Macrophage phenotypes and monocyte subsets after destabilization of the medial meniscus in mice

Abstract Macrophages play an important role in the development and progression of osteoarthritis (OA). The aim of this study was to identify macrophage phenotypes in synovium and monocyte subsets in peripheral blood in C57BL/6 mice by destabilizing the medial meniscus (DMM), and the association of macrophage subsets with OA features. DMM, sham, and non‐operated knees were histologically assessed between 1 and 56 days for macrophage polarization states by immunohistochemistry (IHC), cartilage damage, synovial thickening, and osteophytes (n = 9 per timepoint). Naive knees (n = 6) were used as controls. Monocyte and polarized synovial macrophage subsets were evaluated by flow cytometry. CD64 and CD206 levels on IHC were higher at early timepoints in DMM and sham knees compared to naive knees. iNOS labeling intensity was higher in DMM and sham knees than in naive knees from d3 onwards. CD163 expression was unaltered at all timepoints. Even though macrophage polarization profiles were similar in DMM and sham knees, only in DMM knees the presence of iNOS and CD206 associated with synovial thickness, and CD163 staining inversely correlated with osteophyte presence. At day 14, monocyte subset distribution was different in peripheral blood of DMM mice compared with sham mice. In conclusion, monocyte subsets in blood and synovial macrophage phenotypes vary after joint surgery. High levels of iNOS+, CD163+, and CD206+ cells are found in both destabilized and sham‐operated knees, and coexistence with joint instability may be a requirement to initiate and exacerbate OA progression.


| INTRODUCTION
Macrophages are present in the synovial lining of joints and are thought to have a prominent role in cartilage degeneration and the development and progression of osteoarthritis (OA). 1 When tissue macrophages become activated, this can result in a spectrum of phenotypes ranging from pro-inflammatory (M1-like), to antiinflammatory or tissue repair-stimulating (M2-like) macrophages. 2 M1-like macrophages have increased microbicidal activity, and secrete inflammatory factors such as interleukin (IL)−1β, IL-6, and chemokine (C-X-C motif) ligand (CXCL)10. M2a-like tissue-repair macrophages express and secrete for instance IL-1RA, CD206, and C-C motif ligand (CCL)18, and have a role in wound healing, tissue repair, and tissue remodeling. M2c-like macrophages secrete and express IL-10 and CD163, and are generally considered immunosuppressive. 3,4 In several studies, increased levels of proinflammatory cytokines typically secreted by activated macrophages were found in synovial fluid of patients who acquired a traumatic injury and were likely to develop OA. 5,6 Furthermore, the synovial fluid of patients who already had developed post-traumatic OA showed elevated cytokine levels that may be ascribed to accumulations of pro-inflammatory macrophages. 1,7,8 Accumulation of proinflammatory M1-like macrophages has been observed in human OA synovial tissue as well as in the collagenase-induced OA (CIOA) mouse model. 9 The CIOA model is associated with a higher degree and differential course of synovial inflammation compared with surgical models of OA involving destabilization of the medial meniscus. 10 In addition, macrophages were shown to be essential for the progression of OA in CIOA mouse models, 11 as depletion of synovial macrophages was shown to diminish osteophyte formation 12 and reduce cartilage destruction. 13 In addition to synovial tissue-resident macrophages, infiltrating monocytes may also participate in OA pathogenesis. Most tissueresident macrophages have an embryological origin and self-renew when necessary. 14,15 During steady-state conditions, circulating monocytes do not contribute to the majority of peripheral tissue macrophage populations. However, under certain inflammatory conditions, monocytes migrate to affected tissues and differentiate to macrophages 16 and thereby also contribute to pathological processes driving OA progression. 17 Human monocytes can be divided into three subsets based on cell surface receptor expression: classical (CD14 ++ CD16 − ), intermediate (CD14 ++ CD16 + ) and non-classical (CD14 + CD16 ++ ) monocytes. 18 In mice, similar peripheral blood monocyte subsets differing in phenotype and function are distinguished 18 but are often categorized according to two subsets. 19 Mouse classical monocytes have high level expression of cell surface proteins Ly6C, CD62L and C-C motif Chemokine Receptor 2 (CCR2), and are rapidly recruited to the site of infection and inflammation, where they contribute to local inflammatory processes and have proteolytic functions. 20 The second subset, comprising non-classical monocytes, shows low level expression of Ly6C and CD62L, and high expression of C-X3-C Motif Chemokine Receptor 1 (CX 3 CR1). Nonclassical monocytes exhibit patrolling behavior involving adherence to and migration along vascular endothelium, they may promote angiogenesis and tissue repair processes, 20 and can extravasate following tissue injury and initiate early inflammatory responses. 21 It was previously reported that classical monocytes were mobilized to synovium 7 days after induction of OA by collagenase injection into mouse knees. 22 Although perturbation of peripheral blood monocytes subsets has been observed in association with various inflammatory conditions, 23  and received subcutaneous 0.05 mg/kg Temgesic (RB Pharmaceuticals) analgesic 30 min before the procedures. The medial meniscotibial ligament (MMTL) was transected as described by Glasson et al. 25 The contralateral knees underwent a sham procedure which entailed the same procedure, with the exception of transection of the MMTL. To take into account differences in biomechanics and gait, six naive knees were obtained from three 16-week-old mice that were euthanized without OA induction. Since there were no previous findings reported regarding the standard deviation of the presence of macrophages during OA, and the use of ranks as nonparametric readout parameter, no sample size calculation was conducted. About 9-12 animals per group were used here as this is a common average sample size used in OA studies. 26,27 All animals were housed in groups of 3-9 in individually ventilated cages including enrichment under a standard 12 h light/dark cycle at the Experimental Animal  Sections of all knees were immunohistochemically stained in one batch to visualize CD64, inducible nitric oxide synthase (iNOS), CD163, and mannose receptor C type 1/CD206 to identify different macrophage phenotypes.

| Scoring of histopathological features of OA and macrophage phenotypes
Structural cartilage damage was assessed in four quadrants (i.e., medial femoral condyle, medial tibial plateau, lateral femoral condyle, and lateral tibial plateau) of four thionin-stained sections, and evaluated according to a modified Pritzker score 28 to be more suitable for scoring articular cartilage damage in mouse knees, as well as with the OARSI score 29 (Table S1). The Pritzker score was determined by multiplying a grade (0-6) and a stage (0-4; Figure S1). Both scoring methods were applied using four sections per quadrant, accounting for a total of 16 scores throughout the entire knee joint. The maximum score of each quadrant was summed to determine the total maximum damage in the knee with a maximal possible score of 96 for the Pritzker score and 24 for the OARSI score. Osteophytes were assessed on the medial and lateral side of the knees and were described as either cartilaginous or bony. Synovial thickness was measured medially and laterally at the height of the parapatellar recess at three locations in four sections using NDP.view v2.6.8 (Hamamatsu), accounting for a total of 24 measurements per knee. The mean of all measurements was used to determine the average synovial thickness of the entire knee.
Of the sections immunohistochemically stained for presence of CD64, iNOS, CD163, and CD206, six to seven sections per knee of the same anatomical regions were ranked based on the increasing intensity in the entire knee. Meaning that all knees including contralateral knees and naive knees, were ranked from least intensely to most intensely stained evaluated by bright field microscopy. Knees that exhibited the same amount of positivity were assigned the median rank of the number of equally ranked knees. This resulted in a maximum score of 111 for CD64, iNOS, and CD163, and 112 for CD206. These numbers varied due sample loss during harvesting, sectioning, or staining. All scorings and rankings were performed in a blinded manner by two independent researchers.

| Flow cytometric analysis of peripheral blood monocytes and synovial tissue
To obtain longitudinal data regarding the monocyte subsets in peripheral blood, OA was induced by DMM in a second set of mice and the sham procedure was performed in a third set of mice (11-12 mice per group).
Peripheral blood was obtained from the facial vein 1, 7, 14, 28, and 56 days after joint surgery. Blood taken 7 days before surgery was considered as baseline. About 50 µl of whole blood was pre-incubated with purified rat antimouse CD16/CD32 (Mouse BD Fc Block; BD Biosciences) for 5 min on ice, and stained for cell surface expression of CD45, CD11b, CD115, CD62L, and Ly6C (All from BioLegend; Table S2)    and OARSI 29 scoring method, and appeared to be generally mild over time ( Figure S4). scoring methods had a significant, strong correlation with each other (ρ s = 0.639) for the total cartilage damage within the joint ( Figure S5). Since the Pritzker score is more sensitive to mild structural changes, this score was used for further analysis.

| Statistics
After DMM surgery, total cartilage damage at end-point was significantly more severe in the knees with DMM than in shamoperated knees ( Figure 1A). The difference between naïve joints at 16 weeks of age and DMM-operated knees was evident, though it did not reach significance (p = 0.08). In addition, cartilage damage

| Profiles of macrophage phenotypes after DMM surgery and their correlation with OA features
The intensity of CD64 staining, a pan-macrophage marker, was higher in DMM than in naive knees (Figure 2A) during the first three days. in DMM knees, the intensity of iNOS staining in the synovial membrane, indicative of of M1-like macrophage polarization, was significantly higher than in naive knees during almost the entire experiment ( Figure 2B). CD163, expressed by several resident macrophage populations and upregulated upon M2c-like macrophage activation, was present throughout the experiment, but the intensities did not differ significantly from naive knees ( Figure 2C).

CONFLICT OF INTERESTS
All the authors declare that there are no conflict of interests.