Background
Osteoarthritis (OA) is a chronic degenerative disease of the joints and currently its high prevalence increases even further due to aging and an increasingly obese population [
1‐
3]. Current treatment options focus on targeting the symptoms and are limited to analgesics and anti-inflammatory drugs, with many patients eventually having to undergo joint replacement surgery [
4‐
6]. Novel therapies that focus on preventing joint damage are warranted. A better insight into the etiology and disease progression is therefore needed [
3,
4].
The etiology of OA is multi-factorial, and it is considered to be a disease of the whole joint in which the synovium also plays an important role [
7‐
9]. Synovial activation is clearly present in more than 50% of patients with OA and contributes to the pathophysiology and clinical symptoms [
6‐
10]. Systemic and local synovial inflammation is increasingly recognized to be involved in joint pathologic change [
6,
11,
12]. Although most types of leucocytes have been described to be present within the inflamed OA synovium, monocytes and macrophages are thought to be the predominant cell types driving the pathologic change [
13‐
15]. OA is characterized by joint damage, which leads to the release of proteins or alarmins such as S100A8/A9 promoting activation of monocytes/macrophages within the synovium followed by an inherent release of inflammatory cytokines, such as IL-1β, TNF-α and additional release of S100 alarmins [
7,
9,
10,
16]. The most prominent proteins from the S100 family that are released during OA are S100A8 (myeloid-related proteins: MRP8) and S100A9 (MRP14), which belong to the group of damage-associated molecular patterns proteins (DAMPs) or alarmins, and play a crucial role in innate immunity [
17‐
19]. S100A8 and S100A9 form heterodimers under low calcium conditions within the cell and assemble into (S100A8/A9)
2 hetero-tetramers in the presence of calcium [
20]. When myeloid cells are stressed S100A8/A9 is secreted and binds to the toll-like receptor (TLR)4 receptor promoting pro-inflammatory effects [
19,
21]. We previously described that S100A8/A9 proteins are expressed for prolonged periods in collagenase-induced OA (CiOA), a model driven by synovial inflammation, and that they are important stimulators of tissue pathology [
22‐
24]. In addition, patients with OA are characterized by high levels of S100A8/A9 in blood and synovium, and baseline serum levels of patients with symptomatic OA predict development of joint destruction 2 years thereafter [
22,
23]. Moreover, S100A8/A9 promotes the migration of monocytes, which when activated, are also important producers of S100A8/A9 thereby forming a positive feedback loop [
25,
26].
Monocytes are recruited from the bone marrow (BM) towards the site of inflammation via a combined action of adhesion molecules, e.g. LFA-1 and VCAM, and chemokines, such as (C-C motif) ligand 2 (CCL2) that is known to bind the C-C chemokine receptor type 2 (CCR) [
27‐
30]. In mice, two functionally distinct monocyte subpopulations are described. The pro-inflammatory Ly6C
high monocytes, which express high levels of CCR2, are involved in removing debris, and the Ly6C
low monocytes, which express high levels of CX3CR1, are suggested to be involved in repair processes as they release anabolic factors like vascular endothelial growth factor (VEGF) and transforming growth factor (TGF)-β [
31,
32]. Ly6C
high monocytes reside in the BM and upon release from the BM into the circulation can either stay in their state or transform into Ly6C
low monocytes [
32,
33]. Once in the blood, both Ly6C
high as Ly6C
low monocytes can be mobilized to the peripheral tissue in response to chemokines [
34]. Ly6C
high monocytes are mainly attracted by CCL2 and Ly6C
low monocytes by CX3CL1, which can bind to CCR2 and CX3CR1, respectively [
35]. When the monocytes arrive at the injured site, they can differentiate into M1-like pro-inflammatory or M2-like anti-inflammatory macrophages, dependently on the environmental cues [
33]. The aim of our study is to investigate whether local production of S100A8/A9 in the inflamed joint and their subsequent release into the circulation is involved in recruitment of the different monocyte cell populations to the joint in inflammatory CiOA.
Discussion
Recent studies have shown that synovitis significantly contributes to the development of pathologic change in the joint during inflammatory osteoarthritis [
6]. Previously, we demonstrated that the alarmin S100A8/A9 plays an important role in synovitis and joint pathologic change in CiOA [
22]. In the present study, we further explored the involvement of S100A8/A9 in the recruitment of monocyte subpopulations towards the synovium in CiOA. Our findings suggest that prolonged S100A8/A9 production during induction of inflammatory OA locally leads to strongly elevated mobilization of predominantly Ly6C
high monocytes into the joint in OA and from the BM.
The monocyte is the major inflammatory cell type within the synovium throughout the course of inflammatory OA [
13‐
15]. In the mouse, two types of monocytes have been described. Ly6C
high monocytes, expressing high levels of CCR2 and guided by chemokine CCL2 are involved in removing tissue debris and produce mainly pro-inflammatory cytokines like IL-1β and S100A8/A9. In contrast, Ly6C
low monocytes, expressing high levels of CX3CR1 and attracted by CX3CL1, release anabolic factors like VEGF and TGF-β, involved in repairing joint tissue [
31,
32].
The relative number and the interplay between Ly6C
high and Ly6C
low monocytes within the synovium determine the severity of inflammation and regulate further development of pathologic change in the tissues. Previous studies in our laboratory showed that selective removal of resident lining macrophages prior to induction of CiOA largely inhibited monocyte cell influx and development of joint destruction [
40,
42].
In OA, resident synovial macrophages that cover the surface of the synovium are the first cells cleaning the extracellular matrix (ECM) fragments released by damaged joint tissue [
43]. These fragments are recognized by TLRs and scavenger receptors, leading to activation of synovial macrophages thereby releasing high levels of CCL2. This elevates influx of Ly6C
high monocytes favoring further removal of tissue debris.
On the other hand, a shift towards higher numbers of Ly6C
high monocytes within the inflamed synovium may foster pathologic change in the joint by elevated release of pro-inflammatory cytokines. However, only low levels of cytokines like TNF-α, IL-1β, and IL-6 were observed during the first phase of CiOA and probably are only marginally involved in driving pathologic change in the joint. Induction of CiOA in mice lacking IL-1α/β had no effect on either synovitis or on joint pathology [
44]. The outcome was similar using IL-6
-/- and TNF-α
-/- mice (manuscript in preparation). In contrast, alarmin S100A8/A9 was very strongly upregulated within the synovium and measured in significant amounts for prolonged periods up till day 42 after CiOA induction. The heterodimer is released by monocytes and activated macrophages and stimulates nearby synovial cells via TLR4 to release chemokines that attract monocyte populations [
45,
46].
In the present study we found that levels of CCL2, attracting Ly6Chigh monocytes, are much higher in day 7 CiOA compared to saline-injected controls in WT mice, resulting in increased presence of Ly6Chigh monocytes in the synovium. In addition, CCL2 expression was strongly enhanced in the synovium after i.a. injection of S100A8, pointing towards an important role for S100A8/A9 in this process. Whereas in synovium at day 7 CiOA, expression of CCR2/CCL2, characteristic of Ly6Chigh monocytes, was comparable in both WT and S100a9-/- mice, the Ly6Clow-attracting chemokine CX3CL1 and Ly6Clow monocyte population was raised in the synovium of S100a9-/- mice. Since Ly6Chigh monocytes were not elevated in S100a9-/- synovium, but the Ly6Clow monocytes were, S100A8/S100A9 may favor the presence of Ly6Chigh monocytes within the inflamed synovium, not only by chemotactic attraction but also by suppressing their differentiation into the Ly6Clow population thereby maintaining the monocyte population into a more pro-inflammatory state.
Recently it was shown that monocytes have a memory, which is triggered by TLR4 ligands like lipopolysaccharide (LPS) [
47]. This memory is initiated by epigenetic programming and may drive monocytes into a cell type that sustains its pro-inflammatory characteristics [
48]. S100A8/A9 is an important TLR4 ligand released in high levels for prolonged periods in the joint in OA and may be a major mediator driving this epigenetic programming. Ly6C
high monocytes are high producers of S100A8/A9, which may form a positive feedback loop. Finally Ly6C
high and Ly6C
low monocytes are possibly able to differentiate into M1-like and M2-like macrophages, respectively [
49,
50]. The local environment within the synovium strongly influences the signature of the monocytes and their differentiation into their mature forms [
50,
51].
BM and spleen form important reservoirs for Ly6C
high monocytes, which become available under inflammatory conditions. Local induction of CiOA causes high levels of S100A8/A9 within the synovium for prolonged periods; up to 5 μg/mL have been measured in synovial washouts of CiOA synovium [
52]. These proteins leak from the joint into the blood and elevated levels of S100A8/A9 levels were measured throughout the course of OA until the endpoint at day 42. Other pro-inflammatory cytokines like IL-1β, TNF-α, or IL-6 were not detected within the blood.
Interestingly local induction of CiOA in the knee joints of WT mice induced a significant efflux of Ly6C
high monocytes from the BM into the blood when compared to saline-injected WT mice, which was absent at CiOA day 7 in S100a9
-/- mice. This indicates that low-grade local joint inflammation in OA is able to initiate release of Ly6C
high monocytes from the BM. This is in agreement with an earlier study showing that locally induced low-grade inflammation in the heart in myocardial infarction [
32,
53] induced efflux from the BM resulting in elevated local influx of Ly6C
high monocytes within the lesions. Which mechanisms that are initiated within the inflamed joint drive monocyte efflux from the BM is momentarily under investigation. In contrast to conventional pro-inflammatory cytokines, which were undetectable in the blood, S100A8/A9 levels were very high during the course of CiOA. S100A8/A9 in the blood is not only transported as a free protein but also inside extracellular vesicles (EVs). EVs, released by activated immune cells like monocytes/neutrophils, have been shown to contain S100A8/A9 [
54,
55]. They are released at local inflammatory sites and additionally transported to distant areas where they can affect other cells by either fusion with their membrane or when recognized by receptors [
54,
55]. In addition, mechanisms like sympathetic nerve signaling may explain part of the systemic effects observed in the BM. Earlier studies suggested that myocardial infarction liberates monocytes from the BM due to sympathetic nerve signaling thereby boosting atherosclerosis in which pathologic change is highly regulated by the influx of monocyte subpopulations [
53].
Myeloid precursors in the BM differentiate into Ly6C
high monocytes driven by growth factors like PU.1, and M-CSF. Local inflammation promotes the release of Ly6C
high monocytes from the BM into the circulation where they either remain in that state or are transformed into Ly6C
low monocytes [
32]. Interestingly, S100a9
-/- mice did not have elevated efflux of monocytes at CiOA day 7. S100A8/A9 is also abundantly expressed within the BM. Systemic inflammatory triggers and/or sympathetic nerve signaling may promote local release of S100A8/A9 heterodimer within the BM stimulating migration of Ly6C
high monocytes to the blood. This is supported by our findings that expression levels of LFA-1 (Integrin-β2), VCAM, L-selectin, PECAM1, and VE-cadherin were significantly decreased in the BM of S100a9
-/- mice (Additional file
1: Figure S2). It is described that all these adhesion molecules contribute to leucocyte adhesion and migration in the BM [
34,
56]. S100A8/A9 alarmins are strong activators of the beta-2 integrin on myeloid cells and have been shown to be important in transendothelial migration of phagocytes [
26,
57‐
59]. Moreover, previous studies have already shown that S100A8/A9 drives primary BM expansion of myeloid-derived suppressor cells (MDSC) driven by the S100A9/CD33 pathway thereby altering hematopoiesis [
60].
Apart from the BM the spleen also forms an important reservoir for storage of Ly6C
high monocytes [
61] and may contribute to sustaining synovitis in OA. Although it is described that acute inflammation promotes monocyte release from the spleen [
62], we did not find any effect on the ratio of monocyte populations in the spleen of mice at CiOA day 7. This is in agreement with an earlier study in which it was found that in a more severe model of arthritis, splenectomized mice did not differ in ankle swelling or clinical score compared to non-splenectomized mice [
63], suggesting that monocytes released from the spleen do not contribute to synovial inflammation in arthritis.
Although BM efflux of monocytes was significantly raised at CiOA day 7, no effect on the ratio of monocyte subsets was observed within the blood. Although the data as presented in this manuscript strongly suggest a role for S100A8/A9 in the efflux of cells from the BM and into the inflamed joint, we cannot prove a causal relation for this mechanism. Monocytes are released from the BM as Ly6Chigh monocytes and the ratio of Ly6Chigh and Ly6Clow may be rapidly balanced by systemic factors. Tracking studies using fluorescent-labeled monocytes are in progress and will answer to what extent Ly6Chigh monocytes released from the BM are able to reach the joint in OA.
OA is characterized by cartilage and bone destruction, which is related to synovitis [
6]. A privileged presence of Ly6C
high monocytes within the synovium leads to prolonged production of inflammatory cytokines (particularly S100A8/A9) and proteases, which may contribute to joint destruction. Earlier studies showed that S100A8/A9 is crucial in mediating cartilage and bone destruction within the CiOA model [
22]. S100A8/A9 stimulates chondrocytes to produce matrix metalloproteinases (MMPs) thereby degrading the surrounding ECM [
64], which drives synovitis in a positive feedback loop. In addition, S100A8/A9 directly stimulated osteoclastogenesis and strongly increased osteoclast-mediated bone destruction. The human analogs of Ly6C
high and Ly6C
low in the mouse are classical CD14++CD16- and non-classical CD14 + CD16+, respectively. Investigating the balance of these monocyte subsets within inflammatory OA in synovium and BM may give more insight into how synovitis and joint destruction are connected to this crippling disease.
In summary we found that S100A8/A9 is a crucial alarmin involved in favoring recruitment of Ly6Chigh monocytes into the CiOA knee joint. Local production of S100A8/A9 attracts Ly6Chigh monocytes and may suppress their differentiation into Ly6Clow monocytes. Moreover production of S100A8/A9 within the bone marrow may further promote efflux of Ly6Chigh monocytes into the circulation. Ly6Chigh monocytes are potent producers of S100A8/A9 keeping the monocyte population in a pro-inflammatory state and forming a positive feedback loop. Understanding the underlying inflammatory process in the synovium in OA may lead to new therapeutic targets and inhibiting S100A8/A9 may be an interesting therapeutic target to improve the outcome of OA pathology.