Fig. S1. Proper spindle organization requires correct Pav-KLP localization. (A) The anti-Pav-KLP antibody specifically recognizes the Pav-KLP polypeptide (∼100 kDa) in immunoblots. 0-2 h Drosophila embryonic high speed supernatant (HSS) (lane 1), AMPPNP supernatant (line 2) and MT pellet (lane 3) separated by SDS-7% PAGE and probed with the Pav-KLP antibody. In the presence of AMPPNP Pav-KLP is present only in the MT pellet and not in the supernatant. (B) Localization of Pav-KLP in the absence (left 2 columns) and presence (right 3 columns) of anti-PavKLP antibody. All embryos were injected with rhodamine-tubulin. In each case a large field of view is shown on the left and individual spindles are shown on the right. After Pav-KLP inhibition, spindles that retain some protein are able to complete mitosis (distal region), while spindles without detectable PavKLP become disorganized and do not complete mitosis (proximal region). Tubulin, green and Pav-KLP, red. Bars, 10 m.

Fig. S2. Effect of PavKLP inhibition on chromosome segregation. (A) Anti-PavKLP antibody injection into GFP-His expressing embryo causes a gradient of defects with the more severe ones next to the injection site (top of dotted line) and the less severe ones away from it (bottom of dotted line). In some cases, at the injection site, chromosomes did not complete segregation generating fused nuclei. Bar, 10 m. (B) A representative spindle after PavKLP inhibition in a GFP-ROD expressing embryo (previously injected with rhodamine-tubulin) shows defects in kinetochore-to-pole movement and no anaphase elongation. Control spindle on right. We observed that kinetochores moved poleward at a slower rate (0.06±0.02 m/sec in anti-PavKLP embryo versus 0.1±0.02 m/sec in control). Bar, 2 m. (ROD and His, red and Tubulin, green).

Fig. S3. The cellularization of embryos requires Pav-KLP activity. Alexa Fluor WGA, injected into the peri-vitellin space was used as a cortical membrane marker. The growing membrane incorporates the injected fluorescent WGA making it easy to follow the membrane. Kymographs, taken during cellularization along the red lines, show normal membrane growth in control (bottom), but almost no growth in anti-Pav-KLP antibody injected embryos (top). Time course about 40 minutes. Arrowheads point to the membrane front at the start (left) and at the end (right). Some marker is internalized in vesicles, which move around and lead to irregular tracks in the kymographs.

Fig. S4. Virtual Cell model of furrow growth. (A) Reactions used in the Virtual Cell model of furrow growth. (B) Schematic diagram of the geometry and boundary conditions used in the Virtual Cell model. (C) The velocity field for the MT-based vesicle transport.

Fig. S5. The model shows that vesicle deposition at the tip only does not fit the experimental data on furrow growth (orange curve). The model also shows that deposition all along the furrow length fits the later, fast phase well, but not the initial, slow phase of growth: when the furrow is too short, too few MTs ‘bump into’ the furrow and too few vesicles are delivered. In the later stage, when the furrow is very long, the MTs do not reach the basal end of the furrow and the vesicles are mostly delivered into the apical part of the furrow at almost constant rate, which explains the linear growth.

Fig. S6. Effect of changing different parameters on furrow growth. The model parameters that determine furrow growth are the speed of transport, |v|, the diffusion coefficient, D, and the attachment and detachment rates of the motor, k⁰_att and k_det, respectively. To understand the effect of these parameters on furrow growth, we performed a parameter scan by varying one parameter at a time and keeping all others as listed in supplementary material Table S1. (A) Effect of the motor velocity on furrow growth. A smaller transport speed leads to a smaller flux of attached vesicles at the furrow, and thus slower furrow growth. The furrow growth is very sensitive to the transport speed: decreasing the speed by ∼50% leads to an almost complete disappearance of furrow growth, in agreement with experimental observations in Pav-KLP inhibited embryos, i.e. no furrow growth during cellularization. The figure shows only variations of the speed down from 0.1 m/sec because upward variations had only the obvious effect of accelerating transport, and because it is unlikely that molecular motors can accelerate, while they can easily be slowed down by physical obstacles. (B) Effect of the diffusion constant on furrow growth: decreasing it has no effect, so we do not show the results. Increasing the diffusion coefficient leads to vesicles dispersing more from the MTs when they detached resulting in a smaller flux of attached vesicles and a shorter furrow. However, even for this faster random mobility, the furrow growth is not very sensitive to the diffusion: increasing the diffusion constant by a factor of 10 decreases the growth rate by ∼50% only. (C) Effect of the attachment rate on furrow growth: increasing this rate did not affect furrow growth (not shown); decreasing this rate had a small effect on furrow growth. (D) Effect of the detachment rate on furrow growth: We chose the detachment rate so that 1/k_det=L/|v| because at this detachment rate the vesicle has a significant probability of being delivered to the furrow before it detaches. Increasing the detachment rate slows down the transport tremendously, which was confirmed by our simulations (not shown). On the other hand, decreasing the detachment rate did not affect the furrow growth significantly. (E) Effect of the average MT length on furrow growth. In the realistic range of MT lengths>4 m, furrow growth is not very sensitive to the average MT length. (F) During the slow stage furrow growth can be explained by the force-limited unfolding of the wrinkled membrane.

Table S1. Model parameters.