Human fetal mesenchymal stem cells secretome promotes scarless diabetic wound healing through heat‐shock protein family

Abstract The high mortality rate of patients with diabetic foot ulcers is urging the appearance of an effective biomedical drug. Senescence is one of the major reasons of aging‐induced decline in the diabetic wound. Our previous studies have demonstrated the anti‐senescence effect of secretomes derived from human fetal mesenchymal stem cells (hfMSC). The present study tends to explore the potential role of hfMSC secretome (HFS) in wound healing through anti‐aging. Meanwhile, we try to overcome several obstacles in the clinical application of stem cell secretome. A verticle bioreactor and microcarriers are employed to expand hfMSC and produce the HFS on a large scale. The HFS was then subjected to lyophilization (L‐HFS). The PLGA (poly lactic‐co‐glycolic acid) particles were used to encapsulate and protect L‐HFS from degradation in the streptozotocin (STZ)‐induced diabetic rat model. Results showed that HFS‐PLGA significantly enhanced wound healing by promoting vascularization and inhibiting inflammation in the skin wound bed. We further analyzed the contents of HFS. Isobaric tag for relative and absolute quantitation (ITRAQ) and label‐free methods were used to identify peptides in the secretome. Bioinformatics analysis indicated that exosome production‐related singling pathways and heat‐shock protein family could be used as bio‐functional markers and quality control for stem cell secretome production.


| INTRODUCTION
Diabetes mellitus is one the most common metabolic diseases leading to impaired chronic wound conditions. The 5-year mortality rate for patients suffering from DFU (diabetic foot ulcer) is even higher than that of prostate or breast cancer. 1 7 The function of the secretome is differed and depends on the origin of the cell. 8 Human first trimester (within 3-month implantation) mesenchymal stem cells (hfMSC) exhibited longer telomere and higher activity of telomerase allowing prolonged culture in vitro. 9 Our previous study found that transferring only one mitochondrion from hfMSC to the recipient cell is sufficient to induce anti-aging gene expression. 10 The secretome derived from hfMSC showed an anti-senescence effect on adult cells. 11,12 Meanwhile, in contrast to salamanders who regenerate whole limbs after amputation even in adulthood, scarless healing in humans is only seen in fetal skins. 13,14 The fetal wound healing process is different from the adult in many aspects, such as growth factors, gene expression profiles, extracellular matrix (ECM), and inflammatory responses. [15][16][17] Thus, studying the unique proteins or factors of hfMSC secretome (HFS) may shed a light on the scarless healing in diabetic wounds.
Several studies have proven the effect of cell secretomes on tissue repair and regeneration. [18][19][20][21][22] Stem cell secretomes hold great potential in clinical application. The present study focus on various obstacles that stand in the way from bench to bedside. The traditional monolayer culture system is not sufficient to meet the requirement of the amount of the secretome on a large scale. The limiting surface area in monolayer culture reduces the production rate of secretome as well as prolongs the time to expand the stem cell. The bioreactor is an ideal choice for scaling up the manufacturing of HFS. The presence of microcarriers provides enough surface areas for cell attachment while shear stresses caused by agitation stimulate cell proliferation. 23 Phosphate buffered saline (PBS) Biotech (California, USA) has developed a vertical-peddle bioreactor providing optimized stir power and liquid shear force to the cells on microcarriers, 24 which has been used for the successful expansion of MSCs and inducible pluripotent stem cells. 25,26 The PBS bioreactor system was used in the present study, to produce HFS on a large scale. MSCs released at least 44 biological factors (include TIMP2, FAS, MIP-3b, TRAIL R4, etc). 27 The bioactivities of the factors are of most importance in contributing to the function of the secretome. Reciprocal freeze and thaw of secretome during storage and transportation may compromise the function of HFS.
Freeze-drying is the most wide-accepted method to preserve the activity of protein drugs. 28 In our study, the bioreactor-produced HFS is subject to lyophilization making it become an off-the-shelf product.
Systematic injection of secretome will require a large amount of the factors while the side effect is hard to predict. Thus, local administration of secretome as a free drug is the most efficient way to achieve effective dosage. However, the retention time of drugs in the wound area is limited due to the blood flow and enzyme degradation. Biomaterial-based particles have been used to retain and release the drug over time. Polymeric particle is the optimized option that can reach the therapeutic effect with reduced the effective dosage. 29 PLGA (poly lactic-co-glycolic acid) particle has been used for the controlled delivery of small molecule drugs and proteins. Especially, PLGA encapsulated drugs have been widely used in clinics for reducing inflammation. 30 In our study, we encapsulated HFS in PLGA which is a mature drug release system to retain the bioactivity of HFS and prolonged the required amount of secretome during wound healing.
Recently, a study summarized clinical studies related to secretomes which can be found on clinical.gov, only three of them were completed without the availability of the result. Until now, there is no clear regulations have been stated by FDA. The age of the donor or the tissue sources of the cell is different, leading to the composition of the secretome depending on the variable cell source. 31 Allogenic secretome therapy requires the standardization of internal quality control to calibrate its effective dosage as the inter-donor variation cannot guarantee the therapeutic effect. In the present study, we compare the proteomic profile of the secretome from fetal stem cells and adult stem cells for anti-aging-related proteins which can explain the superior function of the HFS. We used a high throughput proteomic method (ITRAQ) to compare HFS with HAS (human adult stem cell secretome). The data were analyzed using Cytoscape and ClueGO for bioinformatics prediction, 32 Our study identified that the heatshock protein family showed their pivotal role in promoting wound healing and angiogenesis. These factors have the potential to be biomarkers and quality control of secretome production on large scale.

| Large-scale expansion of hfMSC in vertical-wheel bioreactor
We use a vertical-wheel bioreactor combined with collagen-coated microcarriers. We first optimized the ratio of the number of hfMSC to F I G U R E 1 Large scale expansion of human fetal mesenchymal stem cells (hfMSCs) in a vertical-wheel bioreactor. (a) Cell viability of various density of cells was measured using Alamar blue test. The data were normalized to negative control of all time points. Starting cell density was 1 Â 106 cells/100 ml. N = 3; (b). Live/dead cell staining results showed that fetal MSCs and microcarriers formed aggregates and more than 90% of the cells were alive at Day 14 culture; (c). Cell metabolic index changes during culture. The glucose concentration kept on decreasing during culture, indicating that cells were actively proliferating. After change into serum-free media, at Day 15, lactate concentration was lower in the hfMSCs culture suggesting hfMSC have a unique pattern of energy metabolism, N = 3; (d). During 14 days culture, ELISA results showed that the concentration of TGFβ2, VEGF, MMP1, and MMP2 were significantly higher in media of hfMSCs than those in haMSCs media; Statistic: Mann-Whitney U (unpaired, nonparametric T-tests). N = 3, *p < 0.5. Brightness and contrast were adjusted for clear demonstrated contents of the representative pictures. TGFβ2, Transforming growth factor beta-2; VEGF, Vascular endothelial growth factor; MMP, matrix metalloproteinase. the weight of microcarriers. A series of cell densities were tested. The various numbers of hfMSCs and the microcarrier were mixed and co-cultured in 96 well plates. The Alamar blue result indicated that when cell density was higher than 500 cells/100 μl medium (equal to 500 cells/0.03 g microcarriers), the hfMSCs kept proliferating at least for 1 week (Figure 1a). We, therefore, chose 1000 cell/0.03 g microcarriers as our standard culture condition. And the culture medium will be half changed every 7 days. Usually, 1-2 weeks of culture is sufficient for hfMSCs and microcarriers to form visible aggregates, live/dead cell staining showed that more than 90% of cells were alive (Figure 1b). In bioreactor culture, the formation of the aggregates suggested that the culture system has reached its limits. A F I G U R E 2 Lyophilized human fetal mesenchymal stem cell secretome (L-HFS) rejuvenated human adult skin cells into a fetal skin cells-like phenotype. (a) Effects of L-HFS on the proliferation of human skin cells. L-HFS promoted cell proliferation of human keratinocytes and slightly inhibited the cell proliferation of human fibroblasts at Day 1, but the cell proliferation of fibroblasts was enhanced at Day 3; (b) Scratch assay results demonstrated that L-HFS significantly promoted gaps closure of human keratinocytes and fibroblasts. (c) Quantitative result of the gap area. The data represent two-time points, 24 and 48 h after scratch. Areas were calculated by image J. N = 3, *p < 0.05; (d) Fibroblast populated collagen lattice (FPCL) assay showed that HFS treatment reduced the shrinkage of FPCL. The surface area was measured by image J. (e) Quantitative results of FPCL surface area, n = 3, *p < 0.05; (f) Gross view of H&E staining of the FPCL slice. The morphology of L-HFS group is flattened while the control group appear obvious contraction. Scale bar = 500 μm; (g) Sirius red staining and polarized microscopy examination showed that there was more green color in the L-HFS treatment group compared with the control group, indicating the higher intensity of type III collagen. The green color indicates type III collagen, while the yellow color indicates type I collagen; (h and i) Ratio of type I collagen to type III collagen based on the color intensity was measured by image J. Immunohistochemistry staining showed decreased expression of pro-type III procollagen suggesting the maturation of type III collagen was enhanced after L-HFS treatment. Scale bar = 200 μm; Statistic: Mann-Whitney U (unpaired, nonparametric T-tests). N = 3, *p < 0.5. Brightness and contrast were adjusted for clear demonstrated contents of the representative pictures beads-to-beads passage is required to expand the culture scale. In 100 ml reaction volume, about 5 Â 10 6 human bone marrow mesenchymal stem cells were seeded. The cell number increased to nearly 3 Â 10 7 cells after 10-day cultures. We then calculate the doubling time of hfMSC based on the number of cells. The result shows that, in bioreactor culture, the doubling time of hfMSC is around 36 h which is similar to that of the monolayer culture.
Thus, we choose 2 weeks as our endpoint to collect the secretome when we observed the formation of the aggregates. To monitor the metabolic status of cells during culture, we tested the glucose and lactate concentration of the culture media. Human adult mesenchymal stem cells (haMSCs) were used as control. Compared to haMSCs, hfMSCs utilized more glucose during culture, indicating that hfMSCs have a more active metabolic status. On Day 15 in culture, despite F I G U R E 3 Legend on next page. the cell numbers of hfMSCs being significantly higher than that of haMSCs, they only produced 1/10 of lactate in contrast to that of haMSC ( Figure 1c). We performed an ELISA test to measure several well-established factors during 14 days of culture. hfMSCs produced more TGF-β2, PDGF-BB, MMP1, and MMP2 in contrast to haMSCs at all-time points (Figure 1d). Therefore, hfMSCs have higher proliferation ability and higher metabolic rate, they could produce higher concentrations of growth factors in HFS. The collected secretome was then subjected to lyophilization to keep the bioactivity of the factors.

| Lyophilized HFS treatment rejuvenated human adult skin cells into a fetal skin cells-like phenotype
The migration and proliferation of skin cells are crucial for skin regeneration. The effects of lyophilized HFS (L-HFS) on human keratinocytes and fibroblasts migration and proliferation were tested.
Alamar blue assay was used to test the optimal dosage of L-HFS. An effective concentration window (10-50 ng/μl) was selected as the optimal dosage for the next steps ( Figure 2a). Scratch assay con- The L-HFS treatment group had less type I collagen and more type III collagen (Figure 2g,h). Procollagen type III is the precursor of collagen type III. We have also found the expression of pro-collagen type III in the L-HFS treated group was significantly decreased, suggesting that L-HFS treatment promotes type III collagen maturation. More pro-collagen type III has been transformed into collagen type III in the L-HFS group (Figure 2g,i). The fetal skin cell produces more type III collagen than the adult skin cell. The L-HFS treatment may suppress the scar formation and increase the ratio of type III/type I collagen in the adult skin cell. These results suggest that L-HFS may turn the adult cell into fetal cell status.

| HFS encapsulated in PLGA particles retained bioactivities of HFS in vitro
In the wound area, direct application of HFS powder cannot retain its bioactivities for long due to blood flushing or enzyme degradation.
We used PLGA particles to encapsulate L-HFS powder (HFS-PLGA). Crystal violet were dissolved in 33% acetate and absorbance was measured at 570 nm. HFS-PLGA group showed significantly higher absorbance compare to the PBS-PLGA group. N = 3; (g) Represented view of staining result of KOC slides and quantitative result. H&E staining showed that HFS-PLGA increased the thickness of the keratinocyte layer in KOC. (h). Immuno-staining showed that HFS-PLGA decreased pro-collagen type III staining. The quantitative data was measured by image J; I-J. Immuno-fluorescence staining showed that HFS-PLGA increase the expression of cytokeratin 10 and cytokeratin 14 in KOC model. The quantitative data were measured by image J; K. Sirius red staining and polarized microscopy examination showed a higher ratio of collagen III to collagen I in HFS-PLGA group. The quantitative data were measured by image J based on the polarized signaling, green color represents type III collagen and yellow color represents type I collagen; Statistic: Mann-Whitney U (unpaired, nonparametric T-tests). N = 3, *p < 0.5. Brightness and contrast were adjusted for clear demonstrated contents of the representative pictures significantly increased the thickness of the keratinocyte layer ( Figure 3g). We also found the cornified pearl in the keratinocyte layer in the HFS-PLGA group. This result suggested that HFS-PLGA promoted keratinocyte maturation compared to the control group.
Immunohistochemistry staining showed that the HFS-PLGA group had reduced expression of type III pro-collagen ( Figure 3h). We use F I G U R E 4 Legend on next page. cytokeratin 10 (CK10) and cytokeratin 14 (CK14) as two proteins marker of keratinocyte maturation. CK10 is localized in the suprabasal layer and superficial layer of the epidermis of fetal skin. 33 CK14 is localized in the basal layer to interact with skin stem cells corresponding to the proliferation and differentiation of keratinocytes. 34 We performed the IF staining on KOC slides of CK14 and CK10 to indicate the maturation process of human keratinocytes (Figure 3i  Based on FEPs, a protein-protein interaction (PPI) network was constructed in cytoscape. The interaction information was extracted from the STRING database. We then used CytoHubba, a plugin of cytoscape, to identify hub proteins. A hub network was established by combining the hub proteins and their first-neighbor proteins. All the proteins within the hub network were subject to Metascape for visualization and further analysis. We identified six undestroyable sub- Metascape also performed GO enrichment based on hub-network, the proteins were enriched on several biological processes (BPs) in the HFS (Figure 5d). Among which VEGFR1 pathway was highlighted supporting that HFS has strong pro-angiogenic properties. We further performed GO enrichment based on the regulators and the effectors in ClueGo (Figure 5e). The enriched terms were divided into three categories. The first category was only enriched with the regulators such as spliceosome, proteins targeting ER, and COVI-coated vehicles. The second category was co-enriched with the regulators and the effectors, including proteins in mitochondria, lysosome, and tRNA aminoacylation for protein translation, responsible for signaling transduction from the nucleus event to the cellular level. The third category is only enriched by the effectors, two terms related to focal adhesion and endosome transport were highlighted. FEPs enriched on the abovementioned terms suggested that hfMSCs had higher activities in cell proliferation, migration, extracellular secretion, and protein synthesis and degradation compared to those of haMSCs. The GO results suggested that exosome secretion and protein degradation are the two F I G U R E 5 ITRAQ analysis revealed that the exosome, Heat-shock proteins, and 14-3-3 proteins family play a core role in the fetal enriched proteins (FEP) network. (a). The expression heatmap of differentially expressed proteins (DEPs). The genes were pre-selected base on Wilcoxon signed-rank test. Only significantly highly expressed proteins (p < 0.05)were included in the heatmap. Rows represent individual proteins identified and relatively quantified by iTRAQ. Columns represent the individual samples. For bioreactor culture, the intensity scale is green, black and red. Green indicates lower and red higher expression in protein level. For monolayer culture, the intensity scale is blue, white, and yellow. In summary, the bioinformatic result revealed that the exosomes, 14-3-3 family proteins, and heat-shock proteins might be the main contributors to the biological functions of the HFS, and they might be potential biomarkers for quality controls of HFS production.

| Intracellular exosomes, YWHAH, and HSPA8 regulate angiogenesis
We performed quantitative polymerase chain reaction (qPCR) to validate the expression of the candidate proteins selected from the bioinformatic analysis. The results showed that YWHAG, SFN, YWHAH, HSPH1, HSPD1, HSPA8, and NEDD4 were highly expressed in F I G U R E 6 Gene ortholog (GO) network of regulator and effector genes. (a) ClueGO analysis was applied to the regulators and the effectors genes, respectively. The GO terms were shown as nodes and linked based on the kappa score (≥0.3). Only the most significant proteins enriched in each term were shown in the plots. The node size represents the enrichment significance. The red color represents the regulator proteins, the green color represents the effector proteins. The proportions of the two types of genes in mixed terms was visited as green and red color in one node. The distribution of two groups visualized on network based on their kappa score level (≥0.3), where only the label of the most significant term per group is shown as the smallest nodes. The larger node size represents the enrichment significance of this term hfMSCs (Figure 8a Figure 8d,e). These results indicated that the exosomes, HSPA8, and YWHAH in the HFS might be the angiogenic regulators, and the soluble HSPA8 and YWHAH might be used as biomarkers for quality control for HFS production.

| DISCUSSION
The present study used bioreactors and microcarriers to culture expand the hfMSC on large scale. HFS is lyophilized for long-term storage, easy to transport and use. The L-HFS were encapsulated into PLGA particles for slow-releasing purposes, and HFS-PLGA significantly promoted wound healing in STZ-induced diabetic rats. Thus, our study successfully developed a bioproduct that might facilitate the treatment of diabetic foot ulcers. PDGF-BB is an FDA-approved biological drug for wound healing management, it promotes healing through stimulating granulation tissue formation, enhancing reepithelialization and vascularization as well as promoting collagen production. 35 In the present study, we found that HFS exerted comparable if not better effects than PDGF-BB. Cell source, in vitro expansion, and quality control, are three major aspects that decide whether the mesenchymal stem cell is suitable to be an Off-the-shelf bioproduct. Our study tried to prove several new concepts focusing on improving the production of stem cell secretome, which is a new direction of stem cell therapy. The MSCs have been isolated from various tissues, bone marrow, 36 adipose tissue, 37 39 The fetal cell may tend to use oxidative phosphorylation pathway as a major energy supply. This was supported by our previous study that transferred fetal cell mitochondria triggered metabolic reprogramming of the adult cell. 10 The metabolic status of the MSC might correlate with cellular senescence. Less acidified microenvironments might play an important role in promoting in vitro expansion.
The fetal cell properties are not only beneficial for cell proliferation.
It is also important for tissue repair in terms of scarless healing in skin wounds. One crucial event during skin wound healing is fibroblastmyoblast transformation. Myoblast's contraction helps wound closure.
However, excessive myoblasts contraction could lead to scar formation. 40 Contractures limit the normal growth of skin cells and impede the fully functional recovery of skin wounds. Scarless healing in the early gestation stage in mammals was reported in 1971, and the intrinsic properties of the fetal cells mainly contribute to the anti-scar ability. 41 It has been known that fetal skin mainly consists of type III collagen while adult skin mainly consists of type I collagen. Excessive deposition of type I contributes to scar formation. A high ratio of Type III to Type I collagen indicates better anti-contraction potential. 42 Our data have demonstrated that HFS treatment inhibited fibroblast contraction and increased the ratio of typeIII/type I collagen. This result suggests that HFS might endow some fetal cell properties to the adult skin cell.
There are two major types of bioreactors that have been used to scale up the MSC expansion. The hollow fiber-based system 43 and microcarrier-based system. 44 These two systems both have been used for the culture of various types of MSC. Compared with monolayer culture, bioreactor culture could produce 15-20 fold of the cell using the same volume of the culture medium. The stem cell properties could be maintained during the large-scale expansion. 45 The hollow fiber system is static culture while the microcarrier system requires the stirred tank to maintain the refreshment of oxygen and nutrition.
In our study, we choose the vertical stirred tank to scale up the hfMSC since the tank is more convenient to collect the secretome as more as possible. The culture medium did not need to be refreshed too often during the culture period.
The microcarrier system is more widely accepted than the hollow fiber system. Our present study chooses a small bioreactor with a 100 ml culture volume. We proved the feasibility of expanding MSC in the bioreactor. The secretome we collected under this culture condition kept its regenerative function unchanged. The bioreactorderived HFS showed a similar rejuvenating effect as the secretome derived from monolayer culture. A recent study has scaled up the MSC culture in a 50-L bioreactor using a microcarrier system. 46 We may scale up the current system in the future. The safety and effectiveness of MSCs therapy are still under debate, there are concerns that injection of cells, as well as cellular components, may induce side effects, such as immune responses and risks of infection and disease transmission. It is well accepted that MSCs culture conditioned media contain many biological active factors is responsible for modulation of inflammation, cell recruitment, matrix deposition, and cell proliferation. 47 MSCs could release anti-inflammatory factors such as IL-4, -6 and -10, and TGF-β. 48,49 Furthermore, MSCs stimulate the proliferation and migration of various types of cells involved in wound healing via paracrine factors. 50 Thus, MSCs-derived secretory factors could be a reliable sources for promoting wound healing. Moreover, our study analyzed the protein content in HFS through proteomics and bioinformatic analysis. We had identified several key signaling pathways that might contribute to the unique properties of hfMSC. Among these, heat-shock proteins and 14-3-3 proteins were identified as two critical families corresponding to multiple cellular functions. We found that HSPA8 and YWHAH, unique contents of HFS, significantly promote angiogenesis, were crucial in wound healing. We also found multiple biological processes correlate with the biogenesis of exosomes in hfMSC. The 14-3-3 protein family was also regarded as a sub-family of heat-shock protein.
Usually, they exert their function together with the other heat-shock proteins to protect the cell from physiological stress. 51,52 The crucial role of YWHAH and HSPA8 in our protein network indicated that heat-shock protein might play an important role in the superior role of hfMSC. Furthermore, the heat-shock protein and cellular senescence were closely correlated with each other. The fetal cells might have a unique cellular defensive mechanism helping to fight against senescence.
Collectively, our study elucidated novel factors regulating angiogenesis from the hfMSC secretome. The presence of exosomes, HSPA8, and YWHAH might be taken as unique biomarkers for quality control of HFS, this is important for the standardization of production of MSC secretome production.
Stem cell secretome is a mixture, current studies focused on verifying its biological functions. The mechanisms behind these functions remained largely unknown, which limits the wider clinical applications of MSCs secretome. One possible research direction is to analyze the bioactive contents of the secretome in detail. In the present study, we used a high-throughput proteomics technique and determined that the main components of secretomes are exosomes. Exomere is a newly founded subclass of exosomes, that mainly contains proteins related to the glycolysis pathway and mTORC pathway, suggesting that exomere is involved in regulating energy metabolism and cell proliferation. 53 To date, the biological functions of exomere derived from mesenchymal stem cells are unknown and there are no universal biomarkers for exomere. The role of exomere in tissue repair is interesting. One important finding of our study was that we confirmed that HSPA8 mediated wound healing. In HFS, we had identified that HSPA8 was highly expressed, which had also been founded in the tumor as exomere. 53 Thus, HFS contains exomere that might regulate cell energy metabolism and other activities.
The Warburg effect is a well-known phenomenon in that cells tend to utilize the tricarboxylic acid (TCA) pathway for energy metabolism under physiological conditions and switch to anaerobic glycolysis when they are under stress such as in tumorigenesis and anaerobic glycolysis is regarded as a more efficient manner to use the limited ATP. 54 The production of lactate reflects the activity of glycolysis. Our ITRAQ analysis indicated that HFS contained factors that regulated glycolysis such as HSPA8. However, hfMSCs produced extremely lower lactate in bioreactor culture compared to the haMSCs, suggesting that hfMSCs had unique and superior metabolic properties. Recently, studies reported that glycolysis activity was up-regulated during tissue repair 55 and the glycolysis signals released from the tumor cells stimulated the growth of surrounding mesenchymal tissues. 56 Tissue repair and tumor growth-share many common signaling pathways such as angiogenesis and energy consumption. Topical application of HFS may directly regulate glycolysis and promote angiogenesis during wound healing. Therefore, targeting the glycolysis pathway may be a new therapeutic strategy for promoting tissue regeneration.
Stem cell secretome has the potential to be industrialized. However, our production platform is still in its imperfect. Our study is currently limited to the laboratory. A larger scale is needed to meet the requirements of industrialization. We need to enlarge the culture volume from 100 ml Alamar blue is used to measure the cell proliferation ability, We have calculated the cell number before and after plating at two-time points, 5-and 10-days post cell seeding. At each stage, samples were collected for quantification. In brief, 5 ml of well-mixed cellladen microcarrier suspension was transferred into 15 ml sample tubes and allowed to settle. Samples were then washed twice with 5 ml PBS to remove the residue medium. After discarding the supernatant, 2 ml 0.25% Trypsin-EDTA was added to the microcarriers to allow the cells to detach from the microcarriers. 10uL cell suspension was then applied to the hemocytometer.

| Protein identification of secretome from a fetal and adult cell
All proteomics analyses were sent to BGI (Shenzhen, China). Three biological replicates were carried out for each sample. We used the previously described 61 ITRAQ method to identify proteins for monolayer cultured secretome. Briefly, the HFS will be precipitated by acetone at À20 C overnight to obtain a reduced and alkylated protein mixture. A 100 μg of each sample solution is digested with Trypsin Gold (Promega, Madison, WI, USA). After digestion, peptides are labeled with an 8-plex iTRAQ reagent (Applied Biosystems, USA) according to the manufacturer's protocol. An LC-20AB HPLC pump system (Shimadzu, Kyoto, Japan) was used to separate each sample into 20 fractions. Each fraction was re-suspended, then the peptides are subjected to nanoelectrospray ionization followed by tandem mass spectrometry (MS/MS) in a QEXACTIVE (Thermo Fisher Scientific, San Jose, CA, USA) coupled online to the HPLC. Raw data files acquired from the Orbitrap will be converted into MGF files using Proteome Discoverer 1.2 (PD 1.2, Thermo-Fisher, USA). Protein identifications will be performed by using Maxquant with a search engine.
For protein identification, the parameters followed the published study. 62 We use a label-free method to identify protein in the powder of L-HFS. Due to protein concentration being reduced after reconstitution. It is no longer suitable for the ITRAQ procedure. BGI used a customized procedure to enhance the sensitivity of protein detection.

| Cell proliferation and migration assays
The cell proliferation was determined with Alamar blue assay (Life Technologies, USA) as previously reported. 11 Scratch assay was used for cell migration assay according to the previously described methods. The ability of HFS to recruit MSCs was also evaluated by transwell assay as previously described. 65 In brief, haMSCs were seeded in transwell inserts at a density of 5 Â 10 4 cells/cm 2 with 600-μl serum-free medium. The HFS at various concentrations were then added into the bottom chambers of the 24-well plates. The cells were cultured for another 16 h, and residual cells on the upper surfaces of the inserts were removed with cotton swabs, while the migrated cells on the lower surfaces were stained with 0.5% crystal violet and counted.

| Primary culture of human keratinocytes and fibroblast
Primary human keratinocytes were prepared as previously described with a few modifications. 66 Normal human keratinocytes and fibroblasts harvested from pediatric foreskins were used. Briefly, the foreskin was surgically removed and cut into 1.0 Â 0.5 cm strips, and incubated with trypsin/EDTA solution at 4 C overnight. The epidermis was then separated and keratinocytes were collected by centrifugation at 1400 rpm for 5 min and 2 Â 10 6 keratinocytes were seeded into a type IV collagen-coated plate and cultured in complete K-SFM medium (Gibco, USA) at 37 C.
For fibroblast culture, cells were prepared according to previous report. 67 The dermis was mechanically separated from the epidermis.
The dermis was digested with dispase overnight and finely minced.
The resultant cell suspension was transferred to culture dishes and subculture in DMEM supplemented with 10% FBS. Cells were stored in liquid nitrogen.

| Fibroblast-populated collagen lattice
FPCL was prepared according to previously described protocols with minor modifications. 66 In brief, six-well plates were precoated with 1% agarose. Two milliliters of FPCLs containing 1.2 Â 10 5 cells and 1.25 mg/ml type I collagen in complete medium (DMEM supplemented with 10% FBS) were cast in the plates. The gels were then polymerized at 37 C for 30 min. Afterward, the gels were gently detached from the agarose surface to allow contraction, and 2 ml of complete media with or without L-HFS was added per well.

| Keratinocyte organotypic cultures
The organotypic 3D keratinocyte-fibroblast co-culture was prepared according to previously described protocols. 68

| Diabetic rat wound healing model
We have taken the gender effect into account when we designed the experiment. According to the literature, the sensitivity to STZ is significantly (p < 0.001) higher in male rats compared to female. The estrogen could protect pancreatic β cell from apoptosis induced by oxidative stress. 69 We choose male rats in this study to increase the success rate of the STZ model. To establish a diabetic rat model, streptozotocin (STZ, Sigma-Aldrich, USA) prepared in 0.1 M citrate buffer (pH 4.5) was injected intraperitoneally at the dose 50 mg/kg as described previously. STZ could induce diabetes within 3 days by destroying the β cells. 70 The blood glucose was measured at 5 and 7 days after injection to confirm the onset of hyperglycemia. On day 10 following STZ injection the confirmed diabetic rats were used to create skin wounds according to the methods previously reported. 71 In PDGF-BB is used as a "positive" control in this study. A sterile dressing was used to cover the splint to allow the tissue and dressing to have no contact with each other. The dressing could prevent dehydration and contamination of the wound, the dressing was changed every 3 days.

| Histological analysis and immunostaining examinations
All the control and treatment group pictures will be merged into one picture before starting. This step allows regulating two groups' parameters together to avoid objective bias.
We use the following steps to measure the A.O.D (average optical density).
1. The channel of the JPEG picture will be turned into RGB 8-bit format.
2. Use the ROI tool to mark all the samples.
3. The threshold will be manually regulated based on one experience. 4. Set measurements-tick the box: area / mean gray value / integrated density / limit to threshold. 5. Use the multi-measure function to measure all the ROI together.

| Statistical analysis
All the quantitative data were presented as mean and standard deviation. After checking of normal distribution by Kolmogorov-Smirnov test, all parameters were analyzed by ANOVA and post hoc Turkey's HSD. For histological scoring, non-parametric Mann-Whitney U tests were used for comparisons between groups. The statistical analysis is calculated by SPSS (version 11; SPSS Inc, Chicago, IL, USA) and the level of significance is considered at p < 0.05.

| CONCLUSION
In conclusion, we have developed a method to produce secretomes