Suppression of KIF3A inhibits triple negative breast cancer growth and metastasis by repressing Rb-E2F signaling and epithelial-mesenchymal transition

2.1 Tissue samples and cell lines

Ninety-eight female patients with primary TNBC treated at the Affiliated Hospital of Qingdao University between 2011 and 2013 participated in this study. TNBC was defined as invasive breast carcinoma with ER and PR staining in less than 1% of the tumor cells by immunohistochemistry and without HER2 overexpression. All patients did not receive chemotherapy or radiotherapy before surgery and the related clinical information (Table 1) was obtained from all patients with written consent. Thirty-two of the patients who had axillary lymph node dissection had lymph node metastasis. Specimens from cancer tissue and adjacent normal tissue were obtained during the surgery and were immediately dipped in 10% formalin for immunohistochemistry (IHC) analysis. Nine patients had both cancer and adjacent tissue sufficiently large so that a portion of the tissue was immediately frozen in liquid nitrogen and stored for RNA/protein analysis. This study was reviewed and approved by the Institutional Medical Ethics Committee of the Qingdao University Affiliated Hospital.

Human TNBC cell lines MDA-MB-231, BT549, MDA-MB-468 and BT20 were purchased from the Chinese Academy of Science (Shanghai, China). The cells were routinely cultured in DMEM (HyClone) supplemented with 10% FBS (Gibco) at 37℃ in 5% CO₂.

2.2 Immunohistochemistry analysis

The procedure of IHC staining was performed as described previously. Anti–KIF3A (abcam, ab11259, dilution at 1:300), anti–phospho-Rb, anti–E2F1, anti–Cyclin E1, anti–Cyclin D1, anti–P21, anti–ZEB1, anti–E-cadherin, anti–vimentin, anti–MMP-9 and anti–MMP-2 (Bioworld Technology, all at dilution 1:200) were incubated overnight at 4°C. Each slide was photographed with a digital camera on an inverted Olympus IX81 microscope. The staining evaluation was made according to the semi–quantitative scoring system, as described previously. Briefly, the scores of positive tumor cell proportion (0, none; 1, <1/100; 2, 1/100 to 1/10; 3, 1/10 to 1/3; 4, 1/3 to 2/3; and 5,> 2/3) and staining intensity (0, none; 1, weak; 2, intermediate; and 3, strong) were added up to obtain a final total score, ranging from 0 to 8.

2.3 Plasmid construction and generation of stable cell lines

Suppression of KIF3A expression was performed by shRNA interference. KIF3A-shRNA recombinant lentiviruses and the negative control (Scr-shRNA) were purchased from HanBio. Uninfected cells were used for empty control (Mock). The target sequences of the KIF3A-shRNA were as follows: 1#:5′- GAT CCGCCCAGACAAGATGATTGAAATGCAATTCAAGAGATTGCATTTCAATCATCTTGTCTGGGTTTTTTC-3′, 2#:5′-GATCCGACTATGCTGATGGCTGCAAAGTCATTCAAGAGATGACTTTGCAGCCATCAGCATAGTCTTTTTTC-3′, 3#:5′-GATCCGCGTTCTGCAAAGCCTGAAACTGTATTCAAGAGATACAGTTTCAGGCTTTGCAGAACGCTTTTTTC-3′. Cells were infected with lentivirus particles and were selected in medium containing puromycin. MDA-MB-231 and BT549 cell stably expressing KIF3A-shRNA, Scr-shRNA plasmid were named KIF3A-shRNA and Scr-shRNA. The overexpression of KIF3A was performed by KIF3A overexpression vector plasmid (KIF3A-pEX), the empty plasmid was used as a negative control (Vector). Plasmids were purchased from GeneChem. All constructs were confirmed by sequencing. MDA-MB-468 cells transfected with KIF3A overexpression vector plasmid and empty plasmid were named KIF3A-pEX and Vector.

2.4 Real-time RT-PCR analysis

The fresh tissue total RNA was isolated using RNAiso reagent (Takara Bio, Japan) as instructed by the manufacturer. The next procedure was performed as described previously.39 Forward and reverse primer sequences for KIF3A and GAPDH were summarized in Table 2.

2.5 Protein preparation and western blot analysis

Protein preparation and western blot assay were performed as described previously.39 Antibodies used were as follows: anti–KIF3A, anti–vimentin (Santa Cruz Biotechnology, dilution at 1:1000), anti–Cyclin D1, anti–Cyclin E1, anti–P21, anti–E2F1, anti–E-cadherin, anti–MMP-2, anti–ZEB1 (abcam, dilution at 1:1000), anti–MMP-9 (Cloud-Clone Corp, dilution at 1:1000), anti–β-actin (TransGen Biotech, dilution at 1:4000) and anti–phospho-Rb antibodies (Abcam, dilution at 1:5000).

2.6 MTT assay and colony-formation assay

The MTT assay procedure was described previously.39 Cell viability was measured by using the MTT assay. Chemotherapeutic drugs doxorubicin (Solarbio) and cisplatin (Solarbio) were used. For colony formation assay, cells (3 × 10³ cells per well) were seeded into 6-well plates for 2 weeks under regular culture conditions. Colonies formed were fixed with methanol for 30 minutes and were sequentially stained with 0.5% crystal violet for half an hour.

2.7 Cell cycle analysis

Cells were washed with ice-cold PBS and harvested by trypsinization. After centrifugation for 5 min, the cells were washed with ice-cold PBS and fixed with 70% ethanol overnight at 4°C. RNaseA (20 μg/mL) was added to the cells for 30 min (at 37°C). Propidium iodine (50 µg/mL) was added to the cells in the dark, which were analyzed by flow cytometry (BD Accuri C6).

2.8 Cell migration and invasion assay

A total of 8 × 10⁴ cells per well were suspended in serum-free medium and loaded onto the upper compartment of the chamber coated with Matrigel (BD Biosciences). After being incubated at 37°C for 24 hours, the invasive cells that migrated through the Matrigel to the medium containing 15% serum in the lower compartment were stained with crystal violet. The numbers of invaded cells were counted in five random microscopic fields (100×). For migration assays, 4 × 10⁴ cells were plated on the upper chambers coated without Matrigel. The assay was then performed as for the invasive assay.

For the scratch test, cells in each group were seeded into 6-well plates (5 × 10⁵ cells/well). Up to near confluence of the cell monolayer, a sterile pipette tip (200 µL) was used to draw a line across each well, followed by PBS washing to remove the debris or detached cells. The images of the scratch under a microscope were observed at 0 and 48 hours after scratching. Scratch widths were calculated using the Image J software. Cell migration distance equates to scratch width at 0 hour − scratch width at 48 hours.

2.9 Xenograft assays in nude mice

Scr-shRNA and KIF3A-shRNA MDA-MB-231 cells (4 × 10⁶) were inoculated subcutaneously into the right back areas of 6-week-old female BALB/c nude mice (n = 5 for each group), respectively. Tumor size and weight were measured every week and tumor volume was calculated with the formula: volume (mm³) = (width² × length)/2. Other twenty 6-week-old female BABL/c mice were randomly assigned to two groups (10 mice per group) and injected with Scr-shRNA and KIF3A-shRNA MDA-MB-231 cells (2 × 10⁶) via the tail vein, respectively. The mice were killed 6 weeks later. The lungs of the mice were excised for H&E staining. Lung metastasis was quantified by counting the number of tumor foci in 10 randomly selected high-power fields under a microscope. Animal experiments were approved by the Animal Ethics Committee of Qingdao University, China.

2.10 Statistical analysis

All statistical analyses were performed using SPSS 23.0 software. All values were presented as mean ± SD. Wilcoxon’s test was used for non–normal distributed data. The Student’s t test was used for data that were normally distributed. Differences were considered statistically significant at P < 0.05 and P < 0.01.

3.1 KIF3A mRNA and protein expression

We detected the KIF3A mRNA and protein levels in nine paired TNBC tissues and adjacent tissues by using western blot and real-time RT-PCR, respectively. The results showed that both mRNA (Figure 1A,B, P < 0.01) and protein levels (Figure 1C,D, P < 0.01) were significantly higher in TNBC tissues compared with that in adjacent tissues, suggesting that KIF3A was overexpressed in the TNBC tissues.

3.2 Immunohistochemistry assay

Immunohistochemistry staining showed that 70 out of 98 TNBC cases (71.4%) expressed higher levels of KIF3A (Figure 2A,B) than the adjacent tissues (Figure 2C,D). Strong positive staining is observed in the cytoplasm and membrane of tumor cells, and weak staining is observed in the normal duct epithelial cells (5.653 ± 0.719 vs 4.143 ± 0.974, Table 1, P < 0.001, Wilcoxon’s test). We also found that 25 of 32 cases (78.1%) showed stronger KIF3A expression in the metastatic cancer cells in the lymph node (Figure 2E) than in the primary cancer tissues (Figure 2F) (7.031 ± 0.740 vs 6.031 ± 0.647, Table 1, P < 0.001). Higher KIF3A expression was also observed in the primary tumors with lymph node metastasis than those without lymph node metastasis (Table 1, P < 0.01). In addition, patients with recurrence of carcinoma had higher KIF3A expression than those without recurrence (Table 1, P < 0.05). There were no significant differences between groups for age, tumor size or grade (Table 1).

3.3 Effect of KIF3A in different triple negative breast cancer cell lines

Western blot analyses of the KIF3A expression in MDA-MB-231, MDA-MB-468, BT549 and BT20 TNBC cell lines are shown in Figure 3A. Due to the higher expression of KIF3A, MDA-MB-231 and BT549 cells were chosen for silencing KIF3A gene expression by using KIF3A shRNA1#, 2# and 3#. The KIF3A-shRNA 3# and 2# were more effective (Figure 3B, Figure S1) and were used for stably expressing KIF3A-shRNA cell selection. Real-time RT-PCR (Figure 3C) and western blot analysis revealed that the expression of KIF3A mRNA and protein were obviously silenced. MDA-MB-468 cell line had the lowest expression of KIF3A protein (Figure 3A). Therefore, it was used to overexpress KIF3A (Figure 3D), and western blot analysis showed that the expression peak of KIF3A appeared at 48 h after KIF3A-pEX transfection.

3.4 Silencing of KIF3A suppresses tumor cell growth and induces G0/G1 phase arrest

To determine the effect of KIF3A depletion on cell proliferation, MTT and colony-formation assays were performed. The growth curves were plotted in 5 days according to the OD values. The results showed that the growth of the KIF3A-shRNA (3#) group was significantly slower than that of the Scr-shRNA group in both MDA-MB-231 and BT549 cells. The growth of the KIF3A-pEX group was significantly faster than that of the Vector group in MDA-MB-468 cells (Figure 4A, P < 0.01). As shown in Figure 4B, the colony numbers were 293.20 ± 20.93 versus 174.80 ± 46.26 in MDA-MB-231 cells (P < 0.01) and 251.20 ± 23.76 versus 142.20 ± 40.55 in BT549 cells (P < 0.01) before and after KIF3A knockdown. The colony numbers were 151.40 ± 68.58 versus 292.00 ± 75.59 in MDA-MB-468 cells (P < 0.05) before and after KIF3A overexpression. These results suggested that silencing of KIF3A inhibited cell proliferation.

To further validate the effect of KIF3A on the cell cycle, the flow cytometry technique was used. The cell phase distribution of Scr-shRNA and KIF3A-shRNA (3#) MDA-MB-231 cells was as follows: G1 phase: 50.36 ± 1.66% vs 55.94 ± 0.80% (P < 0.01); S phase: 30.53 ± 2.39% vs 22.81 ± 0.77% (P < 0.01); G2/M phase: 19.11 ± 2.38% vs 21.26 ± 1.15%. The cell phase distribution of Scr-shRNA and KIF3A-shRNA (3#) BT549 cells was as follows: G1 phase: 48.66 ± 0.79% vs 53.42 ± 1.08% (P < 0.01); S phase: 26.86 ± 1.53% vs 20.72 ± 1.12% (P < 0.01); G2/M phase: 24.48 ± 1.22% vs 25.86 ± 1.97%. Compared with Scr-shRNA group cells, the proportion of cells in the G1 phase was significantly increased with a decrease in the proportion of cells in the S phase in KIF3A-shRNA (3#) group cells (Figure 4C), suggesting induction of G0/G1 arrest. To confirm the above findings, cell cycle analysis was also performed in KIF3A overexpression MDA-MB-468 cells.

3.5 Silencing of KIF3A inhibits the G1/S transition by repressing the Rb-E2F signaling pathway

To further explore the mechanism the effect of KIF3A on the cell proliferation and cell cycle, the expression of phospho-Rb, E2F1, Cyclin D1, CyclinE1 and P21 was examined by western blot (Figure 4D). Phospho-Rb, E2F1, CyclinD1 and CyclinE1 were decreased and P21 was increased in the KIF3A-shRNA (3#) group of MDA-MB-231 and BT549 cells. Phospho-Rb, E2F1, CyclinD1, CyclinE1 were increased and P21 was decreased in the KIF3A-pEX group of MDA-MB-468 cells. These results further suggested that the G1/S transition was suppressed after silencing of KIF3A by repressing the Rb-E2F signaling pathway.

3.6 Silencing of KIF3A inhibits tumor growth in nude mice

To further explore the tumor-forming capacity of KIF3A in vivo, a human TNBC xenograft nude mice model was established. We found that the tumor size (842.67 ± 52.08 mm³ vs 1344.00 ± 91.02 mm³, P < 0.05, Figure 5A,B) and weight (206.80 ± 20.70 mg vs 418.90 ± 34.55 mg, P < 0.01, Figure 5C) in KIF3A-shRNA (3#) cell xenografts were significantly decreased compared with the Scr-shRNA MDA-MB-231 cell xenografts. Subsequently, the expressions of KIF3A, phospho-Rb, E2F1, Cyclin E1, Cyclin D1, and P21 in xenografts were confirmed by IHC staining. The data showed an upregulation of P21 while downregulation of phospho-Rb, E2F1, CyclinE1, and CyclinD1 in KIF3A-shRNA (3#) cell xenografts (Figure 5D), which were consistent with the results in TNBC cell lines. It is obvious that suppression of KIF3A in MDA-MB-231 cells could inhibit tumor formation and growth.

3.7 Silencing of KIF3A inhibits cell migration and invasion by inhibiting epithelial-mesenchymal transition in triple negative breast cancer cell lines

The metastatic potential of a tumor is dependent on the ability of tumor cells to migrate to distant sites. Transwell assays were performed in vitro to investigate the effects of KIF3A on the migration and invasion of TNBC cell lines. The result of the transwell assay showed that the numbers of migrated MDA-MB-231 cells were 222.80 ± 14.96 and 63.60 ± 8.14 in Scr-shRNA and KIF3A-shRNA (3#). The numbers of migrated BT549 cells were 313.00 ± 13.44 and 178.80 ± 10.23 in Scr-shRNA and KIF3A-shRNA (3#). These data suggested that cell migration was inhibited by KIF3A knockdown in MDA-MB-231 and BT549 cells (Figure 6A, P < 0.01). Furthermore, consistent with migration results, it was confirmed via the invasion assay results that the cell’s invasive ability was progressively suppressed by KIF3A depletion (Figure 6B, P < 0.01).

Similar to the results of the transwell assays, the results of scratch tests showed that migration distance was significantly reduced in the KIF3A-shRNA (3#) group of MDA-MB-231 cells (57.57 ± 17.84 µm vs 107.20 ± 18.92 µm, P < 0.01, Figure 6C) and BT549 cells (136.30 ± 24.83 µm vs 199.50 ± 22.29 µm, P < 0.01, Figure 6C) compared to Scr-shRNA groups.

To confirm the above findings, migration and invasion assays were also performed in KIF3A overexpression MDA-MB-468 cells and the results showed that the number of migrated MDA-MB-468 cells (364.40 ± 16.04 vs 218.40 ± 19.76, P < 0.01, Figure 6D) and the number of invasive MDA-MB-468 cells (359.40 ± 21.73 vs 182.40 ± 47.19, P < 0.01, Figure 6D) was increased in the KIF3A-pEX group compared to the Vector group. The KIF3A-pEX group showed significantly increased migration distance compared to the Vector group (170.00 ± 18.57 µm vs 81.73 ± 22.84 µm, P < 0.01, Figure 6E).

We also found that depletion of KIF3A led to morphological change of the TNBC cells. MDA-MB-231 and BT549 cells became shortened and more adherent to each other due to silencing of KIF3A (Figure 6F). In contrast, overexpression of KIF3A in MDA-MB-468 cells resulted to elongated morphological change and mesenchymal-like properties (Figure 6F).

The EMT is a key step in cancer metastasis and invasion.40, 41 ZEB1, a key transcription factor, has been identified as mediating the initiation and execution of EMT. In addition, MMPs, a family of more than 28 enzymes, are thought to play a crucial role in tumor metastasis on the basis of their ability to degrade the extracellular matrix (ECM) and have been found to be upregulated in nearly all of the tumor types.42 MMP-2 and MMP-9, the two major members of the MMPs family, have been reported to be significant factors in breast cancer invasion and metastasis.43, 44

To further investigate the effect of KIF3A in EMT, E-cadherin, vimentin (two hallmarks of EMT), ZEB1, MMP-2, MMP-9, and KIF3A were detected by western blot (Figure 6G). The expression level of E-cadherin was increased, whereas the expression level of ZEB1, vimentin, MMP-2, and MMP-9 was decreased in KIF3A-shRNA (3#) compared to Scr-shRNA group cells in MDA-MB-231 and BT549 cells. This correlation was further confirmed by KIF3A overexpression in MDA-MB-468 cells, where decreased E-cadherin expression and increased ZEB1, vimentin, MMP-2, and MMP-9 expressions were observed. These results indicated that KIF3A depletion might suppress cell migration and invasion by inhibiting EMT in TNBC cells. In addition, ZEB1 is required for KIF3A-mediated EMT.

3.8 KIF3A depletion inhibits metastasis in vivo

Nude mice were injected with KIF3A-shRNA (3#) MDA-MB-231 cells via the tail vein to confirm the effect of KIF3A knockdown on tumor migration and invasion in vivo. Body weight of the mice did not differ (Figure 7A) after 6 weeks of injection. However, compared with the control group, the lungs from the KIF3A-shRNA (3#) group were significantly lighter (Figure 7B). Only 5 of 10 mice from the KIF3A-shRNA (3#) group developed lung metastasis, while all 10 mice presented with lung metastasis in the control group. We also found that the number of tumor foci in the control group was much more than that in the KIF3A-shRNA (3#) group (Figure 7C-E). Furthermore, the expression of E-cadherin was increased and expressions of vimentin, ZEB1, MMP-2 and MMP-9 were reduced in KIF3A-shRNA (3#) cell xenografts compared with the control group based on IHC staining (Figure 7F), which were in accord with the results in TNBC cell lines. These data showed that KIF3A knockdown significantly inhibits TNBC metastasis in vivo and in vitro.

3.9 KIF3A overexpression correlates with chemoresistance

To evaluate whether KIF3A expression modulates chemoresistance, the cell viability assay showed that silencing of KIF3A evidently decreased the IC50 for doxorubicin in MDA-MB-231 cells (1.59 ± 0.46 μg/mL vs 2.51 ± 0.65 μg/mL, P < 0.05, Figure 8A), whereas silencing of KIF3A had little effect on cisplatin resistance in MDA-MB-231 cells (14.58 ± 5.86 μg/mL vs 14.98 ± 3.20 μg/mL, P > 0.05, Figure 8B). We also found that KIF3A overexpression evidently increased the IC50 for doxorubicin in MDA-MB-468 cells (0.83 ± 0.03 μg/mL vs 0.64 ± 0.03 μg/mL, P < 0.05, Figure 8C), whereas KIF3A overexpression had little impact on cisplatin resistance (1.74 ± 0.37 μg/mL vs 1.72 ± 0.15 μg/mL, P > 0.05, Figure 8D). However, KIF3A knockdown had little impact on doxorubicin and cisplatin resistance in BT549 cells (0.84 ± 0.19 μg/mL vs 1.04 ± 0.07 μg/mL, 2.61 ± 0.43 μg/mL vs 2.34 ± 0.27 μg/mL, P > 0.05, Figure 8E,F). These results suggested that knockdown of KIF3A might restore chemosensitivity of doxorubicin in TNBC but not to cisplatin.