The MuvB complex binds and stabilizes nucleosomes downstream of the transcription start site of cell-cycle dependent genes

Abstract

The chromatin architecture in promoters is thought to regulate gene expression, but it remains uncertain how most transcription factors (TFs) impact nucleosome position. The MuvB TF complex regulates cell-cycle dependent gene-expression and is critical for differentiation and proliferation during development and cancer. MuvB can both positively and negatively regulate expression, but the structure of MuvB and its biochemical function are poorly understood. Here we determine the overall architecture of MuvB assembly and the crystal structure of a subcomplex critical for MuvB function in gene repression. We find that the MuvB subunits LIN9 and LIN37 function as scaffolding proteins that arrange the other subunits LIN52, LIN54 and RBAP48 for TF, DNA, and histone binding, respectively. Biochemical and structural data demonstrate that MuvB binds nucleosomes through an interface that is distinct from LIN54-DNA consensus site recognition and that MuvB increases nucleosome occupancy in a reconstituted promoter. We find in arrested cells that MuvB primarily associates with a tightly positioned +1 nucleosome near the transcription start site (TSS) of MuvB-regulated genes. These results support a model that MuvB binds and stabilizes nucleosomes just downstream of the TSS on its target promoters to repress gene expression.

Fig. 1. LIN9 and LIN37 scaffold the MuvB complex.

a Domain architecture of human LIN9 and LIN37, with regions of predicted and validated structure shown as blocks. The conserved LIN37 CRAW domain and LIN9 DIRP domain structures are determined here. MBD is the MYB-binding domain. b Schematic model for subunit interactions within MuvB. c The indicated tagged subunits or MuvB complex (GST-LIN52, Strep-RBAP48, His-LIN54^504–749, GST-LIN37, and GST-LIN9^94–542) were expressed in Sf9 cells and extracts were precipitated with resin capturing the indicated tag. Proteins were visualized with coomassie staining. *Indicates impurities or degradation observed in some RBAP48 expressions. These bands are not pulled out from the tandem purification. The experiment was repeated three times with similar results. d HCT116 cells were transfected with plasmids encoding the indicated FLAG-tagged mouse protein. FLAG-tagged proteins were precipitated from extracts using anti-FLAG antibody and visualized with anti-FLAG immunoblotting and immunoblotting with antibodies that recognize RBAP48, LIN37, LIN54, and p130. A biological replicate of the experiment was performed, and results were similar.

Fig. 2. Structure of the MuvBN subcomplex.

a Overall structural model. b Alignment of LIN9 sequences from H. sapiens, C. japonica, D. rerio, D. melanogaster, S. purpuratus, and C. intestinalis. The (*) marks residues that contact RBAP48, the (.) marks residues that contact LIN37, and the (@) marks residues that contact both. c Close-up view of one interface between LIN9 and RBAP48. d Location of histone H3 and histone H4 peptide binding sites on RBAP48. When bound to LIN9, the H4 sites is blocked while the H3 site is mostly accessible. The model was generated from PDB IDs: 2YBA and 3CFV.

Fig. 3. LIN37 CRAW domain binds both LIN9 and RBAP48.

a Interactions of LIN9 and LIN37 with the RBAP48 insertion loop. b Alignment of LIN37 sequences from organisms as in Fig. 2b. Residues that contact LIN9 (.), RBAP48 (*), and both LIN9 and RBAP48 (@) are indicated. c Close-up view of the LIN9-LIN37 interface. d The indicated FLAG-GFP control and FLAG-LIN9 WT and mutants were expressed by transient transfection in arrested HCT116 cells. Proteins were immunoprecipitated using anti-FLAG beads and the indicated proteins visualized by western blot. The 3x LIN9 mutant is E125A/W126A/F127A and the 4x LIN9 mutant is R174A/R175A/F180A/F181A. e Same as (d) but expressing the indicated RBAP46 and RBAP48 WT and mutant proteins. These immunoprecipitation experiments were performed each with a biological replicate, and results were similar.

Fig. 4. MuvB binds to histone peptides and nucleosomes.

Fluorescence polarization (FP) measurements of association between recombinant MuvB and dye-labeled histone peptides (a) or nucleosomes (b). The data are shown as mean values from three replicates with the standard deviation (SD) as error bars. Data are fit assuming a single binding constant. c Tabulated affinities and error estimates from global fitting of all data across three replicates. Source FP data are provided as Source data.

Fig. 5. MuvB stabilizes nucleosomes on a reconstituted cell-cycle gene promoter.

a Reconstitution of the TTK promoter with nucleosomes and MuvB. At top is a schematic of the promoter fragment used in these experiments with the transcription start site (TSS) and CHR site indicated. The indicated proteins were refolded with a 461 bp fragment of DNA and the samples were analyzed by an agarose gel with ethidium bromide staining. The assay was replicated twice with similar results. b Schematic representation of the protocol for cross-linking and imaging (left) and example electron microscopy micrograph (middle) with corresponding traced molecules (right). c Histograms showing the fraction of DNA molecules containing the indicated number of nucleosome-sized bubbles for a sample of 100+ analyzed DNA molecules. The MuvB concentration used in the reaction was 1.3 μM, whereas low[MuvB] corresponds to 0.15 μM. The number at the top of the box indicates the average number of nucleosome-sized bubbles per DNA molecule with 95% confidence interval in brackets. These statistics were computed using bootstrapping with 10,000 iterations of resampling. Histograms were fit to a Poisson distribution and conditions were compared using an exact-Poisson test. p-values for comparisons of average number of bubbles (λ of Poisson distribution) across conditions are reported below the histogram.

Fig. 6. MuvB associates with nucleosomes in DREAM-regulated gene promoters in arrested HCT116 cells.

a Diagram of the MNase-ChIP experiment designed to enrich nucleosome-sized DNA fragments that interact with MuvB complexes. b Gene ontology (GO) analysis of the enriched DNA sequences following MNase digestion and precipitation from arrested cell extracts. Gene groups with p-values < 1 × 10⁻³⁰ are labeled. A complete list of GO terms is provided as Source data. c Top enriched genes with DNA sequences that co-precipitated with Strep-LIN9. Experiments were performed using WT or LIN37^−/− HCT116 cells, WT LIN9 or LIN9^4x, and in arrested or cycling cells. The number of genes with an enrichment >4.7-fold are indicated for each experiment. The list of enriched genes can be found on the NCBI GEO database (accession GSE189435). The annotated list of DREAM genes used as a cross-reference is provided as Source data. d Genome browser tracks corresponding to the CCNB2, FOXM1, and ORC6 promoters. The number of DNA sequence reads is plotted for the input (gray) and Strep-LIN9 precipitated DNA samples. These data correspond to one replicate performed in arrested HCT116 cells. Data for other replicates and experiments are shown in Supplementary Fig. 6. The transcription start site (TSS) in each gene (base of orange arrow) along with the position of the DREAM-binding DNA motif relative to the TSS are indicated.

Fig. 7. The sharp positioning of the MuvB-bound +1 nucleosome correlates with gene repression.

a Heatmaps of relative read density across all known DREAM genes that show >4.7-fold enrichment in the Strep-LIN9 precipitant and correspond. Both input and precipitant sequencing data are shown for the same gene set and in the same order. At the bottom are the aggregated and normalized read intensities across all genes shown in the heat map for the input and precipitated DNA data sets. Data from HCT116 (blue) and HCT116-LIN37^−/− (pink) are shown. The observation of more peaks with periodicity corresponding to nucleosomes (i.e., the +2 and +3 nucleosome peaks) in the HCT116-LIN37^−/− input data set is consistent with DREAM genes being active during Nutlin-3a-induced quiescence of those cells. b Comparison of three example gene tracks showing the number of DNA sequence reads in Strep-LIN9 precipitated samples from HCT116 (blue, left axis) and HCT116-LIN37^−/− (pink, right axis) cells. The TSS (orange arrow) and DREAM-binding DNA motifs are indicated. c Overall model showing organization of DREAM and the association between MuvBN and the +1 nucleosome, which we propose mediates gene repression.