In Vivo Identification of Inducible Phosphoacceptors in the IKK (cid:1) /NEMO Subunit of Human I (cid:2) B Kinase*

Transcription factor NF- (cid:2) B plays a pivotal regulatory role in the genetic programs for cell cycle progression and inflammation. Nuclear translocation of NF- (cid:2) B is controlled by an inducible protein kinase called IKK, which earmarks cytoplasmic inhibitors of NF- (cid:2) B for proteolytic destruction. IKK contains two structurally related catalytic subunits termed IKK (cid:3) and IKK (cid:4) as well as a noncatalytic subunit called IKK (cid:1) /NEMO. Mutations in the X-linked gene encoding IKK (cid:1) can interfere with NF- (cid:2) B signaling and lead to immunodeficiency disease. Although its precise mechanism of action remains un-known, IKK (cid:1) is phosphorylated in concert with the in-duction of NF- (cid:2) B by the viral oncoprotein Tax and the proinflammatory cytokine tumor necrosis factor (cid:3) (TNF). We now demonstrate that TNF-induced phosphorylation of IKK (cid:1) is blocked in cells deficient for IKK (cid:4) but not IKK (cid:3) . Phosphopeptide-mapping experiments with metabolically radiolabeled cells indicate that IKK (cid:4) phosphorylates human IKK (cid:1) at Ser-31, Ser-43, and Ser-376 following the enforced expression of either the Tax oncoprotein or (cid:2) is required for signal-dependent phosphorylation of IKK (cid:4) in vivo . Using a combination of site-directed mutagenesis and phosphopeptide mapping, we have also monitored changes in the phosphorylation status of IKK (cid:4) in metabolically radiolabeled cells. Results from these biochemical experiments indicate that human IKK (cid:4) is phosphorylated at Ser-31, Ser-43, and Ser-376 in response to cellular stimulation with either TNF or the HTLV1 Tax oncoprotein. Minimal deletion of the zinc finger domain of IKK (cid:4) attenuates this inducible response. We conclude that IKK (cid:3) mediates phosphorylation of IKK (cid:4) at both amino-terminal and carboxyl-terminal sites via a zinc finger-dependent mechanism. The observed in vivo pattern of signal-dependent phosphorylation at distal sites in the protein may reflect post-translational control of IKK (cid:4) at two distinct levels.

Transcription factor NF-B and other dimeric members of the Rel polypeptide family regulate the expression of multiple genes involved in inflammation, immunity, mitosis, and cell survival (1)(2)(3). Biologic inducers of NF-B include the proinflammatory cytokines tumor necrosis factor ␣ (TNF) 1 and interleukin-1 (IL-1), the lipopolysaccharide (LPS) component of Gram-negative bacteria, and the Tax oncoprotein of human T-cell leukemia virus type 1 (HTLV1) (4 -7). Each of these signal-dependent responses is controlled by labile cytoplasmic inhibitors of NF-B such as IB␣ and a multicomponent IB kinase called IKK (8). The core IKK holoenzyme contains two catalytic subunits termed IKK␣ and IKK␤. Following cellular stimulation, the IKK␤ catalytic subunit phosphorylates IB␣, leading to degradation of the inhibitor and nuclear translocation of NF-B. Signal-dependent activation of IKK␤ is triggered by phosphorylation of two serine residues in its "T loop" regulatory domain (9). This modification appears to involve either autophosphorylation or phosphoryl group transfer from an upstream IKK␤ kinase to the same acceptor sites following cellular stimulation (9).
More recent studies of the IB kinase complex have identified a noncatalytic component called IKK␥ (also known as NEMO, IKKAP1, or FIP-3) (10 -13). IKK␥ is required for signaldependent activation of IKK␤, leading to its assignment as an essential regulatory subunit of the enzyme (10,12,14). Small deletions or point mutations in the gene encoding IKK␥ can cause skin inflammation or humoral immunodeficiencies in humans (15). In terms of structural organization, sequences within the NH 2 -terminal half of IKK␥ mediate its interaction with IKK␤ (11). In contrast, the COOH-terminal half of IKK␥ is required for signal-dependent regulation of IB kinase activity, suggesting that IKK␥ links IKK␤ to upstream activators (11,12). Despite all of these findings, the mechanism of IKK␥ action remains elusive (16). In this regard, we and others (9,(17)(18)(19) have recently shown that IKK␥ is phosphorylated in response to NF-B agonists such as TNF and the Tax oncoprotein of HTLV1. Considering the key role that phosphorylation plays in the mechanism for IKK␤ activation, these findings suggest that IKK␥ subunit phosphorylation is important for proper regulation of the NF-B signaling pathway.
To extend these fundamental observations, we conducted new experiments that address the mechanism of IKK␥ phosphorylation and the relevant phosphoacceptor sites. In this report, we demonstrate that endogenous IKK␤ but not IKK␣ is required for signal-dependent phosphorylation of IKK␥ in vivo. Using a combination of site-directed mutagenesis and phosphopeptide mapping, we have also monitored changes in the phosphorylation status of IKK␥ in metabolically radiolabeled cells. Results from these biochemical experiments indicate that human IKK␥ is phosphorylated at Ser-31, Ser-43, and Ser-376 in response to cellular stimulation with either TNF or the HTLV1 Tax oncoprotein. Minimal deletion of the zinc finger domain of IKK␥ attenuates this inducible response. We conclude that IKK␤ mediates phosphorylation of IKK␥ at both amino-terminal and carboxyl-terminal sites via a zinc finger-dependent mechanism. The observed in vivo pattern of signal-dependent phosphorylation at distal sites in the protein may reflect post-translational control of IKK␥ at two distinct levels.

EXPERIMENTAL PROCEDURES
Reagents-Polyclonal anti-IKK and anti-HA antibodies (H-470, FL-419, and Y-11) were purchased from Santa Cruz Biotechnology, Inc. Monoclonal antibodies for IKK␥, the FLAG epitope, and the T7 tag were purchased from BD Biosciences, Sigma, and Novagen, respectively. Polyclonal anti-IKK␤ antibodies were provided by Nancy Rice (NCI, National Institutes of Health, Bethesda, MD). Rabbit polyclonal antibodies specific for amino acids 321-353 of Tax were provided by Bryan Cullen (Duke University) (20). Expression vectors for Tax, TNF-R1, and epitope-tagged subunits of IKK have been described previously (20 -23). Site-directed mutations were generated using the QuikChange kit (Stratagene) as specified by the manufacturer.
Metabolic Radiolabeling and Subcellular Fractionation-Murine embryonic fibroblasts (MEFs) derived from mice lacking either IKK␣ or IKK␤ have been described previously (24,25). MEFs and human 293T cells (26) were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, 2 mM glutamine, and antibiotics. Human embryonic kidney 293T cells were transfected using the calcium phosphate method (27). At 36 h post-transfection, 293T cells were labeled for 4 h with [ 32 P]orthophosphate (1 mCi/ml, ICN) in phosphate-free Dulbecco's modified Eagle's medium (Mediatech). Cytoplasmic extracts were prepared from recipient cells by detergent lysis (28) in the presence of phosphatase and protease inhibitors (5). Extracts were subjected to immunoprecipitation with anti-FLAG M2 antibodies coupled to agarose in ELB buffer (250 mM NaCl, 50 mM HEPES, 5 mM EDTA, 0.1% Nonidet P-40) (5). MEFs were incubated overnight in Dulbecco's modified Eagle's medium containing 0.5% fetal bovine serum and then labeled in phosphate-free medium for 4 h with [ 32 P]orthophosphate (1 mCi/ml) prior to agonist treatment. Cytoplasmic extracts were prepared as described above, precleared with polyclonal anti-HA antibodies bound to protein A, and immunoprecipitated with polyclonal anti-IKK␥ antibodies. Resultant complexes were washed sequentially with ELB buffer containing 2 M urea and then with 150 mM NaCl, 10 mM sodium phosphate pH 7.2, 0.1% SDS, 0.5% sodium deoxycholate, 1% Nonidet P-40. Immunoprecipitates were fractionated by SDS-PAGE and transferred to polyvinylidine difluoride (PVDF) membranes. Phosphoproteins were identified by autoradiography and analyzed by immunoblotting using an enhanced chemiluminescence system (Pierce) (5). Resolved phosphoproteins were quantitatively analyzed with a Storm 860 PhosphorImager (Amersham Biosciences).
Phosphopeptide Mapping and Phosphoamino Acid Analysis-Phosphoproteins were separated by SDS-PAGE, transferred to PVDF membranes, and visualized by autoradiography. Membrane sections containing radiolabeled IKK␥ were excised, incubated with methanol, and then blocked with 50 mM NH 4 HCO 3 containing 0.1% Tween 20 (Bio-Rad) for 30 min at room temperature. Phosphoproteins were digested in situ with N-tosyl-L-lysine chloromethyl ketone (TLCK) treated chymotrypsin (8 g, Worthington) in 125 l of NH 4 HCO 3 (37 o C, 12 h) (29). Released peptides were subjected to two-dimensional phosphopeptide mapping as described by Boyle et al. (30) excluding the performic acid oxidation. Phosphoproteins were first separated by electrophoresis on TLC plates at pH 1.9 (30) using the Hunter thin-layer electrophoresis system (model HTLE-7000, CBS Scientific). Separation in the second dimension was performed by ascending chromatography in n-butanol (37.5%), pyridine (25%), and acetic acid (7.5%) (30). Resolved phosphopeptides were visualized by autoradiography using Biomax MS high speed film (Eastman Kodak Co.). Phosphoaminoacid analysis was performed by hydrolysis of membrane-bound IKK␥ with 6 N HCl as described previously (30).
IB Kinase Assays-Cytoplasmic extracts were immunoprecipitated with anti-T7 antibodies in the presence of ELB buffer. IB kinase activity was measured as described previously (5,31) in a reaction mixture containing ATP (10 M), [␥-32 P]ATP (5 Ci), and recombinant glutathione S-transferase protein fused to amino acids 1-54 of IB␣. Radiolabeled products were fractionated by SDS-PAGE, transferred to PVDF membranes, and visualized by autoradiography.

IKK␤ Is Required for in Vivo Phosphorylation of IKK␥-Two
interactive components of IKK termed IKK␤ and IKK␥ are essential for TNF-induced phosphorylation and degradation of IB␣ (7). Phosphoryl group transfer to IB␣ is mediated by IKK␤, which is itself activated via a phosphorylation-dependent mechanism (9). However, the functional requirement for IKK␥ in this context remains an enigma (16). In this regard, we recently found that IKK␥ also serves as a phosphorylation substrate in cells expressing the Tax oncoprotein of HTLV1, a potent pathologic inducer of IB kinase activity (19). Accordingly, signal-dependent phosphorylation of IKK␥ may regulate the function of this noncatalytic subunit in the NF-B signaling pathway.
To extend our findings with Tax, we monitored IKK␥ subunit phosphorylation in MEFs following treatment with proinflammatory mediators. In initial experiments, wild type MEFs were metabolically radiolabeled with [ 32 P]orthophosphate and then stimulated with either TNF, IL-1, or LPS. Endogenous IKK␥ complexes were purified by immunoprecipitation, fractionated FIG. 1. IKK␥ phosphorylation is mediated by IKK␤ but not IKK␣. A, immortalized MEFs were starved overnight in low serum (0.5% fetal bovine serum), radiolabeled with [ 32 P]orthophosphate for 4 h, and then stimulated with TNF (40 ng/ml, 10 min), IL-1␤ (10 ng/ml, 10 min), or LPS (15 g/ml, 30 min). IB kinase complexes were isolated from cytoplasmic extracts with polyclonal rabbit antibodies specific for the IKK␥ subunit. Resultant complexes were fractionated by SDS-PAGE and analyzed for 32 P incorporation by autoradiography (top panel). IKK␤ (middle panel) and IKK␥ (bottom panel) protein content was monitored by immunoblotting with the appropriate subunit-specific antibodies. B, WT MEFs or mutant MEFs lacking the indicated IKK subunits were radiolabeled with [ 32 P]orthophosphate for 4 h and stimulated with TNF as described in panel A. IKK complexes were isolated by immunoprecipitation with IKK␥-specific antibodies, fractionated by SDS-PAGE, and analyzed for 32 P incorporation (top panel). IKK polypeptide content was determined by immunoblotting with subunit-specific antibodies (lower panels).
by SDS-PAGE, and analyzed by sequential autoradiography and immunoblotting. As shown in Fig. 1A (top panel), TNF induced significant phosphorylation of IKK␥ as compared with untreated controls (top panel, lanes 1 and 2). Similar results were obtained with IL-1 and LPS (lanes 3 and 4). Changes in the phosphorylation status of IKK␥ could not be attributed to differences in either IKK␤ or IKK␥ protein content (lower panels). As expected, IKK␤ was also phosphorylated and activated under these stimulatory conditions (top panel and data not shown). We conclude that phosphorylation of endogenous IKK␥ is induced not only by Tax but also by multiple proinflammatory agonists of NF-B in murine fibroblasts.
In prior studies, we found that IKK␤ has the capacity to phosphorylate a recombinant IKK␥ substrate in vitro (19). This finding raised the possibility that in vivo phosphorylation of IKK␥ is mediated by IKK␤ or the structurally related IKK␣ catalytic subunit. To test this hypothesis, MEFs deficient for either IKK␣ or IKK␤ were cultured in the presence of [ 32 P]orthophosphate and stimulated with TNF under conditions leading to optimal IB kinase activity (data not shown). We then prepared IKK immunocomplexes and monitored the core subunits for phosphoprotein content. As shown in Fig. 1B (top panel), treatment of wild type MEFs with TNF led to a 13.6-fold increase in IKK␥ phosphorylation (lanes 1 and 2). This inducible response was readily detected in MEFs lacking IKK␣ (lanes 3 and 4), whereas IKK␥ phosphorylation was almost completely blocked in MEFs lacking IKK␤ (lanes 5 and 6). Coupled with our prior in vitro results (19), we conclude that IKK␤ mediates phosphorylation of IKK␥ under physiologic sig-naling conditions. However, we cannot exclude the possibility that IKK␣ plays a secondary role in IKK␥ phosphorylation.
IKK␥ Is Phosphorylated on Multiple Serines-A prerequisite for understanding the functional consequences of IKK␥ subunit phosphorylation is to identify the relevant IKK␤-responsive acceptor sites. To address this important question, expression vectors for IKK␤, IKK␥, and HTLV1 Tax were transfected into 293T cells. After metabolic radiolabeling with 32 P i , ectopic IKK␥ complexes were immunopurified and analyzed by SDS-PAGE. As shown in Fig. 2A, phosphoryl group transfer to IKK␥ was significantly increased in the presence of Tax relative to the basal level of IKK␥ phosphorylation in Tax-deficient cells. Importantly, IKK␤ was chronically phosphorylated and activated by Tax under these experimental conditions, permitting us to capture sufficient quantities of stably modified IKK␥ for subsequent phosphoamino acid and phosphopeptide-mapping analyses. As shown in Fig. 2B, these biochemical studies revealed the presence of phosphoserine in IKK␥ but no evidence for signal-dependent phosphorylation of either threonine or tyrosine residues.
Based on the corresponding cDNA sequences, human IKK␥ contains a total of 26 serine residues, 19 of which are conserved in murine IKK␥ (10 -13). Accordingly, IKK␥ may contain more than one inducible phosphoacceptor site. To address this possibility, human 293T cells were programmed with expression vectors for IKK␥ and IKK␤ in the presence or absence of a Tax effector plasmid. Radiolabeled IKK␥ was purified from recipient cells and digested with chymotrypsin, and the resultant peptides were resolved by two-dimensional phosphopeptide FIG. 2. IKK␥ contains multiple serine phosphoacceptors. A, 293T cells (2 ϫ 10 6 ) were transfected with vectors for FLAG-tagged IKK␤ (60 ng), Tax (0.3 g), and human IKK␥ (T7-tagged, 60 ng). Cells were radiolabeled with [ 32 P]orthophosphate for 4 h. Ectopic IKK complexes were isolated using anti-FLAG antibodies, washed at high stringency, and fractionated by SDS-PAGE. Resolved proteins were subjected to sequential autoradiography (lanes 1 and 2) and immunoblotting with IKK␤-and IKK␥-specific antibodies (lanes 3 and 4). B, 293T cells were transfected and metabolically labeled as described in A. Ectopic IKK␥ was immunopurified, resolved by SDS-PAGE, and hydrolyzed in 6 N HCl. The resultant amino acids were resolved by two-dimensional electrophoresis. Positions of unlabeled phosphoamino acid standards are indicated. C, human 293T cells were transfected and metabolically labeled as described in A. Radiolabeled IKK␥ was resolved by SDS-PAGE, transferred to a PVDF membrane, and digested in situ by chymotrypsin. Peptides were separated by sequential electrophoresis and chromatography on TLC plates and detected by autoradiography. The four major phosphopeptides in wild type IKK␥ are indicated with numbers (PP-1, PP-2, PP-3, and PP-4). Species labeled T appear to derive from phosphorylation of a serine residue in the T7 epitope tag of IKK␥, which has no detectable effect on the distribution or intensity of the four major IKK␥ phosphopeptides (see Fig. 6). mapping (30). As shown in Fig. 2C, IKK␥ derived from Taxexpressing cells contained four major chymotryptic phosphopeptides, which we designated as PP-1, PP-2, PP-3, and PP-4. Radiolabeling efficiencies for the same set of IKK␥ phosphopeptides were significantly reduced in parallel studies with Tax-deficient cells, thus confirming the signal-dependent nature of these post-translational modifications (Fig. 2C). Considering that there are 14 potential sites for chymotryptic cleavage in IKK␥, our biochemical mapping data strongly suggest that Tax induces phosphorylation of IKK␥ at multiple serine residues in vivo.
Identification of Inducible Phosphoacceptors in IKK␥-To assign specific phosphoacceptors in IKK␥, we next used sitedirected mutagenesis to replace individual serine residues with alanine in the full-length protein. In this regard, sequences in the NH 2 -terminal region of IKK␥ (amino acids 1-120) are necessary for its interaction with the IKK␤ catalytic subunit (11,32). To determine whether the IKK␤ binding domain of IKK␥ is subject to signal-dependent phosphorylation, 293T cells were programmed with expression vectors for Tax, IKK␤, and human IKK␥ containing alanine replacements at serine residues 17, 31, 43, 68, and 85. All five of these serines are conserved between mouse and human IKK␥ (10 -13). After labeling recipient cells with 32 P i , IKK␥ proteins were isolated, digested with chymotrypsin, and analyzed by two-dimensional phosphopeptide mapping. As shown in Fig. 3 (upper panels), point mutations at Ser-31 and Ser-43 in IKK␥ eliminated two of the four major phosphopeptides identified in control-mapping experiments with wild type IKK␥ (PP-1 and PP-2, respectively). In sharp contrast, replacement of either Ser-17, Ser-68, or Ser-85 with alanine yielded a phosphopeptide fingerprint that was indistinguishable from the pattern for wild type IKK␥ following chymotrypsin digestion (Fig. 3, lower panels). These in vivo results indicate that IKK␤ mediates inducible phosphorylation of IKK␥ at Ser-31 and Ser-43, whereas other local serine residues in the IKK␤ binding domain of IKK␥ are spared from modification.
In contrast to the NH 2 -terminal region of IKK␥, sequences in the COOH-terminal half of the protein (amino acids 320 -419) appear to couple IKK to upstream signal (11,12). In the case of human IKK␥, this region contains potential serine phosphoacceptors at positions 341, 346, 350, 364, 376, 377, 383, and 387. Based on their corresponding phosphopeptide fingerprints, mutations affecting the first four members of this set did not significantly change the phosphorylation status of IKK␥ in Tax-expressing cells (data not shown). Accordingly, we engineered alanine replacements into IKK␥ at the cluster of serines positioned between amino acids 370 and 390 (denoted as S376A, S377A, S383A, and S387A). As shown in Fig. 4 (top panels), point mutations at either Ser-383 or Ser-387 had no significant effect on distribution of the four major phosphopeptides identified in studies with wild type IKK␥. However, in vivo phosphorylation of S376A in Tax-expressing cells yielded a pattern of chymotryptic peptides that was clearly distinct from the wild type IKK␥ fingerprint, specifically with respect to PP-3 and PP-4 (Fig. 4, bottom panels). Of particular interest, the mutation of Ser-376 was associated with the appearance of two new phosphopeptides (designated PP-3Ј and PP-4Ј), suggesting the presence of compensatory phosphoacceptors in the same chymotryptic fragments. Consistent with this hypothesis, alanine replacement of Ser-377 in mutant S376A eliminated the appearance of PP-3Ј and PP-4Ј (mutant S376/377A), whereas the disruption of Ser-377 alone was without affect (mutant S377A). These data indicate that Ser-376 is the primary target for phosphorylation in the COOH-terminal region of IKK␥ and the vicinal serine at position 377 serves a compensatory role. Taken together with data shown in Fig. 3, our mapping experiments strongly suggest that Ser-31, Ser-43, and Ser-376 are the major phosphoacceptor sites in human IKK␥.
To confirm this interpretation, we next engineered expression vectors for IKK␥ containing serial mutations at the identified serine phosphoacceptors and introduced them into 293T cell transfectants along with Tax and IKK␤ effector plasmids. IKK␤ immunocomplexes were then isolated from 32 P i -labeled recipients and analyzed for phosphoprotein content. As shown in Fig. 5, simultaneous disruption of Ser-31 and Ser-43 in the NH 2 -terminal region of IKK␥ led to a significant reduction in subunit-specific phosphorylation (mutant S31/43A, lane 3). Simultaneous disruption of Ser-376 and compensatory Ser-377 in the COOH-terminal half of IKK␥ also attenuated 32 P i incorporation into the protein (mutant S376/377A, lane 4). Consistent with the biochemical phenotype of these double point mutants, the defect in substrate-radiolabeling efficiency was further evident (ϳ19% wild type control) when all four of the identified serine targets in IKK␥ were disrupted in combination by sitedirected mutagenesis (lane 5). Observed changes in the phosphorylation status of IKK␥ were not attributed to fluctuations in ectopic protein expression, because each of the phosphorylation-defective mutants of IKK␥ were comparably expressed in metabolically radiolabeled cells (Fig. 5, bottom panels). More-

FIG. 3. Identification of phosphoacceptors in the IKK␤ binding domain of IKK␥.
Human 293T cells were transfected with vectors for Tax (0.3 g), FLAG-tagged IKK␤ (60 ng), and T7tagged IKK␥ (60 ng) containing alanine replacements at serine residues 17, 31, 43, 68, and 85 (mutants S17A, S31A, S43A, S68A, and S85A, respectively). Cells were radiolabeled with [ 32 P]orthophosphate for 4 h. Ectopic IKK proteins derived from Tax-expressing cells were immunoprecipitated with anti-FLAG antibodies and analyzed by twodimensional phosphopeptide mapping as described in the legend for Fig. 2. Immunoblotting analysis confirmed that wild type and mutant IKK␥ proteins were comparably expressed in transfected cells (data not shown). over, these results could not be attributed to changes in the overall structural integrity of IKK␥, because each of the phosphorylation-defective mutants retained the capacity to form stable complexes with IKK␤ (Fig. 5, bottom panels). Coupled with the two-dimensional phosphopeptide mapping data (Figs.  3 and 4), we conclude that Ser-31, Ser-43, and Ser-377 in human IKK␥ are the three major targets for signal-dependent phosphorylation in Tax-expressing cells.
Tax and TNF Converge on the Same Set of IKK␥ Phosphoacceptors-Similar to the Tax oncoprotein of HTLV1, cellular stimulation with the proinflammatory cytokine TNF leads to rapid phosphorylation of both IKK␤ and IKK␥ (Fig. 1A) (9,(17)(18)(19). We have previously shown that Tax mediates these IKK subunit modifications by interacting directly with IKK␥, which functions as an essential molecular adaptor for Tax in HTLV1-infected cells (19,33). In contrast, TNF acts indirectly on IKK via binding to TNF-R1, its cognate cell-surface receptor. In turn, IKK and other accessory proteins are recruited to the cytoplasmic tail of TNF-R1, resulting in catalytic activation of IKK␤ (34,35). Given these divergent signaling mechanisms, we reasoned that Tax and TNF might target distinct phosphoacceptors in IKK␥.
To explore this possibility, we transfected 293T cells with expression vectors for IKK␤, IKK␥, and either TNF-R1 or Tax. We then conducted comparative phosphopeptide mapping studies on IKK␥ derived from metabolically radiolabeled recipients. As shown in Fig. 6 (left panel), each of the four major phosphopeptides characteristics of wild type IKK␥ (PP-1, PP-2, PP-3, and PP-4) were identified in control mapping experiments with Tax-expressing cells. Despite the distinct mechanism by which TNF activates IKK signaling, the pattern of IKK␥ subunit phosphorylation induced by TNF-R1 was indistinguishable from the phosphopeptide fingerprint of IKK␥ derived from Tax-expressing cells (Fig. 6, right panel). The complete overlap between IKK␥ phosphoacceptors targeted in the Tax versus TNF-R1 backgrounds was confirmed in subsequent mixing experiments (data not shown). These biochemical data along with results shown in Fig. 1 strongly suggest that IKK␤-FIG. 6. Tax and TNF signaling pathways target identical phosphorylation sites in IKK␥. Human 293T cells (2 ϫ 10 6 ) were transfected with vectors for FLAG-tagged IKK␤ (60 ng), IKK␥ (60 ng), and either Tax or TNF-R1 (300 ng). The IKK␥ expression vectors contained an alanine-to-serine substitution within the T7 epitope tag (see Fig. 5 legend). Cells were radiolabeled with [ 32 P]orthophosphate for 4 h. Ectopic IKK complexes were isolated using anti-FLAG antibodies and fractionated by SDS-PAGE. Radiolabeled IKK␥ was analyzed by twodimensional phosphopeptide mapping as described in the legend to Fig. 2.

FIG. 4. Identification of phosphoacceptors in the COOH-terminal half in IKK␥.
Human 293T cells were programmed with expression vectors for FLAG-tagged IKK␤ (60 ng), Tax (0.3 g), and T7-tagged forms of IKK␥ (60 ng) containing alanine replacements at serine residues 376, 377, 383, or 387 (mutants S376A, S377A, S383A, and Ser387A, respectively). Control experiments were performed under identical transfection conditions with expression vectors for WT IKK␥ or a variant containing alanine rather than serine at positions 376 and 377 (mutant S376/377A, lower right panel). Transfected cells were radiolabeled with [ 32 P]orthophosphate for 4 h. IKK␥ proteins were isolated from Tax-expressing cells and analyzed by two-dimensional phosphopeptide mapping as described in the legend for Fig. 2.   FIG. 5. Ser-31, Ser-43, and Ser-376 are the major phosphorylation sites in IKK␥. Human 293T cells were transfected with vectors for Tax (0.3 g), FLAG-tagged IKK␤ (60 ng), and the indicated T7tagged forms of IKK␥ (60 ng) containing alanine replacements at Ser-31 and Ser-43 (mutant S31/43A), Ser-376 and Ser-377 (mutant S376/ 377A), or all four of these sites (All). To eliminate phosphorylation in the T7 tag fused to IKK␥ (MASMTGGQQ), the serine at position 3 was replaced with alanine in all of the constructs. Cells were radiolabeled with [ 32 P]orthophosphate for 4 h. Ectopic IKK complexes were isolated using anti-FLAG antibodies, fractionated by SDS-PAGE, and analyzed autoradiography (top panel). Relative protein levels were determined by immunoblotting with antibodies specific for the indicated species (lower panels). directed phosphorylation of IKK␥ at Ser-31, Ser-43, and Ser-376 is an integral step in the TNF/NF-B signaling axis.
Phosphorylation of IKK␥ Is Dependent on Its Zinc Finger Domain-Despite the apparent capacity of IKK␤ to phosphorylate both IB␣ and IKK␥, these two substrates share no obvious signature sequences that might underlie a common mechanism for IKK␤ targeting specificity. However, recent studies (36,37) have indicated that the COOH-terminal zinc finger (ZF) domain of IKK␥ is required for NF-B activation by a subset of IKK-inducing agents. This ZF motif lies within the final 30 residues of the full-length protein (419 amino acids) and contains no serines, which serve as the primary phosphoacceptors sites in IKK␥ (Fig. 2B).
To determine whether the ZF motif is important for IKK␤directed phosphorylation of IKK␥, we metabolically radiolabeled 293T cells following transfection with expression vectors for Tax, IKK␤, and either wild type IKK␥ (IKK␥.WT, amino acids 1-419) or a minimal deletion mutant of IKK␥ lacking the ZF motif (IKK␥.ZF, amino acids 1-396). Ectopic IKK complexes were immunopurified from recipient cells, fractionated by SDS-PAGE, and then monitored for phosphoprotein content. As shown in Fig. 7A (upper panel), Tax readily stimulated phosphorylation of IKK␥.WT in the presence of IKK␤ (lanes 1 and 2). Under the same Tax-inducing conditions, the phosphorylation of IKK␥.ZF was significantly attenuated (lanes 3 and 4). The ZF mutation had no detectable effect on Tax-induced phosphorylation of IKK␤ (Fig. 7A, upper panel), IB kinase activity (Fig. 7A, lower panel), IKK protein expression (Fig. 7B, upper  panel), or the formation of Tax⅐IKK complexes (Fig. 7B, lower  panel). These metabolic radiolabeling experiments indicate that the ZF motif is required for efficient signal-dependent phosphorylation of IKK␥.

DISCUSSION
In quiescent cells, nuclear translocation of NF-B is prevented by its association with labile cytoplasmic inhibitors such as IB␣ (38). The inhibitory action of IB␣ on NF-B is subject to regulation by a signal-dependent mechanism involving three key phosphorylation steps. In the first step, the IKK␤ catalytic subunit of a multicomponent IB kinase is activated via phosphorylation of T loop regulatory sequences positioned in its catalytic domain (9). In the second step, activated IKK␤ phosphorylates amino-terminal serines in IB␣, which marks the cytoplasmic inhibitor for degradation and enables NF-B to enter the nucleus (39). In turn, NF-B stimulates the expression of multiple genes involved in immunity and inflammation (1,3). In the third step, IKK␤ is phosphorylated at serines clustered in its COOH-terminal region, which apparently serves to attenuate IB␣ kinase activity (9). These findings highlight the crucial regulatory role that phosphorylation plays at three distinct levels in the process of NF-B signal transduction.
Emerging evidence indicates that IKK␥ is the target for a fourth integral phosphorylation step in this process (9, 18 -20). In our studies, we have found that IKK␥ is chronically phosphorylated in cells expressing the HTLV1 Tax oncoprotein, which interfaces directly with the IB kinase complex ( Fig. 2A) (20). As well, this mechanism is operative in Tax-deficient cells following exposure to three different mediators of inflammation, including TNF, IL-1, and bacterial LPS (Fig. 1A). We have also found that IKK␥ is phosphorylated in vitro in the presence of activated IKK␤, which is the key enzyme subunit responsible for IB␣ phosphorylation and degradation (19). Consistent with these in vitro results, metabolic radiolabeling experiments with mutant MEFs indicate that IKK␤ is required for IKK␥ subunit phosphorylation (Fig. 1B). Taken together, these data indicate an important biochemical interplay between IKK␤ and IKK␥ that may contribute to the control of NF-B signaling during the cellular response to immune and inflammatory mediators.
A prerequisite for understanding the functional consequences of IKK␥ subunit phosphorylation is to identify the relevant acceptor sites. To achieve this experimental objective, we first conducted phosphopeptide mapping studies of IKK␥ in metabolically radiolabeled cells expressing the Tax oncoprotein of HTLV1. Unlike the transient action of TNF and other cytokines, Tax stimulates chronic phosphorylation of IKK␤, enabeling us to capture a stably modified form of IKK␥ (Fig. 2A). These biochemical studies indicated that Tax induces the phosphorylation of human IKK␥ at Ser-31, Ser-43, and Ser-376 (Figs. [3][4][5]. All three of these serine phosphoacceptors are found in murine IKK␥ as well (10). Similar results were obtained in phosphopeptide mapping experiments conducted with the type 1 TNF receptor (Fig. 6), which is known to activate IKK and NF-B when expressed ectopically in mammalian cell transfectants (40). In this regard, Prajapati et al. (17) have reported in vitro evidence for the phosphorylation of murine IKK␥ at Ser-369, which corresponds to Ser-376 of human IKK␥. However, results obtained from our in vivo mapping studies did not correlate with other putative phosphoacceptors identified by Prajapati et al. (17), presumably reflecting significant in vitro versus in vivo differences in IKK␥ substrate targeting by IKK␤.
The three major phosphoacceptors identified in our studies reside in distal regions of IKK␥. Whereas Ser-31 and Ser-43 are positioned in the IKK␤ binding domain of human IKK␥, Ser-376 lies downstream in the COOH-terminal domain of IKK␥ that is required for signal-dependent activation of IKK␤ (11,12). Thus far, we have no compelling evidence that inducible phosphorylation of these IKK␥ domains is important for their assigned functions. Specifically, immunoprecipitation experiments indicated that phosphorylation-defective mutants of IKK␥ lacking Ser-31 and Ser-43 retain the capacity to form stable complexes with IKK␤ (Fig. 5). Moreover, IKK␤ is efficiently phosphorylated in cells expressing mutants of IKK␥ that lack Ser-376 and Ser-377 (Fig. 5), suggesting that IKK␤ activity is unaffected as well. Regarding this latter observation, prior transfection studies (18) have revealed modest increases in IB kinase activity when IKK␤ is coexpressed with murine IKK␥ containing alanine substitutions at Ser-369 and Ser-375 (equivalent to human Ser-376 and Ser-383). However, we have been unable to detect significant changes in the wild type pattern of IB kinase activity in preliminary studies with the phosphorylation-defect mutants of human IKK␥ described in Fig. 5 (data not shown). Accordingly, it may be difficult to ascribe a functional phenotype to these phosphorylation-defective mutants of IKK␥ solely on the basis of in vitro kinase assays.
An unexpected finding to emerge from our studies is that in vivo phosphorylation of human IKK␥ is contingent upon the presence of its COOH-terminal ZF domain (Fig. 7A). We suspect a role for the ZF domain in substrate recognition, because its removal from IKK␥ had no detectable effect on IKK␤-mediated phosphorylation of IB␣ (Fig. 7A). Consistent with this proposal, ZF-deleted forms of IKK␥ are impaired for phosphorylation by a constitutively active mutant of IKK␤ (data not shown). The observed mechanistic relationship between the ZF motif and IKK␥ phosphorylation may help explain why mutations in the ZF motif of IKK␥ interfere with the regulated action of NF-B under certain stimulatory conditions (36,37). This relationship may also have clinical implications. Specifically, mutations in the X-linked gene encoding IKK␥ can cause anhidrotic ectodermal dysplasia with severe immunodeficiency (15). In the majority of affected patients, the IKK␥ gene has missense mutations or small deletions that selectively target the ZF domain of the protein (amino acids 390 -419) (15). Although the underlying signaling defects remain unclear, our data raise the intriguing possibility that some of these ZF mutations will interfere with inducible phosphorylation of IKK␥ in the disease state.
In summary, we demonstrate here that IKK␤ is required for in vivo phosphorylation of IKK␥ in intact cells. Taken together with our prior in vitro data (19), we conclude that IKK␤ mediates IKK␥ phosphorylation under physiologic signaling conditions. We have also conducted phosphopeptide mapping studies to monitor signal-dependent changes in the phosphorylation status of IKK␥ in metabolically radiolabeled cells. These biochemical experiments indicate that both Tax and TNF induce phosphorylation of human IKK␥ at Ser-31, Ser-43, and Ser-376, which are fully conserved in murine IKK␥. Signal-dependent phosphorylation of these target sites in IKK␥ is contingent upon the presence of its COOH-terminal zinc finger domain, a frequent target for mutations in anhidrotic ectodermal dyspla-sia with severe immunodeficiency disease. This requirement suggests an important regulatory role for IKK␤-mediated phosphorylation of IKK␥ in the immune system. Resolution of this key question awaits more detailed gene-targeting studies with the phosphorylation-defective mutants of IKK␥ identified in the present report.