Changes in subcellular structures and states of Pumilio1 regulate the translation of target Mad2 and Cyclin B1 mRNAs

Temporal and spatial control of mRNA translation has emerged as a major mechanism for promoting diverse biological processes. However, the molecular nature of temporal control of translation remains unclear. In oocytes, many mRNAs are deposited as a translationally repressed form and are translated at appropriate timings to promote the progression of meiosis and development. Here, we show that changes in structures and states of the RNA-binding protein Pumilio1 regulate the translation of target mRNAs and progression of oocyte maturation. Pumilio1 was shown to bind to Mad2 and Cyclin B1 mRNAs, assemble highly clustered solid-like aggregates, and surround Mad2 and Cyclin B1 RNA granules in mouse oocytes. These Pumilio1 aggregates were dissolved by phosphorylation prior to the translational activation of target mRNAs. Stabilization of Pumilio1 aggregates prevented the translational activation of target mRNAs and oocyte maturation. Together, our results provide an aggregation-dissolution model for the temporal and spatial control of translation.


Introduction 1
Diverse biological processes including meiosis, embryonic development and neuronal 2 plasticity are promoted by translational activation of dormant mRNAs at appropriate 3 timings and places (Buxbaum et al., 2015;Doyle and Kiebler, 2011;Martin and 4 Ephrussi, 2009;Mendez and Richter, 2001). This temporal control of translation has 5 been most extensively studied in oocyte meiosis. Fully grown vertebrate oocytes are 6 arrested at prophase I of meiosis and accumulate thousands of translationally repressed 7 mRNAs in the cytoplasm (de Moor et al., 2005;Kotani et al., 2017;Masui and Clarke, 8 1979). In response to specific cues such as hormones, oocytes resume meiosis and are 9 arrested again at metaphase II. This process is termed oocyte maturation and is 10 necessary for oocytes to acquire fertility. For proper progression of oocyte maturation, 11 hundreds of dormant mRNAs are translationally activated in periods specific to distinct 12 mRNAs (Chen et al., 2011). Of these, Cyclin B1 mRNA, which encodes the regulatory 13 subunit of maturation/M-phase-promoting factor (MPF), is translated in meiosis I, and 14 the newly synthesized Cyclin B1 proteins in this period are prerequisite for the 15 progression of meiosis (Davydenko et al., 2013;Hochegger et al., 2001;Kondo et al., 16 2001;Kotani and Yamashita, 2002;Ledan et al., 2001;Polanski et al., 1998). 17 Translational activation of the dormant mRNAs including Cyclin B1 has been shown 18 to be directed by the cytoplasmic polyadenylation of mRNAs, which is mediated by the 19 cytoplasmic polyadenylation element (CPE) in their 3' UTR (McGrew et al., 1989;20 Sheets et al., 1994). The CPE-binding protein (CPEB) functions in both repression and 21 direction of the cytoplasmic polyadenylation (Barkoff et al., 2000;de Moor and Richter, 22 1999;Gebauer et al., 1994;Tay et al., 2000). Although many dormant mRNAs contain 23 CPEs, they are translated in different periods during oocyte maturation, indicating that 24 there must be additional mechanisms to determine the timings of translational activation 25 of distinct mRNAs. However, the molecular and cellular mechanisms by which 26 translational timings of hundreds of mRNAs are coordinated remain unclear. 27 Pumilio1 (Pum1) is a sequence-specific RNA-binding protein that belongs to the 28 Pumilio and Fem-3 mRNA-binding factor (PUF) family, which is highly conserved in 29 eukaryotes from yeast to human (Spassov and Jurecic, 2003;Wickens et al., 2002). Pum 30 was identified in Drosophila as a protein that is essential for posterior patterning of 31 embryos (Lehmann and Nussleinvolhard, 1987) and it was shown to repress the 32 translation of target mRNAs in a spatially and temporally regulated manner (Asaoka-33 P granules are cytoplasmic granules that consist of mRNAs and RNA-binding 23 proteins and have been shown to behave as liquid droplets with a spherical shape in C. 24 elegance embryos (Brangwynne et al., 2009). In addition, several RNA-binding proteins 25 that are assembled into stress granules were shown to produce liquid droplets in vitro 26 and in cultured cells (Lin et al., 2015;Molliex et al., 2015). Although phase changes in 27 these liquid droplets into solid-like assemblies have been linked to degenerative 28 diseases (Li et al., 2013;Weber and Brangwynne, 2012), more recent studies have 29 demonstrated the assembly of solid-like substructures within stress granules (Jain et al., 30 2016;Shiina, 2019;Wheeler et al., 2016), suggesting physiological roles of the solid-31 like assemblies. However, biological function of phase changes of protein assemblies 32 from liquid to solid states and vice versa remains to be explored. 33 In this study, we identified Mad2 mRNA as one of the Pum1-target mRNAs in mouse 1 oocytes and found that Mad2 and Cyclin B1 mRNAs were distributed as distinct 2 granules in the cytoplasm. Interestingly, Pum1 was assembled into solid-like aggregates 3 exhibiting highly clustered structures, and these aggregates surrounded Mad2 and 4 Cyclin B1 RNA granules. The Pum1 aggregates dissolved in an early period after 5 resumption of meiosis by phosphorylation, resulting in a liquid-like state and 6 translational activation of Mad2 and Cyclin B1 mRNAs. These results provide an 7 aggregation-dissolution model, accompanied by phase changes of RNA-binding 8 proteins, for temporal and spatial control of mRNA translation. The results also showed 9 the physiological importance of phase changes of proteins in RNA regulation. 10 11

Expression of Mad2 is translationally regulated during mouse oocyte maturation 13
Mad2 has been shown to function as a component of spindle assembly checkpoint 14 proteins to accurately segregate chromosomes in meiosis I of mouse oocytes (Homer et 15 al., 2005). However, how Mad2 is accumulated in oocytes remains unknown. To clarify 16 the mechanism of Mad2 accumulation in meiosis I, we first analyzed the expression of 17 Mad2 mRNA in mouse oocytes. Although two splicing variants of Mad2 mRNA were 18 isolated using purified RNAs from ovaries (Fig. 1A), quantitative PCR,and in 19 situ hybridization analyses showed that the short version of Mad2 mRNA was dominant 20 in oocytes (Figs. 1A-B and S1A). FISH analysis showed that Mad2 mRNA was 21 distributed in the oocyte cytoplasm by forming RNA granules (Fig. 1C). The amount of 22 Mad2 as well as that of Cyclin B1 increased after resumption of meiosis (Fig. 1D). Mad2 mRNA is a Pum1-target mRNA and forms granules distinct from Cyclin B1 31

RNA granules 32
We then assessed the mechanism by which the translation of Mad2 mRNA is temporally 33 regulated. Since Mad2 mRNA was translated in a period similar to that for Cyclin B1 1 mRNA and contains several putative Pumilio-binding elements (PBEs) in its 3'UTR 2 ( Fig. S2A), we investigated whether Pum1 binds to Mad2 mRNA by using an 3 immunoprecipitation assay followed by RT-PCR. Mad2 and Cyclin B1 mRNAs, but not 4 a-tubulin and b-actin mRNAs, were detected in precipitations with an anti-Pum1 5 antibody, while neither of them was detected in precipitations with control IgG (Fig.  6 2A), indicating that Pum1 targets Mad2 mRNA as well as Cyclin B1 mRNA. From 7 these results, we speculated that both mRNAs were assembled into the same granules. 8 However, double FISH analysis showed that the two mRNAs formed distinct granules 9 ( Fig. 2B). Notably, granules containing Mad2 or Cyclin B1 mRNA were found to be 10 distributed close to each other (Fig. 2B, arrows). 11 Time course analysis showed that the number of Mad2 RNA granules was decreased 12 at 4 h (prometaphase I) and that the granules had almost completely disappeared at 18 h 13 (metaphase II) after resumption of meiosis, being consistent with the changes in Cyclin

Pum1 forms aggregates that surround target mRNAs 20
To further assess the mechanism by which translation of Mad2 and Cyclin B1 mRNAs 21 is temporally regulated by Pum1, we analyzed the distribution of Pum1 in the oocyte 22 cytoplasm. Immunofluorescence analysis showed that Pum1 was ununiformly 23 distributed in the cytoplasm of immature oocytes and appeared to form aggregates in 24 highly clustered structures (Fig. 3A). FISH analysis showed that Pum1 aggregates 25 surrounded and partially overlapped Cyclin B1 and Mad2 RNA granules (Fig. 3B). To 26 assess the molecular mechanisms of Pum1 aggregation, we then examined the 27 distribution of GFP-Pum1 and mutant forms of Pum1 by injecting mRNA into mouse 28 oocytes. GFP-Pum1 was distributed in a way similar to that of endogenous Pum1, i.e., it 29 appeared to form highly clustered aggregates ( Fig. 3C and D) surrounding Cyclin B1 30 and Mad2 RNA granules (Fig. S2B). Pum1 contains a glutamine/asparagine (Q/N)-rich 31 domain (Fig. 3E), also identified as a prion-like domain (Fig. S2C;Lancaster et al., 32 2014), which is thought to promote highly ordered aggregation of proteins (Lancaster et 33 We next examined the properties of GFP-Pum1 in mouse oocytes by FRAP analysis. 23 As a control, GFP-Pum1∆QN was analyzed. After photobleaching, the fluorescence of 24 GFP-Pum1 and GFP-Pum1∆QN gradually recovered (Fig. 4B). The fluorescence 25 recovery curves were fitted to a double exponential association model. The half time of 26 recovery (t1/ 2) of the first fraction of GFP-Pum1 was rapid, while that of the second 27 fraction of GFP-Pum1 was slow (Fig. 4C, left), suggesting that a part of Pum1 forms 28 large complexes. Moreover, a critical finding was that a significant fraction of GFP-29 Pum1 (40.7% ± 8.6%, n = 12) showed immobility (not recovering after 30 photobleaching), while only a small fraction of GFP-Pum1∆QN (13.6% ± 5.5%, n = 14) 31 was static (Fig. 4B and C,right). Thereby, the Q/N-rich region promotes the assembly 32 of Pum1 into highly ordered aggregates in an immobile state. 33 We further analyzed the properties of Pum1 by permeabilizing oocytes with 1 digitonin. GFP rapidly diffused out of the oocytes after permeabilization ( Fig. 4D and 2 E). In contrast, the structure and intensity of GFP-Pum1 aggregates persisted after 3 permeabilization ( Fig. 4D and E). Taken together, the immunofluorescence, 4 ultracentrifugation, FRAP and permeabilization analyses demonstrate that Pum1 5 assembles into aggregates in a solid-like state in immature oocytes. A recent study 6 demonstrated that GFP-Pum1 forms solid-like substructures of RNA granules in human 7 culture cells (Shiina, 2019), being consistent with our results in oocytes. had disappeared (Fig. 2C), suggesting a link between translational activation of target 16 mRNAs and Pum1 dissolution. Consistent with these observations, the 17 ultracentrifugation assay showed that a large part of endogenous Pum1 became soluble 18 (69.0% ± 4.4%, n = 3) in mature oocytes, compared with the soluble fraction in 19 immature oocytes (35.2% ± 3.4%, n = 3) (Fig. 4A). FRAP analysis in mouse oocytes 20 indicated that the t1/ 2 of GFP-Pum1 was not significantly different between immature and 21 mature oocytes (Fig. 5B and C,left). In contrast, the percentage of immobile fractions 22 of GFP-Pum1 was significantly reduced in mature oocytes (18.8% ± 6.8%, n = 6) 23 compared with that in immature oocytes (40.7% ± 8.6%, n = 12) (Fig. 5B and C,right). 24 Taken together, the results indicate that Pum1 aggregates dissolve during oocyte 25 maturation and suggest that the change in the property of Pum1 from insoluble, solid-26 like to soluble, liquid-like is crucial for temporal regulation of target mRNA translation. 27 28

Stabilization of Pum1 aggregates prevents the translation of target mRNAs 29
We next assessed whether the change in the property of Pum1 was involved in the 30 translational regulation of target mRNAs. By observing the distributions of truncated 31 forms of Pum1 after resumption of meiosis, we found that the large aggregates of GFP-32 Pum1∆C were stable and persisted until 18 h (Fig. 6A). In contrast, GFP-Pum1∆QN no 33 longer formed aggregates (Fig. S3A), and the aggregates of GFP-Pum1∆N were 1 dissociated within 4 h ( Fig. S3B and Fig. 6A). Consistent with the observations after 2 resumption of meiosis, GFP-Pum1, Pum1∆QN, and Pum1∆N did not affect the 3 progression of oocyte maturation, while GFP-Pum1∆C prevented polar body extrusion 4 ( Fig. 6A and B). Temporal synthesis of proteins is required for proper spindle formation 5 in meiosis I (Davydenko et al., 2013;Kotani and Yamashita, 2002;Polanski et al., 1998;6 Susor et al., 2015). In oocytes expressing GFP-Pum1∆C, meiosis I spindles were 7 defective and synthesis of Mad2 and Cyclin B1 was attenuated ( Fig. 6C and D). These 8 results suggest that insoluble GFP-Pum1∆C inhibited translational activation of Pum1-9 target mRNAs by stabilizing Pum1 aggregates, resulting in failure in spindle formation 10 and polar body extrusion. Since Pum1 targets thousands of mRNAs in the testis and 11 brain (Chen et al., 2012;Zhang et al., 2017), syntheses of many proteins responsible for 12 correct spindle formation would be attenuated in oocytes expressing GFP-Pum1∆C. 13 We then examined the effects of Pum1 inhibition on the progression of oocyte 14 maturation by injecting the anti-Pum1 antibody. To effectively analyze the effect of the 15 anti-Pum1 antibody, we incubated oocytes with 1 µM milrinone, which partially 16 prevents resumption of meiosis. Under this condition, 50-90% of the oocytes underwent 17 germinal vesicle breakdown (GVBD) ( Fig. 6E and F) in a manner dependent on protein 18 synthesis (Fig. 6E). Injection of the anti-Pum1 antibody, but not control IgG, prevented 19 GVBD and dissolution of GFP-Pum1 aggregates ( Fig. 6F and G). The injected anti-20 Pum1 antibody was distributed within the cytoplasm in a way similar to that of 21 endogenous Pum1 (Fig. 6H). These results strongly suggest that the anti-Pum1 antibody 22 inhibited the dissolution of endogenous Pum1 aggregates and thereby prevented the 23 translational activation of Pum1-target mRNAs. 24 25

Pum1 phosphorylation promotes the dissolution of aggregates 26
We finally assessed the mechanism by which Pum1 aggregates are dissolved. As 27 observed in Xenopus and zebrafish (Ota et al., 2011;Saitoh et al., 2018), the 28 electrophoretic mobility of Pum1 was reduced in mature mouse oocytes (Fig. 7A, left). 29 This reduction was recovered by phosphatase treatment (Fig. 7A, right), indicating that 30 Pum1 is phosphorylated during mouse oocyte maturation. Treatment of immature 31 oocytes with okadaic acid (OA), a protein phosphatase 1 and 2A (PP1 and PP2A) 32 inhibitor, induced Pum1 phosphorylation and rapid dissolution of Pum1 aggregates (Fig.  33 7B-D). These results suggest that kinases responsible for Pum1 phosphorylation are 1 present and at least partially active in immature oocytes. Polo-like kinase (Plk) 1 and 4 2 were shown to be present in immature mouse oocytes (Bury et al., 2017;Pahlavan et al., 3 2000). Interestingly, inhibition of Plk4, but not that of Plk1, prevented the dissolution of 4 Pum1 aggregates (Figs. 7C-D and S3C). Inhibition of Plk4 also prevented the 5 phosphorylation of Pum1 (Fig. 7E). These results indicate that Plk4-mediated 6 phosphorylation of Pum1 promotes dissolution of Pum1 aggregates. Recent studies have demonstrated that many of the RNA-binding proteins harbor prion-27 like domains and that some of these proteins have the ability to assemble RNA granules 28 (Decker et al., 2007;Gilks et al., 2004;Reijns et al., 2008). These RNA-binding 29 proteins were shown to promote liquid-liquid phase separation, resulting in the 30 assembly of protein-RNA complexes into droplets (Elbaum-Garfinkle et al., 2015;Lin 31 et al., 2015;Molliex et al., 2015;Nott et al., 2015). These droplets are thought to 32 function as partitions that effectively maintain stability and/or translational repression of 33 mRNAs. In contrast, phase transition of the liquid droplets into solid-like structures 1 such as amyloid fibrils has been thought to contribute to pathological diseases such as 2 amyotrophic lateral sclerosis (ALS) (Li et al., 2013;Weber and Brangwynne, 2012). 3 However, more recently, solid granules were found to assemble during muscle 4 regeneration in a physical state (Vogler et al., 2018). In addition, core regions of stress 5 granules were shown to exhibit solid-like properties (Jain et al., 2016;Shiina, 2019;6 Wheeler et al., 2016). Although these findings suggest the involvement of solid granules 7 in RNA regulation, the physiological importance of the phase changes of protein 8 aggregation from liquid to solid states and vice versa remains unclear. 9 In this study, we demonstrated that Pum1 assembled into aggregates in highly 10 clustered structures through the Q/N-rich region and these aggregates showed solid-like 11 properties in immature oocytes (Figs. 3 and 4). After initiation of oocyte maturation, the 12 Pum1 aggregates dissolved into a liquid-like state (Figs. 4A and 5). The mutant form of 13 Pum1 that lacks the C-terminal PUF domain, Pum1∆C, is expected to be unable to bind 14 to target mRNAs but to have the ability to form assemblies via the Q/N-rich region. 7C-E), suggesting that Plk4 is a kinase responsible for Pum1 phosphorylation and 18 aggregate dissolution. However, other kinases would phosphorylate Pum1, since 19 inhibition of Plk4 activity delayed, but did not completely prevent, the disolution of 20 Pum1 aggregates and Pum1 phosphorylation after initiation of oocyte maturation 21 (unpublished data). Puf3, one of the PUF family proteins in yeast, was shown to be 22 phosphorylated at up to 20 sites throughout the entire region (Lee and Tu, 2015). In 23 addition, we previously showed that Pum1 was phosphorylated at multiple sites in an 24 early period of oocyte maturation in zebrafish (Saitoh et al., 2018). Although the 25 phosphorylation sites responsible for the aggregate dissolution remain to be identified, 26 these results suggest that many sites including those in the Q/N-rich domain might be 27 phosphorylated, resulting in Pum1 aggregate dissolution. 28 29

Subcellular structures of Pum1 and homogenous RNA granules 30
An intriguing finding in this study is that Pum1-target Mad2 and Cyclin B1 mRNAs 31 formed distinct granules in the oocyte cytoplasm, instead of making granules containing 32 both mRNAs (Fig. 2). Pum1 was found to produce highly clustered structures that 33 surrounded both Mad2 and Cyclin B1 RNA granules (Fig. 3). These structures partially 1 resemble those of germ granules in Drosophila embryos, in which mRNAs form 2 homogenous RNA clusters and are spatially positioned within the granules, while RNA-3 binding proteins are evenly distribute throughout the granules (Trcek et al., 2015). 4 These findings suggest the existence of a common mechanism by which each mRNA 5 could be organized into homogenous particles. However, in contrast to our findings, the 6 structures of germ granules were not changed during early stages of embryogenesis and 7 were independent of the control of mRNA translation and degradation (Trcek et al., 8 2015). Therefore, the function of spacially organized structures of germ granules in 9 Drosophila embryos seems to be different from the function of subcellular structures of 10 Pum1 and RNA granules in mouse oocytes. 11 Our results showed that Pum1 aggregates surrounded and overlapped Mad2 and 12 Cyclin B1 RNA granules but were not localized at the center of granules ( into aggregates remain unknown. One possible model is that Pum1 binds to a target 20 mRNA via the PUF domain and subsequently assembles into aggregates via the Q/N-21 rich region. Another possibility is that Pum1 contains two populations; one population 22 binds to target mRNAs and the other functions as structual scaffolds without binding to 23 mRNAs. In addition to the homogenous assembly of Pum1, heterogenous assembly 24 with other RNA-binding proteins may produce aggregates. In any case, the resulting 25 Pum1 aggregates in clustered structures would make compartments that function as 26 regulatory units with related proteins assembled together or separately. These units 27 enable to coordinately regulate the translation of assembled mRNAs. Since Pum1 28 functions in diverse systems and other RNA-binding proteins that harbor prion-like 29 domains may function in a manner similar to that of Pum1, our results will contribute to 30 an understanding of the nature of temporal and spatial control of translation in many 31 cell types of diverse organisms. 32 33

Preparation of ovaries and oocytes 2
All animal experiments in this study were approved by the Committee on Animal 3 Experimentation, Hokkaido University. Mouse ovaries were dissected from 8-week-old 4 females in M2 medium (Sigma). Oocytes were retrieved from ovaries by puncturing the 5 ovaries with a needle in M2 medium containing 10 µM milrinone, which prevents 6 resumption of oocyte maturation. To induce oocyte maturation, the isolated oocytes 7 were washed three times and incubated with M2 medium without milrinone at 37˚C. 8 Alternatively, oocyte maturation was induced by injection of 5 U of hCG 48 h after 9 injection of 5 U of pregnant mare serum gonadotropin into 3-week-old females. For RT-10 PCR and poly(A) test (PAT) assays, ovaries and oocytes were extracted with Trizol 11 reagent (Invitrogen) and total RNA was used for RT-PCR and RNA ligation-coupled 12 RT-PCR. For in situ hybridization analysis, mouse ovaries were fixed with 4% 13 paraformaldehyde in PBS (137 mM NaCl, 2.7 mM KCL, 10 mM Na2HPO4, and 2 mM 14 1 mM dithiothreitol, 100 µM (p-amidinophenyl)methanesulfonyl fluoride, 3 µg/ml 20 leupeptin; pH 7.5) containing 1% Tween20 and 100 U/ml RNasin Plus RNase Inhibitor 21 (Promega). After centrifugation at 15,000 g for 10 min at 4˚C, the supernatant was 22 collected and used for IP. 23 Zebrafish ovaries were dissected from adult females in zebrafish Ringer's solution 24 (116 mM NaCl, 2.9 mM KCl, 1.8 mM CaCl2, and 5 mM HEPES; pH 7.2). Zebrafish 25 oocytes were manually isolated from ovaries with forceps under a dissecting 26 microscope. Oocyte maturation was induced by treatment with 1 µg/ml of 17a,20b-27 dihydroxy-4-pregnen-3-one, a maturation-inducing hormone in fish. For 28 ultracentrifugation analysis, fully grown immature oocytes and oocytes 3 h after MIH 29 stimulation (matured oocytes) were homogenized with an equal volume of ice-cold EB 30 containing 0.2% Tween20. After ultracentrifugation at 90,000 g for 30 min at 4˚C, the 31 supernatant and precipitates were collected and used for immunoblot analysis. 32

RT-PCR and quantitative PCR 1
Total RNA extracted from mouse ovaries or 50 immature oocytes was used for cDNA 2 synthesis using the Super Script III First Strand Synthesis System (Invitrogen). The full 3 length of Mad2 mRNA was amplified with the cDNA and primer sets specific to Mad2, 4 mMad2-f1 (5'-GTA GTG TTC TCC GTT CGA TCT AG-3') and mMad2-r1 (5'-GTA 5 TCA CTG ACT TTT AAA GCT TGA TTT TTA-3'). The amounts of short and long 6 Mad2 mRNAs were quantified by using a real-time PCR system with SYBR green PCR 7 Master Mix (Applied Biosystems) according to the manufacturer's instructions. The 8 short and long Mad2 transcripts were amplified with the cDNA and primer sets to both 9 types of Mad2, mMad2-f2 (5'-GAA TAG TAT GGT GGC CTA CAA-3') and mMad2-10 r2 (5'-TTC CCT CGT TTC AGG CAC CA-3'), and primer sets specific to long Mad2, 11 mMad2-f3 (5'-CTG GAC CAG GAT ATA AAG AAG CG-3') and mMad2-r3 (5'-GCT 12 GTC CTC CCT GCC TCT CT-3'). The signals obtained with distinct primer sets were 13 normalized by standard curves obtained with plasmid DNAs encoding the short or long 14 To analyze the effects of permeabilization on GFP-Pum1 aggregates, the oocytes 1 injected with mRNA encoding GFP or GFP-Pum1 were incubated for overnight at 37˚C 2 with M2 medium containing 10 µM milrinone. After observation under the LSM 5 3 LIVE confocal microscope, the oocytes were transferred to M2 medium containing 4 0.012% digitonin and 10 µM milrinone. The oocytes were then observed under the 5 confocal microscope at the appropriate time points. 6 To analyze the effects of GFP-Pum1∆C on oocyte maturation, the oocytes injected 7 with mRNA encoding GFP or GFP-Pum1∆C were incubated for 18 h at 37˚C with M2 8 medium and then fixed with 4% PFA/PBS for 1 h at 37˚C. The samples were 9 permeabilized with PBS containing 0.1% Triton-X100 for 20 min, followed by 10 incubation with a blocking/washing solution (PBS containing 0.3% BSA and 0.01% 11 Tween20) for 1 h at room temperature. The samples were then incubated with Cy3-12

Puromycin treatment and Pum1 antibody injection 31
To inhibit protein synthesis, oocytes were treated with 20 mM puromycin in M2 32 medium and incubated at 37˚C. The oocytes were collected at appropriate time points 33 after incubation with puromycin for immunoblotting analysis. Two pg of anti-Pum1 1 antibody was injected into fully grown mouse oocytes using the microinjector in M2 2 medium containing 10 µM milrinone. The oocytes were then washed three times and 3 incubated for 18 h at 37˚C with M2 medium containing 1 µM milrinone. To analyze the 4 distribution of GFP-Pum1, 10 pg of the GFP-Pum1 mRNA was co-injected with 2 pg of 5 anti-Pum1 antibody into fully grown mouse oocytes, followed by washing and 6 incubation of oocytes as described above. The distribution of GFP-Pum1 was observed 7 under the LSM 5 LIVE confocal microscope.

Conflict of interest 14
The authors declare that no competing interests exist.