Initiation of Cyclin B Degradation by the 26S Proteasome upon Egg Activation

Goldfish were purchased from a local supplier and maintained at 15°C until use. The 20S and 26S proteasomes were purified from immature goldfish oocytes by conventional column chromatography as described (Tokumoto et al., 1995a,b). Xenopus laevis were obtained from a dealer and maintained at 20°C. Xenopus CSF-arrested egg extracts were prepared by the method of Murray et al. (1989).

Electrophoresis proceeded as described by Laemmli (1970), using 12.5% gels under denaturing conditions. Cyclin B degradation was assessed by immunoblotting against anti-goldfish cyclin B (B63 and B112) monoclonal antibodies (Yamashita et al., 1992b). Immunocomplexes were visualized using the ECL detection kit (Amersham Intl., Arlington Heights, IL).

Electroblotted 42-kD fragments of cyclin B were prepared as described (LeGendre and Matsudaira, 1989). The NH₂-terminal amino acids were determined using a protein sequencer (470A; Applied Biosystems, Chiba, Japan).

Full length (Δ0) and NH₂-terminal truncated (Δ41 and Δ68) goldfish cyclin Bs were produced as described (Hirai et al., 1992; Katsu et al., 1993; Yamashita et al., 1995). Mutant cyclin B in which lysine 57 was replaced by arginine (Δ0K57R) was produced as follows. A cDNA clone encoding full length goldfish cyclin B (Hirai et al., 1992) was mutated using a site- directed mutagenesis system (Mutan K; Takara, Tokyo, Japan), following a strategy based on the method of Kunkel (1985), according to manufacturer's instructions. Double-strand, mutated cDNA was prepared by T3 polymerase using single-strand cDNA and the following oligonucleotide: AAGAAGGAAGTGAGGGTGGCGCCCAAGGTGGAG. This oligonucleotide was designed to produce a mutation site (bold type) and a restriction enzyme site (underlined) as described (Yamashita et al., 1995). Mutant clones were screened by digestion with the restriction enzyme and confirmed by sequencing.

Recombinant proteins were produced in E. coli BL21 (DE3) and purified by SDS-PAGE followed by electroelution from the gel, as described previously (Hirai et al., 1992). ³⁵S-labeled cyclins were produced using a TNT T7-coupled Reticulocyte Lysate System (Promega Biotech, Madison, WI) according to the manufacturer's instructions.

Although the 26S proteasome is a ubiquitin-dependent protease in general (Armon et al., 1990; Driscol and Goldberg, 1990; Kanayama et al., 1992), it also catalyzes an ATP-dependent and ubiquitin-independent proteolysis (Tanaka et al., 1983; Matthews et al., 1989; Murakami et al., 1992). Therefore, we initially investigated whether or not the 26S proteasome can degrade in vitro nonubiquitinated, full length goldfish cyclin B produced in E. coli (cyclin Δ0). Goldfish proteasomes (20S and 26S) were purified from immature goldfish oocytes by sequential chromatography (Tokumoto et al., 1995a,b). SDS-PAGE has demonstrated that the 26S proteasome consists of multiple subunits with a molecular mass ranging from 23.5 to 140 kD, whereas the 20S proteasome includes subunits ranging from 23.5 to 31.5 kD (Fig. 1,A). The 26S, but not the 20S proteasome, digested 49-kD cyclin Δ0 and produced a 42-kD cyclin (Fig. 1,B). Protease inhibitors of microbial origin (antipain, chymostatin, and leupeptin) blocked the cyclin digestion at high (200 μM) but not at low concentrations (50 μM). Proteasome inhibitors (MG115, MG132, and PSI) that inhibit the chymotrypsin-like activity and ubiquitin-dependent protein degradation (Figueiredo-Pereira et al., 1994; Rock et al., 1994; Jensen et al., 1995), blocked the digestion at 50 μM (Fig. 1,C). No digestion proceeded when the 26S proteasome was depleted with an anti-proteasome antibody (Fig. 2, A and B). The reaction was also prevented when ATP was depleted from the reaction mixture (Fig. 2 C). These results indicate that the digestion of cyclin B is not due to a contaminating protease in the 26S proteasome fraction, but is catalyzed by the 26S proteasome itself.

We determined whether the 26S proteasome digests the NH₂- or COOH-terminal region of the cyclin B. We used monoclonal anti-cyclin B63, which recognizes the COOH terminus of cyclin B, and the anti-cyclin B112, which recognizes the NH₂ terminus (Hirai et al., 1992; Katsu et al., 1993). The B63 antibody recognized the 42-kD intermediate and the NH₂-terminal deletion mutants of cyclin B, while the B112 did not react with the intermediate but recognized the full length cyclin B (Fig. 3,B). Therefore, we concluded that cyclin B was digested in the NH₂ terminus. To define the cleavage site, the NH₂-terminal amino acid sequence of the electroblotted 42-kD cyclin was directly determined. The site of cyclin B cleavage by the 26S proteasome was the COOH-terminal peptide bond of lysine 57 (K57; Fig. 3 A).

The NH₂-terminal sequences of cyclins, including a consensus sequence called the destruction box, play a critical role in targeting cyclins for degradation (Glotzer et al., 1991; Lorca et al., 1991; Luca et al., 1991; Kobayashi et al., 1992). We therefore examined the role of NH₂-terminal sequences in cyclin digestion by the 26S proteasome in vitro. We produced two NH₂-terminal truncated cyclins: cyclin Δ41 lacking the destruction box and cyclin Δ68 lacking the destruction box and half of the lysine-rich stretch (Fig. 3,A). Neither cyclin Δ68 nor cyclin Δ41 were digested by 26S proteasome (Fig. 3,B). This finding suggests that the NH₂-terminal region of cyclin B, including the destruction box, supplies an interaction site between cyclin B and 26S proteasome that is necessary for the subsequent cutting of cyclin B at K57. Interaction between the NH₂ terminus of cyclin B and the 26S proteasome was also suggested by an experiment with a truncated protein containing the first 89 amino acids of Xenopus cyclin B2 (B2Nt, a gift from Dr. M.J. Lohka, University of Calgary, Calgary, Canada; Velden and Lohka, 1993). B2Nt inhibited the digestion of cyclin B by the 26S proteasome in a dose-dependent manner (Fig. 4). Control lysozyme, a basic and low molecular weight protein like B2Nt, did not inhibit the cyclin digestion (Fig. 4).

The 26S proteasome also digested native cyclin B that had been isolated as a complex with cdc2, yielding a truncated cyclin of ∼42-kD (Fig. 5,A). This was the same size as the fragment produced by digestion of recombinant, full length cyclin B with the 26S proteasome. The digestion of the cdc2–cyclin B complex with 26S proteasome, however, did not cause kinase inactivation of cdc2 (Fig. 5 B).

Mature goldfish oocytes are arrested at metaphase of meiosis II (metaphase II) and can be activated by the contact with water (Yamashita et al., 1992a,b). Cyclin B was degraded within a few minutes after egg activation (Nagahama et al., 1995). A 42-kD fragment of cyclin B transiently appeared during the initial phase of goldfish egg activation. In a partially purified and highly concentrated fraction from egg extracts, intermediate cyclin B was detected 3 min after egg activation (Fig. 5,C). The monoclonal antibody B112, which recognizes the NH₂-terminal region of cyclin B, did not react with the intermediate (Fig. 6 C). These results suggest that NH₂-terminal digestion of cyclin B by the 26S proteasome is not an artifact of in vitro proteolysis but an initial reaction of cyclin B degradation that proceeds upon egg activation.

Although the results so far suggested that the 26S proteasome interacts with the NH₂-terminal region of cyclin B and then cuts it at K57, digestion of cyclin B by the 26S proteasome in vitro is limited to cleavage of a single peptide bond and does not induce kinase inactivation of MPF (Fig. 5 B). We believe that the incomplete digestion of cyclin B by the 26S proteasome in vitro is due to the absence of factors responsible for further degradation of cyclin B. The most likely candidate is ubiquitin and its ligase. In fact, goldfish oocytes contain high levels of free ubiquitins (Tokumoto et al., 1993b).

We purified and characterized the 26S proteasome from immature Xenopus oocyte extracts (Tokumoto and Ishikawa, 1995). We then examined whether or not this proteasome digests goldfish cyclin B, like the goldfish proteasome can. Xenopus 26S, but not 20S proteasomes, digested the NH₂ terminus of goldfish cyclin B and produced the 42-kD intermediate (Fig. 6, A and C). NH₂-terminal truncated cyclins were not digested by the Xenopus 26S proteasome (Fig. 6 B). These results indicate that the Xenopus 26S proteasome can digest goldfish cyclin B, suggesting a similar role of goldfish and Xenopus proteasomes in the regulation of cyclin degradation. We then examined the involvement of 26S proteasome in cyclin B degradation using a Xenopus cell-free system widely used for cell cycle studies, which contains the complete system necessary for cyclin degradation (Murray et al., 1989).

Cyclin Δ0 was completely degraded within 30 min after adding Ca²⁺ to Xenopus egg extracts, although it was stable in the absence of Ca²⁺ (Fig. 7,A). NH₂-terminal truncated cyclins Δ41 and Δ68 were not degraded in Xenopus egg extracts even after activation with Ca²⁺ (Fig. 7,B). In contrast to cyclins Δ41 and Δ68, the 42-kD cyclin fragment (cyclin Δ57), which had been produced by the prior digestion of cyclin Δ0 with the purified 26S proteasome, was degraded in Xenopus extracts after adding Ca²⁺ (Fig. 8). To exclude the possibility that the 42-kD cyclin fragment remained in complex with the NH₂-terminal portion after digestion with the 26S proteasome, we performed gel chromatography under the buffer conditions that were used for cyclin digestion. When cyclin treated with 26S proteasome was chromatographed, 42-kD cyclin and NH₂-terminal fragment (∼9 kD) were clearly separated (Fig. 9). This result indicates that there is no significant interaction between 42-kD cyclin and the NH₂-terminal fragment after digestion.

Our experiment using E. coli-produced cyclins demonstrated that the 26S proteasome restrictively digests the NH₂ terminus of cyclin B and that only cyclins digestible or digested by 26S proteasome are degradable in Xenopus egg extracts activated with Ca²⁺. To confirm that these results are not artifacts derived from the incorrect form of cyclins produced in E. coli, we performed a similar experiment using cyclin B proteins translated in vitro in rabbit reticulocyte lysate. Cyclin B proteins produced in E. coli and translated in vitro gave the same results. Cyclin Δ0 translated in vitro was digested by the 26S , but not the 20S proteasome (Fig. 10,A), and degraded in Xenopus egg extracts (Fig. 10, C and D). NH₂-terminal truncated cyclins Δ41 and Δ68 were resistant to 26S proteasome digestion (Fig. 10,B) and to degradation in Xenopus egg extracts (Fig. 10, C and D). Furthermore, the point mutant, cyclin Δ0K57R (in which the position of cleavage by the 26S proteasome, lysine 57, was converted to arginine), was neither digested by the 26S proteasome nor degraded in Xenopus extracts (Fig. 10, B–D).

Immediately before the transition from metaphase to anaphase, the kinase activity of MPF is inactivated through the degradation of its cyclin B subunit. The mechanism of cyclin degradation, which must be a highly selective process since few other proteins are degraded only at this time, is poorly understood. Using recombinant goldfish cyclin B and purified 26S proteasomes, we investigated the role of proteasomes in the regulation of cyclin degradation during egg activation. We found that purified 26S proteasome digests not only recombinant cyclin B, but also native cyclin B at its NH₂-terminal portion, producing a 42-kD intermediate form. Since the 42-kD cyclin B appears transiently during the initial phase of normal egg activation, this digestion should not be an artifact but rather an initial step in cyclin B degradation upon egg activation. Using Xenopus egg extracts, we also showed that the initial digestion and further degradation of cyclin B are tightly linked. Cyclins Δ41 and Δ68 that are indigestible by the 26S proteasome are not degraded, whereas cyclins Δ0 and Δ57 digested by the 26S proteasome, are degraded. These findings strongly suggest that the initial cutting of the NH₂-terminal region of cyclin B by 26S proteasome is a prerequisite for the subsequent degradation that leads to the inactivation of MPF at the metaphase/anaphase transition.

The NH₂-terminal sequences of cyclin B, including a consensus sequence that is called the destruction box, play a critical role in targeting cyclins for degradation, since truncated sea urchin (Murray et al., 1989), human (Lorca et al., 1991), and clam (Luca et al., 1991) B-type cyclins missing the first 90, 72, or 97 amino acids, respectively, and clam (Luca et al., 1991) and Xenopus (Kobayashi et al., 1992) A-type cyclins missing the NH₂-terminal 60 or 62 amino acids are resistant to degradation. Each of these truncated cyclins continuously activates cdc2, which prevents cells or cellular extracts from leaving mitosis. A truncated protein containing only the first 89 amino acids of Xenopus cyclin B2 (B2Nt), including sequences essential for cyclin degradation in other species, also inhibited cyclin degradation (Velden and Lohka, 1993). These results indicate interaction of the NH₂-terminal portion of cyclin with the destruction machinery.

In this study, we proposed that the initial reaction of cyclin B destruction is the restricted cleavage of its NH₂-terminal portion (K57 in the case of goldfish cyclin B) by the 26S proteasome. As well as cyclin Δ68 that lacks the cutting site K57, cyclin Δ41 containing K57 was not cleaved, indicating that the NH₂-terminal region affords not only the cutting site but also the interaction site necessary for digestion by the 26S proteasome. This notion was confirmed by inhibiting the cyclin digestion with B2Nt that consists of the first 89 amino acid of Xenopus cyclin B2. Cyclins Δ41 and Δ68 were neither digested by the 26S proteasome nor degraded in Xenopus egg extracts activated by Ca²⁺, whereas cyclin Δ0 was digested at K57 by 26S proteasome and degraded in the extracts. In addition, cyclin Δ57 produced by digesting cyclin Δ0 with the 26S proteasome, was degraded in the extracts. These results suggest that only cyclins that have undergone 26S proteasome digestion at K57 can be degraded upon egg activation.

The mechanism of cyclin B degradation after initial cleavage by the 26S proteasome remains to be determined. The most likely candidate will be ubiquitin-dependent proteolysis. Proteins to be degraded by the ubiquitin pathway are ligated to ubiquitin through their lysine amino acid groups and then degraded by the 26S proteolytic complex (Hershko and Ciechanover, 1982). The first evidence that cyclin B degradation is mediated by ubiquitin-dependent proteolysis was provided by Glotzer et al. (1991). Other support for the involvement of a ubiquitin-dependent pathway in the cyclin degradation arises from the observation that methylated ubiquitin, which prevents the polyubiquitination of proteins destined for degradation, delays cyclin degradation in an extract from clam embryos (Hershko et al., 1991). A complex containing cyclin-selective ubiquitin ligase activity has been identified in clam oocytes (Sudakin et al., 1995). These findings suggest that the cell cycle-specific cyclin degradation is mediated by a ubiquitin-dependent proteolytic system.

The cyclin B subunit of MPF must be ubiquitinated immediately before the onset of its destruction at the metaphase/anaphase transition. The findings of the present study indicate that the restricted cleavage of cyclin B triggers its ubiquitination. It is likely that the digestion of the NH₂-terminal restricted portion by the 26S proteasome (K57 in goldfish cyclin B) changes the cyclin structure available for further chemical modifications including ubiquitination, which leads to the complete destruction of the cyclin at metaphase/anaphase transition. Since cyclin Δ57 was destroyed, whereas cyclin Δ68 was not, the lysine residues between amino acids 58 to 68 constitute the most likely ubiquitination site, and cutting by the 26S proteasome at K57 might be necessary to expose them to ubiquitinating enzymes. This notion should be verified by investigating the difference in the three-dimensional structure of cyclins Δ0, Δ41, Δ57, and Δ68.

Cyclin B is absent in immature (prophase I-arrested) goldfish oocytes. Cyclin B is de novo synthesized during oocyte maturation and forms an MPF complex with extant cdc2, which drives the prophase I-arrested oocytes to metaphase II (mature oocytes; Hirai et al., 1992; Katsu et al., 1993; Yamashita et al., 1995). As shown in this study however, the 26S proteasome purified from immature goldfish oocytes can digest cyclin B. If the initial cleave of cyclin B by the 26S proteasome triggers cyclin destruction as proposed in this study, the question remains why cyclin B is stable during oocyte maturation and in mature oocytes but destroyed upon egg activation. Based on the results obtained from clam oocytes, Sudakin et al. (1995) have suggested that the initiation of cyclin degradation is triggered by ubiquitination caused by the activation of cyclin-selective ubiquitin ligase near the end of M-phase, which targets cyclin B for destruction by the 26S proteasome that is constitutively active during the cell cycle. Contrary to this, we found that the 26S proteasome purified from mature goldfish oocytes cannot digest cyclin B and that at least two subunits in 26S proteasomes from immature and mature oocytes differ (Tokumoto, T., Horiguchi, R., Nagahama, Y., unpublished results). These findings suggest that some inhibitory mechanisms preventing cyclin B degradation proceed on the proteasome itself at least during metaphase II arrest. The amount of the proteasome in egg cytosol also changes dramatically during oocyte maturation and egg activation; the lowest level is in mature metaphase II-arrested oocytes, and there is a transient increase between the first and second meiotic cell cycles and upon egg activation in goldfish (Tokumoto et al., 1993a). Further studies should reveal how the subunit composition of proteasomes and their contents during oocyte maturation and egg activation are involved in controlling the cell cycle by regulating cyclin stability.

Stewart et al. (1994) have shown that binding with cdc2 is necessary for the degradation of Xenopus cyclins A and B2 but not for that of cyclin B1. This implies that the mechanisms of cyclin degradation vary according to the types of cyclins. Since goldfish cyclin B exhibits higher homology to Xenopus cyclin B1 (66%) than to cyclin B2 (50%), the mechanism of cyclin degradation by the 26S proteasome proposed in this study may be specific to cyclin B1.

Our present data suggest that the NH₂-terminal restricted cleavage of cyclin B by 26S proteasome allows cyclin to be ubiquitinated. Compared with the idea that the initiation of cyclin destruction is dependent on the activity of ubiquitin ligase, irrespective of 26S proteasome activity (Sudakin et al., 1995), we propose that cyclin destruction is primarily controlled by the activity of 26S proteasome.

Ubiquitination of cyclins has been well studied genetically and biochemically (for review see Murray, 1995). A cyclin-specific ubiquitin ligase complex, the cyclosome, or APC complex, has been characterized in clam and Xenopus, respectively (King et al., 1995; Sudakin et al., 1995). These ubiquitin ligases (E3) catalyze ubiquitination using a specialized ubiquitin carrier protein (E2). Among the multiple species of E2s, UBC9 is required for cell cycle progression in late G₂ or early M-phase (Seufert et al., 1995). UBC4 protein can ubiquitinate cyclins in Xenopus egg extracts (King et al., 1995). Recently, a novel cyclin- selective UBC family member, E2-C, was reported which can ubiquitinate cyclin B(13-91)/protein A fusion protein in cyclosome-dependent manner (Aristarkhov et al., 1996). These reports have shown that destruction box mutants cannot be ubiquitinated or degraded after extract activation, suggesting that the destruction box is a recognition sequence for the ubiquitinating system. However, we have shown that mutants that lack the proteasome cleavage site cannot be degraded after extract activation. The relative importance of these two processes is unclear because of the discrepancy between in vivo and in vitro results. In vitro, proteasome cleavage and ubiquitination of cyclin seem independent of each other. Destruction box-dependent ubiquitination of cyclin by purified proteins does not depend on the previous proteasome cleavage of cyclin B, and cyclin cleavage by purified proteasome does not depend on previous ubiquitination of cyclin B. But, since mutations that block in vitro ubiquitination and mutations that block proteasome cleavage also block destruction in extracts, the simple conclusion is that the ubiquitination and proteasome cleavage are both necessary for cyclin destruction in extracts. There are, however, no experimental data available at present that provide information on the relative order of these two steps in vivo. Therefore, there are three possible in vivo scenarios. (a) Proteasome cleavage precedes ubiquitination to expose an NH₂-terminal lysine that is a good substrate for ubiquitination. An extreme view would be that destruction box-dependent ubiquitination is an artifact that plays no role in vivo and that the purpose of the destruction box is solely to induce the initial ubiquitin-independent cleavage of cyclin. (b) Destruction box-dependent ubiquitination precedes proteasome cleavage to recruit cyclin to the proteasome by virtue of the proteasome's polyubiquitin binding subunit. An extreme view would be that proteasome cleavage is not necessary in vitro, and mutants like K57R are having a direct effect on destruction box-dependent ubiquitination. (c) There is no required order of proteasome cleavage and destruction box-dependent ubiquitination, although both events would be necessary for efficient cyclin destruction, they could occur in either order. Further studies are necessary to understand the molecular mechanism of cyclin degradation, especially identification of the lysine residue that is destined to be ubiquitinated.

We are grateful to Dr. M.J. Lohka for the B2Nt and to Drs. H.A. Bern and A.W. Murray for critical reading of the manuscript.

This study was supported in part by Grants-in-Aid for Scientific Research on Priority Areas from the Ministry of Education, Science, Sports, and Culture of Japan (07283104 to Y. Nagahama) and by a Grant-in-Aid (Bio Media Program) from the Ministry of Agriculture, Forestry, and Fisheries (BMP 96-II-2-6). This study was performed under the National Institute for Basic Biology Cooperative Research Program (95-112 to T. Tokumoto).