Deciphering the modes of human separase inhibition by securin and CDK1-CCNB1

ABSTRACT

Accurate chromosome segregation depends on tight regulation of the protease separase, which cleaves the ring-shaped cohesin complex that entraps the two sister chromatids. We recently reported structures of human separase bound to its inhibitors securin or the cyclin-dependent kinase 1 (CDK1)-cyclin B1 (CCNB1)-cyclin-dependent kinases regulatory subunit 1 (CKS1) complex and discovered an array of molecular mechanisms that block cohesin-cleavage.

In metaphase of mitosis, sister chromatids align at the equator of the mitotic spindle before being segregated in anaphase. The transition from metaphase to anaphase is a point of no return that is initiated by the dissolution of sister chromatid cohesion through cleavage of the cohesin complex component RAD21 (best known as SCC1).^1,2 Untimely and erroneous chromosome segregation leads to genomic instability and causes aneuploidy and tumorigenesis.

Cohesin, the protein complex that holds the sister chromatids together, is cleaved by an aptly named protease called separase. Because of the irreversible nature of this process, human separase activity is suppressed by binding to three mutually exclusive inhibitors: Securin,³ the cyclin dependent kinase 1 (CDK1)-cyclin B1 (CCNB1)-cyclin-dependent kinases regulatory subunit 1 (CKS1) complex⁴ and the Shugoshin 2 (SGO2)-mitotic arrest deficient 2 like 1 (MAD2L1, best known as MAD2) complex.⁵ These inhibitors are well orchestrated throughout the cell cycle thereby enabling temporal and spatial control of separase activity. However, only securin is conserved from yeast to human, while the other two inhibitors seem to be vertebrate-specific.

Separase was discovered in the late 1990s, but structural studies of the enzyme were reported only recently. First, the C-terminal caspase-like domain of a fungal separase was cocrystallized with a substrate-mimicking peptide to reveal insights into substrate recognition near the catalytic site.⁶ Two subsequent structural studies of yeast⁷ and Caenorhabditis elegans⁸ separase-securin complexes provided insights into the overall architecture of the enzyme, explained the substrate-occlusion inhibitory mechanism of securin using pseudosubstrate sequences, and further rationalized the requirement for an arginine residue in substrates inserted into the catalytic site of separase. Despite these findings, regulation of vertebrate separase remained poorly understood. To decipher and compare the modes of separase inhibition by securin and the CDK1-CCNB1-CKS1 (or CCC) complex, we used cryogenic electron microscopy to determine the structures of separase in complex with either securin or the CCC complex.⁹

Securin is cotranslated with separase and promotes its synthesis. Thus, a recombinant separase-securin complex can readily be prepared in large quantities.⁸ Preparation of the separase-CCC complex however is difficult because binding of securin and the CCC complex to separase is mutually exclusive and securin-free separase is difficult to produce. To resolve this dilemma, we used a neat trick previously described by Rosen and coworkers. Fusing a short C-terminal fragment starting from amino acid 160 of securin to the N-terminus of separase (securin^∆160-separase) stabilizes the enzyme¹⁰ and also allows binding of the CCC complex. Another prerequisite for separase-CCC complex assembly is the phosphorylation of a single serine residue in separase by CDK1⁴. To this end, we generated a single baculovirus for co-expression of CDK1, CCNB1, and CKS1 in insect cells. Coexpression of CDK1-CCNB1 in insect cells produces an active kinase complex. Adding the active CCC complex to securin^∆160-separase, followed by incubation with ATP and Mg²⁺, resulted in a stable separase-CCC complex. The structures of the separase-securin and separase-CCC complex shed light on the molecular basis for the robust control of chromosome segregation.

As in other species,^7,8 human securin is directly embedded in the separase catalytic site and occupies nearby docking sites through pseudosubstrate sequences (Figure 1a). In contrast, the CCC complex binds at the periphery of separase and bridges a clearly defined interdomain cleft, thereby rigidifying three autoinhibitory loop (AILs) segments present in separase itself (Figure 1b). These non-contiguous loops possess motifs that are similar to sequences that can be found in securin and SCC1 and, more importantly, occupy binding sites that are usually recognized by securin or SCC1. Thus, these sequence motifs represent universal separase-binding motifs that have co-evolved in several proteins, and in case of securin and separase prevent SCC1 substrate binding by blocking common binding sites. We show that these AILs partially inhibit separase even in the absence of the CCC complex, as loop deletions increase separase activity.⁹ In brief, while the role of securin is to directly occupy substrate-binding sites on separase, the CCC complex uses flexible non-contiguous loop segments in separase itself to occlude SCC1 binding. These mechanistic differences originate from the fact that the CCC complex does not possess any pseudosubstrate motifs to directly bind to the catalytic site or adjacent binding sites. These novel insights might also assist future structure-guided drug design studies.

Figure 1.

Molecular configuration of separase-securin and separase-CDK1-CCNB1-CKS1 (CCC) complexes. (a) Cartoon representation of two views of the separase-securin complex. Securin (shown in orange) binds in an antiparallel orientation to the entire length of separase (light blue), thereby occupying multiple separase-binding sites. The autoinhibitory loops (AILs) AIL1 and AIL2 are shown in dark blue and gray, respectively. The cell division cycle 6 (CDC6)-like domain of separase and AIL3 are shown in cyan. The cyclin B1 (CCNB1)-binding loop (purple) is flexible in the separase-securin structure and thus drawn as dashed lines. The separase domains are labeled below. (b) Cartoon representation of two views of the separase-cyclin dependent kinase 1 (CDK1)-cyclin B1 (CCNB1)-cyclin-dependent kinases regulatory subunit 1 (CKS1) complex (or CCC complex). Binding of the CCC complex (with CDK1 shown in red, CCNB1 in yellow and CKS1 in pale yellow) to phosphoserine 1126 (pSer1126) of separase is mediated by CCNB1 through a previously unidentified phosphate-binding pocket. Fusing a short C-terminal fragment starting from amino acid 160 of securin (securin^∆160) to the N-terminus of separase stabilizes the enzyme. CCC binding rigidifies the three AILs in separase that consequently occupy binding sites on separase and CDK1-CCNB1, using conserved sequence motifs. Mutual exclusive binding of securin and CCC to separase would result in steric clashes between the AILs and securin. The CCNB1-binding loop entangles CCNB1 with pSer1126 at its center. The N-terminal HEAT repeat domain is flexible in the CCC complex and indicated in dashed lines. Grey asterisks indicate active sites on separase and CDK1

It is possible that the third inhibitory complex, SGO2-MAD2, contains short linear motifs that inhibit separase by directly binding to substrate-docking sites⁵; however, structural studies are needed to prove if a direct or indirect inhibitory mechanism is at play.

CCC binding to separase strictly depends on phosphorylation of a specific separase serine residue.⁴ Our structure now reveals that phosphoserine 1126 engages with a previously unidentified phosphate-binding pocket in CCNB1 (Figure 1b). The residues forming this phospho-residue binding pocket are highly conserved throughout eukaryotic B-type cyclins, but are absent in A-type cyclins.⁹ We therefore conclude that this pocket defines a novel specificity site in B-type cyclins and predict that a variety of cell cycle regulatory proteins may be controlled through this pocket.

Interestingly, the interaction between separase and CDK1-CCNB1 does not just inhibit separase activity but reciprocally inhibits the kinase activity of CDK1, helping to govern CDK1 activity in metaphase and anaphase. When the CCC complex is bound to separase, a pseudosubstrate motif in one of the separase AILs blocks the kinase active site. Thus, separase autoinhibitory loops provide all the sequences required for mutual inhibition of both enzymes in the protease-kinase complex.

While our recent work provides an important step forward, much remains to be learned about the molecular basis of separase function and regulation in human cell division. The mechanism of separase inhibition by the SGO2-MAD2 complex represents just one of many unsolved mysteries surrounding this fascinating cellular enzyme.

Funding Statement

This work was supported by the Foundation Ernst et Lucie Schmidheiny; Novartis Stiftung für Medizinisch-Biologische Forschung [#19B074]; Schweizerischer Nationalfonds zur Förderung der Wissenschaftlichen Forschung [310030_185235].

Disclosure of potential conflict of interest

No potential conflicts of interest were disclosed.