A cytokinesis furrow is positioned by twoconsecutive signalsHenrik Bringmann1& Anthony A Hyman1The position of the cytokinesis furrow in a cell determines therelative sizes of its two daughter cells as well as the distribution oftheir contents. In animal cells, the position of the cytokinesisfurrow is specified by the position of the mitotic spindle1. Thecytokinesis furrow bisects the spindle midway between the micro-tubule asters, at the s ite of the microtubule-based midzone,producing two daughter cells. Experiments in some cell typeshave suggested t hat the midzone positions the furrow2,3,butexperiments in other cells have suggested that the asters positionthe furrow4,5. One possibility is that different organisms and celltypes use different mechanisms to position the cytokinesis furrow.An alternative possibility is that both asters and the midzonecontribute to furrow positioning6,7. Recent work in C. elegans hassuggested that centrosome separation and the midzone are impli-cated in cytokinesis8. Here we examine the relative contributionsof different parts of the mitotic spindle to positioning of thecytokinesis furrow in the C. elegans zygote. By spatially separatingthe spindle midzone from one of the asters using an ultravioletlaser, we show that the cytokinesis furrow is first positioned by asignal determined by microtubule asters, and then by a secondsignal that is derived from the spindle midzone. Thus, the positionof the cytokinesis furrow is specified by two consecutive furrowingactivities.A mitotic spindle contains two structures implicated in cytokinesisfurrow positioning: asters, which are radial arrays of centrosome-nucleated microtubules, and the midzone, which forms between theseparating chromatin at anaphase and consists of non-kinetochorespindle microtubules. A mitotic spindle is an inherently symmetricstructure. A furrow-positioning cue from the asters would positionthe furrow midway between them. A midzone cue would position afurrow at the same place. In order to separate the contributions of themidzone and the asters to cytokinesis, the two structures must bespatially separated. In one-cell C. elegans embryos, the spindle formsin the middle of the cell at metaphase. At anaphase, cortical pullingforces pull on the microtubule asters, separating the two spindlepoles9.We took advantage of these pulling forces to design an experimentin which the position midway between the two asters was differentfrom the position of the spindle midzone. An embryo was observeduntil the onset of anaphase (the time at which the midzone forms10).One aster was then separated from its associated chromatin using anultraviolet laser, creating a cell with one isolated aster and the otheraster still attached to the midzone. Cortical pulling forces moved theasters to opposite poles of the cell, positioning the spindle ‘midzone’roughly one-third of the way along the aster-to-aster-distance(Fig. 1a–d and Supplementary Videos 1, 2). Thus, the position ofthe midzone was different from the position midway between the twoasters. We call this procedure asymmetric spindle severing (ASS).Following ASS, cytokinesis furrow ingression started midwaybetween the asters. However, the furrow did not complete midwaybetween the asters: a second furrow formed at the cell cortex closestto the midzone (Fig. 1e–g and Supplementary Videos 3–5). The twodistinct furrows then met and cytokinesis completed. Thus, twofurrows are observed after ASS, one between the asters and onedirected towards the midzone. Both furrows contribute to the finalposition of the cleavage plane.Using cylindrical cells, it has been shown that the mitotic appa-ratus can induce multiple furrows if it is successively displaced alongthe long axis of the cell11. A furrow can even be produced afterremoving the mitotic apparatus by aspiration before the onset offurrowing12. It could be that the spindle midzone specifies bothfurrows, one before and one after its displacement. Alternatively, theasters might specify the first furrow. To resolve this problem, wecompared the effects of separating either the anterior or the posterioraster. The isolated aster moves further towards the poles of the cellcompared with the aster that is still attached to the midzone. Thus,the position midway between the asters is different after anterior andposterior ASS. This difference is reflected in the difference ofthe position of the first furrow (see Fig. 2, the position of the firstfurrow is 53.7 ^ 0.9% along the embryo length after anterior ASSand 56.5 ^ 0.6% along the embryo length after posterior ASS;mean ^ s.e.m.; P ¼ 0.041). After ASS, the position of the first furrowis thus dependent on the position of the asters.We also shifted the first furrow using monopolar spindles, as it hasbeen shown that either an isolated aster or a monopolar spindleis sufficient to specify a cleavage plane4,13. To generate mono-polar spindles, we repeated ASS and subsequently disintegratedthe separated aster using the ultraviolet laser (Fig. 2 and Supplemen-tary Videos 6, 7). Under these conditions, the first furrow wasestablished further away from the remaining aster (ASS with intactaster, 12.8 ^ 0.2mm; ASS followed by aster disintegration,18.0 ^ 0.7mm), at a position that was not related to the midzoneposition before ASS. The cleavage plane then corrected towards themidzone. The spatial shift of the first furrow away from the intactaster demonstrates that the furrow position is not defined beforespindle severing, and is thus determined by microtubule asters.Taken together, these experiments show that the cytokinesis cleavageplane is specified by two consecutive signals derived from distinctcomponents of the mitotic spindle: the microtubule asters and thespindle midzone.Cleavage furrow formation is driven by actomyosin contraction,but little is known about the molecular mechanisms specifying whereand when contraction will occur. To learn about the molecularcontrol of redundant cleavage plane positioning, we assayed geneswith known and potential roles in cytokinesis for their effects on thetwo furrows. Genes leading to defects in cytokinesis were chosenfrom the literature and genome-wide screens (see SupplementaryTable 1). The gene products were depleted either by RNA interference(RNAi) or using established mutants, subjected to the ASS pro-cedure, and analysed for the specification of the aster-positionedLETTERS1Max Planck Institute for Molecular Cell Biology and