MCF-7 cells were purchased from the China Center for Type Culture Collection and cultured in DMEM/F12 medium (Gibco; Thermo Fisher Scientific, Inc.) supplemented with 10% FBS (Gibco; Thermo Fisher Scientific, Inc.) and 1% penicillin-streptomycin (Sangon Biotech Co., Ltd.) at 37°C in 5% CO₂ atmosphere. Human umbilical vein endothelial cells (HUVECs; cat. no. ATCC^® CRL-1730™; American Type Culture Collection) supplemented with F-12K medium (cat. no. ATCC 30-2004™; American Type Culture Collection) supplemented with 10% FBS (Gibco; Thermo Fisher Scientific, Inc.), 1% penicillin-streptomycin (Sangon Biotech Co., Ltd.), 0.1 mg/ml heparin (cat. no. H3393; Sigma-Aldrich; Merck KGaA) and 1% endothelial cell growth supplement (cat. no. 354006; BD Biosciences) at 37°C in 5% CO₂. Pre-made lentiviral particles expressing a fusion target of green fluorescent protein (GFP)-red fluorescent protein (RFP)-LC3 (cat. no. JM-1314L204H-S) were purchased from Shanghai Genomeditech Co. Ltd. and transfected into MCF-7 cells according to the manufacturer's protocol. In brief, 1×10⁴ MCF-7 cells/well were plated in a 6-well plate and cultured for 12 h to ensure that the cell density reached 30-50% when the virus was infected the following day. Health Gene transgene reagent (Health Gene Technologies) and 3×10⁸ TU/ml lentivirus were used to transfect MCF-7 cells; 6 µg/ml polybrene was added to promote the transfection. After culture for 24 h, the medium was renewed and cultured for another 8 h, then 1.5 µg/ml puromycin (Sigma-Aldrich; Merck KgA) was added to isolate resistant cells. Then, monoclonal transfected cells were obtained and used for subsequent experiments.

BCSLCs, defined as CD44⁺CD24^−/low MCF-7 cells, were isolated using a Miltenyi Microbead kit (Miltenyi Biotec GmbH) according to the manufacturer's instructions and cultured in stem cell medium, comprising serum-free medium supplemented with 20 ng/ml epidermal growth factor (EGF; Gibco; Thermo Fisher Scientific, Inc.), 20 ng/ml basic fibroblast growth factor (bFGF; Gibco; Thermo Fisher Scientific, Inc.) and 10 ng/ml B27 (Gibco; Thermo Fisher Scientific, Inc.). BCSLCs were grown to microsphere formation at 37°C with 5% CO₂ and observed under an inverted fluorescence microscope (magnification, ×20; Leica DM IL LED Fluo; Leica Microsystems GmbH).

Sorted BCSLCs were cultured for 12 h in stem cell medium supplemented with 10 ng/ml VEGF, and VEGF-induced BCSLC and control BCSLCs were analyzed using flow cytometry. Cells were collected and adjusted to a cell density of 5×10⁵ cells/tube. Cells were centrifuged at 300 × g for 10 min at 4°C, the supernatant was discarded, and cells were resuspended in 100 µl buffer plus the corresponding antibody (1:20), and incubated at 4°C for 8 min in the dark. Subsequently, the cells were washed with 1 ml buffer and centrifuged at 300 × g for 10 min at 4°C. The supernatant was discarded, cells were resuspended in 500 µl buffer, stored in the dark on ice, and flow cytometry was performed within 30 min. The following primary antibodies were used: Fluorescein isothiocyanate (FITC)-conjugated anti-human CD44 (cat. no. 130-113-903; Miltenyi Biotec GmbH), phycoerythrin (PE)-conjugated anti-human CD24 (cat. no. 130-098-861; Miltenyi Biotec GmbH) PE-conjugated anti-human CD31 (cat. no. 130-110-807; Miltenyi Biotec GmbH) and FITC-conjugated anti-human CD105 (cat. no. 130-098-778; Miltenyi Biotec GmbH). Flow cytometry was performed using a FACSCalibur flow cytometer (BD Biosciences). The flow cytometry analysis software used was FlowJo™ v10.6.1 (BD Biosciences).

The growth and differentiation status of BCSLCs in vitro was evaluated by a Matrigel formation experiment. Matrigel (BD Biosciences) was thawed, and the blot was added to a 96-well plate at 40 µl/well, with 3 replicate wells. The 96-well plate was then placed in a 37°C incubator for 45 min. After the gel had coagulated, MCF-7 cells and BCSLCs were seeded onto Matrigel at a density of 10⁴ cells/well. The cells were cultured for 12 h in complete or stem cell medium supplemented with or without 10 ng/ml VEGF. Tube formation was observed under an inverted microscope (magnification, ×10; Olympus Corporation).

A total of 5×10⁴ cells/well were plated in a 6-well plate, and cultured in stem cell medium supplemented with or without 10 ng/ml VEGF. After 3 days, the cell were ~80% confluent and sub-cultured. After passaging, the cells were considered second generation of differentiation. The second-generation cell culture solution was used to detect the NO content. The content of NO in the supernatant was measured using the Griess assay protocol as previously described (18). The NO content in HUVEC culture medium was used as the control.

mRFP-GFP-LC3 lentivirus was purchased from Shanghai Genomeditech Co. Ltd. MCF-7 cells were infected with lentivirus as aforementioned and transfected cells were selected for with 1.5 µg/ml puromycin. mRFP-GFP-LC3-transformed BCSLCs were isolated using a Miltenyi Microbead kit. The selected mRFP-GFP-LC3 double-labeled BCSLCs were seeded at a density of 1×10⁴ cells/well in complete culture medium without VEGF, or stem cell culture solution with or without 10 ng/ml VEGF. The localization of RFP and GFP was observed using a confocal laser scanning microscope (Olympus Corporation) at ×100 or ×600 magnification.

The expression of CD105 was detected via RT-qPCR. Total RNA was extracted from cells using TRIzol extraction reagent (Takara Bio, Inc.). For reverse transcription, a PrimeScript™ Reverse Transcription kit (Takara Bio, Inc.) was used according to the manufacturer's instructions. qPCR was performed using SYBR Premix Ex Taq with Tli RNaseH Plus (Takara Bio, Inc.) according to the manufacturer's instructions. GAPDH was used as the internal reference. The PCR reaction conditions consisted of pre-denaturation at 95°C for 30 sec, followed by 40 cycles of denaturation at 95°C for 5 sec and extension at 59°C for 30 sec. The sequence of the qPCR primers were as follows: CD105, forward 5′-AGA GGA CAG GGG TGA CAA GGT-3′, reverse 5′-AAG TGT GGG CTG AGG TAG AGG-3′; and GAPDH, forward 5′-AGA AGG CTG GGG CTC ATT TG-3′, reverse 5′-AGG GGC CAT CCA CAG TCT TC-3′. Expression was measured using the 2^−∆∆Cq method (19).

For the evaluation of endothelial differentiation, after sorting, cells were seeded at 2×10⁵ cells/well in 6-well plates and divided into four groups: Control group; 10 ng/ml VEGF, 10 ng/ml VEGF + 100 nM rapamycin (RAPA; Sigma-Aldrich; Merck KGaA) or 10 ng/ml VEGF + 400 nM bafilomycin A1 (Baf A1; Abcam). Cells were treated for 24 h, and cell morphology and CD105 expression was evaluated.

Short hairpin (sh)-autophagy related 5 (Atg5) was inserted into a lentiviral vector (pGMLV-SC7 RNAi; Shanghai Genomeditech Co. Ltd.). The sequences of the shRNA were: Forward 5′-GCA GAT GGA CAG TTG CAC ACA C-3′, reverse, 5′-AGG TGT TTC CAA CAT TGG CTC A-3′. The transfection of sh-Atg5 lentivirus into MCF-7 was conducted according to the same protocol as the transfection of GFP-RFP-LC3 lentivirus into MCF-7 cells described above. sh-Atg5 BCSLCs were isolated using a Miltenyi Microbead kit, seeded in 6-well plates at a density of 2×10⁵ cells/well, and cultured for 24 h in stem cell medium with or without 10 ng/ml VEGF.

Western blot analysis was used to detect the expression of autophagy-associated proteins. Cells were seeded at a density of 2×10⁵ cells/well in a 6-well plate and treated according to the experimental design. After the end of the experiment, total protein was extracted using a Whole Protein Extraction kit (cat. no. BC3710; Beijing Solarbio Science & Technology Co., Ltd.) and the protein concentration was determined using a Bicinchoninic Acid Protein Assay kit (cat. no. PC0020; Beijing Solarbio Science & Technology Co., Ltd.). Proteins (20 µg/lane) were separated via 12% SDS-PAGE and transferred to a PVDF membrane. Membranes were blocked with 5% skimmed milk for 1 h at room temperature, and subsequently incubated with primary antibodies (1:1,000) at 4°C overnight and incubated for 1 h at room temperature the following day. After washing with TBS-0.05% Tween-20 (TBST), the membrane was incubated with secondary antibody (1:1,000) for 2 h at room temperature. The membrane was washed thoroughly with TBST and signals were visualized using enhanced chemiluminescence reagent (Beijing Solarbio Science & Technology Co., Ltd.). Grayscale analysis was performed using Gel-Pro Analyzer V.4.0 (Meyer Instruments, Inc.). The primary antibodies used were β-actin (cat. no. 4970), LC3 (cat. no. 4108), p62 (cat. no. 5114), Beclin1 (cat. no. 3495), Atg3 (cat. no. 3415), Atg5 (cat. no. 12994), Atg7 (cat. no. 8558), Atg16L1 (cat. no. 8089), Unc-51-like autophagy activating kinase (ULK; cat. no. 8054), phosphorylated (p)-ULK (cat. no. 5869), AMP kinase (AMPK; cat. no. 5832) and p-AMPK (cat. no. 2535); the secondary antibody used was a horseradish peroxidase-conjugated goat anti-rabbit immunoglobulin G antibody (cat. no. 7074). All antibodies were purchased from Cell Signaling Technology, Inc.

To monitor autophagy flux during endothelial differentiation, after sorting, cells were seeded at 2×10⁵ cells/well in 6-well plates and divided into three groups: 0 h VEGF; 24 h VEGF (10 ng/ml); and 24 h VEGF + 1 h chloroquine (CQ; 50 µM; Sigma-Aldrich; Merck KGaA).

Results are presented as the mean ± standard deviation of at least three independent experiments. Data were analyzed by t-test, or by one-way ANOVA followed by Tukey's test post hoc analysis. Statistical analysis was performed using Graph-Pad Prism 6.01 software (GraphPad Software, Inc.). P<0.05 was considered to indicate a statistically significant difference.

Cancer stem cells (CSCs) are a type of tumor cell that are capable of self-renewal and differentiation (20). A previous study showed that CSCs can survive in serum-free medium containing bFGF, EGF, B27 and other growth factors, and gradually form microspheres during culture (21). Cells characterized as CD44⁺CD24^−/low are considered to be breast cancer stem cells. First, CD24^−/low cells were isolated by sorting using immunomagnetic beads, and the expression of CD24 was detected via flow cytometry. As presented in Fig. 1A, the proportion of cells characterized as CD24⁺ in MCF-7 was 53.69 and 46.28% were cells characterized as CD24^−/low; the two cell populations had notable partitions (Fig. 1Ab). The proportion of CD24^−/low cells obtained using immunomagnetic beads was 99.61%, indicating that the purity of the obtained CD24^−/low type cell was high and could be used for the following isolations (Fig. 1Ad). CD44⁺ cells were further sorted from the isolated CD24^−/low cells using immunomagnetic beads, and the results of flow cytometry showed that the proportion of CD44⁺ cells was 99.57% (Fig. 1B). The obtained CD44⁺CD24^−/low cells were considered BCSLCs, and the obtained BCSLCs were identified using flow cytometry; the results showed that the proportion of CD44⁺CD24^−/low cells in BCSLCs was 41.62% (Fig. 1Cc), considerably higher than the proportion of cells obtained from the MCF-7 cells (1.33%; Fig. 1Cb). Therefore, the CD44⁺CD24^−/low cells were enriched using immunomagnetic beads. The obtained BCSLCs were sub-cultured, and the proportion of CD44⁺CD24^−/low cells decreased to 10.93% when cultured up to passage 8 (Fig. 1Ce).

Microspheres gradually formed when BCSLCs were cultured in serum-free medium containing bFGF and EGF. The obtained BCSLCs with CD44⁺CD24^−/low specificity were cultured in stem cell culture medium, and the results showed that microspheres gradually formed when the cells were cultured for 64 h or more (Fig. 1D), suggesting that the obtained BCSLCs exhibited characteristics associated with cancer stem cells.

VEGF was used as an inducer of endothelial differentiation. MCF-7 and BCSLCs were seeded on Matrigel and cultured for 12 h in stem cell medium with or without VEGF to mimic the growth and differentiation of cells in vivo. As presented in Fig. 2A, after incubation with VEGF-containing stem cell culture medium, BCSLCs formed a tubular structure, whereas MCF-7 cells did not.

NO, which is an endothelium-dependent relaxation factor, is secreted by vascular endothelial cells expressing active NO synthase (22). The release of NO can be used as an indicator of endothelial cell viability (23). The results showed that BCSLCs treated with VEGF released significantly higher quantities of NO compared with cells not treated with VEGF (Fig. 2B). The quantity of NO released by induced BCSLCs was higher compared with the levels released by HUVECs. Thus, BCSLCs exhibited a property typically associated with endothelial cells when induced by VEGF.

CD31 and CD105 are used as markers of endothelial cells (24). The BCSLCs were cultured with or without VEGF for 12 h, and the expression of CD105 and CD31 was assessed using flow cytometry. The results showed that in induced cells, the proportion of cells expressing CD105 and CD31 was 78.45 and 50.44%, respectively, significantly higher compared with cells not treated with VEGF (Fig. 2C). These results suggested that after being induced by VEGF, BCSLCs expressed endothelial cell markers.

Autophagy is a dynamic physiological metabolic process in the cell. Detection of autophagy flow can more accurately reflect the changes of autophagy metabolism (25). Chloroquine (CQ) is used as an autophagy inhibitor, as it can alter the pH of the lysosome to abrogate lysosomal function, and inhibit binding of the autophagosome to the lysosome, thus blocking autophagy and resulting in an accumulation of autophagosomes (26). To monitor autophagy flux during endothelial differentiation of BCSLCs, CQ was used to inhibit autophagy, and the expression of Beclin1, p62 and LC3-II was detected via western blotting. The results showed that the expression levels of Beclin1, p62 and LC3-II were significantly increased when the BCSLCs were induced by VEGF for 24 h (Fig. 3A), suggesting that autophagy was increased. When the BCSLSs were treated with CQ for 1 h following treatment with VEGF, the expression of LC3-II and p62 was increased significantly, and the expression of Beclin1 was decreased significantly compared with cells not treated with CQ (Fig. 3A). Treatment with CQ would result in the accumulation of autophagosomes, which would reduce expression of upstream Beclin1 expression via negative feedback regulation (27). The accumulation of LC3-II reflects the increase in autophagy flux when induced by VEGF.

MCF-7 cells stably expressing mRFP-GFP-LC3 were created to further examine the association between autophagy and endothelial differentiation of BCSLCs. Fig. 3B shows the sequencing results. The sequence of the inserted fragment in the recombinant clone was completely identical to the sequence of the target fragment, and thus the construction of the overexpression vector was successful. Fluorescent signals were detected after lentiviral infection, and GFP and RFP were visible, demonstrating successful transfection of mRFP-GFP-LC3 in MCF-7 cells (Fig. 3B). BCSLCs were isolated from the RFP-GFP-LC3-MCF-7 cells, and cultured in complete medium, or stem cell medium with or without VEGF for 24 h, and the cells were observed under a confocal laser scanning microscope (Fig. 3C). After 24 h of cell culture, the green fluorescence intensity of the VEGF-containing group was lower compared with stem cell culture cells. There was an increase in the number of red and yellow spots in the cells cultured in the VEGF-containing stem cell culture medium compared with cells cultured in the complete medium. LC-3 has two forms, LC3-І and LC3-II. LC3-І is diffusely distributed in cells, and LC3-II accumulates on the inner and outer membranes of autophagic vesicles, including autophagy precursors, and autophagosomes, making these autophagic structures appear green in the transfected cells. When autophagosomes fuse with lysosomes, the green fluorescence should be quenched due to the low pH, but the red fluorescence should remain stable (28). Therefore, RFP was used to track LC3, and the weakening of green fluorescence was used to indicate the fusion of autophagosomes and lysosomes. Autophagy flux of BCSLCs in stem cell culture medium containing VEGF was markedly higher compared with cells grown in stem cell culture medium or complete medium. BCSLCs were induced into a spindle shape akin to the typical shape of vascular endothelial cells when treated with VEGF. These data suggested that the initiation of transdifferentiation of BCSLCs was accompanied by an increase in autophagic flux.

Proteins associated with autophagy signaling pathways were detected by western blotting to further determine the changes in autophagy levels during the endothelial differentiation of BCSLCs. The results showed that treatment with VEGF significantly increased the expression levels of ULK1, p-ULK1, AMPK and p-AMPK, all of which are upstream of the autophagy signaling pathway; the expression levels of key proteins involved in the steps between autophagy formation to autophagic lysosome degradation, such as Beclin1, Atg5, Atg16L1, Atg7, Atg3 and p62, were also increased compared with BCSLCs (Fig. 3D). There results suggested that autophagy levels were increased overall during endothelial differentiation of BCSLCs induced by VEGF. The analysis of ULK1, p-ULK1, AMPK, and p-AMPK showed that the ratios of p-AMPK/AMPK and p-ULK1/ULK1 were also increased significantly (Fig. 3Dc), indicating that the AMPK protein was activated, which in turn affects downstream ULK1 expression and phosphorylation, thus initiating autophagy (25).

In order to investigate the association between autophagy and endothelial differentiation of BCSLCs, the autophagy activator RAPA and inhibitor Baf A1 were used to treat BCSLCs separately. As shown in Fig. 4A, cells initially showed an elongated spindle-like shape, similar to endothelial cells when treated with VEGF. When the cells were treated with VEGF and RAPA, the cellular morphology was also elongated and spindle-like, but when the cells were treated with VEGF and Baf A1, the cellular morphology was cobblestone-like similar to that of BCSLCs without VEGF. The results suggest a positive association between autophagy and endothelial differentiation of BCSLCs. Expression of CD105 was assessed using RT-qPCR, and the results showed that treatment with VEGF or VEGF + RAPA increased expression of CD105 significantly compared with BCSLCs not treated with VEGF, and there was no significant difference in CD105 expression levels between these two groups. However, the expression of CD105 was decreased significantly when BCSLCs were treated with VEGF + Baf A1 compared with the VEGF group (Fig. 4B). These results were in agreement with the cellular morphology experiments, and a positive association between autophagy and the endothelial differentiation of BCSLCs was further suggested.

As Atg5 is one of the core proteins involved in autophagosome formation, an Atg5-knockdown MCF-7 cell line was established to further investigate if autophagy was critical for endothelial differentiation of BCSLCs. sh-Atg5-BCSLCs were isolated from sh-Atg5-MCF-7 and were induced with VEGF for 24 h, and the expression of Atg5 and LC3-II was assessed via western blotting. As shown in Fig. 5A, after being induced with VEGF, the expression levels of Atg5 and LC3-II were similar in the control and vector groups, but significantly downregulated in the sh-Atg5 group. This indicated that autophagy was inhibited in sh-Atg5-BCSLCs. When the sh-Atg5-BCSLCs and vector-BCSLCs (negative control) were cultured with 1 ng/ml VEGF for 24 h, the vector-BCSLCs exhibited an elongated spindle-like morphology similar to endothelial cells; however, the majority of the sh-Atg5-BCSLCs retained their cobblestone-like morphology (Fig. 5B). CD105 mRNA expression levels were measured and the results showed that the expression of CD105 was significantly decreased in sh-Atg5-BCSLCs following treatment with VEGF compared with the vector-BCSLCs (Fig. 5C). All the above results suggested that autophagy is required for effective endothelial differentiation of BCSLCs.