ANGPTL4 overexpression inhibits tumor cell adhesion and migration and predicts favorable prognosis of triple-negative breast cancer

Patients with TNBC who were hospitalized at the Department of Medical Oncology or the Department of Breast Oncology of Sun Yat-sen University Cancer Center between January 2007 and December 2016 were retrospectively selected. The inclusion criteria were as follows: (1) patients underwent modified radical mastectomy; (2) the diagnosis of TNBC was confirmed by molecular biology and pathology; and (3) patients had complete follow-up information and pathological specimens available. The appropriate pathological samples for each patient were acquired from the Pathology Department. The final follow-up date was December 2019. Overall survival (OS) was defined as the time between diagnosis and death or the last follow-up visit. Disease-free survival (DFS) was calculated from the time of diagnosis to the date of relapse or metastasis.

Paraffin-embedded tissues were cut at 2 mm thickness. The slides were dewaxed and rehydrated, and endogenous peroxidase activity was blocked. The specimens were boiled in ethylenediaminetetraacetic acid (EDTA) at full power for 5 min and medium heat for 20 min for antigen retrieval. Common goat serum was used to suppress nonspecific binding. Then, polyclonal rabbit anti-ANGPTL4 antibody (diluted 1:100; Abcam, #ab115798, USA) and a secondary antibody were used. Subsequently, we applied horseradish peroxidase. Finally, hematoxylin was used to counterstain the nuclei.

All sections were separately evaluated by two independent pathologists. The percentage was determined as follows: 0–5% was scored as 0, 6–25% was scored as 1, 26–50% was scored as 2, and more than 50% was scored as 3. The intensity was calculated as follows: 0 score for no staining, 1 score for weak staining (light yellow), 2 score for yellowish brown staining, and 3 score for brown staining. The scores of proportion and intensity were added to obtain an overall score, which ranged from 0 to 6 [17]. A receiver operating characteristic (ROC) curve was applied to obtain an optimal cut-off score for overexpression of ANGPTL4 using the 0, 1 criterion.

The breast cancer cell lines used in this research were all from the American Type Culture Collection and were identified by DNA (STR) profiling. MDA-MB-231 cells were grown in DMEM (Gibco, #41965–039) with 10% FBS (Gibco, #10270–106). BT549 cells were grown in RPMI-1640 (Gibco, # A1049101) with 10% FBS. These cell lines were cultured at 37 °C under a humidified atmosphere containing 5% CO₂.

The lentivirus carrying the pEZ-Lv105 plasmid encoding the full-length ANGPTL4 ORF sequence (NM_139314.3) and empty vector were purchased from GeneCopoeia. Lentivirus was transfected into cells, and puromycin (A1113803, Invitrogen; Thermo Fisher Scientific, Inc.) selection (10 μg/ml) started 24 h after transfection. Stable ANGPTL4-overexpressing MDA-MB-231 and BT-549 cells (ANGPTL4 OE) and empty vector cells (Vec) were established from isolated colonies and grown for the next assay. Untreated MDA-MB-231 and BT-549 cells were referred to as the negative control group (NC). The efficiency of ANGPTL4 gene transfection was verified by qPCR and Western blotting.

Total cellular protein was lysed by using RIPA buffer (1:10, Cell Signaling Technology, 9806S). The BCA method (Thermo, USA) was used to calculate the protein concentrations. With a Mini Trans-Blot Electrophoretic Transfer Cell System (Bio-Rad), equal amounts of protein were subjected to gel electrophoresis. Proteins were transferred onto a PVDF membrane (Merck Millipore, Merck KGaA, Darmstadt, Germany). The membrane was blocked with 5% nonfat milk in Tris-buffered saline containing 0.1% Tween-20 for 1 h at room temperature and then incubated overnight at 4 °C with specific primary antibodies. HRP-conjugated anti-rabbit or mouse (Santa Cruz) secondary antibodies were used and visualized using a ChemiDoc imaging system.

Cells were cultured on plastic in 6-well plates until 100% confluence, and a scratched area was created using a blue pipette tip. The gap was photographed at 0 and 16 h. The migration distance was measured in five fields randomly from every triplicate sample and is expressed as the mean versus that of the control.

A Matrigel invasion assay was performed by using Transwell polycarbonate membrane filters with a pore size of 8.0 μm (Costar, Cambridge, MA). Briefly, Matrigel was thawed, diluted and placed into the upper chamber of a 24-well Transwell. After 4 h of incubation for Matrigel gelling, the cells were harvested, resuspended in 1% FBS media at a density of 10⁶ cells/ml and placed onto the Matrigel. The lower chamber was filled with 600 μl of 10% FBS media. After a 6-h incubation at 37 °C, the nonmigrated cells on the upper surface of the membranes were eliminated by cotton swabs. The membranes were fixed with pure methanol and stained with a solution of crystal violet. Cells migrating to the substratum of the membrane were counted.

A 48-well plate was coated with 200 μl/well of 10 μg/ml fibronectin (Sigma, #F0895) in Ca and Mg-free PBS (pH 7.40) at 4 °C overnight and blocked with 200 μl of 1% heat-denatured BSA in PBS for 60 min at 37 °C. The blocking solution was discarded, and the wells were rinsed with PBS three times. Then, 100 μl of adhesion buffer was added to each well. The cells were harvested, and 50,000 cells were plated in each well and incubated for 20–30 min at 37 °C. The wells, except the standard control wells, were washed very gently 3 times until no cells were visible in the BSA-coated control wells. Then, 100 μl of adhesion buffer and 15 μl of MTT dye were added to the wells, and the plate was incubated for 4 h at 37 °C in the dark. Next, the MTT-treated cells were lysed in 100 μl of DMSO, and absorbance was measured at 570 nm on a spectrophotometer.

Total RNA from three groups of MDA-MB-231 cells (231 NC, 231 Vec and 231 ANGPTL4) was reverse transcribed into complementary DNA (cDNA). Then, qRT-PCR amplification was performed using FastStar Universal SYBR Green Master Mix (Roche Diagnostics GmbH, Mannheim, Germany) on an ABI 7500 Real-Time PCR Instrument (Applied Biosystems). The raw cycle number (Ct) values for each specimen were determined by the software supplied with the instrument. The relative gene expression level was computed by ΔCt (ΔCt = Ct_sample − Ct_ACTN). The discrepancy between the ΔCt of the target gene and reference gene was computed by ΔΔCt (ΔΔCt = ΔCt_{target gene} – ΔCt_{reference gene}). Expression fold changes were calculated by 2-ΔΔCt counts.

A total of 161 TNBC patients with a median age of 49 years (range 17–76 years) were included in this study. The median OS time was 82 months (range 0.3–153 months). Until the last follow-up investigation, 47 (29.2%) patients were identified as having recurrence or metastasis, and the disease-specific death rate was 41.8% (81/194).

The subcellular localization and expression of the ANGPTL4 protein were observed and scored by immunohistochemistry (IHC) in various samples including 28 adjacent noncancerous tissues. According to the ROC curve results, the area under the curve was 0.634 (Fig. 1d, p = 0.007). An ANGPTL4 score of 4 maximized the Youden Index (sensitivity [0.655] + specificity [0.563] - 1 = 0.217) as the optimal cut-off score. Thus, patients were split into low (score < 4) and high (score ≥ 4) expression groups.

ANGPTL4 was positively expressed in the cytoplasm of TNBC cells (Fig. 1a). The high expression rate was 59% in TNBC tumor cells, which was lower than that of stromal cells (68%, Fig. 1b), especially adipocytes and epithelial cells (Fig. 1c). Weakly expressed ANGPTL4 was significantly related to the presence of neurovascular invasion (28.8% vs. 15.8%, p = 0.047). Additionally, a higher incidence of relapse or progression was observed in the ANGPTL4 low expression group than in the ANGPTL4 overexpression group (40.9% vs. 21.1%, p = 0.006). Further, the association between ANGPTL4 expression and histological differentiation was marginally significant (p = 0.087). However, in 161 TNBC patients, the ANGPTL4 expression level was not correlated with age, menstrual history, tumor stage or other clinical factors (Table 1).

Forty-seven patients showed relapse or metastasis from the diagnosis until the final follow-up. Univariate survival analyses were performed to explore the prognostic effect of ANGPTL4 expression in TNBC patients. In the weak ANGPTL4 expression group, the median OS time was 135.7 months [95% confidence interval (CI) 55.4–216.0 months]. The 5-year OS and DFS frequencies were 64 and 56%, respectively. Nevertheless, the high ANGPTL4 expression group had a preferable prognosis: the 5-year OS and DFS rates were 82 and 77%, respectively. The Kaplan–Meier survival curves showed that ANGPTL4 overexpression indicated longer OS and DFS than low expression (both p < 0.05, Fig. 1e). In univariate survival analysis, patients with stage III disease, higher T stage, low ANGPTL4 expression, lymph node metastasis and vascular invasion had shorter OS and DFS times. Multivariate analysis indicated that weak ANGPTL4 expression independently predicted poor prognosis (p < 0.05), as did advanced disease (Tables 2 and 3).

First, we determined the expression of ANGPTL4 in different breast cancer cell lines. In luminal BC cell lines such as SKRB3 and MDA-MB-453, the protein expression of ANGPLT4 was higher than that in basal-like BC cell lines such as BT549 and MDA-MB-231 (Fig. 2a), which are considered more invasive than luminal BC cell lines [18]. To further elucidate the role of ANGPTL4, we infected MDA-MB-231 and BT-549 cell lines with lentivirus expressing ANGPTL4, while empty vector served as the corresponding control. The efficiency of overexpression was confirmed by qPCR and Western blot analyses (Fig. 2b). To further assess the effect of ANGPTL4 in TNBC, we detected the impact of ANGPTL4 overexpression on cell migration, invasion and adhesion. In an in vitro scratch-wound assay (Fig. 2c), 16 h after scratching, the percentage of wound closure in the 231 vector group was 69.9 ± 4.0%, while that in the 231 ANG OE group was 38.5 ± 12.7% (p < 0.01). The percentage of wound closure in the 549 vector group was 61.9 ± 5.0%, while that in the BT-549 ANG OE group was 43.9 ± 9.8% (p < 0.05). In the Matrigel invasion assay, the invaded cell number of the 231 vector group was 761 ± 142, while that of the 231 ANG OE group was 245 ± 63 (Fig. 2d, p < 0.05). The results from the wound healing assay and Matrigel invasion assay showed that ANGPTL4 overexpression evidently attenuated the cell invasive and migratory abilities compared with the control. In the cell adhesion assay, the attachment percentage of the 231 vector group was 57.7 ± 4.5%, while that of the 231 ANG OE group was 39.4 ± 6.6% (Fig. 2e, p < 0.05), suggesting that ANGPTL4 overexpression inhibits cell adhesion and attachment, preventing cell migration and invasion.

To further understand how ANGPTL4 inhibits tumor migration, we performed next-generation RNA sequencing on three groups of MDA-MB-231 cell lines: the NC (231 NC), vector (231 vector) and ANGPTL-overexpressing groups (231 ANGPTL4 OE). Differentially expressed gene (DEG) analysis and GO (Gene ontology, http://​www.​geneontology.​org/​) enrichment revealed that genes included in the pathways of adherens junction, blood vessel morphogenesis, extracellular matrix and wound healing were most affected by ANGPTL4 overexpression (Fig. 3a). Real-time quantitative PCR was performed to confirm the results of RNA sequencing. Among 52 genes, ten were significantly downregulated in the 231 ANGPTL4 group compared with the 231 vector group, which was consistent with the RNA sequencing results (Fig. 3b). These genes are BSG (basigin), LGALS3BP (galectin 3 binding), EGFL7–2 (EGF-like domain multiple 7), LAMA4 (laminin subunit alpha 4), MYL6 (myosin light chain 6), COL6A2 (collagen type VI alpha 2 chain), ITGB5 (integrin subunit beta 5), TGFB1 (transforming growth factor beta 1), CST3 (cystatin C), and VAV1 (vav guanine nucleotide exchange factor 1).