Figure 2. Pro-inflammatory power of HMGB1: increased synthesis of chemokine, cytokines, iNOS and decreased synthesis of COL2A1. The Figure was drawn using Servier Medical Art, provided by Servier, licensed under a Creative Commons Attribution 3.0 unported license. It was partly modified and adapted with tests.
However, it was largely demonstrated that alarmins do not act alone but in a synergic way with other factors: in particular, HMGB1 increases the IL-1β-induced synthesis of matrix metalloproteinases (MMPs), a disintegrin and metalloproteinse (ADAM), and iNOS by normal human chondrocytes; FN-f shows similar results but with a minor synergic effect on IL-1β stimulated chondrocytes
[63].
Finally, it is important to underline that OA chondrocytes have shown increased expression of RAGE and TLR4, which are HMGB1 receptors
[54][58]. Apoptosis of chondrocytes results in increased secretion of HMGB1 while senescence induces the accumulation of advanced glycation end products (AGEs); therefore, it is possible to assume that the number of AGEs and HMGB1 in cartilage may favour the interaction with RAGE, thus increasing the secretion of pro-inflammatory cytokines and contributing to cartilage degeneration, further confirming its role in OA
[56].
It was described that HMGB1 expression is also augmented in OA synovial membrane compared with the normal membrane
[61][64][65]. The works of Garcìa-Arnandis et al. and Ke and colleagues have shown that the number of HMGB1-positive OA synoviocytes was higher than that in controls, and HMGB1 positivity was found mostly in the cytoplasm and in the extracellular medium rather than in the nuclei; however, the results of the first paper have not reached statistical significance
[65][66]. Sun and colleagues have reported similar evidence: they have found augmented messenger RNA (mRNA) and protein expressions of HMGB1 and RAGE by synovial tissues from knee OA patients, compared with the control group. HMGB1 positivity was predominantly cytoplasmatic; additionally, the authors demonstrated that HMGB1 and RAGE levels were positively correlated with the X-ray grade of the disease
[64]. HMGB1 was also found to increase, same as mRNA and protein expressions, in synovial samples from OA temporomandibular joint compared with patients with a condylar fracture
[67].
HMGB1 seems to act on synoviocytes in the same way as on chondrocytes. In fact, different works have confirmed the inflammation-enhancing power of HMGB1 on synovial cells. It is widely described that this alarmin could increase interleukins, chemokines, and MMPs productions by OA synoviocytes, acting in complex with other pro-inflammatory mediators, such as LPS, IL-1α, and IL-1β. Specifically, the HMGB1-IL-1β combination induces a higher cytokines production than the HMGB1-LPS complex.
HMGB1 has been shown to significantly increase MMP-13, ADAMTS-5, IL-1β, and IL-6 synthesis in human synovial fibroblasts from OA temporomandibular joint (TMJ) through NF-κB signalling
[68]. It was shown that OA human chondrocytes, stimulated by recombinant dsHMGB1, increased synthesis of NF-κB1 (NF-κB p105 subunit) and NF-κB2 (NF-κB p100 subunit) mRNA, two NF-κB light chain enhancers
[57].
Jiang and colleagues have demonstrated that HMGB1 augmented expression and translocation from nucleus to cytoplasm in human chondrocytes incubated with IL-1β is due to bromodomain containing 4 (BRD4) increased expression in cartilage from OA-affected patients, compared with non-OA-affected patients. Additionally, BRD4 expression was positively correlated with histological OARSI score, and this finding may partially explain the gradually increased expression of HMGB1 with OA progression. This protein belongs to the bromo and extra-terminal (BET) domain family and plays an epigenetic regulatory role because of its capacity to bind acetylated lysine residues of histone tails. In fact, researchers have postulated and reported that BRD4 acts by binding the HMGB1 promoter and upstream non-promoter region, inducing gene transcription.
MCM3AP antisense RNA 1 (MCM3AP-AS1) is a long noncoding RNA primarily described in hepatocellular carcinoma. Gao and colleagues have reported increased expression of this RNA in OA patients compared with controls. They have demonstrated that MCM3AP-AS1 acts by binding miR-142-3p, a miRNA that may improve OA by targeting HMGB1, thus reducing apoptosis and inflammation.
MiR-140-5p, such as miR-142-3p, confirmed improved effects on OA, reducing apoptosis, pro-inflammatory factors, and matrix metalloproteinases synthesis in IL-1β-induced chondrocytes. It acts by targeting the 3′-UTR of HMGB1, and low levels of miR-140-5p correspond to the high expression of HMGB1 found in OA tissues and IL-1β-induced chondrocytes. Wang et al. have reported that miR-140-5p interferes with PI3K/AKT, a demonstrated downstream pathway of HMGB1; this miRNA has reduced levels of phosphorylated PI3K and phosphorylated AKT (Ser473), as well as HMGB1, in IL-1β-mediated chondrocytes. Finally, miR-140-5p overexpression has revealed reduced levels of MMP-1, MMP-3, TNF-α, IL-6, and apoptosis rate, and increased cell viability
[69].
MiR-129-5p is another miRNA with described ability to relieve OA alterations. In vitro tests have shown that it targets 3′-UTR of HMGB1 mRNA, reducing the alarmin release and alleviating inflammatory responses and apoptosis of IL-1β-incubated chondrocytes. In fact, IL-1β-mediated chondrocytes incubated with exosomes poor in miR-129-5p have shown increased HMGB1, TLR4, phosphorylated NF-κB, cyclooxygenase 2 (COX2), iNOS, and MMP13 levels and apoptosis rate, while collagen 2 was downregulated. On the other side, the same cells incubated with exosomes rich in miR-129-5p have revealed inverse results.
Plasmacytoma variant translocation 1 (PVT1) is another long noncoding RNA involved in OA pathogenesis. Meng and colleagues have reported higher PVT1 levels in the serum of OA patients and LPS-stimulated human normal chondrocytes C28/I2 than in the controls. On the other side, the miR-93-5p expression has shown a negative correlation with PVT1, with lower levels in the serum of OA patients and LPS-stimulated C28/I2. The researchers have demonstrated that PVT1 is implicated in OA pathogenesis since PVT1 downregulation, through small interference targeting PVT1 (si-PVT1), relieves cytokines secretion, collagen degradation, and apoptosis rate in LPS-incubated C28/I2.
3.2. Role of IL-33 in Osteoarthritis
IL-33 is another DAMP that seems to have a role in OA pathogenesis. It has been reported that OA chondrocytes show an augmented quantity of IL-33 mRNA, in association with augmented levels of IL-37 and other interleukins, expression of receptors, and MMPs, compared with normal human chondrocytes
[70][71][72]. Additionally, IL-33 protein expression has revealed a greater amount in osteoarthritic cartilage than in normal controls
[62][72]; the same results have been found for IL-37, ST2, TLR2, TLR4, NF-κB, IL-6, TNF-α levels, and numerous degradative enzymes
[70][71]. It is possible to assume that IL-33 may have a role in OA development. Mechanical stress is unanimously considered the most impacting factor in OA onset, and for this reason, mRNA of IL-33, MMP1, and MMP13 has been measured and found to be higher in weight-bearing cartilage compared with non-weight-bearing areas. In order to deeply understand if different pathogenesis could define OA according to the type of joint, Rai and colleagues have performed quantitative reverse transcription polymerase chain reaction (qRT-PCR) and immunofluorescence in the two groups of tissues: hip and knee cartilages. They have evidenced that mRNA and protein levels of IL-33, TLR2, TLR4, NF-κB, MMP2, and MMP9 were higher in the OA knee compared with the OA hip, while IL-37 protein expression was higher in the hip than in the knee cartilage
[70].
IL-33 has also demonstrated a pro-inflammatory power; in fact, this alarmin, such as rHMGB1 and LPS, has increased mRNA levels of IL-33, TNF-α, IL-6, TLR2, TLR4, MMP2, MMP9, NF-κB, HMGB1, and RAGE in incubated normal human articular chondrocytes (NHAC) and has determined M1 phenotype conversion of mediated macrophages. Differently, IL-37 has revealed an anti-inflammatory effect on stimulated NHAC. In fact, this interleukin has reduced mRNA expression of several inflammatory mediators, receptors, and HMGB1 in treated cells compared with non-treated cells. Finally, IL-37 has interfered with the effect of IL-33 since mRNA levels of TLR2, TLR4, IL-6, TNF-α, NF-κB, MMP2, MMP9, RAGE, and HMGB1 decreased in cells incubated with IL-37 followed by IL-33, LPS or rHMGB1 compared with cells incubated only with IL-33, LPS or HMGB1.
Great attention has been directed to the molecular pathway of IL-33; in particular, Li and colleagues have demonstrated that IL-33 is released by chondrocytes after double strand RNAs (dsRNAs) bind TLR3. The researchers have realised a three-step experiment: first, they incubated normal human chondrocytes with supernatant from healthy and injured cartilage lysates and then they compared normal human chondrocytes with supernatant from injured cartilage lysates, with and without RNase A. In the last step, they compared results from human chondrocytes stimulated with supernatant from injured cartilage lysates, knocking down TLR3 and TLR7 with siRNA. They have reported that chondrocytes stimulated with damaged cartilage lysate increase mRNA and protein expression of IL-33, MMP1, and MMP13 and reduce collagen II production. The introduction of RNase A in the lysate reduces this effect, and knocking down TLR3 in stimulated chondrocytes is more effective in reducing IL-33, MMP1, and MMP13 levels and increasing collagen type II expression than silencing TLR7.
3.3. Role of S100B in Osteoarthritis
S100B appears to be also involved in OA pathogenesis. Zhu and colleagues have reported that OA human cartilage samples show augmented levels of S100B, TNF-α, and IL-1β, but more interesting is the described strong correlation between the levels of the alarmin and the two cytokines. Furthermore, the authors have demonstrated that human synovial fibroblasts from healthy knee patients with S100B overexpression and incubated in vitro with LPS increase their protein expression of TNF-α and IL-1β and mRNA and protein levels of fibroblast growth factor receptor 1 (FGFR1). This has been confirmed by the inverse results found with S100B knockdown through small interfering RNA (siRNA). Finally, the researchers have demonstrated that this alarmin acts through FGFR1 signalling; in fact, FGFR1 knockdown markedly reduced TNF-α and IL-1β levels in the conditioned medium of S100B-overexpressing fibroblasts, incubated with LPS
[73]. RAGE is an important multiligand receptor whose role in OA has been recently demonstrated. As well as HMGB1, S100B demonstrates binding ability to this receptor. In fact, it has been demonstrated that human chondrocytes incubated with S100B and HMGB1 increase phosphorylated ratios of ERK-1/2 and p65, one of the NF-κB subunits, and protein levels of MMP-13, an enzyme with a key role in OA, for its digestive ability on type II collagen. All these findings are summarised in the
Figure 3.
Figure 3. S100B and osteoarthritis. After ligand–receptor interaction, the MAP kinases system will be activated (only ERK1/2 is represented in the figure), leading to NF-κB activation and transcription of different genes (cytokines, enzymes) in chondrocytes. The Figure was drawn using Servier Medical Art, provided by Servier, licensed under a Creative Commons Attribution 3.0 unported license. It was partly modified and adapted with tests.
Osteoarthritic joints show high levels of HMGB1, S100B, and IL-33; particularly, HMGB1 is increased in synovium and synovial fluid, as well as in cartilage. It is possible to assume that a primary injury against cartilage terminates with an initial discharge of HMGB1 that gives a start to a series of molecular events, turning all joint tissues into damaged. In fact, cell death leads to the passive release of HMGB1 in extracellular space, and it may act on adjacent cells, augmenting the expression of cytokines, chemokines, and enzymes and reducing the anabolic expression of collagen. Like in chondrocytes’ necrosis and/or apoptosis, extracellular matrix-derived molecules may induce active translocation of HMGB1 from the nucleus to the cytoplasm and then in the extracellular medium.
Afterwards, HMGB1-induced cytokines release, such as IL-1β and TNF-α, may determine augmented synthesis and release of HMGB1
[54][55][60]. This way, a positive feedback circuit may be generated with the initiation and amplification of sterile inflammation.
It is necessary to underline that OA arises from reiterated traumas, especially in joints with an unbalanced weight distribution. Hence, repeated impacts may set reiterated cell deaths and HMGB1 releases, constantly powering the circuit. These concepts are sustained by observations explaining that HMGB1 levels are correlated with histological grade of OA severity
[54][55].
High levels of HMGB1 in synovial fluid of OA joints have been found positively correlated with the radiological grade of severity (Kellgren-Lawrence system)
[74] and clinical manifestations
[66]. For all these reasons, alarmins, specifically HMGB1, may represent a completely new therapeutic target. In vitro, tests have shown that HMGB1 A-box, miR-129-5p, miR-142-3p, miR-140-5p, miR-93-5p, miR-107, chrysin, and glycyrrhizin can counter HMGB1 expression and improve the human osteoarthritic chondrocytes phenotype (rate of apoptosis and catabolic activities)
[60][61][69][75][76][77][78][79].
Nowadays, osteoarthritis is increasingly considered an inflammatory disease with systemic involvement. In a mice model of osteoarthritis, high expression of AGEs and HMGB1 has been found in articular structures and has shown a positive correlation with histological damage in cartilage. Similarly, increased levels of different cytokines, in particular IL-1β, have been revealed in sera of the animals and a positive relationship with joint changes has been reported. Therefore, Kyostio-Moore and colleagues have demonstrated that OA in mice models is a disease with local and systemic inflammatory burden
[80].
Different papers have demonstrated the beneficial effects of HMGB1 blocking in animal models of OA. Aulin and colleagues have realised three different modalities of intra-articular injections of anti-HMGB1 in mice with OA induced by anterior cruciate ligament transection (ACLT) of the left knee joint. They have effectuated a single administration at the time of surgery (time 0) in the first group, two administrations at 0 and 2 weeks after the surgery in the second group, and two administrations at 2 and 4 weeks after the surgery in the third group.
HMGB1 has also been demonstrated to play a role in the pathogenesis of murine arthritis models. Ostberg and colleagues have reported that nuclear retention of this alarmin through oxaliplatin intra-peritoneal treatment relieves collagen type II-induced arthritis in mice
[81]. Kokkola and colleagues have illustrated an example of collagen-induced arthritis in mice and rats. The scholars have reported that polyclonal anti-HMGB1 antibodies and HMGB1 A-box intra-peritoneal therapies, reduce clinical disease activity and histologic damage
[82]. The same results have been obtained by Schierbek and colleagues in their collagen-induced arthritis and spontaneous arthritis in mice with DNase type II and interferon type I receptor deficiencies, treated with intra-peritoneal polyclonal anti-HMGB1 antibodies
[83].