A20 controls RANK-dependent osteoclast formation and bone physiology

Abstract

The anti-inflammatory protein A20 serves as a critical brake on NF-κB signaling and NF-κB-dependent inflammation. In humans, polymorphisms in or near the TNFAIP3/A20 gene have been associated with several inflammatory disorders, including rheumatoid arthritis (RA), and experimental studies in mice have demonstrated that myeloid-specific A20 deficiency causes the development of a severe polyarthritis resembling human RA. Myeloid A20 deficiency also promotes osteoclastogenesis in mice, suggesting a role for A20 in the regulation of osteoclast differentiation and bone formation. We show here that osteoclast-specific A20 knockout mice develop severe osteoporosis, but not inflammatory arthritis. In vitro, osteoclast precursor cells from A20 deficient mice are hyper-responsive to RANKL-induced osteoclastogenesis. Mechanistically, we show that A20 is recruited to the RANK receptor complex within minutes of ligand binding, where it restrains NF-κB activation independently of its deubiquitinating activity but through its zinc finger (ZnF) 4 and 7 ubiquitin-binding functions. Together, these data demonstrate that A20 acts as a regulator of RANK-induced NF-κB signaling to control osteoclast differentiation, assuring proper bone development and turnover.

Keywords: A20; bone; osteoclast - inflammation; osteoporosis.

Figure 1. Myeloid‐specific A20 deficiency promotes osteoclastogenesis

1. Representative micro‐CT pictures of hind legs of 20–30 week old control (A20^WT) and A20^myel‐KO littermates. Note the severe osteoporosis in A20^myel‐KO mice.

2. Trabecular parameters, calculated on micro‐CT scans of hind legs of 20–30 week old control (A20^WT) and A20^myel‐KO littermates. Each dot represents an individual mouse (A20^WT, n = 10; A20^myel‐KO, n = 8). Data are expressed as mean ± SEM. *, **, *** represent P < 0.05, P < 0.01, P < 0.001, respectively (parametric unpaired t‐test).

3. qRT–PCR mRNA expression analysis of osteoclast‐specific genes on RNA isolated from ankles of mice with the indicated genotypes (A20^WT, n = 9; A20^myel‐KO, n = 10). Data are expressed as mean ± SEM. * and ** represent P < 0.05, P < 0.01, respectively (parametric unpaired t‐test).

4. TRAP staining on osteoclast cultures derived from bone marrow cells isolated from tibia of A20^WT and A20^myel‐KO mice and incubated for 7 days with M‐CSF (25 ng/ml) and RANKL (100 ng/ml). Scale bar, 100 μm.

5. qRT–PCR mRNA expression analysis on RNA isolated from bone marrow cultures from A20^WT and A20^myel‐KO mice, incubated with M‐CSF (25 ng/ml) and RANKL (100 ng/ml) for 5 days to induce osteoclastogenesis (A20^WT, n = 3; A20^myel‐KO, n = 3). Data are expressed as mean ± SEM. ** represents P < 0.01 (the Kruskal–Wallis one‐way ANOVA test between indicated genotypes).

Source data are available online for this figure.

Figure 2. Osteoclast‐specific deletion of A20 in mice induces severe osteoporosis and enhanced osteoclastogenesis

1. Levels of TNF and IL‐6 in serum of A20^WT, A20^OC‐KO, and A20^myel‐KO mice. Each dot represents an individual mouse (A20^WT, n = 11; A20^OC‐KO, n = 11 and A20^myel‐KO, n = 14). Data are expressed as mean ± SEM. * represents P < 0.05, ** P < 0.01, **** P < 0.0001, n.s. nonsignificant (parametric one‐way ANOVA between indicated genotypes).

2. Representative micro‐CT pictures of hind legs of 20–30 week old control (A20^WT) and A20^OC‐KO littermates. Note the severe osteoporosis in A20^OC‐KO mice.

3. Trabecular parameters, calculated on micro‐CT scans of hind legs of 20–30 week old control (A20^WT) and A20^OC‐KO littermates. Each dot represents an individual mouse (A20^WT, n = 6; A20^OC‐KO, n = 9). Data are expressed as mean ± SEM. *, **, *** represent P < 0.05, P < 0.01, P < 0.001, respectively; ns, nonsignificant (parametric unpaired t‐test).

4. Functional femur strength was determined by the three‐point bending test and is expressed as the maximum force (Fmax, Newton) bones can resist before breaking, corrected for femur weight, with higher values mirroring stronger bones. Each dot represents an individual mouse (27 week old A20^WT, n = 9; A20^OC‐KO, n = 6). Data are expressed as mean ± SEM. * represents P < 0.05 (the nonparametric Mann–Whitney test between indicated genotypes).

5. Representative pictures of TRAP‐stained sections from the tibia of 20 week old A20^WT and A20^OC‐KO mice. Red arrows indicate osteoclasts. Scale bar, 500 μm.

6. Quantification of the number of TRAP‐positive cells on sections from the tibia of 20 week old A20^WT and A20^OC‐KO mice. Each dot represents an individual mouse (A20^WT, n = 5; A20^OC‐KO, n = 6). Data are expressed as mean ± SEM. * represents P < 0.05 (the nonparametric Mann–Whitney test between indicated genotypes).

Source data are available online for this figure.

Figure EV1. Myeloid‐specific A20 deficiency promotes osteoclastogenesis

1. Representative micro‐CT pictures of cortical bone of 20–30 week old control (A20^WT) and A20^myel‐KO littermates.

2. Cortical parameters, calculated on micro‐CT scans of hind legs of 20–30 week old control (A20^WT) and A20^myel‐KO littermates. Each dot represents an individual mouse.

Data are expressed as mean ± SEM. **, *** represent P < 0.01, and P < 0.001, respectively. ns, nonsignificant (parametric unpaired t‐test). Source data are available online for this figure.

Figure EV2. Osteoclast‐specific A20 deficient mice develop severe osteoporosis

1. Representative micro‐CT pictures of hind legs of 27 week old control (A20^WT) and A20^OC‐KO littermates. Note the severe osteoporosis in A20^OC‐KO mice.

2. Trabecular parameters were calculated on micro‐CT scans of the hind legs of 27 week old control (A20^WT) and A20^OC‐KO littermates. Each dot represents an individual mouse (A20^WT, n = 6; A20^OC‐KO, n = 6). Data are expressed as mean ± SEM. *, ** represent P < 0.05 and P < 0.01, respectively. ns, nonsignificant (the nonparametric Mann–Whitney test between indicated genotypes).

3. Representative micro‐CT pictures of cortical bone of 20–30 week old control (A20^WT) and A20^OC‐KO littermates.

4. Cortical parameters, calculated on micro‐CT scans of hind legs of 20–30 week old control (A20^WT) and A20^myel‐KO littermates. Each dot represents an individual mouse. Data are expressed as mean ± SEM. * represent P < 0.05. ns, nonsignificant (parametric unpaired t‐test).

5. Functional femur strength was determined by the three‐point bending test, expressed as the maximum force (F _max, Newton) bones can resist before breaking, corrected for femur weight (g), with higher values mirroring stronger bones. Each dot represents an individual mouse (16 week old A20^WT, n = 8; A20^OC‐KO, n = 7). Data are expressed as mean ± SEM. ** represents P < 0.01 (the nonparametric Mann–Whitney test between indicated genotypes).

6. Representative pictures of TRAP‐stained sections from the tibia of 50 week old A20^WT and A20^OC‐KO mice. Red arrows indicate osteoclasts. Scale bar, 500 μm.

7. Quantification of the number of TRAP‐positive cells on sections from the tibia of 50 week old A20^WT and A20^OC‐KO mice. Each dot represents an individual mouse (A20^WT, n = 5; A20^OC‐KO, n = 4). Data are expressed as mean ± SEM. * represents P < 0.05 (the nonparametric Mann–Whitney test between indicated genotypes).

Source data are available online for this figure.

Figure EV3. A20^OC‐KO mice do not show increased numbers of osteoclast precursors

1. Bar graphs representing the percentage of osteoclast precursors as measured by flow cytometry in the bone marrow of A20^WT, A20^OC‐KO and A20^myel‐KO animals. Data are expressed as mean ± SEM. * represent P < 0.05 (the two‐sided nonparametric Mann–Whitney test between indicated genotypes).

2. General gating strategy as applied for osteoclast precursor populations described in Source Data for Expanded View.

Source data are available online for this figure.

Figure 3. A20 regulates RANK‐induced NF–ĸB signaling

1. Bone marrow cells isolated from A20^WT and A20^OC‐KO mice were cultured on glass coverslips for 7 days in α‐MEM medium supplemented with M‐CSF (25 ng/ml) alone or with M‐CSF and RANKL (100 ng/ml). On day 7, osteoclasts were stained with TRAP. Scale bar, 50 μm.

2. qRT–PCR mRNA expression analysis on RNA from bone marrow cultures isolated from A20^WT and A20^OC‐KO mice and incubated for 6 days with M‐CSF (25 ng/ml) and RANKL (100 ng/ml; time point 0 represents a 7 day culture with M‐CSF alone). (A20^WT, n = 4–5; A20^OC‐KO, n = 4–5). ** represent P < 0.05, P < 0.01, respectively (Two‐way ANOVA test with Sidak's multiple comparisons between indicated genotypes for each time‐point).

3. qRT–PCR mRNA expression analysis on RNA from BMDM cultures isolated from control (A20^WT) mice (n = 3) incubated for the indicated time‐points with RANKL (100 ng/ml). Data are expressed as mean ± SEM.

4. Measurement of NF–κB luciferase activity in lysates of HEK293T cells transiently transfected with an NF–κB reporter plasmid, an expression plasmid for β‐galactosidase (βgal), an expression plasmid for RANK, and with either an empty vector or a vector expressing wild‐type A20. Cell lysates were analyzed for luciferase, and βgal activity, and values are plotted as Luc/βgal to normalize for possible differences in transfection efficiency. Data are expressed as the mean of technical triplicates ± SD. **** represent P < 0.0001; ns, non‐significant (Two‐way ANOVA test with Sidak's multiple comparisons between indicated genotypes).

5. Western blot analysis of whole cell lysates from BMDMs cultures isolated from A20^WT and A20^myel‐KO mice and stimulated with RANKL (100 ng/ml) for the indicated time periods. Actin is shown as a loading control.

6. Wild‐type and A20 deficient BMDMs were stimulated with GST‐RANKL (1 μg/ml) for the indicated time periods. The RANK signaling complex was immunoprecipitated using glutathione‐sepharose beads and immunoblotted for RANK, A20, TRAF6, HOIP, and Sharpin.

7. BMDMs from A20^WT and A20^myel‐KO mice were stimulated with RANKL (100 ng/ml) for the indicated time periods and immunoblotted to visualize p100 processing to p52. Actin is shown as a loading control.

Source data are available online for this figure.

Figure EV4. No difference in the number of TRAP‐positive osteoclasts or expression of inflammatory cytokines in primary cultures of A20^OC‐KO mice

1. Bone marrow cells isolated from A20^WT and A20^OC‐KO mice were cultured on glass coverslips for 7 days in α‐MEM medium supplemented with M‐CSF (25 ng/ml) and RANKL at different concentrations, as indicated. On day 7, osteoclasts were stained with TRAP.

2. Number of TRAP‐positive osteoclasts after incubation of bone marrow cells with M‐CSF (25 ng/ml) and RANKL at different concentrations, as indicated. Data are expressed as mean ± SEM. n.s., nonsignificant. (the two‐way ANOVA test with Sidak's multiple comparisons between indicated genotypes).

3. Bone marrow cells isolated from A20^WT and A20^OC‐KO mice were cultured on bovine bone slices in α‐MEM medium supplemented with M‐CSF (25 ng/ml) and RANKL (100 ng/ml) for 12 days. Culture supernatant was isolated at the indicated time‐points, and degradation products of C‐terminal telopeptides of type I collagen (CTX‐I) were quantified by ELISA. Data are expressed as mean ± SEM (n = 2 biological replicates with two technical replicates for each, technical replicates within each group are indicated with different symbols).

4. qRT–PCR mRNA expression analysis for inflammatory cytokines on bone marrow cultures from A20^WT and A20^OC‐KO mice, incubated with M‐CSF (25 ng/ml) and RANKL (100 ng/ml) for 6 days (A20^WT, n = 3; A20^myel‐KO, n = 3). Data are expressed as mean ± SEM. ns, nonsignificant (one‐way ANOVA with Tukey correction for multiple comparison).

5. IL‐1β, IL‐6, and TNF levels in the supernatant of bone marrow cultures from A20^WT and A20^OC‐KO mice, incubated with M‐CSF (25 ng/ml) and RANKL (100 ng/ml) for 6 days. Data are expressed as mean ± SEM. ns, nonsignificant (one‐way ANOVA with Tukey correction for multiple comparison).

Source data are available online for this figure.

Figure 4. A20 regulates RANK‐induced NF–κB signaling and osteoclast formation via its ubiquitin binding domains

1. A

   Measurement of NF–κB luciferase activity in lysates of HEK293T cells transfected with an NF‐κB luciferase reporter plasmid, an expression plasmid for β‐galactosidase (βgal), an expression plasmid for RANK, and plasmids encoding different A20 variants: A20^WT, A20^DUB (C103A mutation), A20^ZnF4 (C624A/C627A), A20^ZnF7 (C775A/C779A), and A20^ZnF4ZnF7 (C624A/C627A/C775A/C779A). Cell lysates were analyzed for luciferase and βgal activity, and values are plotted as Luc/βgal to normalize for possible differences in transfection efficiency. Data are expressed as the mean of technical triplicates ± SD. **** represent P < 0.0001; ns, nonsignificant (Two‐way ANOVA test with Sidak's multiple comparisons between indicated genotypes).

2. B

   qRT–PCR mRNA expression analysis on RNA from BMDM cultures isolated from control (A20^WT), A20^KO, A20^ZnF4ZnF7, and A20^DUB mice (with C103R mutation) incubated for the indicated time‐points with RANKL (100 ng/ml). Data are expressed as mean ± SEM. *, **, ***, and **** represent P < 0.05, P < 0.01, P < 0.001, and P < 0.0001, respectively; ns, nonsignificant (Two‐way ANOVA test with Sidak's multiple comparisons between indicated genotypes for each time‐point).

3. C, D

   Western blot analysis of whole cell lysates from BMDMs differentiated from A20^WT and A20^ZnF4ZnF7 mice, stimulated with RANKL (100 ng/ml) for the indicated time periods. Actin is shown as a loading control.

4. E

   BMDMs isolated from control (A20^WT), A20^ZnF7, A20^ZnF4ZnF7, and A20^DUB mice were stimulated with GST–RANKL (1 μg/ml) for the indicated time periods. The RANK signaling complex was immunoprecipitated using glutathione‐sepharose beads and immunoblotted for A20. Actin was used as a loading control.

5. F

   Wild‐type BMDMs were pretreated for 1 h with HOIPIN‐8 (50 μM) or left untreated and stimulated with GST–RANKL (1 μg/ml) for the indicated time periods. The RANK signaling complex was immunoprecipitated using glutathione‐sepharose beads and immunoblotted for A20. Actin was used as a loading control.

Source data are available online for this figure.

Figure EV5. Schematic model illustrating the mechanism by which A20 regulates TNF‐induced inflammatory signaling (left) and RANKL‐induced osteoclast formation (right)

Activation of RANK by its ligand RANKL induces the activation of a signaling pathway leading to the expression of A20 and recruitment of A20 to the RANK receptor complex. A20 inhibits RANK‐induced signaling and expression of osteoclast‐specific genes via its ubiquitin‐binding properties. A20 thus acts as a negative regulator of RANK‐induced signaling, similar to its function in the regulation of inflammatory signaling.