The microRNAs landscape of luminal B breast cancer cells in a three-dimensional microenvironment

TL;DR

This study explores how microRNAs behave in 3D breast cancer cell cultures, revealing new regulatory networks and potential biomarkers for luminal breast cancer.

Contribution

The study identifies miRNA–mRNA–lncRNA networks in 3D breast cancer cultures and links them to tumor-like pathways and potential diagnostic biomarkers.

Findings

3D culture reprograms miRNA expression, with 75 upregulated and 82 downregulated miRNAs compared to 2D.

Four key miRNAs (miR-92a-3p, miR-539-5p, miR-18a-5p, miR-130a-3p) are linked to tumor proliferation and survival pathways.

These miRNAs show diagnostic potential in TCGA datasets with AUC values of ~0.70–0.80.

Abstract

Three-dimensional (3D) culture captures key features of tumor architecture and microenvironmental signaling. Because microRNAs (miRNAs) are central post-transcriptional regulators, we asked whether 3D growth would reveal disease-relevant lncRNA-miRNA-mRNA programs in BT-474 luminal breast cancer cells.

BT-474 cells were profiled with GeneChip miRNA 4.0 microarrays to compare 3D versus 2D conditions. We then performed an integrative analysis consisting of: (i) extracting miRNA–mRNA interactions from miRNet and ENCORI (predicted and curated) for the differentially expressed miRNAs; (ii) identifying inverse miRNA–mRNA pairs (miRNA downregulated with target mRNA upregulated, or the reverse) using public mRNA data from BT-474 3D versus 2D cultures so that pairs were consistent with regulatory repression; and (iii) retaining only those miRNA–mRNA pairs that also appeared in miRNet/ENCORI. In a separate step, we examined public luminal breast tumor datasets versus BT-474 2D cultures to determine which 3D-identified miRNAs showed a similar direction of expression change in luminal tumors, thereby prioritizing signals with putative in vivo relevance. Finally, to enhance biological plausibility, we incorporated an lncRNA layer by leveraging their miRNA “sponge” (ceRNA) activity, and we assembled an integrated miRNA–mRNA–lncRNA network. Key miRNAs from this network were then evaluated for expression in external TCGA cohorts and for diagnostic potential using area-under-the-curve (AUC) analysis.

Our data reveal a reprogramming of miRNA expression between 3D and 2D cultures. Using a threshold of FC > 1.5 and p < 0.05, we found 75 miRNAs upregulated and 82 downregulated in 3D relative to 2D. To strengthen biological interpretation, we integrated our miRNA expression data with public mRNA profiles from BT-474 cells cultured in 3D versus 2D. We then cross-referenced interactions reported in miRNet and ENCORI, retaining only miRNA–mRNA pairs showing inverse regulation. This integrative layer yielded a pathway-level view in which 3D-modulated miRNAs collectively favored proliferative programs, ESR1/estrogen signaling, MAPK/JNK activity, cell-cycle control, and extracellular-matrix remodeling, with relative attenuation of EGFR/TGF-β signaling. Together, the inferred miRNA–mRNA network supports a shift toward a more tumor-like, transcriptionally active state in 3D. To reinforce translational relevance, we cross-referenced public luminal breast tumor datasets with BT-474 2D cultures and identified 14 miRNAs in common, with the same direction of change as observed in 3D cultures. From these, we assembled co-regulatory networks linking miRNAs to their mRNA targets and to lncRNAs. This analysis highlighted four key miRNA nodes: miR-92a-3p, miR-539-5p, miR-18a-5p, and miR-130a-3p, which are connected to 58 mRNAs and modulated by 12 lncRNAs. The hubs converge on pathways for growth and survival via the RTK–MAPK axis, estrogen signaling, and fibronectin/adhesion–mechanotransduction processes, depicting a proliferative, mechano-responsive phenotype. Additionally, these miRNAs emerged as potential biomarker candidates: in TCGA datasets, miR-539-5p, miR-18a-5p, and miR-130a-3p are repressed in luminal breast tumors compared to normal breast tissue, each showing moderate to good diagnostic performance (AUC ~ 0.70–0.80).

Overall, this study highlights the utility of 3D cultures in investigating the biology of breast cancer, emphasizing the critical interplay between lncRNAs, miRNAs, and mRNAs. Our findings provide insights into the discovery of novel potential biomarkers, underscoring the importance of integrating both in vitro 3D culture and in vivo expression data for a more accurate identification of miRNAs with possible applications in cancer therapies.

The online version contains supplementary material available at 10.1186/s41065-025-00586-2.

Figures