miR-146a-5p impairs melanoma resistance to kinase inhibitors by targeting COX2 and regulating NFkB-mediated inflammatory mediators

Abstract

Background: Targeted therapy with BRAF and MEK inhibitors has improved the survival of patients with BRAF-mutated metastatic melanoma, but most patients relapse upon the onset of drug resistance induced by mechanisms including genetic and epigenetic events. Among the epigenetic alterations, microRNA perturbation is associated with the development of kinase inhibitor resistance. Here, we identified and studied the role of miR-146a-5p dysregulation in melanoma drug resistance.

Methods: The miR-146a-5p-regulated NFkB signaling network was identified in drug-resistant cell lines and melanoma tumor samples by expression profiling and knock-in and knock-out studies. A bioinformatic data analysis identified COX2 as a central gene regulated by miR-146a-5p and NFkB. The effects of miR-146a-5p/COX2 manipulation were studied in vitro in cell lines and with 3D cultures of treatment-resistant tumor explants from patients progressing during therapy.

Results: miR-146a-5p expression was inversely correlated with drug sensitivity and COX2 expression and was reduced in BRAF and MEK inhibitor-resistant melanoma cells and tissues. Forced miR-146a-5p expression reduced COX2 activity and significantly increased drug sensitivity by hampering prosurvival NFkB signaling, leading to reduced proliferation and enhanced apoptosis. Similar effects were obtained by inhibiting COX2 by celecoxib, a clinically approved COX2 inhibitor.

Conclusions: Deregulation of the miR-146a-5p/COX2 axis occurs in the development of melanoma resistance to targeted drugs in melanoma patients. This finding reveals novel targets for more effective combination treatment. Video Abstract.

Keywords: BRAF/MEK inhibitors; COX2; Melanoma resistance; miR-146a-5p; microRNA.

Fig. 1

Downregulation of miR-146a is associated with resistance to BRAF/MEKi in melanoma cell lines and its manipulation affects drug response and TRAIL-induced apoptosis. a Venn diagram illustrating common miR associated with BRAF/MEKi resistance in GSE141314, GSE67635 and GSE68841 datasets. b miR-146a shows significant downregulation in melanoma resistant cells (R) compared with their sensitive counterparts (S). logFC: log₂ Fold Change. P-value obtained from differential expression analysis performed with limma package. c Inverse correlation between miR-146a expression levels and IC50 values of combined BRAF/MEKi in melanoma cell lines (Spearman analysis). d Forced expression of miR-146a (+m-miR-146a) increases the effects of BRAF/MEKi treatment in LM16R, LM69 and LM70 cell lines, as shown by reduced cell growth and increased cell cytotoxicity and apoptosis, evaluated by CCK8, LDH and caspase 8 and 3/7 activity. LM69 and LM70 short term melanoma cultures were generated from treatment resistant melanoma lesions surgically excised from two patients. Based on qRT-PCR results, upon transfection miR-146a levels were up to 70-fold higher (range 76–1223) than the levels detectable in cells transfected with control oligos. Data are plotted as fold increase compared to cells transfected with mimic negative control. _*: p < 0.05, _**: p < 0.01, _***: p < 0.0001 by Student’s unpaired t test. e Forced expression of miR-146a increased the effect of treatment with sTRAIL. f Positive correlation between the levels of miR-146a expression and of activated caspase 3/7 induced by sTRAIL treatment (Spearman analysis). RLU: Relative Light Unit. Results shown are representative of 2 experiments performed in triplicate (d, e)

Fig. 2

miR-146a regulates NFkB-related signaling pathways associated with resistance and directly targets PTGS2. a Immunoblot showing reduced levels of NFkB, AKT, p70S6K, and ERK in melanoma cells transfected with synthetic miR-146a (m-miR-146a) compared to mimic negative control transfectants (left). Reduction of phosphorylated proteins in miR-146a transfectants calculated by signal quantification (right). b Reduced expression of miR-146a target genes in transfectants (m-miR-146a: miR-146a mimic; S: mimic scrambled control). c Expression levels of PTGS2 and PTGES genes in resistant cell lines (R) compared to their sensitive counterparts (S). d Positive correlation between PTGS2 expression and IC50 values of BRAF/MEKi in melanoma cell lines (Spearman analysis). e Higher levels of COX2 protein expression in resistant cells (R) compared to their sensitive counterparts (S). Specific signals were quantified and expressed as the ratio of COX2/actin intensity. f Reduction of COX2 protein expression in LM16R resistant cells transfected with miR-146a mimic (m-miR-146a) and increased levels in LM36 parental sensitive cells transfected with miR-146a inhibitor (i-miR-146a). Changes in COX2 expression were calculated by signal quantification. P values were calculated by Student’s unpaired t test in A and F, and by Mann-Whitney U test in B. _*: p < 0.05, _**: p < 0.01, _***: p < 0.0001

Fig. 3

miR-146a spoils resistance of melanoma cells by repressing COX2 expression. a Silencing of COX2 by siRNA transfection (siCOX2) enhanced cell growth inhibition and increased cell cytotoxicity and apoptosis upon treatment with BRAF/MEKi. b Combined treatment with celecoxib increased cell growth inhibition and cytotoxicity by BRAF/MEKi. ●: interaction index = 1. c Increased apoptotic signaling after combination treatment with celecoxib and BRAF/MEKi. Phosphorylated and cleaved apoptotic factors showing increased expression in cells treated with the drug combination compared to BRAF/MEKi treated cells (Fold change > 1.3) by apoptosis signaling antibody array analysis. d Reduction of PGE2 release in culture media by treatment with the COX2 inhibitors celecoxib and NS398, alone or in combination with BRAF/MEKi. All experiments were carried with the LM47R cell line. P values were calculated by Student’s unpaired t test in B, and by one-way ANOVA followed by Bonferroni correction in D. _*: p < 0.05, _**: p < 0.01, _***: p < 0.0001

Fig. 4

Treatment resistant tumors display miR-146a/COX2 axis deregulation. a COX2 immunostaining in tumors excised from patients progressing during BRAF/MEKi therapy (Post) compared to pre-therapy matched melanoma lesions (Pre). Scale bar: 10 μm. b Lower miR-146a and higher PTGS2 and PTGES expression levels in tumors progressing in patients during BRAF/MEKi therapy (Treated, n = 12) compared to tumors from unmatched untreated patients (Untreated, n = 31). _*: p < 0.05, _**: p < 0.01 by Mann-Whitney U test. c Heatmap showing the gene expression pattern of PTGS2, CXCL8, PTGES, IL6, VEGFA, CD274, MCL1 in progressing tumors (Post-treatment) compared to matched pre-therapy lesions (Pre-treatment) from seven metastatic melanoma patients receiving BRAFi therapy

Fig. 5

miR-146a overexpression reduces COX2 and increases drug sensitivity in 3D cultures from BRAF/MEKi-resistant tumors. a Overexpression of miR-146a upon transfection of specific mimic (m-miR-146a) in 3D tumor explants (left) downregulates COX2 expression compared to scrambled-transfected control (S) (right). b Reduced COX2 immunostaining in 3D cultures upon miR-146a forced expression (m-miR-146a) compared to mimic scrambled control (S). c Regulation of PTGS2 and miR-146a target genes (upper panel), and decreased release of CCL2, IL6, IL8 and VEGFA (lower panel) upon treatment with miR-146a mimic (m-miR-146a) and BRAF/MEKi in 3D cultures of resistant tumors in comparison to scrambled-transfected control (S). AU: arbitrary units. d PGE2 release in culture media from 3D tumor explants upon celecoxib treatment. _**: p < 0.01 by Student’s unpaired t test. e Effects of treatment with BRAF/MEKi combined with transfection of miR-146a mimic (m-miR-146a) or of scrambled control (S) in resistant tumors: reduction of COX2 and of Ki67 positive cells and increase of cleaved caspase 3 immunostaining. Scale bars: 10 μm

Fig. 6

Schematic representation of the miR-146a/COX2 axis associated with BRAF/MEKi resistance