Reprogramming miRNAs global expression orchestrates development of drug resistance in BRAF mutated melanoma

Abstract

Drug resistance imposes severe limitations to the efficacy of targeted therapy in BRAF-mutated metastatic melanoma. Although this issue has been mitigated by the development of combination therapies with BRAF plus MEK inhibitors, drug resistance inevitably occurs with time and results in clinical recurrences and untreatable disease. Hence, there is strong need of developing new combination therapies and non-invasive diagnostics for the early identification of drug-resistant patients. We report here that the development of drug resistance to BRAFi is dominated by a dynamic deregulation of a large population of miRNAs, leading to the alteration of cell intrinsic proliferation and survival pathways, as well as of proinflammatory and proangiogenic cues, where a prominent role is played by the miR-199b-5p/VEGF axis. Significant alterations of miRNA expression levels are detectable in tumor biopsies and plasma from patients after disease recurrence. Targeting these alterations blunts the development of drug resistance.

Fig. 1

Changes in the expression of the entire miRNAome characterize evolution of BRAFi-resistance in human melanoma cells. a Schematic representation of Nanostring study in which M14 and WM266 melanoma cells selected for two months with increasing doses of a BRAF inhibitor are subjected to whole miRNAome profiling. b Genomic DNA from M14 and WM266 evaluated by Sanger method confirming BRAF V600 mutational state. c, d Venn Diagrams showing that each selection step is characterized by a distinct set of miRNAs expression changes in M14 (c) and WM266 (d) cells. E, Principal Component Analysis confirming that changes of the entire miRNome expression (n = 800 miRNAs) characterizing different drug sensitivity states. f Venn Diagrams showing the common deregulated miRNAs at each step of the selection. g The number of pathways affected by commonly deregulated miRNAs increasing with stepwise drug exposures. h Bioinformatics analysis revealing the most relevant intracellular pathways affected by deregulated miRNAs

Fig. 2

Shared deregulated miRNAs of drug-resistant melanoma cells affect proliferation and apoptosis in the presence or not of a BRAFi. a, b Effects of miR-204-5p, miR-199b-5p (a), miR-4443 and miR-4488 (b) transient overexpression on cell viability (left panels) and on Caspase 3/7 activation (right panels) in the presence or not of a BRAFi in M14 and WM266 cells. c Specific silencing of miR-4443 and miR-4488 affecting M14S and M14R cell colony formation. d Transient enforced expression of miR-204-5p and miR-199b-5p reducing M14R cell colony formation

Fig. 3

Shared deregulated miRNAs of drug-resistant melanoma cells act on cell intrinsic pathways. a FACS analysis showing that the overexpression of both miR-204-5p and miR-199b-5p induce M14S and WM266S cell apoptosis. b TCGA Skin Cutaneous Melanoma data revealing high co-expression of miR-204-5p/miR-199b-5p associated with a better survival probability (p = 0.0236). c Transient overexpression of miR-204-5p in M14S cells affecting Bcl-2 expression at mRNA (left panel) and protein (right panel) levels. d Enforced expression of miR-199b-5p reducing VEGFA and HIF-1α expression levels in M14S (left panel) and WM266S (right panel). e Venn Diagrams revealing the commonly deregulated target mRNAs of miR-4443 and miR-4488. f miR-4443 and miR-4488 overexpression reducing PTPN14 mRNA in M14S cells. g miR-4443 but not miR-4488 overexpression affecting TIMP2 mRNA in M14S cells

Fig. 4

Simultaneous cotargeting of deregulated miRNAs affects drug-resistant melanoma cell growth. a Different miRNA action on non-redundant pathways relevant for melanoma progression. b–g miR-204-5p + miR199b-5p (b), amiR-4443 + amiR-4488 (c), miR-204-5p + amiR-4443 (d), miR-204-5p + amiR-4488 (e), miR-199b-5p + amiR-4443 (f) and miR-199b-5p + amiR-4488 (g) strongly reducing M14R cell colony formation as compared to single treatments. h Co-expression of miR-204-5p and miR-199b-5p mimics affecting A375DR cell colony formation as compared to single treatments (left panel) and even further when combined with miR-579-3p (right panel)

Fig. 5

Conditioned media from drug-resistant melanoma cells is enriched in several proinflammatory and pro-angiogenic factors. a The levels of 27 cytokines determined in the supernatants from M14S and WM266SS cells and from their resistant counterparts. b, c WM266R and M14R cells conditioned media overproducing several cytokines and chemokines (fold change significance greater than 1.3) divided in three distinct groups with a high, medium and low degree of upregulation. d Cell index (left panel) and Slope induction (right panel) results obtained through xCELLigence Real Time Cell Analysis showing that conditioned media from WM266R melanoma cells inducing cell migration as compared to cell media from sensitive counterparts and CTRL media

Fig. 6

WM266R cells sustain pro-angiogenic stimuli through VEGFA release caused by downregulation of the oncosuppressive miR-199b-5p. a Conditioned media (CM) of drug sensitive vs. drug resistant WM266 cells inducing endothelial tube formation in human umbilical vein endothelial cells (HUVEC) plated on matrigel. b Non-contact co-culture system in which sensitive or BRAFi-resistant WM266 cells grown separated from HUVECs plated on matrigel in an intercup chamber exposed to their secretion products for 4 h. c Enhancing of cord-like structures of HUVECs co-cultured with WM266R cells as compared to HUVECs co-cultured with parental WM266S cells. Scale bar 100 μm. d Avastin (25 μg) or Pazopanib (5μg) inhibing tube formation induced by CM from WM266R melanoma cells; VEGF (200ng/ml) used as positive control. e miR-199b-5p enforced expression reducing the capability to induce tube formation of CM-WM266R cells. f Overexpression of miR-199b-5p inhibiting VEGF protein in WM266R cells observed through Western Blot analysis

Fig. 7

miRNAs signatures characterize the acquisition of MAPKi-resistance in BRAF-mutated melanoma patients. a Total RNA extracted from FFPE samples from 14 matched tumour samples before initiation of targeted therapy and after disease progression PD. b qRT-PCR results as box-whisker plots showing miR-204-5p/miR-199b-5p downregulation and miR-4443/miR-4488 upregulation in relapsing tumours. c Heat-map indicating the correlation of miR-199b-5p and miR-204-5p and anti-correlation to miR-4443 and miR-4488 (identified by yellow squares and by dark blue squares). d Receiver operating characteristic (ROC) curves estimating the predictive value of miRNA cormbinations (from FFPE samples) as markers of drug resistance. e Circulating microRNA extracted from the plasma of 25 melanoma patients before the beginning of therapy and at disease progression. f miR-199b-5p downregulation (higher panel) in the plasma of melanoma patients post-MAPKi treatment as compared to the plasma from untreated patients (pW = 0.17; pt = 0.19); miR-4488 upregulation (lower panel) in patients after MAPKi treatment (pW = 0.023; pt = 0.0039). g ROC curves estimating the predictive value of miR-199b-5p + miR-4488 cormbination as marker of drug resistance (from plasma samples). pt Student’s t test, pW Wilcoxon matched pairs test; pW = Wilcoxon matched pairs test. G, ROC curves estimating the predictive value of miR-199b-5p+miR-4488 cormbination as marker of drug resistance (from plasma samples)