Stage-specific control of neural crest stem cell proliferation by the small rho GTPases Cdc42 and Rac1.

Abstract

The neural crest (NC) generates a variety of neural and non-neural tissues during vertebrate development. Both migratory NC cells and their target structures contain cells with stem cell features. Here we show that these populations of neural crest-derived stem cells (NCSCs) are differentially regulated by small Rho GTPases. Deletion of either Cdc42 or Rac1 in the NC results in size reduction of multiple NC target structures because of increased cell-cycle exit, while NC cells emigrating from the neural tube are not affected. Consistently, Cdc42 or Rac1 inactivation reduces self-renewal and proliferation of later stage, but not early migratory NCSCs. This stage-specific requirement for small Rho GTPases is due to changes in NCSCs that, during development, acquire responsiveness to mitogenic EGF acting upstream of both Cdc42 and Rac1. Thus, our data reveal distinct mechanisms for growth control of NCSCs from different developmental stages.

Full text

Page 1

Cell Stem Cell
Article
Stage-Specific Control of Neural Crest
Stem Cell Proliferation by the Small
Rho GTPases Cdc42 and Rac1
Sebastian Fuchs,1,2Dominik Herzog,2Grzegorz Sumara,2Stine Bu ¨chmann-Møller,1,2Gianluca Civenni,1,2Xunwei Wu,3
Anna Chrostek-Grashoff,4Ueli Suter,2Romeo Ricci,2Joa ˜o B. Relvas,2,6Cord Brakebusch,5and Lukas Sommer1,2,*
1Institute of Anatomy, University of Zurich, CH-8057 Zurich, Switzerland
2Institute of Cell Biology, Department of Biology, ETH Zurich, CH-8093 Zurich, Switzerland
3CBRC, MGH, Harvard Medical School, Charlestown, MA 02129, USA
4Cardiovascular Research Center, University of Virginia Health System, Charlottesville, VA 22908, USA
5Institute of Molecular Pathology, University of Copenhagen, 2100 Copenhagen, Denmark
6Present address: Institute for Molecular and Cell Biology, University of Porto, 4150-180 Porto, Portugal
*Correspondence: lukas.sommer@anatom.uzh.ch
DOI 10.1016/j.stem.2009.01.017
SUMMARY
The neural crest (NC) generates a variety of neural
and non-neural tissues during vertebrate develop-
ment. Both migratory NC cells and their target struc-
tures contain cells with stem cell features. Here we
show that these populations of neural crest-derived
stem cells (NCSCs) are differentially regulated by
small Rho GTPases. Deletion of either Cdc42 or
Rac1 in the NC results in size reduction of multiple
NC target structures because of increased cell-cycle
exit, while NC cells emigrating from the neural tube
are not affected. Consistently, Cdc42 or Rac1 inacti-
vation reduces self-renewal and proliferation of later
stage, but not early migratory NCSCs. This stage-
specific requirement for small Rho GTPases is due
to changes in NCSCs that, during development,
acquire responsiveness to mitogenic EGF acting
upstream of both Cdc42 and Rac1. Thus, our data
reveal distinct mechanisms for growth control of
NCSCs from different developmental stages.
INTRODUCTION
The neural crest (NC) is a transient population of cells in higher
vertebrates that, during embryonic development, emigrates
from the dorsal neural tube to generate most of the peripheral
nervous system and a variety of non-neural structures (Le
Douarin and Dupin, 2003). Among fate-restricted cells, the NC
harbors many cells termed neural crest stem cells (NCSCs)
that display self-renewal capacity and multipotency in clonal
cell culture assays. Intriguingly, very similar cells in terms of
marker expression and potential have also been isolated
from NC-derived structures during fetal development and even
from the adult organism (Delfino-Machin et al., 2007). In vivo
fate mapping revealed a lineage relationship between early,
emigrating (e)NCSCs and NCSCs found at later stages (Wong
et al., 2006). These findings suggest that a fraction of undifferen-
tiated NCSCs must be maintained by self-renewal throughout
development and postnatal stages, while other NC progeny
undergo fate restriction, eventually exit the cell cycle, and differ-
entiate. The mechanisms regulating NCSC maintenance and
proliferation are only partially understood. In Xenopus, the tran-
scriptional regulator Id3 is essential for NC formation from the
neural plate and promotes proliferation and expansion of the
NC domain (Kee and Bronner-Fraser, 2005). At similarly early
stages of murine NC development, combinatorial Wnt and
BMP suppress differentiation and, in the presence of FGF2,
maintain mitotic eNCSCs (Kleber et al., 2005). However, NCSCs
from sciatic nerves and dorsal root ganglia lose Wnt responsive-
ness, pointing to distinct mechanisms of growth control in
different NCSC populations. At later stages, NC-derived stem
and progenitor cells from the gut and the skin are maintained
byendothelin3signalingorbytheactivityofthepolycombgroup
transcriptional repressor Bmi-1 (Bondurand et al., 2006; Molof-
sky et al., 2003; Real et al., 2006), but the relatively mild or
lineage-specific phenotypes of the respective mouse mutants
do not support a general and indispensable role of these factors
in NCSC expansion.
Cdc42 and Rac1 are ubiquitously expressed small Rho
GTPases that act as molecular switches, cycling between an
active GTP-bound and an inactive GDP-bound state (Jaffe and
Hall,2005).RhoGTPasesplayacentralroleinlinkingextracellular
stimuli with the induction of multiple specific downstream effec-
tors resulting in a variety of cellular responses. These include
effects on cell polarity, migration, differentiation, proliferation,
and apoptosis (Jaffe and Hall, 2005). Earlier studies have largely
relied on pharmacological inhibition of Rho GTPases or on
expression of dominant-negative or constitutively active forms.
Although these studies have been fundamental for the under-
standing of basic Rho GTPase biology, these approaches are
potentially nonspecific due to cross-activation or inhibition of
other effectors. Moreover, being central molecules within a
plethora of signaling pathways, Rho GTPase function can signifi-
cantly vary from one cell type to another. To analyze a given Rho
GTPase in a specific cell type in an in vivo setting, conditional
gene targeting of specific Rho GTPases in transgenic mice has,
therefore,becomethemethodofchoice(WangandZheng,2007).
236 Cell Stem Cell 4, 236–247, March 6, 2009 ª2009 Elsevier Inc.

Page 2

Knowledge about the roles of Rho GTPases in NC develop-
mentisstilllimited.RhoVhasbeenidentifiedasanessentialregu-
lator of NC induction in Xenopus (Guemar et al., 2007). Recently,
pharmacologicalinhibitionofCdc42/Rac1wasreportedtoaffect
migrationandneuriteextensionof NC-derivedcellsin theenteric
nervous system (ENS) (Stewart et al., 2007). Here, we performed
gene ablation in the NC and find that Cdc42 and Rac1 are
not required for maintenance, migration, and differentiation of
NCSCs in the early phase of NC development. However, at later
stages when NC cells have reached their initial targets, prolifera-
tion control becomes dependent on Cdc42 and Rac1. Thus,
Cdc42 and Rac1 are essential regulators of NC development
that distinguish NCSCs from different stages.
RESULTS
Inactivation of Cdc42 or Rac1 in NC Cells
ToassessthefunctionsofCdc42andRac1inNCcellsinvivo,we
crossedmicehomozygousforthefloxedalleleofeitherCdc42or
Rac1 (Chrostek et al., 2006; Wu et al., 2006) with mice heterozy-
gous for the respective allele, which additionally expressed the
Cre recombinase under control of the Wnt1 promoter (Danielian
etal.,1998)(Figures1Aand1B).Wnt1-Cre-mediatedrecombina-
Figure 1. Similar Developmental Defects
in NC Derivatives of Cdc42 and Rac1 cko
Embryos
(A and B) Exon 2 (E2) of the Cdc42 locus and Exon
3 (E3) of the Rac1 locus, respectively, are flanked
by loxP sites (floxed allele) and deleted (null allele)
in NC cells upon breeding with Wnt1-Cre mice.
(CandD)LossofCdc42andRac1protein,respec-
tively, in mutant NC cells populating pharyngeal
arch 1 (PA1) at E10.0.
(E) Sagital median cleft in Cdc42 and Rac1 cko
embryos shown on transverse sections at the level
of the frontonasal process.
(F) Mutant DRG are present but appear reduced in
size (white arrows) when compared to DRG in
a control embryo (black arrow).
(G) Failure of aortico-pulmonary septum formation
in the heart outflow tract of cko embryos. Scale
bars: (C) and (D), 100 mm; (E)–(G), 200 mm. Ao,
ascending aorta; PT, pulmonary trunk; TA, truncus
arteriosus.
tion efficiently deletes genes of interest in
virtually the entire NC population (Lee
et al., 2004; Wurdak et al., 2005). Effi-
ciency of recombination was verified
by genomic PCR, amplifying distinct
fragments for wild-type, floxed, or null
alleles of Cdc42 and Rac1, respectively
(Figure S1 available online). In Wnt1-
Cre/Cdc42lox/loxandWnt1-Cre/Rac1lox/lox
embryos, the respective null allele was
only observed in NC target tissue,
whereas PCR amplification from nonre-
combined tissue was negative for the
null allele, demonstrating that expression
of the Cre-recombinase was tissue-
specific. Immunohistochemistry revealed significantly reduced
expression of Cdc42 and Rac1 protein in recombined areas of
the respective mutants (Figures 1C and 1D).
Cdc42 and Rac1 Depletion Cause Multiple Defects
in NC Derivatives
Both Cdc42 and Rac1 conditional knockout (cko) embryos were
recovered up to embryonic day (E)13.5 with a Mendelian ratio,
whereas litters from later developmental stages contained no
mutant embryos (Figure S1). At E10.5, the gross morphology of
Rac1 cko embryos showed no obvious malformations. Cdc42
cko embryos appeared also normal except for slight abnormali-
ties in the mid-hindbrain region, where Wnt1-Cre is also
expressed. These deficiencies appeared similar to those previ-
ously described for the developing central nervous system
(CNS) lacking Cdc42 (Cappello et al., 2006; Chen et al., 2006).
At later stages than E10.5, the phenotypes of Cdc42 and Rac1
cko embryos appeared to be similar, with severe malformations
in NC target tissues evident in both mutants (Figures 1E–1G).
Craniofacial development was drastically impaired, as mani-
fested by a facial cleft. Although initial palatal shelve formation
appeared normal, palatal shelves developed poorly from E10.5
onward and failed to fuse. Constituents of the peripheral nervous
Cell Stem Cell
Small Rho GTPases in Neural Crest Development
Cell Stem Cell 4, 236–247, March 6, 2009 ª2009 Elsevier Inc. 237

Page 3

system (PNS), such as dorsal root ganglia (DRG) and other
ganglia of the sensory-somatic and autonomic nervous system,
were present but appeared all underdeveloped and reduced in
size (Figure 1F; data not shown). During normal development,
the heart outflow tract becomes divided into the aortic and
pulmonary outflow vessels via formation of the aortico-pulmo-
nary septum from the cardiac NC (Figure 1G). In contrast, the
aortico-pulmonary septum failed to form in both mutants leading
to a defect known as truncus arterious (TA), which is often seen
upon impaired development of the NC cell pool that contributes
to heart outflow tract formation (Wurdak et al., 2006).
Emigration of NC Cells from Neural Tube to Initial
Target Structures Appears Not to Be Affected
by Loss of Cdc42 and Rac1
As we observed abnormalities in all NC derivatives examined,
Cdc42 and Rac1 are likely to fulfill general rather than lineage-
specific roles in NC development. Rho GTPase function has
been linked to cell migration and adhesion (Fukata et al., 2003).
Therefore, we first investigated whether lack of Cdc42 and
Rac1 might lead to defective NC cell migration. Upon Wnt1-
Cre-mediated recombination of the ROSA26 Cre reporter allele
(R26R) (Soriano, 1999), b-galactosidase is stably expressed in
NC cells and their progeny, which allows in vivo fate mapping of
control and mutant NC cells (Hari et al., 2002; Ittner et al., 2005).
Using this system, we did not detect major migration deficits of
both Cdc42- and Rac1-deficient NC cells at E10.5 in craniofacial
areas, into the pharyngeal arches (PA), and in the trunk (Figures
2Aand2B).Moreover,atlaterstages,controlandmutantNCcells
havereachedtheirinitialtargets,includingnasofrontalstructures,
the eye, peripheral ganglia and nerves, and the proximal gut
(Figures 2C–2F; data not shown). However, in the mutants,
ganglia were smaller and distal parts of the gut lacked NC cell
colonization, indicative for reduced numbers of NC-derived cells
present in these structures (Simpson et al., 2007).
To confirm and quantify the migratory capacity of control and
mutant NC cells at early stages of NC development, we gener-
ated NC cell explants in vitro from neural tubes isolated at E9.5
(Lee et al., 2004). In this defined cell culture system, the extent
of NC outgrowth can be assessed 20 hr after initiation of the
cultures. The explants formed from Cdc42 and Rac1 cko neural
tubes were highly similar to their respective control explants.
Quantification of the outgrowth area (Hari et al., 2002) did
not reveal differences between control and cko explants
(Figure S2) and the cell density per area was not changed in
mutant explants (data not shown). Thus, impaired emigration
from the neural tube is not a primary cause of the phenotype
observed in Cdc42 and Rac1 cko embryos.
Cdc42 and Rac1 Are Not Required for Differentiation
of NC Cells
Apart from migration, failure of NC cells to properly differentiate
might contribute to the Cdc42 and Rac1 cko phenotype.
Possibly, loss of Cdc42 and Rac1 might interfere with cell-fate
decisions in the NC population, favoring differentiation into
a specific lineage at the expense of other fates. However, as
apparently all examined NC-derived structures were affected,
differentiation might rather be impaired at a global level than at
the level of distinct lineages. Therefore, we investigated whether
Figure2. Cdc42andRac1AreNotRequiredforMigrationofNCCells
to Initial Target Structures
(A and B) In vivo fate mapping of NC cells in E10.5 cko and control (co)
embryos. NC cells expressing b-galactosidase from the R26R Cre-reporter
allele were visualized by whole-mount X-Gal stainings. Cdc42 and Rac1 cko
embryos show normal distribution of NC cells. Arrowheads point to the first
PAs populated by NC cells. Enlarged areas marked by boxes show presence
of DRG and peripheral nerves.
(C–F) Histological sections of E12.5 Cdc42 control and cko embryos stained
with X-Gal show that NC cells have reached various targets such as nasofron-
tal structures (C), the eye (D), dorsal root ganglia (DRG, [E]), and proximal parts
of the midgut ([F], arrowhead). Note DRG size reduction and lack of distal gut
colonization (open arrowheads), pointing to reduced numbers of mutant
versus control NC-derived cells in these structures. Similar histological data
were obtained with Rac1 cko embryos (data not shown). Scale bars, 200 mm.
Cell Stem Cell
Small Rho GTPases in Neural Crest Development
238 Cell Stem Cell 4, 236–247, March 6, 2009 ª2009 Elsevier Inc.

Page 4

different neural and non-neural cell types were present in the
mutant NC tissues. Evidently, loss of Cdc42 or Rac1 was
compatible with overt neuronal differentiation, such as in DRG
and sympathetic ganglia (Figure 3B, data not shown). Early glial
differentiation was marked by the presence of fatty acid binding
protein 7 (Fabp7) in both control and mutant satellite glia
(Figure 3C). Furthermore, mutant NC-derived cells in the outflow
tractregionoftheheartstainedpositiveforsmoothmuscleactin,
indicating that smooth muscle formed correctly (Figures 3D and
3E). In addition to these differentiation markers, presence of
Sox10-positive cells was seen in DRG and peripheral nerves
(Figure 3A). Sox10 is expressed in NC stem and progenitor cells
and in the peripheral glial lineage. All markers investigated were
present at a comparable level to respective control animals in
correlation to the size of the evaluated tissues. Furthermore,
there were no signs of ectopic differentiation into a specific
lineage.
To further assess whether multipotency of NC cells was
impaired,wegeneratedNCexplantsfromCdc42ckoandcontrol
embryos in vitro. 20 hr after initial plating and outgrowth from the
neuraltube,NCexplantswerehighlyhomogeneousfortheNCSC
markers Sox10 and P75 (Figure 3F), but negative for differentia-
tion markers (Paratore et al.,2001; data not shown). Upon induc-
tion of differentiation over several days, mutant NC cells showed
Figure 3. Loss of Cdc42 and Rac1 Does Not
Interfere with Differentiation of NC Cells
(A) Expression of Sox10 in cells within the DRG of
control and cko embryos.
(B) Presence of neuronal differentiation in mutant
DRG.
(C) Satellite glia in DRG express the early glial
marker Fabp7.
(D and E) Although aortico-pulmonary septa in
the mutants fail to form, cardiac NC-derived cells
in the outflow tract region of the heart have gener-
atedsmooth muscle
(arrows). Near adjacent sections were stained
with X-Gal for localization of cardiac crest cells
revealed by in vivo fate mapping using the R26R
Cre-reporter allele. Blood cells are marked with
an asterisk (*).
(F) Cdc42-deficient eNCSCs express the NCSC
markers Sox10 and P75 after outgrowth from the
neural tube.
(G–J) Normal differentiation of Cdc42-deficient
eNCSCs into neuronal (G), glial (H) and smooth
muscle (I) cell types upon induction of differentia-
tion. Cdc42-deficient eNCSCs show normal cyto-
skeletalmorphology(J).Scalebars:(A)–(E),50mm;
(F)-(J), 10 mm.
cells expressing Sma
normal differentiation into neuronal, glial,
and smooth muscle cell types (Figures
3G–3I). Moreover, no abnormalities were
found in the cytoskeletal morphology of
mutanteNCSCs(Figure
together, we did not detect any signs of
reduced or altered
Cdc42- and Rac1-deficient NC cells
in vivo and of differentiation of Cdc42-
3J).Taken
differentiationin
deficient NC cells into diverse derivatives in vitro. We conclude
that Cdc42 and Rac1 do not influence the capacity of NCSCs
to acquire multiple different cell fates and to undergo initial steps
of differentiation.
Cdc42 and Rac1 Are Essential for Mitotic Activity
and Cell-Cycle Control in NC Target Structures
The broad phenotypes in mutant NC derivatives are consistent
with the hypothesis that loss of Cdc42 and Rac1 both lead to
a depletion of the NC cell pool. Accordingly, Cdc42 and Rac1
might play a role in promoting proliferation or they might be
required for survival of NC cells. Assessing apoptosis within
NC tissues at E10.5 and E11.5, however, did not reveal signs
of impaired survival in Cdc42 or Rac1 cko embryos (data not
shown). Several reports have implicated Rho GTPases in cell-
cycle control (Narumiya and Yasuda, 2006). Thus, Cdc42 and
Rac1 might be required for proliferation of NC cells. A reduction
in mitotic activity would lead to a general loss of NC cells,
reducing the overall cell pool size necessary for proper formation
of NC derivatives. To investigate this possibility, we applied an
antibody against phospho-histoneH3 (pH3) to examine prolifer-
ation in PA1 and -2 of Cdc42 and Rac1 cko embryos that
expressed b-galactosidase as a NC lineage label. pH3 is
detected during G2 to M transition of the cell cycle when
Cell Stem Cell
Small Rho GTPases in Neural Crest Development
Cell Stem Cell 4, 236–247, March 6, 2009 ª2009 Elsevier Inc. 239

Page 5

chromatin condensation occurs. Strikingly, while there was no
statistically significant difference at E10.5 (Figures S3A and
S3D), we saw a reduction in the number of pH3-positive cells
by 24.5% (p < 0.05) in NC-derived tissue in Cdc42 mutants
and by 19.6% (p < 0.05) in Rac1 mutants at E11.5 (Figures 4A,
4D, 4G, and 4H). As cell density per area was unchanged
(Figure S4), these results demonstrate a reduced mitotic activity
of mutant NC progenitor cells. Next, we examined whether loss
of Cdc42 or Rac1 induced premature exit from the cell cycle in
NC cells. To this end, a BrdU pulse was given 24 hr before anal-
ysis,followedbyquantificationoftherelativenumberofcellsthat
had incorporated BrdU into their DNA. BrdU-positive cells that
did not express the proliferation marker Ki67 anymore have
exited the cell cycle at the time point of analysis. Apart from
the PA1 and -2, we examined the DRG of control and cko
embryos in order to analyze non-neural as well as neural deriva-
tives of NC origin. At E10.5, no significant changes in cell-cycle
exit were found in DRG and in the PAs of Cdc42 and Rac1 cko
embryos (Figures S3B, S3C, S3E, S3F, and S3G). In contrast,
we observed a drastic increase in cell-cycle exit at E11.5 in
Figure 4. Reduced Mitotic Activity and
Increased Cell-Cycle Exit in NC-Derived
Structures of Cdc42 and Rac1 cko Embryos
at E11.5
(A and D) Reduced numbers of mitotic cells in
mutant PA1 and -2 as detected by pH3 staining
at E11.5. (G and H) PA1 tissue within quantified
regions is largely X-Gal positive, demonstrating
the neural crest origin of the vast majority of cells.
Black and white arrows mark pharyngeal ecto-
derm and core mesoderm. These tissues are not
NC-derived and were excluded from the quantifi-
cation. (B and E) Cell-cycle exit was quantified
24 hr after a BrdU pulse. Cells having exited the
cell cycle show incorporation of BrdU but no
expression of the proliferation marker Ki67. NC
cells that have populated PA1 and -2 show an
increase in cell-cycle exit at E11.5. (C,F, and I)
Cells in mutant DRG show significantly increased
cell-cycle exit at E11.5. Values represent mean ±
SEM. Scale bars, 50 mm. *p < 0.05; ***p < 0.005.
both mutants: the number of DRG cells
that had exited the cell cycle was
increased by 50.1% (p < 0.005) in Cdc42
and by 27.5% (p < 0.05) in Rac1 mutants
(Figures 4C, 4F, and 4I). In PA1 and
-2, overall numbers of cell-cycle exit
were comparatively low. However, the
increase in cell-cycle exit amounted to
401.1% (p < 0.05) in Cdc42 and 123.1%
(p < 0.05) in Rac1 mutants, respectively
(Figures 4B and 4E).
Taken together, we observed a reduc-
tion of mitotically active cells in both
mutants, accompanied by a significant
increase in cell-cycle exit. Interestingly,
these changes did not occur early in
NC development, as mutant embryos at
E10.5 showed no major alterations. This
suggests that during early stages of NC development, NC cell
proliferation might occur independently of Cdc42 and Rac1.
These small Rho GTPases apparently turn into crucial regulators
of proliferation, however, after NC-derived cells have reached
their initial targets.
Stage-Specific EGF Responsiveness Regulating Small
Rho GTPase-Dependent Self-Renewal and Proliferation
Knowledge of the mitogen acting upstream of Cdc42 and Rac1
might be necessary to understand how NC cells become depen-
dent on these small Rho GTPases over time. In the CNS, expres-
sionofepidermalgrowthfactorreceptor(EGFR)increasesduring
development, mediating a transition from mitotic FGF to EGF
responsiveness in neural stem and progenitor cells (Ciccolini,
2001; Lillien and Raphael, 2000). To assess whether similar
sequential changes occur during neural crest development, we
first monitored the spatiotemporal expression of EGFR in vivo.
AtE9.5,EGFRexpressionwasgenerallylow,andSox10-positive
neural crestcellsthat migrated awayfrom theneural tubedid not
display EGFR expression (Figure S5A). At E10.5, EGFR
Cell Stem Cell
Small Rho GTPases in Neural Crest Development
240 Cell Stem Cell 4, 236–247, March 6, 2009 ª2009 Elsevier Inc.

Page 6

expression—although still relatively weak and heterogeneous—
became apparent in DRG, the gut, and PA1 (Figures S5B–S5D).
Strikingly, at E11.5, EGFR was very strongly and broadly ex-
pressed in DRG, the gut, and PA1. Moreover, Sox10-positive
cells in DRG and the ENS coexpressed EGFR. To corroborate
these findings on the cellular level, explant cultures of NCSCs
werepreparedfromdistinctdevelopmentalstages(Experimental
Procedures). Both E9.5 neural tube and E11.5 organotypic gut
explant cultures consisted of undifferentiated cells that ex-
pressed the NCSC markers Sox10/P75 (Figure S6; Paratore
et al., 2001). Moreover, 99.6 ± 0.6% of all P75-positive cells in
gut explants also expressed a4-integrin (data not shown), which
identifies these cells as gut NCSCs (gNCSCs) (Bixby et al., 2002;
Kruger et al., 2003). In agreement with the data obtained in vivo,
eNCSCs from E9.5 neural tube explants did not express EGFR
(Figures 5A and 5B). In contrast, virtually all P75-positive cells
in explants of E11.5 organotypic gut cultures expressed EGFR
(Figures 6A and 6B), demonstrating that, unlike eNCSCs,
Figure 5. Lack of EGF Responsiveness in eNCSCs
(A and B) eNCSC explants were generated by outgrowth from
E9.5 neural tubes. eNCSCs homogeneously express the NCSC
markers Sox10 and P75 (A and B00), but do not express EGF
receptor (EGFR) (B and B0). (B) is an overlay of the stainings shown
in (B0) and (B00).
(C) Following outgrowth from the neural tube, wild-type eNCSCs
were challenged with either FGF2 or EGF. The number of P75-
expressing mitotic eNCSCs in response to these factors was
quantified by pH3 staining. While FGF2 led to an increase in
mitotic cells, EGF-treated mitotic cells remained at levels compa-
rable to nontreated control cells (n.a.). White arrowheads mark
pH3-positive nuclei.
(D) eNCSCs were expanded and maintained in the presence of
Wnt1, BMP2, and FGF2. Numbers of cells expressing the stem
cell marker Sox10 and the proliferation marker Ki67 are compa-
rable in Cdc42 mutant (Cdc42 cko) and control (co), demon-
strating that self-renewal and proliferation of the early population
of NCSCs are not affected. Values represent mean ± SEM. Scale
bars, 10 mm. **p < 0.01.
gNCSCs express EGFR. To assess whether stage-
specific EGFR expression correlated with factor
responsiveness, eNCSCs and gNCSCs were exposed
to FGF2 or EGF in defined culture conditions. In
eNCSCs from neural tube explants, FGF2, but not
EGF, promoted proliferation, as assessed by quantifi-
cation of mitogenic P75-positive cells (Figure 5C). In
contrast, gNCSCs were responsive to both mitogenic
FGF2 as well as EGF (Figure 6C). Thus, EGF is
a stage-specific growth factor during NC develop-
ment.
Since EGF is one of the factors that can lead to
Cdc42 and Rac1 activation (Fanger et al., 1997; Sini
et al., 2004), we tested whether EGF-mediated prolif-
eration in gNCSCs might involve Cdc42 and/or Rac1.
To this end, we measured the relative number of
mitotic cells in explants of gNCSCs derived from
either control, Cdc42 cko, or Rac1 cko E11.5 gut.
Strikingly, proliferation of both Cdc42- and Rac1-
deficient P75-positive cells was reduced to levels
found in the absence of EGF (Figure 6C), demonstrating
that EGF-induced proliferation requires Cdc42 and Rac1. In
contrast, FGF2—that, unlike EGF, is a mitogen for both
eNCSCs and gNCSCs—promoted proliferation independently
of Cdc42 and Rac1. Thus, increased cell-cycle exit and altered
proliferation in Cdc42 and Rac1 mutants in vivo (Figure 4) can
be explained by small Rho GTPase-dependent EGF signaling
active at later, but not early, stage NC development.
The above experiments reveal a mitogenic role of EGF/small
Rho GTPases in cultures containing NCSCs, but they do not
directly address whether actual stem cell self-renewal is
controlled by this signaling cascade. At early stages of NC
development, combinatorial Wnt and BMP signaling, in the
presence of FGF2, suppresses differentiation and maintains
self-renewing multipotent eNCSCs from neural tube explants
(Kleber et al., 2005). As expected from our in vivo analysis
(Figure 2) and given that these early stage cells are not respon-
sive to EGF (Figure 5), deletion of Cdc42 and Rac1, respectively,
Cell Stem Cell
Small Rho GTPases in Neural Crest Development
Cell Stem Cell 4, 236–247, March 6, 2009 ª2009 Elsevier Inc. 241

Page 7

did not affect self-renewing proliferation of eNCSCs (Figure 5D;
Table S1). Indeed, neither the percentage of Sox10-expressing
eNCSCs nor the number of mitotically active cells was dimin-
ished in mutant stem cell cultures (Table S1). Unlike for early
stage eNCSCs, factors instructively promoting stem cell mainte-
nance at later stages of NC development remain to be identified
(Kleber et al., 2005). Nevertheless, the self-renewal capacity of
later stage NCSCs can be monitored in nonadhesive sphere
cultures by serial sphere passaging (Molofsky et al., 2003).
Therefore, to investigate the role of Cdc42 in gNCSC self-
renewal, we generated ENS spheres from control and cko
embryos at E12.5. As the embryonic gut might contain addi-
tional types of sphere-forming cells not related to the NC, we
Figure 6. EGF Acts via Cdc42 and Rac1 to
Promote Cell Proliferation of Gut NC Stem
and Progenitor Cells
gNCSC explants were generated by outgrowth
from E11.5 gut. In addition to the NCSC markers
Sox10 and P75 (A and B00), explant cells homoge-
neously express the EGF receptor (EGFR; [B] and
[B0]). (B) is an overlay of the stainings shown in (B0)
and (B00). (C) Following outgrowth from the gut,
control (co), Cdc42 cko, and Rac1 cko enteric
NC cells were challenged with either FGF2 or
EGF. The number of P75-expressing mitotic
gNCSCs in response to these factors was quanti-
fied by pH3 staining. FGF2 induced proliferation in
both control and mutant cells. However, EGF
induced proliferation only in control, but not
Cdc42- and Rac1-deficient gNCSCs. Pictures for
treatment with EGF are shown. White arrowheads
mark pH3-positive nuclei. (D) Control and Cdc42-
deficient enteric spheres were generated from
E12.5 guts. The percentage of spheres with
a diameter larger than 100 mm is lower in mutants
(1?sphere size, 2?sphere size),indicating reduced
proliferation. In addition, the self-renewal capacity
is significantly reduced in mutants. Values repre-
sent mean ± SEM. Scale bars, 10 mm. *p < 0.05;
**p < 0.01; ***p < 0.005.
tracked
R26R reporter allele so that spheres
formed from recombined NC cells could
be identified based on X-Gal staining. In
agreement with fewer NC-derived cells
being present in cko NC target structures
(Figure 2), the number of primary spheres
generated from Cdc42 cko guts (sphere
frequency) was reduced by 62.3% (p <
0.005) as compared to control spheres.
Importantly, upon subcloning the amount
of mutant secondary spheres that formed
from recombined primary spheres was
decreased by 84.6% (p < 0.05), demon-
strating a drastically decreased self-
renewal capacity of gNCSCs (Figure 6D).
Moreover, while control spheres could
be further subcloned, mutant ENS cells
did not form tertiary spheres (data
NC-derived cellsusingthe
not shown). While sphere passaging reflects stem cell self-
renewal, the sphere size is indicative of progenitor cell prolifer-
ation. Average sphere size was reduced in mutant spheres
(primary spheres: control 137 ± 24 mm, mutant 89 ± 36 mm;
secondary spheres: control 150 ± 50 mm, mutant 92 ± 17 mm).
In particular, spheres with a diameter larger than 100 mm were
reduced by 45.0% in primary and by 43.6% in secondary
mutant spheres, reflecting reduced proliferation of ENS stem
and progenitor cells in the absence of Cdc42 (Figure 6D), con-
firming the data obtained with adhesive cko gut explant cultures
(Figure 6C). ENS sphere formation from Rac1-deficient ENS
cells was equally reduced, and spheres appeared smaller in
size, too (data not shown), pointing to a similar requirement of
Cell Stem Cell
Small Rho GTPases in Neural Crest Development
242 Cell Stem Cell 4, 236–247, March 6, 2009 ª2009 Elsevier Inc.

Page 8

Cdc42 and Rac1 in gNCSC self-renewal and progenitor cell
proliferation.
Cdc42 and Rac1 Appear to Share a Common Pathway,
with Rac1 Being Required for Cdc42 Activation
The similar phenotypes of Cdc42 and Rac1 cko embryos in vivo
and the indistinguishable roles of these small Rho GTPases in
EGF-dependent cell-cycle regulation of gNCSCs suggests that
Cdc42 and Rac1 might involve overlapping signaling pathways.
To address this issue, we first compared alterations in down-
stream signaling events upon Cdc42 versus Rac1 mutation in
theNCbyexaminingtheexpressionprofileofseveralkeyregula-
torsofcell-cycledynamics.Quantitativereal-timeRT-PCR(qRT-
PCR) was carried out with tissue from PA1 and -2 of Cdc42 and
Rac1 cko and control embryos at E11.5. While expression of
CyclinD1 andother regulators was notchanged, P21Cip1expres-
sion levels were significantly increased in both mutants
(Figure S7), suggesting that Cdc42 and Rac1 might act as
suppressors of P21Cip1in NC-derived tissue. qRT-PCR analysis
also revealed that loss of Rac1, but not of Cdc42, resulted in
increased expression of Pten and c-Myc that have been associ-
atedwithregulationofstemcellself-renewal(Groszeretal.,2006;
Wilson et al., 2004). Although for technical reasons we were
unable to confirm on the protein level the ?1.5-fold induction of
P21Cip1mRNA levels in the cko embryos, the overall data are
consistent with the idea that Cdc42 and Rac1 might act in
NC cells through partially overlapping downstream signaling
targets. To directly assess whether Cdc42 and Rac1 pathways
overlap, we examined levels of active (i.e., GTP-bound) forms
of Cdc42 and Rac1 in the respective mutants. To this end,
gNCSCs carrying two floxed alleles of either Cdc42 or Rac1
Figure 7. Cdc42 and Rac1 Appear to Act in a Common
Pathway, with Rac1 Being an Upstream Activator of
Cdc42
Western blots and pull-down assays using GST-PAK-CD
constructs were done with lysates from Adenovirus-Cre-
infected ENS spheres. Infected spheres were wild-type (co)
or carried two floxed alleles for Cdc42 (Cdc42 ko) and Rac1
(Rac1 ko), respectively.
(A) Total Cdc42 was significantly reduced in Cdc42 ko as
compared to control lysates (left panel). Levels of active
GTP-bound Cdc42 were reduced in both Cdc42 ko and
Rac1 ko as compared to control (right panel).
(B) Total levels of Rac1 were significantly reduced in lysates
from Rac1 ko (left panel). Activity of Rac1 was significantly
lower in Rac1 ko, but not in Cdc42 ko, as compared to control
(right panel). ***p < 0.005.
were expanded in sphere cultures in the presence
of EGF, followed by adenovirus-Cre-mediated
gene deletion. Cdc42 ablation resulted in loss of
total Cdc42 protein and of active Cdc42, as ex-
pected, whereas Rac1 activation was unchanged
in Cdc42-deficient as compared to wild-type cells
(Figure 7). Intriguingly, however, loss of Rac1
caused reduced activation of both Rac1 as well as
Cdc42, indicating that Cdc42 activation occurs
downstream of Rac1. Thus, activation of Cdc42
andRac1appearstorelyonoverlappingpathways,
likely explaining the similar phenotypes found upon Cdc42 and
Rac1 inactivation in the NC.
DISCUSSION
Thisreportprovidesthefirstevidenceforacrucialinvolvementof
two small Rho GTPases, Cdc42 and Rac1, in NC development.
Using mice with a deletion of either Cdc42 or Rac1 in the NC,
we show that proliferation and the balance between mitotically
activeandquiescentcellsisCdc42andRac1dependent.Impor-
tantly, growth control during NC development is stage specific:
while self-renewing proliferation of eNCSCs is Cdc42 and Rac1
independent,NCstemandprogenitor cells,afterhavingreached
their initial targets, require these small Rho GTPases for cell-
cycle control. These developmental changes are associated
with altered responsiveness of NC cells to mitogenic growth
factors: eNCSCs emigrating from the neural tube are first
FGF2, but not EGF, responsive. At later stages, however, NC
cells as analyzed in the ENS start to express the EGF receptor
and acquire EGF responsiveness. Thereby, EGF, but not FGF2,
acts through Cdc42/Rac1 to promote proliferation. Thus,
although endowed with a similar set of stem cell features, early
and late NCSCs can be distinguished based on different mech-
anisms regulating proliferation.
Proliferation Control in NC Stem and Progenitor Cells
So far, knowledge on how NC cell proliferation is controlled has
been limited. In Xenopus, the helix-loop-helix transcription
factor Id3 acts as a cell-intrinsic regulator of NC cell proliferation
and survival (Kee and Bronner-Fraser, 2005). However, in
contrast to what we observed for Cdc42 and Rac1 in mice,
Cell Stem Cell
Small Rho GTPases in Neural Crest Development
Cell Stem Cell 4, 236–247, March 6, 2009 ª2009 Elsevier Inc. 243

Page 9

Id3 in Xenopus appears to be involved already at the stage of NC
induction. In mice, several Id genes with overlapping functions
exist, but single and double knockout mice display milder
phenotypes than obtained upon loss of Id3 function in Xenopus.
Thus, a direct implication of Id genes in Cdc42- and Rac1-
dependent NC proliferation appears unlikely. With respect to
extracellular regulators of early NC cell proliferation, FGF2
promotes cell-cycle progression in neural tube explant cultures,
as shown here (Figure 5), and the combined activities of Wnt and
BMP in the presence of FGF2 allows expansion of multipotent
eNCSCs by symmetrical cell division (Kleber et al., 2005). The
latter involves canonical Wnt signaling via its downstream
signaling component b-catenin (S.B.-M. and L.S., unpublished
data), which, in astrocytes and skin progenitor cells, can be
regulated by Cdc42 (Etienne-Manneville and Hall, 2003; Wu
et al., 2006). In NC cells, however, b-catenin activity is only
observed during early stages (Kleber et al., 2005), while it
decreases in postmigratory structures such as DRG that are
affected in Cdc42 and Rac1 cko embryos. Moreover, we
demonstrate that Wnt, BMP, and FGF2 are able to support
eNCSC self-renewal even in the absence of Cdc42 or Rac1
(Figure 5; Table S1). Thus, FGF2 and Wnt/b-catenin signaling
in early NC cells is not associated with Cdc42/Rac1-mediated
cell-cycle control.
At later stages, FGF2 is still a mitogenic signal for NC cells.
However, unlike eNCSCs, later stage NC stem and progenitor
cells also become sensitive to mitogenic EGF due to upregula-
tion of the EGFR (Figure 6; Figure S5). This is reminiscent of
developmental changes occurring during CNS development,
when neural stem cells switch from FGF2 to FGF2/EGF respon-
siveness (Ciccolini, 2001; Lillien and Raphael, 2000). Intriguingly,
while FGF2-induced proliferation remains Cdc42/Rac1-inde-
pendent during NC development, the mitogenic response of
late stage gNCSCs to EGF can be abolished by either Cdc42
or Rac1 inactivation. Thus, EGF can be placed upstream of
both Cdc42 and Rac1 in regulating cell-cycle progression of
NC cells that have reached their initial targets.
The analogous proliferation phenotypes obtained in vivo and
in cell culture upon Cdc42 and Rac1 deletion in the NC raises
the possibility that, downstream of EGF, Cdc42 and Rac1 share
a signaling pathway to activate common targets. Indeed, the
expression of P21Cip1was similarly upregulated in NC tissue
from Cdc42 and Rac1 cko embryos, while the expression of
several other cell-cycle regulators remained unaffected in both
mutants (Figure S7). Conceivably, Cdc42 and Rac1 act through
regulation of P21Cip1to control cell-cycle progression of late
stage NC cells. However, Rac1, but not Cdc42, deletion resulted
in increased levels of Pten and c-Myc expression, indicating
that Rac1 might have a somewhat broader activity spectrum
than Cdc42. This is consistent with our finding that Rac1
appears to act upstream of Cdc42: levels of active Rac1 were
unchanged in Cdc42-deficient NC cells, while active Cdc42
levels were reduced in the Rac1 ko (Figure 7). Such cross-
activation between Cdc42 and Rac1 has previously been
established in fibroblasts, although with reverse hierarchical
order (Czuchra et al., 2005). Thus, Cdc42 and Rac1 can func-
tionally interact, which in the NC appears to involve Cdc42
activation by Rac1 in a shared pathway mediating mitogenic
EGF signaling.
In addition to EGF, several other receptor tyrosine kinases can
signal through Rho GTPases. Of these, PDGFRa plays a role in
cranial and cardiac NC development, which has, however, not
been attributed to the control of proliferation (Tallquist and Sor-
iano, 2003). ErbB2/3 activation by the growth factor neuregulin
(NRG) promotes gliogenesis in NC cells and supports Schwann
cell proliferation and differentiation. Interestingly, NRG-depen-
dent Schwann cell proliferation is mediated by Cdc42, but not
Rac1, pointing to distinct roles of these molecules in Schwann
cell development (Benninger et al., 2007). The cell-lineage-
specific function of NRG and the differential activation of
Cdc42 versus Rac1 by NRG make it improbable that the effects
described in our study involve NRG signaling. Finally, integrin
signaling can also act as an activator of Cdc42 and Rac1
(Schwartz and Shattil, 2000). However, ablation of integrinb1 in
the NC led to a milder phenotype than observed in Cdc42
and Rac1 cko embryos, in that mutant mice were viable up to
3 weeks of age with developmental abnormalities restricted to
the PNS (Pietri et al.,2004). In sum,we cannot exclude that mito-
genic Cdc42 and Rac1 are activated by distinct local signals in
different NC derivatives. Such signals might, for instance, modu-
late mitogenic EGF and contribute to the distinct extent of
cell-cycle exit observed in mutant PA versus DRG (Figure 4).
However, the broad phenotypes observed upon small Rho
GTPase inactivation, similarly affecting virtually all NC deriva-
tives, rather, speak for a general mechanism of Cdc42- and
Rac1-mediated growth control at later stages of NC develop-
ment. According to this model, EGF and Rac1/Cdc42 form
a signaling cascade controlling the pool size of the NC popula-
tion after its emigration from the neural tube, although this has
to be confirmed by conditional inactivation of EGF signaling in
NC derivatives.
Although Cdc42 and Rac1 have been implicated in migration
control in other systems (Fukata et al., 2003), a cellular pheno-
type in the cko embryos became apparent only after E10.5,
and emigration from the neural tube was apparently not affected
in the absence of these small Rho GTPases. The in vivo situation
was recapitulated in vitro by mutant eNCSCs that did not
display any overt migration deficiencies in neural tube explant
cultures. These findings are in agreement with recent reports
demonstrating normal migration ability of Cdc42 and Rac1 null
mutant fibroblasts, respectively (Czuchra et al., 2005; Vidali
et al., 2006). Nevertheless, we cannot exclude a role of Rac1/
Cdc42 in migration, such as during morphogenesis of craniofa-
cial structures at later stages of NC development or in the full-
length colonization of the gut during ENS formation (Stewart
et al., 2007). Indeed, distal gut portions were not or less popu-
lated by NC-derived cells in mutants as compared to the control
(Figure 2), which—in addition to defective gNCSC proliferation
and maintenance—could be associated with migration deficits.
Of note, however, proliferation and directed migration are intri-
cately linked during ENS development, with proliferation being
the key mechanism driving invasion of the gut by NC cells
(Simpson et al., 2007). Thus, small Rho GTPases might elicit
associated functions in migration and proliferation. Importantly,
the amenability of appropriate culture systems allowed us to
demonstrate on the cellular level that Rac1 and Cdc42 regulate
NCSC proliferation cell autonomously and independently of
migration.
Cell Stem Cell
Small Rho GTPases in Neural Crest Development
244 Cell Stem Cell 4, 236–247, March 6, 2009 ª2009 Elsevier Inc.

Page 10

Cdc42 and Rac1 Functions Are Cell-Type
and Stage Specific
Although small Rho GTPases have previously been implicated in
regulating stem cells other than NCSCs, they do not execute
auniversalstemcellfunction.Rather,theyshowfunctionaldiver-
sity depending on the type of Rho GTPase, the cell type, and the
time of action. In the CNS, deletion of Cdc42 leads to progenitor
cell depletion, but this is an effect secondary to gradual loss of
polarity protein complex and adherens junctions (Cappello
et al., 2006). As a consequence, mutant cells fail to self-renew
at the apical site of the neuroepithelium and, therefore, acquire
a neurogenic fate. Such a mechanism unlikely contributes to
the phenotypes described in our study, as expression of polarity
protein complex was unaltered in Cdc42- or Rac1-deficient
postmigratoy NCderivatives (data notshown).Inthehematopoi-
etic system, Rac1 is required for the retention of hematopoietic
stem cells and progenitors (HSC/Ps) in their microenvironmental
niche, but an effect on proliferation was only observed in vitro
and not in vivo (Cancelas et al., 2005). Deletion of Cdc42 in
HSC/Ps also led to impaired retention, but mutant HSC/Ps
showed increased cell-cycle entry, thus increasing the number
of HSC/Ps (Yang et al., 2007). Apparently, in the hematopoietic
system, the function of Cdc42 in cell-cycle control is the inverse
of its role in the NC. In skin progenitor cells, Rac1 plays a very
different role from Cdc42. Similar to its role in late stage NCSCs,
Rac1 is crucial for maintenance of skin stem cells present in the
hair follicle, but possibly not of interfollicular epithelial stem cells
(Benitah et al., 2005; Castilho et al., 2007; Chrostek et al., 2006).
In contrast, unlike in NCSCs, Cdc42 controls differentiation of
progenitor cells into hair follicle cells (Wu et al., 2006). In sum,
there is increasing evidence that the function of a given small
Rho GTPase is specific for the type of stem and progenitor cells
(Wang and Zheng, 2007). The present study now also reveals
stage-specific roles of small Rho GTPases in stem cells. During
NC development, Cdc42 and Rac1 activities mark a transition in
NCSC growth control, acting as critical cell-cycle regulators of
virtually the entire pool of NC cells after emigration from the
neural tube. Similarly, NCSCs from different developmental
stages display intrinsic differences in their responsiveness to
fate-promoting cues (Bixby et al., 2002; Kleber et al., 2005).
However, while signals inducing differentiation are also differen-
tially interpreted depending on the location, the control of prolif-
eration appears to be more general in late stage NC stem and
progenitor cells, leading to broad developmental defects in
NC derivatives upon Cdc42 and Rac1 inactivation.
EXPERIMENTAL PROCEDURES
Generation of cko Mice
Animals homozygous for the floxed Cdc42 allele (Cdc42lox/lox) (Wu et al., 2006)
werematedwithmiceheterozygousforthefloxedallelethatadditionallycarried
the Wnt1-Cre transgene (Danielian et al., 1998) (Wnt1-Cre/Cdc42lox/wt).
Offspring with a Wnt1-Cre/Cdc42lox/loxgenotype were termed Cdc42 cko
embryos, while littermates lacking the Wnt1-Cre transgene or carrying a wild-
type Cdc42 or Rac1 allele did not exhibit any overt phenotype, thus serving
ascontrolanimals.Forinvivofatemappingofrecombinedcells,theR26Rallele
(Soriano, 1999) was bred into Cdc42lox/loxmice, which were mated with Wnt1-
Cre/Cdc42lox/wtmiceto obtainCdc42ckoand controlembryosthatexpressed
b-galactosidase upon Cre-mediated recombination. The same breeding
strategy was used for Rac1lox/loxanimals (Chrostek et al., 2006) to obtain
Rac1ckoandcontrolembryos.GenotypingwasdonebyPCRongenomicDNA.
Histochemistry, Immunofluorescence, and X-Gal Staining
Hematoxylin-eosinstainings(H&E)weredoneonparaffinsections.Immunoflu-
orescence stainings were done on paraffin (7 mm) or cryo (10–12 mm) sections.
For all antibody stainings on sections, except for Cdc42, Rac1, and Sma,
antigen retrieval was done with 10 mM citrate buffer in a steam cooker device
at 110?C for 5 min. Fixation was done with 4% paraformaldehyde (PFA) for
sections,and4%formaldehydeforcellsinPBSfor10minatroomtemperature
with the following exceptions: fixation for Cdc42 staining on cryosections was
with prechilled 4% PFA for 10 min on ice. Fixation for Rac1 stainings on cryo-
sectionswaswithprechilled4%PFA/2%aceticacidinPBSfor10minfollowed
by prechilled ethanol:acetic acid (95:5) for 2 min, all on ice. Antibodies and
substrates used for stainings are given in Table S2. BrdU incorporation was
done according to the manufacturer’s instructions (Roche). X-Gal (Applichem)
stainings were done according to standard protocols.
Cell Culture
eNCSC explant cultures from E9.5 trunk neural tubes were done as described
(Lee et al., 2004). Migratory capacity of eNCSCs in vitro (migration index) was
determined asdescribed (Hari et al.,2002).Cultures were incubated in defined
medium (DM), supplemented with 10 ng/ml FGF2. For eNCSC maintenance
experiments, medium was switched after 20 hr to DM with 10 ng/ml FGF2,
200 ng/ml Wnt1, and 50 ng/ml BMP2. To assess growth factor response of
eNCSCs, medium was withdrawn after 16 hr, cells were starved for 2 hr in
DM, and subsequently pulsed with DM plus 50 ng/ml FGF2 or 50 ng/ml EGF,
respectively, for additional 2 hr. gNCSC explant cultures were obtained from
isolated E11.5guts. Culture conditionswere similar toeNCSCexplant cultures
with the following exceptions: cultures were incubated for 24 hr in DM with
10 ng/ml FGF2, 10 ng/ml EGF, and 10 ng/ml GDNF. To assess growth factor
response of gNCSCs, medium was withdrawn after 24 hr, and cells were
starved for 2 hr in DM and subsequently pulsed with DM plus 50 ng/ml FGF2
or 50 ng/ml EGF, respectively, for an additional 3 hr. gNCSC spheres were
grownfromE12.5controlandmutantENScells.Isolatedgutsweredissociated
into single cells with 1 mg/ml collagenase type I (Worthington) dissolved in
HBSSw/oCaandMg.SelectivegrowthofNC-derivedcellswasaccomplished
with NC-specific sphere (NCS) medium and verified by visualization of the
lineagemarkerb-galactosidase.NCSmediumcontainedDMEM-F12,20ng/ml
FGF2, 20 ng/ml IGF1, 20 ng/ml EGF, 10 ng/ml GDNF (all Peprotech), 1% N2
supplement, 2% B27, 1% Penicillin/Streptomycin (all Invitrogen), 50 mM
2-mercaptoethanol, and 15% chicken embryonic extract. Poly(2-hydroxyethyl
methylacrylate)(Sigma)wasusedtocoatcultureplatestoeliminateadhesionof
cells to the plastic surface. To assess self-renewal, secondary spheres were
cultured from defined numbers of cells obtained from dissociated primary
spheres. The average number of cells/primary spheres was used to calculate
the ratio between secondary and primary spheres.
Quantitative Real-Time RT-PCR
Total RNA from E11.5 PA1/2 cells was isolated with the TRIZOL protocol
(Invitrogen) according to the manufacturer’s instructions, and 2 mg was used
for each cDNA synthesis. cDNA synthesis was performed using Ready-
To-Go You-Prime First-Strand Beads (Amersham). The same amount of
cDNAwasusedforeachPCRreaction.qRT-PCRwasperformedinaLightcyler
(Applied Biosystems) using Master SYBR Green kit (ABgene). Each result
was normalized by the housekeeping 18S gene expression. The primer pairs
used for each PCR can be found in Table S3.
Rac1-GTP and Cdc42-GTP Pull-Down Assays
gNCSC spheresweregrown from E12.5 wild-type, Cdc42lox/lox,and Rac1lox/lox
guts as mentioned above. Adenovirus-Cre-mediated recombination was done
by infecting single cell suspensions of secondary spheres. Infected cells were
cultured to form tertiary spheres. Medium was washed off 4 days after infec-
tion and replaced with virus-free NCS medium for additional 2 days. GST-
PAK-CD constructs were provided by J. Collard (The Netherlands Cancer
Institute, Amsterdam, Holland). Rac1 and Cdc42 activities were measured
as described (Benninger et al., 2007).
Statistical Analysis
Each experiment was performed with at least three independent samples.
Statistical significance was tested with the unpaired Student’s t test.
Cell Stem Cell
Small Rho GTPases in Neural Crest Development
Cell Stem Cell 4, 236–247, March 6, 2009 ª2009 Elsevier Inc. 245

Page 11

SUPPLEMENTAL DATA
The Supplemental Data include seven figures and three tables and can
be found with this article online at http://www.cell.com/cell-stem-cell/
supplemental/S1934-5909(09)00054-X.
ACKNOWLEDGMENTS
We thank A. McMahon and P. Soriano for providing transgenic animals and
K. Wycisk for experimental support. This work was supported by the Swiss
National Science Foundation, the National Center of Competence in Research
‘‘Neural Plasticity and Repair,’’ the Vontobel Foundation, Oncosuisse/Swiss
Cancer League, and the ETH and University of Zurich.
Received: December 5, 2007
Revised: October 2, 2008
Accepted: January 26, 2009
Published: March 5, 2009
REFERENCES
Benitah, S.A., Frye, M., Glogauer, M., and Watt, F.M. (2005). Stem cell deple-
tion through epidermal deletion of Rac1. Science 309, 933–935.
Benninger, Y.,Thurnherr,T.,Pereira, J.A., Krause, S.,Wu,X.,Chrostek-Grash-
off, A., Herzog, D., Nave, K.A., Franklin, R.J., Meijer, D., et al. (2007). Essential
and distinct roles for cdc42 and rac1 in the regulation of Schwann cell biology
during peripheral nervous system development. J. Cell Biol. 177, 1051–1061.
Bixby, S., Kruger, G.M., Mosher, J.T., Joseph,N.M., and Morrison,S.J. (2002).
Cell-intrinsic differences between stem cells from different regions of the
peripheral nervous system regulate the generation of neural diversity. Neuron
35, 643–656.
Bondurand, N., Natarajan, D., Barlow, A., Thapar, N., and Pachnis, V. (2006).
Maintenance of mammalian enteric nervous system progenitors by SOX10
and endothelin 3 signalling. Development 133, 2075–2086.
Cancelas, J.A., Lee, A.W., Prabhakar, R., Stringer, K.F., Zheng, Y., and
Williams, D.A. (2005). Rac GTPases differentially integrate signals regulating
hematopoietic stem cell localization. Nat. Med. 11, 886–891.
Cappello, S., Attardo, A., Wu, X., Iwasato, T., Itohara, S., Wilsch-Brauninger,
M., Eilken, H.M., Rieger, M.A., Schroeder, T.T., Huttner, W.B., et al. (2006).
The Rho-GTPase cdc42 regulates neural progenitor fate at the apical surface.
Nat. Neurosci. 9, 1099–1107.
Castilho, R.M., Squarize, C.H., Patel, V., Millar, S.E., Zheng, Y., Molinolo, A.,
and Gutkind, J.S. (2007). Requirement of Rac1 distinguishes follicular from
interfollicular epithelial stem cells. Oncogene 26, 5078–5085.
Chen, L., Liao, G., Yang, L., Campbell, K., Nakafuku, M., Kuan, C.Y., and
Zheng, Y. (2006). Cdc42 deficiency causes Sonic hedgehog-independent
holoprosencephaly. Proc. Natl. Acad. Sci. USA 103, 16520–16525.
Chrostek, A., Wu, X., Quondamatteo, F., Hu, R., Sanecka, A., Niemann, C.,
Langbein, L., Haase, I., and Brakebusch, C. (2006). Rac1 is crucial for hair
follicle integrity but is not essential for maintenance of the epidermis. Mol.
Cell. Biol. 26, 6957–6970.
Ciccolini, F. (2001). Identification of two distinct types of multipotent neural
precursors that appear sequentially during CNS development. Mol. Cell. Neu-
rosci. 17, 895–907.
Czuchra, A., Wu, X., Meyer, H., van Hengel, J., Schroeder, T., Geffers, R.,
Rottner, K., and Brakebusch, C. (2005). Cdc42 is not essential for filopodium
formation, directed migration, cell polarization, and mitosis in fibroblastoid
cells. Mol. Biol. Cell 16, 4473–4484.
Danielian,P.S.,Muccino,D.,Rowitch,D.H.,Michael,S.K.,andMcMahon,A.P.
(1998).Modification ofgene activityinmouseembryosinuterobyatamoxifen-
inducible form of Cre recombinase. Curr. Biol. 8, 1323–1326.
Delfino-Machin, M., Chipperfield, T.R., Rodrigues, F.S., and Kelsh, R.N.
(2007). The proliferating field of neural crest stem cells. Dev. Dyn. 236,
3242–3254.
Etienne-Manneville, S., and Hall, A. (2003). Cdc42 regulates GSK-3beta and
adenomatous polyposis coli to control cell polarity. Nature 421, 753–756.
Fanger, G.R., Johnson, N.L., and Johnson, G.L. (1997). MEK kinases are
regulated by EGF and selectively interact with Rac/Cdc42. EMBO J. 16,
4961–4972.
Fukata, M., Nakagawa, M., and Kaibuchi, K. (2003). Roles of Rho-family
GTPases in cell polarisation and directional migration. Curr. Opin. Cell Biol.
15, 590–597.
Groszer,M.,Erickson,R.,Scripture-Adams,D.D.,Dougherty,J.D.,LeBelle,J.,
Zack, J.A., Geschwind, D.H., Liu, X., Kornblum, H.I., and Wu, H. (2006). PTEN
negatively regulates neural stem cell self-renewal by modulating G0–G1 cell
cycle entry. Proc. Natl. Acad. Sci. USA 103, 111–116.
Guemar, L., de Santa Barbara, P., Vignal, E., Maurel, B., Fort, P., and Faure, S.
(2007). The small GTPase RhoV is an essential regulator of neural crest induc-
tion in Xenopus. Dev. Biol. 310, 113–128.
Hari, L., Brault, V., Kleber, M., Lee, H.Y., Ille, F., Leimeroth, R., Paratore, C.,
Suter, U., Kemler, R., and Sommer, L. (2002). Lineage-specific requirements
of beta-catenin in neural crest development. J. Cell Biol. 159, 867–880.
Ittner, L.M., Wurdak, H., Schwerdtfeger, K., Kunz, T., Ille, F., Leveen, P.,
Hjalt, T.A., Suter, U., Karlsson, S., Hafezi, F., et al. (2005). Compound develop-
mental eye disorders following inactivation of TGFbeta signaling in neural-
crest stem cells. J. Biol. 4, 11.
Jaffe, A.B., and Hall, A. (2005). Rho GTPases: biochemistry and biology. Annu.
Rev. Cell Dev. Biol. 21, 247–269.
Kee,Y.,andBronner-Fraser,M.(2005).Toproliferateortodie:roleofId3incell
cycle progression and survival of neural crest progenitors. Genes Dev. 19,
744–755.
Kleber, M., Lee, H.Y., Wurdak, H., Buchstaller, J., Riccomagno, M.M.,
Ittner, L.M., Suter, U., Epstein, D.J., and Sommer, L. (2005). Neural crest
stem cell maintenance by combinatorial Wnt and BMP signaling. J. Cell Biol.
169, 309–320.
Kruger,G.M.,Mosher,J.T.,Tsai,Y.H.,Yeager,K.J.,Iwashita,T.,Gariepy,C.E.,
and Morrison, S.J. (2003). Temporally distinct requirements for endothelin
receptor B in the generation and migration of gut neural crest stem cells.
Neuron 40, 917–929.
Le Douarin, N.M., and Dupin, E. (2003). Multipotentiality of the neural crest.
Curr. Opin. Genet. Dev. 13, 529–536.
Lee, H.Y., Kleber, M., Hari, L., Brault, V., Suter, U., Taketo, M.M., Kemler, R.,
and Sommer, L. (2004). Instructive role of Wnt/beta-catenin in sensory fate
specification in neural crest stem cells. Science 303, 1020–1023.
Lillien, L., and Raphael, H. (2000). BMP and FGF regulate the development of
EGF-responsive neural progenitor cells. Development 127, 4993–5005.
Molofsky, A.V., Pardal, R., Iwashita, T., Park, I.K., Clarke, M.F., and Morrison,
S.J. (2003). Bmi-1 dependence distinguishes neural stem cell self-renewal
from progenitor proliferation. Nature 425, 962–967.
Narumiya, S., and Yasuda, S. (2006). Rho GTPases in animal cell mitosis. Curr.
Opin. Cell Biol. 18, 199–205.
Paratore, C., Goerich, D.E., Suter, U., Wegner, M., and Sommer, L. (2001).
Survival and glial fate acquisition of neural crest cells are regulated by an inter-
play between the transcription factor Sox10 and extrinsic combinatorial
signaling. Development 128, 3949–3961.
Pietri, T., Eder, O., Breau, M.A., Topilko, P., Blanche, M., Brakebusch, C.,
Fassler, R., Thiery, J.P., and Dufour, S. (2004). Conditional beta1-integrin
gene deletion in neural crest cells causes severe developmental alterations
of the peripheral nervous system. Development 131, 3871–3883.
Real, C., Glavieux-Pardanaud, C., Le Douarin, N.M., and Dupin, E. (2006).
Clonally cultured differentiated pigment cells can dedifferentiate and generate
multipotent progenitors with self-renewing potential. Dev. Biol. 300, 656–669.
Schwartz, M.A., and Shattil, S.J. (2000). Signaling networks linking integrins
and rho family GTPases. Trends Biochem. Sci. 25, 388–391.
Simpson, M.J., Zhang, D.C., Mariani, M., Landman, K.A., and Newgreen, D.F.
(2007). Cell proliferation drives neural crest cell invasion of the intestine. Dev.
Biol. 302, 553–568.
Cell Stem Cell
Small Rho GTPases in Neural Crest Development
246 Cell Stem Cell 4, 236–247, March 6, 2009 ª2009 Elsevier Inc.

Page 12

Sini, P., Cannas, A., Koleske, A.J., Di Fiore, P.P., and Scita, G. (2004). Abl-
dependent tyrosine phosphorylation of Sos-1 mediates growth-factor-
induced Rac activation. Nat. Cell Biol. 6, 268–274.
Soriano, P. (1999). Generalized lacZ expression with the ROSA26 Cre reporter
strain. Nat. Genet. 21, 70–71.
Stewart, A.L., Young, H.M., Popoff, M., and Anderson, R.B. (2007). Effects of
pharmacological inhibition of small GTPases on axon extension and migration
of enteric neural crest-derived cells. Dev. Biol. 307, 92–104.
Tallquist, M.D., and Soriano, P. (2003). Cell autonomous requirement for
PDGFRalpha in populations of cranial and cardiac neural crest cells. Develop-
ment 130, 507–518.
Vidali, L.,Chen,F.,Cicchetti, G.,Ohta,Y.,and Kwiatkowski,D.J.(2006).Rac1-
null mouse embryonic fibroblasts are motile and respond to platelet-derived
growth factor. Mol. Biol. Cell 17, 2377–2390.
Wang, L., and Zheng, Y. (2007). Cell type-specific functions of Rho GTPases
revealed by gene targeting in mice. Trends Cell Biol. 17, 58–64.
Wilson, A., Murphy, M.J., Oskarsson, T., Kaloulis, K., Bettess, M.D., Oser,
G.M., Pasche, A.C., Knabenhans, C., Macdonald, H.R., and Trumpp, A.
(2004). c-Myc controls the balance between hematopoietic stem cell self-
renewal and differentiation. Genes Dev. 18, 2747–2763.
Wong, C.E., Paratore, C., Dours-Zimmermann, M.T., Rochat, A., Pietri, T.,
Suter, U., Zimmermann, D.R., Dufour, S., Thiery, J.P., Meijer, D., et al.
(2006). Neural crest-derived cells with stem cell features can be traced back
to multiple lineages in the adult skin. J. Cell Biol. 175, 1005–1015.
Wu, X., Quondamatteo, F., Lefever, T., Czuchra, A., Meyer, H., Chrostek, A.,
Paus, R., Langbein, L., and Brakebusch, C. (2006). Cdc42 controls progenitor
cell differentiation and beta-catenin turnover in skin. Genes Dev. 20, 571–585.
Wurdak, H., Ittner, L.M., Lang, K.S., Leveen, P., Suter, U., Fischer, J.A.,
Karlsson, S., Born, W., and Sommer, L. (2005). Inactivation of TGFbeta
signaling in neural crest stem cells leads to multiple defects reminiscent of
DiGeorge syndrome. Genes Dev. 19, 530–535.
Wurdak, H., Ittner, L.M., and Sommer, L. (2006). DiGeorge syndrome and
pharyngeal apparatus development. Bioessays 28, 1078–1086.
Yang, L., Wang, L., Geiger, H., Cancelas, J.A., Mo, J., and Zheng, Y. (2007).
Rho GTPase Cdc42 coordinates hematopoietic stem cell quiescence
and niche interaction in the bone marrow. Proc. Natl. Acad. Sci. USA 104,
5091–5096.
Cell Stem Cell
Small Rho GTPases in Neural Crest Development
Cell Stem Cell 4, 236–247, March 6, 2009 ª2009 Elsevier Inc. 247