Soluble Sema4D cleaved from osteoclast precursors by TACE suppresses osteoblastogenesis

Abstract Bone remodelling is mediated by orchestrated communication between osteoclasts and osteoblasts which, in part, is regulated by coupling and anti‐coupling factors. Amongst formally known anti‐coupling factors, Semaphorin 4D (Sema4D), produced by osteoclasts, plays a key role in downmodulating osteoblastogenesis. Sema4D is produced in both membrane‐bound and soluble forms; however, the mechanism responsible for producing sSema4D from osteoclasts is unknown. Sema4D, TACE and MT1‐MMP are all expressed on the surface of RANKL‐primed osteoclast precursors. However, only Sema4D and TACE were colocalized, not Sema4D and MT1‐MMP. When TACE and MT1‐MMP were either chemically inhibited or suppressed by siRNA, TACE was found to be more engaged in shedding Sema4D. Anti‐TACE‐mAb inhibited sSema4D release from osteoclast precursors by ~90%. Supernatant collected from osteoclast precursors (OC‐sup) suppressed osteoblastogenesis from MC3T3‐E1 cells, as measured by alkaline phosphatase activity, but OC‐sup harvested from the osteoclast precursors treated with anti‐TACE‐mAb restored osteoblastogenesis activity in a manner that compensates for diminished sSema4D. Finally, systemic administration of anti‐TACE‐mAb downregulated the generation of sSema4D in the mouse model of critical‐sized bone defect, whereas local injection of recombinant sSema4D to anti‐TACE‐mAb‐treated defect upregulated local osteoblastogenesis. Therefore, a novel pathway is proposed whereby TACE‐mediated shedding of Sema4D expressed on the osteoclast precursors generates functionally active sSema4D to suppress osteoblastogenesis.

produced in both membrane-bound and soluble forms; however, the mechanism responsible for producing sSema4D from osteoclasts is unknown. Sema4D, TACE and MT1-MMP are all expressed on the surface of RANKL-primed osteoclast precursors.
When TACE and MT1-MMP were either chemically inhibited or suppressed by siRNA, TACE was found to be more engaged in shedding Sema4D. Anti-TACE-mAb inhibited sSema4D release from osteoclast precursors by ~90%. Supernatant collected from osteoclast precursors (OC-sup) suppressed osteoblastogenesis from MC3T3-E1 cells, as measured by alkaline phosphatase activity, but OC-sup harvested from the osteoclast precursors treated with anti-TACE-mAb restored osteoblastogenesis activity in a manner that compensates for diminished sSema4D. Finally, systemic administration of anti-TACE-mAb downregulated the generation of sSema4D in the mouse model of critical-sized bone defect, whereas local injection of recombinant sSema4D to anti-TACE-mAb-treated defect upregulated local osteoblastogenesis. Therefore, a

| INTRODUC TI ON
Semaphorins have been accepted as multipotent factors engaged in several physiological cellular events, such as neuronal growth regulation, immune regulation, epithelial remodelling and angiogenesis. 1,2 However, their functional roles in bone regeneration were only identified in 2011. 3 Amongst the semaphorin family of some 20 sibling molecules, semaphorin 4D (Sema4D) was discovered to suppress the bone anabolic process. 3 More specifically, Sema4D, produced by osteoclasts, downregulates IGF-1-mediated osteoblast activation by binding to plexin B1, a Sema4D receptor. In contrast, anti-Sema4D-mAb could regain lost bone contents in osteoporosisinduced ovariectomized mice, 3 suggesting the translational value of targeting Sema4D in bone lytic disorders.
Sema4D was initially identified as a type-1 membrane protein expressed on immune cells, 4 but its soluble form was also detected in plasma. It is also reported that the soluble form of Sema4D (sSe-ma4D) is secreted by γδ T and cancer cells, 5,6 as well as platelets and osteoclasts. 7 Since no studies have reported spliced form of Sema4D that can be secreted, it is plausible that detected sSema4D results from the cleavage of membrane-bound Sema4D expressed on the cell surface. Important to this study, an elevated level of sSe-ma4D was detected in the gingival crevice fluid (GCF) and serum of patients with inflammatory bone lytic diseases of periodontitis 8 and rheumatoid arthritis, 9 respectively. According to Li Zhu et al., the extracellular domain of Sema4D is shed from the platelet surface by the ectoenzyme TACE (TNF-converting enzyme, also known as ADAM17). 7 On the other hand, it was also reported that Sema4D expressed on the cellular membrane of cancer cells is shed by Membrane Type-1 Matrix Metalloproteinase (MT1-MMP) which has been implicated in cancer metastasis. 6,10 However, the mechanism responsible for producing sSema4D from osteoclasts is unknown.
Therefore, this study aimed to identify the enzyme responsible for shedding Sema4D expressed on osteoclasts and evaluate the functionality of such sSema4D released from osteoclasts.

| Production of sSema4D by RANKL-primed osteoclasts
Upon stimulation of bone marrow mononuclear cells (BMMC: C57BL/6 mouse) with recombinant mouse M-CSF and RANKL (20 ng/ml and 50 ng/ml, respectively, both from BioLegend), Sema4D was expressed on osteoclast precursors in its membrane-bound form (150 kD) and soluble form (120 kD) in W-blot ( Figure 1A). The release of sSema4D in the culture supernatant of RANKL-stimulated BMMCs (osteoclast [OC]-sup) showed its highest level at 48 h, followed by a decline at 72 h. Importantly, the release of sSema4D from RANKL-stimulated BMMCs occurred during the early stage of osteoclast differentiation (up to 72 h) before the start of cell-cell fusion. novel pathway is proposed whereby TACE-mediated shedding of Sema4D expressed on the osteoclast precursors generates functionally active sSema4D to suppress osteoblastogenesis.

K E Y W O R D S
membrane type-1 matrix metalloproteinase, osteoblastogenesis, osteoclasts, semaphorin 4D, sheddase, tumor necrosis factor alpha converting enzyme F I G U R E 1 Sema4D is cleaved by osteoclast TACE to become a soluble form. (A) Upon stimulation of bone marrow mononuclear cells (BMMC: C57BL/6 mouse) with recombinant mouse RANKL (50 ng/ml, BioLegend), Sema4D was expressed on osteoclast cells in its membrane-bound form (150 kD) and soluble form (120 kD) as determined by Western blot. (B) The level of soluble Sema4D showed a peak at 48 h after stimulation with RANKL, diminishing at 72 h. According to sSema4D ELISA (RayBiotech), the amounts of sSema4D detected in the culture supernatant of RANKL-stimulated BMMCs (OC-sup) were proportional to those detected in Western blot. (C) Osteoblastogenesis was induced by the stimulation of MC3T3-E1 cells with ascorbic acid (Vit C) and β-glycerophosphate (β-GP) in the presence or absence of the OC-sup harvested at 48 h with or without anti-Sema4D-mAb or control mAb. After 7-day culture of MC3T3-E1, the alkaline phosphatase (ALP) activity was detected to be reduced by the incubation with OC-sup, which was abrogated by anti-Sema4D-mAb, but not by control mAb. For the detection of mineral deposition stained by Alizarin red (AR), MC-3 T3-E1 cultures were incubated for 3 weeks. Results showed a trend similar to that detected in APL staining. (D) Upon stimulation of BMMCs with RANKL, elevated expression of all three mRNAs for Sema4D, TACE and MT1-MMP was detected by qPCR. (E) In a fluorescence confocal microscopy of osteoclasts stimulated by RANKL, colocalization of Sema4D and TACE was observed in the RANKL-stimulated BMMCs, whereas no colocalization was detected between MT1-MMP and Sema4D. (F) The production of sSema4D from RANKL-stimulated BMMCs was prominently suppressed by TAPI2, an inhibitor of ADAM 8, 10, 12 and TACE, whereas Prinomastat, an inhibitor of MMP-2, MMP-9 and MT1-MMP, showed a reduced capacity to suppress the production of sSema4D, as measured by Western blot (120KD band). (G) The ELISA measurements of TAPI2 and Prinomastat sSema4D in RANKL-stimulated BMMC showed that TAPI2 significantly inhibited sSEMA4D. p < 0.01. (H) The sheddase activity of TACE expressed on RANKL-stimulated BMMCs was confirmed by InnoZyme™ TACE Activity Kit (Sigma-Aldrich). *p < 0.05 and **p < 0.01, by one-way anova, followed by Tukey's post hoc test

| Cellular localization and sheddase activity by TACE and MT1-MMP expressed by RANKL-primed osteoclasts
To explain the generation of sSema4D from its membrane-bound form, two ectoenzymes that are TACE and MT1-MMP have been reported to shed membrane-bound Sema4D. 6,7 Upon stimulation of BMMCs with RANKL, elevated expression of all three mRNAs for Sema4D, TACE and MT1-MMP was detected by qPCR ( Figure 1D).
According to fluorescence confocal microscopy, colocalization of showed a lesser capacity to suppress the production of sSema4D ( Figure 1F,G). The sheddase activity of TACE was elevated in the RANKL-stimulated BMMCs stimulated with RANKL which peaked at 48 h ( Figure 1H). These results suggested that TACE, but not MT1-MMP, may be engaged in the shedding of membrane-bound Sema4D expressed on RANKL-stimulated BMMCs.

| siRNA-based loss-of-function assay to evaluate the role of the sheddases TACE and MT1-MMP in generating sSema4D
Since two chemical inhibitors used for inactivation of TACE

| Effects of anti-TACE-mAb on sSema4D release from RANKL-stimulated BMMCs
We have demonstrated that TACE can shed the membrane-RANKL expressed on T cells and B cells using a polyclonal rabbit anti-human TACE neutralizing antibody. 15 In the present study, anti-mouse TACE-neutralizing-mAb (anti-TACE-mAb, mouse IgG1) was generated by immunizing a peptide antigen specific to mouse TACE (NTCKLLVVADHRFYKMG). The incubation of RANKL-stimulated BMMCs with anti-TACE-mAb diminished the expression of TACE on the surface of osteoclast precursors, compared to that incubated with control-mAb, as monitored by fluorescence confocal microscopy ( Figure 2E). Anti-TACE-mAb also significantly suppressed TACE activity, as well as sSema4D in the OC-sup ( Figure 2F,G).  Figure S1D). These results indicated the possible roles of TACE in generating sSema4D, leading to its downregulation of osteoblastogenesis at the critical-sized bone defect. However, as noted above, anti-TACE-mAb had no effect on mineral deposition at the bone defect site. One plausible explanation could be diminished generation of local sSema4D, thus failing to induce sufficient angiogenesis otherwise required to supply nutrients, calcium and phosphates to osteoblasts. A strong body of evidence supports that a timely coordinated angiogenesis is a crucial event for the successful bone repair 17 and that multifactorial Sema4D possesses angiogenic activity. 18 We have previously reported that TACE produced by lymphocytes in periodontitis sheds membrane-bound RANKL to generate sRANKL which, in turn, promotes osteoclastogenesis. 15 In the present study, we took a step further to prove that TACE has an additional role in the bone remodelling process that of generating sSema4D to inhibit osteoblastogenesis. In conclusion, our study demonstrated that (1)

CO N FLI C T O F I NTE R E S T
The authors declare no conflicts of interest associated with this study.

DATA AVA I L A B I LT Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding authors upon reasonable request.