MicroRNAs (miRNAs) are a unique class of short endogenous RNAs, which have become known in the past few decades as major players in gene regulation at the post-transcriptional level. Their regulatory roles make miRNAs crucial for normal development and physiology in several distinct groups of eukaryotes including plants and animals. The common notion is that miRNAs have evolved independently in those distinct lineages, but recent evidence from non-bilaterian metazoans, plants, and various algae raise the possibility that the last common ancestor of these lineages might already have employed an miRNA pathway for post-transcriptional regulation. In this Review we present the commonalities and differences of the miRNA pathways in various eukaryotes and discuss the contrasting scenarios of their possible evolutionary origin and their proposed link to organismal complexity and multicellularity.
MicroRNAs (miRNAs) are short (∼21–24 nucleotides (nt) long), endogenous, single-strand RNAs derived from hairpin transcripts that regulate gene expression in both animals and plants1,
The lack of sequence homology between miRNA families in plants and animals as well as differences in miRNA biogenesis and mode of action have led to the notion that miRNAs have evolved independently in both kingdoms from an ancient siRNA mechanism that already existed in the last common ancestor of all eukaryotes. However, recent studies relating plant and animal miRNAs unexpectedly raised the old questions: Did the common ancestor of all animals have miRNAs? Do miRNAs of plants and animals share a common origin? How many times did miRNAs evolve? This Review will attempt to re-address these questions.
miRNA sequence homology between distinct lineages
Is the lack of miRNA sequence homology between distinct lineages sufficient to determine convergence of the miRNA pathway? An early phylogenetic comparison of miRNAs revealed over 35 conserved families of miRNAs among bilaterian animals, and the pattern of conservation largely appeared to correlate with the accepted phylogeny. This led to the proposal that miRNAs are rarely lost once they have evolved14,15. Consequently, since no sequence similarity was detected between animal and plant miRNAs, they were considered to have evolved independently16. This line of thought extends also to several other lineages (Fig. 2). For instance, none of the eight miRNAs of the demosponge Amphimedon queenslandica is shared with other non-poriferan animals. Moreover, no conserved miRNAs could be detected between demosponge, calcareous and homoscleromorph sponges. This led to the extreme possibility that the miRNA pathway evolved convergently multiple times even within sponges13,17. Further, the traditional view that sponges are the sister group to all other animal lineages is under debate as several studies point to comb jellies (Ctenophora), which lack miRNAs, being the first animal lineage to diverge18,
Similarly, until recently, no shared miRNAs were observed between plants and green algae, suggesting a convergent origin of the miRNA pathway in these two main groups of Viridiplantae (Fig. 2)21. However, a new study annotating small RNAs of the liverwort Pellia endiviifolia revealed three miRNAs with high similarity to the green alga Chlamydomonas reinhardtii miRNAs22,23. This emphasizes the great importance of sampling many species from each phylum in order to fully perceive the extent of miRNA sequence conservation but also highlights the inherent difficulty of annotating miRNAs with high confidence24. Further, shallow sequencing depth as well as neglecting certain developmental stages and environmental conditions may mask the full miRNA repertoire of an organism and hence possible homology. Regardless of these caveats, the prevailing view is that the miRNA pathways of animals and plants evolved convergently14,16,21,25,
However, we suggest an alternative explanation: it might be possible that sequence turnover rates are high in plants and non-bilaterian animals, leaving no trace of shared miRNA sequences between contemporary lineages. In support of this scenario, plant miRNA genes are born and lost at high rates33,34, hence only a handful of expressed miRNAs are conserved among distant plant lineages35. Comparison of Arabidopsis thaliana and Arabidopsis lyrata small RNAs has shown that 33% of the miRNA families are not conserved between the two species, hence were gained or lost during the ∼10 million years (Myr) since they diverged33. Analysis based on genome data and small RNA sequencing from Capsella rubella, a very close relative of Arabidopsis, estimated that the net flux rate (birth – death) for miRNA genes in Arabidopsis is 1.2–3.3 genes per Myr (ref. 34). A high turnover rate of miRNAs is also suggested in green algae as only one miRNA is conserved between the two green algae Chlamydomonas and Volvox that separated about 200 million years ago (Ma)36,37. Within Bilateria, a recent study demonstrates that despite impressive examples of conservation, miRNA loss is much more common than previously appreciated38. Further, major losses of miRNAs have been reported within flatworms, a rapidly evolving lineage, perhaps as part of a morphological simplification trend39. Regardless, while the turnover of most bilaterian miRNAs might be higher than initially estimated, it is still lower than that of plants and estimated at 0.8–1.6 (or possibly even as low as 0.3) genes per Myr in Drosophila40,41.
Evolution of miRNAs in understudied eukaryotic groups
Because bilaterians have preserved many conserved miRNA families in comparison to plants, it is useful to consider outgroups to bilaterians. Among the non-bilaterians, placozoans and ctenophores do not possess miRNAs18, while in the sponge Amphimedon only eight miRNAs have been identified so far13. The phylum Cnidaria (corals, sea anemones, jellyfish and hydroids) represents the sister group of Bilateria, which branched off about 600 Ma (refs 19,20). A pioneering study of small RNAs in the sea anemone Nematostella vectensis indicated that this species possesses at least 40 miRNAs, yet only miR-100, is shared with Bilateria13. Remarkably, miR-100 is conserved over almost the full length of the miRNA, yet shifted by one nucleotide in the seed region, which was predicted to have significant consequences on the target specificity13. miR-100 function and its targets are not well conserved in Bilateria42 making its extreme sequence conservation in the sea anemone puzzling. However, lack of target conservation is true for most of the well-conserved bilaterian miRNAs, as they are frequently rewired into different genetic networks43. Interestingly, sequencing of small RNAs of the cnidarian Hydra magnipapillata revealed 126 miRNAs with no sequence similarity to those of Bilateria (hence no evidence of miR-100) and only two shared miRNAs with Nematostella44. These results suggest a rather fast turnover of miRNAs within Cnidaria and highlight the risk of misinterpreting lack of miRNA sequence homology as a lack of homology of the miRNA system.
In a recent study of Nematostella small RNAs obtained from several developmental stages, the miRNA complement was expanded to 87 miRNAs45. Interestingly, miR-9422 of Nematostella shares 16 of its positions, including the seed sequence with Arabidopsis miR-156a. This sequence similarity is unlikely to have occurred randomly as the chance of observing such a sequence identity between Nematostella and shuffled Arabidopsis miRNAs is very small (random sampling P-value < 0.01)45. As miR-156 is conserved from mosses to higher plants5,46,47, its similarity to miR-9422 could be an evidence for an miRNA conserved between animals and plants, thus supporting the hypothesis that miRNAs were inherited from the last common ancestor of the two groups. Taken together, in light of the potential high turnover rate, it is questionable whether lack of sequence homology can serve as a proof for multiple cases of convergent evolution of this pathway. Hence, additional features must be taken into account.
Another indication for evolutionary homology of animal and plant miRNA pathways comes from sequencing the genome and the small RNA repertoire of the symbiotic unicellular dinoflagellate Symbiodinium kawagutii, an outgroup to both plants and animals (Fig. 2), which revealed many potential shared miRNAs with plants and animals48. miRNAs have also been described in excavates, a group of unicellular eukaryotes that includes the parasites Giardia lamblia and Trichomonas vaginalis32,49. Several putative miRNA families from excavates have sequence homology to conserved animal and plant miRNAs. However, unlike most plant and animal miRNAs those peculiar excavate miRNAs are produced from small nucleolar RNAs (snoRNAs) and tRNAs and exhibit some additional irregular features. Furthermore, the possibility of horizontal gene transfer from a host to its symbiont or parasite could be an alternative explanation for the potential homology in these cases. Thus, further confirmation of these data from dinoflagellate and excavates is required. Yet, if those small RNAs are verified as bona fide miRNAs, this will be of great importance because it will strongly support the placement of the pathway's origin before the split of plants and animals.
Can miRNA biogenesis clarify the pathway's origin?
The discussion of whether plant and animal miRNA pathways evolved in a common ancestor or independently from an ancient RNAi mechanism must include the differences and similarities in miRNA biogenesis (Fig. 3). miRNAs are synthesized as primary hairpins (pri-miRNAs) and are processed to pre-miRNAs by cutting the hairpin stem, followed by cleaving the hairpin loop3,7. Next, they are loaded on to an Argonaute (AGO) protein and one RNA strand is selected for complementary target mRNA inhibition or cleavage3,7. In animals, the first step is conducted by a specialized microprocessor complex comprised of the RNAse III Drosha with the aid of the RNA binding protein Pasha (DGCR8 in vertebrates), while the second step of cleavage is performed by the RNAse III Dicer (Fig. 3)7. In plants, the Dicer homologue, DICER-LIKE 1 (DCL1), is responsible for both processing events required for miRNA maturation, conducting the two exact same steps in the same order2,28. In both plants and animals Dicer is crucial for processing the precursor miRNA (pre-miRNA) into mature miR/miR* dsRNA duplexes2,7,25,50. However, Dicers are common in many eukaryotes and also take part in non-miRNA activities such as RNAi via endogenous siRNA biogenesis and viral defence25,51 and therefore cannot be used as an argument for the common origin of miRNAs. Phylogenetic and structural analyses indicate that animal Drosha proteins are related to plant and animal Dicers, suggesting that in animals Drosha may have evolved following a duplication of the common ancestor of Dicer and Drosha and further specialised in the first processing step of miRNAs25,52. A very recent study found that miRNAs in Chlamydomonas are processed by DICER-LIKE3 (DCL3), a Dicer homologue that exhibits a domain organization similar to that of Drosha53. This finding hints at a possible evolutionary link between miRNA biogenesis in green algae and animals.
Both Drosha and Pasha homologues were found in Cnidaria13,54 and in two sponge lineages13,17, for which independent evolution of the miRNA pathway has been proposed. Thus, Drosha and Pasha were already present in the common ancestor of all sponges, Cnidaria and Bilateria and probably served miRNA-related functions, although their involvement in the miRNA pathway in the first two lineages remains to be shown. If Drosha and Pasha have an ancestral function in miRNA processing in these lineages, the only explanation for the lack of homology of miRNA sequences in distinct sponge lineages would be high turnover rates of miRNA genes (see above). Nevertheless, it should be noted that Drosha and Pasha are known to be involved also in non-miRNA related functions in mammalian cells that range from ribosomal RNA maturation to cleavage of specific mRNAs (reviewed in ref. 55). The tendency to associate the presence of the microprocessor components with the presence of miRNAs in basally branching lineages may thus be misleading.
The site of miRNA biogenesis seems to differ between plants and animals. In plants, both processing steps by DCL1 take place in the nucleus, while in animals the first step performed by Drosha occurs in the nucleus and the second cleavage by Dicer occurs in the cytoplasm3,16. However, several studies in animals reported the presence of Dicer in the nucleus (reviewed in ref. 55). Whether the nuclear localization of Dicer in animals is a relic of an ancient miRNA-processing pathway or a secondary adaptation is an open question.
In plants and animals Dicer requires protein partners in order to accurately cleave pre-miRNAs (Fig. 3)2,7. In plants, DCL1 is assisted by the RNA binding proteins SERRATE (SE) and HYPONASTIC LEAVES1 (HYL1), both crucial for miRNA biogenesis and development2,56,57. Although an SE homologue, Ars2, is known in animals as a partner of the microprocessor and Dicer58, no HYL1 homologues were found in bilaterian animals. These differences in the biogenesis were taken as additional evidence for an independent evolution of plant and animal miRNAs. However, homologues for both SE and HYL1 were recently reported in the sponge Amphimedon and in several cnidarians including Nematostella, all possessing dsRNA binding motifs54. This suggests that a HYL1-like protein was present in the last common ancestor of plants and animals and was lost in multiple lineages, including Bilateria54. The lack of any known animal Dicer partners such as Loquacious (Loqs), TRBP or PACT7,59,
The miR/miR* duplexes in both plants and animals are similar: they are ∼22 nt long with imperfect complementarity between the two strands and a 2-nt 3′ overhang2,3,7. However, unlike in bilaterian animals, the stem-loop precursor in plants is long and variable16. Interestingly, sponges and slime moulds present longer pre-miRNAs than their bilaterian counterparts13,30. Those longer and more variable miRNA precursors are reminiscent of siRNA precursors. Interestingly, the miR-2024 family in Nematostella has members that are bona fide miRNAs whereas other members of the family show processing variability that is typical for siRNAs (Fig. 1b,c)45. This suggests that transformation of an miRNA into an siRNA or vice versa can happen in animals like in plants62 and that the siRNA-like features of some miRNAs are not sufficient in order to rule out a common origin for plant and animal miRNAs.
Differences can also be found in the genetic structure of miRNA genes in plants and animals. In animals, roughly 50% of miRNA genes are located in clusters, often comprised of different mature miRNAs16,63. In plants, fewer cases of miRNA clusters are found, mostly encoding miRNAs of the same family with noticeable homology16,64. Nevertheless, a few clusters of non-homologous miRNAs were discovered in plants, predicted to target related proteins64. Interestingly, the sea anemone Nematostella has only two repetitive clusters, both of miR-2024, similarly to plants45. Hence, it seems that despite having different frequencies, the same genomic topologies can be found in both kingdoms.
Another genomic pattern separating animal miRNAs from those of plants is the location of miRNA genes. Approximately 30% of animal miRNA genes are located in introns16,65. In contrast, only three tested examples of intronic miRNAs are known so far in plants66,
In both plants and animals, binding of small RNAs to extensively complementary target RNAs induces degradation of the small RNA by adenosine or uracil addition (‘tailing’) and 3′-to-5′ exonucleolytic decay (‘trimming’)1,69. siRNAs in flies and piRNAs in germ cells of all animals are protected from these processes by 2′-O-methylation of their 3′ ends performed by the methyltransferase Hua enhancer 1 (HEN1)1,27. In plants, HEN1 is responsible for the methylation of both siRNAs and miRNAs1,2,70, while bilaterian miRNAs do not undergo this modification27. Interestingly, HEN1 was found to be expressed throughout the body in Nematostella54 and not only in germ cells, suggesting a non-restricted function. Moreover, periodate treatment proves that a major fraction of the Nematostella miRNAs is methylated, similar to plants13,45. One can speculate that the common ancestor of animals and plants possessed methylated miRNAs, a feature later lost in bilaterians possibly due to the loss of high complementarity between miRNAs and their targets (see next section).
miRNA mode of action varies between plants and animals
In both plants and animals siRNAs and miRNAs require a class of AGO proteins to execute gene regulatory functions71,
The degree of complementarity between the miRNA and its mRNA target has been considered a major factor in determining the mode of target repression. High complementarity, as seen in plants, promotes target cleavage by AGO, while seed-matching leads to translational inhibition and is the common mode of action of bilaterian miRNAs74,92,96. The boundaries between plant and animal miRNA mode of action are becoming more blurry as there is more translation inhibition in plants than previously appreciated83,84,97,98. However, even this type of inhibition requires nearly full complementarity as short matches limited to the seed region were shown to confer no target inhibition in plants83,84. Surprisingly, miRNA-mediated translational inhibition in plants also depends on a GW-repeat protein called SUO99. GW182 of animals and SUO of plants do not share detectable homology. However, it is tempting to consider the possibility of a common origin of those two proteins that cannot be detected anymore due to the extremely fast evolution of the GW182 family.
In Bilateria there are only a handful of examples for miRNA-mediated target cleavage due to full complementarity of the miRNA to the target100,101. For instance, in a genome-wide screen, only a single transcript was shown to be cleaved by a perfectly matching miRNA in C. elegans6. Further, the majority of AGO proteins in Bilateria have lost their endonucleolytic activity74,102,103. However, endonucleolytic activity of the vertebrate AGO2 protein was shown to be evolutionarily conserved due to its role in processing a single miRNA, miR-451, which cannot be detected by Dicer43,104,105. Thus, it is plausible that the conservation of this AGO activity in vertebrates is unrelated to target regulation. These are strong indications that target cleavage is far from being a major mechanism of action of bilaterian miRNAs. Recently, Nematostella miRNAs were reported to induce target cleavage by nearly perfect complementarity, more resembling siRNAs and plant miRNAs than bilaterian miRNAs45. Unlike in Bilateria, the target cleavage mechanism seems to be common in Cnidaria as it is also present in Hydra, a cnidarian that diverged from Nematostella at least 550 Ma (ref. 45). The fact that both Cnidaria and plants use target cleavage mechanisms combined with the existence of siRNAs in fungi, plants and animals suggest that cleavage was the ancestral form of small RNA action. The presence of a cleavage mechanism in sponges, the only other non-bilaterian lineage with miRNAs, would strongly support this view. Although experimental evidence is still missing, at least several of the eight miRNAs identified in Amphimedon queenslandica show high complementarity to putative target mRNAs. Additionally, the origin of the miRNA systems in the siRNA mechanism, which is based on slicing, by itself suggests that high complementarity is the ancestral state of the miRNA mode of action. This is also supported by the fact that seed-matching does not promote target inhibition in plants83,84. Those mechanistic differences indicate that AGO holds the guide strand very differently in plants and bilaterian animals leading to different binding characteristics106. It seems that seed-matching is a derived state that appeared in the last common ancestor of all bilaterians. It is likely that this mode of action allowed animals to expand their regulatory networks as one miRNA can potentially inhibit hundreds of targets (Fig. 4)4,32. Those complex networks involving miRNAs might be a reason for the lower turnover rates of miRNAs in bilaterians since the loss of a single miRNA might affect the expression level of numerous targets.
miRNAs and the evolution of complex organisms
In both animals and plants gene regulation by miRNAs allowed organisms to develop more complex gene regulatory networks4,107. Further, miRNAs were theorized by various authors to drive evolution to a multicellular state, making it possible for complex organisms to evolve3,12,
miRNAs are not mandatory for multicellularity as there are multicellular organisms that do not possess miRNAs such as fungi 123, placozoans13 and ctenophores18. In the case of ctenophores, it is possible that their extraordinary large collection of RNA-binding proteins substitutes the functions that miRNAs cover in other animals, allowing this phylum to produce complex cell types such as muscles and neurons20. Is the loss of complex, allegedly integral features feasible? For example, it was suggested that sponges and Placozoa might have lost their nervous system for an adaptive advantage72. A similar logic can explain the possible loss of miRNAs in organisms such as fungi, Placozoa or choanoflagellates. An example for advantageous loss of the RNAi components in fungi can be found in a study on the yeast Saccharomyces cerevisiae. The study demonstrated that several strains that had lost RNAi became more susceptible to killer virus infection when RNAi components were restored73. If losing the entire RNAi machinery is possible, we propose that losing only miRNAs must be a possibility as well. Despite of all this, the alternative scenario arguing that miRNAs evolved independently in plants, animals and other clades cannot be ruled out.
It has been suggested that the relative number of miRNAs is in correlation with the relative level of organism morphological complexity. Although it is generally difficult to define complexity, miRNA complements might have contributed to the complexity of gene regulation13,
Conclusion and future prospects
The rapidly expanding field of non-coding RNA research allows us to constantly re-evaluate the evolutionary origin of miRNAs. This review presents claims for and against a common origin of miRNA in plants and animals. Given the fast turnover in miRNA sequences and significant loss of miRNAs in both plants and animals33,34,38, we consider the lack of miRNA sequence homology known today as not sufficient to conclude on convergent evolution. Instead, the available information provides support for the possibility that the last common ancestor of plants and animals did possess an miRNA system (Fig. 5).
Of course, we cannot rule out convergent evolution. If indeed miRNAs have evolved independently at least nine times among eukaryotes (Fig. 2), it is intriguing to consider what is so unique about them that will support the evolutionary pressures required to allow convergence of such a magnitude. Ctenophores certainly demonstrate that an animal with a wide variety of cell types, including neurons and muscles can exist in the absence of this class of small RNAs18 and single-celled green algae demonstrate that miRNAs are not exclusive for ‘complex’ multicellular organisms29. Further, the lack of classic RNAi components in some fungi and the loss of Pasha (DGCR8) in Placozoa13,105 suggests that losing the protein machinery of the miRNA system is possible (Fig. 5)13.
Taken together, we can suggest two alternative hypotheses regarding the evolution of miRNAs: (1) the miRNA pathways and their mode of action via slicing originated convergently from siRNA mechanisms in the plant and several animal lineages and a system based on seed-matching of miRNA to its targets evolved later, after bilaterians diverged from the rest of animals; (2) the common ancestor of plants and animals already had an miRNA system acting via slicing that was recruited from the siRNA system. Still, in this scenario bilaterian animals also evolved a derived system based on seed-matching. In both scenarios the siRNA system evolved to protect cells from viruses and transposons and was adapted to post-transcriptional regulation of gene expression by targeting mRNAs.
Given the recent new findings, we favour the hypothesis of a common origin of miRNAs in plants and animals (Fig. 5). To further test this idea, biochemical and genetic methods are required to map the mode of action and the biogenesis pathway of miRNAs in groups beyond bilaterian animals and higher plants. We believe that some of the key points to study in the near future should be:
Test whether miRNAs in basally branching lineages such as sponges or algae cleave their targets.
Test whether the cnidarian HYL1 proteins, known to be involved in miRNA biogenesis in plants, are also involved in biogenesis of miRNAs in Cnidaria.
Assess the impact of the slicing versus the seed-matching mechanism on the gene regulatory networks in animals and plants.
Test whether homologous miRNA-related proteins in algae are indeed functional in the miRNA pathway.
Sequence small RNAs at an adequate depth from more eukaryotic lineages and annotate them as miRNAs according to widely accepted guidelines to reveal pattern of evolutionary turnover.
When those five points are met, we might get a better understanding of the evolution of the miRNA pathway in eukaryotes and answer the question of how ancient this system is.
How to cite this article: Moran, Y., Agron, M., Praher, D. & Technau, U. The evolutionary origin of plant and animal microRNAs. Nat. Ecol. Evol. 1, 0027 (2017).
Small RNA research in the Moran lab is supported by a European Research Council Starting Grant (CNIDARIAMICRORNA, 637456) and a Young Investigator Grant by the German–Israeli Foundation for Scientific Research and Development (I-1058-203.7-2013). Research in the Technau group is supported by grants of the Austrian Research Fund FWF (P24858 and P22618).