In situ fluorescence analysis demonstrates active
siRNA exclusion from the nucleus by Exportin 5
Institute for Biophysics, BIOTEC, Dresden University of Technology, Tatzberg 47-51, 01307 Dresden,
Germany and1Cenix BioScience GmbH, Tatzberg 47, 01307 Dresden, Germany
Received December 14, 2005; Revised and Accepted February 9, 2006
Two types of short double-stranded RNA molecules,
namely microRNAs (miRNAs) and short interfering
RNAs (siRNAs), have emerged recently as important
regulators of gene expression. Although these mole-
cules show similar sizes and structural features, the
mechanisms of action underlying their respective
target silencing activities appear to differ: siRNAs
act primarily through mRNA degradation, whereas
most miRNAs appear to act primarily through trans-
lational inhibition. Our understanding of how these
overlapping pathways are differentially regulated
within the cell remains incomplete. In the present
work, quantitative fluorescence microscopy was
used to study how siRNAs are processed within
human cells. We found that siRNAs are excluded
from non-nucleolar areas of the nucleus in an
recognizes key structural features shared by these
and other small RNAs such as miRNAs. We further
established that the Exportin-5-based exclusion of
siRNAs from the nucleus can, when Exp5 itself is
inhibited, become a rate-limiting step for siRNA-
induced silencing activity. Exportin 5 therefore
represents a key point of intersection between the
siRNA and miRNA pathways, and, as such, is of
fundamental importance for the design and inter-
pretation of RNA interference experimentation.
Argonaute-containing complexes are emerging as key regula-
tors of gene expression, both in the context of controlling
developmental programs of gene expression and as a defence
mechanism to protect the genome against viruses and trans-
posons [Reviewed in (1–3)]. These complexes have been
shown to use short double-stranded RNA (dsRNA) molecules
as targeting co-factors, to identify cognate mRNA transcripts
whose expression is to be silenced. The mechanism by which
this silencing occurs is also determined in part by the type of
dsRNA molecule used, and two mechanisms have emerged,
directed by two distinct types of short dsRNA: small inter-
fering RNAs (siRNAs) and micro RNAs (miRNAs).
Documented in organisms from plants to humans, siRNAs
are key intermediates of an evolutionarily conserved, multi-
step pathway for post-transcriptional gene silencing known as
RNA interference (RNAi) (4–6). This process occurs in the
cytoplasm and is guided by the siRNAs which direct the
sequence-specific degradation of targeted mRNAs (7–9).
Naturally occurring siRNAs are produced by Dicer-based pro-
cessing of long dsRNA molecules derived from viral infection
or transposon activity (10–12). These siRNAs consist of a
dsRNA stem of ?19 nt containing 2 nt overhangs with free
hydroxyl groups at both 30ends and phosphates at both 50ends
(6). To act as triggers for mRNA degradation, siRNAs are
assembled with specific proteins including Argonaute 2 to
form the RNA-induced silencing complex (RISC), which cat-
alyzes target mRNA cleavage (7,13–15). The siRNA strand
with the most thermodynamically labile base pairing near
its 5’ end is preferentially used by RISC as the ‘‘guide’’ or
targeting strand, whereas the other ‘‘passenger’’ strand is
degraded (16,17). The target cleavage site in the mRNA is
defined by the siRNA sequence and is located 10 nt upstream
of the nucleotide complementary to the 50-most residue of the
guide strand (18,19).
miRNAs represent a similar yet clearly distinct class of
short RNA species, as these are actually encoded by animal
and viral genomes to direct the post-transcriptional silencing
of target genes, also using RISC-like complexes (20–23). With
hundreds of such molecules having now been identified in
various organisms, miRNAs emerge as 21–25 nt-long
single-stranded hairpin RNAs, derived from larger precursors
that form imperfect stem–loop structures and in most cases,
inhibit translation from target mRNAs in a process that
remains poorly understood (24). Two processing events
lead to mature miRNA formation, as well as to their specific
*To whom correspondence should be addressed. Tel: +49 351 463 40328; Fax: +49 351 463 40342; Email: email@example.com
? The Author 2006. Published by Oxford University Press. All rights reserved.
The online version of this article has been published under an open access model. Users are entitled to use, reproduce, disseminate, or display the open access
version of this article for non-commercial purposes provided that: the original authorship is properly and fully attributed; the Journal and Oxford University Press
only in part or as a derivative work this must be clearly indicated. For commercial re-use, please contact firstname.lastname@example.org
Nucleic Acids Research, 2006, Vol. 34, No. 51369–1380
subcellular localization in animals. The nascent miRNA
transcript (pri-miRNA) is processed by the RNase-III enzyme
Drosha inside the nucleus into an ?70 nt precursor
(pre-miRNA) (25). Pre-miRNAs are then exported from the
nucleus into the cytoplasm by Exportin-5 (Exp5), a Ran-GTP
dependent nucleo–cytoplasmic transporter (26). Once the
pre-miRNA has entered the cytoplasm it is further pro-
cessed by the RNase-III enzyme Dicer to form the mature
Fundamentally, siRNAs and miRNAs share some common
characteristics, e.g. the dsRNA stem, the Dicer dependent
maturation (8,27,28) and the association with Argonaute-
family proteins to direct their gene silencing activity
(14,29–31). It has been shown in Drosophila that miRNAs
and siRNAs are incorporated into different, but similar effec-
tor complexes, where miRNAs associate with AGO1 and
siRNAs with AGO2 as key components of RISC (29,30).
The degree of complementarity between a siRNA or miRNA
and their target mRNA determines, at least in part, the specific
mode of post-transcriptional repression. That is, fully comple-
mentary short RNAs lead to the degradation of the target
mRNA, whereas multiple mismatches in the RNA:mRNA
helix direct translational repression without destabilization of
the mRNA (32).
The detailed analysis of siRNA and miRNA processing is
of fundamental interest to better understand their respective
pathways and to further refine the experimental power of
RNAi techniques. In the present work we employed high-
resolution fluorescence microscopy to quantitatively measure
fluorescence-labelled siRNAs delivered into HeLa SS6 cells
by microinjection. We examined the cellular localization of a
variety of siRNA-like structures, which yielded new insight
into how siRNA molecules are processed in cells. These
observations led to the demonstration that Exp5 is required
for the active exclusion of siRNAs from the nucleus and
thereby directly impacts their biological activity. This study
reveals Exp5-mediated nuclear export as another important
step common to the processing of both naturally encoded
miRNAs and exogenous synthetic siRNAs, and as such, one
that must be taken into careful consideration during RNAi
MATERIALS AND METHODS
Adherent HeLa SS6 cells were cultured at 37?C in DMEM
(Gibco, Invitrogen GmbH) with 10% fetal calf serum (PAA
Laboratories GmbH). Cells were regularly passaged at sub-
confluency and were plated in antibiotic-free medium with
2–3.5 · 104cells/ml density.
For all transfections cells were transfected with Lipo-
fectamine 2000 (Invitrogen GmbH). For the luciferase assays
exponentially growingcells were trypsinized onthe daybefore
transfection and plated into 24-well plates at a density of
5 · 104cells/well in antibiotic-free media. For 6-well plates
and MatTek chambers (MatTek Corporation, USA) cells
were sown one day before transfection at a density of
1.5–3 · 105cells/well in antibiotic-free media. The next
day, the cells were transfected with 100 nM siRNAs with
6 ml Lipofectamine 2000 and 300 ml OptiMEM in 1.5 ml
fresh medium. The medium is replaced by 2 ml fresh growth
medium 3–5 h after transfection.
RNA and DNA oligonucleotides were obtained from IBA
GmbH (Goettingen) and Ambion. For the silencing of
CRM1, Exp5, Pp-luc and the following oligonucleotides
g ¼ guide strand; p ¼ passenger strand
siCRM1: p UGUGGUGAAUUGCUUAUACTT; g GUAU-
siExp5-1: p GCCCUCAAGUUUUGUGAGGTT; g CCU-
siExp5-2: p UGUGAGGAGGCAUGCUUGUTT; g ACA-
siGL2: p CGUACGCGGAAUACUUCGATT; g UCG-
siEGFP: p GCAGCACGACUUCUUCAAGTT; g CUU-
As negative controls a negative siRNA (NegsiRNA) from
Ambion was used and DNAs were obtained from IBA GmbH.
DNA-GL2: p CGTACGCGGAATACTTCGATT; g TCGA-
NegsiRNA: Silencer? Negative Control #1 siRNA.
For the 30–50overhang shift assay
siGL2: g UCG AAG UAU UCC GCG UAC GUG
siGL2: 302ntoh/-30: p CGU ACG CGG AAU ACU UCG
siGL2: blunt/-30: p CAC GUA CGC GGA AUA CUU CGA
siGL2: 502ntOh: p AUC ACG UAC GCG GAA UAC UUC
siGL2: 504ntOh: p ACA UCA CGU ACG CGG AAU ACU
siGL2: 505ntOh: p AAC AUC ACG UAC GCG GAA UAC
siGL2: 506ntOh: p GAA CAU CAC GUA CGC GGA AUA
siGL2: 304ntOh: p GUA AAG CUU CAU AAG GCG CAU
siGL2: 305ntOh: p UGU AAA GCU UCA UAA GGC GCA
siGL2: 306ntOh: p CUG UAA AGC UUC AUA AGG CGC
Duplex siRNAs were prepared by mixing complementary
sense siRNA and antisense siRNA at equimolar ratio,
incubating at 80?C for 2 min followed by a cooling step at
1?C/min to 15?C. The annealing procedure was performed
in the Mastercycler epGradientS (Eppendorf) in 110 mM
K-gluconate; 18 mM NaCl; 10 mM HEPES, pH 7.4 and
0.6 mM MgSO4with 10–60 mM siRNA concentrations. The
quality of the duplex siRNAs was checked by agarose gel
electrophoresis and high-performance liquid chromatography.
The prepared siRNA duplexes were stored at ?20?C.
pEGFP-N1 vector (Clontech) and pDsRed2-N1 vector
Agarose gel electrophoresis
For agarose gel electrophoresis a special high resolving
Metaphor?agarose (Cambrex) was used. All agarose gels
have a 4% (w/v) concentration and were used with 1· TBE
(0.1 M Tris–borate, 2 mM EDTA, pH 8.3) running buffer.
Aliquots of 1 · 10?2nmol labelled siRNA were loaded in
each pocket. The gels were analysed with a Typhoon 9410
Variable Mode Imager (Amersham Biosciences) by using the
1370Nucleic Acids Research, 2006, Vol. 34, No. 5
manufacturer’s settings for Alexa488 (Molecular Probes) and
Cy5 (Amersham Biosciences).
RNA and DNA oligonucleotides were synthesized by
IBA with a 30or 50-amino group via a C6-carbon linker
and were labelled with Alexa Fluor?488 carboxylic acid,
2,3,5,6-tetrafluorophenyl ester (Alexa488-TFP; Molecular
Probes) or Cy5 succinimidyl ester (Cy5-NHS; Amersham
Single-stranded oligonucleotides (0.2 mmol) in 55 ml H2O
were mixed in 20 ml of 500 mM borate buffer, pH 8.5, with 25
ml of 10 mM aminoreactive dye [in N,N-dimethylformamide,
Merck] and incubated at room temperature for 16 h. Labelled
siRNAswere ethanol precipitated[70% (v/v)ethanol,250mM
NaOAc, pH 5.8] four times to remove most of the free unre-
acted dye. The pre-purified siRNAs were loaded on an 18%
denaturating polyacrylamide gel (SequaGel?Sequencing
system; National diagnostics) to separate unlabelled and
labelled strands and unreacted dye. The band with the labelled
RNA is cut out and eluted from the gel slice with 400 ml
of 0.3 M NaCl for 12 h at 4?C. To remove residual gel the
sample was loaded onto a microfilter with 0.2 mm2pore size
Dual-luciferase assays (Promega GmbH) were performed
24–48 h after transfection according to the manufacturer’s
protocol for 24-well chambers and detected with a TD20/20
luminometer (Turner designs). Pp-luc target vector (pGL2-
Control, Promega) was co-transfected with the control vector
Rr-luc (pRL-TK; Promega GmbH).
Lipofectamine 2000 (Invitrogen GmbH) was used for the
triple transfection of the dual luciferase assay vectors pGL2-
control [contains the cDNA of Firefly luciferase (FL)] and
pRL-TK [contains the cDNA of Renilla luciferase (RL)]
together with the siRNAs. The desired amount of each siRNAs
was mixed with 0.9 mg pGL2-control and 0.1 mg pRL-TK. To
circumvent any non-specific effects caused by different
nucleic acid content, all transfections were performed at
equal final nucleic acid concentrations as indicated in the
figures and figure legends. For this adjustment we used a
short double-stranded DNA (dsDNA) (Eg5). The cells were
transfected with the indicated amounts of siRNAs, 100 ml
Opti-MEM and 2 ml Lipofectamine 2000. The medium was
replaced by 500 ml fresh growth medium 3–5 h after trans-
For microinjection, 1.2 · 105HeLa SS6 cells were transfered
onto MatTek chambers coated with poly-D-lysine (0.1 mg/ml;
Sigma) 24 h before microinjection. The micropipette
(Femtotip 2; Eppendorf) was loaded with 20 mM labelled
siRNAs in 110 mM K-gluconate; 18 mM NaCl; 10 mM
HEPES, pH 7.4 and 0.6 mM MgSO4. The micromanipulator
consists of a FemtoJet and InjectMan NI2 which was mounted
directly on an Olympus microscope IX-71. Working pressure
for injection in adherent cells was between 45–90 hPa for 0.1 s
and a holding pressure of 35–40 hPa. For the injection of
plasmids the injection pressure was kept below 90 hPa to
avoid destructive shearing.
Laser scanning confocal fluorescence microscopy
LSM imaging was performed using a commercial Zeiss
ConfoCor 2 laser scanning microscope (Zeiss, Jena, Germany)
with argon ion (488 nm, 25 mW, at 18% of maximum power
output), helium–neon (543, 1 mW, at 30% of maximum power
output) and helium–neon (633 nm, 5.0 mW, at 12% of maxi-
mum power) lasers. A water immersion objective 40·/1.2W
(Zeiss) was used in the epidetection configuration with an
adjustable pinhole set at 71 mm. A band-pass filter trans-
mitting 505–550 nm (Zeiss) was used in the experiments
with Alexa488, while a long-pass filter at 650 nm (Zeiss)
was applied to separate Cy-5 fluorescence. For the experi-
ments with DsRed2 and EGFP a water immersion objective
20·/0.5W Ph2 (Zeiss) was used in the epidetection configura-
tion with an adjustable pinhole set at 73.4 mm for the red
channel and 80 mm for the green channel. A band-pass filter
transmitting 505–530 nm (Zeiss) was used in the experiments
with EGFP, while a long-pass filter at 560 nm (Zeiss) was
applied to separate DsRed2 fluorescence. In order to avoid
saturation of the fluorescence intensity in the scanned images
the detector settings were optimized by using the feature
Range Indicator provided by the Zeiss software (Operating
Manual LSM510; Zeiss, Jena, Germany).
Cell extract and western blot
For western blot analysis the cells were lysed 48 h after siRNA
transfection (100 nM or siRNA amounts are indicated in the
figure) with PLC buffer [1% (v/v) TX100; 10% (v/v) glycerol;
150 mM NaCl; 50 mM HEPES, pH 7.4; 1.5 mM MgCl2; 1 mM
EGTA; 0.2 mM PMSF; 0.1 mg/ml Pepstatin; 0.1 mM Benza-
mide] for 15 min at 4?C. The crude extracts were sonicated
10 times (Bandelin, Sonopuls; duration, 0.1 sec; power, 10%)
and centrifuged 10 min at 14000 r.p.m. (Eppendorf 5417R).
The supernatant was frozen in liquid nitrogen and stored
at ?80?C. The S20 HeLa extract was prepared as described
by Ford and Wilusz (33) except that the final centrifugation
step was performed at 20000 g instead of 100000 g.
For immunoblotting, proteins were run on an 8% PAGE and
subsequently transferred onto Protran?Nitrocellulose Trans-
fer Membrane (Schleicher & Schuell) by semi-dry blotting.
Membranes were blocked for 1 h in phosphate-buffered saline
(PBS) containing 5% milk and 0.05% Tween (PBS-T), incu-
bated for 1 h with a primary antibody, followed by a 1 h
incubation of the secondary antibody linked to horseradish
peroxidase (Novagen). Immunoreactive protein bands were
detected using ECL? western blotting detection reagents
according to the manufacturer’s protocol (Amersham Bio-
sciences) and a documentary station (LAS 3000, Fujifilm).
GAPDH-HRP (Abcam): Clone mAbcam 9484; Dilution
1:40000 (mouse monoclonal)
Alpha-Tubulin (Sigma): Clone DM1A; Dilution 1:100000
(Santa Cruz):H-300; Dilution1:600(rabbit,
Nucleic Acids Research, 2006, Vol. 34, No. 51371
Exportin-5 [kindly provided by U. Kutay (26)]: Dilution
1:1000 (rabbit, polyclonal)
Anti-rabbit-HRP(Dianova): Dilution 1:40000 (goat)
Anti-mouse-HRP (Dianova): Dilution 1:40000 (goat).
Establishment of labelling and delivery conditions for
in situ study of siRNA processing
Single strands of a pre-validated siRNA sequence directed
against Firefly luciferase (siGL2) (5) were synthesized and
subsequently fluorescently labelled on the 30and 50ends
with Alexa488 on the guide strand and Cy5 on the passenger
strand (Figure 1A and B). These single-stranded RNAs
(ssRNAs) were also purified from a polyacrylamide gel and
subsequently annealed to produce the active siGL2 dsRNA
(Materials and Methods and Figure 1B). To examine if the 30
linkeddyes had anyinfluence onsilencing, the double-labelled
siGL2 dsRNA was co-transfected with the reporter plasmid
combination pGL2-Control/ pRL-TK-Control into HeLa SS6
cells (5). Luciferase activities were determined 48 h after
transfection by a dual luciferase assay (Figure 1C) and the
ratios of target to control luciferase were normalized to a
NegsiRNA, which has no significant sequence similarity to
mouse, rat or human gene sequences. For comparison, the
interference ratios of the siGL2 labelled variants were grouped
together for each concentration. Both unlabelled (Figure 1C,
grey bars) and double 30labelled (Figure 1C, open bars)
duplexes showed a similar concentration-dependent silencing
of Firefly luciferase, consistent with previous studies in which
various 30modifications were examined (34,35). Furthermore,
using an siRNA against Renilla luciferase (siTK) and switch-
ing Alexa488 to the passenger strand and Cy5 to the guide
strand also did not change the silencing activity of the siRNA
(data not shown). A 50dual-labelled siGL2 (Figure 1A) was
also tested in this system, since it has been shown that
blocking of the 50end of the guide strand has a strong
Figure 1. Effectof30and 50duallabellingon siRNAactivity.(A) Illustration of differently labelled GL2siRNA(siGL2) withAlexa488andCy5 usedfortargeting
firefly luciferase (pGL2-control). (B) Purification control of the labelled siRNAs (as described in Materials and Methods) on a 4% agarose gel. In the left panel the
The annealeddouble30labelledsiGL2duplex(siGL2-30-Duplex)is shownin lane3. The50labelledvariantsofsiGL2are shownin therightpanel,startingwiththe
guide strand labelled with Alexa488 (siGL2guide50A488), followed by the passenger strand labelled with Cy5 (siGL2passenger50Cy5). The annealed 50double-
labelled siGL2duplex(siGL2-50-Duplex) is shownin lane3. (C) HeLaSS6 cells weretransfected withthe indicatedamountsofdifferentlylabelled siGL2together
with the fixed concentration of the pGL2-Control [Firefly luciferase (FL)] and pRL-TK [Renilla luciferase (RL)] reporter plasmids as described in Materials and
is indicated by the dark grey bar; the 30double-labelled siGL2 is indicated by the open bar; the unlabelled siGL2 is indicated by the grey bar. The plotted data were
averaged from six independent experiments ±SD.
1372 Nucleic Acids Research, 2006, Vol. 34, No. 5
deleterious impact on siRNA silencing activity (7,34). The
50labelled siGL2 showed no significant silencing activity
up to 1 nM and could not silence firefly luciferase beyond
50% even at 50 nM concentrations (Figure 1C, dark grey
bar). These results demonstrate that siRNAs dual labelled
with Alexa488 and Cy5 on the 30termini have no detectible
impact on silencing activity and are therefore a suitable dye
pair for in situ observations. Additionally, it was confirmed
that blocking of the 50end of the guide strand is inhibitory
towards efficient RNAi (7,34,35).
Labelled siRNAs were microinjected directly into the
cytoplasm in order to monitor their subcellular location.
This was done to avoid potential risks of siRNA segregation
within endocytic compartments, as can occur with lipid-based
transfection methods. To confirm that microinjected siRNAs
participate in the RNAi pathway, we developed a dual fluo-
rescent reporter system to quantify target-silencing efficiency.
The reporter plasmids pEGFP-N1 (expressing EGFP protein)
and pDsRed2-N1 (expressing DsRed2 protein) were co-
microinjected into HeLa SS6 cells together with either an
unlabelled siRNA targeting EGFP (siEGFP) or the unlabelled
NegsiRNA with no target (described above). After 48h the
cells were imaged by laser scanning confocal microscopy
(Figure 2A), and all red cells were further analysed. The
ratio of green-expressing cells to red-expressing cells was
calculated (Figure 2B). In three independent injection experi-
ments, fluorescent cells from the negative control (NegsiRNA)
injections exhibited a ratio of ?1.0, indicating that the expres-
sion levels of both DsRed2 and EGFP were not differentially
affected by NegsiRNA (Figure 2A, left panel; Figure 2B,
closed symbols). When the reporter plasmids were injected
together with siEGFP (Figure 2A, right panel; Figure 2B, open
symbols), the ratio of green to red cells decreased with
increasing amounts of siEGFP. This result demonstrated
that siRNAs delivered directly to the cytoplasm via microin-
jection successfully entered the RNAi pathway and mediated
siRNAs are specifically excluded from non-nucleolar
regions of the nucleus
We investigated the impact of several key determinants of
siRNA structure on the subcellular localization of labelled
siRNA-like molecules after microinjection into the cytoplasm
of cultured human cells. We prepared single-labelled siRNAs
(siGL2) as well as labelled controls consisting of ssRNAs
(siGL2-ssRNA), dsDNAs (siGL2-DNA) of the same size
and single-stranded DNAs (siGL2-ssDNA) of the same size
(Materials and Methods). These variants were then micro-
injected at a concentration of 20 mM into the cytoplasm of
HeLa SS6 cells. An intra-needle siRNA concentration of
20 mM was found to give effective silencing activity
(Figure 2B) as well as detectible fluorescence signal intensities
under our recording conditions (Figure 3). The intracellular
concentration of microinjected siRNAs, as measured by fluo-
rescence correlation spectroscopy (FCS), was ?250 nM (data
Within 15 min after microinjection, the labelled siGL2 was
located predominantly throughout the cytoplasm. The siRNA
was excluded from the nucleus, with the exception of faint
fluorescence in the nucleolar region. Although the nucleolar
fluorescene increased over time (open arrow), the siRNA
was excluded from non-nucleolar regions of the nucleus for
>1 h post-injection, irrespective of which siRNA strand was
labelled ((Figure 3A and data not shown). After 1 h, the fluo-
rescence intensity increased slightly in the non-nucleolar
regions of the nucleus, which could be due to degradation
of the labelled siRNAs (Figure 3A). No difference was
observed when labelling with Alexa488 (Figure 3B, left
panel) or Cy5 (Figure 3D, left panel). Similar results were
obtained with other siRNA sequences targeting endogenous
genes (such as the kinesin-related motor protein Eg5), and
with a different cell line (HEK cells; data not shown). In
contrast to the results obtained with siGL2 dsRNA, microin-
jection of labelled siGL2-ssRNA yielded a marked accumu-
lation of fluorescence signal within all areas of the nucleus
regardless of the dye used (Figure 3B and D, right panel) or
the end that was labelled (Figure 3F, left panel). This result
Figure 2. Dual fluorescence-reporter assay for silencing activity analysis of
both pDsRed2-N1 and pEGFP (Ratio 1:1) together with NegsiRNA panel 1 or
siEGFPdirectedagainstEGFP panel2. Recordingofthe confocalpictureswas
after 48 h. Scale bars ¼ 20 mm. (B) Quantitative analysis of silencing after
microinjection with the indicated siRNA concentrations. To normalize the
results the number of red and green cells is divided by the total number of
counted cells (in each experiment >50 cells). NegsiRNA effect on DsRed2 and
Nucleic Acids Research, 2006, Vol. 34, No. 51373
indicated that the strandedness of the RNA was important for
the observed exclusion from the nucleus.
In order to determine whether the type of nucleic acid
backbone is important for the observed localization, we
examined labelled dsDNAs and ssDNAs of equivalent
sizes. As seen in Figure 3 for ssRNA, injection of
siGL2-ssDNA showed fluorescence signals accumulating
throughout the nucleus, whereas the siGL2-dsDNA showed
Figure 3. Subcellular localizations of different types of nucleic acids after microinjection into the cytoplasm. HeLa SS6 cells were cultured in 35 mm dishes with
Localization of the nucleicacids wasmonitoredby confocalmicroscopy.The confocal images representa single plane through the cell and the crosssections of the
stack of the cell are indicated by the lines and illustrated on top and to the right of each image, respectively. (A) Time course of nuclear exclusion of siGL2 labelled
with Alexa488 on the 30end. The time after microinjection is indicated above the images. Nucleoli localization is indicated by an arrow at 45 min. Scale
bars ¼ 20 mm. (B) Subcellular localization of siGL2 labelled on the 30end of the guide strand with Alexa488 in HeLa SS6 cells (left panel) and of the labelled
siGL2-ssRNA (guide strand, right panel). (C) As controls siGL2-dsDNAwere used labelled on the 30end strand with Alexa488 as well as siGL2-ssDNA alone and
controls siGL2-dsDNA was used labelled on the 30end with Cy5 as well as the labelled siGL2-ssDNA and free Cy5 dye. (F) Subcellular localization of differently
single- and double-labelled siGL2-dsRNA with Alexa488 on both 30or 50or either 30and 50ends in HeLa SS6 cells.
1374 Nucleic Acids Research, 2006, Vol. 34, No. 5
a punctate, non-homogeneous distribution in the cytoplasm.
The stability of these various strands in S20 HeLa extracts was
not significantly altered, suggesting that the subcellular
localizations observed after microinjection of labelled
double-stranded nucleic acids were not degradation artefacts
(Supplementary data). In addition, free dye also showed
marked accumulation in the nucleus (Figure 3C and E,
right panel). These results demonstrated that the double-
stranded nature of the siRNA, as well as the ribose backbone
structure, were required for both the observed nuclear exclu-
sion and nucleolar accumulation (Figures 3D, right panel and
Previous RNAi studies showed an influence of the 30and
50overhangs of siRNAs on silencing activity (18,36). We
therefore investigated whether 30or 50end structures were
also necessary for nuclear exclusion. First, to determine
whether free ends are needed, we tested siRNAs dually
labelled on both 30or 50termini, or juxtaposed 30and 50ter-
mini. Dual labelling always resulted in the nuclear exclusion
of the siRNA (Figure 3F, right panels). Next, we examined
the effects of shifting the siRNA’s passenger strand towards
the 30or 50end of the guide strand, thus producing pro-
gressively longer or shorter symmetric 50or 30overhangs,
respectively (Figure 4A). The quality of all annealed
dsRNA preparations was verified by agarose gel electrophor-
esis, which confirmed both the absence of ssRNA contamina-
tion and the expected shifts in duplex mobility corresponding
to the progressive lengthening of the overhanging ends
(Figure 4B). These different siRNA-like variants were then
microinjected into HeLa SS6 cells and analysed after 10–
15 min by confocal microscopy. The resulting ratios of nuclear
to cytoplasmic staining intensities were quantified by taking
mean fluorescence intensity measurements from three 20 mm2
regions within the nucleus and the cytoplasm of injected cells.
In Figure 4C, these nuclear to cytoplasmic fluorescence inten-
sity ratios are shown for each siRNA-like variant, with clear
siGL2 variants were illustrated together with their used abbreviations. For the comparison of the 30and the 50overhangs two labelled strands were used to keep the
The graph shows the ratio of fluorescence intensity nucleus/cytoplasm.
Nucleic Acids Research, 2006, Vol. 34, No. 51375
nuclear exclusion corresponding to values smaller than 0.6.
Such nuclear exclusion was observed with siRNA-like vari-
ants that were either blunt-ended, that contained 50overhangs
of up to 2 nt, or 30overhangs of up to 6 nt (Figure 4C). The
siRNA with a 50overhang of 4 nt (504ntOh) showed a slight
nuclear exclusion initially, but also resulted in marked accu-
mulation in the nucleus (Figure 4C). All siRNA-like variants
containing longer 50overhangs (505ntOh, 506ntOh) immedi-
ately accumulated in the nucleus (Figure 4C), as seen with
ssRNA. The dsRNA variants that were excluded from the
nucleus showed a consistent nuclear to cytoplasmic ratio
of 0.45–0.55 (Figure 4C). Conversely, the 504ntOh duplex
showed a value around 0.9 while the samples containing
505ntOh, 506ntOh and ssRNA showed values around 1.4
(Figure 4C). Nuclear to cytoplasmic ratios were confirmed
by FCS-based concentration measurements of the various con-
structs in the nucleus and cytoplasm (data not shown). These
results indicate that while overhanging ends are not necessary
for nuclear exclusion of siRNA-like molecules, this localiza-
tion pattern is not detectably affected by 30overhangs of up to
6 nt and is strongly inhibited by the presence of 50overhangs
of more than 2 nt.
siRNAs, like pre-miRNAs are exported from
the nucleus by Exportin-5
Our observed nuclear exclusion of siRNAs, as well as its
associated structural determinants, are reminiscent of those
reported previously for certain naturally occurring short
dsRNAs including pre-miRNAs (37,38), which have been
shown previously to be exported from the nucleus by
the Ran-GTP dependent nucleo/cytoplasmic transporter,
Exportin-5 (Exp5) (26,39,40). We therefore tested whether
Exp5-based nuclear export activity was also required for
our observed nuclear exclusion of siRNAs.
To this end, we examined the effects of Exp5 knockdown
on the nuclear exclusion of siRNAs, as well as their silencing
activity. As a control tomonitorthe specificity ofany observed
Exp5 knockdown phenotypes, we also silenced the export
receptor CRM1, which is unable to transport small, structured,
minihelix-containing RNAs (26). Furthermore, we also used
two distinct, previously validated siRNAs targeting Exp5
(siExp5-1 and siExp5-2) and CRM1 (siCRM1) (26), all of
which yielded effective silencing in our experiments, as
confirmed by immunoblotting of the corresponding proteins
(Figure 5A). To quantify and normalize the result, we
developed a two-colour fluorescence assay consisting of 30
Cy5 labelled siGL2 and 30labelled Alexa488 505ntOh
(Figure 4A). These duplexes displayed opposite subcellular
localizations, as the Cy5 labelled siGL2 was excluded
from the nucleus whereas the 505ntOh variant accumulated
throughout the nucleus (Figure 4). Knockdown of CRM1
had no detectable effect on the subcellular localizations of
either siGL2 or 505ntOh (Figure 5B, left panel) compared
with Figure 4C. In contrast, when Exp5 was knocked down
with either siExp5-1 or siExp5-2, siGL2 was found to accu-
mulate together with 505ntOh within all areas of the nucleus,
with perhaps some slight exclusion from nucleoli (Figure 5B,
In order to precisely measure this effect of Exp5
knockdown, we quantified mean fluorescence intensities for
each reporter molecule (siGL2 in red and 505ntOh in green)
from three 20 mm2sampling areas taken in the cytoplasm and
in the nucleus, thus generating normalized, concentration-
independent siGL2 and 505ntOh nuclear to cytoplasmic
ratios. For the CRM1 knock down, the nuclear to cytoplasmic
ratio of the siGL2 was around 0.55, whereas the ratio of
of siGL2 but not of 505ntOh (Figure 5C). After Exp5 knock
down the ratio of siGL2 displayed a ratio of ?3.5, which
means approximately six times higher concentration inside
the nucleus compared with non treated cells or CRM1
(Figures 4 and 5). This high ratio also indicates a 2.5-
fold stronger accumulation of the siGL2 duplexes for the
nucleus in contrast to the 505ntOh ratio of 1.2–1.4 in Exp5
knocked down cells. Additionally, the nuclear/cytoplasmic
ratio of 505ntOh in each knockdown experiment always
resulted in values around 1.2–1.4 (Figure 5C), which is
consistent with the characterization observed in Figure 4.
These results show that Exp5 activity is responsible for
the nuclear exclusion of siRNAs observed in our experiments
and does not detectibly affect the localization of the 505ntOh
Exp5-based nuclear exclusion is a key determinant
of siRNA silencing efficacy
the nucleus represents a contributing factor towards the
efficacy of siRNA-driven target silencing in these cells, we
knocked down Exp5 for 48 hand then measured siRNA silenc-
ing using a dual luciferase assay (Figure 6A). Importantly, for
this portion of the study, all siRNAs were delivered using
standard lipid-based transfection, thus enabling whole popu-
lation analyses and also serving as additional confirmation of
microinjection-based observations under conditions typically
used for RNAi experimentation. To control for the specificity
of any observed Exp5 knockdown phenotypes, siExp5-1 and
siExp5-2 (data not shown), as well as NegsiRNA were all used
at the same concentrations in parallel experiments. The Neg-
siRNA was then also used as a specificity control for siGL2
silencing in the subsequent luciferase assays. As expected,
and siExp5/NegsiRNA) showed similar FL/RL ratios at all
concentrations tested. All positive control samples pre-
treated with the control NegsiRNA and transfected with
siGL2, yielded strong and specific silencing of the targeted
FL, as reflected by a FL/RL ratio of ?25%. Cells pre-treated
with low concentrations of Exp5-targeting siRNAs (1 and
10 nM) (yielding ?98% and ?50% of control Exp5 protein
levels, respectively) gave similar results. However, the knock
down of Exp5 at 100 nM siRNA concentration brought Exp5
protein down to 15% of its control level (that is an 85% knock
down of Exp5) and did exhibit a clearly detectible increase in
FL/RL ratio to 55% (Figure 6B and C). Under these experi-
mental conditions, this result reflects an ?2-fold reduction in
siGL2 silencing efficacy that is directly attributable to Exp5
knockdown. These results also confirm the importance of
Exp5-mediated export for the siRNA silencing effects
shown in Figures 3–5, and is independent of the siRNA deliv-
ery method (i.e. microinjection independent) used in those
1376 Nucleic Acids Research, 2006, Vol. 34, No. 5
With the goal of better understanding how cells differentially
use and process siRNA and miRNA molecules, we first estab-
lished fluorescence labelling conditions that would enable
intracellular detection of biologically active siRNAs in living
cells by confocal microscopy, FCS and LSM. Consistent with
previous reports (7,34,35), we found this to be possible only
with fluorescent labels conjugated to either or both 30ends,
whereas 50labels were found to significantly diminish siRNA
silencing. Standard lipid-based (5) or peptide-based (41) trans-
fection protocols often produce variable results with respect to
probe localization. Therefore, to study the intracellular distri-
bution of labelled siRNAs in living cells, we injected labelled
siRNAs directly into the cytosol. With this approach, the
RNAs successfully entered the RNAi pathway, mediated
sequence-specific target mRNA silencing, and provided con-
sistent and clearly interpretable intracellular fluorescence
patterns in situ.
These patterns revealed a near-total exclusion of labelled
siRNAs from non-nucleolar areas of the nucleus, restricting
them to a homogeneous cytosolic distribution accompanied by
a marked accumulation within nucleoli. Although the latter
observation has been noted in previous studies (41), its physio-
logical relevance remains unclear. The same cytosolic pattern
was observed with all injected siRNAs, irrespective of
whether or not they had an endogenous mRNA target present.
Even at low concentrations, the cytosolic siRNA distribution
did not exhibit any detectible foci such as those noted by
others who observed a co-localization of miRNAs and
Figure 5. Exportin-5isresponsiblefortheexclusionofsiRNAsfromthenucleus.(A)After48htotalproteinextractsfromCRM1,siExp5-1andsiExp5-2(100nM)
transfected cells were immunoblotted with antibodies against Exp5 and CRM1. a-Tubulin was used as a loading control. (B) HeLa SS6 cells were transfected with
labelled with Cy5 and 20 mM 505ntOh (Figure 4A) short dsRNA labelled with Alexa488. The cells were analysed 10–15 min after microinjection by laser scanning
microscopy. Scale bars ¼ 20 mm. (C) Quantification of the translocation process. The confocal images were analysed by using fluorescence area intensity
measurements in the nucleus and in the cytoplasm for the two RNA duplexes. The graph shows the ratio of fluorescence intensity nucleus/cytoplasm.
Nucleic Acids Research, 2006, Vol. 34, No. 51377
imperfectly-targeting siRNA-like molecules with cognate tar-
get mRNAs within so-called processing bodies (42). While
such foci could conceivably have been masked by the rela-
tively high concentrations of labelled siRNAs in our experi-
ments, it also remains possible that siRNAs which mediate
silencing through perfect matching to their target mRNAs do
not accumulate long enough—or at all—at these sites, other-
wise known to be involved in mRNA degradation. This issue
is now being further investigated.
The nuclear exclusion proved to be specific to siRNA-like
molecules, which contrasted with ssRNA and ssDNA mole-
cules of equivalent nucleotide length, or even unconjugated
dyes, all of which accumulated evenly throughout the cytosol
and nucleus. Interestingly, dsDNA strongly accumulated in
punctuate cytoplasmic foci and at the nuclear envelope,
though the relevance of this pattern, if any, remains unclear.
The observed siRNA distribution was thus reminiscent of that
reported by others for certain naturally occurring short dsRNA
molecules, such as adenoviral VA1 RNA and pre-miRNAs
(37,38). During miRNA maturation,pre-miRNAsare exported
from the nucleus by Exp5 (26,39,40). The recognition of pre-
miRNA by Exp5 is dependent upon characteristics of the pre-
miRNA structure (38,43,44) such as a double-stranded stem of
at least 14 nt with either blunt ends, 30overhangs up to 6 nt, or
50overhangs up to 2 nt. Consistent with this, we observed that
our own siRNA-like constructs containing either blunt ends or
30overhangs of up to 6 nt were successfully excluded from the
nucleus, while those with 50overhangs >2 nt were not. This
led us to hypothesize that siRNAs, like pre-miRNAs, are
subjected to an active process of Exp5-based exclusion
from non-nucleolar areas of the nucleus. This was supported
by our finding that the observed nuclear exclusion of siRNAs
is strongly attenuated by silencing of Exp5. Furthermore, our
finding that several siRNA molecules containing completely
distinct sequences were all similarly excluded from the
nucleus indicates that the apparent cytosolic destination signal
is sequence independent. This signal, consistent with the mini-
helix motif serving as an export signal (37), thus appears to be
strictly comprised of those structural characteristics shared by
all siRNAs and pre-miRNAs, and found herein to be recog-
nized for nuclear export by Exp5 (38,43).
As we quantified the effects of Exp5 knockdown on the
siRNA-directed silencing efficiency in cells, we observed
an ?2-fold reduction in silencing of Firefly luciferase upon
knockdown of Exp5, compared with a NegsiRNA control.
This result is contradictory to the findings by Yi et al.
(40,45) who found no effect on silencing after Exp5 knock
down. This is likely due to key differences between experi-
mental conditions used in the two studies. In particular,
Yi et al. (40,45) employed the same siRNA against firefly
luciferase at a concentration of 100 nM, whereas we used
1 nM, thereby making our readout more sensitive in this
case. The level of siGL2 used by Yi et al. (40,45) may
have been high enough to saturate all compartments of the
cell, thereby compensating for the reduced Exp5 activity and
effectively masking its role in RNAi silencing in those
experiments. Indeed, under our own experimental conditions,
the dual luciferase assay showed similar results as those
observed by Yi et al. (40,45), when applied at 100 nM
siGL2 after Exp5 knock down (data not shown). We therefore
conclude that, under conditions of low siRNA concentrations
Figure 6. Exportin-5knockdownleads to reducedsilencingefficiency. (A) An
outline of the experiment is illustrated. First the cells were transfected with
different amounts of siExp5-1 and NegsiRNA and were incubated for 48 h.
siGL2 or NegsiRNA as a control for 24 h. (B) Dual luciferase results were
obtained 24 h after transfection of the pGL2-Control (Firefly luciferase) and
NegsiRNA. Results were grouped together for each siRNA concentration used
for 48 h. The ratios of target to control luciferase were normalized to the
NegsiRNA control transfected for 48 h and for the luciferase assay indicated
in black. The transfection of first the NegsiRNA for 48 h followed by siGL2 is
shown in light grey, whereas the siExp5-1 transfection for 48 h followed by
(C) After 48 h total protein extracts from siExp5-1 transfected cells were
immunoblotted with antibodies against Exp5 and GAPDH-HRP was used
as a loading control. The plotted data were averaged from four independent
1378 Nucleic Acids Research, 2006, Vol. 34, No. 5
often used in standard lipofection-based RNAi experimenta-
tion, the level of Exp5 expression and/or activity in the chosen
cells can represent a rate-limiting factor for target silencing.
Since Exp5 activity also represents a key step in the
processing of endogenous miRNAs, these observations also
warrant the question of whether RNAi experimentation, as
typically carried out in many laboratories today, may in
fact interfere with endogenous miRNA functions. Interest-
ingly, Yi et al. (45) demonstrated that the overexpression
of artificial short hairpin RNAs (shRNAs) can inhibit the
biological activity of endogenous miRNAs, and that this
inhibition can be rescued by Exp5 overexpression. The result-
ing conclusion that endogenous Exp5 activity may represent a
limiting step in the miRNA pathway, especially when the cell
is ‘flooded’ with high concentrations of exogenous shRNA
molecules, is now extended by our present findings linking
Exp5 activity to siRNA silencing. It is therefore likely that
transfected siRNAs can also compete with endogenously tran-
scribed pre-miRNAs for binding to Exp5 and hence affect
miRNA function in cells. While miRNA targeting remains
poorly understood, it is widely believed that most endogenous
miRNAs function as high level ‘managers’ of gene expression
programs, contributing to the maintenance of differentiated
cell states by controlling the levels of large groups of genes
at the same time (2,24). As such, endogenous miRNA deregu-
lation by siRNA ‘flooding’ of Exp5 represents a potential
source for at least some of the complex and unpredictable
off-target effects that have been linked to the use of exces-
sively high concentrations of RNAi silencing reagents (46,47).
It has even been shown that the deregulation of some miRNAs
is associated with certain types of cancers (48), thus further
highlighting the risk that such artefacts could significantly
alter the basic physiology of cells during RNAi experiments.
Importantly, as there is no evidence to date supporting any
dependence of such an Exp5 flooding effect on the specific
sequences of the siRNAs, this type of off-target effect remains
easily controlled for by the use of samples transfected with
equivalent concentrations of ‘scrambled’ or ‘unspecific’ siR-
NAs, such as the NegsiRNA used in the present study.
Because of the fact that siRNAs accumulate ?2.5 times
stronger in the nucleus after Exp5 knockdown compared
with shifted siRNAs, ssRNA, dsDNA, ssDNA and free dye,
this potentially indicates binding to nuclear structures or
complexes of the siRNAs. If siRNAs aberrantly accumulate
in the nucleus, then should we worry about off-target effects
there? Two recent reports have shown specific and potent
RNAi mechanisms in the nucleus (49 ,50). We showed that
siRNAs can shuttle between the cytoplasm and the nucleus,
mediated by diffusion into the nucleus and export out of the
nucleus in a RanGTP/Exp5-dependent fashion. A remnant
nuclear dislocation of a fraction of the siRNA duplexes,
at high concentrations, was confirmed by LSM and FCS.
Furthermore, it has also been shown that siRNAs can induce
transcriptional gene silencing in human cells through a
mechanism that involves DNA methylation [RNA-directed
DNA methylation (RdDM)], an RNAi-related process, that
has been shown to occur in the nucleus and affect gene func-
tion at the level of genomic DNA (51,52). Both the effects of
nuclear RNAi and RdDM were dominant at high concentra-
tions and only if the dsRNAs predominantly entered the
nucleoplasm. This influx of dsRNA molecules into the nucleus
may also result in (non)-specific consequences on gene
expression since nuclear dsRNAs can cause the activation
of non-cytoplasmic RNAi pathways. Previous studies in
Caenorhabditis elegans have shown that the expression of
dsRNA-mediated silencing are mutated (53). This supports
the idea that a nuclear RNAi machinery may be used to regu-
late retrotransposon expression in humans too. These types
of silencing or phenotype effects, if they occur at all, may
or may not be sequence dependent. The use of shRNAs and
stabilized siRNAs should be taken under careful consideration
and must be validated by multiple scrambled sequences.
In addition to these commonly used controls, our findings
further highlight the value of using well-validated RNAi
silencing reagents at their lowest effective concentrations to
minimize the risks of artefacts arising from ‘flooding’ of key
steps common to both the siRNA and miRNA pathways,
as well as cytoplasmic RNAi-related mechanisms. Future
in situ analyses of RNA–protein interactions should help to
further elucidate other common or distinct processes underly-
ing these pathways.
Supplementary Data is available at NAR Online.
We are gratefultoDr UlrikeKutay foranti-Exp5 antibody.We
ments on this manuscript. We also thank members of the
Schwille laboratory, especially Wolfgang Staroske for the
FCS measurements, Drs Nicoletta Kahya and Elke Haustein
for critical discussions as well as Anne Grabner (Cenix
BioScience GmbH) for suggestions and criticisms of this
manuscript. This work was supported by the DFG (SPP
1128). Funding to pay the Open Access publication charges
for this article was provided by the DFG (SPP 1128).
Conflict of interest statement. None declared.
1. Meister,G. and Tuschl,T. (2004) Mechanisms of gene silencing by
double-stranded RNA. Nature, 431, 343–349.
2. Ambros,V. (2004) The functions of animal microRNAs. Nature, 431,
gene regulation. Nature Rev. Genet., 5, 522–531.
4. Fire,A., Xu,S., Montgomery,M.K., Kostas,S.A., Driver,S.E. and
Mello,C.C. (1998) Potent and specific genetic interference by
double-stranded RNA in Caenorhabditis elegans. Nature, 391, 806–811.
5. Elbashir,S.M., Harborth,J., Lendeckel,W., Yalcin,A., Weber,K. and
Tuschl,T. (2001) Duplexes of 21-nucleotide RNAs mediate RNA
interference in cultured mammalian cells. Nature, 411, 494–498.
6. Elbashir,S.M., Lendeckel,W. and Tuschl,T. (2001) RNA interference is
mediated by 21- and 22-nucleotide RNAs. Genes Dev., 15, 188–200.
7. Nyka ¨nen,A., Haley,B. and Zamore,P.D. (2001) ATP requirements and
small interfering RNA structure in the RNA interference pathway.
Cell, 107, 309–321.
for a bidentate ribonuclease in the ignition step of RNA interference.
Nature, 409, 363–366.
Nucleic Acids Research, 2006, Vol. 34, No. 51379
9. Zeng,Y. and Cullen,B.R. (2002) RNA is restricted to the cytoplasm. Download full-text
RNA, 8, 855–860.
10. Billy,E., Brondani,V., Zhang,H., Muller,U. and Filipowicz,W. (2001)
Specific interference with gene expression induced by long,
double-stranded RNA in mouse embryonal teratocarcinoma cell lines.
Proc. Natl Acad. Sci. USA, 98, 14428–14433.
11. Zhang,H., Kolb,F.A., Brondani,V., Billy,E. and Filipowicz,W. (2002)
Human Dicer preferentially cleaves dsRNAs at their termini
without a requirement for ATP. EMBO J., 21, 5875–5885.
12. Provost,P., Dishart,D., Doucet,J., Frendewey,D., Samuelsson,B. and
Ra ˚dmark,O. (2002) Ribonuclease activity and RNA binding of
recombinant human Dicer. EMBO J., 21, 5864–5874.
13. Hammond,S.M., Bernstein,E., Beach,D. and Hannon,G.J. (2000)
An RNA-directed nuclease mediates post-transcriptional gene silencing
in Drosophila cells. Nature, 404, 293–296.
14. Hammond,S.M., Boettcher,S., Caudy,A.A., Kobayashi,R. and
Hannon,G.J. (2001) Argonaute2, a link between genetic and
biochemical analyses of RNAi. Science, 293, 1146–1150.
15. Martinez,J., Patkaniowska,A., Urlaub,H., Lu ¨hrmann,R. and Tuschl,T.
(2002) Single-stranded antisense siRNAs guide target RNA cleavage in
RNAi. Cell, 110, 563–574.
16. Schwarz,D.S., Hutva `gner,G., Du,T., Xu,Z., Aronin,N. and Zamore,P.D.
(2003) Asymmetry in the assembly of the RNAi enzyme complex.
Cell, 115, 199–208.
17. Tomari,Y., Matranga,C., Haley,B., Martinez,N. and Zamore,P.D. (2004)
A protein sensor for siRNA asymmetry. Science, 306, 1377–1380.
18. Elbashir,S.M., Martinez,J., Patkaniowska,A., Lendeckel,W. and
Tuschl,T. (2001) Functional anatomy of siRNAs for mediating
efficient RNAi in Drosophila melanogaster embryo lysate.
EMBO J., 20, 6877–6888.
19. Martinez,J. and Tuschl,T. (2004) RISC is a 50phosphomonoester-
producing RNA endonuclease. Genes Dev., 18, 975–980.
20. Lee,R.C., Feinbaum,R.L. and Ambros,V. (1993) The C.elegans
heterochronic gene lin-4 encodes small RNAs with antisense
complementarity to lin-14. Cell, 75, 843–854.
21. Reinhart,B.J., Slack,F.J., Basson,M., Pasquinelli,A.E., Bettinger,J.C.,
Rougvie,A.E., Horvitz,H.R. and Ruvkun,G. (2000) The 21-nucleotide
let-7 RNA regulates developmental timing in Caenorhabditis elegans.
Nature, 403, 901–906.
22. Ambros,V.,Lee,R.C., Lavanway,A.,Williams,P.T. andJewell,D.(2003)
MicroRNAs and other tiny endogenous RNAs in C.elegans. Curr. Biol.,
Enright,A.J., Marks,D., Sander,C. et al. (2004) Identification of
virus-encoded microRNAs. Science, 304, 734–736.
24. Bartel,D.P. (2004) MicroRNAs: Genomics, biogenesis, mechanism, and
function. Cell, 116, 281–297.
25. Lee,Y., Ahn,C., Han,J., Choi,H., Kim,J., Yim,J., Lee,J., Provost,P.,
Radmark,O., Kim,S. et al. (2003) The nuclear RNase III Drosha initiates
microRNA processing. Nature, 425, 415–419.
26. Lund,E., Guttinger,S., Calado,A., Dahlberg,J.E. and Kutay,U. (2004)
Nuclear export of microRNA precursors. Science, 303, 95–98.
27. Hutvagner,G., McLachlan,J., Pasquinelli,A.E., Balint,E., Tuschl,T. and
Zamore,P.D. (2001) A cellular function for the RNA-interference
enzyme Dicer in the maturation of the let-7 small temporal RNA.
Science, 293, 834–838.
28. Lee,Y., Jeon,K., Lee,J., Kim,S. and Kim,V.N. (2002) MicroRNA
maturation: stepwise processing and sub-cellular localization.
EMBO J., 21, 4663–4670.
29. Caudy,A.A., Myers,M., Hannon,G.J. and Hammond,S.M. (2002)
Fragile X-related protein and VIG associate with the RNA interference
machinery. Genes Dev., 16, 2491–2496.
30. Okamura,K., Ishizuka,A., Siomi,H. and Siomi,M.C. (2004) Distinct
roles for Argonaute proteins in small RNA-directed RNA cleavage
pathways. Genes Dev., 18, 1655–1666.
31. Meister,G., Landthaler,M., Patkaniowska,A., Dorsett,Y., Teng,G. and
Tuschl,T. (2004) Human Argonaute2 mediates RNA cleavage
targeted by miRNAs and siRNAs. Mol. Cell, 15, 185–197.
32. Doench,J.G., Petersen,C.P. and Sharp,P.A. (2003) siRNAs can function
as miRNAs. Genes Dev., 17, 438–442.
extracts that reproduces regulated mRNA stability. Methods, 17, 21–27.
34. Chiu,Y.L. and Rana,T.M. (2002) RNAi in human cells: Basic structural
35. Harborth,J., Elbashir,S.M., Vandenburgh,K., Manninga,H.,
Scaringe,S.A., Weber,K. and Tuschl,T. (2003) Sequence, chemical,
and the effect on mammalian gene silencing. Antisense Nucleic Acid
Drug Dev., 13, 83–105.
Giese,K. and Kaufmann,J. (2003) Structural variations and stabilising
modifications of synthetic siRNAs in mammalian cells. Nucleic Acids
Res., 31, 2705–2716.
37. Gwizdek,C., Bertrand,E., Dargemont,C., Lefebvre,J.C., Blanchard,J.M.,
that directs polymerase III transcripts to the cell cytoplasm. Terminal
minihelix and RNA export. J. Biol. Chem., 276, 25910–25918.
38. Zeng,Y. and Cullen,B.R. (2004) Structural requirements for
pre-microRNA binding and nuclear export by Exportin 5.
Nucleic Acids Res., 32, 4776–4785.
39. Brownawell,A.M. and Macara,I.G. (2002) Exportin-5, a novel
karyopherin, mediates nuclear export of double-stranded RNA binding
proteins. J. Cell Biol., 156, 53–64.
40. Yi,R., Qin,Y., Macara,I.G. and Cullen,B.R. (2003) Exportin-5 mediates
the nuclear export of pre-microRNAs and short hairpin RNAs.
Genes Dev., 17, 3011–3016.
correlation between siRNA localization, cellular uptake, and RNAi in
living cells. Chem. Biol., 11, 1165–1175.
42. Pillai,R.S., Bhattacharyya,S.N., Artus,C.G., Zoller,T., Cougot,N.,
Basyuk,E., Bertrand,E. and Filipowicz,W. (2005) Inhibition of
translational initiation by Let-7 MicroRNA in human cells. Science,
43. Gwizdek,C., Ossareh-Nazari,B., Brownawell,A.M., Doglio,A.,
Bertrand,E., Macara,I.G. and Dargemont,C. (2003) Exportin-5 mediates
nuclear export of minihelix-containing RNAs. J. Biol. Chem.,
44. Gwizdek,C., Ossareh-Nazari,B., Brownawell,A.M., Evers,S.,
Macara,I.G. and Dargemont,C. (2004) Minihelix-containing RNAs
mediate exportin-5-dependent nuclear export of the double-stranded
RNA-binding protein ILF3. J. Biol. Chem., 279, 884–891.
45. Yi,R., Doehle,B.P., Qin,Y., Macara,I.G. and Cullen,B.R. (2005)
Overexpression of Exportin 5 enhances RNA interference mediated by
short hairpin RNAs and microRNAs. RNA, 11, 220–226.
46. Semizarov,D., Frost,L., Sarthy,A., Kroeger,P., Halbert,D.N. and
Fesik,S.W. (2003) Specificity of short interfering RNA determined
through gene expression signatures. Proc. Natl Acad. Sci. USA,
47. Jackson,A.L., Bartz,S.R., Schelter,J., Kobayashi,S.V., Burchard,J.,
Mao,M., Li,B., Cavet,G. and Linsley,P.S. (2003) Expression profiling
reveals off-target gene regulation by RNAi. Nat. Biotechnol.,
48. Gregory,R.I. and Shiekhattar,R. (2005) MicroRNA biogenesis and
cancer. Cancer Res., 65, 3509–3512.
49. Robb,G.B., Brown,K.M., Khurana,J. and Rana,T.M. (2005) Specific and
potent RNAi in the nucleus of human cells. Nature Struct. Mol. Biol.,
50. Langlois,M.A., Boniface,C., Wang,G., Alluin,J., Salvaterra,P.M.,
Puymirat,J., Rossi,J.J. and Lee,N.S. (2005) Cytoplasmic and nuclear
retained DMPK mRNAs are targets for RNA interference in Myotonic
Dystrophy cells. J. Biol. Chem., 280, 16949–16954.
interfering RNA-induced transcriptional gene silencing in human cells.
Science, 305, 1289–1292.
52. Matzke,M.A. and Birchler,J.A. (2005) RNAi-mediated pathways in the
nucleus. Nature Rev. Genet., 6, 24–35.
53. Ketting,R.F., Haverkamp,T.H., van Luenen,H.G. and Plasterk,R.H.
(1999) Mut-7 of C. elegans, required for transposon silencing and RNA
interference, is a homolog of Werner syndrome helicase and RNaseD.
Cell, 99, 133–141.
1380Nucleic Acids Research, 2006, Vol. 34, No. 5