Introduction

Top Abstract Introduction Results Discussion Materials and methods References

Ubiquitin (Ub) is a small polypeptide that is covalently conjugated to protein substrates, thus committing these proteins for degradation (Ciechanover et al. 1978; Hershko et al. 1980; Wilkinson et al. 1980). Ub conjugation is catalyzed by an enzymatic cascade that begins with the ATP-dependent activation of Ub by a Ub-activating enzyme (E1) to form an E1-Ub thiolester. The activated Ub is then transferred to a Ub-conjugating enzyme (E2 or Ubc). Finally, in the presence of a Ub-protein ligase (E3), the carboxyl terminus of Ub is conjugated via an isopeptide bond to a lysine residue of the protein substrate. Processive conjugation of Ub to a previously conjugated Ub results in the formation of multiubiquitin chains that target the protein substrate for degradation by the 26S proteasome (Chau et al. 1989; for review, see Pickart 1997). In most cases, the specificity of protein degradation is determined by the identity of E3, which is operationally defined as a factor that binds to a specific protein substrate and facilitates its multi-ubiquitination.

Ub-dependent proteolysis plays a pivotal role in the regulation of many biological processes, including cell cycle progression, transcription, and signal transduction (for review, see Hochstrasser 1996; Hershko and Ciechanover 1998). The importance of the Ub pathway in the cell cycle is highlighted by recent studies on the degradation of cyclins and cyclin-dependent kinase (Cdk) inhibitors. Exit of cells from mitosis requires the degradation of mitotic cyclins, a step that is controlled by the activation of a 20S E3 complex known as anaphase-promoting complex (APC) or cyclosome (Hershko et al. 1994; King et al. 1995). Interestingly, the degradation of Cdk inhibitors such as Sic1, which triggers the entry of cell cycle into the S phase, is regulated by a distinct mechanism. In this case, phosphorylation of Sic1 at the end of G₁ allows this inhibitor to bind to Cdc4, a protein that contains two structural motifs, an F-box at the amino terminus and seven WD40 repeats at the carboxyl terminus. Through the F-box, Cdc4 tethers phosphorylated Sic1 to Skp1, which in turn binds to Cdc53, which then recruits an E2, Cdc34, to ubiquitinate Sic1 (Feldman et al. 1997; Skowyra et al. 1997). This so-called SCF pathway is also responsible for the ubiquitination of several other substrates (for review, see Patton et al. 1998). In each case, the substrate specificity is determined by the presence of a distinct F-box protein. For example, the F-box/leucine-zipper protein Grr1 binds to phosphorylated Cln1 and Cln2, but not Sic1 (Skowyra et al. 1997).

An F-box/WD40-repeat-containing protein called Slimb was identified recently in a genetic screen for recessive mutations that alter adult patterning in Drosophila (Jiang and Struhl 1998; Theodosiou et al. 1998). Loss of function of slimb causes supernumerary limbs as a result of ectopic activation of the Hedgehog (Hh) and Wnt/Wingless (Wg) pathways. In the Hh pathway, the transcription factor Cubitus interruptus (Ci) is proteolytically processed to a truncated repressor form in the absence of signaling (Aza-Blanc et al. 1997). This processing depends on protein kinase A (PKA) activity, which is antagonized by Hh signaling. By analogy to the SCF pathway in yeast (Feldman et al. 1997; Skowyra et al. 1997) and to the processing of NF-B1/p105 in mammals (Palombella et al. 1994), it was proposed that in the absence of Hh signaling, PKA phosphorylates Ci, thereby targeting Ci for Slimb-dependent processing via the Ub-proteasome pathway (Jiang and Struhl 1998). Similarly, the Wnt/Wg pathway is also regulated primarily through the stability of -catenin/Armadillo (Arm), a putative cofactor of the transcriptional activator Lef1/TCF (Nusse 1997). In the absence of Wnt/Wg, -catenin/Arm is phosphorylated by glycogen synthase kinase-3 (Gsk-3) or Zeste-White 3 (Zw3) and then degraded via the Ub-proteasome pathway (Aberle et al. 1997; Orford et al. 1997). Activation of the Wnt/Wg pathway leads to inhibition of Gsk-3/Zw3, thus allowing for the accumulation of -catenin/Arm to turn on downstream genes in conjunction with Lef-1/TCF. The ectopic activation of the Wnt/Wg pathway and stabilization of Arm in slimb mutant cells suggests that Slimb may be required for the degradation of -catenin/Arm.

The Gsk-3 phosphorylation sites on -catenin are strikingly similar to those of IB, a family of inhibitory proteins that sequester the transcription factor NF-B in the cytoplasm of quiescent cells (for review, see Baldwin 1996; Baeuerle and Baltimore 1996). In response to a variety of stimuli, such as tumor necrosis factor  (TNF), lipopolysaccharide (LPS) and ultraviolet light (UV), IB proteins are phosphorylated rapidly at specific serine residues by a 700-kD protein kinase complex (for review, see Maniatis 1997; Stancovski and Baltimore 1997; Scheidereit 1998). Phosphorylation of IB at serines 32 and 36 targets this inhibitor for ubiquitination at lysines 21 and 22 (Chen et al. 1995; Scherer et al. 1995). Ubiquitinated IB is then degraded specifically by the 26S proteasome, allowing NF-B to translocate into the nucleus. Two closely related E2s, Ubc4/5 and Ubch7/E2-F1, are capable of supporting the ubiquitination of IB in vitro (Alkalay et al. 1995; Chen et al. 1996). However, the E3 responsible for IB ubiquitination has remained unknown.

Recently, a human homolog of Slimb, h-TrCP, was cloned in a yeast two-hybrid screen using human immunodeficiency virus (HIV) Vpu as a bait (Margottin et al. 1998). It was reported that h-TrCP binds specifically to phosphorylated Vpu, which in turn binds to CD4 on T cells, resulting in the degradation of CD4 in the endoplasmic reticulum (ER). This study, however, did not reveal the function of h-TrCP in normal cells (not infected with HIV). The structural and functional properties of Slimb/TrCP described above led us to hypothesize that it is a component of IB-Ub ligase (E3_IB). This hypothesis is strongly supported by the evidence presented in this report.

	Results

Top Abstract Introduction Results Discussion Materials and methods References

TrCP binds to phosphorylated IB

We cloned a mouse homolog of Slimb (mTrCP, GenBank accession no. AF112979) based on two Slimb-related sequences from the mouse EST database (see Materials and Methods). mTrCP is 79% (387/485) identical to Drosophila Slimb and 98% (560/569) identical to the recently cloned human TrCP (hTrCP; Margottin et al. 1998), whose normal physiological function was unknown. As an initial step in determining whether TrCP binds to IB, we synthesized ³⁵S-labeled IB and mTrCP by in vitro translation. IB was phosphorylated by a MEKK1-activated IB kinase complex (Lee et al. 1997) and then incubated with mTrCP. The binding of mTrCP to IB was determined by using a co-immunoprecipitation assay with an IB-specific antibody. As shown in Figure 1A, phosphorylated IB (p-IB) bound to mTrCP, whereas unphosphorylated IB or the phosphorylation-defective IB mutant (S32A/S36A) was unable to bind to mTrCP (Fig. 1A, lanes 1-3). The binding of p-IB to mTrCP was detected under high stringency conditions (1% NP-40, 0.5% deoxycholate, and 0.1% SDS), suggesting a strong interaction. Similar results were obtained when IB was phosphorylated by recombinant IKK expressed from baculovirus-infected insect cells (data not shown). Drosophila Slimb and hTrCP (kindly provided by Dr. Benarous, INSERM, Paris, France) also bound specifically to p-IB (data not shown), suggesting that Slimb/TrCP functions are evolutionarily conserved.

View larger version (91K): [in this window] [in a new window]   Figure 1.   mTrCP binds to phosphorylated IB. (A) In vitro binding. In vitro-translated 35S-labeled IB (lanes 1,2,4,5) or S32A/S36A mutant (lanes 3,6) was phosphorylated by a MEKK1-activated IB kinase complex and incubated with 35S-labeled mTrCP (lanes 1-3) or mTrCPF (lanes 4-6) in RIPA buffer plus 0.1% SDS. Proteins associated with IB were coimmunoprecipitated with an IB-specific antibody and analyzed by SDS-PAGE. Lanes 7-10 show an aliquot of the in vitro-translated proteins, which is ~40% of input in the immunoprecipitation experiments shown in lanes 1-6. (WT) Wild-type IB; (AA) S32A/S36A; (FL) full-length mTrCP (wild type); (F) F-box-deleted mutant of mTrCP; (IP) immunoprecipitation; (IVT) in vitro translation. (B) mTrCP forms a complex with phosphorylated IB, p65, and Skp1 in vivo. 293 cells were transfected with 3 µg of pcDNA3-Myc-mTrCP (lanes 2,3,6,7,10,11) or pcDNA3-Myc-mTrCPF (lanes 4,5,8,9,12,13). After incubating the cells with 20 µM of MG132 for 30 min, the cells were treated with (lanes 3,5,7,9,11,13) or without (lanes 2,4,6,8,10,12) calyculin A (0.1 µM) for 10 min. Cell extracts were immunoprecipitated with antibodies against IB (lanes 2-5), p65 (lanes 6-9), or Myc (lanes 10-13), respectively, and the precipitated proteins analyzed by immunoblotting with antibodies against Myc, Skp1, IB, and p65, respectively. Lane 1 is 10 µg of 293 cell extract that also expresses mTrCP. (C) mTrCP binds directly to p-IB. 293 cells were transfected with pcDNA3 containing Myc-mTrCP, together with Flag-IB (F-IB, lanes 1,2) or Flag-IBN mutant (FN, lanes 3,4). Following treatment of the transfected cells with MG132 and calyculin A, the cell extracts were immunoprecipitated with an anti-Flag antibody (M2, Kodak), and the precipitated proteins analyzed by immunoblotting with antibodies against Myc, Skp1, IB and p65, respectively.

To determine whether mTrCP binds to p-IB in vivo, 293 cells transfected with Myc-tagged mTrCP were stimulated with calyculin A, a cell-permeable phosphatase inhibitor that allows for the accumulation of phosphorylated IB (Chen et al. 1995). To block the degradation of phosphorylated IB, cells were pretreated with the proteasome inhibitor MG132 before the addition of calyculin A. Cell extracts were then immunoprecipitated with an antibody against IB, followed by immunoblotting with anti-Myc. The presence of Myc-mTrCP in the anti-IB precipitates was detected only when cells were stimulated with calyculin A (Fig. 1B, lanes 2,3). Conversely, when cells were immunoprecipitated with anti-Myc, only p-IB but not unphosphorylated IB was coprecipitated (Fig. 1B, lanes 10,11). Thus, mTrCP binds specifically to p-IB in cells.

Both IB and p-IB are bound tightly to NF-B, which is typically a heterodimer of p50 and p65 (for review, see Baldwin 1996). Similarly, the F-box protein TrCP is likely to be part of a SCF complex that also includes Skp1 (Jiang and Struhl 1998; Margottin et al. 1998). Hence, our observation that p-IB binds to TrCP raises the possibility that the NF-B/p-IB complex might associate with a SCF complex that contains both TrCP and Skp1. To test this possibility, we examined the presence of p65 and Skp1 in anti-IB and anti-Myc immunoprecipitates by immunoblotting with respective antibodies (Fig. 1B, lanes 2,3,10,11). Moreover, we immunoprecipitated calyculin A-stimulated cell extracts (see above) with a p65-specific antibody and then immunoblotted the precipitates with antibodies against Skp1, Myc, IB, and p65, respectively (Fig. 1B, lanes 6,7). In each case, when cells were stimulated, all four proteins (p-IB, p65, mTrCP, and Skp1) were detected in the same precipitates (Fig. 1B, lanes 3,7,11), indicating that they are present in the same complex. In contrast, in the absence of calyculin A treatment, while IB remained bound to p65, neither IB nor p65 was present in the anti-Myc precipitates (Fig. 1B, lanes 2,6,10, top two panels). Likewise, when cells were not stimulated, mTrCP remained bound to Skp1, but neither was found in the anti-IB or anti-p65 immunoprecipitates (Fig. 1B, lanes 2,6,10, bottom two panels). These results strongly suggest that phosphorylation of IB leads to the assembly of a multiprotein complex that contains minimally p-IB, p65, TrCP, and Skp1.

It has been proposed that the F-box mediates the binding to Skp1 of several F-box proteins including Cdc4 (Feldman et al. 1997; Skowyra et al. 1997) and hTrCP (Margottin et al. 1998). In an effort to generate a dominant-negative mutant of mTrCP that might allow us to investigate its function in vivo, we deleted the F-box from mTrCP and tested the ability of this mutant (mTrCPF) to bind to p-IB and Skp1, respectively. Deletion of the F-box did not prevent the binding of mTrCP to p-IB either in vitro (Fig. 1A, lane 5) or in vivo (Fig. 1B, lane 5,9,13, top panel). However, this deletion abolished the binding of mTrCP to Skp1 (Fig. 1B, cf. lanes 10 and 12, bottom panel) and compromised the recruitment of p-IB/p65 to a complex containing Skp1 (Fig. 1B, cf. lanes 7 and 9, bottom panel).

Interestingly, both mTrCP and mTrCPF appeared to be phosphorylated when cells were stimulated with calyculin A (Fig 1B, lanes 11,13; also see Fig 1C, lane 2). Furthermore, calyculin A treatment leads to a weak but detectable binding of mTrCPF to Skp1 (Fig. 1B, cf. lanes 12,13, bottom panel). It is possible that when it is phosphorylated, mTrCP itself is recruited for ubiquitination by a SCF complex that includes Skp1. In this case, the F-box may be dispensable, as the binding can be mediated by the interaction between certain phosphorylation sites on TrCP and a SCF complex. Ubiquitination and degradation of other F-box proteins such as Cdc4 and Grr1 within the SCF complexes has been reported recently (Zhou and Howley 1998). We also noted a low level of Skp1 in the anti-IB precipitates from mTrCPF-expressing cells (Fig. 1B, cf. lanes 3 and 5, bottom panel). This may reflect a pool of endogenous p-IB/TrCP/Skp1 complexes that were not displaced by mTrCPF.

We noticed that phosphorylated p65 was also detected following calyculin A treatment (Fig. 1B, lane 7). It is not clear whether PKA or another cytokine-inducible kinase is responsible for the phosphorylation of p65 in this case (Zhong et al. 1997; Wang and Baldwin 1998). Notwithstanding this uncertainty, the observation that p65 is phosphorylated raises the possibility that phosphorylated p65 may bind to TrCP directly, whereas p-IB binds to TrCP indirectly through p65. To test this possibility, we transfected 293 cells with the Myc-tagged mTrCP expression construct, together with Flag-tagged IB (F-IB) or a phosphorylation-defective IB mutant lacking the amino-terminal 36 residues (F-IBN; Brockman et al. 1995). Following treatment with MG132 and calyculin A, cell extracts were immunoprecipitated with a Flag-specific antibody and then immunoblotted with anti-Myc or anti-Skp1 antibodies (Fig. 1C). Like endogenous IB, transfected F-IB associated with Myc-mTrCP and Skp1 when cells were stimulated with calyculin A (Fig. 1C, lane 2). We also observed a low but detectable level of binding between transfected F-IB and mTrCP in the absence of calyculin A treatment (Fig. 1C, cf. lanes 1 and 2). This may be due to overexpression of F-IB, a small but significant fraction of which may be phosphorylated at the signaling sites even in the absence of calyculin A. In contrast to F-IB, transfected F-IBN did not co-immunoprecipitate with either Myc-mTrCP or Skp1 even when cells were treated with calyculin A (Fig. 1C, lanes 3,4). However, F-IBN remained bound to p65. Therefore, p65 is recruited to a TrCP/Skp1-containing complex by virtue of its association with p-IB, implying that p-IB binds to TrCP directly. Further supporting this idea, unphosphorylated p65 was present in the anti-Myc immunoprecipitates, whereas only p-IB coprecipitated with Myc-mTrCP (Fig. 1B, lanes 11,13).

We also found that the electrophoretic mobility of F-IBN was reduced slightly when cells were treated with calyculin A (Fig. 1C, cf. lanes 3 and 4), which is indicative of phosphorylation outside the inducible amino-terminal phosphorylation sites. However, phosphorylation at noninducible sites did not lead to the binding of IB to mTrCP or Skp1, suggesting that the binding of mTrCP to IB is strictly dependent on signaling and that mTrCP does not simply bind to phosphorylated proteins indiscriminately.

Dominant-negative TrCP blocks IB degradation and NF-B activation

The ability of mTrCPF to bind to p-IB, together with its inability to bind to Skp1, suggests that this mutant might be a dominant-negative inhibitor of IB degradation, provided that TrCP is involved in this pathway. To test the in vivo function of mTrCP, 293 cells were transfected with mTrCPF, together with a Flag-tagged IB expression construct. The transfected cells were stimulated with TNF for 30 min, and the degradation of transfected and endogenous IB analyzed by immunoblotting with antibodies specific for Flag or IB, respectively (Fig. 2). With increasing concentration of mTrCPF, there was a concentration-dependent stabilization of the phosphorylated forms of both transfected and endogenous IB. High concentrations of wild-type mTrCP also led to a slight stabilization of p-IB (data not shown). This finding may be explained if some of the overexpressed mTrCP is not incorporated into a functional SCF complex but still binds to p-IB. The sequestered p-IB may be protected from ubiquitination.

View larger version (33K): [in this window] [in a new window]   Figure 2.   TrCP is required for TNF-induced degradation of IB. 293 cells were transfected with pCDNA3-Flag-IB (20 ng), together with increasing concentration of pcDNA3-mTrCPF (lanes 1,2, vector only; lanes 3,4, 0.01 µg; lanes 5,6, 0.1 µg; lanes 7,8, 1 µg). After treatment of cells with or without TNF (20 ng/ml) for 30 min, cell extracts were analyzed by immunoblotting with antibodies against Flag, IB, or Myc, respectively.

To investigate the role of TrCP in NF-B activation, we transfected 293 cells with a mTrCPF expression construct, together with a luciferase reporter gene, which is under the control of three tandem repeats of NF-B binding sites (Fig. 3A). As a control, we also examined the expression of a GAL4-dependent reporter gene. TNF-induced expression of the NF-B reporter, but not GAL4VP16-activated expression of the GAL4 reporter, was severely inhibited with increasing concentration of mTrCPF. This result suggests that TrCP is required for TNF-induced activation of NF-B.

View larger version (20K): [in this window] [in a new window]   Figure 3.   Dominant-negative TrCP inhibits the activation of NF-B-dependent transcription. (A) Effect of mTrCPF on induction of the NF-B-dependent reporter gene by TNF. 293 cells were transfected with pcDNA3-mTrCPF (0.01-3 µg as indicated), together with a NF-B-dependent () or GAL4-dependent () luciferase reporter construct (p[B]3-tk-Luc, or Gal4-E1b-Luc, 10 ng each). The NF-B reporter gene was stimulated with TNF (20 ng/ml) for 6 hr prior to harvest, whereas the GAL4 reporter gene was activated by cotransfection with the strong activator GAL4-VP16 (200 ng). Reporter activity is expressed as fold induction of normalized luciferase activity in stimulated cells relative to that of unstimulated cells. TNF stimulated the NF-B reporter by 14-fold, whereas the GAL4-VP16 activation was >3000-fold. (B) Effect of mTrCPF on the activation of NF-B by IL-1, NIK, MEKK1, and IKK. 293 cells were transfected with pcDNA3-mTrCPF and p[B]3-tk-Luc as described in A, together with 100 ng of NIK, MEKK1, and IKK expression constructs, respectively. Stimulation of cells with IL-1 (10 ng/ml) was carried out for 6 hr prior to harvest. Luciferase activity was determined and normalized as described in Materials and Methods.

Degradation of IB is required for NF-B activation by many different stimuli (for review, see Baldwin 1996). If TrCP is an obligatory component that mediates IB degradation, it is expected that interference of TrCP function should compromise the activation of NF-B by multiple stimuli. To address this possibility, we examined the effects of mTrCPF on the induction of NF-B reporter gene by interleukin-1 (IL-1), NIK, MEKK1, and IKK, respectively (Fig. 3B). In each case, NF-B activation was inhibited markedly by mTrCPF, strongly suggesting that TrCP participates in the activation of NF-B by multiple signaling pathways.

Slimb is required for Dorsal-dependent activation of twist and snail in Drosophila embryos

Although the cell culture experiments shown above support the involvement of TrCP/Slimb in NF-B activation, it is imperative to determine whether the same conclusion can be reached in animal models that are amenable to genetic manipulations. In this regard, Drosophila embryos provide an excellent model system, not only because slimb-deficient embryos can be generated (Jiang and Struhl 1998) but also because there is a highly conserved signaling pathway in Drosophila that is analogous to that of NF-B/IB (for review, see Morisato and Anderson 1995). In Drosophila early embryos, dorsoventral patterning is established by a nuclear concentration gradient of the Dorsal morphogen, a homolog of NF-B. In the ventral region of the Drosophila embryo, local activation of Toll, a homolog of mammalian IL-1 receptor, results in degradation of the IB-like protein Cactus. Consequently, Dorsal translocates into the nucleus where it activates downstream genes that include twist (twi) and snail (sna) (for review, see Morisato and Anderson 1995). To explore the role of Slimb in Dorsal activation in vivo, we generated slimb-deficient embryos and examined the expression of twi and sna by whole-mount in situ RNA hybridization. As shown in Figure 4, wild-type embryos expressed twi and sna in the ventral region (Fig. 4A,C). In contrast, slimb-deficient embryos expressed markedly reduced levels of twi and sna in most of the ventral region (Fig. 4B,D). The residual expression of twi and sna at the posterior pole may be due to modification of the Dorsal/Cactus pathway by terminal signaling, such that reduced dosage of Dorsal is sufficient to activate the polar expression of twi and sna (Ray et al. 1991). This staining pattern in slimb-deficient embryos is reminiscent of what has been described in cactus gain-of-function mutant embryos (Roth et al. 1991), strongly suggesting that Slimb/TrCP is required for Cactus/IB degradation in vivo.

View larger version (64K): [in this window] [in a new window]   Figure 4.   Slimb is required for the Dorsal-dependent activation of twi and sna in Drosophila embryos. Wild-type (A,C) and slimb mutant (B,D) embryos were hybridized with twi (A,B) or sna (C,D) antisense RNA probes to visualize their expression. Embryos were oriented with anterior to the left and dorsal up. Wild-type embryos at the blastoderm stage express the Dorsal target genes twi and sna in the ventral region (A,C). In contrast, slimb mutant embryos at the same stage show diminished expression of both twi and sna in most of the ventral region, with residual expression detectable at the posterior pole (arrows in B and D).

Slimb/TrCP is a component of IB-Ub ligase (E3_IB)

To demonstrate directly that TrCP is a component of E3_IB, we expressed Myc-tagged mTrCP in 293 cells by transient transfection and then purified mTrCP containing complex by immunoprecipitation using a Myc-specific antibody. The immunoprecipitates were used directly as the source of E3 in a reconstituted IB ubiquitination assay (Chen et al. 1995). The reconstituted system contains in vitro-translated ³⁵S-labeled IB phosphorylated by the MEKK1-activated IB kinase complex, recombinant p50/p65 (to form a complex with IB), purified E1, recombinant Ubch5 (which was shown previously to support the ubiquitination of IB in vitro), Ub, and ATP. The presence of Ubch5 in the reaction led to the formation of low molecular mass ubiquitinated IB even in the absence of any E3 (i.e., mono-ubiquitinated IB was evident in Fig. 5A, lanes 1 and 6, but not in lane 4). However, it has been shown that proteins bearing one or a few Ub molecules are poor substrates for the 26S proteasome (Chau et al. 1989). Only high molecular mass conjugates, whose synthesis usually requires E3s, can be degraded efficiently. As shown in Figure 5, addition of immunoprecipitates containing mTrCP to the reconstituted system led to efficient multi-ubiquitination of phosphorylated IB. In contrast, the phosphorylation-defective IB mutant (S32A/S36A) was not ubiquitinated. A control IgG failed to immunoprecipitate any Ub ligase activity from the same extracts. Similarly, immunoprecipitates containing mTrCPF were also unable to support the ubiquitination of p-IB, most likely because of the inability of this mutant to bind to Skp1 (Fig. 5A, bottom panel).

View larger version (29K): [in this window] [in a new window]   Figure 5.   TrCP promotes ubiquitination of phosphorylated IB in vitro. (A) Reconstitution of IB ubiquitination. 35S-labeled IB or mutants were incubated in an ATP-supplemented reconstitution system containing E1, Ubch5, Ub, p50/p65, IB kinase and MEKK1, as described in Materials and Methods. Ubiquitination was initiated by the addition of immunoprecipitated beads containing Myc-mTrCP, which was expressed in 293 cells. Ubiquitinated products were analyzed by SDS-PAGE following immunoprecipitation with an anti-p65 antibody, which coprecipitated IB, as well as ubiquitinated IB (Chen et al. 1995). (Lane 1) No E3 was added; (lane 2) immunoprecipitated Myc-mTrCPF was used in place of Myc-mTrCP; (lane 3) normal rabbit IgG instead of Myc antibody was used to precipitate Myc-mTrCP; (lane 4) no E2 was added; (lanes 5-8) Myc-mTrCP (FL) beads were used to ubiquitinate wild-type and mutant IB as indicated; (lanes 9-12) 293 cell extracts (S100) were used to ubiquitinate wild-type and mutant IB as indicated. An aliquot of the immunoprecipitated beads used in the ubiquitination reaction was analyzed by western blotting using antibodies against Myc (middle panel) or Skp1 (bottom panel). (B) Quantitation of IB ubiquitination. Ubiquitinated IB shown in A was quantitated with the aid of PhosphorImager. Because only multiubiquitinated IB are degraded efficiently by the 26S proteasome, Ubn-IB containing more than two Ub units (n > 2) was quantitated and expressed as the percentage of total radioactivity in each lane.

It has been shown previously that mutation of lysines 21 and 22 of IB impair the ubiquitination and degradation of IB significantly, suggesting that these two lysines are the primary ubiquitination sites on IB (Scherer et al. 1995). We examined whether mTrCP immunoprecipitates could ubiquitinate IB at specific sites in the reconstituted assay. As shown in Figure 5, the IB mutant in which lysines 21 and 22 were mutated to arginine (K21R/K22R) was only weakly ubiquitinated (~40% of wild type; Fig. 5B), consistent with its weaker dominant-negative effect on NF-B activation than the S32A/S36A mutant (Scherer et al. 1995). In contrast, mutation of the nearby lysines 38 and 47 (K38R/K47R) did not impair the ubiquitination of IB in the reconstituted system. Therefore, mTrCP not only binds specifically to p-IB but also promotes the ubiquitination of phosphorylated IB at the physiologically relevant lysine residues in the presence of E1 and E2, thus fulfilling the major criteria for a bona fide E3_IB.

	Discussion

Top Abstract Introduction Results Discussion Materials and methods References

In this report we have shown that TrCP/Slimb exhibits several critical features expected of the substrate recognition subunit of IB-Ub ligase (E3_IB). First, TrCP binds specifically to phosphorylated, but not unphosphorylated, IB, both in vitro and in vivo. Second, a dominant-negative TrCP mutant blocks the signalinduced degradation of IB and the activation of NF-B. Third, Drosophila slimb-deficient embryos fail to activate twi and sna, two genes regulated by Dorsal, the Drosophila homolog of NF-B. Finally, a TrCP-containing complex ubiquitinates phosphorylated IB in the presence of E1 and Ubch5.

TrCP/Slimb is an F-box protein and is likely to function within a SCF complex that serves as an E3 for IB. Deletion of the F-box abolishes the binding of TrCP to Skp1 and also abolishes its ability to support the ubiquitination of p-IB in vitro (Fig. 5A). Three F-box proteins involved in the ubiquitination of cell cycle proteins in yeast, including Cdc4, Grr1, and Met30, function as part of SCF complexes (Patton et al. 1998). In all three cases, the role of F-box is to mediate binding to Skp1, which in turn binds to Cdc53, which recruits an E2 (Cdc34) to the SCF complex. Based on evidence presented in this report and by analogy to the SCF complexes involved in cell cycle, we propose a model suggesting that a TrCP-containing SCF complex is responsible for the signal-induced ubiquitination of IB (Fig. 6). In this model, TrCP associates with Skp1, which in turn binds to a Cdc53-like protein whose identity remains to be determined. When cells are stimulated with NF-B agonists, IB is phosphorylated at serines 32 and 36 by an IB kinase complex. Phosphorylated IB is recruited to the SCF^TrCP complex through its binding to TrCP. An E2, such as Ubch5, binds to the SCF complex and ubiquitinates the nearby IB at lysines 21 and 22. Because of the high-affinity binding of p-IB to TrCP, ubiquitinated IB is not released from SCF and is processively multi-ubiquitinated to form a multi-Ub chain. Multi-Ub chains on IB recruit the 26S proteasome to degrade IB, allowing NF-B to translocate into the nucleus, where it activates target genes.

View larger version (12K): [in this window] [in a new window]   Figure 6.   A proposed SCF pathway for the signal-induced ubiquitination of IB. TrCP/Slimb associates with Skp1 and another protein (possibly Cdc53-like) to form a SCFTrCP complex. Upon phosphorylation of IB by an IB kinase complex, TrCP recruits IB to the SCF complex, allowing the associated E2, such as Ubch5, to ubiquitinate IB. Following ubiquitination, IB is rapidly degraded by the 26S proteasome, allowing NF-B to translocate into the nucleus where it activates gene transcription. NF-B is represented by a heterodimer of p50 and p65, whereas a multiubiquitin chain is represented by the branch structure on IB.

Two components in this SCF^TrCP pathway remain to be identified. First, what is the third component of the SCF^TrCP complex? It is now known that there is a large family of Cdc53-related proteins belonging to the Cullin family (Jackson 1996; Kipreos et al. 1996). Further work is needed to determine whether a mammalian homolog of Cdc53 or a distinct member of the family is involved in the assembly of the SCF^TrCP complex. Second, what is the identity of E2_IB? Although we and others have shown that Ubch5 or Ubch7 can ubiquitinate IB in vitro (Alkalay et al. 1995; Chen et al. 1996), this does not rule out the possibility that other E2s might also be involved. It remains to be determined which E2 (or E2s) functions in the signal-induced ubiquitination of IB in vivo.

Another important question concerns the dynamics of SCF^TrCP complex assembly and substrate binding. Coimmunoprecipitation experiments showed that TrCP is associated with Skp1 regardless of signaling (Fig. 1), suggesting that SCF^TrCP is a preexisting complex in cells and that this complex binds further to p-IB/NF-B to form a larger complex upon signaling. This dynamic assembly process is consistent with the previous report that IB eluted as part of a high molecular weight complex when TNF-stimulated cell extracts were fractionated by gel filtration (Yaron et al. 1997). The dynamic nature of SCF complex assembly is underscored further by the recent finding that several F-box proteins in yeast are short-lived as a result of their own ubiquitination within SCF complexes (Zhou and Howley 1998). The rapid turnover of F-box proteins may provide an opportunity for different F-box proteins to compete for a limited pool of core SCF components such as Skp1. It is not known at present whether and how this combinatorial assembly of SCF complexes is regulated. The rapid degradation of F-box proteins also raises the possibility that these proteins might be limiting factors that control the rate of degradation of their target proteins. This may explain why overexpressed IB is not degraded efficiently in response to signals despite efficient phosphorylation (Traenckner et al. 1995).

Ubiquitination of IB by the TrCP-containing complex in vitro occurs primarily at lysines 21 and 22, and is strictly dependent on its phosphorylation at serines 32 and 36, thus recapitulating the in vivo setting (Fig. 5). Interestingly, the specificity of IB ubiquitination is compromised when it is not bound to NF-B (Z.J. Chen, unpubl.). For example, free IB mutants (i.e., S32A/S36A) can be ubiquitinated in the reconstituted system, albeit much more weakly. It has been shown that free IB is a short-lived protein that can be stabilized by binding to NF-B (Scott et al. 1993). The binding of NF-B to IB is mediated primarily through the ankyrin repeats of IB, which encompasses the bulk of the molecule (Haskill et al. 1991). It is therefore possible that NF-B masks the majority of the IB molecule except for the amino-terminal regulatory sequence, thus rendering the degradation of bound IB dependent on signaling. The binding of NF-B to IB may also minimize the accessibility of lysine residues to ubiquitination enzymes, thereby facilitating the ubiquitination of IB at specific lysine residues. This may explain why IB is one of the few proteins in which ubiquitination sites can be defined (Scherer et al. 1995).

TrCP/Slimb has now been implicated in the ubiquitination of several proteins, including IB/Cactus, -catenin/Arm, CD4 (through HIV Vpu), and Ci. However, IB is the only protein so far shown to be directly ubiquitinated by the TrCP-containing complex in vitro. Strikingly, the phosphorylation sites among Vpu, IB, and -catenin are very similar, with a minimal consensus sequence of DSG-S ( represents a hydrophobic residue). Although the serine residues of these proteins appear to be phosphorylated by distinct kinases, the fates of these proteins after phosphorylation are likely to be the same in terms of their binding to TrCP/Slimb and subsequent ubiquitination and degradation. The only exception is Vpu, which does not appear to be degraded. Instead, it was proposed that Vpu targets its cognate partner CD4 to the ER degradation pathway (Margottin et al. 1998). It will be important to determine whether Vpu or CD4 is ubiquitinated to understand why and how Vpu escapes degradation by the SCF pathway.

The potential involvement of TrCP/Slimb in the NF-B, Hh, and Wg raises the exciting possibility that these divergent pathways may in fact be interconnected and that TrCP/Slimb may be deployed from one pathway to another to allow for integration of these pathways in response to signals. Further studies on TrCP/Slimb may provide another avenue for modulating the activity of these pathways, all of which have been implicated in several human diseases, including cancer.

	Materials and methods

Top Abstract Introduction Results Discussion Materials and methods References

Plasmids, antibodies, and chemicals

cDNAs encoding IB and mutants (S32/36A, N, K21/22R, K38/47R) were gifts of Dr. Dean Ballard (Vanderbilt University, Nashville, TN) and have been described previously (Brockman et al. 1995; Chen et al. 1995; Scherer et al. 1995). A luciferase reporter construct containing three tandem repeats of NF-B binding sites (p[B]₃-TK-Luc) was a gift of Dr. Shigeki Miyamoto (University of Wisconsin, Madison). Rabbit anti-IB (C15 and C21), anti-Myc (A14), anti-Skp1 (H163), and goat anti-p65 (C20G) polyclonal antibodies, as well as mouse anti-Myc monoclonal antibody (9E10), were purchased from Santa Cruz Biotechnology. Mouse anti-Flag antibodies (M2 and M5) were from Eastman Kodak. MG132 (z-Leu-Leu-Leu-H) was purchased from Calbiochem. Calyculin A and okadaic acid were purchased from Alexis.

Proteins

Ub was purchased from Sigma. Ub aldehyde (Ubal) was produced by periodate oxidation of Ubdiol, which was synthesized using the carboxylpeptidase Y method (Lam et al. 1997). Ubch5 was derived from GST-Ubch5 (Chen et al. 1996) by thrombin cleavage and then purified to apparent homogeneity by MonoS fast performance chromatography (FPLC, Pharmacia). E1 was purified from calf thymus by covalent affinity chromatography on Ub-Sepharose. Recombinant (His)₆-MEKK1 was purified from baculovirus-infected Sf9 cells as described previously (Lee et al. 1997; baculovirus provided by Drs. Frank Lee and Tom Maniatis, Harvard University, Cambridge, MA). IB kinase complex was partially purified as described previously, except that calf thymus instead of HeLa cells was used as the source of the kinase. The kinase activity from the Superdex fractions can be activated by ubiquitination (Chen et al. 1996), by MEKK1 (Lee et al. 1997), or by NIK (Z.J. Chen, unpubl.), and was used for in vitro phosphorylation of IB at serines 32 and 36. Further purification and characterization of the IB kinase complex will be reported elsewhere. GST-p50 and p65-His₆ were expressed from Escherichia coli and purified as described previously (Thanos and Maniatis 1992). To allow for the formation of p50/p65 heterodimer, these proteins were mixed in the presence of buffer A (20 mM Tris at pH 7.5, 0.5 mM DTT, 0.1 mM EDTA, 0.1 mM EGTA, 0.1 mM PMSF, 0.1 M NaCl) plus 6 M urea and then dialyzed against the same buffer overnight. The proteins were renatured by step-wise reduction of urea concentration to 4 M, 2 M, and finally no urea in buffer A. In some cases, the proteins were further purified by glutathione chromatography. The functionality of the renatured p50/p65 heterodimer was confirmed by its ability to coimmunoprecipitate in vitro-translated ³⁵S-labeled IB.

cDNA cloning of mTrCP and mTrCPF

A BLAST search of the EST database for mammalian homologs of Slimb (Jiang and Struhl 1998) and Xenopus TrCP (Spevak et al. 1993) identified two clones (AA197590 and AA033076) that together encompass ~370 amino acids corresponding to the carboxyl terminus of the mouse homolog of Slimb/TrCP (mTrCP). PCR primers derived from these two clones (5'GCGGTCGACCGTCAGGACGGACTCTCTGTGG-3' and 5'-AGTGCGGCCGCTTATCTGGAGATGTAGGTGTA-3') were used to amplify a 1.1-kb fragment (mTrCPN) from a mouse E8.5 (embryonic day 8.5) cDNA library in ZAPXR (Stratagene). The 5'-coding region of mTrCP was amplified from the same library by PCR using T7 primer and a reverse mTrCP primer (5'-AGTGCGGCCGCGAGCTTTTTCCACAGCATGCC-3'). A 650-bp fragment was subcloned into BamHI and NotI sites of pBluescript (Strategene) and sequenced. This fragment contains the full-length 5'-coding sequence plus 60 bp of 5'-untranslated sequence. To generate a full-length mTrCP that is fused in-frame with five tandem repeats of amino-terminal Myc epitopes, the amino terminus of mTrCP was amplified by PCR using the following primers: 5'-CGCCCATGGACCCGGCAGAGGCGGTG-3' and 5'-CGCTCTGCCAGGCCTCGCCACAGA-3'. A 600-bp fragment was subcloned into the NcoI and StuI sites of pSK-Myc-slimb (Jiang and Struhl 1998) to replace the amino terminus of Slimb. The carboxyl terminus of Slimb was subsequently replaced with a StuI-NotI fragment (1.1 kb) of mTrCPN. Following partial digestion with KpnI and NotI, the full-length Myc-mTrCP fragment (~1.9 kb) was subcloned into pcDNA₃. The entire co