Site-specific Monoubiquitination of IκB Kinase IKKβ Regulates Its Phosphorylation and Persistent Activation*

Transcription factor NF-κB governs the expression of multiple genes involved in cell growth, immunity, and inflammation. Nuclear translocation of NF-κB is regulated from the cytoplasm by IκB kinase-β (IKKβ), which earmarks inhibitors of NF-κB for polyubiquination and proteasome-mediated degradation. Activation of IKKβ is contingent upon signal-induced phosphorylation of its T loop at Ser-177/Ser-181. T loop phosphorylation also renders IKKβ a substrate for monoubiquitination in cells exposed to chronic activating cues, such as the Tax oncoprotein or sustained signaling through proinflammatory cytokine receptors. Here we provide evidence that the T loop-proximal residue Lys-163 in IKKβ serves as a major site for signal-induced monoubiquitination with significant regulatory potential. Conservative replacement of Lys-163 with Arg yielded a monoubiquitination-defective mutant of IKKβ that retains kinase activity in Tax-expressing cells but is impaired for activation mediated by chronic signaling from the type 1 receptor for tumor necrosis factor-α. Phosphopeptide mapping experiments revealed that the Lys-163 → Arg mutation also interferes with proper in vivo but not in vitro phosphorylation of cytokine-responsive serine residues located in the distal C-terminal region of IKKβ. Taken together, these data indicate that chronic phosphorylation of IKKβ at Ser-177/Ser-181 leads to monoubiquitin attachment at nearby Lys-163, which in turn modulates the phosphorylation status of IKKβ at select C-terminal serines. This mechanism for post-translational cross-talk may play an important role in the control of IKKβ signaling during chronic inflammation.

Initiation of the genetic programs for inflammation and immunity is mediated in part by the inducible transcription factor NF-B (1). In quiescent cells, NF-B is sequestered in the cytoplasmic compartment by virtue of its association with the labile inhibitor IB␣ or other structurally related proteins (2). Biological inducers of NF-B include proinflammatory cytokines such as tumor necrosis factor-␣ (TNF), 4 the lipopolysaccharide component of Gram-negative bacteria, and the Tax oncoprotein of human T-cell leukemia virus type 1 (3)(4)(5). Cellular stim-ulation with these agents leads to activation of a multicomponent kinase called IKK, which phosphorylates IB␣ at Ser-32/Ser-36 (6). In turn, the phosphorylated Ser-32/Ser-36 motif docks with a specific ubiquitin (Ub)-protein ligase complex, resulting in Ub conjugation at nearby Lys-21/Lys-22 (7). Polyubiquitinated IB␣ is then degraded by the 26 S proteosome, enabling NF-B to translocate to its nuclear site of action (6,7).
The capacity of IKK to integrate a wide spectrum of activating cues has evoked significant interest in its structure, function, and signal-dependent regulation (8). The IKK holoenzyme is composed of ␣ and ␤ catalytic subunits, both of which contain kinase, leucine zipper, and helix-loop-helix domains (9). IKK␤ is primarily responsible for phosphorylation of IB␣ in vivo, whereas IKK␣ targets distinct cellular substrates in an alternative pathway for NF-B induction (10). Activation of IKK␤ by TNF, Tax, and other inducers of NF-B is triggered by phosphorylation of this kinase at Ser-177/Ser-181 (11,12). These phosphoacceptors lie in a region of the catalytic domain that shares homology with the regulatory T loop sequences found in mitogen-activated protein kinases and their upstream activators (11). The C-terminal region of IKK␤ (amino acids 670 -705) harbors a second set of TNFinducible phosphoacceptors whose precise function remains unclear relative to Ser-177/Ser-181 (11). Prior studies suggest that modification of the T loop and C-terminal phosphoacceptors is mediated by IKK␤ itself (11,13). Signal-induced phosphorylation and activation of IKK␤ are critically dependent upon its association with a noncatalytic subunit termed IKK␥/Nemo (9). Activated IKK␤ also phosphorylates IKK␥, which may serve to amplify IKK signaling (12,14,15).
More recent studies indicate that IKK␤ and IKK␥ are substrates not only for phosphorylation but also for ubiquitination, although to different extents. Specifically, IKK␥ is polyubiquitinated during the cellular response to microbial products, proinflammatory cytokines, antigen receptor agonists, and genotoxic stress (16). This modification is dependent on the C-terminal zinc finger domain of IKK␥, a frequent target for mutations that can impair NF-B signaling and cause human immunodeficiency disease (17,18). In contrast to IKK␥, IKK␤ is conjugated to a single Ub unit under conditions of chronic stimulation via the Tax oncoprotein or the type 1 receptor for TNF (TNF-R1) (19,20). Monoubiquitin attachment is contingent upon T loop phosphorylation at Ser-177/Ser-181 and disrupted in cells expressing YopJ, a Ub-like protein protease that interferes with IKK/NF-B signaling (19). However, the underlying targeting mechanism and regulatory potential of IKK␤ monoubiquitination remain unknown.
In this study, we provide genetic and biochemical evidence that chronic phosphorylation of IKK␤ at Ser-177/Ser-181 leads to ligation of monoubiquitin at the T loop-proximal residue Lys-163. Replacement of Lys-163 with Arg yields a ubiquitination-defective mutant that is resistant to chronic activation mediated by TNF-R1. This functional defect is not apparent in cells expressing the Tax oncoprotein, suggesting an agonist-specific requirement for monoubiquitination. We also provide evidence that monoubiquitin attachment at Lys-163 regulates the in vivo phosphorylation status of IKK␤ at select TNF-responsive serines located in its C-terminal region (amino acids 670 -705) (11). Comparative studies of in vivo versus in vitro phosphorylated IKK␤ suggest that the Ub-dependent serines are targeted by a kinase other than IKK␤ itself. Taken together, our findings indicate that IKK␤ is regulated by a novel mechanism involving post-translational cross-talk between monoubiquitination and phosphorylation. This mechanism may play an important role in the acquisition of a deregulated IKK␤ phenotype during chronic inflammation (21).

EXPERIMENTAL PROCEDURES
Antibodies-Polyclonal antibodies specific for IKK subunits (H-470, FL-419) and IB␣ (C-21) were purchased from Santa Cruz Biotechnology. Monoclonal antibodies directed against RIP, IKK␤, and IKK␥ were obtained from BD Biosciences. Monoclonal antibodies specific for the HA (Roche Applied Science), FLAG (Sigma), and T7 (Novagen) epitope tags were obtained from the indicated commercial sources. Monoclonal anti-TNF-R1 antibodies (H-5) were purchased from Santa Cruz Biotechnology. Phosphospecific antibodies that recognize modified Ser-181 in IKK␤, Ser-376 in IKK␥, and Ser-32/Ser-36 in IB␣ were obtained from Cell Signaling Technology. Polyclonal anti-IKK␤ antibodies were provided by Nancy Rice (NCI, National Institutes of Health, Bethesda). Polyclonal anti-Tax antibodies were provided by Dr. Bryan Cullen (Duke University).
Plasmids-Expression vectors for Tax, TNF-R1, Ub, and IKK subunits have been described (19). The epitope-tagged derivative of Tax was constructed by PCR-assisted amplification with 5Ј primers that fused sequences encoding the FLAG epitope (MDYKDDDDK) in-frame with the N-terminal coding sequences of Tax. The RIP expression vector was provided by Dr. Adrian Ting (Mount Sinai School of Medicine). The full-length cDNA for IKK␤ harboring Ser 3 Ala substitutions at positions 670, 672, 675, 679, 682, 689, 692, 695, 697, and 705 was provided by Dr. Michael Karin (University of California, San Diego) (11), modified by the addition of FLAG coding sequences, and inserted into pCMV4. Site-directed mutations were introduced into the fulllength cDNA for IKK␤ using the QuikChange (Stratagene) as specified by the manufacturer and confirmed by DNA sequencing.
Immunoblotting and Kinase Assays-Cell extracts, immunocomplexes, or chromatography eluates were fractionated by SDS-PAGE, transferred to polyvinylidine difluoride (PVDF) membranes, and analyzed by immunoblotting using an enhanced chemiluminescence system (Amersham Biosciences) (12). IB␣ kinase activity was measured as described (24) in reaction mixtures containing ATP (10 M), [␥-32 P]ATP (5 Ci), and recombinant glutathione S-transferase protein fused to amino acids 1-54 of IB␣ (GST-IB; 2 g). For in vitro autophosphorylation of IKK␤, IKK␤ immunocomplexes were washed extensively with ELB containing 2 M urea to remove associated proteins and were incubated in GST-IB substrate-deficient reaction mixtures containing [␥-32 P]ATP (10 Ci) and 2.5 M ATP. Reaction products were resolved by SDS-PAGE, transferred to PVDF membranes, and analyzed by sequential autoradiography and immunoblotting.

RESULTS
Differential Ubiquitination of IKK␤ and IKK␥-In prior transfection studies, we found that Tax-induced monoubiquitination of IKK␤ is contingent upon the presence of the regulatory subunit IKK␥ (19). These data suggested that the relevant Ub-conjugating machinery can recognize and modify IKK␤ within the context of a higher order complex containing IKK␥. To test this model, we determined whether Ubconjugated forms of IKK␤ are associated with IKK␥. For these studies, epitope-tagged versions of the two IKK subunits were coexpressed with Tax in mammalian cells and then immunoprecipitated with epitopespecific antibodies. Resultant complexes were dissociated with 6 M guanidine hydrochloride and fractionated by nickel-chelate affinity (Ni-NTA) chromatography to capture His-tagged IKK␤ (19). Ni-NTA eluates were probed on immunoblots (Fig. 1A) for their content of ubiquitinated and phosphorylated IKK␤ (top and middle panels, respectively). Both of these species were readily detected in Ni-NTA eluates derived from IKK␤, IKK␥, or Tax immunoprecipitates (Fig. 1A, lanes 2, 4, and 6). Parallel fractionation studies with extracts from Tax-deficient cells confirmed the signal-dependent nature of each modification (Fig.  1A, lanes 1, 3, and 5). We conclude that IKK␥-bound forms of IKK␤ are targeted for monoubiquitination.
Emerging studies indicate that IKK␥ is polyubiquitinated during the cellular response to TNF, antigen receptor signaling, or genotoxic stress (16). Given our findings that IKK␥-bound forms of IKK␤ acquire a different Ub signature (Fig. 1A), we next investigated whether IKK␥ is conjugated to polyubiquitin or monoubiquitin in Tax-expressing cells. For these studies, extracts from transfected populations were treated with N-ethylmaleimide to disrupt IKK␤-IKK␥ complexes, and then each subunit was isolated by immunoprecipitation (19). Resultant immunocomplexes were fractionated by SDS-PAGE and probed for Ub content on immunoblots. As shown in Fig. 1B, Tax stimulated monoubiquitination of IKK␤ (lanes 1 and 2), whereas IKK␥ was polyubiquitinated in the same cellular background (lanes 3 and 4). In keeping with prior studies of TNF-treated cells (17), removal of the C-terminal zinc finger motif in IKK␥ blocked Tax-induced polyubiquitination (data not shown). Thus, ectopically expressed IKK␤ and IKK␥ are differentially modified with Ub in the presence of Tax, underscoring the subunitspecific nature of the monoubiquitination pathway under investigation.
To determine whether these distinct patterns of subunit modification are recapitulated at physiologic substrate levels, we next monitored Tax-induced ubiquitination of endogenous rather than ectopic IKK. As detected on immunoblots with antibodies to the Ub tag (Fig. 1C, top panels), the endogenous pool of IKK␤ complexes contained a major immunoreactive species of ϳ105 kDa, consistent with the predicted molecular size of monoubiquitinated IKK␤ (lane 4). Parallel studies with control cells lacking either tagged Ub or Tax confirmed the signaldependent nature of this modification (Fig. 1C, lanes 1-3). In keeping with the results shown in Fig. 1B, endogenous IKK␥ was polyubiquitinated in Tax-expressing cells (Fig. 1C, lane 8). Changes in the ubiquitination pattern could not be attributed to Tax-mediated stabilization or degradation of either IKK subunit (Fig. 1C, lower panels). These data provide direct biochemical evidence for differential ubiquitination of IKK␤ versus IKK␥ at the endogenous protein level.
Lys-163 Is the Primary Ub Acceptor Site in IKK␤-Protein ubiquitination is mediated by the sequential action of at least three enzymes (termed E1, E2, and E3) that together catalyze the activation and ligation of Ub to susceptible lysine residues in a substrate (22). A prerequisite for understanding the mechanism and functional consequences of IKK␤ monoubiquitination is to identify the relevant Tax-responsive acceptor site(s). In this regard, recent structural studies indicate that the spacing between a substrate's inducible phosphorylation motif and its Ub acceptor site is a key determinant of ubiquitination efficiency (23). In what may be a related observation, we have found that Tax-induced phosphorylation of T loop residues Ser-177/Ser-181 in IKK␤ is a prerequisite for its monoubiquitination (19).
These mechanistic results raised the possibility that IKK␤ is monoubiquitinated at a Lys residue positioned relatively close to the T loop phosphoacceptors. To identify candidate Ub acceptors, site-directed mutagenesis was used to engineer Arg replacements at each of the 11 Lys residues in the N-terminal kinase domain of IKK␤ ( Fig. 2A). We then programmed Tax-deficient cells with these IKK␤ mutants and monitored their in vivo ubiquitination status by immunoblotting. As shown in Fig. 2B, all of the kinase constructs were comparably expressed (lower panel). With the notable exception of mutant K171R (Lys-171 3 Arg), which was constitutively monoubiquitinated, none of the Lys 3 Arg mutants of IKK␤ were appreciably modified with Ub in Tax-deficient cells (Fig. 2B, top panel). The K171R mutant was also constitutively phosphorylated at T loop residues Ser-177/Ser-181 (data not shown), consistent with our prior studies indicating that the two modifications are linked (19).
Similar experiments were conducted with Tax-expressing cells. As shown in Fig. 2C  impaired for Ub attachment (top panel, lanes 3, 5, 6, and 10). The observed defects in monoubiquitination could not be attributed to Taxinduced changes in IKK␤ protein expression (Fig. 2C, middle panel) or varying levels of Tax expression (lower panel). Moreover, mutations that simultaneously disrupted all of the Lys residues located within the leucine zipper, the helix-loop-helix, or the intervening regions of IKK␤ were without effect, thus confirming specificity (Fig. 2D). These in vivo findings suggested that the kinase domain of IKK␤ contains one or more Ub acceptor sites in addition to its phosphoacceptors at Ser-177/Ser-181. Alternatively, we reasoned that some of the Lys 3 Arg mutations might inhibit ubiquitination by perturbing the structural and/or functional integrity of the IKK␤ kinase domain.
To address this possibility, we investigated whether any of the Lys 3 Arg replacements disrupted the catalytic function of IKK␤.SE, a constitutively active version of IKK␤ containing Glu phosphomimetics at Ser-177/Ser-181 (11,15). For these studies, the IKK␤.SE constructs were produced in 293T cells and then monitored for their capacity to phosphorylate a recombinant IB substrate in vitro. As shown in Fig. 2E (top  panel), IKK␤.SE displayed significant IB kinase activity as compared with a control construct containing Ala substitutions at Ser-177/Ser-181 (lanes 1 and 2). Introduction of Arg at positions 44, 147, or 238 completely abolished the activity of IKK␤.SE (lanes 3, 4, and 6). In sharp contrast, IB kinase activity was fully retained following replacement of Lys-163 with Arg (lane 5). Taken together with Fig. 2C, we conclude that the K44R, K147R, and K238R mutations inhibit ubiquitination by perturbing the structural and/or functional integrity of the IKK␤ kinase domain, whereas the K163R mutation selectively removes the Ub attachment site. Consistent with this interpretation, subsequent studies confirmed that removal of Lys-163 from IKK␤.SE blocks Ub attachment in vivo (see Fig. 5A).
Tax and Proinflammatory Receptors Converge on the Same Ub Acceptor-Similar to the Tax oncoprotein, chronic cellular stimulation through TNF-R1 leads to monoubiquitination of IKK␤ (19). We have shown previously that Tax activates the IKK complex via a direct binding mechanism (24). In contrast, TNF-R1 acts indirectly on IKK via intermediate adaptors, such as RIP, which recruit the kinase to the cytoplasmic tail of TNF-R1 (2,5). Given these divergent mechanisms of action, we reasoned that Tax and TNF-R1 might target distinct Ub acceptors in IKK␤. To explore this possibility, we transfected 293T cells with expression vectors for IKK␤, IKK␥, and either TNF-R1 or Tax. We then monitored in vivo ubiquitination of IKK␤ under each stimulatory condition.
As expected, neither wild type IKK␤ nor the K163R mutant was monoubiquitinated in the absence of chronic cellular stimulation (Fig.  3, top panel, lanes 1 and 2). In keeping with results shown in Fig. 2C, coexpression with Tax led to persistent monoubiquitination of wild type IKK␤, whereas the K163R mutant was defective for this modification (Fig. 3, lanes 3 and 4). Despite the distinct mechanism by which TNF-R1 activates IKK signaling, the pattern of IKK␤ subunit ubiquitination induced by TNF-R1 was indistinguishable from that obtained with Tax-expressing cells (Fig. 3, lanes 5 and 6). Immunoblotting experiments confirmed that Tax, TNF-R1, and IKK substrate targets were efficiently expressed (Fig. 3, lower panels). These biochemical data suggest that monoubiquitination at Lys-163 is a common post-translational modification step in the mechanism of Tax and TNF-R1 action on IKK␤.
Functional Consequences of IKK␤ Monoubiquitination-Both Tax and TNF-R1 stimulate the catalytic activity of IKK␤ via a mechanism involving phosphorylation of its T loop at Ser-177/Ser-181 (11,12). This modification appears to be mediated by IKK␤ itself (13). In turn, acti-vated IKK␤ phosphorylates a set of distinct protein substrates, including the associated regulatory subunit IKK␥ and cytoplasmic inhibitors of NF-B such as IB␣ (6,12,14,15). To determine whether these biochemical steps are dependent on monoubiquitin attachment, we analyzed the in vivo phosphorylation status of K163R, IKK␥, and IB␣ in Tax-expressing cells. For these studies, cytoplasmic extracts from recipient cells were fractionated by SDS-PAGE, and each of the kinase substrates were probed on immunoblots with phosphospecific antibodies. As shown in Fig. 4A (top panels), phosphorylation of the T loop in wild type IKK␤ was potently induced by Tax (lanes 1 and 2). Replacement of Lys-163 with Arg in IKK␤ had no detectable inhibitory effect on T loop phosphorylation (Fig. 4A, top panels, lanes 3 and 4). This finding further confirms the structural integrity of the K163R mutant and assignment of Lys-163 as the primary Ub acceptor site. The K163R mutant also retained the capacity to phosphorylate IKK␥ (Fig. 4A, middle panels, lanes 3 and 4) and IB␣ (lower panels, lanes 3 and 4) in Tax-expressing cells. We conclude that monoubiquitin is dispensable for acquisition of the deregulated IKK␤ phenotype in cells expressing Tax, which bypasses receptor-proximal signal transducers and interfaces directly with IKK (24).
To extend these experimental results to a different class of IKK agonist, studies were conducted with cells expressing TNF-R1 rather than Tax. We also monitored in vivo phosphorylation of IKK␤ substrates in cells programmed with RIP, a downstream signaling intermediate in the TNF-R1 pathway that interfaces directly with IKK (25). As expected (Fig. 4B, lanes 1 and 2), coexpression with TNF-R1 led to chronic T loop phosphorylation of wild type IKK␤ (top panels), enabling the kinase to phosphorylate IKK␥ (middle panels) and IB␣ (lower panels). Replacement of Lys-163 with Arg in IKK␤ significantly reduced its steady-state level of T loop phosphorylation (Fig. 4B, top panels, lanes 3 and 4). Disruption of Lys-163 also impaired the capacity of IKK␤ to phosphorylate IKK␥ and IB␣ in cells expressing TNF-R1 (Fig. 4B, middle and  lower panels, lanes 3 and 4). In sharp contrast, K163R, IKK␥, and IB␣ were all efficiently phosphorylated in vivo following enforced expression of RIP (Fig. 4C). Thus, Lys-163 is required for sustained IKK␤ signaling from cell-surface TNF-R1, whereas RIP and Tax can circumvent this requirement from their sites of action in the cytoplasmic compartment. These results suggest an important mechanistic role for monoubiquitination in stable targeting of IKK␤ to TNF-R1 in the plasma membrane during a chronic inflammatory response.
Role of Monoubiquitin in IKK␤ Phosphorylation at C-terminal Sites-Delhase et al. (11) reported that IKK␤ autophosphorylates at a cluster of C-terminal serines following transient cellular stimulation with TNF. Unlike T loop phosphorylation of IKK␤ at Ser-177/ Ser-181, which greatly stimulates kinase activity, modification of these C-terminal sites is dispensable for the induction of IKK␤ activity in TNF-treated cells (11). To determine whether Ub attachment regulates phosphoryl group transfer to this serine cluster, we performed a series of experiments with IKK␤ substrates containing mutations that disrupt the Ub acceptor (mutant K163R) or the C-terminal phosphoacceptors (mutant M10; Ser 3 Ala replacements at positions 670, 672, 675, 679, 682, 689, 692, 695, 697, and 705) (11). To exclude the potential for any subtle effects on the stoichiometry of T loop phosphorylation, both of the mutations were introduced into a constitutively active form of IKK␤ harboring Glu residues in place of Ser-177/Ser-181 (IKK␤.SE). These Glu residues function efficiently as T loop phosphomimetics (see Fig. 2E) and facilitate chronic phosphorylation of the C-terminal serines in IKK␤ (11).
We first monitored the resultant IKK␤ constructs for Ub attachment following their enforced expression in 293T transfectants. All of the IKK␤ mutants were expressed at comparable levels as demonstrated by immunoblotting (Fig. 5A, middle panel). As shown in the top panel of Fig. 5A, IKK␤.SE was chronically ubiquitinated in the absence of overt cellular stimulation (lane 2). In keeping with the essential role of T loop phosphorylation in Ub conjugation, ubiquitination was blocked when the T loop phosphomimetics were replaced with Ala (mutant IKK␤.SA) (Fig. 5A, lane 1). Likewise, replacement of Lys-163 with Arg in IKK␤.SE impaired Ub attachment (Fig. 5A, lane 3). In contrast, disruption of the C-terminal serine cluster in IKK␤ was without effect (Fig. 5A, lane 4). Thus, C-terminal phosphorylation of IKK␤ is not a prerequisite for monoubiquitination.
To determine whether monoubiquitination is required for C-terminal phosphorylation, transfectants expressing the same set of IKK␤ mutants were metabolically radiolabeled with [ 32 P]orthophosphate. Kinases complexes were isolated from recipient cells by immunoprecipitation, fractionated by SDS-PAGE, and analyzed by sequential autoradiography and immunoblotting. As shown in Fig. 5B (top panel), all of the IKK␤.SE constructs were efficiently phosphorylated relative to IKK␤.SA, a kinase-dead mutant (see Fig. 2E). As such, the radiolabeled kinases were digested in situ with chymotrypsin, and the resultant peptides were resolved on two-dimensional phosphopeptide maps to investigate the modification status of serines positioned in the C-terminal region of IKK␤.
As shown in Fig. 5C, unmodified IKK␤.SE contained at least six major chymotryptic phosphopeptides (designated PP-1 through PP-6). Disruption of the C-terminal serine cluster in IKK␤.SE eliminated PP-2 through PP-6 ( Fig. 5D). In contrast, disruption of the Ub acceptor in IKK␤ completely eliminated PP-3 and significantly altered the content of PP-2 and PP-6 (Fig. 5E). The observed in vivo effect of the K163R mutation on PP-2, PP-3, and PP-6 was highly specific, because replacement of T loop-proximal residue Lys-198 with Arg yielded the same phosphorylation pattern as IKK␤.SE (Fig. 5F). We conclude that Lys-163 is required for proper in vivo phosphorylation of the C-terminal serine cluster in IKK␤.
To investigate whether autophosphorylation generates PP-2, PP-3, and PP-6, we next assessed the in vitro modification pattern of purified IKK␤.SE proteins. For these studies, 293T cells were programmed with IKK␤.SE containing Lys versus Arg at position 163 and grown in media lacking [ 32 P]orthophosphate. Ectopic proteins were immunopurified under highly stringent washing conditions (2 M urea) in order to remove kinases that might loosely associate with IKK␤.SE and phosphorylate its C-terminal serine cluster. Purified IKK␤.SE was then incubated with [␥-32 P]ATP and analyzed by two-dimensional phosphopeptide mapping. In keeping with the in vivo fingerprints, PP-4 and PP-5 were clearly evident following in vitro autophosphorylation of IKK␤.SE (Fig. 5G). Similar results were obtained with its ubiquitination-defective counterpart (Fig. 5H). However, we failed to detect PP-2, PP-3, and PP-6, which were selectively impacted by the Lys-163 3 Arg mutation in vivo (Fig.  5E). These unexpected findings indicate a Ub-dependent subset of C-terminal phosphoacceptors that may be targeted by a kinase other than IKK␤ itself.

DISCUSSION
Current knowledge about the substrate targets and functional range of monoubiquitin in mammalian cell biology is still rudimentary as compared with polyubiquitin, which earmarks many proteins for proteasome-mediated degradation (20,26). In prior studies, we found that chronic signaling mediated by the Tax oncoprotein or the proinflammatory receptor TNF-R1 leads to IKK␤ monoubiquitination, albeit under conditions of IKK subunit overexpression (19). In the present report, we provide several new lines of evidence that underscore the pathophysiological relevance and specificity of this post-translational modification to IKK␤. We demonstrate that (i) monoubiquitinated IKK␤ is integrated into higher order complexes containing the IKK␥ regulatory subunit, (ii) IKK␥ is conjugated to polyubiquitin rather than monoubiquitin in Tax-expressing cells, and (iii) Tax targets endogenous IKK␤ for monoubiquitination in the absence of IKK subunit overex-pression ( Fig. 1). Thus, IKK␤ joins a small but growing roster of cellular proteins that are modified by the addition of a single ubiquitin unit in a signal-dependent manner (20).
In order to understand the regulatory potential of monoubiquitin in chronic IKK␤ signaling, we introduced a comprehensive set of Lys 3 Arg mutations in IKK␤ and then monitored the kinase for monoubiquitination in Tax-expressing cells. Replacement of the T loop-proximal residue Lys-163 with Arg (mutant K163R) yielded a ubiquitination-defective (UD) mutant of IKK␤, whereas conservative Lys 3 Arg mutations at other potential Ub acceptors in the leucine zipper, helix-loophelix, and intervening regions of IKK␤ failed to block this modification step (Fig. 2). More importantly, the K163R mutant retained the capacity to associate with IKK␥ and was targeted for Tax-induced phosphorylation at Ser-177/Ser-181 (Figs. 2E and 4A). These biochemical experiments confirm the structural integrity of Lys-163 as a UD mutant of IKK␤ and reinforce the assignment of Lys-163 as the major Ub acceptor site. K163R was also defective for ubiquitination mediated by chronic TNF-R1 signaling, suggesting that monoubiquitin conjugation to Lys-163 in IKK␤ is a common step in multiple pathways for NF-B induction (Fig. 3).
Recent structural studies indicate that the spacing between a substrate's inducible phosphorylation motif and its Ub acceptor sites is a key determinant of ubiquitination efficiency (23). Consistent with these  ). B, cells were transfected as described in A (except that the Ub expression vector was omitted) and radiolabeled with [ 32 P]orthophosphate for 4 h. IKK complexes were immunopurified with anti-FLAG antibodies and fractionated by SDS-PAGE. Resolved proteins were analyzed by sequential autoradiography (top panel) and immunoblotting with IKK subunit-specific antibodies (lower panels). C-H, cells were transfected and metabolically radiolabeled as described in B. Alternatively, [ 32 P]orthophosphate was omitted from the growth media (G and H). Radiolabeled IKK␤ was immunopurified, resolved by SDS-PAGE, transferred to a PVDF membrane, and digested in situ with chymotrypsin. Peptides were separated by sequential electrophoresis and chromatography on TLC plates. The same protocol was used to process extracts from unlabeled cells, except that immunopurified IKK␤ was subject to in vitro phosphorylation in the presence of [␥-32 P]ATP prior to SDS-PAGE. Radiolabeled chymotryptic fragments derived from IKK␤ following either in vivo phosphorylation (PP-1 through PP-6, top four panels; spots [1][2][3][4][5][6]  spatial constraints, Lys-163 lies in close proximity to the T loop phosphoacceptors of IKK␤. This configuration is highly reminiscent of the distribution of modification sites involved in the phosphorylation-dependent mechanism for ubiquitination of IB␣, a downstream target of IKK␤ (6). By analogy with IB␣ (7), phosphorylation of IKK␤ at Ser-177/Ser-181 may generate a substrate recognition motif, permitting the Ub-conjugating machinery to dock with and monoubiquitinate the kinase at nearby Lys-163. This mechanistic model for IKK␤ monoubiquitination may have broader relevance, because Lys-163 is highly conserved among other cellular enzymes under T loop control, including members of the MAP kinase family of signal transducers (11,27).
To determine whether monoubiquitination of IKK␤ regulates its kinase action on downstream substrates, we monitored the phosphorylating activity of K163R in cells expressing either Tax or TNF-R1. In the presence of Tax, which interfaces directly with IKK (24), K163R mediated efficient phosphorylation of IKK␥ and IB␣ (Fig. 4A). In contrast, this UD mutant was defective for sustained T loop phosphorylation in cells expressing TNF-R1, thus attenuating IKK␥ and IB␣ kinase activity (Fig. 4B). In subsequent studies, we found that the requirement for Ub attachment at Lys-163 was circumvented by RIP, a downstream effector of TNF-R1 that binds to IKK (Fig. 4C). These agonist-specific effects suggest that monoubiquitination facilitates persistent activation of IKK␤ via its stable recruitment to the cytoplasmic tail of TNF-R1. In contrast, the Ub modification is largely dispensable for persistent activation of IKK␤ by agonists that associate directly with the kinase complex, such as Tax and RIP. Further resolution of this issue awaits in vivo gene targeting studies with the UD mutant of IKK␤ identified in the present study.
In response to cellular stimulation with TNF, IKK␤ phosphorylates a cluster of serines located near its C terminus (11). Unlike T loop phosphorylation of IKK␤ at Ser-177/Ser-181, which greatly stimulates kinase activity, modification of these TNF-responsive serines may influence the duration of IKK␤ signaling (11). Our phosphopeptide mapping data provide evidence that the latter phosphorylation step is facilitated by Ub attachment to IKK␤. Two distinct subsets of C-terminal phosphopeptides in IKK␤ were revealed in these studies (Fig. 5). In vivo phosphorylation of one subset was unaffected by the K163R mutation and fully recapitulated in vitro. This biochemical phenotype is in keeping with IKK␤ autophosphorylation as reported by Delhase et al. (11). However, in vivo phosphorylation of the second subset was altered by the K163R mutation, indicating the involvement of monoubiquitin. These secondary phosphorylation sites in IKK␤ were also resistant to in vitro modification, suggesting a requirement for accessory factors.
The precise mechanism underlying Ub-mediated phosphorylation of IKK␤ remains unclear. Our findings may reflect the emerging role of monoubiquitin as an adaptor that mediates protein/protein interactions (28). Assuming that IKK␤-conjugated Ub provides an interaction surface, at least two mechanistic models can be postulated. In one model, formation of IKK␤-Ub conjugates enables the catalytic domain of IKK␤ to physically engage its C-terminal phosphorylation targets. Alternatively, Ub attachment to IKK␤ facilitates the recruitment of a novel IKK␤ kinase, either directly or indirectly via a Ub receptor. Consistent with the latter prospect, recent studies have identified a Ub-binding motif in MEKK1, a MAP kinase kinase kinase (29). Members of this enzyme family have been implicated previously in TNF-induced activation of IKK (30).
In vitro stimulation of mammalian cells with the cytokine TNF leads to rapid T loop phosphorylation and activation of IKK␤, followed by dephosphorylation and inactivation within 30 -60 min (11). Preliminary experiments indicate that this transient response does not enable the stable formation and/or detection of IKK␤-Ub conjugates (supplemental Figs. 1 and 2). In contrast, IKK␤-Ub conjugates were readily detected after reinforced expression of TNF-R1, which evokes sustained T loop phosphorylation and activation of IKK␤ (19). This persistent pattern of IKK␤ signaling has been implicated in chronic inflammatory disease and cancer (31,32). The capacity of TNF-R1 to trigger IKK␤ monoubiquitination was fully recapitulated by the Tax oncoprotein, a pathophysiologic agonist that mediates persistent IKK signaling via a direct binding mechanism (19). Accordingly, formation of stable IKK␤-Ub conjugates may be contingent upon sustained T loop phosphorylation, persistent IB kinase activity, or the action of a late gene product that is selectively expressed under chronic inducing conditions.
In summary, we report here that chronic phosphorylation of the T loop at Ser-177/Ser-181 in IKK␤ leads to monoubiquitination at nearby Lys-163. This close spatial relationship may be attributed to docking of the relevant Ub-conjugating machinery at the T loop phosphoacceptors in IKK␤. We find that Ub modification is required for sustained IKK␤ signaling from the proinflammatory receptor TNF-R1 but not for acquisition of the deregulated IKK␤ phenotype in cells expressing the Tax oncoprotein. Our studies also indicate that monoubiquitination regulates the phosphorylation status of TNF-responsive serines located in the C-terminal region of IKK␤ (11). Interference with Ub attachment selectively alters the phosphopeptide fingerprint of IKK␤ following in vivo but not in vitro phosphorylation, suggesting the involvement of a kinase other than IKK␤ itself. These findings highlight a novel mechanism for post-translational cross-talk between monoubiquitination and phosphorylation that may play an important regulatory role in IKK signaling during a chronic inflammatory response.