Targeted Overexpression of Osteoactivin in Cells of Osteoclastic Lineage Promotes Osteoclastic Resorption and Bone Loss in Mice

This study sought to test whether targeted overexpression of osteoactivin (OA) in cells of osteoclastic lineage, using the tartrate-resistant acid phosphase (TRAP) exon 1B/C promoter to drive OA expression, would increase bone resorption and bone loss in vivo. OA transgenic osteoclasts showed ∼2-fold increases in OA mRNA and proteins compared wild-type (WT) osteoclasts. However, the OA expression in transgenic osteoblasts was not different. At 4, 8, and 15.3 week-old, transgenic mice showed significant bone loss determined by pQCT and confirmed by μ-CT. In vitro, transgenic osteoclasts were twice as large, had twice as much TRAP activity, resorbed twice as much bone matrix, and expressed twice as much osteoclastic genes (MMP9, calciton receptor, and ADAM12), as WT osteoclasts. The siRNA-mediated suppression of OA expression in RAW264.7-derived osteoclasts reduced cell size and osteoclastic gene expression. Bone histomorphometry revealed that transgenic mice had more osteoclasts and osteoclast surface. Plasma c-telopeptide (a resorption biomarker) measurements confirmed an increase in bone resorption in transgenic mice in vivo. In contrast, histomorphometric bone formation parameters and plasma levels of bone formation biomarkers (osteocalcin and pro-collagen type I N-terminal peptide) were not different between transgenic mice and WT littermates, indicating the lack of bone formation effects. In conclusion, this study provides compelling in vivo evidence that osteoclast-derived OA is a novel stimulator of osteoclast activity and bone resorption.


Introduction
Osteoactivin (OA), also known as Dchil (dendritic cell-associated, heparin sulfate proteoglycan-dependent integrin ligand), Gpnmb (glycoprotein non-metastatic melanomal protein B), or Hgfin (hematopoietic growth factor-inducible neurokinin 1), is a type 1 transmembrane glycoprotein [1]. The OA gene, located on human chromosome 7p15.1 or on mouse chromosome 6 [2], has 11 exons with an open reading frame of 1,716 bp that encodes a protein of 572 amino acid residues. It has 13 N-linked glycosylation sites, a heparin binding domain, an integrin-recognition RGD (Arg-Gly-Asp) motif in both its extracellular and intracellular domains, and a polycystic kidney disease (PKD) sequence [1,3]. OA may exist as a 65-kD unglycosylated cellular protein or as multiple glycosylated proteins with molecular size varying from 80-kD to 139-kD [4]. The transmembrane OAs can be proteolytically cleaved at their juxtamembrane region by extracellular proteases, such as ADAMs [5] and MMPs [6], in a process called ectodomain shedding, which results in detachment and release of the extracellular domain to act as cytokines or growth factors [7].
OA is expressed in a wide array of tissues and plays regulatory roles in various cellular functions. Accordingly, OA plays a key regulatory function in endothelial cell adhesion that involves integrin binding [1]. High expression levels of OA protein can be found in the nervous system, basal layer of the skin, germinal cells of hair follicles, and the forming nephrons of the kidney of late mouse embryos [2]. In immune cells, OA expression is associated with cell differentiation, as its expression was detected in differentiated macrophages, lymphocytes, and dendritic cells, but undetectable in proliferating hematopoietic progenitors [8]. OA plays a negative regulator role in activation of macrophages [9] and T lymphocytes [10,11], and functions as an inhibitory immune receptor [10]. In addition, OA is implicated in development of retinal pigment epithelium and iris [12]. OA upregulates expression of matrix metalloproteinase (MMP)-3 and -9 in the infiltrating fibroblasts into denervated skeletal muscle [13]. Overexpression of OA in transgenic mice protects skeletal muscle from severe degeneration and fibrosis caused by long-term denervation [14] and reduces hepatic fibrosis in the injured or diseased liver [15]. The ADAM10-released OA showed potent angiogenic properties [5]. Because of its suggestive functions in cell adhesion, migration, and differentiation in various cell types and tissues, OA has been implicated in physiological and pathophysiological cascades of tissue injury and repair [16]. In addition to its diverse roles in normal cells and tissues, aberrant OA expression is linked to various pathological disorders such as glaucoma [17], kidney disease [18], osteoarthritis [19], and several types of cancer, including: uveal melanoma [20], glioma [21], hepatocellular carcinoma [22], and cutaneous melanoma [23].
In bone, OA was initially discovered by mRNA differential display as a novel osteoblast-specific protein [3]. It was reported that expression of OA is associated temporally with differentiation and maturation of primary rat osteoblasts in vitro [2]. Suppressing the OA functional activity with a blocking antibody reduced differentiation and bone formation activity of rat osteoblasts without affecting their proliferation and viability [24]. Bone marrow mesenchymal cells derived from mutant mice expressing a truncated OA lacking a large portion of the intracellular domain were reported to be defective in their ability to differentiate into osteoblasts in vitro [25]. These findings suggest that osteoblastderived OA has a regulatory role in osteoblast differentiation and bone formation. OA expression in osteoblasts is up-regulated by bone morphogenetic protein (BMP)-2 [26] through up-regulation of binding of homeodomain transcription factors, Dlx3 and Dlx5 to the OA promoter [27]. OA appears to be a key mediator of the BMP-2-induced osteoblast differentiation [26].
Studies from our group [4] and others [28] have demonstrated that mature osteoclasts also express high levels of OA. In fact, the expression level of OA in mature mouse osteoclasts was several-fold in magnitude higher than that in mouse osteoblasts and stromal cells [4,28], indicating that expression of OA in bone is not restricted to osteoblasts. There is in vitro evidence that osteoclastderived OA has a stimulatory role in osteoclast maturation and bone resorption [4]. However, the in vivo function of osteoclastderived OA in bone has not been investigated. The objective of this study was to determine whether osteoclast-derived OA has a regulatory role in bone resorption in vivo by determining the effects of targeted overexpression of OA in cells of osteoclastic lineage with the tartrate-resistant acid phosphatase (TRAP) exon 1B/C promoter to drive transgenic OA expression in bone in vivo. We chose the TRAP-1B/C promoter over the TRAP-1C promoter, which we used previously to generate transgenic mice with targeted overexpression of an osteoclast-specific protein-tyrosine phosphatase in osteoclasts [29], because the TRAP-1B/C promoter is a stronger promoter than the TRAP-1C promoter in vivo [30], although the TRAP-1B/C promoter is less specific for osteoclasts than the TRAP-1C promoter [31].

Results
Generation of transgenic (Tg) mice with OA overexpression in osteoclastic cells OA-Tg founder mice were created by injecting the linearized OA-Tg targeting expression plasmid ( Fig. 1) into C57BL/6 (B6) ova. The efficiency of generating Tg mice with B6 ova was low, and only a total of 12 transgenic pups were obtained from two separate injection procedures. Of these pups, three were shown to be OA transgenic by a PCR-based genotyping assay. Additional genotyping assays revealed that one of these transgenic pups expressed a truncated form of OA lacking most of the intracellular domain and was euthanized. The other two pups were confirmed to be transgenic mice overexpressing the full-length OA in bone. However, one of the OA-Tg lines was lost during breeding. Thus, all subsequent studies were performed in the remaining OA-Tg line. The transgenic founder was crossed with B6 mice to yield an OA-Tg mouse colony of pure B6 genetic background.
Transgenic mice with overexpression of OA in cells of osteoclast lineage (OA-Tg mice) To confirm that osteoclasts of OA-Tg mice indeed overexpressed OA, the relative levels of OA mRNA (by real-time RT-PCR) and cellular OA protein (by Western immunoblots) were measured in marrow-derived osteoclasts of 4-week-old female OA-Tg mice and WT littermates. Fig. 2A shows that the two predominant glycosylated OA protein species (i.e., ,130 kD and ,80 kD), which are the functionally active species [2,4,6], in the OA-Tg osteoclasts were each 2-fold of those in WT osteoclasts. We did not determine cellular levels of the unglycosylated ,50-60 kD OA protein species, because the polyclonal anti-OA antibody recognizes primarily glycosylated species and does not detect unglycosylated OA. Similarly, real-time RT-PCR analysis reveals that the OA mRNA level in OA-Tg osteoclasts was significantly higher (2.4060.51-fold, P = 0.05) than that in WT osteoclasts (Fig. 2C). Osteoblasts and marrow stromal cells also expressed TRAP [4]. To ensure that the use of TRAP-1B/C promoter to drive OA expression would not also result in OA overexpression in osteoblasts, we measured the OA mRNA and protein levels in OA-Tg osteoblasts derived from OA-Tg mice. The level of the predominant glycosylated OA protein, ,80 kD, in OA-Tg osteoblasts was not significantly different from that in WT osteoblasts (Fig. 2B), and the OA mRNA level in OA-Tg osteoblasts was also not significantly different (0.6860.15-fold, p = N.S.) from those in WT osteoblasts (Fig. 2C). Thus, the Tg mice generated with the OA expression construct driven by the TRAP-1B/C promoter did not result in significant OA overexpression in osteoblasts. Fig. 2 also confirms our previous findings that the OA protein level in osteoclasts was at least 2-fold higher than that in osteoblasts [4], as the OA/actin ratio in WT osteoclasts was ,2fold of that in WT osteoblasts.
pQCT analyses of bone phenotype of OA-Tg mice The body weight of OA-Tg mice of either sex at 8 weeks of age was not different significantly from corresponding WT littermates (Table 1). To evaluate the consequence of OA overexpression in cells of osteoclastic lineage in the bone in vivo, we measured (with pQCT) bone area and bone mineral density (BMD) parameters at sites that are rich in trabecular bone (metaphyses) and cortical bone (mid-diaphysis), respectively, in femurs of 8-week-old young adult OA-Tg mice of both sexes ( Table 1). The femur length of female OA-Tg mice was 2.6% shorter (p,0.01) than female WT littermates. In metaphysis at the secondary spongiosia, the Tg mice had significantly 3-6% lower total, sub-cortical and cortical BMD and area (but not trabecular BMD and area) than WT littermates in either sex. The cortical thickness at this bone site of OA-Tg mice of both sexes was also significantly smaller by 8% than WT littermates. At mid-diaphysis, the total, subcortical, and cortical BMD, and total bone area of OA-Tg mice were similarly 3-6% less compared to those of WT mice. The cortical thickness at the mid-shaft of the femur of OA-Tg mice was smaller by 5-6% than those of WT littermates. Accordingly, targeted overexpression of OA in cells of osteoclast lineage led to significantly lower bone mass and BMD at the cortical bone site. There were no apparent sex-related differences in the low bone mass phenotype in these OA-Tg mice.
It has previously been reported that there were more abundant TRAP-positive osteoclasts on bone surface of young growing mice (e.g., 4 and 7 weeks of age) than fully grown adult mice (e.g., 14 weeks of age) [32]. Thus, it is possible that the basal TRAP-1B/C promoter activity in mice may vary with age, which may in turn lead to an age-dependent variation in OA overexpression levels in OA-Tg mice. In this regard, Fig. 3 confirms that there was an agedependent variation in the relative OA overexpression level in OA-Tg mice. The OA expression in the bone of 4 weeks old female OA-Tg mice was 2-to 3-fold of that of age-matched female WT littermates. The overexpression level was increased to 6-fold of WT controls in 8 weeks old female OA-Tg mice. However, the OA overexpression level in the bone of female OA-Tg mice at 15.3 weeks (107 days) of age was reduced to ,3-fold of WT controls.
To test whether this age-related difference in relative OA overexpression levels in bone of OA-Tg mice would result in agedependent differences in the low bone mass phenotype, we also determined the pQCT bone parameters of female OA-Tg mice of 4 weeks of age and of 15.3 weeks of age with respective agematched WT littermates. Table 2 shows that the 4-week-old adolescent OA-Tg mice as well as the 15.3-week-old mature OA-Tg mice each also exhibited similar low bone mass phenotype as those seen in 8-week-old young adult mice (Table 1), although the differences in bone parameters at metaphysis of the younger 4week-old OA-Tg compared to corresponding WT controls did not reach statistically significant levels, presumably due to the relatively large variations. Accordingly, the low bone mass phenotype could also be seen at an age as early as after weanling at 4-week-old and was maintained after they reached adulthood. Two-factor ANOVA analysis confirms that both the age (i.e., growth) and the genotype (i.e., OA overexpression) had significant effects on the various pQCT bone parameters (Fig. 4). However, there was no significant interaction between the genotype and the age of the animals, indicating that the age-related variations in OA overexpression did not have synergistic effects on the low bone mass phenotype.

m-CT analyses of the bone phenotype of OA-Tg mice
To further characterize the low bone mass phenotype, we compared the m-CT bone mineral parameters in femurs of 4week-old female OA-Tg mice with those in femurs of 4-week-old female WT littermates. Measurements at the midshaft, which consists primarily cortical bone, show that the OA-Tg mice had 23.3% lower BV (0.32660.019 mm 3 for Tg mice vs. 0.42560.007 mm 3 for WT mice, p,0.005) and 10.8% lower BV/TV (0.26460.010 mm 3 /mm 3 for Tg mice vs. 0.29660.006 mm 3 /mm 3 for WT mice, p,0.05) than WT littermates. The tissue volume (TV) at this site was not significantly different between the two groups (1.24860.122 mm 3 for Tg mice vs. 1.43660.024 mm 3 for WT mice, p.0.05). These findings confirm the pQCT findings that targeted overexpression of OA in cells of osteoclast lineage caused a significant loss of cortical bone mass. On the other hand, contrary to the pQCT results that showed no significant decreases in trabecular BMD and BMC (Table 1), the m-CT analysis clearly showed that these OA Tg mice showed significant (p,0.05 for each) decreases in BV/TV (231.8%), connectivity density (228.6%), and Tb.Th (214.3%), along with significant (p,0.05 for each) increases in Tb.Sp (+16.8%) and structure model index (SMI, +35.5%) at the trabecular bone-enriched secondary spongiosia site (Fig. 5).
Consequently, it appears that OA overexpression can cause significant bone loss at both cortical and trabecular bone sites. In addition, the significant decreases in connectivity density and Tb.Th, along with an increase in SMI, are also consistent with an increase in bone resorption in these OA-Tg mice.

Effects of targeted OA overexpression in osteoclastic cells on bone resorption and osteoclast activity in vivo and in vitro
To confirm that the low bone mass phenotype in OA-Tg mice was indeed due to an increase in bone resorption, we first determined static histomorphometric bone resorption parameters at the secondary spongiosia of femurs of 4-week-old female OA-Tg mice and WT littermates. Table 3 confirms that OA-Tg mice indeed had significantly lower relative bone volume (%BV/TV, 215%). It also reveals that OA-Tg mice had significant increases in osteoclast number (NOC, +26%), osteoclasts per total bone surface (NOC/BS, +28%), and %TRAP-stained bone surface (TRAP.Pm, +22%). The trabecular thickness (Tb.Th) was significantly reduced by 15%. Although the OA-Tg mice also had lower trabecular number (Tb.N) and larger trabecular spacing (Tb.Sp) than WT littermates, the differences did not reach statistically significant levels, perhaps due in part to the relatively small sample size (n = 5-6 per group). Thus, these findings not only demonstrate that overexpression of OA in cells of osteoclast lineage increased  osteoclastic resorption in vivo, but also confirm the m-CT data ( Fig. 5) that significant bone loss was seen in trabecular bones. We next measured the circulating level of c-telopeptide of type I collagen (a biomarker of bone resorption) in 8-week-old male young adult OA-Tg transgenic mice and WT littermates. The plasma level of c-telopeptide in OA-Tg mice was significantly (p,0.05) higher (by 19.4%) compared to the WT littermates (Fig. 6). This result, along with the increases in NOC, NOC/BS, and TRAP.Pm, indicates that the low bone mass phenotype seen in young adult OA-Tg mice is in a large part due to an increase in bone resorption in vivo.
Because an increase in the size and/or nucleation of osteoclasts is associated with an increase in functional activity, we also determined whether targeted overexpression of OA in cells of osteoclastic lineage would increase the size or nucleation of osteoclasts on the bone surface in vivo. Fig. 7 shows that the average size of osteoclasts (osteoclast size/osteoclast) on bone surfaces at secondary spongiosia of 4-week-old female OA-Tg mice was ,70% (P = 0.029) larger than those of WT littermates. The average osteoclast surface per osteoclast (OC.PM/OC#) was also ,40% (P = 0.046) greater in OA-Tg mice than in WT littermates. Similarly, the number of nuclei per osteoclast was 2-fold (P = 0.012) greater in OA-Tg mice than in WT littermates.
We previously provided in vitro evidence that osteoclast-derived OA plays an important functional role in osteoclast formation and activity in part through promoting the RANKL-dependent fusion and spreading of osteoclasts to form larger and functionally more active osteoclasts [4]. To assess whether overexpression of OA in osteoclastic cells has direct effects on osteoclasts, we compared the relative cell size, TRAP expression level, and the in vitro bone resorption activity of osteoclasts derived from marrow precursors of adult OA-Tg mice with those of corresponding WT osteoclasts. Fig. 8 shows that the marrow-derived TRAP + , multinucleated, osteoclasts of 12-week-old OA-Tg mice were twice as large (Figs. 8A and B), contained more nuclei (Fig. 8A), expressed twice as much TRAP activity (Fig. 8C), and produced twice as large resorption pits (Fig. 8D), as control WT marrow-derived osteoclasts in vitro. In addition, marrow-derived osteoclasts of OA-Tg mice also expressed significantly higher levels of metalloprotease-9 (MMP9) mRNA [a critical degradative enzyme for osteoclastic resorption], calcitonin receptor (CALCR) mRNA [a marker gene for mature osteoclasts], and ADAM12, but not MMP3 mRNA, than WT osteoclasts (Fig. 8E). The phosphotyrosine-527 (PY-527) level of Src in OA-Tg osteoclasts was approximately one-half as that in WT osteoclasts (Fig. 8F). Because the protein-tyrosine kinase (PTK) activity of Src is negatively regulated by its PY-527 level, the PTK activity of Src or its signaling pathway could be twice as active in Tg osteoclasts as in WT osteoclasts. To evaluate whether overexpression of OA in osteoclastic cells might also affect the differentiation and formation of osteoclasts in vitro, we counted the total number of TRAPexpressing, multinucleated (two or more nuclei) osteoclasts formed from treating marrow-derived osteoclast precursors of five 4-weekold female OA-Tg mice and four 4-week-old female WT littermates after 4 days of the RANKL and m-CSF treatment in vitro. While precursors of OA-Tg mice yielded bigger TRAP positive osteoclast-like cells, there was no significant difference in the total number of osteoclasts derived from marrow-derived precursors of OA-Tg mice and from those of WT littermates (Fig. 8G), suggesting that OA in osteoclasts is a potent enhancer of osteoclast fusion and activation but not a major regulator of osteoclast formation.
To determine if siRNA-mediated suppression of OA expression in RAW264.7 cell-derived osteoclast-like cells would yield opposite effects on the cell size (Fig. 9A) and expression of key osteoclastic genes, RAW264.7 cells were pre-treated with two OA-specific siRNAs (denoted as OA siRNA #1 and OA siRNA #2, respectively) for 6 hrs prior to RANKL treatment to form osteoclast-like cells. Both siRNAs suppressed OA mRNA and OA protein expression by .70% and .80%, respectively (data not shown), and each significantly reduced the average cell size of the derived osteoclast-like cells as well as decreased in vitro bone resorption activity (Fig. 9B). Thus, subsequent siRNA studies used only OA siRNA#1. The OA siRNA-treated osteoclast-like cells expressed significantly less MMP9 mRNA and CALCR mRNA (Fig. 9C). The siRNA treatment did not have a significant effect on the NFAT-c1 mRNA [an essential transcription factor for osteoclast differentiation and formation] level, supporting our contention that OA is not an essential regulator of osteoclast differentiation and formation.

Effects of targeted OA overexpression in osteoclastic cells on bone formation in vivo and osteoblast activity in vitro
The shedded OA has been shown to be a potent stimulator of osteoblast differentiation [24][25][26]. To evaluate whether overexpression of OA in TRAP exon 1B/C-expressing cells would release shedded OA protein to act on nearby osteoblasts to stimulate osteoblast differentiation and bone formation, we measured dynamic histomorphometric bone formation parameters at the midshaft of longitudinal sections of femurs of 8-week-old male OA-Tg mice with that of male WT littermates (Table 4). Overexpression of OA in TRAP-expressing cells not only did not increase any of the test histomorphometric bone formation parameters but instead significantly reduced the bone surface and total tetracycline-labeling surfaces. Measurements at the endosteal bone surface of cross-sections of femur at the midshaft of 15.3-weekold female OA-Tg mice also showed significant reduction in endosteal mineralization apposition rate and endosteal bone formation rate when compared to age-matched female WT littermates. To determine whether overexpression of OA in osteoclasts would affect the number of ALP-expressing osteoblasts along the bone surface in vivo, we counted the ALP-expressing osteoblasts on the trabecular bone surface of 4 weeks-old female OA-Tg mice and WT littermates and normalized them against total bone surface or tissue area (Table 5). There were no significant differences in the number of osteoblasts per bone surface (NOB/BS) or the number of osteoblasts per tissue area (NOB/TA). There was also no significant difference in the osteoblast surface per bone surface (OB.PM/BS), indicating that targeted overexpression of OA in cells of osteoclastic lineage did not alter osteoblast activation or increase osteoblastic surface on trabecular bone.
To confirm the lack of bone formation effects, we measured plasma levels of biomarkers of bone formation in the OA-Tg mice. Plasma osteocalcin level of 8-week-old female OA-Tg mice was not different from that of age-matched WT littermates, while the plasma osteocalcin level of 15.3-week-old OA-Tg mice was even slightly (close to, but not statistically significant, P = 0.09) lower than that of age-matched WT mice (Fig. 10A). Similarly, plasma level of another bone formation biomarker, pro-collagen type I Nterminal peptide (PINP), of 15.3-week-old female OA-Tg mice of both sexes was also slightly, but not significantly, reduced when compared to that of age-and sex-matched WT littermates (Fig. 10B).

Discussion
In this study, we have obtained compelling in vivo evidence that osteoclast-derived OA has essential functions in regulating osteoclast activity and in bone resorption. Accordingly, this study shows that targeted overexpression of OA to cells of osteoclastic lineage led to substantial loss of bone mass and density in OA-Tg mice. Although pQCT measurements failed to detect significant loss of bone mass at trabecular bone sites of OA-Tg mice, bone histomorphometry and m-CT analyses each indicate substantial loss of bone mass in trabecular bone. The reason for the failure in detecting bone loss at trabecular bone sites by pQCT is unclear. Nevertheless, based on the m-CT and histomorphometry findings, we conclude that targeted overexpression of OA in cells of osteoclast lineage would lead to loss of bone mass at both cortical   and trabecular sites. Moreover, the findings of significant increases in NOC.PM, TRAP.PM, and Tb.Sp, along with a decrease in Tb.Th, as well as an elevated plasma level of c-telopeptide in OA-Tg mice compared to age-and sex-matched WT littermates strongly indicate that osteoclast-derived OA has essential regulatory roles in bone resorption in vivo. In contrast, histomorphometric bone formation parameters, number of osteoblasts per bone surface length, and plasma levels of osteocalcin and PINP in OA-Tg mice were not different from those in WT littermates. These findings led to our conclusion that the bone loss in OA-Tg mice is due to an increase in bone resorption and not to reduction in bone formation. Two observations in this study are noteworthy. First, in vitro characterization of marrow-derived osteoclasts from OA-Tg mice indicates that marrow-derived osteoclasts of OA-Tg mice were significantly larger, contained more nuclei, and produced larger resorption pits in the in vitro resorption pit formation assay than osteoclasts derived from WT mice. These observations confirm our previous in vitro findings [4] that OA has enhancing effects on osteoclast maturation, fusion, and spreading, and also in their in vitro bone resorption activity. Interestingly, overexpression of OA in osteoclasts upregulated expression of genes associated with osteoclastic activity (e.g., Mmp9 and Calcr), but not those associated with osteoclast formation (e.g., Nfatc1), suggesting that osteoclastderived OA promotes bone resorption primarily through activation of osteoclasts rather than osteoclast formation. This tentative conclusion is supported by both in vivo and in vitro evidence: first, overexpression of OA in cells of osteoclastic lineage did not alter significantly the total number of osteoclasts but increased markedly the percentage of osteoclastic surface (TRAP.PM) on bone surfaces in vivo. Second, while treatment of marrow-derived precursors of OA-Tg mice with RANKL yielded significantly larger TRAPexpressing osteoclasts than those of WT littermates in vitro, the total number of osteoclasts derived from precursors of both OA-Tg and WT mice was not significantly different.
With respect to the potential molecular mechanism by which OA stimulates osteoclast activity, our previous in vitro studies in  marrow-derived osteoclasts have implicated the involvement of OA-induced activation of the integrin signaling (presumably through the RGD-depending OA interaction with integrins) [4]. The integrin signaling in osteoclasts is a key mechanism mediating cytoskeleton reorganization, which is required for all essential processes associated with osteoclast activation, including cell fusion, attachment, polarization, and construction of the sealing zone and ruffled borders [33]. Activation of the Src PTK through dephosphorylation of the inhibitory PY527 residue [34] is one of the critical steps involved in the integrin signaling in osteoclasts [35]. In this regard, the finding that OA-Tg osteoclasts showed significant lower Src PY527 level than WT osteoclasts is consistent with an activation of the Src signaling in OA-Tg osteoclasts. This finding, along with the observations that OA-Tg osteoclasts are larger and more spread out, is consistent with the hypothesis that the OA-induced activation of osteoclast activity may involve Src activation and integrin signaling in osteoclasts. This interesting hypothesis will be evaluated in our future studies.
The second intriguing observation is related to the absence of an increase in bone formation in the transgenic mice. OA belongs to a unique family of transmembrane proteins, which can shed their extracellular domain through proteolysis mediated by extracellular proteases (sheddases) that include various members of the ADAM [5] and MMP [6] families. In bone, the shedded ectodomain of OA has been suggested to function as autocrine/paracrine factor to promote osteoblast differentiation [2,3,[24][25][26][27]. Activated osteoclasts also shed the ectodomain of osteoclast-derived OA, which also acts as autocrine/paracrine factor to stimulate osteoclast activity [4]. We had hypothesized that overexpression of OA in osteoclastic cells would result in increased shedding of ectodomain of osteoclast-derived OA, which then acts as a paracrine osteogenic factor on nearby osteoblast precursors to promote bone formation in vivo. The findings that overexpression of OA in osteoclasts led to upregulation of Mmp9 and Adam12 expression (two extracellular proteases that mediate the shedding of ectodomain of membrane-bound OA) in osteoclasts of OA-Tg   mice supports the assumption of an increased shedding of osteoclast-derived OA from Tg osteoclasts. Therefore, we were surprised by the absence of corresponding increases in bone formation in OA-Tg mice. One of the possibilities for the lack of an increase in bone formation is that the amounts of the shedded ectodomain of osteoclast-derived OA might not have been sufficient to elicit an anabolic response on osteoblasts that would yield an increase in bone formation. Bone marrow mesenchymal cells derived from mutant mice expressing a truncated OA lacking a large portion of the intracellular domain have been shown to be unable to differentiate into osteoblasts in vitro [25]. Thus, an alternative possibility is that the intracellular domain (rather than the shedded ectodomain) of OA in osteoblasts is essential for the osteogenic response. We also cannot dismiss the possibility that the acidic and proteolytic environment of the resorption cavity might have further degraded the shedded ectodomain, rendering it nonfunctional in stimulating the differentiation and activity of nearby osteoblast precursors.
Understanding of the regulation of OA gene expression in osteoclasts could provide important insights into its functional role in osteoclastic resorption. While very little is known about the regulation of OA expression, this study and our previous studies [4] have shown that OA in osteoclastic cells is up-regulated by RANKL in time-and dose-dependent manner. Accordingly, treatment of osteoclasts precursors with RANKL could result in hundreds-or even thousands-fold increase in OA expression. In light of the well established fact that RANKL is a potent stimulator of osteoclast differentiation and activity and also the well known complexity of the RANKL signaling networks in the stimulation of osteoclast differentiation and activity [36], it is possible that OA may yet be another novel mediator of the RANKL signaling mechanism in osteoclasts. However, previous studies showed that OA expression is strongly up-regulated by IFN-c and lipopolysaccharide in macrophages [9], and that BMP-2 markedly stimulates OA expression in rat osteoblasts [26]. Because IFN-c [37] and BMP-2 [38] have each been shown to be capable of acting directly on osteoclast precursors to promote their differentiation and activity, we cannot rule out the possibility that the molecular mechanism of these two effectors to stimulate osteoclast differentiation and bone resorption may also involve in the up-regulation of OA expression in osteoclastic cells. Our future work will evaluate these interesting possibilities.
In summary, this study demonstrates for the first time that the osteoclast-derived OA is a novel regulator of osteoclastic resorption in vivo. This study also provided compelling evidence that targeted overexpression of OA in cells of osteoclastic lineage with TRAP-1B/C promoter, while markedly increased the bone resorption activity of osteoclasts (and thereby bone resorption), but without an appreciable effects on bone formation in vivo. Understanding the mechanism by which OA regulate the osteoclast activity would not only yield important information about its functional role in the overall regulation of osteoclastic resorption, but might also offer significant insights into the pathophysiology of various bone-wasting disorders, including osteoporosis. More importantly, it may also provide novel gene or pathway targets for development of novel and effective therapy for bone-wasting disorders (such as osteoporosis) or fracture repair, as OA or its downstream effectors may be used as screening targets for identifying novel, specific, and safer modulators of bone resorption.

Animals
Osteoclastic-specific OA transgenic (OA-Tg) founder mice were generated in C57BL/6J (B6) genetic background by the Transgenic Core Facility of the University of California at Irvine using an established approach similar to that previously described for PTP-oc transgenic mice [29]. The OA expression is driven by TRAP-1B/C promoter. Very briefly, the pGL3-TRAP-1B/C-Luc plasmid [30] was a generous gift from Dr. A. Ian Cassady of the University of New England in Armidale, New South Wales, Australia. The pGL3-TRAP-1B/C-OA expression plasmid (Fig. 1) was generated by cloning into the pGL3-basic vector containing the TRAP-1B/C promoter at the Kpn1/Hind3 restriction sites and the long PCR-cloned full-length murine OA cDNA at the Hind3 and Xba1 sites. The linearized transgenic plasmid was then injected into B6 ova to generate OA-Tg founder mice.
Identification of OA-Tg mice were accomplished with a PCRbased genotyping assay. Briefly, a small piece of tail tissue was taken from each pup at weaning and genomic DNA was isolated with the Qiagen DNAeasy Blood & Tissue kit (Qiagen, San Diego, CA) according to the supplier's recommended procedure. The quality and quantity of the genomic DNA were analyzed with the nanodrop ND-1000 spectrophotometer (ThermoFisher Scientific, Los Angeles, CA). The PCR-based genotyping reaction was performed using the following set of primers: the forward primer, which corresponds to a unique region of the 59-region of the mouse TRAP-1B/C promoter (from 21,331 of promoter 1C), was 59-TCC TCG GAG AAA ATG CAT CA-39, and the reverse primer, which corresponds to a unique region of the murine OA sequence (from 565 of OA open reading frame), was 59-CTG GCC AAG TGT GTG AAA GA-39. The PCR condition included a 5-min hot start at 94uC and 35 cycles of amplification, each consisting of 30 sec at 94uC, 30 sec at 55uC, and 1 min at 72uC. This was then followed by a final 10 min at 72uC. The PCR product was analyzed by agarose gel electrophoresis. Genomic DNA of transgenic mice but not WT littermates yielded a single PCR band of 875-bp. The animal use component was approved by the Animal Care and Use Committees of both the Jerry L. Pettis Memorial VA Medical Center and the Loma Linda University. The entire study has also been reviewed and approved by the Research and Development committee of the Jerry L. Pettis Memorial VA Medical Center.  -AAT GGG TCT GGC ACC TAC TG-39  59-GGC TTG TAC GCC TTG TGT TT-39 Murine ADAM12