Selective reduction of dormant maternal mRNAs in mouse oocytes by RNA interference

‘Omne vivum ex ovo’ (All living things come from eggs), which is attributed to William Harvey, is probably the first articulation of the current view that the program for early development is established during oogenesis. During oogenesis in the mouse, oocytes grow and acquire the ability to resume and complete meiosis (acquisition of meiotic competence) (Sorensen and Wassarman, 1976; Wickramasinghe et al., 1991), as well as the ability to be fertilized and develop to term (acquisition of developmental competence) (Eppig and O’Brien, 1996). Meiotic maturation and egg activation are accompanied by the recruitment of many maternal mRNAs (Schultz et al., 1979; Schultz and Wassarman, 1977; Van Blerkom, 1981), and presumably some of these direct the synthesis of proteins that are required for the formation of a fertilizable egg that is capable of developing to term. One such mRNA is the Mos mRNA. The mobilization of the Mos mRNA results in the ultimate activation of mitogen-activated protein (MAP) kinase, whose activity is required to maintain arrest at metaphase II (Gebauer and Richter, 1997; Sagata, 1997); oocytes lacking the Mos gene mature to metaphase II but then undergo spontaneous activation, i.e., they emit the second polar body and form a pronucleus (Colledge et al., 1994; Hasimoto et al., 1994). The tissue plasminogen activator (tPA, Plat) mRNA is another maternal mRNA that is recruited during oocyte maturation (Huarte et al., 1987; Vassalli et al., 1989). Although tPA is synthesized during maturation and secreted following fertilization, and then becomes associated with a cell-surface receptor on the embryo (Carroll et al., 1993), Plat knockout mice are viable and fertile, but do display mild perturbations in phenotype, e.g., retardation in neuronal migration (Seeds et al., 1999).

To date, an antisense RNA approach has been the most widely used method to assess the function of maternal mRNAs that are recruited during oocyte maturation. Nevertheless, this approach has problems. For example, an antisense RNA approach has been used to assess the role of Mos mRNA recruitment during maturation. The phenotypes observed range from permitting germinal vesicle breakdown but inhibiting emission of the first polar body (Paules et al., 1989; Zhao et al., 1991), to emission of the first polar body but entering interphase instead of proceeding to and arresting at metaphase II (O’Keefe et al., 1989). In contrast, the phenotype of a Mos null mutant generated by homologous recombination is that the oocytes proceed to metaphase II, but meiotic arrest is not maintained and the eggs spontaneously undergo parthenogenetic activation (Colledge et al., 1994; Hasimoto et al., 1994). This discrepancy between the phenotypes observed by the antisense approach with that of a ‘true’ knockout potentially confounds the use of antisense RNA to study the function of a dormant maternal mRNA. Antisense RNA can also target and destroy the Plat mRNA. The efficacy of destruction of the untranslated Plat mRNA, however, appears restricted to antisense RNA directed towards the 3′ UTR (Strickland et al., 1988). Antisense RNAs directed at other portions of the Plat mRNA are far less effective and can form hybrids only following maturation and the concomitant recruitment of the Plat mRNA. Thus, the efficacy of this approach is compromised by the appropriate selection of the region of the mRNA to be targeted, and this can only be determined experimentally and not a priori.

Recently, RNA interference (RNAi), which employs double-stranded RNA (dsRNA), has been shown to ablate potently the targeted mRNA in a variety of species (Sanchez-Alvardado and Newmark, 1999; Fire et al., 1998; Kennerdell and Carthew, 1998; Li et al., 2000; Lohmann et al., 1999; Misquitta and Paterson, 1999; Ngo et al., 1998; Wargelius et al., 1999). The destruction of the targeted mRNA by dsRNA occurs prior to translation (Fire, 1999; Montgomery et al., 1998; Sharp, 1999; Zamore et al., 2000), and targets exon sequences; dsRNA directed against intron sequences is ineffective (Fire et al., 1998). Genetic approaches in Caenorhabditis elegans have identified genes with homology to eIF-2C, RNase D, and RNA-directed RNA polymerase (Ketting et al., 1999; Tabara et al., 1999; Smardon et al., 2000) that are involved in the RNAi-mediated pathway of mRNA degradation. Very recent studies suggest that a nuclease involved in the destruction of the targeted mRNA contains an essential RNA component containing approx. 25-nucleotide RNAs that are homologous to the dsRNA (Hammond et al., 2000). The processing of the dsRNA to these fragments does not require the presence of the targeted mRNA, and the targeted mRNA is cleaved only in the regions of identity to the dsRNA and at sites that are 21-23 nucleotides apart (Zamore et al., 2000).

We report here that dsRNA directed towards Mos and Plat mRNAs in mouse oocytes effectively results in the specific reduction of the targeted mRNA in both a time- and concentration-dependent manner. Moreover, dsRNA is more potent than either sense or antisense RNA. Targeting the Mos mRNA results in inhibiting the appearance of MAP kinase activity, as well as promoting parthenogenetic activation of the treated cells, and targeting Plat mRNA results in inhibiting production of tPA activity. Effective reduction of the Mos and Plat mRNA is observed with stoichiometric amounts of Mos and Plat dsRNA, respectively. While these studies were in progress, a paper appeared that has reported that oocytes injected with Mos dsRNA undergo egg activation, as evidenced by pronucleus formation (Wianny and Zernicka-Goetz, 2000).

For Mos amplification, a pair of primers was designed based on the cDNA sequence (Accession number J00372). The sequence of upstream Mos primer was 5′-CCATCAAGCAAGTAAACAAG-3′ and the downstream Mos primer was 5′-AGGGTGATTCCAAAAGA-GTA-3′. These primers generated a PCR product that was 535 bp in length and corresponded to the 3′ end of the coding region and the beginning of the 3′UTR. Likewise, for Plat amplification a pair of primers was designed based on the cDNA sequence (Accession number J03520). The sequence of the upstream Plat primer was 5′-CATGGGCAAGCGTTACACAG-3′ and the downstream Plat primer was 5′-CAGAGAAGAATGGAGACGAT-3′. These primers generated a PCR product that was 650 bp in length and corresponded to the middle part of the coding region.

To generate template for transcription in vitro, 5 μg of liver total RNA were reverse transcribed with Superscript II reverse transcriptase (Gibco BRL) according to the manufacturer’s instructions using oligo-dT as the primer. PCR amplification conditions for both Mos and Plat were as follows: initial denaturation at 94°C for 4 minutes was followed by 36 cycles of 94°C for 30 seconds, 60°C for 30 seconds, 72°C for 1 minute and the final cycle had an extended incubation at 72°C for 7 minutes followed by decrease to 4°C. All PCR reactions were performed in either a PE2400 or PE9600 PCR thermocycler.

Gel-purified primary PCR products were diluted 1:500 and re-amplified to produce specific templates to generate sense and antisense transcripts by transcription in vitro. To do this, primers were made that contained an SP6 promoter attached to the 5′ end of both the forward and reverse primers. Following PCR under the above conditions, the secondary PCR products were purified using a Nucleospin Extraction Kit (Clontech). The template (500-1000 ng) was then transcribed with SP6 RNA polymerase (Promega) in order to obtain the corresponding sense and antisense RNAs.

The in vitro transcription products were resolved following electrophoresis in 1.5% NuSieve LM agarose (FMC, Rockland, ME, USA) and the bands corresponding to the sense and antisense single-stranded RNA were purified according to the manufacturer’s protocol. Equimolar amounts of sense and antisense RNA were then annealed in 1 mM Tris-HCl (pH 7.5), containing 1 mM EDTA, or in DEPC-treated water supplemented with 5% RNasin (Promega); similar results were obtained using either procedure. Typically, 2-4 μg of RNA in 30 μl were mixed and heated in 500 ml of boiling water for 1 minute. The sample, still in the water bath, was allowed to cool to room temperature over the course of several hours. The dsRNA was phenol extracted, ethanol precipitated, washed in 75% ethanol and then dissolved in water. Samples were stored in water at −80°C prior to use.

RNA was isolated from oocytes and prepared for RT-PCR as previously described (Temeles et al., 1994). In each case, 0.125 pg of rabbit β-globin mRNA/oocyte was added prior to RNA isolation. The globin mRNA serves as an internal standard for the efficiency of the RT-PCR reactions (Temeles et al., 1994). For each set of gene-specific primers the linear region of semi-log plots of the amount of PCR product as a function of cycle number was determined and a cycle number for each primer pair was selected that was in this linear range; the amount of PCR product under these conditions is proportional to the number of cells used (Manejwala et al., 1991). This method permits the comparison of relative changes in the abundance of a particular transcript (Ho et al., 1995; Latham et al., 1994; Temeles et al., 1994).

Following reverse transcription two oocyte equivalents were used as a template for each PCR reaction. PCR products were labeled with [α-³²P]dCTP (Amersham, 0.25 μCi per 50 μl reaction). PCR amplification conditions for both Mos and Plat were as follows: initial denaturation at 94°C for 2 minutes was followed by 28 (Plat) or 31 (Mos) cycles of 94°C for 30 seconds, 60°C for 30 seconds and 72°C for 1 minute, followed by 4°C until the samples were removed. PCR amplification conditions for globin: initial denaturation at 94°C for 2 minutes was followed by 24 cycles of 94°C for 10 seconds and 62°C for 15 seconds followed by final 4°C. After PCR, the products were subjected to electrophoresis in an 8% polyacrylamide gel. The gel was dried under vacuum for 1 hour at 80°C, exposed in phosphorimager cassette for 4 to 24 hours and the signal was quantified using the Storm 860 PhosphorImager and ImageQuant software (Molecular Dynamics, Sunnyvale, CA, USA).

Fully grown, germinal vesicle (GV)-intact oocytes were obtained from pregnant mare’s serum gonadotropin (PMSG)-primed six-week-old female CF-1 mice (Harlan) and freed of attached cumulus cells, as previously described (Schultz et al., 1983). The collection medium was bicarbonate-free minimal essential medium (Earle’s salt) supplemented with polyvinylpyrrolidone (3 mg/ml) and 25 mM Hepes, pH 7.3. Germinal vesicle breakdown was inhibited by including 0.2 mM 3-isobutyl-1-methyl-xanthine (IBMX). The oocytes were transferred into CZB medium (Chatot et al., 1989) containing 0.2 mM IBMX (CZB+IBMX) and cultured in an atmosphere of 5% CO₂ in air at 37°C. Oocytes were microinjected in bicarbonate-free CZB containing 10 mM Hepes and 0.2 mM IBMX with 5 pl of the corresponding solution; the injections were performed as previously described (Kurasawa et al., 1989). The concentration of the undiluted stock solution was 0.2 μg/μl and injection of 5 pl of either Mos or Plat dsRNA corresponds to 1.7×10⁶ and 1.4×10⁶ molecules, respectively. When single-stranded RNA was injected, it was diluted to a concentration such that injection of 5 pl corresponded to the same number of molecules as when dsRNA was injected. In experiments in which either enzyme activity or phenotype was assayed, microinjected oocytes were cultured in CZB+IBMX for 10 or 20 hours. They were then washed through ten drops of IBMX-free CZB and cultured in CZB until oocyte collection and lysis. In experiments in which mRNA levels were measured, the oocytes were kept in medium containing IBMX for 10, 20, or 40 hours until they were collected and processed for RNA isolation.

tPA activity was assayed by zymography of single oocytes. Immobilon-P (Millipore) was soaked in methanol for 1 minute and then rinsed four times with 50 mM Tris-HCl (pH 8.0) containing 50 mM NaCl. The wet membrane was placed on Whatman paper soaked with this buffer and both were transferred to a 96-well dot blot apparatus (Milliblot Systems, Millipore) with the Immobilon-P facing upwards. A 96-well template was then placed on the stage and single oocytes were transferred in 1-2 μl of CZB medium to the middle of where the wells would form. The apparatus was completely assembled and the wet membrane with the oocytes was then exposed to vacuum suction for 2 minutes. The apparatus was then disassembled and the membrane was immediately applied on the detection gel; the detection gel, which contained 40 μg/ml of plasminogen (Fluka), was prepared as previously described (Vassalli et al., 1984). Zymograms were developed for 12-64 hours at 37°C, scanned with a black background and the lysed area was estimated using the ImageQuant software (Molecular Dynamics).

The activities of both histone H1 and myelin basic protein (MBP) kinases were determined in single eggs as follows: single eggs were transferred in 1.5 μl of culture medium into a tube containing 3.5 μl of double kinase lysis buffer (10 μg/ml aprotinin, 10 μg/ml leupeptin, 10 mM p-nitrophenyl phosphate, 20 mM β-glycerophosphate, 0.1 mM sodium orthovanadate, and 5 mM EGTA). The tubes were immediately frozen on dry ice and stored at −80°C until the assay was performed. The kinase reaction was initiated by the addition of 5 μl of double kinase buffer (24 mM p-nitrophenyl phosphate, 90 mM β-glycerophosphate, 24 mM MgCl₂, 24 mM EGTA, 0.2 mM EDTA, 4.6 mM sodium orthovanadate, 4 mM NaF, 1.6 mM dithiothreitol, 60 μg/ml aprotinin, 60 μg/ml leupeptin, 2 mg/ml polyvinyl alcohol, 2.2 μM protein kinase A inhibitor peptide (Sigma), 40 mM 3-(n-morpholino) propanesulfonic acid (MOPS), pH 7.2, 0.6 mM ATP, 2 mg/ml histone (type III-S, Sigma), 0.5 mg/ml MBP) with 500 μCi/ml [γ-³²P]ATP (3000 Ci/mmol; Amersham). To determine the background level of phosphorylation for H1 and MBP, 5 μl of double kinase lysis buffer was added instead of the egg lysate. The reaction was conducted for 30 minutes at 30°C and terminated by the addition of 10 μl double-strength concentrated SDS-PAGE sample buffer (Laemmli, 1970) and boiling for 3 minutes. Following SDS-PAGE, the 15% gel was fixed in 10% acetic acid/30% methanol, dried and exposed to a phosphorimager screen for 16 to 24 hours. Scanning and quantification of the signal were performed using a Storm 860 PhosphorImager and ImageQuant software (Molecular Dynamics, Inc., Sunnyvale, CA). For each experiment, the mean value of H1 or MBP kinase activities for oocytes microinjected with sense Mos RNA was arbitrarily set as 100% and the values obtained in the other groups of eggs were expressed relative to this amount.

dsRNA for either Mos or Plat was prepared by hybridizing equimolar amounts of gel-purified, single-stranded sense and antisense transcripts that were generated by transcription in vitro of appropriate templates containing an SP6 promoter. In each case, the RNA was directed towards a coding portion of the transcript. Following hybridization, the single-stranded RNAs were essentially totally converted to dsRNA, as evidenced by the absence of any visible staining in the region of the gel that corresponded to single-stranded species (Fig. 1). The quantitative nature of hybridization permitted use of the dsRNA without any need for gel purification of the dsRNA species.

Mos and Plat are two maternal mRNAs that are recruited during oocyte maturation (see Introduction). We selected to target the Mos mRNA since a Mos null oocyte has a defined phenotype, i.e., the oocyte matures to metaphase II, but rather than arresting at metaphase II, it undergoes spontaneous egg activation. In addition, it is possible to measure MAP kinase activity in a single oocyte; MAP kinase activity reflects the mobilization of Mos mRNA (see below). We also selected to target Plat mRNA, which like Mos, is a moderately abundant mRNA; it has been estimated that an oocyte contains approx. 10,000 transcripts each of Mos and Plat (Huarte et al., 1987; Keshet et al., 1988). In addition, it is also possible to measure tPA activity in single oocytes.

Oocytes were injected with approx. 10⁶ molecules of either Plat or Mos dsRNA that was directed towards the coding region of each transcript; this corresponds to approx. 10 nM final concentration (see Discussion). The oocytes were then cultured in medium containing IBMX to inhibit resumption of meiosis; a decrease in cAMP is associated with resumption of meiosis and including the membrane-permeable phosphodiesterase inhibitor IBMX in the medium prevents the decrease in cAMP and thus the resumption of meiosis (Schultz et al., 1983). Following culture, RNA was isolated and the relative amount of Plat and Mos transcripts were determined by a semi-quantitative RT-PCR assay that permits quantification of relative changes in transcript abundance. Prior to RNA isolation, a known amount of rabbit globin mRNA was added; this served as a control for RNA recovery, and for the efficiency of the RT-PCR (Temeles et al., 1994).

Oocytes injected with Mos dsRNA displayed a marked reduction in the amount of Mos transcript (approx. 80%), relative to water-injected or uninjected controls (Fig. 2, compare lane 2 with lanes 4 and 5). Likewise, oocytes injected with Plat dsRNA displayed an approx. 90% reduction in the amount of Plat transcript relative to the control (Fig. 2, lane 3). Specificity of this effect was demonstrated by the finding that Mos dsRNA did not reduce the abundance of Plat mRNA, and reciprocally, that Plat dsRNA did not reduce the abundance of Mos mRNA (Fig. 2). Results of these experiments indicated that the machinery for RNAi-mediated degradation of the targeted endogenous mRNA is present and functions in mouse oocytes.

In other systems, antisense RNA can be ineffective. For example, injection of C. elegans with antisense RNA directed towards the unc-22 gene does not result in the mutant twitching phenotype, whereas dsRNA does (Fire et al., 1998). Nevertheless, antisense RNA can be effective in degrading oocyte mRNAs (Strickland et al., 1988). Accordingly, we determined the effect of sense and antisense Mos and Plat RNA on targeting the cognate oocyte transcript.

Oocytes were injected with approx. 10⁶ copies of either sense, antisense or dsRNA, and incubated for 20 hours in medium containing IBMX, before the RNA was isolated and transcript abundance determined. As anticipated, dsRNA directed towards either Mos or Plat mRNA resulted in the reduction of the targeted mRNA, whereas the untargeted transcript remained essentially intact (Fig. 3A). As also anticipated, injection of sense RNA resulted in little, if any, decrease in the abundance of either the targeted or nontargeted mRNA. Injection of either Mos or Plat antisense RNA, however, did result in a decrease in the targeted, but not in the nontargeted, mRNA (Fig. 3A). Little, if any decrease in the targeted mRNA was observed when the amount of injected Mos or Plat antisense RNA was decreased by 10-fold (Fig. 3B). In contrast, this amount of dsRNA was effective in decreasing the amount of the targeted mRNA (Fig. 3B), e.g., the Mos dsRNA resulted in an approx. 85% decrease in Mos mRNA, and Plat dsRNA resulted in an approx. 30% decrease in Plat mRNA. Results of these experiments suggest that dsRNA is more effective in targeting mRNAs than antisense RNA.

In the experiments described above, the oocytes were injected with approx. 10⁶ molecules of dsRNA and cultured for 20 hours prior to determining the relative amount of targeted transcript. In order to determine further characteristics of the RNAi effect, the concentration- and time-dependence of this effect were determined. Oocytes were injected with 10⁶, 10⁴, or 10² molecules of either Mos or Plat dsRNA, and then incubated for 10, 20, or 40 hours prior to determining the relative abundance of the endogenous Mos and Plat transcripts (Fig. 4). For both Mos and Plat dsRNA-injected oocytes, the targeted message was destroyed in both a time- and concentration-dependent manner. In all cases, the nontargeted mRNA was not destroyed (data not shown).

Injection of 10⁶ or 10⁴ molecules of Mos dsRNA resulted in a substantial reduction in the amount of Mos mRNA, such that by 20 hours more than 75% of the mRNA was degraded; 10² molecules of injected Mos dsRNA had little, if any effect over the 40-hour timecourse. Although 10⁶ molecules of injected Plat dsRNA also dramatically reduced the amount of Plat mRNA, the kinetics of Plat mRNA degradation were slower, when compared with those obtained for Mos dsRNA. In addition, 10⁴ molecules of Plat dsRNA was not nearly effective as 10⁴ molecules of Mos dsRNA. Similar to Mos dsRNA, the 100 molecules of injected Plat dsRNA was ineffective in reducing the amount of Plat mRNA.

The experiments described above documented that both Mos and Plat dsRNA could result in the degradation of the targeted mRNA in a concentration- and time-dependent manner. We next demonstrated that the reduction of the targeted mRNA resulted in loss of formation of the encoded protein. The translation of Mos mRNA that initiates at the onset of oocyte maturation results in synthesis of MOS, which in turn activates MAP kinase kinase by phosphorylation (Gebauer and Richter, 1997; Sagata, 1997). MAP kinase kinase, which is a dual-specificity kinase, then phosphorylates MAP kinase on Thr183 and Tyr185 in the mammal, which in turn results in MAP kinase activation (Nishida and Gotoh, 1993). MAP kinase, which is a component of cytostatic factor (CSF) and is required to maintain metaphase II arrest, is frequently assayed by measuring the phoshorylation of MBP. Concomitant with germinal vesicle breakdown is the activation of p34^cdc2/cyclin B kinase (MPF) (Gebauer and Richter, 1997; Sagata, 1997), which is routinely assayed by phosphorylation of histone H1. In the mouse, MPF activation precedes MAP kinase activation by about 1-2 hours, and both activities reach maximal levels in the metaphase II-arrested egg (Verlhac et al., 1993). Following fertilization, MPF activity declines prior to MAP kinase activity (Moos et al., 1995; Verlhac et al., 1993).

Oocytes were injected with either Mos dsRNA, antisense RNA or sense RNA and cultured for 20 hours in IBMX-containing medium, then transferred to IBMX-free medium. The oocytes then matured to metaphase II, at which time both MAP and p34^cdc2/cyclin B kinase activities were assayed simultaneously in single eggs. As expected, sense RNA did not inhibit either kinase activity when compared with uninjected or water-injected eggs (data not shown). In contrast, both Mos antisense and dsRNA inhibited MAP kinase activity, although a greater degree of inhibition was observed with dsRNA (Fig. 5). This result was consistent with Mos dsRNA eliciting a greater decrease in Mos mRNA than Mos antisense RNA (Fig. 3). Although Mos antisense RNA did inhibit MAP kinase activity, the level of p34^cdc2/cyclin B kinase, i.e., histone H1 kinase, was reduced by only about 25% relative to control sense-injected or uninjected oocytes, while a 70% decrease was observed in the dsRNA-injected oocytes. This reduced amount of histone H1 kinase activity in the dsRNA-injected oocytes was a consequence that in numerous cases these eggs underwent spontaneous egg activation, which results in a decrease in histone H1 kinase activity. In contrast, the antisense-injected oocytes never underwent egg activation.

Although Mos antisense RNA, which did result in a decrease in Mos mRNA, could inhibit MAP kinase activation, the results presented in Fig. 3 indicate that dsRNA is a more potent inhibitor than antisense RNA. Accordingly, oocytes were injected with 1/10 the amount of Mos sense, antisense, or dsRNA, cultured for 20 hours in IBMX-containing medium and then matured to metaphase II by transferring them to IBMX-free medium. The eggs were then assayed for both MAP and p34^cdc2/cyclin B kinase activities simultaneously in individual eggs. Whereas both Mos sense and antisense RNA did not inhibit the appearance of MAP kinase activity (p34^cdc2/cyclin B kinase activity was also high in these eggs), Mos dsRNA still elicited a dramatic inhibition in MAP kinase activity, and a corresponding decease in p34^cdc2/cyclin B kinase activity (Fig. 6). These results strengthen the conclusion that Mos dsRNA is more potent than Mos antisense RNA in promoting the reduction of the endogenous Mos mRNA.

tPA is synthesized during oocyte maturation and its activity can be assayed in single oocytes by zymography (Huarte et al., 1985, 1987; Strickland et al., 1988). We observed only a single band (M_r 72,000) when metaphase II-arrested eggs were used; no activity was observed in GV-stage oocytes (data not shown). The presence of a single activity responsible for generating the lytic zone permitted analysis of tPA activity by simply spotting an oocyte/egg on a membrane, which was then overlaid with an agarose gel containing non-fat dry milk and plasminogen. The area of the lytic zone was linear as a function of time after a lag, which probably reflected the time to activate the zymogen cascade and degrade enough substrate to be visible to the eye (Fig. 7).

We assayed the effect of Plat sense, antisense and dsRNA on tPA activity in matured oocytes. Oocytes were injected with approx. 10⁶ molecules of each RNA and cultured in IBMX-containing medium for 20 hours prior to initiating maturation by transfer to IBMX-free medium. Culture for 18 hours resulted in the production of metaphase II-arrested eggs that were then assayed for tPA activity. Injection of either antisense RNA or dsRNA resulted in a dramatic reduction in the amount of tPA activity, when compared with sense-injected oocytes (Fig. 8, black bars). The ability of Plat antisense RNA to inhibit the production of tPA activity following maturation is consistent with its ability to target the destruction of Plat mRNA (see Fig. 3A and Strickland et al., 1988), as well as its ability to inhibit translation of the Plat mRNA (Strickland et al., 1988). Nevertheless, injection of Plat sense RNA also modestly inhibited the production of tPA activity, although to a lesser extent than either Plat antisense or dsRNA (Fig. 8, black bars). We observed that Mos sense RNA, which doesn’t target the Plat mRNA, also resulted in a 50% decrease in tPA activity but had no inhibitory effect on the activation of MAP kinase (data not shown). The molecular basis underlying this inhibitory effect of sense RNA on the generation of tPA activity remains unresolved.

The results presented in Fig. 3 indicated that Plat dsRNA was more potent than Plat antisense RNA in targeting the reduction of endogenous Plat mRNA. As expected, a ten-fold dilution of Plat antisense RNA resulted in levels of tPA activity similar to that of sense-injected oocytes, whereas injection of Plat dsRNA still promoted a dramatic inhibition (Fig. 8, white bars). Similar to the results obtained with Mos dsRNA, these results confirmed that Plat dsRNA was more potent than Plat antisense RNA in promoting the reduction of the endogenous Plat mRNA.

We have demonstrated that RNAi is an effective and efficient method to inhibit the translation of maternal mRNAs that are recruited during oocyte maturation. The reduction of the targeted mRNA, namely Mos and Plat, is specific, i.e., a nontargeted mRNA is not destroyed, and is both time- and concentration-dependent; mRNA levels can be decreased by up to 90%. In addition, dsRNA is more effective than antisense RNA. The reduction of Mos mRNA led to a failure in MAP kinase activation that normally accompanies oocyte maturation. A consequence of this failure is that metaphase II arrest was not maintained and that the eggs underwent parthenogenetic activation with the concomitant decrease in H1 kinase activity. Likewise, reduction of the endogenous Plat mRNA inhibited the production of tPA following oocyte maturation. A recent report also found that injection of Mos dsRNA results in parthenogenetic activation of mouse eggs (Wianny and Zernicka-Goetz, 2000); it was not shown in that study, however, that the Mos RNA is selectively degraded and that MAP kinase failed to activate. In the mouse, RNAi, which entails microinjection of the dsRNA, should prove far superior to antisense approaches that have been used in the past, but with variable success. It should be noted that culture of oocytes in medium containing either Plat or Mos dsRNA (2 μg/μl) does not reduce the amount of the targeted mRNA (P. S., P. S. and R. M. S., unpublished). Thus, in contrast to lower species such as C. elegans (Fire et al., 1998) and planaria (Sanchez-Alvarado and Newmark, 1999), in which injection of the dsRNA into the animal results in the reduction of the targeted mRNA, mouse oocytes apparently lack this uptake mechanism or, if it is present, it is very inefficient.

About 1.5×10⁶ molecules of either Plat or Mos dsRNA were injected when undiluted dsRNA was used. This corresponds to an intracellular concentration of 10 nM, as the volume of an oocyte is approx. 250 pl. This concentration is similar to that required to ablate frizzled function in Drosophila embryos (Kennerdell and Carthew, 1998). In those experiments approx. 0.2 fmole of dsRNA was injected into syncytial blastoderm embryos whose volume is approx. 7.3 nl and this corresponds to approx. 25 nM. Concentrations of 10 nM dsRNA are effective in an in vitro system that supports the destruction of the targeted mRNA (Tuschl et al., 1999). Significant reduction of both Mos and Plat mRNAs are also observed when only 10,000 molecules of Mos or Plat dsRNA are injected. As the oocyte contains about 10,000 each of these transcripts (Huarte et al., 1987; Keshet et al., 1988), the reduction of the endogenous mRNA appears to be very efficient. A catalytic mechanism is possible, as in other systems the number of dsRNA molecules per cell is likely to be less than the number of endogenous transcripts. For example, in C. elegans, injection of 60,000 unc-22 dsRNA into adult animals results in the twitching phenotype in approx. 100 progeny (Fire et al., 1998). unc-22 expression commences when the embryos contain about 500 cells, by which time the injected dsRNA would be diluted to only a few molecules per cell. Alternatively, the recent finding that an RNA-directed RNA polymerase is implicated in RNAi (Smardon et al., 2000) could provide an amplification mechanism that accounts for the efficacy of stoichiometric or substoichiometric amounts of dsRNA to promote the efficient reduction of the targeted mRNA.

When approx. 10⁶ or 10⁴ molecules of either Mos or Plat dsRNA are injected, the kinetics of Plat mRNA degradation are slower than that for Mos mRNA. As it has been estimated that oocytes contain approx. 10 000 transcripts of each of these mRNAs, the difference in kinetics of mRNA degradation may reflect that the Mos mRNA is more accessible to be targeted for destruction. It should be borne in mind, however, that estimate of the number of transcripts is relatively crude, and hence the difference in kinetics of mRNA degradation may reflect differences in transcript abundance, i.e., there is less Mos mRNA than Plat mRNA.

Both Mos and Plat antisense RNA are also effective in reducing the amount of endogenous Mos and Plat mRNA, respectively. Nevertheless, on a molar basis, the antisense RNA is not as effective as dsRNA. For example, Mos dsRNA more effectively inhibits the activation of MAP kinase when compared with Mos antisense RNA; parthenogenetic activation and the concomitant reduction in histone H1 kinase activity are only observed in Mos dsRNA-injected eggs, and not in Mos antisense RNA-injected eggs. This suggests in turn that MAP kinase activity must be reduced below a threshold level at which MAP kinase activity is almost absent, in order to make the eggs exit metaphase II arrest and enter interphase. Moreover, when the amount of injected Mos dsRNA and antisense RNA are reduced 10-fold, Mos dsRNA is still highly effective in inhibiting the increase in MAP kinase activity whereas Mos antisense RNA is essentially ineffective. Thus, Mos dsRNA is more efficient than Mos antisense RNA. A similar situation is also found with Plat antisense and dsRNA. Injection of 10⁶ molecules of Plat antisense or dsRNA results in both destroying the Plat mRNA and inhibiting the increase in tPA activity that accompanies oocyte maturation. In contrast, injection of 10⁵ molecules of Plat antisense RNA results in little reduction of the endogenous mRNA and little inhibition in the increase in tPA activity, while Plat dsRNA still results in the reduction of the endogenous mRNA and inhibition of the appearance of tPA activity.

The increased potency of dsRNA when compared with antisense RNA could, in principle, reflect differences in their stability, i.e., dsRNA is more stable than antisense RNA. This possibility is minimized by the observation that in a Drosophila cell lysate that supports RNAi-mediated mRNA destruction both capped antisense and capped dsRNA are stable but only the capped dsRNA is active (Tuschl et al., 1999). Moreover, results of recent experiments suggest that processing the dsRNA to discrete 20-25 nucleotide fragments is part of the mechanism that leads to destruction of the targeted mRNA (Hammond et al., 2000; Zamore et al., 2000). In fact, asRNA can give rise in an in vitro system to small amounts of stable 20-25 nucleotide fragments (Zamore et al., 2000). This could account for the activity, albeit reduced, of antisense RNA, relative to dsRNA.

dsRNAs in mammalian cells typically activate protein kinase PKR that phosphorylates and inactivates eIF2a (Fire, 1999). The ensuing inhibition of protein synthesis ultimately results in apoptosis. This sequence-independent response may reflect a form of primitive immune response, since the presence of dsRNA is a common feature of many viral lifecycles. Mouse oocytes, however, clearly lack this response, as oocyte maturation beyond germinal vesicle breakdown requires protein synthesis (Wassarman et al., 1976), which probably reflects a requirement for cyclin B synthesis, and the oocytes injected with dsRNAs resume meiosis and mature to metaphase II. Preimplantation mouse embryos also lack the response, as embryos injected with dsRNAs develop to the blastocyst stage (Wianny and Zernicka-Goetz, 2000). When the embryo acquires this response is unknown. It is not known if oocytes and preimplantation embryos contain PKR activity, which could account for the lack of the response. The lack of this response, however, cannot be attributed to a deficiency in the cell death machinery, because both oocytes (Perez et al., 1999) and preimplantation embryos (Brison and Schultz, 1997; Handyside and Hunter, 1986; Pierce et al., 1989; Weil et al., 1996) can undergo apoptosis.

The lack of this response to dsRNA may confer a selective advantage by minimizing reproductive wastage. Both oocytes and preimplantation embryos are exposed to viruses. Viral exposure throughout the lifespan of the female could deplete the pool of oocytes and compromise her reproductive capacity, because oocytes do not proliferate. The preimplantation embryo is also susceptible to viral infection from viruses present in the female reproductive tract. Infection of an early cleavage stage preimplantation embryo that results in blastomere death could result in a blastocyst containing an insufficient number of inner cell mass cells, and hence be incapable of development to term.

RNAi clearly offers several advantages to the current methods that employ generation of null mutants by homologous recombination, which requires (1) generating a suitable targeting construct, (2) selecting homologous recombination events in ES cells, (3) injecting blastocysts with these ES cells, and (4) establishing pure breeding lines from the chimeric offspring. The RNAi response will also likely be far more efficient and consistent than the antisense RNA approach that has been used with very inconsistent results in the mouse oocyte and embryo. Moreover, hypomorph phenotypes may become manifest, as RNAi does not appear to result in the total ablation of the targeted mRNA. Such hypomorph phenotypes may be as informative (or more informative) than the corresponding null mutation by providing novel insights into the presence of thresholds and/or the function of a gene. For example, as described above, a critical amount of MOS activity appears required for the development of a level of MAP kinase activity that is sufficient to maintain metaphase II arrest, a result consistent with a recently proposed switch mechanism for MAP kinase activation, as well as other cellular switches (Ferrell, 1999). In addition, modest changes in the levels of Oct4 (Pou5f1 – Mouse Genome Informatics) expression may also function as a developmental switch by regulating the fate of embryonic stem cells, e.g., high levels lead to differentiation into primitive endoderm and mesoderm, intermediate levels lead to pluripotent stem cells and reduced levels result in trophectoderm (Niwa et al., 2000).

As more dormant maternal mRNAs are identified, RNAi will be a valuable tool with which to study their function in oocyte maturation, fertilization and egg activation, and development. Moreover, the method can also be used to study the function of genes that are expressed in the early embryo, since dsRNA can inhibit the function of zygotically expressed genes (Wianny and Zernicka-Goetz, 2000; P. S., P. S. and R. M. S., unpublished). Whether dsRNA can also lead to DNA methylation of the targeted gene and result in long-term repression of transcription, as apparently occurs in plants (Wassenegger et al., 1994), is unknown.

This research was supported by a grant from the NIH (HD 22681 to R.M.S.). The authors would like to thank Johanna Mestach for help with preparing the dsRNA.