Pum1 is a maternal effect gene required for normal preimplantation development
To investigate whether maternal PUM1 has a role during preimplantation development, we set up four mating schemes using Pum1+/+/Pum1−/− male and female mice (Fig. 1a) as follows: Pum1+/+ females with Pum1+/+ males (Cross I); Pum1−/− females with Pum1+/+ males (Cross II); Pum1−/− females with Pum1−/− males (Cross III); Pum1+/+ females with Pum1−/− males (Cross IV). The females were superovulated, mated with the appropriate males, and embryos were obtained at 1.5 days postcoitum (dpc) and 3.5 dpc. At 1.5 dpc, the number of unfertilized MII oocytes and embryos of different stages was recorded. The fertilization rate (% fertilized MII oocytes/total number of MII oocytes and embryos) observed were as follows: Cross I (+/+ x +/+) 81.3% fertilization; Cross II (−/− x +/+): 76% fertilization; Cross III (−/− x −/−): 43% fertilization and Cross IV (+/+ x −/−): 67.1% fertilization. These results indicate that Pum1−/− males are associated with a decrease in fertilization rate, consistent with our previous study that Pum1−/− males have defective spermatogenesis and comprised fertility [19].
To understand whether maternal PUM1 has a function during preimplantation development, we examined the progeny from the four crosses at 1.5 dpc and 3.5 dpc. Figure 1b, c shows the percentage and total numbers of embryos at differing developmental stages at 1.5 dpc and 3.5 dpc. Figure 1b shows that in Cross II and III, where there is a lack of maternal PUM1, there is a perturbation in embryogenesis, with significantly lower percentages of embryos at the two-cell stage in both crosses as compared to the positive control (Cross I). In Cross II, there was a significantly higher percentage of four-cell embryos than Cross I, as well as a greater percentage of fragmented embryos.
Similarly, in Cross III, most of the remaining embryos were fragmented. In Cross IV, where maternal PUM1 is present the proportion of two-cell embryos did not differ from Cross I. Moreover, in 3.5 dpc embryos from Cross II and III, the absence of maternal PUM1 continued to impede their development such that there was a significantly lower percentage of embryos that reached blastocyst stage as well as a higher percentage of fragmented embryos (Fig. 1c).
In contrast, a significantly greater proportion of embryos from Cross IV became blastocysts, and less were fragmented than embryos from Cross I. Taken together, the results show that maternal PUM1 is essential for normal preimplantation development. Moreover, the presence of zygotic PUM1 cannot rescue the delay in preimplantation development due to the lack of maternal PUM1. Therefore, PUM1 is a maternal effect gene.
Maternal PUM1 is dispensable for oocyte nuclear maturation
To investigate whether the maternal effect of PUM1 starts as early as during oocyte maturation, we examined oocyte maturation in Pum1−/− female mice. Prior studies have shown that a decrease in the mammalian Pum2 transcript can lead to a delay in meiotic progression of mouse oocytes [4], yet the role of PUM1 in meiotic progression is unknown. Therefore, to understand whether nuclear maturation is normally occurring in Pum1−/− oocytes, we performed a series of in vitro maturation experiments with Pum1+/+ and Pum1−/− germinal vesicle (GV) oocytes. 100% of Pum1+/+ and Pum1−/− GV oocytes underwent germinal vesicle breakdown (GVBD). 94.3% of Pum1+/+ and 86.9% Pum1−/− GV oocytes matured into MII oocytes, respectively (Fig. 2a). Furthermore, 68.8% of Pum1+/+ and 74.3% Pum1−/− MII oocytes extruded a polar body and no increase in degeneration of oocytes was observed. Also, GVBD time course was similar between Pum1+/+ and Pum1−/− oocytes (Fig. 2b). These similar outcomes between Pum1+/+ and Pum1−/− GV oocytes indicate that, surprisingly, Pum1−/− GV oocytes have no delay in meiotic maturation. Thus, PUM1 is not required for the process of nuclear maturation during meiotic maturation.
Maternal PUM1 affects global mRNA dynamics during oocyte maturation
Given that PUM1 depletion does not affect oocyte nuclear maturation yet has a strong maternal effect on preimplantation development, we wondered if maternal PUM1 affected cytoplasmic maturation, a process whereby the oocyte accumulates and stores the necessary RNA/protein to achieve developmental competence. One major pathway used by the oocyte to attain cytoplasmic maturation is selective mRNA degradation [3]. PUM1 could carry out selective mRNA degradation by directly binding to the PRE-consensus on the 3′UTR of target mRNAs. This binding could lead to the shortening of the poly-A tail by PUM1 associated deadenylase complexes and therefore lead to a decrease in mRNA stability. Furthermore, PUM proteins are known to repress translation [33]. However, due to the insufficient material (oocytes/preimplantation embryos), biochemical analysis of translation is not feasible. Therefore, in this study, we focused on the role of PUM1 in regulating target mRNA stability. Our analysis also allowed us to reveal the indirect effects of PUM1 on other mRNAs.
To test the hypothesis that PUM1 affects the mRNA degradation/stabilization during the oocyte-to-embryo transition, we carried out RNA-seq analysis during the period of transcriptional quiescence which occurs from GV oocytes to two-cell embryo stage, since any transcriptome changes seen would reflect the PUM1 regulation of mRNA stability.
To overcome the limited availability of viable oocytes obtainable from Pum1−/− females, we used a low input RNA-seq approach using oligo-dT primers to study the transcriptome changes during the GV to MII transition in both Pum1+/+(WT) and Pum1−/− (KO) oocytes. Three biological replicates for each condition were performed. All WT biological replicates showed high correlation (Fig. 3a) and therefore reflected the excellent reproducibility of our biological replicates.
First, we compared the transcriptome of WT and KO GV oocytes. Only 30 differentially expressed genes were found between WT and KO GV oocytes, with seven genes downregulated and 23 genes upregulated (Fig. 3b). This indicates that there were no significant differences in transcriptome between WT and KO oocytes at the GV stage. This result was expected because transcriptional control is the primary mechanism used by growing oocytes to regulate their development. Hence PUM1 is unlikely to affect the transcriptome at this stage.
Next, we investigated whether there were transcriptome differences between WT and KO MII oocytes. 350 genes were significantly differentially expressed by at least twofold in KO MII oocytes when compared to WT MII oocytes, with 248 upregulated and 102 downregulated (Fig. 3c). GO analysis revealed that the genes that are upregulated in KO MII oocytes are enriched for ATP biosynthesis and DNA synthesis. These processes are usually downregulated in MII oocytes as they transit to become metabolically quiescent during metaphase arrest (Fig. 3d). Interestingly, genes that are downregulated in KO MII oocytes are enriched for negative regulation of reactive oxygen species (Fig. 3d). This dysregulation of reactive oxygen species could partly explain an increase in degenerate MII oocytes in Pum1−/− females found in our prior studies [20]. Together, these data show that, at the MII stage, PUM1 gains a more substantial role in regulating mRNA stability during this period of transcriptional silence.
To estimate how many of the up- and downregulated transcripts are likely direct target of PUM1, we examined their 3′UTR for PRE. Only 23 of the 102 (22.3%) downregulated genes and 31 of the 248 (12.5%) upregulated genes had at least one predicted 3′UTR PRE-consensus sequence. This further indicates that most of the dysregulated mRNAs are due to the indirect effect of PUM1.
We next analyzed the mRNA dynamics during oocyte maturation by examining changes in transcriptome during the GV-to-MII transition in WT and KO oocytes. At the time of this study, preparing a library for RNA-seq using a low number of oocytes was only feasible using oligo dT-priming. Therefore, the interpretation of the following data could be confounded by differences in poly-A tail length. To validate our approach of using low input RNA seq to investigate mRNA dynamics during GV to MII transition, we compared a subset of our data on WT oocytes during the GV-to-MII transition with a microarray study performed by Su et al. [3]. These authors used multi-start primers instead of oligo d-T primers to minimize the effect of poly-A tail length on the analysis and they validated their findings using comprehensive qPCR analysis. They chose five groups of genes to validate with qPCR. We compared the fold change from our analysis with that of the microarray and qPCR validation (Additional file 1: Fig. S1A–D). One of the five groups were granulosa-specific genes Has2, Ptgs2, Ptx3, and Tnfaip6, which were either not detectable or had no significant changes across GV to MII transition. As expected, none of these genes were found in our WT dataset. Another group included oocyte-specific genes (Additional file 1: Fig. S1A), in the microarray study, Bmp15, Gdf9, H1foo, Mater, and Zar1. They showed no significant changes in oocyte maturation, but Fgf8 was found to be significantly degraded. Interestingly, their qPCR analysis showed significant degradation of all these genes. In our dataset, only Mater was found to be significantly degraded, and Fgf8 was not found in our dataset. Another group of genes were polyadenylated transcripts during oocyte maturation, Ccndbp1, Gd6pdx, Mos, Plat and Spin1 (Additional file 1: Fig. S1B). Su et al. [3] postulated that any method that was biased by poly-A tail length would potentially show a false increase in transcript level in MII such as transcripts which are polyadenylated. In the microarray study, none of the genes showed a significant increase in transcript in MII, and their subsequent qPCR analysis showed all transcripts were degraded. Hence, our data was overall consistent with the microarray, apart from Mos which showed a significant increase in transcript level. Su and colleagues also analyzed a group of transcripts that are known to be degraded in MII oocytes, in the microarray all transcripts were significantly lower in MII oocytes and confirmed by their qPCR analysis. Similarly, in our analysis, the same transcripts except Exosc8 and Polr2b were significantly lower in MII. In the microarray there was a group of transcripts which showed ‘upregulation’ in MII oocytes in the microarray study and these were shown to be artifacts of the microarray as their qPCR analysis showed both genes were significantly lower in MII. Taken together, our RNA seq data was consistent with the qPCR data performed by Su et al. (2007) in this group.
In summary, the above analyses verified that our approach is a valid methodology to study the transcriptome of the GV to MII transition and that most of our data are unlikely to be significantly impacted by poly-A tail length despite using oligo-dT primers.
We then analyzed our data for the mRNA dynamics during oocyte maturation. In WT oocyte, 60% (2945) of mRNAs were unchanged, 22% (1070) were significantly lower (> twofold) in MII oocytes than in GV oocytes which represent mRNA degraded, and 18% (868) were significantly higher (> twofold) in MII oocytes which represent transcripts which are stabilized (Fig. 3e). In PUM1-KO oocyte maturation, 73% (3467) of transcripts were unchanged, 15% (710) were degraded, and 12% (544) were stabilized (Fig. 3f). Thus, the absence of PUM1 led to an increase in the number of unchanged transcripts during oocyte maturation and fewer transcripts being degraded or stabilized. These global differences between the transcriptomes of WT and PUM1 mutant oocytes during oocyte maturation indicates that PUM1 has a function in regulating mRNA stability during oocyte maturation.
We then focused our analysis on the transcripts that were degraded during oocyte maturation to understand the biological pathways regulated by PUM1. We compared the transcripts degraded in WT and PUM1-KO oocytes during the GV-to-MII transition (Fig. 4a). We classified the degraded transcripts into three classes (Fig. 4b): Class 1 transcripts underwent normal degradation in KO oocytes during the GV-to-MII transition; Class 2 transcripts did not undergo normal degradation in KO oocytes during the transition; Class 3 transcripts were normally degraded in WT oocytes during maturation but not found in KO oocytes. There were no transcripts found to be normally degraded in WT but activated in KO. Of the 1070 transcripts found to be degraded in WT oocytes during the transition, 578 transcripts (54%) were also degraded in KO oocytes (Class 1). However, 439 transcripts (41%) were no longer degraded in KO oocytes (Class 2), and 53 transcripts (2%) were no longer detected in KO oocytes (Class 3; Fig. 4c). Thus, 41% of genes are not correctly degraded during KO oocyte maturation. This result indicates that PUM1 has a major function in regulating mRNA turnover during oocyte maturation.
GO analysis of the transcripts in Class 2 revealed pathways enriched for translation, protein synthesis, etc. (Fig. 4d). This makes sense because the oocyte should become quiescent during normal oocyte maturation, so processes such as translation and protein synthesis should be winding down. Our results show that lack of maternal PUM1 could dysregulate this developmental programming towards quiescence.
Maternal PUM1 appears to directly target only a small number of mRNAs during oocyte maturation
To further distinguish whether the dysregulation of mRNA degradation during the maturation of the KO oocytes was a direct and/or indirect consequence of the absence of PUM1, we identified bioinformatically transcripts with putative 3′UTR PRE motifs. Of the 1070 transcripts that are degraded in WT oocytes during maturation, only 161 transcripts (15%) had one or more predicted PRE motif, but 909 (84.9%) of transcripts did not have a PRE. Among the 161 PRE-containing transcripts, 79 of these were Class 1, 72 transcripts were Class 2, ten transcripts were Class 3 (Fig. 4e). Among the 909 transcripts without PRE, 498 transcripts were Class 1, 366 transcripts were Class 2, and 45 transcripts were Class 3 (Fig. 4f). These results indicate that PUM1 may directly degrade a small number of mRNAs, which then lead to an indirect effect on the abundance of many other mRNAs as detected in our analysis.
We then examined the role of PUM1 in transcripts found to be at higher levels in both WT and KO oocytes. Less overlap between WT and KO oocytes were seen in this cohort of transcripts (Fig. 5a). We similarly classified these transcripts into Classes 4–7 (Fig. 5b). Class 4 transcripts are equally stabilized in MII vs. GV in both WT and KO; Class 5 transcripts are dysregulated in the KO oocyte maturation; Class 6 transcripts are stabilized in WT but not in present in the KO dataset; Class 7 transcripts are stabilized in WT but degraded in KO. In Class 4 there were 328 transcripts, 409 transcripts in Class 5, 118 transcripts in Class 6 and 13 transcripts in Class 7 (Fig. 5c). Of note, there was a surprisingly greater number of genes dysregulated in the KO than that of the degraded genes. This suggests a more prominent role of PUM1 in the stabilization of transcript rather than degradation. GO analysis of Class 5 transcripts showed that there was an enrichment for genes involved in mRNA processing, cell division, mRNA metabolic processes (Fig. 5d). This result indicates that the oocyte is stockpiling selective RNAs which will be required during early embryogenesis and that PUM1 has a significant role in regulating this process.
We performed a similar bioinformatic analysis looking at the stabilized transcripts with regards to the presence or absence of PRE. Interestingly, there was a higher number of stabilized transcripts with PRE (289 transcripts 33%) than that in degraded transcripts (15%). 579 (77%) stabilized transcripts had no PRE. Of the 289 transcripts with at least one PRE, 116 transcripts were Class 4, 129 were Class 5, 42 were Class 6 and 2 were Class 7 (Fig. 5e). Of the 579 transcripts with no PRE, there were 212 transcripts in Class 4, 272 in Class 5, 84 in Class 6 and 11 in Class 7 (Fig. 4f). Again, there was no evident enrichment for transcripts with or without PRE being more perturbed by the absence of PUM1.
In summary, our data indicate that maternal PUM1 does have an impact on the global transcriptome during oocyte maturation. PUM1 regulates genes involved in processes that are vital for developmental competence but not meiotic genes or those involved in nuclear maturation. It is still possible, however, that PUM1 regulates the translation of many mRNAs during oocyte maturation by translational regulation, which will not be detected by our analysis.
Maternal PUM1 affects global mRNA dynamics during early preimplantation development
If maternal PUM1 regulates the stability of maternal mRNAs, we would predict that the pool of mRNAs present from fertilization to maternal–zygotic activation would be affected by the absence of PUM1. Therefore, we performed RNA seq analysis of m+z+ two-cell embryos from Cross I (Pum1+/+ self-mating), m−z+ two-cell embryos from Cross II (Pum1−/− females mating with Pum1+/+ males) and m−z− two-cell embryos from Cross III (Pum1−/− self-matings) (Fig. 1a). All samples were sequenced in triplicate, which showed good reproducibility of the WT two-cell replicates (Additional file 2: Fig. S2A). The transcriptomes of WT MII oocytes were very different from those of WT two-cell embryos. These data reveal a significant change in transcriptome between MII oocytes and two-cell embryos.
We then compared the transcriptome difference among the three types of two-cell embryos to investigate the function of maternal PUM1 in regulating the transcript dynamics in two-cell embryos. Zygotic PUM1 is not maximally expressed until four-cell stage. Therefore any transcriptome differences prior to this are likely to be mostly under maternal PUM1 control. There were 331 downregulated transcripts and 316 upregulated transcripts when comparing m+z+ (WT) and m−z+ (HET) two-cell embryos (Additional file 3: Fig. S3B). There were 199 downregulated/transcripts and 237 upregulated/transcripts in m−z− (KO) compared to m+z+ (WT) two-cell embryos (Additional file 2: Fig. S2C). Interestingly, there were less differentially expressed genes when comparing m−z+ (HET) two-cell and m−z− (KO) two-cell transcriptomes, with only 120 transcripts downregulated and 115 upregulated transcripts (Additional file 3: Fig. S3D). Because m−z+ (HET) two-cell and m−z− (KO) two-cell embryos were more similar in their transcriptome than with m+z+ (WT) two-cell embryos. These results indicated that maternal PUM1 has a significant influence on the RNA stability even after fertilization.
To explore the mRNA dynamics during MII to two-cell transition which would be likely to be regulated by maternal PUM1, we compared the transcriptome of WT MII oocytes to m+z+ two-cell embryos, KO MII to m−z+ two-cell embryos, and KO MII to m−z− two-cell embryos. Similar to the oocyte data, lacking maternal PUM1 lead to fewer transcripts being stabilized and degraded (Fig. 6a) and more transcripts remaining unchanged across the developmental transition. Therefore, PUM1’s function in regulating the mRNA pool is consistent from GV to two-cell transition.
To isolate the transcripts that are controlled by maternal PUM1, identification of genes inappropriately regulated during transition from KO MII oocytes to m−z+ two-cell embryos was performed. 48.8% (793) of transcripts normally degraded during the WT MII-to-m+z+ transition were also found to be similarly degraded in the KO MII-to m−z+ transition (Fig. 6b). However, 40.5% (540) of transcripts were not appropriately degraded during the MII-to-2-cell transition in m−z+ (Fig. 6b). GO analysis of the transcripts that were not appropriately degraded in m−z+ two-cells showed enrichment in genes involved in cell division and cell cycle (Fig. 6d). This result indicates that PUM1 is fine-tuning genes involved in development and interestingly more of m−z+ embryos had accelerated development to four-cell embryos (Fig. 1b).
We then examined how many mRNAs that are normally stabilized during the MII-two-cell embryo transition and how many of such transcripts are affected when maternal PUM1 is depleted. 41.6% (717) of transcripts normally stabilized during the WT MII-to-m+z+ two-cell transition were similarly regulated in the KO MII-to-m−z+ two-cell transition (Fig. 6c). However, 38.1% (657) of transcripts that were stabilized during transition from WT MII oocytes to m+z+ two-cell embryos are no longer stabilized during the KO MII oocytes to m−z+ two-cell transition (Fig. 6c). GO analysis of these misregulated transcripts revealed enrichment for genes with a function in ATP synthesis and regulation of translation initiation (Fig. 6e). This analysis indicates that maternal PUM1 regulates the stability of transcripts that are important in the conversion of a metabolically quiescent oocyte to a metabolically active two-cell embryo.
In summary, our results show that PUM1 positively and negatively regulates the stability of different maternal mRNAs during early embryogenesis but to a less extent than during oocyte maturation.
Maternal PUM1 regulates Cdk1 during the oocyte-embryo transition
Next, we focused our analysis on genes that could be direct targets of PUM1 and could contribute to the observed phenotype. Additional file 4: Table S1 shows the list of PRE-containing mRNAs that are differentially expressed between WT and KO MII oocytes. Cyclin-dependent kinase 1 (Cdk1) was the most overexpressed candidate gene in KO MII. Cdk1 mRNA contains one PRE-element in the 3′UTR and was present at much higher levels in KO MII oocytes. Cdk1 is a member of the Ser/Threonine kinase family and has been shown to be an essential regulator of meiotic resumption in mouse GV oocytes [34] and also required during preimplantation development [35]. During the WT oocyte maturation, the Cdk1 transcript levels were significantly lower in MII oocytes than in GV (log2(− 6.3) fold change), therefore showing that Cdk1 transcripts are normally degraded during oocyte maturation. During KO oocyte maturation, the Cdk1 transcript was decreased but only by log2(− 1.59) in MII. These data indicate that the absence of PUM1 leads to reduced degradation of Cdk1 transcript. To examine the increase of CDK1 at the protein level, we performed immunostaining of Pum1+/+ and Pum1−/− MII oocytes with anti-CDK1 antibody (Fig. 7a, b), which showed that Pum1−/− MII oocytes have a significant increase (20%) in CDK1 protein levels compared to control oocytes.
Interestingly, Cdk1 is a major ZGA gene, and Cdk1 null mutant embryos fail to develop into morula and blastocyst stages [35]. Given that these crucial events are tightly regulated, we hypothesized that abnormal Cdk1 mRNA dynamics during the oocyte-embryo transition could be detrimental to preimplantation development. Therefore, we investigated the dynamics of Cdk1 from MII to two-cell stage. There was a significantly higher level of Cdk1 transcript in m+z+ 2 cell embryos as compared to WT MII (log2(7.82)) oocytes. This is expected given it is a major ZGA transcript, whereas comparing KO MII to m−z+ and m−z− two-cell embryos, there was a smaller increase (log2 (3.01) and log2 (3.02) respectively). The levels of Cdk1 transcript was not different between m+z+, m−z+ and m−z− two-cell embryos. Therefore, the less prominent increase in Cdk1 transcript in the embryos lacking maternal PUM1 is likely due to suboptimal degradation of Cdk1 transcript in KO MII oocytes. This effect is then sustained through early preimplantation development. Hence, we speculate that the different Cdk1 mRNA dynamics during the oocyte–embryo transition in the absence of maternal PUM1 could possibly explain the observed abnormal preimplantation development.
Both maternal and zygotic PUM1 are essential for postnatal survival
Having shown that maternal PUM1 functions during early preimplantation development, we further investigated whether PUM1 is required after this developmental stage. We collected data on litter sizes and survival in Cross I, II, and III as shown in Additional file 3: Fig. S3. Interestingly, the absence of maternal PUM1 partially affects the postnatal survival of progeny (Cross II). However, the deficiency of both maternal and zygotic PUM1 lead to 100% postnatal lethality of progeny (Cross III). This suggests that both maternal and zygotic PUM1 are both required for postnatal survival.