Chromosome Segregation and the Metaphase-to-Anaphase Transition

Chromosome segregation is fundamentally a protein destruction problem, not a signaling problem in the conventional sense. From S phase onward, sister chromatids are held together along their length by cohesin, a ring-shaped protein complex built from the subunits SMC1, SMC3, SCC1 (RAD21/Mcd1), and SCC3. The SMC1–SMC3 heterodimer forms a V-shaped coiled-coil arm, and SCC1 bridges the two ATPase heads to close the ring, topologically entrapping the two sister DNA molecules like a twist-tie around two strands of rope. Cohesin is loaded onto DNA during S phase and largely persists until mitosis, holding sisters together against the outward pulling forces of the spindle and generating the tension that the spindle assembly checkpoint (SAC) reads out as "correct attachment." The enzyme that opens this ring is separase, a cysteine protease that cleaves SCC1 at a specific site, destroying the ring's integrity and releasing the sisters simultaneously across the entire genome. Separase is dangerous — premature cleavage would tear the genome apart at the wrong time — so the cell keeps it in check by two overlapping brakes: an inhibitory chaperone protein called securin, which binds separase directly and blocks its active site, and inhibitory phosphorylation of separase by Cdk1–cyclin B. Anaphase onset requires removing both brakes, but the trigger event that cascades into removing them is destruction of securin (and cyclin B) by the ubiquitin–proteasome system. The central engine of this destruction is the Anaphase-Promoting Complex/Cyclosome (APC/C), a large (~1.2 MDa, 14–19 subunit) multi-subunit RING-type E3 ubiquitin ligase. The APC/C itself is not sufficient; it requires an activating coactivator subunit, Cdc20, that recognizes destruction motifs (D-box, KEN-box) on substrates like securin and cyclin B and presents them to the catalytic core for polyubiquitination. Polyubiquitinated securin and cyclin B are then degraded by the 26S proteasome within minutes. This is why the transition is switch-like: proteolysis is irreversible (you cannot un-degrade a protein), it can be made essentially all-or-none by positive feedback and threshold effects, and it can be held in check by a single well-defined inhibitory circuit — the spindle assembly checkpoint — until the moment every chromosome is correctly attached.

Step 1 — Cohesin holds sisters together under tension. During S phase, cohesin rings are loaded around each pair of newly replicated sister DNA molecules. By metaphase, cohesin persists mainly at centromeres (arm cohesin has already been largely removed in vertebrates by the prophase pathway, which strips cohesin from chromosome arms via Polo-like kinase 1 and Aurora B phosphorylation, protected at centromeres by the protein Shugoshin/Sgo1, which recruits the phosphatase PP2A). Bipolar attachment of sister kinetochores to microtubules from opposite spindle poles pulls the sisters apart, and centromeric cohesin resists this pulling, generating measurable inter-kinetochore tension (on the order of tens of piconewtons per kinetochore pair). This tension is the physical signal that satisfies the spindle assembly checkpoint.

Step 2 — The SAC restrains APC/C-Cdc20 until attachment is complete. As described in The Spindle Assembly Checkpoint, unattached or tension-less kinetochores generate the mitotic checkpoint complex (MCC), composed of Mad2, BubR1, Bub3, and Cdc20 itself, which sequesters Cdc20 and directly inhibits APC/C. Even a single unattached kinetochore among 46 (in a human cell) can generate enough MCC to hold the entire genome in metaphase, because the checkpoint signal is diffusible and amplifies catalytically at each unattached kinetochore.

Step 3 — Satisfying the checkpoint releases active APC/C-Cdc20. When the last kinetochore achieves proper bi-oriented attachment and tension, MCC production stops, existing MCC is disassembled (partly via the AAA-ATPase p31^comet and TRIP13, which extract Mad2 from Cdc20), and free Cdc20 accumulates. Cdc20 binds the APC/C core, docking substrates bearing destruction-box (D-box: RxxLxxxxN) or KEN-box motifs into the catalytic center, where the E2 enzymes UBE2C (UbcH10) and UBE2S build polyubiquitin chains on target lysines.

Step 4 — APC/C-Cdc20 ubiquitinates securin, marking it for destruction. Securin is a preferred, rapidly-degraded APC/C-Cdc20 substrate. Polyubiquitinated securin is recognized by the 26S proteasome and destroyed within a few minutes of checkpoint satisfaction.

Step 5 — Free separase becomes catalytically active. Securin destruction liberates separase from its inhibitory grip. Simultaneously, cyclin B — also an APC/C-Cdc20 substrate — is degraded, lowering Cdk1 activity and removing the second inhibitory phosphorylation on separase (dephosphorylated by PP2A/PP1-family phosphatases). Only when both brakes are released is separase fully active. This dual-lock mechanism (securin binding + Cdk1 phosphorylation) provides a fail-safe: securin destruction alone is not quite sufficient for maximal separase activity, adding robustness against accidental triggering.

Step 6 — Separase cleaves cohesin, and chromatids separate. Active separase cleaves SCC1/Rad21 at two specific sites, springing open the cohesin ring. Because cohesin removal happens essentially simultaneously across all 92 sister chromatid pairs (in a human diploid cell) within a minute or two, all chromosome pairs separate synchronously — this is what produces the crisp, coordinated anaphase movement rather than a ragged, one-chromosome-at-a-time release.

Step 7 — Poleward movement: anaphase A and anaphase B. Once sisters are physically freed, two distinct, overlapping movements segregate them:

• Anaphase A: kinetochore microtubules depolymerize predominantly at their kinetochore-proximal (plus) ends, and the kinetochore itself, coupled to the shrinking microtubule via the Ndc80 complex and ring/fibril-forming couplers (the Dam1/DASH ring in yeast; a biased-diffusion "hook" mechanism involving Ska complex in vertebrates), moves poleward as the tube shortens, "reeling in" the chromosome. Some depolymerization also occurs at spindle poles (minus-end, "Pac-Man" flux).
• Anaphase B: the spindle poles themselves move apart, driven by plus-end-directed kinesin-5 motors sliding antiparallel interpolar microtubules apart, and by cortical dynein pulling on astral microtubules from the cell cortex, further increasing the separation between the two chromosome sets.

Together anaphase A and B can separate sister sets by tens of micrometers within several minutes, at rates on the order of 1–2 micrometers per minute for anaphase A chromosome-to-pole movement in many animal cells.

Yeast genetics identified securin, separase, and the destruction pathway. In budding yeast (Saccharomyces cerevisiae), Andrew Murray, Doug Koshland, and colleagues in the early-to-mid 1990s isolated mutants that failed to separate sister chromatids properly. The key genes were PDS1 (precocious dissociation of sisters), whose product turned out to be securin, and ESP1 (extra spindle poles), whose product is separase. Loss-of-function mutants in ESP1 arrested with sister chromatids stuck together at anaphase-like spindle configurations, even though the spindle elongated — showing that spindle elongation (anaphase B) and chromatid separation are mechanistically separable. Deleting PDS1 caused cells to separate chromatids prematurely, before proper spindle attachment, demonstrating that Pds1/securin normally restrains separation. Combining these results with epistasis experiments (esp1 mutations are epistatic to pds1 deletion — i.e., you still need Esp1 even without Pds1) placed securin upstream of, and as an inhibitor of, separase.

Xenopus egg extracts reconstituted the entire pathway biochemically. Marc Kirschner's and Andrew Murray's laboratories used cell-free extracts from Xenopus laevis eggs — a system that can be driven synchronously through mitosis in a test tube — to show directly that cyclin B destruction is necessary and sufficient to drive exit from mitosis, and, together with work from Jan-Michael Peters' laboratory identifying the APC/C as a ubiquitin ligase, established that APC/C-dependent ubiquitination targets both cyclin B and securin for 26S proteasomal degradation. Adding non-degradable cyclin B to extracts blocked anaphase and mitotic exit; adding purified APC/C with Cdc20 to extracts triggered securin and cyclin B destruction and cohesin cleavage.

Direct biochemical proof that separase cleaves cohesin. Kim Nasmyth's laboratory provided the decisive biochemical demonstration in 1999–2000: they showed that the yeast cohesin subunit Scc1/Mcd1 is cleaved at two specific sites at the metaphase-to-anaphase transition, that this cleavage requires Esp1 (separase), and that mutating the cleavage sites in Scc1 to be protease-resistant blocks sister chromatid separation even though the rest of the cell cycle proceeds — chromosomes remained glued together while the rest of mitosis (spindle elongation, cyclin destruction) continued normally. This is the single most direct causal proof: block the one specific proteolytic cut, and chromatid separation alone fails, nothing else does.

FRET biosensors visualized the trigger live, in single human cells. More recently (2000s), FRET (Förster resonance energy transfer)-based separase activity sensors and securin-GFP degradation reporters, imaged live in mammalian tissue culture cells, showed that securin destruction and the loss of cohesin-mediated tension at kinetochores occur within a very narrow time window — seconds to a couple of minutes — precisely coincident with visible anaphase onset, and that this event is essentially synchronous across all chromosomes in a single cell, exactly as predicted from the "all cohesin cut at once by one activated protease" model.

Checkpoint-defective mutants confirmed the gatekeeping logic. Mutants lacking Mad2 or Bub1 (components of the spindle assembly checkpoint) undergo anaphase onset (securin/cyclin B destruction, cohesin cleavage) even with unattached or maloriented kinetochores, producing missegregation and aneuploidy — showing that the SAC's job is precisely to withhold APC/C-Cdc20 activation, not to control any other step in the pathway. Together, these lines of evidence — yeast epistasis genetics, cell-free biochemical reconstitution, direct demonstration of a single non-cleavable-cohesin phenotype, and live-cell imaging of the trigger in real time — constitute as rigorous a causal chain as exists in cell biology for any regulatory switch.

Example 1 — Interpreting a non-cleavable cohesin experiment. A researcher engineers yeast cells whose only copy of SCC1 encodes a protein in which the two separase cleavage sites are mutated to be protease-resistant (call this scc1-noncleavable). These cells are otherwise wild-type for Esp1, Pds1, and the APC/C. Prediction and reasoning: Because separase can no longer cut cohesin, sister chromatids should remain glued together indefinitely, even though upstream events (checkpoint satisfaction, APC/C-Cdc20 activation, securin and cyclin B destruction, separase activation) proceed normally. You would predict:

• Securin and cyclin B are still destroyed on schedule (their degradation does not require cohesin cleavage — it is upstream).
• Cdk1 activity still drops, so many "exit from mitosis" events (nuclear envelope reformation attempts, spindle disassembly cues) may partially occur.
• The spindle can still elongate somewhat (anaphase B does not strictly require chromatid separation to begin, though full segregation fails), but sister chromatids stay paired and are dragged, unsegregated, toward one pole or torn.
• The cell should arrest or show catastrophic missegregation, because the physical link (cohesin) — not the enzyme (separase) or its regulator (securin) — was the target mutated.

This is exactly the phenotype Nasmyth's lab observed: it proves cohesin cleavage, not merely separase activation or securin loss, is the proximal cause of chromatid separation. If instead sisters separated normally in this mutant, that would falsify the model — it does not, confirming causality.

Example 2 — A quantitative checkpoint-override thought experiment. Consider a human cell with 46 chromosomes (92 sister kinetochore pairs) entering mitosis. Suppose a drug partially inhibits Mps1 kinase (required for SAC signaling) such that only 95% of kinetochores can generate checkpoint signal, while 5% of kinetochores (randomly distributed, on average about 4–5 kinetochore pairs per cell) remain "invisible" to the checkpoint due to drug action, even if unattached. Question: Will this cell proceed through a normal anaphase, and what is the expected consequence? Reasoning: The SAC operates as an essentially digital, single-kinetochore-sufficient block: even one unattached, checkpoint-competent kinetochore can generate enough MCC to prevent APC/C-Cdc20 activation for the entire cell (the "one kinetochore, whole-cell" logic, confirmed by laser-ablation and micromanipulation experiments removing single kinetochores). But if 5% of kinetochores are chemically blinded from signaling regardless of attachment status, those 4–5 pairs may remain unattached or mono-oriented at the time the rest of the checkpoint is satisfied and anaphase fires. Since cohesin cleavage is genome-wide and cannot be selectively withheld from specific chromosomes, all cohesin — including on the mis-attached chromosomes — is cut simultaneously. The unattached or incorrectly attached chromosomes will not segregate correctly: they may lag in the spindle midzone, move to the wrong pole, or be split unequally, generating aneuploid daughter cells missing or gaining whole chromosomes. Numerically: if ~4 chromosome pairs (out of 46) are at risk per division, and each has roughly even odds of missegregating when checkpoint-blind, you'd expect measurable rates of aneuploidy for those specific chromosomes across a cell population — a scenario resembling how Mps1 inhibitors are used experimentally to induce controlled chromosome missegregation, and analogous to how some tumor cells with chronically weakened SAC signaling generate ongoing chromosomal instability (CIN).

The modern understanding of this pathway is the product of several converging laboratories rather than a single discoverer, but Kim Nasmyth stands out as the figure most closely associated with establishing cohesin and separase as the direct executors of chromatid separation. Working first at the Vienna Biocenter and later at Oxford, Nasmyth built his career on budding yeast genetics, using the classic tools of conditional lethal mutants, epistasis analysis, and — crucially — direct biochemical demonstration of cleavage products on Western blots. His laboratory's identification of cohesin's ring subunits (SMC1, SMC3, SCC1, SCC3) and demonstration that separase cleaves SCC1 at defined sites, with cleavage-site mutants blocking separation while leaving the rest of mitosis intact, turned a genetic pathway into a mechanistic, molecularly explicit model. Nasmyth's style exemplified a hallmark of rigorous cell biology: never stopping at "gene X is required for process Y," but pushing to ask what the gene product physically does to the substrate, and proving it with a cleavage-site mutant that isolates exactly one biochemical step. His parallel work (with collaborators including Frank Uhlmann, who performed the decisive cleavage experiments) also clarified that cohesin's "gluing" function is topological — a ring encircling DNA — rather than a simple sticky protein-protein interaction, a structural insight later confirmed by electron microscopy and single-molecule studies of purified cohesin.

Cancer biology and chromosomal instability. Many solid tumors show chronic missegregation of whole chromosomes (chromosomal instability, CIN), often traced to weakened spindle checkpoint signaling, cohesin gene mutations (e.g., STAG2 mutations in bladder cancer and other cancers), or overexpression of separase, which has itself been found amplified in some breast and other cancers and can act as an oncogene when overactive, driving aneuploidy that fuels tumor heterogeneity and drug resistance.

Anti-cancer drug development. Because rapidly dividing tumor cells depend heavily on accurate mitotic checkpoint function and spindle mechanics, drugs targeting mitotic kinases (Mps1/TTK inhibitors, Aurora kinase inhibitors, PLK1 inhibitors) are being developed to selectively kill cancer cells by forcing premature or catastrophic anaphase in cells that should have arrested — directly exploiting the checkpoint-APC/C-separase axis described here.

Reproductive medicine and aneuploidy screening. Cohesin deteriorates with maternal age in mammalian oocytes (cohesin is loaded once, before birth, and is never replenished in the non-dividing oocyte), weakening chromosome-holding-together over decades of arrest in meiosis I. This is now understood as a major mechanistic contributor to the sharply age-related rise in human trisomies (e.g., trisomy 21), informing genetic counseling and prenatal aneuploidy screening protocols.

Congenital cohesinopathies. Mutations in cohesin loading and regulatory genes (e.g., NIPBL, causing Cornelia de Lange syndrome) illustrate that cohesin has essential roles beyond sister chromatid cohesion — including gene expression regulation via chromatin loop formation — connecting this pathway to developmental disorders.

Agricultural and plant biology. Understanding cohesin and separase regulation in crop species informs efforts to engineer altered meiotic recombination and chromosome behavior for accelerated breeding programs, including attempts to create clonal (apomictic) seed production by manipulating meiotic segregation machinery, foreshadowed by the differences covered in Meiosis Versus Mitosis.

Basic cell biology teaching and diagnostics. Understanding this switch also underlies interpretation of karyotyping and chromosome missegregation assays used in clinical cytogenetics and in studies of environmental or chemical mutagens that interfere with spindle or cohesin function, connecting directly back to concepts in The Mitotic Spindle and Its Assembly and The Spindle Assembly Checkpoint.

Ask the tutor: Why doesn't destruction of cyclin B alone (without securin destruction) trigger anaphase, and what does that tell us about the dual-lock design of separase regulation?