CD10-Bound Human Mesenchymal Stem/Stromal Cell-Derived Exosomes Possess Immunomodulatory Cargo and Maintain Cartilage Homeostasis under Inflammatory Conditions

1. Introduction

In inflammatory joint diseases, including certain phenotypes of osteoarthritis (OA), the synovium and IFP tissues serve as origin of pro-inflammatory and articular cartilage degradative molecules, as well as a source of the pain-transmitting, immune and inflammation modulator neuropeptide, Substance P (SP)[1,2,3,4,5,6,7,8]. In fact, OA-related knee pain may be driving the association with time-to-mortality[9]. Given the current challenges of identifying disease-modifying treatment strategies for patients with OA[10], novel alternatives such as MSC-based therapeutic approaches have yielded encouraging initial clinical results. Early-stage clinical trials using freshly-isolated or culture-expanded MSC derived mainly from bone marrow and adipose tissues have demonstrated clinical superiority when compared with current alternatives such as hyaluronic acid intra-articular placement[11]. Therefore, to reverse these detrimental pathological cascades, we originally focused on MSC-based immunomodulatory therapies, as their paracrine effects actively modulate immune, inflammatory, and fibrotic events[12,13,14,15]. We recently reported that human IFP-MSC after transiently engrafting into joint areas of active synovitis/IFP fibrosis show a potent anti-inflammatory/analgesic phenotype by actively degrading SP via neutral endopeptidase CD10[16,17].

Concern regarding the potential side effects and increased immunogenicity with MSC-based therapies has led researchers to pursue cell-free treatments focusing on the MSC secretome, especially their exosomes. Exosomes are nanosized (50–200 nm) extracellular vesicles generated via the endosomal pathway[18], and secreted by numerous cells in response to their surrounding milieu. Thus, their contents (i.e., cargo) and lipid shell may carry information that reflects particular changes in the parental. According to our data, the in vitro SP degradation capacity of the IFP-MSC was fully recapitulated by their supernatant alone, while, abrogated with a CD10 inhibitor, suggesting the release of active exosome-bound CD10 from the cells[17]. On this basis, further investigations revealed that IFP-MSC derived exosomes (IFP-MSC EXOs) show distinct miRNA and protein immunomodulatory profiles. Specifically, IFP-MSC EXOs infusion into the knee in an acute synovial/IFP inflammation rat model resulted in robust macrophage polarization towards an anti-inflammatory therapeutic M2 phenotype within the synovium/IFP tissues[19]. Overall, preclinical studies have demonstrated that MSC EXOs have strong immunomod-ulatory properties, particularly through the action of miRNAs, which may be capable of targeting the immune system and modulating angiogenesis[10].

We and others have isolated and characterized exosomes from various MSC sources (i.e. bone marrow, umbilical cord, adipose tissues, endometrium), and our results supports a therapeutic potential through a strong anti-inflammatory, anti-fibrotic and angiogenesis-remodeling capacities[19,20,21]. Our hypothesis is that CD10 expression levels in IFP-MSC are directly related to the immunomodulatory and chondroprotective effects of their derived EXOs. On this basis, CD10High EXOs show potent molecular immunomodulatory profile which significantly affect synoviocytes functionality in inflammatory conditions in vitro. Importantly, chondroprotective effects of CD10High EXOs were not only observed in chondropellets/synoviocytes cocultures under inflammatory conditions in vitro, but also in an animal model of acute synovitis/IFP fibrosis by retaining articular cartilage structure/composition. These observations provide a rationale for further testing of a viable MSC EXO-based therapeutic modality for synovitis/IFP fibrosis as well as chronic conditions such as OA where articular cartilage degradation is a critical component of the disease.

2. Materials and Methods

Isolation, Culture and Expansion of IFP-MSC

All experiments using human cells were performed in accordance with relevant guidelines and regulations. Human IFP-MSC were isolated from IFP tissue obtained from de-identified, non-arthritic patients (n=3; two males 46 and 67 years old, and one female 44 years old) undergoing elective knee arthroscopy at the Lennar Foundation Medical Center at the University of Miami. Informed consent was obtained from all participants. All procedures were carried out following approval by the University of Miami IRB not as human research (based on the nature of the samples as discarded tissue). IFP tissue (5-10 cc) was mechanically dissected and washed repeatedly with Dulbecco’s Phosphate Buffered Saline (DPBS; Sigma Aldrich, St Louis, MO, USA), followed by enzymatic digestion using 235 U/ml Collagenase I (Worthington Industries, Columbus, OH, USA) diluted in DPBS and 1% bovine serum albumin (Sigma) for 2 hours at 37^oC with agitation. Enzymatic digestion was inactivated with complete media with DMEM low glucose GlutaMAX (ThermoFisher Scientific, Waltham, MA, USA) containing 10% fetal bovine serum (FBS; VWR, Radnor, PA, USA), washed and seeded at a density of 1x10⁶ cells/175 cm² flask in two different complete media: human platelet lysate (hPL) and chemically-reinforced (Ch-R) media. Complete hPL medium was prepared by supplementing DMEM low glucose GlutaMAX with 10% hPL solution (PL Bioscience, Aachen, Germany). Complete Ch-R medium was prepared by mixing Mesenchymal Stem Cell Growth Medium 2 with supplement provided according to manufacturer’s instructions (PromoCell, Heidelberg, Germany). At 48h post-seeding, non-adherent cells were removed by DPBS rinsing and fresh media were replenished accordingly.

All MSC were cultured at 37°C 5% (v/v) CO2 until 80% confluent as passage 0 (P0), then passaged at a 1:5 ratio until P3, detaching them with TrypLE™ Select Enzyme 1X (Gibco, ThermoFisher Scientific) and assessing cell viability with 0.4% (w/v) Trypan Blue (Invitrogen, ThermoFisher Scientific). Specifically, Ch-R and hPL IFP-MSC cultures yielded CD10High and CD10Low IFP-MSC, respectively.

Immunophenotype

Flow cytometric analysis was performed on P3 IFP-MSC (n=3). Briefly, 2.0 x 10⁵ cells were labelled with CD10 monoclonal antibody (Biolegend, San Diego, CA, USA) and the corresponding isotype controls. All samples included a Ghost Red Viability Dye (Tonbo Biosciences, San Diego, CA, USA). Data (20.000 events collected) were acquired using a Cytoflex S (Beckman Coulter, Brea, CA, USA) and analyzed using Kaluza analysis software (Beckman Coulter).

Quantitative real-time PCR (qPCR)

RNA extraction was performed using the RNeasy Mini Kit (Qiagen, Frederick, MD) according to manufacturer’s instructions. Total RNA (1μg) was used for reverse transcription with SuperScript™ VILO™ cDNA synthesis kit (Invitrogen), and 10 ng of the resulting cDNA was analysed by qPCR using QuantiFast SYBR Green qPCR kit (Qiagen) and a StepOne Real-time thermocycler (Applied Biosystems, Foster City, CA). For each target, human transcript primers were selected using PrimerQuest (Integrated DNA Technologies, San Jose, CA) (Supplementary Table S1). All samples were analysed in triplicate. Mean values were normalized to GAPDH, expression levels were calculated using the 2^−ΔΔCt method and represented as the relative fold change of the primed cohort to the naïve (=1).

A pre-designed 90 gene Taqman-based mesenchymal stem cell qPCR array (Stem Cell Technologies, Supplementary Table S1) was performed (n=2) using 1000 ng cDNA per IFP-MSC sample and processed using StepOne Real-time thermocycler (Applied Biosystems). Data analysis was performed using Stem Cell technologies qPCR online analysis tool (Stem Cell Technologies). Sample and Control Ct Values were expressed as 2^−ΔΔCt (with 38 cycles cut-off point). The expression levels were represented in bar plots ranked by transcript expression levels on a log-transformed scale of sample compared to control cohorts. Bar plots were color-coded by the functional class of genes (namely Stemness, MSC, MSC-related/Angiogenic, Chondrogenic/Osteogenic, Chondrogenic, Osteogenic, Adipogenic). A t-test (unpaired, two-tailed test with equal variance) is used in all statistical analysis and p-values were corrected for multiple comparisons by the Benjamini-Hochberg procedure. CD10High and CD10Low IFP-MSC groups were compared and presented in bar plots.

Isolation and validation of IFP-MSC derived exosomes

CD10High and CD10Low IFP-MSC derived exosomes (EXOs) were isolated from IFP-MSC conditioned media by a stepwise ultracentrifugation method and CD63-immunomagnetic purification. Briefly, conditioned media from IFP-MSC groups cultured in exosome-depleted Ch-R or exosome-depleted hPL media[22] are filtered through a 0.22µm filter to remove debris and large vesicles, and differentially centrifuged for 2,000xg for 10 min, 10,000xg for 30 min, and ultracentrifuged for 120,000xg for 16hr[23]. Pre-enriched exosome preparations were incubated with the Dynabeads®-based Exosome-Human CD63 Isolation/Detection Reagent (Invitrogen) and using a magnetic separator, exosome preparations are further purified. Samples from each group were assessed for biophysical and biochemical characterization[19].

The functional assessment of IFP-MSC EXOs was performed in a concentration corresponding to exosomes secreted from 1 x 10⁶ IFP-MSC. For IFP-MSC EXOs tracking, exosomes were stained with PKH26 red fluorescent membrane staining kit (Fluorescent Cell Linker Kits, Sigma) according to manufacturer’s instructions and co-cultured with target cells in functional assessments.

miRNA profile of IFP-MSC EXOs

miRNA was extracted from CD10High and CD10Low EXOs using Total Exosome RNA and Protein Isolation Kit (Thermo Fisher Scientific) according to manufacturer’s instructions. Total exosome miRNA (1μg) was used for first-strand cDNA synthesis with All-in-One miRNA First-Strand cDNA Synthesis Kit (GeneCopoeia, Rockville, MD).

Pre-designed human MSC exosome 166 miRNA qPCR arrays (GeneCopoeia) were performed using 1000 ng cDNA per IFP-MSC sample (n=2), and processed using StepOne Real-time thermocycler (Applied Biosystems, LLC). Data analysis was performed using qPCR result with GeneCopoeia’s online Data Analysis System (http://www.genecopoeia.com/product/qpcr/analyse/). Mean values were normalized to small nucleolar RNA, C/D box 48 (SNORD48), expression levels were calculated using the 2^−ΔCt method. Putative miRNA interactomes were generated using a miRNet centric network visual analytics platform (https://www.mirnet.ca/). The miRNA target gene data were collected from well-annotated database miRTarBase v8.0 and miRNA-gene interactome network refining was performed with 2.0 betweenness cut-off. Values (with 34 cycles cut-off point) were represented in a topology miRNA-gene interactome network using force atlas layout and hypergeometric test algorithm. miRDB online database (http://mirdb.org) for prediction of functional miRNA targets has been used to correlate highly expressed target genes in macrophages and synoviocytes with specific miRNAs identified by IFP-MSC EXOs miRNA profiling. MirTarget prediction scores are in the range of 0–100% probability, and candidate transcripts with scores ≥ 50% are presented as predicted miRNA targets in miRDB[24].

Synoviocyte inflammation assay

Passage 1 synoviocytes (SYN) were expanded in synoviocyte medium (ScienCell, Carlsbad, CA). Synoviocyte/exosome (SYN/IFP-MSC EXOs) co-cultures were performed using CD10High and CD10Low EXOs for each sample (n=2). Co-cultures were fed with synoviocyte medium + TIC inflammatory/fibrotic cocktail (15 ng/ml TNFα, 10 ng/ml IFNγ, 10 ng/ml CTGF) for 72h.

CCK-8 cytotoxicity assay (Cell Counting Kit-8, Sigma) was performed in SYN TIC and IFP-MSC EXOs/SYN TIC according to manufacturer’s instructions. CD10High and CD10Low EXOs cytotoxicity was determined by measuring optical densities of individual wells at 450nm (SpectraMax M5 spectrophotometer, Molecular Devices, San Jose, CA).

RNA extraction from SYN TIC cultures was performed using the RNeasy Mini Kit (Qiagen, Frederick, MD) according to manufacturer’s instructions. Total RNA (1μg) was used for reverse transcription with SuperScript™ VILO™ cDNA synthesis kit (Invitrogen). A pre-designed 88 gene Human Synoviocyte array (GeneQuery™ Human Synoviocyte Cell Biology qPCR Array Kit, ScienCell) was performed using 1000 ng cDNA per culture and processed using StepOne Real-time thermocycler (Applied Biosystems, LLC). Mean values were normalized to ACTB housekeeping gene, expression levels were calculated using the 2−ΔCt method with 34 cycles cut-off. Values were represented in dot plots as the relative fold change of the CD10High EXOs/SYN TIC or CD10Low EXOs/SYN TIC to SYN TIC (reference sample, 2−ΔCt= X sample/X reference sample).

The functional enrichment analysis was performed using g:Profiler (version e108_eg55_p17_9f356ae) with g:SCS multiple testing correction method applying significance threshold of 0.05[25]. The colors for different evidence codes and for log scale are described in Supplementary Table S2.

Chondropellets/Synoviocytes co-culture assay

Chondrogenic differentiation (0.25x10⁶ IFP-MSC/pellet) was induced for 15 days with serum-free MesenCult-ACF differentiation medium (STEMCELL Technologies Inc, Vancouver, Canada). Chondropellets were harvested and chondropellets/synoviocytes transwell co-cultures were performed with and without CD10High EXOs (n=2). Co-cultures were fed with synoviocyte medium + TIC inflammatory/fibrotic cocktail (15 ng/ml TNFα, 10 ng/ml IFNγ, 10 ng/ml CTGF) for 72h.

On day 3, chondropellets were harvested for histology and molecular profiling. For histological analysis, chondropellets were cryosectioned and 6-μm frozen sections were stained with hematoxylin and eosin (Sigma), and 1% toluidine blue (Sigma) for semi-quantitative assessment of pellet structure and chondrogenic differentiation, respectively. For molecular profile analysis, RNA extraction from chondropellets was performed using the RNeasy Mini Kit (Qiagen, Frederick, MD) according to manufacturer’s instructions. Total RNA (1μg) was used for reverse transcription with SuperScript™ VILO™ cDNA synthesis kit (Invitrogen). A pre-designed 88 gene Human Osteoarthritis array (GeneQuery™ Human Osteoarthritis and Cartilage repair qPCR Array Kit, ScienCell) was performed using 1000 ng cDNA per culture and processed using StepOne Real-time thermocycler (Applied Biosystems, LLC). Mean values were normalized to ACTB housekeeping gene, expression levels were calculated using the 2−ΔCt method with 34 cycles cut-off. Values were represented in a stacked bar plot as the relative fold change of the chondropellets with CD10High EXOs to the chondropellets without CD10High EXOs treatment (reference sample, 2−ΔCt= X sample/X reference sample).

The functional enrichment analysis was performed using g:Profiler (version e108_eg55_p17_9f356ae) with g:SCS multiple testing correction method applying significance threshold of 0.05[25]. The colors for different evidence codes and for log scale are described in Supplementary Table S2.

Mono-iodoacetate model of acute synovial/IFP inflammation

All animal experiments were performed in accordance with relevant guidelines and regulations. The animal protocol was approved by the Institutional Animal Care and Use Committee (IACUC) of the University of Miami, USA (approval no. 21-030 LF) and conducted in accordance to the ARRIVE guidelines[26]. Twelve (#12) 10-week-old male Sprague Dawley rats (mean weight 250 g) were used. The animals were housed to acclimate for 1 week before the experiment initiation. One rat was housed per cage in a sanitary, ventilated room with controlled temperature, humidity, and under a 12/12 hour slight/dark cycle with food and water provided ad libitum.

Acute synovial/IFP inflammation was generated by intra-articular injection of 1 mg of mono-iodoacetate (MIA) in 50 µl of saline into the right knee. Under isoflurane inhalation anesthesia, rat knees were flexed 90º and MIA was injected into the medial side of the joint with a 27G needle using the patellar ligament and articular line as anatomical references. Four (4) days later, a single intra-articular injection of CD10High EXOs in 50 µl of Euro-Collins solution (MediaTech) was performed, having as control: 1) rat knees receiving MIA but not IFP-EXOs (Only MIA group); and 2) healthy rat knees. Animals were sacrificed at day 4 after IFP-EXOs injection (d8 in total). This short exposure to MIA has been shown to induce inflammatory changes within the synovium and adjacent IFP[27].

Cytochemical staining and Lubricin (PRG4) immunolocalization in situ

Rat knee joints were harvested by cutting the femur and tibia/fibula 1 cm above and below the joint line, muscles were dissected and removed, and joints were fixed with 10% neutral buffered formalin (Sigma-Aldrich) for 14 days at room temperature. Knee joints were decalcified, cut along the sagittal plane in half, embedded in paraffin, and serial 4 μm sections obtained. Toluidine blue staining was performed to evaluate the sulfated glycosaminoglycans content of articular cartilage. Masson’s trichrome staining was performed to evaluate the collagen type and content of the articular cartilage. Microscope images of cytochemically stained tissues were acquired using x10 and x20 objectives Leica DMi8 microscope with Leica X software (Leica). Histochemical staining quantitative analysis was evaluated in 3 rat knees per condition and 4 microscopy fields per knee with Fiji/ImageJ software.

For anti-lubricin (anti-PRG4) immunofluorescence staining, sections were permeabilized with 1x PBS + 0.2% Triton X-100 for 20 minutes at room temperature and incubated with blocking buffer (1x PBS + 0.1% Triton X-100 with 1% bovine serum albumin) for 1 hour at room temperature. In between different treatments sections were washed with 1x PBS. Mouse anti-rat PRG4 monoclonal antibody (Sigma) was prepared in blocking buffer (1:500) and sections were incubated at 4 °C overnight. Sections were washed with 1x PBS + 0.01% Triton X-100 and incubated for 2 hours with secondary antibody Alexa Fluor647 conjugated goat anti-mouse IgG antibody (Thermo Fisher Scientific) at room temperature. Controls were incubated with secondary antibody only. All sections were rinsed with 1x PBS, mounted in prolong gold antifade reagent with DAPI (Invitrogen), and microscope images were acquired using x20 objective Leica DMi8 microscope with Leica X software (Leica).

Statistical analysis

Normal distribution of values was assessed by the Kolmogorov-Smirnov normality test. In the presence of a non-normal distribution of the data, one-way or two-way ANOVA were used for multiple comparisons. All tests were performed with GraphPad Prism v7.03 (GraphPad Software, San Diego, CA). Level of significance was set at p<0.05.

3. Results &Discussion

Distinct molecular profiles of IFP-MSC based on CD10 protein expression level

CD10 is a surface neutral endopeptidase expressed in multiple cell types including MSC [28,29], with enzymatic activity neutralizing various signalling substrates including Substance P (SP)[30,31]. Specifically, SP is a neuropeptide associated with nociceptive pathways that is secreted by sensory nerve fibres in the synovium and IFP tissues. Upon its secretion it actively affects local inflammatory/immune and fibrotic responses by modulation of cell proliferation, activation and migration to sites of inflammation, and the expression of recruiting chemokines and adhesion molecules[3,4]. Importantly, for the onset and progression of human inflammatory joint diseases SP-secreting sensory nerve fibres predominate over sympathetic fibers in anterior knee pain[7,8,32], whereas SP secretion is increased in synovial fluid during joint inflammation[3]. On this basis, we have previously reported that IFP-MSC can efficiently degrade substance P and suppress T cell proliferation in vitro, whereas upon intra-articular infusion in vivo IFP-MSC can dramatically reverse signs of synovitis and IFP fibrosis[33,34]. Specifically, our in vivo animal studies revealed a correlation between CD10 expression magnitude in IFP-MSC and the reduction or even absence of SP+ fibers in areas of active inflammation and fibrosis for both the synovium and IFP tissues. Based on the correlation between CD10 positivity and therapeutic outcomes, levels around 70% or more should represents an effective therapeutic outcome [33].

In the present study, IFP-MSC cultured in regulatory-complaint media (Ch-R and hPL) media showed similar fibroblast-like morphology but importantly had different CD10 expression levels (Figure 1A). Ch-R cultured IFP-MSC had 96.5±5.6% expression whereas hPL cultured IFP-MSC had 59.2±15.7% expression. According to the abovementioned 70% threshold for an effective therapeutic outcome in vivo, we named Ch-R cultures as CD10High (more therapeutic) and hPL cultures as CD10Low (less therapeutic).

Molecular profiling of CD10High versus CD10Low IFP-MSC revealed that 25 out of 90 genes tested were more highly expressed in CD10High IFP-MSC with 8 genes being more than 2-fold higher (KDR, COL10A1, NGFR, PROM1, ALPL, EGF, BMP7, ITGAX). Interestingly, genes tested were grouped in phenotype-related cohorts with MSC-associated, Chondrogenic/Osteogenic, and MSC cohorts showing overall the most prominent fold expression change between CD10High and CD10Low IFP-MSC cultures (Figure 1B). Functionally, KDR (VEGFR2) and COL10A1 genes, which are more than 10-fold higher expressed in CD10High IFP-MSC are strongly related to robust pro-chondrogenic and pro-angiogenic MSC actions[35,36]. Interestingly, studies show that the JNK-EGR1 signalling axis promotes TNF-α-induced endothelial differentiation of MSC via VEGFR2 expression[35]. COL10A1 is a specific marker of hypertrophic chondrocytes and is critical for endochondral bone formation, as mutation and altered COL10A1 expression is often accompanied by abnormal chondrocyte hypertrophy in many skeletal diseases[36]. From the highly expressed genes (more than 2-fold), NGFR (CD271) and PROM1 (CD133) have important roles in pro-angiogenic, immunomodulatory, and chondrogenic MSC actions. In vivo studies showed that CD271⁺ MSC and CD133⁺ MSC infusion in ischemia animal models result in increased pro-angiogenic and decrease pro-inflammatory signalling [37,38]. Several studies have associated high NGFR gene and protein expressions in MSC with their increased chondrogenic capacity in vitro and osteochondral cartilage repair in vivo [39,40,41]. Pro-chondrogenic and pro-angiogenic actions can be also attributed to MSC with high expression levels of ALPL, EGF, BMP7, and ITGAX genes[42,43,44]. Of note, MCAM (CD146) gene higher expressed in CD10High IFP-MSC is directly related to increased immunomodulatory MSC actions in vitro and in vivo [45].

In contrast, CD10Low IFP-MSC molecular profile showed more than 2-fold higher expression of 45 genes compared to CD10High IFP-MSC profile. Specifically, FGF9, FGF10, ACAN, IGF-1 genes seem to be a characteristic molecular signature for CD10Low IFP-MSC as they are expressed more than 150-fold higher, respectively. These genes are involved in proliferation and differentiation MSC signaling. Studies showed that FGF9 and FGF10 expression acts as an important regulator of early chondrogenesis partly through the AKT/GSK-3β signaling pathway[46,47]. High ACAN gene expression is related to pro-chondrogenic MSC action as its protein is a key GAG-containing proteoglycan in cartilage and plays an important role in stabilizing the ECM in articular cartilage[48]. Also, a recent study showed that IGF-1 transfected MSC have chondroprotective effects in an OA animal model[49].

Overall, CD10 protein levels in MSC strongly affect their molecular profile. CD10High IFP-MSC had a more distinct molecular profile consisting of only 8 genes that were significantly higher expressed (more than 2-fold) compared to CD10Low IFP-MSC. Furthermore, CD10High IFP-MSC molecular profile is related to robust pro-chondrogenic and pro-angiogenic MSC actions.

CD10High and CD10Low EXOs possess immunomodulatory miRNA cargo

Similar to our previous study[19], upon ultracentrifugation and CD63⁺ immunoselection, both CD10High and CD10Low EXOs show high purity for CD9 (>90%) and <200nm sizes (data not show). From 166 MSC-related miRNAs analyzed, 154 and 151 miRNA cargo were present in CD10High (Figure 2A) and CD10Low eMSC EXOs (Figure 3A), respectively. In CD10High EXOs, 9 miRNAs cargo were predominant (hsa-miR-146a, hsa-miR-4466, hsa-miR-1290, hsa-miR-6089, hsa-miR-1246, hsa-miR-3665, hsa-miR-7975, hsa-miR-4516, hsa-miR-4454) whereas in CD10Low EXOs, 6 miRNAs cargo (hsa-miR-146a, hsa-miR-6089, hsa-miR-4466, hsa-miR-3665, hsa-miR-4454, hsa-miR-7975) were predominant. Furthermore, 2 of the commonly highly present miRNAs (hsa-miR-146a and hsa-miR-6089) are potent regulators of immune system, and specifically T cells and macrophages activation and polarization. The first highly present miRNA cargo, hsa-miR-146a, belongs to the hsa-miR-146 family of genes that are expressed in response to pro-inflammatory stimuli as negative feedback to control inflammation. Specifically, studies showed that hsa-miR-146a negatively regulates the adaptive immunity by modulating adaptor protein (AP)-1 activity and IL-2 expression in T cells, as well as immune cell activation and cytokine production[50,51]. The second highly present miRNA cargo, hsa-miR-6089 inhibit the activation of macrophages and regulates the generation of IL-6, IL-29, and TNF-α by directly controlling TLR4 signaling[52]. Reactome analysis of miRNAs predominant in CD10High EXOs and CD10Low EXOs showed their involvement in the regulation of six gene groups related to: gene expression, immune system, NGF/PDGF/Wnt pathways, metabolism of proteins, cell cycle, and cellular responses to stress (Figure 2B and Figure 3B).

Overall, most detected miRNAs were commonly present in IFP-MSC EXOs however 5 miRNAs were specific for CD10High EXOs (hsa-miR-451a, hsa-miR-374a-5p, hsa-miR-525-5p, hsa-miR-499a-5p, hsa-miR-369-3p) and 2 miRNAs were specific for CD10Low EXOs (hsa-miR-132, hsa-miR-218-5p) (Figure 4A). Moreover, the histological profile of the synovium in OA preclinical models[17] and often in OA patients[53] is characterized by synovial lining hyperplasia and IFP fibrosis. The main cellular types contributing to these histological changes are synoviocytes and macrophages polarized towards the pro-inflammatory classical M1 phenotype (reviewed in[54]). On this basis, macrophage phenotype manipulation has been proposed as a potential OA therapy given that initial data, including results from recent studies in our laboratory, indicate polarization of macrophages towards an alternative anti-inflammatory M2 phenotype and IFP fibrosis clearance[16,19].

In the present study, in silico analysis revealed a functional correlation of identified miRNAs in CD10High EXOs and CD10Low EXOs with known highly expressed target genes in macrophages and synoviocytes (Figure 4B). For CD10High EXOs, the prediction scoring system revealed CCL2 as a target for hsa-miR-374a-5p (92% probability), CD163 as a target for hsa-miR-369-3p (67% probability), IL-10 as a target for hsa-miR-374a-5p (68% probability), TIMP-2 as a target for hsa-miR-369-3p (53% probability), and PRG4 as a target for hsa-miR-369-3p and hsa-miR-374a-5p (56% and 66%, respectively). Interestingly, hsa-miR-369-3p and hsa-miR-374a-5p CD10High EXOs cargo are potent regulators of anti-inflammatory M2 macrophages phenotype (CCL2, CD163, IL-10, TIMP-2)[16,55,56] and lubricin protein production from synoviocytes (PRG4)[57]. For CD10Low EXOs, prediction scoring system revealed TAC1, ARG1, PRG4 as targets for hsa-miR-218-5p (84%, 67%, 80%, respectively). Of note, hsa-miR-218-5p CD10Low EXOs cargo is a potent regulator of inflammation/pain (TAC1), anti-inflammatory M2 macrophages phenotype (ARG1), and lubricin protein production from synoviocytes (PRG4)[16,17,57]. Overall, from a clinical standpoint the potent functionality of these miRNAs strongly support the notion of developing novel cell-free therapeutics for inflammation/fibrosis reversal based on CD10 signature of IFP-MSC EXOs.

CD10High and CD10Low EXOs anti-inflammatory effects on synoviocytes

Synovial lining hyperplasia is a main characteristic of IFP/synovium inflammation in OA. In general, there are two types of resident synoviocytes contributing to this pathological conditions, type A (macrophage-like synoviocytes) and type B (fibroblast-like synoviocytes) are mainly responsible for maintenance of synovial homeostasis[54,58]. Herein, we separately investigated the functional effects of CD10High and CD10Low EXOs on stimulated synoviocytes in vitro to identify the optimal therapeutic IFP-MSC EXOs scheme for in vivo application.

In co-cultures, both CD10High and CD10Low EXOs were internalized by TIC-primed SYN (SYN TIC) and their proliferation was attenuated (Figure 5A,B). This finding suggests that IFP-MSC EXOs, irrespective of the CD10 levels in their parental cells, could attenuate synovial lining hyperplasia upon their in vivo intra-articular infusion. Similarly, we and others have reported that SYN exposure to MSC EXOs results in suppressed healthy-SYN[19] and rheumatoid arthritis-SYN[59] activation, migration, and invasion in vitro.

At the molecular level, SYN TIC cultures gene expression was strongly affected by both CD10High and CD10Low EXOs. From 88 genes analyzed, most expression levels were down regulated upon exposure to IFP-MSC EXOs (Figure 5C). Specifically, CD10High EXOs increased the expression levels of 45 genes in SYN TIC with only 13 genes being more than two-fold higher (PTGS2, MDM2, EREG, KRT7, IGF1, ESM1, IL15, ITGA4, VIM, POSTN, CSNK1D, CD14, ALB). Reactome analysis of highly expressed genes in CD10High EXOs/SYN TIC showed their involvement in the regulation of 5 molecular pathways related to the: regulation of cell population proliferation, cellular response to stimulus, prostaglandin biosynthetic process, regulation of neuroinflammatory response, regulation of protein phosphorylation (Figure 5D). Also, CD10Low EXOs increased the expression levels of 19 genes in SYN TIC with only 9 genes being more than two-fold higher (PTGS2, EREG, MDM2, KRT7, NGF, POSTN, CD14, ESM1, ITGA4). Reactome analysis of highly expressed genes in CD10Low EXOs/SYN TIC showed their involvement in the regulation of 5 molecular pathways related to the: regulation of cell population proliferation, regulation of protein phosphorylation, cellular response to stimulus, prostaglandin biosynthetic process, and regulation of cellular senescence (Figure 5D).

From the commonly expressed genes, PTGS2 (COX2) is the highest expressed (more than 500-fold) after IFP-MSC EXOs exposure compared to SYN TIC cultures alone. Importantly, PTGS2 plays crucial role in the promotion of prostaglandin E2 (PGE2) biosynthesis pathway in fibroblast-like cells (such as MSC) indicating its strong involvement in cells’ immunomodulatory capacity in vivo [60]. On this basis, via COX2-dependent PGE2 secretion fibroblast-like cells can polarize macrophages to an alternative anti-inflammatory M2 phenotype, attenuate T cell proliferation and differentiation to Th17, and increase T cells differentiation to Tregs (reviewed in [60]). In parallel, a pioneering study showed an COX2-dependent PGE2 regulated suppression of fibroblasts proliferation[61] that in our data could explain the reduction of synoviocytes proliferation upon exposure to IFP-MSC EXOs. From the highly expressed genes (more than 50-fold), EREG and MDM2 have important pro-survival and trophic roles in synoviocytes. Studies showed that EREG (Epiregulin), a ligand for epidermal growth factor receptor (EGFR), can affect EGFR signaling which is critical for maintaining the superficial layer of articular cartilage and preventing osteoarthritis initiation[62]. Studies showed that MDM2 higher gene expression in fibroblast-like cells result in higher extracellular matrix accumulation (collagens and fibronectin)[63]. However, MDM2 which is an ubiquitin ligase and binds p53 transcriptional factor for apoptotic pathway, has been associated with rheumatoid arthritis promotion[64]. Although, this is only a part of the pleiotropic IFP-MSC EXOs actions on synoviocytes, these findings invite studies involving OA.

Except the commonly highly expressed genes, CD10High EXOs exposed SYN TIC show additional expression of genes (such as IGF-1 and ALB) involved in synovial fluid enrichment with anabolic factors for cartilage homeostasis. IGF-1 gene expression is associated with insulin-like growth factor (IGF) signaling pathway activation that can promote chondrocyte proliferation, enhance matrix production, and inhibit chondrocyte apoptosis[65]. Also, high ALB gene expression is related to increased production from synoviocytes of albumin, the most common protein in synovial fluid, which plays an important role in the lubrication of the cartilage, thereby protecting the joint cartilage surface from the degradation[66]. Therefore, due to more potent and pleiotropic effects of CD10High EXOs on synoviocytes at the molecular level, we used CD10High EXOs as our platform for further in vitro and in vivo experimentation.

CD10High EXOs-treated chondrocytes show chondroprotective molecular profile in inflammatory conditions in vitro

At the early onset of inflammatory joint diseases, synovium/IFP anatomical and functional unit[67] serve as a site for immune cell infiltration, neovascularization, origin of pro-inflammatory and articular cartilage (AC) catabolic molecules[11,68]. In this pathological microenvironment, chondrocytes become sensitive to pro-inflammatory mediators, especially tumor necrosis factor-α (TNFα) and interleukin-1β (IL-1β), which initiate the degradation of cartilage extracellular matrix mediated by matrix-metalloproteinases (MMPs), a disintegrin and metalloproteinase with thrombospondin motifs[11]. Herein, we simulated this pro-inflammatory microenvironment in vitro by co-culturing MSC-derived chondropellets with inflamed synioviocytes (Figure 6A).

Initially, IFP-MSC were successfully induce towards chondrogenesis in 3-D micromass pellets cultures for 15 days. On this basis, we totally acknowledge that despite MSCs’ ability to differentiate toward the chondrogenic lineage, their terminal phenotype is reminiscent of one following endochondral bone formation rather than articular cartilage chondrocytes. However, we and others have demonstrated that on day 15 MSC have adequately differentiated and possess most of major characteristics of chondrocytes without entering hypertrophy and mineralization in vitro [33,69,70]. Subsequently, our generated chondropellets and SYN were co-cultured with and without CD10High EXOs under inflammatory/fibrotic TIC conditions for 3 days. Chondropellets histological analysis revealed that although CD10High EXOs-treated and non-treated have similar structure, CD10High EXOs treatment result in better chondrocyte differentiation and sulfated glycosaminoglycans production in vitro (Figure 6B).

At the molecular level, CD10High EXOs-treated chondropellets showed distinct transcriptional signature with higher expression of 19 genes (TGFB1, BMP7, WNT8A, COL1A2, GDF11, GADD45B, WNT9A, SOX9, BMP6, TNFSF11, RUNX2, DKK1, ACAN, COL1A1, MAPK8, WNT7A, NOTCH1, WNT4, TLR2) out of 88 genes tested compared to non-treated chondropellets (Figure 6C). Reactome analysis of highly expressed genes in CD10High EXOs-treated chondropellets showed their involvement in the regulation of 10 chondroprotection- and cartilage homeostasis-related molecular pathways (Figure 6D). Specifically, 7 genes showed more than 2-fold higher expression (TGFB1, BMP7, WNT8A, COL1A2, GDF11, GADD45B, WNT9A) in CD10High EXOs-treated chondropellets. Functionally, TGFB1, BMP7, and WNT8A genes, which are more than 5-fold higher expressed are strongly related to robust cartilage homeostasis effects. Both, TGFB1 and BMP7 are major articular cartilage homeostasis growth factors as they arrest chondrocytes differentiation at an early stage of hypertrophy and inhibit expression of COL10A1, VEGF, MMP13, and BGLAP (osteocalcin)[43,71]. Interestingly, studies showed that canonical Wnt signaling, via Wnt8A gene expression, skews TGFB signaling in chondrocytes towards anabolic signaling via ALK1 and Smad 1/5/8[72]. From the more than 2-fold higher expressed genes, COL1A2 and GDF11 genes are strongly related to chondrocytes proliferation/differentiation and extracellular matrix synthesis[73,74]. Of note, study showed that in normal articular cartilage approximately 9% of chondrocytes are positive for COL1A2 (collagen type I) protein whereas significant increase in COL1A2 is observed in cartilage degeneration[74]. Finally, GADD45B expression enhances COL10A1 transcription via the MTK1/MKK3/6/p38 axis in terminally differentiating chondrocytes[75], whereas WNT9A signaling is required for joint integrity and regulation of Indian hedgehog (Ihh) during chondrogenesis[76]. Overall, our data show that CD10High EXOs possess potent chondroprotective and anabolic effects for chondrocytes in vitro.

CD10High EXOs show chondroprotective and anabolic effects for articular cartilage in vivo

We have previously reported that functionally IFP-MSC EXOs can significantly reduce synovitis/IFP fibrosis and, also affect macrophage polarization towards an anti-inflammatory therapeutic M2 phenotype in inflammatory conditions in vivo [19]. Herein, an acute synovial/IFP inflammation rat (MIA) model was used to test the chondroprotective capacity of CD10High EXOs upon intra-articular infusion to reverse cartilage degeneration in vivo (Figure 7A).

Collagen type II and aggrecan are two major components of the extracellular matrix of healthy articular cartilage. Alterations in cartilage extracellular matrix composition and structure are characteristics of OA. Specifically, during OA progression degrading cartilage proteins such as matrix metalloproteinases and aggrecanases result in decreased collagen type II/aggrecan contents, and loss of cartilage integrity[77,78]. In our study, the diseased group (MIA only) demonstrated strong cartilage degeneration signs by reduced staining for sulfated proteoglycans compared to healthy rat knees (Figure 7B,C, upper panels). According to the Udo et al. cartilage scoring system (0-5) for rat OA in the MIA model[27], only the diseased group developed grade 3 cartilage erosion (≤50% of joint surface). Importantly, CD10High EXOs intra-articular infusion resulted in significantly (p<0.05) reduced cartilage degeneration with only minor cartilage depressions compared to diseased group (average scores 0.4±0.5 and 1.7±1.1, respectively). Masson’s Trichrome staining has also been used as a suitable histochemical method[79] to evaluate collagen compositional changes. In general, diseased group showed increased cartilage discoloration (stained as red) compared to other groups. In contrast, CD10High EXOs intra-articular infusion resulted in significantly (p<0.05) increased collagen composition similar to the healthy control group (Figure 7B,C, middle panels). Similar to our findings, previous studies using intra-articular infusions of MSC EXOs derived from various tissue sources such as bone marrow, amniotic fluid, and umbilical cord, reported the anabolic effects of EXOs to enhance matrix synthesis of collagen type II and sulfated glycosaminoglycans (reviewed in [80]). However, in our study for the first time we demonstrate the chondroprotective effects of CD10High EXOs derived from IFP tissue to articular cartilage.

The avascular nature of articular cartilage limits the capacity for perivascular MSC to migrate from the neighboring microenvironment (i.e. subchondral bone) and restore inflammation/tissue imbalance. However, a pool of resident cartilage-derived stem/progenitor cells (CSPC) is present in the upper zone cartilage zone, associated with the expression of proteoglycan 4 (PRG4) encoding for proteoglycan 4 or lubricin[81,82]. Interestingly, studies showed that CPSC is an intermediate cell population between MSC and differentiated chondrocytes which can be stimulated to induce tissue healing and homeostasis at different stages of cartilage degradation in vivo [83,84]. On this basis, we performed in situ PRG4 protein expression evaluations to determine phenotypic cellular changes promoted in articular cartilage in disease and after intra-articular CD10High EXOs infusion (Figure 7B and 7C, bottom panels). In our study, the healthy group showed a rigid distinct PRG4+ CSPC upper layer on articular cartilage. In the diseased group, PRG4 expression exhibited a reduction in the upper layer of articular cartilage, and simultaneously a significant (p<0.05) increase in the intermediate zone randomly distributed within chondrocytes. This phenomenon can be attributed to the fact that intermediate zone chondrocytes increase the expression of PRG4 in order to replenish the significant loss from the upper cartilage surface during disease progression. Notably, CD10High EXOs intra-articular infusion resulted in significantly (p<0.05) preservation of PRG4 expression on the upper cartilage surface and only minor expression from the intermediate zone chondrocytes. This effect can be attributed not only to the general CD10High EXOs immunomodulatory/chondroprotective miRNA profile (Figure 2), but also more specifically to the presence of hsa-miR-369-3p and hsa-miR-374a-5p miRNAs cargo that strongly affect PRG4 gene expression in target cells (Figure 4B). Therefore, CD10High EXOs treatment suggest robust chondroprotective and anabolic effects from articular cartilage in vivo.