Age‐related differences in the translational landscape of mammalian oocytes

Abstract Increasing maternal age in mammals is associated with poorer oocyte quality, involving higher aneuploidy rates and decreased developmental competence. Prior to resumption of meiosis, fully developed mammalian oocytes become transcriptionally silent until the onset of zygotic genome activation. Therefore, meiotic progression and early embryogenesis are driven largely by translational utilization of previously synthesized mRNAs. We report that genome‐wide translatome profiling reveals considerable numbers of transcripts that are differentially translated in oocytes obtained from aged compared to young females. Additionally, we show that a number of aberrantly translated mRNAs in oocytes from aged females are associated with cell cycle. Indeed, we demonstrate that four specific maternal age‐related transcripts (Sgk1, Castor1, Aire and Eg5) with differential translation rates encode factors that are associated with the newly forming meiotic spindle. Moreover, we report substantial defects in chromosome alignment and cytokinesis in the oocytes of young females, in which candidate CASTOR1 and SGK1 protein levels or activity are experimentally altered. Our findings indicate that improper translation of specific proteins at the onset of meiosis contributes to increased chromosome segregation problems associated with female ageing.


| INTRODUC TI ON
The quality of oocytes (female germ cells) is an essential factor for successful sexual reproduction. Mammalian oocyte development is a complex and long process beginning during embryogenesis and then arresting in the first meiotic prophase. It is not until puberty that oocytes can finally be recruited for ovulation, by reinitiating meiosis and culminating in eventual fertilization. It is accepted that chromosomal aneuploidy, a consequence of chromosomal segregation errors during meiosis, is one of the most common causes of poor oocyte quality, leading to embryo lethality or severe developmental disabilities (Savva, Walker, & Morris, 2010). Importantly, it has been reported that most chromosomal segregation aberrations take place during the first meiotic division, occurring after meiotic resumption and Nuclear Envelope Break Down (NEBD) (Hassold & Hunt, 2001).
Precisely, it is shortly after NEBD when chromosome condensation and assembly take place at the newly forming meiotic spindle (Schuh & Ellenberg, 2007).
Chromosomal aneuploidy in human oocytes is not a rare event, exemplified by a 20% incidence in 32-year-old women that further increases with age, reaching 60%-80% in women aged 42 (Capalbo, Hoffmann, Cimadomo, Maria Ubaldi, & Rienzi, 2017;Jones & Lane, 2013;Kuliev, Zlatopolsky, Kirillova, Spivakova, & Cieslak Janzen, 2011). This so called "age-related aneuploidy" in oocytes has long been a focus in the study of human reproduction, as women in economically advanced societies are trending to delay their pregnancies (Molina-García et al., 2019). However, we are still far from fully understanding the mechanisms underlying this phenomenon.
In order to provide enhanced insight into the dysfunctional molecular mechanisms behind age-related aneuploidy, the transcriptomes of oocytes from young and advanced aged females have been compared, using various animal models (Labrecque & Sirard, 2014), of which mouse has proven to be a suitable one (Jones & Lane, 2013b;Sebestova, Danylevska, Dobrucka, Kubelka, & Anger, 2012). These studies have revealed that there are actual significant differences in the transcriptomes of oocytes from aged females, mainly associated with decreased levels of gene transcripts involved in managing oxidative stress, chromatin structure, genome stability and spindle structure. Nonetheless, another crucial feature of oocyte development must be taken into account.
Namely, that oocytes proceed through meiotic maturation and to the post-fertilization zygote stage in the absence of de novo transcription (De La Fuente et al., 2004;. Accordingly, growing oocytes that become arrested at prophase I synthesize large amounts of RNA that is stored for future use during the transcriptionally silent period after meiotic resumption (Fulka, First, & Moor, 1998). Precisely, chromosome condensation and segregation take place in an environment where regulation of translation is a paramount gene expression control mechanism. As such, conventional transcriptome analysis, although undoubtedly useful, might not offer a sufficiently accurate picture of oocyte gene expression on the protein level, as the whole transcriptome includes large pools of stored RNAs not undergoing active translation. To overcome this limitation, polysomal fractionation of oocyte samples can be performed in order to isolate those RNAs which are bound to ribosomes and hence are more likely to be translated into proteins with functional roles during specific stages of meiosis. However, mammalian oocytes are a scarce source of material, which is a big limiting factor to such classical fractionation techniques that typically require millions of cells (Mašek, Valášek, & Pospíšek, 2011). Thus, unsurprisingly, only a few experiments of this kind have been reported (Chen et al., 2011;Potireddy, Vassena, Patel, & Latham, 2006;Scantland et al., 2011). For this purpose, we recently developed a Scarce Sample Polysomal-profiling (SSP-profiling) method, which allows the obtaining of polysomal RNAs with high reliability from as little as 200 oocytes .
In this study, we present what to our knowledge is the first thorough analysis of ribosome-bound mRNAs in mouse oocytes at the post-NEBD stage within the context of maternal age. Moreover, we validate several candidate mRNAs exhibiting altered translation in the oocytes of more advanced age females and identify novel gene transcripts and proteins related to mammalian oocyte development and aneuploidy.
2 | RE SULTS 2.1 | Genome-wide analysis reveals differential translation of cell cycle players in oocytes from aged females Most oocyte meiotic errors from aged females occur in meiosis I (Hassold & Hunt, 2001). To test the effect of maternal age on oocyte quality, we performed cold shock treatment on oocytes from young (2 months old, YF) and aged (12 months old, AF) female mice followed by assessment of spindle morphology and chromosome alignment. As expected, after a short time of cold culture, YF oocytes showed organized spindles and chromosomes with abundant signal for kinetochores while most spindles in AF oocytes were partially or totally disassembled ( Figure S1). An observation in agreement with the findings of others (Pan, Ma, Zhu, & Schultz, 2008;Sebestova et al., 2012). Taking into account the fact fully grown oocytes are transcriptionally silent and that the transcriptome of GV oocytes does not significantly change with maternal age (Pan et al., 2008), we sought to analyse oocyte translation in relation to maternal age. Interestingly, we have previously analysed global translational rates between different age groups but we did not find any significant differences . Therefore, we hypothesized that the quality decrease in AF oocytes could be due to changes in the translation of specific transcripts rather than altered global translational levels.
Consequently, to provide a detailed mechanistic insight into the increased aneuploidy rates of AF oocytes, we compared those polysome-bound mRNAs of oocytes from young and aged females by Next Generation Sequencing. The translatome comparison was carried out on oocytes which underwent NEBD (3 h after 3-Isobutyl-1-methylxanthine (IBMX) removal); the stage at which chromosomes condense, the new meiotic spindle is forming and transcription is absent. Four individual replicates were used for YF samples and three for AF (200 post-NEBD oocytes/sample). Using the SSP-profiling method  combined with RNA-Seq, we analysed mRNA content in polysomal and non-polysomal fractions from oocytes of the two different female groups.
Analyses of the YF-and AF-derived fractions did not reveal any significant differences between the two age groups regarding polysomal distribution (Figure 1a). Significant decreases of ribosomal subunits in polysomal fractions were induced by pre-treatment with Ethylenediaminetetraacetic acid (EDTA), compared to both the YF and AF samples ( Figure 1a). As EDTA sequesters magnesium ions to disrupt ribosomal subunit assembly (Scantland et al., 2011), thus reducing polysome levels, this important technical control provides confidence that the RNAs in polysomal fractions of untreated YF and AF oocytes are indeed bound to ribosomes.
Next, for each sample, we pooled the collected fractions into either non-polysomal (NP, fractions containing free RNAs or RNAs bound to monosomes) or polysomal fractions (P; fractions containing polysome-bound RNAs). This was followed by ribodepleted RNA-Seq of NP and P fractions to detect genome-wide translational differences between the two age groups.

F I G U R E 1
Genome-wide analysis shows differential translation of cell cycle regulators in oocytes from aged females. (a) qRT-PCR analysis of the distribution of 18S rRNA between non-polysomal (NP) and polysomal (P) fractions in oocytes of young female (YF), aged female (AF) mice and in oocytes with disrupted polysomes (YF + EDTA). Data are represented as the mean ± SEM, *p < 0.05; ns, nonsignificant; according to Student's t test. (b) Volcano plot displaying gene transcripts differentially enriched in P fractions of oocytes from YF and AF groups; highlighting those with FC > 10 and adjusted-p < 0.1 (red), with only FC > 10 (green) and the rest (grey). Data points with gene names refer to the selected target genes with possible function in female meiosis. See also Based on the RNA-Seq data output, we were able to identify differentially translated mRNAs ( Figure 1b). Analysis of P fractions with transcripts with at least bigger than 1 Fragments per Kilobase of exon model per Million reads mapped (FPKM) in one group revealed 3589 (35.06%) mRNAs with more than a five-fold comparative difference in enrichment between YF and AF oocytes (FC > 5) and 1006 (9.83%) mRNAs with an enrichment fold change above 10 (FC > 10) ( Figure 1b and Table S1). From those mRNAs with FC > 10, 623 mRNAs were more abundant in the P fractions of YF oocytes and 383 in P fractions of AF samples. In addition, 37 polysomal differentially enriched mRNAs were statistically significant between the two groups, as calculated by DEseq2 (Love, Huber, & Anders, 2014) ( Figure 1b and Table S1). We performed Over-Representation Analysis (ORA) on the polysomal and highly differentially enriched transcripts exhibiting FC > 10 between the two oocyte groups, thus allowing more meaningful results as they conform a larger gene set ( Figure 1c and  Figure S2 and Table S3).
Clusters 2, 5, 6 and 7 appear to comprise gene transcripts with differential translation regulation between the YF and AF groups, as despite having similar abundancies in non-polysomal (NP) fractions they exhibit opposing enrichment trends in the corresponding polysomal (P) fractions.
In summary, we used genome-wide analysis to generate a comprehensive dataset of specific mRNAs with differential polysomal abundancy in oocytes from different maternal age-related oocytes that resumed meiosis.

| Specific genes transcripts with different polysomal occupancy between YF and AF oocytes positively correlate with protein expression
To identify candidate genes potentially functionally involved in agerelated aneuploidy, we selected 10 gene transcripts with different polysomal occupancy between YF and AF datasets (Deseq2, p < 0.1 and/or FC > 10). We primarily focused on transcripts that  Table S4 for the list of antibodies. (c) Western Blot quantification of protein expression for selected genes. Values obtained were normalized to ACTB and YF group was set as reference (100%). Data are represented as the mean ± SEM of at least three independent experiments; N = 30 oocytes per sample; *p < 0.05, according to Student's t test. See also Table S4 for the list of antibodies * * * * favouring the AF group), we were not able to detect significant differences by ddPCR between the two groups. Moreover, we did speculate whether the levels of these transcripts would vary not only in the polysomal fractions but throughout the whole transcriptome as a potential consequence of an age-related differential transcription during the previous oocyte growth period (rather than a differential translation at post-NEBD stage). We therefore extracted total RNA  Table S4 for the list of antibodies. (b) Immunocytochemistry quantification of protein fluorescence in oocytes from YF (n ≥ 48) and AF (n ≥ 23) oocytes in at least three independent experiments. Data are presented as the percentage of oocytes with a fluorescence intensity higher than the average in each experiment; *p < 0.05, ***p < 0.001; according to Fisher's exact test, error bars represent 95% confidence intervals by the adjusted Wald method better understand their role in meiosis resumption, we detected localization of selected differentially translated genes from YF and AF oocytes by immunocytochemistry (ICC). Interestingly, we found all four candidate proteins localized in the area of the newly forming spindle but in distinct patterns along or near ß-tubulin (Figure 3a and Figure S4a). SGK1 and EG5 were found in patches at the multipolar spindle, and CASTOR1 and AIRE were both localized in a similar pattern to newly forming spindle (Figure 3a and Figure S4a). This Previously, we reported that the timing of meiotic progression of AF-derived oocytes is faster than in their younger counterparts by about 30 min . Therefore, we asked whether such meiotic progression timing differences may promote translational differences between age groups. We analysed the expression of AIRE and EG5 proteins (by WB and ICC, respectively) in YF oocytes collected 30 min later than the conventional assay time (3 h post IBMX removal) for YF and AF oocytes. We found that additional cultivation did not significantly alter the protein expression to levels previously detected in the AF group ( Figure S5). This indicates that the observed translational differences are inherent to translation of selected transcripts in AF oocytes and not due to timing differences.
Collectively, these results confirm the spindle localization of selected candidate proteins and that their differential expression in AF-derived oocytes (versus YF counterparts) may play a role in spindle assembly, chromosome alignment and cytokinesis.

| Perturbation of CASTOR1 and SGK1 protein levels/activity leads to meiotic abnormalities
The sum of the described results indicates that at the post-NEBD stage, there are a number of mRNAs that exhibit differential polysomal occupancy and consequently altered protein expression levels in AF oocytes compared to YF derived counterparts. We decided to experimentally perturb the protein levels of two selected candidates with opposing differential polysomal occupancy in AF oocytes (CASTOR1 and SGK1, Figures 1b and 3b). As CASTOR1 protein has higher expression levels in AF oocytes, we performed overexpression of the protein in YF oocytes, by microinjection of  Figure S6). Subsequently, we assayed meiotic progression of YF oocytes treated with vehicle or SGK1 inhibitor.
Additionally, we also down-regulated Sgk1 mRNA ( Figure S7) by double-stranded RNA (Ds-Sgk1) microinjection prior to in vitro oocyte maturation. MII oocytes with down-regulated Sgk1 mRNA levels had increased frequencies of chromosome misalignment or cytokinetic defects compared to control oocytes (injected with only H2b:gfp RNA; Figure 4f,g). The effects of Ds-Sgk1 were less pronounced than those obtained after inhibitor treatment, however, with positive correlation.
In summary, by perturbing YF oocytes to mimic the protein expression or activity of CASTOR1 and SGK1 of AF oocytes, we have shown that these proteins have roles involved in meiosis I chromosome alignment and cytokinesis, which are intrinsically impaired in AF-derived oocytes.

| DISCUSS ION
Although several models have been proposed to explain how the age-associated increase in aneuploidy could occur, further investigation is needed, considering the medical relevance of maternal age-related aneuploidy. Previous results (Jones & Lane, 2013b;Sebestova et al., 2012) have shown that the mouse is a suitable model system to assess the molecular basis for human age-associated increase in aneuploidy. A general view is that the quality of the cytoplasm, which includes the transcriptome, is compromised in aged oocytes.
However, Pan et al. (2008) have shown that perturbations in the global transcriptome of GV oocytes from females of advanced age are actually minimal.
In this study, we have tackled the old problem of oocyte age-re- Research in this study was performed using oocytes at the post-NEBD stage. This specific meiotic time-point was deliberately selected, since the process of first meiotic division is being initiated: the meiotic arrest at prophase I is released, chromosomes have recently condensed, the new spindle has begun to assemble (Schuh & Ellenberg, 2007) and translational reprogramming is occurring Susor et al., 2015). Additionally and importantly, chromosomal mis-segregation is reported to preferentially occur during the first meiotic division. Therefore, uncovering translational differences at this stage between YF and AF oocytes harbours a high potential to better understand age-related aneuploidy (Ottolini et al., 2015). Earlier studies dealing with whole oocyte transcriptome analysis in correlation with maternal age have focused mainly on the GV and/or MII stages as they represent the two meiotic arrest points. For example, Pan et al. (2008) performed microarray analysis in order to obtain the whole transcriptome of oocytes from young and aged mice females at the GV and MII stages, and reported very little differences in the transcriptome between the age groups at the GV stage (5% with FC > 2) while the transcriptome of MII stage oocytes differed more significantly and correlated with maternal age (33% with FC > 2). A similar trend was observed by Reyes et al. (2017) in a transcriptome study comparing human GV and MII oocytes. However, the transcriptome at the post-NEBD stage remains unstudied, possibly because whole transcriptome analysis would reveal few differences between the oocytes from different maternal ages, as was previously reported in GV oocytes (Pan et al., 2008). Here, with the advantage of being able to reliably assay polysome-bound mRNAs, we have been able to detect mRNAs that are actively translated. Thus, we have begun to unveil how the specific oocyte translatome changes with maternal age and Intriguingly, the respective proteins of all transcripts localize to the newly forming spindle, which is essential for chromosome alignment and cytokinesis. EG5 is involved in spindle bi-polarity establishment, microtubule sliding and has been thoroughly studied in mitosis and meiosis (Mann & Wadsworth, 2019). Despite previous whole transcriptome studies failing to report overt differences in Eg5 mRNA expression between YF and AF oocytes (Grøndahl et al., 2010;Hamatani, Falco, et al., 2004;Pan et al., 2008;Reyes et al., 2017), our SSP-Profiling approach demonstrates a significant increase in Eg5 transcript in the polysomes of AF oocytes that then translates to corresponding increased protein expression levels. Interestingly, Liu et al., (2010) and Castillo, Morse, Godfrey, Naeem, and Justice (2007) have shown in other tissue models that overexpression of EG5 can induce aneuploidy and tumorigenesis.
AIRE is another interesting candidate with a possible role in age-related aneuploidy in mice. Although it has been mainly studied for its role as a transcription factor in immune tolerance control (Björses, Aaltonen, Horelli-Kuitunen, Yaspo, & Peltonen, 1998), it is known Aire-deficient female mice display infertility albeit without a direct study of oocyte quality (Jasti et al., 2012). Furthermore, an interaction between AIRE and spindle-associated proteins has been reported to be essential for mitotic spindle assembly in stem cells. In the same study, overexpression of a truncated version of AIRE leads to defective spindles in stem cells (Gu, Lambert, Cockburn, Gingras, & Rossant, 2017). Here, we also find AIRE is localized at the newly forming oocyte meiotic spindle and that Aire mRNA is highly translated in aged mice post-NEBD oocytes compared to their young counterpart. Due to AIRE's main studied role as a transcription factor, the presence of AIRE protein in transcriptionally silent oocytes is not entirely clear. However, as our data demonstrate further research could explain its function in spindle assembly and cytokinesis, with particular emphasis on oocytes from aged females.
The third protein we have identified and validated as up-regulated in the AF oocytes is CASTOR1. In the presence of arginine, CASTOR1 binds to this amino acid, which prevents its inhibitory effect on GATOR2, ultimately allowing mTOR1 activation (Saxton, Chantranupong, Knockenhauer, Schwartz, & Sabatini, 2016).
CASTOR1 localization to the spindle might be connected to mTOR1, which also accumulates to the spindle and its inhibition/down-regulation leads to meiotic defects (Guo et al., 2018;Susor et al., 2015).
Therefore, it is possible that localization of mTOR1 to the assembling spindle, under the control of a strong upstream regulator such as CASTOR1, might be important for its temporal and/or spatial regulation. Indeed, when we experimentally increased CASTOR1 levels in YF oocytes, in order to mimic the increased protein expression as detected in AF oocytes, the most frequently observed anomaly was chromosomal misalignment but also cytokinesis errors, resulting in oocytes devoid of DNA. We note, however, that the increased CASTOR1 levels are several fold higher than those seen in AF oocytes, because it is technically very difficult to achieve levels closer to those seen in AF oocytes. While the latter phenotype is obviously incompatible with further development, chromosome misalignment may lead to aneuploidy and is frequently observed in oocytes from aged females, both mouse and human (Liu & Keefe, 2008;Van den Berg et al., 2011erg et al., 2011Sebestova et al., 2012). Taken together, these results reveal CASTOR1 as a possible factor relevant to age-related chromosomal aneuploidy in oocytes.
In contrast to CASTOR1, SGK1 protein expression is lower in AF oocytes compared to YF oocytes. SGK1 is known to regulate the activity of ion channels, solute carriers, enzymes and transcription factors (Lang, Stournaras, Zacharopoulou, Voelkl, & Alesutan, 2018). SGK1 protein kinase activity can be triggered by insulin, follicle stimulation hormone (FSH), corticosterone and thrombin, and is mediated by several upstream kinases including PI3K, PDK1 and mTORC2 (Kobayashi & Cohen, 1999;Pearce, Sommer, Sakamoto, Wullschleger, & Alessi, 2011). Surprisingly, despite its previously described localization in the plasma membrane, we observed it in the region of the post-NEBD forming oocyte spindle. Paralleling our results, SGK1 levels have also been reported to decline with human ageing (Harries et al., 2012). Moreover, we found that inhibition of SGK1 activity and down-regulation of Sgk1 mRNA in YF oocytes visibly leads to cytokinetic errors. Such errors are common in AF-derived oocytes as evidenced by microtubule defects (Nakagawa & FitzHarris, 2017;Volarcik et al., 1998;Yun, Lane, & Jones, 2014).
Altogether, our study has generated a genome-wide database of mRNA transcripts that occupy post-NEBD oocyte polysomes in both young and aged mouse oocytes. Furthermore, we have demonstrated the existence of differential polysomal mRNA occupancy in young and aged oocytes, suggesting a different translational program in post-NEBD oocytes that is dependent on maternal age and correlates with reduced quality of aged oocytes (see scheme in

| Cell lysis and SSP-profiling
All steps for this protocol were detailed in Masek et al. (2020)

| Western blot
Samples of 20-35 oocytes were lysed in 10 µl of 1xReducing SDS  Table S4 and showed target proteins, Figure S8)

| Statistical analysis
The tools used to determine if the differences between groups were statistically significant were either the Student's t test on the averages and SEM or Fisher exact test with a 95% confidence interval using adjusted Wald method; *p < 0.05 considered as statistically significant (labelled with a star), ***p < 0.005 was considered as highly statistically significant (labelled with three stars).

ACK N OWLED G M ENTS
We would like to thank Jaroslava Supolikova and Marketa Hancova for their valuable assistance with many experiments and Martin Anger for kindly providing the H2B:gfp plasmid. We also acknowl- Infrastructures.

CO N FLI C T O F I NTE R E S T
The authors certify that they have NO conflict of interest of any kind in the subject matter or materials discussed in this manuscript/ article.