Models of melanoma have been difficult to develop due to the differences in anatomy and function of human versus mouse skin. Melanocytes in human skin are aligned on the basement membrane and are dispersed among the epidermal keratinocytes, whereas in mouse skin, melanocytes are absent from the epidermis and situated deep in hair follicles and dermis. When newborn mice are exposed to UVB, they can transform more readily because the melanocytes have not yet descended from the epidermis into the hair follicles. Because of the major architectural and functional differences between mouse and human skin, our laboratory had started ~15 years ago to graft human skin to immunodeficient mice to better mimic the micro-environmental conditions in human skin. In subsequent studies, we exposed the human skin in a two-step carcinogenesis model to an initiator (DMBA) followed by an initiator/promoter (UVB) according to established protocols in mouse skin carcinogenesis. We could readily develop in the human skin squamous but only rarely melanocytic lesions and then only after 1 year of UVB exposure. When we switched to a growth factor as 'initiator', bFGF, we saw more frequent and more profound lentiginous lesions, i.e., the atypical melanocytes were lined along the basement membrane zone and rarely invaded the dermis. The encouraging results prompted us to use three growth factors for over-expression in the skin at the same time, bFGF, SCF and ET-3. All are known mitogens for melanocytes. This combination led to a major breakthrough in melanoma model development because we now induced melanocytic lesions in over 90% of the treated skins within 3 weeks that were histologically diagnosable as melanomas after 4 weeks in 17 of the 50 specimens (34%) examined. A limitation of the model is that once the growth factors were omitted, the lesions regressed suggesting that additional events are necessary for malignant transformation. This now seek to extend this model in order to induce malignant transformation so that melanocytes fulfill all six 'hallmarks of cancer'. This is possible because an additional technical advance has been made enabling us to graft artificial skin created in vitro (known as skin reconstructs, organotypic skin cultures, or skin equivalents) to mice. In this model the melanocytes align along the basement membrane as in natural human skin (this grafting technique is well known among wound healing experts. The organotypic cultures allow us to manipulate the expression/activity of specific genes prior to generation of the skin reconstructs. Specifically, we can increase expression of genes, down-regulate their expression, or introduce genes with specific mutations. Thus, we can theoretically 'rebuild' the epigenetic and genetic events leading to melanoma development and progression and thereby begin dissect and understanding the mechanisms of the disease process. The striking similarities between lesions developing in the experimental model and those occurring in patients make us confident that the model is suitable for studying early stages of melanomagenesis, which are difficult if not impossible to study using patients' material alone.
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
4.2. Human skin reconstructs and their
grafting to mice. Human skin can be rebuilt
in vitro by first embedding fibroblasts
in collagen and then layering on top melanocytes
and keratinocytes (34, in press). When the
keratinocytes are exposed to air, they differentiate
to form multiple layers. Skin reconstructs
are being used in medicine for covering
wounds of patients. Skin reconstructs show
clearly the effects of dysbalance in homeostasis
if growth factors are overexpressed prior
to inclusion of cells in the reconstructs
(Fig. 5). bFGF expressed in melanocytes
leads to poor adhesion of the melanocytes
to the basement membrane and stimulation
of the fibroblasts. The dysregulation of
the homeostatic balance is more pronounced
if SCF is overexpressed in keratinocytes
whereas melanocytes were transduced with
the bFGF genes.
While skin reconstructs in vitro have a
maximum life-span of approximately 1 month,
survival is increased to several months
or years when grafted to living hosts. Figure
6 shows a pigmented human skin reconstruct
on an immunodeficient SCID mouse. Histological
and immunohistochemical analyses revealed
that i) host vessels and immune cells infiltrated
the dermis of the grafts, ii) a papillary
morphology of the upper dermis developed,
iii) the differentiation of the epidermis
improved, and iv) the basement membrane
matured (reviewed in ref. 8). Likely, not
yet defined diffusible factors from the
host's microenvironment support the morphology
and viability of the grafted human skin
reconstructs. Fibroblasts that are transduced
with bFGF and embedded in collagen produce
increased levels not only of bFGF but also
of other growth factors that are mitogens
for melanocytes, for example ET-3 (see Fig.
2). If ET-3 is overexpressed in fibroblasts,
EGF receptor and neuropilin are upregulated
as determined by microarray experiments,
whereas chloride channel proteins, matrix
metalloproteinases, and matrix proteins
are downregulated suggesting that the induction
of ET-3 leads to a complex cascade of gene
activation and inactivation with potential
consequences for melancoyte adhesion to
the basement membrane. The rationale for
switching the transformation model from
intact skin to skin reconstructs is provided
in the Design where we also discuss potential
pitfalls and alternatives.
