Drosophila Cks30A interacts with Cdk1 to target Cyclin A for destruction in the female germline

Passage through the cell cycle must be precisely regulated during development. A number of conserved cell cycle regulators have been shown to act at important developmental transitions in various organisms, including Drosophila (for a review, see Lee and Orr-Weaver, 2003). Cks, or Suc1, is a small cell cycle regulator that was first identified by its ability to interact genetically and physically with the Cyclin-dependent kinases (Cdks) in yeasts(Hadwiger et al., 1989; Hayles et al., 1986). Members of this family were subsequently found in many multicellular organisms,including Xenopus, Caenorhabditis elegans and humans (reviewed by Bartek and Lukas, 2001; Pines, 1996). In addition to interacting with Cdks, Cks proteins share an anion-binding domain implicated in binding to the phospho-epitope Ser/pSer/pThr/X, found on many mitotic proteins; and a domain that mediates a conformational switch between a monomeric and a dimeric form in vitro.

Despite the strong conservation of these proteins, studies in different animal models have revealed a surprising number of distinct roles at different points in the cell cycle. Studies of the Cks2 homolog in Xenopus(Xe-p9) indicate a role in the G2-M transition and the metaphase/anaphase transition, possibly by linking the Cdk to its substrates(Patra and Dunphy, 1996; Patra and Dunphy, 1998; Patra et al., 1999; Spruck et al., 2003). Mammalian Cks2 and C. elegans CKS-1 are also required for the metaphase/anaphase transition (Polinko and Strome, 2000; Spruck et al.,2003). Cks1 in mammals plays a seemingly unrelated role in promoting S-phase progression as an adaptor protein that links the SCF^Skp2 complex to one of its substrates, the Cdk inhibitor p27(Ganoth et al., 2001; Spruck et al., 2001). Recent work in yeast has revealed yet another function for Cks. Saccharomyces cerevisiae Cks1 was found to recruit the 19S and 20S proteosome to the promoter of the CDC20 gene to promote its transcription late in mitosis(Morris et al., 2003). In metazoans, which all appear to have two Cks genes, it is not yet clear how the two Cks genes carry out these diverse roles, and to what degree there is functional redundancy between them.

Female meiosis in Drosophila represents an excellent system for studying the developmental control of cell cycle progression. Female meiosis progresses through well-characterized transitions that are linked to development of the oocyte. With the aim of identifying genes required for completion of female meiosis in Drosophila, we analyzed a collection of maternal effect mutants (Schupbach and Wieschaus, 1989) that arrest early in embryogenesis. We found that the remnants (rem) gene is required for female meiosis and for mitosis in the early embryo, and that this gene corresponds to the Drosophila Cks gene at cytological location 30A (Cks30A). In addition to its maternal role, Cks30A appears to function with Cdk1 to prevent S-phase in G2-arrested histoblasts in the larva. Cks30Ainteracts with Cdk1/cyclin complexes, and this interaction is necessary for its function. One of the key functions of Cks30A is to mediate the destruction of Cyclin A, and we present evidence that this is through an effect on the activity of Cortex, a germline-specific adaptor for the Anaphase-Promoting Complex (APC). We also found that a closely related Cks gene, Cks85A, plays a distinct role in Drosophila, and the two genes cannot substitute for each other in vivo.

rem^RA74 and rem^HG24 were generated in our laboratory (Schupbach and Wieschaus, 1989). Cdk1 alleles Cdk1^E1-23 and Cdk1^E1-24(Stern et al., 1993) were provided by Christian Lehner. The distal breakpoint of Df(2L)30AC was mapped between CG9582 (2L:9017635 and Toll-4 (2L:9081671) by PCR on homozygous embryos. By the same method, the proximal breakpoint of Df(2L)gamma25 was mapped between G-gamma30A (2L:9248035) and CG13111 (9365826). This allowed mapping of rem to a region of 348 kb, containing 19 genes. Cks30A^KO was made by homologous recombination (Rong et al.,2002), removing the start codon from Cks30A. No alternative start sites exist upstream or downstream. Cks rescue constructs were all made by PCR amplifying wild-type genomic DNA corresponding to the coding region and inserting downstream of a sequence encoding the FLAG epitope tag in pUASp. Mutant forms of Cks30A were made using QuikChange Site-Directed Mutagenesis Kit (Stratagene). Cks85A (CG9790) null mutants were generated by imprecise excision of the P-element EP2346, which is upstream of the first exon. Cks85A^ex15 is deleted for all of the promoter sequence and most of the coding sequence of the gene. Northern blots were performed on mRNA purified by Trizol (Sigma) followed by polyATtract (Promega). RT-PCR to cyclin A was performed for 15 cycles on total RNA, using radiolabeled primers to allow detection of products.

Immunoprecipitations were performed on frozen 0-8 hour embryos or 0-4 hour embryos from females carrying a single copy of nanos-Gal4VP16(Van Doren et al., 1998) and a single copy of a given FLAG-tagged Cks in 10 volumes of 50 mmol/l Tris (pH 8),150 mmol/l NaCl, 1 mmol/l CaCl₂ and 0.1% Tween 20 supplemented with complete mini protease inhibitor (Roche) and type I and II phosphatase inhibitors (Sigma). Anti-FLAG beads and FLAG peptide for elution were from Sigma. Antiserum against Drosophila Cks30A was raised in guinea pig(Pocono Rabbit Farms) against full-length His-tagged Cks30A expressed in bacteria, and was affinity purified against GST-tagged Cks30A. This was used at 1/100 on western blots. Other antibodies for western blots were rabbitα-Cdk2 at 1/500, rabbit α-Fzr (1/500) and rabbit α-Cyclin B3(1/1000) (all gifts from Christian Lehner), mouse α-Fzy (from Ian Dawson), mouse α-Cyclin B and α-Cyclin A (both from Developmental Studies Hybridoma Bank) at 1/10, mouse α-FLAG M5 (Sigma) at 1/20 and rabbit α-PSTAIR (Santa Cruz) at 1/1000. Relative Cyclin A levels were determined by loading 1×, 4×, 8×, 12× and 16×dilutions of Cks30A^KO extracts and comparing band intensity (using IP Image software) to 1× of wild-type extract.

To visualize early events in female meiosis, mature oocytes from hand-dissected ovaries were activated in vitro(Page and Orr-Weaver, 1997). Later stages of meiosis were observed in fixed embryos from 0-20 minute collections. For most embryo stainings, timed collections were dechorionated in bleach, fixed in 1:1 heptane/methanol. For phospho-histone H3 staining,embryos were fixed in 3.7% formaldehyde in PBST/heptane. Embryos and ovaries were mounted in tetrahydronapthelene after incubating in methanol followed by isopropanol. Antibodies were rat α-Tubulin (Cappel) at 1/200, rabbitα-Cnn (gift of Thomas Kaufman) at 1/500, mouse α-Lamin Dmo(Development Studies Hybridoma Bank) at 1/20, rabbit α-phospho-histone H3 (Upstate) at 1/500, rabbit α-Caspase 3 (Cell Signaling Technologies)at 1/500. DNA was visualized with Oligreen (Molecular Probes) at 1/500. Fluorescence in-situ hybridization (FISH) against X- and Y-chromosome-specific repeat sequences was performed according to Dernburg(Dernburg, 2000).

The first alleles of rem were isolated in a screen for female sterile mutations (Schupbach and Wieschaus, 1989). Embryos derived from females homozygous or hemizygous for rem^HG24 or rem^RA74arrest development at the onset of embryogenesis, typically with a single metaphase mitotic spindle (Fig. 1A,B; Table 1,40-80 minutes after egg-laying). Some embryos progress through a limited number of divisions before arresting, again in a metaphase-like state(Fig. 1C; Table 1). Several observations indicated that these metaphase-arrested spindles were zygotic and not derived from the female meiotic spindle. First, they were positioned in the egg interior, where the zygotic nuclei are typically found. Second, these spindles contained centrosomes (Fig. 1B,C), suggesting that they were organized by the male pronucleus,because centrosomes in the embryo derive from the male sperm aster. Third,spindles were not detected in eggs from unfertilized rem mutant females (Fig. 1D). To confirm that rem eggs were fertilized and to determine if pronuclear fusion occurred before arrest, we performed FISH against X- and Y-chromosome sequences. In wild type, as expected, half the embryos were positive for the Y-probe, and all nuclei in these embryos also labeled positively for the X-probe (Fig. 1E). In eggs from rem^HG24 homozygous mutants, approximately 50% contained Y-signal (17/38), indicating that embryos were successfully fertilized. Of the embryos fertilized by Y-carrying males, the majority (13/17) had spindles containing both X and Y chromosomes (Fig. 1F), indicating that most, although not all, embryos completed pronuclear fusion.

The partial failure of pronuclear fusion could be due to a failure to complete meiosis. In meiosis in Drosophila females (reviewed by Foe et al., 1993), the oocyte nucleus is arrested in prophase for most of oogenesis. This arrest is broken in stage 12: nuclear envelope breakdown (NEB) occurs and the meiosis I spindle is assembled. Ovulation triggers resumption of meiosis. The meiosis I spindle rotates 90° and then rapidly enters anaphase and assembles the tandem meiosis II spindle (Fig. 2C,E). While female meiosis is being completed, the male aster forms. At completion of meiosis, the four meiotic products and the male pronucleus synchronously reform nuclear envelopes and one female pronucleus (typically the one closest to the male pronucleus) moves towards the male aster(Fig. 2G). When the nuclei are closely apposed, the male centrosome divides, migrates around the two pronuclei, and assembles the first zygotic spindle.

In rem, the first events of female meiosis are unaffected. The karyosome appeared normal, the timing of NEB and appearance of this spindle was also unaffected in the majority of mutant ovaries(Fig. 2A), although approximately 7% had abnormal spindles and failed chromosome congression on the spindle (Fig. 2B). Consistent meiotic defects were always detected upon ovulation and entry into the second meiotic division. Rotation of the spindle failed in the majority of oocytes (data not shown) and assembly of the meiosis II spindle was abnormal. The most common defect in rem mutants was a kink and/or separation of the two tandem spindles (Fig. 2D). In more severe cases, the meiotic spindle broke down and chromosome segregation was severely disrupted(Fig. 2F). Despite these errors, the female meiotic products appeared to exit meiosis, reforming nuclear envelopes, and pronuclear fusion occurred(Fig. 2H), consistent with the FISH results.

In wild type, after completion of meiosis, the three post-meiotic nuclei that do not join with the male pronucleus undergo nuclear envelope breakdown and arrest in a metaphase-like state. Chromosomes are condensed and label for the prophase/metaphase marker Phospho-Histone H3, and are associated with a radial aster of microtubules (Fig. 2I). In rem, chromosomes condensed and stained positive for phospho-histone H3, but the polar body microtubule aster did not form, and instead a shell of microtubules surrounded the chromatin(Fig. 2J).

Rem mutants fail to complement Df(2L)30AC but complement Df(2L)gamma 27 (Schupbach and Wieschaus, 1989; Lane and Kalderon, 1993). We refined the location of these deficiency breakpoints molecularly, and this allowed us to map rem to a region containing 19 genes (see Materials and methods). One probable candidate in the region is the cell cycle regulator, Cks, located in genomic position 30A(Cks30A). We found that both rem alleles contained a mutation in Cks30A, remarkably both disrupting the same conserved amino acid residue, proline 61 (P61) (Fig. 3A). rem^HG24 had a P61S substitution, while rem^RA74 resulted in a P61L change. Both alleles behaved similarly in all phenotypes tested, and therefore we will refer to them both as Cks30A^P61 mutants. We confirmed that remindeed corresponded to Cks30A by rescuing the rem mutant phenotype with a UAS-FLAG-tagged Cks30A transgene under control of the maternal driver, nanos-Gal4VP16(Table 2).

As both existing Cks30A alleles possess the same single amino acid substitution, these may not represent the loss-of-function phenotype for this gene. To establish the null phenotype of Cks30A, we created a knockout allele by homologous recombination(Rong et al., 2002). As expected, this allele, Cks30A^KO, made no detectable protein on western blots probed against an affinity-purified Cks30A-specific antiserum (Fig. 3D). Like Cks30A^P61 mutants, Cks30A^KO are homozygous viable and male fertile, but maternal effect lethal. Whereas Cks30A^P61 frequently progressed through a number of aberrant cell cycles, most Cks30A^KO embryos arrested with a single spindle (Fig. 4A, Table 1), indicating a more severe metaphase arrest phenotype.

Cks30A^KO embryos also displayed a pronounced delay in entry into mitosis. In a 0-20 minute collection of embryos, only 29% had mitotic spindles, compared with 74% in wild type(Table 1). Instead, they typically contained two separated microtubule asters, one of which was associated with condensed chromatin (Fig. 4B, arrow; Table 1). In many cases, a thin spindle connected the two asters (data not shown), suggesting that these were centrosomes that had duplicated in advance of mitosis but had not yet assembled a mitotic spindle. In 20-40 minute collections, 54% of embryos had assembled a mitotic spindle, and in 40-80 minute collections, 80% of embryos had a single mitotic spindle. The others appeared to arrest with separated centrosomes that underwent a small number of extra divisions (Fig. 4C). Therefore, Cks30A^KO is delayed in, or arrests before, mitotic spindle assembly. This delay is apparent but less pronounced in CksS30A^P61 embryos(Table 1).

As in Cks30A^P61, meiosis in Cks30A^KO progressed normally up until the resumption of meiosis following egg activation, and a similar phenotype was observed in which the tandem meiosis II spindle was kinked or separated. In addition, Cks30A^KO mutants displayed a pronounced delay in metaphase of meiosis II (Fig. 4B, Table 1). Cks30A^P61 mutants also displayed a slight delay in completing meiosis (Table 1). The delayed exit from female meiosis in Cks30A^KO embryos would be predicted to result in a failure to complete nuclear fusion. Indeed,FISH to X and Y chromosomes revealed that most embryos (11/12) did not complete joining of the male and female pronuclei(Fig. 4D). The occasional failure to complete nuclear fusion in Cks30A^P61 embryos could similarly be explained by the observed partial delay in exit from female meiosis (Table 1).

In addition to a maternal effect lethal phenotype, all cks30Amutants have a defect in abdominal cuticle deposition indicative of a partial disruption of larval abdominal histoblast development(Fig. 5A,B). This phenotype was variable, but was expressed in all alleles, and at similar strengths. In wild type, the abdominal histoblasts arrest in G2 of the cell cycle through most of larval development and then re-enter the cell cycle after puparium formation. In Cdk1 and escargot mutants, histoblasts fail to arrest in G2 and instead enter an endocycle(Hayashi, 1996; Hayashi et al., 1993). Myb mutants produce a similar phenotype, although these specifically disrupt resumption of mitosis in pupae(Katzen et al., 1998). To determine if either process is disrupted in Cks30A, we looked for genetic interactions between Cks30A and Cdk1 and between Cks30A and myb. While reducing myb activity in a Cks30A mutant background did not produce a stronger phenotype (data not shown), Cks30A in combination with loss of a single copy of Cdk1 produced a very strong effect(Fig. 5C). Similarly, a weak allele of Cdk1 displayed a mild abdominal phenotype at 18°C, but this was strongly enhanced by loss of a single copy of Cks30A (data not shown). These genetic interactions suggest that Cks30A interacts with Cdk1 to prevent polyploidy in larval histoblasts.

All multicellular organisms that have been sequenced to date have two Cks genes. We identified a second Drosophila Cks gene at chromosomal location 85A (Cks85A) (Fig. 3A,B) that encodes a protein with 61% identity and 76% similarity to Cks30A, and has an additional C terminal 21 amino acids. While Cks30A is expressed mainly in the female germline and early embryo, Cks85A is expressed throughout development(Fig. 3C). To determine the function of Cks85A, we generated a null allele by imprecise excision of a nearby P-element (see Materials and methods). This mutant(cks85A^ex15) grew more slowly than wild type and was lethal at the third instar/pupal transition. This lethality was rescued by a UAS-FLAG-Cks85A transgene under control of the zygotic driver, daughterless-Gal4 (data not shown), indicating that the lethal phenotype is due to loss of Cks85A activity. Using the FRT-ovoD system to generate Cks85A germline clones, we found that Cks85A was not required maternally, and double mutants with Cks30A in the germline produced a Cks30A mutant phenotype (data not shown). Cks85A also did not appear to have any obvious mitotic defects in zygotic tissues, either alone or in double mutants with Cks30A (data not shown). Cks85A also failed to dominantly enhance the abdominal phenotype of Cks30A mutations (data not shown). To determine whether the two Cks proteins can functionally replace each other, we attempted to rescue either mutant by constitutively expressing the other. While UAS-FLAG-Cks30A expressed with the nanos-Gal4VP16 driver could rescue a Cks30A^KO mutant, UAS-FLAG-Cks85Acould not (Table 2). Similarly,while UAS-FLAG-Cks85A under the control of daughterless-Gal4 could rescue cks85A^ex15, UAS-FLAG-Cks30A could not (data not shown). These observations indicate that the two Cks proteins are functionally distinct and cannot compensate for each other.

In histoblasts, ck30A interacts genetically with Cdk1 and both the meiotic phenotype and early mitotic arrest could reflect a role in regulating Cdk1 activity. To determine whether Cks30A and Cks85A interact physically with Cdks in Drosophila, we performed co-immunoprecipitation experiments with FLAG-tagged Cks30A and Cks85A proteins expressed in embryos (Fig. 6). FLAG-Cks85A interacted strongly with Cdk1 and Cdk2, and also co-immunoprecipitated the mitotic cyclins, A, B and B3(Fig. 6). FLAG-Cks30A also co-precipitated Cdk1, although much less efficiently than FLAG-Cks85A, and very weakly co-precipitated Cdk2 in embryos(Fig. 6). Cyclin A was not detected in FLAG-Cks30A immunoprecipitates, while Cyclin B and B3 were found at low concentrations (Fig. 6). Therefore, Cks85A was found in complexes with Cdk1 and the mitotic cyclins, as well as the S-phase kinase, Cdk2; while Cks30A associated at lower levels with Cdk1-Cyclin B and Cdk1-Cyclin B3 complexes.

The two mutant alleles, Cks30A^HG24 and Cks30A^RA74 disrupt an amino acid, P61, that lies in a region of the protein implicated in Cdk binding(Bourne et al., 1996). To determine if the loss of function phenotype in these alleles could be due to a failure to bind to Cdks, we constructed a FLAG-Cks30A^P61L mutant(corresponding to the mutation in Cks30A^RA74). As expected, this failed to rescue the Cks30A^KO mutant(Table 2). We then performed immunoprecipitations with FLAG-Cks30A^P61L and found that it was strongly impaired for Cdk binding, supporting the idea that Drosophila Cks30A functions with Cdks to promote female meiotic progression and mitotic exit in embryos.

Cks proteins all share an anion-binding domain that can interact with the phospho-epitope Ser/pSer/pThr/X. This domain can interact with partially phosphorylated Cdk substrates and has been suggested to recruit them to the Cdk to allow more phosphorylations to occur. To test the significance of this domain, we altered two residues in the domain(Bourne et al., 2000). The mutant FLAG-Cks30A^K10ES50E completely failed to rescue a Cks30A^KO mutant (Table 2), indicating that the anion-binding domain is essential for Cks30A function. As expected, this mutation had no effect on Cdk interaction(data not shown).

Cdk1 activity during mitosis is necessary for the activation of the APC. In the cellularized Drosophila embryo, the APC targets the mitotic Cyclins A, B and B3 sequentially for degradation, and this degradation is necessary for mitotic exit (Sigrist et al., 1995). APC-mediated Cyclin degradation has also been implicated in anaphase progression in the early syncytial divisions(Su et al., 1998; Huang and Raff, 1999). Therefore, the metaphase arrest in Cks30A mutants could be due to a requirement for Cks30A in the Cdk1-dependent activation of the APC. To test this possibility, we examined cyclin levels in Cks30A^KO mutants. Cyclins B and B3 occurred at wild-type levels in Cks30A mutants, but the levels of Cyclin A were much higher than in wild type (Fig. 7A). This was not due to a secondary effect of cell cycle arrest, as it was also observed in ovaries from Cks30A mutants(Fig. 7A). It was also not due to increased transcription, as cyclin A mRNA levels were equivalent in wild-type and Cks30A mutants(Fig. 7B). In cellularized embryos, the introduction of non-degradable Cyclin A results in a metaphase delay or arrest (Sigrist et al.,1995), and therefore it is likely that the metaphase arrest in Cks30A mutants was a result of the failure to degrade Cyclin A. If so, it may be possible to relieve the requirement for Cks30A by reducing Cyclin A levels. Indeed, when Cyclin A gene dosage was halved, we saw a significant rescue of Cks30A mutants(Fig. 7C).

The destruction of mitotic cyclins is mediated by the APC and an associated subunit, CDC20, that is thought to target the APC to its substrates (reviewed by Peters, 2002). In Drosophila, three CDC20 homologs, Fzy, Fzr (Rap) and Fzr2 have known roles in cyclin degradation in vivo (Dawson et al., 1995; Jacobs et al.,2002; Sigrist et al.,1995; Sigrist and Lehner,1997). Of these, only Fzy appears be active in the syncytial embryo (Jacobs et al., 2002; Raff et al., 2002). Like Cks30A mutants, hypomorphic alleles of fzy result in a metaphase arrest in the first mitotic division of the embryo(Dawson et al., 1993),suggesting the possibility that Cks30A regulates Cyclin A levels through the APC^Fzy. However, we found that Cyclin A levels were not affected in ovaries or embryos from fzy mutant females (data not shown). Based on homology, the cortex gene encodes a fourth, distantly related Drosophila CDC20 protein (Chu et al., 2001). Mutations in cortex result in a meiotic phenotype similar to that in Cks30A, in which meiosis is arrested at metaphase of the second division(Lieberfarb et al., 1996; Page and Orr-Weaver, 1996). We examined cyclin levels in cortex mutant ovaries and found that, as in Cks30A, those of Cyclin A but not Cyclins B and B3 were elevated(Fig. 7D and data not shown). Therefore, Cks30A may regulate maternal Cyclin A levels through an effect on the activity of APC^Cortex.

In S. cerevisiae, the role of Cks1 in mitotic exit appears to be primarily as a transcription factor, functioning with the 19S and 20S proteosome to promote CDC20 transcription in metaphase(Morris et al., 2003). To determine if Cks30A is required for the transcription of cortex or other Cdc20 genes, we performed northern analysis on Cks30A mutant ovaries with probes for cortex and fzy, and western analysis using a Fzr-specific antiserum. All three Cdc20 genes were expressed at normal concentrations(Fig. 7E), and therefore Cks30A does not function at the level of transcription of maternal Drosophila Cdc20 genes but more likely affects the activity of the APC^Cortex.

We have found that the maternal effect lethal gene, remnants,corresponds to one of the two Drosophila Cks genes (Cks30A). Analysis of two hypomorphic alleles and a null allele made by homologous recombination confirmed that Cks30A is not essential for cell cycle regulation in most tissue types. Rather, Cks30A functions in specialized cell cycles: the abdominal histoblast divisions, female meiosis and the syncytial divisions of the early embryo.

Cks30A mutants displayed a strikingly simple mitotic phenotype:most embryos from mutant females arrested in metaphase of the first mitotic division. Similarly, Cks30A mutants displayed a pronounced delay or arrest in metaphase of female meiosis II. Therefore, there is a common requirement for Cks30A in metaphase to anaphase progression in female meiosis II and in early embryonic mitosis. The second meiotic division is similar to a mitotic division in that it involves the segregation of sister chromatids, and therefore Cks30A may be part of a conserved machinery that is required for both of these processes.

An alternative explanation for the mitotic arrest in Cks30Amutants is that it is a secondary effect of a prior failure in meiosis or pronuclear fusion. However, our FISH experiments indicate that this mitotic arrest occurs even in embryos that successfully underwent pronuclear fusion. Also, mutations in α-Tubulin67C that block pronuclear fusion do not lead to a mitotic arrest in the embryo(Komma and Endow, 1995),indicating that these events are not coupled.

In addition to a crucial role in exit from mitosis, Cks30A is important in at least one aspect of entry into mitosis: spindle formation. Cks30A^KO mutants were severely delayed in assembly of the first mitotic spindle. Cks30A was also required for proper assembly of the female meiotic spindle and the specialized spindle-like microtubule aster of the polar bodies. Therefore, Cks30A appears to be required at two points in the mitotic (or meiotic) cell cycle: in prometaphase for spindle assembly and at the metaphase-to-anaphase transition.

These dual roles for Cks30A in meiosis appear to be at least partially conserved in other metazoans. Xenopus Cks2 (Xe-p9), like Cks30A, is required for the metaphase-to-anaphase transition in meiosis II (Patra and Dunphy,1998). Xenopus Cks2 is also required in vitro for entry into mitosis (Patra and Dunphy,1996), and this may be related to the in vivo requirement for Cks30A in spindle assembly in meiosis and mitosis. Cks genes in other eukaryotes also appear to have related but distinct functions in meiosis. Mouse cks2 is essential for anaphase progression in meiosis I of both male and female meiosis (Spruck et al.,2003), while in C. elegans cks-1 is not required for entry into anaphase, but is necessary for proper chromosome segregation in meiosis I (Polinko and Strome,2000), possibly reflecting a role in meiotic spindle assembly. Therefore, Cks genes appear to share a common requirement in entry into and exit from meiosis in different eukaryotes.

The two roles for Cks30A appear to reflect a conserved role in promoting Cdk1 activity (reviewed by Pines, 1996). Cdk1 is the central mitotic Cdk, and its kinase activity on specific mitotic proteins is required for entry into mitosis, including spindle assembly, and in maintaining the metaphase state. Cdk1 activity is also required for exit from mitosis through activation of the APC, which in turn promotes anaphase progression by targeting mitotic cyclins for destruction (reviewed by Nigg, 2001). We have found that in Drosophila Cks30A and Cdk1 interact in vivo, and that mutations in Cks30A that disrupt Cdk1 binding are compromised for activity in vivo. Cks30A also interacts genetically with Cdk1 in another cell type, the abdominal histoblasts. Therefore, our genetic and physical evidence supports the conclusion that the observed interaction with Cdk1 is required for Cks30A function.

Cks30A is required for the destruction of Cyclin A in the ovary and in the syncytial embryo and two observations argue that it is this failure to degrade Cyclin A that results in the observed delay or arrest in metaphase of meiosis II and in mitosis. First, a similar metaphase arrest is seen in cellularized embryos expressing non-degradable Cyclin A(Sigrist et al., 1995), while syncytial embryos with a slight excess of Cyclin A (approximately 1-3×wild type) due to mutations in grapes display a metaphase delay(Su et al., 1999). Second, the Cks30A mutant phenotype can be partially rescued by lowering Cyclin A levels. The CDC20 homolog, cortex, is also required for exit from meiosis II (Page and Orr-Weaver,1996; Lieberfarb et al.,1996), and we have found that cortex is also required for Cyclin A destruction in the ovary. These results argue that Cks30A and the APC^Cortex function in the same pathway leading to Cyclin A destruction, although we cannot rule out the possibility that Cks30A and Cortex act in independent pathways to promote Cyclin A destruction.

In vitro studies of Cks2 in Xenopus have led to a model in which Cks bound to Cdk1 recruits phosphorylated Cdk1 substrates to the kinase,allowing these substrates to be more efficiently recognized and thereby further phosphorylated by Cdk1 (Patra and Dunphy, 1996; Patra and Dunphy, 1998; Patra et al.,1999). The CDC27 and CDC16 components of the APC are key targets of Cks-Cdk1 phosphorylation in Xenopus. While it is not yet clear how APC phosphorylation leads to its activation, there is evidence that one of the effects of phosphorylation is to stimulate CDC20 binding to the APC(Kraft et al., 2003). Therefore it is possible that in Drosophila Cks30A-Cdk1 phosphorylates the APC, and this phosphorylation specifically stimulates the association of Cortex with the APC. Alternatively, Cks30A-Cdk1 may directly phosphorylate and activate Cortex.

In mammalian cells and in the cellularized embryo, the completion of mitosis depends on the sequential destruction of the three mitotic cyclins by the APC^Fzy (Sigrist et al.,1995). Cyclin A is destroyed first in prometaphase, dependent on APC^Fzy activity. Although the APC^Fzy is active, the spindle checkpoint is thought to inhibit its activity on Cyclin B. Upon spindle assembly, the checkpoint is relieved and APC^Fzy can mediate Cyclin B destruction (Peters,2002).

It now appears that some but not all aspects of anaphase progression are conserved in the second meiotic division and in the nuclear divisions of the syncytial embryo. Although overall Cyclin B levels do not oscillate during the early syncytial divisions (cycles 1 to 8)(Edgar et al., 1994), Cyclin B appears to undergo local degradation on the mitotic spindle at anaphase, and the injection of stabilized Cyclin B into syncytial embryos results in an early anaphase arrest (Huang and Raff,1999; Su et al.,1998). By contrast to the early anaphase arrest upon Cyclin B stabilization, the injection of an APC-inhibiting peptide into early embryos results in a metaphase arrest (Su et al.,1998). Our results suggest that this metaphase arrest is due to the failure of APC^Cortex-mediated Cyclin A destruction. Like Cyclin B levels, Cyclin A levels do not oscillate detectably in cycles 1 to 7,although unlike Cyclin B, this appears to be due to a balance between constant destruction and new protein synthesis(Edgar et al., 1994). Despite this difference, it remains possible that local oscillations in Cyclin A and Cyclin B could drive these syncytial cell cycles.

While the importance of cyclin destruction may be conserved in the early embryo, the means by which the cyclins are destroyed appears to be different. In cellularized embryos, the APC^Fzy is responsible for the sequential destruction of all three mitotic cyclins(Dawson et al., 1995; Sigrist et al., 1995). In the syncytial embryo, Fzy is not required for Cyclin A destruction. Cortex, a diverged, female germline-specific CDC20, targets Cyclin A for destruction,but has no detectable effect on Cyclin B or B3 levels in the syncytial embryo. It remains possible that Cortex is responsible for the destruction of local pools of maternal Cyclin B (and possibly B3). Alternatively, the known maternal requirement for fzy may reflect a role in the local destruction of these cyclins. This would suggest a model in which the germline utilizes two CDC20 homologs, Cortex and Fzy, to mediate the sequential destruction of Cyclins A, B and possibly B3 in the syncytial embryo. Further work will be needed to test this model. It is also not clear if Cyclin A is the only target of the APC^Cortex and if the APC^Cortex is the only target of Cks30A-Cdk1. In addition to metaphase arrest, Cks30A mutants have spindle assembly delays or defects, a phenotype that has not been observed in other cell types to result from a failure to degrade Cyclin A. Interestingly, cortex mutants also have abnormal meiosis II spindles and fail to assemble a mitotic spindle around the male pronucleus (Lieberfarb et al.,1996; Page and Orr-Weaver,1996), suggesting the possibility that the sole function of Cks30A in the female germline is to activate Cortex. Like Cks30A, C. elegans cks-1 and mouse cks2 appear to be predominantly required for meiosis (Polinko and Strome,2000; Spruck et al.,2003), and this may also reflect specific roles in activating meiosis-specific APC complexes. The histoblast requirement for Cks30A, in contrast, is unlikely to represent a role in Cortex activation (or subsequent Cyclin A destruction), as cortex mutants,either alone or in combination with Cks30A, have no effect on abdominal development (Page and Orr-Weaver, 1996; Lieberfarb et al., 1996) (A.S. and T.S., unpublished).

A specific requirement for Cks30A in activation of the maternal-specific APC^Cortex would explain why Cks30A is essential for anaphase progression in female meiosis and the syncytial embryo but not in most cell types. We can rule out an alternative possibility, that Cks30A is functionally redundant with the other Drosophila Cks, Cks85A. Cks85A mutants alone or in combination with Cks30A, do not have obvious defects in exit from mitosis (A.S. and T.S., unpublished). Furthermore, while closely related to Cks30A, Cks85A cannot replace Cks30A when expressed in the female germline, and Cks30Acannot replace Cks85A when expressed zygotically. Therefore, we conclude that the two Drosophila Cks genes have distinct and non-overlapping functions. Recently, Cks85A was found to interact with a Drosophila Skp2 homolog in a genome-wide yeast two-hybrid screen(Giot et al., 2003). Two residues on Cks1 have recently been found to be crucial for Skp2 binding in vitro (Seeliger et al., 2003),and these residues are conserved or similar in Drosophila Cks85A (see Fig. 3A). If indeed Cks85A represents the Drosophila Cks1 ortholog, it is perhaps not surprising that Cks30A cannot functionally replace Cks85A, as it has been found that Cks2 orthologs cannot bind Skp2 in vitro(Ganoth et al., 2001; Spruck et al., 2001). However,it is unexpected that Cks85A cannot substitute for Cks30A in vivo. To date all Cks proteins tested can stimulate the Cdk-dependent phosphorylation of mitotic proteins in vitro, and the mouse Cks1 can functionally replace Cks2 in vivo(Spruck et al., 2003). The failure to rescue Cks30A mutant phenotypes cannot be due to an inability of Cks85A to interact with Cdks, as Cks85A binds Cdks with even greater affinity than does Cks30A. It is possible that, analogous to the Cks1/Skp2 interaction, Cks30A has an as-yet-to-be-identified partner that is necessary for its mitotic activities. Cks85A would, therefore, be unable to carry out the mitotic activities because of an inability to bind this putative Cks30A partner.

In conclusion, we find that Drosophila Cks30A is crucial for Cdk1 activity in spindle assembly and anaphase progression in female meiosis and early embryonic mitosis, and at least part of this activity appears to be to regulate Cyclin A levels. Cks30A functions non-redundantly with another closely related Drosophila cks, Cks85A.

We are grateful to Kim McKim and people in his laboratory for help with the FISH protocol. Thanks to Christian Lehner for fly stocks and antibodies, to Thomas Kaufman for antibodies and to Kent Golic for fly stocks and plasmids. Thanks to Joe Goodhouse for confocal expertise, and thanks to Gordon Gray for fly media. We thank Stephane Larochelle for critical reading of the manuscript, and members of the Schüpbach laboratory for valuable discussions. This work was supported by the Howard Hughes Medical Institute and US Public Health Service Grant PO1 CA41086. A.S. is supported by a Research Fellowship from the National Cancer Institute of Canada.