BIOLOGICAL OVERVIEW

A distinct hallmark of eukaryotes is their use of a microtubule-based spindle to segregate their genetic information onto two daughter cells during cell division. This mechanism requires regulated sister chromatid cohesion. Sister chromatids must remain in association after DNA replication so that they can be recognized as such and oriented in the mitotic spindle during prometaphase. However, after their correct bipolar orientation in the mitotic spindle, cohesion has to be resolved so that sister chromatids can be segregated to opposite poles during anaphase. Drosophila Pimples (Pim) and Three rows (Thr) are required for sister chromatid separation in mitosis and associate in vivo. Neither of these two proteins shares significant sequence similarity with known proteins. However, Pim has functional similarities with securin proteins. Like securin, Pim is degraded at the metaphase-to-anaphase transition and this degradation is required for sister chromatid separation. Securin binds and inhibits separase, a conserved cysteine endoprotease. Proteolysis of securin at the metaphase-to-anaphase transition activates separase, which degrades a conserved cohesin subunit, thereby allowing sister chromatid separation. To address whether Pim regulates separase activity or functions with Thr in a distinct pathway, a Drosophila separase homolog (Sse) has been characterized. Sse is an unusual member of the separase family. Sse is only about one-third the size of other separases and has a diverged endoprotease domain. However, genetic analyses show that Sse is essential and required for sister chromatid separation during mitosis. Moreover, Sse associates with both Pim and Thr. Although this work shows that separase is required for sister chromatid separation in higher eukaryotes, in addition, it also indicates that the regulatory proteins have diverged to a surprising degree, particularly in Drosophila (Jäger, 2001).

Because regulated sister chromatid cohesion is an essential element of eukaryotic cell divisions, its molecular basis is expected to be conserved. Most of the current mechanistic understanding of how sister chromatid cohesion is established during S phase, maintained until the end of metaphase, and resolved at the onset of anaphase, has been obtained with yeast. In budding yeast, the cohesin protein complex is assembled on chromatin during S phase and is required for holding sister chromatids together until the end of metaphase. At the metaphase-to-anaphase transition, the Scc1p/Mcd1p subunit of the cohesin complex is proteolytically cleaved, which allows sister chromatid segregation during anaphase. The Esp1p protease (the yeast homolog of separases: for reviews see Pellman, 2001; Nagao, 2002; Ross, 2002), which is responsible for Scc1p cleavage, is kept inactive until metaphase by an inhibitory subunit, the Pds1p anaphase inhibitor. The timely activation of Esp1p at the onset of anaphase results from degradation of Pds1p by the anaphase-promoting complex/cyclosome (APC/C)-dependent pathway. The APC/C acts as a ubiquitin ligase that is regulated by the spindle assembly checkpoint (Jäger, 2001 and references therein).

Observations in other species support the notion that the mechanisms controlling sister chromatid cohesion are evolutionarily conserved. In particular, analyses in fission yeast and initial studies in vertebrates have given analogous results as described above for budding yeast. Proteins homologous to Scc1p have been shown to become cleaved at the metaphase-to-anaphase transition. Moreover, the separases, Esp1p like proteases, are all regulated by inhibitory protein subunits (named securins), which are degraded by the APC/C pathway at the end of metaphase (Jäger, 2001 and references therein).

Beyond these similarities, however, higher eukaryotes have evolved specific regulatory variations and additions. The majority of the cohesin complexes is dissociated from vertebrate chromosomes already during prophase and independent of separase activity. This early dissociation of cohesin during prophase might be required to allow chromosome condensation, which is far more extensive in higher eukaryotes than in budding yeast. The minor amount of cohesin, which remains on chromosomes until the onset of anaphase, appears to be concentrated in the centromeric region. In Drosophila, the protein MEI-S332 has been suggested to mediate this maintenance of cohesin specifically in the centromeric region. Whereas the dissociation of these remaining cohesin complexes from HeLa chromosomes has been shown to be accompanied by Scc1p cleavage that can be induced in vitro by immunoprecipitated activated separase (Waizenegger, 2000), a separase requirement for sister chromatid separation in higher eukaryotes has not yet been demonstrated directly. Moreover, the securin proteins that have been identified in budding yeast (Pds1p), fission yeast (Cut2p), and vertebrates (PTTG) do not share significant sequence similarity except for the presence of D-boxes, which target the proteins for APC/C-dependent mitotic degradation. This mitotic destruction appears to be required for sister separation in vertebrates also. However, it remains a possibility that securins have evolved to regulate proteins in addition to separase (Jäger, 2001).

Analysis of the two Drosophila genes, three rows and pimples, that do not share significant similarity with known genes, has indicated that at least in Drosophila, sister chromatid separation also involves distinct, nonconserved components. Loss of pim and thr function completely blocks the separation of sister chromatids, primarily within the centromeric region, but it does not inhibit cell cycle progression (D'Andrea, 1993; Philp, 1993; Stratmann, 1996). After each cell cycle, therefore, a doubled number of chromosome arms emanating from a common centromeric region is displayed in these mutants during mitosis. The indistinguishable mutant phenotypes argue for a common function. Consistently, Pim and Thr have been found to form a complex (Leismann, 2000) in vivo (Jäger, 2001).

Despite the lack of significant sequence similarities with known proteins, Pim has been shown to have clear functional similarities with securin proteins. Pim is degraded during mitosis via the APC/C pathway, and a nondegradable Pim mutant as well as high levels of wild-type Pim inhibit sister chromatid separation during mitosis (Leismann, 2000). Therefore, Pim might also bind and regulate a Drosophila separase. However, Pim is known to bind to Thr, which clearly does not have the structural features of separases. Pim and Thr, therefore, might either both regulate a Drosophila separase or function in a distinct pathway. To address this issue, a Drosophila separase has been identified and characterized (Jäger, 2001).

Indeed, Pim and Thr both bind to Drosophila Separase, which is required for sister chromatid separation. Interestingly, the Drosophila Sse sequence is highly diverged, lacking some features conserved in homologs from trypanosomatids to vertebrates. These results show, therefore, that the decisive role of separase in the control of sister chromatid separation has been conserved during evolution of higher eukaryotes. Nevertheless, the surprising degree of divergence of separase and regulatory proteins indicates that regulation is highly evolved, particularly in Drosophila (Jäger, 2001).

Drosophila Sse contains a C-terminal region with significant similarity to a cysteine endoprotease domain, which is found in the C-terminal region of all separases. Whereas this Sse region includes an invariant histidine as well as the putative catalytic cysteine residue that is required for function, it diverges significantly from the other separases in two additional conserved sequence blocks. One of these blocks is functionally important in budding yeast and has been proposed to represent a Ca²⁺-binding motif (Uzawa, 1990; Jensen, 2001). The Drosophila Sse sequence does not contain this Ca²⁺-binding motif. If separase activity is regulated by binding of Ca²⁺ to the conserved region in the C terminus, then Drosophila Sse activity may be regulated by Ca²⁺ binding to (an) accessory protein(s). Another striking difference is the smaller size of Sse when compared with other separases. However, genetic analysis clearly shows that Sse is required for sister chromatid separation (Jäger, 2001).

Sse mutants unable to express functional Sse zygotically complete embryogenesis presumably using the maternal Sse contribution. During the larval stages, however, the mitotically proliferating imaginal cells are specifically affected in these mutants. Cytological analysis of larval brains confirmed the findings first described by Gatti (1989) for the mutant l(3)13m-281, which reflects a complete loss of zygotic Sse function. Mitotic cells in Sse mutant larvae contain endoreduplicated chromosomes with supernumerary arms all connected primarily in a centromeric region. Such chromosomes are also observed in embryos with mutations in the genes pim or thr, which are required for sister chromatid separation (D'Andrea, 1993; Stratmann, 1996), and which encode proteins that bind to Sse (Jäger, 2001).

It is conceivable that in pim and thr mutants, Sse is destabilized, resulting in the failure to separate sister chromatids, which would explain the requirement of Pim and Thr function for sister chromatid separation. However, Western blot analyses of extracts prepared from pim and thr mutant embryos show that Sse is still present in these mutants. The possibility is favored that in these mutants a regulatory function is affected, resulting in the absence of Sse activity (Jäger, 2001).

The pim, thr, and Sse mutant phenotypes argue that Sse activity is required primarily for sister chromatid separation within the centromeric region. In contrast, entry into mitosis, including assembly of a mitotic spindle, chromosome condensation, and congression into a metaphase plate do not appear to depend on Sse activity. Moreover, the degradation of mitotic cyclins and exit from mitosis (chromosome decondensation, spindle disassembly, nuclear envelope formation) appear to occur with normal kinetics, even though sister chromatids fail to be separated and segregate to the spindle poles (D'Andrea, 1993; Philp, 1993; Stratmann, 1996). The defects in these mutants, therefore, do not appear to be detected by an efficient checkpoint mechanism comparable with the mitotic exit network of budding yeast (Cohen-Fix, 1999; Tinker-Kulberg, 1999). Cytokinesis is also attempted in pim and thr mutants, but cannot be completed. Whether the failure to complete cytokinesis is simply a consequence of the presence of nonseparated chromosomes within the equatorial plane or whether Sse activity is directly involved in cytokinesis, is not known and difficult to resolve. Another unresolved and difficult issue at present is the potential involvement of Sse activity in the control of anaphase spindle dynamics (Kumada, 1998; Uhlmann, 2000; Jensen, 2001); this has been suggested by analyses in yeast (Jäger, 2001).

The early onset of phenotypic abnormalities in pim and thr mutants reflects the rapid disappearance of maternally contributed wild-type products. This disappearance is rapid because Pim and Thr are both partially degraded during exit from mitosis (Stratmann, 1996). In contrast, the late onset of phenotypic abnormalities in Sse mutants indicates that Sse is a stable protein. In fact, Sse degradation during mitosis could not be detected by immunoblotting experiments. Unfortunately, the antibodies do not allow Sse detection by immunofluorescence, which might be more sensitive and could also provide information on subcellular localization. Nevertheless, the present evidence strongly argues against the idea that Sse activity is regulated by Sse degradation. Partial cleavage of human separase has been observed during exit from mitosis, but its significance is not yet known (Waizenegger, 2000). This mitotic cleavage of human separase occurs upstream of the conserved endoprotease domain and might therefore represent a mode of regulation that is not conserved (Jäger, 2001).

During the divisions of Drosophila embryogenesis, cleavage of Sse has not been detected. Similarly, no cleavage has been detected of the Drosophila homolog of the yeast cohesin subunit Scc1p (A. Herzig, C. F. Lehner, and S. Heidmann, unpubl., reported in Jäger, 2001). As in vertebrate cells, most of the Drosophila Scc1p homolog has also been shown to dissociate from chromosomes already during prophase (Warren, 2000). However, some can be visualized in the centromeric region of metaphase chromosomes until the onset of anaphase. It is assumed that cleavage by Sse is responsible for the subsequent disappearance of this centromeric pool and that the sensitivity of the immunoblotting experiments is insufficient to detect the cleavage of this minor fraction. So far, it has been impossible to show Sse protease activity directly (Jäger, 2001).

The fact that the majority of Scc1 is clearly not cleaved during mitosis in higher eukaryotes indicates that the regulation of Sse activity within the cell is presumably complex and targeted to the centromeric region. Although much further work remains to be done to understand Sse regulation in detail, some insights can be derived from analysis of the interactions of Sse with Pim and Thr. Pim associates with Sse in vivo. This result further supports the role of Pim as a Drosophila securin. According to two-hybrid experiments, Pim binds to the N-terminal region of Sse. The yeast securins also interact with the N-terminal regions of the separases (Kumada, 1998; Jensen, 2001). Surprisingly, however, neither the securins nor the N-terminal separase regions display sequence conservation (Uzawa, 1990; Zou, 1999). Equally surprising is the finding that Thr is required for the association of Pim with Sse. Moreover, for efficient complex formation, Pim also requires to contact Sse, because Pim² is not efficiently incorporated in a trimeric complex, despite its ability to bind to Thr. It is assumed therefore that a trimeric Pim-Thr-Sse complex is formed during interphase and present during entry into mitosis (Jäger, 2001).

Although the analysis of the binding site for Pim in Sse is difficult in vivo because of the dependency on Thr, two-hybrid experiments indicate that Pim and Thr might bind to the same region within Sse. A model of the trimeric complex takes this into account by showing that Pim, and not Thr, contacts Sse at its N terminus, and Thr can only bind to this region of Sse after Pim has been degraded. However, Pim and Thr interact with each other by use of binding sites that are distinct from those that contact Sse. Sse activity might be inhibited in the trimeric complexes because Pim prevents Thr from providing an activating contact to Sse by competitive binding within the same Sse region. The mitotic degradation of Pim might then give way to the activating Thr-Sse interaction, resulting in Sse activity at the onset of anaphase (Jäger, 2001).

Some discrepancies are noted between two-hybrid results and those obtained in vivo by coimmunoprecipitation. Particularly, the strong interaction between Pim and Sse observed in yeast contrasts with the low abundance of Pim in Sse immunoprecipitates when almost no Thr is present. It is speculated that additional levels of regulation of complex formation may be present in the Drosophila embryo, which are lacking in yeast. Nevertheless, the Pim-Sse complex formation observed in yeast is likely to reflect a significant interaction, since Pim² does not associate with Sse in yeast, as it also fails to do in the embryo (Jäger, 2001).

It is speculated that an ancient separase gene might have broken into two genes during the evolution of Drosophila (Leismann, 2000). Accordingly, Thr might correspond to the nonconserved N terminus and Sse to the conserved C-terminal endoprotease domain of the other separase proteins. This hypothesis predicts that the longer separase proteins might have two binding sites for securin. The published interaction studies do not rigorously exclude this possibility. The fact that the C-terminal acidic region of the Drosophila securin Pim interacts with Thr, whereas the C-terminal acidic regions of yeast securins interact with the N-terminal nonconserved separase regions (Kumada, 1998; Jensen, 2001) is consistent with this hypothesis (Jäger, 2001).

These suggestions are speculative. Moreover, important issues are not resolved by these hypotheses. For instance, they do not address why Pim is required for sister chromatid separation. Pim might be responsible for targeting Sse to specific subcellular locations, just as the yeast securins are required for spindle association or nuclear localization of the associated separases (Kumada, 1998; Jensen, 2001). Furthermore, if in fact a complex pathway controls Sse activation, the inactivation process is also an important issue that needs to be addressed. The requirement for an especially efficient Sse inactivation during the extremely rapid syncytial division cycles at the onset of embryogenesis might explain the particular high divergence of Sse and its regulators in Drosophila. This mechanism of regulation might be typical for insects and Sse might therefore be an interesting target for insecticide compounds (Jäger, 2001).

PROTEIN STRUCTURE

Amino Acids - 634

Structural Domains

A search of the genome sequence for a Drosophila homolog for a separase homolog turned up a single gene (CG10583) with significant similarity to the known separase genes. Comparison of cDNA and genomic sequences revealed the structure of this gene, which will be designated as Separase (Sse). The predicted protein product (Sse) has 634 amino acids and a calculated molecular mass of 72.9 kD. Thus, Sse appears to be much smaller than separase homologs from other organisms, which range in size between 150 and 230 kD. Several findings support the size prediction. The Sse upstream region has virtually no coding potential and three stop codons are present in frame upstream of and close to the presumptive translational start in several independent cDNAs. One of these short cDNAs prevents the phenotype resulting from a complete loss of Sse function when expressed in Sse mutants. Moreover, antibodies against Sse detect a protein with an apparent molecular mass of ~75 kD (Jäger, 2001).

The sequence similarity of separase homologs is restricted to the C-terminal part. This domain includes two invariant residues, a histidine and a cysteine, surrounded by regions typically found in cysteine proteases of the CD clan. This presumptive catalytic dyad is also present in Sse. However, two additional sequence blocks within this C-terminal domain, which are highly conserved among separase family members, are divergent in Sse. As a consequence, Sse is the most distant member in the separase family tree (Jäger, 2001).

date revised: 10 November 2002

Home page: The Interactive Fly © 1995, 1996 Thomas B. Brody, Ph.D.

The Interactive Fly resides on the
Society for Developmental Biology's Web server.