Tuning Transcription Factor Availability through Acetylation-Mediated Genomic Redistribution

Abstract

It is widely assumed that decreasing transcription factor DNA-binding affinity reduces transcription initiation by diminishing occupancy of sequence-specific regulatory elements. However, in vivo transcription factors find their binding sites while confronted with a large excess of low-affinity degenerate motifs. Here, using the melanoma lineage survival oncogene MITF as a model, we show that low-affinity binding sites act as a competitive reservoir in vivo from which transcription factors are released by mitogen-activated protein kinase (MAPK)-stimulated acetylation to promote increased occupancy of their regulatory elements. Consequently, a low-DNA-binding-affinity acetylation-mimetic MITF mutation supports melanocyte development and drives tumorigenesis, whereas a high-affinity non-acetylatable mutant does not. The results reveal a paradoxical acetylation-mediated molecular clutch that tunes transcription factor availability via genome-wide redistribution and couples BRAF to tumorigenesis. Our results further suggest that p300/CREB-binding protein-mediated transcription factor acetylation may represent a common mechanism to control transcription factor availability.

Keywords: DNA-binding affinity; E-box; MITF; acetylation; bHLH-LZ; melanocyte; melanoma; transcription factor.

Graphical abstract

Figure 1

Genome-wide Binding by MITF (A) Genome browser screenshots derived from ChIP-seq using anti-HA antibody of 501mel cells stably expressing ectopic HA-tagged MITF. (B) Consensus motif for the most significant 900 genome-wide MITF-binding sites predicted from 60-bp regions around peak summits generated by MEME. (C) The proportion of peaks with or without a 5′-TCA(T/C)GTGN-3′ motif at different peak heights. (D) Relationship between motif frequency and peak height as in (C). (E) Sequences associated with a selection of differentiation or non-differentiation-associated MITF target genes. (F) Box and whisker plots of peak height related to motif. Center of notches indicates the median. Green box indicates range of peak heights within which lie a set of well-characterized differentiation-associated genes in addition to many other non-differentiation genes. See also Figure S1 and Table S1.

Figure 2

MITF Can Be Acetylated (A) Indicated expression vectors were transfected into Phoenix cells and input and anti-FLAG immunoprecipitates western blotted. (B) Western blot of 501mel cells treated with 200 nM TPA for indicated times. (C) Western blot of extracts from cells transfected with BRAF and/or p300 expression vectors. (D) Western blot of Phoenix cells transfected with indicated vectors and HA-MITF, ±20 μM U0126 immunoprecipitated using anti-HA antibody. (E) Schematic showing the melanocyte-specific MITF-M(+) isoform. The five acetylated lysine residues identified in MITF-M peptides by mass spectrometry are indicated below. ERK, p38, and RSK phosphorylation sites are indicated above with the CBP/p300-binding site. (F) MITF DNA-binding domain-DNA co-crystal structure showing the MITF K243-DNA phosphate-backbone contact. (G) Conservation of K243 between bHLH and bHLH-LZ family members. (H) Peptide array containing indicated residues as 14-amino-acid peptides immobilized on a cellulose membrane probed with rabbit anti-acetyl-K243 antibody. (I) Western blot using anti-acetyl K243 or anti-MITF antibodies of immunoprecipitated GFP-MITF expressed alone or with co-transfected CBP or p300. (J) Western blot using anti-acetyl K243 or anti-MITF antibodies of HIS-tagged MITF purified with nickel beads. All samples were from the same blot. See also Figure S2.

Figure 3

K243 Status Determines MITF DNA-Binding Affinity (A) Comparison of circular dichroism (CD) spectra of bacterially expressed and purified MITF WT and mutant DNA-binding domains. The mean residue ellipticity is plotted in dg × cm² × dmol⁻¹ against the wavelength (in nm). CD spectra show the mutations cause no major structural changes. (B) DNA-binding affinity of bacterially expressed and purified MITF WT and mutant DNA-binding domains determined using fluorescence anisotropy. Representative titration curves of each fluorescein-labeled oligonucleotide with MITF WT and mutants. The anisotropy values are the average of triplicate measurements from which the baseline corresponding to the anisotropy of the free fluorescent probe was subtracted. (C) The dissociation constants of MITF WT and mutants on oligonucleotides containing four different recognition sequences determined by fluorescence anisotropy.

Figure 4

K243 Controls MITF Function In Vivo (A) Complementation of neural crest MITFa-null nacre zebrafish using MITF WT and K238 (equivalent to K243 in human MITF) mutants (left) and quantification of numbers of melanocytes (right). The dots in the plots represent numbers of melanocytes in each rescued embryo with at least one melanocyte. See also Figure S3. (B) Western blot of 501mel cells stably expressing HA-MITF WT and mutants (from the same gel). (C) Tumor formation after subcutaneous inoculation of indicated cell lines into athymic nude mice. (D) Example tumors. (E) Tumor size over time using indicated cell lines. Error bars indicate S.E.M.

Figure 5

K243 Status Determines MITF Genome-wide Distribution (A) Heatmap of MITF WT and K243 mutant average tag density derived from two biological replicate ChIP-seq experiments of HA-tagged MITF expressed using 0 or 20 ng doxycycline centered on WT occupied regions (20 ng doxycycline). (B) Numbers of ChIP peaks called using HA-tagged MITF WT or mutants induced using 0 or 20 ng doxycycline. See also Table S2. (C) Read coverage of two replicates for each of the WT and K243 mutant ChIP-seq experiments expressed using 0 or 20 ng doxycycline centered around peak coordinates of the WT at 5-bp binning intervals. Numbers on the x axis indicate distance from center of the peak (in bp). (D) Genome browser screenshots of indicated loci showing HA-tagged WT and mutant MITF ChIP-seq profiles from iMITF cell lines expressing HA-tagged MITF at 0 or 20 ng doxycycline as indicated. (E) Box and whisker plots showing peak score for two replicate (R1 and R2) ChIP-seq experiments for the WT and two K243 mutants related to the indicated motifs. Expression of HA-MITF WT and mutants induced at 0 or 20 ng doxycycline. Colored line indicates median, and black line indicates mean. See also Figure S4.

Figure 6

Live-Cell Single-Molecule Tracking (SMT) of HALO-Tagged MITF (A) HALO-tagged MITF expression vectors. NLS indicates the nuclear localization sequence. Δbasic lacks residues required for DNA binding. (B) Exemplary frames of SMT movies using WT and mutant HALO-tagged MITF, collected at 100 fps (see also Videos S1, S2, S3, and S4). Scale bar, 5 μm. Labeling with 100 pM Halotag JF 594 allows particle densities in the range of a few molecules per frame. (C) SMT movies were tracked to generate a distribution of single-molecule displacements between consecutive frames that was fit with a three-component model (one immobile component and two diffusing components) to provide quantitative estimates for WT MITF and mutants shown in (D) and (E). Cmp, component. (D) Quantitative estimates derived from SMT using WT and mutant HALO-tagged MITF for the fraction of molecules in each state. Error bars indicate SD. (E) Quantitative estimates of the diffusion coefficients of free molecules. For MITF WT, $\text{Δ}$basic, K243Q, and K243R, respectively,$N_{cells}=20,6,15,15;N_{displacements}=17802,2684,16422,12999$. Error bars indicate SD. (F) Summary derived from the SMT analysis of proportion of MITF calculated to bind high- versus low-affinity sites. (G) Electrophoretic mobility shift assay (EMSA) using bacterially expressed and purified WT and mutant MITF DNA-binding domains, a 30-bp TCACGTGA-motif-containing probe, and competition with 4-fold dilutions of SSD (10 μg to 2.3 fg). Bound DNA is shown. Probe was in excess in all reactions. (H) EMSA as in (G) with competition by indicated competitor oligonucleotides at 3, 10, and 30 ng. Bound DNA is shown. See also Figures S5 and S6 and Videos S1, S2, S3, and S4.