At the metaphase–anaphase transition, a multisubunit ubiquitin ligase called the anaphase-promoting complex or cyclosome (APC/C) in complex with its mitosis-specific activator Cdc20 mediates the ubiquitination of securin and cyclin B [1
]. Degradation of securin and cyclin B activates separase, which cleaves the Scc1 subunit of cohesin and triggers sister-chromatid separation [1
]. Premature sister-chromatid separation leads to aneuploidy, which contributes to cancer progression [3
]. In response to the existence of sister chromatids that lack attachment of spindle microtubules at their kinetochores, a cell-cycle surveillance system called the spindle checkpoint inhibits APC/CCdc20
through multiple mechanisms, stabilizes securin and cyclin B, and delays the onset of anaphase [2
]. The spindle checkpoint protein Mad2 binds directly to Cdc20 in mitosis and is essential for checkpoint-dependent inhibition of APC/C [6
]. Binding of Mad2 to Cdc20 requires Mad1, an upstream regulator of Mad2 that binds to Mad2 throughout the cell cycle [9
]. Both Mad1 and Cdc20 contain similar short peptide motifs that mediate Mad2 binding [11
]. Either inactivation or hyperactivation of Mad2 promotes tumorigenesis in mice [12
], highlighting the importance of proper Mad2 regulation in vivo. A series of biochemical, cell biological, and structural studies has established that Mad2 is a highly unusual two-state protein and that the Mad1-assisted conformational switch between these two states is central to Mad2 regulation [5
In an early study, Fang, et al. [8
] showed that recombinant purified Mad2 has two natively folded conformers, a monomer and a dimer, in the absence of ligand binding or covalent modification. The Mad2 dimer can form tetramers at high concentrations. The Mad2 dimer, but not the monomer, is active in APC/C inhibition in Xenopus
egg extracts. Furthermore, the Mad2 monomer blocks the function of the Mad2 dimer in a dominant-negative manner. Structural studies were subsequently carried out to explain this striking two-state behavior of Mad2. The structures of the Mad2 monomer and Mad2 in complex with either Mad1 or an unnatural peptide ligand called Mad2-binding peptide 1 (MBP1) that mimics the Mad2-binding motifs of Mad1 or Cdc20 were determined [11
]. These structures revealed that the Mad2 monomer has a globular domain and a flexible C-terminal tail. A Mad2 mutant with its C-terminal tail deleted (Mad2ΔC
) is an open Mad2 (O-Mad2) monomer, is incapable of binding to Cdc20, and inhibits the activity of wild-type Mad2 in a dominant-negative manner. Mad2 undergoes a dramatic conformational change upon ligand binding. The peptide ligands are trapped by the C-terminal region of Mad2 in a manner similar to the way that passengers are restrained by the seat belts in automobiles.
The Mad2 point mutant, Mad2R133A
, has two distinct monomeric conformers in the absence of ligands, which allowed us to determine the structure of both natively folded conformers of Mad2R133A
, termed N1-Mad2/open Mad2 (hereafter referred to as O-Mad2) and N2-Mad2/closed Mad2 (C-Mad2), by nuclear magnetic resonance (NMR) spectroscopy [17
]. (We initially named these two conformers N1-Mad2 and N2-Mad2. To avoid confusion, however, we have decided to adopt the nomenclature of De Antoni et al. [18
].) The structure of unliganded C-Mad2 closely resembles that of Mad1- or Cdc20-bound C-Mad2 except that the ligand-binding site is vacant. O-Mad2 can spontaneously convert to C-Mad2 with slow kinetics (t1/2
= 9 h at 30 °C) [17
]. Furthermore, cytosolic Mad2 in human cells is an O-Mad2 monomer [17
]. Monomeric C-Mad2R133A
, but not O-Mad2R133A
, is active in APC/CCdc20
inhibition. In addition, O-Mad2 and C-Mad2 can form an asymmetric O-Mad2–C-Mad2 (O–C) dimer that is less active in APC/CCdc20
], explaining why Mad2ΔC
(which only adopts the O-Mad2 conformation) can block the activity of wild-type Mad2 in a dominant-negative manner. Finally, Mad1 facilitates the conversion of O-Mad2 to C-Mad2 in vitro [17
]. Mad2 is targeted to unattached kinetochores by Mad1 and turns over rapidly at the kinetochores as revealed by fluorescence recovery after photobleaching (FRAP) studies [9
]. These studies suggest that Mad1 activates Mad2 at kinetochores by facilitating the structural conversion of O-Mad2 to C-Mad2.
More recent FRAP studies revealed that only about 50% of kinetochore-bound Mad2 undergoes fast exchange with its cytosolic pool [22
], suggesting that there is a stably bound pool of Mad2 at the kinetochores. Musacchio and coworkers then showed that this stably kinetochore-bound pool of Mad2 forms a tight complex with Mad1 and adopts the C-Mad2 conformation [16
]. The Mad1–Mad2 core complex recruits cytosolic O-Mad2 to kinetochores through asymmetric O–C Mad2 dimerization.
All available data thus support the following main framework to explain the mechanism by which Mad1 assists the binding of Mad2 to Cdc20 () [14
]. In this model, Mad2 has two distinct conformations of roughly equal free energy: a latent O-Mad2 and an active C-Mad2. The Mad1–Mad2 core complex recruits another copy of cytosolic O-Mad2 to kinetochore through O–C Mad2 dimerization. O-Mad2 bound to the Mad1–Mad2 core complex undergoes a conformational change to adopt a short-lived, high-energy intermediate conformation (I-Mad2). (I-Mad2 was previously referred to as O*-Mad2. To avoid confusion, we will use the unified nomenclature described in [24
].) I-Mad2 can be directly passed onto Cdc20 from the Mad1–Mad2 core complex. Alternatively, at least a fraction of I-Mad2 converts to unliganded C-Mad2, which dissociates from Mad1. Because Mad1 is a homodimer, two C-Mad2 molecules dissociated from Mad1 are expected to form a symmetric C-Mad2–C-Mad2 (C–C) Mad2 dimer. These unliganded C-Mad2 species are more active for Cdc20 binding and APC/C inhibition. Chemical shift perturbation experiments had initially suggested that, upon binding to C-Mad2, O-Mad2 undergoes a large conformational change to become I-Mad2 [23
]. The structure of the asymmetric O-Mad2–C-Mad2 dimer has, however, revealed that O-Mad2 bound to C-Mad2 has virtually the same conformation as does free O-Mad2 [25
]. Thus, I-Mad2 is not the stable conformation of O-Mad2 bound to C-Mad2, but rather a high-energy state with a finite lifetime. The existence and nature of I-Mad2 remain to be established.
Model for Mad1-Assisted Mad2 Activation during Checkpoint Signaling
In this study, we performed systematic mutagenesis studies of human Mad2 and obtained Mad2 mutants that preferably adopt the closed conformation. We determined the crystal structure of one such mutant, Mad2L13A, demonstrating unequivocally that C-Mad2 can form a symmetric C–C dimer in vitro. Using NMR spectroscopy, we showed that the wild-type Mad2 can form both an asymmetric O–C dimer and a symmetric C–C dimer. Mad2L13A, which predominantly exists as the symmetric C–C Mad2 dimer, is functional in cells and is active in APC/CCdc20 inhibition in vitro. Finally, the Mad1–Mad2 core complex enhances the conversion of O-Mad2 to C-Mad2. These findings provide further mechanistic insights into the conformational activation of Mad2 by Mad1 in the spindle checkpoint.