Histograms of ChIP fragments across the Pparg locus, normalized to fragments per 10 million aligned reads, for each of the profiled histone modifications and TFs at four time points during L1 adipogenesis. All histograms are shown on the same scale and high values were truncated as necessary. See also Figure S1-S2 and Tables S1-S2.

(A) Fractions of genes associated with at least one adipocyte (Ad), pre-adipocyte (Pre) or invariant H3K27ac region in L1s, conditional on changes in expression levels in adipocytes (max of days 2 and 7) relative to pre-adipocytes (max of days -2 and 0).
(B) Fractions of genes that showed ≥2-fold up- (left) or down-regulation (right) in L1s, conditional on the maximal enrichment score of associated H3K27ac regions.
(C) Fractions of genes that showed ≥2-fold up- (left) or down-regulation (right) in L1s, conditional on the number of associated H3K27ac regions.
(D) Fractions of genes associated with at least one H3K27ac region in hASCs, conditional on their changes in expression levels in adipocytes (max[day 3/9]) relative to pre-adipocytes (max[day -2/0]).
(E) Fractions of genes that showed ≥2-fold up- (left) or down-regulation (right) in hASCs, conditional on the maximal enrichment score of associated H3K27ac regions.
(F) Fractions of genes that showed ≥2-fold up- (left) or down-regulation (right) in hASCs, conditional on the number of associated H3K27ac regions.
(G) Fraction of genes associated with at least one adipocyte-specific H3K27ac region in L1s or hASCs, conditional on the ratio of their changes in expression levels during L1 and hASC adipogenesis.
See also Figure S3 and Tables S3.

(A) Summary of PPARγ binding sites in L1 and hASC adipocytes. For each quartile of ChIP enrichment scores, the columns show (from left to right) the percentage of sites located ≤ 2 kb from a known promoter; sites overlapping a PPARγ motif; sites overlapping H3K4me3/me2/me1 and/or H3K7ac in adipocytes and in pre-adipocytes; sites that could be mapped to an orthologous region in the other genome; and mapped sites that were also bound by PPARγ in the other model.
(B) Motifs learned ab initio from sequences +/- 100 bp from the top 800 PPARγ binding sites in L1s (ranked by enrichment scores). Virtually identical motifs were learned from hASCs.
(C) Correlations between PPARγ binding, the presence of a conserved motif instance, and open chromatin marks in human genomic regions orthologous to L1 PPARγ binding sites.
(D) Fractions of genes that were up-regulated ≥2-fold in L1s or hASCs, conditional on association with a PPARγ binding site. ‘Orthologous’ PPARγ binding sites could be mapped to an orthologous region also bound in the other model. ‘Dynamic’ PPARγ binding sites increased H3K27ac enrichment ≥5-fold.
(E) Fractions of genes that were up-regulated ≥2-fold in L1s or hASCs, conditional on association with ‘dynamic’ PPARγ binding sites.
(F) Annotation enrichment analysis of orthologs associated with PPARγ binding sites and up-regulated ≥2-fold. P-values are Benjamini-corrected.
(G) Summary of CTCF binding sites in L1 and hASC adipocytes.
(H) Motif learned ab initio from sequences +/- 100 bp from the top 800 CTCF binding sites in L1s (ranked by enrichment scores). A virtually identical motif was learned from hASCs.
(I) Correlations between CTCF binding, the presence of a conserved motif instance, and open chromatin marks in human genomic regions orthologous to L1 CTCF binding sites.
See also Figure S4 and Table S4.

(A) Chromatin state maps of the ~300 kb Cd36/Gnat3 locus from L1 pre-adipocytes (day -2) and adipocytes (day 7). Cd36 has three known promoters (P1-P3). Asterisk, start of the protein-coding sequence.
(B) Reporter assays. Each dot shows the ratio of normalized luciferase expression (RLU) from plasmids carrying distal fragments upstream of P2 or P3 over the estimated basal activity of the promoter. Fragments from six distinct distal sites (E1-E6) showed ≥2-fold mean enhancement of expression from P3 (orange dots, top). Three of these (orange dots, bottom) also showed ≥2-fold mean enhancement of expression from P2. E5 was present within two overlapping fragments. Error bars show standard errors of the means.
(C) 3C. Each dot shows the cross-linking frequency of a HindIII fragment to P3 (top) or P2 (bottom) in adipocytes relative to pre-adipocytes. Error bars show standard errors of the means.
See also Figures S5-S6 and Table S5.

(A) Genomic and chromatin state maps from L1 (top) and hASC (bottom) adipocytes. Orthology tracks show regions mapped from the mouse to the human genome (pink) and vice versa (red). Grey vertical lines highlight orthologous sites, those terminated by X highlight sites that could not be mapped. Orange dots show the E1-E6 enhancers identified in L1s.
(B) Expanded view of E5/E6 shows the locations of PPARγ motifs (blue bars) and transposons (grey/black) in the genomic sequences. The PPARγ motif underlying the peak of the L1 ChIP-Seq signal lies within a rodent-specific LINE/L1 fragment (arrow).
(C) Expanded view of an upstream region shows CTCF ChIP-Seq signals at non-orthologous sites separated by ~2.4-5 kb in L1s and hASCs.
(D) Alignments of the sequences underlying the two non-orthologous sites in (C) shows that the underlying motifs (blue bars) are not conserved.

TF motifs with the highest relative enrichment in adipocyte- (right) and pre-adipocyte-specific (left) H3K27ac regions. The top 400 L1 adipocyte and pre-adipocyte H3K27ac regions (ranked by enrichment scores) that could be mapped to orthologous locations with H3K27ac in hASCs were used. Each mammalian TRANSFAC (M prefix) and UniPROBE (U prefix) motif was matched and assigned adipocyte/pre-adipocyte enrichment ratios in the underlying mouse and human sequences (corrected for length and composition). The ‘ratio’ columns show the maximal (right) or minimal (left) enrichment ratio from mouse and human for non-redundant motifs with consistent enrichment ratios in the two genomes. The ‘candidates’ columns show genes or gene families expressed in L1 cells that are known to recognize each of the motifs. See also Table S6.

(A) L1 pre-adipocytes were transduced with retrovirus expressing PLZF or SRF [pMSCV empty vector (EV) as control] and induced to differentiate. The cells were subjected to oil Red O staining at the indicated time points.
(B) mRNA levels relative to 36B were assessed by RT-qPCR (mean±SD; n=4) at the indicated time points. ** P<0.01; *** P<0.001.
(C) Protein levels were assessed by Western blotting at the indicated time points.
(D) L1 pre-adipocytes were transduced with a retrovirus expressing control shRNA (shLuc), PLZF (shPLZF), or SRF (shSRF). The cells were subjected to oil Red O staining at the indicated time points.
(E) mRNA levels relative to 36B4 were assessed by RT-qPCR (mean±SD; n=4) at the indicated time points. *, P<0.05; **, P<0.01; ***, P<0.001.
(F) Protein levels were assessed by Western blotting at the indicated time points. See also Figure S7.