Introduction

Osteoarthritis (OA) is a degenerative multifactorial joint disease, characterized by progressive joint failure with pain and disability. OA involves breakdown and loss of articular cartilage, bone deformation, and synovial inflammation [1, 2]. Studies have shown a direct role for inflammatory factors in OA pathogenesis, which now define OA as no longer simply the result of “wear and tear” biomechanical processes [3]. The interaction between trauma and chronic inflammation has been described as a “triggering factor” that activates the immune responses in OA pathogenesis [4].

Synovial Inflammation and OA

OA has traditionally been considered a pathological response to abnormal joint loading and mechanics but low-grade synovitis is now recognized as a common finding [5•]. Many authors agree that in OA patients, the grade of synovitis is frequently associated with: increased local cartilage damage and pain [6]. Depending on the diagnostic technique and the selection criteria for OA patients, synovitis may occur in the majority of patients. Localized proliferative and inflammatory changes of synovium occur in up to 50% of OA patients suggested by arthroscopic studies [7]. The prevalence of synovial hypertrophy detected by ultrasound (US) in OA of the knees was analyzed by meta-analysis and found that synovial hypertrophy was present in 41.5% (95% CI 26.3 to57.7) of patients with knee OA compared to 14.5% (95% CI 0–58.81) of control subjects without OA [8]. MRI assessment of a thousand knee OA patients revealed synovial inflammation, measured by synovial hypertrophy and synovial enhancement, in 60% of OA patients [9]. Whether synovitis is causative of OA or a consequence of joint failure is still not clear; however, synovitis and bone marrow lesions, detected by contrast- enhanced MRI (ceMRI) or Doppler US, have revealed that synovitis and/or bone marrow lesions are predictive of incident OA and the severity of synovitis predicts OA progression [10,11,12,13,14,15,16]. These data are highly suggestive that synovitis may play a pathological role in OA.

Histopathology of OA Synovium

Synovium is constructed of an outer layer, subintima, and an inner layer, intima, and functions to produce and retain synovial fluid in the joint. Typically, the subintima is composed of dense fibrous tissue, type I collagen, adipose tissue, and includes lymphatic vessels, nerve fibers, and microvascular blood vessels [17]. The intima is usually 1–4 cells thick and includes synoviocytes, macrophages, and fibroblasts [17]. In OA, synovial lining hyperplasia, sublining fibrosis, and neovascularization of the stroma occur with infiltration of predominantly macrophages and T cells and some mast cells, B cells, plasma cells, and NK cells [5•]. However, another study suggests that the grade of macrophage infiltration is increased in early OA [18]. Unlike RA synovitis, lymphoid aggregates and germinal center formation are rarely seen in OA synovial samples [19]. Cumulatively, these studies support a model of OA development that includes local damage to the articular cartilage, release of danger-associated molecular patterns (DAMPs) that stimulate macrophage activation, release of pro-inflammatory cytokine and chemokines that lead to recruitment of more macrophages and lymphocytes, increase in angiogenesis, and promotion of chondrocyte dysregulation that leads to secretion of metalloproteinases, pro-inflammatory cytokines, and prostaglandins that induce more cartilage destruction [20] (Fig. 1).

Fig. 1
figure 1

Components of low-grade inflammation associated with osteoarthritis. Multiple inflammatory pathways and mechanisms contribute to the pathogenesis of OA. Tissue injury from trauma, mechanical disturbances, and other mediators including obesity predispose the joint to complement activation and cartilage destruction releasing disease-associated molecular patterns (DAMPS) that can stimulate macrophages and chondrocytes via toll-like receptors (TLRs) to secrete chemokines like CCL2, aggrecanases, metalloproteinases, and pro-inflammatory cytokines like TNF-α and IL-1β. Additional macrophages, T cells, and mast cells are recruited leading to mild synovitis and continued cartilage degradation

Full size image

The Role of Chemokines in OA

There are two well-documented events in OA and their manifestation as synovitis. The first is the infiltration of monocytes and macrophages in the synovial tissue and the second is the increase of cytokines and chemokines in the synovial tissue and synovial fluid [21, 22••]. The infiltration of immune cells in the synovium of OA primarily reflects migration rather than local proliferation and therefore is dependent on soluble mediators, especially chemokines [23]. Chemokines are a family of low-molecular-weight secretory proteins that induce immune cell migration as well as cell activation, angiogenesis, and pain responses and inflammation [23]. Currently, four subfamilies of chemokine receptors have been classified based on the chemokine ligands to which they bind. The relationship between chemokines and their receptors is complex because a given chemokine can bind to several receptors, and each receptor may bind to multiple chemokines. In OA, chemokines are produced by a variety of cells including synovial macrophages and fibroblasts, chondrocytes, and osteoblasts activated by inflammatory mediators or by mechanical stress [23]. Chemokines expressed by human chondrocytes include IL-8/CXCL-8, GROα/CXCL-1, MCP-1/CCL-2, RANTES/CCL-5, MIP-1α/CCL-3, and MIP-1β/CCL-4, and some of them are over-expressed in OA [24, 25].

CCL2 is also known as monocyte chemotactic protein 1 (MCP-1) and has been extensively studied in human OA and animal models of OA [26]. Synovial fluid levels of CCL2 positively correlate with pain and physical disability in patients with OA [27]. Similarly, OA-related pain is significantly decreased in mice deficient in the receptor for CCL2, CCR2, in a destabilized medial meniscus (DMM) model of OA [28]. In this study, cartilage damage and proteoglycan loss were not severe and were not significantly different between wild type and CCR2-deficient mice [28]. However, a recent study showed significant decreases in DMM-induced OA in mice lacking CCR2 or CCL2 [21]. Protection from DMM-induced OA was associated with significantly reduced joint macrophage infiltration [21]. Furthermore, synovial fluid from OA patients had significant levels of CCR2 ligands, including CCL2, CCL7, and CCL8, and synovial biopsies had abundant CCR2+ macrophages that lined, invaded, and were associated with OA cartilage erosions [21]. Importantly, blockade of CCL2/CCR2 signaling significantly attenuated macrophage accumulation, synovitis, and cartilage damage induced by DMM in mice [21]. A third study found that mice deficient in CCR2 or CCL2 had decreased synovial inflammation and delayed pain behavior but only CCR2-deficient mice were protected from developing OA at 20 weeks [29]. Human articular chondrocytes express CCR2, and CCL2 induces increased expression of MMP-3 and proteoglycan loss in vitro [26]. Highlighting the possible role of CCL2 in human OA, a single nucleotide polymorphism study of CCL2 gene found both OA disease-associated and OA protective variants [30]. Taken together, these data support a significant role for CCL2/CCR2 in trauma-induced OA but further studies are needed to establish a role of CCR2 in age or high-fat diet-induced OA.

CCR5 and its ligands CCL5 (RANTES), CCL4 (MIP-1β), and CCL3 (MIP-1α) have also been evaluated in OA [26]. CCL5 and CCL4 levels have been reported to be increased in the OA synovial fluid [31]. However, a more recent paper found no increase in CCL5 or CCL3 in OA samples that had high levels of CCL2 [21]. This study also showed that mice deficient in CCR5 or CCL5 were not protected from DMM-induced OA [21]. In contrast, Takebe et al. found that CCR5-deficient mice were partially protected from DMM-induced OA, with reduced cartilage damage but similar synovitis and bone changes as WT mice [32]. Further studies are needed to determine a role of CCR5 or its ligands in OA.

Other chemokines, such as CXCR2, may have a more homeostatic role in articular cartilage [33]. CXCR2 and its ligand CXCL6 are expressed in adult healthy articular cartilage; CXCL6 is retained in articular cartilage within the extracellular matrix and CXCL6 is not detected in advanced OA samples [33]. Mice with CXCR2 deficiency developed more severe DMM-induced OA with increased chondrocyte apoptosis compared to WT mice [33]. However, CXCR2-deficient mice did not develop spontaneous OA. Similarly, disruption of CXCR1/2 in human and CXCR2 signaling in mouse chondrocytes leads to decreased extracellular matrix production, reduced expression of chondrocyte differentiation markers, and increased chondrocyte apoptosis [34]. Additional studies are needed to determine if exogenous CXCL6 can restore articular chondrocyte health.

The Role of Complement in OA

The complement system is one of the “first lines of defense” in innate immunity, initiating a systemic danger response to cope with an insult [4]. It is composed by over 30 proteins, capable of inducing opsonization of pathogens, and initiating phagocytosis of its own cells and tissues when improperly regulated by inhibitors, autoantibodies, or mutations in complement components [35,36,37]. Complement activation can also increase vascular permeability, recruitment of phagocytic cells, augmentation of acute phase response, and stimulation of B and T lymphocytes [38, 39]. Three distinct pathways can activate the complement cascade. The classical and lectin pathways have their own specific pattern recognition molecules (PRMs), whereas the alternative pathway is activated by a spontaneous hydrolysis of C3 [40,41,42]. All pathways converge into the C3 and C5 convertases, enzymes responsible to mediate the formation of the membrane attack complex (MAC), composed by C5b-C9 [43]. This complex has been extensively demonstrated to lyse target cells through pore formation when deposited on cellular membranes [44]. Importantly, the complement system is involved in physiological and pathological processes in cartilage [45, 46].

Components of the classical (C1s and C4a) and alternative (factor B) pathways, C3 and C5, and the components of MAC are reported to be highly expressed in the synovial membrane and synovial fluid from OA patients with meniscal tears and cartilage degeneration [47,48,49,50,51]. The activation and deposition of complement factors are found in cartilage from OA patients as well as in animal models of OA [24]. Chondrocytes have been shown to synthesize complement components and synthesis can be upregulated by pro-inflammatory cytokines such as IL-1β and TNF-α in OA [51, 52]. Other components of the complement, such as C5a receptors, have been reported to be upregulated on the surface of chondrocytes in OA but not to the same extent as in RA [39]. MAC formation is induced in response to extracellular matrix (ECM) proteins, such as aggrecans, chondroadherin, fibromodulin, osteoadherin, COMP, and type II collagen [53, 54]. However, within the joint, MAC activation is directly or indirectly inhibited by proline/arginine-rich end leucine-rich repeat protein (PRELP), the NC4 domain of type IX collagen, biglycan, decorin, and COMP [42, 55]. This duality may explain why no correlation was seen between cartilage protein fragments and complement proteins in synovial fluid from patients with knee injury [40]. However, a series of recent studies clearly demonstrate the pathological function of the complement system in trauma-induced OA in mouse models.

DMM in mice deficient in C5 or C6 have reduced joint pathology, with less cartilage loss, osteophyte formation, and synovitis [53]. Chondrocytes from C5-deficient mice after DMM surgery also demonstrated lower mRNA expression levels of Jun and Fos pro-inflammatory transcriptional factors, whose expression is induced by MAC in OA [51, 56]. Mice deficient in MAC-inhibitor, CD59a, developed more severe OA as compared to WT mice using three mouse models of OA [53]. In addition, blocking complement activity with a pharmacological drug, such as CR2-fH, a fusion protein that inhibits activation of C3 and C5, attenuates the development of joint damage. In the same way, another study showed that carboxypeptidase B (CPB) appeared to have a protective effect against joint erosion in rheumatoid arthritis and OA by inactivating the complement system [57]. The same group observed that high levels of CPB in the synovial fluid from individuals with OA were associated with high levels of pro-inflammatory cytokines, and complement components which correlated positively with levels of MAC [53, 57]. Similar findings were seen in Cpb2-deficient mice that developed greater cartilage damage than WT mice and had a greater number of osteophytes and degree of synovitis [57]. In in vitro studies of complement activation assays, the CBP-treated serum suppressed the formation of MAC as well as MAC-induced hemolysis, suggesting that CPB has an anti-inflammatory role in OA [57]. Taken together, these data indicate that complement activation and regulation play a central role in trauma-induced OA.

The Role of Macrophages in OA

Macrophage infiltration is common in OA synovial hyperplasia and up to 90% of end-stage knee OA samples have significant infiltration of CD68+ macrophages [19]. Macrophages are the most abundant cells in OA synovium and produce pro-inflammatory cytokines including IL-1β and TNF-α known to induce cartilage breakdown [24]. Seventy-six percent of OA knees have accumulation of macrophages detected in vivo using etarfolatide labeling and SPECT-CT analysis of activated macrophages and the quantity of activated macrophages in the joint is significantly associated with knee pain (R = 0.60, p < 0.0001), joint space narrowing (R = 0.68, p = 0.007), and osteophytes (R = 0.66, p = 0.001) [58•]. These results are highly suggestive that synovitis with activated macrophage recruitment to the joint is pathogenic in OA.

Activated macrophages may develop in response to pathogen-associated molecular patterns (PAMPs) and endogenous DAMPS. Pattern recognition receptors (PRRs) are germ-line encoded innate immune receptors that recognize exogenous PAMPS and endogenous DAMPs. Major families of PRRs include Toll-like receptors (TLR), C-type lectin receptors (CLRs), nucleotide binding and oligomerization domain (NOD)-like receptor family, pyrin and HIN200 domain-containing (PYHIN) family, the RIG-1-like receptor (RLR) family, and oligoadenylate synthase (OAS) proteins. PRRs activate signaling pathways that collectively induce the production of cytokines as well as activation of NLR family, pyrin domain containing 3 (NALP3) inflammasomes, driving activation of caspase-1 and generating biologically active Il-1β and IL-18 and inducing pyroptosis or inflammatory cell death [59]. Dysregulated NALP3 inflammasome responses are commonly found in human diseases including Parkinson’s disease, Alzheimer’s disease, type 2 diabetes, gout, and obesity [59]. Multiple crystals, including uric acid (UA), calcium pyrophosphate, basic calcium phosphate (BCP) crystals such as hydroxyapatite (HA), activate inflammasomes leading to secretion of IL-1β and IL-18 within the joint and likely contributes to OA. NALP3 protein is over-expressed 5.4 fold in OA synovium compared to non-OA synovium [60]. HA crystal deposition and prevalence in joint fluid are found in up to 60% of OA patients and correlate with OA severity [61,62,63]. HA crystals activate the inflammasome in synovial macrophages, leading to IL-1β and IL-18 release into the synovial fluid targeting chondrocytes and synovial lining cells to up regulate cartilage-degrading enzymes and suppress extracellular matrix synthesis, resulting in joint destruction [64]. However, clinical trials using anti- IL-1β therapies have been disappointing [65]. Recent studies have shown that BCP crystals induce Syk, PI3K, and MAPK activation leading to IL-1β, MMP1, and generation of DAMP antigen S100A8 in human macrophages [66]. Syk inhibition effectively prevented these cellular responses and suggests that treatment with Syk inhibitors, several of which are currently in clinical trials for cancer and autoimmune disease, may be advantageous to prevent crystal- and DAMP- induced inflammation and cartilage damage.

The Role of T Cells in OA

A pathological role of T cells in OA is still uncertain but a significant body of literature indicates OA synovium has a rich population of T cells compared to healthy synovium [67, 68]. In OA synovium, T cells are second only to macrophages in frequency and may account for 20–25% of the inflammatory cells [69]. CD4+ T cells are enriched within synovial aggregates compared to CD8+ T cells [70] and OA synovial tissue has increased CD4+/CD8+ ratios approaching 5:1 in OA compared to 2:1 in healthy synovium [67, 68, 70, 71]. Unilateral versus bi-compartmental OA synovial membrane inflammatory cellular infiltrates also differ, with predominantly CD14+ macrophages in unilateral and both macrophages and CD4+ T cells in bilateral OA [72]. OA synovial T cells also have decreased expression of CD3zeta+ mRNA and protein compared to CD3epsilon suggesting that they are chronically stimulated [73]. When adjusted for age, sex, and body mass index (BMI), synovial CD4+ T cells, but not other immune cells, are associated with the pain visual analog scale (VAS) [74]. Peripherally, OA patients also have altered ratios of circulating CD4+/CD8+ T cell ratios [75]. Peripheral blood T cell populations are also altered in aged OA patients [71, 76, 77]. These data indicate that OA is associated with alterations in T cells locally within the joint and peripherally.

In both RA and OA, T regulatory (Tregs) cells (CD4+CD25+/high CD127/−) are increased in peripheral blood and synovial fluid, while Tregs are significantly elevated in RA synovium (p = 0.0335) compared to OA synovium [78]. Interestingly, the Tregs present in RA and OA synovium are very similar, displaying an activated effector memory phenotype (CD45RO + RA-, CD69, CD62L) and Treg functional markers CD152, CD154, CD274, CD279, and GITR [74]. However, despite the increased frequency of CD4 + CD25 + Foxp3+ Tregs in the blood of OA patients, these Tregs have lower secretion of IL-10, likely due to decreased expression of T cell inhibitory receptor T cell immunoglobulin and mucin domain-containing-3 (Tim-3) [79]. Another study found increased Tim-3 + CD4+ T cells in the peripheral blood of OA patients and found that galectin 9, a ligand for Tim-3, was increased in the synovial fluid in less severe grade 2 OA patients compared to grade 4 OA patients [80]. In the synovial fluid, surface Tim-3 was also higher on CD8+ T cells and CD14+ monocytes from grade 2 OA patients and lower in grade 4 OA patients [80]. Importantly, galectin 9 induced Tim-3 + CD4+ Th1 cell apoptosis, suggesting that galectin 9 may inhibit T cell-induced inflammation in early OA [80].

The role of effector T cells including Th1, Th2, Th9, Th22, and Th17 in OA has been recently reviewed [68]. Although pro-inflammatory Th1 cells, producing interferon-γ (IFN-γ) and TNF-α, are present in the sublining layer of synovial membranes of OA, they are much less frequent than in RA synovium [81,82,83]. Th2 cells, induced by IL-4, producing IL-10, IL-4, IL-5, and IL-13, are infrequent in OA synovium and synovial fluid [68]. There is controversy on alternations in OA synovium with some studies showing similar peripheral Th17 cell percentages as healthy controls [84, 85] and while another has shown an increased frequency or presence in OA synovium [77, 81]. Recent work investigated the percentage of T follicular helper (TFH) cells (CXCR5+, CD4+) and serum IL-21, IL-17A, and IFN-γ in 40 OA patients compared to 12 health controls [86]. CXCR5+, CD4+ TFH cell populations were increased in OA patients and the percent of TFH increased with OA grade [86]. Furthermore, OA patients compared to healthy controls had significantly elevated serum IL-21, IL-17A, and IFN-γ levels. Overall, the expression of IL-21 + TFH cells positively correlated with OA disease activity, CRP levels, and WOMAC. The newly described Th9 cells (CD4+ CD8- IL9+), producing IL-9, positively correlate with Western Ontario and McMaster Universities Osteoarthritis Index in OA patients and may represent a marker of disease activity [77]. Collectively, these studies are suggestive that T cells are dysregulated in OA but more data are needed to confirm a pathological role of individual subsets of T cells in OA.

The Role of Other Immune Cells in OA

OA synovial tissue includes other immune cells including B cells, plasma cells, mast cells, and NK cells [67, 74, 87]. While many studies have shown B cells present in low levels in OA, one study has shown moderate to strong B cell staining in 15% of OA patients while 54% had no B cell staining [88]. However, studies are lacking showing a correlation between B cell infiltration and OA progression or severity. NK cells are present in limited numbers in OA synovium and no evidence supports a significant role of NK cells in OA [89]. Synovial biopsies from 56 symptomatic OA and 49 RA patients revealed significantly more mast cells in the OA samples and a positive association between number of mast cells and KL grade [87]. Mechanical loading can induce mast cell degranulation [90]; however, activated mast cells as indicated by substance P are found in a minority (7%) of OA samples compared to 41% of RA synovial biopsies [91]. Additional studies are needed confirm a possible pathological role of mast cells or B cells in OA development.

Conclusion

Recent studies have provided significant mechanistic advances into the role of inflammation and immune cells in the pathophysiology of OA. Traumatic OA initiation clearly requires both complement activation and macrophage infiltration. Future studies aimed at testing local inhibition of complement activation or preventing macrophage infiltration in the immediate post-injury phase will help to determine if targeting these factors will reduce the progression to end-stage OA. Activated macrophages and T cell synovial infiltration predict pain and disease progression in OA but further studies are needed to identify how this low-level inflammation contributes to cartilage damage.