Figure 1.
Global identification of PPARG and RXR binding sites in human macrophages.
A) Table displaying the number identified PPARG peaks. PPARG/RXR peaks represent PPARG peaks that are supported by enrichment in the RXR ChIP-Seq library (RXR here represents RXRA, RXRB and RXRG). B) PPARG and RXR binding profiles across the locus for PDK4 in THP-1 cells. Plotted are the tag counts obtained from the respective ChIP-Seq libraries. C) Distribution of PPARG/RXR binding sites relative to annotated genes obtained from UCSC Genome Browser (built hg18/NCBI36; RefGene table). D) Motif identified de novo at PPARG/RXR binding sites using CisFinder.
Figure 2.
PPARG binding is poorly conserved between human and mouse macrophages.
A) Overlap of PPARG bindings sites between human and mouse macrophages. Comparison is based on murine binding sites lifted over to the human genome. 1548 out of 1961 PPARG binding sites in mouse aligned to the human genome. B) Tag counts from human PPARG library at different genomic loci in the human genome and mouse genome. Retained binding sites, human-specific sites and mouse-specific binding sites. Mouse and Human-specific sites in the human and mouse genome refer to the orthologous loci of mouse-specific or human-specific sites in the original genomes. For better visualization outliers were omitted from plot. C) Sequence conservation at human-specific and retained PPARG/RXR sites. Shown is the distribution of PhastCons scores for both categories. Significance was calculated using two-tailed t-test D) Pie chart summarizing the proportion of PPARG/RXR site that are retained and/or show sequence conservation (i.e. overlap with PhastCons element). E) Proportion of PPARG/RXR sites in human macrophages containing a PPARG motif compared to the proportion of sites with motif after liftOver to the mouse genome. The orthologous regions in the mouse genome are separated into PPARG bound and not bound. ‘Random’ shows the expected motif frequency for randomly distributed intervals with a matched size distribution. F) Distribution of PPARG/RXR binding sites in regard to TSS of RefGenes. Displayed are the distributions of human-specific and conserved PPARG/RXR sites.
Figure 3.
Identification of human-specific and shared PPARG/RXR target genes.
A) Grouping of PPARG/RXR targets genes in human macrophages based on PPARG binding in mouse. Displayed is the number of genes that are human-specific, indirectly, and directly shared PPARG/RXR target genes. Only PPARG binding sites in proximity to genes (<100 kb to TSS) were taken into consideration. B) and C) Enrichment of PPARG binding in homologous regions proximal to SLAMF9/Slamf9 and NR1H3/Nr1h3 in human and mouse macrophages (upper and lower panel, respectively). SLAMF9/Slamf9 represents an indirectly shared PPARG target gene while NR1H3/Nr1h3 represents a directly shared target gene. Browser tracks for mouse are shown in reversed direction to facilitate easier comparison between human and mouse.
Figure 4.
Conservation reveals functional PPARG/RXR target genes.
A) Association of PPARG/RXR binding sites with RSG regulated genes in THP-1 cells. Significance of enrichment over background was calculated using Fisher's exact test. B) Venn diagram representing the overlap between PPARG/RXR bound genes and RSG regulated genes across the different conservation categories. Indicated are the numbers of genes exclusive to the respective gene sets. C) Proportion of non-conserved, indirectly and directly shared target genes that are induced by RSG. Significance was calculated using Fisher's exact test. D) Bar plot showing the ratio of expected versus observed number of genes associated with the biological process category ‘lipid metabolic processes’ obtained from PANTHER for human-specific, indirectly and directly shared target genes.
Figure 5.
Composition of PPARG bound cis-regulatory modules is conserved between human and mouse macrophages.
A) Overlap between PPARG/RXR and PU.1 ChIP-seq peaks. Significance of overlap was calculated using proportion test. B) PPARG/RXR ChIP-Seq enrichment at PPARG/RXR sites without and with PU.1 overlap. C) Proportion of PPARG/RXR binding sites in human and mouse macrophages that are co-occupied by PU.1. D) Venn-diagram depicting the numbers of species-specific and retained PU.1 binding sites in human and mouse macrophages.
Figure 6.
Pu.1 potentially restricts binding site selection for PPARG during binding site turnover.
A) Scheme depicting a potential scenario for PU.1-associated PPARG binding site turnover. B) Average numbers of PU.1 binding sites in proximity to human-specific, indirectly shared, and directly shared PPARG target genes (<100 kb of TSS). Significance was calculated using two-tailed t-test. C) Proportion of conserved PU.1 binding sites at PPARG/RXR-PU.1 binding sites in human macrophages. Comparison was made between sites at human-specific and indirectly shared targets and significance was calculated using Fisher's exact test D) Human PPARG/RXR binding sites co-bound by PU.1 and adjacent to indirectly shared genes were split into sites containing conserved PU.1 binding sites and human-specific PU.1 binding sites, respectively. PPARG and PU.1 motifs were identified at orthologous loci in human and mouse. E) Shown is the locus for a PPARG/RXR binding site in human macrophages adjacent to ALOX5AP and its orthologous region in mouse. Binding for PU.1 and PPARG is shown at orthologous regions in human and mouse. Sequence alignments demonstrate conservation and loss/gain of binding motifs.