Age-dependent gain of alternative splice forms and biased duplication explain the relation between splicing and duplication

Abstract

We analyze here the relation between alternative splicing and gene duplication in light of recent genomic data, with a focus on the human genome. We show that the previously reported negative correlation between level of alternative splicing and family size no longer holds true. We clarify this pattern and show that it is sufficiently explained by two factors. First, genes progressively gain new splice variants with time. The gain is consistent with a selectively relaxed regime, until purifying selection slows it down as aging genes accumulate a large number of variants. Second, we show that duplication does not lead to a loss of splice forms, but rather that genes with low levels of alternative splicing tend to duplicate more frequently. This leads us to reconsider the role of alternative splicing in duplicate retention.

Evidence for widespread subfunctionalization of splice forms in vertebrate genomes

“…Likewise, they argue that genes that are expressed at low levels demonstrate low levels of alternative splicing and high rates of duplication. These arguments provide a mechanism for the preduplication bias first proposed by Roux and Robinson-Rechavi (2011). That is to say, the apparent reduction in alternative splicing found in large gene families is not due to gene duplication, rather it is a consequence of short genes and lowly expressed genes being more likely to duplicate while being less likely to undergo alternative splicing.…”

“…However, there are several potentially confounding factors. First, it is known that gene families with many paralogs undergo very low rates of alternative splicing (Kopelman et al 2005;Su et al 2006;Jin et al 2008;Chen et al 2011;Roux and Robinson-Rechavi 2011). Therefore, if the exon divergent paralogs are enriched in genes that belong to these large gene families, the importance of exon structure divergence would necessarily be called into question as the association between exon structure divergence and lower rates of alternative splicing may be spurious, due to the influence of gene family size.…”

“…Su et al (2006) demonstrate that in humans there is a reduction in alternative splicing shortly after gene duplication. This reduction appears to be transient, as increasing time since duplication correlates with increasing levels of alternative splicing (Su et al 2006;Roux and Robinson-Rechavi 2011). This means that if exon divergent paralogs are enriched in young duplicates, the influence of duplication age may make it appear as though exon structure divergence between paralogs results in lower levels of alternative splicing.…”

“…In fact, it would seem as though gene families of intermediate size undergo the most alternative splicing. This is difficult to reconcile with the hypothesis that gene duplication can substitute for alternative splicing, but suggests that less stringent purifying selection acting on newly formed duplicates may allow for the acquisition of novel splice forms (Roux and Robinson-Rechavi 2011). They also argue that the apparent reduction in alternative splicing found in the largest gene families is not due to gene duplication reducing levels of alternative splicing but is the product of a preduplication bias predisposing genes with initially low levels of alternative splicing to duplicate more frequently (Studer and Robinson-Rechavi 2009;Roux and Robinson-Rechavi 2011).…”

“…This negative correlation was presumed to be the product of subfunctionalization reducing alternative splicing levels in duplicate genes (Su et al 2006). Others, however, have questioned whether the inverse relationship between alternative splicing and gene duplication can be explained by subfunctionalization or whether the negative correlation between the two is even valid (Talavera et al 2007;Jin et al 2008;Chen et al 2011;Roux and Robinson-Rechavi 2011;Grishkevich and Yanai 2014). Of note, Roux and Robinson-Rechavi (2011), using a more complete data set than either Kopelman et al (2005) or Su et al (2006), find that the largest gene families display the lowest levels of alternative splicing; however, singletons do not have the highest levels of alternative splicing.…”