheroids treated with vehicle (DMSO), 15 nM BTZ alone, or combined with 10 μM U0126 or 1 μM PLX4032. BTZ was only added after 24 hours, for a total of 24 hours, whereas the other treatments were applied for the total 48 hours. (f) Flow cytometric quantification of % cell death in melanoma cells cultured in 2D. Cell line: WM164; treatment: vehicle (DMSO), 10 nM BTZ, 0.1 μM PLX4032, or PLX + BTZ for 48 hours. Drugs were added simultaneously. Samples were compared to DMSO+BTZ to determine statistical differences. Error bars = mean ± SEM. n = 3. (g) As in f with the following variables: cell line: WM164; treatment: 10 μM U0126, 0.1 μM PLX4032, or combination. U0126 was added after 24 hours. PLX4032 was present for the entire 72 hours. (h) Image analysis of the % G1, early S, and S/G2/M FUCCI-C8161 cells cultured in 2D after 24-hour treatment with 10 μM TMZ. n = 3. Values are given as mean ± SEM. (i) As in f with the following variables: cell line: C8161; treatment: 10 μM U0126, 40 μM TMZ, or combination. TMZ was added after 24 hours. U0126 was present for the entire 72 hours. (j) As in f with the following variables: cell line: C8161; treatment: 10 μM TMZ, 60–80 μM U0126, or combination. U0126 was added after 24 hours. TMZ was present for the entire 72 hours. (k) As in f with the following variables: cell line: C8161; treatment: 10 μM TMZ, 10 nM BTZ, or combination. BTZ was added after 24 hours. TMZ was present for the entire 72 hours. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. BTZ, bortezomib; ns, not significant; SEM, standard error of the mean; TMZ, temozolomide.
Flow cytometry of 2D-cultured cells indicated that U0126- or PLX4032-induced G1 arrest inhibited bortezomib-induced apoptosis (Figure 4b–d). Furthermore, flow cytometry of 2D-cultured FUCCI cells and imaging of FUCCI spheroids demonstrated that U0126 or PLX4032 prevented bortezomib-induced S/G2/M arrest (Figure 4e and Supplementary Figure S4d). As expected, this was not the case for PLX4032/bortezomib combination in C8161 cells (Figure 4d and e, and Supplementary Figure S4d). Interestingly, although PLX4032 alone was not cytotoxic for BRAFWT cells (C8161, MelRM), PLX4032/bortezomib combination increased cell death and G2 arrest in BRAFWT cells, compared to bortezomib alone (Figure 4d and Supplementary Figure S4d and e). This may be due to off-target effects of PLX4032 or the paradoxical activation of MAPK in BRAFWT cells (
). Typically, pERK was slightly increased after 48 hours of PLX4032 treatment in C8161 cells (either alone or in combination with bortezomib). However bortezomib alone did not significantly alter pERK levels (data not shown).
To confirm that G1 arrest-inducing pretreatment was necessary for the rescue effect in BRAF-mutant WM164 cells, we added PLX4032 simultaneously with bortezomib. Indeed there was no rescue from bortezomib-induced apoptosis (Figure 4f). However, because both drugs alone induced some cell death and the combination had no additive effect, some protection from bortezomib-mediated apoptosis may have been possible. Some cells may undergo PLX4032-induced G1 arrest before bortezomib becomes effective, which is supported by the reduction of bortezomib-induced G2 accumulation when simultaneously combined with PLX4032 (see Supplementary Figure S4f).
To determine whether pharmacologically induced G1 arrest can rescue MAPKi cytotoxicity, we pretreated WM164 cells with low-dose PLX4032 to cause G1 arrest but minimal cytotoxicity, then treated with U0126 for another 48 hours (Figure 4g). Although the amount of cell death was similar between single-agent PLX4032 and U0126 treatments, there was a significant increase in cell death for the combination. This suggests that rather than rescuing cells from U0126-induced cell death, PLX4032-induced G1 arrest sensitizes cells to U0126-induced cell death.
To explore whether pharmacologically induced G1 arrest protected from G2 arrest-inducing drugs other than bortezomib, we tested the effect of U0126 pretreatment on temozolomide-induced cytotoxicity. Temozolomide caused G2 arrest of C8161 after 24 hours of treatment (Figure 4h). U0126 pretreatment protected cells from temozolomide-induced G2 arrest and cell death (Figure 4i and Supplementary Figure S4g). WM164 and 1205Lu were resistant to temozolomide. At the high concentrations required to induce cell death (>100 μM), very little G2 arrest was observed. Melanoma cell resistance to temozolomide has been described previously (
Finally, to determine the effect of G2 arrest on MAPKi cytotoxicity, we pretreated C8161 cells with low-dose temozolomide, then added high-dose U0126 for another 48 hours to induce cell death. In this case, the cells were not protected from U0126-induced cell death (Figure 4j). U0126 appeared to be able to overcome the temozolomide-induced G2 arrest and drive cells into G1 arrest (see Supplementary Figure S4h). Cells pretreated with low-dose temozolomide were also not protected from bortezomib-induced cell death (Figure 4k). In fact, the combination enhanced cell death.
Environmentally induced G1 arrest inhibits bortezomib-induced cell death but enhances MAPK inhibitor cytotoxicity
As an environmental approach of inducing G1 arrest, we serum-starved WM164 cells under hypoxia for 24 hours (Figure 5a). This environmentally induced G1 arrest reduced bortezomib-induced cell death (Figure 5c), indicating that multiple methods of arresting cells in G1 result in protection from bortezomib cytotoxicity. In contrast, environmentally induced G1 arrest did not protect cells from but rather increased U0126-induced cell death (Figure 5e). This indicates that G1 arrest is not protective for MAPKi cytotoxicity.
Figure 5Environmentally induced G1 arrest inhibits bortezomib-induced cell death but enhances MAPK inhibitor cytotoxicity. (a) Image analysis of the % G1, early S, and S/G2/M phase WM164 cells cultured in two dimensions (2D) in normoxia with normal medium or after 24 hours of hypoxia and serum-starvation. Values are given as mean ± SEM. n = 4. (b) Image analysis of the % red G1, early S, and S/G2/M phase C8161 cells cultured in 2D in control or in confluent/starved wells after 48 hours. Values are given as mean ± SEM. n = 3. (c) FUCCI-WM164 cells cultured in 2D were pretreated with either hypoxia and serum-starvation for 24 hours, or control normoxia and full serum conditions. Flow quantification of the % dead cells was performed after 48-hour treatment with 15 nM bortezomib or vehicle control. Values are given as mean ± SEM. n = 4. (d) FUCCI-C8161 cells cultured in 2D were pretreated with either subconfluency with normal medium or confluency with serum-starvation for 48 hours. Flow quantification of the % dead cells was performed after an additional treatment with 10 nM bortezomib (BTZ), 40 μM TMZ, or vehicle control for 48 hours. Values are given as mean ± SEM. n = 3–4. (e) As in c but treatment with 20 μM U0126. Values are given as mean ± SEM. n = 3. (f) As in d but treatment with 60–80 μM U0126. Values are given as mean ± SEM. n = 3. (g) WM164 cells were cultured in 2D in normoxia with normal medium, or after hypoxia and serum-starvation for 24 or 48 hours. Western blotting was then performed. Blots are representative of n = 3 independent experiments. *P < 0.05; **P < 0.01; ***P < 0.001. ns, not significant; SEM, standard error of the mean; TMZ, temozolomide.
C8161 cells, confluent and serum-starved for 48 hours, arrested in G1 (Figure 5b). This approach of environmentally induced G1 arrest also resulted in protection from bortezomib- and temozolomide-induced cell death (Figure 5d). Again, environmentally induced G1 arrest did not protect C8161 cells from MAPKi cytotoxicity (using high-dose U0126) but instead enhanced it (Figure 5f).
To investigate the molecular mechanisms underlying the changes in drug sensitivity of hypoxic/starved G1-arrested cells, we performed immunoblotting on serum-starved WM164 cells under hypoxia for 24 hours or 48 hours (Figure 5g). NOXA levels were decreased under hypoxia, which may contribute to the resistance to bortezomib-induced apoptosis. No changes in pERK levels relative to total ERK were seen under hypoxia (data not shown), indicating that altered MAPK signaling under hypoxia is not responsible for the enhanced sensitivity to MAPKi.
Discussion
We have established a model to study the effect of the cell cycle on drug sensitivity in real time. Validating this approach, we confirmed that bortezomib induces time- and dose-dependent G2 arrest resulting in death of melanoma cells, as previously shown (
). In this study we demonstrated that bortezomib induces not only G2 but also G1 arrest and, importantly, causes apoptosis preferentially of G2-phase cells, likely via NOXA. Surviving G1-arrested cells reentered the cell cycle upon bortezomib removal and regained sensitivity to bortezomib-induced G2 arrest and apoptosis. Both pharmacologically and environmentally G1-arrested melanoma cells are resistant to bortezomib and temozolomide but are sensitized to MAPKi, indicating cell cycle phase-specific drug sensitivity.
Cell cycle-mediated resistance has previously been demonstrated for the taxanes, which stabilize microtubules and induce G2/M arrest followed by apoptosis (
). Together with our data, these findings indicate that cell cycle phase-specific drug resistance is a general escape mechanism that occurs in various cancer types and a range of chemotherapies.
We recently showed that in drug combinations, one drug can sensitize to the other but not necessarily the converse (
). In this study we demonstrated that reversing the order of the drug combination may impact on treatment efficacy. For example, whereas pretreatment of melanoma cells with MAPKi resulted in resistance to temozolomide or bortezomib, pretreatment with temozolomide did not result in resistance to MAPKi.
Consistent with the synergistic activity of bortezomib with temozolomide against melanoma in mice (
Augmenting chemosensitivity of malignant melanoma tumors via proteasome inhibition: implication for bortezomib (VELCADE, PS-341) as a therapeutic agent for malignant melanoma.
), we showed that pretreatment of melanoma cells with temozolomide, which induces G2 arrest, results in increased bortezomib-induced cytotoxicity, suggesting that sequential combination of bortezomib with other G2 phase-arresting drugs may be an effective therapeutic strategy for patients with metastatic melanoma. Supporting this idea, synchronizing myeloma cells in early S phase using reversible CDK4/CDK6 inhibition resulted in enhanced bortezomib-induced cytotoxicity compared to synchronization of cells in G1 (
). In contrast, a phase I trial with combined bortezomib and temozolomide treatment resulted in only one of 19 advanced melanoma patients achieving a partial response (
A phase I trial of bortezomib with temozolomide in patients with advanced melanoma: toxicities, antitumor effects, and modulation of therapeutic targets.
). However, drugs in this trial were administered simultaneously. Therefore, sequential rather than simultaneous treatment with temozolomide or other G2-phase targeting drugs before administration of bortezomib may be more effective.
Importantly, we demonstrated that although bortezomib and temozolomide effectively induce death of proliferating melanoma cells, they are ineffective against either drug-induced or hypoxia/serum-starved/confluency-induced G1-arrested cells, as well as the G1-arrested core of 3D spheroids. This may explain why the combination of bortezomib and the pan-RAF inhibitor sorafenib in a recent clinical trial was ineffective for the treatment of melanoma (
), as sorafenib may have quickly induced G1 arrest and hence neutralized the effect of bortezomib. Furthermore, drugs that specifically target cells in G2 phase may be universally less effective than G1-phase targeting drugs in vivo due to the initiation of G0/G1 arrest as part of a general stress response to drug treatment or hypoxic, nutrient-poor conditions within a tumor (
). However, the level of hypoxia may be important. We used moderate hypoxia (1%), whereas a previous study using severe hypoxia (<0.2%) showed that hypoxic HeLa cells were more sensitive to bortezomib due to induction of endoplasmic reticulum stress pathways (
), so hypoxia-induced G1 arrest may only increase sensitivity to MAPKi in the short term.
A highly proliferative tumor may also respond better to G2 phase-targeting drugs than a slow growing tumor. C8161, which was most sensitive to bortezomib and temozolomide, is the fastest growing cell line and spends the least amount of time in G1 phase (
The molecular mechanisms underlying cell cycle phase-specific resistance to bortezomib and temozolomide are not fully elucidated. We demonstrated that NOXA, known to promote bortezomib-induced apoptosis (
), is lower in both cycling G1-phase cells and hypoxia-induced G1-arrested cells. Although NOXA is up-regulated during bortezomib treatment as expected, we now demonstrate that NOXA is primarily up-regulated in S/G2/M phase but remains low in G0/G1 phase of bortezomib-treated melanoma cells. Thus, it is possible that decreased levels of NOXA have a protective effect on G1-phase cells.
In contrast to the protective effect of G1 arrest on G2-phase targeting drugs, we showed that G1 arrest increased the sensitivity of melanoma cells to MAPKi, which induce G1 arrest and apoptosis (
), may be effective in combination with G1- but not with G2-phase targeting therapies. These data also further support the rationale for combining BRAF and MEK inhibitors for the treatment of melanoma, which have been shown in numerous clinical trials to improve progression-free survival compared to single-agent treatment (
Dabrafenib and trametinib versus dabrafenib and placebo for Val600 BRAF-mutant melanoma: a multicentre, double-blind, phase 3 randomised controlled trial.
). However, the emergence of resistance in response to continuous treatment with these drugs indicates that G0/G1 arrest may confer a drug-tolerant phenotype that primes the cell for development of permanent resistance mechanisms and reactivation of proliferative signaling. Exposure to a sublethal dose of PLX4032 for 12 days (with cells remaining in G1 arrest) can lead to multidrug tolerance, in which cells become resistant to subsequent treatment with the MEK inhibitor GSK1120212 or other drugs (
). The progression of melanoma from drug tolerance to resistance was observed in tumors from patients treated with the BRAF inhibitor dabrafenib, which subsequently became insensitive to treatment with the MEK inhibitor trametinib (
Combined BRAF (Dabrafenib) and MEK inhibition (Trametinib) in patients with BRAFV600-mutant melanoma experiencing progression with single-agent BRAF inhibitor.
). Therefore, to prevent the progression of melanoma from drug tolerance to resistance, it may be necessary to allow a treatment-free period in the dosing protocol to reduce the general stress response within melanoma cells responsible for the development of resistance.
This study demonstrates that consideration of cell cycle-mediated resistance to bortezomib, temozolomide, and MAPKi must be taken into account when planning melanoma combination therapies, timing of dosing schedules, and choice of drug therapies in solid tumors. These results may extend to other drug therapies that cause cell death via cell cycle arrest and should be investigated in future studies.
Materials and Methods
Cells and cell culture
The human melanoma cell lines C8161, WM164, and 1205Lu were genotypically characterized (
) (with 4% fetal bovine serum instead of 2% fetal bovine serum) and authenticated by short tandem repeat fingerprinting (QIMR Berghofer Medical Research Institute, Herston, Australia). MelRM were grown in DMEM supplemented with 10% fetal bovine serum as previously described (
). This model mimics in vivo tumor architecture and microenvironment and is used for investigating the growth, invasion, and viability of melanoma cells (
). The lentivirus was produced by cotransfection of human embryonic kidney 293T cells; high-titer viral solutions for mKO2-hCdt1 (30/120) and mAG-hGem (1/110) were prepared and used for cotransduction into three biologically and genetically well-characterized melanoma cell lines (see section “Cells and cell culture”); and subclones were generated by single-cell sorting (
2D and 3D cultures were treated with the following drugs at indicated doses and compared to DMSO-treated controls: the proteasome inhibitor bortezomib (Janssen Cilag, North Ryde, NSW, Australia), MEK1/2 inhibitor U0126 (Sigma-Aldrich, St. Louis, MO), selective BRAF inhibitor PLX4032 (Active Biochem, Maplewood, NJ). For pretreatment, U0126 and PLX4032 doses were chosen so as to induce G1 arrest without significant cell death (
MTS (3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium) viability assays were conducted as described in the Supplementary Materials (
FUCCI-WM164 cells were plated to 30% confluence in 10-cm dishes. The unstarved/normoxia control was incubated in complete medium at 21% O2 and the serum-starved/hypoxia group in the same medium without fetal bovine serum and bovine insulin at 1% O2 (CB210, Binder, Germany). Twenty-four hours later, cells were observed for G1 arrest (FUCCI) before addition of drugs or vehicle. After 48 hours of treatment, live/dead cell analysis was performed.
Confluency and serum-starvation assay
FUCCI-C8161 cells were plated at a density of 2.5 × 104 (low confluency) or 3 × 105 cells per well (high confluency) in six-well plates in normal medium. Cells were allowed to adhere, and the high-confluency wells were changed to serum-free medium. Cells were allowed to grow for 48 hours and were observed for G1 arrest (FUCCI). Drugs were added for another 48 hours before live/dead cell analysis.
Flow cytometric cell cycle, live/dead, and annexin V analyses
Flow cytometric cell cycle and live/dead analyses were conducted as described previously (
Imaging of live FUCCI cells before extraction for flow analysis
Live FUCCI cells cultured in six-well plates were imaged on a Nikon-300 (Nikon, Tokyo, Japan) inverted fluorescence microscope using 10× or 20× objectives or a Delta Vision Elite microscope using a 20× objective (GE Healthcare Life Sciences, Cleveland, OH).
Imaging of live FUCCI cells for dose-response assays
Cells were cultured in 96-well imaging plates (BD, Franklin Lakes, NJ) and imaged on a Pathway 855 high-content bioimager (BD) using a 10× objective.
Cell cycle image analysis
FUCCI-red, -green, and -yellow cells in merged spheroid z-stacks or single-plane 2D culture images were quantified using automated image analysis (Volocity software, Perkin Elmer, Waltham, MA) as previously described (
). Alternatively, images obtained on the BD Pathway 855 were analyzed using BD Attovision software, and red and green cell intensities were exported to flow cytometry standard files for analysis using FlowJo (TreeStar, Ashland, OR).
Live time-lapse imaging and cell tracking
Live imaging of 2D cultured cells was performed on the Delta Vision Elite microscope with 30-minute intervals for 48–72 hours. Cells were maintained at 37°C, 5% CO2. Movies were started approximately 30 minutes after drug addition or medium change. Cell tracking was performed using manual tracking in Volocity software. Individual cells were detected based on emitted fluorescence, and cell death was ascertained by cell morphology (blebbing and loss of adherence).
Statistical analysis
For group comparisons (normal distribution) one-way analysis of variance followed by the Tukey or Dunnett test was used. For comparisons of two samples, the Student t test (normally distributed) was used. A difference was considered significant if P < 0.05 (*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001). Experiments were usually repeated at least three times independently.
Conflict of Interest
The authors state no conflict of interest.
Acknowledgments
We thank Atsushi Miyawaki, RIKEN, Wako-city, Japan, for providing the FUCCI constructs; Meenhard Herlyn and Patricia Brafford, The Wistar Institute, Philadelphia, and Peter Hersey, Centenary Institute, for providing cell lines; Andrea Anfosso, Amanda Chen, and the Imaging and Flow Cytometry Facility at the Centenary Institute for outstanding technical support; and Crystal Tonnessen, Loredana Spoerri, Frank Vari, and Maher Gandhi, University of Queensland, for intellectual input. NKH is a Cameron fellow of the Melanoma and Skin Cancer Research Institute, Australia. KAB is a fellow of Cancer Institute New South Wales (13/ECF/1-39). DSH is a David Sainsbury Fellow of the National Centre for the Replacement, Refinement and Reduction of Animals in Research. This work was supported by project grants RG 09-08 and RG 13-06 (Cancer Council New South Wales); 570778 and 1051996 (Priority-driven collaborative cancer research scheme/Cancer Australia/Cure Cancer Australia Foundation); 08/RFG/1-27 (Cancer Institute New South Wales); and APP1003637 and APP1084893 (National Health and Medical Research Council).
Augmenting chemosensitivity of malignant melanoma tumors via proteasome inhibition: implication for bortezomib (VELCADE, PS-341) as a therapeutic agent for malignant melanoma.
The mitogen-activated protein/extracellular signal-regulated kinase kinase inhibitor AZD6244 (ARRY-142886) induces growth arrest in melanoma cells and tumor regression when combined with docetaxel.
Combined BRAF (Dabrafenib) and MEK inhibition (Trametinib) in patients with BRAFV600-mutant melanoma experiencing progression with single-agent BRAF inhibitor.
Dabrafenib and trametinib versus dabrafenib and placebo for Val600 BRAF-mutant melanoma: a multicentre, double-blind, phase 3 randomised controlled trial.
A phase I trial of bortezomib with temozolomide in patients with advanced melanoma: toxicities, antitumor effects, and modulation of therapeutic targets.
Efficacy and side effects of dacarbazine in comparison with temozolomide in the treatment of malignant melanoma: a meta-analysis consisting of 1314 patients.
30862-4&id=gr1.jpg){kind=link}
30862-4&id=gr2.jpg){kind=link}
30862-4&id=gr3.jpg){kind=link}
30862-4&id=gr4.jpg){kind=link}
30862-4&id=gr5.jpg){kind=link}