Searching for the genes that make humans different from other mammals has led David Haussler and colleagues to a novel RNA gene that is expressed in early neural embryogenesis. HAR1 (human accelerated region 1) has evolved much faster in the human lineage than in other vertebrates and is expressed in a region of the brain that develops in a unique way in humans. The human-specific mutations cause changes to the secondary structure of the encoded RNA molecule.

Most sequence differences between species are selectively neutral. To find the substitutions that set humans apart from chimpanzees, the authors searched for regions of the chimpanzee genome that are highly conserved in other amniote species but are diverged in humans. High conservation across other species implies that purifying selection is preventing genetic drift, and that, therefore, the human sequence has changed through adaptive evolution.

The authors identified 49 regions that showed human-specific accelerated evolution, 96% of which are in non-coding regions. The most accelerated region, HAR1, has undergone 18 substitutions since the human–chimpanzee common ancestor, compared with an expected 0.27 given the rate in other species. Resequencing studies indicated that the changes in the human lineage are probably more than 1 million years old, rather than the result of a recent selective sweep, and so might have been important in the emergence of modern humans.

HAR1 is part of a larger region that encodes two oppositely transcribed RNAs — HAR1F and HAR1R. HAR1F is expressed early in gestation in part of the neocortex that develops more extensively in humans than chimpanzees. It is also expressed in other parts of the brain later in gestation, and in the adult brain, ovaries and testes. HAR1R, on the other hand, is only expressed in the adult brain and testes, where it could be involved in antisense regulation of HAR1F.

The authors modelled the secondary structures of the RNAs encoded by HAR1F in humans and chimpanzees, and tested these predictions in vitro. The structures are unlike that of any known RNA, and there is an important difference between the two species: a particular double-stranded helix seems to be longer in the human HAR1F, and two adjacent helices might even be completely absent. Also, the human-specific mutations favour G and C over A and T, and therefore increase the stability of the extended helix.

These results offer a tentative explanation of how an adaptive change at the sequence level contributes to a human-specific neuronal phenotype, and are intriguing in that they implicate an RNA molecule. Much work needs to be done to confirm this explanation, flesh out the mechanistic detail and discover other sequence changes that are likely to underlie uniquely human properties.