Substrate Recognition and Ubiquitination of SCFSkp2/Cks1 Ubiquitin-Protein Isopeptide Ligase*

p27, an important cell cycle regulator, blocks the G1/S transition in cells by binding and inhibiting Cdk2/cyclin A and Cdk2/cyclin E complexes (Cdk2/E). Ubiquitination and subsequent degradation play a critical role in regulating the levels of p27 during cell cycle progression. Here we provide evidence suggesting that both Cdk2/E and phosphorylation of Thr187 on p27 are essential for the recognition of p27 by the SCFSkp2/Cks1 complex, the ubiquitin-protein isopeptide ligase (E3). Cdk2/E provides a high affinity binding site, whereas the phosphorylated Thr187 provides a low affinity binding site for the Skp2/Cks1 complex. Furthermore, binding of phosphorylated p27/Cdk2/E to the E3 complex showed positive cooperativity. Consistently, p27 is also ubiquitinated in a similarly cooperative manner. In the absence of p27, Cdk2/E and Cks1 increase Skp2 phosphorylation. This phosphorylation enhances Skp2 auto-ubiquitination, whereas p27 inhibits both phosphorylation and auto-ubiquitination of Skp2.

p27 was first identified as an inhibitor for cyclin A, D, and E-dependent kinases (1,2). It has now become clear that it positively regulates Cdk4/cyclin D while negatively regulating Cdk2/A 2 and Cdk2/E (3). At the G 1 /S transition, phosphorylation of p27 on Thr 187 leads to its ubiquitination and subsequent degradation, resulting in the activation of Cdk2/E or Cdk2/A and commitment of the cell to the S phase (4). Overexpression of p27 in certain cancer cells prevents DNA replication and tumor formation in nude mice (5), whereas the loss of p27 has been reported in many human cancers and predicts a poor prognosis in breast, prostate, colon, gastric, lung, and esophageal cancers (6). Increased degradation of p27 is likely to be one of the mechanisms leading to abnormally low levels of the p27 protein in cancer cells. In several types of cancers, there is a strong correlation between the loss of p27 and induction of Skp2, a subunit of the SCF Skp2/Cks1 (Skp1, Cul1, Roc1, Skp2, Cks1 complex) ubiquitin E3 ligase that targets p27 for ubiquitination and degradation (7,8). Skp2 itself has consistently been shown to be an oncogene, and increased expression of Skp2 has been reported in many human cancers (9,10).
In the process of protein ubiquitination, the ubiquitin-activating enzyme E1 first forms a thiol ester bond with the C-terminal carboxyl group of Gly 76 on ubiquitin via an active site cysteine, activating ubiquitin for nucleophilic attack (11). The activated ubiquitin is then transferred to an ubiquitin-conjugating enzyme (E2), forming a thiol ester bond between the C-terminal carboxyl group and an active cysteine on E2. An ubiquitin E3 ligase binds to both the substrate and E2 and facilitates the transfer of the ubiquitin from E2 to the substrate. Ubiquitin is attached to the substrate through an isopeptide bond formed between Gly 76 of ubiquitin and the ⑀-amino group of a substrate lysine residue. Polyubiquitin chains are formed via isopeptide bonds between the carboxyl group of Gly 76 on one ubiquitin molecule and the ⑀-amino group of Lys 48 (or less frequently, Lys 63 ) on another ubiquitin molecule. The substrate specificity of ubiquitination is mainly provided by different E3 ligases. Substrate recognition by E3 is often the rate-limiting step for protein degradation in cells (8).
The SCF Skp2/Cks1 E3 ligase recognizes phosphorylated p27 in the p27/Cdk2/E complex and targets it for ubiquitination and degradation at the G 1 /S transition (12,13). SCF Skp2/Cks1 is composed of five subunits: Cks1, Skp2, Skp1, Cul1, and Roc1. Cul1 serves as a scaffold connecting the catalytic core with the substrate-binding site. It is an elongated protein with a stalk at its N terminus and a globular domain at its C terminus (14,15). The RING finger protein Roc1 binds the C-terminal globular domain of Cul1, forming a catalytic core that recruits the ubiquitin-conjugating enzyme E2. Skp1 binds to the N terminus of Cul1. Skp2 is connected to Cul1 by the interaction between an F box on Skp2 and an F-box recognition site on Skp1. The ten C-terminal leucine-rich repeats of Skp2 form a curved blade with a concave surface lined with ␤ strands. A 39-amino acid tail C-terminal to the leucine-rich repeats packs into the concave surface and serves as a platform for binding Cks1 (16). The C terminus of Skp2 and Cks1 form the substrate recognition core of the SCF Skp2/Cks1 E3 ligase. Cks1 is a small Cdk2-binding protein essential for SCF Skp2 -mediated p27 ubiquitination (17,18). Structural (16,19) and mutagenesis (20,21) studies indicate that Cks1 has a Cdk2-binding site, a Skp2-binding site, and a phosphate-binding site and that all three sites are required for p27 ubiquitination. Although the phosphate-binding site on Cks1 binds directly to phosphorylated Thr 187 on p27, the Skp2/ Cks1 interface provides another binding site for Glu 185 on p27. Skp2 greatly enhances the affinity between Cks1 and a C-terminal p27 phospho-peptide (21), indicating that interaction of Glu 185 with Skp2/Cks1 is also essential for substrate recognition by the SCF Skp2/Cks1 ubiquitin ligase.
In addition to phosphorylation on Thr 187 , p27 must be in complex with Cdk2/E to be efficiently ubiquitinated by SCF Skp2/Cks1 (22). It has been shown that Cdk2/E or Cdk2/A plays a noncatalytic role in the ubiquitination of p27 by SCF Skp2/Cks1 (23,24). Using an in vitro reconstituted p27 ubiquitination system with purified protein components, we found that a phospho-p27 peptide containing phosphorylated Thr 187 can block p27 ubiquitination; however, the affinity of the phospho-p27 peptide is 1000-fold less than that of the phosphorylated p27 in complex with Cdk2/E. We developed an HTR-FRET assay to quantitatively measure the interaction between phosphorylated p27/ Cdk2/E and the SCF Skp2/Cks1 complex. We found that phosphorylated p27/Cdk2/E binds to the Skp2/Cks1 complex in a cooperative manner with a K d of 44 nM. Both the phospho-p27 peptide and Cdk2/E can compete for Skp2/Cks1 binding to phosphorylated p27/Cdk2/E with different affinities. We also found that Cdk2/E phosphorylates Skp2 in a Cks1-dependent manner in vitro. This phosphorylation increases auto-ubiquitination of Skp2, which is inhibited by p27.
Preparation of Protein Components-His-E1 and His-UbcH3 were expressed in Escherichia coli and purified on Ni 2ϩ chelate resin. The SCF Skp2 complex was affinity purified via glutathione-agarose chromatography from Sf9 cells co-infected with recombinant baculoviruses expressing GST-Skp2, His-Skp1, His-Cul1, and Roc1. The baculoviruses for GST-Skp2, His-Skp1, his-Cul1, and Roc1 were kindly provided by Dr. Michele Pagano (New York University School of Medicine). p27/ Cdk2/E, FLAG-p27/Cdk2/E, FLAG-p27 T187A/Cdk2/E, and Cdk2/E complexes were purified via Ni 2ϩ chelate chromatography from Sf9 cells co-infected with baculoviruses expressing human p27 (with or without a N-terminal FLAG tag, YKD-DDDKG), Cdk2, and His-tagged cyclin E. In FLAG-p27 T187A, Thr 187 was mutated to alanine through site-directed mutagenesis. Human Cks1 was expressed in E. coli with an S tag (KETA-AAKFERQHMDS) at the N terminus and a Myc tag (EQKLI-SEEDL) at the C terminus and purified via S resin affinity chromatography (Novagen). Subsequently, the S tag was removed from Cks1 by cleavage with thrombin. All of the above proteins were dialyzed into Ub dialysis buffer (30 mM Tris-HCl, pH 7.5, 20% glycerol, 1 mM DTT) and stored in small aliquots at Ϫ80°C. Ubiquitin was labeled with biotin by incubating 100 mg of Ub (Sigma U6253) with 12.5 mg of EZ-link TM Sulfo-NHS-LC-Biotin in phosphate-buffered saline on ice for 2h. The average stoichiometry of labeling was determined by liquid chromatography/mass spectrometry to be 1-2 biotin group/ubiquitin molecule. The biotinylated ubiquitin was dialyzed three times against 4 liters of 10 mM Hepes, pH 8.0 for 2-4 h at 4°C and stored in small aliquots at Ϫ80°C.
Mass Spectrometry Analysis of Phosphorylated p27/Cdk2/E-Phosphorylated p27/Cdk2/E (phospho-p27) was prepared by incubating 0.1 mg/ml Cdk2/E (ϳ1.25 M) with 0.1 mg/ml (ϳ1 M) p27/Cdk2/E or FLAG-p27/Cdk2/E at room temperature for 2 h in kinase buffer (40 mM Tris-HCl, pH 7.5, 10 mM MgCl 2 , 1 mM DTT, 1 mM ATP). Aliquots were stored at Ϫ80°C. For liquid chromatography/mass spectrometry analysis, p27/ Cdk2/E or phosphorylated p27/Cdk2/E were precipitated by chloroform/methanol to remove detergent. Precipitated p27 was resuspended in 5% formic acid and 5% acetonitrile and injected onto a ZQ liquid chromatography/mass spectrometry system. To determine phosphorylation sites, p27 from different samples were subjected to SDS-PAGE. The gel was stained with SimplyBlue SafeStain (Invitrogen). p27 proteins were digested in-gel with trypsin as described (25). Tryptic digests were analyzed on an MS4 liquid chromatography system interfaced to an LTQ linear ion trap mass spectrometer (Thermo Electron) equipped with a nanotrap source (Michrom Bioresources). Tryptic peptides were first captured and desalted on a C18 trapping cartridge followed by separation on a MAGIC C18 0.1 ϫ 150 mm nanotrap column. The peptides were eluted with a 120-min gradient from 5 to 65% acetonitrile in 0.1% formic acid and 0.005% heptafluorobutyric acid. During acquisition, continuous data dependent MS 3 scanning was performed via the following steps: 1) survey MS scan of peptide ions, 2) MS/MS peptide fragmentation scan of the top five most intense ions, and 3) an MS 3 scan (MS/MS/MS) of peptides triggered by neutral loss of 49 or 33 m/z values from the peptides selected for fragmentation in step 2). Raw MS files were processed with MASCOT Distiller then submitted to an internal MASCOT server (version 2.1) for data base searching using the Human IPI data base (June 1, 2006). Cyclin-dependent kinase inhibitor 1B (IPI00006991) was the top protein identified. Relative phosphorylation ratios were calculated as percentages by determining the ion current intensities from extracted ion chromatograms for the ions corresponding to phosphorylated and nonphosphorylated peptides.
In Vitro Phosphorylation of Skp2-The SCF Skp2 complex (1 M) was incubated with Cdk2/E (1 M), Cks1 (1 M), and ATP (1 mM) in the kinase buffer described above at 30°C for 2 h. In some experiments, 20 M staurosporine was added to the reaction to inhibit the Cdk2/E activity.
In Vitro Ubiquitination Assay-Unless specifically indicated, p27 was ubiquitinated in a 15-l reaction by incubating phospho-p27 (40 nM) with E1 (40 nM), E2 (5 M), SCF Skp2 (25 nM), Cks1 (25 nM), and Bio-Ub (27.8 M) in Ub buffer (40 mM Tris-HCl, pH 7.5, 5 mM MgCl 2 ) containing 1 mM DTT and 0.5 mM ATP at room temperature for 1 h. For peptide competition experiments, p27 ubiquitination was performed in the presence of phosphorylated or nonphosphorylated p27 peptide at various concentrations. For analysis of the ubiquitination of p27 by SDS-PAGE, the assay mixture was mixed with an equal volume of 2ϫ SDS loading buffer containing 200 mM DTT. Following SDS-PAGE, the proteins in the gel were transferred to a polyvinylidene difluoride membrane and probed with anti-p27 or anti-Skp2 antibody. The signal on the membrane was detected and quantified by a densitometer or an Odyssey infrared imaging system from LI-COR Biosciences (Lincoln, NE).
Measurement of Interaction between FLAG-p27 and GST-Skp2-For interaction of FLAG-tagged p27 with Skp2 or Cks1, 20 nM of phosphorylated FLAG-p27/Cdk2/E, FLAG-p27/ Cdk2/E, FLAG-p27 T187A/Cdk2/E, Cdk2/E, or buffer control was incubated with 20 nM of SCF Skp2 complex and 20 nM of Cks1 in the presence of 30 nM of XL665 anti-FLAG and 1 nM of LANCE Eu anti-GST or LANCE Eu anti-Myc in a binding buffer containing 50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 0.05% Tween 20, and 1 mM DTT at room temperature for 2-4 h. The FRET signal was measured on an Analyst HT plate reader from Molecular Devices Corp (Sunnyvale, CA) as fluorescence of Cy5 emission at 665 nM/fluorescence of Eu emission at 620 nm ϫ 10,000 after excitation of europium at 340 nm. To determine the K d between phosphorylated FLAG-p27/Cdk2/E and the SCF Skp2/Cks1 complex, increasing amounts of phosphorylated FLAG-p27/Cdk2/E were incubated with 20 nM SCF Skp2 and 20 nM Cks1 in the presence of 1 nM LANCE Eu Anti-Skp2 and 100 nM XL665 anti-FLAG. For competition experiments, 20 nM of phosphorylated FLAG-p27/Cdk2/E was incubated at room temperature for 2-4 h with 20 nM of SCF Skp2 complex and 20 nM of Cks1 in the presence of increasing amounts of phosphorylated p27/Cdk2/E, phospho-or nonphospho-p27 peptide, or Cdk2/E and 1 nM LANCE Eu anti-Skp2 and 40 nM XL665 anti-FLAG. The interaction between FLAG-p27 and GST-Skp2 was measured as above.
Anti-FLAG Immunoprecipitation-Phosphorylated FLAG-p27/Cdk2/E and unphosphorylated FLAG-p27/Cdk2/E (50 ng/l, 100 l) were incubated either on ice or at 78°C for 20 min (heat inactivation). The heat inactivation samples were then centrifuged at 15,000 for 5 min at room temperature in a tabletop microcentrifuge. 90 l of the supernatant for each sample was incubated with 20 l of EZView anti-FLAG M2 affinity agarose for 2h at 4°C. The agarose beads were washed six times with phosphate-buffered saline containing 0.1% Tween 20 and subjected to SDS-PAGE and Western blotting with the indicated antibodies.

RESULTS
Reconstitution of the Ubiquitination of p27 in Vitro-An in vitro p27 ubiquitination assay was first reported by Pagano et al. (26) using cell extracts as a source of enzymes essential for p27 ubiquitination. Since then, all of the essential components for efficient in vitro p27 ubiquitination have been identified (17,18). All of these were recombinantly expressed and purified ( Fig. 1A and Table 1). To generate a robust in vitro p27 ubiquitination assay, p27 in complex with Cdk2/E (p27/Cdk2/E) purified from Sf9 cells was used as the substrate. We found that this complex was weakly ubiquitinated (Fig. 1B, lane 2) in vitro even in the presence of Cks1. However, when extra Cdk2/E FIGURE 1. A, a schematic representation of the p27 ubiquitination complex. B, Cdk2/E significantly increases the ubiquitination of p27/Cdk2/E. GST-Ub was used in the assay. C, Cks1 is essential for p27 ubiquitination in vitro. p27/Cdk2/E was ubiquitinated with or without Cdk2/E or Cks1 under standard conditions described under "Experimental Procedures." p27 was detected by Western blot with anti-p27 antibody (BD PharMingen).

Biochemical Studies of SCF Skp2/Cks1 E3 Ligase
was added to the reaction, p27 ubiquitination was greatly increased (Fig. 1B). This suggests that the p27/Cdk2/E purified from Sf9 cells was not fully phosphorylated and that additional Cdk2/E was required to phosphorylate the p27/Cdk2/E. As expected, the in vitro p27 ubiquitination is dependent on the presence of Cks1 (Fig. 1C). p27 phosphorylation was confirmed by Western blot with a phospho-Thr 187 -specific anti-p27 antibody (Fig. 2, lane 13).
Cdk2/E increased phosphorylation on Thr 187 of p27 in a dosedependent manner. Phosphorylated substrate was prepared for routine ubiquitination and binding assays by incubating p27/ Cdk2/E (100 ng/l) with additional Cdk2/E (100 ng/l) in the presence of ATP and Mg 2ϩ .
Mass spectrometry was used to further quantify p27 phosphorylation. Analysis of intact p27 protein as well as a peptide digestion of p27 in both the unphosphorylated and phosphorylated forms indicated that only a small portion of p27 in p27/ Cdk2/E purified from insect cells was singly phosphorylated, whereas the majority of p27 was unphosphorylated. In contrast, p27 in the phosphorylated p27/Cdk2/E complex was either singly or doubly phosphorylated with the single phosphorylation species being the dominant form (data not shown). Data from peptide mapping indicated that Thr 187 was ϳ90% phosphorylated in phosphorylated p27/Cdk2/E, whereas the corresponding peptides recovered from the digestion of unphosphorylated p27 were not phosphorylated. An additional phosphorylation site was identified at Ser 106 in the phosphorylated p27 digestion, although it does not appear to be a consensus site for Cdk2 (27). Therefore, the p27 in the complex of p27/Cdk2/E purified from Sf9 cells was unphosphorylated; upon in vitro Cdk2/E phosphorylation under our experimental conditions, it became almost completely phosphorylated on Thr 187 .
Phosphorylation on Thr 187 Is Essential but Not Sufficient for High Affinity Binding between p27/Cdk2/E and SCF Skp2/Cks1 -It has been shown that phosphorylation on Thr 187 of p27 is essential for p27 ubiquitination (22). Previous studies have also revealed the importance of Cdk2/E or Cdk2/A for p27 ubiquitination (22)(23)(24). However, none of these studies had directly compared the impact of Thr(P) 187 and Cdk2/E or Cdk2/A on the substrate recognition by the SCF Skp2/Cks1 E3 ligase in a quantitative manner.
An HTR-FRET assay using FLAG-tagged p27 and GSTtagged Skp2 to measure the interaction between SCF Skp2/Cks1 E3 ligase and phosphorylated p27/Cdk2/E was developed (Fig.   3A, top left panel). FLAG-tagged p27 or FLAG-tagged p27 T187A mutant were co-expressed and co-purified with Cdk2 and His-Cyc E via Ni 2ϩ -nitrilotriacetic acid resin and subsequently phosphorylated by Cdk2/E under the conditions described above. Phosphorylated FLAG-p27/Cdk2/E was incubated with the SCF Skp2/Cks1 E3 ligase in the presence of europium-labeled anti-GST and XL665-labeled anti-FLAG antibodies. Interaction of the substrate and the E3 generates a FRET signal between the Eu donor and XL665 acceptor after the excitation of Eu at 320 nm. A strong FRET signal was generated when phosphorylated FLAG-p27/Cdk2/E was incubated with both Cks1 and the SCF Skp2 complex (Fig. 3A, top left panel). In the absence of either Cks1 or the SCF Skp2 complex, no FRET was generated, indicating that the FRET signal specifically measures the interaction between p27 and Skp2. Unphosphorylated FLAG-p27/Cdk2/E or T187A mutant did not generate significant FRET signal when incubated with Cks1 and the SCF Skp2 complex, confirming that phosphorylation on Thr 187 of p27 is essential for the interaction of p27 with Skp2/Cks1 complex. The Myc tag on the C terminus of Cks1 enables the measurement of the interaction between p27 and Cks1 (Fig. 3A, top right panel) as well as the interaction between Skp2 and Cks1 (Fig. 3A, bottom panel) in the same reactions. Similar to the interaction of Skp2 and p27, the interaction of p27/Cks1 depends on the presence of the SCF Skp2 complex. On the other hand, p27/Cdk2/E or Cdk2/E has no impact on the Skp2/Cks1 interaction.
It has been shown that p27 is heat stable, but the activity of Cdk2/E is heat-sensitive (1,23). Phosphorylated FLAG-p27/ Cdk2/E was incubated at 75°C to heat-inactivate Cdk2/E. As shown in Fig. 3B, phosphorylated FLAG-p27/Cdk2/E interacted with Skp2 in the presence of Cks1 and produced a FRET signal; however, heat inactivation of phosphorylated FLAG-p27/Cdk2/E destroyed this interaction. The addition of fresh Cdk2/E to the heat-inactivated phosphorylated FLAG-p27 partially recovered the interaction. Although the levels of total p27, p27 phosphorylated at Thr 187 , Cyc E, and Cdk2 remained the same in samples following heat inactivation, Cdk2 and Cyc E no longer formed complex with FLAG-p27 after heat inactivation (Fig. 3, B and C). Thus, (i) phosphorylation on Thr 187 is essential but not sufficient for the substrate binding of the SCF Skp2/Cks1 E3 ligase and (ii) Cdk2/E in the p27/Cdk2/E complex provides the second binding site for the SCF Skp2/Cks1 complex.
Cooperative Binding of Phosphorylated p27/Cdk2/E and SCF Skp2/Cks1 -To determine the dissociation constant (K d ) between phosphorylated p27/Cdk2/E and the SCF Skp2/Cks1 complex, phosphorylated FLAG-p27/Cdk2/E was incubated with SCF Skp2 and Cks1 at different concentrations. The data fit well to a sigmoidal binding curve (Fig. 4) but not to a hyperbolic binding curve (data not shown). The sigmoidal binding curve is a signature for positive cooperativity between two binding sites (28). This indicates that binding of one p27/Cdk2/E to the E3 ligase enhances the binding of another p27/Cdk2/E to the E3 ligase. There was no previous evidence suggesting that each Skp2/Cks1 complex binds to two p27/Cdk2/E complexes. Moreover, the structure of the Skp1-Skp2-Cks1-p27 peptide complex indicates that, although there are multiple sites contributing to the interaction between each p27/Cdk2/E complex FIGURE 2. Phosphorylation of Thr 187 on p27 by Cdk2/E. p27/Cdk2/E (100 ng/l) was incubated with increasing amount of Cdk2/E at room temperature for 1h. Total p27 and p27 phosphorylated on Thr 187 were detected by Western blotting using anti-p27 antibody (BD PharMingen) and anti-phospho-Kip (Thr 187 ; Upstate Biotechnology, Inc.) that specifically recognize phosphorylated Thr 187 . MAY 25, 2007 • VOLUME 282 • NUMBER 21

JOURNAL OF BIOLOGICAL CHEMISTRY 15465
and each SCF Skp2/Cks1 complex, there is only one binding site for p27/Cdk2/E per Skp2/Cks1 complex. One explanation for the cooperativity is that the SCF Skp2/Cks1 complex forms a dimer and binding of phosphorylated p27/Cdk2/E to one E3 in the dimer increases the affinity of the other E3 for its substrate. The apparent K d for phosphorylated p27/Cdk2/E with the SCF Skp2/Cks1 complex is ϳ40 nM. Consistent with previous observations, unphosphorylated p27 and the T187A mutant in complex with Cdk2/E have much lower affinities to SCF Skp2/Cks1 .
To determine the apparent K m for substrate in the p27 ubiquitination assay (Fig. 4B), increasing amounts of phosphorylated p27/Cdk2/E were incubated with E1, E2, SCF Skp2 , Cks1, and Bio-Ub as described under "Experimental Procedures." Ubiquitination of p27 was detected as a ladder of p27 with increasing molecular weights in the gel via Western blot using an anti-p27 antibody. The ubiquitinated p27 was quantified with a densitometer and plotted against the total p27 concentration (Fig.  4B, right panel). The concentration dependence of total p27 ubiquitination also displayed a sigmoidal curve, suggesting that substrate binding is the rate-limiting step in this p27 ubiquitination assay. The apparent K m for phosphorylated p27/Cdk2/E was ϳ20 nM.
Cdk2/E Provides a Higher Affinity Binding Site than Thr(P) 187 -Because both Cdk2/E and phosphorylation of Thr 187 on p27 are essential for the substrate binding of the SCF Skp2/Cks1 complex, the binding affinity of each of the two sites was assessed using a competition assay. Increasing amounts of Cdk2/E, the C-terminal peptide of p27 with or without phosphorylation on Thr 187 , phosphorylated p27/Cdk2/E, or p27/Cdk2/E were used to compete with the binding of phosphorylated FLAG-p27/Cdk2/E to the SCF Skp2/Cks1 complex (Fig. 5A). Phosphorylated p27/ Cdk2/E was the most potent competitor with an IC 50 of 0.11 M ( Table 2). Cdk2/E and p27/Cdk2/E were less potent inhibitors than phosphorylated p27/Cdk2/E, with IC 50 values of 1.2 and 4.2 M, respectively. Although the unphosphorylated C-terminal peptide of p27 did not inhibit the binding between FLAG-p27 and Skp2, the phosphorylated peptide showed weak inhibition with an IC 50 of 80 M. These data indicate that Cdk2/E contributes the major binding strength, whereas the C-terminal phospho-p27 peptide alone has very weak binding to Skp2/ Cks1. However, Thr(P) 187 increases the binding affinity between p27/Cdk2/E and Skp2/Cks1 significantly.

Auto-ubiquitination of Skp2 Requires Cks1 and Cdk2/E but Is
Inhibited by p27-Skp2 is weakly ubiquitinated in the presence of E1, E2, and SCF Skp2 E3 (Fig. 6A, lane 5). The presence of Cks1 or Cdk2/E alone did not significantly enhance its ubiquitination (Fig. 6A, lanes 6 and 7); however, the presence of both Cks1 and Cdk2/E significantly increased Skp2 ubiquitination (Fig. 6A, lane 2, and data not shown). p27 has no effect on Skp2 ubiquitination in the absence of Cks1 (Fig. 6A, lane 4); however, it significantly decreases Skp2 ubiquitination in the presence of Cks1, whereas p27 is strongly ubiquitinated (Fig. 6B, lane 1). These data suggest that p27 competes better for the ubiquitination machinery than does Skp2.
Cdk2/E Phosphorylates Skp2 in a Cks1-dependent Manner in Vitro-GST-Skp2 in the SCF Skp2 complex purified from Sf9 cells was detected as a single band after SDS-PAGE using either anti-GST (Fig. 7A, lane  8) or anti-Skp2 (data not shown). When it was incubated with Cdk2/E, Cks1, ATP, and MgCl 2 , GST-Skp2 migrated as two bands (Fig. 7A, lane 2). One band migrated at approximately a similar rate to untreated GST-Skp2, whereas the other band migrated at a slower rate. This band shift of GST-Skp2 could be caused by Cdk2/E-mediated phosphorylation because it not only depends on the presence of ATP and MgCl 2 but also can be blocked by staurosporine, a kinase inhibitor (Fig. 7B). Mass spectrometry analysis of the GST-Skp2 bands confirms that the slowly migrating band contains one extra phosphate group at the N terminus of Skp2 (data not shown). Cks1 is required for the phosphorylation of Skp2 by Cdk2/E. This phosphorylation had a positive effect on Skp2 auto-ubiquitination because staurosporine, which blocked the Cks1-dependent phosphorylation of Skp2, also inhibited the Skp2 auto-ubiquitination (Fig. 8A, lanes 1 and 2). The presence of p27/Cdk2/E also inhibited both Skp2 phosphorylation and ubiquitination (Fig. 8A, lanes 5 and 6).

DISCUSSION
Cdk2/E Promotes p27 Binding to the SCF Skp2 E3 Ligase Complex-Recognition of phosphorylated p27 by SCF Skp2/Cks1 E3 ligase is one critical step in p27 ubiquitination and therefore has been the focus of many studies. Previous studies have shown that Cdk2/E or Cdk2/A plays a noncatalytic role in p27 ubiquitination in addition to phosphorylating p27 at Thr 187 . However, the relative role of Thr(P) 187 and Cdk2/E in the binding of phosphorylated p27/Cdk2/E to the SCF Skp2/Cks1 complex is not well understood.
In this study, we measured an apparent K d for the binding between phosphorylated p27/Cdk2/E and the SCF Skp2/Cks1 complex to be ϳ40 nM and an apparent K m for in vitro ubiquiti- A, determination of the dissociation constant for the phosphorylated p27/Cdk2/E and the SCF Skp2/Cks1 complex. 500 nM of phosphorylated FLAG-p27/Cdk2/E, FLAG-p27/Cdk2/E, or FLAG-p27 T187A/Cdk2/E (as described in the legend to Fig. 3) were serially diluted at a ratio of 1:1.2 in the kinase buffer and incubated with an equal volume of a binding mixture containing 2ϫ binding buffer, SCF Skp2 (40 nM), Cks1 (40 nM), Eu anti-GST (2 nM), and XL665 anti-FLAG 200 nM). After incubation at room temperature for 2 h, the interaction between FLAG-p27 and GST-Skp2 was measured as the FRET signal (10,000 ϫ Fluorescence @ 665 nm/Fluorescence @ 620 nm) generated between Eu anti-GST and XL665 anti-FLAG. Background was the FRET signal in the absence of p27/Cdk2E. The data were fit using the following formulas: y ϭ m1/((m2/x) m3 ϩ 1), where x ϭ concentration of phosphorylated FLAG-p27/Cdk2/E; y ϭ FRET signal subtracting background; m1 ϭ maximal binding signal detected; m2 ϭ apparent K d ; and m3 ϭ Hill slope. B, titration of phosphorylated p27/Cdk2/E in p27 ubiquitination assay. Phosphorylated p27/Cdk2/E was prepared as described under "Experimental Procedures" and serially diluted 1:2 in assay dilution buffer (30 mM Tris-HCl, pH 7.5, 19.5% glycerol, 1 mM DTT). 5 l of phosphorylated p27/Cdk2/E at each dilution were incubated with 10 l of an assay mixture containing E1, E2, SCF Skp2 , Cks1, and Bio-Ub at room temperature for 1h. The final concentration of each component was indicated under "Experimental Procedures." Ubiquitination of p27 was measured via Western blot using anti-27 antibody (BD PharMingen) and quantified on a densitometer. The data were fit using the same formula described above where m2 ϭ apparent K m . MAY 25, 2007 • VOLUME 282 • NUMBER 21 nation of phosphorylated p27/Cdk2/E to be ϳ20 nM. Using a quantitative HTR-FRET assay, we showed that phosphorylation of Thr(P) 187 is essential but not sufficient for this high affinity association and confirmed that Cdk2/E provides another binding site. We further evaluated the relative contribution of Cdk2/E and Thr(P) 187 to the substrate binding of the E3 complex in a competitive binding assay. Phosphorylated and unphosphorylated p27/Cdk2/E inhibit the binding between FLAG-p27 and Skp2/Cks1 with IC 50 values of 0.11 and 4.2 M, respectively, indicating that Thr(P) 187 significantly increases the binding affinity. Cdk2/E and the phospho-p27 peptide inhibit the binding of the phosphorylated p27/Cdk2/E to the SCF Skp2/Cks1 complex with IC 50 values of 1.2 and 80 M, respec-tively. Therefore, Cdk2/E provides a stronger binding site, whereas the phospho-p27 provides a weaker binding site for the SCF Skp2/Cks1 complex. Consistent with this, the phospho-p27 peptide inhibits the ubiquitination of phosphorylated p27/ Cdk2/E (40 nM) by the SCF Skp2/Cks1 E3 complex with an IC 50 of 40 M. The recent crystal structure of the Skp1-Skp2-Cks1- FIGURE 5. A, competition of phosphorylated FLAG-p27/Cdk2/E binding to the SCF Skp2/Cks1 complex with p27 C-terminal peptides and Cdk2/Cyc E. Phosphorylated FLAG-p27/Cdk2/E (40 nM) was incubated with SCF Skp2 (20 nM), Cks1 (20 nM), Eu anti-GST (1 nM), and XL anti-FLAG (40 nM) in the presence of increasing amount of p27 peptides, Cdk2/E, p27/Cdk2/E, or phosphorylated p27/Cdk2/E. The data were fit with the following IC 50 formulas: y ϭ 100/(1 ϩ (x/ml) m2 ), where x ϭ concentration of each competitor; y ϭ percentage of p27 bound to the SCF Skp2/Cks1 complex; m1 ϭ IC 50 ; m2 ϭ Hill slope. B, a phsopho-p27 peptide blocks ubiquitination of p27/Cdk2/E. Ubiquitination of phosphorylated p27/Cdk2/E was performed under standard conditions in the presence of increasing amount of phospho-p27 peptide. p27 was detected by Western blot (left) and ubiquitinated p27 was quantitated by a densitometer (right). The data were fit with the same IC 50 formulas described above.    phospho-p27 peptide complex clearly illustrates the binding interface on the Skp2/Cks1 complex for the phospho-p27 peptide (16,29). Only two residues near the C terminus of p27 are involved in the substrate targeting. Hao et al. (16) also determined the K d between the phospho-p27 peptide and Skp2/Cks1 to be 7 M using isothermal titration calorimetry.

Biochemical Studies of SCF Skp2/Cks1 E3 Ligase
Our data indicate that Cdk2/E provides a stronger binding site for phosphorylated p27 on Skp2/Cks1 than does the phosphorylated p27 peptide alone. Additionally, the binding affinity of unphosphorylated p27/Cdk2/E is much weaker than phosphorylated p27/Cdk2/E as shown in both the Skp2/Cks1 and p27/Cdk2/E binding assays and the competition experiment (Figs. 3A, 4A, and 5A and Table 2). It has also been shown that phosphorylation of Thr 187 at the C terminus of p27 is essential for p27 ubiquitination, and unphosphorylated p27/Cdk2/E is not efficiently ubiquitinated by the SCF Skp2/Cks1 E3 ligase (22). One possible explanation for these observations is that there are two distinct binding events that are required for efficient ubiquitination of p27 by the SCF Skp2/Cks1 ligase. The first site recognizes the p27/Cdk2/E complex at a point distal to the phosphorylation site on Thr 187 (Fig. 9A). The second site is phospho-Thr 187 , which yields the high affinity complex required for efficient ubiquitination by the catalytic machinery of the SCF Skp2/Cks1 ligase. Therefore, we propose a mechanism whereby there is an anchoring complex formed between the Cdk2/E portion of the p27/Cdk2/E complex followed by binding the phospho-Thr 187 residue for final positioning at the ubiquitination site (Fig. 9A). Via the above mechanism, unphosphorylated p27/Cdk2/E cannot be efficiently ubiquitinated, although it binds to the E3 ligase. Finally, we did observe the interaction between unphosphorylated p27/Cdk2/E and the SCF Skp2/Cks1 E3 ligase when binding studies were performed under low stringency conditions (data not shown).
How does Cdk2/E promote p27 binding to its E3 ligase? Our data, together with data from many other groups (16,20,21,23,24), provide sufficient evidence to indicate that Cdk2/E promotes p27 binding to its E3 via bridging the N terminus of p27 to Cks1 (Fig. 9A). This model is based upon the following evidence: (i) Cks1 binds to Skp2, Cdk2, and Thr(P) 187 on p27 via three distinct sites, and all three binding sites are essential for SCF Skp2/Cks1 -mediated p27 ubiquitination (16,20,21). (ii) An E63Q mutation of Cks1 destroys the Cks1 and Cdk2 interaction (19). This mutation has a minimal effect on the binding of Cks1/ Skp2 to the C-terminal phospho-p27 peptide (21) but greatly reduces the binding of Cks1/Skp2 to phosphorylated fulllength p27 (20), suggesting that the Cdk2/Cks1 interaction is critical for the binding of full-length p27 to E3 but not for the C-terminal phospho-p27 peptide. (iii) Interaction of the N terminus of p27 with Cdk2/E or Cdk2/A appears to be required for p27 ubiquitination (23,24). (iv) Although the Cdk2/E or Cdk2/A complexes are essential for p27 ubiquitination, neither Cyc E nor Cdk2 alone is sufficient to promote the ubiquitination of p27 (23). (v) Structures of the Cdk2-Cyc E complex, Cdk2-Cyc A complex, and Cdk2-Cks1 complex indicate that Cyc E/A and Cks1 bind to distinct sites on Cdk2 and that Cdk2 provides the link between Cks1 and Cyc E/A (23,30,31). (vi) Cdk2/E provides a stronger binding site for substrate recognition than the C-terminal binding motif of p27 containing Glu 185 and Thr(P) 187 . Therefore, Cks1 recruits p27 to its E3 ligase by securing both the C terminus and N terminus of p27. Cks1 binds Thr(P) 187 and Glu 185 through its anion-binding site and a Skp2/Cks1 interface. It also brings the N terminus of p27 to the E3 ligase through the interactions between Cks1 and Cdk2, Cdk2 and Cyc E, and Cyc E and the N terminus of p27.
Thus, our work, as well as that of others, has demonstrated that substrate recognition of the SCF Skp2/Cks1 E3 ligase is an organized process in which simultaneous and cooperative interaction among multiple subunits are required for the highest affinity between the substrate and the E3 ligase.
The SCF Skp2/Cks1 E3 Ligase Forms a Dimer-The binding kinetics for phosphorylated p27/Cdk2/E to the SCF Skp2/Cks1 complex were found to be sigmoidal. This is recapitulated in the p27 ubiquitination assay. The sigmoidal binding curves indicate that the interaction of one site facilitates the binding at another site. Because there is one substrate-binding site per SCF Skp2/Cks1 complex, our data suggest that the SCF Skp2/Cks1 complex forms a dimer and that binding of one substrate to the E3 enhances the binding of the second substrate to another E3 in the dimer. Based upon the crystal structures of SCF complexes (Skp1, Cul1, and F-box protein complexes), it was proposed that SCF facilitates the ubiquitination of its substrate through positioning of the substrate lysine(s) in proximity to the E2 active site and therefore increasing the effective concentration of the substrate at the E2 active site (32). However, the crystal structures have illustrated that there is a 32-50 Å gap between the substrate lysine and the active site on E2, which is a long distance for the E2 to span to transfer its charged ubiquitin to a substrate bound to Skp2/Cks1. If the SCF Skp2/Cks1 E3 ligase forms a dimer, the catalytic core on one E3 complex may be brought into close proximity to the substrate bound on another E3 complex. Therefore, the ubiquitination process could be an intermolecular transfer of ubiquitin from one ligase complex to the other dimer in the complex, rather than an intramolecular transfer within one ligase complex (Fig. 9B). In fact, another F box-containing protein, ␤-Trcp1, forms a homodimer or a heterodimer with ␤-Trcp2 (33). However, we have observed hyperbolic instead of sigmoidal kinetics for IB␣ ubiquitination in vitro, suggesting that there is no cooperativity between the two substrate-binding sites in the SCF ␤-Trcp1 E3 dimer (34).
Ubiquitination and Phosphorylation of Skp2 Is Inhibited by p27-Skp2 is phosphorylated and auto-ubiquitinated in vitro. The Skp2 phosphorylation is mediated by Cdk2/E and is dependent on Cks1. This phosphorylation is required for Skp2 auto-ubiquitination. p27, on the other hand, decreases both Skp2 phosphorylation and ubiquitination. The physiological significance of Skp2 phosphorylation and ubiquitination is not known yet. The protein level of Skp2 is regulated by ubiquitination during the cell cycle. In late mitosis and the G 1 phase, Skp2 is maintained at a low level through APC/C Cdh1 -mediated ubiquitination (35). More cellular studies need to be done to find out whether Skp2 auto-ubiquitination plays a role in the down-regulation of the Skp2 protein level at the end of S phase before APC/C Cdh1 is activated.