Formerly described as junk and parasitic, who would have expected that transposable elements would help us out from a potential evolutionary embarassment — the low gene number in the human genome? Ast and colleagues have shown that inclusion of Alu elements in exons promotes alternative splicing and therefore genome diversity. Using bioinformatic and experimental approaches, they now identify the mutational steps that create 5′-splice sites in alternatively spliced Alu elements.

The surprisingly low gene number and much higher protein number made alternative splicing an obvious process that could account for our biological complexity. Mouse and human sequence comparisons have indicated that alternative splicing is often associated with recent exon creation and/or loss. The authors previously showed that ∼5% of human alternatively spliced exons are derived from Alu elements — short primate-specific retrotransposons, of which humans have ∼1.4 million copies. These Alu exons (AExs) have evolved from intronic Alu elements. But how can a 'free' intronic Alu element turn into an exon that is alternatively spliced?

To answer that, the authors compared AEx sequences with those of their intronic ancestors. They focused on the 5′-splice site: >98% of human introns begin with GT and very few with GC, although these are the ones that are supposed to be mainly involved in alternative splicing. Strikingly, the most significant change in AExs was at position 2 of the intron, where a C→T transition creates a canonical GT 5′-splice site. Comparing over 300,000 sequences showed that positions 2 and 5 in the intron are those that matter.

But how is the alternative splicing of AExs regulated? Using site-directed mutagenesis, the authors found that alternative splicing of AEx is possible because C at position 2 of the 5′-splice site unpairs from the U1 small nuclear RNA (snRNA); its interaction with the 5′-splice site is crucial for constitutive splicing. Moreover, it seems that positions 3 and 4 of the intron control the level of exon inclusion, whereas G at position 5 ensures that the 5′-splice site is selected.

Both the in silico and ex vivo approaches point to the same conclusion — that the decay of CG dinucleotides in the human genome, as a result of hypermethylation, drives Alu exonization by creating new splice sites. Might this be an unexpected by-product of dampening down Alu transposition, given that mutating these CG dinucleotides renders the Alu retrotransposase inactive?

The authors provide more food for thought on genome evolution and organismal complexity by inviting us to consider the possibility that our genomes are littered with pre-exonic Alu elements, poised to be exonized. As such, they might serve as a reservoir for human-specific exons that might adapt to perform new functions, thereby promoting speciation of the human lineage.