Processing of p105 is inhibited by docking of p50 active subunits to the ankyrin repeat domain, and inhibition is alleviated by signaling via the carboxyl-terminal phosphorylation/ ubiquitin-ligase binding domain.

Processing of the p105 precursor to generate the p50 subunit of the nuclear factor kappaB transcription factor is an exceptional case in which the ubiquitin system is involved in limited processing rather than in complete destruction of the target substrate. A Gly-rich region "stop" signal in the middle of the molecule along with a neighboring downstream ubiquitination, and probably an E3 anchoring domain, have been demonstrated to be important for processing. In addition, we have shown that IkappaB kinase-mediated phosphorylation of the C-terminal domain leads to recruitment of the SCF(beta)-TrCP ubiquitin ligase with subsequent accelerated ubiquitination and processing/degradation of the precursor (Orian, A., Gonen, H., Bercovich, B., Fajerman, I., Eytan, E., Israël, A., Mercurio, F., Iwai, K., Schwartz, A. L., and Ciechanover, A. (2000) EMBO J. 19, 2580-2591). Here we show that processing of p105 molecules that contain more then four ankyrin repeats, but lack the C-terminal phosphorylation/ubiquitin ligase binding domain, is strongly inhibited by docked p50 subunits. Inhibition is caused by interference with the function of the proteasome, as conjugation is not affected. Inhibition is alleviated after IkappaB kinase phosphorylation of the C-terminal domain leads to accelerated, beta-TrCP-mediated ubiquitination and processing/degradation of p105. We suggest that under basal conditions, slow generation of p50 probably involves the mid-molecule ubiquitination/E3 recognition motif. Following stimulation, the C-terminal domain is involved in rapid processing/degradation of p105 with release of a large amount of the stored subunits that now become transcriptionally active.

The NF-B 1 dimeric transcription factors play key roles in basic processes such as regulation of the immune and inflammatory responses, development and differentiation, malignant transformation, and apoptosis (for recent review, see Ref. 1). Certain active subunits of NF-B are generated from inactive precursor molecules via limited, ubiquitin-and proteasomemediated processing. One established case is that of p50 that is generated from the p105 precursor (2)(3)(4). p50 is derived from the N-terminal domain of the molecule, whereas the C-terminal, ankyrin repeat-containing domain, is degraded (2). The processed subunits typically heterodimerize with members of the Rel family of regulators such as p65 (RelA), RelB, or c-Rel to generate the active heterodimeric transcription factor. According to the current model, binding of a member of the IB family of inhibitors generates an inactive heterotrimeric complex that is sequestered in the cytosol. Following stimulation, specific IB kinases (IKKs) are activated and phosphorylate IB on specific Ser residues. Phosphorylation leads to recruitment of the SCF ␤-TrCP ubiquitin ligase complex, rapid polyubiquitination, and subsequent degradation of the inhibitor by the 26 S proteasome. Following degradation of IB, active NF-B is translocated to the nucleus and initiates specific transcription (for recent review, see Ref. 5).
The ubiquitin pathway is involved, via specific degradation of short lived regulatory proteins, in regulation of a broad array of cellular processes. Among these are cell cycle progression, differentiation and development, and the immune and inflammatory responses. Degradation of a protein via the ubiquitin system involves two successive steps: first, processive formation of a polyubiquitin chain that is covalently anchored to the target substrate and second, degradation of the tagged protein by the 26 S proteasome. Conjugation involves activation and subsequent transfer of ubiquitin from the ubiquitin-activating enzyme, E1, to a member of the ubiquitin carrier proteins, E2s, family of enzymes. In most cases, E2 transfers the activated ubiquitin moiety to an ⑀-NH 2 group of an internal Lys residue in the target substrate that is specifically bound to a member of the ubiquitin-protein ligase family of proteins, E3. Subsequent processive conjugation of additional activated ubiquitin molecules to previously attached molecules generates the polyubiquitin chain that serves as a degradation signal for the 26 S proteasome. Several classes of E3s have been described, among them are the SCF complexes that recognize mostly phosphorylated proteins. The currently known tetrameric complexes are composed of Skp1, Cullin1⅐Cdc53, and Rbx1⅐Roc1⅐Hrt1, which are common to all SCFs, and a variable F-box protein that serves as the substrate-recognizing subunit (for recent reviews on the ubiquitin system and the proteasome, see Refs. 6 and 7, respectively; for a recent review on SCF complexes, see Ref. 8). The F-box protein is involved in recognition of the phosphorylated substrate. ␤-TrCP, for example, recognizes the common motif DS(P)GDS(P) shared by IB␣, IB␤, ␤-catenin, and HIV-Vpu (5).
The mechanisms involved in limited processing of the p105 precursor protein have been partially elucidated. Lin and Ghosh (9) have demonstrated that the GRR that spans residues 376 -404 in human p105 is essential for processing and serves as a processing stop signal for the 26 S proteasome (10). Several single residues that reside upstream to the GRR and are involved in proper folding of p50 are also essential for processing, most probably via inhibiting unfolding and entry into the proteasome (11). Processing requires also an additional adjacent downstream domain that contains Lys residues 441 and 442, which are important for ubiquitination, and an acidic region (residues 446 -454) that may function as an E3 recognition motif (10). These findings suggest that processing requires at least two motifs, a physical stop signal(s) and a ubiquitination/E3 recognition site. Fan and Maniatis (2) have shown that a truncated form pf p105, p60, can be processed to p50. Lin and colleagues (12) have shown that p105 can be processed cotranslationally, and synthesis of the complete molecule is not required for generation of p50. Taken together, these studies (2, 9 -12) suggest that all of the motifs that are required for processing in the resting cell are contained within the first ϳ550 amino acid residues. Other studies have suggested a role for phosphorylation of the C-terminal domain of p105 in regulated, signal-induced processing/degradation of the molecule (13,14). Heissmeyer and colleagues (15) have shown that IKK-mediated phosphorylation of Ser residues that reside in a sequence that spans amino acid residues 922-933 leads to rapid degradation of p105. We have recently shown that this IKK-mediated phosphorylation leads to recruitment of the SCF ␤-TrCP ubiquitin ligase. Consequently, the molecule is ubiquitinated and rapidly processed. A certain proportion of the molecules are completely degraded (16). A later study by Heissmeyer and colleagues has corroborated the finding that TrCP is indeed the E3 (17). The motif DS 923 VCDS 927 (17) is similar to the targeting motifs in IB␣, ␤, and ⑀, and in ␤-catenin and HIV-Vpu (5).
Interestingly, processing of p100, the gene product of nfB2, to yield the p52 subunit is mediated by a similar mechanism. Like p105, part of it may occur cotranslationally and requires the GRR (18). A recent study has demonstrated that phosphorylation of Ser residues 867 and 870 which is mediated by NIK (NF-B-inducing kinase) is required for processing (19). Although it is not clear whether NIK phosphorylates p100 directly, it is interesting to note that the phosphorylation site is similar to that of p105, IB␣, IB␤, ␤-catenin, and HIV-Vpu, where the two Serines are interspaced by three residues. A novel motif, processing inhibitory domain (PID), which resides between the ankyrin repeat domain and the phosphorylation site, negatively regulates processing (19).
It should be noted that the C-terminal segment of p105 which resides between the GRR and the IKK/TrCP motif contains seven ankyrin repeats. Active NF-B subunits such as p50 and p65 dock to this region, inhibit processing of the precursor molecule, and are sequestered in an inactive storage form in the cytoplasm. Thus, the ankyrin repeat domain serves as an inhibitor of NF-B activity (20 -22). Harhaj and colleagues (23) have shown that processing of newly synthesized p105 molecules is more efficient compared with that of the accumulated form that is already associated with p50. Accelerated, signal-induced processing/degradation leads to release of the docked active factors (15) and probably to generation of additional active subunits from the processed precursor (16). However, the mechanisms that underlie p50/p65-mediated inhibition of p105 processing and its subsequent signal-induced alleviation have remained obscure.

Materials
Materials for SDS-PAGE and Bradford reagent were from Bio-Rad. L-[ 35 S]methionine (Ͼ1,000 Ci/mmol; ϳ50 mCi/ml) for in vitro translation and prestained molecular weight markers were obtained from Amersham Pharmacia Biotech. Tissue culture sera and media were from Biological Industries, Bet hemeek, Israel, or from Sigma. Rabbit anti-NF-B1 p50 antibody that recognizes both p105 and p50, was from Santa Cruz, and peroxidase-conjugated goat anti-rabbit antibody was from Jackson ImmunoResearch Laboratories. Ubiquitin, dithiothreitol, ATP, ATP␥S, phosphocreatine, creatine phosphokinase, 2-deoxyglucose, glutathione, glutathione immobilized to agarose beads, isopropyl-␤-D-thiogalactopyranoside (IPTG), and Tris buffer were from Sigma. Hexokinase and Fugene TM 6 transfection reagent were from Roche Molecular Biochemicals. HEPES buffer and protease inhibitors mixture were from Calbiochem. Reagents for enhanced chemiluminescence (ECL) were from Pierce. A wheat germ extract-based coupled transcription-translation kit was from Promega. Restriction and modifying enzymes were from New England Biolabs. Oligonucleotides were synthesized by Biotechnology General, Rehovot, Israel. All other reagents were of high analytical grade.

Methods
Cell Lines-COS-7 and HeLa cells were grown at 37°C in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum and antibiotics (penicillin/streptomycin).
Plasmids and Construction of Mutants-The human WT and ⌬918 -934 p105 cDNAs used for in vitro translation (pT7␤105) and for transient transfection (in pCI-neo) were described previously (4,10,16). cDNAs for in vitro translation of proteins that code for the different species of truncated p105s which carry an increasing number of ankyrin repeats were generated by linearization of the cDNA that codes for p105-WT with the indicated restriction enzymes. The linear cDNAs were then incubated in the coupled transcription-translation mixture ( Fig. 1; the appropriate restriction sites and enzymes are marked). To be able to linearize also the cDNA that codes for intact p105-WT, an XhoI site was inserted downstream of the stop codon of the WT molecule. For cell expression, stop codons were introduced in the place of the appropriate restriction sites by site-directed mutagenesis using the QuikChange TM kit (Stratagene) (see Fig. 1; the number of residues in each construct is shown in brackets). These constructs are p105-Tth111 (1-545), p105-AatII (1-677), p105-ScaI (1-715), and p105-WT. These proteins contain 0, 4, 5, and 7 ankyrin repeats, respectively. p105-WT contains also the C-terminal domain that is phosphorylated by IKK and binds ␤-TrCP. p50-XbaI (1-502) for cell expression cloned into the Rc-CMV (cytomegalovirus) vector was obtained from Dr. Alain Israël, Pasteur Institute, Paris, France. For convenience, we designated proteins used for both in vitro translation and in vivo expression in cells according to the restriction site used. p105-Spe (1-340) for bacterial expression was generated by in-frame cloning of the cDNA fragment that codes for the first 340 amino acids of p105 into the pGEX3 vector (Amersham Pharmacia Biotech). A stop codon was introduced instead of the Spe site. p105-DHFR was generated by in-frame cloning of the full-length coding sequence of DHFR into the BsmI-NdeI site (653-794) after removal of the fragment that codes for residues 653-794. This resulted in removal of a sequence that contains ankyrin repeats 4-7. p105-⌬918-934 for both in vitro translation and cell expression, constitutively active IB kinase ␤ (IKK␤-SSϾEE), and ⌬F-box human ␤-TrCP (⌬F-␤-TrCP) for cell expression were as described (16). Sequence of all constructs was confirmed using the Applied Biosystems 310 autosequencer.
Expression of Recombinant p50-Spe in Bacteria-The p50-Spe protein was induced in BL21 Escherichia coli by isopropyl-␤-D-thiogalactopyranoside. After adsorption of the cell extract onto immobilized glutathione, the purified fusion protein was eluted by free glutathione according to the manufacturer's instructions.

Preparation of Cell and Nuclear and Cytoplasmic Extracts-HeLa
cell extract for monitoring in vitro conjugation and processing was prepared by hypotonic lysis as described previously (10,16). Nuclear and cytoplasmic extracts were prepared by hypotonic lysis and Nonidet P-40 extraction (cytosol) followed by treatment with a high glycerol/ high salt nuclear extraction buffer as described (24).
In Vitro Processing of p105-All p105 proteins were translated in vitro using the wheat germ extract-based coupled transcription-translation kit in the presence of L-[ 35 S]methionine according to the manufacturer's instructions. Processing of labeled p105 to p50 was monitored in HeLa cell extract as described (4,10,16). When p50-Spe was added, it was preincubated for 5 min at 30°C along with the labeled substrate prior to the addition of the remaining components of the processing reaction. After incubation, reaction mixtures were resolved via SDS-PAGE (10%). Gels were dried and analyzed by PhosphorImager (Fuji, Japan).
In Vitro Conjugation Assays-Adducts of ubiquitin with p105 or the p105 derivatives were generated in crude HeLa cell extract in an assay similar to that described for processing but with the following modifications as described previously (10, 16): 0.5 g of ubiquitin aldehyde, a specific inhibitor of certain isopeptidases, was added to the reaction mixture, and ATP␥S was used instead of ATP and the ATP-regenerating system.
Transient Transfections and Processing of p105 in Cells-COS-7 cells were transiently transfected with ϳ3 g of the WT or the various p105 mutant/deleted cDNAs. Where indicated, cells were cotransfected along with ϳ1.5 g of each of the cDNAs coding for the constitutively active IKK␤, p50-XbaI, and/or ⌬F-␤-TrCP. An empty vector was added, when necessary, to maintain an equal amount of DNA in all transfections. Transfection was carried out using the Fugene reagent. 48 h after transfection, cells were lysed, and cytosolic and nuclear fractions were prepared as described (24). Protein aliquots from each fraction representing an equal number of cells were resolved via SDS-PAGE (10%) and blotted onto nitrocellulose paper. Processing of p105 was monitored via Western blot analysis using anti-p50 antibody and ECL.
Protein Concentration-Protein concentration was determined according to Bradford (25) using bovine serum albumin as a standard.

Processing of C-terminally Deleted p105s Is Inhibited with
Increasing Number of Ankyrin Repeats-It has been shown that active NF-B subunits dock to the C-terminal domain of p105 and probably inhibit processing (15, 20 -23). Therefore, we wanted to test the hypothesis that the efficiency of processing is inversely correlated with the number of ankyrin repeats to which the active subunits dock. Further, we wanted to test whether alleviation of inhibition and supply of active subunits, derived from both the docked proteins and processed p105, require signal-induced targeting of p105 via the C-terminal domain of the molecule. To test these hypotheses, we constructed a series of p105 deletion mutants that contain an increasing number of ankyrin repeats, yet lack the C-terminal domain ( Fig. 1). We tested these constructs for both processing and conjugation in a cell-free reconstituted system. As can be clearly seen in Fig. 2, processing of the different C-terminally deleted p105 precursors is inhibited in proteins that contain more then four repeats. Thus, p105-Tth111 that does not contain any repeat is processed efficiently (13%). Processing of p105-AatII that contains four repeats is inhibited only slightly (12%). In contrast, processing of p105-ScaI that contains five repeats is much less efficient (6%). Yet, processing of p105-WT that contains seven ankyrin repeats, but also the intact Cterminal targeting domain, proceeds in a highly efficient manner (24%). It should be noted that even under basal, nonstimulated conditions, the C-terminal domain of p105 is phosphorylated (16,26), which leads to recruitment of ␤-TrCP and accelerated processing/degradation of p105 without additional stimulation (16). This may explain the high efficiency of processing of p105-WT in the HeLa extract.
Ankyrin Repeat-dependent Inhibition of p105 Processing in Vitro Is Probably Mediated by Anchored p50 Subunits, and Inhibition Is Alleviated by the Presence of the C-terminal Signaling Motif-We surmised that inhibition of processing which is observed with the increasing number of ankyrin repeats (Fig.  2) is caused by the presence of free p50 and other active NF-B subunits in the HeLa cell extract in which the processing reactions are carried out. To corroborate this hypothesis directly, we added bacterially expressed p50 to a cell-free proteolytic system. As can be seen in Fig. 3A, increasing the concentration of the free subunit had no effect on processing of p105-Tth111. The different p105 proteins were constructed and designated as described under "Experimental Procedures." Numbers in brackets denote the amino acid residues coded by the corresponding cell expression vectors. In these constructs, a stop codon was inserted after the C-terminal residue. Restriction enzymes denote the cleavage sites used for linearization of the corresponding cDNAs that were utilized for in vitro transcription/translation. For convenience, because the number of residues coded by the two groups of vectors is identical, we designated proteins used for both in vitro translation and cell expression according to the restriction site used. The different domains of p105 and the sequence coding for DHFR are marked and annotated in the figure and described under "Experimental Procedures." Processing of p105-AatII was inhibited only slightly. In contrast, processing of p105-ScaI was inhibited almost completely. Because the WT protein is processed efficiently in the presence of a complete cohort of ankyrin repeats (Fig. 2), we further hypothesized that inhibition is alleviated by signaling that is mediated by the C-terminal motif that leads to recruitment of the SCF ␤-TrCP E3 complex, ubiquitination, and processing/ degradation of the precursor molecule. To test this hypothesis, we monitored processing of p105-WT and p105-⌬918 -934 in the presence of an increasing concentration of exogenous p50. As can be seen in Fig. 3B, processing of the WT precursor was not affected, whereas that of the signaling motif-deleted protein was inhibited almost completely.
The Ankyrin Repeat Domain and Anchored p50 Do Not Interfere with Conjugation of p105-To dissect the mechanism(s) that underlies inhibition of processing by the ankyrin repeat domain and anchored p50 subunits, we examined the effect of exogenously added p50 on the efficiency of conjugation of the different p50 precursors. As can be seen in Fig. 4A, p105-Tth111, p105-ScaI, and p105-WT are equally conjugated. This similar conjugation efficiency is in striking contrast to the efficiency of processing: processing of p105-ScaI is significantly lower compared with that of its two other counterparts (Fig. 2). Similarly, conjugation of p105-ScaI was not affected by the addition of exogenous p50 (Fig. 4B). This is again in contrast to the inhibitory effect of added p50 on p105-ScaI processing (Fig.  3A). Not surprisingly, exogenous p50 had no effect on conjugation of p105-⌬918 -934 (Fig. 4C). This is again in sharp contrast to its strong inhibitory effect on processing of the signaling motif-deleted protein (Fig. 3B). Taken together, these findings suggest that anchored p50 probably affects processing by interfering with proteasomal activity (see "Discussion").
p50 Inhibits Processing of p105 in Vivo, and Inhibition Is Dependent on the Presence of the Ankyrin Repeat-containing Domain-To test the effect of the ankyrin repeat domain and anchored p50 subunits on processing of p105 in vivo, we overexpressed p50-XbaI in cells that do not express the NF-B cascade proteins. p50-XbaI cannot generate p50 cotranslationally because it is probably too short to allow folding of its processed tail and recognition by the proteasome (10,12). This molecule contains 502 amino acid residues and can be resolved readily from the native p50 subunit (ϳ435 residues) that is generated from p105. As can be seen in Fig. 5A, overexpressed p50 does not affect processing of p105-Tth111 (11%; compare lanes 1 and 2 with lanes 3 and 4). In contrast, exogenously expressed p50 inhibits processing of p105-WT by 2-fold (compare lanes 5 and 7). Interestingly, a large amount of p105-Tth111 is found in the nucleus (compare lanes 1 and 2 and  lanes 3 and 4) because the molecule lacks the ankyrin repeat domain. Consequently, the p50 subunits that usually anchor to the repeat domain cannot dock which leads to exposure of the NLS and to its subsequent translocation to the nucleus. Similarly, and probably for the same reason, a large proportion of the expressed p50 (p50-Xba) also migrates to the nucleus (com- pare lanes 3 and 4). Finally, the p50 subunit that is generated from p105-Tth111 is also distributed between the cytosol and the nucleus (compare lane 1 with lane 2 and lane 3 with lane 4). In contrast, the vast majority of p50 that is generated from To test further the role of the ankyrin repeat domain in binding and sequestering p50 in the cytosol, we generated p105 in which we replaced the last four ankyrin repeats with DHFR. The resulting protein contains only the first three repeats. The DHFR coding sequence, which replaces the ankyrin 4 -7 repeat domain, contains no ankyrin repeat, yet its insertion generates a p105 species with a molecular mass similar to that of the WT protein. Also, it enables testing of the effect of methotrexateinduced folding on processing, and examination of the possibility that the drug-induced folding can replace the docked p50 subunits in inhibiting p105 processing. As can be seen in Fig.   FIG. 4. The number of ankyrin repeats in p105 and the concentration of exogenously expressed p50 do not affect conjugation of the precursor molecules. Panel A, p50 precursors with an increasing number of ankyrin repeats are conjugated equally. In vitro translated and L-[ 35 S]methionine-labeled p105-Tth111 (no ankyrin repeats), p105-ScaI (five ankyrin repeats), and p105-WT (contains seven ankyrin repeats but also the C-terminal IKK phosphorylation and ␤-TrCP anchoring domain) were subjected to conjugation in a cell-free system in the presence of HeLa cell extract. Panel B, conjugation of p105-Tth111 and p105-ScaI is not inhibited by an increasing concentration of exogenous p50. In vitro translated and L-[ 35 S]methionine-labeled p105-Tth111 (no ankyrin repeats) and p105-ScaI (five ankyrin repeats; both proteins lack the C-terminal IKK phosphorylation and E3 anchoring domain) were subjected to conjugation in a cell-free system in the presence of HeLa cell extract and increasing concentrations of bacterially expressed p50 as indicated and as described under "Experimental Procedures" and in the legend to Fig. 3. Panel C, conjugation of p105-WT and p105-⌬918 -934 that lacks the IKK phosphorylation and ␤-TrCP anchoring domain is not inhibited by an increasing concentration of exogenous p50. In vitro translated and L-[ 35 S]methionine-labeled p105-WT (seven ankyrin repeats) and p105-⌬918-934 (contains seven ankyrin repeats but lacks the C-terminal IKK phosphorylation and E3 anchoring domain) were subjected to conjugation in a cell-free system in the presence of HeLa cell extract and increasing concentrations of bacterially expressed p50 as indicated and as described under "Experimental Procedures" and above. Adducts were resolved and visualized as described under "Experimental Procedures." Samples in panels A and C were resolved in 7.5% acrylamide gels, whereas in panel B, separation was carried out in a 10% acrylamide gel. Conj. denote conjugates.
FIG. 5. p50 inhibits in vivo processing of ankyrin repeat-containing, but not of ankyrin repeat-lacking p105s. Panel A, processing of p105 and nuclear translocation of p50 are inhibited by the presence of an ankyrin repeat domain that binds p50. COS-7 cells were transfected with cDNAs coding for p105-Tth111 (lanes 1-4) or p105-WT (lanes 1-8) as indicated. Control cells were mock transfected (lanes 9 and 10). When indicated, p50-Xba was cotransfected. 48 h after transfection, cells were harvested, and nuclear and cytosolic fractions were isolated as described under "Experimental Procedures." Aliquots of cytosolic and nuclear extracts representing an equal number of cells were resolved via SDS-PAGE, blotted onto nitrocellulose paper, and proteins were visualized using anti-p50 antibody and ECL as described under "Experimental Procedures." Data were quantified and are presented (percent of processing; p50/p50 ϩ p105). C denotes cytosolic fraction, and N denotes nuclear fraction. Sites of migration of all proteins are indicated. Panel B, substitution of four ankyrin repeats with DHFR abrogates the ability of p50 to inhibit processing of p105 and does not prevent nuclear translocation of free p50. COS-7 cells were transfected with cDNAs coding p105-DHFR (lanes 1-4; the DHFR replaces ankyrin repeats 4 -7) or p105-WT (lanes 5-8) in the absence or presence of cDNA coding for p50-Xba as indicated. 48 h after transfection, cells were harvested, and proteins were analyzed as described in the legend to panel A. 5B, all of the p50 that is generated from p105-DHFR migrates to the nucleus (compare lanes 1 and 2). Exogenously expressed p50 does not affect processing of p105-DHFR, and almost all of the population of the cellular p50, both the endogenously generated and the exogenously expressed, migrates to the nucleus (compare lanes 3 and 4). Interestingly, a large amount (ϳ50%) of p105-DHFR is transported to the nucleus as well, suggesting that nuclear translocation of P105 is inhibited, at least partially, by the anchored p50 subunits; species that cannot dock p50, even if they are similar in length to p105, can be translocated to the nucleus. Thus, anchoring of p50 prevents not only processing but also nuclear targeting. Our initial experiments indicate that addition of methotrexate has only a small effect on processing of p105-DHFR (not shown). This may suggest that the inhibitory effect of docked p50 on proteasomal processing of p105 is specific, and this effect cannot be mimicked by tight folding of the molecule induced by methotrexate. If however methotrexate induces tight folding of the C-terminal domain, the lack of effect of the drug on processing may suggest that processing occurs following folding of the molecule and entry of the loop that contains the processing site into the catolytic chamber of the proteasome. That in contrast to a model that suggested processive degradation that starts from the C-terminal residue and proceeds to the processing point (11).
We noted that in cells (Fig. 5, see also Figs. 6 and 7), unlike in the reconstituted cell-free system (Figs. 2 and 3), p50 is able to inhibit processing of p105-WT. It is possible that in the HeLa extract, p105-WT is phosphorylated by the endogenous IKK, which alleviates the inhibition caused by either endogenous or exogenously added p50. In COS cells that do not express the NF-B activation cascade, p50 is inhibitory. The finding that processing of p105-Tth111 and p105-DHFR is not affected by overexpression of p50 (Fig. 5A, lanes 1-4; Fig. 5B, lanes 1-4) rules out the possibility that the expressed p50 interferes with processing by sequestering certain factors necessary for this proteolytic event, such as elements of the ubiquitin-conjugation machinery.
The C-terminal IKK Phosphorylation/␤-TrCP Binding Domain Is Required to Alleviate p50-induced Inhibition of Proc-essing of p105 in Vivo-We (16) and later Heissmeyer and colleagues (17) have shown recently that IKK-mediated phosphorylation of specific Ser residues within the C-terminal domain of p105 leads to recruitment of the SCF ␤-TrCP ubiquitin ligase with subsequent polyubiquitination and rapid processing/degradation of the precursor molecule. Heissmeyer and colleagues (15) have shown that IKK-mediated degradation of p105 releases docked p50 that interacts with Bcl-3 to generate the trimeric p50⅐p50⅐Bcl-3 active transcription factor. We hypothesized that the mechanism that underlies liberation of the docked inhibitory subunits involves IKK and ␤-TrCP-mediated targeting of p105. As can be seen in the experiment depicted in Fig. 6A, processing of p105-WT in cells is accelerated significantly in the presence of active IKK. Furthermore, as can also FIG. 7. ␤-TrCP is required along with IKK to alleviate p50induced inhibition of processing/degradation of p105. COS-7 cells were transfected with cDNA coding for p105-WT (lanes 1-8) along with cDNAs coding for p50-Xba (lanes 3-8), constitutively active IKK␤-SSϾEE ( lanes 5-8), and human ⌬F-␤-TrCP (lanes 7 and 8). 48 h after transfection, cells were harvested, and generation and nuclear translocation of p50 were monitored using Western blot analysis as described in the legend to Fig. 6A.   FIG. 6. The C-terminal IKK phosphorylation and ␤-TrCP anchoring domain is required for in vivo stimulation-induced processing/ degradation of p105 and for alleviation of inhibition of processing by ankyrin repeat-anchored p50 subunits. Panel A, IKK stimulates processing of p105 and subsequent translocation of the resulting p50 to the nucleus. COS-7 cells were transfected with cDNA coding for p105-WT and when indicated, with cDNA coding for constitutively active IKK. 48 h after transfection, cells were harvested, and nuclear and cytosolic fractions were isolated as described under "Experimental Procedures." Protein aliquots derived from an equal number of cells were resolved via SDS-PAGE, blotted onto nitrocellulose paper, and proteins were visualized using anti-p50 antibody and ECL as described under "Experimental Procedures." C denotes cytosolic fraction, and N denotes nuclear fraction. Sites of migration of p105-WT and p50 are indicated. Panel B, IKK alleviates p50-induced inhibition of processing of p105-WT but not of p105-⌬918-934 and allows nuclear translocation of the generated p50. COS-7 cells were transfected with cDNAs coding for p105-WT (lanes 1-6) or p105-⌬918-934 (lanes 7-12) and when indicated, with cDNAs coding for constitutively active IKK (lanes 5, 6, 11, and 12) and p50-Xba (lanes 3-6 and 9 -12). 48 h after transfection, cells were harvested, and generation and nuclear translocation of p50 were monitored using Western blot analysis as described in the legend to panel A. Quantified data are also presented (percent of processing; p50/p50 ϩ p105). be seen in Fig. 6B, processing/degradation of most of p105 enables the generated p50 to migrate to the nucleus because it has lost its cytosolic anchor. To identify the region within p105 which mediates IKK activity, we transfected cells with either p105-WT or p105-⌬918 -934, which lacks the C-terminal phosphorylation domain. As can be seen in Fig. 6B, expression of p50 inhibits processing of both p105-WT and p105-⌬918 -934 (compare lane 1 with lane 3 and lane 7 with lane 9; data are also quantified). Expression of constitutively active IKK␤ leads to alleviation of inhibition and translocation of the generated p50 (as well as of the exogenously expressed p50-Xba) to the nucleus (compare lanes 3 and 4 with lanes 5 and 6). In contrast, the inhibitory effect of p50 on the processing of p105-⌬918 -934 (compare lane 9 with lane 7) cannot be alleviated by IKK (compare lane 11 with lane 9).
Last, it was important to examine the role of ␤-TrCP in the kinase-mediated alleviating effect. We used ⌬F-␤-TrCP, a dominant negative species of the ligase, which can bind the substrate but cannot recruit the other components of the ubiquitin conjugation machinery. Consequently, it cannot target the substrate for degradation (16). As can be clearly seen from the experiment depicted in Fig. 7, p50 inhibits processing (compare lane 1 with lane 3) of p105-WT. IKK alleviates the inhibition (compare lanes 5 and 6 with lanes 3 and 4). This alleviation was abolished, however, by concomitant expression of ⌬F-␤-TrCP (compare lanes 7 and 8 with lanes 5 and 6): the dominant negative ligase inhibited the stimulated IKK-mediated induced processing/degradation and generation of p50. DISCUSSION We have shown previously that p105 is targeted for processing/degradation by two distinct ubiquitin system recognition motifs. The first motif, which is adjacent to the GRR, contains two Lys residues that serve as ubiquitin anchors, and a downstream acidic domain that may serve as an E3 binding site. This motif is probably involved in basal/constitutive processing/degradation that occurs in nonstimulated, resting cells and provides the cell with the low amount of p50 required for its activity under these conditions (10). Processing of p105 under these conditions may occur cotranslationally (12,18), and this site may be involved in targeting the molecule via this unique process. The second, C-terminal, recognition motif undergoes signal-induced IKK-mediated phosphorylation (15) with subsequent recruitment of the SCF ␤-TrCP ubiquitin ligase, polyubiquitination, and accelerated processing/degradation (16). Involvement of the SCF ␤-TrCP ligase complex in the process has been corroborated later by Heissmeyer and colleagues (17). We have shown that the two motifs are targeted via two different E3s (16) and most probably, via distinct E2s as well (not shown). Several proteins have been described which are targeted via two distinct motifs and conjugation enzymes. Among them are p53, which is targeted by Mdm2 after DNA damage (27), and by the E6/E6⅐AP ligase complex in human papillomavirus-transformed cells (28). IB␣ is degraded within the context of a signaling complex following phosphorylation of Ser residues 32 and 36 (for review, see Ref. 5). The free inhibitor is probably degraded following CK II-mediated, constitutive phosphorylation of Ser-293 (29). The yeast mating type transcriptional regulator MAT␣2 is targeted by two motifs, Deg1 and Deg2, and two E2 enzymes, Ubc6 and Ubc7 (30); however, the identity of the E3(s) has remained obscure. Similarly, the model protein lysozyme is targeted by E2-14 kDa/E3␣ following recognition of the N-terminal residue and also by members of the UbcH5 family and a yet to be identified E3 that recognizes a downstream motif (31). Although the distinct motifs and conjugating enzymes that govern stability of p53 and IB␣ operate under different pathophysiological conditions, the physiological significance of the involvement of two sites in the degradation of MAT␣2 and processing/degradation of p105 has remained obscure.
In the current study we show that processing (Fig. 2) of a p105 precursor that contains more than four ankyrin repeats but lacks the C-terminal signaling domain is significantly less efficient compared with processing of p105-WT and p105 species that contain fewer than four repeats. We hypothesized that the inhibition is caused by docking of free NF-B subunits present in the HeLa cell extract to the ankyrin repeat domain. To test this hypothesis, we studied the effect of exogenously added p50 on p105 processing in vitro. In a reconstituted cellfree system, exogenous p50 inhibits processing (Fig. 3) of precursor molecules that contain four or more ankyrin repeats (panel A) but lacks the IKK C-terminal phosphorylation and TrCP binding domain (panel B). Mechanistic analysis shows that the exogenous p50 does not affect conjugation (Fig. 4) but probably interferes with proteasomal processing of the conjugated proteins (Fig. 4). A similar interference had been reported for several GRR-containing proteins such as Epstein-Barr virus nuclear antigen 1 (32), p105 (9,10), and the polyglutamine-containing mutated ataxin 3 (33). In the case of p105, interference with proteasomal degradation may result from the inability of the core protein, to which a cluster of additional large molecules is bound, to unfold in a manner that will allow entry into the proteasomal catalytic chamber. The findings in the cell-free system have been corroborated in vivo as well. As can be seen in Fig. 5, processing of the ankyrin repeat-free molecule p105-Tth111 (panel A) or of a p105 molecule that contains only three repeats and in which the remaining repeats have been replaced with DHFR (panel B), is not affected by exogenous overexpressed p50. In contrast, processing of p105-WT is inhibited (panels A and B; see also Figs. 6 and 7). As noted above, the discrepancy between the lack of p50-mediated inhibition of processing of the WT protein in the cell-free system (Fig. 3) compared with p50-induced inhibition in the cell (Figs. 5-7) can be attributed to the presence of an active kinase in the HeLa extract (16,26 and references therein). Importantly, inhibition of processing of the WT protein in cells could be relieved by expression of a catalytically active IKK␤. Alleviation required an intact C-terminal targeting domain ( Fig. 6) along with the ␤-TrCP ubiquitin ligase (Fig. 7).
Based on these results, we propose a model trying to explain the requirement for two distinct recognition signals for processing/degradation of p105 under different physiological conditions. According to the model, a nascent p105 polypeptide chain can be initially processed cotranslationally. Processing may require the basal/constitutive recognition motif. The p50 that was generated under these conditions is docked to an emerging ankyrin repeat domain in a p105 molecule that has not been processed yet. Docking of several additional free NF-B subunits hinders recognition of these p105 molecules by the proteasome and may inhibit cotranslational processing. The completely synthesized p105 along with the docked subunits serve as an inactive storage for these subunits. After stimulation, the C-terminal domain is phosphorylated. This modification leads to recruitment of the SCF ␤-TrCP ubiquitin ligase, which results in polyubiquitination and subsequent processing/degradation of p105 with release of the docked molecules and an additional p50 subunit generated from the precursor. These subunits serve as a source for an active transcriptional regulator in stimulated cells. Several experimental lines of evidence lend support this model, including progressive inhibition of processing with increasing number of ankyrin repeats and alleviation of inhibition that requires an intact C-terminal domain, IKK, and ␤-TrCP E3. However, several other important problems remain unsolved. The mechanism(s) that underlies selection of p105 molecules that undergo cotranslational processing from those that are synthesized to completion is not known. It is possible that cotranslational processing is the default mechanism that provides the resting cell with the small amount of p50 required for its basal needs. This occurs after stress/stimulation when all of the endogenous p50 has been depleted. Once a small amount of p50 is synthesized, it binds to the emerging nascent p105 chain and inhibits further generation of p50 which will occur now only after stimulation and targeting of the C-terminal domain.
An important observation relates to the distribution of p105 and p50 in the cytosol and nucleus. p105-WT appears to be sequestered solely in the cytosol (see. e.g. Fig. 5A, lanes 5 and  7). In contrast, p105 that lacks the last four ankyrin repeats is translocated to the nucleus (Fig. 5B, lanes 2 and 4). It is possible that the docked p50 subunits do not allow migration to the nucleus, possibly by sterically hindering the NLS. Thus, docking of active NF-B subunits to p105 not only sequesters them in an inactive storage form, but also ensures that they will not be translocated inadvertently to the nucleus.
Although clearly not the main focus of this study, it was suggested by Heissmeyer and colleagues (15,17) that phosphorylation of the C-terminal domain and recruitment of the SCF ␤-TrCP E3 complex leads to complete degradation of p105 and not to increased processing. The researchers argue, although they do not provide direct experimental evidence, that the increased amount of p50 observed under these conditions is attributed to increased transcription of p105 induced by IKK. Our experiments indicate that whereas stimulation leads also to complete degradation of a certain proportion of p105, accelerated processing accompanied by a significant increase in generation of p50 from p105 also occurs (Ref. 16 and Figs. 6 and 7). (i) In pulse-chase experiments, we could not detect increased synthesis of p105 (or of p50) in cells that express also a constitutively active form of IKK␤ (16). Whereas it can be argued that even after a short pulse in the presence of IKK, some of the p105-WT is rapidly degraded, this cannot be the case for p105-⌬918 -934, which cannot be processed/degraded following IKK stimulation. Yet, we could not observe increased synthesis of the mutant molecule in the presence of IKK (16; see also Fig. 6B; compare lane 9 with lane 11). Even if IKK-mediated stimulation of transcription of p105 does occur, several lines of evidence strongly support the notion that IKK/TrCP also stimulates processing. (ii) In a cell-free reconstituted system, processing and not degradation of labeled p105-WT is strongly inhibited by a phosphopeptide that spans the IKK phosphorylation domain and inhibits TrCP (16). Immunodepletion of SCF ␤-TrCP has a similar inhibitory effect (16). (iii) In a recent study, Xiao and colleagues (19) have shown that p100 is processed in a NIK-mediated manner (although NIK is not necessarily the p100 kinase), following phosphorylation of Ser residues 867 and 870, which are similar to p105 Ser residues 824 and 827 targeted by IKK (17). Although neither the sites nor the kinase in the two molecules is identical, it is likely that they are processed in a similar manner. (iv) Last, it is hard to conceive that the degradation machinery that acts under stimulation is completely different from the one that acts under basal conditions and can overcome completely the GRR barrier/ processing stop signal (9, 10) and the upstream residues that confer tight folding upon the p50 domain (11). The GRR physically interferes with proteasomal degradation regardless of the molecule in which it is inserted. For example, it can completely protect IB␣ from the rapid and efficient tumor necrosis factor-␣-induced degradation (34). As noted, it will be also difficult to overcome the tight folding of p50 which is dependent on structural characteristics distinct from the GRR (11). It appears that the property of limited degradation/processing is intrinsic to p100/p105 and not to the enzymatic machinery involved. It is possible, however, that after stimulation and utilization of the C-terminal phosphorylation site, the GRR/p50 folding does not function efficiently, and complete proteolysis of p105 can "overcome" partially these barriers.