Current Concepts of Osteomyelitis

Staphylococcus aureus arriving at the bone surface via iatrogenic or hematogenous routes can readily adhere to soft tissue, bone, or metal implants. The bacterium may achieve this through binding to extracellular matrix (ECM) proteins via microbial surface components recognizing adhesive matrix molecules, such as collagen-binding protein and bone sialoprotein binding protein. Herein, S. aureus employs multiple survival strategies to not be affected by immune cells and therapies (Figure 1A). Adherent S. aureus may multiply, aggregate, and form microcolonies,^, which are also known as staphylococcal abscess communities (SACs) (Figure 1D).^,

SACs are not exclusively found in bone tissue; they have been observed in skin, kidney, renal, and brain tissues.^{,  ,}  In most cases, the SACs form the center of an abscess structure with surrounding fibrin deposits.^,, Specifically, S. aureus can form fibrin by promoting polymerization of fibrinogen and secreting enzymes, such as coagulase and von Willebrand factor–binding protein, which activate endogenous prothrombin and contribute to fibrin formation. This fibrin network surrounding the bacterial SACs protects the bacteria from invasion and clearance by immune cells, such as PMNs, causing the immune cells to gather around the bacterial nidus (Figure 1D).^{,,  ,} 

Staphylococcus aureus adhering to implanted devices or sequestra may form an even more complex structure known as a biofilm (Figure 1D).^, Bacteria in biofilms are less susceptible to antibiotics, because of several factors, including reduced oxygen levels and metabolism. Furthermore, S. aureus resident in biofilm may secrete the so-called extracellular polymeric substance matrix consisting of self-produced polysaccharides and proteins and possibly extracellular DNA from dead bacterial cells, forming a matrix that functions as a physical barrier to immune cell infiltration.

Besides clustering in SACs or into biofilm, S. aureus can also invade the osteocyte-lacuno canaliculi networks within bone (Figure 1D).^, Recently, invasion of S. aureus in the submicron channels buried deep within the dense mineral matrix of cortical bone has been discovered.^, Invasion of S. aureus within the canalicular network could be a mechanism to promote persistence and chronic infection, with the potential to limit access to immune cells. This novel persistence mechanism was originally identified in a mouse model of implant-associated osteomyelitis, and was subsequently confirmed in a human S. aureus diabetic foot infection.^, This discovery is particularly concerning in the context of S. aureus osteomyelitis, as these canalicular networks may be impenetrable by immune cells, and bacteria can possibly survive in this space for a long period of time, using bone matrix as a nutrient source. However, this has not been proved yet.

As another defense mechanism, S. aureus has a wide range of toxins that target host cells. These toxins include exfoliative toxins, pore-forming toxins, and superantigens. The superantigen toxic shock syndrome toxin 1 has been associated with bone infection. It has been shown that toxic shock syndrome toxin 1 can promote osteoclastogenesis and bone resorption activity of osteoclasts in vitro. Also, pore-forming toxins have been linked to bone infection. Staphylococcus aureus pore-forming toxins can be subdivided into leukotoxins, hemolysin-α, hemolysin-β, and phenol-soluble modulins; and these proteins affect host cell membrane integrity. For phenol-soluble modulins, it has been demonstrated in vitro to have a cytotoxic effect on osteoblasts, whereas the hemolysin-α caused both osteoblast and osteoclast cell death in vitro. Hemolysin-β may be involved in phagosomal escape, as shown in vitro in PMNs, and in vivo it stimulated biofilm formation. Furthermore, human osteomyelitis patients infected with an S. aureus strain that can secrete the leukotoxin Panton-Valentine leucocidin had a more aggressive and more difficult to treat infection.^,

In addition, S. aureus can invade and survive intracellularly in professional phagocytes,^, as well as nonprofessional phagocytes (Figure 1D).^, Staphylococcus aureus triggers phagocytic internalization by expressing fibronectin-binding proteins (A and B), adhering to fibronectin, and connecting to α₅β₁ integrins on macrophages or neutrophils. After internalization, S. aureus evades cell death in these cells by persisting within vacuoles or by inhibiting phagolysosomal fusion. It can also infect and survive within non-professional phagocytes, such as primary human osteoblasts in vitro, mouse osteoclasts in vitro and in vivo, and human osteocytes in vitro and ex vivo. This intracellular persistence provides the pathogen with crucial protection needed against the onslaught of the immune system and antibiotic treatments. Staphylococcus aureus–infected human osteoblasts may also mature into osteocytes and remain infected. To survive intracellularly, S. aureus frequently adopts a dormant small colony variant phenotype, characterized by slow growth and reduced metabolic activity. The numerous mechanisms by which S. aureus is able to survive intracellularly within the bone niche for long periods is a primary cause of chronic and recurrent osteomyelitis.

The presence of bacteria within the bone tissue triggers a host response, which encompasses an innate immune response primarily driven by PMNs, macrophages, and adaptive responses mediated by T cells, B cells, and pathogen-specific antibodies (Figure 1B).

First, on bacterial recognition, resident macrophages in bone,^, osteocytes, and osteoblasts^, all appear able to secrete chemoattractants to initiate an influx of immune cells to the site of infection. An influx of PMNs during acute osteomyelitis occurs in both humans^, and in rodent osteomyelitis models.^, Inflammatory macrophages [myeloid related protein 8 (MRP8)/MRP14 positive] and CD4 T cells, potentially activating PMNs, have also been observed in humans; and many necrotic immune cells are present in rodent osteomyelitis models.^,, PMNs can efficiently kill planktonic S. aureus via phagocytosis, oxidative bursts, and production of antimicrobial peptides, whereas secretion of proinflammatory cytokines and chemokines, such as tumor necrosis factor (TNF)-α, IL-1β, CXCL2, CXCL3, and others, activates and recruits PMNs, which ultimately leads to pathogen clearance. PMNs and macrophages also elicit direct host defense responses by forming neutrophil extracellular traps, to trap bacteria, which are eventually cleared by immune cells.^,

As the infection persists and becomes chronic, bacteria tend to form a biofilm phenotype, and the influx of viable PMNs decreases drastically, as demonstrated in mice and observed in humans with chronic osteomyelitis.^,, In humans, most cells present at the site of chronic infection are wound-healing M2 macrophages (CD163 positive), accompanied by a small number of CD8 T cells and plasma cells.^,, The predominant M2 macrophages are inefficient in phagocytosing bacteria within the biofilm, and they tend to promote profibrotic environments and wound healing responses, generating abscesses during chronic osteomyelitis infections.^,

Adaptive immune responses against bone infections include both T- and B-cell responses. Unfortunately, pathogens, such as S. aureus, have evolved numerous evasion mechanisms toward these responses, resulting in chronic osteomyelitis. For instance, in a porcine osteomyelitis infection model, it was observed that the antibody responses against intracellular S. aureus in biofilms are skewed to a predominantly type 1 and type 17 helper T cell biased immune response, which cannot effectively clear intracellular pathogens. Staphylococcus aureus can also efficiently manipulate B cells, affecting their survival and function via the secretion of staphylococcal protein A (SpA), which associates with the Fcγ and Fab domains of certain antibodies,^{,  ,}  blocking antibody-mediated phagocytosis and simultaneously causing proliferative B-cell apoptosis.^,

In addition, pathogen-specific antibodies produced by circulating plasmablasts and plasma cells are often not protective against chronic bone infections.^, Although further studies are needed to fully understand this, it may be that due to SpA, interference antibodies secreted against S. aureus do not confer protection against reinfection or chronic musculoskeletal infections or that these antibodies are nonneutralizing antibodies. In fact, anti–S. aureus IgG responses against certain antigens can lead to mortal outcomes. Nonetheless, these antibody responses can be useful diagnostic and prognostic biomarkers for identifying orthopedic infections.

Bone is a mineralized organic matrix containing osteocytes, bone-forming osteoblasts, and bone-resorbing osteoclasts. All three bone cells are impacted directly and indirectly by S. aureus (Figure 1C).

Directly, SpA binding to TNF receptor-1 on osteoblasts results in an increase in apoptosis and a decrease in differentiation and calcium deposition of the osteoblasts.^, Moreover, internalization of S. aureus through fibronectin-binding protein A/B–α5β1 integrin bridging affects osteoblast viability and functioning.^, Both S. aureus internalization and SpA binding cause decreased bone formation and inhibition of matrix mineralization.^, Conversely, S. aureus infection increases periosteal bone formation by osteoblasts (as shown in rabbits) (Figure 2A) compared with noninfected controls (Figure 2B).

Osteoclasts up-regulate their bone resorption capacity because of TNF and epidermal growth factor receptor activation through SpA secreted by S. aureus. This leads to resorption lacunae formation and necrotic bone pieces, as observed in biopsies of human osteomyelitis patients (Figure 2C)^, and in in vivo osteomyelitis models (Figure 2D). Indirectly, osteoclasts are activated and increase osteolysis activity by osteoblasts, osteocytes, and PMNs. These cell types secrete receptor activator of NF-κB ligand (RANK-L), which drives osteoclastogenesis and activates osteoclasts to resorb bone. Osteocytes do so in response to a neighboring osteocyte that underwent apoptosis (eg, due to invasion of its lacunae by S. aureus). Moreover, osteoblasts up-regulate RANK-L expression when SpA is bound to TNF receptor-1 and by bacterial internalization,^, whereas PMNs up-regulate RANK-L secretion by toll-like receptor 4 activation. PMNs also drive osteoclastogenesis and bone resorption via osteoclast-mediated secretion of IL-8. Another contributor to osteoclastogenesis and osteoclast activity is the persistent inflammatory environment itself. This occurs initially because of the secretion of the proresorptive cytokines IL-6, TNF-α, and IL-1β by immune cells and osteoblasts,^, and subsequently because of hypoxia resulting from the persistent inflammation.

A crucial role of osteocytes is to mature and maintain the mineralized matrix, which is accomplished by their expression of enzymes capable of reversibly removing mineral and remodeling the organic phase of bone matrix, a process described as osteocytic osteolysis or perilacunar remodeling.^, The involvement of this process during osteomyelitis is currently poorly described, although matrix metalloproteinase expression was observed to be induced in S. aureus infected human osteocytes, suggesting that osteocytic osteolysis is affected by S. aureus. Another interesting function of osteocytes is their potential role in the recruitment of immune cells. A recent study demonstrated that human osteocyte-like cultures exposed to S. aureus resulted in the differential expression of >1500 genes, including the robust induction of a large number of chemokines and cytokines. Although classic PMN chemoattractants, such as CXCL1 and chemokine (C-C motif) ligand 5, were detected, CXC chemokine receptor 3 (CXCR3)-binding chemokines CXCL9, CXCL10, and CXCL11 were also expressed in abundance, suggesting the potential participation of osteocytes in the adaptive immune response to bacterial infection by recruiting cytotoxic and/or suppressive T-lymphocyte subsets to the infected sites. Further studies of the influence of the osteocyte in this regard will be of interest.

Taken together, S. aureus infection enhances osteoclastic bone resorption,^,, possibly osteocytic osteolysis of bone, and inhibits bone formation,^, leading to an overall loss in bone tissue.^,

Biofilm growth in vivo can be mimicked with conventional models, such as microtiter plate-based models or flow displacement biofilm models, as reviewed recently. These models can be used for antimicrobial compound testing and measuring bacterial colonization/biofilm formation on various substrates. One of the most common methods currently used to assess anti-biofilm efficacy is the minimum biofilm eradication concentration assay, which is a 96-well biofilm system using polystyrene pegs. Bacterial biofilms grown on the pegs can be simultaneously challenged with multiple antibiotic combinations at different concentrations for assessing the bactericidal and/or bacteriostatic efficacies of these anitmicrobials.

Conventional bacterial models can also be used to examine bacterial colonization on materials such as polymethyl methacrylate and the efficacy of antibacterial coatings. Examples of orthopedic implant-related materials and coatings that have been tested for bacterial colonization have recently been reviewed. An interesting antibacterial coating that has been tested is a tissue plasminogen activator-containing coating to activate plasminogen and increase fibrin degradation. Fibrin can form a layer on biomaterials and promote adherence of pathogens to the biomaterial. Fibrin is also a component of the S. aureus biofilm matrix that facilitates antibiotic resistance due to poor penetration of the antibiotic into the biofilm. It was shown that the tissue plasminogen activator-containing coating reduced bacterial adherence to the biomaterial. In addition, adherent bacteria were more susceptible to antibiotics because the bacteria were not protected by a fibrin matrix. In a mouse model where S. aureus–infected implants were placed subcutaneously, the coating prevented biofilm-related infection.

In an effort to increase the complexity and relevance of in vitro studies, bacterial cocultures with host immune or bone cells identified as key players in osteomyelitis have also been performed. For this review, a coculture is defined as a culture that combines bacteria with at least one host cell type.

Multiple groups have examined the effects of bacteria, usually S. aureus, on osteoblasts using two-dimensional cell culture models. To prevent bacterial overgrowth in static cultures, several techniques are routinely employed to remove extracellular bacteria. These include the use of antibiotics or the S. aureus–lysing enzyme lysostaphin, as well as rinsing to remove unbound bacteria. A variety of osteoblast coculture models have been developed using rodent and human cell lines, as well as human primary osteoblastic cells. Studies using S. aureus–infected human osteoblast cultures reported that the host cells underwent rapid and dramatic cell death after infection. One study examined the host cell response to several S. aureus strains at a fixed multiplicity of infection and showed that human primary osteoblasts exposed acutely or for short periods of time did not undergo cell death. Although relatively few intracellular bacteria were recovered, the primary cells secreted detectable levels of innate immune cell–relevant chemokines and cytokines, indicating the potential of osteoblasts to participate in innate immune responses and the utility of this model for studying this phenomenon. A study using bone explant-derived cells from the femoral heads of patients undergoing hip replacement surgery found that infections for up to 48 hours generated only low-level chemokine and cytokine responses, which the authors interpreted as indicating that osteoblasts may serve to internalize bacteria but not contribute significantly to the innate immune response. It is possible that matching the source of human primary cells and the pathology under investigation (osteomyelitis) may influence experimental outcomes. More specifically, periprosthetic joint infection most often occurs in patients treated for primary osteoarthritis, and osteoblastic cells derived from these donors display different phenotypic and behavioral qualities, such as aberrant in vitro mineralization, to those derived from nonosteoarthritis patients treated for fragility fractures of the hip. Furthermore, in hip osteoarthritis patients, the femoral head is usually diseased and the cells derived from this site may, therefore, be aberrant in their responses ex vivo. Thus, a site more distal from the joint (eg, the intertrochanteric region of the proximal femur) may be a more suitable source of disease-naïve cells. In a study using human osteoarthritis proximal femur-derived osteoblasts differentiated to an osteocyte-like stage, no cell death effect was observed in response to S. aureus infection for up to 30 days. Staphylococcus aureus formed small colony variant associated with the up-regulation of sigma B activity, consistent with establishment of a persistent intracellular infection. This correlated with observations of S. aureus bacteria inside viable osteocytes in clinical periprosthetic joint infection bone specimens. This model allows the study of both the host response and adaptation of the bacteria to intracellular infection.

Other studies have incorporated foreign biomaterials into the infection model. A typical application is the coculture of S. aureus and osteoblasts on a biomaterial surface to model the so-called race for the surface. Herein, the idea is that if host cells colonize the biomaterial first, bacterial adhesion is prevented. One way to study the race for the surface is by seeding a flow chamber with both staphylococci and osteoblasts. By using this method, different coatings that prevent bacterial adhesion and subsequently promote more host cell attachment can be studied, such as a coating containing the antibiotic levofloxacin. Biomaterials coated with levofloxacin had fewer adherent S. aureus compared with a non–levofloxacin-coated equivalent, and this enabled colonization by preosteoblasts.

The effect of S. aureus on osteoblast-induced osteoclastogenesis has also been studied in cocultures. These studies revealed that bacterial surface proteins could drive osteoclast formation because formaldehyde-fixed S. aureus induced RANK-L expression and IL-6 secretion by osteoblasts. Subsequently, this imbalances bone remodeling in favor of bone resorption. Furthermore, cocultures of S. aureus with osteoclasts have been performed; osteoclasts were seeded onto an inorganic crystalline calcium phosphate matrix mimicking bone, in presence of S. aureus. The infection promoted multinuclear osteoclast formation, which had a cellular area fourfold higher than noninfected osteoclasts and an increased bone resorption capacity, resulting from activation of the NF-κB pathway by S. aureus. Therefore, targeting osteoclast activity using antiresorptive drugs, such as bisphosphonates or denosumab (a monoclonal antibody targeting RANK-L), may be a means to prevent infection-induced osteolysis. Bisphosphonates or denosumab is effective in patients with mandibular osteomyelitis,^, but systemic administration of these antiresorptive drugs can also cause osteonecrosis of the jaw,^, indicating that further studies into antiresorptive drugs as a treatment option for osteomyelitis are required.

Bacterial biofilms can also be cocultured with immune cells. Immature and mature biofilms have been cocultured with PMNs to assess phagocytosis of biofilm-resident bacteria and the migration of PMNs to the biofilm. PMNs migrated toward the biofilm and engaged in phagocytosis of the biofilm, especially when the biofilm was in an immature state (<6 days old). Mature biofilm was less sensitive to PMN attack than immature biofilm because 15-day–old biofilm was subjected to significantly less phagocytosis by PMNs than 2- and 6-day–old biofilms. A possible reason for this may be that the ECM covering biofilm matures over time, thus preventing PMNs from reaching the bacteria in resident mature biofilm.

To our knowledge, only one multicellular model involving bacteria cocultured with both bone and host immune cells has been reported. To investigate competition for the surface of a polymethyl methacrylate plate, bacteria (S. aureus, Staphylococcus epidermidis, or Pseudomonas aeruginosa) were cultured in a flow chamber with an osteoblast cell line in the presence or absence of macrophages. It was shown that colonization of the polymethyl methacrylate plate by osteoblasts did not increase in the presence of macrophages, and it was primarily colonized by bacteria. This is in line with clinical observations where, despite host cell presence, bacteria win the race for the surface. The presence of macrophages did prolong the survival of osteoblasts in the multicellular cultures with either S. aureus or P. aeruginosa, and osteoblasts were able to grow and spread in the presence of low-virulence S. epidermidis. Future studies using such a multicellular model and testing different biomaterials would be of interest.

Although conventional models are of great use, they can only model the following aspects of osteomyelitis: biofilm formation and interactions of bacteria with one or multiple host cell types (Table 1). To resemble bone tissue, fibrous encapsulation, complex interactions between bacteria and multiple host cells, and osteomyelitis-induced bone abscesses with a necrotic core, more sophisticated systems, such as 3D in vitro systems, will be of immense value.