document

© 2001 Macmillan Magazines Ltd

brief communications

NATURE CELL BIOLOGY VOL 3 MARCH 2001 http://cellbio.nature.com 321

The cell-cycle regulatory protein Cks1 is required for SCFSkp2-mediated ubiquitinylation of p27

Dvora Ganoth*, Gil Bornstein*, Tun K. Ko†, Brett Larsen‡, Mike Tyers‡, Michele Pagano†§ and Avram Hershko*¶

*Unit of Biochemistry, B. Rappaport Faculty of Medicine, Technion-Israel Institute of Technology, Haifa, 31096, Israel†Department of Pathology and Kaplan Comprehensive Cancer Center, New York University School of Medicine, New York 10016, USA

‡Program in Molecular Biology and Cancer, Samuel Lunenfeld Research Institute, Mount Sinai Hospital, 600 University Avenue, Toronto, M56 1X5, Canada.§e-mail: [email protected]

¶e-mail: [email protected]

The cyclin-dependent kinase (CDK) inhibitor p27 isdegraded in late G1 phase by the ubiquitin pathway1,allowing CDK activity to drive cells into S phase2.Ubiquitinylation of p27 requires its phosphorylation at Thr187 (refs 3, 4) and subsequent recognition by S-phasekinase associated protein 2 (Skp2; refs 5–8), a memberof the F-box family of proteins that associates with Skp1,Cul-1 and ROC1/Rbx1 to form an SCF ubiquitin ligasecomplex9. However, in vitro ligation of p27 to ubiquitincould not be reconstituted by known purified componentsof the SCFSkp2 complex. Here we show that the missingfactor is CDK subunit 1 (Cks1), which belongs to the high-ly conserved Suc1/Cks family of proteins that bind tosome CDKs and phosphorylated proteins and are essen-tial for cell-cycle progression. Human Cks1, but not othermembers of the family, reconstitutes ubiquitin ligation ofp27 in a completely purified system, binds to Skp2 andgreatly increases binding of T187-phosphorylated p27 toSkp2. Our results represent the first evidence that an SCFcomplex requires an accessory protein for activity as wellas for binding to its phosphorylated substrate.

Understanding of the molecular mechanisms of p27 degradationis of considerable importance, as destabilization of p27 occursin many aggressive human cancers (reviewed in ref. 10). We5

and others6,7 have shown that degradation of p27 requires the SCFSkp2

ubiquitin-ligase complex; however, an important piece of the puzzlehas hitherto been missing. We reported that whereas purified recom-binant Skp2 greatly stimulates ubiquitin ligation of p27 in Skp2-depleted extracts, this activity cannot be reconstituted with purifiedcomponents of the SCFSkp2 complex5, indicating the involvement ofone or more further components in this process. We therefore soughtto isolate these components from extracts of HeLa cells.

We first fractionated these extracts on DEAE–cellulose into twofractions — fraction 1, which did not bind to the resin, and fraction2, which contained all proteins that bound to the column and wereeluted with high salt concentration11. We tested the ability of thesefractions to promote ligation of 35S-labelled p27 to methylatedubiquitin (MeUb) in the presence of purified recombinant Skp1,Skp2, Cul1, ROC1, Cdk2–cyclin E, E1 and the Cdc34 E2 protein.MeUb is preferred to native ubiquitin for the ligation assay3,because it terminates formation of polyubiquitin chains and thuscauses the accumulation of easily detectable, discrete, low-molecu-lar-mass derivatives rather than a ‘smear’ of polyubiquitinylatedp27. Addition of fraction 2 did not stimulate p27 ubiquitinylation

in this assay (data not shown). In contrast, addition of fraction 1strongly stimulated ubiquitin ligation of p27 in the presence of thepurified components described above (Fig. 1a, lane 3). Fraction 1by itself had no activity. The activity of fraction 1 was not destroyedby heating at 90 °C for 10 min (Fig. 1a, lane 4). However, the activefactor is a protein, as indicated by the observation that incubationof heat-treated fraction 1 with trypsin completely destroyed itsactivity (see Supplementary Information). While these experimentswere in progress, it was reported that ligation of p27 to ubiquitinrequires fraction 1, and it has been proposed that Nedd8 is theactive component in this fraction12. Nedd8 is a highly conservedubiquitin-like protein that is ligated to different cullins, and liga-tion of Nedd8 to Cul-1 has been shown to stimulate (although it isnot absolutely required) the activity of the SCFβ-TrCP complex in lig-ation of ubiquitin to IκBα (refs 13–15). We found, however, thatrecombinant purified Nedd8 could not replace the factor fromfraction 1 in promoting p27–ubiquitin ligation and that fraction 1was still required after conversion of Cul1 to the Nedd8-modifiedform (see Supplementary Information). We conclude that the fac-tor required to reconstitute the activity of the purified SCFSkp2 com-plex is a heat-stable protein other than Nedd8.

We then purified the stimulating activity in fraction 1 to near-homogeneity (see Supplementary Information) and identified theprotein as human Cks1. Cks1 is a member of the Suc1 (suppressorof Cdc2 mutation)/Cks family of cell-cycle regulatory proteins.Suc1 (ref. 16) and Cks1 (ref. 17) were discovered in fission and bud-ding yeast, respectively, as essential gene products that interact withCDKs. Homologues from different species share extensive sequenceconservation, and the two human homologues can functionallysubstitute for Cks1 in budding yeast18. Crystal structures of the twohuman homologues and the fission yeast Suc1 have shown thatthey share a four-stranded β-sheet that is involved in binding to aCDK catalytic subunit19,20. In addition, they share a highly con-served phosphate-binding site, positioned on a surface that is con-tiguous to the CDK catalytic site in the Cks–CDK complex19. Cksproteins are involved in several cell-cycle transitions, including theG1-to-S-phase transition, entry into and exit from mitosis20, butthe molecular basis of their different activites is not well under-stood. With the exception of Cln2/Cln3–Cdk1 complexes frombudding yeast, which are activated by Cks1 (ref. 21), Cks proteinsdo not directly affect the catalytic activity of CDKs. However, Cksproteins can promote phosphorylation by CDKs at several sites insome substrates. It has been proposed that by simultaneously bind-ing to a partially phosphorylated protein and to a CDK, Cks pro-teins increase the affinity of the kinase for the substrate and thus



NATURE CELL BIOLOGY VOL 3 MARCH 2001 http://cellbio.nature.com322

accelerate subsequent multiple phosphorylations20. Indeed, Cksproteins promote CDK-catalysed multiple phosphorylations ofsubunits of the cyclosome/APC22,23, as well as G2/M-phase regula-tors such as Cdc25, Myt1 and Wee1 (ref. 24). These activities mayprovide at least a partial explanation of the functions of Cks pro-teins in exit from mitosis and entry to mitosis; however, how Cksproteins regulate G1 phase is not well understood.

We confirmed the identification of Cks1 as the factor that isrequired for p27–ubiquitin ligation by replacing it with recombi-nant protein. Cks1 produced by in vitro translation (Fig. 1b, lane 3)or bacterially expressed, purified Cks1 (Fig. 1b, lane 6) effectivelyreplaced the factor in this reaction. Surprisingly, this action wasfound to be specific for Cks1 and was not shared by other membersof the Cks/Suc1 family of proteins. Thus, human Cks2 (which is81% identical and 90% similar to Cks1) as well its homologue in fis-sion yeast, Suc1, were completely inactive in this reaction, eitherwhen produced by in vitro translation (Fig. 1b, lane 4) or as bacteri-ally expressed purified proteins (Fig. 1b, lanes 7, 8). Purified recom-binant Cks2 and Suc1 did not stimulate p27–ubiquitin ligation even

when added at up to 50-fold higher concentrations, despite beingfunctional, as demonstrated by their ability to promote multiplephosphorylation of Cdc27 by Cdk1 (see SupplementaryInformation). The combined evidence thus indicates that the activ-ity of Cks1 in p27–ubiquitin ligation is specific and is not shared byother members of this protein family.

We next investigated how Cks1 promotes ligation of ubiquitinto p27. Cks1 does not seem to be required for the activity of allmammalian SCF complexes. In the well-characterized case of SCFβ-

TrCP, the purified complex carries out robust ubiquitinylation of IκBin vitro25. We also found that addition of Cks1 had no influence onthe rate of ligation of ubiquitin to phosphorylated IκBα by purifiedSCFβ-TrCP (data not shown). It seemed more likely that Cks1 isspecifically involved either in the activity of the SCFSkp2 complex orin some other process that is necessary for p27–ubiquitin ligation.As p27 must be phosphorylated by Cdk2 at T187 for recognition bythe SCFSkp2 complex5,7 and as Cks proteins may stimulate the pro-tein-kinase activity of some, but not all, CDK–cyclin complexes21, itseemed possible that Cks1 stimulates phosphorylation of p27 byCdk2. However, as shown in Fig. 2a, p27 was rapidly phosphorylat-ed by Cdk2–cyclin E in the absence of Cks1, and addition of Cks1had no significant influence on this process. The idea that Cks1 actsat a step subsequent to the phosphorylation of p27 is supported bythe fact that when purified p27 was first phosphorylated by incu-bation with Cdk2–cyclin E and [γ-32P]ATP, its subsequent ligationto MeUb still required Cks1 (Fig. 2b).

We next investigated whether the step that is affected by Cks1 isthe binding of phosphorylated p27 to Skp2. In these experimentswe used the Skp2–Skp1 complex instead of Skp2, because in theabsence of Skp1, recombinant Skp2 was not expressed in an abun-dant soluble form in insect cells. We have previously detected low,but significant, levels of binding of 35S-labelled, in vitro-translatedp27 to Skp2–Skp1 (by immunoprecipitation with an anti-Skp2antibody), which was dependent upon its phosphorylation at T187by Cdk2–cyclin E5. Using a similar procedure for the current study,we found that binding of p27 to Skp2–Skp1 is greatly stimulated byCks1 (Fig. 2c, compare lanes 2 and 3). This activity required phos-phorylation of p27 at T187, as the non-phosphorylatable mutantp27(T187A) did not bind even in the presence of Cks1 (Fig. 2c,lanes 4, 5). To determine whether this activity of Cks1 also occursin a completely purified system devoid of reticulocyte lysate that ispresent in preparations of in vitro-translated p27, we carried out asimilar experiment using bacterially expressed, purified p27 thathad been phosphorylated by [γ-32P]ATP. In this case there was somenon-specific binding of phosphorylated p27 to anti-Skp2–proteinA beads in the absence of Skp2. Nonetheless, a marked stimulationof the specific binding of [32P]p27 to Skp2–Skp1 by Cks1 could beobserved (Fig. 2d). We conclude that Cks1 greatly stimulates bind-ing of phosphorylated p27 to Skp2.

We sought to gain insight into the problem of how Cks1 pro-motes binding of phosphorylated p27 to the Skp2–Skp1 complex.One possibility is that Cks1 acts as an adaptor that binds to Skp2and to phosphorylated p27. We tested this model by incubation ofin vitro-translated, 35S-labelled Cks1 and Cks2 in the presence orabsence of Skp2–Skp1, which was followed by immunoprecipita-tion with anti-Skp2 antibodies. Strong binding of [35S]Cks1 to theSkp2–Skp1 complex was observed (Fig. 3a, lane 2). Under similarconditions, [35S]Cks2 did not bind to Skp2–Skp1 (Fig. 3a, lane 4).As the Skp2–Skp1 complex was used in these experiments (becauseof the lack of recombinant native Skp2), we investigated whetherCks1 binds to Skp1 in the absence of Skp2. We incubated [35S]Cks1either with histidine-tagged Skp1 (His6–Skp1) or withSkp2–His6–Skp1 complex, and estimated the extent of binding toNi–NTA–agarose beads. Cks1 bound strongly to Skp2–His6–Skp1,but not to His6–Skp1 (Fig. 3b). We therefore conclude that humanCks1 binds specifically to the Skp2–Skp1 complex, probablythrough Skp2.

If Cks1 acts as an adaptor that mediates binding of Skp2 to its

[35S]p27

p27–(Ub)n

[35S]p27

p27–(Ub)n

a

b

Fraction 1

Cul-1–ROC1

Skp2–Skp1

Fraction 1, heated

–

+

+

+

+

+

+

–

–

+

+

–

–

–

–

–

–

+

–

+

No

addi

tion

Fact

orC

ks1

IVT

Cks

2 IV

TR

etic

. lys

.C

ks1

Suc

1

Cks

2

1 2 3 4 5

1 2 3 4 5 6 7 8

Figure 1 Identification of Cks1 as the factor required for p27–ubiquitin liga-tion. a, Assay of ligation of [35S]p27 to MeUb (see Methods). Where indicated, frac-tion 1 (5 µg protein) or heat-treated fraction 1 (~50 ng) was added. The upperbracket indicates a ladder of bands of relative molecular mass >27,000, corre-sponding to polyubiquitinated p27; the lower bracket indicates unconjugated[35S]p27. b, Cks1, but not other Cks proteins, is required for p27–ubiquitin ligation.Where indicated, the following proteins were added: ‘Factor’, 0.02 µl of pooled frac-tions 28 and 29 from the peak of the Superdex column, which is the final step ofpurification of the factor required for p27 ubiquitinylation (see SupplementaryInformation); ‘Cks1 IVT’, 0.3 µl of in vitro-translated Cks1; ‘Cks2 IVT’, 0.3 µl of invitro-translated Cks2; ‘Retic. lys.’, 0.3 µl of reticulocyte lysate translation mix;Cks1, Cks2 and Suc1, 2 ng of the corresponding bacterially expressed, purifiedproteins. In vitro-translated, 35S-labelled Cks1 and Cks2 in lanes 3 and 4 are not vis-ible as they migrated off the gel.




substrate, it would be expected to bind to phosphorylated p27and/or to its complex with Cdk2–cyclin E. We have previouslyshown that Skp2 binds to a phosphopeptide that corresponds to thecarboxy-terminal 19 amino acids of p27, with phosphorylated thre-onine at position 187 (ref. 5). Here we found that binding of Skp2

to phosphopeptide–sepharose beads (but not to control beads thatcontained an identical, but unphosphorylated, p27-derived pep-tide) is greatly increased by Cks1 (Fig. 3c). These findings indicatethat binding to this phosphopeptide can serve as a valid tool tostudy the Cks1-assisted interaction between Skp2 and p27. Using

[32P]p27

Control + Cks1

No a

ddito

nFa

ctor

Cks

1

p27(MeUb)2

p27(MeUb)1

a

b

+– + + +Skp2–Skp1

+–Cks1 –– + WT

T187

A

WT T187A Inputs

Bound

+– +Skp2–Skp1

Cks1 –– + Inpu

t

Bound

c

d

[35S]p27

[32P]p27

[32P]p27

Time (min) 5 20 5 20

1 2 3 4

1 2 3

1 2 3 4 5 6 7

Figure 2 Cks1 increases binding of phosphorylated p27 to Skp2. a, Cks1does not affect phosphorylation of p27 by Cdk2–cyclin E. Purified p27 was phos-phorylated as described in Methods, except that mixtures were incubated at 20 °Cfor the indicated times. Where indicated, 2 ng of purified Cks1 was added. Samples(1 µl) were used for SDS–PAGE and autoradiography. b, Cks1 acts at a stage thatis subsequent to phosphorylation of p27. [32P]p27 was prepared as described inMethods. Where indicated, 0.02 µl of ‘Factor’ (see Fig. 1b) or 1 ng of purifiedrecombinant human Cks1 was added. Using this purified system, we did notobserve MeUb-conjugated proteins larger than their respective di-ubiquitinylated

forms, as opposed to the four or five conjugates observed using in vitro-translated[35S]p27 (Fig. 1). Ubiquitin may be ligated to only two Lys residues in p27, and thelarger conjugates may contain short polyubiquitin chains (derived from ubiquitinpresent in reticulocyte lysates) terminated by MeUb. c, Cks1 enhances binding ofp27 to Skp2–Skp1, depending on phosphorylation at T187. Binding of 35S-labelledwild-type (WT) or mutant (T187A) p27 to Skp2–Skp1 (see Methods). Where indicat-ed, 1 ng of purified Cks1 was added to the incubation. Inputs show 5% of the start-ing material. d, Cks1 increases binding of [32P]p27 to Skp2–Skp1. The experimentwas similar to that in c, except that [35S]p27 was replaced with [32P]p27.

– + +Skp2–Skp1

[35S]Cks1 Inputs

Bound to α-Skp2

Cks

2

–

[35S]Cks2

Skp2–His6–Skp1

His6–Skp1

[35S]Cks1 boundto Ni–NTA

– +

– +–

–

Inpu

t

a

b

P–p

27 b

eads

[35S]Cks1

Inpu

t

p27

bead

sd

c

P–p

27 b

eads

[35S]Skp2

Inpu

tp27

bead

s

Cks1 – + + +–

1 3 4 6

Cks

1

2

1 3 42 1 3 42

1 3 4 525

Figure 3 Binding of Cks1 to Skp2 and phosphorylated p27. a, Cks1, but notCks2, binds to Skp2–Skp1. Binding of 35S-labelled Cks1 or Cks2 to Skp2–Skp1was assayed using a similar procedure to that used to assay binding of p27 toSkp2–Skp1 (see Methods), except that Cdk2–cyclin E, ATP and the ATP-regenerat-ing system were omitted. Where indicated, 1 µl Skp2–Skp1 was added. b, Cks1does not bind to Skp1. Binding of [35S]Cks1 to His6–Skp1 or Skp2–His6–Skp1 (1 µleach) was determined as in a, except that Ni–NTA–agarose beads (10 µl; Quiagen)were used for precipitation. In both a and b, inputs show 5% of the starting materi-al. c, Cks1 stimulates binding of Skp2 to p27 phosphopeptide. Sepharose beads,

to which was bound either a peptide corresponding to the 19 C-terminal amino acidresidues of p27 (p27 beads), or a similar peptide containing phosphorylated T187(P–p27 beads), were prepared as described5. In vitro-translated [35S]Skp2 (3 µl)was mixed with 15 µl of the corresponding beads in the absence (lanes 1, 3) orpresence of 10 ng (lane 4) or 100 ng (lanes 2, 5) Cks1. After rotation at 4 °C for 2h, beads were washed 4 times with RIPA buffer. d, Cks1 binds to p27 phosphopep-tide. [35S]Cks1 (2 µl) was mixed with the indicated beads, and beads were treatedas in c. Inputs in c and d show 10% of the starting material.



the same p27-derived peptide beads, we observed significant bind-ing of [35S]Cks1 to phosphorylated p27 peptide, but not to unphos-phorylated p27 peptide (Fig. 3d). These findings indicate that Cks1binds directly to phosphorylated T187 in p27 and show that thepresence of Cdk2–cyclin E is not required for binding of Skp2 tophosphorylated p27.

We have shown that Cks1 is essential for ubiquitin ligation ofp27 and that it greatly enhances binding of T187-phosphorylatedp27 to the Skp2 component of the SCF complex. The specificity ofthe binding of Cks1 to Skp2 may account for its specific function inp27 ubiquitinylation. Mutagenesis and structural studies are need-ed to elucidate the molecular details of the interaction of Cks1 withSkp2 and with phosphorylated p27, as well as any conformationalchanges in Skp2 and Cks1 that are elicited by their interaction.Although Cks1 binds directly to the p27 phosphopeptide (presum-ably through its phosphate-binding site), it may also bind toCdk2–cyclin E, with which phosphorylated p27 is associated.Although the presence of Cdk2–cyclin E is not obligatory, suchmultiple contacts between SCFSkp2 and phosphorylated p27, medi-ated by Cks1, may greatly increase the affinity of ubiquitin ligase forits specific substrate.

Although our evidence for the function of Cks1 in p27–ubiqui-tin ligation is the result of in vitro reconstitution experiments, it hasalso been found that degradation of p27 is inhibited in Cks1–/– mice(S. Reed, personal communication). The combined biochemicaland genetic evidence thus indicates an essential function of Cks1 inubiquitinylation and degradation of p27. This function may pro-vide a further mechanism by which degradation of p27 in the cellcycle is regulated. Expression of Cks1 messenger RNA in humancells oscillates markedly during the cell cycle, being negligible in G1phase and rising to high levels in S phase18. Furthermore, downreg-ulation of Cks1 transcripts has been reported in epithelial cells afterinhibition of growth by transforming factor-β (ref. 26). It is possi-ble that, in G0/G1 phase, degradation of p27 is prevented not onlyby low levels of Skp2 and cyclin E, but also by the absence of Cks1.Thus, Cks1 may have an essential function in the programmeddegradation of p27, and therefore in the progression of cells fromG0/G1 phase to S phase.

MethodsProteins.His6-tagged p27 and Cdc34 were expressed in Escherichia coli and were purified by nickel–agarose

chromatography. Cks2 and p13Suc1 were expressed in bacteria and purified by gel-filtration chromatog-

raphy. His6–Skp1–Skp2, His6–Skp1–β-TrCP, His6–cyclin E–Cdk2 and His6–Cul-1–ROC1 were produced

by co-infection of 5B insect cells with baculoviruses encoding the corresponding proteins and were

purified by nickel–agarose chromatography as described3,5. The approximate concentrations of recom-

binant proteins in these preparations were (in pmol µl–1): Skp1, 5; Skp2, 0.5; Cul-1, 4; ROC1, 1; cyclin

E, 8; Cdk2, 1.5. Purified recombinant human Nedd8 was a gift from C. Pickart (John Hopkins Univ.),

and purified recombinant human Cks1 was a gift from S. Reed (Scripps Inst.). Purified, glutathione- S-

transferase (GST)-tagged IκBα(1–154) and its constitutively active kinase IKKβS177E,S181E were provided

by Z-Q. Pan (Mount Sinai Sch. Med., New York). p27, Skp2 and Cks proteins labelled with 35S were

prepared by in vitro transcription–translation, using the TnT Quick kit (Promega) and 35S-methionine

(Amersham).

Assay of p27–ubiquitin ligation.Unless otherwise stated, the reaction mixture contained the following in a volume of 10 µl: 40 mM

Tris–HCl pH 7.6, 5 mM MgCl2, 1 mM dithiothreitol (DTT), 10% (v/v) glycerol, 10 mM phosphocrea-

tine, 100 µg ml–1 creatine phosphokinase, 0.5 mM ATP, 1 mg ml–1 soybean trypsin inhibitor, 1 µM

ubiquitin aldehyde, 1 mg ml–1 methylated ubiquitin, 1 pmol E1, 50 pmol Cdc34, 0.25 µl Skp2–Skp1,

0.25 µl Cul-1–ROC1, 0.1 µl cyclin E–Cdk2, 0.5 µl [35S]p27 and the indicated additions. After incuba-

tion at 30 °C for 1 h, samples were subjected to SDS–PAGE and autoradiography. Ligation of IκBα to

ubiquitin was assayed as described27, except that baculovirus-expressed, purified Skp1–β-TrCP was

used (5 pmol Skp1, ~1 pmol β-TrCP).

Preparation of 32P-labelled, purified p27 and assay of its ubiquitinylation.Purified p27 (0.18 µg) was incubated for 60 min at 30 °C with 0.25 µl Cdk2–cyclin E in a reaction

mixture containing the following in a volume of 10 µl: 50 mM Tris–HCl pH 7.6, 5 mM MgCl2, 1 mM

DTT, 10% glycerol, 1 mg ml–1 soybean trypsin inhibitor, 1 µΜ okadaic acid and 100 µM [γ-32P]ATP

(~50 µCi); this preparation is referred to as [32P]p27. Ligation of p27 to MeUb was assayed as

described above, with the following changes: [35S]p27 was replaced with [32P]p27, the concentration of

unlabelled ATP was increased to 2 mM (for more complete isotopic dilution of labelled ATP present in

the preparation of [32P]p27), and 1 µM okadaic acid was added.

Assay of binding of p27 to Skp2–Skp1.The reaction mixture contained the following in a volume of 10 µl: 40 mM Tris–HCl pH 7.6, 2 mg ml–1

BSA, 1 µl [35S]p27, 1 µl Cdk2–cyclin E, 1 µl Skp2–Skp1, and MgCl2, ATP, DTT, phosphocreatine and

creatine phosphokinase at concentrations similar to those described above for the p27–ubiquitin liga-

tion assay. After incubation at 30 °C for 30 min, 6 µl of Affi-prep–protein A beads (Bio-Rad), to which

polyclonal rabbit antibody against full-length Skp2 (ref. 5) had been covalently linked using dimethyl

pimelimidate28, was added. Samples were rotated with anti-Skp2–protein A beads at 4 °C for 2 h, and

beads were then washed 4 times with 1-ml portions of RIPA buffer28. After elution using SDS elec-

trophoresis sample buffer, samples were subjected to SDS–PAGE and autoradiography.

RECEIVED 8 NOVEMBER 2000; REVISED 11 DECEMBER 2000; ACCEPTED 4 JANUARY 2001;PUBLISHED 15 FEBRUARY 2001.

1. Pagano, M. et al. Science 269, 682–685 (1995).

2. Sherr, C. J. & Roberts, J. M. Genes Dev. 13, 1501–1512 (1999).

3. Montagnoli, A. et al. Genes Dev. 13, 1181–1189 (1999).

4. Nguyen, H., Gitig, D. M. & Koff, A. Mol. Cell. Biol. 19, 1190–1201 (1999).

5. Carrano, A. C., Eytan, E., Hershko, A. & Pagano, M. Nature Cell Biol. 1, 193–199 (1999).

6. Sutterluty, H. et al. Nature Cell Biol. 1, 207–214 (1999).

7. Tsvetkov, L. M., Yeh, K. H., Lee, S., Sun, H. & Zhang, H. Curr. Biol. 661–664 (1999).

8. Zhang, H., Kobayashi, R., Galaktionov, K. & Beach, D. Cell 82, 915–925 (1995).

9. Kipreos, E. & Pagano, M. Genome Biol. 1, 3002.1–3002.7 (2000).

10. Slingerland, J. & Pagano, M. J. Cell. Physiol. 183, 10–17 (2000).

11. Hershko, A., Heller, H., Elias, S. & Ciechanover, A. J. Biol. Chem. 258, 8206–8214 (1983).

12. Podust, V. N. et al. Proc. Natl Acad. Sci. USA 97, 4579–4584 (2000).

13. Furukawa, M., Zhang, Y., McCarville, J., Ohta, T. & Xiong, Y. Mol. Cell. Biol. 20, 8185–8197 (2000).

14. Read, M. A. et al. Mol. Cell. Biol. 20, 2326–2333 (2000).

15. Wu, K., Chen, A. & Pan, Z. Q. J. Biol. Chem. 275, 32317–32324 (2000).

16. Hayles, J., Beach, D., Durkacz, B. & Nurse, P. Mol. Gen. Genet. 202, 291–293 (1986).

17. Hadwiger, J. A., Wittenberg, C., Mendenhall, M. D. & Reed, S. I. Mol. Cell. Biol. 9, 2034–2041

(1989).

18. Richardson, H. E., Stueland, C. S., Thomas, J., Russell, P. & Reed, S. I. Genes Dev. 4, 1332–1334 (1990).

19. Bourne, Y. et al. Cell 84, 863–874 (1996).

20. Pines, J. Curr. Biol. 11, 1399–1402 (1996).

21. Reynard, G. J., Reynolds, W., Verma, R. & Deshaies, R. J. Mol. Cell. Biol. 20, 5858–5864 (2000).

22. Patra, D. & Dunphy, W. G. Genes Dev. 12, 2549–59 (1998).

23. Shteinberg, M. & Hershko, A. Biochem. Biophys. Res. Commun. 257, 12–18 (1999).

24. Patra, D., Wang, S. X., Kumagai, A. & Dunphy, W. G. J. Biol. Chem. 274, 36839–36842 (1999).

25. Tan, P. et al. Mol. Cell 3, 527–533 (1999).

26. Simon, K. E., Cha, H. H. & Firestone, G. L. Cell Growth Differ. 6, 1261–1269 (1995).

27. Chen, A. et al. J. Biol. Chem. 275, 15432–15439 (2000).

28. Harlow, E. & Lane, D. in Antibodies. A Laboratory Manual (eds Harlow, E. & Lane, D) (Cold Spring

Harbor Laboratory Press, Cold Spring Harbor, New York, 1988).

ACKNOWLEDGEMENTSWe thank C. Segal for technical assistance, S. Reed for sharing results before publication, Z-P. Pan, J.Pines, C. Pickart, S. Reed and E. Yeh for reagents, R. Piva and A. Carrano for their contribution to thiswork, and J. Bloom and L. Yamasaki for critical reading of the manuscript. M.P. also thanks L.Yamasaki and L. Bragi for their continuous support. This work was supported by grants from the IsraelScience Foundation and the Human Frontier Science Program Organization (HFSPO; to A.H.), by anIrma T. Hirschl Scholarship and grants from the HFSPO and the NIH (to M.P), and by grants fromthe National Cancer Institute of Canada (to M.T). Part of this work was done during the stay of A.H.at New York Univ. Sch. Med. (on sabbatical leave), where he was supported by a UICCYamagiwa–Yoshida Memorial International Cancer Study Grant.Correspondence and requests for materials should be addressed to M.P.. Supplementary Information isavailable on Nature Cell Biology’s website (http://cellbio.nature.com) or as paper copy from theLondon editorial office of Nature Cell Biology.



supplementary information


The factor from fraction 1 is a protein. The activity of fraction 1was not destroyed by heating at 90 °C. However, the active factor isa protein, as indicated by the observation that incubation of heat-treated fraction 1 with trypsin completely destroyed its activity(Fig. S1, lane 2). In this experiment, we terminated incubation offraction 1 with trypsin by addition of excess soybean trypsininhibitor (STI), to prevent proteolytic damage to the other compo-nents of the system, which were added after trypsin treatment. STIefficiently blocked trypsin activity, as was shown in a control exper-iment in which we added STI to heated fraction 1 before incubationwith trypsin (Fig. S1, lane 3). In this incubation, there was no sig-nificant reduction in p27–ubiquitin ligation.The factor from fraction 1 is not Nedd8. While these studies werein progress, it has been reported1 that ligation of p27 to ubiquitinrequires fraction 1, and it has been proposed that Nedd8 is theactive component in fraction 1 (ref. 1). Nedd8 (called Rub-1 inyeast) is a highly conserved, ubiquitin-like protein that is ligated todifferent cullins, including Cul-1 (reviewed in ref. 2). Ligation ofNedd8 to Cul-1 has been shown to stimulate (although it is notabsolutely required) the activity of the SCFβ-TrCP complex in ligationof ubiquitin to IκBα3–5. As we used 35S-labelled p27 produced by invitro translation in reticulocyte lysates for the above-describedexperiments, and as reticulocyte lysates contain the enzymesrequired for ligation of Nedd8 to cullins6, it was possible that underthese conditions Nedd8 could be ligated to Cul-1. We found, how-ever, that recombinant purified Nedd8 could not replace the factorfrom fraction 1 in promoting p27–ubiquitin ligation (Fig. S2a). Toexamine this problem further, we purified the enzymes that ligateNedd8 to Cul-1 by affinity chromatography onGST–Nedd8–sepharose (see Methods). Incubation of Cul-1 withNedd8 and its purified conjugating enzymes converted ~50% ofCul-1 molecules to the Nedd8-conjugated form, which migratesmore slowly in SDS–polyacrylamide-gel electrophoresis(SDS–PAGE; Fig. S2b). The slower-migrating form indeed containsNedd8, as was verified by immunoblotting with a specific anti-Nedd8 antibody (data not shown).

We then tested the activity of these preparations of Nedd8-con-

jugated and unmodified Cul-1 in ubiquitinylation of p27 in thepresence or absence of heat-treated fraction 1. To do this, we usedbacterially expressed, purified p27 as the substrate, rather than 35S-labelled p27 translated in reticulocyte lysate, because reticulocytelysates also contain the enzyme(s) that rapidly cleave(s) the amide

Untre

ated

Tryp

sin, S

TI

STI, try

psin

[35S]p27

1 2 3

Figure S1 The heat-stable factor is sensitive to trypsin action. Heat-treatedfraction 1 (~0.1 mg ml–1) was incubated at 37 °C for 60 min with 50 mM Tris–HClpH 8.0, in the absence (lane 1) or presence (lane 2) of 0.6 mg ml–1 of TPCK-treatedtrypsin (Sigma T8642). Trypsin activity was terminated by addition of 2 mg ml–1

soybean trypsin inhibitor (STI). In lane 3, STI was added 5 min before a similar incu-bation with trypsin. Samples corresponding to ~50 ng of heat-treated fraction 1were then assayed for stimulation of p27–ubiquitin ligation.

No ad

dition

Frac

tion

1

Nedd8

p27–(Ub)n

[35S]p27

1 2

Cul-1–Nedd8

Cul-1

Cul-1–Nedd8Fraction 1

Cul-1–+

––+

+++

–––

–––

++–

–

a

b

c

1 2 3 4 5 6

p27–(MeUb)2

p27–(MeUb)1

p27

*

Figure S2 The heat-stable factor is not Nedd8 and is required after modifi-cation of Cul-1 by Nedd8. a, Purified Nedd8 does not replace the factor in stim-ulation of p27–ubiquitin ligation. Where indicated, ~50 ng of heat-treated fraction 1or 100 ng of purified recombinant human Nedd8 was added to the p27–MeUb liga-tion assay. b, Ligation of Nedd8 to Cul-1. Cul-1–ROC1 (3 µl) was incubated with 10µg Nedd8 and 20 µl of purified Nedd8-conjugating enzymes in a 100-µl reactionmixture containing Tris pH 7.6, MgCl2, ATP, phosphocreatine, creatine phosphoki-nase, DTT, glycerol and STI at concentrations similar to those used for thep27–ubiquitin ligation assay. A control preparation of Cul1–ROC1 was incubatedunder similar conditions, but without Nedd8-conjugating enzymes. After incubationat 30 °C for 2 h, samples of control (lane 1) or Nedd8-modified (lane 2) prepara-tions were separated on an 8% SDS–polyacrylamide gel and immunoblotted with ananti-Cul-1 antibody (Zymed). c, SCFSkp2 complex containing Nedd8-modified Cul-1still requires the factor from fraction 1 for p27–ubiquitin ligation. p27–MeUb ligationwas assayed as in ‘Experimental procedures’, except that 35S-labelled p27 wasreplaced with bacterially expressed purified p27 (20 ng), and Cul-1–ROC1 wasreplaced with 2 µl of unmodified or Nedd8-modified Cul-1–ROC1 (see above). Afterincubation 30 °C for 60 min, samples were separated on a 12.5% SDS–polyacry-lamide gel, transferred to nitrocellulose and blotted with an anti-p27 monoclonalantibody (Transduction Laboratories). Asterisk shows a crossreacting protein.




linkage between Nedd8 and Cul-1 (data not shown). Ligation ofp27 to MeUb was followed by immunoblotting with a monoclonalanti-p27antibody. Using this purified system and in the presence ofheat-treated fraction 1, significant formation of mono-ubiquitiny-lated, and less of di-ubiquitiynylated, derivatives of p27 was pro-moted by unmodified Cul-1 (Fig. S2c, lane 2). Using the purifiedsystem, we did not observe MeUb-conjugated proteins larger thantheir respective di-ubiquitinylated forms, as opposed to the four orfive conjugates observed with in vitro-translated [35S]p27 (Fig. 1).Ubiquitin may be ligated to only two Lys residues in p27, and thelarger conjugates may contain short polyubiquitin chains (derivedfrom ubiquitin present in reticulocyte lysates) terminated byMeUb. With Cul-1 conjugated to Nedd8, a modest stimulation ofubiquitinylation of p27 was observed, with a particular increase information of the di-ubiquitin derivative (Fig. S2c, lane 3). In dif-ferent preparations of Cul-1, Nedd8 ligation increased the overallrate of p27–ubiquitin ligation 1.5–3.0-fold. The basal activity of

p27–ubiquitin ligation observed with unmodified Cul-1 was notdue to its significant modification by Nedd8 in insect cells (fromwhich baculovirus-expressed Cul-1 was purified), because similaractivity was observed with a mutant Cul-1 in which Lys 720 at itsspecific Nedd8-ligation site2 was changed to Arg (data not shown).It has also been shown that elimination of Nedd8 modification bya similar mutation significantly reduces, but does not abolish, theactivity of SFCβ-TrCP in ubiqutinylation of IκBα3–5. Importantly,supplementation of fraction 1 was still required for p27–MeUb lig-ation even in the presence of Nedd8-modified Cul-1 (Fig. S2c, lanes5, 6). Similar results were obtained when MeUb was replaced withnative ubiquitin, except that in the latter case high-molecular-masspolyubiquitin derivatives of p27 were formed (data not shown).Thus, our data do not support the conclusions of Podust et al.1 andwe conclude that the active component in fraction 1 is not Nedd8.Purification of the factor and its identification as Cks1. We puri-fied the factor from fraction 1 (see Methods). Fig. S3a shows thefinal step of purification on a gel-filtration column. Activity elutedas a sharp peak at an apparent relative molecular mass of ~10,000(Mr 10K). SDS–PAGE and silver staining of column fractionsrevealed the presence of a single protein of Mr ~10K (Fig. S3b).Elution of the this protein peak coincided with elution of the peakof activity in fractions 27 and 28. However, a similar-sized proteincontinued to be eluted in fractions 30 and 31, in which activitydeclined markedly. To identify the protein(s), we subjected samplesfrom fractions 28 (peak activity) and 31 (subsequent to peak activ-ity) to mass-spectrometric sequencing of tryptic peptides. A tryp-tic peptide of the sequence QIYYSDKYDDEEFEYR, correspondingto amino acids 5–20 of human Cks1, was detected in the Mr ~10Kprotein from both of these fractions. We do not know the reason forthe difference in the activity of the Cks1 protein in these differentfractions, although Cks1 in fraction 31 may be a denatured con-former that exhibited altered exclusion properties in the gel-filtra-tion column.Activity of Cks/Suc1 proteins. To investigate whether all Cks/Suc1proteins used in this study were functional, we examined their activ-ities in promoting multiphosphorylation of cyclosome/anaphase-promoting complex (APC) by Cdk1–cyclinB7,8. Cdk1-catalysedhyperphosphorylation of Cdc27, a subunit of the cyclosome/APC,was markedly stimulated by all three recombinant Cks/Suc1 pro-teins used in this study (Fig. S4). This was indicated by the reductionin levels of the unphosphorylated form of Cdc27 and by its conver-sion to several hyperphosphorylated forms that migrated moreslowly in SDS–PAGE (Fig. S4, lanes 3–5). This large electrophoreticshift, which was promoted by all recombinant Cks/Suc1 proteins,required the activity of Cdk1–cyclin B (Fig. S4, lane 6). All threebacterially expressed Cks/Suc1 proteins used in this study were atleast 95% homogeneous, as indicated by SDS–PAGE andCoomassie-blue staining (data not shown).

a

b

40

35

30

25

20

15

10

5

020

25Fraction 26 27 28 3330 31 3229

22 24 26 28 30 32 34 36

20 8.5

Fraction

21

14

7

[35 S

]p27

liga

ted

to M

eUb

(%)

Mr (K)

Figure S3 Purification of the factor required for p27–ubiquitin ligation andits identification as Cks1. a, Final step of purification by gel-filtration chromatog-raphy. The peak activity from the MonoS step (see Methods) was applied to aSuperdex 75 HR 10/30 column (Pharmacia) equilibrated with 20 mM Tris–HCl pH7.2, 150 mM NaCl, 1 mM DTT and 01% Brij-35. Samples (0.5 ml) were collected ata flow rate of 0.4 ml min–1. Column fractions were concentrated to 50 µl by cen-trifuge ultrafiltration (Centricon-10, Amicon). Samples (0.004 µl) of column fractionswere assayed for their ability to stimulate p27–ubiquitin ligation. Results were quan-tified by phosphorimager analysis and are expressed as the percentage of [35S]p27converted to ubiquitin conjugates. Arrows indicate the elution position of markerproteins of the indicated Mr values × 103. b, Silver staining of samples (2.5 µl) fromthe indicated fractions of the Superdex 75 column, resolved on a 16% SDS–poly-acrylamide gel.

1 2 3 4 5 6

– + + + + –– – – – + –– – – + – ––

Cdk1–cyclin B

Cdc27

Suc1Cks2Cks1 – + – – +

Figure S4 All bacterially expressed Cks/Suc1 proteins stimulate multiphos-phorylation of the Cdc27 subunit of cyclosome/APC. Cyclosomes from S-phase HeLa cells were partially purified13 and were incubated with 500 units ofSuc1-free Cdk1–cyclin B8 as described13. Where indicated, 10 ng µl–1 of the corre-sponding Cks/Suc1 protein was added. Samples were subjected to immunoblottingwith a monoclonal antibody against human Cdc27 (Transduction Laboratories).




MethodsPurification of Nedd8-conjugating enzymes.Purified recombinant human Nedd8 was a gift from C. Pickart. A mixture of Nedd8-conjugating

enzymes (E1-like APP-BP1-Uba3 heterodimer and E2-like Ubc12; refs 6, 9) was co-purified from

lysates of rabbit reticulocytes using a ‘covalent affinity’ chromatography procedure similar to that used

to purify E2s10, except that unfractionated reticulocyte lysate was applied to a column of

GST–Nedd8–sepharose (5 mg ml–1). After washing with 1 M KCl, all proteins bound to immobilized

Nedd8 by thiolester linkages were co-eluted with a solution containing 20 mM dithiothreitol (DTT).

The DTT eluate was concentrated by ultrafiltration to ~10% of the original volume of reticulocyte

lysate. This preparation exhibited strong activity in ligation of Nedd8 to Cul-1, without any detectable

hydrolase activity that removed Nedd8 from Cul-1.

Purification of the factor required for p27–ubiquitin ligation.A frozen pellet from 50 g of HeLa S3 cells (National Cell Culture Center) was a gift from A.

Ciechanover. Cells were disrupted using a nitrogen cell-disruption bomb (Parr, Moline, Illinois) as

described11, except that the buffer also contained 10 µg ml–1 chymostatin and 5 µg ml–1 aprotinin. The

extract was centrifuged at 15,000g for 20 min and supernantants were centrifuged again at 100,000g for

60 min. The supernatant was subjected to fractionation on DEAE–cellulose as described10, except that

2,500 mg of protein was loaded on 250 ml of resin. The fraction that was not adsorbed to the resin

(fraction 1) was collected and was concentrated by centrifuge ultrafiltration to ~10 mg ml–1. Fraction 1

(100 mg protein) was subjected to heat treatment at 90 °C for 10 min. The sample was kept on ice for

30 min, and the precipitate was then removed by centrifugation (10,000g, 15 min). Roughly 99% of the

protein was removed by heat treatment. The supernatant was concentrated by ultrafiltration and was

then applied to a MonoS HR 5/5 column (Pharmacia) equilibrated with 50 mM Tris–HCl, 1 mM DTT

and 0.1% (w/v) Brij-35 (Boehringer). The column was washed with 15 ml of the above buffer and was

then eluted with a gradient of 0–200 mM NaCl. Activity in column fractions was monitored using the

p27–ubiquitin ligation assay in the presence of purified SCFSkp2 components (see below). The peak

fractions of activity were eluted at ~30–40 mM NaCl. The peak-containing factor activity was pooled,

concentrated by centrifuge ultrafiltration and was subjected to the final step of gel-filtration chro-

matography on Superdex-75 (Fig. S3a).

Mass-spectrometric sequencing.The Mr 10K protein from the final step of purification (Fig. S3b) was excised and digested in gel as

described12. Mass-spectrometric analysis was carried out using a Sciex QSTAR mass spectrometer

(MDS-Sciex, Concord, Ontario, Canada). A tryptic peptide of Mr 2163.5 was fragmented from doubly

and triply charged species and yielded a complete match to residues 5–20 of human Cks1.

1. Podust, V. N. et al. Proc. Natl Acad. Sci. USA 97, 4579–4584 (2000).

2. Yeh, E. T., Gong, L. & Kamitani, T. Gene 248, 1–14 (2000).

3. Furukawa, M., Zhang, Y., McCarville, J., Ohta, T. & Xiong, Y. Mol. Cell. Biol. 20, 8185–8197

(2000).

4. Read, M. A. et al. Mol. Cell. Biol. 20, 2326–2333 (2000).

5. Wu, K., Chen, A. & Pan, Z. Q. J. Biol. Chem. 275, 32317–32324 (2000).

6. Osaka, F. et al. Genes Dev. 12, 2263–2268 (1998).

7. Patra, D. & Dunphy, W. G. Genes Dev. 12, 2549–2559 (1998).

8. Shteinberg, M. & Hershko, A. Biochem. Biophys. Res. Commun. 257, 12–18 (1999).

9. Gong, L. & Yeh, E. T. J. Biol. Chem. 274, 12036–12042 (1999).

10. Hershko, A., Heller, H., Elias, S. & Ciechanover, A. J. Biol. Chem. 258, 8206–8214 (1983).

11. Montagnoli, A. et al. Genes Dev. 13, 1181–1189 (1999).

12. Shevchenko, A., Wilm, M., Vorm, O. & Mann, M. Anal. Chem. 68, 850–858 (1996).

13. Yudkovsky, Y., Shteinberg, M., Listovsky, T., Brandeis, M. & Hershko, A. Biochem. Biophys. Res.

Commun. 271, 299–304 (2000).

Documents

document