Drosophila securin destruction involves a D-box and a KEN-box and promotes anaphase in parallel with Cyclin A degradation

Sister chromatid separation during exit from mitosis requires separase. Securin inhibits separase during the cell cycle until metaphase when it is degraded by the anaphase-promoting complex/cyclosome (APC/C). In Drosophila, sister chromatid separation proceeds even in the presence of stabilized securin with mutations in its D-box, a motif known to mediate recruitment to the APC/C. Alternative pathways might therefore regulate separase and sister chromatid separation apart from proteolysis of the Drosophila securin PIM. Consistent with this proposal and with results from yeast and vertebrates, we show here that the effects of stabilized securin with mutations in the D-box are enhanced in vivo by reduced Polo kinase function or by mitotically stabilized Cyclin A. However, we also show that PIM contains a KEN-box, which is required for mitotic degradation in addition to the D-box, and that sister chromatid separation is completely inhibited by PIM with mutations in both degradation signals.

reached the correct bi-orientation within the mitotic spindle. The fact that securin stabilization by the mitotic spindle checkpoint prevents premature sister chromatid separation has been clearly shown in budding yeast (Ciosk et al., 1998;Yamamoto et al., 1996). By contrast, premature sister chromatid separation has not been observed in human cells lacking securin when arrested during mitosis (Jallepalli et al., 2001), suggesting that sister chromatid separation in these cells involves additional securin-independent pathways.
In addition to mitotic securin degradation, other levels of regulation have been implicated in the temporal control of separase activity and of sister chromatid separation. In budding yeast, phosphorylation of the Scc1 cohesin subunit by the Cdc5/Polo-kinase provides regulation at the substrate level (Alexandru et al., 2001). After phosphorylation, Scc1 is a better substrate for the yeast separase Esp1. Regulated Scc1 phosphorylation therefore provides sufficient temporal control of sister chromatid separation when the securin Pds1 is absent and Esp1 is constitutively active throughout the cell cycle. Cohesin phosphorylation by Polo-like kinase also controls the separase-independent dissociation of cohesin complexes from higher eukaryotic chromosomes during mitotic prophase (Losada et al., 2002;Sumara et al., 2002).
In higher eukaryotes, where separase is required for the removal of those cohesin complexes that remain on chromosomes until the metaphase-to-anaphase transition (Hauf et al., 2001), cyclin-dependent kinase 1 (Cdk1) has been proposed to inhibit separase activity in parallel to securin (Stemmann et al., 2001). Separase is phosphorylated by cyclin-Cdk1 complexes and thereby inhibited in cells arrested by the mitotic spindle checkpoint. High levels of non-degradable Cyclin B have been shown to inhibit sister chromatid separation in Xenopus egg extracts and in PtK1 cells (Stemmann et al., 2001;Hagting et al., 2002). In Drosophila embryos, expression of non-degradable Cyclin A, which is found exclusively in Cdk1(Cdc2) and not in Cdk2(Cdc2c) complexes, delays sister chromatid separation significantly (Jacobs et al., 2001;Kaspar et al., 2001;Parry and O'Farrell, 2001;Sigrist et al., 1995). Cdk1 inactivation resulting from the APC/C-dependent proteolysis of the mitotic cyclins at the metaphase-to-anaphase transition therefore presumably leads to separase activation, similar to securin degradation. Securinand Cdk1-dependent separase regulation might be largely redundant, explaining why human cells display only very subtle defects in the absence of securin function.
In securin-expressing cells, mitotic sister chromatid separation is generally assumed to be strictly dependent on mitotic securin degradation. However, the corresponding evidence is from experiments involving overexpression of securin variants with mutant degradation signals. By contrast, our experiments in Drosophila embryos involving expression at physiological levels raised the possibility that sister chromatid separation might not depend on PIM degradation (Leismann et al., 2000). Therefore, we have further analyzed the role of PIM degradation.
We have previously shown that PIM contains a novel D-box variant that functions as a mitotic degradation signal (Leismann et al., 2000). D-boxes that form an RxxLxxxxN consensus sequence (Peters, 1999) were initially identified in B-type cyclins where they are required for mitotic destruction. The D-box variant identified in PIM starts with a K instead of an R. Apart from the D-box, a different destruction signal, the KEN-box, can also mediate APC/C-dependent degradation of various proteins (Pfleger and Kirschner, 2000;Peters, 2002). A KEN-box has recently been shown to contribute to the mitotic degradation of human securin (Hagting et al., 2002;Zur and Brandeis, 2001). Here, we show that PIM contains a functional KEN-box as well. Moreover, we show that physiological levels of PIM with mutations in both the D-and the KEN-box do not support sister chromatid separation, in contrast to our previous findings with D-box mutants. Sister chromatid separation in Drosophila, therefore, might well be strictly dependent on securin degradation. However, we also show genetic interactions arguing for the presence of additional, securinindependent regulation of sister chromatid separation.
The transgenes gpim-myc and gpim dba -myc, in which expression is directed by the pim + regulatory region, have been described previously (Leismann et al., 2000;Stratmann and Lehner, 1996). For the generation of gpim dba lines, we first enzymatically amplified a fragment from pKS+gpim dba -myc with the primers RS33 (5′-TT-CAATACGTAGGCGCC-3′) and OL59 (5′-CAAGGAAAACAC-CGGCATTAATTG-3′). The resulting 300 bp amplification product with the D-box mutations was used as a primer, which was extended after hybridization to the plasmid pKS+S2P4B1. The newly synthesized strands were annealed and ligated, yielding pKS+S2P4B1 dba . Its insert was excised by BamHI and transferred into the germ line transformation vector pCaSpeR 4 (Pirrotta, 1988).
To generate the g>stop>pim, g>stop>pim kenadba , and g>stop>pim kenadba -myc lines, we first constructed pKS+5′FRTgpim3′ by inserting a fragment with a FLP recombinase target sequence (FRT) amplified from the plasmid pKB345 (kindly provided by K. Basler, University of Zurich) with the primers RS62 (5′-GCGA-GATCTACCGGGGGATCTTGAAGTTC-3′) and RS63 (5′-CGCGG-ATCCATTTTTGTACCCAGCTTCAAAAGCGC-3′) after digestion with BglII and BamHI into the BamHI site of pKS+gpim. pKB345 contains a 2.4 kb Asp718 fragment with 3′UTR and transcriptional terminator sequences from the heat-shock protein gene, hsp70 (stop cassette) flanked by two FRT sites (Struhl and Basler, 1993). For the introduction of the stop cassette along with the second FRT site into pKS+5′FRTgpim3′, we first transferred the Asp718 fragment from pKB345 into pKS+, in which the BamHI site had been destroyed by religation after filling the restricted site. The BamHI site within the insert fragment of the resulting plasmid was also eliminated. The plasmid was then used as a template for enzymatic amplification with the primers OL83 (5′-GCCGGTGTGCTGACGCATGTGAAG-3′) and RS63 (5′-CGCGGATCCATTTTTGTACCCAGCTTCAAAAG-CGC-3′). The amplification product containing the 1500 bp stop cassette followed by the downstream FRT site was digested with BglII and BamHI and ligated into the BamHI site of pKS+5′FRTgpim3′. Ligation in the correct orientation resulted in pKS+>stop>gpim, a plasmid with an intact stop cassette flanked by FRT sites in front of the start codon. Its Asp718-NotI insert fragment was transferred into the corresponding sites of pCaSpeR 4. To introduce the KEN-box mutations, the corresponding pim region was amplified using the primers OL13 (5′-CCATCTCTAGAAAAGTGCCGC-3′) and OL84 (5-GCCGGTGGCAGCCGCGTTTAAAATCGGATC-3′) from the template pUASTpim kena -myc. To add the D-box mutations, the resulting product was used as a primer for an additional polymerase chain reaction in combination with a second primer OL8 (5′-ATTAGTAGTACAAAGATACCTAGC-3′) and the template pKS+ gpim dba -myc. The fragment with the kena and dba mutations was used to replace the wild-type sequence in pKS+>stop>gpim either by using BamHI and SnaBI (in the case of g>stop>pim kenadba ) or BamHI and BglII (in the case of g>stop>pim kenadba -myc). The final pCaSpeR 4 constructs were again obtained by transposing Asp718-NotI fragments. All constructs were verified by DNA sequencing and used for Drosophila germ line transformation according to standard procedures.
Fixation of embryos and immunolabeling was performed essentially as described previously (Leismann et al., 2000). Eggs were collected for 2 hours and aged at 25°C from the following crosses: -pim 1 /CyO, P{w + , ftz-lacZ} females with pim 1 , gpim dba II.1/CyO, P{w + , ftz-lacZ} males (aging 4.5 hours) or with pim 1 , gpim dba -myc II.5/CyO, P{w + , ftz-lacZ} males (aging 12 hours) By contrast, at this stage, cells in the CNS are hardly affected in polo mutants that do not express PIM dba (C,D; polo -) or in the polo + siblings that express PIM dba (A,B; pim dba ). B, D and F show high-magnification views with the CNS from the embryos displayed in A, C and E, respectively. (G-J) Using prd-GAL4 and UAS-CycA∆1-53, stabilized Cyclin A was expressed in alternating embryonic segments. Expressing segments are indicated by arrowheads in G and I or by white lines in H and J, which display high-magnification views of epidermal regions. DNA staining indicates that the metaphase delay caused by stabilized Cyclin A is prolonged in embryos that also express the stabilized securin PIM dba under the control of the pim + regulatory region. Compared with embryos without PIM dba (G,H; CycA∆N), metaphase plates (arrows) in regions with stabilized Cyclin A are enriched in embryos that also express PIM dba (I,J; pim dba CycA∆N).
To analyze a potential synergism between gpim dba and reduced polo + function, all embryos with a strong abnormal central nervous system (CNS) phenotype were first identified on the basis of the DNA labeling before genotypes were assigned on the basis of the anti-βgalactosidase labeling. Thereby 80% of the embryos with a strong abnormal CNS phenotype were found to be polo 10 homozygotes with gpim dba II.1. An additional 10% were polo 10 homozygotes without gpim dba II.1, and 10% were polo + siblings with gpim dba II.1.
To analyze a potential synergy between gpim dba and expression of mitotically stabilized Cyclin A, embryos with UAS-CycA∆1-53 III.2 and prd-GAL4 at the stage of mitosis 16 were scored for the presence of gpim dba II.1 and for a strong enrichment of metaphase plates in prd-GAL4-expressing epidermal segments compared with the intervening segments. Although 82% of the embryos with gpim dba II.1 displayed a strong metaphase enrichment, only 25% of the embryos without gpim dba II.1 were comparably affected.

Results
Embryos homozygous for pim null mutations but equipped with a maternal pim + contribution from pim heterozygous mothers progress normally through the initial embryonic cycles. Entry into mitosis 15 and progression to metaphase are still normal. Moreover, the transition from metaphase to anaphase is triggered as well, as evidenced by the degradation of the mitotic Cyclins A, B and B3. However, sister chromatid separation during mitosis 15 is completely inhibited (Stratmann and Lehner, 1996). This block of sister chromatid separation after the exhaustion of the maternal pim + contribution is almost completely prevented when pim embryos inherit a gpim dba -myc transgene (Leismann et al., 2000). gpim dba -myc drives expression of PIM with C-terminal myc epitopes and a mutant D-box (AKPAGNLDA instead of KKPLGNLDN). PIM dba -myc is stable during mitosis according to confocal immunofluorescence microscopy, in contrast to PIM-myc with the wild-type D-box. gpim dba -myc expression is controlled by the normal pim regulatory region, and the resulting level of PIM dba -myc before mitosis 15 was found to be comparable to wild-type PIM levels. Because stabilized PIM dba -myc protein promoted sister chromatid separation in pim mutants, it appeared that sister chromatid separation is not dependent on degradation of the Drosophila securin PIM (Leismann et al., 2000). Analogous experiments with a gpim dba transgene driving expression of a D-box mutant PIM version without myc epitopes also revealed rescue of mitosis 15 in pim mutants (data not shown), excluding the possibility that sister chromatid separation in the presence of stabilized PIM dba -myc occurs simply because C-terminal myc epitopes specifically abolish the inhibitory PIM function.
Instead of being required during each mitosis, PIM degradation might be important to keep protein levels below a critical threshold. We have previously shown that already moderate overexpression of wild-type pim (about fivefold) is sufficient to block sister chromatid separation. Moreover, although gpim dba rescued sister chromatid separation during mitosis 15 and 16 in pim mutants, it did not allow later divisions (data not shown), perhaps because the levels of stabilized PIM dba had built up beyond the critical threshold.
Journal of Cell Science 116 (12) Fig. 2. Stabilized Cyclin A does not inhibit PIM-myc degradation. Stabilized Cyclin A was expressed in alternating segments using prd-GAL4 and UAS-CycA∆1-170 (which results in a more extensive metaphase delay than UAS-CycA∆1-53, shown in Fig. 1). In addition, the embryos expressed PIM-myc under control of the pim regulatory region throughout the epidermis. Epidermal regions with expression of stabilized Cyclin A on the right but not on the left side of the dashed line are shown at high magnification after labeling with a DNA stain (A, B; DNA, red in B), anti-Cyclin B (B,C; CYCB, green in B) and anti-myc (D, PIM-myc). Arrested metaphase cells in regions with stabilized Cyclin A were found to lack PIM-myc (see arrowheads).
If degradation of the securin PIM was not an obligatory process required during each mitosis, separase bound to securin would be expected to have sufficient basal activity to allow sister chromatid separation. In this case, premature sister chromatid separation during interphase and early mitosis would have to be prevented by securinindependent regulation. As securin-independent regulation at the level of Scc1 phosphorylation by Cdc5/Polo kinase has been described in yeast, we analyzed whether a reduction in polo function enhances the effects of stabilized PIM dba . Within the CNS of polo-mutant embryos, we observed many abnormal cells with very large polyploid nuclei, when these embryos also carried gpim dba (Fig.  1E,F). Similar abnormal cells were almost never observed in either polo + sibling embryos with gpim dba (Fig. 1A,B) or in polosibling embryos without gpim dba (Fig. 1C,D). In the presence of stabilized PIM dba , therefore, the remaining level of maternal polo + contribution is no longer sufficient to mask phenotypic abnormalities in polo-mutant embryos. Moreover, reduced polo + function enhances the effects of stabilized PIM dba .
In addition to Scc1 regulation by Cdc5/Polo kinase, vertebrate Cdk1 has been shown to regulate separase independently of securin (Stemmann et al., 2001). The effects of stabilized Cyclin A in Drosophila embryos (Sigrist et al., 1995;Jacobs et al., 2001) are consistent with the finding that vertebrate Cdk1 phosphorylates and thereby inhibits separase. Mutant Cyclin A versions that cannot be degraded during mitosis delay progression through the embryonic cell divisions during metaphase before sister chromatid separation. Therefore, Drosophila Cyclin A-Cdk1 complexes might inhibit separase activity. Accordingly, the effects of stabilized Cyclin A∆1-53 are expected to be enhanced by expression of stabilized PIM dba . Labeling with antibodies against tubulin (data not shown) and a DNA stain clearly revealed an increased number of metaphase figures in epidermal regions of embryos expressing both Cyclin A∆1-53 and PIM dba (Fig. 1I,J), compared with embryos expressing only Cyclin A∆1-53 (Fig.  1G,H). The stabilized Cyclin A∆1-53 therefore results in a more pronounced metaphase delay in the presence of the stabilized PIM dba .
In principle, stabilized Cyclin A might delay cells in metaphase because it results in an inhibition of PIM degradation during mitosis. However, cells delayed in metaphase by stabilized Cyclin A∆1-170 no longer contained PIM-myc according to immunolabeling experiments, whereas metaphase cells that do not express Cyclin A∆1-170 were always positive for PIM-myc (Fig. 2). We conclude, therefore, that the metaphase delay induced by stabilized Cyclin A does not result from delayed PIM degradation.
The phenotypic interactions between stabilized PIM dba and Polo or Cyclin A are consistent with the notion that separase complexed with non-degradable securin might have sufficient activity to allow sister chromatid separation and that the timing of this process is controlled by pathways other than securin degradation. However, the observed sister chromatid separation in PIM dba -expressing cells might also be supported by residual mitotic PIM dba degradation. A KEN motif, which is found close to the N-terminus in all of the securins (Fig. 3A), might allow some limited mitotic PIM dba degradation, escaping detection by confocal microscopy as applied in our previous experiments.
To determine whether the KEN motif of PIM functions as a degradation signal, we analyzed the mitotic stability of a myctagged PIM version with a mutant KEN-box (PIM kena -myc with AAA instead of KEN). PIM kenamyc, and PIM-myc for control, were expressed in the anterior region of embryos during cycle 14, as described previously (Leismann et al., 2000). Immunolabeling at the stage of mitosis 14 indicated that PIM kena -myc is largely stable throughout mitosis (Fig. 3F-I), in contrast to PIM-myc, which was detected before but not after the metaphase-to-anaphase transition ( Fig. 3B-E). Progression beyond the metaphase-to-anaphase transition was monitored by the labeling of DNA and Cyclin B, which is rapidly degraded when cells enter anaphase. Our results show that the KEN-box is required and that the variant D-box (KKPLGNLDN), which is still present in PIM kena -myc, is not sufficient for normal mitotic PIM degradation.
Overexpression of PIM kena -myc resulted in mitotic defects. Normal anaphase and telophase figures were not observed in PIM kenamyc-positive cells that had progressed beyond the metaphaseto-anaphase transition according to the absence of anti-Cyclin-B labeling. Instead of pairs of well-Journal of Cell Science 116 (12) (Q-T) Expression of g>pim kenadba in pim + embryos allows normal proliferation during the early mitotic divisions but not during the late divisions in the CNS. DNA staining at stage 14 reveals the presence of many large polyploid abnormal nuclei (arrowheads) in the CNS of g>pim kenadba embryos (R,T; g>pim kenadba ), which are absent in control siblings (Q,S; pim + ). S and T show highmagnification views with the CNS from the embryos displayed in Q and R, respectively. (U) Coimmunoprecipitation experiments show that PIM kenadba -myc associates normally with SSE and THR. Anti-myc immunoprecipitates (IP anti-myc) isolated from extracts (extract) of embryos expressing Cdk1-myc (Cdk1-myc), PIM kenadba -myc (pim kenadba -myc) or PIM-myc (pim-myc) were probed by immunoblotting for the presence of SSE (SSE), THR (THR), Cyclin B (CYCB) and tubulin (TUB). separated telophase daughter nuclei, which were readily observed in Cyclin-B-negative regions in the PIM-myc control experiments (Fig. 3D,E, arrowheads), Cyclin-B-negative regions of PIM kena -myc-expressing embryos displayed decondensing metaphase plates or chromatin bridges between partially separated nuclei (Fig. 3H,I arrows). These abnormalities caused by PIM kena -myc were indistinguishable from those previously observed with PIM dba -myc which has been shown to inhibit sister chromatid separation (Leismann et al., 2000).
Sister chromatid separation is also inhibited by strong overexpression of wild-type PIM-myc (Leismann et al., 2000). By contrast, at low physiological expression levels, PIM-myc and, remarkably, also the stabilized versions PIM dba -myc (Leismann et al., 2000) and PIM kena -myc ( Fig. 3J-L), can promote sister chromatid separation in pim mutants.
To analyze the function of PIM with mutations in both D-and KEN-box, we constructed additional transgenes (g>stop>pim kenadba and g>stop>pim kenadba -myc), allowing the expression of PIM kenadba or PIM kenadba -myc under the control of the normal pim regulatory region. To establish chromosomal insertions of these potentially detrimental transgenes, we inserted a stop cassette flanked by FLP recombinase target sites (>stop>) into the 5′ untranslated region. This stop cassette was eventually excised by transmitting the established insertions via males expressing FLP recombinase specifically in spermatocytes. Expression of the paternally recombined transgenes (g>pim kenadba and g>pim kenadba -myc) started at the onset of zygotic expression during cycle 14 of embryogenesis. Expression of g>pim kenadba and g>pim kenadba -myc in pimmutant embryos did not allow sister chromatid separation during mitosis 15 (Fig. 4M-O and data not shown). Instead of normal mitotic figures, which were readily apparent in pim + sibling embryos (Fig. 4A, arrows), only decondensing metaphase plates were observed during exit from mitosis (Fig.  4M, arrowheads). Thus, pim-mutant embryos expressing g>pim kenadba and g>pim kenadba -myc displayed the same phenotype as pim mutants without transgene (Leismann et al., 2000) (and data not shown) or with the non-recombined g>stop>pim kenadba transgene ( Fig. 4I-K).
Control experiments with g>stop>pim transgenes encoding wild-type PIM showed that expression after stop-cassette removal was sufficient to promote normal sister chromatid separation in pim mutants (Fig. 4E-G). Moreover, additional control experiments showed that the recombined g>pim kenadbamyc transgene was expressed as expected. Anti-myc immunoblotting clearly showed expression (data not shown), and co-immunoprecipitation experiments (Fig. 4U) indicated that the PIM kenadba -myc protein associates efficiently with Separase (SSE) and Three rows (THR), a Drosophila protein known to form trimeric complexes with SSE and PIM (Jäger et al., 2001). In addition, although g>pim kenadba -myc expression in pim + sibling embryos had little effect during the initial embryonic cell divisions (mitosis 14-16), it resulted in a severe mutant phenotype in the CNS where additional cell divisions occur (Fig. 4T). Wild-type PIM therefore appears to protect cells from the effects of PIM kenadba -myc but only as long as the latter has not yet accumulated to high levels.
In summary, our experiments with g>pim kenadba and g>pim kenadba -myc in pim mutants show that sister chromatid separation does not occur in the presence of physiological levels of the double mutants PIM kenadba and PIM kenadba -myc, in contrast to our findings with the single mutants PIM dba , PIM dba -myc and PIM kena -myc.

Discussion
Sister chromatid separation during the metaphase-to-anaphase transition in Drosophila is strictly dependent on accumulation of the securin PIM (Stratmann and Lehner, 1996). Because we do not understand why Drosophila PIM has to accumulate to allow sister chromatid separation, we cannot evaluate whether PIM versions with mutations in both the D-and the KEN-box (PIM kenadba and PIM kenadba -myc) are still capable of providing this positive function. However, we emphasize that these versions still bind normally to the known partners SSE and THR.
Mutations in either the D-or the KEN-box result in significant stabilization of PIM protein during mitosis. Neither the D-nor the KEN-box, therefore, are sufficient for normal degradation during the embryonic cell divisions in Drosophila. Similar observations have been described for human securin (Hagting et al., 2002;Zur and Brandeis, 2001). However, in contrast to Drosophila, mitotic degradation of human securin still occurs quite effectively when either only the D-or the KEN-box is intact. The D-and KEN-boxes of Drosophila PIM, therefore, might function less independently than the corresponding motifs in human securin. Eventually, the understanding of D-and KEN-box function will require structural analyses of their interactions with Fizzy/Cdc20 and Fizzy-related/Cdh1, which recruit proteins with these degradation signals to the APC/C (Burton and Solomon, 2001;Pfleger et al., 2001). Fizzy and Fizzy-related are clearly both involved in PIM degradation, at least indirectly, as PIM is stabilized in both fizzy and fizzy-related mutants (Leismann et al., 2000) (data not shown).
Under the assumption that PIM kenadba and PIM kenadba -myc are still capable of providing the positive PIM function, our results with these stabilized mutants suggest that PIM must be degraded during each and every mitosis to allow sister chromatid separation. Although not detectable by confocal microscopy, the single mutants PIM dba and PIM kena might not be completely stable in mitosis. After low-level expression in pim-mutant embryos, residual mitotic degradation of singlemutant proteins might free some separase activity sufficient for sister chromatid separation. Similar results have been observed with the fission yeast securin Cut2, which is completely stabilized in a Xenopus extract destruction assay by mutations in either of the two D-boxes, and yet, low-level expression of single-but not double-mutant proteins is able to complement growth of cut2-ts strains at the restrictive temperature (Funabiki et al., 1997). We emphasize that even in wild-type cells, mitotic PIM degradation appears to be far from complete, and it can be speculated that it is the PIM protein of a special pool of separase complexes that is more efficiently degraded, perhaps on kinetochores or during transport on spindles towards kinetochores. At high expression levels of PIM with or without single mutations, free excess of this securin might rapidly re-associate and inhibit the activated separase, resulting in the observed block of sister chromatid separation.
Our results also point to alternative pathways that might regulate separase activity and sister chromatid separation independently of PIM degradation. As in yeast, the success of mitosis in cells with reduced separase function is dependent on Polo kinase in Drosophila embryos. Moreover, as expression of mitotically stabilized Cyclin A versions result in a metaphase delay without inhibiting PIM degradation, Cyclin A appears to contribute independently of PIM to the inhibition of premature sister chromatid separation. Even though it remains to be analyzed whether Polo kinase and Cyclin A-Cdk1 act during Drosophila divisions as proposed for Polo homologs (Alexandru et al., 2001) and vertebrate Cyclin B-Cdk1 (Stemmann et al., 2001), our results indicate that separase and sister chromatid separation are unlikely to be regulated exclusively by securin degradation.