Learning the cis sequence elements that determine AP-1 monomer specificity (ChIP-seq data sets)

Regulation of gene expression is mediated by combinations of DNA binding transcription factors that work in concert to recruit transcriptional machinery. Each cell type expresses hundreds of sequence-specific transcription factors, many of which recognize identical or similar DNA sequences. Such factors can play both redundant and non-redundant roles, but mechanisms determining overlapping or distinct biological outcomes are largely unknown. Here, we implement a machine learning approach to investigate how local combinations of sequence motifs influence the genome wide binding patterns of different members of the AP-1 transcription factor family in macrophages. Significant motifs associated with family member specific binding patterns were validated by assessing effects of motif mutations in different strains of mice. We further confirmed the prediction of PPARg to be preferentially associated with the specific binding pattern of cJun using PPARg knockout macrophages. Together, our results provide evidence that unique binding patterns of AP-1 family members result in part from the corresponding unique ensembles of nearby regulatory elements embedded within enhancers and promoters, and that these elements can be identified by machine learning models trained using genomic sequence. ChIP-seq targeting transcription factors from the AP-1 family (Atf3, cJun, Fos, Fra2, JunB, and JunD) and PPARg were performed in mouse thioglycollate elicited macrophage cells derived from two strains of mice (C57BL6/EIJ and BALB/cJ) in vehicle treatment and/or KLA treatment with corresponding input experiments and biological replicates. Additional ChIP-seq experiments targeting the transcription factors PU.1 and CEBPa were performed to validate previous studies using mouse thioglycollate elicited macrophage cells in vehicle treatment and KLA treatment. ChIP-seqs for Atf3, cJun, and JunD were also performed in immortalized bone marrow derived macrophage cells derived from C57BL6/EIJ mice (also with biological replicates).