IL-1β Induces Pathologically Activated Osteoclasts Bearing Extremely High Levels of Resorbing Activity: A Possible Pathological Subpopulation of Osteoclasts, Accompanied by Suppressed Expression of Kindlin-3 and Talin-1

Osteoclasts are bone-resorbing cells derived from hematopoietic stem cells. These cells not only play a central role in bone remodeling by secreting “clastokines” (1) but also play important roles in the induction of angiogenesis and regulation of hematopoiesis (1). The proper balance between osteoclast-mediated bone resorption and osteoblast-mediated bone formation is critical for keeping normal bone metabolism. Marked augmentation in enhanced osteoclast recruitment occurs in the inflammatory environments associated with rheumatoid arthritis or periodontal disease, resulting in the severe pathological bone destruction (2–4). At sites of inflammation-induced pathological bone destruction, it is considered that osteoclasts could modify their quality in bone-resorbing activity to be in the pathologically activated state. Toh et al. (5) have reported on an extremely high expression of CSF-1 receptor in osteoclasts performing pathological bone destruction in rheumatoid arthritis. In respect to osteoclast precursors, cells expressing membrane Ag Ly6C present in bone marrow tissues differentiate into osteoclasts responsible for pathological bone destruction (6). It has also been reported that monocytes and dendritic cells express membrane surface molecules different from normal cells under inflammatory circumstances (7). Therefore, it is possible that osteoclasts causing pathological bone destruction constitute some unique subpopulation distinct from that of normal osteoclasts performing normal bone remodeling.

In the terminal differentiation of normal osteoclasts, multinucleated osteoclasts recognize bone matrix proteins containing the RGD sequence of bone matrix protein such as osteopontin via integrin αvβ3 to adhere to bone surface (8). Osteoclasts are then markedly polarized to form ruffled borders at the bone-contacting side (apical side). In this terminal differentiation step (also called “activation”) of normal osteoclasts, Schmidt et al. (9) found that Kindlin-3–mediated activation of integrin β3 is essential for osteoclasts expressing bone-resorbing activity. Kindlin-3, a cytoskeleton regulatory protein, binds to cytoplasmic tail of the integrin β3 and activates this integrin in cooperation with Talin-1, which also binds to integrin β3. In addition, activation of integrin αvβ3 through binding to its ligand is known to lead to the phosphorylation and eventual activation of DAP12, a membrane surface molecule essential for osteoclast differentiation (10, 11).

Meanwhile, lines of evidence have shown that IL-1 plays an important role in bone destruction associated with arthritis (1, 2, 4). Two types of IL-1 receptors, the stimulatory receptor (IL-1R1, type 1) and the suppressive receptor (IL-1R2, type 2), are known. We have previously elucidated the expression of both types of receptors in osteoclasts in vivo. IL-1 receptor type 1 and type 2 are expressed in a similar level in osteoclasts present in the normal healthy bone; in contrast, IL-1 receptor type 1 (stimulatory receptors) is preferentially expressed in pathologically activated osteoclasts (PAOCs) performing severe bone destruction with markedly suppressed expression level of IL-1 receptor type 2 (inhibitory receptors) (12).

We have developed the pH-sensitive fluorescence probe Rh-PM (Acidi-Fluor ORANGE-NHS) to detect cancer cells and to explore proton (H⁺) secretion of several cell types (13). Rh-PM emits fluorescence only when this probe is exposed to acid environments, so that imaging of H⁺-secreting cells involving osteoclasts can be possible.

Here we established experimental conditions to form PAOCs by use of bone resorption system in combination with the application of fluorescence probes. We also searched for membrane molecules selectively expressed in PAOCs, which would realize the specific regulation of pathological bone destruction through regulating the identified molecule expressed on PAOCs. We have found that IL-1β–induced osteoclasts showed extremely high H⁺ secretion and resorbing activity. We have also detected marked difference in the regulation of adhesion properties in IL-1β–induced osteoclasts as well as in the pattern of surface protein expression in PAOCs.

Male ddY mice were purchased from Kyudo (Tosu, Japan). All animal experiments were performed according to the guideline for Care and Use of Animals of Kyushu University.

Recombinant human M-CSF and recombinant human soluble receptor activator of NF-κB (RANK) ligand (RANKL) were obtained from PeproTech (Rocky Hill, NJ). Human IL-1α, human IL-1β, human IL-6, human IL-8, human IL-17, and human TNF-α were purchased from Miltenyi Biotec (Bergisch Gladbach, Germany). MEM α (α-MEM), serum (FBS), penicillin streptomycin glutamine (100×), and TrypLE Select (10×) were obtained from Life Technologies (Grand Island, NY).

The femur and tibia were aseptically removed from 5-wk-old male ddY mice, and were cut to remove epiphyses. Using a syringe with a 26 gauge needle, serum-free α-MEM was injected into the bone samples from the end of the metaphysis and then bone marrow tissue was flushed out and bone marrow cells were harvested. Bone marrow cells were cultured in the presence of M-CSF (10 ng/ml) in α-MEM containing 10% FBS (Life Technologies, Invitrogen). After incubation in a CO₂ incubator (37°C, 95% air, 5% CO₂) for 24 h, nonadherent cells were collected and cultured in the presence of M-CSF (20 ng/ml) in a 100-mm Falcon tissue culture dish for 3 d to obtain bone marrow macrophages (BMMs). BMMs were detached from the culture dish with TrypLE Select. Cells (1.5 × 10⁵ cells/ml, 2 ml) were cultured in the presence of M-CSF (20 ng/ml) and RANKL (50 ng/ml) with various inflammatory cytokines in 35-mm Falcon tissue culture dishes to form osteoclasts. To determine optimal concentrations of various inflammatory cytokines, BMMs were seeded in a 96-well plate, cultured for 3 d, and then subjected to tartrate-resistant acid phosphatase (TRAP) staining. TRAP-positive cells with three or more nuclei were counted as the number of osteoclasts, and the results are presented as mean ± SEM. Student t tests were performed, and p < 0.05 was considered to be significant.

PBMCs were obtained from healthy volunteers according to the protocol approved by Kyushu University Institutional Review Board for Clinical Research (Protocol number 29-149). Briefly, peripheral blood was collected from the cutaneous vein of the forearm. Blood was immediately mixed with 2 volumes of saline (0.9% NaCl) containing 2 mM EDTA. This diluted blood (6 ml) was gently layered on 3 ml of Lymphoprep (GE Healthcare) in 15 ml conical centrifuge tube and then centrifuged at 800 × g for 20 min at room temperature with no brake. Mononuclear cells present in the white band formed between Lymphoprep and saline were collected by Pasteur pipette. After washing in PBS two times, cells were suspended in α-MEM containing 15% FBS. Aliquots were taken and diluted in 5% acetic acid and counted by use of hemocytometer. Cells were cultured in α-MEM containing 15% FBS, 25 ng/ml M-CSF, and 50 ng/ml RANKL in 96-multiwell culture plates (2 × 10⁶ cells/ml or 1 × 10⁶ cells/ml, 150 μl per well). Whole culture medium was changed every 3–4 d of culture. Cells were fixed and stained for TRAP at around 2 wk of culture. TRAP-positive cells bearing more than three nuclei were counted.

Small columnar dentin pieces were purchased from Asashima-Ryubundo (Fukuoka, Japan). One end of the columnar dentin (Asashima Ryubun-do, Fukuoka, Japan) was embedded and fixed in resin for dental impression tray (Shofu, Kyoto, Japan), and then cut with a hard-tissue slicer (Ernst Leitz, Wetzlar, Germany) to yield 1-mm thick dentin slices with 6 mm diameter; the surface of the dentin was then polished and stored in 70% ethanol till use.

Nonadherent bone marrow cells obtained from mouse tibia and femur were cultured in the presence of M-CSF for 3 d and then detached using TrypLE Select. Cells were further stimulated with RANKL and M-CSF as described above. After 48 h, cells present on the 35-mm tissue culture dish were collected using TrypLE Select and 6 μl were seeded on each dentin slice. After incubation for 30 min at room temperature, α-MEM containing 10% FBS was added, and cells were further cultured in α-MEM containing 10% FBS involving no additional factors for 5 d. Osteoclasts present on dentin slices were stained for TRAP and observed with Keyence Biozero microscope to confirm the number of osteoclasts. Osteoclasts were completely removed from the dentin slice in an ultrasonic bath. After hematoxylin staining, resorption cavities were observed with Keyence Biozero microscope. Thereafter, the surface of dentin slices was observed with scanning electron microscopy (Hitachi S-3400N). The number of osteoclasts on the dentin slice and resorption area were analyzed by Mann–Whitney U test, and p < 0.05 was considered to be significant.

Resorption cavities were stained with hematoxylin and the total area of dentin resorption was measured using the BZ-II image analysis application of Keyence Biozero microscope. A dentin slice was placed on a glass slide and imaged at regular intervals. From images obtained, a single image was constructed with the image joint function, and the total area formed on each dentin slice was estimated using the protocol of the software BZ-II image analysis. The dynamic cell counting function was then used, and areas intensely stained with hematoxylin were selected using the color extraction function. After refining the area selection using the small particle removing function, the count run button was selected to measure the area. The size measurement button was pushed to display the area measurement, which was in pixel units. The calibration checkbox was checked to convert the unit of area to square micrometers.

Hematoxylin-stained dentin slices were washed in an ultrasonic bath for 2 min, and air-dried in a clean bench. Four dentin slices were carefully placed on a scanning electron microscopy aluminum specimen holder (attached with M4 screws of Φ15 × 5 mm; Okenshoji, Tokyo, Japan) using double-sided adhesive tape. Images of the scanning microscopy were taken at 15.0 V.

BSA (Calbiochem) was dissolved in 0.2 M phosphate buffer (pH 8.5) to be the final concentration 1 mg/ml. One milliliter of the BSA solution was mixed with 3 or 8 μl of the Rh-PM-NHS (10 mM in DMSO), and then incubated at room temperature for 1 h. BSA conjugated with Rh-PM-NHS was purified by use of PD-10 column. Specific activity was determined by measuring the UV absorption (A280) and absorption of Rh-PM (A544).

Surface coating of dentin slices with BSA conjugated with Rh-PM (BSA-Rh-PM) was performed as follows. BSA-Rh-PM (20–30 μl per slice) was put on the air-dried dentin slice in a clean bench and left to stand for 2 h. A nuclear fluorescence labeling of osteoclasts by Hoechst 33342 was performed by adding 1/300 volume of Hoechst 33342 solution to the cell suspension prepared from the osteoclast cultures that had been induced to differentiate for 2 d in 35-mm dishes. The cells were harvested and seeded on a dentin slice coated with BSA-Rh-PM. After 24 h, the dentin slice was transferred into a glass-bottom dish, and nuclei of osteoclasts (stained with Hoechst 33342: blue fluorescence 408 nm) and H⁺ produced by osteoclasts (Rh-PM: red fluorescence 561 nm) were observed using a C2si confocal laser microscope (Nikon, Tokyo, Japan). Total nuclear volume and the total volume of the signal from the reaction between H⁺ and BSA-Rh-PM were measured as the parameters representing the number of cells and the activation of osteoclasts, respectively. All images were obtained and analyzed using NIS-Elements AR 4.00.06 software (Nikon).

Total RNA was extracted from osteoclasts derived from BMMs by use of ISOGEN (Nippon Gene, Tokyo, Japan), followed by synthesis of cDNA with ReverTra Ace qPCR RT Kit (Toyobo, Osaka, Japan). Real-time PCR reactions were performed using THUNDERBIRD SYBR qPCR Mix (Toyobo) with a LightCycler 96 instrument (Roche, Basel, Switzerland), by using specific primers for each gene (Supplemental Table I). mRNA levels were normalized to β-actin expression.

Western blotting was carried out by a semidry procedure. Osteoclasts were formed in 100-mm Falcon tissue culture dishes as described above in the presence or absence of IL-1β. After washing the cells once with PBS, cells were lysed in RIPA buffer (150 mM NaCl, 1% v/v NP-40, 0.5% DOC, 50 mM Tris-HCl (pH 8), 0.1% SDS) containing 1/100 volume of Sigma protease inhibitor mixture P1860. Cell lysates were collected using cell scrapers and collected material was placed in a 1.5-ml microtube and agitated on ice with a shaker for 30 min at 4°C. After centrifugation of the cell lysates for 15 min at 14,000 rpm at 4°C, supernatants were collected. Cell lysates were subjected to SDS-PAGE using 7.5% gel and transferred to polyvinylidene difluoride blotting membrane (GH Healthcare). The membrane was blocked with 0.3% skim milk for 1 h to prevent nonspecific binding. The membrane was incubated in 5% skim milk containing anti–Kindlin-3 Ab (×500 dilution; Cell Signaling Technology) or anti–Talin-1 Ab (Cell Signaling Technology) as the primary Ab for 48 h in a 4°C environment. After washing with TBS-Tween (0.1%) (10 min × 3 times), the membrane was treated with HRP-conjugated anti-rabbit IgG Ab (Cell Signaling Technology) as the secondary Ab for 1 h and then the membrane was treated with the luminogenic reagent Luminata Crescendo Western HRP Substrate (Merck Millipore) for 5 min, and imaged with ImageQuant Las4000 (GH Healthcare).

Biotinylation of membrane surface proteins was performed by using ECL Protein Biotinylation Module (GH Healthcare). After forming osteoclasts in 100-mm dishes under conditions of M-CSF+RANKL or M-CSF+RANKL+IL-1β, cells were washed twice with PBS and biotinylated as follows. After removal of PBS, 2 ml of 40 mM bicarbonate buffer (pH 8.6) was added to each 100-mm dish. Eighty microliters of a biotinylation reagent (Biotinylation reagent in the module) was added followed by gentle orbital shaking at 4°C for 30 min to the biotinylated cell surface proteins. After removing the biotinylation buffer, cells were rinsed in PBS two times and then lysed in 1 ml of the lysis buffer (0.5 M EDTA, 50% NP-40, 50% glycerol, 100 mM sodium vanadate, 200 mM β-glycero-phosphate, 100 mM PMSF, 100 mM DTT, 100 mM sodium pyrophosphate, 0.2 M NaF, 0.5 M phosphate buffer pH 7.4, 0.01 volume of Sigma protease inhibitor mixture P1860). Cells were mildly shaken with the lysis buffer using an orbital shaker in a cold room at 4°C for 20 min to extract the cell surface proteins. Cell lysates were collected in a 1.5-ml tube using a scraper and centrifuged at 12,000 rpm for 10 min at 4°C and the supernatants were collected. After adding 5× sample buffer and 2-ME, proteins recovered in the supernatant were subjected to SDS-PAGE using 10% gel. After transferring proteins to the membrane, biotinylated cell surface proteins were detected by anti-avidin Ab conjugated with HRP.

Student t test and Mann–Whitney U test were used for statistical analyses, and p < 0.05 was considered to be statistically significant.

Effects of various inflammatory cytokines on osteoclastogenesis were examined. We hypothesized that osteoclastogenesis under the pathological environment yields osteoclasts that cause pathological bone destruction. We also assumed that various inflammatory cytokines could create such pathological environments. To know the optimal condition of each cytokine on pathological osteoclastogenesis, various concentrations of cytokines were added to the in vitro system for evaluating osteoclastogenesis, in which M-CSF and RANKL were used to induce osteoclasts, and effects of each inflammatory cytokines on osteoclastogenesis were examined. The highest number of osteoclasts was formed when 100 ng/ml of IL-1α, 0.5 ng/ml of IL-1β, 50 ng/ml of IL-6, 10 ng/ml of IL-8, 1 ng/ml of IL-17, or 50 ng/ml of TNF-α was added, respectively (Fig. 1). When osteoclastogenesis was carried out with a mixture of cytokines of the respective optimal concentrations indicated, the number of osteoclasts formed was in a low level in comparison with control cultures with no inflammatory cytokines (data not shown). Therefore, concentrations of cytokines that gave the maximal number of osteoclasts when each cytokine was used alone were chosen as the candidate conditions of the pathological environment. Based on the results shown in Fig. 1, we have fixed the concentration of each inflammatory cytokine as: IL-1α, 100 ng/ml; IL-1β, 0.5 ng/ml; IL-6, 50 ng/ml; IL-8, 10 ng/ml; IL-17, 1 ng/ml; and TNF-α, 50 ng/ml.

Using BMMs, different concentrations (final concentrations) of the factors were added to the osteoclastogenesis system (M-CSF+RANKL). The cells cultured for 3 d were reseeded on dentin slices followed by culture in α-MEM containing 10% FBS in the absence of cytokines. Fig. 2A shows the number of TRAP-positive osteoclasts observed on dentin slices after 5 d of culture. IL-1α and IL-1β showed a significantly high number of osteoclasts on dentin slices (Fig. 2A). Concerning the resorption area, a marked resorption activity was observed with cultures treated with IL-1β (Fig. 2B). Significant resorption activity was also observed in cultures treated with IL-1α and IL-8. Fig. 2C shows the data demonstrating the resorption area per one osteoclast, indicating that osteoclasts formed under the stimulation with IL-1α, IL-1β, IL-8, IL-17, or TNF-α significantly increased the resorbing activity per one osteoclast compared with the control culture (M-CSF+RANKL). Among them, IL-1β showed the highest activity. IL-1β–stimulated osteoclasts showed the highest values in all data: the number of osteoclasts observed on dentin slice, total resorption area formed on dentin slice, and resorption area per one osteoclast. Therefore, we have fixed the culture condition giving rise to the most activated type of osteoclasts as “M-CSF+RANKL+IL-1β” in α-MEM containing 10% FBS.

To know the bioactivity of IL-1β on human osteoclastogenesis, we have examined the effect of IL-1β on osteoclast formation from human PBMCs as shown in Supplemental Fig. 1. Osteoclast formation tended to be stimulated slightly by the low concentration of IL-1β in some PBMC donors, which was similar to the murine system, whereas osteoclastogenesis was markedly inhibited in other donors. IL-1β response on human osteoclastogenesis seems to be different among the human population.

The dentin slices used for Fig. 2 were observed with scanning electron microscopy to confirm resorption lacunae formed on dentin slices. Compared to normal osteoclasts induced in normal conditions (M-CSF+RANKL), osteoclasts induced in the presence of IL-1β (M-CSF+RANKL+IL-1β) formed markedly high numbers of resorption lacunae with typical morphology on dentin slices (Fig. 3).

To examine whether IL-1β directly stimulates osteoclasts to resorb dentin, we have added IL-1β after reseeding osteoclasts on dentin slices. When IL-1β (final concentration 0.5 ng/ml) was added to osteoclasts adhered to the dentin surface, no significant promoting effect on dentin resorption was observed when compared with that of normal osteoclasts (induced by M-CSF+RANKL) (Fig. 4A). We next examined the resorbing activity of osteoclasts induced in the presence of IL-1β. Although additional factors were not added during resorption on dentin slices, an extremely high level of resorption activity (∼10-fold) was observed in osteoclasts formed in the presence of IL-1β in comparison with normal osteoclasts formed in the presence of M-CSF+RANKL (Fig. 4B). These results clearly demonstrate that osteoclasts with extremely high bone-resorbing activity were formed by the exposure to the inflammatory cytokine IL-1β during osteoclastogenesis.

Osteoclasts formed in the presence or absence of IL-1β were subjected to fluorescence labeling of nuclei with Hoechst 33342 and then seeded onto dentin slices coated with BSA-Rh-PM. After 24 h, nuclei (blue fluorescence 408 nm) and produced H⁺ (red fluorescence 561 nm) were observed using confocal laser microscope followed by reconstructing the three-dimensional images. Abundant signal of red fluorescence originated from the reaction of BSA-Rh-PM with H⁺ was observed for osteoclasts induced with IL-1β, indicating a significantly increased level of H⁺ production in IL-1β–induced osteoclasts compared with that of normal osteoclasts (Fig. 5). The total volume of nuclei was measured as a parameter for estimating the cell number. The total volume of the signal (red fluorescence) from the reaction between H⁺ and BSA-Rh-PM was measured as a parameter of osteoclast activity. Fig. 6 shows a comparison of the H⁺ production level between IL-1β–induced osteoclasts and normal osteoclasts. IL-1β–induced osteoclasts produced a dramatically higher level of H⁺ compared with osteoclasts formed in the absence of IL-1β (normal osteoclasts) (Fig. 6A). IL-1β treatment showed no influence on the total nuclear volume, indicating the number of osteoclasts seeded on dentin slices was not significantly different in both experimental groups (in the presence or absence of IL-1β) (Fig. 6B). As shown in Fig. 6C, red fluorescence signal reflecting H⁺ production per one osteoclast was increased by IL-1β treatment to approximately 3-fold. These data clearly demonstrate that IL-1β–induced osteoclasts bear high potential to secrete H⁺.

To confirm the position of osteoclasts in the resorption experiments of Fig. 5, cell surface lipids were further stained with DiO, which emits green fluorescence. Dentin slices were coated with BSA-Rh-PM as described in 2Materials and Methods so that red fluorescence is emitted when Rh-PM reacts with H⁺ produced by osteoclasts. After staining of nuclei with Hoechst 33342 and cell membrane with DiO, H⁺ production of osteoclasts induced with or without IL-1β was observed by use of confocal laser microscopy (Fig. 7). In control osteoclasts untreated with IL-1β, secreted H⁺ localized at the marginal ring-like portion of the osteoclast (Fig. 7A). In contrast, in IL-1β–induced osteoclasts, an extremely strong signal was observed and the red fluorescence reflecting secretion of H⁺ tended to localize to the corresponding portions of each nuclei (Fig. 7A). Three-dimensional observation showed that the positive area was the surface of dentin just beneath the portion of each nuclei. Fig. 7B is the tomography image (horizontal sections of various heights) of the data in Fig. 7A and a schematic diagram of this analysis is provided in Fig. 7C. These data demonstrate a marked difference in the localization pattern of secreted H⁺ between normal osteoclasts and IL-1β–induced osteoclasts, PAOCs.

Osteoclastic bone resorption is dependent on integrin, mainly on integrin αvβ3. To know whether proton secretion of PAOCs is also dependent on integrin, we have added RGDS peptide, which inhibits integrin-mediated events, to cultures of osteoclasts on dentin slices, which were coated with Rh-PM-conjugated BSA. As shown in Supplemental Fig. 2, proton secretion of normal osteoclasts (osteoclasts formed in the presence of M-SCF and RANKL) was completely inhibited by the addition of RGDS peptide almost by 100%. Proton secretion of PAOCs (MNCs formed in the presence of M-CSF, RANKL, and IL-1β) was also suppressed by RGDS peptide by 89.8%. Such incomplete inhibition of proton secretion by PAOCs could reflect a difference in the nature of these osteoclasts in comparison with normal osteoclasts.

We next examined the effect of IL-1β treatment on the adhesion signaling present in osteoclasts. As Kindlin-3 is a key molecule in the terminal differentiation of osteoclasts (8), we have examined the expression of Kindlin-3 and its related molecule Talin-1 in normal osteoclasts and IL-1β–induced osteoclasts. As shown in Fig. 8A and 8B, normal osteoclasts express high levels of Kindlin-3. Normal osteoclasts also express Talin-1. In contrast, expression levels of Kindlin-3 and Talin-1 were quite low in osteoclasts formed in the presence of IL-1β. In addition, IL-1β–induced osteoclasts showed different morphology when compared with that of normal osteoclasts (Fig. 8C). IL-1β–induced osteoclasts at day 4 tended to bear an irregular shape with the higher thickness having high TRAP activity. Real-time PCR analysis demonstrated that the level of mRNA expression of integrin β3 in PAOCs was almost equal to that of normal osteoclasts (Supplemental Fig. 3). In these analyses, we could not detect any significant difference in the expression of osteoclast marker genes (Dc-stamp, Oc-stamp, Calcitonin receptor, Oscar, Csfr1, Atp6v0d2, Ctsk, Trap) in PAOCs. These data suggest a possibility that IL-1β–induced osteoclasts used the different adhesion signaling molecules from that of normal osteoclasts. These data could imply that IL-1β–induced osteoclasts might form a subpopulation distinct from the population of normal osteoclasts.

To detect membrane surface molecules induced by IL-1β, membrane surface proteins of osteoclasts or osteoclast precursors were labeled with biotin. After extracting the cell surface membrane proteins using the lysis buffer, proteins were subjected to Western blotting, and we detected whole biotinylated cell surface proteins with avidin conjugated with HRP. Fig. 9 shows the comparison of the expression pattern of the cell surface proteins between control osteoclasts (IL-1β −) and IL-1β–induced osteoclasts (IL-1β +). Although no difference of the expression pattern was detected in osteoclast precursors, IL-1β–induced protein band around 50–60 kDa was detected in mature osteoclasts (Fig. 9).

Inflammatory cytokines have been reported to be involved in enhanced osteoclast formation and activation in inflammatory diseases, such as periodontal disease, rheumatoid arthritis, and multiple sclerosis (2–4, 14). We have postulated that pathological osteoclasts can be formed in the unique pathological conditions. In this study, IL-1α, IL-1β, IL-6, IL-8, IL-17, and TNF-α were selected to reproduce the pathological environment in vitro (1, 14). To determine the condition to induce osteoclasts with the highest bone-resorbing activity, osteoclasts were cultured on dentin slices, and their resorption activity was examined. Osteoclast-involving cells formed under stimulation with various cytokines were seeded on dentin slices and resorbing activity was assessed in conditions containing no inflammatory cytokine. IL-1β showed the highest values in all assays, the number of osteoclasts formed on dentin, the total resorption area on dentin, and the resorption area per one osteoclast, among cytokines tested. Therefore, we have chosen M-CSF+RANKL+IL-1β for the in vitro culture condition to induce PAOCs. There are lines of background evidence relating to the highest activity of IL-1β in our current experiments. It has been reported that IL-1 stimulates osteoclastogenesis (15, 16). We have previously reported that IL-1 receptor type 1 (stimulatory receptor) is highly expressed in osteoclasts present in bone destruction sites in rats with adjuvant-induced arthritis (12). It has also been noted previously that IL-1α activates and prolongs survival of osteoclasts (17). In periodontal disease, IL-1 activation is considered to be involved in the inflammation spreading in the gingival connective tissue, occurrence of attachment loss, and alveolar bone destruction due to enhanced osteoclastogenesis (18). A lysine-specific gingipain produced by the periodontal pathogen Porphyromonas gingivalis has been known to be able to degrade cytokines and has been reported to cause pathological bone destruction by degrading osteoprotegerin, a decoy receptor of RANKL, but it does not degrade IL-1β, which results in the promotion of osteoclastogenesis (19). There are other reports including those showing that IL-1 activates proteolytic enzymes, such as collagenase and plasminogen, and is involved in the reduction of bone mass (20).

Osteoclasts express IL-1 receptors and their resorption activity is enhanced through the action of IL-1α and IL-1β in the presence of RANKL in vitro (21). Jimi et al. (22) have shown that IL-1 induces multinucleation and bone resorption activity of osteoclasts in the absence of osteoblasts and stromal cells. Moreover, IL-1 plays a crucial role in bone destruction in animal experimental models of rheumatoid arthritis (23, 24). These reports are likely to be related to our data demonstrating the highest values observed for all three parameters evaluated: total number of osteoclasts formed on dentin, total resorption area, and the area of resorption per one osteoclast.

In the current study, functional enhancement was not observed when IL-1β was added to normal osteoclasts only during resorption, strongly suggesting that IL-1β does not directly stimulate osteoclastic function. Extremely high bone-resorbing activity was observed when precursor cells were treated with IL-1β during osteoclast formation only before replating on dentin slices. These data could be related to the activation of ERK (15) and JNK (25) signaling by IL-1β treatment during osteoclast differentiation, so that the formed mature osteoclast population was shifted to express pathological resorption activity. Our current observation supports the hypothesis that PAOCs are formed under the abnormal pathological bone microenvironments (Fig. 10).

Our current study using the pH-sensitive fluorescence probe Rh-PM clearly demonstrates that osteoclasts induced with IL-1β secrete abundant H⁺ on dentin slices. Because IL-1 receptors have been reported to activate signaling molecule TRAF6 (26) synergistically with TNF receptors, H⁺ production is likely to have been elevated through TRAF6. Therefore, it is considered that IL-1β binds to IL-1 receptors expressed on osteoclasts and activates TRAF6 pathway that was already initiated by the action of RANKL-RANK signaling, which results in the further activation of NF-κB and NFATc1 and induction of marked osteoclast-specific genes. In addition, IL-1β has been reported to enhance cathepsin K release (15). Lines of evidence described in these reports could explain some mechanism for our current observations concerning the abundant H⁺ secretion by PAOCs formed in the presence of IL-1β. We have shown a marked increase in the acidity at the apical side of PAOCs compared to that of normal osteoclasts. In our current observations, H⁺ localized at the marginal area of the large flattened normal osteoclasts almost corresponding to the positions of podosomes (27). In contrast, osteoclasts formed in the presence of IL-1β secrete abundant H⁺ against the surface of dentin at multiple portions just beneath the area of the nuclei; therefore, one osteoclast is presumed to bear several large areas of H⁺ secretion. IL-1β–induced osteoclasts are likely to possess several resorption areas in each osteoclast, which could contribute to an extremely high resorption activity in PAOCs induced in the presence of IL-1β. Time-lapse imaging showed spatial changes of resorption cavities, indicating that the region of bone resorption gradually changes over time within one osteoclast (28). In our current study, we have examined the effect of RGDS peptide, which blocks integrin-mediated events, on proton secretion from osteoclasts adhered on dentin slices. In these experiments, normal osteoclasts showed complete inhibition by RGDS peptide, whereas PAOCs showed marked but partial inhibition (by 89.8%) (Supplemental Fig. 2). Such a difference in the inhibition by RGDS peptide could be attributed to the difference in the potential nature of these osteoclasts.

Studies in the Kindlin-3–deficient mice clarified the potential activation of integrin αvβ3 by Kindlin-3 and phosphorylation of DAP12 is essential for ruffled border formation and activation of normal osteoclasts (8, 9). Moreover, Kindlin-3–deficient mice present symptoms of osteopetrosis accompanied with severe bleeding in many organs and leukocyte deposition (29). These are due to the fact that integrins of hematopoietic cells cannot be fully activated (29). Also in humans, patients with mutated Kindlin-3 who developed leukocyte adhesion deficiency type III disease have been reported to develop osteopetrosis (30), due to the functional defects in osteoclasts. Findings in the current studies concerning the morphology of the PAOCs indicating high motility of osteoclasts formed in IL-1β could be related to these reports on Kindlin-3–deficient mice. Although Kindlin-3 expression is required for integrin activation, Talin-1 is essential for Kindlin-3 to exert its functions (31). Talin-1 is known to bind to β integrin tails and F-actin, and regulate affinity of β integrin (32) and it is essential for integrin activation (31). Talin-1 is involved in the regulation of static ⇄ dynamic conversion of integrin states in inside-out/outside-in integrin signaling (33). Upon Talin-1 binding to the β integrin tail, integrin is converted to the high-affinity state, which allows binding of Kindlin-3 to β integrin. This event occurs in the initial stage of integrin activation. In the terminal differentiation of normal osteoclasts, Talin-1 binds to the tail of β integrin and converts it into a form with high affinity to Kindlin-3, which allows activation of integrin β-chains. Activated integrin is clustered, and signals are transmitted downstream via outside-in signaling (33), resulting in the formation of ruffled borders and acquirement of bone resorption activity. In the current study, expression levels of Talin-1 and Kindlin-3 were markedly suppressed in IL-1β–induced osteoclasts with extremely high resorbing activity. IL-1β–induced marked reduction in the protein levels of Kindlin-3 and Talin-1 could cause the abnormality in integrin β3–mediated functions in PAOCs. Marked attenuation in the Kindlin-3–mediated activation of integrin β3 could be associated with the pathological bone destruction. In response to the gene expression of integrin β3, we could not detect significant effects of IL-1β on integrin β3 mRNA synthesis. Our data suggest that IL-1β treatment selectively suppresses the level of intracellular signaling molecules without influencing the level of surface integrin. Further studies are required to reveal the molecular entity of Kindlin-3/integrin β3–independent activation of osteoclasts in the pathological environments.

Our hypothesis is that PAOCs are formed in the pathological microenvironments of bone tissues. Resident macrophages are differentiated into specific types: from common progenitor cells in response to specified tissue microenvironments through the expression of sets of gene expression required for certain resident macrophages (34, 35). As osteoclasts are classified into one type of resident macrophage specific to bone tissues (34), it might be possible to consider that IL-1β–induced osteoclasts could form the subpopulation distinguished from normal osteoclasts. In the current study, we have successfully detected a difference in the expression pattern of membrane surface molecules in osteoclasts induced in the presence of IL-1β, although we could not so far identify it in the molecular level. There are several papers noting that osteoclasts showing pathological bone destruction are different from normal osteoclasts at the progenitor or precursor stage. Charles et al. (6) demonstrated that bone marrow cells expressing CD11b^−/loLy6C^hi give rise to osteoclasts performing pathological bone destruction in mice with collagen-induced arthritis. Fcγ receptor IV expression in osteoclast precursors enhances osteoclast differentiation and plays an important role in bone destruction of inflammatory arthritis (36–39). In addition, it has been reported that osteoclast precursor cells cultured and allowed to mature in an inflammatory environment have an enhanced bone resorption activity and can cause arthritis (40, 41). Gordon et al. (7) have shown the difference in the expression pattern of cell surface molecules between resident cells and inflammatory cells in monocytes and dendritic cells. In the case of osteoclasts, it is reasonable to think that PAOCs form the different subpopulation from normal osteoclasts. In our current study, we have detected cell surface molecules in mature osteoclasts, which are detected only in osteoclasts induced with IL-1β. Although Toh et al. (5) have reported an extremely high expression of M-CSF receptors on human osteoclasts induced in the pathological bone microenvironments in human rheumatoid arthritis patients, almost no other reports have been published concerning the expression of cell surface molecules induced by pathological factors. Detailed analysis of the cell surface molecules specifically expressed on this possible pathological subpopulation is essential to realize a highly specific regulation of PAOCs without affecting normal osteoclasts. Identification and detection of the unique osteoclast precursors specific to the pathological bone destruction in circulating peripheral blood is also important to establish a novel diagnosis of the next generation to assess the level of pathological bone resorption.

In the current study, we have examined the effect of IL-1β on osteoclastogenesis from human PBMCs. IL-1β showed regulatory effects on human osteoclastogenesis. In some cases, IL-1β stimulated osteoclastogenesis, whereas in other cases, IL-1β markedly suppressed osteoclastogenesis. The latter observation is similar to the results described in the previous report by Lee et al. (42). IL-1β is supposed to regulate differently among human populations. In future clinical trials, it would be ideal to perform pretests concerning the IL-1β signaling using PBMCs obtained from patients with bone diseases.

We thank Dr. S. Tamura of Nikon Instech Co., Ltd. for technical support.

The authors have no financial conflicts of interest.