CDK4 inhibition and doxorubicin mediate breast cancer cell apoptosis through Smad3 and survivin

Abstract

Cyclin D1/CDK4 activity is upregulated in up to 50% of breast cancers and CDK4-mediated phosphorylation negatively regulates the TGFβ superfamily member Smad3. We sought to determine if CDK4 inhibition and doxorubicin chemotherapy could impact Smad3-mediated cell/colony growth and apoptosis in breast cancer cells. Parental and cyclin D1-overexpressing MCF7 cells were treated with CDK4 inhibitor, doxorubicin, or combination therapy and cell proliferation, apoptosis, colony formation, and expression of apoptotic proteins were evaluated using an MTS assay, TUNEL staining, 3D Matrigel assay, and apoptosis array/immunoblotting. Study cells were also transduced with WT Smad3 or a Smad3 construct resistant to CDK4 phosphorylation (5M) and colony formation and expression of apoptotic proteins were assessed. Treatment with CDK4 inhibitor/doxorubicin combination therapy, or transduction with 5M Smad3, resulted in a similar decrease in colony formation. Treating cyclin D overexpressing breast cancer cells with combination therapy also resulted in the greatest increase in apoptosis, resulted in decreased expression of anti-apoptotic proteins survivin and XIAP, and impacted subcellular localization of pro-apoptotic Smac/DIABLO. Additionally, transduction of 5M Smad3 and doxorubicin treatment resulted in the greatest change in apoptotic protein expression. Collectively, this work showed the impact of CDK4 inhibitor-mediated, Smad3-regulated tumor suppression, which was augmented in doxorubicin-treated cyclin D-overexpressing study cells.

Keywords: :

• CDK4
• Smad3
• TGFβ
• apoptosis
• breast cancer
• doxorubicin
• survivin

Introduction

Breast cancer is the most common malignancy affecting women, with over 200 000 cases diagnosed and 40 000 patients dying each year from this disease.¹ The advent of targeted therapeutics has significantly impacted breast cancer outcomes for patients with erbB2 overexpressing cancers. Based on this bench to bedside success, the discovery of additional cancer cell targets is actively being pursued with a specific focus on cell cycle components, including mitogenic cyclins. Cyclin D1 is overexpressed at the mRNA and protein levels in up to 50% of breast cancers.^2-4 Cyclin D1 is primarily overexpressed in estrogen receptor positive (ER⁺) tumors, and this overexpression is associated with poor outcomes and decreased relapse-free survival.^5,6 As such, it is one of the most commonly overexpressed oncogenes in breast cancer and is a potentially significant therapeutic target.

Cyclins are the regulatory subunits of cyclin-dependent kinases (CDKs). Cyclin/CDK complexes permit cells to transition from the G₁ to the S phase of the cell cycle. The activities of these complexes are modulated by the binding of CDK inhibitors (CDKis), including p15, p16, p21, and p27, which can sequester CDKs or bind and inhibit cyclin/CDK complexes. Cyclin D forms active complexes with either CDK4 or CDK6, which initiate the phosphorylation of the tumor suppressive retinoblastoma (Rb) family of proteins.⁷ Hyperphosphorylation of Rb by cyclin D/CDK4 or 6 inhibits Rb from sequestering members of the E2F transcription factor family, which then drives the transcription of genes encoding the proteins required for G₁/S-phase transition and S-phase progression.⁷ Thus, cyclin D overexpression contributes to loss of cell cycle control, facilitating oncogenic progression.⁸ Furthermore, murine studies have shown that the continued presence of active CDK4 complexes plays a key role in mammary tumor growth.^9,10

Cyclin D/CDK4 complexes are also involved in cell cycle control through the phosphorylation and regulation of members of the transforming growth factor-β (TGFβ) superfamily.^11,12 Several members of the TGFβ superfamily have crucial roles in mammary gland physiology, with the Smads functioning as downstream mediators of this signaling pathway.¹³ Intact canonical TGFβ/Smad3 signaling has previously been linked to tumor suppressive cytostatic and pro-apoptotic events in early stage breast cancer.^14,15 Simultaneously, TGFβ/Smad3 signaling has been shown to promote oncogenic progression through the induction of epithelial-to-mesenchymal transition (EMT) in advanced stage breast carcinoma. Based on these opposing actions in early and later stage disease, TGFβ/Smad3 signaling can have dichotomous actions in breast oncogenesis.¹² Canonical TGFβ signaling occurs through the phosphorylation of Smad3 at the C-terminus by the TGFBRI receptor. However, CDKs 4/2, in addition to other kinases, can also noncanonically phosphorylate Smad3 at multiple sites located primarily in the linker region of the protein.¹⁶ This noncanonical phosphorylation of Smad3 can result in decreased tumor suppression of the Smad3 protein associated with increased c-myc activity and inhibition of CDKis.^17,18 Conversely, transfection of the Smad3 protein mutated at five CDK phosphorylation sites (5M Smad3), was shown to restore Smad3 activity and resulted in lower c-myc mRNA levels and higher levels of the CDKi p15.^17,18 Treatment with a CDK4i also resulted in increased Smad3 activity in cyclin D overexpressing breast cancer cells. Collectively, this data suggests that CDK4 inhibition could be a targeted treatment strategy for patients whose tumors overexpress cyclin D by promoting Smad3-regulated cell cycle arrest.

Pan-CDK inhibitors have been utilized in phase I solid tumor clinical trials, yet efficacy has thus far been modest, potentially associated with both the lack of tumor cyclin profiling and nonspecific CDK inhibition implemented in these trials.¹⁹ A more thorough understanding of the role of specific cyclins and their CDK complements, in addition to the directed study of CDK inhibitors in specific cyclin-overexpressing cancers, is necessary to reveal the therapeutic potential of these agents. To date, limited study of combination CDK inhibitor/chemotherapy has demonstrated a partial response for patients in clinical trials of advanced solid tumors.¹⁹ Overall, a more rational and focused implementation of CDK inhibitor therapy, combined with cytotoxic chemotherapy, could result in improved cytotoxic efficacy and improved patient outcomes.

In previous in vitro studies, CDK inhibitors and chemotherapeutics have independently induced cytostatic and apoptotic effects on cancer cells. Doxorubicin, one of the most widely implemented antitumor drugs used for breast cancer treatment, exerts cytostatic changes via increased expression of p21 in a p53-dependent manner, and apoptotic events through DNA intercalation, inhibition of DNA topoisomerase II, and induction of DNA double strand breaks.^20,21 CDK inhibitors exert cytostatic effects through restoration of cell cycle checkpoints.¹⁹ Though less well understood, it has been suggested that CDK inhibitor-associated apoptosis is related to a decrease in CDK-mediated phosphorylation of Rb and persistent E2F activation, which may induce apoptosis in cancer cells.²² Elucidating the mechanisms by which CDK inhibition facilitates apoptosis would allow for more appropriate synergism with current chemotherapeutic treatments.

The apoptotic process is both positively and negatively regulated by pro- and anti- apoptotic factors, respectively. Survivin is a member of the inhibitor of apoptosis proteins (IAPs), and plays a critical role in both inhibition of apoptosis and control of cell division.^23,24 The role of survivin in several different malignancies has been examined, including bladder, blood, liver, colon, pancreas, prostate, and breast.^25-31 Specifically, in breast cancer, survivin expression has been linked to early recurrence and decreased survival rates.^32,33 Survivin expression has also been associated with the promotion of cancer cell survival through the inhibition of pro-apoptotic stimuli resulting in chemoresistance.^34,35 In a prostate cancer model, survivin downregulation was shown to occur in a TGFβ/Smad2/3-dependent manner resulting in increased apoptosis.³⁶ This mechanism involved the binding and interaction of the hypophosphorylated Rb-E2F4 complex and C-terminal phosphorylated Smad3 on the survivin promoter, inducing transcriptional repression. XIAP is a potent mammalian caspase inhibitor that binds and inhibits proteolytic activity of caspases 3, 7, and 9 in normal cells. High XIAP levels have been associated with aggressive, high grade breast cancers.³⁷ An antagonist of XIAP, second mitochondria-derived activator of caspase (Smac/DIABLO), is released from the mitochondria upon intrinsic apoptotic stimuli, and binds and inhibits XIAP, which results in the release of caspase3. Caspase 3 then cleaves PARP1, preventing DNA repair and resulting in apoptosis. To date, the impact of noncanonical phosphorylation of Smad3 has not been studied in the context of survivin, XIAP, and Smac/DIABLO expression.

We have previously shown that pharmacological CDK inhibition can augment Smad3-regulated gene expression. In the current study, we implemented MCF7 cells, a prototypical ER+ breast cancer cell line. We also engineered a cyclin D overexpressing variant of this cell line to reflect the high number of ER⁺ breast cancers that overexpress this cyclin. We then sought to determine if CDK inhibition and chemotherapy could impact Smad3-mediated cell/colony growth and apoptosis, potentially through the regulation of survivin expression. Evidence of this relationship would provide novel insight into the mechanism behind the growth arrest and apoptosis induction mediated by both CDK inhibition and chemotherapy through a convergence on Smad3 signaling.

Results

CDK4i and doxorubicin treatment decreases cell proliferation

We have previously shown that transfecting breast cancer cells with a CDK4 phosphorylation site mutant Smad3 construct or treatment with CDK4i-affected cell cycle regulation.¹⁷ To further examine the impact of CDK4 inhibition and chemotherapy on cyclin D overexpressing cell proliferation, we engineered a model of cyclin D overexpression in ER⁺ breast cancer cells. To this end, MCF7 cells were first stably transfected with either control or cyclin D1 plasmids. An MTS assay was then used to assess the effect of treatment with CDK4i and doxorubicin on proliferation of control MCF7/V (Fig. 1A) and cyclin D1 overexpressing (CD1) MCF7 cells (Fig. 1B). Although the two study cells lines were initially plated at equal densities, the MCF7/CD1 cells showed a higher absorbance at Day 0, indicating a greater initial degree of cellular metabolic activity, as expected, for the cyclin D-overexpressing study cells. When compared with control treatment, both MCF7/V and MCF7/CD1 breast cancer cells treated with CDK4i alone resulted in a significant decrease in cell proliferation over a period of 9 d (P < 0.05). Additionally, treatment of both cell lines with doxorubicin resulted in a greater decrease (P < 0.01), with the most significant decrease in cell proliferation occurring after combination treatment (P < 0.001). We also found that, after normalization to the non-treated cells, for the study cells treated with combination therapy, the proliferation rate was effectively reduced by 23% for the MCF7/CD1 cells, when compared with the proliferation rate for the MCF7/V cells. We observed a similar decrease in proliferation after treatment with CDK4i and doxorubicin using an alternate p53 mutated ER⁺ breast cancer cell line (T47D) and a cyclin D1 overexpressing cell line derivative (data not shown). As cyclin D1 overexpression is more commonly found in wild-type p53 ER⁺ tumors,³⁸ we pursued further work in MCF7 cells, which express wild-type p53.

Figure 1. Treatment with combination therapy results in the lowest proliferation rates and highest levels of apoptosis. (A) MCF7/V and (B) MCF7/CD1 cells were treated over the course of 9 d with control DMSO, 400 nM CDK4i, 30 nM doxorubicin, or combination therapy, and an MTS assay was used to measure cell number at the indicated time points. Absorbance for CDK4i, doxorubicin, and combination treated cells was compared with control treated cells within each cell line using a Student t test to obtain P values. Absorbance values are shown ± SE. (C) MCF7/V and (D) MCF7/CD1 cells were treated over the course of 2 d with control DMSO, 400 nM CDK4i, 30 nM doxorubicin, or combination therapy. Cells were fixed and stained using a TUNEL assay and cells undergoing apoptosis were counted and normalized to control treated cells to obtain fold changes. Experiments were repeated 3 times and representative results are shown. * indicates P < 0.05.

CDK4i and doxorubicin treatment induces apoptosis

We next assessed if combination therapy also impacted induction of apoptosis. MCF7/V and MCF7/CD1 cells were treated with control DMSO, CDK4i, doxorubicin, or combination therapy for 48 h and apoptosis was measured by TUNEL staining. When compared with control cells, we found a modest increase in apoptosis in MCF7/V cells (Fig. 1C) following individual CDK4i or doxorubicin treatment, with the highest level of apoptosis observed in response to treatment with combination therapy. For MCF7/CD1 cells, when compared with control treatment, individual CDK4i and doxorubicin treatments resulted in increased levels of apoptosis, with combination therapy resulting in the highest level of apoptosis. Overall, treatment of MCF7/CD1 cells (Fig. 1D) showed a trend toward a greater impact on apoptosis when compared with MCF7/V cells (although nonsignificant). These data further support the effect of combination therapy on induction of cancer cell death.

Transduction of Smad3 phosphorylation site mutants and CDK4i/chemotherapy treatment results in decreased colony formation in 3D culture

We implemented a more biologically relevant 3D culture model to expand upon the MTS and TUNEL 2D data and more accurately capture the impact of the study treatments in a 3D environment. While our 2D and 3D results are relatively comparable, previous studies have also characterized breast cancer cells to behave distinctly in 2D vs. 3D culture.³⁹ To this end, MCF7/V and MCF7/CD1 cell lines were transduced with control CS2 vector, WT Smad3, or 5M Smad3 constructs to assess the impact of Smad3 action and CDK4i/chemotherapy treatment on study cell colony growth. We have previously confirmed the presence of nuclear and cytoplasmic Smad3 and phosphorylated Smad3 in the study cell lines. We have also previously shown the relative increase in CDK4 kinase activity found in MCF7/CD1 cells when compared with MCF7/V cells.¹⁷ Currently, we confirmed expression of Smad3 and cyclin D1 in both study cell lines. As expected, Smad3 expression increased in the study cells transduced with WT or 5M Smad3 (Fig. 2A). Expression of cyclin D1 also increased in MCF7/CD1 cells when compared with MCF7/V (Fig. 2A). Following transduction, tumor colony formation was assessed in Matrigel at days 4 and 12. When examining the control treated cells, MCF7/V cells (Fig. 2B) formed smaller colonies when compared with MCF7/CD1 cells (Fig. 2C) after 12 d, demonstrating the effect of cyclin D overexpression on increased cellular proliferation in this setting. For MCF7/V cells on day 12, transduction of WT Smad3 and 5M Smad3 constructs resulted in a similar decrease in colony area over time (P < 0.05), when compared with transduction of the CS2 vector. While WT Smad3 transduction resulted in a significant reduction in colony growth in MCF7/V cells, in the context of cyclin D overexpression (MCF7/CD1 cells), this growth suppression was not observed. Notably, when compared with transduction with the CS2 vector, MCF7/CD1 cells transduced with the 5M Smad3 construct showed a significant decrease in colony area on day 12 (P < 0.0001) vs. MCF7/CD1 cells transduced with the WT Smad3 construct. These findings support the role of intact Smad3 signaling, resistant to CDK phosphorylation, in the suppression of tumor cell proliferation.

Figure 2. Treatment with CDK4i and doxorubicin combination therapy or transduction with 5M Smad3 suppresses colony growth. (A) MCF7/V and MCF7/CD1 cells were transduced with CS2 vector, WT Smad3, or 5M Smad3. Protein was extracted and immunoblotted with Smad3, cyclin D, and GAPDH. (B) MCF7/V and (C) MCF7/CD1 cells were transduced with CS2 vector control, WT Smad3, or 5M Smad3 and treated over the course of 12 d with control DMSO, 400 nM CDK4i, 30 nM doxorubicin, or combination therapy. Colony area was measured at the indicated time points using Metamorph software. Experiments were repeated 3 times and representative results are shown. Error bars indicate ± SE. * indicates P < 0.05 for comparisons between vector transduced cells and WT and 5M transduced cells treated with control. ^ indicates P < 0.05 for comparisons between control treatment and study treatments for each transduced construct within each cell line.

We next compared the impact of the study treatments to control within the CS2 vector, WT and 5M Smad3-transduced cells. For the MCF7/CD1 CS2 vector transduced cells, all treatment conditions resulted in a more significant decrease in colony size (P < 0.0001), when compared with vector-transduced MCF7/V cells (P < 0.05). For the WT and 5M-transduced MCF7/CD1 cells, combination treatment resulted in a significant decrease in colony size (P < 0.01), while no significant change was seen after combination treatment for the WT and 5M-transduced MCF7/V cells. Lastly, for both study cell lines, transduction of the 5M Smad3 construct resulted in a similar decrease in colony size when compared with treatment with the CDK4 inhibitor, suggesting that restoration of Smad3-mediated tumor suppressive activity could be achieved through the pharmacological inhibition of CDK-mediated phosphorylation of Smad3.

Expression of apoptotic proteins is affected by CDK4i/doxorubicin combination therapy

To further examine the potential relationship between CDK4i, doxorubicin, and the induction of apoptosis, we explored multiple candidate factors contributing to this mechanism. The Proteome Profiler Human Apoptosis Antibody Array was employed to characterize expression of critical apoptosis-related proteins after CDK4i and doxorubicin treatment, as well as combination therapy. The impact of treatment on expression levels of five selected proteins is shown for MCF7/V (Fig. 3A) and MCF7/CD1 (Fig. 3B) cells. Levels of Fas, a key protein in the extrinsic apoptotic pathway, were not significantly altered upon CDK4i treatment in either study cell line, but were increased with doxorubicin and combination therapy. Levels of the CDKi p21 increased in a similar manner after CDK4i or doxorubicin treatment in MCF7/V cells, with the greatest increase observed in response to combination therapy. While CDK4i treatment led to a minimal increase in expression of p21 in MCF7/CD1 cells, doxorubicin alone and combination treatment resulted in a greater (and similar) increase in p21 expression. Of particular interest, survivin and XIAP levels significantly decreased and Smac/DIABLO levels generally increased after treatment with the study drugs alone and in combination. Collectively, these findings point toward the contribution of CDK4i/doxorubicin to the repression of survivin and the potential to facilitate apoptosis in the study cells. Acknowledging the limitations of high-throughput assays, we then used standard immunoblotting to further delineate the changes found in the array studies. We confirmed a significant decrease in survivin protein levels after combination treatment, when compared with control, for the MCF7/V and MCF7/CD1 cells (Fig. 4A). Additionally, expression of XIAP significantly decreased in MCF7/CD1 cells upon combination CDK4i/doxorubicin treatment (Fig. 4A). While MCF7/V cells responded to individual and combination treatment with a trend toward an increase in Smac/DIABLO expression, no significant change in Smac/DIABLO expression was detected in either the MCF7/V or CD1 cells using whole cell lysates. Once activated, Smac/DIABLO is released from the mitochondria into the cytoplasm to bind XIAP. Thus, we next examined if the sub-cellular localization of Smac/DIABLO was altered upon treatment. We found that treatment with CDK4i/doxorubicin resulted in significantly decreased Smac/DIABLO expression in the mitochondrial fraction, and increased expression in the cytoplasmic fraction for MCF7/CD1 cells (Fig. 4B). p21 expression increased in both doxorubicin and combination-treated MCF7/V and MCF7/CD1 cells, while pRb levels decreased in both MCF7/V and MCF7/CD1 combination-treated cells (Fig. 4A).

Figure 3. The array shows that CDK4i and doxorubicin therapy affect expression of proteins associated with apoptosis. (A) MCF7/V and (B) MCF7/CD1 were treated with control DMSO, 400 nM CDK4i, 30 nM doxorubicin, or combination therapy. Protein was collected and applied to the array. Protein expression was quantified using Multigauge software. The array was repeated 3 times and representative results are shown. Error bars indicate ± SE.

Figure 4. Survivin expression decreases upon treatment with CDK4 inhibitor and doxorubicin alone or in combination. (A) MCF7/V and MCF7/CD1 cells were treated with DMSO, 400 nM CDK4i, 30 nM doxorubicin, or combination treatment for 48 h. Protein was extracted, followed by immunoblotting with survivin, XIAP, Smac/DIABLO, p21, pRb, and β-Actin antibodies. (C) MCF7/V and MCF7/CD1 cells were treated with DMSO, 400 nM CDK4i, 30 nM doxorubicin, or combination treatment for 48 h. Mitochondrial and cytoplasmic fractions were isolated, followed by immunoblotting with Smac/DIABLO, Cox IV, and GAPDH antibodies. Experiments were repeated 3 times and representative results are shown. * indicates P < 0.05.

Intact Smad3 signaling and doxorubicin treatment alter apoptotic protein expression

To determine if survivin expression was regulated by Smad3 action, we transduced MCF7/V and MCF7/CD1 cells with vector control, WT Smad3, or 5M Smad3 constructs. We then assessed the expression of survivin, XIAP, and SMAC/Diablo in the transduced cells in the presence or absence of doxorubicin. For both MCF7/V (Fig. 5A) and MCF7/CD1 (Fig. 5B) cells, WT or 5M Smad3 transduction did not result in significant changes in survivin expression. XIAP expression increased upon 5M Smad3 transduction, in MCF7/V, but not MCF7/CD1, cells. Smac/DIABLO expression decreased upon WT Smad3 transduction in the MCF7/V cells, attributed to the negative feedback mediated by TGFβ signaling after Smad3 overexpression (Fig. 5A).⁴⁰ We then tested for potential synergy between doxorubicin treatment and transduction with the 5M Smad3 construct. To this end, we found that doxorubicin treatment and expression of the Smad3 5M protein resulted in a significant decrease in survivin and XIAP expression in MCF7/CD1 cells (Fig. 5B). Correspondingly, a significant increase in Smac/DIABLO expression was seen when 5M Smad3-transduced MCF7/CD1 cells were treated with doxorubicin.

Figure 5. Intact Smad3 signaling and doxorubicin treatment result in the greatest reduction in survivin expression. (A) MCF7/V and (B) MCF7CD1 cells were transduced with CS2 vector control, WT Smad3, or 5M Smad3 and then treated with either DMSO control or 30 nM doxorubicin for 48 h. Protein was extracted, followed by immunoblotting with survivin, XIAP, Smac/DIABLO, and β-actin antibodies. Experiments were repeated 3 times and representative results are shown. * indicates P < 0.05.

Discussion

We sought to determine the effect of CDK4 inhibition, alone or in combination with doxorubicin, on cell proliferation, colony growth, and apoptosis in an in vitro model of cyclin D overexpressing ER⁺ breast cancer. While cell proliferation was effectively reduced in response to treatment with CDK4i or doxorubicin, a greater effect was found when the therapies were combined. Additionally, cyclin D-overexpressing breast cancer cells showed the highest level of apoptosis induction following combination CDK4 inhibitor and doxorubicin treatment. 3D Matrigel culture studies expanded upon the proliferation/apoptosis findings in an environment that allowed the breast cancer cells to grow into colonies, more closely reflecting in vivo cancer growth. Colony growth was decreased after treatment with CDK4i. Cells transduced with 5M Smad3 construct showed a similar decrease in colony size, demonstrating that pharmacologically or genetically inhibiting CDK-mediated Smad3 phosphorylation resulted in breast cancer cell growth inhibition. Consistent with the proliferation/apoptosis assays, combination CDK4i/doxorubicin treatment also resulted in the smallest colony size. Lastly, we identified pro- and anti-apoptotic factors including survivin, XIAP and Smac/DIABLO, to be critical proteins associated with CDK4i/doxorubicin treatment-induced cellular changes.

We have previously shown that cyclin D-overexpressing ER⁺ breast cancer cells transfected a CDK4 phosphorylation-resistant Smad3 construct or treated with CDK4i resulted in increased cdki p15 and decreased c-myc mRNA expression.¹⁷ Additionally, we had found that treatment with CDK4i, as well as transfection with 5M Smad3, resulted in higher Smad3 reporter activity, and CDK4 inhibition-induced cell cycle arrest.¹⁷ These results prompted us to further investigate the impact of CDK4i on cancer cell proliferation and apoptosis. Previous studies with PD0332991, a CDK4/6 inhibitor, showed the greatest inhibition of proliferation occurring in ER+ breast cancer cells.⁴¹ PD0332991 also inhibited tumor growth in the setting of hormone resistant ER+ breast cancer cells in vivo.⁴² Our current work adds to this prior data supporting the anti-proliferative impact of specific CDK4 inhibition in ER⁺ breast cancer. Although results from previous clinical trials using single-agent pan-CDKis have been inconsistent, the outcomes associated with combination CDKi therapy with cytotoxic drugs have been more encouraging.^19,43 Our in vitro proliferation and 3D tumor colony formation data provide further pre-clinical evidence of the potential of these agents. Furthermore, clinical trials implementing CDK inhibitors have indiscriminately included patients irrespective of their tumor cyclin expression profile, and, to date, one clinical trial (NCT00721409) examining the efficacy of letrozole with and without PD0332991, has specifically targeted ER⁺ breast tumors.⁴³ The recently published results of the randomized phase II PALOMA-1 trial are very promising with combination letrozole and PD0332991 resulting in significantly improved progression free survival (20.2 mo) when compared with treatment with letrozole alone (10.2 mo).⁴⁴ Our work suggests that future clinical studies using CDK inhibitors may also benefit from stratifying patients with greater discrimination directed toward cyclin expression status.

Smad3 is a transcription factor that regulates the expression of genes critical to several phenotypic endpoints, including cell cycle control and apoptosis.⁴⁵ The 3D tumor colony data in this study showed that the effect of WT Smad3 vs. 5M Smad3 transduction on colony growth was nearly equivalent for the control MCF7/V cells. It follows that for MCF7/V cells, basal cyclin D/CDK4 activity was not likely sufficient to repress WT Smad3, and thus transduction of both plasmids resulted in a similar significant decrease in colony area. However, in the setting of control-treated, cyclin D-overexpressing MCF7/CD1 cells, 5M Smad3 transduction was more effective in inhibiting colony growth when compared with transduction with the WT construct. These findings suggest that WT Smad3 is phosphorylated and inhibited by the highly active CDK4 in these cells. Conversely, cells transduced with the 5M Smad3 construct were resistant to inhibitory CDK4 phosphorylation, and likely retained tumor suppressive actions including repression of c-myc and promotion of cdki p15 transcription. Additionally, the impact of individual and combination treatment was mitigated in WT Smad3- or 5M Smad3-transduced MCF7/V cells, but remained intact particularly for the WT Smad3-transduced MCF7/CD1 cells. These data provide further evidence for the potential of CDK4 inhibitors to reduce tumor cell growth through the restoration of Smad3-mediated cell cycle control in cyclin D overexpressing breast cancer cells. Lastly, prior work has shown that cyclin D overexpressing cancer cells were more sensitive to doxorubicin.⁴⁶ Consequently, irrespective of Smad3 expression status, overexpression of cyclin D1 in the MCF7/CD1 cells should result in increased cell division, and thus greater sensitivity to the cytotoxic effects of doxorubicin chemotherapy, found in Figure 2C.

The induction of cellular apoptosis concomitant with cell cycle arrest is a key mechanism by which optimal treatments induce breast cancer regression. Doxorubicin is a widely used anthracycline that has strong antineoplastic activity.²¹ We observed a greater increase in apoptosis in response to CDK4i/doxorubicin combination therapy, with MCF7/CD1 cells showing the highest levels of apoptosis after treatment. Additionally, it has previously been shown that doxorubicin treatment activates TGFβ signaling in breast cancer cells through the induction of C-terminal phospho-Smad3 expression.⁴⁷ As C-terminal phosphorylation of Smad3 enables subsequent nuclear translocation, doxorubicin treatment may augment the expression of apoptotic proteins after WT or 5M Smad3 transduction, in part, by increasing the amount of nuclear Smad3. To further examine potential apoptotic factors linked to these findings, we used a Proteome Profiler Human Antibody Array to perform a high-throughput analysis of 35 apoptosis-related proteins. Among these, p21, Fas, survivin, XIAP, and Smac/DIABLO exhibited differential expression patterns as a result of cyclin overexpression and the study treatments. It has been reported that high survivin expression is an independent predictor of poor prognosis for breast cancer patients, and survivin is not typically detectable in healthy adult tissues with low cell turnover.^24,48 Increased survivin levels are also associated with repression of apoptosis and chemoresistance, while inhibition of survivin has resulted in enhanced chemotherapy-mediated apoptotic induction.⁴⁹ As such, survivin may be an important therapeutic target.^24,48

Furthermore, high expression of other factors noted in our array study, including Smac/DIABLO, have been previously shown to enhance the apoptotic effects associated with chemotherapy, including doxorubicin, in multiple breast cancer cell lines.⁵⁰ Expression levels of survivin and Smac/DIABLO mRNA were also found to be reciprocal in primary breast cancer tissues and benign human tissues, respectively.⁵¹ Smac/DIABLO inhibits members of the IAP family, and overexpression of this protein was shown to potentiate the effectiveness of chemotherapy in breast cancer cells previously resistant to caspase-mediated apoptosis.⁵¹ Specifically, Smac/DIABLO directly interacts with and inhibits XIAP. Conversely, overexpression of XIAP can repress the release of Smac/DIABLO from the mitochondria.⁵² In our studies, the cytoplasmic expression of Smac/DIABLO significantly increased in CD1 overexpressing breast cancer cells treated with combination CDK4i/doxorubicin (Fig. 4B), which correlated with significantly lower levels of XIAP (Fig. 4A). Thus in MCF7/CD1 cells, high XIAP expression may be preventing the release of Smac/DIABLO from the mitochondria until a treatment-induced reduction of XIAP occurs. In MCF7/V cells, treatment with CDK4i/doxorubicin did not result in significant changes in XIAP expression, and correspondingly, no significant changes in Smac/DIABLO localization occurred. Recent reports have linked XIAP to expression of CDK4 and cyclin D,^53,54 supporting the observed discrepancies in Smac/DIABLO expression/localization and study treatment impact found for the MCF7/V and MCF7/CD1 cells. In addition, the reciprocal expression of survivin and Smac/DIABLO has been described where after induction of apoptotic stimuli, survivin was released from the mitochondria to then bind and inhibit Smac/DIABLO.⁵⁵ This resulted in decreased Smac/DIABLO available to bind and inhibit other IAPs, such as XIAP, allowing XIAP to directly interact with caspases and block cell death (Fig. 6). Thus, combination CDK4i/doxorubicin treatment may also limit therapeutic resistance through the inhibition of survivin and the promotion of Smac/DIABLO expression.

Figure 6. Proposed model for impact of CDK4i and doxorubicin on cancer cell cycle arrest and apoptosis. Cyclin D overexpression mediates CDK4 phosphorylation of Smad3, resulting in lower expression of the Smad3-regulated cdkis p15 and p21 and higher expression of c-myc and survivin. Increased survivin levels inhibit apoptosis through the regulation of Smac/DIABLO and XIAP. CDK4i treatment results in cell cycle arrest and induction of apoptosis through inhibition of CDK4-mediated phosphorylation of Smad3 and the consequent increase of p15 and p21 expression and decrease of c-myc and survivin expression. Doxorubicin treatment results in cell cycle arrest and apoptosis, in part, through the induction of p21 and the direct action of p21 on inhibition of survivin. The figure was made with tools from http://www.proteinlounge.com.

Yang et al. have also previously shown that Smad3- and Rb/E2F4-mediated induction of apoptosis occurred through the transcriptional repression of survivin, and that decreased levels of Rb/E2F4 complex formation correlated with higher levels of survivin.³⁶ It is possible that these increased survivin levels resulted from Rb inactivation, which occurs primarily through CDK mediated phosphorylation. Accordingly, we found that in MCF7/V and MCF7/CD1 cells, combination CDK4i/doxorubicin treatment resulted in decreased pRb expression. The role of doxorubicin in regulation of survivin has also been studied, in which a doxorubicin-induced and p53-mediated increase of p21 expression were shown to inhibit survivin in HepG2 human hepatocellular carcinoma cells.⁵⁶ The reduction in survivin and XIAP and increase in Smac/DIABLO expression, found after the addition of doxorubicin treatment to 5M Smad3-transduced cells, further demonstrate how combination therapy can amplify the efficacy of either agent alone through a potential Smad3, Rb/E2F4, and doxorubicin-induced p53/p21 mediated repression of survivin and subsequent increase in apoptosis.

Combination CDK inhibitor and doxorubicin therapy has not been specifically studied in the context of Smad3 signaling in breast cancer cells. Previous work found that this combination therapy may be suboptimal in the context of hormone insensitive breast cancer cells, where CDK inhibition may protect the RB-proficient cancer cells from the impact of doxorubicin.⁵⁷ However, in our studies implementing a hormone sensitive MCF7 breast cancer model, combination therapy was found to be effective. Simultaneously, synergism of Seliciclib, a pan-CDK inhibitor, and doxorubicin has also been observed in a breast cancer MCF7 xenograft model.⁵⁸ In that study, synergism was found to occur through the induction of cell cycle arrest associated with the upregulation of p21. Our immunoblotting studies also showed significant induction of p21 for both the MCF7/V and CD1 cells treated with doxorubicin. As, stated, the impact of doxorubicin was expected as this agent induces expression of p21 in a p53-dependent manner.²⁰ Furthermore, both MCF7/V and MCF7/CD1 cells treated with combination CDK4i/doxorubicin showed significantly increased p21 expression. Expression of p21 has also been linked to Smad3 in association with Sp1 transcription factor activation.⁵⁹ Thus, as we have previously shown, CDK4i-mediated restoration of canonical Smad3 action may also contribute to the induction of p21, particularly for the MCF7/V cells.¹⁷ These data highlight the benefit of selective treatment strategies that employ combination therapy, rationally impacting several signaling pathways, to collectively contribute to cancer cell cytostatic and cytotoxic events. To this end, Fas, which also increased after combination therapy in our array studies, is a potentially important protein for future work associated with the mediation of extrinsic apoptosis.

Collectively, our work demonstrates the potential therapeutic benefit of combining CDK4i with chemotherapy in an ER⁺ cyclin D-overexpressing breast cancer model. We showed that decreased cell proliferation, smaller colony formation, and higher apoptosis levels occurred in breast cancer cells treated with combination CDK4i/doxorubicin. We also found that these tumor suppressive events are, in part, regulated by Smad3, and that pharmacological and genetic inhibition of CDK4-mediated Smad3 phosphorylation result in decreased cell proliferation and increased apoptosis. We have previously shown that noncanonical phosphorylation of Smad3 by CDKs affects the expression of p15 and c-myc in breast cancer cells. The current study points toward expression of survivin as an additional key protein regulated by Smad3 signaling, contributing to important work focusing on cancer cell specificity, co-activators, and co-repressors, and also allowing for greater comprehension of the broad-spectrum of influences mediated by the TGFβ superfamily. Further study of CDK-mediated regulation of survivin may have implications for cyclin D-overexpressing ER⁺ breast cancers in which chemoresistance may be associated with high survivin levels. Overall, the incorporation of CDK inhibition into chemotherapeutic regimens may result in decreased survivin levels, increased apoptosis, and tumor suppression, and contribute to improved patient outcomes.

Methods

Cell culture and generation of stable cell lines

MCF7 cells were obtained from the American Type Culture Collection and maintained in DMEM-F12 according to supplier recommendations. Culture medium was supplemented with 1% Pen Step/Glutamine, 1% nonessential amino acids, 1% sodium pyruvate, and 10% fetal bovine serum. Human cyclin D1 cDNA was inserted into the cytomegalovirus-driven pcDNA-DEST40 expression vector (Invitrogen) containing a neomycin resistance gene to obtain the pEXPR-cyclin D1-expressing plasmid. To create the vector control and cyclin D1-overexpressing (CD1) cell lines, cells were stably transfected with empty vector (pcDNA-DEST40) or pEXPR-cyclin D1 plasmids. Two days after transfection, stable clones were selected in medium containing 500 mg/mL G418 (Invitrogen). The resistant clones were maintained in G418 for 21 d. Cyclin D1 overexpression was verified by immunoblot analysis. Study cells stably expressing empty vector or cyclin D1 were maintained in the same manner as the parental lines.

Antibodies and reagents

Antibodies used were as follows: anti-cyclin D1, anti-Smad3, anti-pSmad3 (Ser423/425), anti-XIAP, anti-Smac/DIABLO, anti-COXIV, anti pRb (Cell Signaling), anti-CDK4, anti-GAPDH, anti-survivin, anti-p21 (Santa Cruz), and anti-β-actin (Sigma). CDK4 inhibitor II (IC₅₀ value of 200 nmol/L) was from EMD Biosciences and doxorubicin (IC₅₀ value of 0.1 μmol/L) was from Sigma. Each drug was resuspended in dimethyl sulfoxide (DMSO), and DMSO was used as the control treatment. Where indicated, cells were treated with the following doses: 0.05% DMSO, 400 nM CDK4i (based on prior dose response studies),¹⁷ and 30 nM doxorubicin.

Cell proliferation

The effect of CDK4 inhibitor, doxorubicin, or combination treatment on MCF7 cell proliferation was examined using the MTS (3-[4,5-dimethylthiazol-2-yl]-5-[3-carboxymethoxyphenyl]-2-[4-sulfophenyl]-2H-tetrazolium, inner salt) assay in the CellTiter 96 aqueous non-radioactive cell proliferation assay kit (Promega), performed according to the manufacturer's instructions. Briefly, study cells were seeded at a density of 2500 cells/100 μL media in triplicate wells of a 96-well plate and incubated for 24 h in 5% CO₂/air at 37 °C. The medium was aspirated and replaced with 100 μL fresh medium containing treatment regimen or control DMSO. All drugs were reconstituted in MCF7 media. Cells were incubated with a media change and cell proliferation assessed at days 0, 2, 5, 7, and 9. After a 2-h incubation, the plates were read at 490 nM in a Biorad Microplate Reader. Experiments were repeated 3 times and representative results are shown. Drug concentrations were based on initial dosage studies.¹⁷

Apoptosis

Apoptosis was measured for each of the treatment groups using the Terminal deoxynucleotidyl transferase mediated dUTP Nick End Labeling (TUNEL) assay. Cells were treated for 48 h as described above with control DMSO, CDK4 inhibitor, doxorubicin, or combination therapy. The supernatant was collected to isolate any floating apoptotic cells. Following trypsinization, all cells were spun down, rinsed with cold phosphate buffered saline (PBS), and fixed in 1 mL 10% neutral buffered formalin. Fixed cells were embedded into paraffin blocks and TUNEL analysis was performed at the Pathology Core Facility of Northwestern University. Impact of treatments was normalized to control treatment within each cell line. Experiments were repeated 3 times and representative results are shown.

3D cell culture

MCF7/V and MCF7/CD1 cells were infected with GFP expressing lentiviral constructs containing CS2 vector control, WT Smad3, or 5M Smad3. The 5M Smad3 expression plasmid has been described previously, and contains multiple CDK phosphorylation site mutations (T8/T179/S204/S208/S213) that result in the inhibition of CDK4/2 phosphorylation.¹¹ This Smad3 mutant was a generous gift from Dr Fang Liu (Rutgers, State University of New Jersey). GFP-positive cells were isolated on the Beckman Coulter Elite at the Northwestern Flow Cytometry core facility. After multiple passages of stable cell lines, sorted Smad3 expressing cells were trypsinized and resuspended in growth medium containing 2% Matrigel (BD Bioscience). Cells were seeded at a density of 1000–1500 cells/well in a BD Matrigel Matrix 96-well plate (BD Bioscience) containing an underlying 1-mm thick bed of Matrigel. Cells were treated as described above. Cells received fresh medium every 48 h. Pictures of growing colonies were taken on days 4 and 12. Average colony size was measured using MetaMorph software (Molecular Devices). Experiments were conducted three times, and representative results are shown.

Apoptosis antibody array

The Proteome Profiler Human Apoptosis Antibody Array Kit (R&D Systems, Inc.) was used to determine the relative levels of apoptosis-related proteins in the treated cells. According to manufacturer instructions, 400 µg of protein lysate from treated cells was incubated with array membranes for 4 d. Membrane images were taken using an LAS 3000 imaging system and quantitative analysis was performed with MultiGauge software (FujiFilm). Numerical values corresponding to the signal intensity of each spot on the array were exported into a spreadsheet file. The background signal was subtracted from total signal values of each spot and each pair of duplicate spots was averaged for each protein in each treatment group. Experiments were repeated 3 times and representative results are shown.

Immunoblotting

Where indicated, cells were either treated as described above for 48 h or transduced with lentivirus expressing CS2⁺ control vector, WT Smad3, or 5M Smad3 mutant construct (5M Smad3). Thereafter, cells were scraped, pelleted, and rinsed with ice-cold PBS, then lysed in ice-cold M-PER (Thermo Scientific) containing protease inhibitors. Cellular lysates were spun, supernatants were recovered, and the protein concentration was determined using the Bio-Rad protein assay kit. For the lysates subjected to mitochondrial fractionation, cells were scraped and homogenized in sucrose buffer (250 mM sucrose, 10 mM TRIS-HCl, 0.1 mM EGTA) and spun at 100 × g for 10 min. The supernatant was collected in a new tube and spun at 10 000 × g for 10 min. The supernatant (cytoplasmic fraction) was transferred to another tube and the pellet (mitochondrial fraction) was suspended in 50 uL of sucrose buffer. Cox IV, a mitochondrial protein, and GAPDH, a cytoplasmic protein, served as the control for the mitochondrial fractionation. Fifty micrograms of lysate was diluted in 2× SDS-PAGE sample buffer (1:1,v/v), electrophoresed, and transferred to a PVDF membrane. Membranes were blocked with 5% milk or 5% BSA in TBS-T at room temperature for 1 h and incubated with appropriate primary antibody at 4 °C overnight. After rinsing with TBS-T, the membrane was incubated with secondary antibody in 5% milk or 5% BSA for 1 h at room temperature. Protein bands were visualized by an ECL detection system (Supersignal, Thermo Scientific). For reblotting, membranes were agitated with Restore PLUS Stripping Buffer (Thermo Scientific) for 15 min at room temperature. Experiments were repeated 3 times and representative results are shown.

Statistical analysis

Results were presented as mean values ± SE and analyzed by the unpaired, two tailed Student t test. P < 0.05 was considered statistically significant.

Abbreviations:
ER	=	estrogen receptor
CDK	=	cyclin-dependent kinase
CDKi	=	cyclin-dependent kinase inhibitor
Rb	=	retinoblastoma
TGFβ	=	transforming growth factor beta
IAP	=	inhibitor of apoptosis
EMT	=	epithelial-to-mesenchymal transition
DMSO	=	dimethyl sulfoxide
Smac	=	second mitochondria-derived activator of caspase

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.

Acknowledgments

We thank Dr Stacey Tobin and Megan Novak for their editorial assistance and Dr Vamsi Parini for assessment of TUNEL staining.

Financial Support

J.S.J. is a Lynn Sage Scholar supported by the NIH K22 CA138776 and R01GM097220 research grant, the Central Surgical Association Foundation, the Society of Surgical Oncology, the Saslow family, and A Sister’s Hope. E.T. is a Dr John N. Nicholson Fellow, a Chicago Biomedical Consortium Scholar supported by the Searle Funds at the Chicago Community Trust, and is also supported by the NIH/NCI training grant T32CA09560 and the National Cancer Institute Kirchstein-NRSA Fellowship F31 CA168106-02.