Building and Repairing the Skin Epithelium: Cell Polarity, Cytoskeleton and Asymmetric Cell Divisions
At gastrulation (embryonic day 7.5, E7.5), the ectoderm moves from inside the embryo to the surface. There, it forms the neuroectoderm that will invaginate to generate the neural crest and nervous system, and the epidermis, which begins as a monolayer of unspecified progenitors. These progenitors will progress not only to form the mature, stratified epidermis, but also its appendages, including hair follicles, sebaceous glands, sweat glands, and even mammary glands. The ectoderm also develops the oral epithelium of the head and neck region.
As the embryo grows, epidermal divisions are initially parallel relative to the plane of the epithelium. As epidermal stratification begins, spindle orientations become oblique, giving rise to one basal and one suprabasal daughter (Williams et al., 2014). Basal cells make and secrete extracellular matrix (ECM) composed of laminin 5, collagen IV and fibronectin, which act as receptors for transmembrane integrins composed of α and β subunits. In epidermis, one of the two major integrin heterodimers is α3β1, essential for engaging focal adhesion kinase (FAK) and mobilizing actin dynamics to assemble the ECM into an underlying basement membrane (Raghavan et al., 2000; 2003; Schober et al., 2007). The other is α6β4, which constitutes the transmembrane core of hemidesmosomes, ultrastructurally elaborate structures essential for strong adherence to the underlying basement membrane (Guo et al., 1995; Dowling et al., 1996). α6β4 engages BPAG2 (collagen XVII) a transmembrane collagen that reinforces this anchorage. On the cytoplasmic surface, α6β4 engages BPAG1, a similar cytoskeletal linker protein which in turn bind the keratin cytoskeleton whose filaments are composed of keratins 14 and 5 (Dowling et al., 1996; Guo et al., 1995). The K14-K5 keratin network provides an essential mechanical framework to these developing basal epidermal cells (Vassar et al., 1989; Coulombe et al., 1991). Both α3β1 and α6β4 utilize laminin 5 as their primary ligand.
During embryonic development, the monolayer of epidermal progenitors begins to polarize its integrins towards the basement membrane. This polarity is accompanied by the elaboration of hemidesmosomes. Polarity is also accompanied by increased cell-cell “adherens” junctions composed of the core transmembrane component, E-cadherin, and its associated proteins α-catenin and β-catenin. β-Catenin links E-cadherin to α-catenin, while α-catenin associates with filamentous actin and with actin-associated proteins (Vasioukhin et al., 2000; 2001a; Kobielak et al., 2004; Livshitz et al., 2012). Suprabasally, the cytoskeletal network and its adhesions strengthen through formation of desmosomes, whose core desmosomal cadherins bind plakoglobin (related to β-catenin), which binds desmoplakin (related to BPAG1), which binds keratin filaments (Kouklis et al., 1993; Gallicano et al., 1998; Vasioukhin et al., 2001b).
The maturation of epithelial junctions and enhanced epidermal polarity is typified by the localization of polarity proteins, Par3, Par6 and atypical protein kinase C (aPKC), at the apical actin cortex (Luxenburg et al., 2011; 2014). These maturation steps are also associated with a change in spindle orientation. By E15.5, 70% of mitotic spindles become polarized perpendicularly to the underlying basement membrane (Lechler et al., 2005; Williams et al., 2014). This shift in alignment from parallel/oblique to perpendicular enhances epidermal stratification and spinous cell formation, typified by expression of keratins 1 and 10 (Fuchs and Green, 1980; Cheng J et al., 1992). Genetic mutations in β1, α-catenin and proteins involved in actin dynamics show that cell substratum adhesion, cell-cell adhesion, and polarization of the cortical actin cytoskeleton are all essential for this perpendicular orientation of spindle microtubules (Lechler et al., 2005; Luxenburg et al., 2011; 2014).
Probing mechanism, we discovered that these asymmetric cell divisions utilize an ancient mechanism that has been previously been described to characterize the differential specification of daughter cell fates that occurs in the early C. elegans embryo and Drosophila neuroblast. These divisions asymmetrically partition fate determinants such that one daughter cell retains progenitor status, while the other differentiates (reviewed by Williams and Fuchs, 2014).
The core components of this pathway are LGN and mINSC, which become stabilized soon after the asymmetrically dividing cell enters mitosis (Lechler and Fuchs, 2005). mINSC binds to Par3, placing it at the apical inner surface of the cortical actin. LGN binds to Gαi, a G-coupled protein receptor which is also apically localized. Additionally, LGN can bind mINSC to link the Gαi and Par3 complexes. LGN can also bind to NuMA, a protein that can bind astral microtubules and link them to the apical cortical complex (Lechler and Fuchs, 2005; Williams et al., 2011). Although LGN cannot bind to mINSC and NuMA simultaneously, this can easily be met if as is likely, multiple subunits are involved in forming the complex.
Indeed, our recent genetic studies show that the Par3 and the Gαi complexes are both required for polarizing the perpendicular orientation of the spindle in these asymmetrically dividing cells, as is LGN (Williams et al., 2011; 2014). Additionally, this mechanism is integrally linked to Notch signaling, as it is for Drosophila neuroblasts (Williams et al., 2011; 2014). Importantly, Notch signaling is essential for epidermal differentiation and governs the transition between basal and spinous cell fates (Blanpain et al., 2006). Notably, genetic defects in LGN alone or combined loss of mInsc and Gαi lead to mostly planar cell divisions, and accompanying defects in epidermal differentiation (Williams et al., 2011; 2014).
Another feature worthy of mention here is that during interphase, epidermal basal cells apically localize their centrin-rich microtubule organizing center (basal body). The oldest centriole of the basal body grows a primary cilium. Primary cilia are known to enhance SHH signaling, which plays a key role in hair follicle morphogenesis. Thus it is curious that epidermal cells display a primary cilium as early as E10.5 in development, and that ciliary defects impact not only hair follicle formation but also epidermal differentiation, a program associated with Notch signaling (Ezratty et al., 2011; 2016). Notably, Notch receptors colocalize to the ciliary surface, and presenilin, necessary for processing and activating Notch, localizes to the basal body. Although we don’t yet know whether the localizations of Notch components are functionally relevant, the role for epidermal cilia seems to deviate from the classical mechanisms involving SHH.
In the adult, the epidermis must protect against harmful microbes on the outside and must retain body fluids. To provide this surface barrier, the epidermis must constantly rejuvenate itself. It performs this function by maintaining an inner layer of living, multiplying cells, which include stem cells. When a basal cell ceases to divide and begins its journey to the skin surface, it embarks on a program of terminal differentiation. As spinous cells form, they protect themselves by producing a mechanically durable, dense intracellular framework of filaments composed of keratin. Through abundant intercellular junctions, differentiating cells form continuous, interconnected cellular sheets. Later, as the spinous cells generate the granular layers, the cells assemble an indestructible proteinaceous envelope that serves as a scaffold on which lipids are extruded and organized to form the epidermal barrier. Cells reaching the skin surface lose all their metabolic and transcriptional activity. These squames are sloughed, replaced by inner cells differentiating and moving outward. Every two weeks, the epidermis is nearly brand new (Blanpain and Fuchs, 2009; Fuchs, 2012).
How the adult epidermis balances growth and differentiation during normal homeostasis is still unfolding. The simplest mathematical models of homeostasis are in agreement with the notion that asymmetric cell divisions persist in the adult, but are largely planar rather than perpendicular. How this shift comes about and whether LGN and/or mInsc are still involved in the process are unclear. However, in the adult as in the embryo, Notch signaling still functions in balancing dividing and differentiating cells, and the program of differentiation still operates in columnar units of upwardly moving cells.
A major question for the future will be to dissect the molecular controls that ensure that during embryogenesis the basal layer matches its proliferation with its needs for expansion and establishing the stratified epithelium, while in the adult it matches its proliferation with needs for maintaining the continual rejuvenation of the epithelium to maintain the surface barrier. While much work still needs to be done, we’ve begun to make inroads by conducting an in vivo lentiviral screen for genes which when knocked down, perturb the balance of growth and differentiation in the epidermis. Interestingly, one of the hits emerging from this screen was a gene that perturbed spindle orientation in the dividing epidermal progenitors (Asare et al., 2016). Adding further intrigue, the gene encodes a protein that has been implicated in association of peroxisomes with mitotic spindle, providing a tantalizing link for future exploration between organelle inheritance and cell divisions in the epidermis.
Central to the issue of how epidermal stem cells balance growth and differentiation is also how they coordinate cytoskeletal dynamics to balance cell-substratum and cell-cell junctions. Recently, we showed that when epidermal cells lose α-catenin, a number of adherens junction-associated proteins relocalize to integrin-mediated cell substratum junctions (Livshits et al., 2012; Vasioukhin et al., 2000; 2001). This results in elevated integrin/focal adhesion kinase (FAK)/Src/MAPK activity and enhanced proliferation. Conversely, if E-cadherin levels are elevated, overall cellular growth is impaired (Jamora et al., 2005). Moreover, in the absence of FAK, the downstream effector of integrin signaling, loss of α-catenin results in enhanced apoptosis of cells (Schober et al., 2007; Livshits et al., 2012).
The antagonistic relation between adherens junctions and integrin signaling provides a convenient mechanism by which the balance between proliferation and differentiation might be controlled in the epidermis. As the basal layer becomes more crowded, intercellular junctions would be expected to increase, while cell-substratum adhesions would be expected to decline. Another key component could be YAP/TAZ, a co-regulator of the TEAD family of DNA binding proteins that function prominently in promoting epidermal proliferation (Zhang et al., 2011). As recently shown by the Camargo and Piccolo labs, the transition of YAP/TAZ from cytoplasm to nucleus can be promoted by three distinct mechanisms: mechanical forces, WNT signaling, or reductions in α-catenin. This potentially places YAP/TAZ regulation at the crossroads of the mechanisms balancing epidermal proliferation and differentiation, and future studies will help to illuminate this.
One of the fascinating aspects of epithelial tissues is their ability to repair themselves after wounding. This process necessitates a shift in the balance of proliferation and differentiation, and as discussed in the previous section, it also involves changes in the stem cell microenvironment, triggering a cascade of different interactions with immune cells and other cell types. However, it also entails the ability of cells to repolarize and remodel their cytoskeleton and utilize it for migration. Only a few details have emerged regarding this latter process, and there is much left to be done. The skin is an excellent model for studying this process because of the ability of its stem cells to grow in culture under conditions where they migrate, adhere and assemble into cellular sheets. When coupled with conventional transgenic technology and our newfound lentiviral transduction system, it is also straightforward to transduce epidermal cells in vivo with fluorescently tagged proteins and then carry out live imaging of skin explants to examine how epithelial cells can migrate within a sheet during wound-repair (Vaezi et al., 2002; see also Heller et al., 2014; Movie 1). With the development of new and improved microscopy methods, we’re also able to increasingly shift to carrying out such studies in living mice.
The signaling pathways and cellular polarizations that occur at a wound front share many of the players involved in epidermal polarization during tissue morphogenesis. In this regard, WNT signaling has been broadly implicated in wound-repair as have the Par3, Par6 and atypical protein kinase C (aPKC) complex. The laboratory of the late Alan Hall initially discovered that this complex polarizes at the cellular front of a scratch wound of astrocytes in culture, and we’ve discovered similar roles for these proteins in wounded endodermal and epithelial sheets. Although the link to WNT signaling remains elusive, the polarization of the Par complex can be triggered by Cdc42, a small GTPase. To effectively migrate, cells must also polarize and stabilize their microtubules in a directed fashion towards the wound front. This process entails actin-microtubule binding proteins, including APC and ACF7 (Kodama et al., 2003).
My laboratory has been especially interested in ACF7, found not only in mammals, but also flies and worms. Although lethal when mutated, ACF7 functions in cultured endodermal cells to polarize and stabilize the Par complex and microtubules at a wound front (Kodama et al., 2003). Using conditional targeting, we specifically ablated ACF7 in the skin epidermis. Without ACF7, microtubules cannot grow and move along actin fibers that associate with α3β1 integrins at the leading edge of the keratinocyte (Movie 2; Wu et al., 2008). When we specifically ablate ACF7 in the hair follicle stem cells, the stem cells do not efficiently exit their niche, migrate into the wound bed and repair wounds in vivo (Wu et al., 2011). Probing mechanism, we further discovered that the microtubule binding site of ACF7 is riddled with GSK3β-kinase phosphorylation sites, which need to be inactivated in order for the negatively charged tubulin tails to efficiently recognize and bind ACF7 (Wu et al., 2011). Notably, WNT signaling is known to inactivate GSK3β, providing a potential link between this signaling pathway, ACF7 and the stabilization and polarization of microtubules in a directed fashion.
Whether ACF7 also functions in the polarization of spindle microtubules that occurs in asymmetric cell divisions is an intriguing notion, which remains to be investigated. However, recently, we uncovered another link between WNT signaling and microtubule organization. During embryogenesis, WNT signaling provokes the clustered organization of unspecified embryonic skin progenitors into placodes within the plane of the epithelium. The first cell divisions in these WNThi progenitors that lead to the formation of a hair bud are exclusively perpendicular: the daughter remaining attached to the basement membrane remains WNThi; the displaced daughter is WNTlo and adopts a new fate, eventually becoming a HF stem cell (Ouspenskaia et al., 2016). Using lentiviral-mediated genetic approaches, we showed that it is not asymmetric divisions per se that are essential, but rather the generation of daughters that are differentially WNThi and WNTlo, a feature which in embryogenesis is achieved through an asymmetric division. These new findings provide new insights and again, underscore the importance of digging deeper into the mechanisms involved. Overall, these recent studies exemplifies how we can exploit our knowledge of skin stem cells to elucidate the complex biological processes of tissue morphogenesis and wound-repair.