Cyclin D1 cooperates with p21 to regulate TGFβ-mediated breast cancer cell migration and tumor local invasion

Human breast cancer cell lines MDA-MB-231 (hereafter referred to as MDA) and SCP2 (provided by Dr. Joan Massagué) were cultured in DMEM containing 10% fetal bovine serum (FBS) and 2 mM L-glutamine. SUM149PT, SUM159PT and SUM229PE (provided by Dr. Stephen P. Ethier) were plated in F-12 HAM'S nutrient mixture (HyClone Laboratories, Inc., Logan, UT, USA) containing 5% FBS, 5 µg/ml insulin (Sigma-Aldrich, St. Louis, MO, USA), and 1 µg/ml hydrocortisone (Sigma). SUM1315MO₂ were cultured in F-12 HAM'S nutrient mixture (HyClone) containing 5% FBS, 5 µg/ml insulin (Sigma), and 10 ng/ml epidermal growth factor (EGF) (Sigma). All cell lines were grown at 37°C in 5% CO₂. Before stimulation with 5 ng/ml TGFβ1 (PeproTech, Rocky Hill, NJ, USA), cells were serum-starved overnight. For cell transfection, flag-tagged p21 cDNA (Addgene plasmid 16240), HA-tagged cyclin D1 cDNA (Addgene plasmid 11181), scrambled and cyclin D1 siRNAs (Sigma) were transfected using Lipofectamine™ 2000 (Invitrogen, Carlsbad, CA, USA), according to the manufacturers' protocols.

Protein extraction buffer containing 10 mM Tris-HCl, pH 7.5, 5 mM EDTA, 150 mM NaCl, 30 mM sodium pyrophosphate, 50 mM sodium fluoride, 1 mM sodium orthovanadate, 1% Triton X-100 and protease inhibitors (1 mM phenylmethylsulfonyl fluoride, 10 µg/ml leupeptin hydrochloride, 10 µg/ml aprotinin and 10 µg/ml pepstatin A) were freshly prepared and kept at 4°C before cell lysis. After cell lysates were centrifuged at 14,000 rpm for 15 minutes at 4°C, the concentration of total protein was quantified using a BCA protein assay kit (Thermo Scientific, Rockford, IL, USA). Cell lysates were boiled with 6× sodium dodecyl sulfate (SDS) Laemmli sample buffer for five minutes and subjected to immunoblot using mouse anti-p21 and rabbit anti-cyclin D1 antibodies (1:1,000 dilution, Santa Cruz Biotechnology, Santa Cruz, CA, USA). p21 and cyclin D1 were immunoprecipitated overnight at 4°C using their respective antibodies and followed by the addition of protein G-Sepharose beads (GE Healthcare Bio-Sciences, Piscataway, NJ, USA) for one hour at 4°C. The immunocomplexes were washed four times with cold lysis buffer and then subjected to Western blot.

Cell migration was performed as previously described [44]. Briefly, 50,000 cells per well were cultured in Essen ImageLock 96-well plates (Essen Bioscience, Ann Arbor, MI, USA). The confluent cell layers were scratched to generate a wound using the Essen Wound maker. Cells were then treated in the presence or the absence of 5 ng/ml of TGFβ1. The images/videos of the wound were automatically taken at the exact same location using the IncuCyte™software (Essen Bioscience). Wound width, wound confluence or relative wound density were automatically measured by the IncuCyte software.

Transfected cell suspensions (40,000 cells/well) were seeded in 24-well Cell Culture Inserts (8-µm pore size; BD Biosciences, Mississauga, ON, Canada). After 24 hours incubation, the cells that migrated to the bottom of the membrane were fixed with 3.7% formaldehyde for 10 minutes and then labeled with the near-infrared fluorescence DNA binding dye DRAQ5 (2 µg/ml in PBS) at 37ºC for 5 minutes. The fluorescence intensity of migrated cells was measured at 700 nm in a near-infrared fluorescence imager (Odyssey CLX, LI-COR Biosciences - Biotechnology, Lincoln, NE, USA) using the Image Studio software (LI-COR Biosciences - Biotechnology).

For the invadopodia formation assay, cells were grown on top of eight-well chamber slides coated with 100 µl growth factor-reduced Matrigel. After TGFβ treatment for 24 hours, cells were fixed with 3.7% formaldehyde for 10 minutes, permeabilized in 0.1% Triton X-100 for 3 minutes, and blocked for 1 hour in 2% bovine serum albumin (BSA). Fixed cells were incubated with primary antibodies against p21, cyclin D1, F-actin and vimentin for one hour and followed by the secondary antibodies Alexa Fluor®568 goat anti-rabbit IgG and Alexa Fluor®488 goat anti-rabbit (1:800 dilution; Invitrogen) for one hour. Nuclei were stained with DAPI (Invitrogen). Confocal analysis was performed using a Zeiss LSM 510 Meta Axiovert confocal microscope (Carl Zeiss, Oberkochen, Baden-Württemberg, Germany) using the 63× objective.

The animal study and SCP2 cells used in the mice model were approved by the McGill ethics committee (University Animal Care Committee, UACC) and all the experimental animal protocols were in accordance with the McGill University Animal Care. Four- to six-week old female Balb/c nude mice (Charles River Laboratories International, Wilmington, MA, USA) were used as a model for assessing mammary tumor formation and local invasion. An anesthetic cocktail of ketamine (50 mg/kg), xylazine (5 mg/kg) and acepromazine (1 mg/kg) was injected intramuscularly into mice (six per group). Fifty thousand parental SCP2 cells or p21 and cyclin D1 double knockdown SCP2 cells in 100 μl of saline (20% Matrigel) were injected into the mice mammary fat pads using a 30-gauge needle. Tumor growth and size were measured using a caliper. At eight weeks post-injection, mice were sacrificed and mammary tumors with surrounding skin and tissues were fixed in 10% neutral-buffered formalin for one day. Sections of mammary tumor were embedded in Tissue-Tek O.C.T. (VWR International, Radnor, PA, USA) compound and 9 µm thick sections were stained with hematoxylin and eosin to assess local advanced features, including skeletal muscle, mammary fat pad, and lymphovascular invasion as well as skin ulceration. Images of the tumors were photographed by light microscope using 10× and 20× objectives.

For intratibia injections, parental and p21/cyclin D1-depleted SCP2 cells (2.0× 10⁶) were injected intramuscularly into the left tibia of two group mice (six per group). The mice were monitored weekly for tumor burden. Digital radiography of hind limbs of all animals was used to monitor the development of skeletal lesions at four, six and eight weeks post-injection in a MX-20 cabinet X-ray system (Faxitron X-ray Corp.). On week 8, radiographs of anesthetized mice were taken.

The difference between groups was analyzed using Student's t-test, and *P <0.05 was considered statistically significant.

We have previously shown that TGFβ's pro-migratory and invasive effects are mediated through the induction of p21 in highly migratory triple negative breast cancer cells [44]. Due to the fact that p21 is a universal regulator of cyclin/CDKs, this prompted us to investigate whether additional cell cycle regulators downstream of TGFβ are involved in this process. Moreover, because multiple studies have suggested that the oncogenic effects of cyclins may not be simply due to enhanced tumor cell growth or proliferation but may also involve tumor promoting functions [45‐48], we examined the effect of TGFβ on protein expression levels of cyclins A, B1, D1 and D2 in the human aggressive breast cancer cell lines MDA and its metastatic sub-progeny SCP2 [49]. As shown in Figure 1A, we found that TGFβ significantly increased cyclin D1 protein expression in a time-dependent fashion. The effect of TGFβ on cyclin D1 expression was specific, as protein levels of G1 and S phase regulators cyclin D2 and A remained unchanged in response to TGFβ stimulation. The M phase cyclin B1 was barely detectable. As a positive control, we measured the expression of p21, which we have previously shown to be potently induced by TGFβ in MDA cells [44]. TGFβ induced the expression of p21 in a similar temporal expression pattern as cyclin D1 in these breast cancer cells. To assess whether TGFβ regulates cyclin D1 at the transcriptional level, we measured mRNA levels of cyclin D1 by quantitative PCR in SCP2 cells stimulated with TGFβ for 2, 6 and 24 hours. Induction of cyclin D1 mRNA by TGFβ was already detectable at 2 hours and was sustained for up to 24 hours (Figure 1B). These results highlight cyclin D1 as a novel TGFβ downstream target gene in human breast cancer cells.

To determine whether there was an association between TGFβ induction of cyclin D1 and TGFβ's pro-migratory effect, we measured the mRNA level of cyclin D1 in a panel of triple negative breast cancer cell lines which are either insensitive (SUM1315) or responsive (SUM149 and SUM159) to TGFβ-mediated cell migration and invasion [44]. Interestingly, TGFβ potently and persistently up-regulated cyclin D1 mRNA in the highly migratory cell lines SUM149 and SUM159, but not in the TGFβ-insensitive SUM1315 cell line (Figure 1C). Together, these results indicate that TGFβ-induced cyclin D1 expression correlates with TGFβ-induced p21 gene expression and cell migration, thus, suggesting that cyclin D1 may be associated with p21 and participate in TGFβ tumor-promoting functions in breast cancer cells.

The intracellular localization of cyclin D1 is important for its function and is, therefore, tightly regulated [50]. Constitutive accumulation of cyclin D1 in the nucleus has been shown to promote tumor transformation [51]. To determine whether TGFβ regulates cyclin D1 nuclear localization, we assessed the localization of cyclin D1 in MDA and SCP2 cells treated with or without TGFβ for 24 hours by confocal immunofluorescence microscopy. Cyclin D1 was predominantly found in the cytosol in unstimulated cells, whereas it appeared to be primarily retained within the nucleus after treatment with TGFβ (Figure 2A).

We have previously shown that TGFβ induces protein expression and nuclear localization of p21 in triple negative breast cancer cells [44]. The concurrent TGFβ effect on p21 and cyclin D1 prompted us to determine whether these molecules co-localize within the nucleus in response to TGFβ. As shown in Figure 2B, TGFβ facilitates nuclear co-localization of cyclin D1 and p21 in MDA cells. The simultaneous induction and co-localization in the nucleus of cyclin D1 and p21 by TGFβ suggested that they may be physically associated with each other. To address this, we performed co-immunoprecipitation of p21 and cyclin D1 in MDA and SCP2 cells treated with or without TGFβ for 6 or 24 hours. As shown in Figure 2C, TGFβ stimulated the interaction between endogenous p21 with cyclin D1 in a time-dependent fashion in MDA and SCP2 cells. Reciprocal immunoprecipitation experiments confirmed that endogenous cyclin D1 specifically interacts with immunoprecipitated p21 in response to TGFβ in MDA cells (Figure 2D, left panel). Moreover, the induction of complex formation between endogenous cyclin D1 and p21 was also observed in both SUM149 and SUM159 cells (Figure 2D, middle and right panels). Collectively, these results indicated that TGFβ stimulates the formation of a complex between cyclin D1 and p21 in triple negative basal-like breast cancer cells.

Given that TGFβ enhanced cyclin D1 and p21 expression and complex formation in these human metastatic breast cancer cells, we investigated whether the TGFβ pro-migratory effect is mediated through cyclin D1. To address this, SCP2 cells were transfected with scrambled (Scr) siRNA or cyclin D1 siRNA. Cell migration in response to TGFβ was assessed by the scratch/wound healing assay coupled to quantitative time-lapsed imaging for up to 24 hours. As shown in Figure 3A, TGFβ-induced cyclin D1 protein expression in the SCP2 cells transfected with Scr siRNA was blocked in cells transfected with cyclin D1 siRNA. As shown in Figure 3B, C, while TGFβ stimulated rapid wound closure in SCP2 cells transfected with the Scr siRNA, this effect was delayed in SCP2 cells depleted of cyclin D1. TGFβ-induced wound closure was not affected by the mitotic inhibitor mitomycin C, suggesting that the effect of TGFβ on cell migration was independent of cell proliferation (Figure 3D). We further assessed the role of cyclin D1 downstream of TGFβ-mediated cell migration, using a Transwell migration assay. As shown in Figure 3E, F, knocking down cyclin D1 inhibited the TGFβ pro-migratory effects, consistent with what observed with the wound healing assay (Figure 3B, C).

To then address whether cyclin D1 and p21 have any synergistic effect, p21 and cyclin D1 cDNAs were overexpressed alone or in combination and the TGFβ effect on cell migration was examined using both the wound healing and Transwell migration assays. As shown in Figure 3G, overexpression of cyclin D1 or p21 alone had little or no potentiation effect on TGFβ-induced cell migration. However, overexpression of both proteins clearly increased/potentiated the TGFβ effect, suggesting that these two proteins synergize their effect downstream of TGFβ. This is consistent with our main finding and conclusion, showing that the two proteins cooperate to regulate TGFβ-mediated breast cancer cell migration and tumor local invasion. Together, these results demonstrate that cyclin D1 is required for TGFβ-mediated migration in breast cancer cells.

Cyclin D1 has previously been reported to regulate cellular migration in primary bone macrophages, mouse embryo fibroblasts (MEFs), and breast cancer cells [52‐54]. For instance, cyclin D1-deficient MEFs display a more spread phenotype, and an increased cell adhesion and actin stress fiber formation through inhibition of thrombospondin 1 and ROCK signaling [53]. Therefore, we examined whether cyclin D1 effects on cellular structure and actin organization contribute to TGFβ-mediated cancer cell migration. To this end, SCP2 cells transfected with either Scr or cyclin D1 siRNAs were stimulated with TGFβ and the dynamics of actin organization were assessed by staining with the fluorescently labeled F-actin marker phalloidin and mesenchymal intermediate filament vimentin. As shown in Figure 4A, vimentin filaments co-localized with F-actin at the leading edge of aggressive SCP2 cells transfected with Scr siRNA, which displayed an elongated phenotype in response to TGFβ. Interestingly, cyclin D1-deficient cells were rounded and exhibited more epithelial-like phenotype. Furthermore, suppression of cyclin D1 expression not only prevented the elongation of vimentin filaments, but also the co-localization with F-actin at the cell edge.

Vimentin is required for the elongation of invadopodia subcellular structures, which are three-dimensional actin-rich protrusions [55]. Invadopodia are selectively found in invasive cancer cells and are important for the degradation of the ECM [56]. As cyclin D1 affects vimentin distribution, we investigated whether cyclin D1 could regulate invadopodia formation. SCP2 cells transfected with either Scr or cyclin D1 siRNAs were seeded on top of growth factor-reduced Matrigel and treated with or without TGFβ. Whereas Scr-transfected SCP2 cells stimulated with TGFβ showed increased F-actin-bundled protrusion and invaded into the Matrigel, this phenotype was completely abolished by knocking down cyclin D1 expression (Figure 4B). All together, these results defined novel functions for cyclin D1 as a TGFβ downstream target that is required for this growth factor to mediate vimentin elongation, induction of a migratory morphological phenotype, and the formation of invasive subcellular structures in metastatic breast cancer cells.

Overexpression of p21 and cyclin D1 is correlated with poor prognosis and aggressiveness in breast cancer. To address the importance of p21 and cyclin D1 on breast cancer development in vivo, we injected either SCP2 control or double p21 and cyclin D1 knockdown cells into the mammary fat pads of female Balb/c nude mice to monitor primary tumor growth and local invasiveness. Silencing p21 and cyclin D1 expression using siRNAs significantly reduced the rate of primary tumor formation and tumor size (Figure 5A). As depletion of p21 alone did not affect tumor formation in a Xenograft transplantation in vivo model [55], it is likely that the observed phenotype on tumor formation in the double knockdown is mediated by cyclin D1. This is in agreement with previous studies showing that depletion of cyclin D1 prevented tumor development in oncogenic HER2 overexpressing transgenic mice [57‐59]. Importantly, three out of six mice in the control group had tumors ulcerating through the overlaying skin, while all the mice in the double knockdown group had intact skin. Breast tumor with ulcerated skin has been clinically classified as locally advanced breast cancer. All tumors were taken with the overlaying skin and surrounding tissues and subjected to hematoxylin and eosin staining. As shown in Figure 5B, the deep tumor margins in the control group were less distinct, invading nearby structures, including skeletal muscles and the mammary fat pad, and showed frequent lymphovascular invasion. However, the tumor margins in the knockdown group were well encapsulated with a non-invasive nature. In addition, we performed immunohistochemistry on primary mammary tumor derived from animals injected with parental and p21/cyclin D1-depleted SCP2 cells. We assessed the expression of the TGFβ-regulated gene PTGS2, which we have previously shown to be involved in mediating the TGFβ effect on cell migration and invasion [44]. As shown in Figure 5C, using tumors from four different mice in each group, we found expression of PTGS2 to be clearly higher in parental tumors compared to p21/cyclin D1-depleted tumors, further confirming that the p21/cyclin D1-depleted tumors displayed less invasive features.

To investigate the role of p21 and cyclin D1 on the development of bone osteolytic lesions, parental and double knockdown SCP2 cells were injected intramuscularly into the left tibia of two groups of nude mice. As shown in Figure 5D, following X-ray examination of the bones, both group of mice developed secondary tumors that caused severe osteolytic bone lesions, suggesting that p21/cyclin D1 do not affect the later stages of bone metastasis. Collectively, these results indicate that while p21 and cyclin D1 are required for breast cancer cells to acquire an invasive phenotype, their effects are primarily occurring at the earlier stages of tumor metastasis, namely induction of local cell invasion from the tumor to the surrounding tissues. This is also consistent with previous work, showing that depletion of p21 alone did not affect the development of bone osteolytic lesions [44].