Epidermal Barriers

Abstract

The epidermis functions as a physical barrier to the external environment and works to prevent loss of water from the skin. Numerous factors have been implicated in the formation of epidermal barriers, such as cornified envelopes, corneocytes, lipids, junctional proteins, proteases, protease inhibitors, antimicrobial peptides, and transcription factors. This review illustrates human diseases (ichthyoses) and animal models in which the epidermal barrier is disrupted or dysfunctional at steady state owing to ablation of one or more of the above factors. These diseases and animal models help us to understand the complicated mechanisms of epidermal barrier formation and give further insights on epidermal development.

Numerous factors are required for skin epidermal barrier development, including structural and junctional proteins, lipids, and transcription factors. Their impairment can lead to ichthyoses—disorders defined by thick, scaly skin.

The skin is the largest organ in the body and is constantly confronted with external stimuli, such as microorganisms, chemicals, heat, and water. Ever since vertebrates started their terrestrial life, skin has had the ability to retain water and to prevent liquids from penetrating into the internal organs. The epidermal barrier is indispensable to the maintenance of organisms; many examples in experimental animals and humans have shown that severe defects in epidermal barrier formation lead to an early demise.

A great number of factors play roles in barrier development, including structural and junctional proteins, lipids, and transcription factors. This makes it difficult to determine which factor or factors are affecting barrier function through their impairment. It is noteworthy that barrier development is one aspect of the development of the epidermis itself.

The ichthyoses are a heterogeneous group of disorders characterized by thick, scaly skin (Oji et al. 2010; Schmuth et al. 2013). “Icthy” comes from the Greek word for fish. Congenital ichthyosis patients and animal models of these conditions generally show epidermal barrier defects caused by mutations in the genes encoding epidermal barrier components or key regulators. The thick, scaly skin of ichthyoses is thought to result from the skin compensating for barrier dysfunction (Segre 2006). This review presents some ichthyosis diseases and animal models with skin barrier dysfunction to illuminate the factors that are relevant to epidermal barrier formation.

ASSESSMENT OF EPIDERMAL BARRIER FUNCTION

Various methods have been used to evaluate epidermal barrier function in animal models and humans. As is true for other experimental procedures, each method has advantages and disadvantages (Indra and Leid 2011). In any case, researchers must recognize two central aspects of barrier function: “outside-in” barrier and “inside-out” barrier (Elias et al. 2008a; Proksch et al. 2008). “Outside-in” denotes the epidermal function of hindering the infiltration of external materials through the epidermis to the area inside. “Inside-out” denotes the ability of the epidermis to prevent the leakage of water and electrolytes from inside. As many of the epidermal components are involved in both outside-in and inside-out barrier function, it is not surprising to find many experimental animals with impairment of both barrier functions.

Transepidermal water loss (TEWL) is a simple measure of the quantity of water that passes from inside the body through the epidermis to the atmosphere by evaporation and diffusion (Fluhr et al. 2006; Indra and Leid 2011). Skin barrier impairment presents as increased TEWL. The method of taking TEWL as a proxy for barrier function has been used for more than three decades and can be applied not only to animal skin but also to human skin (Elias and Brown 1978; Hammarlund and Sedin 1979; Wilson and Maibach 1980; Aszterbaum et al. 1992). However, when a probe is used for TEWL measurement, the score can vary depending on the probe site (Chilcott and Farrar 2000). In addition, the ambient temperature and humidity greatly affect TEWL. For these reasons, researchers must be cautious in interpreting the results of TEWL, especially for human patients. TEWL reflects only inside-out barrier function and not outside-in barrier function.

Lanthanum tracer assay is another way of evaluating inside-out barrier function (Elias and Friend 1975; Elias and Brown 1978). In this assay, lanthanum nitrate is injected intracutaneously into experimental animals and specimens are later observed by electron microscopy. Lanthanum tracer percolates outward to the stratum granulosum (SG) and stops at tight junctions (TJs). The tracer does not reach the stratum corneum (SC) in wild-type skin (Elias and Brown 1978). In contrast, certain barrier impairments, including tight junction abnormalities, allow the tracer to reach the SC via intercellular spaces.

Similar to lanthanum tracer assay, the surface biotinylation assay was developed. In this assay, biotin instead of lanthanum is injected into animal skin (Merzdorf et al. 1998; Furuse et al. 2002). Specimens are incubated with avidin conjugated with fluorescein and visualized by fluorescent microscopy.

Dye penetration assay is the gold standard for evaluating outside-in epidermal barrier function. In this method, animal fetuses or neonates are simply soaked in a dye such as X-Gal, Lucifer Yellow, toluidine blue, or hematoxylin (Hardman et al. 1998; Indra and Leid 2011). Barrier acquisition presents as non-dye-stained area. The dye penetration assay clearly shows E16.5 as the time point of barrier formation in mice (Hardman et al. 1998). Along with this, the dye penetration assay gives locational information on skin barrier development.

There are other ways of evaluating barrier function through physical and chemical challenges to the skin. These include tape stripping and the application of acetone or detergent. Generally these methods are used when experimental animals have only subtle barrier defects or do not show any barrier dysfunction at steady state. When tape stripping or chemical application is used, barrier recovery rates are calculated (Ghadially et al. 1996). Except where otherwise noted, this review addresses experimental animals or human diseases that lead to physiological barrier dysfunction.

COMPONENTS CONTRIBUTING TO EPIDERMAL BARRIER FORMATION

As described earlier, there are many factors involved in epidermal barrier function. As it is almost impossible to cover all of the factors, this review focuses on the following components: (1) cornified envelope and corneocytes, (2) TJs and other junctional proteins, (3) lipids, (4) proteases and protease inhibitors, (5) antimicrobial peptides, and (6) transcription factors. Table 1 summarizes the selected mouse models with epidermal barrier defects.

Table 1.

Selected mouse models with physiological barrier defects

Gene	Gene product	Mutation	Human counterpart	References
Cornified envelope, corneocytes
Inv/Ppl/Evpl	Involucrin/periplakin/envoplakin	Null	-	Sevilla et al. 2007
Tgm1	Transglutaminase-1	Null	LI, CIE	Matsuki et al. 1998; Nakagawa et al. 2012
Flg	Filaggrin	Null	IV	Kawasaki et al. 2012
Krt1	Keratin-1	Null	EI	Roth et al. 2012
Krt10	Keratin-10	Null	EI	Reichelt et al. 2001
Krt16	Keratin-16	Null	PC	Lessard and Coulombe 2012
Junctional proteins
Cldn1	Claudin-1	Null	IHSC	Furuse et al. 2002
Dsc1	Desmocollin-1	Null	-	Chidgey et al. 2001
Cx43	Connexin-43	Carboxy-terminal truncation	HLHS, ODDD	Maass et al. 2004
Cdh1	E-cadherin	Null-C	-	Tunggal et al. 2005
Lipids
Fatp4	Fatty acid transport protein-4	Null, Null-C	IPS	Herrmann et al. 2003; Moulson et al. 2003; Herrmann et al. 2005
Abca12	ABC transporter A12	Null	HI, LI, CIE	Smyth et al. 2008; Yanagi et al. 2008; Zuo et al. 2008
Gba	β-Glucocerebrosidase	Null	Gaucher syndrome type II	Holleran et al. 1994
Ugcg	UDP-glucose ceramide glucosyltransferase	Null-C	-	Amen et al. 2013
Cers3	Ceramide synthase-3	Null	ARCI	Jennemann et al. 2012
Elovl4	Elongation of very long chain fatty acids-like-4	Null	ISQMR	Vasireddy et al. 2007
Elovl1	Elongation of very long chain fatty acids-like-1	Null	-	Sassa et al. 2013
Aloxe3	Lipoxygenase-3	Null	CIE	Krieg et al. 2013
Aloxe12	12(R)-lipoxygenase	Null	CIE	Epp et al. 2007; Moran et al. 2007
Cgi58 (Abhd5)	Comparative gene identification-58 (abhydrolase domain-containing-5)	Null	DCS	Radner et al. 2010
Proteases and protease inhibitors
Spink5	LECTI (lymphoepithelial Kazal-type inhibitor 5)	Null	NS	Yang et al. 2004; Descargues et al. 2005; Hewett et al. 2005
St14	Matriptase	Null	IHS	List et al. 2002
Prss8	Prostasin	Null	-	Leyvraz et al. 2005
Casp14	Caspase-14	Null	-	Denecker et al. 2007
Antimicrobial peptides
Cramp	Cathelicidin-related antimicrobial peptide	Null	-	Nizet et al. 2001
Transcription factors
Klf4	Krippel-like factor-4	Null	-	Segre et al. 1999
Gata3	Gata-binding protein-3	Null-C	-	de Guzman Strong et al. 2006
Taf10	Transcription initiation factor TFIID subunit 10	Null-C	-	Indra et al. 2005
Grhl3	Grainy head-like-3	Null	-	Ting et al. 2005
Arnt	Aryl hydrocarbon receptor nuclear translocator	Null-C	-	Takagi et al. 2003; Geng et al. 2006
Cebpa/Cebpb	CCAAT/enhancer binding protein (C/EBP), α and β	Null-C	-	Lopez et al. 2009

ARCI, autosomal recessive congenital ichthyosis; CIE, congenital ichthyosiform erythroderma; DCS, Dorfman-Chanarin syndrome; EI, epidermolytic ichthyosis; HI, harlequin ichthyosis; HLHS, hypoplastic left heart syndrome; IHS, ichthyosis hypotrichosis syndrome; IHSC, ichthyosis-hypotrichosis-sclerosing cholangitis; IPS, ichthyosis prematurity syndrome; ISQMR, ichthyosis, spastic quadriplegia, and mental retardation; IV, ichthyosis vulgaris; LI, lamellar ichthyosis; NS, Netherton syndrome; Null-C, conditional null; ODDD, oculodentodigital dysplasia.

Cornified Envelope and Corneocytes

The formation of corneocytes and the cornified envelope (CE) is the final step of epidermal differentiation. The CE is formed through crosslinking of insoluble membranous proteins by transglutaminase (Candi et al. 2005). Major components of the CE include involucrin, periplakin, envoplakin, and loricrin (Rice and Green 1977, 1979; Mehrel et al. 1990; Ruhrberg et al. 1996; Ruhrberg et al. 1997). Unexpectedly, mice with loss of each of these proteins did not show overt phenotypic changes (Maatta et al. 2001; Aho et al. 2004; Djalilian et al. 2006), other than loricrin-deficient mice, which developed transient ichthyosiform skin in the neonatal period that disappeared at several days after birth (Koch et al. 2000). In line with the results of the knockout (KO) mice, there have been no reports on human mutations in involucrin, periplakin, or envoplakin genes, although single-base-pair-insertion mutations in the gene encoding loricrin are known to cause loricrin keratoderma in humans, which is characterized by erythrokeratoderma and digital constriction bands (pseudo-ainhum) (Ishida-Yamamoto 2003). A further transgenic approach to delineating the role of CE components in the epidermal barrier failed to find apparent phenotypic changes in double-KO mice for involucrin/periplakin or periplakin/envoplakin or envoplakin/involucrin (Sevilla et al. 2007). Barrier defects were observed only when the three proteins (involucrin, periplakin, and envoplakin) were ablated at the same time (Sevilla et al. 2007). The triple-KO mice were viable but developed dry, ichthyotic skin, reflecting delayed permeability barrier acquisition as shown by dye penetration assay. The impaired epidermal barrier of the triple-KO mice might be accounted by for by (1) the fragility of the cornified envelope, (2) reduced epidermal lipid content, and (3) aberrant filaggrin processing. Intriguingly, the triple-KO mice were characterized by dermal inflammatory cell infiltrates, mostly composed of CD4⁺ T cells, and loss of epidermal dendritic T cells (Sevilla et al. 2007). This inflammatory phenotype in the barrier-defective mice might be consistent with the fact that atopic dermatitis can develop in people with filaggrin mutations, which is discussed below.

In contrast to KO mice for components of the CE, the ablation of transglutaminase, the enzyme crosslinking CE proteins, is detrimental. Mutations in TGM1, which encodes transglutminase-1, lead to lamellar ichthyosis (LI) or congenital ichthyosiform erythroderma (CIE) in humans (Huber et al. 1995; Russell et al. 1995; Laiho et al. 1997). Tgm1-null mice were shown to have severe barrier defects and neonatal lethality (Matsuki et al. 1998; Kuramoto et al. 2002). These mice had erythrodermic skin with defective CE assembly, as well as disintegrated lipid bilayers between CEs (Matsuki et al. 1998; Kuramoto et al. 2002). The same research group recently generated knock-in mice with a Tgm1 R142C mutation that is equivalent to TGM1 R143C in human LI or CIE patients (Nakagawa et al. 2012). Those mice showed severe barrier defects and neonatal lethality (Nakagawa et al. 2012). The results of Tgm1 transgenic mice underscore the indispensable role of transglutaminase in epidermal barrier formation.

Filaggrin is known to bundle keratin filaments and to be incorporated into corneocytes (Steinert et al. 1981). Loss of filaggrin expression has been observed in the epidermal keratinocytes of ichthyosis vulgaris (IV) patients (Sybert et al. 1985; Nirunsuksiri et al. 1998). However, the many repeats in its sequence hindered the sequencing of FLG, which encodes filaggrin. The McLean group succeeded in sequencing FLG and discovered loss-of-function mutations in IV patients (Smith et al. 2006). As it was known that IV is a risk factor for atopic dermatitis (AD), the same group sequenced FLG in AD patients and found FLG mutations (Palmer et al. 2006). Since then, many reports have addressed the frequency of FLG mutations in AD patients (Rodriguez et al. 2009; McAleer and Irvine 2013).

Spontaneous mutant mice called “flaky-tail mice” have been known to show dry, flaky skin and to show abnormal filaggrin processing (Presland et al. 2000). Sequencing technology revealed that flaky-tail mice have a frameshift mutation in Flg (Fallon et al. 2009). Since then, the flaky-tail mouse has been used as an AD model for its spontaneous dermatitis phenotype (Oyoshi et al. 2009; Scharschmidt et al. 2009b; Moniaga et al. 2010). Unexpectedly, however, Flg^−/− mice show no overt phenotypic changes, including spontaneous dermatitis (Kawasaki et al. 2012). Skin barrier defects in the steady state were not confirmed in Flg^−/− mice except in the outside-in barrier against liposome, whereas enhanced irritant and allergic contact dermatitis responses were observed in Flg^−/− mice on hapten application (Kawasaki et al. 2012). Recently, pyrosequencing of flaky-tail mice revealed another homozygous nonsense mutation in the gene encoding mattrin (Sasaki et al. 2013; Saunders et al. 2013). This mutation was reported to be involved in the spontaneous dermatitis phenotype in the mice (Sasaki et al. 2013; Saunders et al. 2013). A common missense variant in the orghologous human gene, MATT, was identified in AD patients (Saunders et al. 2013). Taken together, it may be that the filaggrin mutations in AD patients do not cause spontaneous dermatitis but, rather, contribute to increased percutaneous immune response caused by penetration of antigens, haptens, and irritants into the deeper SC. In contrast, it may be that the mattrin mutations in AD patients induce spontaneous dermatitis as well as elevated immune response on percutaneous challenge.

The implications of human and murine experiments on CE and filaggrin are as follows: (1) loss of one CE component or filaggrin can be compensated by other proteins and does not lead to severe barrier defects, (2) the ablation of several CE components (exemplified in Inv/Ppl/Evpl triple-KO mice) causes barrier defects, (3) in contrast, transglutaminase, which crosslinks CE proteins, is essential for epidermal barrier formation, and (4) disturbance of the CE (as seen in Inv/Ppl/Evpl triple-KO mice and Tgm1-null mice) affects not only the CE but also the lipid contents in the SC.

Keratins are the largest group of intermediate filaments and are expressed in epithelial cells (Schweizer et al. 2006). Keratins are classified into type I (acidic) and type II (basic to neutral). Type I and type II keratins form noncovalent heteropolymers. The specificity of keratin subtypes in each cell has been known (Schweizer et al. 2006). In the epidermis, basal keratinocytes express keratin5/keratin14 (K5/K14) heteropolymers, whereas the keratinocytes in the spinous and granular layers preferentially possess keratin1/keratin10 (K1/K10) (Moll et al. 1982; Nelson and Sun 1983; Roop et al. 1983). Dominant-negative mutations in KRT1 or KRT10 (encoding K1 or K10) cause epidermolytic ichthyosis in human, which is characterized by erythroderma and widespread blister formation at birth (Cheng et al. 1992; Rothnagel et al. 1992). Several mice with keratin modifications have been used to investigate keratin’s involvement in the epidermal barrier. Krt10-null mice show no severe phenotypes and are viable, which is explained by the compensatory expression of K5/K14 in suprabasal layers (Reichelt et al. 2001). Dye penetration assay revealed a focal outside-in barrier defect in the forepaw pad skin of Krt10-null neonates. In contrast to Krt-10 null mice, Krt1-null mice show perinatal lethality (Roth et al. 2012). Krt1-null have elevated TEWL compared with the wild type, although the outside-in barrier in these mice is intact (Roth et al. 2012). The increased fragility of the CE might be responsible for barrier dysfunction in Krt1- or Krt10-null mice (Roth et al. 2012). In contrast to the results for Krt1-null mice, Krt1/Krt10 double-KO mice do not show altered inside-out barrier by biotin permeability assay (Wallace et al. 2012). The discrepancy between Krt1-null mice and Krt1/Krt10 double-KO mice might be accounted for by the assays used in each experiment.

Dominant-negative mutations in the gene encoding keratin 16 (KRT16) cause pachyonychia congenita in humans, characterized by hypertrophic nail dystrophy and palmoplantar keratoderma (PPK) (McLean et al. 2011). Krt16-null mice were reported to develop PPK (Lessard and Coulombe 2012). Intriguingly, the front paws of adult Krt16-null mice show a focal loss of outside-in barrier using dye-penetration assay corresponding to hyperkeratotic lesions, suggesting that hyperkeratosis is a compensatory process resulting from barrier dysfunction (Lessard and Coulombe 2012). The cause of focal barrier disruption is not clear, even though the authors showed that there is reduced expression of filaggrin in the affected area of Krt16-null mice (Lessard and Coulombe 2012), which may not necessarily explain barrier disturbance as discussed earlier in Flg-null mice (Kawasaki et al. 2012).

Tight Junction and Other Junctional Proteins

The borders of epithelial cells are occupied by several junctional devices, including tight junctions (TJs), desmosomes, gap junctions (GJs), and adherent junctions (AJs). The role of TJs in the epidermal barrier is highlighted in claudin-1-deficient mice (Furuse et al. 2002). Claudin-1 has been identified as a TJ protein (Furuse et al. 1998), and copolymers of heterogeneous claudins make TJ strands (Furuse et al. 1999). Claudin-1-deficient mice show neonatal lethality (Furuse et al. 2002). As expected, disruption of normal TJs results in inside-out barrier defects in claudin-1-deficient mice as revealed by increased TEWL and positive biotin permeability assay (Furuse et al. 2002). Furthermore, outside-in barrier dysfunction in these mice is observed; it possibly results from altered intercellular lipid composition (Sugawara et al. 2013). The cause of the altered lipid composition in these mice is not clear. The fact that claudin-1 gene mutations lead to human ichthyosis-hypotrichosis-sclerosing cholangitis (IHSC syndrome) corroborates the importance of this molecule in the epidermal barrier (Hadj-Rabia et al. 2004). Surprisingly, overexpression of claudin-6 under involucrin promoter in mice (INV-Cldn6 mice) produces neonatal lethality with outside-in and inside-out barrier defects, a phenotype similar to that of claudin-1-deficient mice (Turksen and Troy 2002). These complicated results indicate that differential expression of, or the balance of, claudin proteins might be relevant in the tightness of the TJ barrier (Kubo et al. 2012).

Desmosomes are another junctional device involved in cell adhesion, and disturbed desmosomal components have been shown to cause skin, hair, and heart abnormalities in humans (Brooke et al. 2012; Petrof et al. 2012). Desmocollins and desmogleins are the desmosomal cadherins that extend into the extracellular area and adhere to neighboring cells (Delva et al. 2009). Desmocollin1-deficient mice show flaky skin and impaired outside-in/inside-out epidermal barrier (Chidgey et al. 2001). However, this barrier formation may not be due directly to desmocollin-1 in the epidermal barrier; it could simply be a consequence of the acantholytic skin in these mice (Chidgey et al. 2001). Mutations in the gene encoding desmocollin-1 in humans have not been reported.

GJs mediate the exchange of ions, metabolites, and second messengers between neighboring cells (Maeda and Tsukihara 2011). Mutations in the gene encoding connexin-43 (Cx43), a major GJ protein, are known to be responsible for oculodentodigital dysplasia (Paznekas et al. 2003) and hypoplastic left heart syndrome (Dasgupta et al. 2001). Mice in which the C-terminus of Cx43 was ablated died shortly after birth (Maass et al. 2004). Unexpectedly, neonates of these mice show scaly skin and constriction bands around the tails as well as outside-in barrier defects shown by toluidine blue dye penetration assay (Maass et al. 2004). The cause of the epidermal barrier impairment is unclear, although these mice show aberrant filaggrin processing.

E-cadherin is one of the adhesive components of AJs and is important for tissue morphogenesis and polarity (Perez-Moreno et al. 2003; Yonemura 2011). Keratinocyte-specific E-cadherin-deficient mice show intact outside-in barrier function (Tunggal et al. 2005). However, these mice have impaired inside-out barrier (shown by TEWL measurement and biotin penetration assay) because of aberrant tight junction proteins possibly resulting from the loss or reduced presence of activated atypical PKC (Tunggal et al. 2005).

Lipids

Corneocytes and CEs are surrounded by intercellular lipids (ceramides, cholesterols, and free fatty acids) that function as a potent epidermal barrier. Lipids are stored in lamellar bodies in stratum granulosum (SG) keratinocytes and are secreted at the SC–SG interface through cornification steps. Perturbations in lipid metabolism and transport lead to severe barrier dysfunction in humans and animals. Several review papers have recently addressed lipid metabolism disorders (Elias et al. 2008b, 2011, 2012; Feingold and Jiang 2011; Rizzo 2011). In addition to discussing the major players in lipid metabolism and diseases caused by dysfunctional metabolic pathways, this section addresses recent advances in our understanding of the roles of enzymes in facilitating very-long-chain fatty acid elongation and in synthesizing ceramide on the epidermal barrier.

Free fatty acids are transported across cell membranes in several ways (Khnykin et al. 2011). FATP4, encoding fatty acid transport protein 4, was reported to be the causative gene of ichthyosis prematurity syndrome in humans (Klar et al. 2009). The “wrinkle-free” mouse is a natural mutant of Fatp4 that is characterized by neonatal lethality, barrier defects and absence of wrinkles (Moulson et al. 2003). Fatp4-null mice were also generated and confirmed to have phenotypes similar to those of wrinkle-free mice, including outside-in and inside-out epidermal barrier dysfunction (Herrmann et al. 2003). Transgenic mice in which Fatp4 deficiency is inducible in the epidermis show impaired epidermal barrier (Herrmann et al. 2005). In contrast, keratinocyte-specific overexpression of Fatp4 restores the skin phenotype of wrinkle-free mice (Moulson et al. 2007). As the proportion of very-long-chain fatty acids (VLFA; ≥C26) in epidermal ceramides is significantly decreased in wrinkle-free mice and Fatp4-null mice and the reduced VLFA in these mice is rescued through keratinocyte-specific overexpression of Fatp4, the fatty acid chain length of skin lipids is thought to be regulated by Fatp4 activity to maintain normal skin integrity (Herrmann et al. 2003; Moulson et al. 2007).

ABCA12 is a member of the superfamily of ATP-binding cassette (ABC) transporters (Annilo et al. 2002). ABCA12 (ABC transporter A12) has been characterized as a key protein in keratinocyte lipid transport (Akiyama 2011). ABCA12 is expressed in lamellar bodies and is thought to mediate lipid transport into extracellular spaces via lamellar bodies to establish the lipid barrier in cornified layers (Akiyama et al. 2005). ABCA12 mutations have been reported in lamellar ichthyosis, congenital ichthyosiform erythroderma, and harlequin ichthyosis in humans (Lefevre et al. 2003; Akiyama et al. 2005; Kelsell et al. 2005; Natsuga et al. 2007; Akiyama 2010). Abca12-deficient mice show outside-in and inside-out barrier defects and neonatal lethality, which reproduces the phenotype of harlequin ichthyosis, the most severe congenital ichthyosis subtype, in humans (Smyth et al. 2008; Yanagi et al. 2008; Zuo et al. 2008). Abca12-deficient mice have defective lipid efflux represented as decreased ω-O-acylceramides as well as increased accumulation of glucosylceramides (Smyth et al. 2008; Zuo et al. 2008). This defective lipid efflux leads to malformation of the intercellular lipid layers, which mostly accounts for the severe barrier defects in Abca-deficient skin, although aberrant filaggrin processing and fragile CE are also observed in Abca12-deficient mice (Smyth et al. 2008; Yanagi et al. 2008; Zuo et al. 2008; Akiyama 2011).

The hydrolysis of glucosylceramides to ceramides is catalyzed by β-glucocerebrosidase (Beutler 1992). Mutations in the human β-glucocerebrosidase gene cause Gaucher’s disease type II, which is characterized by cranial nerve and brain abnormalities and hepatosplenomegaly (Tsuji et al. 1987). Targeted disruption of murine β-glucocerebrosidase leads to outside-in and inside-out barrier dysfunction, highlighting the importance of the content of glucosylceramides and ceramides in the intercellular lipid layers (Holleran et al. 1994). This fact is corroborated by the disrupted water permeability barrier seen in tamoxifen-induced deletion of glucosylceramide-synthesizing enzyme UDP-glucose ceramide glucosyltransferase (UGCG) in keratinocytes (Amen et al. 2013).

The ceramide synthases (CerS) are a group of enzymes that acylate dihydrosphingosine to produce dihydroceramide (Mizutani et al. 2009). Among CerS subtypes, CerS3 is involved in the production of ceramides with VLFA (≥C26) (Jennemann et al. 2012). Ablation of CerS3 causes neonatal lethality and severe skin barrier defects (both outside-in and inside-out), whereas CerS2-deficiency produces phenotypically normal mice (Jennemann et al. 2012). These results indicate the essential role of ceramides with very-long-chain acyl moieties in epidermal barrier function. It is not surprising that mutations in CERS3 were recently reported to be responsible for human autosomal recessive congenital ichthyosis (Eckl et al. 2013; Radner et al. 2013).

As observed in CerS3-deficient mice, VLFA have been implicated in skin development and barrier formation (Kihara 2012). Fatty acid elongation steps involve the production of 3-ketoacyl-CoA from acyl-CoA and malonyl-CoA (Jakobsson et al. 2006). This reaction is catalyzed by fatty acids elongases (Elovl1-7 in mammals) (Kihara 2012). Among Elovl family members, Elovl1 and Elovl4 have been shown to be involved in the elongation of VLFA (Guillou et al. 2010; Ohno et al. 2010). Elovl4-deficient mice show neonatal lethality as well as severe barrier dysfunction (outside-in), which might be explained by depletion of VLFA (≥C28) in both the ceramide and free fatty acid fractions in the epidermis and by loss of ω-O-acylceramides, which are the major components of the hydrophobic lipid bilayers between corneocytes (Vasireddy et al. 2007). Mutations in ELOVL4 were reported in patients suffering from ichthyosis with intellectual disability and spastic quadriplegia (Aldahmesh et al. 2011), the symptoms of which are very similar to those of Sjogren-Larsson syndrome (De Laurenzi et al. 1996). Moreover, Elovl1-deficient mice also show neonatal lethality and outside-in/inside-out epidermal barrier defects with depletion of VLFA (≥C26), and they also show increases in ≤C24 fatty acids (Sassa et al. 2013). Interestingly, Elovl1 was shown to coordinate with CerS3 to produce VLFA (≥C26) (Sassa et al. 2013). Human ELOVL1 mutations have not been reported.

Mutations in ALOXE3 (encoding lipoxygenase-3) and ALOX12B (encoding 12(R)-lipoxygenase) have been detected in CIE patients through linkage analysis (Jobard et al. 2002). Lipoxygenase-3 and 12(R)-lipoxygenase have been implicated in the catalysis of oxygenation of linoleate followed by the esterase-catalyzed hydrolysis of the oxidized linoleate, eventually leading to covalent coupling of the lipids to the cross-linked proteins of the CE (Zheng et al. 2011). Aloxe3-deficient and Alox12b-deficient mice show severe barrier defects (both outside-in and inside-out) accompanied by altered lipid composition (Epp et al. 2007; Moran et al. 2007; Krieg et al. 2013).

Dorfman-Chanarin syndrome is a rare autosomal recessive disorder characterized by congenital ichthyosiform erythroderma and intracellular accumulation of triacylglycerol droplets in many tissues (Dorfman et al. 1974; Chanarin et al. 1975). Linkage analysis identified CGI-58 (also called ABHD5) as a causal gene for Dorfman-Chanarin syndrome (Lefevre et al. 2001). CGI-58 recombinant protein was shown to show acyltransferase activity on lysophosphatidic acid (Ghosh et al. 2008). Cgi-58-deficient-mice have severe outside-in barrier defects as well as triacylglycerol accumulation and depletion of acylceramide (Radner et al. 2010).

Proteases and Protease Inhibitors

Serine proteases have been implicated in the process of keratinocyte differentiation and desquamation (Ovaere et al. 2009). The main roles of serine proteases are (1) the processing of profilaggrin into filaggrin monomer and natural moisturizing factor, and (2) the degradation of junctional proteins (desmosomes and corneodesmosomes) during desquamation steps. The cascades and main players in the proteases and inhibitors have been discussed in a recent review paper (Ovaere et al. 2009).

The first breakthrough that gained attention was the discovery of SPINK5 (serine protease inhibitor Kazal-type 5) mutations in Netherton syndrome patients (Chavanas et al. 2000; Hovnanian 2013). Netherton syndrome is a rare genodermatosis that is characterized by trichorrhexis nodosa, congenital ichthyosiform erythroderma, and atopic diathesis (Hovnanian 2013). SPINK5 encodes LECTI (lymphoepithelial Kazal-type inhibitor 5) expressed in stratified epithelia (Magert et al. 1999; Bitoun et al. 2003). Spink5-deficient mice show outside-in and inside-out epidermal barrier dysfunction as well as the following: (1) increase in the proteolytic processing of profilaggrin into filaggrin monomers (Descargues et al. 2005; Hewett et al. 2005), (2) abnormal degradation of desmoglein-1 and desmoplakin (Descargues et al. 2005), and (3) premature proteolysis of corneodesmosin (Yang et al. 2004; Descargues et al. 2005). Kallikreins are a family of serine proteases, and many kallikrein members are expressed in human epidermis (Lundwall and Brattsand 2008). As LEKTI is a potent inhibitor of multiple serine proteases, Spink5-deficient epidermis leads to kallikrein 5 (KLK5) and KLK7 hyperactivity, accounting for the degradation of desmosomal proteins (Descargues et al. 2005). The fact that transgenic mice in which KLK7 is overexpressed have increased TEWL confirms the role of LEKTI in epidermal barrier formation (Ny and Egelrud 2004). Furthermore, another target of LEKTI was found to be elastase-2 (Ela2) through mass spectrometry analysis (Bonnart et al. 2010). Overexpression of Ela2 under the control of involucrin promoter produces both outside-in and inside-out barrier dysfunction in mice (Bonnart et al. 2010).

Matriptase and prostatin are serine proteases that are known to be involved in epidermal terminal differentiation (Netzel-Arnett et al. 2006; Ovaere et al. 2009). Matriptase is thought to activate prostatin to mediate subsequent processes that lead to filaggrin monomer formation and the generation of natural moisturizing factors (NMFs) (Ovaere et al. 2009; Miller and List 2013). Mutations in ST14 encoding matriptase have been reported in patients with autosomal recessive congenital ichthyosis with hypotrichosis (Basel-Vanagaite et al. 2007; Alef et al. 2009). Matriptase-deficient mice show outside-in and inside-out barrier defects accompanied by hypoplastic lamellar bodies and loss of filaggrin monomer (List et al. 2002, 2003). Furthermore, matriptase activates prokallikreins (proKLK5 and proKLK7), whereas LEKTI targets neither matriptase nor prostatin (Sales et al. 2010). In accordance with this fact, the epidermal barrier defects in Spink5-deficient mice are compensated for by abrasion of matriptase (Sales et al. 2010).

Similar to matriptase-deficient mice, prostasin-deficient mice have both outside-in and inside-out barrier defects (Leyvraz et al. 2005; Netzel-Arnett et al. 2006) that are consistent with the fact that frizzy mouse and hairless rat, natural mutants of Prss8 (encoding prostasin), show abnormal barrier function (Spacek et al. 2010; Frateschi et al. 2012), although PRSS8 mutations in human patients have not been reported. It is noteworthy that overexpression of prostasin in basal keratinocytes also leads to elevated TEWL in the transgenic mice (Frateschi et al. 2011). These facts indicate that a balance between proteases and protease inhibitors is crucial for epidermal barrier formation.

Caspase-14, an aspartate-specific protease, has recently been implicated in filaggrin degradation to NMFs (Hoste et al. 2011). TEWL in caspase-14 deficient mice is elevated compared with controls, although the outside-in barrier is intact (Denecker et al. 2007). Caspase-14 gene mutations have not been documented in human ichthyosis patients.

Antimicrobial Peptides

Skin faces external pathogens as a surface tissue; thus, it expresses various kinds of antimicrobial peptides (AMPs) constitutively or in response to physical stresses. Excellent review papers have been published on AMPs (Lai and Gallo 2009; Gallo and Hooper 2012). This section briefly discusses key findings on AMPs and skin barrier function.

Many proteins have been listed as having antimicrobial properties in skin, including β defensins, cathelicidin-related antimicrobial peptide (Cramp in mice, LL-37 in humans), S100A7 and S100A8 (Lai and Gallo 2009). Cramp was shown to protect skin from infection by invasive bacteria such as group A Streptococcus pyogens, as revealed by infection experiments on Cramp-deficient mice (Nizet et al. 2001). Cramp is stored in lamellar bodies in granular layer keratinocytes and secreted at the SC–SG interface (Braff et al. 2005). Although Cramp-deficient mice show normal levels of TEWL, lanthanum penetration assay reveals inside-out barrier dysfunction of Cramp-deficient epidermis in electron microscopy (Aberg et al. 2008). This is explained by abnormal lamellar body formation and immature intercellular lipid layers in Cramp-deficient mice (Aberg et al. 2008). This is consistent with the relationship between Cramp expression pattern and barrier status. In a recent paper, techniques for impairing barrier function, including psychological stress, were reported to decrease Cramp expression in the skin (Rodriguez-Martin et al. 2011).

Transcription Factor

Although various kinds of enzymes, structural proteins and lipid metabolisms are involved in epidermal barrier formation, some transcription factors are known to regulate those players and eventually govern the development of the epidermal barrier.

The first example is krippel-like factor 4 (Klf4). Klf4 is expressed in the epidermis, and Klf4-deficient mice were reported to have outside-in barrier defect and neonatal death (Segre et al. 1999). Microarray analysis of wild-type and Klf4-deficient mice identifies connexin 26 (Cx26) as a strongly upregulated gene in Klf4-deficient epidermis. Klf4 binds to the proximal promoter of Cx26 and regulates its transcription level (Djalilian et al. 2006). Cx26 overexpression under involucrin promoter in mice produces severe barrier defects that mimic those of Klf4-deficient mice (Djalilian et al. 2006). In contrast, induced expression of Klf4 under keratin 5 promoter accelerates epidermal barrier formation in mice (Jaubert et al. 2003). These results indicate that Klf4 is one of the major transcription factors that decide the fate of epidermal barrier development in the fetus.

Gata3 is a member of the GATA family of transcription factors in the epidermis (Kaufman et al. 2003). Agpat5, a lipid acyltransferase gene, is the direct target of Gata3 (de Guzman Strong et al. 2006). Keratinocyte-specific Gata3-deficient mice show outside-in and inside-out barrier defects together with decreased ω-O-acylceramides (de Guzman Strong et al. 2006).

Taf10 is a component of the TFIID complex, which is essential for the initiation of transcription by RNA polymerase II (Mohan et al. 2003). Keratinocyte-specific ablation of Taf10 in mice reveals severe outside-in and inside-out barrier dysfunction, which might be attributable to the fact that Taf10 regulates the expression of Tgm1, Klf4, and Cldn1(Indra et al. 2005).

Grainy head-like 3 (Grhl3) is another example of a transcription factor that is involved in epidermal barrier formation. Grhl3-deficient mice have severe outside-in and inside-out barrier defects, which is corroborated by the fact that Grhl3 controls the expression of Tgm1 (Ting et al. 2005).

Aryl hydrocarbon receptor nuclear translocator (Arnt) is a basic protein that belongs to the Period-Arnt-single-minded (PAS) family. Keratinocyte-specific Arnt-deficient mice show neonatal death and severe outside-in and inside-out barrier dysfunction (Takagi et al. 2003; Geng et al. 2006). Lipid composition is perturbed in these mice. Arnt regulates the expression of epidermal differentiation genes, although the direct target of Arnt has not been clarified (Robertson et al. 2012).

CCAAT/enhancer binding protein (C/EBP) α and β are transcription factors expressed in the epidermis (Maytin and Habener 1998; Oh and Smart 1998). Whereas C/EBP-β-deficient mice had mild epidermal hyperplasia, epidermis-specific C/EBP-α-deficient mice showed no skin phenotypic abnormalities in steady state (Zhu et al. 1999; Loomis et al. 2007). Keratinocyte-specific ablation of both C/EBP-α and C/EBP-β revealed hyperproliferation of basal kerationocytes and halted terminal differentiation (Lopez et al. 2009). These mice showed abnormal inside-out skin barrier function but normal outside-in barrier (Lopez et al. 2009).

CONCLUDING REMARKS

This article has briefly reviewed skin-barrier-defective animal models and related human diseases. As seen in the section on lipids, disturbance of lipid metabolism is detrimental and leads to early demise in transgenic mice. Even when epidermal proteases or SC components are ablated, the alteration of lipid composition is observed. Not only do these results highlight the central role of lipids in the epidermal barrier, but they also show the complexity of barrier formation. Furthermore, the epidermal barrier is maintained by a fine balance among the major players; an imbalance of protease/protease inhibitors and overexpression/ablation of tight junction proteins leads to barrier dysfunction.

Recent studies have indicated that the barrier is not simply a fortress keeping pathogens out, but is an integrated interface composed of SCs, keratinocytes structural components and immune cells mediating antigen presentation (Kubo et al. 2012). This is exemplified in AD, which is characterized by epidermal barrier dysfunction and cutaneous sensitization. Nevertheless, the nature of AD remains unclear, because barrier defects allow the penetration of allergens leading to inflammation and inflammation can induce skin barrier dysfunction (Kubo et al. 2012). The key to solving this complicated problem might be in the observation that AD patients with high IgE levels have FLG mutations more frequently than AD patients with low IgE levels (Kabashima-Kubo et al. 2012). This suggests that primary barrier defects induce allergic reaction in a subclass of AD (extrinsic AD), whereas other AD patients have primary immunological abnormalities without barrier impairment (intrinsic AD).

As many mutant mice have become available from library sources (Ayadi et al. 2012), we will be able to add many more genes to the list of players in epidermal barrier formation in the near future. Future tasks in research on epidermal barrier function are not limited to the identification of unknown barrier components. Skin infection and epidermal barrier dysfunction promise to rank among the hottest topics in the context of atopic dermatitis, in which staphylococcal infections are frequent (Kong et al. 2012). In fact, one barrier-defective mouse has been shown to have altered skin microbiota compared with controls (Scharschmidt et al. 2009a). This research field will be supported by the Human Microbiome Project (2012a,b).

Another topic will be the relationship between epidermal barrier function and predisposition to skin cancer. Although there is controversy over the risk of skin cancer in atopic dermatitis patients, ichthyosis patients (especially with those Keratitis-Ichthyosis-Deafness (KID) syndrome, congenital ichthyosiform erythroderma/lamellar ichthyosis, or Netherton syndrome) might have increased risk of skin cancers, including squamous cell carcinoma (Natsuga et al. 2011). Carcinogenesis experiments on barrier-defective mice could provide clues to understanding the pathomechanisms underlying skin cancer in inherited ichthyoses, given that the model mice do not show neonatal lethality.

From the perspective of managing ichthyosis disorders, we have only conservative treatments, such as emollients. Once the mechanisms of epidermal barrier formation are further elucidated, novel and more sophisticated treatment options will be developed.