HISTORICAL PERSPECTIVE FOCUS ON CELL CYCLE AND DNA DAMAGE A brief history of error Andrew W. Murray The spindle checkpoint monitors chromosome alignment on the mitotic and meiotic spindle. When the checkpoint detects errors, it arrests progress of the cell cycle while it attempts to correct the mistakes. This perspective will present a brief history summarizing what we know about the checkpoint, and a list of questions we must answer before we understand it. Mutating a base pair may be regarded as misfortune to lose or gain a whole chromosome looks like carelessness. Most single base pair changes are nearly neutral, even in organisms with compact genomes, but a lost or gained chromosome has major effects. Loss of chromosomes leads to the death of haploid cells and exposure of diploid cells to previously recessive mutations. Because many genes are expressed in proportion to the number of copies of the gene, gaining or losing chromosomes leads to genetic imbalance. Genetic carelessness alters human lives. Errors in chromosome segregation in meiosis can lead to spontaneous abortion and Down syndrome (an extra copy of chromosome 21). Chromosome loss in mitosis exposes mutations that inactivate tumour suppressor genes and has a major role in the initiation and progression of cancer. Eukaryotic cells stave off genetic error by detecting and eliminating its progenitors: non-traditional base pairs and other forms of DNA damage, and chromosomes that have not lined up properly on the spindle�������the bipolar structure that segregates the chromosomes at mitosis and meiosis. When Boveri fertilized sea urchin eggs with multiple sperm, he produced spindles with three poles, showed that this led to dramatic errors in chromosome segregation in meiosis, and suggested that similar errors caused cancer1. The beginning of the 20th century saw the first clue that drugs like colchicine could affect mitosis2. Studies in the Andrew W. Murray is at Harvard University, Molecular and Cellular Biology, 52 Oxford Street, Northwest Science Building, Cambridge, Massachusetts 02138, USA. e-mail: awm@mcb.harvard.edu 1930s and 1940s showed that these compounds or cold would arrest plant3 and animal4 cells in mitosis. Later studies identified tubulin as the target of these treatments5, and showed that tubulin polymerized to form the microtubules6 that chromosomes moved along as they first attached to and then segregated on the spindle. Attachment and movement depend on the kinetochore, a complicated protein machine built on the centromeric DNA. This structure performs multiple functions: binding to microtubules, altering their behaviour, and generating signals that control the passage through mitosis and meiosis7 The 1970s and 1980s saw competition between two views of the cell division cycle8. One school, led by yeast geneticists, showed that many events in the cell cycle could not happen until the preceding event had been completed. The other, led by embryologists, argued that DNA replication and mitosis were induced by a relentless biochemical oscillator that rose and fell whether or not either process was successfully completed. The two views were reconciled by Weinert and Hartwell���s discovery of the DNA damage checkpoint9, which detects DNA damage, generates a signal that arrests the cell cycle oscillator and activates pathways that repair DNA damage. In most cells, the checkpoints can easily restrain the oscillator, but in the first divisions of a fertilized frog egg, the signal from the checkpoint is far too weak to restrain the oscillations produced by a microlitre of cytoplasm10,11. Checking for unattached chromosomes For both replication and mitosis, error- correcting mechanisms were discovered before checkpoints. In 1969, Nicklas and Koch showed that insect spermatocytes could correct chromosome alignment errors (Fig.��1)12. They produced the errors by micromanipulating the paternal and maternal copies of a chromosome so that they were attached to the same pole of the meiotic spindle. Cells detect the error because they respond to the level of tension that the chromosomes ���feel���, which is high when they are attached to opposite poles and pulling against each other, and low when they are attached to the same pole12. When tension is low, one of the chromosomes detaches from the pole, giving it a chance of achieving the correct orientation. Nicklas wondered whether the cell might wait to make sure that the errors had been corrected, but he decided that his data weren���t strong enough to answer this question. Recent experiments using individual kinetochores purified from budding yeast show that they attach more strongly to microtubules when this linkage is placed under tension13. Determining whether this direct influence of tension is the only mechanism that makes relaxed kinetochores let go of microtubules is an important subject for future research (see below). In 1990, Hoyt14 and Li15 independently produced genetic evidence for the existence of a spindle checkpoint, the control circuit that monitored the attachment and alignment of chromosomes on the mitotic spindle. Their two genetic screens identified non-overlapping genes: three BUB (budding uninhibited by benzimidazole) genes for Hoyt and three MAD (mitotic arrest deficient) genes for Li. One of these genes, BUB2, later turned out to be involved in a different checkpoint, which keeps cells from finishing anaphase and completing cell division until one of the two spindle pole 1178 NATURE CELL BIOLOGY VOLUME 13 | NUMBER 10 | OCTOBER 2011 �� 2011 Macmillan Publishers Limited. All rights reserved

FOCUS ON CELL CYCLE AND DNA DAMAGE H I S T O R I C A L P E R S P E C T I V E bodies has reached the cortex of the bud16,17. Four of the checkpoint genes, MAD1, MAD2, BUB1 and BUB3, are widely conserved and have been found in plants, fungi, animals and protozoa. The ensuing years have revealed additional checkpoint components. Two of them, the protein kinases Mps1 and Aurora B (Ipl1 in yeast), have other essential roles in the budding yeast cell cycle18,19, explaining why they were missed in the original screens. Having uncovered the checkpoint, the next important question was asking what it sensed. In 1994, Rieder et��al. showed that there was a constant lag between the attachment of the last kinetochore to microtubules and the start of chromosome segregation20. More importantly, laser ablation of the last unattached kinetochore caused cells to enter anaphase after exactly the same lag21. Experiments in meiotic cells22 and budding yeast23 showed that kinetochores that were attached to microtubules but relaxed could also activate the checkpoint. I discuss how cells sense relaxed kinetochores below. The spindle checkpoint arrests cells in mitosis or meiosis by blocking protein degradation. The anaphase-promoting complex (APC) targets mitotic cyclins and securin (the protein that restrains separase, the molecular scissors that destroy the linkage between sister chromosomes) for ubiquitylation and proteolysis, allowing anaphase to proceed. Genetic24,25 and biochemical26 evidence showed that the spindle checkpoint arrests cells by interacting with Cdc20, which is an essential cofactor for the APC. Biochemical models The spindle checkpoint has been studied by genetics, biochemistry, structural biology and microscopy on living cells. The ultimate goal is to discover how the kinetochore produces a biochemical signal that inhibits the APC. The most attractive hypothesis, from Musacchio and Salmon, revolves around the ability of Mad2 to adopt two different conformations, open and closed, and to bind Mad1 and Cdc20 (ref.��27). In this model, free Mad2 exists in the open conformation, which can���t bind Mad1 or Cdc20. The closed Mad2 conformation can bind to Mad1 or Cdc20, and to free Mad2, thus converting it from the open conformation to the closed one. The model proposes that when Mad1 is away from the kinetochore, it is bound to closed Mad2, but this Mad2 is prevented from interacting with other Mad2 molecules. When the kinetochore is not bound to microtubules, some difference in its structure allows it to recruit Mad1���Mad2 complexes and allows them to convert free molecules of Mad2 into the closed conformation, which can diffuse away and inhibit Cdc20. Somewhat confusingly, there is a second model for the biochemistry of the checkpoint. It appeals to a mitotic checkpoint complex (MCC) made up of Mad2, Mad3, Bub3, BUBR1 and Cdc20 (both BUBR1 (humans) and Mad3 (yeast) are proteins that are relatives of Bub1). This complex was discovered in extracts from mitotic cells and the first reports argued that its presence and activity were constant throughout the cell cycle, but that unattached kinetochores somehow kept the APC from recovering from its inhibition by the APC (ref.��28) later studies argue that the checkpoint regulates the assembly of the MCC (ref.��29) and that ATP hydrolysis and the activity of p31- comet are required for MCC disassembly and activation of the APC (refs. 30,31). One way of reconciling the two models is to argue that the binding of the closed form of Mad2 to Cdc20 induces Mad3/BUBR1 to bind to the Mad2��� Cdc20 complex and that the Mad3/BUBR1 brings with it Bub3. In this model (Fig.��2), conformational change at the kinetochore would allow the catalytic conversion of Mad2 from the open to the closed form, the closed Mad2 would initiate the formation of the MCC, and cooperative binding of Mad2 and Mad3/ BUBR1 to Cdc20 would be required to inhibit the APC. Unanswered questions Despite the efforts of many labs, we understand the spindle checkpoint poorly. Complete knowledge means knowing how the kinetochore senses microtubule binding and tension, how events at the kinetochore produce a soluble signal, and how this signal interferes with the APC in a manner that blocks the destruction of some of its substrates (securin and cyclin B) but not others (cyclin A). These questions raise the following more detailed questions, which start with the ones I see as most important: How does a kinetochore know it lacks microtubules? Studies using fluorescent proteins show that a number of proteins, including Mad1 and Mad2, are recruited to kinetochores that lack microtubules32 and this recruitment lies at the core of the conformational change model. But what changes at the kinetochore to allow this recruitment: is there competition between microtubules and checkpoint proteins for the same binding sites, or does microtubule binding produce conformational changes that affect the binding of checkpoint proteins at some distant site? Resolving this question will require reconstituting the checkpoint in��vitro. Although one report of an in�� vitro assay using mammalian chromosomes has been published33, using it to biochemically dissect what happens at the kinetochore will be a challenge simply because kinetochore proteins constitute a small fraction of the proteins in the assay. The ideal assay would use purified kinetochores, from an organism that is easy to genetically manipulate, activating a cocktail of checkpoint proteins to inhibit the APC, all with amounts of material that would permit biochemical analysis of individual proteins. How does the kinetochore know whether it���s under tension? The simplest possibility is that the linkage between kinetochores and microtubules is intrinsically tension-sensitive, Micromanipulation Spontaneous kinetochore detachment Artificial opposing force stabilizes attachment Grasshopper meiosis I spindle Unstable Stable Figure 1 Tension at kinetochores regulates their attachment to microtubules. Nicklas and Koch12 micromanipulated meiotic chromosomes so that the paternal and maternal chromosomes were attached to the same pole of the meiotic spindle. In the absence of opposing forces on the two kinetochores (black circles), one of the two kinetochores detaches from the pole. If it reattaches to the same pole, the cycle is repeated, but if it attaches to the opposite pole, the kinetochores experience tension again and the attachment becomes stable. Experimentally applying tension, using the micromanipulation needle, produces the same stabilization. NATURE CELL BIOLOGY VOLUME 13 | NUMBER 10 | OCTOBER 2011 1179 �� 2011 Macmillan Publishers Limited. All rights reserved