Did Neanderthals and Denisovans have our de novo genes?

Abstract

Gene duplication provides a profusion of raw material for evolutionary innovation (Lynch and Conery 2000). While most duplicates rapidly become unrecognizable, some, e.g., those that are immediately useful or those that after a period of relaxed selection gain unique roles, are retained and thereby expand a genomes protein-coding repertoire. Ohno (1970) famously remarked that without gene duplication, the creation of metazoans, vertebrates, and mammals from unicellular organisms would have been impossible. Such big leaps in evolution, he argued, required the creation of new gene loci with previously nonexistent functions (Taylor and Raes 2004). Recently, another source of new genes has been recognized: Though rare compared to gene duplication, it seems clear that new genes can emerge from formerly non-coding sequences, the ‘de novo’ protein-coding genes (Zhao et al. 2014, and references therein). In 2009, Knowles and McLysaght reported the discovery of three human de novo genes. These genes (CLUU1, C22orf45, and DNAH10OS) were transcribed and translated and, while the translations had no BLASTp hits in any other genome, orthologous non-coding sequences from five out-group primates were recognizable. Comparisons among these sequences exposed the human-specific ORF ‘enabler’ mutations. In a similar survey of primate genomes, transcriptomes and proteomes, Wu et al. (2011) compiled a list of 60 human-specific de novo genes. Database revisions confuse matters to some extent, with many genes (i.e., annotations) in Wu et al.’s list appearing and disappearing over time, but there seems to be little doubt that de novo origin plays a significant role in genome evolution. More recently, Xie et al. (2012) described 11 human-specific de novo genes in their investigation into the evolution of protein-coding genes from long non-coding RNAs. In all three studies, the authors suggested that these genes might be responsible for the evolution of humanspecific traits. High-quality Denisovan and Neanderthal genomes have recently been reported, and here, I use these resources to determine whether eight of the eleven humanspecific de novo genes reported by Xie et al. (2012), those encoding only one exon, were truly human specific. The Xie et al. list of human-specific de novo genes, including two of those reported by Knowles and McLysaght, was used for several reasons: Xie et al. showed that expression patterns differed between the ancestral noncoding RNA genes and the orthologous coding sequences, which is indicative of a genome-level ‘recognition’ of the enabler mutations. They also included a larger set of outgroup sequences helping to ensure that human de novo genes were not actually old protein-coding genes that had ’died’ in several out-group species. Finally, they compared dn/ds ratios to show that the out-group sequences were not experiencing codon-level evolutionary constraints. To this set of eight de novo genes I added CLLU1 from the Knowles and McLysaght study (Table 1), which was excluded from the final list of Xie et al. because the enabler mutation (‘DA’, see below), though absent from chimpanzee, gorilla, gibbon, and macaque, was observed in orangutan. The Denisova Cave in the Altai Mountains of southern Siberia was the source of the bones used for the Neanderthal and Denisovan genome projects (Meyer et al. 2012; Prufer et al. 2013; Reich et al. 2010). Both bones were estimated to be between 30,000 and 50,000 years old. For the Altai Neanderthal, bam files (which contain reads J. S. Taylor (&) Department of Biology, University of Victoria, PO Box 1700, Station CSC, Victoria, BC V8W 2Y2, Canada e-mail: taylorjs@uvic.ca