Argonaute protein identity and pairing geometry determine
cooperativity in mammalian RNA silencing
JENNIFER A. BRODERICK,1WILLIAM E. SALOMON,2SEAN P. RYDER,2NEIL ARONIN,3
and PHILLIP D. ZAMORE2,4,5
1Program in Neuroscience, University of Massachusetts Medical School, Worcester, Massachusetts 01605, USA
2Department of Biochemistry and Molecular Pharmacology, University of Massachusetts Medical School, Worcester, Massachusetts 01605, USA
3Department of Medicine, University of Massachusetts Medical School, Worcester, Massachusetts 01605, USA
4Howard Hughes Medical Institute, University of Massachusetts Medical School, Worcester, Massachusetts 01605, USA
Small RNAs loaded into Argonaute proteins direct silencing of complementary target mRNAs. It has been proposed that
multiple, imperfectly complementary small interfering RNAs or microRNAs, when bound to the 39 untranslated region of a
target mRNA, function cooperatively to silence target expression. We report that, in cultured human HeLa cells and mouse
embryonic fibroblasts, Argonaute1 (Ago1), Ago3, and Ago4 act cooperatively to silence both perfectly and partially com-
plementary target RNAs bearing multiple small RNA-binding sites. Our data suggest that for Ago1, Ago3, and Ago4, multiple,
adjacent small RNA-binding sites facilitate cooperative interactions that stabilize Argonaute binding. In contrast, small RNAs
bound to Ago2 and pairing perfectly to an mRNA target act independently to silence expression. Noncooperative silencing by
Ago2 does not require the endoribonuclease activity of the protein: A mutant Ago2 that cannot cleave its mRNA target also
silences noncooperatively. We propose that Ago2 binds its targets by a mechanism fundamentally distinct from that used by the
three other mammalian Argonaute proteins.
Keywords: Argonaute; cooperativity; microRNA; miRNA; RNAi; siRNA; RNA silencing
In plants and animals, small silencing RNAs such as small
interfering RNAs (siRNAs) and microRNAs (miRNAs) pro-
vide the specificity determinants for Argonaute proteins. A
small RNA guide bound to an Argonaute protein is called
the RNA-induced silencing complex (RISC) (Hammond
et al. 2000; Hannon 2002; Du and Zamore 2007; Matranga
and Zamore 2007; Ghildiyal and Zamore 2009); binding of
RISC to the 39 untranslated region (UTR) of an mRNA si-
lences its expression (Lee et al. 1993; Wightman et al. 1993;
Olsen and Ambros 1999; Lai 2002; Doench et al. 2003;
Grimson et al. 2007). Argonaute proteins are structural
homologs of RNase H that typically cleave their target RNAs
after the nucleotide paired to the tenth base of the small
RNA guide (Elbashir et al. 2001a,b; Tolia and Joshua-Tor
2007). Cleavage requires three key amino acids—D, D,
H—that form a magnesium-binding catalytic triad, which
promotes nucleophilic attack by hydroxide on the phos-
phodiester bond (Kanaya et al. 1996; Haruki et al. 2000;
Martinez and Tuschl 2004; Schwarz et al. 2004; Song et al.
2004; Rivas et al. 2005).
The human genome encodes four Argonaute paralogs—
Ago1, Ago2, Ago3, and Ago4—and most cultured mamma-
lian cell lines express all four proteins, albeit in different
proportions (Meister et al. 2004). Of the four mammalian
Argonautes, only Ago2 retains the ability to catalyze site-
specific, small RNA-directed endonucleolytic target cleavage
(Liu et al. 2004; Meister et al. 2004). Like Ago2, Ago3
contains an apparent catalytic triad, but unlike Ago2, it lacks
endoribonuclease activity. For Ago1 and Ago4, there is no
catalytic triad, explaining their lack of endoribonuclease
activity (Meister et al. 2004; Rivas et al. 2005; Azuma-Mukai
et al. 2008). Extensive, but not complete, complementarity
between a small RNA guide and an mRNA is required for
Argonaute-catalyzed target cleavage (Hutva ´gner and Zamore
2002; Schwarz et al. 2002; Haley and Zamore 2004; Liu et al.
2004; Meister et al. 2004; Rivas et al. 2005). In contrast, small
RNAs with only partial complementarity to their target
mRNAs, especially those bearing mismatches near the cleav-
age site, cannot direct endonucleolytic cleavage of their
Article published online ahead of print. Article and publication date are
RNA (2011), 17:1858–1869. Published by Cold Spring Harbor Laboratory Press. Copyright ? 2011 RNA Society.
target (Holen et al. 2002) but instead reduce the stability of
the target mRNA (Guo et al. 2010) and, in some conditions,
cause translational repression (Doench et al. 2003; Doench
and Sharp 2004).
Experimental and computational analyses suggest that a
single miRNA can regulate hundreds of genes, because a tar-
get mRNA need only pair with the seed sequence of a small
RNA—comprising nucleotides 2 through 7 or 8—to recruit
RISC and promote repression (Lewis et al. 2003; Doench
and Sharp 2004; Rajewsky and Socci 2004; Brennecke
et al. 2005; Krek et al. 2005; Lewis et al. 2005; Lim et al.
2005; Grimson et al. 2007; Baek et al. 2008; Selbach et al.
2008; Friedman et al. 2009). Multiple, partially comple-
mentary small RNAs, when bound to the 39 UTR of a
luciferase reporter target mRNA, may function cooperatively
to repress its translation (Doench et al. 2003; Bartel and
Chen 2004), and most mRNAs contain multiple potential
miRNA-binding sites in their 39 UTRs (Lee et al. 1993;
Wightman et al. 1993; Reinhart et al. 2000; Abrahante
et al. 2003; Lin et al. 2003; Bartel
2004; Grimson et al. 2007; Friedman
et al. 2009). However, the molecular
basis for cooperativity in small RNA
silencing remains unknown.
Here, we show that both the nature
of siRNA:mRNA target pairing and the
identity of the Argonaute protein to
which the small RNA is bound deter-
mine whether multiple target sites act
cooperatively to recruit RISC. Small
RNAs that pair perfectly to multiple
target sites silenced noncooperatively
when the small RNA guide acts through
Ago2, whereas silencing directed by
either perfectly or imperfectly pairing
small RNAs bound to Ago1, Ago3, or
Ago4 acted cooperatively to silence
mRNA bearing multiple small-RNA-
binding sites. Cooperativity required ad-
jacent sites. Surprisingly, noncooperative
silencing by perfectly pairing small RNAs
bound to Ago2 did not require target
cleavage, as catalytically inactive mutant
Ago2 silenced essentially as well as wild
type. Finally, we find that computation-
ally predicted modes of miRNA:target
pairing required far more small RNA to
achieve repression than more extensively
but still incompletely paired small RNA
ing of RISC to multiple adjacent sites,
combined with high intracellular concen-
trations of miRNAs, allows robust regu-
lation of mRNA targets by Ago1, Ago3,
At least three distinct regulatory mechanisms could explain
the enhanced silencing of reporter mRNAs containing mul-
tiple miRNA-binding sites (Fig. 1). A cooperative binding
model posits that the binding of a miRNA:Argonaute pro-
tein complex to one site increases the affinity of a second
miRNA:Argonaute complex for an adjacent site (Fig. 1A).
In this model, the binding of the first bulged siRNA would
have a higher dissociation constant, KA
binding events, KB
D; we predict that the amount of
siRNA required to silence a reporter would decrease with
an increasing number of target sites as cooperativity be-
tween bound Argonautes increases. Such cooperativity in
small RNA-directed silencing might arise from direct
interactions between adjacent Argonaute proteins. Alterna-
tively, a pair of Argonaute proteins might be bridged by
one or more additional proteins. In a cooperative function
D, than subsequent
FIGURE 1. Potential sources of cooperativity in the repression of a target mRNA by the small
RNA-directed Argonaute complex, RISC. (A) Cooperative binding. RISC binding at multiple
target sites increases site occupancy by mutually stabilizing subsequent binding of RISCs. (B)
Cooperative function. RISC binding at multiple sites may increase the likelihood that
repressive factors, such as nucleases, are recruited to the mRNA. (C) Multiple independent
sites. Each RISC functions independently, so the multiple sites increase the probability of
repression but do not influence each other.
Requirements for cooperativity in RNA silencing
model, multiple miRNA:Argonaute complexes bind to the
target mRNA independently, but the interaction of one
miRNA:Argonaute complex could recruit binding proteins
which block translation of the target mRNA or decrease the
stability of the target (Fig. 1B). Historically, such protein
targets of RISC have been envisioned to include compo-
nents or regulators of the ribosome but more likely corre-
spond to factors that promote accumulation of the target
RNA in a P-body, where it would be degraded (Liu et al.
2005; Rehwinkel et al. 2005; Eulalio et al. 2007; Parker and
Sheth 2007; Guo et al. 2010). In the cooperative function
model, we predict that the presence of three bulged siRNAs on
the target would have a lower inhibitory constant, KABC
for the presence of two (KAB
amount of siRNA required to silence a reporter would de-
crease with increasing number of target sites occupied by
Argonautes and/or a protein factor X, until the concentration
of Argonaute or factor X becomes limiting. Finally, in a
multiple independent sites model, each miRNA:Argonaute
complex binds and acts independently, but the presence of
multiple miRNA-binding sites in the target increases its ef-
fective miRNA occupancy: I.e., the probability that the target
mRNA is bound by at least one miRNA is increased by the
presence of multiple sites (Fig. 1C). Such statistical effects
cause the macroscopic binding constant, K, representative of
all possible combinations of target mRNA with n sites where
at least one site is occupied, to be determined by the statistical
factors of identical microscopic binding constants, k, to give
a fractional saturation of target: [1/n]k (Cantor and Schimmel
1980). If we assume that the IC50is governed by binding and
that the microscopic binding constant for a single site is
essentially identical to the macroscopic binding constant for
the one-site target, then we expect the IC50for the three-site
target to be 1/3(IC50, one-site target).
In contrast to the cooperativity ascribed to miRNA-
directed changes in mRNA stability or translation, small
RNA-guided target cleavage—that is, RNAi—is thought
to be noncooperative, with each RISC acting independently
at each complementary site on the target mRNA. The pres-
ence of multiple, independent small RNA-binding sites in
a target would increase its effective occupancy by RISC: the
probability that the target mRNA is cleaved by at least one
molecule of RISC is increased by the presence of multiple
To evaluate the efficacy of silencing and the extent of
cooperativity directed by a small silencing RNA bound at
one or multiple sites on an mRNA, we established an exper-
imental system comprising six Renilla luciferase reporter
plasmids, each expressing an mRNA bearing one to six
identical, adjacent target sites in its 39 UTR (Fig. 2, left).
We tested four siRNAs whose guide strands pair to dif-
ferent extents with the target sites (Fig. 2, top). The four
siRNAs enabled evaluation of four siRNA:mRNA target
RNA binding modes—perfect pairing, bulged pairing
(mismatched at positions 9 and 10 of the guide strand),
i) or one (KA
i) bulged siRNA; the
seed pairing with supplemental 39 pairing (matching the
target at positions 2–8 and 13–16 of the guide strand), and
seed-only pairing (paired only at positions 2–8 of the guide
strand). Each siRNA duplex was designed to ensure
preferential loading of its guide strand into RISC (Schwarz
et al. 2003). Because all of our experiments comparing
distinct modes of miRNA:target pairing employed a com-
mon target reporter mRNA, our strategy avoids differ-
ences in local target mRNA structure that might confound
For each of the 24 reporter-siRNA combinations tested in
HeLa cells, we calculated the concentration of siRNA re-
quired to achieve half-maximal silencing (IC50) and the Hill
coefficient (nH), a measure of cooperativity, using dose-
response data from at least 12 independent experiments,
each evaluating silencing at $10 siRNA concentrations and
spanning a 2000-fold concentration range. For each siRNA,
we confirmed that the siRNA was inherently active by vali-
dating its ability to silence a Renilla luciferase reporter con-
taining a single, fully complementary siRNA-binding site
(Fig. 3). For the four siRNAs, the mean IC50 values 6
standard deviation for the corresponding perfect, single-site
reporter mRNA ranged from 0.27 6 0.22 nM to 1.33 6 0.78
nM, establishing that all four siRNAs were active.
Silencing by perfect pairing at multiple sites
is not cooperative
Next, we targeted each reporter for silencing by a perfect
siRNA to determine if increasing the number of target sites
reduced the amount of siRNA needed to silence the re-
porter. Based on the multiple independent sites model, we
anticipated that a reporter mRNA bearing more target sites
would be more likely to recruit RISC and would, therefore,
show a reduced IC50for a fully complementary (‘‘perfect’’)
siRNA. Instead, our data suggest that RISC neither binds
nor functions appreciably better when the target contained
multiple sites (Fig. 2). In fact, reporter mRNAs bearing
three (IC50= 0.75 6 0.93 nM), four (IC50= 0.25 6 0.09
nM), five (IC50= 0.41 6 0.33 nM), or six (IC50= 0.30 6
0.14 nM) perfect sites had essentially indistinguishable
IC50values and were silenced only slightly better than a re-
porter bearing a single perfect site (IC50= 0.63 6 0.25 nM).
None of the six reporters displayed positive cooperativity
with the perfectly matched siRNA, with the Hill coefficients
ranging from nH= 0.8 6 0.2 (six-site reporter) to 1.2 6 0.2
(four-site reporter). None of the Hill coefficients were sig-
nificantly different from nH= 1 (P-value > 0.05) (Supple-
mental Fig. S1). Moreover, when the same perfect siRNA
was used to silence a reporter bearing three sites separated
by 19 nt, the IC50and the Hill coefficient were similar to
the mRNA reporter with a single site (IC50= 0.37 6 0.39
nM; nH= 1.0 6 0.2) (Fig. 4).
In general, silencing by the perfect siRNA was well
described by a sigmoidal curve with a Hill coefficient of
Broderick et al.
RNA, Vol. 17, No. 10
one, irrespective of the number of sites (Fig. 2, left). We
conclude that, when guided by a perfectly pairing siRNA,
each RISC acts independently from other RISCs that bind
to nearby target sites.
Multiple bulged sites act cooperatively
Central internal loops—or ‘‘bulges’’—typically block siRNA-
or miRNA-directed target cleavage, the most potent post-
FIGURE 2. Extent of pairing and target site number determine both efficacy and cooperativity in small RNA-directed silencing in HeLa cells.
Silencing of a Renilla luciferase reporter mRNA bearing 1–6 target sites in its 39 UTR, relative to a firefly luciferase internal control, was
determined at different siRNA concentrations. Pairing between the siRNA guide (red) to the 39 UTR sites (black) is shown at top. IC50and Hill
coefficient (nH) were calculated for each dose-response curve. Throughout this study, values are reported as mean 6 standard deviation for IC50
values and nH; error bars indicate standard error for $12 biological replicates. The curves correspond to the concentration-dependence of
silencing expected for the mean IC50and nHvalues.
Requirements for cooperativity in RNA silencing
transcriptional silencing mechanism (Holen et al. 2002; Du
et al. 2005; Dahlgren et al. 2008; Huang et al. 2009). Consis-
tent with that view, effective silencing by a small RNA that
forms a central bulge when paired to its target site required
a higher concentration of siRNA than did the corresponding
perfect siRNA, even when comparing multiple bulged sites to
a single perfect site (Table 1; Fig. 2). In fact, we were unable
to achieve half-maximal silencing of a Renilla luciferase re-
porter bearing one or two bulged sites even at 20 nM
transfected siRNA. For a reporter bearing six bulged sites,
the IC50was nearly three times greater than that for perfect
sites. Unlike silencing mediated by perfect sites, silencing via
bulged sites showed positive cooperativity, with a Hill coef-
ficient of 2.5 6 0.8 (P = <0.0001 for six sites) (Table 1).
Seed matches and supportive pairing
Although bulged sites have been shown to effectively silence
both reporter mRNAs and endogenous genes (Zeng et al.
2002; Doench et al. 2003; Doench and Sharp 2004), they
rarely occur for natural miRNAs and their endogenous
targets (Lewis et al. 2003; Vella et al. 2004; Yekta et al. 2004;
Brennecke et al. 2005; Lewis et al. 2005; Grimson et al.
2007). Instead, miRNAs generally pair with the mRNAs
they regulate at positions 2–8 of the guide strand—the seed
sequence (Lewis et al. 2003; Brennecke
et al. 2005; Krek et al. 2005; Lewis et al.
2005). Additional base pairs between
the mRNA and miRNA positions 13–16
(Grimson et al. 2007) and target adeno-
sines flanking the seed match sequence
at position 1 (t1A) and 9 (t9A) (Lewis
et al. 2005) enhance the likelihood that a
miRNA will regulate a putative mRNA
We tested seed-matched (t1A) sites
with supplemental 39 pairing and seed
only (t1A) sites for their ability to reg-
ulate reporter mRNA bearing one to six
siRNA-binding sites. A seed match plus
supplemental 39 pairing required far
more siRNA to achieve silencing equiv-
alent to the bulged sites (Fig. 2). For ex-
ample, the bulged siRNA regulated the
three-site reporter with an IC50= 1.9 6
0.5 nM, whereas half-maximal silencing
for the same reporter with the siRNA
pairing with both the seed and supple-
mental 39 nucleotides could not be
achieved even using 20 nM siRNA.
With six sites in the reporter, the siRNA
with seed plus supplemental 39 pairing
achieved an IC50= 3.7 6 1.4 nM. The
seed siRNA was even less potent, reach-
ing half-maximal silencing at a siRNA
concentration of 10 6 2.4 nM only for the reporter mRNA
with six sites; the IC50could not be reliably determined for
mRNAs with fewer than six sites. Our data suggest that the
intracellular concentration of functional miRISC exceeds
the RISC concentration we achieved using transfected,
synthetic siRNA duplexes.
Most studies of small RNA-directed silencing report the
extent of repression (‘‘fold-repression’’) for a single concen-
tration of small RNA. To permit comparison of our data to
those in the published literature, we used our data to
calculate the observed ‘‘fold-repression’’ of the multiple-site
reporters by the seed-only siRNA (Supplemental Table S1).
Like Grimson et al. (2007) before us, we observe a $1.4-fold
repression of targets bearing two or more small RNA-bind-
ing sites when using the seed-only siRNA. For three or more
sites, the observed repression, which ranged from 1.8- to 4.8-
fold, was significantly different from that predicted by a
multiple, independent sites model (0.002 # P # 0.04)
(Supplemental Table S1).
Cooperativity requires adjacent target sites
Silencing for bulged sites displayed positive cooperativity
for all multiple-site reporter mRNAs for which we could
FIGURE 3. siRNA validation in HeLa cells. Each siRNA was functional in silencing a reporter
containing a single perfect target site. (A) Perfect siRNA. (B) Bulged siRNA. (C) siRNA with
seed plus supplementary 39 pairing (nt 13–16). (D) siRNA with only seed pairing. The curves
correspond to the concentration-dependence of silencing expected for the mean IC50and nH
values (6 standard deviation) calculated from three independent trials.
Broderick et al.
RNA, Vol. 17, No. 10
measure the IC50and Hill coefficient. To test if the sites
need to be adjacent in order to observe positive coopera-
tivity, we altered the sequence of every other target site in
the six-site Renilla luciferase mRNA to create a three-site
reporter in which 19 nt separate each site targeted by the
siRNAs (Fig. 4A). Silencing of this expanded three-site
reporter mRNA by the bulged siRNA required >15-fold more
siRNA and showed no evidence of cooperativity (IC50$ 20
nM; nH= 0.8 6 0.1) (Fig. 4A) relative to the reporter mRNA
in which the three sites were adjacent
(IC50= 1.3 6 0.8 nM; nH= 1.6 6 0.4)
In theory, these data might reflect
reduced target-site accessibility in the ex-
panded three-site reporter (Brown et al.
2005; Ameres et al. 2007; Tafer et al.
2008). We view this as unlikely. First,
both the adjacent and expanded three-
site reporters were silenced equally well
by a perfectly pairing siRNA (IC50 =
0.37 6 0.39 nM versus 0.22 6 0.15 nM)
(Fig. 4C,D). Second, antisense oligonu-
cleotide-directed RNase H cleavage at
each of the target sites occurred with
similar rates (6.1, 6.0, and 6.6 nM min?1
for sites 1, 2, and 3, respectively) (Sup-
plemental Fig. S2). Finally, the RNase H
cleavage kinetics fit better to a model of
dent, sequential model for cleavage at each
site (Supplemental Fig. S2).
Silencing by the perfectly paired siRNA
was noncooperative for both the ex-
panded (nH= 1.0 6 0.2) and original
(nH= 1.1 6 0.1) three-site reporter
mRNAs (Fig. 4C,D). The observation
that cooperative silencing by a small RNA
requires that fewer than 19 nt separate
the RISC-binding sites to promote effi-
cient, cooperative silencing suggests that
cooperativity springs from interactions between adjacent
ment of proteins involved in subsequent steps in repressing
Ago2-RISC binding prevents cooperative silencing
A simple explanation for why Ago2 acts noncooperatively
to silence a multiple-site reporter with a perfect guide is
FIGURE 4. Cooperative binding of RISC requires adjacent target sites in HeLa cells. Three
sites spaced 19 nt apart (A) require more siRNA to achieve half-maximal silencing, compared
to three adjacent sites (B), and act noncooperatively. In contrast, a perfectly matched siRNA
silences a three-site reporter with sites separated by 19 nt (C) or a reporter with three adjacent
sites (D) with equal efficacy and without detectable cooperativity. The three adjacent-site
experiments in this figure were performed independently from those in Figure 2. A one sample,
two-tailed Student’s t-test was used to calculate the P-values at 95% confidence for the Hill
coefficients to determine if nHwas significantly different from the null hypothesis: nH= 1 (i.e.,
TABLE 1. Concentration-dependence and cooperativity for distinct siRNA:target pairing modes using reporters bearing one to six
Number of sites
Perfect BulgedSeed plus 13–16 Seed only
0.63 6 0.25
1.99 6 0.63
0.75 6 0.93
0.25 6 0.09
0.41 6 0.33
0.30 6 0.14
0.9 6 0.1
0.9 6 0.2
1.1 6 0.1
1.2 6 0.2
1.2 6 0.3
0.8 6 0.2
1.9 6 0.5
0.83 6 0.61
0.81 6 0.20
0.87 6 0.45
1.3 6 0.5
1.7 6 0.5
1.7 6 0.8
2.5 6 0.8
6.3 6 3.3
3.7 6 1.4
2.5 6 0.9
2.3 6 0.810.0 6 2.41.9 6 0.5
(N.D.) not determined. IC50values (nM) and Hill coefficients (nH) of the fitted curves are reported as mean values 6 standard deviation for the
IC50and Hill coefficients for at least 12 trials.
Requirements for cooperativity in RNA silencing
that silencing reflects the endonucleolytic activity unique to
mammalian Ago2. To test this idea, we evaluated silencing
of the six-site reporter mRNA in three mouse embryonic
fibroblast (MEF) cell lines derived from an Ago2 knockout
mouse: Ago2?/?MEFs, Ago2?/?MEFs reconstituted with
mouse Ago2, and Ago2?/?MEFs reconstituted with a mu-
tant mouse Ago2 in which aspartic acid 669 was changed to
alanine (D669A) (O’Carroll et al. 2007). The D669A mu-
tant Ago2 cannot cleave an RNA target (Liu et al. 2004). In
Ago2?/?MEF cells, the perfect siRNA and the bulged
siRNA were both cooperative: nH
suggest that Ago1, Ago3, and Ago4 bind cooperatively to
a reporter mRNA bearing multiple small RNA-binding
sites, irrespective of the nature of small RNA:target pairing.
As expected, repression mediated by a perfectly pairing
siRNA was noncooperative (nH
MEFs reconstituted with overexpressed Ago2 (Figs. 5B, 6).
In contrast, silencing directed by the bulged siRNA in the
Ago2-reconstituted cells was cooperative (nH
0.2; P = 0.02) (Fig. 5B). To test whether the apparent
cooperativity observed in reconstituted Ago2 MEF cells was
caused by Ago2 overexpression, we measured silencing in
perfect= 1.6 6 0.4; P = 0.03,
bulged= 1.8 6 0.3; P = 0.006 (Fig. 5A). These data
perfect= 1.0 6 0.1) in Ago2?/?
bulged= 1.5 6
Ago1?/?MEF cells, which express far less Ago2 mRNA and
protein than reconstituted Ago2?/?MEF cells (Figs. 5D, 6;
Supplemental Table S2). (All four Argonautes are expressed
in the HeLa line we used [Supplemental Fig. S3].) We de-
tected no cooperativity for silencing by the perfect siRNA in
the Ago1?/?MEFs (nH
by the bulged siRNA was cooperative (nH
P = 0.003), suggesting that Ago2 is capable of cooperative
silencing (Fig. 5D).
To test whether Ago3 or Ago4 contributes to the
cooperativity that we observed for a bulged siRNA in the
Ago1?/?MEFs, we used siRNAs to deplete Ago3 and Ago4
mRNAs before transfecting the reporter plasmids and
bulged siRNA. Ago4 mRNA was reduced 50% compared
to Ago1?/?MEF cells transfected with a control siRNA. By
qRT-PCR, we detected Ago3 mRNA two threshold cycles
after detection of Ago1 mRNA in the Ago1?/?MEF cells,
indicating that its expression is probably functionally
inconsequential in our analysis (data not shown). The level
of Ago3 protein in Ago1?/?MEFs was very low, and our
attempts to reduce it further by RNAi were unsuccessful
(Supplemental Fig. S4A). Under these conditions, nHfor
silencing by the bulged siRNA was not significantly different
perfect= 1.1 6 0.1). However, silencing
bulged= 1.7 6 0.2;
FIGURE 5. Silencing in Ago2?/?MEFs or Ago2?/?MEFs reconstituted with mouse Ago2 or catalytically inactive, mutant Ago2D669Aor Ago1?/?
MEFs. (A) In the absence of Ago2, silencing by a perfect site (nH= 1.6 6 0.4; P = 0.03) is equally cooperative as a bulged site (nH= 1.8 6 0.3; P =
0.006). (B) Mouse Ago2 expression restored noncooperative silencing by the perfect siRNA (black; nH= 1.0 6 0.1); silencing directed by a bulged
siRNA became less cooperative (red; nH= 1.5 6 0.2; P = 0.02) than in the absence of Ago2 (red in A; nH= 1.8 6 0.3). (C) Catalytically inactive
mouse Ago2D669Alikewise restored noncooperative silencing by a perfect siRNA (black; nH= 1.1 6 0.1), but silencing by the bulged siRNA (red;
nH= 1.5 6 0.3; P = 0.04), was cooperative. (D) In the absence of Ago1, silencing by the perfect siRNA was not cooperative (black; nH= 1.1 6 0.1),
but silencing by the bulged siRNA was cooperative (red; nH= 1.7 6 0.2; P = 0.003). A one sample, two-tailed Student’s t-test was used to calculate
the P-values at 95% confidence for the Hill coefficients to determine if nHwas significantly different from the null hypothesis: nH= 1 (i.e.,
Broderick et al.
RNA, Vol. 17, No. 10
from the null hypothesis (noncooperative binding), although
the Hill coefficients for the bulged and perfect siRNAs were
significantly different (nH
0.1; P = 0.03) (Supplemental Fig. S4B).
As a final test of the idea that noncooperative silencing
reflects target cleavage, we analyzed silencing directed by a
perfectly pairing siRNA in Ago2?/?MEFs reconstituted
with D669A mutant Ago2. In the cells reconstituted with
catalytically inactive Ago2, the single-site reporter was not
silenced by the perfect siRNA (Supplemental Fig. S5).
Surprisingly, cells reconstituted with catalytically inactive
Ago2 exhibited noncooperative silencing of the six-site
reporter by a perfect siRNA (nH
silencing by a bulged siRNA displayed positive cooperativity
result suggests that target cleavage per se is not required for
noncooperative silencing mediated by Ago2. Rather, both
the identity of the Argonaute protein and the nature of
pairing between the small RNA and its target determine if
RISC bound to multiple sites in the 39 UTR of an mRNA can
collaborate to generate cooperativity in silencing.
bulged= 1.8 6 0.5 versus nH
perfect= 1.0 6
perfect= 1.1 6 0.1), while
bulged= 1.5 6 0.3; P = 0.04) (Fig. 5C). This unexpected
Only Ago2-RISC can repress a reporter
with nonadjacent sites
In HeLa cells, the mRNA with the expanded target sites (Fig.
4A) was less efficiently silenced by a bulged siRNA than an
mRNA in which the three sites were adjacent (Fig. 4B). We
propose that RISCs bound to adjacent sites collaborate to
achieve efficient silencing. Is Ago2 required to silence an
mRNA in which the small RNA binding sites cannot
collaborate? We tested silencing of the expanded three-site
reporter mRNA by the perfect and bulged siRNAs in the
Ago2?/?MEFs. Silencing of the expanded three-site re-
porter was completely dependent on Ago2; little or no si-
lencing was observed in the Ago2?/?MEFs for either type
of small RNA:target pairing (Fig. 7C,D). In contrast, both
perfect and bulged siRNAs cooperatively silenced the re-
porter bearing three adjacent small RNA-binding sites in
the Ago2?/?MEFs (Fig. 7A,B). Notably, in the absence
of Ago2, silencing by the perfect siRNA of the reporter
containing three adjacent sites was highly cooperative
spaced sites, only Ago2 can silence at the intracellular RISC
concentration achieved at the highest amount of siRNA
transfected, likely because, in the absence of cooperativity,
the intracellular concentration of Ago1-, Ago3-, and Ago4-
RISC is less than the KDfor target binding for these
Reconstituting the Ago2?/?MEFs with either wild-type or
catalytically inactive mouse Ago2 rescued silencing of the
expanded three-site reporter (Fig. 7E–J). Silencing showed no
significant cooperativity for the perfect siRNA in MEFs
reconstituted with wild-type (nH
catalytically inactive Ago2 (nH
by the bulged siRNA was cooperative for both the wild-type
and the catalytically inactive Ago2 MEFs (nH
P = 0.03) (Fig. 7H). Intriguingly, in the absence of Ago1,
silencing of the expanded three-site reporter by the bulged
siRNA was highly cooperative (nH
perfect= 2.1 6 0.3; P = 0.007). We conclude that, for widely
perfect= 1.1 6 0.1) or
perfect= 1.2 6 0.2). Silencing
bulged= 1.6 6 0.1; P = 0.02) (Fig. 7F)
bulged= 1.3 6 0.2;
bulged= 2.5 6 0.2; P = 0.006)
In our assays, Ago2 noncooperatively silenced mRNAs
bearing multiple, perfectly complementary small RNA-
binding sites, even when its endoribonuclease activity was
inactivated by mutation. This finding is surprising because
we and others have assumed that endonucleolytic cleavage
by Ago2 explained its lack of cooperativity in silencing
when guided by a perfectly pairing siRNA. Clearly, a more
complex explanation is warranted. We suggest that the
Ago2 conformation associated with perfect small RNA:target
pairing precludes protein:protein interactions, causing both
nearby and adjacent binding sites to act independently.
Alternatively, Ago2 protein, when guided by a small RNA
that pairs extensively with its target mRNA, might be
bound by proteins that prevent its association with factors
Silencing via multiple small RNA-binding sites is likely
always cooperative for Ago1, Ago3, or Ago4, irrespective of
the type of pairing between the small RNA and its target.
We suggest that these noncatalytic Argonaute proteins
adopt a single conformation when bound by different
FIGURE 6. Ago1 and Ago2 protein levels in MEF cells. Ago2 was
detected by Western blotting using a rabbit anti-Ago1 antibody that
recognizes both mouse and human Ago1 and a rabbit anti-Ago2
antibody that recognizes both mouse and human Ago2. Ago protein
levels were normalized to actin, and the level of Ago protein in wild-
type MEFs was set to 1. Data are mean 6 standard deviation for three
trials. Inset shows representative data from a single experiment.
Requirements for cooperativity in RNA silencing
modes to their mRNA targets or that the conformations
produced by both perfect and bulged small RNAs are com-
patible with protein:protein interactions between adjacent
RISC molecules. Intrinsic differences between Ago2 and
Ago1, Ago3, and Ago4 may dictate the combination of Ago
proteins capable of cooperative silencing.
Cooperativity versus statistical effects for closely
apposed target sites
We find that multiple, imperfect binding sites need to be
surprisingly close in order to mediate cooperative silencing.
Our data fundamentally agree with previous reports but
differ quantitatively in the precise inter-site distance that
supports cooperativity (Grimson et al. 2007; Bartel 2004.
We note that the precise inter-site distance that supports
cooperative interactions may reflect the intracellular con-
centrations of Argonaute proteins and associated factors, as
well as the local structure or sequence of the mRNA target.
When testing silencing of a two-site reporter, Grimson et al.
observed that a seed-matched siRNA transfected at 25 nM
cooperatively silenced a reporter bearing two 39 UTR target
sites spaced 8–40 nt apart (counting the number of nucle-
otides between the 39 end of the first site and the 59 end of
the second site); expanding the distance to 56 nt disrupted
cooperative silencing (Grimson et al. 2007). We note that
these authors defined cooperativity as an excess of silencing
when the observed repression for a two-site reporter was
compared to the product of the repression observed for each
site acting alone. Enhanced silencing measured in this way
may correspond to true cooperativity or may simply reflect
the statistical effects6of multiple independent sites. Our Hill
analyses distinguish between these two possibilities and
suggest that RISCs bound to adjacent sites cooperate to con-
fer greater silencing than would be expected from statistical
Supplemental 39 pairing reduces the amount
of small RNA required to repress an mRNA
The amount of siRNA required to repress an mRNA target
is determined by the number of small RNA-binding sites,
the spacing of the sites, and the extent of complementarity
beyond the seed sequence at each site. Compared to an
mRNA in which the siRNA seed sequence alone paired with
the small RNA-binding sites, an mRNA in which seed
pairing was supplemented with additional 39 base pairs
FIGURE 7. In the absence of Ago2, effective silencing requires
adjacent sites. (A,B) Both perfect (nH= 2.1 6 0.3; P = 0.007) and
bulged (nH= 1.5 6 0.3; P = 0.04) adjacent sites were silenced
cooperatively in the absence of Ago2. (C,D) In Ago2?/?MEFs, three
target sites spaced 19 nt apart did not silence the reporter. (E,F)
Expressing mouse Ago2 in the Ago2?/?MEFs allowed three distant
sites to silence the reporter. (G,H) Expressing catalytically inactive,
mutant Ago2D669Aalso allowed three distant sites to silence the
reporter. (I,J) In the Ago1?/?MEFs the three distant sites silenced the
reporter. A one sample, two-tailed Student’s t-test was used to
calculate the P-values at 95% confidence for the Hill coefficients to
determine if nHwas significantly different from the null hypothesis:
nH= 1 (i.e., noncooperative).
6Statistical effects result from the simultaneous occupancy of multiple,
independent binding sites of similar affinities even in the absence of
cooperativity (Cantor and Schimmel 1980). This effect, with increasing site
occupancy, causes a steep threshold response that appears nonadditive. In
contrast, cooperativity results in the concerted loading of sites at a lower
overall concentration of ligand (e.g., miRISC) and a sharp dose-response
to a relatively small increase in ligand concentration.
Broderick et al.
RNA, Vol. 17, No. 10
required slightly less siRNA to achieve comparable re-
pression, particularly for multisite target RNAs (Table 1;
Fig. 2). These data reinforce the view that has emerged
from previous computational analyses of miRNA target
binding: 39 supplemental pairing provides a small but mea-
surable increase in the affinity of a small RNA for its target
(Grimson et al. 2007).
Nonetheless, it is striking how much more siRNA is needed
to regulate a target containing small RNA-binding sites with a
seed-match only or a seed-match plus supplementary pairing,
compared to a target containing sites that fully pair to the
small RNA but for a central bulge. While our current data do
not permit direct estimation of the binding affinity of a small
RNA for a reporter mRNA within a cell, they suggest that one
explanation for the remarkably high intracellular abundance
of some miRNAs is that most miRNAs bind weakly to the
mRNAs they regulate.
MATERIALS AND METHODS
Renilla luciferase vector pRL-TK (Promega) containing target sites
for CXCR4 and a modified linker sequence (Doench et al. 2003) was
mutated from TAG to CTC (lower case letters in the oligonucle-
otides) at nt 387–389 of the Renilla luciferase open reading frame to
generate mismatches with seed positions 5, 6, and 7 of the siRNA
guide strand by PCR-directed mutagenesis using DNA oligonucle-
Mutagenesis was confirmed by sequencing, and then 59 phos-
phorylated oligos containing the target sites and pairing to create
appropriate ends were cloned into the XbaI and ApaI sites of the
mutant pRL-TK. Supplemental Table S3 lists the sequences of the
DNA oligonucleotides used to construct target sites. psiCheck2
(Promega) reporters were constructed by digesting psiCheck2
with NheI and NotI and inserting the 39 UTR target site-con-
taining NheI–NotI fragment from the mutant pRL-TK vectors.
Supplemental Table S3 lists the oligonucleotide sequences used to
generate plasmid reporters. Dual Reporter Luciferase assays were
conducted using Dual Luciferase Assay Reagents (Promega) in
a Veritas Microplate Luminometer (Turner Biosystems) according
to the manufacturer’s directions.
Cell culture and transfection
HeLa CCL2 cell cultures were maintained at 37°C and 5% CO2in
DMEM (Invitrogen) supplemented with 10% heat-inactivated
FBS (Invitrogen) and 50 U/mL penicillin and streptomycin
(Invitrogen). MEF cells were cultured in DMEM (Invitrogen)
supplemented with 15% heat-inactivated FBS (Invitrogen), 50 U/
mL penicillin and streptomycin (Invitrogen), 0.1 mM NEAA
(Invitrogen), 2 mM glutamine (Invitrogen). Cells were seeded at
a density of 0.1 3 106cells per well in 24-well plates in DMEM
(Invitrogen) containing 10% heat-inactivated FBS (Invitrogen).
Twenty-four hours later, cells were washed three times in 500 mL
PBS (Invitrogen), and then 400 mL DMEM with serum was added
to each well. Renilla luciferase plasmid (0.025 mg), firefly luciferase
plasmid pGL3 (0.025 mg), and 20 nM siRNA were mixed with
99 mL DMEM and 1 mL DharmaFECT Duo (Dharmacon) per well.
A control siRNA (CXCR4) was used to equalize the total amount of
siRNA in each transfection. Cells are incubated with 0.5 mL final
volume of DMEM plus serum containing 100 mL of transfection
reagent nucleic acid mixture for 24 h.
Single-stranded guide and passenger siRNA strands (Dharmacon)
(Supplemental Table S4) were annealed by incubating 10 mM of
each strand in 500 mL annealing buffer (100 mM potassium
acetate, 30 mM HEPES-KOH, pH 7.4, 2 mM magnesium acetate)
for 1 min at 90°C, followed by 1 h at 37°C.
Cells were washed once in 500 mL PBS and lysed in 100 mL of
Passive Lysis Buffer (Promega) at room temperature for 20 min in
24-well plates. For each well, 10 mL lysate was read in triplicate
using dual luciferase reagents (Promega) in a Turner Biosystems
luminometer controlled by Veritas software (Turner). Renilla lu-
ciferase activity for each concentration of transfected siRNA was
normalized to the corresponding firefly luciferase activity.
The individual biological replicates for normalized Renilla lucif-
erase activity versus siRNA concentration were fit using Igor Pro
6.10 (Lake Oswego, OR) to the Hill equation to determine IC50
and nH. Fitting was weighted using the standard error of each
mean value. Throughout this study, the standard deviation is
reported for mean IC50and nHvalues. The Hill coefficients from
each replicate were subjected to the Student’s t-test to determine
P-values at 95% confidence using GraphPad Prism (La Jolla, CA).
To test if the individual Hill coefficients from the replicates of each
experiment followed a Gaussian distribution, data were subjected
to the Kolmogorov–Smirnov, D’Agostino-Pearson omnibus, and
Shapiro-Wilk normality tests. By all three tests, all Hill coefficient
data was normally distributed. The P-values at 95% confidence were
calculated using an unpaired, one sample, two-tailed Student’s t-test
(GraphPad Prism) to test whether nHwas significantly different
from the null hypothesis that nH= 1 (i.e., noncooperative). An
unpaired, two-tailed Student’s t-test with Welch’s correction at 95%
confidence, which does not assume equal variances, was used to test
the significance of differences in nHfor a perfect and bulged siRNA.
For non-normally distributed fold-repression data, we used the
nonparametric Wilcoxon Signed-Rank test at 95% confidence to
Forty micrograms cell lysate in cell lysis buffer (50 mM Tris-HCl,
150 mM NaCl, 1% v/v NP-40) containing Complete, mini, EDTA-
Requirements for cooperativity in RNA silencing
free protease inhibitor (Roche) were separated by 4%–20%
HEPES-SDS-PAGE and transferred at 4°C in Tris-bicine buffer
to nitrocellulose membrane overnight at 30 V. Membranes were
blocked in 5% w/v milk-TBST (100 mM Tris Cl pH 7.5, 150 mM
NaCl, 0.1% TWEEN 20) for 1 h and incubated overnight at 4°C
with primary antibody diluted in 3% milk-TBST. Rabbit anti-
human and mouse Ago2 antibody (Cell Signaling Technologies) (Li
et al. 2010) or rabbit anti-human and mouse Ago1 antibody (MBL
International) were diluted 1:1000 and rabbit anti-actin antibody
(Bethyl Laboratories) was diluted 1:5000. After three 5-min washes
in TBST, the membranes were incubated 1 h with secondary goat
anti-rabbit HRP-conjugated antibody (GE Healthcare) diluted
1:10,000. After five 5-min washes in TBST, the membranes were
incubated for 5 min in Super Signal West-Dura Extended Duration
Substrate (Pierce). Chemiluminescent signal was recorded using an
Supplemental material is available for this article.
We thank Do ´nal O’Carroll for his generous gift of MEF cell lines;
Brian Farley and Herve ´ Seitz for help with curve fitting; the
Aronin and Moore laboratories for reagents and equipment; Alicia
Boucher, Karen Logan, Wayne Wilkin, Tiffanie Covello, and Gwen
Farley for extraordinary technical support; and current and past
members of the Zamore lab for advice and critical comments on the
manuscript. This work was supported in part by grants from the
National Institutes of Health (GM62862 and GM65236) to P.D.Z.
Received April 14, 2011; accepted July 14, 2011.
Abrahante JE, Daul AL, Li M, Volk ML, Tennessen JM, Miller EA,
Rougvie AE. 2003. The Caenorhabditis elegans hunchback-like gene
lin-57/hbl-1 controls developmental time and is regulated by
microRNAs. Dev Cell 4: 625–637.
Ameres SL, Martinez J, Schroeder R. 2007. Molecular basis for target
RNA recognition and cleavage by human RISC. Cell 130: 101–112.
Azuma-Mukai A, Oguri H, Mituyama T, Qian ZR, Asai K, Siomi H,
Siomi MC. 2008. Characterization of endogenous human Argo-
nautes and their miRNA partners in RNA silencing. Proc Natl Acad
Sci 105: 7964–7969.
Baek D, Villen J, Shin C, Camargo FD, Gygi SP, Bartel DP. 2008. The
impact of microRNAs on protein output. Nature 455: 64–71.
Bartel DP. 2004. MicroRNAs: Genomics, biogenesis, mechanism, and
function. Cell 116: 281–297.
Bartel DP, Chen CZ. 2004. Micromanagers of gene expression: The
potentially widespread influence of metazoan microRNAs. Nat Rev
Genet 5: 396–400.
Brennecke J, Stark A, Russell RB, Cohen SM. 2005. Principles of
microRNA-target recognition. PLoS Biol 3: e85. doi: 10.1371/
Brown KM, Chu CY, Rana TM. 2005. Target accessibility dictates the
potency of human RISC. Nat Struct Mol Biol 12: 469–470.
Cantor CR, Schimmel PR. 1980. Biophysical chemistry: Part III: The
behavior of biological macromolecules (Their biophysical chemistry:
Pt. 3). W. H. Freeman, New York.
Dahlgren C, Zhang HY, Du Q, Grahn M, Norstedt G, Wahlestedt C,
Liang Z. 2008. Analysis of siRNA specificity on targets with
double-nucleotide mismatches. Nucleic Acids Res 36: e53. doi:
Doench JG, Sharp PA. 2004. Specificity of microRNA target selection
in translational repression. Genes Dev 18: 504–511.
Doench JG, Petersen CP, Sharp PA. 2003. siRNAs can function as
miRNAs. Genes Dev 17: 438–442.
Du T, Zamore PD. 2007. Beginning to understand microRNA
function. Cell Res 17: 661–663.
Du Q, Thonberg H, Wang J, Wahlestedt C, Liang Z. 2005. A
systematic analysis of the silencing effects of an active siRNA at
all single-nucleotide mismatched target sites. Nucleic Acids Res 33:
Elbashir SM, Lendeckel W, Tuschl T. 2001a. RNA interference is
mediated by 21- and 22-nucleotide RNAs. Genes Dev 15: 188–200.
Elbashir SM, Martinez J, Patkaniowska A, Lendeckel W, Tuschl T.
2001b. Functional anatomy of siRNAs for mediating efficient
RNAi in Drosophila melanogaster embryo lysate. EMBO J 20:
Eulalio A, Behm-Ansmant I, Schweizer D, Izaurralde E. 2007. P-body
formation is a consequence, not the cause, of RNA-mediated gene
silencing. Mol Cell Biol 27: 3970–3981.
Friedman RC, Farh KK, Burge CB, Bartel DP. 2009. Most mammalian
mRNAs are conserved targets of microRNAs. Genome Res 19: 92–
Ghildiyal M, Zamore PD. 2009. Small silencing RNAs: An expanding
universe. Nat Rev Genet 10: 94–108.
Grimson A, Farh KK, Johnston WK, Garrett-Engele P, Lim LP, Bartel
DP. 2007. MicroRNA targeting specificity in mammals: Determi-
nants beyond seed pairing. Mol Cell 27: 91–105.
Guo H, Ingolia NT, Weissman JS, Bartel DP. 2010. Mammalian
microRNAs predominantly act to decrease target mRNA levels.
Nature 466: 835–840.
Haley B, Zamore PD. 2004. Kinetic analysis of the RNAi enzyme
complex. Nat Struct Mol Biol 11: 599–606.
Hammond SM, Bernstein E, Beach D, Hannon GJ. 2000. An RNA-
directed nuclease mediates post-transcriptional gene silencing in
Drosophila cells. Nature 404: 293–296.
Hannon GJ. 2002. RNA interference. Nature 418: 244–251.
Haruki M, Tsunaka Y, Morikawa M, Iwai S, Kanaya S. 2000. Catalysis
by Escherichia coli ribonuclease HI is facilitated by a phosphate
group of the substrate. Biochemistry 39: 13939–13944.
Holen T, Amarzguioui M, Wiiger MT, Babaie E, Prydz H. 2002.
Positional effects of short interfering RNAs targeting the human
coagulation trigger Tissue Factor. Nucleic Acids Res 30: 1757–1766.
Huang H, Qiao R, Zhao D, Zhang T, Li Y, Yi F, Lai F, Hong J, Ding X,
Yang Z, et al. 2009. Profiling of mismatch discrimination in RNAi
enabled rational design of allele-specific siRNAs. Nucleic Acids Res
Hutva ´gner G, Zamore PD. 2002. A microRNA in a multiple-turnover
RNAi enzyme complex. Science 297: 2056–2060.
Kanaya S, Oobatake M, Liu Y. 1996. Thermal stability of Escherichia
coli ribonuclease HI and its active site mutants in the presence and
absence of the Mg2+ion. Proposal of a novel catalytic role for
Glu48. J Biol Chem 271: 32729–32736.
Krek A, Grun D, Poy MN, Wolf R, Rosenberg L, Epstein EJ,
MacMenamin P, da Piedade I, Gunsalus KC, Stoffel M, et al.
2005. Combinatorial microRNA target predictions. Nat Genet 37:
Lai EC. 2002. Micro RNAs are complementary to 39 UTR sequence
motifs that mediate negative post-transcriptional regulation. Nat
Genet 30: 363–364.
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, Shih IH, Jones-Rhoades MW, Bartel DP, Burge CB. 2003.
Prediction of mammalian microRNA targets. Cell 115: 787–798.
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.
Broderick et al.
RNA, Vol. 17, No. 10
Li L, Yu C, Gao H, Li Y. 2010. Argonaute proteins: Potential Download full-text
biomarkers for human colon cancer. BMC Cancer 10: 38. doi:
Lim LP, Lau NC, Garrett-Engele P, Grimson A, Schelter JM, Castle
J, Bartel DP, Linsley PS, Johnson JM. 2005. Microarray analysis
shows that some microRNAs down-regulate large numbers of
target mRNAs. Nature 433: 769–773.
Lin SY, Johnson SM, Abraham M, Vella MC, Pasquinelli A, Gamberi
C, Gottlieb E, Slack FJ. 2003. The C. elegans hunchback homolog,
hbl-1, controls temporal patterning and is a probable microRNA
target. Dev Cell 4: 639–650.
Liu J, Carmell MA, Rivas FV, Marsden CG, Thomson JM, Song JJ,
Hammond SM, Joshua-Tor L, Hannon GJ. 2004. Argonaute2 is
the catalytic engine of mammalian RNAi. Science 305: 1437–1441.
Liu J, Rivas FV, Wohlschlegel J, Yates JR, Parker R, Hannon GJ. 2005.
A role for the P-body component GW182 in microRNA function.
Nat Cell Biol 7: 1161–1166.
Martinez J, Tuschl T. 2004. RISC is a 59 phosphomonoester-producing
RNA endonuclease. Genes Dev 18: 975–980.
Matranga C, Zamore PD. 2007. Primer. Small silencing RNAs. Curr
Biol 17: R789–R793.
Meister G, Landthaler M, Patkaniowska A, Dorsett Y, Teng G, Tuschl
T. 2004. Human Argonaute2 mediates RNA cleavage targeted by
miRNAs and siRNAs. Mol Cell 15: 185–197.
O’Carroll D, Mecklenbrauker I, Das PP, Santana A, Koenig U, Enright
AJ, Miska EA, Tarakhovsky A. 2007. A Slicer-independent role for
Argonaute 2 in hematopoiesis and the microRNA pathway. Genes
Dev 21: 1999–2004.
Olsen PH, Ambros V. 1999. The lin-4 regulatory RNA controls
developmental timing in Caenorhabditis elegans by blocking
LIN-14 protein synthesis after the initiation of translation. Dev
Biol 216: 671–680.
Parker R, Sheth U. 2007. P bodies and the control of mRNA
translation and degradation. Mol Cell 25: 635–646.
Rajewsky N, Socci ND. 2004. Computational identification of micro-
RNA targets. Dev Biol 267: 529–535.
Rehwinkel J, Behm-Ansmant I, Gatfield D, Izaurralde E. 2005. A
crucial role for GW182 and the DCP1:DCP2 decapping complex
in miRNA-mediated gene silencing. RNA 11: 1640–1647.
Reinhart BJ, Slack FJ, Basson M, Pasquinelli AE, Bettinger JC, Rougvie
AE, Horvitz HR, Ruvkun G. 2000. The 21-nucleotide let-7 RNA
regulates developmental timing in Caenorhabditis elegans. Nature
Rivas FV, Tolia NH, Song JJ, Aragon JP, Liu J, Hannon GJ, Joshua-
Tor L. 2005. Purified Argonaute2 and an siRNA form recombinant
human RISC. Nat Struct Mol Biol 12: 340–349.
Schwarz DS, Hutva ´gner G, Haley B, Zamore PD. 2002. Evidence that
siRNAs function as guides, not primers, in the Drosophila and
human RNAi pathways. Mol Cell 10: 537–548.
Schwarz DS, Hutva ´gner G, Du T, Xu Z, Aronin N, Zamore PD. 2003.
Asymmetry in the assembly of the RNAi enzyme complex. Cell
Schwarz DS, Tomari Y, Zamore PD. 2004. The RNA-induced
silencing complex is a Mg2+-dependent endonuclease. Curr Biol
Selbach M, Schwanhausser B, Thierfelder N, Fang Z, Khanin R,
Rajewsky N. 2008. Widespread changes in protein synthesis
induced by microRNAs. Nature 455: 58–63.
Song JJ, Smith SK, Hannon GJ, Joshua-Tor L. 2004. Crystal structure
of Argonaute and its implications for RISC slicer activity. Science
Tafer H, Ameres SL, Obernosterer G, Gebeshuber CA, Schroeder
R, Martinez J, Hofacker IL. 2008. The impact of target site ac-
cessibility on the design of effective siRNAs. Nat Biotechnol 26:
Tolia NH, Joshua-Tor L. 2007. Slicer and the Argonautes. Nat Chem
Biol 3: 36–43.
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 39 UTR. Genes Dev 18: 132–137.
Wightman B, Ha I, Ruvkun G. 1993. Post-transcriptional regulation
of the heterochronic gene lin-14 by lin-4 mediates temporal
pattern formation in C. elegans. Cell 75: 855–862.
Yekta S, Shih IH, Bartel DP. 2004. MicroRNA-directed cleavage of
HOXB8 mRNA. Science 304: 594–596.
Zeng Y, Wagner EJ, Cullen BR. 2002. Both natural and designed
micro RNAs can inhibit the expression of cognate mRNAs when
expressed in human cells. Mol Cell 9: 1327–1333.
Requirements for cooperativity in RNA silencing