The Spindle Assembly Checkpoint

The spindle assembly checkpoint is a surveillance pathway, conceptually parallel to the DNA damage checkpoints, but built around a mechanical rather than a chemical lesion: an unattached or non-bioriented kinetochore. The kinetochore is the proteinaceous structure assembled at the centromere of each sister chromatid, and it is the structure that physically couples chromosomes to spindle microtubules described in the article on The Mitotic Spindle and Its Assembly. When a kinetochore has not yet captured microtubules from a spindle pole, or has captured them in a way that does not generate tension across the sister-kinetochore pair, it acts as a signaling platform that recruits a defined set of checkpoint proteins. These proteins do not simply sit and block anaphase locally. Instead, the unattached kinetochore functions as a catalytic assembly line: it converts an inactive, "open" conformation of the protein Mad2 into a "closed" conformation that can bind and sequester Cdc20, the activating cofactor of the APC/C. This conformational templating reaction, discovered by biochemical and structural work, explains how a signal generated at one chromosome among dozens can spread throughout the cytoplasm rather than staying confined to that chromosome. The diffusible product of this reaction, together with Bub3 and BubR1 (Mad3 in budding yeast), forms the mitotic checkpoint complex (MCC), a soluble, four-subunit assembly that acts as a potent, stoichiometric inhibitor of APC/C-Cdc20. The APC/C is a large, multi-subunit RING-type E3 ubiquitin ligase, and Cdc20 is one of two WD40-repeat activator subunits (the other, Cdh1, acts later, after mitotic exit). Active APC/C-Cdc20 ubiquitinates two critical substrates: securin, an inhibitor of the protease separase, and cyclin B, the activating partner of Cdk1. Ubiquitination targets both for destruction by the 26S proteasome. Destroying securin liberates separase to cleave the cohesin subunit Scc1/Rad21, releasing sister chromatid cohesion and permitting anaphase. Destroying cyclin B inactivates Cdk1 and drives mitotic exit. The SAC, by inhibiting APC/C-Cdc20, prevents both events simultaneously, holding the cell in a stable metaphase-like state until the inhibitory signal is switched off.

Step 1 — The unattached kinetochore as a signaling platform. An unattached kinetochore recruits, in a hierarchical order, the kinase Mps1, the Rod-Zw10-Zwilch (RZZ) complex, and the Mad1-Mad2 core complex. Mps1 phosphorylates the outer kinetochore protein Knl1 at MELT repeats, creating docking sites for Bub3-Bub1 and Bub3-BubR1. This assembles the full checkpoint platform: Mad1-Mad2 (via RZZ and the adaptor Spindly, or via direct kinetochore receptors), Bub1, BubR1, and Bub3, all concentrated at the unattached kinetochore's outer corona, a fibrous expansion of the kinetochore that dramatically increases the surface available for signal generation.

Step 2 — The Mad2 template reaction (Mad2 "conformational dimerization"). Mad2 exists in two folded states: an open conformation (O-Mad2), which is the resting cytosolic form, and a closed conformation (C-Mad2), which is generated when Mad2 wraps around a short peptide motif in Mad1 or in Cdc20. Kinetochore-bound Mad1-C-Mad2 acts as a receptor that recruits soluble O-Mad2, converts it to C-Mad2 bound to Cdc20 (Cdc20-C-Mad2), and releases this product into the cytoplasm. Because the kinetochore-bound Mad1-C-Mad2 is regenerated after each catalytic cycle, a single unattached kinetochore can churn out many molecules of Cdc20-Mad2 per second, amplifying a local structural cue into a cytoplasm-wide chemical signal.

Step 3 — Assembly of the mitotic checkpoint complex (MCC). Cdc20-C-Mad2 then associates with BubR1-Bub3 (itself catalyzed or stabilized by kinetochore-templated reactions) to form the four-subunit MCC: Mad2-Cdc20-BubR1-Bub3. The MCC is the actual diffusible "wait anaphase" signal. It binds a second, separate molecule of Cdc20 that is itself already engaged with the APC/C, effectively creating a pseudo-substrate complex that occludes the D-box co-receptor site on the APC/C and blocks ubiquitin transfer to substrates.

Step 4 — APC/C inhibition. Structural studies (cryo-EM reconstructions of APC/C-MCC) show that the MCC engages APC/C at two points: the Mad2-Cdc20 moiety binds where a substrate would normally dock, while BubR1 contributes a pseudo-substrate KEN-box motif that further locks the ligase in an inactive configuration. With APC/C-Cdc20 inhibited, securin and cyclin B are not ubiquitinated, separase remains bound and inactive, cohesin is not cleaved, and anaphase cannot begin.

Step 5 — Satisfaction and silencing. As microtubules capture kinetochores and biorientation is achieved, tension is generated across sister kinetochore pairs (because the two sisters are pulled toward opposite poles while held together by cohesin). This tension is thought to stretch the outer kinetochore, physically separating Aurora B kinase (centromere-localized, part of the chromosomal passenger complex) from its outer-kinetochore substrates, reducing phosphorylation of Knl1 and other checkpoint scaffolds. Checkpoint proteins are stripped from properly attached kinetochores by two main routes: dynein-dynactin motors transport Mad1-Mad2, RZZ, and Spindly poleward along microtubules, physically removing them from the kinetochore; and the phosphatase PP1/PP2A, recruited to attached kinetochores, dephosphorylates Knl1 and other Mps1 substrates, disassembling the recruitment platform. Once the last kinetochore is silenced, MCC production stops, existing MCC is disassembled (with help from the AAA-ATPase TRIP13 and p31^comet, which pry open C-Mad2 back to O-Mad2), and free APC/C-Cdc20 becomes active, triggering securin and cyclin B destruction and, with them, anaphase onset and mitotic exit.

Downstream consequence: APC/C, securin, and separase. The final output of the checkpoint is not a direct block on chromosome movement, but an indirect one, achieved through control of proteolysis. This is worth tracing carefully because it is the crux of why the checkpoint's molecular target (APC/C-Cdc20) is causally sufficient to control anaphase timing.

The idea that a single unattached kinetochore could generate a diffusible signal, rather than exert only a local, cell-autonomous block, was not obvious, and it took several decisive experiments to establish it.

Micromanipulation and single-chromosome ablation (Rieder, Cole, Khodjakov, and colleagues, 1990s). Conly Rieder and colleagues used micromanipulation in living newt lung epithelial cells (PtK1 cells), a classic system for spindle imaging because of their large, flat cells and few, large chromosomes. They used a fine glass needle to reposition a single unattached chromosome so that it lagged near a spindle pole while all other chromosomes were correctly bioriented at the metaphase plate. Cells with one unattached chromosome delayed anaphase for hours, far longer than the roughly 20-30 minute normal metaphase duration in these cells, showing that a single unaligned kinetochore is sufficient to block anaphase for the entire cell, not merely for itself. When the unattached kinetochore was destroyed using a focused laser microbeam (a technique refined by Rieder and Khodjakov), thereby eliminating the source of the inhibitory signal without allowing it to attach, the checkpoint delay was abolished and the cell proceeded into anaphase even though that chromosome had never bioriented. This was the critical control: it distinguished "the checkpoint is satisfied because the kinetochore attached" from "the checkpoint is satisfied because the signal-generating structure no longer exists," and it directly demonstrated that the unattached kinetochore itself is the source of an inhibitory signal, since removing the structure (rather than fixing the attachment) also released the block.

Genetic identification of MAD and BUB genes (Hoyt, Totis, Roberts, 1991; Li and Murray, 1991). Andrew Hoyt's and Andrew Murray's laboratories independently carried out genetic screens in the budding yeast Saccharomyces cerevisiae for mutants that failed to arrest in the presence of spindle-depolymerizing drugs (benomyl or nocodazole). Wild-type yeast, when treated with these microtubule poisons, arrest robustly in a metaphase-like state (large-budded cells with an undivided nucleus) because the checkpoint holds them there indefinitely. Hoyt's screen identified three genes, MAD1, MAD2, and MAD3 (mitotic arrest deficient), whose mutants failed to arrest and instead proceeded through a lethal, chromosome-missegregating mitosis in the presence of nocodazole. Murray's group, using a related screen, identified BUB1, BUB2, and BUB3 (budding uninhibited by benzimidazole) with the same phenotype. These genes turned out to encode the core checkpoint machinery described above (Mad1, Mad2, Mad3/BubR1, Bub1, Bub3), and this genetic identification, combined with later biochemical work purifying and reconstituting the pathway, is the foundation of the entire field.

Biochemical purification of the mitotic checkpoint complex (Sudakin, Chan, Yen, 2001). Vlad Sudakin, Grace Chan, and Tim Yen isolated from mitotically arrested HeLa cell extracts a soluble complex, purified by chromatography, that contained Mad2, BubR1, Bub3, and Cdc20 in roughly stoichiometric amounts, and showed that this purified complex, added back to APC/C-containing extracts, potently inhibited APC/C-dependent ubiquitination of cyclin B and securin in vitro, at concentrations far below those of total cellular Mad2 or Cdc20. This directly demonstrated that a defined, stable, diffusible complex, rather than a diffuse cloud of individual checkpoint proteins, is the biochemical inhibitor of the APC/C, and it gave the complex its name, the mitotic checkpoint complex (MCC).

Live imaging of Mad2 dynamics and FRAP (Howell, Salmon, and colleagues, 2000; Shah, Cleveland lab). Ben Howell and Ted Salmon used GFP-Mad2 and fluorescence recovery after photobleaching (FRAP) in living cells to show that Mad2 at kinetochores turns over extremely rapidly, with a half-time of roughly 20-25 seconds at unattached kinetochores, consistent with a catalytic, flux-through model of signal generation rather than a static, stably bound pool. This kinetic behavior matched the biochemical template model in which Mad2 is continuously recruited, converted, and released, rather than accumulating and staying put, providing cell-biological support for the catalytic-conversion mechanism inferred from structural and biochemical studies of Mad1-Mad2 and Cdc20-Mad2. Together, these four lines of evidence, cell-biological micromanipulation, forward genetics, biochemical reconstitution, and live quantitative imaging, converge on the same model: unattached kinetochores generate, through a catalytic templating reaction, a diffusible protein complex that stoichiometrically inhibits the APC/C, and this is both necessary and sufficient to delay anaphase.

Example 1 — Predicting checkpoint duration from kinetochore attachment kinetics. Suppose a human cell entering mitosis has 46 chromosomes (92 kinetochores), and each kinetochore captures microtubules from a spindle pole independently, with capture events following a roughly exponential waiting-time distribution with mean time to correct, bioriented attachment of 4 minutes (a simplification, since capture in vivo is accelerated by mechanisms such as microtubule search-and-capture bias and Ran-GTP gradients, but reasonable for illustration). Question: what is the expected time until the last (92nd) kinetochore achieves biorientation, and why does this matter for checkpoint logic? For n independent exponential processes each with rate λ = 1/4 min⁻¹, the expected time to the maximum (the slowest of n) is approximately (1/λ) × Hₙ, where Hₙ is the n-th harmonic number. For n = 92, Hₙ ≈ ln(92) + 0.577 ≈ 4.52 + 0.577 ≈ 5.1. So the expected time to the last attachment is roughly 4 min × 5.1 ≈ 20 minutes. This is close to the observed duration of a normal, unperturbed prometaphase-to-metaphase interval in cultured mammalian cells (roughly 15-30 minutes), which is itself set almost entirely by the time needed for the slowest kinetochore to biorient, precisely because the checkpoint is exquisitely sensitive to even one unattached kinetochore among 92. This illustrates the core logic point: the checkpoint does not average or vote across kinetochores, it takes the minimum (the worst-attached kinetochore) as the constraint on the whole cell, which is why the "wait anaphase" signal must be a diffusible, amplified product rather than a local block whose strength depended on how many kinetochores were still unattached.

Example 2 — A clinical/experimental reasoning scenario: taxol, the checkpoint, and cell fate. Paclitaxel (Taxol) stabilizes microtubules and prevents their dynamic depolymerization, which perturbs normal spindle dynamics and interferes with proper, tension-generating kinetochore-microtubule attachments, even though microtubules are still present (unlike nocodazole, which depolymerizes them outright). Consider two hypothetical patient-derived tumor cell lines treated with a clinically relevant taxol dose: Line A has a fully intact SAC (wild-type MAD2, BUBR1, functional Mps1). Line B carries a hypomorphic MAD2 mutation reducing Mad2 kinetochore recruitment by roughly 80%. Predicted outcome for Line A: taxol-induced attachment errors are detected robustly; MCC continuously reforms at the residual unattached or tensionless kinetochores; cells arrest in mitosis for many hours. Prolonged mitotic arrest in most cell types eventually triggers "mitotic slippage" or intrinsic apoptotic pathways (via slow degradation of cyclin B even under checkpoint activity, and stress signaling), so a substantial fraction of Line A cells die, which is the mechanism by which taxol kills cancer cells clinically. Predicted outcome for Line B: with only about 20% of normal Mad2 recruitment, MCC production per unattached kinetochore is drastically reduced, checkpoint signaling is weak, and cells with unattached or maloriented kinetochores nonetheless satisfy a weakened checkpoint and enter anaphase prematurely. This produces gross chromosome missegregation (aneuploidy) rather than prolonged arrest, and because these cells exit mitosis quickly rather than lingering in a pro-apoptotic arrested state, they are comparatively taxol-resistant, even though their genomes are more scrambled. This scenario mirrors real clinical and experimental observations that partial SAC weakening can paradoxically promote drug resistance by allowing rapid mitotic exit, illustrating why SAC status is an active area of investigation as a predictive biomarker for taxane chemotherapy response.

Andrew Murray, working at the University of California, San Francisco in the early 1990s using cell-free Xenopus egg extracts and later at MIT with yeast genetics, was among the key figures who established that the cell cycle contains modular, genetically separable surveillance systems, and his 1991 paper with Ming-Wei Li identifying the BUB genes was one of the two founding genetic screens (alongside Andrew Hoyt's independent MAD gene screen at Johns Hopkins) that gave the spindle assembly checkpoint its molecular identity. Murray had already, with Marc Kirschner, developed the Xenopus extract system that made it possible to reconstitute cell cycle oscillations biochemically outside living cells, work that provided the conceptual and technical scaffold, cyclin destruction driving mitotic exit, onto which the checkpoint's mechanism of action (inhibiting that same destruction machinery) would later be mapped.

The spindle assembly checkpoint is central to the mechanism of action of taxanes (paclitaxel, docetaxel) and vinca alkaloids (vincristine, vinblastine), two major classes of chemotherapy that perturb microtubule dynamics, activate the SAC, and drive tumor cells into prolonged, often lethal, mitotic arrest. Because the outcome of that arrest, death versus slippage-and-survival, depends on the strength and duration of checkpoint signaling, SAC component expression (Mad2, BubR1) is studied as a candidate biomarker for predicting chemotherapy response and resistance in breast, lung, and other solid tumors. Weakened SAC signaling is also found in a subset of human cancers and is a proposed contributor to chromosomal instability (CIN), the persistent missegregation of whole chromosomes that generates the aneuploid karyotypes characteristic of many aggressive tumors, linking a basic mitotic surveillance mechanism directly to tumor evolution and drug resistance. Beyond oncology, small-molecule inhibitors of Mps1 (the apical kinase that nucleates checkpoint signaling) are in preclinical and early clinical development as chemotherapy sensitizers, designed to override the SAC in cancer cells that would otherwise survive taxane-induced arrest by slippage, forcing them instead into catastrophic, error-ridden divisions. In reproductive biology, SAC function (or its natural weakness) in mammalian oocytes is an active research area, since human oocytes have an unusually lax spindle checkpoint that is thought to contribute to the high rate of maternal-age-associated aneuploidy underlying conditions such as Down syndrome and to early pregnancy loss. In basic cell biology and agriculture, understanding SAC mechanics informs work on genome stability in crop breeding programs that manipulate ploidy, and in synthetic biology efforts to engineer controlled cell cycle arrest.

Ask the tutor: How does Aurora B distinguish a correctly bioriented kinetochore from an incorrectly attached one at the molecular level, and why does spatial separation from its substrates matter more than its total activity?