Several immune cell populations are involved in cartilage damage, bone erosion, and resorption processes during osteoarthritis. The purpose of this study was to investigate the role of NK cells in the pathogenesis of experimental osteoarthritis and whether and how neutrophils can regulate their synovial localization in the disease. Experimental osteoarthritis was elicited by intra-articular injection of collagenase in wild type and Cxcr3−/− 8-wk old mice. To follow osteoarthritis progression, cartilage damage, synovial thickening, and osteophyte formation were measured histologically. To characterize the inflammatory cells involved in osteoarthritis, synovial fluid was collected early after disease induction, and the cellular and cytokine content were quantified by flow cytometry and ELISA, respectively. We found that NK cells and neutrophils are among the first cells that accumulate in the synovium during osteoarthritis, both exerting a pathogenic role. Moreover, we uncovered a crucial role of the CXCL10/CXCR3 axis, with CXCL10 increasing in synovial fluids after injury and Cxcr3−/− mice being protected from disease development. Finally, in vivo depletion experiments showed that neutrophils are involved in an NK cell increase in the synovium, possibly by expressing CXCL10 in inflamed joints. Thus, neutrophils and NK cells act as important disease-promoting immune cells in experimental osteoarthritis and their functional interaction is promoted by the CXCL10/CXCR3 axis.

Natural killer cells are innate cytotoxic lymphocytes able to kill virus-infected, aberrant, or transformed cells (1, 2). Moreover, NK cells can exert immunoregulatory functions and promote inflammation thanks to their capacity to rapidly release cytokines and growth factors (35). NK cells have been found as a prominent population in the leukocyte infiltrate of the synovial tissue of osteoarthritis patients undergoing total joint replacement, and in other arthropathies (68). However, their role in joint diseases is poorly understood.

Osteoarthritis is the most common joint disease worldwide, mainly affecting the elderly population. Secondary disease can develop in younger individuals as a consequence of an irregular recovery process after joint/bone trauma. Progressive loss of articular cartilage, resulting from the altered balance between degradation and synthesis, leads to abnormal bone remodeling also involving subchondral bone outgrowths at the joint edge that generates osteophytes (9, 10). This unbalance is the result of several pathogenic mechanisms, including enzymatic degradation of the extracellular matrix, defects in new matrix formation, chondrocyte death, and abnormal activation and hypertrophic differentiation of cartilage cells.

Increasing evidence supports a pivotal role of synovitis in osteoarthritis. Synovitis is often found in osteoarthritis patients with active disease at early as well as advanced stages, indicating that a plethora of ongoing immune processes perpetuates local tissue damage and leads to chronic joint inflammation (1113). Significant amounts of inflammatory mediators, including the cytokines TNF, IL-1β, and IL-6, and the chemokines CXCL8 and CCL2 are produced by the synovial cells during osteoarthritis (1418). Although several groups characterized the immune cells infiltrating the synovial tissue of end-stage osteoarthritis patients, the cellular immune players of the inflammatory processes underlying the disease are still unclear (7, 19, 20). NK cells and neutrophils can be a source of proinflammatory mediators, degradation enzymes and growth factors, triggering not only cartilage damage but also altering bone metabolism and repair. Although their function has been linked to several inflammatory diseases, including arthropathies, it is still unclear how and at which stage a particular population migrates to the synovial fluid/synovium and contributes to cartilage/bone damage or remodeling. In addition, several in vitro and in vivo studies have demonstrated the ability of neutrophils to modulate NK cell biology (2124).

In the current study, we investigated the role of chemokines, small structurally related cytokines that exert fundamental roles in leukocyte migration, in the orchestration of reciprocal neutrophil–NK cell interaction during experimental osteoarthritis. To evaluate the relevance of these cell populations in disease progression, the dynamics of their recruitment into joints was analyzed in the early phases of collagenase-induced osteoarthritis (CIOA) (25), in parallel with the chemokine pattern in the synovial fluid, and CIOA was followed in neutrophil- or NK cell–depleted and Cxcr3−/− mice.

Female wild type (WT) Ly5.1 (CD45.1+) and Ly5.2 (CD45.2+) C57BL/6 mice (Charles River, Calco, Italy) and Ly5.2 Cxcr3−/− (B6.129P2-Cxcr3tm1Dgen/J) mice (Jackson Laboratory, Bar Harbor, ME) were housed in filter-top cages in the animal facility of the histology unit (Sapienza University of Rome) under standard conditions (temperature, diet, and water ad libitum). All animal studies were designed according to Animal Research: Reporting of In Vivo Experiments, approved by the Italian Ministry of Health according to art. 7 D.lgs. 116/92 and conducted in accordance with National Guidelines for Animal Care and Use (D.lgs. 116/92). CIOA was elicited in anesthetized mice (8- to 12-wk old) by a single intra-articular (i.a.) injection of 10 U/10 μl of bacterial collagenase in endotoxin-free PBS with Ca2+/Mg2+ (Sigma-Aldrich, St. Louis, MO), as previously described (26). Control mice were i.a. injected with vehicle.

mAbs directly conjugated to FITC, PE, PerCP 5.5, allophycocyanin, PE-cyanine (cy)7, allophycocyanin-eFluor 780, allophycocyanin-cy7, or biotin, and specific for the following Ags (clone name in parentheses) were used in this study: NK1.1 (PK136), CD3ε (145-2C11), CD11b (M1/70), Ly6G (1A8), F4/80 (CI:A3-1), CD69 ([1H].2F3), CD45.2 (104), CD45.1 (A20). Abs and PE-cy7, allophycocyanin-cy7 conjugated streptavidin were purchased from Pharmingen (Becton Dickinson, San Diego, CA), eBioscience (San Diego, CA), and BioLegend (San Diego, CA). F4/80 was from AbD Serotec (Kidlington, U.K.).

To deplete neutrophils in vivo, mice were injected i.p. with 100 μg of anti-Ly6G mAb (clone 1A8; BioXCell, West Lebanon) in 200 μl of endotoxin-free PBS at day −1, +1 and +3 of CIOA induction. NK cells were depleted by i.p injection of 100 μg of anti-NK1.1 mAb (clone PK136; BioXCell) in 200 μl of endotoxin-free PBS at day −2, 0 and +2 of CIOA induction.

At week 1 and 4 after collagenase injection, total knee joints were dissected and processed as previously described (27), stained by Safranin O and Toluidine blue methods and CIOA development was defined in a blinded protocol by the Grade-Stage scoring system recommended by Osteoarthritis Research Society International (28). A tartrate-resistant acid phosphatase (TRAP) kit (Sigma-Aldrich) was used to identify activated osteoclasts. Van Gieson staining was performed to evaluate collagen elastic fibers in cartilage and subchondral bone (27). The classical Jenner-Giemsa method was used to distinguish osteoblasts (basophilic cytoplasm, blue) from osteoclasts (eosinophilic cytoplasm, purple) in subchondral bone marrow (BM) (29). Photos were captured at different magnifications and image analyses to quantify the intensity of collagen staining and to count osteoblasts/osteoclasts in the gated region were accomplished by Image J 1.42 software (Research Services Branch, National Institutes of Health, Bethesda, MD) (30).

Synovial fluids were obtained by injecting 25 μl PBS/1 mM EDTA twice in the intact knee cavity and flushing, centrifugation, and collection of the supernatant. The levels of the following chemokines were analyzed in synovial fluids, by Milliplex MAP (Millipore, Darmstadt, Germany) and Mouse Chemokine 9plex (eBioscience) kits: KC, CXCL1/GROα, CCL2/MCP-1, CCL4/MIP-1β, CCL3/MIP-1α, CXCL10/IP-10, CCL5/RANTES, and CCL7/MCP-3. Chemokine expression is shown as total picograms in two knee joints.

Spleen, blood, and BM were collected and processed as previously described (31). Synovial fluid cells were obtained by washing the knee cavity with 25 μl PBS/1 mM EDTA twice, followed by extensive washing after careful removal of patella and frontal ligaments. Cells were washed and resuspended in staining buffer (PBS without Ca2+/Mg2+, 0.5% BSA, 2 mM EDTA, and 0.05% NaN3) and incubated for 10 min at 4°C with Fc-blocking (24G2) mAb. Then, cells were stained with the indicated fluorochrome-conjugated mAbs for 25 min at 4°C. Flow cytometric analysis was performed using FACSCanto II (Becton Dickinson) and data were elaborated using FlowJo Version 8.5.2 software.

For macrophage and neutrophil isolation, cells were harvested from pooled synovial fluids (n = 7 mice), spleens (n = 2), or BM (n = 2) of mice 15 h after PBS or collagenase injection. Synovial fluid (entire population), spleen, and BM cells (5 × 106/ml) were labeled with fluorochrome-conjugated Ly6G, F4/80 and CD11b mAbs and Ly6G+CD11b+ (neutrophils), and F4/80+CD11b+ (macrophages) cells were sorted using FACSAria (Becton Dickinson). Purity of the isolated cell populations was always ≥98%.

RNA was isolated from sorted populations using a total RNA minikit (Geneaid, Taiwan) and reverse transcribed to cDNA. Real time quantitative PCR was performed with the following primers from Applied Biosystem (Foster City, CA): mouse CXCL10 (Mm00445235_m1) conjugated with fluorochrome FAM and β-actin (Mm01205647_g1) or hypoxanthine phosphoribosyltransferase 1 (Mm00446968_m1) conjugated with fluorochrome VIC. cDNAs were amplified in duplicate using specific TaqMan Gene Expression Assays (Applied Biosystems).

Sorted neutrophils and macrophages were plated on poly-l-lysine–coated multichamber glass plates (2 × 104 cells per chamber). Cells were then fixed with 4% paraformaldehyde, permeabilized with 0.1% Triton-X-100 for 5 min, and stained with the goat anti-CXCL10 polyclonal Ab (R&D, Minneapolis, MI), then FITC-conjugated rabbit anti-goat IgG (Sigma-Aldrich) diluted in PBS and Hoechst Stain solution (Sigma-Aldrich). After extensive washing, cover slips were mounted using SlowFade Gold reagent (Life Technologies) and acquired using an ApoTome Observer Z.1 microscope with ×40/0.75 NA Plan-Neofluar objective and an Axiocam MR equipped with Axio-Vision Version 4.6.3 software for image acquisition (all from Carl Zeiss). Images were processed with Photoshop Version 7 software (Adobe Systems).

C57BL/6 mice CD45.2 were irradiated with 900 rads (two 450 rad doses with an interval of 3 h). After 1 d, irradiated mice were reconstituted by i.v. injection of 1 × 107 BM cells from CD45.1 WT and CD45.2 CXCR3 KO mice at 1:1 ratio. Seven weeks later, reconstituted mice were injected i.a. with collagenase at both knees and sacrificed at day 1 and 3. Cells from synovial fluid, blood, BM, and spleen were stained with anti-NK1.1, -CD3, -CD19, or anti-F4/80, -Ly6G, -CD11b mAbs coupled with CD45.1 and -CD45.2 specific mAbs and analyzed by flow cytometry. To analyze differences between CXCR3−/− and WT NK cell numbers, we determined the CD45.2:CD45.1 ratio of NK1.1+CD3 cells and their respective numbers within tissues. The obtained ratio and numbers were normalized to that of B cells, which do not express CXCR3, used as reference population as described (31).

Unpaired Student t test was used to compare data from experimental groups after FACS and ELISA. Histological analysis was performed in a blinded manner with an interclass correlation ≥0.8 and Kendall’s coefficient of concordance r ≥0.5 showing the degree of agreement between the observers. Histological scores were compared by Mann–Whitney U test. Data were analyzed using Graphpad Prism Software (GraphPad Software, La Jolla) version 6.01. A p value < 0.05 was considered statistically significant.

To identify and characterize the innate immune cells involved in CIOA development, synovial fluid cells of PBS and collagenase injected mice were analyzed and quantified by flow cytometry. Synovial fluid cellularity and CD45+ cell numbers increased at 6 h, day 1, and day 3 following osteoarthritis induction, and returned to basal levels at day 7 (Supplemental Fig. 1A). As shown in Fig. 1A, neutrophils (Ly6G+CD11b+), macrophages (F4/80+CD11b+), and NK cells (NK1.1+CD3) represented a large proportion of CD45+ cells in the synovial fluid, where they were recruited with different kinetics. Neutrophils accumulated at very early time points, reaching a peak at 6 h and progressively decreasing thereafter; macrophage numbers peaked at day 1 and decreased at day 3, whereas NK cells were detectable at 6 h and progressively increased, reaching their maximum at day 3. Higher levels of CD11b on neutrophils and of CD69 on NK cells from CIOA mouse synovial fluid, as compared respectively to control synovial fluid and to the CIOA blood counterpart, indicated an activated phenotype (Fig. 1B). The marked reduction of synovial fluid cells at day 7 corresponded to a deeper infiltration in the joint tissues, consistent with histological identification of cell infiltrates in synovial membrane, proteoglycan/collagen loss in cartilage and osteoclast activation (Supplemental Fig. 1B). In the remodeling phase occurring at week 4, focal cellular nodules were identified together with increased proteoglycan/collagen fibers in cartilage, decreased activation of osteoclasts and osteophyte formation, characteristic features of abnormal bone repair (Supplemental Fig. 1B, 1C)

FIGURE 1.

Neutrophils, macrophages, and NK cells are recruited to collagenase-injected joints with different kinetics. Synovial cells from knee joints of PBS- and collagenase-injected mice were collected and analyzed by flow cytometry, using anti-CD45.2, -NK1.1, -CD3, -Ly6G, -F4/80, -CD11b, and -CD69 specific mAbs. (A) Number and frequency of neutrophils (Ly6G+CD11b+), macrophages (F4/80+CD11b+) and NK cells (NK1.1+CD3) among CD45+ cells are shown in the representative dot plots after gating and in the graphs (column represents mean value ± SEM, n ≥ 8 mice per group in three independent experiments). (B) CD11b expression level (MFI, median fluorescence intensity) on mature neutrophils and percentage of CD69+ NK cells. Graphs on the left represent the mean value ± SEM of three independent experiments (n ≥ 8 mice per group) whereas the histogram plots on the right show one representative sample (IC, isotype control). Student t test, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

FIGURE 1.

Neutrophils, macrophages, and NK cells are recruited to collagenase-injected joints with different kinetics. Synovial cells from knee joints of PBS- and collagenase-injected mice were collected and analyzed by flow cytometry, using anti-CD45.2, -NK1.1, -CD3, -Ly6G, -F4/80, -CD11b, and -CD69 specific mAbs. (A) Number and frequency of neutrophils (Ly6G+CD11b+), macrophages (F4/80+CD11b+) and NK cells (NK1.1+CD3) among CD45+ cells are shown in the representative dot plots after gating and in the graphs (column represents mean value ± SEM, n ≥ 8 mice per group in three independent experiments). (B) CD11b expression level (MFI, median fluorescence intensity) on mature neutrophils and percentage of CD69+ NK cells. Graphs on the left represent the mean value ± SEM of three independent experiments (n ≥ 8 mice per group) whereas the histogram plots on the right show one representative sample (IC, isotype control). Student t test, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Close modal

To analyze the role of neutrophils and NK cells in CIOA onset and progression, mice were depleted of Ly6G+ or NK1.1+ cells according to the protocols shown in Supplemental Fig. 2A or were left untreated. Histological analysis at week 1 showed that disease ameliorated in the absence of either cell type (histological score, Fig. 2A). Details are presented in Fig. 2B, which shows more intact joint architecture with reduced cell infiltration (indicative of reduced synovitis), chondrocyte apoptosis and glycosaminoglycan loss in cartilage of neutrophil-depleted CIOA mice (Safranin O and Toluidine blue staining). Neutrophil removal also resulted in a diminished osteoclast number and activation in subchondral bone (TRAP staining in Fig. 2B), which was confirmed by measuring TRAP activity in joint bone tissue homogenates (Supplemental Fig. 2B). In regards to NK cell-depleted CIOA mice, the decreased histological score at day 7 (Fig. 2A) was due to increased glycosaminoglycans and extensive chondrocyte proliferation in cartilage, and osteophyte formation at joint edges (Safranin O, Toluidine blue staining, Fig. 2B). Furthermore, depletion of NK cells in CIOA mice abrogated osteoclast activation in subchondral bone (TRAP staining, Fig. 2B) and increased osteoblast-related ALP activity in the synovium (ALP enzymatic activity assay of joint bone tissue homogenates, Supplemental Fig. 2B). Altogether our results show that NK cell and neutrophil accumulation in the synovial fluid was detrimental for tissue repair and joint integrity as its inhibition can restore the balance between destruction and repair processes in osteoarthritis.

FIGURE 2.

Neutrophils and NK cells have a disease-promoting role in CIOA. (A) Graph represents mean value of histological score ± SEM in non-depleted versus NK cell and neutrophil depleted CIOA mice (n ≥ 6 mice per group in two independent experiments). Mann–Whitney U test, **p < 0.01, ***p < 0.001, ****p < 0.0001. (B) Paraffin-embedded knee sections of mice non-depleted (left, control PBS, or CIOA) and depleted of Ly6G+ cells (middle) or of NK1.1+ cells (right) were stained by Safranin O, Toluidine blue, TRAP, or Jenner-Giemsa methods. Representative photomicrographs of Safranin O stained knee joint sections made at frontal plane direction show lateral side of femur (F) and tibia (T). Arrows indicate cell infiltration (1); inflammation of synovial membrane (2); glycosaminoglycan loss (3); chondrocyte apoptosis (4); lack of inflammation in synovial membrane (6); normal staining intensity for glycosaminoglycans (7); lack of apoptotic chondrocytes (8); increased glycosaminoglycan density (10); and chondrocyte proliferation (12); osteophyte formation at joint edges (13). In subchondral bone (SB), arrows show activated osteoclasts (Os) near to BM trabeculae (5); lack of activated osteoclasts (9); decreased number of activated osteoclasts (14).

FIGURE 2.

Neutrophils and NK cells have a disease-promoting role in CIOA. (A) Graph represents mean value of histological score ± SEM in non-depleted versus NK cell and neutrophil depleted CIOA mice (n ≥ 6 mice per group in two independent experiments). Mann–Whitney U test, **p < 0.01, ***p < 0.001, ****p < 0.0001. (B) Paraffin-embedded knee sections of mice non-depleted (left, control PBS, or CIOA) and depleted of Ly6G+ cells (middle) or of NK1.1+ cells (right) were stained by Safranin O, Toluidine blue, TRAP, or Jenner-Giemsa methods. Representative photomicrographs of Safranin O stained knee joint sections made at frontal plane direction show lateral side of femur (F) and tibia (T). Arrows indicate cell infiltration (1); inflammation of synovial membrane (2); glycosaminoglycan loss (3); chondrocyte apoptosis (4); lack of inflammation in synovial membrane (6); normal staining intensity for glycosaminoglycans (7); lack of apoptotic chondrocytes (8); increased glycosaminoglycan density (10); and chondrocyte proliferation (12); osteophyte formation at joint edges (13). In subchondral bone (SB), arrows show activated osteoclasts (Os) near to BM trabeculae (5); lack of activated osteoclasts (9); decreased number of activated osteoclasts (14).

Close modal

To clarify the tissue origin of the joint-infiltrating cells, we asked whether the cell increase in CIOA synovial fluid was associated with changes in the cellular pools from BM, spleen, and circulation. To this aim, we compared neutrophil and NK cell tissue distribution at 6 h, day 1, 3, and 7 following collagenase versus PBS injection. When analyzing BM, we found reduced frequency and numbers of neutrophil (6 h and day 1) and NK cells (day 1 and 3), which coincided with their respective increase in the synovial fluid (Fig. 3A, Supplemental Fig. 3); both cell populations remained constant in the spleen, whereas the numbers of peripheral blood neutrophils increased and of NK cells decreased at day 1. These results are indicative of a rapid mobilization of BM neutrophils, followed by a delayed mobilization of BM NK cells into the bloodstream after CIOA induction, likely in charge of replacing and enriching the circulating immune cell pools entering in the synovium. In regard to residual cells in BM, neutrophils showed reduced expression levels of the activation marker CD11b, whereas NK cells displayed progressively increased CD69 expression from 6 h to day 3 (Fig. 3B). This suggests that neutrophils exit from BM in an activated state, whereas NK cells exit in a resting CD69 state and are activated after migration to the synovium.

FIGURE 3.

Joint inflammation influences cell distribution in BM and peripheral blood. Blood, BM, and spleen were collected at 6 h, day 1, 3 and 7 from control (PBS) or CIOA mice and cell suspensions were stained with anti-Ly6G, anti-NK1.1, anti-CD3, anti-CD11b, and anti-CD69 specific Abs. (A) Graphs show the number of NK cells (NK1.1+CD3) and neutrophils (Ly6G+CD11b+). (B) CD11b expression level was analyzed in BM neutrophils following collagenase injections. NK cell activation state was assessed by analysis of CD69 expression in BM. Histograms show mean ± SEM of CD11b (MFI, median fluorescence intensity) expression in BM neutrophils and frequency of CD69+ NK cells (n ≥ 4 mice per group), whereas the right panels show histogram plots from representative samples at 6 h, 1 and 3 d (IC, isotype control). Student t test, *p < 0.05, **p < 0.01.

FIGURE 3.

Joint inflammation influences cell distribution in BM and peripheral blood. Blood, BM, and spleen were collected at 6 h, day 1, 3 and 7 from control (PBS) or CIOA mice and cell suspensions were stained with anti-Ly6G, anti-NK1.1, anti-CD3, anti-CD11b, and anti-CD69 specific Abs. (A) Graphs show the number of NK cells (NK1.1+CD3) and neutrophils (Ly6G+CD11b+). (B) CD11b expression level was analyzed in BM neutrophils following collagenase injections. NK cell activation state was assessed by analysis of CD69 expression in BM. Histograms show mean ± SEM of CD11b (MFI, median fluorescence intensity) expression in BM neutrophils and frequency of CD69+ NK cells (n ≥ 4 mice per group), whereas the right panels show histogram plots from representative samples at 6 h, 1 and 3 d (IC, isotype control). Student t test, *p < 0.05, **p < 0.01.

Close modal

The overall changes of immune cell tissue distribution indicated the involvement of chemotactic mechanisms in innate immune cell increase in the synovial fluid. The pattern of chemokine expression in synovial fluid from control and CIOA mice was thus analyzed to identify the crucial molecules involved in cell recruitment to the destabilized CIOA joint (Fig. 4A): KC and CXCL1 (ligands for CXCR2) mainly increased at 6 h, in accordance with neutrophil infiltration. Similarly, the macrophage chemotactic molecules CCL2 and CCL7 (shared ligands for CCR2) were found elevated at 6 h and gradually declined at later times. CCL4 and CCL5 (shared ligands for CCR1 and CCR5) were detectable only in the CIOA group at 6 h and day 1. Among other chemokines acting on NK cells, CXCL10 (a CXCR3 ligand) levels increased at 6 h reaching a peak at day 1, which coincided with the accumulation of neutrophils and macrophages in the synovial fluid. Furthermore, CXCL10 mRNA could be found in the synovium up to day 7, suggesting that this chemokine is still produced although undetectable in synovial fluids, and may contribute to the early synovial tissue immune cell infiltration (data not shown).

FIGURE 4.

Cell accumulation in synovial fluid is associated to inflammatory chemokine expression during CIOA. (A) Synovial fluids from control and CIOA mice were collected by washing the synovial cavity with PBS, as described in the 2Materials and Methods section. Analysis of chemokine content was performed by multiplex assays for the chemokines indicated. Histograms show mean values of chemokine total amount (picogram) in two knees ± SEM (n ≥ 4 mice per group in three independent experiments). (B) Neutrophils and macrophages from synovial cells, splenocytes, and BM were stained with anti-Ly6G, -F4/80, and -CD11b specific Abs and purified by FACS sorting for RNA extraction. Histograms represent the fold increase in CXCL10 mRNA levels in CIOA synovium-derived neutrophils and macrophages, compared with corresponding cells in BM and spleen (SP), respectively of PBS- and collagen-treated mice. Data were obtained from two independent experiments. (C) Immunofluorescence and microscopic analysis of sorted cells for intracellular CXCL10 expression. Original magnification ×400. Student t test, *p < 0.05, **p < 0.01.

FIGURE 4.

Cell accumulation in synovial fluid is associated to inflammatory chemokine expression during CIOA. (A) Synovial fluids from control and CIOA mice were collected by washing the synovial cavity with PBS, as described in the 2Materials and Methods section. Analysis of chemokine content was performed by multiplex assays for the chemokines indicated. Histograms show mean values of chemokine total amount (picogram) in two knees ± SEM (n ≥ 4 mice per group in three independent experiments). (B) Neutrophils and macrophages from synovial cells, splenocytes, and BM were stained with anti-Ly6G, -F4/80, and -CD11b specific Abs and purified by FACS sorting for RNA extraction. Histograms represent the fold increase in CXCL10 mRNA levels in CIOA synovium-derived neutrophils and macrophages, compared with corresponding cells in BM and spleen (SP), respectively of PBS- and collagen-treated mice. Data were obtained from two independent experiments. (C) Immunofluorescence and microscopic analysis of sorted cells for intracellular CXCL10 expression. Original magnification ×400. Student t test, *p < 0.05, **p < 0.01.

Close modal

The cellular sources and the role of CXCL10 expression in NK cell accumulation in synovial fluids and in disease progression were further investigated as it was the only chemokine active on NK cells whose expression levels persisted after day 1. Because neutrophils and macrophages are documented to produce CXCR3 ligands (3235), we collected synovial cells 15 h post collagenase injection in the knee, when neutrophils and macrophages were equally represented (data not shown). Afterwards, CXCL10 mRNA expression was analyzed in highly purified synovial Ly6G+CD11b+ mature neutrophils and F4/80+CD11b+macrophages. Interestingly, CXCL10 mRNA expression was markedly increased in synovial neutrophils in comparison with BM neutrophils (Fig. 4B). CXCL10 mRNA was also up-regulated in CIOA synovial macrophages, but less than in neutrophils. Intracellular CXCL10 protein was detected by immunofluorescence, confirming the ability of neutrophils to produce this chemokine (Fig. 4C).

Collectively, these results describe a process in which neutrophils, recruited as a consequence of tissue damage, are activated in the CIOA synovium and produce CXCL10.

Given that neutrophils and NK cells were sequentially recruited in the synovial fluid during experimental osteoarthritis, and that neutrophil-produced CXCL10 could promote NK cell migration, we asked if neutrophils were involved in the accumulation of NK cells in CIOA joints by analyzing neutrophil-depleted CIOA mice. Despite neutrophil removal, synovial total cellularity did not change at day 1, and only decreased at day 3 (Fig. 5A). Macrophages were the most abundant population infiltrating the joints of neutrophil-depleted CIOA mice with no change in recruitment kinetics, suggesting that the depletion of neutrophils does not affect macrophage recruitment. On the contrary, neutrophil depletion reduced NK cell accumulation at day 3, but not day 1 (Fig. 5A). Previous evidence indicates that NK cell function is impaired in mice lacking neutrophils and thus NK cell defective migration to the synovium could be attributable to a direct or indirect effect (24). Nevertheless, our preliminary evidence demonstrates that NK cells purified from healthy mice have a reduced in vitro chemotaxis toward synovial fluids of neutrophil-depleted CIOA mice as compared with the undepleted counterpart (Supplemental Fig. 2C), indicating that neutrophil-produced or induced chemotactic factors in synovial fluids are indeed important for NK cell localization.

FIGURE 5.

Neutrophil depletion leads to reduced NK cell numbers and CXCL10/IP-10 protein levels in the synovium. (A) Synovial cells from control (PBS), CIOA, and neutrophil-depleted CIOA (anti-Ly6G) mice were collected at the time points indicated after disease induction and analyzed by flow cytometry after staining for anti-NK1.1, anti-F4/80, and anti-CD11b specific Abs. Graphs show total synovial cellularity (×104, left), macrophages (F4/80+CD11b+ cells, ×104, middle) and NK cells (NK1.1+CD3, ×103, right). Each circle represents a single mouse, whereas the horizontal bar represents the mean value. Data were obtained by at least three independent experiments. (B) Chemokines in the synovium of neutrophil-depleted mice injected with collagenase were compared with those found in non-depleted CIOA mice by multiplex analysis. Histograms show mean ± SEM. At least eight mice per group, in three independent experiments were used. Student t test, *p < 0.05, **p < 0.01.

FIGURE 5.

Neutrophil depletion leads to reduced NK cell numbers and CXCL10/IP-10 protein levels in the synovium. (A) Synovial cells from control (PBS), CIOA, and neutrophil-depleted CIOA (anti-Ly6G) mice were collected at the time points indicated after disease induction and analyzed by flow cytometry after staining for anti-NK1.1, anti-F4/80, and anti-CD11b specific Abs. Graphs show total synovial cellularity (×104, left), macrophages (F4/80+CD11b+ cells, ×104, middle) and NK cells (NK1.1+CD3, ×103, right). Each circle represents a single mouse, whereas the horizontal bar represents the mean value. Data were obtained by at least three independent experiments. (B) Chemokines in the synovium of neutrophil-depleted mice injected with collagenase were compared with those found in non-depleted CIOA mice by multiplex analysis. Histograms show mean ± SEM. At least eight mice per group, in three independent experiments were used. Student t test, *p < 0.05, **p < 0.01.

Close modal

To investigate how neutrophils could affect NK cell recruitment to the joint, we analyzed the chemokine microenvironment in the synovial fluid of neutrophil-depleted mice. Although the chemokine directing neutrophil recruitment, KC, was increased in neutrophil-depleted CIOA mice, we found an impaired increase of several chemokine levels, including CCL2, CCL7, and CXCL10, at day 1 that persisted at day 3 (Fig. 5B).

To understand if CXCL10/CXCR3 axis was involved in NK cell recruitment in the inflamed synovium, we analyzed cell accumulation in CXCR3−/− mice joints at the peak of NK cell increase, i.e., at day 3 after CIOA induction. Both macrophage and NK cell numbers and frequency were reduced by more than 50% in CXCR3-deficient mice, despite constant CD45+ cell and neutrophil numbers, suggesting that CXCR3 is important for recruitment/accumulation of specific cell populations in the CIOA joints (Fig. 6A).

FIGURE 6.

Cxcr3 deficiency dampens NK cell and macrophage accumulation in synovial fluid and protects mice from experimental osteoarthritis. (A) Synovial fluid was collected at day 3 from WT and CXCR3−/− CIOA mice. Synovial cells were collected and stained with mAbs against CD45.2, NK1.1, CD3, Ly6G, F4/80, and CD11b and analyzed by flow cytometry. Graphs represent mean value ± SEM of the number of CD45+ cells, neutrophils (Ly6G+CD11b+), macrophages (F4/80+CD11b+) and NK cells (NK1.1+CD3) (n = 8 knees per group). Student t test: *p < 0.05, **p < 0.01. (B) Seven weeks after the generation of mixed CD45.1+ (WT) and CD45.2+ (Cxcr3−/−) BM chimeras, mice were injected with collagenase or PBS and sacrificed at day 1 and 3 to analyze BM, spleen, and synovial fluid. Graphs show the number of WT and Cxcr3−/− macrophages (F4/80+CD11b+) and NK (NK1.1+CD3) cells in synovial fluid (n ≥ 5 mice per group in two independent experiments). Student t test: *p < 0.05. (C) WT and Cxcr3−/− mice were injected i.a. with collagenase and knee joints were dissected after 1 and 4 wk and assessed by histology. Representative photomicrographs show Safranin O, TRAP and Jenner-Giemsa stained sagittal knee joint sections from WT and Cxcr3−/− mice at 1 wk after injection with PBS or collagenase (CIOA). The lack of cartilage (C) destruction in femur (F) or tibia (T) and of osteoclast (Os) activation are demonstrated along with presence of activated osteoblasts (Ob) in subchondral bone (SB). Arrows point proteoglycan (PG) loss in WT mice and intact cartilage (1) in Cxcr3−/− mice (Safranin O staining), activated osteoclasts (Os) and osteoblasts (Ob). Graph represents mean value of the histological score ± SEM of PBS (n = 4) and CIOA (n = 4) groups. Mann–Whitney U test, **p < 0.01, ***p < 0.001.

FIGURE 6.

Cxcr3 deficiency dampens NK cell and macrophage accumulation in synovial fluid and protects mice from experimental osteoarthritis. (A) Synovial fluid was collected at day 3 from WT and CXCR3−/− CIOA mice. Synovial cells were collected and stained with mAbs against CD45.2, NK1.1, CD3, Ly6G, F4/80, and CD11b and analyzed by flow cytometry. Graphs represent mean value ± SEM of the number of CD45+ cells, neutrophils (Ly6G+CD11b+), macrophages (F4/80+CD11b+) and NK cells (NK1.1+CD3) (n = 8 knees per group). Student t test: *p < 0.05, **p < 0.01. (B) Seven weeks after the generation of mixed CD45.1+ (WT) and CD45.2+ (Cxcr3−/−) BM chimeras, mice were injected with collagenase or PBS and sacrificed at day 1 and 3 to analyze BM, spleen, and synovial fluid. Graphs show the number of WT and Cxcr3−/− macrophages (F4/80+CD11b+) and NK (NK1.1+CD3) cells in synovial fluid (n ≥ 5 mice per group in two independent experiments). Student t test: *p < 0.05. (C) WT and Cxcr3−/− mice were injected i.a. with collagenase and knee joints were dissected after 1 and 4 wk and assessed by histology. Representative photomicrographs show Safranin O, TRAP and Jenner-Giemsa stained sagittal knee joint sections from WT and Cxcr3−/− mice at 1 wk after injection with PBS or collagenase (CIOA). The lack of cartilage (C) destruction in femur (F) or tibia (T) and of osteoclast (Os) activation are demonstrated along with presence of activated osteoblasts (Ob) in subchondral bone (SB). Arrows point proteoglycan (PG) loss in WT mice and intact cartilage (1) in Cxcr3−/− mice (Safranin O staining), activated osteoclasts (Os) and osteoblasts (Ob). Graph represents mean value of the histological score ± SEM of PBS (n = 4) and CIOA (n = 4) groups. Mann–Whitney U test, **p < 0.01, ***p < 0.001.

Close modal

To determine if the role of CXCR3 on NK cell recruitment in the synovial fluids of CIOA mice was direct, we generated mixed CD45.1+ WT and CD45.2+Cxcr3−/− BM chimeras. We found that Cxcr3−/− NK cells had a significant advantage in BM engraftment in agreement with our previous observations (Supplemental Fig. 4A, right panel) (36), whereas no differences could be observed for neutrophils and monocytes (data not shown). Osteoarthritis induction similarly promoted Cxcr3−/− and WT macrophage increase in synovial fluids at all time points analyzed. Cxcr3−/− synovial fluid NK cell number was significantly higher at day 3 but not at day 1 as compared with WT NK cells (Fig. 6B), but this increase was due to higher tissue accumulation of NK cells in Cxcr3−/− mice (i.e., in spleen during CIOA, Supplemental Fig. 4A), as demonstrated by normalizing the number of NK cells in synovial fluid with those in spleen (ratio synovial fluid/spleen: 8.7 ± 1.6 × 10−4 in WT versus 8.7 ± 1.7 × 10−4 in Cxcr3−/−; n = 9). On the other hand, activation of CXCR3-deficient NK cells at day 1 was markedly reduced in the synovium, as shown by an impaired increase of CD69 expression (Supplemental Fig. 4B).

As the CXCL10/CXCR3 axis is involved in the NK cell increase in synovial fluids, and given the pathogenic role of NK cells in CIOA, we analyzed if CXCR3 absence could affect disease development. To this aim, we analyzed disease progression in Cxcr3−/− mice by histological assessment of collagenase-injected joints 1 and 4 wk following the treatment. Interestingly, the histological score of 1-wk old CIOA Cxcr3−/− mice was comparable to that of PBS-injected WT and Cxcr3−/− mice (Fig. 6C). Safranin O staining showed intact cartilage with lack of proteoglycan loss and absence of activated osteoclasts in CIOA Cxcr3−/− mice (Fig. 6C). Histological evaluation demonstrated the accumulation of activated osteoclasts in subchondral bone (TRAP staining) of WT but not in Cxcr3−/− CIOA mice, whereas activated osteoblasts were detected in Cxcr3−/− CIOA mice. No signs of ongoing or resolved disease were evident at 1 wk. Cartilage abnormalities were reduced at 4 wk in Cxcr3−/− mice, and were accompanied by some changes in bone metabolism, as demonstrated by positive Alizarin S red staining (Fig. 6C, Supplemental Fig. 4C). Overall, these results support a crucial role of CXCR3 for disease development (Fig. 7).

FIGURE 7.

Joint injury induces recruitment of neutrophils in the synovium. In addition to directly promoting tissue damage, they can produce CXCL10, which is a crucial mediator in the OA process. This chemokine plays its detrimental role by different mechanisms: it induces recruitment of macrophages (directly) and NK cells (indirectly) and it directly activates NK cells, both cells promoting OA progression. In particular, NK cells are able to interfere with tissue repair mechanisms, delaying the restoration of tissue integrity and contributing to disease severity.

FIGURE 7.

Joint injury induces recruitment of neutrophils in the synovium. In addition to directly promoting tissue damage, they can produce CXCL10, which is a crucial mediator in the OA process. This chemokine plays its detrimental role by different mechanisms: it induces recruitment of macrophages (directly) and NK cells (indirectly) and it directly activates NK cells, both cells promoting OA progression. In particular, NK cells are able to interfere with tissue repair mechanisms, delaying the restoration of tissue integrity and contributing to disease severity.

Close modal

In this study we identified a crucial role of neutrophils and NK cells in the promotion of osteoarthritis using a mouse model characterized by high synovial activation (CIOA), which was shown to resemble the disease in patients displaying high serum levels of inflammatory mediators (37, 38). In addition, we found that neutrophils produce CXCL10 in inflamed synovial fluids, thus contributing to NK cell localization/activation in the joints, and to disease establishment. We found that abrogated immune cell infiltration in CIOA after NK cell depletion or in Cxcr3−/− mice directly impacted on disease severity.

Experimental osteoarthritis induced by enzymatic damage of knee joint develops in two phases: an early phase (until week 1) of cartilage destruction/bone resorption, and a late phase (observed at week 4) of extensive bone remodeling. We observed that cell infiltration and inflammation coexist with both destruction and remodeling processes. Neutrophils were the first cells recruited in the joint (6 h), followed by macrophages (day 1). NK cells were also found in synovial fluids where they accumulated with delayed kinetics (day 3). Thus, the alterations induced rapidly in CIOA caused the increase, likely due to fast recruitment, of damage-sensing cells in synovial fluids in a sequential process likely triggered by extracellular matrix breakdown. Both neutrophils and NK cells decreased in BM, coincidently to their increase in the synovial fluid, suggesting that BM is a major source of immune cells recruited to the inflamed joint (Fig. 3A), as previously documented in other pathologies (39, 40). Thus, understanding the mechanisms regulating BM immune cell mobilization in osteoarthritis may enable to better control the amount and type of cells directed to the joint.

Our data also indicate that neutrophils and NK cells are deleterious for joint integrity as Ab-mediated depletion of either cell type ameliorates disease. The neutrophil pathogenic role is likely mediated by their ability to promote inflammation, and bone destruction due to production of factors directly regulating osteoclast function and bone resorption (41). NK cells were shown to induce osteoclast differentiation by receptor activator of NF κ-B ligand and M-CSF upon activation and to promote osteoblast apoptosis when activated by IL-15, so interfering with bone metabolism toward promotion of bone degradation (8, 42).

In regard to the mechanisms of NK cell and neutrophil recruitment in the joint, accumulating reports have provided evidence of the presence of several chemokines in synovial fluids from osteoarthritis patients that were associated with metalloprotease production and cartilage abnormalities (43, 44). However, no studies have correlated the levels of chemokines in synovial fluid with recruitment/activation of immune cells in experimental osteoarthritis. We found increased levels of neutrophil, macrophage, and NK cell chemotactic molecules in CIOA joints, with their expression paralleling infiltration of the corresponding cell types.

Among the NK cell chemotactic molecules we found, CXCL10 was also present in the joints of osteoarthritis patients at the time of diagnosis or in those who were undergoing total joint replacement (7, 45). Its expression was linked to osteoclastogenic processes mediated by receptor activator of NF κ-B ligand in other arthropathies (46, 47). In addition, this chemokine is crucial for NK cell trafficking in several organs during homeostatic and inflammatory conditions and was shown to increase NK cell degranulation in vitro (36, 4850). A role for CXCL10 on NK cells in osteoarthritis is also supported by the evidence that infiltrating synovial tissue NK cells from osteoarthritis patients uniformly expressed its receptor CXCR3 (7). Thus, the transient and early elevation of chemokines, among which CXCL10 appears to be a key regulator in osteoarthritis, strongly suggests a pivotal role of these molecules in the first wave of leukocyte recruitment to synovium, driving inflammatory osteoarthritis. On the other hand, the role of CXCL10 may be more limited in diseases displaying reduced synovitis as our preliminary results showed only a slight increase of CXCL10 mRNA levels 1 wk after destabilization of mouse medial meniscus, an osteoarthritis model characterized by low synovial activation (data not shown).

We found that in vivo neutrophil removal led to reduced levels of CXCL10 in the synovial fluid and to reduced number of NK cells at day 3, whereas the other candidate CXCL10-producing cells, macrophages, were unchanged at all time points analyzed. Together, these two results strongly suggest that neutrophils are responsible for the early NK cell recruitment/activation in the synovium via CXCL10 production during CIOA (Fig. 7). Correspondingly, the genetic ablation of Cxcr3 diminished NK cell and macrophage number in the synovial fluid at day 3 post collagenase injection, showing that CXCR3 contributes to NK cell accumulation. Indeed, CXCR3 mRNA levels were upregulated at 1 wk after disease induction, suggesting that CXCR3+ cells infiltrate the synovium or that resident cells upregulate CXCR3 expression (data not shown).

Our BM chimera experiments indicate that CXCR3 directly affects NK cell activation in synovial fluid, which is consistent with several known CXCL10-mediated functions (8, 46), whereas its role in NK cell synovial localization is likely linked to CXCR3 activation on other cell populations (Figs. 5B, 7).

Finally, Cxcr3−/− mice failed to develop CIOA, showing that the CXCL10/CXCR3 axis is a key factor for disease induction. This led us to speculate that, in addition to the role in NK cell recruitment/activation, the CXCL10/CXCR3 axis may be at the crossroads of several pathogenic processes, and to propose that the CXCR3 blockade can be considered a novel target for therapeutic intervention of this highly disabling disease.

We thank Dr. Andrea Ponzetta for help in sample collection and interpretation of results.

This work was supported by Inter-Pasteurien Concerted Actions Grant A05_11, France, and grants from the Ministero dell’Istruzione, dell’Università e della Ricerca–Fondo per gli Investimenti della Ricerca di Base (Futuro in Ricerca program and Grant MIUR-L.297 FAR), Istituto Italiano di Tecnologia, and the Sapienza University of Rome.

The online version of this article contains supplemental material.

Abbreviations used in this article:

BM

bone marrow

CIOA

collagenase-induced osteoarthritis

i.a.

intra-articular

TRAP

tartrate-resistant acid phosphatase

WT

wild type.

1
Cerwenka
A.
,
Lanier
L. L.
.
2001
.
Natural killer cells, viruses and cancer.
Nat. Rev. Immunol.
1
:
41
49
.
2
Cooper
M. A.
,
Colonna
M.
,
Yokoyama
W. M.
.
2009
.
Hidden talents of natural killers: NK cells in innate and adaptive immunity.
EMBO Rep.
10
:
1103
1110
.
3
Martín-Fontecha
A.
,
Thomsen
L. L.
,
Brett
S.
,
Gerard
C.
,
Lipp
M.
,
Lanzavecchia
A.
,
Sallusto
F.
.
2004
.
Induced recruitment of NK cells to lymph nodes provides IFN-gamma for T(H)1 priming.
Nat. Immunol.
5
:
1260
1265
.
4
Vivier
E.
,
Tomasello
E.
,
Baratin
M.
,
Walzer
T.
,
Ugolini
S.
.
2008
.
Functions of natural killer cells.
Nat. Immunol.
9
:
503
510
.
5
Fauriat
C.
,
Long
E. O.
,
Ljunggren
H. G.
,
Bryceson
Y. T.
.
2010
.
Regulation of human NK-cell cytokine and chemokine production by target cell recognition.
Blood
115
:
2167
2176
.
6
de Matos
C. T.
,
Berg
L.
,
Michaëlsson
J.
,
Felländer-Tsai
L.
,
Kärre
K.
,
Söderström
K.
.
2007
.
Activating and inhibitory receptors on synovial fluid natural killer cells of arthritis patients: role of CD94/NKG2A in control of cytokine secretion.
Immunology
122
:
291
301
.
7
Huss
R. S.
,
Huddleston
J. I.
,
Goodman
S. B.
,
Butcher
E. C.
,
Zabel
B. A.
.
2010
.
Synovial tissue-infiltrating natural killer cells in osteoarthritis and periprosthetic inflammation.
Arthritis Rheum.
62
:
3799
3805
.
8
Söderström
K.
,
Stein
E.
,
Colmenero
P.
,
Purath
U.
,
Müller-Ladner
U.
,
de Matos
C. T.
,
Tarner
I. H.
,
Robinson
W. H.
,
Engleman
E. G.
.
2010
.
Natural killer cells trigger osteoclastogenesis and bone destruction in arthritis.
Proc. Natl. Acad. Sci. USA
107
:
13028
13033
.
9
Arden
N.
,
Nevitt
M. C.
.
2006
.
Osteoarthritis: epidemiology.
Best Pract. Res. Clin. Rheumatol.
20
:
3
25
.
10
Felson
D. T.
2006
.
Clinical practice. Osteoarthritis of the knee.
N. Engl. J. Med.
354
:
841
848
.
11
Sellam
J.
,
Berenbaum
F.
.
2010
.
The role of synovitis in pathophysiology and clinical symptoms of osteoarthritis.
Nat. Rev. Rheumatol.
6
:
625
635
.
12
Berenbaum
F.
2013
.
Osteoarthritis: when chondrocytes don’t wake up on time.
Arthritis Rheum.
65
:
2233
2235
.
13
Sokolove
J.
,
Lepus
C. M.
.
2013
.
Role of inflammation in the pathogenesis of osteoarthritis: latest findings and interpretations.
Ther. Adv. Musculoskelet. Dis.
5
:
77
94
.
14
Moos
V.
,
Fickert
S.
,
Müller
B.
,
Weber
U.
,
Sieper
J.
.
1999
.
Immunohistological analysis of cytokine expression in human osteoarthritic and healthy cartilage.
J. Rheumatol.
26
:
870
879
.
15
Scanzello
C. R.
,
Umoh
E.
,
Pessler
F.
,
Diaz-Torne
C.
,
Miles
T.
,
Dicarlo
E.
,
Potter
H. G.
,
Mandl
L.
,
Marx
R.
,
Rodeo
S.
, et al
.
2009
.
Local cytokine profiles in knee osteoarthritis: elevated synovial fluid interleukin-15 differentiates early from end-stage disease.
Osteoarthritis Cartilage
17
:
1040
1048
.
16
Kobayashi
M.
,
Squires
G. R.
,
Mousa
A.
,
Tanzer
M.
,
Zukor
D. J.
,
Antoniou
J.
,
Feige
U.
,
Poole
A. R.
.
2005
.
Role of interleukin-1 and tumor necrosis factor alpha in matrix degradation of human osteoarthritic cartilage.
Arthritis Rheum.
52
:
128
135
.
17
Borzi
R. M.
,
Mazzetti
I.
,
Macor
S.
,
Silvestri
T.
,
Bassi
A.
,
Cattini
L.
,
Facchini
A.
.
1999
.
Flow cytometric analysis of intracellular chemokines in chondrocytes in vivo: constitutive expression and enhancement in osteoarthritis and rheumatoid arthritis.
FEBS Lett.
455
:
238
242
.
18
Kapoor
M.
,
Martel-Pelletier
J.
,
Lajeunesse
D.
,
Pelletier
J. P.
,
Fahmi
H.
.
2011
.
Role of proinflammatory cytokines in the pathophysiology of osteoarthritis.
Nat. Rev. Rheumatol.
7
:
33
42
.
19
Bondeson
J.
,
Blom
A. B.
,
Wainwright
S.
,
Hughes
C.
,
Caterson
B.
,
van den Berg
W. B.
.
2010
.
The role of synovial macrophages and macrophage-produced mediators in driving inflammatory and destructive responses in osteoarthritis.
Arthritis Rheum.
62
:
647
657
.
20
Klein-Wieringa
I. R.
,
de Lange-Brokaar
B. J.
,
Yusuf
E.
,
Andersen
S. N.
,
Kwekkeboom
J. C.
,
Kroon
H. M.
,
van Osch
G. J.
,
Zuurmond
A. M.
,
Stojanovic-Susulic
V.
,
Nelissen
R. G.
, et al
.
2016
.
Inflammatory cells in patients with endstage knee osteoarthritis: a comparison between the synovium and the infrapatellar fat pad.
J. Rheumatol.
43
:
771
778
.
21
Oberlies
J.
,
Watzl
C.
,
Giese
T.
,
Luckner
C.
,
Kropf
P.
,
Müller
I.
,
Ho
A. D.
,
Munder
M.
.
2009
.
Regulation of NK cell function by human granulocyte arginase.
J. Immunol.
182
:
5259
5267
.
22
Pliyev
B. K.
,
Kalintseva
M. V.
,
Abdulaeva
S. V.
,
Yarygin
K. N.
,
Savchenko
V. G.
.
2014
.
Neutrophil microparticles modulate cytokine production by natural killer cells.
Cytokine
65
:
126
129
.
23
Costantini
C.
,
Micheletti
A.
,
Calzetti
F.
,
Perbellini
O.
,
Tamassia
N.
,
Albanesi
C.
,
Vermi
W.
,
Cassatella
M. A.
.
2011
.
On the potential involvement of CD11d in co-stimulating the production of interferon-γ by natural killer cells upon interaction with neutrophils via intercellular adhesion molecule-3.
Haematologica
96
:
1543
1547
.
24
Jaeger
B. N.
,
Donadieu
J.
,
Cognet
C.
,
Bernat
C.
,
Ordoñez-Rueda
D.
,
Barlogis
V.
,
Mahlaoui
N.
,
Fenis
A.
,
Narni-Mancinelli
E.
,
Beaupain
B.
, et al
.
2012
.
Neutrophil depletion impairs natural killer cell maturation, function, and homeostasis.
J. Exp. Med.
209
:
565
580
.
25
van der Kraan
P. M.
,
Vitters
E. L.
,
van Beuningen
H. M.
,
van de Putte
L. B.
,
van den Berg
W. B.
.
1990
.
Degenerative knee joint lesions in mice after a single intra-articular collagenase injection. A new model of osteoarthritis.
J. Exp. Pathol. (Oxford)
71
:
19
31
.
26
Ivanovska
N.
,
Dimitrova
P.
.
2011
.
Bone resorption and remodeling in murine collagenase-induced osteoarthritis after administration of glucosamine.
Arthritis Res. Ther.
13
:
R44
.
27
Dimitrova
P.
,
Toncheva
A.
,
Gyurkovska
V.
,
Ivanovska
N.
.
2012
.
Involvement of soluble receptor activator of nuclear factor-κB ligand (sRANKL) in collagenase-induced murine osteoarthritis and human osteoarthritis.
Rheumatol. Int.
32
:
1317
1325
.
28
Pritzker
K. P.
,
Gay
S.
,
Jimenez
S. A.
,
Ostergaard
K.
,
Pelletier
J. P.
,
Revell
P. A.
,
Salter
D.
,
van den Berg
W. B.
.
2006
.
Osteoarthritis cartilage histopathology: grading and staging.
Osteoarthritis Cartilage
14
:
13
29
.
29
Jobke
B.
,
Milovanovic
P.
,
Amling
M.
,
Busse
B.
.
2014
.
Bisphosphonate-osteoclasts: changes in osteoclast morphology and function induced by antiresorptive nitrogen-containing bisphosphonate treatment in osteoporosis patients.
Bone
59
:
37
43
.
30
Shi
L.
,
Liu
S.
,
Wang
D.
,
Wong
H. L.
,
Huang
W. H.
,
Wang
Y. X.
,
Griffith
J. F.
,
Leung
P. C.
,
Ahuja
A. T.
.
2012
.
Computerized quantification of bone tissue and marrow in stained microscopic images.
Cytometry A
81
:
916
921
.
31
Ponzetta
A.
,
Sciumè
G.
,
Benigni
G.
,
Antonangeli
F.
,
Morrone
S.
,
Santoni
A.
,
Bernardini
G.
.
2013
.
CX3CR1 regulates the maintenance of KLRG1+ NK cells into the bone marrow by promoting their entry into circulation.
J. Immunol.
191
:
5684
5694
.
32
Hamilton
T. A.
,
Bredon
N.
,
Ohmori
Y.
,
Tannenbaum
C. S.
.
1989
.
IFN-gamma and IFN-beta independently stimulate the expression of lipopolysaccharide-inducible genes in murine peritoneal macrophages.
J. Immunol.
142
:
2325
2331
.
33
Cassatella
M. A.
,
Gasperini
S.
,
Calzetti
F.
,
Bertagnin
A.
,
Luster
A. D.
,
McDonald
P. P.
.
1997
.
Regulated production of the interferon-gamma-inducible protein-10 (IP-10) chemokine by human neutrophils.
Eur. J. Immunol.
27
:
111
115
.
34
Tecchio
C.
,
Micheletti
A.
,
Cassatella
M. A.
.
2014
.
Neutrophil-derived cytokines: facts beyond expression.
Front. Immunol.
5
:
508
.
35
Chaturvedi
V.
,
Ertelt
J. M.
,
Jiang
T. T.
,
Kinder
J. M.
,
Xin
L.
,
Owens
K. J.
,
Jones
H. N.
,
Way
S. S.
.
2015
.
CXCR3 blockade protects against Listeria monocytogenes infection-induced fetal wastage.
J. Clin. Invest.
125
:
1713
1725
.
36
Ponzetta
A.
,
Benigni
G.
,
Antonangeli
F.
,
Sciumè
G.
,
Sanseviero
E.
,
Zingoni
A.
,
Ricciardi
M. R.
,
Petrucci
M. T.
,
Santoni
A.
,
Bernardini
G.
.
2015
.
Multiple myeloma impairs bone marrow localization of effector natural killer cells by altering the chemokine microenvironment.
Cancer Res.
75
:
4766
4777
.
37
van Lent
P. L.
,
Blom
A. B.
,
Schelbergen
R. F.
,
Slöetjes
A.
,
Lafeber
F. P.
,
Lems
W. F.
,
Cats
H.
,
Vogl
T.
,
Roth
J.
,
van den Berg
W. B.
.
2012
.
Active involvement of alarmins S100A8 and S100A9 in the regulation of synovial activation and joint destruction during mouse and human osteoarthritis.
Arthritis Rheum.
64
:
1466
1476
.
38
Schelbergen
R. F.
,
de Munter
W.
,
van den Bosch
M. H.
,
Lafeber
F. P.
,
Sloetjes
A.
,
Vogl
T.
,
Roth
J.
,
van den Berg
W. B.
,
van der Kraan
P. M.
,
Blom
A. B.
,
van Lent
P. L.
.
2016
.
Alarmins S100A8/S100A9 aggravate osteophyte formation in experimental osteoarthritis and predict osteophyte progression in early human symptomatic osteoarthritis.
Ann. Rheum. Dis.
75
:
218
225
.
39
Eash
K. J.
,
Greenbaum
A. M.
,
Gopalan
P. K.
,
Link
D. C.
.
2010
.
CXCR2 and CXCR4 antagonistically regulate neutrophil trafficking from murine bone marrow.
J. Clin. Invest.
120
:
2423
2431
.
40
Bernardini
G.
,
Sciumè
G.
,
Santoni
A.
.
2013
.
Differential chemotactic receptor requirements for NK cell subset trafficking into bone marrow.
Front. Immunol.
4
:
12
.
41
Wright
H. L.
,
Moots
R. J.
,
Bucknall
R. C.
,
Edwards
S. W.
.
2010
.
Neutrophil function in inflammation and inflammatory diseases.
Rheumatology (Oxford)
49
:
1618
1631
.
42
Takeda
H.
,
Kikuchi
T.
,
Soboku
K.
,
Okabe
I.
,
Mizutani
H.
,
Mitani
A.
,
Ishihara
Y.
,
Noguchi
T.
.
2014
.
Effect of IL-15 and natural killer cells on osteoclasts and osteoblasts in a mouse coculture.
Inflammation
37
:
657
669
.
43
Han
J. H.
,
Choi
S. J.
,
Kurihara
N.
,
Koide
M.
,
Oba
Y.
,
Roodman
G. D.
.
2001
.
Macrophage inflammatory protein-1alpha is an osteoclastogenic factor in myeloma that is independent of receptor activator of nuclear factor kappaB ligand.
Blood
97
:
3349
3353
.
44
Endres
M.
,
Andreas
K.
,
Kalwitz
G.
,
Freymann
U.
,
Neumann
K.
,
Ringe
J.
,
Sittinger
M.
,
Häupl
T.
,
Kaps
C.
.
2010
.
Chemokine profile of synovial fluid from normal, osteoarthritis and rheumatoid arthritis patients: CCL25, CXCL10 and XCL1 recruit human subchondral mesenchymal progenitor cells.
Osteoarthritis Cartilage
18
:
1458
1466
.
45
Erdem
H.
,
Pay
S.
,
Musabak
U.
,
Simsek
I.
,
Dinc
A.
,
Pekel
A.
,
Sengul
A.
.
2007
.
Synovial angiostatic non-ELR CXC chemokines in inflammatory arthritides: does CXCL4 designate chronicity of synovitis?
Rheumatol. Int.
27
:
969
973
.
46
Kwak
H. B.
,
Ha
H.
,
Kim
H. N.
,
Lee
J. H.
,
Kim
H. S.
,
Lee
S.
,
Kim
H. M.
,
Kim
J. Y.
,
Kim
H. H.
,
Song
Y. W.
,
Lee
Z. H.
.
2008
.
Reciprocal cross-talk between RANKL and interferon-gamma-inducible protein 10 is responsible for bone-erosive experimental arthritis.
Arthritis Rheum.
58
:
1332
1342
.
47
Lee
E. Y.
,
Lee
Z. H.
,
Song
Y. W.
.
2009
.
CXCL10 and autoimmune diseases.
Autoimmun. Rev.
8
:
379
383
.
48
Grégoire
C.
,
Chasson
L.
,
Luci
C.
,
Tomasello
E.
,
Geissmann
F.
,
Vivier
E.
,
Walzer
T.
.
2007
.
The trafficking of natural killer cells.
Immunol. Rev.
220
:
169
182
.
49
Taub
D. D.
,
Ortaldo
J. R.
,
Turcovski-Corrales
S. M.
,
Key
M. L.
,
Longo
D. L.
,
Murphy
W. J.
.
1996
.
Beta chemokines costimulate lymphocyte cytolysis, proliferation, and lymphokine production.
J. Leukoc. Biol.
59
:
81
89
.
50
Maghazachi
A. A.
,
Al-Aoukaty
A.
,
Schall
T. J.
.
1996
.
CC chemokines induce the generation of killer cells from CD56+ cells.
Eur. J. Immunol.
26
:
315
319
.

The authors have no financial conflicts of interest.

Supplementary data