Identification of Smad2/3 binding regions in HepG2 cells. A, Smad2/3 binding to the SERPINE1/PAI-1 locus in HepG2 cells. MAT scores were plotted at the SERPINE1 and HPRT1 loci to obtain a graphical representation of Smad2/3 binding in these regions. Significant Smad2/3 binding regions as determined by detection of p values of 10⁻⁴ are shown by black bars. B, percent input values of Smad2/3 binding compared with input genome as determined by ChIP-qPCR. Cells were treated with 120 pm TGF-β for 1.5 h. Cells were cross-linked sequentially with dimethyl adipimidate and formaldehyde. Error bars represent S.D.

Comparison of the Smad2/3 binding regions among different cell lines. A, Venn diagrams showing the overlaps of Smad2/3 binding regions in HaCaT, HepG2, and Hep3B cells. Numbers in the circles indicate percentages of the Smad2/3 binding region of each cell line (red, HaCaT; black, HepG2; blue, Hep3B). B, identification of a motif conserved in HepG2-specific Smad binding regions. Partial genomic sequences within 250 bp from the peak positions of HepG2-specific Smad2/3 binding regions (n = 2,955) were analyzed using the CisGenome Gibbs motif sampler. Default parameters were used for the calculation, except for the numbers of motifs to be identified (n = 10). Matrix datum of the motif calculated by CisGenome was graphically shown using the SegLogo function of the R software. C, HNF4α-binding motif that matched the predicted motif in HepG2-specific Smad2/3 binding regions. The JASPAR CORE data base was used to identify known transcription factor binding motifs similar to the calculated matrix data in B. An HNF4α motif (ID: MA0114.1) was identified as the most similar motif with a comparison score of 21.3, which reached 96.9% of the potential maximal score. D, frequencies of the HNF4α-binding motif in Smad2/3 binding regions. Presence of the HNF4α-binding motif in each Smad2/3 binding region (within 250 bp from the peak signal position) was determined using CisGenome. Frequencies of the motif in either HepG2- or HaCaT-specific Smad2/3 binding regions were then calculated. As a control, matched genomic regions to HaCaT-specific Smad2/3 binding regions were obtained using CisGenome, and the frequency of the HNF4α motif was determined. E, expression of the HNF4α protein and phosphorylation of Smad2/3 in HaCaT and HepG2 cells. Cells were treated with TGF-β for 1.5 h, and the expression of each protein was determined by immunoblotting (IB).

Identification of HNF4α binding regions in the presence and absence of TGF-β stimulation. A, graphical representation of HNF4α binding to the APOA4/APOC3/APOA1 gene loci. Sequence read numbers of 100-bp sliding window were plotted for HNF4α ChIP-seq samples. Smad2/3 bindings as determined by ChIP-chip analysis were shown in the upper two panels as in Fig. 1A. Black bars represent significant binding regions (FDR, <0.1%). B, Venn diagrams showing overlap between TGF-β-treated and untreated HNF4α binding regions. HNF4α binding regions were determined for each sample (FDR, <0.1%). HNF4α binding regions that have overlapping regions within 500 bp from their positions of maximum read numbers were considered as shared binding regions. C, changes in the HNF4α binding to APOC3 and APOA1 loci were quantitatively determined by ChIP-qPCR analysis. Error bars, S.D. D, frequencies of in vivo HNF4α binding to the Smad2/3 binding regions. Percentages of HNF4α binding within 250 bp from the peak signal position of Smad2/3 binding regions were calculated for the indicated Smad2/3 binding groups. E, frequencies of canonical Smad-binding elements in HNF4α binding regions compared with Smad binding regions in HepG2 cells. A Smad-binding element, M00974.SMAD that was identified as an enriched motif in HepG2-specific Smad2/3 binding regions using CEAS (see text), was selected for calculation. CisGenome was used for mapping of the motif. Presence of the motif for each HNF4α binding region and Smad2/3 binding region was determined using PerlScript.

Effect of knockdown of HNF4α on TGF-β-induced gene expression in HepG2 cells. A, confirmation of HNF4α-knocked down samples for microarray analysis. HepG2 cells were transfected with HNF4α siRNA and treated with 120 pm TGF-β for the indicated times and harvested. Expression of HNF4α was determined by RT-qPCR and normalized by glyceraldehyde-3-phosphate dehydrogenase (GAPDH). siNC, negative control siRNA. B, down-regulation of HNF4α protein expression by siRNA. C, phosphorylation levels of Smad2/3 by using HNF4α siRNA. The top two panels show phosphorylation of Smad2 and Smad3. The 3rd panel indicates the expression of total Smad2/3, and the bottom panel is a loading control. IB, immunoblot. D, heat map of the TGF-β-induced expression of target genes of Smad2/3 and HNF4α and effect of HNF4α siRNA. Target genes that have overlapping binding regions of Smad2/3 and HNF4α were sorted by their induction of probe signal values by TGF-β stimulation for 1.5 h and are represented by color bars in the 1st column, using the TM4 microarray software (60). Relative expression of these genes in HNF4α siRNA samples to the control siRNA is shown in the 2nd column. In addition, a list of genes whose expressions changed more than 1.5-fold is shown in the right panel with their expression changes. E, heat map of target genes of TGF-β with Smad2/3 binding regions common to HNF4α at 24 h after TGF-β stimulation are shown as in D. Genes whose expressions were changed more than 2-fold are shown in the right panel.

Smad2/3 and HNF4α bindings in the MIXL1 locus. A, Smad2/3- and HNF4α-enriched regions in the MIXL1 locus are shown as in Fig. 3A. B, HepG2 cells were treated with 120 pm TGF-β for 1.5 h, fixed in formaldehyde, and harvested. Smad2/3 binding to the MIXL1 locus was verified by ChIP-qPCR. HPRT1 served as a negative control. C, HepG2 cells were treated with or without 120 pm TGF-β for 1.5 h, and ChIP-qPCR analysis of the MIXL1 locus using anti-HNF4α was performed as in B. n.s., not significant. D, effects of knockdown of HNF4α on TGF-β-induced expression changes of MIXL1. HepG2 cells were transfected with control siRNA (siNC) or siHNF4α, treated with 3 ng/ml TGF-β for the indicated times, and harvested. HNF4α expression was quantified by RT-qPCR. *, p < 0.05; error bars, S.D.

Identification of regulatory elements important for TGF-β-induced transactivation of the MIXL1 promoter. A, schematic representation of HNF4α-binding motifs in the Smad2/3 and HNF4α binding region of the MIXL1 promoter. Promoter reporters with mutations in their HNF4α-binding motifs are shown in the lower panel. B, activation of the MIXL1 gene promoter by TGF-β and effects of mutations in putative HNF4α-binding motifs. HepG2 cells were transfected with the MIXL1 promoter and its mutants and treated with TGF-β for 24 h. C, conserved Smad-binding elements (SBEs) of the MIXL1 promoter. Four Smad-binding elements that were conserved between mouse and human (SBE 1–4) are shown with their relative positions from the transcription start site. Nucleotide sequences of Smad-binding elements and their mutations used in D are also shown. WT, wild-type; mut, mutant. D, effect of mutations in Smad-binding elements on TGF-β-induced transcriptional activity of MIXL1 promoter. Cells were treated as in B, and luciferase activities were determined. *, p < 0.05 compared with WT without TGF-β; **, p < 0.05 compared with SBE4 mutant, without TGF-β; n.s., not significant compared with WT and SBE2 mutant, without TGF-β; error bars, S.D.

Roles of HNF4α on Smad2/3 binding and transcriptional activity of MIXL1 promoter. A, effects of HNF4α knockdown on transactivation of the MIXL1 promoter. HepG2 cells were transfected with siRNAs 1 day before transfection with the reporter constructs. siNC, negative control siRNA. B, effects of exogenous HNF4α on transactivation of the MIXL1 promoter. HNF4α was exogenously expressed in HaCaT cells, and transcriptional activity of MIXL1 reporter was determined. The lower panel shows the protein expression of HNF4α. IB, immunoblot. C, HNF4α (variant 2, RefSeq ID: NM_000457) or its C115R (CR) mutant, which does not bind to DNA, was overexpressed in HepG2 cells. The lower panel shows the protein expression of HNF4α and its mutant. D, effect of HNF4α siRNA on Smad2/3 binding to the MIXL1 locus. HepG2 cells were transfected with siRNAs 24 h before TGF-β stimulation. Cells were fixed 1.5 h after treatment, and ChIP-qPCR was performed as in Fig. 1B. Error bars, S.D.; *, p < 0.05; n.s., not significant.