Structure of the intact basement membrane in skin. Top: Electron micrograph showing the two major layers comprising the basement membrane, lamina lucida (LL) and lamina densa (LD), that separate the epidermis from the dermis (d). Bar: 0.5 µm. (Source: Uitto et al. 1996). Bottom: Schematic diagram defining various tissues in the zone of the basement membrane
5.2.2 In Vitro Synthesis of an Epidermis
It makes intuitive sense to attempt covering a skin defect with an epidermis that has been synthesized in vitro . This approach has been used extensively for the treatment of shallow wounds. A great deal of research activity has been expended toward preparation of such an epidermal graft. Grafts based on cultured KCs have been variously referred to, in the literature, as KC sheets, cultured epithelia (CE), cultured epithelial autografts (CEA), or cultured autologous keratinocytes (CAK). The generic acronym, KC sheets, will be used consistently below. Major questions addressed in this section are: How are KC sheets synthesized? What happens when KC sheets are grafted on a dermis-free defect?
Approximately 90 % of the epidermal cells in mammals are KCs. The three sources of KC for eventual cultivation have been employed as follows: (1) KC sheets formed from epidermal explants, (2) suspensions or pellets of disaggregated KC, and (3) KC sheets formed from dissociated cells. A comprehensive review of the evolution of methodology for preparation of KC sheets has been compiled (Compton 1994).
During earlier studies, it had been established that suspensions of epithelial cells could be isolated from skin by trypsinization (Billingham and Reynolds 1952; Billingham and Russell 1956). Epithelial cells isolated from the skin of chick and mouse embryos were shown to be capable of spontaneous reaggregation and reconstruction of epidermal structures when placed directly on the chorioallantoic membrane of the chick. Reaggregation also took place in flasks or in roller tube cultures. When transplanted into a skin graft site on an autologous rabbit host, cell cultures multiplied, differentiated, and formed an epidermis .
Some independent methods for culturing KC have been historically developed in several laboratories with a variety of animal models (mice, Yuspa et al. 1970; rats, Bell et al. 1981a, b; dogs, Eisinger et al. 1980; guinea pigs, Hefton et al. 1983; rabbits, Lui and Karasek 1978; pigs, Eisinger et al. 1984) as well as in humans (Rheinwald and Green 1975a, b; Freeman et al. 1976; Eisinger et al. 1979; Peehl and Ham 1980; Boyce and Ham 1983, 1985). Some of the experimental problems faced by workers who cultivated KCs were contamination of potential epidermal grafts with dermal cells (Billingham and Reynolds 1952; Cruickshank et al. 1960; Delecluse et al. 1974; Prunieras et al. 1976, 1979), cell viability following grafting (Billingham and Medawar 1950, 1951; Karasek 1968; Prunieras 1975), and relatively low cell expansion factors (Karasek 1968; Igel et al. 1974; Regnier and Prunieras 1974) . Expansion of epithelial cells on a dermal substrate or on a plastic film by a factor of 50 over a 3-week period provided complete cover of full-thickness skin defects in rabbits (Igel et al. 1974). In an early clinical application of this concept, autologous epithelial cells were cultured in vitro on an irradiated porcine dermis substrate; the resulting epidermis was detached from the substrate and placed on burn wounds in direct contact with the wound bed. Although the technique was reported to be successful, it could not generate large amounts of epithelium; nor could the epithelial sheet be separated from the culture dish without being damaged.
Many of these problems were overcome with the development of in vitro methods for rapid, serial subcultivation of KC from disaggregated cell suspensions, resulting in expansion factors of over 10,000 within 3–4 weeks (Rheinwald and Green 1975a, b). This development has had considerable impact and has been refined by several groups since it was first introduced (De Corte et al. 2012; Dragunova et al. 2012; Zare al. 2013; Rasmussen et al. 2013) .
KCs are usually isolated from skin biopsies by enzymatic treatment that removes the dermal components and dissociates the coherent epidermal tissue. They have been cultured in media that had typically comprised a combination of fetal bovine serum and defined media, including hormones and growth factors (Eisinger et al. 1979; Boyce and Hansbrough et al. 1988; Cooper et al. 1993). Other media have also been employed, including medium incorporating lethally irradiated fibroblasts (Rheinwald and Green 1975a, b; Barrandon et al. 1988; Carver et al. 1993b), medium harvested from fibroblast cultures (Green and Rheinwald 1977), or medium completely free of dermal components (Eisinger et al. 1979). The availability of oxygen has also been recognized as an important factor in the synthesis of an epidermis (Prunieras 1975) .
The original requirement for use of a feeder layer or for calf serum (Rheinwald and Green 1975a, b) has been shown to be unnecessary following development of serum-free and feeder-layer-free conditions for keratinocyte culture (Rosdy and Clauss 1990; Coolen et al. 2007; Radtke et al. 2009). The use of chemically defined medium MCDB 153 and of inert substrates has been particularly instrumental in early studies where it was shown that a fully differentiated epidermis can be synthesized in vitro under relatively simple conditions (Rosdy and Clauss 1990; Rosdy et al. 1993) .
Synthesis of a coherent, intact sheet of stratified epithelium, typically four to six cell layers thick, kept together by desmosomal attachments has been readily achieved under relatively simple conditions. The level of maturity of the epidermis synthesized in vitro has typically been moderately high; however, keratinization has not always been achieved. Epidermal maturity obtained in vitro has significantly increased when the keratinocyte sheet was grafted on a dermis-free defect. For example, in a well-known study (Carver et al. 1993b), the epidermis cultured in vitro consisted of five to six layers of flattened, undifferentiated cells joined by desmosomes, with sparse keratin filaments running parallel to the long axis of the flattened cells. The plasma membrane of the basal cell layer was in some regions closely apposed to the synthetic polymeric substrate supporting the structure mechanically and formed attachment structures, including hemidesmosomes with sub-basal plates underneath. Use of the enzyme dispase to detach the KC sheets from the culture flasks resulted in the disappearance of these attachment structures . However, following grafting on a dermis-free defect, the epidermis continued maturing, as evidenced by the time-dependent increase in number of hemidesmosomes, average length of a desmosome and of a sub-basal plate (Carver et al. 1993b). The grafted epidermis induced synthesis of a BM as described in the next section. Similar observations have been made by others (Aihara 1989; Cooper et al. 1993) .
Although the maturity of the neoepidermis has been discussed so far with emphasis on keratinization, at least one study has shown that maturation of epidermal attachment structures also depends critically on the identity of the substrate. When stratified epithelium was grown on collagen gels, hemidesmosomes were not synthesized; however, use of a surface consisting of a reconstituted BM led to synthesis of hemidesmosomes (Lillie et al. 1988) .
We conclude that a partly mature epidermis can be synthesized in vitro by condensation of disaggregated KC. There is no requirement for the presence of fibroblasts or of any dermal component in order to form a partly differentiated, relatively immature epidermis; however, there appears to be a temporary requirement for a nondiffusible (solid-like) substrate on which cells are plated, eventually becoming stratified and keratinized. Contact with surfaces of certain connective tissues or their analogs induces maturation of a stratified epidermis very effectively in vitro; however, significant, though not complete, epidermal maturation occurs even on glass or plastic surfaces .
5.2.3 Structure of Basement Membranes
Below, we briefly review the structure of the BM region in skin in order to follow the details of the relevant synthetic processes . The terms “basal lamina” and “BM” have been used interchangeably in the literature, leading to considerable confusion, until it was realized that all three layers seen with the electron microscope represent the single layer (lamina densa or basal lamina) seen with the light microscope (Fig. 5.1, Burkitt et al. 1993). It has been recommended that the term basal lamina should be confined to its meaning as just one of the layers, i.e., lamina densa, as originally employed (Martinez-Hernandez 1988; Burkitt et al. 1993).
As mentioned in an earlier chapter, the BM of an organ is an avascular, cell-free tissue layer interspersed between a layer of avascular epithelia (tissues that cover or line all body surfaces, cavities, and tubes) and a layer of stroma (vascularized connective tissue, or “supporting” tissue). Several roles of BMs have been identified in different organs. These include functions such as that of a boundary that restricts transfer of cells and molecules (Farquhar 1981), an anchorage matrix for epithelial cells and a mechanically competent adhesive-like layer that binds the epithelia to the stroma (Furthmayr 1988; Uitto et al. 1996), a scaffold that facilitates tissue repair after injury (Vracko 1974; Woodley and Briggaman 1988), as well as several specialized roles during differentiation and growth (Hay 1981). Although subtle variations in both composition and assembly of components have been observed in BMs of different organs and species (Kefalides and Alper 1988; Breitkreutz et al. 2009), there are strong similarities that appear to overshadow the differences (Burkitt et al. 1993). There are significant thickness differences depending on anatomical site or species (Furthmayr 1988). Also, the mature glomerular BM in the kidney, as well as segments of the alveolar BM in the lung are three layered (trilaminar) (Martinez-Hernandez 1988). Furthermore, there is evidence that BM defects that are lethal during development vary between tissues and organs (Breitkreutz et al. 2013). However, there are strong similarities in composition between BMs of species as different as Drosophila and mouse (Fessler et al. 1984) . The BMs in skin and peripheral nerves, the two organs that are treated in detail in this volume, are very similar in composition and structure .
The first layer of the BM, 20–40 nm in thickness, is next to the cell membrane of the innermost (basal cell) layer of the epithelia, and is the electron-lucent lamina lucida that consists primarily of the glycoprotein laminin (Fig. 5.1, bottom).The intermediate layer is electron-dense, about 40–50 nm in thickness (lamina densa); it consists primarily of type IV collagen. Adjacent to it is an electron-lucent reticular (fibroreticularis) layer that merges with the fibers of the underlying stroma; in skin, this layer comprises fibers of type VII collagen (anchoring fibrils) that are connected to the dermis by specific structures (anchoring plaques) (Briggaman and Wheeler 1975; Carver et al. 1993b). Hemidesmosomes are discrete plaques inside the layer of epithelial cells closest to the BM (basal cells); they serve to anchor the basal cells to the BM by means of keratin filaments (tonofilaments) and by connections to junctions in the BM (sub-basal plates) (Burkitt et al. 1993). Although the BM is frequently described as consisting of three zones (lamina lucida, lamina densa, and reticular layer), authors have occasionally described the hemidesmososmes with their tonofilaments, a zone about 20–40 nm in thickness, as a fourth zone (Woodley and Briggaman 1988; Carver et al. 1993b). In addition to types IV and VII collagen, BMs contain fibronectin, heparan sulfate proteoglycan, chondroitin sulfate proteoglycan, nidogen/entactin, α1-microglobulin, thrombospondin, and tenascin (Rigal et al. 1991; Breitkreutz et al. 2009, 2013).The total thickness of the BM is only about 100 nm, typically one tenth the thickness of the epidermis (Briggaman and Wheeler 1975). In skin, the BM surface is topologically similar to the surface of a filled egg carton; viewed in planar cross section, the BM appears as an undulating line (rete ridges) (Fig. 5.1, top) .
5.2.4 Synthesis of a Skin BM
An early in vitro synthesis of the BM of skin has been reported (Briggaman et al. 1971) . In this study, a partial-thickness skin graft (epidermis attached to a partial-thickness dermis) was treated with trypsin at low temperature (4 °C), leading to the separation of epidermis from dermis. The separation appeared to have occurred sharply and uniformly at the lamina lucida, as shown by ultrastructural observations of the isolated tissues. The outer surface (basal cell membrane) of the isolated epidermis was lined with hemidesmosomes containing tonofilaments; no lamina densa or anchoring fibrils were seen. The isolated dermis showed an intact lamina densa and anchoring fibrils. Recombination of the isolated epidermis and dermis, followed by incubation, was then pursued in an effort to find out whether a BM would form at the new interface of the two tissues. In order to eliminate the possibility of contamination of the new interface by residual BM from the previous interface, the freshly cut dermal layer was turned around from its normal position (inverted dermis) so that the trypsinized surface would not be in contact with the isolated epidermis. The isolated epidermis was then applied on the surface of the inverted dermis and the recombined bilayer was placed on the chorioallantoic membrane of the chick embryo, yielding the BM (Briggaman et al. 1971) .
To find out whether cells from the dermis participated in the synthesis of BM, the investigators combined the epidermal layer with the dermal layer both in a viable and in a nonviable state. The latter was prepared by repeated freezing and thawing of the dermal layer, followed by a demonstration of lack of dermal cell viability (Briggaman et al. 1971). In the presence of either viable or nonviable dermis, it was observed that lamina densa was synthesized in 3 days, not in continuous fashion but only focally next to the intact hemidesmosomes of the epidermal basal cells. Between 5 and 7 days after combination of the two tissue layers, lamina densa became progressively more dense and continuous at the epidermal–dermal interface. The results showed that dermal viability was not required for synthesis of lamina densa, supporting the epidermal origin of this layer in the BM. In contrast, anchoring fibrils were synthesized in the presence of viable but not nonviable dermis, suggesting the dermis as the hypothetical origin of these structures (Briggaman et al. 1971; Woodley and Briggaman 1988). (However, evidence presented below has suggested that anchoring fibrils originate in the epidermis; Carver et al. 1993b) .
In two other in vitro studies, elements of the BM were synthesized by culturing epidermal cell suspensions on a collagen gel, i.e., in the absence of a dermal layer) (Mann and Constable 1977; Taniguchi and Hirone 1983). Hemidesmosomes were synthesized along the plasma membrane of epidermal cells at the interface with collagen gel and synthesis of a continuous lamina densa was eventually observed; however, anchoring fibrils were not reported (Taniguchi and Hirone 1983).
In vivo synthesis of a complete BM was based on use of meticulously prepared defect surfaces in skin, free of traces of BM and underlying dermis (Woodley et al. 1988a; Aihara 1989; Carver et al. 1993b; Cooper et al. 1993). As the BM layers are very thin, methods for detection of the main macromolecular components of BM, namely, laminin, type IV collagen, and type VII collagen, have been largely based on use of electron microscopy. These methods are destructive and the number of observations has been typically limited to a total of three to four per study with sequential observations sometimes separated by gaps as large as 1–2 weeks. In spite of these limitations, a useful qualitative record of BM synthesis in vivo is available .
An early demonstration that grafted KC sheets can induce synthesis of a complete BM was made in the context of a clinical study (Aihara 1989). Four patients with burn wounds were grafted with cultured KC sheets immediately after excision of the burn surface to the level of the fat or fascia (i.e., on a dermis-free defect). Just before grafting, the cultured KC sheet lacked hemidesmosomes and BM-like structures. At day 9 following grafting, there were no hemidesmosomes or structures resembling the lamina densa; however, by day 42, occasional hemidesmosomes with an associated lamina densa were observed, providing evidence of a discontinuously synthesized BM. The epidermis had become highly differentiated and cornified by day 42. By day 150, formation of micropapillae, structures which are normally associated with formation of rete ridges, was poor; however, the author concluded that the three characteristic layers of the BM, that is, lamina lucida, lamina densa, and the reticular layer (anchoring fibrils), had become continuous by that time (Aihara 1989) .
Somewhat different results were obtained in another clinical study in which the skin defects of four patients were prepared by excising down to fascia before grafting with the KC sheets (Woodley et al. 1988a). Observations made at day 135 showed that hemidesmosomes, a lamina lucida, and a lamina densa were present; however, anchoring fibrils and a component of the lamina densa (7-S sites of type IV collagen) were consistently absent. The lamina densa was discontinuous, absent, or reduced except under the occasional hemidesmosomes.
The role of KC sheets in synthesis of a BM was described in a detailed study with dermis-free skin defects in a porcine model (Carver et al. 1993b). During preparation of these defects, the authors explicitly reported the excision of skin and fat, including all epidermal appendages, down to muscle fascia; this was a demonstration that the defect surface was initially free both of BM structures and of dermis. The KC sheets (Leigh et al. 1987) comprised five to six layers of flattened, undifferentiated cells, containing sparse tonofilaments and joined by desmosomes; there was no evidence of synthesis of BM structures at the completion of this in vitro stage (Carver et al. 1993b) .
In contrast to these in vitro findings, progressive synthesis of a BM was observed after the KC sheets were grafted on the dermis-free defect (Carver et al. 1993b). BM was synthesized between the KC sheet graft and the muscle fascia. Newly synthesized laminin, type IV collagen, as well as type VII collagen in the BM region were all demonstrated from day 7 onward by staining with monoclonal antibodies. A discontinuous lamina lucida and lamina densa were initially observed opposite newly formed hemidesmosomes. Within 10 days, the basal lamina became continuous while the number of hemidesmosomes reached normal levels. Maturation of the epidermis was considered complete at day 16, the time when the authors first reported formation of the outermost horny layer (stratum corneum) . The morphological data supported the conclusion that the newly synthesized anchoring fibrils originated with keratinocyte and not fibroblasts. During the BM maturation process, the number of hemidesmosomes in the membrane of the basal cell layer (basal plasma membrane) of the epidermis increased continuously and reached the number found in normal skin in 10 days. Although a fully stratified epidermis was formed during the period of the study, the length of individual hemidesmosomes did not reach normal size. In summary, during the 27-day period of observation, a highly developed BM was synthesized in vivo (Carver et al. 1993b). The detailed data confirming the synthesis of a BM in this model are presented in Table 5.1.
Morphological characterization of in vivo synthesized basement membrane (Carver et al. 1993b)
Normal basement membrane
Synthesized basement membranea
Keratinocytes with many tonofilaments
Keratinocytes with many tonofilaments
Hemidesmosome number per 10 µm of basal plasma membrane = 12.83 ± 0.84b
12.79 ± 0.98c
Hemidesmosome length per 10 µm of basal plasma membrane (HD length) = 3.43 ± 0.23 µm
2.45 ± 0.52 µmc
Sub-basal dense plate length per 10 µm of basal plasma membrane (SBDP) = 1.40 ± 0.18 µm
1.30 ± 0.28 µmc
SBDP/HD length = 0.406 ± 0.032
0.535 ± 0.055c
Individual desmosome length = 0.268 ± 0.014 µm
0.191 ± 0.035 µmc
Monoclonal antibody (MA) staining for laminin: yes
MA staining for type IV collagen: yes
MA staining for type VII collagen: yes
Anchoring fibrils: fine, plentiful
Thicker fibrils; apparently more numerous than normal
Collagen bundles beneath reticular layer: mature and well organized
Not well organized; less mature
Rete ridges: well formed
No rete ridges
Resistance of epidermis to mild abrasion: high
Similar findings were reported in an independent study (Cooper et al. 1993). KC sheets were grafted on dermis-free defects in athymic mice, prepared by full-thickness skin excision at the lateral side, sparing the pannicuus carnosus muscle underneath. Electron microscopy, based on use of highly specific antibody markers, as well as light microscopy, were used to observe the formation of laminin and type IV collagen in the BM. Laminin was synthesized by day 10; however, little or no type IV collagen could be detected at that time. Both lamina lucida and lamina densa were discontinuous by day 20 and anchoring fibrils were observed to be minimally present. At day 42, the epidermis was mature, except at the basal cell level; it was also flat, lacking rete ridges. A continuous basal lamina was observed at that time; however, staining for type IV collagen was very light and apparently discontinuous (Cooper et al. 1993) .
A much simpler pathway toward synthesis of the BM was demonstrated by culturing second-passage normal human KC for 14 days in a chemically defined medium on an inert polycarbonate filter substrate at the air-liquid interface (Rosdy et al. 1993). No dermal tissue or cells (fibroblasts) were employed in this in vitro protocol. The authors prepared the primary keratinocyte cultures in serum-free conditions; KC were then subcultured in a chemically defined medium before resuspension in a simpler defined medium and inoculation into either a cellulose filter or a polycarbonate filter. A differentiated epidermis was synthesized on the artificial substrates that was similar to living epidermis in the human adult, comprising 25 cell layers with the characteristic stratification pattern of the mature tissue. Electron microscopy showed a BM comprising a lamina lucida and a lamina densa on the surface of the polycarbonate filters. In addition, multiple hemidesmosomes with sub-basal dense plates were synthesized and numerous anchoring filaments were attached to the lamina densa. Several protein components of the BM were identified, including several noncollagenous components of anchoring filaments, heparan sulfate proteoglycan, laminin, type IV collagen, and tenascin. In contrast, type VII collagen, the essential component of anchoring fibrils, was identified inside the cytoplasm of the first layer of epidermal cells, evidence that it had been synthesized, but had not been secreted and deposited. The authors hypothesized that synthesis of anchoring filaments may require either a physically smoother substrate or the presence of dermal factors (Rosdy et al. 1993). Studies of the integrins that serve to attach basal KCs to the dermal matrix (Ghalbzouri et al. 2005) or of the different roles of fibroblasts and KCs in synthesizing laminin, and type IV and VII collagens (Lee and Cho 2005) have thrown additional light into the synthesis of the skin BM .
The combined results of the four in vivo studies (Woodley et al. 1988a; Aihara 1989; Carver et al. 1993b; Cooper et al. 1993) lead to the conclusion that a relatively mature epidermis and a BM with almost completely physiological structure can be synthesized within less than 30 days following grafting of KC sheets on dermis-free defects. There is considerable variability in the rate of the synthetic processes as well as in the morphological identification of the final structures. On the other hand, a much simpler in vitro process led to synthesis of a mature epidermis and a virtually complete BM (anchoring fibrils synthesized but not expressed) (Rosdy et al. 1993) .
5.2.5 Origins of Mechanical Failure of the Dermoepidermal Junction
In early studies of full-thickness skin defects with rodents, the dermal layer was completely excised and postage stamp-sized epidermal sheets (Thiersch grafts), free of dermal elements, were grafted on the underlying muscle. Epithelial outgrowth from the margins of the small grafts occurred through the first 2 weeks following grafting, eventually resulting in a single homogeneous sheet of epidermis. At this point, the epidermal sheet appeared to be adequately bonded to the defect surface and a system of rete ridges was observed histologically at the interface with the underlying defect tissue, which had become highly vascularized. By the third week, however, the epithelial grafts became progressively detached and the rete ridges eventually disappeared (Billingham and Reynolds 1952; Billingham and Russell 1956).
In a related experimental series, suspensions of epidermal cells, free of dermal elements, were pipetted onto the defect surface from which, as before, dermal elements had been removed. During the first 2 weeks, the epithelial cells proliferated and covered the defect with a confluent layer that closely resembled that achieved by the use of sheets of epidermis as grafts. By the third week, however, these confluent epidermal sheets showed the same lack of attachment to the defect surface (Billingham and Reynolds 1952; Billingham and Russell 1956). These authors had observed the consequences of mechanical failure (avulsion) of the bond between the graft and the defect surface.
Since these early studies, methodology for culturing KC sheets has been greatly advanced but the propensity of KC sheet grafts to mechanical failure when grafted on dermis-free defects apparently has not diminished. Frequent failure of grafted KC sheets has been observed by several independent investigators, both in animal models as well as clinically (Eldad et al. 1987; Latarjet et al. 1987; Carver et al. 1993b; Cooper et al. 1993; Kangesu et al. 1993b; Orgill et al. 1998; Fang et al. 2013). We will examine the possible reasons for such failure; in addition to its clinical significance, such failure sheds light on the synthetic processes that are activated when KC sheets are grafted on dermis-free defects.
Any graft, whether on a skin defect or a defect in another organ, is subject to detachment by small, usually uncontrolled, mechanical forces. These are typically exogenous shear forces present during experimental or clinical handling of the grafted defect; in addition, shrinkage stresses, arising from severe dehydration, can cause a skin graft to become detached (Yannas and Burke 1980). Whenever these exogenous mechanical forces become sufficiently large, or generally when the intrinsic strength of bond between graft and defect surface is sufficiently frail, the bond fails. Observers often describe this failure as a “spontaneous” loss of the graft or as lack of graft “take” or as a detachment (avulsion).
It has been occasionally suggested that KC sheet grafts have been avulsed after being displaced by the contracting dermal edges of the defect, especially in rodent models where contraction is a dominant mode of defect closure (Billingham and Reynolds 1952; Banks-Schlegel and Green 1980; Ogawa et al. 1990). Data from two studies can be used to test this suggestion. Both were carried out in swine, a model in which contraction is a much less dominant mode of defect closure than in the rodent. In one study, the dermis-free defects were grafted with KC sheets as described above (Carver et al. 1993b); in the other, the grafts were placed on a defect that was prevented from contracting by use of a specially built rigid frame (splint) (Kangesu et al. 1993b). Extensive mechanical failure of KC sheet grafts was observed in both studies. The data showed that avulsion occurred even in the absence of contraction. The hypothesis that failure of KC sheet grafts was due to contraction of defect edges was clearly not supported by the data.
In skin, the BM is located between the epidermis and the dermis, and it is often modeled as an efficient adhesive layer that keeps the epidermis and dermis (the adhints, in the analogy of an adhesive joint) securely bonded together. A direct demonstration of the contribution of the BM to the mechanical stability of skin was made by preparing specimens of the dermis with and without a BM, followed by incubating KC sheets in contact with these two surfaces. The presence of a BM in the first group of specimens was confirmed by staining for laminin and type IV collagen. In this vitro study, when the dermis lacked a BM, the KC sheet could be pulled away from its surface with negligible force; instead, the KC sheet was torn, suggesting a strong bond, when it was pulled from a dermis which had a BM (Guo and Grinnell 1989).
Another hypothesis for failure can be based on the documented inability of KC sheets to synthesize the undulating BM pattern (rete ridges) that characterizes the normal epidermis (Carver et al. 1993b). The presence of an intact, extensive rete ridge pattern has been associated with resistance to shear and peel forces (Briggaman and Wheeler 1975; Lavker 1979). As before, the dermoepidermal junction in skin is modeled simply as an adhesive joint, in which the BM plays the role of the adhesive while dermis/epidermis are the two adhints. Other factors remaining constant, the strength of the adhesive bond increases with the BM surface area (interfacial area for adhesion). This model predicts qualitatively that, in the absence of rete ridges, the extensive interfacial area of the BM is lowered significantly and the strength of the adhesive joint is accordingly reduced to the point where mechanical failure occurs much more readily. The readiness with which suction blisters can be raised on the skin of elderly subjects can be accounted for, according to this model, by the known absence in these subjects of a rete ridge pattern (Kiistala 1972; Lavker 1979).
In a clinical investigation the structural basis for the fragility of KC sheet grafts was studied by observing just where the failure occurred following formation of a standard blister both in the epidermis synthesized by grafting KC sheets and in normal skin. Blisters formed much more readily in an area grafted with a KC sheet than in intact skin. In addition, the cleavage plane of the blister at the site of KC sheet grafting was below the lamina densa of the BM while failure occurred above it in normal skin. The BM zone beneath the KC sheet grafts was found to lack a component of type IV collagen, known as 7-S sites, as well as anchoring fibrils that are present in normal skin (Woodley et al. 1988a). Results from a five-patient study showed that the tissue layer that formed in the subepidermal region contained most of the major macromolecular components of connective tissue; exceptions were the paucity of elastin fibers and poor organization of the protein linkin (microthread-like fibers). It was suggested that structural abnormalities of skin were responsible for the observed fragility of skin that formed following grafting of KC sheets (Woodley et al. 1990). Additional evidence showing that grafting of full-thickness skin defects with KC sheets leads to strongly delayed synthesis of the BM , as late as within 4–5 weeks, was described in a study of burn patients (Mommaas et al. 1992).
The morphological interpretation of detachment of the epidermis was pursued in some detail in the swine model (Carver et al. 1993b). The two surfaces resulting from avulsion of grafted KC sheets were observed by electron microscopy. The cleavage plane was found to lie between the reticular layer of the BM and the uppermost part of the granulation tissue of the two-week-old defect surface. Specifically, basal KC, lamina lucida, lamina densa and anchoring fibrils were all attached to the avulsed epidermis, while collagen fibers remained with the fibroblasts in the granulation tissue in the defect. The authors concluded that the mechanical weakness of the dermoepidermal junction was due to lack of integration of dermal collagen fibers with anchoring fibrils in the reticular layer of the BM. In contrast, a study of normal skin controls showed that an abundance of dermal collagen fibers was intertwined with the anchoring fibrils of the BM. The description of the sub-epidermal region at day 27 after grafting KC sheets showed that collagen synthesis had taken place and that new capillaries had also formed very close, within 20 mm, to the epidermis; however, collagen fiber bundles were not well organized immediately beneath the BM. The authors eventually concluded that the reported clinical problems with attachment of KC sheet grafts was related both in maturation delay of the BM as well as in poor integration with collagen of the wound bed (Carver et al. 1993b).
In another study, in which the KC sheets were grafted on dermis-free defects in athymic mice, half of the keratinocyte grafts showed blistering at days 20 and 42 (Cooper et al. 1993). Large areas of separation of the epidermis from the subepidermal region were observed over the 20-day period following grafting with the keratinocyte sheets; however, by day 42 no separation was seen at the subepidermal region. At day 42, light microscopy revealed a persistently immature epidermis without rete ridge formation while immunohistochemical staining for type IV collagen showed discontinuous staining, consistent with disruption of the BM at the points of discontinuity. Electron microscopy showed little evidence of anchoring fibrils as well as a discontinuity in the basal lamina by day 20; however, at day 42, the basal lamina had become continuous. The morphology of the subepidermal region was not described in this study; however, blood vessels were observed underneath the keratinocyte graft at day 10 (Cooper et al. 1993). Later reports of clinical studies based on use of cultured epithelial autografts have focused on the vulnerability of these grafts (low “take”) and the low long-term durability of the outcome (Wood et al. 2006; Atiyeh and Costagliola 2007; Fang et al. 2013).
In summary, the data showed that avulsion of KC sheet grafts was caused by a critical structural flaw: the lack of a mechanically competent bond between the anchoring fibrils and collagen fibers in the subepidermal layer. Most studies discussed above showed that anchoring fibrils were in fact synthesized underneath the keratinocyte sheets that had been grafted on a dermis-free defect; in contrast, a well-vascularized, thick dermis was not synthesized. The structural defect responsible for avulsion could therefore be lack of synthesis of the dermis. In other words, the adhesive joint may have failed mechanically because one of the two adhints (the dermis) was either missing or, at least, was inadequately synthesized. In contrast, a sufficiently dense mass of collagen fibers, which normally becomes enmeshed with the anchoring fibrils of the BM, is present both in normal skin and in epithelialized dermal scar, thereby preventing mechanical failure in either tissue structure.
This conclusion was further supported by a number of studies in which, prior to being grafted with KC sheets, the full-thickness skin wound was grafted with either a dermal allograft or a collagen scaffold that induces synthesis of dermis. In early studies dermal allograft was grafted on full-thickness skin wounds in animals or humans with thermal injuries, followed by removal of the necrotic allograft epidermis by abrasion several days later and resurfacing the exposed dermis with a suspension of disaggregated syngeneic KC. No graft loss was observed in these studies (Heck et al. 1985; Cuono et al. 1986, 1987; Langdon et al. 1988). It was reasoned that removal of the epidermis from the allograft eliminated most of the cells expressing alloclass II antigens, leaving behind a viable allogeneic dermal bed that successfully integrated KC cultures without rejection (Cuono et al. 1987). We note these studies among the first using decellularized matrices to circumvent the problem of organ rejection. Many more studies with decellularized matrices have been performed since then with a variety of organs, as described in a later chapter. In later studies KC suspensions were replaced with cultured epidermal autografts (CEA) that were placed over dermal allografts with similar positive results (Hickerson et al. 1994; Sheridan et al. 2001; Sood et al. 2010; Fang et al. 2013). Further support to the hypothesis that cultured epithelia autografts require a dermal bed in order to prevent avulsion was provided in a study where the dermal bed was synthesized in situ with the dermis regeneration template (DRT) rather than allografted. In this study, avulsion of KC sheets was not observed in the presence of DRT; in contrast, KC sheets were avulsed after being grafted on a DRT-free and dermis-free defect (Orgill et al. 1998).
5.2.6 Synthetic Potential and Limitations of Keratinocyte Sheet Grafts
The literature of KC sheet grafting has been strongly focused on formation of a highly differentiated epidermis and a complete BM underneath the epidermis in dermis-free defects. The majority of evidence supports the conclusion that the BM, including anchoring fibrils, derives from the epidermis (Regauer et al. 1990; Carver et al. 1993b). Photographs of histological cross-sections have typically not been extended below the BM. The histological evidence demonstrates the ability of KC sheet grafts to synthesize a BM with physiological structure, including anchoring fibrils. However, there is general lack of information about the possible induction of dermis regeneration in such studies.
Although the synthesis of a dermis has not been a priority in this area of research, it is worthwhile to examine the evidence in some detail for incidental references to such a synthesis. In an early study, a very thin layer of unidentified connective tissue, with collagen fiber axes oriented parallel to the plane of the epidermis, was evident at day 108 underneath the epidermis that formed when human cultured KC sheets were grafted in the athymic mouse (Banks-Schlegel and Green 1980). In another study with the athymic mouse, a small, “scar-like lesion” was reported at day 14, after the defect had contracted and the graft had been avulsed (Ogawa et al. 1990). An unidentified connective tissue was reported underneath the epidermis at day 42 in the athymic mouse (Cooper et al. 1993). In a study of the swine skin defect, very few, poorly organized collagen bundles were observed underneath the BM at day 27 (Carver et al. 1993b). In other studies of KC grafting on a dermis-free defect, the electron-microscopic (ultrastructural) finding in the subepidermal region was an immature dermis consisting of a few thin collagen fibers; the optical-microscopic (histological) finding was a connective tissue layer with collagen fibers highly oriented in the plane of the epidermis, reminiscent of scar (Eldad et al. 1987; Latarjet et al. 1987; Woodley et al. 1990; Orgill et al. 1998).
In summary, there is no evidence that a normal dermis is synthesized when cultured epithelial autografts (CEA, KC sheets) are grafted on a dermis-free defect. The evidence suggests strongly that, in the absence of a dermal bed, application of CEA on a full-thickness skin wound does not lead to satisfactory bonding onto the underlying tissues (typically muscle), eventually leading to avulsion of the CEA.
5.3 Synthesis of the Dermis
Having described methods for synthesizing the epidermis and the BM, we now turn to the dermis, the most important nonregenerative tissue in skin. We start with a detailed description of the dermis.
5.3.1 Structure and Function of the Dermis
The dermis is the inner layer of skin and consists of two zones. Immediately under the epidermis is the papillary dermis, comprising relatively thin collagen fibers, loosely packed, as well as the upward projections of the dermis into the epidermis (dermal papillae) with their content of vascular loops. The papillary dermis also contains fine axonal connections of unmyelinated sensory nerves that end at the epidermis. The main bulk of the dermis is the reticular layer that lies underneath the papillary dermis. It comprises highly interlacing (reticular) collagen fibers that are thicker and more closely packed than those in the papillary dermis. The mechanical strength and substantial deformability of the dermis is enhanced by the presence of elastin fibers. While the collagen fibers are highly crystalline microfibrils that stretch to a modest extent, elastin fibers are much thinner, noncrystalline (amorphous), and deform extensively, almost as much as if they were rubber bands. The combined mechanical reinforcement by these two types of fibers makes the dermis a very robust tissue (Burkitt et al. 1993) with strongly nonlinear stress–strain behavior that has been modeled in terms of the geometry of the collagen fibers (Comninou and Yannas 1976).
The dermis supports the epidermis in at least two vital ways. First, it provides a tough base that can repeatedly absorb substantial mechanical forces of various types, including shear, tensile, and compressive forces, that would have caused an unsupported epidermis to fail. Second, it incorporates a rich vascular system that is required for the metabolic support of the avascular epidermis. The blood supply of the dermis becomes intimately available to the epidermis at the dermal papillae. In addition, the dermis provides thermoregulatory control to the organism, as well as a tactile sensation.
There are several skin appendages in the dermis, including hair follicles, sweat glands, and oil-secreting (sebaceous) glands, that are embryonically derived from the epidermis (Burkitt et al. 1993). The adipose layer underneath the dermis, the hypodermis (subcutis) is often considered to be part of the dermis (Young et al. 2006). In some areas of the body (e.g., scalp) the hypodermis contains the lower parts of many hair follicles.
5.3.2 In Vivo Synthesis of the Dermis Using the Cell-Free Dermis Regeneration Template
Dermis was partially synthesized when the DRT was grafted on a dermis-free defect in the adult guinea pig either as a cell-free scaffold (Yannas and Burke 1980; Yannas 1981) or as a keratinocyte-seeded scaffold (Yannas et al. 1981, 1982b; Orgill 1983; Yannas et al. 1989; Murphy et al. 1990). This observation was confirmed in the adult swine model (Compton et al. 1998) and in clinical trials with humans (Burke et al. 1981; Heimbach et al. 1988). In all these cases the dermis was imperfectly regenerated as it lacked adnexa (hair follicles, sweat glands, etc.) In this section we will focus on use of the cell-free DRT; ensuing sections will describe studies with the cell-seeded DRT.
The commercially available product is Integra™, manufactured by Integra LifeSciences.
A dramatic delay in onset of contraction of dermal edges, amounting to about 20 days relative to the ungrafted defect, was observed in the guinea pig study when cell-free DRT was grafted on it (Fig. 5.2; curve labeled DRT)(Yannas 1981). Following this initial delay in onset of contraction, the cell-free scaffold was degraded, contraction eventually started and was responsible for closure of most of the defect area; the remainder, about 12 % of initial defect area, eventually closed by epithelialization. Underneath this epithelialized layer was a small mass of connective tissue that was tentatively labeled “neodermis” on the evidence that it comprised loosely packed collagen fibers. These features stood in contrast to scar tissue, comprising tightly packed collagen fibers, that formed in ungrafted controls underneath the newly epithelialized area (Yannas 1981; Yannas et al. 1981 1982b).
Contraction kinetics following grafting with cell-free and cell-seeded DRT. The dermis-free defect in the guinea pig was grafted with a analog B, an ECM analog identical to DRT in structure except with average pore diameter 450 µm (inactive ECM analog), b keratinocyte-seeded dermis regeneration template (KC + DRT); c cell-free DRT. Cell-free DRT delayed contraction but did not arrest it; eventually, only a small mass of dermis was synthesized. KC-seeded DRT arrested contraction at 35–40th day and the defect perimeter continued increasing at a rate higher than predicted by animal growth to yield a partial skin regenerate (appendages missing) occupying two thirds of initial defect area at day 200. (Source: Yannas et al. 1989)
Subsequent studies confirmed that loss of regenerative activity of the ECM analog occurred when either the chemical composition, half-life for degradation or average pore diameter were each displaced from a rather narrow range, as described further in Chaps. 8 and 9 (Yannas et al. 1982b, 1989). The combined results suggested very strongly that the ECM analog induced regeneration of a dermis in the guinea pig and that regeneration did not occur unless the structure of the ECM analog was tightly controlled within narrow limits (Yannas et al. 1989; Murphy et al. 1990). These results suggested the specific name DRT for the active ECM analog.
The magnitude of defect contraction on the regenerative activity of DRT was studied by comparing closure of defects in the guinea pig and the swine, two species which show different wound contraction behavior. As expected from data in Table 5.3, a dermis-free defect in the swine spontaneously closed by contraction to a lower extent than in the guinea pig. It was observed that DRT significantly delayed the onset of contraction in both animal models . At the end of the study, at day 21, histological data showed a new bed of thick collagen bundles, randomly oriented, resembling dermis rather than scar, both in the guinea pig and the swine models (Orgill et al. 1996). By day 21, grafted defects in the guinea pig had closed largely by contraction while only about 12 % of the defect area was closed by an epithelialized dermis. In swine defects, contraction had virtually stopped by day 21, and defect closure was mostly completed by epithelial migration from the defect edges and over the newly synthesized dermis (Orgill et al. 1996). It was concluded that the mass of dermis synthesized in the presence of DRT in the swine model (the animal model with defects that contracted less) was higher than in the guinea pig.
In two later studies with the swine model, the cell-free DRT was grafted on full-thickness skin defects and was studied as a control of its keratinocyte-seeded version (see below). The time allocated for study of the unseeded DRT, about 2 weeks, was insufficient for complete re-epithelialization of the newly synthesized dermis from the wound edges. A well-vascularized dermis with an extensive network of capillaries had formed inside the defect (Butler et al. 1998). In a 15-day study of the cell-free DRT in the swine model, the latter was degraded and replaced by a densely cellular connective tissue with a high degree of vascularity that resembled an immature dermis (neodermis) (Compton et al. 1998). The synthesis of BM was not directly examined in these protocols; however, the available evidence suggested that, prior to eventual reepithelialization from the edges of the defect, BM had not been simultaneously synthesized over the dermis in the presence of the keratinocyte-free DRT.
Angiogenesis of DRT occurs spontaneously, although with a delay of several days, following grafting; this fact has been studied and put to use in a number of ways in surgical settings. Studies have shown that angiogenesis initially appeared in DRT at day 7 following grafting, and peaked between day 7 and 14 (Shaterian et al. 2009); DRT was found capable of active neovascularization and capable of serving as an effective dermal substitute (DS) in avascular wounds (Baynosa et al. 2009); also DRT induced angiogenesis in flap prefabrication, and therefore could be used to support vascularization survival of a vascular prefabricated skin flap (Yan et al. 2011).
Clinical studies of DRT with burn patients also repeatedly showed synthesis of a dermis on full-thickness skin defects. These wounds had been excised down to muscle fascia early following injury (Burke et al. 1974) prior to grafting with DRT. With massively burned patients, rather than waiting for the much slower epidermal migration from the edges of the defect to cover the very large defects, the neodermal layer was covered with a thin, autologous epidermal graft, largely free of dermis and about 0.10–0.15 mm in thickness within a few weeks after grafting (ten patients, Burke et al. 1981; 106 patients, multicenter trial, Heimbach et al. 1988). Detailed histological study of biopsies from the second clinical study showed that DRT fibers gradually disappeared, the newly synthesized collagen fibers became more coarse, and a distinction between papillary and reticular layers of the dermis appeared in the tissue layer that the authors reported as the “intact dermis.” Scar formation was not observed either at the gross or the histological level at any time during the course of healing. No skin appendages were evident; and rete ridges were not observed in this study (Stern et al. 1990). A companion clinical immunological study showed a very small rise in immunological activity in patients’ sera in reaction to the macromolecular components of DRT, bovine skin collagen and chondroitin 6-sulfate. The overall conclusion from the clinical study was that DRT presented few, if any, humoral immunological problems to patients (Michaeli and McPherson 1990). Other clinical studies of DRT with massively burned patients have emphasized follow-up of pediatric patients over a 6-year (Burke 1987; Tompkins et al. 1989) or 10-year period (Sheridan et al. 1994). The new integument was reported to be free of restrictions to joint function, indicative of absence of contractures; most interestingly, the new skin had the ability of growing as the child grew (Burke 1987; Sheridan et al. 1994). A study with 11 patients who were being treated with DRT for deep hand burns showed that the treated skin sites were flexible and supple and did not adhere to the deeper layers, thereby permitting free articular and functional movement (Dantzer et al. 2003). The burned breast was reconstructed with DRT and the authors reported the presence of elastin fibers throughout the neodermis, as well as superior patient satisfaction to treatment with thick split-thickness grafting (Palao et al. 2003). Objective evaluation of skin resulting from treatment with the DRT and with split-thickness skin grafts (SSGs) in six burn patients, using an instrumented suction device (Cutometer) showed that the elastic properties of sites grafted with DRT were comparable to normal skin while those treated with SSG were not (Nguyen et al. 2010). Two views of the new skin with patients following use of DRT are shown (Figs. 5.3 and 5.4).