IL-17–Mediated M1/M2 Macrophage Alteration Contributes to Pathogenesis of Bisphosphonate-Related Osteonecrosis of the Jaws

Nitrogen-containing bisphosphonates, a widely used anti-bone resorptive agent, has been associated with osteonecrosis of the jaw (ONJ), a complication of significant impacts in both medical and dental communities (1, 2). Bisphosphonate related osteonecrosis of the jaw (BRONJ) is defined as exposed necrotic bone in the oral cavity that does not heal after appropriate intervention over 8 weeks in the absence of radiotherapy (2, 3). The incidence of BRONJ in patients with cancer receiving high doses of intravenous bisphosphonates such as zoledronic acid and pamidronate (%) ranged from 0.8% to 12%, a rate much higher than that in patients with osteoporosis on oral bisphosphonates treatment. Among the cancer group, highest prevalence of bisphosphonate-related osteonecrosis of the jaw (BRONJ) has been reported in patients with multiple myeloma, followed by breast and prostate cancer (4–7). To date, even though several risk factors, including invasive dental procedure, infection, mechanical trauma to the jaw bone, and concomitant use of immunosuppressive and chemotherapy drugs, have been implicated in the etiology of BRONJ (4, 5), its underlying mechanisms remains largely unknown.

Macrophages play important role in innate immune response and are an essential component of the wound-healing cascade (8–10). Macrophages can be converted to M1 phenotype upon stimulation of T_H1 cytokines such as IFN-γ or lipopolysaccharide (LPS), whereas their M2 counterpart, also known as wound-healing macrophages, are polarized in response to T_H2 cytokines such as interleukin (IL)-4 or IL-13 (11). M1 macrophages are generally proinflammatory based on their production of nitric oxide (NO), reactive oxygen species (ROS), interleukin (IL)-12, and tumor necrosis factor (TNF)-α; whereas M2 macrophages more likely adapt an anti-inflammatory role, characterized by IL-10 and TGF-β (12–14), contributing to the resolution of inflammation and tissue modeling (15, 16). Given the importance of macrophages in wound healing and the fact that these cells share similar cell lineage as osteoclasts, it is plausible to hypothesize that macrophage function may potentially be altered by bisphosphonates treatment and therefore, contributes to the defective healing in BRONJ (17). Most recent studies have suggested a potential effect of zoledronic acid in the polarization of tumor-associated macrophages (TAM) in vitro (18, 19); it remains to be determined comprehensively whether and how the function and phenotypes (M1 vs. M2) of macrophages participate in the pathogenesis of BRONJ.

Interleukin (IL)-17A, a signature cytokine of the T-helper cell subset, T_H17, has been indicated in the development of numerous inflammatory, autoimmune diseases, tumors, and in host defense against bacterial and fungal infection (20). Besides T_H17 cells, γδ T cells and macrophages are capable to produce IL-17 (21–25). IL-17 can promote recruitment of macrophage and induce their cytokine/chemokine production, and thus capable of mediating a link between acquired and innate immunity, specifically, T-cell and macrophage functions (26–30). Most recently, we have shown that the administration of zoledronic acid can induce BRONJ-like lesions in mice, in part, by suppressing the adaptive regulatory T cells (Tregs) and activating the inflammatory T_H17 cells (31); this immunomodulatory effect linking elevated IL-17 and the prolonged inflammatory state associated with delayed wound healing potentially serve the underlying mechanism of BRONJ-like lesions.

In the present study, we first showed an elevated expression of IL-17 in the inflammatory socket wounds obtained from BRONJ-affected patients with cancer with a distinct histologic feature of increased M1 and decreased M2 macrophage infiltration. Using a BRONJ-like disease murine model, we observed a correlation between an elevated IL-17 level and a combination of increased M1 macrophages and decreased M2 macrophages (an increased M1/M2 ratio) in both long bone marrow and local necrotic lesions of the jaw bone. Interestingly, administration of IL-17–neutralizing antibodies or laquinimod, a small-molecule drug that exhibits inhibitory effects on IL-17 activity (32–34), led to a decrease in M1/M2 ratio and concomitantly, a reduction in the incidence of BRONJ-like lesion. In addition, systemic infusion of ex vivo expanded M2 macrophages significantly decreased serum IL-17 levels and inhibited BRONJ formation. Mechanistically, we showed that IL-17 promoted IFN-γ–induced M1 macrophage polarization by enhancing the phosphorylation of STAT-1; on the other hand, this proinflammatory suppressed IL-4–induced M2 macrophage conversion through downregulation of STAT-6 phosphorylation. Therefore, we describe for the first time substantial evidence that IL-17–mediated alteration of M1/M2 macrophage polarization can contribute to the pathogenesis of BRONJ.

C57BL/6J mice (female, 8–10 week-old, Jackson Laboratory), beige nude/nude Xid (III; female, 8–10 week-old, Harlan) were used in this study. All animal experiments were carried out under an Institutionally approved protocol for the use of animal research at University of Southern California (USC; USC #10874, #10941, and #11327).

The mucosal tissues bordering the extraction sockets were collected from patients undergoing tooth extraction under different conditions (n = 5): (i) BRONJ: patients manifested BRONJ lesion that were undergoing surgical debridement and tooth extraction for this condition; (ii) Non-BRONJ: patients with history of bisphosphonates treatment without clinical BRONJ who underwent tooth extraction for nonrestorable indication; (iii) Periodontitis: patients with diagnosis of inflammatory gum disease, specifically, active periodontitis with resultant tooth loss; and (iv) Normal: healthy patients who underwent routine dental extraction for other noninflammatory mucosa condition such as dental crowding indication. The BRONJ group includes 2 patients with multiple myeloma, 2 patients with breast cancer, and one patient with osteoporosis treated with intravenous or oral bisphosphonates, respectively. The study is approved by the Institutional review board (IRB#HS-08-00281) at University of Southern California (Los Angeles, CA).

Murine and human macrophage colony-stimulating factor (M-CSF), IFN-γ, IL-4, IL-13, and IL-17 were purchase from PeproTech. Rat phycoerythrin (PE)-conjugated anti-mouse CD11b and fluorescein isothiocynate (FITC)-conjugated anti-mouse F4/80, mouse PE-conjugated anti-human CD14 and CD68, and FITC-conjugated anti-human CD206 and CD4 were purchased from BioLegend. Rabbit polyclonal antibodies for human and mouse iNOS and arginase-1 were from Santa Cruz. A neutralizing antibody against mouse IL-17A (clone MM17F3) was from eBiosciences. LPS from Escherichia coli 055:B5, and phorbol 12-myristate 13-acetate (PMA) were obtained from Sigma-Aldrich

The BRONJ-like lesions in mice were induced according to the protocol we reported recently (31). Briefly, wild-type C57BL/6 mice or beige nude/nude Xid (III) burdened with multiple myeloma cells (5TGM) were intravenously injected with zoledronate (Zometa, 125 μg/kg, Novartis Oncology) twice a week via the tail vein. One week after zoledronate injection, the first maxillary molar teeth were extracted under deep anesthesia by intraperitoneal injection of ketamine (Ketaject, 100 mg/kg, Phoenix) and xylazine (AnaSed, 20 mg/kg, LLOYD Laboratories Pharmaceutical, Inc.). A total of 6 doses of zoledronate were administered for 3 consecutive weeks. Untreated mice with tooth extraction were used as controls. Clinical evaluation of extracted tooth socket and determination of the incidence of BRONJ-like lesions were done according to the previous study (31).

Bone marrow cells were flushed out from mice bone marrow cavity of femurs and tibias with PBS containing heat-inactivated 3% FBS (Equitech-Bio) and antibiotics (100 U/mL penicillin and 100 μg/mL streptomycin; Biofluids). All nuclear cells were seeded at 15 × 10⁶ onto 100 mm culture dishes (Corning) and initially incubated for 3 hours under 37°C at 5% CO₂ condition. Nonadherent cells were removed by washing the dishes twice with PBS. The attached cells were continuously cultured in complete Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% FBS, 100 U/mL penicillin, and 100 μg/mL streptomycin, and murine M-CSF (10 ng/mL) for 6 days. Macrophages were incubated for 48 hours in the presence of either 20 ng/mL IFN-γ or 10 ng/mL IL-4 to polarize macrophages into M1 and M2 macrophages, respectively. M1 macrophages were characterized by the expression of inducible nitric oxide synthase (iNOS) and secretion of TNF-α in response to LPS (100 ng/mL) stimulation for 24 hours, whereas M2 macrophages were characterized by the expression of arginase-1 and the secretion of IL-10 after stimulation with LPS (100 ng/mL) for 24 hours (13). One day after tooth extraction, cells were collected and intraperitoneally (i.p.) administered into mice (1.0 × 10⁶/mice).

One day after tooth extraction, 0.1 mg of mouse monoclonal anti-murine IL-17–neutralizing antibody (1 mg/mL) was intravenously injected into mice twice a week for 2 weeks while mice received subclass-matched control antibody (1 mg/mouse) served as control. In addition, one day after tooth extraction, laquinimod (TEVA Pharmaceuticals Industries, Ltd), a quinoline-3-carboxamide, was dissolved in purified water and administered daily (25 mg/kg) by oral gavage for 2 weeks (32–34). Mice without any treatment (control) and only with tooth extraction served as controls.

Paraffin-embedded and frozen sections of socket soft tissues from patients with BRONJ were prepared and standard hematoxylin and eosin staining and dual-color immunofluorescence studies using specific primary antibodies for human CD68, iNOS, and arginase-1 were conducted. Frozen sections of mice tissues including BRONJ-like lesions, spleens, and long-bone marrows were immunostained with specific antibodies for mice F4/80 and iNOS or arginase-1 to recognize M1 and M2 macrophages as described before (35). Isotype-matched control antibodies (eBiosciences) were used as negative controls. For semiquantification, positive signals in at least 6 random high-power fields (HPF) were visualized, counted, and expressed as percentage of total 4′, 6-diamidino-2-phenylindole (DAPI)-positive cells (mean ± SD).

Cell lysates or homogenate tissue lysates (50–100 μg of total protein) were separated on SDS-polyacrylamide gel and electroblotted onto nitrocellulose membrane (Bio-Rad). After blocking with TBS/5% nonfat dry milk, the membrane was incubated with antibodies against mouse p-STAT-1 (Tyr701; Cell Signaling), pSTAT-6 (Tyr641; Millipore), or arginase-1(Santa Cruz Biotech, Inc.) followed by incubation with a horseradish peroxidase (HRP)-conjugated secondary antibody, and the signals were visualized by enhanced chemiluminescence detection (ECL; Pierce). The blots were also reprobed with a specific antibody against β-actin (Sigma).

The concentration of IL-10, TNF-α, and IL-12 (p70) in supernatants of cultured cells was detected using ELISA kits (BioLegend) according to the manufacturer's procedures.

All data are expressed as mean ± SEM from at least 3 independent experiments. Differences between experimental and control groups were analyzed by 2-tailed unpaired Student t test using SPSS. P values less than 0.05 were considered statistically significant.

We have recently reported that intravenous administration of zoledronate, a potent nitrogen-bisphosphonate, induces BRONJ-like condition in mice undergoing tooth extraction; this adverse event is correlated with a decrease in Tregs and a concomitant increase in T_H17 cells and IL-17 production (31). Here, using immunofluorescence staining of the mucosal tissues bordering the nonhealing extraction socket obtained from BRONJ patients, we confirmed that IL-17–positive cells were dramatically increased at the local extraction sockets of patients manifested BRONJ lesion that were undergoing surgical debridement for this condition, as compared with that from control patients without clinical BRONJ (Fig. 1A and B). The control group included: (i) patients with diagnosis of inflammatory gum disease, specifically, active periodontitis with resultant tooth loss; and (ii) patients with history of bisphosphonate treatment who did not present with clinical BRONJ at the time of tooth extraction for nonrestorable indication. These IL-17–positive cells were positive for CD4⁺ and CD68⁺, indicating T cell, specifically T_H17 subset, and macrophage lineage, respectively. The elevated IL-17 expression was also confirmed by ELISA using tissue lysates obtained from the above patient groups (Fig. 1C).

To investigate the phenotype of macrophage, specifically in the context of wound healing, we examine the distribution of M1 and M2 macrophages at the BRONJ socket site. A remarkable infiltration of CD68⁺ macrophages was noticed in mucosal tissues adjacent to the extraction sockets of periodontitis affected tooth as compared to clinically healthy tissues (P < 0.01) of patients exposed or not exposed to bisphosphonates treatment (Fig. 2A–E); likewise, abundant CD68⁺ macrophages were also observed in mucosal tissues of BRONJ extraction sockets relative to those of non-BRONJ (P < 0.05; Fig. 2E). Among the general surge in CD68⁺ macrophages activity, we noticed a remarkable shift in iNOS-positive (M1) over CD206⁺ (M2), resulting in an increased M1/M2 ratio, at the nonhealing sockets of patients with history of zoledronate treatment that were diagnosed with or without BRONJ, as compared with those of periodontitis-affected patients that were naïve of zoledronate exposure (P < 0.01; Fig. 2A–D and 2F and G). Of note, polarization of M1/M2 ratio was more pronounced at the mucosal tissue bordering the extraction sockets of BRONJ relative to non-BRONJ patients (Fig. 2F and G). The abundant activity of M1 macrophages was also confirmed using a quick cell smear screening approach obtained from the socket of patients with BRONJ (Supplementary Fig. S1). These results suggest that a combination of enhanced IL-17 and M1/M2 activity may situate the wound environment more of a prolonged inflammatory state, rendering a generalized delay in the healing of the open extraction socket and subsequently, osteonecrosis of the jaw.

To determine whether the enhanced T_H17 and IL-17 activity is limited to the local mucosal tissue bordering the nonhealing extraction socket of BRONJ mice, we tested the effect of elevated serum IL-17 on the development of BRONJ. Here, we conducted a series of human serum transfusion in our murine model using samples collected from patients with BRONJ undergoing surgical debridement versus non-BRONJ patients with history of bisphosphonates treatment or healthy controls. We measured serum IL-17 level in BRONJ-affected patients, range 100–200 pg/mL (182.5 ± 12.4 pg/mL), in non-BRONJ patients, range approximately 50 pg/mL (58.2 ± 9.8 pg/mL), as well as in healthy controls, less than 30 pg/mL (25.4 ± 6.3 pg/mL; n = 5/group; Supplementary Fig. S2A). During the course of bisphosphonate induction of BRONJ, serum sample from each patient group was intravenously infused into C57BL/6 mice one day after tooth extraction and manifestation of BRONJ-like condition was assessed as previously described (31). Interestingly, we observed development of exposed bone (ONJ) or BRONJ-like condition in mice that underwent systemic infusion of serum from patients who manifested active BRONJ, and these ONJ mice showed relatively high serum levels of murine IL-17 and increased percentage of M1 macrophage or M1/M2 ratio as detected in the long bone marrow (Supplementary Fig. S2B–E). Despite the seeming correlation between infused human serum IL-17 and ONJ incidence in our murine model, we cannot rule out other factors that potentially contribute to this effect. Further studies are ongoing to address this finding.

As it is well established that BRONJ occurred predominantly in the jaw bone, we examined whether there is any difference in IL-17 level in bone marrow derived from jaw versus long bone. Interestingly, we found a remarkably increase in T_H17 cells and IL-17 expression at both mRNA and protein levels in jaw bone marrow as compared with long bone marrow of C57BL/6 mice, whereas Treg cells and IL-10 activities were relatively suppressed (Supplementary Fig. S3A–F); similar trends of increased T_H17 and decreased Treg cells were consistently observed in bone marrow compartment of jaw versus long bone in other species, C3H mice (Supplementary Fig. S3G and S3H), and rats (Supplementary Fig. S3I and S3J). We next ask whether the differential IL-17 activity affect bone healing at different skeletal sites. Consistent with previous findings, we showed that IL-17 elevation correlated with high incidence of exposed bone or ONJ in zoledronate-treated wild-type C57BL/6 or Bg ZID (III) nu/nu mice (31) following tooth extraction. However, when subjected nu/nu mice to surgically induced bone cavitation in the appendicular bone, specifically the tibia, there was apparently no difference in bone healing between zoledronate-treated and non-treated groups (Supplementary Fig. S4). When evaluating jaw bone healing, residual necrotic alveolar bone was evident with absolutely no signs of bone filling in the extraction socket of the zoledronate-treated mice (Supplementary Fig. S4A); whereas, in the long bone cavitation defect, similar callus formation and bone filling was observed in mice that manifested clinical BRONJ in the jaw bone as compared with nontreated control (Supplementary Fig. S4B). Taken together, the inherent increase in T_H17 activity unique to the jaw bone, the open tooth extraction wound exposed to the oral microbial environment versus the closed and sterile wound of the long bone cavitation, and specifically its oral mucosa immunology may explain the limited manifestation of BRONJ at this bone type.

We further ask whether zoledronate directly affect systemic IL-17 level using our established model of BRONJ-like disease that showed a high correlation between IL-17 level and incidence of BRONJ in wild-type C57BL/6 mice (31). Consistent with previous findings, zoledronate treatment augmented serum level of IL-17 (Fig. 3A) over the entire course of treatment; more importantly, mice that manifested BRONJ-like condition displayed a much higher level of serum IL-17 than those without BRONJ (P < 0.01; Fig. 3B). At the local mucosa of the nonhealing socket, elevated expression of IL-17 mRNA was detected as confirmed by real-time PCR (Fig. 3C). These results further support the notion that an elevated IL-17 expression at both systemic and local levels contributes to the development of BRONJ-like condition.

It has been suggested that the impaired wound-healing characteristic of BRONJ may be attributed to macrophage dysfunction (17); however, up to date evidence remains lacking. Here, using our established BRONJ-like murine model, we explore the potential involvement of macrophages in IL-17–mediated inflammation as a mechanism of impaired healing in osteonecrosis of the jaw bone. To this purpose, we initially analyzed mRNA expression of iNOS and arginase-1, a marker for M1 and M2 macrophages, respectively, in mucosal tissues bordering extraction socket of mice following treatment with zoledronate. As shown in Fig. 3D–F, an overall elevation of both iNOS and arginase-1 mRNA expression was detected, with a more remarkable increase in iNOS than arginase-1 (P < 0.05), suggesting a dominance of M1 macrophages in the nonhealing extraction socket of BRONJ. To further confirm this, we examined the local and systemic distribution of M1 and M2 macrophages in C57BL/6 mice after zoledronate treatment. Using dual-color immunofluorescence staining of mucosal tissue sections obtained from ONJ sockets, we showed a significantly increased percentage of CD11b⁺ cells expressing iNOS (M1 macrophages) and a decreased percentage of CD11b⁺ cells expressing arginase-1 (M2 macrophages; P < 0.01) (Fig. 3G and H). To determine whether zoledronate has direct effect on macrophage phenotype switching and function, we exposed bone marrow-derived macrophages (BMDM) from BRONJ and non-BRONJ mice under M1 and M2 macrophage induction conditions, respectively. Our results indicated that a higher percentage of M1 macrophages and a lower percentage of M2 macrophages were induced from BMDMs of zoledronate-treated mice that manifested BRONJ-like conditions, as compared with those of non-BRONJ mice (P < 0.05), or nontreated mice (P < 0.01) (Supplementary Fig. S5A–D). Taken together, these findings support for the first time, an immune underlying mechanism, linking the elevated IL-17 activity and a shift in M1/M2 macrophage ratio to the development of BRONJ-like condition in the murine model.

Recent studies have reported a significantly elevated baseline and induced frequency of T_H17 cells in peripheral blood mononuclear cells and an increased serum IL-17 levels in multiple myeloma patients (36), implying a critical role of T_H17 cells in the genesis of multiple myeloma bone lytic diseases (37). In addition, it has been well documented that the highest prevalence of BRONJ occurs in patients with multiple myeloma receiving zoledronate treatment (4, 5). To test the potential contribution of IL-17 in the pathogenesis of BRONJ in multiple myeloma, we induced BRONJ-like condition in a well-established multiple myeloma murine model (Fig. 4A–C) adapting the protocol used in wild-type C57BL/6 mice as previously published (31). Manifestation of BRONJ-like phenotype followed tooth extraction was confirmed by clinical examination (Fig. 4B). Overall, multiple myeloma mice displayed a significantly elevated basal level of serum IL-17 as compared with wild-type C57BL/6 mice (Fig. 4D vs. Fig. 3B), and zoledronate treatment further augmented serum IL-17 levels above the baseline (P < 0.05). Consistent with the incidence of BRONJ in zoledronate induced wild-type C57BL/6 (31), in multiple myeloma mice we observed a baseline disease of 20% (Fig. 4C), which correlated with similar serum IL-17 level in both disease models (Fig. 3B; Fig. 4D). Intravenous administration of zoledronate into multiple myeloma mice caused a remarkably higher incidence of BRONJ (∼80%) as compared with wild-type C57BL/6 mice (Fig. 4B and C). The 2-to 3-fold increase in ONJ incidence in drug-induced multiple myeloma mice corresponded to the augmented serum IL-17 level (Fig. 4D); furthermore, consistent with previous findings in wild-type mice, we observed an increased M1 and a decreased M2 macrophage activity, specifically a significant shift in M1/M2 ratio, in mucosa of multiple myeloma mice sustained BRONJ-like condition (P < 0.05; Fig. 4E and F).

To further explore the critical role of IL-17 and altered M1/M2 macrophage balance in the development of BRONJ-like condition, we tested several therapeutic approaches modifying macrophage function or targeting IL-17 using: (i) adoptive transfer of ex vivo expanded M1 and M2 macrophages; (ii) treatment with a popular small-molecule inhibitor of IL-17 activity (laquinimod); and (iii) treatment with a specific IL-17–neutralizing antibody (Fig. 5A). We showed that adoptive transfer of M1 macrophages had no obvious effect on serum IL-17 levels and BRONJ incidence (data not shown); on the contrary, infusion of M2 macrophages significantly suppressed the incidence of BRONJ in multiple myeloma mice (Fig. 5B–D) and serum levels of IL-17 (Fig. 5E). In addition, we showed that oral administration of laquinimod reduced the incidence of BRONJ (Fig. 5B–D), and simultaneously led to a decrease in serum IL-17 levels (Fig. 5E) and M1/M2 macrophage ratio in long bone marrow of multiple myeloma mice (Fig. 5F and G). Convincingly, administration of IL-17–neutralizing antibody led to a remarkable decrease in M1/M2 ratio in long bone marrow (Fig. 5F and G), and concomitantly, a significant reduction in the incidence of BRONJ in multiple myeloma mice (Fig. 5B–D). Taken together, these results have provided substantial evidence that an elevated IL-17 level and a shift in M1/M2 macrophage polarization contribute to the delayed healing in zoledronate induced osteonecrosis of the extracted socket in murine jaw bone.

We next explored the effects of IL-17 on polarization of M1/M2 macrophages in vitro and their potential mechanisms. To this end, BMDMs from C57BL/6 mice or human macrophages (THP-1) were cultured under M2 macrophage induction condition in the presence or absence of IL-17 for 48 hours. Exposure to exogenous IL-17 enhanced murine and human M1 macrophage polarization as represented by an increased secretion of proinflammatory cytokines TNF-α and IL-12 (P < 0.05; Fig. 6A and B). As INF-γ–mediated STAT-1 signaling pathway plays a central role in polarization of M1 macrophages (38), we then examined whether exogenous IL-17 has any effect on INF-γ–mediated activation of STAT-1. We showed that IFN-γ induced phosphorylation of STAT-1 (Tyr701) at 2 and 4-hour time intervals, and even a more pronounced induction of p-STAT1 was noticed at 12 and 24 hours after IFN-γ stimulation (Fig. 6C). Semiquantitative analysis showed that addition of IL-17 enhanced IFN-γ–induced phosphorylation of STAT-1 by about 2-fold at 4 hours, and by about 1.5-fold at 12 and 24 hours, respectively (Supplementary Fig. S6). These results suggest that IL-17 promotes M1 macrophage polarization possibly by enhancing INF-γ–mediated activation of STAT-1 signaling pathway.

We then examined the effects of exogenous IL-17 on M2 macrophage polarization and its underlying mechanism. Using a series of studies with BMDMs from C57BL/6 mice or human macrophages (THP-1), we applied M2 macrophage polarization condition in the absence or presence of IL-17. Our results showed that pretreatment with IL-17 dramatically inhibited murine and human M2 macrophage polarization as represented by a reduced secretion of the anti-inflammatory cytokine IL-10 (Fig. 6D and E) and a decreased expression of arginase-1in mice BMDMs (Fig. 6F), a specific marker for M2 macrophages. We next examined the effects of exogenous IL-17 on IL-4–mediated STAT-6 signaling pathway in the polarization of M2 macrophages. The data indicated that pretreatment with IL-17 significantly suppressed the phosphorylation of STAT-6 in mice BMDMs stimulated by IL-4 (Fig. 6G), suggesting that IL-17 inhibited M2 macrophage polarization by directly suppressing IL-4–mediated activation of STAT-6 signaling pathway. Further in-depth studies are underway to delineate the underlying mechanism of IL-17–mediated activation of STAT-1/STAT-6 balance and how they contribute to M1 and M2 switching in delayed healing in drug-induced osteonecrosis of the jaw bone.

Over the last decade, osteonecrosis of the jaw (ONJ) has emerged as a devastating and debilitating condition of oncologic patients receiving treatment with high doses of antiresorptive agents (bisphosphonates and denosumab; ref. 39). Epidemiologic data from the latest cohort studies indicate that 88% of the 2408 ONJ cases were associated with intravenous therapy primarily with zoledronate alone, or zoledronate sequentially with pamidronate, whereas only 261 cases (11%) had ever received oral bisphosphonate treatment. Among these ONJ cases, the majority of them (89%) were oncologic patients, with multiple myeloma accounting for 43%, breast cancer for 32%, prostate cancer for 9%, and other cancers for 5% (40). To date, several thousands of ONJ cases have been reported in the literature; however, the pathophysiology or etiology of this condition remains largely unknown. Even though several potential mechanisms have been implicated in the pathophysiology of BRONJ, including oversuppressed alveolar bone turnover, toxicity to oral mucosal tissues, altered angiogenesis and deregulated immune functions that are directly or indirectly caused by bisphosphonate therapy (17, 39), none of them can delineate the full spectrum of the pathophysiology of BRONJ. Conceivably, both epidemiologic/retrospective and clinical prospective studies have identified several major risk factors for BRONJ including the type, the route, and the cumulative dose of bisphosphonates, preexisting infection, oral trauma/injury such as dentures and tooth extraction, the presence of metastases and obesity, and other medical morbidity conditions (17, 39). Inspite of these factors identified so far that are related to BRONJ prevalence, no direct causal effects have been established due to the lack of evidence-based studies and a well-accepted BRONJ-like animal model that can recapitulate the major pathophysiologic characteristics of BRONJ in patients.

Most recently, we have established a BRONJ-like model in mice by intravenous administration of zoledronate in conjunction with the addition of immunosuppressive agent dexamethasone preceding tooth extraction, which displayed similar hallmarks of human diseases including the persistent presence of BRONJ-like clinical, radiographic, and histologic features at the extraction site (31). More importantly, we showed that the BRONJ-like condition was closely associated with an altered immune homeostasis, specifically a deficiency in Tregs and a significant increase in T_H17 cells and IL-17 level in peripheral blood, which might contribute to the prolonged inflammation at the local extraction sites (31). In the present study, an elevated IL-17 expression was also observed in the serum and local mucosal tissues bordering the nonhealing extraction socket of BRONJ patients (Fig. 1 and Supplementary Fig. S2A). We further showed that an elevated serum level of IL-17 was tightly correlated to an increased incidence of BRONJ-like conditions in wild-type C57BL/6 mice (Fig. 3) and multiple myeloma mice (Fig. 4). Moreover, blocking IL-17 activity using specific neutralizing antibody significantly reduced the incidence rate of BRONJ in multiple myeloma mice (Fig. 5), thus further confirmed the essential role of IL-17 in the pathogenesis of this condition. These findings further support the notion that an elevated IL-17 level may serve as a major risk factor in patients sustaining BRONJ. Consistent with our findings, in both murine model and human BRONJ disease, a recent study reported that patients with multiple myeloma experienced an elevated serum level of IL-17 and other related proinflammatory cytokines, including IL-21, IL-22, and IL-23 (36). However, an earlier short letter described decreased serum IL-17 levels in patients with multiple myeloma following administration of bisphosphonates and those presented with ONJ as compared with nontreated multiple myeloma or multiple myeloma patients without ONJ (41). This seemingly controversial observation needs to be addressed in light of the well-recognized immunostimulatory effect of the aminophosphates and related compounds (42, 43). Clinical studies by our group using increased sample size for ONJ and non-ONJ patients in oncologic patients with multiple myeloma or other type of cancers are in progress to further confirm the changes in IL-17 levels both locally and systemically, and to determine whether these levels correlate with ONJ phenotypes.

As a potent proinflammatory cytokine, IL-17 is principally produced by a subset of CD4⁺ T_H17 cells, but recent evidence has shown that it can also be produced by innate immune cells, including γδ T cells, macrophages, and mast cells under certain pathophysiologic settings (21–25). Accumulating evidence has shown that IL-17–mediated inflammation is crucial to the pathogenesis of several inflammatory and autoimmune diseases including rheumatoid arthritis, multiple sclerosis, systemic lupus erythematosus, inflammatory bowel disease, and psoriasis (20). Most recently, several lines of evidence have showed the important role of IL-17–mediated inflammation in bone loss or destruction-related diseases (37, 44–46). IL-17 drives irreversible bone loss or destruction by directly promoting osteoclastogenesis (37, 47) and/or by interplaying with other proinflammatory cytokines like TNF-α to enhance the expression of several important inflammatory mediators (48). In addition, the interplay between IL-17 and macrophages may exaggerate IL-17–mediated inflammatory responses. On one aspect, classically activated or M1 macrophages can produce cytokines such as IL-6 and IL-23 that are essential for T_H17 differentiation (11, 20); on the other aspect, IL-17 can promote the recruitment and proinflammatory cytokine/chemokine production of macrophages (26–30), suggesting a potential role of IL-17 in the induction of M1 macrophages. Under certain conditions, secretion of IL-17 by macrophages in response to proinflammatory or allergic stimuli can play a major role in the inflammatory process (22, 23). In this study, we have provided the first line of evidence that an elevated IL-17 expression was closely correlated with an increased M1/M2 macrophage ratio at the local mucosal tissue of nonhealing extraction socket of BRONJ patients, whereby the elevated IL-17 level at the local sites was shown to be attributed to both T_H17 cells and macrophages (Figs. 1 and 2). Using mice BRONJ-like disease model, we have shown that an elevated IL-17 production in mice with BRONJ is closely associated with a significant increase in the percentage of M1 macrophages and a decrease in M2 macrophages, which is represented as an increased M1/M2 ratio (Fig. 3 and Supplementary Fig. S5); moreover, blocking of IL-17 activity by using a neutralizing antibody or small immunosuppressive molecule such as laquinimod reversed the alteration in M1/M2 macrophages ratio and concomitantly reduced the incidence of BRONJ in mice (Fig. 5). Convincingly, adoptive transfer of M2 macrophages led to a remarkable decrease in serum levels of IL-17 and a corresponding reduction in the incidence of BRONJ in multiple myeloma mice (Fig. 5). In vitro studies further showed that exogenous IL-17 promoted M1 macrophage polarization via augmenting IFN-γ–mediated activation of STAT-1 signaling pathway, and inhibited M2 macrophage polarization by interfering IL-4–mediated activation of STAT-6 signaling pathway (Fig. 6). These findings have provided compelling evidence that IL-17–mediated alterations in M1/M2 macrophages contribute a crucial role in the pathophysiology of BRONJ-like disease.

Zoledronic acid and other bisphosphonates are capable of suppressing bone loss via inhibition of osteoclast activity (1–3). Several evidences describe the close interplay between T_H17 cells and M1 macrophages (11, 20; 26–30); it remains largely unknown whether these interactions participate in zoledronate-mediated IL-17 increase. Recent studies have shown that suppressor of cytokine signaling-3 (SOCS3) can repress the proinflammatory M1 macrophage phenotype, whereas SOCS3 deficiency promotes M1 macrophage polarization and inflammation via an increased production of proinflammatory cytokines IL-1β, IL-6, IL-12, and IL-23, rendering a microenvironment conditioned for the differentiation of T_H1 and T_H17 cells (49, 50). Interestingly, zoledronic acid has been shown to induce M2-like TAMs toward an M1 phenotype (51) and enhances proinflammatory cytokine production through inhibition of SOCS3 expression in macrophages (19). These findings could explain the mechanisms whereby zoledronic acid drives T_H17 differentiation, at least in part, through promoting M1 macrophage polarization and activation (18). On the basis of these studies and our current findings, we postulate a positive feedback loop between zoledronate-induced IL-17 elevation and M1/M2 macrophages switch, regulated in part via STAT-1/STAT-6 activation, as an underlying mechanism in BRONJ pathogenesis (Supplementary Fig. S7). However, it is noteworthy that microbial flora and its biofilm, a common feature of ONJ (52), can also affect differentiation and function of both macrophages and T_H17 cells. Further in-depth studies are required to delineate how zoledronic acid and the oral microbial flora can dependently and/or independently modulate the complex network among T_H17, M1/M2 macrophages and other types of immune cells in ONJ. In addition to zoledronate, recent clinical trials have reported that patients treated with denosumab, a monoclonal antibody against human receptor activator of NF-κB ligand (RANKL) popularly prescribed as an antiresorptive drug in patients with osteoporosis and certain cancers, experienced similar incidence of ONJ (39, 53). The fundamental question whether denosumab induced ONJ shares similar pathways as that of bisphosphonate-related ONJ, particularly their effects on the interplays between T_H17 cells and M1/M2 macrophages, remains to be determined.

A growing body of evidence has shown that activation M1 macrophage within tumor microenvironment can boost inflammatory reaction as a mechanism for immune surveillance of cancer cells. This same mechanism is contributory to the prolonged inflammatory state associated with delayed wound healing, as observed in the refractory soft tissue collected from BRONJ site in human studies (Fig. 2). The TAMs with similar characteristics of M2 macrophages can suppress immune response and promote angiogenesis and metastasis of tumor cells, thus facilitating cancer progression, or attenuating inflammation, and enhancing wound healing (54). Recently, both in vitro and in vivo studies have shown that tumor macrophages are potential targets of bisphosphonates (18, 19, 51), possibly via the manipulation of macrophage phenotypes in tumor microenvironment, and by similar mechanisms, sustain prolonged inflammation leading to an increased susceptibility to BRONJ. Epidemiologic study has identified several risk factors associated with ONJ; in patients with cancer receiving high-dose intravenous bisphosphonates for the prevention and treatment of skeletal-related complications, treatment with other antitumor drugs including glucocorticoid (dexamethasone), chemotherapeutic, and/or antiangiogenic drugs incur additional risks; as such, it should be further explored whether bisphosphonates and these antitumor drugs have any cooperative effects on the immune system, specifically on the interplay between T_H17 cells and M1/M2 macrophages, as these immune cells potentially function as a “double-edge” sword within the tumor microenvironment to enhance both antitumor immunity and inflammatory responses (18, 39). Such mechanisms inadvertently result in prolonged inflammation, and thus delayed healing or bone necrosis manifested as BRONJ.

In summary, we have shown for the first time to our knowledge that a combination of elevated IL-17 activity and a shift in M1/M2 macrophages ratio, contributes to the pathophysiology of BRONJ-like condition in both human and murine models. Blocking IL-17/M1 macrophage axis using small molecules or neutralizing antibodies can offer novel therapeutic modality for preventing and treating BRONJ condition in the vulnerable patients with cancer.

No potential conflicts of interest were disclosed.

Conception and design: Q. Zhang, I. Atsuta, S. Shi, A.D. Le

Development of methodology: Q. Zhang, I. Atsuta, C. Chen, A.D. Le

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): I. Atsuta, S. Liu, C. Chen, S. Shi, S. Shi, A.D. Le

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): Q. Zhang, C. Chen, S. Shi, A.D. Le

Writing, review, and/or revision of the manuscript: Q. Zhang, S. Shi, A.D. Le

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): Q. Zhang, C. Chen, S. Shi, A.D. Le

Study supervision: S. Shi, A.D. Le

This work was supported by NIH Research Grant, R01DE 019932, Oral and Maxillofacial Surgery Foundation (OMSF) Research Grant, USC Institutional funding (CTSI and Zumberge awards), and the Schoenleber funding support.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.