Reconstitution of the oocyte nucleolus in mice through a single nucleolar protein, NPM2

In multicellular organisms, germ cells are the only cells that carry genetic information from one generation to the next. During germ cell formation, reprogramming of chromatin occurs to reset the parental epigenetic memory for the creation of new individuals (Saitou et al., 2012). Not only chromatin but also cellular organelles, such as centrosomes and mitochondria, transform their morphology; however, the significance and the mechanism of organelle transformation during gametogenesis are not well-understood (Sutovsky and Schatten, 1998). As gametogenesis reaches the final phase, the nucleolus transforms into a highly compact and fibrillar structure in oocytes and lacks rDNA transcription (Chouinard, 1971). After fertilization, the morphology of the nucleolus gradually changes from an oocyte type to a somatic one, accompanied by the development of active rDNA transcription as early embryonic development proceeds (Geuskens and Alexandre, 1984).

The nucleolus is the largest sub-nuclear structure and is generally considered the site for ribosome biogenesis. However, because the oocyte nucleolus does not exhibit transcription of rRNA, it had been regarded as a remnant of the nucleolus structure during oogenesis or a precursor of the nucleolus for early embryonic development. The first clear clue in regard to the oocyte nucleolus function came from Npm2-knockout experiments, which showed that Npm2-deficient oocytes lacked the nucleolus and were unable to proceed with the correct development in mice (Burns et al., 2003). In amphibians, nucleoplasmin/NPM2 has been shown to have a dual function − i.e. acting as a histone chaperone by binding to histone H2A-H2B (Kleinschmidt et al., 1985; Laskey et al., 1978) and also as an exchanger of sperm-specific basic proteins, such as protamine in mammals, for maternal histones H2A-H2B to promote sperm chromatin decondensation (Ohsumi and Katagiri, 1991; Philpott and Leno, 1992; Philpott et al., 1991). In mice, however, protamine removal and sperm chromatin decondensation occur independently of maternal histone deposition (Inoue and Zhang, 2014) and Npm2-null zygotes have been shown to form an apparently normal paternal pronucleus (Burns et al., 2003). Thus, in mammals, the role of NPM2 in sperm chromatin decondensation remains a mystery. Moreover, the molecular mechanisms of NPM2 in oocyte nucleolus formation are still largely unknown.

To examine the role of the oocyte nucleolus more directly, we previously have generated mouse oocytes that lack the nucleolus by using micromanipulation in a process called enucleolation (Ogushi et al., 2008). Embryos derived from enucleolated oocytes were able to synthesize proteins and to replicate DNA but failed to develop past the first few cleavages, suggesting that the oocyte nucleolus has pivotal functions that are little related to ribosome biogenesis in early embryonic development (Ogushi et al., 2008). Moreover, the critical time point, when the nucleolus is required for early embryonic development, appears to be at the early step of pronucleus formation (Kyogoku et al., 2014; Ogushi and Saitou, 2010). Zygotes from enucleolated mouse oocytes have been shown to exhibit mis-localization of centromeric proteins (Ogushi and Saitou, 2010), delay in enlargement of the paternal pronucleus (Inoue et al., 2011), alteration of DNA replication timing and abnormal deposition of fas death-domain-associated protein (DAXX), a histone H3.3 chaperone protein for repetitive regions of the genome (Fulka and Langerova, 2014; Voon and Wong, 2016). In mouse Npm2-null zygotes, hypo-acetylated histone H3 on the heterochromatin surrounding nucleoli is lost (Burns et al., 2003). Moreover, both zygotes from enucleolated and Npm2-null oocytes show abnormal mitosis exit (Burns et al., 2003; Fulka and Langerova, 2014; Ogushi and Saitou, 2010). However, it is unknown which of those defects described above directly cause the developmental arrest in embryos from enucleolated oocytes and what functions and components in the oocyte nucleolus are required for early-embryonic development.

To better understand these issues, we here analyze the detailed progression of the first cell cycle in parthenotes (i.e. embryos derived from unfertilized oocytes through parthenogenetic activation rather than sperm entry) or zygotes from enucleolated and Npm2-null oocytes by using a combination of immunocytochemistry and live-cell imaging, and show that, in mice, the absence of a nucleolus in oocytes leads to destabilization of the chromatin at centromeric and/or pericentromeric regions and, consequently, to delay of mitosis and chromosome mis-segregation. Moreover, we show by using proteomics analysis that NPM2 is the main component of the oocyte nucleolus, and protein expression of NPM2 alone rescues the defects in early embryonic development caused by enucleolation.

The oocyte nucleolus in a zygote might play a key role in chromatin organization because it is surrounded by pericentromeric heterochromatin (Ogushi and Saitou, 2010). To determine the physiological relevance of heterochromatin anchoring to the nucleolus, we performed enucleolation (Ogushi et al., 2008) and then examined its impact on chromatin organization in parthenotes or zygotes by using antibodies against histone H3 tri-methylated at lysine 9 (H3K9me3), which marks maternal chromatin (Arney et al., 2002) and pericentromeric heterochromatin (Probst et al., 2007), and CREST (calcinosis, Raynaud's phenomenon, esophageal dysmotility, sclerodactyly, telangiectasia) autoimmune serum which labels mouse centromeres (Broccoli et al., 1990). For whole-DNA staining, 4′,6-diamidino-2-phenylindole, dihydrochloride (DAPI) was used. Examination of the chromatin configuration in the parthenotes from enucleolated oocytes allowed us to identify any potential influence of enucleolation on paternal chromatin decondensation in the zygotes (Inoue et al., 2011).

Enucleolated oocytes that had been injected again with a nucleolus served as a control. In control parthenotes/zygotes, the intensely DAPI-positive heterochromatin was anchored to the nucleoli and formed a ring-like structure, whereas in parthenotes/zygotes from enucleolated oocytes the foci that formed were similar to chromocenters. The latter are usually observed in interphase nuclei of somatic cells but never occur in embryos before transcriptional activation of the embryonic genome (Fig. 1A and Fig. S1A) (Probst et al., 2010). Chromocenter-like foci in pronuclei from enucleolated oocytes were always strongly stained for H3K9me3, indicating pericentromeric heterochromatin. Moreover, there were one or two atypical diffuse CREST signals (>2.0 μm²) in pronuclei of both parthenotes and zygotes from enucleolated oocytes (Fig. 1A,B and Fig. S1A,B). This indicates that centromere proteins detected by CREST serum, such as CENP-A, -B and -C, were mis-loaded onto chromatin. The number of CREST-positive foci was not significantly different from that in control oocytes, although the numbers in enucleolated parthenotes/zygotes were more divergent (Fig. 1C and Fig. S1C). This result suggests that centromere clustering, which is important for the spatial genome organization and gene expression (Tjong et al., 2016; Wijchers et al., 2015), was slightly aborted in parthenotes/zygotes from enucleolated oocytes. Therefore, the lack of nucleoli in pronuclei resulted in a disorganization of chromatin architecture establishment at centromeric/pericentromeric regions. Since we observed similar defects both in parthenotes and in zygotes from enucleolated oocytes, these defects were not caused by an abnormality of sperm chromatin decondensation.

The centromeric/pericentromeric chromatin is an essential region of DNA for the kinetochore assembly during mitosis (Fukagawa and Earnshaw, 2014). We, therefore, observed the progression through the first mitosis in parthenotes/zygotes from enucleolated oocytes in live-cell imaging by using mRNAs of histone H2b fused to mCherry (H2b-mCherry) as a chromosome marker and by using antibodies against histone H3 phosphorylation at S10 conjugated to Alexa-Fluor-488 (Fab311-Alexa488) as a chromosome condensation marker (Hayashi-Takanaka et al., 2009). Parthenotes from enucleolated oocytes showed a delayed entry into mitosis (1085±195 min) compared to those from control oocytes (933±74 min), but exhibited similar kinetics of histone H3 phosphorylation at S10 to that in the control group (Fig. 1D,E and Table S1). Eighty-three percent (15/18) of them showed delayed anaphase onset and underwent cell division with lagging chromosomes, resulting in 2-cell embryos with micronuclei. The total length of the first mitosis in parthenotes from enucleolated oocytes was 236±91 min, versus 177±43 min in the control oocytes (Fig. 1E and Table S1). Zygotes from enucleolated oocytes also exhibited a delayed entry into mitosis, which is consistent with the previous report (Fulka and Langerova, 2014), and longer duration of the first mitosis with lagging chromosomes but to a more-severe degree compared to the parthenotes from enucleolated oocytes (Fig. S1E,F and Table S1). The number of the chromosomes in parthenotes/zygotes from enucleolated oocytes was 40 (=2n) on average, suggesting that enucleolated oocytes underwent meiosis correctly, and chromosome individualization and condensation at the beginning of the first mitosis (Fig. S1F). However, chromosomes in parthenotes/zygotes from enucleolated oocytes exhibited elongated heterochromatin formation at their proximal ends that are rich in repetitive elements such as telomeres, major and minor satellites, and ribosomal genes (Fig. 1F and Fig. S1G). As a consequence, when chromosomes were segregated at anaphase, formation of lagging chromosomes was observed (Fig. S1H). Moreover, DNA damage in zygotes from enucleolated oocytes was checked using antibodies against histone H2AX phosphorylation at serine 139 (γ-H2AX) (House et al., 2014). A certain number of γ-H2AX foci were observed not only in zygotes from enucleolated oocytes but also in those from control oocytes; however, diffuse γ-H2AX signals that colocalized with CREST diffuse signals, were only observed in zygotes from enucleolated oocytes (Fig. S2). Considering that the defect of chromosome segregation only occurs in parthenotes/zygotes from enucleolated oocytes, we conclude that regions showing diffused signals recognized by both γ-H2AX and CREST are regions crucial for correct chromosome segregation but that the remaining DNA damage regions were not. Collectively, these findings indicate that absence of an oocyte nucleolus causes two mitosis defects − i.e. delay in the first mitosis and abnormal chromosome segregation − that are probably caused by impaired pericentromeric heterochromatin formation. Moreover, as chromosomes in the enucleolated oocytes were segregated correctly during meiosis, removal of a large structure from the nucleus was not a primary cause of the chromosome segregation defects.

Previous studies reported that some somatic-type nucleolar proteins localize at the oocyte nucleolus only when oocytes were treated with proteinase K or in response to applying an antigen-retrieval protocol (Fulka and Langerova, 2014; Shishova et al., 2015). NPM2 is the only protein whose endogenous localization at the oocyte nucleolus has been observed without any treatment (Inoue and Aoki, 2010). The dearth of reliable information is, at least partly, due to the tight compaction of the oocyte nucleolus, which makes it difficult to detect its components by using immunohistochemistry or other analytical methods (Fig. 2A). To reveal the components of the oocyte nucleolus, isolated nucleoli were digested with the enzymes proteinase K, DNase I and RNase A. The nucleolus treated with proteinase K solution was disassembled within 20 s after treatment. The nucleolus treated with RNase A solution was slightly disassembled after 30 min of treatment, while the nucleolus treated with DNase I solution and that treated with the control solution maintained their structures (Fig. 2B, left). We confirmed the DNase I and RNase A activity by using plasmid DNA and in vitro-synthesized RNA as substrates. Both DNase I and RNase A completely digested 0.5 μg substrates soon after treatment, suggesting that both enzymes have high digestion activity, but could not disassemble the oocyte nucleolus efficiently (Fig. 2B, right). Thus, proteins are structurally important for the oocyte nucleolus and, as reported previously, RNAs are also included in the oocyte nucleolus (Shishova et al., 2015).

To identify protein components that are important for the structure of the oocyte nucleolus, we isolated nucleoli from germinal vesicle (GV)-oocytes in two independent sets of experiments (collecting 1200 and 2006 nucleoli each) and determined their protein composition by mass spectrometry (MS) analyses (Fig. 2C,D and Table S2). The proteins found in the control group (ZP1, ZP2, ZP3, ACTB), those involved in the metabolic processes at the cytosol (TKT, LDHB) and HSPA5, which is involved in the unfolded protein response at the endoplasmic reticulum (Hao et al., 2009), were removed from the candidate list. The localization of candidate proteins identified by MS analysis was assessed by the expression of N-terminal eGFP-tagged fusion constructs in GV-oocytes, and NPM2, NCL, SSRP1, and NOLC1 were found to be localized at the nucleolus (Fig. 2E). Because the enucleolation procedure isolates not only nucleoli but also components of the nucleoplasm and cytoplasm from oocytes, the MS analysis also identified nucleoplasmic and cytoplasmic proteins. Moreover, we confirmed that NPM2 was, indeed, located at the oocyte nucleolus by western blotting and immunostaining of cryosections, which revealed NPM2 antigens in the middle of the nucleolus structure (Fig. 2F,G).

Next, we analyzed in greater detail the Npm2-null oocytes, which lack an apparent nucleolus structure in their GVs and pronuclei (Burns et al., 2003). Chromocenter-like foci were also observed in Npm2-null parthenotes/zygotes, and there were one or two diffuse CREST signals in each pronucleus (Fig. 3A,B and Fig. S3A,B). The average number of centromere foci was significantly reduced in the Npm2-null parthenotes/zygotes compared with that in heterozygous (het) oocytes (Fig. 3C and Fig. S3C). Thus, centromere proteins were mis-loaded onto the chromatin, and centromere clustering was aborted in Npm2-null parthenotes/zygotes.

Live cell imaging of the first mitosis confirmed that Npm2-null parthenotes/zygotes underwent the first mitosis cell cycle at the same time as Npm2-het parthenotes/zygotes, but showed lagging chromosomes at the first anaphase and formation of abnormal 2-cell embryos with micronuclei (Fig. 3D,E; Fig. S3D,E and Table S1). Thus, the lack of NPM2 caused chromosome mis-segregation during the first mitosis even though the start time of mitosis was appropriate. These results suggest that chromosome lagging during first mitosis is induced by loss of NPM2 in oocytes, and mitosis delay is triggered by the oocyte nucleolus factor triggered by the oocyte nucleolus factor, but not by NPM2. In Npm2-null parthenotes/zygotes, the number of chromosomes was 40 (=2n), but the elongated heterochromatin formation at the end of chromosomes was also observed in parthenotes/zygotes from Npm2-null oocytes, and subsequently the chromosomes were segregated at anaphase, resulting in the formation of lagging chromosomes (Fig. 3F and Fig. S3F-H). Taken together, these results led us to conclude that the nucleolar loss leads to a disorganization of higher-order chromatin structure in pronuclei and frequent chromosome mis-segregation during the first mitosis.

The chromosome segregation defects during the first mitosis, such as chromosome lagging, micronucleus formation and elongated heterochromatin formation at the proximal region of chromosomes at the first mitosis in Npm2-null oocytes were similar to those in enucleolated oocytes, and NPM2 was among the most abundant proteins detected, with more than double the number of spectral counts compared to other proteins identified by MS analysis. Thus, we hypothesize that NPM2 is the main and predominant component in the oocyte nucleolus and, therefore, decided to examine whether NPM2 is sufficient to rescue the defects of the enucleolated oocytes.

We first attempted to identify a set of conditions under which the structure and the function of the nucleolus in Npm2-null oocytes is reconstituted by expressing NPM2 again. Thus, we injected Npm2 mRNA into Npm2-null oocytes at second metaphase (MII), and then used these oocytes for IVF and in vitro development (IVD). We found that injection of Npm2 mRNA at a concentration of 1.2 µg/µl, which yielded a much smaller level of NPM2 expression compared to that observed in Npm2-het oocytes, resulted in blastocyst formation in only 7% of the injected oocytes (4/54) (Fig. 4A,B). We, therefore, went on to inject 11 μg/μl of Npm2 mRNA into Npm2-null oocytes, resulting in an NPM2 expression comparable to that in Npm2-het oocytes (Fig. 4B and Fig. S4A). Expression of NPM2 by injection of Npm2 mRNA at MII persisted until blastocyst formation (Fig. 4B and Fig. S4B). Then, we fertilized the Npm2-null oocytes injected with 11 μg/μl of Npm2 mRNA and found that they formed nucleoli of apparently normal morphology in their nuclei at zygote and 2-cell stages (Fig. 4C top and D). We noticed small amorphous structures in the nuclei of Npm2-null oocytes but their electron-densities and morphological properties were different from those of the oocyte nucleoli (Fig. 4D bottom). Remarkably, we found that the reconstituted nucleoli in the Npm2-null zygotes were much larger than those in the controls (Fig. 4C top and D). Considering that the level of NPM2 in the Npm2-null oocytes injected with Npm2 mRNA was higher than that in Npm2-het oocytes (Fig. 4B and Fig. S4A), we conclude that levels of synthesized NPM2 is proportional to the size of the nucleolus. Importantly, 56% (27/48) of the Npm2-null oocytes injected with Npm2-mRNA at 11 μg/μl reached the blastocyst stage, and 51% (31/61) developed to full term, which is comparable to the rate of blastocyst formation and of development to full-term live births in Npm2-null oocytes injected with the isolated nucleoli (Fig. 4A and C bottom; and Table 1). These results indicate that NPM2 expression at MII establishes a situation that rescues the integrity of the nucleolus structure and deals with functional deficiencies of early embryonic development in embryos from Npm2-null oocytes.

Next, we investigated whether NPM2 protein expression alone would be sufficient to reconstitute the nucleolus in embryos from enucleolated oocytes. The critical time point at which the oocyte nucleolus is required is the early step of pronucleus formation before the activation of zygotic genes, when zygotes are not able to newly synthesize mRNAs (Aoki et al., 1997; Kyogoku et al., 2014; Ogushi and Saitou, 2010). Therefore, we injected Npm2-mRNA (11 μg/μl) into enucleolated oocytes at MII and these oocytes injected showed higher levels of NPM2 protein expression than control oocytes injected with an eGfp mRNA (Fig. 5A), were successfully fertilized and, subsequently, formed large nucleoli in their pronuclei of zygotes (Fig. 5B,C). With the expression of NPM2 in zygotes from enucleolated oocytes, the shape of the heterochromatin was restored from a chromocenter-like type to a peri-nucleolar ring type and abnormal diffused CREST signals were not observed (n=28) (Fig. 5C). We also investigated the nascent transcription of total RNA and rRNA by using a bromouridine (BrUTP)-incorporation assay in the absence or presence of the RNA polymerase II inhibitor α-amanitin (Zatsepina et al., 2000). In late 2-cell embryos from enucleolated and Npm2-mRNA-injected oocytes, BrUTP incorporation was detected 30 min after BrUTP injection both in the absence and presence of α-amanitin (Fig. S5). Moreover, consistent with a previous report, nascent rRNA transcription was also observed in 2-cell embryos from enucleolated oocytes (Fulka and Langerova, 2014). Therefore, with or without an oocyte nucleolus, the embryos undergo zygotic gene activation and resume rRNA transcription successfully. Strikingly, 48% (33/69) of embryos from enucleolated oocytes injected with Npm2-mRNA reached the blastocyst stage (Fig. 6A,B) and 21% (16/78) of the 2-cell embryos from NPM2-rescued enucleolated oocytes developed to full term (Table 1). This demonstrates that NPM2 expression at MII is in itself capable to reconstitute the structural integrity of the nucleolus and to rescue the defects in embryos from enucleolated oocytes.

NPM2 consists of an N-terminal core domain followed by a central region comprising two acidic domains (A2, A3) separated by a nuclear localization signal, and also a C-terminal cluster of basic amino acids that is called the the K-rich motif (Fig. 7A). NPM2 has been reported to form homo-pentamers in a manner that depends on the N-terminal core domain (Platonova et al., 2011), whereas the K-rich motif was shown to be required for its localization at the nucleolus (Inoue and Aoki, 2010). To identify the domain of NPM2 that is responsible for oocyte nucleolus formation and rescue of early embryonic development, we constructed two truncation mutants of NPM2, one lacking a K-rich motif (ΔC16) and another lacking a core domain (ΔCore). NPM2 tagged with eGFP at the N-terminus localized predominantly to the nucleolus in oocytes at the GV-stage and in 2-cell embryos but both eGFP-ΔC16 and eGFP-ΔCore concentrated at the nucleoplasm and were excluded from the nucleolus, suggesting that both domains are required for oocyte nucleolus localization and formation (Fig. S6A).

Npm2-wildtype and Npm2-het zygotes formed one or several nucleoli in their nuclei 12 h post-fertilization (hpf) and 24 hpf, respectively (Fig. 7B). Consistent with the notion that the protein levels of NPM2 are proportional to the size of the nucleolus, the nucleolus in Npm2-het zygotes, which showed a reduced NPM2 level, was smaller than that in Npm2-wild zygotes (Fig. 7B,C and Fig. S6B). Neither of the NPM2-truncation mutants rescued the defects in nucleolus formation in zygotes or in early embryonic development of Npm2-null oocytes, whereas the full-length NPM2 did (Fig. 7B-D and Table S3). Moreover, when ΔC16 or ΔCore mRNA was injected into Npm2-het oocytes, these oocytes, although fertilized, did not form correct chromosomes at the first mitosis and arrested at the 2- to 8-cell stage, suggesting that these truncation mutants dominantly negative affect early embryonic development (Fig. 7D and E). We found that full-length NPM2 and ΔC16 form pentamers both in exogenously expressed HEK293T cells and in the oocytes (Fig. S6C), whereas ΔCore did not. Notably, ΔC16 and ΔCore reduced formation of endogenous NPM2 pentamers, indicating that interruption of endogenous NPM2 pentamer formation by ΔC16 or by ΔCore expression in Npm2-het oocytes has dominant-negative effects in chromosome formation and early embryonic development. Alternatively, it is possible that expression of ΔCore, which has a highly acidic region, induced aberrant chromosome decondensation by non-specific binding to basic proteins, such as histones. We conclude that both the N-terminal and the C-terminal domains of NPM2 are essential for the structural integrity and function of the nucleolus.

In this report, we show that proteins are necessary for structural integrity of the oocyte nucleolus and demonstrate by proteomics analysis that NPM2, NCL, SSRP1, and NOLC1 contribute to the nucleolus. We also provide evidence that the oocyte nucleolus can be reconstituted structurally and functionally by the expression of NPM2 only. NPM2 belongs to the NPM family of proteins, which consists of NPM1, NPM2 and NPM3. NPM1 and its binding protein NPM3 are nucleolar phosphoproteins and, in somatic cells, are involved in diverse processes, such as ribosome biogenesis and maintenance of genomic stability (Colombo et al., 2011; Okuwaki et al., 2001). NPM2 shares the conserved NPM-core domain at N-terminus with NPM1 and NPM3; however, nucleoplasmin, the NPM2 ortholog in amphibians, has little effect on ribosome biogenesis but is well known as a histone chaperone for sperm chromatin decondensation (Laskey et al., 1978; Philpott et al., 1991). In our study, both parthenotes and zygotes from enucleolated and Npm2-null oocytes showed defects in centromere clustering and in centromeric/pericentromeric chromatin organization in their pronuclei. Therefore, in contrast to Xenopus nucleoplasmin, which is required for paternal chromatin remodeling, mouse NPM2 is required for both paternal and maternal chromatin remodeling, especially at the centromeric/pericentromeric regions. Because of the complexity of biochemical experiments when using mammalian oocytes, there are few biochemical data of mammalian NPM2 to specify its molecular function. However, one can speculate about the function of NPM2 by comparing it to NPM1 in somatic cells and to nucleoplasmin-like protein (NPL) in Drosophila, since both proteins have the conserved region of the NPM-core domain and also form a spherical body (Liu et al., 2006; Mitrea et al., 2016). For example, NPM1, has the ability to bind to the centromere-specific nucleosome CENP-A and is required for proper kinetochore and microtubule attachments (Amin et al., 2008; Dunleavy et al., 2009; Foltz et al., 2009). Moreover, the deletion of NLP in S2 cells showed similar defects to those in parthenotes/zygotes from enucleolated and Npm2-null oocytes (Liu et al., 2006; Padeken et al., 2013), and NLP binds to CTCF and the centromere-specific nucleosome CenH3, as well as to the NCL homolog Modulo, whose binding is required for centromere clustering to the nucleolus. Since NCL is one of the oocyte nucleolus proteins that was identified in our proteomics analysis as well as in other studies (Fulka and Langerova, 2014; Shishova et al., 2015), it is highly possible that NPM2 binds to NCL and is involved in centromere clustering to the nucleolus and regulation of the centromeric/pericentromeric chromatin organization in interphase, which is required for correct chromosome segregation.

Our study, pericentromeric heterochromatin in the pronuclei of parthenotes/zygotes from enucleolated and Npm2-null oocytes showed atypical diffuse CREST signals and chromocenter-like structures that are never observed in embryos before the embryonic genome activation. It is known that the expression of major satellite repeats induces the reorganization of pericentromeric heterochromatin from peri-nucleolar rings into chromocenters (Probst et al., 2010). However, Fulka and Langerova (2014), have reported the transcriptional repression of major satellites in zygotes from enucleolated oocytes, which is inconsistent with the phenotype of ectopic formation of chromocenters. Therefore, the question whether the nucleolus/NPM2 is involved in the reorganization of pericentromeric heterochromatin via transcriptional regulation of major satellite repeats remains to be answered. Moreover, if so, how do the nucleolus or NPM2 control this process.

Consistent with previous reports, we found that the embryos without oocyte nucleolus components have the ability to resume transcription of rRNA and to synthesize proteins in the same manner as control embryos do, suggesting that the oocyte nucleolus has little effect on ribosome biogenesis (Fulka and Langerova, 2014; Kyogoku et al., 2014; Ogushi et al., 2008). This naturally raises the question, what is the function of the oocyte nucleolus? Our study did not provide a clear answer but recent reports by the groups of Brangwynn, Hyman, and Rosen have provided some suggestions. Based on their observations, these groups claimed that non-membranous organelles that consist of nucleic acids and proteins form liquid-like droplets through phase separation from surrounding environments, a phenomenon that is important for the spatiotemporal control of proteins while maintaining their dynamics with the surrounding organelles or nucleo-cytoplasm (Banani et al., 2017; Bergeron-Sandoval et al., 2016; Brangwynne, 2013; Shin et al., 2017). In somatic cells, NPM1 − one of the NPM2 homologs − undergoes phase separation in a concentration-dependent manner together with the arginine-rich proteins and rRNAs that are required for ribosome biogenesis. This event is critical for structural organization of the somatic nucleoli and, presumably, required to control the activity of ribosome biogenesis (Feric et al., 2016; Holmberg Olausson et al., 2014; Mitrea et al., 2016). In our study, we found that NPM2 has the ability to form a spherical body together with other proteins and RNAs in a concentration-dependent manner. However, it remains to be elucidated whether this NPM2 body has biophysical and functional properties that are similar to those of the liquid-like body, and whether NPM2-body formation during oogenesis occurs due to an enhancement of reactions among components of the oocyte nucleolus or because NPM2 and binding components from surrounding environments are excluded in order to prevent non-favorable reactions in the interphase nucleus.

The molecular function of NPM2 is still unclear but we have clarified here the timing of the NPM2 requirement during oogenesis and early embryogenesis. NPM2 is expressed in all stages of oocytes − from primordial follicles to antral follicles and in those of embryos from zygotes to morulae (Burns et al., 2003; Inoue and Aoki, 2010), but NPM2 protein expression in oocytes at MII rescued the defects in early embryonic development in Npm2-null oocytes. Thus, NPM2 does not have any roles in oocyte growth but is required for the processes after fertilization. Considering that the oocyte nucleous is required at the timing of the early step of pronucleus formation and the nucleolus can be reconstitued by NPM2 protein expression, it appears that NPM2 is required during this step, i.e. when chromatin remodeling and nuclear body reassembly occur (Kyogoku et al., 2014; Ogushi and Saitou, 2010).

We have demonstrated here that the oocyte nucleolus structure can be reconstituted by expressing a single protein − NPM2 – and that an NPM2-based nucleolus structure serves as a key platform to organize pericentromeric heterochromatin in order to correctly segregate parental chromosomes.

All animals were treated with appropriate care according to the ethics guidelines of RIKEN, Kyoto University (Japan) and the University of Oxford (UK). Npm2 heterozygous (Npm2-het) and null (Npm2-null) mice were obtained by crossing Npm2-het females and Npm2-null males (Burns et al., 2003).

Oocytes at the germinal vesicle (GV)-stage were collected from ovaries of B6DBA2F1 females at 8−12 weeks of age at 44−48 h after injection with 7.5 IU equine chronic gonadotropin (eCG; ASKA Pharmaceutical). Fully grown GV-oocytes were released from ovarian follicles by puncture with needles in modified human tubal fluid (HTF) medium (Nippon Medical & Chemical Instruments) supplemented with 200 μM N6, 2′-O-dibutyryladenosine 3′,5′-cAMP sodium salt (dbcAMP; Sigma) and 100 μM 3-isobutyl-1-methylxanthine (IBMX; Sigma), and their cumulus cells were removed by gentle pipetting. For the in vitro maturation (IVM) medium, minimum essential medium (MEM; Sigma) containing 10% fetal bovine serum (FBS; Equitech Bio), 250 μg/ml sodium pyruvate and 8 μg/ml kanamycin was used. GV-oocytes were cultured for 3-4 h in IVM medium containing 200 μM dbcAMP and 100 μM IBMX at 37°C under an atmosphere of 5% CO₂, and then manipulated with a micromanipulator equipped with a PIEZO drive (Prime Tech). Enucleolation (Fulka Jr et al., 2003; Ogushi et al., 2008) and enucleation (Fulka and Fulka, 2007) were performed in modified HTF medium with 200 μM dbcAMP, 100 μM IBMX and 10-15 μg/ml cytochalasin D (Sigma) under an inverted microscope (Olympus IX71) equipped with a micromanipulator (Narishige). Enucleolated oocytes that had been reinjected with a nucleolus served as a control. After 1 h of recovery culture in IVM medium with dbcAMP and IBMX, the control and enucleolated oocytes were washed and cultured with IVM medium to resume meiosis. Nucleoli isolation was performed by enucleolation, and isolation of GVs and cytoplasts was done by enucleation; these samples were used for enzyme digestion, MS analysis or immune-blot analysis.

For in vitro fertilization (IVF), spermatozoa were obtained from the caudal epididymides of ICR males at 10-14 weeks of age and cultured for 2 h in HTF medium (Quinn et al., 1985) to induce capacitation. After 18-20 h of IVM, control and enucleolated oocytes at second metaphase (MII) were inseminated with capacitated spermatozoa at a concentration of 2.0×10⁷ sperm/ml for 4-6 h in HTF medium. In Npm2-het/-null mice, oocytes at MII were collected from oviducts of 8−12-week-old females that had been injected with 7.5 IU of eCG followed 48 h later by 7.5 IU of human chorionic gonadotropin (hCG; ASKA Pharmaceutical), and inseminated with spermatozoa at the concentration of 2.0×10⁶ sperm/ml for 4-6 h in HTF medium. For parthenogenetic activation, oocytes at MII were artificially activated by treating them with modified Tris-buffered medium (20 mM Tris, 113 mM NaCl, 3 mM KCl, 11 mM glucose, 5 mM Na pyruvate, and 2 mg/ml BSA) or Ca²⁺-excluding CZB medium (Chatot et al., 1989) containing 5−10 mM SrCl₂ and 5 μg/ml cytochalasin B for 4 h. For intracytoplasmic injection (ICSI), epididymal spermatozoa were capacitated in CZB medium for 1 h, and resuspended in HEPES-CZB medium containing 12% polyvinylpyrrolidone (Wako Chemicals). Oocytes at MII were placed in HEPES-CZB, and a sperm head isolated by several PIEZO pulses was injected into oocytes under an inverted microscope equipped with a micromanipulator. These oocytes were then transferred to K⁺- supplemented simplex optimised medium (KSOM; Millipore) or CZB medium in order to assess the progression of early-embryonic development.

Cultured oocytes and parthenotes/zygotes were fixed in 2% paraformaldehyde (PFA) in PBS or in 20 mM Tris-HCl, pH 7.5, and 150 mM NaCl (KB) for 30 min. After permeabilization with 0.2% Triton X-100 in PBS or in KB overnight at 4°C, oocytes and parthenotes/zygotes were incubated for 1 h in PBS or KB with 5 mg/ml BSA at room temperature, and incubated with primary antibodies overnight at 4°C. Following three washes with PBS or KB, Alexa-Fluor-labeled secondary antibodies (Invitrogen) were used for the detection of signals and DNA was counterstained with 14.3 μM 4′,6-diamidino-2-phenylindole, dihydrochloride (DAPI). The following primary antibodies were used: anti-HISTONE H3 (trimethyl K9) (Abcam #ab8898, 1:250-400), CREST serum (a kind gift from Y. Muro, Nagoya University, Japan; 1:2000), anti-GFP (Nacalai Tesque, GF090R, #04404-84, 1:200), anti-Phospho-HISTONE H2AX (S139) (γ−H2AX, Cell Signaling #2577, 1:200) and anti-NPM2 (Burns et al., 2003; 1:200−1:400). The samples were mounted with Vectashield mounting medium (Vector Laboratories) and observed under an Olympus FV-1000 confocal microscope or a Zeiss LSM780 confocal microscope. For detection and analysis of pronuclei in parthenotes/zygotes, samples were scanned in three dimensions and sectioned into z-series of 0.3−1.0 μm slices; then the slices were combined and maximally projected by Image J. The area of centromere foci in pronuclei was determined by using the threshold function and were calculated by using Image J. The foci with an area of 0.2−2.0 μm² were defined as normal and those with an area of >2.0 μm² were defined as diffuse.

The zona pellucida of parthenotes or zygotes at the first mitosis was removed by treatment with acidic Tyrode's solution (Sigma); then, the parthenotes/zygotes were placed on a glass slide that had been dipped in a solution of 1% PFA in destilled water, pH 9.2, containing 0.15% Triton X-100 and 3 mM dithiothreitol. Following overnight fixation in a humid chamber, the slide was washed in 0.4% Photoflo (Kodak) in destilled water and dried for 30 min at room temperature. The samples were washed in PBS and then incubated with anti-HISTONE H3 (trimethyl K9) antibody (1:250−1:400) overnight at 4°C. After washing with PBS, the samples were incubated with Alexa-Fluor-488-conjugated donkey anti-rabbit IgG for 1.5 h at room temperature and mounted with Vectashield containing DAPI (Vector Laboratories).

To visualize the chromatin and chromosome condensation status, oocytes at MII were injected with H2b-mCherry mRNA at 200 ng/μl, and Fab311 antibody conjugated to Alexa-Fluor-488 (Fab311-Alexa488) at 1 μg/μl, which recognizes the phosphorylated Ser10 of histone H3. After injection, oocytes were cultured for 2−4 h in IVM medium to express mRNAs. Following artificial activation or ICSI, oocytes were transferred to 4 μl drops of CZB medium on a glass-bottomed dish and placed in an incubator (MI-IBC; Tokai Hit) at 37°C under 5% CO₂ in air. Observation was performed under an inverted microscope (IX-71; Olympus) equipped with a Nipkow disk confocal unit (CSU-10; Yokogawa Electric Corp.). This system was housed in a dark room at 30°C. Device control and image analysis were performed using MetaMorph software (Molecular Devices). For time-lapse observations, images were taken at 10 min intervals. At each time point, 51 fluorescent images were taken 2 μm apart at the z-axis for optical sectioning.

To determine the components of the oocyte nucleoli, isolated nucleoli were treated with proteinase K (Roche) at 2 μg/10 μl, Turbo DNase I (Ambion) 1 U/ 10 μl and RNase A at 0.25 U/10 μl in modified HTF medium without BSA, and were observed under an inverted microscope (Olympus IX71). To check the activity of DNase I and RNase A, 0.5 μg circular plasmid DNA, pCS2-DsRed and 0.5 μg in-vitro synthesized RNA, DsRed-Pias I in 10 μl of modified HTF medium without BSA were treated with each enzyme as described above and kept in liquid nitrogen. After digestion, samples were treated with proteinase K at 200 μg/ml for 10 min at 50°C and enzymes were inactivated by boiling for 5 min at 95°C. DNA and RNA were separated by electrophorese using 1% agarose gel.

The nucleoli isolated from GV-oocytes by enucleolation were kept in evacuated zona pellucidae and used for mass spectrometry (MS) analysis (n=2; the numbers of nucleoli and those of zona pellucidae were 1200 or 2006 and 20 or 33, respectively). Nucleolar proteins were loaded on a NuPAGE™ 4−12% Bis−Tris gel and separated in 3-(N-morpholino)propanesulfonic acid (MOPS) buffer until SERVA Blue G reached the middle of the gel; then gel was stained with Coomassie G-250 (Invitrogen) and those areas of the gel containing proteins was cut into nine sections. After in-gel digestion with modified trypsin, the resulting peptides were analyzed by liquid chromatography (LC)-MS by using an LXQ linear ion trap mass spectrometer equipped with a nano-ESI source (Thermo Finnigan). From one sample, analysis of LC-MS was performed twice for technical duplication. The ion spectrum data were screened against the NCBI non-redundant protein database with the Mascot program (Matrix Science) and the International Protein Index database (mouse IPI http://www.ebi.ac.uk/IPI/IPImouse.html) was used to identify the proteins (Nozawa et al., 2010).

All protein cDNA identified by MS analysis was amplified from mouse ovary cDNAs by PCR, then subcloned into pCS2 or pCS2-eGfp (a kind gift from J. Lee, Kobe University, Japan). Npm2-truncated mutants were kindly provided by F. Aoki (Tokyo University, Japan) and subcloned into pCS2 or pCS2-eGfp. pcDNA3-H2b-mCherry was generously provided by T. Abe and T. Fujimori (RIKEN CDB, Japan). pHTN was purchased from Promega (#G7721). After confirmation of the sequence in each plasmid, 5 μg of plasmids were linearized by an appropriate restriction enzyme shortly after the sequence of the poly-A signal and transcribed in vitro for 2−4 h at 37°C by using an mMESSAGEmMACHINE kit containing the appropriate RNA polymerase (Ambion). Following Turbo DNase I digestion for 15 min at 37°C, RNA was precipitated by lithium chloride, and resuspended in nuclease-free H₂O. Each RNA was aliquoted and stored at −80°C until use.

Oocytes at the GV-stage that had been cultured for 2−3 h in IVM medium with dbcAMP and IBMX, were injected with mRNAs at 500 ng/µl. They were then further cultured for 4−6 h to induce mRNA expression. In the case of embryos, oocytes at MII were injected with mRNAs at 50−1000 ng/μl, cultured an additional 4 h for mRNA expression and then activated and/or fertilized. eGFP signals were observed under an inverted microscope (Olympus IX71) during epi-fluorescence illumination provided by a Xenon lamp.

Ovaries were collected from ICR females at 3−4 weeks and washed three times in PBS containing 0.2% Tween20 (PBST) for 10 min at 4°C. Ovaries were soaked in a sucrose in PBS gradient of 10−30%, then embedded in optimal cutting temperature (OCT) compound (Sakura Finetek), frozen and sectioned at a thickness of 8−9 μm at −20°C. Air-dried sections were washed three times in PBS and fixed with 2% PFA in PBS for 30 min. After three washes in PBS containing 0.01% polyvinyl alcohol (PBS-PVA), samples were permeabilized in 0.1% Triton X-100 in PBS-PVA, incubated in the blocking solution (3% BSA, 0.1% Tween20, PBS-PVA) and then incubated with antibodies against NPM2 (1:200) or NPM1 (Sigma, FC82291, #B0556, 1:200) in staining solution (1% BSA, 0.1% Tween20, PBS-PVA) overnight at 4°C. After washing with PBS-PVA three times, samples were incubated in staining solution containing secondary antibodies. Samples were then washed three times with PBS-PVA and mounted with Vectashield containing DAPI.

For nucleolus reconstitution experiments, Npm2 mRNA, eGfp mRNA, Halo mRNA or the nucleolus from GV-oocytes was injected into control/enucleolated or Npm2-het/-null oocytes at MII with the assistance of a PIEZO impact drive unit that allowed the use of an injection pipette with a slightly wider bore. After 4 h of culture in IVM medium, the injected oocytes were subjected to IVF and IVD as described above.

Two-cell embryos arising from Npm2 mRNA or nucleolus-injected oocytes were transferred into the oviducts of d-0.5 pseudo-pregnant ICR female mice at 8−12 weeks of age. Caesarian section was performed at E19.5 and surviving pups were fostered by ICR foster mother that had given birth the previous or the same day.

In general, we followed the procedure developed by Zatsepina et al. (2000). Briefly, to detect nascent transcripts of total RNA or rRNA, 1 pl of 100 mM 5-bromourudine 5′-triphosphate (BrUTP) in an injection buffer (140 mM KCl, 2 mM PIPES pH 7.4) with or without 50 μg/ml α-amanitin was injected into one blastomere of 2-cell parthenotes 38-40 h after artificial activation. Another blastomere was used as a negative control of BrUTP incorporation. Only for the detection of rRNA nascent transcripts, 2-cell parthenotes were cultured for 30 min in KSOM containing 10 μg/ml α-amanitin prior to injection of BrUTP. 30-45 min after injection, parthenotes were fixed in 3% PFA in PBS-PVA for 30 min and permeabilized in 0.5% Triton X -100in PBS for 10 min. They were then incubated for 1 h in PBS with 5 mg/ml BSA at room temperature, and reacted with a primary antibody, anti-Bromodeoxyuridine (Roche, BMC9318, #1170376001, 1:50), overnight at 4°C. Following three washes with PBS-PVA, parthenotes were incubated with solution containing Alexa-Fluor-labeled secondary antibodies (Invitrogen), and DNA was counterstained with DAPI. Finally, samples were washed three times with PBS-PVA and mounted using Vectashield.

Samples were lysed in NuPAGE sample buffer (Invitrogen) at 70°C for 10 min, loaded onto NuPAGE™ 4-12% Bis-Tris gel, separated in MOPS or 2-(N-morpholino)ethanesulfonic acid (MES) buffer, and blotted onto an Immobilon-P transfer membrane (Millipore). After blocking with 10% non-fat dry milk in 1×TBS containing 0.1% polysorbate 20, membranes were probed with α-tubulin (Sigma, DM1A, #T9026, 1:2500−1:10,000), GFP (Nacalai Tesque, 1:1000), CREST (1:5000−1:10,000) and NPM2 (1:5000). Primary antibodies were detected with appropriate secondary antibodies conjugated with HRP (Jackson Immunoresearch) followed by detection using a Chemi-Lumi One Super (Nacalai) or SuperSignal West Femto detection system (Thermo Fisher Scientific).

HEK293T cells were cultured in DMEM culture medium (Invitrogen) supplemented with 10% fetal calf serum and 8 μg/ml kanamycin. Cells at the concentration of 2.0×10⁵ were seeded in a six-well dish, cultured for 24 h and transfected with 3 μg plasmid DNA using Lipofectamine 2000 (Invitrogen) according to manufacturer's instructions. Following 24 h of transfection, the transfected cells were extracted with sample buffer for immunoblot analysis.

Samples were fixed by 4% paraformaldehyde and 2% glutaraldehyde overnight at 4°C. After several washes with PBS, samples were embedded in 3% agarose-PBS. They were then post-fixed using 2% OsO₄ in 100 mM phosphate buffer for 2 h at room temperature, dehydrated in a series of graded ethanol solutions, embedded in epoxy resin and polymerized at 60°C for 3 days. Tissue samples were cut into ultra-thin sections (70 nm) on an ultramicrotome EM UC6 (Leica). Following staining with uranyl acetate and lead citrate staining solution (Sigma), the ultra-thin sections were viewed under an H-7650 transmission electron microscope (Hitachi-Hightech).

We thank Fugaku Aoki for the Npm2-truncation mutant plasmids, Jibak Lee for the pCS2 and pCS2-eGfp plasmids, Takaya Abe and Toshihiko Fujimori for the pcDNA3-H2b-mCherry plasmid, Yoshinao Muro for CREST serum, Kaori Yamanaka and Fumiaki Itoi for their assistance in live-cell imaging, Koji Nagao for MS analysis, and Bungo Akiyoshi, Takayuki Hirota and members of the Saitou and Nasmyth labs for their critical input. We are immensely indebted to Akira Nakamura for guidance with the molecular biological procedures and to Jonathan Godwin and Kim Nasmyth for their kind support of revision experiments.