(A) Heatmap of differentially expressed miRNAs in melanoma patient tissue samples comprising primary melanomas, lymph node metastases and distant cutaneous metastases (ou) (n=24), generated using expression data from TaqMan® low density miRNA arrays. Colors represent normalized Ct values (blue for high Ct values indicates low expression and red for low Ct values indicates high expression of miRNAs). Dendrograms are based on hierarchical clustering with Euclidean distance. (B) miR-638 expression in PM and MM and (C) PM of varying thickness (mm). (D) miR-126* expression in PM and MM and (E) PM of varying thickness (mm). Paired t-test and Pearson's coefficient (Pearson's r) were used to determine significance of correlation between miR-638 expression and tumor thickness. (F, G) Expression levels of miR-638 were analysed in primary melanocytes (n=3) in comparison with (F) PM and MM tissue samples (n=27). (G) miR-638 expression in 9 different melanoma cell lines (HT144, 1F6, Mewo, SK-Mel-147, SK-Mel-19, SK-Mel-28, SK-Mel-29, A375 and BRO). All miRNA expression analyses were performed using TaqMan® qRT-PCR technology. RNU48 expression was used as universal reference control for miRNAs.

(A) XTT cell proliferation assay using SK-Mel-28 cells transiently overexpressing a scrambled-control or miR-638 (mean ± S.E.M; n=3). (B) Cell cycle analysis for SK-Mel-28 cells transiently overexpressing a scrambled-control or miR-638. (mean ± S.E.M.; n=3) (C) Colony formation assay after low density seeding of SK-Mel-147 cells overexpressing a scrambled-control or miR-638 (mean ± S.E.M; n=3). (D) Soft agar assays using SK-Mel-28 cells transiently overexpressing a scrambled-control or miR-638 (mean ± S.E.M; n=3). Arrows indicate colonies of growing cells. (E) Migration assays were performed for SK-Mel-147 cells transiently overexpressing a scrambled-control or miR-638. Microscopic pictures were taken at indicated time points (mean ± S.E.M, n=3). NT; scrambled. (F-H) In vivo metastasis (lung colonization) experiments with stably miR-638 transduced or control transduced SK-Mel-147 cells injected into the lateral tail vein of NSG mice (n=5 per group). (F) Macroscopic pictures of mouse lungs 21 days after tail vein injection. (G) Hematoxylin&Eosin (H&E)-stained sections of lung metastases (right; magnification x4). (H) Percentage of lung area covered by metastatic tumor areas analysed in the H&E stained lung sections. A minimum of 3 lung sections per mouse were quantitated (mean ± S.E.M.). All biological assays were performed in triplicates and repeated twice upon individual transfections and assay measurement.

(A) Melanoma cells were transiently transfected with miR-638, antagomir-638 or control-antagomiR and analyzed using whole-genome cDNA microarrays. The Venn diagram at the top displays the overlap between down-regulated and up-regulated genes upon miR-638 and antagomiR-638 overexpression, respectively (n = 3,199). The middle Venn diagram displays the intersection of miRWalk database predicted miR-638 targets and strongly differentially regulated mRNAs (total log₁₀ fold change (FC) > 2; n = 86). Finally, a subset of targets predicted by at least three different algorithms as high-confidence targets of miR-638 regulation was identified and used in subsequent analyses (n = 30). (B) Heatmap of differentially regulated putative target genes for miR-638. Colors represent fold changes of relative gene expressions of miR-638 or antagomiR-638-transfected cells in comparison with mock transfected control cells (blue color indicates high expression and red color indicates low expression). (C) TP53INP2 mRNA expression analysis for primary melanocytes (P-mel 1 and 2), human fibroblasts (Hu.Fib), and melanoma cell lines, respectively, using TaqMan® gene expression assays. Normalized Ct values were quantitated and compared with the mean TP53INP2 expression values, respectively, across all cell lines. Results are expressed as percentage of relative expression. (D) SK-Mel-147 cells were transfected with a scrambled-control (Co) or miR-638. Fourty eight hours after transfection TP53INP2 mRNA expression was analysed using TaqMan® gene expression assays. GAPDH was used as reference control (mean ± S.E.M). (E) SK-Mel-147 melanoma cells were transfected with scrambled-control (Co) or miR-638. After 48 h of transfection, protein expression for TP53INP2 was analysed using immunoblotting. β-actin was used as loading control. (F) Reporter assay in SK-Mel-147 cells co-transfected with a scrambled miRNA control or miR-638 and with either wild type or single (1X) or double (2X) mutated TP53INP2 3′-UTR cloned in a dual-luciferase constructs (mean ± S.E.M; n= 4). (G) XTT cell proliferation assay using SK-Mel-147 cells transiently overexpressing control plasmid or TP53INP2 cDNA co-transfected with a control miRNA or miR-638 (mean ± S.E.M; n=3). The UV-absorptions were measured at 24 h and 48 h at 492 nm (H) Matrigel invasion assays were performed for SK-Mel-147 cells transiently overexpressing a control plasmid or TP53INP2 cDNA co-transfected with a control miRNA or miR-638. Microscopic pictures were taken at 48 h (mean ± S.E.M, n=3). All biological assays were performed in triplicates and repeated twice individual transfections and assay measurements.

(A) Genes de-repressed by antagomiR-638 treatment of melanoma cells were subjected to functional enrichment analysis, using DAVID bioinformatics tool. Different pathway candidates regulated by miR-638, including p53 signalling and lysosomal pathways, were identified. (B) Microscopic pictures of SK-Mel-147 cells overexpressing control-antagomiR (Co) or antagomiR-638 (Antag-638), showing confluency of melanoma cells after 72 h of transfection. (C, D) Human SK-Mel-147 melanoma cells were transfected with control-antagomiR (Co), or antagomiR-638 (Antag-638). At indicated time points, cells were stained for annexin-V/propidium iodide (PI), and apoptotic cells were analysed by flow cytometry. (C) Representative histograms indicating the proportion of annexin V/PI-stained and unstained SK-Mel-147 cells. (D) Graphs indicate percentage of dead cells (mean ± S.E.M). (E) Human SK-Mel-147 cells were transfected with control-antagomiR, antagomiR-638. Expression levels of the indicated apoptosis-related proteins were analysed by immunoblotting. (F) Fluorescent lysosomal staining (lysotracker red) of SK-Mel-147 cells transfected with control-antagomiR or antagomiR-638. Cell nuclei are stained with DAPI (x40). (G) SK-Mel-28 cells co-transfected with an EGFP-LC3B containing construct and a control-antagomiR or antagomiR-638. After 24 h of transfection cells were either left untreated or treated with 100 nM rapamycin for another 24 h, after which they were fixed and microscopic images were taken (x60). (H) The percentage of EGFP-LC3B puncta-positive cells was quantitated (mean ± S.E.M). All biological assays were performed in triplicates and repeated twice upon individual transfections.

(A) Putative transcription factor binding sites in the 10 kb region upstream of the miR-638 transcription start site were analysed using the MIR@NT@N V1.2.1 framework. The number and percentage of binding sites at methylation-prone CpG islands (CGI) are also included. (B) SK-Mel-147 melanoma cell lines were transfected with scrambled-control (Co) or a TFAP2A-specific siRNA (TFAP2A siRNA). Expression of miR-638 was measured using TaqMan® qRT-PCR. (C) SK-Mel-147 cells were transfected with the indicated oligomers. One day after transfection, the transfected cells were treated with 3 μM of 5-Aza for 24 h. At this point, complete growth medium was replenished and cells were incubated for another 24 h. Expression of miR-638 was measured using TaqMan® qRT-PCR (mean ± S.E.M.). RNU48 was used as a reference control. (D) Chromatin immunoprecipitation (ChiP) was performed with DMSO or Aza-treated SK-Mel-147 melanoma cell lysates using specific antibodies against AP-2α or isotype-IgG as control. The precipitated DNA was used to amplify a 1 kb sequence upstream of miR-638 transcriptional start site using specific primer pairs. (E) SK-Mel-147 cells were transfected with scrambled-control, TFAP2A siRNA or miR-638. Fourty eight hours after transfection, TFAP2A mRNA expression was analysed using TaqMan® gene expression assays (mean ± S.E.M.). (F) AP-2α protein levels were analysed using immunoblotting after transfection of SK-Mel-147 cells with the indicated oligomers. β-actin (Actin) was used as the loading control. All biological assays were performed in triplicates and repeated twice upon individual transfections.

(A) A regulatory map describing the miR-638-mediated regulation of apoptosis through its interaction with AP-2α and p53. (B) Model simulation showing the dependency of miR-638 (blue) and AP-2α (green) expression on the methylation of the AP-2α binding site. The simulations suggest an “all-or-nothing” transition in expression levels of miR-638 and AP-2α for intermediate methylation levels. This shift can affect the expression of apoptosis-related genes whose expression is controlled by p53 and AP-2α, here indicated as Apop_Tgt (red, scaled 5-fold to facilitate visualization). (C) Simulation representing the expression of pro-apoptotic targets of p53 and AP-2α under increasing levels of cellular stress (CS) for low (blue) and high (green) methylation levels in the AP-2α binding site. (D) Surface plot representing steady-state levels of pro-apoptotic targets determined by varying CS and AP-2α binding site methylation levels.