An Intermediate Level of BMP Signaling Directly
Specifies Cranial Neural Crest Progenitor Cells in
Jennifer A. Schumacher.¤, Megumi Hashiguchi., Vu H. Nguyen., Mary C. Mullins*
Department of Cell and Developmental Biology, University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania, United States of America
The specification of the neural crest progenitor cell (NCPC) population in the early vertebrate embryo requires an elaborate
network of signaling pathways, one of which is the Bone Morphogenetic Protein (BMP) pathway. Based on alterations in
neural crest gene expression in zebrafish BMP pathway component mutants, we previously proposed a model in which the
gastrula BMP morphogen gradient establishes an intermediate level of BMP activity establishing the future NCPC domain.
Here, we tested this model and show that an intermediate level of BMP signaling acts directly to specify the NCPC. We
quantified the effects of reducing BMP signaling on the number of neural crest cells and show that neural crest cells are
significantly increased when BMP signaling is reduced and that this increase is not due to an increase in cell proliferation. In
contrast, when BMP signaling is eliminated, NCPC fail to be specified. We modulated BMP signaling levels in BMP pathway
mutants with expanded or no NCPCs to demonstrate that an intermediate level of BMP signaling specifies the NCPC. We
further investigated the ability of Smad5 to act in a graded fashion by injecting smad5 antisense morpholinos and show that
increasing doses first expand the NCPCs and then cause a loss of NCPCs, consistent with Smad5 acting directly in neural
crest progenitor specification. Using Western blot analysis, we show that P-Smad5 levels are dose-dependently reduced in
smad5 morphants, consistent with an intermediate level of BMP signaling acting through Smad5 to specify the neural crest
progenitors. Finally, we performed chimeric analysis to demonstrate for the first time that BMP signal reception is required
directly by NCPCs for their specification. Together these results add substantial evidence to a model in which graded BMP
signaling acts as a morphogen to pattern the ectoderm, with an intermediate level acting in neural crest specification.
Citation: Schumacher JA, Hashiguchi M, Nguyen VH, Mullins MC (2011) An Intermediate Level of BMP Signaling Directly Specifies Cranial Neural Crest Progenitor
Cells in Zebrafish. PLoS ONE 6(11): e27403. doi:10.1371/journal.pone.0027403
Editor: Bruce Riley, Texas A&M University, United States of America
Received September 16, 2011; Accepted October 16, 2011; Published November 15, 2011
Copyright: ? 2011 Schumacher et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits
unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This work was supported by an National Institutes of Health (NIH) grant to MCM (GM56326), NIH training grants (T32 HD0075165 and T32 GM07229-28)
to JAS, and an NIH predoctoral fellowship to VHN. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the
Competing Interests: The authors have declared that no competing interests exist.
* E-mail: firstname.lastname@example.org
. These authors contributed equally to this work.
¤ Current address: Division of Developmental Biology, Cincinnati Children’s Hospital Medical Center, Cincinnati, Ohio, United States of America
Neural crest cells are a multipotent population derived from
embryonic ectoderm. During neurulation neural crest cells
undergo an epithelial-to-mesenchymal transition, delaminate from
the dorsal neural tube, and migrate throughout the embryo,
contributing to a variety of tissues including craniofacial skeleton,
pigment cells, and the peripheral nervous system. In both frog and
chick, juxtaposition of explanted neural and non-neural ectoderm
gives rise to neural crest cells [1–3]. Consistent with this neural-
non-neural tissue interaction generating the neural crest, the
zebrafish fate map reveals that neural crest cells are derived from
lateral regions of the gastrula, where prospective neural tissue
meets prospective epidermis . Similarly in Xenopus, fate map
analysis reveals that the prospective neural crest population lies
adjacent to the dorsolateral marginal zone at an early gastrula
stage . Consistent with these studies, the earliest genes
specifically expressed within the neural crest progenitor cells
(NCPC), e.g. snail, AP2, and foxd3, are localized to lateral regions of
the neural plate adjacent to the non-neural ectoderm [6–12].
Gain-of-function studies in chick and Xenopus have addressed
the molecular nature of the signals that are involved in the
induction of the neural crest. These studies have implicated Bone
Morphogenetic Protein (BMP) signaling, among other signals such
as Wnt and FGF, as necessary in this inductive process [9,13,14].
BMPs are postulated to pattern the ectoderm of zebrafish and
Xenopus in a gradient fashion, such that high levels of activity
induce epidermis, intermediate levels induce neural crest, and the
absence of BMP activity is required for neurectoderm formation.
In support of this idea, when zebrafish embryos are treated with a
high concentration of dorsomorphin, a small molecule that inhibits
type I BMP receptor activity, neural crest cells are absent, whereas
a low concentration of dorsomorphin causes expansion of neural
crest cells . When Xenopus animal caps are excised and treated
with intermediate levels of Noggin, they express the early neural
crest marker slug, although this also requires the presence of FGF
. These results indicate that modest attenuation of endogenous
BMP signaling can lead to neural crest induction. Other evidence
for a BMP signaling gradient in the ectoderm, and evidence for an
intermediate level of BMP signaling patterning lateral regions of
PLoS ONE | www.plosone.org1 November 2011 | Volume 6 | Issue 11 | e27403
the embryo, particularly neural crest, comes from genetic analysis
in zebrafish [16–18]. In the strongly dorsalized swirl/bmp2b
mutant, foxd3, AP2, and snail expression in neural crest during
somitogenesis is absent, consistent with a requirement for BMP
signaling in neural crest specification. In more weakly dorsalized
somitabun (sbn)/smad5 and snailhousety68a(snh)/bmp7a mutants, neural
crest is greatly and moderately expanded, respectively, suggesting
that these mutants retain an intermediate level of BMP signaling in
an expanded region sufficient to specify neural crest .
However, the extent of the expansions has not been characterized,
nor has the residual signaling in these mutants been demonstrated.
Furthermore, the gradient model predicts that neural crest
progenitors directly respond to the intermediate level of BMP
signaling, however, this has not been addressed experimentally.
Here, we quantified the effects of reduction in BMP signaling on
the number of neural crest cells by counting the number of Foxd3-
positive cells in wild-type, swirl, sbn, and snhty68aembryos to show
that the expansion of the neural crest domain is not due to
impaired morphogenesis but rather an increase in neural crest cell
number. We modulate BMP signaling levels by over-expression of
BMP antagonists in wild type and various mutant conditions to
demonstrate that different levels of BMP signaling remain in sbn
and snhty68amutants. We further investigate the ability of Smad5 to
act in a graded fashion by injecting smad5 antisense morpholinos
and show that its dose-dependent loss recapitulates the BMP
mutant phenotypes, consistent with Smad5 acting directly in
neural crest progenitor specification. Using Western blot analysis,
we show that P-Smad5 levels are reduced in smad5 morphants in a
dose-dependent manner, consistent with an intermediate level of
BMP signaling acting through Smad5 to specify the neural crest
progenitors. Finally, we perform chimeric analysis to show that
BMP signaling is directly required within neural crest progenitor
cells for their specification. Together these results add substantial
evidence to a model in which graded BMP signaling acts as a
morphogen to pattern the ectoderm, with an intermediate level
responsible for neural crest specification.
Reduction of BMP signaling in the swr, sbn, and snh
mutants affects NCPC specification
We reported previously that the NCPC domain is decreased in
swr/bmp2b, and increased in sbndtc24/smad5 and snhty68a/bmp7a
mutant embryos . This analysis was done during early
somitogenesis stages, several hours after foxd3, the earliest neural
crest marker, is expressed. To investigate if these defects are due to
defects in NCPC specification, we examined foxd3 expression at the
end of gastrulation (bud stage) in these BMP pathway mutants. We
found that NCPCs were greatly reduced to absent in swr/bmp2b
mutants (Fig. 1B); greatly expanded in sbn/smad5 (Fig. 1C); and
moderately expanded in snhty68a/bmp7a mutant embryos (Fig. 1D)
compared to wild-type (Fig. 1A). These results are consistent with
the hypothesis that mutations in BMP pathway components affect
specification, rather than maintenance of neural crest.
BMP signaling has been implicated in both promoting and
inhibiting cell proliferation (reviewed in [19–22]). To determine if
the expanded number of NCPC observed in sbn mutants reflects an
increase in their proliferation, we examined the expression of
Phospho-Histone H3, a marker of proliferating cells, in wild-type
and sbn/smad5 mutant embryos. By gross inspection, we did not
detect a difference between the mutants and wild type at any point
throughout gastrulation (50%, 70% and 90% epiboly stages, and
bud stage, data not shown). We counted the number of Phospho-
Histone H3 positive cells at mid-gastrulation (70-80% epiboly) in
wild-type (n=3, Fig. 1E) and sbn/smad5 mutants (n=3, Fig. 1F) and
found no correlation between expanded NCPCs and an increase in
proliferation (Fig. 1G). Hence, an increase solely in cell proliferation
cannot account for the large increase in NCPCs in sbn/smad5
mutants. Rather these results are consistent with an enlarged
domain of cells that is specified as NCPC in sbn/smad5 mutants.
The number of NCPC is increased in sbn/smad5 and snh/
In 5-somite stage wild-type embryos, the anterior neural crest
population is 2–3 cell layers thick, and thins to a single cell layer in
the posterior (data not shown). It has previously been shown that
dorsal convergence is impaired in several BMP pathway mutants
[23,24], thus the apparent increase in the neural crest population
in these mutants could be due in part or entirely to a failure of the
NCPCs to converge into a multilayer tissue rather than an actual
increase in cell number. To determine the number of NCPC in
sbn/smad5 and snhty68a/bmp7a mutants compared to wild-type, we
counted the number of Foxd3-positive NCPCs (Fig. 1H–J). Foxd3
protein localizes to the nucleus, allowing one to easily count
individual cells, particularly in regions with multiple cell layers
. Foxd3 protein detection is delayed compared to foxd3 RNA.
Thus, we counted Foxd3-positive nuclei in 2-somite stage
embryos, the earliest time point exhibiting strong Foxd3
expression (Fig. 1H). The average number of neural crest cells in
wild-type was 403 (n=4), and in snhty68awas 1160 (n=4), or a 2.8-
fold increase over wild-type. sbn/smad5 embryos averaged 1965
neural crest cells (n=4), or a 4.8-fold increase over wild-type.
Thus, the larger apparent neural crest domain in snhty68a/bmp7a
and sbn/smad5 mutant embryos reflects a larger number of
NCPCs. The NCPCs in these mutants also occupy a larger area
than expected based on their increased number due to reduced
dorsal convergence of the NCPC in these mutants, i.e. most of the
NCPC were found in a single rather than a multi-cell layer.
Together, these results support a model in which an intermediate
level of BMP signaling is present in a larger domain in these
mutants than in wild-type embryos.
Reducing BMP signaling in wild-type embryos dose-
dependently affects NCPC phenotypes
Based on our previously proposed model, we predict that
moderate or strong inhibition of BMP signaling in wild-type
embryos will lead to an expansion or loss of NCPC, respectively.
To test this prediction, we decreased BMP signaling in wild type
embryos by injecting mRNA encoding either a truncated Xenopus
BMP receptor (tBR, ) or zebrafish chordin (Fig. 2A, ), an
extracellular BMP antagonist . Both of these methods yielded
similar results. As expected, over-expression of either inhibitor
produced a range of dorsalized phenotypes. We classified weakly
dorsalized embryos that displayed a roughly normal NCPC
phenotype as ‘‘wild-type’’. Upon injection of 50 pg of chordin
mRNA into wild-type embryos (n=89), we observed that 67% of
the embryos exhibited a wild-type (WT) NCPC phenotype, 21%
the ‘‘snh’’ and 12% the ‘‘sbn’’ phenotype (Fig. 2A, see Fig 1A–D for
the classification of ‘‘wild-type’’, ‘‘snh’’, ‘‘sbn’’, and ‘‘swr’’
phenotypes). When we increased the amount of chordin mRNA
injected to 200 pg (n=87), 40% of the embryos displayed the
‘‘swr’’ phenotype; 32% the ‘‘sbn’’; 18% the ‘‘snh’’ phenotype; and
only 10% exhibited a wild-type NCPC phenotype.
The sbndtc24mutation used in these neural crest studies is an
antimorphic allele of smad5, raising the possibility that the large
expansion of neural crest in these embryos is due to dominant-
negative effects interfering with Smad proteins that are used by
Intermediate BMP Directly Specifies Neural Crest
PLoS ONE | www.plosone.org2November 2011 | Volume 6 | Issue 11 | e27403
other signaling pathways acting in NCPC specification. To
confirm that the sbndtc24phenotype is due to a reduction in
BMP-responsive Smad5 activity only, and to determine if Smad5
itself can act in a dose-dependent fashion, we injected various
amounts of a previously described translation-blocking smad5
antisense morpholino oligonucleotide into wild-type embryos .
Injection of 2.5 ng of smad5 MO1 led to embryos with a range of
moderately to largely expanded neural crest populations, whereas
4 ng of smad5 MO1 led to a large expansion or loss of neural crest
cells (Fig. 2B). Taken together, these results indicate that reducing
BMP signaling dose-dependently recapitulates the NCPC pheno-
types observed in the BMP mutants. Furthermore, the results
indicate that the gradient of BMP signaling that specifies NCPCs
acts primarily through Smad5, and that Smad5 itself can act in a
graded fashion to specify NCPCs.
The NCPC phenotypes of swr/bmp2b, sbn/smad5, and
snh/bmp7a mutants correlate with distinct residual BMP
We previously proposed a model in which an intermediate level
of BMP signaling specifies NCPCs, and that BMP signaling is
lowered below this level in swr/bmp2b mutants, resulting in a loss of
NCPCs. Furthermore, we predict that different intermediate
amounts of residual BMP signaling are present in sbn/smad5 and
snh/bmp7a mutants, leading to the great and moderate expansion
of NCPCs, respectively, in these mutants . To test our
hypothesis that the NCPC phenotypes of swr/bmp2b, sbn/smad5,
and snh/bmp7a reflect the relative amounts of BMP signaling in
these mutants, we decreased or increased the amount of BMP
signaling in these mutants and examined the effects on NCPCs.
Having established that tBR and chordin over-expression can
recapitulate the BMP mutant phenotypes in a dose-sensitive
fashion, we used this technique to further decrease BMP signaling
in the BMP pathway component mutants. We predicted that
decreasing BMP signaling in sbn/smad5 or snhty68a/bmp7a mutants
would phenocopy the NCPC phenotype of swr mutants or sbn and
swr mutants, respectively, in a dose-dependent manner. Fig. 3A
shows a representative experiment in which we injected tBR into
embryos from a cross between two sbn/smad5 heterozygous fish.
Since sbn/smad5 is a fully penetrant dominant maternal-effect
mutation , all offspring from this cross exhibit the mutant
phenotype. In the uninjected group (n=48), 90% of the embryos
exhibit a ‘‘sbn’’ NCPC phenotype in which there is a great
Figure 1. Reduction of BMP signaling in BMP pathway mutants affects the number of NCPC specified. foxd3 expression at bud stage in
wild-type (A), swr (B), sbn (C), and snhty68a(D). P-histone H3 expression at 80% epiboly stage in wild-type (E) and sbn (F). (G) Quantification of the
number of P-histone H3 positive cells in wild-type (n=3) and sbn (n=3). Foxd3 protein expression at the 2-somite stage in wild-type (H), sbn (I), and
Intermediate BMP Directly Specifies Neural Crest
PLoS ONE | www.plosone.org3 November 2011 | Volume 6 | Issue 11 | e27403
expansion of NCPC and 10% display a ‘‘snh’’ NCPC phenotype, a
moderate expansion of NCPC. However, injection of 150 pg of
tBR into these mutants (n=86) resulted in a reduction in the
number of NCPCs. Only 27% of embryos exhibited the ‘‘sbn’’
NCPC phenotype, whereas the majority (73%) showed a great
reduction of NCPCs, resembling swr mutants.
It is possible that the loss of foxd3 expression reflects a general
deleterious effect on development and not a stronger dorsalization
of sbn/smad5 mutant embryos. To address whether over-expression
of tBR caused a general reduction in gene expression, we examined
the expression of krox20, a marker of prospective hindbrain
rhombomeres 3 and 5 . We found that sbn/smad5 mutant
embryos injected with the tBR mRNA expressed krox20 and, in
fact, both rhombomeres were greatly expanded in these embryos
in a manner similar to that of swr/bmp2b mutants (data not shown,
). From these results, we conclude that there is residual BMP
signaling in sbn/smad5 embryos, thus reducing BMP signaling in
these embryos phenocopies the ‘‘swr’’ NCPC phenotype.
We next reduced the level of BMP signaling in snh/bmp7a
mutant embryos to ask whether this would result in the NCPC
phenotype observed in sbn/smad5 or swr/bmp2b mutants. We
injected embryos from a cross between two rescued homozygous
Figure 2. Reduction of BMP signaling in wild-type embryos
causes expansion or loss of NCPC in dosage-sensitive manner.
foxd3 expression in chd mRNA injected embryos (A) and smad5 MO
injected embryos (B) at the end of gastrulation. (A) Injection of a low
dose of chordin mRNA (50 pg) generates weaker NCPC phenotypes
(WT=normal or very mild expansion, ‘‘snh’’ = moderate expansion),
whereas a high dose (200 pg) leads to strong phenotypes (‘‘sbn’’=large
expansion, or ‘‘swr’’ =loss). (B) Injection of a low 2.5 ng dose of smad5
MO leads to ‘‘snh’’ and ‘‘sbn’’ phenotypes. Injection of a high 4 ng dose
of smad5 MO leads to ‘‘sbn’’ and ‘‘swr’’ phenotypes exclusively.
Figure 3. NCPC domains when BMP signaling is reduced in
somitabun and snailhouse and increased in swirl embryos. foxd3
expression at the end of gastrulation in tBR mRNA injected embryos of
sbn(A)andof snh (B), andsmad5 mRNA injectedintoswr embryos (C). (A)
Injection of tBR mRNA into sbn mutants leads to the majority of embryos
displaying a ‘‘swr’’ phenotype. (B) Injection of a low 15 pg dose of tBR
mRNA into snhmutants leads tonearlyequal numbers of‘‘snh’’ and‘‘sbn’’
phenotypes. Injection of a higher 100 pg dose leads to the majority of
embryos displaying the ‘‘sbn’’ phenotype and also a percentage
displaying a stronger ‘‘swr’’ phenotype. (C) Injection of a low 30 pg dose
of murine smad5 mRNA results in nearly half of embryos displaying a
‘‘sbn’’ phenotype. Injection of a higher 150 pg dose results in a small
percentage of embryos displaying the ‘‘swr’’ phenotype, and the rest of
the embryos divided between ‘‘sbn’’, ‘‘snh’’, and WT phenotypes.
Intermediate BMP Directly Specifies Neural Crest
PLoS ONE | www.plosone.org4November 2011 | Volume 6 | Issue 11 | e27403
snh/bmp7a adult fish; thus all progeny display a mutant phenotype.
As expected, the uninjected embryos exhibited a moderate
expansion of NCPC typical of the ‘‘snh’’ phenotype (Fig. 3B).
When we injected a low amount (15 pg) of tBR mRNA (n=61),
52% of the embryos showed the ‘‘sbn/smad5’’ NCPC phenotype;
whereas only 48% exhibited the ‘‘snh’’ NCPC phenotype (Fig. 2B).
When we injected a high amount (100 pg) of tBR mRNA, the
strength of the NCPC phenotype increased: 17% of the snh
homozygous embryos exhibited the ‘‘swr’’ NCPC phenotype; 71%
displayed the ‘‘sbn’’ phenotype; and only 12% showed the ‘‘snh’’
phenotype seen in the uninjected siblings. Thus, we find that
lowering BMP signaling in snhty68amutant embryos can phenocopy
the great expansion of the NCPCs observed in sbn/smad5 mutants,
and that a further reduction results in loss of NCPCs in a small
percentage of embryos. We note here that with all of these
experiments, we consistently found embryos that appeared
intermediate between either the "swr" and ‘‘sbn" or the "sbn" and
"snh " NCPC phenotypes. In these cases, we classified the embryos
into the group they most closely resembled.
We also increased BMP signaling in a dose-dependent manner
in the swr/bmp2b mutant by over-expressing smad5 mRNA to
determine if this could produce the NCPC phenotype of sbn/smad5
and snhty68a/bmp7a mutants. Following injection of 30 pg of murine
smad5 mRNA (n=82), we found that 44% of the embryos
displayed the expanded NCPC phenotype of ‘‘sbn’’ mutants, and
the remaining mutants appeared unchanged. When we increased
the amount to 150 pg (n=93), we observed a range of phenotypes
from the "swr" to wild-type NCPC phenotypes (Fig. 3C).
Therefore, we conclude that increasing BMP signaling in swr/
bmp2b mutants leads to an expansion of NCPC in a dosage-
Phosphorylated Smad1/5 levels correlate with the
strength of NCPC phenotype in smad5 morphant
To further examine the extent of reduction in BMP signaling in
smad5 morphant embryos, we examined phosphorylated Smad1/5
(P-Smad1/5) levels. We injected smad5 translation-blocking
morpholinos (MOs) into wild-type embryos at different concen-
trations and examined P-Smad1/5 levels by Western blot analysis
at the 60% epiboly stage, just after cranial NCPC are specified
. For each injection dose, a fraction of the embryos were used
for Western blot analysis and the remaining embryos were
examined for the NCPC phenotype.
Compared with wild-type embryos, all smad5 morphant
embryos displayed lower P-Smad5 levels that correlated with the
smad5 MO dose injected. Embryos injected with 6ng of smad5
MOs had no detectable P-Smad1/5 compared to embryos
injected with 5 ng, 4 ng and 3 ng of smad5 MOs, which had
progressively increasing levels of P-Smad1/5 (Fig. 4A).
We examined expression of foxd3, one of the earliest markers of
NCPC, at the end of gastrulation in the remaining embryos of
each group to determine the extent of expansion of, or loss of
NCPC at each smad5 MO dose. This allowed us to correlate the
NCPC phenotype with P-Smad1/5 levels during gastrulation.
Injection of a low dose of smad5 MOs (3 ng) caused primarily WT
and ‘‘snh’’ phenotypes, whereas higher doses of smad5 MOs (5 ng,
6 ng) resulted in ‘‘sbn’’ and ‘‘swr’’ phenotypes almost exclusively
(Fig. 4B, 4C). We find that increasing amounts of smad5 MOs leads
to decreasing P-Smad1/5 levels, which correlates with expansion
of NCPCs at intermediate P-Smad1/5 levels and then a loss of
NCPCs at undetectable P-Smad1/5 levels, consistent with the
BMP mutant results.
BMP signaling is autonomously required for neural crest
progenitor cell specification
Models of BMP gradient activity predict that each cell type
within the gradient field responds directly to the level of BMP
signaling, thus neural crest progenitor cells are expected to
respond directly to an intermediate level of BMP signaling. The
direct response of NCPCs to BMP signaling has not been
addressed in any organism. We addressed the cell autonomy of
BMP signaling in NCPCs by placing donor cells that cannot
respond to a BMP signal into a wild-type host environment where
BMP signaling is intact and determined if the donor cells can
express Foxd3. If the donor cells can express Foxd3, this indicates
that BMP signaling is not acting directly to specify NCPCs.
Alternatively, if donor cells do not express Foxd3 this would
indicate that BMP signaling directly induces NCPCs.
To generate donor cells that cannot respond to BMP signaling,
we used a combination of smad5 morpholinos to inhibit Smad5
translation. We then transplanted 5–15 cells from blastula-stage
donors to a region above the margin of a blastula-stage wild type
host embryo, in a region predicted to become neural crest based
on fate-mapping studies . Donor embryos were analyzed
individually for foxd3 mRNA expression, and only chimeras
derived from donors that completely lacked foxd3 expression were
further analyzed. As a control, we transplanted wild-type cells into
the same region of blastula-stage wild-type host embryos, and
examined Foxd3 expression at the 3-somite stage. In 7 of 7
chimeras analyzed, these wild-type donor cells readily expressed
Foxd3 (Fig. 5A–D). In 8 of 9 embryos with smad5 morphant clones
in the neural crest area, donor cells did not express Foxd3. Clones
in two separate embryos from two different donors are shown in
Figure 5E–N. In the single embryo in which donor cells expressed
Foxd3, donor cells became neural crest cells only ectopically in a
more ventral region to the normal domain (Fig. 5O–Q). Ectopic
neural crest induction was never observed with wild-type donor
cells. It is possible that the morphant-derived cells retained a slight
amount of Smad5, insufficient in the donor to specify neural crest
but sufficient to weakly respond to a surrounding high level of
BMP signaling ventrally in the chimera; responding at the lower
level appropriate for neural crest rather than epidermis, which
would normally occupy the domain. Taken together, these results
indicate that NCPCs directly respond to BMP signaling.
A low level of BMP signaling is required for specification
of neural crest progenitors
Temporal inhibition of BMP signaling using Tg(hsp70:chd) 
and treatment of embryos with dorsomorphin at different time
points  show that BMP signaling acts in cranial neural crest
specification at an early gastrula stage. Cells become committed in
the zebrafish embryo between the shield stage and 80% epiboly
[33,34]. NCPC phenotypes are evident in BMP pathway mutants
at bud stage, shortly after cell commitment, suggesting that BMP
signaling is involved in the primary specification of these cells
rather than maintenance of this cell type.
We previously reported that there are no differences in cell
death between wild-type and BMP pathway mutants during
blastula and gastrula stages . Thus, cell death likely does not
contribute to reduction of NCPCs in swr/bmp2b mutants. Our
phospho-histone H3 data revealed no correlation between the
increase in number of NCPCs in sbn/smad5 mutants and cell
proliferation. Taken together, these results indicate that cell
proliferation and cell death do not contribute to the expansion or
Intermediate BMP Directly Specifies Neural Crest
PLoS ONE | www.plosone.org5 November 2011 | Volume 6 | Issue 11 | e27403
Figure 4. Levels of P-Smad1/5 correlate with the strength of NCPC phenotype in smad5 morphant embryos. (A) P-Smad1/5 levels in
smad5 morphant embryos. Embryos injected with increasing higher concentrations of smad5 MOs show decreasing P-Smad1/5 levels relative to
uninjected controls. Actin was used as a loading control. (B, C) Expression of foxd3 in embryos injected with increasing doses of smad5 MOs. Injection
of a low dose of smad5 MOs (3 ng) leads to WT, ‘‘snh’’ and ‘‘sbn’’ phenotypes, whereas injection of higher doses of smad5 MOs (5 ng, 6 ng) lead to
‘‘sbn’’ and ‘‘swr’’ phenotypes. The embryos used for in situ hybridization of foxd3 in Fig. 4B are from the same batch of injected embryos used for
Western blotting in Fig. 4A.
Intermediate BMP Directly Specifies Neural Crest
PLoS ONE | www.plosone.org6 November 2011 | Volume 6 | Issue 11 | e27403
loss of NCPCs, and that these phenotypes reflect the specification
of NCPCs based on the level of BMP activity in each mutant.
Based on our results, we propose a model to explain the NCPC
phenotypes displayed by BMP pathway mutants, shown in
Figure 6. Our results indicate that a low level of BMP signaling
specifies NCPC. This level would be found throughout much of
sbn/smad5 mutant embryos, thereby directing ectodermal cells
located in ventral and lateral regions of this mutant towards a
NCPC fate. Our data suggest that the level of BMP signaling in
swr/bmp2b mutants is lower than in sbn/smad5 mutants, very likely
near or at the threshold of BMP activity required for NCPC
specification, since weak swr/bmp2b mutants retain a few NCPCs.
In the case of snhty68a/bmp7a mutant embryos, our results indicate
that the level of BMP signaling is highest among the three mutants,
but lower than in wild-type embryos. If a low level of BMP
signaling specifies the NCPCs, then our model would predict a
moderate expansion of NCPCs in snhty68a/bmp7a mutant embryos,
consistent with their phenotype.
Integrating our model with other studies of NCPC
Experiments in amphibians and chick in which neural and
epidermal tissues are juxtaposed either in vivo or in vitro led to a
model in which neural crest is induced at the border between
neural and non-neural ectoderm [2,3]. In these experiments,
neural tissue is a source of the secreted BMP antagontists Chordin,
Figure 5. BMP signaling is required cell autonomously in NCPCs for their specification. (A) Z-projection of confocal sections showing that
wild-type donor cells transplanted into wild-type hosts readily express the neural crest marker Foxd3 at the 3-somite stage. (B,C,D) Single confocal
section of the embryo in A. Foxd3 (B) and lineage tracer rhodamine dextran (C) are found in the same cells (D, arrowheads). (E) Z-projection of
confocal sections showing that smad5 morphant donor cells within the neural crest region of a wild-type host do not express the neural crest marker
Foxd3. (F, G, H, I, J, K) Single confocal sections of the cells indicated by arrows in E. Foxd3 (F, I) and lineage tracer (G, J) do not colocalize (H, K). (L, M,
N) Single confocal section of a different host embryo containing smad5 morphant cells (Rho) within the neural crest region that do not express Foxd3.
(O, P, Q) Z-projection of confocal sections of chimera in which donor cells were induced as ectopic neural crest. Foxd3 (O) and lineage tracer (P)
colocalize (Q) in a patch of cells (arrowhead) located ventrally from the normal neural crest domain (asterisk).
Intermediate BMP Directly Specifies Neural Crest
PLoS ONE | www.plosone.org7 November 2011 | Volume 6 | Issue 11 | e27403
Noggin, and Follistatin. It is likely that these antagonists diffuse
across the border into epidermal tissue and inhibit the BMP
ligands there to establish an intermediate level of BMP activity at
the border, which would then induce neural crest formation. This
scenario mimics the proposed mechanism for formation of the
BMP activity gradient in vivo, which generates a low level of BMP
signaling in lateral regions of the gastrulating embryo, precisely
where NCPC are located. Thus, our results are consistent with the
previously proposed model. Importantly, our chimeric analysis
extends this model to show that neural crest progenitor cells
require an intact BMP signaling pathway cell autonomously,
revealing for the first time that NCPCs directly respond to BMP
In zebrafish Wnt signal reception is also required autonomously
within neural crest cells, and loss of wnt8 results in a loss of neural
crest . It is possible that an intermediate level of BMP signaling
makes prospective neural crest cells competent to receive a second
inducing signal, perhaps a Wnt signal. In support of this idea,
conjugation of Xenopus wild type animal caps to animal caps
expressing chordin weakly induces neural crest markers snail and
Xslug, whereas conjugation of animal caps expressing chordin to
those expressing wnt8 strongly induces Xslug expression . In
Xenopus embryos, Gbx2, the earliest factor in neural crest
induction, is a direct target of Wnt signaling . Furthermore,
during gastrulation in zebrafish, wnt8 expression extends from the
margin towards the animal pole in lateral regions of the embryo
, where BMP signaling is predicted to be present at
intermediate levels, thus making it a good candidate for the
Wnt signaling is also a posteriorizing factor, and along with FGF
and retinoic acid, may specify neural crest in the BMP-induced
competent lateral regions by nature of their posteriorizing activity.
Consistent with this, anterior lateral ectoderm, or neural fold,
which normally becomes forebrain, not neural crest, can be
induced to express neural crest markers by treatment with Wnt,
FGF, or retinoic acid (RA) . Knockdown of FGF and RA
signaling strongly reduce expression of midkine-b, which regulates
cell specification at the neural plate border, whereas activation of
Wnt signaling enhances the expression . However, b-catenin
can also expand neural crest in whole embryos without poster-
iorizing neural tissue . Our results are consistent with either of
The interpretation of graded BMP signaling
Our observation that BMP signaling is required cell autono-
mously by neural crest cells highlights the interesting question of
how a subset of cells within a gradient field can interpret a specific
level of signal to activate the appropriate downstream targets. The
mechanism for BMP gradient interpretation is poorly understood
in vertebrates. Our result that reduction of P-Smad1/5 levels leads
to expansion and loss of neural crest in a concentration-dependent
manner is consistent with a scenario in which differential
activation of downstream targets at least in part leads to a distinct
domain of neural crest gene expression. However, foxd3 is not
known to be a direct target of the BMP pathway, thus there are
likely other downstream targets through which an intermediate
level of BMP signaling leads to foxd3 expression in neural crest.
We previously proposed that the gastrula BMP gradient
generates a pattern of nested gene expression in ventral and
lateral regions of the embryo, which in turn would reciprocally
regulate each other to generate more restricted domains of
expression by the end of gastrulation. In this model AP2 expression
becomes restricted to lateral regions of the neural plate by the end
of gastrulation . It is possible that AP2 provides a link between
BMP signaling and neural crest induction. In support of this
notion, ectopic expression or morpholino knockdown of AP2 leads
to induction or reduction, respectively, of the neural crest specific
genes Slug and Sox9 in Xenopus . Furthermore, simultaneous
Figure 6. BMP gradient model for NCPC specification. The Y axis shows BMP signaling levels. The X axis indicates position along the
dorsoventral axis. The threshold range of BMP signaling that specifies NCPC is shown in yellow. The intersection of the gradient with the threshold
range for NCPC specification leads to NCPC formation in a lateral region in the size domain shown. In WT, the gradient of BMP signaling reaches a
high level ventrally and NCPCs are located in a lateral region of the embryo where BMP signaling levels are low. The region of NCPCs specified in WT
is shown with black stripes over the yellow area. In snh, the BMP signaling gradient is lower than WT. Therefore, the NCPCs in snh (blue striped area)
are slightly expanded compared to wild type and are located in a more ventral region than WT. In sbn, the BMP signaling gradient is lower than snh.
The NCPCs are located in a more ventral region than snh and the NCPCs in sbn (red striped area) are greatly expanded compared to wild type. In the
swirl/bmp2b mutant, BMP signaling level is absent or very low, leading to the great reduction or absence of NCPCs.
Intermediate BMP Directly Specifies Neural Crest
PLoS ONE | www.plosone.org8 November 2011 | Volume 6 | Issue 11 | e27403
loss of two AP2 genes in zebrafish, tfap2a and tfap2c, reveals a cell
autonomous requirement for AP2 in NCPC specification [39,40].
Another potential link between BMP signaling and neural crest
gene expression is the Msx family, which has been shown to be a
direct target of BMP signaling [41,42]. msxB expression is affected
in BMP mutants in a similar, yet even more sensitive manner than
The results we present here provide evidence that an
intermediate level of BMP signaling specifies neural crest
progenitor cells. We also present the first evidence that BMP
signaling directly specifies neural crest progenitor cells. These
results will advance the further study of the action of BMP
signaling in neural crest specification and its interaction with other
pathways regulating neural crest specification.
Materials and Methods
All of the zebrafish studies were performed in accordance with,
and with approval from, the Institutional Animal Care and Use
Committee of the Office of Regulatory Affairs at the University of
Fish strains and breeding
Incrosses were performed between heterozygous fish of the
following mutants: swrtc300a, swrta72, swrtdc24, snhty68a, and sbndtc24
[30,44]. Embryos were also obtained from incrosses between
rescued homozygous swrtc300a, swrta72, and snhty68aadults. swr
(bmp2b) and snh (bmp7a) mutant embryos were determined at bud
stage or early somitogenesis stages by their elongated, American
football-shaped morphology. All progeny of sbn (smad5) heterozy-
gous females are mutant, due to the dominant maternal-effect
nature of this mutation .
In situ hybridization, immunohistochemistry, cell
Whole-mount in situ hybridization was performed as previously
described using the foxd3 probe (previously called fkd6, ).
Foxd3 protein was detected using a 1:1000 dilution of Foxd3
rabbit antisera in 10% NCS-PBST (10% fetal bovine serum, 1%
DMSO, 0.1% Tween 20 in PBS) as described . Following
several washes in PBST, embryos were developed using diamino-
benzidine according to the manufacturer’s directions (Vector
Labs). For Foxd3 cell counting experiments, stained embryos were
fixed in 4% paraformaldehyde-PBST for two days at room
temperature, then cleared in benzylbenzoate:benzylalcohol (2:1),
de-yolked using watchmaker forceps, and flattened onto glass
slides in Canada Balsam. Embryos were viewed using Nomarski
optics on a Leica Axioskop. Phospho (P)-histone H3 staining was
performed as previously described . For P-histone H3 cell
counting experiments, stained embryos were cleared in 100%
glycerol. To count the number of P-histone H3 positive cells in the
whole embryo without duplication, several nuclei were first
selected as landmarks in the embryo and images taken. Then
the embryo was rotated slightly and several new nuclei were
selected for landmarks and images taken. Rotation of the embryo,
selection of landmarks and image acquisition were repeated until
returning to the first landmark. Images were captured using a
Kontron digital camera and processed using Adobe software using
the nuclei landmarks to overlap images. Cells were counted from
mRNA and morpholino injections
mRNA encoding tBR , chordin , mSmad5 , and
zebrafish smad5 was synthesized using the SP6 mMessage
mMachine Kit (Ambion). mRNA injections were directed into
the yolk of 1–4 cell stage embryos, and were each performed at
least three times, all with results similar to experiments shown.
smad5 morpholino antisense oligonucleotides (Gene Tools) were
injected into the cell of 1-cell stage embryos from a cross of sbnm169
heterozygotes for transplant experiments and into wild-type
embryos for all other experiments. Concentration of MOs in
Fig. 4 shows total combined concentration of equal amounts of
smad5 MO1 and smad5 MO3. The concentrations are different
from Fig. 2B, due to different smad5 morpholino production lots
used. smad5 MO1 sequence is ATGGAGGTCATAGTGCTG-
GGCTGC, smad5 MO3 sequence is GCAGTGTGCCAGGAA-
Western Blot Analysis
Phospho-Smad1/5 Western blots were performed as previously
described [32,47], except that the primary antibody was applied
for one overnight period at a 1:500 dilution. After the P-Smad1/5
Western, P-Smad1/5 antibody was stripped from the membrane
and Actin antibody was re-probed, as a control.
Donor embryos were injected with a combination of smad5
MO1 (3 ng) and smad5 MO3 (2–3 ng), as well as a 2.5% solution
of lysine-fixable rhodamine-dextran and 2.5% solution of
biotinylated dextran. Approximately 5–15 cells at the blastula
stage were transplanted into the marginal region of blastula-stage,
unlabelled hosts. Donor embryos were fixed at the bud stage and
analyzed for the presence of neural crest by foxd3 in situ
hybridization. Host embryos were fixed at the 3–4 somite stage.
Foxd3 was detected using a 1:500 dilution of Foxd3 rabbit antisera
 in 20% NCS-PBST followed by a 1:500 dilution of anti-rabbit
Alexa 488 in 10% NCS-PBST. Donor cells were visualized by
rhodamine fluorescence. Embryos were mounted in Vectashield
and imaged using a Zeiss LSM 510 confocal microscope. Images
were processed using ImageJ and Adobe software.
We thank J. Dutko and Y. Langdon for comments on this manuscript. We
thank past and present members of the fish facility staff for excellent fish
care and facility maintenance.
Conceived and designed the experiments: JAS MH VHN MCM.
Performed the experiments: JAS MH VHN. Analyzed the data: JAS
MH VHN MCM. Contributed reagents/materials/analysis tools: JAS MH
VHN MCM. Wrote the paper: JAS MH VHN MCM.
1. Dickinson ME, Selleck MA, McMahon AP, Bronner-Fraser M (1995) Dorsalization
of the neural tube by the non-neural ectoderm. Development 121: 2099–2106.
2. Moury JD, Jacobson AG (1990) The origins of neural crest cells in the axolotl.
Dev Biol 141: 243–253.
3. Selleck MA, Bronner-Fraser M (1995) Origins of the avian neural crest: the role
of neural plate-epidermal interactions. Development 121: 525–538.
4. Woo K, Fraser SE (1995) Order and coherence in the fate map of the zebrafish
nervous system. Development 121: 2595–2609.
Intermediate BMP Directly Specifies Neural Crest
PLoS ONE | www.plosone.org9November 2011 | Volume 6 | Issue 11 | e27403
5. Steventon B, Araya C, Linker C, Kuriyama S, Mayor R (2009) Differential
requirements of BMP and Wnt signalling during gastrulation and neurulation
define two steps in neural crest induction. Development 136: 771–779.
6. Essex LJ, Mayor R, Sargent MG (1993) Expression of Xenopus snail in
mesoderm and prospective neural fold ectoderm. Dev Dyn 198: 108–122.
7. Furthauer M, Thisse C, Thisse B (1997) A role for FGF-8 in the dorsoventral
patterning of the zebrafish gastrula. Development 124: 4253–4264.
8. Luo T, Lee YH, Saint-Jeannet JP, Sargent TD (2003) Induction of neural crest
in Xenopus by transcription factor AP2alpha. Proc Natl Acad Sci U S A 100:
9. Mayor R, Morgan R, Sargent MG (1995) Induction of the prospective neural
crest of Xenopus. Development 121: 767–777.
10. Odenthal J, Nusslein-Volhard C (1998) fork head domain genes in zebrafish.
Dev Genes Evol 208: 245–258.
11. Sasai N, Mizuseki K, Sasai Y (2001) Requirement of FoxD3-class signaling for
neural crest determination in Xenopus. Development 128: 2525–2536.
12. Thisse C, Thisse B, Postlethwait JH (1995) Expression of snail2, a second
member of the zebrafish snail family, in cephalic mesendoderm and presumptive
neural crest of wild-type and spadetail mutant embryos. Dev Biol 172: 86–99.
13. Garcia-Castro MI, Marcelle C, Bronner-Fraser M (2002) Ectodermal Wnt
function as a neural crest inducer. Science 297: 848–851.
14. LaBonne C, Bronner-Fraser M (1998) Neural crest induction in Xenopus:
evidence for a two-signal model. Development 125: 2403–2414.
15. Kwon HJ, Bhat N, Sweet EM, Cornell RA, Riley BB (2010) Identification of
early requirements for preplacodal ectoderm and sensory organ development.
PLoS Genet 6.
16. Barth KA, Kishimoto Y, Rohr KB, Seydler C, Schulte-Merker S, et al. (1999)
Bmp activity establishes a gradient of positional information throughout the
entire neural plate. Development 126: 4977–4987.
17. Nguyen VH, Schmid B, Trout J, Connors SA, Ekker M, et al. (1998) Ventral
and lateral regions of the zebrafish gastrula, including the neural crest
progenitors, are established by a bmp2b/swirl pathway of genes. Dev Biol
18. Schmid B, Furthauer M, Connors SA, Trout J, Thisse B, et al. (2000) Equivalent
genetic roles for bmp7/snailhouse and bmp2b/swirl in dorsoventral pattern
formation. Development 127: 957–967.
19. Blair SS (2007) Wing vein patterning in Drosophila and the analysis of
intercellular signaling. Annu Rev Cell Dev Biol 23: 293–319.
20. Orlova VV, Liu Z, Goumans MJ, ten Dijke P (2011) Controlling angiogenesis by
two unique TGF-beta type I receptor signaling pathways. Histol Histopathol 26:
21. Roussel MF, Hatten ME (2011) Cerebellum development and medulloblastoma.
Curr Top Dev Biol 94: 235–282.
22. Ye L, Mason MD, Jiang WG (2011) Bone morphogenetic protein and bone
metastasis, implication and therapeutic potential. Front Biosci 16: 865–897.
23. Myers DC, Sepich DS, Solnica-Krezel L (2002) Bmp activity gradient regulates
convergent extension during zebrafish gastrulation. Dev Biol 243: 81–98.
24. von der Hardt S, Bakkers J, Inbal A, Carvalho L, Solnica-Krezel L, et al. (2007)
The Bmp gradient of the zebrafish gastrula guides migrating lateral cells by
regulating cell-cell adhesion. Curr Biol 17: 475–487.
25. Lewis JL, Bonner J, Modrell M, Ragland JW, Moon RT, et al. (2004) Reiterated
Wnt signaling during zebrafish neural crest development. Development 131:
26. Graff JM, Thies RS, Song JJ, Celeste AJ, Melton DA (1994) Studies with a
Xenopus BMP receptor suggest that ventral mesoderm-inducing signals override
dorsal signals in vivo. Cell 79: 169–179.
27. Miller-Bertoglio VE, Fisher S, Sanchez A, Mullins MC, Halpern ME (1997)
Differential regulation of chordin expression domains in mutant zebrafish. Dev
Biol 192: 537–550.
28. Piccolo S, Sasai Y, Lu B, De Robertis EM (1996) Dorsoventral patterning in
Xenopus: inhibition of ventral signals by direct binding of chordin to BMP-4.
Cell 86: 589–598.
29. Lele Z, Bakkers J, Hammerschmidt M (2001) Morpholino phenocopies of the
swirl, snailhouse, somitabun, minifin, silberblick, and pipetail mutations. Genesis
30. Mullins MC, Hammerschmidt M, Kane DA, Odenthal J, Brand M, et al. (1996)
Genes establishing dorsoventral pattern formation in the zebrafish embryo: the
ventral specifying genes. Development 123: 81–93.
31. Oxtoby E, Jowett T (1993) Cloning of the zebrafish krox-20 gene (krx-20) and its
expression during hindbrain development. Nucleic Acids Res 21: 1087–1095.
32. Tucker JA, Mintzer KA, Mullins MC (2008) The BMP signaling gradient
patterns dorsoventral tissues in a temporally progressive manner along the
anteroposterior axis. Dev Cell 14: 108–119.
33. Grinblat Y, Gamse J, Patel M, Sive H (1998) Determination of the zebrafish
forebrain: induction and patterning. Development 125: 4403–4416.
34. Woo K, Fraser SE (1998) Specification of the hindbrain fate in the zebrafish.
Dev Biol 197: 283–296.
35. Li B, Kuriyama S, Moreno M, Mayor R (2009) The posteriorizing gene Gbx2 is
a direct target of Wnt signalling and the earliest factor in neural crest induction.
Development 136: 3267–3278.
36. Villanueva S, Glavic A, Ruiz P, Mayor R (2002) Posteriorization by FGF, Wnt,
and retinoic acid is required for neural crest induction. Dev Biol 241: 289–301.
37. Liedtke D, Winkler C (2008) Midkine-b regulates cell specification at the neural
plate border in zebrafish. Dev Dyn 237: 62–74.
38. Wu J, Yang J, Klein PS (2005) Neural crest induction by the canonical Wnt
pathway can be dissociated from anterior-posterior neural patterning in
Xenopus. Dev Biol 279: 220–232.
39. Hoffman TL, Javier AL, Campeau SA, Knight RD, Schilling TF (2007) Tfap2
transcription factors in zebrafish neural crest development and ectodermal
evolution. J Exp Zool B Mol Dev Evol 308: 679–691.
40. Li W, Cornell RA (2007) Redundant activities of Tfap2a and Tfap2c are
required for neural crest induction and development of other non-neural
ectoderm derivatives in zebrafish embryos. Dev Biol 304: 338–354.
41. Suzuki A, Ueno N, Hemmati-Brivanlou A (1997) Xenopus msx1 mediates
epidermal induction and neural inhibition by BMP4. Development 124:
42. Takahashi T, Guron C, Shetty S, Matsui H, Raghow R (1997) A minimal
murine Msx-1 gene promoter. Organization of its cis-regulatory motifs and their
role in transcriptional activation in cells in culture and in transgenic mice. J Biol
Chem 272: 22667–22678.
43. Tribulo C, Aybar MJ, Nguyen VH, Mullins MC, Mayor R (2003) Regulation of
Msx genes by a Bmp gradient is essential for neural crest specification.
Development 130: 6441–6452.
44. Nguyen VH, Trout J, Connors SA, Andermann P, Weinberg E, et al. (2000)
Dorsal and intermediate neuronal cell types of the spinal cord are established by
a BMP signaling pathway. Development 127: 1209–1220.
45. Murphey RD, Stern HM, Straub CT, Zon LI (2006) A chemical genetic screen
for cell cycle inhibitors in zebrafish embryos. Chem Biol Drug Des 68: 213–219.
46. Suzuki A, Chang C, Yingling JM, Wang XF, Hemmati-Brivanlou A (1997)
Smad5 induces ventral fates in Xenopus embryo. Dev Biol 184: 402–405.
47. Mintzer KA, Lee MA, Runke G, Trout J, Whitman M, et al. (2001) Lost-a-fin
encodes a type I BMP receptor, Alk8, acting maternally and zygotically in
dorsoventral pattern formation. Development 128: 859–869.
Intermediate BMP Directly Specifies Neural Crest
PLoS ONE | www.plosone.org10November 2011 | Volume 6 | Issue 11 | e27403