TGFβ reprograms TNF stimulation of macrophages towards a non-canonical pathway driving inflammatory osteoclastogenesis

Abstract

It is well-established that receptor activator of NF-κB ligand (RANKL) is the inducer of physiological osteoclast differentiation. However, the specific drivers and mechanisms driving inflammatory osteoclast differentiation under pathological conditions remain obscure. This is especially true given that inflammatory cytokines such as tumor necrosis factor (TNF) demonstrate little to no ability to directly drive osteoclast differentiation. Here, we found that transforming growth factor β (TGFβ) priming enables TNF to effectively induce osteoclastogenesis, independently of the canonical RANKL pathway. Lack of TGFβ signaling in macrophages suppresses inflammatory, but not basal, osteoclastogenesis and bone resorption in vivo. Mechanistically, TGFβ priming reprograms the macrophage response to TNF by remodeling chromatin accessibility and histone modifications, and enables TNF to induce a previously unrecognized non-canonical osteoclastogenic program, which includes suppression of the TNF-induced IRF1-IFNβ-IFN-stimulated-gene axis, IRF8 degradation and B-Myb induction. These mechanisms are active in rheumatoid arthritis, in which TGFβ level is elevated and correlates with osteoclast activity. Our findings identify a TGFβ/TNF-driven inflammatory osteoclastogenic program, and may lead to development of selective treatments for inflammatory osteolysis.

Fig. 1. TGFβ priming switches the action of TNF to effectively drive osteoclastogenesis.

a Human osteoclast differentiation determined by TRAP staining (left) and the relative area of TRAP-positive multinuclear osteoclasts (MNCs, ≥3 nuclei/cell) per well (right) in the cell cultures using human CD14(+)-monocytes treated with or without TGFβ for 3 days, followed by TNF stimulation for 6 days in the presence or absence of TGFβ. Mono-Mφ: monocyte to macrophage stage. TRAP-positive cells: red. (n = 5/group). b Relative area of TRAP-positive-MNCs per well in human TNF-induced osteoclast differentiation with or without TGFβ priming (n = 20/group). c Quantitative real-time PCR (qPCR) analysis of mRNA expression of the indicated genes during osteoclastogenesis using human CD14(+)-monocytes treated with or without TGFβ priming for 3 days, followed by TNF stimulation for the indicated days (n = 5/group). d Von Kossa staining (left) and the resorption area (%) (right) of human osteoclast cultures induced by TNF with or without TGFβ priming (n = 5/group). Mineralized area: black; resorption area: white. e Human TNF-induced osteoclastogenesis using human CD14(+)-monocytes treated with or without TGFβ priming for 3 days, followed by TNF stimulation for 6 days in the presence or absence of recombinant OPG (100 ng/ml). Left: TRAP staining; Right: quantification of the relative area of TRAP-positive-MNCs/well (n = 5/group). f Phagocytosis of zymosan particles (left: zymosan staining; right: quantification of zymosan-containing cells) in the cell cultures using human CD14(+)-monocytes with or without TGFβ priming, followed by TNF stimulation for one day. Fluorescent zymosan particles: green; nuclei: blue; F-actin: red. (n = 6/group). g qPCR analysis of mRNA expression of IL1B and IL6 using human CD14(+)-monocytes treated with or without TGFβ priming for 3 days followed by TNF stimulation for the indicated times (n = 5/group). h Schematic of TGFβ priming effects on TNF-mediated macrophage response. a, c, e, g **p < 0.01; ***p < 0.001; ****p < 0.0001; n.s. not statistically significant by two-way ANOVA with Bonferroni’s multiple comparisons test. b, d, f **p < 0.01; ****p < 0.0001 by two-sided Student’s t test. Error bars: a, c, e, g, Data are mean ± SD. b, d, f Data are mean ± SEM. Scale bars: a, e, f 200 µm; d 100 µm. Source data are provided as a Source Data file.

Fig. 2. Loss of TGFβ signaling suppresses inflammatory osteoclastogenesis and bone resorption.

a Osteoclast differentiation using BMMs derived from WT and Tgfbr2^ΔM mice was stimulated with RANKL for 3 days. Left: TRAP staining. Right: relative area of TRAP-positive-MNCs (≥3 nuclei/cell) per well. TRAP-positive cells: red. (n = 5/group). b, c μCT images (b) and bone morphometric analysis (c) of trabecular bone of the distal femurs isolated from the indicated 12-week-old-male littermate mice (n = 5/group). d, e Osteoclast differentiation determined by TRAP staining (d, left) and the relative area of TRAP-positive MNCs per well (d, right), and qPCR analysis of mRNA expression of Ctsk, Acp5, and Dcstamp (e) in the cell cultures, in which the bone marrow of WT and Tgfbr2^ΔM mice was primed with or without TGFβ for 4 days, followed by TNF stimulation for 3 days. TRAP-positive cells: red. d, e: n = 5/group. f–h μCT images (f, left), the quantification of the resorption area (f, right), TRAP staining of bone surface (g, left) and histological sections (g, right), and histomorphometric analysis of the slices of the calvarial bones obtained from 12-week-old-male WT and Tgfbr2^ΔM mice after PBS or TNF injection to the calvarial periosteum daily for 5 days (n = 8/group in f, n = 6/group in h). i–k μCT images of the surface of tarsal joints (red arrows: bone erosion) (i), TRAP staining (j) and histomorphometric analysis (k) of the tarsal joint sections obtained from the indicated 12-week-old female mice with PBS injection as the control or the littermate female mice that developed K/BxN serum-induced arthritis (n = 6/group). BV/TV, bone volume per tissue volume; BMD, bone mineral density; Tb.N, trabecular number; Tb.Sp, trabecular separation; Tb.Th, trabecular thickness; ES/BS, erosion surface/bone surface; Oc.S/BS, osteoclast surface/bone surface; N.Oc/B.Pm, number of osteoclasts per bone perimeter. a, c ns: not statistically significant by two-sided Student’s t test. d, e, f, h, k **p < 0.01; ***p < 0.001; ****p < 0.0001; ns: not statistically significant by two-way ANOVA with Bonferroni’s multiple comparisons test. Error bars: a, c–f, h, k Data are mean ± SD. Scale bars: a, d, j 200 µm; b, g 100 µm; f, i 1.0 mm. Source data are provided as a Source Data file.

Fig. 3. TGFβ priming reprograms the action of TNF on the transcriptome in macrophages from inflammatory towards osteoclastic gene expression.

a–h RNA-seq analysis of the mRNAs from human CD14(+) monocytes treated with or without TGFβ priming for 3 days, followed by TNF stimulation for 1 or 6 days. n = 3 biological replicates for each condition. a Pathway analysis of TNF-inducible genes at 24 h in the non-priming condition. b Pathway analysis of TNF-inducible genes at day 6 in the TGFβ priming condition. c–e Heatmaps of TNF-induced type-I IFN response genes (c), chemokine genes (d), and osteoclast genes (e) regulated by non-priming and TGFβ priming at the indicated times. Row z scores of CPMs were shown in the heatmaps. f–h Gene set enrichment analysis of TNF-inducible type-I IFN response genes (f), chemokine genes (g), and osteoclast genes (h) regulated by non-priming and TGFβ priming ranked by NES. i–j qPCR analysis of mRNA expression of IFIT1, IFIT2, MX1, STAT1, CCL5, CXCL9 and CXCL10 using human CD14(+) monocytes treated with or without TGFβ priming for 3 days, followed by TNF for 1 day (n = 5/group). i, j ***p < 0.001 by two-way ANOVA with Bonferroni’s multiple comparisons test. Error bars: i, j Data are mean ± SD. Source data are provided as a Source Data file.

Fig. 4. TGFβ priming regulates chromatin accessibility and histone modification to suppress the inflammatory action of TNF and facilitate osteoclastogenesis.

Human CD14(+)-monocytes were treated with or without TGFβ priming for 3 days, followed by TNF stimulation for 0 or 1 day. a Volcano plot of ATAC-seq analysis of TNF-induced differentially accessible peaks at day 1 (gray dots) with significant (FDR < 0.01) and greater than twofold changes between non-priming and TGFβ-priming conditions. Data are from two biological replicates. Blue dots: peaks associated with ISGs. Red dots: peaks associated with osteoclast and TGFβ signaling genes. b Pathway analysis of genes associated with the significantly differentially accessible peaks identified in a. c, d Normalized ATAC-seq tag-density (heatmap in c) and tag counts (boxplots in d) of differentially accessible peaks associated with non-priming or TGFβ-priming genes. e Boxplots showing normalized ATAC-seq tag counts of differentially accessible peaks associated with ISGs or osteoclast genes. f Heatmap of normalized H3K4me3 Cut&Run-seq tag-density of the differentially accessible peaks associated with non-priming or TGFβ-priming genes. g Average tag density profile of normalized H3K4me3 Cut&Run-seq peaks associated with ISGs or osteoclast genes. Blue: Non-priming; Red: TGFβ-priming. h Heatmap of normalized H3K27me3 Cut&Run-seq tag-density of the differentially accessible peaks associated with non-priming or TGFβ-priming genes. i Average tag-density profile of normalized H3K27me3 Cut&Run-seq peaks associated with ISGs or osteoclast genes. Blue: Non-priming; Red: TGFβ-priming. j Heatmap of normalized H3K27ac Cut&Run-seq tag-density of the differentially accessible peaks associated with non-priming or TGFβ-priming genes. k Boxplots showing normalized H3K27ac Cut&Run-seq counts of peaks associated with ISGs or osteoclast genes. l Average tag-density profile of normalized H3K27ac Cut&Run-seq peaks associated with ISGs or osteoclast genes. Blue: Non-priming; Red: TGFβ-priming. m, n Representative IGV tracks displaying normalized tag-density profiles for ATAC-seq, H3K4me3, H3K27me3, and H3K27ac Cut&Run-seq signals at ISG (m) and osteoclast gene loci (n). d, e, k, m, n Data are representative of 2 biological replicates. Data are presented as normalized tag density ±2 kb (c) or 5 kb (f, h, j) around peak centers. d, e, k *p < 0.05; ***p < 0.001, ****p < 0.0001; ns, not statistically significant by two-way ANOVA with Bonferroni’s multiple comparisons test. Boxes represent data within the 25th to 75th percentiles. Whiskers depict the range of min to max. Horizontal lines within boxes represent median values. Source data are provided as a Source Data file.

Fig. 5. TNF-induced IRF1-IFNβ-ISG axis is suppressed by TGFβ priming.

a De novo motif-enrichment analysis of ATAC-seq peaks associated with non-priming or TGFβ-priming genes. Random background regions serve as a control. b IGV track displaying normalized tag-density profiles for ATAC-seq, H3K4me3, H3K27me3, and H3K27ac Cut&Run-seq signals at IFNB1 locus. c qPCR analysis of mRNA expression of IFNB1 using human CD14(+)-monocytes treated with or without TGFβ priming for 3 days, followed by TNF stimulation for the indicated times (n = 5/group). d ELISA analysis of IFNβ levels in the cell culture medium from TNF-induced osteoclastogenesis with or without TGFβ priming (n = 5/group). e ELISA analysis of IFNβ levels in the serum from the WT and Tgfbr2^ΔM mice after TNF-induced supracalvarial osteolysis (n = 5/group). f IGV track displaying normalized tag-density profiles for ATAC-seq, H3K4me3, H3K27me3, and H3K27ac Cut&Run-seq signals at IRF1 locus. g, h qPCR analysis of mRNA expression of IRF1 (n = 5/group) (g) and immunoblot analysis of the expression of IRF1 (h) using human CD14(+)-monocytes treated with or without TGFβ-priming for 3 days, followed by TNF stimulation for the indicated times. p38 was used as a loading control. i Immunoblot analysis of the expression of IRF1 using human CD14(+)-monocytes treated with or without TGFβ-priming for 3 days, followed by TNF or RANKL stimulation for 4 hr. p38 was used as a loading control. j Osteoclast differentiation was determined by TRAP staining (left) and the relative area of TRAP-positive-MNCs/well (right) in the cell cultures, in which the bone marrow of WT and Irf1^−/− mice was primed with or without TGFβ for 4 days, followed by TNF stimulation for two days. TRAP-positive cells: red. (n = 5/group). k–l qPCR analysis of mRNA expression of the indicated osteoclast genes (k) and ISGs (l) during osteoclastogenesis using the WT and Irf1^−/− cells with or without TGFβ-priming followed by TNF stimulation for the indicated times (n = 5/group). c, d, e, g, j, k, l **p < 0.01; ***p < 0.001; ****p < 0.0001; ns, not statistically significant by two-way ANOVA with Bonferroni’s multiple comparisons test. Error bars: c, d, e, g, j, k, l Data are mean ± SD. j 200 µm. Source data are provided as a Source Data file.

Fig. 6. TGFβ priming promotes IRF8 ubiquitination and degradation in response to TNF.

a, b Immunoblot analysis (a) and qPCR analysis (n = 5/group) (b) of IRF8 expression using human CD14(+) monocytes treated with or without TGFβ priming for 3 days, followed by TNF stimulation at the indicated times. p38 was used as a loading control (a). c Immunoblot analysis of the expression of IRF8 using human CD14(+) monocytes treated with or without TGFβ priming for 3 days, followed by TNF stimulation together with DMSO or MG132 (25 µM) for the indicated times. p38 was used as a loading control. d Immunoblot analysis of the expression of IRF8 using human CD14(+) monocytes treated with or without TGFβ priming for 3 days, then CHX (50 µM) for 30 min, followed by TNF stimulation together with DMSO or MG132 (25 µM) for the indicated times. p38 was used as a loading control. e Ubiquitination of IRF8 in the human CD14(+) monocytes treated with or without TGFβ priming for 3 days, followed by TNF stimulation together with DMSO or MG132 (25 µM) for the indicated times. Cell lysates were immunoprecipitated with anti-IRF8 antibody followed by immunoblotting with anti-Ub antibody. IP, immunoprecipitation; IB, immunoblotting. b Data are mean ± SD. Source data are provided as a Source Data file.

Fig. 7. B-Myb is a previously unrecognized osteoclastogenic regulator specifically involved in TGFβ-priming/TNF-mediated osteoclastogenesis.

a RNA-seq analysis and comparison of mRNA expression induced by RANKL or TGFβ-priming/TNF in human CD14(+)-monocyte cultures. Bottom: pathway analysis of non-DEGs between RANKL and TGFβ-priming/TNF conditions. b Volcano plot of the DEGs from a. Blue dots: genes more highly expressed in RANKL-induced-condition; red dots: genes more highly expressed in TGFβ-priming/TNF-condition (adjusted p < 0.001 and FC > 4). c De novo-motif-enrichment analysis of ATAC-seq peaks associated with OC genes in TGFβ-priming/TNF condition. d IGV track displaying the indicated seq signals at MYBL2 locus. e, f qPCR analysis of MYBL2 expression (n = 5/group, e) and immunoblot analysis of B-Myb (f) in human CD14(+)-monocytes treated with/without TGFβ-priming for 3 days, followed by TNF stimulation. g Immunoblot analysis of B-Myb in human CD14(+)-monocyte-derived macrophages transfected with LNAs. h, i, l TRAP staining and relative area of TRAP-positive-MNCs/well (h), and qPCR analysis of the indicated gene expression (i, l) in human CD14(+)-monocyte cultures treated with/without TGFβ for 3 days, transfected with the indicated LNAs, and followed by TNF stimulation for 6 days (h, i) or 1 day (l). (n = 5/group). j Immunoblot analysis of B-Myb in human CD14(+)-monocyte-derived macrophages stimulated with RANKL. p38 was used as a loading control (f, g, j). k TRAP staining and relative area of TRAP-positive MNCs/well in human CD14(+)-monocyte-derived macrophages transfected with the indicated LNAs, followed by 3-day-RANKL stimulation. (n = 5/group) m TRAP (upper) and Immunohistochemical staining of B-Myb (middle, brown) on calvarial slices from TNF-induced osteolysis model in 12-week-old-male mice. Nuclei: blue. Arrows: osteoclasts. n = 3. n–p μCT images and the quantification of resorption area (n), TRAP staining of bone surface (o, left) and histological sections (o, right), and histomorphometric analysis (p) of the calvarial slices from 12-week-old-male mice after PBS or TNF injection to calvarial periosteum daily for 5 days (n = 5/group). e, h, i, l, n, p *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001; ns, not statistically significant by two-way ANOVA with Bonferroni’s multiple comparisons test. k ns, by two-sided Student’s t test. Error bars: e, h, i, k, l, n, p Data are mean ± SD. Scale bars: h, k 200 µm; m 10 µm; n 1.0 mm; o 100 µm. Source data are provided as a Source Data file.

Fig. 8. Distinct TGFβ level/activity contributes to different osteoclastic activity in RA and SLE patients.

a Volcano plot of microarray analysis of the mRNA expression in the PBMCs isolated from SLE and RA patients. Blue dots: DEGs are more highly expressed in SLE PBMCs. Red dots: DEGs are more highly expressed in RA PBMCs. b, c Pathway analysis of the enriched DEGs in SLE (b) and RA (c). Note: c the upper TGFβ signaling Pathway ID is hsa04350 and the lower TGF-β Signaling Pathway ID is WP366. d Heatmaps of mRNA expression of the genes involved in osteoclasts, TGFβ signaling, Type-I IFN response, and Chemokine genes in the SLE and RA PBMCs. Row z-scores of Normalized Signal Intensity were shown in the heatmaps. n = 82 for SLE patients and n = 84 for RA patients. e Gene set enrichment analysis of DEG set (osteoclast, TGFβ signaling, type-I IFN response, and chemokine) in SLE and RA PBMCs. f Scatter plot showing the significantly positive correlation between the osteoclast activity and TGFβ activity. g Normalized Signal Intensity of TGFB1, IFNB1, IRF1, IRF8, FOXM1, and MYBL2 in SLE and RA PBMCs obtained from microarray data. n = 82 for SLE patients and n = 84 for RA patients. g *p < 0.05, **p < 0.01 by Welch’s t test (two-sided). Source data are provided as a Source Data file.