A Lox/CHOP‐10 crosstalk governs osteogenic and adipogenic cell fate by MSCs

Abstract Accelerated marrow adipogenesis has been associated with ageing and osteoporosis and is thought to be because of an imbalance between adipogenic and osteogenic differentiation of mesenchymal stem cell (MSCs). We have previously found that lysyl oxidase (Lox) inhibition disrupts BMP4‐induced adipocytic lineage commitment and differentiation of MSCs. In this study, we found that lox inhibition dramatically up‐regulates BMP4‐induced expression of CCAAT/enhancer binding protein (C/EBP) homologous protein 10 (CHOP‐10), which then promotes BMP4‐induced osteogenesis of MSCs both in vitro and in vivo. Specifically, Lox inhibition or CHOP‐10 up‐regulation activated Wnt/β‐catenin signalling to enhance BMP4‐induced osteogenesis, with pro‐adipogenic p38 MAPK and Smad signalling suppressed. Together, we demonstrate that Lox/CHOP‐10 crosstalk regulates BMP4‐induced osteogenic and adipogenic fate determination of MSCs, presenting a promising therapeutic target for osteoporosis and other bone diseases.

risk related to higher BMP doses, have also been reported. 12 The above studies indicate that enhancing pro-osteogenic and blocking pro-adipogenic effects with a lower dose of BMPs may be effective for clinical use of BMPs in bone disorders.
Lysyl oxidase (Lox) is synthesized and secreted as a 50 kD pro-enzyme (Pro-Lox) into the extracellular environment followed by enzymatic cleavage yielding the 32 kD mature and active Lox enzyme (Ma-Lox) and the 18 kD pro-peptide (Lox-PP). 13 The mature, 32 kD Lox catalyses the cross-linking of elastin and collagen, which is indispensable for the structural integrity and function of bone tissue. 14 We have previously reported that Lox promotes BMP2/4-induced adipocytic lineage commitment of C3H10T1/2 MSCs and that knockdown of Lox disrupts this commitment process. 10 Therefore, based on the theoretical inverse balance between adipogenic and osteogenic programming, 6 it raises a question whether Lox inhibition would promote BMP4-induced osteogenesis of MSCs. Here, we discovered that Lox inhibition promotes the osteogenic fate decisions of MSCs by modulating the expression of CCAAT/enhancer binding protein (C/EBP) and homologous protein 10 (CHOP-10), with pro-osteogenic Wnt/β-catenin signalling activation. Therefore, our studies open up opportunities for novel therapeutic intervention in bone diseases.

| Animal studies
Mice were housed in a controlled environment (12 hours light/dark cycle, 60%-70% humidity). For high-fat diet-induced obesity and osteoporosis, 6-week-old male C57BL/6J mice were fed a high-fat diet (HFD, SLACOM, 51% kcal from fat) for 10 months, with mice fed a normal chow diet (NCD, SLACOM, 10% kcal from fat) used as controls. To examine the effect on bone of Lox inhibition when combined with elevated levels of BMP4 in vivo, we used previously described Fabp4-driven, male BMP4-Tg mice, 15 which also constitutively overexpress BMP4 in bone marrow adipose tissue, providing a higher concentration of BMP4 in local bone environment. We screened the mice by PCR using primers (Fabp4-BMP4 tg: cagtgatcattgccagggagaacc; gcctcctagcaggacttggcta), control mice were non-Tg littermates. In this study, to avoid the influence of oestrogen on bone formation, 16,17 only the male mice was used. For β-aminopropionitrile (BAPN, Sigma-Aldrich, St. Louis, MO, USA) administration, 6-week-old male BMP4-Tg mice were daily injected ip with BAPN (100 mg/kg/d) or PBS control for 2 weeks. 18 Following treatment, the right femurs of mice were subjected to micro-CT analysis. For all in vivo experiments, 3-5 technical replicates were performed in each independent experiment. All animal experiments were approved by the Animal Care and Use Committee of Fudan University Shanghai Medical College and followed the National Institute of Health guidelines on the care and use of animals.

| Cell culture and induction of commitment/ differentiation
Inguinal white adipose tissue (iWAT) was obtained from 6-to 8-week-old male C57BL/6J mice. Fat pads were minced and digested for 40 minutes at 37°C (1 mg/mL Collagenase IV (Sigma-Aldrich, St. Louis, MO, USA) in DMEM). The cell suspension was passed through a 100-μm filter and centrifuged at 500 × g for 5 minutes at 4°C. The SVF pellets were resuspended in F12/DMEM with 10% foetal bovine serum (FBS). C3H10T1/2 mesenchymal stem cells were cultured in DMEM containing 10% calf serum. Primary bone marrow stromal cells (BMSCs) isolated from 6-week-old male C57BL/6J mice were cultured in a fresh alpha-minimum essential medium (α-MEM) containing 10% FBS. When BMSCs reach 80%-90% confluent, they were passaged and used in the experiments below.

| RNA interference
Stealth siRNA duplexes specific for mouse Lox were designed and synthesized by Invitrogen. The sequence for successful Lox RNAi knockdown was GCGGAUGUCAGAGACUAUGACCACA. 10 Stealth siRNA-negative control duplexes with similar GC content were used as control. SVF or C3H10T1/2 cells were transfected at 30%~50% confluence with siRNA oligonucleotides using Lipofectamine RNAi MAX according to the manufacturer's instructions (Invitrogen, Carlsbad, CA, USA).

| Western blotting
Both cell and tissue extracts were generated using lysis buffer containing 50 mmol/L Tris-HCl (pH 6.8), 2% SDS, 100 mmol/L NaF, 1 mmol/L PMSF and a phosphatase and protease inhibitor mixture

| Q-PCR
Total RNA was isolated using Trizol reagent (Thermo Fisher  Table 1.

| Oil Red O staining
C3H10T1/2 stem cells were induced to adipocyte differentiation as described above. On day 8, the cells were washed three times with PBS (phosphate-buffered saline) and then fixed for 10 minutes with 3.7% formaldehyde. Oil Red O (0.5% in isopropyl alcohol) was diluted with water (3:2), filtered through a 0.45-μm filter and incubated with the fixed cells for 1 hour at room temperature. The cells were then washed with water, and the stained fat droplets in the adipocytes were visualized by light microscopy and photographed.

| In vivo bone formation assay
C3H10T1/2 cells were treated as described above and then cultured in the above-mentioned osteogenic differentiation medium for 3 days. Approximately 4 × 10 6 cells were mixed with 40 mg hydroxyaptite-tricalcium phosphate (HA-TCP) powders (Sigma-Aldrich) and subcutaneously implanted into the armpit of 6-week-old BALB/c nude mice (n = 4). Implants were harvested after 4 weeks.

| Micro-CT
The 4% paraformaldehyde-fixed femurs were subjected to micro-CT analysis. Bone parameters of the femur metaphysis were quantified ex vivo using high-resolution, X-ray micro-computed tomography were then used to quantify bone micro-architecture.

| Statistical analysis
Results were expressed as mean ± SEM. Comparisons between groups were determined by Student's t test, or ANOVA. P < 0.05 was considered statistically significant. All experiments were repeated at least three times, and representative data are shown.

| CHOP-10 up-regulation contributes to the enhanced osteogenic lineage commitment by Lox inhibition
It has been previously reported that CHOP-10 inhibits terminal adipocyte differentiation in 3T3-L1 preadipocytes 23 and enhances osteoblastic differentiation in ST-2 stromal cells. 24

| Wnt/β-catenin signalling is involved in the enhanced osteogenesis by Lox inhibition or CHOP-10 overexpression
Next, we investigated the molecular mechanisms by which Lox and CHOP-10 regulates unbalanced adipogenesis and osteogenesis. We previously reported that BMP4 activates Smad and p38 MAPK signalling to induce Lox, which promotes adipocyte lineage commitment. 10 Here we found that Lox inhibition decreased phosphorylation of Smad1/5/8 and p38 MAPK induced by BMP4 ( Figure 3A). In contrast, the Wnt/β-catenin signalling pathway, which is central to bone development and homoeostasis, 3,25 was greatly activated by Lox inhibition ( Figure 3B). We showed that Lox inhibition increased GSK-3β phosphorylation at Ser 9 and up-regulated β-catenin, while Axin and total GSK-3β were both decreased ( Figure 3B), suggesting higher Wnt activity by Lox inhibition through negative regulation of destructive complexes. Wnt ligands promoting osteogenesis, such as Wnt1, Wnt3a, Wnt5a and Wnt10a, were also elevated by Lox inhibition in SVFs from mice inguinal adipose tissue ( Figure S4). While the Wnt inhibitor Dkk1 was down-regulated, Wnt effector Tcf4 was increased by Lox inhibition ( Figure S4). It has been previously demonstrated that IWR-1-endo, an inhibitor of Wnt pathway, induces Axin2 protein levels and promotes β-catenin degradation by stabilizing Axin-scaffolded destructive complexes. 26 Here, we found

| Lox inhibition enhances bone formation in vivo
The above results have shown that Lox inhibition enhances BMP4induced osteogenesis in vitro. As we mentioned earlier, C3H10T1/2 cell line behaves similarly to mesenchymal stem cells, 20  to undergo commitment in cell culture and then implanted s.c. into athymic mice, conditions under which they give rise to tissue that is indistinguishable from endogenous tissue. 8,9,21 As expected, the expression of osteogenic markers Col1α1 and Ocn was increased in the Lox RNAi group, while adipogenic markers PPARγ and 422/aP2 were decreased ( Figure 4A).

Next, we investigated whether Lox inhibition in conjunction with
BMP4 could enhance bone formation in vivo. Here, we use Fabp4driven male BMP4-Tg to do the experiment for the following reason.
On the one hand, BMP4 is a secretory protein and has an endocrine effect. 27 Therefore, BMP-4 overexpressed by the adipose tissues might be released into serum and reach all tissues including bone.
On the other hand, BMP4 might be also overexpressed in bone of  Figure 4E). However, decreased bone formation was observed in wild-type mice with BAPN injection ( Figure S5). Therefore, we have concluded that Lox inhibition, together with BMP4, enhances bone formation in vivo.

| Lox is up-regulated in obesity-induced bone loss
It has been demonstrated that male C57BL6 mice fed a high-fat diet was associated with a reduced rate of bone formation and turn-

| DISCUSSION
Bone is an active tissue, undergoing continuous remodelling by orchestrated processes of osteoclastic bone resorption and osteoblastic bone formation. 33 Various conditions, such as osteoporosis, A complicated and controversial relationship exists between obesity and bone health. BMP4 function in bone is complex. 36 Recently, it has been demonstrated that increased circulating levels of BMP4 in obese human subjects and diet-induced obesity (DIO) mice, 37,38 which is often accompanied with osteoporosis, suggesting a osteoporosis promotion role of BMP4. Furthermore, disruption of signalling through BMPR1A in adult osteoblasts or osteoclasts [39][40][41] increases bone mass provides evidence that alteration of the physio-  Figure S1) and in vivo ( Figure S5), which was consistent with the previous study that Lox inhibit osteoblast differentiation.
We have previously reported that BMP4 promotes adipocyte lineage commitment at a concentration that can also promote the osteogenic lineage commitment found in this study. Here, we  Figure 4). Therefore, BAPN in conjunction with BMP4 prevents adipogenesis while accelerating BMP4-induced osteogenesis, presenting a potential avenue to ameliorate the detrimental effects of BMP4 while preserving its beneficial actions. In general, we found that Lox inhibition facilitates BMP4 and CHOP-10 induced osteoblast lineage commitment by impeding adipocytic lineage commitment of MSCs.
It has been reported that Chop null mice exhibit decreased bone formation, indicating that CHOP-10 is crucial for osteoblastic function in vivo. 46 Here, we have demonstrated that CHOP-10 stimulates both basal and BMP4-induced osteogenesis ( Figure 2D-F and S3A), in agreement with previous reports using ST-2 stromal cells. 24 However, a previous study showed impaired osteoblastic function and osteopenia in transgenic mice overexpressing CHOP-10 because of increased osteoblast apoptosis. 47 We also found that CHOP-10 promotes BMP4-induced osteogenesis by hampering commitment to the adipocyte lineage of MSCs ( Figure S3B,C). Additionally, we identified a reciprocal regulation between Lox and CHOP-10 ( Figure 2G-I), providing possibilities for precisely regulating CHOP-10 so as to balance adipogenic and osteogenic differentiation. Generally, CHOP-10 is necessary for normal osteoblast differentiation, but when in excess and under specific conditions in vivo, it could be detrimental to bone homoeostasis.
The effects of Wnt/β-catenin signalling on bone mass have been well-established in both mouse models and human patients, 48 showing both pro-osteogenic and anti-adipogenic effects. 6 In addition, there is an intricate crosstalk between Wnt/β-catenin and BMPs, both functioning as master regulators of osteogenesis. BMPs, in combination with Wnt, induce MSCs to commit to osteoblastic lineage and enhance the pool and function of mature osteoblasts.
Accordingly, BMP antagonists like Gremlin bind and suppress BMP signalling and activity in osteoblastic lineage cells, which tempers Wnt signalling, 49 whereas deletion or down-regulation of Gremlin sensitizes osteoblastic cells to the actions of BMP and Wnt. 50 Our data show that Wnt/β-catenin signalling is activated to stimulate BMP4-induced osteogenesis by Lox inhibition and CHOP-10 overexpression (Figures 3 and S4). However, CHOP-10 has also been reported to inhibit Wnt/TCF signals in response to Wnt-8 in human embryonic and colon cancer cell lines. 51 This contradiction may be because of how the function and regulation of CHOP-10 varies among different cell lines and models studied.
The impact of obesity on bone health has been controversial for a long time. Obesity was traditionally viewed to be beneficial to bone health. 52 However, increasing evidence has shown that obesity could be a risk factor for osteoporosis. 52,53 Here, we showed a significant bone loss in HFD-induced obesity mice ( Figure 5), in line with reports that obesity is associated with bone loss. 52 We also discovered an inverse correlation between Lox and bone markers in both NCD and HFD mice ( Figure 5). We detected higher expression of Lox, lower expression of CHOP-10 and Ocn and decreased bone formation in HFD mice femurs ( Figure 5D), in line with reports that obesity is associated with decreased bone mass. 52 However, the mechanism underlying how Lox up-regulates CHOP-10 expression is yet to be uncovered.
In general, our study uncovered a novel Lox/CHOP-10 crosstalk governing the osteogenic and adipogenic cell decisions of MSCs. We demonstrated that Lox inhibition greatly enhances BMP4-induced osteogenesis, while suppressing the pro-adipogenic effect of BMP4 ( Figure 5E). This could advance the clinical use of BMPs in bone diseases. Our study also provides a new path towards advancing the fundamental understanding of reciprocal connections between fat and bone tissue, presenting Lox as a promising target in prevention of bone and fat diseases.