Vitamin D Receptor Inhibits Nuclear Factor κB Activation by Interacting with IκB Kinase β Protein*

Background: 1,25(OH)2D3 inhibits NF-κB activation by an undefined mechanism. Results: Vitamin D receptor protein binds to IKKβ protein, blocking TNFα-induced IKK complex formation and NF-κB activity. Conclusion: The vitamin D receptor suppresses NF-κB activation by directly interacting with IKKβ. Significance: This is a novel mechanism whereby 1,25(OH)2D3-VDR inhibits NF-κB. 1,25-Dihydroxyvitamin D (1,25(OH)2D3) is known to suppress NF-κB activity, but the underlying mechanism remains poorly understood. Here we show that the vitamin D receptor (VDR) physically interacts with IκB kinase β (IKKβ) to block NF-κB activation. 1,25(OH)2D3 rapidly attenuates TNFα-induced p65 nuclear translocation and NF-κB activity in a VDR-dependent manner. VDR overexpression inhibits IKKβ-induced NF-κB activity. GST pull-down assays and coimmunoprecipitation experiments demonstrated that VDR physically interacts with IKKβ and that this interaction is enhanced by 1,25(OH)2D3. Protein mapping reveals that VDR-IKKβ interaction occurs between the C-terminal portions of the VDR and IKKβ proteins. Reconstitution of VDR−/− cells with the VDR C terminus restores the ability to block TNFα-induced NF-κB activation and IL-6 up-regulation. VDR-IKKβ interaction disrupts the formation of the IKK complex and, thus, abrogates IKKβ phosphorylation at Ser-177 and abolishes IKK activity to phosphorylate IκBα. Consequently, stabilization of IκBα arrests p65/p50 nuclear translocation. Together, these data define a novel mechanism whereby 1,25(OH)2D3-VDR inhibits NF-κB activation.

nuclear translocation, blocks NF-B DNA binding, increases IB␣ levels, or stabilizes IB␣ protein (10,12,14,15,17,18). It has also been shown that 1,25(OH) 2 D 3 suppresses RelB transcription (19) and reduces p105/p50 and c-rel protein levels (20). Interestingly, p65 has been reported to physically interact with liganded VDR to modulate the transactivating activity of the VDR (21). Despite all these reports, a convincing mechanism to explain the relatively rapid inhibitory action of vitamin D hormone on NF-B activity is lacking. Particularly, how vitamin D increases or stabilizes IB␣, the most critical step in NF-B regulation, remains unexplainable. In this report we elucidate a novel molecular mechanism by which 1,25(OH) 2 D 3 -VDR attenuates NF-B activation. Our data demonstrate that the VDR protein is able to directly interact with IKK␤ protein to block the canonical NF-B activation pathway.

EXPERIMENTAL PROCEDURES
Cell Culture and Transfection-HEK293 and RAW264.7 cells were purchased from the ATCC. Generation of VDR ϩ/Ϫ and VDR Ϫ/Ϫ mouse embryonic fibroblasts (MEFs) were reported previously (10). All cells were cultured in DMEM supplemented with 10% FBS at 37°C and 5% CO 2 . Cell transfection was carried out using Lipofectamine 2000 (Invitrogen) according to the instructions of the manufacturer. Cells were treated with 10 ng/ml recombinant mouse TNF␣ (Millipore) and/or 20 nM 1,25(OH) 2 D 3 unless indicated otherwise.
Western Blot Analyses-Proteins were separated by SDS-PAGE and electroblotted onto Immobilon-P membranes. Western blot analyses were carried out as described previously (22). The following antibodies were used in this study. Anti-IKK␣/␤, anti-p-IKK␣/␤, anti-IKK␣, anti-IKK␤, anti-IKK␥, and anti-HA were obtained from Santa Cruz Biotechnology (Santa Cruz, CA). Anti-FLAG and anti-␤-actin were obtained from Sigma.
Luciferase Reporter Assays-HEK293 cells or MEFs were cotransfected with pNF-B-Luc, pCI-HA-p65, or pCI-HA-IKK␤ (or its N-or C-terminal constructs) and pcDNA-VDR plasmids (or its N-or C-terminal constructs) using Lipofectamine 2000 (Invitrogen). Transfected cells were treated with TNF␣ in the presence or absence of 1,25(OH) 2 D 3 as indicated in each experiment. After 24 h, the cells were lysed, and luciferase activity was determined using the Dual-luciferase reporter assay system (Promega) as reported previously (9). Luciferase activity was normalized to the Renilla luciferase activity, which served as an internal control for transfection efficiency.
GST Pull-down Assays-GST-hVDR fusion protein was generated using the pGEX-4T-1 plasmid as reported previously (9). IKK␣, IKK␤, p50, and p65 proteins were synthesized in the presence of [ 35 S]methionine using an in vitro transcription and translation system (Promega). GST or GST-hVDR beads were incubated with 35 S-labeled IKK␣, IKK␤, p50, or p65 overnight. In some experiments, 20 nM 1,25(OH) 2 D 3 was included in the incubation. After being washed five times, the beads were spun down and dissolved in Laemmli sample buffer. After being boiled for 5 min, the proteins were resolved using SDS-PAGE and visualized by autoradiography.
IKK Assays-IKK complexes from whole-cell extracts were precipitated with anti-IKK-␥ antibodies (Santa Cruz Biotechnology) and protein A/G-Sepharose beads (Millipore). After 2 h of incubation, the beads were washed with lysis buffer and then assayed in a kinase assay mixture containing 50 mM HEPES (pH 7.4), 20 mM MgCl 2 , 2 mM DTT, 20 Ci [␥-32 P]ATP, 10 mM unlabeled ATP, and 2 g of GST-IB␣ (amino acids 1-54) substrate (Clontech). After incubation at 30°C for 30 min, the reaction was terminated by 5 min of boiling in loading sample buffer. Finally, the proteins were resolved by 10% SDS-PAGE, and the radiolabeled substrate bands were visualized by autoradiography. To determine the total amount of IKK␤ in each sample, 50 g of the whole-cell extracts were resolved by 10% SDS-PAGE, electrotransferred to a nitrocellulose membrane, and blotted with anti-IKK␤ antibody.
Statistical Analysis-Data values were presented as mean Ϯ S.D. Statistical comparisons were carried out using unpaired two-tailed Student's t test or one-way analysis of variance as appropriate, with p Ͻ 0.05 being considered significant.

Vitamin D Blocks TNF␣-induced NF-B Activation-We
first performed luciferase reporter assays to confirm the inhibitory effect of 1,25(OH) 2 D 3 on NF-B. As shown in Fig. 1, in MEF cells transfected with the pNF-B-Luc reporter plasmid, TNF␣ drastically induced NF-B luciferase activity. This induction was suppressed markedly by 1,25(OH) 2 D 3 cotreatment in VDR ϩ/Ϫ MEFs but not in VDR Ϫ/Ϫ MEF cells (Fig. 1A). Immunostaining showed that TNF␣-induced p65 nuclear translocation was blocked by overnight 1,25(OH) 2 D 3 pretreatment in VDR ϩ/Ϫ MEF but not in VDR Ϫ/Ϫ MEF cells (Fig. 1, B and C), confirming the requirement of VDR for inhibition of NF-B activation. Inhibition of TNF␣-induced NF-B activity by 1,25(OH) 2 D 3 was also observed in HEK293 cells (not shown). Interestingly, a short exposure (1-2 h) of the VDR ϩ/Ϫ MEF cells to 1,25(OH) 2 D 3 was sufficient to block TNF␣-in-duced p65 nuclear entry (Fig. 1, D and E), suggesting that it is unlikely that this inhibitory action involves a transcriptional event, which usually takes at least several hours. Indeed, 1,25(OH) 2 D 3 blocked TNF␣-induced degradation of IB␣ in cells, and this activity was not affected by actinomycin D, an inhibitor of RNA synthesis (Fig. 1F). Moreover, 1,25(OH) 2 D 3 treatment did not significantly alter the mRNA levels of NF-B components IKK␣, ␤, ␥, IB␣, and p65 (Fig. 1G). These data confirmed that it is unlikely that 1,25(OH) 2 D 3 suppresses NF-B activation by a transcriptional mechanism. Interestingly, VDR overexpression in cells by transfection was suffi-cient to suppress TNF␣-induced NF-B activity dose-dependently in the absence of 1,25(OH) 2 D 3 (Fig. 1H), suggesting that, at high concentrations, VDR can suppress NF-B in a ligandindependent manner. It appears that the rapid blockade of p65 nuclear translocation can be explained by VDR interaction with p65, as reported previously.
It is well known that overexpression of p65 or IKK␤ induces NF-B activity in the absence of extracellular stimuli. We found that VDR cotransfection was unable to attenuate p65-induced NF-B activity in HEK293 cells, regardless of 1,25(OH) 2 D 3 treatment (Fig. 1I), but it markedly suppressed IKK␤-induced FIGURE 1. 1,25-dihydroxyvitamin D rapidly attenuates NF-B activation in a VDR-dependent manner. A, NF-B luciferase reporter assays. VDR ϩ/Ϫ and VDR Ϫ/Ϫ MEFs transfected with the pNF-B-Luc reporter were treated with TNF␣ (10 ng/ml) and/or 1,25(OH) 2 D 3 (20 nM) (1,25VD) as indicated for 24 h before measuring luciferase activity. ***, p Ͻ 0.001. B and C, effects of 1,25(OH) 2 D 3 on p65 nuclear translocation. VDR ϩ/Ϫ and VDR Ϫ/Ϫ MEFs were pretreated with vehicle or 1,25(OH) 2 D 3 overnight, followed by 2 h of TNF␣ stimulation as indicated. The cells were immunostained with anti-p65 antibodies (B), and VDRpositive nuclei were quantified in each cell type (C). Note that p65 nuclear translocation could not be blocked by 1,25(OH) 2 D 3 in VDR Ϫ/Ϫ MEFs. ***, p Ͻ 0.001 versus VDR ϩ/Ϫ . D and E, rapid inhibition of p65 nuclear translocation by 1,25(OH) 2 D 3 . VDR ϩ/Ϫ MEFs were not treated (control) or pretreated with 1,25(OH) 2 D 3 for 0, 1, or 2 h as indicated, followed by 2 h of TNF␣ stimulation. Intracellular p65 location was assessed by immunostaining with anti-p65 antibodies (D), and VDR-positive nuclei were quantified in each treatment (E). Nuclei were stained with DAPI. ***, p Ͻ 0.001 versus controls. NF-B activity in a VDR dose-dependent manner, even in the absence of 1,25(OH) 2 D 3 , although 1,25(OH) 2 D 3 treatment further increased the inhibitory activity of VDR (J). Similar results were observed in MEF cells (not shown). These observations are inconsistent with the assumption that VDR-p65 interaction arrests p65 translocation, leading to inhibition of NF-B activity, but raise a possibility of VDR-IKK␤ interaction in this regulatory process.
VDR Physically Interacts with IKK␤ Protein-That VDR regulates biological activities by interacting with other regulatory proteins has been well documented previously. For example, our previous work showed that 1,25(OH) 2 D 3 -activated VDR binds to CREB and suppresses renin gene transcription by blocking the formation of CREB-CREB-binding protein-p300 complex on the CRE site in the renin gene promoter (9). VDR binds to ␤-catenin protein to inhibit its nuclear translocation in colon cancer cells, thus blocking the transduction of the oncogenic signal of ␤-catenin to the nuclei (8). To explore the apparently non-transcriptional mechanism whereby 1,25(OH) 2 D 3 suppresses NF-B activity, we performed GST pull-down assays to examine the protein-protein interaction between VDR and NF-B components. Interestingly, purified GST-VDR fusion protein ( Fig. 2A) was able to pull down 35 S-labeled IKK␤ protein strongly in vitro (Fig. 2B). This interaction was not altered substantially by the presence of 1,25(OH) 2 D 3 (data not shown), consistent with the above observation that, at high concentrations, VDR suppressed NF-B even in the absence of 1,25(OH) 2 D 3 (Fig. 1, H-J). Surprisingly, given the previously reported VDR-p65 interaction (21), we barely detected any pull-down of 35 S-labeled p65 protein by GST-VDR under the same condition (Fig. 2B). There appeared to be some weak interaction between GST-VDR and p50 or IKK␣ (data not shown). The latter was not unexpected, given that IKK␣ and IKK␤ share extensive structural homology. Together, these data suggest that VDR may target IKK, not p65, to inhibit NF-B activation.
The strong association between VDR and IKK␤ prompted us to focus on this interaction. Co-IP assays showed that, in HEK293 cells transfected with the FLAG-VDR plasmid, anti-FLAG antibodies were able to coprecipitate endogenous IKK␤ and that this action was enhanced markedly in the presence of 1,25(OH) 2 D 3 (Fig. 2C). When both FLAG-VDR and IKK␤ were overexpressed in HEK293 cells by transfection, anti-FLAG antibodies were able to coprecipitate IKK␤ without TNF␣ and 1,25(OH) 2 D 3 stimulation (Fig. 2D). Furthermore, in untransfected cells, anti-VDR antibodies were able to weakly coprecipitate IKK␤ in the absence of 1,25(OH) 2 D 3 . However, the VDR-IKK␤ interaction was enhanced greatly in the presence of TNF␣ and 1,25(OH) 2 D 3 (Fig. 2E). Through co-IP assays, we also observed 1,25(OH) 2 D 3 -induced VDR-IKK␤ interaction in RAW264.7 cells, a macrophage cell line (Fig. 2F), indicating that this protein-protein interaction is not cell-specific and also occurs in immune cells. These data confirm that the physical association between VDR and IKK␤ occurs within cells and that this interaction can take place independently of 1,25(OH) 2 D 3 at high protein concentrations. Consistent with the notion that 1,25(OH) 2 D 3 binding is not required, we observed that overexpression of hVDR mutants at R274L and R391C within the ligand-binding domain (LBD) (Fig. 3A) with extremely low 1,25(OH) 2 D 3 affinity were still able to block TNF␣-induced NF-B activation, regardless of 1,25(OH) 2 D 3 treatment (Fig. 3B). However, under normal physiological conditions where intracellular VDR levels are usually very low in most cell types, particularly in immune cells, VDR needs ligand activation to down-regulate NF-B activity. This is the basis to explain why 1,25(OH) 2 D 3 treatment suppresses NF-B activity.
VDR and IKK␤ Interact at Their C Terminus-VDR contains an N-terminal DNA-binding domain, a C-terminal LBD, and a hinge region between them (6) (Fig. 3A). To define which domain in the VDR molecule interacts with IKK␤, we generated plasmid constructs that express an HA-tagged N-terminal DNA binding domain (VDR-N, amino acids 1-119) and C-terminal hinge and LBD (VDR-C) of hVDR (amino acids 119 -427) (Fig. 3A). Co-IP experiments showed that in HEK293 cells transfected with FLAG-IKK␤ and HA-VDR, HA-VDR-N, or HA-VDR-C, anti-FLAG antibodies were able to pull down HA-VDR-C but not HA-VDR-N (Fig. 3C). Conversely, anti-HA antibodies coprecipitated FLAG-IKK␤ only in cells cotransfected with HA-VDR or HA-VDR-C and not in cells transfected with HA-VDR-N (Fig. 3D). These results indicate that the C-terminal hinge and LBD fragment of VDR protein interacts with IKK␤.
To define the domain in the IKK␤ molecule that interacts with the VDR, we generated plasmids expressing an HA-tagged IKK␤ N-terminal fragment between amino acids 1-346 and a C-terminal fragment between amino acids 341-756, respectively (Fig. 3E). In HEK293 cells cotransfected with FLAG-hVDR and HA-IKK␤-N or HA-IKK␤-C, anti-FLAG antibodies were able to coprecipitate HA-IKK␤-C but not HA-IKK␤-N (Fig. 3F). These results indicate that the VDR interacts with the C-terminal portion of IKK␤ protein in cells. Together, these data reveal that VDR and IKK␤ interaction occurs at their C-terminal portions.
The C Terminus of VDR Is Functional in the Regulation of NF-B-Because the VDR C terminus binds to IKK␤, a key question that needs to be addressed is whether VDR-C is able to suppress NF-B activity. As expected, both VDR-N and VDR-C lacked transactivating activity in VDRE-Luc reporter assays (Fig. 4A). By NF-B luciferase reporter assays, however, we observed that VDR-C, but not VDR-N, was able to attenuate , and the precipitates were blotted (IB) with anti-HA antibodies (C) or anti-FLAG antibodies (D) as indicated. As controls, these precipitates were also blotted with the same antibodies as shown in the lower panels in C and D. Note that IKK␤ interacts with VDR-C. E, schematic of IKK␤ protein and its N-terminal and C-terminal constructs (IKK␤-N and IKK␤-C). F, HEK293 cells were cotransfected with FLAG-VDR and HA-IKK␤-N or HA-IKK␤-C. Cell lysates were precipitated with anti-FLAG antibodies, and the precipitates were blotted with anti-HA antibodies. The input lysates were blotted with anti-HA or anti-FLAG antibodies, respectively, as indicated at the bottom. Note that the VDR interacts with IKK␤-C and not IKK␤-N.
IKK␤-induced NF-B activity in HEK293 cells, similar to fulllength VDR and that the inhibitory activity of both the VDR and VDR-C was enhanced in the presence of 1,25(OH) 2 D 3 (Fig. 4B). To eliminate the potential confounding effect of the endogenous mouse VDR, we asked whether reconstitution of VDR Ϫ/Ϫ MEF cells with VDR-C would be able to restore the ability to suppress NF-B activation. Using different plasmid doses, we observed that transfection of VDR Ϫ/Ϫ MEFs with 0.1 g of hVDR construct/well reconstituted intracellular VDR to a level comparable with that seen in VDR ϩ/Ϫ MEFs (Fig. 4C). Therefore, we performed VDR Ϫ/Ϫ MEF cell transfection using the same dose (0.1 g/well) of VDR, VDR-N, VDR-C, or control empty vector to avoid overexpression (Fig. 4D). Interestingly, in VDR Ϫ/Ϫ MEFs, VDR and VDR-C, but not VDR-N, were able to attenuate TNF␣-induced NF-B activity (Fig. 4E) and IL-6 upregulation (F), and this attenuation was enhanced when the cells were treated with 1,25(OH) 2 D 3 (E and F). This was not surprising because VDR-C has a LBD for 1,25(OH) 2 D 3 binding. Together, these results demonstrate that reconstitution of VDR Ϫ/Ϫ cells with the C terminus of hVDR to a physiological level is sufficient to block TNF␣ induction of NF-B activity and IL-6 expression. Because VDR-C has no DNA binding domain, these observations provide very compelling evidence that VDR-IKK␤ interaction can regulate biological actions independently of VDRE.

DISCUSSION
Vitamin D inhibition of NF-B has been reported frequently in the literature, but the exact molecular mechanism remains poorly understood. It has been documented previously that 1,25(OH) 2 D 3 arrests the nuclear translocation of p65/p50 and suppresses the degradation of IB␣ protein (10,14,25). Because the VDR can interact directly with p65 (21), the most commonly held mechanism to explain vitamin D inhibition of NF-B is that the VDR-p65 interaction blocks the nuclear translocation of p65/p50 (10,26). This mechanism, however, cannot explain how 1,25(OH) 2 D 3 stabilizes IB␣, which was reported in many studies (10,12,14,17,18) and is well known as a critical step in the inhibition of NF-B. In fact, in many studies (14,26), including this one, VDR-p65 interaction was not detectable, suggesting that VDR-p65 interaction is weak, if there is any. Therefore, there likely exist other mechanisms to explain the stabilization of IB␣.
In this report, we present evidence that VDR binds to IKK␤ to block NF-B activation. We showed that VDR-IKK␤ interaction blocks the formation of the IKK complex and, hence, reduces IKK␤ phosphorylation. As a result, the IKK enzymatic activity to phosphorylate IB␣ is abrogated, consequently diminishing IB␣ ubiquitylation and degradation. This mechanism explains well how 1,25(OH) 2 D 3 stabilizes IB␣. The direct consequence of reduced IB␣ degradation is the retention of the p65/p50 heterodimer in the cytoplasm, leading to decreased NF-B transcriptional activity. Thus, this model also explains the blockade of p65/p50 nuclear translocation. We conclude that this is a major mechanism whereby 1,25(OH) 2 D 3 -VDR inhibits NF-B activation.
Our data show that VDR-IKK␤ interaction occurs in the C-terminal portions of both molecules. These mapping studies provide compelling evidence that confirms the interaction between the VDR and IKK␤ proteins. We demonstrated that reconstitution of VDR Ϫ/Ϫ cells with the C terminus of hVDR to a physiologically relevant level is sufficient to block the induction of NF-B activity and IL-6 expression by TNF␣. Because VDR-C has no DNA binding domain, this result confirms that VDR-IKK␤ interaction can generate a biological consequence independently of the VDRE. Although VDR-IKK␤ interaction does not require 1,25(OH) 2 D 3 at high protein concentrations under some artificial conditions (e.g. in the case of cell transfection), 1,25(OH) 2 D 3 is able to enhance this interaction in cells. The VDR C-terminal region that interacts with IKK␤ remains responsive to 1,25(OH) 2 D 3 treatment because VDR-C contains the LBD. Under normal physiological conditions, intracellular VDR levels are usually low in most cell types, particularly in immune cells. Therefore, we believe that, physiologically, VDR needs ligand activation to block NF-B activity. This is the basis for the observation that 1,25(OH) 2 D 3 treatment suppresses NF-B activity.
As a ligand-activated transcription factor, the VDR usually interacts with cis-DNA elements (VDRE) in gene promoters to activate gene transcription. This mechanism is used in most stimulatory regulation of vitamin D actions. The mechanisms for negative regulation, however, are more complicated and diverse. For instance, the VDR/retinoid X receptor heterodimer and VDR homodimer inhibit the formation of the NFAT-1/AP-1 transcriptional complex in IL-2 and GM-CSF promoters to inhibit these gene expressions (27,28). VDR binds to a negative VDRE (nVDRE) in PTH and PTHrP gene promoters and works with Ku antigen to suppress these genes (29). Ligand-activated VDR can also recruit corepressors, such as NCoR, Alien, and SMRT, to mediate transcriptional repression (30,31), and 1,25(OH) 2 D 3 suppresses Cyp27b1 transcription via interaction with VDIR (32). Given these diverse inhibitory mechanisms, it is not surprising that 1,25(OH) 2 D 3 -VDR downregulates NF-B by protein-protein interaction with IKK␤ because this kind of regulatory mode has been observed in the regulation of the PKA/CREB and Wnt/␤-catenin pathways. We reported previously that 1,25(OH) 2 D 3 -activated VDR binds to CREB and inhibits renin gene transcription by blocking the formation of the CREB-CBP⅐p300 complex on the CRE site in the renin gene promoter (9). In the case of the Wnt/␤-catenin pathway, liganded VDR binds to ␤-catenin protein to inhibit its nuclear translocation in colon cancer cells, thus blocking the transduction of the oncogenic signal of ␤-catenin to the nuclei (8). Detailed mapping studies reveal that the interaction between VDR and ␤-catenin occurs between the VDR activator function 2 (AF-2) domain of the VDR and the ␤-catenin C terminus (33). Our domain mapping in this study provided evidence to confirm the interaction between the VDR and IKK␤ at the C terminus of both proteins, but more detailed mapping is needed to further narrow down the interacting domains in each molecule. Given the wide spectrum of NF-B activities that affect numerous biological processes, the inhibitory mechanism of vitamin D on NF-B reported here could have broader implications than we now recognize, which warrants more investigation in the future. Finally, because the VDR interacts physically with different cell signaling proteins described above, the questions whether these intracellular interactions are mutually competitive in biological regulations and whether different interactions have different physiological or pathological implications also warrant further studies.