osteoblasts leads to osteopenia associated with increased osteoclast activity (G

lass et al. 2005; Holmen et al. 2005). Similarly, deletion of -catenin in the mes
enchymal progenitors of osteoblasts leads to a substantial reduction in the mine
ralization of both intramembranous and endochondral bones (Day et al. 2005; Hill
et al. 2005). At a cellular level, deletion of -catenin activity reduces osteobl
ast mineralization capacity in vitro. Thus, maintaining -catenin activity is crit
ical for the development and maintenance of bone mass. However, increased -caten
in activity is also seen to have deleterious effects. Activating -catenin in oste
oblasts by conditional deletion of the 3rd exon containing the inhibitory GSK-3 p
hosphorylation sites leads to both osteoblast-intrinsic osteopetrosis and the de
velopment of spontaneous acute myeloid leukemia due to altered hematopoeitic nic
he function (Glass et al. 2005; Kode et al. 2014). Taken together, these finding
s emphasize that -catenin activity in osteoblasts must be carefully tailored to m
aintain homeostasis of skeletal and hematopoeitic systems. However, how this is
accomplished in osteoblasts has yet to be fully elucidated.
-catenin activity is primarily regulated by stimulation with a subset of the WNT
family of ligands, the so-called canonical WNTs. Activation of these leads to sup
pression of a constitutively active destruction complex containing the adaptor p
roteins APC and Axin alongside the kinases GSK-3 and CK1a (Clevers and Nusse 2012
). CK1a phosphorylates -catenin S45, priming subsequent GSK-3 phosphorylation of
S33, S37 and T41. GSK-3-phosphorylated -catenin is recognized by the F-box contai
ning ubiquitin E3 ligase TrCP, leading to its ubiquitination and rapid degradation
by the 26S proteasome (Jiang and Struhl 1998). Release of this constitutive de
gradation of -catenin by WNT stimulation allows -catenin accumulation in the nucle
us and transcriptional activity.
MEKK2 (MAP3K2) is a member of the MEK Kinase (MEKK) group of MAP3Ks, and early o
verexpression studies demonstrated that MEKK2 has the capacity to activate a num
ber of downstream MAPK pathways, including ERK1/2, JNK, p38 and ERK5 (Blank et a
l. 1996; Cheng et al. 2000; Sun et al. 2003; Zhang et al. 2006). However, mice
with a germline deletion of MEKK2 display intact ERK, p38 and JNK activation in
T cells (Guo et al. 2002). Instead, MEKK2 collaborates with the highly homologo
us MEKK3 to modulate responses to TGF by phosphorylating the regulatory linker se
quences in SMAD2 and SMAD3 (Chang et al. 2011). In contrast to these findings i
n T cells, JNK activation was impaired in the MEKK2-deficient mouse embryonic fi
broblasts (MEFs) (Kesavan et al. 2004), Knockdown of MEKK2 in HEK293 cells atten
uated the activation of ERK, but not JNK or p38, in response to hyperosmotic str
ess (Maruyama et al. 2010). Thus, as has been previously observed for other MAP
3Ks in osteoblasts, the pathways engaged by MEKK2 are highly context dependent a
nd vary by both cell type and stimulus.
Here we present evidence that MEKK2 mediates an alternative pathway for the stab
ilization of -catenin in osteoblasts. MEKK2-deficient mice show a significant dec
rease in both bone mass and osteoblast activity. Through an unbiased mass spectr
ometry-based affinity purification analysis, we found that ?-catenin is a substr
ate of MEKK2, with MEKK2-mediated phosphorylation of ?-catenin acting to stabili
ze ?-catenin through the recruitment of the deubiquitinase (DUB) USP15, which ac
ts downstream of FGF2. Thus, MEKK2 mediates a novel -catenin pathway that is a ke
y determinant of bone formation in vivo.
Results
MEKK2-deficient mice display reduced bone mass and impaired osteoblast activity.
We initially identified MEKK2 as a positive regulator of osteoblast differentiat
ion in a shRNA screen of ~1,500 human kinases, phosphatases, and receptors using
a human mesenchymal stromal cell (hMSC) system (Zou et al. 2013). To explore t
he contribution of MEKK2 to osteoblast biology, expression of active MEKK2 was e
xamined in osteoblasts in vivo. Immunohistochemistry for MEKK2 and phospho-MEKK2
were performed, demonstrating that MEKK2 was both expressed and present in the
active phosphorylated state in osteoblasts in vivo (Fig. 1A,B). Ablation of MEKK
2 and phospho-MEKK2 staining in Mekk2-/- femurs, confirmed the specificity of th
e immunohistochemical stain. Next, the contribution of MEKK2 to the in vivo regu

lation of bone mass was assessed in WT and Mekk2-/- mice by CT analysis (Fig. 1C,
D). Bone mass was significantly reduced in the long bones and calvaria of Mekk2/- mice, suggesting that MEKK2 is important for both endochondral and intramembr
anous ossification. Accordingly, dynamic histomorphometry revealed a defect in o
steoblast activity, as indicated by decreased mineral apposition rate (Fig. 1E).
Bone formation rate was also reduced without a statistically significant altera
tion in osteoblast or osteoclast numbers (Fig. 1E; Supplemental Fig. 1A,B). Cons
istent with the reduction in bone formation rate, serum levels of the marker of
total osteoblast synthetic capacity, procollagen type I N-terminal peptide (P1NP
), were reduced, whereas levels of crosslinked collagen type I C-terminal telope
ptide (CTX), a marker of total osteoclast resorption activity, showed no signifi
cant difference (Fig. 1F). Moreover, expression of an osteoblast marker gene, os
teocalcin (Ocn) was markedly decreased in vivo (Fig. 1G). Though MEKK2 is expres
sed in chondrocytes of the growth plate, significant alterations in growth plate
morphology or the growth of Mekk2-/- mice were not detected (Fig. 1A). Thus, ME
KK2 regulates bone homeostasis in vivo by promoting osteoblast activity without
affecting development of osteoclasts and chondrocytes at a phenotypically detect
able level.
To examine the role of MEKK2 in osteoblast differentiation in vitro, cal
varial osteoblasts (COBs) were isolated from WT and Mekk2-/- pups and cultured u
nder osteoblast differentiation conditions. As observed by Von Kossa staining,
mineralization activity was significantly reduced in Mekk2-/- COBs (Fig. 1H), an
d Mekk2-/- COBs displayed a decrease in expression of osteoblast markers such as
Col1a1, bone sialoprotein (Bsp) and Ocn (Fig. 1I). Taken together with the ost
eoporotic phenotype of Mekk2-/- mice, these results indicate that MEKK2 is criti
cal