SETD6 mediates selective interaction and genomic occupancy of BRD4 and MITF in melanoma cells

Abstract

Aberrant transcriptional programs mediate malignant transformation of melanoma, the most aggressive form of skin cancer. The lysine methyltransferase SETD6 has been implicated in regulating transcription, cell adhesion, migration, and other processes in various cancers; however its role in melanoma remains unexplored. We recently reported that SETD6 monomethylates the BRD4 at K99 to selectively regulate transcription of genes involved in mRNA (messenger RNA) translation. Here, we observed that BRD4 methylation at K99 by SETD6 occurs in melanoma cells. Knockout of SETD6 or a point mutation at BRD4-K99 disrupts BRD4 genomic occupancy. In addition, we show that SETD6 interacts with MITF, a master transcription factor in melanocytes and melanoma, and influences the genomic distribution of MITF. Mechanistically, we uncover a novel chromatin-localized interaction between BRD4 and MITF in melanoma. Our data suggest that BRD4 binds MITF in melanoma cells and that this interaction is dependent on both SETD6-mediated methylation of BRD4 and MITF acetylation. This chromatin complex plays a pivotal role in selective recruitment of BRD4 and MITF to different genomic loci in melanoma cells.

Graphical Abstract

Figure 1.

SETD6 regulates gene expression and cellular properties of melanoma cells. (A) Kaplan–Meier survival curve for high versus low levels of SETD6 in melanoma patients (TCGA database, P= .039), created using R2 genomics analysis and visualization platform (https://hgserver1.amc.nl/). (B) Heatmap showing up- and downregulated genes from RNA-sequencing analysis of two SETD6 control and four KO SKmel147 cells independent clones. Yellow and blue colors represent high and low expression levels, respectively. (C) Selected pathways from GO analysis of the DEGs were analyzed. Circle size represents the count of DEGs related to each pathway. (D) Colony-formation assay for SETD6 CT and KO SKmel147 cells. Images of colonies stained with crystal violet and crystal violet-stained cells were dissolved in 2% SDS and the absorbance at 550 nm was measured, whereby error bars represent the SEM. Statistical analysis was performed for two experimental repeats using one-way ANOVA (**P< .01). (E) Adhesion assay for SKmel147 parental, SETD6 CRISPR CT, and KO cells. Adherent cells stained with crystal violet, dissolved, and analyzed as described in panel (D).

Figure 2.

SETD6 interacts with BRD4 and methylates it in melanoma cells. (A) Kaplan–Meier survival curve for high versus low levels of BRD4 in melanoma patients (TCGA database, P= 4.27e−07) created using R2 genomics analysis and visualization platform. (B) Chromatin extract from SKmel147 cells were immunoprecipitated with anti-BRD4 antibody, followed by WB with the indicated antibodies. Input: levels of BRD4, SETD6, and histone-3 (loading control) in the total chromatin extracts. (C) Cellular methylation assay whereby SETD6 CT and KO cells were transfected with Flag BRD4 WT or Flag BRD4 K99R plasmids. Cell lysates were immunoprecipitated with preconjugated pan-methyl A/G agarose beads, and proteins in the immunoprecipitate and input samples were detected by WB with indicated antibodies. (D) WB analysis for CT and two SETD6 KO cells (KO1 and KO3) with the indicated antibodies. BRD4-K99me1 (U292-FT): an antibody that specifically recognizes methylated BRD4 at K99. (E) Colony-formation assay for SKmel147 cells stably expressing empty, Flag-BRD4 WT, and Flag-BRD4 K99R. Error bars represent the SEM. Statistical analysis was performed for two experimental repeats using one-way ANOVA (ns = not significant, *P< .05). (F) Adhesion assay for SKmel147 cells stably expressing empty, Flag-BRD4 WT, and Flag-BRD4 K99R. Adherent cells stained with crystal violet, dissolved, and analyzed as described in panel (F) (***P< .001, ****P< .0001).

Figure 3.

BRD4 K99 methylation by SETD6 affects its genomic distribution. (A) Heatmaps showing CUT&RUN read densities for IgG (NC) and BRD4 across BRD4 genomic peaks in SETD6 control and KO SKmel147 cells. (B) Heatmaps showing CUT&RUN read densities for IgG (NC) and Flag-BRD4 across Flag-BRD4 genomic peaks in SKmel147 cells expressing Flag-BRD4 WT or Flag-BRD4 K99R. (C) Venn diagram showing common genes for BRD4 in SETD6 CT cells and Flag in BRD4 WT cells as identified in the CUT&RUN analyses. (D) Representative binding profiles of four genomic regions.

Figure 4.

SETD6 interacts with MITF and regulates its genomic distribution. (A) ChEA for the 364 shared peaks presented in Fig. 3C . Blue points represent a significant TF. Smaller gray points represent nonsignificant terms. (B) Heatmaps showing CUT&RUN read densities for IgG (NC) and MITF across MITF genomic peaks in SETD6 control and KO SKmel147 cells. (C) Venn diagram showing common genes for BRD4, MITF in SETD6 CT cells and Flag-BRD4 in Flag-BRD4 WT cells as identified in the CUT&RUN analysis. (D) Representative binding profiles of three genomic regions.

Figure 5.

BRD4 interacts with MITF in a SETD6-dependant manner. (A) Left: Representative images and signal quantification (PLA dots per nucleus, AU) of PLA detecting BRD4 and Flag-MITF proximity in SKmel147 cells compared to negative control (NC) with no primary BRD4 antibody. Red dots represent PLA signal for MITF–BRD4 proximity. Scale bar = 10 µm. Right: Quantification of PLA signal per sample. Statistical analysis was performed using Student’s t-test (****P< .0001). (B) Left: Representative images of PLA detecting BRD4 and MITF proximity in SKmel147 SETD6 CT and KO cells. Red dots represent PLA signal for MITF–BRD4 proximity. Scale bar = 10 µm. Right: PLA signal quantification (PLA dots per nucleus, AU) for each sample. Statistical analysis was performed using Student’s t-test (****P< .0001). (C) Left: Representative images and signal quantification (PLA dots per nucleus, AU) of PLA detecting GFP-BRD4 and MITF in SKmel147 cells overexpressing GFP-BRD4 WT and GFP-BRD4 K99R. Red dots represent PLA signal for MITF–BRD4 proximity. Scale bar = 10 µm. Right: Quantification of PLA signal per sample. Statistical analysis was performed using Student’s t-test (****P< .0001). (D) Chromatin extract from SKmel147 cells overexpressing Flag-BRD4 WT or K99R were immunoprecipitated with anti-Flag antibody, followed by WB with the indicated antibodies. Input: levels of MITF, Flag, and histone-3 (loading control) in the total chromatin extracts.

Figure 6.

Acetylation of MITF increases its interaction with BRD4. (A) Acetylation assay in cells. SKmel147 cell lysates were subjected to IP with preconjugated pan-acetyl lysine A/G agarose beads. Proteins in the immunoprecipitate and input samples were detected by WB with the indicated antibodies. (B) Representative images and signal quantification (PLA dots per nucleus, AU) of PLA detecting BRD4 and MITF proximity in SKmel147 cells that were untreated or treated with SAHA for 4 h. Red dots represent a PLA signal for MITF–BRD4 proximity. Scale bar = 10 µm. Right: Quantification of PLA signal per sample. Statistical analysis was performed using Student’s t-test (****P< .0001).

Figure 7.

The BD of BRD4 binds to acetylated MITF. (A) SKmel147 cells were transfected with FLAG-MITF-WT and treated with 1 μM JQ1, where indicated. Chromatin fractions were then immunoprecipitated (IP) with a Flag antibody followed by WB with the indicated antibodies. Beads (B) served as negative control for the IP. (B) Representative images and signal quantification (PLA dots per nucleus, AU) of PLA to detect the proximity of Flag-BRD4 and MITF in SKmel147 cells overexpressing Flag-BRD4 WT, Flag-BRD4 N140A, or Flag-BRD4 N140F mutants. Red dots represent a PLA signal for MITF–BRD4 proximity. Scale bar = 10 µm. Right: Quantification of PLA signal per sample. Statistical analysis was performed using Student’s t-test (****P< .0001).

Figure 8.

BRD4 interacts with acetylated MITF by its BD in a SETD6-dependent manner. Schematic model illustrating the proposed SETD6–BRD4–MITF axis.