A novel strategy to engineer pre-vascularized 3-dimensional skin substitutes to achieve efficient, functional engraftment

Autologous split-thickness skin grafts are the preferred treatment for excised burn wounds, but donor sites for autografting are often limited in patients with extensive burns. A number of alternative treatments are already in use to treat large burns and ulcers. Despite intense efforts to develop tissue-engineered skin, delayed or absent vascularization is one of the major reasons for tissue-engineered skin engraftment failure. To overcome these problems, we developed a scaffold-free 3-dimensional (3D) skin substitute containing vascular networks that combine dermal fibroblasts, endothelial cells, and epidermal keratinocytes based on our layer-by-layer cell coating technique. We transplanted the pre-vascularized 3D skin substitutes onto full-thickness skin defects on severe combined immunodeficiency mice to assess their integration with the host tissue and effects on wound healing. We used non-vascularized 3D skin substitutes as a control. Vessels containing red blood cells were evident in the non-vascularized control by day 14. However, blood perfusion of the human-derived vasculature could be detected within 7 days of grafting. Moreover, the pre-vascularized 3D skin substitutes had high graft survival and their epidermal layers were progressively replaced by mouse epidermis. We propose that a novel dermo-epidermal 3D skin substitute containing blood vessels can promote efficient reconstruction of full-thickness skin defects.


Results
Characteristics of scaffold-free pre-vascularized 3D skin substitutes. Over approximately 15 days of cell culture, including cell expansion periods, we fabricated pre-vascularized dermo-epidermal 3D skin substitutes without using exogenous scaffolds or biomaterials (Fig. 1). In our system, GFP-expressing HUVEC demonstrated elongated morphology and multicellular structures within the dermal layer by day 4 of cultivation. Over the following days, the vessel-like structures increased in number and length (data not shown). By day 7, mature vessels with lumens were visible (Fig. 2a). In our experiments, cells were cocultured at high density (>1 × 10 7 cells/cm 2 ) to construct scaffold-free 3D skin substitutes. We evaluated the constructs in cell viability assays. Compared to the number of green-stained viable cells, there were relatively few red-stained non-viable cells during cultivation. Although a few dead cells appeared after 4 days, the number of live cells remained much higher even after 7 days of cultivation (Fig. 2b). www.nature.com/scientificreports www.nature.com/scientificreports/ To evaluate the ratio of HUVEC to FN-G-coated fibroblasts for optimal cell density, we seeded 0.1 to 1 × 10 5 HUVEC per culture insert while keeping the number of FN-G-coated NHDF constant (1 × 10 7 cells). Coculture of FN-G-coated NHDF and HUVEC successfully induced the formation of vessel-like networks in the reconstructed dermis consisting NHDF, which varied based on the fibroblast-to-endothelial cell ratio, after 7 days of keratinocyte differentiation (Fig. 3a). The vessel density and branching index (the number of branches per total vessel area) were all maximized by seeding 1 × 10 5 HUVEC (Fig. 3b,c). On the other hand, vessels did not form in the absence of FN-G-coated NHDF (data not shown).
To prepare the pre-vascularized 3D skin substitutes for transplantation, HUVEC (0.2 × 10 5 cells) and FN-G-coated NHDF (1 × 10 7 cells) were seeded and the surface of the dermal-like layer was inoculated with keratinocytes. Following culture at the air-liquid interface for 7 days, we observed little contraction of the structure of the 3D skin (Fig. 4a). Histological examinations of the pre-vascularized 3D skin substitutes revealed a well-stratified, cornified epidermis and a homogenous dermal component with dense collagen fibers (Fig. 4b,c). Laminin 5, an essential component of the basement membrane structure, was detected in a continuous line at the dermal-epidermal junction after maturation in vitro (Fig. 4d). Coculture of HUVEC and FN-G-coated NHDF seeded with keratinocytes for 9 days resulted in the formation of CD31 + lumenized vessels within the dermal component (Fig. 4e). The constructs were approximately 320 μm thick with a pre-vascularized dermis of 200 μm.

Engineered human blood vessels were functional in vivo.
To demonstrate the functionality of the constructed vascular plexuses in vivo, we transplanted pre-vascularized 3D skin substitutes onto the excisional full-thickness wounds of immunodeficient mice. On day 7, transillumination analysis revealed that the artificial dermal substitute-treated wounds appeared more transparent than the non-vascularized 3D skin substitutes (control), whereas both control-and pre-vascularized substitute-treated wounds contained blood vessels growing from the surrounding host tissue. In addition, gross hyperemia and more abundant invasion of the vasculature were observed in the wounds treated with pre-vascularized substitutes compared to the wounds treated with the non-vascularized control substitutes (Fig. 5a,c,e). Histological examinations revealed that, as early as 7 days after grafting, the wounds treated with the pre-vascularized substitutes contained blood-perfused vessels (Fig. 5d), whereas the wounds treated with non-vascularized control and artificial dermal substitutes did not contain perfused vessels. In addition, some ingrown blood vessels of host origin were observed in the wound borders at day 7 (Fig. 5b,f). The vessel densities in the wounds were assessed morphometrically after immunohistochemical staining for murine and human CD31. In both control and pre-vascularized substitutes, mouse-derived CD31 + vessels were localized at the border of substitutes at 7 days after grafting and distributed across the wounds at 14 days after grafting, indicating angiogenesis into the 3D skin substitutes. In contrast, in the pre-vascularized substitutes, engineered human-derived CD31 + vessels were consistently observed overall the substitutes during experiments (Fig. 6a). The vessel density was significantly higher in the pre-vascularized substitute-treated www.nature.com/scientificreports www.nature.com/scientificreports/ wounds than in the control-treated wounds (p < 0.01, Fig. 6b). Furthermore, immunofluorescence staining with human-and mouse-specific CD31 antibodies indicated the vessels formed by both engineered human-(green) and host-derived (red) endothelium at 14 days after grafting in the pre-vascularized substitutes (Fig. 6c).  www.nature.com/scientificreports www.nature.com/scientificreports/ Visual observation in our in vivo grafting model showed that the control substitute-grafted sites exhibited exsiccation and detached from the wound site by day 14. Although 11 of 15 mice treated with non-vascularized control substitutes exhibited epidermolysis by day 7 after grafting ( Fig. 5b), all control-treated wounds were completely re-epithelialized in mice at day 14 ( Fig. 7b). In contrast, the pre-vascularized substitutes adhered to the wound tissue, which led to good healing with minimal wound contraction ( Fig. 7a,b). Masson's trichrome staining revealed mature collagen fiber (bright blue and densely packed) were observed in pre-vascularized substitutes, and the collagen level was similar to the host tissues. In contrast, the wounds of control-substitutes showed immature collagen (light blue and loosely packed) surrounded the fibroblasts (Fig. 7c). Moreover, we confirmed that the wound engraftment in mice was derived from the engineered human 3D skin substitutes by staining with human histocompatibility complex HLA-ABC. The engineered human vessels and Malpighian layer of pre-vascularized treated wound expressed uniformly HLA-ABC, indicating that the vasculature and epidermis are of human origin. In contrast, HLA-ABC positive cells in the control-treated wound were distributed on the dermis (Fig. 7d-f). Most importantly, the dermal and epidermal components derived from the grafted pre-vascularized substitutes survived 14 days post-transplantation, although the majority of control grafts were sloughed and immature.

Discussion
In this study, we demonstrated that scaffold-free 3D skin substitutes with blood vessel networks were functional and reliable upon transplantation onto mouse full-thickness wounds. We successfully developed the novel skin substitutes using an LbL cell coating technique, which formed thin FN-G films on the cell surface. Our bioengineering approach made it possible to generate the pre-vascularized LbL 3D skin substitutes via approximately 2-week organotypic cultivation. This study is the first to demonstrate the potential application of scaffold-free pre-vascularized 3D skin substitutes for full-thickness skin defects.
Early excision of the wound eschar and subsequent coverage with skin substitutes are widely accepted as the primary goals in extensive burn treatment 19 . Bioengineered skin substitutes act first as protective dressings to limit infection and fluid loss, and further function as skin equivalents to provide temporary wound coverage or permanent skin replacement. Recently, various types of skin substitutes and techniques have become available to aid in wound coverage and healing 20 . However, even with increased research, the clinical www.nature.com/scientificreports www.nature.com/scientificreports/ application of tissue-engineered skin remains rare. One of the main reasons is the difficulty in vascularizing the tissue-engineered skin. Fibroblast-derived ECM components are essential for vascular formation and cell migration, adhesion, differentiation, and proliferation 21 . Recently, vascularization of skin constructs has been studied by seeding and cultivating vessel-forming cells on scaffolds 6,9,22,23 . FN-G coating of the cell surface by LbL assembly, which promotes cell-to-cell interaction and homogeneous stacking, is critical to the fabrication of the scaffold-free 3D-tissues 12 . In addition, coculturing the endothelial cells with FN-G-coated fibroblasts may enhance the formation of vasculature within a specific 3D tissue. In our study, vascularization was achieved in a short-term coculture system comprising endothelial cells and supportive FN-G-coated fibroblasts in the absence of exogenous materials such as scaffolds or growth factors. We feel that our unique approach has advantages over previous strategies for the generation of a human vessel network within a dermal layer.
In general, cultured human keratinocytes appear to promote optimal wound re-epithelialization and stimulate the regeneration of a compact neodermal component 24 . As an alternative to autologous split-thickness skin grafting, grafting of tissue-engineered skin substitutes composed of dermis and epidermis has been developed; the tissue-engineered skin substitutes act as a barrier, dermal matrix, and well-vascularized wound bed, and promote healing 25 . However, a lack of initial vascularization after grafting diminishes successful engraftment. Indeed, we clearly observed blood vessels only at the graft periphery, with limited numbers in the grafts, of non-vascularized 3D skin substitutes 7 days after grafting. In contrast, most of the pre-vascularized constructs were fully perfused 7 days after grafting. In addition, human vessel networks connect to host vasculature via the ingrowth of host vessels into the 3D skin substitutes at 14 days after grafting. The rapid perfusion of the grafted constructs by host blood in such a short time will likely alleviate tissue hypoxia and promote cell survival. As expected, we detected HLA-ABC + cells in the grafted dermal and epidermal layers of the pre-vascularized substitutes, which indicated that the pre-vascularized 3D skin substitutes had a high graft survival rate. Shrinkage in in vitro human skin models is a critical drawback because it induces tissue necrosis, restricted graft survival, and faster degradation 26 . Indeed, a well-vascularized wound bed is essential for the survival of the skin graft and graft contraction. In this study, rapid perfusion also induced increases in collagen deposition and dermal thickness; we hardly observed graft contraction, which indicated that our pre-vascularized 3D skin substitutes are promising tools to promote the healing of full-thickness skin defects.
Although engrafted non-vascularized control substitutes have abundant fibroblasts, which secrete paracrine factors 27,28 , they undergo little neovascularization, resulting in a poor wound bed. In addition, epidermolysis occurred in the majority of non-vascularized substitutes at 7 days post-transplant, although they promoted complete re-epithelialization by 14 days. By contrast, we did not observe epidermal sloughing of the pre-vascularized www.nature.com/scientificreports www.nature.com/scientificreports/ 3D skin substitutes. Moreover, the wounds of control-substitutes showed that poor collagen deposition due to immature or inactive fibroblasts highly expressing HLA-ABC antigens, which may have been caused by an insufficient blood supply. The significantly higher vessel density in pre-vascularized substitutes at the early phase was www.nature.com/scientificreports www.nature.com/scientificreports/ associated with an enhanced maintenance, viability, and functional engraftment of 3D skin substitutes. Thus, our findings suggested that a preformed vasculature is a prerequisite for successful grafting.
Integration of the blood vasculature into tissue-engineered constructs can be crucial for the long-term viability and survival of the constructs in vivo 29 . In our strategy, the vascular network within the dermal layer was formed with a high proportion to HUVEC in the coculture system. However, endothelial cells were seeded at the lower density for the fabrication of the transplantable 3D skin because the grafts with a high number of HUVEC developed large subcutaneous and dermal hematomas. Hematomas can be a source of infection, so skin substitutes must be designed to prevent wound hematomas. Although the rapidity of graft vascularization is important for protection against infection to facilitate effective skin graft integration, future studies must address other factors that affect the potential of our approach for enhancing tissue survival and resistance against wound infection.
In summary, we demonstrated the potential of scaffold-free pre-vascularized LbL 3D skin as an implantable tissue substitute. The tissue constructs comprising endothelial cells and a high density of FN-G-coated fibroblasts promoted the development of vessel networks in vitro, accelerated the rate of anastomosis with host vessels, and aided healing in vivo. Our strategy to transplant a dermo-epidermal skin substitute containing blood vessels may facilitate the efficient reconstruction of full-thickness skin defects.
Preparation of vascularized 3D skin substitutes. The 3D skin containing the vascular network was constructed via an LbL technique that enabled the coating of individual cells with FN-G ECM films on NHDF surfaces without damage. Briefly, NHDF were alternately incubated for 1 min with 0.04 mg/mL fibronectin (FN) or gelatin (G) solutions in phosphate-buffered saline (PBS) and rinsed with PBS between each coating step. The coating procedures were performed a total of 9 times (FN: 5 times; G: 4 times). Then, to construct the dermal layer, FN-G-coated NHDF (viability >97%, 1 × 10 7 cells/insert) and HUVEC (0.1 × 10 5 , 0.2 × 10 5 , or 1 × 10 5 cells/insert) or FN-G-coated NHDF alone were seeded onto 12-well culture inserts (1.12-cm 2 membrane growth area, 0.4-μm pores; Corning Life Science, Tewksbury, MA, USA) pre-coated with FN; the inserts were cultured in DMEM containing 5% FBS for 24 hours. In other experiments, GFP-HUVEC were seeded to facilitate the monitoring of vessel formation during the cultures. After 24 hours coculture of FN-G-coated NHDF and HUVEC, type IV collagen solution (from human placenta, 0.2 mg/mL in PBS; Sigma-Aldrich, St. Louis, MO, USA) was dropped onto the upper surface of the dermis, and the constructs were incubated for 30 min. Subsequently, HEKn (passage 4, 1 × 10 6 cells/insert) were seeded onto the collagen type IV-coated dermal layer, then cultured for 24 hours in growth medium (5% FBS/DMEM and EpiLife ® at a ratio of 1:1), and air-liquid interface-cultured for 7 days to drive epidermal differentiation, stratification, and cornification in growth media supplemented with 25 μg/ mL ascorbic acid. The medium in the lower chamber was replaced every day; flooding of the top chamber was avoided. For in vivo experiments, we seeded 0.2 × 10 5 HUVEC per construct.
Cell viability assay in 3D skin substitutes. We determined cell viability in the 3D skin substitutes using the LIVE/DEAD Viability/Cytotoxicity kit (L-3224; Molecular Probes, Eugene, OR, USA) according to the manufacturer's protocol. Briefly, at 2, 4, and 7 days after organotypic coculture, we rinsed the 3D skin substitutes (n = 3 for each time point) 3 times with PBS, then stained with 2 μM calcein AM (which stained live cells green) and 0.5 μM ethidium homodimer (which stained dead cells red) in PBS for 30 min at room temperature. After incubation, the samples were washed twice with PBS and imaged immediately using a Biozero BZ-X700 fluorescence microscope (Keyence, Osaka, Japan).
We acquired 3 images of different portions of each triplicate sample for each time point. Still images of the middle section of the 3D skin substitute (~100 to 120 µm from the culture insert membrane) were captured and serial micrographs were later combined for z-stacked compilation images. The experiment was performed twice.
Transplantation of tissue-engineered 3D skin substitutes. All surgical procedures were conducted according to protocols approved by the National Defense Medical College Animal Care and Use Committee (Permit number: 17012). Pre-vascularized 3D skin substitutes were transplanted to excisional wounds in C.B-17 SCID mice. Eight-week-old male mice (body weight, 24.2 ± 3.7 g; Charles River Laboratories, Kanagawa, Japan) were randomly divided into 3 groups (n = 15 animals/group). The excisional wound splinting model was employed as previously described 30 . In brief, mice were anesthetized with an intraperitoneal injection of ketamine (100 mg/kg) and xylazine (10 mg/kg), then their dorsal surfaces were shaved with electric clippers and treated with a depilatory agent to remove remaining hair. Under sterile conditions, a circular full-thickness excisional wound (8-mm diameter) was created on the middle dorsum of each mouse using a biopsy punch. A donut-shaped silicone splint (0.5-mm thick, 12-mm inner diameter, 20-mm outer diameter) was placed around the perimeter of the wound and fixed to the skin with cyanoacrylate glue (Aron Alpha, Toagosei, Tokyo, Japan) and interrupted 6-0 nylon sutures (Fig. 1). The splint was used to minimize wound contraction. The skin substitutes were cut according to the dimensions of the full-thickness wound, and each wound was treated with a different substitute: www.nature.com/scientificreports www.nature.com/scientificreports/ pre-vascularized 3D skin, non-vascularized 3D skin (negative control), or acellular synthetic bilayer skin, which is a collagen-glycosaminoglycan sponge with a temporary silicone rubber epidermal barrier. The grafts were covered with a non-adherent silicone gel dressing (SI-AID, ALCARE, Tokyo, Japan) and wrapped with a semipermeable adhesive film (OPSITE, Smith & Nephew, Largo, FL, USA) to protect the wound site and prevent desiccation. The animals were housed individually. After 7 (n = 7/group) or 14 (n = 8/group) days, the mice were euthanized and tissue biopsies of the grafted areas were collected. Transillumination stereomicroscopy. To visualize the vascular network by transillumination, the implants and surrounding skin were excised at day 7 after grafting; specimens were then quickly placed on a Petri dish and observed under an inverted microscope.
Histology and immunostaining. To observe the vessel network formed by HUVEC in vitro after 7 days of culture, we stained whole-mount 3D skin substitutes with a fluorescently labeled anti-CD31 antibody. The constructs (n = 3) were fixed in 4% paraformaldehyde for 15 min, then permeabilized in PBS with 0.2% Triton ™ X-100 for 15 min. The tissues were blocked with 1% bovine serum albumin for 1 h, then incubated with a mouse anti-human CD31 antibody (1:100; M0823, clone JC70A; Dako, Carpinteria, CA, USA). The tissues were incubated with Alexa Fluor ® 546-conjugated goat anti-mouse IgG (1:500; Invitrogen) for 1 h. Immunofluorescence images of the stained vessels were taken with a BZ-X700 microscope. Vascular profiles were characterized by positive staining for human CD31 in structures with an identifiable vascular lumen (>5 μm). The vascular area density and number of branching points were analyzed with AngioTool software (National Cancer Institute, Rockville, MD, USA). All analyses were carried out manually.
The grafted sites were fixed in 10% neutral buffered formalin overnight, then dehydrated and embedded in paraffin. Hematoxylin and eosin, trichrome, and immunohistochemical/immunofluorescent staining were performed on 4-μm-thick tissue sections. Trichrome staining was carried out using the Modified Gomori's One-Step Trichrome Staining Kit (Biocare Medical, Concord, CA, USA) and immunohistochemical staining was performed with the Vector ImmPRESS ™ peroxidase polymer Anti-mouse or Anti-rabbit IgG reagent kits Statistical analysis. The data are expressed as the mean ± standard deviation (SD). Statistical analyses were performed by unpaired Student's t-test or one-way analysis of variance (ANOVA) with Tukey's multiple comparison post hoc test using GraphPad Prism 7.0 software (La Jolla, CA, USA). P-values < 0.05 were considered significant.