MKP-1 Is Essential for Canonical Vitamin D-Induced Signaling through Nuclear Import and Regulates RANKL Expression and Function

Abstract

Vitamin D_3, and its most active form, 1,25(OH)₂D₃, are well known to stimulate osteoclastogenesis through stromal cell induction of the receptor activator of nuclear factor-κB ligand (RANKL). MAPK phosphatase-1 (MKP-1) is a phosphatase classically known to negatively regulate the innate immune response through dephosphorylation of p38, ERK, and c-Jun N-terminal kinase activity. This paper describes a new function of MKP-1 in permitting genomic 1,25(OH)₂D₃ signaling and downstream osteoclastogenesis through RANKL. Initially, quantitative RT-PCR (qRT-PCR) and immunoblot analysis comparing bone marrow stromal cells (BMSC) revealed that 1,25(OH)₂D₃-induced vitamin D receptor (VDR), cytochrome P 45024a1, and RANKL mRNA expression and protein were significantly attenuated or absent in MKP-1^−/− BMSC. Immunoblot analysis from cellular fractions of wild type and MKP-1^−/− BMSC stimulated with 10⁻⁷ m 1,25(OH)₂D₃ revealed retinoid X receptor (RXR)α nuclear import was impaired in MKP-1^−/− BMSC, whereas VDR import was not. Proximity ligation assays revealed that baseline VDR-RXRα heterodimer translocation was unchanged, yet 1,25(OH)₂D₃-induced nuclear translocation of VDR-RXRα heterodimers was reduced in MKP-1^−/− BMSC. A functional consequence was observed as BMSC from MKP-1^−/− mice treated with 1,25(OH)₂D₃ and cocultured with RAW 264.7 cells had a 91% decrease in osteoclastogenesis and a 94.5% decrease in mineralized matrix resorption compared with wild-type cocultures (P < 0.01). These results reveal an unexpected, permissive role for MKP-1 in canonical 1,25(OH)₂D₃ signaling via VDR-RXRα heterodimer nuclear import and downstream osteoclastogenesis through stromal cell RANKL expression.

Since its discovery as an antirachitic factor, vitamin D has been widely studied for its role in skeletal homeostasis (1). The active form of this steroid hormone, 1,25-(OH)₂D₃, and the vitamin D receptor (VDR) have been well studied for their roles in bone mineralization and turnover. It is widely accepted that the downstream genomic events of 1,25-(OH)₂D₃-bound VDR are cell specific and can significantly alter the formation and activity of osteoblasts and osteoclasts. A human deficiency in this hormone can result in rickets in children and osteomalacia in adults (2, 3). The importance of the VDR as a 1,25-(OH)₂D₃ receptor is highlighted by type II vitamin D-dependent rickets (OMIM 277440) whereby mutations in the DNA-binding domain of the VDR result in a rickettsial phenotype that is clinically similar to vitamin D deficiency (4).

The classic genomic mechanism by which 1,25-(OH)₂D₃-ligated VDR regulates gene expression in osteoblasts is through ligand-initiated heterodimerization with retinoid X receptor (RXR)-α, translocation to the nucleus, and binding of specific DNA regions, namely vitamin D-responsive elements (VDRE) and coactivator recruitment (5, 6). RXRα is a nuclear receptor family member that regulates the action of many endocrine processes through heterodimerization with a variety of hormone receptors, including thyroid receptor, peroxisome proliferator-activated receptor, VDR, and other retinoic acid receptors (7). RXRα has been shown to have a profound role in VDR genomic signaling, in part by regulating subcellular localization of the RXRα-VDR heterodimer (8). The mouse VDR gene has two primary intronic VDR-RXR binding VDRE at +20 and +29 kb and therefore induces its own expression in the presence of 1,25-(OH)₂D₃ (9, 10). VDRE have been found on the promoter regions of many genes involved in calcium homeostasis including receptor activator of nuclear factor-κB ligand (RANKL), an essential cytokine in the bone-resorbing process of osteoclastogenesis (11–13). Five mouse and human VDRE have been identified upstream of the transcriptional start site of the RANKL gene (TNFS11), with the most distal and potent murine VDRE (mRLD5) occurring at −76 kb (11, 14). Along with PTH as one of the primary endocrine inducers of RANKL in active osteoblasts, vitamin D and its antagonists have been widely studied for their therapeutic potential in bone pathologies (15–17). Recently, both 1,25-(OH)₂D₃ and exogenous RANKL stimulation of primary bone marrow-derived macrophage cells have been shown to induce the immune regulatory protein MAPK phosphatase-1 (18, 19).

The MAPK phosphatase (MKP) family of proteins has the ability to negatively regulate MAPK activity by dephosphorylating MAPK proteins. MAPK phosphatase-1 (MKP-1) is a phosphatase that negatively regulates the innate immune response through dephosphorylation of p38, ERK, and c-Jun N-terminal kinase (JNK) activity in response to various stimuli (20–23). This effect has been demonstrated by our laboratory where lipopolysaccharide-challenged MKP-1^−/− mice exhibit a significant increase in osteoclastogenesis and subsequent alveolar bone loss compared with wild-type controls (24). Recently, MKP-1 has gained acceptance as a potential pharmacological target for the treatment of obesity and metabolic disorders (25).

Although MAPK activity has also been shown to inhibit genomic vitamin D activity through RXRα modification (26, 27), the role of MKP in vitamin D signaling and steroid receptor translocation is currently unknown. Limited studies on MKP-1's effect on skeletal homeostasis reveal that female, but not male, MKP-1^−/− mouse femurs have decreased trabecular density at 8 wk, with a decrease in mineral apposition rate and bone formation rate (18). Although these studies strongly suggest a role for MKP-1 in skeletal homeostasis and maintenance, there has yet to be a clearly defined mechanism for its involvement in these processes. This paper presents a new, highly novel role for MKP-1 whereby it is required in both the genomic 1,25-(OH)₂D₃-induced VDRE binding of the RXRα-VDR heterodimer and downstream target gene expression.

Materials and Methods

Reagents

General biochemicals were obtained from Sigma Chemical Co. (St. Louis, MO). 1,25-(OH)₂D₃ was obtained from Enzo Life Sciences (Farmingdale, NY). Life Technologies MEM, α-MEM, and fetal bovine serum were purchased from Invitrogen (Carlsbad, CA), as were oligonucleotides used in EMSA and chromatin immunoprecipitation (ChIP) analysis. ON-TARGETplus pooled MKP-1 small interfering RNA (siRNA) was obtained from Dharmacon (Lafayette, CO). Anti-RXRα [sc-553(x)] and anti-VDR [sc13133(x)] were obtained from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). [α-³²P]dCTP (BLU513Z250UC) (6000 Ci/mmol) was obtained from PerkinElmer Corp. (Wellesley, MA).

Animals

Wild-type (WT) and MKP-1-knockout mice were purchased from Bristol-Myers Squibb Pharmaceutical Research Institute and bred at the Medical University of South Carolina and were treated in accordance with the standards of humane animal care. These mice are maintained on a mixed C57/129 background and housed under specific pathogen-free conditions with food and tap water ad libitum.

Cell culture

Primary bone marrow cells were harvested by flushing the femurs and tibia of 5- to 10-wk-old mice with α-MEM supplemented with 100 U/ml penicillin and 100 μg/ml streptomycin and 10% heat-inactivated fetal bovine serum. Cells were passed through a 40-μm filter into a 100-mm cell culture plate. After 24 h of incubation, nonadherent cells were removed, and the plate was washed vigorously with PBS. Cells were washed again at 6 d and at 10 d the cells were scraped and plated at a density of 5 × 10⁵ cells per well in a 12-well plate. Before 1,25-(OH)₂D₃ or vehicle (EtOH) treatment, bone marrow stromal cells (BMSC) and UAMS-32 cells were deinduced for 6 h with 0.3% fetal bovine serum.

RT-quantitative PCR (qPCR)

Whole-cell BMSC were harvested in TRIzol, and RNA was harvested according to protocol. Total RNA was quantified and 500 ng were reverse-transcribed into cDNA, which was used in qPCR performed in a TaqMan system using an Applied Biosystems StepOnePlus RT-PCR Real-Time Thermal Cycler, optimized Assays-On-Demand primers for the target genes MKP-1, RANKL, VDR, cytochrome P450 (CYP)24a1, and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) for normalization.

siRNA studies

All siRNA duplexes from purchased from Dharmacon RNA Technologies (Lafayette, CO) and electroporation-based transfection was performed using the Neon Transfection System obtained from Invitrogen.

Western blot analyses

Whole-cell samples were harvested in RIPA buffer, and 20 μg total protein were loaded onto 10% SDS-PAGE gels and transferred to nitrocellulose for immunoblotting to detect RXRα and VDR protein expression. Nuclear and cytoplasmic fractions were isolated using the NE-PER kit from Thermo Scientific (Rockford, IL) according to manufacturer's instructions. Digital images were acquired and quantified using Quantity One software from Bio-Rad Laboratories, Inc. (Hercules, CA).

EMSA

WT and MKP-1^−/− nuclear extracts (6 μg) were isolated (as described above) and were pretreated on ice with 10⁻⁷ m 1,25-(OH)₂D₃ and 1 μg polydeoxyinosinic deoxycytidylic acid for 20 min in a binding buffer [25 mm Tris-HCl (pH 8.0), 5% glycerol, 0.5 mm dithiothreitol, 100 mm KCl]. α-³²P-labeled mouse osteopontin (mOP) VDRE oligonucleotide (mOP VDRE; 100 fmol) was added and incubated for 20 min at room temperature (forward, 5′–3′ GGGTACAAGGTTCACGAGGTTCACGTCTTA). Indicated lanes were also incubated with anti-RXRα antibody (sc-553x), recognizing the NH₂-terminal domain of RXRα, anti-VDR antibody (sc13133x), specific to VDR amino acids 344–424, or control rabbit and mouse IgG.

ChIP assays

Primary BMSC were isolated and cultured as described above and plated on 150-mm cell culture plates and treated with or without 10⁻⁷ m 1,25-(OH)₂D₃ for 6 h before cells were cross-linked with 1% formaldehyde α-MEM. ChIP experiments were performed using reagents supplied following manufacturer's instructions supplied with the ChIP-IT Express kit from Active Motif (Carlsbad, CA). Chromatin pellets were sonicated to an average DNA size of 250-1000 bp (determined by gel electrophoresis) using a Fisher Model 100 Sonic Dismembranator on ice (Fisher Scientific, Pittsburgh, PA). Incubation with the indicated antibodies was performed overnight at 4 C. DNA fragments were subjected to PCR using primers designed to amplify the murine RANKL D5 region located −76 kb from the RANKL transcription start site (TSS) (forward, 5′-AAGGGTCAAGAGGGGCCTGACT; reverse, 5′-TGTCCCGGTCTCTGTCCCTGG) and a negative control region at −50 kb from the TSS designated IS4 (forward, 5′-CCTCCTATCTGTTTTACTGACGTT; reverse, 5′-CACATAGGCAGAAAGTTGAAAAGC) (11). Input DNA was acquired before immunoprecipitation and was loaded at 10% to determine the presence of gene fragments for PCR evaluation. PCR for both primer sets were run for 35 cycles and resolved on 2% agarose gels.

Immunofluorescence and in situ proximity ligation assay (PLA)

WT and MKP-1^−/− BMSC were seeded at 3 × 10⁵ cells per well in eight-well chamber slides (Labtek, Scotts Valley, CA). After a 24-h attachment period and 6 h in 0.5% α-MEM, cells were treated with vehicle control (ethanol) or 10⁻⁷ m 1,25-(OH)₂D₃ for 1 h. Medium was removed, and intact cells were fixed and permeabilized with −20 C MeOH for 8 min. Anti-RXRα (sc-553) and anti-VDR (sc13133) antibodies were used in immunofluorescence detection of nuclear/extranuclear RXRα and VDR native proteins, and in the PLA assay at a 1:500 dilution to detect the nuclear/extranuclear localization of the native VDR-RXRα heterodimer. Incubation with appropriate PLA probe (antimouse or antirabbit antibody), antibody blocking, ligation of probes via DNA ligase, DNA polymerase amplification, and slide mounting were performed using the proprietary solutions supplied in the Duolink kit from Olink Biosciences (Uppsala, Sweden) following manufacturer's instructions and supplied reagents (28). Negative control experiments (either anti-RXRα or anti-VDR were replaced with same species control IgG) were performed in parallel. Both immunofluorescence and PLA slides were analyzed using a Leica SP5 inverted confocal laser microscope (Leica Microsystems SA, Deerfield, IL) with a ×40 oil immersion objective. Z-stacks were taken at a density of one photo/0.5 μm, and the z-range was limited to the dimension of nuclear 4′,6-diamidino-2-phenylindole staining. Z-stacks were overlaid, and image processing and analysis were done using the Leica confocal software. Punctate PLA signals were quantified by a blinded observer.

Osteoclast coculture analysis

WT and MKP-1^−/− BMSC were depleted of CD11b⁺ cells using a Miltenyi autoMACS Pro separator magnetic cell sorter (Miltenyi Biotec, Bergisch Gladbach, Germany) to reduce the RANKL-induced osteoclastogenesis in uncommitted primary macrophage populations (29), and plated (5 × 10⁴ cells per well) in four-well chamber slides or Osteologic discs (Becton Dickinson, Franklin Lakes, NJ). CD11b⁻ BMSC were pretreated with 10⁻⁸ m 1,25-(OH)₂D₃ for 2 d before 1 × 10⁴ RAW 264.7 cells were added to each well. (10⁻⁸ m) 1,25-(OH)₂D₃ and half medium was changed every other day for 7 d [tartrate-resistant acid phosphatase (TRAP) staining] or 10 d (pit assay). Images of cultured cells were digitally captured using a Nikon TS100 inverted microscope and a Nikon 5.1 megapixel camera (Nikon, Melville, NY). Osteoclast or pit measurements were enumerated by blinded technicians. Pit areas were quantified with the assistance of NIH ImageJ version 1.44.

TRAP staining

Osteoclast cocultures were fixed with 10% glutaraldehyde and stained with TRAP as described using an Acid Phosphatase, Leukocyte TRAP Kit (Sigma).

Statistical analysis

All values presented in the figures are expressed as the mean ± sem unless otherwise stated. All statistical analyses were performed with GraphPad PRISM version 4 (GraphPad Software Inc., San Diego, CA). Differences between groups were evaluated by two-way ANOVA or Student's two-tailed t test. Significance was determined at P < 0.05.

Results

MKP-1 is required for RANKL expression in BMSC

Previously published reports from our laboratory indicate that MKP-1^−/− mice have increased inflammatory bone loss in response to subgingival lipopolysaccharide (LPS) injections due to an elevation in cytokine-induced osteoclastogenesis (24). We hypothesized that this was due to an increase in RANKL expression and accordingly tested this hypothesis by treating primary WT and MKP-1^−/− BMSC with the inflammatory cytokines IL-1β and TNF-α, as well as LPS from oral microbes Aggregatibacter actinomycetemcomitans and Porphyromonas gingivalis, and 1,25-(OH)₂D₃. Although we observed a 2-fold elevation in baseline RANKL mRNA expression, the only treatment in MKP-1^−/− BMSC that was significantly different from WT was the 1,25-(OH)₂D₃ signaling. Surprisingly, we observed a 12-fold decrease in 1,25-(OH)₂D₃-stimulated RANKL mRNA expression in MKP-1^−/− BMSC (Fig. 1A). To test whether this finding correlated with soluble RANKL protein expression, primary WT and MKP-1^−/− BMSC were incubated with 10⁻⁸ m 1,25-(OH)₂D₃ for 48 h, and an ELISA was performed for secreted/membrane cleaved RANKL in the culture medium (Fig. 1B). Consistent with the RT-qPCR findings, soluble RANKL protein expression was also decreased in MKP-1^−/− BMSC in response to 1,25-(OH)₂D₃. Subsequent experiments sought to determine the mechanism underlying these observation and determine the role of MKP-1 signaling in 1,25-(OH)₂D₃-mediated stromal/osteoblast gene expression.

Fig. 1.

MKP-1 is required in BMSC for 1,25-(OH)₂D₃ (D3) induction of RANKL. A, RT-qPCR analysis of RANKL mRNA expression in response to 18 h treatment with inflammatory stimuli and 10⁻⁸ m 1,25-(OH)₂D₃. B, ELISA of soluble RANKL expression from cultured media containing indicated cytokines, LPS, or 10⁻⁸ m 1,25-(OH)₂D₃ for 48 h (n = 4; ***, P < 0.001). Aa, Aggregatibacter actinomycetemcomitans; Pg, Porphyromonas gingivalis.

RT-qPCR analysis of 1,25-(OH)₂D₃ target gene expression in MKP-1^−/− BMSC

Although data from RT-qPCR and ELISA analysis showed that MKP-1 is necessary for 1,25-(OH)₂D₃-stimulated RANKL expression in BMSCs, we sought to determine whether the requirement of MKP-1 in 1,25-(OH)₂D₃ signaling was limited to RANKL induction. In addition, we wanted to understand if MKP-1-mediated RANKL inhibition was immediate or occurred over several hours (30). To address these questions, we treated WT and MKP-1^−/− BMSC with 10⁻⁸ m 1,25-(OH)₂D₃ for 0, 2, 4, 8, 12, 18, and 24 h and then performed RT-qPCR for three well-established 1,25-(OH)₂D₃ target genes: RANKL, VDR, and CYP24a1 (Fig. 2A). As expected, this kinetic assay showed that all three 1,25-(OH)₂D₃ target genes in WT BMSC had a peak expression between 12 and 18 h. Conversely, although MKP-1^−/− BMSC exhibited similar baseline expression up to 4 h, peak mRNA expression in the 12- to 18-h range did not occur, indicating a defect in genomic 1,25-(OH)₂D₃ signaling. Western blotting for VDR expression after 24-h 1,25-(OH)₂D₃ treatment indicated that MKP-1^−/− BMSC had a decrease in VDR protein expression as well (Fig. 2, B and C). These data are consistent with the notion that MKP-1 was required for 1,25-(OH)₂D₃ genomic signaling.

Fig. 2.

MKP-1 is necessary for maximal genomic expression of 1,25-(OH)₂D₃ target genes RANKL, VDR, and CYP24a1. A, WT and MKP-1^−/− BMSC were stimulated with 10⁻⁸ m 1,25-(OH)₂D₃ for 0, 2, 4, 8, 12, 18, and 24 h. RT-qPCR results above were normalized to GAPDH expression and shown as fold change relative to WT control (n = 3). B,) Western blot analysis for VDR after 24 h 10⁻⁸ m 1,25-(OH)₂D₃ stimulation indicates that MKP-1 is required for 1,25-(OH)₂D₃ induction of VDR protein expression. C, Densitometric analysis of VDR expression normalized to GAPDH (n = 3; *, P < 0.05, **, P < 0.01).

MKP-1 is required for the maximal 1,25-(OH)₂D₃ target gene expression and RANKL promoter activity in UAMS-32 stromal/osteoblast cells

Because primary BMSC are not a homogenous population of cells, and 1,25-(OH)₂D₃ target gene expression in osteoblasts has been well described (6, 31), we wanted to confirm our hypothesis that MKP-1 plays a role in stromal osteoblast development. We transfected the UAMS-32 mouse stromal/osteoblast cell line with 100 nm pooled MKP-1 siRNA using an optimized electroporation protocol and were able to knock down MKP-1 mRNA expression by 70% after an 18-h 1,25-(OH)₂D₃ treatment (Fig. 3A). MKP-1 knockdown had no effect on unstimulated RANKL, VDR, and CYP24a1 (data not shown), yet RT-qPCR revealed that 1,25-(OH)₂D₃-stimulated CYP24a1 and RANKL mRNA expression was significantly decreased, and VDR mRNA expression was actually found to be below baseline transcript levels (Fig. 3A). These findings demonstrate that MKP-1 is necessary for 1,25-(OH)₂D₃ target gene expression in stromal/osteoblast cells and that the decrease in their expression, as shown in Fig. 2, is due, at least in part, to the permissive role of MKP-1 in stromal/osteoblast cells.

Fig. 3.

MKP-1 is needed for maximal 1,25-(OH)₂D₃ target mRNA expression and RANKL promoter activity in a UAMS32 Luc-reporter stromal cell line. A, UAMS-32 stromal cells were treated with 10⁻⁸ m 1,25-(OH)₂D₃ for 18 h followed by qPCR analysis. B, MKP-1 siRNA-transfected UAMS-32 stromal cells were stimulated with 10⁻⁸ m 1,25-(OH)₂D₃ for 18 h, followed by qPCR analysis. (n = 4; *, P < 0.05; **, P < 0.01). SCR, Scrambled.

Understanding that MKP-1 is required for 1,25-(OH)₂D₃ target mRNA expression, we examined whether MKP-1 signaling had an effect at the transcription level. We performed a similar MKP-1 siRNA experiment with UAMS-32 cells that contained a bacterial artificial chromosome-based transcriptional reporter containing the entire mouse RANKL gene with 120 kb of RANKL 5′-flanking region and 14-kb 3′-flanking region, as described previously (32). As positive controls, reporter activity of these cells was significantly increased by PTH (data not shown) and 1,25-(OH)₂D₃ treatment, with a 50% decreased response upon VDR siRNA treatment (data not shown). Time course experiments revealed maximal 1,25-(OH)₂D₃ induction of RANKL gene promoter activity at 18 h (our unpublished data). Using the same conditions as shown in Fig. 3A, we demonstrated that the MKP-1 knockdown in osteoblasts significantly decreased 1,25-(OH)₂D₃-induced RANKL promoter activity by 57% (P = 0.05; Fig. 3B).

Gel mobility shift and ChIP analysis reveals that VDR-RXRα heterodimer VDRE binding is attenuated in MKP-1^−/− BMSC

Having determined that MKP-1 is necessary for the genomic expression of several 1,25-(OH)₂D₃ target genes and for RANKL gene promoter activity, we explored the possibility that MKP-1 permits the VDR-RXRα heterodimer's ability to access/bind VDRE in BMSC nuclear extracts. We treated primary WT and MKP-1^−/− BMSC with 10⁻⁷ m 1,25-(OH)₂D₃ for 1 h and performed an EMSA with the harvested nuclear extracts (containing phosphatase inhibitors) and the mouse osteopontin (mOP) oligonucleotide containing the VDRE sequence GGTTCAxxxGGTTCA, which strongly interacts with the VDR-RXRα heterodimer (9, 33). Incubation with the α-³²P-labeled mOP oligonucleotide yielded a relatively equal shift with both WT and MKP-1^−/− extract that was abolished using nonradiolabeled mOP competition. Incubation with antibodies against VDR and RXRα consistently produced a strong supershift in WT extracts that diminished the original shift (Fig. 4). Using MKP-1^−/− nuclear extracts, these RXRα and VDR antibodies consistently produced a visibly fainter supershift, whereas incubation with control rabbit IgG had no effect. In EMSA assays in which BMSC were only treated with vehicle control, a VDR and an RXRα supershift was observed in WT cells only (our unpublished data). ChIP analysis was used to test VDR and RXRα's ability to bind the distal (−76 kb) RANKL VDRE mRLD5 in WT and MKP^−/− BMSC containing the binding sequences GAGTCAccgAGTTGA and GGTTGCctgAGTTCA (9). Although immunoprecipitates show that VDR is strongly induced to bind to the mRLD5 region by 1,25-(OH)₂D₃ in both genotypes, we only observe RXRα DNA binding in WT BMSC. These data provide evidence that MKP-1 is required for both maximal 1,25-(OH)₂D₃-induced nuclear binding of VDR and RXRα to a consensus VDRE and 1,25-(OH)₂D₃-induced binding of endogenous RXRα to the RANKL VDRE mRLD5.

Fig. 4.

EMSA analysis shows 1,25-(OH)₂D₃-stimulated MKP-1^−/− BMSC have a decreased VDRE binding capacity. A, The α-³²P-labeled mOP VDRE was incubated with 1,25-(OH)₂D₃-treated nuclear extracts from WT and MKP-1^−/− BMSC and electrophoresed through a nondenaturing gel. R, RXRα supershifts; V, VDR antibody supershifts; *, the putative mOP-RXRα/VDR heterodimer complex. Results are representative of three independent experiments. B, ChIP analysis of the RANKL VDRE enhancer region located −76 kb from the RANKL TSS (mRLD5) and control non-VDRE region located −50 kb from the RANKL TSS (IS4). Results are representative of three independent experiments. Ctrl, Control.

1,25-(OH)₂D₃-induced nuclear import of the VDR-RXRα heterodimer is attenuated in MKP-1^−/− BMSC

After the observation that there is a decreased capacity for VDRE binding in MKP-1^−/− BMSC nuclear extracts, we sought to determine whether MKP-1^−/− BMSC have an altered subcellular localization of RXRα and VDR. We treated WT and MKP-1^−/− BMSC with 1,25-(OH)₂D₃ for 1 h, isolated the nuclear and cytoplasmic fractions, and subjected the protein extracts to Western blotting for RXRα and VDR. The results indicate that 1,25-(OH)₂D₃ induces a VDR translocation from the cytoplasmic to the nuclear fraction in both cell genotypes, although this VDR translocation may be delayed in MKP-1^−/− BMSC, as evidenced by the relatively small reduction in cytoplasmic VDR upon 1,25-(OH)₂D₃ stimulation (Fig. 5A). The results show that upon 1,25-(OH)₂D₃ treatment, RXRα translocates freely into the nucleus from the cytosol in WT BMSC, presumably as a VDR-RXRα heterodimer. In MKP-1^−/− BMSC, however, there is no detectable translocation, as evidenced by the lack of an increase in RXRα upon 1,25-(OH)₂D₃ treatment (Fig. 5, B and C). These findings were supplemented by fluorescent immunocytochemical experiments in which 1,25-(OH)₂D₃ stimulation caused visible VDR nuclear translocation in both genotypes, but only RXRα translocation in WT BMSC (Fig. 5D). Although MKP-1 does not appear to inhibit 1,25-(OH)₂D₃-induced VDR translocation, these data provide clear evidence that MKP-1 may be required for RXRα nuclear translocation.

Fig. 5.

MKP-1 is required for 1,25-(OH)₂D₃-induced RXRα nuclear import. A, Representative Western blot of cytoplasmic/nuclear expression of VDR and RXRα from BMSC treated with 10⁻⁷ m 1,25-(OH)₂D₃ for 1 h. B, Densitometric quantification of RXRα and VDR protein levels normalized to GAPDH or lamin A/C. C, Densitometric comparison of VDR and RXRα nuclear import in primary WT and MKP-1^−/− BMSC and total relative VDR and RXRα present in nuclear and cytoplasmic fractions. D, Immunofluorescence cytochemistry of WT and MKP-1^−/− BMSC for RXRα and VDR protein localization. White scale bar, 10 μm. (n = 3; *, P < 0.05). Ctrl, Control; Veh, vehicle.

When this 1-h 1,25-(OH)₂D₃ stimulation was repeated and WT and MKP-1^−/− BMSC were subjected to a PLA using both anti-RXRα (rabbit) and anti-VDR (mouse) antibodies, we observed that there was little difference in unstimulated total or nuclear PLA signals and an insignificant difference in the total 10⁻⁷ m 1,25-(OH)₂D₃-stimulated PLA signals formed; yet the number of nuclear signals after 10⁻⁷ m 1,25-(OH)₂D₃ stimulation was 64.9% lower (P < 0.05) in the MKP-1^−/− BMSC. Negative controls stimulated with 10⁻⁷ m 1,25-(OH)₂D₃ and incubated with anti-RXRα only or anti-VDR only plus species control IgG yielded few PLA signals (Fig. 6, A and B). Taken together, use of the PLA system suggests that endogenous 1,25-(OH)₂D₃-induced RXRa-VDR heterodimer nuclear translocation in primary BMSC is highly dependent upon MKP-1.

Fig. 6.

MKP-1 is required for 1,25-(OH)₂D₃-induced nuclear translocation of the VDR-RXRα heterodimer. WT and MKP-1^−/− BMSC were treated with vehicle control or 10⁻⁷ m 1,25-(OH)₂D₃ for 1 h and were subjected to PLA with antibodies against VDR (mouse) and RXRα (rabbit); confocal images were captured for quantification. A, Representative confocal images of PLA signals in WT and MKP-1^−/− BMSC. B, Quantification of total and nuclear PLA signals in the Z-range of the DAPI stained nuclei for the indicated treatments. White scale bar, 10 μm (n = 3; *, P < 0.05). Ctrl, Control.

MKP-1 is required in stromal cells for 1,25-(OH)₂D₃ induction of osteoclastogenesis

The findings suggest that BMSC are dependent on MKP-1 for their ability to induce osteoclastogenesis in response to a 1,25-(OH)₂D₃ stimulus. To test this prediction, we used an osteoclast coculture system (34, 35) in which WT and MKP-1^−/− BMSC were magnetically depleted of CD11b⁺ cells to reduce primary macrophage contamination. They were then cultured with the RAW264.7 cells, after which they were treated with 10⁻⁸ m 1,25-(OH)₂D₃ ± recombinant osteoprotegerin. After 7 d of treatment, TRAP staining was performed and osteoclasts (TRAP⁺, ≥ 3 nuclei) were enumerated (Fig. 7A). Cocultures on hydroxyapatite discs were incubated for 10 d to allow for matrix resorption, and pit area and pit number were quantified (Fig. 7, B and C). As anticipated, MKP-1^−/− BMSC and RAW264.7 cell cocultures had a significant reduction in the number of osteoclasts formed, as well as a decrease in the number of nuclei/osteoclasts (Fig. 6A). When was osteoprotegerin was added, the osteoclastogenesis was almost completely blocked in both cocultures. The area of resorbed matrix per hydroxyapatite disc was also significantly reduced in the MKP-1^−/− coculture, as was the number of circumscribed areas of resorption. Taken together, these coculture data demonstrate a functional role for MKP-1 in permitting the 1,25-(OH)₂D₃ induction of RANKL-mediated osteoclastogenesis in BMSC that is predicted by MKP-1's role in canonical 1,25-(OH)₂D₃ signaling.

Fig. 7.

MKP-1 is required for maximal 1,25-(OH)₂D₃-induced osteoclastogenesis. WT and MKP-1^−/− BMSC were cocultured with RAW264.7 cells to test the osteoclast-inducing potential of 1,25-(OH)₂D₃-stimulated stromal cells. A, Quantification of osteoclasts per well and nuclei per osteoclast. Coculture osteoclast-resorbing activity was quantified on osteologic discs. B, Representative pit images and C) pits per well and total pit area per well. (n = 3; P < 0.05; **, P < 0.01; ***, P < 0.001). Ctrl, Control; OPG, osteoprotegerin.

Discussion

In addition to cellular interactions via cytokines, skeletal and immune cellular processes display significant cross talk through common transcription factors, signaling molecules, and membrane receptors (39). MKP-1 is one such immune regulatory protein that has been studied recently for its additional role in osteoclastogenesis and mineralization (18, 40). Immunologically, MKP-1 is primarily known for its role in the negative regulation of inflammatory cytokine responses through dephosphorylation of p38 and JNK MAPK. For a stress-induced gene historically thought of as a key negative regulator for many cellular processes (41–43), evidence that MKP-1 actually permits RANKL expression truly represents a paradigm shift in our understanding of MKP-1 signaling.

Previous work in our laboratory focused on the role of MKP-1 in attenuating chronic inflammatory bone loss generated in experimental periodontal disease models. TRAP staining of LPS-treated maxilla indicated that MKP-1^−/− mice had increased localized osteoclastogenesis, and micro-computed tomography confirmed that there was increased bone loss (24). Likewise, gain of function studies showed that adenoviral delivery of MKP-1 attenuated IL-6 expression, inflammatory infiltrate, and osteoclastogenesis (44). We followed up on this finding in cultured primary BMSC by investigating the expression of RANKL in response to a IL-1β, TNF-α, and LPS challenge from two periodontal pathogens known to stimulate RANKL along with a positive control of 1,25-(OH)₂D₃. Because the presence of MKP-1 has been found to attenuate the expression of inflammatory cytokines (45, 46), the finding that MKP-1 permits RANKL expression was quite surprising, and supports a novel role for MKP-1 in 1,25-(OH)₂D₃-induced gene expression.

Not knowing whether our initial finding was limited to RANKL only, our data indicate that 1,25-(OH)₂D₃-induced VDR and CYP24a1 mRNA expression [two well-established 1,25-(OH)₂D₃ targets] are highly dependent on the presence of MKP-1 signaling. Thus, the role of MKP-1 in 1,25-(OH)₂D₃ target gene expression is not restricted only to RANKL and may be common for the 1,25-(OH)₂D₃ genomic signaling pathway. Although we cannot completely rule out a role for MKP-1 in vitamin D's rapid, nongenomic membrane effects, no evidence for this activation exists for these genes, and similar mRNA expression levels were seen at 0, 2, and 4 h, indicating that MKP-1 plays a primary role in the genomic vitamin D pathway (30). Additionally, because all three vitamin D target genes (RANKL, VDR, and CYP24a1) are inhibited in MKP-1^−/− BMSC at approximately the same time, baseline VDR protein levels are similar in both genotypes, and cycloheximide treatment did not rescue 1,25-(OH)₂D₃ induction (data not shown), we do not suspect that decreased target-gene expression results from a primary defect in VDR expression.

Although evidence suggests that osteoblasts are not the only cells that express RANKL (47), it has been established that cells of stromal osteoblast origin are the primary RANKL-expressing cells (37, 47, 48). To confirm that the MKP-1 has this novel role in stromal cells, we knocked down MKP-1 expression in a stromal/osteoblast cell line (UAMS-32), which revealed a significant decrease in RANKL, VDR, and CYP24a1 mRNA after 1,25-(OH)₂D₃ treatment. These data were very similar to our observations in primary BMSC, leading us to believe that stromal osteoblasts were a significant contributor. To test whether this consequence was due to a decrease in promoter activity, we repeated this siRNA and 1,25-(OH)₂D₃ treatment on the same cell line containing a bacterial artificial chromosome-based transcriptional reporter construct containing the entire mouse RANKL gene with 120 kb of RANKL 5′-flanking region and 14 kb 3′ flanking region (32). The result that MKP-1 siRNA was able to knockdown 1,25-(OH)₂D₃-induced RANKL gene promoter activity in UAMS-32 cells supports a role for MKP-1 in regulating both 1,25-(OH)₂D₃ RANKL mRNA and genomic RANKL activation in stromal osteoblasts. Similar MKP-1 silencing data showing reduced 1,25-(OH)₂D₃-induced VDR and CYP24a1 expression suggest that MKP-1 signaling may be required for other vitamin D target genes' expression as well. Based on these findings, we suspected that MKP-1 regulates 1,25-(OH)₂D₃ genomic action by permitting VDRE promoter activity, and that the results obtained using primary BMSC are due, in part, to a contribution from stromal/osteoblast cells.

This suspicion was strengthened because 1,25-(OH)₂D₃-stimulated MKP-1^−/− BMSC nuclei had a decreased capacity for VDRE binding. The mOP oligonucleotide is an established VDRE consensus sequence, and its binding by VDR and RXRα through EMSA has been used to independently simulate genomic 1,25-(OH)₂D₃ genomic action (9, 33). Previous work by Kim et al. (11) showed the mRLD5 located at −76kb is the primary site of RANKL gene activation by 1,25-(OH)₂D₃-induced VDR-RXRα heterodimer binding in mouse osteoblast cells. Based on this work, our ChIP results suggest that MKP-1 may be required for 1,25-(OH)₂D₃-induced RXRα species binding at the mRLD5 region. The observed increase in 1,25-(OH)₂D₃-induced VDR binding at mRLD5 in MKP-1^−/− cells may represent an increase in VDR-VDR homodimer binding or weaker VDR monomer the VDRE binding of which is strengthened via cross-linking. Either way, this new VDR species binding in MKP-1^−/− cells likely occurs as compensation for a decrease in available nuclear VDR-RXRα heterodimer. Because coregulatory complex recruitment and transcriptional activation depends on the VDR-RXRα heterodimer (49, 50), we predict that these findings result in the observed decrease in RANKL gene activation and could cause the decrease in VDR and CYP-24a1 expression through a similar mechanism. This disruption in VDRE binding could depend on MKP-1's function to prevent a defect in VDR-RXRα heterodimer conformation, formation, or in its nuclear translocation.

The possibility that MKP-1 is involved in 1,25-(OH)₂D₃-induced nuclear trafficking was addressed initially by immunoblot analysis and confocal microscopy. Experimental evidence presented reveals that 1,25-(OH)₂D₃ induces a VDR translocation from the cytoplasmic to the nuclear fraction in both cell genotypes, although this VDR translocation may be delayed in MKP-1^−/− BMSC, as evidenced by the relatively small reduction in cytoplasmic VDR upon 1,25-(OH)₂D₃ stimulation. The results show that RXRα translocates freely into the nucleus from the cytosol in WT BMSC, presumably as a VDR-RXRα heterodimer. However, despite slightly higher baseline RXRα protein in MKP-1^−/− BMSC, there is no detectable translocation, as evidenced by the unchanging protein levels after 1,25-(OH)₂D₃ treatment. Therefore, given that RXRα dictates VDR-RXRα heterodimer translocation (51), and the VDR-RXRα heterodimer is known to have a significantly higher affinity for genomic VDRE than the VDR protein alone (8), this suggests that the decreased RXRα nuclear translocation contributes to the inhibition of VDR-RXRα VDRE binding observed in vitro by EMSA and endogenously by ChIP. Although RXRα is not known to translocate in response to 1,25-(OH)₂D₃ as a monomer, we confirm this by PLA, which tests specifically for RXRα-VDR heterodimer localization. Although there was little difference in unstimulated total or nuclear PLA signals and an insignificant difference in the total 10⁻⁷ m 1,25-(OH)₂D₃-stimulated PLA signals, the number of nuclear signals after 10⁻⁷ m 1,25-(OH)₂D₃ stimulation is significantly lower in the MKP-1^−/− BMSC. These data support observations obtained from nuclear translocation studies and indicate that MKP-1 is required for 1,25-(OH)₂D₃-induced nuclear translocation of the VDR-RXRα heterodimer. Although not significant, the decrease in total PLA signals observed in MKP-1^−/− cells leaves open the possibility that MKP-1 may also contribute to 1,25-(OH)₂D₃ ligand-initiated VDR-RXRα heterodimer formation. Although we cannot rule out that MKP-1 potentially preserves RXRα or VDR baseline expression or half-life, VDR and RXRα total protein levels were very similar in both genotypes; therefore, this should be considered a minor contribution to the observed phenotype, if at all. These data show that MKP-1 is necessary for the proper nuclear transport of VDR-RXRα heterodimer that is a likely contributor to the observed decreased capacity for VDR-RXRα VDRE binding in MKP-1^−/− BMSC.

Having established that stromal osteoblast cells have a diminished capacity for 1,25-(OH)₂D₃-induced RANKL promoter activity, mRNA, and soluble protein expression, we predicted that BMSC are dependent upon MKP-1 for their ability to induce osteoclastogenesis in response to a 1,25-(OH)₂D₃ stimulus. The reduced osteoclastogenesis, nuclei/osteoclasts and mineralized matrix resorption in the primary BMSC and RAW264.7 coculture system is a predictable result given the insufficient stromal osteoclastogenic signal (RANKL) in response to 1,25-(OH)₂D₃. Because osteoclastogenesis was almost completely blocked by osteoprotegerin in both cocultures, the observed reduction in osteoclast formation/activity in MKP-1^−/− cocultures is likely due to reduced 1,25-(OH)₂D₃-induced RANKL expression in the stromal population.

Future studies will focus on potential phosphorylation targets of MKP-1 in genomic 1,25-(OH)₂D₃ signaling. MKP-1 is known to dephosphorylate and deactivate the p38, JNK, and ERK pathways, and so future studies will focus on the potential that our observations stem from an increase in MAPK phosphorylation/activation in the absence of MKP-1. A potential complicating factor in MKP-1^−/− BMSC is that the already reduced nuclear RXRα population may also have an altered structural conformation, due to posttranslational modification(s) that would limit its VDRE binding ability. Previous studies have shown that MAPK activation can directly inhibit genomic vitamin D activity through RXRα phosphorylation by JNK at serine 260 in human keratinocytes (26), and by ERK in the AF-1 domain in human prostate epithelial cells (27); however, the mechanism for how RXRα phosphorylation inhibits genomic vitamin D signaling was not shown. Given this knowledge and our findings that VDR-RXRα heterodimer translocation is impaired, it is possible that MKP-1 may be necessary to prevent a phosphorylation of RXRα, or phosphorylation of heterodimerized VDR, which impairs the 1,25-(OH)₂D₃-initiated translocation of the VDR-RXRα heterodimer, contributing to diminished VDRE binding and decreased 1,25-(OH)₂D₃ target gene expression.

This work provides the first evidence that MKP may regulate RXRα cellular trafficking and that these effects have functional consequences in vitamin D signaling. An additional consequence of these findings is that MKP-1 may have more global consequences in RXR heterodimer partners; therefore, the action of thyroid hormone, the prostanoids, and retinoids may also have altered signaling in the absence of MKP-1, which may contribute to the phenotype of MKP-1^−/− mice. Additionally, the induction of MKP-1 mRNA by 1,25-(OH)₂D₃ and the need for MKP-1 in 1,25-(OH)₂D₃ signaling is suggestive of a positive feedback role for MKP-1.

These findings support a novel role for MKP-1 in permitting canonical 1,25-(OH)₂D₃ signaling via VDR-RXRα heterodimer nuclear import in BMSC. If MKP-1 in humans shows similar regulation of genomic 1,25-(OH)₂D₃ signaling, then these results may provide a scientific basis for future therapies whereby MKP-1 is targeted to control downstream 1,25-(OH)₂D₃ target gene expression. Taken together, we provide a clear link between a new permissive role for MKP-1 in 1,25-(OH)₂D₃ genomic signaling and its functional consequences in osteoclastogenesis via RANKL, while presenting mechanistic evidence for MKP-1 as a new player in osteoimmunology.

NURSA Molecule Pages^†:

• Nuclear Receptors: VDR;

• Ligands: 1,25-dihydroxyvitamin D₃.