In this preclinical study, the authors manufactured batches of Poieskin^® under GMP-compatible conditions using cells harvested from patient biopsies. Human skin biopsies for Poieskin^® production and human split-thickness skin graft (HSTSG) controls were surgical residues removed during plastic surgeries on healthy volunteer patients (BIOPSKIN study number NCT04925323).

All animal experiments were conducted in accordance with the Aix-Marseille University Institutional Animal Care and Use Committee (CE14, Aix-Marseille University) and the French Ministry of Research (project authorization #33645 on 29 November 2021), according to the European Union directive 2010/63/EU and the recommendations of the Declaration of Helsinki. The study was approved by the National Animal Care and Ethics Committee (no. APAFIS 2020012015402650). Trained and authorized operators performed all animal experiments.

Poieskin^® wound healing efficacy was compared to HSTSG as a control group. Tissue was harvested from healthy donor patients during mammaplasty procedures (after obtaining written informed consent) using an electrical dermatome (Acculan 3Ti^® Dermatome, Aesculap, Hazelwood, MO, United States) set to harvest 0.2–0.3-mm-thick skin, according to the STSG defined thickness range. HSTSG was collected 1 hour before the grafting procedure and was transferred in a sterile container with 0.9% NaCl solution to the center for in vivo transplantation.

Human dermal fibroblasts and epidermal keratinocytes were extracted from a skin biopsy. We used two 4-cm² skin biopsies to produce two 40-cm² skin substitutes. Keratinocytes and fibroblasts were isolated from human donor skin. The hypodermis compartment was manually deleted, and the skin biopsy was cut into small pieces. The epidermis and dermis were separated after overnight incubation at 4°C in 2.4 U/mL of dispase. Collagenase NB5 was used at a concentration of 0.25 U/mL for fibroblast isolation and trypsine 0.05% for keratinocyte separation.

The viability and concentration of keratinocytes and fibroblasts in viable nucleated cells were checked after each trypsination using an automated cell counter Luna X7 (Logos Biosystems, Villeneuve-d’Ascq, France).

After primary seeding, keratinocytes and fibroblasts were detached using trypsin‐EDTA and cryopreserved in 10% DMSO + 90% FBS for storage at −150°C. For tissue production, keratinocytes and fibroblasts were thawed and further cultured and amplified ex vivo.

Keratinocytes were grown on an allogeneic feeder layer of irradiated human fibroblasts (WCB-3C, AP-HM, Marseille, France) and cultured in a keratinocyte medium containing 58% Dulbecco–Vogt modified Eagle’s medium, 30% Ham’s F12, 2 mM GlutaMAX, 10% bovine FetalClone II serum, 10 ng/mL human epidermal growth factor, and 20 μg/mL gentamicin.

Fibroblasts were grown in fibroblast media containing alphaMEM, 5% platelet lysate, 2 UI/mL heparin, 100 U/mL penicillin, and 20 μg/mL gentamicin.

In preparation for dermis bioprinting, human dermal fibroblasts (passage 3) were suspended in DPBS to obtain a bioink containing 12.5 million viable nucleated cells/mL for LAB. Meanwhile, bovine collagen ink was formulated at 4 mg/mL using 10X DPBS, water for injection (WFI), and NaOH for extrusion bioprinting at 4°C. Dermal architecture consisted of depositing three layers of collagen and fibroblasts by extrusion and LAB, respectively.

After bioprinting, the dermis was cultured for 5 days in dermis maturation media containing alphaMEM supplemented with 3% platelet lysate, 2 UI/mL heparin, 100 U/mL penicillin, 20 μg/mL gentamicin, and 50 μg/mL laroscorbine.

Keratinocytes (passage 2) were suspended in DMEM to obtain a bioink containing 70 million viable nucleated cells/mL for LAB.

After bioprinting keratinocytes onto the bioprinted dermis, the skin constructs were cultured under immersed conditions for 2 days in a dedicated medium containing 58% Dulbecco–Vogt modified Eagle’s medium, 30% Ham’s F12 , 3% platelet lysate, 10 ng/mL human epidermal growth factor, 100 U/mL penicillin, 20 μg/mL gentamicin, and 50 μg/mL laroscorbine.

Cells were expanded before being formulated into bio-inks and bioprinted using an NGB bioprinter, in a GMP-compatible process, which has been defined by the use of GMP reagents and culture media for all the steps of the Poieskin^® manufacturing process and the conformity to the acceptance criteria on the final Poieskin^® product (size = 40 cm² ± 10%; dermal thickness >200 μm; histological score >50; and sterility of culture media including the specific absence of mycoplasms). Two 40-cm² DESs were produced through a multistep protocol (Figure 1A). First, the multilayer dermal compartment was sequentially obtained by printing human primary fibroblast patterns on a bovine collagen (SYMATESE, Chaponost, France) layer printed by bioextrusion (Figure 1B). Second, the epidermal compartment was obtained by printing human primary keratinocyte patterns on the dermal equivalent (Figure 1C). Then, both 40-cm² DESs were matured at the air–liquid interface (ALI) for the final epidermal differentiation for 6 days with a culture medium containing 58% Dulbecco–Vogt modified Eagle’s medium, 30% Ham’s F12, 3% platelet lysate supplemented with 0.8% human serum albumin, 0.4 μg/mL hydrocortisone, 0.12 UI/mL insulin, 100 U/mL penicillin, and 20 μg/mL gentamicin. We prepared a 1.6X ALI medium and then mixed it with agarose to obtain a final medium of 1X ALI medium and agarose 0.75%. We allowed it to gelify and then placed the Poieskin on top of it (it does not sink in the gelified medium) so that the air–liquid interface is maintained.

On day 6 of the ALI, the DESs were placed on gelified ALI media at room temperature for 6 h and transported from the manufacturing facilities (Poietis, Pessac, France) to the animal facilities (CERIMED, Marseille, France) for grafting. Before grafting, the DESs were cultured at the ALI with fresh ALI media for 24 h. Throughout fabrication, in-process characterizations were performed. Eight areas of 1.5 cm² were selected for the grafting procedure, and eight additional areas of the same dimensions were used for control tests (Figure 1D).

Mechanical behavior was assessed by a senior plastic surgeon whose conformity was defined to be robust enough not to break when packaged, manipulated, and applied to the surgical site.

Control samples from 40-cm² Poieskin^® to be grafted were analyzed to characterize the production features, as well as sections of the wound samples 16 days after grafting. Thickness measurements were made on the histology images using ImageJ software (National Institute of Health, Bethesda, United States).

The quality of the DES was assessed from histological sections after staining with Masson’s trichrome (described in Section 2.11). POIETIS developed a scoring grid to evaluate dermal and epidermal structures, excluding skin appendages, in order to quantitatively assess the quality of the evaluated samples. The semiquantitative score and associated parameters are presented in Supplementary Table S1.

The study was designed to demonstrate the non-inferiority of Poieskin^® compared to the reference HSTSG in a relevant animal model (Figure 2). Immunosuppressed female NMRI-Foxn1^nu/nu mice (aged 7 weeks and weighing 20–23 g) were purchased from Janvier Laboratories (Genest-Saint-Isle, France). They were allowed to acclimatize for 2 weeks before the experiments. Eight mice received a Poieskin^® graft, while eight others received HSTSG on their back immediately after a cutaneous excision of 1 × 1.5 cm. The primary endpoint was the area measurement of the graft taken on standardized photographs in a 16-day follow-up. Moreover, the following secondary endpoints were evaluated: area percentage of wound contraction, superficial vascularization assessed by laser Doppler imaging, local inflammation by PET imaging, and histological analysis after euthanasia.

The initiation of the surgical procedure in mice was synchronized to the end of Poieskin^® maturation and surgical harvesting of HSTSG, in order to perform both treatments on the same day, without delay, under comparable conditions. The mice received a subcutaneous injection of buprenorphine at a dose of 0.1 mg/kg 30 min before general anesthesia by sevoflurane initiation in an induction chamber (8%; 1 L/min). Anesthesia was maintained via a nose cone (2.5%–3%; 0.5 L/min) with spontaneous ventilation on an electric heater carpet at 37°C. Protective collyrium was then administered. The procedure began by cleaning with iodine after marking a 1 × 1.5 cm longitudinal rectangular cutaneous area to excise on the back of the mice. Before incision, the mice received long-term local analgesia by a subcutaneous back-forward injection of ropivacaine at 4 mg/kg around the surgical site. Then, surgical ×2.5 magnifying loupes were used in the surgical procedures.

The surgical protocol consisted of three steps: i) realization of an acute full-thickness longitudinal 1 × 1.5 cm excision using a scalpel no. 15. Any continuous bleeding was coagulated using bipolar forceps. Scarring was followed by immediate covering (with Poieskin^® or HSTSG). Grafts were custom-sized according to the defect (1.5 cm²) from larger samples, using a scalpel no. 23, withdrawn from a 40-cm² Poieskin^® specimen, or from a fresh human split-thickness skin sample for the HSTSG group. They were manipulated using microsurgical tweezers (without teeth); ii) application of an immediate graft on the skeletal muscles by plan and fixation surgical methods to the skin with 14 separate Prolene^® 6–0 (Ethicon Inc., Johnson & Johnson, New Brunswick, NJ, United States) stitches using a microsurgical holder and scissors; iii) application of custom-sized dressing on the graft of a non-adherent, non-absorbent UrgoTul^® interface (Laboratoires URGO, Chenôve, France) and then a non-woven surgical pad was fixed using an adhesive Tegaderm^® bandage (3 M, Saint Paul, MN, United States). We checked before awakening that the dressing did not induce any hindrance to chest expansion, limb mobility, and fecal or urinary tract.

During postoperative surveillance, the mice were placed under a heating lamp until awakening and then placed into an individually ventilated cage in a loose housing room. The general health condition was monitored daily. Analgesic subcutaneous injections of buprenorphine (0.1 mg/kg/day for the first week and then 0.05 mg/kg/day) could be administered, if necessary. No antibiotic was administered. Dressings were first changed on day 2 and then on days 5, 8, 10, 12, and 15 under general anesthesia.

Standardized photographs were taken preoperatively (day 0) on days 2, 5, 8, and 12 and finally on day 16 immediately before the mice were euthanized. The camera (Canon PowerShot SX220 HS, Tokyo, Japan) was placed vertically 35 cm above the operating site, and the view was zoomed to ×2, taking a centimeter ruler in the range in order to harmonize the area measurement performed using ImageJ software (National Institute of Health, Bethesda, United States). We analyzed the evolution of engraftment and retraction ratios. The engraftment ratio is defined as the vascularized epithelialized graft area divided by the total scarring area at the same time. On the other hand, the retraction ratio is calculated as the total scarring area at a specific moment divided by the total scarring area on day 0.

Laser Doppler perfusion imaging (PIM2, Perimed, Craponne, France) was used to assess skin graft perfusion after surgery on days 2, 6, 12, and 16 post-surgery under sevoflurane anesthesia. The mice were induced with 8% sevoflurane in air, followed by 3% sevoflurane in air, and placed on a 37°C heated carpet. The results were expressed as an optical density ratio of the graft area to a healthy back skin area.

The mice were injected with 6.4 ± 1.5 MBq/100 µL [18 F]-FDG in the caudal vein on days 5 and 15 after surgery. MicroPET/CT images were acquired 60 min after injection during a 20-min imaging session. MicroPET/CT imaging was performed using a nanoScan PET/CT camera (Mediso, Budapest, Hungary) under isoflurane anesthesia. The mice were induced with 4% isoflurane in air, followed by 1.5% isoflurane in air (Isovet from Piramal, Voorschoten, Netherlands). Quantitative region-of-interest (ROI) analysis of the PET signal was performed on attenuation- and decay-corrected PET images using VivoQuant software v4.0patch1 (Invicro, Boston, MA, United States). Tissue standardized uptake values were expressed as the ratio of the graft area to the healthy back skin area.

For both techniques (laser Doppler and [18F]-FDG MicroPET/CT), the screened areas were divided into several ROIs. Unaffected skin on the neck and lumbar area served as controls to assess vascularization and local inflammation evolution of the grafts.

On day 0, control DESs were fixed with 4% (w/v) formaldehyde for 8 h at room temperature. At the end of the 16-day follow-up, the mice were euthanized with a lethal overdose of sevoflurane inhaled into an induction chamber associated with cervical dislocation. The surgical sites were then entirely harvested, taking laterally 3-mm-width margins and vertically from the underlying musculoskeletal system to the cutaneous envelope, in order to collect an exhaustive sample for histological analysis. Sections of the wound samples were immediately fixed with 4% (w/v) formaldehyde at room temperature for 48 h and maintained at 4°C for 72 h.

Fixed samples were dehydrated in successive ethanol and xylene solutions prior to paraffin embedding using a Histokinette (Leica, Wetzlar, Germany). Sections of 5 μm were generated using a microtome (Leica) and then stained with Masson’s trichrome dyes before imaging by light microscopy (Eclipse Ts2, Nikon, Tokyo, Japan). The histological characteristics of all grafts, Poieskin^® and HSTSG, were analyzed by an independent skilled pathologist who was blinded to the samples, using NDP2 viewing software (Hamamatsu Photonics, Hamamatsu, Japan) and a semi-quantitative scoring system as previously described.

Previously processed slides were permeabilized with 1X citrate buffer at pH 6 for 20 min at 98°C. Non-specific interactions were limited by incubation with a 2% (w/v) BSA-DPBS solution (bovine serum albumin, Merck, Burlington, Massachusetts, United States; Dulbecco’s phosphate-buffered saline, Dutscher) for 10 min. The samples were then incubated overnight at 4°C with primary antibodies, namely, K5 (1/200, Abcam, AB52635), K10 (1/200, Merck Millipore MAB3230), Coll1 (1/500, Abcam, Ab138492) (specific for human and bovine collagen 1 and does not stain mouse collagen 1), αSMA (1/200, Abcam, Ab5694) (Prost-Squarcioni et al., 2008) (Ruiter et al., 1993), and loricrin (1/200, Abcam, Ab85679) antibodies (Matsui and Amagai, 2015). After washing, secondary goat anti-rabbit Alexa Fluor 488 (1/200, Thermo Fisher Scientific, 10729174), secondary goat anti-rabbit Alexa Fluor 594 (1/200, Molecular Probes, R10477), and goat anti-mouse Alexa Fluor 488 (1/200, Thermo Fisher Scientific, 10256302) antibodies were added for 1 h at room temperature. The samples were counterstained with DAPI (1/1000, PK-CA707-40043, PromoCell, Heidelberg, Germany). Immunofluorescence images were captured using the Eclipse Ts2 microscope (Nikon) and analyzed using ImageJ software (National Institute of Health, Bethesda, United States).

The protein structure of the samples was analyzed by immunohistochemistry. Bovine type I collagen, which was used as a biomaterial in Poieskin^®, allowed for the identification of the location of Poieskin^® and HSTSG. Cytokeratin 5 staining highlighted basal- and spinous-layer keratinocytes of human origin. The association of K10, K5, and loricrin staining revealed the epidermal maturation gradient, as they represent the basal, supra-basal, and corneal layers, respectively. αSMA was used to identify myofibroblasts or pericytes that could reveal neovascularization with a circular shape surrounding a lumen.

Statistical analysis was performed using GraphPad Prism 4.0 (GraphPad Software, La Jolla, CA, United States). The data that support the findings of this study are available on request from the corresponding author. The normality of data distributions was assessed using Shapiro–Wilk and Kolmogorov–Smirnov tests, and all data passed the normality test. Two-way ANOVA statistical tests were used for the analysis with multiple factor interactions (FDG-CT, graft take, and graft retraction). The engraftment ratio was defined as the vascularized epithelialized graft area divided by the total scarring area at the same time, whereas the retraction ratio corresponded to the total scarring area during evaluation divided by the total scarring area on day 0. A mixed-effect analysis was used for the analysis of laser Doppler imaging because values for performing two-way ANOVA were missing. Ultimately, ordinary one-way ANOVA was used for comparing the histological scorings. Graphs display the mean ± standard deviation (SD).