MEKK2 mediates an alternative β-catenin pathway that promotes bone formation

Abstract

Proper tuning of β-catenin activity in osteoblasts is required for bone homeostasis, because both increased and decreased β-catenin activity have pathologic consequences. In the classical pathway for β-catenin activation, stimulation with WNT ligands suppresses constitutive phosphorylation of β-catenin by glycogen synthase kinase 3β, preventing β-catenin ubiquitination and proteasomal degradation. Here, we have found that mitogen-activated protein kinase kinase kinase 2 (MAP3K2 or MEKK2) mediates an alternative pathway for β-catenin activation in osteoblasts that is distinct from the canonical WNT pathway. FGF2 activates MEKK2 to phosphorylate β-catenin at serine 675, promoting recruitment of the deubiquitinating enzyme, ubiquitin-specific peptidase 15 (USP15). USP15 in turn prevents the basal turnover of β-catenin by inhibiting its ubiquitin-dependent proteasomal degradation, thereby enhancing WNT signaling. Analysis of MEKK2-deficient mice and genetic interaction studies between Mekk2- and β-catenin-null alleles confirm that this pathway is an important physiologic regulator of bone mass in vivo. Thus, an FGF2/MEKK2 pathway mediates an alternative nonclassical pathway for β-catenin activation, and this pathway is a key regulator of bone formation by osteoblasts.

Keywords: MAPK; MEKK2; beta-catenin; bone; osteoblasts.

Fig. 1.

MEKK2 promotes osteoblast activity in vivo and in vitro. (A and B) Immunohistochemistry for MEKK2 (A) and phospho-MEKK2/3 (B) at 40× and 100× magnification, respectively on 1-wk-old mouse tibias. (C) Quantification of BV/TV and cortical thickness (C.Th) in the femur of 4-wk-old Mekk2^+/+ and Mekk2^−/− mice. All error bars indicate SEM. P < 0.05 was considered statistically significant by two-tailed, unpaired Student’s t test. *P < 0.05 and **P < 0.01; n = 5 mice per group. (D) 3D reconstructions of calvaria of 4-wk-old Mekk2^+/+ and Mekk2^−/− mice. (E) Histomorphometric analysis of tibias from 8-wk-old Mekk2^+/+ and Mekk2^−/− mice; n = 5 mice per group. (Upper) Photomicrographs of dual-labeled trabecular bones. (Lower) Quantification of BFR and MAR. *P < 0.05. (F) Serum levels of P1NP and CTX in 8-wk-old Mekk2^+/+ and Mekk2^−/− mice; n = 5 mice per group. *P < 0.05; N.S, not significant. (G) In situ hybridization for Ocn on tibias from 1-wk-old Mekk2^+/+ and Mekk2^−/− mice. (Magnification: 40×.) (H and I) Primary calvarial osteoblasts were isolated and placed under differentiation conditions for 21 d. (H) Von Kossa staining was performed to determine osteoblast mineralization activity. (I) After 7 d of culture, the expression of osteoblast marker genes was analyzed by RT-PCR. **P < 0.01; ***P < 0.001.

Fig. S1.

Characterization of the skeletal phenotype of Mekk2^−/− mice. (A) Histomorphometric analysis of tibias from 8-wk-old Mekk2^+/+ and Mekk2^−/− mice. N.Ob/B.Pm, number of osteoblasts per bone perimeter. (B) TRAP staining was performed on femurs from 4-wk-old Mekk2^+/+ and Mekk2^−/− mice. (C) Serum was collected from 4-wk-old Mekk2^+/+ and Mekk2^−/− mice, and an ELISA was run to determine OPG levels. N.S, not significant.

Fig. 2.

MEKK2 increases β-catenin stability. (A) GST or GST-MEKK2 was incubated with lysates of WNT10-expressing ST2 cells, immunoprecipitated with anti-GST antibody–conjugated agarose, and immunoblotted with an anti–β-catenin antibody. (B) HEK293 cells were transfected with vector or Flag–β-catenin along with HA-MEKK2, and lysates were immunoprecipitated with anti-Flag antibody–conjugated agarose and immunoblotted for the indicated antibodies. (C) COBs were lysed, immunoprecipitated with anti–β-catenin or an IgG control and protein G-conjugated Dynabeads, and immunoblotted with anti-MEKK2 antibody. (D) GST–β-catenin was incubated with purified HA-MEKK2 (WT or KD mutant), immunoprecipitated with anti-GST antibody–conjugated agarose, and immunoblotted with the indicated antibodies. Input indicates loading controls for purified HA-MEKK2. (E) C3H10T1/2 cells were transfected with different amounts of MEKK2 along with TOPflash-luciferase and Renilla. Luciferase activity was measured after 48 h of transfection and normalized to Renilla. P < 0.01 by one-way ANOVA. (F) Primary COBs isolated from Mekk2^+/+ and Mekk2^−/− mice were transfected with TOPflash luciferase and Renilla and were cultured under osteoblast differentiation conditions for 6 d. Luciferase activity was normalized to Renilla. ***P < 0.001. (G) HEK293 cells were transfected with vector or HA-MEKK2 (WT or KD mutant) along with Flag–β-catenin and EGFP, and lysates were immunoblotted with the indicated antibodies. EGFP was used as a transfection control. (H) Mekk2^+/+ and Mekk2^−/− COBs were cultured under osteoblast differentiation conditions for 7 or 14 d, and lysates were blotted with the indicated antibodies. (I) Immunohistochemistry for β-catenin on tibias from 1-wk-old Mekk2^+/+ and Mekk2^−/− mice. (J) HEK293 cells were transfected with vector or HA-MEKK2 along with Flag–β-catenin, and β-catenin stability was determined by pulse-chase labeling with [³⁵S]methionine followed by autoradiography. (K) Quantification of BV/TV and cortical thickness in the femur of the following strains: Mekk2^+/+;Ctnnb^fl/fl (^+/+), Mekk2^+/−;Ctnnb^fl/fl (^+/+), Mekk2^+/+;Ctnnb^fl/+;Prx1 (^+/−), and Mekk2^+/−;Ctnnb^fl/+;Prx1 (^+/−) mice. (n = 5 mice per group). In both the cortical thickness and BV/TV comparisons, P < 0.05 by two-way ANOVA for both the Mekk2 and Ctnnb groups and also for the interaction between these groups. **P < 0.01, ***P < 0.001, Bonferroni-corrected Student’s t tests.

Fig. S2.

MEKK2 is dispensable for JNK activation and responses to BMPs in osteoblasts. (A) Primary COBs were isolated from Mekk2^+/+ and Mekk2^−/− pups, cultured under osteoblast differentiation conditions for 7 d, and immunoblotted with the indicated antibodies. (B) Alternatively, expression of Jun and Fra1 was analyzed by RT-PCR analysis. P = N.S., not significant. (C and D) Primary COBs were cultured under osteoblast differentiation conditions in the presence or absence of 100 ng/mL of BMP2/7 for 7 d, and BMP-induced osteoblast differentiation was determined by ALP activity (C) and ALP staining (D). In C, P = N.S. for all comparisons between Mekk2^+/+ and corresponding Mekk2^−/− samples. P values for two-way t tests of samples ± BMP2/7 are indicated: ***P < 0.001.

Fig. S3.

MEKK2 regulates β-catenin activity and stability. (A) C3H10T1/2 cells were transfected with different amounts of MEKK2 (KD) or MEKK2 (WT) along with TOPflash-luciferase and Renilla. Luciferase activity was measured 48 h after transfection and normalized to Renilla. (B, Upper) A diagram showing the truncated mutants of MEKK2. (Lower) C3H10T1/2 cells were transfected with MEKK2 along with TOPflash-luciferase and Renilla. CT, C-terminal; FL, full length; NT, N-terminal. (C) Immunohistochemistry for β-catenin on tibias from 1-wk-old Mekk2^+/+ and Mekk2^−/− mice. (Magnification: 40×.) (D and E) Primary COBs were isolated from Mekk2^+/+ and Mekk2^−/− pups and were cultured under osteoblast differentiation conditions for 7 d. (D) Expression of β-catenin was analyzed by RT-PCR. (E) Cells were treated with 10 μM MG132 for 8 h and immunoblotted with anti–β-catenin antibody. (F) HEK293 cells were transfected with vector (vec) or HA-MEKK2 along with Flag–β-catenin, and β-catenin stability was determined by pulse-chase labeling with [³⁵S]methionine followed by autoradiography and ImageJ analysis. (G) Mass spectra data demonstrating identification of serine 675 as the site of MEKK2-mediated phosphorylation in β-catenin.

Fig. 3.

MEKK2 stabilizes β-catenin via phosphorylation of S675. (A) GST or GST–β-catenin was incubated with purified HA-MEKK2 (WT or KD mutant), and MEKK2 kinase activity was analyzed by an in vitro kinase assay. (B) Purified HA-MEKK2 was incubated with GST or GST–β-catenin [WT or S675A (SA) mutant], and MEKK2 kinase activity was analyzed by in vitro kinase assay. (C) Vector or β-catenin (WT or S675A mutant) was transfected into C3H10T1/2 cells along with TOPflash-luciferase and Renilla in the absence or presence of MEKK2. Luciferase activity was measured after 48 h of transfection and normalized to Renilla. P < 0.05 by one-way ANOVA. P < 0.05; P values for Bonferroni-corrected Student’s t tests: **P < 0.01; ***P < 0.001. (D) β-Catenin–deficient COBs were reconstituted with Flag–β-catenin WT or S675A mutant via lentivirus-mediated delivery and cultured under osteoblast differentiation conditions for 7 d. Lysates were immunoblotted with the indicated antibodies. (E) HEK293 cells were transfected with Flag–β-catenin WT or S675A mutant along with HA-MEKK2, and β-catenin stability was determined by pulse-chase labeling with [³⁵S]methionine followed by autoradiography. (F) HEK293 cells were cotransfected with Flag–β-catenin WT or S675A mutant along with different amounts of HA-ubiquitin and were treated with 10 µM MG132 for 8 h before lysis. Ubiquitinated β-catenin was immunoprecipitated by anti-Flag antibody–conjugated agarose and immunoblotted with anti-HA antibody. (G) An in vitro ubiquitination assay of β-catenin was performed using recombinant proteins. Activated His–β-catenin was incubated with the recombinant SCF^SKP1 complex, ubiquitin, ATP, E1, and E2 in the presence or absence of recombinant MEKK2. Ubiquitinated β-catenin was immunoprecipitated by Ni-NTA agarose, subjected to SDS/PAGE, and immunoblotted with an anti-ubiquitin antibody. S675 phosphorylation of β-catenin by MEKK2 was confirmed by immunoblotting. (H) WT immortalized COBs were lysed, immunoprecipitated with anti–β-catenin or an IgG control and protein G-conjugated Dynabeads, and immunoblotted with the indicated antibodies.

Fig. S4.

MEKK2 regulates β-catenin stability via S675 phosphorylation. (A) C3H10T1/2 cells were transfected with different amounts of β-catenin (WT or S675A mutant) along with TOPflash-luciferase and Renilla. Luciferase activity was measured 48 h after transfection and normalized to Renilla. Expression of β-catenin was analyzed by immunoblotting with anti–β-catenin antibody. (B) β-Catenin–deficient COBs were reconstituted with Flag–β-catenin WT or S675A mutant via lentivirus-mediated delivery and were cultured under osteoblast differentiation conditions for 7 d. Transcript levels of Flag–β-catenin were measured by RT-PCR. P values for Bonferroni-corrected two-way Student’s t tests are indicated: **P < 0.01; N.S., not significant. (C) HEK293 cells were transfected with Flag–β-catenin (WT or S675A mutant) along with HA-MEKK2, and β-catenin stability was determined by pulse-chase labeling with [³⁵S]methionine followed by autoradiography and analysis in ImageJ.

Fig. S5.

MEKK2 functions independently from other known β-catenin kinases in osteoblasts. (A) C3H10T1/2 cells were transfected with a construct encoding MEKK2 along with TOPflash-luciferase and Renilla. Twelve hours after transfection, cells were incubated with different concentrations of the PKA inhibitor H-89, and luciferase activity was measured and normalized to Renilla. (B) Primary COBs isolated from Mekk2^+/+ and Mekk2^−/− pups were stimulated with 10 μM forskolin at different time points, and lysates were immunoblotted with the indicated antibodies. (C) GST–β-catenin was incubated with recombinant GSK3β (Top) or CK1 (Middle) in the presence or absence of MEKK2, and β-catenin phosphorylation by GSK3β or CK1 was analyzed by immunoblotting with antibodies specific to phospho-β-catenin. (Bottom) GST–β-catenin was incubated with recombinant MEKK2 in the absence or presence of GSK3β, and MEKK2-induced phosphorylation of β-catenin was analyzed by immunoblotting with an anti–phospho-β-catenin (S675) antibody. (D) Primary COBs isolated from Mekk2^+/+ and Mekk2^−/− pups were lysed and immunoblotted with the indicated antibodies.

Fig. S6.

β-Catenin interacts with USP15. HEK293 cells were transfected with vector or β-catenin (WT or S675A mutant) along with Flag-USP15, immunoprecipitated with anti-Flag antibody–conjugated agarose, and immunoblotted with the indicated antibodies.

Fig. S7.

MEKK2 and USP15 interact with components of the Axin complex. HEK293 cells were transfected with the indicated constructs, immunoprecipitation was performed as indicated, and the subsequent immunocomplex was immunoblotted with the indicated antibodies.

Fig. 4.

USP15 promotes β-catenin stability and osteoblast differentiation. (A) USP15 or a vector control was transfected into HEK293 cells along with TOPflash-luciferase and Renilla in the absence or presence of β-catenin. Luciferase activity was measured 48 h after transfection and normalized to Renilla. P < 0.05 by one-way ANOVA. P values for Bonferroni-corrected Student’s t test: **P < 0.01. (B) C3H10T1/2 cells were infected with lentivirus expressing control or Usp15 shRNAs, and the resulting cells were transfected with Flag–β-catenin, TOPflash-luciferase, and Renilla in the presence or absence of MEKK2 WT or KD mutant. Luciferase activity was measured 48 h after transfection and normalized to Renilla. P < 0.05 by one-way ANOVA. P values for Bonferroni-corrected Student’s t tests: *P < 0.05. (C) HEK293 cells were transfected with Flag–β-catenin along with vector or USP15, and β-catenin stability was determined by pulse-chase labeling with [³⁵S]methionine followed by autoradiography. Numbers indicate the ratio of the densitometric measurement of the corresponding band to the value at time = 0. (D) HEK293 cells were infected with lentiviruses expressing control or Usp15 shRNAs. The resulting cells were transfected with Flag–β-catenin and MEKK2, and β-catenin stability was determined by pulse-chase labeling with [³⁵S]methionine followed by autoradiography. (E) Human MSCs were infected with the indicated lentiviruses and were cultured under osteoblast differentiation conditions for 6 d before immunoblotting. (F) An in vitro ubiquitination assay of β-catenin was performed using recombinant proteins. Activated His–β-catenin was incubated with recombinant SCF^SKP1 complex, ubiquitin, ATP, E1, and E2 in the presence or absence of recombinant MEKK2 or USP15. S675 phosphorylation of β-catenin by MEKK2 was confirmed by immunoblotting. (G) An in vitro interaction assay of ubiquitinated β-catenin with USP15 was performed using recombinant proteins. Flag-tagged ubiquitinated β-catenin was incubated with recombinant USP15 in the presence or absence of recombinant MEKK2, immunoprecipitated with anti-Flag antibody–conjugated agarose, and immunoblotted with the indicated antibodies. MEKK2-mediated β-catenin S675 phosphorylation was confirmed by immunoblotting with anti–phospho-β-catenin (S675) antibody. (H and I) Human MSCs were infected with lentiviruses expressing control or Usp15 shRNAs, and the resulting cells were cultured under osteoblast differentiation conditions for 21 d. (H) Alizarin red staining was performed to determine osteoblast mineralization activity. (I) After 10 d of culture, the expression of osteoblast marker genes was analyzed by RT-PCR. For each graph, P < 0.05 by one-way ANOVA. P values for Bonferroni-corrected Student’s t tests: *P < 0.05; **P < 0.01; ***; P < 0.001.

Fig. S8.

USP15 regulates the ubiquitination of β-catenin. (A) The efficiency of shRNA construct targeting human Usp15 was determined by RT-PCR and immunoblotting. (B) HEK293 cells were infected with lentiviruses expressing control or Usp15 shRNAs and then were transfected with HA-ubiquitin (HA-Ub) and Flag–β-catenin. Ubiquitinated β-catenin was immunoprecipitated with anti-Flag antibody–conjugated agarose and was immunoblotted with anti-HA antibody.

Fig. 5.

FGF2 activates MEKK2 to stabilize β-catenin in osteoblasts. (A) Primary COBs isolated from Mekk2^+/+ and Mekk2^−/− mice were stimulated with FGF2 for the indicated times, and lysates were blotted with the indicated antibodies. (B) Primary WT COBs were stimulated with FGF2 (25 ng/mL) or IGF1 (25 ng/mL) for 15 min. Cell lysates were immunoprecipitated with anti–phospho-MEKK2/3 antibody and protein G-conjugated Dynabeads and were immunoblotted with the indicated antibodies. (C) C3H10T1/2 cells were transfected with TOPflash-luciferase and Renilla, and 24 h after transfection cells were stimulated with the indicated ligands for 24 h. Luciferase activity was measured subsequently and normalized to Renilla. P < 0.05 by one-way ANOVA. P values for Bonferroni-corrected Student’s t test: **P < 0.01; ***P < 0.001. (D, Left) C3H10T1/2 cells were transfected with TOPflash-luciferase and Renilla and were stimulated with WNT3a in the absence or presence of DKK1 for 24 h. (Right) C3H10T1/2 cells were transfected with TOPflash-luciferase and Renilla along with a construct encoding MEKK2 or a vector control, and then cells were incubated with 1 μg/mL of DKK1. P = N.S. for comparison of DKK1 versus vehicle-treated cells in either the vector control or MEKK2-transfected groups. (E) A diagram of the FGF2/MEKK2/β-catenin pathway in osteoblasts.

Fig. S9.

FGF2 acts upstream of MEKK2 in osteoblasts. (A) Primary WT COBs were stimulated as indicated and immunoblotted with anti–phospho-β-catenin antibody. GAPDH was used for a loading control. (B) Primary WT COBs were stimulated as indicated. Cell lysates were analyzed using a Phos-tag gel (Wako) and immunoblotted with an anti-MEKK2 antibody. The arrow indicates phospho-MEKK2.