CSF-1R promotes vasculogenic mimicry via epithelial-mesenchymal transition in nasopharyngeal carcinoma cells

Cell lines and cultures

The non-metastatic human NPC 6–10B cell line (Sun Yat-sen University, Guangzhou, China) was used in this study. Fetal bovine serum (FBS; Gibco, USA) 10 % in RPMI 1640 medium (Gibco, USA) was used to cultivate cells in a humid incubator at 37 °C and 5 % CO₂.

Lentiviral vectors construction and transfection

The lentivirus overexpressing CSF-1R that carries the enhanced green fluorescent protein (EGFP) gene was purchased from GeneChem Co., Ltd. (Shanghai, China). The LV-CSF-1R vector was used to transfect the 6–10B cells (LV-CSF-1R group). Transfected LV-negative control vector (LV-Ctrl group) and untransfected vector (Mock group) were used as control groups. A lentiviral infection with a multiplicity of infection (MOI) of 15 was enabled for 18 h after the 6–10B cells were injected in 12 well plates. Stably infected cells were identified after 72 h of incubation by adding 3 g/mL puromycin (Solarbio, Beijing, China). After at least 10 consecutive generations, transfection efficiency was assessed based on EGFP protein expression by using inverted fluorescence microscopy (Olympus, Japan). Quantitative real-time PCR (RT-qPCR) and Western blotting were used to demonstrate the upregulation of CSF-1R expression.

Quantitative real-time-PCR

For total RNA extraction, the 6–10B cells were collected and put into TRIzol Reagent (Invitrogen, CA, USA). First, the reaction strand synthesis of cDNA was performed using the FastKing RT kit (GeneChem Co., Ltd., Beijing, China). Next, the SuperReal PreMix Plus SYBR Green PCR kit (Tiangen), CSF-1R primers (CSF-1R forward, 5′-TCT GGT CCT ATG GCA TCC TC-3′, reverse, 5′-GAT GCC AGG GTA GGG ATT C-3′), and β-actin primers (β-Actin forward, 5′-TGA CGT GGA CAT CCG CAA AG-3′, reverse, 5′-CTG GAA GGT GGACAG GGA-3′) in the ABI 7500 Real-Time PCR system (Thermo Fisher Scientific, USA) were used for quantitative PCR analysis. Lastly, the 2^−ΔΔCq method was applied to quantify relative RNA levels, with β-actin serving as a normalizing factor [20].

Western blotting

Protease inhibitors were diluted 100:1 with RIPA lysis buffer (Beyotime, Shanghai, China) and added to the 6–10B cell culture dishes for protein extraction by lysis. Proteins were transferred to polyvinylidene difluoride membranes after electrophoresis in 8 % sodium dodecyl sulfate-polyacrylamide gels separated them (EMD Millipore, Bedford, MA, USA). The membranes were incubated with the primary antibody for 16 h at 4 °C after being blocked in 5 % nonfat milk for 1 h. Orders were placed with Cell Signaling Technology for all antibodies (Danvers, MA, USA). A nti-CSF-1R (1:1,000; #67455), anti-E-cadherin (1:1,000; #3195), anti-HIF-1α (1:1,000; #36169), anti-MMP-2 (1:1,000; #40994), anti-MMP9 (1:1,000; #13667), anti-phosphoinositide 3-kinase (PI3K; 1:1,000; #4255), anti-Akt (1:1,000; #9272), anti-p-Akt (1:1,000; #4060S), and anti-β-actin (1:1,000; #4970) were the primary antibodies. Protein bands were transferred to species-specific secondary antibodies (1:3,000; #7076S) conjugated to horseradish peroxidase polymer (HRP) the following day, and they were incubated for 2 h at 37 °C. Finally, bands were detected with enhanced chemiluminescence reagent (ECL; Meilun, Dalian, China) in an Odyssey FC system (LI-COR Biosciences, USA) and then quantified by ImageJ (NIH, Bethesda, MA, USA) using β-actin as an internal control.

Three-dimensional culture

Matrigel (354230, BD Biosciences, USA) and pipette tips were pre-chilled before performing the tube-formation assays [21]. Next, 10 µl Matrigel was added to the angiogenic slides (Ibidi, USA) for pretreatment. The Matrigel layer in each well was covered with a suspension of 6–10B cells in serum-free media (2 × 10⁵ cells/mL, 50 μL), which was then incubated for 12 h at 37 °C. Five areas were randomly photographed with a microscope and the average number of tubes was calculated.

Transwell assay

Transwell chambers on 24-well plates that were either pre-coated with Matrigel or left uncoated (8-μm pore size, Corning, MA, USA) were used for migration and invasion tests. The 6–10B (8-μm pore size, Corning, MA, USA) were used for migration and invasion tests (8-μm pore size, Corning, MA, USA). The 6–10B cells were resuspended in serum-free media, and 500 μL of the medium containing 15 % FBS was introduced to the bottom chamber while 200 μL of the cell suspension (5 × 10⁵ cells/ml) was pulled into the Transwell chamber. The cells were allowed to migrate or invade for 48 or 72 h before being entirely removed from the top of the Transwell chambers with cotton swabs. 4 % paraformaldehyde-fixed, and 0.1 % crystal violet (Shanghai Biyuntian Biotechnology Co., Ltd., China) stained the cells that had migrated or invaded for 30 min, respectively. The cells that had migrated or invaded were fixed in 4 % paraformaldehyde and stained for 30 min with 0.1 % crystal violet (Shanghai Biyuntian Biotechnology Co., Ltd., China). Finally, using a microscope, these cells were counted in five random regions of each well.

Animal models

The Ethics Committee of the People’s Hospital of the Guangxi Zhuang Autonomous Region authorized the use of animals in this study in compliance with institutional policies for the treatment of laboratory animals (KYGZR-2013-06, China). We bought 15 female BALB/c nude mice (4–5 weeks old; 15–21 g) from the Guangxi Medical University Laboratory Animal Center (SCXK(Gui)2014-0002, China). Mice were kept in a specialized pathogen-free (SPF) laminar flow chamber at a constant humidity level of 55 ± 15 % and temperature of 20–23 °C. The animals were divided into three groups of five at random. NPC models were established by subcutaneous injection of a 0.2-mL suspension of Mock, LV-Ctrl, and LV-CSF-1R overexpressing 6–10B cells (5 × 10⁶ cells/mL) into the right dorsum of nude mice. Animals were monitored for disease progression and health status every 3 days, with care in noting any signs of distress or loss of health. The removed tumors were fixed in 10 % formalin and then implanted in paraffin after the mice were euthanized after three weeks.

CD34-PAS dual staining

Serially sectioned, dewaxed, and rehydrated mouse tissue samples were then used to perform thermally induced antigen repair in a pressure cooker with sodium citrate buffer. The tissues were treated with 3 % hydrogen peroxide to eliminate endogenous peroxidase. Sections were then sealed with 2 % BSA for 20 min at 37 °C, after which they were incubated for 16 h at 4 °C with diluted anti-CD34 antibody (1:200; #ab81289; Abcam, Cambridge, UK). Following that, the sections were treated for 30 min at 37 °C with an HRP-conjugated secondary antibody. The sections were then treated with Schiff’s solution for 20 min before being stained with 3-3′ diaminobenzidine (DAB), 0.5 % periodate solution (PAS), and then Schiff’s solution [21]. Lastly, the slices were re-stained with hematoxylin. The number of VM was calculated after five randomly selected locations from each section were photographed and examined under a microscope.

Statistical analysis

The data analyses were conducted using GraphPad Prism 9 (GraphPad Software, CA, USA) and SPSS 22.0 (IBM Corp., NY, USA). Data are expressed as mean ± standard deviation. One-way ANOVA was performed to compare the means of different groups, and the LSD t-test was used for pairwise comparisons between groups. The threshold for statistical significance was established at p<0.05 for all two-sided statistical tests.

Stable CSF-1R overexpression in 6–10B cell line

To explore CSF-1R the impact on NPC cell lines. To investigate the effect of CSF-1R on NPC cell lines. In previous experiments [22], we evaluated CSF-1R expression in three NPC cell lines (5–8F, 6–10B, and CNE-2), and found that only 6–10B had relatively low CSF-1R expression. Therefore, in this study, 6–10B cells were selected to stably overexpress CSF-1R. Meanwhile, the sequence design for knocking down CSF-1R failed; therefore, knockdown experiments were not carried out.

We employed a lentivirus-mediated CSF-1R RNA (LV-CSR-1R group) transfection technique to upregulate CSF-1R expression levels and produce stably expressed cell lines to analyze the phenotypic changes (Figure 1A). As controls, control-RNA-infected (LV-Ctrl group) and uninfected (Mock group) cells were employed. By using RT-qPCR and western blotting, we examined the expression of CSF-1R in 6–10B cells. The results showed strong and stable expression of CSF-1R in CSF-1R RNA-transfected 6–10B cells as compared with the LV-Ctrl and Mock groups (p<0.001), and there was no statistical significance between the LV-Ctrl and Mock groups (Figure 1B and D).

Figure 1:

CSF-1R expression is abnormally high in 6–10B cells. (A) Transfected 6–10B cells were observed by fluorescence microscopy (magnification, ×200). (B–D) RT-qPCR and western blotting were used to confirm CSF-1R expression in 6–10B cells. Mock and LV-Ctrl are the control samples for LV-CSF-1R. Data as mean ± SD (n=3). ***p<0.001.

Alteration of CSF-1R expression affects tube formation in vitro

To observe whether CSF-1R overexpression affects the tube-formation ability, we next performed the tube formation assay. The results of 3D culture demonstrated the appearance of tube formation following CSF-1R high expression, which were tumor cells arranged in an arc (p<0.01), as illustrated in Figure 2A and B. LV-Ctrl and Mock groups of cells were oval and hardly formed complete tube structures. These results confirmed that CSF-1R induced tube formation in NPC cells in vitro.

Figure 2:

CSF-1R overexpression increases VM formation in vitro. (A) VM channels in three-dimensional cultures of cells overexpressing CSF-1R (magnification, ×200). (B) The number of tubes that formed in each group was statistically analyzed. Data as mean ± SD (n=3). **p<0.01. ***p<0.001.

6–10B cell migration and invasion are promoted by high expression of CSF-1R

VM formation is crucial for the migration and invasion of tumor cells [23]. Figure 3A and B show that after high expression of CSF-1R, the number of cells migrating to the lower lumen was considerably higher than in the Mock and LV-Ctrl groups (p<0.01). Furthermore, the invasion assay indicated a more significant cell invasion potential in the LV-CSF-1R group (p<0.00). No significant differences were observed between the Mock and LV-Ctrl groups (Figure 3C and D). The findings imply that NPC cell invasion and migration are mediated by CSF-1R.

Figure 3:

6–10B cells are driven to migrate and invade by CSF-1R. (A and C) Transwell assay (magnification, ×100) results show that upregulation of CSF-1R influences migration and invasion. (B and D) The bar chart shows the statistical outcomes. Data as mean ± SD (n=3).**p<0.01; ***p< 0.001.

CSF-1R induced VM formation in animal models

6–10B cells were injected subcutaneously into an immunocompromised mouse to create a xenograft cancer model to further support the effect of CSF-1R on VM formation in vivo. VM in the tumors was counted using CD34-PAS double staining. As shown in Figure 4A, LV-CSF-1R tumors generated significantly more VM channels than the LV-Ctrl and Mock groups (p<0.001). No significant differences were observed between the Mock and LV-Ctrl groups (Figure 4B). Our data suggest that CSF-1R overexpression in NPC cells promotes VM.

Figure 4:

Overexpression of CSF-1R accelerated the VM effect in tumor tissues. (A) CD34-PAS double staining of each group in vitro. Red arrows show VM; Black arrows indicate endothelial vessels; top panel, magnification: 200×; Bottom panel, magnification: 40×. (B) Statistical analysis of VM formation in xenograft tumors. Data as mean ± SD (n=3). ***p< 0.001.

CSF-1R promoted EMT by the PI3K/AKT pathway

The epithelial-mesenchymal transition (EMT) is critical for tumor development, and its stimulation causes tumor cell invasion and spread [24]. Research has confirmed that tumor cells can form VM through the EMT process [25]. Therefore, we examined whether the key gene CSF-1R affects the EMT process by western blotting. In the current research, overexpression of CSF-1R substantially reduced the levels of E-cadherin protein and raised the levels of HIF-1α, MMP-2, and MMP-9 proteins in comparison to the Mock and LV-Ctrl groups (Figure 5A). Taken together, the above data confirm the involvement of CSF-1R in the EMT process of NPC.

Figure 5:

CSF-1R overexpression increases the protein levels involved in EMT, VM, and the PI3K/AKT signaling pathway in 6–10B cells. (A and B) Western blot and quantitative analysis revealed the expression of EMT and VM-related proteins (HIF-1α, MMP-2, MMP-9, and E-cadherin), as well as the PI3K/AKT signaling pathway (PI3K, AKT, p-AKT). Data as mean ± SD (n=3). ***p< 0.001. (C) Interaction network of CSF-1R with PIK3AP1, PIK3CA, PIK3CB, PIK3CD and PIK3CG proteins predicted by string website.

We searched the string database and predicted the existence of protein interactions of CSF-1R with PIK3AP1, PIK3CA, PIK3CB, PIK3CD, and PIK3CG (Figure 5C). Moreover, the PI3K/Akt signaling axis regulates EMT, which is linked to tumor development [26]. Consequently, we explored changes in the PI3K/AKT signaling axis. Western blotting revealed that PI3K, AKT and p-AKT expressions were raised in the LV-CSF-1R group (Figure 5B). In conclusion, these findings imply that overexpression of CSF-1R can activate this signaling pathway in NPC cells, which mediates EMT to promote VM formation (Figure 6).

Figure 6:

Schematic diagram showing that CSF-1R overexpression induces epithelial-mesenchymal transition promoted vasculogenic mimicry through the PI3K/AKT signaling pathways in nasopharyngeal carcinoma.