Phosphorylation of Serine 468 by GSK-3β Negatively Regulates Basal p65 NF-κB Activity*

The activity of NF-κB is controlled at several levels including the phosphorylation of the strongly transactivating p65 (RelA) subunit. However, the overall number of phosphorylation sites, the signaling pathways and protein kinases that target p65 NF-κB and the functional role of these phosphorylations are still being uncovered. Using a combination of peptide arrays with in vitro kinase assays we identify serine 468 as a novel phosphorylation site of p65 NF-κB. Serine 468 lies within a GSK-3β consensus site, and recombinant GSK-3β specifically phosphorylates a GST-p65-(354–551) fusion protein at Ser468 in vitro. In intact cells, phosphorylation of endogenous Ser468 of p65 is induced by the PP1/PP2A phosphatase inhibitor calyculin A and this effect is inhibited by the GSK-3β inhibitor LiCl. Reconstitution of p65-deficient cells with a p65 protein where serine 468 was mutated to alanine revealed a negative regulatory role of serine 468 for NF-κB activation. Collectively our results suggest that a GSK-3β-PP1-dependent mechanism regulates phosphorylation of p65 NF-κB at Ser468 in unstimulated cells and thereby controls the basal activity of NF-κB.

The activity of NF-B is controlled at several levels including the phosphorylation of the strongly transactivating p65 (RelA) subunit. However, the overall number of phosphorylation sites, the signaling pathways and protein kinases that target p65 NF-B and the functional role of these phosphorylations are still being uncovered. Using a combination of peptide arrays with in vitro kinase assays we identify serine 468 as a novel phosphorylation site of p65 NF-B. Serine 468 lies within a GSK-3␤ consensus site, and recombinant GSK-3␤ specifically phosphorylates a GST-p65-(354 -551) fusion protein at Ser 468 in vitro. In intact cells, phosphorylation of endogenous Ser 468 of p65 is induced by the PP1/PP2A phosphatase inhibitor calyculin A and this effect is inhibited by the GSK-3␤ inhibitor LiCl. Reconstitution of p65-deficient cells with a p65 protein where serine 468 was mutated to alanine revealed a negative regulatory role of serine 468 for NF-B activation. Collectively our results suggest that a GSK-3␤-PP1-dependent mechanism regulates phosphorylation of p65 NF-B at Ser 468 in unstimulated cells and thereby controls the basal activity of NF-B.
NF-B is a dimeric transcription factor that plays an important role in the immune response, cell survival, and cancer. NF-B activity is controlled at two levels: (i) by proteasome-dependent generation of DNA-binding subunits and (ii) by regulation of its nuclear function. In the recent years, evidence has accumulated that post-translational modifications of the DNA-binding subunits add another level of regulation for the function of NF-B (1)(2)(3).
A number of protein kinases have been shown to phosphorylate the strongly transactivating subunit p65 at Ser 276 (4), Ser 311 (5), Ser 529 (6), and Ser 536 (7), and, with the exception of Ser 529 , phospho-specific antibodies have confirmed phosphorylation of endogenous p65 at these sites. The relevance of regulatory phosphorylations is also evident from the analysis of cells lacking the protein kinases GSK-3␤ (8), TBK1/NAK (9, 10), IKK⑀ (11), NIK (12), and PKC (13), which show an intact IB phosphorylation but an impaired expression of NF-B target genes. Nonetheless, the overall number of p65 phosphorylation sites is not yet known as is the number of all potential p65 protein kinases. As an example, we and others (14 -17) have recently shown that at least six distinct kinases converge on phosphorylation of p65 at Ser 536 . The molecular mechanisms and the biological consequences of p65 phosphorylation are currently a focal point of intense research (2,3). Most p65 phosphorylation sites are located in the COOH-terminal part of the Rel homology domain and in the COOH-terminal transactivation domains (2,3).
The involvement of GSK-3␤ in activation of NF-B as suggested by gene deletion has been a surprising finding, as NF-B activating stimuli such as IL-1, TNF, or phorbol ester will inactivate GSK-3␤ by phosphatidylinositol 3-kinase/AKT-mediated phosphorylation of its NH 2 terminus (18 -20). Furthermore, it is suggested that in NF-B activation the role of GSK-3␤ is non-redundant as the closely related enzyme GSK-3␣ cannot compensate for the loss of GSK-3␤ (8,18). In contrast with the results derived from knock-out mice, in neuronal cells expression of an active form of GSK-3␤ suppresses NF-B activity by inhibiting IB kinase (IKK) 1 and stabilizing IB (21,22). GSK-3␤ has also recently been shown to phosphorylate a GST-p65 fusion protein in vitro, but the relevant site(s) have not been determined and it has remained unclear if p65 is a physiological substrate for GSK-3␤ in vivo (23). Using a peptide array-based approach (24) we detected a protein kinase activity that specifically phosphorylated Ser 468 of p65 NF-B. Experiments presented here strongly suggest that GSK-3␤ is a physiological Ser 468 protein kinase and we imply this phosphorylation site in negative control of NF-B. In the light of the opposing findings regarding the role of GSK-3␤ in NF-B signaling our results close an important gap in the understanding of the role of GSK-3␤ in regulation of p65 activity.
Peptide Arrays-The peptide array containing p65 NF-B peptide spots was generated following SPOT synthesis (25). 180 peptide fragments of 15 amino acid residues in length and overlapping by 12 residues were generated such that the entire p65 NF-B protein sequence was covered. These peptides were chemically synthesized as an array of spots on an amino-polyethylene glycol-modified cellulose membrane (AC-S01, AIMS Scientific Products GmbH, Braunschweig, Germany) as described previously (26). All peptides are NH 2 -terminal acetylated and remain covalently attached to the membrane via their carboxyl termini.
Plasmids and Transfections-The expression plasmid for the p65 TAD, pGEX-p65-(354 -551) was a kind gift of H. Sakurai, Toyama, Japan. GST fusion proteins were expressed in bacteria and purified on GSH-Sepharose using standard procedures. pMT7-p65 NF-B has been published (27) and NF-B (3)luc contained three NF-B-binding sites upstream of a luciferase cDNA. pSV-␤-gal coding for SV40 promoter driven ␤-galactosidase was from Promega. p65 Ϫ/Ϫ cells cells were seeded in 6-well plates and transfected at 70 -80% confluence using Rotifect (Roth) according to the manufacturer's instructions.
Preparation of Cell Extracts-For the preparation of whole cell extracts cells were lysed directly in SDS-PAGE sample buffer. DNA was sheared by brief sonification, and soluble proteins were recovered after centrifugation of lysates at 15,000 ϫ g for 15 min at 4°C. For in vitro kinase assays cells were lysed in 10 mM Tris, pH 7.05, 30 mM NaPP i , 1% Triton X-100, 2 mM Na 3 VO 4 , 50 mM NaF, 20 mM ␤-glycerophosphate and freshly added 0.5 mM phenylmethylsulfonyl fluoride, 0.5 g/ml leupeptin, 0.5 g/ml pepstatin, 400 nM okadaic acid. After 10 min on ice, lysates were clarified by centrifugation at 10,000 ϫ g for 15 min at 4°C.
Nuclear and cytosolic extracts were prepared as described previously (27). The protein concentration of cell extracts was determined by the method of Bradford, and samples were stored at Ϫ80°C.
In Vitro Kinase Assays-For the kinase assay shown in Fig. 1A, 10 l of cell lysate (50 g of protein) was added to 1 g of recombinant protein substrates (GST-p65-(354 -551) or mutants thereof) in 10 l of H 2 O and 10 l of kinase buffer (150 mM Tris, pH 7.4, 30 mM MgCl 2 , 60 M ATP, 4 Ci of [␥-32 P]ATP). After 15 min at 30°C in vitro phosphorylated GST-p65 fusion proteins were purified on GSH-Sepharose prior to SDS-PAGE as described by Holtmann et al. (28). Then, SDS-PAGE sample buffer was added, and proteins were eluted from the beads by boiling for 5 min. After centrifugation at 10,000 ϫ g for 5 min, supernatants were separated on 10% SDS-PAGE. Phosphorylated proteins were visualized by autoradiography.
For in vitro phosphorylation of immobilized peptides the peptide arrays were incubated with 2 ml of cell extract (4.7 mg of protein), 2 ml of cell lysis buffer, and 2 ml of kinase buffer (150 mM Tris, pH 7.4, 30 mM MgCl 2 , 10 M ATP, 200 Ci of [␥-32 P]ATP). After 30 min at 30°C membranes were washed twice in phosphate-buffered saline, once in 8 M urea, 1% SDS, 0.5% ␤-mercaptoethanol for 30 min at 40°C, twice in H 2 O, and three times in EtOH. Air-dried membranes were autoradiographed at 4°C or at room temperature.
For the experiments shown in Fig. 2A 100 units of recombinant GSK-3␤ was incubated with 1 g of GST-p65 fusions proteins in 1ϫ GSK-3␤ reaction buffer (New England Biolabs) supplemented with 20 M ATP and 2.5 Ci of [␥-32 P]ATP in a total volume of 30 l for 30 min at 30°C. Reactions were stopped by the addition of SDS-PAGE sample buffer and phosphorylation of proteins visualized as described above. For detection of phosphorylated proteins by immunoblotting as shown in Fig. 2B 500 units of GSK-3␤, 50 ng of GST-p65 fusion proteins, and 135 M ATP were used in the kinase reaction, and radioactive ATP was omitted. Western blotting and site-directed mutagenesis were performed as described by Buss et al. (14).

RESULTS AND DISCUSSION
To investigate the occurrence of IL-1-inducible phosphorylation sites within the transactivating COOH terminus of p65, whole cell extracts isolated from unstimulated and IL-1-treated HeLa cells were incubated with a recombinant GST-p65-(354 -551) fusion protein in the presence of [␥-32 P]ATP. These experiments revealed constitutive and IL-1-inducible protein kinase activities. Mutations in Ser 529 and Ser 536 , the sites that have been found as targets for the hitherto identified p65 TAD kinases IKK␣ and IKK␤ (7), casein kinase II (6), TBK1 (14, 29), RSK1 (16), and IKK⑀ (14), did not completely abolish in vitro phosphorylation of GST-p65, suggesting the existence of fur-ther p65 phosphorylation site(s) and kinases (Fig. 1A). To profile phosphorylation sites in p65 the complete p65 coding sequence arrayed as overlapping 15-mer peptides was subjected to in vitro kinase assays with cell extracts from IL-1 stimulated HeLa cells. This approach revealed more than 20 peptides that were phosphorylated in vitro (Fig. 1B), including some that contained already identified phosphorylation sites such Ser 536 (Fig. 1B, peptides H1-H3). 19 peptides that are indicated by white circles in Fig. 1B contained potentially novel phosphorylation sites and were selected for further study. To identify the phosphorylated amino acids within these 19 peptides, all threonine, serine, and tyrosine residues within these sequences were systematically mutated and tested for phosphorylation (Fig. 1C). In many peptides these mutations did not change in vitro phosphorylation (Fig. 1C). Some peptides showed reduced phosphorylation upon mutation, and we are currently analyz-

FIG. 1. Identification of Ser 468 as novel phosphorylation site of p65 NF-B.
A, HeLa cells were stimulated for 10 min with IL-1 (10 ng/ml) or left untreated. Cell extracts were assayed in duplicates for kinases that phosphorylate wild type (wt) GST-p65-(354 -551) or a mutant in which Ser 529 and Ser 526 were mutated to alanine (SS529,536AA) by in vitro kinase assays (ka). GST-p65 fusion proteins were purified by GSH-Sepharose, separated on SDS-PAGE, and analyzed by autoradiography. Equal recovery was confirmed by Coomassie Brilliant Blue staining (CBB) of gels. B, cell extracts from IL-1-stimulated cells prepared as described for A and [␥-32 P]ATP were incubated with a membrane-bound array of 180 peptides (A1 to H5) that cover the entire p65 coding sequence (amino acids 1-551). Membranes were washed and phosphorylation of peptides by cellular kinases detected by autoradiography. 19 peptides selected for further analysis of phosphorylation sites are indicated by white circles. The amino acid sequence of the peptide contained in spot G5 that includes serine 468 (underlined) is shown. C, in the 19 peptides identified in B Ser, Thr, and Tyr residues were systematically mutated to non-phosphoacceptor residues (Ala), and these peptides were synthezised as an array (A1 to D25) on a cellulose membrane. In vitro kinase assays were performed as described for B. As indicated in the inserted table peptides D19 to D25 contained amino acids 463-477 of p65 and systematic alanine mutations in all possible phosphorylation sites (underlined).
To directly address the question whether this consensus sequence is phosphorylated by recombinant GSK-3␤ in vitro, the purified kinase was incubated with a GST-p65-(354 -551) substrate protein and a GST-p65-(354 -551-Ser 468 -Ala) control protein where the phosphorylation site was point mutated. GSK-3␤ efficiently phosphorylated the GST-p65-(354 -551) protein, while the GST-p65-(354 -551-Ser 468 -Ala) mutant was phosphorylated to a minor extent ( Fig. 2A), revealing GSK-3␤ as a serine 468 kinase. To obtain evidence for GSK-3␤-mediated p65 serine 468 phosphorylation by an independent experimental approach, the products of the in vitro kinase assays were analyzed using an antibody specifically recognizing the phosphorylated form of serine 468. These experiments clearly confirmed that GSK-3␤ phosphorylated specifically Ser 468 , as no signal with the antibody was obtained using the S468A mutant protein as substrate or by omitting the substrate (Fig.  2B). Thus the experiments shown in Fig. 2 identify p65 Ser 468 as a specific GSK-3␤ phosphorylation site in vitro. The residual phosphorylation observed in the radioactive kinase assay ( Fig.  2A) and the detection of three proteins bands of different mobility on SDS-PAGE by immunoblotting with the phospho-Ser 468 -specific antibody (Fig. 2B) leave the possibility that the p65 TAD might contain additional GSK-3␤ sites as previously suggested by Schwabe and Brenner (23). To investigate whether GSK-3␤ phosphorylates Ser 468 in vivo, HeLa cells were stimulated with IL-1, TNF, or phorbol ester plus ionomycin, but phosphorylation of endogenous Ser 468 was only faintly activated (data not shown). Intriguingly, treatment of cells with the PP1/PP2A Ser/Thr phosphatase inhibitor calyculin A (32) for 30 min strongly induced Ser 468 phosphorylation. These results can be reconciled with a recently suggested model whereby PP1 dephosphorylates GSK-3␤ at the NH 2 -terminal serine 9, thereby activating the kinase (33). In this model, mutual control is ensured by GSK-3␤-mediated inhibition of the protein I-2, a negative regulator of PP1, thus maintaining GSK-3␤ in an active state in unstimulated cells. This mechanism also implies that inhibition of GSK-3␤ accelerates inactivation of PP1 by I-2 (33). We therefore analyzed if inhibition of PP1 affects phosphorylation of GSK-3␤ in the cells employed in this study. Incubation of HeLa cells with calyculin A triggered the phosphorylation of GSK-3␣ at serine 21 and of GSK-3␤ at serine 9, as revealed by immunoblotting with a phospho-spe- cific antibody that recognizes both enzymes (Fig. 3, compare  lanes 1 and 2).
In this situation, addition of increasing amounts of LiCl, a specific inhibitor of GSK-3 (33), further increased phosphorylation of GSK-3␤ and in parallel suppressed phosphorylation of Ser 468 in a dose-dependent fashion (Fig. 3, lanes 3-5). Simultaneous detection of phosphorylated GSK-3␣ with the phosphospecific antibody showed that LiCl also impaired phosphorylation of GSK-3␣ (Fig. 3). However, as GSK-3␣ has not been implicated in in NF-B activation (18), its role in p65 phosphorylation was not further investigated in our study.
In parallel, calyculin A also induced phosphorylation and degradation of IB␣ (Fig. 3, lane 2), which is in accordance with earlier studies (34). These effects support the conclusion that a low level of kinase activity of GSK-3 and of IKKs is maintained in unstimulated cells whose effect on substrate phosphorylation is counteracted by high levels of phosphatase activity. With respect to p65, these experiments strongly implicate GSK-3␤ in phosphorylation of endogenous Ser 468 and also suggest that a calyculin A-sensitive protein phosphatase regulates both NH 2 -terminal phosphorylation of GSK-3␤ and Ser 468 phosphorylation of p65. Based on the experiments reported by Zhang et al. (33) we suggest that PP1 is the phospho-Ser 468 phosphatase. In our model a high level of active, dephosphorylated GSK-3␤ not only increases phosphorylation of Ser 468 of p65 but at the same time also activates PP1, which dephosphorylates Ser 468 .
To characterize the role of Ser 468 functionally we ectopically expressed wild type p65 or the S468A mutant in p65-deficient cells. Under these conditions both proteins were expressed to comparable levels and were detectable within the cytoplasm and the nucleus (Fig. 4A). Interestingly, in p65-deficient cells the S468A mutant showed a slightly faster mobility on SDS-PAGE compared with wild type p65 (Fig. 4A), a phenomen that was not observed when both proteins were transcribed and translated in vitro (data not shown). This observation is in line with a loss of GSK-3␤-mediated phosphorylation. When expressed in p65-negative cells, the S468A mutant displayed an about 4-fold increase in activation of a cotransfected NF-B reporter gene as compared with wild type p65 (Fig. 4B). Expression of the S468A mutant also enhanced p65 activity induced by IL-1, TNF, or phorbol 12-myristate 13-acetate (data not shown). These experiments suggest that Ser 468 phosphorylation by GSK-3␤ has a negative regulatory role for p65 activity the mechanism of which awaits further investigation.
In summary, we provide strong evidence that Ser 468 is a new phosphorylation site of p65 NF-B and identify GSK-3␤ as the kinase that phosphorylates this residue in the absence of extracellular ligands. Ser 468 phosphorylation may also play a role in response to certain specific stimuli as we found during the course of this work that T-cell costimulation induces Ser 468 phosphorylation. However, the kinase that mediates this effect and its biological significance remain to be identified (35).
Here, we suggest a function for GSK-3␤-mediated phosphorylation of Ser 468 in negative regulation of p65. Collectively, these results predict that the balance of active GSK-3␤ and PP1 determines the phosphorylation status of Ser 468 in unstimulated cells (see Fig. 5). In conjunction with other potential GSK-3␤ phosphorylation sites Ser 468 may contribute to the altered constitutive activity of NF-B that has been observed in chronic inflammatory disease (36) and in different tumors (37). Thus Ser 468 phosphorylation maybe another crucially important determinant for the outcome of NF-B activation in inflammation and cancer.  (1) and by that recently reported by Zhang et al. (34) (2) suggests that GSK-3␤ phosphorylates Ser 468 of p65 NF-B and in parallel activates the phosphatase PP1, which dephosphorylates both Ser 468 of p65 and Ser 9 of GSK-3␤. Hence, the relative activity of PP1 and of GSK-3␤ determines the level of phospho-Ser 468 -dependent p65 activity in unstimulated cells.