Background

MicroRNAs constitute a family of small non-coding RNA molecules (of 20–25 nucleotides length) that do not code for protein. Most of them are present in the introns while some exist in the exons of a gene. They play a major role in gene expression regulation and RNA silencing. The dysregulation of any miRNA may be associated with diseases like cancer or heart disease. Similarly, the isomiRs that are defined as the variants of miRNA sequence could have evolved through various cellular processes. The sequence variation in the isomiRs arises through different processes including addition or deletion etc. in the canonical miRNA sequence. In this review, we will discuss isomiRs, their biogenesis, functional involvement in various diseases and some databases that can be used for their prediction.

MicroRNA

MicroRNAs (miRNAs) are the post-transcriptional regulators of gene expression in most of the eukaryotic cells. These miniature RNAs belong to a family of small (~ 20–22 nucleotides in length) non-coding RNA-molecules that are expressed in wide range of organisms including plants, animals, and worms [1]. The miRNA-mediated RNA-interference was first discovered in 1993 by Ambros and coworkers while working on lin-4 gene involved in larval development of Caenorhabditis elegans (C. elegans) [2, 3]. The expression of about 30% of the protein-coding genes [4,5,6,7] is regulated by these non-coding RNA sequences through binding to the 3′-untranslated regions (3’-UTR) of specific mRNAs.

Over a thousand of miRNA-encoding genes are present in the human genome [8], which are directly or indirectly associated with more than 30% of the protein coding genes [9, 10]. The miRNAs precursors are not found within the coding regions of transcripts or the corresponding antisense strand, instead, they are present in clusters within intergenic regions and introns of protein-coding genes with some exceptions such as hsa-miR-935, hsa-miR-302 [11]. The biosynthesis of miRNAs differs in plants and animals. In plants, the entire process of biogenesis of primary-miRNA (pri-miRNA) to mature miRNA occurs in the nucleus by RNase III enzyme DCL1 (Dicer-like 1). While in animals, the process occurs in the cytoplasm as well as in nucleus by two different enzymes: Drosha/Pasha in nucleus and Dicer in the cytoplasm [3, 12]. A vast array of miRNA-repertoire has been identified and reported in different animal and plant species. Recently, in our laboratory, we have identified novel bubaline-miRNAs (using small RNA sequencing and analysis) [13,14,15] and also by biocomputational analysis of whole genome shotgun sequences [16]. It has also been reported that target genes of differentially expressed miRNAs (from healthy as well as diseased individuals) can have a critical role in innate immunity and TLR (Toll like receptors) signaling pathways [14]. An array of isomiRs is the isoforms of a miRNA indicating that these might be functionally important. The present review aims at elaborating the biogenesis and features of isomiRs, and their roles in biological processes, with special emphasis on diseases in animals.

IsomiRs

A single miRNA could differ by a small number of base changes (due to insertion/deletion or substitution) at the 5′ or 3′ termini, resulting in the formation of isomers of specific miRNA, termed as 5′ or 3′ isomiRs, respectively [17]. Thus, isomiRs are derived by imprecise cleavage or any change in processing of mature miRNA from primary transcripts by Drosha and Dicer enzymes and nucleotide addition at 3′ end [18]. IsomiRs can also be generated through RNA editing and single nucleotide polymorphisms (SNPs) from the canonical (any molecular sequence that represents the consensus sequence of that particular molecule) miRNA sequence [18, 19].

Prior to proceeding further on isomiRs, it is necessary to discuss in brief about the nomenclature of miRNAs and isomiRs and their features. There are different ways of writing the names of miRNAs based on the nomenclature. For example- “MIR” refers to the gene that encodes miRNA, “miR” refers to the mature form of miRNA and “mir” refers to the pre-miRNA and the pri-miRNA [20]. Similarly, there are various terms used for isomiRs on basis of their appearance and for analysis purpose. Table 1 adumbrates some of these terminologies that pertain to isomiRs (Table source: https://bioinfo2.ugr.es/miRanalyzer/miRanalyzer_tutorial.html).

Table 1 Terminologies related to isomiRs
Full size table

Biogenesis of isomiRs

IsomiRs are generated by post-transcriptional modifications of the corresponding miRNA sequence. isomiRs have diverse roles in animals, plants, and viruses. To understand the biology of isomiRs, it’s required to adumbrates the process of miRNA production in animals. miRNA biogenesis occurs through different steps in the cytoplasm and nucleus. The process starts in the nucleus where precursor miRNA (pre-miRNA) is transcribed from miRNA coding gene by RNA polII/III, which is then cleaved into shorter sequences by Drosha/DGCR8 complex to form primary-miRNA. The pre-miRNA is then exported from nucleus to cytoplasm by Exportin-5. Mature miRNA is formed from this precursor miRNA through miRNA/miRNA duplex formation. This precursor miRNA is also involved in the formation of isomiRs [21, 22]. Alternatively different pathways can also be involved that can alter the sequenceof canonical miRNA. Further, the isomiRs can be 5′-isomiRs or 3′-isomiRs (Fig. 1). It is evident that the ‘miRNA: miRNA duplex’ can form through complementary nucleotide base pairing of miR-5p and miR-3p generated from pre-miRNA [19].

Fig. 1
figure 1

Flow-diagram to demonstrate the biogenesis of miRNA and its variants (isomiRs) in normal adult cells of an animal (The sequence used here is arbitrary for explanation purpose only)

Full size image

Based on variation in the canonical sequence, isomiRs can be template dependent or non-template dependent. In the template dependent, 5′ or 3′ nucleotide shifting occurs due to exoribonuclease, which ultimately results in the formation of isomiRs. On the other hand, in non-template dependent variation results in post-transcriptional RNA editing and tailing. Adenosine to Inosine (A to I) is commonly occurring RNA editing and diverse set of nucleotidyltransferases are responsible for post-transcriptional tailing. Both these events carry out the formation of isomiRs [23] (Fig. 2).

Fig. 2
figure 2

Different methods of generation of isomiRs from canonical miRNA. During miRNA biogenesis, the canonical miRNA can have template dependent and non-template dependent variations which result in synthesis of miRNA variants known as isomiRs

Full size image

It has been consistently observed that the 5′-ends of the miRNAs are uniform that results in the formation of 5′-isomiRs due to different characterized 5′-ends. Thus, in comparison to miRNA with a single seed, the miRNA with 5′-isomiRs could have a significantly diverse target range and functional influence. These types of isomiRs have been recognized in many species including fruit fly (Drosophila melanogaster), mice (Mus musculus) and human [24]. The significance of 5′-isomiR expression on miRNA target repertoires was studied on vertebrate miR-142-3p by Manzano and coworkers [25]. They reported that 5′-isomiRs that differ from each other by 1 nucleotide can have highly divergent or convergent target ranges. Similarly, any variation in the 3′ end of miRNA can alter the stability and the efficiency of target repression of miRNA [26] (Fig. 2).

Evolutionary perspective of isomiRs

The evolutionary pattern of different miRNA families in divergent species can provide information on functional adaptations and associations in various species of plants and animals. Some of the researchers have provided information on the evolutionary perspective of isomiRs in different taxa. A recent study conducted in vivo on mammalian cells has shown that, Dicer can alternatively process these miRNAs to generate 5′-isomiRs. This alternative Dicer processing can be the result of the sliding bulge in the pre-miRNAs indicating that secondary structure of miRNAs can be responsible for the generation of 5′-isomiRs [27]. Tan and team [28] demonstrated through Northern blot analysis that a proportion of individual miRNA in human may vary spatially. Besides, the team also analyzed the isomiRs from miRBase and inferred that during evolution, canonical miRNAs have been replaced by 5′ isomiRs. This finding supports the hypothesis that isomiRs are the procreation of evolutionary relices of the miRNA genes. The expression profiles of miRNA and isomiRs can be regulated through arm selection (miRNAs may be derived from different arms) or arm switching (two arms usually exhibits dynamic expression patterns). It has been reported that two arms miR-3p and miR-5p contribute to different evolutionary/expression patterns of miRNA and isomiRs [29]. The structural regions within the miRNA (mature miRNA or seed loop) can be under evolutionary pressure. However, the miRNAs that are present within miRNA clusters (miRNA in close proximity of other miRNA) can share similar structures including seed sequences. Gene ontology and analysis of miRNA clusters can provide insight information regarding molecular functions of these clusters. However, evolutionary conservation of isomiRs warrants further in depth study and exploration [30].

Importance of isomiRs

IsomiRs act as potential regulatory molecules and are associated with repression of the target-mRNAs [31]. Several isomiRs are directly involved in post-transcriptional gene silencing, and could also affect the homologous miRNA itself. For example, the isomiRs may increase stability of miRNA and modify the effectiveness of miRNAs through RISC (RNA-induced silencing complex) during maturation process [32]. The small RNA (sRNA) deep-sequencing data showed that isomiRs are differentially expressed in developmental genes and tissues of fruit fly (Drosophila melanogaster). The addition of ‘A’ or ‘U’ at 3′ end of miRNA during early or late embryogenesis suggests that the stability of miRNA or miRNA: target interactions get strengthened. [33].

It has been reported that conserved target sites for isomiRs can be predicted by using biocomputational analysis. A study using immunoprecipitation showed that isomiRs can be incorporated into argonaute proteins and from the luciferase assay it has been observed that isomiRs are different from their canonical miRNA on a functional basis [34]. To study the features and regulatory targets of mature miRNAs, a comprehensive analysis was performed on Arabidopsis. The computational analysis revealed that isomiRs play a key role in gene regulation via terminal heterogeneity, which ultimately enhances the specificity of target gene silencing. The analysis of degradome data available in this experiment indicates the biological role of isomiRs in target cleavage. Further, the comparison of novel predicted and validated target genes with commonly targeted mRNA genes from Gene Ontology (GO) demonstrated that the validated targets are bound to isomiRs along with canonical miRNA [35]. IsomiRs have also been identified in human lymphoblastoid cell lines whose expression was population specific as well as gender-dependent. It has been reported that these isomiRs participate in RNA interference (RNAi) pathway through their association with Argonaute silencing complex [36].

Chan and colleagues [37] reported that not all, but some isomiRs may share common mRNA targets. They investigated isoforms of miR-31 (miR-31H, miR-31P and miR-31 M) that differ slightly at 5′ and 3′ ends. The study investigated concordant and discordant regulation, displayed by 6 known target genes (CEBPα, E2F2, STK40 etc) of isomiR-31. The results revealed that in cell-based systems, isomiRs exhibit similar and disparate regulation of target genes. Moderate level of isomiR expression has been observed between Dicer-independent miRNA and non-dominant miRNA, suggesting complex miRNA maturation process at isomiR level [38]. The study performed by Mercey and colleagues [39] on human miR-34/449 family suggested that isomiR variants which differ by single canonical counterpart can share biological functions indicatind additional mechanism by which regulation of complex biological function can be perfectly and easily employed by miRNA machinery.

Role of isomiRs in human diseases

The relationship of isomiR expression and disease progression is not clear till date. However, some reports are there which show some association of isomiRs with different diseases. In the case of Alzheimer’s disease (AD), a significant change in miRNA isoforms was found between early and late stages of the disease. An entropy based MIH5 model was introduced to identify effects of dysregulation of miRNA isoforms at 5′end. The results indicated that as compared to the expression-based method, the entropy-based method is most stable to detect miRNA related to AD [40]. Using miR-183-5p in breast cancer and normal breast data sets of 2 races (white and black) from the Cancer Genome Atlas repository (https://www.cell.com/pb-assets/consortium/pancanceratlas/pancani3/index.html), it has been reported that as compared to archetype miRNA, full isomiR profile from known and novel human-specific miRNA can provide better results to distinguish between normal and tumorous tissueas its isomiRs were upregulated in breast cancer related white women but not black. Also, a distinct impact of these isomiRs on cellular transcriptome has been identified due to overexpression of isomiRs in MDAMB- 231 cells followed by microarray analysis [41].

The overexpression of canonical miRNA hsa-miR-140-3p and its 5’isomiR-140-3p has been reported in breast cancer patients. Reduced cell viability was observed in breast cancer cell lines (MCF10A, MDA-MB-468, and MDA-MB-231) due to higher expression of 5’isomiR-140-3p. The cell cycle analysis showed that 5’isomiR-140-3p caused decreased cell migration and cell cycle arrest at G0/G1 phase. The data revealed that 5’isomiR-140-3p contributes to tumor-suppressive effects by reducing breast cancer proliferation and migration, [42]. In another study, small RNA (sRNA) sequencing databases were used to study the expression of miRNA and isomiRs by correlating with gender difference. It was determined that the expression of miRNA can vary between different tissues and genders. From the statistical analysis they concluded that there is a significant difference between the expression of miRNA and isomiRs in tumor and normal tissues in both sexes. The study suggested that screening of miRNA/isomiRs associated with the disease could be affected by gender difference [43].

Babapoor and team [44] identified isomiR sequence of miRNAs that were deregulated in cutaneous melanomaand found that miR-451a was functioning as a tumor suppressor in gastrointestinal cancer cells and glioma cells. They concluded that miR-451a was involved in melanoma progression while the mature form of this isomiR, miR-451a.1 was associated with amelanotic melanoma [44]. Zhang and coworkers [45] developed a new method to discover the catalogue of isomiRs in association with cancer progression in human. Differentially expressed isomiRs were detected using DEseq algorithm, followed by rank based MANOVA. The expression pattern between normal and tumorous tissues when compared using MANOVA algorithm and the biological functions of isomiRs when elucidated using functional enrichment analysis. There was significant inconsistency in the expression of multiple isomiRs derived from same miRNA locus in normal and tumor samples .

The miRNA transcriptional response of human dendritic cells to various mycobacterium infections showed that bacterial infection has a strong impact on the cellular immune response of host which can alter the expression and proportion of miRNA isoforms. A total of 1595 isomiRs corresponding to 235 miRNAs were detected with alteration in the expression due to bacterial infection using DESeq algorithm (which is based on negative binomial distribution). isomiRs. It has been observed that due to bacterial infection, as compared to start site, the end regions of isomiRs show great variability. The miRNA repertoire involved in providing immunity to Mycobacterium tuberculosis was also identified using deep sequencing [46, 47]. The results of transcriptome and miRNA analyses of human peripheral blood mononuclear cells (PBMCs) were also used to identify the isomiR profile. The results from data entered in miRBase indicated that the most abundant isomiR sequences did not match the reference miRNA sequence. This specifies that there is a dynamic change in the relative expression level of isomiRs derived from the same precursor (which may vary depending on cell type and its differentiation status) [48].

IsomiRs in livestock

A very limited research has been conducted on role of isomiRs in livestock. The study on tooth morphogenesis in miniature pigs (Sus scrofa) revealed that out of 11 unique miRNA sequences, a total of five (mir-103, mir-107, mir-133a, mir-133b, mir-127) belong to isomiR families and play an important role in developmental stages of teeth including incisors, canines, bicuspids and molars [49]. The miRNA study during bovine oestrous cycle revealed the presence of isomiRs in bovine plasma using sRNAbench. A total of 655 isomiRs were identified from 134 canonical miRNA with modifications on 3′-end, 5′-end and also in the middle of canonical miRNA. Some of the isomiRs were showing higher expression level than their canonical sequence suggesting that some bovine miRNA do not correspond to their isoforms [50]. In case of cattle some isomiRs were showing 3′ or 5′ end variation, but some isomiRs, for example, miR-125-p showed both 3′ as well as 5′ end variations. IsomiRs have also been found in porcine muscle cell study (miR-423) and in longissimus muscle of sheep (miR-96) at 3′ end variants [51]. The results from Illumina deep sequencing of pig miRNA and isomiRs sequence revealed its role during early pregnancy [52]. The miR-127-3p is one of the highly expressed miRNA in retina and RPE/Choroid. The deep sequencing study conducted on mouse retina indicated the presence of similar level of miR-127-3p and its isomiRs [53]. In a recent study novel miRNAs and their orthologs has been detected in left ventricular wall of rat heart [54]. Transcriptome sequencing analysis performed on murine HL-1 cells identified that 5′ isomiRs target the genes that are involved in cardiovascular disease. In the same study 2 identical genomic loci of miR-133a 5′ isomiR (i.e. miR-133a-1 and miR-133a-2) were identified in mammals that process different loci [55].

IsomiR databases

IsomiR Bank (http://mcg.ustc.edu.cn/bsc/isomir/) is a free online database created to integrate detected isomiRs. According to the data provided by Zhang and coworkers [56], a total of 308,919 isomiRs collected from 4706 mature miRNA are present in isomiR Bank. This bank provides the analysis of target prediction and enrichment for evaluating isomiRs effects on target selection. Another online platform miR-isomiRExp has been developed recently to analyze the expression of specific miRNA at miRNA/isomiR level. This software can reveal functional characteristics and can provide the whole mechanism involved in maturation and processing of miRNA/isomiR. This platform also provides information regarding deregulated miRNA loci and detailed isomiR sequence [57].

There are various tools available to predict and quantify known and novel miRNA including miRDB (http://mirdb.org/), miRFinder (http://www.bioinformatics.org/mirfinder/), and miRDeep2 (https://www.mdc-berlin.de/content/mirdeep2-documentation). Another software miRPro (https://sourceforge.net/projects/mirpro/ › Browse) has been developed recently which is able to predict novel miRNA and can quantify the known miRNA. This software can also detect isomiRs, which is not possible with miRDeep2. It also includes genome annotation based read count, optional seed region check, expression quantification of miRNA and arm switching detection [58]. It has been reported that many tools do not provide information regarding isomiRs of specific miRNA and conserved miRNA-mRNA interaction sites. To overcome these problems Urgese and colleagues [59] introduced a novel software named isomiR-SEA. This software characterizes the seed presence of miRNA in input tags and evaluates the position of mismatches. Thus it recognizes isomiRs and characterizes the interaction sites for miRNA-mRNA.

Conclusion

Isoforms of miRNA may provide detailed information about the specific cell or tissue type. The miRNAs and their variants can be used in disease association and detection in different species by studying differential expression pattern. However, more detailed and in-depth studies are warranted to explore the usability of isomiRs as markers for important diseases in animals and humans.