ICOS-Ligand Triggering Impairs Osteoclast Differentiation and Function In Vitro and In Vivo

The inducible costimulator ligand or B7h (CD275, also known as ICOSL, B7H2, B7-RP1, GL50) belongs to the B7 family of surface receptors and binds ICOS (CD278), which belongs to the CD28 family (1–5). ICOS is expressed by activated T cells, whereas B7h is expressed by a wide variety of cell types, including B cells, macrophages, and dendritic cells (DCs). However, B7h is also expressed by cells of nonhemopoietic origin such as vascular endothelial cells (ECs), epithelial cells, and fibroblasts, and many tumor cells. The main known function of B7h is the triggering of ICOS, which acts as a costimulatory molecule for activated T cells by modulating their cytokine secretion and, particularly, increasing the secretion of IFN-γ (in humans), IL-4 (in mice), IL-10, IL-17, and IL-21 (in both species) (6–11). However, recent reports have shown that the B7h–ICOS interaction can trigger bidirectional signals able to modulate also the response of the B7h-expressing cells. In mouse DCs, this B7h-mediated reverse signaling induces partial maturation with prominent augmentation of IL-6 secretion (12). In human DCs, it modulates cytokine secretion, promotes the capacity to cross-present endocytosed Ags in class I MHC molecules, and inhibits adhesiveness to ECs and migration (13, 14). B7h stimulation also inhibits the adhesiveness and migration of ECs and tumor cell lines in vitro and development of experimental lung metastases in vivo (15, 16). These effects are accompanied by decreased phosphorylation of ERK and p38 in ECs, decreased phosphorylation of focal adhesion kinase, and downmodulation of β-Pix in ECs and tumor cells. Moreover, triggering of B7h potentiates signaling via several pattern recognition receptors in human DCs through a signaling pathway involving the adaptor protein receptor for activated C kinase 1 and the kinases protein kinase C (PKC) and JNK (14).

The aim of our research was to extend these analyses by investigating the expression and function of B7h in osteoclasts (OCs), which derive from the monocyte lineage, similarly to DCs. OCs are giant cells formed by the cell–cell fusion of monocyte-macrophage precursors and are characterized by multiple nuclei, abundant vacuoles, and lysosomes; they play a key role in bone remodeling, which also involves osteoblasts (OBs) and osteocytes. OCs differentiate from monocytes under the influence of M-CSF and the receptor activator of NF-κB (RANK) ligand (RANKL) (17–21).

The OC function is stimulated by the triggering of the RANK expressed on the membrane of OCs by RANKL. In healthy bone, RANKL is mainly expressed by OBs as a surface receptor in response to bone-resorbing factors, and it is cleaved into a soluble molecule by metalloproteinases. Moreover, RANKL is also expressed by stromal cells, lymphocytes, and macrophages, which can support OC function during inflammation. Osteoprotegerin (OPG) is a soluble decoy receptor of RANKL secreted by OBs and stromal cells; OPG prevents RANK stimulation by inhibiting its binding to RANKL and impairs osteoclastogenesis (22). The binding of M-CSF to its CSF 1 receptor (c-fms) on OC progenitors upregulates expression of RANK on these cells and is essential for osteoclastogenesis (23). OC differentiation includes cell polarization with formation of ruffled membrane and sealing of the OCs to the bone to form a sealing zone, or clear zone, that separates the resorption lacunae from the surround. This is the secretion site of acid, tartrate-resistant acid phosphatase (TRAP), cathepsins, and metalloproteinases leading to demineralization of the inorganic component of the bone and hydrolysis of its organic components (17, 18). During physiological remodeling, after bone resorption, OBs are recruited within the resorption site, and the resorption lacuna is filled with bone matrix secreted by these cells; the matrix will then be mineralized by precipitation of hydroxyapatite crystals (21).

Increased OC activity leads to bone loss and can be detected in conditions such as osteoporosis, rheumatoid arthritis (RA), and other autoimmune diseases, in which a key role has been ascribed to inflammatory cytokines and adaptive immunity (24). Moreover, some neoplasia involving immune cells, such as multiple myeloma, are characterized by intense focal bone erosions ascribed to high expression of RANKL by stromal cells and, possibly, myeloma cells. Furthermore, bone metastases of solid cancer may be osteolytic through expression of a soluble form of RANKL (25, 26).

Several inflammatory cytokines, such as TNF-α, IL-1, IL-6, and M-CSF, upregulate RANKL expression and stimulate OC function (27–29). A key role is played by Th17 cells secreting IL-17, which induces the expression of RANKL in OBs and synovial cells. Moreover, IL-17 supports recruitment of several immune cell types producing cytokines and other proinflammatory molecules supporting OC differentiation and activity (30).

Our research analyzed the effect of B7h triggering by ICOS-Fc on OC differentiation and function both in vitro and in vivo. The results showed that monocyte-derived OC-like cells (MDOCs) express B7h during their differentiation, and that B7h triggering reversibly inhibits OC differentiation and function both in vitro and in vivo.

PBMCs were separated from human blood samples obtained from healthy donors, who gave their written informed consent, by density gradient centrifugation using the Ficoll-Hypaque reagent (Limpholyte-H; Cedarlane Laboratories, Burlington, ON, Canada). MDOCs were prepared from CD14⁺ monocytes isolated with the EasySepHuman CD14 Negative Selection Kit (STEMCELL Technologies, Vancouver, BC, Canada). Monocytes (0.5 × 10⁶) were plated in a 24-well plate and cultured for 21 d in a differentiation medium composed of DMEM (Invitrogen, Burlington, ON, Canada), 2 mM of l-glutamine, 10% FBS (Invitrogen), recombinant human M-CSF (25 ng/ml; R&D Systems, Minneapolis, MN), and RANK-L (30 ng/ml; R&D Systems). The differentiation medium was changed every 3 d. At different times (Supplemental Fig. 1A), cells were treated with 1 μg/ml of either ICOS-Fc (a fusion protein of the extracellular portion of the human ICOS fused to the human IgG1 Fc portion) or ICOS-msFc, composed of the human ICOS fused to the mouse IgG1 Fc. Controls were performed using ^F119SICOS-Fc, carrying the F119S substitution in the human ICOS amino acid sequence. For analysis, MDOCs were detached from the plates using the TrypLE express reagent (Life Technologies, Carlsbad, CA) for 15 min before using a cell scraper (31). Cell viability detected by trypan blue exclusion assay was >98%.

The OCs phenotype was assessed by immunofluorescence and flow cytometry (BD Biosciences, San Diego, CA) using the FITC-, PE-, and allophycocyanin-conjugated mAbs to CD14 (Immunotools, Friesoythe, Germany), cathepsin K (Bioss, Woburn, MA), and B7h (R&D Systems). Cathepsin K was evaluated after cell permeabilization using the FIX and PERM kit (Invitrogen).

Actin and B7h staining were performed on cells cultured on glass coverslips, fixed with 4% paraformaldehyde, and then permeabilized with 5% FBS, 1% BSA, and 0.1% Triton X-100. The cells were then stained with anti-B7h rabbit polyclonal Abs (Bioss) or preimmune rabbit Ig followed by Texas Red–conjugated secondary anti-rabbit Ig (Invitrogen), or with tetramethylrhodamine B isothiocyanate (TRITC)–conjugated phalloidin (Sigma-Aldrich, St. Louis, MO) in a solution of 0.1% Triton X-100, 1% BSA, and 2% FBS. Nuclear chromatin was stained with the fluorescent dye DAPI-dihydrochloride (Sigma-Aldrich). Stained cells were mounted with Slow-FADE (Light AntiFADE Kit; Molecular Probes Invitrogen) and observed by means of a fluorescence Leica DM 2500 fluorescence microscope equipped with a DFC7000 camera (all from Leica Microsystems, Milan, Italy); data were analyzed with Leica QWin Plus V 2.6 imaging software.

MDOCs were lysed in 50 mM of Tris-HCl (pH 7.4), 150 mM of NaCl, 5 mM of EDTA, and 1% Nonidet P-40 with phosphatase and protease inhibitor cocktails (Sigma-Aldrich). Then, 30 μg of proteins was run on 10% SDS-PAGE gels and transferred onto Hybond-C extra nitrocellulose membranes (GE Healthcare, Piscataway, NJ). The membranes were then probed with Abs to B7h (Bioss), phospho and total for p38 MAPK, Erk1,2, JNK, phospho-PKC (Cell Signaling Technology, Danvers, MA), β-Pix (Millipore, Billerica, MA), and β-actin (Sigma-Aldrich), followed by HRP-conjugated secondary Abs (Sigma-Aldrich). The bands were detected via chemiluminescence using the VersaDoc Imaging System (Bio-Rad Laboratories, Hercules, CA).

Monocytes (1 × 10³/well) were seeded in 96-well plates and incubated at 37°C, 5% CO₂, for 3 d. Cells were treated with ICOS-Fc (1 μg/ml) in complete medium with or without M-CSF (25 ng/ml). After 3 d of incubation, viable cells were evaluated by 2,3-bis[2-methoxy-4-nitro-5sulphophenyl]-2H-tetrazolium-5carboxanilide (Sigma-Aldrich) inner salt reagent at 570 nm, as described by the manufacturer’s protocol. The controls (i.e., cells that had received no treatments) were normalized to 100%, and the readings from treated cells were expressed as percentage of controls.

Total RNA was isolated from MDOCs cultures at day (T) 7, T14, and T21 from mice bone tissue (total limbs including scapula and pelvis), using TRIzol reagent (Invitrogen). RNA (500 ng) was retrotranscribed using the QuantiTect Reverse Transcription Kit (Qiagen, Hilden, Germany). DC-STAMP, OSCAR, NFATc1, and B7h expression were evaluated with a gene expression assay (Assay-on Demand; Applied Biosystems, Foster City, CA). The GAPDH gene was used to normalize the cDNA amounts. Real-time PCR was performed using the CFX96 System (Bio-Rad Laboratories) in duplicate for each sample in a 10 μl final volume containing 1 μl of diluted cDNA, 5 μl of TaqMan Universal PCR Master Mix (Applied Biosystems), and 0.5 μl of Assay-on Demand mix. The results were analyzed with a ΔΔ threshold cycle method.

Monocytes (0.5 × 10⁶/ml) were plated in 24-well Osteo Assay Surface culture plates (Corning, Corning, NY) and differentiated to MDOCs as described earlier adding the ICOS reagents at T14 or T21. As a control, monocytes were cultured also in the absence or presence of M-CSF alone. Supernatants were then collected and the calcium level was evaluated by a calcium colorimetric assay kit (Sigma-Aldrich). Moreover, erosion of the synthetic osteo-surface was visualized after staining with a modified Von Kossa method. In brief, cells were removed from the Osteo Assay Surface by incubation with 5% sodium hypochlorite for 5 min. Then, 300 μl of silver nitrate solution at 5% (w/v) was added to each well and the plate was incubated for 30 min in darkness at room temperature. Wells were then washed with distilled water and treated with 300 μl of balanced formalin at 5% (w/v) for 5 min at room temperature. After washing with distilled water, wells were aspirated and air-dried before imaging analysis (32).

Alternatively, monocytes (0.5 × 10⁶/ml) were plated in 96-well plates containing dentin disks (Pantec Srl, Torino, Italy) and differentiated to MDOCs as described earlier adding the ICOS reagents at T14 or T21. To analyze erosion pit formation on the dentin disk surface, we aspirated the medium from the wells and added 0.25% trypsin for 15 min. Then, wells were washed twice with distilled water, incubated with 0.25 M ammonium hydroxide, and stained with 0.5% toluidine blue followed by 2N NaOH. Individual pits or pit clusters were observed using a microscope at ×25 to ×100 magnification and analyzed with a specific program (33).

TRAP activity was assessed using the Acid Phosphatase kit (Sigma-Aldrich) according to the manufacturer’s instructions.

For RANKL-induced osteoporosis, we used 49-d-old C57BL/6 female mice, and groups were composed from the same littermates. Soluble RANKL (1 mg/kg; GenWay Biotech, San Diego, CA) was injected i.p. daily for 3 d (34) alone or in combination with 100 μg of ICOS-msFc or ^F119SICOS-Fc. Control mice were injected with PBS or 100 μg of ICOS-msFc or ^F119SICOS-Fc, but not with RANKL. The mice were sacrificed 4 h after the last injection, and tibias and femurs were collected for analysis.

For ovariectomy (OVX)-induced osteoporosis, we used 56-d-old C57BL/6 female mice, and groups were composed from the same littermates. Bilateral OVX was performed in mice anesthetized with a mix of Zoletil (Zolazepam, 60 mg/kg) and Xylazine (20 mg/kg) i.p., as reported previously (35, 36). The sham control group was subject to the same surgical procedures except for removal of the ovaries. One day after surgery, the mice were treated with seven i.p. injections (one every 4 d for 4 wk) of either PBS or msICOS-msFc (400 μg). They were sacrificed 4 d after the last injection, and organs and bones were collected for analysis.

Bone samples were fixed at room temperature for 2 d in 4% phosphate-buffered formaldehyde (pH 6.9), and un-decalcified bones were dehydrated in ethanol before a three-step impregnation was performed in methylmethacrylate monomer (Merck, Darmstadt, Germany) for 3 d. Sections were cut on a Leica SP 1600 Saw Microtome and mounted on polyethylene slides.

The cut was performed on the long axis of the bone in the femur and tibia metaphysis for trabecular bone and at the middiaphysis for cortical bone. The sections were stained with light green (Merck) and acid fuchsin (Sigma-Aldrich). Micrographs at ×4, ×10, ×20, and ×40 magnification were acquired and examined in a single blind analysis. Quantitative histomorphometric analysis was performed using micrographs obtained from six sections from each mouse (three sections per leg) and analyzed with Leica imaging software (Qwin Plus V 2.6; Leica).

Measurements of cortical bone included total area of the bone (TA) and medullary area (MA), whereas the mineralized bone area was calculated as TA − MA and expressed as percentage of mineralized bone in the TA. Measurements of trabecular bone were made in a fixed area of 0.17 mm² within which the MA was measured and the percentage of mineralized bone calculated.

The mice were bred under pathogen-free conditions in the animal facility of the Department of Health Sciences, and they were treated in accordance with the University Ethical Committee. The study was approved by the Bioethics Committee for Animal Experimentation of the University of Piemonte Orientale (protocol number 3/2014). Human blood samples were obtained from healthy donors who gave their written informed consent in accordance with the Declaration of Helsinki.

Statistical analyses were performed using ANOVA with Dunnett’s test using GraphPad Instat Software (GraphPad Software, San Diego, CA). Statistical significance was set at p < 0.05.

MDOCs were obtained by culturing CD14⁺ monocytes for 21 d in differentiation medium containing M-CSF and RANKL. To assess the MDOC differentiation, we evaluated the cell morphology by optical microscopy and expression of surface CD14, marking monocytes, and B7h and intracellular cathepsin K, marking OCs, by three-color immunofluorescence and flow cytometry performed at the beginning (T0) and the end (T21) of the MDOC differentiation culture and at the intermediate (T14) stage. The immunophenotypic analysis showed that, from T0 to T21, the cells downregulated CD14 and upregulated cathepsin K as expected (17–21, 23). The proportion of total cells expressing B7h was ∼30% at T0 and slightly decreased to ∼20% at T21. However, ∼30% of cathepsin K⁺ cells expressed B7h and ∼75% of B7h⁺ cells expressed cathepsin K at T21 (Fig. 1A). The morphological analysis showed that the cells acquired a spindle-like morphology at T14 (data not shown) and enlarged and fused in multinuclear cells at T21, as expected (Fig. 1B). Gating of the cytofluorimetric analyses on the cells with the highest size and granularity showed that these cells expressed B7h at T21 (Fig. 1C). Moreover, expression of B7h was assessed at T21 by indirect immunofluorescence and microscopy analysis. Cells were stained after fixing and permeabilization to stain nuclei and detect both extracellular and intracellular B7h, because both expressions have been detected in other cell types (37, 38). The results showed that B7h was expressed in both mononuclear and multinuclear cells (Fig. 1C). Finally, to confirm expression of B7h, we analyzed its mRNA by real-time PCR and the protein by means of Western blot at T7, T14, and T21. The results confirmed that B7h expression decreased during MDOC differentiation, but it was maintained at T21, especially at the protein level (Fig. 1D).

Because B7h is expressed during the MDOC differentiation culture, we evaluated the effect of B7h triggering on differentiating MDOCs using ICOS-Fc. To assess the specificity of the ICOS effect, we also treated cells with either ^F119SICOS-Fc, which is a mutated form of ICOS incapable of binding B7h, or ICOS-msFc, in which the human ICOS is fused with a mouse Fcγ portion to minimize interaction with the human Fcγ receptors. In preliminary experiments, we assessed the ICOS-Fc effect on monocyte proliferation by performing a 2,3-bis[2-methoxy-4-nitro-5sulphophenyl]-2H-tetrazolium-5carboxanilide assay on monocytes cultured for 3 d in the presence and absence of M-CSF with and without ICOS-Fc. The results showed that ICOS-Fc had no effect on monocyte proliferation in any culture condition (Supplemental Fig. 1).

Treatment of differentiating MDOCs was started at either T0 or T14 of the culture by adding the ICOS reagents to the differentiating medium. The culture was continued until T21 to perform the T^0–21 and T^14–21 treatments (Supplemental Fig. 1).

The results showed that the T^0–21 treatment with ICOS-Fc or ICOS-msFc powerfully inhibited MDOC differentiation. At T10, cells displayed a round shape (data not shown), and at T21, they acquired a spindle-like morphology and showed a decreased formation of multinuclear TRAP⁺ cells (expressed as percentage TRAP⁺ cells) (Fig. 2A, 2B). Cytofluorimetric analysis showed minimal downregulation of CD14 and upregulation of cathepsin K (Fig. 2C). By contrast, cells treated with ^F119SICOS-Fc did not display any difference from untreated cells, thus exhibiting the typical progression toward the MDOC morphology and phenotype.

Because a key aspect of OCs is cytoskeleton organization to form the ruffle border at the erosion area delimited by the sealing zone, we analyzed the effect of the ICOS reagents on the cell actin organization by intracellular staining of T^0–21-treated MDOCs with TRITC-phalloidin, which binds actin. Untreated MDOCs were giant cells with podosomes concentrated in well-organized dense actin rings, and actin was polarized with a pattern typical of the sealing zone delimiting the erosive lacuna of OCs. By contrast, in cells treated with ICOS-Fc or ICOS-msFc, cells were smaller, with podosomes arranged in noncircular clusters across the cell body, and actin displayed a perinuclear distribution in a typical F-acting ring without signs of polarization. Cells treated with ^F119SICOS-Fc displayed a pattern similar to that of untreated cells (Fig. 2D).

Because OC differentiation is marked by upregulation of DC-STAMP, OSCAR, and NFATc1 expression, we assessed the effect of the T^0–21 treatment with the ICOS reagents on the expression of these genes by real-time PCR at T7, T14, and T21. The results showed that ICOS-Fc and ICOS-msFc decreased expression of all these mRNAs compared with untreated cells and cells treated with ^F119SICOS-Fc (Fig. 3).

The T^14–21 treatment with either ICOS-Fc or ICOS-msFc showed a substantial slowing down of MDOC differentiation because, at T21, cells displayed decreased cell size and nuclei pyknosis, a decreased ability to adhere to the culture wells, more cells with one nucleus only or three or more nuclei, and fewer cells with less than three nuclei compared with untreated cells. Moreover, several cells displayed a star-like morphology that was not detected in untreated cells. Cytofluorimetric analysis showed a slight decrease of cathepsin K upregulation and a striking decrease of CD14 downregulation, so that CD14⁻ cathepsin K⁺ cells were ∼1% in the ICOS-Fc–treated cultures versus >50% cells in the control cultures. By contrast, cells treated with ^F119SICOS-Fc were similar to untreated cells (Fig. 4). Actin staining and TRAP assay performed on these cells at T21 showed data similar to those detected in the T^0–21 treatment (Supplemental Fig. 2).

To assess reversibility of the ICOS-Fc effect, we treated cells with the different ICOS reagents at T7, washed at T14, and then incubated to T21 in the absence of the ICOS reagents (T^7–14 treatment) (Supplemental Fig. 1). The results showed that the T^7–14 treatment induced a morphology, phenotype, actin staining, and TRAP activity converging on that displayed by untreated cells (Supplemental Fig. 3). By contrast, cells that, after the T14 washing, were cultured again in the presence of the ICOS reagents (T^7–21 treatment) displayed features similar to those described earlier for the T^14–21 treatment (data not shown).

Treatment of already differentiated MDOCs was performed by treating cells at T21 with the ICOS reagents and analyzing them after 3 d (T24) to perform the T^21–24 treatment (Fig. 5). The results showed that the T^21–24 treatment with ICOS-Fc or ICOS-msFc induced a striking decrease in the sizes of cells and nuclei, and a decreased expression of cathepsin K compared with untreated cells. By contrast, the T^21–24 treatment with ^F119SICOS-Fc did not display any effect.

To assess reversibility of the ICOS effect, we washed T^21–24-treated cells at T24 and incubated them for 1 (T25) or 4 d (T28) in a differentiation medium in the absence of ICOS-Fc. The results showed that cells treated with ICOS-Fc or ICOS-msFc and then grown in the absence of ICOS-Fc started to enlarge, upregulated cathepsin K at T25, and displayed a MDOC-like morphology converging on that displayed by untreated cells at T28. By contrast, cells that had been untreated or treated with ^F119SICOS-Fc maintained their morphology and phenotype at T25 and T28 (Fig. 5).

Analysis of cell viability by the trypan blue exclusion test showed that cells were viable in all of these culture conditions (data not shown).

To assess the effect of ICOS on the osteolytic activity of MODCs, we evaluated their ability to promote calcium release from crystalline calcium phosphate in vitro. MDOC differentiation was induced in wells coated with a synthetic surface made of an inorganic crystalline calcium phosphate mimicking living bone material in the presence and absence of the ICOS reagents using T^14–21 and T^21–24 protocols. At the end of the culture, cells were washed and cultured for a further 24 h in fresh medium, with or without the ICOS reagent, and release of calcium was then assessed in the culture supernatants using a colorimetric assay. The results showed that the T^14–21 and T^21–24 treatments with ICOS-Fc or ICOS-msFc significantly decreased the calcium release compared with untreated MDOCs, whereas ^F119SICOS-Fc did not display any effect (Fig. 6A). Moreover, we stained the surface wells with a modified Von Kossa method to visualize the erosion of the synthetic osteo-surface. Microscopic analysis showed a massive erosion in wells containing untreated MDOCs or MDOCs treated with ^F119SICOS-Fc, whereas erosion was inhibited in wells containing MDOCs treated with ICOS-Fc and minimal in wells containing monocytes cultured in the absence or presence of M-CSF alone (Fig. 6B).

To confirm these data, we assessed the effect of ICOS on the osteolytic activity of MODCs on dentin disks. MDOC differentiation was induced in wells containing dentine disks in the presence and absence of the ICOS reagents using T^14–21 and T^21–24 protocols. At the end of the culture, the disks were stained with toluidine blue, and erosion pits were visualized with a microscope and quantified with a dedicated software. The results showed that the T^14–21 and T^21–24 treatments with ICOS-Fc significantly decreased the erosion compared with untreated MDOCs, whereas ^F119SICOS-Fc did not display any effect (Fig. 6C).

To investigate the effect of B7h triggering at the signaling level, we treated MDOCs at T21 in the absence and presence of either ICOS-Fc or ICOS-msFc or ^F119SICOS-Fc (5 μg/ml) for 30 min, and then assessed the expression level of phospho-p38, phospho-Erk1,2, phospho-JNK, phospho-PKC, and β-Pix, which previous works have found to be modulated by B7h triggering in different cell types (13, 15, 16); expression of total p38, Erk1,2, JNK, and β-actin was assessed as the control. The results showed that phospho-p38 was upregulated by treatment with both ICOS-Fc and ICOS-msFc, but not with ^F119SICOS-Fc. By contrast, no effect was detected on phospho-Erk1,2, phospho-JNK, β-Pix, and phospho-PKC (Fig. 7).

Finally, we assessed the effect of B7h triggering in vivo by using two mouse models of osteoporosis. First, 49-d-old female C57BL/6 mice were injected i.p. daily for 3 d with RANKL (1 mg/kg) alone or in combination with either an msICOS-huFc (formed by the mouse ICOS and the human Fc) or ^F119SICOS-Fc (100 μg/mouse). The mice were sacrificed 4 h after the last injection. Histological representative images of cortical and trabecular bone stained with fuchsin and light green that evidenced fibrous and medullary tissues in red and mineralized bone in green are reported in Fig. 8A and Supplemental Fig. 4. Morphometric measurements of mineralized bone tissue in the cortical and trabecular bone showed a marked bone loss in the RANKL-injected mice compared with control mice (Fig. 8B), as expected (34). This bone loss was significantly inhibited by cotreatment of mice with RANKL plus msICOS-huFc, but not with RANKL plus ^F119SICOS-Fc. By contrast, treatment of mice with msICOS-huFc or ^F119SICOS-Fc alone in the absence of RANKL had no effect on the proportion of bone area.

Second, 56-d-old female C57BL/6 mice received surgery with or without OVX (sham controls) and, after 24 h, were injected i.p. every 4 d for 4 wk with either PBS or a total mouse ICOS-Fc (formed by the mouse ICOS and the mouse Fc). The mice were sacrificed 4 d after the last injection. Morphometric measurements of mineralized bone tissue showed a marked bone loss in the PBS-injected mice, and the bone loss was significantly inhibited by the treatment with ICOS-Fc. In the sham mice, no bone loss was detected, and treatment with ICOS-Fc did not show any effect (Fig. 9A, 9B). To evaluate whether the ICOS-Fc effect was ascribable to decreased OC activity, we evaluated expression of DC-STAMP and NFATc1 in the mRNA extracted from these bones by real-time PCR. The results showed that the treatment with ICOS-Fc significantly decreased the levels of DC-STAMP and NFATc1 compared with the controls. In the sham mice, levels of DC-STAMP and NFATc1 were significantly lower than in control OVX mice and were not modulated by treatment with ICOS-Fc (Fig. 9C).

Bone remodeling is a complex process managed by OBs and OCs, and the immune system is involved in regulating the function of these cells through the activity of cytokines and surface receptors. This article has described a novel pathway involved in lymphocyte–bone cell interactions by showing that the binding of ICOS, expressed by activated T cells, to its ligand B7h, expressed by MDOCs, inhibits MDOC maturation and function. These effects were detected using ICOS-Fc, and they were specific because they were not displayed by ^F119SICOS-Fc incapable of binding B7h.

The effect on MDOC differentiation was detected by treating cells with ICOS-Fc during the in vitro differentiation of monocytes to MDOCs. ICOS-Fc almost completely blocked the differentiation when treatment was started at the beginning of the 3-wk differentiating culture, but it inhibited the differentiation also when the treatment was started in the last week, as shown by the decreased cell multinuclearity and the arrest of acquirement of the OC features induced by treatment with ICOS-Fc. This effect was not due to cell toxicity because cell survival was normal even when cultures were prolonged for a fourth week (data not shown). Moreover, the effect was reversible, because interruption of the treatment in the last week of culture allowed cells to restart the OC differentiation path. The arrest of differentiation was accompanied by an altered organization of the actin cytoskeleton which, in ICOS-Fc–treated cells, displayed a perinuclear distribution in an F-acting ring without the signs of polarization typical of the sealing zone delimiting the erosive lacuna detected on OCs. In line with these findings, cells treated with ICOS-Fc displayed decreased expression of TRAP, OSCAR, DC-STAMP, and NFATc1, and decreased osteolytic activity in vitro.

A second key finding was the effect of ICOS-Fc on already differentiated MDOCs, in which treatment with ICOS-Fc induced a striking decrease in the sizes of cells and nuclei and osteolytic activity in vitro without substantial effects on cell viability. Again, the effect was reversible, because cells re-enlarged and reassumed the OC phenotype upon interruption of the treatment.

In differentiated MDOCs, B7h-mediated signaling seemed to involve phosphorylation of the p38 MAPK, which marks a difference from DCs in which B7h signaling involves JNK and PKC (14). Another difference was that MDOCs did not show the downmodulation of β-Pix that had instead been detected in DCs and tumor cell lines (13, 16). These signaling differences parallel functional differences displayed by these cell types because B7h triggering supports DC function by costimulating cytokine secretion in activated DCs, whereas it inhibits differentiation of OCs (12–14). The effect of B7h on p38 is in line with the key role ascribed to p38 in osteoclastogenesis, because, on the one hand, RANKL-induced osteoclastogenesis involves activation of the ERK, JNK, and p38 pathway, and on the other hand, OPG performs part of its antiosteoclastogenic activity by inducing p38 phosphorylation and altering the balance among the ERK, JNK, and p38 pathways needed for osteoclastogenesis (39–41).

These effects on MDOC differentiation and function in vitro were supported by the in vivo results showing that treatment with ICOS-Fc strikingly inhibits the systemic bone resorption induced in mice by high doses of soluble RANKL or OVX (34).

The observation that the ICOS–B7h interaction can modulate OC function is in line with the notion that several components of the immune system, including T cells, are able to modulate bone formation. Moreover, bone loss is a common feature of several chronic inflammatory and autoimmune diseases because the risk for osteoporosis is increased in patients with RA, inflammatory bowel disease, and systemic lupus erythematosus, and aggressive localized bone destruction can be a feature of certain autoimmune diseases, cancers, and infections (42, 43). In RA patients, the localized bone losses may involve inflammatory cytokines such as IL-1, IL-6, and TNF-α, which are abundant in synovial fluid and synovium, and can induce RANKL on synovial fibroblasts and stromal cells. Moreover, RANKL expressed by T and B cells in the synovial tissue and fluid can be involved in OC activation (44). Intriguingly, even the osteoporosis caused by menopausal estrogen deficiency may involve increased production of inflammatory cytokines such as TNF-α and IL-17, and increased RANKL expression on B and T cells (45).

In the immune control of bone formation, a key role has been ascribed to Th cells. Th1 and Th2 cells secrete IFN-γ and IL-4, respectively, which are antiosteoclastogenic cytokines (43). By contrast, Th17 cells express high levels of RANKL and secrete IL-17, inducing expression of RANKL on mesenchymal cells and recruitment of inflammatory cells (43). Moreover, Th17 cells also secrete IL-22, which may induce OB differentiation, enhancing bone formation at sites of inflammation. The cells may also act through use of surface receptors, because their CD40L can stimulate CD40 expressed on stromal cells, inducing them to upregulate RANKL and downmodulate OPG expression, and thereby support OC function (46). Moreover, regulatory T cells (Tregs) can inhibit osteoclastogenesis by release of TGF-β and surface expression of CTLA4, whose interaction with B7.1 and B7.2 on monocyte prevents their differentiation to OCs (47). The role of Tregs in inhibiting OC differentiation is in line with the antiosteoclastogenic activity of ICOS because not only do subsets of Tregs express high levels of ICOS, but also ICOS triggering is involved in Treg differentiation (48). A cooperative role of ICOS and CTLA4 in inhibition of OC function would be intriguing because both molecules belong to the CD28 family, bind surface receptors belonging to the B7 family, and are involved in Treg function. The antiosteoclastogenic activity displayed by ICOS and CTLA4 may counteract the pro-osteoclastogenic activity displayed by T cells by expression of RANKL, CD40L, and IL-17, and the overall effect may depend on the balance between these different signals (49). These findings open a novel field in the pharmacological use of agonists and antagonists of the ICOS–B7h system, which to date have been envisaged as immune modulators mainly in the fields of autoimmune diseases and antitumor immune response.

A patent application (Italy: 102015000018209; international: PCT/IB2016/052903) has been submitted to the Italian Patent Office for the use of ligands of the B7h receptor in the treatment of osteopenia and osteoporosis. The authors report no other conflicts of interest.