|Home | About | Journals | Submit | Contact Us | Français|
Periodontal disease is characterized by both inflammation and bone loss. Advances in research in both these areas have led to a new appreciation of not only each field but also the intimate relationship between inflammation and bone loss. This relationship has resulted in a new field of science called osteoimmunology and provides a context for better understanding the pathogenesis of periodontal disease. In this review, we discuss several aspects of the immuno-inflammatory host response that ultimately results in loss of alveolar bone. A proposal is made that periodontal inflammation not only stimulates osteoclastogenesis but also interferes with the uncoupling of bone formation and bone resorption, consistent with a pathologic process. Furthermore, arguments based on experimental animal models suggest a critical role of the spatial and temporal aspects of inflammation in the periodontium. A review of these findings leads to a new paradigm to help explain more fully the impact of inflammation on alveolar bone in periodontal disease so that it includes the effects of inflammation on uncoupling of bone formation from resorption.
It is widely recognized that bacteria initiate periodontal disease. Early studies by Löe and others (Löe et al., 1965; Theilade et al., 1966) demonstrated a direct relationship between bacterial plaque accumulation and gingival disease. In these studies, tissues without bacterial plaque had little inflammation, while those with plaque exhibited a strong inflammatory reaction. This process was reversible, indicating a direct correlation with bacterial plaque. Other studies in animals demonstrated a relationship among microbial plaque, inflammation, and periodontal bone loss (Keyes and Jordan, 1964; Saxe et al., 1967; Lindhe et al., 1975). These early studies established the bacterial etiology of gingivitis and periodontal diseases involving bone loss.
More recent studies have not only reinforced the bacterial etiology of periodontal disease but have also emphasized the role of inflammation in the pathologic process. These studies were performed in animal models to establish a cause-and-effect relationship between bacteria and the initiation of periodontal disease. In one animal model, a ligature is tied around the teeth, causing plaque accumulation and facilitating bacterial penetration into the gingival, which leads to inflammation and alveolar bone resorption (Graves et al., 2008). In contrast, the placement of ligatures in gnotobiotic rats does not cause significant increases in gingival inflammation or periodontal bone loss (Rovin et al., 1966), again demonstrating the essential role of bacteria as an initiating trigger. Further studies with other animal models also reinforce the relationship between bacteria and periodontal disease. For example, treatment of animals with antibiotics or topical application of chlorhexidine reduces the bacterial load and significantly reduces bone resorption (Weiner et al., 1979; Kenworthy and Baverel, 1981). In contrast to reducing bacteria, an increase in bacterial load enhances periodontal disease (Nagahata et al., 1982). In other animal models, the inoculation of periodontal pathogens into the oral cavities of rodents induced bone loss. In several studies, the introduction of Porphyromonas gingivalis into the oral cavity by oral gavage induced alveolar bone resorption in the mouse (Baker et al., 1994, 1999, 2000; Lalla et al., 1998). Similarly, oral gavage with Actinobacillus actinomycetemcomitans (Garlet et al., 2006) or Tannerella forsythia (Sharma et al., 2005) has been reported to stimulate periodontal bone loss. Introduction of A. actinomycetemcomitans in rats leads to colonization and the loss of alveolar bone (Schreiner et al., 2003; Fine et al., 2005). Thus, experimental studies in animal models support the human clinical trials implicating bacteria in the initiation of inflammation and periodontal disease (Reddy et al., 2003; Kirkwood et al., 2007).
It has been well established that manipulation of the host response can attenuate periodontal bone loss (Graves, 2008). When the host response is reduced either through the application of inhibitors or through genetic manipulation, the severity of bone loss stimulated by periodontal pathogens is typically reduced, even if antibacterial defenses may be weakened. This suggests that the host response plays an essential role in bacteria-induced periodontal bone resorption. There are several classes of molecules that activate a host response that can stimulate osteoclastogenesis either directly or indirectly, including lipid-based mediators such as prostaglandins or leukotrienes, cytokines, and chemokines (see Fig. 1 and Table 1). One of the first studies to demonstrate a cause-and-effect relationship examined the impact of a non-steroidal anti-inflammatory drug on periodontal bone loss in the beagle dog (Williams et al., 1985). This study demonstrated that flurbiprofen over a 12-month period significantly decreased naturally occurring alveolar bone loss in dogs compared with control animals treated with placebo. Thus, products of the cyclooxygenase pathway of arachidonic acid metabolism are significantly involved in tissue destruction induced by periodontal disease. Prostaglandins are elevated in gingival crevicular fluid from patients with localized aggressive periodontitis, and P. gingivalis stimulates leukocyte infiltration concomitant with elevated PGE-2 levels and increased cyclooxygenase-2 expression in vivo (Pouliot et al., 2000). Human clinical trials suggest that cyclooxygenase inhibitors are useful in inhibiting periodontal disease but have not been widely adopted because of side-effects (Paquette et al., 2000). More recent evidence indicates that omega-3 fatty acid products, lipoxins, and resolvins reduce inflammation and alveolar bone loss in experimental periodontitis (Hasturk et al., 2006).
Cytokines also play a prominent role in bacteria-induced periodontal bone resorption (summarized in Table 1). In a non-human primate model, inhibitors of two prominent pro-inflammatory cytokines, interleukin-1 (IL-1) and tumor necrosis factor (TNF), reduced periodontal bone loss and loss of attachment compared with that in animals treated with vehicle alone (Assuma et al., 1998; Delima et al., 2001, 2002). Experiments were carried out in a Macaca fascicularis primate model in which P. gingivalis- soaked silk ligatures were tied around the posterior teeth. The ligature results in greater plaque accumulation coupled with the addition of P. gingivalis to the oral bacteria to induce periodontal disease. IL-1 and TNF were specifically inhibited by the local injection of soluble receptors to these cytokines into the gingiva. The inhibition of TNF and IL-1 with specific antagonists reduced the inflammatory cell infiltrate that forms close to bone and inhibited bone resorption. Thus, IL-1 and TNF contribute significantly to the pathologic bone loss in periodontal disease. In a follow-up study with the same model, IL-1 and TNF blockers inhibited the loss of connective tissue attachment (Delima et al., 2001). Moreover, inhibition of IL-1 alone had an effect, significantly reducing both the “migration” of an inflammatory infiltrate toward bone and the loss of alveolar bone.
From the above results, the authors suggest that inflammation associated with gingivitis is actively protective, since blocking further up-regulation of the host response with IL-1/TNF inhibitors reduced host destruction. This is in contrast to situations where the absence of TNF-α and IL-1 renders the tissue more susceptible to bacterial infection. For example, in endodontic infections where there is no pre-existing inflammatory response in the dental pulp (as there is in the gingiva), exposure of the naïve dental pulp to oral pathogens results in significantly greater destruction when TNF receptor signaling is absent (Chen et al., 1999). In contrast, periodontal disease is reduced in similar TNF-receptor-deficient mice (Garlet et al., 2007). Thus, in experimental periodontitis, inhibition or knockout of inflammatory mediators typically reduces the severity of the disease process, whereas blockage or deletion of the same genes in animals with endodontic lesions typically enhances the pathologic process and disease severity (Assuma et al., 1998; Baker et al., 1999; Chen et al., 1999; Huang et al., 2001; Garlet et al., 2007; De Rossi et al., 2008). This difference may be due to the protective presence of a low-level inflammatory response that is always present in the superficial gingiva close to teeth, whereas this pre-existing leukocyte response is not present in naïve dental pulp.
IL-1 and TNF stimulate the formation of osteoclast-like cells in vitro (Devlin et al., 1998). IL-1 stimulates osteoclastogenesis and bone resorption, largely through up-regulation of receptor activator for nuclear factor-κB (RANK) ligand, while TNF can stimulate osteoclastogenesis directly or indirectly through RANK ligand (Wei et al., 2005). RANK ligand inhibitor, osteoprotegerin (OPG), may also be produced, so the ratio of RANK ligand to OPG is an important consideration. Lymphocytes are thought to be a particularly important source of RANK ligand in periodontal disease, while in physiologic bone remodeling, RANK ligand appears to be principally from bone-lining cells (Teng et al., 2000; Kawai et al., 2006; Rauner et al., 2007). Inhibition of RANK ligand caused a decrease in alveolar bone loss in several models of periodontal disease (Teng et al., 2000; Han et al., 2006; Jin et al., 2007). In one study, the participation of CD4+ T-cells in mediating periodontal bone loss was demonstrated (Teng et al., 2000). Furthermore, the administration of the RANK ligand inhibitor, osteoprotegerin, in these mice reduced osteoclastogenesis and inhibited alveolar bone destruction. Another group has determined that B-cells contribute to periodontal bone loss in the absence of T-cells through the production of RANK ligand (Han et al., 2006). Transfer of B-cells from rats immunized against A. actinomycetemcomitans, followed by injection of the periodontal pathogen, significantly increased periodontal bone loss compared with transfer of B-cells from non-immunized animals. Increased bone loss was inhibited by the injection of osteoprotegerin. The role of RANK ligand in a ligature-induced model of periodontitis has been examined (Jin et al., 2007). In this study, rats were administrated subcutaneously an OPG-Fc fusion protein to block RANK ligand. By µCT and histomorphometric analysis, reduced osteoclastogenesis and significant preservation of alveolar bone volume were observed among OPG-Fc–treated rats compared with the controls. In humans, there have been several clinical trials testing the RANKL-inhibitor Denosumab, a human monoclonal antibody specific for RANKL (Lewiecki, 2009). Denosumab increases bone density and reduces fracture risk in post-menopausal women and inhibits osteoclast-mediated damage caused by rheumatoid arthritis (Lewiecki, 2009).
Mitogen-activating protein (MAP) kinase inhibitors also reduce lipopolysaccharide-induced alveolar bone loss, consistent with the role of MAP kinase in mediating the effects of pro-inflammatory signals (Rogers et al., 2007). Application of a p38 MAPK inhibitor reduced LPS-induced osteoclast formation and periodontal bone loss, which is consistent with in vitro results demonstrating that upstream regulators of p38 MAP kinase, MKK3 and MKK6, are required for IL-1beta and TNF-alpha-induced RANK ligand expression in bone marrow stromal cells (Rossa et al., 2008).
Manipulation of the host response through genetic studies further supports the relationship between the host response and periodontal bone loss. To identify the role of adaptive immune response in periodontal bone loss, a murine model has been used with oral infection by a Gram-negative anaerobic bacterium, P. gingivalis. Severe combined immunodeficient mice, which lack B- and T-lymphocytes, had much less bone compared with normal control mice, suggesting that the adaptive immune response plays a critical role in periodontal bone loss. To further identify the lymphocyte subtype involved, several immune-deficient mice were examined, including MHC II CD4+ T-cell-deficient, MHCI CD8+ T-cell-deficient, and natural killer T-cell-deficient mice. Only MHC class II-responsive CD4+ T-cell-deficient mice had diminished bone loss compared with normal controls (Baker et al., 1999). Thus, CD4+ T-cells but not CD8+ T-cells in this study were critical for bacteria-induced alveolar bone loss in response to oral infection of P. gingivalis.
To investigate the role of pro-inflammatory cytokines, researchers have examined mice with genetic deletion of IL-6 and IFN-γ. Introduction of P. gingivalis to IFN-γ and IL-6 knockout mice stimulated less periodontal bone loss compared with that in normal control mice, indicating that these cytokines contribute to periodontal disease progression. Although IFN-γ directly inhibits osteoclastogenesis, the results suggest that the stimulatory effect of IFN-γ on monocytes and lymphocytes overrides the direct inhibitory effect on osteoclasts and contributes to P. gingivalis-induced bone loss (Yang et al., 2002; Gao et al., 2007). In other studies, it was shown that IFN-γ was necessary in preventing systemic dissemination of a periodontal infection, and that the absence of IFN-γ reduced the levels of inflammatory cytokines and recruitment of leukocytes to the periodontium, supporting the role of IFN-γ as contributing to bacteria-induced periodontal bone loss (Garlet et al., 2008). Likewise, the role of TNF receptor signaling has been investigated by study of the response of mice with genetic deletion of TNF receptors to P. gingivalis backspace in a calvarial model and inoculation of the oral cavity with A. actinomycetemcomitans (Graves et al., 2001; Garlet et al., 2006). In the calvarial model, experimental mice with targeted deletion of TNF receptor-1 and TNF receptor-2 showed significant reduction of PMN infiltration and osteoclast numbers in response to injection of P. gingivalis compared with the wild-type mice (Graves et al., 2001). In addition, this study established that P. gingivalis stimulated fibroblast apoptosis in vivo through a mechanism that involved up-regulation of TNF-α. An alternative approach to investigating the role of TNF in periodontal disease was recently studied with adeno-associated virus vector to deliver a TNF blocker consisting of the TNF receptor 2 and the Fc domain of IgG1 (Cirelli et al., 2009). In this model, P. gingivalis was injected into the gingiva of rats with and without gene therapy treatment. The gene therapy blocked LPS-stimulated alveolar bone loss and osteoclastogenesis and reduced the local levels of several pro-inflammatory cytokines. In a periodontal model, the absence of TNF receptor-1 signaling in mice inoculated with A. actinomycetemcomitans reduced the levels of chemokines and their receptors, and reduced matrix metalloproteinase and RANK ligand mRNA levels. Interestingly, the antibacterial defense was also reduced, which was associated with the detection of higher levels of A. actinomycetemcomitans. Even with greater levels of A. actinomycetemcomitans, there was less bone loss, since the impact on the pro-inflammatory levels and reduced inflammatory infiltrate was relatively more important. Taken together, these studies involving manipulation of the host response through inhibitors or genetics demonstrate that the host inflammatory response to bacteria mediates periodontal bone loss.
The studies discussed above have provided insight into the role of the host response with regard to its being protective vs. destructive (i.e., contributing to tissue loss). For example, the finding that reducing the host response would lead to less bone loss was initially counter-intuitive, since it is widely known that the host response is protective. For example, a compromised host response in leukocyte adhesion deficiency, Chédiak-Higashi syndrome, Papillon-Lefèvre, and acquired immune deficiency syndrome leads to greater susceptibility to periodontal disease (Nualart Grollmus et al., 2007; Yin et al., 2007). Animal studies also support the notion that the host response is protective. For example, adoptive transfer of A. actinomycetemcomitans-specific T-cell clones has been reported to protect against A. actinomycetemcomitans-induced periodontal bone loss (Yamashita et al., 1991), and immunization against P. gingivalis significantly reduces bone loss in a non-human primate model (Persson et al., 1994). In addition, cytokines such as IL-17 play a protective role in limiting bone loss by P. gingivalis (Yu et al., 2007). This study demonstrated that the enhanced bone loss observed in IL-17 receptor-deficient mice is a result of a defective ability to stimulate neutrophil migration. This failure was linked to defective chemokine production in IL-17 receptor-deficient mice and demonstrated that IL-17 plays an important role in the host defense against oral pathogens.
Taken together, the studies above indicate that the host response has both protective and destructive functions in the periodontium. This apparent contradiction has led to the hypothesis that the host response has dual roles and is more complicated than simply being assigned a “protective” or “destructive” role.
From the above results, it is clear that the immune response is critical in protecting the host from oral pathogens, yet at the same time is intimately involved in the destructive process. One way to resolve this apparent discrepancy in opposing roles is to consider a potentially important aspect of the pathological process, the spatial orientation of the host response. A characteristic of human gingiva is the ubiquitous presence of an inflammatory infiltrate in the gingiva adjacent to a tooth surface (Page and Schroeder, 1976; Yu and Graves, 1995). This infiltrate is present even in the absence of clinically obvious inflammation and is associated with gingivitis but not necessarily periodontitis. Bacteria or their products continually interact with gingival epithelium to stimulate a host response (Handfield et al., 2008). In experimental animal models, bone loss is initiated when the inflammatory infiltrate moves closer to the bone surface. For example, in a non-human primate model of experimental periodontitis, the temporal movement of an inflammatory front toward the crest of alveolar bone is associated with increased osteoclast formation and is substantially reduced by specifically blocked-in IL-1 and TNF (Graves et al., 1998). Thus, when inflammation is contained to the sub-epithelial connective tissue, the result is gingivitis, and when the inflammatory infiltrate moves closer to bone, osteoclastogenesis is induced. Thus, the central issue may not be the degree of inflammation or the qualitative nature of the inflammation, but may well be a consequence of where the inflammation is located in relationship to the bone. If the immune response is able to keep bacteria or their products from penetrating deeply into connective tissue, the inflammation and the host inflammatory response will be confined to the sub-epithelial space, and gingivitis is the clinical outcome. However, the progression of the inflammatory infiltrate toward bone so that it is within a critical distance to the bone surface will result in the activation and formation of osteoclasts and, subsequently, bone loss (Figs. 2A, ,2B).2B). One very significant consequence of this concept is that it provides a mechanistic basis for the universal dimensions of biologic width (Gargiulo et al., 1961). Building on the findings of Gargiulo and co-workers (Gargiulo et al., 1961), Page and Schroeder described biologic width as a 2.5-mm zone between the cemento-enamel junction and bone. They concluded that distances > 2.5 mm are caused by bacterial invasion of gingival connective tissue (Page and Schroeder, 1981), and that the closer the inflammatory infiltrate was to the bone, the greater was the amount of bone degraded (Schroeder and Lindhe, 1980; Rowe and Bradley, 1981). Thus, the proximity of the inflammatory infiltrate to bone may be an important consideration. It is likely that the spatial location of the inflammatory infiltrate will be dependent upon the depth to which bacteria or their products have penetrated the connective tissue. Thus, the spatial aspect of the host response to the bacterial challenge may contribute to the clinical manifestation of the disease (gingivitis vs. periodontitis) and the extent of tissue loss.
In a healthy adult with physiologic tissue turnover, an episode of bone resorption is followed by an equivalent amount of new bone formation, a well-accepted process referred to as coupling (Parfitt, 1982). In other words, bone is masterfully programmed to repair itself through the coupling of bone formation to bone resorption. This balanced bone formation and bone resorption require adequate availability of osteoblast precursor cells, osteoblast differentiation, and the formation of bone matrix in response to coupling signals. Under pathological conditions, however, an equivalent amount of bone formation does not occur, so that resorption and formation are uncoupled (Parfitt, 1982). Thus, an episode of alveolar bone resorption in a healthy adult should, under normal circumstances, be followed by an equivalent amount of bone formation. However, in periodontitis there is a failure to form bone, resulting in net bone loss. Therefore, the pathologic process that leads to net periodontal bone loss logically resides in the failure to form bone, either the inability to form compensatory bone in the quantity of bone resorbed (incomplete coupling) (Fig. 1C) or the inability to form bone at all subsequent to the bone resorption episode (total uncoupling). Thus, growth and differentiation factors that are produced during bone resorption (see Table 1) may not be sufficient to stimulate complete coupling. Alternatively, inflammatory mediators may interfere with the bone-forming process, leading to incomplete coupling. A list of molecules that could mediate the effects of inflammation on uncoupling is provided in Table 2. Although this review focuses on bone, it is recognized that, for periodontal regeneration to occur, the formation of new cementum and periodontal ligament must also take place (Foster et al., 2007).
The inflammatory process that leads to osteoclastogenesis and bone resorption may also be responsible for the failure to form an adequate amount of new bone, i.e., inflammation causes uncoupling of bone formation following bone resorption. We propose that if inflammation is prolonged (temporally) so that it is present during the transition from the bone resorption to formation phase, inflammation will interfere with the elaboration of new bone. Again, evidence for this proposal can be found in animal studies. For example, injection of P. gingivalis into the connective tissue of a normal host leads to a brief episode of inflammation. This inflammatory process results in bone resorption followed by bone formation. When the inflammation is prolonged by a condition such as diabetes, so that the inflammatory process is extended into the phase of bone formation, the capacity to form new bone is significantly reduced (Al-Mashat et al., 2006; Liu et al., 2006a). Furthermore, if TNF production is inhibited in these same animals, the level of inflammation during this period is reduced, and, consequently, the ability to form new bone is improved in the diabetic animals (Liu et al., 2006a). This is strong evidence suggesting that the prolonged inflammation in diabetic animals interferes with bone formation in the periodontium following an episode of bone resorption (Liu et al., 2006b). This interpretation is additionally supported by evidence that the application of cytokines in vivo not only stimulates bone resorption but also limits bone formation. For example, injection of either TNF-α or IL-1β causes bone resorption and inhibits the coupling process by reducing new bone formation (Bertolini et al., 1986; Nguyen et al., 1991). Therefore, several lines of animal experimentation support the concept that inflammation uncouples bone formation from bone resorption. Thus, inflammation not only stimulates the formation of osteoclasts and bone resorption, but also affects bone by altering the function of osteoblasts and limiting reparative bone formation. In addition, it is possible that osteocytes are affected by inflammation and could affect the coupling process (Sims and Gooi, 2008). For example, the death of osteocytes is linked to the production of factors that stimulate bone remodeling. In addition, osteocytes produce sclerostin, a potent inhibitor of bone formation (Sims and Gooi, 2008).
Several other disease processes involve both inflammation and bone uncoupling. For example, similar to periodontal disease, osteoporosis is also characterized by uncoupling due to deficient bone formation following bone resorption. It is widely recognized that ovariectomy in the absence of estrogen replacement leads to bone resorption and loss of bone mass through the action of cytokines such as IL-1 or TNF (Weitzmann and Pacifici, 2006). In a recent study, Chang and colleagues investigated the impact of inflammation on the maintenance of bone mass in osteoporosis (Chang et al., 2009). The goal of these studies was to determine what would happen if the capacity of osteoblasts to respond to inflammatory signals was blocked, but other cell types could still respond. This was accomplished by deleting NFκB, a transcription factor that mediates gene expression stimulated by inflammatory molecules in osteoblasts, but not other cell types, through genetic manipulation. When NFκB was specifically blocked in osteoblasts, the capacity to form bone was enhanced without change in inflammation and without affecting osteoclast activities. This study also revealed NFκB affected bone formation through its effect on Fos-related antigen-1 (Fra-1), an essential transcription factor involved in bone matrix formation. These results suggest that inflammatory mediators stimulate NFκB activity, which reduces Fra-1 expression and limits the production of bone matrix by osteoblasts. Thus, inflammation affects osteoblasts by limiting the capacity to form bone and, as a result, plays an important role in the uncoupling process of post-menopausal osteoporosis.
As in osteoporosis and periodontal disease, the loss of bone mass occurs in inflammatory arthritis. In arthritis, the inflammatory aspect of the disease is thought to cause destruction of joints and to induce lytic lesions in the peri-atricular bone which is not adequately repaired by bone coupling (Walsh et al., 2009). This may be partly caused by TNF-α, based on studies in which over-expression of TNF-α in transgenic mice caused the formation of lesions typical of rheumatoid arthritis (Keffer et al., 1991). In vivo administration of monoclonal antibody specific for TNF-α prevented the disease process, suggesting a critical role for TNF in the pathogenesis of arthritis. The failure to form an adequate amount of new bone, i.e., uncoupling, has been linked to reduced osteoblast differentiation caused by inflammatory cytokines (Diarra et al., 2007). Diarra and colleagues reported that TNF-α stimulates production of Dickkopf-1 (DKK-1), which suppresses bone formation by inhibiting the WNT pathway. Their findings showed that DKK-1, a negative regulator of WNT pathway, is up-regulated by TNF stimulation through TNF receptor-1 and P38 MAPK signaling. This up-regulated DKK-1 not only promotes bone resorption but also effectively blocks bone formation and repair in the diseased joint. This provides a mechanistic basis for understanding how TNF-α can limit the ability to form bone by inhibiting osteoblast differentiation. It is consistent with previous reports that inflammatory cytokines prevent bone formation in vitro (Stashenko et al., 1987).
Pro-inflammatory cytokines, such as IL-1β and TNF-α, contribute to uncoupling in bone by reducing bone formation. To clarify the mechanisms through which inflammatory cytokines can affect bone formation, several aspects of bone formation have been studied. These may include the impact of inflammatory mediators on osteoblast precursors, number of osteoblasts, or osteoblast matrix-producing activity. Decreased osteoblast numbers can be achieved from reduced proliferation or enhanced apoptosis of osteoblasts or their precursor cells, both of which may be affected by inflammation. The effects of pro-inflammatory cytokines on cultured human osteoblast proliferation have been examined in vitro (Frost et al., 1997). TNF-α stimulated proliferation of osteoblasts at low doses, while it significantly inhibited proliferation of osteoblasts at high doses or prolonged treatment. This result suggests that excessive and prolonged TNF-α may have a destructive effect on bone by inhibiting its anabolic activity, and reinforces the concept that there may be an important spatial and temporal relationship between inflammation and bone coupling. In addition to affecting proliferation, TNF-α also inhibits the differentiation of osteoblast precursor cells (Gilbert et al., 2000). Treatment of TNF-α has been shown to reduce the differentiation of fetal calvaria precursor cells and MC3T3-E1 clonal pre-osteoblasts to the osteoblast phenotype, and reduces formation of mineralizing nodules in vitro. One mechanism through which this may occur is inhibition of a critical osteoblast differentiation factor, RUNX2, by TNF-α (Gilbert et al., 2002). This appears to be due to signaling through TNF receptor-1 (Gilbert et al., 2005). Thus, TNF-α inhibits RUNX2 activity, which in turn inhibits osteoblast differentiation.
Osteoblast survival is thought to be an important aspect of bone coupling. One of the causes of bone uncoupling could be diminished osteoblast or osteoblast precursor survival through induced apoptosis. Pro-inflammatory cytokines may directly stimulate osteoblast or osteoblast precursor apoptosis or indirectly affect it by stimulating expression of Fas, a potent pro-apoptotic mediator (Tsuboi et al., 1999). In periodontal bone remodeling, periodontal ligament cells are an important source of osteoblast precursors (Lin et al., 1994). TNF-α-induced apoptosis of periodontal ligament cells may affect the pool of osteoblast precursors (Thammasitboon et al., 2006). Apoptosis of mouse pre-osteoblastic MC3T3-E1 cells is induced by combined treatment of TNF-α, IL-1β, and IFN-γ and can be reduced by specific inhibitors of p38 MAP kinase (Kuzushima et al., 2006). More recently, we have found that MC3T3 apoptosis stimulated by pro-inflammatory cytokines is mediated by the pro-apoptotic transcription factor, forkhead box-O1 (FOXO1) (Behl et al., 2008). FOXO1 regulates expression of pro-apoptotic genes, including Fas-associated, via death domain (FADD) and caspases-3, -8, and -9. Moreover, increased apoptosis of MC3T3 cells is significantly reduced by knockdown of FOXO1 by small interfering RNA (Behl et al., 2008). FOXO1 belongs to the forkhead-O family, consisting of FOXO1, FOXO3, and FOXO4, which regulate cell death and cell cycle, and modulates the response to oxidative stress. Interestingly, FOXO1 has been linked to the regulation of serum glucose levels by modulating osteocalcin (Rached et al., 2010).
Another mechanism for uncoupling is reduced function of osteoblasts mediated by diminished production of bone matrix proteins. Bone matrix has organic and inorganic components, with the majority of the former consisting of type I collagen as well as other proteins and including osteocalcin and alkaline phosphatase. TNF-α reduces collagen production and alkaline phosphatase activity in cells obtained from fetal rat parietal bone (Centrella et al., 1988). In addition, production of non-collagen bone matrix proteins is inhibited by pro-inflammatory cytokines. TNF-α and IL-1β induce a two- to three-fold reduction in osteocalcin synthesis in osteoblastic cells (Taichman and Hauschka, 1992). Since TNF appears to interfere with bone formation through multiple mechanisms, the TNF specific inhibitor, etanercept, has been tested for its effect on improving the coupling process. Etanercept promotes BMP-2-induced ectopic bone formation when applied either systemically or locally in vivo (Eguchi et al., 2009). Similarly, inhibition of TNF with etanercept reduces apoptosis of bone-lining cells and significantly improves bone formation following P. gingivalis-induced bone resorption (Liu et al., 2006a). Thus, it is clear that pro-inflammatory cytokines such as IL-1β and TNF-α can affect bone formation and have a negative impact on bone coupling.
There is evidence that inflammation may affect coupling associated with bone resorption induced by periodontal pathogens. When a periodontal pathogen enters connective tissue, an inflammatory response is induced. It has been shown that activation of the acquired immune response by a periodontal pathogen inhibits bone coupling by limiting the formation of new bone following resorption (Leone et al., 2006; Behl et al., 2008). In these experiments, one group of mice was immunized against P. gingivalis. The host response to subsequent challenge of immunized and non-immunized mice with P. gingivalis was significantly greater in the immunized group. Interestingly, the immunized mice had significantly higher numbers of osteoclasts and higher levels of bone resorption. However, they also had greater apoptosis of bone-lining cells, fewer bone-lining cells, and reduced capacity to form new bone compared with non-immunized mice. One mechanism through which this may occur is enhanced cell death of osteoblastic cells or their precursors by the activation of pro-apoptotic transcription factors such as FOXO1 (Behl et al., 2008). Similarly, prolonged inflammation caused by diabetes appears to have a negative effect on bone coupling by stimulating death of bone-lining cells (Al-Mashat et al., 2006; Liu et al., 2006a), and this mechanism has been proposed in limiting new bone formation in other disease processes (Jilka, 2007).
Thus, several diseases that involve inflammation and bone metabolism exhibit uncoupling of bone resorption and bone formation. In fact, inflammation and bone metabolism have received so much attention that a new area of science has developed, called osteoimmunology (Arron and Choi, 2000). This interdisciplinary field integrates the disciplines of immunology and bone biology and provides a new context for the understanding of chronic diseases involving inflammation and bone, such as periodontal disease. The studies reviewed above have allowed us to propose a general model for periodontal disease, whereby inflammation stimulated by bacteria leads not only to osteoclast formation and subsequent bone resorption, but also to diminished bone formation. We propose that the spatial location of inflammation in relationship to bone and its duration are both critical features of the disease. With regard to the temporal component, if the inflammation is brief and does not extend into the phase of bone formation, the impact on periodontal bone loss may be relatively small, since there should be a considerable amount of coupling and new bone formation. However, if the inflammation lasts to the period of bone formation, then uncoupling is likely to occur, and the capacity to repair the resorbed bone is likely to be compromised. There also appear to be several mechanisms through which the inflammatory response may interfere with bone formation. One mechanism may be stimulated cell death of bone-lining cells, reducing the number of available cells or precursors that can form new bone. Alternatively, cytokines associated with the inflammation may inhibit the differentiation of osteoblasts from their precursors and thus limit bone formation. A third mechanism is reduced production of bone matrix due to the impact of inflammatory mediators. There is evidence that each of these mechanisms may play a role in various bone pathologies characterized by uncoupling of bone resorption and formation (Al-Mashat et al., 2006; Liu et al., 2006a; Diarra et al., 2007; Behl et al., 2008; Wang et al., 2009). All three mechanisms would reduce the numbers of osteoblastic cells available to form bone to equal and balance the amount of bone resorbed. This, in turn, would lead to uncoupling and greater periodontal bone loss.
The second consideration is spatial. If the inflammation is localized to the sub-epithelial space, it is unlikely to damage the underlying bone. However, if the inflammatory infiltrate progresses toward bone at some critical distance, the proximity of the inflammatory mediators to the bone surface will stimulate the recruitment of osteoclast precursors, osteoclastogenesis, and bone resorption. If the infiltrate persists near bone, we would predict that uncoupling will occur and result in net bone loss as observed clinically. Alternatively, if the inflammation near bone is transitory, we would expect that coupling would occur and new bone formation would occur, leading to little or no net bone loss.
For years, clinicians have identified periodontal disease on radiographs by examining the amount of bone loss that occurs around the dentition. Through a better understanding of inflammation and bone metabolism, a new paradigm exists to help explain periodontal bone loss.
We thank Juanita Hobson for help in preparing this manuscript. Support from NIDCR awards 5R01 DE017732 and 7R01 DE18307 is acknowledged. D. Graves was not involved in the review process of this manuscript.