A vitamin D receptor-Ser/Thr phosphatase-p70 S6 kinase complex and modulation of its enzymatic activities by the ligand.

We provide evidence of a cross-talk between nuclear receptor and Ser/Thr protein phosphatases and show that vitamin D receptor (VDR) interacts with the catalytic subunit of protein phosphatases, PP1c and PP2Ac, and induces their enzymatic activity in a ligand-dependent manner. PP1c specifically interacts with VDR but not retinoic acid receptor alpha and retinoid X receptor alpha in yeast. Although VDR-PP1c and VDR-PP2Ac interaction is ligand-independent in vivo, 1alpha,25-dihydroxy-vitamin D(3) induces VDR-associated phosphatase activity. Further, VDR modulation of PP1c/PP2Ac activity results in a rapid and specific dephosphorylation and inactivation of their substrate, p70 S6 kinase (p70(S6k)). Finally, we demonstrate that the endogenous VDR, PP1c or PP2Ac, and p70(S6k) are present in a ternary complex in vivo, and the interaction of p70(S6k) with the VDR-PP complex is modulated by the phosphorylation state of the kinase. Since p70(S6k) is essential for G(1)-S transition, our results provide a molecular basis of 1alpha,25-dihydroxyvitamin D(3)-induced G(1) block in colon cancer cells.

Vitamin D receptor (VDR), 1 a sequence-specific ligand-dependent transcription factor belonging to the family of nuclear receptors, mediates biological actions of 1␣,25-dihydroxy-vitamin D 3 (1,25(OH) 2 D 3 ). VDR heterodimerizes with retinoid X receptor (RXR), and at the molecular level VDR-RXR heterodimers induce gene expression via interaction with vitamin D response elements present in the promoter regions of responsive genes (1). This mode of action is known as the "genomic action" of VDR. However 1,25-(OH) 2 D 3 also induces gene ex-pression by a mechanism distinct from its classical mode of action. For example, 1,25-(OH) 2 D 3 -induced expression of monocytic differentiation markers CD14 and CD11b in THP-1 cells requires phosphatidylinositol 3-kinase (PI 3-kinase) via liganddependent interaction of VDR with the regulatory (p85) subunit of PI 3-kinase (2). Similarly estrogen receptor interacts with the p85 regulatory subunit of the PI 3-kinase where estrogen receptor-PI 3-kinase interaction leads to the activation of protein kinase B/AKT and endothelial nitric-oxide synthase (3). It thus appears that cross-talk between nuclear receptors and other signal transduction pathways can lead to either induction of gene expression in a nuclear receptor-responsive element-independent manner or to extranuclear/non-genomic induction of enzymatic activities that are physiologically important, for example, in explaining the vasoprotective effects of estrogen (3). Further, nuclear receptor ligands (dexamethasone, triiodothyronine, and retinoic acid) also induce a rapid dephosphorylation of c-Jun N-terminal kinase independently of their transcriptional activation (4). Therefore, nuclear receptors appear to have a functional role both outside and inside the nucleus. Ser/Thr phosphatases are implicated in the regulation of a wide variety of cellular functions, namely metabolism, transcription, translation, development, cell growth, and differentiation (5). There are two major and structurally related families of Ser/Thr phosphatases, termed PP1 and PP2A. PP1c and PP2Ac are the catalytic subunits of PP1 and PP2A, respectively, that associate with various regulatory and target subunits, thereby generating distinct holoenzymes with unique localizations, specificities, and cellular functions (6).
In this report, we provide evidence of a cross-talk between VDR and Ser/Thr phosphatases and show that VDR interacts with PP1c in a ligand-dependent manner in yeast. Although the association of VDR and PP1c/PP2Ac is ligand-independent in vitro and in vivo in mammalian cells, 1,25(OH) 2 D 3 induces the Ser/Thr phosphatase enzymatic activity of these phosphatases. This induction leads to a rapid and specific dephosphorylation at the Thr-389 residue of p70 S6 kinase (p70 S6k ), resulting in inactivation of this kinase. By co-immunoprecipitation, we also demonstrate the presence of a ternary complex containing VDR-PP-p70 S6k , where PP represents PP1c or PP2Ac. Using rapamycin, which specifically prevents Thr-389 phosphorylation, we show that the Thr-389-dephosphorylated p70 S6k does not efficiently interact with the VDR-PP complex, suggesting that p70 S6k phosphorylation state modulates its interaction with the VDR-PP complex. Since p70 S6k is critical for G 1 -to-S phase transition, our observations provide a molecular basis of VDR ligand-induced G 1 block in colon/epithelial cancer cells.

EXPERIMENTAL PROCEDURES
Yeast Two-hybrid Screening-The bait plasmid was generated by introducing a PCR-amplified VDR-LBD fragment (amino acids 89 -427) in-frame with Gal4 DNA binding domain into EcoRI-BamHI sites of the plasmid pGBKT7 (CLONTECH, Palo Alto, CA). Transformation of yeast was carried out by using the Yeastmaker transformation kit (CLONTECH). A colony from pGBKT7-VDR-transformed yeast strain PJ69-2A was grown for 18 h in 50 ml of SD/Trp Ϫ medium at 30°C with shaking at 270 rpm. After a growth of 0.8 at A 600 , cells were pelleted (1000 ϫ g, 5 min) and resuspended in 5 ml of the medium to give a final concentration of more than 1 ϫ 10 9 cells/ml. The entire pGBKT7-VDRtransformed PJ69-2A culture was combined with 1 ml of Y187 cells pretransformed with a mouse embryo cDNA library in pGADT7 expression vector (CLONTECH) in a 2-liter sterile flask. YPDA/kan (50 ml) * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Microcystin Affinity and Phosphatase Activity-T47D and Caco-2 cells were passaged in phenol red-free, high glucose Dulbecco's modified Eagle's medium supplemented with 10% charcoal-stripped fetal calf serum. For microcystin chromatography, Caco-2 or T47D cells were treated with 1,25(OH) 2 D 3 (10 Ϫ7 M) for 30 min, washed with cold PBS, and scraped in MC buffer (20 mM Tris-HCl, pH 7.5, 0.5 mM EGTA, 2 mM MgCl 2 , 2 mM DTT, 10% glycerol, 1 mM phenylmethylsulfonyl fluoride, and 1ϫ mammalian mixture of protease inhibitors (Sigma)). Cells were centrifuged (1000 rpm) at 4°C for 5 min, resuspended in MC buffer (3 ml), and subjected to freeze-thaw. Cell extracts were passed through a 21-gauge needle, and debris was removed by centrifugation (4000 rpm) for 10 min. Protein concentration was measured using BCA reagent (Pierce). The control sample was preincubated with 5 nM free microcystin for 1 h at 4°C. Cell extract (1 mg of protein) was mixed with 30 l of microcystin-Sepharose beads (50% slurry) (Upstate Biotechnology, Lake Placid, NY) pre-equilibrated with MC buffer and incubated overnight with shaking at 4°C. Beads were washed with cold MC buffer containing 150 mM NaCl, resuspended in 10 l of 2ϫ protein loading dye, and electrophoresed on a 10% SDS-polyacrylamide gel together with aliquots from the unbound fraction (20 g each). 1,25(OH) 2 D 3 was present in all washes and buffer for the treated samples. For phosphatase activity, cell extracts (600 g of protein) were incubated overnight at 4°C with agarose-conjugated rabbit anti-VDR antibody (Santa Cruz Biotechnology, Santa Cruz, CA), and beads were washed four times with washing buffer (25 mM Tris-HCl, pH 7.0, 0.1 mM EDTA, 5 mM DTT, 0.01% Brij35) supplemented with 150 mM NaCl and 1,25(OH) 2 D 3 where needed. Beads were resuspended in washing buffer, and aliquots (40 l) were used for phosphatase activity assay or Western blotting. Ser/Thr phosphatase activity was measured using the Ser/Thr phosphatase assay kit (New England Biolabs, Beverly, MA).
Analysis of p70 S6k Phosphorylation-Cells were treated with 1,25(OH) 2 D 3 (10 Ϫ8 M) or vehicle for 5, 15, or 30 min, rinsed once with cold PBS containing phosphatase inhibitor mixture I (Sigma), and scraped off in the same buffer. Cells were centrifuged at 450 ϫ g for 5 min and resuspended in 5 volumes of hypotonic buffer (10 mM HEPES, pH 7.9, 10 mM KCl, 1 mM DTT with protease and phosphatase inhibitors) for 5 min. Igepal-640 was added (0.6% final concentration), and cells were vortexed (10 s) and centrifuged (10,000 ϫ g, 30 s). The supernatant was kept as the cytoplasmic fraction. Protein extracts (30 g) were subjected to 10% SDS-PAGE and Western blotting with various p70 S6k antibodies (Cell Signaling, Beverly, MA). Immunoprecipitation of VDR-or PP1c-bound p70 S6k was performed on 1 mg of cytoplasmic extract as described for the phosphatase assay.

RESULTS AND DISCUSSION
To gain a better understanding of the mechanism of action of 1,25(OH) 2 D 3 , a yeast two-hybrid system was used to identify proteins that interacted with VDR in the presence of the ligand. The sequence of VDR harboring the D-E regions of the receptor (amino acids 89 -427) was used to prepare the bait construct (pGBKT7-VDR), and cDNAs encoding VDR-interacting proteins were isolated from a 17.5-day mouse embryo library. Upon characterization, most of the cDNA clones, whose protein products interacted with VDR in a ligand-dependent manner, were found to be mouse RXR␣, -␤, and -␥ (Fig. 1A). Among other VDR-interacting protein cDNAs, a clone, H1, was isolated that contained a 2.2-kb insert corresponding to the cDNA sequence of full-length mouse PP1c. Neither the bait nor the fish vector was active when expressed alone in yeast cells. However, ␤-galactosidase activity in yeast was observed only when pGBKT7-VDR was expressed with pGADT7-PP1c in the presence but not in the absence of 1,25(OH) 2 D 3 (Fig. 1, A and C). Ligand-dependent interaction of VDR-LBD with PP1c was also found to be dose-dependent (Fig. 1B), and the interaction was also observed with full-length VDR but not with RAR␣ and RXR␣ (Fig. 1, C and D).
To further explore VDR-PP1c interaction, a GST-VDR-LBD fusion protein (containing amino acids 90 -427) was purified from pGEX3X-VDR-LBD-transformed bacterial cells using glutathione-Sepharose beads. In vitro translated [ 35 S]methioninelabeled PP1c was retained by the glutathione bead-bound GST-VDR-LBD. In contrast to the yeast two-hybrid results, in vitro interaction between GST-VDR-LBD and PP1c was found to be ligand-independent ( Fig. 2A). No interaction was observed between PP1c and glutathione-bound GST protein. To identify the VDR interaction domain of PP1c, deletion mutants of PP1c were generated and assayed for interaction in GST pull-down experiments. Deletion of up to 177 amino acids in the N-terminal portion of PP1c did not prevent binding to VDR-LBD ( Fig.  2A). Interestingly the N-terminal part contains a number of amino acids essential to the catalytic function of PP1c, suggesting that the integrity of the catalytic site is not required for interaction with VDR. Deletion of 123 amino acids from the C terminus of PP1c significantly reduced but did not abolish its binding to VDR (Fig. 2B), suggesting that although the C terminus of PP1c provides the major interacting region, similar

VDR-Ser/Thr Protein Phosphatase Interaction 24848
to its interaction with other regulatory proteins, multiple regions are involved in VDR-PP1c interaction (Fig. 2B).
To verify that endogenous PP1c was physically associated with endogenous VDR, we purified PP1c-associated proteins using microcystin-Sepharose (an affinity resin for PP1c and PP2Ac) beads (8). Cell extracts obtained from control or 1,25(OH) 2 D 3 (10 Ϫ7 M)-treated Caco-2 cells were incubated overnight at 4°C with microcystin-Sepharose beads. Following extensive washing, microcystin bead-bound proteins were separated by SDS-PAGE and analyzed by Western blotting using anti-VDR antibodies. A VDR-PP1c interaction was observed in Caco-2 cells that was not dependent on the presence of the VDR ligand (Fig. 2C, a). Preincubation of Caco-2 extract with 0.5 nmol of free microcystin resulted in a loss of detectable VDR and PP1c, suggesting that VDR-PP1c association was specific (Fig. 2C, b). Further, microcystin chromatography of control and 1,25(OH) 2 D 3 (10 Ϫ7 M)-treated Caco-2 and T47D cell extracts followed by Western blotting with anti-VDR antibodies showed VDR-PP1c interaction in Caco-2 but not T47D cells (Fig. 2D), suggesting that VDR-PP association is cell contextdependent. Moreover, RXR␣ was not detected in the VDR-PP complex retained on microcystin-Sepharose beads (Fig. 2D).
The presence of ligand-independent interaction between VDR and PP1c in GST and microcystin systems but ligand dependence in yeast prompted us to examine the possibility that 1,25(OH) 2 D 3 induced the enzymatic activity of VDR-bound Ser/Thr protein phosphatases. A ligand-mediated increase in phosphatase activity could explain the increase in VDR gene expression in yeast. Therefore, we measured VDR-associated Ser/Thr phosphatase activity. Anti-VDR antibody could pull down Ser/Thr protein phosphatase enzymatic activity from Caco-2 cell extracts (Fig. 3A). Further, 30-min treatment of cells with 1,25(OH) 2 D 3 (10 Ϫ8 M) resulted in an enhancement of VDR-associated phosphatase activity (Fig. 3A). The specificity of VDR-associated Ser/Thr phosphatase activity was demonstrated by incubation of the VDR-IP reaction with microcystin (6 nM) that completely abolished the VDR-associated enzymatic activity (Fig. 3A). Microcystin, a cyanobacterial toxin, is a potent inhibitor of both PP2Ac and PP1c (9). Aliquots from the VDR-IP used to measure phosphatase activity were subjected to Western blot analysis to monitor differences in VDR-associated Ser/Thr phosphatases between control and 1,25(OH) 2 D 3treated samples. In immunoprecipitations, VDR antibody could pull down not only PP1c but also PP2Ac, and identical levels of PP1c and PP2Ac proteins were detected in control and treated cell extracts (Fig. 3B). These results demonstrate that the increased Ser/Thr phosphatase activity associated with VDR-IP derived from 1,25(OH) 2 D 3 -treated cells did not result from increased recruitment of Ser/Thr phosphatases to the receptor. In contrast, as expected, 1,25(OH) 2 D 3 treatment of Caco-2 cells resulted in increased immunoprecipitation of RXR␣ by anti-VDR antibodies (Fig. 3B). Taken together these Incubation of immune complexes with microcystin (6 nM) abrogated both ligand-dependent (black bar) and ligand-independent (striped bar) VDR-associated Ser/Thr phosphatase activity. The amount of Ser/Thr phosphatase activity present in 5 g of control and treated Caco-2 cell extract is also shown. Data are representative of three independent experiments performed in triplicate. B, ligand-independent association of VDR with PP1c and PP2Ac. Immunoprecipitations were performed from 1,25(OH) 2 D 3 -treated (ϩ) or untreated (Ϫ) Caco-2 cell extracts with VDR antibodies. After extensive washing, the immune complexes were analyzed by SDS-PAGE and immunoblotting with PP1c, PP2Ac, RXR␣, and p70 S6k antibodies. The relative intensity of the bands in arbitrary units is also shown. C, ligand-dependent Thr-389 dephosphorylation of cytoplasmic p70 S6k . Immunoblot analysis were performed on 1,25(OH) 2 D 3 -treated (5, 15, or 30 min) or untreated (Ϫ) Caco-2 cell cytoplasmic extracts using anti-p70 S6k Thr-389, anti-p70 S6k Thr-421/ Ser-424, anti-p70 S6k Ser-411, anti-p70 S6k , and anti-mTORSer-2448 antibodies. D, phospho-Thr-389-dependent interaction of VDR-PP complex and p70 S6k Thr-389. 1,25(OH) 2 D 3 -treated (D3), rapamycin-treated (rap), or untreated (Ϫ) Caco-2 cell cytoplasmic extracts were immunoblotted and analyzed using anti-p70 S6k Thr-389 and anti-p70 S6k antibodies. Immunoprecipitations were performed from the same extracts with anti-VDR or anti-PP1c antibodies. The immune complexes were analyzed by SDS-PAGE and immunoblotting with anti-p70 S6k Thr-389 or anti-p70 S6k antibodies. Data are representative of three independent experiments. E, a model of VDR-PP1c/PP2Ac-p70 S6k complex and the modulation of its activity by the VDR ligand are presented. IB, immunoblot.

VDR-Ser/Thr Protein Phosphatase Interaction 24849
data suggest that 1,25(OH) 2 D 3 increases the enzymatic activity of VDR-associated Ser/Thr phosphatases in Caco-2 cells. A number of cellular mechanisms dependent on Ser/Thr phosphorylation and relevant to 1,25(OH) 2 D 3 action could be affected by increased VDR-bound phosphatase activity. The effect of 1,25(OH) 2 D 3 on G 1 -S phase transition is of particular interest since Ser/Thr phosphatases complex with a number of key regulators of cell cycle progression. PP2Ac is known to dephosphorylate and inactivate p70 S6k (10 -12), a Ser/Thr kinase that induces translation of mRNAs containing 5Ј-terminal oligopyrimidine sequences, and is essential for G 1 -S progression (12). PP2Ac is also shown to form a complex with p70 S6k , thus indicating the presence of a Ser/Thr phosphatase-kinase signaling module inside the cell (10,11,13). Further, phosphorylations at Thr-229, Ser-371, and Thr-389 are essential for activation of p70 S6k (12). Since 1,25(OH) 2 D 3 induces G 1 arrest in epithelial cells (14), we looked for changes at the phosphorylation state of p70 S6k after treatment of Caco-2 cells with the VDR ligand. Caco-2 cytoplasmic extracts probed with phospho-Thr-389-specific p70 S6k antibodies showed dephosphorylation of p70 S6k after 5, 15, and 30-min treatments of cells with 1,25(OH) 2 D 3 (Fig. 3C). In contrast, treatment with 1,25(OH) 2 D 3 did not change either the protein level of p70 S6k or the level of phosphorylated p70 S6k at Thr-421/Ser-424 or Ser-411 positions (Fig. 3C). The rapamycin-sensitive FRAP/mTOR kinase is implicated in phosphorylating p70 S6k at Thr-389 upon serum stimulation, raising the possibility that 1,25(OH) 2 D 3 could act by decreasing FRAP/mTOR activity. Activity of this kinase correlates to its phosphorylation at Ser-2448 (15). Analysis of Caco-2 cell extracts did not reveal any decrease in FRAP/mTOR phosphorylation level using a phospho-Ser-2448specific antibody (Fig. 3C), suggesting that decreased Thr-389 phosphorylation on p70 S6k was not secondary to reduced FRAP/ mTOR activity.
Although PP2Ac-p70 S6k interaction is well documented (10 -12), PP1c-p70 S6k interaction has not been reported. As shown in Fig. 3D, p70 S6k could be immunoprecipitated using anti-PP1c antibody, suggesting the presence of a PP1c-p70 S6k signaling module. We next tested the possibility that VDR-PP1c or -PP2Ac complexes also contain p70 S6k , thereby providing an opportunity for PP2Ac or PP1c to dephosphorylate and thus inactivate p70 S6k . Anti-VDR antibody could pull down Thr-389phosphorylated p70 S6k in the absence but not in the presence of the ligand (Fig. 3D). Further, p70 S6k did not directly interact with VDR since in vitro translated p70 S6k was not retained by GST-VDR-LBD bound to glutathione beads (data not shown). Dissociation of p70 S6k from VDR under conditions where Thr-389 was mainly dephosphorylated (via activation of associated PP1c or PP2Ac in the complex) suggested that the interaction could be dependent upon the phosphorylation state of p70 S6k . To support this hypothesis, Caco-2 cells were treated with rapamycin (20 nM, 45 min), and VDR immunoprecipitates were analyzed by using phospho-Thr-389-specific antibody. Since rapamycin preferentially inhibits FRAP/mTOR, the upstream kinase of p70 S6k (16), as expected rapamycin treatment of Caco-2 cells inhibited phosphorylation of p70 S6k at Thr-389 (Fig. 3D). This inhibition resulted in a significant dissociation of p70 S6k from VDR, supporting the idea that phospho-Thr-389 preferentially interacted with the VDR-PP complex in the absence of the ligand (Fig. 3D). These results suggest that VDR may act as a platform to compartmentalize VDR-PP1c-p70 S6k and VDR-PP2Ac-p70 S6k complexes to the same intracellular location or bring them in close proximity so that ligand stimulation could result in induction of Ser/Thr phosphatase activity, which in turn may result in dephosphorylation and inactivation of p70 S6k . Therefore, in addition to vitamin D-dependent induction of p21 gene expression (17), VDR ligand-mediated dephosphorylation and inactivation of p70 S6k may also play a role in 1,25(OH) 2 D 3 -induced G 1 arrest of colon cancer cells. Alternatively the VDR-PP1c/PP2Ac-p70 S6k pathway may contribute to ligand-mediated inhibition of proliferation of epithelial cancer cells that do not show p21 up-regulation after 1,25(OH) 2 D 3 treatment (18).
Epidemiological studies show that vitamin D confers significant protection against colorectal cancer (19) and increased VDR expression is observed in cancerous lesions than normal colon (20,21). Therefore, as shown in our model (Fig. 3E), it is tempting to speculate that in colon cancer cells, at least a part of the cytoplasmic complement of VDR is compartmentalized as VDR-PP1c/PP2Ac-p70 S6k signaling module. Increased VDR levels coupled with low 1,25(OH) 2 D 3 may inhibit the activity of PP1c/PP2Ac in this complex, thus rendering the associated p70 S6k active by virtue of its Thr-389 phosphorylation. This is evident from Fig. 3A since VDR-associated Ser/Thr phosphatase activity is drastically inhibited in the absence of the ligand. VDR ligand, upon binding to LBD, appears to impart a conformational change in PP1c/PP2Ac proteins, thus rendering them active in terms of their enzymatic activity (Fig. 3A), which is evident by decreased Thr-389 phosphorylation, inactivation, and dissociation of VDR-associated p70 S6k (Fig. 3, C  and D). The final result of this sequence of events is cell cycle arrest at the G 1 -S transition state. Accordingly, VDR ligands inhibited the proliferation of colon cancer cells in vitro and reduced tumorigenesis in vivo (22,23). Finally, since downregulation of PP2Ac activity is one of the key steps in cellular transformation (24,25), our observation that 1,25(OH) 2 D 3 regulates Ser/Thr phosphatase activity might contribute to elucidating the role of VDR ligands in controlling growth and differentiation of normal and cancer cells.