News & Views | Published:

Cell division

Timing the machine

Nature volume 430, pages 840842 (19 August 2004) | Download Citation


During cell division everything must happen at the right time, or errors occur. A common cellular control device, protein phosphorylation, is now shown to time the assembly of a key part of the division machinery.

Cells divide and thereby multiply. This fundamental process is central to the development and survival of all organisms, and mistakes in it are responsible for a plethora of human diseases, from Down's syndrome to cancer. Accordingly, cell division — also known as mitosis — has received much attention from biologists. This attention has led to the discovery and analysis of a cycle of events that influences key regulatory proteins1, but the mechanisms by which these proteins in turn influence the machinery of mitosis are less well understood. Writing on page 908 of this issue, Mishima et al.2 help to mitigate this disparity, describing direct links between cell-cycle regulators and the cell-division machinery.

The central machine in cell division is the bipolar mitotic spindle, an apparatus that partitions the duplicated genome of a mother cell equally into two daughter cells1. The spindle is composed largely of microtubules — relatively rigid but highly dynamic tubes formed by the polymerization of tubulin proteins. Microtubules are nucleated by two microtubule-organizing centres (also called centrosomes in animal cells), one at each spindle pole (Fig. 1). Spindle microtubules, in conjunction with associated proteins, capture and separate the duplicated genome, which comes in the form of long DNA molecules known as sister chromatids.

Figure 1: The mechanics of cell division.
Figure 1

The mitotic spindle in a dividing animal cell is composed of microtubules (red lines) projecting from each of two spindle poles (also known as centrosomes; red circles). Some microtubules capture sister chromatids (blue), shown separating during the anaphase period of the cell cycle. Some microtubules interdigitate midway between the two spindle poles, where they are cross-linked through the activity of the centralspindlin complex to form the central spindle (green box). Mishima et al.2 have shown how the phosphorylation and dephosphorylation of a key component of centralspindlin ensures that the central spindle assembles at the correct time.

Spindle assembly begins during genome duplication, when the single centrosome that a cell inherits at birth also duplicates. The resulting two centrosomes migrate apart, grow, and nucleate more microtubules, which radiate out in all directions. The growing, or ‘plus’, ends of some microtubules capture pairs of sister chromatids, each sister at first remaining bound to its duplicate. Eventually, all sister chromatids are captured, with the two sisters in each pair connected to opposite poles. Subsequently, the protein-based glue between paired sisters is dissolved, and poleward forces move them apart — a stage of mitosis known as anaphase.

During anaphase a remarkable transition in spindle structure occurs. Many of the microtubules projecting from each pole do not capture chromatids, and some instead interdigitate, as their plus ends grow from opposite poles and pass each other. During anaphase, these ‘antiparallel’ microtubules are gathered into bundles, in a region of overlap midway between the two poles, forming a structure that is referred to variously as the spindle midzone, the spindle interzone or the central spindle (Fig. 1). This intriguing structure seems to be important late in mitosis, promoting the ingression of a membrane furrow that ultimately partitions the dividing cell into separate daughters — a process called cytokinesis3.

Studies of both roundworm (Caenorhabditis elegans) and mammalian cells have shown that assembly of the central spindle requires a two-protein complex dubbed centralspindlin4. One of its protein constituents is a member of the kinesin family of motor proteins, called ZEN-4 in C. elegans and MKLP1 in humans. The other is a signalling protein of the Rho family — CYK-4 in C. elegans, MgcRacGap in mammals. These two proteins form a complex that crosslinks microtubules of opposite polarity. When functional centralspindlin is lost, the central spindle does not assemble and cytokinesis is defective.

Although the requirement for centralspindlin is well established, how its crosslinking activity is restricted to anaphase, and the functional significance of this restriction, has remained unknown. Mishima et al.2 now present data suggesting that a cell-cycle regulator adds phosphate groups to (phosphorylates) centralspindlin to prevent assembly of the central spindle before anaphase. Dephosphorylation by another regulator then promotes assembly during anaphase.

Kinesins, such as ZEN-4 and MKLP1, are typically dimers that bind microtubules through two head regions, which can ‘walk’ along a single microtubule5. The head regions might also allow centralspindlin complexes to bundle microtubules, by binding to and crosslinking antiparallel microtubules. A kinesin's head regions are joined to neck regions, and Mishima et al. focus on amino acids that are found in the necks of both ZEN-4 and MKLP1 and that look like targets for phosphorylation.

The presence of these conserved potential phosphorylation sites, and the ability of the neck region to promote motor activity in vitro, compelled Mishima et al. to test the functional importance of neck phosphorylation. Their data show that a cell-cycle regulator called Cdk1–cyclin B phosphorylates the neck regions of both proteins. To examine the consequences of this phosphorylation, the authors performed both in vitro and in vivo tests. Kinesins use chemical energy obtained from the hydrolysis of adenosine triphosphate (ATP) molecules to power their movement on microtubules. Phosphorylation of the ZEN-4/MKLP1 neck greatly lowered this ATP hydrolysis activity in vitro, and reduced the affinity of the proteins for microtubules. So, as long as Cdk1–cyclin B is active, microtubule crosslinking is prevented. This would delay the assembly of the central spindle until anaphase, when Cdk1–cyclin B is known to be inactivated.

To address the functional significance of this phosphorylation in vivo, Mishima et al. expressed an altered MKLP1 in cultured mammalian cells. In this altered protein the amino acid alanine, which cannot be phosphorylated, replaces amino acids that would normally be targeted by Cdk1–cyclin B. This alteration resulted in premature bundling of microtubules before anaphase, and interfered with sister-chromatid segregation. So, neck phosphorylation does indeed seem to prevent premature assembly of the central spindle, and might thereby facilitate the proper capture and segregation of sister chromatids.

To complete their story, the authors also found that a cell-cycle phosphatase called CDC14 can dephosphorylate the ZEN-4/MKLP1 necks. In mutant C. elegans cells lacking functional CDC14, ZEN-4 failed to localize to the central spindle during anaphase, presumably because the protein's neck could not be dephosphorylated. In contrast, as would be expected if neck phosphorylation prevents central-spindle assembly, ZEN-4 that had been altered by alanine substitution to prevent neck phosphorylation did localize to the central spindle when CDC14 was not functional. Thus, by targeting the neck region of the centralspindlin kinesin, cell-cycle kinase and phosphatase activities can restrict assembly of the central spindle to anaphase.

Mishima and colleagues' extensive analysis2 documents a direct link between cell-cycle regulatory proteins and a key change in the structure of the mitotic spindle. Nevertheless, data just published by another group6 indicate that our understanding remains incomplete. Surprisingly, these investigators show that deleting the CDC14 gene in C. elegans does not result in detectable defects in assembly of the central spindle, chromosome segregation or cytokinesis. Instead, mutant worms lacking CDC14 undergo extra, non-lethal cell divisions during larval development, apparently because a protein that is unrelated to ZEN-4 cannot be dephosphorylated. At least in C. elegans, then, CDC14 is not required for cell division. One must conclude that dephosphorylation is either not necessary for cell division or can be accomplished without CDC14. No doubt we can expect further chapters in this intriguing story.


  1. 1.

    , & Nature 422, 746–752 (2003).

  2. 2.

    , , , & Nature 430, 908–913 (2004).

  3. 3.

    Annu. Rev. Cell Dev. Biol. 17, 297–300 (2001).

  4. 4.

    , & Dev. Cell 2, 41–54 (2002).

  5. 5.

    & Nature 422, 759–765 (2003).

  6. 6.

    , , , & Nature Cell Biol. doi:10.1038/ncb1154 (2004).

Download references

Author information


  1. Bruce Bowerman is at the Institute of Molecular Biology, 1370 Franklin Boulevard, University of Oregon, Eugene, Oregon 97403, USA.

    • Bruce Bowerman


  1. Search for Bruce Bowerman in:

About this article

Publication history



Further reading


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.

Newsletter Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing