Signal-induced Ubiquitination of IκB Kinase-β

Initiation of the genetic programs for inflammation and immunity involves nuclear mobilization of transcription factor NF-κB. This signal-dependent process is controlled in part by the β-catalytic subunit of IκB kinase (IKKβ), which marks IκBα and other cytoplasmic inhibitors of NF-κB for proteolytic destruction. The catalytic activity of IKKβ is stimulated by pathologic and physiologic inducers of NF-κB, such as the Tax oncoprotein and proinflammatory cytokines. We now report evidence that these NF-κB inducers target IKKβ for conjugation to ubiquitin (Ub) in mammalian cells. The apparent molecular size of modified IKKβ is compatible with monoubiquitination rather than attachment of a multimeric Ub chain. The modification is contingent upon signal-induced phosphorylation of the activation T loop in IKKβ at Ser-177/Ser-181. The formation of IKKβ-Ub conjugates is disrupted in cells expressing YopJ, a Ub-like protein protease that interferes with the NF-κB signaling pathway. These findings indicate an important mechanistic link between phosphorylation, ubiquitination, and the biologic action of IKKβ.

The inducible transcription factor NF-B is biochemically coupled to cell-surface members of the tumor necrosis factor (TNF) 1 receptor, Toll-like receptor, and immunoglobulin superfamilies (1). NF-B is persistently activated in cells expressing the Tax protein of human T-cell leukemia virus type 1, which has potent oncogenic properties (2). This deregulated pattern of NF-B activity also underlies acute and chronic inflammatory diseases (1). Nuclear translocation of NF-B is controlled by an inducible multicomponent protein kinase, termed IKK, which targets IB␣ and other cytoplasmic inhibitors of NF-B for proteolytic destruction (3). IKK contains two catalytic subunits, designated IKK␣ and IKK␤, as well as a regulatory subunit called IKK␥ (3). IKK␤ and IKK␥ are essential for proteolytic inactivation of IB␣, whereas IKK␣ mediates NF-B subunit processing via an IKK␥-independent mechanism (4).
The catalytic activity of IKK␤ is stimulated by signals that trigger its phosphorylation at Ser-177/Ser-181, such as TNF-␣ (5). These phosphoacceptors lie in a region of the catalytic domain that shares homology with regulatory "T loop" sequences found in mitogen-activated protein kinases and their upstream activators (5). In contrast to its transient pattern of phosphorylation and activation in TNF-stimulated cells, IKK␤ is chronically phosphorylated and activated in cells expressing the Tax oncoprotein (6). Tax-induced activation of IKK␤ is blocked by YopJ (6), a cysteine protease that removes ubiquitin (Ub)-related modifiers from target proteins in mammalian cells (7). This finding with YopJ raised the possibility that IKK␤ might be subject to signal-dependent ubiquitination. In keeping with this possibility, previous studies with partially purified kinase complexes suggested that IKK is conjugated to Ub in vitro (8). However, these pioneering in vitro experiments were conducted prior to molecular cloning of individual IKK subunits (3).
We now demonstrate that IKK␤ is conjugated to Ub in vivo following engagement of either the Tax or TNF signaling pathway. Signal-induced ubiquitination of IKK␤ is disrupted by wild type but not protease-deficient versions of YopJ. Formation of IKK␤-Ub conjugates is dependent on the presence of IKK␥, which is essential for IKK␤ catalytic activity. The apparent molecular size of IKK␤-Ub conjugates is consistent with attachment of a single Ub molecule to IKK␤, precluding efficient recognition by the 26 S proteasome (9). Loss-of-function mutations in IKK␤ that block phosphorylation at Ser-177/Ser-181 prevent its ubiquitination. In sharp contrast, the replacement of these sites with phosphomimetic amino acids yields a gain-of-function mutant that is constitutively ubiquitinated in the absence of an NF-B agonist. We conclude that T loop phosphorylation at Ser-177/Ser-181 generates a conditional Ubtargeting signal in IKK␤. This contingent post-translational mechanism may be applicable to other inducible enzymes under T loop control that lie outside of the NF-B signaling pathway.

EXPERIMENTAL PROCEDURES
Reagents-Polyclonal anti-IKK (H-470, FL-419) and monoclonal anti-TNF-R1 (TNF-R1) antibodies (H-5) were purchased from Santa Cruz Biotechnology. Rat monoclonal anti-HA antibodies were obtained from Roche Applied Science. Monoclonal antibodies specific for FLAG and T7 epitope tags were purchased from Sigma and Novagen, respectively. Rabbit anti-Tax antibodies were provided by Dr. Bryan Cullen (Duke University). Phospho-specific antibodies that recognize IKK␤ following modification at Ser-181 were obtained from Cell Signaling Technology. Expression vectors for Tax, YopJ, Ub, and IKK subunits have been described previously (6,7,11). The expression vector for polyhistidine (His)-tagged IKK␤ was constructed using pCruz His (Santa Cruz Biotechnology). The expression vector for TNF-R1 was provided by Dr. Brian Seed (Harvard University).
Subcellular Fractionation-Human 293T cells were cultured as described and transfected using the calcium phosphate method (6). Whole cell extracts were prepared in radioimmune precipitation buffer (150 mM NaCl, 10 mM sodium phosphate pH 7.2, 0.1% SDS, 0.5% sodium deoxycholate, 1% Nonidet P-40) containing 5 mM N-ethylmaleimide. Cytoplasmic extracts were prepared by detergent lysis in the presence of phosphatase/protease inhibitors (6) and then equilibrated in ELB buffer (50 mM HEPES, pH 7.4, 250 mM NaCl, 5 mM EDTA, 0.1% Nonidet P-40) prior to immunoprecipitation. Immmunocomplexes were washed sequentially with ELB buffer containing 2 M urea and then with radioimmune precipitation buffer. Bound proteins were fractionated by SDS-PAGE, transferred to polyvinylidene difluoride membranes, and detected on immunoblots as described previously (6).
Purification of IKK␤-Ub Conjugates-To perform nickel-chelate affinity chromatography under denaturing conditions (11), cytoplasmic extracts were equilibrated in 3 M guanidine hydrochloride (GuHCl), 150 mM NaCl, and 10 mM imidazole. Alternatively, cells were lysed directly with 6 M GuHCl buffered in 100 mM sodium phosphate, 10 mM Tris, pH 8, 300 mM NaCl, and 10 mM imidazole. Extracts were mixed with Ni-NTA-agarose beads (Qiagen) for 2-4 h at room temperature. Resins were washed with 6 M GuHCl or 8 M urea containing 20 mM imidazole. Bound proteins were eluted by boiling in the presence of 1% SDS and 100 mM dithiothreitol, fractionated by SDS-PAGE, and subjected to immunoblotting as described previously (6).

Signal-induced Ubiquitination of IKK␤ in Vivo-
The ␤-catalytic subunit of IKK is required for proteolytic inactivation of IB␣, a principle cytoplasmic inhibitor of NF-B (3). To determine whether IKK␤ is ubiquitinated in vivo, we programmed 293T cells with expression vectors for Tax, IKK␤, IKK␥, and a HA-tagged form of human Ub. IKK␤ immunocomplexes were prepared from recipient cells and then probed on immunoblots for their HA-Ub content. As shown in Fig. 1A (top), we were unable to detect HA-Ub in IKK␤ immunocomplexes derived from control cells lacking either Tax or HA-Ub (lanes 1 and 2). However, Tax induced significant levels of IKK␤ ubiquitination  in transfectants expressing tagged Ub (lanes 3 and 4). This signal-dependent response could not be attributed to Tax-mediated stabilization of IKK subunits, because the steady-state expression levels of IKK␤ and IKK␥ were comparable in all of the transfectants (lower panels). These data strongly suggested that IKK␤ is subject to Ub conjugation via a signal-dependent mechanism.
To confirm this finding, we used nickel-chelate affinity (Ni-NTA) chromatography to purify IKK␤-Ub conjugates under stringent denaturing conditions (11). For these experiments, 293T cells were transfected with expression vectors for Tax, IKK␥, HA-Ub, and polyhistidine (His)-tagged IKK␤. Histagged IKK␤ was purified from recipient cells on Ni-NTA beads in the presence of 3 M GuHCl, eluted, and probed on immunoblots for HA-Ub. We found no evidence for the formation of immunoreactive IKK␤-Ub conjugates in control cells lacking either HA-Ub or Tax (Fig. 1B, top panel, lanes 1-3). However, in keeping with Fig. 1A, Ub-conjugated IKK␤ was readily detected in extracts derived from cells coexpressing Tax and HA-Ub (lane 4). Similar results were obtained in complementary experiments with expression vectors for His-Ub rather than His-IKK␤, which enabled us to probe the IKK␤ content in affinity-purified Ub conjugates under strong denaturing conditions (Fig. 1C). Based on these results from three independent experimental approaches, we conclude that IKK␤ is subject to signal-dependent ubiquitination in transfected mammalian cells. Thus far, we have been unable to detect this modification in untransfected cells, presumably because of limited expression of the relevant IKK␤ species at the endogenous protein level.
Specificity and Extent of IKK␤ Ubiquitination-Association of IKK␤ with IKK␥ is essential for proteolytic inactivation of IB␣ and nuclear translocation of NF-B (3). To determine whether IKK␥ is required for signal-induced ubiquitination of IKK␤, we conducted experiments with 293T cells expressing Tax, His-Ub, and IKK␤ in the presence or absence of IKK␥. To provide stringent controls for specificity, parallel transfections were performed with point mutants of Tax that are selectively defective for either cAMP-response element-binding protein/ activating transcription factor (Tax-M47) or NF-B (Tax-M22) activation (10). Ub conjugates were purified from cytoplasmic extracts by Ni-NTA chromatography, resolved by SDS-PAGE, and then probed on immunoblots for their IKK␤ content. As shown in Fig. 2A (top), wild type Tax failed to induce IKK␤ ubiquitination in the absence of ectopic IKK␥ (lanes 1 and 2). Tax-M47 (lane 6) but not with Tax-M22 (lane 5), consistent with their differing capacities to activate IKK␤ and NF-B (6, 10). All of the ectopic proteins were efficiently expressed ( Fig. 2A, lower panels). We conclude that IKK␥ is required for signal-dependent ubiquitination of IKK␤.

Provision of IKK␥ in trans rescued this modification step in Tax-expressing cells (lane 4). Formation of IKK␤-Ub conjugates was also observed in experiments with
Proteasome-mediated degradation of a substrate requires its

FIG. 4. YopJ disrupts signal-induced ubiquitination of IKK␤.
Human 293T cells were transfected with vectors for IKK␤ (FLAGtagged, 50 ng), Tax (0.3 g), IKK␥ (T7-tagged, 40 ng), Ub (HA-tagged, 2 g), and either wild type YopJ or a mutant containing an alanine substitution at Cys-172 (FLAG-tagged, 50 ng each). IKK complexes were immunopurified from cytoplasmic extracts with anti-T7 antibodies and immunoblotted with HA-specific antibodies (top panel). Cytoplasmic levels of phosphorylated (P) IKK␤, total IKK␤, IKK␥, Tax, and YopJ were determined by immunoblotting (lower five panels). attachment to a multimeric Ub chain containing at least four covalently linked Ub molecules (ϳ8 kDa each) (9). The presence of this degradation marker is typically revealed as a ladder or a smear on SDS-PAGE gels, which we failed to detect in our experiments with Ub-conjugated IKK␤ (Fig. 1). In keeping with this observation, Tax has no significant effect on the steadystate level of IKK␤ (12). To estimate the molecular size of IKK␤-Ub conjugates under conditions that minimize Ub editing by cellular isopeptidases, proteins were extracted from 293T cells by direct lysis in 6 M GuHCl (11). Ub-conjugated proteins were purified by Ni-NTA chromatography, resolved on an SDS-PAGE gel in conjunction with molecular weight standards, and probed on immunoblots for their IKK␤ content (Fig.  2B). An approximate molecular size of 95 kDa was determined for the unmodified form of IKK␤ derived from Tax-deficient cells (lane 1), whereas purified IKK␤-Ub conjugates derived from Tax-expressing cells migrated primarily as a single ϳ105-kDa species (lane 4). The observed electrophoretic mobility shift was compatible with the attachment of a single Histagged Ub molecule to IKK␤ rather than a multimeric Ub chain. We consider this experimental result to be significant, because emerging studies indicate that monoubiquitin regulates the biologic activity of target substrates via a proteaseindependent mechanism (13).
Role of T Loop Phosphorylation in IKK␤ Ubiquitination-Tax-induced activation of IKK␤ is contingent upon its phosphorylation at Ser-177 and Ser-181 (6). To explore the role of Ser-177/Ser-181 in IKK␤ ubiquitination, 293T cells were transfected with vectors for a mutant of IKK␤ containing alanine replacements at these two sites (IKK␤.SA), IKK␥, Tax, and His-Ub. We then isolated His-Ub conjugates from recipient cells by chromatography on Ni-NTA beads, fractionated them by SDS-PAGE, and probed the resolved proteins on immunoblots for the presence of IKK␤. As shown in Fig. 3A (top), mutations that disrupt phosphorylation of Ser-177/Ser-181 in IKK␤ blocked its ubiquitination in the presence of Tax (lanes 3 and 4). Similar results were obtained with a kinase-dead mutant of IKK␤ that is defective for ATP-binding (IKK␤.KM, lanes 5 and 6). IKK␤ was constitutively ubiquitinated in Tax-deficient cells when Ser-177 and Ser-181 were both replaced with the phosphomimetic glutamic acid (IKK␤.SE, lane 7), which generates a constitutively active IB kinase (6). We conclude that Tax-induced phosphorylation and ubiquitination of IKK␤ are biochemically linked.
Phosphorylation of IKK␤ at Ser-177/Ser-181 is also required for its activation by proinflammatory mediators such as TNF (5). To explore whether proinflammatory signaling pathways for NF-B activation lead to IKK␤ ubiquitination, we programmed 293T cells with expression vectors for His-Ub, IKK␤, IKK␥, and TNF-R1 rather than Tax. As shown in Fig. 3B (top), TNF-R1 induced significant levels of IKK␤ ubiquitination relative to control cells lacking the ectopic receptor (lanes 1 and 2). However, this signal-dependent response was disrupted in cells coexpressing TNF-R1 and IKK␤.SA (lanes 3 and 4). Immunoblotting experiments confirmed that enforced expression of TNF-R1 stimulated IKK␤ phosphorylation and that IKK␤.SA was efficiently expressed (Fig. 3B, lower panels). We conclude that both Tax and TNF-R1 induce IKK␤ ubiquitination via a mechanism involving T loop phosphorylation of this catalytic subunit at Ser-177/Ser-181.
Attachment of Ub to IKK␤ Is Disrupted by YopJ-Tax-induced activation of endogenous IKK␤ is blocked in cells expressing the Yersinia virulence factor YopJ (6). Recent studies indicate that YopJ removes Ub-related modifiers from target proteins in mammalian cells (7). To determine whether YopJ interferes with stable formation of IKK␤-Ub complexes, 293T cells were transfected with expression vectors for Tax, HA-Ub, IKK␤, IKK␥, and YopJ. IKK complexes were immunopurified from recipient cells, resolved by SDS-PAGE, and then probed for the presence of HA-Ub on immunoblots. As expected, Tax stimulated ubiquitination of IKK␤ in YopJ-deficient cells (Fig.  4, top panel, lanes 1 and 2). Coexpression with YopJ completely blocked this signal-dependent modification, which was accompanied by diminished T loop phosphorylation of IKK␤ (lane 3, top two panels). In contrast, a protease-deficient mutant of YopJ containing a Cys 3 Ala substitution in its catalytic triad (7) had no significant effect on signal-induced ubiquitination of IKK␤ (top panel, lane 4). All of the ectopic proteins were efficiently expressed, including the protease-deficient mutant of YopJ (Fig. 4, lower panels). Thus, the capacity of YopJ to disrupt signal-induced ubiquitination of IKK␤ is dependent on its protease function (7). It remains unclear how this protease function also impinges on the phosphorylation status of IKK␤. One possibility is that attachment of Ub to IKK␤ protects the activated T loop from cellular phosphatases, leading to sustained IB kinase activity.
In summary, we have found that IKK␤ is ubiquitinated in response to engagement of either the Tax or TNF signaling pathway. The apparent molecular size of Ub-modified IKK␤ is most compatible with the attachment of a single Ub tag (Fig.  2B), precluding efficient recognition of the conjugate by the 26 S proteasome (9,13). In keeping with prior in vitro studies (8), several lines in vivo evidence indicate a role for Ub in the regulation of IKK␤ catalytic activity. First, ubiquitination of IKK␤ is contingent upon the presence of IKK␥ (Fig. 2A), a regulatory subunit that is essential for proper functional control of IKK␤ (3,4). Second, the modification is disrupted by YopJ (Fig. 4), which interferes with signal-dependent activation of endogenous IKK␤ (6). Third, the formation of IKK␤-Ub conjugates is triggered by phosphorylation of IKK␤ at Ser-177/ Ser-181 (Fig. 3), a prerequisite for its activation by either Tax or TNF (5,6). We propose that phosphorylation at Ser-177/Ser-181 generates a substrate recognition motif, permitting the Ub-conjugating machinery to dock with and monoubiquitinate IKK␤ (14). In turn, this modification regulates the biologic action of IKK␤ via a proteasome-independent mechanism, perhaps involving the intervention of an IKK␤-specific Ub receptor (15). Resolution of this model awaits assignment of the Ub acceptor sites and functional studies with ubiquitination-defective mutants of IKK␤.