Down-regulation of p27 Kip1 by Two Mechanisms, Ubiquitin-mediated Degradation and Proteolytic Processing*

The intracellular level of p27 Kip1 , a cyclin-dependent kinase (CDK) inhibitory protein, is rapidly reduced at the G1/S transition phase when the cell cycle pause ceases. In this study, we demonstrated that two posttranslational mechanisms were involved in p27 Kip1 breakdown: degradation via the ubiquitin (Ub)-proteasome pathway and proteolytic processing that rapidly eliminates the cyclin-binding domain. We confirmed that p27 Kip1 was ubiquitinated in vitroas well as in vivo. The p27 Kip1 -ubiquitination activity was higher at the G1/S boundary than during the G0/G1 phase, and p27 Kip1 ubiquitination was reduced significantly when the lysine residues at positions 134, 153, and 165 were replaced by arginine, suggesting that these lysine residues are the targets for Ub conjugation. In parallel with its Ub-dependent degradation, p27 Kip1 was processed rapidly at its N terminus, reducing its molecular mass from 27 to 22 kDa, by a ubiquitination-independent but adenosine triphosphate (ATP)-dependent mechanism with higher activity during the S than the G0/G1 phase. This 22-kDa intermediate had no cyclin-binding domain at its N terminus and virtually no CDK2 kinase inhibitory activity. These results suggest that p27 Kip1 is eliminated by two independent mechanisms, ubiquitin-mediated degradation and ubiquitin-independent processing, during progression from the G1 to S phase.

sity of the signals they integrate and the central importance of their roles in cell cycle control. These regulatory mechanisms include variations in cyclin levels, positive-and negative-acting phosphorylation of the kinase subunit, and the actions of CDK inhibitors (CKIs) (2). Of these, the CKIs appear to be the most diverse and flexible regulators. Mammalian CKIs are classified into two families: the Cip/Kip and Ink4 families. The former comprises p21 Cip1/Waf1 , p27 Kip1 , and p57 Kip2 , each of which has a conserved domain, called the CDK-binding/inhibitory domain, at its N terminus.
The CKI p27 Kip1 plays a pivotal role in the control of cell proliferation (3)(4)(5). Transition from the G 1 to S phase is promoted by G 1 cyclin/CDK complexes, such as cyclin D/CDK4 and 6 and cyclin E/CDK2, and p27 Kip1 inhibits the activities of these kinases directly by binding to them (6 -9). The elimination of p27 Kip1 during the late G 1 phase is required for G 1 cyclin/CDK complex activation and cell cycle progression from the G 1 to S phase in various cell lines (10 -13). Consistent with this idea is that forced expression of p27 Kip1 blocks cell cycle progression during the G 1 phase, whereas targeted p27 Kip1 mRNA antisense vectors increase the fraction of cells in the S phase. Moreover, p27 Kip1 down-regulation due to enhanced degradation in various malignant neoplasms, such as colorectal, breast, stomach, and non-small-cell lung cancers, was observed (14 -18). Finally, we and others demonstrated that targeted disruption of the mouse p27 Kip1 gene resulted in enhanced growth of mice, multiple organ hyperplasia, and a predisposition to tumors (19 -21). These lines of evidence support the idea that p27 Kip1 is a key molecule that negatively regulates cell cycle progression.
A major question is: how are the intracellular levels of p27 Kip1 regulated in a precisely timed fashion? Previous studies showed that p27 Kip1 mRNA does not fluctuate during the cell cycle, implying the existence of posttranslational machinery that controls the p27 Kip1 expression levels (3,22). Genetic studies on yeast revealed that Sic1, a CKI controlling the G 1 /S transition, like mammalian p27 Kip1 , is degraded specifically by the ubiquitin (Ub)-proteasome system (23)(24)(25). Furthermore, the Ub-proteasome pathway was suggested to be involved in p27 Kip1 degradation in mammals (26). The Ub-proteasome pathway is emerging as a major and universal mechanism that regulates selective and time-controlled elimination of shortlived key regulatory proteins, e.g. cell cycle proteins (cyclins (27,28) and CKIs (29)) and transcriptional activators (IB (30,31), c-Jun (32), p53 (33), ␤-catenin (34), and others). This pathway requires adenosine triphosphate (ATP) and the covalent conjugation of target proteins with multiple Ub molecules (35)(36)(37). This multistep process involves Ub activation by a Ubactivating enzyme (E1), followed by transfer of Ub to a Ub-conjugating enzyme (E2), and the third step is the transfer of Ub to a Ub ligase (E3), which catalyzes the formation of isopeptide bonds between the C-terminal glycine of Ub and the ⑀-amino groups of lysine residues on the target proteins. During subsequent cycles, additional Ub molecules are added to the substrate. Then, multi-ubiquitinated proteins are recognized by the 26 S (1500 kDa) proteasome complex and rapidly degraded into short peptides. 26 S proteasomes are multicatalytic protease complexes containing chymotrypsin-like, trypsin-like, and postglutamyl activities together with ATP.
Therefore, it is important to elucidate the mechanisms responsible for p27 Kip1 breakdown, not only to improve our understanding of cell-growth control, but also for the discovery of new anti-cancer drugs. In this study, we found that p27 Kip1 is down-regulated at the G 1 /S transition point by two pathways: Ub-mediated degradation and a novel ubiquitination-independent processing pathway that abrogates p27 Kip1 function by eliminating its cyclin-binding domain.
Cell Culture and Synchronization of Cells-NIH3T3 cells were cultured in Dulbecco's modified Eagle's medium containing 10% calf serum, and FM3A cells were cultured in RPMI medium containing 10% fetal calf serum (Life Technologies, Inc.). NIH3T3 cells were synchronized by subjecting them to contact inhibition during culture to confluence to arrest them at the G 0 /G 1 phase, then releasing them from contact inhibition, allowing them to progress to the S phase, by replating at a density of approximately 30%. For cell cycle analysis, bromodeoxyuridine-pulsed NIH3T3 cells were fixed with 70% ethanol, treated with 2 N HCl containing 0.5% Triton X-100, neutralized with borax buffer (pH 8.5), subjected to dual color staining with an anti-bromodeoxyuridine monoclonal antibody (mAb) conjugated with fluorescein isothiocyanate (Becton Dickinson Immunocytometry Systems, San Jose, CA) and 5 g/ml propidium iodide, and then analyzed using a FACS-Calibur flow cytometer and Cell Quest software (Becton Dickinson).
Plasmids-Mouse or human p27 Kip1 was subcloned into pGEX-6P (Amersham Pharmacia Biotech UK Ltd., Bucks, United Kingdom) in order to produce bacterially expressed protein. A deletion mutant of p27 Kip1 was constructed by subcloning the polymerase chain reaction fragment containing the C terminus of the p27 Kip1 sequence (38 -198 amino acids) into pGEX-6P. KR mutants were produced by replacing the lysine residues of p27 Kip1 with arginine by site-directed mutagenesis using a QuickChange TM site-directed mutagenesis kit (Stratagene), according to the manufacturer's protocol. A mammalian expression vector encoding p27 Kip1 was constructed by subcloning human p27 Kip1 into pcDNA3.1/Myc-His (Invitrogen, Carlsbad, CA), and the mammalian expression vector encoding HA-tagged ubiquitin was a gift from Dr. Dirk Bohmann (European Molecular Biology Laboratory).
Protein Expression and Purification-Glutathione S-transferase (GST)-tagged p27 Kip1 proteins for the in vitro ubiquitination assay were expressed in Escherichia coli XL1-blue and affinity-purified using glutathione-Sepharose CL-4B (Amersham Pharmacia Biotech), and the GST tag was cleaved using PreScission Protease (Amersham Pharmacia Biotech), according to the manufacturer's instructions. Myc-tagged p27 Kip1 and HA-tagged Ubs were transiently expressed. Each DNA was incubated with LipofectAMINE reagent (Life Technologies, Inc.) in serum-free medium (Opti-MEM; Life Technologies, Inc.) for 30 min at room temperature, after which the mixture was incubated with the required cells for 16 h, followed by incubation in complete medium for 32 h.
Preparation of Cell Extracts-For the in vitro ubiquitination assay, the required cells were washed with phosphate-buffered saline), suspended in double-distilled water, and frozen and thawed three times; the resulting lysate was subjected to centrifugation at 100,000 ϫ g for 4 h at 4°C, and the supernatant (S100PrϪ) was retrieved and frozen at Ϫ80°C. For the in vivo ubiquitination assay and analysis of endogenous p27 Kip1 levels, the required cells were incubated in lysis buffer containing 0.1% Nonidet P-40 on ice for 15 min, cleared by centrifugation at 15,000 rpm for 15 min at 4°C, and the protein concentration of the supernatant was determined by the Bradford method (Protein Assay; Bio-Rad). For the in vitro degradation assay, NIH3T3 cells were homogenized in phosphate-buffered saline followed by centrifugation at 100,000 ϫ g for 1 h at 4°C and the resulting supernatant (S100) was retrieved and frozen at Ϫ80°C.
In Vitro Ubiquitination Assay-Mouse recombinant p27 Kip1 was incubated with the FM3A or NIH3T3 cell extract in the presence of an ATP-regenerating system (50 mM Tris (pH 8.3), 5 mM MgCl 2 , 5 mM ATP, 10 mM creatine phosphate, 0.2 unit/ml creatine kinase) together with 1 mg/ml Ub, 100 g/ml Ub-CHO, 2 mM dithiothreitol, a protease inhibitor mixture (10 g/ml each of antipain, pepstatin, chymostatin, leupeptin, and PMSF) and a proteasome inhibitor mixture (250 M ALLN, 250 M ZLLLal, and 25 M clasto-lactacystin ␤-lactone). The reactions were carried out at 37°C for 1 h and terminated by adding SDS sample buffer, and each reaction mixture was subjected to SDS-PAGE on a 10% gel, followed by immunoblotting analysis with the anti-p27 mAb TDL. In order to confirm that p27 Kip1 had been ubiquitinated, identical reaction mixtures to those described above were subjected to in vitro ubiquitination in the presence of biotinylated Ub (produced using EZ-Link Sulfo-NHS-LC-Biotin (Pierce), according to the manufacturer's protocol) instead of Ub. Each reaction mixture was incubated with the TDL mAb, and the resulting immunoprecipitate was analyzed by immunoblotting with Neutravidin-HRP conjugate (Pierce) as the probe.
In Vivo Ubiquitination Assay-NIH3T3 cells were incubated with or without 100 M ALLN for 12 h in a CO 2 incubator, lysed, and then subjected to Western blotting with the TDL mAb. COS7 cells were transiently transfected with myc-tagged p27 and HA-tagged Ub and 36 h later, 50 M lactacystin was added to the culture, which was incubated for another 12 h in a CO 2 incubator. Then, the cells were lysed, as described above, and subjected to immunoblotting with the anti-c-myc antibody 9E10.
In Vitro Degradation Assay-Mouse recombinant p27 Kip1 was incubated with the NIH3T3 cell extract (S100) at 37°C for 30 or 60 min and subjected to immunoblotting with the required anti-p27 Kip1 antibody (N-20, C-19, or TDL).
In Vivo Degradation Assay-NIH3T3 cells were metabolically labeled with 100 Ci/ml Tran 35 S-label (ICN Pharmaceuticals Inc., Costa Mesa, CA) in methionine-free medium for 3 h, and then chased in complete medium for 0 or 3 h in a CO 2 incubator Cell lysates were immunoprecipitated with the TDL mAb, followed by Protein G-Sepharose (Amersham Pharmacia Biotech) affinity purification and SDS-PAGE, as described above, autoradiography, and quantification using a BAS-2000 imaging analyzer (Fuji Film, Kanagawa, Japan).

Ubiquitination of p27 Kip1 in Vivo and in Vitro-
The results of the previous study suggested that p27 Kip1 is down-regulated by the Ub-proteasome pathway (26). Originally, we developed an assay system for estimating p27 Kip1 ubiquitination activity in vitro. Careful manipulation was required to detect the unstable intermediates; the key factors for detection of ubiquitinated p27 Kip1 were inhibition of isopeptidase activity, a supply of excess Ub, and the exclusion of proteasomes. Isopeptidase, which associates with 26 S proteasomes and hydrolyzes multi-Ub chains, made it difficult to detect ubiquitinated p27 Kip1 . In this study, the addition of Ub-CHO, an isopeptidase inhibitor, to the reaction mixture dramatically improved the amount of ubiquitinated p27 Kip1 detected in vitro (compare lanes 4 and 8 in Fig. 1A). The addition of exogenous Ub to the reaction mixture also increased the formation of Ub-p27 Kip1 conjugates (compare lanes 6 and 8 in Fig. 1A), suggesting that Ub was a limiting factor in this reaction. When GST-Ub (34 kDa) was added to the reaction mixture instead of Ub (8 kDa), a species with a higher molecular mass appeared (compare lanes 8 and 10 in Fig. 1A). Thus, this result confirms that the bands with lower electrophoretic mobilities were ubiquitinated p27 Kip1 , not aggregates of p27 Kip1 or nonspecific products resulting from cross-reaction with the mAb. Proteasomes were removed from the cell cytoplasmic extract by differential centrifugation at 100,000 ϫ g for 4 h, and the proteasome activity of the resulting proteasome-depleted supernatant (S100PrϪ) was very low. The exclusion of proteasomes also dramatically increased Ub-p27 Kip1 conjugate formation (data not shown).
In order to confirm that the protein with the higher molecular mass we detected was ubiquitinated p27 Kip1 , identical reaction mixtures containing biotinylated-Ub instead of Ub were subjected to immunoprecipitation-immunoblotting analysis. The resulting products were immunoprecipitated with the TDL mAb, blotted, and probed with an avidin-HRP conjugate.
Ubiquitinated p27 Kip1 was also detected in vivo. ALLN, which inhibits proteasome and calpain activities, induced the accumulation of Ub-p27 Kip1 conjugates in NIH3T3 cells (Fig.  1C). Ubiquitinated p27 Kip1 was also observed in the transient p27 Kip1 cDNA transfection assay using COS7 cells (Fig. 1D), in which co-expression of Ub and p27 Kip1 led to moderately enhanced ubiquitination of p27 Kip1 (Fig. 1D, lane 2) and the selective proteasome inhibitor lactacystin further enhanced the ubiquitinated p27 Kip1 level (Fig. 1D, lane 3). These data indicate that p27 Kip1 is ubiquitinated in vivo as well as in vitro. Of note, mono-and di-ubiquitinated p27 Kip1 was also accumulated in the ALLN-or lactacystin-treated cells (Fig. 1, C and D, respectively), suggesting that the degradation of the monoubiquitinated p27 Kip1 might also be dependent on the proteasome. We also observed accumulation of a p27 Kip1 fragment with a lower molecular mass (approximately 22 kDa, referred to hereafter as p27⌬22k; Fig. 1, C and D). The proteasome and calpain inhibitor ALLN enhanced the accumulation of both p27 Kip1 and p27⌬22k (Fig. 1C), whereas the specific proteasome inhibitor lactacystin resulted in the decrease in p27⌬22k (Fig. 1D), suggesting that this smaller p27 fragment was produced by proteasomes and subsequently degraded by a calpainlike protease.

FIG. 1. Identification of ubiquitinated p27 Kip1 in vitro and in vivo.
A, exogenous p27 Kip1 protein (exo-p27) was incubated with or without the FM3A cell lysate from which the proteasomes had been removed by ultracentrifugation (S100PrϪ), as described under " Experimental procedures," with an ATP-regenerating system. Ub, GST-Ub, and Ub-CHO were added as indicated. The reaction mixtures were analyzed by immunoblotting with the anti-p27 Kip1 mAb TDL. B, exo-p27 was incubated (lanes 1 and 3) or not incubated (lane 2) in ubiquitination mixture with biotinylated Ub instead of Ub. The reaction mixtures were immunoprecipitated with the TDL mAb (lanes 1 and 2), or the mock mAb (anti-mouse IgG; lane 3) and then immunoblotted and probed with avidin-HRP. The bands marked n.s. represent nonspecific bands, which were also observed in the control preparations (not immunoprecipitated or immunoprecipitated with anti-mouse IgG). C, NIH3T3 cells synchronized at the G 1 /S phase were cultured with (lane 2) or without (lane 1) 100 M ALLN and the cell lysates were analyzed by immunoblotting with the TDL mAb. The levels of ubiquitinated p27 Kip1 and a truncated product (p27⌬22k) were augmented by ALLN treatment. D, COS7 cells were transiently transfected with an expression vector encoding myc-tagged p27 Kip1 with or without an expression vector encoding Ub, treated with or without 50 M lactacystin (as indicated by plus and minus signs above the lanes) and analyzed by immunoblotting with the anti-myc mAb 9E10. Lactacystin treatment increased the amount of ubiquitinated p27 Kip1 .
Cell Cycle-dependent Ubiquitination of p27 Kip1 -During the G 0 /G 1 phase, p27 Kip1 accumulates and then its level decreases as cells progress toward the S phase (10). We examined whether the p27 Kip1 ubiquitination activity is regulated in a cell cycle-dependent manner. First, the correlation between the cell cycle stages and expression levels of p27 Kip1 in NIH3T3 cells was examined. The cell cycle was synchronized at the G 0 /G 1 phase by contact inhibition, the cells were released and allowed to progress to the S phase by replating, and the subsequent progress of the cell cycle was monitored by flow cytometry ( Fig. 2A). Almost all the cells were arrested at the G 0 /G 1 phase by contact inhibition, and, about 12 h after release, the cells progressed from the G 1 to S phase almost synchronously. The number of cells in the S phase increased 12 h after release and decreased after 18 h, and the number in the M phase was maximal after about 21 h. The p27 Kip1 expression level was high in the G 0 /G 1 phase (from 0 to 9 h), declined rapidly at the G 1 /S transition point (between 9 and 12 h), and was minimal after about 18 h (Fig. 2B).
Cell lysates were prepared from cells harvested 0, 10, and 20 h after release, roughly corresponding to the G 0 /G 1 , G 1 /S, and G 2 /M phases, respectively. The p27 Kip1 ubiquitination activities of these lysates were determined by the in vitro ubiquitination assay, as described above under "Experimental Procedures" and in the legend to Fig. 1. The ubiquitination activity was higher in cells at the G 1 /S boundary (10 h) than during any other phase (Fig. 2C, lane 2). Therefore, we concluded that p27 Kip1 ubiquitination activity is regulated in a cell cycle-dependent manner, increases at or near the G 1 /S boundary stage, and declines thereafter. Determination of Ubiquitination Site(s) on p27 Kip1 -In an attempt to determine which lysine residue(s) in p27 Kip1 is/are the target(s) for Ub conjugation, the 13 lysine residues in human p27 Kip1 were substituted by arginine in clusters, as shown in Fig. 3A: KR1 (K25R, K47R, K59R), KR2 (K68R, K73R, K81R), KR3 (K96R, K100R), KR5 (K134R, K153R, K165R), and KR6 (K189R, K190R). These mutants were subjected to the in vitro ubiquitination reaction followed by immunoblotting analysis and probing with the TDL mAb, which recognizes the amino acid (aa) stretch around position 60. The ubiquitination level of KR5 was significantly lower than those of wild-type (wt) p27 Kip1 and KR2, -3, and -6 ( Fig. 3B, lanes  1-12). As KR1, which contained mutations at position 59, did not react with this mAb (Fig. 3B, lanes 3 and 4), another anti-p27 Kip1 mAb, C-19, was used to probe this mutant (Fig.  3B, lanes 13-16). The ubiquitination level of KR1 was comparable to that of wt p27 Kip1 (Fig. 3B, lanes 14 and 16). A timecourse experiment confirmed that the ubiquitination of KR5 was impaired markedly in comparison with that of wt p27 Kip1 (Fig. 3C). As cleavage of the GST tag was incomplete as a result of the purification of recombinant p27 Kip1 , GST-p27 Kip1 remained. The possibility that Ub was conjugated to the GST portion of the GST-p27 Kip1 fusion protein was excluded by the reduction in ubiquitination of KR5 mutant. These results indicate that Ub conjugation of p27 Kip1 targets some of or all the lysine residues at positions 134, 153, and 165.
Proteolytic Processing of p27 Kip1 -We also observed that bands with higher electrophoretic mobilities (p27⌬22k in Fig.  1, C and D) appeared in parallel with p27 Kip1 ubiquitination in vivo and in vitro, suggesting the activity responsible for processing p27 Kip1 in the cell lysates that contributes to the downregulation of p27 Kip1 in parallel with that responsible for Ubdependent degradation. Therefore, we investigated this processing activity. The small fragment shown in Fig. 1D seemed to be a truncated product lacking the N-terminal portion, because the myc epitope attached to the C terminus of p27 Kip1 was retained. We investigated the processing mechanism by incubating recombinant p27 Kip1 protein with the NIH3T3 cell extract and analyzing the reaction product (p27⌬22k in Fig. 4), which was readily detected, suggesting this processing reaction was probably rapid and further degradation of the p27⌬22k fragment was rate-limiting. The molecular masses of the product of exogenous p27 Kip1 and the derivative of endogenous p27 Kip1 in NIH3T3 cells were both approximately 22 kDa, indicating that the reaction in vitro faithfully reproduced p27 Kip1 processing in vivo (Fig. 4A, lanes 1 and 3). The amount of intact p27 Kip1 decreased as the amount of the processed product increased. The sum of p27 Kip1 expression level was also reduced, indicating that the further degradation process from the 22-kDa intermediate occurred. This processing reaction was ATP-dependent, because the addition of an ATP-regenerating system to the reaction mixture promoted it, resulting in a reduction in the amount of wt p27 Kip1 and an increase in p27⌬22k protein production (Fig. 4B, lane 3). In contrast, ATP-␥S, which suppresses ATP regeneration, inhibited the processing activity, as the amount of wt p27 Kip1 protein FIG. 2. Cell cycle-dependent expression and ubiquitination of p27 Kip1 . A, NIH3T3 cells were synchronized at the G 0 /G 1 phase, released so they progressed to the S phase, as described under "Experimental Procedures," and the cell cycle was analyzed by flow cytometry at the indicated times after release. The proportions of cells in the G 0 /G 1 (squares), S (solid circles), and G 2 /M (triangles) phases (as percentages of the total number of cells) are shown. B, the p27 Kip1 expression level at each time point was analyzed by immunoblotting with the TDL mAb. The ␣-tubulin expression level was also shown as an internal control. C, in vitro ubiquitination assay of p27 Kip1 in lysates prepared from NIH3T3 cells cultured for 0, 10, and 20 h after release. did not decreased relative to the control level (Fig. 4B, lanes 2  and 4). Furthermore, the proteasome-specific inhibitor clastolactacystin ␤-lactone significantly suppressed the breakdown of p27 Kip1 (Fig. 4B, lane 5), leading to accumulation of the 27-kDa protein, whereas the proteasome inhibitor ZLLLal, which also inhibits calpain, resulted in the accumulation of both 27-and 22-kDa proteins (Fig. 4B, lane 6). The effects of clasto-lactacystin ␤-lactone and ZLLLal in vitro corresponded with the effects of lactacystin and ALLN in vivo, respectively (Fig. 1, D and C). These results suggest that p27 Kip1 is processed to the p27⌬22k fragment by 26 S proteasomes and then degraded to smaller peptides by a calpain-like protease.
We also tested caspase inhibitors, because p27 Kip1 and p21 Cip1 have been reported to be cleaved by caspase-3 (40) and p21 Cip1 was found to be cleaved by caspase-1 (41) during the progression of apoptosis. However, neither the caspase-1 inhibitor YVAD nor the caspase-3 inhibitor DEVD affected the proteolytic processing of p27 Kip1 (Fig. 4D).
Collectively, our results suggest that p27 Kip1 was degraded by at least two pathways: Ub-proteasome-mediated degradation and proteolytic processing, by which p27 Kip1 was first processed by proteasomes in an ATP-dependent manner to produce an approximately 22-kDa fragment (p27⌬22k) and then degraded to small peptides by a calpain-type protease.
These results indicate the existence of a novel proteolytic processing pathway, in addition to the Ub-proteasome pathway, that regulates intracellular p27 Kip1 expression levels.
We carried out further experiments to determine whether the rapid processing of p27 Kip1 to p27⌬22k affects to the function of p27 Kip1 . First, in an attempt to locate the processing site, three antibodies that recognize different positions of p27 Kip1 were used. The epitope recognized by the anti-p27 Kip1 mAb TDL may be located near the aa at position 60, as discussed above, and the anti-p27 Kip1 Abs N-20 and C-19 were raised against the N (aa 2-21) and C (aa 181-198) termini (Fig. 5B) of p27 Kip1 . The p27⌬22k fragment was detected by C-19 and TDL, but not by N-20 (Fig. 5A), which suggests that the processing site was located on the N-terminal side some distance from position 60. In view of its deduced molecular mass, the processing site seemed to lie near positions 35-40. Therefore, this processing reaction eliminates the cyclin-binding domain.
Cell Cycle-dependent, Ubiquitination-independent Processing of p27 Kip1 -We also examined whether the activity that degraded endogenous p27 Kip1 in NIH3T3 cells in vivo was cell cycle-dependent. As expected, the p27 Kip1 -degradative activity was higher in the S than the G 0 /G 1 phase (Fig. 6, A and B). In the S phase, the amount of p27⌬22k produced increased markedly in parallel with the amount of p27 Kip1 degraded. This result suggests the processing activity probably contributes to the down-regulation of p27 Kip1 as the cell cycle progresses from the G 1 to the S phase. In vitro experiments suggested that the degradative processing was executed by 26 S proteasomes, as this rapid reaction was ATP-dependent and inhibited by proteasome inhibitors.
In order to determine whether this degradative processing reaction was mediated through ubiquitination, wt p27 and its ubiquitination-deficient mutant KR5 were transiently expressed in NIH3T3 cells, resulting in comparable levels of p27⌬22k in both groups, suggesting that the reaction that produced p27⌬22k was ubiquitination-independent. Collectively, our data suggest the existence of a novel mechanism that negatively regulates intracellular p27 Kip1 levels, i.e. a Ub-independent proteolytic processing pathway, in addition to the Ub-mediated degradation pathway. DISCUSSION The p27 Kip1 expression level alters dramatically during the cell cycle, particularly from the G 1 to the S phase, suggesting FIG. 3. In vitro ubiquitination of p27 Kip1 KR mutants. A, lysine residues in p27 Kip1 were substituted by arginine in clusters, as indicated. B, in order to locate the ubiquitination site(s) in p27 Kip1 , recombinant KR mutants (KR1, -2, -3, -5, and -6) and wt p27 Kip1 proteins were subjected to the in vitro ubiquitination assay, as described in the legend to Fig. 1A. The TDL mAb was used for analysis of KR2, -3, -5, and -6, the epitopes of which are located near aa 60; and the anti-p27 Kip1 antibody C-19, which recognizes the C terminus of p27 Kip1 , was used to analyze KR1. C, time-course analysis of ubiquitination with wt p27 Kip1 and the KR5 mutant in vitro . that rapid elimination of p27 Kip1 is probably a prerequisite for the activation of cyclin/CDK kinase complexes and cell cycle progression. The mechanism responsible for the regulation of p27 Kip1 expression remains elusive. By analogy with CKIs in Saccharomyces cerevisiae (Sic1) and Schizosaccharomyces pombe (Rum1), protein degradation via the Ub-proteasome pathway has been speculated to be critical for controlling the expression levels of mammalian CKIs (23)(24)(25). Pagano et al. showed that p27 Kip1 was ubiquitinated and that its ubiquitination was augmented by a Ub-conjugating enzyme (E2), Ubc3, the human homologue of Cdc34 in S. cerevisiae (26). Thus, the regulatory system that controls CKI expression seems to be evolutionally well conserved. We demonstrated ubiquitination of p27 Kip1 using the system we originally designed to assay ubiquitination in vitro. Inhibition of isopeptidase activity (by Ub-CHO), an excess supply of Ub (exogenous Ub or GST-Ub), and elimination of proteasomes (S100PrϪ) were critical for the detection of unstable ubiquitinated-p27 Kip1 . After these experimental modifications, multimeric Ub chains conjugated with p27 Kip1 were visualized clearly and immunoprecipitation/immunoblotting analysis of the reaction mixtures verified that the high molecular weight forms were ubiquitinated p27 Kip1 . We used this assay system to locate the ubiquitination sites and demonstrated that the ubiquitination activity was controlled in a cell cycle-dependent manner. In many biological systems, phosphorylation is a signal for ubiquitination. The phosphorylation of Sic1 CKI in S. cerevisiae is necessary for binding to Cdc4, an F-box protein that recruits the substrate (Sic1) to the Skp1/Cdc53/Cdc34 complex, leading to subsequent ubiquitination. Similarly, phosphorylation of Rum1 CKI in S. pombe was shown to be essential for CKI binding to Pop1 ϩ , the homologue of Cdc4 in S. cerevisiae (25). Furthermore, p27 Kip1 was reported to be degraded in a phosphorylation-dependent manner (42,43). Taken together with our findings, these lines of evidence led us to hypothesize that phosphorylation precedes p27 Kip1 ubiquitination, which is mediated by an Skp1/Cul1/Fbox protein complex. This hypothesis remains to be tested.
In this study, we demonstrated the presence of an alternative p27 Kip1 degradation pathway operating at the G 1 /S transition point of the cell cycle. The Ub-independent p27 Kip1 processing activity produced the 22-kDa fragment of C-terminal p27 Kip1 rapidly, and was ATP-dependent and sensitive to proteasome-specific and chymotrypsin-specific inhibitors, suggesting that 26 S proteasomes conduct this processing reaction, because most of the chymotrypsin-like activity present in the cytosol fractions appears to be attributable to proteasomes (37). This is highly similar to the case of cyclin B1 (49 kDa) that is processed at the N terminus, producing a 42-kDa truncated  1, 4, and 7), 30 (lanes 2, 5, and 8), and 60 (lanes 3, 6, and 9) min and the degradation products were subjected to immunoblotting with the anti-p27 Kip1 mAbs N-20 (lanes 1-3), TDL (lanes 4 -6), and C-19 (lanes 7-9). B, diagram of the deduced cleavage sites in p27 Kip1 and the processed product p27⌬22k (22 kDa). The numerals represent the aa positions, and the portions recognized by N-20, TDL, and C-19 are denoted by bold lines. C, in vitro phosphorylation of histone H1 by cyclin E/CDK2 with [␥-32 P]ATP was performed without (lane 1) or with 0.1, 1, 10, 100, or 1000 nM p27⌬22k (lanes 2-6), or wt p27 Kip1 (lanes 7-11); the reaction mixtures were separated by SDS-PAGE; and the products were detected and quantified using a BAS2000 imaging analyzer. form, by 26 S proteasome in an ATP-dependent and Ub-independent manner (44). Similarly, the rate-limiting enzyme ornithine decarboxylase is down-regulated by 26 S proteasomes in a Ub-independent manner (45), whereas the transcription factor NF-B underwent Ub-dependent proteolytic processing by 26 S proteasomes during cellular maturation (46). The C terminus of the p105 precursor of NF-B is destroyed by ATPdependent processing, leaving its N-terminal p50 fragment active. We located the site of p27 Kip1 processing using three mAbs, the recognition sites of which were known, at approximately aa positions 35-40. This processing reaction eliminated almost the entire cyclin-binding domain from p27 Kip1 and reduced the cyclin/CDK-inhibitory activity of p27 Kip1 . Furthermore, the processing activity increased during the S phase. Uren et al. (47) reported that the intracellular levels of a variant of p27 Kip1 , a C-terminal 22-kDa form, increased in parallel with the multiplication of the DNA content. Therefore, it is highly likely that rapid processing of the intact p27 Kip1 molecule to an inert 22-kDa fragment promotes the progression of the cell cycle from the G 1 to the S phase. It remains to be tested whether this processing allows p27 Kip1 to be ubiquitinated for further destruction by Ub-dependent activity of the 26 S proteasome.
The transcriptional and translational efficiencies of p27 Kip1 expression, as well as post-translational proteolysis, may regulate the intracellular expression levels of p27 Kip1 . In cells near the G 1 /S border stage, the Ub-proteasome and Ub-independent processing pathways may cooperatively promote p27 Kip1 degradation and reduce its level below the threshold necessary to restrain the cell cycle from progressing from the G 1 to S phase. Alternatively, these proteolytic mechanisms may play individual roles in specific pathways. As several studies have indicated that deregulation of p27 Kip1 expression in a variety of malignant neoplasms is due to disturbed proteolytic activity, it is important to elucidate the proteolytic pathways that control p27 Kip1 expression. Better understanding of such pathways may lead to the discovery of anti-cancer drugs with novel modes of action.
FIG. 6. Cell cycle-dependent processing of endogenous p27 Kip1 in vivo. A, NIH3T3 cells were synchronized at the G 0 /G 1 or S phase then metabolically labeled with Tran 35 S-label and chased, as described under "Experimental Procedures." B, the levels of p27 Kip1 (solid bar) and p27⌬22k (shaded bar), as percentages of the intact p27 Kip1 level before degradation (0 h), in each phase are shown. C, NIH3T3 cells were transiently transfected with myc-tagged wt p27 Kip1 and the KR5 mutant in parallel with mock transfection, metabolically labeled with Tran 35 S-label, separated by SDS-PAGE, and analyzed using a BAS2000 imaging analyzer. The molecular masses of myc-p27 and myc-p27⌬22k were 28 and 23 kDa, respectively.