Disassembly of actin and keratin networks by Aurora B kinase at the midplane of cleaving Xenopus laevis eggs

Summary

The large length scale of Xenopus laevis eggs facilitates observation of bulk cytoplasm dynamics far from the cortex during cytokinesis. The first furrow ingresses through the egg midplane, which is demarcated by Chromosomal Passenger Complex (CPC) localized on microtubule bundles at the boundary between asters. Using an extract system, we found that local kinase activity of the AURKB subunit of the CPC caused disassembly of F-actin and keratin between asters, and local softening of the cytoplasm as assayed by flow patterns. Beads coated with active CPC mimicked aster boundaries and caused AURKB-dependent disassembly of F-actin and keratin that propagated ~40μm without microtubules, and much farther with microtubules present. Consistent with extract observations, we observed disassembly of the keratin network between asters in zygotes fixed before and during 1^st cytokinesis. We propose that active CPC at aster boundaries locally reduces cytoplasmic stiffness by disassembling actin and keratin networks. Possible functions of this local disassembly include helping sister centrosomes move apart after mitosis, preparing a soft path for furrow ingression and releasing G-actin from internal networks to build cortical networks that support furrow ingression.

Graphical Abstract

eTOC Blurb

Field et al. use Xenopus egg extracts to analyze interaction between cytoskeleton networks prior to cytokinesis. Activity of Aurora B kinase localized at boundaries between microtubule asters causes local disassembly of actin and keratin networks.

Introduction

Cytokinesis in animal cells requires large scale re-organization of the cytoplasm that involves essentially every cytoskeletal and membrane system. Most studies of cytokinesis mechanics focus on the cortex, where local actomyosin contraction promotes cleavage furrow ingression. The internal cytoplasm also has to re-organize to allow furrow ingression, which may occur either by passive deformation in response to cortical forces, or active re-organization by furrow-independent mechanisms, or both.

Eggs of the frog Xenopus laevis are well suited for investigating the organization of internal cytoplasm prior to cleavage. They are ~1.2mm in diameter, and the 1^st furrow ingresses hundreds of microns over tens of minutes, starting at the animal pole and cutting through the egg midplane. Furrow location is specified by an egg-spanning plane of microtubule bundles coated with the cytokinesis signaling protein complexes CPC and Centralspindlin [1]. This plane initiates during anaphase at the position previously occupied by the metaphase plate, stimulated by proximity to chromatin. It then expands outwards at the boundary between the sister microtubule asters that grow from the poles of the mitotic spindle, and initiates furrow assembly where it touches the cortex [2]. Both CPC and Centralspindlin are required for cytokinesis in every cell type tested [3]. Here we focus on CPC, for which we have excellent tools in Xenopus. A requirement for CPC activity for cytokinesis was shown in several large cell types, including zebrafish [4], sea urchin [5] and Xenopus [6]. CPC is activated by autophosphorylation, which is enhanced by binding to microtubules [7]. We previously proposed that a positive feedback loop between CPC auto-phosphorylation and microtubule bundling and stabilization promotes spread of a CPC-positive state from anaphase chromatin to the egg cortex via microtubule bundles between asters [1, 2].

The cytoplasm of Xenopus eggs contains F-actin and keratin networks with the potential to impede centrosome movement and furrow ingression. From proteomic data, estimated concentrations in eggs are 14 μM actin, 700 nM keratin-8; and little vimentin or other intermediate filament proteins [8]. In extract experiments, ~50% of the actin polymerizes into a dense network of bundles throughout the egg during both mitosis and interphase [9,10]. Keratin assembly increases after fertilization, presumably in response to decreased Cdk1 activity, and cleaving blastomeres contain an organized keratin filament array that is enriched on the cortex [11,12]. The organization of keratin in the deep cytoplasm between microtubule asters is unknown.

The organization of actin and intermediate filaments during cytokinesis has been extensively investigated in smaller somatic cells. Actin polymerization locally increases at the furrow cortex during cytokinesis in response to RhoA signaling [13]. Actin dynamics deeper in the cytoplasm are less well investigated, in part because cortical actomyosin is thought to dominate the mechanical behavior of small cells. Keratin and vimentin networks partially depolymerize during mitosis in response to phosphorylation by CDK1 [14,15] and AURKB, the kinase subunit of the CPC [16,17]. A plausible model is that CDK1 promotes global network disassembly at mitotic entry, while AURKB adds a focused disassembly activity that prevents re-formation at the midplane, which might impede furrow ingression. However, definitive evidence on the role of AURKB is lacking.

Living Xenopus eggs are opaque. For live imaging we used an actin-intact egg extract system [6] to investigate how interphase microtubule asters regulate F-actin and keratin networks. The bulk visco-elastic properties of this system, and the contribution of different cytoskeleton networks to bulk mechanics, were previously probed [18]. Here, we use microscopy to investigate cytoskeleton networks, and dependency relationships between F-actin, keratin and microtuble organization. We observe local disassembly of both F-actin and keratin between asters, dependent on AURKB kinase activity. We investigate the spatial regulation and mechanical consequences of this activity, and discuss possible functions in dividing eggs.

Results

AURKB-dependent disassembly of F-actin bundles at boundaries between asters

Figure 1A and Videos S1 and S2, show a typical experiment. Actin-intact interphase egg extract containing fluorescent probes and artificial centrosomes was spread between passivated coverslips and imaged. See STAR Methods for details on extract preparation. Time is scored relative to spreading and warming to 20°C. F-actin was present as a dense network of relatively homogenous bundles interspersed with occasional F-actin comet tails. In the vicinity of microtubule organizing centers actin bundles became partially entrained to the radial array of microtubules (Figure 1A 8, 18min). CPC was recruited at shared boundaries when asters touched and interacted. AURKB and Prc1/Kif4A activities together block plus end growth and generate sharp boundaries between asters, as previously reported [6,19]. In a new observation, F-actin bundles locally disassembled at the line of CPC recruitment (Figure 1A 18, 21min). Between 40-60min the system reached steady state. F-actin bundles were mostly restricted to islands centered on MTOCs, separated by gaps ~100μm wide centered on CPC bundles (Figure 1A, 48 min). Loss of F-actin at CPC-positive aster boundaries was highly reproducible over >100 fields in >30 batches of extract.

Figure 1. AURKB-dependent disassembly of F-actin at boundaries between asters in egg extract.

A) Control reaction containing probes for F-actin Lifeact-GFP), CPC (Alexa 647-anti-INCENP IgG) and microtubules (Tau peptide-mCherry) imaged by wide-field epifluorescence with a 20x objective. Note recruitment of CPC to anti-parallel microtubule bundles at boundaries between asters, and loss of F-actin bundles where CPC is recruited. Third row: higher magnification views of boxed region from top left panel. Note F-actin comet tails in the cleared region (chevrons, and 3x inserts). See Videos S1 and S2. B) Reaction containing 40 μM barasertib, a selective AURKB inhibitor. Asters grow and contact each other, but CPC recruitment is blocked, and F-actin bundles are not disassembled at aster boundaries. Third row: higher magnification views of boxed region, top left panel. See Video S3.

The selective small molecule AURKB inhibitor barasertib [20] completely blocked CPC recruitment and F-actin disassembly (Figure 1B, Video S3) suggesting that these activities depend on AURKB. F-actin appeared uniform across the whole field except for some initial entrainment to microtubules in the vicinity of organizing centers (Figure 1B). AURKB is required to form sharp microtubule boundaries between asters [6]. Wider, less defined boundaries were still observed at later times in barasertib, perhaps because the “anti-parallel pruning” activity of PRC1E + Kif4A does not depend on AURKB [19].

To probe F-actin disassembly at aster boundaries we imaged at higher spatial-temporal resolution (Figure 2A, Video S4). The areas between the asters starts as an isotropic meshwork. F-actin bundle disassembly began concurrently with appearance of CPC-coated microtubules. During disassembly, remaining bundles tended to align normal to the aster boundary, and some appeared to snap, possibly due to tension see Video S4 Tension and contractility may help focus F-actin into islands at late time points (Figure 1A, 48 min). F-actin comet tails, presumably nucleated by Arp2/3, were often observed moving in boundary regions and within the dense actin networks on either side (bottom panels in Figure 2B, Video S5). Comet formation appeared to temporally correlate with bundle disassembly in 4 different extracts, though this was hard to quantify. The presence of moving comet tails within boundary regions suggests that bundle disassembly may be caused by inhibition of a specific nucleation or stabilization pathway rather than global inhibition of polymerization. Bundle disassembly might favor comet formation due to competition between assemblies for actin subunits [21].

Figure 2. Higher spatiotemporal resolution imaging of F-actin, CPC, and microtubules during and after CPC-positive boundary formation.

(A) Initiation of actin disassembly at a CPC-positive boundary in a reaction containing probes for F-actin Lifeact-GFP), CPC (Alexa 647-anti-INCENP IgG), and microtubules (Tau peptide-mCherry) imaged by spinning disk confocal with a 60x objective. Note alignment of remaining F-actin bundles as the overall density decreases. Third row: higher magnification views of actin in the boxed regions. (B) Later sequence in the same reaction at higher temporal resolution illustrating multiple F-actin comets growing in and around the boundary region. Third row: higher magnification views in the boxed regions showing a typical F-actin comet growing in a boundary region where most F-actin bundles have disassembled. See also Video S4 and Video S5.

AURKB-dependent disassembly of keratin at boundaries between asters

We next examined keratin structure in our aster assembly reactions. Keratin was visualized using labeled anti-keratin IgG [22]. Figure 3A and 3C show keratin in an image sequence collected in parallel with Figure 1A and B, so conditions and timing can be directly compared. Keratin is mostly unassembled in mitotic egg extract, and started to assemble into networks following entry into interphase. This assembly is seen in gradual coarsening of the keratin signal in Figure 3A (top panel) and 3C (bottom panel).

Figure 3. AURKB-dependent disassembly of keratin at boundaries between asters in egg extract.

Panels A and C are from reactions run in parallel with Figure 1 using multi-position time lapse, so timing can be directly compared. A) Control reaction containing probes for Keratin (Alexa 568-anti-keratin IgG), CPC (Alexa 647-anti-INCENP IgG) and growing microtubule plus ends (EB1-GFP). Note partial disassembly of keratin at boundaries between asters where CPC is recruited. Keratin aggregates transiently accumulated at boundary lines, noticeable at 28 min in this example. See Video S6 which shows a different control experiment. B) Images from an experiment similar to A) run on a different day and chosen to highlight a boundary zone that formed later in the reaction, after the keratin network was established. Bulk flow caused the network to move leftwards and upwards in this time series. See Video S7. C) Parallel reaction to A) containing 40 μM barasertib CPC recruitment and keratin disassembly at aster boundaries are both blocked. Keratin clearing in a circular zone near MTOCs is similar to the control reaction in A). Bottom row: higher magnification views of boxed region in the top left panel. Note gradual evolution of the keratin signal from homogeneous aggregates into a network of connected bundles. Similar time-dependent assembly occurred in controls.

Keratin networks did not assemble around artificial centrosomes, as visualized by the ~30μm holes around beads in Figure 3A and 3C. This could be due to transport, spatial occlusion due to vesicle accumulation, or perhaps a chemical gradient. Keratin networks were also disrupted at aster boundaries when CPC was recruited (Figure 3A, 48 min). Keratin disruption was slower than F-actin, and extended less far from the aster boundary, perhaps reflecting the less dynamic nature of keratin filaments. We often noted accumulation of disassembled keratin fragments at the center of aster boundaries (Figure 3A, 3x inset), possibly due to plus end-directed transport of keratin fragments that accumulate when the network is partially dissolved by AURKB activity. Video S6 associated with Figure 5A shows a different control experiment where keratin disassembly between asters is clearly visualized.

Figure 5. Inhibition of cytoskeleton assembly around CPC beads.

7μm (panels A,B) or 2.8μm (panels C-I) Protein A beads coated with anti-INCENP IgG were used to recruit and activate endogenous CPC in extract. 40 μM nocodazole and/or 40mM barasertib were included, as indicated, to inhibit microtubules and/or AURKB. Yellow chevrons indicate CPC beads. Actin (Lifeact-GFP), keratin (Alexa 568-antikeratin IgG), Tubulin (directly labeled with ALEXA-647 NHS ester), CPC (DasraA-GFP) and growing microtubule plus ends (EB1-GFP). A) F-actin clearing around 7μm CPC beads in the absence of microtubules. t=45min. B) F-actin was not cleared when AURKB activity was inhibited. t=45min. Panels A, B are taken from parallel reactions. C) Quantification of F-actin clearing from the experiment shown in A,B. 15 beads analyzed per condition. Dark lines show average radial intensity values, pale regions show the standard deviation of intensity values. Linescans were truncated below 7μm to avoid the strong bead edge signal. D) Large zones of F-actin, microtubules and keratin disassemble around 2.8μm CPC beads when microtubules are present. t=60min. F-actin and microtubules disassembled in zones initially ~50μm in diameter, which grew progressively to 200μm or more. Keratin disassembled over smaller diameters. See Video S9. E) AURKB activity was required for disassembly t=60min. See Table S1. F) Microtubules were required for disassembly t=60min. Panels D, E, F are taken from parallel reactions. See Table S1. G) EB1 comets were present throughput the extract at t=60min, and were lost around CPC beads. Conditions similar to D. H) CPC-coated microtubule bundles around CPC beads. Conditions similar to D except that Dasra-GFP was added to visualize CPC. t= 60min. I) Time lapse sequence of formation of CPC-coated microtubule bundles and F-actin disassembly around CPC beads. Conditions similar to H. See Videos S10 and S11.

The precise morphology of keratin disassembly at aster boundaries depended on the relative kinetics of mitotic exit vs boundary formation. Keratin started to assemble into networks after mitotic exit. At high aster density, CPC-positive boundaries formed by 18 min, before keratin had fully assembled into networks. In this case, keratin fragments tended to accumulate at the center of the boundary (Figure 3A). At low aster density, the keratin network was fully assembled when aster boundaries formed. In this case, the keratin network locally disassembled and ripped apart at the aster boundary when CPC was recruited (Figure 3B, Video S7).

Keratin disassembly at aster boundaries depended on AURKB activity, since it was blocked by barasertib (Figure 3C). Keratin networks were still reduced around MTOCs, and in some cases were slightly enriched at aster boundaries. To quantify keratin disassembly, we scored 28 asters on 4 coverslips under control conditions at 45 min.

Out of 58 boundaries that were positive for CPC, 58 (100%) showed evidence of keratin disassembly. On 4 coverslips run in parallel with 40 μM barasertib, we scored 18 aster boundaries. None were positive for CPC and none showed evidence of keratin disassembly. We did sometimes observe areas of low keratin density which corresponded to organelle aggregates.

The labeled anti-INCENP IgG probe used to visualize CPC in Figures 1&2 promotes auto-activation of CPC [1]. This was convenient for observation of F-actin and keratin disruption because every aster boundary rapidly recruited CPC. To test if the antibody activation of CPC was required for actin and keratin disruption we introduced a GFP fusion of the Dasra subunit of CPC, which does not stimulate CPC auto-activation⁶. Using this probe, some aster boundaries recruit CPC and others in the same field do not, dependent on the initial distance between MTOCs [1]. We imaged aster assembly reactions using Dasra-GFP and analyzed data at a timepoint where ~50% of aster boundaries had recruited CPC. For quantification, we used a Voronoi diagram to define aster boundaries in the microtubule channel, then scored CPC recruitment and actin clearing at these boundaries. 92 aster boundaries were scored on 3 coverslips. Of these, 43 were CPC positive and disassembled actin. 45 were CPC negative and did not disassemble actin. 1 was CPC positive and did not clear actin, and 3 appeared to clear actin but were not CPC positive. Thus, 88/92 boundaries (96%) matched our prediction that actin disassembly occurs close to CPC-coated microtubule bundles, even with a non-activating CPC probe. The kinetics and extent of actin clearing were independent of the CPC probe used.

Boundary regions between asters are less gelled than aster centers

Disruption of F-actin and keratin networks at aster boundaries is likely to cause a change in local mechanics. We probed this by measuring cytoplasmic flows in response to pressure gradients induced by adding and removing a small weight from the top coverslip. Flow was imaged in the DIC channel as coherent movement of mitochondria and other small vesicles, and quantified using Particle Image Velocimetry (PIV) analysis (Figure 4A). When flow was induced after assembly of CPC-positive aster boundaries, it occurred preferentially through boundary regions (cyan arrows, PIV analysis). When F-actin disassembly between asters was blocked by barasertib, preferential flow was not (Figure 4B). See also Video S8. This conclusion is based on 21 fields visually scored on two separate days. In control reactions we observed flow preferentially between aster boundaries in 18 examples, no clear flow in 3, and zero examples with flow through aster centers. In parallel barasertib-containing reactions we observed no locally coherent flow between asters in 15 examples. In most cases compressing the coverslip in barasertib reactions caused the whole field to displace (PIV panel in Figure 4B). In a few cases we observed rips and flows in the gel in random locations. We conclude that the boundary between asters provides a soft path for flow of cytoplasm in response to pressure gradients, and formation of this path depends on local melting of cytoskeleton networks by AURKB activity.

Figure 4. Low resistance to pressure induced flow between asters.

A) Flow in an aster assembly reaction similar to Figure 1A. Fluorescent images were collected immediately before initiating flow. Actin (Lifeact-GFP), Tubulin (Tau peptide mCherry) and CPC (anti-INCENP-Alexa-647). Right panel: a DIC still taken from a 20 sec movie collected immediately after inducing flow with a pressure transient. Cyan arrows indicate the velocity of particle movement, measured by Particle Image Velocimetry (PIV) analysis of two sequential frames from the movie. Flow occurred preferentially through the F-actin-free channels between asters. Peak flow in this example was ~7μm/sec. (B) Parallel reaction to A) containing 40 μM barasertib to inhibit AURKB. AURKB does not localize between asters. Slower flow that was uniform across the whole field was observed following a pressure transient. Cyan arrows are the same scale as in A). See Video S8.

Disassembly of F-actin around CPC beads in extract without microtubules

The CPC-coated microtubule bundles that assemble between asters could interact with F-actin in multiple ways, e.g. via other factors such as motor proteins or MAPs. To test if local concentrations of active CPC in the absence of microtubules were sufficient to disrupt F-actin and keratin networks we prepared Protein A beads coated with anti- INCENP IgG, here after referred to as CPC beads. This IgG binds to the C-terminus of INCENP and both recruits CPC and promotes activation by auto-phosphorylation [7,1].

We imaged F-actin in the presence of nocodazole (Figure 5 A, B), using 7μm beads to maximize local effects. We observed reproducible disassembly of F-actin extending ~40 μm away from each CPC bead (Figure 5A), but not control IgG beads (not shown) or when AURKB activity was inhibited with barasertib (Figure 5B). Initial scoring of beads was by eye. In one experiment 30/31 CPC beads vs 0/59 beads coated with control IgG exhibited local clearing of F-actin at 45min. In a different experiment 22/23 CPC beads vs 0/23 incubated in parallel with barasertib exhibited F-actin clearing at 45min. We quantified the spatial extent of F-actin clearing by computing radial F-actin intensity profiles with respect to distance from the beads (Figure 5C). Clearing extended up to 40 μm from CPC beads with a half-maximal distance of ~15μm, and was blocked by barasertib. We hypothesize a reaction-diffusion system in which an unknown actin regulator visits the beads, is phosphorylated by AURKB, and diffuses away to disassemble F-actin. This hypothesis is based on demonstration of an AURKB reaction-diffusion gradient in dividing cells [23]. Keratin disassembly was weak in assays containing nocodazole, and was observed around 7μm CPC beads in only some experiments.

Disassembly of all three cytoskeleton networks when microtubules are present

In interphase extracts, spontaneous microtubule assembly occurs within 20-30 minutes in the absence of centrosomes [24]. We added 2.8μm CPC beads to interphase extracts and observed zones with reduced density of F-actin, keratin, and microtubules around each bead, starting at 40-60min (Figure 5D, Video S9). Keratin disruption occurred over a smaller radius. Inhibition of cytoskeletal networks around CPC beads was AURKB activity dependent: it was not observed around beads coated with control IgG (not shown), or when an AURKB inhibitor was added (Figure 5E). We quantified disassembly of cytoskeleton networks around CPC beads in the presence and absence of AURKB inhibition in 3 independent experiments (Table S1), and confirmed that local disassembly of F-actin, microtubule and keratin networks was reproducible and AURKB activity dependent. Addition of nocodazole, to match the conditions in Figure 5A, prevented formation of cytoskeleton disassembly zones using smaller beads (Figure 5F). This effect shows that microtubule assembly is required for large zones of disassembly around CPC beads.

To test effects of CPC beads specifically on dynamic microtubule plus ends, we imaged EB1-GFP which tracks growing plus ends. EB1 comets were present throughout the extract at t=60 min, growing inapparently random directions. In the vicinity of CPC-coated beads, EB1 comets were completely missing (Figure 5G). This effect was AURKB dependent (not shown).

Assembly of CPC-coated microtubule bundles near CPC beads

We wondered how microtubules contributed to the formation of large, spreading zones devoid of F-actin bundles, given that the density of microtubules in these zones was low. Using Dasra-GFP, we observed a low density of CPC-coated microtubule bundles in the vicinity of CPC beads. Some bundles appeared disorganized, others were aligned in parallel or radial arrays (Figure 5H). To understand how these bundles form we performed time-lapse imaging (Figure 5I, Videos S10 and S11). We observed small, CPC-coated microtubule bundles forming in the vicinity of a bead, elongating and often coalescing into larger bundles that spread outwards. CPC-coated bundles only formed after the field filled with spontaneous background microtubules, and only formed near CPC beads. We hypothesize bundles form by CPC-promoted aggregation of spontaneous microtubules, but cannot rule out local nucleation. Once CPC-coated bundles formed, F-actin bundles started to disassemble. Accumulation of more CPC-coated bundles around each bead over time correlated with increased diameter of the zone of F-actin and keratin disassembly (Videos S10 and S11). We hypothesize that spreading of CPC-coated bundles away from a CPC bead mimics the autocatalytic CPC-spreading reaction that relays the position of the metaphase plate to the cortex to position the furrow in eggs (see discussion).

Disruption of keratin organization ahead of the cleavage furrow in intact zygotes

Figures 1-5 used an extract system. To examine these phenomena in vivo we analyzed intact zygotes using fixation, clearing and laser scanning confocal imaging (Figure 6). We visualized microtubules, CPC and keratin using published methods. No method we tested allowed visualization of F-actin. We focused on zygotes fixed before and during 1^st cleavage (70-100min post fertilization). We observed clear evidence for disassembly of keratin at the boundary between sister asters in zygotes fixed prior to cleavage initiation (Figure 6A) and ahead of ingressing furrows in zygotes fixed during cleavage (Figure 6B, C). Keratin disassembly was spatially coincident with CPC at the boundary between asters, as expected from extract observations. The presence of yolk platelets and lipid droplets made it difficult to visualize extended filaments, but the appearance of keratin at boundaries between asters was quite similar in fixed zygotes and the extract system. Disassembly occurred over a zone ~30μm wide centered on the aster boundary. In zygotes that had not yet initiated cleavage we usually observed a thin plane of keratin aggregates at the center of the boundary region (Figure 6A). These resembled accumulation of keratin aggregates at the center of the boundary region in extract (Figure 3A). After furrow initiation (Figure 6B,C) the network organization of keratin was more pronounced throughout the zygote, and its disruption near CPC-coated microtubule bundles was more obvious. We conclude that keratin network disassembly occurs at boundaries between sister asters in zygotes, spatially coincident with CPC enrichment. Disassembly occurs hundreds of μm, and hundreds of seconds, ahead of the advancing furrow.

Figure 6. Keratin disassembly ahead of cleavage furrow ingression in fixed zygotes.

Xenopus zygotes were fixed 70-100min post fertilization, bleached, stained and imaged by confocal microscopy in a clearing solvent. Second image pair in each row: 4x view of boxed region. A) 70min post fertilization, egg viewed from animal pole. The asters have grown to ~65% of the egg radius, cleavage has not yet initiated. Keratin bundles are not yet pronounced in the bulk cytoplasm. Keratin appears partly disassembled in a ~30μm zone at the aster boundary. Aster boundaries with a line of keratin fragments in the center are seen, resembling those seen in extract (Figure 2A) B) 90min post fertilization, animal pole at top left. This zygote recently initiated cleavage. Keratin bundles are evident in the higher magnification view (different focal plane). Note partial clearing and accumulation of aggregates at the center of the aster boundary region, similar to A). C) 100min post fertilization, animal pole at top. The furrow has ingressed further than B), and keratin bundles are more pronounced away from the aster boundary. Large keratin bundles also accumulate at the cortex (not shown). Disassembly of the keratin network ahead of the ingressing furrow is evident, especially in the higher magnification view.

Discussion

Figure 7A illustrates our current understanding of molecular pathways at the boundary between two asters in frog eggs. It is an extension of our previous model [2] for autocatalytic spreading of CPC-coated microtubules at aster boundaries (red arrows) with the addition of F-actin and keratin disassembly (green inhibitory arrows). The blue inhibitory arrows show the block to microtubule plus end growth at aster boundaries. New experiments with CPC beads (Figure 5H,I and Videos S10, S11) strengthen the autocatalytic spreading model by showing that beads coated with active CPC activity induce or stabilize formation of nearby CPC-coated microtubule bundles. Using F-actin disassembly in the absence of microtubules as an endogenous reporter of AURKB activity (Figure 5A), we estimate the distance for half-action of a hypothetical reaction-diffusion system emanating from CPC beads at ~15μm (Figure 5C). This value is important for building models of CPC spreading at aster boundaries. We hypothesize that in zygotes, an autocatalytic reaction-diffusion system helps the CPC signal spread between microtubule bundles normal to their axis.

Figure 7. Model for CPC recruitment and functions at aster boundaries.

A) Summarizes molecular activities of the CPC, B-D) Illustrate possible functions of F-actin and keratin disassembly activities. A) Red arrows indicate a positive feedback loop that promotes recruitment and spreading of the CPC on microtubule bundles at aster boundaries [7, 2]. Green and blue negative arrows indicate AURKB-dependent inhibition of all three cytoskeletal systems, presumably via reaction-diffusion mechanisms. B) Disassembly of cytoskeletal networks between microtubule asters may help centrosomes and nuclei move apart following anaphase. C) Disassembly of cytoskeletal networks may help steer the ingressing furrow. D) Disassembly of bulk F-actin may supply the ingressing plasma membrane with subunits to build the new cortex, in a model where different F-actin assemblies compete for subunits.

The mechanism of local disassembly of F-actin by AURKB (green inhibitory arrow in Figure 7A) is not known. AURKB stimulates furrowing when it reaches the cortex, and did not inhibit what we assume are Arp2/3 actin comet tails at aster boundaries (Figure 2 A,B). Thus, it must negatively regulate bulk F-actin by a selective mechanism. One possibility is negative regulation of a formin that nucleates bulk cytoplasmic bundles, perhaps a homolog of FMN2 which nucleates bulk cytoplasmic actin in mouse eggs [25]. High magnification observations suggest that tension-driven sliding of F-actin away from regions of high AURKB activity may also contribute to loss of F-actin bundles at aster boundaries (Figure 2 and Video S4).

How AURKB destabilizes keratin filaments (green inhibitory arrow in Figure 6A) is also unknown. Direct phosphorylation by AURKB is plausible. Phosphorylation of intermediate filament polypeptides is a known mechanism for negatively regulating their assembly during cell division [15], and AURKB sites on keratins 5&14 during cytokinesis were recently mapped [17]. Forces exerted on keratin by the receding F-actin network may also contribute to splitting the keratin network at aster boundaries.

The biological function of microtubule plus end growth inhibition by CPC at aster boundaries is presumably to keep the asters separate, and to generate a sharp boundary that positions the cleavage furrow [6]. The function of F-actin and keratin disassembly by the CPC is currently unknown. We propose three non-exclusive models: help sister centrosomes move apart after mitosis by softening the cytoplasm between them (Figure 7B), provide a soft path to help the furrow ingress (Figure 7C) and/or free up G-actin subunits to build the cortex of the ingressing furrow and new plasma membrane (Figure 7D). The latter could be important given that different F-actin networks are known to compete for G-actin subunits [21]. We are currently analyzing centrosome movement in eggs treated with actin depolymerizing drugs to help distinguish these models.

Finally, our work may hold lessons for organization of syncytial organisms. The frog zygote between anaphase and furrow ingression can be viewed as a temporary syncytium, where both nuclei are replicating. The aster-dependent organization of cytoplasm into distinct islands observed in egg extract (Figure 1A, 2A) conceptually resembles the organization of early Drosophila embryos. During the syncytial blastoderm stage, the cytoplasm of Drosophila embryos is organized into discrete islands of cytoplasm centered on MTOCs, called energids, which function like autonomous cells in many respects [26, 27]. How energids are kept separate in the absence of plasma membrane barriers is unsolved. Our data shows that AURKB kinase activity localized on microtubule bundles at aster boundaries partitions Xenopus egg cytoplasm into discrete islands. It does so by generating gaps in the actin and keratin networks and by inhibiting microtubule plus end growth that would otherwise blur the boundary. We propose that AURKB, or functionally analogous kinases, may play similar roles in organization of proliferating syncytia.

STAR METHODS

CONTACT FOR RESOURCE SHARING

Further information and request for resources not already available should be directed to and will be fulfilled by the Lead Contact, Christine M Field (christine_field@hms.harvard.edu)

EXPERIMENTAL MODEL AND SUBJECT DETAILS

The Xenopus laevis frogs (adult females, ~ 3 years old) used in these studies are part of the Xenopus colony bred and maintained at Harvard Medical School under the direction of Dr. Marc Kirschner. The frogs are housed in flow-through ponds with at least 1 gallon of RODI treated water per frog. The frogs receive 100% water changes 3 days a week and are allowed at least 3 months to replenish oocytes between induced egg-laying cycles. Husbandry facilities and all experimental protocols involving frogs were reviewed and approved by the Harvard Medical School's Care and Use Committee. All studies using Xenopus laevis followed the guidelines of the U.S. Department of Health and Human Service for the Care and Use of Laboratory Animals, and all experiments were performed in accordance with national regulatory standards and ethical rules, Mitchison Lab IACUC Protocol number IS00000519-3.

METHOD DETAILS

Xenopus Egg Extracts

Mitotic (Cytostatic Factor, CSF) extract

Freshly made, actin-intact CSF egg extract is the starting point all experiments. Detailed protocols for extract preparation were described previously [28]. Briefly, when laid, unfertilized Xenopus eggs are arrested in metaphase of meiosis-II by CSF. Eggs were de-jellied, packed into tubes by gentle centrifugation to remove buffer, then crushed by sedimentation at 10,000g for 15 min at 4°C. This spin stratifies the egg contents into layers containing fat, cytoplasm and yolk. Undiluted cytoplasm, still arrested in mitosis by CSF, is collected with a syringe. Actin-intact mitotic extract is highly contractile[10]. Many researchers add cytochalasin-D during the crushing spin to prevent actin polymerization. Actin-intact extract has no cytochalasin-D addition. To prevent keratin aggregation, we stored extract at 10-12° C, then cooled briefly to 0° C before setting up reactions. Extract stored this way was typically usable for ~ 8 hrs.

Interphase (I-phase) extracts

The mitotic arrest induced by CSF arrest was released by the addition of calcium (40 μM final). This calcium transient triggers exit from mitosis by mimicking fertilization. The added calcium is rapidly sequestered by the abundant ER in the extract. We converted a small aliquot of extract to interphase just before each imaging experiment, and generally did not image for more than 65 min following calcium addition. A description of the cell cycle state of our extract system can be found in [29]. Judging by immunoblots of proteins that report on Cdk1 phosphorylation status, we conclude that calcium addition caused the extract to begin to enter the interphase state within a few minutes, but that the full interphase state was not reached until approximately 15-20 min after calcium addition.

Interphase aster assembly reactions

Methods are similar to those described previously [28, 29]. In a typical experiment fluorescent probes were added to CSF extract on ice. Calcium chloride was added to 0.4 mM final to trigger exit from mitosis, the reaction was mixed well, incubated at 18 °C for 5 min and returned to 0 °C. Anti-AURKA beads were added as MTOCs, and the reaction divided up for drug or vehicle addition. 6.5 μl aliquots were spotted onto the surface of a 22mm² glass coverslip coated with PEG-poly-lysine mounted on a 4-place metal holder. An 18 mm² similarly coated coverslip was immediately place on top. The resulting 20mm thick squash was sealed with molten VALAP and imaging was started immediately. Typically, we imaged 2-4 reactions in parallel using a 20x dry objective at a room temperature of 18-20°C, collecting sequences at 3-4 positions per each reaction for 1 hr. T=0 is the time the reaction was squashed and warmed to RT. Depending on the density of nucleating sites (anti-AURKA beads) asters typically grew into contact at 10-15 min and formed CPC-positive boundaries a few minutes later. Asters and boundaries were stable until ~ 60 min, and eventually decayed with loss of organized microtubules.

Anti-AURKA beads (Artificial Centrosomes)

Protein A Dynabeads (2.8 micron) are washed three times with Wash Buffer and then coated with saturating amounts of anti-AurkA IgG for 30 min at room temperature.Beads with antibody attached are then washed twice with Storage Buffer and stored at 4 degrees C. Beads are stable for months. Wash Buffer: 50 mM KCl, 10 mM K-HEPES pH 7.7, 1 mM MgCl₂, 1 mM EGTA. Storage Buffer: Wash Buffer plus 0.05% sodium azide, 100 mg/ml bovine serum albumin (BSA). BSA prevents bead aggregation and nonspecific adsorption; however, it should not be added prior to antibody binding to the beads due to possible contamination with bovine IgG [28, 30].

Fluorescently labeled tubulin

We standardly use fluorescently labeled bovine brain tubulin as an imaging probe for microtubules, adding it at 100-300 nM final concentration. Labeled frog egg tubulin gives better signal to background images but is harder to prepare [31]. Bovine brain tubulin is prepared as in [32]. Fluorescent labeling on random lysine residues on polymerized microtubules was performed as in [33]. We typically use Alexa 488, 568 and 647 NHS esters from Life Technologies for labeling. We also use a Tau-based fusion protein to image microtubules. See protein expression section below and [34]. High quality labeled tubulin can also be purchased from Cytoskeleton Inc.

Fluorescent protein probe expression and purification

Lifeact-GFP (actin), human EB1-GFP (growing microtubule plus ends), DasraA (CPC) and Tau-microtubule binding domain (microtubules) were all poly-His tagged constructs that were expressed in E coli, purified by affinity of the His tag for immobilized cobalt followed by gel filtration, and used without cleaving off the tag. A similar expression and purification protocol was used for all these His-tagged probe proteins. Expressing E coli cells were lysed by sonication in ice-cold Lysis buffer (50mM NaPi, pH 7.7, 500mM NaCl, 0.1mM MgCl2, 1% Triton X-100, 1mM dithiothreitol (DTT), 1mM phenylmethylsulfonyl fluoride (PMSF). Clarified lysates were batch incubated for 1.5 hr at 4°C with HisPur Cobalt resin (~2 mL/L of culture). After elution in Lysis Buffer containing 300mM imidazole, proteins were gel filtered using a Superdex 75 Increase 10/300 GL column equilibrated with 10mM K-HEPES, pH 7.7, 150mM KCl, and 1mM dithiothreitol (DTT). Proteins were concentrated by centrifugation against a 10K cut off filter as needed. Typical final concentrations were 1-10 mg/ml. They were then supplemented with 10% (w/v) sucrose, aliquoted, and flash frozen in liquid nitrogen for storage at −80°C. We typically prepared a 20x working stock in CSF egg extract on the day of use. These working stocks could be frozen-thawed once without loss of activity of the probe of the final reaction.

The tau-based microtubule probe SH-TMBD-mCherry was a gift from Jay Gatlin (University of Wyoming, Laramie, WY). Protein was expressed in BL21 (DE3)pLysS cells. For more information about SH-TMBD-mCherry see [34]

For information about Lifeact see [35,36]. The construct was a gift from David Burgess (Boston College, Newton, MA). The construct was amplified and isolated the pEGFP-N 1 mammalian expression vector and TOPO cloned into the E coli expression vector pEXP5-CT.

The EB1-GFP-His construct was a gift from Kevin Slep (UNC Chapel Hill, NC).The construct was cloned into E. coli expression vector Rosetta2(DE3)pLysS. [6]

The X. laevis DasraA gene (IRBHp990G0393D, Source Bioscience, UK was subcloned into pGex-4T-1 with an N-terminal 6x-eGFP tag, creating His-GFP-DasraA. That construct was cloned into E. coli expression vector Rosetta2(DE3)pLysS. [6]

Directed labeled antibodies

We use Protein G UltraLink resin or Affi-Prep Protein A resin. 40 μL of 50% resin slurry is loaded into a 200 μL polypropylene pipette tip (end previously sealed with heat).Resin washed 3x with 100 μL of 0.15 M NaCl, 10 mM K-HEPFS (pH 7.7) giving a packed resin bed of ~20 μL. A purified antibody is passed through the resin bed 3-5x sequentially to bind. For 20 μL of packed resin we typically label 20-100 μg IgG per tip. Resin is washed 3x with 100 μL 200 mM K-HEPES pH 7.7. Dye is dissolved in DMSO at ~50 mM. 0.5 μL of dye stock is diluted into 25 μL of 200 mM K-HEPES pH 7.7 and immediately loaded onto the resin bed. Incubate tip at room temperature for ~ 30-60 min. A 2^nd aliquot of dye stock is diluted and added to the resin and incubated as above. Wash the bed 5x with 0.15 M NaCl, 10 mM K-HEPES pH 7.7. Labeled IgG is then eluted with 5× 20 μL aliquots of 200 mM acetic acid. Mix each aliquot of eluate is immediately with 5 μL of 1 M Tris-Cl pH 9 to neutralize, and cool to 4 °C. A more detailed protocol is available in [28].

Passivated Coverslips

Coverslips: 18 × 18 or 22 × 22 mm were coated with poly-lysine PEG (PLL(20)-g{3.5}-PEG(2); SuSOS Chemicals) using a simplified cleaning method. Coverslips were dipped in 70% ethanol, flamed, cooled and placed on a droplet of 100 mg/ml poly-lysine- PEG in 10 mM HEPES, pH 7.4, on parafilm for 15-30 min. They were then washed twice with distilled water for 5 min each, dried with a jet of nitrogen and used the same day. For coverslips used in Figure 4, the top coverslip (not used for imaging) was 50 μg/ml poly-lysine-PEG and 50 μg/ml poly L-Lysine in 10 mM HEPES, pH 7.4. A more detailed protocol is available in [28].

Cytoplasmic flow experiments

To induce flow through the squash preparation we waited until CPC-positive boundaries had formed between asters, then placed a 10 gm weight (a bolt) on one side of the coverslip. After 30 sec the bolt was removed and a DIC movie was collected. Fluorescent images were collected immediately before and after inducing flow. For analysis of flow see Quantification and Statistical Analysis section.

Anti-INCENP bead (CPC bead) experiments

Protein A coated beads (2.8 μm Dynabeads or 6-8 μm polystyrene beads) were saturated with anti-INCENP IgG overnight and then washed with CSF-XB (10 mM K-HEPES, pH 7.7, 100 mM KCl mM MgCl₂, 0.1 mM CaCl₂, 50 mM sucrose and 5 mM EGTA). These CPC beads were added to interphase extract and imaged as per aster assembly reactions above. We typically waited 45 min and then imaged multiple fields up to 80 min final. When testing for Aurora kinase B dependence, inhibitors were added at the beginning of the reaction. See Table Sup1.

Zygote Immunofluorescence

Embryos were fixed and stained as previously described [6]. Briefly, embryos were fixed in 50 mM EGTA, ~ pH 6.8, 10% H2O, 90% methanol for 24 hrs at room temperature with gentle shaking. Prior to staining, embryos were rehydrated in a series of steps--- 25%, 50%, 75% and 100% TBS (50 mM Tris, pH 7.5, 150 mM NaCl/methanol: 15 min per step with gentle shaking. Embryos were then hemisected in TBS on an agarose cushion using a small piece of razor blade. Embryos were bleached overnight in a solution of 1% H2O2, 5% formamide, 0.5x SSC (75 mM NaCl and 8 mM sodium citrate, pH 7). Embryos were incubated with directly labeled antibodies for at least 24 hours at 4° C with very gentle rotation. Antibodies were diluted in TBSN (10 mM Tris-Cl, pH 7.4, 155 mM NaCl, 1% IGEPAL CA-630),1% BSA, 2% FCS and 0.1% Sodium Azide. After antibody incubation, embryos were washed in TBSN for at least 48 hr (with several solution changes) then washed 1 X in TBS and 2X in methanol (methanol washes for 20 min each). Embryos were cleared in Murray Clear solution (benzyl benzoate/benzyl alcohol 2:1) and mounted in metal slides (1.2 mm thick). The slides have a hole in the center. The hole was closed by attaching a coverslip to the bottom of the slide using heated parafilm.

Microscopy

Figures 1 and 3: Reactions were imaging using a 20x Plan Apo 1.4 NA dry objective on a Nikon Ti2 wide field, inverted microscope with a Perfect Focus System using a Nikon DS-Qi2 camera.

Figure 2: Reactions were imaged using a 60X oil immersion lens on a Yokogawa CSUX1 spinning disk confocal on a Nikon Ti inverted microscope with a Perfect Focus System and a Hamamatsu ORCA-R2 cooled CCD camera.

Figure 4: Reactions were imaged using a 40x oil immersion objective on a Nikon Ti2 wide field, inverted microscope with a Perfect Focus System using a Hamamatsu Flash4.0 V2+ sCMOS camera.

Figure 5 A,B and G: Reactions were imaged by wide-field fluorescence using a 10x or 20 x Pan Apo 1.4 NA dry objective on a Nikon Ti2 wide field, inverted microscope with a Perfect Focus System using a Hamamatsu Flash4.0 V2+ sCMOS camera.

Figure 5 E, F and H: Reactions were imaged using a 10x or 20 x Pan Apo 1.4 NA dry objective on an upright Nikon Eclipse 90i microscope with a Hamamatsu ORCA-ER cooled CCD camera.

Figure 5I: Reactions were imaged using a 60X oil immersion lens on a Yokogawa CSUX1 spinning disk confocal on a Nikon Ti inverted microscope with a Perfect Focus System and a Hamamatsu ORCA-R2 cooled CCD camera.

Figure 6: Fixed Xenopus embryos were imaged using a Nikon Ti-E inverted microscope with a Nikon A1R point scanning confocal head using 10x dry and 20x multi-immersion objectives.

QUANTIFICATION AND STATISTICAL ANALYSIS

Quantification of F-actin intensity near CPC beads

Intensity profiles of Lifeact-GFP signal around CPC beads were quantified using the Fiji [37] plugin “Radial Profile Extended” by Philippe Carl (https://imagej.nih.gov/ij/plugins/radial-profile-ext.html) with an integration angle of +/− 180 degrees (a full circle). Radial intensity profiles were normalized by dividing by the background intensity, in an unperturbed region far from the bead. The background intensity was estimated as the mean intensity of the last 10 points in each profile, about 80 mm from the bead. As a result, all radial intensity profiles approached the same background value. 15 beads analyzed per condition. Dark lines show average radial intensity values, pale regions show the standard deviation of intensity values. Line-scans were truncated below 7μm to avoid the strong bead edge signal.

Particle image velocimetry of cytoplasmic flow

Cytoplasmic flows in response to pressure transients were calculated using PIVlab 1.41, an open source MATLAB toolbox [38]. The PIV algorithm was FFT window deformation, and for the first pass the interrogation area was 128 px and the step was 64 px, and for the second pass the interrogation area was 64 px and the step was 32 px. We rejected vectors using a standard deviation filter with a threshold of 7 standard deviations, and we did not interpolate missing data. For display, we showed every second vector.

Supplementary Material

2

Video S1, AURKB-dependent disassembly of F-actin at boundaries between asters, related to Figure 1A A typical aster assembly reaction. Interacting asters recruit AURKB kinase to microtubule bundles between them. The appearance of AURKB on microtubule bundles correlates with the disassembly of F-actin.

3

Video S2, AURKB-dependent disassembly of F-actin at boundaries between asters, related to Figure 1A A higher magnification video of a portion of Videos S1.

4

Video S3, Aster assembly and interaction in extract with barasertib addition, related to Figure 1B The same reaction as in Videos S1 with barasertib, an AURKB kinase inhibitor added to the extract. F-actin does not disassemble.

5

Video S4, Two asters interacting, higher spatio-temporal resolution, related to Figure 2A Higher spatio-temporal resolution imaging of two asters interacting, Aurora B kinase recruitment between them and F-actin disassembly.

6

Video S5, Two asters interacting, higher spatio-temporal resolution, related to Figure 2B Examines an aster-aster interaction zone in the same aster assembly reaction as Video S4 but 9 minutes later. Images are collected at a much faster rate.

7

Video S6, Keratin disassembly at boundaries between asters, related to Figure 3A Interacting asters recruit AURKB kinase to microtubule bundles between them. The appearance of AURKB on microtubule bundles correlates with the disassembly of keratin.

8

Video S7, Keratin disassembly at boundaries between asters, related to Figure 3B A different keratin disassembly reaction than Video S6. Here asters interacted at a later time point. At that time, keratin structure is more developed.

9

Video S8 Cytoplasmic flow between asters is dependent on AURKB activity, related to Figure 4 Differential interference contrast microscopy (DIC) imaging of the movement of cytoplasm between asters in extract plus and minus barasertib, an AURKB kinase inhibitor.

10

Video S9, Disassembly of cytoskeleton networks around CPC beads, related to Figure 5D The disassembly of F-actin, keratin and microtubules around CPC beads.

11

Video S10, Formation of CPC-coated microtubule bundles near CPC beads, related to Figure 5I The appearance of bundles of CPC-coated microtubule bundles near CPC beads Figure 5I is made up of stills from this video.

12

Video S11, Formation of CPC-coated microtubule bundles near CPC beads, related to Figure 5I Another example of the appearance of CPC-coated microtubule bundles near CPC beads.Video S11 is from the same reaction as Video S10, but from a different region of the slide.

Highlights.

Aurora kinase B activity between asters disassembles actin and keratin networks Cytoskeleton disassembly results in reduced cytoplasmic stiffness between asters Aurora B kinase bound to beads caused local disassembly of cytoskeleton networks Microtubule binding locally amplifies Aurora kinase B activity