Discussion
Other than whole genome duplication, the complexity of vertebrate genomes builds upon many unique sequence and functional features but one of them is genome expansion that compounds with the expansion of gene and intron sizes. There are three essential ways to increase genome sizes.18,19 The first is to increase the number of genes through genome and gene duplications. The second and also the foremost important mechanism is gene size expansion through intron size and number increases.20 The final way is the expansion of intergenic sequences and auxiliary chromosomal structures. With regard to the diversity of RSs and insertion/expansion mechanisms, we classified intron expansion into two categories: TE-driven and SS-driven,2,21 and speculated that they may play distinct roles in the intron size expansion of mammalian genomes. First, the profiles of TE insertions can be classified at levels of species and lineages, such as primates, large mammals, and rodents, and we did observe similar modes within lineages and distinctions among lineages. However, exceptions do exist as the rodents are not always cohesive—guinea pig behaves differently from mouse and rat concerning many RS counts. Second, we would like to emphasize the effect of RS expansion event rather than copy number counts, and we hope to see a clear and direct picture that correlates intron size variation with RS insertion.
In general, both TEs and SSs are reported to be non-randomly distributed among eukaryotic genomes.1,21–23 On one hand, there is strong negative selection to protect essential sequences in genomes for the transmission of basic genetic information in a relative shorter evolutionary time scale, such as protein-coding sequences or exons. On the other hand, RSs are indispensable as the prime power and raw materials for genomes to evolve for better fitness, to generate complexity and diversity, and to promote speciation and population dynamics.2,24 Therefore, RSs have strong influences on gene expression and regulation indirectly through variations in intron length and content.10,13 One mechanism shared by all the studied vertebrates is that both TE and SS insertions increase intron size but the strength of the former is much greater than that of the latter. In fact, after eliminating RS insertions in all introns, we observed that the tendency of length increase in the four intron classes remains the same. In other words, the large introns remain large in size even without RS insertions in all four intron classes and so do small introns. However, the introns of anole and chicken genomes are exceptional, where the intron size definitions may shift or not be clearly distinguishable between large and small when RS insertions are removed from the intron sequences (data not shown). We observed a non-random and unbalanced expansion mechanism of intron size evolution: larger introns tend to grow faster than smaller ones when introns are enlarged to a certain size or over a specific threshold. Furthermore, we investigated relationship and mechanism of TE- or SS-driven intron expansions. Satellites can increase intron size at an early or primitive stage as they change intron size in a relatively limited scale, but transposons are capable of increasing intron size in a larger (such as LINEs) and more massive (such as LTRs in multiple insertions) scale and thus have stronger influence on intron size expansion. Most importantly, we observed a synergy between TE-driven and SS-driven insertions, providing a greater degree of intron expansion
To understand the possible roles of RS families on gene and intron size expansions, we paid special attention on intron length and positioning within a transcript and on functional enrichment in the context of TE- vs. SS dichotomy among species and lineages. For instance, we found that TS-containing introns have a 5′-end bias in all vertebrates but zebrafish and that the RS-free (or the N class) introns have a 3′-end bias in all mammals but platypus. We have recently identified distinct functional profiles of genes at different evolving rates in primates, large mammals, and rodents,25 and in this study we used a similar classification scheme to investigate protein-coding genes with RS-driven intron expansion. For instance, DNA transposon-containing introns tend to be smaller in fraction, larger in size, and biased toward 5′-end enrichment in mouse and rat. We also pointed out that genes with TE-free introns are enriched in both development and transcription and genes with SS-containing introns are mostly immunity-related in primates and large mammals.13 We also extracted function categories in nervous systems for mammalian genes possessing SS-containing introns since microsatellite alternations may lead to neurological disorders.26 Previous studies proposed that microsatellites are unevenly positioned within different regions of protein-coding genes such as UTRs, exons, and introns, and they may play functional roles in regulating gene expression, splicing, mRNA export, and response to external environment.27 Most SSs that we studied are microsatellites, and we demonstrated that there are functional biases in SS-insertions, such as promoter-related regulatory genes as one of the major categories. In addition, SSs preferentially reside in heterochromatins at or near centromeres and telomeres, where transcriptional activities are rarely discovered. However, if detected, the genes are usually development-related and involved in epigenetic regulation and DNA methylation; the latter two lead to the alteration of chromatin state and may in turn regulate the expression of SS-containing noncoding RNAs.28,29 We concluded that combined or independent effects of species/lineage-specific TEs and SSs may play an important role in functional differentiations of intron-containing protein-coding genes. At present, the sequence-similarity-based RS library is mostly composed of known TEs, especially the collection of mammal-specific sequences. As increasing number of completed high-quality non-mammalian vertebrate genomes are being sequenced, together with the help of de novo identification technologies,30,31 there should be more novel species-specific TEs discovered, adding stronger validation power to the current study.
It is vital for us to track down the precise timing of intron evolution and expansion, such as in a context of lineages, especially the number of introns per gene and the length variation of introns.32 Spliceosomal introns are the great majority in vertebrate genomes, albeit opposing hypotheses on the origin of introns, “intron-early” and “intron-late”, which argue that introns of this particular type is either more ancient or late comers.33 Further analyses on genomes based on taxonomy suggested that intron loss is the dominant phenomenon with position- and phase-specificity in modern mammals and perhaps large amount of intron gains occurred at the early stage of animal evolution,34–36 and recent study has found several cases of intron gains happened in the ancestor of placental mammals in transposon-derived domestication-related genes.37 Moreover, gene length is correlated with gene expression levels and breaths and is affected by RS insertions, such as L1 and MIR.38 Housekeeping genes are often highly-expressed and harbor smaller introns to reduce the processing cost of transcription, including time and energy. In contrast, tissue-specific genes are often lowly-expressed and harbor larger introns, requiring more effective and complex regulatory elements.38,39 Our data, based on a RS-centric stratification approach, showed that intron expansion is strongly influenced by not only RS types but also insertion timing, and the latter is manifested as species-specific propagation of distinct RSs. A comparative study concerning the five teleost genomes indicated that zebrafish experienced an ancient large-scale RS-induced intron expansion, and RS profiles of such expansion is rather distinct from the other four fishes with relatively lower insertion frequency.40 Based on these observations, we suspect that the RS content diversity that we observed among vertebrate introns or genes may not be straightforward to characterize with regard to precise timing as the samples we used are still in a limited scope. Insertions of both TEs and SSs should avoid making damages to key regulatory sequences, such as the splice sites, the branch point, the polypyrimidine tract, and other uncharacterized functional elements, and have potential co-evolving patterns with neighbouring sequences;41 and in particular, TEs (eg, SINEs) facilitate the splicing of larger introns via the formation of secondary structure in mammals.42 TE- and SS-derived RSs are forced to cluster or locate in intronic regions and seldom occur in core regulatory regions that are constantly under strong positive or negative selections.