Inducible Degradation of IκBα by the Proteasome Requires Interaction with the F-box Protein h-βTrCP*

Activation of NF-κB transcription factors requires phosphorylation and ubiquitin-proteasome-dependent degradation of IκB proteins. We provide evidence that a human F-box protein, h-βTrCP, a component of Skp1-Cullin-F-box protein (SCF) complexes, a new class of E3 ubiquitin ligases, is essential for inducible degradation of IκBα. βTrCP associates with Ser32–Ser36 phosphorylated, but not with unmodified IκBα or Ser32–Ser36phosphorylation-deficient mutants. Expression of a F-box-deleted βTrCP inhibits IκBα degradation, promotes accumulation of phosphorylated Ser32–Ser36 IκBα, and prevents NF-κB-dependent transcription. Our findings indicate that βTrCP is the adaptor protein required for IκBα recognition by the SCFβTrCP E3 complex that ubiquitinates IκBα and makes it a substrate for the proteasome.

NF-B transcription factor is regulated by IB proteins of which IB␣ is the main and best characterized member (1)(2)(3). Proteasome-mediated degradation of IB␣ releases NF-B and allows its localization in the nucleus (4 -9). Phosphorylation of Ser 32 -Ser 36 residues and subsequent ubiquitination of IB␣ are prerequisites to make the protein susceptible to proteasome attack (10 -13). While the kinase complex accounting for IB␣ phosphorylation has been recently characterized (14 -20), factors necessary for ubiquitination and targeting of IB␣ to the proteasome remain unknown. Covalent attachment of polyubiquitin to substrate proteins involved a cascade of ubiquitin transfer reactions with E1, 1 a ubiquitin-activating enzyme, and E2 a ubiquitin-conjugating enzyme that operates in conjunction with a specificity factor E3 (21)(22)(23)(24)(25)(26). It has been suggested that E3 functions in substrate recognition and E2 positioning. Although E2 enzymes belonging to the Ubc4/Ubc5 family can ubiquitinate IB␣ in vitro, the E3 responsible for the signalinduced ubiquitination of IB␣ remains to be identified (10,27). A novel class of E3 is represented by the Skp1-Cullin-F-box protein complexes (SCFs) (28 -33). The core component of these newly identified E3s is Skp1, which assembles with different F-box proteins and has been shown in human cells to interact selectively with CUL-1, but not with other Cullin proteins belonging to the Cdc53 family (34 -36). The role of the F-box proteins in these SCF complexes is to recruit phosphorylated substrate proteins to trigger their ubiquitination (28 -33). Like the other members of the F-box protein family, human ␤TrCP, which we recently identified (37), has a modular organization with an F-box motif involved in proteasome targeting through interaction with Skp1, and a seven WD repeats binding domain for interaction with substrate proteins (see Fig. 1A). We hypothesized that ␤TrCP could be the F-box adaptor protein allowing recruitment of IB␣ by a SCF E3 ubiquitin-protein ligase complex that ubiquitinates IB␣ and makes it a substrate for degradation by the proteasome.

EXPERIMENTAL PROCEDURES
Cell Lines, Transfections, and Infections-Subconfluent cells of the 293 human embryo kidney cell line were transfected by Lipofect-AMINE TM Plus (Life Technologies, Inc.) with the indicated reporter plasmids and pcDNA3 vectors expressing ␤TrCP proteins or with no ␤TrCP insert. Fluorigenic substrate luciferin served to quantify luciferase reporter gene expression in cytoplasmic extracts obtained by lysis in phosphate buffer containing 1% Nonidet P-40. The pcDNA3-␤TrCP or -␤TrCP⌬F constructs are described in Ref. 37. ␤TrCP and ␤TrCP⌬F coding sequences were amplified by polymerase chain reaction and inserted in fusion with the Myc/His double tag in the pcDNA3.1 Myc/ HisA vector (Invitrogen). SV5-tagged wild type or SV5-tagged S32A/ S36A phosphorylation-deficient mutant IB␣ are described in Ref. 11, 3Enh-B-ConA and ConA luciferase reporter plasmids are described in Ref. 38. RSV luciferase reporter plasmid was purchased from Invitrogen. ␤TrCP or IB␣ were inserted in fusion with the LexA DNA binding domain and the Gal4 activation domain, respectively, as described in Ref. 37. Construction and use of recombinant SFV was carried out as described before (39 -42). Briefly, 293 cells were infected at a multiplicity of 5 for 6 h in serum-free medium with SFV particles carrying myc-tagged ␤TrCP proteins. Expression of ␤TrCP proteins and viral nucleocapsid protein was verified by Western blot analysis, and 100% infection efficiency was confirmed by immunofluorescence (data not shown). The same results as those shown in Fig. 2B were obtained in HeLa cells infected with SFV.
Antibodies and Reagents-The 10B monoclonal antibody directed against the amino terminus of IB␣ was described in Ref. 43; the polyclonal antibody specifically recognizing IB␣ phosphorylated at serine residue 32 is from New England Biolabs (9241S), and the monoclonal anti-myc antibody was from Santa Cruz Biotechnology (hybridoma 9E10, SC-40). For co-immunoprecipitation experiments (Fig. 3, A  and B), 200 g of cytoplasmic extract were incubated with anti-mycagarose conjugates (SC-40 AC from Santa Cruz) for 60 min at 4°C. Precipitated beads were washed 10 times in phosphate-buffered saline containing 1% Nonidet P-40 and both protease and phosphatase inhibitors. Antibody-antigen complexes were disrupted by boiling in gel loading buffer (Pierce). Precipitated proteins were fractionated by SDS-* This work was supported in part by the European Union Concerted Action BIOMED II (ROCIO II project), Agence Nationale de Recherche sur le SIDA, SIDACTION, and Association pour la Recherche sur le Cancer (France). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
§ Supported by a fellowship of the German Academic Exchange Service (Program HSP III).
Electrophoretic Mobility Shift Assay-The electrophoretic mobility shift assay was performed with 4 g of nuclear extract incubated for 15 min at room temperature with a [␥-32 P]ATP-labeled, double-stranded oligonucleotide containing the HIV-1 long terminal repeat binding site for NF-B (5Ј-ACAAGGGACTTTCCGCTGGGACTTTCCAGGGA-3Ј). Samples were analyzed in nondenaturing 6% polyacrylamide gels. Competition experiments were performed by adding a 40-fold molar excess of homologous, unlabeled oligonucleotide to each sample prior to addition of the radiolabeled probe.

RESULTS AND DISCUSSION
To investigate the putative role of ␤TrCP in the regulation of NF-B activation, we first assessed the effect of wild type (␤TrCP) and a F-box deleted ␤TrCP (␤TrCP⌬F) (Fig. 1A) on the transcriptional activity of a NF-B-dependent (3Enh-B-ConA) promoter driving a luciferase reporter gene (10,11). Expression of ␤TrCP resulted in a 2-to 3-fold increase in the activity of the 3Enh-B-ConA promoter in cells of the 293 human embryo kidney cell line stimulated by either tumor necrosis factor (TNF) or okadaic acid (OKA), two well characterized inducers of NF-B (Fig. 1B, left panel). Similarly, the low level of constitutive activation of NF-B found in unstimulated cells (compare 3Enh-B-ConA to ConA) is also enhanced by expression of ␤TrCP (Fig. 1B, left panel). In sharp contrast to ␤TrCP, expression of the ␤TrCP⌬F mutant massively and consistently inhibited NF-B-dependent transcription by more than 90% of the levels induced in TNF-or OKA-stimulated cells transfected with an insertless, control plasmid (pcDNA3) (Fig. 1B, left  panel). Importantly, expression of the transdominant negative mutant ␤TrCP⌬F fully prevented localization of NF-B to the nucleus upon cell activation. This finding is in keeping with an increased stability of inhibitor IB proteins that anchor NF-B in the cytoplasm in an inactive form. The failure of ␤TrCP⌬F to modify the activity of a RSV promoter indicates that ␤TrCP⌬F does not affect NF-B independent mechanisms of transcription ( Fig. 1B, left panel).
Thus, on the one hand, the capacity of ␤TrCP to enhance both basal or signal-induced activation of NF-B and, on the other hand, the specific inhibition of NF-B dependent transcription promoted by ␤TrCP⌬F, strongly suggest that ␤TrCP is an essential component of the NF-B activation pathway. In support of this assumption, we found that, concomitant with the enhancement of NF-B-dependent transcription, overexpression of ␤TrCP in 293 cells stimulated by TNF induced an accelerated degradation of IB␣ ( Fig. 2A, middle panel). In contrast, expression of the ␤TrCP⌬F mutant either from a eukaryotic vector ( Fig. 2A, right panel) or a SFV replicon (Fig.  2B, right panel), stabilized IB␣ and delayed the kinetics of IB␣ degradation. Moreover, the presence of ␤TrCP⌬F promoted the accumulation of slow migrating forms of IB␣ characteristic of the phosphorylation of residues Ser 32 and Ser 36 required for subsequent ubiquitination and degradation of IB␣ by the proteasome (Fig. 2, A and B, right panel). Thus, these findings exclude an inhibitory effect of ␤TrCP⌬F in the transduction pathway leading to phosphorylation of IB␣ and suggest that stabilization of IB␣ is due to the blockade of a post-phosphorylation event in the metabolism of the inhibitor.
Slow migrating forms of IB␣ predominated following induction with TNF but could even be detected in unstimulated cells (Fig. 2, A and B, right panel, time 0). This latter phenomenon likely reflects the inhibitory effect of ␤TrCP⌬F on basal breakdown of IB␣ and is in keeping with the low and constitutive NF-B-dependent transcription observed in 293 cells (Fig. 1B,  left panel).
Characterization of the slow migrating band of IB␣ ( Fig. 2A,  right top panel), as a phosphorylated form of IB␣ accumulated  (10,11). Equal amounts of the pcDNA3 plasmid without ␤TrCP cDNA insert were transfected as a control. The amount of wild type and mutant ␤TrCP proteins was assessed by Western blot analysis of cytoplasmic extracts to ensure that wild type and mutant ␤TrCP proteins were expressed at comparable levels (not shown). After transfection, cells were left untreated (NS) or were stimulated with either 5 ng/ml TNF or 75 ng/ml OKA for 6 h. Experiments were repeated three times in 293 cells and confirmed in HeLa cells. Results of a representative experiment are shown. Luciferase activity is expressed as relative luciferase units (RLU) per g of protein and results from subtracting the background signal from the values obtained for each sample. Comparative electrophoretic mobility shift assay of nuclear extracts from HeLa cells expressing either ␤TrCP or the ␤TrCP⌬F from SFV. A radiolabeled oligonucleotide encoding the NF-B consensus was used as a probe to bind transcription factors. A SFV replicon without insert was used as a control. The asterisk indicates where competitor cold oligonucleotide was added to demonstrate the specificity of NF-B/DNA interaction.
in ␤TrCP⌬F expressing cells, was accomplished ( Fig. 2A, middle row of panels) using a polyclonal antibody that does not recognize the unphosphorylated form of IB␣ but specifically reacts with Ser 32 -Ser 36 -phosphorylated IB␣ which accumulates in the presence of the proteasome inhibitor Z-LLL-H following induction with TNF (Fig. 2C) (44).
To ascertain whether the regulatory effect of ␤TrCP on IB␣ metabolism requires association with IB␣, carboxyl-terminal c-myc-tagged variants of either ␤TrCP or ␤TrCP⌬F were expressed in HeLa cells from SFV (Fig. 3A) or in 293 cells from eukaryotic expression vectors (Fig. 3B), and their co-precipitation with IB␣ was investigated. Cytoplasmic extracts of cells treated or not with TNF and the proteasome inhibitor Z-LLL-H, were incubated with an anti-myc tag antibody bound to protein FIG. 3. ␤TrCP interacts with Ser 32 -Ser 36 phosphorylated but not with unmodified or phosphorylation-deficient S32A/S36A IB␣. A, ␤TrCP selectively co-precipitates phosphorylated IB␣ proteins. HeLa cells were infected as described in Fig. 2B with SFV expressing a ␤-galactosidase (␤Gal), wild type (␤TrCP), or ⌬F (␤TrCP⌬F) myc-tagged proteins (11). Before harvesting, cells were exposed to TNF for 15 min in the presence of proteasome inhibitor (Z-LLL-H ϩ TNF) or left untreated (Ϫ). Proteins from total cytoplasmic lysates were either separated in denaturing gels and probed with ␣-IB␣-S32-P antibody (top panel) or immunoprecipitated (IP ␣-Myc) with myc antibody-agarose conjugates (13). Precipitated proteins were probed with either the ␣-IB␣-S32-P antibody (middle panel) or the 10B monoclonal antibody (␣-IB␣, bottom panel) that recognizes both unmodified and phosphorylated IB␣ (13). The migration pattern of anti-myc-precipitated (IP ␣-Myc) IB␣ is compared with that of IB␣ in total cytoplasmic extracts (control HeLa cell extracts, purchased from New England Biolabs) of untreated (Ϫ) or TNF-treated (ϩ) HeLa cells detected by either ␣-IB␣-S32-P or ␣-IB␣ antibodies (right, panel A). B, ␤TrCP proteins fail to co-precipitate with a S32A/S36A phosphorylation-deficient IB␣ mutant. 293 cells were co-transfected with pcDNA3-␤TrCP or -␤TrCP⌬F DNA vectors and either SV5-tagged wild type (lanes 1-6) or SV5-tagged S32A/S36A phosphorylation-deficient mutant IB␣ proteins (lanes 7-10) deficient in cell activation-induced serine phosphorylation (11). A pcDNA3 plasmid expressing ␤-galactosidase was used as a control (lanes 1 and 2). Cytoplasmic proteins were precipitated by anti-myc antibodies and probed (WB) with ␣-IB␣-S32-P or ␣-IB␣ antibodies indicated under the gels. Migration of either tagged (SV5-IB␣) or endogenous (End-IB␣) IB␣ is indicated. Long and short exposures using the ECL kit are shown. C, interaction of ␤TrCP with wild type but not with the phosphorylation-deficient S32A/S36A IB␣ in the twohybrid system. ␤TrCP proteins or control Ras protein were fused to the Escherichia coli LexA binding domain. IB␣ proteins or control Raf protein were fused to the Gal4 activation domain. The yeast reporter strain L40 expressing the indicated hybrid protein pairs was analyzed for histidine auxotrophy and ␤-galactosidase expression. D, wild type IB␣ is constitutively phosphorylated in yeast. L40 strain yeast colonies expressing the Gal4 activation domain fused to Raf, IB␣ wild type, or IB␣ S32A/S36A were lysed, and protein extracts were blotted onto nitrocellulose. Left, detection with ␣-IB␣ antibody. Right, detection with ␣-IB␣-S32-P antibody.

FIG. 2. Expression of the ␤TrCP⌬F mutant inhibits degradation and promotes accumulation of Ser 32 -Ser 36 phosphorylated IB␣.
A, stability of IB␣ analyzed in 293 cells exposed to TNF for different lengths of time in the presence of ␤TrCP proteins. 293 cells were transiently transfected with ␤TrCP, ␤TrCP⌬F, or control pcDNA vectors. After 36 h, cells were stimulated with TNF in the presence of 100 g/ml cycloheximide, the protein synthesis inhibitor. Cytoplasmic proteins were separated in SDS-denaturing polyacrylamide gels, transferred onto nitrocellulose membrane, and probed with the 10B monoclonal antibody directed against the amino terminus of IB␣ (top panel), a polyclonal antibody specifically recognizing IB␣ phosphorylated at serine residue 32 (␣-IB␣-S32-P antibody, middle panel), or a monoclonal anti-myc antibody (bottom panel) directed against the carboxylterminal myc tag of heterologous ␤TrCP proteins (see Ref. 13 for antibodies). Immunodetection was performed using an ECL chemiluminescence kit (Amersham Pharmacia Biotech). B, stability of IB␣ in TNF-treated HeLa cells infected with Semliki forest virus replicons (SFV) (12) expressing the ␤TrCP⌬F mutant (␤TrCP⌬F) or not (SFV control). After 6 h of infection, cells were treated with cycloheximide and stimulated with TNF for the indicated times. Cytoplasmic extracts were processed as described above and probed with the 10B antibody. C, specific recognition of Ser 32 -Ser 36 phosphorylated IB␣ by the ␣-IB␣-S32-P antibody (13). Western blot analysis of recombinant (rIB␣, 10 ng) and cytoplasmic IB␣ obtained from HeLa untreated (Ϫ) or treated with TNF for 15 min. Left, detection of IB␣ proteins with 10B antibody. Right, detection with the ␣-IB␣-S32-P antibody.
A-coated agarose beads. Proteins precipitated by the anti-myc antibody were probed with anti-IB␣ antibodies (Fig. 3A, middle and bottom panels). In cells expressing either the wild type or the mutated counterpart of ␤TrCP, a single band of coprecipitated IB␣ was recognized by a monoclonal antibody that detects both native and phosphorylated forms of IB␣ (Fig.  3A, bottom panel, lanes 4 -6). This band migrates with a pattern characteristic of the typical upshift induced by phosphorylation of IB␣ Ser 32 and Ser 36 . Western blot analysis using the antibody specifically recognizing the phosphoserine Ser 32 -Ser 36 IB␣ confirmed that the proteins detected are phosphorylated at the critical residues that permit subsequent ubiquitination of IB␣ (Fig. 3A, middle panel, lanes 4 -6). No unphosphorylated form of IB␣ was detected in immunoprecipitates (Fig. 3A, lower panel). Detection of phosphorylated IB␣ did not require TNF induction when the mutant ␤TrCP⌬F was expressed (Fig. 3A, middle panel, lane 5). Moreover, larger amounts of endogenous phosphorylated IB␣ coprecipitated with ␤TrCP⌬F, compared with wild type ␤TrCP (Fig. 3A, compare lanes 6 with lane 4), confirming that the ␤TrCP⌬F mutant acts as a transdominant negative regulator of both constitutive and TNF-induced proteolysis of IB␣.
To confirm the selective association of phosphorylated IB␣ and ␤TrCP, we performed experiments using a S32A/S36A mutant of IB␣ lacking the capacity to be phosphorylated by cell activation signals promoting NF-B activation (3). The presence of a 15-amino acid SV5 carboxyl terminus tag allows distinction of endogenous from transiently expressed wild type or S32A/S36A IB␣. Despite expression of similar amounts of wild type or S32A/S36A-SV5 tagged proteins (data not shown), only the endogenous and wild type SV5-tagged IB␣ (Fig. 3B,  lanes 4 -6), but not the SV5-S32A/S36A phosphorylation-deficient mutant (Fig. 3B, lanes 7-10) were able to associate with either ␤TrCP or ␤TrCP⌬F. Expression of either wild type or phosphorylation-deficient tagged IB␣ proteins from SFV was consistently detected in more than 90% of cells (data not shown). The high infection efficiency of this system allows us to conclude that co-precipitation of the endogenous, but not the S32A/S36A-SV5 IB␣ (Fig. 3B, lanes 7-10), reflects the incapacity of the phosphorylation-deficient mutant to compete for binding to IB␣.
Further evidence of IB␣ interaction with ␤TrCP was provided by the yeast two-hybrid system (carried out as described previously (37)). We observed that ␤TrCP fused to the LexA DNA binding domain (LexA-␤TrCP) associates specifically with IB␣ fused to the Gal4 activation domain (Gal4AD-IB␣), as detected by histidine auxotrophy or ␤-galactosidase expression (Fig. 3C). The interaction between IB␣ and ␤TrCP is likely accounted for by the existence in yeast of phosphorylated IB␣ as shown by the recognition of the Gal4AD-IB␣ hybrid by the antibody that specifically reacts with Ser 32 -Ser 36 phosphorylated IB␣ (Fig. 3D). This hypothesis is reinforced by the fact that the Gal4AD-IB␣ S32A/S36A mutant, which is not recognized by the anti-phosphoserine IB␣ antibody (Fig. 3D), did not associate with LexA-␤TrCP (Fig. 3C).
We have previously documented that ␤TrCP is a component of a SCF complex involved in HIV-1 Vpu-mediated CD4 degradation. The WD domain of ␤TrCP is responsible for the interaction with Vpu (37). Although ␤TrCP is involved in both Vpu-mediated CD4 and IB␣ proteolysis, it should be stressed that important differences between the two degradation pathways exist. Indeed, if both Vpu and IB␣ can be phosphorylated at serine in DSGXXS motifs, phosphorylation of Vpu occurs constitutively (45)(46) while that of IB␣ requires activation of cell signaling. Furthermore, and in contrast to IB␣, Vpu has not yet been characterized as a substrate for ubiquitination or degradation by the proteasome. No human protein recognized as ubiquitination substrate by an SCF complex and undergoing degradation by the proteasome has been documented so far. IB␣ phosphorylated at critical serine residues represents the first example of this kind of substrate (8).
In the SCF complexes with E3-ubiquitin-protein ligase activity, the F-box component allows specific recognition of substrates (28 -33). The existence of a large number of F-boxcontaining proteins revealed by genome sequencing and the combinatorial interactions of SCF components that belong to different protein families (Cullin, E2 ubiquitin-protein conjugating enzymes) suggest that the growing family of E3 ligases is composed of a large number of different SCF complexes. This diversity has likely hampered the identification of the E3 ligase responsible for ubiquitin conjugation of IB␣ required for IB␣ degradation and NF-B activation. While precise characterization of the ubiquitin conjugating activity associated with SCF ␤TrCP is still missing, our findings provide evidence that ␤TrCP is ultimately responsible for recognition of phosphorylated IB␣ by the SCF complex.
As shown for other F-box proteins in yeast, ␤TrCP could target different substrates as well as IB␣ to the proteasome. This hypothesis is sustained by the recent discovery that, Slimb, the Drosophila homolog of ␤TrCP, may be involved in an as yet unidentified step of regulation of the wingless and hedgehog pathways (47). However, it cannot be assumed from this that ␤TrCP is a broad, universal adaptor for ubiquitinated substrates. Indeed, in the SCF complex, F-box proteins determine and restrict substrate recognition by the proteasome. Thus, the yeast F-box protein Cdc4 is able to selectively bind phosphorylated Sic1, but not phosphorylated Cln1 or Cln2, whereas Grr1, another yeast F-box protein, shows selective association with the latter substrates but not with Sic1- (30 -32).
In conclusion, our findings characterize ␤TrCP as a F-box protein, which selectively associates with Ser 32 -Ser 36 phosphorylated, but not unmodified, IB␣. ␤TrCP represents the SCF adaptor which ultimately accounts for recognition of phosphorylated IB␣ by the ubiquitination machinery and allows targeting of the ubiquitinated inhibitor to the proteasome (see model in Fig. 4). While this manuscript was completed and sent for review, a report by Yaron et al. (48) was published that supports the involvement of ␤TrCP in IB␣ degradation and NF-B activation. Thus, ␤TrCP can be considered as a new target for pharmacological intervention in the physiopathological processes regulated by NF-B.