Artificial skin in perspective: concepts and applications

Authors

  • Carla A. Brohem,

    1.  Department of Clinical Chemistry & Toxicology, School of Pharmaceutical Sciences, University of São Paulo, São Paulo, Brazil
    Search for more papers by this author
  • Laura B. da Silva Cardeal,

    1.  Department of Clinical Chemistry & Toxicology, School of Pharmaceutical Sciences, University of São Paulo, São Paulo, Brazil
    Search for more papers by this author
  • Manoela Tiago,

    1.  Department of Clinical Chemistry & Toxicology, School of Pharmaceutical Sciences, University of São Paulo, São Paulo, Brazil
    Search for more papers by this author
  • María S. Soengas,

    1.  Melanoma Group, Centro Nacional de Investigaciones Oncológicas (Spanish National Cancer Research Center) Madrid, Spain
    Search for more papers by this author
  • Silvia B. de Moraes Barros,

    1.  Department of Clinical Chemistry & Toxicology, School of Pharmaceutical Sciences, University of São Paulo, São Paulo, Brazil
    Search for more papers by this author
  • Silvya S. Maria-Engler

    1.  Department of Clinical Chemistry & Toxicology, School of Pharmaceutical Sciences, University of São Paulo, São Paulo, Brazil
    Search for more papers by this author

S. S. Maria-Engler, e-mail: silvya@usp.br

Summary

Skin, the largest organ of the human body, is organized into an elaborate layered structure consisting mainly of the outermost epidermis and the underlying dermis. A subcutaneous adipose-storing hypodermis layer and various appendages such as hair follicles, sweat glands, sebaceous glands, nerves, lymphatics, and blood vessels are also present in the skin. These multiple components of the skin ensure survival by carrying out critical functions such as protection, thermoregulation, excretion, absorption, metabolic functions, sensation, evaporation management, and aesthetics. The study of how these biological functions are performed is critical to our understanding of basic skin biology such as regulation of pigmentation and wound repair. Impairment of any of these functions may lead to pathogenic alterations, including skin cancers. Therefore, the development of genetically controlled and well characterized skin models can have important implications, not only for scientists and physicians, but also for manufacturers, consumers, governing regulatory boards and animal welfare organizations. As cells making up human skin tissue grow within an organized three-dimensional (3D) matrix surrounded by neighboring cells, standard monolayer (2D) cell cultures do not recapitulate the physiological architecture of the skin. Several types of human skin recombinants, also called artificial skin, that provide this critical 3D structure have now been reconstructed in vitro. This review contemplates the use of these organotypic skin models in different applications, including substitutes to animal testing.

General architecture of the skin

Epidermal/dermal layers

The organization of the skin into epidermal and dermal layers that differ in thickness, strength, and flexibility allows for a structured architecture that provides a variety of skin functions. The outermost layer of the skin, known as the epidermis, serves as an impermeable boundary between the environment and the body. The underlying dermis is formed of strong connective tissue, which is rich in collagen and confers its characteristic flexibility to skin. Epidermis and dermis are separated by the extracellular matrix (ECM), known as the basal lamina (Ajani et al., 2007; Balasubramani et al., 2001; Godin and Touitou, 2007; Horch et al., 2005).

The epidermal layer, derived from embryonic ectoderm, is made up of cells generated by proliferating keratinocytes of the stratum basale that move upwards while they differentiate (Figure 1). The continuous process of proliferation, differentiation and, ultimately, cell death and shedding, allows compartmentalization into a number of strata representing different stages of keratinocyte maturation (Balasubramani et al., 2001; Schulz et al., 2000; Stark et al., 2004a). Besides keratinocytes, which account for about 80% of epidermal cells, the epidermis is also composed of the pigment-producing melanocytes, Merkel cells, which are thought to play a sensory role (Boyce and Warden, 2002; Feliciani et al., 1996), and specialized dendritic Langerhans cells, which have an essential role in the skin immune defense system (Phillips, 1998; Régnier et al., 1998, 1999).

Figure 1.

 Drawing showing skin components and layers. Epidermis: containing melanocytes and keratinocytes that are able to differentiate and form the different strata (corneum, lucidum, granulosum and germinativum). Dermis: formed by fibroblasts embedded in a matrix. Hypodermis: containing the adipose tissue.

As shown in Figure 1, ongoing keratinocyte cell division begins in the innermost stratum basal layer of the stratum germinativum and pushes daughter cells apically upwards toward the next spinous stratum, where cells become dense. As they progress toward the interface with the environment, they move into the granulosum stratum, where they accumulate lipid granules, critical for maintenance of the water barrier. The loss of the nucleus in differentiating keratinocytes now leads to a flattened or horny morphology, with only keratin remaining. Pigmentation is imparted by the addition of melanin, produced by melanocytes and transferred to keratinocytes in the final sublayer of the stratum lucidum. The most external layer, known as the stratum corneum, represents the final result of keratinocyte differentiation, and is formed by completely differentiated dead keratinocytes interspersed with intercellular lipids (mainly ceramides and sphingolipids) (Ajani et al., 2007).

Basement membrane and other constituents of the skin

Epidermal development during embryogenesis, wound healing, and regeneration, as well as normal homeostatic maintenance depends on interactions between the epithelia and the connective dermal tissue (Ajani et al., 2007; Balasubramani et al., 2001; Schulz et al., 2000; Stark et al., 2004a). The basal stratum of the epidermis is intimately related to the underlying dermis through a system of protein fiber connections along a micro-architecture network of rete ridges. These constitute a well characterized basement membrane, which anchors the epidermal layer to the loose connective tissue of the underlying dermis. This basement membrane also stabilizes the overlying epidermis mechanically by means of its hemi-desmosomal structure (Balasubramani et al., 2001; El Ghalbzouri et al., 2005; Marinkovich et al., 1992; Moll and Moll, 1998; Schulz et al., 2000).

Basement membrane is mainly made up of collagen IV, laminin, proteoglycans, and glycosaminoglycans, as well as growth factors with a wide variety of biological activities. This dermo-epidermal basement membrane strictly controls the traffic of bioactive molecules in both directions and is able to bind a variety of cytokines and growth factors, thus representing a reservoir for controlled release during physiological remodeling or repair processes (Iozzo, 2005). The dermal matrix, which lies underneath the basement membrane, provides energy and nutrition to the overlying epidermis and imparts considerable strength to skin by virtue of the arrangement of collagen fibers. The collagenous mesh work is interlaced with elastin fibers, fibronectin, proteoglycans, glycosaminoglycans (GAGs), predominantly hyaluronic acid, and other components (Balasubramani et al., 2001). In addition to the dense matrix there are dermal fibroblasts, cells of the immune system, nerve endings, sweat and sebaceous glands, hair follicles, blood vessels, and endothelial cells. The dermal fibroblasts are known to have numerous functions in the synthesis and deposition of ECM components. Moreover, fibroblasts play an important role in proliferation and migration, as well as autocrine and paracrine interactions with neighboring cells (Wong et al., 2007).

The need for in vitro 3D skin models

Interactions between an individual cell, its immediate neighbors, and the ECM are responsible for the control of cell behavior (Bissell et al., 1982; Grinnell, 1976, 2008; Lin and Bissell, 1993; Smalley et al., 2006; Yang et al., 1987). Therefore, cells grown in 2D monolayers cannot capture the relevant complexity of the in vivo micro-environment (Mazzoleni et al., 2009). For example, it has long been suggested that cells cultured on 2D substrates such as culture plates, lose a myriad of important signals, key regulators, and tissue phenotypes. Cells growing in 3D have different cell surface receptor expression, proliferative capacity, extracellular matrix synthesis, cell density, and metabolic functions (Bissell et al., 1982; Grinnell, 1976, 2008; Horning et al., 2008; Lin and Bissell, 1993; Mazzoleni et al., 2009; Smalley et al., 2006; Yang et al., 1987). Thus, 2D monolayer models not only fail to reproduce the complex and dynamic environments of in vivo tissues, but can also to some degree trigger false findings by forcing cells to adapt to an artificial, flat, and rigid surface. Growing numbers of studies report differences in phenotype, cellular signaling, cell migration, and drug responses when the same cells are grown under 2D as opposed to 3D culture conditions (reviewed by Mazzoleni et al., 2009).

In vitro studies have shown that dermal fibroblasts secrete soluble factors that diffuse to the overlying epidermis and can influence keratinocytes to induce the production of basement membrane proteins or melanogenic factors (Balasubramani et al., 2001; El Ghalbzouri et al., 2002a; Wong et al., 2007). Keratinocytes in monoculture produce only a thin epidermal layer and, without mesenchymal support, undergo apoptosis after about 2 weeks in culture (Wong et al., 2007). Dermal fibroblasts promote not only keratinocyte proliferation, but also the development of identifiable keratinocyte layers. Consequently, upon combination of postmitotic dermal fibroblasts (feeder cells) and epidermal keratinocytes, properly stratified epithelia fail to form in simple 2D feeder-layer co-cultures. Only in advanced 3D in vitro systems do keratinocytes develop well ordered epithelia (El Ghalbzouri et al., 2002a; Stark et al., 2004a; Wong et al., 2007), thus offering an opportunity to more closely recapitulate artificial human skin.

Therefore, strategies that allow the reconstruction of artificial human skin equivalents in a 3D setting, including both dermal and epidermal components, would provide versatility as well as answers to physiological questions that cannot be answered solely in the context of monolayer tissue culture. Models that recapitulate the basic architecture of the human skin would supply sites for studying cell–cell interactions and effects of the stromal environment in the regulation of melanogenesis, proliferation, and differentiation of keratinocytes, as well as re-epithelialization processes after wounding. These skin models also present a relevant platform for the creation of cosmetic, photoaging, and cancer models, as well as an excellent system for pharmacological analyses. Furthermore, the human skin equivalents can be engineered easily with specific genetic alterations in either dermal or epidermal compartments to determine how the outcome of these modifications may be modulated by the stromal environment. These skin models also present time- and cost-effective alternatives to the use of laboratory animals that are currently being used for such studies.

While murine animals have routinely been used as experimental models for skin biology and skin cancer (Mancuso et al., 2009; Paulitschke et al., 2010; Youssef et al., 2010), the dissimilar architecture of mouse and human skin severely limits this approach. To begin with, the epithelium of fur-covered mouse models is densely packed with hair follicles that are synchronized for the first few months of life. In contrast, humans possess larger interfollicular regions with sparse hair follicles that are asynchronous in nature. Furthermore, adult murine dermis is thin and the epidermis typically comprises only three layers, with a high rate of turnover, whereas human dermis is quite thick and epidermis is generally 6–10 layers thick (Figure 2). Human skin melanoctyes reside in the basal layer of the epidermis, whereas they are located mainly in dermal hair follicles in mice. Additionally, a cutaneous muscle layer, the panniculous carnosus, which is present in mice but non-existent in humans, has led to findings that are not consistent among species (Donahue et al., 1999).

Figure 2.

 Optical microscopy showing the differences between human and mouse skin. Note that the murine dermis is thin and the epidermis comprises typically only 3 layers, whereas human dermis is quite thick and epidermis is generally 6–10 layers thick. HE staining. 60× magnification.

Skin responsiveness and functionality are also distinct in mouse and humans. For instance, wounding responses in mouse skin allow effective regeneration of tissue, whereas damage in human skin leads to scar tissue or hypertrophic dermal skin lesions known as keloids, which are absent in mouse skin (Khorshid, 2005). The function of epithelium is also dissimilar between mouse and human skin. Mouse skin provides less of a water barrier and displays higher percutaneous absorption than that of human skin, thus limiting topical drug-delivery studies (Menon, 2002).

Generation of skin equivalents

Multiple approaches

Several types of human skin equivalents that closely resemble normal human skin tissue have now been successfully reconstructed in vitro (Table 1). These skin equivalents consist mainly of keratinocytes that are cultured on a dermal substitute for approximately 2 weeks at the air–liquid interface, which allows for epidermal differentiation and development of stratified layers (Figure 3) (El Ghalbzouri et al., 2002a,b, 2008; Pruniéras et al., 1983; Stark et al., 2004a). Although major importance is assigned to the epidermal layer of skin equivalents, dermal substitutes are widely used in clinics for severe burns and chronic wounds (Boyce and Warden, 2002; Fimiani et al., 2005). These dermal substitutes include cell-free dermal matrices such as de-epidermized dermis (DED), inert filters, fibroblast-populated collagen matrices, and lyophilized collagen-GAG membranes (Boyce and Warden, 2002; El Ghalbzouri et al., 2002a,b, 2008; Fimiani et al., 2005; Stark et al., 2004a).

Table 1.   Different strategies to generate artificial skin
Dermal layerEpidermal layerSupplemental factorsReferences
PresenceTypeCellular componentsCellular componentsTime of differentiation
+DED
Filter inserts
Collagen gels
NHF (in collagen matrices)NHK
HaCaT
1–3 weeks5% bovine calf serum
0.5 μM hydrocortisone
0.1 μM isoproterenol
0.5 μg/ml insulin
1 ng/ml EGF
0.1 nM cholera toxin
Boelsma et al., 1999
+Collagen type INHFNHK7 days Asselineau et al., 1985
Bernerd and Asselineau, 1997, 1998;
Pageon and Asselineau, 2005
+Rat tail collagen type IMDCK for basement membrane formation (removed before keratinocyte seeding)REK (neonatal rat epidermal kerationcytes)5 days Ajani et al., 2007
+ NHF mouse 3T3 fibroblastNewborn NHK2–3 weeks24.3 μg/ml adenine
5% Fetalclone II
5 μg/ml insulin
0.4 μg/ml hydrocortisone
1 nM cholera toxn
10 ng/ml EGF
Dubé et al., 2010



El Ghalbzouri et al., 2008
Filter insert (covered or not with collagen type IV) NHK17 days5% calf serum
1 μM hydrocortisone
1 μM isoproterenol
0.1 μM insulin
100 μM L-serine
1 μMdL-α-tocopherol-acetate and a lipid supplement containing
25 μM palmitic acid, 15 μM linoleic acid, 7 μM arachidonic acid and 24 μM bovine serum albumin
0.01 μM l-carnitine
1 μM dl-α-tocopherol-acetate
25 μM palmitic acid
30 μM linoleic acid
7 μM arachidonic acid
50 μg/ml ascorbic acid
1 ng/ml EGF
 
+Calf collagen type INHFNHK + NHM (20:1)Over 1 week Liu et al., 2007
+Collagen type I (Gen-col block collagen)NHFNHK + NHM (4:1)7 daysInserts immersedSouto et al., 2009
+DED
Collagen type I
NHF
NMF
NHK + NHM (20:1)8 daysSCF
human HGF
Endothelin 1 or 3
Cario-André et al., 2006
+Collagen type INHF
3T3-MEF
NHK
HaCaT
7–10 days (human)
10–14 days (mouse)
Calf serum
Insulin
EGF
Cholera toxin
Adenine
Hydrocortisone
Ascorbic acid
TGF-α (only in HaCaT)
Stark et al., 2004a
+Hyalograft 3-DNHFNHKUp to 12 weeksFAD medium
10% fetal bovine serum
0.1 M cholera toxin
0.4 μg hydrocortisone
50 μg L-ascorbic acid
200 U/ml/day aprotinin
Stark et al., 2004b, 2006;
Boehnke et al., 2007
+Collagen type I Matrigel NHK2–3 weeksAprotinin solutionSobral et al., 2007
Figure 3.

 Optical microscopy of human facial skin and human artificial skin. The thickness of the epidermis and dermis is very similar. Human artificial skin presents all epidermal layers, showing the differentiation of this epithelium. HE staining. 60× magnification.

Engineered skin bioproducts containing biopolymers are also being used as grafts in the treatment of patients with excised burns, burn scars, and congenital skin lesions (Boyce and Warden, 2002). In addition, materials such as collagen–glycosaminoglycan matrices, allogeneic dermis, and synthetic polymers have been used as replacements for specific skin layers (Balasubramani et al., 2001). Summarizing, there are a variety of ‘skin models’ that can be broadly categorized into (i) those containing only epidermal components, (ii) grafts consisting of dermal components alone, and (iii) full-thickness composite grafts containing both epidermal and dermal components. In the literature, these systems have been referred to as 3D skin, reconstructed skin, skin equivalents, artificial skin, organotypic culture of skin, skin substitutes, or skin grafts. It should be noted that the last two terms are most commonly used for bioengineered products that use human or animal components such as cultured cells or collagen.

The past few years have seen an increase in the number of commercially available skin models: e.g. EpiDermTM (MatTek, Ashland, MA, USA), EpiskinTM (previously manufactured by Episkin, Chaponost, France, and now by L’Oreal, SkinEthic, Nice, France), Apligraf® (Organogenesis Inc., Canton, MA, USA) and the models engineered by SkinEthic (Boelsma et al., 2000; Ponec, 2002; Welss et al., 2004). As these models are used for commercial purposes, morphological studies have been performed to demonstrate a relevant multilayered epithelium, and the presence of characteristic epidermal ultrastructures and markers of epidermal differentiation (e.g. keratins, fillagrin, involucrin) (El Ghalbzouri et al., 2002a,b, 2008).

In an editorial about the development of new skin equivalents, Phillips (1998) describes the historical evolution of skin substitutes. In 1975, Rheinwald and Green described a methodology for the in vitro cultivation and serial subculture of epidermal cells that were able to produce viable keratinocyte sheets. This technique was critical in the development of tissue culture technology, and has improved the cultivation of large quantities of keratinocytes in vitro (Phillips, 1998). Once keratinocytes could be propagated in large quantities, the first step towards successful epidermal differentiation was the exposure of normal human keratinocyte cultures at the air–liquid interface, in order to induce terminal differentiation resulting in a multilayered stratified tissue (Figure 4) (Régnier et al., 1998, 1999). Keratinocyte differentiation and consequent programmed cell death ultimately led to formation of the uppermost stratum corneum in the skin equivalent models, which is formed continuously during this process (Figure 4).

Figure 4.

 Preparation model of artificial skin. First step is the preparation of the dermal equivalent, consisting of type I collagen and fibroblasts. After polymerization of this layer, keratinocytes (K) and melanocytes (M) are plated. At 24 h after plating and after the contraction of collagen gel, the entire structure is transfered to a steel grid to form the air–liquid interface. After 2 weeks on the interface there is the formation of artificial skin. HE staining. 40× magnification.

Studies on the effectiveness, metabolic transformation, and potential pathologic effects of a huge variety of topical products have only been possible in vitro due to the presence of the stratum corneum in the reconstructed epidermis of artificial skin (Ajani et al., 2007; Boyce and Warden, 2002; El Ghalbzouri et al., 2008; Ponec, 2002). For safety reasons, multiple tests for topical products have been developed to detect changes in the integrity, morphology, viability, and release of pro-inflammatory mediators in this layer (Ponec, 2002). One point to be mentioned is that after a prolonged time in culture (6–7 weeks), the in vitro epidermis remains viable and displays all the signs of a normal differentiation program, but the thickness of the stratum corneum gradually increases with time. This can be a critical issue in some tests, as the gradual thickening of the stratum corneum may lead to higher resistance to environmental stimuli (Ponec, 2002). Thus, these models remain imperfect, and may be intrinsically variable. Nevertheless, despite these limitations, they have already provided much information about dermal–epidermal interactions, cell–cell and cell–matrix interactions, responses of dermal and epithelial cells to biological signals and pharmacological agents, as well as effects of drugs and growth factors on skin reconstruction processes (Fimiani et al., 2005).

Importance of the microenviroment

ECM

The ECM, a gel-like medium produced by the surrounding fibroblasts, is the largest component of normal skin and confers to the skin its unique properties of elasticity, tensile strength, and compressibility. Both dermal fibroblasts and epidermal cells secrete the two critical classes of ECM molecules – fibrous proteins and proteoglycans. Strength, flexibility, and resilience are imparted by fibrous structural proteins, including collagens, elastin, and laminin. In turn, highly hydrated proteoglycans provide cushioning support to cells in the ECM.

The ECM can contribute to the microenvironment specifically through its mechanical features, providing support and anchorage for cells. The ECM also influences tissue segregation as well as regulation of intracellular communication via signaling pathways and its ability to bind growth factors, enzymes, and other diffusible molecules. Interactions between cells and the ECM are extremely important for processes such as normal cell growth and differentiation. For example, in acute wounds, the provisional wound matrix, containing fibrin and fibronectin, provides a scaffolding to direct cells into the injury, as well as stimulating them to proliferate, differentiate, and synthesize new ECM.

An increasing number of studies have identified that it is essential to maintain the composition and structural organization of the ECM (i.e. the number of cells in the dermal layer) to obtain normal skin tissue organization in 3D models (Chioni and Grose, 2008). The cells of the underlying dermal layer do not act simply as a structural passive framework, but rather influence epithelial migration, differentiation, growth, and attachment (Cooper et al., 1991; Phillips, 1998). In the case of skin reconstructs, a living dermal component can be vital to the cultured skin, and therefore the addition of a dermal element should be carefully considered in the design of any such product. For example, the composite graft (epidermal and dermal elements) showed significant advantages over the epidermal sheet graft in the closure of full-thickness wounds (Cooper et al., 1991).

Dermal substrates

Two types of dermal substrates are used for the generation of reconstructed epidermis: (i) acellular structures, which can be either inert filters or de-epidermized dermis (DED), and (ii) a cellular substrate composed of fibroblast-populated collagen matrix (Ponec et al., 1997). Skin reconstructs have been described as containing only acellular dermal substrate, forming a stratified epidermis of three to four viable cell layers, and must be supplemented with various growth factors including epidermal growth factor (EGF), keratinocyte growth factor (KGF), and/or insulin-like growth factor (IGF). However, keratinocyte proliferation is stimulated and epidermal morphology is improved in the presence of fibroblasts (El Ghalbzouri et al., 2002a,b; Wong et al., 2007) (Table 1). In skin reconstructs, viable fibroblasts are shown to increase epidermal growth and spreading, as well as enhance the formation of basement membranes proteins, thus improving the attachment of the epidermal layer (Cooper et al., 1991).

Fibroblast and its cellular interactions

Another important point to consider when manufacturing skin reconstructs for a long period of time is the contraction of the collagen matrix in the presence of fibroblasts (Bell et al., 1979). The contractility of this matrix depends on the number of fibroblasts and the time-frame from the generation of the dermal layer to the subsequent seeding of keratinocytes (El Ghalbzouri et al., 2002b). Some groups have been trying to improve the skin model by avoiding collagen contraction, while retaining the presence of fibroblasts. El Ghalbzouri et al. (2002b) used a centrifugal seeding method to incorporate different numbers of fibroblasts into the DED. They observed that fibroblast initially had a stimulatory effect on induced keratinocyte proliferation, which decreased at later time points in cell culture. They also showed that the interaction between keratinocytes and fibroblasts induced the production of growth factors by the fibroblasts that provided key signals for the induction of appropriate epidermal differentiation. The factors produced by fibroblasts (influencing keratinocyte proliferation in either a negative or a positive manner) were dependent on fibroblast density and culture time (El Ghalbzouri et al., 2002b).

In an attempt to preserve the epidermal homeostasis and increase the lifespan of skin, avoiding dermal shrinkage and increasing tissue stability, some authors used matrices other rather than collagen as sponges or scaffolds (Auxenfans et al., 2008; Boehnke et al., 2007; Stark et al., 2004b, 2006). In a system that used a dermal equivalent constituted by an esterified hyaluronic material (Hyalograft-3D) colonized with fibroblast, Stark et al. (2004b, 2006) improved long-term growth and differentiation for at least 12 weeks. Differentiation and ultrastructure markers were used to show the viability of the system in studies such as skin regeneration (Boehnke et al., 2007) and homeostasis (Stark et al., 2006).

Other studies have shown the effect of fibroblast density on keratinocyte proliferation and correct epidermal proliferation, by demonstrating that fibroblasts produce collagen types IV and VII, laminin 5 and nidogen, all of which contribute to basement membrane formation. Furthermore, fibroblasts secrete cytokines such as transforming growth factor beta (TGF-β), which stimulates keratinocytes to synthesize basement membrane components, including collagen types IV and VII (Wong et al., 2007).

The importance of the role of fibroblasts in keratinocyte differentiation was also demonstrated, showing that fibroblasts produce different patterns of cytokine release during their differentiation. Moreover, differentiated fibroblasts induce the production of the highest levels of keratinocyte growth factor and TGF-β1 (Nolte et al., 2008). It was also shown that fibroblasts from different body sites have different functional properties, which may affect their suitability for dermal substitutes. As Nolte et al. (2008) pointed out, it is important to consider these facts in future in vivo human studies in tissue-engineered dermal substitutes.

Additional cellular components

Fibroblasts are not the only cells that are important in the dermal layer of a skin reconstruct. The 3D culture can also contain other cell types that will enrich the dermal microenvironment, such as myofibroblasts, endothelial cells, inflammatory cells, and adipocytes. A wound-healing study used human mesenchymal stem cells seeded together with dermal fibroblasts in reconstructed skin to show that human mesenchymal stem cells could contribute to the wound-healing process (Chioni and Grose, 2008).

Addressing skin model limitations

Phototypes

Comparisons of cells in monolayer cultures versus skin reconstructs indicate that the 3D models may also address aspects of skin pigmentation. Melanocytes in skin reconstructs can be present as single cells at the basal layer of the epidermis, reflecting the distribution of human skin (Meier et al., 2000). Furthermore, melanin production is more active in melanocytes growing in 3D than in 2D (Nakazawa et al., 1998), demonstrating autocrine and paracrine signaling networks mainly between both melanocyte and keratinocyte cell–cell interactions in the skin (Costin and Hearing, 2007).

Melanocytes, however, have not been included in standard skin grafts. In these cases, even after successful transplantation, skin grafts remain depigmented and thus present an aesthetical dilemma (Balasubramani et al., 2001; Boyce and Warden, 2002; Liu et al., 2007). As cutaneous pigmentation is a result not only of melanin synthesis by melanocytes, but also of the transfer of melanosomes to neighboring keratinocytes, it is critical that melanocytes establish correct interactions with surrounding keratinocytes (Cario-André et al., 2006; Fitzpatrick and Szabo,1959; Nakazawa et al., 1998; Régnier et al., 1999).

One fact that remains to be discussed is that epidermal pigmentation is dependent upon the phototype of melanocytes. In basal conditions, intra-individual skin pigmentation varies according to the phototype of melanocytes, which is regulated mainly by melanocortin-1 receptor (MCIR) and remains the major determinant of the pigmentation phenotype in skin (Rees, 2003). The MC1R gene encodes a seven-transmembrane G-protein-coupled receptor that regulates the quantity and quality of melanin produced via activation of adenyl cyclase and production of cyclic AMP (Mountjoy et al., 1992). Activation of this signal transduction pathway ultimately leads to activation of various genes, most notably microphthalmia transcription factor (MITF), responsible for the expression of numerous enzymes and differentiation factors of the melanogenic cascade (Levy et al., 2006) as well as the survival of migrating melanoblasts (Steingrímsson et al., 2004).

The different skin phototypes (types I–VI) can to some extent be reproduced in vitro by selecting the donor for melanocyte isolation. Melanocytes derived from a darkly pigmented individual will generally give rise to a reconstructed epidermis phenotypically pigmented, independent of the origin of keratinocytes. For this reason, the use of melanocytes has to be taken into consideration not only in skin grafts for clinical applications but also for the modulation of melanogenesis by pro-pigmenting or de-pigmenting agents (Cario-André et al., 2006; Régnier et al., 1999). The use of pigmented skin equivalents has also been adopted to obtain a better understanding of congenital hyperpigmentary disorders (Okazaki et al., 2005).

Cario-André et al. (2006) have shown that there can also be a dermal modulation of human epidermal pigmentation. Epidermal reconstructs from phototypes II–III were grafted on the back of immunotolerant Swiss nu/nu mice; depending on the presence of colonizing human or mouse fibroblasts, they developed a non-regular pigmentation pattern. The authors also demonstrated that when human white Caucasoid split-thickness skin was xenografted onto the Swiss nu/nu mouse strain it phenotypically appeared black within 3 months, revealing a phototype VI pattern of melanin distribution. This demonstrates that melanocyte proliferation and melanin distribution/degradation can be influenced by fibroblast secretion and acellular dermal connective tissue (Cario-André et al., 2006).

Immortalized cell lines versus primary cells in artificial skin

Primary cells can be used to evaluate differences in the epithelial maturation depending on the phototype of the donor and on the anatomical location of the tissue of origin. However, primary cells have a limited lifespan, and the number of cells cannot be increased to produce the large amounts that may be needed to generate multiple reconstructs for statistical analyses. Immortalized cell lines could improve the reproducibility and consistency of skin models, reducing intra- and inter-laboratory variations. Thus, established lines would make tight regulatory procedures possible for biological and clinical studies, as well as industrial applications (Boelsma et al., 1999, Stark et al., 2004a).

The spontaneously immortalized human keratinocyte line, HaCaT, is one of the most frequently used keratinocyte cell lines because of its highly preserved differentiation capacity. HaCaT cells are able to form a differentiated, ordered, structured, and functional epidermis when transplanted onto subcutaneous tissue of athymic mice (Boelsma et al., 1999; Ponec, 2002; Schoop et al., 1999). These HaCaT cells grown at the air–liquid interface initially develop a multilayered epithelium. However, during the course of culture, marked alterations in tissue architecture become apparent, with disordered tissue organization, including the presence of rounded cells with abnormally shaped nuclei (Boelsma et al., 1999; Schoop et al., 1999; Stark et al., 2004a).

Although many studies used HaCaT cells in skin reconstruction procedures, several indicated that HaCaT cells have a limited ability to produce an organized mature cornified epithelium (Boelsma et al., 1999). Despite this, this reconstructed epidermis is often considered functional and is used as a model for elucidation of the molecular mechanisms regulating keratinocyte growth and differentiation, making it popular for use in pharmacotoxicological studies (Schoop et al., 1999; Stark et al., 2004a). Nevertheless, the inability to generate stratum corneum must be considered when cosmetologic applications are the main focus.

The immune system: how to reconstitute this response in vitro?

Another important element to be considered in skin equivalent models is the lack of cells of the immune system. In addition to keratinocytes, supra-basal layers in the human skin contain Langerhans cells (LC), in the proportion of approximately 30 000 cells/cm². These cells correspond to CD34+ dendritic cells of hematopoietic origin, displaying Birkbeck granules and CD1a surface antigens (Facy et al., 2004; Régnier et al., 1997). LC cells are responsible for immune system functions in the skin, capturing allergens of low molecular weight that bind to the skin for possible antigen processing and T-cell recruitment. Allergen-induced epidermis cytokines, such as tumor necrosis factor alpha (TNF-α) and interleukin (IL)-1b, are involved in the migration of these Langerhans cells, which, when stimulated, present CD86 surface markers (Facy et al., 2004; Tiznado-Orozco and Orea-Solano, 2004).

However, the integration of LC into in vitro models has remained a challenge, as these cells, unlike keratinocytes and melanocytes, cannot be subcultured and expanded in vitro (Régnier et al., 1998, 1999). A skin reconstruction composed of both dermal and epidermal layers containing not only keratinocytes, but also the pigmented and immune system constituents, would be an excellent model to assess not only mechanisms of epidermal and dermal cell–cell interactions but also the role of each cell type in the skin immune response (Régnier et al., 1998).

In 1998, Régnier and collaborators demonstrated that in vitro-generated dendritic cells/Langerhans cells from CD34+ hematopoietic progenitors seeded into a reconstructed epidermis gave rise to resident dendritic epidermal LC, expressing major histocompatibility complex (MHC) class II and CD1a antigens, also containing characteristic Birbeck granules. Confocal laser scanning microscopy also revealed a cell morphology identical to that observed for LC in vivo. Interestingly, when purified CD34+ hematopoietic progenitor cells not exposed to factors such as granulocyte-macrophage colony-stimulating factor (GM-CSF) and TNF-α, were co-seeded directly with keratinocytes and melanocytes onto the dermal equivalent, the keratinocytes were able to induce the maturation of these hematopoietic progenitors into epidermal LC (Régnier et al., 1998).

Facy et al. (2004) have now proposed that in a complete reconstructed epidermis, similar to Régnier et al. (1998), the reactivity of Langerhans cells from CD34+-derived dendritic cells is comparable to that in vivo. The authors also call attention to the fact that this complete model has the potential to be produced at an industrial level and may be an alternative model for estimating the sensitizing potential of new chemicals. Facy et al. (2004) pointed out the importance for research studies to investigate Langerhans cell biology in vitro.

Applications

Alternatives to animal testing and regulations

One major concern in the production and use of novel chemical reagents that are to be applied to the skin is their capacity to cause acute skin irritation upon contact. Hence, new procedures have been established for the safe handling, packaging and transport, as well as for general safety assessment of these reagents (Fentem et al., 2001; Ponec, 2002). These reagents are often evaluated for their irritant potential by the application to animals, followed by observations of visible changes including erythema and edema. However, testing for skin irritation in animals is not always predictive of human responses. Furthermore, these irritability tests may cause pain and discomfort to the experimental animals (Boelsma et al., 1999; Ponec, 2002; Welss et al., 2004).

The European Cosmetics Association (COLIPA) is committed to replacing animal testing as well as improving the prediction of irritants in the cosmetic and toiletry industry without compromising human safety or environmental protection (de Silva, 2002; Welss et al., 2004). In 1992, COLIPA coordinated the efforts of the cosmetics industry to develop alternative methods to animal testing for the safety assessment of cosmetics and hence created the Steering Committee on Alternatives to Animal Testing (SCAAT) (de Silva and European Federation of the Cosmetics Industry, Steering Committee on Alternatives to Animal Testing, 2002). In 1994, the European Centre for the Validation of Alternative Methods (ECVAM) organized a workshop on the potential use of non-invasive methods in the safety assessment of cosmetic products, and COLIPA created a specific task force to create guidelines that would reflect the protocols and practices of industry (Boelsma et al., 2000; Fentem et al., 2001; de Silva and European Federation of the Cosmetics Industry, Steering Committee on Alternatives to Animal Testing, 2002; Welss et al., 2004).

Skin reconstructs are now being proposed as an alternative method to animal testing for assessing irritation, corrosiveness, phototoxicity, and genotoxicity of various reagents (Curren et al., 2006; Hu et al., 2009; Welss et al., 2004). These reconstructs are favored over monolayer keratinocyte cultures for multiple reasons, including the ability to test compounds with low water-solubility, as well as the fact that the concentrations inducing irritant responses in monolayer keratinocyte cell cultures are extremely different than concentrations in vivo, and the cultures typically obtained from monolayer cell tests have been difficult to correlate with the in vivo situation (MacNeil, 2007; de Silva and European Federation of the Cosmetics Industry, Steering Committee on Alternatives to Animal Testing, 2002; Welss et al., 2004). Percutaneous absorption, which cannot be properly assessed in monolayer conditions, is another critical factor in considering the safety assessment of cosmetic ingredients, as this governs the margin of safety for the ingredients studied (de Silva and European Federation of the Cosmetics Industry, Steering Committee on Alternatives to Animal Testing, 2002).

Once skin reconstructs appeared as an alternative method to animal testing, various companies provided reconstituted human epidermal in vitro skin equivalents, and the number of distributors is increasingly growing.

During 1999 and 2000, ECVAM commissioned a validation study of five in vitro tests for acute skin irritation. This study specifically addressed aspects of protocol refinement (phase I), protocol transfer (phase II), and protocol performance (phase III), in accordance with the scheme defined by ECVAM (Botham, 2004; Fentem et al., 2001; Portes et al., 2002). Results indicated an acceptable intra-laboratory reproducibility. However, inter-laboratory reproducibility was of concern, and the predictive ability of all methods was also inadequate. One explanation for the insufficient reproducibility and false positives is the common formation of an imperfect barrier function in reconstructed epidermis. Furthermore, the use of various experimental protocols which differ in the concentrations of the applied irritants used, the time of exposure, and the time of incubation made it difficult to create a standardized protocol for assessing irritation in vitro (Welss et al., 2004). Protocol refinements and improvements in 2007 enabled two skin reconstruct models to be validated by ECVAM for use as alternative methods in skin irritation tests (Botham, 2004; Fentem and Botham, 2002; Fentem et al., 2001; Hoffmann et al., 2008; Portes et al., 2002).

A promising new use of 3D human skin models for the evaluation of genotoxicity of topically applied compounds and formulations is also underway. Curren et al. (2006) developed a micronucleus assay employing the EpiDerm 3D human skin model, which is now in the process of validation (Hu et al., 2009).

Photoaging model: the UV effect

Sun exposure causes various deleterious cutaneous effects, leading to short-term responses such as erythema, sunburn, and suntan. Long-term effects include skin cancers and premature photoaging (Bernerd and Asselineau, 1998; Bernerd et al., 2000).

Solar UV light reaching Earth is a combination of both UVB (290–320 nm) and UVA (320–400 nm) wavelengths. Although UVB irradiation has received more attention, an increasing number of studies are now emphasizing the harmful effects of UVA (Bernerd and Asselineau, 1998, 2008). However, for ethical reasons, experimental chronic UV exposure assays pose a challenge in humans.

Conventional monolayer cultures do not reproduce accurately the physiological conditions for studying UV exposure (Bernerd and Asselineau, 1998, 2008). Instead, the full 3D skin model composed of dermal and epidermal equivalent layers may be suitable for determining specific biological effects induced by UVB and UVA irradiation (Bernerd and Asselineau, 1998; Bernerd et al., 2000). As the reconstructed skin model is able to differentiate epidermal horny layers, compounds or sunscreens can be topically applied on the skin, mimicking a more realistic situation (Bernerd et al., 2000). For a proper UV-irradiation study of topically applied sunscreens considering the UV effects on photoaging, the use of pigmented reconstructed epidermis containing melanocytes is crucial (Nakazawa et al., 1998; Régnier et al., 1999).

The solar protection factor (SPF) that corresponds to erythema prevention is one of first features for evaluating sunscreen protection. After 24 h of UVB exposure, typical sun-burned cells (SBC) are formed in the epidermis, corresponding to the clinical appearance of erythema. SBC correspond to apoptotic keratinocytes, which allow elimination of cells strongly damaged by UVB irradiation. Previously, erythema could not be assessed in vitro, but sunscreen efficiency can now be measured in reconstructed skin in vitro by observation and counting of UVB-induced SBC (Bernerd et al., 2000).

The use of a full skin reconstruct model in UV studies is essential as the major skin target of UVB is the epidermis, whereas UVA exposure mainly affects the dermis. UVB causes significant alterations in keratinocyte differentiation processes, whereas UVA induces apoptosis in fibroblasts located in the superficial area of the dermal equivalent (Bernerd and Asselineau, 1998, 2008).

Skin penetration is a key point in the evaluation of a potential skin sunscreen. The ability of these compounds to penetrate the skin will depend on the time and concentration required to reach the desired target site. The stratum corneum is the main barrier against skin penetration, and efficient penetration of this barrier depends on specific processes (Ponec, 2002).

The skin reconstructs are also an efficient model not only to test the SFP of a sunscreen, but also to study cell and ECM modifications provoked by photoaging. Photoaged skin contains notorious changes observed in the uppermost epidermal compartment, but also in the deep dermal compartment of the skin, such as degradation of the connective tissue, decrease in collagen content, and accumulation of abnormal elastic tissue characterizing solar elastosis (Bernerd and Asselineau, 1998; Bernerd et al., 2000). Moreover, photoaging is associated with the appearance of advanced glycation end products (AGEs). AGEs are new residues created by cross-linked formations that are produced by a non-enzymatic glycation reaction in the extracellular matrix of the dermis. AGEs are now considered one of the factors responsible for loss of elasticity and other properties of the dermis during aging (Pageon and Asselineau, 2005). In a study that compared the histological results obtained within the reconstructed skin containing native collagen and collagen modified by glycation, no significant differences were observed in morphological structure except for the reduction of dermal thicknesses in the glycated sample. Pageon and Asselineau (2005) also concluded that this system is a promising model to observe the effects of aging on ECM elements and may provide a tool for evaluating the efficacy of AGE inhibitors (Pageon and Asselineau, 2005).

Other studies involving aging, the ECM, and evaluation of new molecules are also utilizing skin reconstructs as a study model. Sok et al. (2008) showed that a C-xylopyranoside derivative, C-β-d-xylopyranoside-2-hydroxy-propane (C-xyloside), induced the neo-synthesis of matrix proteins such as glycosaminoglycans and heparan sulfate proteoglycans, and also restored dermal epidermal junction integrity, suggesting that it may have beneficial effects on aged skin (Sok et al., 2008).

Pharmacological applications

In the field of pharmacology, drug discovery is generally dependent upon the predictive capacity of cell-based assays (Mazzoleni et al., 2009). Most frequently, the efficacy of anti-cancer drugs is tested in 2D monolayer cells cultured on plates during the initial drug development and discovery phase. However, huge differences are observed when these drugs are tested in vivo. These differences may be the result of different cell surface receptors, proliferation kinetics, ECM components, cellular densities, and metabolic functions of 2D-maintained cells (Horning et al., 2008).

As mentioned above, pharmacological penetration is an important limitation in pharmacological compounds that need to reach deep layers of the skin (Godin and Touitou, 2007; Régnier et al., 1993). Skin from cadavers has previously been used in drug transport studies, but limited availability and large variations between specimens have now increased the application potential of skin reconstruct models (Pasonen-Seppänen et al., 2001). The 3D model has permeability characteristics and metabolic activity resembling that of native skin. This is critical, as metabolic activity may affect the permeability of some drugs and their potential for research on irritation, toxicity, and keratinocyte differentiation (Godin and Touitou, 2007; Pasonen-Seppänen et al., 2001; Régnier et al., 1993).

One of the concerns regarding the skin reconstruct model is its lack of skin appendages, including pilosebaceous units, hair follicles, and sweat glands. Because of this lack, this model provides much lower barrier properties than that found in whole skin. Consequently, while the skin reconstruct model is superior to a monolayer model, the kinetic parameters of skin permeation obtained from these studies must still be considered an overestimation when compared to the flux across human skin (Godin and Touitou, 2007).

Skin cancer

Cancer is a heterogeneous disease whose initiation and progression is tightly modulated by cell–cell and cell–matrix interactions. For these reasons, the use of 3D culture models has been steadily increasing in studies of tumor biology (Chioni and Grose, 2008).

In invasive skin cancers such as melanoma, skin reconstructs are very convenient for modeling not only the growth and progression of melanoma cells in a 3D microenvironment, but also for studying the communication among melanoma cells and surrounding epidermal keratinocytes and dermal fibroblasts (Berking and Herlyn, 2001; Smalley et al., 2006). The use of artificial skin has shown that surrounding fibroblasts are recruited by the primary melanoma and provide survival signals in the form of altered ECM deposition and growth factors, as well as stimulating the production of matrix metalloproteinases, promoting tumor cell invasion (Haass and co-workers 2005; Smalley et al., 2006). Fernández et al. (2006) used human artificial skin to test the efficacy of bortezomib, an anti-tumoral drug acting on the proteasome, in a 3D model. Yu and co-workers (2009) evaluated the role of BRAF mutation and p53 inactivation during transformation of a subpopulation of primary human melanocytes, and observed the formation of pigmented lesions reminiscent of in situ melanoma in artificial skin reconstructs. In addition, artificial skin has been used to screen the therapeutic potential of oncolytic adenoviruses in melanocytic cells. Organotypic 3D culture models are also used for different tumor types including breast, prostate, and ovarian cancer (Chioni and Grose, 2008).

Skin models have also been used for genetic and functional analyses of early stages of tumor development. Normal melanocytes in this model remained singly distributed at the basement membrane. In the radial growth phase of melanoma, proliferation and migration of the cancer cells in the dermal reconstruct and tumorigenicity in vivo were observed when cells were transduced with the basic fibroblast growth factor gene. In the vertical growth phase the cells were able to invade the dermis and an irregular basement membrane was formed. In metastatic melanoma, cells rapidly proliferated and aggressively invaded deep into the dermis, in a growth pattern very similar to the pattern in vivo (Meier et al., 2000).

These skin cultures are excellent models to assess melanoma–keratinocyte interactions (Hsu et al., 2000), as well as study keratinocyte-derived lesions.

A study characterized the migratory pattern of a squamous cell carcinoma cell line (HaCaT-II-4) when E-cadherin expression was suppressed. The authors were able to show that loss of cell adhesion enabled migration of single, intra-epithelial tumor cells between normal keratinocytes that were essential for initial stromal invasion (Alt-Holland et al., 2008). Boccardo et al. (2004) used organotypic cultures of human keratinocytes to evaluate the effects of TNF-α in cells that expressed HPV-18 oncogenes. Another example of the utilization of organotypic culture in epithelial tumor models is described by Hoskins et al. (2009), who studied Fanconi anemia (FA). They described the growth and molecular properties of FA-associated cancers (FANCA)-deficient versus FANCA-corrected HPV E6/E7 immortalized keratinocytes in monolayer and organotypic epithelial raft cultures (Hoskins et al., 2009).

Skin disorders and clinical applications

Skin reconstructs are currently being tested and used in clinics for several skin pathologies. Disorders which may benefit from the development of human skin equivalents include psoriasis (Barker et al., 2004; Gazel et al., 2006; Konstantinova et al., 1996; Tjabringa et al., 2008), vitiligo (Bessou et al., 1997; Cario-André et al., 2007), keloids (Butler et al., 2008; Chiu et al., 2005), nevus (Gontier et al., 2002, 2004) and genodermatoses such as xeroderma pigmentosum (Bergoglio et al., 2008; Bernerd et al., 2001; Ergün et al., 2002; Herlin et al., 2009) and epidermolysis bullosa (De Luca et al., 2009; Ferrari et al., 2006; Mavilio et al., 2006; Wong et al., 2007).

The best temporary skin replacement for full-thickness burns, considering structure and function, is the cadaver allograft skin (Hansen et al., 2001; Phillips, 1998; Wong et al., 2007). However, difficulties associated with handling and transport of the material, as well as the possible risk of disease transmission, have limited the availability and use of this model (Hansen et al., 2001; Phillips, 1998; Wong et al., 2007). There has therefore been a rapid increase in the use of split-thickness skin grafts or cultured epidermal grafts in clinical procedures in the past few years (Phillips, 1998; Wong et al., 2007).

While the main use of homologous skin grafts, reconstructs or related bio-products has traditionally been in the treatment of severe burns and skin disorders, various new clinical and experimental approaches in which this type of skin substitute can be useful have recently emerged. The main new clinical indications for skin allografts include skin loss, surgical wounds, and genodermatoses (Bernerd et al., 2001; De Luca et al., 2009; Ergün et al., 2002; Ferrari et al., 2006; Fimiani et al., 2005; Herlin et al., 2009; Wong et al., 2007).

Two key factors were considered essential for the utilization of skin substitutes in clinical applications: the ability to grow keratinocytes in vitro and the increasing practice of early wound excision in the extensively burned patient (Cooper et al., 1991).

Human skin substitutes are now commercially available from different companies with different compositions; however, they present a number of challenges regarding preclinical safety evaluation. In particular, for clinical use, the companies must demonstrate that the skin grafts do not carry pathogens, that they lack immunogenicity, that they show normal physiological functions, and that they have no potential for tumorigenicity (Nemecek and Dayan, 1999).

Final comments

As artificial skin reconstructs are becoming robust and have been validated for many applications, there has been a push towards their use in the clinic (i.e. skin repair following wounding or burning). Besides these applications, organotypic skin provides an amenable setting for functional analyses of genetically altered cells in the context of physiological cell–cell and cell–matrix interactions. There is also a wide range of possibilities for studies of UV-induced carcinogenesis and aging, as well as efficacy and selectivity of chemotherapeutic agents. In addition to melanoma, or frequent tumors of keratinocytes (i.e. basal cell or squamous cell carcinomas), other skin disorders such as vitiligo and psoriasis, in which cell–cell contact is a preponderant factor, can also be studied using three-dimensional skin modes. Advances in the stem cell field offer the exciting possibility of generating large numbers of donor-matched backgrounds for autologous transplants, as well as a platform for in depth analyses of the physiological impact of various skin cell precursors in health and disease.

Acknowledgements

We are especially grateful to Dr. Monique Verhaegen (Dermatology Department, University of Michigan, USA), for the critical reading of the manuscript and her precious suggestions. This study was supported by FAPESP (2006/50479-7 and 2008/58817-4), CNPq, CAPES and PRP-USP. M.S. is supported by NIH R01 CA107237, CA125017; Spanish Ministry of Science and Innovation SAF 2008-1950 and institutional grants from the Spanish Association against Cancer.

Ancillary