The COP9 Signalosome Promotes Degradation of Cyclin E during Early Drosophila Oogenesis

Developmental Cell, Vol. 4, 699–710, May, 2003, Copyright 2003 by Cell Press

The COP9 Signalosome Promotes Degradation of Cyclin E during Early Drosophila Oogenesis Sergey Doronkin, Inna Djagaeva, and Steven K. Beckendorf* Department of Molecular and Cell Biology University of California, Berkeley 401 Barker Hall Berkeley, California 94720

Summary The COP9 signalosome (CSN) is an eight-subunit complex that regulates multiple signaling and cell cycle pathways. Here we link the CSN to the degradation of Cyclin E, which promotes the G1-S transition in the cell cycle and then is rapidly degraded by the ubiquitinproteasome pathway. Using CSN4 and CSN5/Jab1 mutants, we show that the CSN acts during Drosophila oogenesis to remove Nedd8 from Cullin1, a subunit of the SCF ubiquitin ligase. Overexpression of Cyclin E causes similar defects as mutations in CSN or SCFAgo subunits: extra divisions or, in contrast, cell cycle arrest and polyploidy. Because the phenotypes are so similar and because CSN and Cyclin E mutations reciprocally suppress each other, Cyclin E appears to be the major target of the CSN during early oogenesis. Genetic interactions among CSN, SCF, and proteasome subunits further confirm CSN involvement in ubiquitin-mediated Cyclin E degradation. Introduction The programmed, rapid, and substrate-specific degradation of proteins by the ubiquitin-proteasome pathway plays an important role in the finely balanced regulation of protein activity in the cell (reviewed by Hershko and Ciechanover, 1998). Targeting of proteins for destruction is accomplished by covalent attachment, through an enzymatic cascade, of polymers of the small protein ubiquitin to the protein substrate. These polyubiquitin tags serve as a label for proteins that must be degraded after completion of their duties. Polyubiquitin-tagged proteins are then recognized and degraded by the 26S proteasome, a large protease complex. Ubiquitin-mediated degradation is a major feature of the cell cycle. Many cell cycle regulators, including both the positive cyclins and the negative cyclin-dependent kinase inhibitors (CKIs), are degraded at specific points in the cycle by ubiquitin addition and digestion by the proteasome. The specificity and timing of ubiquitin attachment to particular protein targets is the responsibility of E3 ubiquitin ligase complexes, acting together with E1 ubiquitin-activating enzymes and E2 ubiquitinconjugating enzymes (reviewed by Deshaies, 1999). During the cell cycle, the SCF E3 complex is required for the transition from G1 to S phase. The SCF includes three core subunits, Skp1, Cullin1 (Cdc53 in yeast), and Rbx1 (HRT1 in humans), which assemble with variable *Correspondence: [email protected]

F box proteins into distinct protein complexes. The F box proteins serve as substrate-specific binding proteins (Feldman et al., 1997; Skowyra et al., 1997). Genetic and biochemical studies have identified Cyclin E as an SCF substrate (Dealy et al., 1999; Moberg et al., 2001; Strohmaier et al., 2001; Koepp et al., 2001). Cyclin E binds to and activates the cyclin-dependent kinase Cdk2 and promotes the transition from G1 to S phase (Knoblich et al., 1994; Richardson et al., 1995). The timely appearance and disappearance of Cyclin E in the cell is tightly regulated by ubiquitin-dependent proteolysis (Clurman et al., 1996; Won and Reed, 1996). Failure in the oscillation of Cyclin E levels can lead to accelerated entry into S phase, genetic instability, and tumorigenesis (Spruck et al., 1999; Keyomarsi and Herliczek, 1997). Activity of the SCF complex is in turn regulated by the COP9 signalosome (CSN), a highly conserved, eightsubunit complex. The CSN promotes cleavage of the ubiquitin-like molecule Nedd8 from the cullin component of the SCF (deneddylation; reviewed by Schwechheimer and Deng, 2001; Bech-Otschir et al., 2002). Although the exact physiological role of the CSN remains unclear, the complex or its subunits have been implicated in diverse biological processes. The CSN is required for activation of the Jun transcription factor, stabilization of nuclear hormone receptors, and interactions with integrins. The CSN or its individual subunits regulate the mitotic cell cycle by ubiquitination, nuclear export, and degradation of the CDK inhibitor p27kip1 (Tomoda et al., 1999; Yang et al., 2002). CSN-associated kinase activity promotes degradation of p53, thereby allowing cell cycle progression (Bech-Otschir et al., 2001). In Drosophila, CSN5/CSN plays a critical role in oogenesis and eye development (Suh et al., 2002; Oron et al., 2002; Doronkin et al., 2002). Mutations in CSN5 lead to defects in embryonic axis formation due to activation of a meiotic checkpoint and subsequent inhibition of Gurken translation (Doronkin et al., 2002). We have begun to study the developmental roles of the CSN during Drosophila oogenesis (Doronkin et al., 2002). Oogenesis provides an excellent system for studying the regulation of the mitotic cell cycle within a cell lineage (reviewed by Deng and Lin, 2001). Oocytes are produced in the ovary by progression through a series of parallel tubes known as ovarioles. The most anterior part of each ovariole, called the germarium, is a specialized structure that maintains the proliferation and differentiation of germline and somatic stem cells, generating the functional units of oogenesis, the egg chambers. Oogenesis commences in region 1 at the anterior end of the germarium when a stem cell divides to produce a new stem cell and a distinct daughter cell called a cystoblast. This germline cystoblast divides asymmetrically four times to produce a 16-cell cyst. Because cytokinesis during these divisions is incomplete, the 16 resulting cystocytes remain interconnected by cytoplasmic bridges called ring canals. In region 2 of the germarium one cell is selected to become the

Developmental Cell 700

oocyte, while the other 15 cells differentiate into polyploid nurse cells. In each ovariole, the germline cysts are arranged in a linear order that corresponds to their developmental age. These features offer a valuable model for studying the role of cell divisions in cellular differentiation and patterning in a defined lineage. In this paper, we report that CSN5/CSN mediates SCFdependent degradation of Cyclin E. We found that mutations in CSN or SCF subunits or ectopic expression of Cyclin E disrupt ovarian development and cause abnormal cyst divisions and apoptosis. We demonstrate that different amounts of reduction in CSN and SCF activity can lead to opposite effects on the cell cycle. Our findings reveal a strong regulatory connection between the CSN, cell cycle control, and fly development. Results CSN5 Protein Can Be Detected in Dividing Germline Cells Previously, we reported that in Drosophila ovaries CSN5 mRNA accumulates in germline cells, starting early in the germarium (Doronkin et al., 2002). To begin investigating CSN5 function in early oogenesis, we immunostained ovaries from wild-type females with anti-CSN5/Jab1 antibodies. CSN5 appears to be expressed in all cells of the germarium, both germline and somatic, but at higher levels in the germline stem cells and in the mitotically active cysts in region 1 of the germarium (Figures 1A and 1B). Coincidence between the region of higher CSN5 expression and the region of permanent mitotic activity suggested that CSN5 might be required for progression through the cell cycle in ovarian development. We therefore wanted to test whether reduction of CSN5 can affect mitosis and provoke defects in oogenesis. CSN5 Is Required for Germline Development To analyze how mutations in CSN5 affect oogenesis, we used several CSN5 alleles and examined viable hypomorphic combinations or homozygous germline clones (see Experimental Procedures). Ovaries for all tested CSN5 alleles display similar multiple abnormalities that we have grouped into four categories. First, some egg chambers displayed features of apoptotic cell death: cell shrinkage, nuclear condensation and then fragmentation, and extensive membrane blebbing (Figure 1D). Positive staining using the TUNEL assay confirmed that loss of CSN5 could induce germ cell apoptosis. We found occasional apoptotic egg chambers at all stages of oogenesis (data not shown). In CSN5 mutants, the number of germline cells per egg chamber was variable, a result that might be caused either by an incorrect number of cell divisions or by fusion of neighboring egg chambers. In CSN5 mutants, we found both kinds of defective egg chambers. There were egg chambers with 32 germline cells and five ring canals associated with the oocyte (Figure 1E), as well as egg chambers with two, four, and eight germline cells and appropriately reduced numbers of ring canals (Figure 1F). These define the second phenotypic class, egg chambers with apparent defects in the regulation of mitosis.

There were also egg chambers resulting from fusion of adjacent chambers. In this third phenotypic class, sometimes there were multiple germline cysts in the same, aberrant egg chamber (Figure 1G). In addition, we occasionally found giant, tumorous-like ovarioles filled with large numbers of germline cysts that were not separated by follicle cells. Although we have yet to analyze this phenotypic class further, it suggests that communication between the germline and the somatic follicle cells is disrupted. The fourth mutant phenotype showed changes in the apparent ploidy of the nurse cells. In some CSN5 germline clones, nurse cell nuclei were much larger than normal and sometimes the nuclei within a particular chamber differed dramatically in size (Figure 1F). Although each of the CSN5 alleles caused all of these abnormal phenotypes, the frequency and severity varied widely. For the CSN5 null allele, there was an interesting, time-dependant change in the phenotype. Flies carrying CSN5N germline clones can lay a few eggs during a short period of time starting 5–6 days after heat shockinduced recombination. The first eggs were morphologically normal and sometimes even hatched. Eggs laid on the next day or two were very small and showed dorsal appendage defects. After that, no more eggs were laid. Consistent with this observation, flies carrying marked clones, dissected 5–6 days after heat shock, contained some late-stage homozygous egg chambers. In older females, fewer late, homozygous egg chambers could be found, and finally development of CSN5 null germline cells stopped completely. These observations suggest that wild-type CSN5 made in the germline stem cells before recombination can perdure for several days. There would be enough CSN5 to support the formation of a few normal eggs, but as stem cell divisions continued to dilute the remaining protein, it would become insufficient, and progressively more abnormal egg chambers would be produced. Consistent with this phenotypic progression, weak mutants or strong mutants shortly after heat shock showed a predominance of extra cystocyte divisions. At later times, strong mutants showed a reduced number of cystocytes and often polyploidy. The final phenotype was complete arrest of oogenesis at the beginning of germline development, probably before the very first mitotic division (Figures 2D–2F). This arrest is indeed CSN5 specific and can be prevented by expression of a CSN5 heat shock transgene. We also analyzed three different missense mutations that affect the JAMM/MPN⫹ metalloprotease domain of CSN5, the domain that is required for deneddylation of cullins (Cope et al., 2002; Maytal-Kivity et al., 2002). All three caused typical CSN5 mutant phenotypes, and the strongest, CSN5quo2, had a severe effect on oogenesis, similar to a CSN5 null mutation. CSN5 and CSN4 Mutations Cause Similar Phenotypes and Interact Genetically To test for CSN5 cooperation with other subunits of the CSN, we analyzed mitotic defects in CSN4 mutants. Flies carrying CSN4 mutant germline clones could also initially lay a few eggs, probably from the very first homozygous egg chambers. Similar to CSN5 null mutants,

CSN Promotes Cyclin E Degradation in Drosophila 701

Figure 1. CSN5 Is Required for Oogenesis (A and B) CSN5 protein expression and localization during early oogenesis. (A) In wild-type germaria, CSN5 protein accumulates strongly in region 1. (B) In CSN5 mutant germline clones (GLC), expression of CSN5 is reduced. In all figures, anterior is oriented to the left. (C) Wild-type ovaries. DNA is visualized with DAPI staining. (D–G) CSN5 mutant egg chambers display a variety of phenotypes: (D) apoptosis, (E) extra or (F) fewer nurse cells, and (G) fusion of adjacent egg chambers. (H–J) CSN5/CSN4 double heterozygotes demonstrated similar abnormalities: (H) apoptosis, (I) extra cell divisions, or (J) fused egg chambers. (K and L) CSN5 and ago interact genetically, resulting in (K) fewer or (L) extra cell divisions. (M) Cyclin E overexpression mimics the CSN5 loss-of-function phenotypes: fewer or extra mitotic divisions, fusion of egg chambers, and massive apoptosis. (N and O) CSN5 and CSN4 interact genetically with Rpn6, a subunit of the proteasome, and display similar defects including changes in the number of cell divisions, fusion of egg chambers, and apoptosis.

oogenesis in CSN4 mutant clones was progressively disrupted at earlier and earlier stages until it also arrested at the first mitotic division (Figures 2G–2I). CSN5 and CSN4 also showed a dominant genetic interaction. Whereas singly heterozygous ovaries were normal, ovaries from doubly heterozygous CSN4/⫹; CSN5/⫹ flies displayed phenotypes similar to homozygous CSN5 mutant clones (Figures 1H–1J). We found egg chambers with an incorrect number of germline cells, fused egg chambers, and cell death, which in some conditions could affect more than half of the egg chambers. Thus, it is likely that in oogenesis, CSN5 and CSN4 act together as subunits of the signalosome.

The Mitotic Defects in CSN5 Mutants Are Independent of Meiotic Checkpoint Control As we reported previously, CSN5 participates in oocyte axis determination through effects on the meiotic DNA repair checkpoint (Doronkin et al., 2002). Loss of CSN5 activates this checkpoint and leads to reduced translation of the EGFR ligand Gurken. Because Gurken signaling is required for both dorsal-ventral and anterior-posterior patterning, both primary axes of the oocyte are disrupted in CSN5 germline clones. These axis defects

are completely suppressed by mutations in either mei41, the Drosophila ATR homolog that governs the checkpoint, or mei-W68, the Drosophila spo11 homolog that is necessary for the formation of DNA double-strand breaks during recombination (Doronkin et al., 2002). It seemed possible that some of the mitotic or apoptotic defects in CSN5 mutants might be a consequence of meiotic checkpoint activation. To check this hypothesis, we analyzed several hypomorphic CSN5 mutants in a mei-W68 or mei-41 mutant background. However, there was no suppression of the germ cell number or cell death phenotypes by mutations in either mei-W68 or mei-41. We conclude that these defects are not a consequence of meiotic checkpoint activation.

CSN5 Mutations Affect Fusome Development To investigate cyst formation and differentiation in CSN5 germaria, we stained wild-type and CSN5 ovaries with anti-Hts antibody to highlight the fusome that connects all the cells of a cyst through the ring canals. Fusome development is essential for germline cyst formation (Lin et al., 1994; de Cuevas et al., 1996). In CSN5 mutant germaria the fusome was often less branched, and sometimes there were more individual fusomes than in

Developmental Cell 702

Figure 2. CSN5 and CSN4 Mutations and Cyclin E Overexpression Affect Cystocyte Division and Fusome Development (A–I) Immunostaining of wild-type and CSN5 and CSN4 mutant germaria with fusome-specific antibody to Hts. DNA is stained with DAPI. (A) In wild-type, spherical spectrosomes, precursors of the fusome, are located at the anterior tip of the germarium. During cystocyte divisions, the fusome becomes a branched structure. (B) For some CSN5 mutant germline clones, the germarium contains unbranched fusomes not only at the anterior, but also more posteriorly. (C) CSN5 germline clones sometimes have an increased number of cysts that contain fusomes. (D–I) CSN5 or CSN4 null clones visualized by the absence of GFP. For (D)–(D″), (E)–(E″), and (F)–(F″), the channels are partially separated to facilitate comparison of the clonal boundaries with the enlarged nuclei and spectrosomes. In these extreme phenotypes, cell division is arrested and the cells become extremely polyploid. These germline cells often contain huge, unbranched spectrosomes (D and E). (G) Retarded fusome development in CSN4 null clone (arrowhead). (J and K) Cyclin E overexpression can inhibit cell proliferation: (J) reduced number of cystocytes and enlarged cell nuclei (arrowhead) in region 1-2A demonstrate mild levels of inhibition of cell division; and (K) severe defects block cell division and cause strong polyploidization.

wild-type germaria. Furthermore, spherical spectrosomes (fusome precursors) were frequently found in more posterior regions of the germaria, probably indicating retarded fusome development (Figure 2B). As previously mentioned, CSN5 null mutant clones eventually cease mitotic divisions and often become enormously polyploid. Along with the increase in DNA, these cells often contain oversized spectrosomes or structures similar to a fragmented fusome, indicating dramatic changes in fusome development (Figures 2D and 2E). Some mutant clones lacked spectrosomes/ fusomes (Figure 2F). Usually, these clones were found a significant time after heat shock and were localized in ovarioles with no subsequent germline development. CSN4N mutant clones show similar undifferentiated cysts

with enlarged cell nuclei and defective fusome development (Figures 2G–2I). These data suggest that the intact CSN complex is required for proper cyst divisions and fusome development. The polyploid, nondividing germ cells may be the germline stem cells. We never see more than three of these large polyploid cells in a particular germarium, and they retain contact with somatic cells that probably correspond to the basal and terminal filament cells of normal germaria. Reduction of CSN5 Alters the Temporal Control of Mitosis To check for CSN5 involvement in S phase progression, we used BrdU incorporation to monitor DNA replication.

CSN Promotes Cyclin E Degradation in Drosophila 703

Figure 3. Cell Cycle Progression Is Affected in CSN5 Mutant Ovaries (A) In wild-type germaria, strong BrdU incorporation is limited to region 1. (B) CSN5 mutant germaria show strong incorporation of BrdU in an expanded region and in an increased number of cells. (C) Phospho-histone H3, a mitosis marker, is present in dividing cystoblasts in region 1 in wild-type germaria. (D) In CSN5 mutant germaria, the region of phospho-histone H3 accumulation is expanded and contains more cells. (E) Cyclin A is periodically expressed during mitotic cycles in region 1 in wild-type germaria. (F) In CSN5 mutant germaria, Cyclin A expression is sometimes prolonged. (G and H) Western blots revealed slightly reduced levels of Cyclin A and Cyclin B in CSN5 mutant ovaries.

In wild-type, strong BrdU incorporation was restricted to region 1 in the germarium (Figure 3A). A significantly weaker signal can be seen in later stages due to endoreplication in nurse cells. Under the same conditions, there were S phase cells in more posterior regions of hypomorphic CSN5 germaria, and sometimes a large number of cells were simultaneously in S phase (Figure 3B). The M phase marker anti-phospho-histone H3 is also normally restricted to region 1 of the germarium (Figure 3C). In hypomorphic CSN5 germaria, this marker was often seen more posteriorly in agreement with the changes in S phase (Figure 3D). These data indicate, in agreement with the fusome data, that mutations in CSN5 can delay mitotic progression leading to improper cyst development.

CSN5 Regulates Cyclin E Degradation Because CSN5 mutations affect cell cycle progression in the ovary, we tested whether the cell cycle regulators Cyclins A, B, and E were affected. There was little effect on Cyclins A or B other than a posterior expansion of expression as seen for other markers of mitotic progression (Figures 3E–3H). In contrast, Cyclin E patterning and accumulation were highly abnormal. In 75 wild-type

germaria, high levels of Cyclin E were found in the mitotically active region 1 (Figure 4A). In region 2, cells enter endoreplication, and Cyclin E expression was reduced to barely detectable levels. In 134 CSN5 mutant germaria, the region of strongest Cyclin E expression was often broader and included more cells, consistent with results for Cyclin A and cell cycle markers, indicating delayed mitosis. However, in contrast to Cyclins A and B, Cyclin E accumulated at higher than normal levels in CSN5 mutant germline cells in region 1 (Figures 4B and 4C). During endoreplication in wild-type nurse cells, Cyclin E levels oscillate asynchronously, so that only a few cells per cyst show high Cyclin E at the same time. In contrast, in CSN5 mutant clones, cells expressed Cyclin E at nearly uniform levels, indicating a disruption of Cyclin E oscillation (Figures 4E and 4F). Giant nuclei in CSN5 null mutant germaria almost always showed Cyclin E expression (Figure 4D). An anti-Cyclin E Western blot of whole, wild-type ovaries shows two clusters of bands corresponding to the two Drosophila isoforms Cyclin EI and Cyclin EII (Figure 4G; Richardson et al., 1993). A slowly migrating band was seen in ovaries containing CSN5L4032 germline clones, consistent with the accumulation of a modified form of Cyclin E. Phosphatase treatment increased the

Developmental Cell 704

Figure 4. Accumulation of Cyclin E Is Increased in CSN5 Mutants (A–F) Immunostaining for Cyclin E. (A) In wild-type, Cyclin E is expressed strongly in the mitotically active region 1 in the germarium. (B and C) Hypomorphic CSN5 mutant germaria contain an increased number of cells expressing Cyclin E at high levels and this region extends further posteriorly than in wild-type. (D and D″) CSN5 null mutant polyploid nuclei in the germarium accumulate Cyclin E. (E) In egg chambers at later stages, wild-type nurse cell nuclei express Cyclin E periodically. (F) In CSN5 mutant nurse cell nuclei, Cyclin E can be detected more frequently and in more cells per egg chamber, suggesting that cycling is affected. (G) Western blots of wild-type and CSN5 germline clone ovaries immunostained for Cyclin E. Heat shock-induced overexpression of Cyclin EI and Cyclin EII were used as mobility markers to distinguish the isoforms. CSN5 mutants accumulated a low mobility band. Alkaline phosphatase treatment showed that this band is a phosphorylated derivative of Cyclin EII.

mobility of this band to correspond to Cyclin EII, suggesting that a phosphorylated form of Cyclin EII accumulates in CSN5 mutant cells, although there is little or no effect on Cyclin EI. Accumulation of phosphorylated Cyclin E is the expected result if phosphorylationdependent ubiquitination by the SCF were blocked (Strohmaier et al., 2001; Koepp et al., 2001). Taken together, these observations indicate a requirement for CSN5 in the regulation of Cyclin E, possibly through phosphorylation-dependent degradation by the proteasome. To test more directly whether CSN5 regulates Cyclin E degradation, we used heat shock-driven constructs

to transiently overexpress Cyclin EII with or without concurrent CSN5 overexpression. Degradation of Cyclin EII was then monitored with time after heat shock. In CSN5⫹ ovaries, the overexpressed Cyclin EII was strongly reduced by 2 hr and absent by 4 hr after heat shock (Figure 5A). In CSN5 hypomorphic mutants, the overexpressed Cyclin EII was more stable, with substantial amounts still present at 2 hr and some remaining at 4 hr (Figure 5B). Note that the level of CSN5 protein in these ovaries was lower than in CSN5⫹ but remained easily detectable. Unfortunately, analysis of stronger CSN5 mutants, which might have given a stronger effect, was impossible due to their lethality. Most strikingly, we found that

CSN Promotes Cyclin E Degradation in Drosophila 705

This conclusion is strengthened by genetic interactions between CSN5 and CycE loss-of-function mutations. Hypomorphic, viable CycE mutant flies have rough eyes and have ovarian defects that include a reduced number of cystocyte divisions (Secombe et al., 1998; Lilly and Spradling, 1996; see Table 1). Mutations in CSN5 dominantly suppress both ovarian and eye defects for these CycE mutations (Table 1 and data not shown). Conversely, CycE mutations dominantly suppress the CSN5 mutant ovarian phenotype and even increase the viability of CSN5 mutants. These results emphasize the specificity of the CSN regulation of Cyclin E degradation. The many other potential CSN targets appear to have little effect on the cystocyte divisions.

Figure 5. Alterations in CSN5 Expression Regulate the Stability of Cyclin E (A) In CSN5⫹ ovaries, overexpressed Cyclin E is rapidly degraded. (B) With reduced expression of CSN5, overexpressed Cyclin E is more slowly degraded. (C) Simultaneous overexpression of Cyclin E and CSN5 dramatically increases Cyclin E degradation. Our preliminary results suggest that the reduction in Cyclin E mobility as it disappears is at least in part due to ubiquitin conjugation (data not shown).

simultaneous overexpression of CSN5 and Cyclin EII significantly increased Cyclin EII degradation (Figure 5C). Now the overexpressed Cyclin EII was strongly reduced by 1.5 hr and gone by 3 hr. Transient overexpression of Cyclin EI along with CSN5 showed that Cyclin EI degradation is also CSN5 dependent (data not shown). We conclude that CSN5 stimulates Cyclin E degradation. We also overexpressed Cyclin E in wild-type flies to determine its effects on oogenesis when the CSN was not mutant. Flies carrying either a heat shocked CycEI or CycEII construct were kept at 29⬚C to induce permanent overexpression of Cyclin E at moderate levels. Interestingly, flies for both lines showed a range of oogenesis defects that was very similar to the range of CSN5 mutant defects. Most Cyclin E overexpressing ovarioles contained defective egg chambers with abnormal cystocyte divisions, abnormal polyploidy, apoptosis, or several of these defects (Figure 1M). Importantly, we found that in some cases, overexpression of Cyclin E can inhibit and finally stop cyst division and cause enormous polyploidization (Figures 2J and 2K). The similarities between these phenotypes and those caused by loss of CSN5 or CSN4 suggest that Cyclin E is the major target of the CSN in early oogenesis.

Mutations in CSN5 and CSN4 Lead to Accumulation of Neddylated Cullin1 Degradation of phosphorylated Cyclin E can be mediated by the SCF ubiquitin ligase complex (Dealy et al., 1999; Moberg et al., 2001; Strohmaier et al., 2001; Koepp et al., 2001). Because phosphorylated Cyclin E accumulates in CSN5 mutants, we wanted to test whether CSN5 regulates Cyclin E degradation by acting through the SCF. CSN function has been linked to the SCF by its ability to remove the ubiquitin-like protein Nedd8 from the SCF subunit Cullin1 (Lyapina et al., 2001). To see whether this deneddylation activity is required in vivo, we analyzed the effect of CSN5 and CSN4 mutations on Cullin1 and Nedd8 proteins. Western blots of protein from wild-type, CSN5, or CSN4 larvae revealed an elevated level of Cullin1 protein with slower mobility in the mutants (Figure 6A). The reduced mobility is probably caused by covalent attachment of Nedd8 to Cullin1 because reprobing the blot showed that Nedd8 comigrated with the slower form of Cullin1 and accumulated in CSN5 or CSN4 homozygous mutants (Figure 6B). Note that accumulation of neddylated Cullin1 in CSN5 null mutants is stronger than in hypomorphic CSN5 mutants. Larvae homozygous for CSN5quo1, a missense mutation in the JAMM metalloprotease domain of CSN5, also accumulated neddylated Cullin1 (data not shown). These results demonstrate in vivo that deneddylation of Cullin1, and by inference the activity of the SCF ubiquitin ligase, require a functional CSN complex. Immunostaining of ovaries from wild-type and CSN5 germline clones revealed a remarkable difference in the subcellular localization of Cullin1. Although Cullin1 is found mostly in the nucleus in wild-type ovarioles, it accumulates in the cytoplasm in CSN5 mutants (Figures 6C and 6D). This unusual cytoplasmic localization may correspond to the accumulation of neddylated Cullin1 seen in the Western blots of CSN5 mutants. Thus, CSN5 may regulate the subcellular localization of Cullin1 and possibly the entire SCF complex. CSN and SCF Act in the Same Pathway Strong mutants for cullin1 (lin19) and Nedd8 cause larval lethality (Ou et al., 2002). To evaluate further the role of these genes in oogenesis, we generated Nedd8 and cullin1 germline clones. These clones showed variable, time-dependant ovarian defects similar to the CSN5 or CSN4 mutant phenotype. However, even at early times

Developmental Cell 706

Table 1. Mutations in CSN5 and Cyclin E Reciprocally Suppress Each Other Germline Cells per Egg Chamber (Percentage) Genotype A

B C D E

JP

CycE CycEJP; CSN5quo2/⫹ CycEJP; CSN5quo3/⫹ CycEJP/CycEAR95 CycEJP/CycEAR95; CSN5L4032/⫹ CycEJP/CycEAR95 CycEJP/CycEAR95; CSN5L4032/⫹ CSN5L4032/CSN5ex21 CycEAR95/⫹; CSN5L4032/CSN5ex21 CSN5L4032/CSN5ex13 CycE05206/⫹; CSN5L4032/CSN5ex13

o

25 C 25oC 25oC 18oC 18oC 29oC 29oC 18oC 18oC 18oC 18oC

4 Cells

8 Cells

16 Cells

32 Cells

0 0 0 0 0 0.5 0 0 0 0 0

6.8 0 0.4 37.7 24 85 76.9 0 2.1 0 0.6

93.2 100 99.4 62.3 76 14.5 23.1 95.1 97.3 89.9 96

0 0 0.2 0 0 0 0 6.3 0.6 10.1 3.4

(A–C) CSN5 mutations dominantly suppress the eight-cell phenotype of hypomorphic CycE mutations. (D and E) Reduction of Cyclin E suppresses the extra cell division caused by CSN5 mutations. More than 500 egg chambers were analyzed and counted for each mutant genotype.

after heat shock, we never saw any eggs derived from flies carrying cullin1 or Nedd8 mutant germline clones, suggesting that Cullin1 and Nedd8 proteins may not persist long after the generation of homozygous clones. Similar to CSN5 and CSN4 null mutant clones, both Nedd8 and cullin1 mutant germline clones may be arrested at early stages of oogenesis, sometimes producing highly polyploid, one-cell cysts (data not shown). Furthermore, we found strong dominant genetic inter-

actions between cullin1 and either CSN5 or CSN4. Ovaries from doubly heterozygous females display various defects: cell death, abnormal number of nurse cells per egg chamber, and loss of stalk cells (data not shown). We did not detect a dominant genetic interaction between Nedd8 and either CSN5 or CSN4. This fact is not surprising because in CSN5 mutants, we found an accumulation of neddylated Cullin1, and a modest reduction in Nedd8 expression in double heterozygotes

Figure 6. CSN5 and CSN4 Are Required for Deneddylation of Cullin1 in Drosophila (A and B) Western blot analysis of wild-type, cullin1, CSN5, and CSN4 mutants with antibodies to Cullin1 and Nedd8. Cell extracts were prepared from wild-type, cullin1(lin19) homozygous larvae, and CSN5 and CSN4 mutant homozygous larvae that were marked by the absence of GFP expression. (A) Neddylated Cullin1 accumulates in CSN5 and CSN4 mutants. Cullin1 is strongly reduced in cullin1 mutants. (B) Nedd8 bands coincide with the neddylated Cullin1 bands in (A). (C and D) CSN5 alters cellular localization and accumulation of Cullin1 protein. (C and C⬘) In wild-type germaria, Cullin1 (green) is mostly nuclear and coincides with DAPI (blue) staining (arrows). A cluster of mitotically active germline cells in region 1 accumulates Cullin1 at elevated levels. (D and D⬘) In CSN5 germline clones, there are reduced levels of Cullin1 in region 1. More posteriorly, Cullin1 accumulates largely in the cytoplasm (arrows).

CSN Promotes Cyclin E Degradation in Drosophila 707

Figure 7. ago Mutants Have Elevated Cyclin E Levels, Cell Cycle Defects, and Enlarged Heterochromatic Regions in Their Nuclei (A–A″⬘ and B–B″⬘) ago mutant clones marked by the loss of GFP preferentially accumulate Cyclin E. (C–C″) Some ago mutant clones display a lack of mitotic divisions and enlarged nuclei. (D) Wild-type nurse cell nuclei show regions of compact heterochromatin (arrowheads). (E) In ago mutant clones, regions of compact heterochromatin are greatly enlarged.

would be expected to suppress the CSN5 mutant phenotype. F Box Protein Archipelago Participates in Cyclin E Degradation in Oogenesis The Drosophila F box protein Archipelago (Ago) was proposed to target Cyclin E for ubiquitin-mediated degradation in imaginal discs (Moberg et al., 2001). We used the hypomorphic alleles ago1, ago3, and ago4 to test for a similar role in Cyclin E degradation during oogenesis. Immunostaining showed that ago mutant clones marked by lack of GFP persistently accumulate Cyclin E at high levels (Figures 7A and 7B). With one addition, these clones showed a similar range of phenotypes as those seen in CSN5 or CSN4 mutants or after overexpression of Cyclin E. Some mutant cysts had extra nurse cells and some had fewer than normal, and many were degenerating. Some cysts had been arrested after the stem cell division and some of these single-cell cysts were polyploid (Figure 7C). Prominent in the ago clones was a phenotype we had not previously noticed. Cyclin E

accumulation in ago clones correlated with significantly DAPI-bright regions in nurse cell nuclei (Figures 7D and 7E). Because these regions are likely to include heterochromatic sequences that are usually underreplicated during endoreplication (Lilly and Spradling, 1996), their enlargement may indicate a more complete replication of both heterochromatic and euchromatic sequences in ago clones. Although we do occasionally see enlarged heterochromatin-rich regions in CSN5 mutants and after Cyclin E overexpression, this phenotype is stronger in ago mutants, possibly suggesting a more specific role for ago in regulation of late replication. In addition to the similar phenotypes between ago mutants and CSN, Nedd8, or cullin1 mutants, we found dominant interactions between ago and CSN mutants. CSN5/ago and CSN4/ago double heterozygotes showed familiar ovarian defects: extra cystocyte divisions, fewer divisions but higher ploidy, and apoptotic egg chambers (Figures 1K and 1L). These defects were very similar to the CSN5 mutant phenotype and to defects in oogenesis induced by Cyclin E overexpression. In addition, these

Developmental Cell 708

double heterozygotes had enlarged heterochromatic regions in nurse cell nuclei, suggesting mutual CSN-ago control of late replication. CSN5 and CSN4 Interact Genetically with Rpn6, a Proteasome Mutation The regulatory lid of the proteasome is an eight-subunit complex that is closely related to the CSN. It appears to be necessary for the removal of ubiquitin side chains from the target protein as it is fed into the barrel of the proteasome for proteolysis (Verma et al., 2002). We tested a mutation in the RPN6 subunit of the regulatory lid for genetic interactions in oogenesis with CSN4 and CSN5 mutations. Both double heterozygotes showed a strong interaction and the full range of CSN5-like ovarian defects, including apoptosis, incorrect number of mitotic divisions, and fusions of neighboring egg chambers (Figures 1N and 1O). Discussion The CSN Regulates Cyclin E Degradation Our results show that the CSN regulates the turnover of Cyclin E. Decreased expression of CSN5 increased Cyclin E stability, and increased CSN5 expression stimulated rapid Cyclin E degradation. CSN5 mutations and Cyclin E overexpression produce remarkably similar oogenesis defects, suggesting that the major role of CSN5 during mitosis in early oogenesis is to regulate Cyclin E levels. This finding was further confirmed by the reciprocal suppression between CSN5 and CycE mutations. This level of specificity is remarkable given the wide variety of signaling, transcriptional, and cell cycle regulatory roles reported for the CSN (Schwechheimer and Deng, 2001). Although its role in Drosophila development is just beginning to be understood, the CSN clearly has multiple roles here, too. In addition to Cyclin E turnover, Drosophila CSN is implicated in DNA repair, both at the level of meiotic recombination in the germarium and during larval development (Doronkin et al., 2002; Oron et al., 2002). There are also indirect suggestions that the CSN or some of its individual subunits are required for ecdysone responses or for blood cell development (Oron et al., 2002). In addition, demonstrated roles for Drosophila Nedd8 or SCF subunits in degradation of ␤-catenin/Armadillo and Cubitus interruptus (Jiang and Struhl, 1998; Ou et al., 2002) may foreshadow CSN involvement in those processes. Although these multiple roles might suggest pleiotropic rather than specific phenotypes, the CSN may have a limited number of targets in any particular tissue at a particular time in development. The CSN Promotes SCF Activity The effect of the CSN on the activity of the SCF complex has been controversial. Although Nedd8 modification of Cullin1 stimulates SCF activity (Read et al., 2000; Morimoto et al., 2000), the opposite process, deneddylation, was also shown to be important for SCF function and cell cycle progression (Lyapina et al., 2001; Schwechheimer et al., 2001; Cope et al., 2002). For example, point mutations in the JAMM domain of the S.

cerevisiae CSN5 homolog Rri abolish its deneddylation activity and enhance the growth defect shown by ts alleles of SCF genes (Cope et al., 2002). These results led to the proposal that repeated cycles of neddylation and deneddylation are required for the sustained activity of the SCF (Lyapina et al., 2001; Schwechheimer et al., 2001; Cope et al., 2002). However, a recent gain-offunction analysis suggested that deneddylation by the CSN inhibits degradation of the SCF target p27kip1 (Yang et al., 2002). Our results strongly support the idea that deneddylation of cullin1 by the CSN is necessary for activity of the SCF complex. CSN mutations have the same, not opposite, effects on oogenesis as do Nedd8, cullin1, or ago mutations. CSN5 and CSN4 mutations also interact dominantly with cullin1 and ago mutations, further suggesting that the CSN works along with the SCF to promote Cyclin E degradation. These requirements for the CSN appear to demand its deneddylase activity, as the CSN5quo2 mutation, with a single amino acid substitution in the metalloprotease domain, behaves similarly to a CSN5 null. Cycles of neddylation and deneddylation might control the association of an F box protein with an E3 ubiquitin ligase core complex or the association of a ubiquitinloaded E2-conjugating enzyme with the E3 complex (Schwechheimer and Deng, 2001). Neddylation might also affect Cullin1 stability as suggested by the Cullin1 accumulation that we see in CSN5 mutants and its reduction in Nedd8 mutants (Ou et al., 2002). CSN Affects Cullin1 Intracellular Localization Conjugation of Nedd8 to cullins may regulate not only their activity, but also their subcellular distribution. Shuttling between the nucleus and cytoplasm has been proposed as a regulatory mechanism for E3 ubiquitin ligases when the target protein is ubiquitinated in the nucleus (Lee et al., 1999). Our results showing that in CSN mutants, Nedd8-modified Cullin1 accumulates in the cytoplasm suggest that neddylation may be one way to regulate shuttling. Neddylation might favor nuclear export of Cullin1, and nuclear CSN would be required to remove Nedd8 and prevent export. Alternatively, neddylation might prevent Cullin1 nuclear import, and recycling of SCF into the nucleus would require cytoplasmic CSN. On either model, the CSN would be an important regulator of SCF activity. For example, modulation of SCF nuclear shuttling might affect the timing of Cyclin E degradation and entry into S phase of the cell cycle. Disruption of CSN-Dependent Cyclin E Turnover Can Either Activate or Block the Cell Cycle One of the important results of the current work is the demonstration that the CSN regulates the cell cycle in ovaries primarily through the turnover of Cyclin E. The apparent perdurance and gradual dilution of wild-type CSN5 protein in genetically null germline clones showed that reduced Cyclin E degradation affected both cell division and DNA replication. Slight reductions caused an extra division of the cystocytes. In contrast, continuous or strong accumulation of Cyclin E in null CSN5 mutants was able to reduce or stop cell divisions though often allowing endoreplication to continue. This switch

CSN Promotes Cyclin E Degradation in Drosophila 709

from overproliferation to inhibition of cell divisions was sometimes visible in a single CSN5 mutant ovariole as the wild-type CSN5 protein was diluted by stem cell divisions. These observations support the view that different Cyclin E levels can lead to distinct and sometimes opposite effects. Mutations in Drosophila ago, the C. elegans gene cul1, or the F box-encoding lin23 have been shown to cause increased cell proliferation, suggesting a critical role for SCF in regulating cell divisions (Kipreos et al., 1996, 2000; Moberg et al., 2001). We also found extra cell division to be a frequent phenotype produced by mutations in CSN5, CSN4, cullin1, ago, or by overexpression of Cyclin E. However, SCF and CSN mutations have also been shown to cause the opposite effect on the cell cycle. In null mutant clones of cullin1 or Nedd8, cell proliferation in Drosophila eye disks is arrested (Ou et al., 2002). Similarly, we found that loss of CSN5, CSN4, Cullin1, or ago inhibits and finally stops cell proliferation and often leads to enlarged nuclei. The abundance of Cyclin E and giant polyploid nuclei were also present in mice that were mutant for cul1 (Dealy et al., 1999; Wang et al., 1999). Elevated levels of Cyclin E that may give cells a proliferative advantage are found in many human tumors (Keyomarsi and Herliczek, 1997). In many of these tumors the Cyclin E gene itself is amplified. However, among breast and ovarian cancer cell lines that overexpress Cyclin E protein without amplification, several lines have mutations in hCDC4, the human homolog of archipelago, suggesting that SCF[hcdc4] acts to suppress tumor formation (Strohmaier et al., 2001; Moberg et al., 2001). Our results suggest that the CSN might have a similar effect. In summary, these genetic and functional relationships between the CSN, the SCF, and the proteasome link these complexes in the regulation of Cyclin E degradation during normal development. When either the CSN or SCF are disrupted, the periodic degradation of Cyclin E is prevented, and cell cycle deregulation ensues. Experimental Procedures Fly Strains Canton S and w1118 were used as standard strains. CSN5L4032 is a hypomorphic allele resulting from a P element insertion 24 bp upstream of the cDNA (Freilich et al., 1999). CSN5quo1, CSN5quo2 [l(3)89Cf M255], CSN5quo3 [l(3)89Cf M203], CSN5N, CSN4N, and hs-CSN5 were gifts from G. Suh. In contrast to published results (Oron et al., 2002), we found that CSN5quo1 and CSN5quo3 do not cause early oogenesis arrests. ago1, ago3, ago4, and CycEJP were obtained from I. Hariharan, and flies carrying hsCycEI and hsCycEII constructs were a gift from H. Richardson. Other stocks were obtained from the Bloomington Drosophila Stock Center. Genetics CSN5 homozygous mutant germline clones were produced by using the dominant-female-sterile, FLP/FRT technique (Chou and Perrimon, 1992). Females of genotype w; CSN5* P{neoFRT}82B/TM3B, Sb were mated with males of genotype w hsFLP; ovoD1 FRT82B/ TM3B, Sb to generate germline clones or with w hsFLP; P{neoFRT}82B P{Ubi-GFP(S65T)nls}3R/TM3B, Sb to generate GFPmarked clones. Their progeny were heat shocked as third instar larvae or early pupae for 2 hr at 37⬚C for 2 consecutive days to induce FLP expression. For rescue of the CSN5 mutant defect, females of genotype w hsCSN5; CSN5N P{neoFRT}82B/TM3B, Sb were mated with males of genotype w hsFLP; ovoD1 FRT82B/TM3B,

Sb. Clones mutant for other genes were generated in a similar manner. Flies with genotype w hsFLP; P{neoFRT}42D CSN4N or k08018/ CyO or w hsFLP; P{neoFRT}42D lin19k01207/CyO were mated with w; P{neoFRT}42D P{Ubi-GFP(S65T)nls}2R/CyO. Flies with genotype w hsFLP; P{FRT(w[hs])}G13 CSN4k08018/CyO were mated with P{FRT (w[hs])}G13 P{ovoD1-18}2R/CyO. Flies with genotype w; ago* P{neoFRT} 80B/TM6B were mated with w hsFLP; P{Ubi-GFP(S65T)nls}3L P{neoFRT}80B/TM3. For pulse-chase experiments, flies carrying an hsCycE construct were transferred to 38⬚C for 30 min and then dissected at 30 min intervals. w1118 flies were used as a control in genetic interaction crosses. Crosses were usually maintained at three different temperatures: 18⬚C, 25⬚C, and 29⬚C. For most crosses, genetic interactions were most obvious at either 18⬚C or 29⬚C. Flies carrying a GFP balancer were used for selection of homozygous mutant larvae. Immunohistochemistry and Western Blots For immunostaining and Western blots the following antibodies were used: mouse monoclonal anti-CSN5/Jab1 (GeneTex), rabbit polyclonal anti-Cullin1 (Zymed), rabbit polyclonal anti-phospho-histone H3 (Upstate Biotechnology), and mouse monoclonal anti-Cyclin E (a gift from H. Richardson). DNA replication was assayed by incorporation of BrdU and detected with mouse monoclonal anti-BrdU (Becton Dickinson). Acknowledgments We thank Michael Levine for critically reading the manuscript. We greatly appreciate the generous gifts of reagents and fly stocks by Cheng-Ting Chien, Daniel Chamovitz, Iswar Hariharan, Christian Lehner, Patrick O’Farrell, Helena Richardson, Daniel Segal, and Greg Suh. We are also grateful to the Berkeley Drosophila Genome Project and the Bloomington Drosophila Stock Center. Received: December 3, 2002 Revised: February 26, 2003 Accepted: February 28, 2003 Published: May 5, 2003 References Bech-Otschir, D., Kraft, R., Huang, X., Henklein, P., Kapelari, B., Pollmann, C., and Dubiel, W. (2001). COP9 signalosome-specific phosphorylation targets p53 to degradation by the ubiquitin system. EMBO J. 20, 1630–1639. Bech-Otschir, D., Seeger, M., and Dubiel, W. (2002). The COP9 signalosome: at the interface between signal transduction and ubiquitin-dependent proteolysis. J. Cell Sci. 115, 467–473. Chou, T.B., and Perrimon, N. (1992). Use of a yeast site-specific recombinase to produce female germline chimeras in Drosophila. Genetics 131, 643–653. Cope, G.A., Suh, G.S., Aravind, L., Schwarz, S.E., Zipursky, S.L., Koonin, E.V., and Deshaies, R.J. (2002). Role of predicted metalloprotease motif of Jab1/Csn5 in cleavage of Nedd8 from Cul1. Science 298, 608–611. Clurman, B.E., Sheaff, R.J., Thress, K., Groudine, M., Roberts, J.M. (1996). Turnover of cyclin E by the ubiquitin-proteasome pathway is regulated by cdk2 binding and cyclin phosphorylation. Genes Dev. 10, 1979–1990. Dealy, M.J., Nguyen, K.V., Lo, J., Gstaiger, M., Krek, W., Elson, D., Arbeit, J., Kipreos, E.T., and Johnson, R.S. (1999). Loss of Cul1 results in early embryonic lethality and dysregulation of cyclin E. Nat. Genet. 23, 245–248. de Cuevas, M., Lee, J.K., and Spradling, A.C. (1996). ␣-spectrin is required for germline cell division and differentiation in the Drosophila ovary. Development 122, 3959–3968. Deng, W., and Lin, H. (2001). Asymmetric germ cell division and oocyte determination during Drosophila oogenesis. Int. Rev. Cytol. 203, 93–138.

Developmental Cell 710

Deshaies, R.J. (1999). SCF and Cullin/Ring H2-based ubiquitin ligases. Annu. Rev. Cell Dev. Biol. 15, 435–467.

SCF(␤(TrCP))-dependent ubiquitination of I␬B␣. Mol. Cell. Biol. 20, 2326–2333.

Doronkin, S., Djagaeva, I., and Beckendorf, S.K. (2002). CSN5/Jab1 mutations affect axis formation in the Drosophila oocyte by activating a meiotic checkpoint. Development 129, 5053–5064.

Richardson, H.E., O’Keefe, L.V., Reed, S.I., and Saint, R. (1993). A Drosophila G1-specific cyclin E homolog exhibits different modes of expression during embryogenesis. Development 119, 673–690.

Feldman, R.M., Correll, C.C., Kaplan, K.B., and Deshaies, R.J. (1997). A complex of Cdc4p, Skp1p, and Cdc53p/cullin catalyzes ubiquitination of the phosphorylated CDK inhibitor Sic1p. Cell 91, 221–230.

Richardson, H., O’Keefe, L.V., Marty, T., and Saint, R. (1995). Ectopic cyclin E expression induces premature entry into S phase and disrupts pattern formation in the Drosophila eye imaginal disc. Development 121, 3371–3379.

Freilich, S., Oron, E., Kapp, Y., Nevo-Caspi, Y., Orgad, S., Segal, D., and Chamovitz, D.A. (1999). The COP9 signalosome is essential for development of Drosophila melanogaster. Curr. Biol. 9, 1187–1190. Hershko, A., and Ciechanover, A. (1998). The ubiquitin system. Annu. Rev. Biochem. 67, 425–479. Jiang, J., and Struhl, G. (1998). Regulation of the Hedgehog and Wingless signalling pathways by the F-box/WD40-repeat protein Slimb. Nature 391, 493–496. Keyomarsi, K., and Herliczek, T.W. (1997). The role of cyclin E in cell proliferation, development and cancer. Prog. Cell Cycle Res. 3, 171–191. Kipreos, E.T., Lander, L.E., Wing, J.P., He, W.W., and Hedgecock, E.M. (1996). cul-1 is required for cell cycle exit in C. elegans and identifies a novel gene family. Cell 85, 829–839. Kipreos, E.T., Gohel, S.P., and Hedgecock, E.M. (2000). The C. elegans F-box/WD-repeat protein LIN-23 functions to limit cell division during development. Development 127, 5071–5082. Knoblich, J.A., Sauer, K., Jones, L., Richardson, H., Saint, R., and Lehner, C.F. (1994). Cyclin E controls S phase progression and its down-regulation during Drosophila embryogenesis is required for the arrest of cell proliferation. Cell 77, 107–120. Koepp, D.M., Schaefer, L.K., Ye, X., Keyomarsi, K., Chu, C., Harper, J.W., and Elledge, S.J. (2001). Phosphorylation-dependent ubiquitination of cyclin E by the SCFFbw7 ubiquitin ligase. Science 294, 173–177. Lee, S., Neumann, M., Stearman, R., Stauber, R., Pause, A., Pavlakis, G.N., and Klausner, R.D. (1999). Transcription-dependent nuclearcytoplasmic trafficking is required for the function of the von HippelLindau tumor suppressor protein. Mol. Cell. Biol. 19, 1486–1497. Lilly, M.A., and Spradling, A.C. (1996). The Drosophila endocycle is controlled by Cyclin E and lacks a checkpoint ensuring S-phase completion. Genes Dev. 10, 2514–2526. Lin, H., Yue, L., and Spradling, A.C. (1994). The Drosophila fusome, a germline-specific organelle, contains membrane skeletal proteins and functions in cyst formation. Development 120, 947–956. Lyapina, S., Cope, G., Shevchenko, A., Serino, G., Tsuge, T., Zhou, C., Wolf, D.A., Wei, N., Shevchenko, A., and Deshaies, R.J. (2001). Promotion of NEDD-CUL1 conjugate cleavage by COP9 signalosome. Science 292, 1382–1385. Maytal-Kivity, V., Reis, N., Hofmann, K., and Glickman, M.H. (2002). MPN⫹, a putative catalytic motif found in a subset of MPN domain proteins from eukaryotes and prokaryotes, is critical for Rpn11 function. BMC Biochem. 3, 28. Moberg, K.H., Bell, D.W., Wahrer, D.C., Haber, D.A., and Hariharan, I.K. (2001). Archipelago regulates Cyclin E levels in Drosophila and is mutated in human cancer cell lines. Nature 413, 311–316. Morimoto, M., Nishida, T., Honda, R., and Yasuda, H. (2000). Modification of cullin-1 by ubiquitin-like protein Nedd8 enhances the activity of SCF(skp2) toward p27(kip1). Biochem. Biophys. Res. Commun. 270, 1093–1096. Oron, E., Mannervik, M., Rencus, S., Harari-Steinberg, O., NeumanSilberberg, S., Segal, D., and Chamovitz, D.A. (2002). COP9 signalosome subunits 4 and 5 regulate multiple pleiotropic pathways in Drosophila melanogaster. Development 129, 4399–4409. Ou, C.Y., Lin, Y.F., Chen, Y.J., and Chien, C.T. (2002). Distinct protein degradation mechanisms mediated by Cul1 and Cul3 controlling Ci stability in Drosophila eye development. Genes Dev. 16, 2403–2414. Read, M.A., Brownell, J.E., Gladysheva, T.B., Hottelet, M., Parent, L.A., Coggins, M.B., Pierce, J.W., Podust, V.N., Luo, R.S., Chau, V., and Palombella, V.J. (2000). Nedd8 modification of cul-1 activates

Schwechheimer, C., and Deng, X.W. (2001). COP9 signalosome revisited: a novel mediator of protein degradation. Trends Cell Biol. 11, 420–426. Schwechheimer, C., Serino, G., Callis, J., Crosby, W.L., Lyapina, S., Deshaies, R.J., Gray, W.M., Estelle, M., and Deng, X.W. (2001). Interactions of the COP9 signalosome with the E3 ubiquitin ligase SCFTIRI in mediating auxin response. Science 292, 1379–1382. Secombe, J., Pispa, J., Saint, R., and Richardson, H. (1998). Analysis of a Drosophila cyclin E hypomorphic mutation suggests a novel role for cyclin E in cell proliferation control during eye imaginal disc development. Genetics 149, 1867–1882. Skowyra, D., Craig, K.L., Tyers, M., Elledge, S.J., and Harper, J.W. (1997). F-box proteins are receptors that recruit phosphorylated substrates to the SCF ubiquitin-ligase complex. Cell 91, 209–219. Spruck, C.H., Won, K.A., and Reed, S.I. (1999). Deregulated cyclin E induces chromosome instability. Nature 401, 297–300. Strohmaier, H., Spruck, C.H., Kaiser, P., Won, K.A., Sangfelt, O., and Reed, S.I. (2001). Human F-box protein hCdc4 targets cyclin E for proteolysis and is mutated in a breast cancer cell line. Nature 413, 316–322. Suh, G.S., Poeck, B., Chouard, T., Oron, E., Segal, D., Chamovitz, D.A., and Zipursky, S.L. (2002). Drosophila JAB1/CSN5 acts in photoreceptor cells to induce glial cells. Neuron 33, 35–46. Tomoda, K., Kubota, Y., and Kato, J. (1999). Degradation of the cyclin-dependent-kinase inhibitor p27Kip1 is instigated by Jab1. Nature 398, 160–165. Verma, R., Aravind, L., Oania, R., McDonald, W.H., Yates, J.R., III, Koonin, E.V., and Deshaies, R.J. (2002). Role of Rpn11 metalloprotease in deubiquitination and degradation by the 26S proteasome. Science 298, 611–615. Wang, Y., Penfold, S., Tang, X., Hattori, N., Riley, P., Harper, J.W., Cross, J.C., and Tyers, M. (1999). Deletion of the Cul1 gene in mice causes arrest in early embryogenesis and accumulation of cyclin E. Curr. Biol. 9, 1191–1194. Won, K.A., and Reed, S.I. (1996). Activation of cyclin E/CDK2 is coupled to site-specific autophosphorylation and ubiquitin-dependent degradation of cyclin E. EMBO J. 15, 4182–4193. Yang, X, Menon, S., Lykke-Andersen, K., Tsuge, T., Xiao, D., Wang, X., Rodriguez-Suarez, R.J., Zhang, H., and Wei, N. (2002). The COP9 signalosome inhibits p27(kip1) degradation and impedes G1-S phase progression via deneddylation of SCF Cul1. Curr. Biol. 12, 667–672.