Following challenge under defined in vitro conditions, macrophages activate into a spectrum of phenotypes, where the extremes are characterized by cells exhibiting classical pro-inflammatory (M1) or pro-resolving/healing (M2) responses[51]. In vivo, macrophages respond to variable environmental stimuli, exhibiting a dynamic range of phenotypes that varies by tissue and health status[12,27,51-53]. In synovial joints, classically activated (M1-like) macrophages from the osteoarthritic synovium impede ex vivo chondrogenesis of synovial progenitor cells, while alternatively activated (M2-like) macrophages are required for efficient chondrogenesis[40,54] and inflammation resolution, improving clinical and histological signs of joint disease[49,55]. Collectively, these findings suggest that promoting the M2-like response in diseased joints may provide a therapeutic tool to favor inflammation resolution and tissue repair. Taken together, these observations suggest that harnessing the response of macrophages to enhance the ability of the joint to resolve inflammation can help in the treatment of OA and other joint diseases.

Until recently, specific data about macrophage phenotypes in OA were limited to in vitro studies, end stage OA or experimental animal models[21,40,54,56,57]. Comparative analysis of phenotypical responses between macrophages from healthy and osteoarthritic joints included inferences from other types of arthritic conditions[58]. OA related reports were limited to low amounts of macrophages in synovial fluid, which shed from the synovial membrane after hyperactivation or mechanical separation, and thus are unlikely to reflect the response within the synovium[59,60].

To better understand the roles of macrophages in vivo, we recently assessed the patterns of macrophage phenotype activation in the synovium from healthy equine joints (n = 29) and compared to that of joints with naturally occurring OA (n = 26)[47]. We evaluated the synovial expression of widely used M1-, M2-like and pan macrophages markers [cluster of differentiation (CD)14 (pan-marker), CD86 (M1-like), CD206, and IL-10 (M2-like)] and correlated these findings with synovial histology. Joints with moderate OA were selected to represent a disease stage where the cellular response in the synovium is high, largely attributable to synovial macrophages and also more likely to respond favorably to treatment[61]. We then correlated the findings to cytokine/chemokine profiles in the synovial fluid. Macrophage phenotypes in the synovium were not as clearly defined as they are in vitro. All macrophage markers were expressed with minimal differences between OA and normal joints. However, in OA joints these markers increased proportionate to synovial inflammation, especially CD86. These findings were associated with hyperplasia of the synovial lining, which reflects increased macrophage recruitment and activation in response to injury[47,62].

Among 12 cytokines assayed in that study[47], concentrations of stromal cell-derived factor (SDF)-1 and IL-10 were lower in synovial fluid from osteoarthritic joints, while macrophage chemoattractant protein (MCP)-1 was higher. Upon inflammatory stimuli, synovial fibroblasts release high concentrations of MCP-1. Elevated MCP-1 in synovial fluid from osteoarthritic joints, combined with hyperplasia of the synovial intima characterizes a fibroblast-mediated recruitment of macrophages to the synovium in response to injury[63]. On the other hand, SDF-1 is an essential chemokine in the recruitment of macrophages during inflammation resolution[47,64-66]. Macrophages are the main source of IL-10[7,67], an essential cytokine for cartilage homeostasis and tissue repair[8,9,68,69]. In the face of increased macrophage recruitment, the observation of decreased concentrations of SDF-1 and IL-10 suggests that mechanisms of macrophage-mediated inflammation resolution may be compromised or overwhelmed during OA.

In conclusion, macrophage phenotypes observed in vivo are more diverse and complex than the clear classifications described through in vitro studies[27,39,56,57]. In vivo, synovial macrophages are neither M1 nor M2, but a hybrid phenotype that overall exerts a homeostatic response to injury by driving inflammation to counteract tissue aggressors and trigger tissue repair[27,41,42,46,55,70-74]. Cytokines and cell surface markers frequently used for assessing phenotype identity (M1- or M2-like) in vitro are actually building blocks of a multifaceted immune response and must be prudently analyzed together and in consideration to the dynamic stages of the inflammatory reaction, including its resolution[11,73,75]. Why endogenous recruitment from myeloid macrophages to the joint often fails in recovering synovial homeostasis is not yet known; however, epigenetic mechanisms derived from chronic synovial inflammation seem to affect the performance of local macrophages[76-78]. Developing approaches to recover the homeostatic mechanisms from healthy macrophages may yield improved therapeutic options to prevent or resolve joint inflammation in OA joints and re-establish joint health.

During the last decade, several reports have demonstrated successful use of BMNC in the treatment of chronic synovitis and joint disease[25,46,55,101-104]. Early studies suggesting BMNC could be an effective therapy for joint disease evaluated its effect on “tennis elbow” (elbow lateral epicondylitis), and found a significant improvement using a standardized patient-rated evaluation. Unfortunately, the experimental design did not include an assessment of control groups or a comparative therapy[101]. More recently, Goncars et al[102], evaluated the clinical efficacy of a single joint injection with BMNC to improve pain and other symptoms of moderate knee OA [Kellgren-Lawrence (KL) stage II–III] in a cohort of 28 patients that were followed for 12 mo[104]. The clinical assessments were based on the Knee osteoarthritis outcome score (KOOS) and the Knee society score (KSS) at baseline, 3 mo, 6 mo, and 12 mo after injection, and compared to a patient group (n = 28) treated with 3 consecutive weekly injections of sodium hyaluronate[104]. Significant improvement (P < 0.05) was observed in the BMNC-treated group for all scores at all time points, with special superiority in the KOOS pain subscale for BMNC over the hyaluronic acid-treated group at 6 and 12 mo.

In a second study, the same authors focused on correlating the changes in knee OA symptoms (KL grade II and III) with magnetic resonance imaging (MRI) changes following treatment with a single BMNC injection in a slightly larger cohort (n = 34)[102]. Clinical outcomes were again analyzed using the KOOS and KSS systems at baseline, 3 mo, 6 mo, and 12 mo after injection. Results from MRI were measured using the Whole Organ Magnetic Resonance Imaging Score at baseline and 6-7 mo post injection. Remarkably, 97% of joints treated with BMNC exhibited clinical improvement that was still perceived 12 mo post-treatment in at least 65% of the patients, which still exhibited a difference of more than 10 KOOS points from baseline. Improvements were significant (P < 0.05) for both clinical and imaging scoring systems. Of note, MRI findings revealed that most patients had improvements in synovitis (63%) and bone marrow edema (57 %), but most remarkably, improvements in cartilage were also observed in 38% of patients (Figure 2). No imaging data were available for hyaluronan-treated patients from the previous study. The authors concluded from the second study that a single dose of BMNC (45.56 × 10⁶ ± 34.94 × 10⁶ BMNC containing 1.04 × 10⁶ ± 1.61 × 10⁶ CD34+ hematopoietic stem cells) reduced clinical signs of moderate knee OA, and in some cases, decreased degenerative changes of the joint tissues. Of note, a positive correlation between cell quantity (total BMNC or CD34+ cells) and the improvement in OA symptoms was not observed[102].

Open in New Tab   Full Size Figure   Download Figure

Figure 2 Clinical (Knee injury and osteoarthritis outcome score and Knee society score, n = 34) and imaging (magnetic resonance imaging, n = 30) outcomes from patients with knee Osteoarthritis treated with bone marrow mononuclear cells over 12 mo post-treatment[102]. A and B: Significant improvements (P < 0.05) were observed for Knee injury and osteoarthritis outcome score’ (A) and Knee Society Score (B) scores during the 12-mo follow-up period and most were sustained over time; C: Changes in bone marrow edema, cartilage and synovitis detected by magnetic resonance imaging 6 mo after treatment with bone marrow mononuclear cells. Citation: Goncars V, Kalnberzs K, Jakobsons E, Enģele I, Briede I, Blums K, Erglis K, Erglis M, Patetko L, Muiznieks I, Erglis A. Treatment of Knee Osteoarthritis with Bone Marrow–Derived Mononuclear Cell Injection: 12-Month Follow-up. Cartilage 2019; 10 (1): 26-35. Copyright © The Author(s) 2018. Published by SAGE Publications. KOOS: Knee injury and osteoarthritis outcome Score; KSS: Knee society score; QoL: Quality of life; ADL: Activities of daily living; BME: Bone marrow edema.

Our group recently completed a couple of studies aimed at understanding how normal and inflamed synovial joints respond to injection with BMNC[55], and how BMNC respond to an inflamed environment to produce a positive effect[46]. Special attention was paid to the macrophage component in BMNC. First, using a well-established equine model, we induced synovitis in both radiocarpal joints of 6 horses using lipopolysaccharide. Following 8 h, at the peak of the acute inflammatory response, one inflamed radiocarpal and one normal tarsocrural joint received BMNC injection (20 x 10⁶ viable BMNC, about 79% viability). Saline was injected in the contralateral joints. Synovial fluid was collected at 1 d, 4 d, and 6 d post treatment for cytology, flow cytometry for expression of macrophage markers, and cytokine quantification. Six days post BMNC therapy, following euthanasia, joints were assessed for gross pathology, and the synovium was harvested for histology and immunohistochemistry targeting markers of macrophage activation. At 4 d following treatment with BMNC, inflamed joints exhibited 24% more macrophages with 10% higher counts of IL10+ cells than saline-treated controls. Overall, BMNC-treated joints showed gross and analytical improvements in synovial fluid and synovial membrane, with increasing pro-resolving macrophages and IL-10 concentrations in synovial fluid compared to saline-treated controls. Inflamed joints treated with BMNC were histologically equivalent to healthy joints, whereas saline-treated controls remained abnormal[55] (Figure 3).

Open in New Tab   Full Size Figure   Download Figure

Figure 3 Gross and analytical findings from normal and inflamed equine joints treated with bone marrow mononuclear cells[55]. A: Improvements in synovial fluid associated with decreased cellularity and red cell contamination at 96 h compared to Phosphate buffered saline (PBS)-treated controls; B: Improvements in the synovium were reflected by decreased intra-and peri- and articular synovial hemorrhage and edema, in bone marrow mononuclear cells (BMNC)-(black arrowheads) compared to; C: PBS-treated joints (white arrowheads); D: There was a marked increase in synovial fluid interleukin 10 concentrations at 24 h in inflamed joints, which was much higher in BMNC-treated joints as compared with PBS-treated joints; E and F: BMNC-treated joints exhibited lower scores for all histological aspects of inflammation, although these were only significant (P < 0.05) for vascularity and the composite score(F); Citation: Menarim BC, Gillis KH, Oliver A, Mason C, Ngo Y, Were SR, Barrett SH, Luo X, Byron CR, Dahlgren. Autologous bone marrow mononuclear cells modulate joint homeostasis in an equine in vivo model of synovitis. FASEB J 33, 14337–14353. Copyright © The Author(s) 2019. Published by Wiley Online Library in cooperation with the Federation of American Societies for Experimental Biology. LPS: Lipopolysaccharide; PBS: Phosphate buffered saline; BMNC: Bone marrow mononuclear cells.

To understand how BMNC induced such therapeutic effects, we studied the in vitro response of BMNC to culture in normal (SF) and inflamed autologous synovial fluid (ISF) using cells and synovial fluid from the same horses from the just described in vivo study[55]. Equine BMNC cultured in SF or ISF (n = 8 horses) developed into macrophage-rich cultures exhibiting phenotypes similar to macrophages native to synovial fluid from healthy joints. BMNC confluence (cell proliferation) was ultimately higher in ISF (about 100%) than SF (about 25%). BMNC cultured in SF or ISF were neither M1- nor M2-like but exhibited a range of hybrid phenotypes with a pro-resolving response, characterized over time by decreasing secretion of IL-1β, gradually increasing secretion of IL-10 and IGF-1, and increasing counts of IL-10⁺ macrophages. These changes were sustained over ten days and were more evident in ISF, suggesting that macrophage-mediated mechanisms of homeostasis were conserved over time and were likely favored by the gradual increase in cell proliferation. A combined assessment of data from our in vivo and in vitro studies suggests that joint injection with BMNC can increase the number of synovial macrophages and magnify the macrophage- and IL-10-associated mechanisms of joint homeostasis impaired during the progression of OA[45,46]. Moreover, BMNC therapy preserved, both in vivo and in vitro, the production of cytokines required for tissue repair (PGE₂, IL-10 and IGF-1), the same ones generally impaired by corticosteroids[46,55].

To more fully assess the effects of BMNC therapy in the treatment of naturally occurring equine OA, we conducted a small clinical study evaluating the response of joints (metacarpophalangeal and carpi) with moderate OA to BMNC therapy in 19 adult horses[25]. In the absence of a KL-equivalent scoring system for equine OA, scores were defined by consensus between 6 experienced clinicians. At baseline, horses were subjected to a clinical and musculoskeletal exam, including objective gait analysis at the trot on days 0, 7, and 21 post-treatment with either saline, BMNC (20 x 10⁶ viable cells) or a commonly used dose of triamcinolone (4 mg/joint). After treatment, all horses were rested in stalls for 1 wk, followed by 7 d of stall rest with hand walking, and then a resumption of normal activities in the third week. Lameness decrease was significant (P < 0.05) and consistent only in the BMNC-treated group (between 7 d and 21 d) and associated with increasing counts of macrophages in the synovial fluid. Combined, findings from all of the above-mentioned studies support that BMNC therapy can boost the homeostatic mechanisms of synovial macrophages critical for inflammation resolution.

The use of bone marrow components in the treatment of musculoskeletal conditions dates back from over a century ago and is the foundation for bone marrow grafting as today’s most used techniques for the treatment of non-union fractures[105-107] and osteochondral healing[90,91,108,109]. Newer methods based on the original grafting studies aim at providing means for cells from the subchondral bone marrow to reach the osteochondral defect through microfracture or subchondral bone drilling[109-113]. For therapeutic purposes, removal of most red blood cells and bone spicules from bone-marrow aspirates through centrifugation was a further refinement of the technique because these components were identified as causing severe joint inflammation[114] and heterotopic bone formation[87]. The product resulting from centrifugation of bone marrow aspirate, rich in myeloid leukocytes and other hematopoietic and mesenchymal progenitor cells, is called bone marrow aspirate concentrate (BMAC). The leukocytes included in BMAC, granulocytes and monocytes, not only play crucial roles in inflammation, but also contribute to cellular mechanisms driving tissue repair[11,115-117]. However, the association of neutrophils with chronic inflammation and delayed repair lead clinicians and scientists to avoid their inclusion in biological joint therapies[115]. After removing bone spicules, red blood cells and granulocytes, the mononuclear fraction from the bone marrow (BMNC) is what is left and retains the osteochondral repair properties of bone marrow[90,94]. This is true even when the subchondral bone is not drilled or picked to allow subchondral bone marrow to reach the defect, suggesting that the cells responsible for osteochondral repair are within that BMNC fraction of the bone marrow aspirate[92,118].

The extent of which myeloid macrophage progenitors directly participate in osteochondral repair remains to be defined, but findings from other tissues suggest it goes beyond phagocytic and paracrine activities. Following damage and the subsequent neutrophil-driven acute inflammation that triggers wound healing, the injury site is further populated by mononuclear cells[117,119,120]. These cells display specific monocyte/macrophage markers and are responsible for removing neutrophils by efferocytosis and for the early production of collagen and enzymes required for remodeling the developing granulation tissue, which is largely composed by macrophages[49,121,122]. Following this initial phase, these same cells expressing macrophage markers trans differentiate into cells displaying a different set of markers typically seen in fibroblasts. Cells displaying this ability to trans differentiate are called fibrocytes, are abundant in the bone marrow, and have a central, intrinsic role in tissue development and repair[121-123]. Ongoing research may unveil the transcriptional switches driving these cells to trans differentiate from myeloid to mesenchymal phenotypes, uncovering targets to work on the optimization of cell-based therapies for cartilage regeneration.

Isolation of BMNC from bone marrow aspirates is traditionally done by gradient centrifugation. However, the volume of bone marrow aspirate reported for processing of BMNC varies from 5 to 60 mL and are extrapolated from data optimized for the isolation of MSC[124,125]. Lack of BMNC-specific data may adversely impact the cellular composition of the final BMNC product. To identify an optimal volume of bone marrow aspirate for BMNC isolation, Correa-Letelier et al[126], compared the concentration of BMNC in fractioned bone marrow aspirates in 16 horses. Bone marrow aspirates were obtained in 5 consecutive 5 mL fractions, and the mononuclear cell content of each fraction was quantified using differential cell counts. The total number (median ± QD) of mononuclear cells in the first three fractions (first 15 mL) contained 93.2% of the BMNC in the total 25 mL aspirated, with an average cell viability of 97.5 ± 1.2% (Figure 4). To the authors’ knowledge, similar comparative data for bone marrow aspiration in people is not available. The authors emphasize that, in horses, it is necessary to obtain at least the first 15 mL of bone marrow aspirate from the sternum when the sample is intended for BMNC isolation. Importantly, exceeding the total volume of 25 mL increases the peripheral blood content in the bone marrow aspirate, which has a poor concentration of the pro-resolving cells that are typically found in the bone marrow niche[125,126]. The result is a relative dilution of the desired BMNC with peripheral cells. One useful parameter in assessing if the bone marrow aspirate volume is excessive and diluted by peripheral blood is the platelet count, given that the concentration of platelets in the bone marrow is very low[96]. A detailed protocol for harvesting equine bone marrow aspirates and processing BMNC with minimal resources have been described elsewhere[96]. While only 20-40 x 10⁶ BMNC are used for joint injection, a bone marrow aspirate of 25 mL yields 150-200 x 10⁶ BMNC and the exceeding cells can be cryopreserved for future treatments without compromising BMNC function[127].

Open in New Tab   Full Size Figure   Download Figure

Figure 4 Mononuclear cell content in five consecutive bone marrow aspirates (5 mL each). The first three fractions represent 93.2% of the total 25 mL (the five fractions together) emphasizing the need to obtain at least the first 15 mL of bone marrow aspirate. BMNC: Bone marrow mononuclear cells.