β-Catenin activity in the dermal papilla of the hair
follicle regulates pigment-type switching
David Enshell-Seijffersa,1, Catherine Lindona, Eleanor Wua, Makoto M. Taketob, and Bruce A. Morgana,1
aCutaneous Biology Research Center, Harvard Medical School and Massachusetts General Hospital, Charlestown, MA 02129; andbDepartment of
Pharmacology, Graduate School of Medicine, Kyoto University, Yoshida-Konoé-cho, Sakyo, Kyoto 606-8501, Japan
Edited* by Clifford J. Tabin, Harvard Medical School, Boston, MA, and approved October 26, 2010 (received for review May 27, 2010)
The switch between black and yellow pigment is mediated by the
interaction between Melanocortin receptor 1 (Mc1r) and its antag-
onist Agouti, but the genetic and developmental mechanisms that
modify this interaction to obtain different coat color in distinct en-
vironments are poorly understood. Here, the role of Wnt/β-catenin
signaling in the regulation of pigment-type switching was studied.
Loss and gain of function of β-catenin in the dermal papilla (DP) of
the hair follicle results in yellow and black animals, respectively.
β-Catenin activity in the DP suppresses Agouti expression and acti-
vates Corin, a negative regulator of Agouti activity. In addition,
β-catenin activity in the DP regulates melanocyte activity by a
mechanism that is independent of both Agouti and Corin. The co-
ordinate and inverse regulation of Agouti and Corin renders pel-
age pigmentation sensitive to changes in β-catenin activity in the
DP that do not alter pelage structure. As a result, the signals that
specify two biologically distinct quantitative traits are partially
uncoupled despite their common regulation by the β-catenin path-
way in the same cells.
lutionary change, in part because of the knowledge of genes in-
volved in pigmentation and their developmental interactions,
and in part because strong selective pressure drives dramatic and
quantifiable variation in closely related populations adapting to
different environments (1). In several examples studied, this var-
iation is driven by modulation of a receptor-ligand system that
regulates pigment-type switching (2–7). Activity of Mc1r pro-
motes the production of black pigment (eumelanin), whereas
inhibition of Mc1r activity shifts the balance toward the produ-
ction of yellow pigment (pheomelanin) (8). In the absence of
both agonists and antagonists, basal activity of Mc1r is sufficient
for signaling that supports black pigment production in mice (9,
10). Mc1r activity is augmented by agonists such as α-MSH (11–
13), whereas production of yellow pigment requires the antago-
nistic binding of Agouti to Mc1r (14). The effect of Agouti on
pigment type switching depends on two additional components,
Attractin and Mahagonin, that are epistatically downstream of
Agouti and upstream of Mc1R, and together with it comprise the
Agouti signaling pathway (15–19).
In mouse pelage, pigment production and deposition are re-
stricted to the hair follicle and hair shaft, respectively. During the
active growth phase (anagen) of the mature hair follicle, pigment
is synthesized by melanocytes resident in the hair bulb at the base
of the follicle and adjacent to the dermal papilla (DP), a spe-
cialized mesenchymal component of the hair follicle that plays
important roles in controlling follicle morphogenesis, stem cell
activity, hair shaft formation, and pigmentation (20–22). Kera-
tinocytes in the hair bulb that give rise to the inner layers of the
hair shaft take up pigment from nearby melanocytes as part of
their differentiation program, leading to the formation of pig-
Mc1r receptor is specifically expressed on the surface of
melanocytes throughout the growth phase of the hair cycle. In
oat-color variation and adaptation is a model system for
studying the genetic basis of phenotypic diversity and evo-
contrast, a sharp peak of Agouti expression occurs in DP cells
during the early growth phase of the hair cycle (20, 21, 23). This
peak generates a narrow window in which binding of Agouti
sufficient to suppress Mc1r activity occurs while the distal seg-
ment of the hair shaft is formed. The resultant provisional switch
to pheomelanin deposition generates a subapical yellow band in
an otherwise black hair. Despite the predominance of black
pigment, the presence of lighter pigment in the hair tip creates
the overall appearance of a mottled brown hair coat that pro-
vides adaptive coloration in the natural environment (1). Modest
variations in the length of this apical pheomelanin band can dra-
matically alter coat appearance and represent one mechanism by
which adaptive coloration changes occur (2, 7, 21).
The interaction between Mc1r and Agouti is modified by other
genes. Pomc encodes the precursor of α-MSH, which binds to
Mc1r and both directly augments its activity and competitively
inhibits Agouti binding (11, 14). β-Defensin also binds to Mc1r
binding to Mc1r, but the direct interaction between β-defensin
and Mc1r does not by itself change Mc1r signaling (24, 25). Corin
encodes a transmembrane serine protease that is expressed spe-
cifically in the DP and modifies Agouti signaling by narrowing the
period of effective Agouti activity downstream of Agouti expres-
sion (21). In the absence of Corin, Agouti activity is prolonged
and the yellow band is extended leading to lighter coat color.
The DP-specific expression of Agouti and Corin illustrates the
important role the DP plays in controlling melanocyte behavior
and pigmentation. In contrast with the sustained gene expression
changes in the DP during the growth phase, the transient peak of
Agouti expression reveals an additional level of transcriptional
regulation within that phase that contributes to the regulation of
pigment type switching. Several signaling pathways such as BMP,
TGF, Notch, SHH, and Wnt/β-catenin are known to operate in
the hair bulb and may modulate the activity of the DP to regulate
melanocyte behavior (26). Here, the role of Wnt/β-catenin ac-
tivity in the DP in the regulation of signals that direct pigment
type switching in melanocytes was studied.
Ablation of β-Catenin in the DP Results in Yellow Coat Color. We
have reported that when the β-catenin gene was specifically de-
leted in the DP during the midanagen phase by using a DP-
specific cre line (Cor-cre), dramatic reductions in hair growth are
observed (22). In these experiments, performed in a functional
absence of Agouti (a/a), subtle effects on hair color (Fig. S1) are
partially masked by defects in hair coat structure and cycling.
Author contributions: D.E.-S. and B.A.M. designed research; D.E.-S., C.L., and E.W. per-
formed research; M.M.T. contributed new reagents/analytic tools; D.E.-S. and B.A.M. an-
alyzed data; and D.E.-S. and B.A.M. wrote the paper.
The authors declare no conflict of interest.
*This Direct Submission article had a prearranged editor.
1To whom correspondence may be addressed. E-mail: email@example.com.
edu or firstname.lastname@example.org.
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.
| December 14, 2010
| vol. 107
| no. 50www.pnas.org/cgi/doi/10.1073/pnas.1007326107
However, when the same experiments are performed in the
presence of Agouti (A/a), the mottled brown coat is converted to
a yellow color (Fig. 1). In this Cor-cre line, cre recombinase ac-
tivity is first detected at postnatal day (P)3 (22). Therefore, hair
follicle development, including the recruitment of melanocytes
to the developing follicle, occurs in the presence of an intact
β-catenin gene in mice of the genotype Cor-cre/+;Ctnnb1Del/Flox.
Deletion of the floxed β-catenin allele occurs during the early to
midanagen phase (P3–P8) of the hair cycle (22) as Agouti ex-
pression declines from its peak at P3 to basal levels.
As in wild-type mice, the longer guard hairs are black in the
mutant (Fig. 1 A and B), but the undercoat is dramatically
lightened. The undercoat is composed of three hair types. Within
the awl population, 90% of the hairs are completely black in
wild-type mice (Fig. 1C; Y = 0). In contrast, only 10% in the
mutant are black and the majority exhibit a yellow band that
extends to the base of the hair shaft (Fig. 1 C and F; Y = 1).
Their apical tips remain black as in wild-type mice, consistent
with the timing of cre activity during midanagen (Fig. 1F). It is
noteworthy that a small population of mutant awls have a broad
yellow band but nevertheless switch back to black pigment at the
base of the hair (Fig. 1C; 0.5 ≤ Y <1). Whether the result of
mosaic excision in the DP of this subset of follicles, or a differ-
ential requirement for β-catenin signaling, these follicles dem-
onstrate that changes in the strength of β-catenin signaling may
shift the balance toward pheomelanin production but need not
result in an absolute block to eumelanin production.
All wild-type and mutant zigzag hairs start with a black tip
followed by a subapical yellow band (Fig. 1E). However, the
pheomelanin band is extended in mutant zigzag hairs, both in
absolute terms and as a fraction of total hair length. In most
mutant zigzags, the pheomelanin band extends past the first
oblique bend, indicating a prolonged period of pheomelanin
production (Fig. 1D). In contrast with awls, only a small pro-
portion of mutant zigzag hairs exhibit yellow pigment deposition
all of the way to the basal end (Fig. 1 D and E; Y = Club).
β-Catenin Activity in the DP Suppresses Agouti Expression. Real-time
PCR analysis of wild-type and mutant whole-skin preparations
from P1–10 revealed the pattern and levels of Agouti expression
are altered in the mutant (Fig. 2A). Both wild-type and mutant
show a similar bell-shaped curve of Agouti levels peaking at P4,
before alteration in β-catenin activity in the DP. However, the
drop of Agouti expression in wild type is sharper and settles at
basal levels 10-fold lower than that of the mutant. This change in
expression was confirmed by in situ hybridization (Fig. 2 B and
C). Agouti transcripts are readily observed at P4 in the DP of
both wild-type and mutant mice (Fig. 2B). In contrast, Agouti
transcripts are detected at P8 in mutant DP only when the de-
tection reaction was prolonged (Fig. 2C, Left), whereas the basal
level of Agouti expression in wild-type P8 mice was not detected
under these conditions (Fig. 2C, Right). Regardless of genotype,
no Agouti transcripts are detected in some follicles at P4, con-
sistent with the presence of completely black guard and awl hairs
in both wild-type and mutant mice.
β-Catenin Activity in the DP Activates Corin Expression. Pheomelanin
production in the presence of the low levels of Agouti expression
observed in the mutant at later stages is unexpected. Corin nor-
mally inhibits Agouti activity when Agouti transcript levels are
low (21). Real-time PCR analysis of RNA prepared from whole
skin from P1–10 was performed for Corin and Prss12 (Fig. 2D).
Prss12 is a DP-specific gene whose expression remains unaltered
in DP cells lacking β-catenin sorted from P9 mice (22) (see also
Fig. S2). As expected, no change in Prss12 expression between
wild type and mutant was observed from P1–10. In contrast,
Corin expression in the mutant is reduced from midanagen on-
wards. This decrease was also confirmed by real-time PCR analysis
of purified DP cells FACS-sorted from P9 mice (Fig. S2) and by
immunostaining for Corin in P8 mice (Fig. 2 E and F). Thus,
β-catenin signaling in the DP promotes darker coat color both by
suppressing Agouti and enhancing Corin expression. Further-
more, as Agouti expression remains unaltered in Corin mutants
(21), β-catenin suppression of Agouti is not mediated by Corin.
β-Defensin and Pomc also inhibit Agouti activity and changes
in the expression of these genes might also underlie the efficient
inhibition of Mc1r activity in the mutant. Real-time PCR of
whole-skin preparations revealed both β-defensin and Pomc ex-
pression levels remain unaltered in mice lacking β-catenin in the
DP throughout the early to midanagen phase (Fig. S3) and, thus,
suggests these genes are not involved in the observed phenotype.
β-Catenin Activity in the DP Promotes Black Pigment Production by an
Additional Mechanism That Is Independent of both Agouti and Corin.
The expression of genes required in the melanocyte for eumela-
Mc1r activity, and their levels provide a more direct assessment of
changes in activity of the Agouti/Mc1r pathway. In wild-type mice
with a functional allele of Agouti (A/a), the expression levels of
Dct, Tyrp1, and Silver are repressed during early anagen when
Agouti is high, whereas their transcript levels increase during the
progression through anagen when Agouti levels decline (Fig. 3A,
yellow lines), consistent with their role in eumelanogenesis and
Three-week-old wild-type (WT) and mutant (Mut) mice are shown after the
first hair cycle. In B, higher magnification of the frame in A is shown to re-
veal the yellow undercoat and black guard hairs. (C) Distribution of awl hairs
according to the basal extension of the pheomelanin band (mean ± SD).
Y indicates the position of the proximal boarder of the pheomelanin band
along the distal-proximal axis of the hair. Y = 0 denotes completely black
hairs, and Y = 1 represents pheomelanin extension all of the way toward the
base of the hair. Note that ≈90% of awls in wild type are completely black.
(D) Distribution of zigzag hairs according to the basal extension of the
pheomelanin band (mean ± SD). Y is defined as in C with 1 unit representing
1 segment. (E) Examples of wild-type and mutant zigzag hairs with Y < 1 and
Y = club, respectively. (F) Examples of wild-type and mutant awls with Y <
0.5 and Y = 1, respectively.
β-catenin ablation in the DP results in yellow coat color. (A and B)
Enshell-Seijffers et al.PNAS
| December 14, 2010
| vol. 107
| no. 50
their transcriptional activation by Mc1r signaling. In contrast,
expression levels of these genes remain repressed in late anagen
in the mutant, in line with the changes in Agouti and Corin levels.
On a nonagouti (a/a) background, the suppression in eumelano-
genic gene expression observed during early anagen in A/a mice is
absent (Fig. 3A, black lines), confirming that expression of these
genes is inhibited during this period by Agouti activity. However,
deletion of β-catenin in the DP during midanagen results in sig-
nificant reduction in the RNA levels of these genes in the absence
of Agouti (Fig. 3A and Fig. S4).
The changes in RNA levels observed by real-time PCR anal-
ysis on whole skin from nonagouti mice could reflect a change in
the number of melanocytes. Double immunostaining for Mitf
and Tyrosinase was used to identify melanocytes and to score
their numbers at P8 (Fig. 4 A and B). A reduction of 22% in the
number of melanocytes in the mutant was observed, suggesting
that β-catenin activity in the DP controls melanocyte number by
an Agouti-independent mechanism. However, follicular kerati-
nocytes represent a significant fraction of cells in whole skin, and
this fraction is reduced in mice lacking β-catenin in the DP as
a result of decrease in proliferation of matrix keratinocytes (22).
Consequently, the RNA yield from mutant skin is reduced 24%
relative to that from an identical area of wild-type skin (Fig. 4C).
The consequent enrichment in the fraction of RNA derived from
melanocytes by RNA normalization in the real-time PCR anal-
ysis roughly compensates for the reduction in melanocyte num-
ber. This compensation likely explains the lack of change in Mitf
RNA levels between wild-type and mutant skin (Fig. 3A) and
suggests that alterations in the RNA levels derived from eume-
lanogenic genes reflect changes in gene expression per melano-
cyte that are independent of Agouti.
Mutant mice on a nonagouti background continue to produce
eumelanin, but their coat color is distinct from wild type (Fig.
S1). Because structural changes and lack of regeneration may
contribute to the duller appearance of the mutant hair coat,
pigment content was analyzed directly in hair from the first hair
cycle of nonagouti wild-type and mutant mice (27). Mutant hair
samples contain only 56% of the total melanin found in an equal
weight of wild-type hair (Fig. 4D). Note that the size of mutant
hairs is at most half of that of wild type (Fig. 1 E and F; see also
ref. 22) and, thus, 1 mg of mutant hairs corresponds to hair
produced by twice the number of follicles that produce 1 mg of
promotes eumelanogenesis by an Agouti- and Corin-independent mecha-
nism. (A) Real-time PCR analysis compares eumelanogenic gene expression
between mice lacking β-catenin in the DP (Mut) to littermate controls (WT)
on an Agouti (A/a; yellow lines) or nonagouti (a/a; black lines) background
(mean ± SEM). For statistical analysis, see Fig. S4. (B) Eumelanogenic gene ex-
pression in nonagouti mice homozygous or heterozygous for a mutant allele
of Corin between P7–10, a period when robust changes in eumelanogenic
gene expression are observed in nonagouti mice lacking β-catenin in the DP.
β-catenin activity in the DP positively regulates a novel pathway that
(A) Real-time PCR analysis of whole-skin preparations from P1–10 compares
the RNA levels of Agouti between wild type (WT) and mutant (Mut) (mean ±
SD). In both genotypes, Agouti expression declines dramatically after the
peak to stable levels, but these levels are 10-fold higher in the mutant. (B) In
situ hybridization readily detects Agouti transcripts (blue) in the DP of wild-
type and mutant P4 skins. Saturated signals are obtained after 1 d of de-
tection. In both genotypes, follicles with black pigment and no detectable
Agouti transcript can be identified (arrowheads). (C) In situ hybridization for
Agouti in wild-type and mutant P8 skins. After 6 d of detection, Agouti
transcripts are observed only in the mutant. Insets in the upper left corners
show higher magnification of hair bulbs to illustrate the presence and ab-
sence of Agouti transcript in the DP of mutant and wild type, respectively.
Rare follicles with black pigment deposition and lack of Agouti expression
are observed in the mutant (Lower Left Inset). (D) Real-time PCR analysis
monitors the RNA levels of Corin and Prss12 from P1–10 in wild-type and
mutant mice (mean ± SD). (E and F) Immunostaining of Corin in P8 wild-type
(E) and mutant (F) mice. The same follicle is shown in the left and right
images with YFP (green) marking the DP at Left and Corin staining (red) at
Right. Blue labels nuclei.
β-catenin in the DP regulates Agouti and Corin expression inversely.
| www.pnas.org/cgi/doi/10.1073/pnas.1007326107 Enshell-Seijffers et al.
wild-type hairs. Therefore, this analysis assays the melanin pro-
duction from a higher number of melanocytes in the mutant,
even after taking into the account the reduction in melanocyte
number per follicle. It clearly illustrates that a phenotypic con-
sequence of the Agouti-independent reduction in eumelanogenic
gene expression is reduced accumulation of melanin in the
Although the effect of Corin ablation on pigmentation is only
observed in the presence of Agouti (21), the possibility of similar
cryptic changes in the expression of the eumelanogenic genes in
nonagouti mice and a consequent Agouti-independent mecha-
nism of Corin action had not been evaluated. To address this
question, eumelanogenic gene expression was compared be-
tween nonagouti mice homozygous or heterozygous for a mutant
Corin allele (Fig. 3B). No change in gene expression was de-
tected in mice lacking Corin, suggesting that Corin action pro-
motes eumelanogenesis by interfering with Agouti activity and
not by circumventing the pathway by some Agouti-independent
mechanism such as augmenting Mc1r or Pomc activities. Fur-
thermore, this result demonstrates that alterations in expression
of Dct, Tyrp1, and Silver observed in the absence of β-catenin in
the DP of a/a mice are both Agouti- and Corin-independent.
These observations also reveal a heretofore unanticipated third
signaling component from the DP that depends on β-catenin
activity in these cells and acts on melanocytes to increase the
expression levels of eumelanogenic genes.
Constitutively Activated β-Catenin in the DP Results in Black Mice. In
wild-type mice, Corin levels are relatively constant over the ana-
gen phase, whereas Agouti transcript levels peak dramatically at
P4. The loss of function experiments establish a role for β-catenin
activity in suppressing Agouti and sustaining Corin expression
afterthe normal peak of Agouti,but the role ofβ-catenin signaling
in Agouti regulation during the peak remains unclear. The levels
of Agouti transcripts in the absence of β-catenin at late anagen are
dramatically lower than those during the peak, implying that an
independent regulator drives peak expression. Nevertheless,
β-catenin regulation may limit the height and width of this peak
and, thereby, contribute to specifying pheomelanin bandwidth.
This hypothesis could not be tested directly in this experimental
model because deletion of the β-catenin gene during the first hair
cycle occurs after the peak in Agouti expression, whereas follicle
regeneration is sufficiently defective during the second hair cycle
to preclude analysis (22). However, this hypothesis predicts that
increased levels of activated β-catenin in the DP would suppress
Agouti during its peak expression and result in darker hairs. It is
also consistent with the lack of pheomelanin band in most awls of
wild-type mice, because it has been suggested that β-catenin sig-
naling is higher in the DP of awls than that of zigzag hairs (28).
To explore the effect of higher levels of activated β-catenin on
Agouti expression, the Cor-cre line was used in conjunction with
a conditional allele of β-catenin in which exon3 is flanked by loxP
sites (Ctnnb1Flox3) (29). Exon3 encodes a domain that marks
β-catenin for targeted degradation upon phosphorylation, and
deletion of this exon results in production of a constitutively
activated β-catenin protein. No gross change in hair structure
was observed in mice of the genotype Cor-cre/+;Ctnnb1Flox3/+.
No pigmentation phenotype was observed at the end of the first
hair cycle (Fig. 5A), consistent with prevalent activation of the
conditional allele only after Agouti expression has already
dropped below levels sufficient for pheomelanin production.
However, DP cells harboring the activated allele persist through
the hair cycle and follicles regenerated during the second cycle
express the activated allele throughout the growth phase. The
majority of hairs formed during this cycle lack a pheomelanin
band and are completely black (94 ± 3.9% as opposed to 23 ±
3.8% in controls) (Fig. 5 B and C). In situ hybridization con-
firmed the suppression of Agouti expression in mice expressing
constitutive activated β-catenin in the DP during the stages when
Agouti transcripts would normally be at peak levels (Fig. 5 D–G
and Fig. S5). Immunostaining for Corin reveals that although
Corin levels are undetectable at telogen and high during mid-
anagen in both wild type and mutant, Corin levels are sub-
stantially higher in the mutant during the pulse of Agouti
expression in early anagen (Fig. 5 H and I). Thus, both gain and
loss of function experiments illustrate the key role β-catenin
plays in regulating a genetic network that controls pigment-type
switching (Fig. 5J) and demonstrate that varying levels of
β-catenin signaling in the DP can dictate a wide range of coat
This study reveals a genetic network that regulates pigment type
switching. β-Catenin activity in the DP inversely controls the
expression of both Corin and Agouti to coordinately regulate
their levels. As Corin inhibits Agouti activity, this inverse regu-
lation amplifies the effects of changes in β-catenin activity in the
DP on coat color. This analysis also reveals the presence of an
additional level of regulation to control eumelanogenic gene
expression in the melanocyte that depends on β-catenin activity
in the DP. However, the identity and mechanism of action of this
component remain unknown.
The coordinate reduction in hair bulb size and melanocyte
number in the mutant may be explained by an indirect mecha-
nism in which an altered trophic environment of the hair bulb,
whether directly influenced by the DP or indirectly influenced
by keratinocytes, contributes to the regulation of melanocyte
number. The more specific reduction in eumelanogenic gene
expression within melanocytes may also be an indirect response
or may represent a factor expressed by the DP that either acts as
an agonist of Mc1r or enhances the activity of known agonists of
Mc1r such as Pomc-derived α-MSH (Fig. 5J). Alternatively, this
factor may act independently of the Mc1r signal transduction
cascade to promote black pigment production or modify a
(A) Confocal images of P8 wild-type (Upper) and mutant (Lower) follicle
immunostained for Tyrosinase (Left) and Mitf (Center). (B) Melanocyte
number (mean ± SD) was scored by counting Mitf+ cells per follicle per
section in P8 mice. Three hundred follicles from 3 mice per genotype were
analyzed. Rd, reduction in perentage relative to wild type. Two-tailed un-
paired Student’s t test was used (**P < 0.0001). (C) Three dorsal skin biopsies
of 12.6 mm2along the anterior-posterior axis from 11 wild-type and 8 mu-
tant P8 mice were obtained by using skin-biopsy punches of 2-mm radius to
prepare and measure total RNA yield (mean ± SD). Two-tailed unpaired
Student’s t test was used (**P < 0.0001). (D) Absorbance at 500 nm (A500) was
measured for total melanin extracted from 1 mg of hair (mean ± SD). Hair
coat was harvested at P20 after the first hair cycle from 9 mice per genotype.
Two-tailed unpaired Student’s t test was used (**P < 0.0001).
β-Catenin ablation in the DP results in reduced melanocyte activity.
Enshell-Seijffers et al.PNAS
| December 14, 2010
| vol. 107
| no. 50
downstream component of Mc1r signaling. Additional genetic
studies will be required to distinguish between these possible
mechanisms. Until this factor is identified, technical constraints
prevent us from determining whether, like Corin and Agouti, the
activity of this third mechanism is sensitive to levels of β-catenin
activity in early anagen DP in the range that still promotes
normal hair growth. If so, it would act with Corin to further
amplify the response of the pigment system to small changes in
β-catenin activity in the DP. If not, modification of its activity by
a mechanism other than β-catenin activity in the DP would be
expected to shift the set point of pigmentation and could thereby
alter the range of coat color phenotypes that might be attained
by changes in Agouti and Corin expression in response to changes
in β-catenin signaling in the DP.
The genetic alterations that underlay coat color variation il-
luminate mechanisms that drive evolutionary change. Genetic
analysis of coat color variation in natural environments has re-
peatedly identified variants of the Mc1r and Agouti genes as
sources of phenotypic diversity (2–4, 6, 7). Although the central
roles that both Mc1r and Agouti play in pigment switching con-
tributes to this phenomena, the fact that both genes have no
known function outside of pigmentation and are therefore
largely free of other constraints on variation is also relevant. The
requirement for β-catenin in a broad array of tissues and de-
velopmental events places it at the opposite extreme of this
continuum. However, although this general requirement may
constrain the accumulation of mutations in β-catenin itself, the
vast complexity of inputs modifying the canonical Wnt signal
transduction pathway in which this protein functions as a node
provides a wide array of opportunities for genetic divergence less
subject to constraint. The requirement of β-catenin activity in the
DP for hair morphogenesis (22) sets lower limits to the contin-
uum along an effective signaling gradient to generate a lighter
but otherwise normal hair coat. In contrast, the fact that maximal
levels of activated β-catenin in the DP do not grossly affect hair
structure provides no upper limit on the strength of pathway
activation for effective darkening of the hair coat. The comple-
mentary regulation of Corin and Agouti amplify the impact of
more modest changes in Wnt signaling activity on hair pigment.
These mechanisms allow changes within one range of β-catenin
activity to selectively modify hair pigmentation, while changes in
an overlapping range modify both hair structure and color. In
a littermate wild-type control (WT) are shown. In A, the hair coat after the first hair cycle (P20) is shown. In B, the dorsal fur was clipped at P20 to eliminate the
hairs formed in the first cycle. In C, a newly formed hair coat after the second hair cycle (P50) is observed. (D and E) Composite figures composed of tiled
micrographs from single sections show in situ hybridization for Agouti during the early anagen phase of the second hair cycle (P24) and reveal Agouti ex-
pression in wild-type mice (E) and its suppression in ΔEX3 mice (D). (F and G) Three times higher magnification of wild-type and ΔEX3 follicles are shown.
Although the wild-type follicle is in a slightly later stage of the anagen phase than the ΔEX3 follicle, wild-type follicles at earlier stages express detectable levels
of Agouti as well (see Fig. S5). (H and I) Immunostaining for Corin during early anagen (P24) reveals higher Corin levels in the DP of ΔEX3 mice. The same follicle
is shown in Upper and Lower with YFP (green) marking the DP in Upper and Corin staining (red) in Lower. When the anti-Corin antibody was diluted 1:800 (H),
Corin is reliably detected in ΔEX3 but not wild-type follicles. At lower dilutions (I, 1:200), weak staining is also observed in follicles from the same wild-type
mouse. (J) Schematic representation of the genetic network that controls pigment-type switching. Grey wavy arrows represent alternative mechanisms by which
the third signaling component may act to promote black pigment production (Discussion).
Increased β−catenin activity in the DP results in black coat color. (A–C) A mouse expressing constitutively activated β−catenin in the DP (ΔEX3) and
| www.pnas.org/cgi/doi/10.1073/pnas.1007326107 Enshell-Seijffers et al.
this way, dermal papilla niche cells exploit the same signal trans-
duction pathway to direct production of signals that regulate two
apparently independent biological processes.
Materials and Methods
Mice, in Situ Hybridization, Immunostaining, and Melanocyte Counts. Mice used
in this study and detailed procedures are described in SI Materials and
Methods. For in situ hybridization, frozen sections were hybridized with dig-
labeled RNA probe corresponding to nt 126–613 of Agouti (NM_015770). For
immunostaining, fixed-skin sections were incubated with rabbit polyclonal
anti-corin (21) diluted 1:800 or 1:200, mouse monoclonal anti-Mitf (30) di-
luted 1:10, and rabbit polyclonal anti-Tyr (31) diluted 1:500. For melanocyte
count, skin sections were double immunostained for Mitf and Tyr and used
to score Mitf-positive cells in follicle bulbs with a clear zone of Tyr staining.
Hair Shaft Analysis. Hairs were plucked at the end of the first cycle at P20 and
mounted on slides. To collect hairs formed in the second cycle, the hair-coat
was shaved at P20 and newly formed hairs were plucked at P50 after the end
of the second cycle. Hair shafts were photographed as described (21).
Chemical analysis of hair shaft for total melanin was performed as de-
Real-Time PCR. Middorsal skins of wild-type and mutant mice from P1–P10
were collected and used to prepare RNA. Normalized RNA quantities were
reverse transcribed by using random hexamer primers and SuperScript First-
Strand synthesis system III (Invitrogen). For real-time PCR, primer pairs from
SuperArray were used and differences between samples were quantified
based on the ΔΔCt method.
ACKNOWLEDGMENTS. We thank David E. Fisher and Vincent Hearing for
providing the C5 anti-Mitf and αPEP7 anti-Tyr antibodies, respectively, and
R. Czyzewski for technical assistance. This work was supported by National In-
stitute of Arthritis and Musculoskeletal and Skin Diseases Grant 1R01AR055256
1. Hoekstra HE (2006) Genetics, development and evolution of adaptive pigmentation in
vertebrates. Heredity 97:222–234.
2. Nachman MW, Hoekstra HE, D’Agostino SL (2003) The genetic basis of adaptive
melanism in pocket mice. Proc Natl Acad Sci USA 100:5268–5273.
3. Hoekstra HE, Hirschmann RJ, Bundey RA, Insel PA, Crossland JP (2006) A single amino
acid mutation contributes to adaptive beach mouse color pattern. Science 313:
4. Steiner CC, Weber JN, Hoekstra HE (2007) Adaptive variation in beach mice produced
by two interacting pigmentation genes. PLoS Biol 5:e219.
5. Anderson TM, et al. (2009) Molecular and evolutionary history of melanism in North
American gray wolves. Science 323:1339–1343.
6. Kingsley EP, Manceau M, Wiley CD, Hoekstra HE (2009) Melanism in peromyscus is
caused by independent mutations in agouti. PLoS ONE 4:e6435.
7. Linnen CR, Kingsley EP, Jensen JD, Hoekstra HE (2009) On the origin and spread of an
adaptive allele in deer mice. Science 325:1095–1098.
8. Barsh G, Gunn T, He L, Schlossman S, Duke-Cohan J (2000) Biochemical and genetic
studies of pigment-type switching. Pigment Cell Res 13(Suppl 8):48–53.
9. Slominski A, et al. (2005) Preservation of eumelanin hair pigmentation in
proopiomelanocortin-deficient mice on a nonagouti (a/a) genetic background.
10. Smart JL, Low MJ (2003) Lack of proopiomelanocortin peptides results in obesity and
defective adrenal function but normal melanocyte pigmentation in the murine
C57BL/6 genetic background. Ann N Y Acad Sci 994:202–210.
11. Geschwind II (1966) Change in hair color in mice induced by injection of alpha-MSH.
12. Healy E, et al. (2001) Functional variation of MC1R alleles from red-haired individuals.
Hum Mol Genet 10:2397–2402.
13. Jackson IJ, Budd PS, Keighren M, McKie L (2007) Humanized MC1R transgenic mice
reveal human specific receptor function. Hum Mol Genet 16:2341–2348.
14. Ollmann MM, Lamoreux ML, Wilson BD, Barsh GS (1998) Interaction of Agouti protein
with the melanocortin 1 receptor in vitro and in vivo. Genes Dev 12:316–330.
15. Gunn TM, et al. (1999) The mouse mahogany locus encodes a transmembrane form of
human attractin. Nature 398:152–156.
16. Phan LK, Lin F, LeDuc CA, Chung WK, Leibel RL (2002) The mouse mahoganoid coat
color mutation disrupts a novel C3HC4 RING domain protein. J Clin Invest 110:
17. He L, et al. (2003) Spongiform degeneration in mahoganoid mutant mice. Science
18. He L, Eldridge AG, Jackson PK, Gunn TM, Barsh GS (2003) Accessory proteins for
melanocortin signaling: attractin and mahogunin. Ann N Y Acad Sci 994:288–298.
19. He L, et al. (2001) A biochemical function for attractin in agouti-induced pig-
mentation and obesity. Nat Genet 27:40–47.
20. Millar SE, Miller MW, Stevens ME, Barsh GS (1995) Expression and transgenic studies
of the mouse agouti gene provide insight into the mechanisms by which mammalian
coat color patterns are generated. Development 121:3223–3232.
21. Enshell-Seijffers D, Lindon C, Morgan BA (2008) The serine protease Corin is a novel
modifier of the Agouti pathway. Development 135:217–225.
22. Enshell-Seijffers D, Lindon C, Kashiwagi M, Morgan BA (2010) β-catenin activity in the
dermal papilla regulates morphogenesis and regeneration of hair. Dev Cell 18:
23. Vrieling H, Duhl DM, Millar SE, Miller KA, Barsh GS (1994) Differences in dorsal and
ventral pigmentation result from regional expression of the mouse agouti gene. Proc
Natl Acad Sci USA 91:5667–5671.
24. Kerns JA, et al. (2007) Linkage and segregation analysis of black and brindle coat
color in domestic dogs. Genetics 176:1679–1689.
25. Candille SI, et al. (2007) A -defensin mutation causes black coat color in domestic
dogs. Science 318:1418–1423.
26. Sharov AA, et al. (2005) Bone morphogenetic protein (BMP) signaling controls hair
pigmentation by means of cross-talk with the melanocortin receptor-1 pathway. Proc
Natl Acad Sci USA 102:93–98.
27. Ozeki H, Ito S, Wakamatsu K, Hirobe T (1995) Chemical characterization of hair
melanins in various coat-color mutants of mice. J Invest Dermatol 105:361–366.
28. Driskell RR, Giangreco A, Jensen KB, Mulder KW, Watt FM (2009) Sox2-positive
dermal papilla cells specify hair follicle type in mammalian epidermis. Development
29. Harada N, et al. (1999) Intestinal polyposis in mice with a dominant stable mutation of
the beta-catenin gene. EMBO J 18:5931–5942.
30. King R, et al. (1999) Microphthalmia transcription factor. A sensitive and specific
melanocyte marker for MelanomaDiagnosis. Am J Pathol 155:731–738.
31. Jiménez M, Tsukamoto K, Hearing VJ (1991) Tyrosinases from two different loci are
expressed by normal and by transformed melanocytes. J Biol Chem 266:1147–1156.
Enshell-Seijffers et al. PNAS
| December 14, 2010
| vol. 107
| no. 50