The orientation of cell division is a key process that underlies epithelial morphogenesis. In polarized epithelia, the spindle can be oriented either parallel or perpendicular to the plane of the epithelium, resulting in symmetric or asymmetric cell divisions. Symmetric cell divisions underlie tissue spreading or directional tissue morphogenesis, depending on whether the spindle is randomly oriented within the epithelial plane (z axis) or assumes a fixed planar orientation. Fixed planar orientations are observed during fish and frog neurulation controlled by the PCP pathway or Cdc42, respectively (Kieserman and Wallingford, 2009; Quesada-Hernández et al., 2010
). Here, we have used the epithelium of the early frog gastrula as a model system to study the least-understood mechanism of symmetric cell divisions, when the spindle is held parallel to the plane of the epithelium but does not assume a fixed orientation within that plane. Such divisions would be important in cases where the epithelium spreads in all directions. The frog gastrula provides a good model system for this because during gastrulation it undergoes epiboly, where the epithelium spreads from the animal to the vegetal pole to cover the entire embryo.
We have shown here that in the early stages of epiboly, epithelial cells in the animal pole divide symmetrically. Spindles exhibit rapid rotation in the x/y axis until anaphase and settle in a variable direction within the epithelial plane, consistent with the requirement of the epithelium to spread in all directions. From a mechanistic point of view, we found that the spindles are positioned by balancing counteracting forces contributed by microtubules/Myo10 on the apical side and actin/myosin-2 on the basal side (Figure S5
). In the absence of one or other force, the spindle is repositioned closer to the apical or basal side, respectively, whereas in the absence of both, the spindle is positioning at highly variable points along the apicobasal axis.
Our findings move away from a simple, static, model of spindle positioning whereby spindle location is determined solely by anchoring to a specific cortical landmark, such as an adherens junction (den Elzen et al., 2009; Lu et al., 2001; Marthiens et al., 2010
), to a more dynamic system based on antagonistic forces. Interestingly, spindle positioning based on a dynamic balance of forces has also been recently reported for the meiotic spindle in mouse oocytes (Yi et al., 2011
) However, our results do not exclude a role for cell-cell junctions in spindle position. One possibility is that they act upstream of the force balancing mechanism we describe here, perhaps by providing the polarity cues necessary to set up such a mechanism. Furthermore, it is also possible that cell junctions assume increased functional importance in mature tissue, rather than in early embryonic epithelia or in cells cultured on artificial substrates.
We find that ablation of either microtubules or Myo10 causes the spindle to reposition apically, suggesting that microtubules and Myo10 provide a basally directed force to position the spindle. What mechanism can explain this? We think clues come from studying the organization of microtubules in these cells; we observe an enrichment of microtubules on the apical side of the cell. It could be that this dense microtubule network functions simply as an apical barrier, preventing the spindle from approaching the apical surface. Alternatively, and/or in addition to a barrier function, this network may exert a pushing force to actively position the spindle. Although we cannot conclusively distinguish between these two possibilities at present, our findings that spindles reposition basally in LatB treatment, MHC-B morphants, and with expression of full-length Myo10 and the microtubule-binding IQT-Myo10 strongly suggest that the microtubule network does exert a basally directed force, in addition to any barrier function. This apical microtubule network is likely to be dynamic because its organization is altered by Myo10 and MHC-B knockdown. It would be interesting to investigate how the dynamic assembly and disassembly of microtubules, which generate forces in other systems (reviewed in Dogterom et al., 2005
), contribute to the properties of this network.
Our studies show that Myo10 is required for the formation/maintenance of the apical microtubule network, with knockdown of Myo10 causing a reduction in the apical enrichment of microtubules, which is restored, along with spindle position, when rescued with full-length Myo10. Moreover, we find that the microtubule-binding tail of Myo10 is vital for spindle positioning because the GFP-IQT construct can rescue the morpholino phenotype. In some studies of Myo10 function, the GFP-IQT construct has been shown to act as a dominant negative, presumably because it can still dimerize but cannot function as an actin motor because it lacks the head domain (Cox et al., 2002; Zhang et al., 2004
). This is not the case for Myo10's function in spindle position because expression of GFP-IQT in a control background has the opposite effect on spindle position to the Myo10 MO (basal rather than apical position), and the GFP-IQT rescue of the morphant phenotype restores rather than randomizes spindle position. Together, we believe that these results reflect the microtubule rather than actin dependence of Myo10 function in spindle position. Indeed, in spindle position, Myo10 actually functions antagonistically to actin, a role that is consistent with Myo10's function in spindle structure, where F-actin and Myo10 work antagonistically to maintain mitotic spindle length (Woolner et al., 2008
). Thus, Myo10 plays crucial roles in the organization of the mitotic spindle, at several levels. A key challenge for the future will be to determine exactly how Myo10 fulfills these functions.
Our photoactivation and pharmacological perturbation experiments demonstrate that the apically directed spindle positioning force depends on actomyosin contraction. Our results are consistent with a model whereby apical actomyosin contraction provides an apically directed force to position the spindle, similar to that which occurs during spindle positioning in oocytes (Schuh and Ellenberg, 2008
). However, we also see an apically directed flow of actin filaments, which is dependent on myosin-2 activity. This presents the possibility of a second, nonmutually exclusive model, whereby the spindle is linked to this flow and carried apically.
How is the spindle connected to this flow? Astral microtubules are likely to provide the primary means by which the spindle can connect to the moving cortex. However, because we have shown that spindles move apically in Noc-treated embryos, this suggests that a further, microtubule independent, connection exists. Recent work using live imaging has revealed the presence of dynamic actin cables that reach between the cortex and the spindle both in Xenopus
embryos and mammalian cells (Fink et al., 2011; Mitsushima et al., 2010; Woolner et al., 2008
). These cables could provide a possible link between the spindle and the flowing cortex, even when astral microtubules are lost, and will therefore be an important avenue for future investigation.
The observation that actin and myosin-2 function together to position the spindle bears striking similarities to the force-generating role seen for actomyosin during centrosome separation in single cells (Rosenblatt et al., 2004
). Overall, our findings highlight the fact that the actin cytoskeleton is an active, force-generating, contributor to spindle positioning. In previous models, cortical F-actin has been thought to provide a passive substrate in which to anchor dynamic microtubules (Kunda and Baum, 2009
), whereas our model suggests a much more dynamic role. Together, these findings highlight the importance of studying the role of actin and actin-based motors during mitosis, an area that, historically, has been dominated by the study of microtubules and their motors (Kunda and Baum, 2009; Sandquist et al., 2011
What advantages might this mechanism of spindle positioning confer to epithelial tissues? First, we suggest that it allows the spindle to maintain flexibility within the epithelial plane during mitosis, such that the spindle settles in different directions during anaphase, whereas at the same time maintaining a parallel orientation. In this model, the flexibility of direction would allow the epithelium to spread in all directions, whereas the parallel orientation would maintain the epithelial organization. Indeed, in the absence of both microtubule and actomyosin forces, the spindle exhibits rapid “tumbling” movements during mitosis and fails to maintain a parallel orientation. As a consequence, the ectoderm is thickened (data not shown), and the apical ectodermal surface is reduced. Second, we speculate that the molecularly distinct nature of the forces, microtubule/Myo10 on the apical side and F-actin/myosin-2 on the basal side, may endow polarized cells with the inherent ability to vary these forces independently. Finally, we suggest that the dynamic nature of this mechanism may offer an advantage in allowing the spindle to respond rapidly to dynamic cues in the local environment, such as changes in tissue tension. This may be particularly important for marrying cell division plane with tissue shaping during rapid morphogenetic events in embryogenesis or wound healing. Our findings provide a framework of dynamic force interactions, within which some of these ideas can be further tested.