Secretome profile of TNF-α-induced human umbilical cord mesenchymal stem cells unveils biological processes relevant to skin wound healing

Despite a comprehensive understanding of the fundamental mechanisms of acute and chronic wound repair, persistent nonhealing wounds remain a cause of morbidity and mortality globally which impose a significant burden on society. Generally, the healing of cutaneous wounds requires an intricate cascade of phases to replace injured tissues and re-establish supportive structures. Such complex interactions involve the dermal and epidermal cells, the extracellular matrix (ECM) and the nervous and vascular components of the damaged and surrounding skin. However, when there is a failure of injured skin to proceed through an orderly and timely process and in conjunction with an underlying disease state, chronic wounds may occur [1,2]. The ideal strategy for wound management is aimed at accelerating healing and preventing scar formation. Despite huge advances in medical care and the variety of products available, there are still a large number of wound cases that do not recover and form lesions. Thus there is a clear need for the development of alternative wound therapies that improve healing and reduce scarring. Stem cell therapy has emerged as an attractive therapeutic option to improve tissue repair and treat severe skin disorders in clinical cases [3,4].

Mesenchymal stem cells (MSCs) are multipotent, self-renewing, culture-expandable cells that can be isolated from a variety of tissues. Over the years, MSCs have become one of the promising fields of research for cell-based therapies and regenerative medicine with regard to their unique immunomodulatory properties, proangiogenic characteristics and antiapoptotic activities. Moreover, MSCs can migrate to the site of injury and secrete trophic factors that mediate tissue repair and regeneration of damaged cells [5,6]. In the past decade, human umbilical cord-derived MSCs (hUCMSCs) have started to gain attention in clinical research largely due to the umbilical cord being a medical waste and the collection of umbilical cord being noninvasive, resulting in very minimal ethical concern. In addition, hUCMSCs possess low immunogenicity, which allows them to be utilized in allogeneic transplantation without any rejection in the host body [7].

MSC therapy has been shown to have great potential for the treatment of wounds by accelerating the healing process; it has been proposed that their direct incorporation into regenerating tissues is the main mechanism by which MSCs exert their beneficial effects [3,4]. Of late, evidence is suggesting that the therapeutic efficacy of MSCs is somehow associated with the paracrine effects of its bioactive molecules, which comprise cytokines, growth factors and exosomes, collectively known as the secretome [8,9]. The composition of the MSC secretome is greatly affected by the local microenvironment and the host’s inflammatory conditions and can be modulated by external stimuli and specific preconditioning, including cytokine treatments, hypoxia and 3D culture systems [10,11]. In response to particular stimuli, MSCs secrete plentiful trophic factors and convey regulatory messages to the recipient and immune cells, thereby regulating tissue homeostasis and coordinating tissue regeneration and repair [9,12]. Thus further exploration of the protein constituents of the MSC secretome is crucial to lay a strong foundation for a better understanding of the therapeutic roles of MSCs.

This study aims to identify the proteins secreted by hUCMSCs upon induction with TNF-α in mimicking the inflammatory condition and environment of the human host and to subsequently investigate the effects of TNF-α on hUCMSCs’ secreted proteins, and to explore the wound-healing properties of the TNF-α-induced hUCMSC secretome.

Materials & methods

hUCMSCs & HaCaT cell culture

hUCMSCs were isolated from umbilical cord samples of full-term, healthy babies with written consent from both parents, as previously described [13]. In brief, the umbilical cord blood was drained and the cord cut open lengthwise, and the cord tissues were shredded and enzymatically digested using collagenase (Worthington Biochemical, NJ, USA) for 1–2 h at 37°C. Next, the digested umbilical cord was washed twice with phosphate-buffered saline (PBS) before being cultured in low-glucose DMEM (Gibco, NY, USA) with 10% human serum and several antibiotics: 100 U/ml penicillin, 100 μg/ml streptomycin and 0.25 μg/ml amphotericin (Gibco). The cells were expanded in proprietary growth medium kept at 37°C in a 5% CO₂ and 95% air incubator. After 3 days, nonadherent cells were discarded and replaced with new growth medium until the cells reached 80% confluency. Then the hUCMSCs were subcultured until the required cell number was achieved. The culture-expanded cells were cryopreserved until further use in the following experiments.

The HaCaT keratinocyte cell line (Elabscience, TX, USA) was maintained under exponential growth in DMEM supplemented with 10% fetal bovine serum (Gibco) and 1% Glutamax™ (Gibco) at 37°C and in a humidified atmosphere containing 5% CO₂. The medium was changed every 2–3 days; cells were harvested with TrypLE™ (Gibco) upon reaching 80% confluency and reseeded at 2.5 × 10⁴/cm².

Characterization of hUCMSCs

The isolated hUCMSCs were characterized using immunophenotyping with a Human MSC Analysis kit (BD Biosciences, CA, USA) and tridifferentiation assays (adipogenesis, chondrogenesis and osteogenesis) as reported previously [14]. The full protocol for the immunophenotyping and tridifferentiation assays is provided in Supplementary Table 1.

Preparation of hUCMSC secretome

hUCMSCs from three different umbilical cords were seeded at 5 × 10⁵ cells/T175 flask and allowed to grow until they reached 80% confluency. Prior to the generation of the secretome, old growth medium was discarded and the cells were washed thoroughly three times with PBS to remove serum proteins. The cells were then cultured in basal medium (DMEM without phenol red; Gibco) with or without 250 U/ml recombinant human TNF-α (Miltenyi Biotec, Bergisch Gladbach, Germany). After 48 h of incubation, the supernatant of the culture or conditioned media (secretome) was collected and filtered with a 0.2-μm filter to remove all cellular debris. hUCMSCs cultured in DMEM without phenol red and in the absence of TNF-α served as the uninduced control, while hUCMSCs primed with TNF-α served as the induced secretome. The secretomes from all three biological replicates were stored at -80°C until they were ready to use for further analysis. A summary of the workflow adopted in this study is illustrated in Figure 1.

Protein concentration

The total protein concentration of the collected secretome was determined by bicinchoninic acid protein assay kit (Abcam, Cambridge, UK) according to the manufacturer’s instructions. Briefly, in a 96-well plate, bovine serum albumin standards were prepared accordingly and the secretome samples were diluted in a 1:5 ratio. Next, a working solution was added into each well and incubated for 30 min at 37°C. The optical density was measured at 562 nm. A standard curve graph was plotted to determine the protein concentration of the secretome.

In-solution protein digestion

Approximately 100 μg of total protein was resuspended in 100 μl of 50 mM ammonium bicarbonate (pH 8.0), followed by the addition of 100 μl Rapigest™SF, 0.1% (Waters Corporation, MA, USA). The protein was then concentrated in a speed vacuum concentrator for 20–30 min and incubated at 80°C for 15 min. Protein was reduced in the presence of 5 mM dithiothreitol at 60°C for 30 min before being alkylated with 10 mM iodoacetamide at room temperature for 45 min in the dark. Proteolytic digestion was initiated by adding trypsin at a ratio of 100:0.5 (protein to trypsin) and incubating at 37°C overnight. At the end of the digestion step, tryptic digestion and Rapigest activity were terminated with 1 μl of concentrated trifluoroacetic acid at 37°C for 20 min. The tryptic peptide solutions were kept at -80°C after being centrifuged at 14,000 r.p.m. for 20 min before liquid chromatography–tandem mass spectrometry (LC–MS/MS) analysis.

Nanoflow electrospray ionization LC–MS/MS analysis & protein identification

The samples were sent to National Institutes of Biotechnology Malaysia (NIBM) and analyzed using MS Dionex Ultimate 3000 RSLCnano setup with Orbitrap Fusion™ Tribrid™ mass spectrometer (Thermo Fisher Scientific, MA, USA), as previously described [15]. Briefly, an aliquot of 3 μl of the digested extract was loaded onto a C18 reversed-phase column (EASY-Spray Column Acclaim PepMap™ C18 100 Å, 2 μm particle size, 50 μm internal diameter × 15 cm) at 35°C. The sample was eluted in 0.1% formic acid in acetonitrile, with a gradient of 5–40% in 91 min, 2 min to 85%, 3 min at 85%, back to 5% in 1 min and 4 min at 5% at a flow rate of 250 nl/min. Each biological replicate of TNF-α-induced and uninduced hUCMSC secretome was run in three technical replicates. The peptide spectra were acquired using the orbitrap MS ranging from 310 to 1800 m/z. The chromatograms were analyzed using Thermo Scientific™ Proteome Discoverer™ Software Version 2.1 with the SEQUEST^® HT search engine.

Protein identification was carried out using Percolator^® where data were searched against a Uniprot^®Homo sapiens database and all peptides were validated using the Percolator algorithm, based on q-value with a 1% strict false discovery rate (FDR) and 5% relaxed FDR criteria [16]. Carbamidomethylation was selected as the fixed modification parameter and trypsin as the digestive enzyme with up to two missed cleavages. For protein quantification, the samples area was generated based on two unique and razor peptides which implied a greater confidence of protein identity.

Statistical analysis of proteins was performed using Perseus software v. 1.5.3.2 (Max Planck Institute of Biochemistry, Munich, Germany). Protein files with three technical replicates of three biological replicates in text format from Proteome Discoverer were uploaded to the Perseus software. The data were log2-transformed to stabilize the variance and scale normalized to the same mean intensity across the technical replicates. The mean for three biological replicates of induced and uninduced secretome was grouped in the same matrix and filtered for the valid values of at least two proteins, thus eliminating the proteins that were only present in one biological replicate. Finally, all biological replicates from all samples were grouped under the same matrix and missing values in the data, which represented low abundance measurement, were imputed with a random number drawn from a normal distribution. In addition, differentially expressed proteins between the induced and uninduced hUCMSC secretomes were detected using a t-test. The p-value was adjusted for multiple testing using a permutation-based FDR. A protein was considered to be significantly differentially expressed between two conditions with an adjusted p-value < 0.05 and a t-test difference ≤ -2.0 or ≥2.0.

Protein functional annotation & enrichment analysis

The identified proteins in TNF-α-induced and uninduced hUCMSC secretomes were validated using the Database for Annotation, Visualization and Integrated Discovery functional annotation tools (v. 6.8; https://david.ncifcrf.gov/) by importing their gene names to be searched against H. sapiens background. The Search Tool for the Retrieval of Interacting Genes/Proteins (v. 11.5) was used to map the proteins for interaction network visualization, and Create-A-Venn software (http://bioinformatics.psb.ugent.be/webtools/Venn/) was used to produce a Venn diagram that calculated the intersections to produce a graphical output by submitting the validated gene symbols for both groups’ proteomes. Gene ontology (GO) annotation and functional enrichment analysis were performed using FunRich (v. 3.13; www.funrich.org/) with the UniProt database for H. sapiens to identify the enriched biological processes annotated to the respective proteins.

In vitro scratch wound assay

For the following in vitro experiments, only the induced and uninduced secretome generated from one biological replicate were used. In a six-well plate, HaCaT keratinocytes were seeded at a density of 7.5 × 10⁵ cells/ well and were maintained at 37°C and 5% CO₂ for 48–72 h to permit cell adhesion and expansion until the formation of a confluent monolayer. These cells were then scored with a sterile pipette tip to leave a scratch of approximately 0.5–1.0 mm in width. The old medium was discarded and the cells were washed three times thoroughly with PBS to remove any dislodged cells and serum proteins. The wells were supplemented with 1 ml of either: blank DMEM without phenol red (control medium); TNF-α-induced secretome; or uninduced secretome.

The wound closure at 0, 24 and 48 h after the scratch was monitored. The images were captured with the EVOS™ cell imaging system (Invitrogen, CA, USA) and were then analyzed using ImageJ software (NIH, MD, USA) to measure the width of the scratch at the defined time points along its length; that is, at 0.5, 0.75 and 1.0 mm along the horizontal axis of the image (which equated to the left, middle and right of the field of view). After 48 h, the supernatants from all groups were collected for subsequent ELISA and Sircol™ collagen assay. The wound closure rates were calculated from four independent experiments, and the results were normalized with the 0-h time point.

Cell proliferation assay

The effect of the secretome on HaCaT cell proliferation was assessed using the (3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyltetrazolium bromide) tetrazolium (MTT) reduction assay. HaCaT cells were seeded in a 96-well plate at a seeding density of 2.5 × 10⁴ cells per well and incubated at 37°C and 5% CO₂ for 24 h to permit cell adhesion. The old medium was discarded and replaced with either: blank DMEM without phenol red (control medium); TNF-α-induced secretome; or uninduced secretome. After 48 h of incubation, MTT solution was added to each well at a final concentration of 0.5 mg/ml and incubated for 3 h. Then the media were removed and dimethylsulfoxide was added to dissolve the formazan product, followed by an absorbance reading at 570 nm using an Infinite^® M200 PRO microplate reader (Tecan, Männedorf, Switzerland). The relative percentage of viable cells at 48 h after the treatment was normalized to the DMEM without phenol red control well. The proliferation assay was performed in triplicate in three independent experiments.

Sircol collagen assay

Secreted collagen levels from the supernatant collected in three independent scratch wound assays were evaluated using a Sircol collagen assay kit (Bioclor, Carrickfergus, UK) according to the manufacturer’s instructions. Briefly, the Sircol reagent was mixed with the standards or samples (diluted to a 1:1 ratio with distilled water) to form collagen–dye complexes. After centrifugation, the supernatant was removed and ice-cold acid–salt wash reagent was layered to get rid of the unbound dye. Another centrifugation step was performed to remove the supernatant. Then an alkaline reagent was added to dissolve the bound dye, and the absorbance was read at 555 nm using the Infinite M200 PRO microplate reader. A standard curve graph was plotted to determine the collagen concentration in the samples. The assay was performed in duplicate.

ELISA multiarrays

The concentrations of cytokines and growth factors in the TNF-α-induced and uninduced secretomes and the supernatants collected from the scratch wound assay were analyzed by commercial ELISA multiarrays (R&D Systems, MN, USA), according to the manufacturer’s instructions. The cytokines and growth factors analyzed included MMP-13, TIMP-1, IL-10, IL-6, IGFBP-1, VEGF, FGFb, PDGF-BB, ANG-1 and TSP-2.

Statistical analysis

Statistical analysis was performed using SPSS Statistics v. 23.0 (IBM Corp., NY, USA) Quantitative data such as protein concentration, wound closure rate and ELISA multiarray results, from at least three technical replicates, were reported as mean ± standard deviation. Statistical significance was accepted at p < 0.05, p < 0.01, p < 0.001 and p < 0.0001.

Results

hUCMSCs fulfilled the minimal set of standard criteria defined by the International Society for Cellular Therapy

The Mesenchymal and Tissue Stem Cell Committee of the International Society for Cellular Therapy has released a set of minimal criteria that should be demonstrated for a cell to be referred to as a human MSC [17]. The guidelines state that MSCs must be: tissue culture plastic adherent; positive for surface markers CD105, CD73 and CD90, and negative for CD45, CD34, CD14 or CD11b, CD79α or CD19 and HLA-DR; and able to differentiate to adipocytes, osteoblasts and chondroblasts. In this study we showed that the isolated and expanded hUCMSCs were plastic adherent when maintained in standard culture conditions, and the cells appeared to be spindle-shaped and fibroblast-like (Figure 2A). Following this, immunophenotyping of hUCMSCs by flow cytometry demonstrated high expression levels of CD73, CD90, CD105 and CD44 surface markers (>99%) and low expression levels for negative markers including CD34, CD45, CD11b, CD19 and HLA-DR (Figure 2B), thereby eliminating the likelihood of contamination with hematopoietic cells. Subsequently, the in vitro ability of hUCMSCs to differentiate into adipocytes, osteoblasts and chondroblasts was confirmed (Figure 2C). The hUCMSCs were capable of differentiating into adipocytes, as indicated by the accumulation of neutral lipid droplets and vacuoles after staining with oil red O. For osteoblast differentiation, after 2–3 weeks of culture, the cells were driven to osteogenic differentiation and clearly showed an increase in calcium accumulation, as revealed by Alizarin Red S staining. Finally, for chondroblast differentiation, the cell pellets displayed chondrogenic properties, as evidenced by the Alcian blue staining that detects highly negatively charged glycosaminoglycans in the localized matrix of chondrocytes, and periodic acid Schiff that stains the lower levels of negatively charged glycosaminoglycans and collagen in the interterritorial matrix. The hUCMSCs thus fulfilled the minimal set of criteria defined by the committee and were used in the subsequent experiments.

Protein quantification & profiling of the TNF-α-induced & uninduced hUCMSC secretome

The total protein concentrations in the TNF-α-induced and uninduced secretomes were 1.81 ± 0.35 and 1.84 ± 0.36 mg/ml, respectively, as determined by bicinchoninic acid assay. The samples were tryptic digested and analyzed using Nanoflow electrospray ionization LC–MS/MS. When the search was done against the Uniprot H. sapiens database, with at least two unique peptides and a strict 1% FDR, a total of 506 and 480 proteins were identified in at least two out of three replicates in the induced and uninduced secretomes, respectively. Further validation of the protein lists to their respective gene symbols yielded 258 and 262 proteins in the TNF-α-induced and uninduced hUCMSC secretomes, respectively. The distinctive protein–protein interaction networks of the identified proteins for both conditions are shown in Supplementary Figures 1 & 2. Next, Venn diagram analysis identified 215 proteins that were shared between both secretomes, while 43 and 47 proteins were unique in the induced and uninduced secretomes, respectively (Figure 3A). When statistical analysis of the t-test of common proteins was visualized in the volcano plot based on the t-test difference between induced and uninduced hUCMSC secretomes, corresponding to the relative quantification (Figure 3B), it was found that eight different proteins were significantly over-represented in each secretome (Supplementary Table 2), making an overall number of 51 and 55 proteins that were differentially expressed in induced and uninduced hUCMSC secretomes, respectively.

GO analysis of TNF-α-induced & uninduced hUCMSC secretomes identifies proteins associated with inflammation, cellular development & tissue repair

Functional enrichment analysis was performed by submitting the gene names annotated to the respective proteins to FunRich (v. 3.13) with the UniProt database for H. sapiens to identify the enriched biological processes of total proteins in TNF-α-induced and uninduced hUCMSC secretomes. In the TNF-α-induced secretome, which comprised a total of 258 proteins, the top three GO enriched terms were inflammation-related and cellular development-related terms such as ‘inflammatory response’, ‘angiogenesis’ and ‘extracellular matrix organization’ (Figure 3C). As a large number of proteins in the uninduced secretome were also present in the induced secretome, it was not surprising to find that the enriched biological processes in the uninduced secretome were associated with cellular development and differentiation, including ‘extracellular matrix organization’ followed by ‘positive regulation of fibroblast proliferation’ and ‘acute-phase response’ (Figure 3D).

To further elucidate the diversity between the secretomes produced under different conditions, GO analysis was also performed on the differential proteins that were uniquely present in the induced and uninduced secretomes. Of the 51 unique proteins in the induced secretome, the biological processes were mainly involved in inflammation-related and tissue repair-related functions, including inflammatory response (14.6%), immune response (12.5%), chemotaxis (8.3%), angiogenesis (6.3%), chemokine-mediated signaling pathway (6.3%), leukocyte chemotaxis (6.3%), monocyte activation (4.2%), positive regulation of macrophage differentiation (4.2%) and regulation of several cytokines such as IL-12 (2.1%) and IL-6 (2.1%) (Figure 4A). In particular, these enriched functions are associated with the scarless wound healing process. Processes related to cell adhesion (12.5%) and ECM organization (10.4%) were also detected. On the other hand, the enriched biological processes identified in the 55 unique proteins of the uninduced secretome were found to be related to the inherent properties of MSCs that are important for stem cell maintenance and differentiation, such as mitotic cell cycle (7.7%), receptor-mediated endocytosis (7.7%) and osteoblast differentiation (5.8%) (Figure 4B).

The TNF-α-induced secretome contained higher levels of growth factors & cytokines

To further dissect how an inflammatory microenvironment caused by TNF-α affects the secretion of soluble factors by hUCMSCs, both the induced and uninduced secretomes were assayed by ELISA multiarray. Among the analyzed markers, the results showed significantly increased levels of FGFb and PDGF-BB, inflammatory protein IL-6 and angiogenic factor VEGF in the TNF-α-induced secretome compared with the uninduced secretome (Figure 5). Under normal microenvironments, these biomarkers are released by hUCMSCs at a lower level. After priming with TNF-α for 48 h, the levels of FGFb, PDGF-BB, IL-6 and VEGF were elevated from 30.6 to 50.3 pg/ml, 10.7 to 14.3 pg/ml, 996.6 to 2529.5 pg/ml and 61.9 to 156.0 pg/ml, respectively. In contrast, the levels of other upregulated factors such as MMP-13, TIMP-1, IL-10, IGFBP-1, Ang-1 and TSP-2 were comparable in both induced and uninduced secretomes (data not shown).

The hUCMSC secretome, specifically the TNF-α-induced secretome, significantly accelerated wound closure in HaCaT keratinocytes

To correlate the protein profiling and multiarray results, we experimentally demonstrated the wound healing properties of the secretome in an in vitro scratch assay using HaCaT cells (Figure 6). Representative images of wound closure at 0, 24 and 48 h for three different groups are shown in Figure 6A. In the presence of the induced secretome, the migration rate of HaCaT cells was significantly increased as compared with the control medium. At the 24-h time point, the wound closure rates of HaCaT cells were 13% (± 7%) in the control medium and 39% (± 4%) in the presence of induced secretome (p < 0.001). Similarly, at the 48-h time point, the wound closure rates of HaCaT cells were 17% (± 8%) in the control medium and 55% (± 7%) with the induced secretome (p < 0.001) (Figure 6B).

Conversely, at the 24-h time point, the uninduced secretome-treated group did not show a significant rate of wound closure (24 ± 10%), but at the 48-h time point, the wound closure rate was significantly higher (34 ± 9%), as compared with control medium (Figure 6B). More importantly, the wound closure at 48 h was observed to be notably greater in the induced secretome-treated group as compared with the uninduced secretome group. In short, the results demonstrated accelerated migration and wound closure of HaCaT cells when treated with the induced secretome.

To examine the probable role of the secretome in stimulating cell proliferation during wound closure, an MTT assay was performed to determine the viable number of HaCaT cells in different conditions. There was no significant increase in the HaCaT cell number at 48 h in the induced secretome and uninduced secretome groups versus control (Supplementary Figure 3A). The results thus demonstrated that the wound closure observed was not due to HaCaT cell proliferation in situ. We also did not observe noteworthy effects of the secretome on collagen formation in the wound healing assay (Supplementary Figure 3B).

The hUCMSC secretome further stimulated the release of inflammatory growth factors by HaCaT cells during the wound healing & repair process

The resulting supernatants of the in vitro scratch assay were collected, and quantification of specific protein factors associated with wound healing was carried out by an ELISA multiplex assay. The concentrations of key proteins including FGFb, PDGF-BB, IL-6, VEGF and MMP-13 were significantly elevated in both of the secretome-treated groups as compared with the control-treated group, with the highest level detected in the TNF-α-induced secretome-treated group (Figure 7). Even though these proteins were initially present in the hUCMSC secretome (Figure 5), further release of PDGF-BB, VEGF and MMP-13 by HaCaT cells during wound closure was observed. In the scratch assay, on treatment of HaCaT cells with the induced secretome, elevated levels of PDGF-BB (from 14.4 to 259 ng/ml), VEGF (from 156.04 to 5488.6 ng/ml) and MMP-13 (from 1032.4 to 52326.6 ng/ml) were detected. A similar observation was seen in the uninduced secretome-treated group, whereby the levels of PDGF, VEGF and MMP-13 increased from 10.7 to 90.1 ng/ml, from 61.9 to 3402.7 ng/ml and from 1449.3 to 25426.3 ng/ml, respectively. Collectively, in response to the scratch injury, the hUCMSC secretome further stimulated the release of key factors by HaCaT cells to facilitate wound closure. Specifically, in the induced secretome-treated group, the highest protein levels detected correlated with the fastest wound closure, thereby confirming their essential roles in enhancing and coordinating the scarless wound healing and repair process.

Discussion

Impaired cutaneous wound recovery leading to chronic wounds remains a significant medical issue. In view of that, interest has arisen tremendously in the clinical translation of the MSC secretome for wound repair. The MSC secretome, loaded with trophic factors and bioactive molecules, is capable of providing a regenerative microenvironment that influences the molecular activities during tissue repair to facilitate wound recovery [9]. The present work examined the protein profiling of the hUCMSC secretome, generated under a normal and an inflammatory environment, and demonstrated its therapeutic effect in accelerating wound closure.

Database interrogation and functional enrichment analysis of the uninduced hUCMSC secretome revealed proteins that reflect the characteristics of inherent MSCs for self-renewal and lineage-directed differentiation. In the culturing of hUCMSCs to mimic the natural microenvironment, GO analysis disclosed that a large proportion of the secreted proteins are involved in the modulation of ECM organization. The ECM not only provides structural supports for cell attachment, but also forms a stem cell microenvironment, or niche, which is essential for cell proliferation, cell–cell communication, cell survival and preservation of stemness [18,19]. In addition, proteins related to ECM organization are shown to direct cell behavior and cell fate by modulating stem cell differentiation and preventing cell death [20,21].

GO enrichment analysis also discovered a group of proteins related to the acute-phase response – prominent systemic reactions of cells in response to changes in the local microenvironment [22,23]. During the preparation of the uninduced secretome, the culture condition was shifted from a serum-inclusive to a serum-free microenvironment, in which only a minimal amount of essential amino acids were supplied to avoid any discrepancy caused by the serum. As a result, the local homeostasis of hUCMSCs was slightly affected by the reduction of nutrients and a change in cell metabolism. In response to the mild disturbances, the hUCMSCs secreted additional proteins and enzymes that were related to representative GO annotations of the acute-phase response to restore and maintain cell homeostasis. These secreted proteins are mediators and inhibitors of inflammation and have beneficial roles in healing, ECM remodeling and adaptation to stimulus [24,25].

Other GO terms that were found to be functionally enriched in the uninduced hUCMSCs were related to important biological processes that maintain cell homeostasis and regulate cellular development. Our results were consistent with previous findings by Shin et al., who analyzed the secretome from four diverse human MSC sources (bone marrow, adipose tissue, placenta and Wharton’s jelly) [26]. They reported on the functional analyses of the respective MSCs’ secretome profiles and revealed that despite their different origins, the MSC secretomes shared similar functional characteristics, including ECM organization, cellular homeostasis, cell migration, cellular development, cell proliferation, antiapoptosis and metabolic processes, which are compatible with our data. Their study further showed that MSC secretomes from various sources, with distinct protein profiling, eventually led to the same physiological functions through different pathways, thereby suggesting that the therapeutic functions of MSCs are related to the paracrine factors in the secretome [26].

Considering the inherent interaction between immune cells and MSCs in areas of tissue damage or inflammation, it is becoming crucial to boost the capacity of MSCs to effectively regulate and induce beneficial immune responses following an injury. Plenty of studies have evolved into studying and characterizing MSC secretomes stimulated under different settings to mimic acute and chronic inflammatory conditions. These studies have suggested that priming of MSCs appears to be an effective strategy to enhance their therapeutic effects in association with vital processes involved in tissue regeneration, including but not limited to fibrosis, immunomodulation, angiogenesis and stimulation of progenitor cells [9]. TNF-α plays a crucial pathogenetic role as the central proinflammatory cytokine; a high level of TNF-α has been linked to various inflammatory conditions and diseases, including trauma, sepsis, infection, rheumatoid arthritis, cardiovascular disease, respiratory disease, renal disease and others [27]. In the present study, hUCMSCs were exposed to TNF-α to mimic the proinflammatory conditions of the human host, and to prime the cells to produce respondent biomolecules that are useful in an inflammatory environment. Apart from ECM-related terms, GO enrichment analysis of the TNF-α-induced secretome revealed the majority of the secreted proteins to be functionally annotated in the inflammatory response and tissue repair. A large number of these cytokines and chemokines were discovered to potentially facilitate the wound healing process – and specifically the inflammatory, proliferative and remodeling phases – which is characterized by local vasodilation, extravasation of plasma into the intercellular area, accumulation of white blood cells, filling and contraction of the wound margin, generation of new vessels, reorganization of collagen fibers and epithelialization [28].

Additionally, we analyzed the GO of enriched biological processes of unique proteins in both the uninduced and induced secretomes. In the uninduced secretome, the differential proteins were mainly annotated for hUCMSCs’ traits in the regulation of cell growth, metabolism and fate commitment; for example, mitotic cell cycle, osteoblast differentiation, positive regulation of apoptotic cell clearance, and cell redox homeostasis. Interestingly, GO analysis also discovered the regulation of TGF-β activation, which plays a key role in the recruitment of stem/progenitor cells to participate in the tissue regeneration and remodeling process [29].

On the other hand, GO analysis indicated that the differential proteins in the TNF-α-induced hUCMSC secretome were functionally annotated to inflammatory response, ECM and tissue development as well as angiogenesis. Upon exposure to TNF-α at the site of injury, hUCMSCs modulate the innate immune response through the integration of multiple intercellular signals of cytokines and chemokines, to initiate the wound healing process and preserve homeostasis [30,31]. Consequently, chemokine-mediated signaling pathways are activated, followed by immune-cell infiltration, cell adhesion and activation [32,33]. Angiogenesis is the formation of new blood vessels which occurs immediately after tissue injury. This process is often regulated by the ECM microenvironment via modulation of integrin receptor expression that enables the ECM to play a crucial role in cell–cell adhesion to the ECM for new blood vessel growth [34,35]. Overall, this enriched function of the secretory proteins reflects the main events involved in the wound healing process, thus suggesting the role of the hUCMSC secretome in the healing of injured tissues and maintaining proper tissue homeostasis and function in response to inflammation.

The secretome was generated from serum-free hUCMSC cultures, thus eliminating any discrepancy caused by serum [36] and at the same time allowing the identification of specific proteins released by hUCMSCs in response to inflammation. Subsequently, we dissected the specific cytokines and growth factors in the inflammation microenvironment induced by TNF-α using ELISA and detected significantly high concentrations of FGFb, PDGF-BB, IL-6 and VEGF, as compared with the secretome generated under normal conditions. These factors were also previously detected in the secretome of MSCs from bone marrow [37] and adipose tissue [38–40], with proven functional application in angiogenesis, preventing tissue fibrosis and inflammation, and various phases of wound healing [26,41].

To corroborate our hUCMSCs’ TNF-α-induced secretome with previous findings, a direct wound healing assay was carried out to study the effects of both uninduced and TNF-α-induced secretomes on scratch-induced injury in HaCaT keratinocytes. HaCaT is an immortalized cell line exclusive of potential donor variations, making it a suitable in vitro model to study the effects of the hUCMSC secretome on skin cell behavior [42]. We have shown that the uninduced hUCMSC secretome significantly promoted scratch wound closure in HaCaT keratinocytes, similar to findings reported previously in human dermal fibroblasts and epidermal keratinocytes [40,43–45]. Furthermore, HaCaT cells treated with the TNF-α-induced hUCMSC secretome had a superior wound closure rate as compared with those treated with the uninduced secretome and noninduced control medium. The reason is attributed to the high level of key proteins – specifically FGFb, PDGF-BB, IL-6 and VEGF – in the TNF-α-induced secretome, which provides an enhanced microenvironment to facilitate the migration and other molecular activities during the wound healing process. Our results further demonstrated that the secretome did not affect HaCaT cell viability, suggesting that the wound closure is primarily a migratory effect driven by the hUCMSC secretome and is independent of HaCaT cell proliferation. These observations were consistent with a previous study which concluded that MSC-conditioned medium improved wound closure by increasing the cell migration rate instead of inducing cell proliferation or differentiation to replace the damaged cells in wounded tissues [43].

Subsequently, we investigated the expression levels of the biomarkers released during the wound closure process. In response to the scratch injury, the growth factors and cytokines in the secretome stimulated HaCaT cells to further release the essential proteins to facilitate the wound migration and closure. Given that the HaCaT cells treated with the TNF-α-induced hUCMSC secretome had the fastest wound closure rate, it was not surprising that we detected the highest levels of key proteins in this sample. Apart from FGFb, PDGF-BB, IL-6 and VEGF, we also identified a significantly high level of MMP-13 in the supernatant. Under normal conditions, matrix metalloproteinases are usually minimal, while in injured conditions, they are actively stimulated [46], suggesting that MMP-13 activity is important in angiogenesis and keratinocyte migration and re-epithelialization during wound repair [47,48].

Based on our data and previously published literature, we illustrated the functions of the key factors at various stages of wound healing. During an injury that results from cut, abrasion, surgical incision, ulceration or severe burn, PDGF-BB, consisting of two B subunits, is important to recruit the first inflammatory cells, neutrophils, to the injured site, followed by the chemotaxis of macrophages, fibroblasts and smooth-muscle cells [49,50]. Then, VEGF and IL-6 stimulate vasodilation and vascular permeability, thus facilitating the recruitment and infiltration of immune cells, such as additional neutrophils, macrophages, mast cells and T lymphocytes, to the wound area to degrade and remove bacteria and foreign particles [51]. As an endothelial cell-specific mitogen, VEGF maintains the proliferative capability and delays senescence in endothelial cells to aid endothelialization and neovascularization [52]. On the other hand, IL-6 is an important key regulator of inflammation and timely resolution of wound healing, specifically in the differentiation of helper T cells, progression into the proliferative phase, growth and migration of keratinocytes and induction of VEGF and TGF-β1 secretion, which are crucial for angiogenesis and collagen deposition, respectively [29].

At the proliferation stage of wound healing, granulation tissues are structured to fill the wound, by which additional cells are attracted and migrated to the wound area, including MSCs, fibroblasts, endothelial cells and smooth-muscle cells [50,51]. This is when soluble factors such as VEGF, PDGF-BB and FGFb promote endothelialization and neovascularization, epithelial cell migration and proliferation, ECM re-establishment and epithelialization of the wound surface [50,52,53]. FGFb, with mitogenic and chemoattractant properties, also modulates the wound bed migration (endothelial cells, fibroblasts and keratinocytes) and stimulates collagen production, deposition and degradation [28]. Concomitantly, these factors are accountable for the angiogenesis surrounding the wound to provide adequate oxygen, nutrients and mediators to the granulation tissues and simultaneously remove the unwanted metabolic waste [52,53].

The last stage of wound healing, the remodeling phase, involves the transition of granulation tissue to scar, whereby replacement with type I collagen, wound contraction, ECM reorganization and angiogenesis reduction take place. MMP-13 plays a pivotal role in regulating the breakdown of granulation tissue by modulating myofibroblasts’ function to secrete ECM molecules, regulate proteolysis to degrade the matrix and affect wound contraction [46,48,54]. Along with the closure of the wound and decrease in the number of vessels, the granulation tissue is eventually replaced by a scar. Meantime, FGFb can induce apoptosis of granulation tissue and intervene in the formation of myofibroblasts, which prevents the onset of chronic inflammation that drives the development of keloids and hypertrophic scars [55–57].

A previous study has demonstrated the effect of conditioned medium from UCMSCs (UCMSCs-CM) in accelerating dermal fibroblast migration and proliferation in vitro. Intradermal administration of UCMSCs-CM also enhanced healing with lesser scars in mice models with excisional wounds when compared with sham controls [45]. In another study, UCMSCs-CM, which was loaded with VEGF, PDGF and KGF, significantly accelerated wound closure in mice with diabetic wounds [58]. Apart from that, the therapeutic effect of secretomes from various sources of MSCs (adipose tissue, bone marrow, umbilical cord, amniotic membrane and Wharton’s jelly) in in vivo animal models with skin burns, excisional wounds, diabetic wounds and radiation dermatitis – and in clinical trials of skin scar, ablative fractional laser, diabetic foot ulcers, bullosis diabeticorum, epidermolysis bullosa and chronic plantar ulcers in leprosy – have been summarized in review papers. These studies collectively reported on the efficacy of MSC secretomes in accelerating wound healing and inducing skin regeneration in various wounded animal models and human samples, as compared with control groups [41,49,53].

Additionally, it is inspiring that one of the main factors found in the hUCMSC secretome, human PDGF-BB, is an approved growth factor therapy. In 2005 recombinant PDGF-BB was the first growth factor permitted by the US FDA (under the name becaplermin) for topical administration in the treatment of lower-extremity diabetic neuropathic ulcers [59]. In the respective phase III randomized, placebo-controlled, double-blind study, a topical gel formulation of recombinant PDGF-BB (becaplermin) or placebo, in combination with standardized good wound care, was given to patients with type 1 or type 2 diabetes with chronic ulcers. The application of becaplermin gel significantly increased the frequency of complete wound closure, with a shorter duration to achieve complete closure of diabetic neuropathic ulcers, without causing any severe side effects [60,61]. This accomplishment, together with numerous other published studies, has paved the way for future therapeutic application of other potential recombinant growth factors, and probably the hUCMSC secretome that comprises a combination of proteins, in the treatment of various diseases.

We boldly speculate that the hUCMSCs’ induced secretome, immersed in a sterile medical gauze and directly applied to open wounds, can promote recovery. This is likely because a moist environment favors the exchange of active soluble molecules, including the secreted growth factors, thereby enhancing wound healing [62]. However, medical gauze requires frequent renewal, which incurs additional cost, time and use of equipment. As technology advances, other types of wound dressings have been introduced, and the nanopeptide hydrogel system is one of the potential delivery tools for secretome application. Hydrogel polymers display similar properties to ECM, acting like a tissue structure to control the distribution and sustained release of bioactive molecules [63,64]. A previous study has reported that MSC-conditioned medium encapsulated into nanoparticle hydrogels had increased metabolic activities, without losing its functional properties [65]. Applying the same concept, our hUCMSC secretome, or the secreted growth factors, could be integrated into a nanopolypeptide hydrogel to facilitate protein delivery and elongate secretome residence duration at the wound area, thereby promoting the therapeutic effects. Nevertheless, further studies are required to examine the hydrogel composition, type of polypeptides, cell infiltration and distribution and the incorporation of secretome under the optimized conditions.

Conclusion

Scarless wound healing is important for a complete recovery in patients who have severe medical conditions such as major burns, surgical incisions and trauma. In the past decade, increasing evidence has shown the roles of paracrine factors in contributing to MSCs’ therapeutic properties under different conditions. In this study we have established the secretome profiling of hUCMSCs in an inflammatory state activated by TNF-α. The proteome secreted in response to TNF-α was found to be mainly involved in immunomodulation as an attempt to redress the inflammatory condition set forth by TNF-α. Further analysis disclosed a list of candidate proteins associated with inflammation, angiogenesis and ECM remodeling, which contribute to promoting the therapeutic effects of the hUCMSC secretome. These proteins have also been reported previously to have antifibrotic (antiscar) effects while promoting and/or accelerating acute and chronic wound healing. We then showed that the TNF-α-induced secretome rapidly enhanced wound closure with potentially scarless outcomes, consistent with previously reported data. A further detailed investigation is warranted to validate the present data on the role of the cytokine-induced hUCMSC secretome in promoting tissue regeneration to access its future clinical usefulness. Nevertheless, our findings are able to serve as a comprehensive basis to dissect the paracrine effects of hUCMSCs in prospective functional studies and to provide an essential prerequisite for the development of efficient and safe stem cell-based therapies.