Brf1 posttranscriptionally regulates pluripotency and
differentiation responses downstream of Erk
MAP kinase
Frederick E. Tan
a,b
and Michael B. Elowitz
a,b,1
a
Howard Hughes Medical Institute and
b
Division of Biology and Departments of Bioengineering and Applied Physics, California Institute of Technology,
Pasadena, CA 91125
Edited by Gideon Dreyfuss, University of Pennsylvania, Philadelphia, PA, and approved March 25, 2014 (received for review November 6, 2013)
AU-rich element mRNA-binding proteins (AUBPs) are key regula-
tors of development, but how they are controlled and what
functional roles they play depends on cellular context. Here, we
show that Brf1 (
zfp36l1
), an AUBP from the Zfp36 protein family,
operates downstream of FGF/Erk MAP kinase signaling to regulate
pluripotency and cell fate decision making in mouse embryonic
stem cells (mESCs). FGF/Erk MAP kinase signaling up-regulates
Brf1, which disrupts the expression of core pluripotency-associ-
ated genes and attenuates mESC self-renewal without inducing
differentiation. These regulatory effects are mediated by rapid
and direct destabilization of Brf1 targets, such as Nanog mRNA.
Enhancing Brf1 expression does not compromise mESC pluripo-
tency but does preferentially regulate mesendoderm commitment
during differentiation, accelerating the expression of primitive
streak markers. Together, these studies demonstrate that FGF sig-
nals use targeted mRNA degradation by Brf1 to enable rapid post-
transcriptional control of gene expression in mESCs.
stem cell biology
|
AU-rich element RNA-binding proteins
|
developmental mechanisms
|
developmental signaling pathways
|
gene regulation dynamics
A
U-rich element mRNA-binding proteins (AUBPs) represent
an important class of regulators required for the proper
development of embryonic and adult tissues in the mouse (1),
but whether they have developmentally important roles in mouse
embryonic stem cells (mESCs) remains unclear. A recent pro-
teomic survey identified more than 500 mRNA-binding proteins
in mESCs, several of which are AUBPs (2). Independent of the
micro-RNA pathway, AUBPs are known regulators of splicing,
mRNA stability, translat
ional efficiency, and RNA transport (3),
and could provide an additional layer of developmental regulation
that complements other pluripotency and self-renewal mecha-
nisms. AUBPs are essential in many developmental systems, such
as during hematopoiesis, neurogenesis, germ cell commitment, and
placental morphogenesis (4
–
6). Their absence or misregulation can
be lethal and often promotes disease progression (7
–
9). Despite
growing interest in the many functions of AUBPs, their regulation
and function in mESCs remains poorly understood.
The expression and activity of AUBPs is known to be regu-
lated by growth factor signaling in many cellular contexts (10,
11). In mESCs, the FGF/Erk MAP kinase signaling pathway is
a central regulator of self-renewal, pluripotency, and differenti-
ation (12, 13). Although much is known about the developmental
effects of FGF/Erk MAP kinase signaling inhibition or activation
(14, 15), the regulatory mechanisms used downstream of Erk1/2
often remain unclear. Various transcriptional, posttranscriptional,
and posttranslational mechanisms are engaged to regulate target
genes (16). As part of this signaling cascade, AUBPs could me-
diate rapid signaling-dependent responses, but this potential role
has not been investigated.
Here, we show that the expression of Brf1 (
zfp36l1
) is regu-
lated by FGF/Erk MAP kinase signaling in both pluripotent and
differentiating mESCs. Brf1 is a member of the Zfp36 AUBP
family that plays critical roles throughout mouse development.
Without Brf1, embryos die in utero at approximately embryonic
day 10.5 (E10.5) as a result of allantoic, placental, and neural
tube defects (9), and its absence in adults promotes leukemia
(17). In mESCs, Brf1 binds AU-rich sequences in many pluri-
potency-associated mRNAs, including Nanog, to regulate their
localization and abundance. This regulation broadly perturbs the
core transcription factor network without inducing differentiation,
but compromises the capacity to self-renew. In differentiation-
stimulating conditions, Brf1 enhances the expression of primitive
streak markers, indicating an accelerated commitment to mesen-
doderm. Together, these data identify targeted mRNA degrada-
tion by Brf1 as a mechanism through which the biology of mESCs
is regulated and controlled by FGF/Erk MAP kinase signals.
Results
Erk MAP Kinase Signaling Regulates the Expression of Zfp36 AUBPs.
AUBP expression in mESCs has been documented by several
groups (2, 9, 18). The known sensitivity of AUBPs to growth
factors suggested that these proteins could be regulated by FGF/
Erk MAP kinase signaling (11, 19). To explore this potential
regulatory connection, we first profiled the transcriptome of E14
mESCs using high-throughput sequencing to determine which
AUBPs are actively expressed (Fig. 1
A
and
Dataset S1
). We
identified several classes of AUBPs, including (
i
) members of the
Significance
Intercellular signaling pathways strongly regulate gene ex-
pression in uncommitted precursor stem cells, but the mecha-
nisms through which these signaling pathways regulate gene
targets often remain unclear. We address this question in
mouse embryonic stem cells (mESCs) and highlight the impor-
tance of AU-rich element mRNA-binding proteins as regulatory
intermediates of intercellular signaling. We show that the
FGF/Erk MAP kinase signaling pathway strongly influences the
expression of Brf1, a member of the Zfp36 protein family that
is known to bind and destabilize its mRNA targets. Brf1 phys-
ically binds many pluripotency and differentiation-associated
mRNAs. Moderate changes in its expression compromise self-
renewal capacity and bias fate commitment, thus providing
a posttranscriptional link between intercellular signaling ac-
tivity and gene expression in mESCs.
Author contributions: F.E.T. and M.B.E. designed research; F.E.T. performed research;
F.E.T. contributed new reagents/analytic tools; F.E.T. and M.B.E. analyzed data; and F.E.T.
and M.B.E. wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission.
Data deposition: The data reported in this paper have been deposited in the Gene Ex-
pression Omnibus (GEO) database,
www.ncbi.nlm.nih.gov/geo
(accession no.
GSE40104
).
1
To whom correspondence should be addressed. E-mail: melowitz@caltech.edu.
This article contains supporting information online at
www.pnas.org/lookup/suppl/doi:10.
1073/pnas.1320873111/-/DCSupplemental
.
E1740
–
E1748
|
PNAS
|
Published online April 14, 2014
www.pnas.org/cgi/doi/10.1073/pnas.1320873111
Zfp36 protein family [TTP (
zfp36
), Brf1 (
zfp36l1
), and Brf2
(
zfp36l2
)], which are known to play critical roles during hema-
topoiesis by destabilizing Cytokine and Notch-Delta signaling-
associated mRNAs; (
ii
) members of the Hu protein family [HuR
(
elavl1
), HuB (
elavl2
)], which stabilize their mRNA targets and
are known to actively regulate germ cell development; and (
iii
)
Auf1 (
hnrnpd
), which can stabilize or destabilize mRNA and
modulate inflammation in the adult mouse (20).
To determine whether any of the detected AUBPs was regu-
lated by FGF/Erk MAP kinase signaling, we measured changes
in their expression in response to pharmacological inhibitors of
MEK1/2. We discovered that TTP and Brf1 responded strongly
to MEK1/2 inhibition, with mRNA levels down-regulated greater
than twofold after 5 and 10 h (Fig. 1
B
). Brf2, Auf1, and KHSRP
mRNA levels were also slightly down-regulated. Interestingly,
three out of five of these responding genes are members of the
Zfp36 protein family (Fig. 1
B
, red bracket).
Changes in Erk MAP Kinase Signaling Lead to Transient and Sustained
Zfp36 Responses.
We explored the regulatory connection between
Zfp36 AUBPs and FGF/Erk MAP kinase signaling further by
measuring how TTP, Brf1, and Brf2 responded to short and long
periods of MEK1/2 inhibition (Fig. 2
A
). Incubation with MEK1/2
inhibitor resulted in a rapid reduction in TTP, Brf1, and Brf2
Changes in AUBP expression in response
to MEK1/2 inhibitor (CI-1040)
Fold Change in mRNA
B
0.13
0.25
0.50
1.00
2.00
TTP
Brf1
Brf2
Auf1
HuR
HuB
KHSRP
Tia1
Tial1
HnrnpA0
HnrnpA1
HnrnpA2
HnrnpA3
HnrnpC
5 hours
10 hours
A
log
10
(FPKM)
3
2
1
0
-1
4
0
300
100
500
400
200
Genes
Transcript Abundance in E14
mESCs
Pou5f1
Fgf5
Hnf4
Gata4
Sox2
Nanog
Fig. 1.
Erk MAP kinase signaling regulates AUBP expression in mESCs. (
A
) Distribution of average transcript abundances (FPKM) from two total RNA bi-
ological replicates. Highlighted are notable mESC pluripotency and differentiation genes (gray triangles), which were used to distinguish active
ly expressed
genes (mean FPKM
≥
5, 7,194 genes) from the gene expression background (mean FPKM
<
5, 11,119 genes). FPKM
=
5 denoted by red dashed line. Well-
characterized and actively expressed AUPBs are highlighted (orange triangles). (
B
) Profiling changes in AUBP expression using RT-qPCR in response to
a pharmacological inhibitor of MEK1/2 (CI-1040/PD184352) after 5 and 10 h. Members of the Zfp36 family are highlighted in the red bracket.
Strain FD6 (
fgf4
-/-
) R1 mESCs
-+ -+ -+
Fgf4
Heparin
(5 hours)
Fold Change in
mRNA Level
TTP
Brf1
Brf2
1
0.5
2
Brf2
Brf1
TTP
GAPDH
0
10
20
30
Hours
Protein Expression in V6.5 mESCs
aer MEK1/2 inhibion (CI-1040)
(1.00)
(1.09)
(1.90)
(3.60)
(1.00)
(0.26)
(0.28)
(0.14)
(1.00)
(0.69)
(1.25)
(1.27)
Cyto. Nuc.
Cyto. Nuc.
tubulin
TATA bp
TTP
Brf1
Brf2
Control
Fgf4
Heparin
(5 hours)
Strain FD6 (
fgf4
-/-
) R1 mESCs
Hours
0.13
0.25
0.50
1.00
2.00
0102030
Fold Change in mRNA
mRNA Expression in V6.5 mESCs
aer MEK1/2 inhibion (CI-1040)
TTP
Brf1
Brf2
Strain FD6 (
fgf4
-/-
) R1 mESCs
Fgf4
Heparin
(5 hours)
-+
Erk1/2
pThr202/pTyr204
Erk1/2
BC
DEF
CI-1040/
PD184352
Fgf
Raf
MEK
Erk
AUBPs
FgfR
Fgf4
A
Fig. 2.
FGF/Erk MAP kinase signaling regulates the expression of Zfp36 RBPs. (
A
) Cartoon depicting the potential relationship between FGF/Erk MAP kinase
signaling and AUBP expression. Whereas the addition of Fgf4/heparin activates the FGF/Erk MAP kinase pathway, culturing cells with 5
μ
M PD184352/CI-1040
inhibits Erk1/2 activation. (
B
) A 30-h RT-qPCR time course of TTP, Brf1, and Brf2 mRNA level changes in response to MEK1/2 pharmacological inhibitors (
±
SEM;
n
=
3 for all time points), and (
C
) corresponding Western blots. (
D
) Western blot staining for Erk1/2 and phospho-Erk1/2 (Thr202/Tyr204) showing pathway
activation in cells stimulated with 10 ng/mL FGF4 plus 10
μ
g/mL heparin for 15 min. (
E
) Changes in TTP, Brf1, and Brf2 after 5-h incubation with or without
FGF4/heparin as indicated (
±
SEM;
n
=
3). (
F
) Western blot profiling of changes in TTP, Brf1, and Brf2 protein level in
fgf4
−
/
−
mESCs after stimulation with FGF4/
heparin for 5 h. To compare changes in intracellular localization, proteins were harvested sequentially as cytoplasmic (cyto) or nuclear (nuc) frac
tions, with
β
-tubulin and TATA-binding protein serving as localization controls. Also see
Fig. S1
A
–
D
.
Tan and Elowitz
PNAS
|
Published online April 14, 2014
|
E1741
DEVELOPMENTAL
BIOLOGY
PNAS PLUS
mRNA level as gauged by reverse transcription
–
quantitative
PCR (RT-qPCR) (Fig. 2
B
). Brf1, Brf2, and, to a lesser extent,
TTP mRNA level changes were significant within 1 h of inhibitor
treatment (
Fig. S1
A
) and continued to decrease after 7.5 h of in-
hibition (Fig. 2
B
). However, after 10 h, TTP and Brf2 mRNA ex-
pression recovered, whereas Brf1
expression remained suppressed
(Fig. 2
B
). Because the pharmacological inhibitor provides con-
tinuous suppression of Erk MAP kinase signaling (Fig. 2 and
Fig.
S1
B
), these data indicate that TTP and Brf2 mRNA respond
only transiently (
t
<
10 h) to changes in Erk MAP kinase sig-
naling, whereas Brf1 mRNA maintains a sustained response to
the level of Erk MAP kinase signaling.
Protein level changes were also rapid, with a 30% reduction in
TTP and a 50% reduction in Brf1 within 1.5 h of inhibitor
treatment (Fig. 2 and
Fig. S1
C
). However, whereas Brf1 protein
levels continued to fall for the remainder of the time course,
reaching
∼
10-fold less protein by 30 h, TTP protein levels re-
covered and increased above DMSO-treated controls (Fig. 2
C
).
These data indicate that the regulation of TTP protein becomes
distinct from the regulation of TTP mRNA at later times
(compare Fig. 2
B
and
C
). We note that, in other cellular con-
texts, direct phosphorylation of TTP protein by Erk1/2 has been
shown to reduce its stability (10). Furthermore, Zfp36 AUBPs
also contain AU-rich sequences in their own mRNAs, which
enable direct autoregulation and cross-regulation (Fig. 2 and
Fig.
S1
D
). These mechanisms could explain why TTP protein and
mRNA levels respond differently after prolonged MEK1/2 in-
hibition. In contrast, Brf2 protein levels were much less affected,
dropping slightly at 10 h, and then recovering at later time
points. Thus, at both the mRNA and protein levels, Brf2
Brf1
1x
+ Brf1 siRNA
Brf1
4x
Expression of Pluripotency
Factors Relave to Brf1
1x
Fold Difference
0.5
1.0
2.0
Esrrb
Nanog
Pou5f1
Rex1
Sox2
Tbx3
Brf1
1x
Brf1
4x
Brf1
1x
+ Brf1 siRNA
Hours aer LIF removal
Fold Change
Esrrb
Nanog
Pou5f1
Rex1
Sox2
Tbx3
0.13
0.25
0.50
1.00
2.00
02448
02448
02448
02448
02448
02448
0.13
0.25
0.50
1.00
2.00
04896120144
Fold Difference
YFP(+)/YFP(-) of Brf1
4x
relave to Brf1
1x
(Self-Renewal Assay)
Hours aer inial plang
Brf1
Gapdh
Brf1
1x
+ Brf1 siRNA
Brf1
1x
Brf1
4x
(1.00)
(3.98)
(0.23)
Perturbaons to
Brf1 Expression
C
DE
F
0.50
1.00
1.50
0.00
0.25
0.50
0.75
1.00
024487296
Relave to FGF4
expression at t = 0
Fold
Change
Fold Change
Hours aer LIF removal
FGF4
and
FGF5
FGFR1
Spred2
and
Spry2
0.00
0.25
0.50
0.75
1.00
FGFR inhibitor
MEK inhibitor
Untreated
Hours aer LIF removal
(inhibitor added at
t
– 3 hours)
Fold Change
Brf1 mRNA level during differenaon
and its dependence on FGF/Erk
0 2448 7296
A
B
0.00
0.25
0.50
0.75
1.00
Fig. 3.
Brf1 expression compromises mESC self-renewal. (
A
) Changes in Brf1 expression after LIF removal (blue line; light blue bounding boxes at each time
point represent
±
SEM;
n
=
3). The effect of FGF or Erk MAP kinase signaling inhibition on Brf1 expression at each time point is indicated (white and black bars,
resulting from a 3-h inhibitor treatment) (
±
SEM;
n
=
3). (
B
) Changes in FGF4, FGF5, FGFR1, Spred2, and Spry2 mRNA expression after LIF removal (
±
SD;
n
=
2).
(
C
) Brf1 expression was profiled in E14 carrying a CMV-H2B-YFP expression cassette (Brf1
1
×
), Brf1
1
×
treated with Brf1 siRNAs, and E14 carrying a CMV-Brf1/
T2A/H2B-YFP expression cassette (Brf1
4
×
). A Western blot of Brf1 protein shows that siRNA knockdown reduced Brf1 expression by 75% after 36 h, and stable
CMV-driven Brf1 expression increases Brf1 protein levels by fourfold relative to Brf1
1
×
. Chemiluminescent intensities were normalized to Gapdh signal as
loading control. (
D
) Changes in self-renewal of Brf1
4
×
relative to Brf1
1
×
. Differences in proliferation rate were gauged by changes in the ratio of YFP(
+
)
transgene-expressing cells to YFP(
−
) wild-type cells. For
t
=
48 and 96 h (
±
SEM;
n
=
4). For
t
=
120 and 144 h (
±
SEM;
n
=
3). (
E
) Changes in the expression of
core pluripotency genes via RT-qPCR after Brf1 siRNA knockdown in Brf1
1
×
(gray bars) or enhanced Brf1 expression in Brf1
4
×
(blue bars) over 36 h (
±
SEM;
n
=
2). (
F
) Profiling changes in the expression of core pluripotency genes via RT-qPCR during the early stages of differentiation (15% serum, without LIF) in Br
f1
1
×
treated with Brf1 siRNAs (dashed black line), Brf1
1
×
(solid black line), and Brf1
4
×
(solid blue line) (
±
SEM;
n
=
2).
E1742
|
www.pnas.org/cgi/doi/10.1073/pnas.1320873111
Tan and Elowitz
responds more weakly to these perturbations (compare Fig. 2
B
and
C
). These results indicate that AUBP levels respond to
changes in Erk MAP kinase signaling with different kinetics.
To further validate these findings, we checked whether up-
regulating FGF signaling could produce opposite results to in-
hibition (Fig. 2
A
). We added FGF4/heparin to
fgf4
−
/
−
R1 mESCs
(strain FD6), to activate Erk MAP kinase signaling (Fig. 2
D
)
(21). TTP, Brf1, and Brf2 mRNA levels increased within 5 h of
ligand addition, with similar changes at the protein level (Fig. 2
E
and
F
). These changes occurred specifically within the cyto-
plasmic compartment, consistent with a role for these AUBPs in
regulating targeted degradation of mature mRNAs (Fig. 2
F
).
Together, these results indicate Zfp36 protein expression responds
rapidly to both increases and decreases in FGF/Erk MAP kinase
signaling activity, leading to both transient (TTP, Brf2) and sus-
tained (Brf1) regulatory responses.
Enhancing Brf1 Compromises mESC Self-Renewal.
Among the Zfp36
AUBPs, FGF/Erk MAP kinase signaling most strongly regulated
the expression of Brf1 in pluripotent conditions (Fig. 2). During
differentiation, Brf1 regulation was also dynamic and continued
to be regulated by FGF/Erk MAP kinase signaling (Fig. 3
A
). The
approximately twofold reduction in Brf1 expression over 4 d of
LIF withdrawal (Fig. 3
A
) tracked concomitant changes in FGF4
expression (Fig. 3
B
), and the down-regulation of the Erk MAP
kinase target genes Spred2 and Spry2 (Fig. 3
B
).
To determine the functional effect of Brf1 on pluripotent and
differentiating cells, we perturbed Brf1 expression using siRNAs,
which produced an approximately fourfold decrease in Brf1
protein relative to wild type. We also created stable transgene-
mediated overexpression cell lines, which increased Brf1 protein
levels approximately fourfold above wild-type levels (Fig. 3
C
).
For transgene expression in mESCs, clones expressing H2B-YFP
(Brf1
1
×
) or Brf1-T2A-H2B-YFP (Brf1
4
×
) were derived for these
studies. Of these, one control Brf1
1
×
clone that expressed wild-
type levels of Brf1 protein, and one Brf1
4
×
clone expressing
approximately fourfold more Brf1 protein, was chosen for
further analysis.
To quantify changes in mESC self-renewal brought about by
Brf1 overexpression, we cocultured YFP(
+
) Brf1
4
×
clones with
wild-type YFP(
−
) E14 mESCs. In this assay, any change in self-
renewal ability manifests as changes in relative proliferation rate,
and hence, a change in the YFP(
+
)/YFP(
−
) ratio (22). Com-
pared with Brf1
1
×
, cocultures with Brf1
4
×
exhibited a significant
proliferation defect, with the YFP(
+
)/YFP(
−
) ratio reduced by
∼
20% every 48 h (Fig. 3
D
). In these cells, the expression of
several core pluripotency genes is altered (Fig. 3
E
). However,
most remain pluripotent in conditions with LIF plus serum (see
below). Removing LIF rapidly initi
ates differentiation, and during
the first 48 h, Brf1 siRNA knockdown in Brf1
1
×
or Brf1 over-
expression in Brf1
4
×
produced only modest effects on the rate at
which some markers of pluripotency were down-regulated (Fig. 3
F
).
Enhancing Brf1 Expression Accelerates Mesendoderm Differentiation.
In contrast to the mild effect of Brf1 expression on the down-
regulation of pluripotency factors (Fig. 3
F
), the up-regulation of
Mesendoderm
Neurectoderm
Primive Ectoderm
Extra-embryonic
Endoderm
mESCs
(Oct4, Nanog, Rex1)
(Fgf5, Gbx2, Nodal)
(Sox1, Gbx2)
(Dab2, Gata6, Hnf4a)
(T, Gsc, Mixl1, FoxA2)
A
Brf1
1x
+Brf1 siRNA
Brf1
4x
Brf1
4x
+ Brf1 siRNA
B
15% FBS without LIF, 3 days: (Pro
motes Mesendoderm Differenaon)
Fold Difference
10
0
10
2
10
3
10
1
10
-1
10
-2
Dab2
Gata6
Hnf4a
FoxA2
T
Gsc
Mixl1
Wnt3A
Nkx2-5
Tbx6
Twist
Nodal
Fgf5
Gbx2
10
-1
10
1
10
2
10
0
Fold Difference
Dab2
Gata6
Hnf4a
FoxA2
T
Sox1
Fgf5
Gbx2
C
N2B27, 3 days: (Pro
motes Neurectoderm)
D
Brf1
1x
Brf1
4x
N2B27 media
Sox1 Protein, 5 days
50
200
250
300
0
150
100
10
2
10
3
10
5
10
4
Arbitrary Fluorescence Units
Brf1
1x
Brf1
4x
2
°
only
Counts
Brf1
1x
Brf1
4x
50
100
150
200
0
10
2
10
3
10
4
10
5
Counts
15% FBS without LIF media
Brachyury (T) Pr
otein, 3.5 days
Brf1
1x
Brf1
4x
2
°
only
Arbitrary Fluorescence Units
Fig. 4.
Brf1 expression promotes mesendoderm differentiation. (
A
) Cartoon diagram showing the developmental potential of mESCs and associated lineage-
specific markers. (
B
) Brf1
1
×
and Brf1
4
×
were cultured in mesoderm differentiation media (15% serum without LIF) for 3 d. Fold difference in the expression of
several lineage-specific differentiation markers in Brf1
4
×
relative to Brf1
1
×
after 3 d of differentiation (
±
SEM;
n
=
3). (
C
) Brf1
1
×
and Brf1
4
×
were cultured in
neurectoderm differentiation media (N2B27) for 3 d. Fold difference in the expression of several lineage-specific differentiation markers in CMV-
Brf1 E14
relative to control E14 after 3 d of differentiation (
±
SEM;
n
=
3). (
D
)(
Left
) Brachyury immunostain and associated flow cytometry data after 3.5 d of dif-
ferentiation (
n
=
2,000 cells). (
Right
) Sox1 immunostain and associated flow cytometry data after 5 d of differentiation (
n
=
1,500 cells).
Tan and Elowitz
PNAS
|
Published online April 14, 2014
|
E1743
DEVELOPMENTAL
BIOLOGY
PNAS PLUS
differentiation markers is strongly affected by Brf1 (Fig. 4
A
).
After 3 d of LIF withdrawal, we observed a striking bias in gene
expression when comparing Brf1
4
×
to Brf1
1
×
cultures. LIF
withdrawal generally promotes mesoderm differentiation (23).
Indeed, Brachyury (T) was up-regulated in differentiating Brf1
1
×
and Brf1
4
×
cultures. However, Brachyury expression was 100-fold
greater in the Brf1
4
×
cell line (Fig. 4
B
). Transfecting Brf1
1
×
cultures with Brf1 siRNAs produced the opposite effect, down-
regulating Brachyury expression approximately fourfold relative
to untreated controls (Fig. 4
B
). Furthermore, siRNAs against
Brf1 could attenuate the up-regulation of Brachyury in Brf1
4
×
,
indicating that this regulation resulted specifically from Brf1
overexpression. In agreement with these results, flow cytometry
profiling indicated that a larger fraction of Brf1
4
×
cells expressed
Brachyury protein by 84 h compared with Brf1
1
×
(Fig. 4
D
).
These findings were further supported by the up-regulation of
mesendodermal markers Goosecoid (Gsc), Mixl1, and Wnt3A
(Fig. 4
B
), indicating that Brf1 accelerated commitment to mes-
endodermal fates. Ectoderm markers (Nodal, Fgf5, and Gbx) were
not affected, whereas extraembryonic and definitive endoderm
markers (Gata6, Hnf4a, and FoxA2) showed weaker basal ex-
pression levels that responded differentially to Brf1 (Fig. 4
B
).
Brf1 expression did not influence neural differentiation. Be-
cause serum inhibits neural differentiation (24), we cultured cells
in N2B27 serum-free media without LIF and BMP4. After 3 d in
this media, most markers of differentiation appeared to be un-
affected by Brf1 (Fig. 4
C
). For example, Sox1 mRNA and pro-
tein were readily detected, but its expression levels were similar
in Brf1
1
×
and Brf1
4
×
cultures (Fig. 4
C
and
D
). Interestingly,
even in N2B27, the basal expression of Brachyury was up-regu-
lated in a Brf1-dependent manner. Thus, Brf1 appears to mainly
affect mesendodermal differentiation pathways.
Brf1 Binds Pluripotency-Associated mRNAs.
To better understand
the mechanistic basis for these developmental effects, we pro-
filed possible Brf1 mRNA targets in mESCs. To determine
which actively expressed genes are bound by Brf1, we adapted
a previously developed RNA immunoprecipitation sequencing-
based (RIPseq) assay that could selectively enrich target mRNAs
using an affinity-purified polyclonal antibody against Brf1 (Fig.
5
A
;
Materials and Methods
) (25). In parallel, we performed a
negative control using a nonspecific rabbit IgG. RNA from sam-
ples and controls were then processed and analyzed using high-
throughput sequencing.
To provide a quantitative measure of antibody-mediated en-
richment, we computed a statistic, denoted
E
RIP
for each actively
expressed transcript (
Materials and Methods
).
E
RIP
represents the
amount of mRNA coprecipitated with Brf1 protein over non-
specific background levels (
Materials and Methods
). Genes with
high
E
RIP
values were more likely to have AU-rich elements
(AREs) in their 3
′
-UTR (Fig. 5
B
). For example, considering the
transcripts most enriched by Brf1 immunoprecipitation (positive
outliers,
E
RIP
>
1.226, 418 genes), 25.1% contained the minimal
full consensus ARE and 60.0% contained the minimal partial
consensus ARE. These percentages represent a threefold to
fourfold increase in ARE abundance relative to their frequency
among all protein coding genes (
Table S1
and
Datasets S2
–
S4
).
Moreover, several of the most highly enriched target genes were
previously characterized as direct targets of Zfp36 proteins (e.g.,
Ier3, Mllt11, and Pim3), including Zfp36 proteins themselves
(26, 27). Interestingly, based on our definition of the minimal
ARE element, many highly enriched target genes do not contain
consensus AREs. However, the existence of noncanonical (al-
though still poorly characterized) AU-rich sequences has been
documented and could explain the enrichment of these mRNAs
(28). Thus, the RIPseq assay can selectively enrich for mRNAs
containing AREs.
Several pluripotency-associated factors were detected in the
Brf1-RIP fraction, potentially explaining the developmental effects
of Brf1 overexpression. For example, the core pluripotency regu-
lators Nanog (
E
RIP
=
0.58) and Klf2 (
E
RIP
=
7.15) were both
within the top quartile of enriched targets. Nanog broadly inhibits
mESC differentiation, and its ex
pression is reduced as cells lose
pluripotency and commit to extraembryonic and somatic cell line-
ages in culture (29, 30). Klf2, along with Klf4 and Klf5, inhibits
mesendoderm differentiation. Knockdown of Klf factors up-
regulates primitive streak markers, as well as Cdx2, a gene
expressed in trophectoderm and extraembryonic mesoderm (31).
Also consistent with a role for Brf1 in promoting mesen-
doderm, the pluripotency factors Kdm4c (
E
RIP
=
0.63) and Zfp143
(
E
RIP
=
0.99) were enriched in the RIPseq assay. Knockdown of
the lysine methyl-transferase K
dm4c is known to up-regulate mes-
endoderm and extraembryonic mesoderm markers (32). Zfp143
coordinates with Oct4 to transcriptionally activate Nanog. siRNA
knockdown of Zfp143 rapidly initiates differentiation and pro-
motes the expression of Fgf5, Cdx2, and Cdh3, which are expressed
in trophectoderm and cells that commit to extraembryonic me-
soderm (33). Understanding the regulation of these mRNAs may
provide additional mechanistic insights into the Brf1-dependent
control of gene expression in mESCs.
Brf1 Binds Nanog mRNA in Vitro.
To corroborate these RIPseq
results, we assayed for direct binding of Brf1 to an enriched
mRNA, in this case, Nanog (Fig. 6
A
). Previous work has shown
that Nanog is strongly regulated by FGF/Erk MAP kinase sig-
naling (34), and these effects are mediated, in part, by direct
regulation of the Nanog promoter (35). Posttranscriptional reg-
ulation by Brf1 would provide an alternative mechanism to re-
press Nanog expression.
We conducted a protein pull-down assay using RNA as bait
(Fig. 6
B
,
Left
). Wild-type RNA and variants with ARE sequence
mutations were expressed in vitro and hybridized at their 3
′
ends
to DNA oligonucleotides coupled to magnetic microbeads. These
RNA
–
microbead conjugates were then incubated with crude cyto-
plasmic protein extracts, and all proteins capable of binding
hybridized RNAs were magnetically isolated and purified for
further analysis.
We used the 3
′
-UTR sequence of IL-2 as a positive control,
because it contains clusters of ARE sequences that are bound
and regulated by Zfp36 AUBPs (
Fig. S2
A
) (36). Western blots
showed that Brf1 could be purified from mESC lysates using
a conjugated wild-type IL-2 sequence, but not using a mutant
IL-2 lacking known AREs (Fig. 6
B
). Two Brf1 bands of different
sizes were detected, possibly indicating the purification of dif-
ferent posttranslationally modified forms of Brf1 (Fig. 6
B
).
We next repeated the assay using Nanog mRNA as bait. The
3
′
-UTR of Nanog mRNA is
∼
1 kb and contains three potential
ARE elements: one full consensus (site 1 in Fig. 6
A
) and two
partial nonconsensus sequences (sites 2 and 3 in Fig. 6
A
).
Western blotting of these protein pull downs indicated that Brf1
bound to wild-type Nanog mRNA (Fig. 6
B
). Mutating the full-
consensus ARE (site 1) significantly reduced Brf1 binding. Re-
moving the remaining two partial nonconsensus AREs (sites 2
and 3) did not appear to further reduce the Brf1 signal. In
contrast, the presence or absence of AREs did not affect the
binding of other RNA-binding proteins (RBPs). For example,
addition of an androgen receptor 3
′
-UTR sequence, which
contains a poly(C) RNA-binding protein 1 (PCBP1) site, to
Nanog mRNA permitted isolation of PCBP1 protein (
Fig. S2
B
).
This binding was not affected by the presence or absence of
Nanog
’
s AREs. In contrast, a Nanog mRNA containing only
a 120-nt poly(A) without a PCBP1 site did not bind PCBP1
protein. Together, these results confirm that Brf1 binds specifi-
cally to Nanog mRNA in an ARE-dependent manner.
E1744
|
www.pnas.org/cgi/doi/10.1073/pnas.1320873111
Tan and Elowitz
FGF/Erk MAP Kinase Signaling Regulates Nanog mRNA Half-Life
Through its AREs.
To understand how Brf1 binding impacts
Nanog expression, we measured the Nanog mRNA half-life, with
or without AREs, and with or without FGF signaling, in
fgf4
−
/
−
mESCs (Fig. 6
C
). We determined that the half-life of Nanog
mRNA is 2.5
±
0.4 h (
±
SEM) without FGF4/heparin and 1.5
±
0.3 h (
±
SEM) with Fgf4/heparin (Fig. 6
C
,
Top
). However,
removing all 3
′
-UTR ARE sites [
Δ
ARE (1
–
3)] protected Nanog
from increased degradation by FGF signaling (Fig. 6
C
,
Middle
),
indicating that FGF destabilizes Nanog mRNA through its
AREs. Additionally, when we analyzed H2B-YFP mRNA as a
negative control, we observed no effect of FGF signaling on half-
life (Fig. 6
C
,
Bottom
).
Unexpectedly, mutant Nanog mRNA [
Δ
ARE (1
–
3)] exhibited
a shorter half-life than wild type (2 h instead of 2.5 h; Fig. 6
C
). In
the absence of FGF, we would have expected that the half-life of
mutant and wild-type mRNAs to be the same. However, it is
possible that Nanog
’
s AREs are subject to regulation by other
protein factors that have different effects. For example, AREs
could associate with stabilizing AUBPs, such as HuR, whose
activity is independent of FGF/Erk MAP kinase signaling in
mESCs (Fig. 1
C
). Primary sequence changes may have also af-
fected mRNA stability independent of any particular AUBP.
Despite this difference, the above data clearly show that Nanog
’
s
AREs are necessary for any posttran
scriptional regulation by FGF.
To show that FGF regulates Nanog through Brf1, we con-
ducted an epistasis test in
fgf4
−
/
−
R1 mESCs. We transfected
mock or Brf1 siRNAs, and measured changes in Brf1 and Nanog
mRNA level both in the presence and absence of FGF4/heparin.
Relative to siRNA control (Fig. 6
D
, first column), Brf1 siRNAs
caused a slight (
∼
10%) down-regulation of Brf1 and a corre-
sponding increase in Nanog mRNA levels (Fig. 6
D
, second col-
umn). These data indicate that Brf1 is expressed, albeit at a
lower level, even when FGF signaling is absent (
Fig. S2
C
) and
that alternative pathways support its basal expression. We con-
firmed that FGF signaling remains the dominant regulator of
Erk in mESCs (Fig. 2
D
) and that Erk MAP kinase signaling is
the main driver of Brf1 expression (
Fig. S2
C
).
Addition of FGF4/heparin ligand increased Brf1 and de-
creased Nanog by greater than twofold within 5 h (Fig. 6
D
, third
column). In agreement with its role as a regulatory intermediate,
the presence of Brf1 siRNAs reduced this regulation, yielding
a smaller up-regulation of Brf1 and a smaller down-regulation of
Nanog (Fig. 6
D
). We note that the inability of Brf1 siRNAs to
fully block the down-regulation of Nanog can be partly explained
by its limited knockdown efficiency (
∼
75%) (Fig. 3
C
) but could
also reflect Brf1-independent regulatory mechanisms.
FGF/Erk MAP kinase signaling thus regulates the dynamic
expression of Zfp36 AUBPs in mESCs, which in turn affect the
stability of key mRNA targets. In particular, we show that Brf1
(
zfp36l1
) has the capacity to broadly impact the regulation of
core pluripotency-associated transcription factors, down-regulate
self-renewal, and promote lineage-specific commitment during
differentiation. In this way, Brf1 provides a specific molecular
link between FGF/Erk MAP kinase signaling and the regulation
of gene expression in mESCs.
Discussion
AUBPs modulate developmental gene expression in response to
intercellular and intracellular signals. Some specific examples of
this regulation include the repression of Notch1 in response to
PI3K/Erk MAP kinase signals during T-cell development (17),
Enrichment of Acvely Expressed mRNAs aer
Brf1 anbody mediated RIP
A
B
Raw Sequencing
Reads
Uniquely Mapped
Reads
Total RNA 1
Total RNA 2
Brf1/2 RIP Sample 1
Brf1/2 RIP Sample 2
Rabbit IgG RIP Control 1
Rabbit IgG RIP Control 2
41,504,591
31,255,823
37,909,362
27,965,675
23,262,834
38,725,737
31,195,095
22,623,281
7,075,298
5,157,037
2,244,609
3,910,793
RNA
sequencing
Mapping
Bowe
Protein Coding
Transcript
Annotaon
Analysis:
eXpress
Cytoplasmic
Protein/RNA
extracon
Immunoprecipitaon
of RNA Binding
Proteins with mRNAs
RNA extracon
1
2
3
4
5
6
RIP-Sequencing Method
-2
-1
0
1
2
3
4
5
0
50
100
150
200
250
Percenle (x) of
E
E
R
I
P
Values
Full ARE
Paral
ARE
genes
x ≥ 89 %le (Posive Outliers)
25.1%
60.0%
418
x ≥ 75 %le (Top Quarle)
13.8%
38.9%
1798
75 %le ≥ x ≥ 25 %le
5.8%
18.8%
3597
x ≤ 25 %le (Boom Quarle)
6.7%
20.5%
1799
Oct4
Genes
E
RIP
Fig. 5.
Brf1 binds other pluripotency-associated mRNAs. (
A
)Methodused
for RNA
–
RBP immunoprecipitation and RNA sequencing (RIPseq). Sequencing
statistics for two total RNA samples, two Brf1/2 antibody RIP-derived samples,
and two rabbit IgG RIP controls are presented. (
B
) Distribution of
E
RIP
values
for actively expressed genes (
n
=
7,194 genes). The location of several notable
pluripotency associated transcripts is highlighted. Box plot statistics: Median
(red line,
E
RIP
=
0.17), lower quartile boundary (
E
RIP
=
−
0.04), upper quartile
boundary (
E
RIP
=
0.47), and statistical outliers [median
±
1.5
×
(upper quartile
–
lower quartile)]. Table: Frequency of full ARE motifs (-UUAUUUAUU-) and
partial ARE motifs (-UAUUUAU-) among genes classified as
E
RIP
outliers or in
different
E
RIP
quartiles.
Tan and Elowitz
PNAS
|
Published online April 14, 2014
|
E1745
DEVELOPMENTAL
BIOLOGY
PNAS PLUS
repression of TNF-
α
in response to p38/Erk MAP kinase signals
during inflammation (37), and the stabilization of p21 in re-
sponse to ATR/ATM kinase activation after DNA damage (38).
Here, we demonstrate AUBPs mechanistically connect FGF/Erk
MAP kinase signaling to the regulation of pluripotency, self-re-
newal, and differentiation in mESCs. More specifically, control
over Brf1 expression rapidly regulates the expression of key
pluripotency-associated genes, reduces the capacity to self-renew,
and enhances mesendoderm differentiation upon LIF withdrawal.
Why has Brf1 been selected for implementing FGF-dependent
cellular responses in mESCs? One possibility is that, owing to its
rapid transcriptional response to FGF signaling and its short
protein and mRNA half-life (
∼
1.5 and
∼
1 h, respectively), Brf1 is
capable of tracking dynamic changes in Erk MAP kinase activity.
Brf1 directly affects mRNA abundance and provides stem cell
populations with a mechanism to quickly respond to changes
in FGF signaling, without necessarily altering underlying tran-
scriptional states. However, the benefit of these dynamical prop-
erties to the biology of pluripotent or differentiating mESCs still
remains unclear. Brf1 also provides a regulation that is similar to
miRNAs, which also provide mRNA-
level repression while main-
taining flexibility in target selection. Interestingly, Zfp36 AUBPs
and miRNAs are known to cooperate in regulating some mRNA
targets (39), indicating a potential point of convergence between
these two regulatory mechanisms. Future work will address why
this system is particularly well adapted to serve as a regulator in
mESCs, and as a mediator of FGF/Erk MAP kinase signaling.
Whether Brf1 plays a similar role in the embryo is unclear. We
note that the regulatory effects of Brf1 in cell culture mimic the
developmental response to FGF/Erk MAP kinase signaling at
different stages of embryonic development. For example, FGF4
signaling promotes extraembryonic endoderm differentiation
in the inner cell mass partly by destabilizing the expression of
pluripotency genes (40). Brf1 may participate in this process by
repressing Nanog and other pluripotency regulators. At a slightly
later developmental stage, and one that is more relevant to
mESCs, both FGF4 and FGF8 are required for mesoderm in-
duction (41, 42). The expression of Brachyury, a regulator of
mesoderm morphogenesis, is enhanced by FGF4 and FGF8
signaling through unknown mechanisms (43). Our results show
that Brf1 expression during differentiation in mESCs similarly
enhances Brachyury and other primitive streak markers. This
regulatory connection could explain why Brf1 knockout mice
also exhibit the same gross defects in chorio-allantoic fusion,
neural tube closure, and placental organization at midgestation
(E11) as FGFR2 knockouts, an indication of shared regulation and
function (9, 44). Although the regulation of FGF/Erk MAP kinase
signaling likely differs between the embryonic and cell culture
context, these observations implicate Brf1 as an important FGF/Erk
MAP kinase-inducible regulator of development in both systems.
Materials and Methods
RT-qPCR Experiments.
Total RNA was harvested from samples and controls
using Qiazol Reagent (Qiagen) or an RNeasy Mini Kit (Qiagen). RT of iso-
Δ
Δ
ARE sites 1, 2 and 3
wildtype
Δ
ARE cluster 1
Δ
ARE cluster 1 and 2
wildtype
Δ
ARE site 1
Brf1
3’UTR
AAAAAAA
m
7
G
1
2
3
Nanog CDS
1
2
3
Nanog ARE
Elements
Nanog mRNA
A
BC
No Treatment
+ Fgf4/Heparin
fgf4-/-
mESCs
Relave Expression
Nanog mRNA
0.1
1.0
0123
Nanog
Δ
ARE (1,2,3) mRNA
0.1
1.0
0123
H2B-YFP mRNA
Hours
0.1
1.0
0246
Wildtype Nanog mRNA half-life:
(with FGF) :
1.5 ± 0.3 hours
(without FGF) :
2.5 ± 0.4 hours
p= 0.029
D
ARE
mutant
WT
Protein Pull-Down
Assay
RBP
0.25
0.50
1.00
2.00
Fold Change in mRNA
Brf1 knockdown compromises
FGF-dependent downregulaon of Nanog
FGF4/Heparin
Control siRNA
Brf1 siRNA
─
+
─
─
─
+
+
+
─
+
─
+
Brf1
Nanog
Fig. 6.
Nanog is bound and regulated by Brf1. (
A
) Cartoon illustration of Nanog mRNA indicating the relative location and sequence of AREs in the 3
′
-UTR.
ARE mutants were generated by changing the core (-ATTTA-) motif to (-AGGGA-). (
B
)(
Left
) Schematic diagram of the assay used to extract RBPs from crude
protein extracts. Beads are conjugated to mRNAs with or without AREs. (
Right
) Western blot for Brf1 protein, present in IL-2 and Nanog mRNA protein
isolates. ARE site mutations reduced the extraction of Brf1 from crude protein lysates for both IL-2 sequence control and Nanog mRNA. For more informa
tion
on IL-2 sequence, see
Fig. S2
A
. For more information on assay-specific controls, see
Fig. S2
B
.(
C
) Changes in mRNA half-life were profiled in the presence (blue
data points) or absence (black data points) of FGF signaling in
fgf4
−
/
−
mESCs for wild-type Nanog mRNA (
Top
), mutant Nanog mRNA with deleted AREs [
Δ
ARE
(1
–
3)] (
Middle
), and H2B-YFP mRNA (
Bottom
)(
±
SEM;
n
=
4). The red dashed lines indicate one-half of the initial mRNA concentration. (
D
) Transfection of Brf1
siRNAs can compromise the regulation of Nanog by FGF4/heparin in
fgf4
−
/
−
R1 mESCs (
±
SD;
n
=
2). See also
Fig. S2
C
.
E1746
|
www.pnas.org/cgi/doi/10.1073/pnas.1320873111
Tan and Elowitz
lated mRNA into cDNA was accomplished using iScript cDNA Synthesis Kit
(Bio-Rad). A single qPCR reaction was composed of 0.5
μ
L of cDNA, primers or
primers with probes, and qPCR reaction mix (diluted to a final volume of
10
μ
L). For qPCR experiments using TaqMan/hydrolysis probes (5
′
dye: fluo-
rescein amidite; 3
′
quencher: Zen/Iowa Black FQ), cDNAs were profiled
with SsoFast Probes Supermix Reagent (Bio-Rad) using the manufacturer-
recommended protocol. In brief, we used a two-step thermocycling protocol
(an initial 30-s 95 °C melt, followed by 40 cycles of 5-s 95 °C melt and 10-s 60 °C
anneal/extend). For mRNA half-life qPCR experiments using primers only,
cDNAs were profiled with SsoFast EvaGreen Supermix Reagent (Bio-Rad)
using the manufacturer-recommended protocol. In brief, we used a two-step
thermocycling protocol (initial 30-s 95 °C melt, followed by 40 cycles of 5-s 95 °C
melt and 10-s 55 °C anneal/extend), terminating with a postamplification
melt curve analysis (initialized at 60 °C, and increased at 0.5 °C increments
every 10 s to 95 °C). All measurements were made using a Bio-Rad CFX96
Real-Time PCR System (Bio-Rad). See
primer and probe characterization
(
Table S2
).
Cell Lines and Cell Culture.
fgf4
−
/
−
mESCs (strain FD6) were a kind gift from
Dr. Angie Rizzino (University of Nebraska Medical Center, Omaha, NE).
Cultures were routinely passaged in complete ES culture medium [15%
(vol/vol) FBS, 1,000 U/mL LIF, nonessential amino acid (NEAA), sodium py-
ruvate, and
β
-mercaptoethanol (
β
ME) in DMEM] in the absence of feeders.
Mesoderm differentiation media contained the following: 15% (vol/vol) FBS,
NEAA, sodium pyruvate, and
β
ME in DMEM (23); neurectodermal differen-
tiation media contained the following: N2B27 serum free media (45).
Reagents, Antibodies, Signaling Inhibitors, and siRNAs.
Reagents included the
following: Qiazol reagent (Qiagen); iScript Kit (Bio-Rad); SsoFast Probes
Supermix (Bio-Rad).
Antibodies included the following: mouse monoclonal (L34F12) anti-p44/
p42 (Cell Signal; catalog #4696; 1:2,000); rabbit monoclonal (D13.14.4E) anti-
phospho-p44/p42 (Cell Signal; catalo
g #4370; 1:2,000); mouse anti-TBP
(Abcam; catalog #ab818
; 1:1,000); rabbit anti
–
β
-Tubulin (Abcam; catalog
#ab6046; 1:1,000); rabbit polyclonal anti-Brf1/2 (Cell Signal; catalog #2119;
1:1,000); rabbit anti-Zfp36 (Protein Tech Group; catalog #12737-1-AP; 1:500);
rabbit anti-hnRNP E1 (Cell Signal; catalog #8534; 1:500); goat anti-Brachyury
(R&D Systems; catalog #AF2085; 1:200); rabbit anti-Sox1 (GeneTex; catalog
#GTX62974; 1:200); mouse anti-Gapdh (Abcam; catalog #ab8245; 1:5,000).
Signaling inhibitors included the following: CI-1040 (also known as
PD184352; Axon; 5
μ
M); PD173074 (Sigma; 100 ng/mL).
siRNAs included the following: TTP siRNA (Santa Cruz Biotechnology; 10
nM); Brf1 siRNA (Santa Cruz Biotechnology; 10 nM); Brf2 siRNA (Santa Cruz
Biotechnology; 10 nM); All Stars Negative Control siRNA (Qiagen; 10 nM).
Measurement of mRNA Half-Life.
mESC cultures were cotransfected with
a reverse Tet (rTet) expression plasmid (PGK-H2B-mCherry/T2A/rTet) and
either CMV(2xTetO)-H2B-YFP or CMV(2xTetO)-Nanog with and without ARE
site mutations at 19:1 ratio by mass. Culture media was then transitioned to
media containing FGF4 and heparin (10 ng/mL and 10
μ
g/mL, respectively) or
PBS to determine the regulatory effect of ERK MAP kinase signaling in
fgf4
−
/
−
cells
. To stop transcription from the CMV-TO promoter, doxycycline was
added to a concentration of 1
μ
g/mL to permit binding of rTet to TetO
sequences. Changes in the abundance of H2B-YFP or Nanog mRNA relative
to Gapdh, Sdha, and Tbp housekeeping genes was measured using RT-qPCR.
To distinguish from endogenous Nanog mRNA, we developed a primer set
that specifically recognizes the 5
′
-UTR of Nanog expressed from the
CMV(2
×
TetO) promoter.
Isolation of RBPs from Crude Cytoplasmic Lysates.
IL-2 and Nanog RNAs were
produced in vitro using T7 Ampliscribe (Epicentre Technologies). These RNAs
were then hybridized to biotin-DNA oligonucleotides at their 3
′
end. Two
hundred picomoles of IL-2 or Nanog RNA were used for 250 pmol of biotin-
DNA oligonucleotide. Hybridizatio
n reactions were added to 1 mg of
streptavidin magnetic agarose beads (New England Biolabs). Crude cyto-
plasmic extracts used for protein pull-down assays were obtained using
NE-PER reagents (Thermo Scientific). RNA/DNA
–
bead conjugates were in-
cubated in crude cytoplasmic extract for 1 h at 4 °C, washed five times with
Binding Buffer (20 mM Tris
·
HCl, pH 7.5, 50 mM KCl, 1 mM EDTA, 1 mM DTT,
0.5% Triton X-100 with RNase inhibitors) and incubated in High Salt Elution
Buffer (Binding Buffer plus 1 M NaCl) to collect RNA-bound protein frac-
tions. Protein pull downs were analyzed via Western blot.
Preparation of Cells for Flow Cytometry.
Cultures were harvested with 0.25%
trypsin-EDTA, dispersed into a cell suspension, and added to an equal volume
of 4% formaldehyde. Cells were fixed for 5 min, quenched, and gently
pelleted at 500
×
g
for 30 s. Supernatants were removed and cell pellets
were resuspended in 2.5 mg/mL BSA in 1
×
HBSS. The resulting cell suspen-
sions were incubated overnight at 4 °C before flow cytometry analysis. For
antibody-stained cell suspensions, cell pellets containing fixed cells were
resuspended in 10% FBS in PBS to block, gently pelleted, and stained with
appropriate primary and secondary an
tibodies at 4 °C. Before flow cytometry
analysis, cells were pelleted and resuspended in 2.5 mg/mL BSA in 1
×
HBSS.
All samples, stained and unstained, were analyzed with a Miltenyi Biotec
VYB flow cytometer. Compensation and background correction were
applied postacquisition.
RNA/RBP Immunoprecipitation and RNA Sequencing.
RBP
–
mRNA complexes
were isolated using a Magna-RIP kit (Millipore). Briefly, cytoplasmic extract
from
∼
1
×
10
7
E14 mESCs was distributed equally among two samples and
two controls. For sample reactions, 5
μ
gof
α
-Brf1/2 antibody was used for 50
μ
L of magnetic protein A/G beads. For control reactions, 5
μ
g of rabbit IgG
with no immunoreactivity was used for 50
μ
L of magnetic protein A/G beads.
After stringency washes and proteinase K digestion, RNA was isolated using
Qiazol reagent.
RNA/RBP immunoprecipitation (RIP)-purified RNAs and total RNA from E14
mESCs were prepared for sequencing using a TruSeq RNA Sample Prep kit.
RNAs were fragmented to generate lengths of
∼
200 nt, reverse transcribed
with random hexameric primers to generate double-stranded DNA, blunted,
adenylated, and ligated to Illumina sequencing adapters (150 bp). DNA
fragments were gel separated, and all fragments running at 350 bp were
extracted and amplified. Amplified DNA fragments were then sequenced
using an Illumina HiSeq2000.
RNA-Sequencing Analysis Tools and Methods.
Raw sequencing reads were
trimmed (of 13 nt from 5
′
end) before Bowtie mapping using a mouse
transcript annotation containing only protein coding genes (18,313 genes),
derived from NCBI37/mm9 genome build. Mapping statistics were generated
using eXpress (46). For enrichment analysis, fragments per kilobase exon
permillionmappedreads(FPKM)wereusedasameasureoftranscript
abundance (47).
We computed a statistic (
E
RIP
,
n
) that represents the degree to which the
abundance of the
n
th transcript is enriched by antibody-mediated RIP. This
statistic uses differences in a transcript
’
s abundance after Brf1/2 antibody RIP
(mean FPKM
BRF,
n
) compared with rabbit IgG RIP (mean FPKM
IgG,
n
), nor-
malized by their initial abundance in total RNA before RIP (mean FPKM
Total,
n
).
This difference was further normalized by a penalty factor (
P
), which
accounts for a transcript
’
s tendency to be nonspecifically purified, and is thus
a saturating function of transcript abundance in the IgG control experiment:
E
RIP
,
n
=
A
BRF
,
n
−
A
IgG
,
n
P
,
where
A
BRF
,
n
=
mean FPKM
BRF,
n
mean FPKM
Total,
n
,
A
IgG
,
n
=
mean FPKM
IgG,
n
mean FPKM
Total,
n
,
P
=
median
A
IgG
+
A
IgG
,
n
:
Although
A
IgG
,
n
captures the level of nonspecific association of a transcript
with assay components (i.e., protein A/G beads, rabbit IgG, etc.), nonspecific
association of transcripts with immunoprecipitated RBP/RNA complexes
could not be independently quantified, and could contribute a background
to these
E
RIP
,
n
values.
High-Throughput Sequencing Data.
Raw sequencing data discussed in this
publication were deposited in the National Center for Biotechnology In-
formation Gene Expression Omnibus database (accession no. GSE40104).
ACKNOWLEDGMENTS.
We thank Dr. Kathrin Plath (University of California,
Los Angeles) and Dr. Angie Rizzino (University of Nebraska Medical Center)
for the kind donation of cell culture reagents; Dr. Azim Surani (University of
Cambridge) for fruitful discussions; Rochelle Diamond, Diana Perez, and Josh
Verceles from the Caltech Flow Cytometry Facility; Igor Antoshechkin and
Vijaya Kumar from the Millard and Muriel Jacobs Genetics and Genomics
Laboratory at Caltech; Leah Santat, Yaron Antebi, Joe Markson, James
Tan and Elowitz
PNAS
|
Published online April 14, 2014
|
E1747
DEVELOPMENTAL
BIOLOGY
PNAS PLUS
Linton, Pierre Neveu, John Yong, Zakary Singer, Julia Tischler, Joe Levine,
and Sandy Nandagopal for fruitful discussions. This work was supported by
a Human Frontiers Science Program Grant (RGP0020/2012), the Weston
Havens Foundation, and the David and Lucille Packard Foundation. F.E.T.
was supported by the National Defense Science and Engineering Graduate
Research Fellowship.
1. Colegrove-Otero LJ, Minshall N, Standart N (2005) RNA-binding proteins in early
development.
Crit Rev Biochem Mol Biol
40(1):21
–
73.
2. Kwon SC, et al. (2013) The RNA-binding protein repertoire of embryonic stem cells.
Nat Struct Mol Biol
20(9):1122
–
1130.
3. Glisovic T, Bachorik JL, Yong J, Dreyfuss G (2008) RNA-binding proteins and post-
transcriptional gene regulation.
FEBS Lett
582(14):1977
–
1986.
4. Okano HJ, Darnell RB (1997) A hierarchy of Hu RNA binding proteins in developing
and adult neurons.
J Neurosci
17(9):3024
–
3037.
5. Baou M, Norton JD, Murphy JJ (2011) AU-rich RNA binding proteins in hematopoiesis
and leukemogenesis.
Blood
118(22):5732
–
5740.
6. Wiszniak SE, Dredge BK, Jensen KB (2011) HuB (elavl2) mRNA is restricted to the germ
cells by post-transcriptional mechanisms including stabilisation of the message by
DAZL.
PLoS One
6(6):e20773.
7. Chi MN, et al. (2011) The RNA-binding protein ELAVL1/HuR is essential for mouse
spermatogenesis, acting both at meiotic and postmeiotic stages.
Mol Biol Cell
22(16):
2875
–
2885.
8. Katsanou V, et al. (2009) The RNA-binding protein Elavl1/HuR is essential for placental
branching morphogenesis and embryonic development.
Mol Cell Biol
29(10):2762
–
2776.
9. Stumpo DJ, et al. (2004) Chorioallantoic fusion defects and embryonic lethality re-
sulting from disruption of Zfp36L1, a gene encoding a CCCH tandem zinc finger
protein of the Tristetraprolin family.
Mol Cell Biol
24(14):6445
–
6455.
10. Bourcier C, et al. (2011) Constitutive ERK activity induces downregulation of triste-
traprolin, a major protein controlling interleukin8/CXCL8 mRNA stability in mela-
noma cells.
Am J Physiol Cell Physiol
301(3):C609
–
C618.
11. Amit I, et al. (2007) A module of negative feedback regulators defines growth factor
signaling.
Nat Genet
39(4):503
–
512.
12. Lanner F, Rossant J (2010) The role of FGF/Erk signaling in pluripotent cells.
De-
velopment
137(20):3351
–
3360.
13. Nichols J, Smith A (2011) The origin and identity of embryonic stem cells.
De-
velopment
138(1):3
–
8.
14. Hamazaki T, Kehoe SM, Nakano T, Terada N (2006) The Grb2/Mek pathway represses
Nanog in murine embryonic stem cells.
Mol Cell Biol
26(20):7539
–
7549.
15. Ying QL, et al. (2008) The ground state of embryonic stem cell self-renewal.
Nature
453(7194):519
–
523.
16. Dailey L, Ambrosetti D, Mansukhani A, Basilico C (2005) Mechanisms underlying dif-
ferential responses to FGF signaling.
Cytokine Growth Factor Rev
16(2):233
–
247.
17. Hodson DJ, et al. (2010) Deletion of the RNA-binding proteins ZFP36L1 and ZFP36L2
leads to perturbed thymic development and T lymphoblastic leukemia.
Nat Immunol
11(8):717
–
724.
18. Lu JY, Schneider RJ (2004) Tissue distribution of AU-rich mRNA-binding proteins in-
volved in regulation of mRNA decay.
J Biol Chem
279(13):12974
–
12979.
19. Doller A, Pfeilschifter J, Eberhardt W (2008) Signalling pathways regulating nucleo-
cytoplasmic shuttling of the mRNA-binding protein HuR.
Cell Signal
20(12):2165
–
2173.
20. Barreau C, Paillard L, Osborne HB (2005) AU-rich elements and associated factors: Are
there unifying principles?
Nucleic Acids Res
33(22):7138
–
7150.
21. Wilder PJ, et al. (1997) Inactivation of the FGF-4 gene in embryonic stem cells alters
the growth and/or the survival of their early differentiated progeny.
Dev Biol
192(2):
614
–
629.
22. Ivanova N, et al. (2006) Dissecting self-renewal in stem cells with RNA interference.
Nature
442(7102):533
–
538.
23. Nishikawa SI, Nishikawa S, Hirashima M, Matsuyoshi N, Kodama H (1998) Progressive
lineage analysis by cell sorting and culture identifies FLK1
+
VE-cadherin
+
cells at
a diverging point of endothelial and hemopoietic lineages.
Development
125(9):
1747
–
1757.
24. Ying QL, Nichols J, Chambers I, Smith A (2003) BMP induction of Id proteins suppresses
differentiation and sustains embryonic stem cell self-renewal in collaboration with
STAT3.
Cell
115(3):281
–
292.
25. Keene JD, Komisarow JM, Friedersdorf MB (2006) RIP-Chip: The isolation and iden-
tification of mRNAs, microRNAs and protein components of ribonucleoprotein com-
plexes from cell extracts.
Nat Protoc
1(1):302
–
307.
26. Lai WS, Parker JS, Grissom SF, Stumpo DJ, Blackshear PJ (2006) Novel mRNA targets
for tristetraprolin (TTP) identified by global analysis of stabilized transcripts in TTP-
deficient fibroblasts.
Mol Cell Biol
26(24):9196
–
9208.
27. Tchen CR, Brook M, Saklatvala J, Clark AR (2004) The stability of tristetraprolin mRNA
is regulated by mitogen-activated protein kinase p38 and by tristetraprolin itself.
J Biol Chem
279(31):32393
–
32400.
28. Sarkar B, Xi Q, He C, Schneider RJ (2003) Selective degradation of AU-rich mRNAs
promoted by the p37 AUF1 protein isoform.
Mol Cell Biol
23(18):6685
–
6693.
29. Chambers I, et al. (2003) Functional expression cloning of Nanog, a pluripotency
sustaining factor in embryonic stem cells.
Cell
113(5):643
–
655.
30. Mitsui K, et al. (2003) The homeoprotein Nanog is required for maintenance of
pluripotency in mouse epiblast and ES cells.
Cell
113(5):631
–
642.
31. Jiang J, et al. (2008) A core Klf circuitry regulates self-renewal of embryonic stem cells.
Nat Cell Biol
10(3):353
–
360.
32. Loh YH, Zhang W, Chen X, George J, Ng HH (2007) Jmjd1a and Jmjd2c histone H3
Lys 9 demethylases regulate self-renewal in embryonic stem cells.
Genes Dev
21(20):
2545
–
2557.
33. Chen X, Fang F, Liou YC, Ng HH (2008) Zfp143 regulates Nanog through modulation
of Oct4 binding.
Stem Cells
26(11):2759
–
2767.
34. Chazaud C, Yamanaka Y, Pawson T, Rossant J (2006) Early lineage segregation
between epiblast and primitive endoderm in mouse blastocysts through the Grb2-
MAPK pathway.
Dev Cell
10(5):615
–
624.
35. Santostefano KE, Hamazaki T, Pardo CE, Kladde MP, Terada N (2012) Fibroblast
growth factor receptor 2 homodimerization rapidly reduces transcription of the
pluripotency gene Nanog without dissociation of activating transcription factors.
J Biol Chem
287(36):30507
–
30517.
36. Ogilvie RL, et al. (2005) Tristetraprolin down-regulates IL-2 gene expression through
AU-rich element-mediated mRNA decay.
J Immunol
174(2):953
–
961.
37. Deleault KM, Skinner SJ, Brooks SA (2008) Tristetraprolin regulates TNF TNF-alpha
mRNA stability via a proteasome dependent mechanism involving the combined ac-
tion of the ERK and p38 pathways.
Mol Immunol
45(1):13
–
24.
38. Wang W, et al. (2000) HuR regulates p21 mRNA stabilization by UV light.
Mol Cell Biol
20(3):760
–
769.
39. Jing Q, et al. (2005) Involvement of microRNA in AU-rich element-mediated mRNA
instability.
Cell
120(5):623
–
634.
40. Frankenberg S, et al. (2011) Primitive endoderm differentiates via a three-step
mechanism involving Nanog and RTK signaling.
Dev Cell
21(6):1005
–
1013.
41. Boulet AM, Capecchi MR (2012) Signaling by FGF4 and FGF8 is required for axial
elongation of the mouse embryo.
Dev Biol
371(2):235
–
245.
42. Sun X, Meyers EN, Lewandoski M, Martin GR (1999) Targeted disruption of Fgf8
causes failure of cell migration in the gastrulating mouse embryo.
Genes Dev
13(14):
1834
–
1846.
43. Schulte-Merker S, Smith JC (1995) Mesoderm formation in response to Brachyury
requires FGF signalling.
Curr Biol
5(1):62
–
67.
44. Xu X, et al. (1998) Fibroblast growth factor receptor 2 (FGFR2)-mediated reciprocal
regulation loop between FGF8 and FGF10 is essential for limb induction.
De-
velopment
125(4):753
–
765.
45. Ying QL, Stavridis M, Griffiths D, Li M, Smi
th A (2003) Conversion of
embryonic stem cells
into neuroectodermal precurso
rs in adherent monoculture.
Nat Biotechnol
21(2):183
–
186.
46. Roberts A, Pachter L (2013) Streaming fragment assignment for real-time analysis of
sequencing experiments.
Nat Methods
10(1):71
–
73.
47. Trapnell C, et al. (2010) Transcript assembly and quantification by RNA-Seq reveals
unannotated transcripts and isoform swi
tching during cell differentiation.
Nat Biotechnol
28(5):511
–
515.
E1748
|
www.pnas.org/cgi/doi/10.1073/pnas.1320873111
Tan and Elowitz