2013-Diversifying MicroRNA Sequence and Function

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 14

REVIEWS

Diversifying microRNA sequence


and function
Stefan L. Ameres1 and Phillip D. Zamore2
Abstract | MicroRNAs (miRNAs) regulate the expression of most genes in animals, but we are
only now beginning to understand how they are generated, assembled into functional
complexes and destroyed. Various mechanisms have now been identified that regulate
miRNA stability and that diversify miRNA sequences to create distinct isoforms.
The production of different isoforms of individual miRNAs in specific cells and tissues may
have broader implications for miRNA-mediated gene expression control. Rigorously testing
the many discrepant models for how miRNAs function using quantitative biochemical
measurements made in vivo and in vitro remains a major challenge for the future.

Argonaute For more than a decade, miRNAs have enthralled the animals8. In the plant kingdom, the miRNA pathway
(AGO). Proteins that are guided biological community. Almost everything about micro- presumably evolved before multicellularity, because the
to mRNA targets by small RNAs (miRNAs) is small. At ~22 nucleotides long, they unicellular alga Chlamydomonas reinhardtii produces
silencing RNAs. AGO proteins are among the shortest functional eukaryotic RNAs. They miRNAs that are similar to those of higher plants9.
can also serve as scaffolds to
bind secondary silencing
were discovered in a diminutive organism, Caenorhabditis For several model animals and plants, as well as
factors such as the GW‑repeat- elegans, and their physiological relevance in regulating humans, we now know the majority of miRNAs and these
containing protein GW182. plant and animal gene expression was underappreciated are inferred to be functional on the basis of their evolu-
for nearly a decade1,2. Finally, miRNA­s repress most of tionary conservation. But we are only beginning to under-
Primary miRNAs
the genes they regulate by just a tiny amount. Yet, col- stand the divers mechanisms that contribute to miRNA
(pri-miRNAs). Polyadenylated,
7‑methylguanosine-capped
lectively miRNAs affect nearly all cellular pathways, from production and assemble them into complexes with pro-
RNA polymerase II transcripts developmen­t to oncogenesis. teins. For example, although it is established that miRNAs
containing a stem–loop The primary repository for miRNA sequences and direct Argonaute (AGO) proteins to repress mRNAs, the
structure that serves as a annotations, miRBase, debuted with just 218 miRNA loci3. mechanism of repression remains intensely debated.
substrate for the RNase III
enzyme Drosha in animals and
Since then, the application of high-throughput sequencing Here, we review the mechanisms that diversify miRNA
Dicer-like 1 (DCL1) in plants. to miRNA discovery in >193 different plant and anima­l sequence and function. We briefly outline the pathways
They are processed to liberate species has swelled that number to >21,264 loci that that generate miRNAs and regulate their production,
a precursor miRNA and two produce >25,141 mature miRNAs. The human genome activity and longevity. We describe the mechanisms by
unstable, single-stranded
alone comprises >1,500 hairpin structures that produce which miRNA sequences and functions can be altered, as
by‑products.
detectable small RNAs, although the authenticity and well as the processes that destroy rogue and old miRNA­s.
functional importance of many of these miRNA candi- Finally, we consider the evidence for different models
1
Institute of Molecular dates remain to be established4. Selective pressure during of how miRNAs function and present a quantitativ­e
Biotechnology of the Austrian
Academy of Sciences (IMBA),
evolution seems to have maintained the pairing between framewo­rk for evaluating these models.
Dr. Bohr-Gasse 3, 1030 miRNAs and more than half of all human protein-coding
Vienna, Austria. genes5, suggesting that these tiny riboregulator­s control Making miRNAs
2
Howard Hughes Medical the expression of most human proteins6. Most miRNAs derive from longer, intramolecularly
Institute and University of
The absence of miRNAs in fungi and the distinct double-stranded RNAs (dsRNAs; termed primary miRNA­s
Massachusetts Medical
School, 55 Lake Avenue sequences, precursor structures and mechanisms of pro- (pri-miRNAs)) that are sequentially cleaved into shorter
North, Worcester, duction for plant and animal miRNAs suggest that the intermediates by specialized ribonuclease III (RNase III)
Massachusetts 01605, USA. miRNA pathway evolved at least twice7. Small silencing enzymes that partner with specific dsRNA‑binding
e‑mails: stefan.ameres@ RNAs, including miRNAs, are present in the earlies­t protein­s 10,11 (FIG.  1) . Both transcriptional and post-
imba.oeaw.ac.at; phillip.
zamore@umassmed.edu
diverging extant lineage of animal life, the sponge transcriptiona­l mechanisms regulate miRNA biogenesis
doi:10.1038/nrm3611 Amphimedon queenslandica, suggesting that the emer- (for a detailed discussion, see REFS 12,13). A few miRNA­s
Published online 26 June 2013 gence of miRNAs herald the dawn of multicellular are produced by alternative pathways that replace

NATURE REVIEWS | MOLECULAR CELL BIOLOGY VOLUME 14 | AUGUST 2013 | 475

© 2013 Macmillan Publishers Limited. All rights reserved


REVIEWS

Standard pathway Alternative pathway standard biogenesis steps with RNA processing control
RNA Pol II from pre-mRNA splicing or RNA degradation pathways
m7Gppp
(for further details, see REF. 14).

miRNAs are transcribed by RNA polymerase II. RNA


mRNA splicing poly­merase II (Pol II) produces long pri-miRNAs from
independent genomic transcription units or from the
introns of protein-coding genes15–22. For miRNAs that
Drosha
arise from introns, splicing is not required for their pro-
Pri-miRNA
AAAA(n) Branched Mirtron
duction23, and the co‑transcriptional processing of pri-
m7Gppp miRNA into precursor miRNAs (pre-miRNAs) does not
Pasha or
Pri-miRNA DGCR8 Lariat debranching affect the splicing of the host pre-mRNA24,25.
Nucleus cropping
Drosha crops pri-miRNAs into shorter pre-miRNAs.
In animals, the RNase III enzyme Drosha converts pri-
Cytoplasm miRNA­s into pre-miRNAs, which are ~60 nucleotide
Exportin 5
stem–loop structures. Drosha excises one or more pre-
miRNAs from the pri-miRNA26–29. Thus, pri-miRNAs can
produce several pre-miRNAs. Polycistronic miRNAs
Pre-miR-451
Loqs or allow a single promoter to drive co‑expression of multi-
TRBP ple miRNAs and, in some cases, mediate the coordinated
Pre-miRNA Dicer expression of target factors in a single pathway 10.
Drosha acts as a component of a larger complex, the
microprocessor. This nuclear complex seems to exist
in at least two forms, a ~600 kDa complex of unknown
H
O
2′

miRNA–miRNA* function and a smaller heterodimer comprising Drosha


AGO2
duplex and its dsRNA-binding protein, named DGCR8 in mam-
mals and Pasha in other animals28–31. The collaboration
H
O
2′

Pre-miRNA of Pasha with Drosha defines a general paradigm in


HSC70 and HSP90 cleavage and small RNA biogenesis: dsRNA-binding proteins part-
maturation
AGO ner with RNase III enzymes to restrict the substrates
Pre-RISC
they process, increase their affinity for a substrate or
improve cleavage site accuracy. The subcellular distribu-
miRNA* tion of miRNA-producing enzymes and their substrates
implies that animal pri-miRNA­s are cropped in the
Mature RISC RISC nucleus, whereas pre-miRNAs are processed in the cyto-
plasm26,32–36. The nuclear transport receptor exporti­n 5
(known as Ranbp21 (accession number CG12234)
in flies) recognizes the ends and the stem of the pre-
m7Gppp miRNA37 and exports the pre-miRNA from the nucleus
(n)AAAA ORF
to the cytoplasm via the nuclear pore38–41.
Figure 1 | MicroRNA biogenesis.  In the standard microRNA (miRNA) biogenesis
Nature Reviews | Molecular Cell Biology
pathway, primary miRNA (pri-miRNA) transcripts are processed by Drosha in the Dicing makes miRNA–miRNA* duplexes. In the cytoplas­m,
nucleus and by Dicer in the cytoplasm. The pri-miRNA, which is transcribed by RNA a second RNase III enzyme, Dicer, liberates a ~22 nucleo-
polymerase II (Pol II), begins with a 7‑methylguanosine cap (m7Gppp) and ends with a
tide miRNA–miRNA* duplex from the pre-miRNA42–45.
3ʹ poly(A) tail. The pri-miRNA contains a stem–loop structure that is cleaved in the
nucleus by the endonuclease Drosha together with its double-stranded RNA In flies, Dicer‑1 cleaves pre-miRNAs, whereas Dicer‑2
(dsRNA)-binding protein parner DGCR8 (in mammals) or Pasha (in flies). The resulting generates siRNAs46. Like Drosha, Dicer‑1 recognizes
precursor miRNA (pre-miRNA) is exported from the nucleus by exportin 5 and then defined RNA structures and then cleaves at a fixed dis-
further cleaved by the endonuclease Dicer together with its dsRNA-binding partner tance away from the base of the pre-miRNA stem, cutting
TRBP (transactivation-response RNA-binding protein; in mammals) or Loquacious off the loop to produce ~22 nucleotide mature miRNA–
(Loqs; in flies) to liberate a miRNA–miRNA* duplex. Supported by the HSC70–HSP90 miRNA* duplexes47,48. Like cropping of pri‑miRNA­s,
chaperone machinery, this duplex is loaded into an Argonaute (AGO) protein as a ‘dicin­g’ of pre-miRNAs can also be bypassed. For exam-
dsRNA. Subsequent maturation steps expel the miRNA*, producing a mature ple, the hairpin of pre-miR-451 in zebrafish and mice is
RNA-induced silencing complex (RISC). Alternative pathways (shown on the right) too short to be recognized by Dicer; instead, it is directly
typically replace individual steps of miRNA precursor processing. Pri-miRNA cropping loaded into the RNA-induced silencing complex (RISC) for
can be replaced with nucleases from other cellular pathways, including the general
further processing by AGO2 into a mature miRNA49–51.
RNA degradation machinery or pre-mRNA splicing factors. In such instances, the
pri-miRNA is generated from a branched mirtron structure that undergoes lariat
debranching. In the specific case of pre-miR‑451, the pre-miRNA escapes Dicer Partner proteins shape Dicer function. Continuing the
processing after nuclear export and is instead directly loaded into the AGO2 protein, theme of RNase III enzymes that partner with dsRNA-
which triggers its maturation into a single-stranded miRNA. 2ʹ OH, 2ʹ hydroxyl group; binding proteins, Drosophila melanogaster Dicer‑1 part-
HSP, heat shock protein; ORF, open reading frame. ners with two isoforms of the dsRNA-binding protein

476 | AUGUST 2013 | VOLUME 14 www.nature.com/reviews/molcellbio

© 2013 Macmillan Publishers Limited. All rights reserved


REVIEWS

Flies and mammals Flies Plants

miRNA siRNA miRNA


– Hen1 + Hen1 – HEN1 + HEN1
and siRNA

HESO1 HEN1
AGO1 AGO2 AGO2
UUUUU 2′O-CH3
2′O-CH3
Highly complementary Tailing HEN1
target mRNA miRNA* miRNA*
Hen1
2′ OH 2′ O-CH3
NNN ? ? 3′-to-5′
exonuclease
Tailing 3′-to-5′exonucleolytic
trimming AGO
2′O-CH3 ? 2′O-CH3
Small RNA turnover siRNA stabilization Small RNA turnover Small RNA stabilization
Figure 2 | Small RNA turnover by tailing and trimming.  In the tailing-and-trimming pathway, the presence of a highly
complementary target RNA triggers both tailing and trimming of microRNAs (miRNAs),Natureultimately
Reviews | decreasing theirBiology
Molecular Cell
abundance in flies and mammals (left panel). The identity of the exonuclease is not known (denoted by the question
mark in the figure). In flies, Argonaute 2 (AGO2)‑bound siRNAs are protected from tailing and trimming because the
methyltransferase Hen1 (Hua enhancer 1) methylates siRNAs at the 2ʹ position of the 3ʹ terminal nucleotide. In the
absence of Hen1, siRNAs are subjected to target RNA-directed tailing and trimming (middle panel). Unlike fly Hen1,
plant HEN1 methylates both siRNA and miRNA duplexes before they are loaded into AGO. As in flies, loss of HEN1 in
plants prompts small RNA tailing, which is mediated by the plant terminal nucleotidyltransferase (TNTase) HEN1
SUPPRESSOR 1 (HESO1), ultimately resulting in 3ʹ-to‑5ʹ exonucleolytic decay (right panel). The normal triggers for
small RNA decay in plants are unknown.

Ribonuclease III
(RNAse III). Double-stranded
RNA-specific endoribonuclease Loquacious (Loqs), termed Loqs‑PA and Loqs‑PB. but detectable decrease in the expression of an mRNA.
that generates products with These are required for efficient miRNA processing in Seed matches can occur in any region of an mRNA but
two nucleotide 3ʹ overhangs, a flies because they increase the affinity of Dicer‑1 for pre- are more likely to decrease mRNA expression when they are
5ʹ phosphate and a 3ʹ hydroxyl
group.
miRNAs52–55. Conversely, the Dicer‑2 partner protein in the 3ʹ untranslated region (3ʹ UTR)64–67. Because the
R2D2 prevents Dicer‑2 from processing pre-miRNAs, region used to create the seed is so short, more than half
dsRNA-binding protein restricting it to long dsRNA substrates56. of all protein-coding genes in mammals are regulated by
Proteins containing Alternative partner proteins refine the substrate spec- miRNAs, and thousands of other mRNAs seem to have
double-stranded RNA-binding
ificity of individual Dicer proteins, and this may explain experienced negative selection to avoid seed matches with
domains, which are ~70
amino acid motifs that bind how organisms such as worms and mammals produce miRNAs that are present in the same cells5,68–70.
to RNA helices via their miRNAs and siRNAs with only a single enzyme.
2ʹ hydroxyl group and Consistent with this, mammals have two Dicer Target RNAs alter miRNA stability
phosphate backbone. partners, transactivation-response RNA-binding pro- Binding of the AGO–miRNA complex to an mRNA
Precursor miRNAs
tein (TRBP) and protein kinase R‑activating protein not only regulates gene expression (discussed in detail
(pre-miRNAs). Stem–loop RNAs (PACT)57–59. below) but also alters the stability of the miRNA itself. For
comprising a single-stranded In addition to their role in defining Dicer substrate example, binding of complementary target RNAs in flies
loop that connects two partially specificity, Dicer partner proteins can alter the position and mammals can direct miRNA destruction. miRNA
complementary sequences.
of Dicer-mediated cleavage within a pre-miRNA55,60. For decay elicited by target binding was first reported for
These sequences pair to form a
predominantly double-stranded example, TRBP, but not PACT, changes the site at which antagomirs, which are chemically modified, synthetic
stem. Pre-miRNAs typically mammalian Dicer cleaves a few pre-miRNAs, such as oligonucleotides used to inhibit miRNA function in cells
have a 5ʹ phosphate and a pre-miR‑132. In flies, Loqs‑PB changes the length and and in vivo71. We now know that antagomirs, as well as
two nucleotide 3ʹ overhang, seed sequence of the miRNAs generated by Dicer‑1 from highly expressed RNAs containing sites with extensive
allowing them to serve as
substrates for Dicer.
a small subset of pre-miRNAs. complementarity to a miRNA, trigger miRNA degra-
dation (FIG. 2). This pathway involves both the addition
Drosha A seed for target recognition. After the sequential pro- of adenosine or uracil to the miRNA (‘tailing’) and the
The nuclear RNase III cessing of miRNA precursors, one of the two strands 3ʹ-to‑5ʹ exonucleolytic resection of the miRNA 3ʹ end
endonuclease in animals that
of the miRNA duplex guides AGO proteins to comple- (‘trimming’)72–74. Target RNA-directed tailing and trim-
cleaves the base of a stem–loop
structure contained in primary mentary mRNA sequences to repress their expression ming of miRNAs require greater complementarity than
microRNAs (pri-miRNAs) to (see REFS 61,62 for detailed reviews of AGO loading). The is typically found between miRNAs and their targets72,74.
produce a precursor miRNA. major determinant for AGO binding to its target mRNA is This may be because the structural rearrangements
Collaborates with mammalian a 6–8 nucleotide domain at the 5ʹ end of the miRNA. associated with target binding provide the tailing and
DGCR8 (or Pasha in other
animals), which is its
AGO associates with this region to create the ‘seed’ (for trimming enzymes access to the miRNA 3ʹ end, which
double-stranded RNA-binding details, see REF. 63). Sequences that are complementary would otherwise reside in the AGO PAZ domain75–79.
protein partner. to the seed (‘seed matches’) suffice to trigger a modest Alternatively, all miRNAs may be susceptible to tailing

NATURE REVIEWS | MOLECULAR CELL BIOLOGY VOLUME 14 | AUGUST 2013 | 477

© 2013 Macmillan Publishers Limited. All rights reserved


REVIEWS

Dicer and trimming when bound to target RNAs, but the that they inappropriately accumulate in AGO1 com-
An RNase III endonuclease that slower dissociation of miRNAs from high-affinity sites plexes. As a consequence, they would not acquire a
is predominantly found in the makes highly complementary RNAs more efficient at protective 2ʹ-O‑methyl modification. Their binding to
cytoplasm of animal cells and triggering modification and degradation of miRNAs. fully complementary viral RNAs will instead mark them
the nucleus of plant and some
fungal cells. Dicer proteins
The endogenous functions of target RNA-directed as foreign sequences that have inappropriately entered
liberate mature microRNA– tailing and trimming of miRNAs in animals remain the miRNA (self) pathway. The tailing and trimmin­g
microRNA* duplexes from elusive, and the proteins that mediate this pathway pathwa­y would then initiate their elimination.
pre-microRNAs and siRNA are unknown. However, tailing and trimming can be
duplexes from long
triggered by transcribed miRNA inhibitors, and use Target RNAs can stabilize miRNAs. In C. elegans, the
double-stranded RNAs.
of these inhibitors in adult mice phenocopies miRNA 5ʹ-to‑3ʹ exoribonuclease XRN‑2 limits miRNA abun-
siRNAs loss‑of‑function mutations, suggesting that this pathway dance. However, miRNAs in RISC are protected from
The ~21 nucleotide small is functionally relevant for miRNA-mediated regulation XRN‑2 when bound to a partially complementary target
RNAs that mediate RNA of gene expression in vivo74,80,81. Moreover, some viruses RNA94. Similarly, persistence of siRNAs in cell culture
interference in plants, animals
and some fungi. Typically
use a similar strategy to interfere with host miRNA is enhanced by the presence of a fully complementary
produced by Dicer processing function during infection (for details, see REFS 82,83). targe­t mRNA95. So, perhaps target binding recruits
of long double-stranded RNA To what extent target RNA-directed miRNA degradation miRNA protective factors or translocates miRNAs to
(dsRNA) precursors encoded in contributes to the regulation of miRNA function in vivo a subcellular location that lacks small RNA-degrading
the genome (endo-siRNAs) or
remains to be determined. But it is tempting to speculate nucleases such as XRN‑2. Alternatively, target RNAs
from exogenous dsRNA
sources (exo-siRNAs) such that the tailing-and-trimming pathway has contributed may protect miRNAs from a release factor that would
as viruses. to the evolution of target sites in mRNAs in animals, as expel them from RISC and expose them to subsequent
mRNAs rarely possess sufficient complementarity to destruction94. Supporting this idea, a partially comple-
RNA-induced silencing miRNAs to trigger miRNA destruction68,69. mentary target RNA also protects worm miRNAs from
complex
(RISC). A ribonucleoprotein
In D. melanogaster, target RNA-directed small RNA destruction by XRN‑1, another 5ʹ-to‑3ʹ exonuclease96.
complex that consists of a decay is restricted to the miRNA pathway. siRNA­s in XRN1 may also degrade miRNAs in animals97.
small RNA guide strand bound flies and PIWI-interacting RNAs (piRNAs) in all ani-
to an Argonaute protein. mals are protected from tailing and trimming by Diverse strategies to generate isomirs
RISC mediates all RNA
2ʹ-O‑methylation of their 3ʹ ends, a modification that Although miRBase catalogues miRNAs as single
silencing pathways, and it can
also include auxiliary proteins is introduced during AGO loading by the methyltrans- sequences, the availability of small RNA sequence data
that extend or modify its ferase HEN1 (Hua enhancer 1)72,84–86. Flies use these from various organisms, tissues and cell types shows
function. protected siRNAs to target AGO2–RISC complexes for that miRNAs comprise multiple isoforms (also known
the repression of extensively complementary targets. as isomirs ) 4,98–104. Such sequence heterogeneity may
Seed sequence
A nucleotide motif in the
In the absence of Hen1, these targets promote the tail- arise from imprecise precursor cropping or dicing,
5ʹ domain of all small silencing ing, trimming and depletion of complementary endo­ terminal trimming or the addition of non-templated
RNAs, which is organized by genous siRNA­s86. HEN1 was first discovered in plants, in nucleotide­s98,103,104 (FIG. 3).
Argonaute to determine which it methylates both miRNAs and siRNAs87,88. Loss
target-RNA recognition.
of HEN1 in plants provokes uridylation and 3ʹ-to‑5ʹ exo- Trimming produces 5ʹ miRNA isoforms. Because the
PIWI-interacting RNAs nucleolytic decay of miRNAs and siRNAs88–91. Plant and 5ʹ end of a miRNA defines its seed sequence, a single
(piRNAs). Small silencing RNAs, animal HEN1 enzymes catalyse the same chemical reac- nucleotide shift at this site will radically alter its target
25–35 nucleotides long, that tion92, but plant small RNAs are methylated as duplexes repertoire. The 5ʹ ends of miRNAs are further con-
bind PIWI clade Argonaute before their loading into AGO88,92,93, whereas animals strained because AGO loading typically selects for
proteins in animals and silence
germline transposons. They are
methylate only a subset of single-stranded small RNAs, miRNAs with distinct 5ʹ nucleotides99,102,105–107. Changes
thought to derive from typically piRNAs. In flies, and presumably in arthropods in the 5ʹ end of a miRNA can therefore alter the effi-
single-stranded RNA precursors more generally, HEN1 methylates siRNAs as the last step ciency of AGO loading. miRNA 5ʹ isoforms comprise
and do not require RNase III in AGO2–RISC assembly 85, but mammalian siRNAs are less than 10% of all miRNA reads in humans, flies and
enzymes for their maturation.
thought not to be HEN1 substrates. worms4,99,108. Nonetheless, some miRNAs produce two
Exoribonuclease or more 5ʹ isomirs that repress distinct target mRNAs4,55.
Enzymes that successively Tailing and trimming distinguish endogenous from for- One of the most striking examples is D. melanogaster
remove nucleotides from either eign small silencing RNAs. In flies, highly complemen- miR‑210, which exists as two nearly equally abundant
the 3ʹ (3ʹ-to‑5ʹ exoribonuclease) tary target RNAs have not been found for miRNA­s bound 5ʹ isoforms98.
or the 5ʹ end (5ʹ-to‑3ʹ
exoribonuclease) of RNA.
to AGO1 (the miRNA-binding AGO protein in flies). miRNA 5ʹ isoforms can also arise from differential
They catalyse phosphodiester Animal miRNAs generally lack a protectiv­e 2ʹ-O‑methyl processing of paralogous pre-miRNAs, each of which pro-
bond cleavage using water modification (FIG. 2). Introduction of RNAs with exten- duces a single miRNA sequence, such as D. melanogaster
(releasing nucleotide sive complementarity to one of these miRNAs triggers pre-miR‑2‑1 and pre-miR‑2‑2 (REF. 99) (FIG.3a).
monophosphates) or inorganic
tailing and trimming of the miRNA. Thus, in flies, Hen1 Other paralogous hairpins, such as human pre-
phosphate (releasing
nucleotide diphosphates) and the tailing-and-trimming pathway might collabo- miR‑133a‑1 and pre-miR‑133a‑2, generate two pre-
as a nucleophile. rate to distinguish self from non-self small RNAs. For dominant isoforms 4. Some extreme examples of
example, anti-viral siRNAs bound to AGO2 (the siRNA- 5ʹ  heterogeneity were discovered in mammals and
Isomirs bindin­g AGO protein in flies) bind and destroy viral chicken­s for a set of mirtron­s that contain a long 5ʹ exten-
microRNA variants containing
sequences that deviate from
RNA without risk of tailing and trimming, because they sion connecting the hairpin to the 5ʹ splice site4,109,110.
miRBase-annotated or most have a 2ʹ-O‑methyl-modified 3ʹ end. During viral infec- After splicing, the 5ʹ extension is trimmed ‘sloppily’ to
frequently observed species. tion, non-self-directed siRNAs could be so abundant produce several mature miRNAs.

478 | AUGUST 2013 | VOLUME 14 www.nature.com/reviews/molcellbio

© 2013 Macmillan Publishers Limited. All rights reserved


REVIEWS

a 5′ miRNA isoforms
miR-2-1
Paralogous Seed
miRNAs
miR-2-2 5′ 3′

Lariat 5′-to-3′ exonucleolytic


debranching maturation
? +1 –1

Branched mirtron
5′ tailed mirtron

Seed
b Internal miRNA isoforms A-to-I editing
ADAR 5′ 3′
A A
I
c 3′ miRNA isoforms I
Lariat 5′ to 3′ exonucleolytic
debranching maturation

Branched mirtron
3′ tailed mirtron
Seed
Exosome

5′ to 3′ exonucleolytic 5′ 3′
miRNA* maturation
AGO1 Trimming

24 nt 22 nt A
Nibbler Tailing
UU
-2 +2
ZCCH6, ZCCHC11 Dicer
or GLD-2
U U
Pre-miRNA Loqs or Monouridylated
Distributive TRBP miRNA duplex
ZCCHC11 LIN28 DIS3L2
UUU UUU

Processive Pre-miRNA decay

Figure 3 | Mechanisms and consequences of microRNA isoforms.  Precursor microRNAs (pre-miRNAs) and mature
Nature Reviews | Molecular Cell Biology
miRNAs can be modified by the addition, editing or subtraction of nucleotides, resulting in distinct miRNA isoforms.
Such isomirs can be classified as 5ʹ, internal and 3ʹ isoforms. a | 5ʹ miRNA isoforms can be produced by differential
processing of paralogous miRNAs, as exemplified by the miR‑2 family in Drosophila melanogaster or the exonucleolytic
processing of 5ʹ tailed mirtrons. This results in miRNAs with different seed sequences, which are the primary sequence
determinants for target RNA binding. Consequently, different 5ʹ miRNA isoforms regulate different sets of target RNAs.
The identity of the exonuclease that targets 5ʹ tailed mirtrons is not known (indicated by the question mark). b | Internal
miRNA isoforms are rare and almost exclusively generated by editing of adenosine to inosine by the enzyme ADAR
(adenosine deaminase acting on RNA). c | The 3ʹ end of a miRNA shows the greatest heterogeneity, which is a result of
both exonucleolytic trimming and the addition of nucleotides (typically adenine or uridine) to the 3ʹ end (tailing).
Exonucleolytic trimming is required for the efficient processing of 3ʹ tailed mirtrons, as well as ~25% of all miRNAs in flies.
Dicer-1 initially makes these miRNAs as ~24 nucleotide miRNA duplexes. Upon loading into Argonaute 1 (AGO1), such
‘long’ miRNAs are trimmed by the 3ʹ-to‑5ʹ exoribonuclease Nibbler to produce mature ~22 nt miRNAs. In mammals,
monouridylation of selected pre-miRNAs by ZCCHC11, GLD-2 and/or ZCCHC6 enhances substrate recognition and
processing by Dicer. In the case of miRNAs derived from the 3ʹ arm of such pre-miRNAs, monouridylation is propagated
into the mature miRNAs, contributing to its 3ʹ heterogeneity. In the presence of the pre-miRNA-binding protein LIN28,
ZCCHC11 is a processive enzyme, oligouridylating pre-miRNAs that contain a LIN28‑binding motif. Unlike
monouridylation, oligouridylation inhibits Dicer-mediated processing and enhances pre-miRNA decay by the
3ʹ-to‑5ʹ exoribonuclease DIS3L2. Loqs, Loquacious; TRBP, transactivation-response RNA-binding protein.

Internal miRNA isoforms. Modifications that change correspond to deamination of adenosine to inosine
Mirtrons the internal sequence of miRNAs are rare and occur at by ADAR (adenosine deaminase acting on RNA) 115
Intron-derived precursor frequencies that are similar to sequencing errors4,98,111. (FIG. 3b). ADAR enzymes also edit long dsRNAs, pro-
microRNAs excised from However, for a few individual miRNA species, as much moting their unwinding and preventing them from
primary miRNAs by the
splicing machinery and a
as 80% of all sequence reads for the miRNA contain at triggering RNAi116–118, as well as local double-stranded
lariat-debranching enzyme least one base change from the corresponding genomic structures. Indeed, these local structures may have
instead of Drosha. sequence4,98,112–114. Such editing events nearly always evolved to facilitate ADAR-mediated repair of mRNAs

NATURE REVIEWS | MOLECULAR CELL BIOLOGY VOLUME 14 | AUGUST 2013 | 479

© 2013 Macmillan Publishers Limited. All rights reserved


REVIEWS

that lack the appropriate protein-coding sequence. pre-miRNA and prevents efficient substrate recogni-
Because ADAR enzymes are preferentially expressed tion by Dicer, whereas the oligo(U) tail serves as a decay
in neural tissues of flies and mammals, miRNAs in signal for the Perlman syndrome exonuclease DIS3L2
the brain are often edited. The biological significance (REF. 132).
of miRNA editing outside the seed sequence, however, Surprisingly, in the absence of LIN28, ZCCHC11
remains to be established. and its close relative ZCCHC6 monouridylate a selected
class of pre-miRNAs, including several members of the
Trimming can produce 3ʹ miRNA isoforms. miRNA iso- pre-let‑7 family (class II pre-miRNAs). Monouridylation
forms often differ at their 3ʹ ends99,100,102,103,108. miRNAs enhances dicing, because it restores the two nucleotide
with a common 5ʹ end but different 3ʹ ends are presumed 3ʹ  overhang of pre-miRNAs that were imprecisely
to target the same mRNAs, but they may differ in the cropped by Drosha133.
extent of target repression or they might have different Mature miRNAs can also serve as substrates for
half-lifes. Recent studies implicate 3ʹ-to‑5ʹ trimming as TNTases, but the triggers and biological consequence of
one major source of miRNA 3ʹ heterogeneity. this type of modification are less well-established. For
In D. melanogaster, ~40% of AGO1‑bound miRNAs example, the pre-miRNA tailing enzyme ZCCHC11
are trimmed at their 3ʹ termini, irrespective of whether apparently also adds uridine to mature mammalian
they are produced via the canonical or the mirtron path- miRNAs such as miR‑26, and this somehow facilitates
way 104. In extreme cases, such as miR‑34 or miR‑317, cytokine expression134. GLD2 adds one adenosine to
four to six variant isoforms accumulate that differ at miR‑122, which is the most abundant miRNA in the
their 3ʹ ends104,119,120. This 3ʹ trimming requires the liver 135. miR‑122 derives from the 5ʹ arm of pre-miR‑122,
3ʹ-to‑5ʹ exoribonuclease Nibbler 119,120 (FIG. 3c). so GLD2 must act on the mature miRNA after dicing.
Nibbler mediates trimming of more than a quarter of In mice lacking GLD2, both miR‑122 adenylation and
all fly miRNAs. It acts predominantly on miRNAs that miR‑122 abundance decrease, suggesting that adenyla-
arise from pre-miRNAs containing bulges or internal tion enhances miR‑122 stability. GLD2 adds tails to
loops that force Dicer or Drosha to generate an atypically several other miRNAs in mammals, although this does
long product. After the miRNA is loaded into AGO1, not always increase their stability 136. A protective func-
Nibbler trims these long miRNAs into shorter isoforms, tion for miRNA adenylation has also been reported in
perhaps to ensure that their 3ʹ ends nestle in the AGO1 plants137.
PAZ domain. Human miRNAs also seem to be trimmed In the absence of the methyltransferase HEN1,
by a Nibbler-like enzyme120. The C. elegans homologue uridylation of mature plant miRNAs by the TNTase
of Nibbler, MUT‑7, is required for transposon silencing, HEN1 SUPPRESSOR 1 (HESO1) triggers miRNA
RNAi and co-suppression121–124. But a role for MUT‑7 in decay 88,89,138,139. Loss of HESO1 partially restores small
the miRNA pathway has not been reported. RNA function and suppresses developmental pheno-
types in HEN1 mutant plants138,139 (FIG. 2). Because HEN1
Tailing: a special form of 3ʹ heterogeneity. The addi- methylates miRNAs before they bind AGO proteins, the
tion of non-templated nucleotides by terminal nucleotidy­l substrates for tailing are inferred to be duplexes.
transferase­s (TNTases) (tailing) gives rise to a special form In the green alga C. reinhardtii, loss of the TNTase
of 3ʹ heterogeneity in miRNAs and other types of RNA. MUT68 reduces small RNA uridylation and increases
All TNTases are members of the DNA polymerase‑β miRNA and siRNA abundance140. In vitro, MUT68 tails
superfamily that includes poly(A) polymerase, which synthetic small RNAs, and this promotes their degra-
adds the poly(A) tail to mRNAs. TNTases typically add dation by the catalytic exosome subunit RRP6 (ribo­
adenosine or uridine in vivo125. The consequences of tail- somal RNA-processing protein 6). Like loss of MUT68,
ing can vary. In bacteria, the poly(A) tail triggers mRNA loss of RRP6 stabilizes small RNAs in vivo. MUT68 is
decay, whereas in eukaryotes it enhances mRNA stability also implicated in the degradation of the 5ʹ target RNA
and translation126. Moreover, uridylation has been impli- fragments produced by siRNA- or miRNA-directed
cated in the maturation and turnover of coding and RISC cleavage141, which have previously been reported
non‑codin­g RNAs127. to serve as substrates for tailing in Arabidopsis thaliana
Several studies have recently implicated TNTases in and mammals142. Perhaps enzymes that can tail either
the regulation of miRNAs, through uridylation of their miRNAs or their targets are both recruited when AGO
3ʹ ends (FIG. 3c). These include ZCCHC6 (also known as proteins bind an RNA.
TUT7) and ZCCHC11 (also known as TUT4), which
are both zinc-finger- and CCHC domain-containing AGO creates the seed
proteins, and GLD2 (also known as TUT2 and PAPD4), AGO proteins create a seed in any small RNA sequence
which is a poly(A) polymerase domain-containing- they bind. Mutations in the seed block target cleavage
protein. For example, the TNTase ZCCHC11 oligo­ by plant miRNAs in extracts143 and in vivo144. In addi-
uridylates pre-let‑7 in mouse embryonic stem cells128,129. tion, such mutations relieve the repression of highly
Terminal nucleotidyl Oligouridylation requires the formation of a ternary complementary mRNAs by miRNAs and siRNAs in
transferases complex, consisting of ZCCHC11, pre-let‑7 and the animals145,146, almost certainly because a seed match is
(TNTases). Template-
independent polymerases that
pluripotency factor LIN28, which directly binds to a required for high-affinity binding between a miRNA
add nucleotides to the 3ʹ ends sequence motif in the loop of pre-let‑7 (REFS 130,131). and its target 147. Structural analyses of archael, bacterial,
of nucleic acids. Tailing masks the two nucleotide 3ʹ overhang of the yeast and human AGO proteins reveal how AGO creates

480 | AUGUST 2013 | VOLUME 14 www.nature.com/reviews/molcellbio

© 2013 Macmillan Publishers Limited. All rights reserved


REVIEWS

the seed79,148–154. Upon AGO binding, the seed bases How do miRNAs regulate mRNA expression? Some
are exposed to the solvent and pre-ordered in a quasi- AGO proteins cleave highly complementary target
helical structure that ‘pre-pays’ the entropic penalty RNAs, a process that is catalysed by their RNase H‑like
that comes with nucleic acid pairing 151. Structural and PIWI domain, which positions a pair of Mg 2+ ions at
kinetic data suggest that the seed nucleates binding to the scissile phosphate. This endonucleolytic mechanism
complementary sequences in the target RNA (the seed derives from the ancestral RNAi pathway, in which target
match) and that pairing of the target with sequences 3ʹ to ‘slicing’ provides an anti-viral and anti-transposon
the seed can only occur subsequently, after a structural defence.
rearrangemen­t of the AGO protein79,147,150,152,155. In plants, highly complementary miRNA target
This nucleation theory was recently modified after sites trigger mRNA cleavage143,171–176, but at least some
the discovery that a set of miR‑124 targets in the mouse miRNAs can block translation177,178 (FIG. 4). By contrast,
brain seem to lack full seed complementarity. Instead, just a few miRNA–target pairs in mammals have suf-
they contain one ‘bulged’ nucleotide across from posi- ficient complementarity to direct AGO2 to cleave the
tions 5 and 6 of the miRNA seed156 (FIG. 4). In these cases, target 170,179–181. Even for these atypical examples, the con-
an initial seed–target interaction across five nucleotides tribution of miRNA-directed target cleavage to the over-
has been proposed to rearrange into a six nucleotid­e all decrease in target mRNA abundance is unknown.
helix containing one additional unpaired (that is, flipped Moreover, three of the four human AGO clade proteins
out or bulged) target base. Such sites constitute more are catalytically inactive; only a miRNA bound to AGO2
than 15% of all AGO-bound sequences in the mouse can cleave a highly complementary target RNA. This
brain, are evolutionary conserved and mediate target suggests that AGO proteins more often mediate target
mRNA repression156. These studies are consistent with gene expression control through a mechanism that is
in vitro analyses of AGO in the eubacterium Thermus independent of RNA endonucleolytic cleavage.
thermophilus, which tolerates a flipped out base in The overwhelming majority of miRNAs recognize
the seed-matching sequences of its target 150, as well their targets solely through their seed sequence, and such
as thermo­dynamic studies on the recognition of let‑7 partial complementarity does not permit target cleavage
target­s in worms157. even when the miRNA is bound to AGO2. Nonetheless,
Pairing between a target and miRNA bases that are 3ʹ animal miRNAs typically trigger destruction of the
to the seed — that is, the ‘supplementary’ base pairs from mRNAs they bind182–190. Quantitative high-throughput
miRNA position 13 to 16 — can decrease the off rate sequencing and proteomics methods for measuring
(koff) of RISC from a target RNA, which may explain why mRNA and protein abundance as well as new methods
such additional pairing can enhance target repression64,147 that measure ribosome occupancy of mRNAs have only
(FIG. 4e). Moreover, features other than base pairing can fuelled the debate on how miRNAs regulate their targets.
have a substantial effect on target regulation and will Ribosome profiling experiments in mice suggest
contribute to the quality of a miRNA-binding site. These that mRNA destruction by a process distinct from
include: an adenosine across from the first nucleotide of endonucleolytic cleavage, rather than translational
the miRNA70,158; the accessibility of a target site64,159–161; repression, explains the overwhelming majority
and the position of a target site within a 3ʹ UTR64. mRNA- (>84%) of miRNA-directed repression in mammals188.
binding proteins add an additional layer of complexity, Ribosome profiling quantitatively determines the posi-
as they can compete with and perhap­s even enhance tions of ribosomes on cellular mRNAs with codon-
miRNA binding 162–169. level resolution for the entire transcriptome191. Such
Alternative modes of miRNA target recognition have data have reinforced the view that most of the decrease
been reported, but these seem to have a modest impact in steady-state protein abundance that is observed
on gene regulation: fewer than 10% of all target sites are during miRNA-mediated repression reflects mRNA
predicted to include base pairs that compensate for an decay 188,189,192. Nonetheless, a small but significant frac-
incomplete seed match, and even fewer sites lack seed tion (11–16%) of miRNA repression seems to derive
matches altogether 63,170 (FIG. 4e). miRNAs can also con- from a block in translation190.
tain sequence motifs that regulate their stability or intra- Although mRNA destruction is necessarily mutually
cellular localization (BOX 1). Nonetheless, the association exclusive with translational repression, studies in zebrafish
of a seed sequence with the target mRNA remains the suggest that translational repression can precede mRNA
predominant mechanism for target recognition. degradation. Zebrafish miR‑430 is the predominant
miRNA expressed at the onset of zygotic transcription.
The great miRNA function debate miR‑430 clears maternal mRNAs from the embryo, a
Despite a plethora of high-quality studies examining the process that is prevented in mutants lacking both mater-
biochemistry, biology and genomics of miRNA-directed nal and zygotic Dicer 185,193. Ribosome profiling of wild-
mRNA regulation, how miRNAs repress or activate gene type and Dicer mutant fish shows that miR‑430 initially
expression in animals remains controversial. There is reduces ribosome occupancy of target mRNAs — which
general agreement that miRNAs regulate gene expres- represses their translation — and then, two hours later,
sion post-transcriptionally, but several mechanisms have triggers mRNA destruction193. The initial decrease pri-
been proposed to explain how, ranging from repression marily reflects a change in the rate of translational initia-
of translational initiation or elongation to acceleration of tion, a mechanism of miRNA action previously observed
general mRNA decay processes (FIG. 4c,d). for fly AGO1 in  vitro 194–196 and in cell extracts197,198.

NATURE REVIEWS | MOLECULAR CELL BIOLOGY VOLUME 14 | AUGUST 2013 | 481

© 2013 Macmillan Publishers Limited. All rights reserved


REVIEWS

a Endonucleolytic cleavage b
AGO m7Gppp Plants (frequent) 10,11
(n)AAAA miR-171 5′-UGAUUGAGCCGCGCCAAUAUC-3′
SCL6 mRNA 3′-...ACUAACUCGGCGCGGUUAUAG...-5′

(n) AAAA Animals (rare) 10,11


miR-196a 5′-UAGGUAGUUUCAUGUUGUUGGG-3′
UUUU
HoxB8 3′ UTR 3′-...AUCCGUCAAAGUACAACAACCU...-5′
XRN1 or XRN4
Exosome

c Translational repression
Block to translation e
initiation Plants (rare?)
miR-172a/c 5′-AGAAUCUUGAUGAUGCUGCAU-3′
SCL6 mRNA 3′-...UCUUAGGACUACUACGACGUC...-5′
m7Gppp
(n)AAAA
ORF
Animals (canonical seed match site; most frequent)
2 8
lin-4 5′-UCCCUGAGACCUCAAGUGUGA-3′

Change in repressive mechanism over time?


lin-14 3′ UTR 3′-...AGGGACUCAACCUAAUUUUCU...-5′
AA

A Deadenylation
A Animals (G-bulge site; less frequent?)
Recruitment of 2 8
translation blockers miR-124 5′-UAAGG-CACGCGGUGAAUGCC-3′
Mink1 3′ UTR 3′-...GUUCCGGUGAUGUAACUCUUU...-5′

(n)AAAA
Animals (3′ supplementary site; less frequent?)
2 7 13 16
miR-2 5′-UAUCACAGCCAGCUUUGAGGAGC-3′
grim 3′ UTR 3′-...UUAGUGUUACGCGAAACUAACUC...-5′
d mRNA turnover Decapping and
5′-to-3′ decay
Animals (3′ compensatory site; rare)
2 8
let-7 5′-UGAG-GUAGUA-GGUUGUAUAGUU-3′
AA
lin-41 3′ UTR 3′-...ACUCACAUCUUGCCAACAUAUUUU...-5′
A Deadenylation
A

Figure 4 | microRNA function in plants and animals.  a | Most plant microRNAs (miRNAs) and a few animal miRNAs
direct endonucleolytic cleavage (slicing) of their mRNA targets. The 5ʹ‑to‑3ʹ exoribonuclease XRN4
Nature Reviews in plants and
| Molecular CellXRN1
Biology
in animals, together with the major cellular 3ʹ‑to‑5ʹ exonucleolytic complex, the exosome, subsequently degrade the
sliced mRNA fragments. The 3ʹ end of the 5ʹ cleavage product is frequently uridylated, perhaps to mark these fragments
for decay. b | miRNA-directed endonucleolytic cleavage of mRNAs requires extensive complementarity between the
miRNA and its target site (representative examples in plants and mammals are shown). Endonucleolytic cleavage occurs
at the phosphodiester bonds across from nucleotides 10 and 11 of the miRNA, counted from the miRNA 5ʹ end. Although
slicing seems to be the dominant mode of action for miRNAs in plants, highly complementary sites in the transcriptome
of animals are rare. c | In animals, miRNAs were originally proposed to repress translation of an open reading frame (ORF).
Biochemical studies have suggested that miRNAs have a role in blocking translational initiation, in poly(A) tail
shortening  or in the recruitment of protein cofactors that can interfere with translation. d | In many cells and tissues,
miRNA-directed translational repression is indistinguishable from mRNA destruction via decapping and 5ʹ‑to‑3ʹ decay.
This has led to the suggestion that miRNAs directly target mRNAs for decay. Another possibility is that the inhibition of
translation by miRNAs (part c) triggers subsequent mRNA decay, and the temporal delay between these two effects can
vary depending on the surveillance mechanisms in place in particular cellular contexts (depicted by blue arrow on the
right). e | Like animal miRNAs, plant miRNAs may regulate target mRNAs via mechanisms other than endonucleolytic
cleavage. But the underlying miRNA–target RNA interaction (for example, miR‑172–SCL6) is indistinguishable from sites
that trigger cleavage (part b). In contrast to plants, animal miRNAs are almost always imperfectly complementary to the
target mRNAs they regulate. The seed sequence (miRNA nucleotides 2 to 8; sometimes interrupted by a specific
G‑bulge) is the major determinant for target binding and often suffices to trigger mRNA repression. Additional pairing
of miRNA nucleotides 12 to 16 can sometimes bolster seed binding (3ʹ supplementary sites). Only in rare cases can an
imperfectly matching seed sequence be compensated for by extensive complementarity between the miRNA 3ʹ region
and the target site (3ʹ compensatory sites). Binding of animal miRNAs to partially complementary target sites generally
results in translational inhibition and/or mRNA decay via the mechanisms shown in part c and part d. HoxB8, homeobox B8;
UTR, untranslated region.

482 | AUGUST 2013 | VOLUME 14 www.nature.com/reviews/molcellbio

© 2013 Macmillan Publishers Limited. All rights reserved


REVIEWS

Box 1 | microRNA sequence motifs


Although the precise mechanism by which overlap-
ping non-coding transcripts decrease gene expression
Sequence motifs within a microRNA (miRNA) can direct its localization to particular remains unknown, miR‑373 may similarly destabilize or
sites in a cell or enhance its turnover. For example, a hexamer motif in miR‑29b triggers block the function of non-coding transcripts that cross
its nuclear import241. Nuclear miR‑29b may regulate gene expression in the nucleus or, the promoter regions of protein-coding genes.
alternatively, the nuclear import of miR‑29b may alter target gene expression in the
The requirement for miR‑122 in hepatitis C virus
cytosol by sequestering the miRNA and effectively decreasing its concentration in
the cytoplasm. This hexamer motif is necessary and sufficient for nuclear import,
replication is a particularly intriguing example of activa-
as insertion of this sequence into another miRNA directs its nuclear import. tion by a miRNA: binding of miR‑122 to complementary
Several sequence motifs have been linked to miRNA turnover. In the absence of such sites in the 5ʹ end of the virus genome is required for effi-
motifs, miRNAs are long-lived, with half-lives as long as 5 days242 or even 2 weeks243. cient viral replication and enhances internal ribosome
For example, a seven nucleotide motif at the 3ʹ end of miR‑382 triggers its decay in entry (IRES) site-directed translation207. The mechanism
HEK293 cells97. Furthermore, destruction of miR‑16 family miRNAs upon re‑entry into of this activation is not known, nor is it clear whether
the cell cycle after serum starvation relies on both the seed sequence and a sequence miRNAs have similar effects on viral gene expression
at the 3ʹ end of the miRNAs244. Uridylation of transiently transfected miR‑29b and as they do on mammalian mRNAs. Nonetheless, the
miR‑29c at positions 9–11 elicits their rapid turnover in HeLa cells, a phenomenon that finding that a readily inhibited host factor is required
seems to be conserved in the fly miRNA bantam245. For miR‑29b and miR‑29c, the
for viral replication makes miR‑122 one of the most
miRNA–miRNA* duplex seems to be targeted for degradation rather than the mature
Argonaute (AGO)-bound miRNA. The proteins that recognize the sequence motifs
promisin­g drug targets for treating hepatitis C208,209.
controlling miRNA localization or stability and whether such recognition occurs while A more unusual and direct form of miRNA activa-
the miRNA is still bound to AGO are not yet known. tion of gene expression has been described following cell
cycle arrest. Upon serum starvation, miR‑369‑3 binds
an AU‑rich element in the 3ʹ UTR of the gene encodin­g
The timing of mRNA poly(A) tail shortening, which was tumour necrosis factor (TNF). miR‑369‑3 normally
previously proposed to explain miR‑430‑directed transla- directs repression but, in these specific conditions,
tional repression in the same model system185, is similar to it seems to promote translation210,211. A similar phenom-
the effects on mRNA decay but not to the initial decrease enon has also been observed for the translational activa-
in translational initiation193. Together with previous bio- tion of ribosomal protein genes, although miR‑10a was
chemical studies198–201 and more recent kinetic analysis proposed to bind non-seed-matching sites in the 5ʹ UTR
in a D. melanogaster cell line202, these studies point to a of mRNAs212.
testable, unifying hypothesis: miRNAs initially repress
mRNA translation and later trigger mRNA decay, per- Apostasy or prophecy? Not only are the mechanisms
haps in response to the block to translation rather than by which miRNAs repress gene expression debated but
any specific property of miRNAs or AGO proteins. The also who represses whom. The crux of the issue is the
timing of decay might vary depending on the context: seemingly modest effect that miRNAs have on the vast
in cells with robust surveillance mechanisms, mRNAs majority of mRNAs they target. A generally accepted
bound by miRNAs would be rapidly degraded, whereas in hypothesis posits that many miRNAs ‘tune’ the expres-
early embryos, translationally repressed mRNAs would be sion of most targets, with only a small number of target­s
stable. Mutations in the mRNA surveillance or degrada- experiencing a large change in mRNA or protein abun-
tion machinery should prove useful in testing these ideas. dance63. The biological importance of such tuning is
We note that all forms of this ‘translational-repression- thought to emerge in specific environmental conditions
triggers-mRNA-decay’ hypothesis imply biochemical that are not typically encountered by animals reared in a
coupling between the rate of translation and the stability laboratory. Supporting this view, the role of miR‑7 in
of an mRNA. Proof of such coupling will require rigorous, ensuring correct patterning of the D. melanogaster eye
quantitative in vivo measurements of the rates of transla- is only apparent in sensitized genetic backgrounds or
tion and decay for mRNAs targeted by miRNAs in the when the fly experiences ‘wild swings’ in temperature
presence and absence of miRNA function. during eye development 213. Similarly, the loss of indi-
vidual miRNA­s or miRNA families in C. elegans rarely
miRNAs turn me on? miRNAs generally repress gene produces an obvious phenotypic defect214.
expression, but they have also been linked to gene acti- An alternative view proposes that most ‘tuning targets’
vation. Often, the mechanism of activation is indirect, are actually ‘titrating targets’ that moderate the effective
with repression of a repressor leading to increased concentration of the miRNA available to bind a smaller
expression of specific transcripts. One such case occurs set of highly regulated targets215. The concept of titrat-
during neuronal differentiation in frogs, chickens and ing sites is supported by the finding that artificial, long,
mammals203, in which miR‑128 in the brain targets highly abundant RNAs containing miRNA-binding sites
components of the nonsense-mediated decay machin- can act as competitive inhibitors of specific miRNAs.
ery and thereby increases the abundance of mRNAs Indeed, the first specific miRNA inhibitors were synthetic
involved in neuronal differentiation and brain function. anti-miRNA oligonucleotides216,217. Similarly, transgenic
A second example may be the activation of trans­ or transfected genes containing multiple miRNA-binding
cription by miR‑373 (REF.  204) . In other contexts, sites can inhibit the function of miRNAs in animals218–220.
repression of non-coding transcripts that cross the Such miRNA ‘sponges’ typically contain non-natural
promoters of overlapping genes have been reported to miRNA-binding sites that are designed with central
increase the abundance of the sense gene transcript205,206. mismatche­s to block target cleavage.

NATURE REVIEWS | MOLECULAR CELL BIOLOGY VOLUME 14 | AUGUST 2013 | 483

© 2013 Macmillan Publishers Limited. All rights reserved


REVIEWS

Biologically irrelevant Biologically relevant


[miRNA]<<Kd [miRNA]>>Kd
AGO AGO

m7Gppp m7Gppp
(n)AAAA
ORF (n)AAAA
ORF
No occupancy [Target sites] < [miRNA] [Target sites] > [miRNA]
Frequent? Rare?

Strong regulation Weak regulation

Figure 5 | A quantitative framework for microRNA function.  In this model, the relative abundance of microRNAs
(miRNAs) and their targets inside the cell dictates the regulatory outcome. At concentrations
Nature Reviewsbelow the affinity
| Molecular of Biology
Cell
Argonaute (AGO)-bound miRNAs for their targets (Kd), occupancy of miRNAs at mRNA targets is negligible and, thus, the
biological function of these miRNAs is irrelevant. If present at concentrations above the Kd (this is typically true at least
for the most abundant miRNAs in a given cell type or tissue), virtually every target site will be bound by a miRNA. In this
case, the cumulative abundance of all target sites in the cell, relative to the abundance of the miRNA, will determine the
extent to which each individual target mRNA is regulated. Fewer total target sites than the corresponding miRNA will
prompt a larger extent of regulation compared with miRNAs that are less abundant than their target sites.

miRNA sponges occur naturally in plants and act to their dissociation constant (Kd) values for seed match
competitively inhibit miRNA repression of other targets sites 147. Nematode and human miRNAs can reach
in a phenomenon termed target mimicry 221. For exam- 50,000 molecules per cell, with some being as abundant
ple, phosphate starvation in A. thaliana induces the as the spliceosomal U6 small nuclear RNA (snRNA)229,230.
expression of both miR‑399 and the non-coding RNA By contrast, the 5,000 most abundant mRNA species in
INDUCED BY PHOSPHATE STARVATION 1 (IPS1). nematodes are thought to be present at just ~100 mol-
IPS1 binds miR‑399 through a site that contains cen- ecules each per cell229. Many miRNAs therefore out-
tral mismatches that protect IPS1 from AGO-catalysed number any single mRNA target by as much as 500‑fold.
cleavage. IPS1 thereby reduces the amount of miR‑399 However, each animal miRNA has the potential to regu-
available to repress its target mRNA PHOSPHATE 2. late hundreds of mRNAs, some of which contain more
Transgenic non-cleavable miRNA sponges modelled after than one miRNA‑binding site. Consequently, the num-
IPS1 inhibit individual miRNAs in plants221,222. ber of some miRNAs may be similar to the number of
Notably, the titrating targets model leaves unexplained miRNA‑bindin­g sites, suggesting that a model in which
the apparent conservation of miRNA-binding sites in a miRNA is regulated by numerous titrating targets is
titrating targets: how would so many sites, each of which biochemically more feasible than one in which individual
contributes just a little to reducing the free pool of the target mRNAs are key. Moreover, computer simulations
miRNA, be under positive evolutionary selection? Should using experimentally measured Kd and koff values suggest
such sites not come and go over evolutionary time, with that only miRNAs that are least likely to provide meaning-
only their number and not their locations within any ful target repression in a given cell type could be subject to
specific mRNA being conserved? A more speculative competition by highly abundant, individual non-coding
and thermodynamically implausible version of this idea RNAs or mRNAs bearing miRNA-binding sites147.
proposes that individual target RNAs — not the collec-
tive abundance of miRNA seed matches — can titrate A quantitative framework for miRNA and target RNA
the amount of miRNA that is available to repress their function. The biochemical and biophysical properties of
targets223. For example, protein-coding and non-coding AGO-bound miRNAs provide a quantitative framework
transcripts have been predicted to regulate the dosage- for understanding the reciprocal function of miRNAs and
sensitive tumour suppressor gene PTEN by competing their targets according to their relative abundance inside
with the PTEN transcript for binding to a similar set of the cell (FIG. 5). When a miRNA is dilute, that is, at concen-
miRNAs224. A decrease or loss of competing transcripts, trations lower than the affinity of AGO-bound miRNA­s
for example derived from the pseudogene PTENP1 or for their targets, the overall impact of the miRNA on
the protein-coding gene zinc-finger E‑box-binding gene expression is expected to be negligible147. This pre-
homeobox 2 (ZEB2), would increase the amount of diction is supported by miRNA sensor assays in cultured
miRNAs available to repress PTEN and hence accelerate monocytes, which revealed that only the most abundant
tumorigenesis225–228. miRNAs are able to change the expression of a reporter
Among the objections to this idea are the findings construct containing a miRNA-binding site231.
that many miRNAs are more abundant than the sum Highly abundant miRNAs (present at intracellular
of their individual mRNA targets and that the concen- concentrations that vastly exceed the Kd for miRNA
tration of miRNAs in vivo are probably far greater than binding to a typical binding site on an mRNA target) are

484 | AUGUST 2013 | VOLUME 14 www.nature.com/reviews/molcellbio

© 2013 Macmillan Publishers Limited. All rights reserved


REVIEWS

predicted to be bound to target mRNA. The extent of reg- complementarity to miR‑7 and is proposed to function
ulation, that is, the number of sites on each target mRNA as a ‘super-sponge’ to regulate miR‑7 function in the neu-
occupied by a miRNA, is then determined by the overall ronal tissues in which both the miRNA and the circRNA
abundance of target sites in the transcriptome. In this are co‑expressed. Although the biological function and
case, significant regulation of miRNA activity through the general scope of this type of regulation are unclear,
competitive transcripts is in principle possible, but only ectopic expression of CDR1as in zebrafish, which natu-
if the overall abundance of the competitor is much higher rally produce miR‑7 but not CDR1as, impairs midbrain
than the total number of miRNAs. Experimentally, this development, recapitulating the loss‑of‑function pheno-
can be achieved by either expressing artificially high lev- type of miR‑7 (REF. 240). The high abundance and strik-
els of sponge transcripts220,232–234 or by vastly increasing ingly high numbers of miRNA-binding sites in CDR1as
the number of miRNA-binding sites in each transcript219. are consistent with its proposed function in miRNA inhi-
For endogenous mRNAs or non-coding transcripts, bition. More intriguingly, its circular architecture might
it is unlikely that sufficiently high concentrations can prevent its degradation by miRNAs, particularly if such
be achieved to allow competitive inhibition of abundant miRNA-directed destruction requires ribosome binding
miRNAs, particularly because miRNA binding to the or recruitment of exonucleases.
competitor transcripts should decrease their steady-state
abundance via one of the mechanisms highlighted above. Conclusions and perspectives
Nevertheless, miRNA decay induced by viral infection This year marks the twentieth anniversary of the discovery
suggests that mechanisms exist that efficiently inhibit of miRNAs1,2. In 1993, it would have been hard to imagine
miRNA function by triggering their decay upon inter­ the extent and complexity of miRNA-mediated regulation
action with target RNAs235–237, perhaps because enzymatic of gene expression. Those studying these tiny riboregu-
degradation and not simple competition renders this pro- lators can deservedly be proud of our current catalogue
cess efficient72–74,236. Because viruses only contribute the of miRNAs, their mRNA targets and mechanisms of
target RNA signal for miRNA decay, and the molecular biogenesis and sequence diversification. Although many
machinery seems to be provided by the host cell itself, it details surely remain to be discovered, what we most lack
is tempting to speculate that this process may be used to is a deeper understanding of the biological functions of
regulate miRNA function through cellular transcripts in miRNAs. Are unifying principles and deeper explana-
uninfected cells, but biological examples of this have not tions for the role that miRNAs have in coordinating gene
yet been identified. expression during development and environmental expe-
rience to be found? Or does much of what we now view
Squaring the circle. An unexpected, natural solution to as miRNA-mediated regulation or seemingly purposeful
the competitive miRNA inhibitor paradox seems to have mechanisms to diversify miRNA sequences simply cor-
now been identified in the form of previously overlooked respond to neutral experiments in evolution, the results
circular RNAs (circRNAs)238–240. circRNAs are generated of which we will be unable to observe? We believe that the
by head‑to‑tail splicing of exons. The human circRNA, answers to these and other questions about miRNAs lie in
CDR1as (antisense to the cerebellar degeneration-related rigorous computational, biochemical and genetic tests of
protein 1), contains 63 conserved sites with imperfect quantitative models.

1. Lee, R. C., Feinbaum, R. L. & Ambros, V. The C. elegans 8. Grimson, A. et al. Early origins and evolution of 19. Corcoran, D. L. et al. Features of mammalian
heterochronic gene lin‑4 encodes small RNAs with microRNAs and Piwi-interacting RNAs in animals. microRNA promoters emerge from polymerase II
antisense complementarity to lin‑14. Cell 75, Nature 455, 1193–1197 (2008). chromatin immunoprecipitation data. PLoS ONE 4,
843–854 (1993). 9. Molnar, A., Schwach, F., Studholme, D. J., e5279 (2009).
2. Wightman, B., Ha, I. & Ruvkun, G. Thuenemann, E. C. & Baulcombe, D. C. miRNAs control 20. Rodriguez, A., Griffiths-Jones, S., Ashurst, J. L. &
Posttranscriptional regulation of the heterochronic gene expression in the single-cell alga Chlamydomonas Bradley, A. Identification of mammalian microRNA
gene lin‑14 by lin‑4 mediates temporal pattern reinhardtii. Nature 447, 1126–1129 (2007). host genes and transcription units. Genome Res. 14,
formation in C. elegans. Cell 75, 855–862 10. Kim, V. N., Han, J. & Siomi, M. C. Biogenesis of small 1902–1910 (2004).
(1993). RNAs in animals. Nature Rev. Mol. Cell Biol. 10, 21. Baskerville, S. & Bartel, D. P. Microarray profiling of
References 1 and 2 report the discovery of the first 126–139 (2009). microRNAs reveals frequent coexpression with
miRNA, lin‑4, and propose that it regulates mRNA 11. Axtell, M. J., Westholm, J. O. & Lai, E. C. Vive la neighboring miRNAs and host genes. RNA 11,
expression post-transcriptionally. différence: biogenesis and evolution of microRNAs 241–247 (2005).
3. Griffiths-Jones, S., Grocock, R. J., van Dongen, S., in plants and animals. Genome Biol. 12, 221 (2011). 22. Aboobaker, A. A., Tomancak, P., Patel, N.,
Bateman, A. & Enright, A. J. miRBase: microRNA 12. Siomi, H. & Siomi, M. C. Posttranscriptional regulation Rubin, G. M. & Lai, E. C. Drosophila microRNAs
sequences, targets and gene nomenclature. Nucleic of microRNA biogenesis in animals. Mol. Cell 38, exhibit diverse spatial expression patterns during
Acids Res. 34, D140–D144 (2006). 323–332 (2010). embryonic development. Proc. Natl Acad. Sci. USA
4. Chiang, H. R. et al. Mammalian microRNAs: 13. Carthew, R. W. & Sontheimer, E. J. Origins and 102, 18017–18022 (2005).
experimental evaluation of novel and previously mechanisms of miRNAs and siRNAs. Cell 136, 23. Kim, Y. K. & Kim, V. N. Processing of intronic
annotated genes. Genes Dev. 24, 992–1009 (2010). 642–655 (2009). microRNAs. EMBO J. 26, 775–783 (2007).
5. Friedman, R. C., Farh, K. K., Burge, C. B. & Bartel, D. P. 14. Yang, J. S. & Lai, E. C. Alternative miRNA biogenesis 24. Ballarino, M. et al. Coupled RNA processing and
Most mammalian mRNAs are conserved targets of pathways and the interpretation of core miRNA transcription of intergenic primary microRNAs. Mol.
microRNAs. Genome Res. 19, 92–105 (2009). pathway mutants. Mol. Cell 43, 892–903 (2011). Cell. Biol. 29, 5632–5638 (2009).
Shows that more than half of all protein-coding 15. Lee, Y. et al. MicroRNA genes are transcribed by RNA 25. Kataoka, N., Fujita, M. & Ohno, M. Functional
genes in mammals have been evolutionarily selected polymerase II. EMBO J. 23, 4051–4060 (2004). association of the Microprocessor complex with the
to maintain pairing with miRNAs, indicating that 16. Cai, X., Hagedorn, C. H. & Cullen, B. R. Human spliceosome. Mol. Cell. Biol. 29, 3243–3254
most of the protein-coding transcriptome is microRNAs are processed from capped, (2009).
regulated by miRNAs. polyadenylated transcripts that can also function as 26. Lee, Y. et al. The nuclear RNase III Drosha initiates
6. Bartel, D. P. MicroRNAs: genomics, biogenesis, mRNAs. RNA 10, 1957–1966 (2004). microRNA processing. Nature 425, 415–419
mechanism, and function. Cell 116, 281–297 17. Xie, Z. et al. Expression of Arabidopsis MIRNA genes. (2003).
(2004). Plant Physiol. 138, 2145–2154 (2005). 27. Denli, A. M., Tops, B. B., Plasterk, R. H., Ketting, R. F.
7. Jones-Rhoades, M. W., Bartel, D. P. & Bartel, B. 18. Ozsolak, F. et al. Chromatin structure analyses identify & Hannon, G. J. Processing of primary microRNAs by
MicroRNAs and their regulatory roles in plants. miRNA promoters. Genes Dev. 22, 3172–3183 the Microprocessor complex. Nature 432, 231–235
Annu. Rev. Plant Biol. 57, 19–53 (2006). (2008). (2004).

NATURE REVIEWS | MOLECULAR CELL BIOLOGY VOLUME 14 | AUGUST 2013 | 485

© 2013 Macmillan Publishers Limited. All rights reserved


REVIEWS

28. Gregory, R. I. et al. The Microprocessor complex 53. Forstemann, K. et al. Normal microRNA maturation 80. Hsu, S.‑H. et al. Essential metabolic, anti-
mediates the genesis of microRNAs. Nature 432, and germ-line stem cell maintenance requires inflammatory, and anti-tumorigenic functions of
235–240 (2004). Loquacious, a double-stranded RNA-binding domain miR‑122 in liver. J. Clin. Invest. 122, 2871–2883
29. Landthaler, M., Yalcin, A. & Tuschl, T. The human protein. PLoS Biol. 3, e236 (2005). (2012).
DiGeorge syndrome critical region gene 8 and its 54. Jiang, F. et al. Dicer‑1 and R3D1‑L catalyze 81. Tsai, W.‑C. et al. MicroRNA‑122 plays a critical role in
D. melanogaster homolog are required for miRNA microRNA maturation in Drosophila. Genes Dev. 19, liver homeostasis and hepatocarcinogenesis. J. Clin.
biogenesis. Curr. Biol. 14, 2162–2167 (2004). 1674–1679 (2005). Invest. 122, 2884–2897 (2012).
30. Han, J. et al. The Drosha–DGCR8 complex in primary 55. Fukunaga, R. et al. Dicer partner proteins tune the 82. Libri, V., Miesen, P., van Rij, R. P. & Buck, A. H.
microRNA processing. Genes Dev. 18, 3016–3027 length of mature miRNAs in flies and mammals. Cell Regulation of microRNA biogenesis and turnover by
(2004). 151, 533–546 (2012). animals and their viruses. Cell. Mol. Life Sci. http://
31. Han, J. et al. Molecular basis for the recognition of 56. Cenik, E. S. et al. Phosphate and R2D2 restrict the dx.doi.org/10.1007/s00018‑012‑1257‑1 (2013).
primary microRNAs by the Drosha–DGCR8 complex. substrate specificity of Dicer‑2, an ATP-driven 83. tenOever, B. R. RNA viruses and the host microRNA
Cell 125, 887–901 (2006). ribonuclease. Mol. Cell 42, 172–184 (2011). machinery. Nature Rev. Microbiol. 11, 169–180
32. Wu, H., Xu, H., Miraglia, L. J. & Crooke, S. T. Human 57. Chendrimada, T. P. et al. TRBP recruits the Dicer (2013).
RNase III is a 160‑kDa protein involved in complex to Ago2 for microRNA processing and gene 84. Pelisson, A., Sarot, E., Payen-Groschene, G. &
preribosomal RNA processing. J. Biol. Chem. 275, silencing. Nature 436, 740–744 (2005). Bucheton, A. A novel repeat-associated small
36957–36965 (2000). 58. Haase, A. D. et al. TRBP, a regulator of cellular PKR interfering RNA-mediated silencing pathway
33. Billy, E., Brondani, V., Zhang, H., Muller, U. & and HIV‑1 virus expression, interacts with Dicer and downregulates complementary sense gypsy
Filipowicz, W. Specific interference with gene functions in RNA silencing. EMBO Rep. 6, 961–967 transcripts in somatic cells of the Drosophila ovary.
expression induced by long, double-stranded RNA in (2005). J. Virol. 81, 1951–1960 (2007).
mouse embryonal teratocarcinoma cell lines. Proc. 59. Lee, Y. et al. The role of PACT in the RNA silencing 85. Horwich, M. D. et al. The Drosophila RNA
Natl Acad. Sci. USA 98, 14428–14433 (2001). pathway. EMBO J. 25, 522–532 (2006). methyltransferase, DmHen1, modifies germline
34. Provost, P. et al. Ribonuclease activity and RNA 60. Lee, H. Y. & Doudna, J. A. TRBP alters human piRNAs and single-stranded siRNAs in RISC.
binding of recombinant human Dicer. EMBO J. 21, precursor microRNA processing in vitro. RNA 18, Curr. Biol. 17, 1265–1272 (2007).
5864–5874 (2002). 2012–2019 (2012). 86. Ameres, S. L., Hung, J. H., Xu, J., Weng, Z. &
35. Lee, Y., Jeon, K., Lee, J. T., Kim, S. & Kim, V. N. 61. Czech, B. & Hannon, G. J. Small RNA sorting: Zamore, P. D. Target RNA-directed tailing and
MicroRNA maturation: stepwise processing and matchmaking for Argonautes. Nature Rev. Genet. 12, trimming purifies the sorting of endo-siRNAs between
subcellular localization. EMBO J. 21, 4663–4670 19–31 (2011). the two Drosophila Argonaute proteins. RNA 17,
(2002). 62. Kawamata, T. & Tomari, Y. Making RISC. Trends 54–63 (2011).
36. Zeng, Y. & Cullen, B. R. Sequence requirements for Biochem. Sci. 35, 368–376 (2010). 87. Park, W., Li, J., Song, R., Messing, J. &
micro RNA processing and function in human cells. 63. Bartel, D. P. MicroRNAs: target recognition and Chen, X. CARPEL FACTORY, a Dicer homolog, and
RNA 9, 112–123 (2003). regulatory functions. Cell 136, 215–233 (2009). HEN1, a novel protein, act in microRNA metabolism in
37. Okada, C. et al. A high-resolution structure of the pre- 64. Grimson, A. et al. MicroRNA targeting specificity in Arabidopsis thaliana. Curr. Biol. 12, 1484–1495
microRNA nuclear export machinery. Science 326, mammals: determinants beyond seed pairing. Mol. (2002).
1275–1279 (2009). Cell 27, 91–105 (2007). 88. Yu, B. et al. Methylation as a crucial step in plant
38. Yi, R., Qin, Y., Macara, I. G. & Cullen, B. R. Exportin‑5 65. Forman, J. J., Legesse-Miller, A. & Coller, H. A. microRNA biogenesis. Science 307, 932–935 (2005).
mediates the nuclear export of pre-microRNAs and A search for conserved sequences in coding regions Reports, for the first time, 2ʹ-O-methyl modification
short hairpin RNAs. Genes Dev. 17, 3011–3016 reveals that the let‑7 microRNA targets Dicer within of small RNAs, which is a stabilizing chemical
(2003). its coding sequence. Proc. Natl Acad. Sci. USA 105, modification that is added to all plant and some
39. Lund, E., Guttinger, S., Calado, A., Dahlberg, J. E. & 14879–14884 (2008). animal small RNA classes.
Kutay, U. Nuclear export of microRNA precursors. 66. Gu, S., Jin, L., Zhang, F., Sarnow, P. & Kay, M. A. 89. Li, J., Yang, Z., Yu, B., Liu, J. & Chen, X. Methylation
Science 303, 95–98 (2004). Biological basis for restriction of microRNA targets to protects miRNAs and siRNAs from a 3ʹ-end uridylation
40. Bohnsack, M. T., Czaplinski, K. & Gorlich, D. the 3ʹ untranslated region in mammalian mRNAs. activity in Arabidopsis. Curr. Biol. 15, 1501–1507
Exportin 5 is a Ran GTP-dependent dsRNA-binding Nature Struct. Mol. Biol. 16, 144–150 (2009). (2005).
protein that mediates nuclear export of pre-miRNAs. 67. Forman, J. J. & Coller, H. A. The code within the code: 90. Yang, Z., Ebright, Y. W., Yu, B. & Chen, X. HEN1
RNA 10, 185–191 (2004). microRNAs target coding regions. Cell Cycle 9, recognizes 21–24 nt small RNA duplexes and deposits
41. Zeng, Y. & Cullen, B. R. Structural requirements for pre- 1533–1541 (2010). a methyl group onto the 2ʹ OH of the 3ʹ terminal
microRNA binding and nuclear export by Exportin 5. 68. Farh, K. K. et al. The widespread impact of nucleotide. Nucleic Acids Res. 34, 667–675 (2006).
Nucleic Acids Res. 32, 4776–4785 (2004). mammalian microRNAs on mRNA repression and 91. Abe, M. et al. WAVY LEAF1, an ortholog of
42. Bernstein, E., Caudy, A. A., Hammond, S. M. & evolution. Science 310, 1817–1821 (2005). Arabidopsis HEN1, regulates shoot development by
Hannon, G. J. Role for a bidentate ribonuclease in the 69. Stark, A., Brennecke, J., Bushati, N., Russell, R. B. & maintaining microRNA and trans-acting small
initiation step of RNA interference. Nature 409, Cohen, S. M. Animal microRNAs confer robustness to interfering RNA accumulation in rice. Plant Physiol.
363–366 (2001). gene expression and have a significant impact on 154, 1335–1346 (2010).
43. Hutvágner, G. et al. A cellular function for the RNA- 3ʹUTR evolution. Cell 123, 1133–1146 (2005). 92. Huang, Y. et al. Structural insights into mechanisms of
interference enzyme Dicer in the maturation of the 70. Lewis, B. P., Burge, C. B. & Bartel, D. P. Conserved the small RNA methyltransferase HEN1. Nature 461,
let‑7 small temporal RNA. Science 293, 834–838 seed pairing, often flanked by adenosines, indicates 823–827 (2009).
(2001). that thousands of human genes are microRNA targets. 93. Vilkaitis, G., Plotnikova, A. & Klimasauskas, S. Kinetic
44. Grishok, A. et al. Genes and mechanisms related to Cell 120, 15–20 (2005). and functional analysis of the small RNA
RNA interference regulate expression of the small 71. Krutzfeldt, J. et al. Silencing of microRNAs in vivo with methyltransferase HEN1: the catalytic domain is
temporal RNAs that control C. elegans developmental ‘antagomirs’. Nature 438, 685–689 (2005). essential for preferential modification of duplex RNA.
timing. Cell 106, 23–34 (2001). 72. Ameres, S. L. et al. Target RNA-directed trimming and RNA 16, 1935–1942 (2010).
45. Ketting, R. F. et al. Dicer functions in RNA interference tailing of small silencing RNAs. Science 328, 94. Chatterjee, S. & Grosshans, H. Active turnover
and in synthesis of small RNA involved in 1534–1539 (2010). modulates mature microRNA activity in Caenorhabditis
developmental timing in C. elegans. Genes Dev. 15, References 71, 72, 94 and 96 show that target- elegans. Nature 461, 546–549 (2009).
2654–2659 (2001). RNA binding can alter miRNA stability in mammals, 95. Song, E. et al. Sustained small interfering RNA-
46. Lee, Y. S. et al. Distinct roles for Drosophila Dicer‑1 flies and worms. References 72 and 94 help mediated human immunodeficiency virus type 1
and Dicer‑2 in the siRNA/miRNA silencing pathways. explain how the presence or absence of targets inhibition in primary macrophages. J. Virol. 77,
Cell 117, 69–81 (2004). may destabilize miRNAs. 7174–7181 (2003).
47. Zhang, H., Kolb, F. A., Brondani, V., Billy, E. & 73. Baccarini, A. et al. Kinetic analysis reveals the fate of a 96. Chatterjee, S., Fasler, M., Bussing, I. & Grosshans, H.
Filipowicz, W. Human Dicer preferentially cleaves microRNA following target regulation in mammalian Target-mediated protection of endogenous microRNAs
dsRNAs at their termini without a requirement for cells. Curr. Biol. 21, 369–376 (2011). in C. elegans. Dev. Cell 20, 388–396 (2011).
ATP. EMBO J. 21, 5875–5885 (2002). 74. Xie, J. et al. Long-term, efficient inhibition of 97. Bail, S. et al. Differential regulation of microRNA
48. Zhang, H., Kolb, F. A., Jaskiewicz, L., Westhof, E. & microRNA function in mice using rAAV vectors. Nature stability. RNA 16, 1032–1039 (2010).
Filipowicz, W. Single processing center models for Methods 9, 403–409 (2012). 98. Berezikov, E. et al. Deep annotation of Drosophila
human Dicer and bacterial RNase III. Cell 118, 57–68 75. Yan, K. S. et al. Structure and conserved RNA binding melanogaster microRNAs yields insights into their
(2004). of the PAZ domain. Nature 426, 468–474 (2003). processing, modification, and emergence. Genome
49. Pase, L. et al. miR‑451 regulates zebrafish erythroid 76. Lingel, A., Simon, B., Izaurralde, E. & Sattler, M. Res. 21, 203–215 (2011).
maturation in vivo via its target gata2. Blood 113, Structure and nucleic-acid binding of the Drosophila 99. Ruby, J. G. et al. Evolution, biogenesis, expression,
1794–1804 (2009). Argonaute 2 PAZ domain. Nature 426, 465–469 and target predictions of a substantially expanded
50. Cheloufi, S., Dos Santos, C. O., Chong, M. M. & (2003). set of Drosophila microRNAs. Genome Res. 17,
Hannon, G. J. A dicer-independent miRNA biogenesis 77. Lingel, A., Simon, B., Izaurralde, E. & Sattler, M. NMR 1850–1864 (2007).
pathway that requires Ago catalysis. Nature 465, assignment of the Drosophila Argonaute2 PAZ 100. Wu, H. et al. miRNA profiling of naïve, effector and
584–589 (2010). domain. J. Biomol. NMR 29, 421–422 (2004). memory CD8 T cells. PLoS ONE 2, e1020 (2007).
51. Yang, J. S. et al. Conserved vertebrate mir‑451 78. Lingel, A., Simon, B., Izaurralde, E. & Sattler, M. 101. Morin, R. D. et al. Application of massively parallel
provides a platform for Dicer-independent, Nucleic acid 3ʹ-end recognition by the Argonaute2 sequencing to microRNA profiling and discovery in
Ago2‑mediated microRNA biogenesis. Proc. Natl PAZ domain. Nature Struct. Mol. Biol. 11, 576–577 human embryonic stem cells. Genome Res. 18,
Acad. Sci. USA 107, 15163–15168 (2010). (2004). 610–621 (2008).
52. Saito, K., Ishizuka, A., Siomi, H. & Siomi, M. C. 79. Wang, Y., Sheng, G., Juranek, S., Tuschl, T. & 102. Seitz, H., Ghildiyal, M. & Zamore, P. D. Argonaute
Processing of pre-microRNAs by the Dicer‑1– Patel, D. J. Structure of the guide-strand-containing loading improves the 5ʹ precision of both microRNAs
Loquacious complex in Drosophila cells. PLoS Biol. 3, Argonaute silencing complex. Nature 456, 209–213 and their miRNA* strands in flies. Curr. Biol. 18,
e235 (2005). (2008). 147–151 (2008).

486 | AUGUST 2013 | VOLUME 14 www.nature.com/reviews/molcellbio

© 2013 Macmillan Publishers Limited. All rights reserved


REVIEWS

103. Rajagopalan, R., Vaucheret, H., Trejo, J. & Bartel, D. P. 128. Heo, I. et al. TUT4 in concert with Lin28 suppresses 151. Parker, J. S., Parizotto, E. A., Wang, M., Roe, S. M. &
A diverse and evolutionarily fluid set of microRNAs in microRNA biogenesis through pre-microRNA Barford, D. Enhancement of the seed-target
Arabidopsis thaliana. Genes Dev. 20, 3407–3425 uridylation. Cell 138, 696–708 (2009). recognition step in RNA silencing by a PIWI/MID
(2006). 129. Hagan, J. P., Piskounova, E. & Gregory, R. I. Lin28 domain protein. Mol. Cell 33, 204–214 (2009).
104. Westholm, J. O., Ladewig, E., Okamura, K., Robine, N. recruits the TUTase Zcchc11 to inhibit let‑7 maturation 152. Schirle, N. T. & MacRae, I. J. The crystal structure of
& Lai, E. C. Common and distinct patterns of terminal in mouse embryonic stem cells. Nature Struct. Mol. human Argonaute2. Science 336, 1037–1040 (2012).
modifications to mirtrons and canonical microRNAs. Biol. 16, 1021–1025 (2009). 153. Elkayam, E. et al. The structure of human Argonaute‑2
RNA 18, 177–192 (2012). References 128 and 129 report that LIN28‑bound in complex with miR‑20a. Cell 150, 100–110 (2012).
105. Czech, B. et al. Hierarchical rules for Argonaute pre-miRNAs act as substrates for the TNTase 154. Nakanishi, K., Weinberg, D. E., Bartel, D. P. &
loading in Drosophila. Mol. Cell 36, 445–456 ZCCHC11, thereby establishing the mechanism for Patel, D. J. Structure of yeast Argonaute with guide
(2009). inhibition of pre-miRNA processing by uridylation. RNA. Nature 486, 368–374 (2012).
106. Okamura, K., Liu, N. & Lai, E. C. Distinct mechanisms 130. Heo, I. et al. Lin28 mediates the terminal uridylation 155. Wang, Y. et al. Nucleation, propagation and cleavage
for microRNA strand selection by Drosophila of let‑7 precursor microRNA. Mol. Cell 32, 276–284 of target RNAs in Ago silencing complexes. Nature
Argonautes. Mol. Cell 36, 431–444 (2009). (2008). 461, 754–761 (2009).
107. Ghildiyal, M., Xu, J., Seitz, H., Weng, Z. & 131. Joo, C., Fareh, M. & Kim, V. N. Bringing single- 156. Chi, S. W., Hannon, G. J. & Darnell, R. B. An
Zamore, P. D. Sorting of Drosophila small silencing molecule spectroscopy to macromolecular protein alternative mode of microRNA target recognition.
RNAs partitions microRNA* strands into the RNA complexes. Trends Biochem. Sci. 38, 30–37 (2013). Nature Struct. Mol. Biol. 19, 321–327 (2012).
interference pathway. RNA 16, 43–56 (2010). 132. Chang, H.‑M., Triboulet, R., Thornton, J. E. & 157. Long, D. et al. Potent effect of target structure on
108. Ruby, J. G. et al. Large-scale sequencing reveals Gregory, R. I. A role for the Perlman syndrome microRNA function. Nature Struct. Mol. Biol. 14,
21U‑RNAs and additional microRNAs and endogenous exonuclease Dis3l2 in the Lin28–let‑7 pathway. 287–294 (2007).
siRNAs in C. elegans. Cell 127, 1193–1207 (2006). Nature 497, 244–248 (2013). 158. Nielsen, C. B. et al. Determinants of targeting by
109. Babiarz, J. E., Ruby, J. G., Wang, Y., Bartel, D. P. & 133. Heo, I. et al. Mono-uridylation of pre-microRNA as a endogenous and exogenous microRNAs and siRNAs.
Blelloch, R. Mouse ES cells express endogenous key step in the biogenesis of group II let‑7 microRNAs. RNA 13, 1894–1910 (2007).
shRNAs, siRNAs, and other Microprocessor- Cell 151, 521–532 (2012). 159. Kertesz, M., Iovino, N., Unnerstall, U., Gaul, U. &
independent, Dicer-dependent small RNAs. Genes Reveals that monouridylation of specific pre- Segal, E. The role of site accessibility in microRNA
Dev. 22, 2773–2785 (2008). miRNAs enhances Dicer-mediated processing by target recognition. Nature Genet. 39, 1278–1284
110. Glazov, E. A. et al. A microRNA catalog of the restoring a canonical two nucleotide 3ʹ overhang, (2007).
developing chicken embryo identified by a deep which is required for efficient substrate recognition 160. Tafer, H. et al. The impact of target site accessibility on
sequencing approach. Genome Res. 18, 957–964 by Dicer. the design of effective siRNAs. Nature Biotech. 26,
(2008). 134. Jones, M. R. et al. Zcchc11‑dependent uridylation of 578–583 (2008).
111. Ebhardt, H. A. et al. Meta-analysis of small RNA- microRNA directs cytokine expression. Nature Cell 161. Obernosterer, G., Tafer, H. & Martinez, J. Target site
sequencing errors reveals ubiquitous post- Biol. 11, 1157–1163 (2009). effects in the RNA interference and microRNA
transcriptional RNA modifications. Nucleic Acids Res. 135. Katoh, T. et al. Selective stabilization of mammalian pathways. Biochem. Soc. Trans. 36, 1216–1219
37, 2461–2470 (2009). microRNAs by 3ʹ adenylation mediated by the (2008).
112. Blow, M. J. et al. RNA editing of human microRNAs. cytoplasmic poly(A) polymerase GLD‑2. Genes Dev. 162. Kedde, M. et al. RNA-binding protein Dnd1 inhibits
Genome Biol. 7, R27 (2006). 23, 433–438 (2009). microRNA access to target mRNA. Cell 131,
113. Landgraf, P. et al. A mammalian microRNA expression 136. Burroughs, A. M. et al. A comprehensive survey of 1273–1286 (2007).
atlas based on small RNA library sequencing. Cell 3ʹ animal miRNA modification events and a possible 163. Bhattacharyya, S. N., Habermacher, R., Martine, U.,
129, 1401–1414 (2007). role for 3ʹ adenylation in modulating miRNA targeting Closs, E. I. & Filipowicz, W. Relief of microRNA-
114. Kawahara, Y. et al. Redirection of silencing targets by effectiveness. Genome Res. 20, 1398–1410 (2010). mediated translational repression in human cells
adenosine-to‑inosine editing of miRNAs. Science 315, 137. Lu, S., Sun, Y. H. & Chiang, V. L. Adenylation of plant subjected to stress. Cell 125, 1111–1124 (2006).
1137–1140 (2007). miRNAs. Nucleic Acids Res. 37, 1878–1885 (2009). 164. Huang, J. et al. Derepression of microRNA-mediated
115. Hundley, H. A. & Bass, B. L. ADAR editing in double- 138. Zhao, Y. et al. The Arabidopsis nucleotidyl transferase protein translation inhibition by apolipoprotein B
stranded UTRs and other noncoding RNA sequences. HESO1 uridylates unmethylated small RNAs to trigger mRNA-editing enzyme catalytic polypeptide-like 3G
Trends Biochem. Sci. 35, 377–383 (2010). their degradation. Curr. Biol. 22, 689–694 (2012). (APOBEC3G) and its family members. J. Biol. Chem.
116. Knight, S. W. & Bass, B. L. The role of RNA editing by 139. Ren, G., Chen, X. & Yu, B. Uridylation of miRNAs by 282, 33632–33640 (2007).
ADARs in RNAi. Mol. Cell 10, 809–817 (2002). HEN1 SUPPRESSOR1 in Arabidopsis. Curr. Biol. 22, 165. Mishima, Y. et al. Zebrafish miR‑1 and miR‑133
117. Habig, J. W., Aruscavage, P. J. & Bass, B. L. In 695–700 (2012). shape muscle gene expression and regulate
C. elegans, high levels of dsRNA allow RNAi in the 140. Ibrahim, F. et al. Uridylation of mature miRNAs and sarcomeric actin organization. Genes Dev. 23,
absence of RDE‑4. PLoS ONE 3, e4052 (2008). siRNAs by the MUT68 nucleotidyltransferase 619–632 (2009).
118. Wu, D., Lamm, A. T. & Fire, A. Z. Competition promotes their degradation in Chlamydomonas. Proc. 166. Elcheva, I., Goswami, S., Noubissi, F. K. &
between ADAR and RNAi pathways for an extensive Natl Acad. Sci. USA http://dx.doi.org/10.1073/ Spiegelman, V. S. CRD‑BP protects the coding region
class of RNA targets. Nature Struct. Mol. Biol. 18, pnas.0912632107 (2010). of βTrCP1 mRNA from miR‑183‑mediated
1094–1101 (2011). 141. Ibrahim, F., Rohr, J., Jeong, W.‑J., Hesson, J. & degradation. Mol. Cell 35, 240–246 (2009).
119. Liu, N. et al. The exoribonuclease Nibbler controls Cerutti, H. Untemplated oligoadenylation promotes 167. Goswami, S. et al. MicroRNA‑340‑mediated
3ʹ end processing of microRNAs in Drosophila. Curr. degradation of RISC-cleaved transcripts. Science 314, degradation of microphthalmia-associated
Biol. 21, 1888–1893 (2011). 1893 (2006). transcription factor mRNA is inhibited by the coding
120. Han, B. W., Hung, J. H., Weng, Z., Zamore, P. D. & 142. Shen, B. & Goodman, H. M. Uridine addition after region determinant-binding protein. J. Biol. Chem.
Ameres, S. L. The 3ʹ-to‑5ʹ exoribonuclease Nibbler microRNA-directed cleavage. Science 306, 997 (2004). 285, 20532–20540 (2010).
shapes the 3ʹ ends of microRNAs bound to Drosophila 143. Tang, G., Reinhart, B. J., Bartel, D. P. & Zamore, P. D. 168. Jafarifar, F., Yao, P., Eswarappa, S. M. & Fox, P. L.
Argonaute1. Curr. Biol. 21, 1878–1887 (2011). A biochemical framework for RNA silencing in plants. Repression of VEGFA by CA‑rich element-binding
References119 and 120 describe the enzyme Genes Dev. 17, 49–63 (2003). microRNAs is modulated by hnRNP L. EMBO J. 30,
responsible for generating much of the 3ʹ 144. Mallory, A. C. et al. MicroRNA control of PHABULOSA 1324–1334 (2011).
heterogeneity observed in D. melanogaster in leaf development: importance of pairing to the 169. Toledano, H., D’Alterio, C., Czech, B., Levine, E. &
miRNAs, revealing that exonucleolytic trimming of microRNA 5ʹ region. EMBO J. 23, 3356–3364 Jones, D. L. The let‑7–Imp axis regulates ageing of
miRNAs after assembly into AGO proteins (2004). the Drosophila testis stem-cell niche. Nature 485,
promotes the formation of active complexes. 145. Haley, B. & Zamore, P. D. Kinetic analysis of the RNAi 605–610 (2012).
121. Ketting, R. F., Haverkamp, T. H., van Luenen, H. G. & enzyme complex. Nature Struct. Mol. Biol. 11, 170. Shin, C. et al. Expanding the microRNA targeting
Plasterk, R. H. Mut‑7 of C. elegans, required for 599–606 (2004). code: functional sites with centered pairing. Mol. Cell
transposon silencing and RNA interference, is a 146. Ameres, S. L., Martinez, J. & Schroeder, R. Molecular 38, 789–802 (2010).
homolog of Werner syndrome helicase and RNaseD. basis for target RNA recognition and cleavage by 171. Llave, C., Xie, Z., Kasschau, K. D. & Carrington, J. C.
Cell 99, 133–141 (1999). human RISC. Cell 130, 101–112 (2007). Cleavage of Scarecrow-like mRNA targets directed by
122. Ketting, R. F. & Plasterk, R. H. A genetic link between 147. Wee, L., Flores-Jasso, C. F., Salomon, W. & a class of Arabidopsis miRNA. Science 297,
co‑suppression and RNA interference in C. elegans. Zamore, P. D. Argonaute divides its RNA guide into 2053–2056 (2002).
Nature 404, 296–298 (2000). domains with distinct functions and RNA-binding 172. Dunoyer, P., Lecellier, C. H., Parizotto, E. A.,
123. Grishok, A., Tabara, H. & Mello, C. C. Genetic properties. Cell 151, 1055–1067 (2012). Himber, C. & Voinnet, O. Probing the microRNA and
requirements for inheritance of RNAi in C. elegans. Reports detailed kinetic analyses of AGO-directed small interfering RNA pathways with virus-encoded
Science 287, 2494–2497 (2000). small RNA–target RNA interactions in flies and suppressors of RNA silencing. Plant Cell 16,
124. Sijen, T. & Plasterk, R. H. Transposon silencing in the mice. Establishes the biochemical basis for siRNA 1235–1250 (2004).
Caenorhabditis elegans germ line by natural RNAi. and miRNA function, which is determined by the 173. Souret, F. F., Kastenmayer, J. P. & Green, P. J. AtXRN4
Nature 426, 310–314 (2003). relative abundance of small RNAs and their degrades mRNA in Arabidopsis and its substrates
125. Rissland, O. S., Mikulasova, A. & Norbury, C. J. targets. include selected miRNA targets. Mol. Cell 15,
Efficient RNA polyuridylation by noncanonical poly(A) 148. Song, J. J., Smith, S. K., Hannon, G. J. & Joshua-Tor, L. 173–183 (2004).
polymerases. Mol. Cell. Biol. 27, 3612–3624 Crystal structure of Argonaute and its implications for 174. German, M. A. et al. Global identification of
(2007). RISC slicer activity. Science 305, 1434–1437 (2004). microRNA–target RNA pairs by parallel analysis of
126. Dreyfus, M. & Regnier, P. The poly(A) tail of mRNAs: 149. Parker, J. S., Roe, S. M. & Barford, D. Structural RNA ends. Nature Biotech. 26, 941–946 (2008).
bodyguard in eukaryotes, scavenger in bacteria. Cell insights into mRNA recognition from a PIWI domain– 175. Addo-Quaye, C., Eshoo, T. W., Bartel, D. P. &
111, 611–613 (2002). siRNA guide complex. Nature 434, 663–666 (2005). Axtell, M. J. Endogenous siRNA and miRNA
127. Scott, D. D. & Norbury, C. J. RNA decay via 3ʹ 150. Wang, Y. et al. Structure of an Argonaute silencing targets identified by sequencing of the
uridylation. Biochim Biophys Acta 1829, 654–665 complex with a seed-containing guide DNA and target Arabidopsis degradome. Curr. Biol. 18, 758–762
(2013). RNA duplex. Nature 456, 921–926 (2008). (2008).

NATURE REVIEWS | MOLECULAR CELL BIOLOGY VOLUME 14 | AUGUST 2013 | 487

© 2013 Macmillan Publishers Limited. All rights reserved


REVIEWS

176. Jones-Rhoades, M. W. & Bartel, D. P. Computational 200. Zdanowicz, A. et al. Drosophila miR2 primarily targets 225. Poliseno, L. et al. A coding-independent function of
identification of plant microRNAs and their targets, the m7GpppN cap structure for translational gene and pseudogene mRNAs regulates tumour
including a stress-induced miRNA. Mol. Cell 14, repression. Mol. Cell 35, 881–888 (2009). biology. Nature 465, 1033–1038 (2010).
787–799 (2004). 201. Fabian, M. R. et al. Mammalian miRNA RISC recruits 226. Karreth, F. A. et al. In vivo identification of tumor-
177. Lanet, E. et al. Biochemical evidence for translational CAF1 and PABP to affect PABP-dependent suppressive PTEN ceRNAs in an oncogenic BRAF-
repression by Arabidopsis microRNAs. Plant Cell 21, deadenylation. Mol. Cell 35, 868–880 (2009). induced mouse model of melanoma. Cell 147,
1762–1768 (2009). 202. Djuranovic, S., Nahvi, A. & Green, R. miRNA-mediated 382–395 (2011).
178. Brodersen, P. et al. Widespread translational gene silencing by translational repression followed by 227. Cesana, M. et al. A long noncoding RNA controls
inhibition by plant miRNAs and siRNAs. Science 320, mRNA deadenylation and decay. Science 336, muscle differentiation by functioning as a competing
1185–1190 (2008). 237–240 (2012). endogenous RNA. Cell 147, 358–369 (2011).
References 177 and 178 propose that, despite 203. Bruno, I. G. et al. Identification of a microRNA that 228. Sumazin, P. et al. An extensive microRNA-mediated
the large number of extensively complementary activates gene expression by repressing nonsense- network of RNA–RNA interactions regulates
cleavage targets for plant miRNAs, translational mediated RNA decay. Mol. Cell 42, 500–510 (2011). established oncogenic pathways in glioblastoma. Cell
repression may be a widespread mechanism by 204. Place, R. F., Li, L. C., Pookot, D., Noonan, E. J. & 147, 370–381 (2011).
which plant miRNAs repress gene expression. Dahiya, R. MicroRNA‑373 induces expression of 229. Lim, L. P. et al. The microRNAs of Caenorhabditis
179. Yekta, S., Shih, I. H. & Bartel, D. P. MicroRNA-directed genes with complementary promoter sequences. elegans. Genes Dev. 17, 991–1008 (2003).
cleavage of HOXB8 mRNA. Science 304, 594–596 Proc. Natl Acad. Sci. USA 105, 1608–1613 (2008). 230. Chang, J. et al. miR‑122, a mammalian liver-specific
(2004). 205. Li, L. C. et al. Small dsRNAs induce transcriptional microRNA, is processed from hcr mRNA and
180. Davis, E. et al. RNAi-mediated allelic trans-interaction activation in human cells. Proc. Natl Acad. Sci. USA may downregulate the high affinity cationic amino
at the imprinted Rtl1/Peg11 locus. Curr. Biol. 15, 103, 17337–17342 (2006). acid transporter CAT‑1. RNA Biol. 1, 106–113
743–749 (2005). 206. Schwartz, J. C. et al. Antisense transcripts are targets (2004).
181. Karginov, F. V. et al. Diverse endonucleolytic cleavage for activating small RNAs. Nature Struct. Mol. Biol. 231. Mullokandov, G. et al. High-throughput assessment
sites in the mammalian transcriptome depend upon 15, 842–848 (2008). of microRNA activity and function using microRNA
microRNAs, Drosha, and additional nucleases. Mol. 207. Jangra, R. K., Yi, M. & Lemon, S. M. DDX6 (Rck/p54) sensor and decoy libraries. Nature Methods 9,
Cell 38, 781–788 (2010). is required for efficient hepatitis C virus replication but 840–846 (2012).
182. Bagga, S. et al. Regulation by let‑7 and lin‑4 miRNAs not for internal ribosome entry site-directed 232. Care, A. et al. MicroRNA‑133 controls cardiac
results in target mRNA degradation. Cell 122, translation. J. Virol. 84, 6810–6824 (2010). hypertrophy. Nature Med. 13, 613–618 (2007).
553–563 (2005). 208. Jopling, C. L., Yi, M., Lancaster, A. M., Lemon, S. M. & 233. Gentner, B. et al. Stable knockdown of microRNA in
183. Jing, Q. et al. Involvement of microRNA in AU‑rich Sarnow, P. Modulation of hepatitis C virus RNA vivo by lentiviral vectors. Nature Methods 6, 63–66
element-mediated mRNA instability. Cell 120, abundance by a liver-specific microRNA. Science 309, (2009).
623–634 (2005). 1577–1581 (2005). 234. Haraguchi, T., Ozaki, Y. & Iba, H. Vectors expressing
184. Lim, L. P. et al. Microarray analysis shows that some 209. Lanford, R. E. et al. Therapeutic silencing of efficient RNA decoys achieve the long-term
microRNAs downregulate large numbers of target microRNA‑122 in primates with chronic hepatitis C suppression of specific microRNA activity in
mRNAs. Nature 433, 769–773 (2005). virus infection. Science 327, 198–201 (2010). mammalian cells. Nucleic Acids Res. 37, e43
185. Giraldez, A. J. et al. Zebrafish miR‑430 promotes 210. Vasudevan, S., Tong, Y. & Steitz, J. A. Switching from (2009).
deadenylation and clearance of maternal mRNAs. repression to activation: microRNAs can up‑regulate 235. Cazalla, D., Yario, T. & Steitz, J. Down-regulation of a
Science 312, 75–79 (2006). translation. Science 318, 1931–1934 (2007). host microRNA by a herpesvirus saimiri noncoding
186. Rehwinkel, J. et al. Genome-wide analysis of mRNAs 211. Vasudevan, S. & Steitz, J. A. AU‑rich-element- RNA. Science 328, 1563–1566 (2010).
regulated by Drosha and Argonaute proteins in mediated upregulation of translation by FXR1 and 236. Marcinowski, L. et al. Degradation of cellular
Drosophila melanogaster. Mol. Cell. Biol. 26, Argonaute 2. Cell 128, 1105–1118 (2007). mir‑27 by a novel, highly abundant viral transcript is
2965–2975 (2006). 212. Orom, U. A., Nielsen, F. C. & Lund, A. H. important for efficient virus replication in vivo. PLoS
187. Wu, L., Fan, J. & Belasco, J. G. MicroRNAs direct rapid MicroRNA‑10a binds the 5ʹUTR of ribosomal protein Pathog. 8, e1002510 (2012).
deadenylation of mRNA. Proc. Natl Acad. Sci. USA mRNAs and enhances their translation. Mol. Cell 30, 237. Libri, V. et al. Murine cytomegalovirus encodes a
103, 4034–4039 (2006). 460–471 (2008). miR‑27 inhibitor disguised as a target. Proc. Natl
188. Baek, D. et al. The impact of microRNAs on protein 213. Li, X., Cassidy, J. J., Reinke, C. A., Fischboeck, S. & Acad. Sci. USA 109, 279–284 (2012).
output. Nature 455, 64–71 (2008). Carthew, R. W. A microRNA imparts robustness 238. Salzman, J., Gawad, C., Wang, P. L., Lacayo, N. &
189. Selbach, M. et al. Widespread changes in protein against environmental fluctuation during Brown, P. O. Circular RNAs are the predominant
synthesis induced by microRNAs. Nature 455, 58–63 development. Cell 137, 273–282 (2009). transcript isoform from hundreds of human genes in
(2008). 214. Miska, E. A. et al. Most Caenorhabditis elegans diverse cell types. PLoS ONE 7, e30733 (2012).
190. Guo, H., Ingolia, N. T., Weissman, J. S. & Bartel, D. P. microRNAs are individually not essential for 239. Hansen, T. B. et al. Natural RNA circles function as
Mammalian microRNAs predominantly act to decrease development or viability. PLoS Genet. 3, e215 (2007). efficient microRNA sponges. Nature 495, 384–388
target mRNA levels. Nature 466, 835–840 (2010). 215. Seitz, H. Redefining microRNA targets. Curr. Biol. 19, (2013).
191. Ingolia, N. T., Ghaemmaghami, S., Newman, J. R. & 870–873 (2009). 240. Memczak, S. et al. Circular RNAs are a large class of
Weissman, J. S. Genome-wide analysis in vivo of Hypothesizes that in animals many endogenous animal RNAs with regulatory potency. Nature 495,
translation with nucleotide resolution using ribosome miRNA target site-containing transcripts (‘titrating’ 333–338 (2013).
profiling. Science 324, 218–223 (2009). targets) may function to titrate miRNA activity 241. Hwang, H. W., Wentzel, E. A. & Mendell, J. T. A
192. Hendrickson, D. G. et al. Concordant regulation of rather than serving as targets for regulation. hexanucleotide element directs microRNA nuclear
translation and mRNA abundance for hundreds of 216. Meister, G., Landthaler, M., Dorsett, Y. & Tuschl, T. import. Science 315, 97–100 (2007).
targets of a human microRNA. PLoS Biol. 7, Sequence-specific inhibition of microRNA- and siRNA- 242. Gantier, M. P. et al. Analysis of microRNA turnover in
e1000238 (2009). induced RNA silencing. RNA 10, 544–550 (2004). mammalian cells following Dicer1 ablation. Nucleic
193. Bazzini, A. A., Lee, M. T. & Giraldez, A. J. Ribosome 217. Hutvagner, G., Simard, M. J., Mello, C. C. & Acids Res. 39, 5692–5703 (2011).
profiling shows that miR‑430 reduces translation Zamore, P. D. Sequence-specific inhibition of small 243. van Rooij, E. et al. Control of stress-dependent cardiac
before causing mRNA decay in zebrafish. Science 336, RNA function. PLoS Biol. 2, e98 (2004). growth and gene expression by a microRNA. Science
233–237 (2012). 218. Ebert, M. S., Neilson, J. R. & Sharp, P. A. MicroRNA 316, 575–579 (2007).
Proposes, together with reference 190, that animal sponges: competitive inhibitors of small RNAs in 244. Rissland, O. S., Hong, S. J. & Bartel, D. P. MicroRNA
miRNAs predominantly function by triggering mammalian cells. Nature Methods 4, 721–726 (2007). destabilization enables dynamic regulation of the
mRNA decay. Reference 193 also suggests that 219. Loya, C. M., Lu, C. S., Van Vactor, D. & Fulga, T. A. miR‑16 family in response to cell-cycle changes. Mol.
translational repression may precede mRNA decay. Transgenic microRNA inhibition with spatiotemporal Cell 43, 993–1004 (2011).
194. Rehwinkel, J., Behm-Ansmant, I., Gatfield, D. & specificity in intact organisms. Nature Methods 6, 245. Zhang, Z. et al. Uracils at nucleotide position 9–11 are
Izaurralde, E. A crucial role for GW182 and the 897–903 (2009). required for the rapid turnover of miR‑29 family.
DCP1:DCP2 decapping complex in miRNA-mediated References 216–219 describe strategies to Nucleic Acids Res. 39, 4387–4395 (2011).
gene silencing. RNA 11, 1640–1647 (2005). competitively inhibit miRNAs for loss-of-function
195. Behm-Ansmant, I. et al. mRNA degradation by studies. Acknowledgements
miRNAs and GW182 requires both CCR4:NOT 220. Ebert, M. S. & Sharp, P. A. MicroRNA sponges: The authors thank the members of the Zamore and Ameres
deadenylase and DCP1:DCP2 decapping complexes. progress and possibilities. RNA 16, 2043–2050 laboratories for helpful discussions and comments. The
Genes Dev. 20, 1885–1898 (2006). (2010). Ameres laboratory is funded by the Austrian Academy of
196. Braun, J. E., Huntzinger, E., Fauser, M. & Izaurralde, E. 221. Franco-Zorrilla, J. M. et al. Target mimicry provides a Sciences and the Austrian Federal Ministry of Economy,
GW182 proteins directly recruit cytoplasmic new mechanism for regulation of microRNA activity. Family and Youth (BMFWJ).
deadenylase complexes to miRNA targets. Mol. Cell Nature Genet. 39, 1033–1037 (2007).
44, 120–133 (2011). 222. Todesco, M., Rubio-Somoza, I., Paz-Ares, J. & Competing interests statement
197. Thermann, R. & Hentze, M. W. Drosophila miR2 Weigel, D. A collection of target mimics for The authors declare no competing financial interests.
induces pseudo-polysomes and inhibits translation comprehensive analysis of microRNA function in
initiation. Nature 447, 875–878 (2007). Arabidopsis thaliana. PLoS Genet. 6, e1001031
198. Iwasaki, S., Kawamata, T. & Tomari, Y. Drosophila (2010). FURTHER INFORMATION
Argonaute1 and Argonaute2 employ distinct 223. Salmena, L., Poliseno, L., Tay, Y., Kats, L. & Phillip D. Zamore’s homepage:
mechanisms for translational repression. Mol. Cell 34, Pandolfi, P. P. A ceRNA hypothesis: the Rosetta stone http://www.umassmed.edu/zamore/index.aspx
58–67 (2009). of a hidden RNA language? Cell 146, 353–358 Stefan L. Ameres’ homepage:
199. Mathonnet, G. et al. MicroRNA inhibition of (2011). http://www.imba.oeaw.ac.at/research/stefan-ameres
translation initiation in vitro by targeting the cap- 224. Tay, Y. et al. Coding-independent regulation of the The microRNA database: http://www.mirbase.org
binding complex eIF4F. Science 317, 1764–1767 tumor suppressor PTEN by competing endogenous ALL LINKS ARE ACTIVE IN THE ONLINE PDF
(2007). mRNAs. Cell 147, 344–357 (2011).

488 | AUGUST 2013 | VOLUME 14 www.nature.com/reviews/molcellbio

© 2013 Macmillan Publishers Limited. All rights reserved

You might also like