Introduction

The entire eukaryotic genome is pervasively transcribed and yields to a myriad of non-protein-coding RNA species with complex expression and regulation patterns [1]. These non-coding RNAs (ncRNAs) contribute to evolutionary processes of the genome complexity with increasing amounts of information [2–5]. They have multiple roles in biological processes, including occupy critical edges and nodes in physiological networks, frequently involved in feedback loops [6]. The function of them is more difficult and complexity than we thought, but they will contribute to complex genomic pervasive transcription on cell biology and evolution [7].

As non-protein-coding sequences, most ncRNAs tends to be only weakly constrained than mRNAs through evolutionary processes. The potential function of ncRNA is still enigmatic, although functional ncRNAs are identified as regulatory molecules and play roles in multiple biological processes, such as a class of small (~22-nts) ncRNAs, microRNAs (miRNAs). miRNAs have been identified as crucial regulators of gene expression in multiple physiological and pathological processes [8–12]. Recently, long non-coding RNAs (lncRNAs), a large proportion of non-coding transcripts (longer than 200-nts), have been studied as another class of ncRNAs. The lncRNAs may also play a role in tumorigenesis, such as H19 [13–15]. Indeed, dysregulation of many ncRNAs contribute to pathological processes via versatile functional roles. Different ncRNAs may have close relationships according to their location distributions and sequence similarity. For example, H19 is a host gene of miR-675 and imprinted and endogenously expressed, and H19 can function as pri-miRNA of miR-675 [16, 17].

LncRNAs and miRNAs are widely studied because of their crucial regulatory roles in biological processes and evolutionary players, especially for the well-conserved small miRNAs. Although they are characterized as ncRNAs and negative regulatory molecules, they are thought to have different evolutionary patterns. Here, to further understand evolutionary and crosstalk of miRNA–lncRNA, we performed an integrated analysis of miR-675 and Hg19 genes across different mammalians. Simultaneously, the neighboring Igf2 gene of H19 was also involved in the integrative analysis. The study will enrich the ncRNA study, especially for the potential crosstalk of miRNA–lncRNA and mRNA–ncRNA, which provides potential interaction in the coding-non-coding RNA regulatory network.

Results

The long non-coding RNA, H19 gene, was located on different chromosomes with various length distributions across mammalians (Table 1). The lncRNA included the small non-coding RNA, mir-675 gene. Although both of H19 and mir-675 were identified as ncRNAs, they showed inconsistent nucleotide divergence and evolutionary patterns. Compared to small miRNAs, H19 was involved in higher levels of nucleotide transition/transversion, insertion/deletion across different animal species (Table 2). The hypothesized pri-miRNA sequences showed similar nucleotide divergence pattern with H19, while both pre-miRNA and mature miRNAs were well-conserved phylogenetically as expected. Indeed, compared to other broadly conserved miRNA gene families (such as let-7 gene family), the miR-675 was identified as a poorly conserved and mammalian-specific gene family.

Table 1 Hg19 information in different mammalians
Full size table
Table 2 Average frequencies over all species based on different sequences
Full size table

Nucleotide compositions indicated that guanine (G) was the dominant nucleotide in the lncRNA, while cytosine (C) was the most dominant nucleotide in pri-miRNA (Fig. 1a, b). Interestingly, distributions of nucleotides in mouse and rat were inconsistent with other animals based on H19 and hypothesized pri-mir-675, but the diverse distributions were moderated in pre-mir-675. The mature small RNAs showed inconsistent nucleotide compositions with other longer sequences, although miR-675-5p/3p were the part sequences of them (Fig. 1). The two mature miRNAs showed inverse nucleotide compositions, which ensured the stem-loop structure of pre-miRNA.

Fig. 1
figure 1

Frequency distributions of nucleotide composition. Frequency distributions of nucleotides (A, T(U), C and G across different animal species; distributions of average nucleotide pair frequency; percentage of average nucleotide pair frequency) of H19 (a), hypothesized pri-mir-675 (b), pre-mir-675 (c), miR-675-5p (d) and miR-675-3p (e) sequences. H19 and pri-mir-675 show consistent distribution patterns, and Rodentia (mmu and rno) restricts divergence of nucleotide composition than other mammals

Full size image

Reconstructed phylogenetic trees showed that pre-mir-675 showed inconsistent phylogenetic relationships between H19 and pri-mir-675 sequences (Fig. 2, Fig. S1). No significant divergence could be detected between trees of H19 and pri-mir-675. The evolutionary analysis of H19 and pri-mir-675 indicated that mouse (mmu) and rat (rno) had larger genetic distances with other animal species, but the rodentians were clustered together with primates in evolutionary relationships of pre-mir-675. All the precursor miRNAs showed stable stem-loop structures with various minimum folding free energy (MFE) and loop regions (Table S1). Although both the two arms could yield potential negative regulatory molecules, miR-675-3p was involved in higher level of nucleotide divergence than miR-675-5p (Fig. 3). miR-675-5p was identified as the original annotated mature miRNA, and had higher or dominant expression level than its partner [16]. Phylogenetic networks indicated that miR-675-3p had more complex evolutionary pattern, and was involved in more median vectors (hypothesized sequence types in reconstructed network) (Fig. 3b).

Fig. 2
figure 2

Phylogenetic relationships based on different sequences. a NJ tree of H19 sequence. b NJ tree of hypothesized primary mir-675 sequences (pri-675). c NJ tree of precursor miR-675 sequences (pre-675). Phylogenetic tree of pre-675 sequences show inconsistent distributions with H19 and pri-675 sequences

Full size image
Fig. 3
figure 3

Sequence diversity and phylogenetic networks of mir-675. a Multiple sequence alignment of mir-675 and the two mature miRNAs (miR-675-5p and miR-675-3p). Compared to other well-conserved miRNAs, miR-675 sequences are poorly conserved across different animal species. They are always involved in larger genetic distances via nucleotide divergences. b The two products show inconsistent evolutionary patterns

Full size image

Further, according to predicted target mRNAs of miR-675-5p/3p, dog (cfa) and cow (bta) showed more private or specific target mRNAs, although other species shared more common targets (Table S2). Strikingly, the ever thought as passenger strand, miR-675-3p, indicated more predicted potential target mRNAs than miR-675-5p. Functional enrichment analysis suggested that some target mRNAs were involved in important biological processes, such as lysine degradation, taurine and hypotaurine metabolism, valine, leucine and isoleucine biosynthesis in human, and etc. (Table S2). These results indicated that miR-675 played a versatile biological role via negatively regulating multiple target mRNAs.

Moreover, to understand the evolutionary pattern of host gene of lncRNA, we simultaneously analyzed the Igf2 gene. The Igf2 gene might be located in the upstream region or downstream of the H19 gene across mammalians (Tables 1, 3). Compared to related ncRNAs, including lncRNA and miRNA, protein-coding genes were also involved in higher level of nucleotide substitutions, insertions/deletions (Fig. S2; Table S3). The Igf2 gene showed similar phylogenetic relationships with pre-miRNAs and mature miRNAs (Figs. S1, S2).

Table 3 Igf2 genes in different mammalians
Full size table

Discussion

Non-protein-coding RNAs (ncRNAs), including miRNAs and lncRNAs, are major players that increase organismal complexity as robust regulatory molecules. These regulatory RNA species are involved in complex overlapped expression and regulation patterns [1], which contributes to coding-non-coding RNA regulatory network. Deregulated expressed ncRNAs may also play a role in physiological and pathological processes, even in tumorigenesis.

The longer (lncRNA) and small (miRNA) ncRNAs may be involved in potential relationships. For example, some lncRNAs have been identified as potential pri-miRNAs to yield mature miRNAs, such as H19 gene [16, 17]. The lncRNA is imprinted with maternal expression and plays a role in tumorigenesis [13–15], and simultaneously it is also identified as a primary transcript of miR-675 [16, 17]. Although distributions of average nucleotide pair frequency in H19, hypothesized pri-mir-675, pre-miRNA and mature miR-675 are quite similar, while frequency distributions of nucleotides are various (Fig. 1). The diversity indicates the bias of nucleotide compositions with different sequence lengths, which may contribute to functional need as well as nucleotide compositions in miRNAs. H19 contains miR-675, which suggests close relationships between miRNA and lncRNA. The phenomenon is not random, and other imprinted ncRNAs might also give rise to miRNAs [16]. The interesting relationships may increase cross-talk between different ncRNAs, and further enrich coding-non-coding RNA regulatory network via their versatile biological roles.

Although the miRNA is located in the H19 gene, phylogenetic relationships based on H19, hypothesized pri-miRNA and pre-miRNA show various distributions, especially for between pre-miRNA and pri-miRNA/H19 (Figs. 1, 2 and Fig. S1). The longer sequences are always prone to be involved in more nucleotide substitutions, insertion/deletion (Table 2), although some special regions are well-conserved (Fig. S3). We found that internal miRNA do not show special conserved region than other regions, except for special nucleotide compositions (Fig. 1). Indeed, the miRNA is a mammalian-specific species according to the annotated miRNAs in the miRBase database (Release 19.0), and it has been identified as a poorly conserved gene family (Fig. 3). Both miR-675-5p and miR-675-3p can be detected nucleotide divergence across mammalians, even though in the nucleotides 2–8 (the typical “seed sequences” of miRNA) (Fig. 3a). The inconsistent “seed sequences” lead to various target mRNAs, which simultaneously enriches the more potential versatile biological roles of the miRNA family. The two mature miRNAs also show specific evolutionary networks, and miR-675-3p is involved in more nucleotide substitutions and median vectors (Table; Fig. 3b). These results implicates that the two miRNAs are differentially evolved that is mainly derived from the functional pressures. As the reported canonical miRNA sequence, miR-675-5p is dominantly expressed as a crucial regulatory molecule, but its partner is always rarely expressed although it is also potential active regulatory molecule. The two miRNAs may have important roles in multiple biological processes via interacting with their potential target mRNAs (Table S2).

Imprinted genes in mammalian are always clustered with lncRNAs [18]. The imprinting mechanism of Igf2-H19 is conserved in therians, but the distributions can be flexible across different animal species (Tables 1, 3). Between the coding and non-coding genes, Igf2 and H19 show similar evolutionary patterns, although lncRNAs are always thought to be poorly conserved (Table 2; Figs. 1, 2, Figs. S1, S2; Table S3). The main reasons may be derived from the imprinting mechanism and functional implication and evolutionary pressure. Compared to the two longer coding and non-coding sequences, the internal miRNA, is also well-conserved. However, compared to other broadly conserved miRNAs, the mammalian-specific miRNA species is identified as a poorly conserved gene family. On contrast, the H19 lncRNA is conserved in evolutionary process, although the lncRNAs are always thought as poorly conserved RNA molecules. The H19 and mir-675 show inconsistent distributions of the MFE structures, despite both of them show mountain plots (Fig. S4). Specially, although both of the two arms of mir-675 can yield mature miRNAs (miR-675-5p/3p), the two regulatory molecules show inconsistent evolutionary patterns (Fig. 3). The identified as canonical miRNA (miR-675-5p) is more conserved than ever thought as passenger strand of miR-675-3p. Accumulating evidences have shown that miRNA passenger strands, also termed miRNA* sequences, may be potential regulatory molecules and have potential biological roles in specific developmental stages [19–23]. However, perhaps because of dynamic and temporal functional need, miRNA* sequences are always involved in rapid evolutionary processes inter- and extra-species [21]. Functional analysis based on predicted target mRNAs also suggests versatile biological roles of miR-675 (Table S2). Collectively, the evolutionary process of imprinted cluster may be slightly diverged that is mainly derived from functional implication, although all of them are well-conserved across complex evolutionary patterns inter- and extra-species.

Materials and methods

All the H19 sequences in the study were obtained from the GenBank (Table 1). The pre-miRNA of miR-675 were collected from the miRBase database (Release 19.0, http://www.mirbase.org/) [24]. Although H19 can be thought to be a pri-miRNA of miR-675, to further understand the evolutionary patterns of miRNAs, we also hypothesized the pri-miRNA sequences through extending bilaterally to 500 nt of pre-miRNAs. The miRNAs in some species (such as Bos taurus and Canis familiaris) were not annotated or reported. We hypothesized these pre-miRNAs, pre-miRNAs and mature miRNAs by homology search, because the small ncRNAs are well-conserved phylogenetically [9, 25]. The secondary structures of pre-miRNAs were estimated in MiPred web server [26]. The secondary structure of H19 and pri-miRNAs were predicted in RNAfold WebServer (http://rna.tbi.univie.ac.at/cgi-bin/RNAfold.cgi) [27].

Phylogenetic relationships of H19, pri-/pre-mir-675 and miR-675-5p/3p were reconstructed with MEGA 5.10 [28] using Neighbor-Joining method, SplitsTree 4.10 [29] using Neighbor-Net method [30], and Network 4.6.1.0 (http://www.fluxus-engineering.com/) using Median-Joining method. These sequences were aligned with Clustal X 2.0 [31]. The nucleotide composition and nucleotide pair frequency were estimated in MEGA software. Further, target mRNAs of miR-675-5p/3p were predicted by TargetScan program [32], and functional enrichment analysis was performed using CapitalBio Molecule Annotation System V4.0 (MAS, http://bioinfo.capitalbio.com/mas3/).