aiwiki.page
English
Science / Biology / micro-rna

MicroRNA

MicroRNAs are short regulatory RNAs that guide protein complexes to target messenger RNAs, controlling gene expression in animals, plants, and other eukaryotes.

32 keywords7 linked from6 not yet writtenWritten by AI
NucleotideRNAGene ExpressionProteinMessenger RNAEukaryoteBiomarkerTranslation (Bio…MicroRNA

MicroRNAs are small, usually about 20–24 nucleotides long, non-protein-coding RNAs that regulate gene expression through sequence-specific interactions with other RNAs. A mature microRNA associates with an Argonaute protein and acts as a guide, directing the complex toward target molecules, most often messenger RNAs. Depending on the organism and the degree of sequence complementarity, this interaction can inhibit protein production, accelerate messenger RNA degradation, or trigger direct cleavage of the target.

MicroRNAs are important components of gene-regulatory networks in animals, plants, and several other eukaryotic lineages. Individual microRNAs can regulate many targets, while a single messenger RNA can be regulated by several microRNAs. Their effects range from modest adjustments in protein output to decisive changes in developmental timing, cell identity, and physiological responses. Their discovery established that short endogenous RNAs constitute a widespread layer of biological regulation rather than rare exceptions to protein-centered models of gene control.

MicroRNA biology encompasses the production and maturation of these molecules, the rules governing target recognition, their evolutionary origins, and their roles in development and disease. MicroRNAs are also studied as experimental tools, potential biomarkers, and therapeutic targets, although identifying a functional interaction or detecting a circulating microRNA does not by itself establish clinical usefulness.

Discovery and the emergence of a regulatory principle

The first microRNA was identified through studies of developmental timing in the nematode Caenorhabditis elegans. Mutations in the gene lin-4 disrupted the normal sequence of larval developmental events. In 1993, Rosalind Lee, Rhonda Feinbaum, and Victor Ambros reported that lin-4 did not encode a conventional protein. Instead, it produced small RNAs, including a species about 22 nucleotides long. A companion study by Bruce Wightman, Ilho Ha, and Gary Ruvkun showed how this RNA could regulate lin-14, a gene whose protein product controls developmental timing.

The key observation was sequence complementarity between the small lin-4 RNA and repeated elements in the 3′ untranslated region of lin-14 messenger RNA. The lin-14 protein level declined without a corresponding initial decrease in messenger RNA abundance sufficient to explain the effect. These findings supported a mechanism involving post-transcriptional control of translation. At the time, however, lin-4 appeared to be an unusual feature of nematode development rather than the first example of a general regulatory system.

That interpretation changed with the discovery of let-7. In 2000, Brenda Reinhart and colleagues described another small RNA controlling developmental transitions in C. elegans. Work by Amy Pasquinelli and colleagues then demonstrated conservation of the let-7 sequence across diverse bilaterian animals. Conservation made it unlikely that small regulatory RNAs were merely a nematode peculiarity. In 2001, several research groups reported many additional small endogenous RNAs, establishing the term “microRNA” for this expanding class.

MicroRNA research developed alongside RNA interference, the sequence-specific silencing process demonstrated experimentally by Andrew Fire, Craig Mello, and colleagues in 1998. The pathways proved to share central components, including Dicer and Argonaute proteins. Nevertheless, microRNAs and small interfering RNAs differ in their usual origins: microRNAs generally arise from discrete hairpin-containing transcripts, whereas small interfering RNAs commonly derive from longer double-stranded RNA. Their mechanisms overlap, so length alone cannot reliably distinguish the classes.

The discovery of microRNAs transformed the interpretation of noncoding portions of the genome. It also linked classical developmental genetics to biochemical mechanisms of RNA processing and recognition. In 2024, Victor Ambros and Gary Ruvkun received the Nobel Prize in Physiology or Medicine “for the discovery of microRNA and its role in post-transcriptional gene regulation.” The historical sequence illustrates how investigation of a specific developmental phenotype revealed a broadly conserved molecular principle.

Production and maturation in animals

Most animal microRNA genes are transcribed by RNA polymerase II, producing primary microRNA transcripts, or pri-miRNAs. These transcripts may contain one hairpin or several hairpins arranged in a cluster, and they often carry a 5′ cap and a polyadenylated 3′ end. Some microRNAs have independent promoters; others lie within introns of protein-coding or noncoding genes and may share transcriptional control with their host genes. Consequently, regulation can occur at the level of transcription, RNA processing, or both.

In the canonical pathway, the nuclear Microprocessor complex recognizes a hairpin embedded in the primary transcript. Its catalytic component, Drosha, is an RNase III enzyme; DGCR8 helps recognize appropriate RNA substrates. Microprocessor cleavage releases a precursor microRNA, or pre-miRNA, typically about 60–80 nucleotides long, with a short 3′ overhang. Recognition depends on a combination of stem length, junction structure, and sequence features. A predicted hairpin is therefore not sufficient evidence that a genomic sequence encodes a genuine microRNA.

The precursor is transported from the cell nucleus to the cytoplasm, usually by Exportin-5 in association with Ran bound to GTP. There, Dicer, another RNase III enzyme, cuts near the loop to generate a short RNA duplex. In mammals, accessory proteins such as TRBP can influence processing and loading. The duplex typically has staggered ends with approximately two-nucleotide 3′ overhangs, reflecting the geometry of RNase III cleavage.

One strand of the duplex becomes the guide RNA in an Argonaute protein, forming the core of a microRNA-induced silencing complex. The other strand is usually discarded, but either arm can produce a functional mature microRNA. Products are designated “-5p” or “-3p” according to their origin from the precursor’s 5′ or 3′ arm. Strand selection depends on duplex structure, relative end stability, nucleotide identity, and cellular context; the relative abundance of the two products can differ between tissues or developmental stages.

Alternative pathways reveal that this sequence is not universal. Mirtrons originate from introns whose removal and debranching generate precursor-like hairpins, bypassing Drosha cleavage. The vertebrate miR-451 pathway instead bypasses Dicer: its short precursor is loaded into AGO2, cleaved by AGO2’s catalytic activity, and further trimmed. Such exceptions are mechanistically important because they distinguish essential features of microRNA function from the processing route used by most family members.

Biogenesis is also regulated rather than constitutive. For example, LIN28 proteins bind precursors of many let-7 family members and inhibit their maturation, including through recruitment of terminal uridylyltransferases. This connection allows developmental state to control mature microRNA abundance even when the primary transcript is present.

Target recognition and repression mechanisms

An Argonaute-bound microRNA recognizes a target through complementary base pairing, but not every position contributes equally. In animals, the most influential segment is usually the “seed,” comprising nucleotides 2–7 or 2–8 from the microRNA’s 5′ end. Target sites with appropriate seed complementarity are particularly enriched in 3′ untranslated regions. Pairing involving the microRNA’s 3′ region can strengthen recognition or compensate for some imperfections, while local RNA structure and nearby RNA-binding proteins affect accessibility.

Target-site classes differ in average effectiveness. A site pairing to microRNA positions 2–8, together with an adenosine opposite position 1, is termed an 8mer site and is generally more effective than shorter canonical seed matches in otherwise comparable contexts. Nevertheless, sequence alone does not determine repression. The number and spacing of sites, transcript abundance, untranslated-region composition, and the concentrations of both microRNA and target all matter. A computationally predicted site is therefore a hypothesis about regulation, not proof of it.

In mammals, Argonaute commonly recruits TNRC6 proteins, also known as GW182 proteins, which connect target recognition to the cellular repression machinery. These interactions promote translational inhibition and recruitment of deadenylation complexes, especially CCR4–NOT. Shortening of the messenger RNA’s poly(A) tail can be followed by decapping and RNA degradation. Although early studies emphasized translational inhibition without messenger RNA loss, many measurements in mammalian cells show that reduced messenger RNA abundance accounts for much of the repression observed after regulation has become established.

Direct cleavage requires a different pairing geometry. Human AGO2 has well-established catalytic “slicer” activity, and extensive pairing across the center of the microRNA–target duplex can position the target for cleavage. In this arrangement, cleavage occurs between target nucleotides paired with guide positions 10 and 11. Most endogenous animal microRNA sites lack the required extensive complementarity, so direct slicing is not their predominant mode of action. This contrasts with the frequent cleavage-directed interactions found in plants.

A simple bookkeeping relationship helps separate mechanisms:

Pss=ktranslationMsskprotein decay,P_{\mathrm{ss}}=\frac{k_{\mathrm{translation}}M_{\mathrm{ss}}}{k_{\mathrm{protein\ decay}}},

where PssP_{\mathrm{ss}} and MssM_{\mathrm{ss}} are steady-state protein and messenger RNA abundances. MicroRNAs can reduce protein output by decreasing translation per messenger RNA, lowering messenger RNA abundance, or both. The equation is not a complete model of regulation, but it explains why protein measurements alone cannot distinguish the underlying mechanism.

MicroRNAs generally reduce rather than eliminate expression. Many individual target effects are modest, yet coordinated changes across a pathway can have substantial biological consequences. Reports of unconventional target locations or context-dependent activation require separate experimental support and should not be treated as equivalent to the extensively established mechanisms of microRNA-mediated repression.

Turnover, sequence variants, and regulatory networks

Mature microRNAs are often comparatively stable because Argonaute protects them from degradation. Their abundance nevertheless reflects several processes: transcription of their genes, processing efficiency, loading into Argonaute, degradation, and dilution as cells divide. A stable microRNA may persist after its gene stops being transcribed, whereas a rapidly dividing population can lose it through dilution even without accelerated molecular decay. Measurements of primary, precursor, and mature RNA can therefore yield different pictures of regulation.

MicroRNA sequences are not always represented by a single exact molecular species. Variants called isomiRs arise through alternative cleavage, terminal trimming, nucleotide addition, and sometimes RNA editing. A shift at the 5′ end changes the seed and can redirect the target repertoire; a change confined to the 3′ end may instead influence stability, Argonaute association, or supplementary pairing. Some sequence heterogeneity is biologically regulated, while some apparent variation results from technical errors. Establishing function requires more than detecting a rare sequence variant.

Targets can themselves influence microRNA lifetime. In target-directed microRNA degradation, particular target RNAs pair extensively with a microRNA in a configuration that destabilizes the silencing complex rather than simply undergoing conventional repression. In a well-characterized mammalian pathway, the ZSWIM8 ubiquitin ligase promotes degradation of Argonaute, exposing the microRNA to destruction. The long noncoding RNA CYRANO provides a prominent example of an RNA that promotes degradation of miR-7. These mechanisms make the regulatory relationship bidirectional: targets can determine the abundance of their regulators.

MicroRNAs also participate in feedback and feed-forward circuits with transcription factors. A transcription factor may activate both a protein-coding gene and a microRNA that restrains that gene’s output, producing an incoherent feed-forward circuit. In other arrangements, a microRNA and a transcription factor repress one another, potentially sharpening transitions between cellular states. Such circuit descriptions are useful, but their actual behavior depends on quantitative parameters rather than diagram topology alone.

The proposal that RNAs can regulate one another by competing for shared microRNAs is often called the competing endogenous RNA hypothesis. Competition is physically possible, but its physiological importance depends on relative concentrations, binding-site affinity, and the total pool of accessible sites. Adding a small number of sites to a cell containing a much larger effective target pool may produce little measurable redistribution. Large experimental overexpression can therefore demonstrate competition without showing that the same effect occurs at normal abundance.

These quantitative considerations help explain why a microRNA can have a strong effect in one tissue and little detectable effect in another. The relevant variable is not simply whether the microRNA exists, but whether its concentration and target environment permit biologically consequential occupancy of specific sites.

Plant pathways and evolutionary diversity

Plant microRNAs share the basic principle of Argonaute-guided recognition with animal microRNAs, but their maturation and target interactions differ substantially. In the model plant Arabidopsis thaliana, the Dicer-like protein DCL1 performs successive nuclear processing steps, assisted by proteins including HYL1 and SERRATE. Processing commonly occurs in nuclear structures called dicing bodies. Thus, the canonical plant pathway does not divide precursor cleavage between Drosha in the nucleus and Dicer in the cytoplasm.

After processing, the methyltransferase HEN1 modifies the terminal ribose of the small-RNA duplex by adding a 2′-O-methyl group. This modification protects plant microRNAs against destabilizing terminal modifications and degradation. The mature guide is loaded into an Argonaute protein, frequently AGO1. The identity of the guide’s first nucleotide contributes to sorting among plant Argonaute proteins, although loading specificity also depends on other features of the RNA and protein.

Plant microRNAs usually pair more extensively with their targets than animal microRNAs do. Many therefore guide endonucleolytic cleavage of target messenger RNAs. Translational repression also occurs and can make an important contribution, so the contrast is not an absolute division between plant cleavage and animal translational inhibition. Plant target sites frequently occur within coding regions, whereas functional animal sites are especially common in 3′ untranslated regions.

Conserved plant microRNA families regulate major developmental and physiological processes. The miR156 family represses SQUAMOSA PROMOTER BINDING PROTEIN-LIKE transcription factors and contributes to age-dependent developmental transitions. The miR172 family regulates APETALA2-like factors involved in developmental phase changes and floral patterning. Other families connect growth to environmental conditions: miR399 participates in phosphate homeostasis, while miR395 regulates components of sulfur assimilation. These examples show how small RNAs can coordinate nutrient status with developmental programs.

MicroRNA-like regulatory systems also occur outside animals and land plants, but the evolutionary relationships among these systems require care. Short hairpin-derived RNAs have arisen in different eukaryotic lineages, and some fungi possess small regulatory RNAs described as microRNA-like RNAs. Shared participation of ancient RNA-silencing proteins does not establish that all present-day microRNA pathways descend unchanged from a single ancestral system.

Within lineages, new microRNA genes can emerge from genomic sequences capable of forming appropriate hairpins, including duplicated sequences and some transposable elements. Many young candidates are weakly expressed or not retained over long evolutionary periods. Deeply conserved families, by contrast, often show strong sequence constraint, particularly in their seed regions. Evolution can preserve an ancestral microRNA while changing its expression pattern or target repertoire, allowing an old regulatory molecule to acquire new biological roles.

Roles in development and normal physiology

MicroRNAs influence development by controlling the timing, magnitude, and spatial distribution of protein production. The original lin-4 and let-7 examples illustrate temporal regulation: particular cell behaviors must occur at the correct developmental stage, and persistent expression of an earlier regulatory program can disrupt later events. In mammals, let-7 family members are associated with differentiation and restrain targets involved in growth and progenitor-cell programs. Their interaction with LIN28 connects RNA processing to changes in developmental state.

Cell-type-specific expression provides another major organizing principle. miR-1 and miR-133 are prominent in muscle regulatory networks, while miR-124 is abundant in many neurons. These molecules act alongside transcription factors rather than replacing them. miR-124, for example, helps suppress non-neuronal gene-expression programs and interacts with pathways governing alternative RNA processing. Such coordinated regulation can stabilize cellular identity by reducing expression of genes inappropriate for a particular lineage.

The immune system provides well-studied examples of context-dependent microRNA action. miR-155 is induced in several activated immune-cell populations and regulates components of immune signaling and differentiation. miR-146a participates in negative regulatory circuits downstream of inflammatory signaling, including regulation of IRAK1 and TRAF6. These molecules do not have a single invariant “pro-inflammatory” or “anti-inflammatory” function: their effects depend on cell type, stimulus, timing, and which targets are expressed.

MicroRNAs also contribute to metabolic regulation. miR-122 is highly enriched in the liver and influences networks involved in lipid metabolism and hepatocyte biology. miR-33 family members occur within introns of sterol regulatory element-binding protein genes and regulate targets involved in lipid transport and related processes. This genomic arrangement links transcriptional programs controlling lipid synthesis with post-transcriptional regulation of other parts of lipid homeostasis. The specific family members present and their physiological contributions differ among species.

Some of the clearest evidence for physiological importance comes from mutations or targeted deletion. Removing a microRNA can produce an obvious developmental defect, but many individual deletions have subtle effects under standard laboratory conditions. Such findings do not necessarily imply absence of function: closely related family members may compensate, the relevant environmental challenge may be missing, or the phenotype may involve quantitative traits rather than gross anatomy. Conversely, invoking “robustness” without identifying a measurable phenotype does not establish a particular function.

MicroRNAs are therefore best understood as components of larger regulatory systems. They can tune a threshold, reinforce a cell state, restrain inappropriate expression, or coordinate several pathway components. The importance of a particular microRNA depends on experimentally demonstrated relationships among its targets, the affected biological process, and the phenotype.

Human disease and pathological mechanisms

MicroRNA abnormalities can arise from changes in gene copy number, transcription, processing, mature RNA sequence, or target-site sequence. Disease can also alter microRNA expression secondarily by changing tissue composition or cellular state. These possibilities must be distinguished: an association between a microRNA and a disease does not establish that the microRNA initiates the disease or that changing its abundance would correct it.

In cancer, microRNAs can influence proliferation, apoptosis, differentiation, and interactions between tumor cells and surrounding tissue. A landmark example is the miR-15a/miR-16-1 cluster at chromosome region 13q14, which is frequently deleted or downregulated in chronic lymphocytic leukemia. Experimental work linked these microRNAs to regulation of the anti-apoptotic protein BCL2. The miR-17–92 cluster provides a different pattern: increased expression can promote tumor-associated processes in particular cellular and genetic contexts. Labels such as “tumor-suppressive microRNA” and “oncogenic microRNA” describe demonstrated contexts, not universal properties of a sequence.

Rare inherited disorders provide more direct evidence of causation. Mutations in the seed region of miR-96 cause an inherited form of progressive hearing loss. Because the seed determines much of target recognition, such mutations can both weaken normal interactions and introduce new ones. Mutations affecting precursor processing can also be pathogenic; variants in MIR184 have been associated with inherited ocular disorders. These cases illustrate that disease mechanisms can involve either altered target selection or altered production of the mature regulator.

The processing machinery is another point of vulnerability. Pathogenic variants in DICER1 underlie a tumor-predisposition syndrome associated with several characteristic neoplasms. Some tumor-associated DICER1 changes disproportionately affect processing of particular microRNA arms rather than simply abolishing all microRNA production. This distinction matters because a partial or biased processing defect can generate a different regulatory state from complete loss of an essential RNA-processing enzyme.

MicroRNAs also interact with infectious agents. Some viruses, particularly members of the herpesvirus family, encode their own microRNAs and use them to regulate viral or host transcripts. Conversely, host microRNAs can restrict or support viral replication. Hepatitis C virus provides an unusual, well-established example: liver-enriched miR-122 binds near the viral genome’s 5′ end and supports viral RNA accumulation. This interaction differs from the conventional repression of a cellular messenger RNA.

Interpretation of disease-associated microRNA profiles requires attention to cellular composition. A tumor sample contains malignant cells, immune cells, stromal cells, and blood-derived material; a change in bulk microRNA abundance may reflect changing proportions of these populations. Similarly, altered circulating microRNAs can result from tissue injury or blood-cell disruption rather than a specific disease program. Mechanistic conclusions require evidence that connects the relevant microRNA, its targets, and the affected cells.

Experimental identification and functional validation

Reliable identification begins with distinguishing genuine microRNAs from other short RNA fragments. High-throughput sequencing of size-selected RNA libraries reveals the sequences and relative abundance of small RNAs, but short reads alone are insufficient for annotation. A convincing microRNA locus should have a plausible precursor structure and a processing pattern consistent with production of a defined duplex. Fragments of abundant transfer or ribosomal RNAs can otherwise be mistaken for new microRNAs.

Databases serve different purposes. miRBase has provided a widely used repository and naming framework, while MirGeneDB emphasizes manually curated animal microRNA annotations based on evolutionary and processing criteria. Counts differ among resources because the underlying standards differ, not simply because one resource has sampled more molecules. A reported number of human microRNAs should therefore be interpreted in relation to the database, release, and annotation criteria used.

Expression can be measured by northern blotting, quantitative reverse-transcription polymerase chain reaction, hybridization-based assays, or small-RNA sequencing. Northern blotting can distinguish precursor and mature species by size, while PCR-based assays provide sensitive measurement of selected sequences. Sequencing measures many small RNAs simultaneously but introduces biases through adapter ligation, amplification, and mapping. Closely related family members and isomiRs create additional challenges because assays may cross-detect sequences that differ by only one nucleotide.

Target identification requires complementary approaches. Computational algorithms use seed pairing, sequence conservation, site context, and other features to prioritize candidate interactions. Argonaute crosslinking and immunoprecipitation methods identify RNA regions occupied by the silencing machinery, while some related methods capture ligated microRNA–target chimeras. Occupancy strengthens the case for direct contact, but it does not by itself show that the interaction changes gene expression or produces a physiological effect.

Reporter assays test whether a candidate target region confers microRNA-dependent regulation; mutation of the proposed binding site provides an important specificity control. Endogenous validation is stronger when microRNA perturbation changes the native target’s RNA or protein abundance and when editing the native target site alters the response. Genome editing can therefore help distinguish direct regulation from downstream effects. Rescue experiments can further connect a particular target to a phenotype, although a rescue construct lacking normal regulation may create nonphysiological expression.

Gain- and loss-of-function experiments each have limitations. Excess synthetic microRNA can load Argonaute at unnatural concentrations and repress normally irrelevant targets. Antisense inhibitors can have incomplete or sequence-dependent effects, while deletion of a genomic locus may disturb neighboring regulatory elements or a host gene. Strong conclusions usually combine several methods, appropriate controls, and measurements made at biologically relevant expression levels rather than relying on one assay.

Biomarkers, therapeutic strategies, and translational limits

MicroRNAs can be detected in blood and other body fluids because extracellular RNA is protected by proteins, lipid-containing particles, or membrane-bound vesicles. This relative stability has encouraged investigation of circulating microRNAs as biomarkers of tissue injury, malignancy, and other conditions. However, extracellular presence does not establish active secretion or intercellular signaling. Demonstrating a signaling role requires evidence that recipient cells take up a biologically sufficient quantity and that the transferred RNA produces sequence-specific effects.

Biomarker development faces substantial preanalytical challenges. Blood collection methods, processing delays, platelet contamination, hemolysis, RNA extraction, and normalization can all alter measurements. For example, disruption of red blood cells can elevate microRNAs abundant in those cells, generating an apparent disease association unrelated to the proposed tissue source. Serum and plasma are not interchangeable materials, and differences between study groups must not be confounded with differences in collection or processing.

A clinically useful biomarker must perform beyond the dataset in which it was discovered. Relevant requirements include reproducibility, independent validation, defined sensitivity and specificity, and evidence that the measurement adds information to existing clinical variables. A statistically significant difference between small groups is not equivalent to a validated diagnostic test. Profiles involving many microRNAs also require safeguards against overfitting, particularly when the number of measured features greatly exceeds the number of participants.

Therapeutic approaches broadly aim either to inhibit an unwanted microRNA or restore a deficient one. Antisense oligonucleotides can bind mature microRNAs and prevent target recognition. Chemical modifications can improve affinity and resistance to degradation. Replacement strategies use microRNA mimics or expression constructs to increase regulatory activity. Target protectors offer a narrower approach by blocking one microRNA-binding site on a selected transcript, rather than inhibiting all functions of the microRNA.

Human trials have established both feasibility and significant obstacles. Miravirsen, an antisense inhibitor of miR-122, reduced hepatitis C viral RNA in a phase 2a study, supporting the principle that a host microRNA could be pharmacologically targeted. MRX34, a liposomal miR-34a mimic investigated in cancer, entered clinical testing but its trial was terminated after serious immune-mediated adverse events. These examples demonstrate that sequence-based rationale and preclinical activity do not guarantee acceptable safety or clinical benefit.

Delivery remains a central limitation. An agent must reach the relevant tissue, enter the appropriate cells, escape intracellular compartments when necessary, and engage its RNA target without excessive unintended effects. MicroRNAs naturally regulate multiple transcripts, so broad biological consequences may reflect their normal mechanism rather than accidental sequence mismatch. Translational evaluation must therefore distinguish delivery-related toxicity, immune responses to the formulation or RNA, unintended hybridization, and consequences of changing the intended microRNA network.