Schwann Cell Precursors from Nerve
Innervation Are a Cellular Origin
of Melanocytes in Skin
Igor Adameyko,1,8Francois Lallemend,1,8Jorge B. Aquino,1,2Jorge A. Pereira,3Piotr Topilko,4Thomas Mu ¨ller,5
Nicolas Fritz,1Anna Beljajeva,6Makoto Mochii,7Isabel Liste,1Dmitry Usoskin,1Ueli Suter,3Carmen Birchmeier,5
and Patrik Ernfors1,*
1Unit of Molecular Neurobiology, Department of Medical Biochemistry and Biophysics, Karolinska Institute, 17177 Stockholm, Sweden
2Liver Unit, School of Medicine, Austral University, Avenida Presidente Pero ´n 1500 (B1629ODT) Derqui-Pilar, Buenos Aires, Argentina
3Institute of Cell Biology, Department of Biology, ETH Zu ¨rich, CH-8093 Zu ¨rich, Switzerland
4Laboratoire de Ge ´ne ´tique Mole ´culaire du De ´veloppement, Inserm 784, Ecole Normale Supe ´rieure, 45 rue d’Ulm, 75230 Paris cedex 05,
5Neuroscience Department, Max-Delbru ¨ck-Centrum for Molecular Medicine, Robert-Ro ¨ssle-Strasse 10, 13125 Berlin, Germany
6Department of Cell and Molecular Biology, Medical Nobel Institute, Karolinska Institute, Stockholm, Sweden
7Department of Life Science, Graduate School of Life Science, University of Hyogo, 3-2-1 Kouto, Kamigori, Akou, Hyogo 678-1297, Japan
8These authors contributed equally to this work
Current opinion holds that pigment cells, melano-
cytes, are derived from neural crest cells produced
at the dorsal neural tube and that migrate under
the epidermis to populate all parts of the skin. Here,
we identify growing nerves projecting throughout
the body as a stem/progenitor niche containing
Schwann cell precursors (SCPs) from which large
numbers of skin melanocytes originate. SCPs arise
as a result of lack of neuronal specification by
Hmx1 homeobox gene function in the neural crest
ventral migratory pathway. Schwann cell and mela-
nocyte development share signaling molecules
with both the glial and melanocyte cell fates inti-
mately linked to nerve contact and regulated in an
opposing manner by Neuregulin and soluble signals
including insulin-like growth factor and platelet-
derived growth factor. These results reveal SCPs as
a cellular origin of melanocytes, and have broad
implications on the molecular mechanisms regu-
lating skin pigmentation during development, in
health and pigmentation disorders.
Melanocytes, or pigment cells, represent a significant proportion
of the cells in the adult epidermis, with about 800 cells/mm3.
Melanin, which is produced by melanocytes, is the main contrib-
utor to pigmentation and is packaged and delivered to keratino-
cytes by lysosome-like structures called melanosomes. The
skin, as the main barrier to the external environment, relies on
melanocytes to provide photoprotection and thermoregulation
via the production of melanin.
The current view of how melanocytes populate the skin is that
their precursors, the melanoblasts, arise from the neural crest
(NC), which is induced by Bone morphogenic protein (BMP)
the ectoderm almost immediately, migrating in its vicinity
following well defined routes (Dorris, 1938, 1939; DuShane,
1935; Rawles, 1947; Ris, 1941; Twitty, 1936). First, they migrate
dorsolaterally between the dermamyotome and the overlaying
ectoderm and then ventrally through the developing dermis to
their eventual destination in the basal layer of the epidermis and
the hair follicles (Erickson, 1993). This hypothesis accounts for
that neural crest cells (NCCs) commit to a melanocyte fate
already at the level of the neural tube, shortly after NC delamina-
tion. The origin and stereotypic migratory pathway of skin
melanocytes have become widely accepted in the literature
or quail-into-chick transplantation systems, vital dye tracing in
chick and mouse, and genetic targeting to express reporter
markers under melanoblast- and melanocyte-specific protein
promoters in mouse (Bronner and Cohen, 1979; Johnston, 1966;
zijaetal., 1989,1990; Weston,1963). Thesestudiesdescribe the
emergence of melanoblasts between the dermamyotome and
ectoderm. However, the eventual migration within the prospec-
tive dermis to ventral body regions and limbs has not been
fully addressed and it has been observed that melanoblasts
appear precociously and localize at least to the limbs without
an apparent dermal migration (Fox, 1949). The clonal expansion
degree during migration under the epidermis, is thought to give
rise to coat pigmentation (Mintz, 1967) and to underlie the obser-
melanoblasts (Huszar et al., 1991; Wilkie et al., 2002).
366 Cell 139, 366–379, October 16, 2009 ª2009 Elsevier Inc.
and differentiate into neurons and Schwann cell precursors
(SCPs). SCPs are defined as Sox10+NCC-derived cells tightly
associated with neuronal projections during early stages of
embryonic development that are able to migrate long distances
along the nerves (Jessen and Mirsky, 2005). NCCs that acquire
a melanoblast fate emerge lastly at the dorsal neural tube and
migrate following a dorsolateral pathway. The difference in
timing between these waves correlates with the period of
conversion of the epithelial somite into the dermamyotome,
and these spatial and temporal changes in the presence of
instructive signals are believed to instruct the late NCCs to
acquire a melanocyte fate. In chick and mouse, Wnt1 promotes
melanoblast formation and reduces formation of neurons and
glia via b-catenin (Dunn et al., 2000; Jin et al., 2001) in part by
inducing the shared multipotent NCC/melanocyte SRY-related
high-mobility-group domain transcription factor Sox10 (Aoki
et al., 2003; Honore et al., 2003). As a central regulator of
pigment cell development, microphthalmia-associated tran-
scription factor (Mitf) determines a melanocyte fate of multipo-
tent NCCsin partbyits
lineage-specific regulation of the three major pigment enzymes:
tyrosinase, Tyrp1, and Dct as well as other pigmentation factors
(Steingrimsson et al., 2004). In Zebrafish, the Wnt pathway also
promotes a commitment to the melanocyte lineage by a direct
transcriptional regulation of the Mitf gene (Dorsky et al., 2000).
Consistent with a source of Wnt signals from the neural tube
shortly after NC delamination, and hence the current view of
the origin and migration of melanoblasts, the analysis of Mitf ex-
pressing cells has confirmed a commitment of NCCs at the
dorsal neural tube and the presence of Mitf positive cells under-
neath the ectoderm at the level of the dermamyotome during
development (Nakagawa and Takeichi, 1998).
In this study, we confirm the previously described source of
skin melanocytes from the delaminating NCCs at the dorsal
nocytes are also produced from nerves innervating the skin. The
ventral migratory pathway. We show that myelinating Krox20+
Schwann cells normally do not differentiate into melanocytes
but retain the competence to do so. The SCP versus melanocyte
and depends on interactions with Neuregulins and their ErbB3
receptor, revealing a dual role for this signaling system in the
developing nerve. In addition, insulin-like growth factor 1 (IGF1)
and platelet-derived growth factor (PDGF) act in an opposing
manner to Neuregulins promoting, in the later waves melanocyte
differentiation, survival, and expansion from SCPs.
potent transcriptional and
Melanocyte Development Is Associated with SCPs
Melanoblasts are believed to follow a single highly stereotyped
dorsolateral pathway from their origin in the dorsal neural tube
to their final target, the skin. Consistent with this, Mitf+/Sox10+
melanoblasts were seen at the dorsal neural tube in the
Hamburger Hamilton stage (HH) 22 in chick. Labeled cells
were located in the trunk along the lateral edges of the embryo
under the epidermis at the extent of the dermamyotome but
were never located in the ventral body wall or the limbs
(Figure 1A). Between HH24 and 27, these cells were reduced
in number with the remaining few cells being located at dorsal
aspects of the embryo (Figures 1H and 1L, arrowheads).
AtlateHH22,a secondpresence ofMitf+/Sox10+melanoblasts
spatiotemporally distinct from the previously described dorsolat-
eral pathway was observed within the distal ventral ramus of the
spinal nerve (Figures 1B, 1D, and 1E, indicated 2a). Weak levels
alizing with Sox10+SCP nuclei while MITF expression was
stronger in cells associated but not in direct contact with nerves,
suggesting that melanocytes arise from Sox10+SCPs of the
nerves (Figures 1C–1F). At HH24, an increased number of Mitf+/
Sox10+melanoblasts associated with ventral ramus nerves was
observed (Figures 1H, 1J, and 1K). At this stage, when the dorsal
ramus was not clearly established, the presence of a few Mitf+/
and skin that were not in contact with nerves indicated that some
might migrate unguided bynerves through the dermamyotome to
all, this contribution appeared minor as at HH27-29, larger
numbers of Mitf+/Sox10+melanoblasts were observed along the
muscles of the dorsal and lateral trunk(Figures 1M, indicated 2b).
At this stage, melanoblasts appeared in even greater number
around the ventral spinal nerves as compared to HH24 (Figures
Mitf+cells at HH21-22, ventral ramus associated Mitf+cells
increased with time (i.e., HH27-28; Figure 1G).
Chick embryos electroporated in the neural tube with a green
fluorescent protein (GFP) expression construct for NCC tracing
were used to address if melanoblasts produced from spinal
nerves were NC-derived and whether they originated from
SCPs associated with nerve fibers. At HH22, GFP+cells within
the nerves were occasionally Sox10+/Mitf+while GFP+cells
that were very close (i.e., associated with), but not in direct
contact with the nerve, were always Sox10+/Mitf+(Figure 1I
and inset). At HH24 and 27 most Sox10+/Mitf+/GFP+cells were
associated with nerves extending toward the skin (Figures 1H
and 1J–1M) with ‘‘hot spots’’ at distal ends of the dorsal and
ventral ramus and brachial plexus, together with smaller nerve
branches of limbs (Figures 1I–1M and S1A–S1D, summarized
in 1N). Similar results were obtained in the mouse. At E10.5 the
first wave (i.e., dorsolateral pathway) of Mitf+/Sox10+melano-
blasts was evident (Figure 1O). In the chick, a marked reduction
in cells of the first wave was observed asdescribed above, but in
mouse at E11.5, the majority of melanoblasts in the dorsolateral
aspect of the embryo was absent. At E12, a few Mitf+/Sox10+
melanoblasts were closely associated with cutaneously located
spinal nerves of the dorsal rami (Figures 1Q and 1S) with
increasing numbers at E13 (Figure 1R, summarized in S2A).
Quantification confirmed a reduction of melanoblasts from the
first wave and a later increase now associated with nerves
(Figure 1P). Nerve-associated Mitf+cells were confirmed to be
Cell 139, 366–379, October 16, 2009 ª2009 Elsevier Inc. 367
368 Cell 139, 366–379, October 16, 2009 ª2009 Elsevier Inc.
melanoblasts by their coexpression in chick with melanosome
matrix protein (MEBL-1), which specifically recognizes avian
melanocytes and their precursors, and in mouse by the coex-
pression with the tyrosinase family member dopachrome tauto-
merase (DCT), a key enzyme involved in the synthesis of the
melanin pigment (Figure S3) (Jessen and Mirsky, 2005). Com-
bined, these results show that during development emergence
of melanoblasts is closely associated with SCPs of nerves in
temporospatially distinct locations from that previously de-
scribed in the dorsolateral pathway.
Defined and Independent Contribution of Nerve-Derived
We next directly addressed if melanoblasts associated with
nerves are derived from cells within the nerves (i.e., SCPs).
One hemisphere of the chick neural tube together with the adja-
cent DRGs was ablated at the forelimb bud level just after the
first melanocyte wave ceased (HH22; Serbedzija et al., 1989)
(Figures 2A–2D). Mitf+cells in the dorsolateral pathway were
unaffected by the ablation immediately and 6 hr after surgery
(Figures 2N, S4A, and S4B). The embryos were previously elec-
troporated with GFP for cell lineage tracing and were analyzed at
HH28 (Figures 2C–2H). On the experimental side, a reduced
number of Mitf+melanoblasts was observed in the dorsal and
lateral body wall (Figures 2E and F) and an even greater loss in
the limb (Figures 2H, S4C, quantification 2O). This shows that
melanocytes were reduced in numbers as a consequence of
eliminating the ventral migratory pathway.
To exclude that later nerve-associated melanoblasts appear-
ing in the limb were derived from migrating melanoblasts of the
dorsolateral pathway, we next surgically unilaterally ablated
the dorsal and lateral surface of the embryo at HH22, including
the delaminated melanoblasts of the dorsolateral pathway, but
leaving the neural tube, DRG and ventrally migrating SCPs intact
(Figures 2I–2L and S5A). Ablation of the first (dorsolateral) wave
of melanoblasts did not quantitatively affect numbers of melano-
tracing in such ablated animals provided evidence for NC-der-
ived SCPs in the ventral pathway of NC migration as an origin
of melanocytes in the chick (Figures 2M and S5).
Our data suggest that not only melanocytes in the limbs
originate from SCPs migrating along nerves but also a large
number of melanocytes in the dorsal and lateral body wall.
These may emerge partially from Sox10+DRG border cells
but most abundantly from SCPs migrating along the dorsal
ramus of the spinal nerve. To confirm these migratory paths
we performed slice cultures of HH24 and 27 chick embryos
previously electroporated with GFP to trace NC-derived cells.
NCCs occasionally cross the hemisphere and contribute with
cells to the contralateral side and if electroporated at HH13
they often contribute with border cells around the DRG
(Figure S6A). Slice cultures were first prepared from HH24
when the DRG has just coalesced and the dorsal ramus nerve
was not formed. Already after 6 hr in vitro, border cells divided
and started to migrate laterally, and within 18 hr, many of the
cells had migrated through the dermamyotome apparently
independent of nerves (Figures S6A–S6D). When cultures were
instead prepared from HH27 embryos the dorsal ramus was
clearly distinguishable and at this developmental stage the
SCPs migrated along the dorsal ramus to the dermis/epidermis
(Figures S6E–S6H). This identifies that large numbers of SCPs
migrate along nerves toward the skin where some acquire
a melanoblast fate.
PLP+SCPs: A Cellular Origin of Melanocytes
in the Mouse
cellular and molecular identity of the precursors of melanoblasts
in the mouse. Transgenic mice expressing the improved tamox-
ifen (TM)-inducible Cre recombinase (CreERT2) under the pro-
teolipid protein (PLP) promoter that is active specifically in
SCPs and Schwann cells (Leone et al., 2003) were crossed to
the Rosa26-YFP reporter mouse strain that contains a floxed
stop cassette preventing YFP expression. In such mice, TM
treatment results in the excision of the stop cassette leading to
the permanent expression of YFP in SCPs and, hence, also in
all cells derived from SCPs. Injection of TM at E16 (when forma-
tion of new melanocytes has ceased) and analysis of YFP
expression at E17 revealed labeling of Schwann cells within
cutaneous nerves but not in Mitf+melanocytes, confirming that
the PLP-CreERT2 transgene is not active in melanocytes
(0/218 Mitf+cells, n = 4 animals; Figure 3A). When administrating
TM at E11 and analyzing embryos at E13, similar to the experi-
mental conditions (see below), YFP expression in Mitf+melano-
blasts of the dorsolateral migratory pathway at sacral levels of
the embryo was not observed (0/48 Mitf+cells, n = 4 animals,
Figure 1. Association of Melanoblasts with Nerves in Chick and Mouse Development
(A–N) Chick development, transverse sections through the trunk. (A) Mitf+melanoblasts in location of dorsolateral pathway (first wave, indicated by digit #1) at
HH22. (B) Spatially distinct Mitf+cells located at ventral spinal nerves a few hours later (indicated by digit #2a). (C and D) Mitf+melanoblasts within ventral spinal
nerve at late HH22. (E and F) Weak Mitf expression colocalizes with Sox10+cells within nerves (dotted arrows) with increased intensity in cells distal to the nerve
(solid arrows). (G) Quantification of Mitf+cells in indicated spatiotemporal locations (wave 1, dorsolateral pathway; 2a ventral spinal nerve; 2b dorsal spinal nerve;
n= 4animals/stage; error bars represent SEM). (H–K)GFP tracing of NCCsatHH22and HH24 (Dotted arrows point atMitf+/GFP+cells traced inlocation ofspinal
nerve and skin). (L) GFP tracing at HH27. Note increased number of Mitf+cells in proximity to ventral spinal nerve. White bracket indicates a spatial gap between
wave 1 and 2a melanoblasts. (M) GFP traced cells associated with the dorsal spinal rami (indicated by digit #2b). (N) Schematic spatiotemporal representation of
Mitf+melanocyte in the chick. (A–M) Solid arrows identify Mitf+cells in location 2a and arrowheads in 1 and 2b.
(O–S) Mouse development, transverse sections through the trunk. (O) Mitf+melanoblasts in the dorsolateral pathway at E10.5. (P) Quantification of Mitf+cells
associated with peripheral nerves (n = 4 animals/stage; error bars represent SEM). (Q and R) Labeling for Mitf+at E12-13. Note absence of cells in location of
dorsolateral (first) wave and appearance of increasing numbers of labeled cells in proximity to dorsal rami nerves (solid arrows: 2b wave). (S) Mitf and Sox10
double stained cells within dorsal rami nerves, similar to chick.
Abbreviations are as follows: NT, neural tube; DRG, dorsal root ganglion; dm, dermamyotome; dr, dorsal ramus; snv, ventral branch of spinal nerve. The scale
bars represent 50 mm.
Cell 139, 366–379, October 16, 2009 ª2009 Elsevier Inc. 369
Figure 2. A Contribution of Melanoblasts from the Ventral Migratory Pathway
(A–H)Ablationof DRGand nerves, transverse sectionsthroughthe trunk.(A) Schematic representation ofthe surgeryat HH22 eliminatingDRGand nerve-depen-
dent pathways (green represents electroporated parts of the embryo, red circles represent melanoblasts). (B) Section of forelimb at stage when surgery was con-
ducted.NotethepresenceofMitf+melanocytesinthedorsolateralpathway.(C)Whole-embryo view2daysaftersurgery(arrowindicating positionofsurgery).(D)
Ablated side of the embryo showing near complete loss of GFP+nervous tissue. (E–H) Mitf+cells at trunk (E and F) and the dorsal limb level (G and H) of an
operated embryo ([E and G] unoperated side; [F and H] operated side). Note loss of melanoblasts in (H), compared to (G), due to the lack of the ventral migratory
pathway (dotted line marks the prospective position of the ventral spinal nerve).
(I–M) Ablation of skin and dermamyotome, transversal sections through the trunk. (I) Schematic representation of the surgery eliminating the dorsolateral mela-
noblast pathway at HH22. Whole (J), part (K), and section (L) view of operated embryo at time of analysis. (M) Mitf labeling of limb section on operated side. Note
unaffected numbers of Mitf+cells. Orange arrows point at traced GFP+Smelanoblasts.
370 Cell 139, 366–379, October 16, 2009 ª2009 Elsevier Inc.
Figure S7A). PLP and Cre in situ hybridization confirmed their
expression in ventral pathway including SCPs of nerves but not
in the dorsolataral pathway or DCT+melanocytes at E10.5 (Fig-
ure S7). Furthermore, TM injection at E9.5, at the peak of NC
delamination, and triple immunohistochemical staining for YFP,
Cre and DCT at E10.5 confirmed recombination and Cre expres-
sion in cells of the ventral migratory pathway. However, a
complete absence of Cre was observed in DCT+melanoblasts
of the dorsolateral migratory pathway in the trunk (0/136 DCT+
melanoblasts, n = 6 embryos, Figures 3B and 3C). In cranial
nerves which are more advanced in development than the trunk
and therefore could be analyzed in the E10.5 control embryos,
Cre was specifically expressed in SCPs while genetically traced
rapidly lost in cells detaching from the nerves (Figure 3D).
Combined, these data confirm that PLP-Cre is not active in the
NC dorsolateral pathway, melanoblasts or melanocytes. Exami-
nation of the dorsal ramus of the spinal nerve revealed a large
number of Sox10+/Mitf+cells expressing YFP often associated
with cutaneous Tuj1+(bIII-tubulin+) nerves (Figures 3E and 3F).
was63± 3%while thepercentage of Sox10+/Mitf+cellsexpress-
Mitf+cells were derived from SCPs of the nerve. We next ad-
dressed if these melanoblastscontribute to melanocytes at post-
collected at postnatal day (P) 11. Schwann cells of cutaneous
nerves were YFP+(Figures S8C–S8E), consistent with their origin
from SCPs. In addition, all hair follicles analyzed contained pig-
mented YFP+cells in skin of the trunk and extremities (200/200
follicles, n = 4 animals). Pigmented cells in hair follicles that
were genetically traced (YFP+) were double stained for DCT con-
analyzing dermal melanocytes in both limbs and the trunk and
quantification showed that 65.9 ± 3.1% and 58.6 ± 4.1% of the
contained YFP (Figures 3H–3P). Pigmented and DCT+dermal
melanocytes of glabrous skin (foot pad) were also YFP+(Figures
S8A and S8B). These results confirm SCPs as an important
cellular origin of pigmented dermal and hair follicle melanocytes
Hmx1 Regulates Neuronal versus Schwann Cell
Precursor and Melanocyte Fates
Our previous results suggest that DRG border cells and SCPs
along the nerves can be a cellular source of melanocyte precur-
sors. Expression of a number of transcription factors (TFs) iden-
tified in the DRG (Marmigere et al., 2006; Gray et al., 2004) was
The homeobox TF Hmx1 was absent in migrating NCCs at
HH21 (Figure 4A) and increased as the NCCs coalesced into
a DRG at HH24 (Figures 4B–4F). Hmx1 was exclusively localized
to Islet-1+and mutually exclusive with Sox10+cells (marking
neuroblasts and uncommitted NCC/glia progenitors, respec-
tively; Figures 4C–4F) (Montelius et al., 2007), indicating that
the onset of Hmx1 expression in the DRG coincides with the
commitment to a neuronal fate. To assess the role of Hmx1
during DRG development, we developed two independent short
interfering RNAs (siRNAs) against Hmx1. The efficiency of these
siRNA (scrambled) led to a contribution of electroporated cells
that was similar to embryos only receiving GFP vector (Fig-
ure 4G). GFP+cells were seen at HH24 in nerves and the DRG,
often colocalizing with Islet-1 (Figure 4G). Embryos receiving
Hmx1 siRNAs displayed very few electroporated GFP+cells
acquiring a neuronal fate in the DRG as seen by an almost
complete complementary pattern of GFP+cells and neuronal
markers Islet-1 and Tuj1. The electroporated cells were instead
concentrated around the lateral border of the ganglion (i.e.,
DRG border cells, Figures 4H and 4I and S9G–S9H), some of
them Mitf+melanoblasts and Sox10+/Mitf+/GFP+cells were
observed between the DRG and the skin. At HH29 the skin
showed a marked increase of Sox10+/Mitf+/GFP+melanoblasts
compared to the control (scrambled siRNA) condition (Figures
4J–4M), which correlated with an increase in SCPs along dorsal
rami spinal nerves in siHmx1 treated embryos (Figures 4J, 4K,
S9I, and S9J). These findings show that a shift in the glia-
neuronal balance can affect the amount of melanocyte precur-
sors arising from late nerve-dependent pathways and that
Hmx1 may act as a critical transcriptional switch between
neuronal versus glia-melanocyte fates in the ventral NCC
Nerve-Schwann Cell Interactions during Melanoblast
In order to examine the role of nerve contact for differentiation of
SCP-derived melanocytes, the spinal nerves proximal to the
brachial plexus were transected on the GFP electroporated left
side of HH27 chick embryos, and analyzed for Mitf expression
at HH28 and 29. Compared to the unoperated contralateral fore-
limb, the operated limb exhibited a significant increase of Mitf+
cells located in and around the region of the degenerating
peripheral nerve at 12 and 24 hr after axotomization (Figures
5A–5E and S10). Often Mitf+cells were aligned with GFP traces
of decaying electroporated nerve (Figure 5F). The increase of
Mitf+cells was not caused by expansion of any already existing
Mitf+cells since BrdU incorporation in such cells was not
increased compared to control animals (Figure S11). These
results demonstrate that maintenance of SCPs is nerve-depen-
dent, as it has previously been shown (Jessen and Mirsky,
(A–M) Solid arrows mark nerve associated melanoblasts, arrowheads mark melanoblasts of dorsolateral pathway, dotted arrows point GFP+SCPs within nerve.
Abbreviations are as follows: NT, neural tube; DRG, dorsal root ganglion. The scale bars represent 50 mm.
(N) Quantification of Mitf+cells at HH22 following ablation of DRG and nerves. Note that ablation does not affect numbers of Mitf+cells in dorsolateral pathway
(n = 4; error bars represent SEM).
(O) Quantification of Mitf+cells in limbs from intact and operated sides of both ablation studies (n = 4 animals; error bars represent SEM; Students t test ***p <
Cell 139, 366–379, October 16, 2009 ª2009 Elsevier Inc. 371
Figure 3. Genetic Tracing with PLP-CreERT2 Mice Confirms SCPs-Derived Origin of Embryonic and Adult Melanocytes in Mice
(A–D) Transversal section of genetic tracing controls. (A) Mitf labeling of E17 embryo 24 hr after tamoxifen administration. Note that Mitf+skin melanocytes
do not express YFP while Schwann cells of cutanous nerves are positive. (B–D) E10.5 embryo injected with tamoxifen at E9.5 and stained for DCT or Mitf
and Cre. Note expression of Cre and YFP (recombination) in the ventral pathway (B and C) and in SCPs (D), while Mitf+and DCT+melanoblasts are negative
(E–G)Genetic tracingofembryonicmelanocytesbytamoxifenadministration atE11and analysisatE13.(Eand F)StainingforYFP,MitfandSox10orTuj1reveals
recombination in SCPs and traced melanoblasts associated with Tuj1+nerves and dermis (Mitf+cells with punctuated arrows, and Mitf+/YFP+solid arrows).
(G) Quantification of YFP+cells among Sox10+SCPs and MITF+melanoblasts. Note that these numbers represent recombination frequency in nerve and the
proportion of nerve-derived melanoblasts (n = 4 animals).
(H–P) Genetic tracing of postnatal melanocytes by tamoxifen administration at E11 and analysis at P11. Longitudinal (H and I) and transversal (M–P) sections of
hair follicles traced with YFP and stained for DCT, as indicated. (J and K) High magnification of SCP traced (YFP+) pigmented, DCT+pigmented hair follicle (J) and
372 Cell 139, 366–379, October 16, 2009 ª2009 Elsevier Inc.
2005), and that in the absence of signals provided by the nerve
some SCPs instead acquire a melanocyte fate. This opens for
that a competition of rapidly expanding numbers of SCPs for
nerve contact may be a participating mechanism during the
recruitment of melanoblasts populating the skin from nerves
innervating the embryonic body.
Competence of Myelinating Schwann Cells to Generate
Pigmented Cells in the Adult
The transcriptional factor Krox20/Egr2, that starts expression at
E16 in mouse nerves, is critical for the differentiation of immature
Schwann cells into myelinating Schwann cells (Topilko et al.,
dermal (K) melanocytes. (L) Quantification of percent of melanocytes containing YFP in hair follicles and dermis at P11 (n = 4 animals). Note that the number is
similar to the recombination frequency in SCPs.
Abbreviations are as follows: NT, neural tube ; NCC, neural crest cells ; dr, dorsal ramus. The scale bars represent 50 mm and 20 mm (J and K).
Figure 4. Hmx1 Expression Defines the
Balance between Neurogenic and Glia-Mel-
anocytic Fates in the Developing Dorsal
(A and B) Transverse sections of HH21 (A) and
HH24 (B) embryos hybridized with an Hmx1 anti-
sense riboprobe. Note the detection of Hmx1
at HH24) but not during NCC migration (HH21).
(C–F) HH24 forelimb DRG section triple stained for
Hmx1 mRNA (Hmx1 ISH), Sox10, and Isl1. Note
that Hmx1 colocalized with the neuronal marker.
(G–I) Hmx1 siRNA knockdown experiments at
of siRNA receiving (GFP+) cells at border of the
DRG, some expressing Mitf ([H and I] arrowheads
mark dorsolateral melanoblasts).
(J–L) Hmx1 siRNA knockdown experiments at
HH29. Note marked increase of siRNA receiving
(GFP+) cells in skin and which double stain for
Mitf. (K and L) Digit 2b denotes dorsal rami associ-
(M) Quantification of percent of GFP+cells located
in the skin at HH29 (n = 6 animals, p < 0.001; error
bars represent SEM; Students t test ***p < 0.001).
Abbreviations are as follows: NT, neural tube;
DRG, dorsal root ganglion; dm, dermamyotome;
dr, dorsal ramus. The scale bars represent 50 mm.
1994) and acts together with Sox10 to
(Steingrimsson et al., 2004). We ad-
dressed if nerve-derived melanocytes
arise from SCPs prior to their specifica-
tion to promyelinating and myelinating
a knock-in allele for the Cre recombinase
in the Krox20 locus (Krox20Cre/+) (Voicu-
lescu et al., 2000) were combined with
genetic tracing. YFP labeled cells were
found in nerves of the adult mouse in
close association with neurofilament+
axons (Figure 6A), consistent with its
expression in myelinating Schwann cells, as well as in hair folli-
cles and epidermis (Figure 6B), as previously described (Gam-
bardella et al., 2000). YFP was not observed in melanocytes of
hair follicles and in melanocytes of the skin of the adult mouse
(Figures 6B and 6C), showing that nerve-derived melanocytes
do not differentiate during normal development from promyeli-
nating and myelinating Schwann cells.
Our previous data showed a strong correlation between
nerves and Schwann cell phenotype indicating that those cells
staying in contact with nerves retain a SCP state and eventu-
ally differentiate into Schwann cells while cells detaching from
the nerve acquire MITF expression. This raised the issue if
Cell 139, 366–379, October 16, 2009 ª2009 Elsevier Inc. 373
myelinating Schwann cells could have the potential to differen-
tiate into melanocytes if nerve contact is perturbed. We
therefore addressed whether Krox20+
myelinating Schwann cells retain the competence to differen-
tiate into melanocytes. We examined this by surgically cutting
a 5 mm piece of the sciatic nerve, leaving the transected piece
of nerve in position by sutures to connective tissue of the
underlying muscle (Rizvi, et al., 2002). Eighty days later, clus-
ters of highly pigmented cells around muscles extended
for more than 1 cm around the nerve fragment in all animals
(n = 8, Figures 6D–6G). This massive hyperpigmentation was
also observed on the ventral portion of the dermis (Figures
6H and 6I) with pigmented cells occasionally located also
within muscles (Figures 6J and 6K). A large number of pig-
mented melanocytes were YFP+around and within muscles
(Figure 6K), in the dermis (Figure 6L) and within and around
remnants of the denervated sciatic nerve fragment (Figures
6M–6O). These results suggest that during development the
onset of Krox20 expression defines a restriction in fate that
normally precludes their differentiation into pigmented cells,
but that these cells remain competent to form pigmented cells
Figure 5. Excessive Melanocyte Numbers upon Loss
of SCP-Nerve Contact
(A–D and F) Microsurgery-induced nerve axotomization at
HH27 analyzed in transverse sections of HH28 (A and B) and
HH29 (C, D, and F) embryos at the forelimb level, on the unop-
erated (A and C) and operated (B, D, and F) sides. Note the
appearance of Mitf+melanoblasts at the place of the degener-
ating nerve 12 (B) and 24 hr (D and F) after surgery. Remnants
of nerves seen by GFP fluorescence (arrowheads mark the
(E) Quantification of melanoblasts inside of the limb on
sections after microsurgery-induced nerve axotomization
(n = 4 animals; error bars represent SEM; Students t test
**p < 0.01).
(F) Axotomized spinal nerve at high magnification of a GFP-
traced chick embryo. Note Mitf+cells along the decaying
ventral spinal nerve (residual GFP).
if challenged with a new microenvironment result-
ing from a loss of nerve contact.
ErbB3 Signaling between Nerve and
Schwann Cell Precursors Balance Glial
versus Melanocyte Fates
NCCs which migrate along nerves and eventually
adopt a Schwann cell fate rely on axonal signals
for survival, proliferation and differentiation. A neu-
ronally-derived signal regulating survival and prolif-
Brockes, 1984), also named glial growth factor,
Heregulin or Neu differentiation factor, and that
signals through the receptor tyrosine kinase
ErbB2 and ErbB3 heterodimer complex expressed
by Schwann cells. NRG1 appeared as a possible
signaling molecule regulating glial versus melano-
cyte fates because it is produced in sensory and
motor neurons as a molecule inserted into the axonal
membranes. We examined a possible role for NRG1 signaling
in the formation of melanocytes during development in erbB3?/?
mice (Riethmacher et al., 1997). The dorsolateral wave of
Mitf+melanoblasts appearing close to the neural tube was unaf-
fected in erbB3?/?mice at E10 (Figures 7A, 7B, and 7E) while
increased number of melanoblasts was observed around distal
ends of dorsal spinal nerves at E12 (Figures 7C and 7D) and
quantification confirmed an increase of Mitf+cells to 178% that
of control mice (Figure 7E), despite an overall reduction of
SCPs along nerves (Figure S12). Combined, our data show
that Neuregulin signaling regulates not only survival and prolifer-
ation of SCPs along nerves during development, but it also
participates in the decision between Schwann cell versus mela-
Soluble Signals Regulating Melanocyte Development
Several soluble signals have been proposed to regulate the
proliferation and migration of melanocytes (Thomas and Erick-
son, 2008). Development of melanocytes in different time and
location from that previously thought open for soluble signals
374 Cell 139, 366–379, October 16, 2009 ª2009 Elsevier Inc.
partly different from those previously described. Two different
in vitro assays were developed to address the role of previously
unidentified soluble signals. In the first assay progenitor cells of
HH29 DRG can stay in contact with axons and survive indepen-
to identify instructive and proliferative signals (Figures S13A–
S13G). Low numbers of Mitf+melanocytes were seen in control
cultures and NRG1 led to their reduction, consistent with our
previous in vivo data. Insulin-like growth factor 1 (IGF1) and
platelet-derived growth factor (PDGF) which are produced in
the developing nerves (Meier et al., 1999) in addition to the previ-
ously characterized hepatocyte growth factor (HGF), signifi-
cantly increased Mitf+cells. In a combination of NRG1 and
IGF1, the latter could efficiently compete with NRG1, which led
to a presence of significant amounts of melanoblasts (Figures
S13F and S13G).
In the second assay, pieces of the limb nerves containing
SCPs were dissected out from HH29 chick embryos and
cultured for 5 days. Under neutral medium conditions a near
complete death of all SCPs allowed us to assay for survival
effects (Figures S13H–S13L). NRG1 led to marked cell survival
but only a few Mitf+melanocytes (Figures S13J and S13L).
IGF1 led to significant cell survival and development of numer-
ous Mitf+melanocytes and was found to partially compete with
NRG1 signaling (Figures S13I, S13K, and S13L), while PDGF
supported cell survival inefficiently and cultures displayed only
suggest that the Schwann cell produced factors IGF1 and PDGF
may act in an opposing manner to NRG1 during recruitment of
melanoblasts from nerves in the developing embryo.
It has been widely accepted that during development NCCs de-
laminating from the neural tube acquire a melanocyte fate,
migrate dorsolaterally to populate the epidermis and massively
expand in numbers making up as much as 5%–10% of all cells
in the epidermis of the adult. We now provide evidence suggest-
ing that NCCs migrating in the ventral pathway, which previously
were considered to contribute only to glia and neurons of the
peripheral nervous system, are also a source of melanocytes.
Nerve-Derived Schwann Cell Precursors as a Cellular
Source of Skin Melanocytes
Several independent experiments confirm SCPs along nerves as
a cellular source of melanocytes. Immunohistochemical analysis
identified melanoblasts associated with nerves temporospatially
distinct from the dorsolateral pathway. Consistently, cell tracing
Mitf and MEBL expression. Ablation experiments confirmed that
Figure 6. Mature Schwann Cells in the Adult Retain the Potential to
Differentiate into Melanocytes
(A–C) Tracing of cell progenies using a Krox20-Cre locus crossed to a YFP
reporter strain. (A) YFP in myelinating Schwann cells on a transverse section
of the sciatic nerve. (B) YFP in the hair shafts and epidermis. (C) Pigmented
cells in the skin were not positive for YFP. (B–C) Dotted arrows mark GFP+
(D–I) Sciatic nerve surgery experiment. Pigmented melanocytes was found at
80 days after sciatic nerve surgery inside of the thigh (D), connective tissue (F),
and skin (H). (E, G, and I) Pigmented cells were not found in the contralateral
(J–O) YFP tracing of cell progenies using a Krox20-Cre activating strain
following sciatic nerve surgery. Note YFP+melanocytes between muscle
fibers(Jand K),connective tissueanddermis(L),andinthedegeneratednerve
stump of sciatic nerve (M–O). Dotted circle outlines the degenerated distal
nerve stump. Solid arrows mark YFP+melanocytes, arrowheads point YFP-
The scale bars represent 50 mm.
Cell 139, 366–379, October 16, 2009 ª2009 Elsevier Inc. 375
thedorsolateralpathwaydidnotaffect GFPcell tracingofventral
pathway-derived melanocyte numbers while removal of neural
tube and DRG resulted in greatly diminished numbers identifying
SCPs as a cellular origin of melanoblasts in the chick. Some
residual nerves were left in the limbs of ablated embryos which
could be the source of the few remaining melanoblasts found
in this location. SCPs as an origin of melanocytes were inferred
from axotomization experiments both during development and
consistent with our analysis of erbB3?/?mice also showing
increased melanoblast numbers. Further corroboration is data
showing that the homeobox transcription factor Hmx1, ex-
is required for neurogenesis and in its absence, increased
numbers of non-neuronal progenies appear. Such increases re-
sulted in marked elevations of melanocyte numbers in the skin.
We used PLP-CreERT2 mice to genetically trace the progeny of
PLP+SCPs unequivocally identifying SCPs as a cellular origin of
melanocytes during mouse development. This data also shows
that DCT+/Mitf+mature pigmented melanocytes derived from
SCPs appear at cutaneous sites in the postnatal animal.
experiments corroborate our conclusion that SCPs along nerves
innervating cutaneous tissues are an origin of melanocytes.
Our findings agree with, and may explain, some previous
noblasts gives rise to coat pigmentation (Mintz, 1967), more
recent data show that there is a large number of melanoblast
progenitors (Wilkie et al., 2002). Furthermore, when single
NCCs were labeled by genetics, melanocytes coming from
a single originating cell were located in patches in the skin inter-
mingled with unlabeled cells. This shows that the progenitors
disperse and mix but populate the skin in a defined patch (Hus-
zar et al., 1991; Wilkie et al., 2002). It is conceivable that melano-
cytes arising from SCPs in cutaneous nerves could explain the
observed patches of pigmentation which forms when individual
labeled NCCs are studied. This phenotype could appear if cell
division of rapidly migrating NCCs is limited and expansion of
SCP and melanoblast numbers largely takes place within and
around nerves terminating in specific cutaneous locations.
How Is Hmx1 Regulating Neuronal versus Melanocyte
Cell Fates in the Ventral Migratory Pathway?
A melanoblast fate has been intimately linked with the timing of
NC delamination. The NCCs migrate in chain-like structures
with sympathetic neurons forming first, DRG neurons later and
glial cells throughout this period while melanocytes are born
from the last NCCs delaminating from the neural tube. The com-
mitment of NCCs to a neuronal fate takes largely place within the
coalescing ganglion where Neurogenins (Ngn1 and Ngn2) bias
while Isl1 and FoxS1 marks a commitment to a sensory and
neuronal fate, respectively (Zirlinger et al., 2002; Marmigere
and Ernfors, 2007; Montelius et al., 2007). Those cells failing to
initiate neurogenesis remain transiently as Sox10+progenitors
and are largely localized to the border of the ganglion, hence
termed DRG border cells (Montelius et al., 2007). How does
expression of Hmx1 fit into neurogenesis in the DRG and with
the decision between a neuronal versus Sox10+progenitor cell
fate? We find that Hmx1 is not expressed in the migrating
NCCs but is turned on within the DRG, coinciding with Islet-1
expression and that elimination of its expression using RNAi
leads to near complete loss of neurogenesis. Instead, the cells
remain as Sox10+cells located around the lateral surface of
Figure 7. Regulation of SCP versus Melanocyte Fate by Neuregulin
(A–E) Melanocyte development in erbB3?/?and control mice. (A–D) Mitf and
Tuj1 labeling of wild-type and erbB3?/?mice at E10 (A and B) and E12 (C
and D). Arrow in the inset (D) points at melanoblast associated to the nerve
fiber. (E) Quantification of Mitf+melanoblast numbers in control and erbB3?/?
embryos at E10 and E12 (n = 4 animals/group; error bars represent SEM;
Students t test **p < 0.01,). Abbreviations are as follows: DRG, dorsal root
ganglion; dr, dorsal ramus; NT, neural tube. The scale bars represent 50 mm.
376 Cell 139, 366–379, October 16, 2009 ª2009 Elsevier Inc.
the DRG and shortly thereafter many acquire Mitf expression.
The results imply Hmx1 as critical for decisions between neu-
ronal and non-neuronal fates and also raise the possibility that
it could repress the melanocyte fate and therefore in its absence,
the cells in our experiments acquire a melanocyte fate. Although
we can not exclude this possibility, it seems more likely that the
melanocyte fate is secondary to the failure in neurogenesis.
Differentiation of border cells into melanocytes could either be
a consequence of instructive signals imposing this fate on the
multipotent Sox10+border cells or alternatively a default differ-
entiation in the absence of instructive signals for other fates.
In invertebrate chordates, the ascidian urochordate Ecteinas-
cidia turbinate, NC-like cells differentiate only into pigment cells.
This opens for that evolution of NCCs begins with the generation
of pigment cells and have gained additional functions to gen-
erate the full repertoire of cell types of the vertebrate NC lineage
(Jeffery et al., 2004). This would imply that within NCCs, the
melanocyte lineage isa default fate and the diversity of cell types
coordinated signals maintaining their multipotency and instruct-
ing their differentiation into different cell types.
The Multipotent Schwann Cell Precursor
Early embryonic nerves are exclusively built from axon bundles
and NC-derived cells. The NC-derived cells are intimately asso-
ciated with the nascent nerves and start migrating their way
Once NCCs associate with axons of nascent nerves, they are
considered as SCPs and express markers such as cadherin
ness to survival factors compared with NCCs (Jessen and Mir-
sky, 2005). We find that this population of cells may also be
the source of many melanocytes in the skin, suggesting that
these cells share characteristics with multipotent stem cells.
Hence, the growing nerves might be considered as stem/pro-
genitor cell niches from which via inductive recruitment diverse
cell types could be generated. We identify melanocytes as one
such cell type, but the NC contributes with numerous cell types
to a large number of tissues during embryogenesis. Future
studies will have to address the role of the nerve SCPs as
a cellular source for differentiated cell types in other tissues
The competence of Krox20+myelinating Schwann cells
derived from the deafferented adult nerves to differentiate into
melanocytes in our experiments is likely a consequence of
that cessation of proliferation in mature myelinating Schwann
cells is reversible, as these cells can de-differentiate and re-
enter the cell cycle upon nerve injury (Stewart et al., 1993;
Dupin et al., 2003). A developmental origin of melanocytes
from SCPs, a competence of Krox20+Schwann cells and
possibly of undifferentiated SCPs that could remain in the adult
to generate melanocytes, may open insights in the numerous
observed hypo- and hyper-pigmentation disorders caused by
changes in the number of melanocytes, in mechanisms of re-
pigmentation and in the association between neurological
disorders and changes in skin pigmentation. For instance, neu-
rofibromatosis type 1 patients which develop peripheral nerve
tumors consisting mainly of Schwann cells and nerve sheath
tumors also show skin hyperpigmentation (Fetsch et al., 2000;
Weinreb et al., 2007). Recently, genotoxic stress was shown
to result in melanocyte stem cell (MSC) depletion and irrevers-
ible hair graying due to their unscheduled differentiation (Ino-
mata et al., 2009). The role of SCPs in the formation of new
melanocytes in the adult, their role during age-related hair
graying and the impact of genotoxic stress on a putative contri-
bution of melanocytes from SCPs in the adult remains to be
Nerve-SCP Interactions and Soluble Signals Regulating
Our results suggest that melanocyte development shares
several signaling molecules with Schwann cell development.
NRG1 plays an important role as an axonal signal for survival
and proliferation of SCPs (Jessen and Mirsky, 2005). NRG1 is
evidently not only a survival and proliferation signal for SCPs,
given that the direct contact between the nerve and SCPs that
is a prerequisite for Neuregulin signaling in SCPs, appears to
suppress a melanocyte fate. In erbB3?/?mice we observed
a significant increase of melanoblasts in and around the devel-
oping nerves despite significant loss of SCPs. Further, in DRG
cultures containing border cells and in SCP cultures, NRG1 effi-
ciently suppressed melanocyte differentiation.
The dependence on NRG1 for cell survival of SCPs ends with
the transition to immature Schwann cells at which time the
Schwann cells produce and respond to the survival factors
IGF1 and PDGF (Meier et al., 1999). Our results show that IGF1
and PDGF efficiently instruct a differentiation of SCPs along
a melanocyte cell lineage and that IGF/PDGF signaling and Neu-
regulin signaling maycompetewith opposingresults.Hence, itis
conceivable that as glial precursors expand in numbers Neure-
gulin signaling becomes limiting due to competition for nerve
contact. In parallel, as increasing amounts of SCPs differentiate
into immature Schwann cells and levels of IGF1/PDGF increase,
the balance of Neuregulin and IGF1/PDGF signaling is shifted in
favor of melanocyte differentiation. A similar mechanism might
occur in the adult during nerve damage which leads to increased
expression of IGF1/PDGF (Meier et al., 1999) and a reduced or
complete loss of the nerve-derived Neuregulin signaling.
The generation and characterization of PLP-CreERT2, erbB3?/?and Egr2Cre/+
R26R-YFP (Krox20YFP) mice have been previously described (Leone et al.,
2003; Maro et al., 2004; Riethmacher et al., 1997).
In Situ Hybridization and Immunohistochemistry
Embryos were fixed in 4% paraformaldehyde/PBS, cryoprotected and
sectioned at 14 mm thickness. In situ hybridization experiments were per-
formed as previously described (Marmigere et al., 2006). Detailed protocols
and information on antibodies, immunohistochemistry and ISH are provided
as Supplemental Data.
In Ovo Transfection
In ovo microinjections and electroporation of plasmids and siRNAs were
carried out as previously described (Marmigere et al., 2006). siRNA sequences
are detailed in the Supplemental Data.
Cell 139, 366–379, October 16, 2009 ª2009 Elsevier Inc. 377
BrdU Administration In Ovo
BrdU in a concentration of 100 mM was applied to chick embryos in ovo,
followed by incubation for 12 hr at 38?C with subsequent fixation.
DRGs were removed entirely from the brachial to the first thoracic segments at
HH22. For the dorsolateral skin epithelial-mesodermal ablation, the dorso-
lateralpresumptive skin together withtheunderlyingmesodermand thedorsal
part of the dermamyotome were removed from the brachialto the first thoracic
segments at HH22. For the axotomization, spinal nerves of the brachial plexus
were sectioned on the left side of HH27 embryos. A more detailed description
is outlined in the Supplemental Data.
Nerve Transection in Adult Mice
Theright sciatic nerves ofeight 4-week-old Krox20YFPmicewereexposed and
ligated at mid-thigh level with sterile suture silk, and the nerve was cut distally
to the suture. Animals were sacrificed after 80 days. A more detailed descrip-
tion is outlined in the Supplemental Data.
DRG and Nerve Cultures
HH29 DRGs and nerves were cultured in 24-well plates in melanocytic differ-
entiation medium. See Supplemental Data for details.
Slice Tissue Cultures
Chick embryos were transfected in ovo at HH13 with the GFP plasmid (1 mg/ml)
together with siRNA4 and siRNA42 (1.5 mg/ml each). Preparation of slice tissue
cultures from HH24 and HH27 chick embryos followed the method of Lopez-
Bendito et al. (Lopez-Bendito et al., 2006) adapted for chick embryos. Details
are included in the Supplemental Data.
Supplemental data include Supplemental Experimental Procedures, Supple-
mental References, and thirteen figures and can be found with this article
online at http://www.cell.com/supplemental/S0092-8674(09)01043-5.
We thank Michael Wegner, Vince Hearing, and Yoshio Wakamatsu for the
generousgifts of antibodiesdirected against Sox10, Dct and MEBL-1, respec-
tively. This work was supported by the Swedish Research Council, the
Swedish Foundation for Strategic Research, Linne ´ grants (CEDB and DBRM
grants), the Swedish Cancer Foundation, Swedish Child Cancer Foundation,
the Swedish Brain Foundation, Bertil Ha ˚llsten Research Foundation, EU FP7
MOLPARK collaborative project, and ERC advanced grant (232675) to P.E.,
and Knut and Alice Wallenberg Foundation (CLICK Imaging Facility). I.A. was
supported by the Swedish Research Council. F.L. was supported by the
Swedish Medical Research Council (K2007-77PK-20285-01-6) and the Euro-
pean Union (Marie Curie MEIF-CT-2006-039237). U.S. was supported by the
Swiss National Science Foundation and the National Center for Competence
in Research Neural Plasticity and Repair.
Received: July 1, 2008
Revised: April 3, 2009
Accepted: July 22, 2009
Published: October 15, 2009
and Saint-Jeannet, J.P. (2003). Sox10 regulates the development of neural
crest-derived melanocytes in Xenopus. Dev. Biol. 259, 19–33.
Bronner, M.E., and Cohen, A.M. (1979). Migratory patterns of cloned neural
crest melanocytes injected into host chicken embryos. Proc. Natl. Acad. Sci.
USA 76, 1843–1847.
Dorris, F. (1938). The production of pigment in vitro by chick neural rest. Arch
Entwicklungsmech Organ 138, 323–334.
Dorris, F. (1939). The production of pigment by chick neural crest in grafts to
the 3-day limb bud. J. Exp. Zool. 86, 315–345.
Dorsky, R.I., Raible, D.W., and Moon, R.T. (2000). Direct regulation of nacre,
a zebrafish MITF homolog required for pigment cell formation, by the Wnt
pathway. Genes Dev. 14, 158–162.
gene transfer demonstrates Wnt1 role in melanocyte expansion and differen-
tiation during mouse development. Proc. Natl. Acad. Sci. USA 97, 10050–
Dupin, E., Real, C., Glavieux-Pardanaud, C., Vaigot, P., and Le Douarin, N.M.
(2003). Reversal of developmental restrictions in neural crest lineages: transi-
tion from Schwann cells to glial-melanocytic precursors in vitro. Proc. Natl.
Acad. Sci. USA 100, 5229–5233.
DuShane, G. (1935). An experimental study of the origin of pigment cells in
amphibian. J. Exp. Zool. 72, 1–31.
Erickson, C.A. (1993). From the crest to the periphery: control of pigment cell
migration and lineage segregation. Pigment Cell Res. 6, 336–347.
Fetsch, J.F., Michal, M., and Miettinen, M. (2000). Pigmented (melanotic)
neurofibroma: a clinicopathologic and immunohistochemical analysis of 19
lesions from 17 patients. Am. J. Surg. Pathol. 24, 331–343.
Fox, M.H. (1949). Analysis of some phases of melanoblast migration in the
barred Plymouth Rock embryos. Physiol. Zool. 22, 1–22.
Gambardella, L., Schneider-Maunoury, S., Voiculescu, O., Charnay, P., and
Barrandon, Y. (2000). Pattern of expression of the transcription factor Krox-
20 in mouse hair follicle. Mech. Dev. 96, 215–218.
Gray, P.A., Fu, H., Luo, P., Zhao, Q., Yu, J., Ferrari, A., Tenzen, T., Yuk, D.I.,
Tsung, E.F., Cai, Z., et al. (2004). Mouse brain organization revealed through
direct genome-scale TF expression analysis. Science 306, 2255–2257.
development of the prospective neural crest in Xenopus embryos. Dev. Biol.
Huszar, D., Sharpe, A., Hashmi, S., Bouchard, B., Houghton, A., and Jaenisch,
R. (1991). Generation of pigmented stripes in albino mice by retroviral marking
of neural crest melanoblasts. Development 113, 653–660.
Inomata, K., Aoto, T., Binh, N.T., Okamoto, N., Tanimura, S., Wakayama, T.,
Iseki, S., Hara, E., Masunaga, T., Shimitzu, H., and Nishimura, E.K. (2009).
Genotoxic stress abrogates renewal of melanocyte stem cells by triggering
their differentiation. Cell 137, 1088–1099.
like cells form body pigmentation in a urochordate embryo. Nature 431,
Jessen, K.R., and Mirsky, R. (2005). The origin and development of glialcells in
peripheral nerves. Nat. Rev. Neurosci. 6, 671–682.
Jin, E.J., Erickson, C.A., Takada, S., and Burrus, L.W. (2001). Wnt and BMP
signaling govern lineage segregation of melanocytes in the avian embryo.
Dev. Biol. 233, 22–37.
Johnston, M.C. (1966). A radioautographic study of the migration and fate of
cranial neural crest cells in the chick embryo. Anat. Rec. 156, 143–155.
Le Douarin, N. (1973). A biological cell labeling technique and its use in exper-
mental embryology. Dev. Biol. 30, 217–222.
Lemke, G.E., and Brockes, J.P. (1984). Identification and purification of glial
growth factor. J. Neurosci. 4, 75–83.
Leone, D.P., Genoud, S., Atanasoski, S., Grausenburger, R., Berger, P.,
Metzger, D., Macklin, W.B., Chambon, P., and Suter, U. (2003). Tamoxifen-
inducible glia-specific Cre mice for somatic mutagenesis in oligodendrocytes
and Schwann cells. Mol. Cell. Neurosci. 22, 430–440.
378 Cell 139, 366–379, October 16, 2009 ª2009 Elsevier Inc.
Lopez-Bendito, G., Cautinat, A., Sanchez, J.A., Bielle, F., Flames, N., Garratt, Download full-text
A.N., Talmage, D.A., Role, L.W., Charnay, P., Marin, O., and Garel, S. (2006).
Tangential neuronal migration controls axon guidance: a role for neuregulin-
1 in thalamocortical axon navigation. Cell 125, 127–142.
Mackenzie,M.A.,Jordan,S.A., Budd,P.S., andJackson,I.J.(1997).Activation
of the receptor tyrosine kinase Kit is required for the proliferation of melano-
blasts in the mouse embryo. Dev. Biol. 192, 99–107.
Marmigere,F., Montelius, A., Wegner, M., Groner, Y., Reichardt, L.F., and Ern-
fors, P. (2006). The Runx1/AML1 transcription factor selectively regulates
developmentand survival of TrkA nociceptive sensory neurons.Nat. Neurosci.
Marmigere, F., and Ernfors, P. (2007). Specification and connectivity of
neuronal subtypes in the sensory lineage. Nat. Rev. Neurosci. 8, 114–127.
Maro, G.S., Vermeren, M., Voiculescu, O., Melton, L., Cohen, J., Charnay, P.,
and Topilko, P. (2004). Neural crest boundary cap cells constitute a source of
neuronal and glial cells of the PNS. Nat. Neurosci. 7, 930–938.
Meier, C., Parmantier, E., Brennan, A., Mirsky, R., and Jessen, K.R. (1999).
Developing Schwann cells acquire the ability to survive without axons by es-
tablishing an autocrine circuit involving insulin-like growth factor, neurotro-
phin-3, and platelet-derived growth factor-BB. J. Neurosci. 19, 3847–3859.
Mintz, B. (1967). Gene control of mammalian pigmentary differentiation.
I. Clonal origin of melanocytes. Proc. Natl. Acad. Sci. USA 58, 344–351.
Montelius, A., Marmigere, F., Baudet, C., Aquino, J.B., Enerback, S., and Ern-
fors, P. (2007). Emergence of the sensory nervous system as defined by Foxs1
expression. Differentiation 75, 404–417.
Nakagawa, S.,and Takeichi, M.(1998).Neuralcrestemigrationfrom theneural
tube depends on regulated cadherin expression. Development 125, 2963–
Rawles, M. (1947). Origin of pigment cells from neural crest in the mouse
embryo. Physiol. Zool. 20, 248–270.
Riethmacher, D., Sonnenberg-Riethmacher, E., Brinkmann, V., Yamaai, T.,
Lewin, G.R., and Birchmeier, C. (1997). Severe neuropathies in mice with tar-
geted mutations in the ErbB3 receptor. Nature 389, 725–730.
Ris, H. (1941). An experimental study on the origin of melanophorese in birds.
Physiol. Zool. 14, 48–66.
Rizvi, T.A., Huang, Y., Sidani, A., Atit, R., Largaespada, D.A., Boissy, R.E., and
Ratner, N. (2002). A novel cytokine pathway suppresses glial cell melanogen-
esis after injury to adult nerve. J. Neurosci. 22, 9831–9840.
Serbedzija, G.N., Bronner-Fraser, M., and Fraser, S.E. (1989). A vital dye anal-
ysis of the timing and pathways of avian trunk neural crest cell migration.
Development 106, 809–816.
Serbedzija, G.N., Fraser, S.E., and Bronner-Fraser, M. (1990). Pathways of
trunk neural crest cell migration in the mouse embryo as revealed by vital
dye labelling. Development 108, 605–612.
Steingrimsson,E.,Copeland, N.G.,and Jenkins,N.A.(2004). Melanocytes and
the microphthalmia transcription factor network. Annu. Rev. Genet. 38,
synthesis rate in the Schwann cell lineage in vivo are correlated with the
precursor–Schwann cell transition and myelination. Eur. J. Neurosci. 5,
Thomas, A.J., and Erickson, C.A. (2008). The making of a melanocyte: the
specification of melanoblasts from the neural crest. Pigment Cell Melanoma
Res 21, 598–610.
Topilko, P., Schneider-Maunoury, S., Levi, G., Baron-Van Evercooren, A.,
Chennoufi, A.B., Seitanidou, T., Babinet, C., and Charnay, P. (1994). Krox-20
controls myelination in the peripheral nervous system. Nature 371, 796–799.
Twitty, V. (1936). Correlated genetic and embryological experiments on tritu-
rus. I and II. J. Exp. Zool. 74, 239–302.
Weinreb, I., Labonte, S., Goli, H., Chetty, R., and Ghazarian, D. (2007). Cuta-
neous melanocytoneuroma: the first case of a distinctive intraneural tumor
with dual nerve sheath and melanocytic differentiation. J. Cutan. Pathol. 34
(Suppl 1), 52–57.
Weston, J.A. (1963). A radioautographic analysis of the migration and localiza-
tion of trunk neural crest cells in the chick. Dev. Biol. 6, 279–310.
Wilkie, A.L., Jordan, S.A., and Jackson, I.J. (2002). Neural crest progenitors of
the melanocyte lineage: coat colour patterns revisited. Development 129,
Voiculescu, O., Charnay, P., and Schneider-Maunoury, S. (2000). Expression
pattern of a Krox-20/Cre knock-in allele in the developing hindbrain, bones,
and peripheral nervous system. Genesis 26, 123–126.
Zirlinger, M., Lo, L., McMahon, J., McMahon, A.P., and Anderson, D.J. (2002).
Transient expression of the bHLH factor neurogenin-2 marks a subpopulation
of neural crest cells biased for a sensory but not a neuronal fate. Proc. Natl.
Acad. Sci. USA 99, 8084–8089.
Cell 139, 366–379, October 16, 2009 ª2009 Elsevier Inc. 379