Stimulation by Parathyroid Hormone of Interleukin-6 and Leukemia Inhibitory Factor Expression in Osteoblasts Is an Immediate-early Gene Response Induced by cAMP Signal Transduction

Parathyroid hormone and other agents that stimulate bone resorption function, at least in part, by inducing osteoblasts to secrete cytokines that stimulate osteoclast differentiation and activity. We previously demonstrated that parathyroid hormone induces expression by osteoblasts of interleukin-6 and leukemia inhibitory factor without affecting the 16 other cytokines that were examined. We also showed that stimulation of osteoclast activity by parathyroid hormone is dependent on activation of the cAMP signal transduction pathway and secretion of interleukin-6 by osteoblasts. In the current study, we demonstrate that the rapid and transient stimulation of interleukin-6 and leukemia inhibitory factor is inhibited by actinomycin D and superinduced by protein synthesis inhibitors, the classical characteristics of an immediate-early gene response. Moreover, activation of cAMP signal transduction by parathyroid hormone and parathyroid hormone-related protein is necessary and sufficient to induce both interleukin-6 and leukemia inhibitory factor. In addition, cAMP analogues as well as vasoactive intestinal peptide and isoproterenol, two neuropeptides that stimulate bone resorption by activating cAMP signal transduction in osteoblasts, also induce interleukin-6 and leukemia inhibitory factor in these cells. Taken together with our previous results, this study suggests that interleukin-6 is crucial for stimulation of bone resorption not only by parathyroid hormone, but also by parathyroid hormone-related protein, vasoactive intestinal peptide, and beta-adrenergic agonists, like isoproterenol.

Excessive skeletal loss is caused by a disturbance in the normal balance between bone resorption by osteoclasts, and bone formation by osteoblasts. Knowledge of the mechanisms that regulate these processes is fundamental to the under-standing of the pathogenesis of bone loss that occurs in conditions such as osteoporosis. In addition to forming bone, osteoblasts appear to mediate the effect of parathyroid hormone (PTH) 1 and other resorptive agents (1,2). This concept is best supported by evidence that many agents, including PTH, stimulate bone resorption by osteoclasts that are cultured in the presence of osteoblasts but not by osteoclasts that are cultured alone. In most cases, resorption is also stimulated if the osteoblasts are replaced by conditioned media from osteoblasts activated by the resorptive agents. It is therefore thought that the primary effect of the resorptive agents is to stimulate osteoblasts to produce soluble cytokines that activate the osteoclasts.
We previously examined the effect of PTH on expression by osteoblasts of mRNAs encoding 18 cytokines, and found that of these only interleukin-6 (IL-6) and leukemia inhibitory factor (LIF) were stimulated by the hormone (3,4). Maximal stimulation was approximately 50-fold for IL-6 and approximately 10-fold for LIF. PTH also stimulates secretion by osteoblasts of both IL-6 (3-9) and LIF (10) proteins. Lack of stimulation of IL-6 and LIF by PTH in particular cell preparations is therefore likely due to investigation of osteoblasts at different developmental stages or lack of sensitivity of the methods used. Thus, such results should not be interpreted to mean that PTH does not in general stimulate IL-6 or LIF in osteoblasts (reviewed in Ref. 3).
Our previous detailed time course studies showed that stimulation by PTH of IL-6 and LIF mRNA expression was both rapid and transient in MC3T3-E1 osteoblastic cells, primary rat osteoblasts, and in vivo osteoblasts (3,11). IL-6 and LIF mRNA levels were maximal 30 -60 min after stimulation and returned to baseline within 4 -6 h; a pattern that is typical for immediate-early gene responses (12,13). In the current study, we show that the stimulation of IL-6 and LIF by PTH is inhibited by actinomycin D and superinduced by protein synthesis inhibitors, the distinguishing characteristics of an immediate-early gene response (12,13).
In conjunction with evidence that exogenous IL-6 and LIF can stimulate bone resorption (6, 8, 14 -16), the realization that PTH stimulates IL-6 and LIF expression led to the suggestion that production of these cytokines might be involved in stimulation of resorption by a variety of agents, including PTH (3). In confirmation of this hypothesis, we have recently shown that an antibody directed against the IL-6 receptor can block osteoclast activation induced by PTH (17). The importance of IL-6 is underscored by recent findings that implicate the cytokine in the increased bone resorption that occurs during estrogen withdrawal, Paget's disease, rheumatoid arthritis, multiple myeloma, hypercalcemia of malignancy, hyperparathyroidism, and Gorham-Stout disease (reviewed in Ref. 18).
PTH-related protein (PTHrP), which is responsible for most cases of hypercalcemia of malignancy (19), also induces IL-6 secretion by osteoblasts (6,7). PTHrP and PTH utilize the same receptor to activate both the classical adenyl cyclase pathway as well as the phospholipase C pathway (20,21). Adenyl cyclase produces cAMP to activate protein kinase A, while phospholipase C produces inositol 1,4,5-trisphosphate and diacylglycerol to activate calcium/calmodulin-dependent kinase and protein kinase C, respectively. In order to fully understand the mechanism of action of PTH and PTHrP, it is important to clarify which signal transduction pathways are involved in the response of interest. PTH-(3-34), a truncated form of PTH, activates phospholipase C without stimulating cAMP production and can, therefore, be employed to determine whether cAMP signal transduction is involved in a particular response (22,23). We (17) and others (22) have utilized this partial agonist to show that activation of the cAMP pathway in osteoblasts is required for stimulation of resorption by PTH. Other workers have reached the same conclusion using inhibitors of the cAMP pathway (24,25).
Because stimulation of resorption by PTH depends on activation of the cAMP pathway and on IL-6 production, it is important to determine whether IL-6 production depends on cAMP signaling. Activation of either the protein kinase A, protein kinase C, or intracellular calcium pathways can stimulate IL-6 and LIF expression by osteoblasts (26) 2 or other mesenchymal cells (27)(28)(29)(30). Nonetheless, it is unknown which pathways are required for stimulation of IL-6 by PTH and PTHrP. In this study, we employed PTH-  to distinguish between these pathways. We show here that activation of the cAMP signaling pathway in osteoblasts is required for stimulation of IL-6 and LIF gene expression by PTH and PTHrP. Moreover, osteoblastic cAMP signaling is also required for stimulation of IL-6 and LIF by two other bone resorptive agents, namely vasoactive intestinal peptide (VIP) and isoproterenol.

MATERIALS AND METHODS
All reported data are representative of multiple experiments. PTH, PTHrP, VIP, and their analogues were obtained from Bachem (Torrance, CA); RNA and protein synthesis inhibitors, cAMP analogues, and ␤-adrenergic reagents were from Sigma. All of these agents were screened for endotoxin contamination using the colorimetric Limulus amoebocyte lysate assay (QCL-1000, Whittaker Bioproducts, Walkersville, MD). Endotoxin levels were Ͻ0.0008 Eu/ml for the highest concentration of each hormone used in the experiments, Ͻ0.015 Eu/ml for the highest concentration of 8-Br-cAMP used, Ͻ0.0075 Eu/ml for cycloheximide, and Ͻ0.034 Eu/ml for puromycin. Moreover, none of the stimulatory agents contained endotoxin contamination detectable by our previously described (3,17) functional assay of cytokine induction in MC3T3-E1 cells (see Fig. 9, for an example of this assay).
UMR106-01 rat osteoblast-like osteosarcoma cells (31) were kindly provided by Dr. N. Partridge (St. Louis University). UMR106-01 cells between passage 18 and 24 were harvested using 0.02% EDTA (Sigma) and cultured in phenol red-free (32) minimum essential media (Life Technologies, Inc., Gaithersburg, MD) containing non-essential amino acids (Mediatech, Herndon, VA), 2 mM L-glutamine (Mediatech, Herndon, VA), 10% heat-inactivated fetal bovine serum (HyClone, Logan, UT), 100 units/ml penicillin (Mediatech), and 100 g/ml streptomycin (Mediatech). All media, serum, and additives were from lots which contained the lowest concentration of endotoxin available. For studies of immediate-early gene responses, confluent UMR106-01 cultures were made quiescent by incubation in serum-free media containing 1 mg/ml bovine serum albumin (A8412, Sigma) for 48 h prior to addition of inhibitors, stimulators, and vehicles. For signal transduction studies, stimulators and vehicles were added 2 h after plating (25,000 cells/cm 2 ) UMR106-01 cells in order to mimic the conditions used in our previous studies of signal transduction during PTH-induced osteoclast activation (17). Murine MC3T3-E1 osteoblastic cells (33) were kindly provided by Dr. H. Tanaka (Okayama University) and experiments were performed 1 day after reaching confluence as described previously (3). In experiments using RNA or protein synthesis inhibitors, actinomycin D (10 g/ml), cycloheximide (10 g/ml), or puromycin (100 g/ml) were added immediately prior to addition of PTH. Spent media from both cell lines were consistently negative when assayed for mycoplasma contamination by solution hybridization to a probe complementary to mycoplasma rRNA (Gen-Probe, San Diego, CA).
IL-6 and LIF mRNA levels were assessed by RT-PCR as described previously (3). Murine IL-6 primers were previously described (3) and the rat IL-6 primers were modified versions of the mouse primers based on the published rat sequence (34). LIF primers were previously described (3) and were used for both mouse and rat, as they are based on the murine sequence (35) and have been used by other workers with UMR cells (36). Actin mRNA levels were employed in all experiments as a constitutive control using the PCR primers described in Ref. 37. All of the PCR primer pairs flanked at least one intron to avoid false positives due to amplification of genomic DNA. Controls without cDNA and with known positive cDNAs were used in all PCR reactions. To confirm the identity of the PCR products, we showed that fragments of appropriate sizes were produced by diagnostic restriction enzymes (HpaII, MboI, and HpaII for IL-6, LIF, and actin, respectively) as described previously (3). IL-6 protein was determined in supernatants harvested 24 h after addition of stimulators or vehicles. In these studies, cultures were rinsed twice with phosphate-buffered saline before adding stimulators and vehicles dissolved in the media described above except containing 0.5% fetal bovine serum. IL-6 bioactivity was assessed by measuring proliferation of the B cell hybridoma B13.29 (also known as B9) as described previously (3,7). Purified recombinant murine IL-6 (R&D Systems, Minneapolis, MN) was used in each bioassay to establish a standard curve and serve as a positive control. The presence of IL-6 was confirmed by preincubation of the culture supernatants for 30 -40 min prior to addition of the B9 cells with MP5-17D2, a neutralizing rat monoclonal anti-murine/rat IL-6 antibody obtained from Dr. J. Abrams (DNAX Research Institute) (38). mycin D (12,13). Fig. 1 shows that stimulation of IL-6 and LIF mRNA expression by PTH is inhibited by actinomycin D and that PTH has no detectable effect on stability of either cytokine mRNA. Although these results are consistent with an immediate-early response, conclusive evidence requires examination of protein synthesis inhibitors. Because immediate-early responses are primary effects that do not require synthesis of intermediate proteins, they are not blocked by inhibition of protein synthesis (12,13). Instead, immediate-early responses are superinduced; that is, they are increased in both duration and magnitude (12,13). Although cycloheximide is the classical protein synthesis inhibitor used in studies of immediate-early responses, it has recently been shown to activate the stressactivated protein kinase signal transduction pathway (39,40). Puromycin is a protein synthesis inhibitor with a distinct mode of action (41) that does not activate stress-activated protein kinase signal transduction (40). We therefore investigated the effects of both cycloheximide and puromycin. The two protein synthesis inhibitors gave indistinguishable results in both MC3T3-E1 and UMR106-01 cells. For example, Fig. 2 shows the effect of cycloheximide on MC3T3-E1 cells and Fig. 3 shows the effect of puromycin on UMR106-01 cells. In all cases, inhibition of protein synthesis superinduces the PTH-dependent stimulation of IL-6 and LIF mRNAs. In comparison, the inhibitors have only modest effects in the absence of PTH (Figs. 2 and 3). Taken together with the rapid and transient kinetics and the inhibition caused by actinomycin D, these results demonstrate that stimulation of IL-6 and LIF by PTH is a classical immediate-early gene response.

Stimulation of IL-6 and LIF Expression by PTH Is an
cAMP Signal Transduction Is Necessary and Sufficient for Stimulation of IL-6 and LIF Expression by PTH- Fig. 4 demonstrates that IL-6 and LIF mRNAs are stimulated by PTH in a dose-dependent manner; stimulation is detectable with concentrations as low as 0.025 nM PTH and is maximal at 25 nM. PTHrP also stimulates IL-6 and LIF mRNAs (Fig. 4) with a similar dose dependence (data not shown). To assess if cAMPmediated signal transduction is required for stimulation of IL-6 and LIF mRNA levels by PTH and PTHrP, we examined whether the effects of these agents are mimicked by PTH- . This partial agonist has no detectable effect on expression of IL-6 and LIF mRNAs even at 100 nM (Fig. 4). Since PTH-(3-34) activates phospholipase C but not adenyl cyclase (23,42), these results demonstrate that cAMP signal transduction is required for stimulation of IL-6 and LIF mRNA expression by PTH and PTHrP.
To further study the role of cAMP, we examined whether membrane-permeant analogues of cAMP effect IL-6 and LIF mRNA levels. Fig. 5 shows that 30 -300 M 8-Br-cAMP dose dependently stimulates expression of IL-6 and LIF. In contrast, 300 M 8-Br-cGMP, which was used as a negative control, has no detectable effect (Fig. 5). Expression of IL-6 and LIF is also stimulated by both mono-and dibutyrl-cAMP but not by the inactive analogue, monobutyryl cyclic deoxyadenosine monophosphate (data not shown).
To determine whether receptor-mediated intracellular cAMP is sufficient to induce IL-6 and LIF mRNA expression, we asked whether the cytokines are also induced by other bone resorptive agents that also stimulate cAMP in osteoblasts, namely VIP (43) and isoproterenol (44). VIP dose dependently stimulates expression of IL-6 and LIF (Fig. 6), with detectable and maximal effects at 0.1 and 10 nM VIP, respectively. The same concentrations of [4Cl-D-Phe 6 ,Leu 17 ]VIP (VIP-A), a weak VIP agonist (45), have lesser effects on the level of either cytokine mRNA (Fig. 6). Although VIP is thought to function primarily through the cAMP signal transduction system (46)  this potential complication, we examined the effect of the ␤-adrenergic agonist, isoproterenol, which specifically stimulates cAMP signal transduction (52). Isoproterenol dose dependently stimulates expression of both IL-6 and LIF mRNAs (Fig. 7), with detectable and maximal effects at 0.1 and 10 M, respectively. In contrast, the same concentrations of propranolol, a ␤-adrenergic antagonist which served as a negative control, did not stimulate expression of either cytokine (Fig. 7).
In order to assess whether the increases in IL-6 mRNA illustrated in Figs. 4 -7 are reflected in augmented IL-6 protein secretion, we employed the B9 hybridoma bioassay in conjunction with an antibody that neutralizes IL-6. We previously demonstrated that PTH-(1-34) stimulates IL-6 secretion using this method (3). Fig. 8 shows that all of the agents that stimulate expression of IL-6 mRNA (PTH-(1-34), PTHrP, 8-Br-cAMP, VIP, and isoproterenol) also induce secretion of IL-6 protein. Moreover, comparison to a standard curve constructed using purified recombinant murine IL-6 allowed estimation of the amount of IL-6 in each sample. These values were 0.14 Ϯ 0.009 pg/ml for the control cultures, 77.2 Ϯ 3.4 pg/ml for the PTH-treated cultures, 38.9 Ϯ 2.4 pg/ml for the PTHrP-treated cultures, 35.0 Ϯ 1.0 pg/ml for the 8-Br-cAMP-treated cultures, 4.2 Ϯ 0.04 pg/ml for the VIP-treated cultures, and 3.5 Ϯ 0.1 pg/ml for the isoproterenol-treated cultures. As expected, no effect was observed when the resorptive agents were added directly to the IL-6 bioassay (data not shown). Importantly, the IL-6 bioactivity secreted in response to these agents was reduced by Ͼ93% by an antibody that specifically neutralizes authentic IL-6 (data not shown). In contrast, an anti-IL-11 antibody (11h3/19.6.1 provided by Genetics Institute) had no detectable effect on the bioactivity in the media from cultures that had been treated with the resorptive agents (data not shown).
The results depicted in Figs. 4 -8 show results with UMR106-01 cells to provide compatibility with our previous work demonstrating that stimulation of osteoclast activity by PTH depends on cAMP and IL-6 (17). The same series of stimulators and controls were employed to determine whether the cAMP signal transduction pathway is also essential for induction of IL-6 and LIF mRNA in MC3T3-E1 cells (Fig. 9). As was seen in UMR106-01 cells, all of the agents that are potent stimulators of cAMP signaling (PTH-(1-34), PTHrP, VIP, 8-Br-cAMP, and isoproterenol) strongly increase expression of IL-6 and LIF mRNAs, while the control compounds (PTH-(3-34), VIP-A, 8-Br-cGMP, and propranolol) had little or no effect on expression of IL-6 and LIF mRNAs. Fig. 9 also shows that none of these agents detectably stimulate tumor necrosis factor ␣ mRNA. Thus, stimulation of IL-6 and LIF mRNAs by these agents cannot be attributed to endotoxin contamination, since endotoxin strongly stimulates expression of tumor necrosis factor ␣ mRNA (Fig. 9). This finding confirms the results of the colorimetric endotoxin assay described under "Materials and Methods." DISCUSSION We previously showed that PTH rapidly and transiently stimulates expression of IL-6 and LIF mRNAs (3, 11), reminiscent of classical immediate-early responses (12,13). The cur-  For PTH-treated cultures, 0.03% conditioned media was p Ͻ 0.05 and 0.1-10% were p Ͻ 0.0005; for PTHrP, 0.03% was not significant and 0.1-10% were p Ͻ 0.0005; for 8-Br-cAMP, 0.03% was p Ͻ 0.05 and 0.1-10% were p Ͻ 0.0005; for VIP, 0.3 and 1% were not significant, 0.03 and 0.1% were p Ͻ 0.005, and 3 and 10% were p Ͻ 0.0005; for ISO, 0.3 and 1% were not significant, 0.03 and 3% were p Ͻ 0.05, and 0.1 and 10% were p Ͻ 0.0005. Replicate dishes were incubated with identical stimulators and vehicles for 1 h prior to extraction of RNA and RT-PCR analysis (top).
cAMP Stimulates Immediate-early Expression of IL-6 and LIF rent study demonstrates that induction of these mRNAs by PTH exhibits the other characteristics of immediate-early responses; that is, it is inhibited by actinomycin D and superinduced by protein synthesis inhibitors (12,13). Consistent with our conclusion that this represents an immediate-early gene response, both genes contain multiple 3Ј-AUUUA sequences that direct rapid degradation of the previously accumulated mRNAs (53,54). Moreover, the 3Ј-untranslated regions of both the IL-6 and LIF genes (35,55) contain multiple copies of UUUUGUA, a recently identified motif which is responsible for induction of the immediate-early gene, JE, and is found in many other immediate-early genes (56). Since we also show here that induction of IL-6 and LIF mRNAs by PTH is dependent on cAMP signaling, we hypothesize that the transient nature of these responses, like other examples of transient cAMPstimulated transcription (57,58), is dependent on inactivation of relevant transcription factors by serine/threonine protein phosphatases.
Since stimulation of osteoclast activity by PTH depends on both cAMP signal transduction and IL-6 production (17), it is important to evaluate whether IL-6 production depends on cAMP. In the current study, we demonstrate that cAMP signal transduction is required for stimulation by PTH of both IL-6 and LIF. Moreover, PTHrP, VIP, and isoproterenol, other bone resorptive agents that stimulate cAMP signal transduction in osteoblasts (19,43,44), also induce IL-6 and LIF expression by activating cAMP signal transduction. VIP and isoproterenol also stimulate IL-6 production in anterior pituitary cells (59) and astrocytes (60,61). In further support of a role for cAMP, the promoter region of the IL-6 gene contains multiple regulatory elements that cooperatively mediate induction by cAMP (62) and the LIF promoter contains a consensus AP-2 site (35), which confers cAMP inducibility to other genes (63)(64)(65). Moreover, it has recently been reported that a cAMP response element is required for stimulation of c-fos transcription by PTH in UMR106-01 cells (66).
Although the results presented here demonstrate that the PTH/PTHrP responses depend on cAMP, they do not exclude a role for the phospholipase C pathway. We have, however, been unable to inhibit induction of IL-6 and LIF mRNAs by submaximal concentrations of PTH (500 pM) with staurosporine (10 nM), dantrolene (10 M), or lanthanum (100 M), either alone or in combination (data not shown). These agents block protein kinase C activation (67), intracellular calcium release (68), and uptake of extracellular calcium (69, 70), respectively.
These results are consistent with the demonstration that 1-alkyl-2-methylglycerol, a specific protein kinase C inhibitor, does not block the ability of PTH to stimulate IL-6 secretion (71). We cannot rule out the possibility that the inhibitors were not fully effective and that residual calcium or protein kinase C signal transduction may have been involved in stimulation of IL-6 and LIF by PTH. Nonetheless, the current study demonstrates that if the phospholipase C pathways are involved, they act together with cAMP.
Taken together with our previous work demonstrating that IL-6 is involved in stimulation of osteoclast activity by PTH (17), this study suggests that IL-6 is also involved in stimulation of resorption by PTHrP, VIP, and ␤-adrenergic agonists. All of these agents are thought to stimulate resorption through activation of osteoblast cAMP signal transduction (19,43,44). VIP is particularly interesting in this regard, since high concentrations of this molecule are found near focal bone resorption sites induced by tooth movement (72). Thus, in response to mechanical forces, sympathetic nerves innervating bone and periosteum (73) may secrete VIP, which would stimulate production by osteoblasts of factors such as IL-6 that regulate bone resorption. It has also been proposed that VIP may be involved in the stimulation of bone resorption observed in reflex sympathetic dystrophy (74). Similarity of function between PTH and VIP likely reflect homology between the intracellular signaling domains of the two receptors (75). In fact, it has recently been reported that PTH and VIP antagonists are able to inhibit activation of each other's receptor (76).
In summary, this study demonstrates that induction of IL-6 and LIF by PTH is an immediate-early gene response that depends on cAMP signal transduction. Moreover, PTHrP, VIP, and isoproterenol also induce IL-6 and LIF expression through cAMP signal transduction. These results suggest that stimulation of osteoclast activity by PTHrP, VIP, and isoproterenol, like that by PTH, is mediated by production of IL-6 and possibly LIF. We and others have also shown that many resorptive agents that activate signaling pathways that do not involve cAMP also stimulate production of IL-6 by osteoblasts. These include IL-1, IL-6, LIF, tumor necrosis factor, platelet-derived growth factor, 1,25-dihydroxyvitamin D 3 , transforming growth factor ␤, endothelin, and lipopolysaccharide (reviewed in Ref. 18). IL-1, IL-6, tumor necrosis factor, transforming growth factor ␤, and lipopolysaccharide have also been shown to stimulate LIF expression by osteoblasts (3,4,10,16,17,36,77). Thus, production of IL-6, and perhaps LIF, may be critically involved in stimulation of resorption by a wide variety of agents in addition to those examined in this study that activate cAMP signal transduction. Alternatively, LIF may be involved in stimulation of bone formation by PTH (78) since LIF increases osteoblastic differentiation (36,79) while IL-6 has been reported to inhibit this process (80).