Chapter 46. Epidermal Growth and Differentiation

Although thin, human skin is a marvelously resilient and multifunctional organ. It performs immunomodulatory and thermoregulatory functions, is involved in social, cultural, and reproductive behaviors, and provides broad protection against water loss and environmental insults such as trauma, infection, and exposure to radiation or chemicals. The outermost layer of skin, termed the epidermis, consists of a stratified squamous epithelium and its appendages, including hair follicles, sebaceous, apocrine, and eccrine glands. This chapter discusses epidermal differentiation, with the primary focus placed on keratin filaments that are formed as major structural elements within the epidermis. Defects in epidermal keratins are known to play key roles in a number of important blistering epidermal diseases. Additional major epidermal differentiation markers, including keratohyalin granules and the cornified envelope, are also discussed.

Keratins and Their Classification

Keratins (also known as cytokeratins) are structural proteins that belong to the superfamily of intermediate filament (IF) proteins. They are heterogeneous in size (40–70 kDa) and charge (pI 4.7–8.4), and notoriously insoluble. Sequencing the human genome revealed the presence of 54 functional keratin genes that are nearly perfectly conserved in other mammals.1 The tremendous diversity of keratin genes had not been fully appreciated until the advent of database mining and genomics, and could not be accommodated in the original nomenclature system aptly devised by Roland Moll, Werner Franke and colleagues in 1982.2 In 2006, an international effort culminated in a revised nomenclature (Table 46-1) that accommodates the newly discovered keratins, adheres to the guidelines of the Human and Mouse Gene Organization Gene Nomenclature Committee, and maintains the original designation of keratins devised by Moll and colleagues.1

Sequence homology and gene substructure (number and position of introns) reveal two distinct groups of keratins of roughly equal size, designated type I and II IF genes1,3 (Fig. 46-1A). In Homo sapiens, functional type I and type II keratin genes are clustered on the long arms of chromosomes 17 (Fig. 46-1B) and 12 (Fig. 46-1C), respectively.1,4 Keratin genes are highly conserved across mammals, at the level of their organization, structure, sequence, and regulation.5 Mature filaments contain type I and II keratins in a 1:1 molar ratio.3,6 This requirement underlies the coordinated transcription of type I and II keratin genes. Remarkably, most type I and II keratin genes are regulated in a pairwise, tissue type-related, and differentiation-related fashion.7–9 This is illustrated particularly well in stratified epithelia such as epidermis (Fig. 46-2). Given their large number, differential regulation, and ease of detection (owing to abundance), keratin mRNAs and proteins represent unparalleled markers for staging the fate and differentiation of epithelial cells, under healthy and diseased conditions.

Keratin Proteins Form the Intermediate Filament Network of Epithelial Cells

Despite sequence differences, all keratins display the tripartite domain structure that is typical of IF-forming proteins (Fig. 46-1D). The central domain consists of an extended α helix featuring long-range heptad repeats that mediate coiled-coil dimerization. This “rod” domain is ∼310 amino acids long and is flanked by highly variable sequences at the N- terminal head and C-terminal tail domains (Fig. 46-1D).3 Neither terminal domain exhibits known functional motifs other than the glycine loops seen in epidermal keratins.10 The head and tail domains are readily protease-accessible at the surface of the filament, where they can foster interactions with neighboring filaments, other proteins, or serve as substrates for posttranslational modifications involved in their regulation.11 Given their heterogeneity of size and primary structure, the head and tail domains are expected to make key contributions to the differential function and regulation of keratin proteins in vivo.12

The central rod domain of keratins is the main determinant of self-assembly, with important contributions from the head domain as well.13 Assembly begins with the formation of heterodimers in which the α-helical central rod domains of type I and II keratins are aligned in parallel and perfect register. Heterodimers interact along their lateral surfaces and in an end-to-end fashion to give rise to the 10–12-nm wide filaments (Fig. 46-2A) which, depending on IF protein type and assembly conditions, may contain a variable number of subunits in cross section.13 Mature IFs lack a structural polarity, a direct consequence of the antiparallel orientation of their constituent coiled–coiled dimers. The extraordinary stability of keratin subunits reflects the tightness of interactions between type I and type II keratins.14 Most of the intracellular pool of keratin proteins (>95%) is polymerized.4 There is evidence that keratin IF assembly is initiated at the cell periphery, near the cortical F-actin cytoskeleton, in cultured epithelial cells.15

Keratins form the major IF network in all epithelial cells.2,3,9 The abundance and organization of keratin IFs in vivo differ among epithelia. Keratin proteins are highly abundant (10%–80% of total cellular proteins) in surface-exposed stratified squamous epithelia (e.g., epidermis, oral mucosa, corneal epithelium, etc.).7 In epithelial cells of such tissues, keratin IFs are organized in a pancytoplasmic network extending from the surface of the nucleus to the cytoplasmic periphery, where they are membrane-anchored at sites of cell–matrix and cell–cell adhesion (hemidesmosomes, desmosomes) (Fig. 46-2B). In simple epithelia (e.g., liver, gut, pancreas, etc.), keratins are less abundant. In such tissues, polarized epithelial cells often feature asymmetrically organized keratin IFs concentrated mostly at the cytoplasmic periphery and, particularly, the apical pole.4 Several associated proteins contribute to the organization and regulation of keratin IFs in these various settings.16 Some of these proteins promote the bundling of keratin IFs (e.g., filaggrin, trichohyalin), their association with microtubules and actin microfilaments (e.g., plectin, BPAG isoforms) and/or with desmosomes or hemidesmosomes (desmoplakin, plakophilin, BPAG isoforms, etc.) (Fig. 46-2). Other partners, for example, TRADD, 14–3-3, Akt, reflect the newly discovered participation of keratin IFs in signaling roles.17

Keratin Gene Expression Mirrors Epithelial Differentiation: The Case of Epidermis

Skin provides a beautiful example of the tight relationship that has evolved between keratin gene regulation and epithelial differentiation. More than half of all known keratin genes are expressed in mature mammalian skin tissue alone. The architectural complexity of adult skin epithelia is achieved through a temporally and spatially regulated expression of keratin genes and a number of other epithelial differentiation-related genes.7,18 In the clinical setting, keratin typing is often exploited in diagnosing cancer type, its differentiation status (and therefore prognosis), as well as the origin of the cells forming metastatic foci. This strategy is also applied for diseases other than cancer (see below).4

In “thin” interfollicular epidermis (e.g., trunk; Fig. 46-2), mitotically active cells of the basal layer act as progenitors, and consistently express K5 and K14 as their main keratin pair, along with low levels of K15. Onset of differentiation coincides with the appearance of the K1/K10 pair through a robust transcriptional induction that occurs at the expense of the K5/K14 genes, which are downregulated.7,18 Accordingly, K1/K10 keratins are readily detectable in the lowermost suprabasal layer of epidermis (Fig. 46-2D). The appearance of K1 and K10 correlates with a sudden and dramatic shift in the organization of keratin IFs, which now exhibit significant bundling.19 Another type II gene, K2e, is expressed at a later stage of differentiation, i.e., the granular layer.20

The epidermis of palm and sole skin is specialized for resisting a high degree of mechanical stress, and thus is markedly thicker. This function is reflected in its architecture of alternating stripes of primary and secondary ridges,21 and again, in keratin expression. In the thick, stress-bearing, primary ridges, the major differentiation-specific (type I) K9 is presumed to foster a more resilient cytoskeleton. In the thinner secondary ridges, postmitotic keratinocytes preferentially express the (type II) K6a and (type I) K16 and K17.22 Relative to K1, K9, and K10, the properties of K6a, K16, and K17 likely foster greater cellular pliability, thereby providing flexible “hinge” regions between the more rigid, K1/K9-rich primary ridges.22 While this attractive model remains to be supported by direct experimentation, it is consistent with the dramatic upregulation of K6a, K6b, K16, and K17 that occurs in keratinocytes recruited from wound margins to participate in the restoration of the epidermal barrier following injury.23,24

Epidermal disease states are often accompanied by a deviation from normal terminal differentiation and, not surprisingly, they are almost always accompanied by altered keratin gene expression. For instance, K6a, K6b, K16, and/or K17, normally restricted to wound repair in trunk epidermis, are ectopically induced in psoriasis and related hyperproliferative disorders, nonmelanoma skin cancers, viral infections, and other conditions accompanied by inflammation.23,25,26 Similar “replacement” of the K1 and K10 keratin pair by K6, K16, and K17 occurs when normal human keratinocytes are placed in culture.23,25,27 During early progression toward malignancy, cutaneous squamous cell carcinoma progressively shifts from being K6/K16-positive while maintaining some degree of K1/K10 expression, thus reflecting a differentiated state, to being entirely negative for K1/K10 and positive for the simple epithelial keratins K8/K18, indicative of a less differentiated and more aggressive state.28 These variations in keratin gene expression likely impact the biological properties of keratinocyte in a significant way.

Function of Keratin in the Epidermis and Other Skin Epithelia

A major function fulfilled by keratins and all other IF proteins is to enhance the cell's ability to withstand trauma. This structural support function3,4 is made possible by the unique mechanical properties exhibited by IF networks.24 This function is enhanced by attachment of IFs to adhesion complexes (desmosomes, hemidesmosomes), and to F-actin and microtubules.3,4,16 Partial or complete loss of this function, for example, through inherited mutations, underlies a wide variety of rare diseases that render cells fragile and unable to sustain mechanical stress (Table 46-2). In vivo, the mechanical properties of IF networks likely need to be modulated, in a dynamic fashion, to meet the demands placed on cells by changing physiological circumstances. To a degree, varying needs along a continuum of viscoelastic properties likely account, at least in part, for the dynamic regulation of keratin IF genes and their proteins in vivo.17,24

The recent discovery of nonmechanical functions for keratin proteins has placed the field on an exciting new path.17 In hair follicles, K17 promotes the anagen (growth) phase by attenuating TNF-α-induced apoptosis in matrix keratinocytes.29 In the epidermis, the suprabasally expressed K10 regulate proliferation in the basal layer of epidermis and in sebaceous glands, likely through a noncell-autonomous mechanism,30,31 while K17 cell autonomously regulates protein synthesis and cell size in wound-proximal keratinocytes.32 Keratins influence the melanin pigment distribution and, thus, skin pigmentation.33–35 In polarized epithelia, keratin IFs impact the distribution of organelles, routing of specific outer membrane proteins, and response to stress.36 These newly defined keratin functions, which are just beginning to be understood, involve regulated interactions between keratin proteins and noncytoskeletal proteins, many exerting key roles in specific signaling pathways.16,17,24,36 Interference with these functions could play a role in the pathogenesis of disorders linked to IF gene mutations.4 The discovery of these novel keratin functions provides an opportunity to better understand the diversity and context-dependent regulation of IF genes and their proteins.

Mutations in Epidermal Keratins Underlie Several Inherited Skin Blistering Diseases

In the 1980s, ultrastructural studies showed that epidermal basal keratinocytes of patients with the Dowling-Meara form of epidermolysis bullosa simplex (EBS) contained dense cytoplasmic aggregates,37,38 later shown to contain mispolymerized keratin.39,40 In parallel, reverse genetic studies showed that expression of dominant-negative keratin mutations cause keratin IF aggregation in cultured cells, and epithelial fragility in vivo.41,42 In the early 1990s, the first mutations in keratins K5 and K14 were linked to EBS.39,43,44 In EBS, K5/K14-expressing basal layer keratinocytes literally rupture in response to mild mechanical trauma to the skin. Fragility is occasionally seen in other sites of K5/K14 expression, such as the cornea and oral mucosa.45 Similarly, dominant mutations in the suprabasally expressed K1, K10, and K2e were found to cause epidermolytic hyperkeratosis (EHK),46,47 ichthyosis bullosa of Siemens,48 and related diseases (Table 46-2). Such efforts established that compromising keratin function engenders structural failure and fragility. In specific instances, depending on the disorder, this primary defect is accompanied by enhanced proliferation and hyperkeratosis,4,30 or aberrations in skin pigmentation.33–35 Since then, a broad range of additional diseases affecting either epidermal appendages (e.g., hair, nail) or nonskin epithelia (e.g., oral mucosa, cornea) have been linked to keratin mutations.4,49

The following general principles can be drawn from the large body of data accumulated to date while studying keratin-based disorders. The majority of cases involve single missense mutations acting in a dominant-negative fashion. Small insertions and deletions are also seen with some frequency.49 In this setting, dominance implies that disease-causing mutant keratin proteins do not markedly alter the early stages of assembly (dimer, tetramer formation) up to the step(s) involving subunit incorporation in a growing IF polymer.4 Depending on their nature and location within the keratin protein backbone, these mutations exert a wide range of effects on the assembly or organization of IFs in keratinocytes, with a corresponding impact on the severity of clinical presentation. This concept can be readily illustrated for EBS.50 Mutations altering residues that are highly conserved within the central rod domain tend to dramatically alter the structure of keratin IFs, foster the formation of aberrantly polymerized keratin aggregates, and cause severe disease (as seen in the Dowling-Meara form of EBS). Conversely, mutant proteins that elicit a clinically milder version of the disease (e.g., Weber-Cockayne EBS) affect keratin assembly more subtly, and do not cause keratin aggregation.50 The extent to which a given keratin mutant compromises the function of the entire IF network is also a function of the number and abundance of potentially redundant IF proteins occurring in a given cell, or its ability to overexpress an alternate IF protein.4 While it is assumed that these mutations elicit a cell fragility phenotype in part through their ability to alter cellular mechanics,51–53 there is evidence that they alter keratin regulation, for example, via posttranslational modifications,11 and elicit a cellular stress response.4,54 Treatment options for EBS are currently limited, and are primarily palliative in nature.50 They consist of supportive care for skin, management of skin blisters as they heal so as to prevent infection, and preventive avoidance of mechanical trauma.

Alongside the substantial build up in keratin IFs inside differentiating keratinocytes, two specializations allow the epidermis to build a remarkably effective and mechanically resilient barrier: (1) the cornified envelope (CE), a covalently cross-linked protein polymer that forms underneath the plasma membrane, and (2) an extracellular hydrophobic phase that is made up of specialized lipids synthesized by terminally differentiating keratinocytes.55 Epidermal lipids are discussed in detail in Chapter 47, and will not be discussed further here.

The CE represents a complex assembly of covalently cross-linked proteins that forms underneath, and eventually replaces, the outer keratinocyte membrane in the granular layer of epidermis, as part of the final push to complete terminal differentiation. This 20-nm thick sheath, which is so insoluble and stable it resists extended boiling in the presence of strong denaturants, encases the cell's interior, and significantly contributes to the physicochemical properties of the stratum corneum compartment.56 The extraordinary stability of the CE results in large part from the large number of ϵ-(-γ-glutamyl)-lysine isopeptide bonds between its primary constituents (see below), which are further reinforced by disulfide bridges. These isopeptide bonds are catalyzed by transglutaminases, a family of calcium-dependent enzymes.57 The outer aspect of the CE is covalently linked to ceramides and other specialized lipids that are produced by, and secreted from, differentiated granular keratinocytes, whereas large bundles of tightly packed keratin IFs are cross-linked to its inner aspect.

The major protein constituents of the CE are loricrin, involucrin, filaggrin, elafin, cystatin A, cornifelin, several small proline-rich (SPR) proteins and calcium-binding S100 proteins, and “late-envelope proteins” (LEPs). In addition to these proteins, key components of desmosomes (e.g., desmoplakin, envoplakin, and periplakin) and several type II keratins (K1, K2, K5) are also cross-linked into the CE56–58 (Table 46-3). CE proteins have several interesting features in common. First, many CE proteins are encoded by genes clustered as part of the epidermal differentiation complex (EDC) locus on human chromosome 1q21. In addition to obvious evolutionary implications, this genomic clustering brings up the distinct possibility of coordinated regulation through cis-acting determinants.55,58 Second, many of these proteins are made as precursors that are activated by proteolytic cleavage, calcium binding, or other modifications at the time of CE formation. Third, many CE proteins are made of repeated units that lack an intrinsically defined three-dimensional structure. Fourth, virtually all CE proteins are transglutaminase substrates.55,57

Involucrin, loricrin, and filaggrin, in particular, are well-characterized components of the CE.56,57 Involucrin biosynthesis is initiated in the spinous compartment, soon after the onset of K1/K10 keratin expression, in differentiating epidermal keratinocytes. It is the first major component to be activated and cross-linked into the emerging CE, possibly reflecting a scaffolding role, and is concentrated in the outermost aspect of the mature CE. Profilaggrin and loricrin are synthesized in precursor forms and stored in keratohyalin granules. Loricrin is the major structural component of the CE (∼80% by weight)—this largely unstructured but highly flexible protein features glycine loops as described for type II keratins. Once activated via dephosphorylation and proteolytic cleavage, filaggrin participates in the bundling of keratin IFs in the late granular compartment of epidermis (hence accounting for its incorporation into the CE) and also is degraded to free amino acids that contribute to the cornified cell's ability to retain water. Basic information about other CE components is provided in Table 46-3.56,57 The protein composition and fine structure of the epidermal CE differs to a degree from those of the other stratified squamous epithelia; the underlying significance of this CE heterogeneity is unclear.56,57

The important contribution of the CE toward proper barrier function in epidermis is reflected in the clinical symptoms associated with mutations altering their primary constituents in the context of human diseases55 (Table 46-3). This is particularly well illustrated by the recently documented role of filaggrin mutations in ichtyosis vulgaris, atopic dermatitis (AD), and AD-associated hay fever or asthma (the so-called atopic march; see Chapter 14).59–61 The partial or complete loss of filaggrin protein engendered by frameshift mutations in the corresponding gene compromises the epidermal barrier function and allows for the entry of irritants and allergens into the skin,62 thereby eliciting local (and even systemic) inflammation accompanied by itching, erythema, and flaking reflecting improper cornification (for further information see Chapter 14). Barrier status in the epidermis thus sets our degree of exposure to external elements and contributes to define the relationship that we have with our environment.

In this chapter, we have discussed epidermal differentiation viewed through the prism of keratin gene regulation and CE formation. While it is well established that specific type I and type II keratins are coexpressed as pairs associated with various stages of normal and pathological epidermal differentiation, and that mutations of major epidermal keratins cause various forms of skin blistering diseases, the functional implications of these different keratins, not only in terms of structural roles, but also in terms of their novel involvements in signal transduction and organelle transport, are just beginning to be understood. Additional studies are needed to better understand the roles of regulatory molecules in epidermal proliferation, homeostasis, and disease; and how the keratin filaments function, at the atomic level, in normal and diseased epidermis. These future studies should lead to new modalities, including cell and gene therapies, for a better treatment of some of the devastating skin diseases.