Langerhans cells down-regulate inflammation-driven alveolar bone loss

Abstract

Excessive bone resorption is frequently associated with chronic infections and inflammatory diseases. Whereas T cells were demonstrated to facilitate osteoclastogenesis in such diseases, the role of dendritic cells, the most potent activators of naive T cells, remains unclear. Using a model involving inflammation-driven alveolar bone loss attributable to infection, we showed that in vivo ablation of Langerhans cells (LCs) resulted in enhanced bone loss. An increased infiltration of B and T lymphocytes into the tissue surrounding the bone was observed in LC-ablated mice, including receptor activator of NF-κB ligand (RANKL)-expressing CD4⁺ T cells with known capabilities of altering bone homeostasis. In addition, the absence of LCs significantly reduced the numbers of CD4⁺Foxp3⁺ T-regulatory cells in the tissue. Further investigation revealed that LCs were not directly involved in presenting antigens to T cells. Nevertheless, despite their low numbers in the tissue, the absence of LCs resulted in an elevated activation of CD4⁺ but not CD8⁺ T cells. This activation involved elevated production of IFN-γ but not IL-17 or IL-10 cytokines. Our data, thus, reveal a protective immunoregulatory role for LCs in inflammation-induced alveolar bone resorption, by inhibiting IFN-γ secretion and excessive activation of RANKL⁺CD4⁺ T cells with a capability of promoting osteoclastogenesis.

A connection between inflammation and bone disease has been established in numerous clinical and animal studies (1). Bone loss attributable to inflammation could be found in many diseases, such as rheumatoid arthritis, multiple myeloma, diabetes mellitus, lupus erythematosus, and periodontal disease (2). Such diseases could develop because of infection or autoimmunity; they could be acute or chronic and either local or systemic. Although the mechanisms underlying the bone loss are complex, it appears that they involve disruption of the bone remodeling cycle. Bone formation and resorption is mediated by two cell types: osteoblasts and osteoclasts, respectively. Receptor activator of NF-κB ligand (RANKL), its receptor RANK, and the natural antagonist osteoprotegrin (OPG) are key regulators of the differentiation, activation, and survival of osteoclasts (3). This bone-remodeling cycle is tightly regulated under physiological conditions to guarantee bone homeostasis. Under inflammatory conditions, however, the level of RANKL is increased because of its expression and secretion by inflammatory lymphocytes, which facilitate osteoclastogenesis and bone resorption (4).

Dendritic cells (DCs) are a heterogeneous family of potent antigen-presenting cells with the capability to prime naïve T cells. With regard to inflammation-driven bone loss, attention was given to DCs as cells with an osteoclastogenic potential. DCs derive from the same myeloid precursor as osteoclasts, and the function of both cell types is modulated by common factors. Because of the high developmental and functional plasticity of myeloid cells such as DCs, it has been proposed that under certain inflammatory conditions, DCs could differentiate into osteoclast-like cells with a bone resorption capability (5). Nevertheless, as the major cells activating T cells, it is likely that DCs will also have a critical influence on the intensity and quality of the inflammation, thereby acting as potential osteo-immune players.

Langerhans cells (LCs) are a unique DC subset expressing the C-type lectin receptor Langerin and are located in mucosal-stratified squamous epithelium and skin epidermis (6). The development of transgenic mice allowing conditional ablation of LCs in vivo has contributed to our understanding of the role of LCs in eliciting adaptive immune responses and in inducing and maintaining tolerance. The frontline localization of LCs makes them prone to probe the environment and orchestrate T-cell immunity against innocuous or infectious agents. This led us to hypothesize that LCs could play a role in inflammation-driven bone loss, by regulating the quality of the inflammatory reaction.

Results

In Vivo Ablation of LCs Aggravates Local Inflammation and Enhances Alveolar Bone Loss.

Langerin–diphtheria toxin receptor (DTR) mice were administered i.p. with diphtheria toxin (DT) 7 and 5 d before infection to eliminate LCs, as previously shown (7). Infection with the pathogen Porphyromonas gingivalis was performed via oral gavage, a relevant model for inducing resorption of alveolar bone. Six weeks after the infection, the hemimaxillae were harvested and scanned using micro computed tomography (μCT) to measure alveolar bone volume. Fig. 1A presents representative μCT sections of the second upper molar following infection. The distance between the cemento-enamel junction (CEJ) and alveolar bone crest (ABC) in DT-treated mice was larger compared with mice exposed to vehicle alone or infected mice with no DT treatment. This indicates that the lack of LCs resulted in a considerable resorption of the alveolar crest. When bone morphology was evaluated, an irregular cortical plate with small radiolucent punched-out lesions was observed in the alveolar bone where LCs were ablated, suggesting the occurrence of active bone loss (Fig. 1A). Such lesions were only seen in the ABC of infected mice in the absence of LCs and were not found in the carboxymethylcellulose (CMC)-treated control group. The role of LCs in down-modulating bone resorption was further demonstrated by a 3D quantification of the residual alveolar bone (P < 0.005 compared with infected mice with no DT treatment) (Fig. 1B).

Fig. 1.

Enhanced bone resorption and local inflammation in LC-ablated mice. Langerin-DTR mice were injected i.p. with DT 7 and 5 d before infection. Antibiotic pretreated mice were infected via oral gavage, three times at 2-d intervals, with 4 × 10⁹ cfu of P. gingivalis strain 381 (Pg). Six weeks after the last inoculation, the maxillae of the mice were harvested and examined for alveolar bone loss. (A) Representative μCT sections of the second upper molar demonstrating the impact of LC deletion on bone structure. White arrows indicate lesions in the alveolar bone. CEJ, cemento-enamel junction. (B) 3D quantification of the residual alveolar bone. Data are presented as the volume of alveolar bone in the buccal plate and represent the means of eight mice per group ± SEM. Results are representative of two independent experiments. (C) Three weeks after the last inoculation, gingival tissues were enzymatically digested, stained with antibodies against CD3, CD4, CD8, CD19, and RANKL, and subjected to flow cytometric analysis. Graphs demonstrate the numbers of B and T lymphocytes, as well as T-cell subsets, in the bone-surrounding tissue. (D) Quantification of RANKL-expressing T cells. Data represent the means of five mice per group ± SEM. (E) Numbers of CD4⁺Foxp3⁺ T regulatory cells in the gingiva 5 wk postinfection. Data represent the means of five mice per group ± SEM. **P < 0.0001 compared with infected mice with no DT treatment; *P < 0.05 compared with mice receiving vehicle only. One representative out of three independent experiments is depicted.

To better understand the impact of LC ablation on local inflammation, we analyzed lymphocyte infiltration into the bone-surrounding gingiva 2–3 wk after the first infection. In the absence of LCs, the number of infiltrating lymphocytes increased 3.5- and twofold compared with mice exposed to CMC or P. gingivalis alone, respectively (Fig. 1C). Within the lymphocyte population, B cells were considerably increased in both infected groups, with a two times higher number of B cells in the absence of LCs. In contrast to B cells, higher numbers of T cells were detected only in infected mice lacking LCs (P < 0.0001). The overall percentages of T-cell subsets were comparable across the various groups, with the exception of higher CD4⁺ T-cell frequencies in LC-depleted mice (P < 0.001) (Fig. S1). In line with this finding, CD4⁺ T cells were the major T-cell population infiltrating the inflamed tissue, with an ∼3.2-fold increase in LC-depleted infected mice compared with controls (P < 0.0001) (Fig. 1C). Moreover, we detected increased numbers of CD8⁺ T cells, natural killer (NKT) cells, and γδT cells in the tissue without LCs. Because expression of RANKL on CD4⁺ T cells was reported to regulate osteoclastogenesis and, thus, facilitate bone loss (8), we determined its expression. Intriguingly, the inflamed gingiva contained large numbers of RANKL-expressing CD4⁺ T cells (Fig. 1D). Of note, we could not detect a significant difference in the level of RANKL expression on a per-cell basis between infected mice in the presence or absence of LCs. In addition to CD4⁺ T cells, our analysis revealed that a high percentage of NKT cells and γδT cells expressed RANKL as well (Fig. 1D). We next addressed the presence of T-regulatory (Treg) cells in the gingiva. An approximate twofold decrease in the numbers of CD4⁺Foxp3⁺ T cells was detected in the absence of LCs (Fig. 1E). Collectively, our results indicate that the absence of LCs leads to a massive infiltration of lymphocytes and, thus, to a higher inflammatory response in the bone-surrounding tissue. Moreover, ablation of LCs resulted in reduced numbers of Treg cells and in the infiltration of a large number of RANKL-expressing T cells previously shown to be involved in alveolar bone resorption (9).

Characterization of DC Subsets in the Tissue Surrounding the Alveolar Bone.

As the most potent cells initiating adaptive immunity, DCs are likely major players in diseases involving inflammation-mediated bone loss. It is, thus, essential to examine the capabilities of such tissue-resident DCs in inducing adaptive immunity and controlling local inflammation. Nevertheless, our knowledge regarding the various DC subsets located in the gingiva, the tissue adjacent to alveolar bone, is limited. To characterize these DCs, we isolated the cells from gingival tissue of mice (Fig. S2), by enzymatic digestion and analyzed them by flow cytometry. Based on expression of CD11c, CD103, epithelial cell adhesion molecule (Ep-CAM), Langerin, and MHC class II, we were able to distinguish three distinct DC subsets. Firstly, CD11c⁺Ep-CAM^negCD103^negLangerin^neg cells represent the largest DC population that is equivalent to dermal DCs and termed interstitial (i)DCs (Fig. 2A). Secondly, we identified a population of CD11c⁺Ep-CAM^brightLangerin⁺ cells, which are the equivalents of epidermal LCs that reside in the gingival epithelium (Fig. 2A). Of note, LCs represent only 2–5% of the total CD11c⁺ DCs in the gingiva. Staining with MHC class II and Langerin also confirmed that Langerin-expressing DCs are a minor population (∼3%) in the tissue (Fig. 2B). A low frequency of LCs was also reported in human gingiva (10), demonstrating the high resemblance to mice. Thirdly, we observed that ∼3–4% of the iDC population coexpressed CD103. Interestingly, in contrast to the skin in which CD103⁺ DCs are also Langerin-positive (termed Ln⁺dDCs) (11), these cells in the gingiva did not express Langerin (CD11c⁺Ep-CAM⁻CD103⁺Langerin^neg), thus representing a not yet recognized gingival DC subset (Fig. 2C). To evaluate the capability of gingival-derived DCs to migrate to the LNs under inflammatory conditions, we painted the gingiva with FITC/DBP solution. As demonstrated in Fig. 2D, exposure of the gingiva to the DBP irritant resulted in migration of CD11c⁺FITC⁺ cells to the cervical LN within 48 h. Collectively, these results suggest that murine gingiva contain at least three DC subsets and that LCs correspond to a minor epithelial DC population. In addition, local inflammation facilitates rapid migration of gingival DCs to the cervical draining lymph nodes (LNs) for subsequent activation of T cells.

Fig. 2.

Characterization of DC subsets in the bone-surrounding gingiva. Gingival and ear skin tissues were obtained from B6 mice and enzymatically digested to obtain single-cell suspension. The cells were treated by surface and intracellular staining using various panels of antibodies and analyzed by flow cytometry. (A) Assessment of Ep-CAM and Langerin expression on CD11c⁺ cells. (B) Analysis of MHC class II and Langerin expression on total gingival cell suspension. (C) Analysis was restricted to CD11c⁺ cells and the expression of CD103 versus Langerin was assessed. (D) Gingival tissues of B6 mice (n = 4) were painted with FITC/DBP solution, and 2 d later, the draining LNs were collected and pooled and CD11c⁺ cells were enriched and subjected to flow cytometric analysis. FACS plots are presented confirming the migration of gingiva-derived DCs (FITC⁺CD11c⁺) to the LNs. Results are representative of at least three independent experiments.

Lack of LCs Facilitates the Elicitation of Pathogen-Specific CD4⁺ T Cells.

We have demonstrated that the absence of LCs resulted in a massive infiltration of T cells into the bone surrounding gingiva. To better understand the role of LCs in T-cell activation, we first examined antigen-presenting capabilities of tissue-derived DCs. For this purpose, we have developed an ex vivo assay to evaluate T-cell activation following infection. Mice were infected once i.p. with P. gingivalis and used 4 wk later as a source of T cells in this assay. Another cohort of mice was infected orally as described in Material and Methods, and their draining cervical LNs were collected to isolate various DC subsets by flow cytometry. These purified DCs were cocultured for 72 h with pathogen-specific CD4⁺ or CD8⁺ T cells, and secretion of IFN-γ was analyzed as a readout of DC-mediated T-cell activation. Secretion of IFN-γ in this assay was P. gingivalis-specific, as demonstrated in our validation assay (Fig. S3). Because the oral infection regimen consists of three constitutive inoculations with the pathogen, we analyzed antigen-presenting activity following the first and last inoculations. Fig. 3A illustrates our gating strategy for purifying each DC subsets. Tissue-derived DCs were segregated from LN-resident DCs (LN-DCs) based on the lack of CD8 expression, further divided into CD103⁺ and negative DCs, and the latter were further separated into iDCs and LCs according to Ep-CAM expression (Fig. 3A). Our analysis revealed that following the first infection, orally-derived DCs differ in their ability to stimulate CD4⁺ or CD8⁺ T cells (Fig. 3B). Whereas iDCs were the only subset presenting antigen to CD4⁺ T cells, presentation to CD8⁺ T cells was mediated by iDCs and LCs. Limited presentation was detected by CD103⁺ DCs. Of note, we detected activation of CD4⁺ and CD8⁺ T cells by LN-resident DCs. We next analyzed antigen presentation following the third inoculation and observed only negligible T-cell activation, suggesting that T-cell priming was already completed at this stage (Fig. S4).

Fig. 3.

Role of LCs in T-cell activation. B6 mice were infected orally with P. gingivalis either once (n = 15) or three times (n = 15). Three days after the infection, LNs were collected, pooled, and enriched for CD11c⁺ cells. The cells were stained with antibodies against: CD11c, CD8α, CD103, and Ep-CAM, for further separation by flow cytometry. (A) Flow cytometric plots demonstrating our strategy to identify various DC subsets from the draining LNs. (B) DC subsets purified after a single infection were incubated with pathogen-specific CD8⁺ or CD4⁺ T cells for 72 h. Secretion of IFN-γ to the supernatant, as an indication for T-cell activation by the DCs, were measured by ELISA. Data represent the means ± SD for each group. (C) Langerin-DTR mice, either LC-depleted or not, were infected orally once with P. gingivalis. Three days later, the draining LNs were collected and CD11c⁺ cells were purified by FACSAria for the assay. CD11c⁺ cells from various experimental groups were incubated with purified pathogen-specific CD8⁺ or CD4⁺ T cells for 72 h. Secretion of IFN-γ to the supernatant was measured by ELISA to quantify the level of T-cell activation. Data represent the means of five mice per group ± SEM. One representative out of two independent experiments is depicted.

Next, we tested the impact of LC ablation on T-cell priming. We used the ex vivo assay described above to measure T-cell activation after elimination of LCs in orally infected mice. Following the first inoculation, CD11c⁺ DCs were FACS-sorted and cocultured with pathogen-specific CD4⁺ or CD8⁺ T cells. The absence of LCs resulted in significantly increased secretion of IFN-γ by pathogen-specific CD4⁺ T cells, indicating an elevated activation of these cells by migratory DCs (P < 0.05) (Fig. 3C). In contrast, the lack of LCs had no effect on CD8⁺ T-cell activation. This suggests that LCs are essential for downregulating CD4⁺ T-cell responses.

LC Ablation Increased IFN-γ but Had No Impact on IL-17 or IL-10 Production.

Our data demonstrated that P. gingivalis-infected mice developed more vigorous T and B-cell responses attributable to the lack of LCs. Therefore, we sought to determine whether the nature of these elevated adaptive immune responses was altered in the absence of LCs. To this aim, we collected sera 6 wk after infection and examined the elicitation of pathogen-specific antibodies. High IgG titers were measured in P. gingivalis-infected groups with a modest, but statistically significant, increase in IgG levels in LC-ablated mice (Fig. 4A). Further analysis revealed that this increase was mainly attributable to an elevation in the titers of IgG2c (P < 0.005 at 1:10⁴ dilution) and not IgG1 (P = 0.1 at 1:10⁴ dilution) (Fig. 4A). We also measured the levels of IgA in the saliva, the port of entry of P. gingivalis in our model. As demonstrated in Fig. S5, the levels of IgA were not affected by the absence of LCs. We then analyzed cytokine secretion by splenocytes upon in vitro exposure to a recombinant P. gingivalis antigen (RgpA-Ad) (12). A higher production of IFN-γ was found by splenocytes derived from infected mice that lack LCs compared with LC-competent infected animals (Fig. 4B). Analysis of IL-10 secretion by splenocytes demonstrated that the addition of RgpA-Ad induced a similar production of IL-10 by all of the groups compared with unstimulated splenocytes (Fig. 4C). Finally, we observed no significant differences in the levels of IL-17A secretion in the absence or presence of LCs (Fig. 4D). In concordance, neutrophil infiltration into the gingiva was not affected by the lack of LCs (Fig. S6). Taken together, our analysis suggests that LC ablation augmented the production of IFN-γ in infected mice, whereas no significant impact was found on the secretion of IL-10 and IL-17 cytokines.

Fig. 4.

Production of higher IFN-γ levels and IgG2c antibodies due to the lack of LCS. (A) Six weeks after infection, sera or saliva from Langerin-DTR mice, either LC-depleted or not, were collected and analyzed for pathogen-specific IgG, IgG1, and IgG2c antibodies. Data represent the means of eight mice per group ± SEM. **P < 0.005 and *P < 0.05 compared with infected mice with no DT treatment. (B–D) Three weeks after infection, splenocytes were prepared and incubated with a recombinant adhesive fragment of the RgpA protein (RgpA-Ad) (1 μg/well) for 72 h. Levels of IFN-γ (B), IL-10 (C), and IL-17 (D) in the supernatants were measured by ELISA. Data represent the means of five mice per group ± SEM. **P < 0.005 compared with infected mice with no DT treatment; *P < 0.05 compared with mice receiving vehicle only; ^#P < 0.05 compared with infected mice with DT treatment. Results are representative of at three independent experiments.

Discussion

We demonstrated in this study the critical role of LCs in controlling inflammatory-driven alveolar bone loss following P. gingivalis infection. In the absence of LCs, infected mice developed reduced Treg cell numbers, produced elevated levels of IFN-γ, and activated high numbers of CD4⁺ T cells. Furthermore, considerable numbers of RANKL-expressing cells were found in the bone-surrounding gingiva, in particular CD4⁺ T cells, which resulted in enhanced destruction of alveolar bone. Our results are in line with the well-established role of RANK-RANKL interactions, CD4⁺ T cells, and IFN-γ during experimental periodontitis, the model used in the present study (9, 13–15).

Recent in vitro studies have questioned the role of IFN-γ in inflammation-driven bone loss, because IFN-γ was shown to suppress osteoclastogenesis by inhibiting RANK-RANKL signaling (16, 17). Nevertheless, the bone-destructive function of IFN-γ in vivo seems to overcome these in vitro results (18–21). With regard to experimental periodontitis, IFN-γ was shown to increase the number of RANKL-expressing CD4⁺ T cells and alveolar bone loss in vivo (21). Furthermore, another in vitro study has shown that IFN-γ-producing T cells were capable of inducing monocyte differentiation into osteoclasts by RANK-RANKL signaling, whereas non-IFN-γ producers failed to do so (22). This finding is in line with the vast presence of RANKL-expressing T cells in the gingiva of LC-depleted mice. Th17 cells were also proposed to play a protective role in P. gingivalis-induced alveolar bone loss, by recruiting neutrophils (23). Nevertheless, we observed no impact on IL-17 production in LC-ablated mice, nor on the frequencies of neutrophils in the infected gingiva. It is, thus, likely that the protective role of LCs in our experimental system does not involve Th17 cells. We also detected an increased infiltration of RANKL-expressing NKT and γδT cells in the lack of LCs. These T-cell subsets have been found to be elevated in human periodontitis (24, 25), suggesting that NKT and γδT cells might contribute to bone destruction as a source of RANKL.

Besides T cells, our data identified large numbers of B cells in the bone-surrounding gingiva of infected mice, which further increased by the lack of LCs. Our findings are in concurrence with a recent report demonstrating that B cells outnumber T cells in periodontal lesions (26). In agreement with previous observations regarding the phenotype of B cells in the inflamed gingiva, ∼30% of B cells accumulating in the gingiva of LC-ablated mice expressed RANKL (27). Still, the importance of these B cells in mediating alveolar bone loss is uncertain, as a considerable number of B cells was also found in mice without DT treatment, where no bone loss was observed. It was also demonstrated that B cells are not essential for LPS-induced bone loss, whereas T cells mediated this process (28). We, thus, argue that B cells might contribute to bone loss but are not crucial for this process. T cells, on the other hand, seem to play a critical role in alveolar bone loss, because their number increased only in LC-depleted mice, in which bone resorption was detected. Interestingly, the absence of LCs resulted in higher production of IgG2c antibodies. This is in concurrence with recent work demonstrating higher IgG2c levels in LC-deficient mice following DNA vaccination (29). Of note, in this study, Ln⁺dDCs were proposed to augment IFN-γ production by plasmid DNA-elicited T cells. Although the gingival equivalent of this DC subset was not clearly identified in our analysis, it is intriguing to speculate whether these cells present in extremely low numbers in the gingiva and can impact T-cell function.

Our data suggest that the absence of LCs skewed, to some extent, P. gingivalis-specific immunity toward a Th1-type response. This observation is in agreement with previous in vivo studies involving P. gingivalis-mediated bone loss (13, 30). The capability of LCs to negatively regulate Th1-type pathogen-specific immune responses was recently demonstrated during Leishmania infection (31). Similar to our study, the lack of LCs during infection with Leishmania reduced the generation of Treg cells at the site of inflammation. Indeed, Treg cells were recently reported to attenuate alveolar bone loss (32). Depletion of LCs was also found to dampen cell-mediated immunity following mucosal immunization (33). These observations are consistent with a recent study proposing that LCs are precommitted to induce tolerance (34). Of note, LC ablation had no impact on P. gingivalis clearance (Fig. S7), unlike Leishmania infection, where LC-dependent Th1 response was important for pathogen clearance (31). This suggests that the quality of the inflammatory reaction, rather than bacterial load, influences mostly bone loss, further supporting the concept of LCs as immune regulators of adaptive immunity.

The subsets of DCs mediating antigen presentation to CD4⁺ and CD8⁺ T cells following P. gingivalis infection seems to be quite different. Whereas iDCs play the major role in presenting antigens to both T-cell subsets, presentation to CD8⁺ T cells involved also LCs, LN-DCs, and CD103⁺ DCs. It is surprising that LCs present antigen to CD8⁺ but not to CD4⁺ T cells because LCs reside close to the bacterial plaque, engulf the pathogens, and should be able to present bacterial peptides on MHC class II to CD4⁺ T cells. However, such presentation does not occur. A possible explanation could be that P. gingivalis membrane vesicles have the capability to inhibit IFN-γ-induced MHC class II expression (35). On the other hand, CD8⁺ T-cell activation may depend on cross-presentation, a function previously reported for LCs (36), as well as LN-DCs (CD8⁺ DCs) and CD103⁺ DCs, thus explaining CD8⁺ T-cell activation by these DC subsets in our system (37, 38). This conclusion is further supported by our inability to detect the pathogen in the draining LNs, suggesting that LN-DCs obtained pathogen-derived antigen from incoming LCs/DCs, as has been demonstrated for certain viral skin infections (39). Of note, LC ablation had no impact on the level of antigen presentation to CD8⁺ T cells; nevertheless, an increased number of these cells was observed in the gingiva of LC-depleted mice. This could be explained by the critical role of CD4-help in facilitating CD8⁺ T-cell responses, as well as B-cell activation (40). Thus, inhibition of CD4⁺ T cells by LCs leads to reduced local inflammation and minimal bone destruction.

The accumulation of CD4⁺ T cells in the gingiva for long periods of time might influence resident DCs. Activated CD4⁺ T cells express RANKL and CD40-ligand (CD40L), molecules with a potential to modulate immunity (41). Using transgenic mice that overexpress CD40L or RANKL in keratinocytes, it has been shown that RANKL suppressed autoimmunity induced by CD40L (42). Interestingly, in this system the numbers of epidermal LCs were greatly reduced by CD40L (43), suggesting that CD40L might regulate LC numbers, as was recently suggested also for RANKL (44). Oral DCs have the capability to express RANK (Fig. S8) and, thus, might directly interact with RANKL-expressing CD4⁺ T cells. Although the impact of infiltrating CD4⁺ T cells on gingival LCs is unknown, the well-documented reduction in LC numbers during human periodontitis could be a result of an imbalance in CD40L/RANKL expression by local CD4⁺ T cells (45–47). Gingival DCs might also regulate the function of T cells accumulating in the tissue. The gingiva was proposed to act as a tertiary lymphoid site during periodontitis, where local DCs can activate T cells (48). Such peripheral activation of T cells was shown in the skin as well (49) and could enhance gingival inflammation and alveolar bone loss.

In summary, this study reveals a protective role for LCs in inflammation-driven bone loss initiated by bacterial infection. Through increasing Treg numbers, LCs are proposed to inhibit IFN-γ production and reduce RANKL⁺ CD4⁺ T cells in the inflamed tissue. As a result, bone homeostasis is not disturbed by the local inflammation, and alveolar bone remains intact. Additional experiments involving blocking of RANK-RANKL interactions in the absence of LCs might further enforce the proposed mechanism by which LCs impact bone loss. Beyond increasing our understating of the role of LCs in inflammation-induced bone loss, our data indicate that LCs are promising candidates for future approaches attempting to prevent inflammatory bone diseases without compromising protective antibacterial immunity.

Materials and Methods

Antibodies and Reagents.

Antibodies and reagents are described in SI Materials and Methods.

Mice.

Six- to 12-wk-old knock-in mice expressing human DTR under transcriptional control of the endogenous Langerin/CD207 promoter (Langerin-DTR) were bred in our facility and maintained under specific pathogen-free (SPF) conditions (7). Langerin-DTR mice allow conditional ablation of Langerin-expressing cells in vivo by the administration of DT. The identity of the mice used for experiments was confirmed by genotyping with the following PCR primers: forward, GCCACCATGAAGCTGCTGCCG; and reverse, ATAGTTTA GCGGCCGCTTTACTTGTACAG. C57BL/6 (B6) mice were purchased from Harlan and used at the age of 6–12 wk. Research on mice was approved by the Hebrew University Institutional Animal Care and Ethic Committee.

Ablation of Langerin-Expressing Cells in Vivo.

Langerin-DTR mice were treated i.p. with 1 μg DT (Sigma-Aldrich) in 100 μL of PBS 7 and 5 d before the infection with P. gingivalis. Similar administration of DT into wild-type B6 mice did not influence the examined immune responses or bone properties.

Inflammation-Induced Bone Loss Model.

The details of P. gingivalis infection and μCT analysis of the alveolar bone were performed as described previously (50) (SI Materials and Methods).

FITC Painting.

Painting of the gingiva is described in SI Materials and Methods.

Antigen Presentation Assays.

Antigen presentation assay is described in SI Materials and Methods.

Isolation of Lymphocytes and DCs.

Gingival and skin cells were isolated as explained in SI Materials and Methods.

Cytokine Secretion by Splenocytes.

Mice were infected orally with P. gingivalis as described above, and 2–3 wk later, the spleens were collected from each mouse, and single-cell cultures were prepared. The samples were cultured in complete RPMI medium 1640 (4 × 10⁶ cells/well) in a 96-well plate. The RgpA-Ad antigen was added to the cultures (1 μg/well), the plates were incubated for 72 h, and supernatants were collected. The level of IFN-γ, IL-17, and IL-10 in the supernatants was measured using an ELISA MAX mouse kits (BioLegend) according to the instructions of the manufacturer. Cytokine levels were determined using standard curves of recombinant cytokines and are expressed as picograms per milliliter.

Serum Analysis.

Antibody responses were assessed as described in SI Materials and Methods.

Statistical Analysis.

Data were expressed as means ± SEM. Statistical tests were performed using one-way analysis of variance (ANOVA) and Student t test. P < 0.05 was considered significant.