A growing body of evidence supports the concept of osteoclasts working as the innate immune cells of the bone (23, 24); accordingly, it is expected that they play a role in the framework of osteomyelitis. Bacterial infection has the potential to influence osteoclast formation directly, through receptor ligation on progenitor cells (9), and indirectly, through enhanced cytokine release from neighboring cells (8). In particular, in vitro S. aureus infection of murine bone marrow-derived osteoclast precursors has been shown to induce their differentiation into activated macrophages that actively secrete proinflammatory cytokines, such as CCL5, MIP-1α, MIP-1β, G-CSF, IL-12p40, and MCP-1 (25). These cytokines enhance the bone resorption capacity of uninfected mature osteoclasts and differentiation of uninfected precursors. Moreover, infection of mature osteoclasts directly promotes cell fusion, which contributes to enhance their ability to resorb bone (25). A similar effect has been also demonstrated upon exposure of human monocyte-derived osteoclasts to Staphylococcal superantigen TSST-1, while the Panton-Valentine leukocidin, known as one of the most powerful pore-forming toxins, and hemolysin-α induced osteoclast death, indicating that the clinical presentation and outcome of bone and joint infection could be related, at least partly, to the toxin profile of the S. aureus isolate involved (26). The osteoclast intracellular infection, which has been demonstrated also in vivo (27), has been recently studied exploiting fluorescently labelled bacterial strains visualized by confocal and time-lapse microscopy (28). Intracellular penetration of bacteria occurred in vitro in a short timeframe; bacterial proliferation started two-four hours post-infection, but the bacterial load in the intracellular compartment of infected osteoclasts did not change over time (28). Osteoclast colonization was accompanied by reduced bactericidal potential compared to non-infected osteoclast precursors (29). Moreover, the proliferative capacity of S. aureus within osteoclasts was dependent on the signaling cascade activated by the osteoclastogenic master transcription factor NFATc1, and varied to some extent amongst individual osteoclasts (29).

S. aureus has been shown to infect osteoblasts and downstream mature cells (osteocytes) both in in vitro and in vivo animal models and clinical biopsies (21, 30–33). The interaction of osteoblasts with S. aureus involves several pathogen-derived factors and components of the osteoblast plasma membrane. For example, a major virulence factor of S. aureus, SpA, interacts directly with osteoblasts via TNFR1, and activates intracellular signaling cascades that result into reduced proliferation, enhanced apoptosis, and impaired mineralizing potential of cultured osteoblasts (34). The binding of S. aureus FnBPA and FnBPB to the extracellular matrix protein fibronectin is recognized as another important process in osteoblast infection. Indeed, fibronectin is believed to act as a “molecular bridge” in linking osteoblasts to the pathogen through α5β1 integrin (35). In this respect, the supramolecular structure of fibronectin in the bone extracellular matrix has been recently shown to play an essential role in bacterial uptake by osteoblasts (36).

Whether bacteria internalization in osteoblasts is achieved passively or through an active process has been debated. It is likely that in fact both mechanisms are in place (37, 38). Intracellular infection has been proposed to occur very rapidly upon exposure to S. aureus. Indeed, bacteria have been detected by immunofluorescence microscopy inside murine calvaria osteoblast-like MC3T3-E1 cells as early as 15 minutes after infection, and the rate of bacteria internalization was found to increase over time (39). Invasion likely requires actin rearrangement at the cell surface; in fact, treatment with the actin depolymerization agent Cytochalasin D significantly reduced S. aureus invasion (40).

Mouton and colleagues recently demonstrated that S. aureus internalization (assessed in vitro through a gentamycin/lysostaphin protection assay that allows eliminating adherent and non-adherent bacteria, while sparing intracellular pathogen) induced an impairment in early osteoblast differentiation by interfering with β1 integrin signaling, leading to decreased expression of RUNX2 and COL1α1 and ALP activity. Accordingly, an internalization defective S. aureus strain lacking fnbpA expression did not elicit this effect (41). Consistent with this, in vivo infection with the S. aureus strain capable of cell internalization altered some bone histomorphometric parameters, supporting the hypothesis that osteoblast functions are impaired upon intracellular bacterial infection (41).

Osteoblast colonization by S. aureus stimulates the focal adhesion kinase (FAK)/epidermal growth factor receptor (EGFR) and c-Src signaling pathways by enhancing their phosphorylation in a time-dependent fashion; on the contrary, inhibition of the EGFR/FAK or c-Src signaling pathways significantly reduces the rate of pathogen internalization (39). Consequently, these pathways could be targeted in parallel to standard antibiotic therapy of chronic S. aureus osteomyelitis.

Osteoblast infection by S. aureus has been confirmed also in the presence of invasive MRSA infection of the human MG-63 osteosarcoma cell line, using imaging flow cytometry (IFC) which is more sensitive and reproducible than conventional cell culture methods (42). Pathogen uptake is known to vary depending on the strain irrespectively of antibiotic resistance (43). Statistical analysis of results obtained by IFC assays demonstrated that intracellular persistence capacity of several different MRSA strains over a 24 hour-timeframe was correlated with the total number of infected cells at 24 hour-post-infection and not with the number of bacteria that managed to enter/replicate in each single cell, defined by spot counting after cell transient permeabilization and pathogen staining with a membrane-impermeable green-fluorochrome vancomycin analogue (42). This would suggest that other factors besides the specific clone define the bacteria ability to internalize and persist inside osteoblasts. Future research is needed on this point that is relevant to chronicization of infection.

The encounter of the pathogen with osteoblasts results into release of cytokines and chemokines (as discussed later on) that recruit and activate immune cells; increased RANKL production that sustains osteoclast activity; impaired bone matrix production and mineralization; and ultimately osteoblast death through upregulation of the cell death signal transducer TRAIL and its cell surface death receptors, and concomitant decrease of the decoy receptor OPG (44, 45). All these events contribute to bone loss. Progression from acute to chronic bone and joint infections is accompanied by phenotypic adaptation of the pathogen to a less virulent form, called “small colony variant”. This is characterized by increased intracellular persistence and antibiotics resistance, and reduced cytokine release induction and immune system stimulation (46, 47). In particular, osteocytes have been recently demonstrated to constitute a reservoir for silent or persistent infection owing to reduced antimicrobial capacity to eliminate intracellular bacteria, and higher pathogen survival in the form of small colony variants (11, 48). Moreover, infected osteocytes can also elicit an inflammatory response that contributes to communication with other skeletal cells, immune cell recruitment and bone disruption (11).

The mechanisms described above are specific for S. aureus and not shared by other opportunistic bacteria involved in orthopedic infections, which points to the evolution of bacterial species-specific ways of interaction with eukaryotic cells that need to be further elucidated.

The secretory function is highly relevant in the framework of osteoblast activities. It comprises the release of the diverse components of the extracellular matrix, essentially calcium-deficient apatite and trace elements, for the inorganic part; type I collagen and other types of collagens, and non-collagenous proteins (γ-carboxyglutamate-containing proteins, proteoglycans, glycoproteins, and small integrin-binding ligands N-linked glycoproteins), for the organic part (49). Osteoblasts also release a variety of cytokines, chemokines, and growth factors (i.e., the osteoblast secretome) that regulate osteoclast formation and resorptive activity (e.g., M-CSF, RANKL, OPG, WNT5A, WNT16, IL-6, GM-CSF) (14), enhance the osteoblast anabolic function (e.g., IGFs and IGFBPs) (50), support hematopoiesis (e.g., G-CSF, osteopontin, thrombopoietin, angiopoietin 1, CXCL12, SCF and IL-7) (51) and act as immune modulators (e.g., IL-6, GM-CSF, CCL5) (52). In this respect, an intimate crosstalk takes place in the bone between skeletal cells and the immune system, which also involves the release of cytokines and chemokines in diverse pathophysiological contexts (53). Systemic and local factors such as inflammation and infection elicit marked changes in bone cell functions, as further described in the following sections.

At sites of infection, soluble factors deriving from activated immune and skeletal cells make up an inflammatory milieu that mediates reciprocal regulation of the osteal and immune components and provides host defense against pathogens. In this regard, osteoblasts play an active role, and cannot be regarded as inert niches for bacterial colonization. In fact, not only do they strive to kill intracellular bacteria by increasing the production of reactive oxygen species and oxidative stress, but they also take part in the immune response orchestrated by professional innate immune cells through production of antimicrobial peptides (beta-defensins), as shown in in vitro and in vivo models and ex vivo in human specimens (54, 55). In addition, the osteoblast secretome further boosts the inflammatory response and affect the behavior of skeletal cells (see Figure 2 ).

Cytokine synthesis and release involve several intracellular signaling pathways including the NF-kB pathway, which regulates the secretion of IL-6 and MCP1 (CCL2; see below) (56), and the JNK pathway, which is activated downstream of S. aureus SpA binding to TNFR1 on the osteoblast plasma membrane and leads to increased TLR2 and RANKL protein levels and TNF-α and IL-6 secretion (56).

The transcriptomic profile of infected osteoblasts has been recently comprehensively investigated by RNAseq analysis of FACS-sorted S. aureus-bearing MG-63 cells, and compared to that of non-infected cells and of a mixed cell population comprising both infected and not-infected MG-63 cells (57). Specifically, this work indicated enhanced immune and inflammatory responses in a model of long-term infection, taking advantage of engineered bacterial strains expressing a fluorescent reporter gene that allowed isolation of infected cells 6 days after infection. Top up-regulated genes included several cytokines such as IL-33, IL-32, IL-6, IL-1β, IL-1α, IL-24, G-CSF, TRAIL and TNFSF14, with higher protein levels in culture supernatants. Accordingly, in the Gene Set Enrichment Analysis, several pathways enriched in FACS-sorted infected MG-63 were related to the immune response, including functional categories such as antigen processing and presentation (of note, CD44 and HLA-DR expression has been reported in cultured human osteoblasts) (58) complement and coagulation cascade, platelet activation, Th17 cell differentiation, IL-17 pathway, NOD-like and TLR signal cascades (known to be involved in bacteria recognition), and cytokine signaling (57).

At the protein level, IL-6 has been found to be upregulated in infected murine osteoblasts (in various experimental settings) and human bone tissues (59, 60). Increased IL-6 induces COX-2 and thereby PGE2 and RANKL, which modulate osteoclast recruitment and differentiation, contributing to progressive inflammatory damage and bone loss (61). Enhanced RANKL production by osteoblasts, without concomitant significant change in OPG expression, occurs also downstream of TLR2 recognition of S. aureus; accordingly, RANKL increase is abrogated in osteoblasts from Tlr2 knockout mice. This mechanism would support pronounced bone resorption and periosteal osteoclast formation in S. aureus-infected bones (62).

IL-12 also is secreted by osteoblasts in S. aureus-induced osteomyelitis (59). It has been proposed to strengthen Th1 immune responses and favor elimination of intracellular bacteria (63), a mechanism that could be potentially used to develop novel strategies for infection prevention (64, 65). Since IL-12 in the bone microenvironment promotes myeloid-derived suppressor cell recruitment, it is plausible that osteoblasts contribute to this mechanism (66).

Expression of the highly conserved anti-inflammatory cytokine TGFβ1 has been reported to change in MG-63 cells infected with four different clinically isolated S. aureus strains (43), with 2 of them causing downregulation in the short and one upregulation in the long timeframe (3 and 24 hours post infection, respectively). Among other functions, TGFβ1 stimulates type I collagen production (67), therefore an increase of this factor might explain the abnormal matrix deposition that occurs in the periosteum during an infection.

IL-1β and TNF-α increased in the bone of animals with S. aureus-induced OM (43, 68, 69), and both cytokines stimulated osteoclast maturation and function. However, some findings regarding to IL-1β are contradictory: in vitro infection of murine primary osteoblasts with S. aureus resulted in increased transcription, but not in increased protein synthesis or secretion. Of note, the same was observed for IL-18 (70), a potent inflammatory molecule, structurally and functionally closely related to IL-1β: in fact, it favors osteoclast differentiation by expanding the inflammatory response and inhibits the osteogenic function (71). Owing to the fine line between intense or exaggerated immune response, it is reasonable that IL-1β and IL-18 production is strictly controlled (70), however their regulation deserves further investigation. In parallel, the specific IL-18 inhibitor IL-18BP was upregulated in vitro in human primary osteoblasts 2 hours after S. aureus infection (https://www.ebi.ac.uk/arrayexpress/E-MTAB-6700), while no data are available regarding IL18-BP protein production by osteoblasts in this specific context, to the best of our knowledge. Therefore, at present, activated immune cells, not osteoblasts, are the most likely source of these ILs in the bone microenvironment during bone infections.

Also, while low levels of TNF-α were detected in in vitro differentiated human osteoblasts and in the MG-63 cell line in basal conditions (72), a marked increase was observed in MG-63 upon infection (43, 73). A recent study described a novel signaling cascade comprising TNF-α/miR-129-5p/endothelial nitric oxide synthase (eNOS) in the pathogenesis of osteomyelitis (74). Briefly, TNF-α and miR-129-5p were upregulated while eNOS was downregulated in S. aureus-infected MC3T3-E1 cells and in osteomyelitis patients’ blood. Accordingly, a TNF-α blocker inhibited miR-129-5p and elevated eNOS expression, likely contributing to rescue the mineralization defect caused by S. aureus infection in MC3T3-E1 cells (74).

IL-27 expression has been recently demonstrated to be induced early (on day 1) in the infected bone in a transtibial model of S. aureus-induced osteomyelitis and upon in vitro infection of MC3T3-E1 cells and primary osteoblasts (75). This cytokine likely contributes to host innate immune response in the early phases of the infection by stimulating local neutrophil recruitment and activation.

Finally, there is preliminary evidence of IFN-β secretion by mature murine osteoblasts in response to S. aureus infection (76). This type I interferon has been reported to be a negative regulator of RANKL-mediated osteoclastogenesis by inhibiting the translation of the critical signaling component c-Fos (77). Based on these data, S. aureus would stimulate osteoblasts to release factors with opposing effects on osteoclastogenesis (78); it can be speculated that the production of IFN-β represents a compensatory response to restore bone homeostasis.

Cytokine production from infected osteoblasts has been shown also in the framework of three-dimensional (3D) models of S. aureus-induced osteomyelitis, which more closely reproduce composition and structure of the natural bone compared to conventional 2D culture systems. For example, in a 3D model of osteomyelitis based on the coculture of MC3T3-E1 cells and S. aureus on magnesium-doped hydroxyapatite/collagen I scaffolds, Tnf-α expression increased over time during infection. The same was observed for the long pentraxin Ptx3, a key pattern recognition molecule with emerging roles in bone pathophysiology, known to be induced by TNF-α. On the contrary, Tgf-β(a reported repressor of Ptx3 transcription) decreased over time. In the conditioned medium of the 3D cocultures, TNF-α was not detected, while PTX3 and OPG levels were stable over time (79). Importantly, the expression of selected osteogenic (Bmp2, Alp, Spp1), and antioxidant (Nrf-2, Ho-1) genes were substantially affected by the applied 3D setting, indicating matrix-dependent effects on osteoblasts during an S. aureus infection (79).

In S. aureus osteomyelitis, mRNA and serum protein levels of G-CSF significantly increase in infected patients, contributing to bone loss through suppression of osteoblast function in favor of osteoclast formation (80), and to enhanced phagocytic activity of immune cells. G-CSF is also released directly by osteoblasts upon S. aureus infection, as well as GM-CSF and M-CSF. In fact, GM-CSF and G-CSF mRNA expression and protein secretion were substantially upregulated in cultured mouse and human osteoblasts following interaction with S. aureus (81). Furthermore, these cytokines were induced in unexposed osteoblasts separated from infected osteoblasts by means of a transwell system, thus pointing to a paracrine-autocrine regulatory mechanism. On the contrary, M-CSF secretion increased only in cultures of infected human osteoblasts (81). These CSFs allow osteoblasts to modulate the cellular composition of the bone microenvironment, favoring differentiation of myeloid progenitors towards osteoclasts and various innate immune cell fates (66).

In response to S. aureus infection, osteoblasts also produce chemokines, members of the C-X-C-motif chemokine ligand (CXCL) and CC-motif (CCL) families, such as CCL2, CCL5, CCL7, CCL8, CCL10, CCL11, CCL13, CCL20, CCL26, and CXCL1, CXCL2, CXCL3, CXCL5, CXCL6 and CX3CL1 (57). Chemokines are traditionally known to act as immune cell chemoattractants, recruiting and activating components of the innate and adaptive immunity, however skeletal cells also are endowed with autocrine and paracrine chemokine signaling, which modulates bone turnover (82).

CCL3 and CXCL2 are known inflammatory mediators enhancing osteoclast formation and osteolysis (83). Their expression was documented in samples from osteolytic sites of patients with implant-associated infection (83). In vitro experiments showed that human primary osteoblasts released CCL3 and CXCL2 upon S. aureus infection (83). This indicates that, besides monocytes, also osteoblasts contribute to the sustained local production of these factors and the related enhanced RANKL-dependent bone resorption (82), in line with previous evidence in a mouse model (84) of S. aureus-induced osteomyelitis. Overexpression of CXCL2 would also result into inhibition of osteoblast formation through downmodulation of the ERK1/2 signaling upstream of RUNX2 (85). In accordance with a role of the CCL2-CCR2 axis in S. aureus-induced osteomyelitis, Ccr2-deficient mice had a higher bacterial load than wild type mice in a model of implant-associated S. aureus infection in which a bioluminescent bacterial strain was inoculated directly into the knee joint after implantation of an orthopedic-grade titanium pin. The infection was monitored in vivo by means of bioluminescence and ex vivo by colony-forming unit counting in the infected joint tissue (86). While the specific contribution of osteoblast-derived CCR2 could not be established in this model, the higher bacterial burden in Ccr2-deficient mice was ascribed to lower T cell and myeloid cell infiltration and overall reduced host defense against the pathogen in the absence of a functional CCR2/CCL2 axis.

In vitro S. aureus infection of human primary osteoblasts also causes a strong upregulation and release of CCL5 (also known as RANTES) compared to other cell types, such as endothelial and epithelial cells (87). The levels of this chemokine influence osteoclast and osteoblast formation and function, as highlighted in the CCL5-deficient mouse (88), besides acting as chemoattractant for monocytes/macrophages and T lymphocytes (89).

Additionally, in a mouse model of implant-associated osteomyelitis, CXCL10 and CXCL9 were up-regulated in the infected femurs versus controls at 3- and 14-days post-infection (90), suggesting a possible role of these chemokines (produced also by osteoblasts through TLR4 activation (91, 92)) in the pathological bone turnover occurring during Osteomyelitis. Of note, in clinical samples from S. aureus-infected patients and from a mouse model of MRSA skin infection, CXCL10 and CXCL9 have also been found to enhance the spontaneous release of the virulence factor SpA, however the underlying mechanism and the biological significance of this process are yet to be defined (93). In this regard, there is evidence that extracellular SpA contributes to biofilm formation by S. aureus (39), therefore the ability of certain chemokines to induce its release could paradoxically help the pathogen skip immune recognition via encasing in a protective layer of biofilm (94). A similar process could occur in principle in bone infections, further increasing the complexity of molecular interactions that underpin osteomyelitis.

Moreover, in a mouse model of endodontic infection-induced inflammation that mimics osteomyelitis of the jaw, the chemokines Cxcl5, Cxcl2, and Cxcl13 were among the top upregulated genes in bone lesions (95).

Furthermore, osteoblasts express both CXCL12 (also known as SDF-1) and its receptor CXCR4. This signaling axis has been extensively studied in relation to the bone marrow niche (96) and implicated in skeletal homeostasis (97). Its involvement in bone remodeling during Osteomyelitis is quite likely, though not specifically investigated thus far, to the best of our knowledge.

Finally, the neutrophil chemoattractants CCL5, CXCL1, and CXCL8 and the chemokines related to T cell activation CXCL9, CXCL10, and CXCL11, were strongly upregulated in human-osteocyte-like cells in response to S. aureus invasion, in ex vivo infected human bone and in bone specimens from the infected acetabulum site of patients suffering from periprosthetic joint infection (48). CCL5 and CXCL10 proteins, but not CXCL8 were also confirmed to be secreted by infected osteocyte-like cells.