FOXM1 binds directly to non-consensus sequences in the human genome

Abstract

Background: The Forkhead (FKH) transcription factor FOXM1 is a key regulator of the cell cycle and is overexpressed in most types of cancer. FOXM1, similar to other FKH factors, binds to a canonical FKH motif in vitro. However, genome-wide mapping studies in different cell lines have shown a lack of enrichment of the FKH motif, suggesting an alternative mode of chromatin recruitment. We have investigated the role of direct versus indirect DNA binding in FOXM1 recruitment by performing ChIP-seq with wild-type and DNA binding deficient FOXM1.

Results: An in vitro fluorescence polarization assay identified point mutations in the DNA binding domain of FOXM1 that inhibit binding to a FKH consensus sequence. Cell lines expressing either wild-type or DNA binding deficient GFP-tagged FOXM1 were used for genome-wide mapping studies comparing the distribution of the DNA binding deficient protein to the wild-type. This shows that interaction of the FOXM1 DNA binding domain with target DNA is essential for recruitment. Moreover, analysis of the protein interactome of wild-type versus DNA binding deficient FOXM1 shows that the reduced recruitment is not due to inhibition of protein-protein interactions.

Conclusions: A functional DNA binding domain is essential for FOXM1 chromatin recruitment. Even in FOXM1 mutants with almost complete loss of binding, the protein-protein interactions and pattern of phosphorylation are largely unaffected. These results strongly support a model whereby FOXM1 is specifically recruited to chromatin through co-factor interactions by binding directly to non-canonical DNA sequences.

Fig. 1

Mutation of the FOXM1 DBD inhibits DNA binding. a Sequence alignments of the DBD for a number of Forkhead family members with the secondary structure shown schematically above. The residues used to generate point mutations are outlined in red. (*) conserved amino acids. H1-3 are α-helices, the orange arrows are β strands, and W1-2 are winged domains. b Plot showing relative change of polarization of a fluorescently-labeled (6FAM) dsDNA FKH consensus oligonucleotide upon addition of increasing concentrations of GST-FOXM1 WT or mutant DBD proteins. The FP assay provides a quantitative method and non-disruptive method to determine FOXM1 affinity for target by measuring the fluorescence polarization signals from the FAM-labeled FKH consensus (see Materials and methods). Data are plotted as % binding and show mean ± SD of triplicate experiments. (WT K_d = 1.10 ± 0.02 μM, H287A K_d = 3.04 ± 0.10 μM). c Plots showing relative luciferase activity of a 6X DB-TATA-luciferase reporter in cells transiently transfected with either WT or DBD mutant FOXM1 with the T596A mutation as a positive control. Data are shown as fold induction of luciferase activity following doxycycline induction. d Plot showing fold induction of a luciferase reporter containing a 200 bp sequence taken from the CCNB1 promoter following doxycycline induction of WT and mutant FOXM1 expression. Data represent triplicate experiments ± SD. (*) P <0.05, (**) P <0.01, (***) P <0.001

Fig. 2

Generation of inducible GFP-FOXM1 expressing HEK293 cell lines. a Western blot showing induction of GFP-FOXM1B expression following addition of doxycycline for 24 h while levels of endogenous FOXM1 are unchanged. The blot was probed using antibodies for FOXM1 and ACTB. b Quantitative PCR (qPCR) showing RNA expression levels of total FOXM1 (GFP-FOXM1 and endogenous) and endogenous FOXM1 only (FOXM1-UTR). c qPCR of FOXM1 target genes transcript levels in GFP-FOXM1 cells treated ± doxycycline for 24 h and expressed relative to the levels in the parent HEK293 cells (Negative) for each transcript. Data show triplicate experiments ± SEM. d Western blot showing protein levels of GFP-FOXM1 in WT and mutant cell lines treated ± doxycycline (Dox) at 1 μg/mL for 24 h

Fig. 3

GFP-tagged FOXM1 shows a similar genomic distribution to endogenous FOXM1. a Venn diagram showing the overlap between endogenous FOXM1 and GFP-FOXM1 binding sites in TetR HEK293 cells, using peaks in common in two replicates for each condition. b Motifs enriched in FOXM1 and GFP-FOXM1. Selected motifs identified in FOXM1 are also found in the GFP-FOXM1 peaks (full list in Additional file 1: Table S2). P values and percentages in brackets represent the statistical significance and the percentage of peaks for each motif, respectively. Motif sequences are sorted according to their proportional representation in FOXM1 peaks. c Examples of genomic regions showing three representative sites in common for endogenous FOXM1 and GFP-FOXM1 binding. d CEAS analysis comparing genomic distribution of endogenous FOXM1 binding events to GFP-FOXM1 in HEK293 cells

Fig. 4

Mutation of the DBD of FOXM1 inhibits binding genome-wide. a Hierarchical clustering analysis of the WT and DBD mutant GFP-FOXM1 replicate ChIP-seq datasets showing that the samples separate well into distinct treatment groups. b Venn diagram showing the overlap between binding regions identified by ChIP-seq analysis in 293 cells expressing GFP-FOXM1 WT and the GFP-FOXM1 mutants H287A or R286A. c CEAS analysis comparing genomic distribution of GFP-FOXM1 binding events in WT versus DBD mutants. d Examples of genomic regions showing three representative promoter binding sites in which binding peaks are only identified for GFP-FOXM1 WT and are not present in the DBD mutant cell lines. e Motifs enriched in WT GFP-FOXM1 and the mutant R286A-H287A (full list in Additional file 1: Table S4). P values and percentages in brackets represent the statistical significance and the percentage of peaks for each motif, respectively. Motif sequences are sorted according to their proportional representation in WT GFP-FOXM1 peaks (NS = non-significant)

Fig. 5

Binding of the GFP-FOXM1 DBD mutants is significantly reduced genome-wide. a Heat map comparing binding events in WT FOXM1 and both FOXM1 DBD mutants. The window represents ± 5 kb regions centered on WT GFP-FOXM1 binding events with a second plot (on right) showing the average signal intensity of differential bound peaks. b Three examples of genomic regions showing binding peaks where binding is reduced in the GFP-FOXM1 DBD mutants compared to the WT. (c and d) Differential binding analysis (DBA) was used to identify significantly (FDR <0.05) differentially bound peaks in the WT GFP-FOXM1 compared to H287A DBD mutant or the R286A DBD mutant GFP-FOXM1 cell lines. The red dots represent peaks where FOXM1 binding is significantly increased/decreased compared to the WT

Fig. 6

Proteomic analysis shows that the FOXM1 DBD mutants bind to same proteins as the WT. a Schematic diagram showing RIME analysis to identify FOXM1 co-binding proteins. b Coverage of GFP-FOXM1 and associated high-confidence interacting proteins. Yellow shading indicates regions of peptide coverage and the green shading shows post-translational modifications identified. (Prob = probability, # pep = number of peptides, %Cov = % protein coverage). c Table showing the average number of peptides identified for the six top proteins present in the WT GFP-FOXM1 pull-downs. d Co-immunoprecipitation showing pull-down of B-MYB, LIN9, and TF2B with a GFP antibody in extracts from HEK293 cells expressing WT, H287A, and R286A GFP-FOXM1 DBD mutants and GFP only. e Schematic diagram showing phosphorylation sites identified by Proteome viewer in a WT GFP-FOXM1 RIME sample. The color indicates the identification probability. f Diagram showing the position of previously identified phosphorylation sites in FOXM1b (red indicates serine and green threonine residues). The table indicates sites identified from the RIME analysis in WT and R286A DBD mutant GFP-FOXM1, with novel sites highlighted in green

Fig. 7

FOXM1 transcriptional activity requires direct chromatin interaction involving recruitment to non-consensus sequences. a qPCR analysis of the mRNA transcript levels in GFP-FOXM1 WT or mutant (H287A or R286A) cell lines treated ± doxycycline (1 μg/mL) for 24 h showing the relative change in the levels of FOXM1B, AURKB, CCNB1, CDC25B, CENPF, and PLK1. In each case the data are normalized to the minus doxycycline control. b Binding curves measured by fluorescence polarization analysis (assay details in the Materials and methods section), showing binding affinity of GST-FOXM1B DBD for 16-mer [FAM]dsDNA sequences present in FOXM1 binding peaks from the ChIP-seq dataset compared to the FKH consensus. The plot shows the fraction bound with increasing protein concentration. The table shows the K _d values ± SD determined for each sequence. c Illustration of alternative models proposed the recruitment of FOXM1 to chromatin. (1) Direct DNA binding of FOXM1 at promoter sites containing a FKH consensus motif and interaction with MuvB and B-Myb. (2) FOXM1 is recruited by MuvB complex and does not directly bind to the DNA. (3) FOXM1 binds directly at non-consensus sequences facilitated by interaction with MuvB and B-Myb. Arrow indicates transcription start site of target gene. Data representative of triplicate experiments ± SD. (*) P <0.05, (**) P <0.01, (***) P <0.001