Fig 1.
miRNA genes are transcribed from the genome, resulting in a primary miRNA transcript. Regions of the primary miRNA form a hairpin structure that is recognized by the endonuclease drosha, which cleaves the double-stranded stem region of the hairpin to create a pre-miRNA of ~83 nt in length. The pre-miRNA is exported to the cytoplasm where it is further processed by dicer, which cleaves off the loop region of the hairpin. This results in an approximately 22 to 23 nt double-stranded RNA called the miRNA-miRNA* duplex. The mature miRNA strand is loaded into the RNA-induced silencing complex (RISC), where its 8 nt seed region complementarily base pairs with messenger RNA targets, leading to their downregulation.
Fig 2.
Non-human primate miRNA is poorly characterized.
Only 12 of the ~300 known primate species have any entries in miRBase (release 21). The majority of these miRNA are predicted based only on homology (shown in blue); only four species (chimpanzee, gorilla, orangutan, and rhesus macaque) have sequences that are experimentally validated through RNAseq or other expression analysis (shown in purple). The number of characterized human pre-miRNAs (n = 1881) is still more than twice as large as that of chimpanzee (n = 655).
Fig 3.
Phylogeny of primate genome assemblies included in our study (adapted from [26]).
We selected species that had both a sequenced genome, and fibroblast cell culture available through Coriell Cell Repositories.
Fig 4.
MiRDeep2 scores range from -10 to 10, with a higher number corresponding to increased likelihood that the miRNA is genuine. A cut-off of 0 was used to be included in this study. MiRNA already annotated in miRBase are represented in black and gray: black represents miRNA with experimental validation, and gray represents miRNA previously predicted solely by homology to the human genome that have now been validated in this study. Novel miRNA are shown in a color corresponding to their miRDeep2 score; this score is partially determined by the availability of any previously annotated miRNA, which would inherently result in lower scores for our primates with no information in miRBase.
Fig 5.
Distribution of the number of non-human primate species from our dataset sequenced for a particular miRNA that was previously computational predicted by homology alone.
140/163 (86%) have experimental support from at least two primates. Because of the difficulty distinguishing between paralogs with identical mature sequences, only the paralog with the most coverage from a family of miRNA is shown in this chart.
Fig 6.
Location of variants within mature miRNA across the thirteen primate species sequenced in this study.
The 5’ end of the mature miRNA has an 8 nt “seed region” in positions 1 through 8 that complementary base-pairs with the 3’ untranslated region (UTR) of messenger RNA (mRNA). The 3’ end of the mature miRNA can also have an effect: positions 13–16 are highly conserved, and their proper complementary base pairing to a mRNA target is associated with downregulation [15]. As expected, the vast majority of the variants sequenced in our study appear in positions with relaxed evolutionary constraints (positions 9–12 and 17–23).
Table 1.
Summary of all variants found within the mature region of a miRNA ortholog group.
Fig 7.
Mean pairwise sequence identity compared to the z-score, where a more negative z-score indicates increased structural stability.
Scores below -3 (represented by the dotted black line) generally indicate very stable structures.
Fig 8.
Mean pairwise sequence identity compared to the Structure Conservation Index (SCI).
In general, an SCI near or above the mean pairwise identity indicates structural conservation (dotted gray line). The black line is the linear regression for our data (R2 = 0.1719).
Fig 9.
Alignment of miR-2355 homologous sequences.
The black line indicates species that have experimental support for the transcription of miR-2355, either in miRBase (human, cow) or from this study (chimpanzee, bonobo, gorilla, and orangutan). The red boxes outline the mature and star sequences within the miRNA. Variants only found among the great apes are highlighted in yellow, while all other variants are marked in grey. Humans have experienced a reversion at position 15 of the mature miRNA, restoring that nucleotide to its ancestral state.
Fig 10.
Predicted structure of miR-501 by miRDeep2.
Red indicates the mature miR-501-3p sequence supported by reads, yellow the predicted loop, and blue the predicted star sequence. In mouse lemur and galago, the mature sequence contains a variant (arrow) immediately following the seed region (underlined); this as well as variants outside of the mature sequence appear to alter the overall secondary structure of the hairpin, resulting in the mature sequence and thus the seed region being shifted downstream by 1 nt (circled in black).
Fig 11.
Phylogenetic tree of the predicted pre-miRNAs of the miR-320 family, based on experimentally determined mature sequences.
Paralogs miR-320b and miR-320c are only expressed in apes and Old World monkeys, lacking representation from New World monkeys and Strepsirrhines.
Fig 12.
Alignment of miR-320b1 homologous sequences.
The yellow box denotes the pre-miRNA sequence, red outlines the mature sequence, and variants with respect to humans are marked in grey. miR-320b1 is found in all apes, Old World monkeys, and New World monkeys (only human and marmoset are shown for simplicity). The entire pre-miRNA sequence is clearly absent in Strepsirrhines (aye-aye and galago), despite conservation of flanking sequence, demonstrating an insertion event that took place after the Strepsirrhini suborder split from the rest of the primate lineage.