Triple-negative breast cancer (TNBC) is a highly heterogeneous and recurrent subtype of breast cancer that lacks an effective targeted therapy. To identify candidate therapeutic targets, we profiled global gene expression in TNBC and breast tumor-initiating cells with a patient survival dataset. Eight TNBC-related kinases were found to be overexpressed in TNBC cells with stem-like properties. Among them, expression of PKC-α, MET, and CDK6 correlated with poorer survival outcomes. In cases coexpressing two of these three kinases, survival rates were lower than in cases where only one of these kinases was expressed. In functional tests, two-drug combinations targeting these three kinases inhibited TNBC cell proliferation and tumorigenic potential in a cooperative manner. A combination of PKC-α-MET inhibitors also attenuated tumor growth in a cooperative manner in vivo. Our findings define three kinases critical for TNBC growth and offer a preclinical rationale for their candidacy as effective therapeutic targets in treating TNBC. Cancer Res; 74(17); 4822–35. ©2014 AACR.

Breast cancers are typically classified into several subtypes: Luminal A and B subtypes of breast cancer that correspond to pathologic estrogen receptor (ER)-positive tumors, the HER2 subtype that corresponds to HER2-overexpressing tumors, and triple-negative/basal-like breast cancer (TNBC/BLBC). BLBC makes up about 15% to 20% of breast cancers. Recent studies using clinical samples indicate that BLBC shares more than 80% similarity with TNBC, which is negative for ER, progesterone receptor (PR), and HER2 expression (1). In addition, two major subgroups of TNBC characterized on the basis of gene ontologies and differential gene expression profile have been reported: basal-like TNBC driven by genes enriched in cell cycle, cell division, and DNA damage response; and mesenchymal-like driven by genes involved in cell motility, cell differentiation, and growth factor pathways (2). These findings indicate that TNBCs are highly heterogeneous. Although subtyping allows for better prediction of the response of each subtype to specific molecular targets, the therapeutic benefits in clinical trial are still unclear. Here, we will use TNBC as a general term to broadly represent the TNBC/BLBC subtype.

Patients with TNBC initially respond to conventional chemotherapy, but the disease frequently relapses and leads to worse outcome than patients with hormone receptor-positive subtypes (3). The low survival rate of patients with TNBC is also due to high metastasis rates and lack of effective treatment after a relapse (4). Currently, no effective targeted therapies are available for patients with TNBC because of TNBC's lack of expression of hormone receptors and HER2 amplification (1, 5).

A cancer stem cell (CSC) or tumor-initiating cell (TIC) hypothesis has been proposed to account for treatment failure and recurrence in patients with TNBC (6). Breast TICs (BTIC) make up a small subpopulation of cells inside tumors that are resistant to conventional therapy and are capable of reinitiating tumor growth after treatment (7–9); these cells can be enriched by flow cytometry using specific cell-surface markers such as CD44+ and CD24−/low (10) and ALDH1high (11). Accumulating evidence suggests that BTICs are responsible for tumor initiation, progression, and drug resistance (12, 13). Residual breast tumor cells that survive after conventional treatment may contain a high proportion of cells with both tumor-initiating and mesenchymal features (7, 14). In line with these BTIC features, TNBC also exhibits stem cell signatures and epithelial–mesenchymal transition characteristics with high expression of mesenchymal markers (1, 15). We hypothesized that BTICs and disease recurrence may activate common pathways and contribute to the resistance of TNBC.

On the basis of The Cancer Genome Atlas (TCGA) analysis, the mutation rates of genes other than TP53 (80%) or PIK3CA (9%) are less than 5% in TNBC (16). However, there is no effective way to address loss/mutations of p53 in the clinic at this moment, and the effects of the PI3K pathway inhibitors in clinical trials are yet to be seen (17–20). miRNA-regulated gene expression has been implicated in cancer progression (21), suggesting that in addition to DNA alteration (mutation or amplification), miRNA regulation may contribute to TNBC progression by turning on certain oncogenic gene expression. To find other novel therapeutic targets, we performed an unbiased two-stage bioinformatics analysis (cell and patient based) of TNBC/BTIC (mRNA and miRNA) databases and of patient survival data to identify key oncogenic molecules in TNBC, with a particular focus on those targeted by drugs that are currently in oncology clinical trials. Through this analysis, we identified eight kinases linking TNBC and CSCs.

Drugs

Crizotinib was purchased from LC Laboratories (C-7900), safingol from Matreya LLC (1807) or Sigma (D4681), L-threo-dihydrosphingosine from Santa Cruz Biotechnology (CAS 73938-69-9), and PD-0332991 from Sigma (PZ0199).

Cell lines

MDA-MB-231, Hs578T, BT549, MDA-MB-468, T47D, MCF7, and SK-BR-3 cells were grown in DMEM-F12 medium (Caisson Laboratories, Inc.) supplemented with 10% FBS, penicillin (50 U/mL), and streptomycin (50 U/mL). Immortal normal mammary epithelial cells (MCF10A) were cultured in DMEM-F12 medium supplemented with 5% horse serum, penicillin (50 U/mL), streptomycin (50 U/mL), EGF (20 ng/mL), insulin (10 μg/mL), cholera toxin (1 ng/mL), and hydrocortisone. BT20 cells were grown in DMEM medium (Caisson Laboratories, Inc.) supplemented with 10% FBS, penicillin (50 U/mL), streptomycin (50 U/mL), and nonessential amino acids. Cell lines were validated by short tandem repeat (STR) DNA fingerprinting using the AmpFLSTR Identifiler Kit according to the manufacturer's instructions (Applied Biosystems catalog no. 4322288). The STR profiles were compared with known American Type Culture Collection fingerprints (ATCC.org) and with data in the Cell Line Integrated Molecular Authentication database version 0.1.200808 (22). The STR profiles matched known DNA fingerprints or were unique.

Analysis of public gene expression data

The following databases were used to compare TNBC data with non-TNBC data: E-MTAB-327 (miRNA expression profiling by array of NCI-60 human cancer cell lines), GSE25037 (miRNA expression profiling in CD44+/CD24−/low and non-CD44+/CD24−/low human mammary epithelial cell populations), Cancer Cell Line Encyclopedia (CCLE) data on breast cancer cell mRNA expression (23), and GSE7513 (human breast tumor sample data were sorted to select cells that were CD44+/CD24−/low). The log2 ratio of TNBC versus non-TNBC data from public microarray databases was determined using NetWalker software (24). Gene expression profiles, pathways, and cross-talk were analyzed with the Ingenuity Pathway Analysis System.

Hierarchical clustering

Clustering was analyzed with an integrated pair of programs, Cluster and TreeView (25), for analyzing and visualizing the results of complex microarray experiments, and the median adjusted values were used to create expression files. The hierarchical clustering algorithm used is based closely on the average-linkage method of Sokal and colleagues (26). For any set of TNBC-related kinases, an upper-diagonal similarity matrix was computed by using average-linkage clustering, which contained similar scores for all pairs of genes. This algorithm was determined by computing a dendrogram assembling all elements into a single tree, as described by Eisen and colleagues (25). The software application of this algorithm was obtained from M.B. Eisen Department of Molecular and Cell Biology, University of California Berkeley, Berkeley, CA (25).

Display

The heatmap was represented graphically by coloring each cell on the basis of the measured fluorescence ratio. Log ratios of 0 (a ratio of 1.0 indicates that the genes are unchanged) were colored in black, positive log ratios were colored in red, and negative log ratios were colored in green (with darker colors corresponding to higher ratios).

qRT-PCR

Total RNAs from different breast cancer cells were extracted by using the RNeasy Kit (Qiagen). RNAs were reverse transcribed by using the Superscript II Kit (Invitrogen). RT-PCR was then performed using the iCycler (Bio-Rad). RNA levels were quantitated and normalized to GAPDH by subtracting the cycling threshold for the control from the cycling threshold for the target. Primers were as follows:

  • LYN: 5′-TTCTGGTCTCCGAGTCACTCA-3′ and 5′-GCCGTCCACTTAATAGGGAACT-3′

  • MAP4K4: 5′-CCAATGGCAACTCCGAGTCTGT-3′ and 5′- GGGTCACTGAAGGAATGGGATC-3′

  • FYN: 5′-ACAAAACTGACGGAGGAGAGG-3′ and 5′-GAAGCTGGGGTAGTGCTGAG-3′

  • MET: 5′-TGTGTGGTCCTTTGGCGTGCTC-3′ and 5′-GGCGCATTTCGGCTTTAGGGTG-3′

  • RPS6KA3: 5′-CGAGGTCATACTCAGAGTGCTG-3′ and 5′-ACTGTGGCATTCCAAGTTTGGCT-3′

  • PRKCA: 5′-GCCTATGGCGTCCTGTTGTATG-3′ and 5′-GAAACAGCCTCCTTGGACAAGG-3′

  • CDK6: 5′-GGATAAAGTTCCAGAGCCTGGAG-3′ and 5′-GCGATGCACTACTCGGTGTGAA-3′

  • PIK3CD: 5′-TGCCAAACCACCTCCCATTCCT-3′ and 5′-CATCTCGTTGCCGTGGAAAAGC-3′.

Western blotting

Western blotting was carried out as described previously (27) with anti-LYN (Santa Cruz Biotechnology), anti-MAP4K4 (GeneTex), anti-FYN (Cell Signaling Technology), anti-MET (Cell Signaling Technology), anti-RPS6KA3 (Cell Signaling Technology), anti-PRKCA (BD Biosciences), anti-CDK6 (Abcam), and anti-PIK3CD (GeneTex) antibodies. The membranes were stripped and reprobed with an anti-α-tubulin mouse monoclonal antibody (Sigma) as a loading control.

Patient overall survival analysis

The NKI-295 database (28) was used for analysis of TNBC-related gene expression and overall survival. A gene expression level within the first quartile was defined as positive. Survival curves were estimated by the Kaplan–Meier method, and survival rates in different groups were compared by the log-rank test. The data were analyzed using statistical software SPSS version 17.0. A P value ≤ 0.05 was defined as statistically significant.

Cell viability assay

Cells were seeded in 96-well plates at an initial density of 2 × 103 per well. At each time point, cells were stained with 100 μL of sterile MTT (0.5 mg/mL; Sigma) for 2 hours at 37°C, followed by removal of the culture medium and addition of 100 μL of dimethyl sulfoxide. Absorbance was measured at 570 nm, using 655 nm as the reference wavelength. All experiments were carried out in triplicate.

Anchorage-independent colony growth

Two thousand cells were trypsinized and suspended in 2 mL of complete medium plus 0.3% agar (Sigma). The cell/agar mixture was plated on top of a layer of 1% agar in complete medium. After 21 days, viable colonies that were larger than 0.1 or 0.05 mm in diameter were counted using ImageJ software. Experiments were carried out for each cell line in triplicate.

Colony formation assay

Cells were plated in 6-well plates (2–5 × 102 cells/well) and cultured for 10 days. Colonies were stained with 1% crystal violet for 10 minutes after fixation with 4% paraformaldehyde for 5 minutes.

Dual-drug combination assay

Breast cancer cells were plated in 96-well plates (BD Biosciences) and treated with various concentrations of kinase inhibitors, either alone or in combination, for 48 hours. Cell viability was determined using the MTT assay. Synergistic effects were determined by using the Chou–Talalay method to calculate the combination index (CI; ref. 29).

Flow-cytometric analysis

Cells were harvested, washed with cold PBS, and processed for CD24 and CD44 surface markers. Allophycocyanin-conjugated anti-CD44 (BD Biosciences) and phycoerythrin-conjugated anti-CD24 (BD Biosciences) antibodies were used. To measure the levels of CD44+/CD24−/low cells, MDA-MB-231, MCF7, and BT20 cells stained with fluorescein isothiocyanate-conjugated anti-CD44 antibody and phycoerythrin-conjugated anti-CD24 antibody (both at 1:50) were sorted at the Flow Cytometry and Cellular Imaging Core Facility of The University of Texas MD Anderson Cancer Center (Houston, TX).

Mouse studies

All animal procedures were conducted under the guidelines approved by the Institutional Animal Care and Use Committee at MD Anderson Cancer Center. Athymic mice (Harlan Laboratories) were used as hosts for tumor xenografts. MDA-MB-231 cells with luciferase and GFP expression were used for tumor injection. Mice were divided according to the mean tumor volume in each group when the diameter of tumor size reached 4 mm. On the basis of our experience with in vitro tumorigenic analyses (Fig. 6A and B), we observed synergism when at least one single kinase inhibitor reached 50% of cell killing effect. To find synergistic effect in vivo, the doses of single-drug treatments were also chosen at 50% inhibition. The ED50 (50% effective dose) of crizotinib and safingol was estimated by a dose-titration experiment (data not shown) and comparable with previous reports in other cancer cell lines (28, 29). Crizotinib was prepared by dissolving the drug in 50 mmol/L sodium acetate buffer (pH 4.6; ref. 30) and administered orally at 8 mg/kg every other day (31). L-threo-dihydrosphingosine was prepared by dissolving the drug in 5% dextrose containing lactic acid, with pH adjusted to 4.4 (32), and was injected intravenously at 20 mg/kg twice a week (33). Tumors were measured twice weekly with a caliper, and tumor volume was calculated as (width2 × length)/2. Tumor volume measured during drug treatment was normalized to the volume before treatment. The CI was calculated as survival rate(drugs A + B)/(survival rate(drug A) × survival rate(drug B)).

Patient tissue samples

The TNBC specimens (n = 107) used for our IHC and survival analyses were originally obtained from patients undergoing surgical resection of breast cancer as primary treatment at MD Anderson Cancer Center or Mackay Memorial Hospital (Taipei, Taiwan) between 1995 and 2009. The specimens were used in accordance with the protocols approved by the Institutional Review Board of MD Anderson Cancer Center. Written informed consent was obtained from patients in all cases at the time of resection.

Eight TNBC-related kinases are commonly overexpressed in TNBC subtypes and BTIC populations

To identify cancer-related kinases that are overexpressed in TNBC and BTICs, we performed a two-stage bioinformatics analysis of both cell line data and patient data. The study design is outlined in Fig. 1A. The first stage was a cell-based analysis of data from two mRNA databases (14, 23) and two miRNA databases (34, 35) to identify oncogenes involved in TNBC. Breast cancer gene expression profile data were from the CCLE (23) and NCI-60 datasets (34), and BTIC gene expression profile data were from isolated human breast tumors (14), a mammary epithelial cell line, and primary human mammalian epithelial cells (35), representing the heterogeneity of TNBC. We focused on cancer-related kinases, as they are ideal drug targets, and by filtering miRNA targets with the Ingenuity Pathway Analysis System, we identified 23 kinases that were predicted to be upregulated in BTICs and TNBC (Table 1). We then validated the mRNA expression patterns of these 23 kinases by hierarchical clustering of expression levels in 56 breast cancer cell lines (CCLE) and identified thirteen kinases that had higher expression in TNBC cell lines (n = 26) than in non-TNBC cell lines (n = 30; Fig. 1B).

Figure 1.

Thirteen druggable targets for aggressive TNBC were identified by bioinformatics analysis of TNBC and BTIC databases. A, schematic representation of the TNBC bioinformatics analysis. B, breast cancer cell lines were grouped by mRNA subtype according to ERBB2, ESR1, and PGR mRNA expression levels into non-TNBC (n = 30) and TNBC (n = 26) cells. The graph on the right was generated from original and log2-transformed mRNA expression levels of TNBC-related kinases in breast cancer subtypes. Genes highly expressed in TNBC cells (ERBB2, ESR1, and PGR) are bold faced. The heatmap represents color-coded expression levels of differentially expressed genes in all breast cancer cell lines tested. The color scale ranges from saturated green for log ratios of −2.0 and below to saturated red for log ratios of 2.0 and above. C, twenty TNBC cell lines were subtyped according to previously published criteria (2), and six luminal breast cancer cell lines were selected as controls. The color scale ranges from saturated green for log ratios of −2.0 and below to saturated red for log ratios of 2.0 and above. The average expression levels of the 13 TNBC-related kinases are shown in the top row.

Figure 1.

Thirteen druggable targets for aggressive TNBC were identified by bioinformatics analysis of TNBC and BTIC databases. A, schematic representation of the TNBC bioinformatics analysis. B, breast cancer cell lines were grouped by mRNA subtype according to ERBB2, ESR1, and PGR mRNA expression levels into non-TNBC (n = 30) and TNBC (n = 26) cells. The graph on the right was generated from original and log2-transformed mRNA expression levels of TNBC-related kinases in breast cancer subtypes. Genes highly expressed in TNBC cells (ERBB2, ESR1, and PGR) are bold faced. The heatmap represents color-coded expression levels of differentially expressed genes in all breast cancer cell lines tested. The color scale ranges from saturated green for log ratios of −2.0 and below to saturated red for log ratios of 2.0 and above. C, twenty TNBC cell lines were subtyped according to previously published criteria (2), and six luminal breast cancer cell lines were selected as controls. The color scale ranges from saturated green for log ratios of −2.0 and below to saturated red for log ratios of 2.0 and above. The average expression levels of the 13 TNBC-related kinases are shown in the top row.

Close modal
Table 1.

Twenty-three kinases identified in the cell-based analysis

SymbolEntrez gene nameDrugs
CDK6 Cyclin-dependent kinase 6 PD-0332991, flavopiridol 
DGKA Diacylglycerol kinase, α80 kDa  
EPHA4 EPH receptor A4  
FLT1 fms-related tyrosine kinase 1 (VEGF/vascular permeability factor receptor) Sunitinib, pazopanib, axitinib, CEP 7055, vandetanib 
FYN FYN oncogene related to SRC, FGR, YES Dasatinib 
HIPK2 Homeodomain interacting protein kinase 2  
ITK IL2-inducible T-cell kinase  
JAK1 Janus kinase 1 Ruxolitinib 
KSR1 Kinase suppressor of ras 1  
LYN v-yes-1 Yamaguchi sarcoma viral-related oncogene homolog  
MAP3K8 Mitogen-activated protein kinase kinase kinase 8  
MAP4K4 Mitogen-activated protein kinase kinase kinase kinase 4  
MAP4K5 Mitogen-activated protein kinase kinase kinase kinase 5 Vemurafenib 
MAPK7 MAPK7  
MET met proto-oncogene (hepatocyte growth factor receptor) Crizotinib 
PIK3CD Phosphoinositide-3-kinase, catalytic, delta polypeptide SF 1126, PX-866, NVP-BEZ235, GDC-0941, BKM120, XL147, CAL-101 
PIP4K2A (PIP5K2A) Phosphatidylinositol-5-phosphate 4-kinase, type II, α  
PRKCA Protein kinase C, α L-threo-safingol, ingenol 3-angelate 
PRKCE Protein kinase C, epsilon Ingenol 3-angelate 
RPS6KA3 Ribosomal protein S6 kinase, 90 kDa, polypeptide 3  
STK4 Serine/threonine kinase 4  
TLK1 Tousled-like kinase 1  
TRIO Triple functional domain (PTPRF interacting)  
SymbolEntrez gene nameDrugs
CDK6 Cyclin-dependent kinase 6 PD-0332991, flavopiridol 
DGKA Diacylglycerol kinase, α80 kDa  
EPHA4 EPH receptor A4  
FLT1 fms-related tyrosine kinase 1 (VEGF/vascular permeability factor receptor) Sunitinib, pazopanib, axitinib, CEP 7055, vandetanib 
FYN FYN oncogene related to SRC, FGR, YES Dasatinib 
HIPK2 Homeodomain interacting protein kinase 2  
ITK IL2-inducible T-cell kinase  
JAK1 Janus kinase 1 Ruxolitinib 
KSR1 Kinase suppressor of ras 1  
LYN v-yes-1 Yamaguchi sarcoma viral-related oncogene homolog  
MAP3K8 Mitogen-activated protein kinase kinase kinase 8  
MAP4K4 Mitogen-activated protein kinase kinase kinase kinase 4  
MAP4K5 Mitogen-activated protein kinase kinase kinase kinase 5 Vemurafenib 
MAPK7 MAPK7  
MET met proto-oncogene (hepatocyte growth factor receptor) Crizotinib 
PIK3CD Phosphoinositide-3-kinase, catalytic, delta polypeptide SF 1126, PX-866, NVP-BEZ235, GDC-0941, BKM120, XL147, CAL-101 
PIP4K2A (PIP5K2A) Phosphatidylinositol-5-phosphate 4-kinase, type II, α  
PRKCA Protein kinase C, α L-threo-safingol, ingenol 3-angelate 
PRKCE Protein kinase C, epsilon Ingenol 3-angelate 
RPS6KA3 Ribosomal protein S6 kinase, 90 kDa, polypeptide 3  
STK4 Serine/threonine kinase 4  
TLK1 Tousled-like kinase 1  
TRIO Triple functional domain (PTPRF interacting)  

NOTE: Twenty-three TNBC-related kinases were analyzed through core, compare, and miRNA target filter analysis from 2000–2012 Ingenuity Systems, Inc.

To determine whether these thirteen kinases are associated with a specific TNBC subgroup, we examined their expression levels in 20 TNBC cell lines, including 10 basal-like, 8 mesenchymal-like, and 2 unclassified, according to previously published criteria (2). We found that the average expression levels of these thirteen kinases were elevated in one unclassified, six of ten basal-like, and all eight mesenchymal-like TNBC subtypes compared with expression levels in the luminal breast cancer subtype (Fig. 1C). The results indicated that mRNA expression levels of these 13 kinases are higher in most TNBCs tested, especially in the mesenchymal-like subtype.

To validate the above findings, we compared the mRNA and protein expression levels of the 13 kinases in breast cancer cell lines; among these, the mRNA expression levels of eight (CDK6, FYN, LYN, MET, PRKCA, RPS6KA3, MAP4K4, and PIK3CD) were significantly higher in at least two of four TNBC cell lines (BT549, Hs578T, MDA-MB-231, and MDA-MB-468) compared with expression levels in a non-TNBC cell line (T47D; Fig. 2A and Supplementary Fig. S1). Western blot analysis indicated that expression levels of MET, PKC-α (from PRKCA), RSK-2 (from RPS6KA3), LYN, FYN, and CDK6 were higher in most of the six TNBC cell lines tested than in six non-TNBC cell lines (Fig. 2B). Although the levels of PIK3CD (which encodes PI3K δ) and MAP4K4 mRNA were higher in TNBC cell lines than in non-TNBC cell lines, the protein expression levels were not. Generally, the protein levels of these kinases were overexpressed in most TNBC cell lines tested; therefore, they served as a basis for further experimental analysis.

Figure 2.

Eight TNBC-related kinases are overexpressed in TNBC cell lines and stem cell population. A, qPCR analysis of the expression of the indicated TNBC-related kinases in TNBC cells compared with expression in non-TNBC cells (T47D). B, immunoblot analyses of protein expression levels of TNBC-related kinases in a panel of breast cancer cell lines. qPCR analysis of the expression levels of TNBC-related kinases in stem cell populations (CD44+/CD24−/low) from BT20 (C) and MCF7 (D) breast cancer cells compared with expression in non-stem cell populations (CD24+ CD44low). Data are means with SDs (n = 3). *, P < 0.05; **, P < 0.005.

Figure 2.

Eight TNBC-related kinases are overexpressed in TNBC cell lines and stem cell population. A, qPCR analysis of the expression of the indicated TNBC-related kinases in TNBC cells compared with expression in non-TNBC cells (T47D). B, immunoblot analyses of protein expression levels of TNBC-related kinases in a panel of breast cancer cell lines. qPCR analysis of the expression levels of TNBC-related kinases in stem cell populations (CD44+/CD24−/low) from BT20 (C) and MCF7 (D) breast cancer cells compared with expression in non-stem cell populations (CD24+ CD44low). Data are means with SDs (n = 3). *, P < 0.05; **, P < 0.005.

Close modal

To evaluate the expression levels of the eight kinases in BTICs, we isolated CD44+/CD24−/low stem cells from a TNBC cell line (BT20) by flow cytometry. CDK6, LYN, MAP4K4, MET, PIK3CD, PRKCA, and RPS6KA3 mRNA levels were upregulated in the isolated CD44+/CD24−/low subpopulations from TNBC cell line (BT20 cells; Fig. 2C) compared with the levels in CD44low/CD24+ nonstem cells. Interestingly, FYN, LYN, MAP4K4, PIK3CD, PRKCA, and RPS6KA3 mRNA levels were also upregulated in the isolated CD44+/CD24−/low subpopulations from the non-TNBC cell line (MCF7 cells; Fig. 2D), suggesting that the eight kinases are commonly overexpressed in BTICs. Because the increased expression of the eight kinases is likely an important feature of TNBC, we named them “TNBC-related kinases.”

Expression of TNBC-related kinases correlates with TNBC clinical subtypes

The second stage of our bioinformatics analysis (Fig. 1A) was to determine whether the eight TNBC-related kinases correlated with TNBC clinical subtypes. We examined mRNA expression levels of the TNBC-related kinases in the data from TCGA breast invasive carcinoma patient cohort (16). We used ERBB2, ESR1 (which encodes ER), and PGR (which encodes PR) gene expression patterns to distinguish TNBC patients from non-TNBC patients. The hierarchical clustering algorithm used is based closely on the average-linkage method, indicating that the mRNA expression levels of the eight TNBC-related kinases highly correlated with TNBC (Fig. 3A). Box plots were generated from the original log2-transformed mRNA expression levels of the eight kinases by breast cancer subtype on the basis of the mRNA expression levels of ESR1, PGR, and ERBB2. These findings indicate that expression of the TNBC-related kinases is strongly associated with TNBC versus non-TNBC (Fig. 3B). The histogram PNG files show how cutoffs were set based on unbiased and automatic K-Means algorithm for the separation of TNBC and non-TNBC tumors (Fig. 3C). Taken together, the results of both cell and patient database analyses suggest that TNBC-related kinases likely play important roles in the aggressive behavior of TNBC.

Figure 3.

Cluster analysis identified eight TNBC-related kinases overexpressed in TNBC from the TCGA breast invasive carcinoma patient cohort. A, hierarchical clustering analysis of eight TNBC-related kinases that distinguish TNBC from other human breast tumors (n = 753). Patients with breast cancer were categorized by mRNA subtype as non-TNBC patients (n = 613) and TNBC patients (n = 140) according to expression of ERBB2, ESR1, and PGR (encoding HER2, ER, and PR, respectively). The color scale ranges from saturated green for log ratios of −2.0 and below to saturated red for log ratios of 2.0 and above. Red represents high gene expression (>2), and green represents low gene expression (<2). B, box plot generated from original and log2-transformed mRNA expression levels of TNBC-related kinases in TNBC patients and non-TNBC patients by ESR1, PGR, and ERBB2 mRNA subtyping. C, data were clustered by the standard K-means clustering algorithm to separate ESR1-high and ESR1-low expression, and PGR-high and PGR-low expression tumors. Cutoffs were set automatically (indicated by red line).

Figure 3.

Cluster analysis identified eight TNBC-related kinases overexpressed in TNBC from the TCGA breast invasive carcinoma patient cohort. A, hierarchical clustering analysis of eight TNBC-related kinases that distinguish TNBC from other human breast tumors (n = 753). Patients with breast cancer were categorized by mRNA subtype as non-TNBC patients (n = 613) and TNBC patients (n = 140) according to expression of ERBB2, ESR1, and PGR (encoding HER2, ER, and PR, respectively). The color scale ranges from saturated green for log ratios of −2.0 and below to saturated red for log ratios of 2.0 and above. Red represents high gene expression (>2), and green represents low gene expression (<2). B, box plot generated from original and log2-transformed mRNA expression levels of TNBC-related kinases in TNBC patients and non-TNBC patients by ESR1, PGR, and ERBB2 mRNA subtyping. C, data were clustered by the standard K-means clustering algorithm to separate ESR1-high and ESR1-low expression, and PGR-high and PGR-low expression tumors. Cutoffs were set automatically (indicated by red line).

Close modal

Overexpression of PRKCA, MET, and CDK6 correlates with poor prognosis in patients with TNBC

To identify the most suitable therapeutic targets among the eight TNBC-related kinases for TNBC, we analyzed the relationship between the TNBC-related kinases and overall survival in patients with breast cancer. Using an online Kaplan–Meier plotter (36), three out of the eight kinases (PRKCA, MET, and CDK6) that showed high expression were associated with shorter overall survival in patients with TNBC (P = 0.019, 0.02, and 0.073, respectively; Fig. 4A and data not shown) based on the univariate analysis with selected probes. Consistently, another data analysis from breast cancer cohort NKI-295 (28) with selected cutoffs also indicated that PRKCA, MET, and CDK6 expression levels were correlated with worse overall survival (P = 0.003, 0.086, and 0.002, respectively; Fig. 4B). Furthermore, high expression of PRKCA/MET (P = 0.018), PRKCA/CDK6 (P < 0.0001), or CDK6/MET (P = 0.005) correlated with more adverse survival outcomes than each one alone in patients with breast cancer (Fig. 4C). Interestingly, the survival rates of patients with breast cancer in whom three kinases PRKCA/MET/CDK6 were coexpressed were not worse than in whom two kinases were coexpressed (data not shown), providing a rationale to investigate whether the dual-drug combinations sufficiently impede TNBC progression. Together, these results suggest that expression of PRKCA, MET, or CDK6 has the potential to serve as a prognostic marker and therapeutic target for breast cancer, especially TNBC.

Figure 4.

PRKCA, MET, and CDK6 expression levels are negatively correlated with probability of survival in patients with breast cancer. A, Kaplan–Meier (KM) estimates of overall survival for patients with basal-like breast cancer in relation to PRKCA, MET, or CDK6 mRNA expression level from KM plotter analysis (36) using auto-selected best cutoffs to group patients. A statistical analysis of time-to-event data with the expression of PRKCA (HR, 3.21; 95% confidence interval, 1.15–9), MET (HR, 2.2; 95% confidence interval, 1.11–4.36, and CDK6 (HR, 1.81; 95% confidence interval, 0.94–3.51) were estimated. Kaplan–Meier plots of patient survival stratified by gene expression level (high/low) with selected probes are shown in figure. Black line represents low expression; red line high expression. B, Kaplan–Meier overall survival curves in relation to PRKCA, MET, and CDK6 expression levels in patients with breast cancer. C, Kaplan–Meier overall survival curves in relation to coexpression of PRKCA/MET, PRKCA/CDK6, and CDK6/MET in patients with breast cancer. The NKI-295 database (28) was used for PRKCA, MET, CDK6, PRKCA/MET, PRKCA/CDK6, and CDK6/MET gene expression analysis and overall survival. A gene expression level within the first quartile was defined as positive. Blue line, nonexpression; green line, single-expression; red line, dual-expression. Cum, cumulative.

Figure 4.

PRKCA, MET, and CDK6 expression levels are negatively correlated with probability of survival in patients with breast cancer. A, Kaplan–Meier (KM) estimates of overall survival for patients with basal-like breast cancer in relation to PRKCA, MET, or CDK6 mRNA expression level from KM plotter analysis (36) using auto-selected best cutoffs to group patients. A statistical analysis of time-to-event data with the expression of PRKCA (HR, 3.21; 95% confidence interval, 1.15–9), MET (HR, 2.2; 95% confidence interval, 1.11–4.36, and CDK6 (HR, 1.81; 95% confidence interval, 0.94–3.51) were estimated. Kaplan–Meier plots of patient survival stratified by gene expression level (high/low) with selected probes are shown in figure. Black line represents low expression; red line high expression. B, Kaplan–Meier overall survival curves in relation to PRKCA, MET, and CDK6 expression levels in patients with breast cancer. C, Kaplan–Meier overall survival curves in relation to coexpression of PRKCA/MET, PRKCA/CDK6, and CDK6/MET in patients with breast cancer. The NKI-295 database (28) was used for PRKCA, MET, CDK6, PRKCA/MET, PRKCA/CDK6, and CDK6/MET gene expression analysis and overall survival. A gene expression level within the first quartile was defined as positive. Blue line, nonexpression; green line, single-expression; red line, dual-expression. Cum, cumulative.

Close modal

Dual-drug combinations of PKC-α, MET, and CDK6 inhibitors synergistically inhibit TNBC cell proliferation

Given that expression of two of the three kinases PKC-α, MET, and CDK6 was more highly correlated with patient survival than was expression of any one of the three kinases alone and that PRKCA/MET/CDK6 were not worse than in whom two kinases were coexpressed, we investigated the anti-TNBC efficacies of dual-drug combinations of safingol (a PKC-α and sphingosine kinase 1inhibitor), crizotinib (a MET inhibitor), and PD0332991 (a CDK4/6 inhibitor)—drugs that either are in clinical trials or have been approved by the Food and Drug Administration (and therefore could be readily tested in combination in clinical trials). As shown in Fig. 5A, when two of the three kinase inhibitors were combined, the synergistic cytotoxic effects seemed to be higher in TNBC cells than in non-TNBC cells. In addition, normal epithelial cells (MCF10A) exhibited more resistance of all the dual-drug combinations under similar treatments conditions, suggesting that these dual-drug combinations preferentially kill TNBC cells over non-TNBC cancer cells or normal breast epithelial cells. To further demonstrate whether CSC population is more sensitive to the combination therapy, drug-treated MDA-MB-231 cells were subjected to FACS to determine the percentage of CD44+/CD24−/low population and assess the cells' ability to form mammospheres. As shown in Supplementary Fig. S2A, MDA-MB-231 cells treated with DMSO contained 81 ± 5% CD44+/CD24−/low population, whereas those treated with combination of safingol–crizotinib, safingol–PD0332991, or crizotinib–PD0332991 contained significantly less (34 ± 12%, 24 ± 12%, or 26 ± 10%, respectively). In addition, all three combinations and single drug treatments caused a decrease in mammosphere formation (Supplementary Fig. S2B). Taken together, these findings support the notion that the stem-like MDA-MB-231 cells were more sensitive to these three dual-drug combinations than their non-stem like counterparts.

Figure 5.

Dual-drug combinations of PKC-α, MET, and CDK6 inhibitors synergistically inhibit TNBC cell growth. A, TNBC cells (MDA-MB-231, BT549, and Hs578T), non-TNBC cells (T47D and SK-BR-3), and normal epithelial cells (MCF10A) were used to determine the specificities of the cytotoxic effects of safingol (Sa), crizotinib (Cr), and PD0332991 (PD) in dual-drug combinations. Cells (1 × 103–3 × 103) were plated in 96-well plates, and cell growth was determined by MTT assay 2 days after cells were treated with an inhibitor or a combination of inhibitors. *, P < 0.05. B, CI plots for the dual-drug combinations of inhibitors in six breast cancer cells lines. Cells were treated with combinations of safingol (0, 0.5, 1, 2, 3, 4, or 5 μmol/L), crizotinib (0, 1, 2, 3, 4, 5, or 6 μmol/L), and PD0332991 (0, 1, 2, 4, 6, 8, or 10 μmol/L) for 48 hours. Open and closed symbols represent non-TNBC and TNBC cells, respectively. CI values were plotted as a function of fractional inhibition as determined by MTT analysis and computer simulation (CompuSyn) for the Fa (fraction affected) range of 0.10 to 0.95. CIs of <1, 1, and >1 indicate synergism, additive effect, and antagonism, respectively. At least three independent experiments with three replicates were performed.

Figure 5.

Dual-drug combinations of PKC-α, MET, and CDK6 inhibitors synergistically inhibit TNBC cell growth. A, TNBC cells (MDA-MB-231, BT549, and Hs578T), non-TNBC cells (T47D and SK-BR-3), and normal epithelial cells (MCF10A) were used to determine the specificities of the cytotoxic effects of safingol (Sa), crizotinib (Cr), and PD0332991 (PD) in dual-drug combinations. Cells (1 × 103–3 × 103) were plated in 96-well plates, and cell growth was determined by MTT assay 2 days after cells were treated with an inhibitor or a combination of inhibitors. *, P < 0.05. B, CI plots for the dual-drug combinations of inhibitors in six breast cancer cells lines. Cells were treated with combinations of safingol (0, 0.5, 1, 2, 3, 4, or 5 μmol/L), crizotinib (0, 1, 2, 3, 4, 5, or 6 μmol/L), and PD0332991 (0, 1, 2, 4, 6, 8, or 10 μmol/L) for 48 hours. Open and closed symbols represent non-TNBC and TNBC cells, respectively. CI values were plotted as a function of fractional inhibition as determined by MTT analysis and computer simulation (CompuSyn) for the Fa (fraction affected) range of 0.10 to 0.95. CIs of <1, 1, and >1 indicate synergism, additive effect, and antagonism, respectively. At least three independent experiments with three replicates were performed.

Close modal

To further validate the synergistic killing effects of dual-drug combinations targeting PKC-α, MET, and CDK6, we carried out drug interaction analysis using the Chou–Talalay method to determine the CI (37). TNBC cells were treated with safingol, crizotinib, or PD0332991 alone or in combination (safingol plus crizotinib, safingol plus PD0332991, or crizotinib plus PD0332991) at various concentrations. The CIs of the dual drug combinations are shown in Fig. 5B; CI values of <1, 1, and >1 indicate that the two drugs in the combination have synergistic, additive, or antagonistic effects, respectively. All three dual-drug combinations synergistically inhibited TNBC cell proliferation, especially at the highest EDs, which killed most cancer cells. The greatest synergism (CI) for the safingol–crizotinib combination was observed in MDA-MB-231 and BT549 cells at 95% EDs with CI values of 0.35 and 0.15, respectively. However, for the crizotinib-PD0332991 combination, the greatest synergism observed in MDA-MB-231 and Hs578T cells was only at 50% EDs with CI values of 0.39 and 0.53, respectively (Table 2). According to these results, the combination with the best therapeutic effect was safingol–crizotinib with the strongest synergism, followed by safingol–PD0332991 and crizotinib–PD0332991 in that order. Interestingly, the safingol–crizotinib combination treatment also showed a slight synergism in non-TNBC cancer cells (0.82 at 95% EDs in T47D cells) and strong antagonistic effect in normal epithelial cells. These findings further indicate that these dual-drug combinations, especially safingol–crizotinib, may be effective against TNBC progression.

Table 2.

CIs of dual-drug combinations in different breast cancer cells

CI
Drug combinationED50ED75ED90ED95
MDA-MB-231 
 Sa-Cr 1.00 0.66 0.44 0.35 
 Sa-PD 1.03 0.70 0.57 0.57 
 Cr-PD 0.39 0.76 1.72 3.27 
BT549 
 Sa-Cr 1.96 0.68 0.27 0.15 
 Sa-PD 0.81 0.74 0.70 0.69 
 Cr-PD 1.11 1.02 0.94 0.89 
Hs578T 
 Sa-Cr 0.52 0.60 0.71 0.79 
 Sa-PD 0.58 0.79 1.17 1.56 
 Cr-PD 0.53 1.26 3.20 6.05 
SK-BR-3 
 Sa-Cr 1.23 1.07 0.93 0.84 
 Sa-PD 1.20 1.25 1.30 1.34 
 Cr-PD 1.55 1.45 1.36 1.31 
T47D 
 Sa-Cr 0.99 0.90 0.85 0.82 
 Sa-PD 1.04 0.99 1.06 1.15 
 Cr-PD 1.11 1.18 1.26 1.32 
MCF10A 
 Sa-Cr 3.03 13.10 59.89 168.96 
 Sa-PD 1.42 1.20 1.04 0.96 
 Cr-PD 1.22 1.09 1.01 0.98 
CI
Drug combinationED50ED75ED90ED95
MDA-MB-231 
 Sa-Cr 1.00 0.66 0.44 0.35 
 Sa-PD 1.03 0.70 0.57 0.57 
 Cr-PD 0.39 0.76 1.72 3.27 
BT549 
 Sa-Cr 1.96 0.68 0.27 0.15 
 Sa-PD 0.81 0.74 0.70 0.69 
 Cr-PD 1.11 1.02 0.94 0.89 
Hs578T 
 Sa-Cr 0.52 0.60 0.71 0.79 
 Sa-PD 0.58 0.79 1.17 1.56 
 Cr-PD 0.53 1.26 3.20 6.05 
SK-BR-3 
 Sa-Cr 1.23 1.07 0.93 0.84 
 Sa-PD 1.20 1.25 1.30 1.34 
 Cr-PD 1.55 1.45 1.36 1.31 
T47D 
 Sa-Cr 0.99 0.90 0.85 0.82 
 Sa-PD 1.04 0.99 1.06 1.15 
 Cr-PD 1.11 1.18 1.26 1.32 
MCF10A 
 Sa-Cr 3.03 13.10 59.89 168.96 
 Sa-PD 1.42 1.20 1.04 0.96 
 Cr-PD 1.22 1.09 1.01 0.98 

NOTE: TNBC cells (MDA-MB-231, BT549, and Hs578T), non-TNBC cells (T47D and SK-BR-3), and normal epithelial cells (MCF10A) were used to determine the CI index of safingol (Sa), crizotinib (Cr), and PD0332991 (PD) in dual-drug combinations. CIs of <1, 1, and >1 indicate synergism, additive effect, and antagonism, respectively. CI index <0.9 is labeled in bold.

Combined inhibition of PKC-α and MET suppresses TNBC tumorigenesis in vitro and in vivo

Next, we examined the effects of the dual drug combinations on the tumorigenic potential of TNBC cells. All three combinations efficiently inhibited colony-forming ability in BT549 cells (Fig. 6A). Safingol combined with crizotinib or PD0332991 suppressed anchorage-independent colony formation more efficiently than did any one of the drugs by itself in both MDA-MB-231 and BT549 cells (Fig. 6B). However, crizotinib combined with PD0332991 did not inhibit anchorage-independent growth more efficiently than PD0332991 alone; suggesting that inhibition of PKC-α activity is required for optimal killing of TNBC cells. Indeed, PKC-α was recently identified as a central signaling node in non-CSC to CSC transformation and has been proposed as a potential therapeutic target for breast cancer (38). We further examined PKC-α expression and patient outcomes with a TNBC tissue microarray by IHC analysis, with results indicating that the survival rate was worse in patients with TNBC with high levels of PKC-α expression than those with low levels (P = 0.045; Fig. 6C). Collectively, our findings suggest that PKC-α is a unique prognostic indicator of TNBC patient survival.

Figure 6.

Dual-drug combinations of PKC-α, MET, and CDK6 inhibitors synergistically suppressed clonogenic and anchorage-independent growth in vitro and in an animal model. A, the number of colony-forming cells was determined by clonogenic assay in BT549. Cells were treated with safingol (Sa; 0.75 μmol/L), crizotinib (Cr; 0.75 μmol/L), and PD0332991 (PD; 0.75 μmol/L), separately or in combination, for 10 days. B, the number of colony-forming cells was determined by anchorage-independent growth analysis in MDA-MB-231 and BT549 cells. Cells were treated with safingol (0.5 μmol/L), crizotinib (0.5 μmol/L), and PD0332991 (0.2 μmol/L), separately or in combination, for 21 days. Representative images from the clonogenic and anchorage-independent growth assays (carried out as described in Materials and Methods) are shown. The data on the right are means with SDs for three independent experiments. *, P < 0.05 for differences between combination and single-drug treatments. C, representative images of IHC staining of PKC-α expression in human TNBC tissues (left). TNBC tissue array analysis of overall survival in PKC-α-overexpressing patients. To ensure objectivity, the ACIS III automated cellular imaging system (Dako) was used for tissue scoring and quantification. All correlations between kinase expression levels were analyzed using SPSS version 15.0 statistical software (right). The overall survival durations after surgery were analyzed using a Kaplan–Meier plot. The Cox proportional hazards regression model was used to assess prognostic factors for overall survival. Cum, cumulative. D, seven days after transplantation of MDA-MB-231 cells expressing luciferase and green fluorescent protein, mice were treated with safingol at 20 mg/kg intravenously twice a week and/or crizotinib at 8 mg/kg orally every day. Tumor volume (V) was measured on days 8, 12, 15, 19, and 21 and was normalized to the tumor volume in untreated mice (V0). Error bars, SDs (n = 5). CTL, control; Inh., inhibition. E, body weight of mice described in D. Data represent mean ± SD. n = 5. *, P < 0.05, **, P < 0.01 compared with the control group.

Figure 6.

Dual-drug combinations of PKC-α, MET, and CDK6 inhibitors synergistically suppressed clonogenic and anchorage-independent growth in vitro and in an animal model. A, the number of colony-forming cells was determined by clonogenic assay in BT549. Cells were treated with safingol (Sa; 0.75 μmol/L), crizotinib (Cr; 0.75 μmol/L), and PD0332991 (PD; 0.75 μmol/L), separately or in combination, for 10 days. B, the number of colony-forming cells was determined by anchorage-independent growth analysis in MDA-MB-231 and BT549 cells. Cells were treated with safingol (0.5 μmol/L), crizotinib (0.5 μmol/L), and PD0332991 (0.2 μmol/L), separately or in combination, for 21 days. Representative images from the clonogenic and anchorage-independent growth assays (carried out as described in Materials and Methods) are shown. The data on the right are means with SDs for three independent experiments. *, P < 0.05 for differences between combination and single-drug treatments. C, representative images of IHC staining of PKC-α expression in human TNBC tissues (left). TNBC tissue array analysis of overall survival in PKC-α-overexpressing patients. To ensure objectivity, the ACIS III automated cellular imaging system (Dako) was used for tissue scoring and quantification. All correlations between kinase expression levels were analyzed using SPSS version 15.0 statistical software (right). The overall survival durations after surgery were analyzed using a Kaplan–Meier plot. The Cox proportional hazards regression model was used to assess prognostic factors for overall survival. Cum, cumulative. D, seven days after transplantation of MDA-MB-231 cells expressing luciferase and green fluorescent protein, mice were treated with safingol at 20 mg/kg intravenously twice a week and/or crizotinib at 8 mg/kg orally every day. Tumor volume (V) was measured on days 8, 12, 15, 19, and 21 and was normalized to the tumor volume in untreated mice (V0). Error bars, SDs (n = 5). CTL, control; Inh., inhibition. E, body weight of mice described in D. Data represent mean ± SD. n = 5. *, P < 0.05, **, P < 0.01 compared with the control group.

Close modal

Given that the safingol–crizotinib combination consistently showed synergistic killing effect from multiple in vitro assays, we further evaluated the efficacy of the safingol–crizotinib combination in a TNBC orthotopic xenograft mouse model. In agreement with results from our in vitro assays, the safingol–crizotinib combination was more effective in reducing tumor growth than was either single-drug treatment or vehicle control treatment (Fig. 6D). The inhibition efficiency in mice was 68% for the combination (CI = 0.43) but only 6% (P = 0.006) and 21% (P = 0.002) for safingol alone and crizotinib alone, respectively. We did not observe any significant body weight changes when mice were treated with the safingol or crizotinib alone or in combination (Fig. 6E). These results suggest that the dual-drug combination of PKC-α with MET may be effective against TNBC and could be readily applied for marker-guided clinical trials as inhibitors of the TNBC-related kinases are available in clinical trials or have been approved by FDA.

Our analysis with multiple cell line and patient databases identified eight TNBC-related kinases—MET, FYN, PKC-α, MAP4K4, LYN, PI3K δ, CDK6, and RSK-2—that are commonly overexpressed in TNBC/BTICs. Of these eight kinases, PKC-α, MET, and CDK6 associated with the worst survival outcomes in patients with breast cancer. Dual-drug combinations that target these three kinases synergistically inhibited TNBC cell proliferation and in vitro tumorigenic potential. The combination that showed the strongest synergism (PKC-α and MET inhibition) also synergistically (CI = 0.43; Fig. 6D) attenuated tumor growth in vivo. Our findings suggest that combining drugs already in use may accelerate the availability of marker-guided therapy for TNBC.

Currently, the most promising clinical target for TNBC is the enzyme PARP, a member of the family of nuclear enzymes involved in the detection and repair of DNA damage (39). There are at least five PARP inhibitors that are in clinical trials, and among them, BSI-201 (iniparib; ref. 40) and AZD2281 (olaparib; ref. 41) have been evaluated in multiple clinical trials in women with metastatic breast cancer. However, the results indicated that the reduced rate of tumor regression was observed mostly in patients with the BRCA1 or BRCA2 mutations (42). Thus, efficacy of PARP inhibitors in patients with TNBC is still nuclear (43). Recent encouraging data showed that inhibition of PI3K impairs DNA homologous recombination and sensitizes TNBC without BRCA mutations to poly(ADP-ribose) inhibition (44). However, this mechanism has yet to be tested in patients, and important questions remain unanswered.

TNBC is the most heterogeneous subtype of breast cancer, and because no individual cell line represents the heterogeneity of TNBC, it is difficult to find an appropriate TNBC model. The well-established cell-based systems, such as the BPLER cancer cell line derived from human primary breast epithelial cells, have a TNBC-like phenotype in vitro and are highly enriched for BTICs (45). Although BPLER mimics TNBC, this system is most appropriate for the basal-like TNBC subtype, which accounts for about half of the TNBC cases (46). In identifying 13 TNBC-related kinases that are expressed in most TNBC subtypes (Fig. 1C), including the basal-like and mesenchymal-like subtypes, our study overcame some of the limitations associated with TNBC heterogeneity.

In the absence of major growth stimulation on their surface, TNBC cells may rely on downstream signaling to compensate for the loss of cell-surface signaling to support cell growth. Kinases are ideal target molecules, as they transmit signals, control complex cellular processes, and are frequently activated in cancer by mutation, constitutive activation, or overexpression. A gene expression profiling study of 102 human breast tumors identified several kinases as druggable targets in ER-negative breast cancer (47). The identification of LYN, MAP4K4, MET, and RPS6KA3 kinases in both that study and ours suggests that these kinases are common regulators in TNBC. We further showed that dual-drug combinations using PKC-α, CDK6, and MET kinase inhibitors enhanced the killing on TNBC cells, reduced the tumorigenic phenotype of cells in vitro, and synergistically impeded tumor growth in vivo. These results further attest to the critical role these kinases play in transmitting oncogenic signaling in TNBC. It may be worthwhile to test the efficacy of other dual-drug combinations in an animal model. However, our findings indicate that the combination of PKC-α and MET inhibitors is efficacious against tumors in mice and should be tested in patients with TNBC.

Our study, together with the previous reports, suggests that PKC-α plays an important role in both TNBC and BTICs. An inverse relationship between PKC-α activity and ER expression in human breast cell lines and tumors was firmly established over 25 years ago, with ER-negative cells expressing significantly higher levels of PKC-α than the levels expressed in ER-positive cancer cells (48). PKC-α overexpression has also been suggested to play a role in growth signaling when breast cancer shifts from hormone dependence to hormone independence (49). More recently, PDGFR/PKC-α/FRA1 signaling was identified upon activation of epithelial–mesenchymal transition from nonCSCs to CSCs, and inhibition of either PKC-α or FRA1 was suggested to have potential therapeutic value in aggressive breast cancer (38). Although the role of PKC-α in breast cancer has been established, none of the clinical trials using PKC-α inhibitors specifically targeted TNBC, ER-negative breast cancer, or metastatic breast cancer. Our study has demonstrated an association between PKC-α and TNBC both in cell lines and in patient tumors, and our findings indicate that PKC-α is a unique prognostic marker and a potential therapeutic target for TNBC.

G.B. Mills received commercial research grants from Adelson Medical Research Foundation, AstraZeneca, Critical Outcomes Technology, and GSK, has ownership interest (including patents) in Catena Pharmaceuticals, PTV Ventures, Spindle Top Ventures, and is a consultant/advisory board member for AstraZeneca, Blend, Tau Therapeutics, Critical Outcome Technologies, HanAl Bio Korea, Nuevolution, Pfizer, Provista Diagnostics, Roche, Signalchem Lifesciences, and Symphogen. No potential conflicts of interest were disclosed by the other authors.

Conception and design: Y.-H. Hsu, T.-J. Wu, W.-C. Huang, C.-L. Liu, M.-Y. Wang, A.A. Sahin, M.-C. Hung

Development of methodology: Y.-H. Hsu, L.-C. Chan, G.B. Mills

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): Y.-H. Hsu, L.-C. Chan, Y.-F. Fang, Y. Wu, C.-L. Liu, M.-Y. Wang, M.-K. Chen, D. Yu

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): Y.-H. Hsu, J. Yao, L.-C. Chan, T.-J. Wu, Y. Wei, A.A. Sahin, A. Sood, G.B. Mills, G.N. Hortobagyi, M.-C. Hung

Writing, review, and/or revision of the manuscript: Y.-H. Hsu, J. Yao, J.L. Hsu, A. Sood, G.B. Mills, D. Yu, G.N. Hortobagyi, M.-C. Hung

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): Y.-H. Hsu, T.-J. Wu, W.-C. Huang, Y.-C. Chang, J. Shen, M.-K. Chen, G.B. Mills, G.N. Hortobagyi, M.-C. Hung

Study supervision: Y.-H. Hsu, W.-C. Huang, C.-W. Li, M.-C. Hung

Other (discussed the project with the first author): D. Yu

The authors thank Arthur Gelmis from Scientific Publications at MD Anderson for editing the article and Su Zhang, Jian Guan Shi, Zhenbo Han, Jin-Fong Lee, and Junyan Wang for technical assistance. STR DNA fingerprinting was performed at the Cancer Center Support Grant-funded Characterized Cell Line Core Facility of MD Anderson Cancer Center (NCI grant CA016672). In memoriam, Mr. Tiong Loi Ang for his courageous fight against cancer.

This work was funded, in part by, the NIH (grants CA109311, CA099031, and CCSG CA16672), the National Breast Cancer Foundation, Inc., the Breast Cancer Research Foundation, The University of Texas MD Anderson-China Medical University, and Hospital Sister Institution Fund (M.-C. Hung), the Ministry of Health and Welfare, China Medical University Hospital Cancer Research Center of Excellence (MOHW103-TD-B-111-03; Taiwan), the Program for Stem Cell and Regenerative Medicine Frontier Research (NSC 102-2321-B-039-001; Taiwan), the International Research-Intensive Centers of Excellence in Taiwan (NSC 103-2911-I-002-303), and the Center for Biological Pathways, Postdoctoral Research Abroad Program (grant 101IP999900059 from the National Science Council of Taiwan; Y.-H. Hsu).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Foulkes
WD
,
Smith
IE
,
Reis-Filho
JS
. 
Triple-negative breast cancer
.
N Engl J Med
2010
;
363
:
1938
48
.
2.
Lehmann
BD
,
Bauer
JA
,
Chen
X
,
Sanders
ME
,
Chakravarthy
AB
,
Shyr
Y
, et al
Identification of human triple-negative breast cancer subtypes and preclinical models for selection of targeted therapies
.
J Clin Invest
2011
;
121
:
2750
67
.
3.
Dent
R
,
Trudeau
M
,
Pritchard
KI
,
Hanna
WM
,
Kahn
HK
,
Sawka
CA
, et al
Triple-negative breast cancer: clinical features and patterns of recurrence
.
Clin Cancer Res
2007
;
13
:
4429
34
.
4.
Carey
L
,
Winer
E
,
Viale
G
,
Cameron
D
,
Gianni
L
. 
Triple-negative breast cancer: disease entity or title of convenience?
Nat Rev Clin Oncol
2010
;
7
:
683
92
.
5.
Liedtke
C
,
Mazouni
C
,
Hess
KR
,
Andre
F
,
Tordai
A
,
Mejia
JA
, et al
Response to neoadjuvant therapy and long-term survival in patients with triple-negative breast cancer
.
J Clin Oncol
2008
;
26
:
1275
81
.
6.
Dick
JE
. 
Stem cell concepts renew cancer research
.
Blood
2008
;
112
:
4793
807
.
7.
Li
X
,
Lewis
MT
,
Huang
J
,
Gutierrez
C
,
Osborne
CK
,
Wu
MF
, et al
Intrinsic resistance of tumorigenic breast cancer cells to chemotherapy
.
J Natl Cancer Inst
2008
;
100
:
672
9
.
8.
Park
CY
,
Tseng
D
,
Weissman
IL
. 
Cancer stem cell-directed therapies: recent data from the laboratory and clinic
.
Mol Ther
2009
;
17
:
219
30
.
9.
Gupta
PB
,
Onder
TT
,
Jiang
G
,
Tao
K
,
Kuperwasser
C
,
Weinberg
RA
, et al
Identification of selective inhibitors of cancer stem cells by high-throughput screening
.
Cell
2009
;
138
:
645
59
.
10.
Al-Hajj
M
,
Wicha
MS
,
Benito-Hernandez
A
,
Morrison
SJ
,
Clarke
MF
. 
Prospective identification of tumorigenic breast cancer cells
.
Proc Natl Acad Sci U S A
2003
;
100
:
3983
8
.
11.
Ginestier
C
,
Hur
MH
,
Charafe-Jauffret
E
,
Monville
F
,
Dutcher
J
,
Brown
M
, et al
ALDH1 is a marker of normal and malignant human mammary stem cells and a predictor of poor clinical outcome
.
Cell Stem Cell
2007
;
1
:
555
67
.
12.
Reya
T
,
Morrison
SJ
,
Clarke
MF
,
Weissman
IL
. 
Stem cells, cancer, and cancer stem cells
.
Nature
2001
;
414
:
105
11
.
13.
Rossi
DJ
,
Jamieson
CH
,
Weissman
IL
. 
Stems cells and the pathways to aging and cancer
.
Cell
2008
;
132
:
681
96
.
14.
Creighton
CJ
,
Li
X
,
Landis
M
,
Dixon
JM
,
Neumeister
VM
,
Sjolund
A
, et al
Residual breast cancers after conventional therapy display mesenchymal as well as tumor-initiating features
.
Proc Natl Acad Sci U S A
2009
;
106
:
13820
5
.
15.
Sotiriou
C
,
Pusztai
L
. 
Gene-expression signatures in breast cancer
.
N Engl J Med
2009
;
360
:
790
800
.
16.
Cancer Genome Atlas N
. 
Comprehensive molecular portraits of human breast tumours
.
Nature
2012
;
490
:
61
70
.
17.
Bendell
JC
,
Rodon
J
,
Burris
HA
,
de Jonge
M
,
Verweij
J
,
Birle
D
, et al
Phase I, dose-escalation study of BKM120, an oral pan-Class I PI3K inhibitor, in patients with advanced solid tumors
.
J Clin Oncol
2012
;
30
:
282
90
.
18.
Moreno Garcia
V
. 
A phase I study evaluating GDC-0941, an oral phosphoinositide-3 kinase (PI3K) inhibitor, in patients with advanced solid tumors or multiple myeloma
.
J Clin Oncol
2011
;
29
:
a3021
.
19.
Wagner
A
. 
A first-in-human phase I study to evaluate GDC-0980, an oral PI3K/mTOR inhibitor, administered QD in patients with advanced solid tumors
.
J Clin Oncol
2011
;
29
:
a3020
.
20.
Yap
TA
. 
First-in-man clinical trial of the oral pan-AKT inhibitor MK-2206 in patients with advanced solid tumors
.
J Clin Oncol
2011
;
29
:
4688
95
.
21.
Engels
BM
,
Hutvagner
G
. 
Principles and effects of microRNA-mediated post-transcriptional gene regulation
.
Oncogene
2006
;
25
:
6163
9
.
22.
Romano
P
,
Manniello
A
,
Aresu
O
,
Armento
M
,
Cesaro
M
,
Parodi
B
. 
Cell Line Data Base: structure and recent improvements towards molecular authentication of human cell lines
.
Nucleic Acids Res
2009
;
37
:
D925
32
.
23.
Barretina
J
,
Caponigro
G
,
Stransky
N
,
Venkatesan
K
,
Margolin
AA
,
Kim
S
, et al
The Cancer Cell Line Encyclopedia enables predictive modelling of anticancer drug sensitivity
.
Nature
2012
;
483
:
603
7
.
24.
Komurov
K
,
Dursun
S
,
Erdin
S
,
Ram
PT
. 
NetWalker: a contextual network analysis tool for functional genomics
.
BMC Genomics
2012
;
13
:
282
.
25.
Eisen
MB
,
Spellman
PT
,
Brown
PO
,
Botstein
D
. 
Cluster analysis and display of genome-wide expression patterns
.
Proc Natl Acad Sci U S A
1998
;
95
:
14863
8
.
26.
Sokal
RR
,
Michener
CD
,
Kansas
Uo
. 
A Statistical Method for Evaluating Systematic Relationships
:
University of Kansas
;
Lawrence, KS
; 
1958
.
27.
Bolt
MW
,
Mahoney
PA
. 
High-efficiency blotting of proteins of diverse sizes following sodium dodecyl sulfate-polyacrylamide gel electrophoresis
.
Anal Biochem
1997
;
247
:
185
92
.
28.
van de Vijver
MJ
,
He
YD
,
van't Veer
LJ
,
Dai
H
,
Hart
AA
,
Voskuil
DW
, et al
A gene-expression signature as a predictor of survival in breast cancer
.
N Engl J Med
2002
;
347
:
1999
2009
.
29.
Chou
TC
. 
Drug combination studies and their synergy quantification using the Chou-Talalay method
.
Cancer Res
2010
;
70
:
440
6
.
30.
Kentsis
A
,
Reed
C
,
Rice
KL
,
Sanda
T
,
Rodig
SJ
,
Tholouli
E
, et al
Autocrine activation of the MET receptor tyrosine kinase in acute myeloid leukemia
.
Nat Med
2012
;
18
:
1118
22
.
31.
Zou
HY
,
Li
Q
,
Lee
JH
,
Arango
ME
,
McDonnell
SR
,
Yamazaki
S
, et al
An orally available small-molecule inhibitor of c-Met, PF-2341066, exhibits cytoreductive antitumor efficacy through antiproliferative and antiangiogenic mechanisms
.
Cancer Res
2007
;
67
:
4408
17
.
32.
Kedderis
LB
,
Bozigian
HP
,
Kleeman
JM
,
Hall
RL
,
Palmer
TE
,
Harrison
SD
 Jr
, et al
Toxicity of the protein kinase C inhibitor safingol administered alone and in combination with chemotherapeutic agents
.
Fundam Appl Toxicol
1995
;
25
:
201
17
.
33.
Tan
KB
,
Ling
LU
,
Bunte
RM
,
Chng
WJ
,
Chiu
GN
. 
In vivo efficacy of a novel liposomal formulation of safingol in the treatment of acute myeloid leukemia
.
J Control Release
2012
;
160
:
290
8
.
34.
Sokilde
R
,
Kaczkowski
B
,
Podolska
A
,
Cirera
S
,
Gorodkin
J
,
Moller
S
, et al
Global microRNA analysis of the NCI-60 cancer cell panel
.
Mol Cancer Ther
2011
;
10
:
375
84
.
35.
Chang
CJ
,
Chao
CH
,
Xia
W
,
Yang
JY
,
Xiong
Y
,
Li
CW
, et al
p53 regulates epithelial-mesenchymal transition and stem cell properties through modulating miRNAs
.
Nat Cell Biol
2011
;
13
:
317
23
.
36.
Gyorffy
B
,
Lanczky
A
,
Eklund
AC
,
Denkert
C
,
Budczies
J
,
Li
Q
, et al
An online survival analysis tool to rapidly assess the effect of 22,277 genes on breast cancer prognosis using microarray data of 1,809 patients
.
Breast Cancer Res Treat
2010
;
123
:
725
31
.
37.
Chou
TC
,
Talalay
P
. 
Quantitative analysis of dose-effect relationships: the combined effects of multiple drugs or enzyme inhibitors
.
Adv Enzyme Regul
1984
;
22
:
27
55
.
38.
Tam Wai
L
,
Lu
H
,
Buikhuisen
J
,
Soh Boon
S
,
Lim
E
,
Reinhardt
F
, et al
Protein kinase Cα is a central signaling node and therapeutic target for breast cancer stem cells
.
Cancer Cell
2013
;
24
:
347
64
.
39.
Ashworth
A
. 
A synthetic lethal therapeutic approach: poly(ADP) ribose polymerase inhibitors for the treatment of cancers deficient in DNA double-strand break repair
.
J Clin Oncol
2008
;
26
:
3785
90
.
40.
BiPar Sciences presents interim phase 2 results for PARP inhibitor BSI-201 at San Antonio Breast Cancer Symposium
.
Cancer Biol Ther
2009
;
8
:
2
3
.
41.
Liu
JF
,
Tolaney
SM
,
Birrer
M
,
Fleming
GF
,
Buss
MK
,
Dahlberg
SE
, et al
A Phase 1 trial of the poly(ADP-ribose) polymerase inhibitor olaparib (AZD2281) in combination with the anti-angiogenic cediranib (AZD2171) in recurrent epithelial ovarian or triple-negative breast cancer
.
Eur J Cancer
2013
;
49
:
2972
8
.
42.
Hutchinson
L
. 
Targeted therapies: PARP inhibitor olaparib is safe and effective in patients with BRCA1 and BRCA2 mutations
.
Nat Rev Clin Oncol
2010
;
7
:
549
.
43.
Dent
RA
,
Lindeman
GJ
,
Clemons
M
,
Wildiers
H
,
Chan
A
,
McCarthy
NJ
, et al
Phase I trial of the oral PARP inhibitor olaparib in combination with paclitaxel for first- or second-line treatment of patients with metastatic triple-negative breast cancer
.
Breast Cancer Res
2013
;
15
:
R88
.
44.
Ibrahim
YH
,
Garcia-Garcia
C
,
Serra
V
,
He
L
,
Torres-Lockhart
K
,
Prat
A
, et al
PI3K inhibition impairs BRCA1/2 expression and sensitizes BRCA-proficient triple-negative breast cancer to PARP inhibition
.
Cancer Discov
2012
;
2
:
1036
47
.
45.
Ince
TA
,
Richardson
AL
,
Bell
GW
,
Saitoh
M
,
Godar
S
,
Karnoub
AE
, et al
Transformation of different human breast epithelial cell types leads to distinct tumor phenotypes
.
Cancer Cell
2007
;
12
:
160
70
.
46.
Petrocca
F
,
Altschuler
G
,
Tan
SM
,
Mendillo
ML
,
Yan
H
,
Jerry
DJ
, et al
A genome-wide siRNA screen identifies proteasome addiction as a vulnerability of basal-like triple-negative breast cancer cells
.
Cancer Cell
2013
;
24
:
182
96
.
47.
Speers
C
,
Tsimelzon
A
,
Sexton
K
,
Herrick
AM
,
Gutierrez
C
,
Culhane
A
, et al
Identification of novel kinase targets for the treatment of estrogen receptor-negative breast cancer
.
Clin Cancer Res
2009
;
15
:
6327
40
.
48.
Borner
C
,
Wyss
R
,
Regazzi
R
,
Eppenberger
U
,
Fabbro
D
. 
Immunological quantitation of phospholipid/Ca2+-dependent protein kinase of human mammary carcinoma cells: inverse relationship to estrogen receptors
.
Int J Cancer
1987
;
40
:
344
8
.
49.
Tonetti
DA
,
Chisamore
MJ
,
Grdina
W
,
Schurz
H
,
Jordan
VC
. 
Stable transfection of protein kinase C alpha cDNA in hormone dependent breast cancer cell lines
.
Br J Cancer
2000
;
83
:
782
91
.