| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Reproductive Biology and Medicine Branch (T.K., T.I., N.K., S.R., G.P.C.), National Institute of Child Health and Human Development and Laboratory of Neurochemistry (N.D.A., S.K., Y.-L.Z., H.C.P.), National Institute of Neurological Disorders and Stroke, Bethesda, Maryland 20892; Microarray Facility (Y.W., A.P., E.K.), Advanced Technology Center, National Cancer Institute, Gaithersburg, Maryland 20877; Department of Microbiology and Urology (M.J.G.), New York University Cancer Institute, New York University School of Medicine, New York, New York 10016; and First Department of Pediatrics (G.P.C.), Athens University Medical School, 11527 Athens, Greece
Address all correspondence and requests for reprints to: Tomoshige Kino, M.D., Ph.D., Reproductive Biology and Medicine Branch, National Institute of Child Health and Human Development, National Institutes of Health, Building 10, Clinical Research Center, Room 1-3140, 10 Center Drive MSC 1109, Bethesda, Maryland 20892-1109. E-mail: kinot{at}mail.nih.gov.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
Because glucocorticoids are essential for survival in humans and play numerous important roles in the proper functioning of the CNS and other tissues, several distinct signaling pathways regulate the transcriptional activity of the GR at many levels (2). A major such mechanism is via posttranslational modifications of the GR, which result in changes in its transcriptional activity on responsive genes (6). Indeed, several kinases, such as the cell cycle-related and mitogen-activated kinases, located downstream of cell surface receptors for growth factors and cytokines, phosphorylate specific serine or threonine residues of the GR and change its transcriptional activity (7, 8, 9, 10, 11, 12, 13).
Cyclin-dependent kinase 5 (CDK5) is a member of the cyclin-dependent kinase family of serine/threonine kinases, most of which are key regulators of the cell cycle (14, 15, 16). In contrast to other such CDKs, CDK5, although associated with cyclin-like molecules, has no activity during mitosis but is essential for neuronal morphogenesis and survival (16, 17). CDK5 is expressed ubiquitously in many tissues; however, its activity is restricted primarily to the nervous system due to neuron-specific expression of its activator molecules p35 and p39 (18, 19). In addition to these physiological roles of CDK5, recent evidence suggests that aberrant CDK5 activation caused by proteolytic conversion of p35 to p25 may play a role in the pathogenesis of neurodegenerative disorders, such as Alzheimer’s disease and amyotrophic lateral sclerosis (20, 21, 22, 23, 24, 25). In these conditions, calpain-directed proteolysis of p35 deprives the membrane-associated p35 of a N-terminal myristoylated membrane tether, releasing p25 into the cytoplasm (26, 27).
We searched for potential regulators of the transcriptional activity of the GR in the CNS by performing cyto-trap yeast two-hybrid screening assays using the human GR
ligand-binding domain (LBD) as bait. We found that p35 specifically interacted with this portion of the GR and that GR-associated CDK5/p35 or p25 modulated GR-induced transcriptional activity by phosphorylating multiple serine residues at the N-terminal domain (NTD) of the receptor. Oligonucleotide microarray analyses in rat cortical neuronal cells revealed that CDK5 had gene-specific effects on GR transcriptional activity, positively or negatively regulating GR-induced transactivation and transrepression. Thus, GR appears to be a CDK5 substrate, possibly mediating some of the biological actions of this kinase in the brain.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
). Full-length GR and its LBD interacted with the full-length (1–307) and the C-terminal part (190–307) of p35 in a dexamethasone-dependent fashion, whereas the NTD and DBD of the GR did not. In contrast, the N-terminal part of p35, including its portion proteolytically removed from p35 (amino acids 1–90), did not interact with any of the GR fragments. Thus, the C-terminal part (amino acids 190–307) of p35 or p35 interacts with the GR LBD in a ligand-dependent fashion in this yeast two-hybrid assay.
|
, B and C). The protein complex was precipitated with anti-GR or control antibody, and CDK5 and/or p35 associated with the GR were visualized in Western blots using specific antibodies against them. Both CDK5 and p35 were coprecipitated with anti-GR in a dexamethasone-dependent fashion, but not with the control antibody (Fig. 1B, top two gels). Expression levels of p35, CDK5, and GR were similar throughout the experiment (Fig. 1B, bottom three gels). We further examined precipitation of GR with CDK5/p35 in these cells. As expected, GR was coprecipitated with anti-CDK5 antibody in a dexamethasone-dependent fashion (Fig. 1C, top two gels). These results indicate that the CDK5 and p35 interact with the GR in a ligand-dependent fashion in vivo.
We further examined the interaction of the GR and p35, p25, and CDK5 in glutathione-S-transferase (GST) pull-down assays using GST-fused p35, p25, and CDK5, and in vitro translated GR (Fig. 1D). GR strongly interacted with GST-p35 and -p25 in a dexamethasone-dependent fashion, whereas it was not associated with the GST. GR was also weakly associated with GST-CDK5 in a ligand-dependent fashion (Fig. 1D). It is known that p35 functionally interacts with CDK5 through three portions that are included by its amino acids spanning from 138–291, where N-terminal (amino acids 138–154) and C-terminal (amino acids 287–291) ends are essential for the dominant negative activity of this p35 fragment on p35-induced CDK5 activation (28). Thus, we tested p35 (190–262), a fragment devoid of such essential portions for inhibiting CDK5 activity, in a GST pull-down assay with in vitro translated GR and CDK5. GR physically interacted with p35 (190–262) in a dexamethasone-dependent fashion, whereas CDK5 lost such interaction (Fig. 1E). Taken together, these results indicate that p35 and/or p25 physically interact with ligand-bound GR in vitro through their C-terminal portion enclosed by amino acids 190 and 262.
CDK5 Suppresses GR-Induced Transcriptional Activity in a Kinase-Dependent Fashion
We examined the functional interactions between CDK5, p35, p25, and GR by expressing these molecules in GR-deficient HCT116 cells and measuring GR-induced transcriptional activity (Fig. 2). Independent expression of CDK5 or p25 weakly suppressed GR-induced transcriptional activity on the glucocorticoid-responsive mouse mammary tumor virus (MMTV) promoter in a dexamethasone-dependent fashion in HCT116 cells (Fig. 2A). CDK5 D144N, a kinase-defective CDK5 mutant with transdominant negative activity on wild-type CDK5-mediated kinase activity, weakly increased GR-induced transcriptional activity in a dexamethasone-dependent fashion. Coexpression of the wild-type CDK5 with p25 or p35 strongly suppressed GR transcriptional activity, with p25 demonstrating a much stronger effect than that of p35. Coexpression of CDK5 D144N with p25 or p35, on the other hand, did not induce a suppressive effect on GR transcriptional activity. Indeed, coexpression of this CDK5 mutant with p25 attenuated the suppressive effect of the latter on GR transactivation, whereas its coexpression with p35 further enhanced GR-induced transcriptional activity, similarly to the expression of this mutant by itself. These results indicate that CDK5 suppresses GR-induced transcriptional activity in a kinase activity-dependent fashion. Because individual expression of CDK5 or p25 suppressed GR-induced transcriptional activity, HCT116 cells appear to express endogenous molecules similar to CDK5 and p35/p25, which respectively interact with exogenously expressed p25 or CDK5. These observations are further strengthened by the fact that CDK5 D144N enhanced the transcriptional activity of the GR, possibly antagonizing such endogenous molecules. Because p25 suppressed GR-induced transcriptional activity more strongly than p35, a free cytoplasmic form of p25, not associated with membranes via its N-terminal portion, may be the functional fraction of these molecules that suppresses GR transcriptional activity.
|
). We used p25 instead of p35 because the former is a more active form for suppressing GR transcriptional activity. CDK5/p25 shifted the dexamethasone dose-response curve downward in a dose-dependent manner, indicating that the complex acted at the post-ligand-binding phase of GR-induced transcriptional activity consistent with the other results.
CDK5 Suppresses GR-Induced Transcriptional Activity through Physical Interaction
We examined the importance of physical association between CDK5/p35 and GR on CDK5-induced suppression of GR transcriptional activity in HCT116 cells (Fig. 2C). CDK5 and p35 were efficiently coprecipitated with GR in a dexamethasone-dependent fashion (Fig. 2C, top two gels). We found that overexpression of His-p35 (190–262), a p35 fragment that interacts with the GR but not with CDK5 in a GST pull-down assay, abolished their association in HCT116 cells (Fig. 2C, top two gels). Expression levels of p35, CDK5, and GR were similar throughout the experiment, whereas His-p35 (190–262) expression was detected only in transfected cells (Fig. 2C, bottom four gels). In a reporter assay in HCT116 cells, overexpression of His-p35 (190–262) attenuated CDK5/p35-mediated suppression of GR transcriptional activity on the MMTV promoter in a dose-dependent manner (Fig. 2D). These results indicate that CDK5/35 suppresses GR-induced transcriptional activity through physical interaction with the receptor.
CDK5 Suppresses GR-Induced Transcriptional Activity of the Serum/Glucocorticoid-Inducible Kinase Promoter in HCT116 and COS7 Cells
Because the MMTV is of exogenous, viral origin, we examined the effect of CDK5/p35 on the endogenous glucocorticoid-responsive serum/glucocorticoid-inducible kinase (SGK) promoter in HCT116 and COS7 cells (Fig. 2, E and F). It is known that SGK plays important biological roles in neuronal tissues, such as special memory formation and neuronal dendritic growth/branching (29, 30). Glucocorticoids stimulate the human and rat SGK promoters through tandem GREs located approximately 1 kb upstream of their transcription start sites (31, 32). Dexamethasone stimulated the transcriptional activity of the human SGK promoter by about 3-fold, and CDK5/p25 expression strongly suppressed GR-induced transcriptional activity in HCT116 and COS7 cells, similarly to its effect on the MMTV promoter (Fig. 2, E and F). These results suggest that CDK5/p25 suppresses the transcriptional activity of endogenous glucocorticoid-responsive promoters, which are active in neuronal tissues.
Both the NTD and LBD, But Not the DBD, of the GR Are Necessary for CDK5 to Suppress GR-Induced Transcriptional Activity
Because CDK5/p35 and CDK5/p25 interact with GR at the LBD and suppress GR-induced transcriptional activity, we tested the effects of CDK5 on GR chimeras, which have the GR NTD and/or LBD fused with the GAL4 DBD, to examine the contribution of these domains to CDK5-induced modulation of GR transactivation (Fig. 3A). CDK5/p25 strongly suppressed the transcriptional activity of the G-gal-G, which had the GAL4 DBD instead of the GR DBD, but contained the GR NTD and LBD, on a GAL4-responsive element-driven promoter in HCT116 cells. CDK5, in contrast, did not affect the transcriptional activity of the GR NTD or LBD fused with the GAL4 DBD. These results indicate that both the NTD and LBD of the GR are required for CDK5/p25 to suppress GR-induced transcriptional activity. CDK5/p25 binds GR at the GR LBD, whereas NTD appears to be an effector domain for CDK5/p35. The DBD of the GR is not necessary for CDK5 to suppress the GR transcriptional activity of ligand-activated GR.
|
, top two gels, lanes 5 and 6). However, coexpression of the wild-type CDK5 and p25 strongly phosphorylated these serines in a dexamethasone-dependent fashion (Fig. 3B, top two gels, lanes 3 and 4), whereas CDK5 D144N, a mutant that is defective in its kinase activity, failed to do so (Fig. 3B, top two gels, lanes 1 and 2). These results indicate that CDK5 phosphorylates GR at serines 203 and 211 in vivo.
We examined the transcriptional activity of GR mutants that have alanines at amino acid positions 203 and/or 211, along with the wild-type GR, in reporter assays using the glucocorticoid-responsive MMTV promoter in HCT116 cells (Fig. 3C). We also examined GR mutants T8A, S45A, S226A, S267A, S288A and S395A, the mutated threonine and serine residues of which are in incomplete consensus sequences, because CDK5 also phosphorylates such residues in some of its substrate proteins (37). In preliminary experiments, CDK5/p25 modulated dexamethasone-stimulated transcriptional activity of GR mutants S45A, S203A, S211A, S226A, and S395A, whereas it did not influence that of GR T8A, S267A, and S288A, on the MMTV promoter (data not shown). Thus, we further examined their effects on the GR mutants included in Fig. 3C. CDK5/p25 strongly suppressed wild-type GR-induced transcriptional activity on the MMTV promoter in a dexamethasone-dependent fashion (12.6% of that in the absence of CDK5/p25). GR S203A had more than 2 times stronger transcriptional activity than the wild-type GR, whereas CDK5/p25 suppressed the transcriptional activity of this mutant less (34.6% of the activity observed in the absence of CDK5/p25) than that of the wild-type GR. GR S211A also demonstrated stronger transcriptional activity on the MMTV promoter and weaker responsiveness to CDK5/p25; however, the impact of the S211 mutation was weaker than that of S203. GR S226A, in which serine is not within a perfect consensus sequence, showed similar activity to that of GR S211A, as well as a similar response to CDK5/p25. GR S45A and GR S395A had transcriptional activities similar to that of the wild-type receptor, whereas the suppressive effect of CDK5/p25 on these mutant receptors was blunted (17.2 and 17.4% of the activity observed that in the absence of CDK5/p25, respectively). The GR S203/211A demonstrated stronger transcriptional activity than the wild-type GR and significantly blunted responsiveness to CDK5/p25 (63.0%). Triple replacement of serines at 203, 211, and 226 by alanines further increased the transcriptional activity of the mutant receptor and almost abolished the suppressive effect of CDK5/p25 (87.0% of that observed in the absence of CDK5/p25). Replacement of all serine residues (serines 45, 203, 211, 226, and 395) by alanines completely abolished the CDK5/p25-induced suppressive effect (108.3%). All these GR-related molecules were similarly expressed in HCT116 cells (Fig. 3D). Taken together, these results indicate that serines 203 and 211 of the human GR are major sites for mediating the suppressive effect of CDK5 on GR-induced transcriptional activity, whereas serine 226 further contributes to CDK5-induced suppression. Serines 45 and 395 minimally contribute to the suppressive effect of CDK5.
Based on the results obtained in the reporter assay employing GR serine mutants, we further examined CDK5-induced phosphorylation of GST-fused GR fragments in in vitro kinase assays (Fig. 3E). CDK5 strongly phosphorylated wild-type GST-GR (1–490), which consisted of the NTD and DBD of the GR and GST (Fig. 3E, top panel). As expected, CDK5-induced phosphorylation was significantly attenuated in GST-GR (1–490) S203A and S211A that had a single replacement of the indicated serine residues located in the consensus CDK5 phosphorylation sequence by an alanine. When both of these serines were mutated, CDK5-dependent phosphorylation was strongly reduced. CDK5-induced phosphorylation was also strongly diminished in GST-GR (1–490) S226 and almost disappeared in GST-GR (1–490) S203/211/226A. The phosphorylation was minimally blunted in GST-GR (1–490) S45A and S395A, whereas it was completely abolished in GST-GR (1–490) S45/203/211/226/395A. These GR-related GST-fused proteins were equally expressed (Fig. 3E, middle panel) and preserved their ability to bind Sma- and Mad-related protein (Smad)6, a transcription factor that physically interacts with the GR at amino acids 263–419 and suppresses GR-induced transcriptional activity (38) (Fig. 3E, bottom panel). CDK5 did not phosphorylate the negative control GST at all, whereas it greatly phosphorylated the positive control histone. Thus, serines 203 and 211, along with serine 226 of the human GR, are major phosphorylation sites of CDK5 in vitro, whereas serines 45 and 395 are also minimally phosphorylated by CDK5.
CDK5 Suppresses GR-Induced Attraction of Cofactors to the MMTV Promoter GREs
To further study the mechanism of CDK5-induced suppression of GR transcriptional activity, we examined accumulation of cofactors p300 and sucrose non-fermenting 2 (SNF2), which are, respectively, components of the histone acetyltransferase coactivator and the SWI/SNF complexes (39, 40), on the chromatin-integrated MMTV promoter and the endogenous SGK promoter in chromatin immunoprecipitation (ChIP) assays in HCT116, COS7, and rat cortical neuronal cells (Fig. 4). These protein complexes associate with GRE-bound GR through its AF-1 and contribute to GR-induced transcriptional activity (4). HCT116/MMTV and COS7/MMTV cells were transfected with CDK5- and p25-expressing plasmids together with a GR-expressing plasmid. p300- and SNF2-expressing plasmids were also included when COS7/MMTV cells were tested, because these cells do not express immunoreactive p300 and SNF2 with the antibodies employed. Rat cortical neuronal cells were exposed to roscovitine, a specific CDK5 inhibitor, instead of being transfected with CDK5- and p25-expressing plasmids. After these transfected/treated cells were exposed to dexamethasone for 5 h, ChIP assays were performed by using their respective antibodies, and associated MMTV and SGK GREs were evaluated with the SYBR green real-time PCR using specific primer pairs. The transcriptional cofactor molecules p300 and SNF2 were attracted to MMTV GREs in response to dexamethasone in the absence of CDK5/p25, whereas their accumulation was greatly suppressed in the presence of the wild-type CDK5 and p25 on the MMTV promoter in HCT116 and COS7 cells (Fig. 4, A and B). Expression of kinase-defective CDK5 D144N with p25 did not influence dexamethasone-induced attraction of p300 and SNF2 to MMTV GREs. Dexamethasone treatment stimulated attraction of p300 and SNF2 to SGK GREs, and the CDK5 inhibitor roscovitine further potentiated their attraction in rat cortical neuronal cells (Fig. 4C). These results indicate that CDK5 suppresses the accumulation of cofactors p300 and SNF2 on ligand-activated GR, which is associated with the exogenous MMTV GREs, as well as the endogenous SGK GREs in a kinase activity-dependent fashion.
|
). In the brain, CDK5 is constitutively active (16). Among approximately 55% of total genes (
30,000 independent transcripts), the expression of which was successfully analyzed, roscovitine and dexamethasone treatment, respectively, regulated the expression of 289 and 596 independent genes, with 34 of these genes overlapping (2 SD as threshold) (data not shown). Roscovitine induced up to ±3-fold induction or suppression of mRNA expression, whereas dexamethasone caused over ±10 times alteration in some of the glucocorticoid-regulated genes (Fig. 5A). These results indicate that individual uses of roscovitine and dexamethasone exert distinct transcriptional activities in rat cortical neuronal cells, and that few genes regulated by dexamethasone are also influenced by roscovitine in the absence of dexamethasone. Cluster analysis, however, revealed that mRNA expression of dexamethasone-regulated genes was differentially regulated with roscovitine treatment in the presence of dexamethasone (Fig. 5B). Roscovitine up- (17%) or down- (83%) regulated dexamethasone effect on more than 90% genes the transcriptional activity of which was up-regulated by dexamethasone, indicating that CDK5 represses or facilitates GR-induced transactivation in a gene-specific fashion (Fig. 5B, U/U and U/D). In contrast, roscovitine attenuated dexamethasone effect on the majority of genes the mRNA expression of which was down-regulated by dexamethasone, suggesting that CDK5 primarily antagonizes GR-induced transrepression (Fig. 5B, D/U). We confirmed the results of the microarray analyses using the SYBR green real-time PCR and specific primer pairs (Fig. 6 and Table 1). Roscovitine enhanced dexamethasone-stimulated mRNA expression of the protein phosphatase 1 regulatory subunit 10 and the neuropeptide Y receptor Y1, whereas it repressed dexamethasone-stimulated mRNA expression of the inhibitor of DNA-binding 3 (Id3), RGD1309863, and torsin family 1 member b. Consistent with the results observed in reporter and ChIP assays, roscovitine enhanced dexamethasone-stimulated SGK mRNA expression. In contrast, roscovitine attenuated dexamethasone-induced suppression of mRNA expression of c-fos, nischarin, and the kinesin family member 1A. Dexamethasone did not influence roscovitine-stimulated or -suppressed mRNA expression of metallothionein 1a, heat shock protein 27, or the oligodendrocyte transcription factor 2, indicating that GR did not affect CDK5-mediated gene expression. Thus, CDK5 differentially regulates glucocorticoid-induced gene expression in a gene-specific fashion in rat cortical neuronal cells.
|
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
). The C-terminal part of the p35 enclosed in amino acids from 190–262 binds the GR LBD in a ligand-dependent fashion. GR LBD and NTD act as interactor and effector surfaces for CDK5/p35, respectively. Phosphorylation of the GR attenuated attraction of transcriptional cofactors on GR bound to GREs of the MMTV and SGK promoters. In contrast to the effect of CDK5 on the MMTV and SGK promoters in HCT116 and COS7 cells, roscovitine, an inhibitor of CDK5, differentially modulated the transcriptional activity of the GR on endogenous glucocorticoid-responsive genes in rat cortical neuronal cells, suggesting that the effect of CDK5 on GR-induced transcriptional activity is specific to gene promoter, and possibly, to tissue (Fig. 7B).
|
Glucocorticoids have diverse physiological and pathological activities on the regulation of neuronal functions (2, 46). For example, physiological levels of circulating glucocorticoids are essential for the maintenance of the proper CNS function (46, 47). Indeed, chronic lack of GR action impairs learning and memory functions by changing the consolidation process in genetically modified mice, whereas both acute and chronic elevation/reduction of glucocorticoid levels change arousal and/or attention, as well as cause emotional dysfunction and psychiatric problems in humans (48, 49, 50, 51). At the cellular level, glucocorticoids change many functions of neuronal cells, such as their electrical activity and synthesis/secretion of neurotransmitters and/or neuropeptides, and affect cell degeneration or death (2, 50, 51). Alteration of CDK5 activity in neuronal cells can potentially influence any of these glucocorticoid actions in the CNS. To examine whether the regulatory actions of CDK5 on GR transcriptional activity contribute to the development of neurodegenerative disorders is also particularly interesting, because glucocorticoids have strong activity on consolidation of memory and neuronal cell survival (2, 52). Proteolytically produced cytoplasmic p25, the excessive production of which has been associated with Alzheimer’s disease (23, 53), demonstrated a much stronger effect than p35 in regulating GR transcriptional activity. Thus, it is possible that CDK5 may cause pathological events by altering GR transcriptional activity through aberrant conversion of p35 to p25. Furthermore, alterations of CDK5 activity might explain region-specific regulation of glucocorticoid activity in the CNS. Indeed, glucocorticoids exert different effects in various regions of the brain, and some of their actions are mediated by the homologous but distinct mineralocorticoid receptor, the regulation of which by CDK5 is uncertain at this time (51). Thus, CDK5 might contribute to such location-specific alterations of GR activity in the CNS in many different ways.
GR has several phosphorylation sites, and all of them are located in the AF-1 domain of its NTD (7, 39). Classically, GR is phosphorylated after binding to its ligand, and this may determine target promoter specificity, cofactor interaction, strength and duration of receptor signaling, and receptor stability (7, 36). There are several kinases that phosphorylate the GR in vitro and in vivo; yeast cyclin-dependent kinase p34CDC28 phosphorylates rat GR at serines 224 and 232, which are orthologous to serines 203 and 211 of the human GR, with the resultant phosphorylation enhancing rat GR transcriptional activity in the yeast (11). These residues are also phosphorylated after activation of the GR with agonists or antagonists, and the phosphorylated receptor shows reduced translocation to the nucleus and/or altered subcellular localization in mammalian cells (7, 12). The p38 MAPK phosphorylates serine 211 of the human GR, enhances the transcriptional activity of the GR fragment lacking the LBD, and mediates GR-dependent apoptosis in cells transfected with this fragment (54). p38 MAPK and c-Jun N-terminal kinase also phosphorylate serine 226 of the human GR and suppress its transcriptional activity by enhancing nuclear export of the receptor (8). Threonine 171 of the rat GR is phosphorylated by p38 MAPK and glycogen synthase kinase-3; phosphorylated GR demonstrates reduced transcriptional activity in yeast and human cells; however, the human GR does not have a threonine residue equivalent to that of the rat GR (55). In this report, we demonstrated that CDK5 phosphorylates the human GR at serines 203 and 211, which reside in perfect consensus sequences for CDK5-induced phosphorylation, and suppresses GR-induced transcriptional activity on the MMTV and SGK promoters, by inhibiting accumulation of transcriptional cofactors on GRE-bound GR. Serine 226 is also phosphorylated by CDK5 and contributes to its alteration of GR-induced transcriptional activity, although this serine residue does not reside in a perfect consensus sequence.
Our microarray analyses revealed that the effect of CDK5 on glucocorticoid-induced transcriptional activity is much more diverse and complicated, positively or negatively regulating GR transcriptional activity in a gene promoter-specific fashion. The findings of this study add significant complexity to the glucocorticoid signaling system, which is already dauntingly stochastic (39, 56).
We demonstrated that CDK5, through its kinase activity, suppressed attraction of cofactors by GR bound on GREs of the exogenous MMTV promoter in HCT116 and COS7 cells and of the endogenous SGK promoter in rat neuronal cortical cells. Because phosphorylation-induced transcriptional regulation is promoter specific, it is possible that phosphorylation may facilitate or inhibit, respectively, the attraction of transcription cofactors on glucocorticoid-responsive genes, through as yet undetermined mechanisms that appear to be gene and tissue specific (Fig. 7B). We found that mutations of serines 203 and/or 211 of GR increase the transcriptional activity of these mutant receptors on the MMTV promoter in HCT116 cells, whereas previous reports demonstrated the opposite effects (11, 13, 54). These findings further support our hypothesis that gene promoter/tissue-specific cofactors might regulate the direction of the transcriptional effect induced by GR phosphorylation. Indeed, the tumor susceptibility gene 101, which interacts with the histone acetyltransferase coactivator p300 and inhibits the transcriptional activity of the GR by modulating GR-induced attraction of coactivators, preferentially interacts with a nonphosphorylated form of the GR (7, 55, 57). Thus, the tumor susceptibility gene 101 and/or similar molecules that specifically bind nonphosphorylated or phosphorylated forms of the GR may mediate phosphorylation-dependent transcriptional modulation by changing accumulation of cofactors on promoter-bound GR. Tissue-specific or brain region-specific expression of such molecules might also explain tissue- or region-dependent regulation of GR transcriptional activity by CDK5. Because phosphorylation sites of the GR are clustered in its AF-1 domain, it is likely that AF-1 acts as an interface for phosphorylation-dependent intracellular molecular signals, in contrast to the AF-2 domain, which is located in the LBD and mediates the effects of different interactants of the GR independently of phosphorylation (6, 58).
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
, pRSG-gal-G, and pRSerbA–1 are generous gifts of Dr. R. M. Evans (Salk Institute, La Jolla, CA). pRSerbA–1 was used as a negative control for pRShGR-related plasmids. pRShGR S45A, S203A, S211A, S226A, S395A, S203/211A, S203/211/226A, and S45/203/211/226/395A, which, respectively, express GR with serine to alanine substitutions at indicated residues, were created by using PCR-assisted site-directed mutagenesis. pM-GR-NTD, pM-GRLBD, and pBK-CMV-GR were reported previously (59). pCDNA3-CDK5, -CDK5D144N, p35, and p25, which, respectively, express the wild-type human CDK5, a dominant-negative form of CDK5 on the kinase activity of the wild-type form due to aspartic acid to asparagine substitution at position 144, and the human p35 and p25, were previously reported (60). pCDNA3.1 His/C-p35 (190–262), which expresses His-tagged p35 (190–262), was created by subcloning the corresponding cDNA fragment of p35 into pCDNA3.1 His/C (Invitrogen, Carlsbad, CA). pCMVß-p300-CHA and pSVhSNF2 were gifts from Drs. D. M. Livingston (Dana-Farber Cancer Institute, Boston, MA) and H. Kato (Nippon Roche Research Center, Kanagawa, Japan), respectively. pCDNA3-FLAG-Smad6 was described previously (38). pMMTV-luc and p17mer-tk-Luc, which express luciferase under the control of glucocorticoid-responsive, 4 GRE-containing MMTV promoter and the GAL4 response elements-driven promoter, are gifts from Drs. G. L. Hager (National Cancer Institute, Bethesda, MD) and M.-J. Tsai (Baylor College of Medicine, Houston, TX), respectively. pSGK1-Luc, which has –1899 to +117 of the human SGK1 promoter that includes one functional tandem GRE, was a gift from Dr. C. P. Thomas (University of Iowa College of Medicine, Ames, IA) (32). pGEX-4T3-GR (1–490) and pGEX-2T-CDK5, -p35, and -p25, which, respectively, express GST-fused indicated molecules, were previously reported (28, 61). pGEX-4T3-GR (1–490) S45A, S203A, S211A, S226A, S395A, S203/211A, S203/211/226A, and S45/203/211/226/395A were created by using PCR-assisted site-directed mutagenesis. pGEX4T3-p35 (190–262) was created by subcloning the corresponding cDNA fragment of p35 into pGEX-4T3 (GE Healthcare Bioscience Corp., Piscataway, NJ). pSos-GR (485–777), which expresses human GR LBD fused to the human son-of-sevenless (Sos), was constructed by inserting the cDNA fragment that encodes amino acids 485–777 of the human GR, into pSos (Stratagene, La Jolla, CA). pLexA-GR and -GR-NTD, -DBD, and -LBD were described previously (61). pB42AD-p35 (1–308, 90–308, 90–191) and (190–308) were constructed by inserting the indicated fragment of human p35 cDNA into pB42AD (CLONTECH Laboratories, Inc., Palo Alto, CA). pCDNA3, pLexA, pOP8-LacZ, and pSV40-ß-Gal were purchased from Invitrogen, CLONTECH, and Promega Corp. (Madison WI), respectively.
Yeast Two-Hybrid Screening and Assay
Cyto-trap yeast two-hybrid screening assay was performed by using pSos-GR (485–777) as a bait plasmid in the human fetal brain cDNA library (Stratagene) following the company’s instructions. The yeast two-hybrid assay was performed with the LexA system (CLONTECH), as previously described (62). The ß-galactosidase activity was normalized for O.D. value at 600 nm. Fold induction was calculated by the ratio of adjusted ß-galactosidase values of transformed cells cultured in the presence of galactose/raffinose vs. those in the medium containing glucose.
Cell Cultures and Transfections
Human colon carcinoma HCT116 and African green monkey kidney COS7 cells were maintained in McCoy’s 5A medium and DMEM supplemented with 10% fetal bovine serum, 50 U/ml of penicillin, and 50 µg/ml of streptomycin, respectively. HCT116/MMTV and COS7/MMTV cells, which were stably transformed with pMAM-neo-Luc (CLONTECH) that has the full-length MMTV promoter upstream of the luciferase gene, were maintained in McCoy’s 5A medium or DMEM containing neomycin and the same supplements (38, 62). Neither HCT116, HCT116/MMTV, COS7, nor COS7/MMTV cells contain functional GR. HCT116, HCT116/MMTV, COS7, and COS7/MMTV cells were transfected as previously described (38, 62). HCT116 cells were transfected with the indicated amounts of CDK5 and/or p35-related plasmids, 0.5 µg/well of GR-related plasmids together with 1.5 µg/well of luciferase-expressing reporter plasmid, and 0.5 µg/well of pSV40-ß-Gal. Primary cultures of rat cortical neuronal cells were prepared from embryonic d 18 (E18) rat fetuses and cultured in Neurobasal medium and B27 supplement (Invitrogen) containing 100 U/ml of penicillin, 100 µg/ml of streptomycin, and 2 mM glutamine, as described previously (60). After 7 d of culture, these cells were treated with 10–6 M dexamethasone and/or 20 µM roscovitine (Sigma-Aldrich, St Louis, MO) and used for the coimmunoprecipitation or ChIP assays or purification of total RNA for oligonucleotide microarray analyses.
Regular Coimmunoprecipitation Assay
Rat cortical neuronal cells and HCT116 cells transfected with indicated plasmids were treated with 10–6 M dexamethasone or vehicle for 5 h. Cells were lysed in buffer containing 50 mM Tris-HCl (pH 7.4), 150 mM NaCl, 0.1% sodium dodecyl sulfate, 1% Nonidet P-40, 0.5% sodium deoxycholate, and one tablet/50 ml Complete Tablet (Roche Diagnostics Corp., Indianapolis, IN), and coimmunoprecipitation was carried out as previously described (62). Proteins were immunoprecipitated by anti-hGR, anti-CDK5 antibodies, or control rabbit IgG (Santa Cruz Biotechnology, Inc., Santa Cruz, CA) and the protein-antibody complexes were collected with Protein Agarose A/G PLUS (Santa Cruz Biotechnology, Inc.). Associated proteins were separated in 8 or 10–20% SDS-PAGE gels and blotted on nitrocellulose membranes, and GR-associated CDK5 or p35 was detected by anti-CDK5 or -p35 antibodies, respectively, whereas His-p35 (190–262) was probed with anti-His antibody (Santa Cruz Biotechnology, Inc.). To evaluate endogenously or exogenously expressed GR, CDK5, and p35, 10% of cell lysates used in the coimmunoprecipitation reaction were run on SDS-PAGE gels.
To examine expression levels of the wild-type GR and its serine mutants, plasmids expressing these molecules were transfected into HCT116 cells, and cell lysates were prepared as described above. Expressed GR-related proteins were then separated on an 8% SDS-PAGE gel, and then blotted and visualized with anti-GR antibody (Santa Cruz Biotechnology, Inc.).
GST Pull-Down Assay
35S-labeled human GR, CDK5, and Smad6 were generated by in vitro translation using pBK-CMV-GR, pCDNA3-CDK5, and pCDNA3-FLAG-Smad6, respectively, as templates and tested for interaction with GST-CDK5, -p35, -p25, -p35 (190–262), or -GR (1–490)s, immobilized on glutathione-sepharose beads in the presence or absence of 10–6 M dexamethasone, as previously described (63). After vigorous washing with the buffer, proteins were eluted and separated on an 8% or 10–20% SDS-PAGE gels. Gels were fixed, treated with Enlighting (NEN Life Science Products, Inc., Boston, MA), dried, and exposed to film.
In Vivo Phosphorylation of GR at Serines 203 and/or 211 by CDK5
To examine in vivo phosphorylation of the GR by CDK5, HCT116 cells were transfected with the GR-expressing plasmid in the absence or presence of CDK5 wild-type- or D144N mutant-, and p25-expressing plasmids. After stimulating with 10–6 M of dexamethasone for 5 h, cells were lysed and GRs phosphorylated at serine 203 or 211 was detected using their specific antibodies (12).
In Vitro Phosphorylation of a GR Fragment by CDK5
Expression and purification of the GST and GST-GR (1–490) proteins were performed as described above. Equal amounts of GST-GR (1–490) proteins were mixed with 10 ng of active CDK5 and p35 (Upstate Biotechnology, Inc., Lake Placid, NY) in a buffer containing 20 mM HEPES (pH 7.4), 1 mM EDTA, 10 mM MgCl2, 0.2 mM dithiothreitol, and a protease inhibitor cocktail (2 µl) in a final volume of 50 µl (28). The reaction was started by addition of 100 µM ATP containing 10 µCi of [
-32P]ATP and incubated at 30 C for 30 min. Samples were separated on 8–16% SDS-PAGE gels, and the gels were subjected to autoradiography.
ChIP Assay
ChIP assay was performed in HCT116/MMTV, COS7/MMTV, and rat cortical neuronal cells that have the genomically integrated MMTV-luciferase gene and the endogenous SGK gene, respectively, as previously described (38). Briefly, HCT116/MMTV and COS7/MMTV cells were transfected with CDK5 wild-type- or D114N mutant-, and p25-expressing plasmid together with the GR-expressing plasmid, and exposed to either 10–6 M dexamethasone or vehicle for 5 h. pCMVß-p300-CHA and pSVhSNF2 were cotransfected in COS7/MMTV cells. Rat cortical neuronal cells were incubated with 20 µM roscovitine for 5 h. The cells were then fixed, DNA and bound proteins were cross-linked, and ChIP assays were performed by coprecipitating the DNA-protein complexes with anti-GR, anti-p300, anti-SNF2-h antibodies, or rabbit control IgG (Santa Cruz Biotechnology, Inc.). The promoter regions –219 to –47 of the MMTV long terminal repeat and –1044 to –895 of the rat SGK promoter that contain functional GREs were amplified from the prepared DNA samples using specific primer pairs (MMTV promoter: 5'-AACCTTGCGGTTCCCAG-3' and 5'-GCATTTACATAAGATTTGG-3'; SGK promoter: 5'-CCTCCTCACGTGTTCTTG-3' and 5'-GAAATAAGTCTCGCGCTACAAG-3') in the SYBR green real-time PCR using the SYBR Green PCR Master Mix (Applied Biosystems, Foster City, CA) and a 7500 Real-time PCR System (Applied Biosystems) (62). Obtained threshold cycle values of ChIP samples were normalized for those of corresponding inputs, and their relative precipitations were demonstrated as fold precipitation above the baseline.
Microarray Analysis
Rat cortical neuronal cells were treated with 10–6 M dexamethasone and/or 20 µM roscovitine for 8 h. To avoid inactivation of roscovitine, the compound was added to the medium every 2 h. Total RNA was purified from cells, and probes for the microarray hybridization were prepared from 5 µg of total RNA by using the One-Cycle Target Labeling and Control Reagents Kit (Affymetrix, Inc., Santa Clara, CA). The Rat Genome 230 2.0 Tiled Array was then labeled with prepared probes, washed, and stained with the Affymetrix working station (Affymetrix). Detailed data analysis steps were performed as previously described (Wang, Y., Z.-H. Miao, Y. Pommier, E. S. Kawasaki, and A. Player, manuscript submitted). Briefly, all probe level annotations on the chip were verified before performing analysis by remapping to current Unigene sequences database (March 2006), and only those correctly mapped probes were used for analysis. Samples were examined using the Perfect-Match-only method with MAS background signals subtraction [ref: www.bioconductor.org]. One sample Student’s t test was performed based on comparisons of gene expression values (log ratio) of the same probe set among all replicates with critical value as P 
0.05. Candidate genes were identified by using the z distribution by calculating the 95% cut-off interval.
Expression levels of candidate genes identified in the microarray analysis were further confirmed with the real-time PCR. Briefly, total RNA purified from three independent experiments were reverse transcribed as previously described (62), and mRNAs of the indicated genes were determined in triplicate in the SYBR green real-time PCR using the specific primer pairs listed in Table 1, as described above. Rat acidic ribosomal phosphoprotein P0 was used as an internal control. Their threshold cycle values were normalized for those of ribosomal phosphoprotein P0, and relative mRNA expression levels were demonstrated as fold induction above the baseline. The dissociation curves of the primer pairs used showed a single peak, and after the PCRs, the samples had a single expected DNA band in an agarose gel analysis (data not shown).
Statistical Analysis
Statistical analysis was carried out by ANOVA, followed by Student’s t test with Bonferroni correction for multiple comparisons or unpaired t test with the two-tailed P value.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Disclosure Statement: All authors have nothing to disclose.
First Published Online April 17, 2007
Abbreviations: AF-1, Transactivation function 1; CDK5, cylin-dependent kinase 5; ChIP, chromatin immunoprecipitation; CNS, central nervous system; DBD, DNA-binding domain; GR, glucocorticoid receptor; GRE, glucocorticoid response element; GST, glutathione-S-transferase; LBD, ligand-binding domain; MMTV, mouse mammary tumor virus; NTD, N-terminal domain; SGK, serum/glucocorticoid-inducible kinase; Smad, Sma- and Mad-related protein; SNF2, sucrose non-fermenting 2.
Received for publication August 21, 2006. Accepted for publication April 10, 2007.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
. Mol Cell Biol 26:8357–8370
