Checkpoint signals in grasshopper meiosis are sensitive to microtubule attachment, but tension is still essential.

The spindle checkpoint detects errors in kinetochore attachment to microtubules and delays anaphase if attachment is improper. The checkpoint is activated by attachment-sensitive components including Mad2 and certain phosphorylated proteins detected by the 3F3/2 antibody. We have studied Mad2 and 3F3/2 immunofluorescence in grasshopper spermatocytes. As in other cells, unattached kinetochores are loaded with Mad2 and are highly phosphorylated, whereas after proper attachment, Mad2 is lost and kinetochores are dephosphorylated. What is it about proper attachment that produces these changes--is it microtubule attachment itself or is it the tension from mitotic forces that follows proper attachment? Using micromanipulation, we created an intermediate state, weak attachment, that provides an answer. Weakly attached kinetochores are not under tension and have few kinetochore microtubules. Despite the absence of tension, many weakly attached kinetochores lose their Mad2 and become dephosphorylated. Therefore we conclude that microtubule attachment determines both Mad2 binding and phosphorylation. Nevertheless, tension plays an absolutely essential role. Tension elevates the number of kinetochore microtubules to the level necessary for the complete loss of Mad2 and dephosphorylation from all kinetochores. This gives a reliable 'all clear' signal to the checkpoint, allowing the cell to progress to anaphase.

Dynein is a transient kinetochore component whose binding is regulated by microtubule attachment, not tension.

Cytoplasmic dynein is the only known kinetochore protein capable of driving chromosome movement toward spindle poles. In grasshopper spermatocytes, dynein immunofluorescence staining is bright at prometaphase kinetochores and dimmer at metaphase kinetochores. We have determined that these differences in staining intensity reflect differences in amounts of dynein associated with the kinetochore. Metaphase kinetochores regain bright dynein staining if they are detached from spindle microtubules by micromanipulation and kept detached for 10 min. We show that this increase in dynein staining is not caused by the retraction or unmasking of dynein upon detachment. Thus, dynein genuinely is a transient component of spermatocyte kinetochores. We further show that microtubule attachment, not tension, regulates dynein localization at kinetochores. Dynein binding is extremely sensitive to the presence of microtubules: fewer than half the normal number of kinetochore microtubules leads to the loss of most kinetochoric dynein. As a result, the bulk of the dynein leaves the kinetochore very early in mitosis, soon after the kinetochores begin to attach to microtubules. The possible functions of this dynein fraction are therefore limited to the initial attachment and movement of chromosomes and/or to a role in the mitotic checkpoint.

Tension on chromosomes increases the number of kinetochore microtubules but only within limits.

When chromosomes attach properly to a mitotic spindle, their kinetochores generate force in opposite directions, creating tension. Tension is presumed to increase kinetochore microtubule number, but there has been no direct evidence this is true. We micromanipulated grasshopper spermatocyte chromosomes to test this assumption and found that tension does indeed affect the number of kinetochore microtubules. Releasing tension at kinetochores causes a drop to less than half the original number of kinetochore microtubules. Restoring tension onto these depleted kinetochores restores the microtubules to their original number. However, the effects of tension are limited. Prometaphase kinetochores, when under normal tension from mitotic forces, have about half as many microtubules as they will in late metaphase. We imposed a tension force of 6 x 10(-5) dynes, three times the normal tension, on prometaphase kinetochores. The elevated tension did not drive kinetochore microtubule number above normal prometaphase values. Tension probably increases the number of kinetochore microtubules by slowing their turnover rate. The limited effect of tension at prometaphase kinetochores suggests that they have fewer microtubule binding sites than at late metaphase. The relatively few sites available in prometaphase may be the decisive sites whose binding of microtubules regulates the dynamics of transient kinetochore constituents, including checkpoint components.

The reduction of chromosome number in meiosis is determined by properties built into the chromosomes.

In meiosis I, two chromatids move to each spindle pole. Then, in meiosis II, the two are distributed, one to each future gamete. This requires that meiosis I chromosomes attach to the spindle differently than meiosis II chromosomes and that they regulate chromosome cohesion differently. We investigated whether the information that dictates the division type of the chromosome comes from the whole cell, the spindle, or the chromosome itself. Also, we determined when chromosomes can switch from meiosis I behavior to meiosis II behavior. We used a micromanipulation needle to fuse grasshopper spermatocytes in meiosis I to spermatocytes in meiosis II, and to move chromosomes from one spindle to the other. Chromosomes placed on spindles of a different meiotic division always behaved as they would have on their native spindle; e.g., a meiosis I chromosome attached to a meiosis II spindle in its normal fashion and sister chromatids moved together to the same spindle pole. We also showed that meiosis I chromosomes become competent meiosis II chromosomes in anaphase of meiosis I, but not before. The patterns for attachment to the spindle and regulation of cohesion are built into the chromosome itself. These results suggest that regulation of chromosome cohesion may be linked to differences in the arrangement of kinetochores in the two meiotic divisions.

Kinetochore "memory" of spindle checkpoint signaling in lysed mitotic cells.

The spindle checkpoint prevents errors in mitosis. Cells respond to the presence of kinetochores that are improperly attached to the mitotic spindle by delaying anaphase onset. Evidence suggests that phosphorylations recognized by the 3F3/2 anti-phosphoepitope antibody may be involved in the kinetochore signaling of the spindle checkpoint. Mitotic cells lysed in detergent in the absence of phosphatase inhibitors rapidly lose expression of the 3F3/2 phosphoepitope. However, when ATP is added to lysed and rinsed mitotic cytoskeletons, kinetochores become rephosphorylated by an endogenous, bound kinase. Kinetochore rephosphorylation in vitro produced the same differential phosphorylation seen in appropriately fixed living cells. In chromosomes not yet aligned at the metaphase plate, kinetochores undergo rapid rephosphorylation, while those of fully congressed chromosomes are under-phosphorylated. However, latent 3F3/2 kinase activity is retained at kinetochores of cells at all stages of mitosis including anaphase. This latent activity is revealed when rephosphorylation reactions are carried out for extended times. The endogenous, kinetochore-bound kinase can be chemically inactivated. Remarkably, a soluble kinase activity extracted from mitotic cells also caused differential rephosphorylation of kinetochores whose endogenous kinase had been chemically inactivated. We suggest that, in vivo, microtubule attachment alters the kinetochore 3F3/2 phosphoprotein, causing it to resist phosphorylation. This kinetochore modification is retained after cell lysis, producing a "memory" of the in vivo phosphorylation state.

Mad2 binding by phosphorylated kinetochores links error detection and checkpoint action in mitosis.

The spindle checkpoint must detect the presence of unattached or improperly attached kinetochores and must then inhibit progression through the cell cycle until the offending condition is resolved. Detection probably involves attachment-sensitive kinetochore phosphorylation (reviewed in [1,2]). A key player in the checkpoint's response is the Mad2 protein, which prevents activation of the anaphase-promoting complex (APC) by the Cdc20 protein [3-8]. Microinjection of Mad2 antibodies results in premature anaphase onset [9,10], and excess Mad2 protein causes arrest in mitosis [5,11]. We have previously shown that Mad2 localizes to unattached kinetochores in vertebrate cells, and that this localization ceases as kinetochores accumulate microtubules [10,12,13]. But how is Mad2 binding limited to unattached kinetochores? Here, we used lysed PtK1 cells to study kinetochore phosphorylation and Mad2 binding. We found that Mad2 binds to phosphorylated kinetochores, but not to unphosphorylated ones. Our data suggest that it is kinetochore protein phosphorylation that promotes Mad2 binding to unattached kinetochores. Thus, we have identified a probable molecular link between attachment-sensitive kinetochore phosphorylation and the inhibition of anaphase. The complete pathway for error control in mitosis can now be outlined.

Tension-sensitive kinetochore phosphorylation in vitro.

Many cells have a checkpoint that detects a single misattached chromosome and delays anaphase, allowing time for error correction. Detection probably depends on tension-sensitive kinetochore protein phosphorylation. Somehow, mechanical tension, or some consequence of tension, produces a chemical change, dephosphorylation. The mechanism of tension-mediated dephosphorylation can be approached using an in vitro system. Earlier work showed that the kinetochores of washed chromosomes from a mammalian cell line can be phosphorylated in vitro simply by incubation with ATP and a phosphatase inhibitor. We confirm this for chromosomes from insect meiotic cells. Thus, kinetochores of washed chromosomes from diverse sources contain a complete phosphorylation system: a kinase, a phosphatase and the substrate protein(s). We show that phosphorylation in vitro is sensitive to tension, as it is in living cells. This makes the conditions required for phosphorylation in vitro relevant to the process in living cells. The phosphatase is ruled out as the tension-sensitive component in vitro, leaving either the kinase or the substrate as the sensitive component. We show that a kinase extracted from mammalian cells in mitosis phosphorylates the kinetochores of insect meiotic chromosomes very effectively. The mammalian kinase under-phosphorylates the kinetochore of the insect's X-chromosome, just as the native insect kinase does. This provides a clue to the evolution of a chromosome that is not detected by the checkpoint. The mammalian kinase is not tightly bound to the chromosome and thus functions primarily in solution. This suggests that the substrate's phosphorylatable groups are freely available to outside constituents, e.g. regulators, as well as to the kinetochore's own kinase and phosphatase.

Tension-sensitive kinetochore phosphorylation and the chromosome distribution checkpoint in praying mantid spermatocytes.

Improper chromosome attachment to the spindle can lead to daughter cells with missing or extra chromosomes. Such mishaps are avoided in many cells by a checkpoint that detects even a single improperly attached chromosome. What is detected? A misattached chromosome is not under tension from opposed mitotic forces, and in praying mantid spermatocytes, direct experiments show that the absence of tension is what the checkpoint detects. How is the absence of tension detected? Tension-sensitive kinetochore protein phosphorylation is the most likely possibility. We combined micromanipulation with immunostaining for phosphoproteins in order to study the effect of tension on kinetochore phosphorylation in mantid spermatocytes. We confirm earlier observations on mammalian cells and grasshopper spermatocytes that misattached chromosomes have phosphorylated kinetochore proteins. We also confirm experiments in grasshopper spermatocytes showing that tension alters kinetochore chemistry: tension from a micromanipulation needle causes kinetochore protein dephosphorylation, and relaxation of tension causes kinetochore protein rephosphorylation. Beyond confirmation, our results provide fresh evidence for phosphorylation as the signal to the checkpoint. First, mantid cells are the only ones in which an effect of tension on the checkpoint has been directly demonstrated; by equally direct experiments, we now show that tension affects kinetochore phosphorylation in these same cells. Second, sex chromosome behavior in mantids provides a natural experiment to test the relationship between phosphorylation and the checkpoint. In grasshoppers, an unpaired sex chromosome is normal, its kinetochore is under-phosphorylated, and the checkpoint is not activated. In mantids, exactly the opposite is true: an unpaired sex chromosome is abnormal, its kinetochore is phosphorylated and, as predicted, the checkpoint is activated. We conclude that tension-sensitive kinetochore protein phosphorylation very likely is the essential link between proper chromosome attachment and the check-point, the link that permits potential errors in chromosome distribution to be detected and avoided.

How cells get the right chromosomes.

When cells divide, the chromosomes must be delivered flawlessly to the daughter cells. Missing or extra chromosomes can result in birth defects and cancer. Chance events are the starting point for chromosome delivery, which makes the process prone to error. Errors are avoided by diverse uses of mechanical tension from mitotic forces. Tension stabilizes the proper chromosome configuration, controls a cell cycle checkpoint, and changes chromosome chemistry.

'Anaphase' and cytokinesis in the absence of chromosomes.

Anaphase and cytokinesis are key processes in the segregation of replicated chromosomes to the daughter cells: in anaphase, chromosomes move apart; in cytokinesis, a cleavage furrow forms midway between the separated chromosomes. Some evidence suggests that chromosomes may be involved both in controlling the timing of anaphase onset and in dictating the position of the cleavage furrow. Other evidence indicates that the controlling mechanisms are intrinsic to the spindle and the cell. Here we test these possibilities in grasshopper spermatocytes by observing spindles and cells after removal of chromosomes. We found that both anaphase and cytokinesis occur independently of chromosomes: stage-specific changes occur at an appropriate time and in the correct way, despite the absence of chromosomes. This finding is particularly noteworthy because chromosomes have an important impact on spindle microtubule assembly and the timing of anaphase onset in these cells.

Chromosomes initiate spindle assembly upon experimental dissolution of the nuclear envelope in grasshopper spermatocytes.

Chromosomes are known to enhance spindle microtubule assembly in grasshopper spermatocytes, which suggested to us that chromosomes might play an essential role in the initiation of spindle formation. Chromosomes might, for example, activate other spindle components such as centrosomes and tubulin subunits upon the breakdown of the nuclear envelope. We tested this possibility in living grasshopper spermatocytes. We ruptured the nuclear envelope during prophase, which prematurely exposed the centrosomes to chromosomes and nuclear sap. Spindle assembly was promptly initiated. In contrast, assembly of the spindle was completely inhibited if the nucleus was mechanically removed from a late prophase cell. Other experiments showed that the trigger for spindle assembly is associated with the chromosomes; other constituents of the nucleus cannot initiate spindle assembly in the absence of the chromosomes. The initiation of spindle assembly required centrosomes as well as chromosomes. Extracting centrosomes from late prophase cells completely inhibited spindle assembly after dissolution of the nuclear envelope. We conclude that the normal formation of a bipolar spindle in grasshopper spermatocytes is regulated by chromosomes. A possible explanation is an activator, perhaps a chromosomal protein (Yeo, J.-P., F. Alderuccio, and B.-H. Toh. 1994a. Nature (Lond.). 367: 288-291), that promotes and stabilizes the assembly of astral microtubules and thus promotes assembly of the spindle.

Kinetochore chemistry is sensitive to tension and may link mitotic forces to a cell cycle checkpoint.

Some cells have a quality control checkpoint that can detect a single misattached chromosome and delay the onset of anaphase, thus allowing time for error correction. The mechanical error in attachment must somehow be linked to the chemical regulation of cell cycle progression. The 3F3 antibody detects phosphorylated kinetochore proteins that might serve as the required link (Gorbsky, G. J., and W. A. Ricketts. 1993. J. Cell Biol. 122:1311-1321). We show by direct micromanipulation experiments that tension alters the phosphorylation of kinetochore proteins. Tension, whether from a micromanipulation needle or from normal mitotic forces, causes dephosphorylation of the kinetochore proteins recognized by 3F3. If tension is absent, either naturally or as a result of chromosome detachment by micromanipulation, the proteins are phosphorylated. Equally direct experiments identify tension as the checkpoint signal: tension from a microneedle on a misattached chromosome leads to anaphase (Li, X., and R. B. Nicklas. 1995. Nature (Lond.). 373:630-632), and we show here that the absence of tension caused by detaching chromosomes from the spindle delays anaphase indefinitely. Thus, the absence of tension is linked to both kinetochore phosphorylation and delayed anaphase onset. We propose that the kinetochore protein dephosphorylation caused by tension is the all clear signal to the checkpoint. The evidence is circumstantial but rich. In any event, tension alters kinetochore chemistry. Very likely, tension affects chemistry directly, by altering the conformation of a tension-sensitive protein, which leads directly to dephosphorylation.

The impact of chromosomes and centrosomes on spindle assembly as observed in living cells.

We analyzed the role that chromosomes, kinetochores, and centrosomes play in spindle assembly in living grasshopper spermatocytes by reconstructing spindles lacking certain components. We used video-enhanced, polarization microscopy to distinguish the effect of each component on spindle microtubule dynamics and we discovered that both chromosomes and centrosomes make potent and very different contributions to the organization of the spindle. Remarkably, the position of a single chromosome can markedly affect the distribution of microtubules within a spindle or even alter the fate of spindle assembly. In an experimentally constructed spindle having only one chromosome, moving the chromosome to one of the two poles induces a dramatic assembly of microtubules at the nearer pole and a concomitant disassembly at the farther pole. So long as a spindle carries a single chromosome it will persist normally. A spindle will also persist even when all chromosomes are detached and then removed from the cell. If, however, a single chromosome remains in the cell but is detached from the spindle and kept in the cytoplasm, the spindle disassembles. One might expect the effect of chromosomes on spindle assembly to relate to a property of a specific site on each chromosome, perhaps the kinetochore. We have ruled out that possibility by showing that it is the size of chromosomes rather than the number of kinetochores that matters. Although chromosomes affect spindle assembly, they cannot organize a spindle in the absence of centrosomes. In contrast, centrosomes can organize a functional bipolar spindle in the absence of chromosomes. If both centrosomes and chromosomes are removed from the cell, the spindle quickly disappears.

Mitotic forces control a cell-cycle checkpoint.

Every time a cell divides, the chromosomes must be distributed accurately to the daughter cells. Errors in distribution arise if chromosomes are improperly attached to the mitotic spindle. Improper attachment is detected by a cell-cycle checkpoint in many cells and the completion of cell division is delayed, allowing time for error correction. How is an improperly attached chromosome detected? An absence of tension from mitotic forces is one possibility. Here we test this possibility directly by applying tension to an improperly attached chromosome with a micromanipulation needle. In the absence of tension, the entry into anaphase and the completion of mitosis was delayed by 5-6 hours. When the misattached chromosome was placed under tension, however, the cell entered anaphase in 56 minutes, on average. Tension from mitotic forces or from a micromanipulator's needle evidently signals to the checkpoint that all is in order and that cell division can proceed.

Elements of error correction in mitosis: microtubule capture, release, and tension.

The correction of certain errors in mitosis requires capture and release: new kinetochore microtubules must be captured and old, misdirected ones must be released. We studied capture and release in living grasshopper spermatocytes. Capture is remarkably efficient over a broad range in the angle at which a microtubule encounters a kinetochore. However, capture is inefficient when kinetochores point directly away from the source of properly directed microtubules. Capture in that situation is required for correction of the most common error; microtubule-kinetochore encounters are improbable and capture occurs only once every 8 min, on average. Release from the improper attachment caused by misdirected microtubules allows kinetochore movement and the completion of error correction. We tugged on kinetochores with a micromanipulation needle and found they are free to move less than one time in two. Thus error correction depends on two improbable events, capture and release, and they must happen by chance to coincide. In spermatocytes this will occur only once every 18 min, on average, but a leisurely cell cycle provides ample time. Capture and release generate only change, not perfection. Tension from mitotic forces brings change to a halt by stabilizing the one correct attachment of chromosomes to the spindle. We show that tension directly affects stability, rather than merely constraining kinetochore position. This implies that chromosomes are attached to the spindle by tension-sensitive linkers whose stability is necessary for proper chromosome distribution but whose loss is necessary for the correction of errors.

Odd chromosome movement and inaccurate chromosome distribution in mitosis and meiosis after treatment with protein kinase inhibitors.

Errors in chromosome orientation in mitosis and meiosis are inevitable, but normally they are quickly corrected. We find that such errors usually are not corrected in cells treated with protein kinase inhibitors. Highly inaccurate chromosome distribution is the result. When grasshopper spermatocytes were treated with the kinase inhibitor 6-dimethylaminopurine (DMAP), 84% of maloriented chromosomes failed to reorient; in anaphase, both partner chromosomes were distributed to the same daughter cell. These chromosomes were observed for a total of over 60 h, and not a single reorientation was seen. In contrast, in untreated cells, maloriented chromosomes invariably reoriented, and quickly: in 10 min, on average. A second protein kinase inhibitor, genistein, had exactly the same effect as DMAP. DMAP affected PtK1 cells in mitosis as it did spermatocytes in meiosis: improper chromosome orientations persisted, leading to frequent errors in distribution. We micromanipulated chromosomes in spermatocytes treated with DMAP to learn why maloriented chromosomes often fail to reorient. Reorientation requires the loss of improper microtubule attachments and the acquisition of new, properly directed kinetochore microtubules. Micromanipulation experiments disclose that neither the loss of old nor the acquisition of new microtubules is sufficiently affected by DMAP to account for the indefinite persistence of malorientations. Drug treatment causes a novel form of chromosome movement in which one kinetochore moves toward another kinetochore. Two kinetochores in the same chromosome or in different chromosomes can participate, producing varied, dance-like movements executed by one or two chromosomes. These kinetochore-kinetochore interactions evidently are at the expense of kinetochore-spindle interactions. We propose that malorientations persist in treated cells because the kinetochores have numerous, short microtubules with a free end that can be captured by a second kinetochore. Kinetochores capture each other's kinetochore microtubules, leaving too few sites available for the efficient capture of spindle microtubules. Since the efficient capture of spindle microtubules is essential for the correction of errors, failure of capture allows malorientations to persist. Whether the effects of DMAP actually are due to protein kinase inhibition remains to be seen. In any case, DMAP reveals interactions of one kinetochore with another, which, though ordinarily suppressed, have implications for normal mitosis.

Evolution and the meaning of metaphase.

We used an evolutionary test to ask whether the congression of chromosomes to the spindle equator is important in itself or just a mitotic happenstance. If congression matters, then it might evolve if absent initially. Previous workers established that newly made trivalents, meiotic units of three chromosomes, generally do not congress to the spindle equator. Instead, these young trivalents lie close to the pole to which two of the three chromosomes are oriented. We studied ancient sex-chromosome trivalents that arose hundreds of thousands to several million years ago in several species of praying mantids and one grasshopper. All these old trivalents lie near the spindle equator at metaphase; some of them congress as precisely to the equator as the ordinary chromosomes in the same cells. We conclude that congression evolved independently two or three times in the materials studied. Therefore, the metaphase position of chromosomes midway between the poles appears to matter, but why? In the praying mantids, the evident answer is that metaphase is a quality-control checkpoint. Sometimes the three chromosomes are not associated in a trivalent but rather are present as a bivalent plus an unpaired chromosome, which lies near one pole. Earlier workers showed that such cells are blocked in metaphase and eventually degenerate; this prevents the formation of sperm with abnormal combinations of sex chromosomes. We suggest that the quality-control system would have trouble distinguishing an unpaired chromosome from an uncongressed, newly arisen trivalent, both of which would lie near a spindle pole. If so, the confused quality-control system would block anaphase imprudently, causing a loss of cells that would have produced normal sperm. Hence, we conclude that the congression of the trivalent to the equator probably evolved along with the metaphase quality-control checkpoint. The mechanism of congression in old trivalents is uncertain, but probably involves an interesting force-sensitive regulation of the motors associated with particular chromosomes. We also examined the congression of two newly made quadrivalents when they orient with three kinetochores to one pole and one to the other. As others have described, one of these quadrivalents does not congress, while the other quadrivalent comes closer than expected to the spindle equator. Such variation in the extent of congression may provide materials on which natural selection can act, leading to the evolution of congression. The trivalents of praying mantids are attractive materials for further studies of the mechanism of congression and of the idea that metaphase is a checkpoint for progression through the cell cycle.

Orientation and segregation of a micromanipulated multivalent: familiar principles, divergent outcomes.

We studied the orientation and segregation of a particular quadrivalent in living grasshopper spermatocytes. Quadrivalents were detached from the spindle by micromanipulation, then placed and bent as desired. The detached quadrivalents reattach and orient on the spindle. Their orientation is determined by the same principles that apply to ordinary chromosomes in mitosis and meiosis, but the outcome is different. Certain characteristics of the quadrivalent lead to a variety of orientations rather than the single one typical of ordinary chromosomes. Two kinetochores in the quadrivalent are linked to the others by unusually long, flexible chromosome arms. These kinetochores may face either the same pole or opposite poles and tend to orient initially to the pole toward which they face. Consequently, the initial orientation of the flexibly linked kinetochores is variable, and, moreover, they frequently reorient. In contrast, the other two kinetochores are as rigidly connected as those in a small bivalent and so display the typical back-to-back arrangement. Usually, this arrangement leads quickly to a stable orientation of the two kinetochores to opposite poles. Sometimes, however, the back-to-back arrangement changes to a side-by-side arrangement so that the orientation of both kinetochores to the same pole is favored. The combined effect of this diverse behavior is that the quadrivalent has four stable orientations, each leading to a different distribution of chromosomes in anaphase. The result is genetic chaos. Ironically, this chaos is produced by the same mechanisms that, in ordinary bivalents and mitotic chromosomes, produce a single stable orientation and genetically appropriate chromosome distribution.

The motor for poleward chromosome movement in anaphase is in or near the kinetochore.

I have tested two contending views of chromosome-to-pole movement in anaphase. Chromosomes might be pulled poleward by a traction fiber consisting of the kinetochore microtubules and associated motors, or they might propel themselves by a motor in the kinetochore. I cut through the spindle of demembranated grasshopper spermatocytes between the chromosomes and one pole and swept the polar region away, removing a portion of the would-be traction fiber. Chromosome movement continued, and in the best examples, chromosomes moved to within 1 micron of the cut edge. There is nothing beyond the edge to support movement, and a push from the rear is unlikely because cuts in the interzone behind the separating chromosomes did not stop movement. Therefore, I conclude that the motor must be in the kinetochore or within 1 micron of it. Less conclusive evidence points to the kinetochore itself as the motor. The alternative is an external motor pulling on the kinetochore microtubules or directly on the kinetochore. A pulling motor would move kinetochore microtubules along with the chromosome, so that in a cut half-spindle, the microtubules should protrude from the cut edge as chromosomes move toward it. No protrusion was seen; however, the possibility that microtubules depolymerize as they are extruded, though unlikely, is not ruled out. What is certain is that the motor for poleward chromosome movement in anaphase must be in the kinetochore or very close to it.