To prepare AAM, adipose tissue was obtained through liposuction in human patients at the department of Plastic and Cosmetic Surgery, Nanfang Hospital. The study protocol was approved by the Institutional Review Board of the Nanfang hospital. After obtaining informed consent, human adipose tissue was harvested from four healthy female patients (aged 22–34 years) undergoing abdominal liposuction. Fat was obtained by manual liposuction using a combination of intravenous general anesthesia and local anesthesia. Two-hole blunt-tipped liposuction needles with diameters of 2.5 and 3.0 mm and a lateral hole (2 mm diameter) were used for suction with a 10 mL screw-tipped syringe under low negative pressure. The detailed procedure of AAM preparation is shown in Figure 1A. Briefly, the tissue was subjected to three cycles (1 h each) of freezing and thawing (−80°C to 37°C) in balanced salt solution. Following centrifugation (2,000 × g, 3 min), the fat samples were homogenized using a specially made blender (Shanghai Tiangong Instruments Co., Ltd., Shanghai, People’s Republic of China) at 3,000 rpm for 1 min. After homogenization, the adipose suspension was collected and centrifuged again at 3,000 × g for 3 min to remove the free lipid content. After washing with PBS, the samples underwent 6 h of polar solvent extraction in 99.9% isopropanol to remove the lipid content. After three rinses with phosphate-buffered saline (PBS), the samples were decellularized with aqueous sodium deoxycholate (2%) for 12 h. After three rinses with PBS, the samples were disinfected with 4% ethanol containing 0.1% peracetic acid for 4 h and then rinsed three more times with PBS. The remaining ECM component (i.e., AAM) was soaked in PBS containing 1% penicillin and stored at −20°C.

Human ADGFs were prepared as previously described (He et al., 2019). The obtained adipose tissue was washed with PBS and centrifuged at 1,200 × g for 3 min to remove blood constituents. Then, the adipose tissue was mixed with an equal volume of PBS and physically emulsified by repeated transfer between two 10 mL syringes connected by a female-to-female Luer-Lock connector (1.4 mm internal diameter). Adipose tissue was transferred for 1 min at a constant rate (10 mL/s). After centrifugation at 1,200 × g for 2 min, the liquid portion from emulsified adipose tissue was passed through a 0.20 μm syringe filter (Xiboshi; DIONEX, USA) to remove cell and tissue debris, and was collected as ADGFs.

For loading of ADGFs, heparin and AAM were crosslinked as described previously (Wissink et al., 2001). Briefly, the carboxylic acid groups of heparin were activated with 1-ethyl-3 (3-dimethylaminopropyl) carbodiimide (EDC)/N-hydroxysuccinimide (NHS). One milligram of heparin (H-4784; Sigma, Steinheim, Germany) was activated with 1 mg EDC and 0.6 mg NHS in 500 mL of 0.05 M MES buffer (pH 5.6) for 10 min. The AAM specimen was immersed in this solution, which was then evacuated at 20 mmHg for ∼2 min to remove air from AAM. The reaction was allowed to proceed for 4 h at 37°C, after which AAM was extensively washed with 0.1 M Na₂HPO₄ (2 h), 4 M NaCl (four times in 24 h), and distilled water (five times in 24 h). After the modification procedure, AAM was frozen at −80°C overnight and lyophilized. Native and heparinized AAM was cut into discs (1 × 1 × 0.5 cm³) for experiments. The native and heparinized AAM samples were incubated with 5 mL ADGF solution for 90 min at room temperature and then washed with 5 mL PBS (two times for 5 min) to remove all non-bound ADGFs. Finally, HEP-ADGF-AAM and ADGF-AAM were frozen at −80°C overnight.

Both fat tissue and AAM samples were fixed in 4% paraformaldehyde and embedded in paraffin. Five-micrometer sections were then cut for hematoxylin and eosin (H&E) staining. In addition, residual DNA was extracted using a DNeasy kit (Qiagen, Valencia, CA, USA). DNA content (µg/mg wet weight) was then quantified with a microplate reader (Model 680; Bio-Rad, Hercules, CA, USA) at 260 nm and normalized to the initial wet weight of each sample.

Scanning electron microscopy (SEM) SEM was used to examine decellulized products and normal fats. Samples were fixed in 2% glutaraldehyde for 2 h, dehydrated using an ethanol series (30%–100%), critical point-dried, coated with gold, and examined under a scanning electron microscope (S-3000N; Hitachi Ltd., Tokyo, Japan). SEM images were captured using a digital camera (Canon Inc., Tokyo, Japan).

To visualize the collagen turnover in the acellular products, the collagen components of ADM, AAM, and HEP-ADGF-AAM were labeled immediately after preparation according to a previously described protocol (Correa-Gallegos et al., 2019). Before being applied to wounds, ADM, AAM, and HEP-ADGF-AAM (500 mg) were incubated with 100 µM Alexa Fluor 647 NHS Ester (A20006; Thermo Fisher Scientific, Foster City, CA, USA) for 1 h at 25°C, and then washed three times with PBS.

The wound healing efficacy was evaluated using a full-thickness wound mouse model. Male nude mice aged 8–10 weeks and weighing 20–25 g were purchased from the Laboratory Animal Center of Southern Medical University. For wound healing experiments, mice were anesthetized and two circular full-thickness excisional skin wounds of 8-mm radius were made on the dorsum of mice. A prefabricated 12 mm silicone ring (inner ring radius of 8 mm) was placed around the wound and fixed using 4-0 silk sutures to prevent wound contraction (Lin et al., 2021). To compare in vivo wound healing efficacies, 48 nude mice were randomly divided into the HEP-ADGF-AAM, AAM, ADM (Beijing Jayyalife Biological Technology Co.,Ltd.), and control groups (n = 12 per group). Wounds were untreated in the control group, while covered with ADM, AAM, and HEP-ADGF-AAM dressings in other groups, respectively. All dressings had a diameter of 8 mm, and a thickness of 1 mm. Wounds are treated every 2 days and sterile dressings are changed daily. Photographs of the skin around the wound and healthy skin were taken at 5 and 14 days postoperatively (n = 6 per time point) and samples were taken for further histological analysis. At the end of the study, all mice were anesthetized by intraperitoneal injection of pentobarbital sodium (0.3%, 50 mg/kg). Cervical vertebra dislocation was used for euthanasia and death was confirmed by cessation of the heartbeat. Using the ImageJ software to quantify the wound area, wound healing is represented as follows: residual wound area/original wound area 100%. Researchers were blinded to the analysis of data from different groups.

To assess the adipogenic capacity of these acellular products in vivo, ADM, AAM, and HEP-ADGF-AAM (diameter of 8 mm and thickness of 3 mm) were subcutaneously inserted into the backs of nude mice, and then the skin was closed with 7-0 nylon sutures. Twelve weeks later, the implanted material was harvested and analyzed (n = 6 per group).

Fresh samples of wound skin and subcutaneous tissue were fixed in formalin for histological assessment and immunohistochemical staining. Briefly, formalin-fixed samples were embedded and sliced into 5-μm sections. For histological assessment, tissue sections were deparaffinized in xylene, rehydrated through graded alcohol in phosphate-buffered saline (PBS), and then stained with a hematoxylin-eosin (HE) working solution. For immunohistochemical analysis, sections were dewaxed and rehydrated, then incubated in 3% H₂O₂ to block endogenous peroxidase. Immunohistochemical staining was performed using antibodies against vimentin (1:25, ab92547, Abcam, Cambridge, UK) and perilipin (1:500, GP29; Progen, Heidelberg, Germany). The vimentin + cells and perilipin + adipocytes were evaluated by counting the number of brown-labeled cells from six fields of each slide. For immunofluorescence staining, the sections were incubated with the following primary antibodies: guinea pig anti-mouse Perilipin (Progen, GP29, Germany) and rat anti-CD31 (1:200; Abcam, Cambridge, MA, USA), followed by the corresponding secondary antibodies. After DAPI staining (Sigma, D9542, USA), the sections were observed and photographed using a FV10i-W confocal laser scanning microscope (Olympus, Japan). Angiogenesis was evaluated by counting the number of fluorescent vessel-like structures from six fields of each slide.

Upon immobilization of ADGFs onto native and heparinized AAM, the scaffolds were immersed in 2 mL PBS. Release of VEGF, HGF, bFGF, and EGF from HEP-ADGF-AAM and ADGF-AAM was assessed at 1, 3, 5 and 7 days. Supernatants were collected at designated time points and was analyzed using VEGF, HGF, bFGF, and EGF enzyme-linked immunosorbent assay (ELISA) kits (R&D Systems, Minneapolis, MN, USA) following the manufacturer’s instructions. The experiment was repeated three times.

Fibroblasts were cultured in six-well plates until reaching 100% confluence and a linear gap was generated using a sterile 200-μL pipette tip. Complete medium was then replaced with a serum-free medium supplemented with 2 mL of soak solution obtained by soaking ADM, AAM or HEP-ADGF-AAM within PBS for 24 h or serum-free medium (control group); the cells were then cultured for 0, 12 and 24 h. Three representative images of each scratched area were obtained using a light microscope (Nikon, Japan). Cell migration rate was analyzed using ImageJ software (NIH, MD, USA). All scratch-wound assays were carried out in triplicate.

The Matrigel (diluted 1:1 in PBS) was transferred to a 24-well plate and solidified by incubation at 37°C and 5% CO₂ for 30 min. HUVECs (5 × 10⁵ per well) were then seeded on the Matrigel and treated with 2 mL medium (1:1, soak solution obtained by soaking ADM, AAM or HEP-ADGF-AAM within PBS for 24 h to endothelial cell basal medium). HUVECs treated with angiogenic factors [20 ng/mL vascular endothelial growth factor (VEGF) and 5 ng/mL basic fibroblast growth factor (bFGF)] served as a positive control, while cells treated with 2 mL medium (1:1, PBS to endothelial cell medium) were used as a negative control. Angiogenesis is evaluated by counting the number of tubular structures from six fields of each well. Tube formation assays were repeated three times.

ADSCs (1 × 10⁵ per well) were plated on 6-well plates and cultured overnight at 37°C and 5% CO₂. The next day, cells in each well were rinsed and treated with 2 mL medium (1:1, supernatants from ADM, AAM or HEP-ADGF-AAM soak solution to human ADSC complete growth medium). ADSCs treated with 2 mL medium (1:1, PBS to human ADSC complete growth medium) were used as the negative control. The medium in all the groups was changed every 3 days. Adipogenic differentiation of ADSCs was examined for lipid accumulation by Oil Red O (ORO; Sigma-Aldrich, USA) staining. Cells were photographed using a Nikon E200 microscope and collected for gene expression analysis at days 21. Adipogenic induction assays were repeated three times.

Samples of wound beds and subcutaneous implants were quickly excised, immediately frozen in liquid nitrogen, and stored at −80°C. Total RNA was extracted from 50 mg frozen tissue using a RNeasy Lipid Tissue Mini Kit (Qiagen, Hilden, Germany), according to the manufacturer’s instructions. cDNA was amplified in 40 cycles using the QuantiTect Reverse Transcription Kit (Qiagen, Germany) and Rotor-Gene 3,000 Real-Time PCR Detection System (Corbett Research, Sydney, Australia). GAPDH was used as the reference gene. The primer sequences were as follows: peroxisome proliferator-activated receptor gamma (PPARG), forward 5′-TCG​CTG​ATG​CAC​TGC​CTA​TG-3′ and reverse 5′-GAG AGG​TCC​ACA​GAG​CTG​ATT-3′; CCAAT-enhancer-binding protein alpha (CEBPA), forward, 5′-CTT​GAT​GCA​ATC​CGG​ATC​AAA​C-3′ and reverse, 5′-CCC​GCA​GGA​ACA​TCT​TTA​AGT-3′; vascular endothelial growth factor A (VEGFA), forward 5′-TTA​CTG​CTG​TAC​CTC​CAC​C-3′ and reverse 5′-ACA​GGA​CGG​CTT​GAA​GAT​G-3′; basic fibroblast growth factor (bFGF), forward 5′- AGC​GGC​TGT​ACT​GCA​AAA​ACG​G-3′ and reverse 5′-CCT​TTG​ATA​GAC​ACA​ACT​CCT​CTC-3’. Expression levels were calculated using the 2^−ΔΔCT method.

Data were analysed using IBM SPSS (V22.0; IBM Corp., Armonk, NY, USA). An independent variable t-test was performed to compare the two mean values. For comparison of three or more means, one-way ANOVA was conducted, followed by the post hoc Tukey test. All results were expressed as mean standard deviation. The level of significance was set as p < 0.05.

AAM was prepared as shown in Figure 1A. Scanning electronic microscopy (SEM) showed that AAM had a porous structure, whereas normal fat contained a large volume of connected adipocytes and ADM presented condensed collagen fibers (Figure 1B). Hematoxylin and eosin (H&E) staining showed that AAM only contained ECM and lacked any cellular components (Figure 1C). The residual DNA content of AAM (7.42 ± 2.25 ng/mg) was significantly lower than that of normal adipose tissue (40.65 ± 3.12 ng/mg) (p < 0.001) (Figure 1D), indicating that cellular components were effectively removed from AAM.

HEP-ADGF-AAM was prepared as shown in Figure 2A. To assess the efficiency with which HEP-ADGF-AAM released ADGFs, growth factors released from HEP-ADGF-AAM were quantified using ELISA and compared with those released by ADGF-bound AAM (ADGF-AAM). ELISA of VEGF (Figure 2B), hepatocyte growth factor (HGF) (Figure 2C), bFGF (Figure 2D), and endothelial growth factor (EGF) (Figure 2E) showed that release of these growth factors was sustained after a peak in the HEP-ADGF-AAM group, with relatively high concentrations maintained for up to 7 days, while the concentrations of these growth factors remained low after Day 2 in the ADGF-AAM group.

Acellular wound dressing was applied to a full-layer wound model of mice. Wound healing was monitored on day 5 and day 14, respectively, and measured by the percentage of lesion area closed. Figure 3A shows representative images of full-layer wound beds of mice in different treatment groups. Histological observations (Figure 3A) showed that on day 5, the wounds treated with HEP-ADGF-AAM showed significant re-epithelialization, while virtually no re-epithelialization was observed in the blank control group. On the 14th day, the blank control group still had unhealed wounds. Compared with a blank control group, AAM, ADM, and HEP-ADGF-AAM all promoted wound healing to varying degrees (Figure 3B, p < 0.05). However, there was no significant difference in wound area between the AAM group and the ADM group. Surprisingly, the HEP-ADGF-AAM treatment group achieved almost complete wound healing on day 14, showing the best pro-healing properties. In addition, the level of re-epithelialization induced by HEP-ADGF-AAM at 14 days after injury was significantly higher than that in the control and AAM groups (Figure 3D). HE staining showed (Figure 3C) that granulation tissue, adipose cells, and regenerated skin attachment levels were higher in the HEP-ADGF-AAM group than in the other three groups at day 5 and 14 post-injury. We can see that on day 5, significant subcutaneous fat regeneration can be observed in both ADM and HEP-ADGF-AAM treatment groups. And, unlike the other groups, wound skin in the HEP-ADGF-AAM treated group had more regenerated hair follicles and sebaceous glands at day 14.

After the skin defect is formed, the repair process begins with the coordination of a variety of cells. The directional migration and proliferation of activated fibroblasts from the dermis to the wound is important for the remodeling of the extracellular matrix in the defective skin. For this purpose, we used Vimentin to track fibroblasts in the dermis. Immunohistochemical staining suggested (Figure 4A) that AAM, ADM, and HEP-ADGF-AAM could effectively promote the migration of fibroblasts to the wound bed on the 5th day after injury. Among them, the number of vimentin-positive fibroblasts in the wound bed of the HEP-ADGF-AAM group was significantly higher than that in AAM, ADM, and control groups (p < 0.05) (Figure 4B). In addition, fibroblasts cultured in immersion solution with different acellular products were used for in vitro scratch experiments (Figure 4C). The results showed that after 12 and 24 h, the fibroblast migration rate of the HEP-ADGF-AAM group was significantly faster than that of the other three groups. However, fibroblast migration did not differ significantly between ADM, AAM, and negative controls (Figure 4D).

To explore whether treatment with HEP-ADGF-AAM has the effect of stimulating angiogenesis, we used CD31 immunofluorescence staining to characterize and quantify the number of CD31-positive capillary cells in wounds of different treatment groups. Only a small amount of neovascularization was observed in the blank control group, and treatment with AAM, ADM, and HEP-ADGF-AAM significantly promoted wound angiogenesis (Figure 5A). Quantitative angiogenesis showed that the amount of CD31 in AAM group, ADM group, and HEP-ADGF-AAM group was significantly higher than that in the control group (p < 0.05) (Figure 5B). To further confirm their ability to promote vascular regeneration, mRNA levels of angiogenic genes (VEGF and bFGF) in the wound bed were detected by qRT-PCR. The results showed that these levels in the HEP-ADGF-AAM, AAM, and ADM groups were significantly higher than those in the control group (p < 0.05) (Figure 5C). In addition, we used in vitro tube formation experiments to directly observe the angiogenesis induction ability of HEP-ADGF-AAM on human umbilical vein endothelial cells (HUVECs). In the presence of HEP-ADGF-AAM, the typical tubular structure could be observed after 18h, which was similar to the positive control group and significantly higher than the negative group. However, almost no typical tubular structure was observed when HUVECs were cultured with AAM or ADM throughout the assay process (Figure 5D). Quantitative analysis of capillary tubular structures showed (Figure 5E) that the number of tubular structures in the HUVECsHEP-ADGF-AAM group and positive control group cultured with HEP-ADGF-AAM was significantly higher than that in other groups (p < 0.05).

At Day 5, perilipin + adipocytes were found in the wound bed in all four groups (Figure 6A). The number of perilipin + adipocytes in the wound bed was significantly higher in the HEP-ADGF-AAM group than in the other three groups (p < 0.05) (Figure 6B). Expression of adipogenic genes was tested by qRT-PCR. Expression of PPARγ and CEBP/α in wounds was significantly higher in the HEP-ADGF-AAM group than in the other three groups (p < 0.05) (Figure 6C). To evaluate the adipogenic capacities of the acellular products, they were subcutaneously transplanted into nude mice. After 12 weeks, H&E staining showed that grafts in the ADM group did not contain adipocytes, and this was confirmed by immunofluorescence staining of perilipin. Viable adipocytes (perilipin+) were found in the AAM and HEP-ADGF-AAM groups (Figure 6D). mRNA levels of adipogenic genes (PPARγ and CEBP/α) in the grafts were tested by qRT-PCR. Adipogenic gene expression was significantly higher in the HEP-ADGF-AAM and AAM groups than in the ADM group (p < 0.05) (Figure 6E). Oil Red O staining was performed following culture of ADSCs with the acellular products under adipogenic induction conditions (Figure 6F). Measurement of absorbance at 530 nm showed that HEP-ADGF-AAM improved adipogenic differentiation, while AAM and ADM did not (Figure 6G). In addition, the mRNA level of PPARγ was significantly higher in the HEP-ADGF-AAM group than in the other three groups (p < 0.05) (Figure 6H).