Author contributions: Q.H.: conception and design, manuscript writing, collection and/or assembly of data; J.S., K.O., P.S.L., and C.K.: collection and/or assembly of data; B.J.H.: data analysis and interpretation, collection and/or assembly of data; D.A.H.: data analysis and interpretation, manuscript writing, financial support; R.J.K.: conception and design, data analysis and interpretation, provision of study material or patients, manuscript writing, final approval of manuscript, financial support.
Department of Surgery, University of Calgary, Calgary, Alberta, Canada
Department of Surgery, University of Calgary, Calgary, Alberta, Canada
McCaig Institute for Bone and Joint Health, Calgary, Alberta, Canada
Correspondence: R. Krawetz, Ph.D. McCaig Institute for Bone and Joint Health, Faculty of Medicine, University of Calgary, 3330 Hospital Drive NW, Calgary, Alberta, Canada T2N 4N1. Telephone: 403-210-6268; Fax: 403-270-2772; e-mail: firstname.lastname@example.org
Osteoarthritis (OA) is a multifactorial, often progressive, painful disease. OA often progresses with an apparent irreversible loss of articular cartilage, exposing underlying bone, resulting in pain and loss of mobility. This cartilage loss is thought to be permanent due to ineffective repair and apparent lack of stem/progenitor cells in that tissue. However, the adjacent synovial lining and synovial fluid are abundant with mesenchymal progenitor/stem cells (synovial mesenchymal progenitor cells [sMPCs]) capable of differentiating into cartilage both in vitro and in vivo. Previous studies have demonstrated that MPCs can home to factors such as monocyte chemotactic protein 1 (MCP-1/CCL2) expressed after injury. While MCP-1 (and its corresponding receptors) appears to play a role in recruiting stem cells to the site of injury, in this study, we have demonstrated that MCP-1 is upregulated in OA synovial fluid and that exposure to MCP-1 activates sMPCs, while concurrently inhibiting these cells from undergoing chondrogenesis in vitro. Furthermore, exposure to physiological (OA knee joint synovial fluid) levels of MCP-1 triggers changes in the transcriptome of sMPCs and prolonged exposure to the chemokine induces the expression of MCP-1 in sMPCs, resulting in a positive feedback loop from which sMPCs cannot apparently escape. Therefore, we propose a model where MCP-1 (normally expressed after joint injury) recruits sMPCs to the area of injury, but concurrently triggers changes in sMPC transcriptional regulation, leading to a blockage in the chondrogenic program. These results may open up new avenues of research into the lack of endogenous repair observed after articular cartilage injury and/or arthritis. Stem Cells2013;31:2253–2265
Musculoskeletal diseases represent one of the largest economic burdens to North American health-care systems and are major contributors to long-term disability, chronic pain, and reduced quality of life for a large segment of the population. Osteoarthritis (OA) is a degenerative disorder of joints in which the hyaline cartilage surface is lost over time. OA is quickly becoming the most common disease in the elderly as the global population ages . If current trends continue, it has been suggested that OA will become the fourth leading cause of global disability by the year 2020 . While OA increases with aging, its incidence and prevalence are increasing for many other reasons (active lifestyles, injuries, obesity, and others). Clinically, it is accepted that once a patient is diagnosed with OA, it may only be a matter of time (in more severe cases) before some level of surgical intervention is required to moderate the symptoms (pain, loss of mobility) associated with disease progression. Therefore, understanding the role of endogenous progenitors in the apparent lack of cartilage repair is paramount in developing novel therapeutics to restore the repair capacity of these cells. A logical first step is to identify if and how the injured and/or early OA joint environment impacts endogenous progenitor cell biology.
Articular cartilage is a remarkably resilient tissue that ordinarily lasts a lifetime. Although it was long treated as an inert material, it is clear that articular cartilage is a living tissue that undergoes some level of turnover to maintain homoeostasis under normal conditions . However, the exact mechanism whereby normal articular cartilage regenerates a functional extracellular matrix over time remains obscure. Several recent reports support the existence of a progenitor cell population in the synovium and synovial fluid that may contribute to normal joint maintenance and that might be a source of cells for endogenous articular cartilage repair [4-6].
There is significant evidence suggesting that cells from the synovium may play a role in cartilage repair/maintenance. In specific disease states of the synovium, chondrogenic potential of the endogenous progenitors can be observed. Chondromatosis, a condition in which cartilage forms within synovial tissue , demonstrates endogenous chondrogenic potential of this tissue if certain in vivo cues are misread. Within rheumatoid arthritis (RA) knee joints, synovial lining proliferation (pannus) contains cells that express specific markers of chondrocytes . Furthermore, during normal development of synovial joints, both synovium and cartilage originate from a common pool of progenitor cells , signifying that, even in the temporal and functional separation of these two tissues (cartilage and synovium), synovial mesenchymal progenitor cells (sMPCs) of the synovium do share a common cellular history with the cartilage surface of the joint.
The current view of adult progenitor cells in joint tissues is that they reside in a highly specialized niche in close association with somatic cells. MPCs are thought to mobilize and attempt repair during injury and to maintain normal homeostasis in musculoskeletal system [10-13] as they retain the ability to differentiate into osteoblasts (bone), adipocytes (fat), and chondrocytes (cartilage). MPCs are typically identified by expression of CD146, CD105, CD90, CD73, and CD44, but lacking the expression of CD45, CD34, CD31, and CD11b surface proteins . Using these characteristics, MPC populations have been derived from the synovial membrane and fluid with compelling evidence that these cells may possess cartilage repair potential [5, 15-24]. In comparison with bone marrow derived mesenchymal progenitor cells (bMPCs), sMPCs have increased chondrogenic potential [25-29]. Studies undertaken to explain this observation have shown that sMPCs have increased CD44 (hyaluronan receptor) expression  and natively express enzymes required for hyaluronan synthesis (UDPGD), whereas bMPCs and other MPC populations do not express UDPGD . Evidence from in vivo studies has demonstrated that when partial-thickness defects in the articular cartilage of rabbits are formed, a continuous layer that extends from the synovial membrane contributes to the repair of the cartilage in both native and chondrogenic-induced states [23, 24]. Furthermore, recent experiments have clearly demonstrated that labeled synovial progenitor populations can indeed migrate within the joint and integrate into either a normal or injured articular cartilage surface . Additional in vivo animal studies have also begun to elucidate the role of cytokines and chemokines in the recruitment of stem/progenitor cells to a defect or injury [31-37]. Animals deficient in MCP-1 demonstrate slowed and/or incomplete wound healing , while animals treated with injections of MCP-1 into the heart after injury retain stem/progenitor cells at the injury site for a longer duration . Additionally, stromal cell-derived factor-1 (SDF-1) has been implicated in the homing of stem/progenitors cells to musculoskeletal defects, and stem cells deficient in the receptor for SDF-1 are unable to migrate to the defect in vivo [36, 37].
The complete role of inflammation in OA is still unknown, but elements of inflammation have been directly implicated in the progression of the disease [38-42] and the degeneration of the cartilage surfaces of the joint . Specifically, interleukin 1β (IL-1β), a pro-inflammatory cytokine has been demonstrated to play a central role in the pathophysiology of cartilage damage and degradation in arthritis . In OA patients, increased IL-1β (message and/or protein) has been observed within the synovium , synovial fluid , and cartilage itself . IL-1β has been directly implicated in the inhibition of chondrogenic differentiation [46, 47] and inhibition of specific ECM proteins required for cartilage function . Furthermore, IL-1β is not the only inflammatory cytokine upregulated within synovium during OA, IL-6, IL-8, and TNF-α are all highly expressed and have roles in modulating chondrocyte behavior [49, 50]; however, the direct roles of many OA expressed cytokines on sMPCs chondrogenesis is yet unknown, but a number of studies are beginning to establish links between the inflammatory environment present in the arthritic joint to the ability of resident stem/progenitor populations to effectively undergo chondrogenesis . However, the role of MCP-1 in the recruitment and retention of stem/progenitor cells to a site of injury makes the elucidation of its roles in arthritis and stem cell biology particularly interesting.
In this study, we present evidence supporting the hypothesis that healthy synovial fluid contains synovial mesenchymal progenitor cells (sMPCs) that have the ability to differentiate into chondrocytes, and therefore possibly maintain the functional matrix of the articular surface. However, we suggest that MCP-1 is over-expressed in the synovial fluid of patients with OA and exposure to this molecule triggers changes in sMPC transcriptional regulation, leading to a net loss of chondrogenic differentiation in this cell population.
Materials and Methods
Informed consent to participate was obtained by written agreement. The study protocol was approved by the University of Calgary Research Ethics Board.
Inclusion and Exclusion Criteria
Normal Group (n = 10): Criteria for control cadaveric donations were an age of 40 years or older, no history of arthritis, joint injury or surgery (including visual inspection of the cartilage surfaces during recovery), no prescription anti-inflammatory medications, no co-morbidities (such as diabetes/cancer), and availability within 4 hours of death.
OA Group (n = 10): inclusive criteria were an age of 40 years or older, OA diagnosed based on the American College of Rheumatology criteria with x-ray documentation and no evidence of autoimmune disease or RA.
Luminex Multiplex Array
Synovial fluid was obtained from the joints by aspiration, the cells removed by centrifugation, and the resulting fluid maintained frozen at −80°C. Synovial fluid aliquots were thawed on ice and 20 μl of fluid was diluted 1:5 with the Milliplex running buffer (Millipore, Billercia, MD, www.millipore.com). Sample analysis was performed by Eve Technologies (University of Calgary) using the Human 65-Plex Cytokine/Chemokine Panel (Millipore, Billerica, MD, www.millipore.com) using the Luminex 100 platform (Luminex Corp., Austin, TX, www.luminexcorp.com). All samples were prepared and analyzed according to the manufacturer's instructions included with the kits. All samples were analyzed at least in duplicate and prepared standards were included in all runs.
The quantified data from the Cytokine Luminex arrays were analyzed using Stata 9.2 for Macintosh (Stata, College Station, TX, www.stata.com) and Prism GraphPad five for Macintosh (Prism, LA Jolla, CA, www.graphpad.com) in three steps. Step 1: Principal component analysis (PCA) was applied to the set of synovial fluid cytokine expression values returned by Luminex (normal (n = 10) and OA (n = 10). PCA is a data reduction algorithm that reduces a multivariate data set down to its principal components (PCs). These PCs represent a combination of ideally weighted values that are calculated from the variability of all cytokines investigated, while retaining much of the variability within the original data set so that a data set with many variables can be accurately described by just a few variables. Generally in PCA, PCs are retained if their Eigen value is greater than or equal to 1, however in this study only the first two PCs were investigated visually, and statistically, as they represented the vast majority of the dataset variability. In summary, after PCA each patient's cytokine data (normal and OA) is represented by only two variables: PC1 and PC2. Step 2: PC1 and PC2 values were then investigated separately in patient group comparisons (normal vs. OA using PC1 values and normal vs. OA using PC2 values) using two tailed t-tests (significance accepted at p ≤ .05). Step 3: Factor loadings (the weights of each cytokine on each of the PCs) were plotted for PC1 and PC2.
sMPC Selection and Chondrogenic Differentiation
sMPCs were derived from fresh human synovial fluid (normal and OA each n = 4) by aspiration and centrifugation, and then plated in DMEM:F12 containing 10% fetal bovine serum, 1% non-essential amino acids, 1% pen-strep, and 0.1% β-mercaptoethanol (all Invitrogen, Carlsbad, CA, www.lifetechnologies.com). After 7–14 days in culture, colonies were observed and the cells were sorted using fluorescence-activated cell sorting (FACS) for the CD105+, CD90+, CD73+, CD45−, and CD11b− population. Briefly, cells were stained with antibodies to CD105PerCP-Cy5.5, CD90FTIC, CD73APC, CD45PE, and CD11bPE (all Becton Dickinson: BD, Franklin Lakes, NJ, www.bd.com). The stained cells were sorted using the FACSAria III (BD, University of Calgary, Flow Cytometry Core), the number of cells matching the desired marker profile varied from culture to culture, however, CD105+, CD73+, CD90+, CD45−, and CD11b− population accounted for 50.2%(±8.3%) of the total cells isolated from normal joints, and 74.8%(±6.0%) from joints with OA. To quantify cells matching the sMPC marker profile directly out of fresh synovial fluid, 100 μl of synovial fluid was diluted to 1 ml with Dulbecco's Phosphate-Buffered Saline (DPBS) and passed through a 70-μm strainer (BD). The resultant cells were stained with the same antibodies and processed using the same FACS machine with CountBright absolute counting beads (Invitrogen, Carlsbad, CA, www.lifetechnologies.com) premixed with each sample as per the manufacturer's instructions and were gated separately .
Synovial membrane MPCs were obtained from fresh synovial tissue biopsies (∼3 mm3) recovered at the same time as the synovial fluid. Approximately half of each sample was processed for histology, while the remaining tissue was minced and placed into the same tissue culture conditions described above. At 7–14 days after initial seeding, the cells were strained (70 μm) and processed for FACS as described above.
sMPCs were plated in triplicate (100,000 cells per well per 24 well dish) and exposed to chondrogenic media for 14 days with or without micro-mass aggregation. Aggregation was achieved by placing 100,000 cells in a 1.5-ml sterile tube at 37°C overnight. The chondrogenic media contained the following: 500 ng/ml BMP-2 (Peprotech), 10 ng/ml TGF-β3 (Peprotech, Rocky Hill, NJ, www.peprotech.com),10–8 M dexamethasone, 50 μg/ml ascorbic acid, 40 μg/ml proline, 100 μg/ml pyruvate, and supplemented with insulin, transferrin, and selenium (all Sigma, St. Louis, MO, www.sigmaaldrich.com). Treatment groups contained 2,500 pg/ml of MCP-1(Peprotech, Rocky Hill, NJ, www.peprotech.com) in addition to the chondrogenic media, as this is the mean cytokine concentration found in the synovial fluid of the OA patients. All experiments were performed in at least quadruplicate.
Relative Quantification of Gene Expression
Chondrogenesis of sMPCs was evaluated in each of the treatment groups by quantifying the relative gene expression of the chondrogenic markers Sox9, Col2A1, and Aggrecan (ACAN) using quantitative polymerase chain reaction (qPCR).
Total mRNA from each well was extracted and purified using Trizol in accordance with the manufactures instructions (Invitrogen, Carlsbad, CA, www.lifetechnologies.com) and converted into cDNA according to the High Capacity cDNA Reverse Transcription Kits protocol (Applied Biosystems, Foster City, CA, www.lifetechnologies.com). cDNA of 2 μl was added to a 96-well qPCR plate along with Sox9, Col2a1, ACAN, or MCP-1 TaqMan validated probes/primers and TaqMan Universal PCR Master Mix. In addition, an 18S RNA probe/primer was used as an endogenous control. Two replicates of each time point and control sample were performed, and all cDNA used was from the same corresponding replicate to reduce variability.
Western Blot Analysis
Total protein was collected from sMSCs using a Tris-HCl/SDS based lysis/sample buffer, and separated on a 10% poly-acrylamide gel. The gels were transferred to nitrocellulose membranes and probed with primary antibodies specific to the proteins GAPDH, Sox9, and Collagen 2 (all Santa Cruz, Santa Cruz, CA, www.scbt.com). GAPDH was utilized as a control, as it is constitutively expressed in high levels in most cell types. An infra-red secondary was utilized for detection of the signal with the Odyssey imaging system (LICOR, Lincoln, NE, www.licor.com).
Alcian Blue Staining
The presence of glucosaminoglycans was visually detected using Alcian blue staining . Cells were fixed for 15 minutes at room temperature in 4% paraformadehyde (PFA) then washed three times in 1× phosphate-buffered saline (PBS). An amount of 90% ethanol was added to the plates for 30 minute and left in Alcian Blue stain (2 mg Alcian blue, 0.8 ml water, 16 ml 100% ethanol, and 4 ml glacial acetic acid) for 48 hours at 4°C. The cell cultures were rehydrated with progressively more dilute solutions of ethanol (70, 50, 25, and 20%), placed in PBS, and then visualized using 8× magnification.
Cell Counting and Viability Assessment
The cells were counted using a Bio-Rad TC-10 automated cell counter with trypan blue staining to quantify cell viability (all Bio-Rad). To determine cell proliferation rates, the CyQUANT Cell Proliferation Assay (Invitrogen, Carlsbad, CA, www.lifetechnologies.com) was utilized, briefly, normal, OA, and MCP-1-treated cells, were passaged at the same times (every 4 days) using the same split ratio (1:4). On passages that were being assayed for proliferation, 10,000 cells from each treatment group were plated into 60 mm plates and samples were collected every 24 hours for 8 days and cell number was quantified using a fluorescent standard curve and micro plate reader as described in the manufactures instructions.
Histology and Immunohistochemistry
Synovium biopsied from each individual was fixed in 4% PFA at 4°C overnight and then washed in PBS. The biopsies were then processed using an automatic tissue processer (LIECA, Wetzlar, Germany, www.leica.com) and embedded in paraffin wax. After processing, each tissue biopsy was sectioned (7 μm thickness) using a Leica RM2255. Each biopsy was stained with H&E and also analyzed using immunohistochemistry (IHC). For IHC, serial sections were prepared by deparaffinization and rehydration through ethanol of decreasing concentration. Antigen retrieval was achieved using Proteinase K at a concentration of 0.2 mg/ml. Serum blocking was performed using 1% bovine serum in 10% goat serum. Sections were then treated with a biotin-based primary antibody (CD90, CD271, CD68, MCP-1: all EBiosciences, San Diego, CA, us.ebioscience.com). The sections were then washed and treated with horse-radish peroxidase-conjugated strepavadin. For detection, True Blue Peroxidase Substrate (Mandel Scientific, Guelph, Canada, www.mandel.ca) solution was used. After staining was complete, the sections are washed and mounted onto a slide for visualization using an optical microscope. Positive control tissues were used to test the specificity for each primary antibody.
RNA was extracted using Trizol Reagent (Life Technologies, Inc, Carlsbad, CA, www.lifetechnologies.com) according to the manufacturer's protocol. Total RNA was purified with RNeasy Plus Micro Kit (Qiagen, Germantown MD, www.qiagen.com) to remove genomic DNA. The RNA integrity number (RIN) was measured with Agilent RNA 6000 NanoChips on a 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, www.agilent.com). The quantity was measured with a NanoDrop 1,000 (NanoDrop Technologies, Inc, Wilmington, DE, www.nanodrop.com). A total of 300 ng of each RNA sample with RIN higher than nine was labeled with GeneChip Whole Transcript (WT) Sense Target Labeling Assay (Affymetrix, Santa Clara, CA, www.affymetrix.com) and hybridized to Affymetrix GeneChip Human Gene 1.0 ST Arrays at 45°C for 16 hours. Arrays were stained and washed on Affymetrix GeneChip Fluidics 450 following the manufacturer's protocol and scanned with an Affymetrix GeneChip Scanner 3000 7G System.
Micro-Array Data Analysis
Array data files were generated with GeneChip Command Console Software (AGCC) (Affymetrix, Santa Clara, CA, www.affymetrix.com) and statistical significant analysis was carried out on software of GeneSpring (Agilent Technologies, CA, www.agilent.com). The fold change between normal and OA samples was based on the p < .05 from a t-test (Asymptotic and Benjamini Hochberg FDR).
Telomerase Semi-Quantitative Measurement
To measure the activity of telomerase in sMPCs, the TRAPeze Telomerase Detection Kit (Millipore, Billerica, MD, www.millipore.com) was utilized, specifically; 50 μl PCR reactions in 96-well plates were performed per the manufacturer's recommendations. Homogenates of cells and TSR8 standards, minus-DNA, heat-treated extracts (90°C for 10 minutes) and minus-Taq controls were run in duplicate. In this assay, fluorescent emission products generated through endogenous telomerase activities (detected through Sybr Green) were compared with the amplification of an internal control template (TSK2, detected through sulforhodamine). Next, 40 μl of PCR reaction was diluted 1:30 in 150 mmol/l NaCl and Sybr Green I and sulforhodamine fluorescence recorded using a Perkin Elmer Victor 3 plate reader (excitation/emission wavelengths of 495/516 and 600/620). Data are expressed in Total Product Generated (TPG) units, in which 1 TPG unit corresponds to one femtomole of TSR8 standard. Human embryonic stem cells were utilized as the positive control .
Stripe-Migration Assay and Microscopy
We modified the assay of Pelletier et al.  to deposit a spot of MCP-1 (2500 pg) coated Collagen one (100 μl; Purecol, Advanced Biomatrix, San Diego, CA, ww.advancedbiomatrix.com) on a glass coverslip. Approximately 1000 sMPCs were then deposited in a drop either at a point several millimeters away from the spot of MCP-1. Adhesion was allowed to proceed at 37°C for 1 hour. Coverslips were immersed in culture medium and incubated for 24 additional hours at 37°C. Cells then were fixed and stained for photography under light microscopy using the Ethidium Bromide to label nuclei.
Inflammatory Profiling of Normal and OA Synovial Fluid
To determine which cytokines, chemokines, and/or growth factors are differentially expressed (at the protein level) between normal and OA synovial fluid, native synovial fluid was collected from normal and OA joints and analyzed using the Milliplex Cytokine/Chemokine Discovery Array. Sixty-five different proteins (Supporting Information Table 1) were quantified and analyzed using principle component analysis (Fig. 1A, 1B). It was found that a number of cytokines were differently expressed at the protein level between normal and OA synovial fluid (Fig. 1C), with MCP-1 being one factor that strongly discriminated OA (bars on the right hand side of the plots) from normal synovial fluid (bars on the left-hand-side of the plots) (Fig. 1C). However, before moving to an in vitro based cell assay, IHC was performed on normal and OA synovium to examine co-location of sMPCs cells and MCP-1, as it is unknown if sMPCs within the synovium are exposed to MCP-1 in vivo.
Immunohistochemistry of Synovium
Normal synovial membrane has a uniform appearance when stained with H&E, with the majority of cells observed at the boundary (Fig. 2A). When the normal tissue was stained for presumptive progenitor cells using CD90 (Fig. 2B) or CD271 (Fig. 2C), CD90-positive cells were observed at the tissue boundary, however, CD271 staining was absent. As synovium is known to contain a macrophage population in addition to synovial fibroblasts, the tissue biopsies were also stained with CD68 (Fig. 2E) which recognized a cell population at the boundary as well. When the tissue was stained with an antibody to MCP-1 (Fig. 2F), diffuse signal was observed throughout the tissue. These results were then compared with OA synovial membrane biopsies. OA synovial membrane displays a heterogeneous composition, with an increased number of large blood vessels present (Fig. 2G). CD90 staining was observed at the tissue boundary and throughout the synovium (Fig. 2H), as are CD271-positive cells (Fig. 2I). Intense CD68 (Fig. 2K) and MCP-1 (Fig. 2L) staining was observed throughout the synovium.
Evaluation of sMPC Chondrogenic Differentiation with and without MCP-1
Normal sMPCs (positive controls) were placed in chondrogenic differentiation media in the presence or absence of MCP-1, while OA sMPCs were used as a negative control for chondrogenic differentiation. Positive controls self-assemble into three dimensional tissue constructs that stain positive for Alcian blue  (Fig. 3), while OA sMPCs do not demonstrate this capacity and remained as a monolayer of cells (Supporting Information Table 2). When normal sMPCs were treated with MCP-1 (2,500 pg/ml) during the differentiation period, all treatment groups failed to assemble into alcian blue-positive constructs (Fig. 3; Supporting Information Table 2). To account for the possibility that these factors were altering cell viability, cell viability was quantified and found to be not significantly altered in the presence of MCP-1 (Fig. 3).
Chondrogenesis of sMPCs was also examined using quantitative PCR for chondrocyte specific makers (Sox9, Col2A1, and ACAN) (Fig. 4). As expected, normal sMPCs in the absence of cytokines demonstrated a significant up regulation of Sox9, Col2A1, and ACAN over the course of differentiation (Fig. 4). However, OA sMPCs display decreases in Col2A1 and Sox9 expression concurrent with the lack of morphological changes, however ACAN expression was upregulated similar to the normal controls (Fig. 4). Normal sMPCs treated with MCP-1 (Fig. 4), displayed a similar marker expression profile to OA sMPCs, with the exception of Sox9, which was nearly undetectable on all days of differentiation (Fig. 4). To validate the qPCR data, Western blot analysis was used to determine the protein expression levels for Sox9 and Collagen 2. Aggrecan was not selected as its expression levels did not appear to be dramatically effected within any treatment group. The Western blot analysis provided additional confirmation of the qPCR results. As expected, normal sMPCs induced to differentiate produced Collagen 2 and Sox9 protein (Fig. 4), whereas OA sMSCs demonstrated significantly less expression of Collagen 2 and Sox9 (Fig. 4). Furthermore, Collagen 2 was barely discernible (if at all) in sMPCs differentiated in the presence of MCP-1, which is consistent with the significant down regulation relative to the undifferentiated control observed in the qPCR data. Similarly, Sox9 was also downregulated with MCP-1 treatment. GAPDH was used as a loading control and no major fluctuations in loading were detected. In a previous study, we demonstrated that OA sMPCs still retained the ability to undergo chondrogenesis if the cells are pelleted before differentiation was induced . To test whether MCP-1-treated sMPCs still retain chondrogenic potential, the treated sMPCs were induced to form micro-mass pellets and treated with chondrogenic differentiation media. Based on qPCR (Fig. 4G) and Western blot (Fig. 4H) analysis of Sox9, Collagen 2 and Aggrecan expression which increased during the differentiation period, it does appear that MCP-1-treated sMPCs still retain the ability of undergo chondrogenesis if differentiated with a micro-mass methodology.
To determine whether MCP-1 may be playing a more global role in chondrogenesis, primary human articular chondrocytes were harvested from healthy donors and exposed to the same concentration of MCP-1 for the same duration as the sMPCs. qPCR for Sox9, Col2A1, and ACAN demonstrated very small changes in gene expression of all markers at the end of the treatment window compared with the untreated chondrocyte controls (Supporting Information Fig. 1), possibly signifying that MCP-1 is not directly regulating Sox9, Col2A1, and ACAN expression as human chondrocytes are known to express functional MCP-1 receptors .
Based on these results, we next examined whether MCP-1 was able to alter the transcriptome of normal sMPCs.
Micro-Array Analysis of sMPCs
Micro-array analysis was performed on normal untreated sMPCs, sMPCs from OA synovial fluid and normal sMPCs that had been exposed to MCP-1 (Fig. 5). It was observed that OA sMPCs consistently displayed a number of significantly upregulated and downregulated genes compared with normal sMPCs (±fold difference; Supporting Information Tables 3 and 4), and interestingly, when the normal sMPCs were treated with MCP-1, they took on a transcriptional profile that closely resembled (but not completely matched) the expression profile of OA sMPCs (Fig. 5). Furthermore, a subset of genes were noticeably differentially regulated (greater than 10-fold changes) in MCP-1-treated sMPCs compared with both the normal controls and OA sMPCs (Table within Fig. 5).
Examination of Telomerase Activity in sMPCs
During the experiments with treated and untreated sMPCs, it was noticed that cells treated with MCP-1 demonstrated consistent proliferation over an extended period of culture (at least 10 passages) compared with the untreated parental lines, which normally start to show decreased proliferation capacity at passage eight (data not shown). Therefore, we performed a proliferation assay on early (passage two) and late (passage 15) normal, OA, and MCP-1-treated sMPCs. It was observed that normal, OA, and MCP-1-treated sMPCs displayed similar cell proliferation rates at early passages, whereas normal untreated sMPCs demonstrated a significant decrease in proliferation compared with OA and MCP-1-treated cells during later passages (Fig. 6). Based on these results, the telomerase activity of sMPCs derived from normal and OA synovial membrane and synovial fluid was analyzed (Fig. 6). Quite strikingly, it was observed that sMPC lines treated with physiological levels of MCP-1 significantly upregulated the activity of telomerase, while the normal parental untreated lines demonstrated basal telomerase activity (Fig. 6). OA sMPC lines demonstrated increased basal levels of telomerase activity, and also exhibited an increase in enzyme activity following exposure to MCP-1 (Fig. 6). Based on these results, it was assessed if indeed increased numbers of sMPCs are present within synovial fluid of OA patients and it was observed that on average there are five- to sixfold more sMPCs present in the synovial fluid of patients with OA compared with healthy controls (Fig. 6). Furthermore, it was observed that normal sMPCs treated with 2500 pg/ml MCP-1 started to upregulate the expression of MCP-1 mRNA (Fig. 6)
In vivo, it is thought that cartilage lacks extensive endogenous repair capacity as a definitive stem cell population has not been identified within this tissue. However, a number of studies have demonstrated in animal models that stem cells within the synovial membrane can: (A) migrate to cartilage defects and (B) differentiate within the defect into a tissue that morphologically resembles native cartilage tissue [5, 23, 24]. If cartilage can be repaired by stem cells present within the synovial membrane and/or fluid, then the question that needs to be addressed is why an apparent diminished cartilage repair is observed in individuals with OA.
In this study, synovial progenitor/stem cells were derived from highly characterized normal individuals, as well as patients with OA, to observe if any factors expressed in the synovial fluid may be regulating the ability of sMPCs to differentiate into chondrocytes thereby compromising any potential for endogenous cartilage repair. It was decided to focus on MCP-1 for a number of reasons; this mediator was identified from the Luminex array screen and principle component analysis to be a major discriminator between normal and OA synovial fluid (Fig. 1), it has also been implicated in the pathogenesis of human and rodent arthritis in the current literature [56, 57], antagonists against MCP-1 have also proved effective in reducing cartilage damage in rodent models of adjuvant induced arthritis , and finally, MCP-1 is known to play a role in the homing and migration of adult stem/progenitor cells derived from many tissue types [31, 32], and we have verified that sMPCs do migrate towards an MCP-1 “source” (Supporting Information Fig. 2).
Using physiological levels of MCP-1 found within the synovial fluid of OA patients, we were able to demonstrate that MCP-1 inhibited the chondrogenesis of sMPCs at the gene, protein, and primitive tissue levels. Such findings suggest that MCP-1 may be a potent inhibitor of pathways required to undergo chondrogenesis in synovial progenitors; however, it remains unknown whether MCP-1 is inhibiting chondrogenesis (potentially through inhibition of pre-cartilaginous condensation) or inducing the cells to differentiate down another lineage. Furthermore, in this study, it was observed that while MCP-1 treatment blocked the expression of Sox 9 and Collagen 2 when the cells where cultured under chondrogenic inducing conditions, Aggrecan expression was not affected and increased at the message and protein levels during the differentiation period. Although Sox 9 can directly regulate the Aggrecan expression during chondrogenesis, Aggrecan gene expression can also be regulated through L-Sox 5 and Sox 6 , OP-1, BMPs, and TGF-beta  among other factors, and as our differentiation media contained BMP-2 and TGF-beta, it is possible that one or both of these pathways were able to still induce Aggrecan expression in the presence of MCP-1. Furthermore, it has been shown (both directly and indirectly) that Aggrecan may not be a suitable marker for assaying the chondrogenic differentiation of mesenchymal progenitor/stem cells [61-63]. In specific regards to the role MCP-1 might be playing in the regulation of Sox 9 expression, little to no direct evidence has been presented to demonstrate how MCP-1/CCL2 may affect Sox 9 expression, however, IL-6 has been shown to regulate the expression of Sox 9 in bovine chondrocytes , but not in human chondrocytes . However, in both systems, IL-6 regulated the expression of Collagen 2. Taking into account a number of studies that have demonstrated that IL-6 can stimulate expression of MCP-1 and vice versa [66, 67], it is possible that in sMPCs, MCP-1 treatment may increase the expression of IL-6 which regulates the expression of Sox 9 and Collagen 2, but not Aggrecan [54, 65]. Additionally, MCP-1 treatment downregulated the expression of CCR4-NOT (CNOT4) based on the microarray data, and down regulation of the CCR4 subunit (not to be confused with CCR4, the MCP-1 receptor) disrupts Sox 9 expression in human cancer cells . While this result partially supports our observations in this study, it does suggest an indirect link between MCP-1 and Sox9 signaling that will require further study to understand fully.
Leveraging from our previous work demonstrating that OA sMPCs cannot spontaneously undergo condensation in vitro, but can effectively differentiate into chondrocytes with micro-mass/pellet culture , we have observed a similar effect with MCP-1-treated sMPCs in this study, where the treated cells expressed Sox 9 and Collagen 2 under pellet culture conditions. This could signify that MCP-1 inhibition of chondrogenic differentiation of sMPCs is targeting the pre-cartilaginous condensation step that normally occurs in vivo and is generated in vitro by micro-massing/peletting the cells [69-71]. If this is the case, then this “auto-condensation” ability of sMPCs (apparently lacking in bone marrow MSCs ) may play a significant role in endogenous repair of cartilage; however, for this process to be effectively understood, in vivo studies with appropriate animal models will be necessary.
It is also important to note that we did not observe changes in many of the inflammatory proteins observed in OA synovial fluid in the microarray analysis of MCP-1-treated sMPCs. This could be a result of the inflammatory genes being only moderately differentially regulated (our cut off was ± threefold), or that the sMPCs are not the cell type responsible for secreting these factors into the synovial fluid.
Further study is also required to understand the link between MCP-1 and telomerase, as to our knowledge, there are no studies demonstrating any type of telomerase (expression or activity) regulatory function governed by MCP-1; however, there are numerous studies suggesting that MCP-1 may play a role in cancer cell proliferation, therefore the link between proliferation and even transformation in specific regards to MCP-1 needs to be further elucidated.
Although there is currently little evidence suggesting that MCP-1 plays a role in adult stem/progenitor differentiation/specification, there is significant published evidence demonstrating MCP-1 is essential in specifying monocytes to a mature phenotype (reviewed in ref. ). In relation to adult stem/progenitor cells, MCP-1 is widely accepted to play a significant role in the recruitment, migration, and retention of MPCs to areas of injury within the body [31-34]. Furthermore, there is evidence suggesting that MPCs can also express MCP-1 , and we have also found this to be true in our model system following exposure to MCP-1, suggesting a positive feedback loop (Fig. 6). Disregarding the potential inhibitory effects of MCP-1 on sMPCs, it does make sense that MCP-1 would induce MCP-1 expression to foster recruitment and maintenance of MPCs in an area of injury, in this case. it also is understandable why MCP-1 expression would upregulate telomerase to allow for the expansion of the recruitment stem cell pool. In many adult tissues, this chain of events would be followed by differentiation of the stem cell pool into mature phenotypes to assist in the repair/regeneration of the damaged tissue. However, with the chondrogenic inhibition by MCP-1 on synovial MPCs, this could potentially create a nonproductive cycle where stem cells are recruited, activated (proliferation and telomerase), but are inhibited from assisting in repair (Fig. 7). This theoretically could lead to a buildup of stem cells within the synovial fluid and membrane (which has been reported to occur in OA ), but these cells may not able to contribute to effective repair of the articular cartilage (observed in vitro ). Instead, these sMPCs appear to upregulate the expression of MCP-1 and initiate the cycle all over again, recruiting new MPCs that are unable to differentiate once they reach the joint environment. For this hypothesis to be validated, it will require testing in a in vivo system where MCP-1 expression can be effectively regulated to observe if either abolishment of the MCP-1 signal or adjusting the timing of the signal can lead to a repair response in articular cartilage in an OA model.
In conclusion, we have demonstrated that MCP-1 treatment can inhibit chondrogenesis of sMPCs, while concurrently increasing the activity of telomerase within human synovial MPCs possibly through modification of the transcriptional network within the cells. Understanding the mechanism by which MCP-1 “reprograms” these sMPCs may lead to novel pharmaceutical targets and interventions in OA.
We thank Scott Ewald, Gary Rockl, and the SAOTDP as well as Mark Fritzler and Riley Sullivan of Eve Technologies. We also thank Thomas Kryton for his assistance with microscope slide scanning. These studies were supported by funding from the AIHS Osteoarthritis Team Grant, a research grant from Pfizer, the Canadian Stem Cell Network, and the Canadian Arthritis Network.
Disclosure of Potential Conflicts of Interest
The authors indicate no potential conflicts of interest.