STAT3 Activation in Stromal/Osteoblastic Cells Is Required for Induction of the Receptor Activator of NF-κB Ligand and Stimulation of Osteoclastogenesis by gp130-utilizing Cytokines or Interleukin-1 but Not 1,25-Dihydroxyvitamin D3 or Parathyroid Hormone*

Interleukin (IL)-6-type cytokines stimulate osteoclastogenesis by activating gp130 in stromal/osteoblastic cells and may mediate some of the osteoclastogenic effects of other cytokines and hormones. To determine whether STAT3 is a downstream effector of gp130 in the osteoclast support function of stromal/osteoblastic cells and whether the gp130/STAT3 pathway is utilized by other osteoclastogenic agents, we conditionally expressed dominant negative (dn)-STAT3 or dn-gp130 in a stromal/osteoblastic cell line (UAMS-32) that supports osteoclast formation. Expression of either dominant negative protein abolished osteoclast formation stimulated by IL-6 + soluble IL-6 receptor, oncostatin M, or IL-1 but not by parathyroid hormone or 1,25-dihydroxyvitamin D3. Because previous studies suggested that IL-6-type cytokines may stimulate osteoclastogenesis by inducing expression of the tumor necrosis factor-related protein, receptor activator of NF-κB ligand (RANKL), we conditionally expressed RANKL in UAMS-32 cells and found that this was sufficient to stimulate osteoclastogenesis. Moreover, dn-STAT3 blocked the ability of either IL-6 + soluble IL-6 receptor or oncostatin M to induce RANKL. These results establish that STAT3 is essential for gp130-mediated osteoclast formation and that the target of STAT3 during this process is induction of RANKL. In addition, this study demonstrates that activation of the gp130-STAT3 pathway in stromal/osteoblastic cells mediates the osteoclastogenic effects of IL-1, but not parathyroid hormone or 1,25-dihydroxyvitamin D3.

The adult skeleton undergoes periodic replacement of old bone by new. During this process, old bone is resorbed by osteoclasts and new bone is formed by osteoblasts (1). Both osteoblasts and osteoclasts are derived from precursors origi-nating in the bone marrow. The precursors of osteoblasts are multipotent mesenchymal stem cells, while the precursors of osteoclasts are hematopoietic cells of the monocyte/macrophage lineage (2,3). Osteoclast development depends strictly on support provided by stromal/osteoblastic cells. Moreover, hormones or cytokines that stimulate bone resorption such as 1,25-dihydroxyvitamin D 3 (1,25(OH) 2 D 3 ), 1 parathyroid hormone (PTH), members of the interleukin (IL)-6 family, or IL-1 stimulate osteoclast formation by activating discrete signaling pathways in stromal/osteoblastic cells (4 -6). The extent to which these pathways interact, or are dependent on one another, to stimulate osteoclast formation is largely unknown.
The mechanistic basis of the dependence of osteoclastogenesis on mesenchymal cell differentiation has recently been established by the discovery of a membrane-bound member of the tumor necrosis factor family of cytokines, receptor activator of NF-B ligand (RANKL), which is expressed in committed preosteoblastic cells (7)(8)(9). RANKL binds to a specific receptor (RANK) that is expressed in hematopoietic osteoclast progenitors (7). This interaction is necessary and, together with M-CSF, sufficient for osteoclast formation, since mice lacking RANKL are unable to make osteoclasts and since exogenously provided soluble RANKL and M-CSF stimulate osteoclastogenesis in the absence of stromal/osteoblastic cells (8 -10). In addition, it has been demonstrated that many of the cytokines and hormones that stimulate osteoclast formation also stimulate the expression of RANKL in stromal/osteoblastic cells (9).
Members of the IL-6 cytokine family that are capable of stimulating osteoclast formation include IL-6, complexed with its soluble receptor (IL-6 ϩ sIL-6R), IL-11, oncostatin M (OSM), and to a lesser extent, leukemia inhibitory factor (LIF) (11). Cell surface receptor complexes for each of these cytokines contain the signal-transducing protein gp130 (12). Upon ligand binding, gp130 either homodimerizes or heterodimerizes with the related protein OSM receptor ␤ or LIF receptor ␤, and it is phosphorylated by associated members of the Janus kinase family of tyrosine kinases (13). This event results in tryrosine phosphorylation of several downstream signaling molecules, including members of the signal transducers and activators of transcription (STAT) family of transcription factors (14,15). Phosphorylated STATs in turn undergo homo-and heterodimerization, after which they translocate to the nucleus and activate cytokine-responsive gene transcription (16).
STAT3 is required for several of the cellular responses to gp130 activation, including differentiation of myeloid cells (17)(18)(19) and protection from apoptosis (20). However, other signaling molecules have also been implicated as downstream effectors of gp130 activation. Specifically, STAT1 and, in some cases, STAT5 are phosphorylated following gp130 activation (21)(22)(23) and may turn on a different set of genes than the ones targeted by STAT3 (22). Similarly, gp130 activation can cause activation of the extracellular signal-regulated kinase/mitogenactivated protein kinase pathway (14,24). Post-translational modification of the transcription factor CCAAT/enhancer-binding protein ␤ may also mediate some of the downstream effects of gp130 activation (24). Finally, gp130 activation can lead to activation of the protein-tyrosine phosphatase SHP-2, which acts as a positive regulator of ciliary neurotrophic factor-induced c-fos expression (25).
In the studies presented in this paper, we have investigated whether STAT3 is required for the osteoclastogenic effects of gp130-activating cytokines and/or other stimuli thought to depend on gp130 signaling. We present evidence that STAT3 is an essential downstream effector of the signaling of IL-6-type cytokines during osteoclastogenesis and that the ultimate target of STAT3 during this process is induction of RANKL. In addition, we provide evidence that IL-1 stimulates osteoclastogenesis via gp130-STAT3 activation, whereas 1,25(OH) 2 D 3 and PTH act independently of gp130 or STAT3.

EXPERIMENTAL PROCEDURES
Osteoclast Formation Assay-Nonadherent bone marrow cells were prepared by removing femurs from 30 -90-day-old C57BL/6J mice and flushing the marrow cavity with minimum essential medium-␣ (Life Technologies, Inc.) containing 15% fetal bovine serum (Hyclone). Marrow cells were seeded at a density of 2.5 ϫ 10 5 cells/cm 2 in the same medium and cultured for 48 h, after which nonadherent cells were collected and counted. Nonadherent bone marrow cells were seeded at a density of 2 ϫ 10 4 cells/cm 2 together with the UAMS-32 stromal/ osteoblastic cell line at 5 ϫ 10 3 cells/cm 2 in minimum essential medium-␣ containing 10% fetal bovine serum. For co-cultures involving conditional expression of transduced genes, retrovirally transduced UAMS-32 cells were seeded at 2.5 ϫ 10 3 cells/cm 2 and cultured in the presence or absence of 100 ng/ml doxycycline-HCl (Sigma) for 2 days before the addition of nonadherent bone marrow cells. In either case, osteoclast-inducing hormones or cytokines were added at the indicated concentrations, and the co-cultures were maintained at 37°C in 5% CO 2 for 6 days. On day 3, one-half of the medium was replaced with fresh medium. After 6 days, cells were fixed and stained for tartrate-resistant acid phosphatase (TRAP), and in some cases, they were assayed for the presence of calcitonin receptor as described previously (26). The ability of the osteoclast-like cells to form resorption pits was determined by performing co-cultures on slices of devitalized bovine cortical bone (a kind gift of P. Osdoby). After culturing, the slices were stripped of cells with dilute bleach, and pits were visualized by reflected light microscopy.
DNA Constructions-Retroviral vectors were derived from the LNL-6 vector provided by Dr. Dusty Miller (27). The vector LEN was constructed by mutating the EcoRI site downstream of the extended packaging sequence to a BglII site by linker insertion, followed by removal of the neomycin phosphotransferase gene as a BglII-SalI fragment. A BglII-SalI cassette containing the internal ribosome entry site from the 5Ј-untranslated region of encephalomyocarditis virus fused to the neomycin phosphotransferase gene was then inserted (28). The tTA expression vector, pLEN-tTA, was generated from LEN by the insertion of a 1190-bp BamHI-HindIII fragment containing the tTA coding region (derived from pRetro-Off; CLONTECH) immediately 5Ј of the encephalomyocarditis virus-neomycin phosphotransferase cassette. The selfinactivating retroviral vector pSIN was derived from the retroviral vector LED by deletion of an XbaI-PvuII fragment from the 3Ј-long terminal repeat, which removes the 75-bp direct repeats containing the Moloney murine leukemia virus enhancer (29,30). In addition, the 1383-bp BglII-XhoI fragment was replaced with a 30-bp fragment containing XhoI and NotI sites. A 440-bp XhoI-NotI fragment containing the TetO-CMV promoter (TRE) was removed from the plasmid pRetro-On (CLONTECH) and inserted into the same sites of pSIN to make the vector pST. To generate the Tet-regulated dn-STAT3 expression vector, a 2.5-kilobase pair XhoI-BglII fragment containing the dn-STAT3-FLAG coding region was removed from pY705F-FLAG (a gift from M. Saunders) (31) and inserted downstream of the TRE of pST to yield pST-dn-STAT3. To generate the Tet-regulated dn-gp130 expression vector, a 130-bp DNA fragment consisting of three sequential Myc epitopes followed by a stop codon was inserted into the ScaI site of pmgp130 (a gift from T. Kishimoto) (32). This insertion resulted in truncation of murine gp130 at glutamic acid 756. A 2.4-kilobase pair SacI-MluI fragment containing the truncated, Myc-tagged gp130 coding sequence was inserted downstream of the TRE of pST to yield pST-dn-gp130. The murine RANKL coding sequence was amplified from total RNA, obtained from 1,25(OH) 2 D 3 -treated UAMS-32 cells, using primers generated from the published sequence (7). The complete sequence of the amplified cDNA was verified using a model 377 automated DNA sequencer (Applied Biosystems). To generate the Tet-regulated RANKL expression vector, the full-length cDNA was inserted downstream of the TRE of pST to yield pST-RANKL.
Retroviral Packaging and Infection-Plasmids harboring retroviral constructs (6.5 g/6-cm dish) were transiently transfected into the Phoenix ampho packaging cell line (33) using LipofectAMINE (Life Technologies, Inc.). Supernatants containing viral particles were collected between 48 and 72 h post-transfection, filtered through a 0.45-m filter, and either used immediately or stored at Ϫ80°C. Subconfluent UAMS-32 cells were exposed to viral supernatants in the presence of 4 g/ml polybrene for 6 -12 h and then incubated in fresh culture medium for 12-24 h. The cells were then exposed to aliquots of the same supernatant two additional times before expansion. Infections with the Tet-regulated constructs were carried out in the presence of 100 ng/ml doxycycline.
Transient Transfections-Transient transfection of retrovirally transduced UAMS-32 cells, plated at 5 ϫ 10 4 cells/well in 12-well plates, was performed using LipofectAMINE as described previously (35). The promoter-reporter construct used in this assay, p4xAPRE (a gift from I. Matsumura), contained a firefly luciferase gene controlled by composite promoter consisting of four STAT-binding elements from the ␣ 2 -macroglobulin promoter inserted upstream from a minimal Jun B promoter (19). Luciferase values were normalized to ␤-galactosidase activity resulting from co-transfection of the plasmid pSV␤-gal (Promega).

Isolation of a Stromal/Osteoblastic Cell Line That Supports
Osteoclast Formation-In order to study the role of STAT3 in osteoclastogenesis, it was necessary to obtain a stromal/osteoblastic cell line that supports osteoclast formation in response to IL-6-type cytokines as well as other osteoclast-inductive agents. We previously reported the isolation of several spontaneously transformed murine bone marrow cell lines that exhib-ited phenotypic characteristics of stromal/osteoblastic cells, including expression of osteoblast-specific genes and the ability to form a mineralized matrix in vitro (37). When co-cultured with nonadherent bone marrow cells, as a source of osteoclast precursors, several of these lines supported the formation of cells that were multinucleated and TRAP-positive, phenotypic characteristics of osteoclast-like cells (38). For the studies described here, we selected one of these lines, UAMS-32, based on its ability to support osteoclast-like cell formation in response to several osteoclastogenic agents (Fig. 1A). UAMS-32 cells supported osteoclast-like cell formation when treated with 1,25(OH) 2 D 3 , IL-6 ϩ sIL-6R, OSM, IL-1, or PTH. The presence of calcitonin receptors as well as the ability to excavate resorption pits on devitalized bovine cortical bone (Fig. 1, B and C, respectively) confirmed that the multinucleated, TRAP-positive cells were, in fact, authentic osteoclasts.
Conditional Expression of Dominant Negative STAT3 and gp130 -To determine whether STAT3 is a downstream effector of gp130 in the osteoclast support function of stromal/osteoblastic cells, we sought to express a form of this protein that would block the activity of endogenous STAT3 in UAMS-32 cells. For this purpose, we utilized a mutant STAT3 protein in which the tyrosine at position 705 was changed to phenylalanine (Y705F) (31). This mutation was shown to block phosphorylation of endogenous STAT3 as well as transactivation of a reporter gene by STAT3 and therefore functions in a dominant negative manner (31). The dn-STAT3 has been tagged at the carboxyl terminus with the FLAG epitope to allow specific detection of this protein (Fig. 2). To determine if pathways initiated by gp130 activation mediate the effects of other osteoclast inducing agents, we constructed a dn-gp130. This protein contains a carboxyl-terminal truncation to glutamic acid 756, followed by a Myc tag, thus removing the STAT3 and SHP-2 binding sites from the gp130 cytoplasmic tail (Fig. 2), and is similar to a previously reported dominant negative version of human gp130 (39).
Initially, we attempted to stably transfect UAMS-32 cells with plasmid constructs constitutively expressing either dn-STAT3 or dn-gp130. However, in both cases we were unable to isolate any cell clones that expressed significant levels of the dominant negative proteins or that demonstrated reduced gp130 signaling (not shown). A possible explanation for this result is that at the extremely low cell densities required for cloning, some level of gp130 or STAT3 activation, via an autocrine pathway, may be required for viability or proliferation. In order to circumvent this problem, we expressed these dominant negative proteins in UAMS-32 cells in a conditional manner utilizing the tetracycline-regulated gene expression system (40). This was accomplished by first generating a pool of cells infected with a retroviral construct constitutively expressing the transactivator fusion protein, tTA (pLEN-tTA). This pool of cells was then infected with a second viral construct containing either dn-STAT3 or dn-gp130 cDNA under the control of the tetracycline-responsive promoter (pST-dn-STAT3 or pST-dn-gp130, respectively).
Induction of dn-STAT3 in cells transduced with pST-dn-STAT3 was analyzed by immunoblot with anti-FLAG antibodies. The FLAG-tagged dn-STAT3 was undetectable in extracts of cells grown in the presence of the potent tetracycline derivative, doxycycline (Fig. 3). However, when the cells were cultured in the absence of doxycycline, high level expression of dn-STAT3 was observed. Analysis of the same blot with an antibody that recognized both the wild type and the dominant negative STAT3, which have the same gel mobility, indicated a severalfold increase in band intensity in the absence of doxycycline, suggesting that the dominant negative STAT3 was present in excess of the wild type protein. Similar results were obtained with the cells expressing the Myc-tagged dn-gp130. Immunodetection with an antibody that recognized the wild type, but not the dominant negative, gp130 indicated that dn-gp130 expression did not effect wild type protein levels. To demonstrate that the dominant negative constructs were functional, each pool of infected cells was transiently transfected with a promoter-luciferase construct that is responsive to activated STAT3. In the absence of the dominant negative proteins, IL-6 ϩ sIL-6R or OSM stimulated the activity of this promoter severalfold (Fig. 4). However, induction of either dn-STAT3 (left panel) or dn-gp130 (right panel) blocked this stimulation. These results demonstrate that withdrawal of doxycycline from cultures of these transduced cells results in expression levels of either dn-STAT3 or dn-gp130 that are sufficient to decrease the magnitude of the intracellular signaling initiated by IL-6-type cytokine binding.
STAT3 Is Required for Osteoclast Formation Induced by IL-6type Cytokines or IL-1, but Not by PTH or 1,25(OH) 2 D 3 -To determine if gp130-mediated osteoclast support requires functional STAT3, we analyzed the ability of UAMS-32 cells conditionally expressing dn-STAT3 to support osteoclast formation in response to either IL-6 ϩ sIL-6R or OSM, as well as other osteoclast-inducing agents (Fig. 5, left panel). In the presence of doxycycline, treatment with IL-6 ϩ sIL-6R or OSM resulted in significant levels of osteoclast formation. However, when the co-cultures were performed in the absence of doxycycline, resulting in dn-STAT3 expression, osteoclast formation in response to either IL-6 ϩ sIL-6R or OSM was completely abolished. Strikingly, osteoclast formation in response to IL-1 was also eliminated by expression of dn-STAT3. Expression of dn-STAT3 had no effect on 1,25(OH) 2 D 3 -or PTHinduced osteoclast formation, indicating that dn-STAT3 did not have a generalized negative effect on cell viability but blocked only specific pathways leading to osteoclast support. Essentially the same results were obtained when UAMS-32 cells conditionally expressing dn-gp130 were analyzed in parallel (Fig. 5, right panel).
Osteotropic Agents Stimulate Osteoclast Formation by Stimulating RANKL Expression-Previous studies have demonstrated that osteotropic agents stimulate the production of RANKL and that, together with M-CSF, this protein can initiate the differentiation of hematopoietic precursors into osteoclasts in the absence of stromal/osteoblastic cells (8,9). Nonetheless, these earlier studies have not established whether RANKL is the sole target of gp130 activation required for support of osteoclast formation by stromal/osteoblastic cells. For example, it is conceivable that in addition to inducing RANKL expression, osteotropic agents may stimulate osteoclast formation in part by down-regulating the production of antiosteoclastogenic cytokines such as interferon-␥, IL-18, or IL-4 by stromal/osteoblastic cells (41). To determine if IL-6type cytokines or other osteotropic agents induce support of osteoclast formation solely by stimulating RANKL expression in stromal/osteoblastic cells, we first confirmed that each of these agents was able to stimulate RANKL mRNA expression in UAMS-32 cells (Fig. 6A). Next, RANKL mRNA was conditionally expressed in UAMS-32 cells using the tetracyclineregulated expression system. Induction of RANKL expression by removal of doxycycline was sufficient in itself to dramatically stimulate osteoclast formation in co-cultures using the transduced UAMS-32 cells, overcoming the requirement for stimulation by IL-6-type cytokines or other osteotropic agents (Fig. 6B).
STAT3 Is Required for RANKL Expression Induced by IL-6 ϩ sIL-6R and OSM, but Not by 1,25(OH) 2 D 3 -Given the findings that dn-STAT3 specifically blocked IL-6-type cytokine-or IL-1-induced osteoclastogenesis (Fig. 5) and that conditional expression of RANKL in UAMS-32 cells supplanted the need for osteotropic agents (Fig. 6), we postulated that the dn-STAT3 blocked osteoclast formation by blocking the stimulated expression of RANKL. To determine if this was the case, we treated UAMS-32 cells conditionally expressing dn-STAT3 with IL-6-type cytokines or 1,25(OH) 2 D 3 and measured RANKL expression by Northern blot analysis (Fig. 7, left panel). In the absence of dn-STAT3 (plus doxycycline), 1,25(OH) 2 D 3 , OSM, or IL-6 ϩ sIL-6R stimulated the expression of RANKL mRNA, which was undetectable in cells treated with vehicle. However, in the presence of dn-STAT3 (minus doxycycline), RANKL mRNA stimulation by OSM or IL-6 ϩ sIL-6R was dramatically reduced or completely blocked, whereas 1,25(OH) 2 D 3 -stimulated RANKL expression was unaffected. Similar results were obtained in cells conditionally expressing dn-gp130 (Fig. 7, right panel). Hybridization of these Northern blots with an M-CSF probe demonstrated that this mRNA was constitutively expressed by UAMS-32 cells and was only minimally affected by hormone or cytokine treatment. In an additional experiment, induction of dn-STAT3 in UAMS-32 cells also reduced stimulation of RANKL mRNA by IL-1 but not by PTH (data not shown). DISCUSSION A role for the signaling pathways triggered by gp130 activation in osteoclastogenesis has been strongly suggested by observations that gp130 activation by IL-6, IL-11, OSM, or LIF stimulates osteoclast formation in a variety of in vitro models of osteoclastogenesis (38,42,43). In addition, mice and rats receiving IL-6-neutralizing antibodies or genetically engineered IL-6-deficient mice do not exhibit the expected increase in osteoclast formation and bone loss that occur in their control littermates after gonadectomy (44 -46). Besides its critical role in the bone loss caused by loss of gonadal function, IL-6 seems to play a similar role in several other conditions associated with increased bone resorption as evidenced by increased local or systemic production of IL-6 and the IL-6 receptor in patients with multiple myeloma, Paget's disease, rheumatoid arthritis, Gorham-Stout or disappearing bone disease, hyperthyroidism, primary and secondary hyperparathyroidism, and McCune Albright Syndrome (47). Nonetheless, osteoclast formation is unaffected in eugonadal mice or rats receiving neutralizing IL-6 antibodies or in gp130 knockout mice, suggesting that gp130mediated osteoclast formation may be important only in pathologic states (1).
The results of the in vitro studies presented in this report elucidate for the first time some of the downstream signals triggered by gp130 activation, in stromal/osteoblastic cells, that are essential for osteoclast formation from hematopoietic precursors. Specifically, the data described in this paper demon- strate that activation of the transcription factor STAT3 in stromal/osteoblastic cells is required for gp130-mediated osteoclastogenesis and that the ultimate target of STAT3 for osteoclastogenesis is RANKL. In addition, our results provide evidence to suggest that the gp130-STAT3-RANKL pathway mediates not only the osteoclastogenic effects of IL-6-type cytokines but also those of IL-1. However, in the in vitro system utilized in the study, the osteoclastogenic effects of PTH or 1,25(OH) 2 D 3 were independent of the gp130-STAT3 pathway.
Although IL-11 and LIF have each been shown to stimulate osteoclast formation in other culture systems (11,42,43), these cytokines were unable to do so in our co-culture system. 2 The most likely explanation for this is that UAMS-32 cells lack sufficient amounts of the specific receptors for these cytokines. However, both of these cytokines require gp130 for signal transduction, and both activate STAT3 in cells with IL-11-or LIF-specific receptors (12). Therefore, it is likely that, as is the case with OSM and IL-6 ϩ sIL-6R in our system, STAT3 is required for the induction of RANKL (9) and stimulation of osteoclast formation (11, 42) by IL-11 or LIF in other culture systems.
Since STAT1 and STAT5 can also be phosphorylated after gp130 activation (21,22) and since both may form complexes with STAT3 (16, 23), we cannot rule out the possibility that these other STATs, together with STAT3, participate in gp130mediated osteoclastogenesis. It should be noted, however, that many gp130 responses are normal in STAT1-deficient mice (48) and that STAT5 and STAT3 appear to target different sets of genes (22). In any event, further studies with dominant negative forms of these other STATs will be required to establish whether they play a role in osteoclastogenesis. Similarly, our results cannot exclude the possibility that other gp130 signaling pathways may also be required for the osteoclastogenic response to IL-6-type cytokines, especially since serine phos- phorylation of STATs, possibly mediated by an extracellular signal-regulated kinase-independent pathway (49), is required for full transcriptional activation of certain target genes (50).
Based on work by their group and others, Suda et al. (38) have proposed that osteoclastogenic agents function by activating one of three separate signaling pathways in stromal/osteoblastic cells: the vitamin D receptor (VDR) pathway activated by 1,25(OH) 2 D 3 ; the protein kinase A pathway activated by PTH, prostaglandins, or IL-1; and the gp130 pathway activated by IL-6-type cytokines. However, evidence exists that suggests that IL-1, PTH, or 1,25(OH) 2 D 3 may stimulate osteoclastogenesis in part by stimulating the production of one or more members of the IL-6-type cytokine family. Indeed, we have shown previously that an anti-IL-11 antibody reduced osteoclast formation in response to IL-1, 1,25(OH) 2 D 3 , or PTH in co-cultures of bone marrow and calvarial cells (42). Likewise, Romas et al. showed that anti-gp130 antibodies abolished IL-1-stimulated osteoclast formation and reduced by approximately 50% osteoclast formation induced by PTH or 1,25(OH) 2 D 3 in a murine co-culture system (43). Further, in vitro as well as in vivo studies have implicated IL-6 in the bone resorbing effects of PTH (42,51). Nonetheless, Devlin et al. (52) have found that human osteoclast development in vitro, in response to IL-1 or tumor necrosis factor but not to 1,25(OH) 2 D 3 or PTH, can be attenuated by an IL-6 antagonist.
The demonstration of an essential role of gp130-STAT3 ac-tivation in IL-1-mediated, but not 1,25(OH) 2 D 3 -or PTH-mediated, osteoclast formation in our murine co-culture system is in full agreement with the studies of Devlin et al. but not with the evidence implicating IL-6-type cytokines in the osteoclastogenic effects of 1,25(OH) 2 D 3 or PTH. The two sets of observations may not be mutually exclusive. Indeed, it is well established that activation of gp130 on hematopoietic osteoclast precursors stimulates their proliferation (53,54). Hence, it is possible that the reduction in 1,25(OH) 2 D 3 -or PTH-mediated osteoclast formation by anti-IL-6, anti-IL-11, or anti-gp130 antibodies in the earlier studies resulted from an inhibition of osteoclast precursor proliferation rather than attenuation of signals provided by stromal/osteoblastic cells, i.e. RANKL or M-CSF. In our studies, we inactivated the gp130-STAT3 pathway specifically in a stromal/osteoblastic cell line that supports osteoclast formation so that possible paracrine or autocrine action of IL-6-type cytokines on hematopoietic cells would have been unaffected. In contrast to the suggestion that IL-1 stimulates osteoclast formation via pathways also utilized by PTH (38), the present work in murine cells, together with that of Devlin et al. using human cells, suggests that the osteoclastogenic effects of IL-1 are not mediated by the same pathways as PTH but rather by the gp130 signaling pathway. IL-1 and PTH are both capable of stimulating IL-6-type cytokine production in bone marrow cells or bone marrow-derived cell lines (26, 55-57), apparently via different signaling pathways (57). This evidence, taken together with the finding that gp130 activation is required for IL-1-stimulated but not PTH-stimulated osteoclast formation, suggests that IL-1 (but not PTH) may exert additional effects, e.g. increase the levels of the ligand binding subunits of the receptors for IL-6-type cytokines in stromal/osteoblastic cells. Alternatively, our results cannot rule out the possibility that IL-1 itself may be activating the gp130-STAT3 pathway (58).
RANKL is expressed predominantly in T-lymphocytes and bone cells (7)(8)(9). We have found and reported elsewhere that activation of the gp130 or VDR signaling pathways results in RANKL expression in cells of the osteoblastic lineage but not other mesenchymal cell lineages (37). Since the gp130 and VDR pathways were active in these other cell types, we have reasoned that restricted expression of RANKL in cells of the stromal/osteoblastic lineage must be determined by a stromal/osteoblast-specific factor. In preliminary studies, we have determined that new protein synthesis is required for gp130 or VDR-stimulated RANKL expression. 2 Therefore, it seems that activation of gp130 or VDR on stromal/osteoblastic cells stimulates the expression of a factor(s) that subsequently increases the level of RANKL mRNA. Whether the cell specificity of RANKL expression is due to cell-specific expression of this intermediate protein(s) and/or cell-specific elements in the RANKL promoter is currently under investigation. It is interesting to note, however, that T-lymphocytes and osteoblasts, the two cell types that express high levels of RANKL, are also the two cell types that express high levels of the transcription factor CBFA-1 (59). More intriguingly, we have recently determined that both the murine and human RANKL genes contain two functional CBFA-1 sites and that mutation of these sites abrogates the transcriptional activity of the RANKL promoter (60). Therefore, the cell-specific expression of RANKL in cells of the stromal/osteoblastic lineage might be dictated, at least in part, by the expression of CBFA-1. In addition to explaining the osteoblast-specific expression of RANKL, we believe that the requirement of CBFA-1 for RANKL gene expression may constitute the molecular mechanism of the linkage between osteoblastogenesis and osteoclastogenesis.
In conclusion, the observations reported in this paper demonstrate the importance of STAT3 activation for RANKL induction and the stimulation of osteoclastogenesis by cytokines that utilize gp130 and strongly suggest that the osteoclastogenic properties of IL-1 might be mediated, at least in part, through IL-6-type cytokines. A better understanding of the signaling pathways activated by osteotropic agents may provide specific targets for therapeutic intervention in conditions of excessive or unbalanced osteoclast formation.