Aurora Kinase A and Bcl-xL Inhibition Suppresses Metastasis in Triple-Negative Breast Cancer

Triple-negative breast cancer (TNBC) is a heterogeneous disease that accounts for 10–15% of all breast cancer cases. Within TNBC, the treatment of basal B is the most challenging due to its highly invasive potential, and thus treatments to suppress metastasis formation in this subgroup are urgently needed. However, the mechanisms underlying the metastatic ability of TNBC remain unclear. In the present study, we investigated the role of Aurora A and Bcl-xL in regulating basal B cell invasion. We found gene amplification and elevated protein expression in the basal B cells, which also showed increased invasiveness in vitro, compared to basal A cells. Chemical inhibition of Aurora A with alisertib and siRNA-mediated knockdown of BCL2L1 decreased the number of invading cells compared to non-treated cells in basal B cell lines. The analysis of the correlation between AURKA and BCL2L1 expression in TNBC and patient survival revealed significantly decreased relapse-free survival (n = 534, p = 0.012) and distant metastasis-free survival (n = 424, p = 0.017) in patients with primary tumors exhibiting a high combined expression of AURKA and BCL2L1. Together, our findings suggest that high levels of Aurora A and Bcl-xL promote metastasis, and inhibition of these proteins may suppress metastasis and improve patient survival in basal B TNBC.


Introduction
Breast cancer is the most common cancer in women worldwide, accounting for 31% of all cancer cases and is the second leading cause of cancer-related death at 15% of all breast cancer deaths [1]. Although prognosis for women diagnosed with localized disease is good with a 5-year survival rate of 87%, the 5-year survival rate for those diagnosed with distant metastasis is only 23.4% [2]. Of all breast cancer subtypes, the triple-negative breast cancer (TNBC), which accounts for 10-15% of all breast cancer cases, is associated with worse prognosis compared to other subtypes due to its aggressive nature, higher metastatic rate and early recurrence [3,4]. Treatment of this breast cancer subgroup is very challenging due to the lack of available treatments and tumor heterogeneity [5]. Although chemotherapy has been shown to improve TNBC patient survival, it has limited efficacy after the tumor has metastasized [6]. Importantly, several targeted therapies have recently been approved for TNBC. PARP (poly adenosine diphosphate-ribose polymerase) inhibitors have been approved for BRCA1 and BRCA2 mutation carriers, which account for approximately 20% of TNBC patients [7]. Additionally, an antibody conjugate consisting of an anti-Trop2 antibody conjugated with SN-38 (Sacituzumab govitecan), the active metabolite of irinotecan, was recently approved by the FDA for metastatic TNBC based on promising phase III study results [8,9]. Finally, the addition of the anti-PDL1 pembrolizumab to neoadjuvant chemotherapy has shown to significantly improve event-free survival for high-risk TNBC [10].

TNBC Cells with the Basal B-Like Phenotype Exhibit Gene Amplification and Overexpression of Aurora A and Bcl-xL
Although Aurora A and Bcl-xL have both been linked to cancer formation and metastasis in breast cancer [19,32], their specific role in the ability of basal A and B cells to invade and metastasize remains unclear. To examine the correlation between Aurora A and Bcl-xL expression and metastatic abilities in basal A and basal B TNBC, we first confirmed the epithelial or mesenchymal features of MDA-MB-468, CAL-120 and MDA-MB-231 cells ( Figure 1a) [33] and evaluated their invasion ability in vitro ( Figure 1b). As expected, immunocytochemical analysis using the mesenchymal marker, vimentin and the epithelial marker, EPCAM showed: low vimentin and high EPCAM in MDA-MB-468 cells, consistent with epithelial basal A phenotype; and high-vimentin and low-EPCAM expression in CAL-120 and MDA-MB-231 cells, consistent with a mesenchymal basal B phenotype (Figure 1a). An invasion assay showed that CAL-120 and MDA-MB-231 cells exhibited higher invasive properties than MDA-MB-468 cell lines (Figure 1b). Although the higher invasion ability of CAL-120 cells compared to MDA-MB-468 cells did not reach statistical significance, the differences observed between these two cell lines in vimentin and EPCAM expression support the mesenchymal phenotype of CAL-120 cells. These findings support the classification of MDA-MB-468 cells as basal A subtype and CAL-120 and MDA-MB-231 cells as basal B subtype, and that basal B exhibits higher invasive properties.

Overexpression of Aurora A and Bcl-xL
Although Aurora A and Bcl-xL have both been linked to cancer formation and metastasis in breast cancer [19,32], their specific role in the ability of basal A and B cells to invade and metastasize remains unclear. To examine the correlation between Aurora A and Bcl-xL expression and metastatic abilities in basal A and basal B TNBC, we first confirmed the epithelial or mesenchymal features of MDA-MB-468, CAL-120 and MDA-MB-231 cells (Figure 1a) [33] and evaluated their invasion ability in vitro ( Figure 1b). As expected, immunocytochemical analysis using the mesenchymal marker, vimentin and the epithelial marker, EPCAM showed: low vimentin and high EPCAM in MDA-MB-468 cells, consistent with epithelial basal A phenotype; and high-vimentin and low-EPCAM expression in CAL-120 and MDA-MB-231 cells, consistent with a mesenchymal basal B phenotype (Figure 1a). An invasion assay showed that CAL-120 and MDA-MB-231 cells exhibited higher invasive properties than MDA-MB-468 cell lines (Figure 1b). Although the higher invasion ability of CAL-120 cells compared to MDA-MB-468 cells did not reach statistical significance, the differences observed between these two cell lines in vimentin and EPCAM expression support the mesenchymal phenotype of CAL-120 cells. These findings support the classification of MDA-MB-468 cells as basal A subtype and CAL-120 and MDA-MB-231 cells as basal B subtype, and that basal B exhibits higher invasive properties. Comparable results were obtained with immunocytochemical analysis of these proteins in the three cell lines (Figure 2c). Together, these findings show that TNBC cell lines with the basal B phenotype (basal B TNBC) are more invasive and exhibit higher levels of both Aurora A and Bcl-xL.
A and Bcl-xL in the basal B cell lines than in the basal A cells (Figure 2b), with CAL-120 exhibiting the highest Aurora A expression, while MDA-MB-231 showed the highest Bcl-xL expression. Comparable results were obtained with immunocytochemical analysis of these proteins in the three cell lines (Figure 2c). Together, these findings show that TNBC cell lines with the basal B phenotype (basal B TNBC) are more invasive and exhibit higher levels of both Aurora A and Bcl-xL.

Inhibition of Aurora A and BCL2L1 Reduces Invasion of Basal B TNBC Cell Lines
To further investigate the role of Aurora A in the invasion capability of basal B TNBC cells, we tested a specific Aurora A inhibitor, alisertib and evaluated its effect on cell viability and invasion ( Figure 3). Alisertib IC50 value for CAL120 and MDA-MB-231 cells was determined as 10 and 19.33 µM, respectively, and 20 µM alisertib was used for further experiments in both cell lines ( Figure 3a). Furthermore, alisertib induced an increased Aurora A and Bcl-xL expression and decreased vimentin compared to untreated CAL-120 and MDA-MB-231 cells as evaluated by Western blotting (Figure 3b). Importantly, treatment with alisertib inhibited CAL-120 and MDA-MB-231 cell invasiveness (Figure 3c), consistent with reduced levels of vimentin observed upon treatment (Figure 3b). Although alisertib reduced invasion in MDA-MB-468 cells as well, vimentin levels in this cell line were significantly lower (Figure 3b) which supports the lower invasiveness and aggressiveness of these cells. Together, these findings suggest an important role for Aurora A in regulating invasion in highly aggressive basal B TNBC cells.

Inhibition of Aurora A and BCL2L1 Reduces Invasion of Basal B TNBC Cell Lines
To further investigate the role of Aurora A in the invasion capability of basal B cells, we tested a specific Aurora A inhibitor, alisertib and evaluated its effect on c bility and invasion ( Figure 3). Alisertib IC50 value for CAL120 and MDA-MB-2 was determined as 10 and 19.33 µM, respectively, and 20 µM alisertib was used for experiments in both cell lines ( Figure 3a). Furthermore, alisertib induced an increas rora A and Bcl-xL expression and decreased vimentin compared to untreated C and MDA-MB-231 cells as evaluated by Western blotting (Figure 3b). Importantly ment with alisertib inhibited CAL-120 and MDA-MB-231 cell invasiveness (  As Bcl-xL expression was found to be significantly higher in the more invasiv B cells compared to the less invasive basal A cells (Figures 2b,c and 3b), we furthe tigated the role of BCL2L1 in the invasion capability of basal B TNBC cells. To this e performed siRNA-mediated transient knockdown of BCL2L1 and evaluated its e cell growth and invasion ( Figure 4). Bcl-xL expression was significantly reduced in both cell lines treated with BCL2L1-siRNA compared to cells treated with scr siRNA, as assessed by Western blotting (Figure 4a). We observed decreased cell of CAL-120 and MDA-MB-231 cells 72 h following BCL2L1-siRNA transfection com to cells transfected with the control siRNA, as evaluated by crystal violet assay 4b). Notably, BCL2L1 siRNA-mediated knockdown caused a marked reduction in vasiveness of CAL-120 and MDA-MB-231 cells (Figure 4c). Together, these findin port a role for BCL2L1 in controlling basal B TNBC cells invasion. A schematic pr tion of the role of Aurora A and Bcl-xL in regulating the invasion of basal B cells is in Figure 5. As Bcl-xL expression was found to be significantly higher in the more invasive basal B cells compared to the less invasive basal A cells (Figures 2b,c and 3b), we further investigated the role of BCL2L1 in the invasion capability of basal B TNBC cells. To this end, we performed siRNA-mediated transient knockdown of BCL2L1 and evaluated its effect on cell growth and invasion (Figure 4). Bcl-xL expression was significantly reduced at 72 h in both cell lines treated with BCL2L1-siRNA compared to cells treated with scrambled siRNA, as assessed by Western blotting (Figure 4a). We observed decreased cell growth of CAL-120 and MDA-MB-231 cells 72 h following BCL2L1-siRNA transfection compared to cells transfected with the control siRNA, as evaluated by crystal violet assay (Figure 4b).
Notably, BCL2L1 siRNA-mediated knockdown caused a marked reduction in the invasiveness of CAL-120 and MDA-MB-231 cells (Figure 4c). Together, these findings support a role for BCL2L1 in controlling basal B TNBC cells invasion. A schematic presentation of the role of Aurora A and Bcl-xL in regulating the invasion of basal B cells is shown in Figure 5.

High AURKA and BCL2L1 Expression Correlates with Poor Clinical Outcome in TNBC
To investigate the prognostic value of Aurora A and Bcl-xL in cohorts of TNBC patients, we used the web-based tool Kaplan-Meier plotter to access the correlation between AURKA and BCL2L1 mRNA expression and clinical outcome. High BCL2L1 expression significantly correlated with shorter relapse-free survival (RFS, n = 534, p = 0.0019) and distant metastasisfree survival (DMFS, n = 424, p = 0.0055) in TNBC patients (Figure 6a). High AURKA expression was also associated with shorter RFS and DMFS in the same patient population, but this correlation was not statistically significant (Figure 6b, p = 0.017 and p = 0.13, respectively). Notably, mean expression of combined AURKA and BCL2L1 was calculated, and high AURKA-BCL2L1 mRNA expression showed a significant correlation with low RFS (p = 0.012) and DMFS (p = 0.017) (Figure 6c). Cox regression multivariate analysis including mean expression of combined AURKA and BCL2L1, MKi67 and ESR1 expression showed that combined AURKA and BCL2L1 and ESR1 expression are independent prognostic factors of RFS (HR 1.65, p = 0.0036; HR 1.42, p = 0.041, respectively) and DMFS (HR 1.74, p = 0.013; HR 1.49, p = 0.041, respectively; Table 1.) The KM plotter does not subdivide the basal-like subtypes into basal A and B; thus, we were not able to perform survival analysis of these markers separately in basal A and basal B TNBC. These data support a correlation between high AURKA and BCL2L1 and poor clinical outcome in TNBC patients.

Discussion
TNBC and the partly overlapping molecular subtype basal-like breast cancer represent an aggressive subtype of breast cancer that is associated with high metastatic ability, tumor heterogeneity and lack of effective therapies [4,34]. Although chemotherapy is initially effective, the majority of TNBC, particularly the basal B subtype, will metastasize and exhibit chemotherapy resistance, which limits the efficacy of further therapy and is associated with poor clinical outcomes [35][36][37]. Investigating the mechanisms of metastasis formation is essential for developing effective clinical interventions for this patient population [38]. Aurora A and Bcl-xL have both been linked to the metastatic process in different types of cancer, including breast cancer [19,32], but their specific impact on the ability of TNBC cells to invade and metastasize remains unclear.
Here, we showed gene amplification and protein upregulation of Aurora A and Bcl-xL in basal B TNBC cell lines compared to basal A cells, which correlated with a mesenchymal phenotype and higher invasive capabilities of basal B cells. Other studies have found upregulation of Aurora A and Bcl-xL in breast cancer, including basal-like breast cancer and an association with breast cancer progression through the activation of EMT and tumor stemness [19,20]. However, concomitant high expression of these two molecules has not yet been shown in basal B TNBC. Identifying the vulnerabilities of TNBC is particularly challenging due to its high heterogeneity and lack of major cancer drivers. Thus, identification of the co-expression of Bcl-xL and Aurora A opens the possibility for cotargeting therapeutic strategies in TNBC. Indeed, targeting the anti-apoptotic protein Bcl-xL together with anti-mitotic agents has recently been proposed as an efficient therapeutic strategy for different cancers, including TNBC [39]. In a different study, it was shown that TNBC cells expressing Bcl-xL are sensitive to combined inhibition with a drug targeting Bcl-xL and a drug targeting CDK1/2/4 but not when the Bcl-xL inhibitor was combined with drugs inhibiting FOXM1, CDK4/6, Aurora A or Aurora B [40]. In contrast, our data show that basal B TNBC cell lines co-expressing Aurora A and Bcl-xL are sensitive to Aurora A inhibition with alisertib. Furthermore, we found that treatment with alisertib induces increased expression of Bcl-xL in basal B TNBC cells, further supporting co-targeting both molecules. In line with our findings, it was recently shown that the combined inhibition of Aurora B and Bcl-xL caused synergistic cell growth impairment in cancer cells [41,42].
The Aurora A inhibitor alisertib has been shown to hinder the phosphorylation of Aurora A and induce apoptosis by altering the expression of the BCL-2 family proteins to a pro-apoptotic state through increased expression of pro-apoptotic proteins and decreased expression of anti-apoptotic proteins, such as Bcl-xL [25]. In our study, we observed an increased Aurora A expression after treatment with alisertib in basal B TNBC cells, which can be explained by the accumulation of non-phosphorylated Aurora A. Furthermore, we showed that the reduced metabolic activity induced by alisertib in basal B cells is not due to the induction of apoptosis, as Bcl-xL expression is increased, likely due to a compensation mechanism in response to the growth inhibition caused by the Aurora A inhibitor. Instead, the reduced cell viability can be a result of the cell cycle arrest caused by Aurora A inactivation. Importantly, inhibition of Aurora A and knockdown of BCL2L1 resulted in reduced in vitro invasion of both basal B cells compared to non-treated cells, suggesting that these two targets are important in the metastatic process in mesenchymallike TNBC.
To further investigate the implications of overexpression of Aurora A and Bcl-xL in TNBC breast cancer patients, we assessed the correlation between AURKA and BCL2L1 mRNA expression and clinical outcomes using the web-based tool KM plotter. We found that high-BCL2L1-AURKA correlated with shorter RFS and DMFS. High levels of Bcl-xL or Aurora A have been previously correlated with poor overall survival (OS) and short progression-free survival (PFS) in breast cancer, including TNBC [40,[43][44][45][46], but a correlation between co-expression of AURKA and BCL2L1 and TNBC patient survival has not yet been reported. Of the studies that evaluated these markers and clinical outcomes in TNBC, one study found a correlation between AURKA and OS and PFS in a small population of TNBC patients (n = 122) [44]. Another study showed an association between BCL2L1 levels and OS in a larger TNBC patient population (n = 580), but only one microarray dataset was used, and RFS and DMFS were not evaluated [40]. Furthermore, this study showed that high levels of AURKA either alone or together with BCL2L1 levels were not associated with altered patient survival [40]. Our study is the first, to our knowledge, to show that high co-expression of AURKA and BCL2L1 correlates with poor RFS and DMFS in a large population of TNBC patients from different datasets.
In summary, our evaluation of the association between Aurora A and Bcl-xL copy numbers and expression and epithelial/mesenchymal phenotypes in TNBC shows that the co-expression of Aurora A and Bcl-xL is associated with the invasion capability of TNBC cells in the basal B subtype, and inhibition of both molecules is required to suppress tumor metastasis. Finally, we suggest that combined AURKA and BCL2L1 expression is a prognostic factor in TNBC as tumors exhibiting high expression of both markers correlate with poor outcomes.

Cell Line Culture
The TNBC cell lines MDA-MB-468 and MDA-MB-231 were obtained from ATCC, and CAL-120 was obtained from Leibniz Institute DSMZ-German Collection of Microorganisms and Cell Cultures. All cell lines were grown in Dulbecco's Modified Eagle Medium (DMEM, Sigma, Darmstadt, Germany) supplemented with 10% fetal bovine serum (FBS, Gibco, ThermoFisher Scientific, Waltham, MA, USA) and 1% penicillin/streptomycin (P/S, Gibco, ThermoFisher Scientific, Waltham, MA, USA). All cell lines underwent DNA authentication using Cell ID™ System (Promega, Walldorf, Germany) and mycoplasma testing (Lonza, Basel, Switzerland) before use in the described experiments.

DNA Extraction and CNV Assay
DNA was extracted using the DNeasy Blood and Tissue kit (Qiagen, Hilden, Germany) according to manufacturer instructions. The DNA was diluted in 1× tris-EDTA (TE) buffer to a concentration of 5 ng/L before running TaqMan Copy Number Assay PCR (Applied Biosystems, ThermoFisher Scientific, Waltham, MA, USA) on a StepOnePlus Real-Time PCR system v2.3 (Applied Biosystems, ThermoFisher Scientific, Waltham, MA, USA) in four replicates using AURKA (Hs02938272_cn, Applied Biosystems, ThermoFisher Scientific, Waltham, MA, USA) and BCL2L1 (Hs07178628_cn, Applied Biosystems, ThermoFisher Scientific, Waltham, MA, USA) copy number assays. TaqMan Copy Number Reference Assay RNase P (4403326, ThermoFisher Scientific, Waltham, MA, USA) was used as a control. The threshold was set to 0.2 for all targets, and data were analyzed with CopyCaller Software v2.1 (Applied Biosystems, ThermoFisher Scientific, Waltham, MA, USA).

Cell Proliferation Assay
Transfected cells were seeded (10 4 cells/well) in 96-well plates and incubated for 24, 48 and 96 h at 37 • C in 5% CO 2 for evaluation of cell proliferation using crystal violet-based colorimetric assay [47]. The absorbance was measured in Sunrise™ 500 absorbance reader (Tecan, Männedorf, Switzerland).

Cell Invasion Assay
Cell invasion was measured using the cell invasion assay kit (ECM554, Sigma, Darmstadt, Germany) according to manufacturer instructions. The cells were starved in serumfree growth medium for 24 h before seeding 2.5 × 10 5 cells per insert in a 24-well plate. The plate was incubated for 72 h at 37 • C and 5% CO 2 humidity. For the inhibitor and siRNA-mediated knockdown experiments, cells were first treated with alisertib or BCL2L1-siRNA in medium containing serum for 48 h, after which the media was changed to serum-free growth medium with alisertib treatment. Next day, cells were seeded in inserts, and invasion was evaluated 24 h after seeding. Cell growth medium with serum was used as a chemoattractant. Cells without chemoattractant under the insert were used as a non-invasive control for the invasion assay. Cell invasion was measured with fluorescent light emission (480 nm/520 nm) using a Victor3™ 1420 counter (Perkin Elmer, Waltham, MA, USA).

Kaplan-Meier Plotting
The web tool Kaplan-Meier (KM) plotter (Budapest, Hungary) [48] was used to generate survival curves for TNBC patients depending on mRNA expression (gene chip) of AURKA and BCL2L1. All datasets available in KM plotter were included in the analysis. The inclusion criteria for sample selection were: ER and PR negative by IHC and HER2 negative by array, independently of clinical and pathological characteristics, such as grade, lymph node status or previous treatment. JetSet optimal probe was selected for each gene (probe Id 208,079 for AURKA and probe Id 212,312 for BCL2L1) and the best performing threshold was selected as the cut-off. Mean expression of both genes was used to evaluate the correlation of combined AURKA and BCL2L1 expression. Relapse-free survival (RFS) and distant metastasis-free survival (DMFS) were selected as endpoints.

Statistical Analysis
Statistical significance was calculated using one-way ANOVA, and p-values ≤ 0.05 were considered statistically significant. GraphPad Prism v.8 (GraphPad Software, Inc., San Diego, CA, USA) was used for statistical analysis.  Data Availability Statement: All datasets generated during the study will be made available upon reasonable request to the corresponding author, Henrik Ditzel, email address: hditzel@health.sdu.dk.