Tissue-specific transcriptional regulation has diverged significantly between human and mouse

Abstract

We demonstrate that the binding sites for highly conserved transcription factors vary extensively between human and mouse. We mapped the binding of four tissue-specific transcription factors (FOXA2, HNF1A, HNF4A and HNF6) to 4,000 orthologous gene pairs in hepatocytes purified from human and mouse livers. Despite the conserved function of these factors, from 41% to 89% of their binding events seem to be species specific. When the same protein binds the promoters of orthologous genes, approximately two-thirds of the binding sites do not align.

Figure 1

Strategy to analyze transcription factor-DNA interactions in mouse and human. (A) Panel 1: approximately 8,000 high-confidence human (purple) and mouse (orange) gene orthologs were identified. Panel 2: 60-mer oligonucleotides were designed against a ten kilobase region centered around the complete set of transcription start sites in both species (colored boxes on genome track); orthologous genes with incomplete coverage, low oligonucleotide quality, or substantial gaps in one or both species were removed from the final design (Supplementary Methods). Panel 3: a human 5-array set and a mouse 4-array set capturing the transcription start sites for approximately 4,000 genes in each species were created using these oligonucleotides. (B) Mouse and human hepatocytes were isolated from liver samples and used in chromatin immunoprecipitations, which were hybridized against the array sets. (C) Gene-centric analysis classifies orthologous gene pairs by whether they are not bound in either species (hm), bound uniquely in human (Hm), bound in both species (HM), or bound uniquely in mouse (hM). (D) Peak-specific analysis classifies peaks relative to whether corresponding aligned regions exist in the second species and whether these aligned regions are bound. The four possible outcomes are shown in both the human-to-mouse and the mouse-to-human panels: In the first three cases (i, ii, iii) the aligned locus is present in the arrayed region of the ortholog. Case i (conserved): the aligned regions are bound in both species; Case ii (turnover): the orthologous gene is bound, but not at the aligned locus; Case iii (gain/loss): no binding is detected in the arrayed region of the second species, including the aligned sequence; Case iv (unaligned): the aligned sequence is not present within the arrayed region and therefore we cannot definitively classify the binding event, regardless of the presence or absence of a binding event in the other species.

Figure 2

(A) Number of genes bound by liver master regulators in each species, p-value (using a hypergeometric distribution) that the cross-species overlap is due to random chance, and the THEME-derived binding motifs in human and mouse. (B) The location of binding events varies between species. Here, ChIP enrichments are shown as traces. The 500 base pair sequence underlying the ChIP peak in each species is colored by species (purple human, orange mouse) and aligned with the corresponding sequence in the second species using dashed lines. For clarity, mouse ChIP enrichments are displayed as a negative y-axis, but orientation of the transcription start site is left to right. IGFPB1 is bound by HNF6 in both species, but the binding events do not align. The human sequence aligned to the mouse HNF6 peak in IGFBP1 contains large insertions overlapping a substantial portion of the human first intron (outlined with a dashed orange box), and is not bound by HNF6. (C) Shared binding events are frequently found in non-aligned regions. From left to right: aligned regions (shown as colored boxes) that are bound in both species (Figure 1D, case i); aligned regions present on both human and mouse arrays but bound only in one species (Figure 1D, case ii); regions bound in both species, but lacking aligned sequences on the orthologous array (Figure 1D, case iv, with a binding peak present). Typically, only about a third of the binding events detected in both species occur in sequences that align to each other (Table S3).