PerspectiveCell Biology

A Gradient Signal Orchestrates the Mitotic Spindle

See allHide authors and affiliations

Science  26 Aug 2005:
Vol. 309, Issue 5739, pp. 1334-1335
DOI: 10.1126/science.1117842

Eukaryotic cell division is entrancing when observed through a microscope. In a dramatic prelude, the internal structure of a cell is reorganized and pairs of duplicated chromosomes become arranged in the middle of the cell. Then each pair is separated into chromatids that are segregated one to each side, so that when the cell divides, each “daughter” cell receives an identical set of genes. This formidable feat is achieved by the mitotic spindle, a precision machine made from a bipolar array of microtubules that are focused at each end of the spindle by a centrosome or spindle pole body. Microtubules are themselves dynamic polymers that interact with chromosomes in the middle of the spindle and provide tracks to separate them toward the poles. Without the spindle, cell division would be impossible, and subtle defects in its function are likely to be involved with the genomic instability associated with cancer.

The mechanisms that orchestrate assembly of the mitotic spindle have been somewhat of an enigma. It has been proposed that signals emanating from chromosomes that promote microtubule growth play a key role (1). On page 1373 of this issue, Caudron et al. (2) provide the clearest evidence yet that spindle assembly is coordinated by the generation, at chromosomes, of an intracellular gradient of the active guanosine triphosphate (GTP)-bound form of Ran, a small GTPase of the Ras super-family present in all eukaryotic cells.

The concept of signaling gradients is a familiar one in animal development (3). Release of a diffusible and slowly degraded chemical, or morphogen, from a specific site can produce an extracellular concentration gradient that provides positional information to cells. The effect on a particular cell (for example, inducing differentiation) is determined by the cell's threshold in the response to the graded signal. If there are multiple thresholds, then the gradient can produce patterns of different cell responses. These may be limited to precise concentrations of the morphogen, and hence a precise position within a developing tissue. Intracellular gradients that provide positional cues can be generated through subcellular localization of mRNA, such as the localization of bicoid mRNA at the anterior pole of the Drosophila oocyte. Local translation subsequently produces a gradient of bicoid morphogen during early development (4).

Intracellular gradients can also be generated by enzyme activity. In this case, spatial information is provided by the concentration gradient of a diffusible substrate generated by a fixed enzyme. For instance, if phosphorylation of a diffusible protein is catalyzed by a localized protein kinase, and if the opposing protein phosphatase is dispersed, then a gradient in the phosphorylation status (and therefore in the activity of the substrate protein) can be generated (5). Such a reaction-diffusion process creates a steadily changing gradient that is distinct from sharp concentration differences due to compartmentalization, such as the difference in ion concentration across an impermeable membrane. Previous work has shown, by mathematical simulation combined with molecular experiments using fluorescence reporters, that such a chemical gradient can be generated by phosphorylation of the microtubule-binding protein stathmin (6).

Now, Caudron et al. (2) have used a similar dual approach to examine the gradient of Ran-GTP generated by mitotic chromosomes in Xenopus egg extracts (a system commonly used to study mitosis). Ran-GTP releases factors that are required for spindle assembly from inhibitory complexes that also contain proteins involved in nuclear transport called importins or karyopherins. The gradient of Ran-GTP directed away from chromosomes is proposed to provide a positional signal that causes changes in microtubule dynamics and organizes the spindle around chromosomes. The key assumptions of this model, based on good experimental evidence, are that RCC1 (the nucleotide exchange factor that generates Ran-GTP) is localized predominantly to chromosomes, while the hydrolysis of diffusible Ran-GTP is catalyzed by a largely dispersed GTPase-activating protein, RanGAP1, thereby forming a reaction-diffusion system that can be described mathematically.

The use of mathematical simulation to describe a process is sometimes viewed with skepticism by molecular and cellular biologists who regard such models as drastically oversimplified in terms of the molecules involved and their functions, and lacking in predictive power that would help design further experiments to test the hypothesis. Nonetheless, even skeptics can appreciate the use of mathematics to provide a clear framework on which to organize ideas and formulate general principles. Building on previous simulations (7) and experimental evidence (8), Caudron et al. confirm that a steep gradient of free Ran-GTP is generated around chromosomes in Xenopus egg extracts. In addition, they demonstrate the generation of a longer range gradient of Ran-GTP that is in a complex with importin-β. It is this gradient that the authors suggest is more informative about the concentration of released spindle assembly factors.

Caudron et al. show that microtubule nucleation is highly dependent on a threshold level of Ran-GTP-importin-β being exceeded and that this only occurs in the vicinity of chromosomes. On the other hand, stabilization of the dynamic “plus” (growing) ends of microtubules responds proportionately to Ran-GTP-importin-β concentration and is extended over longer distances, as far as the centrosomes (see the figure). Varying the Ran-GTP-importin-β gradient experimentally by adding regulators of Ran-GTP production or stability disrupts proper bipolar spindle assembly around chromosomes. The inference is not only that the generation of Ran-GTP is required, but that the correct gradient of concentrations is important to provide the proper spatial cues for different reactions during the assembly of the spindle.

One clear prediction of this signaling gradient model is that factors involved in microtubule nucleation or plus-end stabilization will differ in their interaction with importin-β (or that they are complexed to different members of the importin/karyopherin family) and are released at different threshold concentrations of Ran-GTP. This is analogous to thresholds in the response of cells to morphogens. Because microtubule interactions with chromosomes are disrupted by altering the gradient, Caudron et al. propose that the stabilization of microtubules at some distance from chromosomes promotes their growth toward the chromosomes, and this promotes their “capture” by microtubules.

Chromosomes generate a chemical gradient that organizes the mitotic spindle during cell division.

Microtubule nucleation occurs only within a region close to the chromosomes, as a result of an ultrasensitive response to a gradient of Ran-GTP-importin-β. Microtubule stabilization works at greater distances from the chromosomes and responds linearly to the gradient. These responses may reflect differing sensitivities of effectors in each of the processes to Ran-GTP, so that spatial information is provided by the gradient. Half of the assembling spindle is represented.


In an alternative “search-and-capture” model for spindle assembly, stochastic microtubule growth from centrosomes would be stabilized by interactions with specific protein complexes, particularly at kinetochores [structures on chromosomes that are important for chromosome movement toward the spindle poles and to which microtubules are attached (9)]. In contrast to the signaling gradient model, search-and-capture does not require longer range effects of chromosomes on microtubule growth. However, chromatin lacking kinetochore regions can organize spindle assembly in the absence of centrosomes in Xenopus egg extracts (10), and theoretical modeling indicates that the search-and-capture process alone would be too slow (11). Even so, there is recent evidence that, in addition to localization of RCC1 across mitotic chromosomes, a subset of RanGAP1 molecules together with a Ran binding protein called RanBP2 are localized to kinetochores through the interaction of Ran-GTP with the karyopherin Crm1 (12). These localized molecules are necessarily not considered in the simplified model used by Caudron et al., yet they appear to have an important function in kinetochore attachment of microtubules. Indeed, it is likely that Ran has a fundamental role in the interactions of microtubules and chromosomes that is conserved in eukaryotes (13).

The signaling gradient model of Caudron et al. may therefore not be the complete picture. Both the effects of a widespread Ran-GTP gradient and specific localized functions of Ran-GTP could be incorporated in a biased search-and-capture mechanism (11) in which the Ran- GTP gradient permits release of factors that stabilize or nucleate microtubules while the interactions of microtubules with kinetochores are controlled by localized complexes dependent on Ran-GTP. And although Xenopus egg extracts are a very valuable model for mitotic spindle assembly, it remains to be seen if the gradient mechanism is necessary in cells where there are other spatial constraints on spindle assembly and orientation. Nonetheless, the work of Caudron et al. provides the clearest portrait yet of how Ran orchestrates the overall process of mitotic spindle assembly, and it illustrates very nicely how an intracellular chemical gradient can provide spatial information. It will be interesting to see whether this is a general principle for the organization of structures within cells.


View Abstract

Stay Connected to Science

Navigate This Article