Proteomic Analysis Reveals Commonly Secreted Proteins of Mesenchymal Stem Cells Derived from Bone Marrow, Adipose Tissue, and Synovial Membrane to Show Potential for Cartilage Regeneration in Knee Osteoarthritis

Paracrine factors secreted by mesenchymal stem cells (MSCs) reportedly modulate inflammation and reparative processes in damaged tissues and have been explored for knee osteoarthritis (OA) therapy. Although various studies have reported the effects of paracrine factors in knee OA, it is not yet clear which paracrine factors directly affect the regeneration of damaged cartilage and which are secreted under various knee OA conditions. In this study, we cultured MSCs derived from three types of tissues and treated each type with IL-1β and TNF-α or not to obtain conditioned medium. Each conditioned medium was used to analyse the paracrine factors related to cartilage regeneration using liquid chromatography-tandem mass spectrometry. Bone marrow-, adipose tissue-, and synovial membrane-MSCs (all-MSCs) exhibited expression of 93 proteins under normal conditions and 105 proteins under inflammatory conditions. It was confirmed that the types of secreted proteins differed depending on the environmental conditions, and the proteins were validated using ELISA. The results of Gene Ontology and Kyoto Encyclopedia of Genes and Genomes pathway analysis using a list of proteins secreted by all-MSCs under each condition confirmed that the secreted proteins were closely related to cartilage repair under inflammatory conditions. Protein-protein interaction networks were confirmed to change depending on environmental differences and were found to enhance the secretion of paracrine factors related to cartilage regeneration under inflammatory conditions. In conclusion, our results demonstrated that compared with knee OA conditions, the differential expression proteins may contribute to the regeneration of damaged cartilage. In addition, the detailed information on commonly secreted proteins by all-MSCs provides a comprehensive basis for understanding the potential of paracrine factors to influence tissue repair and regeneration in knee OA.


Introduction
Mesenchymal stem cells (MSCs) can be easily obtained from different cell sources such as bone marrow (BM), adipose tissue (AT), and synovial membrane (SM). They have selfrenewal and trilineage differentiation potential [1][2][3][4]. MSCs are also known to secrete various paracrine factors including cytokines, chemokines, growth factors, and extracellular vesicles. Paracrine signalling is a form of cell-to-cell communication in which a cell produces a signal to induce changes in nearby cells, altering the behaviour or differentiation of those cells. Paracrine factors secreted by MSCs induce sur-rounding cells to differentiate into mature cell lines and regulate tissue inflammation or recovery processes [5]. Thus, there is increasing evidence that MSCs play a role in regenerating damaged tissue through the paracrine effect on surrounding cells [2,6,7]. Owing to these characteristics, MSCs are widely studied for the treatment for various diseases. According to the official database of ClinicalTrials.gov, 630 MSC-based clinical trials have been reported, and 10% or more of them are actively being conducted to assess the potential of MSCs for knee osteoarthritis (OA) therapy.
Knee OA is a degenerative disease caused by various factors, including abnormal mechanical stress, ageing, obesity, and genetic factors, and results in inflammation in joints and degradation of cartilage tissue [8]. The cartilage of the knee joint where knee OA occurs is difficult to regenerate after injury due to the nature of cartilage; therefore, therapeutic treatments are limited to conservative care or artificial joint surgery [9,10]. To overcome these therapeutic limitations and augment cartilage regeneration, various tissue-derived MSCs are being used for the development of cell therapy for knee OA treatment [11][12][13][14]. According to the results of recent studies, injecting MSC-conditioned medium (CM) into the anterior cruciate ligament transection rat model is sufficient to mediate cartilage regeneration function [11,[15][16][17]; thus, interest in efficacious knee OA treatments using MSC-derived paracrine factors has increased [14,[18][19][20][21]. However, although several studies have actively investigated paracrine factors secreted by each type of MSC obtained from various tissues under general culture conditions [22][23][24], it is still unclear which paracrine factors affect the regeneration of damaged cartilage.
To elucidate the therapeutic role of paracrine factors secreted by MSCs, it is necessary to identify the factors that are commonly secreted by MSCs under various environmental conditions, since each type of tissue-derived MSC secretes different paracrine factors [24][25][26]. However, little is known about how environmental conditions similar to knee OA affect paracrine factors [27,28]. Therefore, in the present study, we aimed to investigate the major paracrine factors related to cartilage regeneration by identifying the differences in the types and characteristics of commonly secreted proteins by BM-, AT-, and SM-MSCs under various conditions.

Materials and Methods
2.1. Isolation and Culture of Human BM-, AT-, and SM-MSCs. The study was approved by the Institutional Review Board of Yonsei Sarang Hospital (IRB number: YSSR 2020-09-001), and informed consent was obtained from all donors.

Stem Cells International
HyClone) and the undigested tissues were removed using a 70 μm strainer (BD Biosciences, San Diego, CA, USA). The cells obtained by centrifugation at 645 × g for 5 min were plated at a density of 5,000 cells/cm 2 in a 75 cm 2 culture flask with complete medium [29]. Human SM (n = 3) were harvested from the knees of patients with OA during total knee arthroplasty. The tissue was placed on a petri dish and finely minced using sterile scissors. The minced tissue was digested with 0.25% Trypsin-EDTA (HyClone) at 37°C for 30 min. After digestion, 0.3% of collagenase type 1 was added and gently shaken at 37°C for 90 min. Undigested tissues were removed using a 40 μm strainer (BD Biosciences), and cells were obtained by centrifugation at 645 × g for 5 min. The obtained cells were plated at a density of 5,000 cells/cm 2 in a 75 cm 2 culture flask with complete medium [30,31].
All cells used in the experiment were cultured in a humidified incubator at 5% CO 2 at 37°C. The medium was changed every 2-3 days, and cells were subcultured at 80-90% confluence following treatment with 0.25% Trypsin-EDTA at 37°C for 5 min. The cells were washed and harvested by centrifugation at 645 × g for 5 min and then plated at a density of 5,000 cells/cm 2 [32,33]. Passage 4 cells were used for all cell types.

Preparation of CM.
The BM-, AT-, and SM-MSCs were maintained under two conditions based on environmental differences: (i) normal condition: without inflammatory cytokines (tumour necrosis factor-α (TNF-α; PeproTech) and interleukin-1β (IL-1β; PeproTech)), and (ii) inflammatory condition: treated with 10 ng/mL of TNF-α and 10 ng/mL of IL-1β [38][39][40][41]. To prepare CM, cells were cultured in a 75 cm 2 culture flask in α-MEM containing 10% FBS. When MSCs were approximately 80-90% confluent at passage 4, the cells were switched to serum-free α-MEM for overnight incubation and treated with conditions (i) or (ii) for 6 h. Cultures were refed with 15 mL of serum-free α-MEM and incubated for 48 h. Here, we used serum-free medium to avoid interference from albumin-enriched FBS. Then, the CM was collected and filtrated through a 0.2 μm filter to remove cellular debris. Thereafter, the CM was concentrated 25-fold using ultrafiltration units with 3 kDa cutoff filters (Amicon Ultra; Merck Millipore, Watford, UK) at 4000 × g for 1 h [42]. The protein concentration of the CM was measured using the Pierce BCA Protein Assay Kit (Thermo Fisher Scientific, Waltham, MA, USA) according to the manufacturer's instructions.
2.5. In-Solution Digestion. Briefly, 100-200 μg of protein was denatured using 8 M urea in 50 mM heavy carbonate ammonium buffer (pH 7.8) and allowed to react at room temperature for 3 h, followed by reduction using 10 mM dithiothreitol for 2 h at room temperature. The proteins were alkylated with 10 mM iodoacetamide in the dark at room temperature for 1 h and then diluted more than 10-fold using 50 mM ammonium bicarbonate solution. In-solution digestion was carried out by adding trypsin to the protein solution with an enzyme-to-protein ratio of 1 : 50 (w/w) at 37°C for 18 h. Finally, formic acid was added to each sample to stop the reaction. Samples were stored at -80°C until further analysis [43].  The processing workflow consisted of the following nodes: two maximum missed cleavages, peptide length range of 6-144 amino acids, precursor mass tolerance of 10 ppm, fragment mass tolerance of 0.02 Da, cysteine carbamidomethylation as a static modification, and oxidation as a dynamic modification. Peptide validation settings were identified using a target false discovery rate (FDR; strict) for peptide-spectrum match (PSM) of 0.01, target FDR (relaxed) for PSM of 0.05, and peptide filter confidence of at least high level. The identified proteins were associated with Gene Ontology (GO) terms to determine their biological and functional properties. The three main types of annotations, namely, cellular components (CC), molecular functions (MF), and biological processes (BP), were obtained from the GO website at http://www.geneontology.org.
Protein pathways were generated using Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis. In addition, the identified proteins were entered into the STRING database (https://string-db.org/) to predict and visualise the proteinprotein interactions (PPI) under various environmental conditions. A representative network was obtained with high confidence in data settings with a minimum interaction score of 0.7.
2.8. Enzyme-Linked Immunosorbent Assay. Cell culture supernatants were collected as described in the preparation of CM. The concentration of proteins in CM was assessed using a commercially available enzyme-linked immunosorbent assay (ELISA) kit (RayBiotech, GA, USA) according to the manufacturer's instructions [45]. Assay kits for the following proteins were used: human thrombospondin 2 (TSP-2) and human TNF-inducible gene 6 (TSG-6). All measurements were performed in duplicate.  (Figure 1(a)). We performed in vitro differentiation assays to confirm the trilineage differentiation potential of each MSC type. As a result, each MSC type differentiated into adipocytes, osteoblasts, and chondrocytes when cultured in the appropriate differentiation medium (Figure 1(b)). Flow cytometry analysis was performed to confirm that MSCs expressed cell surface antigens. In each MSC type, the positive markers CD73, CD90, and CD105 were expressed at 95% or higher, and the negative markers CD14, CD34, CD45, CD79a, and HLA-DR were expressed at less than 2% (Figure 1(c)). These results demonstrated that the human tissue-derived cells used in the present study had typical MSC characteristics (Supplementary Figure 1).

Commonly Secreted Proteins among the Three Types of
MSCs. LC-MS/MS (n = 3 per tissue source) was performed to identify the proteins secreted by each MSC type under normal or inflammatory conditions (Figure 2(a)). We con-  Table 1).

GO Analysis.
GO analysis was performed to confirm the functional differences between commonly secreted proteins by all-MSCs under normal and inflammatory conditions. As a result, in MF, the percentage of "protein binding" of the secreted proteins under both conditions was confirmed to be 88% or more, while the percentages of other categories were 35% or less (Figure 3(a)). In CC, the percentage of "extracellular" of the secreted proteins under both conditions was confirmed to be 73% or more, and that of the other categories was 50% or less (Figure 3(b)). In BP, the percentage of secreted proteins under both conditions including "regulation of biological process," "response to stimulus," "metabolic process," and "cell organization and biogenesis" was confirmed to be approximately 52% or more, with the percentages of these categories being higher under inflammatory conditions than under normal conditions. The percentages of other categories were confirmed to be 40% or less (Figure 3(c)).

KEGG Pathway
Analysis. KEGG pathway analysis was conducted to identify signalling pathways related to proteins commonly secreted by all-MSCs under normal and inflammatory conditions. We confirmed that the pathway most closely related to proteins secreted under both conditions was "signal transduction." The main pathways were similar under both conditions, but the levels of the secreted proteins were higher under inflammatory conditions than under normal conditions (Figure 4).

Changes in PPI Networks Depending on Environmental
Differences. PPI network analysis was performed using the STRING database to confirm how the interaction networks of commonly secreted proteins by all-MSCs were altered because of environmental differences. Thus, excluding unconnected nodes, the PPI network under normal condition was composed of 102 nodes and 368 edges. The PPI network under inflammatory condition was composed of 92 nodes and 309 edges (interaction score > 0:7), confirming three main clusters under each condition. Under normal conditions, "extracellular matrix (ECM) organization" constituted 13 proteins, primarily including the collagen family, and "regulation of cell differentiation" constituted 29 proteins, including proteins such as insulin-like growth factorbinding protein 3/4/7, gelsolin, and stanniocalcin-2. In addition, "cellular component organization" constituted 5 proteins, including vinculin (VCL) (Figure 5(a)). Under the inflammatory condition, "ECM organization" constituted 10 proteins, including those from the collagen family, and "protein metabolic process" constituted 5 proteins, including matrix metalloproteinase (MMP). Moreover, "response to stimulus" constituted 12 proteins including C-X-C motif chemokine 2/3/6/8 and annexin A1 (Figure 5(b)).

Identification of Proteins with Tissue Regeneration
Potential. We identified proteins related to tissue regeneration among the commonly secreted proteins by all-MSCs under normal and inflammatory conditions. The categories affecting tissue regeneration were classified as anti-inflammation, anti-apoptosis, ECM-cell interactions, homeostasis, 10 Stem Cells International inhibition of MMPs, and regeneration of chondrocytes. Proteins related to tissue regeneration were found to be secreted under all or specific environments. Except for the category of ECM-cell interactions, the number of secreted proteins was larger under inflammatory conditions than under normal conditions. We identified that proteins such as TSG-6 and thrombospondin 1 (TSP-1) are secreted only under inflammatory conditions ( Table 2).

Validation of Commonly Secreted Proteins by Three Types
of MSCs under the Two Environmental Conditions. ELISA was performed to validate the proteins presented in Table 2. We confirmed that TSP-2 was secreted under both environmental conditions, whereas TSG-6 was secreted under inflammatory conditions. The expression level of TSP-2, confirmed to be secreted under the two conditions, was higher under the normal condition than under the inflammatory condition. However, the concentration of TSP-2 in the CM was not significantly different between the normal and inflammatory conditions (Figure 6(a)). The expression level of TSG-6 was surprisingly 15 times higher under the inflammatory condition than under the normal condition ( Figure 6(b)). The control was serum-free medium, and no protein was detected in it.

Discussion
We investigated proteins predicted to exhibit cartilage regeneration potential by confirming the differences and characteristics of proteins commonly secreted by human BM-, AT-, and SM-MSCs, depending on environmental differences. Thus far, cartilage repair studies using MSCs derived from various tissues, including BM, AT, and SM, have reported these cells to exhibit similar cartilage regenerative properties despite differences in paracrine factors secreted by each MSC type [46][47][48]. Based on these results, we hypothesised that there are paracrine factors commonly secreted by MSCs derived from tissues of different origins. Therefore, MSCs were isolated from each tissue; the cultured cells were confirmed to possess MSC characteristics defined   Table 1).
GO analysis was performed to confirm the properties of commonly secreted proteins by all-MSCs under normal or inflammatory conditions. We found that the largest proportion of MF was "protein binding," suggesting that direct regulation of PPIs may be a major regulatory process carried out by proteins secreted by MSCs [49]. BP included "regulation of biological process," "response to stimulus," "metabolic process," and "cell organization and biogenesis," indicating that processes functionally relevant to commonly secreted proteins by all-MSCs play an important role in tissue repair [23,49].
In addition, KEGG pathway analysis was conducted to confirm the biological function of paracrine factors commonly secreted by all-MSCs, and several pathways related to regeneration were verified. Environmental information processing included several pathways such as signal transduction, and the subcategories of signal transduction were identified in 11 pathways (data not shown). Among these pathways, the TGF-β signalling pathway is known to be essential for tissue regeneration by activating Smad signalling

12
Stem Cells International to regulate collagen and aggrecan expression [50,51]. In addition, this pathway has been reported to maintain homeostasis and inhibit the degradation of articular chondrocytes by modulating proinflammatory cytokines, eventually repairing or regenerating damaged cartilage tissue [52]. Taken together, each MSC type secreted proteins related to the TGF-β signalling pathway, and these proteins may contribute to the maintenance and regeneration of damaged cartilage. We confirmed that the processes related to repair and regeneration were more under inflammatory conditions by GO and KEGG pathway analyses. The mechanism underlying this phenomenon was studied using PPIs. We found that ECM organization was a process observed under both environmental conditions. ECM stores and appropriately supplies factors necessary for cell growth and differentiation and supports spontaneous regeneration mechanisms to recover damaged tissues, displaying potential as a therapeutic avenue [53]. Meanwhile, as expected, a "cluster of responses to stimuli" was formed under inflammatory conditions. According to several studies reporting the association of paracrine factors secreted by MSCs under inflammatory conditions, the secretion and therapeutic effects of MSCs may be enhanced by inflammatory stimuli and crosstalk [54][55][56]. In addition, secretion of paracrine factors is reportedly increased in MSCs stimulated with inflammatory factors [41,57]. These findings indicate that paracrine factors contribute to the regeneration of damaged tissue under any condition. Furthermore, it was confirmed that paracrine factors secreted under inflammatory conditions are closely related to the repair and regeneration of cartilage, and several studies have reported them to regulate endogenous cellular responses and mediate cartilage regeneration in damaged tissue [6,58]. Thus, we concluded that paracrine factors released under inflammatory conditions facilitate the recovery and regeneration of damaged cartilage. Table 2 [6, shows the proteins predicted to be involved in the regeneration of damaged tissue. It was confirmed that ECM-cell interactions included collagen α-1(I) chain, decorin, biglycan, lumican, VCL, fibronectin 1, TSP-1, and TSP-2. These proteins are primarily known to constitute cartilage ECM and provide structural support to cells and tissues. Furthermore, they modulate cellular signals that can affect tissue organization, cellular proliferation, matrix adhesion, growth factors, and cytokine responses. They also reportedly protect the surface of collagen type I and II fibres from degradation [63,[92][93][94]. The articular cartilage, a specialised form of hyaline cartilage, is avascular and has a poor capacity for self-repair [95]. Therefore, ECM-cell interactions are considered essential for cartilage regeneration, and the proteins identified herein are similar to those reported in other studies [94,96].
Knee OA development has been observed to increase proinflammatory cytokines such as IL-1β and TNF-α in the joint cavity, which can trigger chondrocyte apoptosis [97,98]. In addition, lactoferrin (LTF) was reported to inhibit IL-1β-mediated chondrocyte apoptosis by regulating the activity of cyclic AMP-responsive element-binding protein through protein kinase B signalling. LTF also promoted car-tilage regeneration by increasing the expression of collagen type II in a rat OA model. Thus, LTF secreted by human MSCs can be anticipated to have anti-apoptotic and regenerative effects on chondrocytes in damaged cartilage. In a study of TSP-2, Jeong et al. [82] showed that intraarticular injection of human umbilical cord blood-MSCs into an osteochondral defect rat model exerts a regenerative effect on damaged cartilage through paracrine factors, of which TSP-2 was the major paracrine player. These findings demonstrate the therapeutic potential of TSP-2-mediated paracrine action of human MSCs, which can regenerate cartilage in knee OA treatment.
Furthermore, inflammation is accompanied by the breakdown of the ECM, the main component of articular cartilage [99]. The synovial fluid of knee OA patients exhibits a high level of proinflammatory cytokines as well as different types of MMPs [100]. Synovitis is mostly associated with gradual progressive damage to cartilage and severe pain. The pain in knee OA may be caused by the activation of nociceptive pathways by nerve growth factor, which in turn may occur due to inflammation. Therefore, inflammation is a major therapeutic target to relieve OA pain, and alleviation of inflammation and suppression of MMPs may be key strategies for knee OA treatment [101]. Interestingly, in our study, we found that TSG-6 and TSP-1 proteins were secreted only under inflammatory conditions. These proteins attenuate proinflammatory cytokines and MMPs by modulating the nuclear factor κB pathway and enhance the production of collagen type II [65,[102][103][104]. TSG-6 and TSP-1 have anti-inflammatory and tissue-protective properties and can enable the regeneration of damaged cartilage.
We confirmed the level of expression depending on the environment for some of the abovementioned proteins. TSP-2 was identified in both conditions, and it was confirmed that its expression level was similar between the two conditions. This suggests that TSP-2 is a paracrine factor that is secreted by MSCs under various conditions and plays an essential role in cartilage regeneration. TSG-6, which was only secreted under the inflammatory condition, showed a dramatically higher expression level under the inflammatory condition than under the normal condition, supporting the result that TSG-6 is induced by stimulation by IL-1 and TNF-α [105].
The limitation of this study is that the identified proteins could not be confirmed to regenerate damaged cartilage through actual studies. However, it was meaningful to check the protein list for paracrine factors secreted by MSCs under normal culture conditions compared to those secreted under inflammatory conditions. Moreover, according to recent studies, the paracrine factors identified in this study have an effect on cartilage regeneration [62][63][64][65][66]. Thus, commonly secreted paracrine factors by each MSC type provide valuable information for understanding the potential of damaged knee OA regeneration.

Conclusions
We identified proteins commonly secreted by human BM-, AT-, and SM-MSCs under normal and inflammatory 13 Stem Cells International conditions and consequently found paracrine factors that are predicted to be closely related to cartilage regeneration under inflammatory conditions. Therefore, our study sheds light on paracrine factors as potential therapeutic options for knee OA.

Data Availability
The experimental data analysed in this study are included in this article. The data used to support the findings of this study are available from the corresponding author upon request.

Conflicts of Interest
The authors declare that there is no conflict of interest regarding the publication of this paper.