Content uploaded by Reuven Agami
Author content
All content in this area was uploaded by Reuven Agami
Content may be subject to copyright.
LETTERS
A Pumilio-induced RNA structure switch in p27-
3 UTR controls miR-221 and miR-222 accessibility
Martijn Kedde1,4, Marieke van Kouwenhove1,4, Wilbert Zwart2, Joachim A. F. Oude Vrielink1, Ran Elkon1
and Reuven Agami1,3,5
Key regulators of 3ʹ untranslated regions (3ʹ UTRs) are
microRNAs and RNA-binding proteins (RBPs)1,2. The p27
tumour suppressor is highly expressed in quiescent cells,
and its downregulation is required for cell cycle entry after
growth factor stimulation3,4. Intriguingly, p27 accumulates in
quiescent cells despite high levels of its inhibitors miR-221
and miR-222 (refs 5, 6). Here we show that miR-221 and
miR-222 are underactive towards p27-3ʹ UTR in quiescent
cells, as a result of target site hindrance. Pumilio-1 (PUM1)
is a ubiquitously expressed RBP that was shown to interact
with p27-3ʹ UTR7,8. In response to growth factor stimulation,
PUM1 is upregulated and phosphorylated for optimal induction
of its RNA-binding activity towards the p27-3ʹ UTR. PUM1
binding induces a local change in RNA structure that favours
association with miR-221 and miR-222, efficient suppression
of p27 expression, and rapid entry to the cell cycle. We have
therefore uncovered a novel RBP-induced structural switch
modulating microRNA-mediated gene expression regulation.
MicroRNAs (miRNAs) are genes involved in normal development and
in cancer, mainly by associating with 3 untranslated regions (3 UTRs)
of messenger RNAs, regulating their expression9,10. In a similar manner
to miRNAs, RBPs can interact with 3 UTRs in a sequence-specific man-
ner and can both stimulate and inhibit gene expression1,2. In particular, a
member of the Caenorhabditis elegans Pumilio family (Puf-9) is required
for 3 UTR-mediated regulation of the let-7 target hbl-1 (ref. 11). By asso-
ciation with hundreds of mRNAs, many coding for cell cycle regulators,
Pumilio RBPs potentially influence expression by an as yet unknown
mechanism7,8. High levels of miR-221 and miR-222 are required in many
different cancer types to inhibit the expression of p27 (CDKN1B; cyclin
dependent kinase inhibitor 1b) and stimulate proliferation5,6. p27 is a
cyclin-dependent kinase inhibitor that negatively regulates cell cycle
progression by association with cyclin-dependent kinase 2 (CDK2) and
cyclin E complexes, resulting in the inhibition of the transition from
G1 to S phase4. Accumulation of p27 protein is required for entry into
quiescence (G0), and, on stimulation with growth factor, p27 levels must
decrease to allow proper S-phase entry3,4.
We asked whether the miR-221/miR-222 cluster is involved in p27
regulation in quiescence, because it is a negative regulator of p27 transla-
tion in many cancer cell types. We therefore examined p27 and miR-221/
miR-222 levels in both quiescent and cycling BJ primary fibroblasts by
RNase protection assays (RPAs), quantitative RT–PCR (qRT–PCR), and
northern blot and expression array analyses. Although p27 protein level
was clearly elevated in quiescent cells, p27 mRNA and miR-221/miR-
222 levels remained constant (Fig. 1a, b; Supplementary Information,
Fig. S1a–d). We next inhibited miR-221 and miR-222 function by using
miR-221 and miR-222 antagomirs (validated in ref. 5). Addition of miR-
221 and miR-222 antagomirs, but not a control antagomir, to cycling
BJ cells resulted in an increase in p27 levels (Fig. 1c). In contrast, addi-
tion of miR-221 and miR-222 antagomirs to quiescent BJ cells did not
affect the level of p27 protein, suggesting that in quiescent cells miR-221
and miR-222 is less functional in suppressing its target, p27. Effective
uptake of antagomirs in quiescent cells was demonstrated by a control
directed against p53 short hairpin RNAs (shRNAs), in BJ-p53kd cells
(Supplementary Information, Fig. S1e)12. Indeed, despite similar p27
mRNA levels (Supplementary Information, Fig. S1b–d), p27 translation
is increased in quiescent cells (Fig. 1d), indicating that the production
of p27 protein is not inhibited in quiescent cells despite the presence of
its miRNA inhibitor.
The activity of miRNAs can be dependent on accessibility to their target
mRNAs13. To examine the association of miR-221 and miR-222 with p27-
3 UTR in quiescent and cycling cells, we immunoprecipitated endog-
enous Argonaute 2 (AGO2, the main component of the RNA-induced
silencing complex (RISC), directing miRNA target inhibition14) and
measured the relative amounts of associated p27 mRNA and miR-221/
miR-222. As controls we used anti-CDK4 antibody for immunoprecipi-
tation, and both glyceraldehyde-3-phosphate dehydrogenase (GAPDH)
and 18S ribosomal RNA for qRT–PCR. Although similar amounts of
1Division of Gene Regulation, The Netherlands Cancer Institute, Plesmanlaan 121, 1066 CX, Amsterdam, The Netherlands. 2Division of Cell Biology II, The
Netherlands Cancer Institute, Plesmanlaan 121, 1066 CX, Amsterdam, The Netherlands. 3Centre for Biomedical Genetics UMCU, Stratenum 3.223, Universiteitsweg
100, 3584 CG Utrecht, The Netherlands.
4These authors contributed equally to this work
5Correspondence should be addressed to R.A. (e-mail: r.agami@nki.nl)
Received 01 July 2010; accepted 12 August 2010; published online 05 September 2010; DOI:10.1038/ncb2105
nature cell biology advance online publication 1
© 2010 Macmillan Publishers Limited. All rights reserved.
LETTERS
AGO2 were expressed and immunoprecipitated (Fig. 1e), less p27 mRNA
was associated with AGO2–miR-221/miR-222 in quiescent cells than in
cycling cells (Fig. 1e; Supplementary Information, Fig. S2a, b). Previous
formaldehyde crosslinking yielded similar results (Supplementary
Information, Fig. S2c). Analysis of a control miRNA (miR-29a) and its
target mRNA (collagen 3A1)15 revealed opposite association ratios, indi-
cating the specificity of this assay (Supplementary Information, Fig. S2d).
These data suggest that p27 mRNA in cycling cells is more accessible for
interaction with miR-221 and miR-222.
Recently, screens for mRNA targets of the RBP Pumilio revealed,
among many genes, p27 (refs 7, 8). The p27-3 UTR harbours two
evolutionarily conserved Pumilio recognition elements (PREs); one
is located close to the miR-221 and miR-222 target sites (Fig. 2a). The
human Pumilio family contains two members, PUM1 and PUM2. We
knocked down PUM1, the most abundant Pumilio family member, in
miR-221/miR-222-expressing HEK293 cells13. Immunoblot analysis
revealed elevated p27 protein levels in cells transfected with either of
two functional PUM1 knockdown constructs (Fig. 2b). These data sug-
gest that PUM1 inhibits p27 expression in HEK293 cells.
Because both PUM1 and miR-221/miR-222 inhibit p27 expression,
we measured the effect of PUM1 on miR-221-induced repression of a
luciferase reporter gene coupled to the 3 UTR of p27 in MCF7 cells,
which endogenously express PUM1 but not miR-221 and miR-222
(ref. 13). As expected, co-transfection of miR-221 resulted in decreased
luciferase activity (Fig. 2c). On knockdown of PUM1 (Supplementary
Information, Fig. S3a), miR-221 function was compromised. No effect
of PUM1 knockdown on the reporter was seen in the absence of miR-
221 (Fig. 2c). In addition, inactivating mutations in the miR-221 and
miR-222 target sites also resulted in a loss of the PUM1 knockdown
effect (Supplementary Information, Fig. S3b). Moreover, mutating the
PREs in p27-3 UTR also compromised the PUM1 knockdown effect,
whereas miR-221 function remained intact (Fig. 2d). The fact that
PUM1 knockdown abolished miR-221 function, but loss of its binding
sites on the p27-3 UTR did not, suggests that PUM1-induced changes
a
Relative p27 mRNA/miR-221
miR-221
miR-125b
Probe
p27
Tubulin
Western
RPA
p27 mRNA
Cyclophilin
b
e
c
CCQQ
CQ
miR-221 miR-222
Relative miRNA level
d
AGO2-miRNA IP:
Antagomir:
Tubulin
p27
221/222
Control
221/222
Control
Cycling Quiescent
CDK4
0 0.5 12 400.5 124
Quiescent Cycling
MG-132 (h):
0.9 1.0 0.90.91.0p27 (fold): 1.1 1.2 1.51.61.0
Total lysates
IP
QC C
QCC
AGO2 CDK4
Tubulin
AGO2
0
0.2
0.4
0.6
0.8
1.0
1.2
0
0.2
0.4
0.6
0.8
1.0
1.2
p27
Probe
Yeast
Cycling
Quiescent
Figure 1 miR-221 and miR-222 are underactive towards p27 in quiescent
cells. (a) RNA was extracted from quiescent and cycling BJ primary
fibroblasts and was subjected to RPA analysis for p27, miR-221 and
miR-125b, with cyclophilin as control. Immunoblots were performed
with antibody against p27, with anti-tubulin as control. Bands were
spliced together from different parts of the same blot as indicated by
the line. (b) The amounts of miR-221 and miR-222 were measured by
qRT–PCR in cycling (C) and quiescent (Q) BJ cells. Error bars represent
s.d. for triplicate reactions. (c) Quiescent and cycling BJs were treated
with cholesterol-conjugated control or miR-221/miR-222 antagomirs.
Immunoblot analysis was performed as in a. (d) Quiescent and cycling
BJs were treated with the proteasome inhibitor MG-132. Time points are
indicated; a densitometric analysis is shown below. Immunoblot analysis
was performed with antibodies against p27 and CDK4. (e) BJ cell extracts
were used for immunoprecipitation analysis with antibodies against AGO2
and CDK4. Immunoblots were performed with antibody against AGO2, with
anti-tubulin as control. The amounts of miR-221 and p27 mRNA were
measured by qRT–PCR in the immunoprecipitates. Results are presented as
relative p27/miR-221 ratio. The ratio in the immunoprecipitates (IP) from
cycling BJs was set to 1. Enrichment factors of miR-221 and p27 mRNA
in AGO2 immunoprecipitates over CDK4 immunoprecipitates are shown
in Supplementary Information, Fig. S2b. Error bars represent s.e.m. for
triplicate reactions. Uncropped images of blots are shown in Supplementary
Information, Fig. S10.
2 nature cell biology advance online publication
© 2010 Macmillan Publishers Limited. All rights reserved.
LETTERS
in mRNA structure are involved in regulating miR-221 function (see
below). Taken together, these results indicate, first, that both PUM1 and
miR-221 inhibit p27 expression post-transcriptionally through sites in
p27-3 UTR, and second, that efficient suppression of p27 expression by
miR-221 requires Pumilio.
We next examined whether PUM1 RNA-binding activity is altered
between quiescent and cycling cells. To test this we developed an assay to
measure the RNA-binding capacity of PUM1 in vivo with a Cy3-tagged
RNA oligonucleotide corresponding to the 5 PRE in the p27-3 UTR
(Cy3-RNA) and a green fluorescent protein (GFP)-tagged PUM1 (Fig. 3a;
Supplementary Information, Fig. S4a, b). These were microinjected into
both quiescent and cycling BJ fibroblasts and revealed by fluorescent con-
focal microscopy. GFP–PUM1 showed a granular localization pattern,
as reported previously8. Similar localization was observed with endog-
enous PUM1 (Supplementary Information, Fig. S4c). These granules are
juxtaposed to P-bodies, which contain miRNAs and repressed mRNAs
and are thought to be sites of translational repression8,16,17. We found
PUM1 and AGO2 to co-localize to granules in both quiescent and cycling
cells, although no direct interaction between the two could be shown
(Supplementary Information, Fig. S4c, and data not shown). As a result
of non-specific adhesion of Cy3-RNA oligonucleotides to chromatin, a
partial nuclear localization was observed. GFP–PUM1 and Cy3-tagged
wild-type RNA (Cy3-wt-RNA) showed strong co-localization in the
cytosol of cycling cells (Fig. 3a, middle panel; Supplementary Information,
Fig. S4b). When GFP–PUM1 and Cy3-wt-RNA were injected together
into quiescent BJ cells (Fig. 3a, top panel), or when an RNA oligonucle-
otide with two nucleotide alterations in the PRE was injected with GFP–
PUM1 into cycling cells, no co-localization was observed (Fig. 3a, bottom
panel). PUM1 co-localization with Cy3-wt-RNA was specific, because it
was not observed with several other RBPs (Supplementary Information,
Fig. S4d, e). Direct and specific binding of PUM1 to wild-type, but not
mutant, RNA is shown in immunoprecipitation-binding assays with radio-
actively labelled probes and PUM1–TAP (tandem affinity purification;
Fig. 3b). These results indicate that the RNA binding of PUM1 is specific
and its RNA-binding capacity, at least towards the p27-3 UTR, is low in
quiescent cells and high in cycling cells.
We also confirmed this conclusion by immunoprecipitations cou-
pled to RPA of endogenous PUM1 in quiescent and cycling BJ cells.
Immunoprecipitation with anti-PUM1 antibody, but not an anti-CDK6
control, from cycling BJ cells confirmed binding of PUM1 to p27
mRNA (Fig. 3c; quantification is shown in Supplementary Information,
Fig. S4f). The cyclophilin RNA-negative control was not enriched in
ab
cd
p27-3ʹ UTR
miR-221/222 Pumilio miR-221/222
13400
303196
miR-221/222 target site Putative pumilio-binding sitemiR-221/222
-An
Control
KD 1
KD 3
Control
KD 1
KD 3
Control
KD 1
KD 3
Relative PUM1 mRNA levels
0
0.2
0.4
0.6
0.8
1.0
Tubulin
p27
Control
PUM1 KD
13
Control Control
Relative luciferase activity
p27-wt-3ʹ UTR
-An
Luciferase
miR-221 miR-221 Control miR-221 Control ControlmiR-221 miR-221 Control miR-221
0
20
40
60
80
100
120
140
Relative luciferase activity
p27 PUM DM 3ʹ UTR
-An
Luciferase
0
20
40
60
80
100
120
140
XX
Figure 2 Pumilio is required for miR-221 and miR-222 function. (a)
Conservation analysis of p27-3 UTR from human to fish25. The positions
of the binding sites for miR-221 and miR-222 and the PREs (consensus
sequence 5-UGUANAUA-3) are marked. (b) HEK293 cells were transfected
with shRNA constructs targeting PUM1 and control. Cells were subjected to
quantitative RT–PCR for PUM1 and actin control. Error bars represent s.d.
for triplicate reactions. Right: immunoblot analysis as in Fig. 1a. KD, knock-
down. (c) MCF7 cells were co-transfected with expression vectors coding for
luciferase coupled to the wild-type p27-3 UTR, miR-221 and hTR control,
and shRNA vectors against PUM1 or control. Relative luciferase activity is
the ratio between firefly luciferase and Renilla control, adjusted to 100%.
A schematic representation of the p27-3 UTR is shown below. Error bars
represent s.d. for triplicate experiments. (d) Luciferase assay performed as
in c, with a luciferase construct coupled to the p27-3 UTR mutated for the
PREs. Error bars represent s.d. for triplicate experiments. Uncropped images
of blots are shown in Supplementary Information, Fig. S10.
nature cell biology advance online publication 3
© 2010 Macmillan Publishers Limited. All rights reserved.
LETTERS
the immunoprecipitations, indicating the specificity of PUM1 bind-
ing. In contrast, in quiescent BJ cells no p27 mRNA was detected in
PUM1 immunoprecipitation. Immunoblot analysis revealed higher
levels of endogenous PUM1 in cycling BJ cells than in quiescent cells
(Fig. 3c). This effect was observed with both endogenous and stable
exogenous tagged PUM1 (Fig. 3d), indicating post-translational modi-
fications. Taken together, our observations show that, on cell cycle entry
from quiescence, PUM1 levels increase and its RNA-binding activity
is turned on.
A study of phosphorylated proteins in HeLa cells reported unchanged
phosphorylation of PUM1 Ser 209 on stimulation with epidermal
growth factor, whereas Ser 714 phosphorylation was rapidly increased
up to about fivefold (Fig. 3e)18,19. To examine whether these phospho-
rylation sites affect the RNA-binding activity of PUM1 in cycling cells,
GFP–PUM1 phospho-mutants (Supplementary Information, Fig. S4a)
were microinjected together with the Cy3-RNA oligonucleotides into
cycling BJ cells. Mutation of Ser 714 to alanine (S714A) decreased the
RNA-binding activity of PUM1 for Cy3-wt-RNA in cycling cells (Fig. 3f,
lower panel), whereas the S209A mutant was as active as wild-type
PUM1 (Fig. 3f, upper panel). Furthermore, a phospho-mimic mutation
of Ser 714 to glutamic acid (S714E) showed persistent RNA-binding activ-
ity in quiescent cells (Fig. 3g). In contrast, the RNA-binding domain of
PUM1 (GFP–PUM1(HD); HD: homology domain) is mostly nuclear,
and the cytoplasmic fraction does not co-localize to Cy3-wt-RNA
(Supplementary Information, Fig. S4g). This suggests that residues out-
side the HD domain are essential for cytoplasmic localization and binding
specificity. Although the results above do not exclude the involvement of
other modification events in the activation process of PUM1, they suggest
that both PUM1 upregulation and phosphorylation of Ser 714 in response
to stimulation with growth factor are necessary and sufficient to increase
the RNA-binding activity of PUM1 in BJ fibroblasts.
Next we examined the effect of Pumilio on endogenous p27 expression
and cell cycle re-entry from quiescence. PUM1 knockdown resulted in
a delayed re-entry into the cell cycle from quiescence (Supplementary
Information, Fig. S5a) despite modest differences in p27 levels (see below);
this can be explained by haploinsufficiency of p27 (ref. 20). We noticed that
the levels of PUM2, a homologue of PUM1 that is also expressed in BJ and
HEK293 cells, increased when PUM1 was suppressed by RNA-mediated
abc
f
g
d
Quiescent
S714E
0.82
0.55
0.19
Rr:
TAP-IP
PUM1
Tubulin
32P-RNA
Probe: WT mt WT mt
Western
e
Rr: RNA Protein
WT
Cy3
WT
Cy3
Cy3
QuiescentCycling
Cycling
GFP–PUM1 RNA-Cy3 Merge
PUM1–GFPPUM1–GFP
PUM1–GFP
RNA-Cy3
RNA-Cy3
RNA-Cy3
0.20
0.31
0.62
Total lysates
IP
Cyclophilin
p27
PUM1
*
RPA
Western
Probe
pf
QC CQCC
-180
Mr(K)
-115
PUM1–TAP
Endogenous PUM1
QC
Tubulin
Time (min):
Summed:
Prole: Increasing
0151020
0.33 0.44 1 1.36 1.45
S714
Time (min):
Summed:
Prole: Not changing
0151020
0.88 0.79 1 1.02 0.92
S209
GFP–PUM1 RNA-Cy3 Merge
PUM1–GFP
PUM1–GFP
PUM1–GFP
Total
TAP-IP
PUM1
PUM1
CDK6
PUM1
PUM1
CDK6
Cycling
S209A
Cycling
S714A
RNA-Cy3
RNA-Cy3
RNA-Cy3
PUM1
GFP
Figure 3 PUM1 RNA-binding activity is enhanced in cycling versus quiescent
cells. (a) Quiescent or cycling BJ primary fibroblasts were microinjected with
GFP–PUM1 constructs and a Cy3-labelled RNA. After incubation overnight,
cells were fixed and revealed by confocal laser scanning microscopy
(CLSM). Co-localization is shown in the merge panel, and the correlations
between GFP and Cy3 signals within the same cell were ascertained with a
scatter plot. Representative pictures are shown. Insets: enlargements of the
highlighted areas. Pearson’s correlation coefficients (Rr) are shown in the
scatter plots. wt, wild-type; mut, mutant. Scale bar, 10 mm. (b) Binding assay
of immunoprecipitated PUM1–TAP from HEK293 cells and 32P-labelled
wild-type (wt) and mutant (mt) p27 RNA. Gel image displays both unbound
and bound probes. Total lysate and immunoprecipitate (TAP-IP) were
analysed by immunoblotting with antibody against PUM1, with anti-
tubulin as control. (c) RPA for p27 mRNA and cyclophilin negative control
was performed on immunoprecipitates of PUM1 complexes and of CDK6
control from cycling (C) and quiescent (Q) BJ cells. pf, protected fragment.
Quantification is shown in Supplementary Information, Fig. S4f. Total
lysates and immunoprecipitates (IP) were analysed by immunoblotting with
antibody against PUM1. Asterisk, loading control. (d) Immunoblot analysis
of endogenous and overexpressed PUM1 on growth factor stimulation with
antibody against PUM1, with anti-tubulin as control. (e) Data adapted from
the PHOSIDA phosphorylation site database showing the phosphorylation
of PUM1 at Ser 209 and Ser 714 along the course of stimulation with
epidermal growth factor (in minutes). (f) Cycling BJ cells were microinjected
with inactivated phospho-mutants of GFP–PUM1 (S209A, S714A) in
combination with Cy3-tagged RNA. The experiment was performed as in a.
(g) Quiescent BJ cells were microinjected with a phospho-mimic mutant
(S714E) of GFP–PUM1, in combination with Cy3-tagged p27 RNA. The
experiment was performed as in a. Uncropped images of blots are shown in
Supplementary Information, Fig. S10.
4 nature cell biology advance online publication
© 2010 Macmillan Publishers Limited. All rights reserved.
LETTERS
interference (Fig. 4a and data not shown). The presence of two PREs in
the 3 UTR of PUM2 could explain this7. Suppression of PUM2, like that
of PUM1, led to an increase in p27 levels and a comparable delay in cell
cycle re-entry, suggesting a redundant activity with PUM1 (Supplementary
Information, Fig. S5b). Knockdown of both PUM1 and PUM2 signifi-
cantly increased p27 protein levels by elevating translation (Fig. 4b) and
halted proliferation (Fig. 4c). Knockdown of PUM1 and PUM2 in BJ cells
caused a delayed entry into S phase on stimulation with growth factor,
whereas BJ cells containing a stable p27 knockdown were insensitive to the
loss of PUM1 and PUM2 (Fig. 4d; Supplementary Information, Fig. S5c).
These results indicate that Pumilio proteins control cell cycle re-entry in
response to growth factors, and that this function is in part mediated by
controlling p27 expression.
Using the secondary structure prediction RNAfold software (Vienna
RNA package version 1.8.3)21, we noticed that the PRE and the miR-221
and miR-222 target site could form a stem-loop structure with consider-
able base-pair probability (Fig. 5a; Supplementary Information, Fig. S6).
We therefore speculated that PUM1-binding to the PRE favours opening
of the stem-loop structure, allowing miR-221 and miR-222 to gain access
to the p27-3 UTR in cycling cells. To study changes in RNA secondary
structure in vivo, we tagged RNA oligonucleotides containing the p27-
3 UTR PRE and the proximal miR-221/miR-222-binding site, with both
3 (fluorescein) and 5 (Cy3) fluorophores. On microinjection of this
RNA, the fluorescein lifetime in cycling BJ cells was significantly longer
than in quiescent cells, as a result of decreased energy transfer (FRET;
fluorescence resonance energy transfer) to the Cy3 fluorophore (Fig. 5b).
This suggests an increased distance between the two fluorophores and
thus a more open conformation of the stem-loop structure in cycling
cells. To examine the potential differences in donor-fluorophore life-
time and to test the specificity of this assay, we microinjected mutant
RNAs with strong and weak predicted secondary structures (Fig. 5b;
Supplementary Information, Fig. S7). A PRE mutant RNA that was ener-
getically more stable than the wild-type RNA (strong mutant) showed
a short fluorescein lifetime in both quiescent and cycling cells, suggest-
ing an unchanged, closed, RNA conformation. In contrast, an energeti-
cally weak structured RNA mutated in the miRNA site (weak mutant)
maintained a longer fluorescein lifetime in both quiescent and cycling
cells, suggesting an open conformation in both conditions. Because the
changes in FRET observed with the wild-type RNA were within the
range indicated by the mutant RNAs, our results imply that the meas-
ured changes in FRET represent actual structural differences. We also
tested changes in luciferase–p27-3 UTR reporter activity on mutation
of either PUM binding site with strong or weak complementarity to the
miRNA sites. Whereas the weak mutant of both PUM sites permitted
ab
cd
PUM1
PUM2
Tubulin
p27
PUM1
02045900204590 min
Control PUM1, PUM2 KD
1.0 1.0 1.0 1.3 1.0 1.3 1.5 1.8
Tubulin
p27
p27 (fold):
p27 KD
p27 KD
Control
PUM1, PUM2 KD
–
–
+
+
Tubulin
p27
Percentage of S-phase cells
Control
PUM1, PUM2 KD
Control p27 KD
0
1
2
3
4
5
6
7
8
9
Control siRNA
PUM1 siRNA
PUM2 siRNA
Figure 4 Pumilio regulates p27-dependent cell cycle re-entry from quiescence.
(a) HEK293 cells were transiently transfected with PUM1 siRNA, PUM2
siRNA and scrambled control siRNA, and immunoblot analysis was performed
with antibodies against PUM1, PUM2 and p27, with anti-tubulin as control.
(b) HEK293 cells transfected with PUM1- and PUM2-siRNA, or control, were
treated with the proteasome inhibitor MG-132. A densitometric analysis at
the indicated time points is shown below. Immunoblot analysis was performed
as in a. (c) HEK293 cells were transfected with shRNA vectors against p27
or control, and with either PUM1- and PUM2-siRNA or control siRNA. After
3 days the cell densities were revealed by staining with Coomassie blue.
(d) Wild-type BJ cells and BJ cells containing a stable p27 knockdown were
transfected with either PUM1- and PUM2-siRNA or control siRNA, and
deprived of growth factors for 72 h. After 16 h of subsequent stimulation
with growth factor, the percentage of cells in S phase was determined by flow
cytometric analysis of bromodeoxyuridine incorporation. The percentage of
cells in G1 and G2/M are shown in Supplementary Information, Fig. S5c. Error
bars represent s.d. for three independent experiments. Immunoblot analysis
of BJ cells stably expressing p27 KD or control was performed with antibody
against p27, with anti-tubulin as control. Uncropped images of blots are
shown in Supplementary Information, Fig. S10.
nature cell biology advance online publication 5
© 2010 Macmillan Publishers Limited. All rights reserved.
LETTERS
miRNA-mediated repression, altering either or both of the PUM sites to
strong mutants abrogated miRNA activity (Supplementary Information,
Fig. S8). These results suggest a functional interaction between Pumilio
and miR-221/miR-222 through their binding sites on the p27-3 UTR
and indicate that both Pumilio sites in p27-3 UTR may contribute to
the miRNA inhibitory structure in vivo.
On PUM1 knockdown in quiescent BJ cells, donor lifetime was not
affected when compared with transfection of control siRNAs, which is
consistent with an inactive state of PUM1 (Fig. 5c). In contrast, knock-
down of PUM1 in cycling cells abolished the increase in donor lifetime,
suggesting that the changes in conformation observed with the wild-
type oligonucleotide are dependent on PUM1 protein. These results are
supported by in vivo crosslinking of BJ cells and RT–PCR with a primer
designed to detect the structured RNA loop specifically. A PCR product
indicating a closed p27-3 UTR conformation was observed in quiescent
cells and in PUM1 and PUM2 knockdown cells but not in cycling cells
(Fig. 5d; Supplementary Information, Fig. S9).
Taken together, our results provide evidence in support of a model in
which, on stimulation by growth factors, Pumilio levels are increased and
RNA-binding activity is further enhanced by phosphorylation induc-
ing a conformational change in the p27-3 UTR (Fig. 5e). These changes
permit a more efficient binding of miR-221 and miR-222 specifically
to their target sites on the p27-3 UTR and tuning of cell cycle progres-
sion by repressing p27 expression. In addition, miRNA upregulation in
PUM1
miR-221
RISC
309
630
236
717
e
c
b
a
Crosslinked
Reversed
Bridge PCR GAPDH
d
P > 0.05
Strong:
Lifetime (ns)
P < 10–5
P > 0.05
QC
F
F
p27-3ʹ UTR:
Long lifetime Short lifetime
F
Base-pair probability
QCQC
Weak:WT:
–3.70–10.20–6.30ΔG:
2.1
2.2
2.3
2.4
2.5
2.6
2.7
2.8
0
F
Lifetime (ns)
Quiescent Cycling
P < 10–5
0
2.2
2.4
2.6
Bridge primerForward primer
miR site
PUM site
p27-3ʹ UTR
Q
C
Control
KD
Q
Control
KD
Growth factor
Stimulation
Low level of PUM1
High level of PUM1
(phosphorylated)
RISC
p27-3ʹ UTR
p27
Quiescence Cycling
miR-221/222
p
n = 27
n = 22
n = 21 n = 24
n = 21 n = 17 n = 19
n = 14 n = 9
n = 21
C
01
Base-pair probability
01
PUM1
Control siRNA
PUM1 siRNA
Figure 5 Pumilio binding alters local p27-3 UTR structure and miR-
221 and miR-222 accessibility. (a) Schematic representation of the
conformation of a region of the p27-3 UTR containing a PRE and a
miR-221/miR-222 site as predicted by RNAfold software. Base-pair
probability is indicated in the key. (b) Quiescent (Q) and cycling (C) BJ
cells were microinjected with short RNAs containing the p27-3 UTR-PRE
and the proximal miR-221/miR-222-binding site, and tagged with both
3 (fluorescein) and 5 (Cy3) fluorophores. The amount of conformational
free energy (DG in kcal mol-1) is listed for the wild-type (WT) and the two
mutant short RNAs (named accordingly ‘strong’ and ‘weak’). Differences
in Cy3 fluorophore lifetime (in ns) due to FRET are shown, and P values
are calculated for the differences in lifetime. Error bars represent s.d.
(c) BJ cells were transfected with PUM1 siRNA and control siRNA and
microinjected with the wild-type short RNA as in b. Error bars represent s.d.
(d) Model representing part of the p27-3 UTR, indicating the sequences
recognized by the bridge primer used for RT and PCR. Ethidium-bromide-
stained gels of PCRs performed on bridge or GAPDH reverse primer-primed
cDNA from crosslinked or crosslink-reversed RNA isolated from quiescent
(Q), cycling (C) and cycling PUM1 and PUM2 knockdown (KD) BJ cells. (e)
Model proposing a role for Pumilio RBPs in mammalian somatic cells. See
the text for details. Uncropped images of blots are shown in Supplementary
Information, Fig. S10.
6 nature cell biology advance online publication
© 2010 Macmillan Publishers Limited. All rights reserved.
LETTERS
response to growth factors has been reported in cancer cells, resulting in
global target downregulation, implying distinct modes of regulation to
achieve target specificity22. Our results reveal a highly conserved, specific
case of complementarity of an RBP target motif to a miRNA-binding site.
To our knowledge, this is the only demonstration of an RBP that modu-
lates miRNA activity by inducing a local structural switch in mRNA.
Considering the generally high conservation of some 3 UTR regions,
we expect that other RBPs may be found to modulate miRNA regulation
of other genes in a similar manner.
METHODS
Methods and any associated references are available in the online version
of the paper at http://www.nature.com/naturecellbiology/
Note: Supplementary Information is available on the Nature Cell Biology website.
ACKNOWLEDGEMENTS
We thank all members of the Agami laboratory for technical help and discussions.
We also thank André Gerber for constructs, Kees Jalink for advice on fluorescence
lifetime imaging microscopy measurements, and R. B. Israel for assistance with
statistical analysis. This work was supported by the EURYI (European research
young investigator award), ERC (European Research Council), KWF (koningin
wilhelmina fonds; Dutch cancer foundation) and Horizon-NWO (Nederlandse
Organisatie voor Wetenschappelijk Onderzoek; R.A.) and an EMBO long-term
fellowship (R.E.).
AUTHOR CONTRIBUTIONS
M.K. and M.v.K. performed most of the experimental work. R.A. supervised the
project. W.Z. performed fluorescence lifetime imaging microscopy and confocal
laser scanning microscopy analyses. J.O.V. provided technical assistance. R.E.
performed bioinformatical analyses. M.K., M.v.K. and R.A. wrote the manuscript.
COMPETING FINANCIAL INTERESTS
The authors declare no competing financial interests.
Published online at http://www.nature.com/naturecellbiology
Reprints and permissions information is available online at http://npg.nature.com/
reprintsandpermissions/
1. Filipowicz, W., Bhattacharyya, S. N. & Sonenberg, N. Mechanisms of post-transcrip-
tional regulation by microRNAs: are the answers in sight? Nature Rev. Genet. 9, 102–
114 (2008).
2. Kedde, M. & Agami, R. Interplay between microRNAs and RNA-binding proteins deter-
mines developmental processes. Cell Cycle 7, 899–903 (2008).
3. Hengst, L. & Reed, S. I. Translational control of p27Kip1 accumulation during the cell
cycle. Science 271, 1861–1864 (1996).
4. Chu, I. M., Hengst, L. & Slingerland, J. M. The Cdk inhibitor p27 in human can-
cer: prognostic potential and relevance to anticancer therapy. Nature Rev. Cancer 8,
253–267 (2008).
5. le Sage, C. et al. Regulation of the p27Kip1 tumor suppressor by miR-221 and miR-222
promotes cancer cell proliferation. EMBO J. 26, 3699–3708 (2007).
6. Lotterman, C. D., Kent, O. A. & Mendell, J. T. Functional integration of microRNAs into
oncogenic and tumor suppressor pathways. Cell Cycle 7, 2493–2499 (2008).
7. Galgano, A. et al. Comparative analysis of mRNA targets for human PUF-family proteins
suggests extensive interaction with the miRNA regulatory system. PLoS ONE 3, e3164
(2008).
8. Morris, A. R., Mukherjee, N. & Keene, J. D. Ribonomic analysis of human Pum1 reveals
cis-trans conservation across species despite evolution of diverse mRNA target sets.
Mol. Cell. Biol. 28, 4093–4103 (2008).
9. Kloosterman, W. P. & Plasterk, R. H. The diverse functions of microRNAs in animal
development and disease. Dev. Cell 11, 441–450 (2006).
10. Voorhoeve, P. M. & Agami, R. Classifying microRNAs in cancer: the good, the bad and
the ugly. Biochim. Biophys. Acta 1775, 274–282 (2007).
11. Nolde, M. J., Saka, N., Reinert, K. L. & Slack, F. J. The Caenorhabditis elegans pumilio
homolog, puf-9, is required for the 3UTR-mediated repression of the let-7 microRNA
target gene, hbl-1. Dev. Biol. 305, 551–563 (2007).
12. Brummelkamp, T. R., Bernards, R. & Agami, R. Stable suppression of tumorigenicity
by virus-mediated RNA interference. Cancer Cell 2, 243–247 (2002).
13. Kedde, M. et al. RNA-binding protein Dnd1 inhibits microRNA access to target mRNA.
Cell 131, 1273–1286 (2007).
14. Pillai, R. S. et al. Inhibition of translational initiation by Let-7 microRNA in human
cells. Science 309, 1573–1576 (2005).
15. van Rooij, E. et al. Dysregulation of microRNAs after myocardial infarction reveals a
role of miR-29 in cardiac fibrosis. Proc. Natl Acad. Sci. USA 105, 13027–13032
(2008).
16. Leung, A. K., Calabrese, J. M. & Sharp, P. A. Quantitative analysis of Argonaute protein
reveals microRNA-dependent localization to stress granules. Proc. Natl Acad. Sci. USA
103, 18125–18130 (2006).
17. Kedersha, N. et al. Stress granules and processing bodies are dynamically linked sites
of mRNP remodeling. J. Cell Biol. 169, 871–884 (2005).
18. Gnad, F. et al. PHOSIDA (phosphorylation site database): management, structural
and evolutionary investigation, and prediction of phosphosites. Genome Biol. 8, R250
(2007).
19. Olsen, J. V. et al. Global, in vivo, and site-specific phosphorylation dynamics in signaling
networks. Cell 127, 635–648 (2006).
20. Fero, M. L., Randel, E., Gurley, K. E., Roberts, J. M. & Kemp, C. J. The murine
gene p27Kip1 is haplo-insufficient for tumour suppression. Nature 396, 177–180
(1998).
21. Hofacker, I. L. Vienna RNA secondary structure server. Nucleic Acids Res. 31, 3429–
3431 (2003).
22. Medina, R. et al. MicroRNAs 221 and 222 bypass quiescence and compromise cell
survival. Cancer Res. 68, 2773–2780 (2008).
nature cell biology advance online publication 7
© 2010 Macmillan Publishers Limited. All rights reserved.
METHODS DOI: 10.1038/ncb2105
METHODS
Constructs and antibodies. MiR-Vec constructs and the pGL3-p27-3 UTR and miR
mutants were described previously5,23. The PREs in the p27-3 UTR were mutated
(weak) to the following sequences using the Stratagene multisite-directed mutagen-
esis kit: PRE1, 5-tgtatata-3 to 5-ggtatgta-3; PRE2, 5-tgtacata-3 to 5-ggtacgta-3
(strong mutants are shown below). Constructs for RPA detection of hTR, cyclophi-
lin, p27 and miR-221 were described previously13; the RPA probe sequence for miR-
125b was 5-CUCAGUCCCUGAGACCCUAACUUGUGAUGUUU-3. Probes
were prepared in accordance with the manufacturer’s instructions (Ambion mirVana
probe construction kit). shRNA for p27 was described previously5; the shPUM1.1
sequence was 5-AATCCAACATGTACTGGAGCA-3, the shPUM1.3 sequence
was 5-AACAGACCACCCCACAGGCTC-3, the shPUM1.4 sequence was
5-AATTCAGCTAATCAACAGACC-3; these were cloned in pRETRO-SUPER.
siRNAs ordered from Ambion were against PUM1 (no. 138317), PUM2 (no. 138319)
and a scrambled control (5-CUGUAGCCGUAUCAAGUCGUUCCUGTT-3)
from Invitrogen. The PUM1–TAP construct was a gift from A. Gerber. The
GFP–PUM1(HD) and GFP–RBP constructs were made by cloning the cDNA
into the Clontech eGFP vector; mutants were made with the Stratagene mul-
tisite-directed mutagenesis kit. All constructs were sequence-verified. The
Cy3-wt-RNA oligonucleotide (5-ACUACCUGUGUAUAUAGUUUUU-3)
and the Cy3-mt-RNA oligonucleotide (5-ACUACCUCUCCAUAUAG
UUUUU-3 were labelled 3 (Dharmacon). Labelled RNA oligonu-
cleotides (3 (fluorescein) and 5 (Cy3)) used for FLIM were wild-type
(5-CUGUGUAUAUAGUUUUUACCUUUUAUGUAGCACAU-3), strong
mutant (5-CUGUGCACAUAGUUUUUACCUUUUAUGUAGCACAU-3) and
weak mutant (5-CUGUGUAUAUAGUUUUUACCUUUUAGGUCGGAGAU-3)
(Dharmacon).
Antibodies used were AGO2 (Transduction Labs and Abcam), actin (Abcam),
p27 (Transduction Labs), CDK4 (C22), p53 (DO1) and CDK6 (Santa Cruz),
PUM1 and PUM2 (Bethyl Labs), tubulin (YL1/2 ECACC), rabbit GFP and bro-
modeoxyuridine (Dako).
Cell culture, transfections, dual luciferase activity analysis and cell cycle profile
analysis. HEK293, MCF7 and BJ primary fibroblast cells were cultured in DMEM
supplemented with 10% heat-inactivated fetal calf serum (FCS) in 5% CO2 at
37 °C. HEK293 cells were transiently transfected by using calcium phosphate
precipitation. MCF7 cells were transfected with Fugene (Roche) for luciferase
analysis with 10 ng of reporter, 5 ng of Renilla control plasmid, 250 ng of either
miR-Vec or miR-Vec control, and 250 ng of knockdown construct for PUM1 or
control. Dual luciferase activity assays were performed 72 h after transfection in
accordance with the manufacturer’s instructions (Promega). BJ cells were trans-
fected with siRNAs in a final concentration of 50 nM with the use of Dharmafect
reagent (Dharmacon), in accordance with the manufacturer’s instructions. To
obtain quiescent BJ cells, cells were cultured for 72 h in DMEM containing 0.25%
FCS. Antagomir sequences were described previously5 and applied to the cells
overnight at a final concentration of 15 mM. The proteasome inhibitor MG-132
was from Sigma, used at a final concentration of 10 mM. For cell cycle profile
analysis, quiescent BJs were stimulated with growth factors. Cell cycle analysis
was performed as described previously24.
Immunoprecipitation, immunoblotting, RNAse protection assays and qRT–
PCR analysis. PUM1 and AGO2 were immunoprecipitated from BJ cell extracts
with GammaBind G Sepharose (GE Healthcare). Beads were preblocked with
yeast tRNA (Invitrogen) and RNase-free BSA (Ambion) and then washed; extracts
were sonicated and cleared in lysis buffer (100 mM KCl, 10 mM Tris-HCl pH 7.5,
0.1% Nonidet P40, 0.5% Tween 20, 5 mM MgCl2, 2 mM b-glycerophosphate,
0.5 mM dithiothreitol, protease inhibitor mixture (Roche Applied Science) and
RNAse-OUT (Invitrogen). Extracts were incubated for 4 h with antibodies against
AGO2, CDK4, CDK6 or PUM1 (1 mg per immunoprecipitation) in a tumbler
placed at 4 °C. Thereafter, beads were washed and a 10% aliquot was used for
immunoblot analysis; from the remainder, RNA was extracted (Trizol, Invitrogen)
to be subjected to RPA or qRT–PCR analysis. PUM1–TAP was precipitated from
transiently transfected HEK293 cells by using rabbit IgG Sepharose (Sigma) in
lysis buffer with 125 mM NaCl instead of KCl as described above. Beads were
washed and incubated for 20 min with 32P-labelled oligonucleotides (wild-type
and mutant, described above) at 30 °C. Beads were washed, and bound RNA and
proteins were revealed on gel.
For immunoblot analysis, extracts were separated on 10% SDS–PAGE gels,
and transferred to Immobilon-P membranes (Millipore). Western blots were
developed with Supersignal (Pierce) or by enhanced chemiluminescence (ECL;
Amersham Biosciences) and exposed to film (Kodak). Densitometric analysis
was performed with AIDA software (Raytest).
RPAs for p27 and cyclophilin were performed with the HybSpeed RPA and
MAXIscript kits (Ambion) as described13. For miRNAs, we used mirVana kits
(Ambion) in accordance with the manufacturer’s instructions13. Northern analysis
was performed with standard protocols and RPA probe for p27.
For mRNA qRT–PCR, cDNA (from 3 mg RNA) was synthesized with
SuperScript III and primed with oligo(dT) in accordance with the manufac-
turer’s instructions (Invitrogen). For combined miRNA and mRNA qRT–PCR,
about 100 ng of input RNA and 20% of immunoprecipitated RNA was used for
cDNA synthesis with random primers from a Taqman High Capacity cDNA
kit (Applied Biosystems), in accordance with the manufacturer’s instructions.
Primers for qPCR were PUM1 (5-AAAAACCTGAGAAGTTTGAATTGT-3
(forward) and 5-GCAAGACCAAAAGCAGAGTTG-3 (reverse))
and COL3A1 (5-AACACGCAAGGCTGTGAGACT-3 (for ward) and
5-GCCAACGTCCACACCAAATT-3 (reverse)); p27, GAPDH and b-actin
primers were as described13. QPCR primers for miR-221, miR-222, miR-29a,
GAPDH and 18S were from Applied Biosystems. Analysis was performed with
SYBR Green PCR master mix or TaqMan UNG master mix (Applied Biosystems)
and Chromo 4 system (Bio-Rad Laboratories).
Crosslink bridge RT–PCRs were performed with bridge reverse
(5-CTTCCCCAAAGTTTATAGGTAG-3) and GAPDH reverse primer; PCR
forward primer was 5-TATAAGCACTGAAAAACAACAACAC-3. BJ cells were
crosslinked for 15 min with 1% formaldehyde (Sigma), inactivated with 330 mM
glycine (Sigma), sonicated and cleared. Cleared lysate was treated for 1 h with
proteinase K (Invitrogen) at 37 °C and inactivated with phenylmethylsulphonyl
fluoride (Sigma). RNA was extracted, and reverse crosslinking was performed
for 1 h at 70 °C.
Fluorescence lifetime imaging microscopy (FLIM). Before FLIM experiments,
cells were grown on coverslips and microinjected with RNA labelled 3 with fluo-
rescein and 5 with Cy3. Subsequently, cells were mounted in bicarbonate-buffered
saline (140 mM NaCl, 5 mM KCl, 1 mM MgCl2, 1 mM CaCl2, 23 mM NaHCO3,
10 mM glucose, 10 mM HEPES at pH 7.3) in a heated tissue-culture chamber
at 37 °C under 5% CO2. FLIM experiments were performed on a Leica inverted
DM-IRE2 microscope equipped with a Lambert Instruments frequency domain
lifetime attachment (Leutingewolde), controlled by the vendor’s LI FLIM software.
Fluorescein was excited with about 4 mW of 488-nm light from a light-emitting
diode modulated at 40 MHz; emited light was collected at 490–550 nm with an
intensified charge-coupled-device camera (CoolSNAP HQ; Roper Scientific). To
calculate the fluorescein lifetime, the intensities from 12 phase-shifted images
(modulation depth about 70%) were fitted with a sinus function, and lifetimes
were derived from the phase shift between excitation and emission. Differences
in lifetimes were assigned P values with Student’s t-test.
CLSM analysis. For CLSM analysis, BJs were microinjected with GFP–PUM1
or its mutants, in combination with Cy3-labelled RNA. After expression of the
GFP–PUM1 overnight, cells were fixed with 3.7% formaldehyde in PBS, and
coverslips were mounted in Vectashield mounting medium (Vector Laboratories).
The specimens were imaged with a Leica TCS SP2 System equipped with a 63×
oil-immersion objective. Endogenous stainings for PUM1 and AGO2 were per-
formed in accordance with the manufacturer’s instructions. Scatter plots for co-
localization analysis were generated with ImageJ WCIF software (http://www.
uhnresearch.ca/wcif).
23. Voorhoeve, P. M. et al. A genetic screen implicates miRNA-372 and miRNA-373 as
oncogenes in testicular germ cell tumors. Cell 124, 1169–1181 (2006).
24. Duursma, A. & Agami, R. p53-dependent regulation of Cdc6 protein stability controls
cellular proliferation. Mol. Cell. Biol. 25, 6937–6947 (2005).
25. Kent, W. J. et al. The human genome browser at UCSC. Genome Res. 12, 996–1006
(2002).
8 nature cell biology advance online publication
© 2010 Macmillan Publishers Limited. All rights reserved.
supplementary information
www.nature.com/naturecellbiology 1
DOI: 10.1038/ncb2105
Figure S1 MiR-221/222 and p27 levels in quiescent (Q) versus cycling
(C) cells. (a) MiR-221 and miR-222 expression analysis on the Exiqon v2
microRNA microarray platform. M represents fold change (log2) as detected
in quiescent versus cycling BJ cells. Hybridization was performed using a
standard protocol (http://microarrays.nki.nl). (b) Northern blot for p27 mRNA
and ethidium bromide staining for 18S ribosomal RNA in quiescent versus
cycling BJ cells. Densitometric analysis resulted in the normalized amounts
displayed below. (c) The amount of p27 mRNA was measured by qRT-PCR
in cycling and quiescent BJ cells. Error bars represent SD from triplicate
reactions. (d) Expression analysis of p27 on the Illumina Sentrix BeadChip
v3 microarray platform. Absolute expression values were obtained with two
probes in quiescent versus cycling BJ cells. The data is a representative of
a duplo experiment. Hybridization was performed using a standard protocol
(http://microarrays.nki.nl). (e) An antago-p53kd was administered to
quiescent and cycling BJ-p53kd cells and immunoblot analysis on BJ-p53kd
and control cells was performed with p53 and actin control antibodies.
Figure S1aFigure S1b
1,04 1
Q C
p27
mRNA
18S
Figure S1d
Figure S1c
Figure S1e
p27 mRNA levels
0
200
400
600
800
1000
C
Q
C
Q
probe 1
probe 2
p27 levels (array)
p27 levels (qPCR)
0.0
0.4
0.8
1.2
C Q
Relative p27 mRNA levels
ctrlp53 kd
quiescentcycling
inhibitor:ctrlshp53ctrlshp53
--
p53
actin
© 2010 Macmillan Publishers Limited. All rights reserved.
supplementary information
2 www.nature.com/naturecellbiology
Figure S2 Relative amounts of miRNAs and associated mRNAs in AGO2 IPs.
(a) The amounts of miR-222 and p27 mRNA were measured by qRT-PCR
in the IPs shown in Fig. 1e. Results are presented as relative p27/miR-
222 ratio. The ratio in the IPs from cycling BJ cells was set to 1. Error bars
represent SEM from triplicate reactions. (b) Enrichment factors and SEM of
miR-221, miR-222, p27 and gapdh control mRNA in AGO2 IPs over CDK4
IPs. (c) qRT-PCR performed as in a, for miR-221 and p27 mRNA in AGO2 IPs
from formaldehyde crosslinked BJ cells. Enrichment factors and SEM of miR-
221, p27 and gapdh control mRNA in AGO2 IPs over CDK4 IPs are shown in
the table. (d) qRT-PCR performed as in a for miR-29a and COL3A1 mRNA.
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
Relative p27 mRNA/miR-221
QC
crosslinked AGO2-miRNA IP:
Figure S2a
Figure S2c
Figure S2b
C
+/-
C
+/-
p27
4.9
2.1
p27
3.2
0.5
miR-221
8.3
0.9
miR-222
3.7
0.5
gapdh
1.4
0.1
gapdh
1.2
0.4
Enrichment in AGO2 IP/CDK4 IP
Figure S2d
C
+/-
p27
4.7
1.3
miR-221
5.0
0.7
gapdh
1.0
0.2
QC
Relative p27 mRNA/miR-222
0.0
0.2
0.4
0.6
0.8
1.0
1.2
AGO2-miRNA IP:
Relative COL3A1 mRNA/miR-29a
QC
0.0
0.5
1.0
1.5
2.0
2.5
AGO2-miRNA IP:
© 2010 Macmillan Publishers Limited. All rights reserved.
supplementary information
www.nature.com/naturecellbiology 3
Figure S3 Effect of PUM1 knockdown on miR-221 mediated regulation towards
the p27-3’UTR. (a) MCF7 cells were transfected with shRNA vectors against
PUM1 or control and selected, immunoblot analysis was performed with PUM1
and tubulin control antibodies. (b) Luciferase assay performed as in Fig. 2c,
with a luciferase construct coupled to the p27-3’UTR mutated for the miR-
221/222 sites. Error bars represent SD from three independent experiments.
Figure S3a
PUM1
tubulin
Figure S3b
relative luciferase activity
ctrl
kd 1
kd 3
ctrl miR-
221
ctrl miR-
221
ctrl miR-
221
p27 miR DM 3’UTR
2
2
2
2
2
2
2
p2
p2
7
7
7
7
7
7
7
7
i
i
i
i
i
i
i
mi
mi
R
R
R
R
R
R
R
R
R
DM
DM
DM
DM
DM
DM
DM
DM
DM
3
3
3
3
3
3
3
3
3
’U
’U
’U
’U
’U
’U
’U
’U
U
TR
TR
TR
TR
TR
TR
TR
TR
TR
-A(n)
luciferase
xx
© 2010 Macmillan Publishers Limited. All rights reserved.
supplementary information
4 www.nature.com/naturecellbiology
Figure S4 Localization and expression level analysis of PUM1 and control RBPs.
(a) Confirmation of expression levels of GFP-PUM1 constructs by immunoblot
for GFP and tubulin. (b) Zoom-in of the inset in merge panel and scatterplots in
Fig. 3a,f,g. (c) Immunostaining for endogenous PUM1 and AGO2 in quiescent
and cycling BJ cells. Colocalisation analyses within the same cell were
performed through a scatter plot. Inset shows zoom-in on the highlighted area.
Scalebar represents 10 mm. Representative pictures are shown. (d) Cycling BJ
cells were microinjected with GFP-RBP control constructs and a Cy3-labeled
RNA as in Fig. 3a. Representative pictures are shown. (e) Confirmation of
expression levels of GFP-PUM1HD and GFP-RBP constructs by immunoblot
for GFP and tubulin. (f) Quantification of RPA signals as shown in Fig. 3c,
quantifications were performed with Phosphoimager software. (g) Quiescent or
cycling BJ cells were microinjected with the GFP-PUM1HD construct and a Cy3-
labeled RNA as in Fig. 3a. Representative pictures are shown.
Figure S4b
Figure S4a
merge
merge
tubulin
GFP
GFP-PUM1
quiescent
cycling
cycling
0.20
0.31
0.62
Rr:
quiescent
S714E
0.82
0.55
0.19
Rr:
cycling
S209A
cycling
S714A
AGO2PUM1merge
quiescent
cycling
Figure S4c
© 2010 Macmillan Publishers Limited. All rights reserved.
supplementary information
www.nature.com/naturecellbiology 5
Figure S4 continued
Figure S4dFigure S4e
GFP
tubulin
-
-
-
-
64
81
114
122
kDa:
Figure S4f
quiescent
cycling
Figure S4g
totIP
RPA
qccqcc
cyclophilin
p27
probe
pf
<0.01 <0.01 <0.01
0.03 <0.01 0.19
p27:
cyclophilin:
IP/total:
Figure S4f
quiescent
cycling
Figure S4g
totIP
RPA
qccqcc
cyclophilin
p27
probe
pf
<0.01 <0.01 <0.01
0.03 <0.01 0.19
p27:
cyclophilin:
IP/total:
Figure S4f
Figure S4g
© 2010 Macmillan Publishers Limited. All rights reserved.
supplementary information
6 www.nature.com/naturecellbiology
0.0
0.4
0.8
1.2
1.6
0
2
4
6
8
10
12
14
ctrl
PUM1 kd
PUM2 kd
PUM1,2 kd
% S-phase
16 hours stimulation
ctrlkd1kd3
Relative % G1 or S-phase
Figure S5b
Figure S5a
S-phase
G1-phase
%
ctrl
PUM1 kd
PUM2 kd
PUM1,2 kd
G1
73.1
74.2
76.2
79.5
S
11.9
9.6
9.9
5.4
G2/M
14.8
16.0
13.4
14.4
18 hours stimulation
wt
%
ctrl
PUM1,2 kd
G1
84.5
85.1
S
6.5
3.0
G2/M
8.4
11.3
p27 kd
%
ctrl
PUM1,2 kd
G1
86.1
85.3
S
7.7
7.1
G2/M
5.8
7.2
Figure S5c
PUM1
tubulin
PUM2
ctrl PUM1kd PUM2kd PUM1,2kd
%
ctrl
PUM1 kd1
PUM1 kd3
PUM1 kd4
G1
62.0
71.9
65.0
77.2
S
33.9
24.4
29.4
23.4
G2/M
2.9
3.0
3.6
2.7
kd4
Figure S5 Loss of Pumilio affects cell cycle progression. (a) BJ cells
containing stable PUM1 knockdowns or control were growth factor deprived
and then stimulated for 16 hours with growth factors. The percentage of cells
in S-phase was determined by flow cytometric analysis of BrdU incorporation.
G1- and G2/M- phase percentages as measured by propidium iodine are
shown below. Error bars represent the SD of triplicate experiments. (b) BJ
cells were transfected with either siPUM1, siPUM2, both, or control siRNA,
and growth factor deprived for 72 hours. After 18 hours of subsequent growth
factor stimulation, the percentage of cells in S-phase was determined by flow
cytometric analysis of BrdU incorporation. G1- and G2/M- phase percentages
as measured by propidium iodine are shown below. Error bars represent the
SD of triplicate experiments. Immunoblot analysis of quiescent BJ cells
stimulated with growth factors for PUM1&2 and tubulin control. (c) G1- and
G2/M- phase percentages as measured by propidium iodine from Fig. 4d.
© 2010 Macmillan Publishers Limited. All rights reserved.
supplementary information
www.nature.com/naturecellbiology 7
Figure S6 Predicted conformation of the complete p27-3’UTR. A schematic representation of the conformation of the complete p27-3’UTR as predicted by
RNAfold software. Base pair probability is indicated in the legend.
Figure S6
Figure S7a
Lifetime (ns)
3.0
3.2
3.4
3.6
3.8
4.0
4.2
4.4
Q
C
Q
C
Q
C
wt
weak
strong
Figure S7b
Base pair probability
15 bp
60 bp
10 bp
50 bp
Native gelDenaturing gel
wtstrongweakwtstrongweakRNA:
n=9n=9n=7n=11n=10n=6
© 2010 Macmillan Publishers Limited. All rights reserved.
supplementary information
8 www.nature.com/naturecellbiology
Figure S6
Figure S7a
Lifetime (ns)
3.0
3.2
3.4
3.6
3.8
4.0
4.2
4.4
Q
C
Q
C
Q
C
wt
weak
strong
Figure S7b
Base pair probability
15 bp
60 bp
10 bp
50 bp
Native gelDenaturing gel
wtstrongweakwtstrongweakRNA:
n=9n=9n=7n=11n=10n=6
Figure S7 Conformation of tagged RNA oligos and single labelled FLIM
control. (a) Gel migration analysis of RNA oligos containing wildtype, strong
and weak mutant p27-3’UTR-PRE and the proximal miR-221/222 binding
site. (b) Quiescent (Q) and cycling (C) BJ cells were microinjected with
short RNAs containing the p27-3’UTR-PRE and the proximal miR-221/222
binding site, and tagged with 3’ (fluorescein) fluorophore. The wildtype,
weak and strong mutants described in the text were used. Cy3 fluorophore
lifetimes (in ns) due to FRET are displayed. Error bars represent SD.
© 2010 Macmillan Publishers Limited. All rights reserved.
supplementary information
www.nature.com/naturecellbiology 9
Figure S8 Functional interaction between Pumilio and miRNA sites in the
p27-3’UTR. Luciferase assay performed as in Fig. 2c, in HEK293 cells
(endogenously expressing miR-221/222), with luciferase constructs coupled
to the p27-3’UTR mutated for the miR-221/222 sites (miR DM), and several
constructs mutated for both Pumilio sites (see schematic representation
below). Pumilio sites 1 and 2 are shown in green, miRNA-221/222 sites
are shown in red, weak and strong PUM site mutants are as described in the
text. Error bars represent SD from three independent experiments.
Figure S8
Figure S9b
PUM1
Q
ctrl Pum1,2 kd
tubulin
PUM2
C
Figure S9a
relative luciferase activity
miR DM wt PUM DM
weak
weak
PUM DM
strong
strong
PUM DM
strong
weak
PUM DM
weak
strong
p27 3’UTR
-A(n)
luciferase
PUM site 1:
PUM site 2:
12
PUM site:
© 2010 Macmillan Publishers Limited. All rights reserved.
supplementary information
10 www.nature.com/naturecellbiology
Figure S9 In vivo crosslinking reveals predicted secondary p27-3’UTR structure.
(a) Schematic representation of the bridge PCR product (yourseq) adapted from
BLAT search function 23 with nucleotide numbers shown. Sequence is shown
below, as expected the miR-221/222 site (red) and the Pumilio site (green) are
missing from the PCR product. (c) Immunoblots showing levels of PUM1&2 and
control tubulin from crosslinked BJ cells from Fig. 5d.
Figure S8
Figure S9b
PUM1
Q
ctrl Pum1,2 kd
tubulin
PUM2
C
Figure S9a
relative luciferase activity
miR DM wt PUM DM
weak
weak
PUM DM
strong
strong
PUM DM
strong
weak
PUM DM
weak
strong
p27 3’UTR
-A(n)
luciferase
PUM site 1:
PUM site 2:
12
PUM site:
© 2010 Macmillan Publishers Limited. All rights reserved.
supplementary information
www.nature.com/naturecellbiology 11
Figure S10 Full scans
Figure 1a
miR-221
miR-125b
RP A
p27 mRNA
cyclophilin
Figure 1e
Figure 1c
tubulin
p27
cyclingquiescent
tot
IP
QCC
QCC
AGO2CDK4
tubulin
AGO2
probe
Figure 3ctotIP
cyclophilin
PUM1
*
RP A
Western
p27
qccqcc
64 kDa
49 kDa
82 kDa
115 kDa
37 kDa
26 kDa
64 kDa
49 kDa
156 bases
103 bases
156 bases
103 bases
Figure 3bTAP-IP
wt mt wt mt
PUM1
tubulin
32P-RNA
20 nt
Western
64 kDa
49 kDa
180 kDa
115 kDa
probe
Figure 3d
PUM1-TAP
PUM1
Q
tubulin
180 kDa
115 kDa
64 kDa
49 kDa
Figure 4a
PUM1
PUM2
tubulin
p27
64 kDa
49 kDa
37 kDa
26 kDa
180 kDa
115 kDa
180 kDa
115 kDa
Figure 4d
tubulin
p27
64 kDa
49 kDa
37 kDa
26 kDa
Figure 4b
PUM1
02045900204590
ctrlPUM1,2 kd
tubulin
p27
64 kDa
49 kDa
37 kDa
26 kDa
180 kDa
115 kDa
26 kDa
p27
Supplementary Figure 10
XX
XX
X
35 bases
35 bases
21 bases
21 bases
180 kDa
115 kDa
64 kDa
49 kDa
37 kDa
26 kDa
© 2010 Macmillan Publishers Limited. All rights reserved.
