Macro effects of microRNAs in plants
Catherine A. Kidner and Robert A. Martienssen
Cold Spring Harbor Laboratory, 1 Bungtown Road, Cold Spring Harbor, NY 11724, USA
MicroRNAs (miRNAs) are 20- to 22-nucleotide frag-
ments that regulate expression of mRNAs that have
complementary sequences. They are numerous and
widespread among eukaryotes, being conserved
throughout evolution. The few miRNAs that have been
fully characterized were found in Caenorhabditis
elegans and are required for development. Recently, a
study of miRNAs isolated from Arabidopsis showed
that here also developmental genes are putative
regulatory targets. A role for miRNAs have in plant
development is supported by the developmental
phenotypes of mutations in the genes required for
miRNA processing.
MicroRNAs (miRNAs) are a recently described class of
small (20–22 nt) RNA molecules. They are produced from
larger precursor transcripts of 90–100 nt, which have a
hairpin structure [1]. They bind to mRNAs that have
complementary sequences, and thus they can regulate
gene expression by modulating transcript stability or
translation of the target mRNA. Recently, miRNAs were
cloned from both Arabidopsis and Schizosaccharomyces
pombe, indicating that miRNAs are present in all
eukaryotes, establishing them as an important, conserved
mechanism of eukaryotic gene regulation [2–4]. The next
few years will be spent determining which genes are being
regulated and how.
What are microRNAs (and why are they important)?
Small RNAs with target specificity conferred by antisense
sequences are already well characterized as regulators of
gene expression in bacterial systems and in phages.
Antisense RNA molecules block translation, promote
decay, cause premature termination of target transcripts,
and can even influence transcriptional regulation and
DNA replication. Whereas phage and transposons usually
generate sense and antisense transcripts from the same
locus, many bacterial antisense RNAs are similar to
eukaryote miRNAs in that they are generated in trans
and match their targets imperfectly [5].
Eukaryote miRNAs were first described in Caenorhab-
ditis elegans, where the lin-4 locus required for correct
developmental timing was shown to produce a 22-nt
miRNA. The lin-4 miRNA binds to the 30 untranslated
region (UTR) of the developmental genes lin-14 and lin-28
and downregulates their translation [6,7]. More recently, a
similar locus, let-7, has also been found to encode an
miRNA [8].
One of the intriguing aspects of these miRNAs was their
similarity to small interfering RNAs (siRNAs). These
small (21–24 nt) RNAs are a hallmark of post-transcrip-
tional gene silencing (PTGS) in plants [9], and similar
RNAs are associated with RNA interference (RNAi) in
animals [10]. RNAi is activated by double-stranded RNAs,
which are then processed into siRNA fragments by
DICER, an RNase III helicase [11]. The siRNAs can then
be amplified by an RNA-dependent RNA polymerase
(RdRP). These siRNAs then bind to complementary
mRNA targets, which activates the RNAi silencing
complex (RISC) to cleave the target mRNA [12]. Some of
the components of RISC have been identified, including
the ARGONAUTE-like proteins, and they seem to be very
evolutionarily conserved (Table 1). The presence of the
‘RNAi pathway’ genes in such a wide range of organisms
suggests they have a crucial endogenous role. The path-
way is certainly effective in defense against viruses and in
control of transposons, but the phenotypes arising from
mutation of some of the genes suggest that this pathway
also regulates developmental genes.
How many miRNAs are there?
Taking a genomics approach and using the features of
DICER products as a guide, six groups have sequenced
20–24-nt RNAs from C. elegans, Drosophila, human and,
most recently, from Arabidopsis and S. pombe [2,3,13–16].
Comparison with genomic sequence reveals that many of
these small RNAs are processed from the 30 end of larger
double-stranded RNA hairpin precursors, as is typical of
miRNAs. However this is not the case for the small RNAs
of S. pombe.
AlltheS.pombesiRNAscorrespondtothesequencesofthe
centromeric repeats [4], and mutations in the genes thought
to be involved in miRNA processing and function relieve
centromericsilencing[17], indicatingthatinthesingle-celled
S. pombe, siRNA has a specific limited role. In Arabidopsis,
however, miRNAs are complementary to a wide variety of
genes. Llave et al. [2], Reinhart et al. [3] and Park et al. [18]
recovered 51 different miRNA species with near perfect
complementarity to open reading frames (ORFs). There was
little overlap between the sets of miRNAs identified by the
three studies, suggesting that large numbers remain to be
discovered. Several of the miRNA precursors identified in
Arabidopsis were also found in the genomic sequence of rice,
indicating they have been evolutionarily conserved for more
than250millionyears.The largenumberofdifferentmiRNA
species in plants and their conservation throughout evol-
ution strongly indicates that miRNAsare important in many
plant gene regulation pathways.Corresponding author: Robert A. Martienssen (martiens@cshl.org).
0168-9525/02/$ - see front matter q 2002 Elsevier Science Ltd. All rights reserved. PII: S0168-9525(02)00011-2
Update TRENDS in Genetics Vol.19 No.1 January 2003 13
http://tigs.trends.com

What are the targets of miRNAs?
Potential targets for the majority of animal miRNAs have
not been identified despite the availability of complete
genomic sequence for C. elegans and Drosophila. In plants,
however, many of the miRNA have near perfect matches to
known genes, allowing a list of potential targets to be
drawn up [2,19].
Some miRNAs matched several members of the
same gene family, showing how multiple transcripts
could be regulated in concert. It is notable that most of
the target genes identified (34 out of 49) are transcrip-
tion factors. Targets include genes controlling all
aspects of plant development. For example, CUP-
SHAPED-COTYLEDONS2, which is involved in shoot
apical meristem formation; members of the SCARE-
CROW gene family, which are required for signaling
and asymmetric cell division; auxin response genes and
genes required for the regulation of floral development,
such as the LEAFY regulator myb33 and APETALA1
regulator SPBL3.
The most intriguing targets are members of the
homeodomain-ZIP gene family, including revoluta,
phabulosa and phavoluta. Intriguingly, dominant alleles
of phabulosa and phavoluta affect leaf polarity and have
lesions in the 19-nt portion of the coding sequence that
matches the miRNA, suggesting that base-pair comple-
mentarity is involved in the gain-of-function phenotype.
Intriguingly, dominant alleles of phabulosa result in
ectopic expression and overexpression of the mRNA [20].
Therefore, target turnover, as well as translation, could be
regulated by miRNA.
How do miRNAs work?
Moving from the identification of targets to finding out how
the miRNAs affect these mRNAs is difficult. Work in
human cell lines suggests that the degree of complemen-
tarity between the miRNA and its target controls whether
the RISC complex causes cleavage of the target mRNA
(exact match), or translational repression (imperfect
match) [21,22]. Unlike the known animal miRNAs, most
plant miRNAs match their targets with near perfect
complementarity within the ORF. Although the production
of miRNAs seems conserved throughout eukaryotes,
plants could have evolved a different series of outcomes
from the process. Alternatively, use of miRNAs to direct
cleavage of extraneous transcripts could simply be much
more common in plants. miRNA-directed cleavage does
occur, as a recent paper from Llave et al. shows that an
exact miRNA match is required for cleavage of the mRNA
of a SCARECROW-like gene [22].
Mutants in processing machinery
Arabidopsis is an ideal system for studying the miRNA
pathway because, as well as identifiable targets, there
are several well-characterized mutations in the path-
way. CARPEL FACTORY (CAF ) encodes a homolog of
dicer, and ARGONAUTE1 (AGO1 ) encodes a homolog
of C. elegans rde1, which is required for RNAi, and of
Drosophila ago2, which is found in the RISC complex
[23,24]. Arabidopsis caf mutants fail to accumulate
miRNAs [3], and ago1 mutants are defective in PTGS
[25]. The caf-1 mutants have small, dark serrated
leaves, short stems, semi-fertile flowers and multiple
carpels (Fig. 1). Strong ago1 alleles also have severe
developmental phenotypes, and we have found weak
alleles of ago1 that cause phenotypes resembling those
of caf-1 mutants (Fig. 1), indicating the two affect
similar targets (Fig. 2). Genetic interactions between
these and other mutants suggest a role for miRNAs in
plant development, particularly in stem cell function
and leaf polarity (C. Kidner et al., unpublished).
Mutations in HEN1 also prevent accumulation of
miRNAs. HEN1 encodes a novel protein and its role
in the miRNA pathway is unknown. However, hen1-1
mutants have very similar phenotypes to caf mutants,
suggesting a similar function to DICER or CAF [18].
Could the miRNAs be signalling molecules?
Although it remains to be seen whether the developmental
defects found in caf and ago1 mutants are mediated by the
putative miRNA targets, Foster et al. [26] have demon-
strated that overexpression of a viral movement protein in
tobacco resulted in radial leaves, a phenotype common to
strong alleles of argonaute and dominant PHB alleles in
Arabidopsis. Movement proteins facilitate plasmodesma-
tal transport of viral RNA. Furthermore, leaf polarity
requires communication between the meristem and the
incipient leaf primordia [27]. One possibility is that
miRNAs produced by the RNAi machinery in the meristem
are trafficked to leaves through plasmodesmata to specify
dorsal (adaxial) and/or ventral (abaxial) domains. Thus,
miRNAs could join peptides, hormones and other small
Table 1. Proteins involved in miRNA gene regulationa
Organism Production (DICER) Amplification (RdRP) Effector? (ARGONAUTE-like)
Gene Endogenous function Gene Endogenous function Gene Endogenous function
Arabidopsis CAF Meristem function SDE1 Leaf shape AGO1 Stem cell function and organ formation
floral development SGS2 ?
Neurospora ? ? Qde-1 ? Qde-2 ?
S. pombe dcr1 Centromeric silencing rdrp1 Centromeric silencing ago1 Centromeric silencing
C. elegans DCR-1 Developmental timing EGO1 Gametogenesis ALG1 Developmental timing and fertility
oogenesis RRF-1 ? ALG2
Drosophila Dicer ? ? AGO1 Viability, nervous system
PIWI Fertility, pleiotropic
aThose genes for which there is direct evidence of their involvement in miRNA regulation are shown in bold. The others are required for RNAi or PTGS, and mutants in most
cause developmental phenotypes, suggesting an endogenous role in control of gene expression [1]. The Schizosaccharomyces pombe genes are required for centromeric
silencing by an RNAi-like mechanism [17].
Update TRENDS in Genetics Vol.19 No.1 January 200314
http://tigs.trends.com