In every dividing cell, a time comes when the two copies of the genome need to be separated. The aptly named enzyme separase springs into action and gets the job done. Unleashing separase at any other time in the life of a cell would be dangerous, so the enzyme is kept well guarded. Human separase is held in check by not one but three mutually exclusive inhibitors. Writing in Nature, Yu et al.1 report structures of human separase in complex with two of these inhibitors. The structures show commonalities but also striking differences. One of the inhibitors snakes along separase to embed itself in the enzyme’s active site. The other forces separase to inhibit itself; at the same time, this inhibitor is itself inhibited by separase in an entangled embrace.
Cell division is studied both for its beauty and for the danger that it represents. When all goes well, new healthy cells are born. But when things go awry, newborn cells inherit faulty copies of the genome and might die or become the seed for cancerous growth. Movies of cell division showing this dramatic process never fail to intrigue, and such films have provided an inspiration that has launched renowned scientific careers (see, for example, ref. 2). In the key scene of cell-division movies, chromosomes split abruptly along their length, separating the two copies of the genome destined for the daughter cells. The major force behind the split is separase, which at this crucial moment cleaves a protein complex called cohesin that serves as ‘glue’ between the genome copies3.
Until this pivotal moment, separase activity is blocked by inhibitors. The best-characterized inhibitor is the protein securin, which begins to bind to separase while that enzyme is still being made4; it even supports separase synthesis5. Genetic and biochemical experiments were the first to hint at the possibility that securin mimics the cohesin substrate and binds to the active site of separase6. Structures of budding yeast and nematode separase in complex with securin confirmed this7,8. One of the structures solved by Yu et al. using cryogenic electron microscopy now shows that the same is true for human separase (Fig. 1).
The big surprise comes with the second structure that Yu et al. solved. This shows separase bound to another of its inhibitors, the cyclin-dependent kinase 1 (CDK1) complex, which consists of the proteins Cks1, CDK1 and cyclin B. This complex is itself a major player in cell division, and it functions by phosphorylating (adding phosphate groups to) hundreds, if not thousands, of different proteins to bring about the cellular changes required for division. To solve this structure, the authors used a neat trick5: they fused separase to a short piece of securin that was long enough to promote separase synthesis but not so long that it impaired binding of the CDK1 complex.
Despite much previous biochemical insight into the interaction between separase and the CDK1 complex9, anyone would have been hard-pressed to imagine the structure that Yu et al. have solved. The same sites in separase that are used to bind securin are occupied, but now by separase itself, which has become autoinhibitory (Fig. 1). However, unlike securin (and probably cohesin), which binds to separase in a linear fashion (as a continuous stretch of protein), the autoinhibitory elements of separase found in this key binding region are non-contiguous and come from three loops, which the authors call autoinhibitory loops (AIL1, AIL2 and AIL3). AIL3 not only autoinhibits separase; it also inhibits CDK1 by binding to its active site. This loop probably binds to every protein in the complex.
A fourth separase loop wraps around cyclin B and contributes to inhibition of the CDK1 complex. At the centre of this separase loop is a well-characterized phosphorylation site that is required for the formation of this complex9. Visualization of this phosphorylation site in the structure revealed a previously unrecognized phosphate-binding pocket in cyclin B. Seeing the strikingly different types of inhibition achieved by securin and the CDK1 complex makes one wonder what sort of inhibition mechanism the third and most recently discovered10 separase inhibitor, SGO2/MAD2, might have up its sleeve.
Separase inhibition by securin is probably universal across eukaryotes (organisms with a nucleus), but inhibition by the CDK1 complex seems to be vertebrate specific. Why different inhibition modes evolved, and how labour is distributed between these inhibition options, remain mysterious. Some mammalian cell types crucially rely on separase inhibition by the CDK1 complex11, highlighting the importance of the new structural data that Yu and colleagues report. All three types of separase complex coexist in human cell lines10. What determines which inhibitor binds to a given separase molecule, and whether separase molecules bound to the various inhibitors execute different functions in the cell once released from inhibition, is unclear. Separase has other roles in cell division beyond that of cohesin cleavage, and perhaps different inhibitors enable spatial or temporal control of separase activity.
Interestingly, although separase needs to be released from its inhibitors to trigger chromosome separation, the CDK1 complex also binds to separase late during cell division, at a time after separase has become active and cohesin has been cleaved. Inhibition of CDK1 in this complex supports the movement of chromosomes into the daughter cells12.
The formation of this late complex requires the enzyme PIN113, which acts on a site in the separase loop that binds to cyclin B. Curiously, assembly of the CDK1 complex bound to separase in the structure that Yu et al. solved did not require PIN1. Human separase not only cleaves cohesin, but also cleaves itself when it becomes active. To solve the structure, the authors made separase catalytically inactive to prevent its auto-cleavage. Did this modification alleviate the requirement for PIN1? And is the reported structure of the CDK1 complex bound to separase representative of a complex found early during cell division, but possibly different from that assembled later on?
Clearly, more remains to be uncovered about the regulation of separase. Further behind-the-scenes footage will be needed before we can not only admire, but also fully understand, cell-division movies.
Yu, J. et al. Nature https://doi.org/10.1038/s41586-021-03764-0 (2021).
Lehmann, R. & Peters, J.-M. Cell 184, 10–14 (2021).
Uhlmann, F., Wernic, D., Poupart, M.-A., Koonin, E. V. & Nasmyth, K. Cell 103, 375–386 (2000).
Hellmuth, S. et al. J. Biol. Chem. 290, 8002–8010 (2015).
Rosen, L. E. et al. Nature Commun. 10, 5189 (2019).
Nagao, K. & Yanagida, M. Genes Cells 11, 247–260 (2006).
Luo, S. & Tong, L. Nature 542, 255–259 (2017).
Boland, A. et al. Nature Struct. Mol. Biol. 24, 414–418 (2017).
Gorr, I. H., Boos, D. & Stemmann, O. Mol. Cell 19, 135–141 (2005).
Hellmuth, S., Gómez-H, L., Pendás, A. M. & Stemmann, O. Nature 580, 536–541 (2020).
Huang, X. et al. Mol. Cell Biol. 29, 1498–1505 (2009).
Shindo, N., Kumada, K. & Hirota, T. Dev. Cell 23, 112–123 (2012).
Hellmuth, S. et al. Mol. Cell 58, 495–506 (2015).
Lin, Z., Luo, X. & Yu, H. Nature 532, 131–134 (2016).
The author declares no competing interests.