Elucidating the Role of MicroRNA-18a in Propelling a Hybrid Epithelial-Mesenchymal Phenotype and Driving Malignant Progression in ER-Negative Breast Cancer

Abstract

Epigenetic alterations that lead to differential expression of microRNAs (miRNAs/miR) are known to regulate tumour cell states, epithelial-mesenchymal transition (EMT) and the progression to metastasis in breast cancer. This study explores the key contribution of miRNA-18a in mediating a hybrid E/M cell state that is pivotal to the malignant transformation and tumour progression in the aggressive ER-negative subtype of breast cancer. The expression status and associated effects of miR-18a were evaluated in patient-derived breast tumour samples in combination with gene expression data from public datasets, and further validated in in vitro and in vivo breast cancer model systems. The clinical relevance of the study findings was corroborated against human breast tumour specimens (n = 446 patients). The down-regulated expression of miR-18a observed in ER-negative tumours was found to drive the enrichment of hybrid epithelial/mesenchymal (E/M) cells with luminal attributes, enhanced traits of migration, stemness, drug-resistance and immunosuppression. Further analysis of the miR-18a targets highlighted possible hypoxia-inducible factor 1-alpha (HIF-1α)-mediated signalling in these tumours. This is a foremost report that validates the dual role of miR-18a in breast cancer that is subtype-specific based on hormone receptor expression. The study also features a novel association of low miR-18a levels and subsequent enrichment of hybrid E/M cells, increased migration and stemness in a subgroup of ER-negative tumours that may be attributed to HIF-1α mediated signalling. The results highlight the possibility of stratifying the ER-negative disease into clinically relevant groups by analysing miRNA signatures.

Keywords: ER-negative breast cancer; chemoresistance; epithelial–mesenchymal transition; hybrid E/M phenotype; microRNA-18a; stem-like cells.

Figure 1

Reduced levels of miR-18a enriches for hybrid EMT/luminal-basal phenotype in ER-negative breast cancer. (A) Expression levels of miR-18a transcripts in ER-negative (n = 105) and ER-positive (n = 170) tumours (part of our cohort) as examined by q-PCR. (B) Change in protein expression of TNFAIP3 between MDA-MB-468/miR-18a/cont and MDA-MB-468/miR-18a/inh and CD49f, Keratin 19 between MDA-MB-231/miR-18a/cont and MDA-MB-231/miR-18a/inh cells. (C) Expression levels of luminality- and basality-associated genes in miR-18a/low tumours of the TCGA and METABRIC cohorts when compared to miR-18a/high tumours in the ER-negative subtype. (D) Mutational spectrum analysis depicting higher PIK3CA mutation load in miR-18a/low in comparison to miR-18a/high ER-negative tumours of METABRIC cohort. (E) Gene signature-based computational analysis characterising luminal, basal, epithelial and mesenchymal scores/signatures in miR-18a/low and miR-18a/high, ER-negative tumours of the TCGA and METABRIC cohorts. (F) Correlation analysis between miR-18a and PGR transcript levels in the miR-18a/high (n = 24) and miR-18a/low (n = 30) ER-negative tumours (part of our cohort) as examined by q-PCR. (G) Representative immunophenotyping images depicting the expression levels of CD44 and CD24 in MDA-MB-468/miR-18a/cont and MDA-MB-468/miR-18a/inh cells as assessed by flow cytometry. (H) Quantitative assessment of CD44 and CD24 expression in MDA-MB-468/miR-18a/cont and MDA-MB-468/miR-18a/inh as assessed by flow cytometry (cumulative result from three independent trials). (I) Representative immunofluorescence images demonstrating the expression of E-cadherin and Vimentin in MDA-MB-468/miR-18a/cont and MDA-MB-468/miR-18a/inh. Arrows represent hybrid epithelial/mesenchymal cells. (J) Quantitative assessment of E-cadherin and Vimentin expression in MDA-MB-468/miR-18a/cont and MDA-MB-468/miR-18a/inh.

Figure 2

Association of reduced miR-18a levels with proliferation and EMT characteristics in ER-negative breast cancer. (A) Association of Ki67 index and miR-18a levels in miR-18a/high and miR-18a/low tumours and (B) association of proliferation probability score and miR-18a in miR-18a/high and miR-18a/low tumours of our cohort. (C) Change in protein expression levels of EMT-associated proteins in MDA-MB-231/miR-18a/cont and MDA-MB-231/miR-18a/inh cells. (D,E) Migratory ability as assessed by wound healing assay in MDA-MB-468/miR-18a/cont vs. MDA-MB-468/miR-18a/inh and MDA-MB-231/miR-18a/cont vs. MDA-MB-231/miR-18a/inh, respectively. (F) Percentage of migration from three independent trials in MDA-MB-468/miR-18a/cont vs. MDA-MB-468/miR-18a/inh and MDA-MB-231/miR-18a/cont vs. MB-231/miR-18a/inh. (G) Association of ZEB2 transcript levels and miR-18a in miR-18a/high and miR-18a/low tumours as assessed by q-PCR. (H) DEGs associated with EMT and cell proliferation in miR-18a/low, ER-negative tumours of TCGA. (I,J) Functional enrichment of DEGs depicting up-regulated and down-regulated pathways in miR-18a/low, ER-negative tumours of TCGA and METABRIC cohorts, respectively. (K,L) Evaluation of the levels of the EMT score derived from a pan-cancer 77 EMT gene signature in miR-18a/high and miR-18a/low, ER-negative tumours of TCGA and METABRIC cohorts, respectively. (M,N) Heat map depicting expression of genes up-regulated and down-regulated during the process of EMT (derived from EMT core list of 130 genes) in miR-18a/low and miR-18a/high groups of ER-negative tumours respectively of TCGA and METABRIC cohorts.

Figure 3

Association of low miR-18a levels with drug response and stemness acquisition in ER-negative breast cancer. (A) Fold change of DEGs associated with drug transporter gene family (ATP-binding cassette transporter) in miR-18a/low, ER-negative tumours of TCGA cohort. (B) Functional enrichment of DEGs depicting up-regulated pathways in miR-18a/low, ER-negative tumours of TCGA. (C) Representative IHC images of Integrin β3-stained sections of residual tumours post-NACT. (D) Association of Integrin β3 protein and miR-18a in ER-negative residual tumours post-NACT. (E) Cell viability of MDA-MB-468/miR-18a/cont and MDA-MB-468/miR-18a/inh after 200 µM paclitaxel treatment. (F) Association of ITGB3 and miR-18a transcript levels and (G) association of stemness score and miR-18a transcript levels in miR-18a/low treatment naïve primary ER-negative tumours as assessed by q-PCR. (H) Change in expression levels of ALDH1A1 between MDA-MB-468/miR-18a/cont and MDA-MB-468/miR-18a/inh. (I) Percentage of cell viability of ALDH1A1/miR-18a/cont, ALDH1A1/miR-18a/inh, DsRed2N1/miR-18a/cont and DsRed2N1/miR-18a/inh groups in MDA-MB-231 after paclitaxel treatment across doses ranging from 10–200 µM. (J) Representative images of spheres obtained from MDA-MB-468/miR-18a/cont and MDA-MB-468/miR-18a/inh cells. (K) Table depicting clonogenicity (stem cell frequency) calculated from extreme limiting dilution assay. (L) Graph depicting clonogenicity from extreme limiting dilution assay performed on MDA-MB-468/miR-18a/cont and MDA-MB-468/miR-18a/inh groups across the same initial seeding dose range of 1–1000 cells.

Figure 4

Correlation of low miR-18a levels with immune suppression in ER-negative breast cancer. (A–C) ESTIMATE-generated Stromal score, Immune score and Estimate score for miR-18a/high and miR-18a/low tumours of TCGA cohort. (D) Graph representing proportion of claudin-low and basal tumour subtypes in miR-18a/low and miR-18a/high, ER-negative tumours of METABRIC series. (E,F) CIBERSORT analysis depicting the proportions of M2 and M1 macrophages in miR-18a/high and miR-18a/low tumours. (G,H) CIBERSORT analysis depicting the proportions of T follicular helper cells in miR-18a/high and miR-18a/low, ER-negative tumours of TCGA and METABRIC cohorts, respectively. (I,J) ImmuneCellAI analysis depicting the proportions of Th2 (Type 2 helper cells) and Tr1 (Type 1 regulatory T cells) in miR-18a/high and miR-18a/low, ER-negative tumours of TCGA series respectively.

Figure 5

Effect of miR-18a inhibition on mice-induced tumours in vivo and association with HIF1A expression. (A) Kaplan–Meier analysis depicting disease-free survival in ER-positive/miR-18a/low, ER-positive/miR-18a/high, ER-negative/miR-18a/low and ER-negative/miR-18a/high tumours of the METABRIC cohort. (B) Expression of HIF1A in ER-positive/miR-18a/low, ER-positive/miR-18a/high, ER-negative/miR-18a/low and ER-negative/miR-18a/high tumours of the TCGA cohort. (C) Increased expression levels of HIF-1α in MDA-MB-468/miR-18a/inh in comparison to MDA-MB-468/miR-18a/cont. (D) Images of tumours harvested from mice injected with miR-18a/antagomiR cells and antagomiR negative control cells post-21-days of injection. (E) Box plot depicting the average weight of the tumours harvested from miR-18a/antagomiR and the antagomiR negative control mice. (F,G) Immunofluorescence to demonstrate the dual positivity of E-cadherin and Vimentin in tumour sections from miR-18a/antagomiR tumour and absence of dual positivity of E-cadherin and Vimentin in antagomiR negative control tumour. (H) Quantitative analysis of immunofluorescence for expression of Vimentin and E-cadherin in MDA-MB-468/miR-18a/cont, MDA-MB-468/miR-18a/inh and MDA-MB-468/miR-18a/inh + CAY10585. (I) Percentage of migration as measured by wound healing assay in MDA-MB-468/miR-18a/cont, MDA-MB-468/miR-18a/inh and MDA-MB-468/miR-18a/inh + CAY10585.