Cell, Vol. 81,279-288, April 21, 1995, Copyright© 1995 by Cell Press A 20S Complex Containing CDC27 and CDC16 Catalyzes the Mitosis-Specific Conjugation of Ubiquitin to Cyclin B Randall W. King,*t Jan-Michael Peters,*t cyclin degradation is required to exit mitosis (Surana et Stuart Tugendreich,$ Mark Rolfe,§ Philip Hieter,$ al., 1993; Holloway et al., 1993). Treatments that interfere and Marc W. Kirschnert with the proteolysis of endogenous cyclin B, however, ar- tDepartment of Cell Biology rest cell division earlier, at anaphase (Holloway et al., Harvard Medical School 1993). This discrepancy can be explained by hypothesiz- Boston, Massachusetts 02115 ing that chromosome segregation and cyclin proteolysis $Department of Molecular Biology and Genetics depend upon common components. Support for this idea Johns Hopkins School of Medicine has emerged recently from studies in budding yeast, in Baltimore, Maryland 21205 which CDC16 and CDC23, genes required for progression §Mitotix, Incorporated through anaphase, have been shown to be required for One Kendall Square the proteolysis of B-type cyclins (Irniger et al., 1995 [this Cambridge, Massachusetts 02139 issue of Cell]). The proteins encoded by these genes form a complex with the CDC27 protein (Lamb et al., 1994), a homolog of which is also required for anaphase progres- Summary sion in mammalian cells (Tugendreich et al., 1995 [this issue of Cell]). Cyclin B is degraded at the onset of anaphase by a Biochemical evidence suggests that cyclin proteolysis ubiquitin-dependent proteolytic system. We have frac- is mediated by the ubiquitin pathway: cyclin B-ubiquitin tionated mitotic Xenopus egg extracts to identify com- conjugates can be observed in mitotic, but not interphase, ponents required for this process. We find that UBC4 Xenopus extracts (Glotzer et al., 1991); mutations in the and at least one other ubiquitin-conjugating enzyme D box that block degradation also interfere with ubiquitina- can support cyclin B ubiquitination. The mitotic speci- tion (Glotzer et al., 1991); and methylated ubiquitin, an ficity of cyclin ubiquitination is determined by a 20S inhibitor of polyubiquitin chain formation, interferes with ~omplex that contains homologs of budding yeast the proteolysis of A- and B-type cyclins in extracts of clam CDC16 and CDC27, Because these proteins are re- eggs (Hershko et al., 1991). quired for anaphase in yeast and mammalian cells, we A complex multistep pathway is required for the covalent refer to this complex as the anaphase-promoting com- attachment of polyubiquitin chains to substrate proteins plex (APC). CDC27 antibodies deplete APC activity, (reviewed by Ciechanover, 1994). The polypeptide ubiqui- while immunopurified CDC27 complexes are sufficient tin is first activated at its C-terminus via thioester formation to complement either interphase extracts or a mixture with El, the ubiquitin-acti~ating enzyme. E1 subsequently of recombinant UBC4 and the ubiquitin-activating en- transfers ubiquitin to a family of ubiquitin-conjugating en- zyme El. These results suggest that APC functions as zymes (E2s), again forming thioester intermediates. Al- a regulated ubiquitin-protein ligase that targets cyclin though certain E2s can transfer ubiquitin directly to sub- B for destruction in mitosis. strates in vitro, the physiologic reaction often requires a third component, termed a ubiquitin-protein ligase or E3. Introduction This component can directly mediate substrate specificity and may also be required to synthesize the polyubiquitin Entrance into mitosis is governed by the protein kinase chain that is presumed to target the substrate for degrada- Cdc2, whose positive regulatory subunits, the mitotic tion by the 26S proteasome complex (reviewed by Peters, cyclins, accumulate throughout interphase. Exit from mito- 1994). In certain cases, substrate specificity is mediated sis requires the inactivation of Cdc2, initiated by rapid by an additional component, such as the human papillo- cyclin B proteolysis that commences at anaphase. Resta- mavirus E6 protein, which interacts with a cellular E3 to bilization in the subsequent interphase enables cyclin B ubiquitinate p53 (Scheffner et al., 1993). to accumulate again, initiating a new mitotic cycle. In the The cyclin ubiquitination reaction is unusual in that it embryonic cell cycle, the regulated activation and inactiva- exhibits specificity at two levels: substrate recognition, tion of mitotic cyclin destruction transforms continuous which is reflected in the requirement for an intact D box, cyclin synthesis into alternating periods of interphase and and temporal control in the limitation of its activity to a mitosis (reviewed by Nasmyth, 1993; King et al., 1994). specific phase of the cell cycle, late mitosis and early G1 Mitotic cyclins contain a short N-terminal sequence, (Hunt et al., 1992; Amon et al:,, 1994). Little is known re- called the destruction box (D box), that is required for their garding the components involved in cyclin B ubiquitina- rapid degradation (Glotzer et a!., 1991). Ectopic expres- tion. Studies in budding yeast have implicated a ubiquitin- sion of nondegradable cyclins arrests the cell cycle with conjugating enzyme, UBC9, in the degradation of both S elevated levels of Cdc2 kinase activity (Murray et al., 1989; and M phase cyclins (Seufert et al., 1995); however, it Ghiara et al., 1991; Luca et al., 1991; Gallant and Nigg, remains unclear whether this enzyme is required for D 1992). This arrest occurs in telophase, suggesting that box-dependent ubiquitination. Fractionation of clam egg extracts has separated two activities distinct from E1 that *These authors contributed equallyto this work. are required for cyclin ubiquitination (Hershko et al., 1994). Cell 280 A Mitotic Extract that it functions as a temporally regulated cyclin-ubiquitin ligase. Peliet Membranest Supernatant 1 Flow L~J O.l-O.6 M Results , through | KCI gradient Q1 Q'2 Cyclin Ubiquitination Activity Requires Both Cell [~ 0'2M I 0"4M Flow 0.6 M KCI KCI Cycle-Regulated and Unregulated Fractions throughIKCt eluate I I , --1 ~ As a substrate for the cyclin ubiquitination reaction, an Q1A QIB iodinated N-terminal fragment of sea urchin cyclin B was used. We monitored the formation of radiolabeled cyclin- B C D E ubiquitin conjugates by SDS-polyacrylamide gel electro- C phoresis (SDS-PAGE) and autoradiography. The ubiquiti- OO OOOO OOO nation and degradation of this protein are dependent upon o ++ ++++ +E++E+ the cell cycle state of crude extracts and require an intact o_~ 58oo 0006 oooa D box (Holloway et al., 1993; data not shown). As a source of factors required for cyclin ubiquitination, concentrated interphase extracts from Xenopus eggs were prepared. Stable mitotic extracts that constitutively degrade cyclin B were then obtained by addition of the nondegradable cyclin B A90 fragment (Glotzer et al., 1991). Our fractionation protocol is shown sche m atically in Fig- ure 1A. As a first step, we prepared pellet, membrane, [12~ and supernatant ($100) fractions by high speed centrifu- gation of diluted crude mitotic extracts. Cyclin ubiquitina- Figure 1. Fractionation of Cyclin Ubiquitination Activity in Mitotic tion activity was recovered in the $100 fraction after recon- Xenopus Egg Extracts into a Regulated and an Unregulated Fraction centration to the original volume, as indicated by the (A) Fractionation scheme. For details see Results. ladder-like appearance of higher molecular mass species (13) Pellet (P), membrane (M), and supernatant ($100) fractions ob- (Figure 1B). Ubiquitin conjugates were observed within tained from mitotic extracts were assayed for their ability to convert a radiolabeled N-terminalfragment of cyclin B ([1251]cyc)into ubiquitin 1-2 min of incubation in the mitotic supernatant and conjugates. reached a steady state within 5 min (data not shown). The (C) Flowthrough (Q1) and eluate (Q2) fractions were obtained from substrate was degraded in this fraction with a half-life of mitotic $100 by ResourceQ chromatographyand assayedindividually 5 rain, similar to that observed in crude extracts (data not or after mixing equal volumes. D box dependencewas tested by addi- shown). No ubiquitination activity was detectable in the tion of a radiolabeledfragment containingtwo D box point mutations (lane marked Q1 plus Q2 with an asterisk). washed pellet or membrane fractions (Figure 1B). (D) Flowthrough(Q1 ~) and eluate (Q2i) fractions were prepared as in We fractionated mitotic $100 by anion exchange chro- (C) using interphase $100 and were tested for their ability to replace matography, using Resource Q as a resin, yielding a Q1 and Q2 derived from mitotic $100. flowthrough fraction (Q1) and a 0.6 M KCI eluate (Q2). (El Fractions Q1 and Q2 were immunodepleted with either purified total mouse IgG control antibody(Q1 c and Q2~) or with purified MPM-2 While neither fraction alone catalyzed cyclin ubiquitina- monoclonal antibody (Qlr" and Q2m) and assayed as in (C). tion, mixing the fractions fully reconstituted activity (Figure 1C). A substrate containing a mutated D box (R42A and A44R) produced only low molecular mass conjugates. To determine whether both of these fractions were mitotically While one activity is found only in mitotic extracts, the regulated, we prepared Q1 and Q2 from interphase $100 other is active throughout the cell cycle and appears to (designated Q1 ~ and Q2~). Cyclin ubiquitination was ob- be an E2. The components required for D box-dependent sewed when Q1 was replaced by Q1 ~, but