REVIEWS

Gene expression regulation

mediated through reversible
m6A RNA methylation
Ye Fu1, Dan Dominissini1,2,3, Gideon Rechavi2,3 and Chuan He1

Abstract | Cellular RNAs carry diverse chemical modifications that used to be regarded as
static and having minor roles in fine-tuning structural and functional properties of RNAs.
In this Review, we focus on reversible methylation through the most prevalent mammalian
mRNA internal modification, N6-methyladenosine (m6A). Recent studies have discovered
protein writers, erasers and readers of this RNA chemical mark, as well as its dynamic
deposition on mRNA and other types of nuclear RNA. These findings strongly indicate
dynamic regulatory roles that are analogous to the well-known reversible epigenetic
modifications of DNA and histone proteins. This reversible RNA methylation adds a new
dimension to the developing picture of post-transcriptional regulation of gene expression.

Epigenetic modifications
Reversible chemical
modifications on DNA and
histones that regulate gene
expression independently of the
genome sequences and that are
heritable through cell division.

Writers, erasers and

readers
Enzymes or proteins that add,
remove or preferentially bind
to the chemical modifications
at designated DNA or RNA
nucleotides and amino acid
residues of histones.
Department of Chemistry
and Institute for Biophysical
Dynamics, The University
of Chicago, 929 East 57th
Street, Chicago, Illinois
60637, USA.
2
Cancer Research Center,
Chaim Sheba Medical Center,
Tel Hashomer 52621, Israel.
3
Sackler School of Medicine,
Tel Aviv University, Tel Aviv
69978, Israel.
Correspondence to C.H.
email:
chuanhe@uchicago.edu
doi:10.1038/nrg3724
Published online
25 March 2014
1

In the central dogma of molecular biology, genetic information flows from DNA to RNA and then to protein.
Reversible epigenetic modifications occur on genomic
DNA15 and histone proteins69 to substantially regulate
gene expression that defines cell status and that affects
cell differentiation and development (FIG.1). Although
both DNA and proteins are subject to reversible chemical tuning, as we pointed out in 2010, a similar process
on mRNA or other forms of RNA as the third component of the central dogma had been missing 10. RNA has
crucial roles in biological systems not only by passing
genetic information from DNA to protein but also by
regulating various biological processes. The diverse
functions of RNA are accompanied by more than 100
chemical modifications1114, although the functions of
most of these RNA modifications have remained a mystery. Most RNA species were thought to be short lived,
and RNA modifications were considered to be static and
unalterable after their covalent attachment. The central role of RNA in gene expression and the intrinsic
chemical reversibility of certain types of RNA methylation prompted us to raise the question of reversible
RNA modifications in gene expression regulation10. In
this Review, we discuss how, only in the past 23years,
N6-methyladenosine (m6A) has been discovered as the
first example of reversible RNA methylation15,16. We
describe the transcriptome-wide distribution of m6A
in mammalian systems17,18, the identification of protein writers, erasers and readers for this dynamic RNA

methylation, and emerging functions for m6A in several

mechanisms of post-transcriptional regulation of gene
expression (FIG.1).

m6A RNA methylation in eukaryotes

Discovery and quantification of m6A in mRNAs and
long non-coding RNAs in eukaryotes. Discovered in the
1970s, m6A is the most prevalent internal modification
in polyadenylated mRNAs and long non-coding RNAs
(lncRNAs) in higher eukaryotes19. m6A is widely conserved among eukaryotic species that range from yeast,
plants, flies to mammals, as well as among viral RNAs
with a nuclear phase2024. The identified sequence content
of m6A obtained from mutational studies and substrate
preference of the methyltransferase enzyme invitro2527
is [G/A/U][G>A]m6AC[U>A>C]. The amount of m6A in
isolated RNA was estimated to be 0.10.4% of that of adenines (that is, ~35m6A sites per mRNA) in mammals19,28
and ~0.25% in meiotic Saccharomyces cerevisiae 29.
The total amount of m 6A in RNA can be probed
by several methods, including two-dimensional thin
layer chromatography30, dot-blot 15 and high-performance
liquid chromatography coupled with triple-quadrupole
tandem mass spectrometry (HPLCQqQMS/MS)15,16. The

femtomole sensitivity achieved by HPLCQqQMS/MS

makes it a quantitative tool for monitoring m6A dynamics; the purity of mRNA is extremely important for this
measurement because ribosomal RNA (rRNA), small
nuclear RNA (snRNA) and tRNA also containm6A.

NATURE REVIEWS | GENETICS

VOLUME 15 | MAY 2014 | 293

2014 Macmillan Publishers Limited. All rights reserved

REVIEWS
Methyltransferase
An enzyme that transfers a
methyl group to its substrate.
Most methyltransferases use
S-adenosyl-l-methionine (SAM)
as the methyl donor.

Two-dimensional thin layer

chromatography
A technique to separate and
identify nucleosides on
cellulose plates according to
their differential migration
patterns in two different
solvents. The nucleoside is
typically radiolabelled for
detection.

High-performance liquid
chromatography coupled
with triple-quadrupole
tandem mass spectrometry
(HPLCQqQMS/MS). A liquid
chromatography method
coupled with triple-quadrupole
tandem mass spectrometry,
which can quantitatively and
simultaneously monitor
multiple molecular species
according to their
fragmentation patterns.

Distribution of m6A in mammalian mRNAs and long

non-coding RNAs. Before 2012, the genome-wide distribution of m6A was unknown, until two independent
studies developed an m 6A RNA immunoprecipitation
approach followed by high-throughput sequencing
(MeRIPseq) to map the m6A RNA methylomes with
a ~100nucleotide resolution17,18. Briefly, the isolated
mRNA is fragmented, immunoprecipitated using an
m6A-targeted antibody, ligated to sequencing adaptors, reverse-transcribed to cDNA, amplified using
PCR and subjected to high-throughput sequencing
(FIG.2a). The resulting maps have shown that m6A is
widely distributed in more than 7,000 mRNA and 300
non-coding RNA (ncRNA) transcripts in human cells,
and is enriched around stop codons, in 3 untranslated
regions (3UTRs) and within internal long exons (FIG.2b).
Additionally, the presence of m6A in introns suggests
that this modification can be added either before or at
the same time as RNA splicing. Many m6A peaks are
well conserved between humans and mice, and dynamic
changes of certain peaks have been observed under different stress conditions. A potential link between m6A
and microRNA-target sites was also suggested18. Several
lncRNAs contain m6A, which indicates that certain
ncRNAs transcribed by RNA polymerase II are also subject to m6A methylation. This approach has since been
widely adopted in studies of transcriptome-wide m6A
distribution.

Central dogma

Chemical modications

DNA replication

Reversible
DNA methylation

m5C

hm5C

DNA

Transcription
?
?

RNA

Reversible
RNA methylation

m6A

hm6A

Translation

Protein

Reversible histone
methylation or
acetylation

Me

Ac

Figure 1 | Reversible chemical modifications that regulate the flow of genetic

Naturefrom
Reviews
Genetics
information. In the central dogma, genetic information is passed
DNA |to
RNA
and then to protein. Epigenetic DNA modifications (for example, the formation of
5methylcytosine (m5C; also known as 5mC) and 5hydroxymethylcytosine (hm5C;
also known as 5hmC)) and histone modifications (for example, methylation (me) and
acetylation (ac)) are known to have important roles in regulating cell differentiation
and development. Reversible RNA modifications (for example, the formation of
N6-methyladenosine (m6A) and N6-hydroxymethyladenosine (hm6A)) add an additional
layer of dynamic regulation of biological processes.

Distribution of m 6A in mRNA in meiotic yeast. In

S.cerevisiae, m6A has an important role in the initiation
of meiosis, which is induced by nitrogen starvation31.
Although yeast cells lack m6A (or contain very little of
it) in mRNA during the mitotic log phase, they begin to
accumulate high levels of m6A during nitrogen starvation29. Genome-wide mapping of m6A has revealed 1,308
methylation sites in 1,183 transcripts in an ndt80deficient
(ndt80/) strain that was arrested during meiotic G2
phase or prophase32. The MeRIPseq approach was further optimized through the use of shorter mRNA fragments and an ime4/ strain (which lacks the inducer of
meiosis4 (Ime4) methyltransferase) as a negative control
to obtain a map of m6A at higher resolution. The methylated transcripts were found to be less structured, and most
of them encode functions that are particularly related to
meiosis. These transcripts have a consensus sequence
of ANRGm6ACNNU (where R denotes A or G, and N
represents any nucleotide), and their methylation sites
are enriched at the 3 ends, which is similar to those of
the mammalian systems. Thus, the distribution pattern
of m6A and the consensus sequence of these sites seem
to be conserved to a large extent from yeast tohumans.
Quantitative detection of the m6A fraction with singlenucleotide resolution. The antibody-based profiling of
m6A could not provide information at single-nucleotide
resolution: although this method determines RNA fragments that harbour m6A, the difficulty in distinguishing
m6A from unmodified adenines by sequencing hinders
the pinpointing of m6A sites within these fragments.
Multiple RRACH (where HdenotesU, A or C) motifs
could be present adjacent to each other, and the modification may also occur at non-consensus sites. In addition, antibody-based profiling cannot reveal the fraction
of cellular RNAs that are modified at each specific site.
Traditionally, radioactive labelling was used to detect
the modification site; however, this procedure is long
and tedious. A digestion-based method has recently
been developed to determine the percentage of m6A at a
specific site with single-nucleotide resolution33. Termed
site-specific cleavage and radioactive labelling followed
by ligation-assisted extraction and thin layer chromatography (SCARLET), this method uses RNase H guided
by a sequence-specific 2OMe/2-H chimeric oligonucleotide to cleave the 5 end of the candidate site for
subsequent labelling and detection. Application of this
method to two lncRNAs and three mRNAs revealed
genuine m6A sites and quantified the methylation fractions (1177%). These results, together with the 20%
of m6A modification that was previously reported in
bovine prolactin mRNA34, indicate that many m6A sites
in mRNA and lncRNA are incompletely methylated. In
fact, the majority of m6A consensus sequence sites are
not methylated in mammalian mRNA17,18.
Methylation on the N6 position of adenosine
slightly reduces the stability of WatsonCrick A:U
base pairing 35, but it does not noticeably block the
extension activities of most reverse transcriptases.
Therefore, methods based on primer extension cannot
be readily used to map precise modification positions.

294 | MAY 2014 | VOLUME 15

www.nature.com/reviews/genetics
2014 Macmillan Publishers Limited. All rights reserved

REVIEWS
a

Me
A

mRNA

a genome scale is still limited, and a method for the

genome-wide mapping of m6A with single-nucleotide
resolution thus remains highly desirable.

Me

An

Fragmentation
to ~100 nucleotides
A

Me
A

Me

A
A

Immunoprecipitation
with m6A-specic
antibodies

Input control

Me
Me

Me

m6A writers in eukaryotes

METTL3 is an active component of the m6A methyltransferase complex in mammalian cells. m6A mRNA
methylation is catalysed by a multicomponent methyl
transferase complex, which was originally isolated as
~200kDa and ~800kDa subcomplexes from HeLa
nuclear extracts25,38. A 70kDa protein METTL3 (known
as MTA70 when identified) was the only known component characterized38. METTL3 is highly conserved in
eukaryotes from yeast to humans (FIG.3a). Knockdown
of METTL3 in HeLa cells led to a ~30% decrease of the
total m6A level, and the same knockdown experiment
in HepG2 cells induced apoptosis, possibly through the
activation of the p53mediated pathway 17,39. A recombinant FLAG-tagged human METTL3 protein has recently
been shown to exhibit a low level of activity by itself.
Other components are required to achieve optimal
activity invitro40.

cDNA library construction and

high-throughput sequencing
m6A RNA immunoprecipitation signal

Signal

m6A peak

Signal

m6A peak

Input signal

Locus

Locus

b
m6A abundance

mRNA

5 transcription
start site

Long exon

Stop codon

Figure 2 | Profiling of m6A in RNA by m6A RNA immunoprecipitation.

Nature Reviews
| Genetics
Antibody-based N6-methyladenosine (m6A) RNA immunoprecipitation
has been
developed to profile the transcriptome-wide distribution of m6A. a | Isolated mRNA is
fragmented to ~100nucleotides, immunoprecipitated using m6A-specific antibodies,
converted to a cDNA library and subjected to high-throughput sequencing.
Comparison between the immunoprecipitated sample and the input sample identifies
m6A signal peaks. b | Transcriptome-wide profiling of m6A in mRNA revealed that m6A
is enriched around stop codons, at 3 untranslated regions and within long exons.
The 5 cap contains the N6,2Odimethyladenosine (m6Am) modification, which can
also be enriched using the m6A-specific antibody. Me, methyl group.

However, Thermus thermophilus DNA polymerase I36

and HIV reverse transcriptase37 show kinetic differences when extending opposite m6A compared with
unmodified adenines, and they could be used to map
m6A positions36,37. The application of this approach at

METTL14 is another active component of the m6A

methyltransferase complex and forms a stable hetero
complex with METTL3. A phylogenetic analysis of the
METTL3 family of methyltransferases in the human
genome identified METTL14 and METTL4 as close
homologues of METTL3 with a conserved motif that
contains either Asp-Pro-Pro-Trp or Glu-Pro-Pro-Leu41
(FIG. 3a) . We found that knockdown of METTL14,
but not METTL4, leads to decreased m6A levels in
HeLa and 293FT cells40. Biochemical characterization
revealed that these two proteins form a stable complex
with a stoichiometric ratio of 1:1. Although the methylation activity of METTL14 is slightly higher than that of
METTL3 invitro, the combination of both methyltrans
ferases leads to a substantially enhanced methylation
activity. This heterodimer also shows a strong preference for the cognate m6A consensus sequence and a
modest preference for less structured RNA invitro.
METTL3 and METTL14 colocalize in nuclear speckles,
and the heterodimer forms the core of the mammalian methyltransferase complex. Additional features
of METTL14 include glycinerich sequences in its carboxyl terminus and a potential coiled-coil in its amino
terminus (FIG.3a), which may participate in protein
protein interactions in nuclear speckles. The binding
sites of METTL3 and METTL14 in substrate RNAs,
as shown by a photoactivatable ribonucleoside-enhanced
crosslinking and immunoprecipitation (PARCLIP) assay,
contain a similar consensus sequence to that known
for m6A (FIG.3b). Interestingly, silencing of the methyltransferase complex led to an increase in the abundance
and half-lives of their target RNAs, which is consistent
with an emerging role for m6A as a negative regulator of gene expression (see below). A related study in
mouse embryonic stem cells (mESCs) also indicates that
METTL3 and METTL14 work as a complex 42.
The discovery of the second active methyltransferase
component in the core complex raises the following

NATURE REVIEWS | GENETICS

VOLUME 15 | MAY 2014 | 295

2014 Macmillan Publishers Limited. All rights reserved

REVIEWS
m6A RNA
immunoprecipitation
An immunoprecipitation
method to selectively enrich
for N6-methyladenosine
(m6A)-containing RNA using
an m6A-targeted antibody.

Nuclear speckles
Nuclear domains located in
the interchromatin regions
of the nucleoplasm and
enriched with pre-mRNA
processing factors.

DPAW

b Consensus motif

SAM-binding

S. cerevisiae Ime4

METTL14
DPPW

SAM-binding

Human METTL3
Coiled-coil

EPPL

SAM-binding

Human METTL14

G-rich
METTL3

c
E. coli AlkB
Coiled-coil

Photoactivatable ribonucleoside-enhanced crosslinking

and immunoprecipitation

Human ALKBH5

(PARCLIP). A biochemical
method that takes advantage
of incorporated photoreactive
ribonucleoside analogues to
identify the binding sites of
RNA-binding proteins in cells.

Human FTO

A-rich
C-terminal domain

AlkB domains Fe(II)-binding

motif (HXDXnH)

Substrate and -KG

binding (RXXXXXR)

WTAP

Extra
loop

METTL3 METTL14
+ other
factors?
H

WTAP

N H

N
N

H
C H
H

N
N

ALKBH5

m6A reader

Functions

m6A

FTO
H

FTO

N C

N
H

OH
N C H
H
N

f6A

hm6A
Readers?

Figure 3 | Reversible m6A methylation of mRNA and other types of nuclear RNA. The N6-methyladenosine (m6A)
Nature Reviews | Genetics
modification is installed by a hetero complex of two methyltransferases METTL3METTL14, assisted by Wilms
tumour1associating protein (WTAP), and can be demethylated by the -ketoglutarate (-KG)-dependent
dioxygenases FTO and ALKBH5. a | Saccharomyces cerevisiae inducer of meiosis4 (Ime4), and human METTL3 and
METTL14 contain the S-adenosyl-l-methionine (SAM)-dependent methyltransferase domain for m6A methylation. The
(D/E)P(P/A)(W/L) active site and the SAM-binding motif are conserved. b | Photoactivatable ribonucleoside-enhanced
crosslinking and immunoprecipitation (PARCLIP) reveals that the binding sites of METTL14 and METTL3 on mRNA
resemble the consensus sequence of m6A in mammalian mRNA. The sequence bound by WTAP moderately overlaps
with those bound by METTL14 and METTL3. c | Mammalian FTO and ALKBH5 contain the active site motif HXDXnH
(where X denotes any amino acid) for Fe(ii) binding, RXXXXXR for both -KG binding and substrate recognition, and an
extra loop that leads to preferential binding of single-stranded over double-stranded nucleic acids68,121,122. Relative to
Escherichia coli AlkB, mammalian ALKBH5 has an aminoterminal alaninerich sequence and a potential coiled-coil
structure that could be important for its localization. FTO contains an extra carboxyterminal domain with a novel fold,
possibly to engage in additional proteinprotein interactions. d | Methylation and demethylation of m6A on RNA are
shown. Whereas ALKBH5 catalyses the direct removal of m6A, FTO can oxidize m6A to N6-hydroxymethyladenosine
(hm6A) and N6-formyladenosine (f6A) sequentially; hm6A and f6A are moderately stable (with half-lives of ~3hours under
physiological conditions) and can be hydrolysed to adenine.

296 | MAY 2014 | VOLUME 15

www.nature.com/reviews/genetics
2014 Macmillan Publishers Limited. All rights reserved

REVIEWS
question: why is the m6A methyltransferase complex
composed of two active components, both of which bind
to the methyl donor S-adenosyl-l-methionine (SAM)?
The hetero complex may allow the selective tuning
of methylation activity through post-translational
modification of each component in order to affect different substrate transcripts, thus having an influence
on different biological pathways. A heterodimer of two
methyltransferase components is required for optimal
activities of several other known RNA methyltransferase complexs4346. Typically, one subunit has a SAMbinding pocket, and the other non-catalytic subunit
either stabilizes the catalytic subunit or enhances its
activity by forming a continuous substrate-binding surface. However, both METTL3 and METTL14 are active.
A crystal structure of this complex will be helpful in
uncovering the synergy between the two enzymes and
the properties associated with each active component.

Yeast two-hybrid screens

A method in which one protein
is fused to the GAL4 activation
domain and the other to the
GAL4 DNA-binding domain,
and both fusion proteins
are introduced into yeast.
Expression of a GAL4regulated
reporter gene indicates that the
two proteins physically interact.

Demethylase
An enzyme that removes a
methyl group from its substrate.

WTAP is the third crucial component of the m6A methyltransferase complex invivo. Yeast two-hybrid screens have
identified FKBP12interacting protein of 37kDa (FIP37;
also known as AT3g54170) in Arabidopsis thaliana47 and
Mum2 in yeast as the partner proteins of the METTL3
homologues in these organisms48. These two proteins
are homologues of the Wilms tumour 1associating protein (WTAP) in humans. WTAP was initially identified
as a splicing factor that binds to the Wilms tumour1
(WT1) protein49, and it is essential for cell cycle progression and early mammalian embryonic development. We
found that knockdown of WTAP leads to a decrease in
the total m6A level in HeLa and 293FT cells40. WTAP
interacts with both METTL3 and METTL14, and colocalizes with the METTL3METTL14 heterodimer in
nuclear speckles to participate in m6A RNA methylation
(FIG.3d). In fact, knockdown of WTAP leads to the largest
decrease in m6A levels in these cell lines, which indicates
that WTAP has important roles in cellular m6A deposition. A PARCLIP assay revealed that WTAP shares
a similar binding sequence of GACU; that is, the sequence
bound by WTAP moderately overlaps with the GGAC
sequence bound by both METTL3 and METTL14
(FIG.3b). As identified by PARCLIP, these targets have
a ~50% overlap with m6A-containing transcripts, which
further indicates that METTL3, METTL14 and WTAP
form the core of the major cellular writer complex of m6A
(REF.40). A large proportion of the binding sites of these
three proteins are found in introns (2934%), which further implies that methylation occurs cotranscriptionally
either before or at the same time as splicing. As WTAP
has been thought to be a splicing factor, a recent study
indicates that the knockdown of WTAP or METTL3
yields different isoforms of m6A-containing transcripts,
which suggests that methylation could affect splicing 50.
How does WTAP enhance the methylation activity
of METTL3 and METTL14 invivo (FIG.3d)? Potentially,
WTAP may help to recruit METTL3 and METTL14 to
their target mRNAs. WTAP has also been shown to be
essential for the nuclear speckle localization of METTL3
and METTL14, which could affect the methylation
efficiency of these proteins50. WTAP, which is known

to interact with many proteins and lncRNAs51, could

also recruit other proteins or enzymes to the methyltransferase complex; these additional factors may affect
the methylation activity and selectivity through direct
interactions or post-translational modifications. Future
research to identify additional factors that interact with
or modify the two methyltransferases is crucial for understanding the selectivity of m6A deposition. We may then
be able to answer questions such as how do cells choose
to methylate certain RNA sites, and how is m6A targeted to 3UTRs and long exons. The potential interplay
between RNA methylation, transcriptional regulation
and splicing could also be further investigated.
Mum2Ime4Slz1 (MIS) complex in yeast mediates
mRNA methylation during meiosis. Ime4 is the homologue of METTL3 in yeast and is crucial for the induction
of yeast sporulation. Two other components of the methylation complex Mum2 and sporulation specific with
a leucine zipper motif protein 1 (Slz1) have been identified through yeast two-hybrid experiments48. Mum2
is homologous to human WTAP, whereas Slz1 lacks
mammalian homologues. Interestingly, the increase in
m6A levels during meiosis is mainly triggered by Ime1
(a master regulator of meiosis), which transcriptionally
induces SLZ1. Ime4 and Mum2 are expressed before the
induction of meiosis, and Slz1 then recruits them from
the cytoplasm to the nucleolus32. This nucleolar localization of the MIS complex is essential for accumulating the
full level of m6A. In contrast to yeast, mammalian cells
lack homologues of Slz1, and the mammalian and plant
methyltransferase complex primarily locates in nuclear
speckles instead of the nucleolus.
METTL3 and WTAP are highly conserved in eukaryotes. In A.thaliana, m6A seems to be mainly found near
3UTRs52; a mutation in MTA (which is the METTL3
homologue in A.thaliana) has been associated with
cell division defects, arrested seeds, reduced apical
dominance and organ abnormality 47. In Drosophila
melanogaster, the METTL3 homologue Ime4 is essential for viability and regulates Notch signalling during
egg chamber development53. In zebrafish, knockdown
of either WTAP or METTL3 leads to multiple developmental defects, and knockdown of both proteins leads
to increased apoptosis50.

m6A erasers in mammals

Demethylation of m6A in RNA by FTO. In 2011, the discovery of -ketoglutarate-dependent dioxygenase FTO
as the first RNA demethylase was an important breakthrough in reigniting investigations of m6A biology 15.
The Fto gene was initially discovered in a deletion of
four genes that led to a fused-toe phenotype in mutant
mice54. In 2007, three independent studies revealed that
a single-nucleotide polymorphism in the first intron of
FTO strongly associates with body mass index and the
risk of obesity in multiple populations5557. In adult mice,
Fto has the highest expression level in the brain, particularly within the hypothalamus58. Deletion or overexpression of Fto in mouse models has been associated with
altered body weight or food intake59,60. Fto also affects

NATURE REVIEWS | GENETICS

VOLUME 15 | MAY 2014 | 297

2014 Macmillan Publishers Limited. All rights reserved

REVIEWS
development: Fto-knockout mice shows increased postnatal lethality and growth retardation59, and a homozygous lossoffunction mutation (Arg316Gln) in the FTO
protein in humans leads to postnatal retardation, as well
as multiple dysmorphisms and malformations61.
FTO is a member of the Fe(ii) and -ketoglutaratedependent AlkB family of proteins that catalyse oxidative
demethylation 58 ; close homologues participate in
epigenetic regulation, such as oxidative DNA demethylation6264 and histone demethylation8. FTO was originally shown to demethylate N3-methylthymidine in
single-stranded DNA58 and N3-methyluridine in singlestranded RNA65 invitro; however, the function of FTO
invivo remained unknown until we discovered that
FTO efficiently demethylates m6A in both RNA and
DNA invitro15. Further experiments showed that silencing of FTO in HeLa and 293FT cells increased total m6A
levels in polyadenylated RNA, and overexpression of
FTO decreased m6A levels on RNA15. FTO is expressed in
dot-like patterns in the nucleoplasm and partially colocalizes with nuclear speckles. These cell-based results,
together with observations that most mammalian cells
and tissues contain very low levels (a few parts per million)
of m6A on genomic DNA, led us to conclude that m6A
on nuclear RNA (including mRNA, lncRNA and possibly other types of RNA) is the main substrate of FTO.
Recent work showed that m6A on three mRNA species
could be demethylated by FTO invivo, and this function
seems to affect neuronal activities66. FTO may also act
as a nutrient sensor, which could modulate its demethylation activities67. It should be noted that although
FTO works preferentially on single-stranded RNA
and DNA, it can still exhibit demethylation activity,
albeit low, towards double-stranded RNA andDNA15.
The crystal structure of the FTO protein reveals
an active domain that is similar to those of other proteins of the AlkB family 68 (FIG.3c). FTO also contains a
Cterminal domain with a novel fold that is distinct from
other proteins of this family. This Cterminal domain
may engage in additional proteinprotein or protein
RNA interactions to affect the function of FTO. The
discovery of FTO as an m6A demethylase strongly suggests functional roles for m6A in human developmental
regulation; however, to achieve the end goal of uncovering the underlying mechanism, a considerable amount
of future work is required to identify the physiological
RNA targets of FTO and to elucidate the functional
consequences of such demethylation.

Oxidative demethylation
A chemical reaction in which
the CH bond of a methyl
group attached to a nitrogen or
an oxygen atom is oxidized to
OH by demethylases, and the
intermediate decomposes to
release the methyl group as
formaldehyde.

Demethylation of m6A in RNA by ALKBH5. ALKBH5

is another protein of the AlkB family that shows efficient demethylation activity towards m 6A in mRNA
and other types of nuclear RNA16,69. ALKBH5 has an
alanine-rich sequence and a potential coiled-coil structure in its Nterminus (FIG.3c), which may be important
for its localization. ALKBH5 knockdown in human cell
lines led not only to increased total m6A levels on polyadenylated RNA but also to accelerated export of these
RNAs from the nucleus to the cytoplasm16. However,
the underlying mechanism is not fully understood.
ALKBH5 and its demethylation activity affect nascent

mRNA synthesis and the rate of splicing 16. Unlike FTO,

direct immunoprecipitation of ALKBH5 has identified
bound RNA substrates16, and ALKBH5 has been shown
to be part of the mRNA-bound proteome70, which suggests a tight interaction with mRNA and other RNA
substrates. ALKBH5 also colocalizes well with nuclear
speckles in an RNase Asensitive manner. Alkbh5 has the
highest expression level in mouse testes. Consistently,
Alkbh5knockout male mice exhibit aberrant spermatogenesis, which is probably a result of altered expression
of spermatogenesis-related genes16.
hm6A and f6A modifications on mammalian mRNA.
While investigating FTO-catalysed demethylation, we observed two unprecedented intermediates, N 6 -hydroxymethyladenosine (hm 6 A) and
N 6-formyladenosine (f 6A), which were generated
through the FTO-catalysed oxidation of m6A (REF.71)
(BOX 1;FIG.3d). The hm6A intermediate is a direct oxidation product of m6A, and f 6A is the further oxidized
product of hm6A. Both hm6A and f 6A can decompose in
water to yield unmethylated adenine and formaldehyde
(from hm6A) or formic acid (from f 6A). To our surprise,
these modifications are metastable under physiological
conditions in neutral buffered solutions at 37C with
half-lives of ~3hours. This observation raised the possibility that both modifications could exist in living
cells and could have functional implications, given that
median mammalian RNA half-lives are ~5hours72,73.
Indeed, using a modified protocol to avoid acid, base
and heating treatments, we have detected the presence
of these modifications in mRNA isolated from mouse
tissue and human cell lines71. The exact sources of these
modifications invivo remain unknown so far; however,
these new modifications carry functional groups that
are distinct from m6A and could substantially affect
RNAprotein interactions.
Although both FTO and ALKBH5 are mainly found
in the nucleus, the possibility that both proteins could
translocate to the cytoplasm under certain circumstances
should not be ruled out. Cytoplasmic RNA may also be
demethylated by these enzymes or by other currently
unknown demethylases. ALKBH5 is only conserved in
vertebrates from fish to humans, whereas FTO is conserved in vertebrates and has homologues in marine
algae74. As there are many Fe(ii) and -ketoglutaratedependent dioxygenases with unknown functions in various organisms75, we should not be surprised to see the
discovery of more m6A demethylases. In addition to its
occurrence in eukaryotic mRNA, m6A also exists in various classes of RNA in eukaryotes, bacteria and archaea,
including rRNA, tRNA and snRNA 14. Furthermore,
chemical modifications can also occur on various nitrogen, carbon and oxygen atoms within the bases and
backbone of RNA13 (BOX 1). These modifications (for
example, methylation) on the heteroatoms oxygen and
nitrogen can, in principle, be enzymatically reversed
through the oxidative demethylation mechanism used
by FTO and ALKBH5 or through nucleophilic substitutions (BOX 1). Demethylases that remove these other RNA
methylations could exist and exhibit functionalroles.

298 | MAY 2014 | VOLUME 15

www.nature.com/reviews/genetics
2014 Macmillan Publishers Limited. All rights reserved

REVIEWS
Box 1 | RNA modifications
Cellular RNA species contain more than 100 chemical modifications with
diverse properties. Chemical modifications of RNA can occur on the N1,
N3, N7 and C8 atoms in both adenine and guanine; C2 and N6 in adenine;
N2 and O6 in guanine; N1, O2, N3 and C5 in cytosine and uracil; N4 in
cytosine and O4 in uracil; as well as on 2O of the ribose backbone and the
OH group of the phosphate backbone (see the figure, part a). These
modifications can modulate hydrophobicity, steric and electrostatic
effects, and hydrogen-bonding abilities of RNA bases and backbones.
Methylation or other forms of alkylation on nitrogen or oxygen atoms can
be removed through either an oxidative or a nucleophilic substitution
mechanism. The oxidative demethylation (see the figure, part b) is best
exemplified by Fe(ii) and -ketoglutarate-dependent dioxygenase
enzymes, which use Fe(ii) as a catalytic centre, O2 as an oxidant and

-ketoglutarate as a cofactor. When the methyl group is linked to

a heteroatom such as nitrogen or oxygen, the oxidation of CH to a
hemiaminal or hemiacetal intermediate destabilizes the CN or CO
bond, respectively, which leads to the demethylated product with the
release of formaldehyde. The hemiaminal intermediate, such as
N6-hydroxymethyladenosine (hm6A), may undergo further oxidation to
produce a formamide, such as N6-formyladenosine (f 6A), which can
decompose in water to yield the demethylated product with the release of
formic acid. The demethylation activity could be modulated by the
effective concentrations of Fe(ii), O2 or -ketoglutarate. The bimolecular
nucleophilic substitution (Sn2) mechanism could also be used to remove
RNA methylation on heteroatoms; however, such a process has yet to be
shown for RNA demethylation (see the figure, part c).

a
NH2
N

NH2

Ribose

NH
N

Ribose
A

HC

(e.g. f6A)

NH

NH2

(e.g. Adenine)

Ribose O

Base

OH

Ribose

Ribose

Phosphate

RNA backbone

OH

(Formic acid)

Oxidation
CH3

H2C

OH

(e.g. m6A)

(e.g. hm6A)

Oxidation

N CH3
H
N

(e.g. Adenine)
+

OH
O

CH3

CH2

N H

:Nu

CH3

Nu

CH3

(Formaldyhyde)

m6A, N6-methyladenosine; Nu, nucleophile.

Nature Reviews | Genetics

m6A reader proteins and effector functions

The discoveries of m 6A RNA demethylation and
demethylases validate our hypothesis that the ubiquitous m6A modification is dynamic and reversible, which
is similar to epigenetic DNA and histone modifications.
The noticeable phenotypes of both FTO and Alkbh5
mutations in humans and mice strongly indicate the
functional importance of this reversible m6A methylation on RNA. For the m6A group to have a biological
function, it needs to be recognized through reading
by specific proteins. This process could resemble the
roles of proteins that read 5methylcytosine (m5C; also
known as 5mC) in DNA, or methylated or acetylated
amino acid residues of histones in order to exhibit
the biological function associated with the modifications and to enable reversible tuning. We can envision
three types of selective reading mechanisms for m6A on
RNA: first, a reader protein could selectively bind to the

m6A-containing RNA; second, the presence of m6A in

a specific sequence could weaken the cognate binding
interaction of an RNA-binding protein; and third, the
presence of m6A may change the secondary structures
of RNA and therefore alter proteinRNA interactions.
YTHDF2 preferentially recognizes m 6A-containing
mRNA and regulates both mRNA stability and localization. Using pulldown experiments, we have identified
three cytoplasmic proteins of the YTH domain family, YTHDF13, as selective m6A-binding proteins in
mammalian cell extracts17,76 (FIG. 4a). The YTH domain
family consists of abundant RNA-binding proteins that
previously had no clear function assigned. We confirmed that mammalian YTHDF proteins preferentially
bind to RNA that contains m6A at the G[G>A]m6ACU
consensus sequence relative to unmethylated RNA of
the same sequence76. Additionally, RNA probes that

NATURE REVIEWS | GENETICS

VOLUME 15 | MAY 2014 | 299

2014 Macmillan Publishers Limited. All rights reserved

REVIEWS
a
S. cerevisiae Mrb1

YTH domain

S. pombe Mmi1

YTH domain

Human YTHDF1

P/Q/N-rich

YTH domain

Human YTHDF2

P/Q/N-rich

YTH domain

Human YTHDF3

P/Q/N-rich

YTH domain

Number of dierent mRNAs

600
Non-m6A
m6A

500
400
300
200
100
0
3

12

15

18

mRNA lifetime (hours)

Translation
P/Q/Nrich

?
Me
An

m7Gppp
RNA

Transcription

YTHDF2 YTH
Me
m7Gppp

An

?
DNA
Localization
(strorage or transport)

Degradation

Figure 4 | Functions of the reader (that is, effector) proteins of m6A. a | The
6
Nature
ReviewsHuman
| Genetics
characterized YTHDF proteins serve as N6-methyladenosine (m
A) readers.
YTHDF13 proteins contain a carboxyterminal YTH RNA-binding domain and an
aminoterminal P/Q/N-rich region. The YTH domain protein is conserved in the fission
yeast Schizosaccharomyces pombe and the budding yeast Saccharomyces cerevisiae.
b | The m6A modification is enriched in mRNAs with shorter half-lives in general,
which supports the proposed main role of m6A in regulating mRNA stability. c | The
m6A-specific RNA-binding proteins are engaged in post-transcriptional regulation
of gene expression. YTHDF2 regulates the methylation (me)-dependent RNA
degradation. Other reader proteins may exist and affect RNA splicing, storage,
trafficking and translation. Data in part b courtesy of X.Wang, laboratory of C.H.

Ribosome profiling
Qualitative and quantitative
sequencing of the RNA attached
to ribosomes as a signature of
genes that are expressed.

Processing bodies
(P-bodies). Distinct foci in the
cytoplasm that are enriched
with RNA degradation factors.

contain adenine, hm6A or f 6A, or that have m6A in nonconsensus sequences have decreased binding affinity.
The enrichment of m6A in RNA immunoprecipitated
with YTHDF13 further supports the role of YTHDF
proteins as m6A-specific RNA-binding proteins.
RNA immunoprecipitation and PARCLIP experiments revealed mostly mRNA as targets of YTHDF2,
in addition to some lncRNA targets. The binding sites
localize around stop codons and at 3UTRs with a

conserved GAC[U>A] motif; thus, the occupancy of

YTHDF2 resembles the distribution pattern of m 6A
on mRNA. Notably, the knockdown of YTHDF2 led to
decreased half-lives of these RNA targets but had minor
effects on the mRNA levels in the actively translating
pool. Ribosome profiling further suggests that YTHDF2
alters ribosome occupancy of its mRNA targets. These
results suggest that YTHDF2 has a role in RNA decay.
Fluorescence immunostaining of YTHDF2 and fluorescence insitu hybridization of its cognate mRNA revealed
that YTHDF2 binds to m6A through the Cterminal
YTH domain and localizes the cognate mRNA to
processing bodies (Pbodies) for accelerated degradation
through its Nterminal Pro/Gln/Asnrich domain (FIG. 4).
The exact RNA degradation mechanism needs to be
further elucidated; however, YTHDF2 binds to mRNA
with shorter poly(A) tails and does not seem to affect the
deadenylation process76.
Several cytoplasmic mRNA decay pathways are
known7786. The YTHDF2mediated mRNA degradation, which affects thousands of mRNA molecules, is
a unique process that is dependent on the methylation
of the target mRNA and could therefore be reversibly
tuned through m6A methylation and demethylation.
This discovery, together with the negative correlation
of m6A with mRNA stability in general as revealed by
knockdown of methyltransferases40, suggests one main
function of m6A on RNA: the regulated degradation
of methylated RNA. This process is mediated through
selective m6A recognition and subsequent relocalization by a reader or effector protein. The control of the
stability of the non-translating pool of mRNA (or other
RNA species) through the YTHDF2dependent mechanism could be important under various circumstances
for the selective elimination of a group of RNAs77.
Interestingly, Mmi1 the homologue of YTHDF proteins in Schizosaccharomyces pombe (FIG.4a) is essential for the elimination of meiosis-specific transcripts
during meiosis87. However, the presence of m6A has not
been reported in S.pombe, which lacks homologues of
METTL3 and METTL14. The potential presence of m6A
in mRNA and its functional roles in S.pombe should be
further investigated.
hnRNPs could be potential nuclear m6A readers. Besides
the YTH domain family of proteins and other cytoplasmic mRNA-binding proteins, pulldown experiments
have also identified proteins of the heterogeneous
nuclear ribonucleoprotein (hnRNP) type as potential
m6A-selective binding proteins17. Known to form ribonucleoprotein granules that could affect mRNA localization and transport, hnRNPs could also block binding of
splicing factors and affect alternative splicing. Additional
experiments are required to investigate connections
between hnRNPs andm6A.
Anti-readers of m6A and m6A-derived modifications.
The presence of the methyl group can also disfavour
binding of an RNA-binding protein to the modified
RNA. This mechanism of anti-reading has yet to be
observed for m6A. The m6A modification is widely

300 | MAY 2014 | VOLUME 15

www.nature.com/reviews/genetics
2014 Macmillan Publishers Limited. All rights reserved

REVIEWS
distributed in the 3UTRs of mRNA transcripts a
region bound by numerous RNA-binding proteins that
regulate mRNA metabolism and translation. It is possible that certain anti-reading mechanisms exist to regulate the fate of methylated mRNA. m6A is also known to
protect RNA from recognition by cellular innate immunity proteins. Toll-like receptor 3 (TLR3) and TLR7 recognize unmodified double-stranded and single-stranded
RNA as invasive RNA species and selectively target them
for degradation88,89. The incorporation of m6A and other
RNA modifications in transfected exogenous RNA can
reduce the recognition by innate immune systems to
prevent unnecessary degradation, which increases
their expression. An anti-reading mechanism possibly
operates in this process.
Indirect reading. Certain RNA modifications, such as
pseudouridine (), are known to cause secondary and
tertiary structural changes90. The m6A modification
reduces the base-pairing energy of A:U only marginally 35, but this difference may shift the equilibrium of
certain secondary and tertiary structures of RNA. The
altered structures could have an effect on the binding
of specific proteins, leading to indirect reading and
regulation. In a recent study of HuR (also known as
ELAVL1) a well-known RNA-binding protein that
affects the stability of many mRNA transcripts in mammalian cells9196 the m6A modification affected the
ability of HuR to bind to different RNA probes invitro42.
In this particular case, the RNA structure altered by
methylation might indirectly contribute to the accessibility of the cognate HuR-binding site. However,
the consensus sequence recognized by HuR invivo is
different from the m6A-containing sequence91,92. The
extent and details of the cellular connection between
HuR and m6A still need to be further investigated.
So far, no cellular example is known for this indirect
reading mechanism.

Biological consequences of m6A

The recent breakthroughs in the discovery and characterization of m6A writers, erasers and readers, together
with the parallel development of high-throughput assays
that profile this methylation on a transcriptome-wide
scale, set the stage and provide tools for functional investigations that aim to identify the mechanisms by which
m6A is translated into biological outcomes. Past studies
that used broad-spectrum methylation inhibitors have
yielded inconclusive results. Now that writer, eraser and
certain reader proteins have been clearly defined, perturbation of these machineries can lead to more specific
phenotypic outcomes and experimental observations
that will help to elucidate the biological roles of m6A and
the underlying mechanisms. An instrumental aspect of
this endeavour will be to categorize the phenotypic levels influenced by m6A. First are effects at the levels of
whole organisms or tissues; studies at these levels could
reveal tissue specificity of m6A, as well as its relevance
to certain diseases and biological processes (for example, development, infertility, carcinogenesis, stemness,
meiosis and circadian rhythm). Second are effects at the

pathway level (for example, the p53-mediated pathway,

Notch signalling, nutrient sensing through mammalian
target of rapamycin complex1 (mTORC1) and apoptosis). Third are roles at the machinery level (for example,
the spliceosome and the nuclear export machinery). Last
are functions at the elementary molecular level on which
all other levels depend (for example, thermodynamics
and proteinm6A interactions). The reader proteins and
their associated recognition mechanisms will be crucial
in revealing and understanding theseroles.
Post-transcriptional regulation through methylationdependent localization of the target transcript. To our
knowledge, the indepth characterization of YTHDF2
as the first m6A reader delineates the first established
molecular pathway mediated by m6A: the binding of
YTHDF2 to thousands of mRNA transcripts (and also
to certain ncRNA transcripts) results in the localization
of bound mRNA from the translatable pool to decay
sites, thereby affecting the translation status and halflife of mRNA76. This discovery has two fundamental
merits. First, it indicates that a main function of m6A
methylation as a reversible mark is to affect mRNA
stability, which fits nicely with the negative correlation between m 6A levels and transcript abundance
observed upon silencing of methyltransferases. In fact,
such methylation generally associates with mRNAs that
have shorter half-lives (FIG.4b), which further supports
this notion. Second, this example illustrates how selective reading of the m6A mark by a binding protein can
affect localization of the target RNA, thus providing
a model that applies to other potential readers which
may broadly affect RNA transport, storage, stability,
translation and splicing.
Although transcriptional regulation has major
roles in regulating gene expression, it is protein levels
that mainly determine biological phenotypes. Protein
production is also subjected to various types of posttranscriptional regulation such as through mRNA
secondary structure, microRNAs or mRNA translational
control which probably contributes as much as, if not
more than, transcriptional regulation to determine cellular protein abundance97103. m6A methylation provides
a new dimension of post-transcriptional gene regulation.
The m6A mark on mRNA could affect the abundance,
localization and use of mRNA, and potentially splicing;
all of these represent central processes that are connected
to protein expression (FIG.4c). We believe that m6A has
a substantial contribution to the post-transcriptional
balance that regulates proteinlevels.
Known examples of RNA methylation in regulating cellular processes. Various circadian RNAs and clock output
transcripts contain the m6A modification104. Inhibition
of m6A formation leads to prolonged nuclear retention of circadian RNAs and thus delays the nuclear exit
of mature period circadian clock 2 (Per2) and aryl
hydrocarbon receptor nuclear translocator-like (Arntl;
also known as Bmal1) mRNAs104. This observation is
consistent with our discovery that deletion of ALKBH5
(which increases m6A levels) in mammalian cells leads

NATURE REVIEWS | GENETICS

VOLUME 15 | MAY 2014 | 301

2014 Macmillan Publishers Limited. All rights reserved

REVIEWS
to reduced nuclear retention time16, although it is not
clear whether m6A has a direct cis role in modulating the
export specifically of m6A-containing RNA molecules or
whether it is an indirect consequence of perturbation to
the RNA export machinery.
The m6A RNA modification is involved in priming
yeast cells to bipotential states and meiosis during nitrogen starvation. Through carefully monitoring methylation profiles at different stages, a recent study suggests
that methylation is important for the kinetic control
of RNAs during the meiotic prophase32. Although no
marked change in half-lives has been observed for the
m6A-containing RNAs, the accessibility of these RNAs
to translation may be modulated through interactions with potential reader proteins. Analogous to the
proposed function of m6A in accelerating both RNA
export and degradation in mammalian cells, m6A may
ensure faster turnover of the RNA transcripts that
are important during the meiotic prophase but that are
harmful and need to be degraded after the exit from
prophase. This example suggests that m6A could globally ensure fast kinetic responses by redirecting RNA
to different organelles and by quickly decreasing the
expression of related proteins. A similar mechanism
could also affect eukaryotic mitosis.
Similarly, another study in mESCs revealed that the
m6A methylation accelerates transcript decay, which
is consistent with the main role that we propose for
m6A, and affects stem cell maintenance and differentiation42. Interestingly, when compared with pluripotency-related genes, developmental regulators were
significantly enriched among target genes of METTL3
and METTL14 in mESCs. In particular, m6A destabilizes developmental regulator transcripts, which may
suggest that methylation is important for maintenance
and differentiation of mESCs. Temporal and spatial
regulation of mRNA is known to have crucial roles in
embryonic development. Post-fertilization, maternal
mRNA needs to be degraded in a programmed manner.
The methylation on mRNA could affect this process
through altering the localization and half-lives of target mRNA transcripts. Such methylation could mark
specific sets of RNA species and therefore differentiate
between maternal and zygotic mRNA in a kinetic manner. Heritable information could perhaps be passed
down to generations of cells through orchestrated RNA
methylation and demethylation activities.
Advantage and specificity of the m6A-based regulation. The first m6A reader protein to be characterized is
known to affect more than 3,000 different mRNA transcripts76. We propose that the reversible RNA methylation pathway, in general, has evolved to affect processes
that involve changes in the expression of large groups of
genes. This property is intimately related to the potential advantages of reversible methylation at the RNA
level. Besides providing increased complexity to the
regulatory network, this mechanism may allow rapid
responses to signalling and stimuli when the expression of a group of proteins (which can range from tens
to thousands) needs to be adjusted in a rapid manner;

Figure 5 | RNA methylation could affect various aspects

of RNA metabolism and mRNA translation, and
regulate protein expression post-transcrptionally.
Whereas N6-methyladenosine (m6A) methyltransferases
and demethylases shape the methylation (me)
landscape, the reader proteins bind to the methylated
RNA and mediate specific functions. Various cellular
processes could be affected by m6A RNA methylation.
In the cell nucleus, m6A may affect RNA export, nuclear
retention and splicing, possibly through interactions of
reader proteins with RNA export, retention and splicing
machineries. After RNAs are exported to the cytoplasm,
YTHDF2 can bind to the m6A-containing RNAs and
direct them to processing bodies (Pbodies) for
accelerated mRNA decay. Pbodies can dynamically
form stress granules, in which RNAs could be stored
and released back to the translating pool. Besides
YTHDF2, other m6A reader proteins may bind to
m6A-containing RNAs to control their transport
and storage, thereby affecting translation. FTO,
-ketoglutarate-dependent dioxygenase FTO; WTAP,
Wilms tumour 1-associating protein.

when the response at the DNA level (that is, transcription) could be too slow; and when the response at
the protein level may require specific interactions or
modifications to tens to thousands of proteins, which
is difficult to achieve. Reversible methylation or other
forms of modifications on mRNA provides the best
option. A specific sequence that can undergo reversible modification, and thus be subjected to regulation,
can be readily included in a group of mRNA transcripts
(for example, at their 3UTRs) and lncRNAs in order
to affect RNA stability, localization and translatability,
as shown in the example ofYTHDF2.

Perspectives
In summary, reversible RNA methylation shares many
of the same characteristics as epigenetic DNA and histone modifications. Expression levels and potentially
post-translational modifications of writers, erasers
and readers can constantly sculpt the RNA methylome (FIG.5), which might in turn affect the eventual
protein expression. Therefore, the reversible chemical
tagging that dynamically controls the outcome of gene
expression occurs in all three main components of the
central dogma. Whereas epigenetic DNA and histone
modifications affect mostly transcriptional events,
reversible RNA methylation mainly has an impact on
regulation of post-transcriptional gene expression and
could directly affect protein production. Indeed, recent
research indicates that cellular protein levels are not
necessarily correlated with the mRNA levels105,106, which
emphasizes the importance of post-transcriptional
regulation of gene expression. Owing to the shared
use of reversible chemical tagging for dynamic gene
expression control, reversible RNA methylation has
been compared with epigenetic DNA and histone
modifications10,107. To also be a true epigenetic mark,
m6A would need to be heritable through cell division;
although this has not yet been demonstrated, such

302 | MAY 2014 | VOLUME 15

www.nature.com/reviews/genetics
2014 Macmillan Publishers Limited. All rights reserved

REVIEWS
METTL14

METTL3
WTAP

Histone modication

Other m6A-binding proteins

DNA modication

RNA polymerase II

An

m7Gppp
Nuclear retention
Nuclear
speckle

ALKBH5
FTO foci

Translation

FTO

m7Gppp

6
Me m A
An

m7Gppp

An

Me
m7Gppp

An

m7Gppp

An
Storage or
transport

Me
m7Gppp

6
Me m A
An

Export

YTHDF2
Translation
Me
m7Gppp

An

m7Gppp

An

m7Gppp

Me
An

m7Gppp

An

m7Gppp

An

Ribosome
Me
m

7Gppp

Me

Translating pool

An
ppp

Me

An

An

Me

Me

Me

m 7G

Me

Translatable pool

Me
m 7Gpp

Me

Me
An

Stress granule

ppp

Me

mG

m 7Gppp

Me

Me
An

mRNA decay

P-body

Nature Reviews | Genetics

NATURE REVIEWS | GENETICS

VOLUME 15 | MAY 2014 | 303

2014 Macmillan Publishers Limited. All rights reserved

REVIEWS
a possibility is conceivable through direct passage
of writer, reader and eraser proteins or methylated
RNA between generations of cells, which is a research
direction that needs to be further explored.
Many challenges lie ahead. It will be important to
clearly define the spatiotemporal properties of m6A
in terms of tissue specificity and in response to external and internal cues. The top priority is to elucidate
the involvement of m6A in RNA degradation, transport, storage, translation and splicing. The first steps
to achieve this goal will involve identifying and elucidating the functions of m6A reader proteins. Some
additional questions are: what is the interplay between
methyltransferases and demethylases that orchestrate
the methylation status of individual sites? How is
methylation and demethylation selectivity achieved?
Is the methylation coupled with transcription, and do
the two processes have mutual interactions? From yeast
to humans, how do the functions of m6A relate to cell
phenotypes and cell behaviour? Could some of the
processes be targeted to regulate biological functions
or to treat human diseases? Further research will answer
some of these questions and reveal fundamental aspects
of m6A biology.
Suzuki,M.M. & Bird,A. DNA methylation landscapes:
provocative insights from epigenomics. Nature Rev.
Genet. 9, 465476 (2008).
2. Kohli,R.M. & Zhang,Y. TET enzymes, TDG and the
dynamics of DNA demethylation. Nature 502,
472479 (2013).
3. Jones,P.A. Functions of DNA methylation: islands,
start sites, gene bodies and beyond. Nature Rev.
Genet. 13, 484492 (2012).
4. Branco,M.R., Ficz,G. & Reik,W. Uncovering the role
of 5hydroxymethylcytosine in the epigenome. Nature
Rev. Genet. 13, 713 (2012).
5. Bhutani,N., Burns,D.M. & Blau,H.M. DNA
demethylation dynamics. Cell 146, 866872 (2011).
6. Strahl,B.D. & Allis,C.D. The language of covalent
histone modifications. Nature 403, 4145 (2000).
7. Shi,Y. Histone lysine demethylases: emerging roles in
development, physiology and disease. Nature Rev.
Genet. 8, 829833 (2007).
8. Klose,R.J., Kallin,E.M. & Zhang,Y. JmjC-domaincontaining proteins and histone demethylation.
Nature Rev. Genet. 7, 715727 (2006).
9. Bird,A. Molecular biology. Methylation talk between
histones and DNA. Science. 294, 21132115 (2001).
10. He,C. Grand challenge commentary: RNA epigenetics?
Nature Chem. Biol. 6, 863865 (2010).
11. Grosjean,H. & Benne,R. Modification and Editing
of RNA (American Society for Microbiology Press,
1998).
12. Grosjean,H.Fine-Tuning of RNA Functions by
Modication and Editing (Springer-Verlag, 2005).
13. Machnicka,M.A. etal. MODOMICS: a database
of RNA modification pathways 2013 update.
Nucleic Acids Res. 41, D262D267 (2013).
14. Motorin,Y. & Helm,M. RNA nucleotide methylation.
Wiley Interdiscip. Rev. RNA 2, 611631 (2011).
15. Jia,G. etal. N6methyladenosine in nuclear RNA is a
major substrate of the obesity-associated FTO.
Nature Chem. Biol. 7, 885887 (2011).
This work describes a major breakthrough of
discovering the first m6A RNA demethylase FTO,
which highlights the possible biological function
of m6A.
16. Zheng,G. etal. ALKBH5 is a mammalian RNA
demethylase that impacts RNA metabolism and
mouse fertility. Mol. Cell 49, 1829 (2013).
This study discovered the second mammalian m6A
demethylase ALKBH5 that affects mouse
spermatogenesis.
17. Dominissini,D. etal. Topology of the human and
mouse m6A RNA methylomes revealed by m6A-seq.
Nature 485, 201206 (2012).
1.

Other intriguing chemical modifications exist on

mRNA and other types of nuclear RNA, such as m5C,
and 2OMe. Some of these modifications are only
a few fold less abundant than m6A on mRNA. They
could also be dynamic and may have important roles
in gene expression regulation, as recently suggested for
108110. Although transcriptome-wide m5C distribution
has been mapped111113, the other two modifications
have yet to be studied using modern sequencing
approaches. Both and 2OMe may have connections
to human diseases, which suggests functional roles114116.
Modifications on tRNA and rRNA can also be dynamic
and could affect the outcome of protein expression. Of
the nine human homologues of RNA demethylases,
ALKBH2 and ALKBH3 are DNA repair enzymes that
use the same oxidative demethylation mechanism to
remove DNA methyl adducts117,118, and ALKBH8 is a
tRNA hydroxylase119,120 that seems to affect tRNA codon
usage, whereas ALKBH1, ALKBH4, ALKBH6 and
ALKBH7 still do not have clearly defined functions.
Some of these homologues might work on nucleic acids
and act as demethylases for other forms of nucleic acid
methylations. We are still at the very beginning of this
new realm of fundamental research.

18. Meyer,K.D. etal. Comprehensive analysis of mRNA

methylation reveals enrichment in 3 UTRs and near
stop codons. Cell 149, 16351646 (2012).
References 17 and 18 revealed, for the first time,
the transcriptome-wide distributions of m6A in
mammalian genomes.
19. Wei,C.M., Gershowitz,A. & Moss,B. Methylated
nucleotides block 5 terminus of HeLa-cell messengerRNA. Cell 4, 379386 (1975).
20. Krug,R.M., Morgan,M.A. & Shatkin,A.J. Influenza
viral mRNA contains internal N6methyladenosine and
5terminal 7methylguanosine in cap structures.
J.Virol. 20, 4553 (1976).
21. Rottman,F.M., Desrosiers,R.C. & Friderici,K.
Nucleotide methylation patterns in eukaryotic mRNA.
Prog. Nucleic Acid. Res. Mol. Biol. 19, 2138 (1976).
22. Beemon,K. & Keith,J. Localization of
N6methyladenosine in the Rous sarcoma virus
genome. J.Mol. Biol. 113, 165179 (1977).
23. Schibler,U., Kelley,D.E. & Perry,R.P. Comparison of
methylated sequences in messenger RNA and
heterogeneous nuclear RNA from mouse L cells.
J.Mol. Biol. 115, 695714 (1977).
24. Wei,C.M. & Moss,B. Nucleotide sequences at the
N6methyladenosine sites of HeLa cell messenger
ribonucleic acid. Biochemistry 16, 16721676 (1977).
25. Narayan,P. & Rottman,F.M. An invitro system for
accurate methylation of internal adenosine residues in
messenger RNA. Science 242, 11591162 (1988).
26. Csepany,T., Lin,A., Baldick,C.J.Jr & Beemon,K.
Sequence specificity of mRNA N6adenosine
methyltransferase. J.Biol. Chem. 265, 2011720122
(1990).
27. Narayan,P., Ludwiczak,R.L., Goodwin,E.C. &
Rottman,F.M. Context effects on N6adenosine
methylation sites in prolactin mRNA. Nucleic Acids
Res. 22, 419426 (1994).
28. Rottman,F., Shatkin,A.J. & Perry,R.P. Sequences
containing methylated nucleotides at 5 termini of
messenger-RNAs possible implications for
processing. Cell 3, 197199 (1974).
29. Bodi,Z., Button,J.D., Grierson,D. & Fray,R.G.
Yeast targets for mRNA methylation. Nucleic Acids
Res. 38, 53275335 (2010).
30. Keith,G. Mobilities of modified ribonucleotides on
two-dimensional cellulose thin-layer chromatography.
Biochimie 77, 142144 (1995).
31. Clancy,M.J., Shambaugh,M.E., Timpte,C.S. &
Bokar,J.A. Induction of sporulation in Saccharomyces
cerevisiae leads to the formation of N6methyladenosine
in mRNA: a potential mechanism for the activity of the
IME4 gene. Nucleic Acids Res. 30, 45094518 (2002).

304 | MAY 2014 | VOLUME 15

32. Schwartz,S. etal. High-resolution mapping reveals a

conserved, widespread, dynamic mRNA methylation
program in yeast meiosis. Cell 155, 14091421 (2013).
This study reveals the dynamics of transcriptomewide m6A changes during yeast meiosis.
33. Liu,N. etal. Probing N6methyladenosine RNA
modification status at single nucleotide resolution
in mRNA and long noncoding RNA. RNA 19,
18481856 (2013).
34. Carroll,S.M., Narayan,P. & Rottman,F.M.
N6methyladenosine residues in an intron-specific
region of prolactin pre-mRNA. Mol. Cell. Biol. 10,
44564465 (1990).
35. Kierzek,E. & Kierzek,R. The thermodynamic
stability of RNA duplexes and hairpins containing
N6alkyladenosines and 2methylthioN6alkyladenosines. Nucleic Acids Res. 31,
44724480 (2003).
36. Harcourt,E.M., Ehrenschwender,T., Batista,P.J.,
Chang,H.Y. & Kool,E.T. Identification of a selective
polymerase enables detection of N6-methyladenosine
in RNA. J.Am. Chem. Soc. 135, 1907919082
(2013).
37. Vilfan,I.D. etal. Analysis of RNA base modification
and structural rearrangement by single-molecule realtime detection of reverse transcription.
J.Nanobiotechnol. 11, 8 (2013).
38. Bokar,J.A., Shambaugh,M.E., Polayes,D.,
Matera,A.G. & Rottman,F.M. Purification and cDNA
cloning of the AdoMet-binding subunit of the human
mRNA (N6adenosine)-methyltransferase. RNA 3,
12331247 (1997).
This pivotal study identifies METTL3 as a
key SAM-binding subunit of the RNA
methyltransferase complex.
39. Bokar,J.A. in Fine-Tuning of RNA Functions by
Modification and Editing 141177 (Springer-Verlag,
2005).
40. Liu,J. etal. A METTL3METTL14 complex mediates
mammalian nuclear RNA N6-adenosine methylation.
Nature Chem. Biol. 10, 9395 (2014).
This paper uncovers the core components of
the m6A RNA methyltransferase complex and
reveals an overall negative correlation between
the levels of m6A mRNA methylation and gene
expression.
41. Bujnicki,J.M., Feder,M., Radlinska,M. &
Blumenthal,R.M. Structure prediction and
phylogenetic analysis of a functionally diverse family of
proteins homologous to the MTA70 subunit of the
human mRNA:m6A methyltransferase. J.Mol. Evol.
55, 431444 (2002).

www.nature.com/reviews/genetics
2014 Macmillan Publishers Limited. All rights reserved

REVIEWS
42. Wang,Y. etal. N6-methyladenosine modification
destabilizes developmental regulators in embryonic
stem cells. Nature Cell Biol. 16, 191198 (2014).
This study discovered that the m6A modification
on mRNA affects embryonic cell differentiation.
43. Alexandrov,A., Martzen,M.R. & Phizicky,E.M.
Two proteins that form a complex are required for
7methylguanosine modification of yeast tRNA.
RNA 8, 12531266 (2002).
44. Chujo,T. & Suzuki,T. Trmt61B is a methyltransferase
responsible for 1methyladenosine at position 58 of
human mitochondrial tRNAs. RNA 18, 22692276
(2012).
45. Ozanick,S., Krecic,A., Andersland,J. &
Anderson,J.T. The bipartite structure of the
tRNA m1A58 methyltransferase from S.cerevisiae
is conserved in humans. RNA 11, 12811290
(2005).
46. Leulliot,N. etal. Structure of the yeast tRNA m7G
methylation complex. Structure 16, 5261 (2008).
47. Zhong,S. etal. MTA is an Arabidopsis messenger
RNA adenosine methylase and interacts with a
homolog of a sex-specific splicing factor. Plant Cell 20,
12781288 (2008).
48. Agarwala,S.D., Blitzblau,H.G., Hochwagen,A. &
Fink,G.R. RNA methylation by the MIS complex
regulates a cell fate decision in yeast. PLoS Genet. 8,
e1002732 (2012).
49. Little,N.A., Hastie,N.D. & Davies,R.C. Identification
of WTAP, a novel Wilms tumour 1associating protein.
Hum. Mol. Genet. 9, 22312239 (2000).
50. Ping,X.L. etal. Mammalian WTAP is a regulatory
subunit of the RNA N6methyladenosine
methyltransferase. Cell Res. 24, 177189 (2014).
51. Horiuchi,K. etal. Identification of Wilms
Tumor1associating protein complex and its role in
alternative splicing and the cell cycle. J.Biol. Chem.
288, 3329233302 (2013).
52. Bodi,Z. etal. Adenosine methylation in Arabidopsis
mRNA is associated with the 3 end and reduced
levels cause developmental defects. Front. Plant Sci.
3, 48 (2012).
53. Hongay,C.F. & Orr-Weaver,T.L. Drosophila Inducer of
MEiosis 4 (IME4) is required for Notch signaling
during oogenesis. Proc. Natl Acad. Sci. USA 108,
1485514860 (2011).
54. Peters,T., Ausmeier,K. & Ruther,U. Cloning of Fatso
(Fto), a novel gene deleted by the Fused toes (Ft)
mouse mutation. Mamm. Genome 10, 983986
(1999).
55. Dina,C. etal. Variation in FTO contributes to
childhood obesity and severe adult obesity. Nature
Genet. 39, 724726 (2007).
56. Frayling,T.M. etal. A common variant in the FTO
gene is associated with body mass index and
predisposes to childhood and adult obesity. Science
316, 889894 (2007).
57. Scuteri,A. etal. Genome-wide association scan
shows genetic variants in the FTO gene are
associated with obesity-related traits. PLoS Genet. 3,
e115 (2007).
58. Gerken,T. etal. The obesity-associated FTO gene
encodes a 2oxoglutarate-dependent nucleic acid
demethylase. Science 318, 14691472 (2007).
59. Fischer,J. etal. Inactivation of the Fto gene protects
from obesity. Nature 458, 894898 (2009).
60. Church,C. etal. Overexpression of Fto leads to
increased food intake and results in obesity. Nature
Genet. 42, 10861092 (2010).
61. Boissel,S. etal. Lossoffunction mutation in the
dioxygenase-encoding FTO gene causes severe growth
retardation and multiple malformations. Am. J.Hum.
Genet. 85, 106111 (2009).
62. He,Y.F. etal. Tet-mediated formation of
5carboxylcytosine and its excision by TDG in
mammalian DNA. Science 333, 13031307
(2011).
63. Ito,S. etal. Tet proteins can convert 5methylcytosine
to 5formylcytosine and 5carboxylcytosine. Science
333, 13001303 (2011).
64. Tahiliani,M. etal. Conversion of 5methylcytosine to
5hydroxymethylcytosine in mammalian DNA by MLL
partner TET1. Science 324, 930935 (2009).
65. Jia,G. etal. Oxidative demethylation of
3methylthymine and 3methyluracil in single-stranded
DNA and RNA by mouse and human FTO. FEBS Lett.
582, 33133319 (2008).
66. Hess,M.E. etal. The fat mass and obesity associated
gene (Fto) regulates activity of the dopaminergic
midbrain circuitry. Nature Neurosci. 16, 10421048
(2013).

67. Gulati,P. etal. Role for the obesity-related FTO gene

in the cellular sensing of amino acids. Proc. Natl Acad.
Sci. USA 110, 25572562 (2013).
68. Han,Z. etal. Crystal structure of the FTO protein
reveals basis for its substrate specificity. Nature 464,
12051209 (2010).
69. Zheng,G. etal. Sprouts of RNA epigenetics:
the discovery of mammalian RNA demethylases.
RNA Biol. 10, 915918 (2013).
70. Baltz,A.G. etal. The mRNA-bound proteome and its
global occupancy profile on protein-coding transcripts.
Mol. Cell 46, 674690 (2012).
71. Fu,Y. etal. FTO-mediated formation of
N6hydroxymethyladenosine and N6formyladenosine in
mammalian RNA. Nature Commun. 4, 1798 (2013).
72. Schwanhausser,B. etal. Global quantification of
mammalian gene expression control. Nature 473,
337342 (2011).
73. Rabani,M. etal. Metabolic labeling of RNA uncovers
principles of RNA production and degradation
dynamics in mammalian cells. Nature Biotech. 29,
436442 (2011).
74. Robbens,S. etal. The FTO gene, implicated in human
obesity, is found only in vertebrates and marine algae.
J.Mol. Evol. 66, 8084 (2008).
75. Iyer,L.M., Tahiliani,M., Rao,A. & Aravind,L.
Prediction of novel families of enzymes involved in
oxidative and other complex modifications of bases in
nucleic acids. Cell Cycle 8, 16981710 (2009).
76. Wang,X. etal. N6methyladenosine-dependent
regulation of messenger RNA stability. Nature 505,
117120 (2014).
This work presents the first m6A reader protein to
be characterized, YTHDF2, and a main function of
m6A: YTHDF2 mediates the m6A-dependent RNA
decay by targeting RNA substrates to Pbodies.
77. Schoenberg,D.R. & Maquat,L.E. Regulation of
cytoplasmic mRNA decay. Nature Rev. Genet. 13,
246259 (2012).
78. Isken,O. & Maquat,L.E. The multiple lives of NMD
factors: balancing roles in gene and genome
regulation. Nature Rev. Genet. 9, 699712 (2008).
79. Sheth,U. & Parker,R. Decapping and decay of
messenger RNA occur in cytoplasmic processing
bodies. Science 300, 805808 (2003).
80. Han,D. etal. IRE1 kinase activation modes control
alternate endoribonuclease outputs to determine
divergent cell fates. Cell 138, 562575 (2009).
81. Marzluff,W.F., Wagner,E.J. & Duronio,R.J.
Metabolism and regulation of canonical histone
mRNAs: life without a poly(A) tail. Nature Rev. Genet.
9, 843854 (2008).
82. Dasgupta,T. & Ladd,A.N. The importance of CELF
control: molecular and biological roles of the CUGBP,
Elav-like family of RNA-binding proteins. Wiley
Interdiscip. Rev. RNA 3, 104121 (2012).
83. Yang,F. & Schoenberg,D.R. Endonuclease-mediated
mRNA decay involves the selective targeting of PMR1
to polyribosome-bound substrate mRNA. Mol. Cell
14, 435445 (2004).
84. Ghosh,S. & Jacobson,A. RNA decay modulates gene
expression and controls its fidelity. Wiley Interdiscip.
Rev. RNA 1, 351361 (2010).
85. He,L. & Hannon,G.J. MicroRNAs: small RNAs with a
big role in gene regulation. Nature Rev. Genet. 5,
522531 (2004).
86. Ameres,S.L. & Zamore,P.D. Diversifying microRNA
sequence and function. Nature Rev. Mol. Cell Biol. 14,
475488 (2013).
87. Harigaya,Y. etal. Selective elimination of messenger
RNA prevents an incidence of untimely meiosis.
Nature 442, 4550 (2006).
88. Kariko,K., Buckstein,M., Ni,H. & Weissman,D.
Suppression of RNA recognition by Toll-like receptors:
the impact of nucleoside modification and the
evolutionary origin of RNA. Immunity 23, 165175
(2005).
89. Kawai,T. & Akira,S. Toll-like receptor and RIGIlike
receptor signaling. Ann. NY Acad. Sci. 1143, 120
(2008).
90. Newby,M.I. & Greenbaum,N.L. Sculpting of the
spliceosomal branch site recognition motif by
a conserved pseudouridine. Nature Struct. Biol. 9,
958965 (2002).
91. Lebedeva,S. etal. Transcriptome-wide analysis of
regulatory interactions of the RNA-binding protein
HuR. Mol. Cell 43, 340352 (2011).
92. Mukherjee,N. etal. Integrative regulatory mapping
indicates that the RNA-binding protein HuR couples
pre-mRNA processing and mRNA stability. Mol. Cell
43, 327339 (2011).

NATURE REVIEWS | GENETICS

93. Srikantan,S. & Gorospe,M. UneCLIPsing HuR nuclear

function. Mol. Cell 43, 319321 (2011).
94. Dormoy-Raclet,V. etal. HuR and miR1192 regulate
myogenesis by modulating the translation of HMGB1
mRNA. Nature Commun. 4, 2388 (2013).
95. Barnhart,M.D., Moon,S.L., Emch,A.W.,
Wilusz,C.J. & Wilusz,J. Changes in cellular mRNA
stability, splicing, and polyadenylation through HuR
protein sequestration by a cytoplasmic RNA virus.
Cell Rep. 5, 909917 (2013).
96. Abdelmohsen,K. & Gorospe,M. Posttranscriptional
regulation of cancer traits by HuR. Wiley Interdiscip.
Rev. RNA 1, 214229 (2010).
97. Ambros,V. The functions of animal microRNAs.
Nature 431, 350355 (2004).
98. Chen,K. & Rajewsky,N. The evolution of gene
regulation by transcription factors and microRNAs.
Nature Rev. Genet. 8, 93103 (2007).
99. Filipowicz,W., Bhattacharyya,S.N. & Sonenberg,N.
Mechanisms of post-transcriptional regulation by
microRNAs: are the answers in sight? Nature Rev.
Genet. 9, 102114 (2008).
100. Parker,R. & Sheth,U. P bodies and the control of
mRNA translation and degradation. Mol. Cell 25,
635646 (2007).
101. Keene,J.D. RNA regulons: coordination of
post-transcriptional events. Nature Rev. Genet. 8,
533543 (2007).
102. Gallego,M. & Virshup,D.M. Post-translational
modifications regulate the ticking of the circadian
clock. Nature Rev. Mol. Cell Biol. 8, 139148 (2007).
103. Eulalio,A., Behm-Ansmant,I. & Izaurralde,E.
P bodies: at the crossroads of post-transcriptional
pathways. Nature Rev. Mol. Cell Biol. 8, 922 (2007).
104. Fustin,J.M. etal. RNA-methylation-dependent RNA
processing controls the speed of the circadian clock.
Cell 155, 793806 (2013).
This study shows that the m6A modification affects
the export of several mRNAs that are related to
the circadian cycle.
105. Khan,Z. etal. Primate transcript and protein
expression levels evolve under compensatory selection
pressures. Science 342, 11001104 (2013).
106. Wu,L. etal. Variation and genetic control of protein
abundance in humans. Nature 499, 7982 (2013).
107. Saletore,Y. etal. The birth of the epitranscriptome:
deciphering the function of RNA modifications.
Genome Biol. 13, 175 (2012).
108. Karijolich,J. & Yu,Y.T. Converting nonsense codons
into sense codons by targeted pseudouridylation.
Nature 474, 395398 (2011).
109. Fernandez,I.S. etal. Unusual base pairing during the
decoding of a stop codon by the ribosome. Nature
500, 107110 (2013).
110. Ge,J. & Yu,Y.T. RNA pseudouridylation: new insights
into an old modification. Trends Biochem. Sci. 38,
210218 (2013).
111. Edelheit,S., Schwartz,S., Mumbach,M.R.,
Wurtzel,O. & Sorek,R. Transcriptome-wide mapping
of 5methylcytidine RNA modifications in bacteria,
archaea, and yeast reveals m5C within archaeal
mRNAs. PLoS Genet. 9, e1003602 (2013).
112. Hussain,S., Aleksic,J., Blanco,S., Dietmann,S. &
Frye,M. Characterizing 5methylcytosine in the
mammalian epitranscriptome. Genome Biol. 14, 215
(2013).
113. Squires,J.E. etal. Widespread occurrence of
5methylcytosine in human coding and non-coding
RNA. Nucleic Acids Res. 40, 50235033 (2012).
114. Bykhovskaya,Y., Casas,K., Mengesha,E., Inbal,A. &
Fischel-Ghodsian,N. Missense mutation in
pseudouridine synthase 1 (PUS1) causes
mitochondrial myopathy and sideroblastic anemia
(MLASA). Am. J.Hum. Genet. 74, 13031308
(2004).
115. Patton,J.R., Bykhovskaya,Y., Mengesha,E.,
Bertolotto,C. & Fischel-Ghodsian,N. Mitochondrial
myopathy and sideroblastic anemia (MLASA):
missense mutation in the pseudouridine synthase 1
(PUS1) gene is associated with the loss of
tRNA pseudouridylation. J.Biol. Chem. 280,
1982319828 (2005).
116. Sahoo,T. etal. PraderWilli phenotype caused by
paternal deficiency for the HBII85C/D box small
nucleolar RNA cluster. Nature Genet. 40, 719721
(2008).
117. Sedgwick,B. Repairing DNA-methylation damage.
Nature Rev. Mol. Cell Biol. 5, 148157 (2004).
118. Mishina,Y., Duguid,E.M. & He,C. Direct reversal of
DNA alkylation damage. Chem. Rev. 106, 215232
(2006).

VOLUME 15 | MAY 2014 | 305

2014 Macmillan Publishers Limited. All rights reserved

REVIEWS
119. Fu,Y. etal. The AlkB domain of mammalian ABH8
catalyzes hydroxylation of
5methoxycarbonylmethyluridine at the wobble
position of tRNA. Angew. Chem. Int. Ed Engl. 49,
88858888 (2010).
120. van den Born,E. etal. ALKBH8mediated formation
of a novel diastereomeric pair of wobble nucleosides
in mammalian tRNA. Nature Commun. 2, 172
(2011).
121. Aik, W. et al. Structure of human RNA
N6-methyladenine demethylase ALKBH5 provides
insights into its mechanisms of nucleic acid recognition
and demethylation. Nucleic Acids Res. http://dx.doi.
org/10.1093/nar/gku085 (2014).

122. Chen, W. et al. Crystal structure of the RNA

demethylase ALKBH5 from zebrafish. FEBS Lett. 588,
892898 (2014).

Acknowledgements

The authors apologize to colleagues whose work was not

cited owing to space limitation. They thank T. Pan, X. Wang,
Y. Yue and J. Liu for discussion. C.H. is supported by the US
National Institutes of Health grants GM071440 and the
EUREKA grant GM088599. This work was also supported
partly by grants from the Israel Science Foundation, the Flight
Attendant Medical Research Institute (FAMRI) and the Israeli
Centers of Research Excellence. S.F. Reichard contributed to
editing of this manuscript.

306 | MAY 2014 | VOLUME 15

Competing interests statement

The authors declare no competing interests.

FURTHER INFORMATION
Modomics a database of RNA modification pathways:
http://modomics.genesilico.pl/
Three-dimensional ribosomal modification maps database:
http://people.biochem.umass.edu/fournierlab/3dmodmap/
main.php
tRNAmod prediction of tRNA modifications: http://crdd.
osdd.net/raghava/trnamod/
ALL LINKS ARE ACTIVE IN THE ONLINE PDF

www.nature.com/reviews/genetics
2014 Macmillan Publishers Limited. All rights reserved