The mitotic spindle checkpoint. In response to kinetochores that have not achieved biorientation, this checkpoint blocks the activation of APC/C, preventing chromosome segregation and mitotic exit.

The two-state Mad2 model. (A) Mad2 binds to Mad1 and adopts the N2' conformation. Upon checkpoint activation, the Mad1–Mad2 complex is recruited to the unattached kinetochores. Another copy of N1–Mad2 is recruited to Mad1, mainly through its binding to N2'–Mad2, and is then converted to N2–Mad2. N2–Mad2 is either directly passed on to Cdc20 from Mad1 or dissociates from Mad1 to form a transient dimeric intermediate that then binds to Cdc20. Because Mad1 exists as a homodimer, three pathways can be envisioned for the transfer of N2–Mad2 from Mad1 to Cdc20: (I) the two loosely bound N2–Mad2 molecules dissociate from Mad1 as a unit; (II) one tightly bound N2'–Mad2 and one loosely bound N2–Mad2 dissociate as a unit; (III) the two tightly bound N2'–Mad2 molecules dissociate as a unit. Pathways I and III require the dimerization of Mad1. Pathway III is equivalent to the so-called “exchange” model (De Antoni et al., 2005). Only pathway I is consistent with FRAP studies in Ptk cells. (B) Upon microtubule attachment, the Mad1–Mad2 complex is depleted from kinetochores. Binding of p31^comet prevents further Mad2 turnover on Mad1 and neutralizes the inhibitory activity of Cdc20-bound Mad2, leading to activation of APC/C, degradation of securin and cyclin B, and exit from mitosis.

The multiple conformations of Mad2. Ribbon drawings of the structures of N1–Mad2, N2–Mad2^R133A, and Mad2 bound to the Mad2- binding motifs of Cdc20 or Mad1 (N2'–Mad2); the corresponding structural elements that undergo large conformational changes are colored orange or yellow. The structure of the wild-type dimeric Mad2 is unknown, but it is very likely a homodimer with each monomer resembling the conformation of N2–Mad2^R133A. Because the R133A mutation disrupts Mad2 dimerization, the αC helix is likely a part of the dimerization interface. As pointed out by Sironi et al. (2002), the COOH-terminal region (orange and yellow) in N2'–Mad2 topologically traps its ligands (MBP1, Cdc20, or Mad1) in a manner analogous to the seat belt of an automobile. For this reason, they referred to N2– and N2'–Mad2 as closed Mad2 (C-Mad2) and N1–Mad2 as open Mad2 (O-Mad2; De Antoni et al., 2005). However, it is somewhat misleading to name N1–Mad2 open Mad2 simply because strand β9 of the “seat belt” in N2– and N2'–Mad2 is absent in N1–Mad2 (a large portion of β9 in N2–Mad2 corresponds to the COOH-terminal tail in N1–Mad2) and because strands β7 and β8 in N1–Mad2 occupy its ligand-binding site.

The Mad2 template model. The symbols used are the same as in Fig. 3. In this model, N2'–Mad2 bound to Mad1 recruits N1–Mad2, which is passed on to Cdc20. The Cdc20-bound N2'–Mad2 recruits another N1–Mad2. In this way, the N2'–Mad2–Cdc20 can self-propagate away from the kinetochores.