A Novel Distal Enhancer Mediates Cytokine Induction of Mouse Rankl Gene Expression

Abstract

Chronic inflammatory states are associated with increased bone loss. This increase is often linked to an elevation in receptor activator of nuclear factor-κB ligand (RANKL), a TNFα-like factor essential to osteoclast formation. In this study, we document the ability of IL-6 in combination with IL-6 soluble receptor (IL-6/IL-6sR) and oncostatin M to induce Rankl expression in stromal cells via signal transducer and activator of transcription 3 (STAT3). We used chromatin immunoprecipitation-tiled DNA microarray analysis to determine sites of action of STAT3 at the Rankl locus and to assess the consequences of binding on histone H4 acetylation and RNA polymerase II recruitment. Both IL-6/IL-6 soluble receptor and oncostatin M stimulated STAT3 binding upstream of the Rankl transcriptional start site. Although previously identified enhancers bound STAT3, a more distal enhancer termed mRLD6 was a particular focus of STAT3 binding. When fused to a heterologous promoter, this enhancer was highly active, containing two functionally active STAT response elements. Importantly, small interfering RNA knockdown of Stat3 mRNA and protein, but not that of Stat1 or Stat5a, was effective in limiting Rankl mRNA up-regulation. Interestingly, although RNA polymerase II and histone H4 acetylation marked many of the enhancers under basal conditions, the levels of both were strongly increased after cytokine treatment, particularly at mRLD6. Finally, mRLD6 was also a target for forskolin-induced cellular response element-binding protein (CREB) recruitment, which potentiated cytokine activity. Our studies provide new insight into mechanisms by which glycoprotein 130 activating cytokines induce RANKL expression.

GP130-activating cytokines regulate the expression of the mouse Rankl gene through a strong distal enhancer that contains two closely spaced STAT3 response elements.

Skeletal homeostasis is accomplished through the maintenance of a delicate balance between osteoblast-mediated bone formation and osteoclast-mediated bone resorption. A disruption in this balance leads to changes in bone mineral density (BMD) that often alters bone strength and can lead to osteoporosis and fractures. Receptor activator of nuclear factor-κB ligand (RANKL) has been identified as the primary cytokine essential for osteoclast formation (1,2). Expressed on the surface of stromal cells, RANKL interacts with receptor activator of nuclear factor-κB (RANK) on osteoclast precursors, stimulating their differentiation into functional osteoclasts and promoting the activation and survival of mature osteoclasts (3,4). Normal RANKL expression stimulates bone breakdown allowing for the removal of damaged bone and its eventual replacement with new matrix during the remodeling process. During chronic inflammatory diseases (CIDs), bone homeostasis is disrupted and favors bone resorption. Increased cytokine production during CIDs increases RANKL expression and decreases BMD (5,6,7). Thus, the regulation of RANKL by cytokines plays an important role in the local osteolysis and systemic osteoporosis associated with many CIDs. Understanding the molecular mechanisms of cytokine-induced Rankl expression not only is relevant to complications of CIDs but also provides a model system for understanding complex transcriptional regulation of gene expression by distal enhancer regions.

Inflammation-induced bone loss is associated with several CIDs and autoimmune diseases including rheumatoid arthritis, inflammatory bowel disease (IBD), and chronic obstructive pulmonary disease among others (8,9,10,11). Rheumatoid arthritis results in site-specific bone loss (12,13) whereas chronic obstructive pulmonary disease (10,11) and IBD are associated with an overall decrease in bone mass in humans. IBD causes a 33% decrease in bone mass in rats (14). It is well established that inflammation leads to increased levels of many proinflammatory cytokines including TNFα, IL-1α, IL-1β, IL-6, IL-11, IL-17, and oncostatin M (OSM) (15,16). Although TNFα is the primary cytokine associated with inflammation-induced bone loss, both TNFα and IL-1α also induce the expression of IL-6 (17). IL-6 levels are inversely correlated with BMD (18,19), whereas both IL-6 and IL-6 soluble receptor (IL-6sR) levels are correlated with rates of bone loss (20). Both observations suggest that this cytokine, its soluble receptor, and other members of the glycoprotein 130 (gp130)-activating cytokine family play an integral part in bone loss initiated by inflammation.

Both stromal cells and T lymphocytes are involved in enhanced RANKL-mediated osteoclast formation. The mechanism of osteoclast activation by osteoblasts is a direct result of the close communication between osteoblasts and osteoclasts in the bone microenvironment (3,4,21), whereas T lymphocytes produce a soluble form of RANKL upon activation (22,23). Although RANKL expression is better characterized in osteoblasts than in T lymphocytes, both sources of RANKL play key roles in osteoclast activation and inflammation-induced bone resorption. With respect to stromal cells, the gp130-activating cytokine family including IL-6, OSM, leukemia inhibitory factor (LIF), and IL-11 also play primary roles in this process (24). RANKL expression is induced in stroma/osteoblasts after exposure to gp130-activating cytokines, specifically IL-6 and OSM. Indeed, binding of STAT3, one of the major transcription factor targets of the gp130-activated cytokines, to the mouse Rankl locus has been observed in osteoblasts with OSM treatment (25,26). Treatment with an IL-6-neutralizing antibody (toxilizumab) has been shown to reduce RANKL expression from stromal cells (27). Contributions from both osteoblasts and T lymphocytes elevate overall RANKL levels, increasing the interaction of RANKL with RANK to initiate the processes of osteoclast differentiation and activation.

The regulation of transcription by gp130-activating cytokines is relatively well understood. OSM and IL-6 bind to their specific sets of surface receptors and form dimers that autophosphorylate the Janus kinases JAK1 and JAK2, which in turn phosphorylate the intracellular receptor tails. Signal transducer and activator of transcription (STAT) transcription factors 1, 3, and/or 5 are recruited to these phosphorylation sites and become phosphorylated as well. Activated STATs form homo- or heterodimers, translocate into the nucleus, bind to DNA regulatory elements comprised of the consensus TTCn_(2-4)GAA and modulate transcription (21,24,28,29). STAT3 has two phosphorylation sites, Tyr705, which is associated with both receptor dimerization and nuclear translocation, and Ser727, which is linked to the recruitment of coactivators (30,31). Phosphorylation at both sites is necessary for complete transactivation.

Transcriptional regulation of RANKL has been extensively studied with respect to 1,25-dihydroxyvitamin D₃ [1,25-(OH)₂D₃] and PTH. Five distal enhancer regions have been identified and shown to be direct targets of activated vitamin D receptor (VDR), cellular response element-binding protein (CREB), or both. The mRLD5a/b region, located −75 to −77 kb upstream of the mouse Rankl transcriptional start site (TSS), is the most highly active with respect to 1,25-(OH)₂D₃ and PTH treatment (32,33). OSM treatment also induces STAT3 accumulation at several of these enhancer regions (mRLD4 and mRLD5), but the overall transcriptional activity at these sites appears to be relatively low (25). Indeed, deletion of a fragment containing the mRLD5a/b region resulted in only minor affects on OSM-induced Rankl reporter regulation. However, deletion of a 10-kb fragment located between −91 and −82 kb upstream of the Rankl TSS in a Rankl BAC clone construct almost completely eliminated OSM-induced Rankl reporter expression (32). These results suggest the possibility that an additional regulatory region upstream of mRLD5 might also exist that could play a dominant role in cytokine-mediated Rankl expression.

In this report, we use chromatin immunoprecipitation-tiled DNA microarray (ChIP-chip) analysis to further understand the regulation of Rankl expression by gp130-activating cytokines. We show that these cytokines induce STAT3 binding to previously defined enhancers and also to a new region −88 kb upstream of the Rankl TSS designated mRLD6. This region contains two STAT regulatory elements that mediate transcription synergistically. This binding alters the levels of both histone H4 (H4) acetylation (H4ac) and induces the recruitment of RNA polymerase II (RNA pol II). We conclude that this novel distal enhancer plays a key role in cytokine-induced Rankl gene expression.

Results

Rankl transcript levels increase with OSM and IL-6/IL-6sR treatment in the ST2 cell line

We have used the ST2 mouse stromal cell line to study the regulation of Rankl gene expression by a variety of hormonal inducers including 1,25-(OH)₂D₃ and the PTH surrogate forskolin (33). We and others have also used this cell line to explore the capacity of cytokines such as OSM to induce Rankl (25). To confirm and extend these observations, we treated ST2 cells with either vehicle or increasing concentrations of OSM, isolated total RNA from these cells 6 h later, and examined the concentrations of Rankl mRNA using real-time RT-PCR. As can be seen in Fig. 1A, OSM strongly induced Rankl expression at 6 h with peak activity associated with 10 ng/ml OSM. In a similar experiment, the gp130-activating cytokine IL-6 was also active in inducing Rankl expression in ST2 cells as seen in Fig. 1B. This activation, however, required the addition of IL-6sR, an observation consistent with previous studies (34,35) that suggest that stromal cells are generally deficient in expression of the IL-6 receptor component that forms an activating complex with membrane gp130 receptors (21). Thus, although neither IL-6 nor IL-6sR alone were capable of inducing Rankl transcripts, a combination of 25 ng/ml IL-6 and 100 ng/ml IL-6sR resulted in peak activation. The level of activation by IL-6/IL-6sR was not as substantial as that provoked by OSM; however, both manifested dose-dependent profiles. We confirm here that OSM induces Rankl expression and show that IL-6/IL-6sR is active as well. Thus, ST2 cells likely provide an excellent model for exploring the mechanistic basis for cytokine-induced Rankl gene expression.

Figure 1.

Induction of Rankl mRNA levels in ST2 cells is enhanced by OSM or IL-6/IL-6sR. A, Induction of Rankl transcript levels by OSM in vitro. ST2 cells were treated with OSM concentrations ranging from 0.5–20 ng/ml for 6 h. B, Induction of Rankl transcript levels by IL-6/IL-6sR in vitro. ST2 cells were treated with 50 ng/ml IL-6 or 100 ng/ml IL-6sR or cotreated with 100 ng/ml IL-6sR and increasing concentrations of IL-6 ranging from 5–50 ng/ml for 6 h. Total RNA was isolated, reverse transcribed, and analyzed by quantitative PCR using primers specific to Rankl and β-actin. Rankl transcript levels were normalized to β-actin. Each value represents the average of three independent experiments ± sem compared with control samples using a one-way nonparametric ANOVA analysis with single variance followed by a Tukey multiple-comparison post-test. *, P < 0.05.

ChIP-chip analysis of ST2 cells treated with OSM and IL-6/IL-6sR identifies a Rankl enhancer region located −88 kb upstream of the Rankl TSS

As summarized earlier, the activity of the gp130- activating cytokines is mediated by the STAT family of transcription factors that include STAT1, STAT3, and STAT5. Indeed, the direct involvement of STAT3 in OSM-induced Rankl expression has been previously established (25), although the specific sites of action of this factor induced by OSM at Rankl gene locus have not been defined. More recently, however, we examined the ability of OSM to induce STAT3 binding to specific regions of the Rankl gene using ChIP analysis (25). This analysis revealed STAT3 binding to the mRLD4 and mRLD5 regions but was restricted by its site-specific nature to previously described Rankl enhancers. To obtain a complete profile of OSM- or IL-6-induced STAT3 binding at the mouse Rankl gene locus, we conducted a ChIP-chip scan from 200 kb upstream of the Rankl TSS to approximately 200 kb downstream of the final 3′ noncoding Rankl exon. ST2 cells were treated with vehicle, 20 ng/ml OSM, or 50 ng/ml IL-6 and 100 ng/ml IL-6sR for 6 h and then subjected to ChIP using antibodies to STAT3, phospho-STAT3 (Tyr705) (pSTAT3), or IgG. Isolated DNA from the precipitated samples was amplified using ligation- mediated PCR, labeled with fluorescence dyes as described in Materials and Methods, and then cohybridized overnight to the custom microarrays. Figure 2 (top) depicts a schematic of the mouse Rankl gene locus on chromosome 14. The locations of previously identified enhancers designated mRLD1-mRLD5a/b are indicated. Figure 2A depicts a data track representing the log₂ ratio of fluorescence obtained from a vehicle-treated sample precipitated with antibodies to either pSTAT3 or IgG (pSTAT3_veh vs. IgG) (basal pSTAT3). Figure 2B depicts similar tracks representing 1) an OSM-treated sample precipitated with antibodies to either pSTAT3 or IgG (pSTAT3_OSM vs. IgG) (total pSTAT3) or 2) OSM-treated vs. vehicle-treated samples precipitated with antibody to pSTAT3 (pSTAT3_OSM vs. pSTAT3_veh) (net inducible pSTAT3). Figure 2C depicts comparable data tracks after induction by either vehicle or IL-6/IL-6sR. Data obtained outside the 77.000- to 77.150-Mb segment were unremarkable. pSTAT3 binding at the Rankl gene locus was strongly induced by both OSM and IL-6/IL-6sR, as can be seen in Fig. 2, B and 2C (tracks 1 and 2). Although both OSM- and IL-6/IL-6sR-inducible pSTAT3 binding was observed at mRLD2, mRLD4, and mRLD5a/b, as described previously for OSM alone (25), the transcription factor was most strikingly recruited to a site that we termed mRLD6 located upstream of mRLD5a/b at −88 kb upstream of the Rankl TSS. Similar observations for both inducers were obtained using an antibody recognizing both the nonphosphorylated and phosphorylated forms of STAT3 (data not shown). The levels of OSM- and IL-6-inducible pSTAT3 are depicted in Fig. 2, B and C (tracks 2). They contrast with levels of pSTAT3 obtained at the Rankl locus in the absence of inducer (Fig. 2A). Interestingly, both inducers appear to stimulate recruitment of pSTAT3 to a site surrounding the proximal Rankl promoter as well. Our results confirm previous findings of the importance of mRLD2, mRLD4, and mRLD5 and also show that the activities of OSM and IL-6/IL-6sR may also be mediated by a new enhancer located even further upstream of the TSS.

Figure 2.

ChIP-chip analysis reveals localization of pSTAT3 to the Rankl gene locus in response to the gp130 cytokines OSM and IL-6/IL-6sR. ST2 cells were treated with 20 ng/ml OSM or 50 ng/ml IL-6 and 100 ng/ml IL-6sR (IL-6/R) for 6 h and then subjected to ChIP analysis using antibodies to pSTAT3-Tyr705 or IgG. Immunoprecipitated DNA was amplified by ligation-mediated PCR, labeled with Cy3 or Cy5, and cohybridized to custom DNA microarrays as described in Materials and Methods. The upper panel depicts the Rankl gene locus. Nucleotide positions (Mb) are shown on chromosome 14 (chr14; February 2006 assembly), whereas the Rankl gene and the positions of the distal enhancer regions are indicated below the chromosomal position and designated by descending gray bands. A, Interaction of pSTAT3 with the Rankl gene locus under basal conditions. The data track represents the log₂ ratios of fluorescence obtained from a vehicle-treated (Veh) sample precipitated with antibodies to either pSTAT3 or IgG (pSTAT3_veh vs. IgG) (basal pSTAT3). B, Interaction of pSTAT3 with the Rankl gene locus after treatment with OSM. The data tracks represent the log₂ ratios of fluorescence obtained from an OSM-treated sample precipitated with antibodies to either pSTAT3 or IgG (pSTAT3_OSM vs. IgG) (total pSTAT3) or samples treated with either vehicle or OSM and precipitated with an antibody to pSTAT3 (pSTAT3_OSM vs. pSTAT3_veh) (net inducible pSTAT3). C, Interaction of pSTAT3 with the Rankl gene locus in response to IL-6/IL-6sR. Data tracks represent the log₂ ratios of fluorescence as in B with IL-6/IL-6sR as the inducer. All peaks highlighted in red represent statistically significant peaks [false discovery rate (FDR), P < 0.05].

OSM and IL-6/IL-6sR induce significant RNA pol II recruitment in the Rankl locus at mRLD6

Our previous studies using site-specific ChIP analysis suggested that RNA pol II was recruited to many of the upstream Rankl enhancers and perhaps to regions even further upstream as well (25). These data provided additional support for the functional validity of these distal enhancers and prompted the hypothesis that they might function as recruitment centers for RNA pol II. To explore both the basal and inducible levels of RNA pol II in high resolution across the Rankl locus, we treated ST2 cells with either OSM or IL-6/IL-6sR and conducted a ChIP-chip analysis as described in Fig. 2 using antibodies to either RNA pol II or IgG. Figure 3A depicts a data track representing the log₂ ratio of fluorescence obtained from a vehicle-treated sample precipitated with antibodies to either RNA pol II or IgG (RNA pol II_veh vs. IgG) (basal RNA pol II). Figure 3B depicts similar tracks representing 1) an OSM-treated sample precipitated with antibodies to either RNA pol II or IgG (RNA pol II_OSM vs. IgG) (total RNA pol II) or 2) OSM-treated vs. vehicle-treated samples precipitated with antibody to RNA pol II (RNA pol II_OSM vs. RNA pol II_veh) (net inducible RNA pol II). Figure 3C depicts comparable data tracks after induction by either vehicle or IL-6/IL-6sR. As can be seen in Fig. 3A, RNA pol II is present at several of the Rankl enhancers under basal conditions, although most prominently at mRLD2. Surprisingly, the enzyme does not accumulate at the TSS. Both OSM and IL-6/IL-6sR induce the recruitment of RNA pol II at enhancers in the Rankl gene locus (see Fig. 3, B and C, tracks 1 and 2). Interestingly, however, although the activity of IL-6/IL-6sR is somewhat less than that observed for OSM, recruitment is limited largely to the mRLD4, mRLD5, and particularly the mRLD6 enhancers. RNA pol II recruitment is substantial at these sites for OSM but more limited to the mRLD6 enhancer after IL-6/IL-6sR treatment, consistent with the reduced ability of IL-6/IL-6sR to induce Rankl transcripts. These observations suggest that the overall activity of the cytokines is restricted to the more distal enhancers with a primary focus on mRLD6. Surprisingly, neither cytokine induced a strong net accumulation of RNA pol II at the Rankl TSS.

Figure 3.

ChIP-chip analysis reveals the presence and recruitment of RNA pol II at the Rankl gene locus in response to the gp130-activating cytokines OSM and IL-6. ST2 cells were treated with 20 ng/ml OSM or 50 ng/ml IL-6 and 100 ng/ml IL-6sR (IL-6/R) for 6 h and then subjected to ChIP analysis using antibodies to RNA pol II or IgG. Samples were prepared as described in Materials and Methods. The upper panel depicts the Rankl gene locus as in Fig. 2. A, Interaction of RNA pol II with the Rankl gene locus under basal conditions. The data track represents the log₂ ratios of fluorescence obtained from a vehicle-treated (Veh) sample precipitated with antibodies to either RNA pol II or IgG (RNA pol II_veh vs. IgG) (basal RNA pol II). B, Interaction of RNA pol II with the Rankl gene locus after treatment with OSM. The data tracks represent the log₂ ratios of fluorescence obtained from an OSM-treated sample precipitated with antibodies to either RNA pol II or IgG (RNA pol II_OSM vs. IgG) (total RNA pol II) or samples treated with either vehicle or OSM and precipitated with an antibody to RNA pol II (RNA pol II_OSM vs. RNA pol II_veh) (net inducible RNA pol II). C, Interaction of RNA pol II with the Rankl gene locus in response to IL-6/IL-6sR. Data tracks represent the log₂ ratios of fluorescence as in B with IL-6/IL-6sR as the inducer. All regions highlighted in red represent statistically significant peaks [false discovery rate (FDR), P < 0.05].

OSM and IL-6/IL-6sR induced significant H4ac in the Rankl locus at mRLD6

Transcription factors are believed to enhance gene expression through the direct recruitment of coactivator complexes, a subset of which function by increasing local levels of histone acetylation. Indeed, we have previously shown that the level of H4ac is increased at specific sites across the Rankl locus in response to 1,25-(OH)₂D₃ or the PTH surrogate forskolin (25,33), although the panoramic nature of this induction was unclear. To assess both basal as well as cytokine-inducible levels of H4ac across the Rankl locus, we treated ST2 cells with vehicle, OSM, or IL-6/IL-6sR and conducted a ChIP-chip analysis as described in Fig. 2 using antibodies to either tetra-acetylated H4 or IgG. Figure 4A depicts a data track representing the log₂ ratio of fluorescence obtained from a vehicle-treated sample precipitated with antibodies to either H4ac or IgG (H4ac_veh vs. IgG) (basal H4ac). Figure 4B depicts similar tracks representing 1) an OSM-treated sample precipitated with antibodies to either H4ac or IgG (H4ac_OSM vs. IgG) (total H4ac) or 2) OSM-treated vs. vehicle-treated samples precipitated with antibody to H4ac (H4ac_OSM vs. H4ac_veh) (net inducible H4ac). Figure 4C depicts comparable data tracks after induction by IL-6/IL-6sR treatment. As can be seen in Fig. 4A, high levels of H4ac are evident under basal conditions across the Rankl locus, centered largely although not exclusively at regions associated with previously identified Rankl enhancers and at the TSS. Interestingly, both OSM and IL-6/IL-6sR promote an increase in H4ac levels as seen in Fig. 4, B and C (tracks 1 and 2), respectively. This increase in H4ac is largely focused at the distal regions comprising mRLD4, mRLD5, and particularly at the distal region mRLD6. The increase in H4ac levels could be due potentially to the recruitment of additional H4ac units to the locus but is more likely due to an increase in the levels of H4 acetylation at the enhancer regions. Despite the absence of RNA pol II, OSM also induced H4ac at the TSS as well. The ability of OSM and IL-6/IL-6sR to induce an increase in acetylation that is largely restricted to mRLD5 and mRLD6 provides increased evidence for a role for these enhancers in cytokine-mediated induction of Rankl expression in stromal-like cells. These data also provide an explanation for the unusual pattern of elevated H4ac levels we observed in previous studies using direct ChIP analysis (25).

Figure 4.

ChIP-chip analysis reveals basal and inducible levels of H4ac at the Rankl gene locus in response to the gp130-activating cytokines OSM and IL-6/IL-6sR. ST2 cells were treated with 20 ng/ml OSM or 50 ng/ml IL-6 and 100 ng/ml IL-6sR (IL-6/R) for 6 h and then subjected to ChIP analysis using antibodies to H4ac or IgG. A, Levels of H4ac with the Rankl gene locus under basal conditions. The data track represents the log₂ ratios of fluorescence obtained from a vehicle-treated sample precipitated with antibodies to either H4ac or IgG (H4ac_veh vs. IgG) (basal H4ac). B, the Rankl gene locus after treatment with OSM. The data tracks represent the log₂ ratios of fluorescence obtained from an OSM-treated sample precipitated with antibodies to either H4ac or IgG (H4ac_OSM vs. IgG) (total H4ac) or samples treated with either vehicle or OSM and precipitated with an antibody to H4ac (H4ac_OSM vs. H4ac_veh) (net inducible H4ac). C, Interaction of H4ac with the Rankl gene locus in response to IL-6/IL-6sR. Data tracks represent the log₂ ratios of fluorescence as in B with IL-6 as the inducer. All regions highlighted in red represent statistically significant peaks [false discovery rate (FDR), P < 0.05].

The Rankl mRLD6 enhancer region shows strong transcription activation in reporter assays

The mRLD6 enhancer region, like other regulatory regions in the Rankl locus, is strongly conserved across multiple species as observed in supplemental Fig. 1 (published as supplemental data on The Endocrine Society’s Journals Online web site at http://mend.endojournals.org). This conservation together with the considerable activity identified at mRLD6, as determined by ChIP-chip analysis, suggests the possibility that the mRLD6 enhancer may play a central role in cytokine-stimulated up-regulation of Rankl gene expression. To explore this possibility further, we cloned an approximately 800-bp segment containing mRLD6 into a TK promoter-luciferase reporter plasmid, introduced this plasmid transiently into ST2 cells, and assessed luciferase activity 20 h after treatment with increasing concentrations of OSM. We also examined similar constructs cloned earlier containing the mRLD1-mRLD4, mRLD5a, and mRLD5b regions (25,33). Figure 5 reveals that although mRLD5a is modestly inducible by OSM above the 2-fold level observed in the parent vector, the plasmid containing the mRLD6 segment is strikingly up-regulated 29-fold in response to OSM. These experimental results indicate that the mRLD6 region not only is capable of binding pSTAT3 (Fig. 2, B and C), but also can regulate transcription after OSM treatment.

Figure 5.

The mRLD6 enhancer region together with the D5a region are transcriptionally active in response to OSM. ST2 cells were transfected with pTK-luc control or the indicated Rankl enhancer region reporter constructs (250 ng) and pCH110-βgal (50 ng) in triplicate. Cells were treated for 20 h with 0.5–20 ng/ml OSM and harvested, and both luciferase activity and β-galactosidase activity were measured. Luciferase activity was normalized to β-galactosidase activity. Each value represents the average of three independent experiments ± sem compared with control samples using a one-way nonparametric ANOVA analysis with single variance followed by a Tukey multiple-comparison post-test. *, P < 0.05.

Two STAT-binding elements in the distal portion of the mRLD6 enhancer region control transcriptional activation synergistically by OSM and IL-6/IL-6sR

In view of the potential importance of this regulatory region in cytokine-induced Rankl expression, we began by searching for STAT regulatory elements. Both the mouse and human RANKL genes are highly conserved, as indicated in supplemental Fig. 1. The mRLD6 enhancer region contains two highly conserved approximately 200-bp segments separated by approximately 350 bp of nonconserved sequence; we cloned both the distal and proximal segment of mRLD6, identified schematically in Fig. 6A, into the TK promoter-luciferase reporter vector, and assessed their activities in response to OSM or IL-6/IL-6sR after transient transfection into ST2 cells. As can be seen in Fig. 6B, only the distal segment of mRLD6 was induced in response to cytokine treatment. Surprisingly, its activity was far greater than that observed with the full-length mRLD6 enhancer. These data suggest that STAT3 elements involved in transcriptional activation of the mRLD6 enhancer region are likely to be contained within the distal conserved portion. Indeed, in silico analysis using TFSearch (http://www.cbrc.jp/research/db/TFSEARCH.html) revealed the presence of two highly conserved potential STAT-binding elements separated by 52 bp as documented in Fig. 6C and similar in sequence to the consensus STAT-binding element TTCn_(2–4)GAA (36). To assess whether these putative elements conferred STAT3-mediated reporter response, we introduced inactivating mutations into each element within the parent distal mRLD6 vector as well as in combination, transfected the constructs into ST2 cells, and assessed their abilities to mediate reporter induction by OSM or IL-6/IL-6sR. As can be seen in Fig. 6, D and E, the introduction of an inactivating mutation in either regulatory element strongly reduced but did not fully eliminate the capacity of the mRLD6 region to mediate OSM (Fig. 6D) and IL-6/IL-6sR (Fig. 6E) induction. Mutations in both elements, however, eliminated all response to either OSM or IL-6/IL-6sR. This striking reduction in inducible response suggests the possibility that both elements function in a synergistic manner to up-regulate transcription. We conclude from these studies that two closely spaced regulatory elements in the mRLD6 enhancer likely provide binding sites for STAT3 and mediate the actions of OSM and IL-6/IL-6sR in cis at the Rankl locus. The overall activity of the mRLD6 enhancer suggests that this region together with the weaker activity manifested by mRLD5a and perhaps mRLD4 may represent the primary means by which cytokines such as OSM and IL-6/IL-6sR induce Rankl gene expression.

Figure 6.

The gp130 cytokine regulation of the mRLD6 enhancer region is mediated by synergistic STAT-binding elements. A, Conserved regions of mRLD6. University of California, Santa Cruz Genome Browser (http://genome.uscs.edu)-derived conservation plot of the Rankl mRLD6 enhancer region (February 2006 assembly) shows location and size of the cloned mRLD6 fragments. B, Transcriptional activity of mRLD6 subfragments. The pTK-luc vector, pTK-mRLD6, pTK-mRLD6 proximal, and pTK-mRLD6 distal reporter constructs (250 ng) were transfected into ST2 cells with pCH110-βgal (50 ng). Cells were treated for 20 h with 20 ng/ml OSM or 50 ng/ml IL-6 and 100 ng/ml IL-6sR, and luciferase activity was assessed and normalized as described in Materials and Methods. C, Conservation of STAT response elements and identification of mutations. Conservation comparison of putative STAT-binding elements located in the mRLD6 region of several species is depicted using the TFSearch algorithm (36). The mutations shown were introduced into the pTK-luc mRLD6 distal enhancer region using site-directed mutagenesis. D and E, STAT3-mediated cytokine activity is abrogated upon mutation of the STAT-binding elements. The pTK-luc vector, pTKmRLD6 distal, pTKmRLD6 distal mut1, pTKmRLD6 distal mut2, or pTKmRLD6 distal mut1/2 reporter constructs (250 ng) were transfected into ST2 cells with pCH110-βgal (50 ng). Cells were treated with 20 ng/ml OSM (D), 50 ng/ml IL-6, or 100 ng/ml IL-6sR or cotreated with 100 ng/ml IL-6sR with IL-6 concentrations ranging from 5–50 ng/ml (E). Cells were harvested after 20 h, and both luciferase activity and β-galactosidase activity were measured. Luciferase activity was normalized to β-galactosidase activity. Each value represents the average of three independent experiments ± sem compared with control samples using a one-way nonparametric ANOVA analysis with single variance followed by a Tukey multiple-comparison post-test. *, P < 0.05. chr14, Chromosome 14.

STAT3 activity is central to OSM-induced Rankl expression and mediated via the STAT elements identified in mRLD6

Previous work has suggested that the actions of OSM and IL-6/IL-6sR to induce Rankl expression are mediated via the STAT3 transcription factor (25,26). These cytokines are known to activate not only STAT3, however, but STAT1 and STAT5a as well. There is also evidence that these factors can function as both homodimers as well as selected heterodimers. Based upon these observations, we explored the possibility that in addition to STAT3, OSM or IL-6/IL-6sR might also regulate STAT1 and/or STAT5a activity in stromal cells as well. In a first set of experiments, ST2 cells were preincubated for 48 h with small interfering RNA (siRNA) pools specific for cyclophilin B (control RNA), Stat1, Stat3, or Stat5a and then treated for an additional 6 h with 20 ng/ml OSM. RNA was subsequently isolated and analyzed for Stat1, Stat3, and Stat5a as well as Rankl mRNA levels. As can be seen in supplemental Fig. 2, whereas the effect of cyclophilin B siRNA was similar to that of the mock transfected cells, siRNA to Stat1, Stat3, or Stat5a reduced levels of each of their respective STAT mRNA targets. Knockdowns of Stat1, Stat3, and Stat5a transcripts were confirmed at the protein level by Western blot analysis (supplemental Fig. 2D), although STAT1 protein levels were modestly reduced with Stat3 and Stat5a knockdown. The presence of OSM strongly induced the basal level of Stat3, an autoregulatory process that has been observed previously (20,37), and modestly up-regulated Stat5a but not Stat1. As seen in Fig. 7A, only the Stat3 knockdown influenced basal as well as OSM-inducible concentrations of Rankl expression. We conclude from this work that OSM functions to induce Rankl expression exclusively via STAT3.

Figure 7.

STAT3 knockdown eliminates OSM-induced Rankl expression via the mRLD6 enhancer region. A, Stat3 knockdown eliminates OSM-induced Rankl transcription. ST2 cells were transfected with 40 nm siRNA as indicated, incubated for 48 h, and then treated with 20 ng/ml OSM for an additional 6 h. RNA was isolated, reverse transcribed, and analyzed by quantitative PCR for Rankl expression. Levels were normalized to β-actin. B, Stat3 knockdown eliminates OSM-induced pTK-luc mRLD6 reporter activity. The pTK-luc mRLD6 distal reporter construct (250 ng) and pCH110-βgal (50 ng) were transfected into ST2 cells in the presence of cyclophilin B (CycloB), Stat1, Stat3, or Stat5a siRNA (40 nm). After a 48-h preincubation, cells were treated overnight with 20 ng/ml OSM or 50 ng/ml IL-6 and 100 ng/ml IL-6sR. Cells were harvested, and both luciferase activity and β-galactosidase activity were measured. Luciferase activity was normalized to β-galactosidase activity. Each value represents the average of three independent experiments ± sem compared with cyclophilin B control samples using a nonparametric t test. *, P < 0.05.

To determine whether the effects of STAT3 were mediated via the STAT elements located in mRLD6, we performed similar knockdown studies of Stat1, Stat3, and Stat5a and examined whether reductions in these transcription factors altered the ability of OSM to induce the pTK-mRLD6 distal reporter vector. ST2 cells were therefore cotransfected with the reporter vector and siRNA to cyclophilin B, Stat1, Stat3, or Stat5a. Cells were treated 48 h later with 20 ng/ml OSM or 50 ng/ml IL-6 and 100 ng/ml IL-6sR and then harvested 20 h after cytokine treatment for assessment of luciferase activity. As can be seen in Fig. 7B, although siRNA knockdown of Stat3 reduced both basal as well as OSM- and IL-6/IL-6sR-inducible activity from the mRLD6 distal reporter, siRNAs for cyclophilin B, Stat1, and Stat5a had no effect. Transfection and activation of mutant pTK-mRLD6 distal constructs produced similar results, although the levels were much reduced (data not shown). We conclude that the activity of OSM and IL-6/IL-6sR at the Rankl gene is mediated by STAT3 via the regulatory elements identified in mRLD6.

Forskolin induces CREB binding to the mRLD6 enhancer and acts synergistically with OSM to induce Rankl gene expression

Earlier studies revealed that many of the Rankl enhancers not only bind the VDR in response to 1,25-(OH)₂D₃ induction but also bind CREB in response to forskolin (25,33). In view of the modular nature of these enhancers, we examined whether mRLD6 might also provide a target for CREB action after forskolin treatment. ST2 cells were treated with either vehicle or forskolin (10⁻⁶ m) for 6 h and then subjected to ChIP-chip analysis using antibodies to either CREB or IgG. Figure 8A depicts data tracks representing 1) a vehicle-treated sample precipitated with antibodies to either CREB or IgG (CREB_veh vs. IgG) (basal CREB), 2) a forskolin-treated sample precipitated with antibodies to either CREB or IgG (CREB_Fsk vs. IgG) (total CREB), and 3) forskolin-treated and vehicle-treated samples precipitated with antibody to CREB (CREB_Fsk vs. CREB_veh) (net inducible CREB). Although CREB levels in the uninduced state were modest and primarily localized to mRLD2, as seen in Fig. 8A (track 1), forskolin strongly induced an increase in total CREB levels at the TSS, at mRLD2, mRLD4, mRLD5, and surprisingly, at mRLD6 as well. Inducible CREB was clearly observed in Fig. 8A (track 3). These data confirm previous observations that suggest the presence of CREB at mRLD2, mRLD4, and mRLD5a/b and highlight a potential role for CREB at mRLD6 (track 2) as well. The functionality of CREB binding at previously discovered enhancers was explored in past work (25). To determine whether the activity of CREB was interactive with STAT3 at mRLD6, we transfected ST2 cells with either the mRLD6 reporter constructs or mutants thereof, treated the cells with OSM, forskolin, or the combination, and evaluated luciferase activity 20 h later. As can be seen in Fig. 8B, forskolin treatment alone produced at best a modest and statistically insignificant increase over basal levels. OSM, on the other hand, strongly induced the mRLD6 reporter, as seen in Figs. 5 and 6. The combination, however, was 2- to 3-fold more effective than OSM alone in inducing the reporter. The activities of the mutant constructs were lower but similar in profile. The inability of forskolin alone to induce STAT3 binding to the Rankl enhancers, as assessed by ChIP-chip analysis (data not shown), suggests that the synergistic response to forskolin and OSM exhibited by the mRLD6 enhancer could be mediated via a direct interaction between CREB and STAT3.

Figure 8.

CREB is recruited to the mRLD6 enhancer region and potentiates cytokine-induced reporter activity. A, ChIP-chip analysis reveals the localization of CREB to the Rankl gene locus in response to forskolin. ST2 cells were treated with forskolin (Fsk) (10⁻⁶ m) for 6 h and then subject to ChIP analysis using antibodies to CREB or IgG. Samples were prepared as described in Materials and Methods. The upper panel depicts the Rankl gene locus as in Fig. 2. Data tracks represent the log₂ ratios of fluorescence obtained from 1) a vehicle-treated sample precipitated with antibodies to either CREB or IgG (CREB_veh vs. IgG) (basal CREB), 2) a forskolin-treated sample precipitated with antibodies to either CREB or IgG (CREB_OSM vs. IgG) (total CREB), and 3) forskolin-treated and vehicle-treated samples precipitated with antibody to CREB (CREB_OSM vs. CREB_veh) (net inducible CREB). All regions highlighted in red represent statistically significant peaks [false discovery rate (FDR), P < 0.05]. B, Forskolin and OSM cotreatment potentiates activity of the mRLD6 enhancer. The pTK-luc vector, pTK-luc mRLD6, pTK-luc mRLD6 distal, pTK-luc mRLD6 mut1, or pTK-luc mRLD6 mut2 reporter constructs (250 ng) and pCH110-βgal (50 ng) were transfected into ST2 cells and the cells treated with 20 ng/ml OSM, Fsk (10⁻⁶m), or both for 20 h. Cells were then harvested, and both luciferase activity and β-galactosidase activity were measured. Luciferase activity was normalized to β-galactosidase activity. Each value represents the average of three independent experiments ± SEM compared with control samples using a nonparametric T test. *, P < 0.05.

Discussion

The role of RANKL as an osteoclastogenic factor in bone metabolism is well established. RANKL is produced on the surface of stromal cells, osteoblasts, and T lymphocytes; interacts with RANK on both osteoclast precursors and mature osteoclasts; and promotes increased differentiation, activation, and survival of this primary bone-resorbing cell. Indeed, increased expression of RANKL is the major contributor to the osteoporotic bone loss associated with many diseases and is linked directly to the decreased BMD associated with most chronic inflammatory diseases. Although it is well known that inflammatory cytokines such as TNFα, IL-1β, and the IL-6-type cytokines induce RANKL expression from stromal support cells, the molecular mechanisms associated with this up-regulation are largely unknown. Using ChIP-chip analysis, we explored this mechanism by tracking the localization of OSM- and IL-6/IL-6sR-activated STAT3 at the Rankl locus, thus identifying regions to which this transcription factor becomes bound. Our studies identified several regions within the upstream distal portion of the Rankl gene that represent direct targets of STAT3 interaction, including previously described enhancers as well as a new region termed mRLD6.

The Rankl regulatory region is highly complex and comprised of multiple enhancers located significant distances upstream of the gene’s TSS (25,33). To fully explore this region after cytokine activation, we turned to ChIP-chip analysis and examined the ability of cytokines to promote the binding of pSTAT3 in response to the gp130-activating cytokines OSM and IL-6/IL-6sR. This approach not only provided an overall panoramic view of the binding of pSTAT3 across the Rankl gene locus but also allowed us to examine levels of histone acetylation and RNA pol II binding. Importantly, we assessed independently both basal and inducible levels of pSTAT3, RNA pol II, and H4ac. Moreover, by directly contrasting the presence of these factors in the absence and presence of inducer, we were also able to obtain a net inducible level of response. Thus, our experiments revealed striking changes in not only pSTAT3 accumulation at target sites but in the status of RNA pol II and H4ac across the gene locus as well. Although our previous studies explored the localization of CREB and pSTAT3 at enhancers identified originally through their interaction with the VDR (25), this new approach enabled a fully unbiased analysis of pSTAT3 and CREB targets. Importantly, we confirmed our previous results while simultaneously identifying an additional unique enhancer termed mRLD6.

As indicated above, we discovered that both OSM and IL-6/IL-6sR induced localization of STAT3 to several regulatory regions of the Rankl gene, three of which had been previously identified (mRLD2, mRLD4, and mRLD5) and a novel site termed mRLD6 as well. The binding of pSTAT3 to these regions prompted a general increase in RNA pol II recruitment over and above that which was already present at the enhancers in the absence of inducer, but particularly at regions including mRLD5 and mRLD6 to which the transcription factor bound. Interestingly, levels of H4ac were also relatively high across the upstream region of the Rankl gene in the absence of cytokine but particularly prevalent at each of the Rankl enhancers previously described (25,33). Moreover, some residual H4ac was also evident at the TSS. Cytokine treatment induced H4ac levels; this increase was focused almost exclusively at enhancers to which pSTAT3 bound, namely mRLD4, mRLD5a/b, and mRLD6. A modest increase was also observed at the TSS. Surprisingly, RNA pol II was never observed at the Rankl TSS regardless of whether cells were treated with the cytokine. The reason for this absence is unclear, although it is possible that the upstream enhancers serve as primary recruitment centers for the transfer of RNA pol II to the TSS via a looping mechanism and that the enzyme does not actually accumulate at the TSS. Regardless, the recruitment of both RNA pol II and changes in H4ac provide additional support for the functionality of these enhancers after STAT3 activation. Interestingly, significant H4ac appears to define the limits of both the Rankl coding and regulatory locus in stromal cells/preosteoblasts, because levels of H4ac decrease upstream of mRLD6 and immediately downstream of the final Rankl exon. Indeed, this decrease in H4ac activity appears to correlate directly with the presence of CCCTC-binding factor (CTCF)- occupied elements that are located at these boundaries as well (Martowicz, M.L., and J.W. Pike, unpublished).

Our studies revealed that activation of Rankl by OSM and IL-6/IL-6sR is mediated exclusively by STAT3. Thus, although a reduction in STAT3 suppressed Rankl expression in response to OSM, a reduction in STAT1 or STAT5a expression had no effect. The activity of STAT3 was mediated by two highly conserved, yet closely spaced response elements located in the distal portion of mRLD6. Both elements appear to contribute to cytokine activation, perhaps in a synergistic manner, because mutation of either of the elements strongly reduced but did not eliminate transcriptional activation. The central importance of this region to Rankl expression is likely highlighted by the observation that OSM induction of a Rankl reporter BAC clone is almost completely abrogated when a large region containing the mRLD6 enhancer is deleted (32). Interestingly, the presence of two adjacent regulatory elements mimics the structural organization of regulatory elements for both the VDR and CREB found at the mRLD5a/b enhancer, both of which are similarly comprised of two closely spaced response elements (25,33). The dramatic increase in activity of two or more elements is well known (38) and may account for the ability of the mRLD5 as well as the mRLD6 enhancer regions to mediate significant activity in response to cytokines, vitamin D, and protein kinase A activators when assayed independently of the Rankl gene. Indeed, all other regions we have examined thus far contain only single regulatory elements (25,33). Although we have not examined the functionality of STAT3-binding sites in the mRLD4 and mRLD5 regions, largely due to their inability to mediate sufficient inducible activity from independent constructs, we were able to identify putative single elements within via in silico analyses (data not shown). Finally, we show that a combination of forskolin and cytokine treatment leads to a synergistic induction at the mRLD6 region. Although it is likely that both STAT3 and CREB mediate this synergy, perhaps through a direct interaction between both factors, we cannot rule out the possibility that other transcription factors induced by forskolin may participate as well.

Although enhancer activities have been reported at the Rankl proximal promoter (39,40,41,42), our results suggest that Rankl regulation is largely mediated by a series of distal enhancers (25,32,33). Because the intergenic region that separates the Rankl gene locus from its upstream neighbor is approximately 180 kb, it is possible that additional enhancers may also exist. Although initially thought to be unique, the presence of enhancers located at distal intergenic sites recapitulates a theme that is emerging for many highly regulated genes (43). Genes that encode both transcription factors as well as key regulatory factors such as peptide regulators, growth factors, and cytokines are particularly good candidates for this highly complex regulation. Examples include such genes as Pax6 (44,45), BMP-2 (46), sclerostin (47), and SOX9 (48) as well as many others. The mechanistic roles of these enhancers to activate their associated genes will require additional investigation. Perhaps they participate in the recruitment of RNA pol II, delivering this key enzyme to the TSS where it functions to produce primary transcripts. It is possible that changes in H4ac that are generally associated with an open chromatin conformation could reflect this architectural transition (49).

Our current data together with previous studies reveal that the Rankl gene is regulated by multiple enhancers. These studies provide new insight into the underlying mechanisms whereby RANKL gene expression is modulated. Interestingly, the locations of these enhancer regions, highly conserved in the human RANKL gene (50), may also provide important clues as to how single- nucleotide polymorphisms (SNPs) can influence BMD in human populations. Recent genome-wide association studies have identified SNPs in five separate genomic regions in Icelandic and other populations that are associated with changes in BMD at the hip and spine and with osteoporotic fracture (51,52). One region is associated with the RANKL gene locus and located adjacent to the mRLD6 region. Several additional SNPs were located in the intergenic region between mRLD6 and RANKL’s upstream neighbor (AKAP11). These observations support regulatory roles for the known Rankl enhancers we have identified and suggest that additional regulatory regions may also exist as well. The biological impact of sequence variants in distal enhancers that influence the binding of specific transcription factors, and thus, the expression of associated genes is linked to a variety of complex diseases (43,53,54). In many cases, SNPs of this type have led to the identification of key enhancer elements and the transcription factors with which they interact (55).

The present studies extend our initial observations that the RANKL gene in both mouse and human cells is largely regulated by a series of novel enhancer elements located upstream of the Rankl TSS from −16 to −88 kb. The studies herein describe the discovery of the most distal of these enhancers, termed mRLD6, and show that this region mediates the up-regulation of the Rankl gene by cytokines such as OSM and IL-6. This up-regulation is facilitated through activation of the transcription factor STAT3, which acts via two STAT regulatory elements, and perhaps in synergy with a third that binds CREB. These studies contribute significantly to our understanding of RANKL gene expression and may lead to new approaches for controlling RANKL gene transcription, thereby providing new therapeutic opportunities for treating inflammation-induced bone loss and/or osteoporosis.

Materials and Methods

Reagents

General biochemicals were purchased from ThermoFisher Scientific (Waltham, MA) and Sigma-Aldrich (St. Louis, MO). Recombinant mouse OSM (495-MO) and recombinant human IL-6 (206-IL) and IL-6sR (227-sR/CF) were purchased from R&D Systems (Minneapolis, MN). Anti-STAT3 (sc-482) antibody and STAT1 E-23 antibody (sc-346) were obtained from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). Anti-pSTAT3-Tyr705 (9145S) and STAT3 antibody (9132S) were purchased from Cell Signaling Technologies (Danvers, MA), anti-H4ac antibody (06-866) and Immobilon-Psq (ISEQ00010) were obtained from Millipore (Billerica, MA), and anti-RNA pol II 8WG16 (MMS-126R) was purchased from Covance (Princeton, NJ). Stat1 (M-058881-01), Stat3 (M-040794-00), Stat5a (M-063202-00), and cyclophilin B siRNA pools were purchased from Dharmacon RNA Technologies (Lafayette, CO). Lipofectamine (18324-012), Plus Reagent (10964-012), MEMα (12000-022), GoTaq Flexi DNA polymerase (M8296), deoxyribonuclease I, and 4–20% Tris-glycine gels (EC60252Box) were purchased from Invitrogen Corp. (Carlsbad, CA). Goat antirabbit IgG horseradish peroxidase (401315) was purchased from Calbiochem (San Diego, CA). High-capacity cDNA reverse transcription kit (4368813) and Power SYBR Green PCR Master Mix (4347660) were purchased from Applied BioSystems (Foster City, CA). Cy3 (N46-1019-CJ1A) and Cy5 (N46-0110-BS2A) 9-mer wobble primers were purchased from TriLink (San Diego, CA). Fetal bovine serum (FBS) (SH30088.03) and penicillin/streptomycin (SV30010) were purchased from Hyclone (Logan, UT). The QuikChange mutagenesis kit (200516-5) was purchased from Stratagene (San Diego, CA). QIAquick PCR purification kits were purchased from QIAGEN (Valencia, CA). NimbleGen hybridization kits (05223474001) and NimbleGen wash buffer kits (05223504001) were purchased from Roche NimbleGen, Inc. (Madison, WI).

Cell culture

Mouse ST2 osteoblastic cells were cultured in MEMα supplemented with 10% FBS, 100 U/ml penicillin, and 100 μg/ml streptomycin.

RNA isolation and analysis

ST2 cells were seeded in six-well plates and were treated upon confluence with OSM or IL-6 and IL-6sR for 6 h. Total RNA was isolated from cells using Tri-Reagent. The isolated RNA was treated with deoxyribonuclease I and reverse transcribed using the high-capacity cDNA reverse transcription kit. The resulting cDNA was then subjected to quantitative PCR analysis.

Quantitative PCR analysis

Real-time PCR was performed on an Eppendorf Realplex using Power SYBR Green PCR Master Mix with standard cycling conditions. The Mastercycler ep realplex software was used for analysis. Standard curves were made by serial dilutions of PCR-amplified cDNA. Primers used in the RNA analyses include Rankl mRNA, 5′-TGTACTTTCGAGCGCAGATG-3′ (forward) and 5′-AGGCTTGTTTCATCCTCCTG-3′ (reverse); β-actin mRNA, 5′-TGTTTGAGACCTTCAACACCC-3′ (forward) and 5′-CGTTGCCAATAGTGATGACCT-3′ (reverse); Stat1 mRNA, 5′-ATTGACCTGGAGACCACCTCT-3′ (forward) and 5′-TGACTGATGAAAACTGCCAACT-3′ (reverse); Stat3 mRNA, 5′-AAAAGGACATCAGTGGCAAGAC-3′ (forward) and 5′-GGGTAGAGGTAGACAAGTGGA-3′ (reverse); and Stat5a mRNA, 5′-ACTCTTCGGGATGGGGACTAT- 3′ (forward) and 5′-CGGTGGAGGCTGTTACTTCTAA-3′ (reverse).

ChIP analysis

ChIP analysis was preformed as previously described (25,33,56,57). Immunoprecipitated DNA was blunt-ended by T4 DNA polymerase, ligated to linkers with the sequence 5′-GAATTCAGATC-3′ and 5′-GCGGTGACCCGGGAGATCTGAATTC-3′ using T4 DNA ligase, and amplified by ligation-mediated PCR with Go Taq Flexi and linear PCR amplifications. PCR purification was performed with the QIAquick PCR purification kit. DNA samples were labeled with Cy3 or Cy5 9-mer wobble primers using Klenow fragment, the reaction then stopped with the addition of EDTA, precipitated with NaCl and isopropanol, washed with 80% ethanol, and then resuspended in water. Cy3- and Cy5-labeled DNA samples as indicated were cohybridized to a custom oligonucleotide microarray using a NimbleGen hybridization kit and a MAUI hybridization system. Microarrays were washed using NimbleGen wash buffer and scanned using an Axon 4000B scanner with GenepixPro version 4.1 software at the appropriate wavelengths. Custom oligonucleotide microarrays were synthesized by Roche-NimbleGen Systems (Madison, WI). The microarray oligonucleotide probes were 50- to 70-mer in length with 65-bp resolution and synthesized using a mask-less array system tiled from 200 kb upstream of the Rankl TSS to 200 kb downstream of the final 3′-coding exon. This custom array contained additional tiled genes not considered in this study. DNA samples were obtained by immunoprecipitation with antibodies specific for STAT3, pSTAT3 (Tyr705), H4ac, RNA pol II, CREB, and a control rabbit IgG. A series of three comparisons were made for each analysis: 1) untreated and IgG control; 2) OSM-, IL-6/IL-6sR-, or forskolin-treated and IgG control; and 3) OSM-, IL-6/IL-6sR-, or forskolin-treated vs. vehicle-treated samples. The aforementioned cohybridizations were analyzed and presented as previously described (33). A sliding window of 700 bp containing at least four of eight probes statistically above background was deemed significant.

Plasmids

Construction of pTK-mRLD1, -D2, -D3, -D4, -D5a, and -D5b reporter constructs prepared in the thymidine kinase luciferase vectors (pTK-luc) was previously described (33). Similar methods were used to prepared the pTK-mRLD6 full (−87842 to −88614), -D6 proximal (−87842 to −88067), and -D6 distal (−88398 to −88614) reporter constructs using primers containing HindIII/BamHI restriction sites. Digested fragments were cloned into the corresponding sites within the pTK-luc vector. The QuikChange mutagenesis kit was used to introduce mutations into the pTK-mRLD6 distal reporter construct.

Reporter transfections

The transient transfection of ST2 cells with reporter constructs was performed as previously described (33). Briefly, ST2 cells were seeded in a 24-well plate. Individual wells were transfected with 250 ng reporter construct and 50 ng pCH110-βgal using Lipofectamine and PLUS reagent in serum-free media. After a 3-h incubation, wells were supplemented with MEMα containing 20% FBS to a final concentration of 10% FBS and varying concentration of OSM, IL-6, IL-6sR, IL-6, and IL-6sR or forskolin. Cells were harvested 20 h later for reporter assays using lysis buffer and both luciferase and β-galactosidase activities were determined as previously described (25,33).

siRNA transfection

The transient transfection of ST2 cells with siRNA targeting cylophilin B, Stat1, Stat3, or Stat5a was performed as previously described (33). Briefly, ST2 cells were seeded in a 6-well plate. Individual wells were transfected using Lipofectamine and PLUS reagent with 40 nm siRNA in serum-free media. After a 3-h incubation, wells were supplemented with MEMα containing 20% FBS to a final concentration of 10% FBS for 48 h. Cells were then treated with 20 ng/ml OSM or 50 ng IL-6 and 100 ng IL-6sR for 6 h. RNA was isolated and analyzed according to previously described protocols (33). Cotransfection of siRNA and reporter constructs was performed according to the siRNA transfection protocol. Briefly, ST2 cells were seeded in a 24-well plate and transfected using Lipofectamine and PLUS reagent with 40 nm siRNA, 250 ng reporter construct, 50 ng pCH110-βgal in serum-free media. After a 3-h incubation, wells were supplemented with MEMα containing 20% FBS to a final concentration of 10% FBS for 48 h. Cells were then treated with 20 ng/ml OSM or 50 ng IL-6 and 100 ng IL-6sR. Cells were harvested 20 h later and luciferase activity assessed using Bright Glo luciferase substrate and normalized to β-galactosidase activity.

Western blot analysis

Intact cells, transfected 48 h earlier with Stat1, Stat3, Stat5a, and cyclophilin B siRNAs (40 nm), were trypsinized, counted, and boiled in loading dye. Equivalent concentrations of cells were electrophoresed on 4–20% Tris-glycine gels and transferred to an Immobilin-Psq membrane for analysis. Blots were probed with antibodies to STAT1 (1:500), STAT3 (1:1000, Cell Signaling), STAT5a (1:500), β-tubulin (1:1000), and goat antirabbit IgG horseradish peroxidase (1:2000). Blots were stripped between analyses using 100 mm BME, 100 mm Tris-HCl (pH 6.8), and 2% SDS.

Statistical analyses

All values are expressed as the mean ± sem. All statistical calculations were performed using GraphPad PRISM version 4 statistical software package (GraphPad Software Inc., San Diegoç CA) using either a one-way nonparametric ANOVA analysis with single variance followed by a Tukey multiple-comparison post-test or a nonparametric t test. OSM-treated samples were contrasted with vehicle-treated controls. IL-6/IL-6sR-treated samples were contrasted with vehicle-treated, IL-6-treated, or IL-6sR-treated controls. Selective siRNA-mediated suppression of Stat1, Stat3, and Stat5a RNAs were contrasted with that obtained using a cyclophilin B siRNA control.