Background
Breast cancer is the most frequent malignancy among women worldwide [
1]. The status of the expression of the estrogen receptor (ER), progesterone receptor (PR), and human epidermal growth factor receptor 2 (HER2) are the most important prognostic markers for invasive breast cancer [
2]. Triple-negative breast cancer (TNBC) cells do not express ER, PR, or HER2 [
3]. TNBC accounts for approximately 15–26% of breast cancer cases worldwide [
4‐
7]. The survival of patients with TNBC is shorter compared with that of patients with other breast cancer subtypes because of the unique genotype and clinical behavior of TNBC [
3]. TNBCs are more likely to be aggressive and have a higher tendency to metastasize to visceral organs. Patients with TNBC do not benefit from endocrine therapy or from anti-HER2 antibody therapy [
8]. Moreover, the chemosensitivity of TNBCs is limited. Despite clinical trials, an efficacious treatment for patients with TNBC is not available [
9,
10].
Antiangiogenic agents such as a anti-vascular endothelial growth factor (VEGF) neutralizing antibody (Avastin, bevacizumab) and inhibitors of VEGF receptor tyrosine kinase activity (sorafenib and sunitinib) are key components of front-line combination regimens for the treatment of various human cancers [
9‐
11]. These agents are used to treat non-small cell lung cancer, renal cell cancer, and hepatocellular carcinoma [
12‐
14] and were used to treat metastatic breast cancers in preclinical and clinical studies [
15]. However, these and other studies indicate that these therapies may have limited efficacy [
16‐
18]. Although these agents inhibit the growth of primary tumors, the responses are usually temporary, and the overall survival of patients is only modestly increased [
19]. Further, when antiangiogenic agents are administered intermittently, for example, sunitinib (4 weeks on and 2 weeks off), tumor regrowth is sometimes observed during drug-free periods or upon termination of treatment [
20,
21].
Given the limited effect of such treatments, several clinical trials of sunitinib or bevacizumab to treat breast cancer were terminated. One study reported increased tumor invasiveness and metastasis after using VEGF inhibitors or inactivating VEGF gene expression in mouse models of cancer [
22]. These reports suggest that the rationale and prospects of antiangiogenic therapies for breast cancer treatment must be re-evaluated. Because of this, we asked two questions as follows: 1. What is the mechanism of antiangiogenic treatment failure? 2. Is there any difference in the responses to anti-VEGF agents of patients with TNBC or non-TNBC?
In 1999, Maniotis
et al. reported the discovery of vasculogenic mimicry (VM), a vascularization of malignant tumors [
23]. VM channels are formed by tumor cells but not by endothelial cells. VM occurs in many aggressive tumors such as melanoma, inflammatory breast carcinoma, prostate carcinoma, ovarian carcinoma, hepatocellular carcinoma, and gastrointestinal stromal tumors [
24‐
28]. Tumors with VM are more aggressive, and patients have a poorer prognosis than those without VM.
We proved that hypoxia induces VM, and uncovered evidence that cancer stem cells (CSCs) may play an important role in VM [
29,
30]. Moreover, administration of antiangiogenic agents induces intratumoral hypoxia, and hypoxia increases the number of CSCs in cell lines derived from glioblastomas and breast cancers [
31]. Based on these results, we hypothesized that intratumoral hypoxia induced by antiangiogenic agents accelerates VM channel formation in TNBC by increasing the population of CSCs, which in turn, causes tumor regrowth, metastases, and treatment failure using antiangiogenic agents. This hypothesis is supported by the results of the present study that includes an analysis of human patients with TNBC and non-TNBC as well as studies conducted in
vivo and in
vitro using mice that develop spontaneous TNBC and nude mice engrafted with human breast cancer cell lines with TNBC and non-TNBC phenotypes.
Discussion
Vasculogenic mimicry occurs in over 10 tumor types [
23,
32] that are characterized as highly aggressive, poorly differentiated, and highly metastatic [
32,
33]. Therefore, patients with tumors characterized by vasculogenic mimicry have poor outcomes compared with those without VM [
24]. Compared with other tumors, TNBCs are larger, higher grade, more aggressive, and they present with lymph node involvement [
4,
7,
34]. We found that VM occurs more frequently in patients with TNBC compared with those with non-TNBC, which strongly supports the conclusion that VM indicates poor prognosis.
Hendrix
et al. proposed that tumor cells with embryonic phenotypes are highly plastic and form VM channels [
33]. The genes that express proteins that contribute to the formation of VM channels are specifically expressed by endothelial and hematopoietic stem cells [
35,
36]. Recent reports indicate that CSCs may be involved in VM in glioblastomas, breast cancers, and hepatocellular carcinomas [
30,
37]. Hepatocellular carcinoma cells that form VM channels express CSC markers such as SOX2 and OCT4 (Sun BC
et al. unpublished observations). Glioma cancer stem cells enriched in the human glioblastoma cell line U87 form VM channels in xenograft transplantation [
37].
Analysis of the gene expression profiles of 587 patients with TNBC shows enrichment of markers specific for stem cells or mesenchymal stem cells [
38]. The epithelial-to-mesenchymal transition (EMT) is important in VM. Moreover, genes encoding proteins associated with the EMT are expressed at high levels in this population of patients [
39]. Moreover, these results suggest that the gene expression fingerprint of TNBC determines the ability of TNBCs to form VM more efficiently compared with non-TNBCs.
Bevacizumab and sunitinib in combination with cytotoxic drugs were administered in phase III trials of patients with metastatic breast cancer, and bevacizumab was evaluated for treating patients with early-stage breast cancer as a neoadjuvant [
40]. The failures of these trials recently provoked several debates regarding the future applications of anti-VEGF agents in breast cancer [
20,
41]. Aside from the failure of anti-VEGF agents to treat breast cancer, the results of clinical and preclinical research show that they have limited efficacy for treating hepatocellular carcinoma, rectal cancer, and glioblastoma [
14]. These treatment failures may be explained as follows [
16,
19]:
(i) some breast cancers are highly angiogenic and express low levels of VEGF;
(ii) numerous proangiogenic growth factors such as PLGF, PDGFBB, and bFGF are present and can be up-regulated to drive angiogenesis when the VEGF pathway is inhibited [
10]; and
(iii) antiangiogenic therapy can increase tumor hypoxia, which induces an increase in HIF-1α expression to levels sufficient to activate genes that encode proteins required for the growth, invasion, and metastasis.
In the present study, sunitinib was administered to TA2 mice engrafted with tumors of the TN phenotype that were derived from TA2 spontaneous breast cancers [
42]. The growth of primary tumors and metastases were typically inhibited by sunitinib treatment, and the survival of the treated mice increased from 40 to 100%. However, the mice suffered from tumor regrowth and metastases when treatment was suspended, which decreased to those of the control group.
Our previous study reveals “three stages of tumor microcirculation” in melanomas [
32]. VM channels, mosaic blood vessels, and endothelial vessels coexist in a malignant tumor and can transform into each other by changes in the tumor microenvironment [
29,
43]. Because VM occurs in TNBC, we investigated the microcirculation patterns of the sunitinib-treated tumors and observed numerous VM channels when endothelial vessels were inhibited by sunitinib. After discontinuing treatment, the number of endothelial vessels increased and were linked to the VM channels.
The VEGF signaling pathway is essential in endothelial cell-dependent angiogenesis. However, VM is independent of VEGF [
32]. For example, when endothelial vessels are blocked by anti-VEGF agents, VM can be triggered to provide blood to promote tumor growth and metastasis. Moreover, VM is responsible for regenerating the endothelial vessels when treatment is discontinued in this study. These results implicate VM in the failure of standard antiangiogenic therapy to kill aggressive tumors. Therefore, devising strategies that combine standard VEGF-targeted therapies or an endothelium-dependent drug with VM-targeted therapies is attracting considerable interest [
44].
A hypoxic tumor microenvironment is the most important inducer of VM [
28,
29,
39]. Consistent with these findings, we show here that sunitinib inhibited endothelial vessels, produced additional hypoxic areas in TNBC-bearing TA2 mice, and increased the number of VM channels. These results are consistent with our previous finding that hypoxia of the ischemic back limb promotes VM in the B16 melanoma mouse model. We also found that CoCl
2-induced hypoxia increased the number of VM-like channels of MDA-MB-231 cells cultured in Matrigel in the present study.
The mechanism responsible for the effects of hypoxia on VM is unknown. The downstream effectors of HIF-1α are associated with angiogenesis, cell proliferation, cell survival, and glucose/iron metabolism. Hypoxia is linked to increased numbers of CSCs in glioblastoma and breast cancer, and CSCs are involved in tumor angiogenesis and VM [
31,
45,
46]. The breast cancer cells of TA2 mice expressed increased levels of CD133 under hypoxia after sunitinib treatment. Further, hypoxia induced an increase in the population of CD133
+ MDA-MB-231 cells in
vivo. Moreover, stem cells that formed spheres survived and expressed CD133 under hypoxia. Only the CD133
+ cells formed VM channels in Matrigel after reoxygenation, suggesting that hypoxia accelerates VM by stimulating the CSC population.
We found that the EMT factor Twist1 induced the expression of VE-cadherin to promote VM in hepatocellular carcinoma [
28]. Hypoxia induces an EMT-like phenotype in cancer cells [
47], and HIF-1α regulates the expression of Twist1 by binding directly to the hypoxia-response element (HRE) in the proximal promoter of Twist1. EMT can induce stem-cell generation by normal and tumor cells [
48,
49]. We show here that MDA-MB-231 cells with up-regulated Twist1 expression increased the CSC population after reoxygenation. Therefore, more VM channels were generated by cells cultured in Matrigel.
Hypoxia did not significantly affect the CSC population in cultures of MCF-7 cells, and Twist1 expression was down-regulated after reoxygenation. Inhibiting Twist1 expression by MDA-MB-231 cells caused the loss of VM channels in Matrigel and decreased the number of CD133
+ cells in hypoxic cultures. MCF-7 cells that expressed Twist1 gained the ability to form VM channels and generate CD133
+ cells. These results highlight the complexity of the mechanism that regulates the EMT and the biology of CSCs. Human mammary epithelial cells are transformed by Twist1 and snail, and they exhibit the characteristics of CSCs, including the formation of mammospheres, colonies in soft agar, and tumors in
vivo [
49]. For example, Borgna
et al. found that MCF-7 cells gain mesenchymal features by enriching for CSCs in short-term mammosphere culture [
50]. Whether the EMT regulates CSCs or CSCs regulate the EMT is unknown.
Therefore, hypoxia induced by sunitinib accelerates VM by increasing Twsit1 expression and the population of CSCs in TNBC. This finding may explain the inefficacy of antiangiogenic agents in certain breast cancers. Most important, Twist1 and related signal transduction pathways may serve as targets for treating TNBC.
Methods
Reagents and cell culture
The primary antibodies used in this study are listed in Additional file
2: Table S1. All secondary antibodies were purchased from Zhongshan Golden Bridge Biotechnology Co., Ltd. (Beijing, China). Sunitinib malate (S-8803) was purchased from LC Laboratories (MA, USA). The Hypoxyprobe-1 Kit (
HP1-1000Kit) was purchased from Hypoxyprobe, Inc. (MA, USA). The human breast cancer cell lines MCF-7 and MDA-MB-231 and breast cancer cells of TA2 mice were cultured in RPMI-1640 medium with 10% FBS, 4 mM L-glutamine, and 1% penicillin-streptomycin. Matrigel (BD Bioscience, NY, USA) was diluted with RPMI-1640 medium.
Patient samples
The Tianjin General Hospital Ethics Committee approved the studies of humans. The patients were informed of the aims, methods, and other details of the present study. All clinical investigations were conducted according to the principles stated in the Declaration of Helsinki. We collected samples from 174 patients with breast cancer with detailed pathological and clinical information. All patients underwent surgery and chemotherapy in Tianjin General Hospital from 1997 to 2004. The median age of the patients was 51.0 years (range, 31–74 years). All patients had invasive breast cancer, and axillary node metastases were present in 76. The diameter of the primary tumor in 31 patients was <2 cm and >5 cm in 24 patients. The follow-up period started at the time of the surgery and ended in December 2008.
Tissue microarrays and scoring methods
Formalin-fixed, paraffin-embedded tissues from the patients were analyzed after H&E staining. Specific tissue samples were chosen to create tissue microarrays with 1-mm cores (1.5-mm between cores). Tissue microarrays were analyzed using IHC according to a standard protocol [
43]. Protein expression levels were quantified according to the method of Sun
et al.[
43]. Staining was scored as follows; 0 = undetectable, 1 = weak, 2 = moderate, and 3 = strong. The number of positive cells out of 100 tumor cells per field was visually evaluated and scored as follows: 0 < 10% positive, 1 < 25%, 2 < 50%, and 3 > 50%. The staining index or the sum of the staining intensity and the positive-cell score were used to determine the result for each sample. A sample was defined as positive when the staining index was >1. VM and endothelial vessels were counted at 400× magnification, and the score for each sample was defined as the average of 10 fields-of-view.
TA2 and nude mouse models of TNBC
Tianjin Medical University Ethics Committee approved the protocols for using animals. All steps were carefully administered to protect the welfare of the animals and prevent their suffering. The Tientsin Albino 2 (TA2) mice were provided by the Animal Center of Tianjin Medical University. TA2 mice develop spontaneous breast cancer with the TN phenotype at high incidence (showed in Additional file
3), and we used these tumors to induce tumors in TA2 mice [
42]. Nude mice were purchased from Beijing HFK Bioscience Company. Approximately 1 × 10
6 TA2 breast cancer, MDA-MB-231, and MCF-7 cells were injected subcutaneously into the backs of 6-week-old female nude mice (N = 20 per group, respectively). Tumors were measured every 2 days, and tumor size was calculated using a standard formula (length × width
2 × 0.52). The TA2 mice with breast cancer were administered sunitinib daily when the tumor reached 0.2 cm
3. Sunitinib was administered to nude mice when the size of tumors induced by engrafted MDA-MB-231 and MCF-7 cells was approximately 0.05 cm
3. Sunitinib (60 mg/kg) was administered orally for 8 days, and distilled water was used as the placebo. Survival was closely monitored daily at least three times. All surviving mice were sacrificed 1 week after treatment was terminated. Pimonidazole-HCl was injected intraperitoneally (60 mg/kg) 60 min before the mice were sacrificed. The primary tumors and metastatic sites in the peritoneal cavity, lungs, liver, spleen, kidneys, and femurs were collected, weighed, and fixed with 4% paraformaldehyde (PFA). All organs and tumors were embedded in paraffin, and 5-μm-thick sections were prepared.
Hypoxic cell culture in vivo
MDA-MB-231 and MCF-7 cells were seeded into 96-well plates or on Matrigel-coated slides. Cells were treated with 40 μg/ml CoCl2 in cell culture medium for 30 h, and then the hypoxic medium was removed and replaced with normal medium. After 40 h, the cells or slides were harvested, and images were acquired using an inverted microscope (ECLIPSE TS100, Nikon).
Western blotting
HIF-1α, Twist 1, and VE-cadherin expression was analyzed using western blotting. Lysates were prepared using a buffer containing 1% SDS, 10 Mm Tris–HCl, pH 7.6, 20-μg/ml aprotinin, 20-μg/ml leupeptin, and 1-mM AEBSF. The protein concentration of lysates was measured using the Bradford method. Approximately 20 μg of protein was separated on an 8% SDS-PAGE gel and electroblotted onto a PVDF membrane. After blocking with 5% fat-free milk in TBS-Tween overnight, the membrane was incubated with primary antibodies overnight at 4°C. After washing with TBS-Tween three times, the membrane was labeled with horseradish peroxidase-conjugated anti-goat IgG (1:1,000) for 1 h at room temperature (RT). Blots were developed using a DAB kit, GAPHD was used as an internal control, and the bands were analyzed using a gel imaging system (Kodak).
Fluorescence-activated cell sorting (FACS) analysis
Suspensions of MDA-MB-231 and MCF-7 cells were fixed in 75% cold ethanol, and 106 cells were incubated with anti-CD133-PE antibody solution or isotype control on ice for 15 min before they were washed, resuspended in staining buffer (2% fetal calf serum in PBS), and analyzed using a FACS Accuri C6 (BD Biosciences). Gates were set with isotype controls for each cell so that <1% of the cell population was false-positive. The labeled cells were then analyzed (10,000 events).
Immunohistochemical and immunofluorescence assays of formalin-fixed, paraffin-embedded tissues
Formalin-fixed, paraffin-embedded tissues were sectioned, dewaxed, and rehydrated using graded concentrations of alcohol. Endogenous peroxidase was blocked using 5% goat serum at RT for 10 min. The sections were heated in a microwave oven in citrate buffer for 20 min. The slides were incubated with primary antibodies overnight at 4°C, washed with PBS, and individually incubated with biotin-labeled or FITC-labeled secondary antibodies. The color was developed using DAB. The sections were counterstained with hematoxylin or DAPI and observed using a microscope (80i, Nikon).
Immunohistochemical detection of CD31 and periodic acid Schiff (PAS) double-staining
After immunohistochemical analysis of sections for CD31 expression, the sections were exposed to 1% sodium periodate for 10 min, washed for 5 min in distilled water, and then incubated for 15 min with PAS at 37°C. The sections were counterstained with hematoxylin and observed using a microscope (80i, Nikon).
Immunofluorescence analysis of cells cultured on Matrigel
MDA-MB-231 and MCF-7 cells cultured on Matrigel-coated slides were washed with PBS twice, permeabilized, and fixed in 2% PFA and 0.1% Triton X100 in PBS buffer at 4°C for 30 min. The slides were then washed three times with PBS and incubated with 10% goat serum in PBS. The cells were then incubated with the primary antibodies at 4°C overnight, washed three times with PBS containing 0.1% Tween-20 for 15 min, and incubated with the secondary antibodies for 2 h at RT. The slides were washed with PBS and mounted using a slow-fade Light Anti-fade Kit (Zhongshan Golden Bridge). All matched samples were photographed using a confocal laser scanning microscope (A1, Nikon).
Expression plasmids and twist1 gene silencing
A full-length Twist1 complementary cDNA was amplified using PCR from a library of normal human embryo cDNA digested with XhoI/EcoRI and subcloned into pcDNA3.1 vectors [
28]. The constructs were confirmed by DNA sequencing. A small interfering RNA (siRNA) kit (pGP-Twist1-shRNA) was purchased from GenePharm (Shanghai, China). The target sequence (5′-AAGCTGAGCAAGATTCAGACC-3′ [siTwist1 nucleotides 505–525]) was used to inhibit Twist1 expression in vitro [
28]. A nonsilencing siRNA sequence (target sequence 5′-AATTCTCCGAACGTGTCACGT-3′) was used as a negative control.
Statistical analysis
SPSS version 11.0 (Chicago, IL, USA) was used to evaluate the data. The χ2 test was performed to assess the pathological and clinical characteristics of the TNBC and non-TNBC groups. The survival of these two groups was evaluated using Kaplan–Meier analysis. The two-tailed Student t test was performed to compare the endothelial vessels of the human breast cancers, tumor growth, metastasis, and CD133+ cells between groups. The survival rate of the tumor-bearing mice was evaluated using Kaplan–Meier analysis. Statistical significance was defined as P < 0.05.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made.
The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
BS conception, design, and final approval of manuscript; DZ conception, design, animal experiments, drafting manuscript; XZ design, IHC analysis, PAS/CD31 double-staining, acquiring images; YM and RJ collected patients’ data; QG and XD prepared sections and tissue blocks; JL and XJ animal treatment and data collection; FL western blotting animal experiments; XL and CZ cell culture; RS and JC IF analysis. All authors read and approved the final manuscript.