Early myeloid-derived suppressor cells accelerate epithelial-mesenchymal transition by downregulating ARID1A in luminal A breast cancer

Abstract

Early myeloid-derived suppressor cells (eMDSCs) are a newly characterized subclass of MDSCs, which exhibit more potent immunosuppressive capacity than classical MDSCs. Previously, we found high eMDSCs infiltration was correlated with poor prognosis of breast cancer, though the regulatory mechanisms have not been fully understood. Here, we constructed a 21-gene signature to evaluate the status of eMDSCs infiltration within breast cancer tissues and found that highly infiltrated eMDSCs affected the prognosis of breast cancer patients, especially in luminal A subtype. We also found that eMDSCs promoted epithelial-mesenchymal transition (EMT) and accelerated cell migration and invasion in vitro. Meanwhile, eMDSCs significantly downregulated ARID1A expression in luminal A breast cancer, which was closely associated with EMT and was an important prognostic factor in breast cancer patients. Moreover, significant changes of EMT-related genes were detected in luminal A breast cancer cells after co-cultured with eMDSCs or ARID1A knock-down and overexpression of ARID1A significantly reversed this procedure. These results implied that eMDSCs might suppress the ARID1A expression to promote EMT in luminal A breast cancer cells, which might provide a new light on developing novel treatment regimens for relapsed luminal A breast cancer after conventional therapies.

Keywords: ARID1A; breast cancer; early-stage myeloid-derived suppressor cells; epithelial-mesenchymal transition; prognostic factor.

FIGURE 1

The 21-genes signature is able to predict the infiltration of eMDSCs in situ.(A) The relative expression of 50 genes that can separate the three groups of eMDSCs^SOCS3KO, CD11b⁺Gr-1⁺ and eMDSCs^fl/fl cells. (B) The relative expression of 21 genes in human PBMC-derived eMDSCs was confirmed by RT-qPCR. The transcriptional patterns of 21 genes are comparable both in humans and mice. CD33⁺ myeloid progenitors isolated from healthy PBMCs were co-cultured with MDA-MB-231 breast cancer cells to induce eMDSCs. (C) T-SNE analysis was used to visually display the classification results and the consistency between the 21-genes signature prediction and immunohistochemical observation was as high as 85.7% (12/14). (D) Kaplan-Meier survival analysis of differences between eMDSCs^high and eMDSCs^low groups in TCGA breast cancer cohort showed that the OS was longer in the eMDSCs^low group than the eMDSCs^high group. (E) Heatmap of the 21 genes expression profiles in the TCGA breast cancer cohort.

FIGURE 2

Highly-infiltrated eMDSCs are significantly correlated with poor prognosis in breast cancer patients (A) Multispectral IF staining analysis showed the infiltration of CD33⁺ cells in tumor sections from 280 cases of breast cancer patients. (B–H) Kaplan-Meier survival analysis of overall survival between eMDSCs^high and eMDSCs^low groups among 4 different subtypes of 280 cases of breast cancer patients.

FIGURE 3

eMDSCs promote migration and invasion of luminal A breast cancer cells via stimulating EMT (A) KEGG enrichment analysis of the differential expressed genes between eMDSCs^high and eMDSCs^low groups divided by the 21-genes signature in TCGA breast cancer cohort. (B) Distribution of FPKM values for TCGA breast cancer samples between eMDSCs^high and eMDSCs^low groups for the three EMT-related genes. (C) Immunoblotting was used to assess the proteins of E-Cadherin, Vimentin, and SNAIL in the indicated breast cancer cells. (D) Immunofluorescence was used to examine the levels of E-Cadherin in EO771, T47D, and MCF7 cells cocultured with eMDSCs. (E) Transwell assays revealed that eMDSCs dramatically improved the migratory and invasion capacities of EO771, T47D, and MCF7 cells.

FIGURE 4

eMDSCs downregulate the expression of ARID1A in luminal A breast cancer cells (A) Gene ontology enrichment analysis of the strongest differential expressed proteins between eMDSCs^high and eMDSCs^low groups. (B) A multispectral IF staining assay showed that the changes of ARID1A expression between eMDSCs^high and eMDSCs^low groups in 140 primary breast cancer samples from cohort 1. (C–G) The ARID1A expression between different groups in (B) was analyzed.

FIGURE 5

eMDSCs accelerate epithelial-mesenchymal transition in luminal A breast cancer cells by downregulating ARID1A (A,B) The mRNA and protein level of ARID1A were determined by RT-PCR and immunoblotting in EO771, T47D and MCF7 cells after cocultured with eMDSCs. (C) Quantitative RT-PCR showing expression changes of ARID1A and EMT-related genes in EO771, T47D and MCF7 cells upon ARID1A knocked-down with the transfection of siRNA or vehicle. (D) Quantitative RT-PCR showing expression changes of ARID1A and EMT-related genes in EO771, T47D, and MCF7 cells cocultured with eMDSCs upon ARID1A overexpression plasmids or vehicle. *p < 0.05, **p < 0.01, ***p < 0.001.

FIGURE 6

eMDSCs promote tumor growth via accelerating EMT process (A) Representative images for xenografts model in SOCS3^fl/fl and SOCS3^KO mice. (B,C) The weight and volume of tumors in the SOCS3^KO group were significantly greater than those in the control group. (D) There was no discernible change in body weight between the SOCS3^KO group and the negative control group. (E) The percentages of eMDSCs were detected by FCM in primary tumor tissues. (F) The images show immunohistochemistry staining for ARID1A, E-Cadherin, and Vimentin in xenografts. *p < 0.05, **p < 0.01.