Background
Although remarkable advances have been made in the evolution of therapeutic strategies and molecular elucidation of breast malignancies, recurrence and distant metastasis are still remained hard to cure and poorly understood [
1,
2]. In 2016, it was reported that there will be 249,260 new breast cancer cases, of which 40,890 cases will be lethal to the U.S. female population [
3]. Metastasis accounts for over 90% of the cases of mortality in cancer patients [
4], and women with distant metastatic lesions have a significantly decreased 5-year survival rate compared with women with localized breast tumors (26.3% versus 98.8%) [
5]. Strikingly, only 6% of the patients were initially diagnosed with metastatic breast cancer, and 20% to 50% of primary breast cancer patients would eventually develop invasive phenotypes [
2]. Although overwhelming experimental evidence suggests that metastasis usually occurs at the advanced stages of cancer, some studies implied that tumor cells could be detected in the circulatory system as early as breast cancer initiation [
6,
7]. However, the phenotype of circulating tumor cells is not necessarily uniform, only a minority of cancer cells could migrate to the distant organs and finally form the metastatic lesions [
8,
9]. Hence, the identification of the metastatic cells that are capable of forming metastatic nodules and seeking targeted therapies to eliminate early metastasis are urgently needed in clinical management.
The discovery of CSCs sheds novel light in limiting metastasis. The CSC hypothesis supports the belief that cancer is largely driven by a subpopulation of malignant cells with the distinct ability of self-renewal, heterogeneous population generation, and tumor initiation [
10]. Further, mounting studies suggest that CSCs exhibit high metastatic potential.
Weng D et al. revealed that early metastatic cells that disseminated in the lungs displayed stem cell markers in the MMTV-PyMT mice model [
9]. Liu et al. compared the gene expression differences between breast CSCs and normal mammary epithelial cells. There were 186 “invasive” genes that were identified, and they revealed a tight correlation to the overall survival and metastasis-free survival in breast cancer and other malignancies, including lung carcinoma, medulloblastoma, and prostate cancer [
11].
Balic M et al. identified the existence of the stem-like cells within the bone marrow of breast cancer patients in the early phase [
12]. All of these findings suggest that CSCs might be the precursors guiding cancer metastasis, and identification of key targets mediating CSCs metastasis is of great significance for improving cancer prognosis.
Target identification with natural phytochemicals has become an important strategy leading to the novel discovery of pathological biomarkers nowadays. Although there still is not a specific CSCs-targeting drug in the clinic at present, a lot of naturally occurring compounds have exhibited promising CSCs-limiting effects. Meanwhile, dietary supplements are specifically favored due to their wide safety profile, focus on multi-targets, and economic applications [
13]. EA is one of the representative dietary polyphenols widely found in fruits and vegetables. Our previous study indicated that EA had significant suppressive effects on breast cancer growth and neoangiogenesis [
14]. Meanwhile, extensive studies also proved EA to be an anti-cancer compound with multiple functions such as antioxidant, apoptosis induction, carcinogen-DNA binding blockage, and metastasis inhibition [
15,
16]. However, little investigation has been carried out to explore its CSCs-limiting effects and the precise molecular target involved.
In the current study, we aimed to determine the direct molecular target of EA and its potential regulation effects on CSCs metastasis. DARTS analysis identified ACTN4 as the direct target of EA, and ACTN4 was found to closely correlate with CSCs’ self-renewal and metastatic abilities, poor survival period, and basal-like phenotype. Mechanism exploration revealed that interruption of ACTN4/β-catenin interaction will result in the activation of β-catenin proteasome degradation. Our findings would not only facilitate present understanding on the critical role of ACTN4 in breast CSCs metastasis, but also benefit for designing potentially more effective therapeutic strategies against breast cancer.
Methods
Chemicals and reagents
EA (over 97% purity) was provided by Alpha Aesar Company (Alfa Aesar, WardHill, MA). Dimethylsulfoxide (DMSO) was used to prepare the stock solution of EA and kept at −20 °C. 3-(4, 5-dimethylthiazol −2-yl) -2, 5-diphenyltetrazolium bromide (MTT), bovine serum albumin (BSA), insulin, 4′,6-diamidino-2-phenylindole (DAPI),
lithium chloride (LiCl), Akt inhibitor LY294002, cycloheximide (CHX) and proteasome inhibitor MG132, and were all purchased from Sigma (St. Louis, MO, USA).
Cell culture
The human breast cancer cell lines MCF-7, BT-549, MDA-MB-231, SKBr3, MDA-MB-453, T47D, BT-474, HCC1937, and MCF-10A were purchased from the American Type Culture Collection (Manassas, VA, USA). The cells were cultured in medium (L15 for MDA-MB-231; RPMI-1640 for BT-549 and MCF-7; DMEM for SKBr3, MDA-MB-453, T47D, BT474, and HCC1937; and F12 for MCF-10A) at 37 °C in a humidified incubator with or without 5% CO2. 10% FBS and 1% penicillin and streptomycin (Gibco Life Technologies, Lofer, Austria) were added as supplements. The CSCs were sorted from MDA-MB-231 and BT-549 cells and maintained in DMEM/F12 medium. B27 (Invitrogen, Carlsbad, CA, USA), 5 μg/ml of insulin, 20 ng/ml of hEGF (BD Bioscience, Bedford, MA, USA), 1% penicillin/streptomycin and 0.4% BSA were added to DMEM/F12 medium for in vitro propagation.
Breast cancer mice models and EA intervention
All animal procedures were performed in accordance with the institutional guidelines for the care and use of laboratory animals approved by the Animal Care and Use Committee of Guangdong Provincial Hospital of Chinese Medicine (the Ethics Approval Number 2015008) and the National Institutes of Health guide for the care and use of Laboratory animals. For MMTV-PyMT transgenic mice, 3-week-old female mice were randomly divided into the vehicle and the EA treatment groups, and EA (50 mg/kg/d) was given by intragastric administration from the Weeks 3–18 (n = 10 mice, total of 100 glands). For xenograft models, 106 cells of MDA-MB-231 or BT-549 with or without genetic alternations were subcutaneously implanted into the mammary fat pads of female NOD/SCID mice at 4-week-old. After the tumor size reached over ~5 × 5 mm, EA (50 mg/kg/d) was given by intragastric administration for an additional 4 weeks. We monitored tumor formation and tumor size twice a week, and removed the tumors at the end-point of experiments for additional studies.
India ink assay
Pulmonary metastatic lesions were imaged and calculated by India ink staining (15% India ink, 85% water, 3 drops of NH4OH/100 ml) through intra-tracheal injection. Feket’s solution (300 ml of 70% EtOH, 30 ml of 37% formaldehyde, and 5 ml of glacial acetic acid) were then prepared and washed the ink-stained lungs overnight to develop the white tumor nodules against a blue lung background.
Immunohistochemistry and Immunofluorescence analysis
Tumor samples were paraffin-embedded and cut into 4 μm sections, and subsequently mounted onto poly-L-lysine-coated slides for immunohistochemistry analysis. The sections were then treated with xylene twice for 10 min and rehydrated with ethanol from 100% to 70% gradually and finally immersed in distilled water. H&E staining was applied to identify the lung metastatic lesions. For immunohistochemistry analysis, the sections were firstly treated with methanol (with 0.3% hydrogen peroxide) for 30 min to inactivate the endogenous peroxide at room temperature. Antigen retrieval was performed by heating the slides in sodium-citrate buffer, followed by permeabilization with 0.2% Triton X-100 for 15 min. After blocking with 10% goat serum, the sections were incubated with indicated primary antibody ACTN4 (Abcam, Cambridge, USA) at 4 °C overnight. DAB detection system (Dako A/S, Glostrup, Denmark) was applied as chromogenic agents according to the manufacturer’s instructions. Finally, sections were counterstained using Mayer’s hematoxylin, dehydrated, cleared, and mounted before examination. With regard to cellular immunofluorescence detection, 4% paraformaldehyde and 0.2% triton X-100 were firstly administrated to cells for 10 min. Following goat serum blocking for 60 min, the samples were co-incubated with primary antibodies including ACTN4 (Abcam, Cambridge, USA) and β-catenin (Cell Signaling Technology, Beverly, MA, USA) at 4 °C overnight, and subsequently labeled with fluorescence-conjugated secondary antibodies for 2 h at room temperature. 4′,6-diamidino-2-phenylindole (DAPI) was finally applied for nuclear staining and the signals were detected with NIKON TS2R fluorescence microscopy.
Flow cytometry analysis
Single cell suspension were prepared and suspended in wash buffer (PBS containing 1% FBS) at a density of 107/ml. Aldehyde dehydrogenase-based cell detection kit (Stem Cell Technologies, Grenoble, France) was used to detect the subpopulation of CSCs. Briefly, Aldehyde dehydrogenase (ALDH) substrate (Bodipy-aminoacetaldehyde) was co-incubated with 106 cells for 45 min at 37 °C. Diethylaminobenzaldehyde (DEAB), an ALDH1 enzyme inhibitor, was added as a negative control. ALDH fluorescence was detected by a 488-nm blue laser. For CD44+/CD24− detection or sorting assay, the cells were incubated with primary antibodies CD44-FITC, CD24-PE, and/or ACTN4-APC (BD Biosciences, San Diego, CA, USA) at 4 °C for 40 min. After washing with PBS, the cells were re-suspended and analyzed with FACSAria SORP (BD Biosciences, San Jose, CA, USA). FlowJo software was applied for data analysis.
Cell proliferation and colony formation assays
MTT assay was conducted according to the manufacturer’s instructions. EA-treated or gene-modified cells were also seeded into 6-well plates at a density of 1000 cells/well to form colonies. After 2 weeks, the colonies were stained with Coomassie Blue and counted.
Wound healing and transwell migration assays
For wound healing assay, the wound gap was recorded at 0, 24, or 48 h after the scratch made with a 10-μl pipette tip. Transwell assay was examined using 8-mm pore Transwell chambers (Milipore, Billerica, MA, USA).
The sorted breast CSCs were subjected into mammosphere formation assay at a density of 1000 cells/well in ultra-low attachment plates. For tumorigenic assay, a series dilution of CSCs (105, 104, 103) were injected into the mammary pads of NOD/SCID mice to compare the tumorigenic ratio and tumor initiating cell frequency calculated by Extreme Limiting Dilution Analysis (ELDA) software.
Western blotting and co-immunoprecipitation analysis
Total protein extraction and western blot analysis were performed as described previously [
17]. Antibodies for western blotting included ACTN4 (Abcam, Cambridge, USA), vimentin, E-cadherin (Abclonal, Cambridge, MA, USA), β-catenin,
p-β-catenin (Ser33/37/Thr41), Akt,
p-Akt (Ser473), GSK3β,
p-GSK3β (Ser9), lamin B, β-actin (Cell Signaling Technology, MA, USA) as well as secondary anti-rabbit or anti-mouse antibodies. The interaction between ACTN4 and β-catenin were analyzed by co-immunoprecipitation analysis according to the instruction provided by Dynabeads Protein G immunoprecipitation kit (Invitrogen, Carlsbad, CA, USA).
Darts
DARTS strategy was applied to identify the precise protein target of EA according to the protocol provided by
Lomenick et al. [
18]. Protein lysates from CSCs were quantified using bicinchoninic acid protein assay. Different concentrations of EA were then co-incubated with the lysates for 60 min at room temperature. The drug-protein interaction system was then treated with pronase (1:50, Roche Applied Science) to undergo proteolysis for 30 min at 4 °C. The lysates were finally denatured and subjected to western blotting analysis. The protective band was visualized by Coomassie bule staining and identified by MALDI-TOF-MS.
In vitro ubiquitination assay
Ubiquitination detection assay was carried out according to the instruction provided by Ubiquitination Kit (UW9920, BioMol). The prepared cell lysates were firstly incubated with 100 nM E1, 2.5 mM UbcH5a as E2 [
19], 20 U/ml of inorganic pyrophophatase (Sigma-Aldrich), 5 mM dithiothreitol, 5 mM Mg-ATP and 2.5 mM biotin-labelled ubiquitin in a 50 ml reaction system at 37 °C. After 4 h incubation, 50 ml of 2 × non-reducing gel-loading buffer was added to quench the reaction and subjected to SDS-PAGE analysis. After the proteins smaller than 70 kDa ran out, the gel was transferred onto PVDF membrane and immunoblotted with β-catenin antibody.
Real-time PCR
TRIzol reagent (Invitrogen, Carlsbad, CA, USA) was applied to extract the total RNA, followed by reverse transcription reaction using the first-strand cDNA synthesis kit (Roche, Mannheim, Germany). SYBR Green kit (Roche, Mannheim, Germany) was utilized to perform real-time PCR analysis on Roche LightCycler 480 detector. PCR reaction condition was set as 95 °C for 10 min followed by 40 cycles of 95 °C for 10 s, 55 °C for 30 s, and 72 °C for 1 min. The target gene expression was calculated by 2
-△△Ct and normalized to the housekeeping gene control. The primers’ sequences were listed in Additional file
1: Table S1.
Plasmids and siRNA construction and transfection
The pENTER vector plasmid carrying ACTN4-cDNA cloning and shACTN4 were purchased from Vigene Biosciences (Jinan, China) and transfected into cells using LipoFiter™ reagent (Hanbio Biotechnology Co, LTD. Shanghai, China). Scrambled plasmids were set as negative control. ACTN4 siRNA and their scrambled ones were bought from Invitrogen (Carlsbad, CA, USA) and transfected by X-tremeGENE siRNA transfection reagent (Roche Diagnostics, IN). The target protein expression was confirmed by western blotting. For the TCF/LEF luciferase assay, the TOPFLASH or FOPFLASH plasmids (Promega, Madison, WI) were transiently transfected into cells with or without EA treatment. pRL-TK plasmid was co-transfected for normalizing the transfection efficiency.
RNA sequencing
RNA preparation, library construction and sequencing on BGISEQ-500 platform was performed at Beijing Genomics Institution (
www.genomics.org.cn, BGI, Shenzhen, China). Genes expression levels were quantified by a software package called RSEM [
20]. NOISeq method was used to screen differentially expressed genes (DEGs) between groups. Gene Ontology (GO) and pathway annotation and enrichment analyses were based on the Gene Ontology Database (
http://www.geneontology.org
/) and KEGG pathway database (
http://www.genome.jp/kegg/), respectively. Statistical analysis was performed and DEGs were selected with the criteria of fold change ≥ 1.2,
P ≤ 0.05.
Statistical analysis
Data analysis was performed with SPSS 13.0 software. The data were expressed as mean ± SD. Student’s t-test was used to compare the statistical difference between groups. The significance of multiple groups was compared using the one-way analysis of variance (ANOVA) followed by the Dunnett’s post hoc test. A value of P < 0.05 was considered significant.
Discussion
DARTS strategy is a novel drug target identification system based on the susceptibility difference to proteolysis between single drug and drug-protein complex [
23]. Compared with other affinity-based target identification methods, the key advantage of DARTS is that it does not require ligand modification. Therefore, DARTS is not limited by chemical structure. Here, we applied the DARTS technique to identify ACTN4 as the direct bound protein of EA in breast CD44
+/CD24
− phenotypes. The successful target identification of DARTS strategy is dependent on two factors: the target of the small molecule should be highly abundant in cells, and the identified protein should not be extremely sensitive or resistant to the proteases applied [
18]. This indicates that ACTN4 should be a highly abundant protein in breast CSCs, and would be strongly protected by EA from proteolysis and resulted in detectable differences presented as clear variable bands in Fig.
3A. In other words, ACTN4 is one of the most abundant and important targets of EA in breast CSCs, and this is not to exclude the existence of any other possible targets of EA in cancer cells. According to literature reports, EA had inhibition effects on multiple targets of cancer cells, such as VEGFR-2 [
14], STAT3 [
28], TGF-β [
29], and NF-κB [
30], etc. However, this is the first study to demonstrate the direct target of EA in cancer cells, and a more comprehensive strategy, such as network pharmacology, might be used to establish the anti-cancer network signaling of EA in the future.
ACTN4, an actin-binding protein, has been described to exist in at least 2 different subcellular locations: the cytosol and nucleus. Shao H et al. proposed ACTN4 was largely responsible for the spreading, motility, and contractility of fibroblasts [
31]. Additionally,
Honda K et al. demonstrated its potent ability to increase cell motility and promote lymph node metastasis in colorectal cancer [
32]. Consistent with their findings, abnormal ACTN4 expression was also correlated to increased tumor invasiveness and metastasis in breast, esophageal, pancreatic, ovarian, and lung carcinomas, indicating that actinin-4 is a promising biomarker for cancer invasion and predictive indicator for patients with metastatic cancer diseases [
33‐
36]. Several studies also identified the crucial role of ACTN4 in transcriptional regulation. ACTN4 transcriptionally potentiated myocyte enhancer factor-2 by antagonizing histone deacetylase [
37]. ACTN4 could also serve as a NF-κB co-activator by interacting with its RelA/p65 subunit [
38]. More importantly, ACTN4 and E-cadherin might compete for the same binding domain of β-catenin [
39]. ACTN4 was found to promote EMT and tumorigenesis by regulating Snail expression and the Akt pathway in cervical cancer [
40]. However, few studies have revealed the mechanisms of ACTN4-mediated CSC activities and tumor metastasis in breast cancer.
In the present study, we demonstrated at least 4 lines of evidence supporting the crucial roles of ACTN4 in facilitating CSCs and metastasis in breast cancer. On the threshold, we identified ACTN4 as the direct target of EA in breast CSCs using the DARTS strategy. Further validation suggested that EA administration significantly suppressed ACTN4 expression in vitro and in vivo in the breast cancer model, accompanied by CSC suppressive effects. Next, increased expression of ACTN4 predicted poor differentiation, high metastatic potential, and short overall survival and disease-free survival durations. ACTN4 expression was strongly expressed in the TNBC subtype, which is usually enriched for CD44
+/CD24
− CSCs. Thirdly, CD44
+/CD24
−/ACTN4
+ phenotypes showed a greater ESA
+ proportion, increased tumorsphere-formation ability, and stronger in vivo tumorigenesis potential in mice when compared with the CD44
+/CD24
−/ACTN4
− populations. Lastly, ACTN4 sustained breast CSC properties mainly by β-catenin stabilization. Furthermore, Charpentier et al. recently identified that breast stem-like cells display higher levels of microtentacles (McTN), tubulin-based protrusions of the plasma cell membrane that aids cell reattachment [
41]. Given that some studies proposed that ACTN4 was preferentially localized in moving structures, such as dorsal ruffles, lamellipodia, and filopodia [
39], it is of great significance to further investigate the expression of ACTN4 expression in McTNs.
In addition, our study clearly validated that ACTN4 establishes a direct association between CSC phenotype and metastatic breast cancer. There mainly exists 2 different strategies for CSC isolation and identification [
42‐
44]. The first one is based on unique surface markers in CSCs. Fluorescence-activated cell sorting analysis (FACS) and/or magnetic-activated cell sorting analysis (MACS) are designed to isolate CSCs expressing specific biomarkers, such as CD44
+, CD24
−, EpCAM
+, and ALDH
hi, etc. [
45,
46]. The second strategy is to utilize biofunctions of CSCs. CSC-like cells can be selected by their resistance to chemodrugs, EMT induction, or 3-D sphere culture systems [
47,
48]. Nevertheless, functional stem-like cells are not real CSCs and possibly require further purification by additional biomarkers. Our study identified that approximately 100% of the basal-like phenotype MDA-MB-231 cells are CD44
+/CD24
− subpopulations, among which CD44
+/CD24
−/ACTN4
+ cells exhibited better CSC-like properties (high self-renewal properties, adherence-promoting capacity, and tumorigenic potential with very few cells) compared with the unfractioned parts. This finding was consistent with recent studies that showed the presence of CD44
+/CD24
− cells does not correlate with tumorigenicity [
49], distant metastasis or clinical outcome [
26]. Thus, it is reasonable that combined utilization of ACTN4 and CD44
+/CD24
− markers might be more sufficient to isolate metastatic cells with CSC properties in breast cancer.