Co-movement of astral microtubules, organelles and F-actin suggests aster positioning by surface forces in frog eggs

How bulk cytoplasm generates forces to separate post-anaphase microtubule (MT) asters in Xenopus laevis and other large eggs remains unclear. Previous models proposed dynein-based organelle transport generates length-dependent forces on astral MTs that pull centrosomes through the cytoplasm, away from the midplane. In Xenopus egg extracts, we co-imaged MTs, endoplasmic reticulum (ER), mitochondria, acidic organelles, F-actin, keratin, and fluorescein in moving and stationary asters. In asters that were moving in response to dynein and actomyosin forces, we observed that all cytoplasmic components moved together, i.e., as a continuum. Dynein-mediated organelle transport was restricted by interior MTs and F-actin. Organelles exhibited a burst of dynein-dependent inward movement at the growing aster surface, then mostly halted inside the aster. Dynein-coated beads were slowed by F-actin, but in contrast to organelles, beads did not halt inside asters. These observations call for new models of aster positioning based on surface forces and internal stresses.


Introduction
Cytokinesis requires drastic reorganization of the cell, and it provides a model for probing cytoplasmic 12 mechanics and principles of sub-cellular organization. Here, we focus on organization of the cytoplasm by 13 MT asters that establish the cleavage plane before furrow ingression in Xenopus laevis eggs. Due to their 14 ~1 mm diameter, these cells provide a system where organization of bulk cytoplasm by MT asters is 15 particularly clear. The first mitotic spindle is centrally located, and much smaller than the egg. After 16 mitosis, a pair of MT asters grow out from the centrosomes, reaching the cortex ~20 min later. These 17 asters are composed of a branched network of short, dynamic MTs ~15 µm long and oriented 18 approximately radially, with plus ends outward (Ishihara et al., 2016;Ishihara et al., 2014). 19 20 Asters have two important organizational and mechanical functions. The first function is to position 21 cleavage furrows. Where the paired asters meet, at the midzone of the egg, the MTs form an antiparallel 22 interaction zone which recruits the chromosomal passenger complex (CPC) and centralspindlin (Field et 23 al., 2015), which define the cleavage plane (Basant et al., 2018;Carmena et al., 2012). The second function 24 of MT asters, and the focus of this paper, is to move centrosomes and nuclei away from the future 25 cleavage plane, so each daughter blastomere inherits one of each. This separation movement transports 26 centrosomes and nuclei hundreds of microns away from the midplane over tens of minutes. Aster growth 27 and centrosome separation movement after anaphase of first mitosis are illustrated in Figs 1A-C. In 28 common with other authors, we often refer to centrosome and aster movement together for simplicity. 29 The reality is more complex due to continuous MT growth and turnover. As centrosomes move away from 30 the midplane, the aster surface adjacent to the midplane remains stationary, while the outer aster surface 31 grows outwards due to a combination of MT polymerization and outward sliding. If the aster were rigid, 32 surface MTs and centrosomes would slide outward at the same rate, but since the aster is better 33 considered a deformable network, this is not a safe assumption. 34

35
The forces that act on MTs to move centrosomes and asters have been extensively investigated and 36 reviewed, and differ between systems (Garzon-Coral et al., 2016;Grill et al., 2005;Kotak et al., 2013;37 Meaders et al., 2020;Xie et al., 2020). In Xenopus and zebrafish zygotes, which are unusually large cells, 38 Wühr and colleagues showed that movement away from the midplane after first mitosis is driven by 39 dynein-dependent pulling forces (Wühr et al., 2010). Since movement occurs before astral MTs reach the 40 cortex, the dynein must be localized throughout the cytoplasm, presumably attached to organelles, but 41 this was not tested. The most prominent model for centrosome movement of this kind proposes that 42 dynein attached to organelles generates pulling forces that increase with MT length, which can potentially 43 account for the force asymmetries required for directed movement (Hamaguchi et al., 1986;Kimura et 44 al., 2011;Tanimoto et al., 2016;Tanimoto et al., 2018;Wühr et al., 2010). In this "length-dependent 45 pulling" model, dynein transports organelles along astral MTs toward the centrosome, then viscous or 46 elastic drag on the organelles imparts a counter force on the MTs, pulling them away from the 47 centrosome. The flux of organelles, and thus the net pulling force, is thought to scale with MT length. 48 Although length-dependent pulling models are widely discussed, many aspects remain unclear. Net forces 49 may not scale with MT length due to hydrodynamic interactions between MTs, which computational 50 studies have shown are significant (Nazockdast et al., 2017). The organelles that anchor dynein in the 51 cytoplasm of large egg cells have not been fully identified and the spatiotemporal distribution of organelle 52 transport has not been mapped. Candidate dynein anchor organelles include the ER, which moves inwards 53 as sperm asters center in sea urchin (Terasaki et al., 1991), acidic organelles which were implicated in 54 nematode eggs (Kimura et al., 2011), and mitochondria which are abundant in early embryos. 55

Centrosome separation and ER distribution in fixed eggs 82
As a first test of how centrosomes and organelles move relative to one another, we fixed frog eggs before 83 first cleavage, stained for tubulin and ER, and imaged by confocal microscopy (Figs 1A-C). Centrosome 84 separation movement is represented by the diagonal lines connecting different eggs in panels B and C. As 85 centrosomes move away from the midplane, the centrioles within them replicate and split, visible as the 86 pair of bright cyan spots within each aster in Fig 1C. We probed the ER distribution by staining for the ER 87 membrane marker Lunapark (LNPK) (Figs 1B',C') and also the luminal marker PDIA3 (not shown) with 88 similar results. The ER was distributed all over the asters, with some enrichment near centrosomes and 89 the cortex. Lack of strong ER enrichment at centrosomes called into question the length-dependent 90 pulling model with the ER as a dynein anchor. However, organelle transport dynamics could not be 91 measured from fixed images, so we turned to an egg extract system for live imaging. 92 (A) Cartoon illustrating MTOC movement away from the CPC-positive midplane before astral microtubules (MTs) reach the cortex in Xenopus laevis eggs. MTs shown in cyan and CPC-positive interaction zone in magenta. Note the CPC is shown in the cartoon panels A and D, but not in the rest of the figure. (B,C) Anti-tubulin immunofluorescence of eggs fixed ~70 and ~85 min post fertilization (pf). Diagonal lines connecting different eggs in panels B and C emphasize centrosome separation movement and growing aster surfaces. (B',C') Anti-LNPK (ER) immunofluorescence of the same eggs. (D) Cartoon illustrating aster separation movement in an extract system. MTs and CPC as in panel A. Asters were reconstituted from artificial microtubule organizing centers (MTOCs) in interphase Xenopus egg extracts. (E,F) MTOCs moved apart as asters grew and interacted with one another over time. Time is defined with respect to perfusing the sample and warming to 20 °C, so the start of aster growth occurred soon after 0 min. (F') A fraction of the ER became enriched around MTOCs, and (F'') F-actin was disassembled locally along interaction zones.

Microtubule organizing center (MTOC) separation movement in egg extract by dynein and actomyosin 93
To model separation movement in a cell free system suitable for live imaging, we filled chambers 94 consisting of two PEG-passivated coverslips spaced ~20 µm apart with actin-intact interphase egg extract 95 containing artificial MTOCs, imaging probes, and drugs. MTOCs nucleated astral MTs. We then imaged 96 aster growth and MTOC movement over ~30 min. For most experiments we used widefield microscopy 97 with a 20x objective to collect data on overall organization and flows, in some cases stitching multiple 98 image fields. To illustrate structural details of the components we studied, Fig S1 and  When asters grew to touch each other, they formed CPC-positive interaction zones (Fig 2A) as previously 107 reported . These reconstitute the CPC-positive zone that forms between asters in 108 eggs (Field et al., 2015). CPC-positive zones cause local disassembly of both MTs and F-actin, which locally 109 softens the cytoplasm (Field et al., 2019), and generates strong anisotropy in both MT and F-actin density. 110 These anisotropies may lead to generation of directed forces on MTOCs by both dynein and actomyosin. 111 Previous work focused on possible consequences of MT length anisotropy on dynein forces (see 112 Introduction). Here, we will focus on consequences of locally softening the cytoplasm at CPC-positive 113 zones on the mechanical response of asters to forces from both dynein and actomyosin. 114

115
To quantify MTOC movement, and determine the role of forces from different motors, we picked random 116 locations and imaged large fields over time in up to four conditions in parallel. Fig 2B and SI Movie 3 show 117 a typical experiment, where only the CPC channel is shown for simplicity. At early times points, the spatial 118 distribution of MTOCs was random and the CPC signal was diffuse, except some signal on the MTOCs. As 119 asters grew and interacted, they recruited CPC to zones between them under all conditions. We quantified 120 MTOC movements with respect to their nearest neighbors, which were defined by the Delaunay 121 triangulation between MTOCs at the earliest time point and followed over the movie (Fig 2B). Red edges 122 indicate when neighboring MTOCs formed a CPC-positive zone between them, and blue edges indicate 123 when they did not. We then measured the maximum separation speed as a function of the initial 124 separation distance between the MTOCs. 125 126 Under control conditions, MTOCs that were initially closer together tended to move farther apart, while 127 those initially farther apart tended to move closer together, as evidenced by the strong negative 128 correlation between the maximum speed of separation movement and starting distance (Fig 2C). We 129 focused on separation movement of MTOCs in separate asters with a CPC-positive zone between them 130 (red points), since this models centrosome separation movement in eggs. 131

132
To test the role of dynein and actomyosin in MTOC movement, we inhibited dynein or fragmented F-actin, 133 separately or together. Inhibiting only one motile system caused a partial block. The role of each motile 134 system was similar, as judged by effects on the slopes of separation speed vs initial distance plots ( Fig 2C). 135 We investigate sites of dynein-based pulling below. We hypothesize actomyosin-based separation 136 movement is driven by actomyosin contraction away from regions of lower F-actin density along 137 interaction zones, and will analyze this model in detail elsewhere. Inhibiting both motile systems 138 completely blocked MTOC movement. Inhibiting CPC recruitment with an AURKB inhibitor also completely 139 blocked MTOC movement (not shown). These findings were qualitatively confirmed by visual inspection 140 and partial analysis of more than 10 experiments using multiple extracts. We interpret these data as 141 showing that MTOC movement in our extract system is driven by a combination of dynein and actomyosin 142 forces. We next focused on analysis of co-movement of astral MTs, organelles and F-actin during aster 143 separation movement. 144 The CPC localized to interaction zones between neighboring asters, blocking mutual interpenetration of MTs and disassembling F-actin locally. Time is defined with respect to perfusing the sample and warming to 20 °C, so the start of aster growth occurred soon after 0 min. (B) Four aster growth reactions were followed in parallel under control vs inhibitor conditions. The first column in each condition shows an early time point, and the second column shows a time point 30 min later. MT growth was similar and CPC-positive zones formed under all conditions (see SI Movie 3). (C) Maximum speed of separation with respect to initial distance between the MTOCs.

ER and F-actin move with MTs in separating asters 145
MTOC separation trajectories were longest, and most unidirectional, when MTOCs were clustered at the 146 initial time point. In these cases, MTOCs moved predictably outwards from the cluster as asters grew out 147 and interacted ( at any given location. Typically, the region near the interaction zones moved ~20% faster than the MTOC, 176 and the leading edge of each aster moved ~20% slower. These PIV-based velocities reflect polymer 177 translocation rather than polymerization, so this spatial variation in velocity shows that the aster does not 178 move as a completely rigid body. Rather, it deforms as a gel, compressing or stretching in response to 179 forces and stresses. Disassembly of F-actin softens the interaction zone, facilitating aster separation 180 movement. 181 . MTOC trajectories are represented by contours colored from blue to yellow. Time is defined with respect to perfusing the sample and warming to 20 °C, so the start of aster growth occurred soon after 0 min. (C) Intensity kymographs along the grey line shown in panels A and B, passing through the MTOCs marked with a white star. To show relative movement of the MTOCs, each row of the kymograph was computationally translated to keep stationary the midpoint between the MTOCs, where the interaction zone formed. Solid curves indicate the MTOCs, the dashed curve indicates the growing aster surface, and the dash-dotted line indicates the interaction zone. (D) Velocity kymographs in the same frame of reference as in panel C. 2D flow fields were measured by particle image velocimetry (PIV), projected onto the line passing through the MTOCs, then the projected velocity of the midpoint between the MTOCs was subtracted, again to show movement relative to the interaction zone. A white color indicates stationary with respect to the midpoint, blue indicates moving to the left, and red to the right. PIV outliers were filtered and shown in beige. (E) Velocity of the MTOCs based on particle tracking, as well as the velocity of ER and F-actin in the neighborhood of the MTOCs based on PIV. (F) Velocity of ER with respect to the moving MTOCs, not with respect to the interaction zone as in panel D.

ER and F-actin move with MTs on coverslips functionalized with dynein 182
To generate a complementary system for dynein-dependent MTOC movement where actomyosin was 183 less important, we artificially anchored dynein to the coverslip via a biologically relevant linkage. This also 184 had the effect of speeding up movement ~10 fold compared to movement away from CPC-positive zones. 185 Endogenous HOOK2, a coiled-coil dynein-dynactin adapter (Reck-Peterson et al., 2018), was recruited to 186 PEG-passivated coverslips via an antibody raised to its C-terminus (Methods). To characterize the antibody 187 and identify HOOK2 interacting proteins, we performed quantitative immunoprecipitation-mass 188 spectrometry (IP-MS) ( Fig S2). We compared 3 conditions: anti-HOOK2 in interphase extracts (3 separate 189 extract repeats), anti-HOOK2 in mitotic extracts (2 repeats), and as negative control, random IgG in 190 interphase extracts (3 repeats). HOOK2 was the most abundant protein recovered on anti-HOOK2 beads. 191 HOOK3 was also detected, consistent with heterodimerization between HOOK family members (Redwine 192 et al., 2017;Xu et al., 2008). In interphase extracts, anti-HOOK2 pulled down multiple subunits of the 193 dynein-dynactin complex, plus known interactors LIS1 and CLIP1. All these dynein-related proteins were 194 greatly reduced in pulldowns from mitotic extracts, suggesting the interaction between HOOK2 and 195 dynein-dynactin is negatively regulated by CDK1 activity. We also noted many potential dynein-dynactin 196 interacting proteins that have not been confirmed by other methods (data available on request). We 197 concluded the HOOK2 antibody offers a physiological linkage to dynein, and we proceeded to test its 198 effects on aster movement. 199 200 Dynein attached to coverslips via HOOK2 generated pulling forces on MTs directed away from the MTOC 201 ( Fig 4A). We previously reported that dynein non-specifically adsorbed to non-passivated coverslips 202 increases the rate of aster growth due to outwards microtubule sliding, but did not move MTOCs (Ishihara 203 et al., 2014). Remarkably, on HOOK2-functionalized coverslips, asters exhibited rapid translational 204 movement in a circular pattern with a diameter of 20-30 µm (Figs 4B,C, SI Movie 4). During this movement, 205 MTOCs moved continuously at ~1 µm/s, comparable to the maximum speed of dynein (Reck-Peterson et 206 al., 2018) ( Fig 4D). This 2D-oscillatory movement was observed in >10 different experiments using 207 different batches of extract, and was blocked by dynein inhibition. We plan to investigate the mechanism 208 of the instability that causes circular motion elsewhere. Here, we used the rapid aster movement as an 209 alternative system to study how ER and F-actin move with respect to moving MTs. Fig 4E shows  continuum. In another experiment, keratin was also advected with moving asters (SI Movie 5). From these 216 observations we conclude that cytoplasmic networks are mechanically integrated inside asters, and 217 cytoplasmic networks move together with moving asters. 218

A small molecule probe is advected with moving asters 219
The high speed and predictability of the oscillatory movement on HOOK2-functionalized coverslips 220 enabled us to ask whether the cytosol was advected with the moving cytoplasmic networks. This question 221 was inspired by recent experiments showing that moving actomyosin gels advect cytosol in Drosophila 222 embryos (Deneke et al., 2019). We functionalized artificial MTOCs with caged fluorescein, linked to the 223 MTOCs via the caging group ( Fig 5A). The fluorescein was uncaged upon shining 395 nm light, 224 simultaneously activating its fluorescence and releasing it from the MTOCs (Figs 5A-C, SI Movie 6). The 225 cloud of photo-released fluorescein dispersed within tens of seconds ( Fig 5D). Rapid diffusive spread of 226 the cloud validated that the fluorescein behaves as a freely diffusing small molecule ( Fig 5E) and enabled 227 estimation of the viscosity of the cytosol at ~6x that of water (Methods), consistent with previous 228 estimates ( Luby-Phelps, 1999;Valentine et al., 2005). We then fit the fluorescein cloud with a 2D Gaussian 229 to track its center of mass. The center of brightness of the diffusing fluorescein cloud was clearly advected 230 with the MTOC (Fig 5F), showing that cytosol advects with moving asters due to hydrodynamic 231 interactions inside asters. Advection of cytosol with moving asters is consistent with poroelastic behavior 232 and places an upper bound of ~100 nm on the effective pore size of cytoplasmic networks in this system 233 (Methods) (Mitchison et al., 2008;Moeendarbary et al., 2013). Similar evidence of advection was found 234 in >10 experiments in 3 extracts. The cloud center and the MTOC did not precisely co-align. This could be 235 due to tracking error, but we suspect that bulk liquid flow driven by forces outside the aster may permeate 236 asters and move the diffusing cloud. 237

Dynein-mediated organelle movement is restricted by F-actin and interior MTs 238
We next investigated which organelles in egg extract recruit dynein, and how they might exert force on 239 asters in a continuum model for aster movement. To facilitate detailed analysis of organelle transport 240 throughout the aster, we imaged isolated asters which remained stationary as they grew. ER and 241 mitochondria are the most abundant organelles in Xenopus egg extracts based on proteomics (Wühr et 242 al., 2014), and acidic organelles were implicated in centrosome movement in C. elegans embryos (Kimura 243 et al., 2011). 244

245
In control extracts with F-actin intact, almost all the ER, mitochondria, and acidic organelles remained 246 evenly distributed over asters as they grew, and a small fraction of the ER accumulated near MTOCs (Figs 247 6A, 7A,D, SI Movies 7,8). The ER intensity around MTOCs increased to ~2 fold higher than the intensity 248 outside the aster (Fig 6A') in >5 examples scored. Although the majority of ER remained stationary, astral 249 MTs did induce a subtle change in the texture of the ER, from coarser outside the aster, to finer and more 250 tubular in appearance inside the aster (SI Movies 7,8). Astral MTs also affected the structure of the F-actin 251 network, from random orientation of filaments outside the aster, to radial alignment of a subpopulation 252 of bundles inside the aster ( Fig 7A) as we reported previously (Field et al., 2019). 253 254 When F-actin was fragmented with Cytochalasin D, all organelles exhibited inward movement (Figs 6B, 255 7G, Fig S3), which was fastest at the growing surface of the aster (Fig 7). Compared to control, a greater 256 fraction of the ER was transported inwards (Fig 6B'), and average transport speeds were an order of 257 magnitude faster with F-actin fragmented than intact (Fig 7). The ER intensity around MTOCs accumulated 258 to ~6 fold higher than the intensity outside the aster and continued to increase with time (Fig 6B'). Due to 259 the burst of movement at the surface of the growing aster, the intensity of organelles was ~30% lower 260 there than outside the aster (Fig 6B'). Compared to control, the texture of the ER was coarser when F-261 actin was fragmented, both inside and outside asters, and MTs appeared more bundled. Mitochondria 262 and acidic organelles moved inwards and accumulated near the MTOC. These organelles appeared to 263 physically associate with ER in higher magnification images (Figs 6B,C, 7G, S3), so all organelles may be 264 physically connected in this system. These findings show that the ER, and perhaps all organelles, recruit 265 dynein, and can move toward the MTOC. Inward movement is restrained by F-actin under control 266 conditions. However, even with F-actin fragmented, the majority of the ER, mitochondria, and acidic 267 organelles were still evenly distributed over the aster. 268 269 We next added CC1 to test for a role of dynein in organelle transport (Fig 6C). We did this with and without 270 Cytochalasin D to fragment F-actin, but only report results with F-actin fragmented because inward 271 movement is much easier to score. With CC1 present, with or without F-actin, organelles moved neither 272 inwards nor outwards, and did not accumulate at MTOCs. This suggests that dynein generates all the 273 inward force on organelles, and that other forces on organelles do not induce significant net transport in 274 this system. Tip-tracking and kinesin forces might escape detection in 20x movies if they only cause local 275 movements, but they did not make a major contribution to overall organelle distributions. 276 Figure 6. Dynein-mediated organelle movement is restricted by F-actin. (A) In control with intact F-actin, a small amount of ER became concentrated around the MTOC, but the majority of the ER and mitochondria remained distributed over the aster (see SI Movie 7). The white arc indicates the growing aster surface, and the box indicates the zoomed region in the lower panels. (A') Average intensity with respect to distance from the MTOC over time, from black to gray. (B) When F-actin was fragmented with Cytochalasin D, a greater fraction of the ER was transported toward the MTOC, and a fraction of mitochondria was transported as well. Higher magnification: ER started to move when MTs indicated by growing +TIPs first grew into the cytoplasm, and ER and mitochondria co-localized with one another. (C) When dynein was inhibited with CC1, the ER was not transported, neither toward nor away from the MTOC.

Dynein-mediated organelle movement is maximal near the aster surface 277
To infer outward forces on MTs as a function of time and location, we needed a measure of the total 278 inward organelle flux. Kymographs and PIV provide direct visualization of movement but have limitations 279 for this inference, because they measure movement of local gradients in fluorescence intensity, not mass 280 transport of organelles. We therefore developed an analysis to infer mass transport of organelles based 281 on flux of fluorescence intensity (analysis described in Fig S4 and Methods). This analysis quantifies the 282 amount of fluorescence signal crossing a given circumference at a given time, normalized by the total 283 fluorescence in a region containing the aster. Inward organelle transport from this analysis can be 284 qualitatively related to outward force on MTs (Discussion). Fig 7 shows examples with ER and acidic 285 organelles. Mitochondria exhibited similar movement as acidic organelles (Fig S3). 286 287 All analysis methods revealed a burst of inward organelle movement when the growing aster surface 288 reached them, followed by slowing down inside asters (SI Movies 8-10). This burst can be visualized as 289 inward diagonal features in kymographs, and high values on the surface diagonal in mass transport and 290 PIV plots. Under control conditions, with F-actin intact, the amount of organelle movement at the aster 291 surface was variable between extracts. Out of 11 extract preps, we observed a burst of inward ER 292 movement at the aster surface in 7 extracts (64%) as in Fig 7C, and observed weaker or no burst in the 293 remaining extracts as in Fig 7F. Factors that seem to lessen the burst of inward movement include higher 294 concentrations of spontaneously nucleated MTs outside the aster, and insufficient passivation of the 295 coverslips, but these two factors do not explain all the examples that did not exhibit a burst. Lack of fast 296 organelle movement in control asters with intact F-actin is consistent with co-movement of cytoplasmic 297 networks in moving asters (Figs 3 and 4). 298 299 When F-actin was fragmented to highlight interactions between organelles and MTs, organelle transport 300 at the growing aster surface was faster, and therefore easier to visualize and quantify. The burst of inward 301 movement near the aster surface was highly reproducible when F-actin was fragmented with Cytochalasin 302 D. Velocity values for ER moving inwards at the aster surface reached ~0.25 µm/s with F-actin fragmented 303 (Fig 7J), and mass transport reached 2% per min (Fig 7I). Mass transport values were more peaked at the 304 surface than PIV values, in part because mass transport takes into account the increase in circumference 305 as the radius increases. However, the PIV values were peaked at the surface, as well, so organelles moved 306 faster near the surface then slower once incorporated into asters. The peak in velocity at the aster surface 307 was highly reproducible with F-actin fragmented and observed in >10 experiments with separate extracts. 308 A smaller fraction of acidic organelles than ER were transported inwards (Figs 7F,I), but with a similar bias 309 toward movement at the surface. The restriction of organelle transport to the aster surface, even with F-310 actin fragmented, suggests that forces from dynein are exerted primarily at the growing aster surface. 311 Figure 7. Dynein-mediated organelle movement is maximal on the aster surface. (A) Stationary asters were grown from isolated MTOCs. The growing aster surface is indicated by a white arc, and the ER was largely distributed but slightly depleted just inside the growing aster surface. The ER exhibited a change in texture from slightly coarser outside the aster to finer inside the aster (see SI Movie 8). (B) Kymographs along a line extending away from the MTOC. The MTOC corresponds to the left column, and the growing aster surface corresponds to the diagonal line where soluble tubulin is depleted upon incorporation into the growing aster. (C) Mass transport map for ER averaged over a quadrant, in the same frame of reference as the kymographs in panel B. Mass transport analysis is described in Fig S4. (D-F) Similar experiment with F-actin intact, in a different batch of extract that exhibited less organelle movement (see SI Movie 9). (G-J) Similar experiment with F-actin fragmented by Cytochalasin D (see SI Movie 10). (J) Average speed based on PIV, in the same frame of reference as panels H,I and averaged over a quadrant. PIV is not shown for control because movement was too slow to be reliably quantified.

Dynein-coated beads move inwards at constant rates throughout asters 312
Slowing of organelle transport upon incorporation into the aster, whether F-actin was intact or 313 fragmented, suggested dynein might be inhibited inside asters, for example by chemical signals. To test 314 this, we turned to an artificial system. 3 µm beads were functionalized with the antibody against the 315 dynein adapter HOOK2 used in Figs 4 and 5. Negative control beads were functionalized with random IgG. 316 We then measured transport of the beads on isolated, stationary asters as in Figs 6 and 7. With F-actin 317 intact, the anti-HOOK2 beads moved inwards at a constant speed of 0.2 ± 0.1 µm/s throughout asters (Figs 318 8A-D, SI Movie 11). When F-actin was fragmented with Cytochalasin D, the anti-HOOK2 beads moved at 319 0.7 ± 0.2 µm/s (Figs 8E-H), faster than with F-actin intact. Thus, artificial dynein-coated beads were slowed 320 by F-actin, like endogenous organelles. However, these beads were transported all the way to the MTOC, 321 unlike organelles which slowed or stopped inside asters. This suggests some brake on dynein movement 322 is recruited to organelles but not to HOOK2-coated beads, or alternatively, that the force on HOOK2-323 functionalized beads is higher. 324

Volume exclusion is unlikely to block organelle movement inside asters 325
Organelles might slow down inside asters because the environment becomes too crowded with other 326 organelles. To investigate volume exclusion by organelles, we quantified the intensity and flux of 327 fluorescent dextran as a marker for the cytosol. As organelles were transported toward MTOCs, 328 fluorescent dextran was displaced away from MTOCs (Fig S5, SI Movie 12), consistent with volume 329 conservation. However, the degree of steric exclusion was fairly small, since the dextran signal was only 330 reduced by ~10%, and exclusion was only observed within ~50 µm of MTOCs, where the ER density is 331 maximal. Outside that central region, the intensity of fluorescent dextran was similar inside and outside 332 asters. We conclude that volume exclusion between organelles may be significant in the immediate 333 neighborhood of MTOCs, but is unlikely to account for organelles becoming stationary inside asters. 334

Discussion
We tracked multiple cytoplasmic networks in moving asters using two different systems to promote 335 movement, and found that organelles, F-actin, keratin, and even a small molecule moved coherently with 336 astral MTs (Figs 3,4,5, SI Movie 5). During aster separation movement, organelles on both the leading and 337 trailing sides moved away from the interaction zone (Fig 3, SI Movie 2). In stationary asters, organelles 338 exhibited a burst of movement at the aster surface, then mostly halted inside asters (Figs 6,7). This burst 339 was much more pronounced when F-actin was fragmented, but even in that situation the majority of 340 organelles were slower or stationary once they joined the body of the growing aster. Thus, asters moved 341 as a near-continuum, with maximal shear between networks at the aster surface, and much less shear 342 between networks inside asters. Co-movement of cytoplasmic networks is consistent with mechanical 343 integration and entanglement between networks. Based on these extract observations, we propose that 344 MTOCs and astral MTs move away from the midplane as a continuum as they grow after first mitosis, 345 carrying organelles and F-actin with them. 346 347 An important question is how well our aster movement systems model centrosome movement in eggs. 348 After anaphase in Xenopus eggs, centrosomes move away from the midplane at ~10 µm/min, which is 349 faster than the aster separation movement in Fig 3, and slower than the dynein-based movement over 350 the coverslip in Fig 4. Thus, neither of our extract movement systems precisely reconstituted the speed of 351 aster movement in eggs, but they spanned a wide range of relevant velocities, which increases our 352 confidence that movement is also coherent in eggs. To test whether coherent movement occurs in large 353 eggs, we re-analyzed movies of aster growth and separation movement after first mitosis in live zebrafish 354 eggs expressing a fluorescent MT binding protein (Wühr et al., 2010) (SI Movie 13). Lipid droplets are 355 visible as large dark objects in these movies. These droplets move rapidly and randomly before the aster 356 contacts them, then slowly outwards once they are embedded inside the aster. Using PIV analysis, we 357 observed outward flow of structure in the MT channel at the same speed as the lipid droplets. This analysis 358 suggests large asters in zebrafish eggs also move away from the midzone as a continuum as they grow 359 after first mitosis. 360 361 Dynein located throughout the cytoplasm is thought to generate the force that moves asters in large egg 362 cells, but where dynein is anchored has been unclear. Here, we found that all the organelles in the extract 363 moved inwards at the aster surface in a dynein-dependent manner, though this movement was only 364 pronounced when F-actin was fragmented. Thus, all the organelles may serve as dynein anchors, either 365 by recruiting dynein directly, or by physical contact with the ER. The ER moved inwards fastest and to the 366 greatest extent, perhaps because it recruits more dynein, or because individual ER tubules are smaller and 367 more deformable than other organelles. We observed no outward movement of organelles when dynein 368 was inhibited with CC1, so dynein is the dominant microtubule motor in egg asters. The identity of the 369 dynein adapter on egg organelles is unknown. Proteomic analysis estimates ~40 nM HOOK2 and ~10 nM 370 HOOK3 in eggs (Wühr et al., 2014), but preliminary experiments failed to implicate HOOK2 in organelle 371 transport. The only other known dynein adapter present at significant concentration is the kinetochore 372 protein Spindly (SPDL1, ~100 nM) (Wühr et al., 2014). In interphase U2OS cells, Spindly binds to the plasma 373 membrane (Conte et al., 2018). It might be recruited to organelles via its farnesyl modification (Holland 374 et al., 2015) or by interaction with the ZW10-containing NRZ complex (Civril et al., 2010;Menant et al., 375 2010). 376 377 Organelles reproducibly exhibited a burst of inward movement when the growing aster surface first 378 contacted them, then slowed or halted upon incorporation into the aster (Figs 6,7). This burst was much 379 more pronounced when F-actin was fragmented, and it was seen in both mass transport and PIV analyses 380 (Fig 7). Most organelles inside asters were stationary, which explains why the density of organelles in the 381 bulk of the aster is similar to that outside the aster, as previously observed (Hara et al., 2015;Wang et al., 382 2013). The molecular mechanism that restricts dynein-mediated organelle movement to the aster surface 383 is unknown. F-actin reduced transport speeds of both organelles (Fig 6,7) and artificial cargoes (Fig 8). We 384 hypothesize this is due to drag forces from F-actin networks increasing the effective viscosity of the system 385 for organelles. Organelle speeds remained maximal at the aster surface even when F-actin was 386 fragmented (Fig 7). We hypothesize this is also due to an increase in effective viscosity in the aster interior, 387 due to increased MT density and/or non-motor interactions between organelles and MTs. In contrast, 388 artificial cargoes moved to the aster center at constant speeds whether F-actin was intact or fragmented 389 (Fig 8). Organelles may recruit less dynein than beads, or they may recruit unidentified proteins which act 390 as a brake on dynein-mediated movement (Gurel et al., 2014). 391 392 Figure 9 illustrates a working model for aster and centrosome movement in eggs based on the hypothesis 393 that dynein-based forces are restricted to the aster surface. For dynein to pull on astral MTs from 394 organelles, the organelles must move toward the midzone, or at least remain stationary. In contrast, our 395 observations showed that organelles inside asters moved away from the midzone, and inward organelle 396 movement was largely restricted to the aster surface (solid red arrows in Figs 9A,B). Surface force 397 propagates through the aster by mechanical stress, and the aster responds as a deformable gel. Inside 398 asters, MTs act both as a substrate for dynein, as well as a brake on organelles, so dynein may thus 399 generate active stress inside asters, which is different than external force on astral MTs. Asters and 400 centrosomes move apart because the midzone is mechanically softer than the rest of the aster. Continual 401 MT polymerization fills the expanding space at the midzone, and the midzone kinesins Kif4A, Kif20A and 402 Kif23 keep the CPC and centralspindlin focused . Since moving asters advected cytosol 403 (Fig 5), we hypothesize outward translocation of the asters generates hydrodynamic forces that displace 404 cytoplasm around the asters and into the midzone (solid beige arrows in Fig 9C). Actomyosin-based forces 405 make a major contribution to separation movement in our model system (Fig 2), and may also contribute 406 in eggs (hollow green arrows in Fig 9C). 407 Actomyosin contraction away from the midzone, where F-actin is locally disassembled, may also generate forces on asters (hollow green arrows). Surface forces propagate through the aster gel by mechanical stress, causing flow away from the softer midzone. This flow is balanced by net MT polymerization at the midzone. (C) Centrosome separation movement in an egg showing location of forces from dynein and actomyosin and resulting movement of the aster gel and cytoplasm outside the aster. Only the flow of MTs is shown for simplicity, but organelles and F-actin inside asters move with MTs as in panel A.
We have not performed detailed simulations of the model in Fig 9, but conceptually it can account for 408 centrosome separation in an aster pair, as well as centering of a single aster in a spherical space. Frog egg 409 asters exhibit branching MT nucleation, which causes the MT density to remain roughly constant at their 410 surface as the aster grows . If dynein force on the aster surface is proportional to 411 microtubule density, or to organelle density, it will scale linearly with surface area. This "force per unit 412 surface area" feature of the model can potentially account for sperm aster centration in eggs and in 413 regions where MT growth is allowed by UV inactivation of colcemid (Hamaguchi et al., 1986). In extracts, 414 asters are almost 2D and likely experience drag from coverslips, so we expect surface forces may be more 415 significant in 3D eggs than in 2D extracts. 416

417
We are not the first to propose aster positioning by surface forces. In early microneedle experiments in 418 echinoderm eggs, Chambers observed that asters behave as a gel and proposed that forces act on their 419 surface (Chambers Jr, 1917). The proposal by Hamaguchi et al. that dynein forces act throughout asters 420 (Hamaguchi et al., 1986) came to dominate recent thinking, but our observations favor the older idea, at 421 least for frog eggs. New experiments are needed to discriminate these models in different systems. The 422 predominance of surface pulling vs length-dependent pulling on asters may depend on the size of the 423 system. We observed dynein-mediated organelle movement relative to MTs over a distance of ~50 µm 424 near the aster surface (Fig 7). This distance corresponds to a relatively thin surface layer in frog egg asters, 425 but it is larger than the cell radius in sea urchin or C. elegans eggs. To understand forces on asters in each 426 system, we think it is important to analyze the movement of all cytoplasmic components. 427

Immunofluorescence 428
Embryos were fixed and stained as described previously (Field et al., 2019). Embryos were fixed in 90% 429 methanol, 10% water, 50 mM EGTA pH 6.8 for 24 h at room temperature with gentle shaking. After 430 fixation, embryos were rehydrated in steps from 75%, 50%, 25%, to 0% methanol in TBS (50 mM Tris pH 431 7.5, 150 mM NaCl) for 15 min each step with gentle shaking. Rehydrated embryos in TBS were cut in half 432 on an agarose cushion using a small razor blade. Before staining, embryos were bleached overnight in 1% 433 hydrogen peroxide, 5% formamide (Sigma-Aldrich #F9037), 0.5x SSC ( Actin-intact, CSF Xenopus egg extract was prepared as described previously (Field et al., 2017). CSF 451 extracts were stored at 4-10 °C and flicked occasionally to disperse membranes. Extracts stored in this 452 way were typically usable for ~8 h. Before each reaction, extracts were cooled on ice to ensure 453 depolymerization of cytoskeletal networks. 454

455
Interphase aster assembly reactions 456 In a typical reaction, fluorescent probes were added to CSF extract on ice. To trigger exit from CSF arrest 457 and entry to interphase, calcium chloride was added to 0.4 mM final concentration. To ensure complete 458 progression to interphase, the reaction was mixed well immediately after calcium addition by gently 459 flicking and pipetting. Extracts were pipetted using 200 µL pipette tips manually cut to a wider bore to 460 reduce shear damage, which can make membranes in the extract appear coarser by eye. Reactions were 461 incubated in an 18 °C water bath for 5 min then returned to ice for 3 min. Next, drugs or dominant negative 462 constructs were added (see Perturbations below), and in some cases reactions were split for direct 463 comparison between control and perturbed conditions. Last, Dynabeads Protein G (Thermo Fisher 464 #10004D) functionalized with an activating anti-Aurora kinase A (anti-AURKA) antibody were added as 465 artificial microtubule organizing centers (MTOCs) (Tsai et al., 2005). For experiments in which asters 466 moved away from one another, unlabeled anti-INCENP antibody was included at a final concentration of 467 4 nM to promote zone formation by activating the CPC. 468

469
Coverslip passivation 470 18 and 22 mm square coverslips were passivated with poly-L-lysine covalently grafted to polyethylene 471 glycol (PLL-g-PEG) (SuSoS #PLL(20)-g3.5-PEG(2)) as described previously (Field et al., 2019). Coverslips 472 were cleaned by dipping them in 70% ethanol, igniting the ethanol with a gas burner, cooling the 473 coverslips for several seconds, then the coverslips were passivated by placing them on a droplet of 0.1 474 mg/mL PLL-g-PEG in 10 mM HEPES pH 7.4 on Parafilm. 18 mm coverslips were placed on 90 µL droplets, 475 and 22 mm coverslips were placed on 110 µL droplets. After 30 min incubation, excess PLL-g-PEG was 476 rinsed by placing coverslips on droplets of distilled water twice for 5 min each, then drying them with a 477 stream of nitrogen gas. To check the passivation, when we focused near the coverslips, we found no 478 evidence of a surface layer of cytoskeletal filaments or organelles adsorbed to the coverslips. Quite the 479 opposite, the density of cytoplasmic networks was typically lower near the coverslips and higher near the 480 midplane between the coverslips, we suspect due to continuous contraction of actomyosin away from the 481 coverslips sustained by continuous diffusion of monomer toward the coverslips. solutions were stored on ice until use, then diluted an additional 10-30 fold into the final reaction. F-actin 520 was imaged with Lifeact-GFP (Moorhouse et al., 2015;Riedl et al., 2008) at a final concentration of 300 521 nM. More details on fluorescent probes are reported in (Field et al., 2017). 522

Perturbations 524
To fragment F-actin, Cytochalasin D (CytoD) was added to a final concentration of 20 µg/mL. CytoD was 525 diluted in DMSO to 10 mg/mL, then diluted 20 fold into extract. This extract working solution was stored 526 on ice until use, then diluted an additional 25 fold into the final reaction. CytoD and other drugs or 527 dominant negative constructs were typically added to actin-intact extracts after cycling to interphase, 528 then reactions were split for direct comparison between control and perturbed extracts. Alternatively, 529 CytoD may be added during extract preparations before the crushing spin, following the classic CSF extract 530 protocol (Murray, 1991). The ER appeared coarser in CytoD extracts than in actin-intact extracts, and the 531 ER appeared to coarsen over time in actin-intact extracts plus CytoD. 532

Preparation of dynein on coverslips 554
Coverslips were passivated following the protocol above but using biotinylated PLL-g-PEG. NeutrAvidin 555 and biotinylated Protein A (GenScript #M00095) were mixed in a 1:1 ratio to a final concentration of 10 556 µM and stored at 4 °C. Just before functionalizing the coverslips, the NeutrAvidin (Thermo Fisher #31000) 557 and biotinylated Protein A mixture (GenScript #M00095) was diluted 42 fold to 240 nM in 1x PBS with 558 0.0025% Tween 20. That concentration was found to be the smallest amount to decrease the surface 559 tension enough to maintain a layer of solution on the coverslips, to reduce damage to the functionalized 560 surfaces due to air-water interfaces when transferring the coverslips from one droplet to another. 561 Coverslips were incubated with the NeutrAvidin and biotinylated protein A mixture at least 30 min on 562 droplets on Parafilm at room temperature. Coverslips were incubated under a box with a damp paper 563 towel, to block room light and to reduce evaporation. After the incubation, coverslips were rinsed twice 564 on droplets of 1x PBS with 0.0025% Tween 20 for 5 min each, then incubated with anti-HOOK2 or random 565 IgG diluted in 1x PBS with 0.0025% to a final concentration of 10 µg/mL at least 30 min. After the 566 incubation with antibody, coverslips were rinsed twice on droplets of 1x PBS with 0.0025% Tween 20, 567 then twice on droplets of distilled water, then swirled in a beaker of distilled water, then gently drying 568 them with a stream of nitrogen gas. Coverslips were often used same day, but could be stored overnight 569 in the dark at 4 °C and used the following day. After perfusing extracts into flow cells and sealing the edges 570 with VALAP, the metal slide holders were chilled for 10 min on a metal block on ice, to allow endogenous 571 HOOK2 and dynein-dynactin time to bind the anti-HOOK2 before the start of aster growth. 572 573

Photo-release of fluorescein from MTOCs 574
Caged fluorescein with -O-CH2-COOH functionality on the caging groups was synthesized as described 575 (Mitchison et al., 1998). Carboxylic acid groups were activated as sulfo-NHS esters in a small reaction 576 containing 2 micromols caged fluorescein, 5 micromols sodium sulfo-NHS and 5 micromols 1-ethyl-3-(3-577 dimethylaminopropyl)carbodiimide (EDC) (Thermo Fisher #22980) in 10 µL of DMSO. After 1 h at room 578 temperature, this reaction mix was added directly to protein coated beads. Direct modification of anti-579 AURKA beads caused loss of nucleation activity, so we first biotinylated beads, then modified with caged 580 fluorescein, then attached anti-AURKA IgG using a NeutrAvidin bridge. Dynabeads Protein A (Thermo 581 Fisher #10002D) were sequentially incubated with goat anti-rabbit whole serum (Jackson 582 ImmunoResearch #111-001-001) then biotinylated rabbit IgG (homemade). They were labeled with the 583 caged fluorescein reaction mix in 0.1 M K HEPES pH 7.7 for 1 h, then washed again. We empirically titrated 584 the amount of reaction mix added such that beads were maximally labeled while still retaining nucleation 585 activity in extract. After labeling with caged fluorescein, beads were incubated sequentially with a mixture 586 of NeutrAvidin and biotinylated protein A, then rabbit anti-AURKA to confer nucleation activity. Pure 587 proteins were added at 10-20 µg/ml and serum was added at 1/20. All binding reactions were incubated 588 for 20 min, and washes were in 1x PBS. Fluorescein was released from beads by exposing the microscope 589 field to full illumination in the DAPI channel (395 nm) for 5 sec. 590 591 Analyses 592 PIV 593 PIVlab (Thielicke et al., 2014) was used to estimate flow fields of cytoplasmic networks based on particle 594 image velocimetry (PIV). Though PIV is primarily used to estimate flow fields based on tracer particles 595 embedded in fluids, PIV has been used to estimate cortical or cytoplasmic flows in C. elegans cortices 596 (Mayer et al., 2010), zebrafish epithelia (Behrndt et al., 2012), and Drosophila embryos (Deneke et al., 597 2019  2D Gaussian fitting of fluorescein photo-released from MTOCs was performed using a nonlinear least 607 squares solver in MATLAB (Nootz, 2020). After photo-release the MTOCs were bright due to uncaged 608 fluorescein that remained bound to the MTOCs. Thus the MTOCs were masked as not to bias the Gaussian 609 fits. We fit expansion of the fluorescein cloud to a model of diffusion, and we assumed a diffusion 610 coefficient of fluorescein in water of 425 µm 2 /s (Culbertson et al., 2002). Advection of cytosol with 611 cytoplasmic networks is consistent with a poroelastic Péclet number VLµ/Eξ 2 greater than unity (Mitchison 612 et al., 2008;Moeendarbary et al., 2013). Given the oscillatory speed V ~1 µm/s ( Fig 4D) and amplitude L 613 ~30 µm (Fig 4C), and assuming a viscosity µ ~6x water ( Fig 5E) and an elastic modulus E ~10 Pa (Valentine 614 et al., 2005), we estimate the upper bound on the effective pore size ξ of cytoplasmic networks in this 615 system is ~100 nm. 616

Analysis of organelle mass transport 618
The flux-based analysis of organelle transport is described in Fig S4. In summary, images were background 619 subtracted and flat field corrected, then a region of interest (ROI) was defined large enough to enclose 620 the aster at all time points, so the total amount of ER in the ROI was conserved. Then, the total intensity 621 was normalized across frames to correct for photobleaching. The net flux of organelle fluorescence 622 intensity toward MTOCs was calculated as described in Fig S4. In particular, the average intensity was 623 calculated in annular bins with a width of 10 µm, then the cumulative total intensity was calculated from 624 the MTOC to outside the aster, then the net flux was calculated at each radial distance by subtracting 625 subsequent cumulative total intensity profiles.  In 3 repeat extracts, we measured 3 conditions: anti-HOOK2 in interphase extracts (columns 1 to 3), anti-HOOK2 in mitotic extracts (columns 4 and 5), and as negative control, random IgG in interphase (columns 6 through 8). HOOK2 conditions were normalized so the amount of HOOK2 was constant. Random IgG conditions were normalized to have the same IgG count as the average IgG count of the HOOK2 columns. Then for each protein, the sum across channels was normalized, so the abundance values represent relative enrichment across the 8 channels. Abundances are shown on a log scale with base 2. We here show dynein subunits (blue), dynactin subunits (red), and other proteins known to interact with dynein-dynactin (yellow, green) that came down in the immunoprecipitations. The interaction between HOOK2 and the dynein-dynactin complex was stronger in interphase (left columns) than in mitosis (middle columns).  Fig S4. (D) Average speed based on PIV, in the same frame of reference as panels B and C, and averaged over a quadrant. In all analysis methods, mitochondria behaved similar to acidic organelles and likewise exhibited a burst of movement at the growing aster surface. Fig 7) (A) From the first frame (top row, horizontal bars) to the second frame (bottom row, vertical bars), consider one unit of ER that moves from left to right (blue), and another unit of ER that moves from right to left (red). (B) Cumulative intensity of ER during the first frame (horizontal bars) and second frame (vertical bars). (C) Difference between the cumulative intensity during the second frame C(t2) (vertical bars) minus the cumulative intensity during the first frame C(t1) (horizontal bars) highlights mass transport of the ER toward the MTOC. (D) There is a net flux of ER toward the MTOC where C(t2) exceeds C(t1) (red, vertical bars), while there is a net flux of ER away from the MTOC where C(t1) exceeds C(t2) (blue, horizontal bars).

SI Movies
SI Movie 1. Dynamic reorganization of cytoplasmic networks during the initial stages of aster nucleation and growth. Related to Fig 1. MTs were labeled with tubulin-Alexa Fluor 647, ER with DiI, and F-actin with Lifeact-GFP. Imaged on a spinning disk confocal with 60x objective lens. Cytoplasmic networks were highly dynamic, and astral MTs dynamically reorganized the ER and F-actin networks. Parts of the ER exhibited abrupt and transient motion toward the MTOC, presumably driven by dynein, and the F-actin transitioned from random to radial entrainment with MTs. SI Movie 2. Co-movement of MTs, ER, and F-actin during aster separation movement. Related to Fig 2A  and Fig 3. MTs were labeled with tubulin-Alexa Fluor 647, ER with DiI, F-actin with Lifeact-GFP, and organelles were shown in differential interference contrast (DIC). All cytoplasmic networks moved together. Note the flow of organelles visible in DIC: inside asters, where the density of F-actin, MTs, and ER was higher, organelles flowed with the asters; whereas along interaction zones between asters where the density of F-actin was lower, organelles flowed in the opposite direction, into the space on the right that was vacated by the asters moving to the left. SI Movie 3. Both dynein and actomyosin contribute to aster separation movement. Related to Fig 2. We compared four conditions: control with F-actin intact, dynein inhibited by CC1, F-actin fragmented by Cytochalasin D, and double inhibition of dynein and F-actin. F-actin was labeled with Lifeact-GFP, ER with DiI, and CPC-positive interaction zones with anti-INCENP-Alexa Fluor 647. MTs grew and CPC-positive zones formed between asters in all conditions. F-actin and ER were imaged instead of MTs because local disassembly of F-actin along CPC-positive interaction zones enables aster separation movement, and inward transport of ER and other organelles is thought to drive dynein-based aster movement. SI Movie 6. Advection of fluorescein with moving asters during oscillatory aster movement. Related to Fig 5. The first frames show MTs labeled with tubulin-Alexa Fluor 647, and the aster filled the region. The next few frames show the caged fluorescein attached to the MTOC. Then, the fluorescein was simultaneously photo-released from the MTOC as its fluorescence was uncaged, releasing a cloud of fluorescent fluorescein around the MTOC. The fluorescein cloud was fit with a 2D Gaussian. The center of the cloud is indicated at the intersection of the red and green lines, and the standard deviation of the cloud is indicated by the black circle. The plots above and to the right indicate the intensity values along the lines, and the black curves show the 2D Gaussian fit along the lines. SI Movie 7. F-actin reduced dynein-based transport of ER and mitochondria on stationary asters. Related to Fig 6. The growing aster is indicated by growing +TIPs labeled with EB1-GFP, ER was labeled with DiI, and mitochondria with TMRE. In control with intact F-actin, some ER accumulated around the MTOC, and little to no mitochondria accumulated around the MTOC. When F-actin was fragmented, a greater fraction of ER and mitochondria were transported toward the MTOC. When dynein was inhibited, organelles were not transported, neither toward nor away from the MTOC. Fig 7. MTs were labeled with tubulin-Alexa Fluor 647, ER with DiI, and F-actin with Lifeact-GFP. The ER exhibited a burst of movement toward the MTOC at the growing aster surface, resulting in transient depletion of the ER intensity near the aster surface. SI Movie 9. Burst of ER and acidic organelle movement at the growing aster surface with F-actin fragmented. Related to Fig 7. Transport of ER and acidic organelles with F-actin fragmented by Cytochalasin D. MTs were labeled with tubulin-Alexa Fluor 647, ER with DiI, and acidic organelles with LysoTracker Red. Unlike in control with F-actin intact, the burst of movement near the aster surface was highly reproducible when F-actin was fragmented with Cytochalasin D.  (Wühr et al., 2010) and analyzed with permission. Microtubules were labeled with microtubule-binding domain of Ensconsin fused to three GFPs (EMTB-3GFP) (Faire et al., 1999;von Dassow et al., 2009). Flows of MTs were estimated by PIV (Methods).