PerspectiveGENE REGULATION

Breakers and blockers—miRNAs at work

See allHide authors and affiliations

Science  24 Jul 2015:
Vol. 349, Issue 6246, pp. 380-382
DOI: 10.1126/science.1260969

MicroRNAs (miRNAs) are small, ~22-nucleotide-long noncoding RNAs. They silence the expression of messenger RNAs (mRNAs) containing complementary sequences (1). The human genome encodes ~1500 miRNAs, each with the potential to bind hundreds of different mRNAs (1). miRNAs regulate many biological processes, and the dysregulation of their expression is linked to various human diseases, including cancer (1). To exert their repressive function, miRNAs associate with the Argonaute family of proteins (AGOs) to form the core of miRNA-induced silencing complexes (miRISCs) (1) (see the figure). In animals, miRISCs silence mRNA expression at two levels, by preventing protein production (translation) and inducing mRNA degradation. Over the past decade, progress has been made in our understanding of the mechanism by which miRISCs induce mRNA degradation, but the question of how miRISCs repress translation remains elusive.

miRISCs were originally thought to repress protein synthesis without detectable changes in mRNA abundance. However, work from many laboratories has established that miRISCs degrade their mRNA targets, thereby reducing protein synthesis, at least in part, indirectly (13). Nevertheless, the possibility that a fraction of targets were regulated only at the translational level without changes in mRNA levels could not be ruled out, raising concerns that an unknown fraction of targets would be overlooked if only mRNA levels were monitored (2, 3). Mechanistically, mRNA degradation was thought to represent a secondary effect of the inhibition of translation, with the primary effect of the miRISC being interference with protein synthesis.

These issues have been in part addressed with the advent of ribosome profiling, which maps the position of translating ribosomes on mRNAs at a global level (2, 3). Ribosome profiling analyses combined with simultaneous detection of mRNA amounts did not yield evidence of a large number of targets that are regulated exclusively at the translational level. These studies revealed that mRNA degradation is the dominant effect of miRNA-mediated regulation at steady state, accounting for most (66 to 90%) of the repression observed in cultured mammalian cells (2, 3).

The mechanism by which miRNAs degrade their mRNA targets occurs through a cellular pathway—the 5′-to-3′ mRNA decay pathway—which also degrades bulk mRNA (1). miRISCs recruit the cellular machinery involved in this pathway to the mRNA target, thereby accelerating its degradation. mRNA degradation is initiated by deadenylation, the removal of the polyadenylate tail at the 3′ end of the mRNA. Deadenylation is catalyzed by the consecutive action of two deadenylase complexes: the PAN2-PAN3 complex and the CCR4-NOT complex (see the figure). Following deadenylation, the cap structure that protects the mRNA 5′ end is removed by the decapping enzyme DCP2 in complex with additional cofactors. Decapping makes the mRNA 5′ end accessible to the major cytoplasmic exonuclease XRN1 that degrades the mRNA from the 5′ end (1, 46).

The Argonaute proteins of the miRISC recruit the two deadenylase complexes to the mRNA. They do so indirectly, through their interaction with a protein of the GW182 family (1, 46). GW182 proteins (also known as TNRC6 proteins in vertebrates) are conserved in animals. These proteins are mainly unstructured and contain multiple glycine (G) tryptophan (W) repeats. They use these repeats to bind to Argonaute proteins and subunits of the PAN2-PAN3 and the CCR4-NOT deadenylase complexes, thereby bridging their interaction (46).

Structures of human Argonaute-2 (AGO2) bound to miRNAs (7), as well as of AGO2 and subunits of the deadenylase complexes (PAN3 and NOT9) bound to tryptophan residues (810), have revealed how miRISCs recognize target mRNA, and subsequently how GW182 proteins bridge the interaction between Argonaute proteins and the deadenylase complexes. Thus, we can now describe in molecular terms a chain of direct interactions leading all the way from miRNA target recognition to mRNA deadenylation, decapping, 5′-to-3′ mRNA degradation, and eventually translational repression (710). Additionally, these structural studies have definitively shown that miRISC actively recruits mRNA degradation factors and thus that mRNA destabilization is a primary effect of silencing.

“Pure” translational repression (that cannot be explained by mRNA degradation) is thought to account for 6 to 26% of the repression of each endogenous target in mammalian cells (2, 3). However, the precise translation-repressive mechanism remains poorly understood.

Ribosomal profiling has been instrumental in simplifying our understanding of silencing by excluding models invoking inhibition of translation elongation, as well as models involving protein degradation. These and additional studies have implicated initiation and the eukaryotic translation initiation factor eIF4A as the target for miRISC inhibition (2, 3, 1113). eIF4A has two paralogs involved in translation in vertebrates: eIF4A1 and eIF4A2 (1113). eIF4A proteins are RNA helicases that unwind secondary structures within the mRNA 5′ untranslated regions (UTRs), allowing the small ribosomal subunit, in association with initiation factors, to scan the 5′ UTR in search of the AUG start codon. Thus, interfering with eIF4A function is likely to reduce or inhibit the initiation of protein synthesis.

MiRNA-mediated silencing.

Animal miRNAs bound to an Argonaute protein (AGO) recognize their mRNA targets by base-pairing to partially complementary binding sites. AGOs interact with GW182 proteins, which recruit deadenylase complexes. The decapping complex is recruited, at least in part, through interactions with the CCR4-NOT complex, which are mediated by the DDX6 protein. In addition, miRNAs repress translation, possibly through DDX6 and/or interference by miRISCs with the function of the initiation factor eIF4A. The decapping complex contains multiple components, which are shown as a single oval labeled EDCs (enhancers of decapping). The CCR4-NOT complex contains at least eight subunits, of which only four are shown. The cap structure is shown as a black dot. ORF: open reading frame; UTR: untranslated region; CAF1: chromatin assembly factor 1. Cyan circles in GW182 represent W-containing motifs.

However, there is debate regarding how this interference is achieved. In one model, miRISC recruits and locks eIF4A2 on the mRNA 5′ UTR, forming a “roadblock” for scanning ribosomes (13). A second model proposes that miRISC dislodges eIF4A1 and eIF4A2 from the silenced mRNA (11, 12). Both models imply that miRNAs can only repress mRNAs that are translated via a scanning and eIF4A-dependent mechanism. Experiments to test this prediction have so far yielded conflicting results.

In the “roadblock” model, eIF4A2 has been proposed to be recruited to miRNA targets through an interaction with NOT1 (13). However, other studies have indicated that NOT1 interacts with DDX6 rather than eIF4A2 (9, 10, 14). Because DDX6 represses translation and its depletion inhibits silencing in human cells (9, 10, 14, 15), its recruitment to miRNA targets could explain how translational repression by miRISC is achieved.

It is becoming increasingly clear that the CCR4-NOT complex is a conserved and central downstream effector of silencing that can elicit all of the effects that the miRISC has on its targets, namely translational repression, deadenylation, decapping, and mRNA decay (4, 5, 9, 10). A possible contribution by additional silencing mechanisms, independent of the CCR4-NOT complex (12), remains an important topic for future work.

A complicating factor in the study of silencing mechanisms that has not yet been addressed is redundancy. Redundancy is manifested at two levels. The first level is due to gene-duplication events that generated four Argonaute and three GW182 paralogs in most vertebrates. Similarly, there are two paralogs for each of the two catalytic subunits of the CCR4-NOT complex, resulting in the assembly of at least four distinct complexes. Another level of redundancy is observed in the network of interactions between the factors involved. For example, deadenylase complexes interact both with each other, and CCR4-NOT interacts with decapping factors independently of GW182 and DDX6. Redundancy offers alternative ways to assemble silencing complexes, which might ultimately differ in their functional outcomes, potentially resulting in distinct molecular mechanisms, which might become dominant in a context-dependent manner.

References

View Abstract

Stay Connected to Science


Editor's Blog

Navigate This Article