|Home | About | Journals | Submit | Contact Us | Français|
Hair follicle development and maintenance require precise reciprocal signaling interactions between the epithelium and underlying dermis. Three major developmental signaling pathways, Wnt, Sonic hedgehog, and NF-κB/Edar, are indispensable for this process and, when aberrantly activated, can lead to skin and appendage neoplasms. Recent data point to protein polyubiquitination as playing a central role in regulating the timing, duration, and location of signaling. Here we review how polyubiquitination regulates the stability and interaction of key signaling components that control hair follicle development and regeneration.
Despite their architectural and functional diversity, epithelial organs, and appendages, such as the vertebrate hair follicle, lung, and prostate share common developmental strategies. Each derives from ectodermal or endodermal progenitors and, through the interaction with the underlying mesenchyme, is launched upon a characteristic program of proliferation, migration, apoptosis, and differentiation to form a mature organ (Gumbiner, 1992; Chuong, 1998; Watt and Hogan, 2000; Fuchs et al., 2004; Chuong et al., 2006). Several highly conserved developmental signaling pathways are crucial for appendage development and maintenance. In the hair follicle, the Edar/NF-κB pathway is required at the earliest stages of hair follicle and epidermal appendage development (Headon and Overbeek, 1999; Botchkarev and Fessing, 2005; Mou et al., 2006). In a similar manner, the Wnt/β-catenin pathway initiates morphogenesis and the onset of hair cycling (Huelsken et al., 2001; Van Mater et al., 2003). Finally, the Sonic hedgehog (Shh)/Gli signaling pathway drives hair progenitor proliferation that is required to build a growing hair follicle (Callahan and Oro, 2001; Oro and Higgins, 2003; Levy et al., 2005).
The essential regulatory mechanisms controlling the intensity and duration of each of these signals on a cell-by-cell basis during hair morphogenesis are poorly understood. Abnormalities in the above pathways result in birth defects, inflammation, or cancer predisposition, reinforcing the need to understand how they are controlled. An emerging theme is the central role for protein polyubiquitination in controlling the timing and location of active morphogenic signaling. As with protein phosphorylation, another common post-translational modification of proteins, recent studies have documented myriad functional changes in response to ubiquitination (Reinstein and Ciechanover, 2006). For instance, polyubiquitination is well documented to lead to recognition by the proteasome and destruction of the protein (Rechsteiner, 1987). However, recent data suggest that polyubiquitination can also regulate protein–protein interactions, leading to altered transcription factor activity and subcellular localization of transcriptional cofactors (Dennis and O'Malley, 2005; Dhananjayan et al., 2005). Finally, new experiments suggest that monoubiquitination may have distinct properties from polyubiquitination and has been associated with vesicular trafficking of proteins during receptor endocytosis (Mukhopadhyay and Riezman, 2007). In this review, we will focus on the emergence of polyubiquitination as a critical mechanism to regulate key hair follicle morphogenic and maintenance pathways.
The covalent modification of proteins with ubiquitin has been well studied and is highly specific and strictly controlled by three distinct protein activities termed E1–E3 (Pickart and Eddins, 2004). Ubiquitin is a highly conserved 76 aa protein expressed in all cells. Cellular ubiquitin is first “activated” by transfer to an E1 and subsequently to one of numerous E2 ubiquitin ligases. Specificity of ubiquitination is conferred when an E3 ligase binds to its substrate protein, facilitating transfer of the activated ubiquitin. The ubiquitin ligase can initiate further rounds of ubiquitin ligation, resulting in a “ubiquitin chain” of multiple ubiquitins.
Two major types of polyubiquitin chains have been identified based on which lysine in ubiquitin is used for chain elongation (Figure 1). Multimeric K48-linked chains of ubiquitin are recognized by the proteasome, which specifically degrades ubiquitinated substrates. The proteasome is a barrel-shaped, multi-protein complex, composed of a 20S core, which houses the highly active proteolytic sites and a 19S “lid” that regulates entry of ubiquitinated proteins. The destruction rate of a protein is roughly proportional to the rate it accumulates mature poly-ubiquitin chains. This rate, in turn, is determined by a competition between the polymerization activity of the E3 ligases and the depolymerization by deubiquitinating enzymes (DUBs).
On the other hand, multimeric K63-linked ubiquitin chains, instead of being efficiently shuttled to the proteasome for destruction, facilitate important protein–protein interactions. Studies comparing K48 and K63 linked proteins indicate that K63-linked ubiquitin-modified proteins can still be recognized by the proteasome, but with reduced affinity (Varadan et al., 2004). This allows the ubiquitinated protein to form novel protein complexes not seen without the modification (Wu et al., 2006). As with K48-linked ubiquitin-modified proteins, specific K63-specific DUBs. The length of K63 ubiquitin chains is controlled by (Brummelkamp et al., 2003; Kovalenko et al., 2003; Trompouki et al., 2003).
Substrate selection is a tightly regulated process. Substrate binding by an E3 ligase is a necessary precondition for ubiquitination. The ligase binds sequence motifs on the substrate called degrons. As many ligases and substrates have been characterized, it is clear that each E3 ligase can often only recognize a single type of degron. Further, many degrons require post-translational “priming” modifications, such as phosphorylation, to be recognized by an E3 ligase. This priming step broadens the opportunities for regulating the stability of a substrate protein through the concerted activity of multiple kinases in different signaling pathways. In addition, a given substrate may be acted upon by multiple ubiquitin ligases (and likely DUBs), either simultaneously or individually in unique biological contexts. This allows the cell to integrate information from many simultaneous signals to give precise temporal and spatial control of protein stability or activity. In the hair follicle, many recent reports illustrate how polyubiquitination can precisely control hair follicle proliferation and differentiation. Here we illustrate the general mechanisms and abnormalities associated with three polyubiquitin-controlled signaling pathways in the hair (Table 1).
One of the best examples of regulation by polyubiquitination in the skin is the control of β-catenin protein levels during Wnt signaling. Numerous studies and recent reviews support a critical role for the Wnt/β-catenin pathway in triggering the earliest events of anagen, the active growth phase of the hair follicle. Continued activation of the Wnt/β-catenin pathway maintains follicle growth, whereas reduction in active β-catenin leads to follicular regression (Silva-Vargas et al., 2005). Given the central importance of β-catenin in Wnt signaling, the levels and stability of this protein are tightly regulated by several destruction pathways. Under most conditions, β-catenin performs an integral role at adherens junctions by acting as a bridge between transmembrane cadherin proteins and the cytoskeleton via α-catenin (Weis and Nelson, 2006). Under non-signaling conditions, β-catenin not associated with adherens junctions is rapidly degraded by the proteasome.
At least two distinct destruction pathways control β-catenin steady state levels. The first is the SCF (Skp1-Cullin1-Fbox) E3 ligase consisting of the scaffolding protein Cullin-1 (Cul1) and the substrate-binding protein β-TrCP (Hart et al., 1999). β-TrCP only recognizes β-catenin that is phosphorylated by Glycogen synthase kinase 3β (GSK3β) at key residues in its N-terminal degron (Salic et al., 2000). This phosphorylation requires the cooperation of the scaffolding proteins Axin and adenomatous polyposis coli protein to bring GSK3β and β-Catenin into close proximity (Hart et al., 1998). The morphogen Wnt blocks β-TrCP mediated ubiquitination via inhibition of GSK3β and a cessation of the ubiquitination reaction. This, in turn, leads to the stabilization and accumulation of β-Catenin, its translocation to the nucleus and activation of Wnt target genes.
A distinct Wnt and β-TrCP-independent destruction mechanism has recently been identified that controls β-catenin levels during cell-cycle regulation. In response to p53 induction, β-catenin levels are decreased by the proteasome using a different E3 ligase known as seven in absentia homolog (Liu et al., 2001; Matsuzawa and Reed, 2001). This ubiquitination event does not require the activity of GSK3β or Wnt, but requires an adapter protein ebi-1. Together, seven in absentia homolog/ebi-1 is thought to maintain low basal levels of the substrate upon p53 activation. Interestingly, a priming kinase such as GSK3β has yet to be found for ebi-1, suggesting that the regulation of destruction complex is mediated by transcriptional upregulation of the ubiquitin ligase components rather than post-translational modification of the β-catenin substrate itself.
Genetic loss of key components of the ubiquitination pathway such as Axin or adenomatous polyposis coli or point mutations at key degron residues in β-Catenin are documented in sporadic and inherited forms of colon cancer as well as hair follicle tumors. Strong nuclear localization of β-catenin is also characteristic of benign pilomatricomas and malignant pilomatrix carcinomas. Sequencing of β-catenin from these hyperproliferative lesions reveals that 2−100% of benign lesions and 100% of malignant lesions harbor stabilizing mutations in β-catenin (Chan et al., 1999; Kajino et al., 2001; Moreno-Bueno et al., 2001; Lazar et al., 2005). These point mutations disrupt the integrity of the N-terminal degron and prevent β-catenin's proteasome-dependent degradation.
Polyubiquitination also plays a critical role in regulating the Shh/Gli pathway during hair morphogenesis. Following the β-catenin-dependent induction of anagen, bulge stem cells proliferate and expand hair follicle-specific progenitor cells in the outer root sheath, a process dependent on Shh. Inhibition of the Shh pathway during embryonic development arrests the follicle at the hair bud stage after placode formation and initial ingrowth (St-Jacques et al., 1998). Furthermore, inhibition of Shh in adult skin prevents anagen hair growth (Wang et al., 2000). Shh hedgehog signaling is mediated in the nucleus by the Gli family of transcription factors. Three Gli family members are present in mammals, with Gli1 and Gli2 acting as transcriptional activators and Gli3 acting almost exclusively as a transcriptional repressor. The sum activity of the activators and repressors determines pathway output. Shh binding to its receptor patched alters the balance of transcriptional activity away from repression and toward transcriptional activation.
Both repressor formation and activator stability intimately depend on ubiquitination and the proteasome (Figure 2). In the absence of Shh, Gli3 repressor formation occurs via proteasome-dependent processing. Repressor formation is dependent upon the concerted activity of at least three kinases: protein kinase A, casein kinase Iε, and GSK3β. Extensive sequential phosphorylation by these kinases on the C-terminus of Gli3 creates an atypical degron, which is recognized and bound by β-TrCP (Tempé et al., 2006). As in the case of β-catenin, binding of β-TrCP leads to ubiquitination of the substrate. Unlike β-catenin, recognition by β-TrCP does not appear to require adenomatous polyposis coli or Axin. Also unlike β-Catenin, Gli3 is not fully degraded but instead is specifically digested by the proteasome (Wang and Li, 2006). The mechanism of partial proteolysis is poorly understood, but results in degradation of the C-terminal activation domain to form a potent transcriptional repressor. However, the end result is similar to Wnt/β-catenin, as activation of the pathway by Shh blocks substrate phosphorylation, inhibits β-TrCP ligase binding, ubiquitination, proteolysis, and repressor formation. This results in the de-repression of Shh target genes.
Although the ubiquitin-proteasome system facilitates Gli3 repressor formation, it also effects the complete degradation of the Gli1 and Gli2 activators. Gli2 appears to be phosphorylated in a manner very similar to Gli3. Pharmacological or genetic inhibition of this C-terminal phosphorylation stabilizes Gli2 (Pan et al., 2006). Gli1 lacks a full complement of phosphorylation sites and therefore may not be regulated in the same way (Huntzicker et al., 2006). However, Gli2 and Gli1 both contain a highly conserved degron that is independent of the C-terminal phosphorylation sites. This degron resembles the canonical β-TrCP-binding site but differs by lacking the critical second serine phosphorylation site. Despite its atypical sequence, this motif is required to mediate binding of Gli1 and Gli2 to β-TrCP (Bhatia et al., 2006; Huntzicker et al., 2006) and its removal results in delayed destruction kinetics and Gli protein stabilization.
Recent data have identified additional regulatory pathways controlling Gli levels Structure-function studies of Gli1 stability led to the identification of a highly conserved degron in the N-terminus of Gli1, although the ligase binding to this site was not determined (Huntzicker et al., 2006). Further, additional ligases that regulate Gli levels have also been found. Data from Drosophila suggest that the MATH (Meprin and TRAF-C Homology) and BTB/POZ (Bric-a-brac, Tramtrack, Broad complex/Polio virus and Zinc finger) domain-containing protein Speckle type Poz Protein binds to the fly Gli homolog cubitus interrupts (Ci) and regulates protein levels. Speckle type Poz Protein has been shown to bind the N- and C-termini of Ci and has also been shown to interact with vertebrate Gli proteins (Kent et al., 2006; Zhang et al., 2006b). The relationship of the degrons and the ligases and their significance remains to be shown.
As Shh plays a critical role in both development and tumor formation, it is not too surprising that alterations in repressor and or activator stability might have serious consequences. Constitutive Gli3 repressor formation is seen in people with a rare developmental abnormality called Pallister–Hall Syndrome. In this syndrome, the activation domain normally removed by the ubiquitin-proteasome system is lost through truncating or frame-shift mutations. This results in constitutive repressor formation and prevents activation of the pathway, leading to limb, neural tube, and hair follicle abnormalities. In contrast to the Wnt pathway where β-catenin mutations are relatively common, mutations that stabilize Gli1 or Gli2 have not been found in human tumor samples, although they are just now being examined. In animal models of Basal cell carcinoma formation in which the Shh pathway is activated, removal of either the β-TrCP degron or the N-terminal conserved region (or both) partially stabilized the Gli1 protein relative to unmutated, wild-type Gli1 protein. Tumors develop 2−4 weeks earlier in mice expressing mutant Gli1 than in those expressing the wild-type protein, demonstrating that the proteasome-dependent destruction of Gli1 is an important means of controlling Shh pathway activity in the skin (Huntzicker et al., 2006).
The NF-κB signaling pathway is a multicomponent pathway that regulates the expression of hundreds of genes involved in many key cellular and organismal processes, including cell proliferation, cell survival, the cellular stress response, innate immunity, inflammation, and ectodermal appendage differentiation. Although NF-κB operates in numerous pathways, in the hair follicle, it is notable for mediating the transcriptional effects of ectodysplasin receptor signaling during hair morphogenesis (Botchkarev and Fessing, 2005). As with other pathways, polyubiquitination regulates NF-κB signaling at almost every step of the signal transduction process. Pathway activation is ultimately determined by the translocation of heterodimeric NF-κB transcription factors to the nucleus. Under non-signaling conditions, this translocation is prohibited by cytoplasmic sequestration of NF-κB proteins by IκB proteins. Pathway activation results in the phosphorylation, ubiquitination by β-TrCP, and destruction of IκB proteins, thus releasing NF-κB for nuclear translocation (Karin and Ben-Neriah, 2000; Ghosh and Karin, 2002).
Overexpression of constitutively active forms of NF-κB leads to cell-cycle arrest in keratinocytes and epidermal hypoplasia. Conversely, overexpression of non-destructable IκB results in NF-κB inactivation, subsequent hyperplasia, and sensitivity to Ras-induced neoplasia (Dajee et al., 2003). The kinase complex responsible for initiating phosphorylation of the IκB proteins is itself regulated by ubiquitination. Assembly of the kinase complex requires the auto-ubiquitination of the tumor necrosis factor-α receptor associated factor 6 (TRAF6). This ubiquitination results in the formation of K63 poly-ubiquitin chains. These poly-ubiquitin chains, as opposed to K48 chains do not lead to proteasome-dependent destruction. This auto-ubiquitination of TRAF6 is opposed by the action of the DUB CYLD, whose activity is inhibited upon pathway activation. Loss of CYLD leads to constitutive pathway activation, characterized by constitutive nuclear localization of NF-κB proteins (Brummelkamp et al., 2003; Kovalenko et al., 2003; Trompouki et al., 2003). Paradoxically, instead of leading to keratinocyte cell-cycle arrest and epidermal hypoplasia, genetic loss of CYLD in humans results in a predisposition to developing cylindromas and other tumors of hair-follicle origin (Bignell et al., 2000; Takahashi et al., 2000). This phenotype cannot be reproduced in mice; however, genetic loss of TRAF6 and upstream signaling components of the Edar-signaling pathway leads to the loss of certain hair types (Cui et al., 2003). Similar mutations in humans lead to multiple defects in all ectodermal appendages.
In this review, we have highlighted recent advances in our understanding of how polyubiquitinated proteins contribute to three critical pathways during hair follicle morphogenesis. Recent data suggest polyubiquitination plays a much broader role in regulating protein function than was previously thought. In addition to simple destruction by the proteasome, polyubiquitination can have several effects including partial proteolysis of functional domains, activation, and scaffolding for the association of protein complexes. Further, with the identification of monoubiquitination in the trafficking of receptors, the functional role of protein ubiquitination in signaling has gained great importance.
This review focuses on E3 ligases that polymerize the ubiquitin chain. However, intense experimentation is now focused on identifying the various classes of DUBs, the enzymes that depolymerize ubiquitin chains. In other tissues, critical roles for removing ubiquitin chains have been documented for the DNA damage response (Zhang et al., 2006a), proliferation, and apoptosis (Li et al., 2005). Closer examination of the functions of these proteins may reveal key functions in regulating hair morphogenic signaling.
The inherent modularity of protein destruction pathways suggests the opportunity for coordinate regulation. As seen in Table 1, GSK3β priming and β-TrCP binding are shared between the Shh and Wnt pathways. This suggests the intriguing possibility of the ability to regulate β-catenin and Gli3 processing simply by regulating GSK3β activity, β-TrCP levels, or DUB activity. Examples from other signaling pathways not mentioned here include the E3 ligase Itch-Numb that, while previously shown to ubiquitinate the hair differentiation regulator Notch (McGill and McGlade, 2003), now appears to also control levels of the Gli proteins (Di Marcotullio et al., 2006). Similarly, Smurf2, previously shown to regulate the BMP pathway in complex with Smad7 (Lo and Massague, 1999; Kavsak et al., 2000), also appears to regulate β-catenin levels (Han et al., 2006). These results suggest that external signals such as cell-cycle regulation or growth factors may coordinately regulate the stability or interaction of several pathways.
Appendage progenitor cells must make minute-to-minute developmental decisions in order to faithfully replicate a particular organ. Previous studies with protein phosphorylation have revealed a host of functions that can be regulated. The study of other modifications, including polyubiquitination, acetylation, and methylation, are just in their infancy, but are likely to illuminate how the pathways controlling hair follicle morphogenesis can be tightly regulated.
This work is supported by NIAMS (2R01AR046786, AEO) and The California Institute for Regenerative Medicine (E.G.H.).
CONFLICT OF INTEREST The authors state no conflict of interest.