Wnt Signalling Pathways in Skin Development and Epidermal Stem Cells

Abstract

Mammalian skin and its appendages constitute the integumentary system forming a barrier between the organism and its environment. During development, skin epidermal cells divide rapidly and stratify into a multilayered epithelium, as well as invaginate downward in the underlying mesenchyme to form hair follicles (HFs). In postnatal skin, the interfollicular epidermal (IFE) cells continuously proliferate and differentiate while HFs undergo cycles of regeneration. Epidermal regeneration is fueled by epidermal stem cells (SCs) located in the basal layer of the IFE and the outer layer of the bulge in the HF. Epidermal development and SC behaviour are mainly regulated by various extrinsic cues, among which Wnt-dependent signalling pathways play crucial roles. This review not only summarizes the current knowledge of Wnt signalling pathways in the regulation of skin development and governance of SCs during tissue homeostasis, but also discusses the potential crosstalk of Wnt signalling with other pathways involved in these processes.

Abbreviations

IEF, interfollicular epidermis;

HF, hair follicle;

HFSC, hair follicle stem cell;

E, embryonic day;

P, postnatal day;

DPP, day post-plucking;

DPG: day post-grafting;

sHG, secondary hair germ

Introduction

Secreted Wnt proteins can stimulate multiple intracellular signalling pathways and act as growth factors that regulate diverse processes, including cell proliferation, differentiation, migration and polarity [1-3]. Among Wnt-stimulated pathways, Wnt/β-catenin signalling is known as an important regulatory pathway that governs developmental processes and fate choices during tissue morphogenesis [4-6]. Deregulation of Wnt/β-catenin signalling has been linked to several human diseases and cancers [5, 7]. In addition to β-catenin-dependent Wnt signalling, Wnt-activated signalling pathways that do not depend on β-catenin are referred as non-canonical Wnt pathways and also play divergent roles in development and cancer [8, 9]. While the role of Wnt/β-catenin signalling has been extensively characterized in many different tissues, studies of non-canonical Wnt pathways so far has mainly stressed their ability of inhibiting Wnt/β-catenin signalling.

In this review, we focus on the cellular processes of skin development and homeostasis to point out the spatial and temporal interconnections of Wnt-dependent signalling pathways. Using epidermal regeneration as an example, we draw attention to the controversial question whether Wnt/β-catenin signalling is essential for stem cell maintenance. Lastly, we discuss the potential crosstalk of Wnt signalling with other pathways that co-orchestrate skin development and stem cell activation.

The Wnt Signalling Pathways

Wnt signalling is one of the major cues directing skin development and maintenance [1, 3]. To drive these cellular activities, Wnt ligand-receptor interactions initiate signalling cascades that can be divided into canonical (β-catenin-dependent) and non-canonical (β-catenin-independent) Wnt signalling. The detailed description of Wnt signalling pathways has been extensively covered elsewhere [3, 4, 10, 11], and we only briefly summarize those pathways below.

Wnt signalling is activated when a Wnt ligand binds to receptor/co-receptor. Wnt proteins are encoded by a family of genes highly conserved across the animal kingdom, and so far 19 Wnt genes have been identified in mouse and human genomes [3]. Post-translational modifications of the Wnt proteins, including glycosylation and palmitoylation, are critical for the secretion and the binding to receptors [12, 13]. Some synthesized Wnt proteins are glycosylated and palmitoylated in the endoplasmic reticulum by the actions of porcupine [13, 14], and transferred into the Golgi, where they are complexed to Wntless (Wls), a transmembrane protein that is essential for Wnt secretion [15, 16]. Wls binds to Wnt proteins and escorts them from the Golgi to the plasma membrane, and then released Wnt proteins bind to target cells in an autocrine or paracrine fashion (Figure 1). Secreted Wnt ligands interact with receptor complexes consisting of the Frizzled (Fz) family receptors and/or co-receptors, such as low density lipoprotein receptor-related protein (LRP)-5/6, receptor tyrosine kinase-like orphan receptor 2 (ROR2), or receptor-like tyrosine kinase (RYK), to activate diverse signalling pathways [17-21]. The interaction of Wnt proteins with their cognate-receptors can be blocked by a number of secreted soluble regulators, including Dickkopf proteins (Dkk), secreted frizzled-related proteins (SFRP), or Wnt inhibitory factor (WIF) [22-25]. Furthermore, receptor-ligand interactions are regulated by modulation of receptor abundance via Rspondins and leucine-rich repeat-containing G-protein coupled receptor (LGR) proteins [10, 26].

Among the Wnt-induced pathways, the β-catenin-dependent canonical Wnt signalling pathway appears to be the most conserved in both vertebrates and invertebrates [27]. The central output of the canonical Wnt (referred as Wnt/β-catenin) pathway is the stabilization of β-catenin, a cytoplasmic/nuclear protein that plays a dual role in adherens junctions and transcriptional regulation [11, 28] (Figure 1). In the absence of Wnt stimulation, free cytoplasmic β-catenin proteins are phosphorylated by a destructive complex, consisting of Axin, casein kinase 1α (CK1α), adenomatous polyposis coli (APC) and glycogen synthase kinase 3β (GSK3β). Phosphorylated β-catenin is then ubiquitinated by a β-transducing repeat-containing protein (β-TrCP) for proteasome-dependent degradation [29, 30].

Once the Wnt ligand binds to the receptor/co-receptor complex, the Dishevelled (Dvl) protein is recruited, leading to the inhibition of the degradation complex, thereby stabilizing β-catenin [31, 32]. Stabilized β-catenin accumulates in the cytoplasm and then enters the nucleus where it acts as transcriptional co-activator for the transcription factor family of lymphoid-enhancing factor/T-cell factor (LEF/TCF) proteins [1, 28, 33]. The β-catenin and LEF/TCF interaction engages transcriptional regulators and histone modifiers, which in turn mediates a plethora of developmental and homeostatic processes.

The β-catenin-independent Wnt signalling can be triggered when the Wnt ligand binds to Fz receptor and the co-receptors, ROR2 or RYK. These non-canonical Wnt signalling cascades can be categorized into Wnt/Calcium (Ca²⁺) and Wnt/Planar cell polarity (PCP) signalling pathways [34-37] (Figure 1). In Wnt/Ca²⁺ signalling, the Wnt ligand-receptor interaction leads to release of  intracellular calcium, which in turn acts as a secondary messenger to activate calmodulin-dependent protein kinase II (CaMKII), calcineurin (CaN) or protein kinase C (PKC) [8, 38]. Among these activated effectors, CaMKII further triggers phosphorylation of TGF-β activated kinase 1 (TAK1), which in turn elevates the activity of nemo-like kinase (NLK), leading to the LEF1–β-catenin/DNA dissociation, thereby inhibiting transcriptional activity of Wnt/β-catenin signalling [39]. In parallel, activated CaN induces nuclear translocation of its downstream effectors, nuclear factor of activated T-cells (NFAT) family proteins, which act as transcriptional regulators [40]. In parallel, NFAT’s transcriptional activity can also be reinforced by mitogen-activated protein kinase p38 upon Wnt stimulation [41].

In addition to the Ca²⁺-mediated signalling cascades, induction of the non-canonical Wnt signalling also activates Rho family small GTPases, including Cdc42, Rac, and RhoA, via the recruitment to the receptor/co-receptor/Dvl complex [42, 43] (Figure 1). Activated Rac and Cdc42 initiate downstream c-Jun N-terminal kinase (JNK) signalling, which leads to transcriptional activation of the activating protein-1 (AP-1) complex, and orchestrates PCP by regulating actin cytoskeleton organization and cell migration [44-46]. Other than Rac and Cdc42, RhoA activation downstream of non-canonical Wnt signalling requires binding of Daam1 to Dvl. The activation of RhoA switches on the activity of the downstream effector, Rho-associated kinase (ROCK), which leads to cytoskeletal rearrangement, thereby regulating PCP and cell migration [47-49].

There are several different Wnt proteins and known receptors in vertebrates, and their expression is spatially and temporally regulated during development. The responses to a given stimulus depend not only on which Wnt is present, but also on which cognate receptor is expressed on the cell. It is likely that one Wnt protein activates a combination of multiple signalling cascades that might act independently or collaboratively. Adding more layers of complexity, cofactors, secreted antagonists and co-receptors of Wnt signalling modulate both canonical and non-canonical actions. It is known that non-canonical Wnt pathways have the ability to inhibit β-catenin-dependent signalling; however, the mechanism underlying this inhibition remains unclear. Proposed explanations include the competition of Wnt ligands for the binding to Fz receptors [37], downregulation of β-catenin by the E3 ubiquitin ligase Siah2 [50], or the inhibition of Wnt/β-catenin transcriptional activity via TAK1-NLK-mediated phosphorylation of TCFs [51, 52].

Wnt signalling in Epidermal Stratification and Hair Follicle Morphogenesis

Mammalian skin is constituted of three primary layers, the epidermis, the dermis and the hypodermis. The epidermis and its derivative appendages, such as hair follicles (HFs), sebaceous glands and sweat glands, function together as a physical barrier that protects the organism from environmental stresses, such as dehydration, irradiation, and pathogenic infection. The underlying dermis contains blood vessels, which nourish the avascular epidermis, and protein fibers that enhance skin mechanical strength. This dermis lays on the hypodermis which is an adipose tissue providing thermal insulation and energy resource. Wnt signalling pathways have been shown to play crucial roles in epidermal development, HF morphogenesis and regeneration, which will be discussed in depth below.

The Stratification of the Developing Epidermis

At the early stages of skin development, dynamic signalling crosstalk occurs between the embryonic epidermis and the dermis. The inter-tissue communication instructs basement membrane formation, stratification of the epidermis and HF induction [53]. During embryogenesis, the neuroectoderm layer of the embryo gives rise to the nervous system and the skin epithelium. The specification of the ectodermal cells into the epidermal fate is driven by Wnt signalling, which inhibits the response of ectoderm to fibroblast growth factors (FGFs). In the absence of FGF signalling, ectodermal cells express bone morphogenetic proteins (BMPs) and start to acquire an epidermal fate [54-56].

Once ectodermal cells commit to an epidermal fate, they differentiate into keratin (K)-expressing cells, namely keratinocytes, and form a basal layer of embryonic epidermis (Figure 2). Keratinocytes in the newly formed embryonic basal layer replace the expression of K8/K18 with K5/K14 [57]. At the initial step of epidermal stratification, the epidermal basal cells give rise to a transient layer of endodermis-like cells, called periderm, which protects epidermal basal cells from constant exposure to the amniotic fluid [58]. The next layer of the epidermis generated between the basal layer and the periderm is called intermediate layer, the development of which is associated with asymmetric cell division of the epidermal basal cells [59]. The intermediate layer cells initially undergo proliferation and then mature into spinous cells expressing K1/K10. The resulting spinous cells subsequently further mature into Involucrin-positive granular cells, which terminally differentiate into Filaggrin- and Loricrin- expressing cornified cells [60-62]. The cornified cells finally form the cornified envelop and fulfill the barrier function of the skin. When the stratification program is completed, the epidermis is composed of an inner layer of basal cells with proliferative potential and suprabasal layers of differentiated cells.

At a molecular level, the epidermal stratification process is orchestrated by several transcriptional regulators and signalling pathways (reviewed in [61, 63]). Although Wnt signalling has been mainly implicated in HF induction during skin development, it has also been recently shown to regulate epidermal stratification. Using a genetic mouse model disrupting Wnt production in the developing epidermal basal cells, the authors found that epidermal Wnt production triggers Wnt-dependent activation of a BMP-FGF signalling cascade in the underlying mesenchyme, and that this mesenchymal activation is essential for feedback regulations in the epidermis to modulate the spinous layer formation [64]. Furthermore, in primary human keratinocytes, Wnt5a acts as an autocrine stimulus to promote extracellular calcium-induced keratinocyte differentiation by coupling with Wnt/β-catenin pathway [65]. Although the underlying molecular networks are not fully understood, extracellular calcium-induced keratinocyte differentiation could be in part regulated by Wnt.

Hair Follicle Formation and Maturation

Hair follicle morphogenesis includes three main stages: hair placode formation, hair follicle organogenesis, and cytodifferentiation [66, 67]. The sequential events during morphogenesis are tightly regulated by signals transmitted between the dermis and the epidermis. Among those signals, Wnt signalling pathway is considered to be the master regulator.

Before any visible hair placode formation, dermal fibroblasts uniformly receive a widespread epidermis-derived Wnt signal. In response, dermal fibroblasts then produce the first dermal Wnt signal that induces aggregation of epidermal basal cells at regularly spaced intervals, leading to the formation of hair placodes (Figure 2). Notably, the pattern of the HF induction is dependent on a competition between slowly diffusible Wnt ligands and faster diffusing Wnt inhibitors [66, 68]. After the initial induction, the developing placode produces Wnt ligands to induce underlying fibroblasts to form a dermal condensate [69, 70]. Concomitantly, the placode continuously grows, invaginates into the underlying dermis, and then joins the dermal condensation to form the first structure of the HF organogenesis, the primary hair germ (HG). The epidermal cells continue to penetrate the forming dermis and generate a multi-layered and elongated column, called the hair peg. Meanwhile, the dermal condensate becomes a spherical dermal papilla (DP). The hair peg thickens at the lower end to form a hair bulb and half of which encloses the elongated dermal papilla. When the hair germ grows into a hair peg, differentiated epidermal layers that will give rise to the hair shaft become visible. As soon as HF down-growth reaches the subcutis, the program of cytodifferentiation initiates. At this point, the DP first becomes thinner and gets totally enclosed; then the sebaceous gland starts forming at the upper part of the HF. Finally, the fully formed hair shaft protrudes from the skin surface and the hair follicle reaches its maximal length. [66].

Wnt/β-catenin signalling is first upregulated uniformly in the upper dermis and then focally in both the hair placode and the underlying dermal condensate, where Lef1 is expressed [69, 71, 72]. When Wnt/β-catenin signalling is turned off either by overexpressing Dkk1 in the skin, by conditional ablation of β-catenin or by transgenic expression of a truncated form of Lef1, hair follicle formation is blocked [73-75]. Consistently, overexpression of a stable form of β-catenin or Lef1 induces de novo HF formation [72, 76, 77]. These findings suggest that the levels of Wnt/β-catenin signalling in the developing epidermis determine the specificity toward hair follicle lineage. Further genetic investigations revealing Wnt signalling effects on the development of skin epidermis and hair follicles are summarized in Table 1.

The development of DP occurring at early HF morphogenetic stages is also dependent on Wnt, and more precisely on Wnt5a [78]. Typically, Wnt5a is expressed in the developing dermis and highly enriched in the dermal condensate. Given that the dermal Wnt5a expression is completely abolished in sonic hedgehog (Shh) mutant skin, the dermal Wnt5a is considered as a target of Shh signalling activated by Wnt/β-catenin signalling in the hair placode. Although Wnt5a deletion in the skin does not lead to detectable defects in HF morphogenesis, Wnt5a-deficient DP cells lose HF-inductive ability in postnatal skin [79], underscoring the role of Wnt5a as an essential dermal signal.

In embryonic murine skin, PCP is established within the entire epidermal plane in an anterior to posterior direction [70]. Over-activation of Wnt/β-catenin signalling in the developing epidermis can disrupt the polarity of hair follicle orientation, and this is likely caused by perturbing the asymmetrical expression of Shh in the developing HF [72, 76]. Non-canonical Wnt pathways can also direct PCP via a Wnt-Fz-Dvl circuitry to activate Rho-ROCK and/or Rac-dependent signalling cascades. A Wnt receptor and PCP protein, Fz6, is asymmetrically localized in the embryonic epidermis and controls HF orientation in mice [80-82]. Although it remains unclear whether Wnt morphogens are needed in the establishment of hair patterning, Wnt/PCP signalling, known to regulate cytoskeletal rearrangement and cell migration, likely serves as a spatial signal to modulate HF orientation during morphogenesis.

Wnt Signalling in Skin Epidermal Stem Cells

Throughout life the skin epidermis is regularly renewed. Skin epidermal SCs, capable of self-renewal and differentiation, provide unlimited sources of cells to maintain tissue homeostasis, as well as to regenerate HFs and repair the epidermis after injury. Skin epidermal SCs are located in the basal layer of the interfollicular epidermis, and also in the bulge region of the hair follicle [83]. While adult IFE basal cells sustain continual proliferation, the HFs undergo cycles of degeneration and regeneration. Both processes require the activation of SCs, which is largely modulated by signalling cues from their microenvironment. Among these signals, Wnt-dependent signalling plays crucial roles in the maintenance, activation and fate determination of the SC populations.

Stem Cells in the Interfollicular Epidermis

The postnatal epidermis is continuously regenerated by the proliferative basal cells of the IFE. This IFE pool gives rise to progenies that will differentiate into suprabasal cells while migrating upwards. During this process, the epidermal cell number remains constant as the number of newly generated cells exactly compensates for the number of cells that differentiate or die [53]. Two models, the hierarchical and the stochastic model, have been proposed to explain how SCs in the basal layer act to replenish the IFE [83]. The hierarchical model suggests that a slow-cycling SC located in each epidermal proliferative unit of the IFE generates short-lived transit-amplifying cells (TACs), which then give rise to differentiated cells. The stochastic model was first proposed by Jones group that the progenitors in the basal layer of the IFE have equal potential to generate daughter cells which remain as progenitors or differentiate into suprabasal cells [84].

Using different experimental approaches, evidence for both the hierarchical and the stochastic model have been presented. On the one hand, IFE lineage tracing experiments support the hierarchical model and show the existence of two proliferative cell populations representing slow-cycling SCs and committed progenitors produced in hierarchical manner. Upon wounding, only SCs are capable of responding to proliferative stimuli and contributing to long-term regeneration of the tissue [85]. In accordance with this finding, a recent study further shows that epidermal label-retaining cells (LRCs) and non-LRCs constitute two SC populations of the IFE. Although these two SC populations are molecularly distinct and produce unique differentiated lineages during homeostasis, they are functionally interchangeable and able to replenish each other’s territories upon wounding [86]. On the other hand, using live imaging techniques, Rompolas and colleagues support the stochastic model and demonstrate that IFE basal cells are born as uncommitted SCs which have equal potential to proliferate or undergo differentiation [87]. In agreement, a recent study using multicolor lineage tracing shows that the IFE can be maintained without a hierarchy of SCs and TACs [88]. The tools used to mark basal cells and examined epidermal regions at different body sites are likely to account for the differences among these studies. Despite the growing evidence from various studies, the existence of a hierarchy between slow-cycling SCs and committed TACs in the IFE remains as an open question.

In the developing epidermis, a high level of Wnt/β-catenin signalling is essential for HF induction. The constitutive attenuation of epidermal Wnt/β-catenin signalling impairs HF formation, but does not impact IFE integrity [73-75, 89, 90]. Notably, K14Cre-driven β-catenin depletion in developing IFE even causes epidermal hyperproliferation [73]. Nevertheless, lineage tracing studies with an inducible Cre recombinase gene driven by Wnt reporter Axin2 (Axin2-Cre) show the presence of Wnt/β-catenin signalling activity in the basal cells of non-hairy epidermis, and that Axin2-marked basal cells are long-term IFE progenitors. When β-catenin is depleted in these cells, the epidermis exhibits severe hypoproliferation [91, 92] (Table 1). The disparate results in epidermal proliferation could be partially explained by the fundamental differences between hairy and non-hairy epidermis because it is thought that epidermal hyperproliferation of hairy skin could partially result from inflammatory response to the HF disintegration [91, 93]. However, such hyperproliferation caused by HF loss is not found in epidermis expressing dominant-negative LEF1 or Wnt inhibitor DKK1 [74, 75]. In addition, given that β-catenin-null SCs are still capable of generating epidermis when chamber grafted [94], it remains unsolved whether Wnt/β-catenin signalling is necessary for the generation and maintenance of IFE.

Hair Follicle Stem Cells

In contrast to the epidermis which continuously regenerates, mature HFs progress through cycles ofgrowth (anagen), degeneration (catagen), and rest (telogen) all through life (Figure 3). During the hair cycle, the upper portion of the HF is permanent while the lower portion undergoes degeneration and then regeneration by the stem cells. Stem cells in the HF can be divided into two populations: one population residing in the outer layer of the bulge, known as bulge hair follicle stem cells (HFSCs); the other population located within the secondary hair germ (sHG) right below the bulge [83]. During telogen, bulge HFSCs remain in quiescence while SC progenies in the sHG are primed, which makes them become the first group of cells getting activated at the anagen onset [95, 96]. At early anagen, bulge HFSCs begin to proliferate and give rise to the outer root sheath (ORS), a population of cells migrating downward from the bulge and feeding into the bulb of the matrix at the base of the newly formed HF, resulting in follicle growth [95, 97]. On the contrary, the sHG develops into the matrix, composed of the pool of TACs that rapidly proliferate and then terminally differentiate to form hair shaft and the inner root sheath (IRS) [97, 98]. As soon as the TACs and the DP move far away from the SC niche, bulge HFSCs return to quiescence [99]. Once reaching catagen, the lower portion of ORS and matrix cells die through apoptosis while upper and middle portions of ORS cells migrate upward to form a new bulge and a new hair germ, respectively [97].

Bulge HFSCs were initially identified as slow-cycling LRCs in the postnatal mouse skin [100, 101]. Recent studies with lineage tracing and live-imaging experiments postulate the early specification of HFSCs during skin development [102, 103]. Xu and colleagues discovered that HFSCs are derived from the upper region of hair placode/hair germ cells which display attenuated Wnt/β-catenin signalling. Elevation of Wnt/β-catenin signalling in embryonic epidermal cells abolishes HFSC specification and suppresses SC marker expression [102]. Furthermore, recent work demonstrated that HFSCs are born from asymmetric divisions of the developing hair bud in which daughter cells differentially display Wnt and Shh signalling [103]. The Wnt-high basal cells maintain slow-cycling and then grow as short-lived progenitors that eventually differentiate; by contrast, the Wnt-low suprabasal cells become SCs and symmetrically expand to form the early HFSC pool [103]. Both studies support the notion that attenuation of Wnt/β-catenin signalling in the early hair bud is a prerequisite for HFSC specification.

In the mature HF, the quiescence and activation of HFSCs are tightly regulated by a balance of BMP and Wnt signals coming from their niche cells [83]. During telogen, the inner bulge K6+ cells secrete high levels of BMP6 and FGF18 [97]. Dermal fibroblasts express BMP4 and subcutaneous mature adipocytes express BMP2 [104]. All of these BMPs serve as inhibitory signals to maintain HFSCs in a quiescence state. Toward the end of telogen, the dermal macro-environment reduces BMP expression. By secreting Noggin, the DP turns BMP signalling off in HFSCs, allowing them to transduce Wnt/β-catenin signalling and thereby promoting anagen onset [95, 104-106]. If the adjacent DPs are ablated, telogen HFs are unable to enter the hair cycle [107, 108], underscoring the necessity of the DP for HFSC activation. In fact, decreased BMP signalling unleashes Wnt signalling activation via upregulation of Wnt ligands and receptors as well as downregulation of Wnt antagonists [109]. Among expressed Wnts, upregulated Wnt7b primes and tips the balance toward Wnt/β-catenin activation and HF growth. Postnatal Wnt7b ablation delays HFSC activation during the telogen-anagen transition [109]. In addition to the DP, adipocyte precursor cells could also indirectly contribute to HFSC activation by modulating DP activity via stimulation of platelet-derived growth factor (PDGF) signalling [110].

At the telogen phase, bulge HFSCs appear to reside in a Wnt-restricted environment as Wnt repressors, such as SFRP1, WIF1 and Dkk3, are highly expressed by HFSCs and the inner bulge K6+ niche cells [78, 95, 97, 100, 111, 112]. In agreement, Wnt/β-catenin signalling reporters, TOPGAL or Axin2-LacZ, are inactive in the telogen HFSCs consistent with the absence of nuclear β-catenin [71, 96, 104]. A recent study reports that some Wnt ligands are expressed by HFSCs and that Axin2, a Wnt/β-catenin signalling target gene, is detectable in the outer bulge compartment throughout telogen, implicating the presence of Wnt/β-catenin signalling in telogen HFSCs [113]. However, it is unclear whether this low level of Wnt/β-catenin activity in quiescent HFSCs is required or functional for HFSC maintenance. Telogen HFSCs depleted for β-catenin or Wls are arrested in a quiescent state, and silent β-catenin- or Wls-null HFSCs can be maintained in their niche for a long time without losing the expression of HFSC markers [91, 96, 114]. Only when arrested β-catenin- or Wls-null HFSCs are stimulated by wounding conditions, e.g. hair depilation, they begin to exhibit the defects in which they fail to differentiate into HF lineage [96, 114]. Such deficiencies are also found in those developing epidermis losing β-catenin, LEF1-β-catenin interaction or Wls, before HFSCs are specified [73, 74, 93, 115]. Taken together, these findings suggest that the low level of Wnt/β-catenin signalling in telogen HFSCs might not be essential to maintain HFSCs as long as they are kept within their native niche; however, the presence of Wnt/β-catenin signalling is critical for activated HFSCs to retain their potency of HF lineage commitment and differentiation.

Signalling Crosstalk of the Wnt pathways in Skin Epidermis

During epidermal development and homeostasis, Wnt signalling pathways function independently or collaboratively with other pathways in order to elicit appropriate cell responses. Some regulatory axes are well studied, and others remain hypothetical and need further investigations. Below we discuss a few selected signalling pathways that cross-interact with Wnt pathways in skin epidermis.

Wnt-BMP signalling crosstalk is a regulatory axis that is required for the development of suprabasal keratinocytes and the cyclical regeneration of the HF. During skin development, epidermal Wnts activate mesenchymal responses by transducing a BMP-FGF signalling cascade in the dermis. Such a Wnt-BMP regulatory loop is crucial for the feedback regulation that controls the stratification processes of developing epidermis [64]. Moreover, at the telogen-anagen transition of the hair cycle, the activity of BMP signalling in HFSCs progressively decreases, which permits the activation of Wnt/β‐catenin signalling [116, 117]. The mechanism by which BMP inhibition regulates canonical Wnt activation promotes HFSC activation and initiates cyclic regeneration of HFs [104, 109].

In early epidermal development, Shh is expressed in the placode of developing hair follicles, and Shh signals can be received by both developing epidermis and underlying dermal condensate. The epidermis expressing stabilized β-catenin displays ectopic HF and Shh expression [76, 118]; conversely, in the epidermis lacking β-catenin, Shh is not expressed [73]. These results indicate that Shh signalling serves as a downstream pathway of Wnt/β-catenin signalling to regulate HF induction. On the contrary, Wnt5a, only expressed in the developing dermal condensate, becomes absent in Shh-null embryos, placing Wnt5a as a downstream target of Shh signalling in HF morphogenesis [78]. Moreover, a recent study uncovers that cell fates of the daughter cells in the developing hair bud are determined by their Wnt signalling activities and responses to Shh stimulation. Wnt-low suprabasal cells, responding to Shh, expand symmetrically and become HFSCs, whereas Wnt-high basal cells, producing but not responding to Shh, grow as short-lived progenitors that eventually differentiate [103]. Collectively, these results demonstrate that Wnt-Shh regulatory axis plays multiple roles in HF induction, morphogenesis and fate choices.

Both Wnt and Notch signalling pathways regulate HF maintenance. The induction of new HFs in adult epidermis by stabilized β-catenin can be inhibited by blocking Notch signalling through Jagged1 deletion or treatment with the γ-secretase inhibitor [119]. This finding places Notch pathway as a downstream pathway of Wnt/β-catenin signalling to determine HF fate. Furthermore, in a culture system when HFSCs are induced to differentiate toward a hair fate, upregulated Lef1 activates the target gene Jagged1, leading to the co-activation of Wnt/β-catenin and Notch signalling pathways [120]. On the contrary, in the stratified IFE, nuclear β-catenin is only detected in some proliferating basal cells while Notch1 is predominately expressed in differentiated suprabasal layer. Depletion of Notch1 causes upregulation of β-catenin-mediated signalling in multiple layers of hyperproliferative IFE, suggesting Notch1 can repress Wnt/β-catenin signalling pathway to restrict its activation to the basal layer [121]. The synergistic and antagonistic effects between Wnt and Notch signalling seem dependent on the cell fate of skin epidermis.

Emerging evidence shows that the Hippo pathway could engage a crosstalk with Wnt/β-catenin signalling to regulate self-renewal of embryonic and intestinal SCs as well as the growth of cancer cells. The transducers of Hippo signalling, YAP/TAZ, act as integral components of the β-catenin destruction complex, and their incorporation orchestrates the Wnt response to regulate SC self-renewal and tissue homeostasis [122, 123]. In cancer cells, downregulation of Hippo signalling correlates with upregulation of β-catenin activity. The proposed mechanism attributed to this correlation is that phosphorylated YAP/TAZ suppresses phosphorylation of Dvl and nuclear translation of β-catenin, thereby inhibiting Wnt/β-catenin signalling [124, 125]. Given that Hippo signalling has been identified as an important regulator of epidermal proliferation [126, 127], the cross-interaction between Hippo and Wnt signalling would be worthwhile further investigation in the regulation of epidermal development and HF regeneration.

Several studies have shown that non-canonical Wnt ligands can inhibit canonical Wnt signalling depending on the cellular context. Ectopic Wnt5a expression causes loss of Wnt/β-catenin signalling and inhibition of hair follicle formation in the developing skin, but gain of Wnt/β-catenin signalling in the meninges [128]. A number of mechanisms for the antagonistic effects of non-canonical Wnt signalling against Wnt/β-catenin signalling have been proposed. For instance, Wnt5a stimulation can lead to increased PKC activity that phosphorylates retinoic acid-related orphan nuclear receptor α (RORα). Phosphorylated RORα binds to β-catenin and forms a transcriptional complex, thereby inhibiting Wnt/β-catenin transcriptional activity [129]. In addition, Wnt5a can also inhibit Wnt3a-induced β-catenin signalling through Ror2 activation [34]. Wnt5a stimulation induces Ror2-Dvl interaction via a CK1-dependent mechanism, and this interaction negatively regulate transcriptional activity of Wnt/β-catenin signalling [130]. Whether similar mechanisms are transduced by non-canonical Wnt ligands during epidermal development and HF regeneration requires further examination.

Conclusion

In this review, we summarized the current knowledge of the Wnt-dependent signalling pathways and their functions in skin development and adult epidermal stem cells. It is perceivable that Wnt signalling is essential at early time points of skin development for epidermal fate specification, HF induction and morphogenesis, and continuously regulates HFSC activation, lineage commitment and differentiation. Although the roles of Wnt signalling in skin epidermis are diverse, it is more and more clear that this diversity is a combined outcome of crosstalk with other signalling inputs in a stage- and context-dependent manner. Multiple Wnt ligands, receptors and differential ligand-receptor compositions add additional layers of complexity to Wnt signalling regulation. Some of Wnt ligands trigger canonical Wnt signalling; others lead to activation of non-canonical Wnt signalling cascades. It remains unclear how cells respond to the stimulation of multiple Wnt ligands in their native microenvironment, and how activated Wnt signalling pathways work collaboratively or antagonistically to orchestrate skin development and HF regeneration. Future investigations should draw more attention to the temporal and spatial signalling cross-interaction required in specific developmental stages and tissue types. It might be worthwhile to investigate which Wnt ligands and receptors are expressed in every cell layers of the epidermis, HFs and their underlying dermal cells. The outputs of Wnt ligand-receptor combinations in individual cell types and their neighboring cells will provide integrative views of signalling crosstalk among Wnt-dependent pathways.

Figure Legends

Figure 1. Wnt signalling pathways. This schematic diagram displays the processes of Wnt secretion, canonical and non-canonical Wnt signalling pathways. Synthesized Wnt protein is palmitoylated by Porcupine in endoplasmic reticulum (ER), and transferred into the Golgi, where Wnt is complexed to Wntless (Wls) for secretion. Wls-bound Wnt protein is then escorted from Golgi to the plasma membrane and released into the extracellular environment where it can bind to target cells. In the absence of Wnt, free cytoplasmic β-catenin is targeted and phosphorylated for its degradation by a destructive complex, composed of the core proteins Axin, CK1α, APC and GSK3β. The subsequent β-catenin degradation is mediated by β-TrCP E3 ubiquitin ligase. Once Wnt ligands bind to receptor Frizzled (Fzd) and co-receptor lipoprotein receptor-related protein 5/6 (LRP5/6), Dishevelled (Dvl) is recruited, leading to the inhibition of the degradation complex. Stabilized β-catenin is accumulated in the cytoplasm and then enters into the nucleus where it acts as a transcriptional co-activator for lymphoid-enhancing factor/T-cell factor (LEF/TCF) transcription factors to activate Wnt-responsive genes. β-Catenin-independent non-canonical Wnt pathways can be categorized into Wnt/Ca²⁺ and Wnt/PCP pathways. Binding of Wnt isoforms to either Fzd or other tyrosine kinase-like receptors, e.g. Ror2, can trigger multiple signalling cascades. In Wnt/Ca²⁺ signalling, Wnt-receptor interaction initiates intracellular calcium release, thereby activating calmodulin-dependent protein kinase II (CaMKII), calcineurin (CaN) or protein kinase C (PKC). Among these effectors, CaMKII triggers TAK1-NLK cascade, which suppresses transcriptional activity of Wnt/β-catenin signalling. In parallel, activated CaN activates nuclear factor of activated T-cells (NFAT) family proteins for transcriptional regulation. Wnt/PCP pathways involve activation of small GTPases Rho, Rac, and Cdc42, and their downstream JNK signalling that regulate cytoskeleton rearrangement and planar cell polarity (PCP). The mechanism underlying β-catenin-independent Wnt signaling may be dependent on the cellular context.

Figure 2. Epidermal stratification and hair follicle formation. During early embryogenesis the surface of the embryo is covered by a single ectodermal layer, called surface ectoderm, which adheres to an underlying basement membrane (BM). Upon Wnt stimulation, ectodermal cells are specified into epidermal fate and form the first keratin-expressing basal layer (BL). At the initial step of epidermal stratification, the epidermal basal cells give rise to a transient layer periderm, which protects basal cells from constant exposure to the amniotic fluid, and then generate intermediate layer via asymmetric cell division. The intermediate layer cells undergo proliferation and then mature into spinous layer (SL), which subsequently develops into granular layer (GL). The granular layer cells in the last step of stratification terminally differentiate into cornified layer (CL) to form cornified envelop and acquire the barrier function. In parallel with epidermal stratification, upon Wnt signal transmission between epidermis and dermis, hair follicles (HFs) start forming from the developing epidermis. Before HFs are induced, dermal fibroblasts uniformly receive a widespread Wnt signal from epidermis. In response, dermal fibroblasts then produced the first dermal Wnt signal to induce aggregation of epidermal basal cells, leading to the formation of hair placode. After the initial induction, the developing placode produces Wnt ligands to induce underlying fibroblasts to form a dermal condensate; at the same time, the placode continuously grows and invaginates into a hair germ (HG). The epidermal cells continue to penetrate the forming dermis and form an elongated column, called hair peg (HP); meanwhile, the dermal condensate becomes a spherical dermal papilla (DP) enclosed by the lower end of HP. Once HG grows into HP, precortical matrix, a differentiated epidermal layer, becomes visible. As soon as HF down-growth reaches the subcutis, the HF starts forming sebaceous gland (SG) at the upper part of HF and DP becomes thinner and gets fully enclosed by the matrix of the HF.

Figure 3. The hair cycle. During the resting phase (Telogen), hair follicle stem cells (HFSCs) residing in the outer layer of bulge remain in quiescence as the inner bulge K6+ cells express high levels of inhibitory signals. At the onset of the growth phase (Anagen), SC progenies located in secondary hair germ (sHG) proliferate and initiate HF regeneration in response to the activating cues produced from the underlying dermal papilla (DP). At early anagen, HFSCs begin to proliferate and give rise to outer root sheath (ORS), a population of cells migrating downward from the bulge and feeding into the bulb of matrix at the base of the newly formed HF. At the same time, sHG develops into matrix, composed of the pool of transient-amplifying cells (TACs) that rapidly proliferate and then terminally differentiate to form hair shaft and inner root sheath (IRS) at later stage of anagen. At Mid-Anagen, matrix progenitors in the pre-cortex region terminally differentiate to form the hair shaft (HS). At the end of anagen when the HF enters a destructive phase (Catagen), the lower part of the HF undergoes apoptosis and the epithelial strand regresses upward. Once the DP is drawn upward towards the bulge/sHG, the HF re-enters telogen. IFE, interfollicular epidermis; SG, sebaceous gland.

Table 1. Summary of mouse models for Wnt signalling pathways in skin epidermis and hair follicles

Target gene	Target tissues	Mutation	Induction	Phenotypes (stages)	References
β-catenin (loss-of-function)	IFE & HFs	K14-Cre; Ctnnb1fl/fl	E9.5	– Placode formation is impaired (E15.5)   – Hair loss and cyst formation (P20) – Hyperproliferative epidermis (P40)	[73]
		K14-CreER;   Ctnnb1fl/fl	P49	– Loss of SC marker – CD34 (P59)   – Fail to generate HF and adapt to sebocyte fate in cysts (P59)	[131]
		K5-rtTA;   TetO-Cre; Ctnnb1fl/fl	P4-60	– Decreased cell proliferation and increased apoptosis of HF matrix (P8)   – Precocious catagen entry (P14) – HF degeneration and loss of HFSCs (P60)	[91]
			P17-30	– HFs fail to enter anagen (P25)
			P50-57+ plucking at P54	– The sHG fails to proliferate (P57; 3DPP)
	HFSCs	K15-CrePR; Ctnnb1fl/fl	P50-54+ plucking at P54	– The sHG fails to proliferate (P59; 5DPP)   – HFs fail to enter anagen (P68; 14DPP)	[91]
			P46-50	– HFs fail to enter anagen (P95)   – HFSCs remain expressing SC makers – CD34, SOX9, TCF4 (P200)	[96]
			P46-50+ plucking at P58	– HFSCs undergo proliferation (P60; 2DPP), but fail to differentiate into HFs (P65; 7DPP)   – HFSCs differentiate into sebocytes (P81; 23 DPP) – Loss of bulge/HFSCs, but with enlarged sebaceous glands (P610, 560DPP)
	Non-hairy IFE   (footpad)	Axin2-CreER; Ctnnb1fl/fl	P21	– Decreased cell proliferation and increased proportion of K10+ differentiated cells in non-hairy IFE (P31)	[92]
		K5-rtTA;   TetO-Cre; Ctnnb1fl/fl	P43-45	– Decreased cell proliferation of non-hairy IFE (P71)	[91]
	Dermal papilla	Cor-Cre; Ctnnb1fl/-	P3 / P7	– Reduced proliferation of HF matrix cells (P10)   – Premature catagen (P16) – Hair shortening and thinning (P60) – Impaired HF regeneration after depilation (P82; 12DPP)	[132]
	Dermal fibroblasts	En1-Cre; Ctnnb1fl/∆	E12.5	– Reduced proliferation of dermal fibroblasts (E14.5)   – Failure in placode initiation and dermal condensate formation (E16.5)	[69]
β-catenin (gain-of-function)	IFE & HFs	K14-∆N87βcat	E9.5 – E14.5	– De novo HF formation from IFE and the upper part of HFs (>P18)   – Trichofolliculoma formation & Pilomatricoma development (Old homozygous mice)	[76]
				– Precocious HFSC activation and hair cycle entry (P55 heterozygous mice)	[131]
		K14-∆N∆Cβcat	E9.5 – E14.5	– Formation of epidermal and sebaceous cysts from HFs (P24-P28)   – Loss of TOPGAL activation in the bulge (P22) – De novo hair-like invagination from IFE (P28)	[133]
		K14- S33Yβcat-ER	P50-64	– Precocious activation and anagen entry (P53)   – Hyperplastic anagen HFs (P64)	[134]
		K14-∆Nβcat   -ER	P42-56	– Precocious activation and anagen entry (P45)   – Abnormal HFs containing cysts (P53) – De novo HF formation from IFE (P50) – Trichofolliculoma formation (P70), but tumors regress after induction stops	[77]
		K14-Cre;   Ctnnb1fl(ex3)/+	E9.5	– Precocious hair follicle induction (E13.5)   – Disrupted HF morphogenesis (E17) – Impaired hair shaft production in grafted skin (21DPG)	[135]
	HFSCs	K19-CreER;   Ctnnb1fl(ex3)/+	P20	– Induction of new hair growths from existing HFs and this induction is independent of DP signals (P32-P39)	[136]
	Dermal fibroblasts	HoxB6CreER;   Ctnnb1fl(ex3)/+	E12.5	– Increased proliferation of dermal fibroblasts (E14.5)   – Increased hair placode size and number (E16.5) – Accelerated differentiation of HF (E17.5)	[69]
LEF1 (loss-of-function)	IFE & HFs	LEF1-/-	Consti-tutive	– Arrest in HF development (P3)   – Reduction in HF number (P3) – Lack of body hairs and whiskers (P16)	[137]
		K14-ΔNLef1	E9.5	– Suppression in hair differentiation and sHG cells adapt sebocyte fate (>P21)	[89]
		K14-ΔNLef1	E9.5	– Hair loss (P42)   – Formation of epithelial cysts (P105) – Development of skin tumors (>P90)	[74]
LEF1 (gain-of-function)	IFE & HFs	K14-Lef1	E9.5	– Irregularly spaced and misoriented HFs (P28)	[72]
TCF (loss-of-function)	IFE & HFs	K14-Cre; Tcf3fl/fl	E9.5	– No apparent epidermal defects	[138]
		K14-myc∆NTcf4	E9.5	– Epidermal differentiation defects and impaired barrier function (P0)
		K14-Cre; Tcf3fl/fl; Tcf4-/-	E9.5	– Initiated HFs could not grow down (P0)   – No hair formation in grafted skin (17DPG) – Failure of long-term IFE maintenance (60DPG)
	HFSCs	K15CrePR; Tcf3fl/fl; Tcf4-/-	30-40   DPG	– HFSC precocious activation (58DPG)   – Precocious entry of hair cycle (65DPG)	[96]
TCF (gain-of-function)	IFE & HFs	K14-Tcf3	E9.5	– Impaired epidermal barrier function (E18.5-P0)   – Epidermal differentiation defects (P0)	[89]
		K14rtTA;   TRE-mycTcf3	P1-5	– Repression of epidermal differentiation and sebaceous gland development (P5)   – Suppression of HF development (d27, 60DPG)	[139]
			P19-22	– HFSCs fail to get activated (P23)	[96]
Wls (loss-of-function)	IFE & HFs	K14-Cre; Wlsfl/fl	E9.5	– Impaired hair placode induction (E14.5)   – Loss of HF formation (E18.5, P7)	[115]
				– Hyperproliferation of IFE basal cells (P16-28)   – Epidermal differentiation and skin barrier defects (P16)	[93]
		K14‐CreER; Wlsfl/fl	P20-27	– Impaired hair cycle entry (P37)   – Hyperplasia of IFE and sebaceous glands (P47) – Defects in HFSC proliferation, but normal for HFSC maintenance (P91) – Failure of depilation-induced HF regeneration (15DPP)	[114]
		K5-rtTA;   TetO-Cre; Wlsfl/fl	P4-18	– Premature regression of HFs (P14)   – External hair loss and formation of cysts (P60) – HFSCs remain expressing SC makers – CD34 and KRT15 (P60)	[91]
	HFSCs	K15-CrePR; Wlsfl/fl	P19-25	– HFSCs remain expressing SC makers – CD34 and KRT15 (P60)	[91]
	Dermal fibroblasts	En1-Cre; Wlsfl/fl	E12.5	– No defects in HF initiation (E14.5)	[69]
Porcupine (loss-of-function)	IFE & HFs	Sox2-Cre; Porcnfl/+	E6.5	– Reduced density of hair placodes (E14.5)   – Partial skin with hair loss (P22)	[90]
DKK1 (gain-of-function)	IFE & HFs	K14-Dkk1	E9.5	– Failure of placode formation (E15.5)   – Lack of HF formation (P16) – No effect in epidermal differentiation (P28)	[75]
		K5-rtTA;   TetO-Dkk1	P21-60	– Decreased HF matrix proliferation (P8)   – Premature HF regression (P14) – HFSCs remain expressing SC makers – CD34 and KRT15 (P24)	[91]
			P50-57+ plucking at P54	– Reduced proliferation in sHG (P57; 3DPP)
			P21-475+   withdrawal for 14d (P489)	– Inhibition of hair growth and hair loss (P475)   – HFSCs remain expressing SC makers – CD34 and KRT15 (P475) – Hair growth inhibition is reversed after stopping DKK1 expression (P489)
DKK2 (gain-of-function)	IFE & HFs	Foxn1-Dkk2	E11.5	– Reduced density of HFs (E14.5)   – New hair growths from existing HFs (P1)	[68]