Nearly 200 different conditions have been described under the term ectodermal dysplasia (ED). EDs are genetic disorders identified by lack or dysgenesis of at least two ectodermal derivatives such as hair, nails, teeth, or sweat glands. Currently, the causative gene has been identified in about 30 different EDs [Priolo and Laganà, 2001; Lamatine, 2003; Itin and Fistarol, 2004; Irvine, 2005]. Hypohidrotic (or anhidrotic) ED (HED or EDA) is the most frequent form of EDs that can be inherited in an X-linked (XL), (MIM 305100), autosomal recessive (AR) (MIM 224900) or (AD) autosomal dominant (MIM 129490) manner. HED is characterized by sparse hair, missing or scanty eye brows and lashes, severe oligodontia, absent or reduced sweating, as well as defects in a number of other ectodermal organs [Mikkola and Thesleff, 2003]. XL-HED was the first ED in which the defective gene was cloned thereby leading to the identification of a novel signaling molecule of the tumor necrosis factor (TNF) superfamily named ectodysplasin (EDA) [Kere et al., 1996; Srivastava et al., 1997]. Autosomal forms of HED are due to mutations in the EDA receptor (EDAR), a novel TNF receptor family member [Monreal et al., 1999; Chassaing et al., 2006]. EDAR binds specifically the A1 isoform of EDA (EDA-A1), but not the EDA-A2 isoform that utilizes a distinct receptor X-linked EDA receptor (XEDAR) for signal transmission [Yan et al., 2000]. EDA-A2/XEDAR pathway has not been associated with ED while autosomal HED may also be caused by mutations in a cytosolic, EDAR -specific adapter molecule named EDAR-associated death domain (EDARADD) [Headon et al., 2001; Bal et al., 2007]. The EDA-EDAR-EDARADD axis is a unique example of naturally occurring mutations in ligand, receptor, and adaptor protein giving rise to the same phenotypic disease affecting development of epidermal appendages. This pathway is highly conserved from fish to man and regulates the development of ectodermal organs in all vertebrate species studied to date [Pantalacci et al., 2008].
Most HED cases carry mutations in EDA—currently over 100 different mutations in the EDA gene have been reported, while only ∼20 and 2 causative mutations are known for EDAR and EDARADD, respectively (The Human Gene Mutation Database: www.hgmd.cf.ac.uk). Almost all of the EDA mutations are thought to be null mutations and show no clear genotype–phenotype correlation [Schneider et al., 2001]. Interestingly, there are several recent reports of families with missense mutations in EDA that are associated with non-syndromic tooth agenesis [Schneider et al., 2001; Tao et al., 2006; Tarpey et al., 2007; Li et al., 2008]. It is likely that these mutations are hypomorphic ones [Schneider et al., 2001] and thus these findings suggest that dental tissues are particularly susceptible to reduction in EDA signaling activity.
LESSONS FROM HED MOUSE MODELS—EDA REGULATES MULTIPLE STEPS IN DEVELOPING ECTODERMAL ORGANS
Mouse models for HED (tabby, downless, sleek, crinkled) were described already in the mid-1900s [Mikkola and Thesleff, 2003]. Cloning of the Eda pathway genes confirmed that Eda, Edar, and Edaradd were mutated in these mice [Srivastava et al., 1997; Headon and Overbeek, 1999; Headon et al., 2001; Yan et al., 2002]. Similar to humans, mutations in mouse Edar and Edaradd may cause both recessive and dominant forms of HED. The defects in Eda, Edar, and Edaradd mutant mice are highly similar to those seen in humans and include tooth and hair abnormalities. Over 20 different glands are affected, and most notably sweat glands, normally found at the footpads, are absent [Mikkola and Thesleff, 2003].
These loss-of-function mouse models have been central to the elucidation of the developmental function of Eda [Mikkola and Thesleff, 2003; Mikkola, 2008]. In addition, several transgenic mice ectopically expressing Eda have been generated during recent years [Cui et al., 2003; Mustonen et al., 2003; Zhang et al., 2003]. Analysis of all these mouse models has shown that Eda regulates organogenesis at multiple levels, from initiation to differentiation [Cui and Schlessinger, 2006; Mikkola, 2008]. Until now, most progress has been made in understanding the role of the Eda pathway in tooth and hair follicle biology while its function in other ectodermal organs is less well recognized. Therefore I will focus on these two organs in what follows.
The onset of skin appendage development is marked by the appearance of focal thickenings of the epithelium known as placodes. In the next stage the placode invaginates to form a bud. The hair follicle bud grows rapidly downwards and encases a cluster of dermal cells that form the dermal papilla while the tooth bud epithelium undergoes more complex folding morphogenesis that resemble first a cap and later a bell in histological (frontal) sections [Mikkola and Millar, 2006].
Studies from several groups show that the formation of placodes is the outcome of very complex reciprocal interactions between different pathways including the Wnt, fibroblast growth factor (FGF), bone morphogenetic protein (BMP) and EDA families [Mikkola and Millar, 2006; Mikkola, 2007]. At this developmental stage expression of Edar and thus, the effects of Eda are strictly confined to placodes [Mikkola, 2008]. In mice, hair placodes form in several waves that give rise to distinct hair types [Dry, 1926; Mann, 1962; Schlake, 2007]. Primary follicles form at embryonic day 14 (E14) and give rise to guard hairs (long, straight hairs that comprise about 2–4% of the coat); secondary follicles appear at E16 and give rise to awl hairs (short, straight hairs; 25–30% of the coat), and tertiary follicles develop during few days around birth and give rise to the two bent hair types: zigzags (several bends; majority of the coat) and auchenes (a single bend, 5–10% of the coat). It has long been known that Eda is essential for the formation of primary hair placodes but curiously, not for the other ones. Recent findings indicate that Edar and a related TNF receptor Troy function redundantly during formation of secondary hair placodes which fail to form in compound Eda/Troy mutant embryos [Pispa et al., 2008]. Although it is not yet clear exactly how Eda regulates placode formation it seems that it acts downstream of a primary inductive signal to stabilize emerging hair placodes. In the absence of Edar signaling only rudimentary pre-placodes form [Mou et al., 2006; Schmidt-Ullrich et al., 2006; Fliniaux et al., 2008], while excess of Eda increases the size of placodes [Mustonen et al., 2004].
Dental (and mammary) placodes develop roughly normally in Eda and Edar deficient embryos [Pispa et al., 1999; Tucker et al., 2000]. Intriguingly, ectopic tooth (and mammary) placodes and consequently supernumerary organs form in Eda gain-of-function mice [Mustonen et al., 2003, 2004] implying that in addition to hair follicles, Eda-A1 is also involved in the initiation of tooth (and mammary) development, but its loss may be compensated for by some other pathway(s). In any case, it appears that Eda regulates the size of the tooth field and thereby the number of teeth. In addition, Eda activity is absolutely essential for correct tooth morphogenesis. In the absence of Eda the tooth bud is smaller and a progressively more hypoplastic enamel organ develops resulting in small teeth with few cusps in adult mice [Mikkola and Thesleff, 2003]. Eda regulates tooth development via a signaling center called the enamel knot where the expression of Edar is confined. The enamel knot is a cluster of non-dividing cells in the cap stage molar that expresses many of the same signaling molecules as the placode [Thesleff et al., 2001]. In Eda null mice, the enamel knot is smaller and consequently most of the growth factors are expressed in diminished areas [Pispa et al., 1999; Laurikkala et al., 2001]. Strangely, the HED tooth phenotype appears to be less severe in mice than in humans [Nieminen, 2009] which may imply that species-specific differences in pathway utilization exist and/or compensatory pathways may be absent in humans. Of the three molars and one incisor present in each mouse jaw quadrant, the incisor and the third molar may be absent in Eda mutant mice, albeit with variable penetrance, while in humans, often most teeth are missing and, as in mice, the existing ones are abnormal (conical) in shape. Mice have only one tooth generation and therefore it is not currently known whether Eda is specifically involved in tooth replacement but studies on the HED canine model suggest that this may be the case [Casal et al., 2007].
Analysis of Eda loss- and gain-of-function mice has also revealed that Eda regulates hair follicle cycling, which is a life-time process where the lower portion of the hair follicle goes through periods of growth, regression, and rest. Here, Eda appears to regulate specifically the length of the growth phase/onset of regression which is proportional to the length of the hair shaft [Mustonen et al., 2003; Fessing et al., 2006]—less Eda, shorter hairs, more Eda, longer hairs. Currently, the role of Eda in regulating the activity of hair follicle stem cells is unknown.
The Eda pathway is also involved in later stages of tooth and hair development. Edar is expressed in pre-ameloblasts, the precursors of enamel producing cells [Tucker et al., 2000; Laurikkala et al., 2001], and overexpression of Eda or Edar causes enamel hypoplasia suggesting that Eda pathway could regulate the differentiation and/or activity of ameloblasts [Mustonen et al., 2003; Tucker et al., 2004; Pispa et al., 2004]. The fine structure of hair shafts of HED patients and Eda deficient mice indicates a role for Eda in hair filament formation. In mice, Eda has a specific role in production of zigzag hairs that constitute the soft undercoat of mice [Schlake, 2007]. Both loss of Eda [Sundberg, 1994] and its continuous overexpression [Cui et al., 2003; Mustonen et al., 2003; Zhang et al., 2003] prevent formation of the zigzag hair type suggesting a need for a strict temporal and quantitative control of Edar activity to produce morphologically normal hair fibers.
Although the relationship between murine and human hair types is not clear [Schlake, 2007], it has become apparent that the Eda pathway is involved also in common variation of scalp hair morphology in human populations [Fujimoto et al., 2008]. A variant of EDAR with a single amino acid (V370A) substitution has experienced strong positive selection in East Asia prior to 10,000 years ago and is now near fixation [Sabeti et al., 2007 and references therein]. The V370A allele is also common in Native Americans but is practically absent in European and African populations. The amino acid substitution is located in the Edaradd –binding domain of Edar and population-based studies have revealed that this variant is associated with increased hair thickness, the “East Asian” hair type [Fujimoto et al., 2008] apparently due to an enhanced Edar signaling output [Bryk et al., 2008; Mou et al., 2008]. As discussed previously, in mouse models increased Eda activity leads to enlarged organ primordia [Mustonen et al., 2003; Mou et al., 2008] which give rise to widened hair follicles and consequently thicker hair fibers [Mou et al., 2008]. The selective pressures that acted on the V370A allele are not known, but they may be related to the hair phenotype (e.g., better insulation, sexual selection), or hair thickness may simply be a by-product of selection acting on a trait in a different organ (e.g., tooth or gland).
NF-κB IN THE EDA PATHWAY
All evidence available suggests that activation of the transcription factor complex nuclear factor-κB (NF-κB) is integral to Edar activity in vivo and that NF-κB signaling is essential for skin appendage development [for review see Mikkola, 2008]. NF-κB is a generic term for dimeric complexes of Rel homology proteins (five in mammals) [Perkins, 2007]. In unstimulated cells, NF-κB is sequestered in the cytoplasm by inhibitory IκB proteins. Upon stimuli, such as Eda, IκB is phosphorylated by the I-κB kinase (IKK), a protein complex that is composed of two kinase subunits (IKK1/α and IKK2/β) and an obligate regulatory component NEMO (or IKKγ). As a result, IκB is degraded and NF-κB released allowing its translocation into the nucleus and regulation of target gene expression [Perkins, 2007].
The molecular details between Edar engagement by Eda and NF-κB activation have been elucidated to some extent. Edaradd links Edar to downstream pathways via TNFR-associated factors (TRAFs), the common adaptor molecules utilized by different TNFRs [Headon et al., 2001; Yan et al., 2002]. TRAFs, in particular TRAF6 (see below) in turn recruit TGFβ-activated kinase 1(TAK1)-binding protein 2 (TAK2) which bridges TRAF6 to TAK1 leading to its activation and subsequent stimulation of IKK by TAK1 [Morlon et al., 2005] (Fig. 1).
In addition to EDA, EDAR, and EDARADD, several other molecules of the EDA signal transduction pathway have been associated with HED either in man or mice (Fig. 1) [Courtois and Smahi, 2006]. Hypomorphic mutations in NEMO or hypermormorphic ones in IκBα cause HED (as well as immunodeficiency and other defects not related to EDAR signaling) in humans [Zonana et al., 2000; Courtois et al., 2003]. Thus far, none of the TRAFs have been associated with ED or any other congenital malformations in humans, but Traf6 null mice have a similar skin appendage phenotype as Eda null mice, as well as bone and immune defects [Naito et al., 2002]. While mice deficient in a single NF-κB gene do not recapitulate the Eda phenotype, compound RelA; c-Rel null mice have hair and tooth abnormalities [Gugasyan et al., 2004], and remarkably, transgenic mice expressing a degradation resistant IκBα (cIκBαΔN mice) have an ectodermal phenotype strikingly similar to Eda mutant mice [Schmidt-Ullrich et al., 2001].
In line with these findings, analysis of NF-κB activity in mice using transgenic reporter constructs has demonstrated prominent reporter expression in developing skin appendages that is dependent on the presence of intact Eda activity [Schmidt-Ullrich et al., 2006; Pispa et al., 2008]. Moreover, NF-κB reporter expression is lost in tooth, whisker, and mammary buds of Traf6−/− embryos [Dickson et al., 2004] suggesting a tight link between EDA/EDAR, TRAF6, NF-κB and HED. These findings, however, do not rule out the possibility that physiologically relevant Eda signaling may also be mediated by downstream players other than NF-κB.
MOLECULAR TARGETS OF EDA
As NF-κB is thought to have no functions other than as a transcriptional regulator, the effects of Eda can be expected to be mediated via its transcriptional targets. Several different approaches have been taken to identify genes downstream of Eda. These include candidate gene approach [Pispa et al., 1999; Mou et al., 2006; Pummila et al., 2007], comparisons of the transcriptomes of Eda deficient, control, and Eda overexpressing adult [Cui et al., 2002] and embryonic skin [Cui et al., 2006; Cui and Schlessinger, 2006], and analysis of differentially expressed genes in embryonic skin explants after a short treatment with recombinant Eda protein [Fliniaux et al., 2008]. These studies have revealed a number of putative target genes some of which have been verified as true transcriptional targets of Eda.
The correct spacing and size of hair follicles is thought to arise as a result of competition between both placode promoting and inhibiting signals emanating from the placode or the underlying condensed mesenchyme [Jiang et al., 2004; Mikkola and Millar, 2006]. BMPs (in particular BMP4) are generally regarded as inhibitors of placode formation. BMP activity is highest in interfollicular cells [Mou et al., 2006] yet BMPs are expressed in the placode itself (and in the dermal condensate) and thus the activity of BMPs needs to be suppressed in pre-placodal cells to allow them to adopt the hair follicle fate [Mikkola and Millar, 2006]. It now appears that this is at least in part mediated by Eda activity as it regulates the expression of two BMP antagonists, Ccn2/ctgf (connective tissue growth factor) and follistatin [Mou et al., 2006; Pummila et al., 2007]. During placode formation, Eda also strongly induces the expression of Dickkopf-4 (Dkk4), a soluble Wnt inhibitor and a target of Wnts itself [Sick et al., 2006]. This was a rather unexpected finding as Wnt pathway activity is essential for the development of all skin appendages and its abrogation via transgenic expression of the related Dkk1 protein completely blocks hair development [Andl et al., 2002]. Although the physiological relevance of Dkk4 in hair follicle development is yet to be clarified these data may indicate that Eda exerts its function by modulating the expression of both activators and inhibitors of placode formation thereby controlling correct patterning of hair follicles. These findings further highlight the complexity of the molecular interactions that regulate the formation of correct hair follicle density.
Another recently identified target gene of Eda is Sonic hedgehog (Shh) [Schmidt-Ullrich et al., 2006; Pummila et al., 2007] which is essential for proper downgrowth of the hair germ [Mikkola and Millar, 2006]. Accordingly, several genes downstream of Shh are downregulated in Eda null skin [Schmidt-Ullrich et al., 2006; Cui et al., 2006]. In addition, it seems that effects of Eda in hair shaft formation are mediated at least in part by Shh whose correct spatial expression pattern is involved in hair bending [Hammerschmidt and Schlake, 2007]. Shh may play a role in the pathogenesis of the HED tooth defects as well, as Shh transcripts are severely downregulated in Eda mutant teeth [Kangas et al., 2004].
Finally, Eda induces the expression of lymphotoxin-β (LTβ) [Cui et al., 2006], a TNF-like ligand with previously identified roles in the immune system. Analysis of LTβ deficient mice indicates that LT β has no morphogenetic role but is involved in hair shaft differentiation and may partially explain the hair fiber abnormalities associated with HED in mice and men [Cui et al., 2006].
The Eda pathway constitutes an exceptional system of one ligand—one receptor pair while in most developmentally important pathways promiscuity is rather a rule than an exception. For example, a particular FGF or Wnt ligand may bind to multiple receptors and vice versa. This fidelity has substantially increased the usefulness of HED mouse models and has facilitated the elucidation of the developmental role of the Eda pathway. Additional information has also been obtained by introducing novel animal models to study HED such as the chick, fish, and dog [Mikkola, 2008]. Consequently, considerable progress has been made during past few years in understanding how Eda regulates skin appendage development. The function of Eda during placode formation has been analyzed in detail, the important role of NF-κB downstream of Edar has been confirmed, and the first transcriptional target genes have been discovered. Downstream targets of Eda are obvious candidates in search for novel genes behind ectodermal dysplasias. It should be remembered though that these genes are likely to be regulated by other pathways and be part of distinct processes during embryogenesis as well. This is exemplified by Shh which is a novel Eda target gene, yet disturbances in the Shh pathway cause very different congenital malformations (see OMIM 142945 and 147250) compared to HED due to involvement of Shh in a plethora of developmental processes.
Despite these important steps forward there are still several open questions. What other genes are regulated by Eda and which of them are most relevant for the pathogenesis of HED? Are the same target genes of Eda involved in the development of different ectodermal organs? For example, our knowledge of sweat gland development is still very rudimentary. A recent study shows that as in hair follicles, Eda is essential for both initiation and post-initiation stages of sweat gland development [Cui et al., 2009]. And what is upstream of Eda, Edar, and Edaradd? Could those genes be novel candidate genes for ectodermal dysplasias? With the increasing interest in the Eda pathway, many of these questions are likely to be answered in upcoming years.
I thank participants of the International Conference on Ectodermal Dysplasia (Charleston, March 2008) for fruitful discussions.