Intron size increase often involves lineage-specific changes in RS contents in the context of genes
To investigate the relationship between intron size and repeat insertion in a comparable fashion, we divided introns into ten size intervals for the convenience of in-depth analysis since in general introns tend to cluster at certain size ranges (Fig. 1). According to the relationships among shape-variable curves from the three repeat types, retrotransposons, DNA transposons, and satellites, we found that RSs of the twelve mammals fell into two basic patterns. The first pattern is SS-rich, including three rodent species and two primitive mammals, and its repeat abundance ranks as retrotransposon > satellite > DNA transposon. The second pattern, including the rest of the seven mammals, has a repeat content order of retrotransposon > DNA transposon ≥ satellite (the subequal sign is true only for macaque). In addition, we observed an up-convex curvature of retrotransposon distribution and an up-concave curvature of DNA transposon and satellite distributions with the exception that the curves of satellite distribution in mouse and rat are near-linear, indicating that SSs play a relatively dominant role in their intron size expansion. As to the difference between the non-mammal vertebrates and the mammals, we found that DNA transposons have higher abundance but decreasing slope with intron size increase than the other two patterns in both zebrafish and frog. However, this phenomenon disappears and changes into lower abundance and an increasing slope with intron size increase in anole and chicken. The abundance of retrotransposons is lower than those of satellites in zebrafish, frog, and anole, and the abundance of retrotransposons is higher than that of satellites but the mode of slope remains the same in chicken and the mode of slope changes into descending in all twelve mammals.
We subsequently tried to find the major TE families that influence intron size in each vertebrate species or lineages by calculating the fraction of introns possessing a particular RS class (Table 1). First, SINEs are supreme in overall abundance among all TEs in mammals. In the primates, Alu and MIR are most abundant. In the two small rodents, mouse and rat, B1, B4, and B2 are most abundant, whereas in guinea pig, the larger rodent of the group, B1 and B4 are most abundant. Second, for the four most abundant TE families in each species, the four large mammals, cow, panda, horse, and elephant, share MIR, L1, and L2, as well as other species-specific TEs that include BovA for cow, tRNA-Lys for panda, and SINE:SINEs that are specific for horse and elephant. MIR is abundant in all twelve mammals; opossum and platypus rank as the top two but the three rodents appear behind all the rest mammals. Third, the three lower vertebrates, chicken, anole, and frog, have CR1, Sauria, and Harbinger as the most abundant TEs, respectively. Zebrafish appears to have the most diverse DNA transposons and they are all quite abundant: DNA:DNA, hAT, hAT-Charlie, TcMar-Tc1, En-Spm, hAT-Ac, and Harbinger. Fourth, concerning satellite sequence classes, we found that all SSs are prevalent in the sixteen vertebrates but mouse, rat, zebrafish, and opossum are more SS-rich among all.
We further identified abundant TE families in each species and have several significant observations (Fig. 2). First, there are near-linear distributions of MIR in introns with a length range of 150 bp–10,000 bp and rapid accumulations of introns over 10,000 bp in the primate and large mammal lineages. In contrast, there is a drastic slowing-down in the rodents, particularly mouse and rat. Aside from this, slowing gains of MIR are also seen in the two primitive mammals. Second, the trends of L1 and L2 insertions over intron sizes are also interesting; the two curves intersect in the large mammals and primates but do not in opossum, where we observe L1 < L2 before and L1 > L2 after the intersections. Third, the distribution of primate-specific Alu repeats has an up-convex curvature, an indication of early saturation and preferred insertions in relatively small introns as compared to LINEs and other SINEs. The rodent-specific B1, in contrast, has a near-linear distribution and is more prevalent than B2 and B4. SINE:ID, unique to mouse and rat, seems more active in rat than in mouse. Fourth, distinctly different from what in other mammals, L2 in platypus behaves similarly to its MIR.