There is emerging evidence that microtubules in nondividing cells can be employed to remodel the intracellular space. This study demonstrates an essential role for microtubules in dorsal closure (DC), which occurs toward the end of Drosophila melanogaster embryogenesis. Dorsal closure is a morphogenetic process similar to wound healing, whereby a gap in the epithelium is closed through the coordinated action of different cell types. Surprisingly, this complex process requires microtubule function exclusively in epithelial cells and only for the last step, the zippering, which seals the gap. Preceding zippering, the epithelial microtubules reorganize to attain an unusual spatial distribution, which is described with subcellular resolution in the intact, living organism. This study provides a clearly defined example where cells of a developing organism transiently reorganize their microtubules to fulfill a specialized morphogenetic task (Jankovics, 2006).

Using time-lapse imaging of DC-stage embryos expressing GFP-tagged forms of tubulin or the plus end tracking protein EB1, it is shown that during DC, antiparallel MTs form stable bundles that align parallel to the D/V cell axis at the apical cell cortex. Within bundles, the MTs remain highly dynamic, which allows them to grow into the cellular protrusions that form at the dorsal surface of the dorsal-most row of epithelial cells (DME) cells. Surprisingly, elimination of these MTs by injection of MT depolymerizing drugs into the embryo or by expressing an MT-severing protein Spastin in a subset of epithelial cells inhibited exclusively the zippering process that completes DC. All other essential processes, the convergence of the two lateral epithelial cell layers, D/V cell polarization of the DMEs, or their actin-based dorsal constriction were not altered (Jankovics, 2006).

How could MT function be linked to zippering at the molecular level? One obvious scenario is that MTs are required for the local delivery of adhesion proteins. MT-based, localized delivery of factors was already shown to be crucial for proper morphogenesis in fission yeast cells and for a number of processes during early Drosophila development. However, zippering is not a simple one-step process. It is proposed to start with the interdigitation of the cell protrusions that form at the dorsal side of the DME cells and that establish the first contacts between equivalent cells of the two opposing epithelial cell layers. Consistent with this, the protrusions are essential for zippering. The initial cell-cell contacts subsequently develop into the known cell-adhesion structures. The possibility for initial interdigitation occurs at the anterior and posterior ends of the dorsal opening where the two cell layers meet. As this possibility also exists in embryos that cannot zipper due to a lack of MTs, it is conceivable that MTs are required for the interdigitation of protrusions. Intriguingly, the absence of MTs considerably affects the number and appearance of cellular protrusions known to be essential for zippering, which provides another possible explanation for the inability of these cells to interact with each other. An MT-mediated increase in protrusion formation could provide DME cells with sufficient interactive surface or interaction time to enable interdigitation. Consistent with this, the stripes of cells lacking MTs in Spastin-overexpression experiments were unable to zipper on their own but eventually managed to establish cell adhesion when forced into close proximity by zippering of the neighboring wild-type cells. To understand MT function during zippering, one may therefore need to ask how MTs modify cell protrusions. As protrusions can still form in the absence of MTs, these do not seem to control the on/off activity of the protrusion-forming machinery but may rather modulate its activity. MTs were previously shown to affect protrusion formation in several cultured cell types but the molecular mechanisms are not clear. It was speculated that MTs may modulate the actin machinery by delivering regulatory factors such as guanine nucleotide exchange factors or GTPase-activating proteins that modulate the actin organizing activities of the Rac1, Cdc42, and Rho GTPases. It is also possible that growing MTs produce pushing forces that support protrusion growth (Jankovics, 2006).

Why do MTs reorganize in such a specific way? It was not possible to answer this question. If delivery of adhesion factors or promoting protrusion formation were indeed the critical functions of MTs, then their orientation parallel to the D/V cell axis would certainly improve delivery to the relevant site. However, because of the antiparallel arrangement, transport of factors is bidirectional, and therefore the factors would also be delivered to the wrong cell ends. In addition, MTs reorganize in all epithelial cells, although most of them are not involved in zippering and therefore do not need delivery of the proposed factors. It is possible that MTs reorganize to fulfill additional, nonessential functions that optimize the DC process. For example, the MTs are required for proper epithelial cell morphology. During convergence, these cells gradually elongate along the D/V axis while thinning out in the apical-basal direction. Elongation coincides with the D/V alignment of the MTs, implying that MT reorganization may be the consequence of cell shape changes. However, cells lacking MTs cannot maintain their shape after an initial elongation phase, suggesting that proper cell morphology is dependent on MT function. Notably, this is not the consequence of defects in polarization, since cell polarity is not affected in the absence of MTs. The role of epithelial cell shape changes in DC is not clear. Their dorsalward stretching may contribute to gap closure, but recent data indicated that it is not essential. The finding that the shape abnormalities resulting from MT depletion did not affect convergence of epithelial cell layers is consistent with this view (Jankovics, 2006).

These results provide further evidence that MT reorganization is not only crucial when cells change from an interphase to a mitotic state but also when they change behavior during development. It is important to understand how such rearrangements are controlled at the molecular level and also how cell-type-specific differences are achieved. DC in Drosophila provides an excellent experimental system in which the molecular mechanisms controlling MT organization can be studied in vivo by live imaging with appropriate subcellular resolution in combination with classical genetics (Jankovics, 2006).