**3.** *In vivo* **and** *in vitro* **pigmented skin models**

Human melanocytes have been cultured selectively for two decades when, in 1982, Eisinger and Marko published their work on selective proliferation of human melanocytes. Their selective culture was based on the proprieties of phorbol ester. Indeed, at a certain concentra‐ tion, this compound is toxic for keratinocytes, but not for melanocytes. By adding it to an epidermal cell solution, it allows the selective proliferation of melanocytes only [90]. In 1986, Topol *et al.* went further and reported the first pigmented human skin equivalent. This equivalent consisted of human neonatal melanocytes plating onto a dermal skin substitute with keratinocytes. They were added before those cells overgrew the dermal equivalent [91]. Since that year, many other models have been developed.

#### **3.1.** *In vivo* **models**

[76]. However, postinflammatory hyperpigmentation can be treated by a peeling surgery, a technique that uses chemical products for destruction of a part of the dermis and/or the epidermis [77]. Finally, diffuse hyperpigmentation in Addison's disease is usually treated with mineralocorticoid and glucocorticoid [78]. These two corticosteroids will compensate for the lack of corticosteroids, reduce the production of ACTH, and thus, reduce the production of melanin pigments. Considering that it is a rare disease, treatments are not very abundant. The development of pigmented skin models could be useful for studying unknown mechanisms involved in these disorders, and for developing more relevant treatments with few side effects.

It is known that there is a real need for human organs available for transplantations [79]. Each year, people die while they are waiting for a compatible organ. Just in Canada, in 2011, a person needing an organ had a 30 to 40% probability of not receiving it [80]. Moreover, there are some problems with allogeneic grafts, such as problems of incompatibility and reject preoccupa‐ tions. To get rid of these problems, a new approach emerged at the end of the 1980s [81]. This approach, called tissue engineering, is a science that combines both biology and engineering expertise. Its goal is to develop biological substitutes for maintaining, repairing or regenerating

Skin substitutes are useful in different fields. They can be grafted onto patients suffering from severe burns or chronic wounds such as skin ulcer [84-85]. They can also be used for funda‐ mental research to analyze skin functional mechanisms. Moreover, skin substitutes can be used for cosmetic testing to replace animal testing [86]. While skin is the interface tissue between human body and exterior environment, it is an organ particularly exposed to chemical and mechanical wounds and to pathological agents. Furthermore, it is an important organ in area and complexity both in structural and functional ways [87]. Consequently, to regenerate a

In recent years, many skin substitutes have been reported, becoming more and more similar to natural human skin, but different from each others, and still not perfectly simulating skin functionality. Those substitutes can be characterized by different factors. Provenance of cells used to produce the substitute can be either allogeneic, xenogeneic or autologous. Presence of autologous cells allows reduction of incompatibility and graft reject problems [88]. The purpose of substitutes can also vary between a permanent, semi-permanent or temporary use [83]. Depending on the use wanted, complexity of substitutes will change and can be either monolayered or bilayered. As there is no perfect substitute that has been developed to this

**2. Challenges for the development of pigmented skin models**

**2.1. Tissue engineering**

768 Regenerative Medicine and Tissue Engineering

**2.2. Skin substitutes**

human organs or tissue, such as skin [82-83].

human skin with all its functionality is a big challenge.

*2.2.1. Skin substitute characteristics*

#### *3.1.1. Spontaneous and induced mutations*

Spontaneous mutation models can be use to study some diseases when the mutations have similarity with human diseases. In melanocyte skin diseases, there are not many spontaneous mutation models available. One of them was developed in 1981 by Smyth which proposed a chicken model to study human vitiligo [92]. Indeed, he developed a mutant line of chickens characterized by a higher than normal spontaneous postnatal cutaneous amelanosis. This affection has similarity to human vitiligo. They are both a consequence of melanocyte de‐ struction. Those chickens are called "chickens of the autoimmune delayed-amelanotic" (DAM) or "Smyth chickens".

In melanomas studies, spontaneous mutations on mice are rare because, even if some chemical agents can induce them, it takes a long time, and the melanomas are not very representative of human ones, with the frequent absence of metastases that are frequent in human [93]. Most of the models found in the literature use other animals. An example is Millikan *et al.* which proposed in 1974 a Sinclair swine model to study pigment tumors [94]. Lesions developed by those swine are similar histologically and clinically to some different human tumors. Indeed, they show flat lesions that can be compared to human junctional neavus and elevated lesions similar to human compound neavus. Some other lesions found in swine are raided blue tumors, the counterpart of human blue neavus, peripheral depigmentation, the counterpart of vitiligo, and ulcerative tumors, the counterpart of melanoma. Previously, in 1966, Cherno‐ zemski has proposed a Syrian hamster model which presented spontaneous or induced by DMBA melanomas [95]. DMBA was also used to induce melanomas on Guinea pigs by Clark *et al.* ten years later [96]. Clark *et al.* have reproduced Edgcomb's and Mitchelich's work of 1963 and had shown that tumors in guinea pigs have some similarities with those in humans, but are not histogenetically the same. In 1989, Setlow *et al.* reported a platyfish-swordtail hybrid model susceptible to melanoma when exposed to UV radiations [97]. This fish had already been used by other authors in the past such as Anders *et al.* who used it in 1984 for melanoma research [98]. Setlow, in his work, studied different strains of this fish and their response to some UV wavelengths to find two that were susceptible to melanomas under UV irradiations. Fish of those two strains developed melanomas that were quite similar to the human ones. The principal difference was the presence of melanophores in fish melanomas. In vertebrates, melanophores represented the last stage of pigmented cells differentiation. The same year, Ley *et al.* also reported an animal model of melanomas induced by UV radiations using South American opossum [99]. In their study, they also used the concept of photoreactivation repair pathway for DNA damages to investigate pyrimidine dimer's implication in melanoma induction. Their research allowed them to make two principal conclusions: first, they con‐ cluded that UV radiations can be used to induce malignant melanoma; second, they came to the conclusion that pyrimidine dimer is involved in melanoma formation, and that the radiations induced DNA damages. Reported spontaneous models of melanoma have been less and less frequent in the recent years probably as a result of the improvement of science in different fields such as gene modifications and *in vitro* models.

#### *3.1.2. Transgenic models*

Transgenic animals can be useful for mimicking some human diseases such as melano‐ ma. In 1991, Bradl *et al.* reported a transgenic mouse with the simian virus 40 (SV-40) controlled by a tyrosinase promoter [93] that promotes ocular and cutaneous melanomas. The melanomas reported were histopathologically the same as their human counterparts. One year later, Iwamoto *et al.* also reported a transgenic mouse model for studying melanocyte tumors. Their model consisted of a mouse metallothionein promoter enhanc‐ er coupled to a *ret* oncogene inductor [100]. They developed four independent mouse lines:

three of those lines were well-predisposed to developing melanocytic tumors, and the other one reported an acceleration of melanogenesis with no clear proliferative disorders. In 1994, Klein-Szanto *et al.* also used the SV40 driven by tyrosinase promoter to develop a transgenic mouse model. Their model allowed them to study the induction of malignant skin melanomas by short ultraviolet radiation exposure and without chemical carcinogen application [101]. Inbred line choice and other factors such as the age of the mice and the intensity of the UV treatment can be modified in the protocol to improve melanoma induction. In 1997, Takayama *et al.* proposed a new transgenic mouse model, also using metallathionein promoter driving, this time an hepatocyte growth factor/scatter factor (HGF/SF) [102]. They wanted to study the oncogenic role of those factors. Their transgen‐ ic mice developed a large variety of tumors, including melanoma. Some tumors, includ‐ ing this one, overexpressed HGF/SF. The same year, Chin *et al.* reported another transgenic mouse model using a different gene. Indeed, their model consisted of H-ras driven by tyrosinase promoter with INK4a knockout mice [103]. Their studies allowed them to conclude that development of melanoma can be accelerated by both the loss of INK4a allele and the activation of Ras. In 2009, Goel *et al.* reported a BRAFV600E transgenic mouse [104]. Indeed, in more than half melanoma cases, there is presence of a mutation that affects BRAF, a protein activated by Ras. Transgenic mice presented benign melanocytic hyperpla‐ sia of which progression to the melanoma stage depended on BRAF expression. In 2011, Meyer *et al.* used *ret* transgenic mice to study melanoma evolution. They also studied inflammatory tumor microenvironment when there is enrichment of myeloid-derived suppressor cells (MDSCs). They concluded from their studies that, before starting an immunological treatment for melanoma, the immune status should be controlled and the immunosuppressive microenvironment should be neutralized [105]. However, transgenic models are not the best for mimicking some cancers such as melanoma because, such as spontaneous mutations, they present a lack of metastase production. Indeed, it is known that the importance or not of metastases in large amount is crucial for the patient's survival. By consequence, it is often a target in treatment development and so their presence is important to have a valuable model.

#### *3.1.3. Xenotransplantation models*

struction. Those chickens are called "chickens of the autoimmune delayed-amelanotic" (DAM)

In melanomas studies, spontaneous mutations on mice are rare because, even if some chemical agents can induce them, it takes a long time, and the melanomas are not very representative of human ones, with the frequent absence of metastases that are frequent in human [93]. Most of the models found in the literature use other animals. An example is Millikan *et al.* which proposed in 1974 a Sinclair swine model to study pigment tumors [94]. Lesions developed by those swine are similar histologically and clinically to some different human tumors. Indeed, they show flat lesions that can be compared to human junctional neavus and elevated lesions similar to human compound neavus. Some other lesions found in swine are raided blue tumors, the counterpart of human blue neavus, peripheral depigmentation, the counterpart of vitiligo, and ulcerative tumors, the counterpart of melanoma. Previously, in 1966, Cherno‐ zemski has proposed a Syrian hamster model which presented spontaneous or induced by DMBA melanomas [95]. DMBA was also used to induce melanomas on Guinea pigs by Clark *et al.* ten years later [96]. Clark *et al.* have reproduced Edgcomb's and Mitchelich's work of 1963 and had shown that tumors in guinea pigs have some similarities with those in humans, but are not histogenetically the same. In 1989, Setlow *et al.* reported a platyfish-swordtail hybrid model susceptible to melanoma when exposed to UV radiations [97]. This fish had already been used by other authors in the past such as Anders *et al.* who used it in 1984 for melanoma research [98]. Setlow, in his work, studied different strains of this fish and their response to some UV wavelengths to find two that were susceptible to melanomas under UV irradiations. Fish of those two strains developed melanomas that were quite similar to the human ones. The principal difference was the presence of melanophores in fish melanomas. In vertebrates, melanophores represented the last stage of pigmented cells differentiation. The same year, Ley *et al.* also reported an animal model of melanomas induced by UV radiations using South American opossum [99]. In their study, they also used the concept of photoreactivation repair pathway for DNA damages to investigate pyrimidine dimer's implication in melanoma induction. Their research allowed them to make two principal conclusions: first, they con‐ cluded that UV radiations can be used to induce malignant melanoma; second, they came to the conclusion that pyrimidine dimer is involved in melanoma formation, and that the radiations induced DNA damages. Reported spontaneous models of melanoma have been less and less frequent in the recent years probably as a result of the improvement of science in

different fields such as gene modifications and *in vitro* models.

Transgenic animals can be useful for mimicking some human diseases such as melano‐ ma. In 1991, Bradl *et al.* reported a transgenic mouse with the simian virus 40 (SV-40) controlled by a tyrosinase promoter [93] that promotes ocular and cutaneous melanomas. The melanomas reported were histopathologically the same as their human counterparts. One year later, Iwamoto *et al.* also reported a transgenic mouse model for studying melanocyte tumors. Their model consisted of a mouse metallothionein promoter enhanc‐ er coupled to a *ret* oncogene inductor [100]. They developed four independent mouse lines:

*3.1.2. Transgenic models*

or "Smyth chickens".

770 Regenerative Medicine and Tissue Engineering

Another type of model that can be used to study human melanoma consists of xenotrans‐ plantation of human skin onto an animal. For models on mice, three different types of this animal are mainly used: athymic nude mice, severe combined immunodeficient mice (SCID) and spontaneous AGR129 mice models. The immune systems of those mice differ from those of normal mice so that their immunological potential is decreased. Athymic nude mice do not have T cells because of the absence of a thymus. SCID mice do not have either T cells, and are also deficient of B cells. AGR129 mice do not have either T or B cells as SCID mice, and also have immature natural killer cells [106]. In 1993, Juhasz *et al*. proposed a xenotransplantation model of human skin grafted onto either nude or SCID mice. Beforehand, human melanomas were injected onto the human skin graft [107]. This model can allow reproduction of human melanomas and their metastases because tumour cells invaded human vessels, and in more than half case, metastases were found in lungs. In 1998 Atillasoy *et al.* proposed another model for studying UV irradiations and chemi‐ cal carcinogen implication in skin melanoma development. Their model consisted of human newborn foreskin grafted onto RAG-1 mice. Mice were separated into four groups and received different treatments. While the first group was the control group and received no treatment, the second received a treatment of DMBA, a chemical carcinogen. The third group received UVB irradiations and the last one, both DMBA and UVB irradiations [108]. Only DMBA treatment was not conclusive, as the only impact was the development of melanocytic hyperplasia in 16%. UVB only treatment was a little better as it caused solar lentigo in 23% and melanocytic hyperplasia in 68%. Combined treatment caused solar lentigo in 38% and melanocytic hyperplasia in 77% and was the only one that generated melanoma in 2.1% after 15 months. This is representative of melanoma incidence in Caucasian Americans that is 1.4%. In the last years, xenotransplantation models were also used to test different treatments. For example, in 2010, Schicher *et al.* used a xenotransplan‐ tation model of a SCID mouse with a human melanoma graft to test treatment of Erloti‐ nib combined with some chemotherapeutics agents [109]. Erlotinib is a treatment already used for non small cell lung cancer. The melanomas treated with a combined treatment showed a higher reduction than those only treated with chemotherapeutic agents.

#### *3.1.4. Other models*

Other testing systems that can be used are injection of mouse melanoma into mice. For that, some mouse melanoma cell lines have been produced. One of these lines originated from one of the rare spontaneous melanomas in mice, an event that occurred on the ear of a C57BL/6 mouse in 1954. It was Fidler *et al.*, who at the beginning of the 1970s really set up the line known as B16 melanoma cell line [110]. Those murine melanoma cells can be used for studying melanoma treatments as reported by some authors. For example, in 2002, Lucas *et al.* injected B16 melanoma in C57BL/6 mice in order to test injection by electroporation of interleukin-12 [111]. They tested an intratumoral and an intramuscular treatment. While the first one was useful for treating tumour in 47% of the cases, the second one was not conclusive. They also tried those two treatments in a nude mouse model, but neither the intratumoral or the intramuscular treatment worked. Those results allowed them to conclude that T cells probably have a role in the melanoma regression, at least in this model. The same year, Garcia-Hernandez *et al.* investigated on the implication of interleukin-10 in the promotion of B-16 melanoma growth [112]. This study allowed them to conclude that IL-10 seems to have a role in this promotion in three fields: first, they simulate the proliferation of the tumour-cells; second, they have implications for the angiogenesis process and third, they are implicated in the immunosuppression. B16 melanoma cells can therefore be useful to many types of studies.

#### **3.2.** *In vitro* **models**

#### *3.2.1. Monolayers*

model can allow reproduction of human melanomas and their metastases because tumour cells invaded human vessels, and in more than half case, metastases were found in lungs. In 1998 Atillasoy *et al.* proposed another model for studying UV irradiations and chemi‐ cal carcinogen implication in skin melanoma development. Their model consisted of human newborn foreskin grafted onto RAG-1 mice. Mice were separated into four groups and received different treatments. While the first group was the control group and received no treatment, the second received a treatment of DMBA, a chemical carcinogen. The third group received UVB irradiations and the last one, both DMBA and UVB irradiations [108]. Only DMBA treatment was not conclusive, as the only impact was the development of melanocytic hyperplasia in 16%. UVB only treatment was a little better as it caused solar lentigo in 23% and melanocytic hyperplasia in 68%. Combined treatment caused solar lentigo in 38% and melanocytic hyperplasia in 77% and was the only one that generated melanoma in 2.1% after 15 months. This is representative of melanoma incidence in Caucasian Americans that is 1.4%. In the last years, xenotransplantation models were also used to test different treatments. For example, in 2010, Schicher *et al.* used a xenotransplan‐ tation model of a SCID mouse with a human melanoma graft to test treatment of Erloti‐ nib combined with some chemotherapeutics agents [109]. Erlotinib is a treatment already used for non small cell lung cancer. The melanomas treated with a combined treatment

showed a higher reduction than those only treated with chemotherapeutic agents.

melanoma cells can therefore be useful to many types of studies.

Other testing systems that can be used are injection of mouse melanoma into mice. For that, some mouse melanoma cell lines have been produced. One of these lines originated from one of the rare spontaneous melanomas in mice, an event that occurred on the ear of a C57BL/6 mouse in 1954. It was Fidler *et al.*, who at the beginning of the 1970s really set up the line known as B16 melanoma cell line [110]. Those murine melanoma cells can be used for studying melanoma treatments as reported by some authors. For example, in 2002, Lucas *et al.* injected B16 melanoma in C57BL/6 mice in order to test injection by electroporation of interleukin-12 [111]. They tested an intratumoral and an intramuscular treatment. While the first one was useful for treating tumour in 47% of the cases, the second one was not conclusive. They also tried those two treatments in a nude mouse model, but neither the intratumoral or the intramuscular treatment worked. Those results allowed them to conclude that T cells probably have a role in the melanoma regression, at least in this model. The same year, Garcia-Hernandez *et al.* investigated on the implication of interleukin-10 in the promotion of B-16 melanoma growth [112]. This study allowed them to conclude that IL-10 seems to have a role in this promotion in three fields: first, they simulate the proliferation of the tumour-cells; second, they have implications for the angiogenesis process and third, they are implicated in the immunosuppression. B16

*3.1.4. Other models*

772 Regenerative Medicine and Tissue Engineering

Monolayer models are characterized by the culture of one cell type, which has previously been extracted from a skin biopsy. Cells can be extracted from normal or lesional skin such as cells of skin affected by hypo- and hyperpigmentation disorders. Monolayer models are useful for studying melanocyte properties and testing different conditions or drugs. In 2008, in a comparative study of melanocytes culture and melanocytes-keratinocytes co-culture, Liu *et al*. demonstrated the effect of melanogenic stimulators (α-MSH and L-tyrosine) and inhibitors (arbutin and hydroxybenzyl alcohols (HBA)) in the two conditions [113]. Results showed that α-MSH and L-tyrosine increased the melanin content of melanocytes, and that the increase was better in the co-culture with keratinocytes. 4HBA and arbutin inhibit the melanogenesis in the two conditions, but, in co-culture, the inhibition was much better than in melanocytes alone. These results suggest that cytokines released by keratinocytes can have an effect on the regulation of melanin synthesis, and that the co-culture model has interesting properties for testing drugs related to the treatment of pigmentation disorders. In the same vein as Liu *et al.*,Criton *et al*. tested 22 N-hydroxy-N-phenylthiourea and N-hydroxy-N-phenylurea ana‐ logues, which could inhibit tyrosinase activity and reduce melanin synthesis on melanocyte culture [114]. Results showed that compound 1 inhibits tyrosinase and reduces 78 % of melanin synthesis. It is a promising candidate for the treatment of hyperpigmentation disorders to replace whitening agents that have undesirable effects. These studies demonstrate that monolayer models allow testing of several conditions, and the possibility of observing melanocytes behavior in some pigmentation disorders.

#### *3.2.2. Collagen gels*

Unlike monolayer models, which are composed of only one cell type, collagen gel models allow formation of a dermis on which more than one cell types can be seeded. Globally, for the construction of a pigmented equivalent with collagen gel, fibroblasts, keratinocytes and melanocytes are extracted from a skin biopsy and are cultured separately. Fibroblasts are seeded onto the collagen gel, then, after few weeks of culture, keratinocytes and melanocytes can be seeded onto fibroblasts and collagen matrixes [115]. These types of equivalent that contain various cell types are useful for studying the interactions between keratinocytes and melanocytes and understanding different mechanisms of pigmentation disorders. Recently, Duval *et al*. developed a pigmented skin model using collagen, such as a dermal matrix that is very representative of normal human skin with melanocytes [116]. They demonstrate that their model has a functional pigmentary system by the presence of melanocytes well-devel‐ oped with melanosomes, the expression of tyrosinase, TRP1 and TRP2, the transfer of mela‐ nosomes containing melanin to keratinocytes and the stimulation of melanin synthesis by α-MSH. This model has most of the normal human skin melanocyte characteristics [9] and seems to be an interesting pigmented skin model for studying cell interactions of the pigmentary system. A study with this type of skin model has observed the response of melanocytes after UV radiation [117]. Archambault *et al*. demonstrated by a comparative study of monolayer melanocyte culture and pigmented skin equivalent that melanocytes have a better capacity for surviving UVR than melanocytes in culture. These results suggest that keratinocytes and fibroblasts secrete factors that enhance melanocytes survival and migration, which could explain UVR-induced pigmentation by melanocytes. Unlike other teams, Freeman *et al*. elaborated another technique of culture with collagen gels [118]. They put a complete skin biopsy, which was affected by a melanoma, onto a collagen gel that contained fibroblasts. This technique allowed conservation of the *in vivo* properties of the melanoma and observation of the proliferation and other characteristics *in vitro*. All of these models composed with a collagen matrix are representative of human skin, and are effective for the study of pigmentation disorders. However, collagen models may not be useful for testing drugs, because they can be absorbed by the collagen and a higher quantity of drugs must be used [119].

#### *3.2.3. De-Epidermized Dermis (DED)*

Such as collagen gels, DED allowed the construction of pigmented skin equivalent that reproduced human skin for studying the pigmentary system. Unlike collagen gels, this method has a native extracellular matrix and a basal membrane that facilitates melanocyte adhesion. Principally, DED preparation is very similar. From a skin biopsy, the epidermis is removed and the dermis is incubated in saline solution or undergoes freezing-thawing cycles to kill cells [120]. After this treatment, the dermis remaining is called a dead de-epidermized dermis, and is ready to be seeded by keratinocytes and melanocytes. In 1993, Todd *et al*. used this model to demonstrate the effect of UV radiation on a pigmented equivalent [121]. They demonstrated that, after UV radiation, there is an increase of TRP1 activity, an increase of pigmentation and an increase of DOPA-positive melanocytes such as observed *in vivo*. In 2000, for a better understanding of melanoma invasion, Dekker *et al*. developed and characterized a DED skin equivalent with four types of melanoma cells [122]. They observed the expression of different integrins that play an important role in the behavior of melanoma cells. Their model is a useful tool for studying melanoma and other mechanisms involved in this cancer. In 2007, Cario-André *et al*. developed a DED skin model with normal and non-lesional vitiligo cells to understand if the loss of melanocytes in vitiligo is caused by a detachment of melanocytes due to stress factors [123]. The model with non-lesional vitiligo cells contains less melanocytes than the model with normal cells, and hydrogen peroxide and epinephrine could be the cause of the detachment of melanocytes. Todd, Dekker and Cario-André models demonstrated that deepidermized dermis is an interesting model for studying different pigmentation disorders and, in comparison with collagen gels, DED produced an epidermis, which is closer to the native epidermis than collagen gels [124].

#### *3.2.4. Commercial models*

Commercial *in vitro* skin models have been developed to reduce tests on animals, replace animal models and refine methodologies. For the cosmetic industry, it allows testing of the toxicity of their products and their pharmaceutical effects on a complete human epidermis and dermis. Several companies produce skin models and some of them produce equally pigmented skin models that are useful for testing photoprotection, whitening agents and repigmentation products. The list of the main commercial pigmented skin models and their principal features is presented in Table 3.


**Table 3.** Different commercial pigmented skin models
