Differential Roles of MAPK Kinases MKK3 and MKK6 in Osteoclastogenesis and Bone Loss

Bone mass is maintained by osteoclasts that resorb bone and osteoblasts that promote matrix deposition and mineralization. Bone homeostasis is altered in chronic inflammation as well as in post-menopausal loss of estrogen, which favors osteoclast activity that leads to osteoporosis. The MAPK p38α is a key regulator of bone loss and p38 inhibitors preserve bone mass by inhibiting osteoclastogenesis. p38 function is regulated by two upstream MAPK kinases, namely MKK3 and MKK6. The goal of this study was to assess the effect of MKK3- or MKK6-deficiency on osteoclastogenesis in vitro and on bone loss in ovariectomy-induced osteoporosis in mice. We demonstrated that MKK3 but not MKK6, regulates osteoclast differentiation from bone marrow cells in vitro. Expression of NFATc1, a master transcription factor in osteoclastogenesis, is decreased in cells lacking MKK3 but not MKK6. Expression of osteoclast-specific genes Cathepsin K, osteoclast-associated receptor and MMP9, was inhibited in MKK3−/− cells. The effect of MKK-deficiency on ovariectomy-induced bone loss was then evaluated in female WT, MKK3−/− and MKK6−/− mice by micro-CT analysis. Bone loss was partially inhibited in MKK3−/− as well as MKK6−/− mice, despite normal osteoclastogenesis in MKK6−/− cells. This correlated with the lower osteoclast numbers in the MKK-deficient ovariectomized mice. These studies suggest that MKK3 and MKK6 differentially regulate bone loss due to estrogen withdrawal. MKK3 directly mediates osteoclastogenesis while MKK6 likely contributes to pro-inflammatory cytokine production that promotes osteoclast formation.


Introduction
Homeostatic maintenance of bone mass and strength requires the concerted actions of bone-forming osteoblasts and boneresorbing osteoclasts [1]. Imbalances in bone remodeling that favor osteoclast differentiation, survival and activation occur in chronic inflammatory diseases such as rheumatoid arthritis (RA) as well as in post-menopausal osteoporosis [2,3]. Pro-inflammatory cytokines such as IL-1, TNF, and IL-6 in chronic inflammation or the loss of estrogen in menopause regulate bone resorption by increasing the expression of Receptor Activator of NF-kB ligand (RANKL) on osteoblasts and stromal cells or by amplifying RANKL-mediated differentiation of monocytes to osteoclasts [4][5][6][7]. The interaction of RANKL with its cognate receptor RANK, expressed on osteoclasts, initiates signaling that enhances osteoclast activity, leading to accelerated bone loss and fractures [8].
RANK and RANKL belong to the TNF receptor superfamily [9,10]. The binding of RANKL to RANK recruits the adapter protein TNF receptor associated factor-6 (TRAF6) to the plasma membrane [11]. RANK, RANKL, and TRAF6 are essential for osteoclastogenesis as mice lacking these molecules have profound defects in bone resorption [11]. The RANK/TRAF6 complex activates several pathways including NF-kB and the MAPKs, JNK and p38, which induce the expression of NFATc1, considered one of the master transcription factors of osteoclastogenesis [12,13]. NFATc1, in conjunction with the transcription factors AP-1, PU.1 and micropthalmia transcription factor (MITF), is required for the expression of osteoclast specific genes such as tartrate-resistant acid phosphatase (TRAP) and Cathepsin K that promote bone resorption and demineralization [14,15].
The p38 family of MAPKs, most notably p38a, is an essential regulator of RANKL-mediated osteoclastogenesis, which makes it a potential therapeutic target for osteoporosis [16]. Several studies have shown that p38 inhibitors prevented inflammatory bone destruction in animal models and in some cases, reversed the existing disease [17][18][19]. MAPK kinases, MKK3 and MKK6, are two upstream enzymes that differentially regulate p38 function [20]. Our previous studies have shown that MKK3 is required for optimal p38a activation while MKK6 regulates pro-inflammatory cytokine production in response to LPS and TNF [21]. In addition, either MKK3-or MKK6-deficiency reduces arthritis severity and joint destruction in passive serum transfer K/BxN arthritis or collagen-induced arthritis [21][22][23]. The goal of this study was to assess the role of MKK3 and MKK6 in p38-regulated bone resorption. Therefore, we evaluated the effect of MKK3and MKK6-deficiency on osteoclastogenesis in vitro and on bone preservation in vivo using an ovariectomy model of osteoporosis. These studies show that blocking either MKK3 or MKK6 has beneficial effects on bone preservation by differentially regulating osteoclastogenesis or function.

Ethics Statement
All experimental animal protocols were reviewed and approved by the University of California San Diego Institutional Animal Care and Use Committee. This study was carried out in strict accordance with the recommendations in the Guide for the Care Figure 1. Effect of MKK-deficiency in osteoclastogenesis in vitro. (A) WT, MKK32/2 and MKK62/2 bone marrow cells were cultured to induce osteoclastogenesis and analyzed by TRAP staining (n = 3/group). The number of osteoclasts were counted in 5 pre-defined locations per well and the data are represented as mean 6 SEM. Numbers of osteoclasts were reduced in SB-treated WT and MKK32/2 cells compared with WT. Osteoclastogenesis in MKK62/2 group were comparable to WT cells. *p,0.05 (B) Phosphorylation of p38 in MKK-deficient osteoclasts. Western blot analysis showed that P-p38 levels were significantly lower in MKK32/2 but not in MKK62/2 compared with WT group. *p,0.05. doi:10.1371/journal.pone.0084818.g001 and Use of Laboratory Animals of the National Institutes of Health. All surgeries were performed under isoflurane vaporizer, and all efforts were made to minimize suffering. Mice were housed using standard housing conditions and allowed access to food and water ad libitum.

Generation of Osteoclasts
Bone marrow was isolated from WT, MKK32/2, and MKK62/2 femora and tibias. The red blood cells were lysed and the remaining cells were counted and seeded in T75 flasks at 1610 6 cells/ml. The cells were cultured for 24 h in aMEM (1X) containing GlutaMAX (Invitrogen, Carlsbad, CA) supplemented with 10% FBS, 5% penicillin-streptomycin, and 30 ng/ml of recombinant murine M-CSF (Peprotech, Rocky Hill, NJ). After 24 hours, the non-adherent cells were seeded in 24-well plates (1610 6 cells/ml) with 30 ng/ml of M-CSF and 50 ng/ml of receptor activator of NF-kB ligand (RANKL) (R&D, Minneapolis, MN). The culture medium was replaced every 72 h and supplemented with M-CSF and RANKL for 5-8 days until multinucleated osteoclasts were observed. In some experiments, cells were cultured for 5 days with 3 mM SB203580 (p38 inhibitor) or DMSO (EMD Chemicals, Gibbstown, NJ).

Osteoclast Differentiation and Activity
Tartrate-resistant acid phosphatase (TRAP) staining: Differentiated cells were stained using the Acid Phosphatase, Leukocyte (TRAP) kit (Sigma-Aldrich, St. Louis, MO). Osteoclasts were defined as multinucleated ($3 nuclei), TRAP positive cells and were counted in 5 pre-defined locations in each well. The data are represented as average count in each well.
Bone resorption assay: Bone marrow monocytes (BMM) were seeded at 3610 5 cells/well for 8 days on an Osteologic TM Bone Cell Culture System (BD Biocoat TM , Bedford, MA) (4). The cell culture system was then subjected to von Kossa staining. Eight defined areas of each well were imaged and the resorption pits were quantified using Image Pro Plus software (Media Cybernetics, Bethesda, MD). The average resorption was calculated per area for each well and the data are expressed as a percentage of resorbed area/total area.

Ovariectomy
At 16 weeks of age, female WT, MKK32/2, and MKK62/2 mice were subjected to ovariectomy or sham-surgery and allowed to recover for 6 weeks. To confirm ovariectomy, leutinizing hormone (LH) was measured in serum by sandwich ELISA (Ligand assay and analysis core lab at the University of Virginia).

Micro-Computed Tomography (micro-CT) and Bone Histomorphometry
Micro-CT images of the tibias were acquired as described previously [13]. Briefly, the tibias were harvested, fixed in 10% neutral buffered formalin for 24 h and stored in 70% ethanol. All quantifications were performed with digital image analysis (OsteoMeasure; Osteometrics, Atlanta, GA). The following parameters were measured: fraction of bone volume of the total Histomorphometric analysis was performed on methacrylateembedded undecalcified plastic sections of bone and were stained using von Kossa's stain and TRAP stain. To measure bone formation rate (BFR), all mice were injected with calcein green solution (30 mg/kg, Sigma, MO) at 10 days and 2 days prior to harvesting. The tibias were prepared as described above and the measurements were performed using an image analysis system (OsteoMeasure) and the bone formation rate was calculated and the data are shown as mm 3 /mm 2 /year.

Gene Expression Analysis
Total RNA was isolated from cells and 500 ng RNA was used for first-strand cDNA synthesis (Life Technologies, NY). Quantitative real-time PCR was performed using Taqman gene expression assay primer/probe sets and GeneAmp 7300 system (Applied Biosystems, Foster City, CA) [24]. Ct values were normalized to hypoxanthine-guanine phosphoribosyl transferase 1 (Hprt1) using the DDCt method.

Statistical Analysis
Comparisons between WT, MKK32/2, and MKK62/2 cells were analyzed by one-way ANOVA and Tukey's or Kruskal-Wallis multiple comparison tests, unless otherwise stated. Ovariectomy data were analyzed by two-way ANOVA and Tukey's multiple comparison test. Data were analyzed using GraphPad

Effect of MKK3-and MKK6-deficiency on Osteoclastogenesis in vitro
To evaluate the role of MKK3 and MKK6 in osteoclast differentiation, we cultured WT and MKK3-and MKK6- Osteoblast numbers were unaltered after OVX in the three groups. (C) Bone formation rate was evaluated by injecting mice with calcein green solution at 10 days and 2 days prior to harvesting. The tibias were evaluated for bone formation rate (n = 3/group). Bone formation rate was unaltered by MKK-deficiency in sham-or OVX-treated mice compared with WT controls. *p,0.05. doi:10.1371/journal.pone.0084818.g005 deficient bone marrow cells in the presence of M-CSF and RANKL. WT cells were cultured in the presence or absence of the p38 inhibitor SB203580 (SB). Osteoclastogenesis was significantly decreased in MKK32/2 (p = 0.0012) and SB-treated WT group (p,0.0001) compared with WT control ( Figure 1A, n = 3 mice/ group). MKK32/2 osteoclast numbers were also significantly lower than MKK62/2 cells (p = 0.0012). Surprisingly, osteoclast numbers in MKK62/2 group were comparable to the WT control. We next determined the level of phosphorylation of p38 (P-p38) in cultured WT, MKK32/2 and MKK62/2 osteoclasts by Western blot analysis. MKK32/2 cells had significantly reduced P-p38 levels compared with WT or MKK62/2 ( Figure 1B, p = 0.0009 and p = 0.005, respectively). These data show that MKK3, but not MKK6, is an important regulator of osteoclast differentiation in vitro.

Bone Resorption Activity in MKK3-and MKK6-deficient Osteoclasts
Bone resorption of MKK-deficient osteoclasts was then determined by their ability to demineralize matrix and form resorption pits. Resorption was visualized using Von Kossa staining ( Figure 2). MKK32/2 osteoclasts had a significantly lower number of resorptive pits than WT (p = 0.0061) or MKK62/2 cells (p = 0.027, Kruskal-Wallis test), suggesting that MKK3 is a key regulator of osteoclast differentiation in vitro.

Gene Expression in MKK32/2 and MKK62/2 Osteoclasts
Osteoclast differentiation requires the expression of a number of well-characterized markers such as NFATc1 and Cathepsin K, which regulate osteoclast function. To determine the effect of MKK-deficiency on these markers, we compared gene expression in WT, MKK32/2, and MKK62/2 osteoclasts by qPCR ( Figure 3). Expression of Cathepsin K and osteoclast-associated receptor (OSCAR) was significantly reduced in MKK32/2 osteoclasts compared with WT (p,0.001, respectively) or MKK62/2 cells (p,0.001, respectively). Similarly, NFATc1 expression was significantly reduced in MKK32/2 cells compared with WT (p = 0.004) or MKK62/2 osteoclasts (p = 0.009). Expression of c-fos and MMP9 (gelatinase B) was inhibited only in MKK32/2 cells (p = 0.037 and p = 0.031, respectively, t-test) compared with WT. In contrast, RANK expression was not decreased in MKK-deficient cells. These data show that in vitro expression of osteoclast-specific markers is inhibited in the absence of MKK3.

Ovariectomy-induced Bone Loss in MKK3-and MKK6deficient Mice
Loss of estrogen in ovariectomized (OVX) mice is associated with increased bone turnover with bone resorption exceeding formation, leading to a net bone loss. Thus far, our data have shown that MKK3-deficiency decreases osteoclastogenesis and bone resorption in vitro, suggesting that a lack of MKK3 might preserve bone in vivo. We therefore examined the effect of MKKdeficiency on bone preservation after ovariectomy. The efficiency of the surgery was confirmed by showing increased levels of luteinizing hormone (LH) in post-surgical sera ( Figure S1). Trabecular bone density at the proximal tibia was evaluated in OVX or sham control mice using micro-CT ( Figure 4A, n = 5/ group).
Micro-CT analysis showed significantly higher bone volume (BV/TV) and trabecular number (Tb.N.) in sham MKK32/2 mice compared with WT sham control ( Figure 4B, BV: p = 0.003, Tb.N: p = 0.006). Surprisingly, sham MKK62/2 mice had a significantly elevated bone volume and Tb.N. compared with sham WT control (BV: p,0.0001, Tb.N: p,0.0001). Together, these data suggest that both MKK3 and MKK6 play a role in physiological bone homeostasis. While a significant bone loss was observed after OVX in all three genotypes, bone volume and Tb.N were modestly elevated in OVX MKK32/2 and significantly increased in MKK62/2 mice compared to the OVX WT control. These results suggest that MKK62/2 mice were partially protected from OVX-mediated bone loss, maintaining a bone volume similar to sham-treated WT controls. Trabecular thickness was virtually unchanged in OVX or sham control WT and MKK2/2 mice. In contrast, trabecular separation (Tb. Sp) was significantly higher after OVX in WT (p = 0.002) and MKK32/2 (p = 0.005) but not in MKK62/2, compared with their respective sham controls. Trabecular separation was significantly lower in OVX MKK32/2 (p = 0.003) and MKK62/2 (p,0.0001) compared with OVX WT mice, indicating a higher bone mass in MKK-deficient mice. Together, the in vivo data demonstrate that MKK3-and MKK6deficiency are associated with a high bone mass phenotype, which is partially maintained after OVX.

Histomorphometric Analysis of Bone in Ovariectomized MKK3-and MKK6-deficient Mice
To further evaluate the effect of MKK3-and MKK6-deficiency in estrogen-deficient mice, undecalcified tibial bone sections from OVX and sham control mice were stained and subjected to histomorphometric analysis ( Figure 5A, B). Consistent with the micro-CT data, bone volume (%BV/TV) was significantly higher in sham-and OVX-treated MKK32/2 (p,0.05 for each) and MKK62/2 mice compared to respective WT controls (p,0.001, for each). OVX led to a reduction of bone volume in the WT, MKK32/2 and MKK62/2 groups (n = 5/treatment, p,0.001, respectively). Nonetheless, bone volume and trabecular number were at least partially maintained in MKK-mutants, indicating a similar bone volume after OVX as sham-treated WT mice. Trabecular thickness (Tb.Th) was similar in WT, MKK32/2, and MKK62/2 mice, both after sham-treatment and OVX.
TRAP staining showed higher osteoclast numbers (NOc/bone perimeter) in OVX WT mice compared with sham-treated WT mice (p,0.001). However, MKK3-and MKK6-deficient mice had significantly lower osteoclast numbers compared with shamtreated or OVX WT mice (MKK32/2: p,0.01, MKK62/ 2:p,0.001). Interestingly, the number of osteoblasts (NOb) and bone formation rate was not altered in MKK3 and MKK6 mutants. These results suggest that MKK-deficiency inhibits osteoclast formation in vivo but does not alter numbers of osteoblasts or bone formation ( Figure 5C). Together, these data show that MKK-deficiency induced high bone mass by lowering osteoclast numbers in vivo.

Discussion
Osteoclast differentiation and activation is regulated by p38 MAPK, at least in part by mediating the effects of proinflammatory cytokines [25,26]. Several studies have shown that p38 inhibitors protect against bone resorption in inflammation, possibly through the direct effects on osteoclasts as well as through inhibition of p38-induced cytokine expression. For instance, p38a activation is essential for TNF-mediated inflammatory bone loss, and a selective p38a inhibitor prevented trabecular bone loss and increased cortical bone area in ovariectomized rats [13,27]. Our previous studies demonstrated clinical benefits of MKK3-and MKK6-deficiency in acute and chronic models of arthritis [21][22][23]. Inflammation-induced bone loss in the joints was markedly reduced in the absence of MKK3 or MKK6, with a corresponding decrease in serum IL-6 and synovial MMP [22]. In the present study, we extend our observations on bone protection by evaluating the effect of MKK3-and MKK6-deficiency on osteoclastogenesis and bone loss due to estrogen withdrawal in ovariectomized mice.
Our data show that MKK3 and MKK6 differentially regulate osteoclastogenesis. MKK3 is required for optimal osteoclastogenesis and function. MKK6 was unable to effectively compensate for the absence of MKK3 in the osteoclasts. MKK6-deficiency, on the other hand, did not affect osteoclastogenesis in vitro. Consistent with our previous studies, P-p38 levels were markedly reduced in MKK32/2 but normal in MKK62/2 osteoclasts [21]. Osteoclast numbers and resorption activity correlated with p38 activation in MKK32/2 and MKK62/2 cells and are in agreement with previous studies [16]. Expression of NFATc1 and c-fos were reduced in MKK32/2 osteoclasts, indicating that p38 function is necessary for the induction of these transcription factors that are involved in osteoclast differentiation [28]. Cathepsin K, OSCAR, and matrix metalloproteinase 9 (MMP9) expression was reduced in MKK3-deficient osteoclasts. This is consistent with the requirement of NFATc1 in the transcription of these osteoclastspecific markers [15,29,30]. Together, these results indicate that MKK3, rather than MKK6, is a principal regulator of RANKLinduced signaling in osteoclastogenesis in vitro. These data are in accordance with other studies showing that p38 regulates the early phases of osteoclast differentiation [31,32]. However, those studies suggested that MKK6, but not MKK3, is necessary for osteoclastogenesis in vitro, where it enhances cell survival but not bone resorption. The differences might be partly due to differences in the methodology, such as the efficiency of inhibiting MKK3 and MKK6 using viral transduction of dominant-negative MKK3 and MKK6 constructs compared with using bone marrow-derived osteoclasts of MKK32/2 and MKK62/2 mice.
Osteoclastogenesis in vivo is regulated by a number of factors such as vitamin D, parathyroid hormone, prostaglandins and most notably, estrogen [33][34][35][36]. Estrogen is a key modulator of both the immune system and bone homeostasis [37]. It regulates bone remodeling by decreasing osteoclast life span by promoting apoptosis [38,39]. Estrogen inhibits expression of RANKL on osteoblasts, stromal cells, and lymphocytes, and increases production of osteoprotegerin (OPG), an inhibitor of RANK signaling [1]. In menopause, the loss of estrogen is associated with elevated levels of IL-1, IL-6, TNF, and M-CSF, which support bone resorption [40,41]. Estrogen deficiency also triggers the differentiation of Th17 cells that produce the pro-inflammatory cytokine IL-17 [42]. IL-17 induces osteoblast and stromal cell production of IL-6, and TNF, which induce RANKL expression, accelerate osteoclastogenesis and lead to osteoporosis [43].
In this study, we showed that both MKK3-and MKK6deficiency is associated with high bone mass and at least partially inhibited bone loss after ovariectomy, even though our in vitro studies did not show major effects of MKK6 on in vitro osteoclastogenesis. Histomorphometric data demonstrated that high bone mass in MKK3 and MKK6-deficiency correlates with reduced osteoclast numbers. The decrease in osteoclast numbers in MKK6-deficient mice could be due to decreased p38-regulated cytokine production, such as IL-6, since a direct effect on osteoclasts was not observed. The bone protection is consistent with our previous studies showing reduced inflammation and joint destruction in MKK32/2 and MKK62/2 mice, where a direct effect on osteoclasts for MKK3 and an indirect effect by blocking cytokines for MKK6 were found [22].
The study also found no major effect on osteoblast number and bone formation rate in MKK3-and MKK6-deficient mice. The regulation of bone homeostasis and skeletal formation was recently evaluated in MKK3-and MKK6-deficient mice [44]. That study showed that both MKK32/2 and MKK62/2 mice had reduced bone mass resulting from defective osteoblast differentiation. The reason for the discrepancy is unclear but might be partly due to strain variations in mice and the different ages of the mice used, 5 weeks compared with 20 weeks for our study.
In conclusion, MKK3 and MKK6 affect bone mass through regulation of osteoclast numbers in vivo. MKK6 and to a lesser extent MKK3 mitigate bone loss in conjunction with estrogen deficiency. While MKK3 likely mediates direct effects on osteoclasts through RANKL signaling, MKK6 likely contributes to the production of pro-inflammatory cytokines such as IL-6 and IL-17 that increase RANKL expression. Targeting upstream p38 kinases, such as MKK3 and MKK6 might modulate bone loss during inflammation and estrogen deficiency.