Human evolution: the non-coding revolution

Abstract

What made us human? Gene expression changes clearly played a significant part in human evolution, but pinpointing the causal regulatory mutations is hard. Comparative genomics enabled the identification of human accelerated regions (HARs) and other human-specific genome sequences. The major challenge in the past decade has been to link diverged sequences to uniquely human biology. This review discusses approaches to this problem, progress made at the molecular level, and prospects for moving towards genetic causes for uniquely human biology.

Fig. 1.

Identification of human accelerated elements. Top: the four different approaches used to identify human accelerated regions. Some key differences include (i) the conserved elements used as candidates to identify HARs (which depend on multiple sequence alignments, methods to detect conservation, and whether human was masked in the alignments), (ii) bioinformatics filters that aim to restrict to non-coding elements and/or remove assembly or alignment artifacts, and (iii) tests used to detect acceleration. Bottom: overlap of the different datasets of human accelerated regions. Abbreviations: ANC accelerated conserved non-coding sequences [20]; HACNS human accelerated conserved non-coding sequences [23]; HTBE human terminal branch elements [21]. HARs include the original HARs [19] and the second generation HARs or 2xHARs [100]

Fig. 2.

Strategies to identify human-specific enhancers. a Methodology used to identify human-accelerated DNSase I hypersensitive sites (haDHSs) in [38]. Black bars, nucleotides that differ from the human sequence; blue bars, sites where all species differ from humans; dotted red lines, DHS and neutral sequence. b Five haDHSs showed different activity with human versus chimpanzee sequences in SK-N-MC cells. c Differentially active haDHS12/DAR12 (asterisk) overlaps previously identified HACNS219 [23] and is located near RNF145, a ring finger gene involved in cellular cholesterol metabolism [101]. d Brains of species studied in [43] (approximately to scale, colors label regions). Tree indicates approximate timing of splits between lineages (millions of years ago). e Workflow and (bottom) fraction of H3K27ac peaks that is differentially enriched (DE) between species per brain region. f Left: Percentage of HARs within conserved regulatory elements (CREs). Right: Most of the 240 HAR-containing CREs that align across species were not DE in human versus macaque and chimpanzee. g H3K27ac tracks for mouse, macaque, chimpanzee, and human cerebellum. Gray, shared enhancers; purple, DE enhancers higher in human versus macaque; red, HAR87. Enhancers gained in mouse or lost in primates (yellow) may compensate for the enhancer gained in primates (light blue) upstream of the CADM1 gene. Abbreviations: CB cerebellum, PFC prefrontal cortex, PcGm precentral gyrus, OP occipital pole, WM white matter, CN caudate nucleus, TN thalamic nucleus, Put putamen. Reproduced partially from [38, 43], with permission

Fig. 3.

Evolutionary mechanisms at the level of gene regulatory regions. a An example of accelerated sequence evolution affecting one enhancer in a locus and leading to gain of an expression domain in the developing forebrain. We characterized one such gain of function HAR in Kamm et al. [57]. b Shadow enhancers are multiple enhancers that direct a similar gene expression pattern and thus overlap in function. As such, they can act cooperatively to confer robustness in different physiological situations [77, 78]. The example shown is based on the work of Lam et al. [78], where two enhancers direct expression to the Arcuate Nucleus and both must be deleted in mice to produce a dramatic change in expression and phenotype. c Enhancer turnover is when one enhancer disappears and a new one appears in the same regulatory region, replacing the lost function. In the example shown by Domené et al. [72], turnover resulted in no net change in expression or phenotype

Fig. 4

Testing HARs and HAR sequence variants for enhancer activity with reporter assays. a Example of a transient transgenic reporter assay to test a HAR (HACNS426) for enhancer activity in mouse embryos. The experiment compares enhancer activity of the major allele (G) to that of the autism-associated minor allele (A) that is never homozygous in healthy controls and is predicted to change transcription factor binding. Top: constructs carrying each of the two HACNS426 alleles fused to the human CUX1 promoter and the GFP reporter gene are separately injected into single-cell mouse embryos. Bottom: to assay enhancer activity, brain slices from embryonic day E16.5 are stained for GFP. The authors observed differences in GFP expression with the G allele (above) versus A allele (below). The major strength of the approach is the spatial and cellular resolution of in vivo measurements, while weaknesses include not being highly quantitative, the use of mouse to compare human and chimpanzee variants, low throughput, and relatively high cost. Adapted with permission from [23]. The study also performed in vitro luciferase reporter assays and showed HANCS426 interacts with the dosage-sensitive CUX1 promoter. b Massively parallel reporter assays (MPRAs) enable thousands of reporter constructs to be tested as a library (top) in which each HAR variant is associated with a unique DNA barcode (such as 20-bp sequence). RNA sequencing of barcodes (bottom) provides a quantitative estimate of the activity of each HAR variant. MPRAs are high throughput, allowing thousands of HARs and variants thereof to be tested, and they are quantitative, enabling detection of single nucleotide differences with moderate effects on expression. Current weaknesses of the technology include being limited to HARs or HAR segments less than 200 bp and being restricted to testing in cell lines or mouse tail vein assays