Substrate recognition by human separase

Abstract

The cohesin complex encircles sister chromatids in early mitosis. At anaphase onset, sister separation is triggered by the proteolytic cleavage of the cohesin subunit SCC1/RAD21 by separase. SCC1 contains two cleavage sites, where cleavage is stimulated by SCC1 phosphorylation. Substrate recognition and cleavage are only partly understood. Here, we determined structures of human separase in apo- or substrate-bound forms that, together with biochemical analysis, provide critical insights into separase cleavage regulation. We verify the first SCC1 cleavage site and reassign the second. We show that substrates, including separase autocleavage sites and the two SCC1 cleavage sites, interact with docking sites in separase, including five phosphate-binding sites. We also describe the interaction between the cohesin subunit SA1/SA2 and separase, which promotes cleavage at the second SCC1 site. Using cross-linking mass spectrometry and cryo-electron microscopy, we propose how cohesin is targeted by human separase. Our work provides an extensive functional and structural framework that explains a key event in cell division.

Fig. 1.. Cleavage of SCC1 at two sites and cryo-EM reconstructions of human separase in apo- or substrate-bound states.

(A) Domain organization of human separase and SCC1. Hs, Homo sapiens. Separase consists of three domains shown as blocks: a HEAT-repeat domain (gray), a TPR-like domain (light blue), and a C-terminal protease domain (blue). The TPR-like domain contains autoinhibitory loops 1 to 3 (AIL1 to 3), shown with lines in dark blue, gray, and cyan, as well as two large flexible insertions, insert 1 (amino acids 1065 to 1152) and insert 2 (amino acids 1278 to 1572) illustrated in black. Three autocleavage sites (green line) are in the insert 2. The catalytic site consists of two conserved residues H2003 and C2029 (red). The N-terminal structural maintenance of chromosome 3 (SMC3)–binding domain and C-terminal SMC1-binding domain of SCC1 are indicated with gray blocks. SCC1 contains two cleavage sites: cleavage site 1 (R172) with nearby substrate motifs (orange) and cleavage site 2 (R427) with nearby substrate motifs (orchid). The SA1/2-binding region of SCC1 is shown in purple. (B) Schematic representation of SCC1 cleavage products resulting from two sites. aa, amino acids; FL, full-length. (C) Cleavage assay using ³⁵S-labeled wild-type SCC1 and mutants as substrates. Results are representative of three independent experiments. WT, wild-type. (D) Fusion strategy for reconstitution of separase-substrate complexes. Left: Inactive apo-separase (separase^C2029S; light gray) with insert 2 shown with a dashed line. AIL1 and autocleavage sites are color coded as in (A). Middle and right: Region containing three autocleavage sites in insert 2 is replaced by SCC1 fragments: SCC1 site 1 (amino acids 100 to 320: orange) or SCC1 site 2 (amino acids 310 to 550; orchid). AIL1 is replaced by a GS linker (ΔAIL1) (8). The C2029S mutation in separase indicates an inactive variant. Red star: the catalytic site. (E to G) Cryo-EM reconstructions of apo-separase [C2029S; (E)] and separase bound to SCC1 [amino acids 100 to 320; (F)] and SCC1 [amino acids 310 to 550; (G)]. The color codes are the same as in (D).

Fig. 2.. Multiple substrate motifs interact with separase.

(A) Views of SCC1 site 1 (amino acids 100 to 320) and site 2 (amino acids 310 to 550), plus autocleavage sites and securin binding across the three domains of separase. Structures of separase^C2029S, securin-separase complex (PDB: 7NJ1), and SCC1^310–550-separase^{C2029S/ΔAIL1} are aligned to SCC1^100–320-separase^{C2029S/ΔAIL1}. Separase is depicted as surface representation in dark gray. SCC1 (orange or orchid), securin (orange red), and autocleavage sites (green) are shown as cartoon. (B) Close-up view of SCC1, securin, and autocleavage site binding near the catalytic site of separase. Hydrogen bonds are indicated with yellow dashed lines. (C) Close-up view of SCC1, securin, and autocleavage site binding to the TPR-like domain of separase. (D) Close-up view of SCC1 and securin binding to the HEAT-repeat domain of separase. (E) Affinity measurement of unphosphorylated SCC1^100–320 and SCC1^350–550 binding to separase^C2029S using fluorescence polarization. Each experiment was repeated three times; data points indicate mean ± SEM. (F) Cleavage assay of ³⁵S-labeled wild-type SCC1 and mutants containing two copies of cleavage site 1 motif (SCC1^{2× site 1}) or two copies of cleavage site 2 motif (SCC1^{2× site 2}). Left: Schematic representation of wild-type SCC1 and mutants. Right: Autoradiograph of SCC1 cleavage. Results are representative of three independent experiments. (G) Top: Quantification of relative abundance of site 1 cleavage fragment (amino acids 1 to 172). Below: Quantification of relative abundance of site 2 cleavage fragments (amino acids 173 to 427 plus amino acids 428 to 631). For normalization, the intensities of the cleavage fragments were divided by the total intensities of all bands in the respective lanes.

Fig. 3.. SA1/2 stimulates SCC1 cleavage at site 2 and cryo-EM structure of separase bound to SCC1-SA2 complex.

(A) Cleavage assay showing stimulation of SCC1 cleavage at site 2 by SA1/2 proteins, tested at two different concentrations (conc.,1 and 4 μM). conc., concentration. Left: Autoradiograph showing cleavage products of ³⁵S-labeled SCC1. Right: Quantification of relative abundance of site 2 cleavage fragments (amino acids 173 to 427 and 428 to 631). For normalization, the intensities of the site 2 cleavage fragments were divided by the total intensities of all bands in the respective lanes. Results are representative of three independent experiments. (B) Schematic representation of the SCC1-SA2-separase complex reconstitution. SCC1 (amino acids 310 to 550) fragment (purple) includes the SA2 binding region and cleavage site 2. SA2 is shown in deep sky blue, and separase is depicted in wheat. (C) Views of the cryo-EM map of the SCC1^310–550-SA2-separase^{C2029S/ΔAIL1} complex. (D) Ribbon representation of the SCC1^310–550-SA2-separase^{C2029S/ΔAIL1} complex highlighting the interaction interfaces between the three proteins. Red box: cryo-EM density showing the interaction interface between SCC1 and the TPR-like domain of separase. Black box: cryo-EM density showing the interaction interface between SCC1, SA2, and the protease domain of separase. (E) Close-up view showing SCC1 binding to the protease domain of separase in the complexes of SCC1^310–550-SA2-separase^{C2029S/ΔAIL1} and SCC1^310–550-separase^{C2029S/ΔAIL1}. The two structures were aligned using separase as a reference. SCC1 in the two complexes, depicted as sticks, is shown in purple and orchid. Side chains of W1777, R2067, and R2101 in the two complexes are shown in wheat and gray, respectively. Hydrogen bonds are indicated with yellow dashed lines. (F) Close-up view of the cleavage site 2 motif of SCC1 binding near the catalytic site of separase. (G) Two views of the interaction interface formed by a short helix of SCC1 (amino acids 413 to 419), C-terminal helices of SA2, and the protease domain of separase.

Fig. 4.. Cohesin targeting by separase is mediated by SMC subunits.

(A) Schematic representation of the strategy for reconstituting the cohesin-separase complex. Full-length SCC1 replaces the autocleavage sites in insert 2 (dashed line) of separase. SMC3 is colored in light and forest green, SMC1 in light and dark red, SCC1 in gray, SA2 in deep sky blue, and separase in a blue gradient. P-sites 1 to 5 are indicated as red dots. (B) Cross-linking MS (XL-MS) analysis of cohesin-separase complex cross-linked using sulfosuccinimidyl 4,4′-azipentanoate (Sulfo-SDA). Circular view showing intermolecular cross-links (top) and intramolecular cross-links (bottom) in the cohesin-separase complex. Each protein is color coded as in (A). Cross-links between the cohesin subunits (SMC1, SMC3, and SCC1) and separase are shown with dark red, green, and blue lines, respectively. Intramolecular cross-links are shown in purple. (C) Cryo-EM map and model of separase bound to SMC3-SCC1. Satisfied Sulfo-SDA cross-links with Cα-Cα distances ≤ 25 Å are shown as orange solid lines. (D) A model illustrating phosphorylation- or SA1/2-dependent regulation and the specificity determinants of cohesin cleavage by separase. APC/C, anaphase-promoting complex/cyclosome.