Reviews

R N A P R O C E S S I N G A N D M O D I F I C AT I O N S

Regulation of microRNA function

in animals
Luca F. R. Gebert and Ian J. MacRae*
Abstract | Since their serendipitous discovery in nematodes, microRNAs (miRNAs) have emerged
as key regulators of biological processes in animals. These small RNAs form complex networks
that regulate cell differentiation, development and homeostasis. Deregulation of miRNA function
is associated with an increasing number of human diseases, particularly cancer. Recent
discoveries have expanded our understanding of the control of miRNA function. Here, we review
the mechanisms that modulate miRNA activity, stability and cellular localization through
alternative processing and maturation, sequence editing, post-​translational modifications of
Argonaute proteins, viral factors, transport from the cytoplasm and regulation of miRNA–target
interactions. We conclude by discussing intriguing, unresolved research questions.

Isomirs
MicroRNAs (miRNAs) are short non-​coding RNAs miRNAs are prognostic markers21,22 or potential targets
Variant forms of a canonical (ncRNAs) of ~22 nucleotides that mediate gene silenc- for novel cancer therapies23. Plant miRNAs24,25, which
miRNA, generated by ing by guiding Argonaute (AGO) proteins to target sites are not discussed here, differ considerably from animal
alternative cleavage during in the 3′ untranslated region (UTR) of mRNAs. AGOs miRNAs in their evolution26, biogenesis27 and function28
biogenesis, RNA editing or
constitute a large family of proteins that use single-​ (reviewed in refs29,30).
non-​templated nucleotide
addition.
stranded small nucleic acids as guides to complemen- In this Review, we first provide an overview of
tary sequences in RNA or DNA targeted for silencing1. miRNA function and regulation. We then discuss in
The miRNA-​loaded AGO forms the targeting module detail the regulation of miRNA function through the
of the miRNA-​induced silencing complex (miRISC), formation of isomirs, the addition of non-​templated
which promotes translation repression and degradation nucleotides, post-​translational modification (PTM) of
of targeted mRNAs2. The first miRNA was discovered miRNA-​binding proteins, miRNA sequestration, modu-
in Caenorhabditis elegans as a short RNA produced by lation by viral factors, transport from the cytoplasm and
the lin-4 gene, which post-​transcriptionally represses the the regulation of interactions between miRNAs and their
lin-14 mRNA3,4. Such small RNAs were widely thought target mRNAs (miRNA–target interactions).
to be unique to nematodes, until they were shown to
be abundant in diverse animal phyla5. This new class of Overview of miRNA function
regulators was subsequently named microRNAs6–8. miRNA genes are transcribed into primary miRNA (pri-​
miRNAs are involved in virtually every cellular miRNA) transcripts and undergo multistep biogenesis,
process and are essential for animal development, cell in which they are processed first into pre-​miRNAs and
differentiation and homeostasis; deletions of the funda- finally into mature miRNAs (Box 1). miRNAs exhibit
mental miRNA biogenesis factors Dicer9 and Drosha10 tissue-​specific expression patterns31, which are primar-
are lethal in mouse embryos. Although the impor- ily regulated transcriptionally32. They are transcribed
tance of miRNAs for embryonic development is well mainly by RNA polymerase II33–35 and can be derived
established11, Drosha and Dicer are involved in other from introns or from long non-​coding RNAs (lncRNAs).
nuclear processes, such as pre-​mRNA splicing, which Pri-​miRNAs can consist of a single mature miRNA or
may also contribute to their deletion phenotypes 12. of clusters of often related miRNAs6,36. miRNAs are
The miRNA repository miRBase currently lists 1,917 grouped into families37 on the basis of the similarity of
precursor miRNAs (pre-​miRNAs) and 2,654 mature their seed sequences; the seed comprises nucleotides 2–8
miRNAs in Homo sapiens13, and more than 60% of (counting from the 5′ end) and is primarily responsible
Department of Integrative human protein-​coding genes harbour predicted miRNA for miRNA targeting of mRNAs38.
Structural and Computational
Biology,The Scripps Research
target sites14. Deregulation of miRNA function is asso- The mature miRNA functions within an AGO
Institute, La Jolla, CA, USA. ciated with numerous diseases15, particularly cancer16,17: protein (Fig. 1a), which comprises a single polypep-
*e-​mail: macrae@[Link] miRNAs can be both oncogenes (called oncomirs)18 and tide chain composed of four characteristic domains:
[Link] tumour suppressors19, although overall downregulation the amino (N)-terminal domain, the Piwi–Argonaute–
s41580-018-0045-7 of miRNA expression is a hallmark of cancer20. Some Zwille (PAZ) domain, the middle (MID) domain and

NATuRe RevIewS | MOleculAR Cell BIOlOgy volume 20 | JANUARY 2019 | 21

Reviews

Box 1 | miRNA biogenesis

The biogenesis of microRNAs (miRNAs) is a multistep process. They are transcribed mainly by RNA polymerase II33–35
as structured primary miRNAs (pri-​miRNAs) and processed into precursor miRNAs (pre-​miRNAs) and finally into
mature-​miRNA duplexes (see the figure). The mature miRNA comprises a 5p strand (arising from the 5′ arm of the
pre-​miRNA hairpin) and a 3p strand. The sequence of pri-​miRNAs can be altered by A-​to-I editing by double-​stranded
RNA-​specific adenosine deaminase (ADAR) proteins, which may affect further biogenesis and the sequence of the
mature miRNA or promote degradation of the pri-​miRNA99.
The pri-​miRNA hairpin is excised in the
nucleus by the Microprocessor complex, Transcription
comprising the RNase III enzyme Drosha252 miRNA gene RNA A-to-I editing by ADAR
and the protein DiGeorge syndrome critical Possible
region 8 (DGCR8)253,254. Drosha recognizes ADAR degradation
the double-​strand RNA–single-​strand RNA 5′ A I
I
junction at the hairpin base, whereas two I I
Pri-miRNA A I A I
DGCR8 proteins bind the stem and ensure 3′
correct cleavage255,256. Alternative cleavage Altered sequence
by Drosha leads to the production of Cleavage by the of precursor
isomirs97,98 (Fig. 2; Supplementary Table 1). Microprocessor
complex Drosha DGCR8 Alternative cleavage leads
Pre-​miRNAs are hairpins of ~70
to isomir formation
nucleotides . The hairpin end features a
6–8
5p
2-nucleotide overhang at the 3′ end, a 5′ DGCR8
phosphate and a 3′ hydroxyl, which are 3p
typical of RNase III products257. Exportin 5 Pre-miRNA
recognizes the overhang and transports
the pre-​miRNA to the cytoplasm258. XPO5
knockout in a human cell line reduced
but did not eliminate nuclear export of
pre-​miRNAs, which suggested that Exportin 5
alternative modes of pre-​miRNA nuclear Nuclear export
export exist262. Nucleus
In the cytoplasm, the RNase III enzyme Alternative export
Dicer259–261 binds the pre-​miRNA by mechanisms possible
Cytoplasm
recognizing the 5′ phosphate, 3′ overhang
and loop structure 263–266
. Dicer is a ‘molecular
ruler′263,267 that cleaves pre-​miRNAs at a
species-​specific length268 and yields a Alternative cleavage leads
mature-​miRNA duplex with another typical to isomir formation
Cleavage 5p
2-nucleotide 3′ overhang257,269. Alternative by Dicer
cleavage by Dicer can also lead to the 3p
production of isomirs98 (Fig. 2;
Supplementary Table 1). In vertebrates, Dicer
cleavage by Dicer is modulated by TAR TARBP or PACT
RNA-​binding protein (TARBP) and by protein
activator of the interferon-​induced protein
Loading on 5p
kinase (PACT); in flies, it is modulated by
Loquacious proteins 103–107
. AGO 5p
One strand of the mature miRNA (the
‘guide’ strand) is loaded into Argonaute (AGO) One strand is loaded as guide
whereas the other strand (‘passenger strand’) AGO • Strand with less stably
paired 5′ end preferred
is discarded270–272. Loading preference is given • A or U preferred as
to the strand possessing the less stably paired 5′-terminal nucleotide
5′ end273–275; Ago2 was also reported to prefer The passenger strand is
an A or U as the 5′-terminal nucleotide276. discarded and degraded

the P-​element induced wimpy testes (PIWI) domain. of the miRNA43,44, which is a feature important for the
Two linker domains (L1 and L2) connect the N-​terminal biogenesis of miR-451 (refs45,46) and for regulation of a
and PAZ domains (L1) and the PAZ and MID domains subset of miRNAs that are extensively paired with their
(L2). The N-​terminal and PAZ domains form one targets47,48. Although some human miRNAs are preferen-
lobe of AGO, and the MID and PIWI domains form tially loaded into specific AGO proteins, many associate
Guide strand the second lobe. The MID and PIWI domains hold the with all AGO proteins49–52.
The strand in the mature 5′ end of the miRNA, whereas the PAZ domain binds miRNA target sites are generally located in the
miRNA duplex that is its 3′ nucleotide39–42. The mammalian genome encodes 3′ UTR of mRNAs; they possess strong complementa-
loaded into an Argonaute
protein and used to identify
four AGO proteins (Ago1–4). Ago2 is the most highly rity to the seed region38, which is the main criterion for
complementary sites in target expressed and the only AGO protein able to cleave a target-​site prediction53–55. The strongest canonical (seed-​
mRNAs. target that is fully complementary to the guide strand matching) target sites are those that complement miRNA

22 | JANUARY 2019 | volume 20 [Link]/nrm

 Reviews

nucleotides 2–8 and have an adenine opposite miRNA guide strand, but by a binding pocket within AGO56,57
Metastable
A stable state in a system that nucleotide 1 (known as ‘t1A’), followed by those comple- (Fig. 1a). Target sites with complementarity to nucleo-
is not the state of least energy. menting nucleotides 2–8 without a t1A and nucleotides tides 2–7 or 3–8 of the miRNA are much weaker but still
2–7 with a t1A54. t1A is not recognized by the miRNA considered canonical54. Structural and single-​molecule
studies suggest that target recognition is achieved by a
a AGO two-​step mechanism in which first nucleotides 2–6 of the
3′ miRNA 3′ Target mRNA seed are pre-​organized by the MID and PIWI domains
Target binding
5′ 5′ in a stacked, helical conformation, with nucleotides 2–4
5′ 3′ exposed to the solvent41. This conformation allows for
rapid initial binding, which is weak and metastable and
becomes long-​lived only if the site presents complemen-
N L1 PAZ L2 MID PIWI tarity to nucleotides 7–8 of the guide strand. Otherwise,
AGO pocket AGO pocket
AGO disengages the site and either laterally diffuses
binds miRNA 5′ nt Supplemental binds miRNA 3′ nt along the mRNA or dissociates from it completely58.
Seed pairing Thus, the mRNA-​binding behaviour of an miRNA is
nt 2–8 nt ~13–16
5′ similar to that of an RNA-​binding protein (RBP)58–60.
miRNA
AGO–target mRNA crosslinking has enabled the
3′ identification of non-​canonical (not seed-​matching)
Target mRNA
binding sites61–63, and a recent large-​scale microarray-​
AGO pocket binds t1 based survey in HeLa cells reported sites with minimal
with a preference for A
seed-​pairing64. However, the biological function of non-​
canonical sites, when considered as a whole, has been
b Stimulation of disputed, as these sites imparted no detectable repression
DCP1–DCP2
mRNA decay complex in meta-​analyses of miRNA and small RNA transfection
AGO–miRNA
data sets54. In addition to the seed, the 3′ half of the
eIF4G
CCR4–NOT m7G miRNA can also contribute to target recognition (par-
complex AAA eIF4E ticularly nucleotides 13–16, which are termed the ‘sup-
PAN2–PAN3 PABPC eIF4A-I
GW182 mRNA
eIF4A-II Inhibition of translation plemental region’) in a subset of sites54,65 and can direct
complex
initiation miRNA family members with the same seed to differ-
miRNA target site ent target sites, as recently shown in C. elegans66 and in
in 3′ UTR
mouse brains63.
c One miRNA can regulate AGO–miRNA binding to the 3′ UTR leads to gene
many mRNAs silencing through translation repression and mRNA decay2,67
(Fig. 1b). The latter involves recruitment by AGO of a
member of the glycine-​tryptophan protein of 182 kDa
(GW182) protein family (in humans, t­ rinucleotide repeat-​
containing gene 6 A protein (TNRC6A), TNRC6B and
AAA AAA AAA TNRC6C)68–71. GW182 interacts with polyadenylate-​
binding protein (PABPC), thereby promoting mRNA
Closely spaced target sites One mRNA can be regulated deadenylation by recruiting the poly(A)-nuclease
promote cooperative repression by many miRNAs
deadenylation complex subunit 2 (PAN2)–PAN3 and
Fig. 1 | Overview of miRNA function and its regulation. a | Mature microRNAs carbon catabolite repressor protein 4 (CCR4)–NOT
(miRNAs) operate as functional units that include an Argonaute (AGO) protein. AGO complexes72–76. Deadenylation promotes decapping by
proteins have four domains, the amino-​terminal domain (N), the Piwi–Argonaute–Zwille the mRNA-​decapping enzyme subunit 1 (DCP1)–DCP2
(PAZ) domain, the middle (MID) domain and the P-​element induced wimpy testes (PIWI) complex73, thereby making the mRNA susceptible to
domain, and two linker regions (L1 and L2). The MID and PIWI domain hold the 5′ end of
rapid degradation by 5′–3′ exoribonuclease 1 (XRN1)77.
the miRNA and arrange it in a helical conformation; the MID domain has a binding pocket
for the 5′-terminal nucleotide. Another binding pocket, in the PAZ domain, holds the
GW182-mediated recruitment of CCR4–NOT also leads
3′-terminal nucleotide39–41. Nucleotides 2–8 from the 5′ end of the miRNA form the seed, to translation repression through recruitment of the
which is crucial for target mRNA recognition38. Seed interactions involve nucleotides 2–8, probable ATP-​dependent RNA helicase DDX6 (refs2,78).
2–7 and 2–6 (ref.54) and can be supplemented by the binding to the MID domain of an Inhibition of translation initiation is also caused
adenine in the target mRNA opposite miRNA nucleotide 1 (t1)54,56,57, or through additional by interfering with the function of eukaryotic initia-
base pairing to nucleotides ~13–16 of the miRNA (the ‘supplemental region’)65. b | miRNAs tion factor 4 A-​I (eIF4A-​I) and eIF4A-​II (refs2,79–81). In
silence gene expression by inhibiting translation at the initiation step, likely through Drosophila melanogaster S2 cell lysates, this activity was
release of eukaryotic initiation factor 4 A-​I (eIF4A-​I) and eIF4A-​II (refs79–81), and by independent of GW182 (ref.80), suggesting the existence
mediating mRNA decay2 through interactions with glycine-​tryptophan protein of 182 kDa of multiple mechanisms of translation inhibition. There
(GW182) proteins68–71. GW182 binds polyadenylate-​binding protein (PABPC) and the
is disagreement on the mechanism of interference with
deadenylation complexes poly(A)-nuclease deadenylation complex subunit 2 (PAN2)–PAN3
and carbon catabolite repressor protein 4 (CCR4)–NOT72–76. Deadenylation is followed by
eIF4A-​I and eIF4A-​II function79–81, but most data indi-
decapping by the complex mRNA-​decapping enzyme subunit 1 (DCP1)–DCP2 (ref.73) and cate that the miRISC induces their dissociation from
5′–3′ mRNA degradation (not shown)77. c | miRNAs form complex networks of interactions, target mRNAs and thereby inhibits ribosome scanning
as one miRNA can target many different mRNAs86, and one mRNA can be regulated by and assembly of the eIF4F translation initiation complex.
many different miRNAs87, with cooperative repression achieved by binding closely spaced A recent investigation in human and D. melanogaster cells
target sites65,89,90. m7G, 7-methylguanylate cap; nt, nucleotide. reported that the AGO Trp-​binding pockets that mediate

NATuRe RevIewS | MOleculAR Cell BIOlOgy volume 20 | JANUARY 2019 | 23

Reviews

Supplemental (nt 13–16)

pri-miRNA
5′
5p arm Drosha cleavage
Dicer cleavage

Seed (nt 2–8)

3p arm
3′
pre-miRNA

5′ end

3′ end
Mechanisms
Length of formation Effects
Canonical ~22 nt in mammals
miRNA (species-specific
variability)
5′ isomirs Longer • 5p: alternative cleavage • Altered seed
by Drosha • Possibly altered supplemental
Shorter • 3p: alternative cleavage pairing
by Dicer • Potential length-dependent effects
3′ isomirs Shorter • 5p: alternative cleavage • Length-dependent effects
by Dicer • Possibly altered supplemental
Longer • 3p: alternative cleavage pairing
by Drosha • Potential effects on localization
• Both: non-templated and turnover
nucleotide addition
5′ and 3′ Same or variable • Alternative cleavage All of the above
isomirs by Drosha and Dicer
Same or variable • Non-templated nucleotide
addition
Polymorphic Same; combination RNA editing • Seed editing may alter activity
isomirs with other isomir (primarily A-to-I) or specificity
types possible • Editing 3′ of the seed may
affect supplemental pairing,
Seed nucleotides (nt 2–8) localization or turnover
Supplemental pairing nucleotides (nt ~13–16)
Shifted terminal nucleotides
Edited nucleotides

Fig. 2 | Isomirs differ in length and sequence and expand the functional repertoire of miRNAs. Isomirs are classified as
5′, 3′ or polymorphic, with combinations possible. Depending on the arm of the microRNA (miRNA) precursors (5p (arising
from the 5′ arm of the precursor miRNA (pre-​miRNA) hairpin) or 3p; see inset) used to produce the mature miRNA, cleavage
by either Drosha or Dicer can result in the formation of the isomir98. 5′ isomirs have shifted seeds and can thereby target a
different set of genes109. The functions of 3′ isomirs are less clear, but there is increasing evidence for their differential
activity113,114. Polymorphic isomirs are generated by RNA editing, mainly by adenosine deaminase acting on RNA (ADAR).
The editing can affect miRNA biogenesis, either by preventing it or by leading to the formation of 5′ isomirs or 3′ isomirs;
if editing alters the seed, it could retarget an miRNA to other mRNAs98,99. nt, nucleotide; pri-​miRNA, primary miRNA.

GW182 binding41,82 are required for translation inhibi- fine-​tune protein expression92 and thereby buffer against
mRNA decay
Controlled mRNA degradation,
tion83. Thus, miRISC-​mediated translation inhibition fluctuations (‘noise’) in the levels of gene expression93.
usually starting with is still incompletely understood.
deadenylation, through Both modes of miRISC-​mediated gene silencing are Modification of miRNA sequence
either 3′–5′ exonucleolytic thought to be interconnected2, and ribosome profiling The interaction of miRNAs with their targets is largely
processing or decapping
assays revealed that mRNA decay is generally respon- based on their seed sequence38,54, and miRNA biogenesis
and 5′–3′ exonucleolytic
processing.
sible for 66–90% of silencing67,84. The observation that is affected by RNA secondary structures that mediate
translation inhibition can be rescued85 but mRNA deg- interactions with RBPs94. Cellular processes that alter the
Trp-​binding pockets radation is irreversible raises the possibility that the sequence of an miRNA or of its precursor therefore can
AGO possesses three pockets regulated pauses, or blocks, in the molecular cascade affect miRNA biogenesis, activity and turnover.
located in the PIWI domain,
which bind tryptophan
leading to mRNA degradation could allow translation
and mediate the interaction repression without mRNA decay. Isomir formation is a regulated process that determines
with GW182. One miRNA can silence hundreds of genes, miRNA activity. Isomirs are mature-​miRNA variants that
although the effect on each gene is generally mild86, differ in length, sequence or both31,95–97. The maturation,
miRNA clusters and multiple miRNAs can regulate the same gene87 stability and turnover, activity or targetome of isomirs can
Multiple miRNAs located in
(Fig. 1c). Furthermore, entire cellular pathways can be vary (Fig. 2). Isomirs are classified into 5′, 3′ or polymor-
close proximity on the genome
and transcribed as a single regulated by individual miRNAs87 or miRNA clusters88. phic (internal) isomirs, depending on the site of variation,
primary miRNA. miRNA binding of neighbouring target sites on a tar- and can be derived from alternative Drosha or Dicer pro-
get mRNA can result in cooperative repression65,89,90, cessing, RNA editing or non-​templated nucleotide addi-
Multivalent protein which might explain why the function of non-​canonical tion (NTA)98,99. A recent study in a human breast cancer
interactions sites can depend on the occupancy of neighbouring cell line reported that many miRNAs had no isoforms
Protein–protein interactions
mediated by multiple, often
canonical sites91. Cooperativity is explained, in part, by whereas miR-21-5p had 43 isoforms100. Isomir expression
fairly weak binding events the formation of multivalent protein interactions of GW182 patterns have been found useful in distinguishing differ-
or points of contact. with AGO proteins82. miRNAs can either switch off or ent types of cancer and represent potential biomarkers101.

24 | JANUARY 2019 | volume 20 [Link]/nrm

 Reviews

The primary determinant of miRNA length reduces ADAR1 expression, thereby reducing the edit-
and sequence is cleavage by Drosha and Dicer 32 ing of miR-455-5p; unedited miR-455-5p represses
(Supplementary Table 1). Alternative processing by cytoplasmic polyadenylation element-​binding protein 1
Drosha was first reported for miR-142 and miR-342 (CPEB1) and other tumour suppressors, thereby promot-
in mouse CD8 T cells102 and shown to be widespread in ing tumour growth and metastasis121. Changes in seed
human cells97. Cleavage by Dicer has been more exten- sequences can also be mediated by the more elusive C-​to-
sively studied; it is modulated by TAR RNA-​binding U editing129. miRNA A-​to-I editing was first observed
protein (TARBP) and protein activator of the interferon-​ in pre-​miR-22 in human and mouse tissues at species-​
induced protein kinase (PACT; also known as PRKRA) specific positions and was predicted to alter miRNA
in vertebrates (Box 1) and by Loquacious in flies103–107. targeting and activity130. In summary, RNA editing is a
Variation in the Dicer cleavage site directly modulates versatile regulatory mechanism that can control the abun-
the seed sequence of 3p miRNA strands (arising from the dance and target specificity of miRNAs, but in light of the
3′ arm of the pre-​miRNA hairpin) and can alter guide-​ relatively low frequency of editing99, further research is
strand selection103,107 (Fig. 2; Supplementary Table 1). Dicer required to characterize its biological importance.
binds TARBP and PACT by the same domain, as revealed
by a recent crystal structure of the binding interface, sug- Non-​templated nucleotide addition. NTA was first
gesting that Dicer is regulated by a mechanism based on suggested to be biologically regulated during D. melano­
binding competition between TARBP and PACT, which gaster embryogenesis131. NTA predominantly involves
produce specific isomirs for some miRNAs107. adenylation or uridylation at 3′ ends and is miRNA-​
Alternative cleavage by Drosha or Dicer can shift specific across tissue types, developmental stages, disease
the position of the seed and is the primary mechanism states and different species132.
of 5′ isomir production (Box 1; Fig. 2), as nucleotide NTA is carried out by several enzymes and can modu­
addition or removal at the 5′ end of mature miRNAs late miRNA stability (Fig. 3a). The poly(A) RNA poly-
is rare98. Cleavage-​directed 5′ end variation can retar- merase GLD2 (also known as PAPD4) mediates 3′-end
get an miRNA, as observed in flies104 and humans108–111, monoadenylation that stabilizes miR-122 in the liver133
and 5′ isomirs can target the same biological pathways and in mouse embryonic fibroblasts134. Poly(A)-specific
together109,112 (Supplementary Table 1). The abundance ribonuclease PARN counteracts this as its depletion in
of 5′ isomirs can vary between cell types, with certain a human liver cancer cell line leads to the appearance of
isomirs being predominant in specific cell types96. By miR-122 with 3′-oligo adenylation and to increased miR-
contrast, 3′ isomirs mostly differ in their stability and 122 stability. CUG triplet repeat RBP1 (CUGBP1; also
turnover (see below), although recent findings point also known as CELF1) binds miR-122 and mediates its deg-
to length-​related effects on targeting and activity. For radation by recruiting PARN (Fig. 3b). A similar mecha­
example, binding to the RNA of hepatitis C virus (HCV; nism regulates miR-93 and miR-652–3p135. By contrast,
see below) of a 21-mer 3′ isomir of miR-122 is weaker a recent analysis of miRNAs in the hippocampus of Gld2-
than binding of longer miR-122 isomirs113, and longer knockout mice revealed reduced levels of 3′ adenylation
3′ isomirs of miR-222 have increased apoptotic activity114. but no effect on miRNA stability136; similarly, only about
half of the investigated miRNAs were destabilized in
Editing the sequence of miRNA precursors. RNA human fibroblasts when GLD2-mediated adenylation was
editing can change the miRNA sequence, generating suppressed137. Moreover, 3′ adenylation by the poly(A)
isomirs, and can also affect biogenesis, leading to 5′ or RNA polymerase PAP-associated domain-containing
3′ variation98,99 (Box 1; Fig. 2). Deamination is the most protein 5 (PAPD5) initiates degradation of miR-21 (ref.138).
commonly observed type of miRNA precursor editing. Compared with 3′ adenylation, 3′ uridylation more
Adenosine deaminase acting on RNA (ADAR; also consistently inhibits miRNA activity. Uridylation by
known as DRADA) converts adenosine into inosine99, terminal uridylyltransferase 4 (TUT4; also known as
which is read as guanosine during splicing and trans- ZCCHC11) of miR-26b in a human adenocarcinoma
lation. Similarly, cytidine deaminase acting on RNA cell line reduces target repression139. Moreover, TUT4-
(CDAR) proteins, better known as members of the mediated uridylation primes numerous miRNAs for
apolipoprotein B mRNA editing catalytic polypeptide-​ degradation during T cell activation140 (Fig. 3a).
like (APOBEC) family, convert cytidine into uracil115. In To date, the enzymes reported to mediate NTA include
vertebrates, ADAR1 and ADAR2 act on double-​stranded GLD2 (refs133,141), PAPD5 (ref.141), poly(A) polymerase-γ
RNA and edit miRNA precursors99. A-​to-I editing sites and poly(A) RNA polymerase, mitochondrial132 (ade-
appear widespread116, and in the human brain ~16% of nylation) and TUT7, TUT4 (ref.139) and TUT1 (uridyl-
pri-​miRNAs were predicted to be edited (on the basis ation). Species-dependent differences in NTA have
of the analysis of 209 pri-​miRNAs). However, editing been observed, with uridylation being more common in
levels at different miRNAs vary considerably99,117. C. elegans and adenylation more common in mouse and
A-​to-I editing of miRNA precursors can interfere with human132. In summary, NTA appears to influence, both
cleavage by Drosha117–121 or Dicer117,122, or, less frequently, positively and negatively, the stability of specific miRNAs
with AGO loading, and editing of the seed can redirect in specific cell types. Recent findings demonstrate that
an miRNA to a new set of targets120,123–126. A-​to-I editing interactions between the miRNA 3′ region and target
of miRNAs is involved in miRNA deregulation in can- RNAs can substantially affect miRNA modification
cer121,127,128. For example, in melanoma, overexpression and turnover (see below), thereby providing a potential
of cAMP-​responsive element-​binding protein (CREB) mechanism for the cellular context dependence of NTA.

NATuRe RevIewS | MOleculAR Cell BIOlOgy volume 20 | JANUARY 2019 | 25

Reviews

a • miR-122 in mouse liver133

• miR-122 in human fibroblasts134
• miR-122, miR-93 and miR-652-3p in Huh-7 cells135
• miR-145-5p, let-7d-5p, let-7i-5p, miR-98-5p, miR-26b in
and miR27a-3p in human fibroblasts and let-7b-5p in vitro137 A549 cells139

PARN U Reduced target

5′ repression
5′ A
TUT4 NTA reduces Degradation by
GLD2 miRNA function unknown nuclease
miR-122 in human NTA stabilizes
liver cancer cell line miRNA TUT4
5′ 5′ U 5′ Multiple miRNAs
upon T cell
Unknown effect activation140
of NTA GLD2 NTA destabilizes
PAPD5 miRNA
5′ A
miR-21 in
5′ A 5′ THP1 cells138
• miR-542-3p, miR-431-5p, let-7f-5p, miR-193b-3p,
miR-34a-5p and miR-154-3p in human fibroblasts
and let-7e-5p in vitro137 3′–5′ degradation
• At least 43 miRNAs in mouse hyppocampus136 by PARN

b For miR-122, miR-93 and miR-652-3p in Huh-7 cells, c

adenylation counteracts degradation by PARN135 Targets with Targets with extensive Fully complementary
seed pairing pairing to miRNA 3′ ends targets
GLD2
miRNA
5′ A 5′ A A ... A 5′ 5′ 3′ 5′

5′
Target Target-directed miRNA Target cleavage by
PARN miRNA degradation unloading catalytically active
AGO proteins
Repression of Nucleotidyl (for example,
A Tailing transferase mammalian Ago2)
5′ target mRNA 5′
CUGBP1 and miRNA
stabilization 5′
Trimming
5′ Exonuclease
CUGBP1 recognizes a PARN is recruited by CUGBP1,
GU-rich region in miR-122 leading to 3′–5′ degradation

Fig. 3 | Non-​templated nucleotide addition and miRNA turnover. PARN. CUG triplet repeat RNA-​binding protein 1 (CUGBP1) interacts with
a | Non-​templated nucleotide addition (NTA) by poly(A) RNA polymerase miR-122 and recruits PARN135. c | Interactions of an Argonaute (AGO)-bound
GLD2 can stabilize some microRNAs (miRNAs) but has no effect on others. miRNA with a target mRNA through the miRNA seed sequence result in
Terminal 3′ adenylation of miR-21 by PAP-​associated domain-​containing translation repression and mRNA degradation. Targets with extensive
protein 5 (PAPD5) promotes exonucleolytic cleavage by poly(A)-specific pairing to miRNA 3′ ends promote tailing 144,151 , trimming 148 and
ribonuclease PARN138. Terminal 3′ uridylation by terminal uridylyltrans- target-directed miRNA degradation142,149,150. A target mRNA that is fully
ferase 4 (TUT4) can reduce the activity of miR-26b139 or prime miRNAs for complementary to a miRNA is cleaved when bound by a catalytic AGO,
degradation following T cell activation140. b | Terminal 3′ adenylation such as mammalian Ago2 (refs44,49), but such pairing can also result in
stabilizes some miRNAs by counteracting 3′–5′ exonucleolytic activity by unloading of the miRNA from AGO154.

miRNA turnover. miRNAs are generally thought to One mechanism of miRNA destabilization is termed
be stable in vivo, but turnover rates are regulated by target RNA-directed miRNA degradation (TDMD)148.
multiple factors and across different miRNAs can In TDMD, target RNAs with extensive complementarity
range from minutes to days 142. Turnover rates are to both 5′ and 3′ regions of an miRNA promote its turn-
miRNA-specific and isomir-specific and thus linked over (Fig. 3c). TDMD is often associated with 3′ NTA,
to miRNA sequence143,144; indeed, a 5′ guanine or cyto- or ‘tailing’, and with 3′–5′ trimming of the degraded
sine is associated with faster turnover rates than a uracil miRNA142,149,150. It has been proposed that 3′ tailing
in this position143. Stability can also be miRNA-specific. extends the AGO-loaded miRNA144,151 until a 3′–5′
For example, miRNAs in mouse fibroblast 3T3 cells nuclease can bind and degrade it148. This hypothesis
are generally stable, except for miR-16 family mem- is consistent with the observation that 3′-remodelled
bers, which are intrinsically unstable145. This insta- isomirs generally coincide with TDMD. However, the
bility allows miR-16 levels to vary with, and thereby hypothesis has recently been challenged by the obser-
help regulate, the progression of the cell cycle. Some vation that tailing of miR-7 by GLD2, which coincides
miRNAs are stable unless specifically degraded in with TDMD of miR-7 in mouse neurons, is not required
response to developmental cues; for example, miR-150 for efficient miR-7 degradation152. Similarly, the 3′–5′
is rapi­dly lost when murine naive T cells differentiate exoribonuclease DIS3-like exonuclease 2 (DIS3L2)
into T helper 1 (TH1) and TH2 lymphocytes146. Turnover was implicated in 3′ trimming associated with TDMD
rates can also depend on tissue context: faster turn­ of miR-27a, but DIS3L2 depletion did not affect
over rates are generally observed in neuronal miRNAs TDMD efficiency151. Thus, it is possible that TDMD
compared with miRNAs in other tissues147. and miRNA 3′ remodelling are distinct processes;

26 | JANUARY 2019 | volume 20 [Link]/nrm

 Reviews

therefore, the molecular mechanism of TDMD remains Several studies report phosphorylation of Ser387
poorly understood. in the L2 region of human Ago2 (refs158–160) (Fig. 4a).
Although viral inducers of TDMD have been known Ser387 phosphorylation is stimulated by the p38
for several years (see below), endogenous inducers of MAPK pathway-mediated stress response (for example,
TDMD have only recently been reported. In zebrafish, by MAP kinase-activated protein kinase 2 in vitro)158.
miR-29b is selectively degraded in the cerebellar granule The proto-oncogene RACγ serine/threonine-protein
cell layer by a lncRNA and in mice by the 3′ UTR of a kinase (AKT3) also phosphorylates Ser387 in vitro and
protein-coding gene153. Both transcripts show extensive in HeLa cells, and its depletion or the expression of an
sequence similarity around a highly complementary tar- S387A Ago2 mutant (which cannot be phosphorylated)
get site for miR-29b. Similarly, the lncRNA Cyrano can led to derepression of a luciferase reporter and weak-
induce TDMD of miR-7, thereby affecting the stability ened the Ago2 interactions with TNRC6A159. Similarly,
of an miR-7-interacting circular RNA (circRNA) (see Ser387 phosphorylation is necessary for interactions
below)152. Furthermore, miR-503, which is inherently between Ago2 and LIM domain-containing protein 1,
unstable in 3T3 cells, can be stabilized by mutating either which is a protein suggested to bridge Ago2 and GW182
its seed or 3′ regions145, raising the possibility that its (ref.161). Blocking Ser387 phosphorylation also reduced
inherent instability is mediated by TDMD. It is possible accumulation of Ago2 in cytoplasmic foci with GW182
that many inherently unstable miRNAs are subjected to (refs158,159) and was suggested to reduce trafficking of
TDMD through yet-to-be-identified target RNAs. Ago2–miRNA complexes to multivesicular endosomes and
A biochemical screen using lysates of HEK239 cells reduce the secretion of exosomes162. Notably however,
(a human embryonic kidney cell line) revealed that a recent investigation found that S387A Ago2 did not
miRNA degradation is promoted particularly by reduce colocalization of GW182 in HeLa cells, indicat-
seedless targets with high complementarity to miRNA ing that additional factors may be involved in enabling
3′ ends, with complementarity of the 3′-terminal nucleo­ the Ago2–GW182 interaction163. The combined data
tide being crucial for degradation150. Combined with indicate that Ser387 phosphorylation promotes miRNA
the observation that the sequences of both the 5′ end function by stimulating the assembly of the miRISC.
and the 3′ end influence turnover rates143,144, this might Tyr393 is adjacent to Ser387 in the L2 region, and
partially explain the tissue-specific and miRNA-specific its phosphorylation is also well documented160,164,165
effects of NTA on stability. Alternatively, highly com- (Fig. 4a). It appears to be mediated by epidermal growth
plementary targets can destabilize the AGO–miRNA factor receptor (EGFR) and stimulated by hypoxic
interaction and promote release of the guide strand154, stress164. Overexpression of oncogenic RAS reversibly
although miRNAs appear to be stabilized by a high inhibits protein-tyrosine phosphatase 1B (also known
abundance of seed-matching targets155. as tyrosine-protein phosphatase non-receptor type 1),
In addition to exonucleolytic miRNA turnover, resulting in Ago2 hyperphosphorylation at Tyr393
Tudor staphylococcal nuclease, which is respon­sible (ref.165). Both studies reported diminished interactions
for degradation of A-to-I edited double-stranded between Ago2 and Dicer and reduced levels of Ago2-
RNA99, was found to target miR-31-5p, miR-29b-3p and associated miRNAs following Tyr393 phosphoryla­tion.
miR-125a-5p in HEK293T cells. Other tested miRNAs Thus, in contrast to Ser387, Tyr393 phos­phorylation
did not appear to be affected, suggesting that this is an appears to negatively regulate miRNA activity by
miRNA-specific mechanism, although how selectiv- inhi­biting the loading of miRNAs into Ago2, thereby
ity is established remains to be determined156. Finally, promoting tumorigenesis.
phosphorylation-dependent regulation of miRNAs was Tyr529 phosphorylation also potentially blocks
recently discovered for the tumour suppressor miR-34. miRNA loading160 (Fig. 4a). Tyr529 binds the 5′ phosphate
The nuclei of four human cancer cell lines were found to and first nucleotide of miRNAs loaded into AGO41,166,167.
Seedless targets contain pools of inactive, mature, single-stranded miR- Tyr529 phosphorylation is therefore expected to pre-
miRNA targets with 34 lacking a 5′ phosphate and not bound by AGO157. clude miRNA binding, as indeed was demonstrated in
considerably reduced
Irradiation of the cells led to 5′ phosphory­lation, export Tyr529 Ago2 mutants160. Tyr529 phosphorylation was
complementarity to the
miRNA seed. to the cytoplasm and loading into AGO of miR-34. The also associated with decreased miRNA binding to Ago2
serine-protein kinase ATM and the RNA kinase Clp1 during macrophage activation168.
Multivesicular endosomes were required for miR-34 phosphorylation. It is unclear A phosphorylation cycle of residues S824–S834 in
Type of late endosome that how widespread this pheno­menon is, but these findings Ago2 regulates the release of target mRNAs169,170 (Fig. 4b).
contains intraluminal vesicles
formed by budding into the
suggest that miRNAs can be maintained in an inactive Target binding leads to phosphorylation of these resi-
lumen of the endosome. Their form and rapidly activated in response to stimuli. dues by casein kinase I isoform-α (CSNK1A1), which
content can be degraded by reduces the affinity of Ago2 for mRNA and enables tar-
fusion with lysosomes or Post-translational modification of AGO get release. The serine/threonine-protein phosphatase 6
released into the extracellular
AGO proteins undergo PTM on multiple residues regulatory ankyrin repeat subunit C (ANKRD52)–
space through fusion with the
cell membrane. (Fig. 4). Phosphorylation is the best characterized AGO catalytic subunit (PPP6C) complex dephosphorylates
PTM and it occurs at three main sites: at the L2 linker on the residues, which primes Ago2 for binding a new tar-
Exosomes Ser387 and Tyr393 of human Ago2; at the miRNA 5′-end get. Interrupting this cycle strongly inhibited miRNA
Type of extracellular vesicle, binding region in the MID domain on Tyr529; and at activity, and the cycle was suggested to prevent overly
~50–150 nm in diameter, that
contains proteins, lipids and
a surface-exposed loop in the PIWI domain known as long association with mRNA targets170. Alternatively, the
RNA and can carry cargo to the eukaryotic insertion or the S824–S834 cluster, which phosphorylation cycle may represent an AGO regulation
target cells. includes Ser824, Ser828, Thr830, Ser831 and Ser834. mechanism that is mediated by mRNA-binding proteins.

NATuRe RevIewS | MOleculAR Cell BIOlOgy volume 20 | JANUARY 2019 | 27

Reviews

a
S387 phosphorylation Y393 phosphorylation Y529 phosphorylation
Increased miRNA-mediated repression; Reduced levels of Ago2- Reduced miRNA loading
reduced trafficking to exosomes associated miRNAs

P P
P P P
P

EGFR In response to
MAPKAPK2
hypoxic stress
Phosphorylation sites or PARylation
of unknown function PARylation of Ago1–4 reduces
AKT3
miRNA activity
T307 P P T303
PAZ PARylation sites
S253 P miRNA remain unknown
5′ PARP
N
P MID PIWI In response to PAR
S798 cell stress
AGO
Type I collagen
prolyl-4 UBC9
b S824–S834 cluster phosphorylation cycle
hydroxylase
CSNK1A1

PP PP
P700 4-hydroxylation K402 sumoylation
Increased Ago2 stability Ago2 destabilization or
increased Ago2 siRNA activity
Target Target
binding release

OH
O S
SU
SUMO
U SUMO PP PP

ANKRD52–PPP6C complex

Fig. 4 | The activity and the stability of miRNA-​induced silencing complex post-​translational modifications include Pro700 (P700) 4-hydroxylation, which
is modulated by post-​translational modifications of Argonaute proteins. increases Ago2 stability171; Lys402 (K402) sumoylation, which was reported to
a | Phosphorylation of Ser387 (S387) in the linker 2 (L2) region of Argonaute either destabilize Ago2 (ref.177) or be required for full small interfering RNA
(AGO) was found to be mediated by MAP kinase-​activated protein kinase 2 (siRNA) activity178; and poly(ADP-​ribosylation) (PARylation), which inhibits
(MAPKAPK2)158 and RACγ serine/threonine-​protein kinase (AKT3)159 in vitro. miRNA activity173,174, presumably by decreasing target accessibility. b | The
Ser387 phosphorylation increases microRNA (miRNA) activity by stimulating S824–S834 cluster in the eukaryotic insertion region of human Ago2
the assembly of miRNA-​induced silencing complexes and reduces undergoes a phosphorylation cycle, which regulates AGO–target interactions.
translocation of Ago2 to multivesicular endosomes and secretion of Phosphorylation of the Ser residues in the cluster by casein kinase I isoform-​α
exosomes162. Phosphorylation of the nearby Tyr393 (Y393), also in the L2 (CSNK1A1) favours target-​mRNA release. Subsequent dephosphorylation by
region, decreases the miRNA–Ago2 association, thereby reducing miRNA the serine/threonine-​protein phosphatase 6 regulatory ankyrin repeat subunit
activity164,165. Tyr 529 (Y529) is located in the middle (MID) domain, near the C (ANKRD52)–catalytic subunit (PPP6C) complex primes Ago2 for the next
miRNA 5′-nucleotide binding pocket, and its phosphorylation prevents round of target binding169,170. EGFR , epidermal growth factor receptor ;
miRNA loading160. No function has yet been assigned to phosphorylation sites N, amino-​terminal domain; OH, hydroxyl group; P, phosphate; PAR, poly
in the Piwi–Argonaute–Zwille (PAZ) domain (S253, T303 and T307) and in the (ADP-ribose); PARP, poly(ADP-​ribose) polymerase; SUMO, small ubiquitin-​like
P-element induced wimpy testes (PIWI) domain (S798)160. Additional AGO modifier; UBC9, ubiquitin carrier protein 9.

The cluster is conserved in human AGO proteins, mouse to increase the stability of Ago2 in mouse and human
Ago2, rat Ago2, zebrafish Ago2, the C. elegans argonaute cells171. Interestingly, Pro700 is adjacent to a Trp-binding
protein (ALG-1) (ref.169) and in the fruitfly Ago1 (but not pocket in Ago2 that is used to interact with GW182
in the fruitfly Ago2, which is used for small interfering (refs41,82); thus, its 4-hydroxylation could potentially alter
RNA (siRNA)-mediated target cleavage)170. the assembly of the miRISC, although this has not yet
Additional Ago2 phosphorylation sites have been been demonstrated.
reported in the PAZ domain (Ser253, Thr303 and Ago1–4 undergo poly(ADP-ribosylation) (PARylation)
Thr307) and the PIWI domain (Ser798), but their in human cell lines172. Poly(ADP-ribose) (PAR) is a
function is currently unknown160 (Fig. 4a). polymer of ADP-ribose units that can be enzymatically
In addition to phosphorylation, other AGO PTMs are added to Asp, Glu and Lys residues. PARylation regulates
known (Fig. 4a). Hydroxylation of Ago2 Pro700 appears the cellular stress response, particularly the formation of

28 | JANUARY 2019 | volume 20 [Link]/nrm

 Reviews

cytoplasmic stress granules, which regulate mRNA trans- hypo­thesis, such a mechanism could alter the affinities
lation and stability172. PARylation of Ago1–4 reduced of different target sites that otherwise seem equivalent on
translation repression and endonucleolytic cleavage the basis of miRNA complementarity alone. Additionally,
and was hypothesized to impair target accessibility173. although this has yet to be demonstrated, ceRNAs that
PARylation-induced downregulation of miRNA activ- promote TDMD could function at lower concentrations
ity was observed in the antiviral response in HEK293 by reducing miRNA abundance189. Thus, although the
cells: cytotoxic interferon-stimulated genes are gener- original ceRNA hypothesis may be generally implausible,
ally heavily regulated by miRNAs, and miRNA inhibi- ceRNA mechanisms may still have a biological role188.
tion through AGO PARylation might enable a cytotoxic The controversy notwithstanding, the list of potential
response to viral infection174. ceRNAs is growing and includes lncRNAs, pseudogenes,
The ubiquitin–proteasome system reduces the abun- mRNAs (Fig. 5b) and specific circRNAs187. In mice, long
dance of the fruitfly Ago1 and mouse Ago2 (ref.175) and intergenic non-protein coding RNA, muscle differenti-
drives global miRNA downregulation during T cell ation 1 (Linc-md1) is proposed to drive myoblast differ-
activation176. Finally, two groups have reported Ago2 entiation by sequestering miR-133 and miR-135, which
sumoylation at Lys402 with contradicting conse- target muscle-specific transcription factors191. Similarly,
quences: sumoylation might destabilize Ago2 (ref.177) in human cells, the untranslated PTEN pseudogene 1
or be required for full Ago2 siRNA-mediated activity178. (PTENP1) derepresses PTEN183. This derepression was
Compared with AGO proteins, less is known about suggested to be a common function of pseudogenes187,
the regulation of other proteins involved in miRNA func- as they regularly share miRNA target sites with their
tion. Tripartite motif-containing protein 65 ubiquitylates parent genes. Recently, the overexpression of the Braf
TNRC6 in HEK293 cells, leading to proteasomal degra- pseudo­gene was found to promote B cell lymphoma in
dation and derepression of miRNA targets179. TNRC6A mice through a possible ceRNA mechanism192. mRNAs
has long been known to be phosphorylated in mamma- have also been suggested to function as ceRNAs; for
lian cells180, and phosphorylation of the PAM2 motif of example, PTEN expression is regulated not just by
TNRC6C was suggested to reduce its interactions with pseudogenes but also by numerous protein-coding ceR-
PABPC1 (ref.181). A recent mass-spectrometry study of NAs193,194 (Fig. 5b). Different types of ceRNA were found
the TNRC6A interactome in HeLa cells revealed a large to crosstalk in a network of various oncogenic pathways
number of binding partners, but it is currently unknown in glioblastoma195.
whether any of these regulate TNRC6A182. circRNAs are enigmatic ncRNAs with dynamic and
complex expression patterns; their function is still poorly
Regulation of miRNAs by sequestration understood 196. The circRNA CDR1 antisense RNA
The competing endogenous RNA (ceRNA) hypothesis183 (CDR1as) is expressed in human and mouse brains197,198.
postulates that an increase in the cellular concentration Because CDR1as contains many binding sites (63–74) for
of an miRNA target RNA would reduce the cytoplasmic miR-7, it acts as an miR-7 sponge when overexpressed197.
availability of the specific miRNA by binding it, thereby In support of the sponge mechanism, depletion of
derepressing other mRNAs that are targets of the same CDR1as in HEK293 cells led to downregulation of miR-7
miRNA; this is similar to the proposed function of target mRNAs198. However, knockout of CDR1as in mice
miRNA sponges184. Thus, according to the ceRNA model, led to a reduction in miR-7 levels and an increase in the
gene expression would be shaped by the global compe- levels of miR-7 target genes in the brain, suggesting that
tition of target RNAs for miRNAs185 (Fig. 5a). The poten- the sponge effect stabilizes miR-7, as none of its binding
tial efficacy of ceRNAs is controversial, as the increased sites in CDR1as possesses the extensive 3′ complementa-
expression of any single ceRNA is expected to increase rity required for TDMD199. The miR-7 molecules can be
only slightly the total number of target sites of an miRNA released from CDR1as through its slicing by miR-671, for
and thus be unlikely to meaningfully affect miRNA acti­ which CDR1as contains a highly complementary target
vity186–189. Two studies measuring the cellular abundance site. Thus, CRD1as may bind to miR-7 in order to help
Stress granules
of miRNAs and miRNA target sites proposed that the to localize it to specific subcellular compartments197–199.
Following global translation
shutdown during the cellular ability of a putative ceRNA to derepress the expression The lncRNA Cyrano was proposed to be responsible for
stress response, cytoplasmic of other target mRNAs depends on miRNA abundance190 the increased turnover of miR-7 upon CDR1as deple-
granules form, which are and/or on target site abundance186. Both studies agreed tion199 and has now been reported to induce TDMD
composed of non-​translating that, in most cases, unphysiologically high levels of a of miR-7 (ref. 152) . Interestingly, Cyrano depletion
mRNAs, translation initiation
factors and regulatory proteins.
ceRNA would generally be necessary to yield a biolog- increased miR-7 levels and led to a decrease in CDR1as
ically meaningful effect186,190. Mathematical modelling levels, in part owing to an increased miR-671-mediated
PAM2 motif of miRNA distribution in the targetome indicates that degradation of CDR1as152.
Poly(A) binding protein a ceRNA must increase the cellular abundance of target Although thousands of circRNAs have been identi-
interacting motif 2 mediates
sites at least twofold to be effective188. On the other hand, fied, only a handful stand out as harbouring large num-
the interaction between
GW182 and PABP. it was also noted that ceRNA efficacy could be increased bers of miRNA target sites200. In addition to CDR1as,
if local concentrations of ceRNAs and/or miRNAs ten different circRNAs derived from zinc-finger genes
miRNA sponges deviated strongly from the cytoplasmic average 188. contain 7–24 target sites for the miR-23, miR-181 or
Transcripts that contain The discovery of the phosphorylation cycle regulating miR-199 families200. Another proposed sponge, the
multiple target sites for a
specific miRNA and bind
AGO activity also raises interesting new possibilities, testis-specific circRNA sex-determining region Y,
miRNAs, thereby derepressing as AGO–target binding might be modulated by possesses 16 sites for miR-138 in mice197, but only 1
the miRNA target mRNAs. mRNA-associated RBPs170. With respect to the ceRNA in humans 200. Thus, although some circRNAs may

NATuRe RevIewS | MOleculAR Cell BIOlOgy volume 20 | JANUARY 2019 | 29

Reviews

interact with miRNAs, this does not appear to be a major single-stranded RNA virus of the Flaviviridae family
function of most of them. that causes acute and chronic liver infection201. Its highly
structured 5′ UTR contains an internal ribosomal
Viral modulation of miRNA activity entry site202, and upstream of it, at the very 5′ end of the
Viruses have evolved to repurpose or modulate host genomic RNA, two binding sites for the liver-abundant
miRNAs for their replication, which often affects miR-122 (ref. 203) (Fig. 5c) . The binding sites recruit
miRNA function (Fig. 5c). The best studied interaction is Ago2–miR-122 to the uncapped 5′ end of the viral
between the HCV and miR-122. HCV is a positive-sense RNA204 to protect it from the cellular antiviral response

a ceRNA hypothesis and controversy

AAA Expression ceRNAs compete
AAA of ceRNAs for miRISCs AAA
AAA
AAA
AAA
miRISC AAA AAA AAA
AAA
AAA AAA AAA AAA
AAA AAA
AAA miRNA AAA
AAA target site AAA

Expression
Expression

miRSCs Gene expression

target mRNAs altered

Gene 1 Gene 2 Gene 1 Gene 2

Main concern: ceRNA expressed at physiological levels is unlikely
to compete with the very large number of other miRNA targets
Other miRNA targets
AAA AAA
AAA AAA AAA
AAA Expression AAA AAA
AAA AAA AAA AAA AAA AAA
AAA AAA of ceRNA AAA AAA
AAA AAA AAA
AAA AAA AAA
AAA AAA AAA AAA
AAA AAA AAA
AAA AAA AAA AAA AAA
AAA AAA AAA AAA AAA
AAA AAA
AAA
AAA AAA AAA AAA AAA AAA ceRNA

b Examples of ceRNAs
Long non-coding RNAs Derepressed mRNAs mRNAs
miR-133
AAA Overexpression of Zeb2
MAML1 Myoblast
AAA differentiation AAA Derepression
AAA MEF2C of PTEN
miR-135 AAA
Linc-md1 AAA
AAA
Nine shared Repression
Pseudogenes PTEN derepressed Depletion of Zeb2 miRNA families
Tumour of PTEN
AAA AAA suppression AAA AAA
PTENP1

c Viral mechanisms modulating miRNA function

HCV miR-122 HVS
5′ UTR of viral RNA contains HVS expresses short ncRNAs to modulate host miRNA
two binding sites for miR-122 activity in infected T cells
IRES AAA miR-27a targets
HSUR1 causes
AAA TDMD of miR-27a derepressed
Viral RNA
genome AAA miR-27a miR-27a AAA
5′ 5′
Endogenous miR-122 AAA
Binding of Ago2–miR-122 stabilizes targets are derepressed HSUR1
the viral RNA by protecting it from
detection and exonucleases
HSUR2 redirects miRISCs miR-16 miR-142-3p
to inhibit apoptosis in
BVDV infected cells
The 3′ UTR of the viral genome HSUR2
contains a miR-17 target site
AAA AAA
host mRNA
Endogenous targets
miR-17 of miR-17 are derepressed
3′ AAA
Binding of miR-17 stabilizes the viral RNA AAA
through an unknown mechanism

30 | JANUARY 2019 | volume 20 [Link]/nrm

 Reviews

Small nuclear RNAs

and exonuclease activity205,206. This functionally seques- sites for host miRNAs: miR-16 (HSUR2), miR-27a
Small non-​coding RNAs in the ters cytoplasmic miR-122, which is a tumour suppres- (HSUR1) and miR-142-3p (both HSUR1 and HSUR2).
nucleus that form complexes sor miRNA responsible for hepatic maintenance and HSUR1 reduces miR-27a levels in infected marmoset
with proteins and are part of potentially explains the link between chronic HCV infec- T cells through TDMD, thereby derepressing miR-27a
the splicing machinery.
tion and an increased risk of developing hepatocellular cellular target mRNAs and promoting T cell activa-
carcinoma207. tion 212. TDMD of miR-27a is required for efficient
Another virus from the Flaviviridae family that binds HVS replication, as viral strains with HUSR1 bearing
host miRNAs is bovine viral diarrhoea virus, which a mutated miR-27a binding site have reduced titres213.
is an economically important cattle virus. Binding of HSUR2 does not deplete the miRNAs it binds but
let-7 and miR-17 at a structured region of the viral RNA instead acts as a tether that recruits AGO–miR-142-3p
3′ UTR promotes viral replication. miR-17 also sta- and AGO–miR-16 complexes to cellular mRNAs
bilizes the viral RNA upon binding and increases its that encode pro-apoptosis factors, thereby induc-
translation through an unknown mechanism. The ing silencing of the tethered mRNAs and preventing
binding of miR-17 to the viral RNA derepresses cellu- apoptosis214.
lar targets of the miRNA, an effect also observed with Analogously to HSUR1 function, TDMD and reduc-
classical swine fever virus208. The detection of similar tion in miR-27a levels were observed in mouse cell lines
interactions between miRNAs and viral genomic RNA and primary macrophages upon murine cytomegalo­
of different members of the Flaviviridae family raises virus infection215, mediated by a transcript harbouring
the question of whether this mechanism might be even a miR-27a target site with substantial 3′-end com-
more widespread in this family, which includes other plementarity 216. Similarly, human cytomegalovirus
health-relevant members such as dengue virus and (HCMV) targets miR-17 and miR-20a, two mem-
zika virus. bers of the miR-17-92 cluster. Degradation of these
Hepatitis B virus, which is a partially double-stranded miRNAs, which is mediated by complementarity with
DNA virus unrelated to HCV, directly downregulates sites in viral ncRNA, accelerates virus production during
miR-122 in liver cells, as this miRNA appears to inhibit HCMV infection217.
viral infection209. A hybrid viral–human transcript210
contains multiple miR-122 binding sites, one of which miRNA transport from the cytoplasm
resembles a TDMD element that depletes cellular As miRNAs function in the cytoplasm, their activity
miR-122 levels and promotes loss of hepatic function can be modulated by transferring them to the nucleus
and liver damage in mice211. or to extracellular vesicles, where they potentially have
Several members of the Herpesviridae family also location-specific functions218.
modulate host miRNA function. Herpesvirus saimiri
(HVS) expresses H. saimiri uracyl-rich RNAs (HSURs), Nuclear localization of miRNAs. miR-21 was the first
which are short ncRNAs with structural similarity miRNA to be found in the nucleus, as 20% of total miR-21
to small nuclear RNAs. Two HSURs harbour binding is present in nuclear extracts49; miR-29b was found to
be predominantly localized to the nucleus owing to a
◂ Fig. 5 | miRNA sequestration by endogenous and viral RNAs. a | The competing
3′ hexanucleotide nuclear localization signal219. Ago2,
endogenous RNA (ceRNA) hypothesis postulates that a newly expressed RNA can Dicer, TARBP and GW182 are present in the nucleus
compete with the already present microRNA (miRNA) targets for cytoplasmic and form complexes. In C. elegans, the AGO homologue
miRNA-induced silencing complexes (miRISCs), potentially leading to derepression of ALG-1 uses mature let-7 to bind pri-let-7 transcripts in
certain genes185. However, a ceRNA is unlikely to lead to gene derepression when the nucleus and promote their biogenesis, thereby cre-
expressed at physiological levels186–190. b | Long non-coding RNAs, pseudogenes and ating a positive feedback loop220. Recent single-molecule
mRNAs can have ceRNA activity. In mice, long intergenic non-protein coding RNA, studies in mammalian cells indicated that nuclear
muscle differentiation 1 (Linc-md1) contains one binding site for miR-133 (blue) miRNAs do not repress complementary targets. The
and two for miR-135 (green). By sequestering these miRNAs, the muscle-specific importance of the hexanucleotide localization signal
transcription factors mastermind-like transcriptional co-activator 1 (MAML1) and of miR-29b was also questioned, as miRNA nuclear
myocyte enhancer factor 2C (MEF2C) are derepressed, thereby promoting myoblast
locali­zation was found to be primarily dependent on
differentiation191. The pseudogene PTEN pseudogene 1 (PTENP1) shares many miRNA
target sites with the tumour suppressor PTEN mRNA and can derepress PTEN in human the nuclear presence of targets155. These findings sug-
cells183. Similarly, the mouse zinc-finger E-box binding homeobox 2 (Zeb2) mRNA can gest that our understanding of nuclear miRNA activity
derepress PTEN193. c | Different viral mechanisms affect miRNA function. Hepatitis C virus is still very partial.
(HCV) harbours two binding sites for miR-122 at the very end of the 5′ untranslated
region (UTR) of its RNA genome203. These recruit Argonaute 2 (Ago2)–miR-122 Circulating miRNAs. Circulating miRNAs are poten-
complexes204 to protect the viral RNA from the cellular antiviral response and the activity tial cancer biomarkers221; their discovery in exosomes
of exonucleases205 and functionally sequester miR-122 and derepress hepatic miR-122 led to the hypothesis that they might contribute to
target mRNAs207. Bovine viral diarrhoea virus (BVDV) is an RNA virus that contains a intercellular signalling222 (Fig. 6a). It remains unclear
binding site for miR-17 in its 3′ UTR; miR-17 binding increases the stability of the viral whether only a small fraction of circulating miRNAs
RNA. The site functionally sequesters the miRNA and derepresses its cellular targets208.
travel within exosomes (~10% or less in plasma)223,224 or
Herpesvirus saimiri (HVS) produces short non-coding RNAs (ncRNAs) termed HVS
uracyl-rich RNAs (HSURs), two of which are known to modulate miRNA function. HSUR1 whether exosomes contain the majority of circulating
binds miR-27a and promotes its target RNA-directed miRNA degradation (TDMD), which miRNAs (83–99% in serum)225. These apparently con-
leads to derepression of miR-27a cellular targets and promotes T cell activation212. trasting observations could be due to technical differ-
HSUR2 binds miR-142-3p and miR-16 and tethers them to cellular target mRNAs, which ences in exosome isolation or to differences between
prevents apoptosis214. IRES, internal ribosome entry site. serum and plasma225.

NATuRe RevIewS | MOleculAR Cell BIOlOgy volume 20 | JANUARY 2019 | 31

Reviews

a b Mechanisms of miRNA RBP–exomotif-

exosomal sorting mediated
Source cell Human T cells RBP-mediated
miR-198 HEK293T cells
5′ GGAG miR-223
lncRNA-mediated
Exosome hnRNPA 2/B1 5′
Prostate cancer cells YB1
AGO miR-149-3p Murine hepatocytes Mouse colon tumours
Diffusion to neighbouring
cells, or to distant cells lncRNA 5′ GGCU miR-193a
Receptor- via circulation lncRNAs harbouring hnRNP Q 5′
miRNA seed targets MVP
independent uptake

Receptor- Ago2-mediated Endosome pathway 3′-nt directed

mediated uptake Human B cells
5′ U
Release of miRNAs or U
AGO–miRNAs into target cells 5′
Target cell
Inhibited by KRAS-
3′-U miRNAs enriched
dependent Ago2 S387
in exosomes compared
phosphorylation
with 3′-A miRNAs
Repression AAA in isogenic colon
of target mRNA cancer cells A
5′
Multivesicular endosome

Fig. 6 | Mechanisms of sorting miRNAs into exosomes. a | MicroRNAs miRNAs into exosomes by binding ‘exomotifs’ at miRNA 3′ ends. The
(miRNAs) can be packaged into exosomes and thus may contribute to exomotif GGAG promotes exosomal sorting of miR-198 by heterogeneous
intercellular signalling. Uptake of exosomes can be receptor-​mediated or nuclear ribonucleoproteins A2/B1 (hnRNP A2/B1) in human primary
receptor-​independent277. Upon entering a target cell, exosome-​delivered T cells233, and the exomotif GGCU promotes hnRNP-​Q-mediated exosomal
miRNAs are thought to regulate target mRNAs222. b | Although the biological sorting of miRNAs in murine hepatocytes234. No exomotifs are known for
function of exosomal miRNAs is still incompletely understood, multiple Y-​box-binding protein 1 (YB1)-mediated sorting of miR-223 in HEK293T
mechanisms direct miRNAs into exosomes. Argonaute 2 (Ago2)-mediated cells235 and for major vault protein (MVP)-mediated sorting of miR-193a in
sorting of specific miRNAs has been reported in isogenic colon cancer cells, mouse colon tumours236. Finally, in human B cells, 3′-adenylated miRNAs are
and phosphorylation of Ago2 Ser387 inhibited loading of some miRNAs into depleted in exosomes whereas 3′-uridylated miRNAs are enriched in
exosomes162. Sorting based on sequence complementarity between exosomes237. Exosomes have been reported to carry proteins, different
miR-149-3p and exosomal long non-​coding RNAs (lncRNAs) was shown in RNAs and miRNA biogenesis components 226 but also Ago2–miRNA
prostate cancer cells232. Exosomal RNA-​binding proteins (RBPs) can direct complexes162 and AGO-​free miRNAs235.

Circulating miRNAs may have regulatory func- nuclear ribonucleoproteins A2/B1 (hnRNP A2/B1) in
tions. Breast cancer cell-derived exosomes contain human primary T cells, thereby directing the miRNAs
pre-miRNAs and the proteins required for cell- to exosomes233. Similarly, the exomotif GGCU, which is
independent miRNA maturation, can induce cell pro- also preferentially located at miRNA 3′ ends, was sug-
liferation in culture226 and might have prometastatic gested to recruit hnRNP Q (also known as SYNCRIP)
properties in vivo227,228. A major concern with these in murine hepatocytes, yielding exosomal secretion of
studies is the technical challenge of clearly distinguishing the bound miRNAs234. Other RBPs proposed to medi-
between miRNA-mediated effects and changes caused ate miRNA sorting into exosomes, such as Y-box-
by other exosomal components229. Moreover, exosomes binding protein 1 (also known as nuclease-sensitive
from five different sources were found to have less element-binding protein 1) for miR-223 in HEK293T
than one miRNA molecule per exosome on average230. cells235 and major vault protein (MVP) for miR-193a in
Nevertheless, a recent study showed that brown adipose colon tumour cells in mice236, lack specific associated
tissue is a major source of exosomal miRNAs in humans recognition motifs. The tumour suppressor miR-193a
and mice in support of a functional role of exosomal is an interesting case because the main reason for its
miRNAs231. Specifically, exosomes derived from a donor sorting into exosomes appears to be its removal from
mouse expressing an exogenous miRNA in brown adi- the cytoplasm236. NTA at miRNA 3′ ends may also
pose tissue were found to repress a reporter gene in the influence sorting, as reported for human B cells and B
liver of an acceptor mouse231. cell-derived exosomes, where adenylation was associ-
A biological function of exosomal miRNAs would ated with miRNA depletion from and uridylation with
require miRNA-specific mechanisms of sorting into miRNA enrichment in exosomes237. Finally, Ago2 Ser387
exosomes (Fig. 6b). One proposed sorting mechanism is phosphorylation by the GTPase KRAS–MAPK pathway
through exosomal lncRNAs and is based on the observed antagonizes exosomal sorting of Ago2 in colon cancer
enrichment in exosomes of prostate cancer cell lines of cells162. Although free, single-stranded miRNAs are
miR-149-3p with lncRNAs harbouring its target sites232. rapidly degraded in the cell and thus are hardly func-
Exosomal RBPs could also promote sorting by directly tional155, some reported sorting mechanisms233,234 focus
Circulating miRNAs interacting with miRNA motifs termed ‘exomotifs’. The on miRNAs, with no mention of AGO proteins, or do not
miRNAs present in circulation
and found either as
exomotif GGAG located at the miRNA 3′ end was sug- detect AGO proteins in exosomes232. Clearly, more stud-
AGO–miRNA complexes or as gested to mediate binding of miR-198 and other exo- ies are required to fully understand the mechanistic and
cargo of vesicles (exosomes). somal miRNAs to the exosomal RBP heterogeneous functional aspects of exosomal miRNAs.

32 | JANUARY 2019 | volume 20 [Link]/nrm

 Reviews

Regulation of miRNA target sites of the added nucleotides that promote miRNA stability
The activity of miRNAs can also be modulated through or degradation nor the mechanism by which the cell
changes in their target sites. RNA editing can alter target-site selects which miRNAs to modify are known. Moreover,
complementarity, and deregulation of target-site editing an extended miRNA 3′ end has been shown to improve
is associated with cancer238,239. Similarly, the t1A binding interactions between miR-122 and HCV genomic RNA113
site on AGO does not recognize N6-methyladenosine and to alter the function of miR-222 to promote apopto-
modification in RNA, raising the possibility that adeno- sis in a breast cancer cell line114, suggesting that 3′ isomir
sine methylation of target mRNAs could subtly modulate variation affects miRNA targeting in ways that remain to
their targeting by miRNAs, although this idea has yet to be understood. Thus, whereas altered seed-targeting has
be tested in a cellular context57. Less subtly, formation of been explored in 5′ isomirs, very little is known about the
mRNA 3′-UTR isoforms can add or remove miRNA target functions and properties of 3′ isomirs.
sites, thereby altering mRNA susceptibility to miRNAs in a The role of mRNA secondary and tertiary struc-
cell-type-specific or tissue-specific manner240. ture in miRNA–target recognition is also largely unex-
RBPs also can modulate miRNA–target interactions. plored. Structures that render target sites inaccessible
Pumilio promotes cell-cycle re-entry of quiescent cells to the miRISC clearly inhibit silencing247,248. However,
by binding the 3′ UTR of the mRNA encoding the miRNA-binding sites have been identified within heav-
tumour suppressor p27 (also known as CDKN1B) and ily structured segments of RNA viruses203,208, raising
remodelling it, thereby exposing target sites for miR-221 the possibility that some structures may not be detri-
and miR-222 (ref.241). AU-rich element binding factor 1 mental for targeting or might even enable recognition
(AUF1; also known as HNRNPD) facilitates interactions by miRNAs. Indeed, effective miR-159 target sites in
of Ago2 with mRNAs in HeLa cells242. The AU-rich ele- Arabidopsis thaliana require an adjoining structural ele-
ment binding protein Hu-antigen R (HuR; also known ment composed of two stem loops249. The interactions
as ELAVL1) derepresses the miR-122-targeted mRNA between RNA structures and AGO and the degree to
encoding high-affinity cationic amino acid transporter 1 which they shape miRNA targeting remain unknown.
(CAT1; also known as SLC7A1) by binding an AU-rich Although the activity of exosomal miRNAs and the
and U-rich region in the mRNA, thereby preventing presence of miRNA-selective exosomal sorting mech-
miRNA function but apparently not miRNA binding85. anisms suggest that exosomal miRNAs participate in
Regulation by HuR appears to be widespread, as over intercellular communication, evidence in a physiologi­
75% of human mRNAs that harbour miRNA-binding cal context remains elusive. The reported scarcity of
sites also possess HuR binding sites, often overlapping miRNAs in exosomes230 and the difficulty of disentangling
or in close proximity243. HuR prevents repression of miRNA-mediated effects from the effects of other exoso-
p53 by miR-125b by binding to the mRNA and causing mal cargoes add to this challenge229. Similarly, although
miRISC dissociation244 and of programmed cell death exosomal Ago2-loaded miRNAs have been detected162,226,
protein 4 (PDCD4) by miR-21 by binding to targets in so were single-stranded miRNAs, which would require
the mRNA and displacing the miRISC and by directly loading into AGO in the target cells 235. Thus, our
binding to miR-21 in MCF cells245. RBPs have also been understanding of exosomal miRNAs is far from complete.
suggested to regulate the Ago2 phosphorylation cycle, Finally, the recent report of molecular condensation
but no specific RBP has been discovered yet to do so170. properties of AGO and TNRC6 (ref.250) connects miRNA
Interestingly, in extracts of most adult mouse tissues, regulation to the growing field of biological phase separ­
miRNAs are found in low-molecular-mass miRISCs ation251. The data demonstrate that miRISCs can form
of ~100 kDa, which is approximately the molecular large molecular condensates in vitro and in living cells,
mass of AGO with its miRNA guide and not bound and it was hypothesized that the ability to form higher
to mRNA. By contrast, in cell lines, most miRNAs are order complexes through molecular condensation may
found in high-molecular-mass miRISCs (up to 2 MDa) allow miRISCs to organize miRNA–target interactions
containing GW182, other proteins and target mRNAs. within the cytoplasm and thereby modulate rates of
Consequently, in transformed cells, miRNA abundance mRNA translation and decay. This hypothesis raises the
may correlate more strongly with miRNA activity than possibility that miRNA activity is regulated through
in primary tissues, where additional regulation of AGO the assembly of the miRISC itself by modulation of the
activity by the cell appears present246. Findings obtained biophysical properties of miRISC components.
in cell lines should therefore be considered with caution. In conclusion, although the first evidence of miRNAs
was discovered over 25 years ago and major advances
Future perspective have been made since, many aspects of the complex
A number of interesting questions remain unanswered mechanisms that govern the activity of these tiny
regarding the regulation of miRNA function. NTA at the 3′ regulators remain to be discovered and explored.
end of miRNAs has been recognized as a widespread pro-
cess131,132. However, neither the tissue-specific importance Published online 14 August 2018

1. Swarts, D. C. et al. The evolutionary journey of 3. Lee, R. C., Feinbaum, R. L. & Ambros, V. The C. elegans mediates temporal pattern formation in C. elegans.
Argonaute proteins. Nat. Struct. Mol. Biol. 21, heterochronic gene lin-4 encodes small RNAs with Cell 75, 855–862 (1993).
743–753 (2014). antisense complementarity to lin-14. Cell 75, 5. Pasquinelli, A. E. et al. Conservation of the
2. Jonas, S. & Izaurralde, E. Towards a molecular 843–854 (1993). sequence and temporal expression of let-7
understanding of microRNA-mediated gene silencing. 4. Wightman, B., Ha, I. & Ruvkun, G. Posttranscriptional heterochronic regulatory RNA. Nature 408, 86–89
Nat. Rev. Genet. 16, 421–433 (2015). regulation of the heterochronic gene lin-14 by lin-4 (2000).

NATuRe RevIewS | MOleculAR Cell BIOlOgy volume 20 | JANUARY 2019 | 33

Reviews

6. Lau, N. C., Lim, L. P., Weinstein, E. G. & Bartel, D. P. 35. Bracht, J., Hunter, S., Eachus, R., Weeks, P. & 62. Grosswendt, S. et al. Unambiguous identification
An abundant class of tiny RNAs with probable Pasquinelli, A. E. Trans-splicing and polyadenylation of miRNA:target site interactions by different types
regulatory roles in Caenorhabditis elegans. Science of let-7 microRNA primary transcripts. RNA 10, of ligation reactions. Mol. Cell 54, 1042–1054
294, 858–862 (2001). 1586–1594 (2004). (2014).
7. Lagos-Quintana, M., Rauhut, R., Lendeckel, W. & 36. Berezikov, E. Evolution of microRNA diversity and 63. Moore, M. J. et al. miRNA-target chimeras reveal
Tuschl, T. Identification of novel genes coding for small regulation in animals. Nat. Rev. Genet. 12, 846–860 miRNA 3′-end pairing as a major determinant of
expressed RNAs. Science 294, 853–858 (2001). (2011). Argonaute target specificity. Nat. Commun. 6,
8. Lee, R. C. & Ambros, V. An extensive class of small 37. Lim, L. P. The microRNAs of Caenorhabditis elegans. 8864 (2015).
RNAs in Caenorhabditis elegans. Science 294, Genes Dev. 17, 991–1008 (2003). 64. Kim, D. et al. General rules for functional microRNA
862–864 (2001). 38. Bartel, D. P. MicroRNAs: target recognition and targeting. Nat. Genet. 48, 1517–1526 (2016).
9. Bernstein, E. et al. Dicer is essential for mouse regulatory functions. Cell 136, 215–233 (2009). 65. Grimson, A. et al. MicroRNA Targeting specificity in
development. Nat. Genet. 35, 215–217 (2003). 39. Wang, Y., Sheng, G., Juranek, S., Tuschl, T. & Patel, D. J. mammals: determinants beyond seed pairing.
10. Chong, M. M. W. et al. Canonical and alternate Structure of the guide-strand-containing argonaute Mol. Cell 27, 91–105 (2007).
functions of the microRNA biogenesis machinery. silencing complex. Nature 456, 209–213 (2008). 66. Broughton, J. P., Lovci, M. T., Huang, J. L., Yeo, G. W.
Genes Dev. 24, 1951–1960 (2010). 40. Nakanishi, K., Weinberg, D. E., Bartel, D. P. & & Pasquinelli, A. E. Pairing beyond the seed supports
11. Pauli, A., Rinn, J. L. & Schier, A. F. Non-coding RNAs Patel, D. J. Structure of yeast Argonaute with guide microRNA targeting specificity. Mol. Cell 64,
as regulators of embryogenesis. Nat. Rev. Genet. 12, RNA. Nature 486, 368–374 (2012). 320–333 (2016).
136–149 (2011). 41. Schirle, N. T. & MacRae, I. J. The crystal structure 67. Guo, H., Ingolia, N. T., Weissman, J. S. & Bartel, D. P.
12. Burger, K. & Gullerova, M. Swiss army knives: of human Argonaute2. Science 336, 1037–1040 Mammalian microRNAs predominantly act to
non-canonical functions of nuclear Drosha and Dicer. (2012). decrease target mRNA levels. Nature 466, 835–840
Nat. Rev. Mol. Cell Biol. 16, 417–430 (2015). 42. Sheu-Gruttadauria, J. & MacRae, I. J. Structural (2010).
13. Kozomara, A. & Griffiths-Jones, S. miRBase: foundations of RNA silencing by Argonaute. 68. Ding, L., Spencer, A., Morita, K. & Han, M. The
annotating high confidence microRNAs using deep J. Mol. Biol. 429, 2619–2639 (2017). developmental timing regulator AIN-1 interacts with
sequencing data. Nucleic Acids Res. 42, D68–D73 43. Diederichs, S. & Haber, D. A. Dual role for argonautes miRISCs and may target the Argonaute protein ALG-1
(2013). in MicroRNA processing and posttranscriptional to cytoplasmic P bodies in C. elegans. Mol. Cell 19,
14. Friedman, R. C., Farh, K. K.-H., Burge, C. B. & regulation of microRNA expression. Cell 131, 437–447 (2005).
Bartel, D. P. Most mammalian mRNAs are conserved 1097–1108 (2007). 69. Liu, J. et al. A role for the P-body component GW182
targets of microRNAs. Genome Res. 19, 92–105 44. Liu, J. et al. Argonaute2 is the catalytic engine in microRNA function. Nat. Cell Biol. 7, 1261–1266
(2009). of mammalian RNAi. Science 305, 1437–1441 (2005).
15. Esteller, M. Non-coding RNAs in human disease. (2004). 70. Meister, G. et al. Identification of novel
Nat. Rev. Genet. 12, 861–874 (2011). 45. Cheloufi, S., Santos, dos, C. O., Chong, M. M. W. & Argonaute-associated proteins. Curr. Biol. 15,
16. Lin, S. & Gregory, R. I. MicroRNA biogenesis pathways Hannon, G. J. A. Dicer-independent miRNA biogenesis 2149–2155 (2005).
in cancer. Nat. Rev. Cancer 15, 321–333 (2015). pathway that requires Ago catalysis. Nature 465, 71. Rehwinkel, J., Behm-Ansmant, I., Gatfield, D. &
17. Bracken, C. P., Scott, H. S. & Goodall, G. J. A 584–589 (2010). Izaurralde, E. A crucial role for GW182 and the
network-biology perspective of microRNA function and 46. Cifuentes, D. et al. A novel miRNA processing pathway DCP1:DCP2 decapping complex in miRNA-mediated
dysfunction in cancer. Nat. Rev. Genet. 17, 719–732 independent of Dicer requires Argonaute2 catalytic gene silencing. RNA 11, 1640–1647 (2005).
(2016). activity. Science 328, 1694–1698 (2010). 72. Braun, J. E., Huntzinger, E., Fauser, M. & Izaurralde, E.
18. Ventura, A. et al. Targeted deletion reveals essential 47. Yekta, S., Shih, I.-H. & Bartel, D. P. MicroRNA-directed GW182 proteins directly recruit cytoplasmic
and overlapping functions of the miR-17~92 family of cleavage of HOXB8 mRNA. Science 304, 594–596 deadenylase complexes to miRNA targets. Mol. Cell
miRNA clusters. Cell 132, 875–886 (2008). (2004). 44, 120–133 (2011).
19. Takamizawa, J. et al. Reduced expression of the let-7 48. Karginov, F. V. et al. Diverse endonucleolytic cleavage 73. Chen, C.-Y. A., Zheng, D., Xia, Z. & Shyu, A.-B.
microRNAs in human lung cancers in association with sites in the mammalian transcriptome depend upon Ago–TNRC6 triggers microRNA-mediated decay by
shortened postoperative survival. Cancer Res. 64, microRNAs, Drosha, and additional nucleases. promoting two deadenylation steps. Nat. Struct. Mol.
3753–3756 (2004). Mol. Cell 38, 781–788 (2010). Biol. 16, 1160–1166 (2009).
20. Lu, J. et al. MicroRNA expression profiles classify 49. Meister, G. et al. Human Argonaute2 mediates RNA 74. Behm-Ansmant, I. mRNA degradation by miRNAs and
human cancers. Nat. Cell Biol. 435, 834–838 (2005). cleavage targeted by miRNAs and siRNAs. Mol. Cell GW182 requires both CCR4:NOT deadenylase and
21. Schwarzenbach, H., Nishida, N., Calin, G. A. & 15, 185–197 (2004). DCP1:DCP2 decapping complexes. Genes Dev. 20,
Pantel, K. Clinical relevance of circulating cell-free 50. Burroughs, A. M. et al. Deep-sequencing of human 1885–1898 (2006).
microRNAs in cancer. Nat. Rev. Clin. Oncol. 11, Argonaute-associated small RNAs provides insight 75. Chekulaeva, M. et al. miRNA repression involves
145–156 (2014). into miRNA sorting and reveals Argonaute association GW182-mediated recruitment of CCR4–NOT through
22. Wang, Y., Goodison, S., Li, X. & Hu, H. Prognostic with RNA fragments of diverse origin. RNA Biol. 8, conserved W-containing motifs. Nat. Struct. Mol. Biol.
cancer gene signatures share common regulatory 158–177 (2011). 18, 1218–1226 (2011).
motifs. Sci. Rep. 7, 1183 (2017). 51. Dueck, A., Ziegler, C., Eichner, A., Berezikov, E. & 76. Fabian, M. R. et al. miRNA-mediated deadenylation is
23. Rupaimoole, R. & Slack, F. J. MicroRNA therapeutics: Meister, G. microRNAs associated with the different orchestrated by GW182 through two conserved motifs
towards a new era for the management of cancer and human Argonaute proteins. Nucleic Acids Res. 40, that interact with CCR4–NOT. Nat. Struct. Mol. Biol.
other diseases. Nat. Rev. Drug Discov. 16, 203–221 9850–9862 (2012). 18, 1211–1217 (2011).
(2017). 52. Wang, D. et al. Quantitative functions of Argonaute 77. Braun, J. E. et al. A direct interaction between
24. Llave, C., Kasschau, K. D., Rector, M. A. & proteins in mammalian development. Genes Dev. 26, DCP1 and XRN1 couples mRNA decapping to 5′
Carrington, J. C. Endogenous and silencing-associated 693–704 (2012). exonucleolytic degradation. Nat. Struct. Mol. Biol.
small RNAs in plants. Plant Cell 14, 1605–1619 53. Betel, D., Koppal, A., Agius, P., Sander, C. & Leslie, C. 19, 1324–1331 (2012).
(2002). Comprehensive modeling of microRNA targets 78. Mathys, H. et al. Structural and biochemical insights
25. Reinhart, B. J., Weinstein, E. G., Rhoades, M. W., predicts functional non-conserved and non-canonical to the role of the CCR4-NOT complex and DDX6
Bartel, B. & Bartel, D. P. MicroRNAs in plants. sites. Genome Biol. 11, R90 (2010). ATPase in microRNA repression. Mol. Cell 54,
Genes Dev. 16, 1616–1626 (2002). 54. Agarwal, V., Bell, G. W., Nam, J.-W. & Bartel, D. P. 751–765 (2014).
26. Mukherjee, K., Campos, H. & Kolaczkowski, B. Predicting effective microRNA target sites in 79. Meijer, H. A. et al. Translational repression and
Evolution of animal and plant Dicers: early parallel mammalian mRNAs. eLife 4, e05005 (2015). eIF4A2 activity are critical for microRNA-mediated
duplications and recurrent adaptation of antiviral RNA 55. Wong, N. & Wang, X. miRDB: an online resource for gene regulation. Science 340, 82–85 (2013).
binding in plants. Mol. Biol. Evol. 30, 627–641 (2012). microRNA target prediction and functional annotations. 80. Fukaya, T., Iwakawa, H.-O. & Tomari, Y. MicroRNAs
27. Kurihara, Y. & Watanabe, Y. Arabidopsis micro-RNA Nucleic Acids Res. 43, D146–152 (2015). block assembly of eIF4F translation initiation
biogenesis through Dicer-like 1 protein functions. 56. Schirle, N. T., Sheu-Gruttadauria, J. & MacRae, I. J. complex in Drosophila. Mol. Cell 56, 67–78
Proc. Natl Acad. Sci. USA 101, 12753–12758 (2004). Structural basis for microRNA targeting. Science 346, (2014).
28. Llave, C., Xie, Z., Kasschau, K. D. & Carrington, J. C. 608–613 (2014). 81. Fukao, A. et al. microRNAs trigger dissociation of
Cleavage of Scarecrow-like mRNA targets directed 57. Schirle, N. T., Sheu-Gruttadauria, J., Chandradoss, S. D., eIF4AI and eIF4AII from target mRNAs in humans.
by a class of Arabidopsis miRNA. Science 297, Joo, C. & MacRae, I. J. Water-mediated recognition Mol. Cell 56, 79–89 (2014).
2053–2056 (2002). of t1-adenosine anchors Argonaute2 to microRNA 82. Elkayam, E. et al. Multivalent recruitment of human
29. Rogers, K. & Chen, X. Biogenesis, turnover, and targets. eLife 4, e07646 (2015). Argonaute by GW182. Mol. Cell 67, 646–658.e3
mode of action of plant microRNAs. Plant Cell 25, 58. Chandradoss, S. D., Schirle, N. T., Szczepaniak, M., (2017).
2383–2399 (2013). MacRae, I. J. & Joo, C. A. Dynamic search process 83. Kuzuoglu-Öztürk, D. et al. miRISC and the CCR4-NOT
30. Borges, F. & Martienssen, R. A. The expanding world underlies microRNA targeting. Cell 162, 96–107 complex silence mRNA targets independently of
of small RNAs in plants. Nat. Rev. Mol. Cell Biol. 16, (2015). 43S ribosomal scanning. EMBO J. 35, 1186–1203
727–741 (2015). 59. Salomon, W. E., Jolly, S. M., Moore, M. J., Zamore, P. D. (2016).
31. Landgraf, P. et al. A mammalian microRNA expression & Serebrov, V. Single-molecule imaging reveals that 84. Eichhorn, S. W. et al. mRNA destabilization is the
atlas based on small RNA library sequencing. Cell 129, Argonaute reshapes the binding properties of its dominant effect of mammalian microRNAs by the time
1401–1414 (2007). nucleic acid guides. Cell 162, 84–95 (2015). substantial repression ensues. Mol. Cell 56, 104–115
32. Ha, M. & Kim, V. N. Regulation of microRNA biogenesis. 60. Klum, S. M., Chandradoss, S. D., Schirle, N. T., Joo, C. (2014).
Nat. Rev. Mol. Cell Biol. 15, 509–524 (2014). & MacRae, I. J. Helix-7 in Argonaute2 shapes the 85. Bhattacharyya, S. N., Habermacher, R., Martine, U.,
33. Lee, Y. et al. MicroRNA genes are transcribed by RNA microRNA seed region for rapid target recognition. Closs, E. I. & Filipowicz, W. Relief of microRNA-mediated
polymerase II. EMBO J. 23, 4051–4060 (2004). EMBO J. 37, 75–88 (2018). translational repression in human cells subjected to
34. Cai, X., Hagedorn, C. H. & Cullen, B. R. Human 61. Helwak, A., Kudla, G., Dudnakova, T. & Tollervey, D. stress. Cell 125, 1111–1124 (2006).
microRNAs are processed from capped, polyadenylated Mapping the human miRNA interactome by CLASH 86. Selbach, M. et al. Widespread changes in protein
transcripts that can also function as mRNAs. RNA 10, reveals frequent noncanonical binding. Cell 153, synthesis induced by microRNAs. Nature 455, 58–63
1957–1966 (2004). 654–665 (2013). (2008).

34 | JANUARY 2019 | volume 20 [Link]/nrm

 Reviews

87. Uhlmann, S. et al. Global microRNA level regulation 112. Cloonan, N. et al. MicroRNAs and their isomiRs proliferative disease. Proc. Natl Acad. Sci. USA 111,
of EGFR-driven cell-cycle protein network in breast function cooperatively to target common biological 11467–11472 (2014).
cancer. Mol. Systems Biol. 8, 1–15 (2012). pathways. Genome Biol. 12, R126 (2011). 139. Jones, M. R. et al. Zcchc11-dependent uridylation of
88. Mestdagh, P. et al. The miR-17-92 microRNA cluster 113. Yamane, D. et al. Differential hepatitis C virus RNA microRNA directs cytokine expression. Nat. Cell Biol.
regulates multiple components of the TGF-β pathway target site selection and host factor activities of 11, 1157–1163 (2009).
in neuroblastoma. Mol. Cell 40, 762–773 (2010). naturally occurring miR-122 3′ variants. Nucleic Acids 140. Gutiérrez-Vázquez, C. et al. 3′ Uridylation controls
89. Saetrom, P. et al. Distance constraints between Res. 45, 4743–4755 (2017). mature microRNA turnover during CD4 T cell activation.
microRNA target sites dictate efficacy and 114. Yu, F. et al. Naturally existing isoforms of miR-222 RNA 23, 882–891 (2017).
cooperativity. Nucleic Acids Res. 35, 2333–2342 have distinct functions. Nucleic Acids Res. 45, 141. Burroughs, A. M. et al. A comprehensive survey of 3′
(2007). 11371–11385 (2017). animal miRNA modification events and a possible role
90. Broderick, J. A., Salomon, W. E., Ryder, S. P., 115. Salter, J. D., Bennett, R. P. & Smith, H. C. The APOBEC for 3′ adenylation in modulating miRNA targeting
Aronin, N. & Zamore, P. D. Argonaute protein identity protein family: united by structure, divergentin function. effectiveness. Genome Res. 20, 1398–1410 (2010).
and pairing geometry determine cooperativity in Trends Biochem. Sci. 41, 578–594 (2016). 142. Rüegger, S. & Großhans, H. microRNA turnover: when,
mammalian. RNA silencing. RNA 17, 1858–1869 116. Blow, M. J. et al. RNA editing of human microRNAs. how, and why. Trends Biochem. Sci. 37, 436–446
(2011). Genome Biol. 7, R27 (2006). (2012).
91. Flamand, M. N., Gan, H. H., Mayya, V. K., Gunsalus, K. C. 117. Kawahara, Y. et al. Frequency and fate of microRNA 143. Guo, Y. et al. Characterization of the mammalian
& Duchaine, T. F. A non-canonical site reveals the editing in human brain. Nucleic Acids Res. 36, miRNA turnover landscape. Nucleic Acids Res. 43,
cooperative mechanisms of microRNA-mediated 5270–5280 (2008). 2326–2341 (2015).
silencing. Nucleic Acids Res. 45, 7212–7225 118. Yang, W. et al. Modulation of microRNA processing 144. Marzi, M. J. et al. Degradation dynamics of microRNAs
(2017). and expression through RNA editing by ADAR revealed by a novel pulse-chase approach. Genome Res.
92. Tsang, J., Zhu, J. & van Oudenaarden, A. deaminases. Nat. Struct. Mol. Biol. 13, 13–21 26, 554–565 (2016).
microRNA-mediated feedback and feedforward loops (2006). 145. Rissland, O. S., Hong, S.-J. & Bartel, D. P. microRNA
are recurrent network motifs in mammals. Mol. Cell 119. Chawla, G. & Sokol, N. S. ADAR mediates destabilization enables dynamic regulation of the
26, 753–767 (2007). differential expression of polycistronic microRNAs. miR-16 family in response to cell-cycle changes.
93. Ebert, M. S. & Sharp, P. A. Roles for microRNAs in Nucleic Acids Res. 42, 5245–5255 (2014). Mol. Cell 43, 993–1004 (2011).
conferring robustness to biological processes. 120. Vesely, C. et al. ADAR2 induces reproducible changes in 146. Monticelli, S. et al. MicroRNA profiling of the
Cell 149, 515–524 (2012). sequence and abundance of mature microRNAs in the murine hematopoietic system. Genome Biol. 6, R71
94. Kim, V. N., Han, J. & Siomi, M. C. Biogenesis of mouse brain. Nucleic Acids Res. 42, 12155–12168 (2005).
small RNAs in animals. Nat. Rev. Mol. Cell Biol. 10, (2014). 147. Krol, J. et al. Characterizing light-regulated retinal
126–139 (2009). 121. Shoshan, E. et al. Reduced adenosine-to-inosine microRNAs reveals rapid turnover as a common
95. Morin, R. D. et al. Application of massively parallel miR-455-5p editing promotes melanoma growth and property of neuronal microRNAs. Cell 141, 618–631
sequencing to microRNA profiling and discovery in metastasis. Nat. Cell Biol. 17, 311–321 (2015). (2010).
human embryonic stem cells. Genome Res. 18, 122. Kawahara, Y., Zinshteyn, B., Chendrimada, T. P., 148. Ameres, S. L. et al. Target RNA-directed trimming
610–621 (2008). Shiekhattar, R. & Nishikura, K. RNA editing of the and tailing of small silencing RNAs. Science 328,
96. Tan, G. C. et al. 5′ isomiR variation is of functional microRNA-151 precursor blocks cleavage by the 1534–1539 (2010).
and evolutionary importance. Nucleic Acids Res. 42, Dicer-TRBP complex. EMBO Rep. 8, 763–769 149. la Mata, de, M. et al. Potent degradation of neuronal
9424–9435 (2014). (2007). miRNAs induced by highly complementary targets.
97. Kim, B., Jeong, K. & Kim, V. N. Genome-wide mapping 123. Kawahara, Y. et al. Redirection of silencing targets by EMBO Rep. 16, 500–511 (2015).
of DROSHA cleavage sites on primary microRNAs and adenosine-to-inosine editing of miRNAs. Science 315, 150. Park, J. H., Shin, S.-Y. & Shin, C. Non-canonical
noncanonical substrates. Mol. Cell 66, 258–269.e5 1137–1140 (2007). targets destabilize microRNAs in human Argonautes.
(2017). 124. Nigita, G. et al. microRNA editing in seed region Nucleic Acids Res. 45, 1569–1583 (2017).
98. Neilsen, C. T., Goodall, G. J. & Bracken, C. P. aligns with cellular changes in hypoxic conditions. 151. Haas, G. et al. Identification of factors involved in
IsomiRs – the overlooked repertoire in the Nucleic Acids Res. 44, 6298–6308 (2016). target RNA-directed microRNA degradation.
dynamic microRNAome. Trends Genet. 28, 544–549 125. Vitsios, D. M., Davis, M. P., van Dongen, S. & Nucleic Acids Res. 44, 2873–2887 (2016).
(2012). Enright, A. J. Large-scale analysis of microRNA 152. Kleaveland, B., Shi, C. Y., Stefano, J. & Bartel, D. P. A.
99. Nishikura, K. A-To-I editing of coding and non-coding expression, epi-transcriptomic features and biogenesis. Network of noncoding regulatory RNAs acts in the
RNAs by ADARs. Nat. Rev. Mol. Cell Biol. 17, 83–96 Nucleic Acids Res. 45, 1079–1090 (2016). mammalian brain. Cell 174, 350–362.e17 (2018).
(2016). 126. Paul, D. et al. A-To-I editing in human miRNAs is 153. Bitetti, A. et al. microRNA degradation by a conserved
100. Telonis, A. G., Loher, P., Jing, Y., Londin, E. & enriched in seed sequence, influenced by sequence target RNA regulates animal behavior. Nat. Struct.
Rigoutsos, I. Beyond the one-locus-one-miRNA contexts and significantly hypoedited in glioblastoma Mol. Biol. 25, 244–251 (2018).
paradigm: microRNA isoforms enable deeper insights multiforme. Sci. Rep. 7, D195 (2017). 154. De, N. et al. Highly complementary target RNAs
into breast cancer heterogeneity. Nucleic Acids Res. 127. Tomaselli, S. et al. Modulation of microRNA editing, promote release of guide RNAs from human
43, 9158–9175 (2015). expression and processing by ADAR2 deaminase in Argonaute2. Mol. Cell 50, 344–355 (2013).
101. Telonis, A. G. et al. Knowledge about the presence or glioblastoma. Genome Biol. 16, 5 (2015). 155. Pitchiaya, S., Heinicke, L. A., Park, J. I., Cameron, E. L.
absence of miRNA isoforms (isomiRs) can successfully 128. Wang, Y. et al. Systematic characterization of A-to-I & Walter, N. G. Resolving subcellular miRNA trafficking
discriminate amongst 32 TCGA cancer types. RNA editing hotspots in microRNAs across human and turnover at single-molecule resolution. Cell Rep.
Nucleic Acids Res. 45, 2973–2985 (2017). cancers. Genome Res. 27, 1112–1125 (2017). 19, 630–642 (2017).
102. Wu, H., Ye, C., Ramirez, D. & Manjunath, N. Alternative 129. Negi, V. et al. Altered expression and editing of 156. Elbarbary, R. A. et al. Tudor-SN-mediated
processing of primary microRNA transcripts by Drosha miRNA-100 regulates iTreg differentiation. endonucleolytic decay of human cell microRNAs
generates 5′ end variation of mature microRNA. Nucleic Acids Res. 43, 8057–8065 (2015). promotes G1/S phase transition. Science 356,
PLoS ONE 4, e7566 (2009). 130. Luciano, D. J., Mirsky, H., Vendetti, N. J. & Maas, S. 859–862 (2017).
103. Lee, H. Y. & Doudna, J. A. TRBP alters human RNA editing of a miRNA precursor. RNA 10, 157. Salzman, D. W. et al. miR-34 activity is modulated
precursor microRNA processing in vitro. RNA 18, 1174–1177 (2004). through 5′-end phosphorylation in response to DNA
2012–2019 (2012). 131. Fernandez-Valverde, S. L., Taft, R. J. & Mattick, J. S. damage. Nat. Commun. 7, 10954 (2016).
104. Fukunaga, R. et al. Dicer partner proteins tune the Dynamic isomiR regulation in Drosophila development. 158. Zeng, Y., Sankala, H., Zhang, X. & Graves, P. R.
length of mature miRNAs in flies and mammals. RNA 16, 1881–1888 (2010). Phosphorylation of Argonaute 2 at serine-387
Cell 151, 533–546 (2012). 132. Wyman, S. K. et al. Post-transcriptional generation of facilitates its localization to processing bodies.
105. Lee, H. Y., Zhou, K., Smith, A. M., Noland, C. L. & miRNA variants by multiple nucleotidyl transferases Biochem. J. 413, 429–436 (2008).
Doudna, J. A. Differential roles of human contributes to miRNA transcriptome complexity. 159. Horman, S. R. et al. Akt-mediated phosphorylation of
Dicer-binding proteins TRBP and PACT in small RNA Genome Res. 21, 1450–1461 (2011). Argonaute 2 downregulates cleavage and upregulates
processing. Nucleic Acids Res. 41, 6568–6576 133. Katoh, T. et al. Selective stabilization of mammalian translational repression of microRNA targets. Mol. Cell
(2013). microRNAs by 3′ adenylation mediated by the 50, 356–367 (2013).
106. Kim, Y. et al. Deletion of human tarbp2 reveals cellular cytoplasmic poly(A) polymerase GLD-2. Genes Dev. 160. Rüdel, S. et al. Phosphorylation of human Argonaute
microRNA targets and cell-cycle function of TRBP. 23, 433–438 (2009). proteins affects small RNA binding. Nucleic Acids Res.
Cell Rep. 9, 1061–1074 (2014). 134. Burns, D. M., D’Ambrogio, A., Nottrott, S. & 39, 2330–2343 (2011).
107. Wilson, R. C. et al. Dicer-TRBP complex formation Richter, J. D. CPEB and two poly(A) polymerases 161. Bridge, K. S. et al. Argonaute utilization for miRNA
ensures accurate mammalian microRNA biogenesis. control miR-122 stability and p53 mRNA translation. silencing is determined by phosphorylation-dependent
Mol. Cell 57, 397–407 (2015). Nature 473, 105–108 (2011). recruitment of LIM-domain-containing proteins.
108. Llorens, F. et al. A highly expressed miR-101 isomiR is 135. Katoh, T., Hojo, H. & Suzuki, T. Destabilization of Cell Rep. 20, 173–187 (2017).
a functional silencing small RNA. BMC Genomics 14, microRNAs in human cells by 3′ deadenylation 162. McKenzie, A. J. et al. KRAS-MEK signaling controls
104 (2013). mediated by PARN and CUGBP1. Nucleic Acids Res. Ago2 sorting into exosomes. Cell Rep. 15, 978–987
109. Manzano, M., Forte, E., Raja, A. N., Schipma, M. J. & 43, 7521–7534 (2015). (2016).
Gottwein, E. Divergent target recognition by 136. Mansur, F. et al. Gld2-catalyzed 3′ monoadenylation of 163. Lopez-Orozco, J. et al. Functional analyses of
coexpressed 5′-isomiRs of miR-142-3p and selective miRNAs in the hippocampus has no detectable effect phosphorylation events in human Argonaute 2.
viral mimicry. RNA 21, 1606–1620 (2015). on their stability or on animal behavior. RNA 22, RNA 21, 2030–2038 (2015).
110. Karali, M. et al. High-resolution analysis of the human 1492–1499 (2016). 164. Shen, J. et al. EGFR modulates microRNA maturation
retina miRNome reveals isomiR variations and novel 137. D’Ambrogio, A., Gu, W., Udagawa, T., Mello, C. C. & in response to hypoxia through phosphorylation of
microRNAs. Nucleic Acids Res. 44, 1525–1540 Richter, J. D. Specific miRNA stabilization by Gld2- AGO2. Nature 497, 383–387 (2013).
(2016). catalyzed monoadenylation. Cell Rep. 2, 1537–1545 165. Yang, M. et al. Dephosphorylation of tyrosine 393
111. Mercey, O. et al. Characterizing isomiR variants (2012). in argonaute 2 by protein tyrosine phosphatase 1B
within the microRNA-34/449 family. FEBS Lett. 591, 138. Boele, J. et al. PAPD5-mediated 3′ adenylation and regulates gene silencing in oncogenic RAS-induced
693–705 (2017). subsequent degradation of miR-21 is disrupted in senescence. Mol. Cell 55, 782–790 (2014).

NATuRe RevIewS | MOleculAR Cell BIOlOgy volume 20 | JANUARY 2019 | 35

Reviews

166. Ma, J.-B. et al. Structural basis for 5′-end-specific 192. Karreth, F. A. et al. The BRAF pseudogene 219. Hwang, H.-W., Wentzel, E. A. & Mendell, J. T. A.
recognition of guide RNA by the A. fulgidus functions as a competitive endogenous RNA and Hexanucleotide element directs microrna nuclear
Piwi protein. Nature 434, 666–670 (2005). induces lymphoma in vivo. Cell 161, 319–332 (2015). import. Science 315, 97–100 (2007).
167. Parker, J. S., Roe, S. M. & Barford, D. Structural insights 193. Karreth, F. A. et al. In vivo identification of tumor- 220. Zisoulis, D. G., Kai, Z. S., Chang, R. K. & Pasquinelli, A. E.
into mRNA recognition from a PIWI domain-siRNA guide suppressive PTEN ceRNAs in an oncogenic Autoregulation of microRNA biogenesis by let-7 and
complex. Nature 434, 663–666 (2005). BRAF-induced mouse model of melanoma. Cell 147, Argonaute. Nature 486, 541–544 (2012).
168. Mazumder, A., Bose, M., Chakraborty, A., 382–395 (2011). 221. Mitchell, P. S. et al. Circulating microRNAs as stable
Chakrabarti, S. & Bhattacharyya, S. N. A transient 194. Tay, Y. et al. Coding-independent regulation of the blood-based markers for cancer detection. Proc. Natl
reversal of miRNA-mediated repression controls tumor suppressor PTEN by competing endogenous Acad. Sci. USA 105, 10513–10518 (2008).
macrophage activation. EMBO Rep. 14, 1008–1016 mRNAs. Cell 147, 344–357 (2011). 222. Valadi, H. et al. Exosome-mediated transfer of mRNAs
(2013). 195. Sumazin, P. et al. An extensive microRNA-mediated and microRNAs is a novel mechanism of genetic
169. Quévillon Huberdeau, M. et al. Phosphorylation of network of RNA-RNA interactions regulates exchange between cells. Nat. Cell Biol. 9, 654–659
Argonaute proteins affects mRNA binding and is established oncogenic pathways in glioblastoma. (2007).
essential for microRNA-guided gene silencing in vivo. Cell 147, 370–381 (2011). 223. Arroyo, J. D. et al. Argonaute2 complexes carry a
EMBO J. 36, 2088–2106 (2017). 196. Barrett, S. P. & Salzman, J. Circular RNAs: analysis, population of circulating microRNAs independent of
170. Golden, R. J. et al. An Argonaute phosphorylation expression and potential functions. Development 143, vesicles in human plasma. Proc. Natl Acad. Sci. USA
cycle promotes microRNA-mediated silencing. 1838–1847 (2016). 108, 5003–5008 (2011).
Nature 542, 197–202 (2017). 197. Hansen, T. B. et al. Natural RNA circles function as 224. Turchinovich, A., Weiz, L., Langheinz, A. & Burwinkel, B.
171. Qi, H. H. et al. Prolyl 4-hydroxylation regulates efficient microRNA sponges. Nature 495, 384–388 Characterization of extracellular circulating microRNA.
Argonaute 2 stability. Nature 455, 421–424 (2008). (2013). Nucleic Acids Res. 39, 7223–7233 (2011).
172. Leung, A., Todorova, T., Ando, Y. & Chang, P. 198. Memczak, S. et al. Circular RNAs are a large class of 225. Gallo, A., Tandon, M., Alevizos, I. & Illei, G. G.
Poly(ADP-ribose) regulates post-transcriptional gene animal RNAs with regulatory potency. Nature 495, The majority of microRNAs detectable in serum and
regulation in the cytoplasm. RNA Biol. 9, 542–548 333–338 (2013). saliva is concentrated in exosomes. PLoS ONE 7,
(2012). 199. Piwecka, M. et al. Loss of a mammalian circular RNA e30679 (2012).
173. Leung, A. K. L. et al. Poly(ADP-ribose) regulates stress locus causes miRNA deregulation and affects brain 226. Melo, S. A. et al. Cancer exosomes perform
responses and microRNA activity in the cytoplasm. function. Science 357, eaam8526 (2017). cell-independent microRNA biogenesis and promote
Mol. Cell 42, 489–499 (2011). 200. Guo, J. U., Agarwal, V., Guo, H. & Bartel, D. P. tumorigenesis. Cancer Cell 26, 707–721 (2014).
174. Seo, G. J. et al. Reciprocal inhibition between Expanded identification and characterization of 227. Fong, M. Y. et al. Breast-cancer-secreted miR-122
intracellular antiviral signaling and the RNAi machinery mammalian circular RNAs. Genome Biol. 15, 101 reprograms glucose metabolism in premetastatic
in mammalian cells. Cell Host Microbe 14, 435–445 (2014). niche to promote metastasis. Nat. Cell Biol. 17,
(2013). 201. Webster, D. P., Klenerman, P. & Dusheiko, G. M. 183–194 (2015).
175. Smibert, P., Yang, J.-S., Azzam, G., Liu, J.-L. & Hepatitis C. Lancet 385, 1124–1135 (2015). 228. Zhang, L. et al. Microenvironment-induced PTEN
Lai, E. C. Homeostatic control of Argonaute stability 202. Otto, G. A. & Puglisi, J. D. The pathway of HCV loss by exosomal microRNA primes brain metastasis
by microRNA availability. Nat. Struct. Mol. Biol. 20, IRES-mediated translation initiation. Cell 119, outgrowth. Nature 527, 100–104 (2015).
789–795 (2013). 369–380 (2004). 229. Tkach, M. & Théry, C. Communication by extracellular
176. Bronevetsky, Y. et al. T cell activation induces 203. Jopling, C. L., Schütz, S. & Sarnow, P. Position-dependent vesicles: where we are and where we need to go.
proteasomal degradation of Argonaute and rapid function for a tandem microRNA miR-122-binding Cell 164, 1226–1232 (2016).
remodeling of the microRNA repertoire. J. Exp. Med. site located in the hepatitis C virus RNA genome. 230. Chevillet, J. R. et al. Quantitative and stoichiometric
210, 417–432 (2013). Cell Host Microbe 4, 77–85 (2008). analysis of the microRNA content of exosomes.
177. Sahin, U., Lapaquette, P., Andrieux, A., Faure, G. & 204. Shimakami, T. et al. Stabilization of hepatitis C virus Proc. Natl Acad. Sci. USA 111, 14888–14893 (2014).
Dejean, A. Sumoylation of human Argonaute 2 at RNA by an Ago2-miR-122 complex. Proc. Natl Acad. 231. Thomou, T. et al. Adipose-derived circulating miRNAs
lysine-402 regulates its stability. PLoS ONE 9, Sci. USA 109, 941–946 (2012). regulate gene expression in other tissues. Nature
e102957 (2014). 205. Sedano, C. D. & Sarnow, P. Hepatitis C virus subverts 542, 450–455 (2017).
178. Josa-Prado, F., Henley, J. M. & Wilkinson, K. A. liver-specific miR-122 to protect the viral genome 232. Ahadi, A., Brennan, S., Kennedy, P. J., Hutvágner, G. &
SUMOylation of Argonaute-2 regulates RNA from exoribonuclease Xrn2. Cell Host Microbe 16, Tran, N. Long non-coding RNAs harboring miRNA seed
interference activity. Biochem. Biophys. Res. Commun. 257–264 (2014). regions are enriched in prostate cancer exosomes.
464, 1066–1071 (2015). 206. Amador-Cañizares, Y., Bernier, A., Wilson, J. A. & Sci. Rep. 6, 24922 (2016).
179. Li, S. et al. TRIM65 regulates microRNA activity by Sagan, S. M. miR-122 does not impact recognition of 233. Villarroya-Beltri, C. et al. Sumoylated hnRNPA2B1
ubiquitination of TNRC6. Proc. Natl Acad. Sci. USA the HCV genome by innate sensors of RNA but rather controls the sorting of miRNAs into exosomes through
111, 6970–6975 (2014). protects the 5′ end from the cellular pyrophosphatases, binding to specific motifs. Nat. Commun. 4, 1–10
180. Eystathioy, T. et al. A phosphorylated cytoplasmic DOM3Z and DUSP11. Nucleic Acids Res. 46, (2013).
autoantigen, GW182, associates with a unique 5139–5158 (2018). 234. Santangelo, L. et al. The RNA-binding protein
population of human mRNAs within novel cytoplasmic 207. Luna, J. M. et al. Hepatitis C virus RNA functionally SYNCRIP is a component of the hepatocyte exosomal
speckles. Mol. Biol. Cell 13, 1338–1351 (2002). sequesters miR-122. Cell 160, 1099–1110 (2015). machinery controlling microRNA sorting. Cell Rep. 17,
181. Huang, K. L., Chadee, A. B., Chen, C. Y. A., Zhang, Y. & 208. Scheel, T. K. H. et al. A broad RNA virus survey 799–808 (2016).
Shyu, A. B. Phosphorylation at intrinsically disordered reveals both miRNA dependence and functional 235. Shurtleff, M. J., Temoche-Diaz, M. M., Karfilis, K. V.,
regions of PAM2 motif-containing proteins modulates sequestration. Cell Host Microbe 19, 409–423 (2016). Ri, S. & Schekman, R. Y-Box protein 1 is required to
their interactions with PABPC1 and influences mRNA 209. Bandiera, S., Pfeffer, S., Baumert, T. F. & Zeisel, M. B. sort microRNAs into exosomes in cells and in a
fate. RNA 19, 295–305 (2013). miR-122 — a key factor and therapeutic target in liver cell-free reaction. eLife 5, e19276 (2016).
182. Suzawa, M. et al. Comprehensive identification of disease. J. Hepatol. 62, 448–457 (2015). 236. Teng, Y. et al. MVP-mediated exosomal sorting of
nuclear and cytoplasmic TNRC6A-associating proteins. 210. Lau, C.-C. et al. Viral-human chimeric transcript miR-193a promotes colon cancer progression.
J. Mol. Biol. 429, 3319–3333 (2017). predisposes risk to liver cancer development and Nat. Commun. 8, 14448 (2017).
183. Poliseno, L. et al. A coding-independent function of progression. Cancer Cell 25, 335–349 (2014). 237. Koppers-Lalic, D. et al. Nontemplated nucleotide
geneand pseudogene mRNAs regulatestumour 211. Liang, H.-W. et al. Hepatitis B virus-human chimeric additions distinguish the small RNA composition
biology. Nature 465, 1033–1038 (2010). transcript HBx-LINE1 promotes hepatic injury via in cells from exosomes. Cell Rep. 8, 1649–1658
184. Ebert, M. S., Neilson, J. R. & Sharp, P. A. MicroRNA sequestering cellular microRNA-122. J. Hepatol. 64, (2014).
sponges: competitive inhibitors of small RNAs in 278–291 (2016). 238. Zhang, L., Yang, C.-S., Varelas, X. & Monti, S. Altered
mammalian cells. Nat. Methods 4, 721–726 (2007). 212. Cazalla, D., Yario, T., Steitz, J. A. & Steitz, J. RNA editing in 3′ UTR perturbs microRNA-mediated
185. Salmena, L., Poliseno, L., Tay, Y., Kats, L. & Down-regulation of a host microRNA by a Herpesvirus regulation of oncogenes and tumor-suppressors.
Pandolfi, P. P. A. ceRNA hypothesis: the Rosetta Stone saimiri noncoding RNA. Science 328, 1563–1566 Sci. Rep. 6, 23226 (2016).
of a hidden RNA language? Cell 146, 353–358 (2011). (2010). 239. Nakano, M., Fukami, T., Gotoh, S. & Nakajima, M.
186. Denzler, R., Agarwal, V., Stefano, J., Bartel, D. P. & 213. Marcinowski, L. et al. Degradation of cellular miR-27 A-to-I RNA editing up-regulates human dihydrofolate
Stoffel, M. Assessing the ceRNA hypothesis with by a novel, highly abundant viral transcript is reductase in breast cancer. J. Biol. Chem. 292,
quantitative measurements of miRNA and target important for efficient virus replication in vivo. 4873–4884 (2017).
abundance. Mol. Cell 54, 766–776 (2014). PLoS Pathog. 8, e1002510 (2012). 240. Nam, J.-W. et al. Global analyses of the effect of
187. Thomson, D. W. & Dinger, M. E. Endogenous 214. Gorbea, C., Mosbruger, T. & Cazalla, D. A viral different cellular contexts on microRNA targeting.
microRNA sponges: evidence and controversy. Sm-class RNA base-pairs with mRNAs and recruits Mol. Cell 53, 1031–1043 (2014).
Nat. Rev. Genet. 17, 272–283 (2016). microRNAs to inhibit apoptosis. Nature 550, 241. Kedde, M. et al. A Pumilio-induced RNA structure
188. Jens, M. & Rajewsky, N. Competition between target 275–279 (2017). switch in p27-3′ UTR controls miR-221 and
sites of regulators shapes post-transcriptional gene 215. Buck, A. H. et al. Post-transcriptional regulation of miR-222 accessibility. Nat. Cell Biol. 12, 1014–1020
regulation. Nat. Rev. Genet. 16, 113–126 (2015). miR-27 in murine cytomegalovirus infection. RNA 16, (2010).
189. Denzler, R. et al. Impact of MicroRNA levels, 307–315 (2010). 242. Min, K.-W. et al. AUF1 facilitates microRNA-mediated
target-site complementarity, and cooperativity on 216. Libri, V. et al. Murine cytomegalovirus encodes a miR- gene silencing. Nucleic Acids Res. 45, 6064–6073
competing endogenous RNA-regulated gene 27 inhibitor disguised as a target. Proc. Natl Acad. (2017).
expression. Mol. Cell 64, 565–579 (2016). Sci. USA 109, 279–284 (2012). 243. Mukherjee, N. et al. Integrative regulatory mapping
190. Bosson, A. D., Zamudio, J. R. & Sharp, P. A. 217. Lee, S. et al. Selective degradation of host microRNAs indicates that the RNA-binding protein HuR couples
Endogenous miRNA and target concentrations by an intergenic HCMV noncoding RNA accelerates pre-mRNA processing and mRNA stability. Mol. Cell
determine susceptibility to potential ceRNA virus production. Cell Host Microbe 13, 678–690 43, 327–339 (2011).
competition. Mol. Cell 56, 347–359 (2014). (2013). 244. Ahuja, D., Goyal, A. & Ray, P. S. Interplay between
191. Cesana, M. et al. A long noncoding RNA controls 218. Leung, A. K. L. The whereabouts of microRNA actions: RNA-binding protein HuR and microRNA-125b
muscle differentiation by functioning as a competing cytoplasm and beyond. Trends Cell Biol. 25, 601–610 regulates p53 mRNA translation in response to
endogenous RNA. Cell 147, 358–369 (2011). (2015). genotoxic stress. RNA Biol. 13, 1152–1165 (2016).

36 | JANUARY 2019 | volume 20 [Link]/nrm

 Reviews

245. Poria, D. K., Guha, A., Nandi, I. & Ray, P. S. 259. Bernstein, E., Caudy, A. A., Hammond, S. M. & 272. Leuschner, P. J. F., Ameres, S. L., Kueng, S. &
RNA-binding protein HuR sequesters microRNA-21 to Hannon, G. J. Role for a bidentate ribonuclease in Martinez, J. RNAi: double-stranded RNA directs the
prevent translation repression of proinflammatory the initiation step of RNA interference. Nature 409, ATP-dependent cleavage of mRNA at 21 to 23
tumor suppressor gene programmed cell death 4. 363–366 (2001). nucleotide intervals. EMBO Rep. 101, 314–320 (2006).
Oncogene 35, 1703–1715 (2016). 260. Grishok, A. et al. Genes and mechanisms related to 273. Schwarz, D. S. et al. Asymmetry in the assembly of the
246. La Rocca, G. et al. In vivo, Argonaute-bound RNA interference regulate expression of the small RNAi enzyme complex. Cell 115, 199–208 (2003).
microRNAs exist predominantly in a reservoir of low temporal RNAs that control C. elegans developmental 274. Khvorova, A., Reynolds, A. & Jayasena, S. D.
molecular weight complexes not associated with timing. Cell 106, 23–34 (2001). Functional siRNAs and miRNAs exhibit strand bias.
mRNA. Proc. Natl Acad. Sci. USA 112, 767–772 261. Hutvagner, G. et al. A cellular function for the Cell 115, 209–216 (2003).
(2015). RNA-interference enzyme Dicer in the maturation of 275. Suzuki, H. I. et al. Small-RNA asymmetry is directly
247. Ameres, S. L., Martinez, J. & Schroeder, R. Molecular the let-7 small temporal RNA. Science 293, 834–838 driven by mammalian Argonautes. Nat. Struct. Mol. Biol.
basis for target RNA recognition and cleavage by (2001). 22, 512–521 (2015).
human RISC. Cell 130, 101–112 (2007). 262. Kim, Y.-K., Kim, B. & Kim, V. N. Re-evaluation of 276. Frank, F., Sonenberg, N. & Nagar, B. Structural basis
248. Kertesz, M., Iovino, N., Unnerstall, U., Gaul, U. & the roles of DROSHA, Exportin 5, and DICERin for 5. Nature 465, 818–822 (2010).
Segal, E. The role of site accessibility in microRNA microRNA biogenesis. Proc. Natl Acad. Sci. USA 113, 277. van Niel, G., D’Angelo, G. & Raposo, G. Shedding light
target recognition. Nat. Genet. 39, 1278–1284 E1881–E1889 (2016). on the cell biology of extracellular vesicles. Nat. Rev.
(2007). 263. MacRae, I. J., Zhou, K. & Doudna, J. A. Structural Mol. Cell Biol. 19, 213–228 (2018).
249. Zheng, Z. et al. Target RNA secondary structure is a determinants of RNA recognition and cleavage
major determinant of miR159 efficacy. Plant Physiol. by Dicer. Nat. Struct. Mol. Biol. 14, 934–940 Acknowledgements
174, 1764–1778 (2017). (2007). L.F.R.G. is supported by an Advanced Postdoc Mobility fellow-
250. Sheu-Gruttadauria, J. & MacRae, I. J. Phase transitions 264. Park, J.-E. et al. Dicer recognizes the 5′ end of RNA ship from the Swiss National Science Foundation, project
in the assembly and function of human miRISC. for efficient and accurate processing. Nature 475, number P300PA_177860. I.J.M. is supported by US
Cell 173, 946–957.e16 (2018). 201–205 (2011). National Institutes of Health grants R01-GM104475 and
251. Banani, S. F., Lee, H. O., Hyman, A. A. & Rosen, M. K. 265. Tsutsumi, A., Kawamata, T., Izumi, N., Seitz, H. & R01-GM115649.
Biomolecular condensates: organizers of cellular Tomari, Y. Recognition of the pre-miRNA structure
biochemistry. Nat. Rev. Mol. Cell Biol. 18, 285–298 by Drosophila Dicer-1. Nat. Struct. Mol. Biol. 18, Author contributions
(2017). 1153–1158 (2011). Both authors contributed to researching, discussing, writing
252. Lee, Y. et al. The nuclear RNase III Drosha initiates 266. Tian, Y. et al. A phosphate-binding pocket within the and revising the article.
microRNA processing. Nature 425, 415–419 platform-PAZ-connector helix cassette of human Dicer.
(2003). Mol. Cell 53, 606–616 (2014). Competing interests
253. Denli, A. M., Tops, B. B. J., Plasterk, R. H. A., 267. MacRae, I. J. et al. Structural basis for double-stranded The authors declare no competing interests.
Ketting, R. F. & Hannon, G. J. Processing of primary RNA processing by Dicer. Science 311, 195–198
microRNAs by the Microprocessor complex. Nature (2006). Publisher’s note
432, 231–235 (2004). 268. Lau, P.-W. et al. The molecular architecture of Springer Nature remains neutral with regard to jurisdictional
254. Gregory, R. I. et al. The Microprocessor complex human Dicer. Nat. Struct. Mol. Biol. 19, 436–440 claims in published maps and institutional affiliations.
mediates the genesis of microRNAs. Nature 432, (2012).
235–240 (2004). 269. Zhang, H., Kolb, F. A., Jaskiewicz, L., Westhof, E. & Reviewer information
255. Nguyen, T. A. et al. Functional anatomy of the human Filipowicz, W. Single processing center models for Nature Reviews Molecular Cell Biology thanks A. Pasquinelli
Microprocessor. Cell 161, 1374–1387 (2015). human Dicer and bacterial RNase III. Cell 118, 57–68 and the other anonymous reviewer(s) for their contribution to
256. Kwon, S. C. et al. Structure of human DROSHA. (2004). the peer review of this work.
Cell 164, 81–90 (2016). 270. Rand, T. A., Petersen, S., Du, F. & Wang, X. Argonaute
257. Nicholson, A. W. Ribonuclease III mechanisms of 2 cleaves the anti-guide strand of siRNA during RISC Supplementary information
double-stranded RNA cleavage. WIREs RNA 5, 31–48 activation. Cell 123, 621–629 (2005). Supplementary information is available for this paper at
(2014). 271. Matranga, C., Tomari, Y., Shin, C., Bartel, D. P. &
258. Okada, C. et al. A high-resolution structure of the Zamore, P. D. Passenger-strand cleavage facilitates Related links
pre-microRNA nuclear export machinery. Science assembly of siRNA into Ago2-containing RNAi enzyme [Link]
miRBase: [Link]
326, 1275–1279 (2009). complexes. Cell 123, 607–620 (2005).

NATuRe RevIewS | MOleculAR Cell BIOlOgy volume 20 | JANUARY 2019 | 37