Fig 2: Osteoblast-specific deletion of SHN3 inhibits articular erosion and systemic bone loss in two models of RA. (A–F) Serum transfer arthritis: Five-week-old Shn3fl/fl and Shn3Prx1 mice were induced with serum transfer arthritis. (A) Shn3 deletion as assessed by RT-PCR in tibial bones (n = 4). (B) Clinical inflammation (Top) and ankle thickness (Bottom) scores show no difference (n = 5 ~ 7/group). (C) Sagittal H&E-stained sections of inflamed ankles. (Scale bars: 500 µm.) T: tibia, Ta: talus, N: navicular bone. (D) Histologic quantification of inflammation and erosion show no difference in inflammation, but protection from erosion in Shn3Prx1 mice compared with Shn3fl/fl mice (n = 4 ~ 5/group). (E) Tibio-talar joint sections were immunostained for OCN (Top) or TRAP (Bottom). Ob, osteoblast; Oc, osteoclast. (Scale bars: 100 µm.) (F) Rankl/Opg mRNA ratios and mRNA levels of Sfrp1, Dkk1, and Sost in Shn3fl/fl and Shn3Prx1 tibias (n = 4). (G–L) TNF-tg model of RA. (G) MicroCT sagittal sections of inflamed ankles (boxes indicate the talus bone) and (H) H&E-stained sections of inflamed midfoot in 12-wk-old Shn3fl/flTNF-tg and Shn3Prx1TNF-tg mice show protection from erosion in the Shn3Prx1TNF-tg mice. (Scale bars: (G) Top, 400 µm; Bottom, 200 µm; (H) 250 µm). (I and J) H&E- (Top, I) and TRAP- (Middle & Bottom, I) stained sections of inflamed tibio-talar joints and (J) histologic quantification of inflammation (Top) and erosion (Bottom) show protection from articular erosion in Shn3Prx1TNF-tg mice compared with Shn3fl/fl mice (J, n = 5 ~ 9). (Scale bars: Top, 250 µm; Middle, 250 µm; Bottom, 100 µm). (K) Shn3 mRNA levels in tibial bones confirm Shn3 deletion (Left) and Rankl/Opg ratios show a significant decrease in Rankl/Opg mRNA levels in Shn3Prx1 and Shn3Prx1TNF-tg mice compared with Shn3fl/flTNF-tg mice (n = 8). (L) MicroCT 3D reconstruction images of femoral trabecular bone mass showing protection from systemic bone loss in Shn3Prx1TNF-tg mice (Left, Scale bars: 500 µm). Quantification of trabecular BV/TV shows significant increases in Shn3Prx1 and Shn3Prx1TNF-tg mice compared with Shn3fl/fl TNF-tg mice (Right, n = 5 ~ 6/group) Ta: talus, II: inflammatory infiltration, P: pannus. Values represent mean ± SD: NS, nonsignificant; **P < 0.01; ***P < 0.001; ****P < 0.0001 by a two-tailed unpaired Student’s t test for comparing two groups (A–D, F, and J) or one-way ANOVA with Dunnett’s multiple comparisons test (K and L).

Fig 3: SHN3 expression in osteoblasts is up-regulated in RA. (A) Immunohistochemistry (IHC) for SHN3 in the inflamed metatarsal bones of RA patients (n = 5). IgG control antibody (Left) shows negative staining. Red arrows indicate staining for SHN3 protein in osteoblasts lining the bone surface (Mid, Top). Staining for SHN3 protein in osteoclasts (Mid, Bottom) is confirmed in a serial section stained for TRAP (blue arrows: Right, Bottom). (Scale bar = 50 µm.) (B and C) Shn3 mRNA expression in the tibial bone and synovium in two mouse models of RA, SKG, and TNF transgenic (tg) mice and in synovium in SKG mice. There is a statistically significant increase in Shn3 mRNA expression in the tibial bone of both SKG and TNF-tg mice compared with bone in nonarthritic wild-type (WT) littermate controls, but no difference in Shn3 mRNA expression in synovium from SKG arthritic mice compared with WT nonarthritic mice (n = 6 ~ 12). (D) Match database analysis (>100,000 analyses) demonstrating the top 10 biological ligands associated with SHN3 expression. (E and F) Human bone marrow stromal cells (BMSCs, E), mouse calvarial osteoblasts (COBs, F), mouse BMSCs (F) and mouse osteocytes (OCY454, F) were treated with TNF, IL-17A or the combination for 6 h and Shn3 mRNA expression was assessed by RT-PCR (n = 4). (G) Mouse bone marrow-derived monocytes (BMMs) were differentiated into preosteoclasts in the presence of M-CSF and RANKL for 3 d, then treated with TNF, IL-17A, or the combination for 6 h. Alternatively, undifferentiated BMMs or B or T cells isolated from mouse spleen were similarly treated. Shn3 mRNA expression was assessed by RT-PCR (n = 3 ~ 4). Data shown in (A–C) and (D–G) are representative of three independent experiments. Values represent mean ± SD: NS, nonsignificant; *P < 0.05; **P < 0.01; ****P < 0.0001 using a two-tailed unpaired Student’s t test for comparing two groups (B and C) or one-way ANOVA with Dunnett’s multiple comparisons test (E–G).

Fig 4: ERK-induced phosphorylation of SHN3 is required for TNF-induced bone loss. (A and B) C3H10T1/2 cells expressing Flag-Shn3 were stimulated with TNF at different time points (A). Cells were pretreated with an inhibitor of p38, ERK, or JNK MAPK for 30 min, then stimulated with TNF for 15 min (B). Cell lysates were immunoprecipitated with Flag-conjugated agarose and immunoblotted with anti-phospho-serine/threonine antibody. Input indicates the immunoblots of pre-IP cell lysates with the indicated antibodies. (C) A cell-free kinase assay using recombinant ERK2 (rERK2) and His-SHN3 (rSHN3, p.50-936 aa) demonstrates ERK-induced phosphorylation of SHN3. (D) Schematic diagram showing SHN3 phosphorylation sites by ERK MAPK (Top, p.630-936 aa of rSHN3). A cell-free kinase assay using rERK2 and rSHN3s (WT, KA, STA) was performed, demonstrating ERK-induced phosphorylation of SHN3 at S810/S811/T851/S911/S913 via the interaction at K886/K887/K888 (Bottom). S810/S811/T851/S911/S913 (STA, ERK-phosphorylation defective mutant) or K886/K887/K888 (KA, ERK-binding defective mutant) were substituted to alanine. Coomassie blue staining indicates loading control of rSHN3. Zn-F, zinc finger domain; BAS, basic-, acidic-, and serine-rich region. (E and F) BMSCs were isolated from Shn3Prx1TNF-tg mice, reconstituted with SHN3 proteins (WT, ?D, KA, STA) via lentivirus-mediated delivery, and cultured under conditions of osteogenic differentiation. ALP activity (E) and expression of Ibsp and Col1a1 (F) were assessed at day 6 of culture. ?D: D domain deletion. (E, n = 5; F, n = 4). (G–I) MicroCT images showing sagittal sections of inflamed ankles in 8-wk-old TNF-tg and Shn3KA/KATNF-tg mice. Red arrows indicate bone erosion sites (G). MicroCT analysis showing trabecular bone mass in the femur. 3D reconstruction images and relative quantification of trabecular bone mass are displayed (H). Representative images of calcein/alizarin red labeling and relative histomorphometric quantification of BFR/BS and MAR are displayed. Yellow arrows indicate the distance between calcein and alizarin labeling (I). (H, n = 6 ~ 8/group; I, n = 7 ~ 10/group). (Scale bars: G, 400 µm; H, 500 µm; I, 50 µm.) (J) Shn3+/+ and Shn3KA/KA BMSCs were transfected with the ß-catenin-responsive reporter gene (Topflash-Luc) along with Renilla, treated with TNF for 24 h, and luciferase activity was measured and normalized to Renilla (n = 5). (K) mRNA levels of the ß-catenin-target gene Lef1 in tibial RNAs of 8-wk-old TNF-tg and Shn3KA/KATNF-tg mice (n = 4). (L and M) Representative longitudinal sections of femurs from 8-wk-old TNF-tg and Shn3KA/KATNF-tg, immunostained for ß-catenin (L) or stained for TRAP (M, Left). Quantification of TRAP-stained osteoclasts per bone perimeter (M, Right). (Scale bars: 50 µm.) (N) Rankl/Opg mRNA ratio in Shn3+/+ and Shn3KA/KA BMSCs following TNF treatment (n = 4). (O) Shn3+/+ and Shn3KA/KA BMSCs were transduced with lentivirus expressing SHN3 proteins (WT, KA, STA), and cultured under osteogenic conditions in the presence of TNF for 6 d. Rankl/Opg mRNA ratio was assessed by RT-PCR (n = 4). Values represent mean ± SD: NS, nonsignificant; *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001 by ordinary one-way ANOVA with Dunnett’s multiple comparisons test (E, F, H, and O) or a two-tailed unpaired Student’s t test for comparing two groups (I, K, and N).

Fig 5: SHN3 deletion inhibits inflammation-induced bone loss in SKG mice. (A) Ten-week-old WT (Shn3+/+), Shn3-/-, SKG, and Shn3-/-SKG mice were injected with curdlan. Ankle thickness and clinical joint inflammation were measured weekly. WT and Shn3-/- demonstrate no inflammation, while SKG and Shn3-/- SKG mice develop equivalent levels of inflammation. (n = 3 ~ 4: WT (Shn3+/+), Shn3-/- groups, n = 9 ~ 10: SKG, Shn3-/-SKG groups). (B) Flow cytometric analysis of immune cell from inflamed synovial tissues isolated from SKG and Shn3-/-SKG mice show equivalent percentages of immune cells (n = 3). (C and D) H&E- and TRAP-stained sections of inflamed ankles (C) and relative quantification of inflammation and erosion (D), demonstrating that Shn3-/-SKG mice are protected from articular erosion. Red arrows indicate TRAP+ osteoclasts. (Scale bars: Left, 200 µm; Right, 50 µm; Ta, talus; II, inflammatory infiltrates; N, navicular bone (n = 9~10/group).) (E) MicroCT images showing sagittal sections of inflamed ankles in SKG and Shn3-/-SKG mice, demonstrating fewer erosions in Shn3-/-SKG. Red arrows indicate sites of bone erosion. (Scale bars: Top, 400 µm; Bottom, 100 µm.) (F) MicroCT analysis showing trabecular bone in the femurs of SKG and Shn3-/-SKG. 3D reconstruction (Top) and relative quantification of trabecular bone volume/total volume (Tb. BV/TV, bottom) (n = 6/group). (Scale bars: 500 µm.) (G and H) Longitudinal femoral sections were immunostained for osteocalcin (OCN) or stained for TRAP (G) and relative quantification of TRAP+ osteoclasts per bone perimeter (G) and OCN+ osteoblasts (Ob) per bone perimeter (B. Pm) (H) is displayed. Red arrows indicate TRAP+ Ocs, blue arrows indicate OCN positive Obs (n = 3 ~ 5/group). (Scale bars: 50 µm.) (I) Serum levels of osteocalcin (OCN, Left) or cross-linked C-terminal telopeptide 1 (Ctx-I, Right) in SKG and Shn3-/-SKG mice 7 wk postinjection of curdlan as measured by ELISA (n = 7 ~ 12/group). Values represent mean ± SD: NS, nonsignificant; *P < 0.05; **P < 0.01; ****P < 0.0001 by a two-tailed unpaired Student’s t test for comparing two groups (A, B, D, and I) or one-way ANOVA with Dunnett’s multiple comparisons test (F and H).