A gelatin/collagen/polycaprolactone scaffold for skin regeneration

Preparation of G∕C∕P and C/P biocomposites

The preparation of the G∕C∕P and C/P biocomposites was listed in Table 1. Aliquots of type B gelatin (from bovine skin, Sigma-Aldrich, St Louis, MO, USA) with specific concentration of collagen (Sigma-Aldrich, St Louis, MO, USA) solution was prepared by dissolving gelatin in double distilled water followed by mixing collagen (dissolved in acetic acid) to produce two ratios of collagen in gelatin/collagen biocomposites respectively. The dissolution of gelatin/collagen was facilitated by stirring using a heating magnetic stirrer in a 25 ml glass shell vial at the temperature of 40 °C. After complete dissolution, aliquots (0.25 ml) of the gelatin/collagen solution were added to 7 ml glass vials followed by elimination of air bubbles and frozen at −20 °C for approximately 45–50 min. In the second stage, samples were transferred to a freezer at −72 °C for 35 min. The frozen samples were placed in a freeze dryer (Edwards Modulyo^®) at −44 °C under 42 mbar vacuum for 24 h to prepare for the G/C mats. Aliquots (0.5 ml) of PCL (Mw=115000; Solvay Interox, Warrington, UK) /dichloromethane (DCM with HPLC grade; Fisher Scientific, Loughborough, UK) solution were added carefully to the freeze dried G/C mats with low collagen content to produce GC_LP_L(G/C:PCL is 1:8) and GC_LP_H (G/C:PCL is 1:20) biocomposites respectively, whereas aliquots (0.5ml) of PCL/dichloromethane (DCM) solution were added to the freeze dried gelatin/ collagen mats with high collagen content to produce GC_HP_L(G/C:PCL is 1:8) and GC_HP_H (G/C:PCL is 1:20) biocomposites respectively. The vials were kept for 30 min and then removed the lids to allow solvent evaporation overnight.

On the other hand, C/P biocomposite was fabricated described by previous study (Hajiali et al., 2011). Briefly, collagen solution was prepared in 1% acetic acid by stirring with a magnetic stirrer overnight at room temperature. Aliquots (0.25 ml) of the collagen solution were added to 7 ml glass shell vials (Fisher Scientific) and frozen at –20 °C for approximately 45–50 min. Samples were transferred to a freezer at –70 °C for 35 min before freeze drying. Aliquots (0.5 ml) of a solution of PCL in DCM were added carefully to the freeze dried collagen mats to prepare 1:8, 1:20 collagen:PCL materials, respectively. The vials were kept stopped for 30 min before removing the lids to allow solvent evaporation overnight. The component ratios of the four G∕C∕P and C/P biocomposites are shown in Table 1.

Protein release from G∕C∕P biocomposites

To determine the release of gelatin and collagen from biocomposites (n = 3), the bicinchoninic acid (BCA) assay was used to estimate the amount of protein release from G∕C∕P biocomposites after incubation in phosphate-buffered saline (PBS) at 37 °C for 12 days. Individual samples were added to 7 ml glass shell vials containing 1 ml PBS and then incubated at 37 °C in a water bath. The release media was replaced completely by fresh PBS periodically and analysed for total protein content using the BCA assay. The absorbance of the calibration samples measured at 562 nm was used to produce a calibration curve used to calculate the protein (gelatin/collagen) concentration of the test samples.

Thermal analysis by differential scanning calorimetry (DSC)

The thermal characteristics of GC_LP_L, GC_LP_H, GC_HP_L, and GC_HP_H, biocomposites with weight between 2–10 mg were recorded using a Perkin-Elmer Pyris Diamond differential scanning calorimeter. All the samples were lightly pressed into the bottom of the pan to ensure good thermal contact. Sealed DSC pans were used in the study. Triplicate samples were heated at a rate of 10 °C/min from 10 °C to 100 °C. Peak melting temperature (Tm) and heat of fusion data for the PCL component of the materials were determined using the built-in software of the DSC. The latter measurement was subsequently used to estimate the percentage crystallinity of PCL in the composites from the reported heat of fusion of 139.5 J/g for fully crystalline PCL (Burke, 1987). Indium was used as a standard.

The tensile strength analysis by a universal testing machine

The mechanical analysis was conducted on GCP and C/P biocomposites using a universal testing machine (Instron) under axial loading. The biocomposite specimens were cut as the suitable size of rectangular block and were carefully clamped at the center of the cross-head with its end faces exactly perpendicular to the longitudinal axis. The crosshead speed of 50 mm/min was applied for this test. The tensile strength (MPa) was calculated as the force at failure divided by the cross-sectional area of the biocomposite. Results were the average from 3–6 measurements and analyzed by one-way analysis of variance and the t-test (P < 0.05).

Observation of porous structure by scanning electron microscopy (SEM)

The scanning electron microscope was used to observe the surface morphology of G∕C∕P biocomposites. Samples were attached to aluminum SEM stubs using carbon tabs (Agar Scientific). Specimens were sputter coated with gold prior to examination using a HITACHI^®S-3000N scanning electron microscope.

Cell culture on G∕C∕P

Primary human epidermal keratinocytes (PHEK) and primary human dermal fibroblasts (PHDF) were isolated and primarily cultured from the donated human foreskin samples after the surgery of circumcision in the study, which were approved by the institutional review board (IRB). The study protocol was reviewed and approved by the Institutional Review Board (IRB) in the Tri-Service General Hospital, R.O.C. (TSGHIRB No.: 100-05-251). Then the written informed consent was obtained from each donor. EpiLife^® HKM (Cat No.M-EPIcf-500 with addition of 0.06 mM calcium chloride and HKGS kit: Cat No.S-001-5) was used for keratinocyte culture. The cell isolation from skin samples, cell expansion, and cell counting followed those described previously (Burke, 1984).

PHEK (child foreskin; P4) were seeded on the top surface of G∕C∕P (1:8 and 1:20) and C/P biocomposites and 24-well tissue culture plastics (TCP) at a cell density of 1.7 × 10⁵ or cells per cm² for up to 9 days, PHDF (adult foreskin; P3-4) were seeded on the same materials at a cell density of 2.0 × 10⁴ cells per cm² for up to 10 days. Human adipose-derived stem cells (ASCs; passages 3–4) were seeded on the biocomposites and 24-well TCP at a cell density of 2.0 × 10⁴ cells per cm² for up to 12 days. Trypsin-EDTA solution was used for cell detachment followed by cell counting at 1, 3, 6, and 9 days for PHEK and cell at 1, 4, 7, and 10 days for PHDF using a Weber’s haemocytometer. The experiments were repeated at least twice with similar results.

The adhesion, growth, and distribution of cells seeded on G∕C∕P biocomposites were investigated using the immunohistochemical assay. PHEK (child foreskin; P4) were seeded on the top surface of G∕C∕P biocomposites (1:8 and 1:20), C/P biocomposites and 24-well tissue culture plastics (TCP) at a cell density of 1.7 × 10⁵ cells per cm² for time intervals up to 9 days. PHEK on G∕C∕P biocomposites (1:8 and 1:20) at 1 and 3 days were labeled with the (primary) monoclonal rabbit anti-human cytokeratin (CK)-14 1:100 (Bioworld Technology), and the (secondary) rhodamine-conjugated polyclonal goat anti-rabbit IgG 1:200 (Chemicon, Billerica, Mass) examined under a fluorescent microscope.

PHDF (adult foreskin; P3-4) were seeded on the top surface of 1:8 and 1:20 GC10, GC25 and collagen:PCL biocomposites and TCP (24-well tissue culture plastics) at a cell density of 2.0 × 10⁴ cells per cm² for time intervals up to 10 days. PHDF on G∕C∕P biocomposites (1:8 and 1:20) at 1 and 3 days were labeled with the (primary) monoclonal mouse anti-human α-tubulin antibody 1:200 (Cat No.sc-5286; Santa Cruz Biotechnology, Inc.), and the (secondary) fluorescein (FITC)-conjugated polyclonal goat anti-mouse IgG 1:100 (Lot No.62686; Jackson ImmunoResearch Laboratories, Inc., Baltimore, MD, USA) and examined under the fluorescent microscope.

Preparation of human adipose-derived stem cells (ASCs)-seeded G∕C∕P biocomposites for animal experiments

Human adipose tissues were obtained using lipoaspirate after syringe-assisted liposuction from abdomens of adult Taiwanese female patients who had no systemic metabolic diseases or lipid disorders. The procedures and protocols were conducted at the Tri-Service General Hospital, Taipei, Taiwan, and were approved by IRB.

The isolated fat tissues were washed using PBS. After washing, the tissues were digested with an equal volume of PBS including 0.075% type I collagenase (Sigma-Aldrich Company Ltd, Poole, UK) at 37 °C for 1 h. After centrifugation at 2,500 rpm for 10 min, the cell pellet was resuspended in Dulbecco modified Eagle medium-low glucose (DMEM-LG; Invitrogen) containing 10% fetal bovine serum (FBS; Invitrogen) and 1% penicillin-streptomycin (Invitrogen), and placed in the incubator. After 1 day of culture, the dishes were washed with Hanks buffered salt solution (HBSS; Thermo Scientific, Agawam, MA, USA) to discard blood cells and nonadhesive cells and fresh culture medium was then added. The medium was refreshed every 3 days. When the cells reached 70% to 90% confluence, the cells were trypsinized (0.25% trypsin; Sigma), neutralized with cultured medium and then passaged at a ratio of 1:3. Cells of passages 3 to 4 were used in the experiments.

G∕C∕P biocomposites were cut to fit 12-well tissue culture plastics and then placed into the wells of TCP. For each well, human adipose-derived stem cells (ASCs) at a cell density of 10⁶ cells per cm² were seeded onto the biocomposite with the addition of cultured medium. After incubation for 24 h, the ASC-seeded biocomposite was washed with HBSS to remove the culture medium and nonattached cells.

Animal model and wound healing experiments

A skin wound healing model of nude mice was employed (Huang et al., 2012). All mice were acquired, housed, and studied under a protocol approved by the Institutional Animal Care and Use Committee of National Defense Medical Center, Taiwan, R.O.C. The three eight-week-old nude mice (BALB/c-nu; BioLASCO) were chosen for each group including GC_LP_H scaffold only and GC_LP_H scaffold seeded with ASCs in this study. All of the surgical instruments were sterilized and the surgical procedures were performed under laminar flow. The surgical sites were sterilized with Easy Antiseptic Liquid 2% (Panion & BF, Taipei, Taiwan) before surgery. After anesthesia, a square of full-thickness cutaneous wound (12 mm ×12 mm) was made by surgery using scalpel on each dorsum of the hind thighs, followed by grabbing, pulling the region with a forceps, and excising of the full-thickness tissue with scissors.

The wounds were divided into the following two groups: blank GC_LP_H biocomposite and ASCs-seeded GC_LP_H biocomposites. The GC_LP_H biocomposites (0.1–0.2 mm thick) for ASCs seeding were used. Both grafts were placed on each wound, sewn with 6 to 8 stitches using NC125L Nylon 5-0 surgical sutures (UNIK, Taipei, Taiwan), and covered with Tegaderm films (3M Health Care, MN) to prevent catching, biting, or wound infections. The wounds were continuously observed for a period of 3 weeks. The area of the wounds was blindly measured twice using ImageJ v1.44p software (http://imagej.nih.gov/ij; NIH, Bethesda, MD, USA). After 2 weeks, the tissues of healed wounds were excised, fixed in 10% formalin for at least 24 h at room temperature, subsequently embedded in paraffin, and sectioned in 5-µm increments. The sections were stained with Gomori’s trichrome staining and examined by an optical microscope (Olympus BX50, Hamburg, Germany) (Luna, 1992). The collagen component of the extracellular matrix deposited in skin substitutes was stained green.

Statistical analysis

Results were presented as the mean ± standard deviation of three replicates for each experiment. Statistical analysis was performed using Statistical Package for the Social Sciences, Version 12.0 (SPSS Inc., Chicago, Illinois). The statistically significant differences between groups were assessed by one-way ANOVA analysis of variance, followed by Tukey post-hoc test. P < 0.05 was considered statistically significant.

Protein release from the G∕C∕P

The compositions and fabrication of G∕C∕P biocomposites are present in Table 1 and the cumulative release of protein from the biocomposites is shown in Fig. 1 and Table 2. For G∕C∕P with a lower collagen content, a higher cumulative protein release was present for GC_LP_L biocomposite in comparison with GC_LP_H group. At 6 h, the former showed about 78% average release, whereas the latter showed 52% average release. The protein was released quickly in 12 h for both biocomposites but slow down afterwards. The cumulative release (w/w) at 24 h was 100% and 76% for two biocomposites respectively. For G∕C∕P with a higher collagen content, a higher cumulative release of protein was found for GC_HP_H biocomposite relative to GC_HP_L biocomposite. The former exhibited a higher average release rate of about 62% average release at 6 h, compared to 52% average release of the latter. The average cumulative release (w/w) after 24 h was 81% and 62% respectively for GC_HP_H and GC_HP_L biocomposites. The cumulative release reached a plateau for both biocomposites after 24 h. PCL is the polymer with slow degradation rate and low water uptake and hence GC_LP_H showed better collagen and gelatin encapsulation due to high PCL content. Therefore, unencapsulated gelatin or collagen were released within 3 days and the remained gelatin or collagen would be released slowly depending on the PCL degradation rate. In addition, the amount of protein release significantly increased by gelatin amount in G∕C∕P biocomposites with low PCL content of polymer but no effect in the those with high PCL content of polymer.

Thermal and tensile analyses for G∕C∕P biocomposites

The thermal properties of G∕C∕P biocomposites (n = 3) are listed in Table 3. For G∕C∕P made with GC_L mats, the melting point of the PCL component was close to the value of 60 °C normally found for pure PCL. The crystallinity of the PCL in these biocomposites (C/P and G∕C∕P) was estimated by comparing the fusion heat with that of 100% crystalline PCL as—139 J/g. From our previous DSC study (Dai et al., 2004), the 1:20 collagen:PCL biocomposite displayed the highest crystallinity of 63.5 ± 3.8%, followed by 48.5 ± 4.4% for 1:8 collagen:PCL biocomposites. However, the crystallinity of the PCL films prepared by solvent casting was only around 54.2 ± 5.4% in comparison with 1:20 w/w collagen:PCL biocomposites at 63.5 ± 3.8% (Table 3). For G∕C∕P biocomposites, the average crystallinity of GC_HP_H was found to increase by 2.9% relative to the solvent cast PCL film. Furthermore, the mean crystallinity of the PCL phase in the higher gelatin/collagen content (1:8 w/w) tended to be lower than that of 1:20 films. However, there was no statistically significance between these groups (Paired t-test: P = 0.173).

The mechanical properties of G∕C∕P vs. C/P are listed in Table 4. The enhanced tensile strength and elongation and lower Young’s modulus were detected in GC_LP_H when compared with C/P biocomposites (1:20 mixture).

Structure of G∕C∕P biocomposites

The SEM images of G∕C∕P biocomposites are shown in Fig. 2. From these image, an irregular pore structure (20–100 µm) was apparent on the surface of G∕C∕P biocomposite. The biocomposites with a lower PCL content (GC_LP_L) exhibited more porous structure. On the other hand, biocomposites with a higher collagen content (GC_HP_L and GC_HP_H) showed less pore structure when compared to those with a lower collagen content (GC_LP_L and GC_LP_H). The GC_HP_L biocomposite with highest content of collagen and lowest content of PCL showed a rough, fibrous surface due to the underlying collagen mat structure. The higher ratio of PCL was applied, the smoother surface was present because smooth overlying areas could be formed by the PCL phase. In addition, the surfaces of biocomposites with higher ratio of collagen seemed to have gel-like material filling the pores. All biocomposites exhibited the rough, fibrous structure of G/C mat overlaid by the smooth PCL phase. Little difference in morphology was observed.

Attachment and proliferation of PHEK, PHDF and ASC on G∕C∕P biocomposites

The attachment and proliferation of PHEK, PHDF and ASCs on G∕C∕P are shown in Fig. 3. The number of PHEK on GC_LP_H and GC_HP_L was greater than that on the other biocomposites at 1 day. The cell density of PHEK was greater on C/P biocomposites (1:8) than on G∕C∕P biocomposites at 3 and 6 days (P < 0.05). PHEK almost had equal cell density on C/P and G∕C∕P biocomposites at 9 days (P > 0.05). Moreover, the GC_LP_L biocomposite appeared to have the greatest number of cells at 9 days. However, no significant differences were present among these groups. The number of PHDF on GC_HP_L was greater than on the other biocomposites after 1 day (P < 0.05). There was no significant difference in the cell number among all groups at 4, 7, and 10 days. Results of cell proliferation on different biocomposites showed that PHEK exhibited faster cell growth than PHDF at 1 day. Meanwhile, the numbers of ASCs on different biomaterials were also investigated. Cell proliferation on GC_LP_H was similar to the other groups in 10 days. Based on these results, PHEK, PHDF and ASCs well attached and proliferated on all G∕C∕P biocomposites.

The morphology of PHEK and PHDF is shown in Fig. 4. Keratinocytes with similar granular shape were observed on the biocomposites. A higher expression of cytokeratin-14, the structural protein of mature keratinocyte, was induced as the incubation time increased. On the other hand, α tubulin, a major component of fibroblast cytoskeleton, was stained and appeared as the spindle-like shape. As the time increased, more α tubulin was expressed in PHDFs. The fluorescent expressions of cytokeratin-14 and α tubulin were parallel to the numbers of attached and proliferated cells shown in Fig. 3. In addition, the α tubulin expression of fibroblasts and cytokeratin-14 expression of keratinocytes seeded on GC_LP_H and GC_HP_L were higher than those of other groups at three days. Therefore, the GC_LP_H biocomposite with the lowest collagen content in the 4 GCP scaffolds was selected as the skin substitute for the pilot animal study because gelatin was a cheaper material than collagen.

In vivo animal study

Rapid wound closure was observed in the both groups covered with GC_LP_H biocomposites in comparison with the control group, as shown in Fig. 5. In both groups covered with GC_LP_H biocomposites, the wound size dramatically decreased apparently was observed from 8 to 12 days and then gradually decreased until 21 days. The rate of wound healing was almost the same for both groups within 16 days. However, the wound healing based on wound area exhibited statistical difference at day 21 (P < 0.05). The incomplete wound closure also observed even up to 21 days for the group covered with GC_LP_H only but not ASC-seeded GC_LP_H. Furthermore, Gomori’s trichrome staining was performed in control, groups covered with GC_LP_H only and ASC-seeded GC_LP_H, as presented in Fig. 6. The open wound covered with GC_LP_H only exhibited loose collagen deposition even after 14 days (Figs. 6B & 6E). In the ASC-seeded GC_LP_H group, complete wound closure with differentiated epidermis and abundant dermal parallel-arranged fibrous collagen deposition was observed at 14 days (Figs. 6C & 6F). At a larger magnification, the ASC-seeded GC_LP_H group showed the largest thickness of epidermis, followed by the GC_LP_H only group, and then the open wound. The thick epidermis layer was obviously observed in the outside portion of the wound for the ASC-seeded GC_LP_H.