RANK-Independent Osteoclast Formation and Bone Erosion in Inflammatory Arthritis

Abstract

Objective: Proinflammatory molecules promote osteoclast-mediated bone erosion by up-regulating local RANKL production. However, recent evidence suggests that combinations of cytokines, such as tumor necrosis factor (TNF) plus interleukin-6 (IL-6), induce RANKL-independent osteoclastogenesis. The purpose of this study was to better understand TNF/IL-6-induced osteoclast formation and to determine whether RANK is absolutely required for osteoclastogenesis and bone erosion in murine inflammatory arthritis.

Methods: Myeloid precursors from wild-type (WT) mice or mice with either germline or conditional deletion of Rank, Nfatc1, Dap12, or Fcrg were treated with either RANKL or TNF plus IL-6. Osteoprotegerin, anti-IL-6 receptor (anti-IL-6R), and hydroxyurea were used to block RANKL, the IL-6R, and cell proliferation, respectively. Clinical scoring, histologic assessment, micro-computed tomography, and quantitative polymerase chain reaction (qPCR) were used to evaluate K/BxN serum-transfer arthritis in WT and RANK-deleted mice. Loss of Rank was verified by qPCR and by osteoclast cultures.

Results: TNF/IL-6 generated osteoclasts in vitro that resorbed mineralized tissue through a pathway dependent on IL-6R, NFATc1, DNAX-activation protein 12, and cell proliferation, but independent of RANKL or RANK. Bone erosion and osteoclast formation were reduced, but not absent, in arthritic mice with inducible deficiency of RANK. TNF/IL-6, but not RANKL, induced osteoclast formation in bone marrow and synovial cultures from animals deficient in Rank. Multiple IL-6 family members (IL-6, leukemia inhibitory factor, oncostatin M) were up-regulated in the synovium of arthritic mice.

Conclusion: The persistence of bone erosion and synovial osteoclasts in Rank-deficient mice, and the ability of TNF/IL-6 to induce osteoclastogenesis, suggest that more than one cytokine pathway exists to generate these bone-resorbing cells in inflamed joints.

Figure 1. BMMs differentiate into osteoclasts upon stimulation with TNFα/IL-6

(A) Representative TRAP stain of BMMs cultured with the indicated cytokines in the presence or absence of OPG, 100X magnification. (B) Quantitation of TRAP-positive multinucleated (≥3) cells (MNCs) generated as in panel A (Each data point represents average of 3 wells, n=3–6 mice per condition). Results in (A) and (B) are representative of 3 independent experiments. (C) qPCR for the indicated markers of osteoclast differentiation on BMMs cultured for 4 days with TNFα alone or with TNFα/IL-6, n=3 mice. (D) TRAP-stain of FACS sorted CD11b^−/lowLy6c^high osteoclast precursors differentiated with TNFα/IL-6 or RANKL, n=3 mice. Results in (C) and (D) are representative of 2 independent experiments. Multiple comparison adjusted p values are reported as follows: *, p<0.05; *p≤0.0003; ****, p≤0.0001 by one-way ANOVA with Sidak’s post-test for multiple comparisons.

Figure 2. TNFα/IL-6-induced OC differentiation and resorption is RANK independent

(A) TRAP stain of Rank^WT and Rank^Δ/Δ BMMs cultured with RANKL or TNFα/IL-6. (B) Quantitation of TRAP-positive MNCs generated as in (A) from triplicate cultures of each biologic replicate, n=3–6 mice per condition. Results in (A) and (B) are representative of 3 independent experiments. (C) Von Kossa stain of calcium phosphate coated plates incubated with Rank^WT and Rank^Δ/Δ BMMs cultured in the presence of the indicated cytokines, n=3 mice per condition. Results are representative of 2 independent experiments (D) CTX ELISA on culture supernatants of Rank^WT and Rank^Δ/Δ BMMs cultured in the presence of the indicated cytokines. Data are pooled from 2 independent experiments, n= 2–6 mice per condition. Multiple comparison adjusted p values are reported as follows: *, p<0.03; ****, p≤0.0001 by Sidak’s multiple comparisons test of a two-way ANOVA (B) and one-way ANOVA (D).

Figure 3. TNFα/IL-6-induced OC differentiation requires IL-6R signaling, NFATc1, DAP12, and cell proliferation

(A) TRAP stain of WT BMMs cultured with TNFα, TNFα/IL-6, or TNFα/IL-6 plus either IgG2a isotype control or anti-IL6R antibody, 100X magnification. Quantitation of TRAP-positive MNCs cultured with TNFα/IL-6 alone, or TNFα/IL-6 with 0.2, 2 and 20μg/mL of IgG2a isotype control or anti-IL6R antibody (right panel). Triplicate cultures of each biologic replicate, n=3 mice, were quantified for each condition. (B) TRAP stain of Nfatc1^WT or Nfatc1^Δ/Δ BMMs cultured with RANKL or TNFα/IL-6 (C) Quantitation of TRAP-positive MNCs generated from Nfatc1^WT or Nfatc1^Δ/Δ BMMs (left panel), or Dap12^+/+ or Dap12^−/− BMMs (right panel), cultured with the indicated cytokines. Triplicate cultures from n=3 mice for Nfatc1^Δ/Δ and n=2 for Dap12^−/− were quantified. (D) Quantitation of TRAP-positive MNCs from BMMs cultured with the indicated cytokines in the presence and absence of HU. Triplicate cultures from n=3 mice were quantified for each condition. Results are representative of 3 independent experiments except for (E). Multiple comparison adjusted p values are reported as follows: *, p<0.02; **, p< 0.005; ***, p <0.0002, ****, p≤0.0001 by Tukey’s (A) and Sidak’s (C) multiple comparisons test of a two-way ANOVA, and Sidak’s multiple comparisons test of a one-way ANOVA (D).

Figure 4. Rank^Δ/Δ mice are not fully protected from inflammatory bone erosion

(A) qPCR for IL6, Osm, and Lif on mRNA generated from the synovial tissue of WT mice 0 and 10 days after the initiation of K/BxN serum transfer arthritis (n=4–5 per time point). (B) Clinical scores (left panel) and change in paw thickness (right panel) from baseline of Rank^WT vs Rank^Δ/Δ mice after induction of K/BxN arthritis (n=10 and 12, respectively). (C) Upper panel shows H&E (left column) and TRAP stains (right 2 columns) of hindpaws from Rank^WT vs Rank^Δ/Δ 14 days after the initiation of K/BxN serum transfer arthritis. Arrows indicate erosion sites; arrowheads indicate osteoclasts in the marrow space of Rank^WT mice. Lower panel shows qualitative scoring of the histology shown for inflammation (left), bone erosion (middle), and TRAP staining (right) (D) Representative microCT images of forepaws from Rank^WT vs Rank^Δ/Δ 14 days after the initiation of K/BxN serum transfer arthritis with graph of the mean micro-CT erosion score for each mouse (lower panel). *, p<0.05, **, p<0.005 by Student’s t test.

Figure 5. Loss of RANK expression in the BM and synovium of Rank^Δ/Δ mice with K/BxN serum transfer arthritis

(A) qPCR for Rank expression in BM mRNA from Rank^WT vs Rank^Δ/Δ mice with arthritis (n=8–10); ****, p<0.0001 by Mann-Whitney test. (B) TRAP-stain of RANKL treated BMMs from Rank^WT vs Rank^Δ/Δ mice with arthritis. (C) Representative images of synovial cultures from Rank^WT vs Rank^Δ/Δ mice with arthritis cultured with RANKL or TNFα/IL-6 stained with TRAP (left panel) with quantitation of TRAP-positive multinucleated (≥3) cells (MNCs) (right panel). Results are representative of three independent experiments with the exception of A, where results are pooled from two independent experiments. Multiple comparison adjusted p values are reported as follows ***, p≤0.0002 by two-way ANOVA with Sidak’s multiple comparison test. E. Serial frozen sections of the hindpaw of a Rank^Δ/Δ mTmG mouse with arthritis stained for TRAP (left panel) or visualized by direct fluorescence (middle panel), 200X. S, synovium; B, bone; C, cartilage; JS, joint space; *, erosion. Digital overlay of TRAP and fluorescence images (right panel), with far left panels showing 3x digitally magnified image of corresponding boxed area (red channel removed for clarity).