Physiologically relevant miRNAs in mammalian oocytes are rare and highly abundant.

miRNAs, ~22nt small RNAs associated with Argonaute (AGO) proteins, are important negative regulators of gene expression in mammalian cells. However, mammalian maternal miRNAs show negligible repressive activity and the miRNA pathway is dispensable for oocytes and maternal-to-zygotic transition. The stoichiometric hypothesis proposed that this is caused by dilution of maternal miRNAs during oocyte growth. As the dilution affects miRNAs but not mRNAs, it creates unfavorable miRNA:mRNA stoichiometry for efficient repression of cognate mRNAs. Here, we report that porcine ssc-miR-205 and bovine bta-miR-10b are exceptional miRNAs, which resist the diluting effect of oocyte growth and can efficiently suppress gene expression. Additional analysis of ssc-miR-205 shows that it has higher stability, reduces expression of endogenous targets, and contributes to the porcine oocyte-to-embryo transition. Consistent with the stoichiometric hypothesis, our results show that the endogenous miRNA pathway in mammalian oocytes is intact and that maternal miRNAs can efficiently suppress gene expression when a favorable miRNA:mRNA stoichiometry is established.

Introduction

MicroRNAs (miRNAs, reviewed in detail in Bartel (2018)) are genome‐encoded ~22‐nt‐long single‐strand small RNAs, which guide post‐transcriptional repression of gene expression. A typical mammalian miRNA biogenesis (reviewed in Kim et al (2009)) involves nuclear processing of primary miRNA transcripts (pri‐miRNA) into small hairpin precursors (pre‐miRNA), which are transported to the cytoplasm. A cytoplasmic pre‐miRNA is cleaved by RNase III Dicer, and a miRNA is loaded onto an Argonaute protein, the key protein component of the effector complex named RISC (RNA‐induced silencing complex). Mammalian genomes encode four AGO proteins, which accommodate miRNAs equally well (Meister et al, 2004) and appear to be functionally redundant in the miRNA pathway (Su et al, 2009). However, AGO2 stands out among the mammalian AGO proteins, as it carries endonucleolytic activity cleaving cognate RNAs perfectly complementary to a loaded small RNA (Liu et al, 2004; Meister et al, 2004). AGO2 with a small RNA thus forms the minimal RISC (holo‐RISC), which traditionally has been associated with the RNA interference pathway (Jinek & Doudna, 2009). While miRNAs guide RNAi‐like endonucleolytic cleavage of perfectly complementary mRNAs by AGO2 as well (Hutvagner & Zamore, 2002; Yekta et al, 2004), a typical miRNA:mRNA interaction occurs through imperfect complementarity involving the “seed” region comprising nucleotides 2 to 8 of the miRNA (Brennecke et al, 2005; Sontheimer, 2005). Target repression through this type of interaction, referred to miRNA‐like hereafter, is slower as it involves weaker and longer association of RISC with the target (Wee et al, 2012; Salomon et al, 2015). It also involves additional proteins, which form the full miRNA‐loaded RISC (miRISC). The key AGO‐binding partner is GW182 adaptor protein, which recruits further protein factors mediating translational repression coupled with deadenylation and decapping (Chekulaeva et al, 2011; Nishihara et al, 2013; Chen et al, 2014; Rouya et al, 2014).

miRNAs were implicated in countless physiological processes and pathologies. Thousands of mammalian miRNAs were annotated (Kozomara et al, 2019), and more than a half of mammalian genes could be directly targeted by miRNAs (Friedman et al, 2009). Yet, miRNAs are dispensable for mouse oocytes and preimplantation development and their activity in oocytes is negligible (Ma et al, 2010; Suh et al, 2010). Notably, analysis of miRNAs in mammalian oocytes revealed their low cytoplasmic concentration (<0.5 nM) (Kataruka et al, 2020) as opposed to functional miRNAs in somatic cells, which are relatively abundant (Bosson et al, 2014; Denzler et al, 2014, 2016). This is consistent with kinetic studies highlighting miRNA concentration as an important factor for efficient miRNA‐mediated repression (Wee et al, 2012; Salomon et al, 2015). For example, analysis of let‐7 miRNA in HeLa cells, whose transcriptome size was estimated to be ~580,000 mRNA molecules/cell (Bishop et al, 1974), determined that a HeLa cell contains ~50,000 let‐7 molecules (Bosson et al, 2014); this equals to ~20 nM cytoplasmic concentration (assuming ~20 μm cell diameter). In contrast, the cytoplasmic mRNA concentration in somatic cells and oocytes is comparable despite somatic cells containing 0.1–0.6 × 106 mRNA molecules while mouse oocytes accumulate a sizable transcriptome of ~27 × 106 mRNA molecules (Bishop et al, 1974; Hastie & Bishop, 1976; Piko & Clegg, 1982; Carter et al, 2005; Marinov et al, 2014; Fan et al, 2015). Maintenance of mRNA concentration in the cytoplasm in oocytes correlates with extended average half‐life of maternal mRNAs (Jahn et al, 1976; Brower et al, 1981; De Leon et al, 1983), while turnover of maternal miRNAs does not appear to be adapted to the oocyte growth resulting in their dilution (Kataruka et al, 2020).

The “stoichiometric model” explaining the loss of physiologically significant activity of the miRNA pathway thus proposes that oocyte’s growth dilutes maternal miRNA concentration to the point where miRNAs become ineffective regulators of the maternal transcriptome. Another explanation was offered by Freimer et al (2018) who proposed that alternative splicing of Ago2 makes the functional AGO2 a limiting factor, which contributes to the observed miRNA inactivity. However, although this mouse‐specific Ago2 regulation could contribute to the negligible miRNA activity in mouse oocytes, it cannot explain low miRNA abundance and inactivity observed in bovine and porcine oocytes (Kataruka et al, 2020). Here, we report identification and analysis of two exceptionally abundant maternal miRNAs in bovine and porcine oocytes, which overcome the diluting effect, exhibit robust repressive activity, and support the stoichiometric model.

Results and Discussion

Extreme abundance of ssc‐miR‐205 and bta‐miR‐10b in oocytes

During analysis of published small RNA‐sequencing (RNA‐seq) data (Roovers et al, 2015; Gad et al, 2019), we noticed exceptional abundance and possible functional relevance of ssc‐miR‐205 and bta‐miR‐10b in porcine and bovine oocytes, respectively (Fig 1A). Even though miR‐205 and mir10b are conserved across vertebrates (Lim et al, 2003), their high maternal expression is not conserved in mammals (Fig 1A). There are minimal sequence differences among the mouse, porcine, and bovine miR‐205 and mir10b miRNA precursors (Fig 1B). Consequently, the secondary structure of ssc‐miR‐205 and bta‐miR‐10b precursors seems not associated with their high abundance in porcine and bovine oocytes, respectively (Figs 1B and EV1A).

Figure 1

Most abundant miRNAs in mammalian oocytes

miRNA abundance in RNA‐sequencing samples from murine, porcine, and bovine oocytes (Graf et al, 2014; Garcia‐Lopez et al, 2015; Gad et al, 2019). Each graph shows on the y‐axis reads per million of 19‐ to 32‐nt reads for miRNAs and on the x‐axis miRNAs ordered by name.

Schematic depiction of miRNA precursors. Precursor sequences and predicted base pairing were adopted from miRbase miRNA annotations (Kozomara et al, 2019). miR‐10b secondary structures in the miRBase contained two different base pairing versions in the loop, the alternative folding is displayed in Fig EV1A.

qPCR quantification of ssc‐miR‐205 and bta‐miR‐10b miRNAs in murine (M.m.), porcine (S.s.), and bovine (B.t.) oocytes, respectively. Shown are the mean values from three independent experiments. Error bars = SD. A detailed information concerning miRNA quantification is provided in the Methods section.

Figure EV1

Additional schemes of pre‐miRNAs and reporter designs

Alternative fold of bovine, porcine, and murine miR‐10b. The fold was predicted in miRBase for bta‐mir‐10b, and the remaining miRNAs were folded accordingly.

Calibration curves with diluted let‐7a and miR‐205 oligonucleotides used for miRNA quantifications.

miR‐205 abundance in murine, bovine, and porcine oocytes (from experiments shown in Fig 1C) recalculated using the miR‐205 oligonucleotide calibration curve. Shown are the mean values from three independent experiments. Error bars = SD.

Since RNA‐seq data provided only relative estimates of miRNA abundance, we used quantitative RT–PCR to determine copy numbers per oocyte (Figs 1C and EV1C): ssc‐miR‐205 was estimated to have ~1.6 million molecules per oocyte (~4.4 nM cytoplasmic concentration) and bta‐miR‐10b ~261,000 molecules per oocyte (~0.5 nM cytoplasmic concentration). Concentrations were estimated for 105 and 120 μm diameters of porcine and bovine oocytes, respectively (Griffin et al, 2006). In mouse oocytes, we have observed functional repression by miRNAs at 1.5 nM but not at 0.3 nM concentration (Kataruka et al, 2020). Thus endogenous ssc‐miR‐205 would be predicted to suppress gene expression in oocytes, while repressive potential of bta‐miR‐10b was unclear.

Endogenous ssc‐miR‐205 and bta‐miR‐10b are active in oocytes

To estimate ssc‐miR‐205 and bta‐miR‐10b activities in oocytes, we produced and microinjected luciferase reporters with perfectly complementary (“perfect”), partially complementary (“bulged” (Doench et al, 2003)), or mutated miRNA binding sites for miR‐205 or miR‐10b (Fig EV2A). While artificial, bulged sites are a convenient counterpart of perfect sites used in reporters to partially distinguish between RNAi‐like cleavage of the target and typical miRNA‐mediated repression (Doench et al, 2003; Schmitter et al, 2006). Since the extensive 3’ base pairing in bulged sites was associated with target‐mediated miRNA degradation (TDMD, reviewed in Fuchs Wightman et al, 2018), we also produced a control reporter 4x‐seed+suppl (Fig EV2B) where the four miRNA binding sites did not base‐pair in the last five nucleotides, minimizing TDMD effects (Sheu‐Gruttadauria et al, 2019).

Figure EV2

Overview of reporters and their miRNA binding sites

Schematic depiction of NanoLuc reporters and sequences of miRNA binding sites.

An additional miR‐205 reporter 4x‐seed‐suppl, which should not be affected by TDMD, is also repressed by endogenous ssc‐miR‐205. The 4×‐seed‐suppl reporter for ssc‐miR‐205 was analyzed together with other reporters shown in Fig 2A where we showed analogous sets of reporters for bta‐miR‐10b and ssc‐miR‐205 analyzed in bovine and porcine oocytes. Each data point represents a value obtained from five microinjected oocytes. Data were collected in two microinjection sessions, each microinjection session used a new batch of in vitro‐transcribed reporter mRNAs. Error bars = SD.

Figure 2

Endogenous miRNA activity is detected in bovine and porcine oocytes

bta‐miR‐10b and ssc‐miR‐205 miRNAs efficiently repress microinjected NanoLuc luciferase reporters carrying a perfectly complementary (1× perfect) miRNA binding site or four partially complementary (4x‐bulged) miRNA binding sites. See Fig EV2A for an overview of reporter design. Error bars = SD. Each data point represents a value obtained from five microinjected oocytes. Data were collected in two microinjection sessions, each microinjection session used a new batch of in vitro‐transcribed reporter mRNAs.

Derepression of predicted ssc‐miR‐205 targets. In three independent microinjection experiments, ssc‐miR‐205 was inhibited by a microinjected antisense oligonucleotide inhibitor and target levels were assessed 24 h after microinjection by RT–qPCR. Error bars = SD.

Inhibition of ssc‐miR‐205 by a microinjected antisense oligonucleotide inhibitor results in reduced development to the blastocyst stage. Shown is development of microinjected oocytes to the blastocyst stage relative to water‐injected oocytes. Five independent experiments (represented by individual data points) were performed, ~50 porcine oocytes were microinjected in each group in each experiment. Difference between let‐7a and miR‐205 inhibitor effects is statistically significant (two‐tailed paired t‐test P‐value = 0.0169). Error bars = SD.

Data information: In panels A–C, asterisks indicate statistical significance (P‐value) of one‐tailed t‐test (*P < 0.05, **P < 0.01, and ***P < 0.001).

Importantly, for miRNA‐targeted reporters, we used NanoLuc luciferase (England et al, 2016), arguably the best tool for determining endogenous miRNA activity in oocytes. With the same miRNA binding sites as previously used in Renilla or firefly reporters, NanoLuc allows to reduce the reporter amount by an order of magnitude to approximately 10,000 reporter mRNA molecules per oocyte (Kataruka et al, 2020), which is well within the physiological range of maternal mRNA abundance (Fan et al, 2015).

Both miRNAs efficiently repressed all targeted reporters demonstrating that ssc‐miR‐205 and bta‐miR‐10b are active and efficiently repressing targets (Figs 2A and EV2C). Consistent with its much higher abundance, ssc‐miR‐205 in porcine oocytes attained much stronger reporter repression than bta‐miR‐10b in bovine oocytes (Fig 2A). These results provide strong evidence for efficient target repression by endogenous miRNAs in mammalian oocytes.

miRNA activities in mammalian oocytes do not appear to be suppressed in sensu stricto. The miRNA pathway appears mechanistically intact, but cytoplasmic miRNA concentrations are much lower than in somatic cells, and consequently, miRNA targets are repressed weakly. It is thus important that analysis of maternal miRNAs would respect physiological concentrations of miRNAs and their targets in order to avoid artifacts generated by non‐physiological ones.

Several studies reported the presence of functional miRNAs in murine, bovine, or porcine oocytes (Chen et al, 2012; Sinha et al, 2017; Wang et al, 2017; Gad et al, 2020). In these studies, experimental support for significant endogenous miRNA activity in oocytes was inferred from correlative effects of miRNA overexpression or use of antisense inhibitory oligonucleotides (“antagomirs”) (Krutzfeldt et al, 2005). For example, Chen et al reported that miR‐27a activity is not suppressed in porcine oocytes (Chen et al, 2012). However, their functional analysis employed microinjection of oocytes with 50 pl of 500 ng/μl of miRNA mimics or inhibitors. That would correspond to more than a billion microinjected oligonucleotides, while an oocyte contains tens of millions of mRNA molecules (Fan et al, 2015). Similarly, miR‐98 function in mouse oocytes was studied using microinjection of 10 pl of 50 μM miRNA or 100 μM inhibitors (Wang et al, 2017). Since cytoplasmic volume of a mouse oocyte is ~260 pl, such experimental design would result in micromolar concentrations of microinjected oligonucleotides in the cytoplasm. This is three orders of magnitude or more than estimated amounts of maternal miRNAs (Kataruka et al, 2020) and an order of magnitude excess over the amount of maternal mRNAs (Fan et al, 2015). The analysis of miR‐130b in bovine oocytes was conducted near the physiological range, as 10 pl of 50 nM miRNA or inhibitors was injected (Sinha et al, 2017). At the same time, this injected amount is higher than the amount of bta‐miR‐10b, the most abundant miRNA in bovine oocytes reported here, and there is no evidence that bta‐miR‐130b would reach similar abundance. In fact, miR‐130b has three orders of magnitude lower abundance than miR‐10b in RNA‐seq from bovine oocytes (Roovers et al, 2015).

Endogenous ssc‐miR‐205 is biologically relevant

After determining the concentrations of ssc‐miR‐205 and bta‐miR‐10b and assessing their activity, we focused on subsequent functional analysis of ssc‐miR‐205, which showed stronger target repression and because porcine oocytes were easily accessible for experimental analysis. To test whether ssc‐miR‐205 suppresses endogenous targets, we inhibited the miRNA in the oocyte with miR‐205 antagomir microinjection and examined the mRNA levels of five selected predicted endogenous targets: Plcb1, Prkab2, Lsm4, Mbip, and Srfbp1 by qPCR.

To test suppression of the endogenous targets, we microinjected fully grown transcriptionally quiescent porcine oocytes (Prather, 1993) with ~5 × 106 molecules of miR‐205 antagomir. This amount corresponds to ~two‐fold excess over endogenous ssc‐miR‐205 molecules and is much lower than amounts used to study miRNAs in oocytes in the studies mentioned above (Chen et al, 2012; Fan et al, 2015). Microinjected fully grown oocytes were cultured for 24 h in the presence of dbcAMP, which prevents resumption of meiosis (Appeltant et al, 2015). As the first wave of maternal mRNA degradation is induced during resumption of meiosis (Svoboda et al, 2015), the use of dbcAMP avoids interference of complex meiotic mRNA degradation with target degradation by miRNAs.

miR‐205 antagomir caused a significant three‐fold to five‐fold increase in mRNA abundance for all predicted targets but not for two non‐targeted control genes (Fig 2B). Derepression of the endogenous targets upon injection of miR‐205 antagomirs complements results from NanoLuc reporter experiments and implies that ssc‐miR‐205 is indeed functional in porcine oocytes and regulates endogenous gene expression in porcine oocytes.

Next, we examined whether ssc‐miR‐205 plays a significant biological role in oocytes and/or early embryos. We microinjected fully grown porcine oocytes with the miR‐205 antagomir (~two‐fold excess over endogenous ssc‐miR‐205 molecules) and allowed them to undergo meiotic maturation, and then parthenogenetically activated their preimplantation development. We opted for parthenogenetic early development because porcine oocytes suffer a high incidence of polyspermy (Wang et al, 1994), while parthenogenotes can progress through the preimplantation development to the blastocyst stage (Hwang et al, 2020). As a negative control, we used let‐7a inhibitor as this miRNA is present but not effective in porcine oocytes (Kataruka et al, 2020).

ssc‐miR‐205 inhibitor‐injected oocytes showed ~50% reduced ability to support development to the blastocyst stage relative to water‐injected control and let‐7a inhibitor‐injected oocytes (Fig 2C). The precise role of miR‐205 inhibition in the phenotype remains unknown. There was no specific stage at which the development of miR‐205 antagomir‐injected oocytes arrested and there are many predicted ssc‐miR‐205 targets. The miRNA could play a role in shaping the zygotic expression or contribute to maternal mRNA degradation. Limitations of the model prevent determining whether miR‐205 antagomir‐injected blastocysts would develop normally or show additional defects, similar to miR‐430 in zebrafish. miR‐430 is uniquely adapted for rapid and massive expression upon fertilization and contributes to maternal mRNA degradation in zebrafish embryos (Giraldez et al, 2006). Deletion of the miR‐430 cluster in zebrafish causes developmental delay and morphological defects at earlier developmental stages, while mutant embryos are able to undergo gastrulation and organogenesis before they die five days after fertilization (Liu et al, 2020). Thus, dissection of the biological role of ssc‐miR‐205 in early porcine embryos will require further analysis of ssc‐miR‐205 effects on maternal and zygotic transcriptomes to reveal the phenotype severity and strengthen the link between the phenotype and ssc‐miR‐205 function.

High abundance of ssc‐miR‐205 correlates with its longer half‐life

Exceptional abundance of ssc‐miR‐205 in porcine oocytes suggests existence of one or more adaptations, which would be unique for ssc‐miR‐205 in porcine oocytes and enable such a high accumulation of this miRNA during porcine oocyte growth. As mentioned above, there are minimal differences among the porcine, bovine, and murine miRNA precursors, which do not have any apparent effect on the secondary structure of the precursor (Fig 1B). This makes unique secondary structure or sequence variability of mature miRNA or miRNA* less likely explanations for ssc‐miR‐205 accumulation.

Porcine oocyte‐specific high abundance of ssc‐miR‐205 could come from unusually high transcription of the miRNA precursor. RNA sequencing from porcine, bovine, and mouse oocytes (Graf et al, 2014; Garcia‐Lopez et al, 2015; Roovers et al, 2015; Horvat et al, 2018; Gad et al, 2019; Jiao et al, 2020) suggests that ssc‐miR‐205 is well expressed. However, transcription at the ssc‐miR‐205 locus does not seem to be strikingly high (Fig 3). Porcine and bovine miR‐205 apparently originate from a nascent transcript of a spliced lncRNA (Fig 3). Production of the miR‐205 pre‐miRNA and splicing of the lncRNA are mutually exclusive because of an intron/exon boundary within the pre‐miRNA (Fig 3). Albeit the boundary appears conserved in mice, RNA‐seq data from oocytes and somatic tissues do not provide evidence for splicing of the mmu‐miR‐205 host transcript (Abe et al, 2015; Veselovska et al, 2015; Sollner et al, 2017; Horvat et al, 2018) (Fig 3). Abundance of the porcine miR‐205 host lncRNA in oocytes is approximately twenty times higher than the miR‐205 host lncRNA level in bovine oocytes, but it still seems rather low (~3 CPM level, Fig 3) to explain production of 1.6 million ssc‐miR‐205 molecules per oocyte. In comparison, highly abundant mRNAs such as Zp3 and Gdf9, which would be expected to have hundreds of thousands of mRNA copies/oocyte, have expression around 700–800 CPM (Gad et al, 2019; Jiao et al, 2020).

Figure 3

Transcription in the miR‐205 miRNA locus from selected species

UCSC browser snapshots show tracks from small and long RNA‐seq analyses of mouse (Garcia‐Lopez et al, 2015; Horvat et al, 2018), bovine (Graf et al, 2014; Roovers et al, 2015), and porcine oocytes (Gad et al, 2019; Jiao et al, 2020). Tracks were constructed as described in the Material and Methods. The y‐scale for small RNAs depicts counts per million (CPM) of mapped 19‐ to 32‐nt reads. The y‐scale for long RNA‐seq depicts CPM of mapped fragments. Gray rectangles and dashed lines represent exon–intron structure and transcript splicing inferred from RNA‐seq data. Although the splice acceptor of the terminal exon appears conserved, murine RNA‐seq data do not support splicing of the primary miR‐205 transcript. Instead, the mmu‐miR‐205 locus produces a detectable antisense spliced maternal lncRNA while the precursor of mmu‐miR‐205 is undetectable. SA, splice acceptor for the last exon of lncRNA originating from the locus.

Other porcine miRNA loci expressed in oocytes carry miRNAs in introns without apparent interference with splicing. These miRNAs show similar or even higher expression of spliced lncRNA precursor. For example, the ssc‐miR‐16 locus produces the second most abundant maternal miRNA. The miRNA is localized in an intron of a spliced lncRNA, which has much higher abundance than the spliced lncRNA at the ssc‐miR‐205 locus (Fig EV3A). A spliced ncRNA transcript from the miR‐17‐92 cluster locus (also known as the “OncomiR” cluster (He et al, 2005)) is another lncRNA with abundance comparable to the ssc‐miR‐205 locus lncRNA (~3 CPM level). However, relative to ssc‐miR‐205, the six miRNAs from the oncomir cluster had lower amounts with considerable variability among the oncomirs (Fig EV3B). Taken together, transcriptome analysis does not explain exceptionally high abundance of ssc‐miR‐205 in oocytes. We cannot rule out that the precursor is exceptionally highly expressed, efficiently processed into mature miRNA while the host lncRNA is unstable and does not accumulate. However, we find it more likely that in addition to transcription other factors contribute to the observed ssc‐miR‐205 abundance in oocytes.

Figure EV3

UCSC browser snapshots depicting transcription in additional miRNA loci in porcine oocytes

The ssc‐miR‐16 locus giving rise to the second most abundant miRNA in the sequencing dataset (Gad et al, 2019).

The porcine miR‐17/92 (Oncomir cluster) locus.

Data information: Tracks from small and long RNA‐seq analyses in oocytes (Gad et al, 2019; Jiao et al, 2020) were constructed as described in Materials and Methods. The y‐scale for small RNAs depicts counts per million (CPM) of mapped 19‐ to 32‐nt reads. The y‐scale for long RNA‐seq depicts counts per million of mapped fragments. Gray rectangles and dashed lines represent exon–intron structure and splicing of transcripts in the locus inferred from RNA‐seq data. All UCSC browser snapshots are oriented such that depicted miRNAs are transcribed from the left to right.

We thus investigated whether ssc‐miR‐205 accumulation could be also regulated at the level of the mature miRNA. We first examined decay of ssc‐miR‐205 in fully grown oocyte since miRNA turnover appeared to be a contributing factor to miRNA dilution in mouse oocytes. We cultured transcriptionally quiescent fully grown oocytes for 10, 20, and 40 h. dbcAMP was added to the medium to prevent resumption of meiosis and maintain fully grown oocytes in the germinal vesicle stage during the experiment, so the changes in RNA metabolism, which are associated with meiotic maturation, would not affect the outcome of the experiment. We observed that ssc‐miR‐205 exhibits higher stability than three other tested miRNAs, particularly ssc‐let‐7a (Fig 4A and B). Let‐7a belongs to medium‐to‐high abundant miRNAs in porcine oocytes (Fig 1A) (Gad et al, 2019). While ssc‐let‐7a levels were reduced by ~70% over the course of 40 h (similar to mouse oocytes (Kataruka et al, 2020)), ssc‐miR‐205 levels were reduced only by ~25% after 40 h (Fig 4A and B). Additional two examined miRNAs, ssc‐miR‐10b and ssc‐miR‐22, were less stable than ssc‐miR‐205 but more stable than ssc‐let‐7a. These data suggest that increased miRNA stability could be one factor contributing to accumulation of ssc‐miR‐205 during oocyte growth.

Figure 4

Selected features of ssc‐miR‐205

ssc‐miR‐205 turnover. Fully grown transcriptionally quiescent oocytes were cultured for indicated periods of time, and selected miRNAs were quantified by RT–qPCR in four (ssc‐let‐7a), five (ssc‐miR‐22), or six (ssc‐mir10b and ssc‐miR‐205) independent experiments. Error bars = SD.

Comparison of mean relative changes of miRNA level at 20 h of culture (see Fig EV4A for 40 h data). ssc‐miR‐205 in oocytes is less than ssc‐miR‐10b, ssc‐miR‐22, or ssc‐let‐7a. Error bars = SD. Asterisks indicate statistical significance (P‐value) of one‐tailed paired t‐test (*P < 0.05, **P < 0.01).

Treatment by sodium periodate followed by qPCR does not support 2’‐OH modification of either ssc‐miR‐205 or bta‐miR‐10b miRNA. Shown are relative levels of the sodium periodate‐treated samples to non‐treated samples, which were set to one. The experiment was performed three times. Error bars = SD. Efficiency of periodate treatment was confirmed using non‐methylated and methylated miR‐221 RNA oligonucleotides (EV4B).

3′ tailing of small RNAs in murine, bovine, and porcine oocytes (Graf et al, 2014; Garcia‐Lopez et al, 2015; Gad et al, 2019). Shown are percentages of miRNAs tailed at 3′ end with mono‐ and oligonucleotides.

Figure EV4

Additional data concerning analysis of miR‐205 stability

ssc‐miR‐205 in oocytes is less degraded after 40 h of culture than ssc‐miR‐10b, ssc‐miR‐22, or ssc‐let‐7a. Shown is comparison of mean relative changes of miRNA level at 40 h. Error bars = SD. Asterisk indicates statistical significance (P‐value < 0.05) of one‐tailed paired t‐test. Selected miRNAs were quantified in four (ssc‐let‐7a), five (ssc‐miR‐22), or six (ssc‐mir10b and ssc‐miR‐205) independent experiments. Error bars = SD.

Control experiment for oxidation by sodium periodate. Shown are levels of the sodium periodate‐treated non‐methylated and methylated miR‐221 RNA oligonucleotides relative to non‐treated samples, which were set to one. The experiment was performed three times, each time as three technical replicates. Error bars = SD calculated from CT value variability.

The cause of increased stability of ssc‐miR‐205 remains unclear, but it does not involve a 2’‐OH modification of the miRNA, such as 2’‐O‐methylation, which is common for plant miRNAs and was reported to stabilize miR‐21‐5p in lung cancer (Liang et al, 2020). Endogenous ssc‐miR‐205 and bta‐miR‐10b miRNAs vanished upon oxidation, suggesting that they do not carry 2’‐O‐methylation (Fig 4C). In contrast, no change was observed in the level of 2’‐O‐methylated miR‐221 oligonucleotide, which served as a positive control (Fig EV4B).

A well‐established mechanism regulating mature miRNA turnover is non‐templated nucleotide addition at their 3’ end (Krol et al, 2010). Monoadenylation was reported to increase miRNA stability (D'Ambrogio et al, 2012) and is a common miRNA modification observed in mouse oocytes, but miRNA abundance and monoadenylation frequency do not correlate (Yang et al, 2016). We observed that 16.4% of ssc‐miR‐205 in porcine oocytes was monoadenylated, which was more than the monoadenylation frequency observed in the next four abundant porcine miRNAs (Fig 4D). However, bta‐mir‐10b, the most abundant bovine oocyte miRNA, had lower monoadenylation level (4.6%) than less abundant bta‐miR‐148 (10.7%) or bta‐miR‐92a (19.1%, Fig 4D). Thus, analysis of non‐templated additions of five most abundant miRNAs in porcine, bovine, and murine oocytes showed inconclusively variable monoadenylation levels. Furthermore, while monoadenylation can contribute to ssc‐miR‐205 accumulation, the ssc‐miR‐205 monoadenylation could also be in part a consequence of high miRNA stability.

Another candidate mechanism for selective ssc‐miR‐205 accumulation is interaction with some maternal RNA molecule, which would increase ssc‐miR‐205 stability. Such an RNA could explain species‐specific and tissue‐specific ssc‐miR‐205 accumulation in porcine oocytes. A precedent for such a regulation could be effect of circular RNA Cdr1 on miR‐7 in the mouse brain (Piwecka et al, 2017). However, search for a putative binding partner among circRNAs, lncRNA, and mRNAs did not yield any outstanding candidate, which would have high abundance and multiple ssc‐miR‐205 binding sites. The strongest candidate was a highly abundant unspliced ~28‐kb‐long piRNA precursor RNA carrying several miR‐205 seed motifs. However, given the length of this RNA, it is hard to envision how this transcript could selectively stabilize ssc‐miR‐205 and no other maternal miRNAs.

Taken together, we show that highly abundant maternal miRNAs ssc‐miR‐205 and bta‐miR‐10b overcome the constraints imposed on miRNA activity by the size of the maternal transcriptome and volume of the oocyte. Our results support the model that the main cause of the apparent maternal miRNA inactivity is unfavorable miRNA:mRNA stoichiometry. In addition, we provide a framework for the identification of functionally relevant miRNAs in the oocyte. ssc‐miR‐205 reported here is the best experimentally documented example of an active maternal mammalian miRNA, which functionally contributes to zygotic development. At the same time, the lack of conserved high abundance of miR‐205 and miR‐10b in mammalian oocytes suggests that these two miRNAs represent unique evolutionary events. Thus, they are exceptions from the common biological insignificance of maternal miRNAs that did not adapt to the diluting effect of oocyte growth.

Materials and Methods

Oocyte collection and microinjection

Porcine and bovine oocytes were obtained from the slaughterhouse material as described previously (Blaha et al, 2015; Kinterova et al, 2019). For reporter injection, a mixture of in vitro‐transcribed firefly and nanoluciferase (NanoLuc) RNA in the ratio of 100,000:10,000 was injected per oocyte with FemtoJet microinjector (Eppendorf). For antisense oligonucleotide inhibitor injection, commercially obtained hsa‐let‐7a‐5p or hsa‐miR‐205‐5p miRCURY LNA miRNA inhibitor (Qiagen, cat# YI04101776 and YI04101508, respectively) was diluted in water and microinjected (~5 × 106 molecules per oocytes).

Injected bovine oocytes were cultured in MPM media (prepared in‐house (Kinterova et al, 2019)) containing 1 mM dbcAMP (Sigma) without a paraffin overlay in a humidified atmosphere at 39°C with 5% CO2 for 20 h. Porcine oocytes were cultured in M‐199 medium (Gibco) supplemented with 1 mM dbcAMP, 0.91 mM sodium pyruvate, 0.57 mM cysteine, 5.5 mM HEPES, antibiotics, and 5% fetal calf serum (Sigma). Injected oocytes were incubated at 38.5°C in a humidified atmosphere of 5% CO2 for 20 h.

Parthenogenetic activation and embryo culture

After maturation, oocytes were washed twice in PXM‐HEPES and activated by exposure to 10 μM ionomycin in PXM‐HEPES for 5 min. This was followed by two washes in porcine zygote medium 3 (PZM 3) supplemented with 2 mM 6‐DMAP and cultivation for 5 h at 38.5°C in 5% CO2. Then, they were washed twice in PZM 3 and cultivated for seven more days in PZM 3 medium.

cDNA synthesis and qPCR

For miRNA analysis, oocytes were washed with M2 media to remove any residual dbcAMP, collected in a minimum amount of M2 media with 1 μl of RiboLock, and incubated at 85°C for 5 min to release the RNA. cDNA synthesis for miRNA analysis was done with the miRCURY LNA RT Kit (Qiagen, cat# 339340) according to the manufacturer’s protocol. qPCR was set using cDNA fraction corresponding to one oocyte equivalent per qPCR. The miRCURY LNA SYBR Green Kit (Qiagen, cat# 339345) was used as per the manufacturer’s protocol, and the reaction was set in Roche LightCycler 480. Each biological replicate was analyzed in a technical triplicate. The following primer sets from Qiagen were used: hsa‐let‐7a‐5p (cat# YP00205727), hsa‐miR‐205‐5p (cat# YP00204487), hsa‐mir‐10b‐5p (cat# YP00205637), and hsa‐mir‐221‐3p (cat# YP00204532). For miRNA quantification, a standard curve was produced using a let‐7a oligonucleotide as described previously (Kataruka et al, 2020). A standard curve was also produced using a miR‐205 oligonucleotide (Fig EV1B) to assure that the let‐7a oligonucleotide used in the previous work and here did not distort ssc‐miR‐205 quantification. Comparison of quantifications showed that miR‐205 oligonucleotide‐based calibration yielded slightly higher estimate of ssc‐miR‐205 (1.91 × 106 molecules, Fig EV1C) than let‐7a oligonucleotide‐based calibration (1.62 × 106 molecules, Fig 1C). Thus, let‐7a oligonucleotide‐based calibration was used for data presentation to be consistent with the previous report (Kataruka et al, 2020).

For mRNA analysis, oocytes were washed and lysed for cDNA synthesis as for miRNA. Reverse transcription with Maxima Reverse Transcriptase (Thermo Scientific) was primed with random hexanucleotides. Maxima SYBR Green qPCR Master Mix (Thermo Scientific) was used for qPCR. qPCR was set using cDNA fraction corresponding to one oocyte equivalent per qPCR in technical triplicates for each biological sample (also in triplicates). Average CT values of the technical replicates were normalized to uninjected control oocytes.

Plasmid reporters

Nanoluciferase plasmid pNL1.1 was obtained from Promega. The Nluc gene was cleaved out and ligated in the phRL‐SV40 plasmid downstream of the T7 promoter. miRNA binding sites were obtained as oligonucleotides from Sigma. Oligonucleotides were annealed, phosphorylated, and cloned downstream of the Nluc gene in XbaI site. The firefly control plasmid was made in‐house with a T7 promoter (oligonucleotide sequences are provided in Table EV1).

In vitro transcription

Linearized pNL‐miR‐205 and miR‐10b‐perfect, bulged, mutant, and firefly plasmids were in vitro‐transcribed, capped, and then polyadenylated by Poly(A) Tailing Kit (Thermo Fisher).

Luciferase assay

Five oocytes were collected per aliquot, and Nano‐Glo Dual‐Luciferase Reporter Assay System (Promega, cat# N1610) was used to measure the samples as per the manufacturer’s protocol.

2’‐OH methylation assay

Oligonucleotides or oocytes (supplemented with a methylated miR‐221 oligonucleotide as an internal normalization control) were incubated in 50 μl Borax buffer with sodium periodate for 30 min at room temperature in dark. Next, 3 μl of 60% glycerol was added and further incubated for 10 min at RT. This was followed by phenol–chloroform extraction and precipitation with isopropanol. The pellet was dissolved in water and used for cDNA synthesis followed by qPCR, which was set in technical triplicates for each biological sample (also in triplicates). Average CT values were first normalized to internal methylated miR‐221 control followed by normalization to untreated sample.

ssc‐miR‐205 predicted target selection

To identify putative ssc‐miR‐205 targets, we used sequences of 3'UTRs of all annotated porcine genes and counted the presence of complementary sequences to ssc‐mir‐205 seeds (seed 6mer UGAAGG, seed 7mer_m8 AUGAAGG, seed 8mer AUGAAGGA, and seed 7mer_1a UGAAGGA). From the top‐scoring genes with most seed matches, we selected four targets based on low but well detectable expression in RNA‐seq data from porcine oocytes (Gad et al, 2019). The rationale for this criterion was that miRNA targets would be expected to have reduced expression. Each of the selected putative targets had seven to nine 6mers and two to four 7mers (Table EV2). In addition, we included Plcb1 among the putative targets as a candidate conserved target because it was the best predicted mmu‐miR‐205 target by TargetScanVert in miRBase (Kozomara et al, 2019) and its porcine transcript carried one conserved 8mer binding site plus an additional 6mer (Table EV2).

Analysis of high‐throughput sequencing data

Small RNA‐sequencing data from mouse (GSE59254 (Garcia‐Lopez et al, 2015)), porcine (GSE122032 (Gad et al, 2019)), and bovine (GSE 64942 (Roovers et al, 2015)) oocytes were mapped to their respective genomes (mouse—mm10, cow—bosTau9, pig—susScr11) as described previously (Demeter et al, 2019). After mapping, 21‐ to 23‐nt‐long reads perfectly aligned to the genome were selected for mature miRNA expression quantification with annotation from miRBase (Kozomara et al, 2019) using featureCounts v2.0.0 program (Liao et al, 2014):

For bovine and porcine samples, miRBase annotation was lifted over from older genome versions using Liftoff (Shumate & Salzberg, 2020).

Long RNA‐seq data from mouse (GSE116771 (Horvat et al, 2018)), porcine (PRJNA548212 (Jiao et al, 2020)), and bovine (GSE52415 (Graf et al, 2014)) fully grown oocytes were mapped to their respective genomes (mouse—mm10, cow—bosTau9, pig—susScr11) using STAR 2.5.3a (Dobin et al, 2013) as previously described (Horvat et al, 2018).

Read mapping coverage was visualized in the UCSC Genome Browser by constructing bigWig tracks from merged replicates using the UCSC tools (Kent et al, 2010).

Author contributions

SK, VK, and MIRK performed the experiments. FH analyzed RNA‐sequencing data. SK, JK, and PS designed the experiments, supervised, and analyzed experimental data. SK and PS wrote the manuscript.

Conflict of interest

The authors declare that they have no conflict of interest.

Expanded View Figures PDF

Click here for additional data file.

Table EV1

Click here for additional data file.

Table EV2

Click here for additional data file.