Actin filaments accumulated in the nucleus remain in the vicinity of condensing chromosomes in the zebrafish early embryo

Actin is a highly conserved protein throughout eukaryotes. In the cytoplasm, actin serves a role in cell migration, stress fiber formation, and cell division by forming a contractile ring during cytokinesis. Actin is also present in nuclei, where it plays a crucial role in transcriptional regulation (Vartiainen et al., 2007; Baarlink et al., 2013; Tian et al., 2016; Sokolova et al., 2018) and in DNA damage repair processes (Chiolo et al., 2011; Belin et al., 2015; Caridi et al., 2019). Actin nuclear localization is mediated through the interplay of nuclear import regulated by importin 9 with cofilin (Dopie et al., 2012) and nuclear export regulated by exportin 6 (Exp6), which transports profilin-actin complexes (Stüven et al., 2003). Whereas filamentous actin (F-actin) is rarely observed in somatic cell nuclei, it is presumed to be involved in regulating the function and structure of undifferentiated cell nuclei (Misu et al., 2017).

In mouse early embryonic nuclei, F-actin facilitates nuclear expansion during the G1 phase (Baarlink et al., 2017) and maintains the integrity of the totipotent state of the cell mass (Okuno et al., 2020). Nuclear actin polymerization via Wave-1 is required for reprogramming somatic nuclei that are transplanted into African clawed frog (Xenopus laevis) oocytes (Miyamoto et al., 2011). Actin has a distinct role in the large nucleus of oocytes, which is also called the germinal vesicle (GV). Xenopus laevis GV contains a high concentration of actin to maintain the large nuclear structure (Clark and Merriam, 1977; Clark and Rosenbaum, 1979; Bohnsack et al., 2006). In mammalian oocytes, the F-actin bundle in GVs stabilizes the chromatin mobility and secures accurate chromosome alignment and segregation (Mogessie and Schuh, 2017; Scheffler et al., 2022; Dunkley et al., 2022). In starfish oocytes, actin mesh forms in the nuclear area after nuclear envelope breakdown (NEBD; Lénárt et al., 2005), and the contraction of the actin meshwork transports chromosomes until they are captured by microtubules (Mori et al., 2011). In addition, F-actin polymerizes underneath the nuclear envelope and accelerates fragmentation of the nuclear envelope during the first stage of NEBD (Mori et al., 2014; Wesolowska et al., 2020).

During Xenopus embryonic development, F-actin accumulates in nuclei up to the blastula stage, when the major wave of zygotic transcription begins, before diminishing at the gastrula stage (Oda et al., 2017). A reconstituted system using Xenopus egg extracts has been used for analyzing the assembly of early embryonic nuclei from de-membraned sperm in vitro (Murray, 1991). In reconstituted nuclei using this system, nuclear actin prevents chromatin aggregation during interphase and facilitates chromosome alignment during prometaphase (Oda et al., 2017). However, the dynamics of nuclear actin during Xenopus embryo development have not been elucidated, because live-cell imaging is hampered by the melanin granules that are present in the cortex of embryos.

In contrast to the African clawed frog (Xenopus laevis), zebrafish (Danio rerio) embryos that share a similar developmental process with the frog are transparent and thus suitable for live-cell imaging. During the early embryo stages of zebrafish, cells have relatively large nuclei and undergo a series of short and synchronized cell cycles without transcription and gap phases (Satoh, 1977; Kimmel et al., 1995; Joseph et al., 2017), which also occurs in Xenopus embryos (Jevtić and Levy, 2015; Newport and Kirschner, 1982; Lucas et al., 2000; Mahbubani et al., 1992).

In zebrafish embryos, transcription from the zygotic gene is observed as early as at the 64-cell stage, and becomes further activated at around the 512-cell to 1000-cell stages (Mathavan et al., 2005; Heyn et al., 2014; White et al., 2017; Pálfy et al., 2017; Vastenhouw et al., 2019). After the 1000-cell stage, the cell cycle elongates and loses synchronicity, which occurs with the onset of differentiation (Kane et al., 1992).

Fluorescently labeled phalloidin has been the standard reagent to stain F-actin in fixed cells (Dancker et al., 1975; Wehland et al., 1977). Live-cell actin-specific probes have also been developed, including fluorescent protein-tagged LifeAct and UtrCH, as well as actin-specific chromobody (Actin-chromobody^®, Rocchetti et al., 2014). UtrCH, which is the Calponin Homology domain of human utrophin (Burkel et al., 2007), is an actin-binding peptide that is more selective to F-actin than LifeAct, which is derived from the yeast actin-binding protein, Abp140 (Riedl et al., 2008). In this report, we mainly used superfolder GFP-tagged UtrCH (UtrCH-sfGFP) to track nuclear F-actin dynamics in zebrafish early embryos. We observed that F-actin accumulates in nuclei during interphase and the actin patches remain in the vicinity of condensing chromosomes during the prophase to metaphase.

To investigate whether F-actin accumulates in nuclei in zebrafish early embryos, we first aimed to detect F-actin by using fixed embryos stained with fluorescently labeled phalloidin. DNA was counterstained with Hoechst dye. Phalloidin signals were enriched in nuclei in the 32-, 128-, and 1000-cell stages, particularly in Hoechst-poor regions (Fig. 1A-C). In sphere-stage embryos, the total nuclear phallodin signals were decreased, whereas filamentous structures were observed (Fig. 1B and C). This observation suggests that F-actin is abundantly present in nucleoplasm during the early stages of zebrafish development, as was observed in Xenopus laevis embryos (Oda et al., 2017). Although the filamentous structures observed at the later stages are of interest, we here focused on the dynamics and function of nuclear F-actin that are abundantly accumulated at the earlier stages.

To visualize nuclear F-actin dynamics in living zebrafish embryos, purified UtrCH-sfGFP (Fig. S1A) was injected into the yolk of one-cell stage embryos. The antigen-binding fragment (Fab) specific for acetylation of histone H3 lysine 9 (H3K9ac), which distributes throughout the chromatin from the early stage of development, labeled with Cy5, was co-injected as a chromatin marker (Sato et al., 2019). The injected embryos were set onto a confocal microscope at the four-cell stage, and three-dimensional time-lapse images were acquired every 90 s from the eight-cell stage to the oblong stage (Fig. 2A). UtrCH-sfGFP signal is enriched in the interphase nuclei of embryos from the eight-cell stage to the 1000-cell stage (Fig. 2B, arrowheads and insets; Fig. 3A; Movie 1). At the oblong stage (3.5 h post-fertilization; hpf), UtrCH signals in nuclei became weaker compared to those in earlier stages (Fig. 2B). These live-cell imaging data are consistent with the observations in fixed embryos.

Time-lapse imaging with 1.5-min intervals (Fig. 3A and Fig. S2B) and measurements of the ratio of nucleus to the cytoplasm intensity, or chromosome area to the cytoplasm during M phase, (N/C ratio) of UtrCH-sfGFP (Fig. 3B) revealed the kinetics of UtrCH-sfGFP during the cell cycle. UtrCH-sfGFP signal began to accumulate in the nucleus at the end of telophase, became more concentrated during interphase, peaking just before mitosis, and then disappeared from mitotic chromosome areas during the 16-cell to 1000-cell stages. After the 1000-cell stage, the mean value of UtrCH-sfGFP N/C ratio reduced to less than 1.0 (Fig. 3B). We also analyzed whether the nuclear F-actin level and dynamics depend on the position in an embryo. UtrCH-sfGFP N/C ratio in cells at the surface enveloping and deep layers of embryos were measured at the 512-cell and sphere stages. Peaks at the onset of NEBD were observed at the 512-cell stage both in the envelope and deep layer cells (Fig. 3C). The N/C intensity ratio was slightly, but insignificantly, higher in the deep layer cells (p=0.25), suggesting that nuclear F-actin formation at this stage is similar in cells in different layers. No nuclear F-actin accumulation was observed in either layer at the sphere stage.

Because UtrCH could stabilize actin filaments and influence actin dynamics (Spracklen et al., 2014), we examined if the injection of UtrCH induces more F-actin formation or intense phalloidin-staining signals. However, the phalloidin signals in UtrCH-sfGFP injected embryos were similar to those in control embryos (Fig. S1C), suggesting that injected UtrCH-sfGFP does not affect F-actin formation under the conditions used in this study. We also used another actin-binding peptide, LifeAct, which binds to globular actin (G-actin) and F-actin (Riedl et al., 2008), as the sfGFP-tagged version (LifeAct-sfGFP) to validate the results obtained with UtrCH. Purified LifeAct-sfGFP (Fig. S1B) was co-injected with Cy3-labeled anti-phosphorylated histone H3 serine 28 (H3S28ph) Fab to highlight chromosomes from the prometaphase to anaphase (Hayashi-Takanaka et al., 2014) and Cy5-labeled anti-H3K9ac Fab to label nuclei (Fig. S1D). LifeAct-sfGFP was concentrated in nuclei, showing >1.0 N/C ratio, during early embryonic stages (Fig. 4A and Fig. S2B). After the 1000-cell stage, the LifeAct-sfGFP N/C ratio reduced to <1.0 (Fig. S2B), as was observed in UtrCH-sfGFP (Fig. 3B). This lowered N/C ratio on average could be attributed to the less nuclear accumulation and/or the asynchronized cell cycle, as cells in the mitotic phase showed a low N/C ratio (Fig. 3 and Fig. S2B).

We then analyzed N/C ratios in individual interphase nuclei by eliminating H3S28ph-positive mitotic cells (Fig. 4B). The cumulative bar chart showed that >90% of interphase nuclei exhibited a LifeAct-sfGFP N/C ratio >1.0 up to the 256-cell stage. At the 512- and 1000-cell stages, the fraction of nuclei with an N/C ratio of >1.0 decreased to ∼80% and ∼65%, respectively (Fig. 4B). After the 1000-cell stage, the fraction of nuclei with an N/C ratio of <1 increased substantially (Fig. 4B, 150 min), reaching ∼90% at the sphere (180 min) and dome (195 min) stages. Taken together with the phalloidin-staining and UtrCH-sfGFP data, these results suggest that actin is gradually concentrated and polymerized in nuclei during the interphase in the early stages of zebrafish embryos, as observed in Xenopus (Oda et al., 2017), and the amount of nuclear actin decreases after the 1000-cell stage.

Since nuclear UtrCH-sfGFP reached a peak just before mitosis (Fig. 3A, 7.5, 9.0, and 10.5 min; Fig. S2A, 7.5, 9.0, and 10.5 min, 4A, 12.0 min), we investigated F-actin dynamics during the late interphase to mitosis in detail. For this purpose, tetramethylrhodamine-labeled 155-kDa dextran (TMR-dextran) was used for monitoring the timing of NEBD; TMR-dextran that was injected into the cytoplasm could not enter the nucleus owing to its large size until the nuclear membrane broke down at the onset of prometaphase. Thus, NEBD was identified by the onset of TMR-dextran leakage into the nucleus. In addition, we used JF646-labeled latency-associated nuclear antigen peptide (JF646-LANA) to track chromatin throughout the cell cycle. The LANA peptide binds to the acidic patch in the nucleosome (Barbera et al., 2006), and a synthetic LANA peptide with polyethylene glycol and fluorescein was shown to label chromatin in living cells (Fujiwara et al., 2021). In the current study, JF646-LANA was synthesized (see Materials and Methods) to be used together with sfGFP- and TMR/Cy3-labeled probes. Mitotic chromosomes visualized with Cy3-H3S28ph Fab were indeed more intensely detected by JF646-LANA than by H3K9ac Fab, whose chromatin binding is decreased by deacetylation during mitosis (Fig. S1D).

To visualize F-actin dynamics at around NEBD, one-cell stage embryos were injected with UtrCH-sfGFP, TMR-dextran, and JF646-LANA, and were imaged every 12.5 s from 256-cell stage. Fig. 5 shows the representative images of cells in the 1000-cell stage (A) and high stage (B) embryos with the changes in N/C ratio of UtrCH-sfGFP and TMR-dextran (C). UtrCH-sfGFP accumulated in nuclei (Fig. 5, −37.5 s to −12.5 s), reaching a peak concomitantly with NEBD as indicated by TMR-dextran diffusing into nuclear areas (Fig. 5, 0 s). After NEBD, UtrCH signals were remained in the vicinity of condensing chromosomes during prometaphase (Fig. 5, 0 s to 25 s), and almost disappeared from the chromosome areas by metaphase (Fig. 5, 37.5 s).

To track microtubule dynamics with F-actin and chromosomes during prophase to metaphase, TMR-labeled tubulin was co-injected with UtrCH-sfGFP and JF646-LANA, and three-dimensional images were acquired every 15 s at the 256-cell (Movie 2), 512-cell (Movie 3), 1000-cell (Fig. 6; Movie 4 for the large area covering many cells and Movie 5 for a magnified view), high (Fig. S3 and Movie 6), sphere (Fig. S4), and dome stages (Fig. S5). The injection of these probes did not affect embryo viability at least up to the prime-6 stage (Fig. S6A). UtrCH-sfGFP was located in the nucleus with enrichment at around the nuclear periphery of the 1000-cell stage embryo before NEBD (Fig. 6, −15 s). Just after NEBD, UtrCH-sfGFP remained in the vicinity of the condensing chromosomes, and microtubule spindles (TMR-Tubulin) grown from centrosomes extended along with the chromosomes (Fig. 6A, 0 s and 15 s; see also Fig. 6B for 3D rendering views). UtrCH-sfGFP signals rapidly disappeared from the nuclear areas during prometaphase to metaphase (Fig. 6A, 30–45 s). Concomitantly with actin diminishment, microtubule spindles appeared to capture chromosomes for their segregation (Fig. 6, 45 s to 90 s). Similar dynamics of actin and tubulin were observed from the 256-cell stage to 1000-cell stages (256-cell stage: Movie 2, 512 cell-stage: Movie 3). In the high stage, the UtrCH-sfGFP signal in the nucleus was much weaker during the interphase (Fig. S3, −45 s to 0 s; Movie 6); however, after the NEBD, the clear actin patches were still observed in the vicinity of condensing chromosomes as observed in the 1000-cell stage (Fig. S3, 15 s and 30 s; Movie 6). In the sphere stage, the timing of UtrCH-sfGFP nuclear accumulation was more transient and the signals were much weaker than those during the 1000-cell stage (Fig. S4, 0 s and 15 s). Only a trace amount of UtrCH-sfGFP was observed in the chromosome area in the dome stage Fig. S5, 30 s and 45 s).

We next examined the relationship between zygotic transcription and nuclear F-actin, because the amount of F-actin was substantially decreased after the oblong stage, as in most somatic cells (Schoenenberger et al., 2005; Hofmann et al., 2009; Xu et al., 2010; Parisis et al., 2017). To test whether zygotic transcription is required to diminish nuclear F-actin, we injected α-amanitin, which directly binds to RNA polymerase II and interferes with its translocation and nucleotide incorporation (Nguyen et al., 1996; Brueckner and Cramer, 2008), into zebrafish embryos (Fig. 7A; Kane et al., 1996; Sato et al., 2019). Embryos that were injected with α-amanitin or the buffer were fixed at the 1000-cell and oblong stages and stained with Hoechst and phalloidin. In 1000-cell stage embryos, in which major ZGA typically begins, phalloidin signals were concentrated in the nuclei, regardless of α-amanitin injection (Fig. 7B, 1000-cell). In contrast, at the oblong stage, nuclear phalloidin signals were slightly enriched in α-amanitin-injected embryos compared to those detected in the control embryos (Fig. 7B, oblong). We then confirmed the effect of α-amanitin on nuclear F-actin in living cells by using UtrCH-sfGFP. In the both amanitin-treated and control embryos, the N/C ratio of UtrCH-sfGFP began to decrease after the high stage (Fig. 7C). However, compared to the control, the N/C ratio in α-amanitin-injected embryos remained relatively high at sphere and dome stages (Fig. 7C). These data suggest that the decrease of the N/C ratio of UtrCH-sfGFP is facilitated by zygotic transcription. The higher N/C ratio of UtrCH-sfGFP might also be associated with the effect of α-amanitin on the cell cycle since α-amanitin treatment slightly shortened the cell cycle length after the 1000-cell stage (from 11th to 13th cell division cycles) (Fig. S6B). The accumulation of nuclear F-actin may contribute to the early entry into mitosis by accelerating the fragmentation of the nuclear envelope (Mori et al., 2014; Wesolowska et al., 2020), although it is difficult to dissect the effect of higher nuclear F-actin from the inhibition of transcription by α-amanitin treatment.

To investigate the physiological role of nuclear F-actin in early embryos, we sought to artificially manipulate its levels. The first attempt to decrease nuclear actin by ectopic expression of Exp6, which is known to export actin from the nucleus to the cytoplasm (Stüven et al., 2003), by injecting mRNA encoding 3×FLAG-tagged zebrafish Exp6 into embryos. However, Exp6-3×FLAG expression induced severe cytokinesis defects but did not decrease nuclear actin levels (Fig. S7). Next, we expressed a polymerization-deficient actin mutant (R62D), in which Arg 62 was substituted to Asp (Posern et al., 2002), tagged with the superfolder Cherry and a nuclear localizing signal (Actin R62D-sfCherry-NLS), to inhibit nuclear F-actin formation. The wild-type actin tagged with sfCherry-NLS (Actin WT-sfCherry-NLS) and sfCherry-NLS were used as controls to examine the effect of the wild-type actin and sfCherry alone. Embryos were co-injected with mRNA encoding Actin R62D-sfCherry-NLS, Actin WT-sfCherry-NLS, or sfCherry-NLS, with UtrCH-sfGFP and Cy5-labeled H3K9ac-specific Fab and were imaged from the eight-cell stage. The N/C intensity ratios of UtrCH-sfGFP, however, did not show obvious differences among embryos expressing sfCherry-NLS, Actin R62D-sfCherry-NLS, and Actin WT-sfCherry-NLS (Fig. S8A and S8B). Expression of Actin R62D tagged with smaller 3xFLAG, instead of sfCherry, also did not affect the N/C ratios of UtrCH-sfGFP (Fig. S8C). Because actin was already abundantly present in nuclei, additional expression of the wild-type or mutant actin might have minimal effects.

In has been reported that cofilin proteins actively depolymerize actin filaments in cytoplasm during the interphase (Vartiainen et al., 2002; Amano et al., 2002; Kaji et al., 2008; Chin et al., 2016; Wioland et al., 2017). Cofilin1 is the non-muscle isoform expressed ubiquitously while cofilin2 is the muscle isoform. We expressed zebrafish cofilin1 and cofilin2, tagged with sfCherry and an NLS, to attempt depolymerizing nuclear actin in embryos, but again the N/C intensity ratios of UtrCH-sfGFP were not affected (Fig. S9). Thus, it was not possible to manipulate the level of nuclear F-actin in our experimental conditions. Nuclear F-actin level might be robustly regulated in zebrafish early embryos.

We demonstrated F-actin accumulation in zebrafish early embryo nuclei by fixed cells using phalloidin and by living cells using fluorescent actin-binding probes. In living embryos from the eight-cell to 1000-cell stages, UtrCH-sfGFP, which selectively binds to F-actin, gradually accumulated in the nucleus during the interphase, thereby reaching the maximum amount at around NEBD. After the high stage, UtrCH-sfGFP was not much concentrated in nuclei, while some filamentous structures were more evidently observed. These data are consistent with previous findings in mouse and Xenopus laevis early embryos (Baarlink et al., 2017; Oda et al., 2017), which indicates that nuclear F-actin accumulation in undifferentiated cells is common in vertebrates (Xu et al., 2010; Miyamoto et al., 2011; Miyamoto et al., 2013; Sokolova et al., 2018; Okuno et al., 2020). After ZGA, the level of nuclear F-actin drops. The decrease appears to be induced by zygotic transcription, as nuclear F-actin was still observed in the sphere-stage embryos that were treated with α-amanitin, an RNA polymerase II inhibitor. Maternal-to-zygotic transition may alter the balance of nuclear import and export factors of actin, as it was previously demonstrated that the actin exporter, Exp6, is not expressed in Xenopus oocytes nor in early embryonic stages, yet it becomes expressed after ZGA (Bohnsack et al., 2006).

In early-stage zebrafish embryos, the peak of nuclear F-actin accumulation was concurrent with the timing of NEBD and F-actin localized in the vicinity of condensing chromosomes during prometaphase before disappearing during metaphase. This actin behavior is similar to that observed in starfish oocytes (Lénárt et al., 2005; Mori et al., 2011; Mori et al., 2014; Wesolowska et al., 2020). In starfish oocyte GV, F-actin is assembled underneath the nuclear membrane and forms spikes toward to the nuclear membranes, which contribute to membrane rupture and NEBD (Mori et al., 2014; Wesolowska et al., 2020). After the germinal vesicle breakdown, actin meshwork catches chromosomes and shrinks like a fishnet to transport them over a long distance toward underneath the starfish oocyte cortex in meiosis II (Lénárt et al., 2005; Mori et al., 2011). In the mouse oocyte, both nuclear F-actin and spindle F-actin protect the chromosomes misalignment and missegregation at the meiosis (Mogessie and Schuh, 2017; Scheffler et al., 2022). It is interesting to speculate that the function of actin in large oocyte cells may partially be retained in fertilized eggs, although it is unknown whether a similar system to starfish is conserved in zebrafish oocytes. During the short cell cycle of zebrafish embryos, the spindle checkpoint is weak until the zygotic genome activation (Ikegami et al., 1997). In relatively large cells during the early embryonic stages, actin patches remained around chromosomes may assist in chromosome congression, just as the actin meshwork in starfish oocytes. It is also possible that actin patches guide microtubule assembly and promote proper and rapid spindle formation, as recent reports have shown that actin supports proper spindle formation in Xenopus gastrula embryos, mouse and human oocytes, and cultured cells (Woolner et al., 2008; Mogessie and Schuh, 2017; Roeles and Tsiavaliaris, 2019; Plessner et al., 2019; Farina et al., 2016; Farina et al., 2019; Kita et al., 2019). However, ‘spindle-actin’ has been depicted with barrel structures composed of fine actin fiber at metaphase, unlike actin patches during the prometaphase of zebrafish embryos. We speculate that the accumulation of F-actin in nuclei in zebrafish early embryos contribute to proper mitotic progression by assisting in NEBD, chromosome congression, and/or spindle assembly under rapid cell cycles with weak checkpoint mechanism.

We took different approaches to decrease the level of nuclear F-actin by expressing Exp6, an NLS-tagged Actin R62D, and NLS-tagged cofilin; unfortunately, however, none of them were successful. Exp 6 and NLS-tagged Actin R62D have been used to decrease nuclear actin and inhibition of nuclear actin polymerization, respectively, in previous reports (Bohnsack et al., 2006; Okuno et al., 2020; Miyamoto et al., 2011; Spencer et al., 2011; Dopie et al., 2012; Parisis et al., 2017; Baarlink et al., 2017). However, we did not observe significant decreases of UtrCH-sfGFP N/C ratios in zebrafish embryos expressing Exp6 and Actin R62D, under the conditions that did not affect cytoplasmic F-actin. Increasing the expression levels of those proteins by injecting more mRNA affected cytoplasmic F-actin causing defects in cytokinesis. We also attempted to decrease nuclear F-actin using NLS-tagged cofilin1 and cofilin2, which are actin-depolymerizing factors (Amano et al., 2002; Vartiainen et al., 2002; Kaji et al., 2008; Chin et al., 2016; Wioland et al., 2017); however, their expression did not affect nuclear F-actin levels. In addition, the expression of NLS-tagged wild-type actin did not increase UtrCH-sfGFP N/C ratio. It can be speculated that nuclear actin is very abundant and the ectopic wild-type actin or R62D mutant may not affect much on the total F-actin level. Also, F-actin formation in the nucleus may be robustly regulated, and expressing factors that assist in actin depolymerization may be compensated when the expression level is mild not affecting the cytoplasmic actin.

To obtain fertilized eggs, 5–10 pairs of AB line WT zebrafish (Danio rerio) males and females were crossed. Embryos were dechorionated using 1 mg/mL pronase from Streptomyces griseus (Sigma, 10165921001) diluted in 0.3× Danieau's buffer [17.4 mM NaCl, 210 μM KCl, 120 μM MgSO₄, 180 μM Ca(NO₃)₂, 1.5 mM HEPES, pH 7.6]. For actin visualization, UtrCH-sfGFP (1.46 fmol; see below) or LifeAct-sfGFP (2.3 fmol; see below) was injected with labeled Fabs [anti-H3K9ac (CMA315; 50 pg), anti-H3S28ph (CMA315; 50 pg); anti-DYKDDDDK tag (FUJIFILM Wako, 012-22384; 50 pg); Sato et al., 2019] and 10 fmol JF646-LANA (see below) in 0.5 nL phosphate-buffered saline (PBS, Takara, T900) containing 0.05% phenol red (Sigma, P5530) as an injection marker, were injected into the yolk of one-cell stage embryos. For NEBD or tubulin visualization, TMR-labeled 155-kDa dextran (250 pg in 0.5 nL PBS, Sigma, T1287) or TMR-labeled tubulin (1.25 ng in 0.5 nL General Tubulin Buffer, Cytoskeleton, TL590M) was injected with UtrCH-sfGFP and JF646 LANA. When mRNA was also injected, mRNA of sfCherry-NLS or 3×FLAG-tagged Actin WT or R62D (125 pg in 0.5 nL water), 3×FLAG-tagged Exp6 (150 pg in 0.5 nL water), sfCherry-NLS-tagged cofilin1 (125 or 400 pg in 0.5 nL water) or sfCherry-NLS-tagged cofilin2 (125 pg in 0.5 nL water) was first injected into the cell of one-cell-stage embryos, and then UtrCH-sfGFP, LifeAct-sfGFP, Fabs, and/or JF646-LANA were injected into the yolk. For transcription inhibition, 250 pg α-amanitin (Merck, 129741) in 0.5 nL PBS was injected into the yolk of one-cell-stage embryos (Kane et al., 1996; Sato et al., 2019). To monitor the transcription inhibition, 530 pg of Cy3-labeled antisense morpholino oligonucleotide specific to miR-430 transcripts (Hadzhiev et al., 2019) in 0.5 nL water was injected into the yolk of the one-cell stage embryos with JF646-LANA. The injected embryos were incubated at ∼25°C up to the four-cell stage and embedded into 0.5% agarose type VII-A low melting temperature (Sigma, A-0701) in PBS on a glass bottom dish (IWAKI, 3791-035) with the animal pole positioned down for live-cell imaging (Fig. 2A).

All zebrafish experiments were approved by the Tokyo Institute of Technology Genetic Experiment Safety Committee (I2018001), and animal handling was carried out according to the guidelines.

For Fig. 1A and B, dechorionated embryos were incubated until fixation at the desired stage with 4% paraformaldehyde in 250 mM HEPES for 3 h at 25°C. After fixation, embryos were washed three times with PBS and stained with 2 μg/ml Hoechst 33342 (Nacalai Tesque, 04929-82) and 20 nM Acti-stain™ 555 phalloidin (Cytoskeleton, PHDH1-A) in PBS containing 0.05% Triton X-100 at 4°C overnight in the dark. Stained embryos were washed three times with PBS before microscopic observation. For Fig. 7B and Fig. S1C, embryos were injected with α-amanitin (250 pg in 0.5 nL, Fig. 7B), UtrCH-sfGFP (1.46 fmol in 0.5 nL, Fig. S1C), or PBS (0.5 nL) as the control and incubated until the 1000-cell or oblong stage and fixed with 4% paraformaldehyde in 250 mM HEPES for 3 h at 25°C. After the fixation, embryos were washed with PBS and embedded in 100 μl 0.5% agarose in PBS and covered with 100 μl 1% agarose to prevent the detachment of embryos from a glass-bottomed dish. Embryos were stained with 2 μg/ml Hoechst 33342 and 20 nM Acti-stain 555 phalloidin in PBS containing 0.05% Triton X-100 in the dark with gentle shaking (Titec, Shake-XR; 40 rpm) overnight. Stained embryos were washed three times with PBS before microscopic observation.

Images were collected using a confocal microscope. For Figs 1, 2, 3, 4B, 7C; Figs S2, S7, S8, S9, and Movie 1, images were acquired using a confocal microscope FV1000 (Olympus) operated under built-in software (FLUOVIEW-ASW, version 4.2) with a UPLSAPO 30× silicone-immersion objective lens (NA 1.05), with 150 μm pinhole, zoom 1.0×, and 488-nm, 543-nm, and 633-nm laser lines on a heated stage at 28°C (Tokai Hit). For live imaging of Figs 2B, 3, 4B, 7C; Figs S2, S6C, S7, S8, S9, and Movie 1, 25 z-stack images (4 μm z-intervals; 512×512 pixels) were taken with 90 s intervals.

For high-speed or higher resolution imaging to produce the images shown in Figs 5 and 6, Figs S1C, S1D, S3, S4, S5 and Movies 2–6, images were taken using a spinning disk confocal system consisting of an inverted microscope (Ti-E, Nikon) with an Apo 40× water-immersion objective lens (NA 1.25; Figs 4, 5, 6, 7A; Figs S1C, S3, S4, S5, and Movies 2–6) or a Plan Apo 25× silicone-immersion objective lens (NA 1.05, Fig. S1D), a spinning disk unit with 40 μm pinhole (Dragonfly, Oxford Instruments), a laser unit with 488-nm, 561-nm and, 637-nm laser lines, and an EM-CCD camera (iXon Life 888, Oxford Instruments), operated under Fusion software (version 2.0 and 2.2, Oxford Instruments). 15–25 z-stack images (1, 2, or 4 μm z-intervals; 1024×1024 pixels) were taken with 12.5, 15, 60, or 90 s time intervals at 25°C.

DNA fragment encoding UtrCH (1-262 of human utrophin, NM_007124) or LifeAct (Riedl et al., 2008) gifted from Dr Mari Iwabuchi (Nagoya University, Japan) was fused with the sfGFP gene with a linker sequence (DPPVAT) and was cloned into a pGEX6P-1 vector that has the N-terminal glutathione S-transferase (GST) tag (Cytiva). E. coli BL21 (DE3) cells harboring the pGEX6P-1-based expression plasmid were grown in 24 ml Plusgrow II (Nacalai Tesque, 08246-86) medium overnight at 25°C. The overnight culture (24 ml) was diluted into 400 ml PlusgrowII medium and further incubated at 25°C for 3 h. The expression of GST-UtrCH-sfGFP and GST-LifeAct-sfGFP was induced in E. coli BL21 (DE3) at 25°C for 7 h by adding 1 mM isopropyl β-D-thiogalactopyranoside (Nacalai Tesque, 19742-94). Harvested bacterial cells (400 ml, 4000×g, 10 min, 4°C) were suspended in 20 ml ice-cold 50 mM Tris–HCl (pH 7.4) buffer containing 1 mg/ml lysozyme (Seikagaku Corporation, 100940) and 1× protease inhibitor cocktail (Nacalai Tesque, 03969-21) and were lysed by sonication (Branson; SONIFIRE 250; output control: 1, duty cycle: 50% for 2 min repeated 14 cycles with 30 s interval) in an ice bath. Triton X-100 (10% in water) was added to the sonicated solution to yield the final concentration at 1% and the mixture was incubated for 30 min at room temperature with a gentle rotation (Titec, RT-50; 30 rpm). After centrifugation (10,000×g for 20 min at 4°C), the supernatant was mixed with 5 ml (bed volume) glutathione sepharose 4B beads (Cytiva, 17075605; prewashed with PBS) and gently rotated (Titec, RT-50; 30 rpm) at 4°C for 20 h in the dark. The beads were collected in a Poly-Prep Chromatography Columns (BioRad) and washed with a 10-column volume of PBS. GST-UtrCH-sfGFP or GST-LifeAct-sfGFP was eluted using a 5-column volume of 10 mM glutathione (Fujifilm Wako Pure Chemical, 073-02013) containing 50 mM Tris–HCl (pH 8.7). The eluate was concentrated down to 1.5 ml using an Amicon Ultra Centrifuge Filter unit (30 kDa cutoff for GST-UtrCH-sfGFP and 10 kDa cutoff for GST-LifeAct-sfGFP; Merck, UFC903024 and UFC901024, respectively). The GST tag was removed by digestion with 8 units/ml PreScission Protease (Cytiva, 27084301) on ice for 24–36 h and then removed through glutathione sepharose 4B beads (5 ml bed volume) packed twice in Poly-Prep Chromatography Columns (BioRad). UtrCH-sfGFP and LifeAct-sfGFP were further purified through gel filtration column chromatography (HiLoad 16/600 Superdex 75 pg, Cytiva, 28989333) using AKTAPrime PLUS (Cytiva, 11001313). After loading 1.5 ml of the protein sample, 30 ml fractions were collected and analyzed by 15% sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis (SuperSep Ace, Fujifilm Wako Pure Chemical, 190-15001). The highly pure fractions were collected (Fig. S1A and S1B), aliquoted, snap-frozen in liquid N₂, and stored at −80°C. Once thawed, recombinant proteins were stable at 4°C for at least 5 days.

Dye-conjugated Fabs were prepared as previously described (Kimura and Yamagata, 2015). In brief, 100 μg purified Fab (anti-H3K9ac: CMA310/19E5, Hayashi-Takanaka et al., 2011; anti-H3S28ph: CMA315/10-20F11, Hayashi-Takanaka et al., 2014; anti-DYKDDDDK tag: Fujifilm Wako, 012-22384) was reacted with N-hydroxysuccinimide ester-conjugated fluorescence dye (Cy5, cytiva, PF11A25001; Cy3, cytiva, PA13105) in 100 mM NaHCO₃ (pH 8.3) in PBS for 1 h at room temperature in the dark. Dye-conjugated Fabs were collected using PD MiniTrap G-25 column (Cytiva, 28918004; pre-equilibrated with PBS) and were concentrated to 1 mg/ml using a 10-kDa cutoff Amicon Ultracell Centrifuge Filter Unit (Merck, UFC5010BK), and stored at 4°C in the dark.

Antisense morpholino oligo nucleotide was labeled as previously described (Sato et al., 2019). In brief, 3′-primary amino-modified morpholino oligonucleotides specific for miR-430 was reacted with N-hydroxysuccinimide ester-conjugated Cy3 in 0.1 M NaHCO₃ (pH 8.3) for 1 h at room temperature in the dark. After the reaction, Cy3-labeled morpholino oligonucleotide was purified using a PD MiniTrap G-25 column.

Preparative high-performance liquid chromatography (HPLC) was conducted by using a JASCO HPLC system equipped with a UV-2075 spectrometer (230 nm), PU-4086 pumps, a DG-4580 degasser, and an MX-2080-32 mixer using a YMC-Triart C18 (20 mm I.D. ×250 mm or 10 mm I.D. ×250 mm) column at 40°C with a gradient of acetonitrile in 0.1% aqueous trifluoroacetic acid (TFA) at a flow rate of 10 or 3.5 ml/min. Analytical HPLC was conducted by using a JASCO HPLC system equipped with a UV-2075 spectrometer, PU-4180 pumps, a DG-4580 degasser, and an MX-2080-32 mixer, using a YMC-Triart Phenyl C18 (4.6 mm I.D. ×150 mm) column with a linear gradient of 2–90% acetonitrile in 0.1% aqueous TFA over 3–15 min.

Poly(ethylene glycol) methyl ethers (Mn=750) were purchased from Sigma-Aldrich (#202495). All protected α-amino acids were purchased from Watanabe Chemical Industries (Hiroshima, Japan) and Peptide Institute (Osaka, Japan). NovaPEG Rink amide resin was purchased from Merck KGaA. Other chemicals were used as received, unless otherwise stated.

The LANA peptide was synthesized on a solid phase using NovaPEG-Rink-amide-resin (Fig. S10). The Fmoc-amino acid was sequentially coupled using the DIC-HOBt method in N,N-dimethylformamide (DMF; three equivalent of each) for 60 min at room temperature after removal of each Fmoc group with 20% piperidine-DMF for 10 min. The N-terminus was capped with 20% acetic anhydride in CH₂Cl₂ for 5 min. The peptide was cleaved from the resin by treatment with TFA in the presence of triisopropylsilane (TIPS) and H₂O (95:2.5:2.5) for 90 min at room temperature, and then was concentrated under reduced pressure, and precipitated with ether to obtain crude peptides, which were purified with preparative HPLC (0% acetonitrile for 5 min, followed by a linear gradient of 0–60% acetonitrile over 30 min with a flow rate of 10 ml/min) to afford the product LANA_5-22-N₃ peptide (Fig. S10B) as white solids after lyophilization. 3,4,5-Tris(PEG₇₅₀)-N-(prop-2-yn-1-yl)benzamide was synthesized (Fig. S10C) following a previously described method (Fujiwara et al., 2021).

To synthesize PEG₇₅₀-LANA (Fig. S10D), Cu-Tris[(1-benzyl-1H-1,2,3-triazol-4-yl)methyl]amine (TBTA) solution in water:tBuOH (1:2, 150 μl; separately prepared by mixing 24 mM CuSO₄ solution in water and 24 mM TBTA solution in tBuOH) was added to a stirred solution of LANA_5-22-N₃ (10 mM in water, 72 μl, 720 pmol) and 3,4,5-tris(PEG₇₅₀)-N-(prop-2-yn-1-yl)benzamide (50 mM in tBuOH, 14.4 μl, 720 pmol). Then, sodium ascorbate aqueous solution (100 mM, 60 μl, 6.00 μmol) was added, and the mixture was stirred at room temperature for 22 h. Insoluble materials were removed by filtration, and the filtrate was purified with preparative HPLC (0% acetonitrile for 3 min, followed by a linear gradient of 0–25% acetonitrile over 2 min, then a linear gradient of 25–80% acetonitrile over 35 min with a flow rate of 3.5 ml/min) to produce PEG₇₅₀-LANA; (Fig. S10E) after lyophilization.

To synthesize PEG₇₅₀-LANA-JF646 (Fig. S10F), 276 μl of DMF was added to a stirred solution of PEG750-LANA (10 mM in water, 40 μl, 400 pmol) and Janelia Fluor 646, N-succinimidyl ester (TOCRIS, 10 mM in DMSO, 44 μl, 720 pmol). Then, 40 μl saturated sodium hydrogen carbonate aqueous solution was added, and the mixture was stirred at room temperature for 5 h. Insoluble materials were removed by filtration, and the filtrate was purified with preparative HPLC (0% acetonitrile for 3 min, followed by a linear gradient of 0–25% acetonitrile over 2 min, then a linear gradient of 25–80% acetonitrile over 35 min with a flow rate of 3.5 ml/min) to produce PEG₇₅₀-LANA-JF646 (Fig. S10G) after lyophilization. The final product was analyzed by a MALDI-TOFMS (Fig. S11, Shimadzu Biotech Axima ToF² spectrometer).

To clone zebrafish genes, RNA was prepared from 24 hpf embryos using TRIzol reagent (Ambion, 15596026). Zebrafish wild-type beta-actin (actin wild-type, GenBank accession number: AF025305), Exp6 (GenBank accession number: BC044132), cofilin1 (GenBank accession number: AY398323), and cofilin2 (GenBank accession number: AY398324) cDNA were amplified from the embryo RNA using PrimeScript II High Fidelity One Step RT-polymerase chain reaction (PCR) Kit (Takara, R026A) using the following primers: actin wild type: (5′-ATGGATGATGAAATTGCCGCACTGGTTGT-3′) and (5′-GAAGCATTTGCGGTGGACGATGGATGGTCC-3′); Exp6: (5′-ATGGCGTCAGAGGAGGCCTCCTTACGGGCC-3′) and (5′-TAACTTCAGCGTTCCCGGCGGGAGGCTCCC-3′); cofilin1: (5′-ATGGCCTCAGGTGTAGC-′3) and (5′-TTAGACAGGCTTCCCCTCC-′3); cofilin2: (5′-ATGGCCTCCGGAGTTAC-3′) and (5′-TCAATCGGTTAGAGGCTTTCC-3′). Amplified cDNA was cloned into pSC plasmid (StrataClone Blunt PCR Cloning Kit, STRATAGENE, 240207). Then, the actin R62D mutant was obtained by inverse PCR using the following primers (5′-GATGAGGCTCAGAGCAAGAGAGGTATCCTGACCCTGAAG-3′) and (5′-CTTCAGGGTCAGGATACCTCTCTTGCTCTGAGCCTCATC-3′). The nucleotide sequence was validated by Sanger sequencing. For in vitro transcription, actin WT and R62D were inserted into the EcoRI and NotI sites of the pcDNA3.1+poly(A)83 plasmid (Yamagata et al., 2005) with sfCherry or 3×FLAG-tag and NLS (see below). Exp6 was inserted into the EcoRI and NotI sites of the pcDNA3.1+poly(A)83 plasmid with 3×FLAG-tag. Cofilin1 or cofilin2 were inserted into the EcoRI and NotI sites of the pcDNA3.1+poly(A)83 plasmid with sfCherry and NLS. The resulting pcDNA3.1_sfCherry-NLS+poly(A)83, pcDNA3.1_actin R62D-sfCherry-NLS+poly(A)83, pcDNA3.1_actin WT-sfCherry-NLS+poly(A)83, pcDNA3.1_actin R62D-3×FLAG-NLS+poly(A)83, pcDNA3.1_actin WT-3×FLAG-NLS+poly(A)83 pcDNA3.1_sfCherry-NLS+poly(A)83, pcDNA3.1_Cofilin1-sfCherry-NLS+poly(A)83, pcDNA3.1_Cofilin2-sfCherry-NLS+poly(A)83 were linearized by XbaI digestion, and pcDNA3.1_Exp6-3×FLAG-NLS+poly(A)83 was linearized by XhoI digestion. The linearized plasmids were treated with 0.2 mg/ml proteinase K (Invitrogen, 25530-015), purified with phenol–chloroform treatment and ethanol precipitation, then processed through in vitro transcription using mMESSAGE mMACHINE T7 Transcription Kit (Ambion, AM1344) according to the manufacturer's instruction. The concentration of synthesized mRNA was measured and adjusted to 50, 250, 300, or 800 ng/μl by using nuclease-free water (Cytiva, SH30538.01) containing 0.05% phenol red (Sigma-Aldrich, P5530-5G) as an injection marker.

Nucleoplasmin 2B (NPM, GenBank accession number: NM_001123007) was amplified from 24 hpf embryo RNA using the PrimeScript II High Fidelity One Step RT-PCR kit (Takara, R026A) and the following primers: 5′-ATGAGCAAAACCGAGAAACC-3′ and 5′-TCACGCCTTCGCCTTC-3′. The nucleotide sequence was validated with Sanger sequencing. NPM cDNA was fused to the C-terminus of sfGFP and cloned into pcDNA3.1 containing a poly(A) tail sequence for mRNA synthesis, as described above. After confirming the nuclear localization of sfGFP-NPM, the potential NLS region containing 41 amino acids at the C-terminus was subcloned into the same vector. The 41 amino acids directed sfGFP to the nuclei, so were used as an NLS in zebrafish embryos.

After dechorionation and injection, zebrafish embryos were incubated at 28°C for 5.3 h. At the 50%-epiboly stage, the yolk was removed by hand using a pair of sharp forceps and the cells were washed three times in 0.3× Danieau's buffer. Ten de-yolked embryos were transferred into a 1.5 ml protein low bind-tube (Eppendorf, 0030108116) with the buffer and the number of the embryos was confirmed under a stereoscopic microscope. A gel-loading tip (QSP, 010-R204S-Q) was used to remove as much of the extra buffer as possible. Then, 10 μl 2× SDS-gel loading buffer (125 mM Tris–HCl, 0.01% bromophenol blue, 4% SDS, 20% glycerol, 100 mM DTT) was added to the embryos, mixed by vortex and boiled at 95°C for 5 min. The 10 embryos were loaded into each lane and separated on a 7.5% polyacrylamide gel (SuperSep Ace, Fujifilm Wako Pure Chemical, 191-14931), then transferred onto a polyvinylidene fluoride membrane (PALL, BSP0161) using a semi-dry blotting system (ATTO). After blocking with Blocking-One (Nacalai Tesque, 0395395), the membrane was incubated with anti-DYKDDDDK antibody (1E6, 1 μg/ml, Fujifilm Wako Pure Chemical, 012-22384) overnight at 4°C. After washing with TBST (0.01% Tween-20, 10 mM Tris, 150 mM NaCl, pH 8.0), the membrane was incubated with horseradish peroxidase-conjugated goat anti-Mouse IgG (1:1000 dilution, Jackson ImmunoResearch, 115-035-146) overnight at 4°C. After washing the membrane with TBST, the signal was developed using Western Lightning Plus-ECL (PerkinElmer, NEL104001EA) and detected using a gel documentation system (ATTO, LuminoGraphII).

The intensity ratio of the nucleus to the cytoplasm depicted in Figs 3B, 4B, 7C, S2, S7, S8, and S9 were measured as previously described (Sato et al., 2019) using MATLAB software (https://github.com/lhilbert/NucCyto_Ratio_TimeLapse). Using this code, the nuclear area was segmented by the Cy5-labeled H3K9ac channel, and the cytoplasm was defined as the region between 3.3 μm and 8.3 μm from the surface of the nucleus. To measure the intensity ratio of the nucleus or chromosomes to the cytoplasm of UtrCH or phalloidin shown in Fig. 3C, 5C, and 7B, and Fig. S1C, a single z-section of the equatorial plane of each nucleus was selected, and the nuclear area was determined based on the JF646- LANA or Hoechst signal. The cytoplasmic area was defined as the region between 0.3 μm and 2.5 μm from the edge of the nucleus using ImageJ FIJI (version 2.1.0). For the cell cycle length measurement in Fig. S6B, a microscopy image analysis software AIVIA (version 11.0.1, Leica) was used to track individual nuclei in 3D.

Drawing line graphs, box plots, and statistical analyses were performed by using R software (version 4.0.3; https://www.r-project.org/). The data distribution of each dataset was first analyzed using Kolmogorov–Smirnov test. For parametric samples, the homoscedasticity was analyzed using F-test, and then Student's t-test (unpaired, two-tailed, Figs 3C, 7B and Fig. S1C; comparing N/C ratio with homoscedasticity) was used. For comparing multiple groups, the one-way factorial ANOVA was used (Fig. S6A, comparing embryo viability).

We appreciate Mari Iwabuchi and Kazuo Yamagata for plasmids, Lennart Hilbert for MatLab code, and Masahiko Harata, Kei Miyamoto and members in Kimura lab for discussion. We thank the Biomaterials Analysis Division, Open Facility Center, Tokyo Institute of Technology for DNA sequencing.