To better understand the differences between t-OCLs and i-OCLs, we performed a comparative RNA-sequencing (RNAseq) analysis between sorted mature (≥three nuclei) MN-OCLs (originating from BM CD11b⁺ monocytic cells) and DC-OCLs (differentiated from BM-derived DCs) representing t-OCLs and i-OCLs, respectively, as already demonstrated (Ibáñez et al., 2016; Madel et al., 2020; Figure 1—figure supplement 1a–b). A total of 906 genes (log₂FC (Fold Change) ≥1; p<0.05) were significantly differentially expressed between the two OCL subsets (Figure 1a), including Cx3cr1 previously identified as a marker of i-OCLs (Ibáñez et al., 2016; Madel et al., 2020). The most differentially expressed genes were related to innate immunity and immune defense responses (Figure 1b–d), confirming our previous observation that i-OCLs and t-OCLs differ in their immune capacity (Ibáñez et al., 2016).

Volcano plot representation showed that the expression of genes involved in OCL resorbing activity as well as in the Rank-Rank-L differentiation pathway was not significantly different in the two OCL populations, except for Nfatc1. However, major differences were found in genes associated with the Ig-like receptor-dependent costimulatory OCL differentiation pathway (Koga et al., 2004; Merck et al., 2004; Figure 1e). Interestingly, annotation of these genes (EnrichR/Kegg analysis) highlighted that they are also linked to pathways related to immune responses and PRRs, notably the C-type lectin receptors (CLRs) (Figure 1f).

These results were confirmed by a global miRNA profiling on the same sorted mature OCL samples (≥three nuclei). Volcano plot visualization confirmed that the two OCL subsets differ also in their miRNA expression pattern (Figure 1g). Differentially expressed miRNAs (FC ≥2, p<0.05, quantitative PCR [qPCR] cycle threshold [CT] values <32) (Table 1) were further validated by quantitative real-time PCR (RT-qPCR) confirming that Mir151, Mir342, Mir146b, and Mir155 were significantly upregulated in i-OCLs, and that Mir185, Mir674, Mir26b, and Mir29a were upregulated in t-OCLs (Figure 1h). Using the miRWalk software that provides information on miRNA-target interactions (Dweep and Gretz, 2015), we carried out an integrative analysis of these miRNAs and genes that were significantly differentially expressed between both OCL subsets to find possible relationships (Table 2). In agreement with the RNAseq analysis, computational analysis of these related genes (EnrichR/ Kegg analysis) also revealed an association with the OCL differentiation pathway as well as the C-lectin-like receptor pathway (Figure 1i).

These data strongly suggest that CLRs could play an important role in the specific properties of i-OCLs and t-OCLs. Analysis of the RNAseq data for the expression of genes involved in the CLR and TLR signaling pathways confirmed major differences between t-OCLs and i-OCLs (Figure 1j). Validation by flow cytometry analysis of mature multinucleated OCLs confirmed that the proportion of OCLs expressing Dectin-1, Mincle, and Tlr2, but not Dectin-2, was significantly higher in i-OCLs compared to t-OCLs (Figure 1k). We also analyzed their expression in OCL progenitors. The proportion of cells expressing Dectin-1, Mincle, and Tlr2 was much higher in CD11c⁺ BM-derived DCs than in BM CD11b⁺ MNs, and the proportion of i-OCL progenitors expressing Dectin-2 was lower than for the other markers (Figure 1l).

The herein investigated OCL populations were derived from purified progenitor cells which are a powerful approach to identify major differences (Ibáñez et al., 2016; Madel et al., 2020), while in vivo, OCLs arise from a mixture of different BM progenitors whose proportions are depending on the pathophysiological conditions (Madel et al., 2019). Thus, we compared BM-derived OCLs from OVX-induced osteoporotic mice in which the proportion of i-OCL increases (Madel et al., 2020) to OCLs derived from SHAM mice. FACS analysis revealed higher proportions of Dectin-1⁺ and Tlr2⁺ OCLs, in OCLs generated from OVX mice compared to those from SHAM mice, while the proportion of Mincle⁺ OCLs was not significantly altered (Figure 2a).

These PRRs share the capacity to sense fungi and control responses to pathogenic or commensal gut mycobiome (Li et al., 2019; Sancho and Reis e Sousa, 2012). Therefore, we treated SHAM and OVX mice by oral gavage with the probiotic yeast Sb (Figure 2b). As expected, OVX mice showed atrophy of the uterus and their body weight increased compared to SHAM control mice, which was not affected by treatment with Sb (Figure 2—figure supplement 1a–b). Micro-computed tomographic (μCT) analysis revealed that Sb-treated OVX mice displayed reduced trabecular bone loss, as assessed by a significantly higher trabecular bone volume fraction (BV/TV) and trabecular number and less significant reduction in trabecular separation compared to untreated OVX mice (Figure 2c–d). Furthermore, administration of Sb resulted in a reduction of TRAcP⁺ OCL surface per bone surface in OVX mice (Figure 2e–f) and a significantly reduced serum level of cross-linked C-telopeptides of type I collagen (CTX), confirming a decrease in OCL activity (Figure 2g). Importantly, Sb treatment did not affect cortical parameters (Figure 2—figure supplement 1b), which could be due to a too short treatment period, as suggested in previous studies using bacterial probiotics (McCabe et al., 2013; Britton et al., 2014; Lawenius et al., 2022) or to a site-specific response to yeast derivates. It had no effect on the number of osteoblasts and osteocytes in vivo (Figure 2—figure supplement 2a–d) and on the mineralization capacity of osteoblasts in vitro (Figure 2—figure supplement 2e).

We then checked whether Sb treatment affects inflammatory parameters and microbiota metabolites that are known to influence bone remodeling (Zaiss et al., 2019). Evaluation of the gut barrier integrity by fluorescein isothiocyanate (FITC)-dextran assay showed a reduction in serum dextran concentration in Sb-treated OVX mice to the level observed in SHAM mice (Figure 3a) confirming the protective effect of Sb on the intestinal barrier as already reported in other pathological contexts (Terciolo et al., 2019). Modifications in the gut microbiome were evaluated by dosage of serum concentrations of metabolites produced by commensal bacteria such as propionate and butyrate, two major short chain fatty acids (SCFA) as well as lactate produced by lactic acid bacteria (LAB). Serum propionate, butyrate, and lactate were reduced in OVX mice, and treatment with Sb reversed this decrease, although not reaching statistical significance for lactate and butyrate (Figure 3b–d).

Interestingly, in Sb-treated OVX mice, we also observed a decreased proportion of BM CD4⁺ Tnf-α-producing T cells that have been reported to be responsible for increased osteoclastogenesis in OVX mice (Cenci et al., 2000; Figure 3e). Lastly, we evaluated i-OCL differentiation by analyzing mature BM-derived OCLs for their expression of Cx3cr1, the previously described marker for i-OCLs (Ibáñez et al., 2016). In Sb-treated OVX mice, the proportion of Cx3cr1⁺ i-OCLs was significantly reduced compared to non-treated OVX mice, while there was no alteration of Cx3cr1⁺ OCLs in SHAM mice (Figure 3f).

Our results showed that the beneficial effect of Sb in OVX mice was in part due to a systemic effect on inflammatory parameters responsible for pathological osteoclastogenesis. However, yeast derivatives such as ß-glucans are known to translocate to the blood and organs through the gut barrier (Isnard et al., 2021; Rice et al., 2005) and can therefore directly affect cells expressing PRR that recognize them. Thus, we determined the direct effect of agonists of these receptors on the differentiation of t-OCLs and i-OCLs. We used curdlan and zymosan as Dectin-1 and Tlr2 agonists and glucosyl-6-tetradecyloctadecanoate (GlcC₁₄C₁₈, a synthetic C6-branched glycolipid) to stimulate Mincle. The formation of t-OCLs from BM CD11b⁺ cells was not affected by any of these agonists (Figure 4a–c), consistent with their low expression of Dectin-1, Tlr2, and Mincle receptors (Figure 1k–l). In contrast, curdlan, zymosan, and GlcC₁₄C₁₈ dramatically inhibited the differentiation of BM-derived DCs into i-OCLs (Figure 4a–c). This effect was not due to an increased apoptosis of the cells in the presence of the agonists (Figure 4—figure supplement 1a–d). To confirm the involvement of Dectin-1, Tlr2, and Mincle in these effects, we used neutralizing antibodies and siRNA. Anti-Dectin-1 and anti-Mincle antibodies reversed the inhibitory effect of curdlan and GlcC₁₄C₁₈, respectively, on i-OCL formation compared to the control isotype (Figure 4d–e). Tlr2 siRNA decreased the expression level of membrane Tlr2, as expected (Figure 4—figure supplement 1e) and was able to abrogate the decrease of i-OCL differentiation induced by curdlan but not by zymosan (Figure 4f).

These data revealed that Dectin-1, Tlr2, and Mincle activation specifically inhibits the formation of i-OCLs without affecting t-OCLs. Additionally, we performed in vitro assays to assess the effect of these agonists on mature i-OCL activity. Our results revealed that, in addition to blocking i-OCL differentiation, agonists of these receptors also reduced their capacity to degrade mineralized matrix (Figure 4g), whereas they do not affect the activity of t-OCLs (Figure 4—figure supplement 1f).

In line with the aforementioned results, the Dectin-1 and Tlr2 agonists (curdlan and zymosan) also strongly reduced the differentiation of OCLs derived from BM cells of OVX mice in vitro, while they had no significant effect on the differentiation of OCLs generated from SHAM control mice (Figure 4h). However, GlcC₁₄C₁₈ reduced OCL differentiation of progenitors from OVX mice but also from SHAM mice (Figure 4h), according to the equivalent expression level of Mincle in these cells (Figure 2a).

The inhibition of the differentiation of i-OCL by a Mincle agonist was contrasting with a recent publication showing that Clec4e^−/− mice deficient in Mincle have increased bone mass and that sensing of necrotic osteocytes by Mincle increases OCL differentiation (Andreev et al., 2020). Accordingly, using µCT analysis, we observed the same increase in BV/TV, trabecular number, and trabecular thickness in Clec4e KO mice compared to controls, as already demonstrated (Andreev et al., 2020; Figure 4—figure supplement 2a). However, in vitro assays revealed that BM cells, BM-DCs, and BM-MNs from Clec4e^−/− mice differentiate into OCLs at least as efficiently as cells from controls (Figure 4—figure supplement 2b), contrasting with data from the literature (Andreev et al., 2020), which could be explained by different protocols used for in vitro OCL differentiation. Moreover, inhibition of i-OCL differentiation by GlcC₁₄C₁₈ was decreased in cells from Clec4e^−/− compared to Clec4e^+/+ mice confirming that Mincle is involved in this inhibition (Figure 4—figure supplement 2c). These data revealed that implication of Mincle in OCL differentiation is much more complex than expected.

Downstream signaling of Dectin-1 and Mincle largely involves activation of the spleen tyrosine kinase, Syk, which is also required for the differentiation and activity of OCLs (Mócsai et al., 2004). Accordingly, we found that BM-derived DCs from CD11cΔSyk mice that have selective depletion of Syk in CD11c⁺ cells (Iborra et al., 2012) failed to differentiate into OCLs (Figure 5a), indicating that Syk expression by DCs is required for i-OCL formation. As expected, FACS analysis showed that addition of curdlan to the OCL differentiation medium rapidly induced Syk phosphorylation (Figure 5b), which was followed at 24 hr by an increased proportion of MHC-II⁺ and CD80⁺CD86⁺ DCs, revealing the maturation of BM-derived DCs (Figure 5c). This treatment simultaneously reduced the proportion of RANK⁺ cells expressing Csf1r (CD115) and FcgRII/III (CD16/32), all required for OCL differentiation (Figure 5d). Furthermore, it also downregulated the expression of Syk, as well as its downstream targets Nfatc1 and Ctsk (Figure 5e). These results show that despite a rapid activation of Syk in BM-DCs treated with M-csf and Rank-L upon addition of curdlan, Syk expression decreases with time, as previously shown (Yamasaki et al., 2014), as well as the capacity of BM-DCs to differentiate into i-OCLs.

To investigate whether yeast probiotics have the same effect as the PRR agonists on the differentiation of i-OCLs, we used Sb-conditioned medium (Sb-CM). Sb-CM completely blocked the differentiation of i-OCL progenitors while it did not affect the formation of t-OCLs, except at very high concentrations (Figure 6a). This inhibitory effect was not associated with an increase in cell apoptosis (Figure 6b). Moreover, Sb-CM also reduced the resorption capacity of OCLs in vitro (Figure 6c). Next, we addressed the involvement of Dectin-1, Tlr2, and Mincle in the inhibitory effect of Sb-CM on i-OCLs as described above. While Tlr2-siRNA had no effect (Figure 6d), anti-Dectin-1 and anti-Mincle blocking antibodies completely abrogated the inhibitory effect of Sb-CM on the differentiation of i-OCLs, demonstrating the prominent role of these receptors in mediating the effect of Sb on inflammatory osteoclastogenesis (Figure 6e). Moreover, as curdlan, Sb-CM strongly stimulated BM-DC maturation (Figure 6f) while it dramatically decreased the proportion of Rank⁺ Csf1r⁺ (CD115⁺) and FcgRII/III⁺ (CD16/32⁺) cells representing OCL progenitors (Figure 6g). These results revealed that derivatives from Sb recognized by PRRs, such as ß-glucans, interfere with the capacity of BM-derived DCs to differentiate into i-OCLs and with the activity of these OCLs.

C57BL/6 mice were purchased from Charles River Laboratory at 4 weeks of age and housed in the Animal Facility of the University Côte d’Azur. Animals were maintained under a 12-hr light/dark cycle, and food and tap water were provided ad libitum. Female 6-week-old C57BL/6 mice were randomly divided into two groups for subsequent bilateral ovariectomy or SHAM surgery and were closely monitored until complete healing. Starting from 2 weeks after surgery, mice received by gavage Sb (Biocodex, Gentilly, France) 3 g/kg of body weight, 3 times per week until the end of the experiment. Six weeks after surgery, mice were sacrificed. CD11cΔSyk mice and CD11c-Cre littermates (Iborra et al., 2012) as well as Clec4e^+/+ and Clec4e^−/− (B6.Cg-Clec4e^tm1.1Cfg) (Wells et al., 2008) female mice were bred in Centro Nacional de Investigaciones Cardiovasculares in SPF conditions. All experiments were approved by the French Ministry of Health, Higher Education and Research (autorization number 8389–2016121216457153 v3) and conducted in accordance with the Institutional Ethics Committee on Laboratory Animals (CIEPAL-Azur, Nice Sophia-Antipolis, France).

i-OCLs and t-OCLs were differentiated in vitro as described previously (Halper et al., 2021; Ibáñez et al., 2016) from 6-week-old C57BL/6 mice (Figure 1—figure supplement 1). Briefly, CD11c⁺ BM-derived DCs cells were obtained by culturing 5×10⁵ BM cells/well in 24-well plates in RPMI medium (ThermoFisher Scientific) supplemented with 5% serum (Hyclone, GE Healthcare), 1% penicillin-streptomycin (ThermoFisher Scientific), 50 µM 2-mercaptoethanol (ThermoFisher Scientific), 10 ng/mL Gm-csf, and 10 ng/mL IL-4 (both PeproTech). The differentiation of t-OCLs and i-OCLs took 4–5 days and 5–6 days, respectively. CD11c⁺ DCs were isolated using biotinylated anti-CD11c (1:200; clone HL3; BD Biosciences) and anti-biotin microbeads (Miltenyi Biotec). iOCLs were differentiated by seeding a total of 2×10⁴ CD11c⁺ DCs/well on 24-well plates in MEM-alpha (ThermoFisher Scientific) including 5% serum (Hyclone, GE Healthcare), 1% penicillin-streptomycin, 50 µM 2-mercaptoethanol, 25 ng/mL M-csf, and 30 ng/mL Rank-L (both R&D) (OCL differentiation medium). For t-OCL culture, 2×10⁵ CD11b⁺ monocytic BM cells that were isolated by biotinylated anti-CD11b (1:100; clone M1/70; ThermoFisher Scientific) and anti-biotin microbeads (Miltenyi Biotec) were seeded per well on 24-well plates in OCL differentiation medium as described above. For OCL differentiation from OVX and SHAM control mice, 5×10⁵ BM cells were cultured per well in 24-well plates in OCL differentiation medium as indicated above. When indicated, differentiation was performed in the presence of indicated concentrations of curdlan, zymosan, or GlcC₁₄C₁₈. Medium conditioned by Sb CNCM I-745 (Sb-CM) was prepared by culture of the yeast in MEM-alpha overnight. The medium was collected and filtered through a sterile 22-µm filter, complemented with serum and antibiotics as described above and used as CM at indicated dilutions. All agonists and Sb-CM were added at day 0 (start of culture) and day 3 when the medium was changed. OCL differentiation was evaluated at the end of the differentiation after TRAcP staining according to manufacturer’s instructions (Sigma-Aldrich). Mature OCLs were enumerated under a light microscope as multinucleated (≥three nuclei/cell) TRAcP⁺ cells. OCL activity was evaluated by seeding OCL progenitors on plates coated with resorbable matrix (Osteoassay) in OCL differentiation medium. All agonists and Sb-CM were added at day 3 when the medium was changed and OCLs started to fuse. Resorbed areas were quantified (ImageJ software version 1.53, NIH, Bethesda, MD) after removing of the cells with water and staining of the mineralized matrix with alizarin red.

RNAseq analysis was performed on mature multinucleated OCLs. After differentiation in vitro, t-OCLs and i-OCLs (five biological replicates, each derived from one mouse) were detached with Accutase, labeled with 5 µg/mL H33342, and sorted on their multinucleation as previously described (Madel et al., 2018; c.f. Figure 1—figure supplement 1b for the gating strategy). Total RNA (100 ng) was extracted from sorted OCLs (RNeasy kit, Qiagen), and directional libraries were prepared (Truseq stranded total RNA library kit, Illumina). Libraries were pooled and sequenced paired-ended for 2×75 cycles (Nextseq500 sequencer, Illumina). 30–40 million fragments were generated per sample, and quality controls were performed. Data were analyzed by two approaches as described previously (Madel et al., 2020). Both gave equivalent results. For the first one, reads were ‘quasi’ mapped on the reference mouse transcriptome (Gencode vM15) and quantified (SALMON software, mapping mode and standard settings) (Patro et al., 2017). Transcript count estimates and confidence intervals were computed using 1.000 bootstraps to assess technical variance. Transcript counts were aggregated for each gene for computing gene expression levels. Gene expression in biological replicates (n=5) was then compared between t-OCLs and i-OCLs using Phantassus and the limma R package (https://ctlab.itmo.ru/phantasus) (Daria Zenkova, 2018). For the second approach, raw FASTQ reads were trimmed with Trimmomatic and aligned to the reference mouse transcriptome (Gencode mm10) with STAR (Dobin et al., 2013) on the National Institutes of Health high-performance computing Biowulf cluster. Gene-assignment and estimated read counts were assessed using HTseq (Anders et al., 2015). Gene expression was compared between t-OCLs and i-OCLs using DESeq2 (Love et al., 2014) with the Wald test (FDR <0.01).

Total RNA from sorted OCLs was extracted using the miRNeasy Micro Kit (QIAGEN), and the procedure automatized using the QIAcube (QIAGEN). The miRNA expression profiles were analyzed on paired samples using the TaqMan Array rodent MicroRNA Card Set v3.0 (TLDA, Applied Biosystems) after pre-amplification steps, according to manufacturer’s instructions. Relative expression and statistical analysis were calculated using the ExpressionSuite software (Applied Biosciences), which included the student’s t-test for sample group comparisons and built Volcano Plot comparing the size of the fold change (biological significance) to the statistical significance (p-value). Dysregulated miRNAs were examined with MirWalk (Dweep and Gretz, 2015), a miRNA database aiming to identify predicted and validated target genes and related pathways. This software provides information on miRNA-target interactions, not only on 3′-UTR, but also on the other regions of all known genes and simultaneously interrogates several algorithms (TargetScan, Miranda, RNA22, and miRWalk). We used a high predictive score with at least three of the four queried algorithms predicting miRNA target genes. Comparison of the expression patterns of 750 miRNAs in i-OCLs and t-OCLs was performed using Ct values <35, difference of at least twofold with a p-value <0.05.

Mature miRNAs of interest were specifically converted into cDNA using TaqMan microRNA reverse transcription kit according to the manufacturer’s protocol (Applied Biosystems). RT-specific primers 5× (ThermoFisher) were multiplexed in a primer pool containing 1% of each diluted in an adequate volume of Tris-EDTA 1× . Pre-amplification step was performed using FAM-labeled specific PCR primers 20× and TaqMan PreAmp Master Mix kit (Applied Biosystems) for 12 cycles. Alternatively, these preliminary steps were performed using Megaplex RT and PreAmp Primers, Rodent pool A (Applied Biosystems) that include specific primers for miRNAs of interest. was performed on diluted pre-amp products using the specific TaqMan PCR primers and TaqMan Universal Master Mix II with no uracile N-glycosylase and run on Viia7 system (Applied Biosystems) in 96-well PCR plates for 40 cycles. Relative miRNA expression was normalized on sno202 expression in murine cells with the 2^−ΔCT method.

Long bones of OVX and SHAM-operated mice were fixed in 4% paraformaldehyde. Bone microarchitecture analysis using high-resolution µCT was performed at the pre-clinical platform ECELLFRANCE (IRMB, Montpellier, France). Cortical and trabecular femora were imaged using high-resolution µCT with a fixed isotropic voxel size of 9 µm with X-ray energy of 50 kV, current of 500 mA, 0.5 mm aluminum filter, and 210 ms exposure time. Quantification of bone parameters was performed on the trabecular region of the proximal part of each femur (1.72 mm long) and on the cortical region (0.43 mm long region centered at the femoral midshaft) on CT Analyzer software (Bruker microCT, Belgium). For visual representation, 3D reconstructions were generated using NRecon software (Bruker µCT, Belgium).

At 1 hr before sacrifice, mice received oral gavage of 3–5 kDa FITC–dextran (Sigma-Aldrich) (60 mg/100 g body weight). FITC-dextran concentration in serum was measured by fluorometry in a fluorimeter (Xenius, SAFAS, Monaco) at 488/525 nm. Standard curve was prepared using dilutions of FITC-dextran in PBS with 20% fetal calf serum.

Tibias were fixed in 4% paraformaldehyde and decalcified for 72 hr in 4.13% EDTA, 0.2% paraformaldehyde pH 7.4, at 50°C in KOS microwave tissue processor (Milestone, Michigan, USA). They were then dehydrated and embedded in paraffin. TRAcP staining was performed as described (Lézot et al., 2015) with Mayer hematoxylin counterstaining on 3-μm thick sections to identify OCLs and three images of four to five biological replicates were analyzed. Immunostaining of osteoblasts and osteocytes was performed with rabbit polyclonal anti-Osterix antibody (ab22552, dilution 1/800; Abcam, Cambridge, UK) and rabbit polyclonal anti-sclerostin antibody (AF1589, dilution1:200; R&D System, Abingdon, UK), respectively, with Gill2 hematoxylin counterstaining. For enumeration of osteoblasts and osteocytes, one image of four to five biological replicates was analyzed. Analyses were performed on the entire trabecular area axcluding the cortical area. Stained sections were automatically numerized (nanozoomer, Hamamatsu photonics) before observation (NDP view virtual microscope, Hamamatsu) and quantification (ImageJ software version 1.53, NIH, Bethesda, MD).

Concentration of lactate, propionate, and butyrate in the serum was determined by ion chromatography analysis after depletion of proteins and lipids with acetonitrile (Sigma-Aldrich). Samples were loaded on a Dionex ICS-5000 Plus system automatic device (ThermoScientific), and elution was performed according to the manufacturer’s protocol. Chromatograms were aligned to standard solutions of each compound individually. Compound concentrations were determined using Chromeleon software (Thermo Scientific) by measuring surface area under the curve of the peaks and were compared to the corresponding ion standard profiles.

Serum crosslaps (CTX) were evaluated by enzyme-linked immunosorbent assay according to the manufacturer’s protocol (RatLaps (CTX-I) EIA, Immunodiagnostic Systems Limited).

CD11b⁺ monocytic BM cells and BM-derived CD11c⁺ DCs cells were analyzed for their expression of Dectin-1 (1:100; clone bg1fpj; ThermoFisher Scientific), Dectin-2 (1:100, clone 17611; R&D Systems), Tlr2 (1:200; clone 6C2; BD Biosciences), and Mincle (1:50, clone 6G5, Invivogen).

For analysis of BM-derived DC maturation, BM-DCs were treated or not with curdlan or Sb-CM at the indicated dose for 24 hr in OCL differentiation medium (containing 25 ng/mL M-csf and 30 ng/mL Rank-L). They were then labeled with anti-CD11c (1:200; Clone N418, eBioscience), MHC-II/IAb (1:200; Clone AF6-120.1, BD Bioscience), CD80 (1:100; Clone 16–10 A1, eBioscience), CD86 (1:100; Clone GL1, BD Bioscience), CD115 (1:100; Clone AFS98, eBioscience), and Rank/CD265 (1:100; clone R12-31, eBioscience). Cells were analyzed by flow cytometry (BD FACSCanto-II, BD Bioscience).

To investigate the Syk phosphorylation after curdlan stimulation, BM-derived DCs were stimulated for 15 min in OCL differentiation medium with the indicated doses of curdlan and subsequently fixed with 2% PFA (Transcription Factor Fixation/Permeabilization kit, eBioscience) over night at 4°C. Surface staining was then performed with anti-MHC-II/IAb (1:100;Clone AF6-120.1, BD Bioscience) and CD11c (1:200; Clone N418, eBioscience) antibodies before the cells were permeabilized with 1× Saponine (1 g/100 mL) for 15 min and stained with Phospho-Syk (Tyr348, clone moch1ct, eBioscience) for 45 min. Cells were washed and acquired on a BD FACSCanto II.

For FACS analysis on OCLs, mature OCLs were detached using Accutase (ThermoFisher Scientific), labeled with 5 µg/mL H33342 and with anti-Dectin-1, Dectin-2, Tlr2, and Mincle antibodies and analyzed after doublet exclusion as multinucleated cells with three or more nuclei as previously described (Madel et al., 2018) for their expression of these markers (see Figure 1—figure supplement 1b for the gating strategy). Cells were analyzed by flow cytometry (BD FACSCanto-II, BD Bioscience).

For intracellular cytokine analysis, T cells isolated from the BM of SHAM and OVX mice were stimulated with phorbol myristate acetate (PMA), ionomycin, and brefeldin A, labeled with anti-CD4 antibody (1:1000; clone RM4-5; BD Biosciences) and fixed with 4% formaldehyde overnight as described (Ciucci et al., 2015). Cells were subsequently stained with anti-Tnf-α antibody (1:400; clone MP6-XT22; ThermoFisher Scientific) in 1× Saponine (1 g/100 mL). Data were acquired using a FACSCanto-II (BD Biosciences). All FACS data were analyzed with FlowJo 10.8.1.

Osteoblastic differentiation of BM-MSC cells from C57Bl/6 mice was performed for 3 weeks in MEM-alpha medium with 10% FBS, 1% penicillin-streptomycin, 50 µM b-mercaptoethanol supplemented with 170 µM L-ascorbic acid, 10 mM β-glycerophosphate, and 0.1 µM dexamethasone (Merck). This differentiation medium (with adjusted concentration of inductors) was supplemented with the indicated % of Sb-CM. Medium was changed every 4 days. At day 21, cultures were fixed with 4% PFA and stained with 2% Alizarin red S (Merck). For the quantification, the Alizarin staining was dissolved in acetic acid (10%), heated 10 min at 85°C, centrifuged, and buffered with ammonium hydroxide. Absorbance at 405 nm was measured (Xenius SAFAS, Monaco).

For inhibition with siRNA, BM-derived DCs were transfected with ON-TARGET SMART pool of four siRNA to mouse Tlr2 (50 nM, Dharmacon, Horizon Discovery, USA) using lipofectamine RNAiMAX (Invitrogen) and were further differentiated into OCLs for 5–6 days either in OCL differentiation medium in the presence of the indicated concentration of agonists, or in Sb-CM as described above.

For inhibition with blocking antibodies, BM-derived DCs were cultured either in OCL differentiation medium containing the indicated concentration of agonists, or in Sb-CM. Anti-Dectin-1 (clone bg1fpj; ThermoFisher Scientific), anti-Mincle (clone 6G5, Invivogen), or control isotype antibodies were added at the indicated concentration at the beginning of the differentiation. Medium was changed at day 3.

For both approaches, OCLs differentiation was evaluated after TRAcP staining, and mature OCLs were enumerated under a light microscope as multinucleated TRAcP⁺ cells.

BM-derived DCs were stimulated for 72 hr with the indicated doses of curdlan. Total RNA was extracted with Trizol according to the manufacturer protocol. RNAs were reverse transcribed (Superscript II, Life Technologies), and RT-PCR was performed using SYBR green and the primers indicated below. Results were normalized to the 36B4 gene with the 2^−ΔCt method. Syk: AACGTGCTTCTGGTCACACA and AGAACGCTTCCCACATCAGG; Ctsk: CAGCAGAGGTGTGTACTATG and GCGTTGTTCTTATTCCGAGC; Nfatc1: TGAGGCTGGTCTTCCGAGTT and CGCTGGGAACACTCGATAGG; Rplp0: TCCAGGCTTTGGGCATCA and CTTTATCAGCTGCACATCACTCAGA.

Data were analyzed and statistics prepared using Graph Pad Prism 9.2 software. Analyses were done using two-tailed unpaired t-test when comparing two groups and ANOVA with multiple comparison test when comparison of more than two groups. Statistical significance was considered at p<0.05, and experimental values are presented as mean ± SD. Biological replicates were obtained from different mice.