Positive selection for the male functionality of a co-retroposed gene in the hominoids

Abstract

Background: New genes generated by retroposition are widespread in humans and other mammalian species. Usually, this process copies a single parental gene and inserts it into a distant genomic location. However, retroposition of two adjacent parental genes, i.e. co-retroposition, had not been reported until the hominoid chimeric gene, PIPSL, was identified recently. It was shown how two genes linked in tandem (phosphatidylinositol-4-phosphate 5-kinase, type I, alpha, PIP5K1A and proteasome 26S subunit, non-ATPase, 4, PSMD4) could be co-retroposed from a single RNA molecule to form this novel chimeric gene. However, understanding of the origination and biological function of PIPSL requires determination of the coding potential of this gene as well as the evolutionary forces acting on its hominoid copies.

Results: We tackled these problems by analyzing the evolutionary signature in both within-species variation and between species divergence in the sequence and structure of the gene. We revealed a significant evolutionary signature: the coding region has significantly lower sequence variation, especially insertions and deletions, suggesting that the human copy may encode a protein. Moreover, a survey across five different hominoid species revealed that all adaptive changes of PSMD4-derived regions occurred on branches leading to human and chimp rather than other hominoid lineages. Finally, computational analysis suggests testis-specific transcription of PIPSL is regulated by tissue-dependent methylation rather than some transcriptional leakage.

Conclusion: Therefore, this set of analyses showed that PIPSL is an extraordinary co-retroposed protein-coding gene that may participate in the male functions of humans and its close relatives.

Figure 1

Gene structure of PIPSL and its polymorphisms. The aqua arrows between top two bars mark the correspondence between two parental genes and PIPSL. "ATG" and "TAG" in black indicate the border of the ORF. It is almost the complete fusion product of PIP5K1A and PSMD4, although the original start codon (the boxed "ATG" in rose) was destroyed due to a human specific deletion [13]. The distance between this original start codon and the current assumed start codon is only 60 bps. The left-hand and right-hand pale blocks mark the sequenced promoter region and 3' UTR region, respectively. The gold arrows indicate from which region polymorphisms are, like 5' UTR, coding region and 3' UTR. "Segregating sites" show the ID of polymorphisms. "Reference position" indicates the position relative to the starting point of sequenced reads. The first base corresponds to 592 bp upstream relative to the transcription start site of PIPSL. "Reference sequence" marks the consensus sequence in those locations with "D" indicating deletions relative to the consensus, with the nucleotides deleted shown in individuals. Letters in uppercase indicate homozygous mutations, while letters in lowercase indicate heterozygous mutations. "031","032" and so on indicate ID of samples. Herein, 031~040, 041~049, 820~914 and 014~089 are samples from African American, Africans in the south of the Sahara, Russian and Chinese, respectively.

Figure 2

The evolutionary process of PSMD4-derived region (top panel) and PIP5K1A-derived region (bottom panel) inferred based on the free ratio model of CODEML. Blue and yellow bars marks ancestral branches leading to human and chimp in PSMD4-derived region and PIP5K1A-derived region, respectively. "P" indicates the parental gene. The number like "5.1/3.1" indicates how many nonsynonymous substitutions and synonymous substitutions occur in this branch, while the number in thicker font like "0.66" indicates Ka/Ks. In addition, we mark all branches with Ka/Ks significantly different with one by "a", which means a p of 0~0.05. Considering the small number of substitutions, we also mark those branches with a marginal significance (p of 0.05~0.1) by "b".