Angiopoietin-like protein 1 antagonizes MET receptor activity to repress sorafenib resistance and cancer stemness in hepatocellular carcinoma

CELL LINES

Hep3B, HepG2, C3A, THLE-2, and 293T cell lines were purchased from the American Type Culture Collection (Manassas, VA, USA), and Huh7, Huh6, and Huh1 cell lines were purchased from the Japanese Collection of Research Bioresources Cell Bank (Osaka, Japan). HA22T/VGH and PLC/PRF/5 cell lines were obtained from the Bioresource Collection and Research Center of the Food Industry Research and Development Institute (Hsinchu, Taiwan). Mahlavu cells were generously donated by Dr. Michael Hsiao (Genomics Research Center, Academia Sinica, Taipei, Taiwan). HCC36 cells were generously donated by Dr. Jang-Yang Chang (National Cheng Kung University, Tainan, Taiwan). All cells were grown in regular Dulbecco's modified Eagle's medium (DMEM; Invitrogen, Waltham, MA, USA) supplemented with 10% fetal bovine serum (Invitrogen), 1,000 U/mL of penicillin, and 100 μg/mL of streptomycin at 37°C in a humidified atmosphere of 95% air and 5% CO₂. To develop sorafenib-resistant cells, Huh7 cells and HepG2 cells were treated stepwise with increasing doses of sorafenib for a long period of time. The resulting sorafenib-resistant cells, Huh7/SR and HepG2/SR, were maintained in culture medium containing sorafenib (Selleckchem, Houston, TX, USA) at a final maximum concentration of 10 (Huh7/SR) and 7.5 μM (HepG2/SR).

CELL APOPTOSIS AND CELL SORTING BY FLOW CYTOMETRY ANALYSIS

For the apoptosis assay, cells were seeded into 6-cm dishes at a density of 3 × 10⁵ cells/dish overnight and then treated with vehicle or 10 μM of sorafenib and incubated for 48 hours. Aliquots of cells were collected and washed once with ice-cold phosphate-buffered saline (PBS) and then fixed with ice-cold 70% ethanol overnight. After fixation, cells were washed with PBS to remove residual ethanol, pelleted, and resuspended in PBS containing 1 μg/mL of propidium iodide (Sigma-Aldrich, St. Louis, MO, USA) and 10 μg/mL of RNase A (Sigma-Aldrich), and then samples were analyzed using an FACSCalibur flow cytometer (BD Biosciences, San Jose, CA, USA). For stem cell marker staining and cell sorting, cells were collected and labeled with a series of primary antibodies for 30 minutes at 4°C in the dark. After being washed with PBS (Invitrogen), labeled cells were analyzed by FACSCalibur flow cytometer (BD Biosciences) for stem cell marker population staining and a FACSAria (BD Biosciences) for cell sorting.

SPHERE FORMATION ASSAY

Single cells were seeded on ultralow attachment plates (Corning Inc., Corning, NY, USA) at a concentration of 1,000 cells/mL in DMEM supplemented with 50 ng/mL of epidermal growth factor (EGF), 20 ng/mL of basic fibroblast growth factor, and 1× B27. Fresh medium was added every week. Spheres were counted at 21 days after seeding (>100 μm in diameter).

SPECIMENS

Tissues used for sorafenib treatment were obtained from National Cheng Kung University (Tainan, Taiwan) with institutional review board approval (A-ER-103-174). Patient specimens had been fixed in formalin and embedded in paraffin before they were archived. Sorafenib responses of patients were defined according to the Response Evaluation Criteria in Solid Tumors and were classified as PD (progressive disease) and SD (stable disease) + PR (partial response).19 We analyzed data for all patients whose tumor, nodes, metastasis and survival data were available (n = 92) through the Taiwan Liver Cancer Network with institutional review board approval and divided the patients into two groups in accord with their mean ANGPTL1 and Slug messenger RNA (mRNA) expression levels.

MICE XENOGRAFT ASSAY

All animal experiments were conducted in accord with a protocol approved by the Laboratory Animal Center of the National Health Research Institutes (Zhunan, Taiwan). Female nonobese diabetic/severe combined immunodeficiency (NOD/SCID) mice (6-8 weeks old) were used for tumor growth in xenograft studies. Approximately 1 × 10⁷ cells/mouse suspended in 100 μL of PBS and Matrigel (1:1; BD Biosciences) were subcutaneously injected into each flank. When tumor volumes reached approximately 200 mm³, as determined by measuring tumor lengths and widths using calipers and calculating their volumes with the formula (1/2[length × width²]) after inoculation, mice were randomly assigned to groups in which they received thrice-weekly oral doses of ether vehicle solution or 25 mg/kg of sorafenib (Selleckchem) for 4 weeks. Mice were treated intravenously with 100 μg/kg of ANGPTL1 recombinant protein (rANGPTL1) twice a week. Tumor volumes were measured every 3 days, and tumor tissues were extracted at indicated times for further analysis.

IN VITRO BINDING ASSAY

Two micrograms of MET-His protein (MET extracellular domain 1-932 amino acids) was mixed with 2 μg of ANGPTL1-GST (glutathione S-transferase) protein in binding buffer and incubated at 4°C for 4 hours, with gentle rotation. Forty microliters of Glutathione Sepharose 4B beads (GE Healthcare Life Sciences, MA, USA) were added to the mixture and incubated at 4°C for 2 hours, with gentle rotation. Beads were washed extensively with binding buffer and boiled. Then, proteins were resolved on sodium dodecyl sulfate/polyacrylamide gel electrophoresis (SDS-PAGE) and analyzed by western blotting.

IMMUNOPRECIPITATION

Cells were lysed by brief sonication in coimmunoprecipitation (Co-IP) buffer (20 mM of Tris [pH 7.4], 150 mM of NaCl, 1 mM of ethylenediaminetetraacetic acid, and 0.5% nonyl phenoxypolyethoxylethanol) supplemented with protease inhibitor cocktails (Sigma-Aldrich). Lysates were centrifuged for 20 minutes at 14,500 rpm, and the resulting supernatant was precleared by incubation with immobilized Protein A/G gel (20 μL; Pierce, Dallas, TX, USA) for 1 hour at 4°C. The precleared supernatant was subjected to overnight immunoprecipitation at 4°C using the indicated antibodies (Supporting Table S1) or control immunoglobulin G (IgG) antibodies. The next day, protein complexes were collected by incubation with 35 μL of immobilized Protein A/G gel for 2 hours at 4°C. Protein complexes were washed five times with Co-IP buffer and eluted by boiling in protein sample buffer under reducing conditions, after which proteins were resolved by SDS-PAGE and analyzed by western blotting.

STATISTICAL ANALYSIS

All statistical analyses were carried out using Prism software (GraphPad, La Jolla, CA, USA). Data are presented as the mean ± SE of at least three independent experiments. A chi-squared test was used to compare clinicopathological data. Comparisons of groups were performed using one-way analysis of variance. A two-tailed Student t test was used to compare data between two groups. P values <0.05 were considered statistically significant.

ELEVATED EXPRESSION OF ANGPTL1 ENHANCES THE SORAFENIB SENSITIVITY OF HCC IN VIVO AND IN VITRO

To determine whether ANGPTL1 is associated with sorafenib resistance in HCC, we examined expression of ANGPTL1 and the half maximal inhibitory concentration (IC₅₀) of sorafenib in HCC cell lines, including Huh7 cells and their sorafenib-resistant counterparts, Huh7/SR (sorafenib resistance) cells (Supporting Fig. S1A,B). A negative correlation between ANGPTL1 expression and the IC₅₀ value of sorafenib was observed in HCC cell lines by enzyme-linked immunosorbent assay (Fig. 1A). To determine the impact of ANGPTL1 on sorafenib sensitivity, we overexpressed ANGPTL1 in relative sorafenib-resistant cells (Huh7/SR and Mahlavu) and found that ANGPTL1 significantly decreased sorafenib resistance (Fig. 1B and Supporting Fig. S1C-E). We also found that knockdown of ANGPTL1 in Huh7 and HepG2 cells increased sorafenib resistance (Fig. 1C and Supporting Fig. S1F-H). To further investigate the role of ANGPTL1 in sorafenib resistance in HCC cells in vivo, HCC cells were genetically modulated of ANGPTL1 and subcutaneously inoculated into NOD/SCID mice. The results showed that tumor expression of ANGPTL1 was more sensitive to sorafenib treatment than control tumors (Fig. 1D and Supporting Fig. S1I, left panel). Tumor xenografts dissected from each mouse were used to confirm ANGPTL1 expression by quantitative reverse-transcriptase polymerase chain reaction (qRT-PCR) analysis (Fig. 1D and Supporting Fig. S1I, right panel). These results demonstrate that ANGPTL1 can overcome sorafenib resistance in HCC.

To determine the clinical significance of ANGPTL1 in HCC patients, we analyzed ANGPTL1 expression in a cohort of 92 HCC specimens (cohort 1). The results showed that ANGPTL1 expression was inversely correlated with tumor size, tumor stage, tumor grade, vascular invasion, and poor prognosis (Table 1; Supporting Fig. S2). We also collected another cohort of 73 HCC specimens (cohort 2) with sorafenib treatment and found that elevated expression of ANGPTL1 was associated with longer survival after sorafenib treatment (Fig. 1E) and was positively correlated with sorafenib responsiveness (Fig. 1F). Taken together, these results indicate that ANGPTL1 expression is inversely correlated with poor sorafenib responsiveness and poor prognosis in patients with HCC.

ANGPTL1 OVERCOMES SORAFENIB RESISTANCE BY SUPPRESSING Slug

Previous studies have shown that EMT is crucial for sorafenib resistance in HCC.7, 8, 20 We found that Huh7/SR cells exhibited EMT characteristics and a spindle-like, fibroblastic morphology (Supporting Fig. S3A). Expression of the epithelial marker, E-cadherin, was lower in Huh7/SR cells, whereas expression of the mesenchymal marker, Vimentin, and the EMT-related transcription factors, Slug and Twist, were higher in Huh7/SR cells than in parental cells (Supporting Fig. S3B). We subsequently investigated whether Slug or Twist is involved in ANGPTL1-mediated sorafenib sensitization. Knockdown of ANGPTL1 increased Slug mRNA and protein expression, but not Snail and Twist (Fig. 2A,C). Overexpression of ANGPTL1 in Huh7/SR cells consistently decreased Slug expression (Fig. 2B,C). Moreover, we found that sorafenib resistance was abolished by Slug knockdown in Huh7/short hairpin (sh)ANGPTL1 cells (Fig. 2D). ANGPTL1-induced sensitization to sorafenib in Huh7/SR cells was repressed by Slug reexpression (Fig. 2E). Similar results were found in HepG2 and Mahlavu cells (Supporting Fig. S3C-H). To further elucidate the physiological role of Slug in sorafenib resistance, we silenced Slug and found recovery of sorafenib sensitivity in Huh7/SR cells (Supporting Fig. S4A-C). Moreover, high Slug expression was positively correlated with poor prognosis in cohort 1 (Supporting Fig. S4D,E) and was associated with shorter survival rates and poorer responses to sorafenib treatment in cohort 2 (Supporting Fig. S4F,G).

We next analyzed the correlation between ANGPTL1 and Slug expression in two cohorts and found that ANGPTL1 expression was inversely correlated with Slug expression (Fig. 2F and Supporting Fig. S5A). Additionally, the Gene Expression Omnibus and Oncomine databases21, 22 also demonstrated an inverse correlation between ANGPTL1 and Slug expression (Supporting Fig. S5B,C). These results suggest that ANGPTL1 overcomes sorafenib resistance by repressing Slug in HCC cells.

Slug RESCUES ANGPTL1-MEDIATED REPRESSION OF CANCER STEMNESS AND TUMOR GROWTH IN HCC CELLS

Emerging evidence suggests that CSC-mediated EMT in HCC cells is critical for tumor metastasis and sorafenib resistance.11, 12 To clarify the effects of ANGPTL1 on CSCs, we examined the sphere-forming abilities of Huh7, Huh7/SR, and HepG2 cells and their derivatives and found that sphere formation was repressed by Slug knockdown (Fig. 3A and Supporting Fig. S6A). Treatment with rANGPTL1 significantly inhibited sphere formation and sorafenib-resistant tumor growth (Supporting Fig. S7A,B). ANGPTL1-mediated repression of sphere formation in Huh/SR/ANGPTL1 cells was reversed by Slug overexpression (Fig. 3A). It is known that CD133 and CD44 are CSC markers of HCC.23, 24 Thus, Huh7/SR cells with high and low CD44 and CD133 expression levels were sorted to determine sorafenib sensitivity. Notably, our data demonstrated that Huh7/SR cells expressing low levels of CD44 or CD133 showed greater sensitivity to sorafenib than parental cells and CD44^High/CD133^High Huh7/SR cells (Supporting Fig. S8). Furthermore, Slug knockdown reduced shANGPTL1-mediated CD133^High and CD44^High cell population and CSC marker (SRY [sex determining region Y]-box 2 [SOX-2], octamer-binding transcription factor 4 [Oct-4], Nanog, and Krüppel-like factor 4 [Klf-4]) expression (Fig. 3B-F and Supporting Fig. S6B-G). Taken together, these results indicate that ANGPTL1 decreases cancer stemness and sorafenib resistance in HCC by suppressing Slug.

ANGPTL1 REPRESSES Slug-MEDIATED SORAFENIB RESISTANCE THROUGH INHIBITION OF THE MET RECEPTOR−AKT/ERK SIGNALING PATHWAY

To identify the mechanisms underlying the regulation of sorafenib resistance in HCC by ANGPTL1, we used a phospho-receptor tyrosine kinase (phospho-RTK) array to analyze the alterations of phospho-RTKs in Hun7/shANGPTL1 cells, Huh/SR/ANGPTL1 cells, and their control cells. The results showed that phosphorylation of the MET receptor was significantly increased in Huh7/shANGPTL1 cells compared with Huh7/shLuc cells. ANGPTL1 overexpression reduced MET receptor phosphorylation in Huh7/SR cells (Supporting Fig. S9A). To determine whether the MET receptor signaling pathway is associated with ANGPTL1-mediated sorafenib resistance, we examined the phosphorylation of AKT and ERK, which are downstream molecules of the MET receptor pathway, and found that these proteins were negatively regulated by ANGPTL1 (Fig. 4A,B). We also found that Huh7/SR cells express higher levels of HGF than Huh7 cells (Fig. 4A,B). However, HGF expression was not affected by ANGPTL1. Next, we found that rANGPTL1 and SU11274 (a MET inhibitor) blocked hepatocyte growth factor (HGF)-induced MET phosphorylation, Slug expression, and sorafenib resistance (Fig. 4C and Supporting Fig. S9B). An inverse correlation between ANGPTL1 and HGF was observed in HCC cell lines (Supporting Fig. S10A-C). We also found that the HGF/MET signaling pathway positively regulates ANGPTL1 expression (Supporting Fig. S10D,E). However, ANGPTL1 did not alter HGF expression (Supporting Fig. S10F). ANGPTL1 expression was also regulated by different stimuli, such as serum, HGF, EGF, and hypoxia, in normal and malignant liver cells (Supporting Fig. S11A-F), but the mechanism underlying this regulation requires further investigation. To confirm the role of MET signaling in the regulation of Slug expression and sorafenib resistance, constitutively active forms of mitogen-activated protein kinase kinase (CA-MEK1) and AKT (Myr-AKT) were transfected into Huh7/SR/ANGPTL1 cells. The results showed that ANGPTL1-mediated repression of sorafenib resistance and Slug expression in Huh7/SR/ANGPTL1 cells was recovered by CA-MEK1 and Myr-AKT (Fig. 4D). We found that sorafenib resistance and Slug expression were reduced by treatment with SU11274, U0126 (a MEK inhibitor), and LY294002 (a phosphoinositide 3-kinase [PI3K] inhibitor) in Huh7/shANGPTL1 cells (Fig. 4E). These data indicate that ANGPTL1 is a negative regulator of the MET receptor-AKT/ERK signaling pathway, which is associated with Slug regulation and sorafenib resistance.

ANGPTL1 REPRESSES Slug-INDUCED SORAFENIB RESISTANCE IN AN Egr-1-DEPENDENT MANNER

A previous study showed that AKT and ERK1/2 are associated with EGF-induced Slug expression through Egr-1 in ovarian cancer cells.25 Our findings indicated that ANGPTL1 represses Egr-1 and Slug expression (Fig. 5A). To determine whether Egr-1 is involved in Slug expression and the subsequent development of sorafenib resistance, Egr-1 expression was knocked down in Huh7/SR cells, which resulted in significant repression of Slug expression and sorafenib resistance (Fig. 5B). Egr-1-induced sorafenib resistance in Huh7 cells was abrogated by Slug knockdown (Fig. 5C). We also found that inhibition of Erg-1 expression in Huh7/shANGPTL1 cells suppressed Slug expression and sorafenib resistance (Fig. 5D). To confirm the role of the MET signaling pathway in Egr-1 regulation, we transfected CA-MEK1 and Myr-AKT into Huh7/SR/ANGPTL1 cells and found that ANGPTL1-mediated repression of Egr-1 expression was recovered by CA-MEK1 and Myr-AKT (Fig. 5E). These results indicate that ANGPTL1-mediated repression of Slug expression occurs through the AKT/ERK-regulated Egr-1 pathway.

ANGPTL1 INTERACTS WITH THE MET RECEPTOR AND COMPETES WITH HGF FOR MET BINDING

ANGPTLs exert their biological effects by binding to integrins in various cancers,16, 26 but the functional receptor of ANGPTL1 in HCC is unknown. To determine whether integrins are associated with ANGPTL1-mediated repression of sorafenib resistance, we first analyzed expression of integrins by qRT-PCR. We found that integrin β₁ expression was reduced in ANGPTL1-overexpressing Huh7/SR cells (Supporting Fig. S12A-I). We also found that integrin β₁ knockdown in Huh7/SR/ANGPTL1 cells did not abolish ANGPTL1-mediated repression of sorafenib resistance (Supporting Fig. S12J-M), suggesting that ANGPTL1-mediated sorafenib resistance may not through integrin β₁.

To determine whether ANGPTL1 interacts with MET to exert its effects, we performed an in vitro binding assay and confirmed that the rANGPTL1-GST protein binds directly to the His-tagged recombinant MET extracellular domain (1-932 amino acids) protein (Fig. 6A). Co-IP assay showed that overexpressed exogenous ANGPTL1-V5 interacted with endogenous MET, but not integrin α₁ and integrin β₁, in Huh7/SR cells (Fig. 6B). To determine whether ANGPTL1 competes with HGF for MET binding, Co-IP assay was performed. Analysis of immunoprecipitates from 293T cells transfected with ANGPTL1-flag, MET-V5, and HGF-HA (hemagglutinin) revealed that ANGPTL1 interacts with the MET receptor and competes with HGF for MET binding (Fig. 6C). These results indicate that ANGPTL1 represses MET phosphorylation through competition with HGF.