Age-associated changes in expression of small, noncoding RNAs, including microRNAs, in C. elegans

Abstract

Small, noncoding RNAs (sncRNAs), including microRNAs (miRNAs), impact diverse biological events through the control of gene expression and genome stability. However, the role of these sncRNAs in aging remains largely unknown. To understand the contribution of sncRNAs to the aging process, we performed small RNA profiling by deep-sequencing over the course of Caenorhabditis elegans (C. elegans) aging. Many small RNAs, including a significant number of miRNAs, change their expression during aging in C. elegans. Further studies of miRNA expression changes under conditions that modify lifespan demonstrate the tight control of their expression during aging. Adult-specific loss of argonaute-like gene-1 (alg-1) activity, which is necessary for miRNA maturation and function, resulted in an abnormal lifespan, suggesting that miRNAs are, indeed, required in adulthood for normal aging. miRNA target prediction algorithms combined with transcriptome data and pathway enrichment analysis revealed likely targets of these age-associated miRNAs with known roles in aging, such as mitochondrial metabolism. Furthermore, a computational analysis of our deep-sequencing data identified additional age-associated sncRNAs, including miRNA star strands, novel miRNA candidates, and endo-siRNA sequences. We also show an increase of specific transfer RNA (tRNA) fragments during aging, which are known to be induced in response to stress in several organisms. This study suggests that sncRNAs including miRNAs contribute to lifespan regulation in C. elegans, and indicates new connections between aging, stress responses, and the small RNA world.

Full text

Age-associated changes in expression of small, noncoding
RNAs, including microRNAs, in C. elegans
MASAOMI KATO,
1
XIAOWEI CHEN,
1,2
SACHI INUKAI,
1
HONGYU ZHAO,
2,3,4
and FRANK J. SLACK
1,5
1
Department of Molecular, Cellular and Developmental Biology, Yale University, New Haven, Connecticut 06520, USA
2
Program in Computational Biology and Bioinformatics, Yale University School of Medicine, New Haven, Connecticut 06520, USA
3
Department of Epidemiology and Public Health, Yale University School of Medicine, New Haven, Connecticut 06520, USA
4
Department of Genetics, Yale University School of Medicine, New Haven, Connecticut 06520, USA
ABSTRACT
Small, noncoding RNAs (sncRNAs), including microRNAs (miRNAs), impact diverse biological events through the control of
gene expression and genome stability. However, the role of these sncRNAs in aging remains largely unknown. To understand the
contribution of sncRNAs to the aging process, we performed small RNA profiling by deep-sequencing over the course of
Caenorhabditis elegans (C. elegans) aging. Many small RNAs, including a significant number of miRNAs, change their expression
during aging in C. elegans. Further studies of miRNA expression changes under conditions that modify lifespan demonstrate the
tight control of their expression during aging. Adult-specific loss of argonaute-like gene-1 (alg-1) activity, which is necessary for
miRNA maturation and function, resulted in an abnormal lifespan, suggesting that miRNAs are, indeed, required in adulthood
for normal aging. miRNA target prediction algorithms combined with transcriptome data and pathway enrichment analysis
revealed likely targets of these age-associated miRNAs with known roles in aging, such as mitochondrial metabolism.
Furthermore, a computational analysis of our deep-sequencing data identified additional age-associated sncRNAs, including
miRNA star strands, novel miRNA candidates, and endo-siRNA sequences. We also show an increase of specific transfer RNA
(tRNA) fragments during aging, which are known to be induced in response to stress in several organisms. This study suggests
that sncRNAs including miRNAs contribute to lifespan regulation in C. elegans, and indicates new connections between aging,
stress responses, and the small RNA world.
Keywords: aging; miRNA; noncoding RNA; C. elegans; modEncode
INTRODUCTION
Aging is defined as the time-dependent degenerative changes
that occur in tissues, cells, and in molecules such as DNA.
For example, loss of muscle mass and strength usually be-
gins at the middle of adulthood and proceeds further during
the later stages of life in diverse animal species, including the
nematode Caenorhabditis elegans (C. elegans)(Herndonetal.
2002; Johnston et al. 2008). These physiological alterations
are correlated with a decline in motility and closely re-
semble sarcopenia in humans (Herndon et al. 2002). Also,
progressive accumulation of a fluorescent pigment, lipo-
fuscin, is observed in aging tissues as a result of the oxida-
tive degradation and autophagocytosis of cellular compo-
nents (Garigan et al. 2002). Its ubiquitous occurrence in a
wide variety of organisms makes it a universal biomarker
for aging. Additionally, it is known that aspects of nuclear
architecture, such as nuclear shape and peripheral het-
erochromatin, are disorganized during aging in C. elegans
and that such features are often observed in premature ag-
ing syndromes, including Hutchinson-Gilford Progeria Syn-
drome (HGPS) in humans as well (Mattout et al. 2006).
It had been thought that age-related functional declines
were passive and not subject to regulation. Yet, it has been
demonstrated in C. elegans that mutations in single genes
can cause a delay in the aging process and extend lifespan.
age-1 and daf-2, both of which encode components in an
insulin signaling cascade, cause more than a twofold in-
crease in lifespan when either activity is reduced (Klass 1983;
Friedman and Johnson 1988; Kenyon et al. 1993; Morris
et al. 1996). Following these pioneering genetic studies in
C. elegans, it has been shown that these longevity genes are
highly conserved in yeast, flies, and mice and that they also
regulate lifespan through the control of, for example, me-
tabolism and the stress response, across phyla (Antebi 2007;
5
Corresponding author.
E-mail frank.slack@yale.edu.
Article published online ahead of print. Article and publication date are
at http://www.rnajournal.org/cgi/doi/10.1261/rna.2714411.
1804 RNA (2011), 17:1804–1820. Published by Cold Spring Harbor Laboratory Press. Copyright Ó2011 RNA Society.

Kenyon 2010). These findings suggest that lifespan is, at
least in part, a genetically determined biological event and
that changes in gene expression may underlie at least some
of the aging processes, including metabolic changes.
Gene expression profiling during aging can provide in-
sight into the molecular basis of the aging process. C. elegans
provides a genetically tractable model for this, with its rel-
atively short lifespan and plentiful genetic resources. Studies
of gene expression changes during aging in C. elegans have
provided clues to understanding mechanisms of age-related
phenomena. For example, the study of Lund et al. (2002)
demonstrated an increase in expression of transposable ele-
ment-derived transcripts during aging, pointing out the po-
tential of increased genome instability or uncontrolled,
unfavorable transcriptional activation with age. Another
study showed that many genes encoding heat shock pro-
teins, which are stress-resistance genes activated by stress,
rise in their expression at the beginning of adulthood and
then decrease toward later stages of aging (Golden and
Melov 2004; McCarroll et al. 2004). Since heat shock
proteins serve as molecular chaperones in protein homeo-
stasis (Morimoto 2008), their decrease in aged animals may
cause an increased level of unfolded proteins, possibly
leading to increased toxicity and stress, finally causing im-
paired cellular function and organismal death. These studies
support the idea that alteration of gene activity might serve
as a biological clock influencing lifespan.
Notably, small noncoding RNAs (sncRNAs), especially
microRNAs (miRNAs), have a great impact on the control
of gene expression through their ability to regulate hun-
dreds of target genes (Stefani and Slack 2008). miRNAs neg-
atively regulate gene expression at the post-transcriptional
level via perfect or partial sequence complementarity to
mRNAs of their target genes. For example, a founding miRNA,
lin-4, accumulates in the early larval stage of C. elegans
development and suppresses the expression of its target gene
lin-14 to initiate a transition in developmental stage (Lee
et al. 1993). Importantly, in a previous study, we found that
the lin-4 miRNA and its target lin-14 are also necessary for
normal lifespan in C. elegans;lin-4 loss-of-function mu-
tants and gain-of-function mutants of its target lin-14 (due
to loss of a lin-4 binding site in its 39untranslated region)
showed a short lifespan, while overexpression of lin-4 miRNA
and loss-of-function mutants of lin-14 resulted in a longer
lifespan (Boehm and Slack 2005), suggesting that miRNA
and miRNA-mediated negative regulation of gene expres-
sion influence longevity. Subsequent miRNA microarray
analysis showed that many mature miRNAs change their
expression during aging in C. elegans (Ibanez-Ventoso et al.
2006). In addition to miRNAs, additional new classes of
sncRNAs identified in recent studies (Ruby et al. 2006; Batista
et al. 2008; Claycomb et al. 2009; Gu et al. 2009; Stoeckius
et al. 2009), such as 21U-RNAs/piRNAs and 22G-RNAs
(endo-siRNAs), might also have important roles in aging
through the maintenance of genome stability and genome
surveillance, but the dynamics of these sncRNA species
through aging have not been examined.
Here, we show, using deep-sequencing technology, that
a significant number of sncRNAs, including miRNAs, change
their expression during aging in C. elegans. In addition, adult-
specific knockdown of argonaute-like gene-1 (alg-1), which is
required for miRNA processing and function, resulted in
a shorter lifespan, supporting our hypothesis that miRNAs
have important roles in normal lifespan regulation. Also,
predicted targets of age-associated miRNAs have known
roles in aging processes, including mitochondrial respira-
tion and protein homeostasis. We also report the identifi-
cation of novel miRNA candidates, miRNA star molecules,
and other age-associated sncRNAs, including transfer RNA
(tRNA) cleavage products, which are known to be induced
in response to stress in several organisms (Thompson and
Parker 2009a). This study uncovers potential roles for miRNAs
and other sncRNAs in aging and age-related events, and we
propose that these RNAs contribute to lifespan regulation
in C. elegans.
RESULTS
In order to reveal the contributions of sncRNAs, especially
miRNAs, to aging in C. elegans, we first examined their ex-
pression changes during aging, using deep-sequencing with
Solexa technology. In this study, we used spe-9(hc88), a tem-
perature-sensitive sterile mutant, to prevent contamination of
RNAs from aging animals with those of developing embryos.
Although spe-9(hc88) mutant animals cannot produce prog-
eny at the restrictive temperature due to a defect in spermato-
genesis (Singson et al. 1998), they still have a lifespan similar to
wild-type N2 animals (Fabian and Johnson 1994). We first
confirmed their normal lifespan at the nonpermissive tem-
perature, 23°C (Fig. 1A), and prepared cDNA libraries for
small RNAs at four different time points during adulthood:
Day 0, Day 5, Day 8, and Day 12 (Day X represented in this
study is defined as days post-L4 [the fourth larval stage]
molt). As shown in Figure 1A, these times represent key
inflection points in the survival curve. The approximate
survival rate of spe-9(hc88) mutants was 100%, >90%, 50%,
and <10% at Day 0, Day 5, Day 8, and Day 12, respectively.
We also produced additional cDNA libraries from RNA
samples that were collected from independently grown
Day 0 and Day 8 spe-9(hc88) mutants in order to test the
reproducibility of the initial experiment.
More than 5 million sequence reads that match to the C.
elegans genome were obtained from each library (Supple-
mental Table S1). Of those, z70% were known, annotated
mature miRNA sequences, z1.5% of reads perfectly
matched 21U-RNAs/piRNAs, z15% mapped to a part of
other known noncoding RNA fragments such as tRNAs,
and the remaining 15% did not map to any annotated regions
and were categorized into a fraction of ‘‘other reads’’ (Fig.
1B; Supplemental Table S1). This fraction includes miRNA-
Small, noncoding RNAs in C. elegans aging
www.rnajournal.org 1805

associated reads such as miRNA ‘‘star’’ strands, and novel
miRNA candidates and possible novel sncRNAs (described
below). We also found that the total levels of both miRNAs
and 21U-RNAs/piRNAs exhibited a gradual decrease during
aging, while other small RNA fragments derived from tRNAs,
ribosomal RNAs (rRNAs), small nucleolar RNAs (snoRNAs),
and small nuclear RNAs (snRNAs) showed a gradual but con-
stant increase in accumulation during aging (Fig. 1B; Sup-
plemental Table S1) (details described below).
A significant number of mature miRNAs change their
expression during aging
As mentioned, most of the aligned deep-sequencing reads
mapped to known, annotated mature miRNAs. We could
detect reads from 149 of 174 annotated mature miRNAs in
our libraries (based on miRBase release 14) (Kozomara and
Griffiths-Jones 2011; Supplemental Tables S2, S3). Of the
known miRNAs we could not detect, some may not ac-
tually encode miRNAs, or their mature strands may have
been mistakenly annotated. For example, we did not clone
sequences corresponding to the annotated miR-356 mature
sequence, but reads corresponding to the mir-356 precur-
sor sequence were found in the stem regions of its hairpin du-
plex, which are likely to be the true mature and star miRNAs
(Supplemental Table S3). Other possibilities are that expres-
sion of these undetectable miRNAs is induced only for a
short time in the life cycle, such as during the 1-cell-stage
embryo (Stoeckius et al. 2009) or might be induced only
under specific conditions or that they are hard to clone by
the adapter-ligation-mediated method used in library prep-
aration for Solexa because of terminal modifications on
RNA molecules (Pak and Fire 2007).
In a previous study, we examined miRNA expression
changes during larval development and in different sexes in
C. elegans and found that the expression of many mature
miRNAs is dynamically regulated during the developmental
stages from embryo to young adult (Kato et al. 2009a).
Therefore, we tested the overall changes in expression of
mature miRNAs over time during aging. Pairwise compar-
isons between samples from aged populations (e.g., Day 5
vs. Day 8, or Day 5 vs. Day 12) showed strong correlations
in their expression changes, while pairwise comparisons
between samples from Day 0 young adult and any other
aged populations exhibited large differences in their ex-
pression changes (Supplemental Fig. S1). These observations
suggest that alterations in expression of miRNAs mostly
occur in the early stages of adulthood rather than in middle
to later stages of lifespan and do not fluctuate greatly from
mid-adulthood.
For this reason, coupled with the availability of replicates
we made in the deep-sequencing experiments (see Supple-
mental Fig. S2A,B for details), we focused on two stages,
Day 0 young adults and Day 8 old adults in order to rank
changes in expression of mature miRNAs during aging. We
identified known miRNAs with statistically significant ex-
pression changes during aging (P#0.05) (Supplemental
Table S2). These include z50 mature miRNAs in total,
which are equivalent to z30% of the 174 known, annotated
mature miRNAs (miRBase release 14). Those with more
than twofold changes in the number of sequence reads
from Day 0 to Day 8 were listed in Figure 2A, which in-
cludes 23 up-regulated mature miRNAs and 16 down-
regulated ones, plus a mature miRNA variant for miR-71
(see the legend for Supplemental Table S2) and two miRNA
star species (see below). As shown in Figure 2A, the most
dramatically up-regulated miRNA was miR-239a, confirm-
ing our earlier report (de Lencastre et al. 2010). Another
one of the miRNAs with the most dramatic up-regulation
during aging was miR-34, which is highly conserved from
FIGURE 1. Lifespan of spe-9(hc88) mutant animals and summary of
deep-sequencing reads. (A) Animals were raised and cultured at 23°C
in the lifespan assay. Mean adult lifespan was 8.62 +/0.05 d. Detailed
results of the lifespan assay are shown in Supplemental Table S4. Error
bars represent standard error (SE) calculated from triplicates. RNAs
were purified at Day 0, Day 5, Day 8, and Day 12 post-L4 molt as
highlighted and used for library preparation for Solexa deep-sequencing
experiments. (B) Proportion of each noncoding RNA species, including
miRNAs, 21U-RNAs/piRNAs, and tRNAs, were analyzed in each aging
sample. Details are shown in Supplemental Table S1.
Kato et al.
1806 RNA, Vol. 17, No. 10

C. elegans to human and involved in the DNA damage
response across phyla (He et al. 2007; Kato et al. 2009b). An
example of the most down-regulated miRNA, lin-4, the
first miRNA found to be necessary for
normal lifespan in C. elegans (Boehm
and Slack 2005), indeed, showed a sig-
nificant reduction during aging. Addi-
tionally, another founding member of
miRNAs, let-7, and its family miRNAs
such as mir-241 and mir-795 (Roush and
Slack 2008), showed a dramatic decrease
in expression during aging (Fig. 2A).
We confirmed these results by bio-
logical replicates of the deep-sequencing
libraries for Day 0 and Day 8, using in-
dependently prepared RNA samples, and
also by quantitative RT-PCR (qRT-PCR)
(Fig. 2B; Supplemental Fig. S2), support-
ing the reliability of our data. We further
performed a simple hierarchical cluster-
ing analysis and found that miRNA
family members have a similar trend in
expression changes during aging (Sup-
plemental Fig. S3). All together, these
observations suggest that, in addition to
lin-4, additional miRNAs might have
important roles in lifespan regulation in
C. elegans.
Adult-specific loss of alg-1 activity
disrupts miRNA expression changes
and results in a shorter lifespan
To test the importance of miRNAs in
aging, we examined how the loss of ac-
tivity of ALG-1 affects normal lifespan.
The alg-1 gene encodes an Argonaute
protein which is necessary for miRNA
maturation and function but is not re-
quired for other known small RNA path-
ways (Grishok et al. 2001; Batista et al.
2008). In this experiment, spe-9(hc88)
animals were first cultured on standard
Escherichia coli OP50 bacteria and then
exposed to feeding RNAi against alg-1
and empty vector L4440 as a control when
they were Day 0 young adults (Supple-
mental Fig. S4). This enabled us to avoid
the possibility that the effect of alg-1 loss
on the lifespan was due to a defect in de-
velopment. We found that animals ex-
posed to alg-1 RNAi displayed a signif-
icantly shorter lifespan compared to the
control (Fig. 3A; Supplemental Table
S4), demonstrating that miRNAs are
necessary for normal lifespan in C. elegans adults. Essen-
tially the same results were obtained when wild-type N2
animals were cultured with alg-1(RNAi) versus control at
FIGURE 2. miRNAs with most increased and decreased expression during aging. (A) Of the
mature miRNAs with statistically significant expression changes during aging (P#0.05), those
with more than twofold changes in the number of reads from Day 0 to Day 8 are shown here.
This list also includes a mature miRNA variant for miR-71 and two miRNA star strands.
miRNAs were sorted by the fold-changes. P-values are minimum P-value of pairwise analyses
among Day 0, Day 5, Day 8, and Day 12 (see Materials and Methods for more details). Two of
the miRNAs with * in their names represent miRNA star species, and sequences with the most
abundant number of reads (Supplemental Table S3) were used for counting as we have done
for the mature miRNAs. Note that the expression of annotated mature miR-71 (19 nt)
appeared to be increased during aging as observed in the qRT-PCR experiments (Supplemental
Fig. S2C). However, its longer form (miR-71_L-form) (23 nt) was more abundant than the
annotated one, and it showed decreased expression during aging (Supplemental Table S3;
details are discussed in the legend for Supplemental Table S2). (B) The results of miRNA
expression changes were confirmed by qRT-PCR. The results were normalized by the average
of the expression of act-3 and ama-1. Error bars for qRT-PCR results shown by dark gray bars
indicate standard deviation (SD). Additional results are shown in Supplemental Figure S2C.
Error bars for deep-sequencing results shown by light gray bars represent the maximum and
minimum values in fold-changes in two replicates, which were calculated after each read was
normalized by the total number of aligned reads in each library.
Small, noncoding RNAs in C. elegans aging
www.rnajournal.org 1807

either 20°Cor23°C (data not shown). This is further
supported by the observations that miRNAs with increasing
expression during aging were disrupted in their expression
by loss of alg-1 activity, while those with decreasing ex-
pression, such as let-7, were not affected (Fig. 3B). Con-
sistent with this conclusion, we recently found that some
of the age-associated miRNAs identified here, indeed, play
important roles in lifespan regulation in C. elegans (de
Lencastre et al. 2010).
Conditions that modify normal lifespan can modulate
age-associated miRNA expression changes
To further understand the contribution of these age-associated
miRNAs to aging, we examined their expression changes
under conditions that modified lifespan. Specifically, we tested
whether conditions leading to a long lifespan could cause
a delay in miRNA expression changes and if conditions
leading to a short lifespan could accelerate their expres-
sion during aging. In C. elegans, an increase in temperature
accelerates the aging processes and reduces lifespan, whereas
a decrease in temperature causes a delay in the aging pro-
cesses and prolongs lifespan (Klass 1977).
In this study, synchronized spe-9(hc88)
animals were first cultured at the non-
permissive temperature (23°C) during
larval development to induce their ste-
rility and then shifted to 15°C and 27°C
when they were Day 0 young adults.
Compared to control animals which
were kept at 23°C during all life stages
including adulthood, animals shifted to
the lower temperature (15°C), indeed,
showed a significantly longer lifespan,
while those exposed to the higher tem-
perature (27°C) resulted in a shorter life-
span (Fig. 4A). Where possible, we puri-
fied total RNAs from Day 0, Day 3, Day
5, Day 8, Day 12, and Day 15 animals
at each temperature condition and ex-
amined changes in expression of age-
associated miRNAs using qRT-PCR.
We found that miRNAs with increas-
ing expression during aging showed a
significant delay in expression change at
the long-lived 15°C condition, while their
expression changes were accelerated at
the short-lived 27°C condition (Fig. 4B,
left; Supplemental Fig. S5). For exam-
ple, miR-34 and miR-239a, both of
which exhibited about a 10- to 15-fold
increase in expression from Day 0 to Day
8 at the standard condition 23°C, reached
a similar fold increase at Day 15 in the
long-lived condition, while they reached
a similar fold increase earlier at Day 3 in the short-lived
condition (Fig. 4B, left). These observations demonstrate
the tight regulation of miRNA expression during aging and
support the idea that maintaining these miRNAs at lower
levels during aging might contribute to a delay in the aging
process and cause longer lifespan or health span. Also, these
results suggest their potential as molecular biomarkers for
aging, although we cannot rule out that some of them may
have been affected by higher temperature rather than short
lifespan, including the case of lin-4 in Figure 4B, right.
In contrast to miRNAs with an increase in expression
during aging, those with a decrease in expression during
aging did not show such a delay; they exhibited the same
rapid decrease in expression in the long-lived condition at
15°C as observed in the standard control condition (Fig.
4B, right). All miRNAs examined, including let-7 and lin-4,
reduced their expression a short time after animals were
shifted to the lower temperature (Fig. 4B, right; Supple-
mental Fig. S5). One possible explanation is that their
down-regulation might be programmed at earlier stages
including during larval development. Alternatively, their ex-
pression might be controlled independently from the length
of lifespan.
FIGURE 3. Lifespan of spe-9(hc88) animals treated with adult-specific RNAi against alg-1.(A)
Animals were exposed to RNAi only during adulthood. Mean adult lifespans for the control
and alg-1 RNAi were 9.14 +/0.33 and 7.89 +/0.15 d, respectively (P#0.0005). Detailed
results of the lifespan assay are shown in Supplemental Table S4. Error bars represent standard
error (SE) calculated from triplicates. (B) Expression changes of age-associated miRNAs were
examined by qRT-PCR. Data were normalized by the average of the expression levels of act-3
and ama-1 genes. Error bars represent SD. Day 0 animals cultured on OP50 bacteria (just
before RNAi exposure) were used as a control.
Kato et al.
1808 RNA, Vol. 17, No. 10

Moreover, we confirmed the tight regulation of age-
associated miRNAs in modified lifespan backgrounds in-
duced by RNAi-mediated knockdown of aging-related genes.
These include daf-2,anddaf-16 and hsf-1, which cause long-
lived and short-lived conditions, respec-
tively, when their activity is diminished
(Fig. 4C; for review, see Antebi 2007;
Kenyon 2010). Similar to those ob-
served in the temperature-shift experi-
ments above, many age-associated miRNAs
with up-regulated expression during
aging, such as miR-239a, exhibited a delay
in expression changes, while those with
down-regulation during aging showed
expression changes independently of their
lifespan conditions (Fig. 4C), although
we found that some of the age-associ-
ated miRNAs were not consistent with
such trends in expression changes, possi-
bly due to their genetic interaction with
the aging-related genes examined (M Kato
and FJ Slack, unpubl.).
Age-associated miRNA expression
is mainly attributed to transcriptional
activity from miRNA promoters
We validated the altered expression of
age-associated miRNAs using transgenic
animals carrying constructs of the miRNA
promoter fused to a GFP marker gene.
Transgenic lines used in this study con-
tained transgenes integrated into chro-
mosomes and were crossed into the spe-
9(hc88) mutant background. We tested
changes in GFP signal intensity during
aging at Day 0, Day 3, and Day 6 of
adulthood. Most of the lines we examined
exhibited similar trends in expression
change shown for the mature miRNAs
from the results of deep-sequencing
and qRT-PCR. For example, mir-34,
which showed a dramatic increase in ex-
pression of the mature miRNA during
aging, exhibited a similar increase in GFP
signals in aged transgenic animals (Fig.
5). For let-7, which showed a marked
reduction of mature miRNA levels dur-
ing aging, we found that GFP signals were
nearly absent in Day 6 animals (Fig. 5).
This suggests that at least a part of these
expression changes can be attributed to
transcriptional activity driven by the
miRNA promoter.
Interestingly, in many cases of miRNAs
up-regulated with age, the GFP signals seem to be expressed
in tissues or cells in which they are not expressed at earlier
stages. The GFP signals from the mir-34Tgfp transgenic line
were detectable de novo in cells on the ventral side in most
FIGURE 4. Expression changes of age-associated miRNAs in conditions that modify lifespan.
(A) Lifespan of spe-9(hc88) animals was assayed at different temperatures, 15°C, 23°C, and
27°C. Mean adult lifespans were 7.62 +/0.15 d, 16.21 +/0.37 d, and 6.11 +/0.17 d at
23°C, 15°C, and 27°C, respectively (P< 0.0001). Error bars indicate SE. Details are shown in
Supplemental Table S4. Total RNAs were purified from Day 0, Day 3, Day 5, Day 8, Day 12,
and Day 15 post-L4 molt as highlighted. (B) Expression changes of age-associated miRNAs
were examined by qRT-PCR. Data were normalized by the average of the expression levels of
act-3 and ama-1 genes. Error bars represent SD. Day 0 animals cultured at 23°C (just before
the temperature shift) were used as a control. Additional results are shown in Supplemental
Figure S5. (C) Similar to the temperature-shift experiment, a delay or acceleration in
expression changes of the age-associated miRNAs was observed in genetically modified
lifespan backgrounds. In this study, spe-9(hc88) animals were exposed to each feeding RNAi,
including the control empty vector (L4440), from the L1 stage, in order to induce a sufficient
RNAi effect.
Small, noncoding RNAs in C. elegans aging
www.rnajournal.org 1809

aging animals (Fig. 5; marked by arrowheads in larger
images in Supplemental Fig. S6B). In another example, the
mir-35-41 miRNA cluster, which is highly expressed in
developing embryos (Lim et al. 2003), exhibited extensive
GFP signals in mid- to late- aging animals, in spite of their
lack of fertilized embryos in the spe-9(hc88) background.
These changes in expression pattern could be due to the
following possibilities: a tightly regulated aging activity;
transgene-specific background expression of GFP signals;
the effect of disorganized tissues with age (e.g., diffusion
of GFP proteins); or uncontrolled transcription because of
an age-dependent loss of proper regulation of gene ex-
pression (Lund et al. 2002). If the GFP signals specifically
observed in aged animals are due to age-related uncon-
trolled transcriptional activation, this erratic activation of
miRNA expression might trigger unfavorable age-related
decline, since miRNAs affect activities of many target genes.
We also examined age-related changes in the promo-
terTGFP signal intensity in the lower temperature-induced
longer lifespan background, as we did for mature miRNAs.
As expected, the GFP signals expressed from miRNA pro-
moterTgfp lines for up-regulated miRNAs in aging (e.g.,
mir-34) exhibited a remarkable delay in their changes in the
long-lived condition at 15°C, compared to those observed
in the control cultured at the standard condition 23°C (Fig.
5; images are shown in Supplemental Fig. S6C). In ad-
dition, the transgenic lines for down-regulated miRNAs in
aging (e.g., let-7) also showed a delay in changes of the GFP
signals in the long-lived condition, unlike the results ob-
served for the mature miRNAs (Figs. 4
and 5; images are shown in Supplemen-
tal Fig. S6C). However, the extent of
delay in changes of the GFP signals de-
tected for down-regulated miRNAs was
not as pronounced, compared to those
detected for up-regulated miRNAs, and
may reflect the stability of the GFP
protein.
Computational miRNA target
prediction combined with
transcriptome analysis reveals
likely candidate genes regulated
by age-associated miRNAs
We predicted candidate targets of age-
associated miRNAs and their possible
biological pathways. miRNAs negatively
regulate their target genes by binding to
complementary sequences in the 39UTR
of mRNAs (Vella et al. 2004). Although
miRNAs had been thought to preferen-
tially affect the level of proteins rather
than mRNAs, it has been shown that
miRNAs also reduce the level of mRNAs
(Bagga et al. 2005; Guo et al. 2010). Indeed, microarray
experiments have proven to be an effective way to find genes
modulated by miRNAs (Johnson et al. 2007). We assumed
that age-associated alterations of miRNA expression levels
might cause a reciprocal trend in the expression change of
the mRNAs of their target genes during aging. Thus, in
addition to the conventional target prediction algorithms
such as miRanda, we combined the expression profiles of
miRNAs and protein-coding genes in aging and age-related
phenotypic information together (details are described in
Materials and Methods) (Fig. 6A) to highlight candidate tar-
get genes with reciprocal expression changes to their cognate
miRNAs during aging. Three hundred and fifty-four
protein-coding genes were predicted as targets for one
or multiple age-associated miRNAs (from Fig. 2A) (with
a maximum of 24 different miRNAs found on one target)
(Supplemental Table S5). Of these 354, 58 genes are known
to be necessary for normal lifespan in C. elegans, suggesting
that these 58 genes are better candidates for regulation by
age-associated miRNAs (these are highlighted in red in Sup-
plemental Table S5). Next, to determine how these pre-
dicted targets contribute to aging, we performed pathway
enrichment analysis with KEGG (Kyoto Encyclopedia of
Genes and Genomes). KEGG is an integrated database
resource providing functional aspects of biological systems,
such as biological pathways in cells and organisms, based
on the molecular information of genes and their products
(Kanehisa et al. 2010). For C. elegans,z1700 genes are
classified into 121 pathways in the KEGG database. In our
FIGURE 5. Age-associated changes in GFP signals expressed from miRNA promoterTgfp
transgenes. The vertical axis represents a fold-change in GFP signal intensity, microscopically
determined from z15–20 individual whole animals at each time point for each line. Error bars
represent SE. The blue-colored and green-colored bars represent the GFP signal intensity
observed in the transgenic lines cultured at the standard temperature 23°C and the long-lived
condition at 15°C, respectively. Scale bars represent 100 mm. Additional results for miRNATgfp
lines were examined, and strain information is shown in Supplemental Figure S6B.
Kato et al.
1810 RNA, Vol. 17, No. 10

pathway analysis, predicted target genes of 33 age-associ-
ated miRNAs were found to be significantly enriched in 22
pathways (P< 0.05) (Fig. 6B; Supplemental Tables S5, S6).
Importantly, these include phagosome and lysosome path-
ways, which contain vacuolar family and proteasome genes,
suggesting that age-associated miRNAs including let-7 and
mir-1 may contribute to aging through the control of the
proteasome system. Also, we found that
genes in the tricarboxylic acid (TCA)
cycle and oxidative phosphorylation,
which form the core for energy produc-
tion in the mitochondria (Wallace 2005),
are predicted as targets of age-associated
miRNAs (Fig. 6B). Indeed, mitochon-
drial function and proteostasis are co-
ordinately modulated during aging by
insulin, target of rapamycin (TOR), and
sirtuin signaling pathways to protect ge-
nomes and cells from reactive oxygen
species and misfolded proteins (Haigis
and Yankner 2010).
miRNA ‘‘star’’ strands preferentially
accumulate in RDE-1, and they
change their expression
during aging
As we mentioned earlier, z70% of aligned
sequence reads mapped to known, anno-
tated mature miRNAs, and z15% map-
ped to unannotated genomic regions,
suggesting that there still might be
age-associated sncRNAs in this frac-
tion. This fraction could include
novel miRNA candidates and miRNA-
related sequences such as miRNA ‘‘star’’
molecules.
The miRNA star sequence is the op-
posite strand to the mature one in a
hairpin duplex derived from the miRNA
precursor, and it has been previously
thought that they are byproducts in
miRNA processing and degraded rap-
idly. However, several recent studies dem-
onstrate that they are incorporated into
Argonaute protein complexes in Dro-
sophila (Czech et al. 2009; Okamura
et al. 2009; Ghildiyal et al. 2010). In-
deed, our deep-sequencing libraries de-
tected a significant number of reads
corresponding to miRNA hairpin-loop
sequences, including many miRNA star
reads. Star reads and their derivatives were
detected in our library for 111 known
miRNAs (miRBase release 14), and their
levels generally decreased during aging, similar to those
of mature miRNAs (Supplemental Table S3). These obser-
vations suggest that star strands can, indeed, exist as stable
RNA molecules in C. elegans.
To investigate in which Argonaute proteins the star
miRNA strands are incorporated, we examined the enrich-
ment of sequence reads corresponding to mature and star
FIGURE 6. Target prediction of age-associated miRNAs and pathway analysis. (A)Target
prediction algorithms combined with gene expression profiles and age-related phenotypes fol-
lowed by the pathway analysis revealed strong candidate targets of age-associated miRNAs and
their possible roles in aging. The expression profiles for protein-coding genes were obtained from
Budovskaya et al. (2008), where they used a spe-9(hc88) background and similar conditions for
RNA preparation, including the time points during aging. The size of each circle reflects the
number of genes identified in each category (but the area of overlapped regions is not to scale
completely). (B) Pathway enrichment analysis with KEGG against predicted target genes identified
their possible roles in aging processes. The results were sorted by the order of enrichment (num-
ber of predicted genes in each pathway). For example, many TCA cycle and related genes were
predicted as targets. These serve an important function in mitochondria together with genes clas-
sified in oxidative phosphorylation (shown by a broken line). The names of gene pathways shown
in black represent genes necessary for the normal lifespan, suggesting that these pathways are
involved in aging. Detailed results are available in Supplemental Tables S5 and S6.
Small, noncoding RNAs in C. elegans aging
www.rnajournal.org 1811

miRNAs in ChIP-Seq (chromatin im-
munoprecipitation followed by deep-
sequencing) libraries made for Argo-
naute proteins, ALG-1, ALG-2, and
RDE-1 (Corre
ˆa et al. 2010) (sequence
data were obtained from the Genome
Expression Omnibus [GEO], accession:
GSE20649). In addition to ALG-1, an-
other Argonaute protein, ALG-2, is
thought to act redundantly with ALG-1
to support it in miRNA processing
(Tops et al. 2006). A third member,
RDE-1, is involved in the double strand
RNA-induced siRNA pathway (Tabara
et al. 1999). Consistent with the pre-
vious report (Corre
ˆa et al. 2010), most
of the reads bound to these Argonaute
proteins were found to be miRNAs.
Notably, we found that the miRNA
reads corresponding to star strands were
significantly enriched in the RDE-1
ChIP-Seq library compared to those in
ALG-1 or ALG-2 (Supplemental Fig.
S7A; Supplemental Table S7) and also
compared to the ratio of mature and
star miRNAs in our deep-sequencing
results from aging samples (mature:
99.85%, and star: 0.15%) (Supplemental
Table S3). Additionally, the accumula-
tion of star miRNAs as well as mature
ones was reduced in rde-1 mutants com-
pared to wild-type animals (Supplemen-
tal Fig. S7B; Supplemental Table S7)
(sequence data were obtained from
GEO, accession: GSE20649) (Corre
ˆa
et al. 2010). A similar reduction of
miRNA star level in rde-1 mutants was
also observed in another library ob-
tained from the GEO, GSE19414 (Gent
et al. 2010) (data not shown). These
results suggest that the star miRNA strands are preferen-
tially sorted into or stabilized in RDE-1 Argonaute protein
in C. elegans.
As observed in our deep-sequencing libraries, in most
cases, the star strands accumulated at much lower levels
than their corresponding mature sequences. However, we
found some cases where both species appeared to accu-
mulate at a similar frequency and others where star reads
were much more abundant than the reads for the mature
miRNA (Fig. 7A; Supplemental Table S3). For each miRNA
with statistically significant expression changes during
agingineithermatureorstarstrand,wetestedthecor-
relation between mature and star miRNA strands over the
time course of aging (Fig. 7B). Of those, seven signifi-
cantly changed the expression of their star strands (red- or
green-colored dots in Fig. 7B). These include miR-788 and
miR-789-2, both of which have more abundant star
strands than mature ones and have a similar trend in ex-
pression change (Figs. 2A and 7B,C, top). Importantly,
there are several cases where the star strands showed
distinct changes in abundance relative to their mature coun-
terparts during aging, including those for miR-34 and miR-
63 (Fig. 7B,C, bottom; Supplemental Table S3), suggesting
the complexity in miRNA processing. Recently, it has been
shown that star miRNA strands also have a capacity to
repress synthetic targets in vitro (Yang et al. 2010).
Although the biological significance of star strands is still
not clear, our findings suggest that they might serve as
regulatory molecules in aging in C. elegans, similar to ma-
ture miRNAs.
FIGURE 7. Changes in expression of miRNA star strands. (A) Mature miRNAs were mostly
more abundant than their star miRNA species (green dots), but in some cases, both mature
and star miRNA strands accumulated at similar frequency (black dots), or star strands were
much more abundant than mature ones (red dots). (B) Correlation of expression changes
between mature and star miRNA strands was tested over the time course of aging (detailed
results are shown in Supplemental Table S3). Each dot indicates the miRNAs for which we
observed statistically significant age-associated expression changes either in their mature or
star strands (mature or star miRNAs with less than 10 reads in total from Day 0 to Day 12 are
not included here since their expression change is less reliable due to an extreme low
abundance). The results were plotted in ascending order by correlation values from left to right.
Of those, blue points represent miRNAs with age-associated changes only in mature strands;
red points represent miRNAs with age-associated expression changes only in star strands;
green points represent miRNAs with age-associated expression changes in both mature and
star strands. As for the five red points, these correspond to miR-229, miR-77, miR-2214, miR-
789-2, and miR-788, from left to right. As for the two green points, these correspond to miR-34
and miR-230. (C) Two examples of miRNAs with age-associated expression changes in their
star strands showed more abundant reads corresponding to star strands than mature ones
(top). The vertical scales on the left represent the number of sequence reads for mature ones,
and those on the right are for star strands. Another two examples showed distinct changes in
abundance during aging between mature and star strands (bottom). Additional miRNAs, such
as miR-54, also appear to have different trends in expression changes between mature and star
strands (Supplemental Table S3), although their expression changes were not statistically
significant due to a lower number of sequence reads.
Kato et al.
1812 RNA, Vol. 17, No. 10

Novel miRNA candidates were
identified from aging animals
In the process of filtering out annotated
sequence reads, we were left with z30,000
unique, unannotated sequencing reads.
These potentially include novel miRNA
candidates and additional age-associated
sncRNAs. Computational prediction with
the miRDeep program (Friedlander et al.
2008) yielded 76 novel miRNA candi-
dates from these unannotated reads (Sup-
plemental Table S8). For 21 of these, we
detected corresponding star strands or
their derivative reads in our data set.
Based on the features, including the pres-
ence of star strands or expression change
during aging (Supplemental Tables S8–
S10), we picked seven novel miRNA
candidates for further verification (Fig.
8A). As shown in Figure 8B, some of the
candidates may fall into known miRNA
families because they had the same ‘‘seed’’
sequences as other known miRNAs. Also,
qRT-PCR experiments showed that some
of them change their expression during
aging (Fig. 8C). Additionally, we con-
structed candidate miRNA promoterTgfp
fusion transgenic lines and detected
GFP signals in adult C. elegans tissues,
such as sensory neurons in the head (Sup-
plemental Fig. S6D), demonstrating that
these sequences are transcribed.
In order to further validate whether
these candidates are bona fide miRNAs,
we examined their expression level in
mutants of alg-1(gk214), which is nec-
essary for miRNA maturation (Grishok
et al. 2001). We found that five of
the candidate miRNAs examined signifi-
cantly reduced their accumulation in
the alg-1(gk214) mutant background, in-
dicating that they are real miRNAs (Fig.
8D, top). However, for two candidates,
405191_adh and 1263561_spe, we could
not validate these as real miRNAs since
they did not show reduced expression in
the alg-1(gk214)oralg-2(ok304)mutants
(Fig. 8D). Although we attempted to
test their expression in an alg-2(ok304);
alg-1(RNAi) background, this genetic combination led to
growth arrest at an early larval stage. Eventually, nine novel
miRNA candidates were officially annotated as miRNAs by
miRBase, which include mir-5545 to mir-5553 (Fig. 8;
Supplemental Tables S8–S10).
Additional sncRNAs are associated with aging
We hypothesized that our deep-sequencing approach would
reveal additional sncRNAs related to the aging process. Of
the z30,000 unique, unannotated reads, the 14,500 unique
FIGURE 8. Characterization of novel miRNA candidates. (A) The secondary structures of
primary miRNA precursors were predicted for novel miRNA candidates using the RNAfold
program. Note that mir-5546 (1128878_adh) was also reported in Stoeckius et al. (2009),
although it has not yet been submitted to miRBase or WormBase. (B) Some novel miRNA
candidates may fall into known miRNA families since they have the same ‘‘seed’’ sequence as
known miRNAs. (C) The age-associated expression changes of novel miRNA candidates were
confirmed by qRT-PCR. The results were normalized by the average of the expression of act-3
and ama-1. Error bars indicate SD. (D) Levels of novel miRNA candidates were examined by
qRT-PCR in mutants of Argonaute family genes [top:alg-1(gk214) and bottom:alg-2(ok304)]
at two different developmental stages—the fourth larval (L4) and young adult. The results were
normalized by the expression level of U18 and standardized to the level in wild-type N2 in each
stage examined. Error bars indicate SD. P-values were calculated by t-test (***: P< 0.0001; **:
P< 0.001; *: P< 0.05).
Small, noncoding RNAs in C. elegans aging
www.rnajournal.org 1813

reads 20 nucleotides (nt) or longer in size were grouped
based on features, such as expression change, chromosomal
location, and nucleotide content (Supplemental Fig. 8A,B;
Supplemental Table S11). We found that z100 of them
significantly changed their expression during aging. Notably,
many of these age-associated reads started with a uracil (U)
or a guanine (G) and include sequence reads of 21 nt starting
with a U (referred to here as 21nt-U-RNA) and sequence
reads of 22 nt and 26 nt starting with a G (referred to here as
22nt-G-RNA and 26nt-G-RNA, respectively) (Fig. 9A).
21nt-U-RNAs contain a subset that includes 21U-RNAs/
piRNAs, a class of sncRNAs involved in transposon silencing
in the germline of flies, C. elegans, and mammals (Batista
et al. 2008; Stefani and Slack 2008). Since typical 21U-
RNAs in C. elegans are preferentially localized on chromo-
some IV and have the core consensus motif ‘‘CTGTTTCA’’
in their upstream regions, we searched for these features
and identified an additional nine novel 21U-RNA candi-
dates. These novel 21U-RNAs, as well as known, annotated
ones, showed a gradual decrease in expression during aging
(Fig. 9B; Supplemental Table S12). Furthermore, although
our remaining 21nt-U-RNAs are not from chromosome IV
and/or do not have the perfect canonical consensus motif,
they still appear to be bona fide 21U-RNAs since z300 of
them show reduced expression in a prg-1 mutant, which is
necessary for 21U-RNAs/piRNAs biogenesis (sequence data
for prg-1 mutants were obtained from GEO, accession:
GSE11735) (Batista et al. 2008; Supplemental Fig. S8C; Sup-
plemental Table S13).
Unlike these 21nt-U-RNAs, which showed a decrease in
expression during aging, both 26nt-G-RNAs and 22nt-
G-RNAs and their derivatives that we cloned appear to be
increased during aging (Fig. 9A, top; Supplemental Table
S11), although for many, their expression levels were quite
low and fluctuated during aging. In
total, we cloned 2082 22nt-G-RNAs
and 525 26nt-G-RNAs in our libraries
(Supplemental Table S14).
These possibly age-associated 26nt-
G-RNAs and 22nt-G-RNAs that we
cloned may correspond to a class of
recently identified endo-siRNAs, 22G-
and 26G-RNAs, which are implicated in
transposon silencing and chromosome
segregation in C. elegans (Claycomb
et al. 2009; Gu et al. 2009). To address
this question, we investigated their ac-
cumulation in mutants of WAGO and
CSR-1 pathways, such as drh-3 and csr-1,
which are necessary for the generation
of these endo-siRNAs. We utilized se-
quence data for these mutants obtained
from GEO, accession: GSE18215 and
GSE18165 (Claycomb et al. 2009; Gu
et al. 2009). 22nt-G-RNAs showed a dra-
matic reduction in their accumulation
in mutants including drh-3, compared
to the control wild-type animals (Sup-
plemental Fig. S8D; Supplemental Table
S14). Also, another abundant RNA spe-
cies in aged animals, 26nt-G-RNAs, ex-
hibited a marked reduction in their ac-
cumulation in mutants including csr-1.
These results indicate that the 22nt-
G-RNAs and 26nt-G-RNAs that we
cloned from an aged animal population
correspond to the recently identified
endo-siRNA species (detailed observa-
tions are discussed in the legend for
Supplemental Fig. S8D).
It is still not clear why these 26nt-
G-RNAs/26G-RNAs and 22nt-G-RNAs/
FIGURE 9. Characterization of additional sncRNA candidates. (A) The total number of
unannotated reads starting with a U or a G was examined in each aging stage. 21nt-U-RNAs
showed a decrease in expression with age, while 26nt-G-RNAs and their related reads appear to
be increased. (B) The total number of 21U-RNA/piRNA reads, including both known and
novel ones, was reduced during aging. (C) tRNA-derived fragments were increased in their
accumulation during aging, and two examples are shown here. The longest bars represent the
annotated mature tRNA sequences, and each shorter bar represents unique sequence reads
corresponding to a part of their mature products. The scores and the color gradation on each
shorter bar indicate the number of sequencing reads for each read found in all aging samples.
The bar graphs represent the expression changes of the total number of tRNA-derived
fragments during aging. All examples are shown in Supplemental Table S15.
Kato et al.
1814 RNA, Vol. 17, No. 10

22G-RNAs seem to be increased during aging and how they
are related to lifespan regulation, since spe-9(hc88) animals
exposed to RNAi against genes involved in their biogenesis,
such as csr-1 and drh-3, did not show obvious lifespan
abnormalities (data not shown). However, it has been
shown that CDE-1 (also known as CID-1 or PUP-1), one
of the factors necessary for the function of CSR-1 22G-RNAs,
has a role in lifespan regulation and thermotolerance, to-
gether with DNA check point proteins, including CHK-1
and CDC-25.1 (Olsen et al. 2006), suggesting the possibility
that these endo-siRNAs may also contribute to aging and
stress response through the maintenance of genome stabil-
ity and genome surveillance.
While the total levels of miRNAs and 21U-RNAs/
piRNAs showed a gradual decrease in accumulation during
aging, other reads corresponding to known noncoding RNAs,
such as tRNAs, rRNAs, and small snoRNAs, showed a
gradual but constant increase in accumulation during aging
(Fig. 1B; Supplemental Table S1). One would expect that
these may be products of degradation since small RNAs
ranging in size from 15–30 nt were selectively purified in
the process of cDNA library preparation. However, at least
in the case of some tRNAs and snoRNAs, they are unlikely
to be simply derived from random degradation because some
specific fragments preferentially accumulate, revealing that
the cleavage events occur at specific sites in the mature
tRNAs and snoRNAs (Fig. 9C; Supplemental Fig. S9;
Supplemental Tables S15, S16). Also, as shown in the bar
graphs in Figure 9C, for different types of tRNAs, their
cleavage products all showed an increase in accumulation
through aging.
It has been reported that tRNA fragments are present
in a wide variety of organisms and often increase under
various stress conditions such as heat and oxidative damage
(Thompson and Parker 2009a). The nucleases responsible
for tRNA cleavage during stress have been identified in
budding yeast and mammalian cells, and both of these
ribonucleases are found to cleave not only tRNAs but also
rRNAs in response to stress (Thompson and Parker 2009b;
Yamasaki et al. 2009). Most recently, it was demonstrated
that stress-induced tRNA cleavage depends on a DNA meth-
yltransferase Dnmt2, and Dnmt2-mediated methylation pro-
tects tRNAs against ribonuclease cleavage in Drosophila
(Schaefer et al. 2010). The biological roles of these novel,
cleaved RNAs remain largely unknown; however, their in-
duction found in this study might be a consequence of
accumulated damage with age.
DISCUSSION
Small, noncoding RNAs, including miRNAs, are key factors
in the control of gene expression and in the maintenance of
genome stability, and here we provide evidence that they
contribute to the regulation of aging processes in C. elegans.
Our expression profiling of sncRNAs using deep-sequenc-
ing technology reveals new molecular features related to
aging. We show in this study that many classes of small
RNAs, including miRNAs, exhibited significant expression
changes during aging in C. elegans. Also, we identified miRNA
star strands, novel miRNAs, tRNA- and snoRNA-derived
short fragments, and endo-siRNAs such as 21U-, 22G-, and
26G-related RNAs as differentially expressed during aging.
One of the most straightforward approaches to un-
derstanding the role of small RNAs in aging is to test the
effects on lifespan of their knockout mutants. This is not
generally a plausible approach for the thousands of 21U-,
22G-, and 26G-RNAs but is plausible for the hundreds
of miRNAs. In fact, in addition to the necessity of alg-
1-mediated miRNA processing in the normal lifespan, we
recently found that some of the miRNAs with changes in
expression during aging result in abnormal lifespan when
they are deleted (de Lencastre et al. 2010). However, z40%
of known miRNAs share significant sequence similarities
(74 of 174 miRNAs [miRBase ver. 14.0] have the same
‘‘seed’’ sequences with other C. elegans miRNAs) (Supple-
mental Table S2), suggesting considerable potential for
functional redundancy in miRNA families. Furthermore, a
single gene may be regulated redundantly by different miRNAs
at the same time, as genes often have multiple miRNA
binding sites. This complexity makes it difficult to reveal
biological roles of individual miRNAs using their genetic
knockouts. However, analyses of miRNA expression and
target prediction, followed by pathway analysis shown in
this study, suggest that the miRNAs identified here function
in known aging processes, such as proteostasis. Further, ex-
pression profiling of miRNAs in longevity mutants or on
RNAi-mediated knockdown of aging-associated genes (e.g.,
daf-2 for long-lived and daf-16 for short-lived conditions)
uncovered additional evidence for miRNA roles in aging.
The miRNA processing machinery yields both mature
and star strands from the same hairpin-loop precursor prod-
ucts, and so one would expect that both mature and star
miRNA strands have the same trend in their expression
patterns. However, we found that some miRNAs have dif-
ferent trends in expression for their mature and star strands
during aging. Additionally, we found that star miRNA strands
preferentially accumulated in a third AGO protein, RDE-1,
which is involved in the siRNA pathway. This feature might
be highly conserved among diverse species since it was also
recently reported in Drosophila that mature miRNA strands
are predominantly sorted into AGO1, and their star strands
accumulate in AGO2, which preferentially acts in the siRNA
pathway, and this strand selection is associated with central
base-pairing mismatches in miRNA precursor duplexes
(Okamura et al. 2009). Also, the same group recently dem-
onstrated that star strands have a capacity to repress
expression of synthetic targets in vitro (Yang et al. 2010).
Although it is not known how strand specificity is achieved
in C. elegans or how important star miRNAs are in any
organisms, the machinery producing distinct age-depen-
Small, noncoding RNAs in C. elegans aging
www.rnajournal.org 1815

dent accumulation patterns of mature/star strands, if any,
will be an important issue since it would potentially affect
aging through strand selection and/or by stability of mature/
star strands.
In terms of miRNA stability, it is also worth investigating
how miRNAs are negatively regulated during aging. Here
we examined changes in expression of miRNAs in temper-
ature-shift-induced modified lifespan backgrounds and
found that miRNAs with decreased expression during aging
rapidly reduced their levels in both normal and long-lived
conditions. Also, we previously found some miRNAs that
changed their expression dynamically during development
(e.g., miR-795, one of the let-7 family miRNAs, which has
its peak expression in the L3 [the third larval stage] and
then is quickly down-regulated toward the young adult
stage) (Kato et al. 2009a). These observations imply a
mechanism that positively reduces levels of mature miRNAs,
potentially conducted by an exoribonuclease like XRN-2
(Chatterjee and Grosshans 2009) or miRNA star strands,
which have a nearly perfect base-pairing match with partner
mature strands, like anti-miRNAs (antisense-miRNA oli-
gonucleotides) (Krutzfeldt et al. 2005).
In addition to miRNA star strands, we identified
additional novel miRNA candidates in our cDNA libraries
for small RNAs. Their expression levels were mostly quite
low, but we found that some of them were expressed in
neurons, including sensory neurons in the head (e.g., ADLs),
suggesting their possible role in the response to environ-
mental stimuli. Although for two of the novel miRNA
candidates, it remains unclear whether they are real miRNAs
or not, since their expression levels were not affected in
either alg-1(gk214)oralg-2(ok304) mutant backgrounds,
they are still interesting candidates to pursue since both of
them showed a decrease in expression during aging. Also,
our validation approach using Argonaute mutants may
reveal an unexpected functional complexity in Argonaute
proteins. Both C. elegans ALG-1 and ALG-2 are members of
the highly conserved RDE-1/AGO1/PIWI family of pro-
teins that regulate post-transcriptional gene silencing, and
it is thought that ALG-1 functions as the primary miRNP,
and ALG-2 acts redundantly with ALG-1 to support it. This
is based on the observation that alg-1(gk214) mutants cause
a vulval phenotype, resulting in premature death at the
young adult stage, while alg-2(ok304) mutants are grossly
normal. Our study of novel miRNA expression in alg-1/2
mutants suggests that some of the miRNAs depend on both
alg-1 and alg-2, and this dependency might vary during
development and possibly during aging as well. Also, we
noticed that one of the novel miRNA candidates that showed
a dramatic reduction in expression in the alg-1(gk214)
mutant background (1277767_adh in Fig. 8D) maps to
a tranposase-coding region, meaning that expression of
a transposase gene-derived small RNA depends on Argo-
naute protein activity. This observation might suggest the
possibility that, like siRNAs, 21U-RNAs/piRNAs, or 22G-
RNAs (Sijen and Plasterk 2003; Batista et al. 2008; Gu et al.
2009), transposon-derived miRNAs also contribute to
genome stability through the control of transposon activity.
In addition to new classes of small noncoding RNAs, we
found an interesting age-associated feature in the known,
classical noncoding RNAs, such as tRNAs and snoRNAs;
cleaved RNA fragments of these noncoding RNAs accumu-
lated during aging. These, especially tRNA-derived shorter
fragments, may have the potential to modulate aging and to
connect aging and the stress response since it has been
shown that tRNA cleavage is induced in response to stress
in several organisms (Thompson and Parker 2009a). We,
indeed, found a dramatic increase in the accumulation of
tRNA cleavage products in C. elegans that were exposed to
stress such as heat shock (M Kato and FJ Slack, unpubl.).
This suggests that stress-induced cleavage of tRNAs is a
conserved, fundamental stress response. Although their
biological function remains largely unclear, some studies
have shown that tRNA-derived fragments were co-purified
with an Argonaute protein complex (Kawamura et al. 2008),
suggesting that such tRNA fragments could potentially
function like siRNAs or miRNAs and regulate the trans-
lational machinery. Indeed, in the case of snoRNAs, recent
reports suggest that snoRNA-derived small RNAs can func-
tion like miRNAs (Ender et al. 2008; Taft et al. 2009). We
also found a significant number of snoRNA-derived short
reads in our libraries which showed an increase in accumu-
lation with age, and indeed, some mature snoRNAs appear
to resemble miRNA hairpin-loop precursors in their sec-
ondary structures (data not shown).
The novel sncRNAs we report in this study, including
cleaved RNA fragments, might also mediate sequence-
specific regulation like miRNAs. Additional study of the
potential targets interacting with these small RNAs might
open up a new field in small RNA-related biology. Also,
several observations suggest that stress-induced RNA cleav-
age and induction of ribonucleases may be involved in
cancer and other diseases (Thompson and Parker 2009a).
Further analysis of small noncoding RNA expression pro-
files and detailed genetic studies will give new insights into
novel molecular networks in aging and common mecha-
nisms in aging and age-related diseases.
MATERIALS AND METHODS
C. elegans strains
The spe-9(hc88) strain was first obtained from CGC (Caenorhabditis
Genetic Center) and then backcrossed nine times to our standard
wild-type N2 lab strain. Animals were maintained for a few gen-
erations without starvation at the permissive temperature, 15°C,
to collect embryos. After embryos were harvested, they were in-
cubated at 23°C in M9 buffer without a food source overnight to
arrest their growth at the early L1 (the first larval) stage.
Synchronized animals were transferred to bacteria-seeded plates
Kato et al.
1816 RNA, Vol. 17, No. 10

and cultured at 23°C to induce their sterility. The aging stages
used for RNA preparation were Day 0, Day 5, Day 8, and Day 12
post-L4 molt, or as shown in each experiment. Day 0 in this study
represents the day of the final molt, z48 h after staged L1 animals
were placed on seeded plates at 23°C. In alg-1 RNAi and tem-
perature-shift experiments, animals were treated with each condi-
tion after Day 0 (young adult stage) of adulthood. The two mutant
strains, alg-1(gk214) and alg-2(ok304), were obtained from the
CGC and maintained at 15°C similar to the spe-9(hc88) strain, but
they were cultured at 15°C during all stages since alg-1(gk214)
mutants are sick and show a vulval bursting phenotype at a higher
temperature. Total RNAs were purified from alg-1/2 mutants 72 h
and 84 h after synchronized L1 stage animals were fed, for L4 and
young adult stages, respectively. For miRNA promoterTgfp fusion
lines, details are shown in Supplemental Figure S6.
Lifespan assays
Lifespan assays were basically conducted as previously described
(Boehm and Slack 2005), except that FUDR (5-fluorodeoxyuri-
dine, a DNA replication inhibitor) was not used in this study.
Similar to those used for RNA isolation mentioned above, syn-
chronized Day 0 young adult animals were transferred into dif-
ferent conditions (e.g., 15°C and on RNAi bacteria). Survival of
animals was examined every day. Animals that crawled off the
plate or burst were excluded from the calculations. Approximately
100 animals were tested on each plate with 2–3 replicates, and at
least two independent assays were performed for each result.
RNA, cDNA library preparation and Solexa
deep-sequencing
spe-9(hc88) animals were first washed with M9 buffer multiple
times in order to remove bacteria. Then, RNAs were purified using
the mirVana miRNA Isolation Kit (Ambion), according to the
manufacturer’s instructions. In the initial deep-sequencing exper-
iment, cDNA libraries were made from 10 mgofRNAsenriched
from a small RNA fraction (<200 nt), which were prepared based
on the manufacturer’s instructions in the mirVana kit. In the
second deep-sequencing experiment (verification sets from Day
0 and Day 8), we used 10 mg of total RNA for constructing cDNA
libraries. Since our sequencing results were essentially identical, we
now confirm that either procedure leads to consistent results. Prep-
aration of cDNA libraries for the Solexa deep-sequencing experi-
ment was carried out using the DGE-Small RNA Sample Prep Kit
ver. 1.0 (Illumina) according to the manufacturer’s instructions.
Specifically, RNAs corresponding to 15–30 nt in size were selectively
purified, and then these were ligated to adapters and amplified by
RT-PCR. The same amount of purified library DNA was captured
on an Illumina flow cell for cluster generation and was sequenced
for 36 cycles on an Illumina Genome Analyzer II following the
manufacturer’s protocols for single-end reads (DGE-Small RNA
Cluster Generation Kit and 36 Cycle Solexa Sequencing Kit). All
our raw deep-sequencing data and processed data are available
from the GEO database (GSE18634).
Quantitative RT-PCR
We used qRT-PCR with TaqMan Small RNA Assays (Applied
Biosystems) for confirmation of the deep-sequencing results and
for examining changes in expression of miRNAs in alg-1 RNAi-
induced abnormal lifespan background, in different environmen-
tal conditions and in the alg-1/2 mutant background, according to
the manufacturer’s instructions. For validating expression of novel
miRNA candidates in the alg-1/2 mutant background, the results
were normalized to the expression level of U18, which is widely
used as a control in TaqMan qRT-PCR. For testing miRNA
expression during aging, we used the expression of act-3 and ama-1
using TaqMan Gene Expression Assay (Applied Biosystems) as
controls, both of which are thought to be constantly expressed in
aging (Evans et al. 2008; Fuhrman et al. 2009). Note that the
results were normalized with the ‘‘average’’ of these multiple stably
expressed reference genes since there is no universally accepted gene
for normalization during aging (Nolan et al. 2006). Also, we used
50–100 ng/ml of total RNAs for novel miRNA candidates and 10
ng/ml for known miRNAs and for controls (e.g., U18) since the
level of expression of novel miRNA candidates was much lower.
The results were analyzed by the delta-delta Ct method, and
P-values were calculated by t-test from delta Ct values of the con-
trol and the target.
Microscopic study of miRNA promoterTgfp
transgenic lines
Age-related changes in GFP signals expressed from miRNA
promoterTgfp lines were examined by microscopic observation.
All strains used were crossed into the spe-9(hc88) background. To
obtain images from a whole animal body, an individual animal
was exposed to a UV light for the same exposure time and with a
focus on the center of each animal based on its pharynx and/or
vulva. Since signals obtained with a longpass GFP (LPGFP) filter
contain both those from intestinal cellular autofluorescence and
transgene-expressing GFP, signals obtained with a TRITC filter
that only reflects age-related autofluorescence were subtracted from
the signals obtained with a LPGFP filter after normalizing the
differences of signal levels in LPGFP and TRITC filters based on
those detected in non-GFP spe-9(hc88) animals at each time point
and each condition. Signal intensities were calculated using the
NIH ImageJ program (http://rsbweb.nih.gov/ij/) from z15–20
animals at each time point of each line.
Computational sequence data analysis
We required perfect sequence matching from the processed,
aligned reads to assess the number of miRNA and other small
RNA reads in each sample if not specified in each computational
analysis. The raw data were aligned to the C. elegans genome using
the SOAP program (ver. 1.10) (Li et al. 2008) with the following
command options: –w 5 –S 3 –A (the 39adapter sequence used)
(maximum two base pair mismatches were allowed as a default
option). We controlled library differences by normalizing to the
total number of reads that matched the C. elegans genome
(WormBase WS190) in each sample. The number of sequence
reads in each sample was finally standardized to the Day 0 sample
in each set of deep-sequencing experiments. The data sets of
known miRNAs and 21U-RNAs were obtained from miRBase
(Release 14.0) and WormBase (WS215), respectively.
For testing the proportion of each noncoding RNA species in
each sample, the number of reads with perfect matches was counted
after searching with the blastn program (ver. 2.2.17). The param-
eters in the blastn program we used are as follows: -e 0.001 -G 5 -E
Small, noncoding RNAs in C. elegans aging
www.rnajournal.org 1817

2 -q -3 -r 1 -W 7 -v 10 -b 10. The miRDeep program was used for
finding novel miRNA candidates by basically following the default
procedure provided with the program (Friedlander et al. 2008),
and the RNAfold program (Vienna RNA package ver. 1.6.5; http://
www.tbi.univie.ac.at/zivo/RNA/) was used for predicting second-
ary structure of primary miRNA transcripts and snoRNAs. All
genome coordinates shown in this study are based on WormBase
genome build WS190.
For clustering analysis of miRNAs, we used the Cluster 3.0
program (uncentered correlation option with normalization) (Eisen
et al. 1998), and the Java TreeView program (Saldanha 2004) was
then used to visualize results.
Deep-sequencing data obtained from the GEO database are as
follows: GSE20649, Argonaute proteins ChIP-Seq libraries, was
used for the analysis of mature and star miRNA species; GSE11735,
prg-1 mutant, was used for the analysis of 21nt-U-RNAs;
GSE18231, mutants in WAGO pathway, and GSE18165, mutants
in CSR-1 pathway, were used for the analysis of 26nt-G-RNAs and
22nt-G-RNAs.
Statistical analysis of deep-sequencing data
To identify differentially expressed miRNAs, pairwised hypothesis
testings for Day 0, Day 5, Day 8, and Day 12 were used. Since we
did not have a complete set of experimental replicates, we took
advantage of two replicates of Day 0 and Day 8. Linear regressions
on mean (m) and variance (s
2
) of 174 miRNAs showed that s
2
=
m
1.7
for both Day 0 and Day 8. Considering that expressions of
different days should have similar distribution patterns (biases
and errors are caused by technical processes), we assumed that all
days follow the normal distribution of N(m,m
1.7
). Based on the
distribution, we used the t-test to calculate the pairwise P-values
of all four stages of aging samples. The final P-value for each
miRNA is the minimum P-value among the six pairwise compar-
isons. To confirm our selection, another method was applied to
our data set. We calculated a time-course P-value across all days,
based on the chi-square distribution after variance stabilizing trans-
formation. Since variance is a power function of mean (s
2
=m
1.7
),
we wanted a transformation of data that had constant variance.
Based on a first-order Taylor series expansion at point mand
supposing s
2
=a
2
m
2b
, we have
fxðÞ=1
a
xb+1
b+1

;
where a=1andb= 0.85. After the transformation of data,
P-values across all days were calculated based on a chi-square
distribution with three degrees of freedom. All differentially ex-
pressed miRNAs identified by the time-course method could also
be identified by the pairwise comparison.
Computational target prediction of age-associated
miRNAs and pathway analysis
To identify the potential targets of age-associated miRNAs, we
combined analysis of the computational target prediction and
gene expression microarray data. First, miRanda (http://www.
microrna.org/microrna/home.do) (John et al. 2004), TargetScan
(http://www.targetscan.org/) (Lewis et al. 2005), and PicTar (http://
www.pictar.org/) (Lall et al. 2006) were used to predict the targets
based on seed matching, cross-species conservation, and targeted
pair energy. Then, to reduce the false positive targets in the pre-
diction, we incorporated the microarray data (Table S3 in Budovskaya
et al. 2008), which have expression levels of protein-coding genes
during aging in C. elegans at the same time points as our small
RNA study. Based on the fact that miRNAs or co-expressed
miRNAs tend to target the genes from the same functional cat-
egories or pathways, we used gene set enrichment analysis to
identify differentially expressed genes. Pathways information was
obtained from KEGG (Kyoto Encyclopedia of Genes and Ge-
nomes; http://www.genome.jp/kegg/.). Then, Fisher’s exact test
was used to find significant pathways. If a gene predicted by one of
the prediction programs mentioned above is also present in the
significant pathways, this gene is considered to be the target of the
corresponding miRNAs. The reduced target list is subjected to
gene set enrichment analysis (by Fisher’s exact test) in order to see
if any interesting pathways are enriched in our predicted targets. A
list of genes involved in or necessary for the normal aging process
is obtained from ‘‘Human Ageing Genomic Resources’’ (http://
genomics.senescence.info/genes/models.html) and based on the GO
Term ‘‘determination of adult life span [GO:0008340].’’
DATA DEPOSITION
Deep-sequencing data from this study has been submitted to the
GEO database: GSE18634.
SUPPLEMENTAL MATERIAL
Supplemental material is available for this article.
ACKNOWLEDGMENTS
We thank Ghia Euskirchen and Hannah Monahan for help with
Solexa sequencing, and Zachary Pincus and Alexandre de Lencastre
for critical reading of this manuscript. We also thank Zhi John Lu,
Mike Wilson, and Nicole Washington for handling of the deep-
sequencing data, and the CGC for strains. We also thank Yale
University Biomedical High Performance Computing Center and
NIH grant RR19895, which funded the instrumentation. M.K.
was partially supported by a postdoctoral fellowship from the
Uehara Life Science Foundation; X.C. was supported by a fellow-
ship from the China Scholarship Council; F.S. was supported by
grants from the NIH to the modENCODE consortium (RFA-
HG-06-006) and R01 AG033921 and from the Ellison Medical
Foundation.
Received March 10, 2011; accepted June 24, 2011.
REFERENCES
Antebi A. 2007. Genetics of aging in Caenorhabditis elegans.PLoS
Genet 3: 1565–1571.
Bagga S, Bracht J, Hunter S, Massirer K, Holtz J, Eachus R, Pasquinelli
AE. 2005. Regulation by let-7 and lin-4 miRNAs results in target
mRNA degradation. Cell 122: 553–563.
Batista PJ, Ruby JG, Claycomb JM, Chiang R, Fahlgren N, Kasschau
KD, Chaves DA, Gu W, Vasale JJ, Duan S, et al. 2008. PRG-1 and
Kato et al.
1818 RNA, Vol. 17, No. 10

21U-RNAs interact to form the piRNA complex required for
fertility in C. elegans.Mol Cell 31: 67–78.
Boehm M, Slack F. 2005. A developmental timing microRNA and its
target regulate life span in C. elegans.Science 310: 1954–1957.
Budovskaya YV, Wu K, Southworth LK, Jiang M, Tedesco P, Johnson
TE, Kim SK. 2008. An elt-3/elt-5/elt-6 GATA transcription circuit
guides aging in C. elegans.Cell 134: 291–303.
Chatterjee S, Grosshans H. 2009. Active turnover modulates mature
microRNA activity in Caenorhabditis elegans.Nature 461: 546–549.
Claycomb JM, Batista PJ, Pang KM, Gu W, Vasale JJ, van Wolfswinkel
JC, Chaves DA, Shirayama M, Mitani S, Ketting RF, et al. 2009.
The Argonaute CSR-1 and its 22G-RNA cofactors are required for
holocentric chromosome segregation. Cell 139: 123–134.
Corre
ˆa RL, Steiner FA, Berezikov E, Ketting RF. 2010. MicroRNA-
directed siRNA biogenesis in Caenorhabditis elegans.PLoS Genet 6:
e1000903. doi: 10.1371/journal.pgen.1000903.
Czech B, Zhou R, Erlich Y, Brennecke J, Binari R, Villalta C, Gordon
A, Perrimon N, Hannon GJ. 2009. Hierarchical rules for Argo-
naute loading in Drosophila.Mol Cell 36: 445–456.
de Lencastre A, Pincus Z, Zhou K, Kato M, Lee SS, Slack FJ. 2010.
MicroRNAs both promote and antagonize longevity in C. elegans.
Curr Biol 20: 2159–2168.
Eisen MB, Spellman PT, Brown PO, Botstein D. 1998. Cluster analysis
and display of genome-wide expression patterns. Proc Natl Acad
Sci 95: 14863–14868.
Ender C, Krek A, Friedlander MR, Beitzinger M, Weinmann L, Chen
W, Pfeffer S, Rajewsky N, Meister G. 2008. A human snoRNA with
microRNA-like functions. Mol Cell 32: 519–528.
Evans EA, Chen WC, Tan MW. 2008. The DAF-2 insulin-like
signaling pathway independently regulates aging and immunity
in C. elegans.Aging Cell 7: 879–893.
Fabian TJ, Johnson TE. 1994. Production of age-synchronous mass
cultures of Caenorhabditis elegans.J Gerontol 49: B145–B156.
Friedlander MR, Chen W, Adamidi C, Maaskola J, Einspanier R,
Knespel S, Rajewsky N. 2008. Discovering microRNAs from deep
sequencing data using miRDeep. Nat Biotechnol 26: 407–415.
Friedman DB, Johnson TE. 1988. Three mutants that extend both
mean and maximum life span of the nematode, Caenorhabditis
elegans, define the age-1 gene. J Gerontol 43: B102–B109.
Fuhrman LE, Goel AK, Smith J, Shianna KV, Aballay A. 2009.
Nucleolar proteins suppress Caenorhabditis elegans innate immu-
nity by inhibiting p53/CEP-1. PLoS Genet 5: e1000657. doi:
10.1371/journal.pgen.1000657.
Garigan D, Hsu AL, Fraser AG, Kamath RS, Ahringer J, Kenyon C.
2002. Genetic analysis of tissue aging in Caenorhabditis elegans:A
role for heat-shock factor and bacterial proliferation. Genetics 161:
1101–1112.
Gent JI, Lamm AT, Pavelec DM, Maniar JM, Parameswaran P, Tao L,
Kennedy S, Fire AZ. 2010. Distinct phases of siRNA synthesis in an
endogenous RNAi pathway in C. elegans soma. Mol Cell 37: 679–
689.
Ghildiyal M, Xu J, Seitz H, Weng Z, Zamore PD. 2010. Sorting of
Drosophila small silencing RNAs partitions microRNA* strands
into the RNA interference pathway. RNA 16: 43–56.
Golden TR, Melov S. 2004. Microarray analysis of gene expression
with age in individual nematodes. Aging Cell 3: 111–124.
Grishok A, Pasquinelli AE, Conte D, Li N, Parrish S, Ha I, Baillie DL,
Fire A, Ruvkun G, Mello CC. 2001. Genes and mechanisms related
to RNA interference regulate expression of the small temporal RNAs
that control C. elegans developmental timing. Cell 106: 23–34.
Gu W, Shirayama M, Conte D Jr, Vasale J, Batista PJ, Claycomb JM,
Moresco JJ, Youngman EM, Keys J, Stoltz MJ, et al. 2009. Distinct
argonaute-mediated 22G-RNA pathways direct genome surveil-
lance in the C. elegans germline. Mol Cell 36: 231–244.
Guo H, Ingolia NT, Weissman JS, Bartel DP. 2010. Mammalian
microRNAs predominantly act to decrease target mRNA levels.
Nature 466: 835–840.
Haigis MC, Yankner BA. 2010. The aging stress response. Mol Cell 40:
333–344.
He L, He X, Lowe SW, Hannon GJ. 2007. MicroRNAs join the p53
network–another piece in the tumour-suppression puzzle. Nat Rev
Cancer 7: 819–822.
Herndon LA, Schmeissner PJ, Dudaronek JM, Brown PA, Listner KM,
Sakano Y, Paupard MC, Hall DH, Driscoll M. 2002. Stochastic and
genetic factors influence tissue-specific decline in aging C. elegans.
Nature 419: 808–814.
Ibanez-Ventoso C, Yang M, Guo S, Robins H, Padgett RW, Driscoll
M. 2006. Modulated microRNA expression during adult lifespan
in Caenorhabditis elegans.Aging Cell 5: 235–246.
John B, Enright AJ, Aravin A, Tuschl T, Sander C, Marks DS. 2004.
Human microRNA targets. PLoS Biol 2: e363. doi: 10.1371/
journal.pbio.0020363.
Johnson CD, Esquela-Kerscher A, Stefani G, Byrom M, Kelnar K,
Ovcharenko D, Wilson M, Wang X, Shelton J, Shingara J, et al.
2007. The let-7 microRNA represses cell proliferation pathways in
human cells. Cancer Res 67: 7713–7722.
Johnston J, Iser WB, Chow DK, Goldberg IG, Wolkow CA. 2008.
Quantitative image analysis reveals distinct structural transitions
during aging in Caenorhabditis elegans tissues. PLoS ONE 3: e2821.
doi: 10.1371/journal.pone.0002821.
Kanehisa M, Goto S, Furumichi M, Tanabe M, Hirakawa M. 2010.
KEGG for representation and analysis of molecular networks
involving diseases and drugs. Nucleic Acids Res 38: D355–D360.
Kato M, de Lencastre A, Pincus Z, Slack FJ. 2009a. Dynamic
expression of small noncoding RNAs, including novel microRNAs
and piRNAs/21U-RNAs, during Caenorhabditis elegans develop-
ment. Genome Biol 10: R54. doi: 10.1186/gb-2009-10-5-r54.
Kato M, Paranjape T, Muller RU, Nallur S, Gillespie E, Keane K,
Esquela-Kerscher A, Weidhaas JB, Slack FJ. 2009b. The mir-34
microRNA is required for the DNA damage response in vivo in C.
elegans and in vitro in human breast cancer cells. Oncogene 28:
2419–2424.
Kawamura Y, Saito K, Kin T, Ono Y, Asai K, Sunohara T, Okada TN,
Siomi MC, Siomi H. 2008. Drosophila endogenous small RNAs
bind to Argonaute 2 in somatic cells. Nature 453: 793–797.
Kenyon CJ. 2010. The genetics of aging. Nature 464: 504–512.
Kenyon C, Chang J, Gensch E, Rudner A, Tabtiang R. 1993. A C.
elegans mutant that lives twice as long as wild type. Nature 366:
461–464.
Klass MR. 1977. Aging in the nematode Caenorhabditis elegans: Major
biological and environmental factors influencing life span. Mech
Ageing Dev 6: 413–429.
Klass MR. 1983. A method for the isolation of longevity mutants in
the nematode Caenorhabditis elegans and initial results. Mech
Ageing Dev 22: 279–286.
Kozomara A, Griffiths-Jones S. 2011. miRBase: Integrating microRNA
annotation and deep-sequencing data. Nucleic Acids Res 39: D152–
D157.
Krutzfeldt J, Rajewsky N, Braich R, Rajeev KG, Tuschl T, Manoharan
M, Stoffel M. 2005. Silencing of microRNAs in vivo with
‘‘antagomirs’’. Nature 438: 685–689.
Lall S, Grun D, Krek A, Chen K, Wang YL, Dewey CN, Sood P, Colombo
T, Bray N, Macmenamin P, et al. 2006. A genome-wide map of
conserved microRNA targets in C. elegans.Curr Biol 16: 460–471.
Lee RC, Feinbaum RL, Ambros V. 1993. The C. elegans heterochronic
gene lin-4 encodes small RNAs with antisense complementarity to
lin-14. Cell 75: 843–854.
Lewis BP, Burge CB, Bartel DP. 2005. Conserved seed pairing, often
flanked by adenosines, indicates that thousands of human genes
are microRNA targets. Cell 120: 15–20.
Li R, Li Y, Kristiansen K, Wang J. 2008. SOAP: Short oligonucleotide
alignment program. Bioinformatics 24: 713–714.
Lim LP, Lau NC, Weinstein EG, Abdelhakim A, Yekta S, Rhoades
MW, Burge CB, Bartel DP. 2003. The microRNAs of Caenorhab-
ditis elegans.Genes Dev 17: 991–1008.
Lund J, Tedesco P, Duke K, Wang J, Kim SK, Johnson TE. 2002.
Transcriptional profile of aging in C. elegans.Curr Biol 12: 1566–
1573.
Small, noncoding RNAs in C. elegans aging
www.rnajournal.org 1819

Mattout A, Dechat T, Adam SA, Goldman RD, Gruenbaum Y. 2006.
Nuclear lamins, diseases, and aging. Curr Opin Cell Biol 18: 335–341.
McCarroll SA, Murphy CT, Zou S, Pletcher SD, Chin CS, Jan YN,
Kenyon C, Bargmann CI, Li H. 2004. Comparing genomic
expression patterns across species identifies shared transcriptional
profile in aging. Nat Genet 36: 197–204.
Morimoto RI. 2008. Proteotoxic stress and inducible chaperone
networks in neurodegenerative disease and aging. Genes Dev 22:
1427–1438.
Morris JZ, Tissenbaum HA, Ruvkun G. 1996. A phosphatidylinositol-
3-OH kinase family member regulating longevity and diapause in
Caenorhabditis elegans.Nature 382: 536–539.
Nolan T, Hands RE, Bustin SA. 2006. Quantification of mRNA using
real-time RT-PCR. Nat Protoc 1: 1559–1582.
Okamura K, Liu N, Lai EC. 2009. Distinct mechanisms for microRNA
strand selection by Drosophila Argonautes. Mol Cell 36: 431–444.
Olsen A, Vantipalli MC, Lithgow GJ. 2006. Checkpoint proteins
control survival of the postmitotic cells in Caenorhabditis elegans.
Science 312: 1381–1385.
Pak J, Fire A. 2007. Distinct populations of primary and secondary
effectors during RNAi in C. elegans.Science 315: 241–244.
Roush S, Slack FJ. 2008. The let-7 family of microRNAs. Trends Cell
Biol 18: 505–516.
Ruby JG, Jan C, Player C, Axtell MJ, Lee W, Nusbaum C, Ge H, Bartel
DP. 2006. Large-scale sequencing reveals 21U-RNAs and addi-
tional microRNAs and endogenous siRNAs in C. elegans.Cell 127:
1193–1207.
Saldanha AJ. 2004. Java Treeview–extensible visualization of micro-
array data. Bioinformatics 20: 3246–3248.
Schaefer M, Pollex T, Hanna K, Tuorto F, Meusburger M, Helm M,
Lyko F. 2010. RNA methylation by Dnmt2 protects transfer RNAs
against stress-induced cleavage. Genes Dev 24: 1590–1595.
Sijen T, Plasterk RH. 2003. Transposon silencing in the Caenorhabditis
elegans germ line by natural RNAi. Nature 426: 310–314.
Singson A, Mercer KB, L’Hernault SW. 1998. The C. elegans spe-9
gene encodes a sperm transmembrane protein that contains EGF-
like repeats and is required for fertilization. Cell 93: 71–79.
Stefani G, Slack FJ. 2008. Small noncoding RNAs in animal de-
velopment. Nat Rev Mol Cell Biol 9: 219–230.
Stoeckius M, Maaskola J, Colombo T, Rahn HP, Friedlander MR, Li
N, Chen W, Piano F, Rajewsky N. 2009. Large-scale sorting of C.
elegans embryos reveals the dynamics of small RNA expression.
Nat Methods 6: 745–751.
Tabara H, Sarkissian M, Kelly WG, Fleenor J, Grishok A, Timmons L,
Fire A, Mello CC. 1999. The rde-1 gene, RNA interference, and
transposon silencing in C. elegans.Cell 99: 123–132.
Taft RJ, Glazov EA, Lassmann T, Hayashizaki Y, Carninci P, Mattick
JS. 2009. Small RNAs derived from snoRNAs. RNA 15: 1233–1240.
Thompson DM, Parker R. 2009a. Stressing out over tRNA cleavage.
Cell 138: 215–219.
Thompson DM, Parker R. 2009b. The RNase Rny1p cleaves tRNAs
and promotes cell death during oxidative stress in Saccharomyces
cerevisiae.J Cell Biol 185: 43–50.
Tops BB, Plasterk RH, Ketting RF. 2006. The Caenorhabditis elegans
Argonautes ALG-1 and ALG-2: Almost identical yet different. Cold
Spring Harb Symp Quant Biol 71: 189–194.
Vella MC, Choi EY, Lin SY, Reinert K, Slack FJ. 2004. The C. elegans
microRNA let-7 binds to imperfect let-7 complementary sites from
the lin-41 39UTR. Genes Dev 18: 132–137.
Wallace DC. 2005. A mitochondrial paradigm of metabolic and
degenerative diseases, aging, and cancer: A dawn for evolutionary
medicine. Annu Rev Genet 39: 359–407.
Yamasaki S, Ivanov P, Hu GF, Anderson P. 2009. Angiogenin cleaves
tRNA and promotes stress-induced translational repression. J Cell
Biol 185: 35–42.
Yang JS, Phillips MD, Betel D, Mu P, Ventura A, Siepel AC, Chen KC,
Lai EC. 2010. Widespread regulatory activity of vertebrate
microRNA* species. RNA 17: 312–326.
Kato et al.
1820 RNA, Vol. 17, No. 10