Fibroblasts in isogeneic skin equivalents persist for long periods after grafting

Current biofabrication methods for vascular tissue engineering and an introduction to biological textiles

Cardiovascular diseases are the leading cause of mortality in the world and encompass several important pathologies, including atherosclerosis. In the cases of severe vessel occlusion, surgical intervention using bypass grafts may be required. Synthetic vascular grafts provide poor patency for small-diameter applications (< 6 mm) but are widely used for hemodialysis access and, with success, larger vessel repairs. In very small vessels, such as coronary arteries, synthetics outcomes are unacceptable, leading to the exclusive use of autologous (native) vessels despite their limited availability and, sometimes, quality. Consequently, there is a clear clinical need for a small-diameter vascular graft that can provide outcomes similar to native vessels. Many tissue-engineering approaches have been developed to offer native-like tissues with the appropriate mechanical and biological properties in order to overcome the limitations of synthetic and autologous grafts. This review overviews current scaffold-based and scaffold-free approaches developed to biofabricate tissue-engineered vascular grafts (TEVGs) with an introduction to the biological textile approaches. Indeed, these assembly methods show a reduced production time compared to processes that require long bioreactor-based maturation steps. Another advantage of the textile-inspired approaches is that they can provide better directional and regional control of the TEVG mechanical properties.

Immune tolerance of tissue-engineered skin produced with allogeneic or xenogeneic fibroblasts and syngeneic keratinocytes grafted on mice

Organs are needed for the long-term replacement of diseased or wounded tissues. Various technologies based on cells seeded in synthetic or biomaterial scaffolds, or scaffold-free methods have been developed in order to produce substitutes that mimic native organs and tissues. For cell-based approaches, the use of living allogeneic fibroblasts could potentially lead to the production of "off-the-shelf" bioengineered organs/tissues. However, questions remain regarding the outcome of allogeneic grafts in terms of persistence of allogeneic cells, tolerance and the host immune reaction against the tissue after implantation. To evaluate graft tolerance of engineered-tissues containing non-autologous fibroblasts, tissue-engineered skin substitutes (TESs) produced with syngeneic, allogeneic or xenogeneic fibroblasts associated with syngeneic, allogeneic or xenogeneic epithelial cells were grafted in mice as primary and secondary grafts. The immune response was evaluated by histological analysis and immunodetection of M2 macrophages, CD4- and CD8-positive T cells, 15, 19, 35 and 56 days after grafting. Tissue-engineered skin composed of non-autologous epithelial cells were rejected. In contrast, TESs composed of non-autologous fibroblasts underlying syngeneic epithelial cells were still present 56 days after grafting. This work shows that TES composed of non-autologous fibroblasts and autologous epithelial cells are not rejected after grafting. STATEMENT OF SIGNIFICANCE: We found that tissue-engineered skin substitutes produced by a scaffold-free cell-based approach from allogeneic fibroblasts and autologous epithelial cells are not rejected after grafting and allow for the permanent coverage of a full-thickness skin wounds. In the field of tissue engineering, these findings open the possibility of selecting a human fibroblastic or stromal cell population based on its biological properties and adequate biosafety, banking it, in order to produce "ready-to-use" bioengineered organs/tissues that could be grafted to any patient without eliciting immune reaction after grafting. Our results can be generalized to any organs produced from fibroblasts. Thus, it is a great step with multiple applications in tissue engineering and transplantation.

Characterization of a Cell-Assembled extracellular Matrix and the effect of the devitalization process

We have previously shown that the Cell-Assembled extracellular Matrix (CAM) synthesized by normal, human, skin fibroblasts in vitro can be assembled in a completely biological vascular graft that was successfully tested in the clinic. The goal of this study was to perform a detailed analysis of the composition and the organization of this truly bio-material. In addition, we investigated whether the devitalization process (dehydration) used to store the CAM, and thus, make the material available "off-the-shelf," could negatively affect its organization and mechanical properties. We demonstrated that neither the thickness nor the mechanical strength of CAM sheets were significantly changed by the dehydration/freezing/rehydration cycle. The identification of over 50 extracellular matrix proteins highlighted the complex composition of the CAM. Histology showed intense collagen and glycosaminoglycan staining throughout the CAM sheet. The distribution of collagen I, collagen VI, thrombospondin-1, fibronectin-1, fibrillin-1, biglycan, decorin, lumican and versican showed various patterns that were not affected by the devitalization process. Transmission electron microscopy analysis revealed that the remarkably dense collagen network was oriented in the plane of the sheet and that neither fibril density nor diameter was changed by devitalization. Second harmonic generation microscopy revealed an intricate, multi-scale, native-like collagen fiber orientation. In conclusion, this bio-material displayed many tissue-like properties that could support normal cell-ECM interactions and allow implantation without triggering degradative responses from the host's innate immune system. This is consistent with its success in vivo. In addition, the CAM can be devitalized without affecting its mechanical or unique biological architecture. STATEMENT OF SIGNIFICANCE: The extracellular matrix (ECM) defines biological function and mechanical properties of tissues and organs. A number of promising tissue engineering approaches have used processed ECM from cadaver/animal tissues or cell-assembled ECM in vitro combined with scaffolds. We have shown the clinical potential of a scaffold-free approach based on an entirely biological material produced by human cells in culture without chemical processing. Here, we perform a comprehensive analysis of the properties of what can truly be called a bio-material. We also demonstrate that this material can be stored dried without losing its remarkable biological architecture.

Engineering a skin replacement

Major skin loss from trauma or burns cannot always be replaced with the patient's own skin. An engineered skin replacement would restore the barrier function of the skin, remain permanently on the wound, and minimize late functional complications of wound contraction. Cultured epithelial autograft (CEA) sheets reproduce the epidermis' function and have been used in burn patients to close large wounds. There are several promising avenues for dermal replacement, but none has yet had wide clinical application.

Viability and Function of Autologous and Allogeneic Fibroblasts Seeded in Dermal Substitutes after Implantation

BACKGROUND

Fibroblast-seeded collagen sponges have been used for the treatment of skin defects and skin ulcers. However, the viability of the fibroblasts after implantation is still unknown. The objective of this study was to investigate the viability and distribution of autologous and allogeneic fibroblasts after implantation and to clarify which type is more effective for wound healing.

MATERIALS AND METHODS

Skin samples of Hartley guinea pigs were retrieved and autologous fibroblasts were isolated and cultured. Fibroblasts isolated from the skin of a Strain2 guinea pig were used as allogeneic fibroblasts. Three full-thickness wounds were created on the backs of guinea pigs and an acellular collagen sponge, a collagen sponge seeded with autologous fibroblasts, and a collagen sponge seeded with allogeneic fibroblasts were transplanted. Before implantation, fibroblasts were labeled with PKH26. The guinea pigs were sacrificed 1, 2, and 3 weeks after implantation. The epithelization and contraction of the wounds were assessed, and the viability and distribution of the seeded fibroblasts were observed in cross sections.

RESULTS

Three weeks after implantation, the PKH26-labeled autologous and allogeneic fibroblasts remained viable. In the wounds covered with the autologous fibroblast-seeded collagen sponge, the epithelization was fastest, and the percent wound contraction was smallest. In contrast, in the wounds covered with allogeneic fibroblasts, the epithelization was slowest and the percent contraction was largest.

CONCLUSION

The allogeneic fibroblasts seeded in the collagen sponge survived and remained viable on the grafted area, but did not accelerate wound healing.

Allogeneic fibroblasts in dermal substitutes induce inflammation and scar formation

In the present study, we compared the use of autologous versus allogeneic fibroblasts in dermal skin substitutes in a porcine wound model. The allogeneic fibroblast populations were isolated from female and a male pig (allo-1, - 2 and - 3) and the controls, autologous fibroblasts, from female graft-recipient pigs (control). The histocompatibility of the three donor pigs with the recipient pigs was determined with a mixed lymphocyte reaction. In two pigs, full-thickness wounds were treated with the fibroblast-seeded dermal substitutes (n = 5 per animal) and immediately overgrafted with meshed split-skin autografts. After 6 weeks, wound contraction was measured by planimetry and scar formation was scored. At 2, 4, and 6 weeks biopsies were taken and evaluated for the presence of inflammatory reactions, myofibroblasts, and scar formation. The mixed lymphocyte reaction of both recipient pigs showed the highest responses on peripheral blood mononuclear cells of the allo-3 donor pig, and was low or negative for allo-1 and allo-2. In all "allogeneic" wounds, more inflammatory cells were observed over time along with inflammatory foci consisting of a mix of lymphocytes and granulomatous cells. After 4 weeks, myofibroblasts were absent in the control wounds, whereas in "allogeneic" wounds, myofibroblasts colocalized with inflammation foci. The final scar tissue of the "allogeneic" wounds showed granulating areas with thin, immature collagen bundles. In contrast, the control wounds showed a dermal tissue with mature collagen bundles organized randomly like in normal skin. The wounds treated with allo-3 fibroblasts showed in both pigs a significant increase in scar formation and wound contraction when compared with control wounds. In conclusion, for optimal restoration of dermal skin function with minimal scar formation, skin substitutes containing autologous fibroblasts are preferred over skin substitutes with allogeneic fibroblasts.

Cultured autologous fibroblasts augment epidermal repair

BACKGROUND

Autologous dermal fibroblasts may be useful in the treatment of skin wounds and for the enhancement of keratinocyte proliferation. This paper addressed the following questions: (1) can cultured fibroblasts (CF) be transplanted as suspensions to full-thickness skin wounds and do they influence wound healing; (2) will the transplanted CF be integrated into the new dermis; (3) can a transgene that encodes a secretable marker, human epidermal growth factor (hEGF), be expressed in the wound fluid by the transplanted CF; and (4) do CF cotransplanted with cultured keratinocytes (CK) influence the rate of wound healing?

METHODS

Suspensions of CF were transplanted alone or together with CK to full-thickness wounds covered with liquid-containing chambers in an established porcine model.

RESULTS

Transplantation of CF accelerated reepithelialization as determined from wound histologies and sequential measurements of protein efflux over the wound surface. CF transfected with a marker gene, beta-galactosidase, resulted in in vivo gene expression and demonstrated that transplanted CF integrated into the developing dermis. Transplantation of hEGF gene-transfected CF resulted in significant hEGF expression in wound fluid. The hEGF levels peaked at day 1 (2450 pg/ml) and then sharply decreased to low levels on day 6. CF cotransplanted with CK led to greater number of keratinocyte colonies in the wound and accelerated reepithelialization as compared with CK alone.

CONCLUSIONS

Transplanted CF integrated into the dermis, accelerated reepithelialization, and improved the outcome of CK transplantation. CF may also be used for the expression of transgenes in wound and wound fluid.

Modification of fibroblast gamma-interferon responses by extracellular matrix

Fibroblasts from scaffold-based three-dimensional human cultures have been demonstrated to colonize ulcer wound beds and persist for at least 6 mo without rejection. This study examines the expression in these cultures of molecules associated with activation of the immune system in acute rejection. Studies in monolayer cultures showed that fibroblasts expressed CD40 at about 10% of the surface density seen in umbilical vein endothelial cells, whereas HLA-DR was undetectable. In these cultures, both molecules were induced by gamma-interferon. In scaffold-based three-dimensional cultures, however, a majority of the fibroblasts showed little induction of CD40 and HLA-DR in response to gamma-interferon, although HLA class I expression was increased. Fibroblasts re- isolated from the three-dimensional cultures and cultured in monolayers recovered HLA-DR induction in response to gamma-interferon. Fibroblasts cultured in an alternative three-dimensional system using collagen gels showed CD40 and HLA-DR induction by gamma-interferon in the same manner as monolayer cultures. Comparison of phosphorylation of signal transducer and activator of transcription 1 on tyrosine-701 showed it to be similar in monolayer and three-dimensional culture, and phospho-signal transducer and activator of transcription 1 moved into the nucleus. Induction of the class II transcription activator was greatly reduced, however. We propose that interaction of fibroblasts with the fibroblast-derived extracellular matrix is an important modulator of gamma-interferon responsiveness and that this interaction may play a role in the low immunogenicity of allogeneic fibroblasts grown on scaffolds.

A novel treatment for venous leg ulcers

Venous ulceration, a relatively common manifestation of venous hypertension, is often difficult to treat. This article reports the authors' experience with a new wound-healing technology using a bilayered, culture-derived human skin equivalent (HSE, Apligraf) for treatment of venous ulcers. In the patients studied, HSE appeared to promote wound healing in three ways: 1) apparent graft "take"; 2) temporary wound closure (persistence of HSE with subsequent wound re-epithelialization from wound margins); and 3) stimulation of host healing without temporary persistence by acting as a biologic dressing. The demonstrated efficacy of HSE suggests that it will prove useful for promoting the healing of venous ulcers.

Advantage of the presence of living dermal fibroblasts within in vitro reconstructed skin for grafting in humans

Methods for serial cultivation of human keratinocytes can provide large quantities of epidermal cells, which have the potential of restoring the vital barrier function of the epidermis in extensive skin defects such as burns. To investigate the value of combining an epidermis with a dermal component, fibroblasts originated from the superficial dermis were used to seed a collagen lattice as described by E. Bell (dermal equivalent). Beginning in 1981, we grafted 18 patients (burns and giant nevi) using 35 grafts 10 x 10 cm in size. In the course of this work, the original technique was modified and improved as experience was gained. We began by using small skin biopsy samples as a source of keratinocytes cultured on a dermal equivalent before grafting in a one-step procedure, but this gave poor cosmetic results, because of a nonhomogeneous epidermalization. We then chose to cover the graft bed using a two-step procedure. The first step consisted of grafting a dermal equivalent to provide a dermal fibroblast-seeded substrate for subsequent in vivo epidermalization by cultured epidermal sheets. Whatever the epidermalization technique used, a living dermal equivalent applied to the graft bed was found to reduce pain, to provide good hemostasis, and to improve the mechanical and cosmetic properties of the graft. A normal undulating dermal-epidermal junction reappeared by 3 to 4 months after grafting and elastic fibers were detectable 6 to 9 months after grafting. As a result of the biosynthesis of these products, the suppleness (e.g., elasticity) of the grafts was closer to that of normal skin than the cicatricial skin usually obtained with epidermal sheets grafted without the presence of living dermal cells. This rapid improvement of the mechanical properties of the graft could be attributed to the presence of fibroblasts cultured from the dermis and seeded into the collagen matrix.

Design of an artificial skin. IV. Use of island graft to isolate organ regeneration from scar synthesis and other processes leading to skin wound closure

Deep skin wounds in the adult mammal close spontaneously by epithelialization, wound contraction, and scar synthesis. In previous wound healing studies, it has been unsuccessfully attempted to separate from each other the natural processes that close wounds. In this study, we attempted to isolate skin regeneration from spontaneous processes of wound closure using "island" grafts. A porous analog of the extracellular matrix, composed of a graft copolymer of type I collagen and chondroitin 6-sulfate, was seeded with uncultured autologous keratinocytes and served to induce regeneration of the dermis and the epidermis. Grafts of the copolymer, measuring 1 x 2 cm, were placed in the center of 5 x 6-cm wounds in guinea pigs. By day 14, the edges of the island grafts were clearly separated from the host epidermis and dermis by a distinct bed of granulation tissue. Histologic study of island grafts on day 14 showed that the copolymer grafts had largely degraded and that a new epidermis and dermis had been synthesized in its place. The thickness of the new epidermis increased as the density of cells seeded into the graft increased. No synthesis of epidermis or dermis was observed in the granulation tissue outside the perimeter of the island grafts. We conclude that island grafting allows the study of early events in skin regeneration in isolation from epithelialization, contraction, and scar synthesis.

Design of an artificial skin. IV. Use of island graft to isolate organ regeneration from scar synthesis and other processes leading to skin wound closure

Deep skin wounds in the adult mammal close spontaneously by epithelialization, wound contraction, and scar synthesis. In previous wound healing studies, it has been unsuccessfully attempted to separate from each other the natural processes that close wounds. In this study, we attempted to isolate skin regeneration from spontaneous processes of wound closure using "island" grafts. A porous analog of the extracellular matrix, composed of a graft copolymer of type I collagen and chondroitin 6-sulfate, was seeded with uncultured autologous keratinocytes and served to induce regeneration of the dermis and the epidermis. Grafts of the copolymer, measuring 1 x 2 cm, were placed in the center of 5 x 6-cm wounds in guinea pigs. By day 14, the edges of the island grafts were clearly separated from the host epidermis and dermis by a distinct bed of granulation tissue. Histologic study of island grafts on day 14 showed that the copolymer grafts had largely degraded and that a new epidermis and dermis had been synthesized in its place. The thickness of the new epidermis increased as the density of cells seeded into the graft increased. No synthesis of epidermis or dermis was observed in the granulation tissue outside the perimeter of the island grafts. We conclude that island grafting allows the study of early events in skin regeneration in isolation from epithelialization, contraction, and scar synthesis.

Characterization of a new in vitro model for studies of reepithelialization in human partial thickness wounds

Reepithelialization of artificial partial thickness wounds made in biopsies of human skin was determined after 3, 5, or 7 d of incubation, submerged or elevated to the air-liquid interface. The biopsies were reepithelialized within 5-7 d, with a more complete epidermal healing in wounds exposed to air. Both types of wounds showed similar time-course in deposition of basement membrane components, as detected by immunofluorescence labeling. Laminin and collagen type VII were deposited underneath the migrating tips, whereas collagen type IV was detected after reepithelialization. Markers of terminal differentiation showed a pattern close to normal in the air-liquid incubated wounds after reepithelialization. Involucrin was detected in the suprabasal regions of the migrating epidermis and thereafter in the upper half of neo-epidermis in the air-liquid incubated wound. Filaggrin could not be detected in the submerged wounds at any time during healing, whereas wounds exposed to air showed a well-differentiated epidermis by Day 7. Tritiated thymidine-incorporation indicated proliferation of epidermal and dermal cells during reepithelialization and a maintained viability, as shown by cultivation of endothelial- and fibroblast-like cells obtained from the dermis 7 d after wounding. Reepithelialization in this human in vitro model is supported by a matrix close to normal with the possibility of extracellular influences and cell-cell interactions and, in addition, the technique is simple and reproducible. Therefore, we suggest this model for studies of regeneration in culture and as a complement to in vivo studies on epidermal healing.

Development of a skin model based on insoluble fibrillar collagen

A biocompatible, 3-dimensional, noncontracting, crosslinked collagen matrix was adapted to promote differentiation of epidermal keratinocytes. To produce the matrix, a 3% wt/wt dispersion of insoluble bovine collagen containing 5 mg polylysine/g collagen in 0.001 N HCl was blended, lyophilized, and crosslinked using a dehydrothermal technique. Matrices 4 cm2 and 3 mm thick were seeded with human dermal fibroblasts (1 x 10(5)/cm2). After 5 days in culture, the matrices were seeded with human epidermal keratinocytes (1 x 10(5)/cm2). The cultures were grown submerged for 1 week and raised to the liquid/air interface for 3 weeks to promote epidermal differentiation. Based on morphology and immunological staining with antibodies for human involucrin, keratin 1 (k1), filaggrin, and loricrin, the state of differentiation of the epidermal layer was nearly equivalent to that seen with cultures grown on contracted collagen lattices produced according to the methodology described in the literature and similar to the pattern produced in normal neonatal foreskin. These results demonstrate the usefulness of an in vitro skin model employing a crosslinked collagen matrix that permits the incorporation of additional covalently linked bioactive molecules during matrix formation.

Cultured epithelial autograft: five years of clinical experience with twenty-eight patients

Cultured epithelial autograft (CEA) has been used as an adjunct in burn wound coverage at the Vancouver Hospital and Health Sciences Centre since 1988, and has been available to all patients admitted with significant burn injuries. During the 5-year period from 1988 to 1992 inclusive, 28 patients treated with CEA survived long enough for assessment. The mean age was 35.3 years with a mean total body surface area burn of 52.2% and a mean total full thickness injury of 42.4%. CEA was applied to wounds covering between 2% and 35% body surface area (BSA; mean 10.4%) after excision to fat or fascia. Most wounds had interim homograft coverage. Preservation of homograft dermis was attempted in three patients at the time of removal without effect. The mean CEA "take" was 26.9% of the grafted area. Eight patients had 50% or greater take and were discharged with between 1 and 19% BSA covered with CEA. Thirteen patients had no take on wounds between 2 and 16% BSA. Overall mortality in burn patients treated at the Vancouver Hospital and Health Sciences Centre from 1988 to 1992 was not significantly different from 1983 to 1987 with the populations being similar in terms of total BSA burns, age, inhalation injury, and homograft availability. When compared to a matched control population from the preceding 5 years, when CEA was not available, there was no significant difference in duration of hospital stay or number of autograft harvests. However, approximately one more debridement without autograft harvest per CEA patient occurred. Timing and depth of wound excision, interim coverage, type of dressing, and wound microbiology were not found to influence good versus poor take. The anterior trunk and thighs were the best recipient sites. Subjective differences between CEA and meshed autograft were noted. The results show that after 5 years of use, CEA engraftment continues to be unpredictable and inconsistent, and hence, it should be used as only a biologic dressing and experimental adjunct to conventional burn wound coverage with split thickness autograft.

New grafts for old? A review of alternatives to autologous skin

Immediate resurfacing of skin defects is a challenging prospect, especially in patients with extensive full-thickness burns. Currently, split-thickness autografts offer the best form of wound coverage, but limited donor sites and their associated morbidity have prompted the search for alternatives. The application of allogeneic skin is restricted by availability and the risk of transmission of infection, whilst synthetic skin substitutes are simply expensive dressings. The problems of limited expansion may be overcome by culturing keratinocytes in vitro. Unlike autologous cells, allogeneic keratinocytes are available immediately, although they survive for less than a week when applied to full-thickness skin defects. Moreover, the absence of a dermal component in these grafts predisposes to instability and contracture. A cross-linked collagen and glycosaminoglycan dermal substitute, covered with thin split-skin grafts or cultured autologous keratinocytes, shows promise in burns patients. An alternative is a collagen matrix populated by allogeneic fibroblasts and overlaid with cultured autologous or allogeneic keratinocytes. The clinical application of cultured grafts remains imperfect but offers the prospect of immediate coverage and massive expansion.

Transplantation of mucosal tissue model composed of rabbit oral mucosal cells

Since skin grafted into the oral cavity does not differentiate into mucosa, its original characteristics remain unchanged. This may cause discomfort to patients treated with such grafted skin, due to the skin's hair follicles and keratinization. This phenomenon is controlled by subepithelial connective tissues, and is thus referred to as the Epithelio-Mesenchymal Interaction. The grafting of mucosa itself would be preferable, but this procedure is severely limited due to the scarcity of suitable tissue. Accordingly, we have recently prepared artificial mucosa in vitro, using fibroblasts and epithelial cells derived from rabbit oral mucosa and collagen gel, and transplanted it into a donor animal. This paper is the first report of a method using artificial mucosa for the reconstruction of mucosal defects.

A new method for studying epidermalization in vitro

A new method for studying epidermalization in vitro is described. It consists of inserting a punch biopsy that serves as a source of epidermis into dermal equivalent freshly made up, with fibroblasts mixed in a collagen matrix. Fibroblasts cling to collagen fibrils and contract the matrix, leading in 3 days to a resistant dermal equivalent holding the punch biopsy firmly in place. At day 5, a culture medium favouring epidermal growth was used and a fringe of a new epidermis appeared around the punch, the area of which grew linearly with time. This new epidermis showed a pattern of differentiation similar to epidermis in vivo, with cuboidal basal cells, keratohyalin granules, membrane coating granules and the expression of the 65-67 kd keratin subset. The method seems to combine the advantages of the explant technique and of classical keratinocyte cultures, providing the researcher with a large quantity of differentiated epidermis, the pharmacologist with simple and quantitative system in which to study modifications of growth and differentiation of epidermis, and the plastic surgeon with a possible material for skin grafting.

Reconstruction of a thyroid gland equivalent from cells and matrix materials

A living thyroid gland equivalent has been fabricated with a cultivated strain of rat thyroid cells (FRTL), dermal or thyroid fibroblasts, and matrix materials. The mixture becomes tissuelike in vitro by virtue of interactions between fibroblasts and collagen. Initially in vitro the thyroid cells are uniformly distributed as single cells and a small number of clusters containing less than 10 cells. When implanted into thyrodectomized hosts the thyroid cells in the tissue lattice become organized into follicles containing a colloidlike material. The follicles were found in clusters in sizes up to 0.3 mm. The clusters are vascularized. It is thought that they arise within clones rather than by an aggregation process. Using an antithyroglobulin antibody it was shown that both thyroid cells and the colloidlike material within follicles reacted positively. Development of follicles was strictly dependent on whether hosts were thyroidectomized. In nonthyroidectomized hosts no follicles were observed. We conclude that an organotypic structure can develop in vivo from a "gland-equivalent" fabricated with adult cells and matrix materials combined in vitro, and that cells cultivated for years in vitro retain the capacity to express differentiated functions in a reconstituted organ equivalent in vivo.