Polarity of the ascidian egg cortex and relocalization of cER and mRNAs in the early embryo

Ascidian oocytes are highly polarized along the animal-vegetal (a-v) axis before fertilization (Sardet et al., 1989; Sardet et al., 1992). This polarity principally concerns two peripheral domains: a mitochondria-rich and ER-poor subcortical layer (the myoplasm); and a cortical endoplasmic reticulum (cER) network (Roegiers et al., 1999). The peripheral region of the egg is also characterized by an (a-v) gradient of maternal mRNAs (called postplasmic mRNAs or, because they make a posterior end mark in zygotes and embryos, PEM mRNAs) including the muscle determinant macho1 (Nishida and Sawada, 2001; Sardet et al., 2005; Sardet et al., 2003; Satou et al., 2002; Yoshida et al., 1996). We have recently shown that, in Halocynthia, the cER network is associated with some postplasmic/PEM RNAs, constituting a cortical domain that we call the cER-mRNA domain throughout this article (Sardet et al., 2003; Sardet et al., 2005). We did not, however, provide a detailed description of domain reorganizations in whole Halocynthia zygotes. Neither did we analyse crucial stages of cortical reorganization before cleavage or describe the simultaneous changes in cytoskeleton organization at the level of the cortex. This prompted us to extend our initial observations made in Halocynthia in two other ascidian species used for research, Ciona intestinalis and Phallusia mammillata, and in particular to include a thorough study of reorganizations of cortical domains and the cytoskeleton after fertilization in these two species.

The cortical cER-mRNA domain and subcortical myoplasm domain (Fig. 1, Fig. 2I) are relocalized in two major phases of reorganizations between fertilization and first cleavage (Roegiers et al., 1999). First, a sperm-triggered actomyosin-driven contraction moves the sperm aster vegetally and concentrates the cER-mRNA domain and the myoplasm in and around the vegetal/contraction pole (Fig. 2IC, cp). The position of this vegetal protrusion defines the future site of gastrulation and dorsal pole (Fig. 2IC, d) of the embryo (Roegiers et al., 1995). Maternal determinants for muscle, endoderm, unequal cleavage and gastrulation are concentrated in this vegetal region after fertilization (Nishida, 1997; Nishida, 2002b; Roegiers et al., 1999). The sperm centrosome is also moved vegetally in the cortex, predicting the position of the future posterior pole (Fig. 2IC, P). After meiosis is complete (20-30 minutes after fertilization), determinants of muscle (macho1) and unequal cleavages, as well as more than two dozen postplasmic/PEM RNAs are displaced posteriorly towards the future posterior pole of the zygote (Chiba et al., 1999; Roegiers et al., 1999; Sardet et al., 2005; Sardet et al., 2003). This phase starts after the male and female pronuclei form and it extends throughout mitosis (25-50 minutes). This second major phase of reorganization is caused mainly by the translocation of the sperm aster with respect to the posterior cortex (Sardet et al., 1989; Roegiers et al., 1999). During first cleavage, determinants and the cortical and subcortical domains are partitioned equally in the posterior region of the two first blastomeres. At the eight-cell stage, a macroscopic cortical structure called the centrosome attracting body (CAB) forms in the most posterior region of vegetal blastomeres (Nishida, 2002b; Nishikata et al., 1999). The CAB is characterized by the accumulation of the cER-mRNA domain, including macho1 (Nishida and Sawada, 2001; Sardet et al., 2003). It is thought that the CAB is involved in asymmetric divisions and acts as a source of factors (probably translation products of localized mRNAs) directing development and differentiation in the posterior region of the embryo (Kobayashi et al., 2003; Kondoh et al., 2003; Nishida, 2002a; Nishida, 2002b; Sardet et al., 2005; Sardet et al., 2003; Yoshida et al., 1998).

In order to analyse the structure of the cortex and its transformation, it is possible to isolate cortical fragments rapidly and reproducibly using simple techniques (Sardet et al., 2002; Sardet et al., 1992). Ascidian zygotes and embryos are ideal for a detailed study of the cortex because they develop synchronously in large numbers, are highly polarized and can be attached to polylysine-coated surfaces to prepare large fields of cortical fragments coming from identifiable regions of zygotes and embryos at all stages. The isolated ascidian oocyte cortex consists of the plasma membrane (PM), adhering cytoskeletal elements and a conspicuous polarized network of cortical rough endoplasmic reticulum (cER), which is covered with ribosomes.

Recently using the Japanese ascidian Halocynthia, our lab and Nishida's lab have shown that one of the most abundant cortical maternal mRNAs, Hr-PEM1, and the muscle determinant Hr-macho1 are bound to the cER network in eggs and embryos. We have suggested that these maternal mRNAs are relocalized together with the cER network during the two major phases of reorganization and segregated into specific blastomeres to direct differentiation and asymmetric division (Sardet et al., 2003). However, in this study, we did not examine other cortical components (MF, MT) or critical stages of relocalization (second major phase). We also wondered whether what we described in Halocynthia roretzi held true for its distant relative C. intestinalis, the cosmopolitan species used by most laboratories.

We now report the evolution of cortical polarity including cytoskeletal elements in zygotes and embryos of C. intestinalis which belongs to the Aplousobranchiata order and the European species P. mammillata belonging to the Phlebobranchiata order (Turon and Lopez-Legentil, 2004). Our results complement our initial observations (Sardet et al., 2003) in the Japanese species H. roretzi which belongs to the Stolidobranchiata, a third order. Our study shows that these three evolutionarily distant ascidian species use similar cytoskeleton-driven repositioning and compaction of the cER network for polarization of cortical maternal mRNAs (including the determinant macho1) and for segregation of these mRNAs in small posterior blastomeres. A part of this story is told in the form of a BioClip `Polarity inside the egg cortex', a multimedia document that can be downloaded from the `Research' section of http://www.bioclips.com/. Further information on the cortex is available in supplementary materials (six films) and on our laboratory web site (http://biodev.obs-vlfr.fr/biomarcell/ascidies/eggcortex.html).

Adults of the ascidians C. intestinalis and P. mammillata were collected near Sète or Roscoff, France. Oocytes and zygotes were dechorionated using either 0.1% trypsin in seawater (pH 8.0) for 3 hours or with 1% thioglycolate, 0.05% pronase in filtered sea water, pH 10.0, for 2-5 minutes (Dumollard and Sardet, 2001; Sardet et al., 2003; Sardet et al., 1989). Oocytes were fertilized and embryos cultured as described (Roegiers et al., 1999).

Mitochondria were labelled for 5-10 minutes with the carbocyanine dye DiOC2(3) (0.5 μg ml^–1) or 1 μM TMRE mitotracker (Molecular Probes) and chromosomes were labelled for 15 minutes with 1 μg ml^–1 Hoechst 33342 (Molecular Probes) (Dumollard and Sardet, 2001; Sardet et al., 1989). For labelling endoplasmic reticulum, we injected a small oil droplet saturated with DiIC16(3) (Speksnijder et al., 1993).

We used a Leica SP2 confocal microscope, and Zeiss Axiophot/Axiovert and Metamorph/Metaview software (Universal Imaging) for digital imaging (Roegiers et al., 1999). Thin-section electron microscopy of properly orientated oocytes and embryos, and fast-freezing deep-etching and replication of cortices were performed as described (Sardet et al., 2003; Sardet et al., 1992).

Oocytes or synchronous populations of zygotes and embryos at different stages were deposited on a polylysine-coated coverslip into a drop of EMC or one of three types of cortex buffer that all gave similar results (Buffer X, CIM buffer and Vacquier buffer) (Sardet et al., 1992). Oocytes, zygotes or embryos attached to a polylysine-coated glass surface were opened and emptied of their cytoplasm with a stream of cortex buffer squirted from a Pasteur pipette, leaving cortical fragments attached to the glass (Fig. 2I). Immunolabellings were done essentially as described (Roegiers et al., 1999). Primary antibodies were rat YL1/2 anti-β-tubulin or mouse anti-β-tubulin antibody (Amersham), and rabbit antibody against the S6 ribosomal subunit (Cell Signaling Technology). Live cortices were also immunolabelled using short incubations (15 minutes) with antibodies and secondary antibodies in Buffer X. The cER network was labelled using DiOC6(3) or DiIC16(3) and microfilaments were labelled using fluorescent phalloidins (rhodamine- or Alexa-633-conjugated) as described (Sardet et al., 1992). Surface sugars of live eggs and zygotes were labelled with rhodamine/concanavalin-A (conA) (50 μg ml^–1, Molecular Probes) for 1 minute before cortex isolation. To detach ribosomes from the surface of the ER, isolated cortices were incubated in KCl-puromycin (Sigma) buffer for 20 minutes as reported (Sardet et al., 1992; Sardet et al., 2003)

In order to determine the origin of the piece of cortex attached to the coverslip with respect to the polarity of oocytes, zygotes or embryos [animal-vegetal (a-v), dorsoventral or anteroposterior], we applied the strategy described previously for oocytes (Sardet et al., 1992). Cells were first labelled with Hoescht 33342, TMRE mitotracker (Molecular Probes) and/or rhodamine/conA. We recorded the position of individual zygotes attached to the coverslip and the orientation of axes, using scratch marks on the coverslip as reference points. We then sheared zygotes, fixed the resulting cortices and labelled them with DiOC6(3) to observe the cER network. Using the scratch marks on the coverslip, we matched the images of the zygote or embryo and the corresponding piece of cortex to determine its origin along the polarized axes.

The fixations and fluorescent (tyramide signal amplification, TSA) in situ hybridization procedures (without extractions to preserve the cER network in cortices) on whole cells and cortices isolated from them were as described (Sardet et al., 2003). Hybridized antisense maternal mRNAs [Ci-PEM1 corresponding to CLSTR1544 and Ci-PEM3 corresponding to CLSTR865 on the Ghost Database (http://ghost.zool.kyoto-u.ac.jp/indexr1.html)] labelled with digoxigenin were revealed using the phosphatase precipitation method, the HNPP fluorescent detection kit (Roche) or the TSA method (Sardet et al., 2003). We have used a nomenclature for cortically localized mRNAs we call postplasmic/PEM RNAs that harmonizes the current terminology (for review, see Sardet et al., 2005).

Experiments were performed on oocytes, zygotes and embryos of Phallusia and Ciona, and they gave similar results. To facilitate understanding, we have selected the best illustrations in either species (specified in figure legends). Phallusia zygotes were easier to handle for cortical orientation experiments and most mRNA localization experiments were done with Ciona.

The myoplasm is a mitochondria-rich, ER-poor domain and forms a 5-7 μm thick subcortical basket lining the vegetal hemisphere and opened in the animal pole region of the oocyte (Fig. 1A1-4,C1-4; see also Movies 1-3 in supplementary material). The animal pole region itself is characterized by a low density of mitochondria and an accumulation of ER surrounding the meiotic spindle (Fig. 1E). A transition zone in which the distribution and density of subcortical mitochondria and the ER network change is clearly visible above the equator (Fig. 1C1-3). The distribution of the cER network in Ciona is seen in confocal tangential sections made 0.5-2 μm under the surface (Fig. 1D1,D2). In the vegetal hemisphere, cortical sections reveal a very dense network of tightly knit tubules and ER cisternae (Fig. 1D1), contrasting with the sparser cortical cER network observed in the animal hemisphere (Fig. 1D2). This cER network can be detected lining the PM in thin-section electron micrographs, which also reveal strands of ER traversing the mitochondria-rich myoplasm and the presence of ER microdomains in the deeper cytoplasm (Fig. 1B). Maternal mRNAs Ci-PEM1 (Fig. 1F1) and Ci-PEM3 (data not shown) are located within 0.5-2 μm of the surface along an a-v gradient best seen in equatorial confocal sections. At the highest possible resolution, the fluorescent in situ hybridization signal appears as a reticulated network (Fig. 1F2-3), suggesting that Ci-PEM1 might be associated with the cER network.

Cortices isolated from oocytes, zygotes and embryos constitute an ideal open cell preparation in which to analyse the spatial distribution of cortical organelles, cytoskeletal elements, macromolecular complexes and mRNAs associated with the cytoplasmic side of the PM (Fig. 2I). Live and fixed cortices isolated from oocytes of Ciona and Phallusia (∼0.5-1.0 μm thick) show an obvious polarity in the organization of the cER network and retain some microfilaments (MFs) and microtubules (MTs) that adhere to the PM or macromolecular complexes attached to it (Fig. 2IIE). Cortical fragments derived from the vegetal hemisphere of oocytes are characterized by a cER network made of tightly knit tubules or sheets (Fig. 2IIA1,A2). The fluorescent signal for Ci-PEM1 RNAs colocalizes with this vegetal cER network (Fig. 2IIC1,C2) as described for Halocynthia (Sardet et al., 2003). Isolated cortical fragments issued from the animal hemisphere are characterized by a sparse tubular cER network (Fig. 2IIA1,A2). Observations of the cER in isolated cortices from Ciona oocytes are in agreement with confocal observations of whole eggs (Fig. 1D1,D2) and previous observations (Sardet et al., 1992). MFs are also distributed along an a-v gradient, being more abundant where the cER network is the most tightly knit (Fig. 2IIA1,A2). Double labelling of membranes with DiIC16(3) and ribosomes (with antibody against S6 ribosomal protein) demonstrates that the cER network retained on isolated cortices is rough ER (Fig. 2IIB1-B3). Ribosomes can be detached from the cER network using KCl-puromycin treatment (Fig. 2IID1,D2) (Sardet et al., 2003; Sardet et al., 1992). That the cER is rough ER is confirmed by the fact that, in electron micrographs (fast-frozen, deeply etched replica in Fig. 2IIE), many particles the size of ribosomes (Fig. 2IIE, yellow) can be detected on the cER (Fig. 2IIE, red).

Fertilization triggers a first major phase of reorganization of the oocyte's cortex which concentrates ConA-labelled surface glycoproteins and glycolipids, the cER-mRNA domain, and the mitochondria-rich myoplasm in the vegetal hemisphere (Fig. 3A,B,D-G). A vegetal/contraction pole (labelled cp/d in the figures) forms in fertilized Ciona eggs (Fig. 3F1,F2,G) as in Phallusia (Roegiers et al., 1995). In both Phallusia and Ciona, the vegetal/contraction pole is not always exactly opposite the polar bodies (marking the animal pole) but can be off by 45-60° (Roegiers et al., 1995) (data not shown). Many cortical fragments isolated during and after the contraction 2-5 minutes after fertilization contain tightly packed sheets of cER network forming a patch (20 μm in diameter, 2-5 μm thick) positioned in the centre of the vegetal/contraction pole (Fig. 3C). Ribosomes associated with cER are abundant in this cER-rich patch (data not shown). Interestingly, actin microfilaments form a ring around the cER-rich patch, which is particularly noticeable at the end of the contraction (Fig. 3C1-C3,H). We also find that surface glycolipids and glycoproteins recognized by ConA in the contraction pole (Fig. 3F1,F2) define a microvillus-rich surface region corresponding to the cortical ER-rich plaque situated beneath the PM (Fig. 3J1,J2). A microvillus-poor zone adjacent to the microvillus-rich zone precisely corresponds to the location of the MF ring (Fig. 3H,J1,J2). Another feature of these postfertilization cortices is the presence of conspicuous MTs emanating from sperm asters, which are probably retained in the isolated cortical fragments through interaction of MTs with the PM or MFs and/or local accumulation of the cER network in the aster's centre (Fig. 3I1,I2).

Sperm aster localization in the cortex after the cortical contraction triggered by the fertilizing sperm represents a first indication of posterior polarization. This polarity is amplified during a second major phase of reorganization (Chiba et al., 1999; Roegiers et al., 1999). The mitochondria-rich myoplasm domain and the cER-mRNA domain, which first accumulate around and in the vegetal/contraction pole, are then progressively translocated and relocalized in three subphases towards the future posterior region and equator between meiosis-II completion and cleavage (Fig. 2I, Fig. 4A-D; see Movies 4 and 5 in supplementary material) (Roegiers et al., 1999; Sardet et al., 1989). We prepared cortical fragments from synchronous zygotes during posterior translocation, identified their provenance (from vegetal, posterior, anterior regions) and labelled them to observe cER, MT, MF and Ci-PEM1 RNAs. We observed that the posterior pole region of isolated cortices was characterized by a more tightly knit network of cER (∼1 μm thick) composed of anastomosed tubules and sheets (Fig. 4H1,H3,I1,I3,J1,J3,K1). This posterior cER network is displaced from the vegetal/dorsal region, where surface ConA binding sites are the most abundant (Fig. 4F). The posterior cortical fragments also retain many short MTs, which probably correspond to the tips of MTs emanating from the posteriorly localized sperm aster (Fig. 4E,K). In whole zygotes, part of the in situ signal for Ci-PEM1 RNAs is situated close to the cortex, but it is also present in deeper subcortical and cytoplasmic locations (Fig. 4B,C) as previously observed in Halocynthia (Sardet et al., 2003). We observed that Ci-PEM1 RNAs are retained in posterior fragments of cortices isolated during the second major phase of reorganisation. Ci-PEM1 is clearly associated with the tightly woven network of cER marking the posterior pole, where the mRNAs form small patches (Fig. 4H,I). The Ci-PEM1 signal is also detected on a subpopulation of ER tubules projected outside the isolated posterior cortex during the shearing process that breaks the egg open (data not shown). This probably corresponds to a part of the cER present in a deeper cytoplasmic location at this stage (Fig. 4A). As with oocyte cortices, we observe a loss of the signal for Ci-PEM1 RNAs after KCl-puromycin treatment (Fig. 4I,J), suggesting that Ci-PEM1 RNAs might be detached from the surface of the cER together with ribosomes.

Owing to the orientations of the first three cleavage planes [along the a-v axis and along the anteroposterior (A-P) axis, respectively], the posterior myoplasm domain and the cER-mRNA domain are both equally partitioned into the two posterior vegetal blastomeres (Fig. 5, B4.1) at the eight-cell stage (Fig. 2IC, Fig. 6D). From the eight-cell stage to the 64-cell stage, postplasmic/PEM RNAs are located in and around the CAB (a cER-rich macroscopic structure sandwiched between the myoplasm and the PM) (Fig. 5) (Iseto and Nishida, 1999; Nishikata et al., 1999; Roegiers et al., 1999; Sardet et al., 2003). The formation and condensation cycle of the CAB structure can be observed using time-lapse confocal imaging of an embryo double labelled for ER and mitochondria (Fig. 5A1-A3; see also Movie 6 in supplementary material), and in fixed embryos and cortices with labelled chromatin and Ci-PEM1 RNAs. The cER-rich CAB domain is thinnest and most extended along the PM at the eight-cell stage during interphase, compacting into a smaller and thicker cortical patch during mitosis (Fig. 5A1-A3,B,C). This change in shape of the CAB from spread out to concentrated repeats as a cycle during each unequal cleavage (data not shown). Observation of the CAB region in Phallusia by electron microscopy (Fig. 5E1,E2) confirms the presence of a high density of ER tubes (rather than vesicles) and of an electron-dense matrix as described previously in Halocynthia and Ciona (Iseto and Nishida., 1999). We characterized the cER-rich zone of embryo cortices, taking advantage of the fact that the CAB is retained in cortical fragments from the posterior face of the cube-like eight-cell-stage embryo. The CAB is easily recognized in isolated cortices as cER-rich plaques (interphase, 20×10 μm, 4-5 μm thick; mitosis, 10×5 μm, 7-8 μm thick) (Fig. 5F,G). Ribosomes are particularly abundant and can stretch along ER strands spread by the force of shear (Fig. 5J). Such isolated CABs often have a hole in their centre possibly indicating the position of the centrosome ripped away from the cortex during the shearing procedure (Fig. 5H1). A variable level of condensation of the cER network in the CAB is also evident in isolated cortices. This probably reflects the cell-cycle position of blastomeres at the time of cortex isolation. Double labelling of the isolated cortices demonstrates that the cER-rich zone retains bundles of MTs enmeshed deep into the cER network (Fig. 5H, one plane of a series of confocal z-axis sections). These MTs probably originate from the centrosome adjacent to the CAB. Actin MFs (Fig. 5I) and ribosomes (Fig. 5J) are also abundant in the isolated CAB. Observations of Ci-PEM1 RNAs after fluorescent in situ localization indicate that they are present in a reticulated network compatible with their cER association as previously described for Halocynthia (Sardet et al., 2003).

We have analysed the reorganizations of the cell cortex in living oocytes, zygotes and embryos, and in cortices isolated from them in two species of ascidian: P. mammillata and C. intestinalis. The most conspicuous cortical changes concern the distribution and relocalization of the submembranous network of endoplasmic reticulum (cER) and of postplasmic/PEM RNAs associated with it (Fig. 6). The polarization and compaction of the cER network follows changes in the activation status of the egg and its progression through meiotic and mitotic cell cycles, and is correlated with cell-cycle-dependant modifications in the abundance and redistribution of cytoskeletal elements (MFs and MTs) and surface topography (ConA-binding sites indicating abundant microvilli). Ciona, Phallusia and Halocynthia, the three main species of solitary ascidians used for research, all seem to use cER-mRNA relocalization and compaction for polarization of the zygote and for the segregation of informational macromolecules (determinants/mRNAs) and the translation machinery (ribosomes/cER) in small posterior blastomeres (Sardet et al., 2005; Sardet et al., 2003; Sardet et al., 1992). This suggests that ascidians species separated by millions of years of evolution share the same cellular and molecular mechanisms to establish developmental axes and translocate and segregate determinants acting on posterior blastomeres in similar ways. Such conservation does not seem to have happened in nematodes, which show great flexibility in cellular mechanisms of sperm-directed axis formation and domain localization (P granules) (Goldstein et al., 1998; Hasegawa et al., 2004). Our previous experiments with Halocynthia (Sardet et al., 2003) strongly suggested that there was a link between the cER network established in the oocyte and the CAB, the macroscopic cortical structure mediating unequal posterior cleavages, which acts as a signalling centre for muscle and posterior development (Nishida, 2002a; Nishida, 2002b).We did not, however, investigate whether the cER-mRNA domain that moved posteriorly was retained as part of the isolated posterior cortex. Our present observations show that the cER network established during oogenesis is undoubtedly a structural precursor of the CAB.

The idea that a specific ER network exists in the cortex of eggs and embryos derives from the fact that, in Halocynthia (Sardet et al., 2003) and Ciona and Phallusia (this paper; F. Prodon et al., unpublished), the original and most abundant postplasmic/PEM maternal RNA (Ci-PEM1) and others (including macho1) are associated with a layer of rough cER attached to the cytoplasmic face of the oocyte's PM (Sardet et al., 1992; Sardet et al., 2005). In contrast to the cER, cytoplasmic ER (which is also mostly rough ER) does not bind Ci-PEM1. After fertilization, another type of cER (formed of sheets and tubes not coated with postplasmic/PEM RNAs) is also present in isolated cortices of zygotes and blastomeres. It is not clear whether this post-fertilization cER is a part of the original oocyte cER from which postplasmic/PEM RNAs have been excluded (perhaps by lateral translation and/or aggregation) or whether it derives from deeper cytoplasmic ER that has formed new connections to the PM after fertilization. Such rearrangements in the ER network are expected because the deeper cytoplasmic ER network is clearly in continuity with the cER (Speksnijder, 1992). Our observations with anti-ribosome antibodies complement our previous conclusions (based on fast-freezing deep-etching electron microscopic observations) that cER is rough ER (Sardet et al., 1992).

The fact that cER remains attached to the PM in cortical fragments isolated using large shearing forces implies that cER has strong attachment sites to the PM. We have visualized discrete attachment feet on the cER forming junctions with the PM in Phallusia (Sardet et al., 1992). Specialized junctions between the PM and sarcoplasmic reticulum (SR) in muscle cells are well documented and PM-ER junctions have been recently isolated and characterized from neurons and astrocytes, where they are suspected to play a role in calcium signalling and the refilling of intracellular ER stores (Lencesova et al., 2004). In these cells, PM microdomains form junctional units with the underlying ER through macromolecular complexes composed of specific subunits of pumps, channels and submembranous proteins. It will be important to see whether such macromolecular complexes are components of the PM-cER junctions we have described in ascidian eggs (Sardet et al., 1992). We have argued that such PM-cER junctions are likely to exist in other eggs (for review, see Sardet et al., 2002). Several recent reports also indicate that specific cortical ER compartments close to the PM are present in yeast and plant cells, where they are thought to play important roles in polarization, membrane-protein insertion (in yeast bud) and mRNA and protein localization (in rice endosperm) (Du et al., 2004; Hamada et al., 2003).

Fig. 6 describes the cellular transformation of the original network of cER-mRNA present in the mature oocyte from a monolayer network of increasing density in the vegetal cortex into a compact posterior cortical mass forming the bulk of the CABs of eight-cell-stage blastomeres. To achieve the relocation and compaction the cER network passes through two major phases of reorganization between fertilization and cleavage. We also detail changes in surface topography (microvilli) and the cortical MF and MT networks.

In mature oocytes arrested in meiotic metaphase I (Fig. 6A), the cER forms a thin (less than 0.5 μm) monolayer made of tubes and sheets whose density increases from the animal to the vegetal pole (Fig. 2A1, Fig. 6) (Sardet et al., 1992). Post-plasmic/PEM RNAs such as PEM1 are evenly distributed over the whole surface of cER network at that stage. MFs are enriched in regions of abundant cER in both Ciona and Phallusia but, as previously noted, we do not know whether this represents the situation in living oocytes (Sardet et al., 1992). A few stable MTs are also present in the isolated cortex and are most frequent in the cER poor zone (animal region). No major differences in surface microvilli (ConA-binding sites) are detected at this stage.

In the zygote (Fig. 6B), the spectacular actomyosin contraction that follows fertilization by sperm drags the cER network, including the postplasmic/PEM RNAs, in the direction of the vegetal hemisphere. This contraction is triggered by calcium (Roegiers et al., 1995). The sperm aster, subcortical myoplasm layer and surface macromolecules are also moved vegetally. This massive `capping' of surface macromolecules together with cortical and subcortical components results in the formation of a contraction/vegetal pole (Roegiers et al., 1999). During and at the end of the contraction, a remarkable ring of MFs forms around the thick (3-5 μm) patch in which cER sheets and tubes, and MFs are intertwined. This cER/MF-rich patch corresponds to the microvillus-rich surface of the contraction pole. The cER/MF-rich patch is surrounded by a MF ring, which is morphologically reminiscent of a cleavage ring and corresponds to a microvillus-free zone on the surface (possibly a zone of membrane insertion, as seen during cytokinesis). Few individual MTs adhere to the isolated cortex but those that do are stable in live cortices. Just after fertilization, large fragments of sperm aster are tethered to the isolated cortex, possibly via the centrosomal accumulation of cER identified as the moving calcium-wave pacemaker site situated in the cortex (Dumollard and Sardet, 2001).

In the zygote (Fig. 6C) after meiosis completion, the sperm aster situated in the future posterior pole of the embryo has extended long MTs, which interact with the posterior cortex and the female pronucleus near the animal pole. An extensive translocation of the aster towards the centre of the egg and rotational movement against the posterior cortex cause a huge displacement of the cER and postplasmic/PEM RNAs associated with it, as well as of the bulk of the myoplasm and entrapped cytoplasmic elements (Chiba et al., 1999; Roegiers et al., 1999). At this stage, some of the cER has moved away from the surface during the rotational and folding motions of the posterior domains [as seen in Halocynthia (Sardet et al., 2003)]. It appears that some of this cER is positioned in cytoplasmic ER folds located within the myoplasm or dragged towards the centrosomal area along astral rays (Roegiers et al., 1999). At this stage postplasmic/PEM RNAs are no longer distributed all over the cER in the cortex but form patches associated with the cER in the posterior region of the zygote. In contrast to the conspicuous MF-rich structures we observed at the vegetal/contraction pole, the mitotic zygote periphery and cortical fragments including the posterior region lacked remarkable MF structures. We have previously observed that this region vibrates at high frequency owing to the buckling of posterior MTs from the sperm aster against the surface (Roegiers et al., 1999). Therefore, it was not surprising to find many short MTs (stable in live cortices) in isolated cortical fragments originating from the posterior region of the zygote. These MTs might represent MT ends in interaction with the cytoplasmic face of the PM and it would be interesting to know whether they are polarized and what motors and microtubule-associated proteins (MAPs) are associated with them. To proceed further, an analysis similar to that conducted to understand the mechanism of the cortical rotation in Xenopus zygotes will be needed (Marrari et al., 2003; Marrari et al., 2004).

In the embryo (Fig. 6C), because the cER and associated mRNAs are located in the vegetal posterior region at the time of first cleavage, they naturally segregate in the two posterior vegetal B4.1 blastomeres and their micromere progeny through a series of stereotyped symmetrical and asymmetrical cleavages. Our observations in Ciona and Phallusia confirm our previous observations in Halocynthia, in which the cER-rich CAB is proportionally bigger (about three times bigger) (Sardet et al., 2003). In addition, we could clearly see that cortical ER and mRNAs in Ciona form an extended cortical plaque at interphase that is subsequently compacted during mitosis just before unequal cleavage takes place. Although, in isolated cortices, a higher density of crisscrossing MFs are situated beneath the PM in the CAB area, we do not know whether this represents the situation in whole embryos because no such cortical differences can be seen in fixed embryos labelled with fluorescent phalloidins (Nishikata et al., 1999) (J. Chenevert, personal communication). MTs from the centrosomal aster are present in great numbers enmeshed within the CAB and it is possible to imagine that multiple motors situated within or on the surface of the CAB (on cER or electron dense matrix material) can manoeuvre and attract the centrosomal MTs to mediate unequal cleavage between the eight-cell and 64-cell stages (Iseto and Nishida, 1999; Nishikata et al., 1999).

Because ER is a multifunctional organelle important for calcium regulation, signalling, mRNA localization and translation, it is interesting to consider what cellular and developmental consequences localization of a large amount of cER-mRNA domain at the poles might have.

cER is endowed with special properties because it initiates calcium signals triggered by the sperm factor even when it is injected in the egg centre (Carroll et al., 2003; Kyozuka et al., 1998). Accumulation of cER in the centre of the sperm aster and in the vegetal/contraction pole are, successively, the sites of two calcium-wave pacemakers that emit 6-12 repetitive calcium waves from fertilization to meiosis completion. These repetitive calcium waves initiated in the zone of cER accumulation are necessary for meiosis completion and also regulate mitochondrial metabolism (Specksnijder et al., 1990; Carroll et al., 2003; Dumollard et al., 2003). Because calcium waves are followed by waves of enzymatic activations [protein kinase C, calmodulin kinase II (Larabell et al., 2004; Markoulaki et al., 2004)], the vegetal/contraction pole, cER-mRNA domain and associated translation machinery will be exposed to the highest levels of these enzymatic activities and calcium signals for 15-20 minutes during completion of meiosis II (Speksnijder et al., 1990). Therefore, during this period, the vegetal/contraction pole might acquire special properties that affect later developmental events taking place at that site, such as gastrulation. A memory of ER-generated calcium wave has been shown to play a role during the postimplantation development of mammals (Ozil and Huneau, 2001).

In contrast to Drosophila and Xenopus oocytes, in which the mechanisms of mRNA localization are being uncovered (Kloc and Etkin, 2005; Kloc et al., 2002), we know very little about the initial localization of postplasmic/PEM RNAs in the cortex during oogenesis. However, our observations in whole zygotes and embryos, and cortices isolated from them in Ciona (this paper) and Halocynthia (Sardet et al., 2003) provide a possible explanation for the relocalization of some mRNA after fertilization, namely that these mRNAs move and relocate together with the cER after fertilization. How a subset of postplasmic/PEM RNAs are bound to cER remains a mystery but our working hypothesis is that they are associated with ER-associated Staufen-containing ribonucleoprotein particles of the type described in mammalian neurons or Xenopus oocytes (Ohashi et al., 2002; Yoon and Mowry, 2004).

In ascidians we know almost nothing of the translational control of cortical mRNAs, except for Cs-PEM3, whose protein product assumes a broader posterior distribution than its mRNA situated in the CAB (Satou, 1999). The distribution of other proteins encoded by postplasmic/PEM mRNAs has not yet been examined. It is very likely that proteins coded by these maternal RNAs and, in particular, the muscle determinant macho1 are synthesized before the 16-cell stage, because macho1 must exert its effect on muscle precursor blastomeres that do not inherit the cER-rich CAB. It is also probable that, like PEM3, most proteins translated from cortically localized postplasmic/PEM RNAs occupy a much larger area in the posterior pole. Recently, a Y-box protein (CiYB1; thought to be involved in storage and translational control of mRNAs in oocytes) and its mRNA (CiYB1) have been shown to be partially colocalized with postplasmic/PEM RNAs in the posterior of the embryo (Tanaka et al., 2004).

As discussed above, the cER/RNA domain and/or other macromolecular complexes contained within the CAB might be responsible for pulling the centrosome and positioning one pole of the spindle near the cortex for the asymmetrical and unequal cleavages that generate posterior micromeres. This macroscopic structure might be related to the structure that repositions the spindle in posterior blastomeres of Caenorhabditis elegans embryos (Skop and White, 1998). Isolated cortical fragments with adhering CABs and associated MTs constitute an ideal open cell preparation to investigate the mechanisms of MT translocation in an ER/MT-motor-rich cortical domain using the strategy developed to reactivate MT translocation in isolated cortices of Xenopus (Marrari et al., 2003; Marrari et al., 2004).

Ascidians are an emerging model for cell and development biology that offers the possibility to routinely isolate cortical fragments from synchronous oocytes, zygotes and embryos (the presence of chorions around oocytes and zygotes renders this approach difficult for the moment in C. elegans and Drosophila). In terms of the nature of the developmental information contained in the cortex, it is clear that, as in Xenopus and Drosophila, maternal mRNAs in the ascidian cortex play an essential role. There are strong indications that, in Xenopus, some of the maternal mRNAs (e.g. XCat2 and Vg1) are associated with ER and ER-bound Staufen (Chang et al., 2004; Dollar et al., 2002; Yoon and Mowry, 2004). In terms of cellular mechanisms of polarity amplification after fertilization, in both ascidians and in C. elegans, establishment of anteroposterior polarity is linked to an actomyosin-dependent surface capping of many peripheral macromolecules (including the PAR3-PAR6-aPKC complex) and organelles (Munro et al., 2004; Schneider and Bowerman, 2003). Nothing, however, is known about reorganizations of cortical ER in the C. elegans zygote. Similarities between Xenopus and ascidian zygotes concern the MT-dependent cortical rotation that, just before mitosis, displaces the cER-mRNA and myoplasm domains posteriorly in ascidians and dorsalizing factors and organelles in Xenopus (Weaver et al., 2003).

Because genomic and transgenic tools are now at hand in ascidians (Satoh et al., 2003), we should be able to elucidate the molecular and cellular mechanisms underlying cortical reorganizations, localization of mRNAs and information processing in the cell cortex.

We thank ARC and AFM foundations, and ACI from the French Research Ministry for their support. We also thank C. Djediat and P. Chang for help with electron microscopy, C. Rouviere for advice about imaging, and J. Chenevert for critical reading of the manuscript.