ArticlePDF AvailableLiterature Review

The Hair Follicle as an Estrogen Target and Source

Authors:

Abstract and Figures

For many decades, androgens have dominated endocrine research in hair growth control. Androgen metabolism and the androgen receptor currently are the key targets for systemic, pharmacological hair growth control in clinical medicine. However, it has long been known that estrogens also profoundly alter hair follicle growth and cycling by binding to locally expressed high-affinity estrogen receptors (ERs). Besides altering the transcription of genes with estrogen-responsive elements, 17beta-estradiol (E2) also modifies androgen metabolism within distinct subunits of the pilosebaceous unit (i.e., hair follicle and sebaceous gland). The latter displays prominent aromatase activity, the key enzyme for androgen conversion to E2, and is both an estrogen source and target. Here, we chart the recent renaissance of estrogen research in hair research; explain why the hair follicle offers an ideal, clinically relevant test system for studying the role of sex steroids, their receptors, and interactions in neuroectodermal-mesodermal interaction systems in general; and illustrate how it can be exploited to identify novel functions and signaling cross talks of ER-mediated signaling. Emphasizing the long-underestimated complexity and species-, gender-, and site-dependence of E2-induced biological effects on the hair follicle, we explore targets for pharmacological intervention in clinically relevant hair cycle manipulation, ranging from androgenetic alopecia and hirsutism via telogen effluvium to chemotherapy-induced alopecia. While defining major open questions, unsolved clinical challenges, and particularly promising research avenues in this area, we argue that the time has come to pay estrogen-mediated signaling the full attention it deserves in future endocrinological therapy of common hair growth disorders.
Regulatory modules relevant for the hair follicle regulation with partial ties to estrogen signaling. ER regulates the depicted genome network at various endpoints (depicted with encircled numbers 1-5). Factors found to be important for hair follicle patterning, cycling, as well homeostasis are richly intertwined. Paracrine factors are liberated within the hair follicle to be perceived by the same cell in an autocrine fashion, or by adjacent cells in a paracrine fashion. These factors control tissue proliferation as lineage-specific within the hair bulge. ERs have been shown to be upstream or downstream of various regulatory connections (depicted by small arrows). Functional interactions with unknown mechanisms are depicted by question marks. Arrows in principle signify activating or inhibiting interactions. [Modified after van Steensel et al. (375) with permission from Editions John Libbey Eurotext Paris.] NOTCH, Notch receptor; Delta, delta ligand of NOTCH receptor; EDA, ectodysplasin; EDAR, EDA receptor; SRC kinase, Rous sarcoma virus tyrosine kinase; APC, adenomatosis polyposis coli tumor suppressor protein; TCF, T cell factor; LEF1, lymphocyte-enhancing factor 1; SHH, sonic hedgehog protein; PTC, patched; SMO, smoothened; Noggin, Noggin protein; E-cadherin, epithelial cadherin adhesion molecule; Wnt, Drosophila wingless homolog (acronym for wingless-type mouse breast tumor virus); Frz, frizzled; GLI, glioma-associated gene; EN-1, engrailed 1; DLX2, distal-less homeobox gene 2.
… 
Content may be subject to copyright.
The Hair Follicle as an Estrogen Target and Source
Ulrich Ohnemus,* Murat Uenalan,* Jose´ Inzunza, Jan-Åke Gustafsson, and Ralf Paus
Department of Dermatology (U.O.), University Hospital Hamburg-Eppendorf, University of Hamburg, D-20246 Hamburg,
Germany; Department of Paediatric Haematology and Oncology (M.U.), Hannover Medical School, D-30625 Hannover,
Germany; Department of Biosciences and Nutrition (J.I., J.-Å.G.), Karolinska Institute, Novum, SE-14186 Stockholm,
Sweden; and Department of Dermatology (R.P.), University Hospital Schleswig-Holstein, Campus Lu¨beck, University of
Lu¨beck, D-23538 Lu¨beck, Germany
For many decades, androgens have dominated endocrine re-
search in hair growth control. Androgen metabolism and the
androgen receptor currently are the key targets for systemic,
pharmacological hair growth control in clinical medicine. How-
ever, it has long been known that estrogens also profoundly alter
hair follicle growth and cycling by binding to locally expressed
high-affinity estrogen receptors (ERs). Besides altering the tran-
scription of genes with estrogen-responsive elements, 17
-es-
tradiol (E2) also modifies androgen metabolism within distinct
subunits of the pilosebaceous unit (i.e., hair follicle and seba-
ceous gland). The latter displays prominent aromatase activity,
the key enzyme for androgen conversion to E2, and is both an
estrogen source and target.
Here, we chart the recent renaissance of estrogen research
in hair research; explain why the hair follicle offers an ideal,
clinically relevant test system for studying the role of sex
steroids, their receptors, and interactions in neuroectoder-
mal-mesodermal interaction systems in general; and illus-
trate how it can be exploited to identify novel functions and
signaling cross talks of ER-mediated signaling. Emphasizing
the long-underestimated complexity and species-, gender-,
and site-dependence of E2-induced biological effects on the
hair follicle, we explore targets for pharmacological interven-
tion in clinically relevant hair cycle manipulation, ranging
from androgenetic alopecia and hirsutism via telogen efflu-
vium to chemotherapy-induced alopecia. While defining ma-
jor open questions, unsolved clinical challenges, and partic-
ularly promising research avenues in this area, we argue that
the time has come to pay estrogen-mediated signaling the full
attention it deserves in future endocrinological therapy of
common hair growth disorders. (Endocrine Reviews 27:
677–706, 2006)
I.
Introduction
A. Why study the role of estrogens in hair biology?
B. The hair follicle is a prototypic and ideal test system for
studying the role of sex steroids and their receptors in
neuroectodermal-mesodermal interaction systems
C. The cycling hair follicle offers a unique, multipurpose
model for studying estrogen biology
D. Estrogen-related hair research is clinically, psycholog-
ically, and commercially highly relevant
II. Hair Follicle Biology: Relevant Key Facts
A. Hair follicle morphogenesis and cycling
B. Molecular controls of hair follicle cycling
III. Cellular and Molecular Mechanisms of Estrogen Action
A. Estrogen synthesis and metabolism
B. Estrogen receptors
C. Nuclear receptor superfamily
D. Estrogen receptor signaling pathways
E. Estrogen-responsive genes and coregulators of ER
signaling
F. Estrogen target tissues
G. Interdependence of estrogen and androgen signaling
pathways
IV. Estrogens in Dermatoendocrinology
A. Effects of estrogens on the skin
V. Estrogens in Pilosebaceous Unit Biology
A. Estrogen synthesis and metabolism in the piloseba-
ceous unit
B. Estrogen receptor expression in the hair follicle
C. Estrogen target genes in the pilosebaceous unit
D. Species-specific differences in estrogen actions on hair
follicle cycling
E. Gender- and location-specific differences in estrogen
actions on the hair follicle
F. Clinical hair growth effects of estrogens
G. Relevant signaling cross talks in the hair follicle
H. Other potentially important signaling cross talks with
intrafollicularly generated hormones
VI. Open Questions and Unmet Clinical Challenges
VII. Conclusions and Perspectives
I. Introduction
F
OR MORE THAN 70 yr, estrogens have been known to
play a role in skin physiology and hair growth control.
In the 1930s, it was first recognized by Dawson that hair
growth and sexual hormones in animals are closely con-
nected, because in clipped guinea pigs the regrowth of the
hair was faster in spayed than in breeding females (1). A few
years later, Emmens (2) and Hooker and Pfeiffer (3) reported
that parenteral administration of estrogenic hormones in-
First Published Online July 28, 2006
*U.O. and M.U. contributed equally to this review.
Abbreviations: AF, Activation function; AP-1, Activating protein-1;
BMP, bone morphogenetic protein; BMPR, BMP receptor; E2, 17
-es-
tradiol; EGF, epidermal growth factor; EGFR, EGF receptor; ER, estrogen
receptor; ERE, estrogen response element; FGF, fibroblast growth factor
5; HGF, hepatocyte growth factor; IFN-
, interferon-
; IR, immunore-
activity; NGF, nerve growth factor; PPAR, proliferator-activated recep-
tor; REA, repressor of estrogen action; SRC, steroid receptor coactivator;
VEGF, vascular endothelial growth factor.
Endocrine Reviews is published by The Endocrine Society (http://
www.endo-society.org), the foremost professional society serving the
endocrine community.
0163-769X/06/$20.00/0 Endocrine Reviews 27(6):677–706
Printed in U.S.A. Copyright © 2006 by The Endocrine Society
doi: 10.1210/er.2006-0020
677
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 03 November 2015. at 05:08 For personal use only. No other uses without permission. . All rights reserved.
hibited hair growth in rats. In the late 1950s, Ebling (4, 5)
published pioneering work in the field, e.g., by showing that
estrogens increase the mitotic rate in the epidermis of both
rodents and man and reduce the size and activity of seba-
ceous glands. This encouraged the concept that “female”
hormones (estrogens, prolactin, or progesterone) influence
the hair growth cycle, because hair growth waves in female
rats were found to lag behind males, whereas such sex-
dependent hair growth differences were absent in gonadec-
tomized animals (6, 7). In fact, we now know that such
supposedly “female” hormones as, e.g., prolactin are also
produced by males, yet are indeed important modulators of
hair growth; they are even synthesized by the hair follicle
itself, clearly do so across the gender barrier, and are general
modulators of epithelial-mesenchymal tissue interactions in
both sexes (8, 9). As we will discuss below, the latter concept
also holds true for 17
-estradiol (E2) (10–12) (Table 1).
The presence of endogenous estrogens in the skin and
differences in estrogen binding affinities in different regions
of the integument were first demonstrated in murine skin by
titrated estradiol in autoradiographic studies (with estrogen-
related radioactivity found to be localized in the epidermis,
dermal fibroblasts, and the hair follicle) (13, 14). In 1978, an
“ER protein” was detected in murine skin, and titrated es-
tradiol was found to bind specifically to the cytosol of cells
from mouse back skin (15). In addition, it was noted that the
estrogen-binding protein translocates from the cytoplasm
into the nucleus of cutaneous cell populations, both in mice
(15) and in humans (16). Spelsberg and co-workers (16) were
the first to isolate and characterize an estrogen receptor (ER)
in human skin in 1980. Furthermore, a reservoir function of
the skin for steroids was proposed, because titrated estradiol
was retained in the sebaceous glands and the stratum cor-
neum for more than 24 h, implying two penetration path-
ways for estrogen to the dermis: one through the stratum
corneum, and the other through the hair canal and hair
sheaths (17). In addition, more recent studies on the expres-
sion of ER
and ER
using RT-PCR (18) and immunohisto-
chemistry (19) have reported that ER
is the predominant
receptor in human skin, with strong expression in epidermis,
dermal fibroblasts, blood vessels, and hair follicle (19), and
human keratinocytes are reported to express both ER
and
ER
, possibly including a membrane ER
(20).
It took surprisingly many years before these intriguing
leads from the literature were picked up and pursued in
appropriate models.
A. Why study the role of estrogens in hair biology?
1. Estrogens and estrogen metabolism are at least as important as
androgens in male and female hair biology. Androgens are rec-
ognized key regulators of normal human hair growth and the
prerequisite for sexual hair and sebaceous gland develop-
ment (21, 22). However, estrogens also profoundly alter hair
growth in practically all mammalian species investigated
and operate as key modulators of hair follicle biology by
binding to high-affinity cognate receptors (ERs) (19, 23–28).
After a long period of relative dormancy, estrogens have
been rediscovered as hair growth modulators throughout the
past decade. This development was stimulated by a seminal
paper by Oh and Smart (26) in 1996, who showed that the
prototypic ER agonist, E2, after topical application, is a very
potent hair growth inhibitor in mice, thus calling our atten-
tion to similar effects that had already been reported many
decades ago in several mammalian species (1–5) (Table 1).
This hair growth-inhibitory activity reported in mice strik-
ingly contrasted with the supposedly hair growth-stimula-
tory topical E2 therapy long practiced in many countries for
the treatment of female pattern androgenetic alopecia (29, 30)
and the hair loss induced by therapy with aromatase inhib-
itors, which lower serum and tissue E2 levels (31, 32). This
apparent contradiction already suggested that E2 effects on
the mammalian hair follicle were likely to be complex, and
species dependent.
Besides the fact that patients with clinical hair growth
disorders will all probably profit from investigations regard-
ing the molecular pathology of these diseases, there are sev-
eral reasons to systematically reexplore the role of estrogens
in human hair growth control (10–12, 25–28, 33): the hair
follicle 1) offers a microcosmic, prototypic tissue interaction
system (12, 34–37) that allows one to dissect and manipulate
both classical (i.e., ER-mediated) and nonclassical pathways
of estrogen signaling under physiological and pathological
conditions; 2) invites one to study the cross talk of pleiotropic
estrogens with multiple other signaling pathways in complex
neuroectodermal-mesodermal interaction systems; and 3) of-
TABLE 1. Estrogen effects on hair and skin in various species
Species Effect Ref.
Rat Hair waves of female rats lag behind males. Emmens 1942 (2)
Mouse Repressing effect on spontaneous hair waves in pregnant
individuals; regrowth after plucking is not altered.
Fraser and Nay 1953 (361)
Mouse, rat,
and dog
Injection of estrogens inhibits hair growth. Emmens 1942 (2); Gardner and DeVita 1940 (268);
Hooker and Pfeiffer 1943 (3)
Rat and dog Topical estrogen application represses hair growth. Emmens 1942 (2); Whitaker and Baker 1951 (362);
Williams et al. 1946 (267)
Human Human hair follicles aromatize androgens to estrogens. Schweikert et al. 1975 (122)
Mouse and rat Intracutaneous injection of estrogens decreases size of
sebaceous glands and epidermal thickness.
Ebling 1957, 1953 (4, 363); Hooker and Pfeiffer 1943 (3)
Mouse Topical estrogens repress anagen, and the antiestrogen
ICI182.780 promotes anagen hair waves.
Oh and Smart 1996 (26)
The hair follicle is a traditional experimental system for studying estrogen effects in many animal species. In humans and various animal
models, effects of estrogen treatment on skin and hair follicles were noted. In the majority of reports, a repressing phenomenon on hair growth
was noted, whereas anagen-promoting effects of estrogens were reported elsewhere.
678 Endocrine Reviews, October 2006, 27(6):677–706 Ohnemus et al. Hair Follicle as an Estrogen Target and Source
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 03 November 2015. at 05:08 For personal use only. No other uses without permission. . All rights reserved.
fers novel insights into previously unknown estrogen func-
tions and target genes (see below for details).
Estrogens are able to modify androgen metabolism within
distinct subunits of the hair follicle (e.g., in the dermal pa-
pilla), diminishing the amount of 5
-dihydrotestosterone
formed after incubation with testosterone (38). It is not yet
known whether this effect is mediated directly by an inhi-
bition of 5
-reductase within the hair follicle or indirectly
through estrogen-induced increased conversion of testoster-
one to weaker androgens (38). Because aromatase, the en-
zyme that converts testosterone to E2 is also found at many
of the sites of ER and androgen receptor expression (39), the
local balance between E2 and androgen levels may serve to
fine-tune E2 and androgen action in their target cells (40).
This is further supported by the growing evidence that ste-
roid receptors can cross talk with one another, showing an
interdependence of estrogen-, progesterone-, and androgen-
receptor signaling pathways (33, 41).
Also, many of the growth and transcription factors, cyto-
kines, and hormones that are currently recognized to control
hair growth (21, 34–36) are themselves modulated by estro-
gens. Thus, it will be far from easy to clearly distinguish
direct from indirect E2 effects on hair growth, even when
ER-null mice are used, because nonclassical E2 effects could
still alter the expression of many genes that have not pre-
viously been shown to have an estrogen response element
(ERE) (12).
B. The hair follicle is a prototypic and ideal test system for
studying the role of sex steroids and their receptors in
neuroectodermal-mesodermal interaction systems
Besides the hair follicle’s most evident function, the pro-
duction of a hair shaft, it is also an attractive tool for studying
basic biological issues such as cellular differentiation and
neuroectodermal-mesodermal tissue interactions (34, 35).
Hair follicles contain both epithelial and mesenchymal com-
partments (Fig. 1), which are cyclically remodeled and whose
interactions drive hair shaft formation and hair follicle cy-
cling (42). Hair follicle induction and morphogenesis depend
on complex bidirectional communication events between the
epithelium and the underlying mesenchyme (43). Mature
hair follicle mesenchyme is organized in two communicating
compartments: the surrounding connective tissue sheath,
and the follicular dermal papilla. The character of both these
mesenchymal regions changes dramatically over the growth
cycle (44). The dermal papilla is an inductive mesodermal
structure that sends and receives morphogenic signals (45,
46). Its activity depends on continuous and intimate inter-
action with the hair matrix epithelium via native extracel-
lular matrix (47). Anagen dermal papilla dissected free of the
epithelial follicle components and inserted into non-hair-
bearing skin has been shown to induce hair follicle formation
from the resident epithelium (48).
For effective hair follicle induction, continuous and close
dermal papilla contact with the receptive follicle epithelium
is needed; if the papilla is separated from a growing follicle
experimentally (47) or developmentally (hr/hr mouse) (49),
follicle growth ceases. Within the theme of epithelial-mes-
enchymal interactions, it is of interest that keratinocytes may,
in turn, act on the mesenchyme: for example, keratinocytes
produce specific factors, which stimulate the growth of pa-
pilla cells (50).
The growth and regression phases of the hair follicle are
modulated by a broad spectrum of hormones such as go-
nadal, thyroid, adrenal cortical, pituitary, and pineal hor-
mones (35, 51–54). Because the pilosebaceous unit (i.e., hair
follicle, sebaceous gland, and arrector pili muscle) (Figs. 1
and 2) expresses all enzymes to generate androgens as well
as estrogens and is able to convert testosterone to estrogen
(35, 55–58), it must not only be viewed as a recipient of signals
from distant transmitters but rather as an organized com-
munity in which the cells emit, receive, and coordinate mo-
lecular signals from a seemingly unlimited number of distant
sources including established endocrine organs (modern and
classical endocrine functions), neighboring tissues (paracrine
and juxtacrine functions), and the pilosebaceous unit itself
(autocrine and intracrine functions) (35, 54, 59).
C. The cycling hair follicle offers a unique, multipurpose
model for studying estrogen biology
Owing to its lifelong cycling activity, each hair follicle
represents a unique stem cell-rich “microcosmos” that has
the ability to completely regenerate itself, based on the in-
teractions of its unique follicular epithelial and mesenchymal
components (34, 36, 37). Epithelial stem cells reside in the hair
follicle’s bulge region, which is localized below the sebaceous
gland (Fig. 1); they can repopulate both the hair follicle and
the interfollicular epidermis (60). Bulge cells are slowly cy-
cling and are quiescent until they receive a signal to leave
their niche and begin dividing and differentiating to support
a new anagen or to repopulate a skin defect (61, 62).
Each follicle develops from a single layer of ectoderm
into a complex miniorgan, constituting a dynamic struc-
ture that shares developmental pathways with many ec-
todermally derived endocrine organs like the pituitary,
adrenal gland, and pancreas (35, 37, 52, 61). Hair follicle
mesenchyme is placed in two intimately communicating
compartments that engage in intercompartmental fibro-
blast trafficking: the connective tissue sheath and the fol-
licular dermal papilla (63). The dermal papilla is separated
from the proximal follicle in telogen but is embraced by the
lower follicle matrix or bulb portion of the follicle during
anagen (34). Morphological changes of the dermal papilla
over the cycle primarily reflect changes in its extracellular
matrix: in anagen, it is rich in mucins; in catagen, the
glycosaminoglycan content is decreased; and in telogen,
its mucin content is scant (34, 64).
The hair follicle cycle is associated with a dramatically
altered cutaneous blood vessel supply. It has been shown
that in species with a synchronized hair cycle, anagen
development is accompanied by an increase in skin per-
fusion due to a rearrangement of the skin vasculature and
a genuine, substantial angiogenesis (65). Therefore, the
hair follicle represents an unusually attractive model for
studying how physiological angiogenesis is controlled by
a complex epidermal-mesenchymal interacting system in
vivo. We still do not know exactly what cellular or mo-
lecular mechanisms control these vascular changes. Be-
Ohnemus et al. Hair Follicle as an Estrogen Target and Source Endocrine Reviews, October 2006, 27(6):677–706 679
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 03 November 2015. at 05:08 For personal use only. No other uses without permission. . All rights reserved.
sides the two major recognized angiogenesis stimulators,
vascular endothelial growth factor (VEGF) and hepatocyte
growth factor (HGF) (65–67), processes of angiogenesis
can generally be modulated by hormonal changes, includ-
ing changes in estrogen levels (68). In fact, E2 reportedly
even stimulates human hair follicle synthesis of VEGF (69).
FIG. 1. Pigmentation of the hair shaft and differentiation of the root sheath. A, The hair follicle is a miniorgan composed of an epithelial
appendage of ectodermal origin embedded into the mesoderm-derived connective tissue and sc fat. B, Pluripotent keratinocytes in the hair matrix
niche constantly differentiate into the hair follicle shaft and root sheath lineages. Arrowheads depict their principal fates after they divide and
exit to participate in the modeling of the dynamic hair follicle components. In parallel, in an intriguingly orchestrated fashion, the melanocytes
residing in the matrix get in touch with the keratinocytes and transfer melanin granules. Interestingly, a chief part of the melanocytes is in
close contact with the dermal papilla, the prominent site of ER
expression.
680 Endocrine Reviews, October 2006, 27(6):677–706 Ohnemus et al. Hair Follicle as an Estrogen Target and Source
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 03 November 2015. at 05:08 For personal use only. No other uses without permission. . All rights reserved.
After a period of epithelial proliferation and differentia-
tion (anagen), follicle growth stops, and catagen begins (Fig.
3). The unknown signals that drive these changes are either
inherent in or delivered to the follicle. Catagen is a highly
controlled process of coordinated cell differentiation and
apoptosis, involving the cessation of cell growth and pig-
mentation, release of the papilla from the bulb, loss of the
layered differentiation of the lower follicle, substantial ex-
tracellular matrix remodeling, and vectorial shrinkage (dis-
tally) of the inferior follicle by the process of apoptosis (34,
70). Pigmentation is strictly coupled to anagen III-VI, and
many factors constituting the driving forces can also operate
as regulators of hair follicle cycling (71, 72). Follicular mel-
anin synthesis and pigment transfer to bulb keratinocytes are
modified by hormonal signals (72).
Although it is still unclear to what extent the immune
system contributes to the control of hair follicle cycling (34,
73), in rats and mice, the latter clearly is associated with
alterations in the skin immune status (74). Anagen hair bulbs
enjoy a relative immune privilege, and several forms of ab-
normal hair loss are associated with a prominent inflamma-
tory cellular infiltrate that attacks the hair follicle (74–76).
The transformation of terminal to vellus hair follicles in an-
drogenetic alopecia is also associated with a discrete infil-
tration of perifollicular macrophages and with mast cell ac-
tivation, which has been proposed to be inherent to the
terminal-to-vellus switch itself (77, 78). Also, mast cells and
macrophages likely play at least an important modulatory
(although nonessential) role in the control of hair follicle
cycle (73, 76; for review, see Ref. 34). Therefore, although this
remains entirely speculative, the well-recognized immuno-
modulatory properties of estrogen (7981) may indeed be
relevant to hair cycle control.
D. Estrogen-related hair research is clinically,
psychologically, and commercially highly relevant
Although hair growth disorders like hair loss and hirsut-
ism are often trivialized, they can profoundly affect a pa-
tient’s quality of life (82). This is evident in women with
androgenetic alopecia, who often report that the onset of hair
loss is associated with considerable anxiety and feelings of
FIG. 2. General conversion of inactive hor-
monal precursors into active sex steroids. Hu-
moral circulating precursors are converted by
cells in the periphery by distinct enzymes. The
aromatase enzyme is central for the produc-
tion of estrogens out of androgen precursors.
Androstenedione and testosterone are con-
verted into either E2 or estrone (E1). The ste-
roid sulfatase (data not shown) catalyzes the
formation of dehydroepiandrosterone (DHEA)
and E1 from their sulfated precursors, and
5
-reductase irreversibly converts testoster-
one into dihydrotestosterone. The steps from
cholesterol to androstenedione are depicted by
the precursors left to the curved arrow.
FIG. 3. The “biological clock” of the murine hair follicle
cycle. Many diseases can be understood as hair follicle cy-
cling disorders. Various cellular effectors are recognized as
hair follicle cycle modulators. The clock-like cartoon is an
idealized view of the transitional states of the hair follicle
growth phases. The hands of the clock are driven by an
unknown intrafollicular self-perpetuating event regulating
the duration of each individual hair cycle phase. Hair follicle
transitions from one phase to the next are manipulated by
an ever-increasing number of factors (34, 36), some of which
are shown here as examples: acceleration () of the specific
hair follicle cycle transition; or deceleration (). BMP-2,
Bone morphogenic protein-2; NTs, neurotrophins; NT-3,
neurotrophin-3; KGF, keratinocyte growth factor; CTSL,
cathepsin-L; GDNF, glial-derived neurotropic factor.
Ohnemus et al. Hair Follicle as an Estrogen Target and Source Endocrine Reviews, October 2006, 27(6):677–706 681
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 03 November 2015. at 05:08 For personal use only. No other uses without permission. . All rights reserved.
diminished attractiveness and helplessness, leading to social
withdrawal (82). A small minority of patients may even
display dysmorphophobia (83, 84). In addition, chemother-
apy-induced alopecia remains one of the most serious un-
solved problems in clinical oncology and has a psycholog-
ically disastrous impact on affected patients and their social
environment, for which a truly satisfactory remedy remains
to be developed in clinical practice (85, 86).
A multibillion dollar industry worldwide caters to the
unmet needs in managing unwanted hair loss (alopecia, ef-
fluvium) and unwanted hair growth (hirsutism, hypertri-
chosis) (87), advertising allegedly hair growth-stimulating
products or procedures such as vitamins, trace elements,
exotic herbs, amino acids, etc., which typically have not been
subjected to professionally designed and executed clinical
trials (88). Although topical formulations containing either
17
-or17
-estradiol have long been successfully employed
for the treatment of androgenetic alopecia, where they ap-
pear to improve the telogen/anagen ratio of scalp hair fol-
licles, this critique also applies here. In addition, phytoestro-
gen-containing preparations are increasingly and
aggressively advertised as hair growth-promoting agents,
despite the absence of sound clinical data to support such
claims.
In any case, however, there are solid clinical, psycholog-
ical, and economic reasons to dissect how, when, and why
estrogens modulate human hair follicle growth in defined
hair follicle populations and skin regions.
II. Hair Follicle Biology: Relevant Key Facts
A. Hair follicle morphogenesis and cycling
To understand the role of estrogens in hair biology, a few
principles need to be kept in mind. Hair shafts are produced
in the hair follicle, which is a specialized skin appendage in
which epithelial, mesenchymal, and neural crest-derived cell
populations collaborate in a stringently coordinated fashion
to generate a pigmented keratin fiber (Fig. 1) (34, 36, 37). The
hair follicle is the only organ that undergoes a lifelong cyclic
transformation, characterized by three distinct stages:
growth (anagen), regression (catagen), and resting (telogen).
Although only a minority of humans with hair growth dis-
orders have a disturbed hair shaft production, most cases of
hair loss seen in clinical practice result from alterations in
hair follicle cycling (87) (Fig. 3). Even the dramatic skin
appendage transformations that remodel a large terminal
hair follicle into a tiny vellus hair follicle are now recognized
to be hair cycle-dependent phenomena (35, 88, 89).
Shortly after completion of hair follicle morphogenesis, the
follicle enters into catagen. In humans, this entry occurs
already in utero; in mice, it happens about 17 d after birth (35,
37, 87). This is followed by a short phase of relative quies-
cence (telogen). Thus, morphologically, hair follicle cycling
begins with catagen, not with the actual growth phase, ana-
gen. The hair follicle’s transformations from telogen through
six stages of anagen and eight stages of catagen, followed
again by telogen, are genetically determined (37, 87). All hair
follicles manifest this cycle, although the duration of the cycle
as well as of the individual phases, and the length of the
individual shafts vary dramatically from site to site (34). In
the human and guinea pig, each follicle has its own inherent
rhythm, and thus the cycles are asynchronous (90), although
small groups of hair follicles on the human scalp are arranged
in so-called follicular units, which appear to link about three
terminal and/or vellus hair follicles into functional units (it
is unclear to what degree cycling within a human follicular
unit is synchronized) (91). In most rodents, large collections
of follicles cycle together, where synchronous follicle growth
occurs in large waves (34).
B. Molecular controls of hair follicle cycling
Although the ultimate oscillator system (“hair cycle
clock”) that drives hair follicle cycling remains unknown, an
ever-increasing list of molecules is now recognized to mod-
ulate normal hair follicle cycling (34–36) (Fig. 3). For exam-
ple, the duration of anagen is prolonged by IGF-I, HGF,
glial-derived neurotropic factor, and VEGF, whereas anagen
is shortened and catagen is induced by fibroblast growth
factor 5 (FGF5), TGF
1 and TGF
2, IL-1
, and interferon-
(IFN-
) (34, 35, 65, 66, 92). One critical question in the context
of the current review, therefore, is to what extent these key
hair cycle modulators are regulated by ER-mediated
signaling.
The expression of these key regulators of hair follicle cy-
cling is under the control of a number of often still undefined
upstream signals, which differ between species and hair
follicle subpopulations. Key examples of these upstream sig-
nals are nuclear factor-
B, members of the Wnt and TGF
/
bone morphogenetic protein (BMP) families as well as their
functional antagonists, Shh, and
-catenin (34–36). Interest-
ingly, many of the same signals that drive hair follicle in-
duction and morphogenesis are reused during anagen de-
velopment (35, 45, 93). Hair follicle pigmentation and active
melanogenesis in the follicle pigmentary unit are strictly
coupled to anagen III-VI (94). Besides locally generated
-melanocyte-stimulating hormone and/or ACTH, con-
trolled changes in the intrafollicular expression of stem cell
factor, nerve growth factor (NGF), and/or HGF are probably
the inducers of melanocyte activity in the hair follicle pig-
mentary unit. Some of these pigmentation-regulatory factors
are also regulators of hair follicle cycling (95–97). Again, the
question arises, whether and how estrogens modulate the
intrafollicular expression of these agents.
In catagen, which is a stringently controlled, apoptosis-
and terminal differentiation-driven process of rapid organ
involution (98), there are two protagonists that regulate nor-
mal apoptosis in the hair follicle: p53 (99, 100), and the prod-
uct of the hairless gene (Hr), a zinc finger transcription factor
(49, 101, 102). Intriguing similarities in the phenotype of
hr-defective hairless mice and of mice with loss-of-function
mutations in the vitamin D receptor (VDR) or retinoid X-
receptor (RXR-
) suggest that Hr, VDR, and RXR-
are all
parts of similar pathways that are critical for activation of the
(as yet undefined) key genes that control the anagen-catagen
transformation (35, 103). FGF5, TGF
1, TGF
2, the neuro-
trophins NT-3, NT-4, and brain-derived neurotrophic factor,
as well as p75NTR signaling, IFN-
, prolactin, and estrogen,
are recognized inducers of catagen (8, 28, 52, 104–112).
682 Endocrine Reviews, October 2006, 27(6):677–706 Ohnemus et al. Hair Follicle as an Estrogen Target and Source
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 03 November 2015. at 05:08 For personal use only. No other uses without permission. . All rights reserved.
During the resting stage (i.e., telogen, a period of relative
biochemical and proliferative quiescence), ER
is maximally
expressed, and E2 serves as a kind of “hair cycle brake” (26,
27). Dramatic shortening of telogen is seen when the hair
follicle chooses the so-called “dystrophic catagen” damage
response and recovery pathway after chemotherapy induced
hair follicle dystrophy and alopecia (113, 114). The damage
response pathways differ in the speed and outcome of hair
follicle recovery and can be manipulated pharmacologically
by application of PTH/PTHrP receptor ligands and cyclo-
sporine A as well as by steroid hormones like dexametha-
sone, calcitriols, or estrogens (27, 113, 115, 116). Thus, not
only hair follicle cycling is profoundly influenced by estro-
gens, but also hair follicle recovery from chemical damage.
A selection of factors currently recognized as prominent
regulators of hair follicle cycling in men or mice is shown in
Fig. 3. In the current context, the key concept is that, in
principle, estrogens may have both direct hair growth-mod-
ulatory effects that target the elusive hair cycle clock and
indirect ones by altering the expression of important hair
growth modulatory factors such as those indicated in Fig. 3
(117, 118). It is on this hair biology background that we are
exploring estrogen functions in the following sections.
III. Cellular and Molecular Mechanisms of
Estrogen Action
A. Estrogen synthesis and metabolism
It helps to recall a few basic facts in estrogen biology to
understand the mechanisms by which estrogens may affect
hair follicle growth and cycling. E2 de novo synthesis starts
from cholesterol precursors. The final step essentially re-
quires androgens as substrates. For the conversion of tes-
tosterone to E2, testosterone is converted to 19-hydroxytes-
tosterone by a monooxygenase (EC 1.14.13.), then to 19-
oxotestosterone, which is then converted to E2 by an
oxidoreductase (EC 1.14.99.).
An alternative route is via 4-androstene-3,17-dione, which
is converted to estrone by a monooxygenase (EC 1.14.13.),
and then by an oxidoreductase to E2 (EC 1.14.99.). Estrone
can be metabolized to E2 by 3
(or 17
)-hydroxysteroid
dehydrogenase (Ref. 119) (EC 1.1.1.51) or estradiol 17
-
dehydrogenase (Ref. 120) (EC 1.1.1.62). The only known
pathway connecting testosterone to E2 is the cytochrome
P-450 enzyme aromatase (EC 1.14.14.1, CYP19A1; ARO)
pathway (Fig. 2). The CYP19 gene is localized on chromo-
some 15. It spans nine coding exons and a few untranslated
exons, upstream of exon II, namely exon I1–I5.
The fact that CYP19A1 (also called aromatase or estrogen
synthetase; ARO) transcripts with specific 5-ends were iso-
lated from various tissues (with all transcripts belonging to
the exon I variant) led to the finding of tissue-specific pro-
moter regulation. These are under an intricate control of
transcription factors in response to gonadotropins, IL-6, IL-
11, and TNF-
. Because exon I is not translated, all proteins
are therefore identical. The flexibility of the system is exem-
plified by the differential regulation displayed by the adi-
pocyte ARO promoter, compared with its bone counterpart
(121). Intriguingly, ARO activity was also found in human
hair follicles (122, 123), and ARO transcripts have been de-
tected in cultured hair follicle fibroblasts and keratinocytes
(69). Paracrine estrogen secretion by hair follicle cells with
ARO activity may be important for hair growth control,
perhaps in a manner that is comparable to the paracrine
activation of ERs by E2 during mammary gland duct mor-
phogenesis (124).
The total E2 production rate of human males has been
calculated to range from 35 to 45
g (0.01300.0165
mol)
per day, of which 15–20% supposedly originates from the
testes (125–127). About 60% of circulating E2 is thought to
arise from peripheral aromatization of testosterone, whereas
20% is formed by reduction of estrone (125, 126). Estrone is
formed by peripheral aromatization of androstenedione,
which partly derives from the adrenal glands and partly from
peripheral conversion of testosterone. Estrone can also be
directly produced and released by the adrenals (126). In
general, the testicular glands control circulating estrogen
levels, as evident from their rapid decline after orchiectomy
(128, 129). In premenopausal women, the main biosynthesis
of estrogen takes place in the corpus luteum. Small amounts
are also produced by the adrenals. During pregnancy, sub-
stantial amounts of estrogens are produced in the placenta as
well. Postmenopausal decline of ovarian production of es-
trogen is partially compensated for by nonovarian conver-
sion of androstenedione in the adrenals, liver, adipose tissue,
skeletal muscle, kidney, and brain (130). Figure 2 depicts
basic biosynthesis and metabolism of estrogens.
In the current context, it should be kept in mind, however,
that the pilosebaceous unit itself, which displays very sub-
stantial aromatase activity, especially in its sebaceous com-
partment (59), is a significant source of estrogen synthesis,
both in men and women. However, it is still far from clear
which percentage of circulating estrogens is provided by
peripheral estrogen synthesis in human skin under physio-
logical and pathological conditions, and how much of this
intracutaneous estrogen synthesis arises from the piloseba-
ceous unit.
B. Estrogen receptors
The effect of estrogens on their target tissues is determined
by: 1) the receptor subtype(s) expressed and their posttrans-
lational status; 2) the balance between corepressors and co-
activators present; 3) the conformational transformation of
the receptor after binding of the ligand; and 4) the interaction
of the final receptor-multiprotein complex with the promoter
of the target genes.
Two distinct isoforms of the ER exist: ER
and ER
(131)
(Fig. 4). Phylogenetic analysis and theoretical reasoning sug-
gest that both isoforms diverged from a putative ancestor
protein. The long time of parallel and divergent evolution
could explain the distinct biological roles of these receptors,
although they still possess large sequence homology (132).
As regards molecular action, they show significant differ-
ences (133–135). A structural overview of human ER
and
ER
is shown in Fig. 4. Interestingly, a constitutively active
ER ortholog without sensitivity to estradiol or related mol-
ecules is expressed in the mollusc Aplysia (136) and may
Ohnemus et al. Hair Follicle as an Estrogen Target and Source Endocrine Reviews, October 2006, 27(6):677–706 683
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 03 November 2015. at 05:08 For personal use only. No other uses without permission. . All rights reserved.
correspond to an early ancestral protein from which ER
,
ER
, and other steroid receptors have evolved.
The human ER
gene spans a length of more than 140 kb.
ER
, like ER
, acts through direct intranuclear binding to
DNA after activation by a ligand. ER
has eight exons. The
ER
protein sequence varies 77–97% between rat, human,
and chicken (137). A detailed overview of the ER
gene
structure in combination with a proposal for a consistent
nomenclature has been published (138). The putative role of
tissue-specific ER
gene promoter regulation in developing
and adult tissue is also discussed there. ER
, like ER
, has
eight exons, whereas the translated protein is shorter. For an
overview of the organization of the ER
gene, see Ref. 139.
Alternative splicing accounts for various subforms of ER
and ER
receptors (140). ER
splicing variants differ in their
untranslated 5-ends. Recently, a 46-kDa variant of ER
was
cloned from endothelial cells (141). It is missing 173 amino
acids in its N-terminal end, whereas the protein still has its
DNA- and hormone-binding domains intact, including a
functional activation function (AF)-2 domain (142). Five iso-
forms of human ER
mRNA were isolated from various
tissues, varying in their C-terminal ends and tissue-depen-
dent expression (143). An 18-amino acid residue ligand bind-
ing domain insertion variant of ER
in rodents acts as a
dominant negative repressor of ER
(144). The ER proteins
are subject to ubiquitinylation and proteosomal degradation
(145). ER splice variants may act as regulators of ER
and/or
ER
(146). These variants that lack sequences coding for
nuclear translocation/nuclear localization (contained within
domain C encoded by exons 2 and 3 and part of domain D
encoded by exons 3 and 4) and/or sequences coding for DNA
binding domain (contained within domain C encoded by
exons 2 and 3) exist in numerous tissues (147, 148).
New estrogen-sensitive entities, e.g., at the plasma mem-
brane or in the endoplasmic reticulum, have been reported,
although the biological significance of these binding sites for
E2 is still under critical discussion (149–154). Stimulation by
estrogens evoke rapid cellular effects that peak minutes after
stimulation, even in multiple cell types and after inhibition
of RNA synthesis, indicating nongenomic mechanisms (155).
Signaling cascades that might be involved include second
messengers such as calcium and nitric oxide, receptor ty-
rosine kinase signaling involving epidermal growth factor
(EGF) receptor (EGFR) and IGF-I receptor, G protein-coupled
receptors, phosphoinositide-3 kinase, serine-threonine ki-
nase, MAPK, nonreceptor kinase steroid receptor coactivator
(SRC), and protein kinases A, B (Akt), and C (156–159).
It should also be mentioned that many studies have sug-
gested that this nongenomic effect is important in nonrepro-
ductive tissues such as brain, bone, and the cardiovascular
system (160). Interestingly, it has been demonstrated that the
effects of sex steroids on prevention of osteoblast apoptosis,
which allegedly are mediated by nongenomic actions in-
volving the MAPK-signaling pathway, appear to be gender
nonspecific. These effects are supposedly mediated by the
ligand (rather than DNA) binding domain of ER
,ER
,or
androgen receptor, and can be transmitted with similar ef-
ficiency irrespective of whether the ligand is an estrogen or
an androgen (161). The role of such gender nonspecific, non-
genomic effects in hair follicle biology is as yet unknown.
C. Nuclear receptor superfamily
ERs are members of the nuclear hormone receptor super-
family (131). Nuclear hormone receptors exhibit a sequential
organization into consecutive domains enumerated A to F.
These have highly specific functions: domains A and B are
required for ligand-independent transactivation; domain C,
with its two zinc-fingers, for DNA-binding; and domain E for
ligand-dependent transactivation, dimerization, and inter-
action with other proteins (162). The sequence of the ER gene
is conserved in all species studied except fish (163). In fish,
only the C and E domains have high homology to ERs of
other species. The DNA-binding domain has the highest
homology between species. The function of the F domain of
ER
and ER
is not fully understood.
In addition to ER
and ER
, the following nuclear recep-
tors are derived from a putative common ancestor protein:
the progesterone receptor, androgen receptor, glucocorticoid
receptor, and mineralocorticoid receptor. Structurally, the
thyroid and retinoid receptors also belong to this receptor
gene superfamily (164). In addition, so-called orphan recep-
tors have been identified, i.e., members of the nuclear recep-
tor gene superfamily that still lack assigned ligands seem to
have important functions (165, 166). Although evolutionary
analysis has invited the speculation that they may be “mo-
lecular fossils” of the prototypical transcription factors with-
out ligand-activating function (167), we may as well just have
failed to identify the appropriate ligands, as exemplified by
selected retinoid orphan receptors. For example, retinoid
orphan receptor-
has been shown to operate as a mediator
of nuclear melatonin signaling (168–170), is expressed at the
gene and protein levels in murine hair follicles, and displays
significant hair cycle-dependent expression changes (53).
FIG. 4. Domains and structure similarity of human ER
and ER
.
The amino terminal A/B domain harbors the AF-1, which enables the
receptor to interact with members of the transcriptional machinery.
The C domain contains two zinc-fingers, important for DNA-binding
and receptor dimerization. The “hinge region” or D domain, gives the
receptor some degree of flexibility between the DNA and the E do-
main. HSP90 also binds to this region, and the nuclear localization
signal is supposed to be here. The carboxy-terminal E/F domain binds
the receptor-specific ligands, is required for nuclear translocation and
receptor dimerization, and modulates the target gene expression with
coregulators. The functional regions have varying degrees of homol-
ogy between the ER
and ER
isoforms, as depicted in the boxes of
the ER
molecule. The functional sites are placed below the domain
boxes.
684 Endocrine Reviews, October 2006, 27(6):677–706 Ohnemus et al. Hair Follicle as an Estrogen Target and Source
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 03 November 2015. at 05:08 For personal use only. No other uses without permission. . All rights reserved.
Thus, it remains a particularly intriguing challenge to ex-
plore the functional importance of the three orphan receptors
(ERR1, ERR2, and ERR3) related to ERs that have been iden-
tified (165, 166, 171, 172). Murine ERR1 is found in adipo-
cytes, muscle, brain, and testis as well as in skin. It is highly
expressed in the ossification zone of mice. Compared with
ER
, it shows a relatively higher level of expression in os-
teoblast-like cells. ERR2 reportedly is restricted to embryo-
logical stages and only very few adult tissues (171).
D. Estrogen receptor signaling pathways
The basis of differential expression of target genes is bind-
ing of transcription factors like nuclear receptors to specific
DNA sequences residing within regulatory promoters (173).
Alternative ways of transcriptional activation by ERs are
shown in Fig. 5. Tissue-specific coregulators are believed to
be important factors in tissue specific effects of nuclear re-
ceptor ligands (174), which may explain, e.g., why breast
cancer cells are inhibited by tamoxifen, whereas this ligand
is growth promoting in the uterus (175). These cofactor pro-
teins are part of a transient ER-multiprotein complex (176),
of which several have been identified (174, 176).
An ER
-associated protein, template-activating factor I
,
regulates transcription of estrogen-responsive genes by
modulating acetylation of ER
, and may also interact with
other nuclear receptors (177). Classical receptor activation is
understood as the ligand binding to the receptor binding
pocket, which causes conformational changes from an inac-
tive into an active receptor state. This facilitates binding of
the receptor to the DNA response motif, in the typical case
a 15-bp palindromic sequence ERE, as a dimer. Alternative
response elements and their proteins may also be associated
with liganded receptor, including the GC-box binding pro-
tein (SP-1), nuclear factor
B (178), or the bipartite c-Jun,
c-Fos complex [activating protein-1 (AP-1)]. It is known that
ER
and ER
regulate some gene promoters with AP-1 sites
in an opposite manner (179). Interestingly, ER
exerts a neg-
ative transcriptional regulation at AP-1 sites when com-
plexed with its natural ligand E2, whereas, in this context,
antiestrogens positively activate gene transcription (133).
In addition to ligand activation, ERs can be regulated by
phosphorylation through polypeptide growth factors such as
EGF and IGF-I (180, 181). For example, murine uterus re-
sponded to cotreatment with anti-EGF antibodies with at-
tenuated E2 response. Furthermore, administration of the ER
antagonist ICI164-384 reduced the uterine response to EGF
(182).
An example of the significance of activation of cytosolic
signal transduction proteins is the role of ERK activation in
regulating osteoblast survival and bone formation (161), or
the role of ERK and phosphoinositide-3 kinase activation on
nitric oxide production in endothelial cells and angiogenesis
(183). Growth factors such as IGF-I, EGF, and TGF
, through
activation of MAPK pathway, regulate phosphorylation of
ER influencing its transcriptional activity. Also, in the ab-
sence of estrogen, ER can be activated by these growth factors
(184). Moreover, in different cell types, estrogen regulates the
expression of EGF, IGF-I, and TGF
, suggesting that these
growth factors are mediators of estrogen action (181). Thus,
given the well-appreciated central role of IGF-I, EGF, and
TGF
in hair follicle biology, the cross talk between peptide
growth factors and ER signaling pathways may be highly
relevant in hair growth control.
E. Estrogen-responsive genes and coregulators of
ER signaling
ERs interact with numerous coregulator proteins, result-
ing in either enhanced or repressed gene expression. Exam-
ples are the coactivator actions on the ligand-binding AF-2
domain. Crystallographic analysis of the ER
ligand-binding
domain has indicated that, upon binding of an agonist, four
of 12
-helices in this receptor domain are rearranged to form
a hydrophobic cleft with docking sites for coactivators im-
portant for AF-2 function (185–188).
According to one hypothetical model of the exchange of
coregulators involved in regulation of genes by ERs (189), in
the unliganded state, ER
may bind to either corepressor or
coactivator complexes. Intracellular signaling (e.g., ligand-
induced receptor activation, posttranslational receptor mod-
ification and activation) may shift this dynamic equilibrium
to favor coactivator complex interaction. When ER is ligand-
stimulated, a series of coregulator complexes bind and ex-
change in a programmed manner until the gene is activated.
FIG. 5. Alternative ways of transcriptional activation of ERs. ER
activation of target genes can take place through distinct intracellular
pathways. Classical receptor activation is believed to happen through
ligand diffusion through cell cytoplasm into the nucleus, where the
ligand induces conformational change of the cognate ER protein to
exhibit a conformation that has high affinity to the ERE residing in
promoters of various cellular target genes. Alternatively, estrogen
phosphorylation events can activate the receptor without ligand after
activation by cellular kinases through growth factor receptor at the
cell surface. Besides the typical ERE sequence, alternative binding at,
e.g., AP-1 sites can take place as the ER complexes with other proteins
to initiate transcription.
Ohnemus et al. Hair Follicle as an Estrogen Target and Source Endocrine Reviews, October 2006, 27(6):677–706 685
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 03 November 2015. at 05:08 For personal use only. No other uses without permission. . All rights reserved.
This involves histone modification, like acetylation, which is
carried out by CBP/p300 and SRCs, followed by formation
of a complex containing BRG-1/BAF57, which unwinds
DNA and changes chromatin, enhancing formation of com-
plexes functioning to activate transcription. TRAP/DRIP
multiprotein complex interacts with RNA polymerase II,
driving transcription. After the process is complete, the pro-
teins involved are ubiquitin-tagged and turned over by 26S
proteosomal degradation.
One particularly interesting corepressor involved in tis-
sue-selective effects of estrogens is the repressor of estrogen
action (REA) (190, 191). It is a 37-kDa ER-selective coregu-
lator, which directly competes with SRC1. It acts on ligand
bound ER and modulates its sensitivity to agonistic as well
as antagonistic ligands (190).
F. Estrogen target tissues
ER
and ER
are found in many organs and cell types.
Besides the skin and hair follicle (Table 2), ERs have been
detected in a wide variety of tissues and cells, such as mam-
mary gland, prostate, testis, placenta, brain pituitary, carti-
lage, adipocytes, osteoblasts, skin, keratinocytes, and fibro-
blasts (191–200). ERs are also found in the classical steroid
hormone-susceptible and -producing organs, e.g., ovarian
granulosa cells and testis (201). During a study of stromal and
epithelial tumors of the ovary, ER
and ER
isoforms were
detected by RT-PCR and Southern blot analysis (202). A wide
expression in malignant and normal ovary has been de-
scribed. ER
was often found, with low intensity but with
high incidence in granulosa cell tumors. A C-terminally
shortened variant of ER
,ER
cx, was widely expressed in all
tumors studied. This variant is a ligand-independent ER
isoform with antagonistic activity to ER
(203).
Variants of ERs are normally found to be coexpressed with
the wild-type receptor. For example, ER
E7 is a natural
splice variant with exon 7 absent. It is insensitive to allosteric
modulation through ligands and coregulators such as p160,
SRC1, and AIB1. ER
E7 heterodimerizes with ER
or ER
and acts in a dominant-negative manner (204). Studies fo-
cusing on the exclusive distribution of these ER variants have
revealed, e.g., the presence of ER
46 in the plasma mem-
brane, cytoplasm as well as the nucleus of estrogen-deprived
endothelial cells (142). mRNA of ER
wild type, and splice
variants (5) were, for example, studied in normal human
breast tissues of 37 women. Sixty-two percent showed co-
expression of ER
wild type and ER
5 variants, whereas
in around 30% of the specimens, exclusively ER
wild type
was detectable. This study suggests an important function of
ER
in normal female breast biology (205). Finally, various
parts of the brain also exhibit ER. Here, ER may be involved
in the feedback or general regulation of various humoral
factors, including estrogens themselves, as well as in other
complex higher functions such as cognition, memory, and
motor control (206).
G. Interdependence of estrogen and androgen
signaling pathways
Estrogens have diverse effects on many tissues in both
males and females, and the majority of these effects are
mediated by both subtypes of ERs: ER
and ER
. However,
there are striking differences with respect to the tissue dis-
tribution of the ER subtypes. For example, ER
appears to
play a major role in mediating estrogen action in the pituitary
and the uterus, whereas a clear role of ER
has been estab-
lished in the ovary, lung, prostate, immune system, and brain
(207).
Given the well-appreciated importance of both andro-
gens and estrogens in hair follicle biology, the interde-
pendence of estrogen and androgen signaling in hair
growth control deserves careful scrutiny. In recent years,
it has become quite clear that the conventional concept of
androgens as male hormones and estrogens as female hor-
mones is an oversimplification. We now know, for exam-
ple, that estrogens play a central role in the skeleton, the
cardiovascular system, and the reproductive tract of males
(208, 209), whereas androgens are important for repro-
ductive function of females (210). The effects of sex ste-
roids on prevention of osteoblast apoptosis appear to be
gender independent (161). As mentioned above, they are
mediated by the ligand (rather than DNA) binding domain
of ER
,ER
, or the androgen receptor. In addition, there
are numerous reports demonstrating that estrogen and
androgen metabolites can interact with both receptor sub-
types (211, 212). Further evidence for this molecular cross
talk comes from studies in mouse prostate, demonstrating
that the testosterone metabolite, 5
-androstane-3
,17
-
diol, binds and activates ER
, thereby reducing androgen
receptor levels (213). Thus, because both androgen recep-
tor and ER
are prominently expressed in the hair follicle
[i.e., in follicular dermal papilla cells (24)], systematic stud-
ies are required to determine whether androgen metabo-
lites, acting via ER
, affect the hair follicle’s mesenchymal
“command center,” i.e., follicular dermal papilla cells [sim-
ilar to what has been reported in mouse prostate (213)].
IV. Estrogens in Dermatoendocrinology
A. Effects of estrogens on the skin
Besides its protective and regulatory functions, the skin is
an important endocrine organ (59, 72, 97, 214). Of all hor-
mones that decline with age, estrogens apparently have the
most dramatic effect on the skin (215), and this occurs in more
than one way. E2 increases in vivo the collagen I and III
content and quality, maintains skin moisture by increasing
acid mucopolysaccharides, glucosaminoglycans, and hyal-
uronic acid and possibly maintains stratum corneum barrier
function (25, 215–217). Estrogen, together with progesterone,
prevents or repairs skin atrophy, wrinkles, and dryness as-
sociated with chronological or photoaging in postmeno-
pausal women by an increase in the number and an im-
provement of the orientation of elastic fibers in the dermis
(81, 215, 216, 218). E2 increases skin thickness, whereas vas-
cularization is enhanced (215).
In a murine model, topical application of estradiol signif-
icantly increased skin thickness, hyaluronic acid synthase
levels, hyaluronic acid content, tissue transglutaminase, and
collagen type I in SKH-1 hairless mouse skin, suggesting a
therapeutic role for ER agonists in wrinkle repair (219). In rat
686 Endocrine Reviews, October 2006, 27(6):677–706 Ohnemus et al. Hair Follicle as an Estrogen Target and Source
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 03 November 2015. at 05:08 For personal use only. No other uses without permission. . All rights reserved.
skin, E2 has effects on mitosis and on differentiation of epi-
dermis and sebaceous glands (5). E2 also stimulates the syn-
thesis, maturation, and turnover of collagen in rat (220) and
guinea pig skin (221). Furthermore, estrogens have been
shown to increase mitotic activity in the epidermis of mice
and induce epidermal thickening (222, 223). However, pro-
longed E2 administration reportedly reduces epidermal
thickness in rats (3).
TABLE 2. Reports of 17ß-HSD isoforms and ER expression in the skin and hair follicle
Species Location Detection method Ref.
Mouse 17
-Hydroxysteroid dehydrogenase (17
-HSD) type 1 mRNA;
specific hybridization signal was seen in the sebaceous
glands, whereas the epidermis, stroma, hair follicles, and
sweat glands were unlabeled.
In situ hybridization using a
35
S-labeled
cRNA probe
364
Human 17
-HSD was mainly demonstrated in the pilosebaceous unit
and epidermal keratinocytes. In hair follicles, 17
-HSD was
histochemically localized to outer root sheath cells (ORS);
freshly plucked anagen hairs mainly containing keratinocytes
from the inner root sheath and ORS expressed very high
levels of 17
-HSD2 and moderate levels of 17
-HSD1.
257, 365–367
Mouse Type 8 17
-HSD; in the dorsal skin, specific labeling was
detected in the stroma.
Hybridization with the radio-labeled
sense probe
367
Human Sebocytes and keratinocytes; sebocytes but not keratinocytes
expressing type 3 17
-HSD.
RT-PCR 366
Human ER
was poorly expressed, being restricted to sebocytes. In
contrast, ER
was highly expressed in the epidermis,
sebaceous glands (basal cells and sebocytes), and exocrine
sweat glands. In the hair follicle, ER
is widely expressed
with strong nuclear staining in dermal papilla cells, inner
sheath cells, matrix cells, and outer sheath cells including
the bulge region.
257
Human ER
was the major steroid receptor expressed in human skin.
It was highly expressed in epidermis, blood vessels, and
dermal fibroblasts, in contrast to ER
. In the hair follicle,
ER
expression was localized to nuclei of outer root sheath,
epithelial matrix, and dermal papilla cells, in contrast to
ER
, which was most prominently expressed in dermal
papilla cells. Serial sections also showed strong nuclear
expression of ER
in the cells of the bulge, whereas ER
was
not expressed. In the sebaceous gland, ER
was expressed in
both basal and partially differentiated sebocytes. ER
exhibited a similar pattern of expression.
Immunohistochemistry 19, 24
Mouse ER
-specific staining of the nuclei of the cells of the dermal
papilla. The expression of the ER in the dermal papilla was
hair cycle-dependent, with the highest levels of expression
associated with the telogen follicle; ER was hair cycle
dependent because there was weaker staining of the dermal
papilla of early anagen follicles and no detectable staining in
dermal papilla of mid- to late-anagen or catagen follicles.
Very light ER staining was observed in the cells of the outer
root sheath in the isthmus of the telogen follicle as well as in
some nuclei of dermal fibroblasts.
Immunohistochemistry 26
Mouse Hair cycle-dependent ER expression: ER
IR had its peak in
telogen follicles within the dermal papilla and the sebaceous
gland, whereas the outer root sheath, epithelial strand, germ
capsule, and bulge region showed weaker IR. In anagen VI,
ER
was seen in the dermal papilla (DP), matrix
keratinocytes, inner and outer root sheath, whereas the
epidermis, sebaceous gland, and the bulge region showed no
ER
IR. In early catagen, ER
IR was seen in the dermal
papilla, matrix keratinocytes, inner root sheath, outer root
sheath, and connective tissue sheath. In late catagen, ER
IR
was restricted to the dermal papilla, germ capsule, outer root
sheath, bulge and sebaceous gland. ER
503 IR was intense
and ubiquitously seen in the dermal papilla, inner and outer
root sheath, matrix keratinocytes, sebaceous gland, bulge
region, and epidermis throughout all investigated stages; the
connective tissue sheath was positive for ER
in anagen VI.
ER
ins IR was weaker but coexpressed throughout the
whole cycle and in the same locations as ER
503.
Immunohistochemistry 28
Expression of ERs and important steroid hormone-metabolizing enzymes within the hair follicle compartments was found in various systems.
Interestingly, ER was found in the most central subcompartments of the hair follicle, the dermal papilla and the bulge. The current data also
suggest a mixed mesenchymal-ectodermal expression.
Ohnemus et al. Hair Follicle as an Estrogen Target and Source Endocrine Reviews, October 2006, 27(6):677–706 687
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 03 November 2015. at 05:08 For personal use only. No other uses without permission. . All rights reserved.
Skin color varies with the menstrual cycle (224, 225). Such
variations may result from a synergistic action of estrogens
and progesterone on the melanogenic activity of epidermal
melanocytes. Similar mechanisms account for the hyperpig-
mentation during pregnancy, which is most prominent in
mammilar skin (224, 225). Estrogens are capable of acceler-
ating the synthesis of melanin; the mechanism is supposed
to be a direct effect of the hormone itself, because the re-
sponse occurs locally when the hormone is applied directly
to the skin (226). Normal human melanocytes become en-
larged and dentritic in culture after 2-d incubation with es-
tradiol (227). The effects of estrogens on melanocyte func-
tions, however, are as yet unclear, because estrogens can
increase epidermal melanocyte cell number, while decreas-
ing melanin content and tyrosinase activity (228). In contrast,
another group observed that estrogens significantly increase
melanin synthesis and tyrosinase activity in normal human
skin melanocytes in vitro (229). Therefore, the actual effects
of estrogens on melanocyte functions may be highly context-
dependent, and may reflect the local predominance of dis-
tinct ER coregulatory elements.
Estrogens have also been suggested to be major regulators
of wound repair, which may reverse age-related impaired
wound healing in human and animal models (216, 230, 231).
Estrogens dampen inflammation (as indicated by a suppres-
sion of the production of proinflammatory cytokines, mac-
rophage migration inhibitory factor, and TNF-
by macro-
phages), and enhance the deposition of collagen I in the
dermis, thus increasing the breaking strength of wounds in
ovariectomized mice (232). The ER complex is able to stim-
ulate the expression of growth factors, such as IGF-I, a mi-
tosis-enhancing protein for keratinocytes (233, 234), which
may also serve as a “guardian of immune privilege” (76).
In vitro, E2 stimulates human keratinocyte proliferation by
promoting the expression of cyclin D2, which induces G
1
to
S phase progression in the cell cycle (235); in addition, it
inhibits oxidative stress-induced apoptosis in keratinocytes
by promoting Bcl-2 expression (236). E2 in vitro enhances
production of NGF, a growth factor, e.g., for neurons and
keratinocytes, in macrophage-like differentiated THP-1 cells
(237) and also induces c-Fos expression in macrophages via
the GPR30/cAMP/protein kinase A signaling pathway; E2
activates NGF transcription from AP-1 elements. Further-
more, in vitro, E2 induces keratinocytes to produce an au-
tocrine growth factor, granulocyte-macrophage colony stim-
ulating factor by increasing both its transcription and mRNA
stability (79). These effects of E2 may combine to enhance
wound reinnervation and reepithelialization (81).
Other authors have suggested that estradiol effects on
keratinocytes are mediated via a membrane ER
that acti-
vates the MAPK pathway. It has been shown that E2 induces
the proliferation of human keratinocytes and stimulates
MAPK activation as well as cyclin D1 expression (238). De-
spite the obvious effect of estrogen on dermal collagen con-
tent (220–222), the underlying molecular mechanisms are
still poorly understood. However, it has been demonstrated
that both ER
and ER
are expressed in human dermal
fibroblasts, which is in line with the concept that E2 effects
in the dermis may occur, at least in part, through direct
regulation of ER-mediated fibroblast functions (239).
Reportedly, the course of a number of chronic inflamma-
tory skin diseases is modulated by estrogens: the inflamma-
tory infiltration in psoriatic lesions may be suppressed by
estrogens (81), whereas atopic dermatitis worsens during
pregnancy (240). CD-1a-positive Langerhans cells are acti-
vated in the skin lesions of estrogen dermatitis (241). The
proliferation of skin hemangioma vascular endothelial cells
in culture is stimulated by E2 (242). Because VEGF transcrip-
tion from AP-2 elements can be enhanced by E2 in vitro via
the G protein-coupled receptor 30/cAMP/protein kinase A
signaling pathway, E2 may promote the development of
granuloma pyogenicum (243).
V. Estrogens in Pilosebaceous Unit Biology
A. Estrogen synthesis and metabolism in the
pilosebaceous unit
Estrogens and androgens are closely related sex steroids
with interconnected metabolism. Their role in skin physiol-
ogy was perceived early, whereas the finding that isolated
human hair follicles have their own repertoire of sex steroid-
metabolizing and synthesizing enzymes (122, 244) represents
a much younger insight. For example, isolated human hair
follicles harbor the enzymes 5
-reductase, aromatase, and
17
-hydroxysteroid dehydrogenase to control estrogen syn-
thesis in the pilosebaceous unit. Aromatase converts the sub-
strates androstenedione to estrone and testosterone to estra-
diol (Fig. 2). Interestingly, both dermal papilla cells and outer
root sheath keratinocytes reportedly synthesize cytochrome-
P450-aromatase (25). Estrone can be converted back and forth
to estradiol by 17
-hydroxysteroid dehydrogenase (for re-
view, see Refs. 23, 25, 59, and 245).
This has initiated a shift of paradigm—from the pilose-
baceous unit as a mere target organ of steroid hormone
activities to the concept that the pilosebaceous unit is an
important site of steroid hormone synthesis and metabolism
(5). Today, the skin is understood to have established its own
para- and autocrine hormonal regulation networks, with the
pilosebaceous unit located at center stage; it is now recog-
nized to be highly sensitive to an ever-expanding list of
hormonal regulators that are generated and/or metabolized
within or in close vicinity to the pilosebaceous unit (8, 9, 52,
53).
B. Estrogen receptor expression in the hair follicle
Until the cloning of a novel gene coding for a second ER,
named ER
, from rat prostate (246) and, thereafter, from
human tissue (247), the consensus was that only one ER
existed: ER
, cloned in 1986 from MCF-7 cells (178, 248). Both
receptors bind E2 with high affinity (228) and bind to classic
EREs in a similar manner (249). Both receptors are detectable
in the skin of humans and rodents with distinct expression
patterns (14–19, 25, 26, 28, 250–253) (Table 2).
Recently, Thornton et al. (20, 24) showed that, in human
scalp skin, ER
is the predominant ER. In human hair fol-
licles dissected from male and female nonbalding scalp skin,
ER
expression was found to be localized to nuclei of outer
root sheath and epithelial matrix keratinocytes as well as of
688 Endocrine Reviews, October 2006, 27(6):677–706 Ohnemus et al. Hair Follicle as an Estrogen Target and Source
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 03 November 2015. at 05:08 For personal use only. No other uses without permission. . All rights reserved.
dermal papilla fibroblasts. In contrast, ER
and the androgen
receptor were only expressed in dermal papilla cells. Serial
sections also showed strong nuclear expression of ER
in the
cells of the bulge, whereas neither ER
nor androgen recep-
tor was detectable. In the sebaceous gland, ER
was ex-
pressed in both basal and partially differentiated sebocytes.
ER
exhibited a similar pattern of expression, whereas the
androgen receptor was expressed in the basal and very early
differentiated sebocytes (19, 24). In this study, there were no
obvious differences in the expression of either ER in male or
female skin. The same group found that in cultured human,
nonbalding scalp dermal papilla cells, the two ERs exhibited
different expression patterns, ER
showing strong nuclear
expression, and ER
granular cytoplasmic expression (24).
This different distribution may contribute to a variable E2
responsiveness (254).
The different expression patterns of ER and androgen re-
ceptor in the hair follicle and their potential biological rel-
evance deserve special attention, because follicular dermal
papilla cells are thought to be the primary target cells within
the hair follicle that mediate the growth stimulating signal of
androgens by releasing growth factors that act in a paracrine
fashion on other cells of hair follicle (255, 256). The exact
pattern of androgen receptor expression in the mesenchyme
and epithelium of human hair follicle remains a matter of
contention, with published results heavily influenced by the
respective methodology employed. However, Thornton et al.
and other authors have reported that no androgen receptor
immunoreactivity is detected in the keratinocytes of the
outer root sheath (including its bulge region) and of the inner
root sheath, whereas the majority of dermal papilla cells
express androgen receptor (257). In contrast, ERs are more
widely expressed, and importantly, ER
is strongly ex-
pressed in the bulge region of the outer root sheath. This
region contains stem cells for hair follicle keratinocytes that
regenerate the follicle during the anagen phase. This suggests
that these epithelial stem cells are targets for estrogen action.
The wide distribution of ER
in human pilosebaceous unit
suggests that estrogens play an important role in the main-
tenance and the regulation of the hair follicle and provides
further evidence for estrogen action in nonclassic target tis-
sues. Recently, it was reported that in cultured dermal pa-
pilla cells from nonbalding male donors, both ER
and ER
showed a consistently higher expression, both at the RNA
and protein levels, in occiput dermal papilla cells compared
with vertex dermal papilla cells (258). With respect to ER
immunoreactivity, we found that, in anagen VI follicles mi-
crodissected from frontotemporal skin, there was a remark-
able distribution difference between male and female hair
follicles from frontotemporal scalp skin: ER
immunoreac-
tivity was found in male scalp hair follicles predominantly
in the matrix keratinocytes, whereas in female hair follicles,
ER
immunoreactivity was predominantly found in the der-
mal papilla fibroblasts (10). These data not only highlight
substantial, previously underappreciated sex-dependent dif-
ferences in ER
expression of an important peripheral E2
target organ, but also underscore the importance of inves-
tigating whether E2 effects on the human hair follicle are
location-dependent, as is well-recognized for the paradoxical
hair growth effects of androgens (64, 259, 260).
Conflicting data have been presented concerning ER ex-
pression patterns in murine hair follicles. It has been reported
that ER
was expressed only in the dermal papilla and outer
root sheath of telogen and early anagen mouse hair follicles
and that ER
was undetectable (26, 250). Recently, however,
we could show that both ER
and ER
as well as the splice
variant ER
ins are expressed throughout the entire, depi-
lation-induced murine hair cycle at both the protein and
RNA levels (28). In addition, hair follicles in late anagen
(anagen VI) were highly sensitive to regulation by topically
applied E2, which rapidly induced premature catagen entry.
Therefore, anagen VI mouse pelage hair follicles must ex-
press fully functional ERs (28).
ER
immunoreactivity peaks in murine telogen follicles
within the dermal papilla and the sebaceous gland, whereas
the inner root sheath and outer root sheath show weaker
immunoreactivity. In anagen VI, ER
immunoreactivity (IR)
is detectable in the outer root sheath and the dermal papilla,
whereas in early catagen it is restricted to the dermal papilla
and the secondary hair germ. In anagen VI follicles, ER
is
weakly positive in hair matrix and outer root sheath, whereas
in catagen and telogen follicles, ER
is expressed in the
dermal papilla, inner root sheath, outer root sheath, and the
sebaceous gland. By RT-PCR, ER
and ER
transcripts can
be detected in telogen, anagen V and VI, and late catagen skin
mRNA extracts. Investigation of ER
knockout mice showed
an accelerated catagen development along with an increase
in the number of apoptotic hair follicle keratinocytes (28).
Taken together, this suggests that the catagen-promoting
properties of E2 in murine skin are mediated by ER
and that
ER
mainly functions as a silencer of ER
action in murine
hair biology. (An additional list on reported expression of
estrogen signaling components is provided in Table 2).
C. Estrogen target genes in the pilosebaceous unit
The classical mechanism of estrogen action involves bind-
ing to its receptors in the nucleus, after which the receptors
dimerize and bind to specific response elements known as
EREs located in the promoters of target genes. However, ERs
can also regulate gene expression without directly binding to
DNA. This occurs through protein-protein interactions with
other DNA-binding transcription factors in the nucleus (261).
About one third of the genes in humans that are regulated by
ERs do not contain ERE-like sequences (262). Candidates of
estrogen target genes with relevance to pilosebaceous biol-
ogy that are activated without ERE promoter include IGF-I,
collagenase, EGF, EGFR, and cyclin D1 (261, 263). Instead,
progesterone receptor, prolactin, and lactoferrin are exam-
ples of relevant target genes in the pilosebaceous unit with
consensus EREs (263–265).
Zouboulis et al. (266) showed that, in sebocytes, the ex-
pression of peroxisome proliferator-activated receptor
(PPAR)
, postulated to be required for androgen-induced
lipogenesis, was down-regulated by the phytoestrogen
genistein, whereas E2 enhanced the metabolism of prosta-
glandin D2 to 12-prostaglandin J2, a natural PPAR
ligand.
Additionally, the same group found that E2 increases IGF-I
synthesis and down-regulates IGF-I receptor expression
(266).
Ohnemus et al. Hair Follicle as an Estrogen Target and Source Endocrine Reviews, October 2006, 27(6):677–706 689
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 03 November 2015. at 05:08 For personal use only. No other uses without permission. . All rights reserved.
Recently, we have employed cDNA microarray to screen
for genes in organ-cultured human scalp hair follicles that
respond to E2 stimulation with transcriptional changes, us-
ing a skin focus chip and comparing the E2 response of male
and female human frontotemporal scalp hair follicles. Of
1300 genes screened, more than 60 E2-responsive genes were
detected. Several genes were modulated equidirectionally in
both sexes (e.g., down-regulation of osteopontin and hevin
highly expressed endothelial venule protein; up-regulation
of cytokeratin type II and bone morphogenetic protein 7).
Intriguingly, however, several genes showed distinct regu-
latory responses in male and female hair follicles: e.g., down-
regulation of filaggrin and FGF receptor 2 in males; up-
regulation of nuclear receptor subfamily 4, group A, member
1 in females; whereas cysteine-rich 61, fos-like antigen 2, and
collagen IV A6 were up-regulated in males, yet down-reg-
ulated in females (10). This reveals that terminal human scalp
hair follicles from one defined region show strikingly dif-
ferent, sex-dependent biological responses to stimulation
with the same ER ligand, strongly advocating gender-tai-
lored management of female vs. male pattern balding (an-
drogenetic alopecia) (10).
D. Species-specific differences in estrogen actions on hair
follicle cycling
E2 has long been recognized to profoundly modulate hair
growth, acting primarily as a hair growth inhibitor in mam-
malian species as diverse as mice, rats, guinea pigs, and dogs
(1, 2, 51, 267, 268) (Table 1). For example, injections of es-
trogens or topical applications of ER agonists effectively in-
hibits spontaneous hair growth in rats and dogs (2, 267). Hair
regrowth after plucking was noted to differ between male
and female rats, and spontaneous hair growth was inhibited
during periods of lactation and pregnancy in mice (269). In
young castrated rats, im injections of estrogens inhibited the
initiation of follicle growth and prolonged the duration of the
entire hair growth cycle, resulting in a fine and sparse pelage
(6). Removal of the ovaries in rats accelerated the passage of
the moult, increased the rate of hair growth and length of
hairs, and accelerated the loss of club hairs. Treatment of
ovariectomized rats with estradiol delayed the initiation of
the wave, slowed its passage, reduced the rate of growth and
definitive length, and delayed the loss of the telogen club
hairs (7). Spayed female rats shed more than 80% of telogen
club hairs within 2 wk of the start of anagen; implantation of
estradiol delayed this process by 3–4 wk. Anagen initiation
has been shown to be delayed by more than 5 wk in the
ventral hair follicles of E2-treated rats (7). Therefore, estra-
diol can act as a brake on hair follicle cycling by delaying the
initiation of anagen and by prolonging the duration of
telogen.
Hale and Ebling (270, 271) showed that estrogens reduce
the rate of growth and the ultimate length of spontaneously
erupting hairs by shortening the anagen period, whereas
ovariectomy of rats tended to advance the spontaneous erup-
tion of successive generations of hairs by shortening each
complete hair cycle. Oh and Smart (26) found that, in mice,
topical E2 administration to clipped dorsal skin arrested hair
follicles in telogen and produced a profound and prolonged
inhibition of hair growth, whereas treatment with the bio-
logically inactive stereoisomer 17
-estradiol did not alter
hair growth. That topical E2 arrests murine pelage hair fol-
licles in telogen was independently confirmed (28). This rep-
resents an important rediscovery of older work that had
already demonstrated a prolongation of telogen by topical E2
application in various rodent species (272, 273).
Vice versa, orchiectomy induces a premature telogen-ana-
gen transition and an anagen wave (223, 274). Application of
E2 to gonadectomized animals inhibits hair follicle growth
and blocks the telogen-to-anagen transition (223). Topical
treatment with the selective ER antagonist ICI182-780, a pure
ER antagonist, reportedly caused premature anagen induc-
tion (26, 273). These results were later confirmed by the same
group in C57BL/6 and C3H male and female mice (272).
More precise analysis of hair follicle cycling in C57BL/6
mice, however, revealed that the ER antagonist ICI182-780
does not prematurely initiate anagen but accelerates anagen
development and anagen wave spreading, once anagen has
been initiated by independent (endogenous) signals (28).
Most recently, we could demonstrate that E2 is also a
potent catagen inducer (27, 28). [This was in contrast to older
studies in different rodent species, where no changes in the
duration of anagen had been noted after E2 application
(273)]. In line with these newly discovered catagen-inducing
properties of topical E2 (27, 28, 114), ER
knockout mice
display accelerated catagen development, along with an in-
crease in the number of apoptotic hair follicle keratinocytes
(28). This suggests that, contrary to previous working hy-
potheses (274), ER
does indeed play a significant role in
murine hair growth control: whereas the catagen-promoting
properties of E2 are mediated via ER
,ER
may mainly
function as a silencer of ER
action in hair biology (28).
Nevertheless, ER
is thought to serve as the predominant ER
in the hair follicle of animals (31, 233, 254).
In mice, ER expression is stringently hair cycle-dependent
(26, 28, 33), and nuclear immunoreactivity was detected for
both ER
and ER
throughout all investigated hair cycle
stages (telogen, anagen VI, catagen) in mice (28) (Table 2).
Together with the fact that the treatment of murine anagen
hair follicles in vivo induces catagen, i.e., a dramatic remod-
eling process of a complex miniorgan (28), this renders ob-
solete the old concept that only telogen follicles express ER
and engage in ER-mediated signaling (26). However, it is
perfectly reasonable to propose that ER-mediated signaling
operates as an endogenous paracrine regulator of the hair
cycle (26). Stringently controlled changes in the expression
and/or activity of ER may, therefore, well be an integral
component of the elusive hair cycle clock (28, 35) (Fig. 3).
Just like glucocorticoids and calcitriols, topical E2 also
promotes the so-called dystrophic catagen response pathway
of chemotherapy-damaged anagen hair follicles (27, 114).
Topical E2 significantly alters the cycling response of murine
follicles to cyclophosphamide, whereas the ER antagonist
ICI182-780 exerts no such effects. Initially, topical E2 en-
hances chemotherapy-induced alopecia by forcing the folli-
cles into the dystrophic catagen response pathway to hair
follicle damage, which allows for a maximally fast secondary
690 Endocrine Reviews, October 2006, 27(6):677–706 Ohnemus et al. Hair Follicle as an Estrogen Target and Source
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 03 November 2015. at 05:08 For personal use only. No other uses without permission. . All rights reserved.
recovery by construction of a new anagen hair follicle. In-
stead, follicles treated with ICI182-780 or vehicle shift into the
dystrophic anagen response pathway. Consequently, the re-
growth of normally pigmented hair shafts after chemother-
apy-induced alopecia is significantly accelerated in the E2-
treated group (27).
However, these findings in mice contrast with long-stand-
ing clinical experience in the topical application of E2 to the
human scalp. For human scalp hair, topical E2 has long been
used in the management of telogen effluvium and androge-
netic alopecia, especially in women (29, 58, 275). Although
this remains to be unequivocally demonstrated in vivo,E2has
been proposed to decrease the telogen rate and to prolong the
anagen phase in human scalp skin, justifying the use of
topical E2 in the management of hair loss characterized pre-
mature catagen entry, such as androgenetic alopecia and
telogen effluvium (29, 30, 88, 275–277).
Such a catagen-inhibitory effect of E2 in human scalp hair
follicles would help to explain the well-established clinical
observation that topical E2 or high systemic E2 levels during
estrogen-based contraception and during pregnancy in-
crease the telogen/anagen ratio, thus notably improving a
preexisting telogen effluvium, and that a telogen effluvium
occurs postpartum (supposedly due to sudden E2 with-
drawal) (51, 276, 277). In pregnant women, the scalp hair
shaft diameter reportedly increases compared with nonpreg-
nant women, although it remains to be determined which
pregnancy-associated factors are responsible for this phe-
nomenon (278). In view of the complex, multiple concomi-
tant endocrine changes during and after pregnancy and lac-
tation (including dramatic fluctuations in gestagen and
prolactin levels), it clearly is very difficult to dissociate
strictly E2-based hair growth effects from those that other
hormones might exert on the human scalp hair follicle in vivo
during this time (10, 12).
Although rodent hair follicles generally respond to E2
stimulation with an inhibition of hair shaft formation, this
is not necessarily true for human scalp hair follicles, at
least in males. Studying the isolated effects of E2 on human
hair growth in vitro (in organ-cultured, microdissected
human anagen hair bulbs), we recently showed that, in
frontotemporal male hair follicles, E2 indeed slightly pro-
longs anagen and stimulates hair shaft elongation (11).
Corresponding studies (279–281) have reported that E2
inhibits hair shaft elongation in vitro or that E2 does not
influence the decay rate of organ-cultured human anagen
hair follicles from occipital scalp skin (measured by mor-
phology and autoradiographic
3
H-thymidine incorpora
-
tion) (279). These partially conflicting reports may become
reconciled, once larger studies with organ-cultured hu-
man scalp hair follicles have been performed that system-
atically distinguish between male, female, frontotemporal,
and occipital follicle populations, and that correlate hair
shaft elongation with hair cycle effects. In addition, these
in vitro studies need to be complemented by and compared
with the results of clinical trials on the scalp hair growth
effects of topical E2, documented by professional photo-
trichogram methodology (88).
E. Gender- and location-specific differences in estrogen
actions on the hair follicle
Besides the estrogen-dependent development of breasts,
the hair follicle is the other most characteristic skin feature
of mammals. The growth of the beard and of pubic and
axillary hair is, in adults of either sex, dependent on the
production of sex steroids (51). The pelage changes as a
mammal grows, and that of the adult often differs markedly
from that of the juvenile animal, a circumstance that may
reflect the changing requirements of heat regulation, cam-
ouflage, sexual and reproductive activity, and social com-
munication (51). As to sexual hair growth, it has been rec-
ognized that not only adrenal androgens but also ovarian
hormones may play a role for the growth of pubic and ax-
illary hair in human females. Pubic hair can still develop in
the presence of preadrenarchal levels of adrenal androgens
in girls with precocious puberty or primary adrenal insuf-
ficiency (282). Females with primary ovarian insufficiency
have very sparse pubic and axillary hair, which can be stim-
ulated to grow with adequate and prolonged estrogenic ther-
apy (283).
Estrogens act, either alone or together with androgens,
directly at the level of the hair follicle in pubic skin to stim-
ulate hair growth. However, in the absence of active andro-
gen receptors, E2 cannot promote sexual hair growth, e.g., in
patients with complete testicular feminization who do not
grow pubic and axillary hair, despite signs of E2 effects in
other tissues (33).
In human occipital scalp hair follicles, E2 may inhibit hair
shaft elongation in both males and females in vitro (279, 280).
However, we found sex-dependent differences of frontotem-
poral scalp hair shaft elongation after E2 treatment in vitro:
in females the hair shaft elongation was inhibited, whereas
E2 significantly stimulated hair shaft elongation in human
frontotemporal anagen hair follicles from male patients in
vitro (12, 281). This corresponded to a significantly up-reg-
ulated proliferation rate of the matrix keratinocytes in the
male frontotemporal scalp hair follicles compared with fe-
male hair follicles (12).
The apparent differences of E2 action on human hair
growth in vivo and in vitro can be reconciled if one considers
that, clinically, even a significant inhibition of the speed of
hair shaft production per time unit on the human scalp after
E2 administration would hardly be noticed by either the
patient or the doctor (unless specifically assessed with highly
sensitive methods). In contrast, if the percentage of telogen
follicles among the approximately 100,000 human scalp hair
follicles changes in favor of anagen follicles under the influ-
ence of topical E2, this would result in an easily apparent
reduction in telogen effluvium—a phenomenon that is
quickly recorded by the patient, especially when assessing
the hair loss after combing or shampooing (37, 87). Thus, the
major, clinically relevant effect of topical E2 on human scalp
hair follicles seems to be the inhibition of catagen (88).
The observed differences in the E2 response of female vs.
male frontotemporal hair follicles raise the question whether
E2 exerts similarly paradoxical, site-dependent effects on
human hair growth (12) as are recognized for androgens (70).
Perhaps, there are location-dependent differences in defined
Ohnemus et al. Hair Follicle as an Estrogen Target and Source Endocrine Reviews, October 2006, 27(6):677–706 691
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 03 November 2015. at 05:08 For personal use only. No other uses without permission. . All rights reserved.
populations of human hair follicles that react to E2 stimu-
lation in a divergent signaling and gene expression response,
similar to the response of beard hair vs. scalp follicles to
androgen stimulation with respect to TGF
1 vs. IGF-I ex-
pression in the dermal papilla (118, 259).
Due to different activities of key enzymes in dermal, epi-
dermal, or sebaceous compartments like aromatase, 17
-
hydroxysteroid dehydrogenase, or steroid sulfatase (57, 69,
123), there may indeed be important regional differences in
the extrafollicular estrogen metabolism of/in hair follicle in
different integumental locations that must be carefully taken
into account when estrogens are applied topically. However,
possibly more important than such differences in local in-
tracutaneous estrogen metabolism are gender-associated dif-
ferences in the hair follicle response to E2 stimulation: fron-
totemporal female scalp hair follicles display most of their
ER
-associated immunoreactivity inside the (mesenchymal)
dermal papilla, whereas ER-like immunoreactivity in male
frontotemporal scalp hair follicles is located in the (epithelial)
hair matrix and outer root sheath (10).
Furthermore, when comparing the distribution pattern of
ER
-like immunoreactivity in male and female scalp hair
follicles from one defined integumental site (frontotempo-
ral), immunohistology suggests striking differences in the
preferential location of ER protein expression; although ER
protein expression in female hair follicles dominated in the
mesenchymal command center of the hair follicle, the dermal
papilla, male scalp hair follicles exhibited a much more
prominent ER
-like immunoreactivity in the hair follicle ep-
ithelium (especially the hair matrix) than in the hair follicle
mesenchyme. However, dermal papilla ER expression in
female scalp follicles was greatly enhanced after hair follicle
stimulation with E2, whereas no such up-regulation of ER
expression was seen in E2-stimulated male scalp hair follicles
(10). In the context of this study strikingly sex-dependent
differences in gene regulation of frontotemporal human
scalp hair follicles in response to E2 stimulation have recently
surfaced from microarray analyses (10).
Although these findings remain to be followed up and
their biological significance for hair growth control remains
to be clarified, they may be interpreted as only the “tip of an
iceberg” of very substantial, previously unknown gender
differences that must be taken into account when exploring
the hair follicle response to E2 stimulation. Given the re-
ported differences in the E2 response between frontotempo-
ral and occipital human scalp hair follicles summarized
above (10, 11, 279–281), it is not unreasonable to expect that
a similar claim can be made for location-dependent differ-
ences in hair follicle responses to ER activation.
F. Clinical hair growth effects of estrogens
There are only a few reports on the use of systemic es-
trogens for hair loss management. Estrogens have been used
for topical treatment of hair diseases for more than half a
century (284) and constitute a firm staple of management
strategies for female pattern androgenetic alopecia in central
Europe (88). Orentreich observed in 1969 a decrease in daily
effluvium during therapy with systemic estrogens (285),
which were reported to increase the proliferation rate, slow
down differentiation, and, thus, postpone telogen effluvium
(286). On this basis, even intralesional stilbene administra-
tion was once recommended for the treatment of alopecia
areata (287). Some studies have reported an increased anagen
and decreased telogen rate after treatment with estrogens,
compared with placebo (288, 289). However, professionally
executed, double-blind, placebo-controlled, randomized,
prospective clinical trials on the efficacy of topical E2 in the
treatment of androgenetic alopecia and non-androgen-
dependent telogen effluvium are still painfully missing.
Since the studies of Hamilton, we know that androgens
play a crucial role in the onset and progression of androge-
netic alopecia (290). Androgenetic alopecia occurs in genet-
ically susceptible individuals and in androgen-sensitive hair
follicles when testosterone is transformed to 5
-dihydrotes-
tosterone by 5
-reductase (88, 291). This conversion is in-
hibited inter alia by E2 (288, 292). For the treatment of an-
drogenetic alopecia in women, solutions containing estradiol
benzoate, estradiol valerate, or 17
-estradiol are commer-
cially available in Europe. Due to unwanted side effects like
gynecomastia, E2 should not be used in men because very
high topical doses seem to be required to obtain measurable
hair growth effects (84), whereas the inactive stereoisomer
17
-estradiol may also be prescribed for men.
Its claimed efficacy for male or female pattern balding
(290), however, remains as yet unsupported by solid, pro-
fessionally designed and executed, prospective, double-
blind, placebo-controlled long-term clinical trials. However,
that 17
-estradiol has been reported to induce aromatase
activity in organ-cultured human anagen hair follicles, with
the consequence of an increased conversion of testosterone
to E2 and androstenedione (293) certainly encourages one to
systematically explore the use of 17
-estradiol for androge-
netic alopecia in men and women.
G. Relevant signaling cross talks in the hair follicle
1. ER target genes. An exhaustive list of factors responding to
estrogen and/or ER signaling is readily available in the com-
prehensive ERGDB database (294)
1
. Table 3 shows a small
selection from this database. Below, we discuss some of these
factors that are already recognized as hair growth-modula-
tory agents.
The association of ER and growth factor receptors includ-
ing their second messengers has been widely accepted. Just
a few important genes regulated by ERs, which also are
recognized for their involvement in hair growth control, are
listed here as examples: progesterone receptor, EGFR, sev-
eral growth factors like IGF-I, TGF-
and TGF-
, cathepsin
D, and several protooncogenes (e.g., c-fos, c-myc, and c-jun),
as well as an array of heat shock proteins (295–297). Some of
these factors are capable of nonclassical ER activation
through modifications such as phosphorylation (298). In cell
culture, for example, the growth factors IGF-I, EGF, and
1
Only a subset of proteins will be presented; these were selected due
to the fact that they are more or less well documented to be involved in
hair follicle morphogenesis, proliferation, or cycling. It should be noted
here that the original data were not collected from skin or skin-derived
cell populations but are from very diverse origin and multiple different
vertebrate species.
692 Endocrine Reviews, October 2006, 27(6):677–706 Ohnemus et al. Hair Follicle as an Estrogen Target and Source
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 03 November 2015. at 05:08 For personal use only. No other uses without permission. . All rights reserved.
TGF-
increase the transcription of ER target genes (298
301). In the following, we list a few E2 targets to illustrate the
rather stunning complexity of E2-induced signaling, without
offering exhaustive coverage of all pathways that might be
relevant for E2-induced effects on the hair follicle.
2. ER and EGF. Through binding to its membrane bound
receptor (EGFR), EGF can partially mimic the E2 stimulation
of uterine growth in ovariectomized mice (301). EGFR may
exert this effect by recruiting ER for its downstream signal-
ing, because ER antagonists can block it (302). Vice versa,
EGF signaling can be reinforced by E2 (302). This cross-
connection is interesting in view of other long-appreciated
hair growth-inhibitory effects of EGF in various species in-
cluding man, which includes the induction of apoptosis-
driven catagen (303–311) (Fig. 6). Interestingly, the EGFR
gene is directly estrogen responsive (312), with multiple ERE
sequences upstream of the transcription start site. Also,
ERBB2 is ER-regulated (313) and exhibits ERE sequences
within its putative promoter region (Table 3). It is, therefore,
interesting to ask to which extent ER activation results in hair
growth effects that are mediated, at least in part, via the
described connection to EGF signaling.
3. ER and MAPK pathway. The MAPKs are promiscuous in-
tracellular signal transducers that are modulated by many
extra- and intracellular receptors and signaling molecules,
including the above mentioned FGF/FGF receptor, EGFR,
TGF family including BMPs, noncanonical Wnt pathway,
and the ERs themselves (314–316) (Fig. 6). Studies with thy-
roid cells (317, 318) indicate that ER indirectly alters the cell
cycle through the MAPK pathway and E2 increases the pro-
liferation rate and enhances expression of cyclin D1 (318).
Intriguingly, this effect was abolishable by coapplication of
the MAPK inhibitor PD098059 to E2-treated cells. In response
to E2, both MAPK isoenzymes ERK1 and ERK2 were strongly
phosphorylated in benign and malignant thyroid cells. This
indicates that, at least in benign and malignant thyroid cells,
ER utilizes the MAPK pathway to interfere with the cell cycle
via cyclin D1. This provides one plausible pathway by which
ER-mediated signaling could affect hair follicle cycling via
cyclin D1 (319).
4. ER and the Wnt pathway. Exciting new connections between
the ER and other important factors critically involved in hair
growth control were found in recent years. This is high-
lighted by the convergence of estrogen signaling with the
Wnt signal transduction pathway (320, 321) (Fig. 6). The
morphogenetic factors involved in this pathway have been
the subject of studies centered on the hair follicle, including
stem cell regulation, hair follicle induction, morphogenesis,
and differentiation (323–326).
The Wnt pathway is composed of a class of lipid-modified
diffusible para- and autocrine factors that act through var-
ious intracellular signals, including the calcium and planar
cell polarity pathways (323). The best characterized is the
canonical pathway, with beta-catenin in its center (324, 325).
Although this protein is involved in the pathogenesis of
many human cancers, under physiological conditions it has
at least two major distinct functions, one as a cytoplasmic
cytoskeleton-adhesion mediator protein placed adjacent to
the cell membrane, within the cadherin complex in the ad-
herens junctions (326) and one through participation in reg-
ulatory networks of a whole array of important factors for
keratinocyte and hair follicle homeostasis, e.g., c-myc (295,
327, 328), cyclin D1 (319), PPAR-
(328), which are docu-
mented to alter hair follicle cycling or homeostasis.
Wise is a secreted Wnt modulator that has been found to
be differentially expressed in inductive dermal papilla cells
vs. cells in culture (329). As ER-regulated targets include the
Wnt pathway, Wise is especially intriguing to study, because
it is expressed at the site where ER
is dominantly located.
Together with the presence of Wise in the bulge region (329),
the site believed to be suppressed in telogen, the connection
between ERs and the Wnt pathway is also repeatedly present
in the hair follicle.
5. ER and the TGF/BMP family. Members of the TGF
/BMP
family and their functional antagonists are recognized as
critical regulators of hair follicle morphogenesis and cycling
(322, 330–332). Therefore, this family is likely to be critically
important for understanding the role of estrogens in hair
biology to follow up the increasing insight into cross-con-
nections between ER-mediated signaling and the TGF
/
BMP family; e.g., BMP genes are principal ER-responsive
genes (333–335) whose transcriptional modulation could
profoundly affect hair growth (Fig. 6).
Activators as well as inhibitors play essential roles in the
control of postnatal skin remodeling and hair follicle growth
(111, 336–339). Artificial loss or gain of BMP signals induces
severe alterations in skin morphogenesis. Normally high
levels of BMP-6 transcripts as well as proteins are expressed
in the suprabasal layer of the murine epidermis from em-
TABLE 3. ER-regulated target genes according to ERGDB (294)
Gene Unigene ID ERE elements upstream of TSS (nt) Refs.
Progesterone receptor (PGR) Hs.368072 None 368
Epidermal growth factor receptor (EGFR) Hs.488293 560, 544 369
IGF-I receptor (IGFIR) Hs.20573 None 370
TGF-
(TGFA) Hs.170009 None 368
TGF-
3 (TGFB3) Hs.2025 None 334
Cathepsin D (CTSD) Hs.121575, Hs.546248 231 nt, 189 nt 371
FOS Hs.25647 3412 nt 372
MYC Hs.202453 2839 nt, 2823 nt 373
ERBB2 Hs.446352 1424 nt, 2972 nt 373
BMP2 3982 nt 333–334
HOX-C6, HOX-D4, HOX-A7, HOX-C8 Hs.820, Hs.386365, Mm.294826, Mm.6167 None, multiple, yes, yes 346, 373
Msx-2 Hs.89404 Multiple 350, 374
Ohnemus et al. Hair Follicle as an Estrogen Target and Source Endocrine Reviews, October 2006, 27(6):677–706 693
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 03 November 2015. at 05:08 For personal use only. No other uses without permission. . All rights reserved.
bryonic day 15.5, whereas BMP-7 mRNA is found in the basal
epidermal layers at late stages of embryonic development
(340–342). BMP-2 and BMP-4 transcripts are more restricted
to the hair follicle epithelium and mesenchyme (340, 343).
Notably, the BMP receptor (BMPR)-1a is localized to murine
epidermis at embryonic day 16.5, whereas the BMPR-1b is
FIG. 6. Regulatory modules relevant for the hair follicle regulation with partial ties to estrogen signaling. ER regulates the depicted genome
network at various endpoints (depicted with encircled numbers 1–5). Factors found to be important for hair follicle patterning, cycling, as well
homeostasis are richly intertwined. Paracrine factors are liberated within the hair follicle to be perceived by the same cell in an autocrine fashion,
or by adjacent cells in a paracrine fashion. These factors control tissue proliferation as lineage-specific within the hair bulge. ERs have been
shown to be upstream or downstream of various regulatory connections (depicted by small arrows). Functional interactions with unknown
mechanisms are depicted by question marks. Arrows in principle signify activating or inhibiting interactions. [Modified after van Steensel et
al. (375) with permission from Editions John Libbey Eurotext Paris.] NOTCH, Notch receptor; Delta, delta ligand of NOTCH receptor; EDA,
ectodysplasin; EDAR, EDA receptor; SRC kinase, Rous sarcoma virus tyrosine kinase; APC, adenomatosis polyposis coli tumor suppressor
protein; TCF, T cell factor; LEF1, lymphocyte-enhancing factor 1; SHH, sonic hedgehog protein; PTC, patched; SMO, smoothened; Noggin,
Noggin protein; E-cadherin, epithelial cadherin adhesion molecule; Wnt, Drosophila wingless homolog (acronym for wingless-type mouse breast
tumor virus); Frz, frizzled; GLI, glioma-associated gene; EN-1, engrailed 1; DLX2, distal-less homeobox gene 2.
694 Endocrine Reviews, October 2006, 27(6):677–706 Ohnemus et al. Hair Follicle as an Estrogen Target and Source
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 03 November 2015. at 05:08 For personal use only. No other uses without permission. . All rights reserved.
present in the suprabasal keratinocytes (338). Mice that lack
a key BMP antagonist, noggin, have around 90% reduced
hair follicles and display increased proliferation of the epi-
dermal compartment with down-regulation of Keratin 10
and misplaced up-regulation of Keratin 14 in the upper epi-
dermal layers (338, 339); overactivation under K14 promoter
leads to increased hair follicle density with various defects in
skin appendage morphogenesis and differentiation (344).
Also, mature hair follicle cycling is regulated by BMPs and
their antagonists. For example, the active hair growth stage,
anagen, is accompanied by down-regulation of the BMP4
and increased noggin mRNA expression in the hair follicle.
Inhibition of BMPs by Noggin protein induces a new hair
growth phase in postnatal telogen skin in vivo, and BMP
signaling regulates hair follicle cycling and differentiation in
mice (337). This underscores the concept that factors con-
tributing to hair follicle morphogenesis often also participate
as factors regulating the hair cycle (338, 339).
Therefore, it is important to note in the current context that
ligands of the TGF-
group, TGF
1, as well as BMP sub-
group, BMP1, -15, -2, -8a, and -8b (333) are recognized as E2
genes that are up-regulated after ER activation. BMP1 and
BMP2 are directly ER-regulated (334, 335), and they have
ERE sequences 1004 and 3982 nucleotides upstream of the
transcription start site, respectively (294). The TGF-
1 gene
was found to be estrogen sensitive in vivo and in vitro (334),
but without detectable ERE sequence in the putative pro-
moter region (294), suggesting an ERE-independent mode of
transcriptional regulation. This ren