Musashi regulates the temporal order of mRNA translation during Xenopus oocyte maturation

The Mos PRE is a target of the progesterone ‘trigger' signaling pathway

To determine if the Mos PRE was a direct target of the MAP kinase-independent progesterone trigger pathway, oocytes were treated with the MAP kinase pathway inhibitor U0126. U0126 significantly attenuated the rate of oocyte maturation (a 4 h delay to GVBD₅₀; Figure 1A) and abolished progesterone-stimulated MAP kinase activation, as previously reported (Gross et al, 2000). However, the initiation of endogenous Mos mRNA polyadenylation occurred with similar kinetics in the presence or absence of U0126, although the length of the poly[A] tail extension was reduced in U0126-treated oocytes. By contrast, CPE-directed polyadenylation of the endogenous cyclin B1 mRNA was significantly delayed in the U0126-treated oocytes, occurring coincident with GVBD. As the β-globin 3′ UTR does not undergo any significant polyadenylation (Charlesworth et al, 2002, 2004), the Mos PRE was inserted into the β-globin 3′ UTR to directly determine if the PRE was sufficient to direct MAP kinase-independent polyadenylation. Initiation of PRE-directed polyadenylation of the β-globin 3′ UTR occurred with similar kinetics in the presence or absence of U0126 (Figure 1A), although some differences in overall length of poly[A] tail extension can be seen between the two treatments. Injection of the PKA inhibitor (PKI) has been shown to induce progesterone-independent oocyte maturation (Huchon et al, 1981; Davidson, 1986). To determine if the PRE was downstream of PKA inhibition, we analyzed PRE-directed polyadenylation following injection of PKI. PKI induced PRE-directed polyadenylation of the β-globin 3′ UTR reporter mRNA, even in the presence of U0126 (Figure 1B). These findings indicate that the Mos PRE is downstream of PKA inhibition and is a target of the MAP kinase-independent progesterone trigger signaling pathway.

Xenopus Musashi is a Mos 3′ UTR interacting protein

We next sought to identify potential RNA sequence-specific binding proteins that control PRE function. The yeast three-hybrid system (SenGupta et al, 1996, 1999; Zhang et al, 1999; Bernstein et al, 2002) was used to screen a X. laevis pACT2 unfertilized egg cDNA library (Clontech) for proteins that specifically interact with a CPE-disrupted Mos UTR (Mos M1 48; Figure 2A). The CPE sequence was disrupted to prevent possible recovery of the CPE binding protein (CPEB1) in the library screen. In total, 1.3 × 10⁷ initial yeast transformants were screened (Supplementary Table I) and two distinct cDNAs (designated A8 and C36) were recovered and shown to reconstitute specific RNA interaction with the Mos 3′ UTR (Figure 2B, M1 48). The protein encoded by the A8 cDNA interacts specifically with the Mos 3′ UTR but does not interact with the PRE sequence and will be described elsewhere (Charlesworth and MacNicol, in preparation).

The pACT2 C36 cDNA was identical to amino acids 17–347 of nervous system-specific RNP protein 1b (Nrp-1b), more recently identified as the Xenopus homolog of the Drosophila Musashi gene (Good et al, 1993, 1998). Xenopus Musashi has two RNA recognition motifs in the N-terminal region (Figure 2C) that are highly conserved (87% amino-acid identity) with the mammalian Musashi1 protein. The Xenopus Musashi protein localizes to the cytoplasm of the oocyte (Good et al, 1993), consistent with a potential role in the regulation of cytoplasmic mRNA translation.

Musashi binds specifically to the PRE in the Mos 3′ UTR

The 24-nt Xenopus Mos PRE (Charlesworth et al, 2002) contained a match to the SELEX-derived murine Musashi RNA binding consensus sequence (G/AU₁₋₃AGU) (Imai et al, 2001), and included a 3′ U residue essential for PRE function (Charlesworth et al, 2002) (Figure 3A, white nucleotides). Electrophoretic mobility shift assays (EMSAs) were performed to verify that Musashi interacts specifically with the Mos 3′ UTR. GST-tagged N-terminal Musashi (GST N-Msi), but not the GST moiety alone or unprogrammed reticulocyte lysate, formed specific complexes with the wild-type Mos UTR probe (Figure 3B, left panel). Musashi did not form a complex with the Mos UTR when the Musashi binding site was disrupted (AUAGU to AUccU; Figure 3B, right panel). Moreover, the Musashi-specific complexes formed with GST N-Msi could be supershifted with GST antisera, but not GFP antisera (Figure 3B, right panel). To delineate the Musashi target sequence within the PRE, we performed a competition experiment using 50-fold molar excess of unlabeled wild-type Mos UTR probe or unlabeled mutant Mos UTRs (Figure 3C). Whereas Musashi-specific binding to the wild-type Mos UTR probe was competed by unlabeled wild-type Mos UTR, Musashi-specific complex formation was not competed with a Mos UTR carrying a disruption in the Musashi binding site (Msi mut) or a Mos UTR lacking the 24-nt PRE (ΔPRE). Two additional RNAs encoding PRE mutations 5′ of the Musashi binding site effectively competed Musashi complex formation (Figure 3C). These findings indicate that Musashi does not bind to the 5′ region of the PRE, although our data do not exclude a role for the 5′ region of the PRE in facilitating Musashi function. Musashi-specific complex formation with the wild-type Mos UTR was not observed with the RNA binding-deficient form of the truncated Musashi protein (N-Msi-bm; Figure 3D), which encodes three amino-acid substitutions in the first RRM domain (Imai et al, 2001). Taken together, these results demonstrate that Xenopus Musashi interacts specifically with the consensus Musashi binding site within the Mos PRE sequence. As an additional demonstration of specific RNA binding, the full-length Musashi (Msi) and N-Msi proteins were able to interact with the endogenous Mos mRNA but not with the IPP2 mRNA that lacks a consensus Musashi binding site in the 3′ UTR (Figure 3E). No association of either Mos or IPP2 mRNAs was detected with the GST moiety alone.

Musashi interaction with the Mos 3′ UTR is necessary for early translational activation

To determine whether the Musashi consensus site was necessary for early, CPE-independent mRNA translational activation directed by the Mos PRE (Charlesworth et al, 2002, 2004), disruptive nucleotide substitutions were introduced into the Musashi consensus sequence (AUAGU to AUccU) and the polyadenylation and translation of reporter mRNAs analyzed. Insertion of the Mos PRE into the heterologous β-globin 3′ UTR confers progesterone-dependent polyadenylation and translational activation (two- to three-fold above β-globin alone; Figure 4A), whereas disruption of the Musashi binding site (PRE/msi mt) abolished this regulation (Figure 4A). These findings indicate that the integrity of the Musashi consensus site is critical for the translational activation directed by the Mos PRE. A role for the Musashi protein in mRNA translational activation in Xenopus oocytes was independently confirmed through the use of the MS2 tethered assay (Gray et al, 2000; Minshall et al, 2001; Collier et al, 2005). As can be seen in Figure 4B, fusion of Musashi to the MS2 RNA binding domain directed a two-fold stimulation of a firefly luciferase reporter mRNA containing 3′ UTR MS2 binding sites. No induction of the luciferase reporter mRNA was observed with the MS2 RNA binding domain alone, nor with a luciferase reporter mRNA lacking MS2 binding sites (ΔMS2; Figure 4B). In these experiments, the MS2-Dazl fusion protein serves as a positive control for translational induction (Collier et al, 2005). The firefly luciferase reporter mRNAs used in the tethered assay lack a consensus polyadenylation hexanucleotide, suggesting that tethering of Musashi can direct translation independently of polyadenylation. To test this directly, the polyadenylation hexanucleotide in the GST β-globin/PRE reporter mRNA was disrupted (AAUAAA to AAgAAA; Fox et al, 1989) to generate GST β-globin/PRE/hex mt. While eliminating progesterone-dependent polyadenylation, disruption of the polyadenylation hexanucleotide reduced, but did not eliminate, PRE-directed mRNA translation (approximately 1.5-fold above β-globin alone; Figure 4C). These findings suggest that the Musashi binding site in the Mos PRE directs both polyadenylation and translation in response to progesterone, but that polyadenylation is not obligatory for PRE-directed mRNA translation.

A truncated form of Musashi blocks Xenopus oocyte meiotic cell cycle progression

We reasoned that expression of a truncated form of Musashi encompassing only the RNA binding domain (N-Msi, aa 1–199) may function in a dominant inhibitory manner to block early Mos mRNA polyadenylation and translational activation in response to progesterone stimulation. Immature oocytes were injected with mRNA encoding either N-Msi or the RNA binding-deficient N-Msi-bm (Figure 3D). Compared to the N-Msi-bm-expressing oocytes, expression of N-Msi dramatically attenuated progesterone-stimulated oocyte maturation (Figure 5B, left panel). To determine the molecular basis for the block to oocyte maturation exerted by N-Msi, oocytes were analyzed for endogenous Mos and cyclin B1 mRNA polyadenylation. In contrast to the polyadenylation of the Mos mRNA in N-Msi-bm-expressing oocytes, Mos mRNA polyadenylation was attenuated in oocytes expressing N-Msi at all time points analyzed (Figure 5B, upper right panel). The polyadenylation of the cyclin B1 mRNA is a late event in progesterone-stimulated maturation and occurs coincident with oocyte GVBD (Sheets et al, 1994; Ballantyne et al, 1997; de Moor and Richter, 1997; Charlesworth et al, 2004). In N-Msi-bm-expressing oocytes, cyclin B1 mRNA polyadenylation occurred coincident with GVBD as expected, whereas expression of N-Msi protein prevented progesterone-stimulated GVBD and ablated cyclin B1 polyadenylation, even at late time points (Figure 5B, lower right panel). Two possible explanations could account for the inhibitory effect of N-Msi on cyclin B1 polyadenylation. First, N-Msi may exert an inhibitory effect by directly binding to the cyclin B1 3′ UTR and blocking CPE-dependent cytoplasmic polyadenylation. Alternatively, the effect of N-Msi may be indirect and occur through inhibition of Musashi-dependent mRNA translation events required for GVBD and/or the activation of CPE-dependent mRNA polyadenylation. To distinguish between these possibilities, oocytes were microinjected with cyclin B1 protein to induce activation of MPF independently of progesterone stimulation (Freeman et al, 1991) and reverse the relative order of MAP kinase and MPF activation (Figure 5A; Charlesworth et al, 2002). N-Msi did not block cyclin B protein-induced maturation and the kinetics of maturation were indistinguishable between N-Msi- and N-Msi-bm-expressing oocytes (Figure 5C, left panel). Cyclin B protein induced the polyadenylation of the endogenous cyclin B1 mRNA in both N-Msi- and N-Msi-bm-expressing oocytes (Figure 5C, right panel). These results indicate that CPE-dependent polyadenylation of the cyclin B1 mRNA can occur in oocytes expressing dominant-negative Musashi if MPF is activated independently of progesterone-triggered early signaling effectors. By contrast, cyclin B-induced Mos mRNA polyadenylation was abrogated in N-Msi-, but not N-Msi-bm-, expressing oocytes (Figure 5C, right panel) demonstrating the specificity of N-Msi-mediated inhibition on PRE-regulated mRNAs irrespective of initiating signal. Mos mRNA deadenylation was observed in the N-Msi-expressing oocytes after GVBD. We conclude that the dominant-negative Musashi specifically blocks early progesterone effector pathways (including Mos mRNA translational activation) and the inhibition of CPE-dependent mRNA polyadenylation is an indirect consequence of a block to upstream Musashi-regulated mRNA translation events. Consistent with a block to Mos mRNA polyadenylation, progesterone-stimulated Mos protein accumulation was abrogated in N-Msi (but not N-Msi-bm)-expressing oocytes (Figure 5D). N-Msi, but not N-Msi-bm, also prevented progesterone-dependent activation of MPF (Figure 5E). In these experiments, the N-Msi and N-Msi-bm mutant proteins were expressed to comparable levels (Figure 5F). Although we cannot exclude additional effects of the truncated Musashi protein (N-Msi) on processes upstream of mRNA polyadenylation and translational activation, the inability of an RNA binding-deficient form of the truncated protein (N-Msi-bm) to block progesterone-stimulated cytoplasmic polyadenylation or cell cycle progression indicates that effect of the dominant inhibitory Musashi is primarily due to interaction with mRNA targets. Taken together, our data provide compelling evidence that Musashi regulates maternal mRNA translation and MPF activation during progesterone-stimulated meiotic cell cycle progression.

Musashi regulates polyadenylation of multiple Xenopus mRNAs

We next determined if Musashi could regulate additional Xenopus mRNAs. The mammalian Musashi consensus sequence (A/G)U₁₋₃AGU (Imai et al, 2001) was identified in 9.4% of Xenopus mRNA in the 3′ UTR database (Mignone et al, 2005). Musashi binding sites were found in the previously characterized early class mRNAs Mos, Aurora A/Eg2, D7 and FGF receptor 1 (Charlesworth et al, 2002, 2004), as well as in the late class cyclin B1 and cyclin A1 mRNAs. In addition, the Eg1/CDK2, Eg3, Xotch (Xenopus Notch), Bub3, and TATA binding protein 2 mRNAs were identified as having a Musashi site within the last 150 nt of the 3′ UTR. Consistent with Musashi- and PRE-directed regulation, the Eg1/CDK2, Eg3, Xotch (Xenopus Notch), Bub3 and TATA binding protein 2 mRNAs were polyadenylated in an early manner, before oocyte GVBD (Figure 6A). The length of the poly[A] tail extension varied between the different mRNAs but was not obviously correlated with position or sequence of the Musashi binding site within the 3′ UTRs. In contrast to the early mRNAs, polyadenylation of the CPE-dependent cyclin A1 mRNA occurred later, coincident with oocyte GVBD.

The progesterone-stimulated polyadenylation of the mRNAs analyzed in Figure 6A was ablated in oocytes expressing the dominant inhibitory N-Msi, but not the N-Msi-bm, protein (Figure 6B, left panels). As the PCR products of the early class mRNAs were reduced at late time points in progesterone-stimulated oocytes expressing N-Msi, our data suggest that these mRNAs become deadenylated. No apparent deadenylation of the late class cyclin A1 mRNA was observed in N-Msi-expressing oocytes. Active PRE-directed polyadenylation may normally preclude access of deadenylation factors or directly oppose deadenylation of early class mRNAs in maturing oocytes. The presence of a polyadenylation hexanucleotide-overlapping CPE, common to both the cyclin A1 and cyclin B1 3′ UTRs, may protect the late class mRNAs from deadenylation (see also Figure 5). With the exception of D7 and cyclin A1, mRNA polyadenylation was abrogated in N-Msi-, but not N-Msi-bm-, expressing oocytes injected with cyclin B protein. Cyclin B protein induced polyadenylation of the D7 and cyclin A1 mRNAs in the presence of N-Msi, suggesting that the CPE sequences in these 3′ UTRs were responsive to MPF signaling. These findings indicate that Musashi regulates the translational activation of multiple maternal mRNAs during progesterone-stimulated Xenopus oocyte meiotic cell cycle progression.

Plasmid construction

A detailed methodology of construction of all plasmids used in this study is provided as Supplementary data.

Yeast three-hybrid screen

The yeast three-hybrid screen was performed as described (SenGupta et al, 1996) in the yeast strain YBZ-1 (Bernstein et al, 2002) using the X. laevis oocyte Matchmaker cDNA Library (Clontech). Initial transformants were plated in the presence of 0.5 mM 3-AT to reduce false positives. A summary of the screen is provided in Supplementary Table 1. To investigate binding specificity by mating, the R40 coat strain, transformed with an empty vector (pIIIA MS2-2.1), a negative control plasmid (pIIIA IRE (iron response element)) or pIIIA MS2-2.1 M1 48 was used. The orientation of the Mos UTR relative to the MS2 sites in the hybrid RNA was critical for the success of the screen.

RNA electrophoretic mobility shift assays

GST fusion proteins were in vitro transcribed/translated using TNT SP6-coupled Reticulocyte Lysate System (Promega). 5′ biotin-labeled RNA oligonucleotide probes were synthesized by Integrated DNA Technologies. An 80 fmol portion of labeled probe was incubated with 1 μl of reticulocyte lysate in binding buffer (50 mM Tris pH 7.5, 20 mM KCl, 150 mM NaCl, 2 mM EGTA, 0.05% NP-40, 6 mM DTT, 8 U RNase OUT; Okabe et al, 2001) in a final volume of 20 μl. Unlabeled competitor RNA was added to 4 pmol (50-fold molar excess). The binding reaction was incubated at room temperature for 20 min and then 0.5 μl of 200 mg/ml heparin was added and incubated for a further 20 min. For supershift analysis, 1 μl of anti-GST antibody (Santa Cruz) or anti-GFP antibody (Molecular Probes) was added 10 min before the end of the incubation. A 5 μl volume of the binding reaction was run on a 6% DNA retardation gel (Invitrogen) and transferred to Biodyne B membranes (Pierce) according to the manufacturer's instructions. Biotinylated RNA was detected using Chemiluminescent Nucleic Acid Detection Module (Pierce) according to the manufacturer's directions, with the modification that incubation with the streptavidin-HRP conjugate was for 40 min. Image collection was performed using an AlphaInnotech ChemiImager.

Tethered assay

Oocytes were injected with 23 ng of appropriate MS2 fusion protein mRNA, 0.1 ng of firefly luciferase mRNA, 0.35 pg of Renilla luciferase control mRNA and incubated for 20 h. Three pools of five oocytes were harvested for each experimental point and lysed in 50 μl of Passive Lysis Buffer (Promega) per oocyte. A 10 μl portion of lysate was analyzed for Renilla and firefly luciferase activity using the Dual-Luciferase Assay System (Promega) on a TD-20/20 Turner Designs luminometer. Mean values and standard deviation were determined for each experimental point, with the ratio of firefly to Renilla luciferase normalized to 1.0 for the MS2 protein. MS2 fusion protein expression was comparable as assessed by Western blot (MS2 antibodies generously provided by Mike Kiledjian).

Oocyte isolation, culture, microinjection and lysate preparation

Xenopus oocyte isolation and culture has been described (Machaca and Haun, 2002). Where indicated, oocytes were pretreated with 50 μM U0126 (Promega) and 0.5% DMSO for 1 h before stimulation. Oocytes were induced to mature with 2 μg/ml progesterone, cyclin B protein (Howard et al, 1999) or by injection of approximately 0.04 U of recombinant rabbit PKI-α (Calbiochem). GVBD was used as an indicator of maturation and assessed by the presence of a white spot on the animal pole. Where indicated, oocytes were segregated at the time when 50% of the oocyte population had reached GVBD (GVBD₅₀) based on whether they had (+) or had not (−) completed GVBD. A 250 ng portion of RNA encoding GST N-Msi or GST N-Msi-bm was injected per oocyte and left for 36 h for protein expression and equilibration with endogenous Musashi. The oocyte culture medium was changed after about 18 h. RNA and protein were extracted from the same pool of oocytes as previously described (Charlesworth et al, 2002). MPF activation was assessed using phospho-CDK1/CDK2 antisera (Cell Signaling). MAP kinase activation as well as Mos and GST protein accumulation was assessed as previously described (Howard et al, 1999).

Polyadenylation assays

RNA ligation-coupled RT–PCR was performed as described previously (Charlesworth et al, 2004). To analyze polyadenylation of reporter constructs, primers to the GST region were used (Charlesworth et al, 2004). The primers used to analyze the novel mRNAs are described in Supplementary data.

mRNA co-association assay

A 25 ng portion of RNA encoding the indicated GST fusion protein was injected into immature oocytes and cultured for 24 h to allow GST fusion protein expression. Pools of 25 oocytes were lysed in 50 mM Tris–HCl (pH 7.5), 150 mM NaCl, 0.5% NP-40, 6 mM DTT, protease inhibitor cocktail (Sigma), PMSF, 10 mM ribonucleoside vanadyl complex (NEB) and 400 U/ml RNase OUT (Invitrogen) and centrifuged to remove cell debris. GST-tagged proteins were affinity purified with glutathione Sepharose for 30 min and washed 5 × 1 ml with the same buffer. RNA was extracted using STAT-60 and analyzed by RNA ligation-coupled RT–PCR.

GST reporter mRNA translation assays

Analyses of wild-type and mutant 3′ UTRs on GST reporter mRNA translation were performed using 0.2 ng reporter RNA as previously described (Charlesworth et al, 2000, 2002).