CPEB-related RNA-binding protein and localization of maternal components in the zebrafish oocyte

Mechanisms of Development 77 (1998) 31–47

Characterization of the zebrafish Orb/CPEB-related RNA-binding protein and localization of maternal components in the zebrafish oocyte Laure Bally-Cuif*, William J. Schatz, Robert K. Ho Department of Molecular Biology, Princeton University, Princeton, NJ 08544, USA Received 9 June 1998; accepted 30 June 1998

Abstract The animal/vegetal axis of the zebrafish egg is established during oogenesis, but the molecular factors responsible for its specification are unknown. As a first step towards the identification of such factors, we present here the first demonstration of asymmetrically distributed maternal mRNAs in the zebrafish oocyte. To date, we have distinguished three classes of mRNAs, characterized by the stage of oocyte maturation at which they concentrate to the future animal pole. We have further characterized one of these mRNAs, zorba, which encodes a homologue of the Drosophila Orb and Xenopus CPEB RNA-binding proteins. Zorba belongs to the group of earliest mRNAs to localize at the animal pole, where it becomes restricted to a tight subcortical crescent at stage III of oogenesis. We show that this localization is independent of microtubules and microfilaments, and that the distribution of Zorba protein parallels that of its mRNA.  1998 Elsevier Science Ireland Ltd. All rights reserved Keywords: RNA localization; Determinant; RNA-binding protein; Orb; CPEB; Zebrafish oogenesis

1. Introduction Understanding the molecular bases of axes establishment is an important challenge of developmental biology. In extensively studied model systems such as Drosophila melanogaster and Xenopus laevis, maternal factors non-ubiquitously localized within the egg have been demonstrated to play a major role in imparting asymmetries. These ‘determinants’ generally consist of mRNAs deposited or anchored in a localized fashion within the oocyte, following local production by accessory cells and/or a passive or active transport to a specific binding site (see for reviews: Micklem, 1995; Gru¨nert and St Johnston, 1996; King, 1996). These factors are responsible for setting up both the anteroposterior (AP) and dorsoventral (DV) axes in Drosophila, and the animal-vegetal (An/Vg) axis in the Xenopus oocyte. An additional step of local factor ‘activation’ is required in Xenopus embryos to specify the DV axis following cortical * Corresponding author. Present address: GSF Forschungszentrum, Institut fu¨r Sa¨ugetiergenetik, Ingolsta¨dter Landstrasse 1, D-85758 Neuherberg, Germany. Tel.: +49 89 31874204; fax: +49 89 31873099; e-mail: [email protected]

0925-4773/98/$19.00  1998 Elsevier Science Ireland Ltd. All rights reserved PII S0925-4773 (98 )0 0109-9

rotation (see for a review Gerhart et al., 1989). Different strategies have been illustrated in other systems. For instance in the Caenorhabditis elegans egg, no localized mRNAs have so far been isolated. Rather, polarities seem established by a later process, relying on the localized degradation or translational control of some ubiquitous mRNAs and on subsequent inductive cell–cell interactions after the first embryonic cleavage divisions (Goldstein et al., 1993; Goldstein and Hird, 1996; Seydoux et al., 1996). It remains unclear at present how widely these different strategies are used within the animal kingdom. As an entry point towards a comparative analysis of axes specification strategies in vertebrates, we have turned to the teleost zebrafish, Danio rerio. In zebrafish, the An/Vg axis is established during oogenesis as it can be observed at laying by the asymmetrical (animal) location of the germinal vesicle (the oocyte nucleus) and of the micropyle (Wacker et al., 1994), a specialized structure of the vitelline envelope allowing sperm entry (Hart et al., 1992; Hart and Donovan, 1983). After fertilization, the egg cytoplasm (endoplasm) will converge to the animal pole, forming a yolk-free blastodisc where meroblastic divisions occur and which will give rise to the embryo proper (see Driever,

32

L. Bally-Cuif et al. / Mechanisms of Development 77 (1998) 31–47

1995; Kimmel et al., 1995 for reviews). The DV axis becomes morphologically visible much later (Schmitz and Campos-Ortega, 1994). However, lithium, UV- and coldsensitive periods (Stachel et al., 1993; Stra¨hle and Jesuthasan, 1993; Jesuthasan and Stra¨hle, 1997) in zebrafish and other teleosts (Oppenheimer, 1936; Tung et al., 1945; Kostomarova, 1969) suggest that the DV axis may also be determined prior to the onset of zygotic transcription (or midblastula transition, 1024-cell stage). Together, these observations suggest that both An/Vg and DV axes in zebrafish are established under the control of maternal factors. As a first step towards the identification of such factors, we initiated a search for zebrafish homologues of molecules involved in the asymmetrical localization, activation or repression, of maternal mRNAs in Drosophila and Xenopus. In fact, these species seem to use conserved determinants only in rare instances (Mosquera et al., 1993), but they share common strategies for the localization or translational regulation of these determinants. In particular, specific RNAand/or cytoskeleton-binding proteins are involved in both species (see for reviews Micklem, 1995; Hesketh, 1996; King, 1996; Macdonald and Smibert, 1996). Here we report the cloning and expression analysis of the zebrafish homologue of Drosophila orb and Xenopus CPEB. orb encodes a maternal RNA-binding protein of the RRM (RNA recognition motif) family which is required at multiple steps of Drosophila oogenesis (Lantz et al., 1992). The Orb protein is present from the earliest stages of germ line development, and first accumulates in the developing oocyte shortly after the formation of the 16-cell cyst. It is later redistributed to the posterior, and subsequently to the anterior, pole of the oocyte (Christerson and McKearin, 1994; Lantz et al., 1994). The most severe orb mutations, orbdec and orbF343, likely protein-nulls, block ovarian development at the 8-cell cyst stage (Lantz et al., 1994). Mutants retaining weak residual orb function, orbF303, are blocked later as they fail to properly differentiate an oocyte within the 16cell egg chamber. Finally, the least affected orb mutants orbmel differentiate an oocyte; however, it is abnormally polarized along both the AP and DV axes (Christerson and McKearin, 1994). Maternal mRNAs such as Bicaudal-D, fs(1)K10, oskar and gurken, involved in establishing these polarities, are mislocalized in the orb mutants (Christerson and McKearin, 1994; Lantz et al., 1994). Together with the distribution of Orb protein in the egg chamber during the course of oogenesis, these phenotypes are consistent with the idea that Orb may be involved in the transport and/or anchoring of specific maternal mRNAs in the oocyte. In parallel to these studies, a maternal Xenopus protein possessing a very similar RNA-binding domain, CPEB, was isolated for its capacity to bind the maturation-type cytoplasmic polyadenylation element (CPE) (Hake and Richter, 1994). The maturation-type CPE is a short U-rich nucleotide stretch (of consensus sequence UUUUAAU) present in the 3′UTR of several mRNAs involved in Xenopus oocyte maturation. The CPE promotes

the cytoplasmic polyadenylation and resultant translational activation of these mRNAs (see Macdonald and Smibert, 1996; Richter, 1996; Hake and Richter, 1997; StebbinsBoaz and Richter, 1997 for reviews). CPEB was in particular proven to be necessary for the polyadenylation of the cmos mRNA and, consequently, for Xenopus oocyte maturation (Stebbins-Boaz et al., 1996). Based on structural homology, we have cloned zorba, the potential zebrafish homologue of orb/CPEB. We demonstrate here that zorba mRNA and protein are maternal and become restricted to a subcortical crescent at the future animal pole during early stages of oocyte maturation. Our results suggest a localized function for the maternal RNAbinding protein Zorba in zebrafish, possibly related to the specification of the animal pole. In addition, we characterized the distribution of other maternal mRNAs in the zebrafish oocyte. Surprisingly, our comparison highlights the existence of several classes of maternal mRNAs redistributing to the animal region at different stages of oocyte maturation. These findings demonstrate that the distribution of the different maternal mRNAs in the zebrafish oocyte is specifically regulated, and also that their accumulation at the animal pole occurs prior to the redistribution of the endoplasm, which follows fertilization.

2. Results 2.1. Cloning of zorba, a zebrafish homologue of Drosophila orb As a first step towards the identification of putative maternal determinants in zebrafish, we looked for potential upstream factors involved in the localization or in the translational regulation of maternal mRNAs. We used degenerate oligonucleotides directed against the RNA-binding domain (RNA-BD)-encoding region of Drosophila orb and Xenopus CPEB to RT–PCR amplify a single 450 bp fragment under low stringency annealing conditions (see Section 4) from pre-midblastula transition zebrafish cDNA. This fragment encoded an RNA-BD comprised of two RRM motifs, overall 62% and 90% similar to those of the Orb and Xenopus CPEB proteins, respectively. The fragment was extended by 3′RACE PCR and the resulting 1.6 kb cDNA fragment was used to screen at high stringency a zebrafish 4–8 cell-stage cDNA library. Ten positive clones were isolated, all encoding parts of the same protein. Four clones, 2.5–3 kb in length, contained a full length 1800 bp ORF, which was preceded by 15 bp of 5′UTR containing an in-frame stop codon, and followed by 1–1.5 kb 3′UTRs differing in their polyadenylation sites (Fig. 1A). The deduced protein sequence, of calculated molecular weight 60 kDa, is aligned in Fig. 1B with Orb and Xenopus CPEB. The RNA-BD of this zebrafish protein contains the same arrangement of two closely linked RRM motifs as found in Orb and CPEB, with the same atypical RNP2 peptide in the

L. Bally-Cuif et al. / Mechanisms of Development 77 (1998) 31–47

second RRM, in which Phe residues are replaced by Arg and Tyr (Lantz et al., 1992) (Fig. 1B). The RNA-BD is followed by a short zinc-finger domain 58% and 67% identical to those of Orb and CPEB, respectively (Fig. 1B). These fea-

33

tures are characteristic of Orb and CPEB among the RRMcontaining RNA-binding proteins. While homology between the three proteins is much lower beyond these domains, significant stretches of similarity are observed in

Fig. 1. (A) Nucleotide and amino acid sequences of the zorba mRNA and protein. The START and STOP codons are in capital letters, an in-frame STOP codon upstream of the ATG is double-underlined, the polyadenylation sites of the different zorba clones sequenced are indicated by the asterisks, and the RNA-binding domain-encoding region is underlined. The 3′UTR domain aligning with that of X.CPEB (C) is in italics. (B) Sequence alignment of the Zorba, Xenopus CPEB (Hake and Richter, 1994) and Drosophila Orb (Lantz et al., 1992) proteins. The RRM motifs of the RNA-binding domains are in italics, with the characteristic RNP-1 and -2 peptides underlined. The atypical S and R residues within the second RNP2 are indicated (asterisks). The C and H residues of the single zinc finger domain located C-terminal to the RNA-BD are underlined. Short stretches of homology outside the RNA-BDs are underlined with a dotted line. (C) Sequence alignment of the 3′UTRs of zorba (position + 2518) and CPEB (position + 2638; Hake and Richter, 1994). Note, underlined, the two conserved CPEs. (D) Temporal expression of zorba, revealed by Northern blot. 10 pg of total RNA from oocytes (Oo), 64-cell (64), sphere (Sph), 50% epiboly (50%), tail bud (tb), 28-h- and 36-h-stage embryos were loaded in each lane and probed with zorba full length clone. Ribosomal RNA, visualized by ethidium bromide staining, was used as an internal standard for quantification of deposited RNA. A single 3 kb mRNA species is detected during oogenesis and early cleavage stages, but not after mid-blastula transition (MBT).

34

L. Bally-Cuif et al. / Mechanisms of Development 77 (1998) 31–47

Fig. 1b–d.

L. Bally-Cuif et al. / Mechanisms of Development 77 (1998) 31–47

35

Fig. 2. Stages of oogenesis in zebrafish (after Selman et al., 1993), showing the appearance of the first morphological animal/vegetal asymmetries: the formation of the micropylar cell at stage III, and the migration of the germinal vesicle towards the anima pole at stage IV. Diameter sizes of the oocytes, and meiotic stages, are indicated. GVBD, germinal vesicle breakdown.

the N-terminal halves of the three proteins, and C-terminal to the zinc finger domain (Fig. 1B). We thus believe that we have isolated the zebrafish orthologue of orb, hence named ‘zebrafish orb-type a’ (Zorba). The extensive similarity of Zorba and CPEB RNA-BDs further suggests that both are RNA-binding proteins of similar specificity. It is therefore interesting to note that the 3′UTRs of zorba and Xenopus CPEB align at position +2518 over a restricted 53 nt stretch comprising two closely linked potential CPE elements (see Fig. 1C), suggesting possible binding of the Zorba protein to its own mRNA (see Section 3). 2.2. zorba mRNA is strictly maternal and is asymmetrically localized in maturing zebrafish oocytes Northern blot experiments were performed to determine the time-course of zorba transcription (Fig. 1D). A single 2.5–3 kb mRNA species was detected, abundant in oocytes and decreasing in pre-mid-blastula transition embryos. No expression was detected at post-mid-blastula transition embryonic stages, at larval stages, and in various adult tissues other than ovaries (including skeletal muscle, brain and testes, not shown). Therefore, zorba is unique in that its expression is restricted to pre-mid-blastula transition stages, since all the other maternally expressed genes described to date in zebrafish also display a zygotic component during embryonic development. This suggests that zorba has a specific maternal function. The spatial distribution of zorba transcripts was determined by whole-mount in situ hybridization. At post-fertilization stages, the zorba message was uniformly distributed

within the blastomeres (not shown). The intensity of the signal then progressively decreased to become undetectable by mid-blastula transition. By contrast, localization of the zorba message followed a dynamic and non-ubiquitous profile during oogenesis (Fig. 3). Oogenesis in zebrafish proceeds through five stages of maturation, identifiable on the basis of oocyte size and of the presence or location of specific subcellular components, such as the germinal vesicle (GV) and yolk granules (Fig. 2, and Selman et al., 1993). At stage III, a micropylar follicle cell (the future site of sperm entry) becomes visible at the future animal pole (see Selman et al., 1993). The GV is located at the center of the oocyte until stage IV, when it migrates to the future animal pole of the oocyte and breaks down. These features are the first morphological asymmetries described during oocyte maturation. Strikingly, we observed that the zorba message was ubiquitous at stages I and II, but gradually accumulated in one restricted spot at the periphery of the oocyte during stage III (Fig. 3A, arrowheads). Expression was maintained in that region at stage IV (Fig. 3B), and then became delocalized at stage V to cover a much broader area presumably encompassing the future blastodisc (Fig. 3C, and see below). To determine the subcellular localization of zorba mRNA accumulation during oocyte maturation, we performed in situ hybridization on ovary sections (Fig. 3D–G). No transcripts were ever detectable in the follicle cells. Transcripts were present in oocytes of all stages, distributed ubiquitously at stages I and II (Fig. 3D), and progressively localized thereafter to a narrow subcortical domain by stage III (Fig. 3D,E). The position of this domain relative to the

36

L. Bally-Cuif et al. / Mechanisms of Development 77 (1998) 31–47

Fig. 3. Expression of zorba during zebrafish oogenesis, as revealed by in situ hybridization on whole-mount (A–C) and sectioned (D–G) oocytes. zorba mRNA is ubiquitous at stages I and II (A, D), and starts to accumulate at the future animal pole (see text) at early stage III (A, D, arrowheads). In D, note the centrally located germina vesicle, and the onset of accumulation of yolk granules, identifying stage II. The germinal vesicle is also visible as a transparent central region in the stage III oocyte shown in A. zorba mRNA is tightly tethered at the future animal pole at late stage III (E; note the central location of the germinal vesicle and the abundant yolk granules) and IV (B, F). It redistributes over the entire future blastoderm at stage V (C). The future animal pole can be morphologically identified at stage IV by the localization of the germinal vesicle (F) and of the micropyle (G). gv, germinal vesicle; m, micropyle. Scale bar = 0.4 mm.

L. Bally-Cuif et al. / Mechanisms of Development 77 (1998) 31–47

future egg axes could not be defined at this stage. However, at stage IV, this subcortical crescent overlayed the site of migration of the GV (Fig. 3F), and lay under the micropyle (Fig. 3G), showing that the position of zorba mRNA localization corresponds to the future animal pole of the egg. Low levels of expression could still be detected at stage IV in the yolk-free oocyte cytoplasm, as well as around the GV(see Fig. 3F). Our data establish that zorba transcripts are localized to the presumptive animal pole of the oocyte from maturation stage III onwards, before morphological A/P determination. This observation demonstrates the existence of an early molecular asymmetry within the zebrafish oocyte which identifies the future animal pole. The localization of the zorba message at stage III cannot be dependent on redistribution of the oocyte endoplasm, which occurs later, after fertilization (or activation) to form a protuberant blastoderm. zorba message localization must therefore rely on a different, earlier mechanism.

37

mRNAs becomes accumulated to the future animal pole at early oogenesis stages, suggesting that this localization is a highly regulated process.

2.3. mRNA localization in the zebrafish oocyte at stage III is a selective phenomenon We wished to determine whether the localization of zorba transcripts reflected a general phenomenon (i.e. occurred with all maternal mRNAs at stage III of oocyte maturation) or was a selective process (i.e. affected only a subset of these maternal mRNAs). We therefore compared the localization of zorba mRNA with that of seven other maternal mRNAs, using ISH on adjacent ovary sections (Fig. 4). The maternally expressed zebrafish genes tested were the following: Vg1 (Helde and Grunwald, 1993), Notch (Bierkamp and Campos-Ortega, 1993), Taram-A (Renucci et al., 1996), Sox19 (Vriz and Lovell-Badge, 1995), gsc (Stachel et al., 1993), PABP (polyA-binding protein gene, L.B-C and R.K.H., unpublished data), and a maternal ActRIIB (L.BC and R.K.H., unpublished data). Overall, we observed from this sampling that maternal mRNAs could be subdivided into three categories: (A) mRNAs which show the same early localization as zorba at stage III (namely: Notch, PABP, and Taram-A), (B) mRNAs which become concentrated in the same restricted animal region slightly later, at early stage IV (namely: Vg1), and, finally, (C) mRNAs which remain ubiquitous at these stages and do not appear to follow any specific localization during oocyte maturation (namely: ActRIIB, gsc, and Sox19). All mRNAs will, however, localize to a very broad region in the animal hemisphere at stage V, corresponding to all or most of the future blastoderms (see zorba on Fig. 3C); mRNAs of categories A and B are thus delocalized between stages IV and V. As an example, Fig. 4 shows the localization of zorba and Notch, and the non-localization of Vg1, gsc and ActRIIB mRNAs at stage III; and Fig. 5 shows the localization of zorba and Vg1 mRNAs to the future animal pole at stage IV (see also Fig. 6). Therefore, it appears that only a subset of maternal

Fig. 4. Comparison of the distribution of zorba (A, C, E, G), Notch (B), Vg1 (D), gsc (F) and ActRIIB (H) mRNAs during zebrafish oogenesis. Adjacent sections of the same stage III oocyte are shown on each line (note, when visible, the central location of the germinal vesicle). At stage III, zorba mRNA accumulates at the future animal pole (arrowhead), Notch mRNA localizes like zorba, most of Vg1 mRNA is still ubiquitous (arrows), gsc mRNA is mostly found subcortically but around the whole oocyte (arrows), and ActRIIB mRNA is mostly central and surrounds the germinal vesicle (not visible on H). Vg1 mRNA will redistribute to the animal pole between stages III and IV (see Fig. 5), and gsc and ActRIIB mRNAs only at stage V and in a loose fashion. All mRNAs show the same profile at this stage (see for instance zorba in Fig. 3C). See text for more examples. gv, germinal vesicle. Scale bar = 0.4 mm.

38

L. Bally-Cuif et al. / Mechanisms of Development 77 (1998) 31–47

Fig. 5. Localization of Vg1 mRNA at the future animal pole of a stage IV oocyte. Ovary sections were hybridized with the zorba (A) or Vg1 (B) probes, and stage IV oocytes were identified by the animal location of the germinal vesicle (gv). Staining of adjacent sections for either mRNA demonstrates that Vg1 mRNA accumulates in the same restricted animal domain as zorba mRNA from stage IV onwards (arrowheads). Scale bar = 0.4 mm.

2.4. Localization of zorba and Vg1 mRNAs to the animal pole is independent of microtubules and microfilaments The non-ubiquitous localization of maternal mRNAs in Xenopus and Drosophila oocytes has been shown to rely on cytoskeletal elements, such as microtubules and microfilaments. For instance, it was shown that Xenopus Vg1 mRNA was transported to the vegetal pole by a microtubule-dependent process, and was then anchored to actin microfilaments present at the vegetal cortex (Yisraeli et al., 1990). Using immunocytochemistry, we could reveal microtubules in the subcortical domain of zebrafish oocytes from stage III onwards (see Fig. 7F), thus extending previous reports (see Hart and Fluck, 1995 for a review). In addition, we used phalloidin-FITC labelling and detected microfilaments immediately underlying the oocyte vitelline membrane from stage II onwards (see Fig. 7H). To determine whether such cytoskeletal elements were involved in localizing zorba and/or Vg1 mRNAs to the animal pole during zebrafish oocyte maturation, we applied microtubule- or microfilament-depolymerizing drugs to oocytes cultured in vitro from early stage III onwards, and assessed the distribution of zorba and Vg1 mRNAs in these oocytes by whole-mount ISH. Maturation stages were estimated at the time of culture by oocyte size, the presence of vitellus and the location of the germinal vesicle, as reported in Selman et al. (1993). Oocytes were placed in culture at early (diameter 0.35–0.5 mm) or late (0.5–0.7 mm) stage III. As a test for our culture conditions, the maturation of late stage III oocytes to stage IV was monitored in the absence of drug treatment. After 5 h of culture, 23% of late stage III oocytes (n = 148) had matured to stage IV (as assessed by their translucent appearance, and by the disappearance of the germinal vesicle), a proportion similar to that reported by Selman et al. (1994). This proportion was brought to 97% (n = 54) after addition of 5 pg/ml DHP hormone for 5 h,

again in agreement with the report by Selman et al. (1994). These results show that our culture conditions are compatible with at least some steps of oocyte maturation from stage III onwards. We then tested the effect of cytoskeletal drugs treatment. In all cases, a low proportion of cultured oocytes (16%, n = 562) showed no detectable expression of either zorba or Vg1 after 5 h. When identically staged oocytes were cultured in parallel after injection with capped lacZ mRNA, a similar proportion (20%, n = 30) showed no beta-galactosidase activity (not shown). We assume that these oocytes were dead at the time of culture, or did not survive the culture conditions, and they were not taken into account in the following analysis of our results. We also verified that the proportion of oocytes maturing to stage IV was unaffected by drug treatment. To test for a role of microtubules in mRNAs localization, isolated oocytes were cultured in the presence of 2 or 20 mg/ ml nocodazole for 5 h. Whole-mount immunocytochemistry using the anti-beta tubulin antibody KMX1 assessed that no polymerized microtubules were detectable after a 20 mg/ml treatment (not shown). However, this treatment did not lead to any change in zorba (n = 44) and Vg1 (n = 46) mRNAs localization at any stage compared to untreated oocytes (n = 26 and 32, respectively) (Fig. 6A–D). Therefore, microtubules do not appear to be a primary component of the mRNAs localization process. To test for a role of actin microfilaments, oocytes were treated with 2.5 or 25 mg/ml cytochalasine B for 5 h. Phalloidin–FITC staining demonstrates a gradual disruption of the subcortical actin network at these doses, with total breakdown obtained at the highest dose (not shown). The distribution of zorba or Vg1 mRNAs in oocytes cultured at early stage III was not modified by these treatments. However, zorba or Vg1 mRNAs distribution was affected when oocytes were treated with cytochalasine B at late stage IV. We observed an enlargement and an increase in the staining

L. Bally-Cuif et al. / Mechanisms of Development 77 (1998) 31–47

39

Fig. 6. Distribution of zorba (A, C, E) and Vg1 (B,D,F) mRNAs in oocytes cultured for 5 h from late stage III onwards and treated with: 20 mM nocodazole to disrupt microtubules (C, D), or 25 mM cytochalasine B to disrupt microfilaments (E, F). (A, B) mock treatment. zorba mRNA is localized in all oocytes studied (A, arrowheads); in comparison, Vg1 localization was consistently noted in only those oocytes that underwent maturation under culture conditions (B, arrowheads). Nocodazole treatments (C, D) had no effect upon the localization of either zorba or Vg1. However, cytochalasine B treatments enhanced the localization of both zorba and Vg1 mRNAs. Note in E and F that the expression domains are wider and more intensely labeled at the animal pole (bars). In addition, note that unlike the oocytes shown in B and D, all oocytes treated with cytochalasine B exhibited strong localization of Vg1; however, cytochalasine B treatments did not significantly affect maturation rates (see text for details). Scale bar = 0.4 mm.

40

L. Bally-Cuif et al. / Mechanisms of Development 77 (1998) 31–47

L. Bally-Cuif et al. / Mechanisms of Development 77 (1998) 31–47

intensity of the zorba mRNA localization spot at the animal pole in all cytochalasine B-treated oocytes cultured at this stage (n = 20, Fig. 6E). In rare instances (10%, Fig. 6E, arrows), the staining spot appeared split. Vg1 mRNA normally localized later than zorba, between stages III and IV. We observed that Vg1 mRNA, like zorba, concentrated to the animal pole in late stage III oocytes treated with cytochalasine B (localization in 28% cases, n = 18, at 2.5 mg/ml cytoB, and in 100% cases, n = 26, at 25 mg/ml cytoB; Fig. 6F). Again, the cytochalasine B-induced localization was wider than normal. After a control culture of late stage III oocytes, Vg1 mRNA was localized in only 22% of cases (n = 46, Fig. 6B), i.e in the oocytes having matured to stage IV. However, the cytochalasine B-induced localization of Vg1 was not related to oocyte maturation, as we verified that the proportion of oocytes maturing to stage IV was not affected by cytochalasine treatment. These results suggest that microfilaments are not involved in the localization of either zorba or Vg1 mRNAs at the animal pole. Rather, a likely interpretation is that the depolymerization of the actin network most probably released an internal non-localized pool of zorba and Vg1 mRNAs (visible in Figs. 3F and 4D), which subsequently accumulated at the animal pole following an actin-independent process. Alternatively, microfilaments might be necessary for the tight tethering of zorba mRNA to a restricted spot at the animal pole. 2.5. zorba mRNA localization is reflected at the protein level To test whether zorba might perform a localized function in the zebrafish oocyte, we wished to determine whether, like its mRNA, the Zorba protein was localized early during oogenesis. Therefore, we raised a polyclonal antibody against the N-terminal (non-RNA-BD) portion of the Zorba protein (see Section 4). Western blot analyses using this antibody revealed a major band of approx. 62 kDa in oocyte extracts (Fig. 7A), a size in agreement with the expected molecular weight of the Zorba protein (60 kDa) and with the size of the in vitro translated product of zorba mRNA in a reticulocyte lysate. No protein is detected by Western blot (Fig. 7A) or immunocytochemistry in postfertilization embryos and in adult tissues except ovaries

41

(not shown). However, the 62 kDa band was specifically detected in extracts of embryos which had been previously injected with capped zorba mRNA (not shown). Further confirmation that the detected protein came from the injected mRNA was obtained by whole-mount immunostaining of embryos injected with a mixture of 2 × 106 kDa fixable rhodamine dextran and capped zorba mRNA: the immunocytochemistry and rhodamine stainings were always largely overlapping (data not shown). Taken together, these results suggest that our antibody specifically recognizes the Zorba protein. The undetectable level of protein between fertilization and mid-blastula transition suggests that at this stage the zorba mRNA (see Fig. 1D) is not translationally active, and/or that the Zorba protein is rapidly degraded. We further used this antibody on ovary sections to determine the distribution of the Zorba protein during oocyte maturation. At early stage I (stage Ia, Selman et al., 1993), no or little protein is detected (Fig. 7D). It becomes detectable at stage Ib (Fig. 7C), and remains ubiquitously distributed within the oocyte cytoplasm until stage II (Fig. 7C). It then follows a gradual accumulation to the future animal pole during stage III (Fig. 7D,E), and remains localized to this pole at stage IV (Fig. 7G). No staining was observed with the pre-immune serum (not shown). Therefore, the distribution of the Zorba protein corresponds to that of its mRNA, suggesting that Zorba may perform a localized function. To determine whether Zorba might be linked to cytoskeletal elements when it becomes localized to the animal pole, the proteins from detergent-soluble and insoluble fractions of stages I–II and III–IV oocytes were isolated under high salt conditions solubilizing most of the yolk proteins. Surprisingly, Western blot analyses of these extracts demonstrate that the Zorba protein is always associated with the insoluble fraction (Fig. 7B) (which contains most of the cytoskeletal elements of the cell, as revealed by microtubule detection, not shown), suggesting that Zorba could be bound at all stages to cytoskeletal components. Double-labellings of oocyte sections with anti-Zorba and anti-beta tubulin antibodies reveal a region of overlap between the two stainings at the site of Zorba protein accumulation from stage III onwards (Fig. 7E,F). However, we were unable to copurify microtubules and Zorba by ultracentrifugation and immu-

Fig. 7. Temporal and spatial distribution of the Zorba protein. (A) Protein extract from oocytes (Oo), 1–4 cell (1–4), sphere (Sigh), 10 somite (10s) and 24-hstage embryos was resolved by electrophoresis, blotted to nitrocellulose and probed with rabbit anti-Zorba antiserum (or pre-immune serum at the oocyte stage (PI)). Zorba protein is detected in oocytes, but not at any stage after fertilization. (B) Detergent-soluble (S) and -insoluble (I) fractions of oocyte extracts at stages I–II (lanes 1–2) and III–IV (lanes 3–4) analysed by the same Western blot procedure. Zorba protein is always associated with the detergent-insoluble fraction. The detection of a minor faster migrating species of approx. 60 kDa in some instances suggests that the 62 kDa band represents a post-translationally modified form of Zorba. (C, D) Immunolocalization of the Zorba protein during oogenesis. Ovary sections were incubated with rabbit anti-Zorba antibody, later revealed with HRP. The protein is absent or in low amounts at stage Ia (D), ubiquitous at stages Ib and II (C), and tethered at the future animal pole from stage III (D, note the central location of the germinal vesicle). (E, F) Double-labelling of an ovary section with anti-Zorba (E) and anti-beta tubulin KMX-1 (F). In stage III oocytes, tubulins and Zorba distributions overlap at the site of Zorba accumulation (arrowhead). (G, H) Double-labeling of an ovary section with anti-Zorba (G) and phalloidin–FITC (H). The limits of the different oocytes are indicated by dotted white lines in G. Note that at stage IV, the Zorba protein (arrowhead) is not in immediate contact with the subcortical actin network.

42

L. Bally-Cuif et al. / Mechanisms of Development 77 (1998) 31–47

noprecipitation (data not shown); thus, whatever the stage considered, the majority of Zorba protein is unlikely to be bound to microtubules. In addition, double labellings at stage III with phalloidin–FITC and anti-Zorba show that the Zorba accumulation spot is located relatively far from the subcortical actin network (Fig. 7G,H). The subcortical accumulation of Zorba at this stage thus may not be due to binding to this actin network. The Zorba protein may accumulate at the animal pole as a result of the previous localization mRNA, ensuring a localized protein production, and/ or as a result of binding to intracellular components other than microfilaments and microtubules. In conclusion, we observed that the Zorba protein specifically accumulates at the animal pole from stage III of oocyte maturation, demonstrating that some maternal proteins can be non-ubiquitously distributed in the zebrafish oocyte. The Zorba protein may thus exert a localized role in the zebrafish oocyte.

3. Discussion 3.1. The zorba gene of zebrafish encodes an RNA-binding protein The zorba gene encodes a protein of 600 amino acids which shows significant homology to Drosophila Orb (40% overall identity) (Lantz et al., 1992) and Xenopus CPEB, and the recently identified mouse CPEB (67% overall identity) (Hake and Richter, 1994; Gebauer and Richter, 1996). In particular, two atypical RRM motifs, closely linked to a single zinc finger, are present in the C-terminal half of all four proteins. The N-terminal halves are much more divergent; however, they show two short stretches of significant homology. This structural conservation strongly suggests that the orb, CPEBs and zorba genes belong to the same family. Suggestive evidence for the existence of a single gene in Drosophila, zebrafish (this report) and mouse (Gebauer and Richter, 1996) suggests that orb, CPEB and zorba are orthologous genes. In support of this conclusion, these genes share expression characteristics as all four are transcribed during oocyte development. The RNA-binding domain has been demonstrated to be one of the diagnostic determinants of the binding specificity of RNA-binding proteins (see Biamonti and Riva, 1994 for a review). The Xenopus and mouse CPEB proteins can recognize the same RNA target in vitro, an A/U rich nucleotide sequence designated as CPE element (Gebauer and Richter, 1996). The similarity between the RNA-binding domains of Zorba and the CPEBs thus strongly suggests that Zorba also recognizes the CPE. Accessory domains of RNA-binding proteins have been proposed to help form supramolecular structures via protein–protein or protein–RNA interactions, thereby modulating the functional specificity of the RNAbinding proteins (see Biamonti and Riva, 1994). In spite of the fact that Zorba, Orb and the CPEBs probably bind simi-

lar target sequences, the divergence of their N-terminal domains may therefore introduce important functional differences between these proteins. Xenopus CPEB has been isolated for its role in the regulation of polyadenylation (Hake and Richter, 1994), and Orb has been hypothesized to play a role in RNA localization (Christerson and McKearin, 1994; Lantz et al., 1994). Whether both proteins share both functions is under investigation. The N-terminal domain of Zorba does not possess any typical motif of classical RNA-binding protein-accessory domains, rendering it difficult to predict its biochemical function. It is, however, more closely related to the N-terminus of the CPEBs than to that of Orb, suggesting that Zorba may also be involved in the control of translation. 3.2. zorba mRNA and protein become localized at early oogenesis stages We found that the distribution of zorba mRNA and protein was biphasic during oocyte maturation. An initial phase of ubiquitous distribution until stage II is followed by a phase of sharp localization from stage III onwards. This is reminiscent of the redistribution of orb mRNA and protein during Drosophila oogenesis, but differs from the non-localization of mCPEB mRNA in the mouse oocyte (Gebauer and Richter, 1996). The distinct phenotypes of the different orb mutant alleles show that Orb has several functions, probably associated with the different phases of Orb localization (Christerson and McKearin, 1994; Lantz et al., 1994). By analogy, it is tempting to speculate that zorba might successively exert different roles in the zebrafish oocyte. The initial ubiquitous phase of zorba expression might correspond to a general role in oocyte development, via the translational regulation of mRNAs necessary for oocyte survival or maturation. Specific oocyte mRNAs whose polyadenylation is controlled by CPEB have been identified in Xenopus (Hake and Richter, 1994; Stebbins-Boaz et al., 1996). CPEB was proven to be crucial for oocyte maturation: immunodepletion of CPEB from Xenopus eggs by the injection of anti-CPEB antibody prevents c-mos polyadenylation and meiotic maturation (Stebbins-Boaz et al., 1996). The second, localized phase of zorba expression might reflect a role in oocyte polarization. zorba belongs to the group of earliest mRNAs to localize to the future animal pole. Localizing and/or locally regulating the translation of target mRNAs, the protein products of which are involved in animal pole specification, might be a key step in the pathway setting up the An/Vg axis. In zebrafish, the animal pole is the region of the oocyte from which the embryo proper will develop. Specifying the animal pole therefore controls the choice between embryonic versus yolk fate. This may be reminiscent of the early role of Orb in Drosophila, when it controls the choice of cell fate between oocyte and nurse cells within the egg chamber (Lantz et al., 1994).

L. Bally-Cuif et al. / Mechanisms of Development 77 (1998) 31–47

Interestingly, the 3′UTRs of zorba and Xenopus CPEB align with 80% identity over a restricted 53 nt stretch including two potential CPEs. Such conservation suggests functional significance, and this domain may mediate the binding of zorba to its own mRNA, thus ensuring its localization or local translational regulation. Production and localization of the Zorba protein would then be reinforced via an autoregulatory loop. Positive autoregulation might also occur in Xenopus. Since no CPE is found in the 3′UTR of mouse CPEB (see discussion in Gebauer and Richter, 1996) and in the fragment of orb 3′UTR necessary to recapitulate orb mRNA distribution in transgenic flies (Lantz et al., 1994), the potential autoregulatory function of members of the Orb/CPEB family might be species-specific. A second potential candidate for Zorba binding is Vg1 mRNA, whose localization to the animal pole slightly follows that of Zorba. The 3′UTR of Vg1 possesses a potential CPE (UUUUUAUU) 14 nt upstream of the polyA addition site (see Fig. 1A in Helde and Grunwald, 1993). Several proteins binding the Vg1 mRNA have been isolated in Xenopus, although none of them appears to have the same molecular weight as full length CPEB (Elisha et al., 1995; Deshler et al., 1997; Mowry, 1997). 3.3. Maternal mRNAs distribution in the zebrafish oocyte is precisely regulated A major outcome of our study is the finding that maternal mRNAs are differentially distributed within the zebrafish oocyte. Some become tightly tethered at the future animal pole as soon as stage III, others between stages III and IV, and a last pool redistributes more loosely to the blastodermal region at stage V. With the unique exception of Vg1, no localization studies of maternal mRNAs had previously been performed in zebrafish. Vg1 mRNA was reported to be ubiquitous in early embryos (Helde and Grunwald, 1993); however, we have extended these studies and we have observed a definite localization of Vg1 to the future animal pole of oocytes. Our finding has several important implications. (i) It demonstrates that the zebrafish oocyte is a polarized cell from much earlier than previously thought. We observed a micropyle, the earliest asymmetry so far described (yet not in the oocyte proper but in the follicle cell layer) on oocytes larger than 0.5 mm in our preparations (see also Hart and Donovan, 1983; Hart et al., 1992). We have been able to now demonstrate the position-specific expression of early localizing mRNAs in younger, 0.3 mm oocytes. (ii) It shows that the redistribution of most maternal mRNAs to the blastoderm probably occurs before, and therefore does not depend on, the redistribution of the endoplasm during and after fertilization. Indeed, all mRNAs tested were localized to the blastoderm at stage V of oogenesis. This suggests that mRNA localization to the embryonic region is not simply a by-product of cytoplasmic streaming. Experimental manipulations of the zebrafish egg have suggested that factors required for

43

DV axis formation are brought from the yolk cell into the blastoderm at the 32-cell stage by cytoplasmic streaming (Jesuthasan and Stra¨hle, 1997). If these factors are mRNAs, they would have to belong to a restricted population with vastly different localization properties to those we have surveyed in this paper. (iii) The localization of some mRNAs and not others suggests that the localization processes are regulated and depend on specific RNA sequences. (iv) The mRNAs which localize early during oogenesis suggest candidate factors and mechanisms for the pathway(s) involved in animal pole specification. From our sampling, these appear rather numerous, but the proportions of localized versus unlocalized mRNAs may turn out to be different once the study is extended to more mRNA species. An important issue raised by our observations is the identity of the factors regulating the differential mRNA localization. All the mRNAs that we studied had an initial ubiquitous distribution, and were never observed in the follicle cells. This suggests that they were transcribed by the oocyte itself, and later localized by an internal process, such as regional stability, regional entrapment, or active transport. The two latter processes have been illustrated in both Xenopus and Drosophila. In Xenopus oocytes, early localized mRNAs such as Xcat2, are primarily trapped in the mitochondrial cloud, and translocate to the vegetal region via the METRO pathway, independently of microtubules and microfilaments (Kloc and Etkin, 1995; Forristall et al., 1995; Kloc et al., 1996). mRNAs localizing later, such as Vg1, are actively transported along microtubules and become anchored to microfilaments at the vegetal cortex (Yisraeli et al., 1990; Forristall et al., 1995; Kloc and Etkin, 1995). In early Drosophila oocytes, a number of mRNAs accumulate along the posterior or anterior cortex in an active movement involving microtubules and probably microtubule motors (see Gru¨nert and St Johnston, 1996). mRNAs localizing later (such as nanos, or oskar after stage 10), i.e. after the dismantling of the microtubule network and the initiation of cytoplasmic streaming, are most likely trapped on a posterior pole anchor (see St Johnston, 1995; Glotzer et al., 1997). Our results following treatments with cytoskeletal drugs in zebrafish suggest that zorba and Vg1 mRNAs are localized independently of the microfilament and microtubule networks. Further, localization is enhanced when microfilaments are depolymerized, suggesting that an internal pool of these mRNAs has been released and accumulates at the animal pole. It is therefore most probable that zorba and Vg1 localizations follow a trapping mechanism. Cytoskeletal elements other than microtubules and microfilaments, or binding to other subcellular components or proteins, may explain entrapment within this specific cytoplasmic domain. Which subcellular factors and mRNA sequences may be involved, and whether these observations can be generalized to other early localizing mRNAs, are important questions which may now be addressed in the zebrafish.

44

L. Bally-Cuif et al. / Mechanisms of Development 77 (1998) 31–47

4. Experimental procedures

4.2. Northern blotting

4.1. Cloning of zebrafish orb-type a (zorba)

Total RNA was extracted according to Chomczynski and Sacchi (1987), electrophoresed (10 mg per well) and transferred onto nitrocellulose membranes (Amersham Hybond N). Hybridization with 106 cpm/ml random prime-32P-labelled zorba full length probe was performed overnight at 42°C in 50% formamide, 50 mN sodium pyrophosphate, 5× SSC, 5× Denhardt’s, 0.5% SDS and 200 mg/ml salmon sperm DNA. The membrane was rinsed at 65°C in 2× SSC, 0.1% SDS, and autoradiographed.

4.1.1. Degenerate RT –PCR and 3′RACE PCR cloning A 450 bp cDNA fragment encompassing most of the zorba RNA-binding domain-encoding region was PCRamplified from pre-mid-blastula transition (,250-cell) zebrafish embryo mRNA using degenerate oligonucleotides directed against the RNA-binding domains of D.orb (Lantz et al., 1992) and Xenopus CPEB (Hake and Richter, 1994). Sequences of the oligonucleotides were as follows: FP 5′ AA(A/G)GTITT(T/C)(T/C)TIGGIGGI(A/G)TICCITGGGA 3′ (D.orb amino acids 577–586), RP1 5′ TGCCAITGCCA(G/A)CAI(C/G)(A/T)IC(TM)(G/A)CA(G/A)AA(G/A)TA 3′ (D.orb amino acids 793–802), RP2 5′ AIGG(G/A)TCIA(T/C)(T/C)TGIAC(T/C)TT(T/C)TTIGT (G/A)AA 3′ (D.orb amino acids 758–768). After reverse transcription from random primers (MMLV Superscript Reverse Transcriptase, Amersham), two rounds of PCR were performed with 100 pmol FP and RP1 primers, then 100 pmol FP and RP2 primers, and the following cycles: 1 min 94°C, 1 min 42°C, 1 min 72°C (2 cycles) and 1 min 94°C, 1 min 48°C, 1 min 72°C (28 cycles). The resulting 450 bp cDNA fragment was sequenced and an internal 5′ non-degenerate oligonucleotide was designed to increase the fragment length by 3′RACE PCR following the method of Frohman (1993). The resulting 1.6 kb fragment, encompassing the 3′ coding and non-coding zorba cDNA, was ligated to the zorba 450 bp RNA-binding domain-encoding fragment using an internal restriction site, and the resulting partial cDNA (1.6 kb) was used for library screening. 4.1.2. Construction and screening of a zebrafish pre-MBT cDNA library PolyA + RNA from 4–8-cell stage zebrafish embryos was prepared using the Fast Track kit (Promega) according to the manufacturer’s instructions. cDNA was synthesized (M-MLV reverse transcriptase, Gibco) from 5 mg of polyA + RNA using a mixture of 1 mg oligo dT primer and 100 ng random hexamers in the first strand synthesis reaction. After ligation to EcoRI adaptors (Novagen), the resulting cDNA mixture was enriched in fragments above 500 bp and ligated to lEX LoxTM arms (Novagen). 150 000 phage plaques of the non-amplified library were screened at high stringency using the zorba probe (see above) labelled by random priming (Boehringer Mannheim kit). Hybridization was performed overnight at 42°C in 50% formamide, 5 × SSC, 5 × Denhardt’s, 0.5% SDS and 200 mg/ml salmon sperm DNA, followed by rinses in 0.2 × SSC, 0.1% SDS at 65°C. Ten positive clones were isolated. The plasmids were excised into the pEX Lox vector according to Novagen’s instructions, and sequenced on both strands using the Sanger method (Amersham Sequenase version 2.0 kit). Sequences were analysed with the GCG software.

4.3. In situ hybridization (ISH) 4.3.1. Probes RNA probes labelled by incorporation of digoxigenin-11UTP (Boehringer Mannheim) were synthesized from the following templates: zorba RNA-binding domain (RT–PCR fragment), zorba 3′ (1.6 kb RACE PCR fragment), and zorba full length, all subcloned in pGEM-T (Promega) or pBS SK(−); full length zebrafish Vg1 cDNA (Helde and Grunwald, 1993); full length zebrafish Notch cDNA (Bierkamp and Campos-Ortega, 1993); zebrafish Sox19 cDNA (Vriz and Lovell-Badge, 1995); full length zebrafish gsc cDNA (Stachel et al., 1993); zebrafish Taram-A cDNA (Renucci et al., 1996); zebrafish PolyA-binding protein (PABP) gene (L.B.-C. and R.K.H., unpublished data): 1.5 kb partial cDNA obtained by PCR amplification from pre-midblastula transition zebrafish RNA; zebrafish maternal ActRIIB gene (L.B.-C. and R.K.H., unpublished data): zebrafish maternal activin receptor type II, 700 bp cDNA fragment obtained from our zebrafish pre-midblastula transition library. 4.3.2. Whole-mount in situ hybridization ISH at post-fertilization stages were performed as described in Prince et al. (1998). For ISH on oocytes, ovaries were dissected out from anaesthetized females, rinsed in PBS, and follicles were manually singled out, staged (see Selman et al., 1993, and Section 2) and fixed in 4% paraformaldehyde (PFA). Stage V oocytes were obtained by squeezing females. After 1 h fixation at room temperature, the oocytes were manually dechorionated using forceps under the stereomicroscope. Oocytes were postfixed in 4% PFA overnight at 4°C, dehydrated in methanol and processed for ISH as above. Following ISH, oocytes or embryos were cleared and mounted in 80% glycerol between bridged coverslips. They were photographed under bright field illumination using an Axioskop photomicroscope (Zeiss).

L. Bally-Cuif et al. / Mechanisms of Development 77 (1998) 31–47

4.3.3. In situ hybridization on paraffin sections Ovaries were fixed in 4% PFA overnight at 4°C, then dehydrated and embedded in Paraplast Plus. 7.5-mm microtome sections were processed for ISH as described in Johnston et al. (1997). The sections were mounted in Cytoseal. They were viewed and photographed in a Zeiss Axioskon. 4.4. Raising and purification of polyclonal anti-Zorba antibodies A His tag-zorba fusion construct was obtained by subcloning the N-terminal kb PstI fragment of zorba full length in frame into pQE31 (Qiagen). The fusion protein was produced in Escherichia coli by IPTG induction according to standard protocols (Qiagen), purified on nickel-agarose columns (Qiagen), dialysed to 150 mM NaCl/10 mM Tris– HCl, pH 7.5, and resuspended in PBS/0.1% SDS. Rabbits were boosted with 250 mg of protein in Freund’s adjuvant every month and bled at regular intervals. Antibody titre was checked on Western blot against zebrafish oocyte extracts (see below). One batch showing high titre was then affinity-purified against the His tagged-Zorba protein bound to nickel/agarose columns according to Gu et al. (1994) and used in Western and immunocytochemistry studies. 4.5. Western blotting analyses The following antibodies were used: affinity-purified polyclonal anti-Zorba diluted to 1/200, and the mAb antibeta tubulin KMX1 (Sigma) diluted to 1/50. 4.5.1. Time course analysis of Zorba expression Extracts of oocytes or embryo were obtained by lysis at 4°C in 250 mM sucrose, 100 mM NaCl, 2.5 mM MgCl2, 10 mM EGTA, 20 mM HEPES, 1% Triton X-100, 10 mM NaF, 1 mM Na3VO4, 10 mg/ml leupeptin, 1 mg/ml aprotinin, 1 mg/ ml pepstatin A, 1mg/ml antipain, 20 mg/ml phenylmethylsulfonyl fluoride. The extracts were then centrifuged at 12 000 rpm for 10 min at 4°C to eliminate debris and yolk granules, and the supernatant was collected. Samples were loaded on a 12% acrylamide–bisacrylamide urea gel at 0.5– 1 oocyte/embryo per lane. Nitrocellulose membrane transfer and antibody incubation were performed according to standard protocols, and the primary antibodies were revealed using HRP-coupled secondary antibodies (Jackson Laboratories) diluted to 1/2000, and enhanced chemiluminescence (Amersham). 4.5.2. Oocyte detergent soluble and insoluble fractions Oocyte detergent soluble and insoluble fractions were obtained as described in Yisraeli et al. (1990). Oocytes were singled out in PBS from whole ovaries, staged, and lysed in cold extraction buffer (0.5% Triton X-100, 10 mM PIPES, pH 6.8, 0.3 M KCl, 10 mM MgAc, 0.5 mM EGTA). Extracts were spun at 4°C for 20 min in a microfuge. The

45

detergent soluble fraction (DSF, supernatant) was recovered and diluted in 1/3 volume of 4 × SDS sample buffer. The detergent insoluble fraction (DIS, pellet) was resuspended in an equal final volume of 1 × SDS sample buffer. The same amount of each fraction, i.e. 0.5–1 oocyte per lane, was loaded on a 12% acrylamide–bisacrylamide urea gel and processed for Western blotting as described above. A parallel gel was dried and Coomassie-stained to ensure that the lysis procedure had solubilized most yolk proteins (not shown). 4.6. Immunocytochemistry and phalloidin labelling 4.6.1. Immunocytochemistry on sections Oocytes or ovaries were fresh frozen at −50°C in isopentane in liquid nitrogen and cryostat sectioned at 12.5 mm. The sections were rinsed in PT (PBS, 0.025% Triton X100), fixed in 4% paraformaldehyde for 30 min at room temperature (RT) and bleached in 6% H2O2 for 1 h at RT. They were incubated in 10 mg/ml proteinase K in PT for 2 min at RT, then in 2 mg/ml glycine for 5 min at RT, and finally blocked in PTBN (2 mg/ml bovine serum albumin (BSA), 2% normal goat serum (NGS) in PT) for 1 h at RT. The primary antibodies (anti-zorba 1/ 200, KMX-1 1/50) diluted in PTBN were applied overnight at 4°C. Secondary antibodies (goat anti-mouse or anti-rabbit IgGs, FITC- or Cy3-coupled, Jackson) were diluted to 1/200 in PTBN and applied for 2 h at RT. The sections were rinsed in PT, dehydrated and mounted in Vectashield (Vector). 4.6.2. Phalloidin labelling Oocytes and ovary sections were prepared and fixed as described above. The sections were then processed in the dark to preserve fluorescence They were rinsed in PT and incubated in 0.1 mg/ml phalloidin–PBS (Sigma) in PBS, 1 g/ l gelatin, 0.025% Triton X-100 for 1 h at RT, rinsed in PT, blocked in PTBN for 2 h at RT and incubated in 1/200 antizorba antibody in PTBN at 4°C overnight. They were revealed using 1/200 goat anti-rabbit, Cy3-coupled secondary antibody (Jackson) as described above. 4.7. Oocyte culture and cytoskeletal drug treatments Oocytes were cultured as described in Selman et al. (1994) with minor modifications. Briefly, ovaries from anaesthetized females were immersed in 60% L-15 medium (Gibco), pH 7.5, 100 mg/ml gentamycin sulfate (Sigma) at 26°C, and follicles were singled out using fine forceps. Oocytes undergoing vitellogenesis (diameter .0.35 mm) (Selman et al., 1993) were arbitrarily separated into two groups: early stage III (0.35–0.5 mm) and late stage III (0.5–0.7 mm), and were cultured at 26°C in the above medium at a density of 15–20 oocytes per ml. After 1 h of culture, damaged oocytes were discarded and treatments were applied. Maturation of late stage III oocytes was

46

L. Bally-Cuif et al. / Mechanisms of Development 77 (1998) 31–47

induced by treatment with 5 mg/ml 17a,20b-dihydroxy-4pregnen-3-one (DHP) (Sigma) for 5 h; ethanol carrier applied in the same proportions was used as a control. Microtubule depolymerization was induced in both oocytes groups by incubation with 2 or 20 mg/ml nocodazole (Sigma) (see Gard, 1991), and microfilament depolymerization was induced by incubation with 2.5 or 25 mg/ml cytochalasine B (Sigma) (see Kloc and Etkin, 1995). Both treatments were applied for 5 h at 26°C DMSO carrier applied in the same proportions was used as a control. The oocytes were subsequently fixed in 4% paraformaldehyde overnight at 4°C, in FGT fix (80 mM KPIPES, pH 6.8, 5 mM EGTA, 1 mM MgCl2, 3.7% formaldehyde, 0.25% glutaraldehyde, 0.5 mM taxol, 0.2% Triton X-100) (Gard, 1991) overnight at 4°C, or were fresh frozen and cryostat sectioned, to be processed for whole-mount ISH, wholemount KMX-1 immunocytochemistry or phalloidin-FITC labelling, respectively.

5. Note added in proof The ActRIIB clone reported in this paper was found to be identical to a full-length zebrafish ActRIIB isolated by Garg et al. (pers. commun.). The zorba sequence reported in this paper has been attributed the following GenBank accession number: AF076918.

Acknowledgements We are indebted to Drs. E. Boncinelli, J.A. CamposOrtega, D. Grunwald, M. Halpern, F. Rosa and S. Vriz for gifts of probes, and to F. Rosa and M. Wassef for providing laboratory facilities in the last stages of this work. We would like to thank F. Rosa, A. Vincent, T. Vogt and M. Wassef for their interest and advice, L. Hake and J. Richter for discussions when we initiated this work, and K. Selman for assistance in oocyte culturing techniques. We are grateful to S. Easter, C. Howley, V. Prince, F. Rosa, A. Vincent and M. Wassef for their critical reading of the manuscript. This work was supported by a donation from the Rathmann Family Foundation to the Molecular Biology Department at Princeton University, by a Basil O’Connor Starter Scholar Research Award from the March of Dimes to R.K.H. who is a Rita Allen Foundation Scholar, and by an EMBO Long Term Fellowship and CNRS to L.B-C.

References Biamonti, G., Riva, S., 1994. New insights into the auxiliary domains of eukaryotic RNA binding proteins. FEBS Lett. 340, 1–8. Bierkamp, C., Campos-Ortega, J.A., 1993. A zebrafish homologue of the Drosophila neurogenic gene Notch and its pattern of transcription during early embryogenesis. Mech. Dev. 43, 87–100.

Chomczynski, P., Sacchi, N., 1987. Single step method of RNA isolation by guanidinium thiocyanate-phenol-chloroform extraction. Anal. Biochem. 162, 156–159. Christerson, L.B., McKearin, D.M., 1994. orb is required for anteroposterior and dorsoventral patterning during Drosophila oogenesis. Genes Dev. 8, 614–628. Deshler, J.O., Highett, M.I., Schnapp, B.J., 1997. Localization of Xenopus Vg1 mRNA by Vera protein and the endoplasmic reticulum. Science 276, 1128–1131. Driever, W., 1995. Axis formation in zebrafish. Curr. Opin. Genet. Dev. 5, 610–618. Elisha, Z., Havin, L., Ringel, I., Yisraeli, J.K., 1995. Vg1 RNA binding protein mediates the association of Vg1 RNA with microtubules in Xenopus oocytes. EMBO J. 14, 5109–5114. Forristall, C., Pondel, M., Chen, L., King, M.-L., 1995. Patterns of localization and cytoskeletal association of two vegetally localized RNAs Vg1 and Xcat2. Development 121, 201–208. Frohman, M.A., 1993. Rapid amplification of complementary DNA ends for generation of full-length complementary DNAs: thermal RACE. Methods Enzymol. 218, 340–356. Gard, D., 1991. Organization, nucleation, and acetylation of microtubules in Xenopus laevis oocytes: a study by confocal immunofluorescence microscopy. Dev. Biol. 143, 346–362. Gebauer, F., Richter, J.D., 1996. Mouse cytoplasmic polyadenylation element binding protein: an evolutionary conserved protein that interacts with the cytoplasmic polyadenylation element of c-mos mRNA. Proc. Natl. Acad. Sci. USA 93, 14602–14607. Gerhart, J., Danilchik, M., Doniach, T., Roberts, S., Rowning, B., Stewart, R., 1989. Cortical rotation of the Xenopus egg: consequences for the anteroposterior pattern of embryonic dorsal development. Development 107 (Suppl.), 37–52. Glotzer, J.B., Saffrich, R., Glotzer, M., Ephrussi, A., 1997. Cytoplasmic flows localize injected oskar RNA in Drosophila oocytes. Curr. Biol. 7, 326–337. Goldstein, B., Hird, S.N., 1996. Specification of the anteroposterior axis in Caenorhabditis elegans. Development 122, 1467–1474. Goldstein, B., Hird, S.N., White, J.D., 1993. Cell polarity in early C. elegans development. Development (Suppl.), 279–287. Gru¨nert, S., St Johnston, D., 1996. RNA localization and the development of asymmetry during Drosophila oogenesis. Curr. Opin. Genet. Dev. 6, 395–402. Gu, J., Stephenson, C.G., Iadarola, M.J., 1994. Recombinant proteins attached to nickel-NTA column: use in affinity purification of antibodies. Biotechniques 17, 257–262. Hake, L.E., Richter, J.D., 1994. CPEB is a specificity factor that mediates cytoplasmic polyadenylation during Xenopus oocyte maturation. Cell 79, 617–627. Hake, L.E., Richter, J.D., 1997. Translational regulation of maternal mRNA. Biochim. Biophys. Acta 1332, M31–M38. Hart, N.H., Donovan, M., 1983. Fine structure of the chorion and site of sperm entry in the egg of Brachydanio. J. Exp. Zool. 227, 277–296. Hart, N.H., Fluck, R.A., 1995. Cytoskeleton in teleost eggs and early embryos: contributions to cytoarchitecture and motile events. Curr. Top. Dev. Biol. 31, 343–381. Hart, N.H., Becker, K.A., Wolenski, J.S., 1992. The sperm entry site during fertilization of the zebrafish egg: localization of actin. Mol. Reprod. Dev. 32, 217–228. Helde, K.A., Grunwald, D.J., 1993. The DVR-1 (Vg1) transcript of zebrafish is maternally supplied and distributed throughout the embryo. Dev. Biol. 159, 418–426. Hesketh, J.E., 1996. Sorting of messenger RNAs in the cytoplasm: mRNA localization and the cytoskeleton. Exp. Cell Res. 225, 219–236. Jesuthasan, S., Stra¨hle, U., 1997. Dynamic microtubules and specification of the zebrafish embryonic axis. Curr. Biol. 7, 31–42. Johnston, S.H., Rauskolb, C., Wilson, R., Prabhakaran, B., Irvine, K.D., Vogt, T.F., 1997. A family of mammalian fringe genes implicated in

L. Bally-Cuif et al. / Mechanisms of Development 77 (1998) 31–47 boundary determination and the Notch pathway. Development 124, 2245–2254. Kimmel, C.B., Ballard, W.W., Kimmel, S.R., Wullmann, B., Schilling, T.F., 1995. Stages of embryonic development of the zebrafish. Dev. Dyn. 203, 253–310. King, M.L., 1996. Molecular basis for cytoplasmic localization. Dev. Genet. 19, 183–189. Kloc, M., Etkin, L.D., 1995. Two distinct pathways for the localization of RNAs at the vegetal cortex in Xenopus oocytes. Development 121, 287– 297. Kloc, M., Larabell, C., Etkin, L.D., 1996. Elaboration of the Messenger Transport Organizer Pathway for localization of RNA to the vegetal cortex of Xenopus oocytes. Dev. Biol. 180, 119–130. Kostomarova, A.A., 1969. The differentiation capacity of isolated loach (Misgurnis fossilis) blastoderm. J. Embryol. Exp. Morphol. 22, 407– 430. Lantz, V., Ambrosio, L., Schedl, P., 1992. The Drosophila orb gene is predicted to encode sex-specific germline RNA-binding proteins and has localized transcripts in ovaries and early embryos. Development 115, 75–88. Lantz, V., Chang, J.S., Horabin, J.I., Bopp, D., Schedl, P., 1994. The Drosophila orb RNA-binding protein is required for the formation of the egg chamber and establishment of polarity. Genes Dev. 8, 598–613. Macdonald, P.M., Smibert, C.A., 1996. Translational regulation of maternal mRNAs. Curr. Opin. Genet. Dev. 6, 403–407. Micklem, D.R., 1995. mRNA localization during development. Dev. Biol. 172, 377–395. Mosquera, L., Forristall, C., Zhou, Y., King, M.-L., 1993. A mRNA localized to the vegetal cortex of Xenopus oocytes encodes a protein with a nanos-like zinc finger domain. Development 117, 377–386. Mowry, K.L., 1997. Complex formation between stage-specific oocyte factors and a Xenopus mRNA localization elembent. Proc. Natl. Acad. Sci. USA 93, 14608–14613. Oppenheimer, J.M., 1936. The development of isolated blastoderms of Fundulus heteroclitus. J. Exp. Zool. 72, 247–269. Prince, V.E., Moens, C., Kimmel, C., Ho, R.K., 1998. Zebrafish Hox genes: expression in the hindbrain region of wild type and mutants of the segmentation gene valentino. Development 125, 393–406. Renucci, A., Lemarchandel, V., Rosa, F., 1996. An activated form of type I serine/threonine kinase receptor TARAM-A reveals a specific signalling pathway involved in fish head organiser formation. Development 122, 3735–3743.

47

Richter, J.D., 1996. Dynamics of poly(A) addition and removal during development. In: Hershey, J.W.B., Mathews, M.B., Sonenberg, N. (Eds.), Translational Control. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, pp. 481–504. Schmitz, B., Campos-Ortega, J.A., 1994. Dorso-ventral polarity of the zebrafish embryo is distinguishable prior to the onset of gastrulation. Roux’s Arch. Dev. Biol. 203, 372–380. Selman, K., Wallace, R.A., Sarka, A., Qi, X., 1993. Stages of oocyte development in the zebrafish Brachydanio rerio. J. Morphol. 218, 203–224. Selman, K., Petrino, T.R., Wallace, R.A., 1994. Experimental conditions for oocyte maturation in the zebrafish Brachydanio rerio. J. Exp. Zool. 269, 538–550. Seydoux, G., Mello, C.C., Pettitt, J., Wood, W.B., Priess, J.R., Fire, A., 1996. Repression of gene expression in the embryonic germ lineage of C. elegans. Nature 382, 713–716. Stachel, S.E., Grunwald, D.J., Myers, P., 1993. Lithium perturbation and goosecoid expression identify a dorsal specification pathway in the pregastrula zebrafish. Development 117, 1261–1274. Stebbins-Boaz, B., Richter, J.D., 1997. Translational control during early development. Crit. Rev. Euk. Gene Expr. 7, 73–94. Stebbins-Boaz, B., Hake, L.E., Richter, J.D., 1996. CPEB controls the cytoplasmic polyadenylation of cyclin, cdk2, and c-mos mRNAs and is necessary for oocyte maturation in Xenopus. EMBO J. 15, 2582– 2592. St Johnston, D., 1995. The intracellular localization of messenger RNAs. Cell 81, 161–170. Stra¨hle, U., Jesuthasan, S., 1993. Ultraviolet irradiation impairs epiboly in zebrafish embryos: evidence for a microtubule-dependent mechanism of epiboly. Development 119, 909–919. Tung, T.C., Chang, C.Y., Tung, Y.F.Y., 1945. Experiments on the developmental potencies of blastoderms and fragments of teleostean eggs separated latitudinally. Proc. Zool. Soc. Lond. 115, 175–188. Vriz, S., Lovell-Badge, R., 1995. The zebrafish Zf-Sox 19 protein: a novel member of the Sox family which reveals highly conserved motifs outside of the DNA-binding domain. Gene 153, 275–276. Wacker, S., Herrmann, K., Berking, S., 1994. The orientation of the dorsal/ ventral axis of zebrafish is influenced by gravitation. Roux’s Arch. Dev. Biol. 203, 281–283. Yisraeli, J.K., Sokol, S., Melton, D.A., 1990. A two step model for the localization of maternal mRNA in Xenopus oocytes: involvement of microtubules and microfilaments in translocation and anchoring of Vg1 mRNA. Development 18, 289–298.