Endoplasmic reticulum stress inhibits 3D Matrigel-induced vasculogenic mimicry of breast cancer cells via TGF-β1/Smad2/3 and β-catenin signaling

Abstract

Endoplasmic reticulum (ER) stress is a cellular stress condition involving disturbance in the folding capacity of the ER caused by endogenous and exogenous factors. ER stress signaling pathways affect tumor malignant growth, angiogenesis and progression, and promote the antitumor effects of certain drugs. However, the impact of ER stress on the vasculogenic mimicry (VM) phenotype of cancer cells has not been well addressed. VM is a phenotype that mimics vasculogenesis by forming patterned tubular networks, which are related to stemness and aggressive behaviors of cancer cells. In this study, we used tunicamycin (TM), the unfolded protein response (UPR)-activating agent, to induce ER stress in aggressive triple-negative MDA-MB-231 breast cancer cells, which exhibit a VM phenotype in 3D Matrigel cultures. TM-induced ER stress was able to inhibit the VM phenotype. In addition to the tumor spheroid phenotype observed upon inhibiting the VM phenotype, we observed alterations in glycosylation of integrin β1, loss of VE-cadherin and a decrease in stem cell marker Bmi-1. Further study revealed decreased activated transforming growth factor β1, Smad2/3, Phospho-Smad2 and β-catenin. β-Catenin knockdown markedly inhibited the VM phenotype and resulted in the loss of VE-cadherin. The data suggest that the activation of ER stress inhibited VM phenotype formation of breast cancer cells via both the transforming growth factor β1/Smad2/3 and β-catenin signaling pathways. The discovery of prospective regulatory mechanisms involved in ER stress and VM in breast cancer could lead to more precisely targeted therapies that inhibit vessel formation and affect tumor progression.

Keywords: TGF-β1; breast cancer; endoplasmic reticulum stress; vasculogenic mimicry; β-catenin.

Fig. 1

VM formation of MDA‐MB‐231 breast cancer cells in 3DMs. (A) Morphology of MDA‐MB‐231 breast cancer cells in 3DMs by light microscope (left; scale bar: 200 μm) and hematoxylin and eosin staining (right; scale bar: 50 μm). (B) 3D reconstruction images of VM phenotype of MDA‐MB‐231 cells in 3D cultures by using a confocal z‐stacking program. Scale bars: 100 μm. (C) PAS staining image of VM phenotype of MDA‐MB‐231 cells in 3D culture. Scale bar: 100 μm. (D) Flow cytometric analysis of surface markers CD24 and CD44 in MDA‐MB‐231 cells. (E) Western blot analysis of VM marker protein VE‐cadherin in MDA‐MB‐231 cells from 2D Matrigel culture, 3DM, and 3D collagen culture (3DC) groups.

Fig. 2

Phenotypic changes of MDA‐MB‐231 cells in 3DM after TM treatment. (A) Light microscope images of MDA‐MB‐231 cells in 3DM and TM‐treated 3DM (3DM+TM). Scale bars: 200 μm. (B) F‐actin (red) and nuclear (blue) DAPI fluorescence images of MDA‐MB‐231 cells in 3DM and 3DM+TM. Scale bars: 100 μm. (C, D) Western blot analysis of VM marker protein VE‐cadherin and stem cell marker Bmi‐1 of MDA‐MB‐231 cells in 2D, 3DM and 3DM+TM. (E) Flow cytometric analysis of surface markers CD24 and CD44 in 3DM MDA‐MB‐231 cells. (F) Flow cytometric analysis of surface markers CD24 and CD44 in 3DM MDA‐MB‐231 cells after TM treatment.

Fig. 3

Changes in glycosylation of integrin β1 after TM treatment. (A) Integrin β1 (green), F‐actin (red) and nuclear (blue) DAPI fluorescence images and merged images of MDA‐MB‐231 cells in 3DM and TM‐treated 3DM (3DM+TM). Scale bars: 100 μm. (B) Western blot analysis of integrin β1 in MDA‐MB‐231 cells from 2D, 3DM and 3D collagen culture (3DC) groups. (C) Western blot analysis of integrin β1 in MDA‐MB‐231 cells before (3DM) after TM treatment (3DMT).

Fig. 4

TM‐induced ER stress activation and the changes of TGF‐β1/Smad2/3 and β‐catenin signaling. (A) Western blot analysis of Bip and CHOP of MDA‐MB‐231 cells in 2D, 3DM and 3DM+TM. (B) Concentrations of activated TGF‐β1 of MDA‐MB‐231 cells in 3DM and 3DM+TM (*P < 0.05). Statistical significance was determined for experimental data by using the Student's t‐test. Error bars indicate SEM. All experiments were repeated three times. (C) Western blot analysis of Smad2/3 and Phospho‐Smad2 of MDA‐MB‐231 cells in 2D, 3DM and 3DM+TM. (D, E) Western blot analysis of total β‐catenin, active β‐catenin (p‐Ser675) and inactive β‐catenin (p‐Ser33/37/Thr41) in 2D, 3DM and 3DM+TM.

Fig. 5

The effects of β‐catenin knockdown on the formation of VM and cellular behaviors. (A) β‐Catenin knockdown inhibited the VM phenotype and reduced the tubular networks of MDA‐MB‐231 cells in 3DM. Scale bars: 200 μm (top); 100 μm (bottom). (B) Western blot analysis of VE‐cadherin of scramble control and β‐catenin knockdown (shβ‐cat1 and shβ‐cat2) groups. (C) Pyrvinium significantly reduced VM formation of MDA‐MB‐231 cells in 3DMs. Scale bars: 200 μm. (D) IC₅₀ of 2D cultured MDA‐MB‐231 cells to TM after β‐catenin knockdown (*P < 0.05). (E) Wound healing assays revealed that knockdown of β‐catenin inhibited cells migration (*P < 0.05). Statistical significance was determined for experimental data by using the Student's t‐test. Error bars indicate SEM. All experiments were repeated three times.