Human Lineage-Specific Transcriptional Regulation through GA-Binding Protein Transcription Factor Alpha (GABPa)

Abstract

A substantial fraction of phenotypic differences between closely related species are likely caused by differences in gene regulation. While this has already been postulated over 30 years ago, only few examples of evolutionary changes in gene regulation have been verified. Here, we identified and investigated binding sites of the transcription factor GA-binding protein alpha (GABPa) aiming to discover cis-regulatory adaptations on the human lineage. By performing chromatin immunoprecipitation-sequencing experiments in a human cell line, we found 11,619 putative GABPa binding sites. Through sequence comparisons of the human GABPa binding regions with orthologous sequences from 34 mammals, we identified substitutions that have resulted in 224 putative human-specific GABPa binding sites. To experimentally assess the transcriptional impact of those substitutions, we selected four promoters for promoter-reporter gene assays using human and African green monkey cells. We compared the activities of wild-type promoters to mutated forms, where we have introduced one or more substitutions to mimic the ancestral state devoid of the GABPa consensus binding sequence. Similarly, we introduced the human-specific substitutions into chimpanzee and macaque promoter backgrounds. Our results demonstrate that the identified substitutions are functional, both in human and nonhuman promoters. In addition, we performed GABPa knock-down experiments and found 1,215 genes as strong candidates for primary targets. Further analyses of our data sets link GABPa to cognitive disorders, diabetes, KRAB zinc finger (KRAB-ZNF), and human-specific genes. Thus, we propose that differences in GABPa binding sites played important roles in the evolution of human-specific phenotypes.

Keywords: ChIP-Seq; GABP; KRAB zinc finger genes; comparative genomics.; human molecular evolution; human-specific binding sites; promoter assay.

FIG. 1.

Overview of the experiments and analyses performed in this study. (A) Schematic representation of ChIP-Seq experiments with a GABPa-specific antibody in HEK293T cells to identify binding sites for GABPa. (B) Peak detection using the peak-calling algorithm available in QuEST. (C) Search for the GABPa consensus binding sequence and deduction of a PWM using the motif discovery tool Multiple Em for Motif Elicitation (MEME). (D) Search for additional binding sites within the peak region using the motif alignment and scan tool MAST. (E) Extraction of multiple species alignments of the MAST-identified binding sites from the UCSC MultiZ 44 vertebrate alignments. (F) Ancestral sequence reconstruction along with the phylogeny of 34 mammalian species from the UCSC 44-vertebrate alignments using the tools PAML and ANCESTORS. (G) GABPa motif matching within reconstructed ancestral binding site sequences using MAST. The dots in panels C and E represent sequence conservation.

FIG. 2.

Comparison of GABPa motifs from different studies and databases. Sequence logos represent the different PWMs.

FIG. 3.

GABPa peaks map close to gene starts and harbor GABPa binding sites residing closely to the peak centers. (A) Distance of the sites contributing to the MEME motif (6,031 of 6,208 in total) to the ChIP peak centers. (B) Number of GABPa motifs found per each ChIP-Seq peak. (C) Distribution of motif occurrences within 200 bp surrounding the ChIP peak centers. (D) Distance of peak calls to the nearest TSSs of UCSC genes within 10 kb, centered on the TSS. (E) ± 400 bp zoom of the peak calls to the nearest TSS. The x axis shows the distance to the nearest TSS in base pairs. Negative values represent upstream, positive values downstream regions.

FIG. 4.

Genomic view of GABPa ChIP-Seq reads spanning the TMBIM6 promoter including GABPa binding site predictions and multiple species alignment of the first exon. ChIP-Seq reads are colored in blue (forward reads) and red (reverse reads). The first exon (5′-UTR) is shown as blue bar with a black arrow indicating the TSS in HEK293T cells as determined by RNA-Seq. GABPa binding site predictions are shown as green and black boxes. Within the blowup in the lower part, including the UCSC multiple species alignment of exon 1, binding sites are shown as sequence logos of the GABPa PWM aligned to their matching positions. Within the alignment, dots indicate identity to the human reference sequence, whereas orange vertical bars indicate bases that are not depicted. Orange numbers below represent the sum of bases not depicted. The blue (C) illustrates the presence of a single cytosine in all nonhaplorhini at the indicated site.

FIG. 5.

Normalized firefly luciferase activities of human, chimpanzee, and rhesus macaque wt and mutated (mut) promoters. Bars represent average firefly to Renilla ratio in black for human (HSA), in gray for chimpanzee (PTR) and in white for macaque (MAC) promoters. For each species, the left bar refers to the wt and the right bar to the mutated promoter. (+binding site) or (−binding site) indicate presence or absence of GABPa binding sites in wt promoters and indicate introduction or disruption of sites in mutated promoters. For each promoter, measured activities were normalized to the construct with the lowest promoter activity level in HEK293T cells (set to one; supplementary table S16, Supplementary Material online). Standard errors were calculated from at least six replicates. (*) indicates significant differences between wt and mutated promoter activities according to a one-tailed Welch’s t-test, while (#) indicates significant difference of wt chimpanzee or macaque promoters compared with human wt activity according to a two-tailed Welch’s t-test (supplementary tables S17–S21, Supplementary Material online). The raw data for all constructs and the statistical significance are available in supplementary material (supplementary tables S17–S21, Supplementary Material online).

FIG. 6.

The introduction and disruption of GABPa consensus binding sites significantly influence reporter gene activities. For each gene, the number of predicted binding sites within 200 bp surrounding the peak centers is indicated. Species are denoted by HSA, Homo sapiens; PTR, Pan troglodytes (chimpanzee); and MAC, Macacca mulatta (macaque). Sequences are shown for wt and mutated sites. Underlined bases indicate differences from the human wt sequence. Mutated bases are colored in green or red indicating generation or disruption of a GABPa binding site, respectively. Green arrows depict higher activity of mutated over wt promoter, red arrows indicate lower activity, and yellow arrows represent no change. Differences in mutated and wt promoter activities are given as log2 ratios of average luciferase to Renilla ratios. Significance levels, as determined by Welch’s t-test for unequal variances, are indicated as (*) P value < 0.05, (**) P value < 0.01, (***) P value < 0.001, and (ns) not significant.