A line of evidence shows that single nucleotide polymorphisms (SNPs) are important functional sequence variations, and SNPs that occur at miRNA loci may have profound effects during the evolution of a species. To date, a survey of the concerted evolution of miRNAs and their binding target sites has been investigated in Arabidopsis (Ehrenreich and Purugganan 2008), and rice (Wang et al. 2010), respectively. In these analyses, a significant lower level of polymorphism in miRNAs was found, indicative of purifying selection in these pre-miRNAs (Ehrenreich and Purugganan 2008; Wang et al. 2010); while the reverse phenomenon was observed in this study when all rice miRNA genes were taken into consideration, where pre-miRNAs accumulated more SNPs than the flanking regions (Figure 2a). The major reason underlying the observed inconsistence may be due to the differences in evolutionary conservation of the selected miRNA samples in different studies, because when only the highly conserved rice pre-miRNAs were adopted, the SNP densities in pre-miRNAs and MmiRNAs decreased significantly (Figure 2b). As known, the miRNA repertoire of each species basically consists of a set of conserved miRNAs as well as many lineage- or species-specific miRNAs (Rajagopalan et al. 2006). During evolution, significantly stronger selective constraints should impose on the deeply conserved miRNAs, whereas most species-specific miRNAs (young miRNAs) are expressed weakly, or in specific conditions or tissue types, more divergent, and tend to lack targets, suggesting a neutrally evolving path. For example, we analyzed further the 55 pre-miRNAs with 10 or more SNPs, particularly the miR2124 family members (osa-miR2124a-f, h-i), and found that these miRNAs are almost rice species-specific. However, using a very stringent filter approach, we found that 44 out of the 177 pre-miRNAs having no SNPs including osa-miR156e-j, osa-miR160a, osa-miR166b, f, osa-miR172a, c, d, osa-miR395j, l, and etc. are highly conserved in plants. These results support the findings in human that species-specific miRNAs that are always lowly expressed are under weaker selective pressures and evolve rapidly (Liang and Li 2009).
In rice, the length of mature miRNAs (MmiRNAs) ranges from 19 to 24 nt, where near half (49.2%) are 21 nt in length, and followed by 22 nt (22.2%) and 24 nt (20.3%). However, 28.7%, 36.5% and 31.9% SNPs were found to be located in the MmiRNAs with 21, 22, and 24 nt in length, respectively. This observation suggests that weaker constraints should work on the 22- and 24-nt-MmiRNAs, leading to the generation of new miRNAs or alteration their functions. In plants, most mature miRNAs start with uridine (U; site 1); sites 10, 11, and 19 are crucial for target recognition and cleavage (Palatnik et al. 2007; Voinnet 2009), which explains the low sequence diversity at these sites (Figure 3). In addition, the nucleotide site 8 is also rather conserved during evolution, suggestive of its potential importance in miRNA-target interactions and/or functions.
Several studies revealed that SNPs in miRNA genes could affect, either increase or disrupt the processing of miRNAs. However, most of these analyses have been performed in human (Sun et al. 2009; Harnprasopwat et al. 2010; Gong et al. 2012). For example, an A/G SNP that is located at the 24-nucleotide downstream of the 3’ end of the mature miR126, can significantly block the processing of pri-miR126 to mature miRNA in the RS4;11 cells (Harnprasopwat et al. 2010). On the contrary, a G/A transition occurring in the stem region of miR510 enhanced the production of mature miRNA (Sun et al. 2009). By analyzing the published documents (Jazdzewski et al. 2008; Sun et al. 2009; Harnprasopwat et al. 2010) as well as large-scale analysis of human pre-miRNAs SNPs, Gong et al. (2012) summarized a possible relationship between the SNPs in miRNAs stem and their effects on the generation of mature miRNAs. Based on this rule, we can infer that most of rice miRNAs would reduce the mature miRNAs production, thanks to the instability of hairpin structures caused by SNPs in stem regions. In contrast, near one third of miRNAs could increase the efficiency of processing pre-miRNAs into mature miRNAs, because the SNPs in stems extensively enhanced the stability of hairpin structures.
As reported, miRNAs can not only down-regulate the expression of proteins at the translational level (Brodersen et al. 2008), but also efficiently guide the cleavage of genes mRNA at the post-transcriptional level (Voinnet 2009). It is estimated that approximately 30% of human mRNA genes may be regulated by miRNAs (Friedman et al. 2009). In plants, the total number of genes targeted by miRNAs is greatly decreased compared with that in animals (Jones-Rhoades and Bartel 2004). Accordingly, plant miRNAs have fewer targets relative to animal miRNAs. For example, most Arabidopsis miRNAs have on average six targets or fewer (Jones-Rhoades et al. 2006), whereas each human miRNA can target hundreds of genes (Gong et al. 2012). Here, we found that each WmiRNA has on average 11 targets in rice, which is significantly fewer than that in human (375 targets per wild-type miRNA; Gong et al. 2012). Interestingly, increasing evidence reveals that miRNAs constitute a large and complex gene regulatory network in plants (Meng et al. 2011b). In most cases, one miRNA can target many different mRNA genes; In contrast, one mRNA can also be regulated by two or more miRNAs. We found for example that both the osa-miR820c and osa-miR395 family members can co-target the LOC_Os03g53230.1, a gene that encodes bifunctional 3-phosphoadenosine 5-phosphosulfate synthetase that is highly abundant in the roots of 7-day-old rice seedlings (GSE6893). Two F-box domain-containing protein genes, LOC_Os12g32630.1 and LOC_Os12g32630.2, are the newly created targets of SNP-miR2104 and SNP-miR2102-5p, suggestive of the gain of additional roles for the two SmiRNAs in development transition and in response to environmental stressors in plants (Jain et al. 2007).
It is generally accepted that the modern rice (Oryza sativa) has been domesticated from its Asian wild ancestor O. rufipogon about 10,000 years ago (Kovach et al. 2007). After re-sequencing tens to hundreds of rice genotypes, Huang et al. (2012b) and Xu et al. (2012) demonstrated that artificial selection should impose on the whole rice genome during the long period time of domestication. During the past decades, more than 11 protein-coding genes controlling domestication traits have been identified in rice (Izawa et al. 2009; Jiao et al. 2010). However, the issue whether miRNA genes were under artificial selection during domestication remains to be further clarified. In this study, twelve putative domestication-related miRNAs were identified by searching and comparing the whole rice genome sequences. In the previous studies, Ehrenreich and Purugganan (2008) and Wang et al. (2007; 2010) found evidence for positive selection on some miRNA loci in Arabidopsis and rice, respectively. Notably, the investigation of the whole miRNA repertoire of cultivated and wild rice (O. rufipogon) has supported the notion that some miRNA genes should have experienced strong artificial selection during evolution (Wang et al. 2012b). Therefore, miRNA genes, like protein-coding genes, could serve as one of the driving forces that are important for rice domestication (Wang et al. 2012b).