Identifying human-specific transcription factor binding site gains and losses that contribute to human-specific phenotypes

Abstract

Identifying the genetic changes responsible for human-specific traits remains a central challenge in evolutionary genomics. Although thousands of modern human-specific variants have been identified through comparisons with Neanderthal, Denisovan, and great ape genomes, the phenotypic consequences of most of these changes remain unknown, particularly for non-coding variants. Here, we predicted transcription factor binding site (TFBS) gains and losses for 16,883 human-specific high frequency variants. We integrated these predictions with experimentally annotated cis-regulatory elements (cCREs), target genes, and phenotype annotations to identify regulatory changes associated with human-specific traits. We identified 3,357 human-specific variants within cCREs that modify TFBS motifs and prioritized variants for which both the target gene and the affected transcription factor are independently associated with the same phenotype. This identified 128 candidate human-specific regulatory variants involving 155 genes and 58 skeletal traits known to differ between modern humans and archaic hominins. Beyond skeletal morphology, enrichment analyses suggested regulatory changes also affect soft tissues that are not represented in the fossil record, including the brain, vocal cords, testes, and other reproductive organs. Independent support for the functional relevance of these candidates comes from their significant enrichment in modern human-derived differentially methylated regions and among variants experimentally shown to alter gene expression in massively parallel reporter assays. Furthermore, candidate variants preferentially occur within unusually large regions depleted of archaic introgression, consistent with the hypothesis that some regulatory changes contributed to reduced fitness following admixture with Neanderthals and Denisovans. Together, our results provide a prioritized catalogue of human-specific regulatory variants and reveal candidate molecular mechanisms underlying traits that distinguish modern humans from archaic hominins.