Research Article

A liquid-like spindle domain promotes acentrosomal spindle assembly in mammalian oocytes

See allHide authors and affiliations

Science  28 Jun 2019:
Vol. 364, Issue 6447, eaat9557
DOI: 10.1126/science.aat9557

A new phase in egg biology

Chromosome segregation typically requires centrosomes, which generate the microtubule spindle. However, mammalian eggs build a spindle and segregate chromosomes without centrosomes. How acentrosomal spindles are organized has remained elusive. So et al. show that centrosomal and microtubule-associated proteins are repurposed into a large “liquid-like meiotic spindle domain” (LISD) in eggs. The domains localized to spindle poles and also extended to the spindle fibers that connect to kinetochores. LISDs formed by phase separation and were required for spindle assembly, serving as reservoirs that locally sequester and mobilize spindle assembly factors within the large egg cytoplasm.

Science, this issue p. eaat9557

Structured Abstract

INTRODUCTION

Mammalian embryos frequently develop abnormally, resulting in miscarriages and genetic disorders such as Down syndrome. The major cause for aberrant embryonic development is chromosome segregation errors during egg meiosis. Unlike somatic cells and male germ cells, eggs segregate chromosomes with a specialized microtubule spindle that lacks centrosomes. Canonical centrosomes consist of a pair of centrioles surrounded by pericentriolar material and are the main microtubule organizing centers in centrosomal spindles. How acentrosomal spindles are organized in mammalian eggs is still poorly understood.

RATIONALE

Despite the absence of centrosomes, mammalian eggs express many centrosomal proteins. We set out to investigate systematically how these centrosomal proteins localize to acentrosomal spindles and organize microtubules in mammalian eggs.

RESULTS

We analysed the localization of 70 centrosomal and spindle-related proteins by combining high-resolution microscopy in live and fixed mouse eggs. Unexpectedly, 19 of these proteins localized to a domain that permeated a large region of the spindle and formed prominent spherical protrusions, which were dynamic, fused with each other, and extended well beyond the spindle poles. The domain included centrosomal proteins (AKAP450, CEP170, and KIZ), centriolar satellite proteins (CEP72, PCM1, and LRRC36), minus-end binding proteins (CAMSAP3 and KANSL3), dynein-related proteins (HOOK3, NDE1, NDEL1, and SPDL1), and proteins that control microtubule nucleation and stability (CHC17, chTOG, GTSE1, HAUS6, MCAK, MYO10, and TACC3). Proteins within this domain were dynamic and could redistribute rapidly throughout the entire spindle region. By combining in vitro and in vivo assays, we found that the domain forms by phase separation and behaves similar to a liquid. We hence termed it the liquid-like meiotic spindle domain (LISD). The LISD was also present in spindles in bovine, ovine, and porcine eggs and is thus widely conserved. Many LISD proteins have been studied extensively in mitosis, yet a similar structure has not been reported in somatic cells, suggesting that the LISD is likely exclusive to acentrosomal spindles in oocytes.

Assembly of the LISD was controlled by the regulatory kinase aurora A and dependent on the aurora A substrate TACC3 as well as the clathrin heavy chain CHC17, which binds to microtubules together with TACC3. Disruption of the LISD via different means released microtubule regulatory factors within this domain into the cytoplasm and led to severe spindle defects. Spindles were smaller and less stable and took longer to segregate chromosomes. Microtubule growth rates were significantly decreased, and their overall turnover was significantly increased. Both the microtubules that bind to the chromosomes’ kinetochores (kinetochore fibers) as well as microtubules that overlap in an antiparallel manner in the spindle midzone (interpolar microtubules) were strongly depleted. Together, these data establish that the LISD is required for efficient microtubule assembly and to form stable acentrosomal spindles.

CONCLUSION

Our data uncover a previously unknown principle of acentrosomal spindle assembly in mammalian eggs: Meiotic spindle assembly is facilitated by a prominent liquid-like domain that contains multiple microtubule regulatory factors and sequesters them in a dynamic manner in proximity to spindle microtubules.

Enriching microtubule regulatory factors in local proximity to the spindle may be particularly important in large cells such as eggs, where they would otherwise be dispersed throughout the cytoplasm. Liquid-liquid phase separation may be an ideal principle for such an enrichment: It sequesters factors within proximity to microtubules but still allows them to diffuse dynamically throughout the spindle. This could help to promote the even distribution of spindle assembly factors throughout the spindle and to titrate their local concentration to drive efficient spindle assembly within the large egg cytoplasm.

Acentrosomal spindle in a mouse egg.

A liquid-like meiotic spindle domain (LISD, cyan) forms prominent spherical protrusions at acentrosomal spindle poles and extends into the spindle region (magenta, right) toward chromosomes (magenta, left). The LISD forms by phase separation and is required for spindle assembly, serving as a reservoir that locally sequesters and mobilizes microtubule regulatory factors within the large egg cytoplasm. Scale bar, 5 μm.

Abstract

Mammalian oocytes segregate chromosomes with a microtubule spindle that lacks centrosomes, but the mechanisms by which acentrosomal spindles are organized and function are largely unclear. In this study, we identify a conserved subcellular structure in mammalian oocytes that forms by phase separation. This structure, which we term the liquid-like meiotic spindle domain (LISD), permeates the spindle poles and forms dynamic protrusions that extend well beyond the spindle. The LISD selectively concentrates multiple microtubule regulatory factors and allows them to diffuse rapidly within the spindle volume. Disruption of the LISD via different means disperses these factors and leads to severe spindle assembly defects. Our data suggest a model whereby the LISD promotes meiotic spindle assembly by serving as a reservoir that sequesters and mobilizes microtubule regulatory factors in proximity to spindle microtubules.

Once every menstrual cycle, an oocyte progresses through the first meiotic division to mature into a fertilizable egg. To this end, the oocyte eliminates half of its chromosomes in a small cell, called a polar body. The remaining chromosomes become aligned in the second metaphase spindle, and the egg is released into the fallopian tube, where it can be fertilized. Upon fertilization, the egg completes the second meiotic division, during which it eliminates half of the remaining sister chromatids into the second polar body. Subsequently, the male and female pronuclei form and progress toward each other, and the mitotic divisions of the embryo begin.

However, mammalian embryos frequently develop abnormally, resulting in miscarriages and genetic disorders such as Down syndrome. The major cause for aberrant embryonic development is aneuploidy in the egg, which results from chromosome segregation errors during oocyte meiosis. Unlike somatic cells and male germ cells, oocytes segregate chromosomes with a specialized microtubule spindle that lacks centrosomes (1). Canonical centrosomes consist of a pair of centrioles surrounded by pericentriolar material and are the main microtubule organizing centers in centrosomal spindles. These organelles serve as the major sites of microtubule nucleation and form the two poles of mitotic spindles. Oocytes have developed mechanisms to nucleate microtubules independently of centrosomes. For instance, in Drosophila oocytes, the augmin complex and subito (kinesin-6) mediate microtubule nucleation from the spindle poles and spindle equator, respectively, by recruiting the γ-tubulin complex (24). The γ-tubulin complex resides at the minus ends of microtubules and serves as a template for microtubule nucleation (5). Chromosomes can also serve as sites of microtubule nucleation, as studied most comprehensively in Xenopus egg extracts: They locally activate the small guanosine triphosphatase Ran, which releases spindle assembly factors from inhibitory binding to importins to promote local microtubule assembly (6).

How acentrosomal spindles are organized in mammalian oocytes is still incompletely understood. Despite the absence of centrosomes, mammalian oocytes express many centrosomal proteins (7). Some of these proteins have been mapped to the acentriolar microtubule organizing centers (aMTOCs) (table S1), which functionally replace centrosomes in mouse oocytes (8, 9). However, a comprehensive map of centrosomal protein localization in oocytes is lacking. Such a map would not only shed light on the roles of centrosomal proteins during oocyte meiosis but may also reveal previously unknown functions and subdomains of centrosomes in other cell types.

RESULTS

Identification of the LISD: A liquid-like meiotic spindle domain

We analyzed the localization of 70 centrosomal and spindle-related proteins in mouse metaphase I oocytes (Fig. 1A; fig. S1, A and B; and table S1). We identified several previously unknown aMTOC components, including CEP120, CP110, DISC1, KIF2B, MCRS1, and TOP2A. Of the 17 centriolar proteins that we examined, only CNTROB, CNAP1, and CP110 localized to aMTOCs, consistent with the absence of centrioles in oocytes (1). Proteins that constitute the pericentriolar material of centrosomes mostly localized to aMTOCs. Several centrosome-associated regulatory kinases and their substrates also localized to aMTOCs, and many of the mapped proteins showed enrichment to the overall spindle region.

Fig. 1 Identification of a previously unknown spindle domain in mammalian oocytes.

(A) Schematic representation of the mouse metaphase I spindle (scheme; LISD in green, aMTOCs in magenta, kinetochores in blue, Golgi in purple, and spindle microtubules in gray). (B) Immunofluorescence images of a mouse metaphase I spindle. Green, LISD (TACC3); magenta, aMTOCs (pericentrin); gray, microtubules (α-tubulin). (C) Immunofluorescence images of metaphase I and II spindles in mouse, bovine, ovine, and porcine oocytes. Green, LISD (TACC3); magenta, microtubules (α-tubulin). Scale bars, 5 μm.

Unexpectedly, we observed a previously undescribed domain that contains 19 proteins, including centrosomal proteins (AKAP450, CEP170, and KIZ), centriolar satellite proteins (CEP72, PCM1, and LRRC36), minus-end binding proteins (CAMSAP3 and KANSL3), dynein-related proteins (HOOK3, NDE1, NDEL1, and SPDL1) and proteins that control microtubule nucleation and stability (CHC17, chTOG, GTSE1, HAUS6, MCAK, MYO10, and TACC3) (fig. S1, A and B, and table S1). This domain permeated a large region of the spindle and formed prominent spherical protrusions that extended well beyond the spindle poles during late meiosis I as well as in the metaphase II spindle (Fig. 1, A to C) and was conserved in oocytes from other mammalian species (Fig. 1C and fig. S2).

Many of the proteins observed within this domain have been studied extensively in mitosis, yet similar structures have not been reported in somatic cells. To investigate whether the domain is a specific feature of acentrosomal spindles in mammalian oocytes, we depleted centrosomes in somatic cells by centrinone treatment (10) (fig. S3, A and B). Notably, mitotic spindles in centrosome-depleted mouse embryonic fibroblasts did not possess a related domain (fig. S3, B to D). This result suggests that the domain is not a general feature of acentrosomal spindles but is likely exclusive to female meiotic spindles. As we outline in detail below, this domain forms by phase separation and behaves similar to a liquid. We will therefore refer to it as the liquid-like meiotic spindle domain (LISD).

AURA, TACC3, and CHC17 are essential for LISD assembly

To investigate LISD assembly, we first examined the role of the regulatory kinases aurora A (AURA), polo-like kinase 1 (PLK1), and polo-like kinase 4 (PLK4), which control spindle assembly in multiple systems, including oocytes (11). Whereas pharmacological inhibition of PLK1 and PLK4 did not affect LISD assembly, inhibition of AURA led to disruption of the LISD (fig. S4, A to E). Of the LISD-associated proteins that we identified, only TACC3 and GTSE1 are known mitotic substrates of AURA (1215). Specific depletion of endogenous TACC3 by Trim-Away fully disassembled the domain (fig. S5, A to D), leading to complete dispersion of multiple LISD-associated proteins, including CAMSAP3 (minus-end stability) (16), chTOG (microtubule nucleation and stability) (17, 18), and GTSE1 (microtubule stability) (19, 20) (fig. S6). Some LISD-associated proteins, such as CHC17 (microtubule stability) (21), HAUS6 (microtubule nucleation) (22, 23), and KANSL3 (minus-end stability) (24), were no longer detected as prominent protrusions and were strongly depleted from the spindle but still showed some residual association with microtubules upon depletion of TACC3 (fig. S6). By contrast, depletion of GTSE1 had only a minor effect on LISD assembly (fig. S7, A and B). LISD assembly was also disrupted upon depletion of CHC17 (fig. S8, A and B), which binds to microtubules together with TACC3 (25, 26). TACC3 truncations (26) that are unable to interact with CHC17 (fig. S8C) or bind microtubules (fig. S8D) were not incorporated into the LISD, consistent with the codependence of LISD assembly on TACC3 and CHC17. Thus, AURA, TACC3, and CHC17 are essential for LISD assembly.

The LISD is a liquid-like assembly

To further analyze LISD assembly, we performed live imaging and immunofluorescence on mouse oocytes at different maturation stages. We used TACC3 as the reporter, as it is a core component of the LISD. The LISD accumulated within the center of the microtubule mass after nuclear envelope breakdown (NEBD), permeated the entire microtubule ball as the spindle grew and the aMTOCs fragmented, and translocated toward the two spindle poles together with the aMTOCs during spindle bipolarization (Fig. 2, A and B). As the bipolar spindle increased in size from early to late metaphase I, the LISD grew and extended beyond the spindle poles (Fig. 2, A and B). The LISD was not only enriched in the pole region but also formed projections toward the midzone in the metaphase I spindle (Fig. 1B and movie S1). Protein-retention expansion microscopy and short-term treatment with the EG5 inhibitor monastrol further demonstrated that the LISD was present throughout the spindle body (Fig. 2C and fig. S9), with enrichments on kinetochore fibers (K-fibers) confirmed by cold-stable assays and immunoelectron microscopy (Fig. 2C and fig. S10, A to D). Also, the LISD components CHC17 and PCM1 showed similar localization patterns as TACC3 throughout oocyte maturation (fig. S11, A to C).

Fig. 2 Dynamics of LISD assembly during meiosis I in mouse oocytes.

(A) Immunofluorescence images of mouse oocytes fixed at different times after nuclear envelope breakdown (NEBD). Gray, microtubules (α-tubulin); green, LISD (TACC3); magenta, aMTOCs (pericentrin). Insets are magnifications of regions outlined by dashed boxes. (B) Still images from time-lapse movies of meiotic maturation of mouse oocytes. Green, LISD (TACC3-mClover3); magenta, aMTOCs (CEP192-mScarlet). Time is given as hours:minutes after NEBD. (C) Single sections of expansion microscopy of a mouse early metaphase I spindle. Gray, LISD (TACC3). Yellow arrowheads highlight spherical protrusions at the spindle poles; white arrowheads highlight the K-fibers. Chr., chromosome. Scale bars, 5 μm.

High-resolution time-lapse microscopy of fluorescently labeled TACC3 further revealed that the spindle pole–associated protrusions of the LISD were dynamic and fused with each other, generating larger protrusions (Fig. 3A). Given that the protrusions of the LISD existed in a region that is largely devoid of microtubules, we explored whether the LISD can be maintained independently of microtubules. Indeed, mouse oocytes acutely treated with the microtubule-depolymerizing drug nocodazole transiently maintained the LISD, even after the bulk of the microtubules had disappeared (Fig. 3B and movie S2). The LISD eventually disappeared and then reassembled into new AURA-dependent spherical condensates after a few minutes (Fig. 3B, fig. S12, and movie S2). Endogenous TACC3 and PCM1 also formed these spherical condensates in oocytes fixed after nocodazole addition (Fig. 3C), demonstrating that they were not an artifact of ectopically expressed, fluorescently labeled TACC3.

Fig. 3 The LISD forms by phase separation.

(A) Stills from time-lapse movies of mouse late metaphase I oocytes. Gray, LISD (TACC3-mClover3). Yellow lines mark the position of xz planes on the corresponding xy planes. Arrowheads highlight fusing LISD protrusions. (B) Still images from time-lapse movies of acutely nocodazole-treated mouse metaphase I oocytes. Green, LISD (TACC3-mClover3); magenta, microtubules (EB3-3×mCherry). Time is given as minutes after 10 μM nocodazole addition. (C) Immunofluorescence images of spherical condensates in acutely 10 μM nocodazole–treated mouse metaphase I oocytes. Gray, microtubules (α-tubulin); green, spherical condensates (TACC3 or PCM1); magenta, aMTOCs (pericentrin). (D) Still images from time-lapse movies of acutely 10 μM nocodazole–treated mouse metaphase I oocytes. Gray, spherical condensates (TACC3-mClover3). Arrowheads highlight fusing spherical condensates. (E) Partial bleaching of TACC3-mClover3 in a spherical condensate in acutely 10 μM nocodazole–treated mouse metaphase I oocytes. The bleached area is outlined by the dashed box. Scale bar, 1 μm. (F) Photoactivation (PA) of different proteins on mouse metaphase I spindles. Spindle poles are outlined with white dashed lines; photoactivated bars are marked with yellow dashed lines. Time is given as minutes after photoactivation with a 405-nm laser. (G) Still images from time-lapse movies of acutely 1,6-hexanediol-treated mouse metaphase I oocytes pretreated with 10 μM nocodazole. Time is given as minutes after 3.5% 1,6-hexanediol addition. Scale bars, 5 μm unless otherwise specified.

Notably, even in the absence of microtubules, the spherical condensates fused with each other and formed larger foci (Fig. 3D and movie S3). This droplet-like behavior was reminiscent of properties of phase-separated structures reported in recent studies (27). We hence investigated whether the LISD might form by phase separation. Phase separation is characterized by the absence of membrane surrounding the phase-separated structure. To test for the absence of membrane, we established a workflow for focused ion beam–scanning electron microscopy (FIB-SEM) of mouse oocytes and imaged the meiotic spindle pole at isotropic resolution in three dimensions (fig. S13A). This analysis confirmed that the LISD was not enclosed by membranes (fig. S13, B and C).

Phase-separated structures are further characterized by their ability to rearrange internally. To test for rearrangement, we performed fluorescence recovery after photobleaching (FRAP) on half of a TACC3-labeled spherical condensate in nocodazole-treated oocytes. The TACC3 signal recovered most prominently directly adjacent to the bleached region, indicating diffusion of fluorescent TACC3 from within the nonbleached region into the bleached region. While the signal recovered in the bleached region, the nonbleached region became dimmer over time, consistent with internal rearrangement within the spherical condensate (Fig. 3E). Consistent with these results, photoactivated bars of TACC3 and PCM1 in intact spindles rapidly spread out bidirectionally within the LISD. By contrast, the photoactivated bar of the microtubule-associated protein TPX2, which is not a LISD component, moved slowly and unidirectionally toward the spindle pole (Fig. 3F). In addition to droplet-like behavior and internal rearrangement, proteins in the LISD and spherical condensates recovered after bleaching (fig. S14, A and B). Notably, proteins in the LISD recovered with similar, rapid kinetics (fig. S14B). By contrast, core aMTOC proteins such as CDK5RAP2, CEP192, and PCNT showed no prominent turnover on aMTOCs in metaphase I spindles (fig. S14C).

Depending on protein dynamics and reversibility, phase-separated structures can be further classified into liquid-like (dynamic and reversible), gel-like (arrested and reversible), and solid-like (arrested and irreversible) categories (28). As proteins in the LISD were highly dynamic (fig. S14B) and the LISD reversibly disassembled and reassembled upon nocodazole addition and washout (Fig. 2B and fig. S15A), we hypothesized that the LISD is liquid-like. To test this hypothesis further, we discriminated liquid-like from solid-like condensates in oocytes by treating them with 1,6-hexanediol, which dissolves liquid-like condensates by disrupting weak hydrophobic interactions (29). The spherical condensates in nocodazole-treated oocytes dispersed within minutes of 1,6-hexanediol addition (Fig. 3G). 1,6-hexanediol treatment also strongly depleted the LISD on intact spindles, though not completely (fig. S15, B and C). Finally, thioflavin T, which stains the amyloid-like interactions that are characteristic of solid-like condensates, did not stain the LISD (fig. S15D). Together, these data strongly suggest that the LISD is a liquid-like condensate.

TACC3 phase-separates via its N terminus

We discovered that the spherical condensates in nocodazole-treated oocytes were dependent on TACC3 but only mildly affected by CHC17 depletion (fig. S16, A and B). We thus tested whether TACC3 can phase-separate on its own in vitro. In the presence of the macromolecular crowding agent polyethylene glycol (PEG), recombinant TACC3 self-assembled into micrometer-sized spherical droplets over a range of pH values, ionic strengths, PEG concentrations, and protein concentrations (fig. S17, A to C). TACC3 droplets were able to wet glass surfaces, fuse into larger droplets, and exclude fluorescently labeled 70-kDa dextran (Fig. 4, A to C), consistent with characteristics of phase separation (27). In further support of liquid-liquid phase separation, TACC3 droplets rapidly recovered after bleaching (Fig. 4D and fig. S17D).

Fig. 4 The N terminus of TACC3 is necessary for phase separation in vitro and in vivo.

(A) Bright-field and fluorescence images of GST-TACC3 droplets in vitro. Inset is the magnification of the region marked by the dashed line box. (B) Still images from time-lapse movies of fusing GST-TACC3 droplets in vitro. Scale bar, 1 μm. Arrowheads highlight fusing GST-TACC3 droplets. (C) Fluorescence images of GST-TACC3 droplets in the presence of 70-kDa dextran in vitro. Green, GST-TACC3; magenta, 70-kDa dextran. Insets are magnifications of regions marked by dashed line boxes. Scale bar, 2.5 μm. (D) FRAP of GST-TACC3 droplets in vitro. Gray, GST-TACC3. The number of analyzed droplets is specified in italics. M.F., mobile fraction. Scale bar, 1 μm. (E) Domain organization of human and mouse TACC3 showing the disordered region [purple; analysis with DisEMBL (100)] and the coiled-coil domain [yellow; analysis with MARCOIL (101)]. (F) Fluorescence images of GST, GST-TACC3, GST-TACC3(ΔTACC), and GST-TACC in vitro. (G and H) Immunofluorescence images of the metaphase I spindle in control, TACC3-depleted, and TACC3(ΔNT)-TACC3–depleted mouse oocytes with and without cold treatment. IgG, immunoglobulin G. Insets are magnifications of regions marked by dashed line boxes. All in vitro assays were performed in pH 6.4 buffer with 150 mM KCl and 12% PEG. Scale bars, 5 μm unless otherwise specified.

In silico predictions suggested that both human and mouse TACC3 are bipartite, with a disordered N terminus and a structured, coiled-coil–containing C terminus (Fig. 4E). Both disordered and coiled-coil domains have been implicated in driving phase separation (30). To address which region of TACC3 mediates phase separation, we purified TACC3 fragments that contained only the coiled-coil TACC domain [amino acids (aa) 594 to 838] or the rest of the protein TACC3(ΔTACC) (aa 1 to 593). TACC3(ΔTACC) retained the ability to form droplets, similar to full-length TACC3 (Fig. 4F). By contrast, the TACC domain self-organized into network-like structures (Fig. 4F). These results reveal that the previously uncharacterized disordered N terminus of TACC3 is necessary and sufficient for phase separation in vitro.

Although the N terminus of TACC3 is required for its phase separation in vitro, it is dispensable for TACC3 localization in Xenopus mitotic extracts (15) and in mitotic cells (26). To determine whether the TACC3 N terminus is essential for LISD assembly, we expressed TACC3(ΔNT) (aa 522 to 838) in TACC3-depleted oocytes. TACC3(ΔNT) lacks the disordered N terminus but still contains the regions required to bind both CHC17 and microtubules. Although TACC3(ΔNT) localized properly to the spindle and associated with K-fibers, as confirmed by cold treatment, it failed to restore LISD assembly (Fig. 4, G and H). Instead, consistent with the behavior of the purified TACC domain in vitro (Fig. 4F), TACC3(ΔNT) and CHC17 assembled network-like structures at the spindle poles in vivo (Fig. 4G). These data indicate that the N terminus of TACC3 is essential for phase separation in vivo. Notably though, the N terminus of TACC3 is not sufficient for TACC3 recruitment to the LISD in vivo, as TACC3(ΔTACC) was not incorporated into the LISD (fig. S8D). Microtubule binding via the TACC domain is apparently required for LISD formation and may help to stabilize the condensates in vivo, as also suggested by the higher resistance of TACC3 condensates to 1,6-hexanediol treatment on intact spindles than in nocodazole-treated oocytes (Fig. 3G and fig. S15, B and C).

The LISD promotes acentrosomal spindle formation by sequestering microtubule regulatory factors

Phase separation can selectively enrich factors to promote reactions or storage (27). Recent studies proposed that the phase-separated Xenopus microtubule-binding protein BuGZ, regulatory kinase PLK4, and the Caenorhabditis elegans centrosomal protein SPD-5 promote microtubule nucleation in centrosomal spindles by locally enriching tubulin dimers by a factor of 4 to 10 (3133). By contrast, LISD protrusions and spherical condensates did not significantly concentrate tubulin dimers (fig. S18, A to E). However, microtubule dynamics were altered in TACC3-depleted oocytes: Microtubule growth rates were significantly decreased, and their overall turnover was significantly increased (fig. S19, A to D). We thus hypothesized that the LISD locally sequesters microtubule regulatory factors to promote spindle assembly. To investigate this hypothesis, we examined acentrosomal spindle assembly upon ablation of the LISD by three different means: inhibition of AURA, depletion of TACC3, and depletion of CHC17, all of which are essential for LISD assembly (see above). Under all conditions, the total microtubule intensity and spindle volume were severely reduced, to about half of the values in control oocytes (Fig. 5, A to C; figs. S20, A to E, and S21, A to I; and movies S4 and S5). Moreover, aMTOCs were no longer scattered around the spindle poles but coalesced into two single foci (figs. S22, A to K, and S23, A to I). Similarly, AURA inhibition and TACC3 depletion in bovine oocytes caused a severe reduction in microtubules as well as spindle assembly defects (Fig. 5, D to F, and fig. S24), in line with the abnormal spindles observed in a previous RNA interference (RNAi) study (34).

Fig. 5 Microtubule loss and defective spindle assembly in TACC3-depleted oocytes.

(A) Stills from time-lapse movies of control and TACC3-depleted mouse oocytes. Green, microtubules (mClover3-MAP4-MTBD); magenta, aMTOCs (CEP192-mScarlet); blue, chromosomes (H2B-miRFP). Time is given as hours:minutes after NEBD. (B and C) Quantification of total fluorescence intensity of microtubules and spindle volume in control and Tacc3-depleted mouse oocytes. (D) Immunofluorescence images of the metaphase I spindle in control and TACC3-depleted bovine oocytes. Gray, LISD (TACC3); green, microtubules (α-tubulin); magenta, chromosomes (Hoechst). (E and F) Quantification of total fluorescence intensity of microtubules and spindle volume in control and TACC3-depleted bovine metaphase I (MI) oocytes. The number of analyzed oocytes is specified in italics. Error bars (shaded areas) represent SD. Scale bars, 5 μm.

In mitotic cells, depletion of TACC3 causes a minor loss of spindle microtubules (17, 23) but results in misaligned chromosomes and metaphase arrest (25, 3538). To uncover LISD-specific functions, we took advantage of TACC3(ΔNT), which can neither phase-separate in vitro nor restore LISD assembly in vivo (Fig. 4G). We do not exclude the possibility that there are additional, phase separation–independent functions associated with the N terminus of TACC3. However, TACC3(ΔNT) has been shown to rescue the phenotypes induced by TACC3 depletion in Xenopus mitotic extracts and in mitotic cells, and the N terminus of TACC3 has hence been suggested to be dispensable for TACC3 function in these systems (15, 26). Notably, expression of TACC3(ΔNT) in TACC3-depleted mouse oocytes did not rescue the reduction in total microtubule intensity, and the spindle volume was still significantly reduced compared with that in wild-type oocytes, albeit not as strongly as in TACC3-depleted oocytes, as microtubules in the spindle appeared less densely packed and more spread out (Fig. 6, A to C, and movie S6). Moreover, spindle bipolarization was severely delayed in TACC3(ΔNT)-TACC3–depleted oocytes relative to wild-type oocytes, and spindles progressed through a prolonged phase of spindle instability and fragmentation before becoming bipolar (Fig. 6A and fig. S25, A and B). Also the association between aMTOCs and spindles in TACC3(ΔNT)-TACC3–depleted oocytes was abnormal (fig. S25C). By contrast, TACC3(ΔNT) rescued progression into anaphase and markedly reduced lagging chromosomes in TACC3-depleted oocytes, demonstrating that the construct is functional (fig. S25, D to G). Together, these data establish that the LISD is required for efficient microtubule assembly and to form stable acentrosomal spindles.

Fig. 6 The N terminus of TACC3 is required for spindle assembly in mouse oocytes.

(A) Still images from time-lapse movies of control, TACC3-depleted, and TACC3(ΔNT)-TACC3–depleted mouse oocytes. Green, microtubules (mClover3-MAP4-MTBD); magenta, aMTOCs (CEP192-mScarlet); blue, chromosomes (H2B-miRFP). Time is given as hours:minutes after NEBD. Yellow asterisks highlight the unfocused spindle intermediate. (B and C) Quantification of total fluorescence intensity of microtubules and spindle volume in control, TACC3-depleted, and TACC3(ΔNT)-TACC3–depleted mouse oocytes. The number of analyzed oocytes is specified in italics. Error bars (shaded areas) represent SD. Scale bars, 5 μm.

K-fibers and interpolar microtubules are depleted upon LISD disruption by TACC3 depletion

To further investigate the role of the LISD in spindle assembly, we examined K-fibers and interpolar microtubules. To assess K-fibers, we briefly placed oocytes on ice, which depolymerizes dynamic microtubules within the spindle while preserving the more stable, kinetochore-bound microtubules. K-fibers were significantly diminished in TACC3-depleted oocytes (Fig. 7, A to C). To assess interpolar microtubules, we treated oocytes with calcium, which preserves both K-fibers and stable interpolar microtubules in mouse oocytes (39). Stable interpolar microtubules were strongly depleted in TACC3-depleted oocytes (Fig. 7D and movies S7 and S8). Quantification revealed a more severe microtubule loss than in the cold-stable assay (Fig. 7, E and F). We also labeled spindles with the microtubule cross-linker PRC1, which associates with interpolar microtubules (40). Consistent with the calcium-stable assay, PRC1-marked microtubules were drastically reduced in TACC3-depleted oocytes (Fig. 7, G and H, and movies S9 and S10). In addition, oocytes in anaphase displayed a significant reduction in the total intensity and volume of the central spindle (fig. S26, A to C), which consists of interpolar microtubules and is largely devoid of K-fibers (41). Thus, both K-fibers and interpolar microtubules are depleted upon LISD disruption by TACC3 depletion.

Fig. 7 Loss of K-fibers and interpolar microtubules in TACC3-depleted oocytes.

(A) Immunofluorescence images of the metaphase I spindle in cold-treated control and TACC3-depleted mouse oocytes. Gray, LISD (TACC3). Green, microtubules (α-tubulin); magenta, kinetochores (ACA); blue, chromosomes (Hoechst). (B and C) Quantification of total fluorescence intensity of cold-stable microtubules and spindle volume in cold-treated control and TACC3-depleted mouse metaphase I oocytes. (D) Immunofluorescence images of the metaphase I spindle in calcium-treated control and TACC3-depleted mouse oocytes. Green arrowheads highlight stable interpolar microtubules; magenta arrowheads indicate K-fibers. (E and F) Quantification of total fluorescence intensity of calcium-stable microtubules and spindle volume in calcium-treated control and TACC3-depleted mouse metaphase I oocytes. (G) Immunofluorescence images of PRC1-3×mClover3 in control and TACC3-depleted mouse metaphase I oocytes. Green, PRC1 (GFP); magenta, TPX2. (H) Quantification of total fluorescence intensity of PRC1-marked microtubules in control and TACC3-depleted mouse oocytes. The number of analyzed oocytes is specified in italics. Error bars (shaded areas) represent SD. Scale bars, 5 μm.

Discussion

Our data uncover a previously unknown principle of acentrosomal spindle assembly in mammalian oocytes: Meiotic spindle assembly is facilitated by a prominent liquid-like domain that contains multiple microtubule regulatory factors and enriches them in a dynamic manner in proximity to spindle microtubules. The domain permeates the spindle poles and forms dynamic protrusions extending outward well beyond the spindle region in mouse, bovine, ovine, and porcine oocytes. Ablating the domain results in microtubule loss and defective spindle assembly.

Enriching microtubule regulatory factors in local proximity to the spindle may be particularly important for spindle assembly in large cells such as oocytes, where they would otherwise be dispersed throughout the large cytoplasmic volume. Liquid-liquid phase separation may be an ideal principle for this purpose: It sequesters factors within proximity to microtubules but still allows them to diffuse dynamically throughout the spindle region. This could also help to control the local concentration of certain spindle-related factors and prevent them from accumulating within specific spindle regions such as the spindle poles or kinetochores but may instead promote their even distribution throughout the entire spindle volume.

Does the LISD play a role in spindle assembly beyond mammalian oocytes? Structures that resemble the LISD have not yet been reported in other cell types (42), although many LISD proteins have been extensively studied in mitotic cells. Given the absence of LISD-related structures in centrosome-depleted mitotic cells, it appears likely that the LISD is an oocyte-specific structure. Notably though, when TACC3 is highly overexpressed in somatic cells, it forms large structures that can associate with the mitotic spindle (43). Whether phase separation also plays a role in the formation of these structures is unclear.

In centrosomal spindles, phase separation has been proposed to enrich tubulin dimers to facilitate microtubule nucleation (3133). However, the LISD does not concentrate tubulin dimers but enriches different microtubule regulatory factors in acentrosomal spindles. One could speculate that an enrichment of tubulin dimers is less important in large oocytes, as tubulin is not limiting in large cytoplasmic volumes (44, 45). Indeed, recent studies imply microtubule dynamics as a key factor in determining spindle size when tubulin is not limiting (46). Mammalian oocytes may use phase separation of microtubule regulatory factors as a mechanism to modulate microtubule dynamics and to thereby promote the assembly of large meiotic spindles. The LISD differs from other phase-separated structures in terms of morphology, composition, and mechanism of action, underscoring that phase separation is a widely used principle that can promote spindle assembly via distinct mechanisms in different species and cell types.

Materials and methods

Preparation and culture of mouse oocytes and follicles

All mice were maintained in a specific pathogen-free environment according to animal ethics guidelines of the Animal Facility of the Max Planck Institute for Biophysical Chemistry and U.K. Home Office regulations.

Oocytes were isolated from ovaries of 8- to 12-week-old FVB/N or CD1 female mice. For some experiments that required large numbers of germinal vesicle (GV) oocytes, the mice were primed with 7.5 IU of pregnant mare serum gonadotropin 48 hours before isolation. Fully grown oocytes of ~75 μm in diameter with a centered GV were maintained at prophase arrest in homemade phenol red-free M2 medium supplemented with 250 μM dibutyryl cyclic AMP (dbcAMP) (Sigma-Aldrich) under paraffin oil (ACROS Organics) at 37°C.

Follicles were isolated from 12-day-old (C57BL × CBA) F1 female mice as previously described (47) with some minor modifications. Briefly, compact follicles of ~100 μm in diameter with a centered oocyte were cultured in MEM-alpha with GlutaMax (Gibco) supplemented with 5% fetal bovine serum (FBS) (Gibco), 0.01 μg/ml ovine follicle stimulating hormone (National Hormone and Peptide Program), 1× insulin/transferrin/sodium selenite (Sigma-Aldrich), and 0.1× penicillin G/streptomycin (Gibco) on collagen-coated inserts (Corning) at 37°C/5% CO2. Medium surrounding the insert was replaced every 3 days. After 8 days of culture, in vitro grown oocytes were denuded and matured in modified M2 medium with 10% FBS instead of 4 mg/ml bovine serum albumin (BSA).

Preparation and culture of bovine, ovine, and porcine oocytes

All ovaries were obtained from local slaughterhouses. Bovine ovaries were transported to the laboratory within 2 hours of retrieval in a thermo-flask. Ovine and porcine ovaries were transported to the laboratory within 1 hour of retrieval in a portable 37°C incubator in M2 medium supplemented with 1 mM dbcAMP. Oocytes were recovered by aspiration of antral follicles with an 18-gauge needle affixed to a 1-ml syringe. For bovine oocytes, 140 μl of 5000 IU/ml heparin (Merck Millipore) was added to every 20 ml of aspirates. Cumulus-oocyte complexes (COCs) were allowed to sediment and then washed extensively with HEPES-buffered medium 199 (for bovine oocytes) or M2 medium (for ovine and porcine oocytes) supplemented with 1 mM dbcAMP. Only fully grown oocytes with a homogeneous cytoplasm and several layers of compact cumulus cells were selected for experiments. Bovine and ovine/porcine oocytes were maintained in prophase arrest in dbcAMP-containing medium at 39° and 37°C, respectively. Surrounding cumulus cells of bovine and ovine/porcine oocytes were removed by vortexing and with 30 nM hyaluronidase (Sigma-Aldrich), respectively. Ovine and porcine oocytes were released into dbcAMP-free M2 medium to resume meiosis.

Cell culture

NIH3T3 cells (ATCC) were cultured in high-glucose DMEM supplemented with GlutaMAX, pyruvate, and 10% calf serum (Gibco). To deplete centrosomes, cells were treated with 300 nM centrinone (Tocris Bioscience) for 12 days. To obtain mitotic spindles, cells were synchronized with 10 μM RO-3306 (Tocris Bioscience) for 1 day and then released into RO-3306-free medium for 45 min.

Immunofluorescence

To obtain mouse and ovine/porcine metaphase I spindles, oocytes were incubated at 37°C for around 7 and 12 hours, respectively, upon release into dbcAMP-free medium. To obtain bovine metaphase I spindles, oocytes were incubated at 39°C/5% CO2 for ~12 hours upon release into dbcAMP-free medium. To obtain mouse, bovine, ovine and porcine metaphase II spindles, oocytes were incubated for ~16 hours upon release into dbcAMP-free medium. None of the oocytes used for immunofluorescence analyses was subjected to live imaging before fixation.

Oocytes and cells were fixed in 100 mM HEPES (pH 7.0, titrated with KOH), 50 mM EGTA (pH 7.0, titrated with KOH), 10 mM MgSO4, 2% methanol-free formaldehyde, and 0.5% triton X-100 at 37°C for 15 to 60 min. Fixed oocytes and cells were extracted in phosphate-buffered saline (PBS) with 0.5% triton X-100 (PBT) overnight at 4°C and blocked in PBT with 5% BSA (PBT-BSA) overnight at 4°C. All antibody incubations were performed in PBT-BSA at 10 μg/ml overnight at 4°C (for primary antibodies) or at 20 μg/ml for 1 hour at room temperature (for secondary antibodies). Primary antibodies used were human anti-centromere antibody (ACA) (FZ90C-CS1058; Europa Bioproducts), rabbit anti-AKAP450 (NBP1-89167; Novus Biological), rabbit anti-ASPM (NB100-2278; Novus Biological), rat anti-α-tubulin (MCA78G; Bio-Rad), mouse anti-AURA (NBP2-50041; Novus Biological), rabbit anti-pT288-AURA (NB100-2371; Novus Biological), rabbit anti-CAMSAP3 (48), rabbit anti-CDK5RAP2 (ABE236; Merck Millipore), mouse anti-pan centrin (04-1624; Merck Millipore), rabbit anti-CEP63 (06-1292; Merck Millipore), rabbit anti-CEP120 (PA5-55985; Thermo Fisher Scientific), rabbit anti-CEP135 (ab75005; Abcam), rabbit anti-CEP152 (ab183911; Abcam), rabbit anti-CEP164 (HPA037606; Sigma-Aldrich), rabbit anti-CEP170 (27325-1-AP; Proteintech), rabbit anti-CEP192 (18832-1-AP; Proteintech), mouse anti-CHC17 (610500; BD Biosciences), rabbit anti-CHC17 (ab21679; Abcam), rabbit anti-chTOG (PA5-59150 and PA5-58763; Thermo Fisher Scientific), rabbit anti-CNAP1 (14498-1-AP; Proteintech), rabbit anti-CP110 (12780-1-AP; Proteintech), rabbit anti-DHC (12345-1-AP; Proteintech), rabbit anti-DISC1 (NB110-40773; Novus Biological), goat anti-GFP (600-101-215; Rockland Immunochemicals), mouse anti-GM130 (610822; BD Biosciences), mouse anti-GMAP210 (611712; BD Biosciences), rabbit anti-GTSE1 (20), rabbit anti-GTSE1 (A302-425A; Bethyl Laboratories), mouse anti-γ-tubulin (11-465-C025; Exbio), rabbit anti-HAUS6 (49), rabbit anti-HOOK3 (NBP2-44279; Novus Biological), rabbit anti-KANSL3 (HPA035018; Sigma-Aldrich), rabbit anti-KIF2A (NB500-180; Novus Biologicals), rabbit anti-KIZ (21177-1-AP; Proteintech), mouse anti-LIS1 (H00005048-M03; Abnova), rabbit anti-MCRS1 (HPA039057; Sigma-Aldrich), sheep anti-MCAK (50), rabbit anti-MKLP2 (51), goat anti-MYO10 (sc-23137; Santa Cruz Biotechnology), rabbit anti-NDE1 (10233-1-AP; Proteintech), rabbit anti-NDEL1 (H00081565-D01P; Abnova), mouse anti-NEDD1 (H00121441-M05; Abnova), rabbit anit-ninein (ABN1720; Merck Millipore), mouse anti-NUMA (610561; BD Biosciences), rabbit anti-pS395-NUMA (3429; Cell Signaling Technology), rabbit anti-NUMA (14951; Cell Signaling Technology), rabbit anti-ODF2 (12058-1-AP; Proteintech), mouse anti-P50 (611002; BD Biosciences), mouse anti-P150 (612708; BD Biosciences), rabbit anti-PCM1 (52), rabbit anti-PCM1 (HPA023374; Sigma-Aldrich), mouse anti-pericentrin (611814; BD Biosciences), rabbit anti-pericentrin (ab4448; Abcam), mouse anti-PLK1 (ab17056; Abcam), mouse anti-pT210-PLK1 (558400; BD Biosciences), goat anti-PLK4 (NB100-894; Novus Biological), rabbit anti-SSX2IP (HPA027306; Sigma-Aldrich), rabbit anti-STIL (ab89314; Abcam), rabbit anti-SPDL1 (53), goat anti-TACC3 (AF5720-SP; R&D Systems), mouse anti-TACC3 (H00010460-M02; Abnoa), rabbit anti-TACC3 (ab134154; Abcam), rabbit anti-TGN46 (ABT95; Merck Millipore), mouse anti-TOP2A (MAB4197; Merck Millipore), and rabbit anti-TPX2 (NB500-179; Novus Biological). Secondary antibodies used were Alexa Fluor 405-, 488-, 568-, or 647-conjugated anti-human IgG, goat IgG, mouse IgG, mouse IgM, rabbit IgG, rat IgG, or sheep IgG (Molecular Probes). DNA was stained with Hoechst 33342 (Molecular Probes).

Optical clearing of bovine, ovine, and porcine oocytes

Oocytes were fixed, extracted, and blocked as for routine immunofluorescence. Before incubation with primary antibodies, lipid droplets in oocytes were cleared with 4000 U/ml lipase from Candida rugose (Sigma-Aldrich) in 400 mM NaCl, 50 mM tris (pH 7.2), 5 mM CaCl2, and 0.2% sodium taurocholate supplemented with complete, EDTA-free Protease Inhibitor Cocktail (Roche) at room temperature for 20 to 40 min.

Protein-retention expansion microscopy

Expansion microscopy was performed as previously described for cell and tissue samples (54) using the Expansion Microscopy Kit (Expansion Technologies).

Expression constructs, mRNA synthesis, protein expression, and purification

To generate constructs for mRNA synthesis, we fused previously published coding sequences with meGFP (Clonetech), mClover3 or 3×mClover3 (55), mPA-GFP (56), 3×CyOFP (57), mCherry (58), mScarlet (59), or miRFP (60) and subcloned them into pGEMHE (61) to obtain meGFP-AKAP450 (62), AURA-meGFP (63), Bbs4-meGFP (64), meGFP-BBS14 (65), TurboGFP-BUGZ (66), mCherry-CDK5RAP2 (67), CEP72-meGFP (68), mScarlet-CEP192 (69), CETN2-meGFP (70), meGFP-centrobin (71), mClover3-CHC17 (21), chTOG-mScarlet (72), mEmerald-CLIP170 (M. Davidson,Florida State University, FL), meGFP-CPAP (73), mClover3-GTSE1 (72), H2B-miRFP (4), meGFP-HOOK3 (74), meGFP-KIF2B (75), LRRC36-meGFP (68), LRRC45-meGFP (76), mClover3-MAP4-MTBD (77), 3×CyOFP-MAP4-MTBD (77), NEDD1-mCherry (78), meGFP-NEK2A (79), meGFP-P50 (80), meGFP-P150 (81), meGFP-PAR6α (82), PCM1-mClover3 (Source BioScience), PCM1-mPA-GFP (Source Bioscience), meGFP-Pericentrin (83), PRC1-3×mClover3 (72), meGFP-rootletin (84), meGFP-SAS6 (85), TACC3-mClover3 (72), TACC3-mPA-GFP (72), mPA-GFP-TPX2 (86), (bovine) TRIM21 (Thermo Fisher Scientific), (mouse) TRIM21 (87), and γ-tubulin-meGFP (88). EB3-3×mCherry was subcloned from pEB3-mCherry (J. Ellenberg, EMBL, Heidelberg, Germany) into pGEMHE. pGEMHE-TACC3(ΔCID)-mClover3, pGEMHE-TACC3(ΔTACC)-mClover3, and pGEMHE-TACC3(ΔNT) were constructed from pGEMHE-TACC3-mClover3 using QuikChange Lightning Multi Site-Directed Mutagenesis Kit (Agilent) or In-Fusion HD Cloning Plus (Clonetech) or Q5 Site-Directed Mutagenesis Kit (NEB). pGEMHE-mCherry-MAP4-MTBD (77), pGEMHE-mPA-α-tubulin (77), pGEMHE-CEP250-meGFP (89) and pGEMHE-mCherry-PLK1 (89) were also use. All mRNAs were synthesized and quantified as previously described (90).

To generate constructs for protein expression, TACC3(ΔTACC) (Δaa 594-838)-His6 and TACC (aa594-838)-His were inserted into pGEX-6p-1 (GE Healthcare), and His6-TACC3-His6 was inserted into pET-28c(+) (Novagen). GST-TACC3-His6, GST-TACC3(ΔTACC)-His6, GST-TACC-His6 were expressed in and purified from BL21(DE3)pLysS Competent Cells (Promega) as previously described (26) with some modifications. Briefly, recombinant proteins were purified first using Ni-NTA Agarose (Qiagen), followed by Q Sepharose Fast Flow (GE Healthcare). His6-TACC3-His was expressed in NiCo21(DE3) (NEB) and purified first using Ni-NTA Agarose, followed by Chitin Resin (NEB).

Short-interfering RNAs

All short interfering RNAs (siRNAs) were purchased from Qiagen. For knockdown of TACC3 using RNAi, a mix of the following siRNAs were used: 5′-AAGTCCTAACATGACCAATAA-3′, 5′-CTGCATGTCTTAAATGACGAA-3′, 5′-AAGACTAAAGTTTAATCTCAA-3′, and 5′-AAGGAAATAGCTGAAGACAAA-3′. AllStars Negative Control (Qiagen) was used as a control.

Microinjection of mouse oocytes, follicles, and bovine oocytes

Mouse oocytes were microinjected with 3.5 pl of mRNAs as previously described (4). mClover3-MAP4-MTBD mRNA was microinjected at a needle concentration of 83.5 ng/μl, PCM1-mPA-GFP mRNA at 165.8 ng/μl, PRC1-3×mClover3 mRNA at 97 ng/μl, TACC3-mClover3 mRNA at 208.1 ng/μl, TACC3-mPA-GFP at 277.4 ng/μl, mPA-GFP-α-tubulin mRNA at 250.5 ng/ μl, 3×CyOFP-MAP4-MTBD mRNA at 166.4 ng/μl, CEP192-mScarlet mRNA at 165.8 ng/μl, EB3-3×mCherry mRNA at 138.8 ng/μl, H2B-miRFP mRNA at 28.4 ng/μl, and Trim21 mRNA at 421 ng/μl. For other mRNAs, they were individually titrated and care was taken that they did not induce phenotypes due to overexpression. Oocytes were allowed to express mRNAs for 3 to 4 hours before released into dbcAMP-free medium for imaging.

Mouse follicles were microinjected with 6 pl of siRNAs at a needle concentration of 2 μM as previously described (47).

Bovine oocytes were microinjected with 7 pl of TACC3-mClover3 mRNA and bovine TRIM21 mRNA at a needle concentration of 208.1 and 421 ng/μl, respectively. Oocytes were allowed to express mRNAs for 3 to 9 hours before released into dbcAMP-free, bicarbonate-buffered BO-IVM (IVF Bioscience).

Peptide preincubation assay

Oocytes were fixed, extracted, and blocked as for routine immunofluorescence. Rabbit polyclonal anti-CHC17, polyclonal anti-GTSE1, and monoclonal anti-TACC3 were preincubated with peptide PQAPFGYGYTAPPYGQPQPGFGYSM (GenScript), recombinant GTSE1 (OriGene), and recombinant GST-TACC3-His6 (homemade), respectively, before applied to oocytes as previously described (90).

Trim-Away in mouse and bovine oocytes

Only affinity-purified antibodies were used in this study for Trim-Away–mediated protein depletion. Rabbit polyclonal anti-CHC17, polyclonal anti-GTSE1, and monoclonal anti-TACC3 were purified as previously described (90). The control IgG used was normal rabbit IgG (12-370; Millipore).

For constitutive Trim-Away in mouse GV oocytes, 3.5 pl of mRNAs and 3.5 pl of antibodies were microinjected as previously described (90). All antibodies were microinjected at a needle concentration of 1 mg/ml with 0.1% NP-40. Target proteins were allowed to be depleted for 3 to 4 hours before the oocytes were released into dbcAMP-free medium for imaging. For acute Trim-Away in mouse metaphase I oocytes, oocytes were first microinjected with 3.5 pl of mRNAs and then released into dbcAMP-free medium after 3 hours. At ~6 hours upon release, oocytes were further microinjected with 3.5 pl of antibodies immediately before imaging.

For constitutive Trim-Away in bovine GV oocytes, 7 pl of bovine TRIM21 mRNA and 7 pl of anti-TACC3 at a needle concentration of 2 mg/ml with 0.1% NP-40 were microinjected. TACC3 was allowed to be depleted for 3 to 9 hours before the oocytes were released into dbcAMP-free, bicarbonate-buffered BO-IVM.

Immunoblotting

Thirty mouse metaphase I oocytes per lane were washed in BSA-free M2 medium and resuspended in 4 μl of BSA-free M2 medium. 12 μl of 1.333× NuPAGE LDS sample buffer (Thermo Fisher Scientific) with 100 mM DTT was then added, and the mixture was immediately snap-frozen in liquid nitrogen for 5 min. Samples were thawed and frozen twice more before heated at 100°C for 5 min. Samples were resolved on a 15-well NuPAGE 4 to 12% bis-tris protein gel of 1.0 mm thickness (Thermo Fisher Scientific) with NuPAGE MOPS SDS Running Buffer (Thermo Fisher Scientific). Proteins were transferred onto a 0.45-μm PVDF membrane with SDS-free Towbin buffer at 200 mA for 2 hours on ice. Blocking and antibody incubations were performed in tris-buffered saline (TBS) with 5% skim milk and 0.1% tween-20. Primary antibodies used were mouse anti-AURA (610938; BD Biosciences), rabbit anti-chTOG, mouse anti-CHC17 rabbit anti-TACC3, and rat-α-tubulin. Secondary antibodies used were HRP-conjugated anti-mouse (P0447; Dako), anti-rabbit (31462; Invitrogen), and anti-rat (sc-2032; Santa Cruz Biotechnology). Blots were developed with SuperSignal West Femto Maximum Sensitivity Substrate (Thermo Fisher Scientific) and documented with Amersham Imager 600 (GE Healthcare). Care was taken that the exposure time did not cause saturation.

Confocal and super-resolution microscopy

For confocal imaging, oocytes were imaged in 1 to 2 μl of M2 medium (for live mouse oocytes) or PBS with 1% polyvinylpyrrolidone (PVP) and 0.5 mg/ml BSA (for fixed oocytes) under paraffin oil in a 35-mm dish with a #1.0 coverslip. Images were acquired with LSM 880 confocal laser scanning microscopes (Zeiss) equipped with an environmental incubator box and a 40× C-Apochromat 1.2 NA water-immersion objective. A volume of 50 μm by 50 μm by 37.5 μm or 35 μm by 35 μm by 37.5 μm centered around the chromosomes was typically recorded. Automatic 3D tracking was implemented for time-lapse imaging with a temporal resolution of 5 to 15 min using Autofocuscreen (91) or MyPiC (92). mClover3 and meGFP were excited with a 488-nm laser line and detected at 493 to 571 nm. CyOFP was excited with a 488-nm laser line and detected at 571 to 638 nm. mScarlet and mCherry were excited with a 561-nm laser line and detected at 571 to 638 nm. miRFP was excited with a 633-nm laser line and detected at 638 to 700 nm. Images of the control and experimental groups were acquired under identical imaging conditions on the same microscope. For some images, shot noise was reduced with a Gaussian filter. Airyscan images were acquired using the Airyscan module on LSM800 and LSM880 confocal laser scanning microscopes (Zeiss) and processed in ZEN (Zeiss) after acquisition. Care was taken that the imaging conditions (laser power, pixel-dwell time, and detector gain) did not cause phototoxicity (for live imaging), photobleaching, or saturation.

Drug addition and washout

All drugs except 1,6-hexanediol were prepared in DMSO (Sigma-Aldrich) as 1000× stocks. To depolymerize microtubules, metaphase I oocytes were treated with 10 μM nocodazole (Sigma-Aldrich). To regrow microtubules, metaphase I oocytes were first washed into 10 μM nocodazole for 45 to 60 min and then washed out into medium without drug. To inhibit EG5, metaphase I oocytes were treated with 100 μM monastrol (Sigma-Aldrich) for 30 to 90 min. To inhibit AURA, PLK1, and PLK4 in metaphase I oocytes, MLN8237 (Selleckchem), BI2536 (Selleckchem) and centrinone were used at 500 nM, 100 nM, and 5 μM, respectively. Previous studies have shown that these drugs do not exhibit nonspecific inhibition on other kinases at the indicated concentrations (5, 89, 93). 1,6-hexanediol (Sigma-Aldrich) was prepared as a 10× stock in medium and used at a final concentration of 3.5%.

Electron microscopy staining, immunoelectron microscopy, and focused ion beam–scanning electron microscopy

For electron microscopy staining, mouse metaphase I oocytes were fixed as previously described for mitotic cells (94) with some minor modifications. Briefly, oocytes were fixed in 100 mM HEPES (pH 7.0, titrated with KOH), 50 mM EGTA (pH 7.0, titrated with KOH), 10 mM MgSO4, 3% sucrose, 3% EM-grade glutaraldehyde, 0.5% methanol-free formaldehyde, and 0.1% tannic acid at 37°C for 1 hour. Fixed oocytes were first stained with 2% osmium tetroxide (Electron Microscopy Sciences)–1.5% potassium ferrocyanide (Electron Microscopy Sciences) in 0.1 M phosphate buffer (pH 7.4) for 1 hour. Oocytes were then incubated with 0.1% tannic acid (Sigma-Aldrich) for 30 min and stained with 2% osmium tetroxide in water for 40 min. After overnight staining with 1% uranyl acetate (SPI-Chem) in water at 4°C, oocytes were further stained with 0.02 M lead nitrate (Merck Millipore)–0.03 M aspartic acid (Sigma-Aldrich) (pH 5.5) for 30 min. Oocytes were washed three times with water for 5 min between every staining step.

For immunoelectron microscopy of TACC3, mouse metaphase I oocytes were microinjected with 7 pl of 0.3 mg/ml US Immunogold-conjugated rabbit Fab anti-TACC3 (Aurion) with 0.1% NP-40. After fixation as described above, silver enhancement was performed with R-Gent SE-EM (Aurion). All following processing steps were performed in a microwave (Ted Pella) and oocytes were washed three times with water for 40 s at 250 W between every staining step. Fixed oocytes were first stained with 2% osmium tetroxide–1.5% potassium ferrocyanide in 0.1 M phosphate buffer (pH 7.4) for 8 min at 100 W (microwave cycling between on and off every 2 min). Oocytes were then incubated with 1% thiocarbohydrazide (Sigma-Aldrich) for 12 min at 100 W (microwave cycling between on and off every 2 min) and stained with 2% osmium tetroxide in water for 12 min at 100 W (microwave cycling between on and off every 2 min). After overnight staining with 1% uranyl acetate in water at 4°C, oocytes were further stained with 0.02 M lead nitrate–0.03 M aspartic acid (pH 5.5) for 12 min at 100 W (microwave cycling between on and off every 2 min).

Oocytes were subsequently dehydrated in a graded ethanol series (10, 30, 50, 75, 90, 100, and 100%) for 40 s at 250 W and infiltrated in a graded series (25, 50, 75, 90, 100, and 100%) of Durcupan resin (Sigma-Aldrich) in ethanol for 3 min at 250 W. Infiltrated oocytes were embedded with minimal amount of resin on top of an Aclar film as previously described (95). Embedded oocytes were cut out and attached to a SEM stub (Science Services) using silver-filled EPO-TEK EE129-4 adhesive (Electron Microscopy Sciences), and were cured overnight at 60°C. Samples were coated with an 8-nm platinum layer using the high-vacuum sputter coater EM ACE600 (Leica) at 35-mA current. Afterwards, samples were placed in the Crossbeam 540 FIB-SEM (Zeiss). To ensure even milling and to protect the surface, a 400-nm platinum layer was deposited on top of the region of interest at 3-nA current. Atlas 3D (Fibics) was used to collect the 3D datasets. Samples were exposed with a 15-nA current, and a 7-nA current was used to polish the cross-section surface. Images were acquired at 1.5 kV with the ESB detector at a grid voltage of 1100 V (5-nm pixel size in xy) in a continuous mill and acquire mode using 700 pA for the milling aperture (5-nm z-step).

After acquisition, images were first aligned using Linear Stack Alignment with SIFT in Fiji (NIH) (96). Datasets were then cropped, inverted and smoothed in Fiji. For better visualization of microtubules, datasets were further subjected to two rounds of Local contrast enhancement (CLAHE) in Fiji. Specific parameters used for the first round were: 127 for blocksize, 100 for histogram bins, and 1.5 for maximum slope. Specific parameters used for the second round were 50 for blocksize, 256 for histogram bins, and 2.0 for maximum slope. To obtain a metaphase I spindle parallel to the imaging plane, the resulting image stacks were rotated and resliced with Interactive Stack Rotation in Fiji. 3D segmentation and surface rendering were performed with Microscopy Image Browser (97) and Imaris, respectively.

Fluorescence recovery after photobleaching (FRAP)

For analyses of protein dynamics, oocytes coexpressing fluorescent reporter(s) of interest with H2B-miRFP were rotated with an unbroken microinjection needle on stage to obtain meiotic spindles parallel to the imaging plane. Rectangular or circular regions of interest (ROIs) were marked and photobleached using the corresponding excitation laser line (405, 488, 561, and/or 633 nm) at the maximum power after the third or fifth time point.

Photoactivation

For analyses of dissipation, oocytes coexpressing mPA-α-tubulin, PCM1-mPA-GFP, TACC3-mPA-GFP, or mPA-GFP-TPX2 with H2B-miRFP were rotated with an unbroken microinjection needle on stage to obtain meiotic spindles parallel to the imaging plane. Rectangular ROIs were marked and photoactivated using the 405-nm laser line at the maximum power after the third or fifth time point.

In vitro droplet assembly and dextran exclusion assay

TACC3 droplets were assembled in an imaging chamber as previously described for SPD-5 (32). For most experiments, 20 μM purified His-tagged proteins or recombinant GST (GenScript) in 50 mM tris (pH 7.4), 500 mM NaCl, 0.5 mM DTT, 1% glycerol, and 0.1% CHAPS (all from Sigma-Aldrich) was added to 25 mM HEPES (pH 6.4), 150 mM KCl, 0.5 mM DTT, and 12% PEG-3350 with 0.1 mg/ml NTA-Atto 647 N (all from Sigma-Aldrich). For dextran exclusion assay, 2.5 mg/ml Oregon Green 488-conjugated dextran of 70,000 M.W. (Molecular Probes) was included during the assembly of the reaction.

Thioflavin T staining

Oocytes were fixed, extracted, and blocked as for routine immunofluorescence. 10 μM thioflavin T was applied to oocytes for 10 min as previously described (98).

Cold-stable assay

Mouse metaphase I oocytes were incubated on ice for 15 min and immediately fixed as for routine immunofluorescence.

Calcium-stable assay

Mouse metaphase I oocytes were incubated in 100 mM PIPES (pH 7.0, titrated with KOH), 1 mM MgCl2, 0.1 mM CaCl2, and 0.1% triton X-100 at 37°C for 15 min and immediately fixed as for routine immunofluorescence.

General quantification

Time-lapse movies of live mouse oocytes were analyzed using Imaris (Bitplane). To determine the stages of meiosis and to score for chromosome segregation defects, the timing of meiotic progression was quantified relative to the time of NEBD. NEBD was defined as the time point when the sharp boundary between the nucleus and cytoplasm disappeared. Chromosome(s) that failed to congress to the metaphase plate before anaphase onset or the end of the video were scored as misaligned chromosomes. Anaphase onset was defined as the last time point of metaphase (5 to 6 min) before chromosome separation was first observed. Chromosome(s) that failed to clear the central spindle region within 10 to 12 and 20 to 24 min of anaphase onset were scored as mildly and severely lagging chromosomes, respectively. The presence of misaligned or lagging chromosome(s) was further confirmed by 3D reconstruction of chromosomes in Imaris as previously described (77).

Quantification of FRAP experiments

Mean intensities of photobleached areas over time were exported from ZEN into Excel (Microsoft) for further processing. The intensity of an unbleached area in the cytoplasm of the same oocyte was subtracted to correct for cytoplasmic background. Background-corrected data were then normalized to the mean intensity of prebleach time points (F0). As the postbleach intensity could not be normalized between individual plots due to variable photobleaching efficiency, plots of intensity (F) against time were fitted to single exponential functions [F(t) = cFet where c is the offset, F is the amplitude of maximum intensity recovered after equilibrium, and τ is the time constant) in OriginPro (OriginLab). Half-times of maximum recovery (t1/2) and mobile fractions were determined by τ × ln(2) and F/(F0F′) (where F′ is the minimum intensity measured immediately after photobleaching), respectively.

Manual and automatic quantification of plus-end growth velocity

To determine the growth rate of plus-ends of astral microtubules, comet velocities were determined using the Kymograph function in Fiji. Tracks were analyzed in Excel.

For plus-ends of microtubules within the spindle, comets were automatically tracked using u-track 2.2 (99) in MATLAB R2017b (MathWorks). For comet detection, low-pass and high-pass Gaussian standard deviations were set to two and six pixels, respectively. For comet tracking, maximum gap to close was set to 0 frame to eliminate discontinuous tracks. As astral microtubules were protruding from the spindle in all directions and their comets should be out of focus before comets of spindle microtubules, tracks longer than five frames were used for further processing. This resulted in, on average, >500 (for control group) and >300 (for experimental group) tracks per oocyte within 2.5 min.

Quantification of photoactivation experiments

To determine dissipation rates of α-tubulin, mean intensities of photoactivated areas over time were exported from ZEN into Excel for further processing. Data were first corrected for cytoplasmic background by subtracting the intensity of the photoactivated areas before photoactivation. Background-corrected data were then normalized to the intensity of the first postactivation time point (F0). Plots of intensity (F) against time were fitted to one-component exponential functions [y = Ae(c x)/τ, where c is the offset, A is the fraction of the component, and τ is the time constant] in OriginPro. Half-times of decay (t1/2) were calculated by τ × ln(2). Single-component exponential fitting was used because double-component exponential fitting was unable to yield two distinct τ, consistent with our previous studies on metaphase II spindles (77). This was likely due to the fact that the fraction of dynamic non-kinetochore-bound microtubules is much more abundant than that of stable kinetochore microtubules in mouse meiotic spindles.

Manual quantification of fluorescence intensities and volume by 3D reconstruction of spindles and aMTOCs

Spindles and aMTOCs were labeled (with MAP4-MTBD and CEP192, respectively) in live and (with α-tubulin and pericentrin, respectively) fixed oocytes. They were segmented by applying a threshold in the corresponding channel using the Surface function of Imaris. Cytoplasmic aMTOCs that were not part of the spindle were manually removed. Depending on the signal-to-noise ratio in each oocyte, a suitable threshold value was selected and maintained for the entire time series. The volumes of spindles and aMTOCs were exported into Excel. The mean and total intensities for spindle proteins (α-tubulin, MAP4-MTBD, PRC1, or TACC3) and aMTOCs proteins (CEP192 or Pericentrin) were exported into Excel.

For live imaging, data from individual oocytes were aligned to the time of NEBD or the onset of microtubule nucleation. To normalize individual dataset, all (mean or sum) intensities for a channel from both the control and experimental groups were divided by the average (mean or total) intensity for that channel from either the control or experimental group at steady state. For some experiments, normalized data from different datasets were interpolated to identical time intervals in OriginPro.

For immunofluorescence, all (mean or sum) intensities for a channel from both the control and experimental groups were normalized to the average (mean or total) intensity for that channel from the control group.

Automatic quantification of fluorescence intensities and volume by 3D reconstruction of spindles

To consistently reconstruct the spindles from different cold-stable assays, an in-house–developed MATLAB script (https://github.com/Emoenni/batch_surfaces) was used for automatic surface creation in Imaris.

Individual sections of a dataset were background-subtracted by a Gaussian smoothed image with a sigma equal to 80% of the background scale given and smoothed with a median filter. The size of the filter was determined by the nearest odd integer to twice the object detail given.

To automatically determine a threshold for the Surface function in Imaris, Otsu’s method was implemented via the “multithresh.m” function in MATLAB. This function allows the computation of multiple thresholds for multilevel images. A single threshold was first computed and the volume of the surface was approximated by multiplying the number of voxels above the threshold with the volume of a voxel. When the volume was not within the range of possible volumes, more thresholds were computed. This was repeated until an appropriate threshold was found or a number of five thresholds was reached. The unaltered values of object detail, background scale, and a filter based on the intensity mean of the second channel were then used for surface creation in Imaris. For spindles, the common center of homogeneous mass of all surfaces was computed, and all surfaces with a center of mass >10 μm from the common one were removed.

Specific parameters used for spindles were: 0.15 μm for object detail, 0.6 μm for background scale, and 50 to 1000 μm3 for volume range.

Statistical analysis

Average (mean) and standard deviation (SD) were calculated in Excel. Statistical significance based on unpaired, two-tailed Student’s t test (for absolute values) and two-tailed Fisher’s exact test (for categorical values) were calculated in OriginPro or Prism (GraphPad). All box plots show median (horizontal black line), mean (small black squares), 25th and 75th percentiles (boxes), 5th and 95th percentiles (whiskers), and 1st and 99th percentiles (crosses). All data are from at least two independent experiments. P values are designated as *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001. Nonsignificant values are indicated as NS.

Supplementary Materials

science.sciencemag.org/content/364/6447/eaat9557/suppl/DC1

Figs. S1 to S26

Table S1

References (102118)

Movies S1 to S10

References and Notes

Acknowledgments: We are grateful to the staff from the Animal Facility and Live-Cell Imaging Facility of the Max Planck Institute for Biophysical Chemistry for technical assistance. We thank E. Bellou, T. Cavazza, M. Daniel, K. Harasimov, and A. Zielinska for help with bovine, ovine, and porcine ovaries; D. G. Booth and T. Ruhwedel for advice on sample preparation for electron microscopy; C. Nardis for help with data visualization; T. Cavazza, S. Cheng, T. M. Franzmann, A. A. Hyman, D. Kamin, P. Lénárt, S. Truckenbrodt, and X. Zhang for helpful discussions; T. Cavazza, P. Lénárt, L. Wartosch, and Life Science Editors for critical comments on the manuscript; M. van Breugel, I. M. Cheeseman, J. Chen, J. Ellenberg, A. M. Fry, T. W. Gadella, J. G. Gleeson, P. Gönczy, R. Z. Qi, M. Lin, J. Lippincott-Schwartz, F. Liska, J. Lüders, S. Munro, K. Mykytyn, E. A. Nigg, J. Pines, K. Rhee, S. J. Royle, D. J. Sharp, T. Stearns, D. Stephens, C. Sütterlin, L. H. Tsai, R. Y. Tsien, P. Wadsworth, and V. V. Verkhusha for cDNAs and constructs; and I. M. Cheeseman, D. A. Compton, N. J. Galjart, R. Gassmann, G. Goshima, G. J. Gorbsky, A. A. Hyman, T. U. Mayer, A. D. McAinsh, A. Merdes, L. Pelletier, M. Takeichi, and L. Wordeman for antibodies. Funding: The research leading to these results has received financial support from the European Research Council under grant agreement no. 337415 and from the Lister Institute of Preventive Medicine. C.S. is a recipient of the Croucher Scholarship for Doctoral Study. A.M.S. is funded by the Cluster of Excellence and Deutsche Forschungsgemeinschaft (DFG) Research Center Nanoscale Microscopy and Molecular Physiology of the Brain (CNMPB). Author contributions: M.S. conceived the study; C.S., K.B.S., and M.S. designed experiments and methods for data analysis; C.S. and K.B.S. performed all experiments and analyzed the data with the following exceptions: A.M.S. prepared electron microscopy samples with C.S. and performed FIB-SEM; E.M. wrote all in-house–developed scripts and plugins and analyzed comet velocities with K.B.S.; D.C. initiated the systematic analysis of protein localization and characterization of the LISD; A.P. purified recombinant proteins and optimized in vitro droplet assembly with C.S.; C.S., K.B.S., and M.S. wrote the manuscript and prepared the figures with input from all authors; W.M. supervised the electron microscopy experiments; and M.S. supervised the entire study. Competing interests: The authors declare no competing interests. Data and materials availability: Plasmids are available from M.S. under a material transfer agreement with the Max Planck Society. All other data needed to evaluate the conclusions in the paper are present in the paper or the supplementary materials.
View Abstract

Stay Connected to Science

Subjects

Navigate This Article