Spatiotemporal localization of germ plasm RNAs during zebrafish oogenesis

Transcription

1 Mechanisms of Development 124 (2007) Spatiotemporal localization of germ plasm RNAs during zebrafish oogenesis Kyoko Kosaka a, Koichi Kawakami b, Hiroshi Sakamoto a, Kunio Inoue a, * a Department of Biology, Graduate School of Science and Technology, Kobe University, 1-1 Rokkodaicho, Nadaku, Kobe , Japan b Division of Molecular and Developmental Biology, National Institute of Genetics, Mishima, Shizuoka , Japan Received 14 November 2006; received in revised form 27 December 2006; accepted 9 January 2007 Available online 13 January 2007 Abstract In zebrafish, primordial germ cells (PGCs) are determined by a specialized maternal cytoplasm, the germ plasm, which forms at the distal ends of the cleavage furrows in 4-cell embryos. The germ plasm includes maternal mrnas from the germline-specific genes such as vasa and nanos1, and vegetally localized dazl RNA is also incorporated into the germ plasm. However, little is known about the distributions and assembly mechanisms of germ plasm components, especially during oogenesis. Here we report that the germ plasm RNAs vasa, nanos1, and dazl co-localize with the mitochondrial cloud (MC) and are transported to the vegetal cortex during early oogenesis. We found that a mitochondrial cloud localization element (MCLE) previously identified in the 3 0 untranslated region (3 0 UTR) of Xenopus Xcat2 gene can direct RNA localization to the vegetal cortex via the MC in zebrafish oocytes. In addition, the RNA-binding protein Hermes is a component of the MC in zebrafish oocytes, as is the case in Xenopus. Moreover, we provide evidence that the dazl 3 0 UTR possesses at least three types of cis-acting elements that direct multiple steps in the localization process: MC localization, anchorage at the vegetal cortex, and localization at the cleavage furrows. Taken together, the data show that the MC functions as a conserved feature that participates in transport of the germ plasm RNAs in Xenopus and zebrafish oocytes. Furthermore, we propose that the germ plasm components are assembled in a stepwise and spatiotemporally-regulated manner during oogenesis and early embryogenesis in zebrafish. Ó 2007 Elsevier Ireland Ltd. All rights reserved. Keywords: Zebrafish; Germ plasm; RNA localization; Oogenesis; Mitochondrial cloud; Cleavage furrows; DAZ-like (dazl); Vasa; Nanos1; Hermes 1. Introduction In many animals, including Caenorhabditis elegans, Drosophila, and Xenopus, maternal mrna localization plays a crucial role in germ cell formation. Primordial germ cells (PGCs) are formed via the inheritance of a specialized cytoplasm called the germ plasm, which contains an electrondense, non-membrane bound structure called the germinal granule (reviewed in Eddy, 1975; Wylie, 1999). In Drosophila, oskar and nanos mrnas, which are important for germ plasm (pole plasm) assembly and germ cell function, are localized to the posterior pole of the oocyte (Ephrussi et al., 1991; Kim-Ha et al., 1991; Wang and Lehmann, 1991). Similarly, in frogs, the germ plasm is present at * Corresponding author. Tel.: ; fax: address: kunio@kobe-u.ac.jp (K. Inoue). the vegetal pole of oocytes (Ikenishi et al., 1974; Mahowald and Hennen, 1971; Williams and Smith, 1971). There are two major pathways for vegetal localization of maternal mrnas during Xenopus oogenesis, the early and late pathways (Kloc and Etkin, 1995). Vg-1 and VegT mrnas, which encode germ layer determinants, are transported to the vegetal cortex via the late pathway in a microtubule-dependent manner. In contrast, maternal mrnas incorporated into the germ plasm, such as Xcat2 (Zhou and King, 1996), Xpat (Hudson and Woodland, 1998), and Xdazl (Houston et al., 1998), utilize the early or messenger transport organizer (METRO) pathway, by associating with a structure called the mitochondrial cloud (MC), also known as the Balbiani body (reviewed in King et al., 2005; Kloc and Etkin, 2005). The germ plasm RNAs localize by non-directed movement and entrapment within the MC (Chang et al., 2004). Many studies have shown that /$ - see front matter Ó 2007 Elsevier Ireland Ltd. All rights reserved. doi: /j.mod

2 280 K. Kosaka et al. / Mechanisms of Development 124 (2007) the 3 0 untranslated region (3 0 UTR) governs mrna localization. For example, the Xcat2 3 0 UTR possesses a mitochondrial cloud localization element (MCLE) and a germinal granule localization element (GGLE) (Kloc et al., 2000; Zhou and King, 1996). Interestingly, Betley et al. (2002) reported that clusters of CAC-containing motifs are a ubiquitous signal for RNA localization in the early and late pathways. The Xcat2 MCLE contains several copies of the UGCAC motif. Deletion of the motifs from the Xcat2 MCLE abolishes localization (Betley et al., 2002). However, little is known about the trans-acting factors involved in targeting the germ plasm RNAs to the MC, although a few protein constituents of the MC have been identified (King et al., 2005). Recently, it was reported that Xpat protein is involved in germ plasm assembly, although homologs have not been identified outside the genus Xenopus (Machado et al., 2005). In zebrafish, maternal mrnas of germline-specific genes (i.e. vasa, nanos1, dead-end (dnd), askopos (kop) and TDRD7) are localized to the distal ends of the first and second cleavage planes at 4-cell stage (Blaser et al., 2005; Köprunner et al., 2001; Mishima et al., 2006; Weidinger et al., 2003; Yoon et al., 1997). Knock-down experiments using antisense morpholino oligos have shown that nanos1 and dnd genes are essential for proper development of germ cells (Köprunner et al., 2001; Weidinger et al., 2003). Our previous study showed that at this stage, DAZ-like (dazl) and bruno-like (brul) mrnas are also localized to the cleavage furrows in addition to the vegetal pole of oocytes and fertilized eggs (Hashimoto et al., 2004). Electron-microscopic analyses showed that germinal granule-like structures are localized at the furrows of 4-cell embryos, and vasa transcripts are embedded in these structures (Knaut et al., 2000). Moreover, ablation of the cytoplasm at the sites results in the loss of PGCs (Hashimoto et al., 2004). These findings demonstrate that the maternally supplied cytoplasm containing localized mrnas at the cleavage furrows in 4-cell stage embryos functions as germ plasm in zebrafish. It is of great interest to learn the mechanisms that govern localization and assembly of the germ plasm-forming RNAs in zebrafish. Recently, Theusch et al. (2006) reported that a first class of germ plasm RNAs, vasa, nanos1, and dnd, are enriched in a wide cortical band at the animal pole in the freshly laid zebrafish egg, whereas a second class of RNAs that includes vegetally localized dazl mrna translocate along the plane of the cortex towards the animal pole. After recruitment to the cleavage furrows, these two classes of RNAs occupy overlapping but distinct regions of the germ plasm (Theusch et al., 2006). Less is known about the distributions of the germ plasm RNAs during oogenesis. Here, we focused on localization of the germ plasm components, vasa, nanos1, and dazl mrnas, during zebrafish oogenesis. To our surprise, we found that both classes of germ plasm RNAs co-localize with the mitochondrial cloud (MC) and are transported to the vegetal cortex during early oogenesis, and that their distributions change during late oogenesis. We provide evidence that in zebrafish and Xenopus, vegetal localization of germ plasm RNAs is directed by the METRO pathway: Xcat2 mrna is localized to the MC when expressed in zebrafish oocytes and the RNA-binding protein Hermes, a constituent of the MC in Xenopus (Zearfoss et al., 2004), is also observed at the MC in zebrafish. The localization of the germ plasm mrnas is directed by their 3 0 UTRs, and by analyzing transgenic fish expressing GFP mrna fused with various truncated forms of dazl 3 0 UTR, we found that dazl 3 0 UTR possesses multiple cis-elements required for localization. These independent elements direct MC localization, anchorage at the vegetal cortex, and localization to embryonic cleavage furrows. Thus, the results of our study suggest that germ plasm components are assembled in a stepwise and spatiotemporally-regulated manner during oogenesis and early embryogenesis in zebrafish. 2. Results 2.1. Germ plasm RNAs co-localize with the mitochondrial cloud in zebrafish oocytes To identify in detail the distribution patterns of germ plasm RNAs during oogenesis, we performed in situ hybridization on serial sections of the zebrafish ovary. As we previously described (Maegawa et al., 1999; Suzuki et al., 2000), dazl RNA can be observed adjacent to the germinal vesicle at stage I and localized to the vegetal cortex of stage II oocytes (Fig. 1). To our surprise, the distributions of vasa and nanos1 transcripts were quite similar to that of dazl RNA during early oogenesis. In stage I oocytes, vasa and nanos1 RNAs were tightly co-localized with dazl RNA, forming an aggregate adjacent to the germinal vesicle (Fig. 1a d). Subsequently, these maternal RNAs became restricted to the vegetal cortex (Fig. 1e h). However, at stage II, the distribution patterns of vasa, nanos1 and dazl became distinct (Fig. 1i l). By stage II, dazl RNA was localized strictly at the vegetal cortex, whereas vasa RNA formed small particles broadly around the vegetal cortex. nanos1 RNA became to be distributed throughout the oocyte. Later in oogenesis, vasa mrna is present at the cortex as reported previously (Braat et al., 1999; Howley and Ho, 2000) (Fig. 1m). However, we could not observe localization of nanos1 (Fig. 1o). Co-localization of vasa, nanos1 and dazl mrnas in stage I oocytes suggests that the same pathway directs each of these germ plasm RNAs to the vegetal cortex during early oogenesis. The distributions of vasa, nanos1 and dazl RNAs in early oocytes are reminiscent of the METRO (messenger transport organizer) pathway, which is involved in germ plasm formation in Xenopus (Kloc and Etkin, 1995). We therefore hypothesized that a METRO-like pathway exists in zebrafish oocytes and controls vegetal localization of maternal RNAs. To address this hypothesis, we first stained zebrafish oocytes with Mitotracker Red, a

3 K. Kosaka et al. / Mechanisms of Development 124 (2007) Fig. 1. The distributions of germ plasm RNAs during oogenesis. In situ hybridization of serial sections of the zebrafish ovary. Distributions of vasa and dazl RNAs (a, b, e, f, i, j, m and n) and nanos1 and dazl RNAs (c, d, g, h, k, l, o and p) are compared in the serial sections of oocyte at stages I (a h), II (i j), and III (m p). Sections were cut at a thickness of 10 lm. Scale bars, 50 lm. mitochondrial probe. As expected, a mitochondrial cloud was observed adjacent to the germinal vesicle, and seemed to move towards the vegetal cortex in stage I oocytes (Fig. 2A). We next asked if germ plasm RNAs co-localize with the mitochondrial cloud (MC). We found that vasa, nanos1 and dazl RNAs co-localize with the MC (Fig. 2B). In contrast, zor-1/zorba RNA, which is localized to the animal pole (Bally-Cuif et al., 1998; Suzuki et al., 2000), did not co-localize with the MC. These results strongly suggest that these germ plasm RNAs are transported to the vegetal cortex via the mitochondrial cloud (MC) in a METRO-like pathway. In Xenopus, the MCLE in the Xcat2 3 0 UTR directs MC localization (Zhou and King, 1996) and several copies of the hexanucleotide UGCAC within the Xcat2 MCLE are necessary for vegetal localization via the METRO pathway (Betley et al., 2002). To compare the function of the mitochondrial cloud in zebrafish with that in Xenopus, we examined if Xcat2 MCLE RNA is localized to the vegetal cortex when expressed in zebrafish oocytes. Using Tol2 transposon-mediated transgenesis (Kawakami et al., 2004), we generated transgenic zebrafish strains that express a GFP mrna fused with MCLE (Fig. 3a). Strikingly, we found that transcripts containing the Xcat2 MCLE were localized at the MC in stage I oocytes and then to the vegetal cortex in stage III oocytes (Fig. 3b). In contrast, transcripts with MCLEDugcac which lacks all six copies of UGCAC (Betley et al., 2002) were distributed throughout the oocyte but were excluded from the mitochondrial cloud at stage I and did not localize to the vegetal cortex at stage III (Fig. 3b). The observation that the Xcat2 MCLE RNA can be localized to the vegetal cortex in a manner that depends on the UGCAC motifs in zebrafish strongly suggests that the mitochondrial cloud has some conserved role in RNA transport between zebrafish and Xenopus Hermes protein is a constituent of the mitochondrial cloud in zebrafish Next we focused on a candidate for the maternal factor that functions in the METRO-like pathway. It has been reported that the RNA-recognition motif (RRM)-type RNA binding protein Hermes is a constituent of METRO in Xenopus (Zearfoss et al., 2004). We identified zebrafish Hermes (zhermes) from a zebrafish heart cdna library and found that the deduced zebrafish Hermes protein has overall 84% similarity in amino acids as compared to the Xenopus homolog (Fig. 4a). Northern blot analysis revealed that the zhermes transcript is abundantly expressed in the ovary (data not shown). We found that a purified antibody against the Xenopus Hermes protein

4 282 K. Kosaka et al. / Mechanisms of Development 124 (2007) data not shown). These results indicate that zhermes is a maternal protein that is localized to the mitochondrial cloud during early oogenesis in zebrafish, probably in association with germ plasm RNA The 3 0 UTRs of vasa, nanos1 and dazl govern RNA localization during oogenesis and early cleavage stages The results of previous studies based on a transgenic approach indicated that vasa and nanos1 3 0 UTRs direct cleavage furrow localization in 4-cell stage embryos (Blaser et al., 2005; Knaut et al., 2002; Krøvel and Olsen, 2002). To ask if RNA localization is also directed by the 3 0 UTRs during oogenesis, we generated transgenic fish that express a reporter GFP mrna fused with vasa, nanos1 or dazl 3 0 UTRs. The results of in situ hybridization using an antisense GFP probe revealed that the pattern of localization of each of the reporter RNAs was the same as that of the corresponding endogenous RNA in oocytes as well as in early embryos (Fig. 5). In contrast, a reporter RNA fused with the SV UTR was not localized (Fig. 5). These observations strongly suggest that the 3 0 UTRs of vasa, nanos1 and dazl RNAs govern the entire process of their localization Localization elements in the dazl 3 0 UTR Fig. 2. Co-localization of germ plasm RNAs with the mitochondrial cloud in zebrafish oocytes. (A) Stage I oocyes were stained with Mitotracker Red580 (mitochondria) and DAPI (DNA). Schematic representation was shown at the bottom of each panel. In panels a and b, a mitochondria cloud was found adjacent to the germinal vesicle (GV), whereas mitochondria staining was observed at or near the vegetal cortex in panel c. (B) The right panel shows mitochondria staining of a section of a stage I oocyte. For the left panel, in situ hybridization was carried out on the same section as that shown in the right panel using the vasa, nanos1, zdazl and zorba/zor-1 probes. Sections were cut at a thickness of 10 lm. Scale bars, 50 lm. (Zearfoss et al., 2004) is able to recognize Myc-tagged zebrafish Hermes protein expressed in embryos (Fig. 4b). Immunoblot analysis showed that the endogenous zhermes protein can be detected as a single band of approximately 21 kda in ovaries and early embryos (Fig. 4c and data not shown). We investigated the distribution of zhermes during oogenesis by immunostaining of ovarian sections. zhermes was abundant at an area adjacent to the germinal vesicle in stage I oocytes (Fig. 4d). At stage II, zhermes was observed at the vegetal cortex. In addition, zhermes protein co-localized with dazl and vasa RNAs (Fig. 4d and To further define localization element(s), we generated transgenic fish expressing GFP reporter mrnas containing various deleted forms of the dazl 3 0 UTR (Fig. 6A). Northern blot analysis showed that each of the mrnas was efficiently expressed in the transgenic fish (data not shown). The transcripts of deletion mutants D1 and D2 (which contain the 5 0 most regions, a and a + b, respectively) were not localized at any stages (Fig. 6B, and data not shown). By contrast, the deletion mutants D3 D9 were localized to the MC in stage I oocytes. As the D6 andd8 transcripts (which contain regions d and b + c, respectively) were efficiently localized to the MC and vegetal cortex, it is likely that there exist at least two distinct regions in the dazl 3 0 UTR that can direct vegetal localization. It is worth noting that the D7 RNA (which contains region c), which was localized to the MC and vegetal cortex during early oogenesis, was not detected at the vegetal pole after fertilization (Fig. 6B). This result suggests that the D7 RNA lacks a region or regions of the dazl 3 0 UTR required for anchoring of the RNA at the vegetal pole and/or for maintenance of vegetal localization. In contrast to the D7 RNA, the D8 RNA was observed at the vegetal pole in 4-cell embryos (Fig. 6B). Taken together, the data show that region b (nucleotides relative to the stop codon) in the dazl 3 0 UTR required for the vegetal anchorage but cannot direct vegetal localization. We also asked if the RNAs were localized at the distal ends of cleavage furrows in 4-cell embryos. The D4 RNAs was clearly localized to the cleavage furrows, as is true for endogenous dazl RNA (Fig. 6B and 7a). However, we

5 K. Kosaka et al. / Mechanisms of Development 124 (2007) Fig. 3. The Xcat2 3 0 UTR directs localization to the mitochondria cloud in transgenic zebrafish oocytes. (a) Schematic representation of the transgenic construct used to express a GFP mrna fused with Xcat2 3 0 UTR. Xcat2 MCLE or Xcat2 MCLEDugcac 3 0 UTR (Betley et al., 2002) was inserted into the downstream site of EGFP in the T2KXIGDin plasmid vector (Kawakami et al., 2004). Expression of GFP mrna is driven by the EF1a promoter. (b) In situ hybridization in transgenic zebrafish oocytes expressing GFP mrna fused with Xcat2 MCLE (upper panel) or MCLEDugcac (lower panel), using antisense GFP RNA as a probe. Left column, expression of the reporter mrna in a stage I oocyte; middle column, the same oocyte counterstained with Mitotracker. Right column, expression of the reporter mrna in a stage III oocyte. Sections were cut at a thickness of 10 lm. Scale bars, 50 lm. could not easily determine if some of the other mutant RNAs were localized at the cleavage furrows when RNAs were detected in embryos using the NBT/BCIP alkaline phosphatase substrate. Further analysis by fluorescein in situ hybridization showed that D3 D6 and D9 RNAs can be observed at the cleavage furrows of 4-cell embryos, whereas D1, D2andD7 RNAs were not localized to the furrows or to the vegetal pole (Fig. 7B, and data not shown). The D8 RNA (the region b + c) was not observed at the cleavage furrows, despite the fact that it was detected at the vegetal pole (Figs. 6B and 7b), suggesting that region a is involved in localization to the cleavage furrow. 3. Discussion 3.1. The mitochondrial cloud transports germ plasm RNAs in Xenopus and zebrafish oocytes In a detailed investigation of the distribution of maternal vasa, nanos1, and dazl mrnas in zebrafish oocytes, we have found that the RNAs are co-localized with the mitochondrial cloud (MC) and delivered to the vegetal cortex during early oogenesis. The results strongly suggest that the MC plays a key role in the distributions of several germ plasm RNAs early in oogenesis, but at later stages, different mechanisms are likely to be involved, as these RNAs show distinct distributions later in oogenesis. For example, dazl mrna is tightly localized at the vegetal cortex (Maegawa et al., 1999; Suzuki et al., 2000), whereas vasa mrna is distributed throughout the oocyte cortex (Braat et al., 1999; Howley and Ho, 2000). Although nanos1 mrna is undetectable at stage III, some fraction of nanos1 mrna may be localized to the cortex. Distinct pathways for RNA recruitment lead to the compartmentalization of the germ plasm such that a first class of germ plasm RNAs such as vasa and nanos1 and a second class, dazl mrna, occupy overlapping yet distinct regions of the germ plasm in 4-cell embryos (Theusch et al., 2006). Thus it is surprising that both classes of mrnas share a common distribution during early oogenesis. One possibility is that one or more common protein factor binds to both types of RNAs early in oogenesis, which is presumably required for the later assembly of germ plasm in the embryo. Interestingly, Selman et al. (1993) have reported that an electron-dense material termed nuage, which has striking similarity to germinal granules in Xenopus and Drosophila, is detected near the nuclear envelope and among clusters of mitochondria in stage IB zebrafish oocytes. Future study should clarify if the two classes of the germ plasm RNAs and Hermes protein are associated with the nuage in oocytes. In addition, it

6 284 K. Kosaka et al. / Mechanisms of Development 124 (2007) Fig. 4. Hermes protein is a constituent of the mitochondrial cloud. (a) Schematic representation of Hermes proteins in zebrafish and Xenopus (Gerber et al., 1999). RRM represents an RNA-recognition motif. The positions of conserved motifs in RRM, RNP2 and RNP1 (Birney et al., 1993) are indicated. Amino acid sequence identity among Hermes proteins in the N-terminal RRM and C-terminal regions is shown. Numbering according to amino acids in the Hermes proteins. (b) Immunodetection of Myc-tagged zebrafish Hermes using an antibody against Xenopus Hermes. The in vitro synthesized mrna encoding Myc-zHermes was injected into 1-cell embryos and whole lysates of injected embryos were collected at 12 hpf (I). Control lysates from uninjected embryos are shown in (U). Two embryo equivalents per lane were loaded onto the gel. Immunoblotting was done with anti-hermes and anti-myc antibodies. As expected, the Myc-Hermes protein was detected at approximately 30 kda. (c) Immunoblot analysis of endogenous Hermes protein in early stage zebrafish embryos (at 1-cell and 4-cell stages). (d) Distribution of dazl RNA (left) and zhermes protein (right) in the serial sections from stage I and II oocytes (upper and lower panels, respectively). Sections were cut at a thickness of 10 lm. Scale bars, 50 lm. would be informative to determine the distributions of nuage-like structures late in oogenesis. On the other hand, there may be an additional pathway for localization of germ plasm RNAs. Another vegetally localized RNA, bruno-like (brul), is incorporated in the germ plasm at the cleavage furrow (Hashimoto et al., 2004), although its localization pattern is similar to the late pathway in Xenopus (Suzuki et al., 2000). It will be interesting to determine if some common mechanism(s) is involved in brul localization. Generation of transgenic zebrafish provides a powerful method for study of maternal mrna localization. In this study, we have shown that the reporter mrnas containing vasa, nanos1, and dazl 3 0 UTRs faithfully reproduce endogenous patterns of RNA distribution. Also using transgenic experiments, others have shown that the vasa 3 0 UTR governs mrna localization to the germ plasm at the cleavage furrows in zebrafish (Knaut et al., 2002; Krøvel and Olsen, 2002). Moreover, when a GFP mrna with the zebrafish vasa 3 0 UTR is injected into Xenopus oocytes, it is localized to the germ plasm, suggesting the mechanisms of RNA localization machinery in Xenopus and zebrafish have been conserved (Knaut et al., 2002). As we were able to show that the Xenopus Xcat2 MCLE can direct MC localization in zebrafish, our study supports and extends the notion that the transport mechanisms have been conserved. And furthermore, our results indicate that the mitochondrial cloud functions in part of the conserved machinery for localization in these two species. In the Xcat2 MCLE, several copies of the UGCAC motif are responsible for MC localization in Xenopus, as a mutant MCLE that lacks the motifs does not localize to the MC (Betley et al., 2002). As demonstrated here, the motif is also required for localization of the reporter mrna containing Xcat2 3 0 UTR in zebrafish. However, the 3 0 UTRs of endogenous zebrafish germ plasm RNAs do not contain multiple repeats of the UGCAC element. In the case of nanos1, we found that the 5 0 most 200 nucleotides are sufficient for specific expression in primordial germ cells (our unpublished observation). A single copy of the UGCAC motif resides in the region and it will be interesting to learn if the motif is required for the localization of the nanos1 transcript. The vasa 3 0 UTR, by contrast, does not contain the UGCAC motif. Although dazl 3 0 UTR does contain one copy of the motif, our results suggest that it is unlikely that the motif plays a key role in MC localization (see below). Thus, at least for vasa and dazl mrnas, it appears that some distinct motif(s) directs MC localization.

7 K. Kosaka et al. / Mechanisms of Development 124 (2007) Fig. 5. The 3 0 UTRs of vasa, nanos1 and dazl RNAs govern localization during oogenesis and early embryogenesis. In situ hybridization of GFP mrna fused with vasa, nanos1, dazl, orsv UTRs expressed in transgenic zebrafish oocytes (stages I and III) and fluorescein in situ hybridization in 4-cell embryos. GFP-nanos1 mrna was expressed by zpc promoter (Onichtchouk et al., 2003), whereas other mrnas were expressed by EF1a promoter. In stage I oocytes, in situ hybridization (RNA) was performed on sections counter-stained with Mitotracker (Mitochondria). Oocyte sections were cut at a thickness of 10 lm. Scale bars, 50 lm. Lateral and animal views of the 4-cell embryo are shown. It is likely that although the UGCAC repeats-dependent mechanism has been maintained in Xenopus and zebrafish, some additional mechanisms are used for MC localization in zebrafish Localization of dazl mrna to the germ plasm occurs in a stepwise manner We were also able to show here that there are at least two regions in the dazl 3 0 UTR that are capable of directing the entire localization process: nucleotides (regions a c) and (region d) relative to the stop codon. However, when the reporter mrna was fused with the either region alone, localization in 4-cell embryos was not as efficient or tight as that observed when the reporter was fused with the full length 3 0 UTR. Thus it seems likely that two regions have additive or cooperative effects on localization. Furthermore, we found that at least three kinds of cis elements are required for the normal pattern of localization of dazl mrna during oogenesis and early embryogenesis (Fig. 8). In other words, the localization process can be subdivided to multiple steps governed by independent regions of the 3 0 UTR. The D7 RNA, which contains region c, was localized to the vegetal cortex via association with the MC, although it does not seem to be anchored at the cortex. The D8 RNA, which contains regions b and c, was localized and anchored stably at the vegetal cortex but was not observed at the cleavage furrows. Finally, the D3 RNA, which contains regions a, b and c, was clearly localized to the furrows. Therefore, we think it likely that the MC localization element resides in region c (nucleotides relative to the stop codon); the vegetal anchorage element resides in region b (nucleotides ); and the element responsible for translocation to the cleavage furrows resides in region a (nucleotides 1 199), although the latter two regions, a and b, cannot function without the prior action of the MC element (region c). Since the UGCAC motif resides in region b (at nucleotides ) in the dazl 3 0 UTR, it seems unlikely that the motif is the MC localization element. It was previously reported that vegetally localized dazl RNA translocates to the animal pole along the cortex and moves to the cleavage furrows during early cleavage stages (Theusch et al., 2006). The finding that the D8 RNA efficiently localized to the vegetal cortex but not the cleavage furrows suggests that the vegetal localization of dazl RNA is necessary but not sufficient for later targeting to the cleavage furrows. One possible explanation of our results is that the D8 RNA lacks the ability to translocate along the cortex after fertilization. Although it remains elusive how dazl RNA translocates to the blastoderm, association with cytoskele-

8 286 K. Kosaka et al. / Mechanisms of Development 124 (2007)

9 K. Kosaka et al. / Mechanisms of Development 124 (2007) Fig. 7. Distribution of a reporter mrna fused with mutant dazl 3 0 UTRs in 4-cell embryos. (a) In situ hybridization of D4 RNA. Animal view of the 4-cell embryo shown in Fig. 6b. (b) Fluorescein in situ hybridization in transgenic 4-cell embryos expressing GFP mrna fused with full-length (FL), D2, D3, D6, or D8 dazl 3 0 UTRs. The furrow region is enlarged. Fig. 8. The dazl mrna localization pathway in zebrafish. Early in zebrafish oogenesis, dazl mrna (red) is transported to the vegetal pole via the mitochondrial cloud (blue), and anchored at the vegetal cortex. Localization of vasa and nanos1 mrnas are also delivered via the MC, although the distribution of these mrnas changes late in oogenesis (not shown in the figure). Upon activation of the egg, vegetally localized dazl mrna translocates and dazl mrna is assembled into the germ plasm at embryonic cleavage furrows. Cis-acting elements in region a c of dazl 3 0 UTR are shown schematically. The UGCAC motif resides in region b (closed triangle). The region d is also sufficient to direct the entire localization process (see Fig. 6). tal structures such as microtubules may be important (Gore and Sampath, 2002; Jesuthasan and Stahle, 1997). It is also possible that although the D8 RNA can translocate along the cortex towards the blastodisc, this mutant RNA cannot associate with cytoskeletal machinery necessary for transport to the cleavage furrows. In support of b this idea, some aggregates of D8 RNA were observed around the blastodisc of transgenic embryos, suggesting that the D8 RNA does translocate along the cortex (our unpublished observation). Clearly, learning more about the mechanisms that direct the distinct steps in mrna localization will be of interest in the future. Fig. 6. Identification of cis-elements required for localization of dazl mrna. (a) Schematic representation of deletions of zebrafish dazl 3 0 UTR (D1 D9) used in transgenic experiments. Numbers represent nucleotide positions relative to the stop codon. Regions of the dazl 3 0 UTR labeled a, b, c and d contain nucleotides 1 199, , , and , respectively. The results of mrna distribution analyses are indicated to the right of the schematic diagram. (+), normal pattern of distribution; ( ), not localized. MC shows MC localization in oocytes. Vg shows localization at vegetal pole in 4-cell embryos. (b) Detection of the GFP mrna fused with mutant 3 0 UTRs in transgenic oocytes (stages I and III) and 4-cell embryos. For stage I oocytes, in situ hybridization was performed on sections counterstained with Mitotracker to detect mitochondria. Oocyte sections were cut at a thickness of 10 lm. Scale bars, 50 lm. A lateral view of the 4-cell embryo is shown.

10 288 K. Kosaka et al. / Mechanisms of Development 124 (2007) Experimental procedures 4.1. Animals and care Zebrafish were raised, maintained and bred as described (Hashimoto et al., 2004). Staging was according to (Kimmel et al., 1995). The AB strain and a local strain were used for microinjections Generation of transgenic zebrafish To generate transgenic fish, we used the Tol2 transposon system (Kawakami et al., 2004). Tol2 transposase mrna was synthesized in vitro as described previously (Kawakami et al., 2004). Approximately 1 nl of DNA/RNA solution containing 25 ng/ll circular DNA of T2KXIGDin and 25 ng/ll Tol2 transposase mrna was injected into fertilized eggs using a microinjector (Narishige IM300) Plasmid construction To construct the EF1a-EGFP-3 0 UTR plasmids, EGFP and the SV40 region of the T2KXIGDin vector (Kawakami et al., 2004) were amplified by PCR using the following primers and the amplified fragments were then cloned into the pbs plasmid. 5 0 EGFP: CGCGGATCCATGGTGAGCAAGGGCGAGGA 3 0 EGFP: TCCCCCGGGTTAATTAATTACTTGTACAGCTCGTC 5 0 SV40: CCGCTCGAGATCCAGACATGATAAGATAC 3 0 SV40: CGGGGTACCAGATCTGATCTAGAGGATCA The 3 0 UTRs of vasa (Mishima et al., 2006) and dazl (Maegawa et al., 2002) were amplified by PCR using the following primers and cloned into the pbs-egfp-sv40 plasmid. The resulting pbs-egfp-3 0 UTR-SV40 plasmids were cut with BamHI and BglII, and EGFP-3 0 UTR fragments were inserted into T2KXIGDin. vasa 5 0 : CCGGAATTCCTGGCCTCACACCTGTTATA vasa 3 0 : CCGCTCGAGGTCACCAGTATCCGTCTTTA dazl 5 0 : CCGGAATTCCCTTCACTACTAGCACAGAC dazl 3 0 : CCGCTCGAGCCAAATTAATTAAACTTTGT The dazl 3 0 UTR deletion mutants were constructed as follows. To generate the D1 3 mutant forms, HindIII XhoI, NdeI XhoI and EcoRV XhoI fragments, respectively, were removed from pbs-egfp-dazl 3 0 UTR- SV40. To generate D4 6 mutant forms, the EcoRI HindIII, EcoRI NdeI, and EcoRI EcoRV fragment were deleted from the 3 0 UTR clone, respectively. To generate D7, the EcoRI NdeI fragment was removed from pbs- EGFP-D3. To generate D8 and D9, the EcoRV XhoI and NdeI EcoRV fragments, respectively, were removed from pbs-egfp-d4. The BamHI BglII fragments of the resulting plasmids were inserted into T2KXIGDin. The MCLE and MCLEDugcac fragments of Xcat2 3 0 UTR from p18. (partial)xcatle WT and ptz18u (partial) Xcat2 MCLE Delta all TGCAC, respectively, (Betley et al., 2002) were cloned into pbs- EGFP-SV40. To generate the zpc-egfp-3 0 UTR plasmids, the zpc promoter region (Onichtchouk et al., 2003) was amplified using the following primers. The amplified fragment was cloned upstream of the EGFP coding sequence in the T2KXIGDin vector replacing the EF1a promoter. 5 0 zpc:ccgctcgagaaaatccccatgacatgctg 3 0 zpc:acgcgtcgacattgcctgctgactaatta The nanos1 3 0 UTR (Mishima et al., 2006) was cloned into pbs-egfp- SV40 using the following primers. The XhoI SalI fragment of pbs-egfpnanos1 3 0 UTR-SV40 was then inserted into the T2KXIGDin vector. The resulting plasmid does not contain the SV40 sequence. 5 0 nanos1: CCGGAATTCAGCGGACATTGATGCTCCGG 3 0 nanos1: CCGCTCGAGAAGTCTAGAGAAAATGTTTA 4.4. Preparation of ovarian sections Ovaries were dissected from anaesthetized female fish. For detection of mitochondrial aggregates, ovaries were immersed in 50 lg/ll Mito- Tracker Red 580 (Molecular Probes) in Leivobitz medium (Sigma) for 10 min at room temperature. After washes with PBST (0.1% Tween 20 in PBS), the ovaries were fixed with MEMFA (0.1 M MOPS, 2 mm EGTA, 1 mm MgSO4, 3.7% Formalin) overnight at 4 C. The ovaries were then washed with PBST and immersed in a graded series of sucrose solutions up to 30%, then embedded in Tissue-Tek (Sakura). Cryostat sections were cut at a thickness of 10 lm using a microtome (Leica CM1900). Oocyte staging was according to Selman et al. (1993) In situ hybridization and immunochemistry In situ hybridization of DIG labeled anti-sense probes was performed as described previously (Maegawa et al., 1999). FastRed was used for fluorescein in situ hybridization. For immunohistochemistry, embryos were fixed with MEMFA overnight at 4 C. After fixation, the embryos were incubated with 4% Block Ace (Dainippon Pharmaceutical) for 1 h, followed by incubation in the blocking solution containing the primary antibody (1/100 dilution) for three hours at room temperature. The embryos were washed with PBST, and incubated with the second antibody (1/500 dilution) overnight at 4 C. After washing in PBST for 30 min, the embryos were observed using a confocal microscope (Zeiss, LSM 5 Pascal) SDS PAGE and immunoblotting By screening a zebrafish heart kzapii cdna library (Stratagene), the hermes cdna was identified (DDBJ Accession No. AB080736). The PCR amplified fragment containing the zhermes coding region was inserted into pcs2+ MT (Rupp et al., 1994). RNA for injection was synthesized using the mmessage mmachine Sp6 in vitro transcription kit (Ambion). RNA was injected into fertilized eggs and lysates were recovered at 12 hpf and mixed with SDS sample buffer (50 mm Tris HCl ph 7.5, 150 mm NaCl, 1 mm EDTA, 0.1% Tween 20, 1 Protein Inhibitor Cocktail (Roche)). Each sample was separated by 12% SDS PAGE and blotted on to a nitrocellulose membrane. The membranes were then incubated with anti-myc monoclonal antibody 9E10 (1/1000 dilution; SantaCruz) or Hermes antibody (1/500 dilution; Zearfoss et al., 2004) in blocking solution (2.5% skimmed milk in PBST), respectively. Anti-mouse IgG conjugated to HRP (Amersham Biosciences) and anti-rabbit IgG conjugated to HRP were used at 1/2000 and 1/1000, respectively, as secondary antibodies and were then detected using ECL Western Blotting Detection reagents (Amersham Biosciences). Acknowledgements We are thankful to Drs. Wolfgang Driever, James O. Deshler, and Malgorzata Kloc for providing plasmids and antibodies, Dr. Karuna Sampath for helpful advice, and Drs. Kunio Yasuda, Yoshiko Takahashi, Naoaki Saito and Hiroshi Sasaki for their help. We also thank to Drs. Toshinobu Fujiwara, Yuji Kageyama, Shingo Maegawa, Yoshiko Hashimoto and Yui Jin, Mr. Shinya Yamamoto, and Ms. Michiyo Nakafuji for their assistance and helpful discussions. This work was supported by Grants-in-Aid from MEXT and JSPS, and in part by Mitsubishi Foundation and Hyogo Science and Technology Association.

11 K. Kosaka et al. / Mechanisms of Development 124 (2007) Kyoko Kosaka was supported by a fellowship from Iue Foundation. References Bally-Cuif, L., Schatz, W.J., Ho, R.K., Characterization of the zebrafish Orb/CPEB-related RNA binding protein and localization of maternal components in the zebrafish oocyte. Mech. Dev. 77, Betley, J.N., Frith, M.C., Graber, J.H., Choo, S., Deshler, J.O., A ubiquitous and conserved signal for RNA localization in chordates. Curr. Biol. 12, Birney, E., Kumar, S., Krainer, A.R., Analysis of the RNArecognition motif and RS and RGG domains: conservation in metazoan pre-mrna splicing factors. Nucleic. Acids Res. 21, Blaser, H., Eisenbeiss, S., Neumann, M., Reichman-Fried, M., Thisse, B., Thisse, C., Raz, E., Transition from non-motile behaviour to directed migration during early PGC development in zebrafish. J. Cell Sci. 118, Braat, A.K., Zandbergen, T., van de Water, S., Goos, H.J., Zivkovic, D., Characterization of zebrafish primordial germ cells: morphology and early distribution of vasa RNA. Dev. Dyn. 216, Chang, P., Torres, J., Lewis, R.A., Mowry, K.L., Houliston, E., King, M.L., Localization of RNAs to the mitochondrial cloud in Xenopus oocytes through entrapment and association with endoplasmic reticulum. Mol. Biol. Cell 15, Eddy, E.M., Germ plasm and the differentiation of the germ cell line. Int. Rev. Cytol. 43, Ephrussi, A., Dickinson, L.K., Lehmann, R., Oskar organizes the germ plasm and directs localization of the posterior determinant nanos. Cell 66, Gerber, W.V., Yatskievych, T.A., Antin, P.B., Correia, K.M., Conlon, R.A., Krieg, P.A., The RNA-binding protein gene, hermes, is expressed at high levels in the developing heart. Mech. Dev. 80, Gore, A.V., Sampath, K., Localization of transcripts of the zebrafish morphogen Squint is dependent on egg activation and the microtubule cytoskeleton. Mech. Dev. 112, Hashimoto, Y., Maegawa, S., Nagai, T., Yamaha, E., Suzuki, H., Yasuda, K., Inoue, K., Localized maternal factors are required for zebrafish germ cell formation. Dev. Biol. 268, Houston, D.W., Zhang, J., Maines, J.Z., Wasserman, S.A., King, M.L., A Xenopus DAZ-like gene encodes an RNA component of germ plasm and is a functional homologue of Drosophila boule. Development 125, Howley, C., Ho, R.K., mrna localization patterns in zebrafish oocytes. Mech. Dev. 92, Hudson, C., Woodland, H.R., Xpat, a gene expressed specifically in germ plasm and primordial germ cells of Xenopus laevis. Mech. Dev. 73, Ikenishi, K., Kotani, M., Tanabe, K., Ultrastructural changes associated with UV irradiation in the germinal plasm of Xenopus laevis. Dev. Biol. 36, Jesuthasan, S., Stahle, U., Dynamic microtubules and specification of the zebrafish embryonic axis. Curr. Biol. 7, Kawakami, K., Takeda, H., Kawakami, N., Kobayashi, M., Matsuda, N., Mishina, M., A transposon-mediated gene trap approach identifies developmentally regulated genes in zebrafish. Dev. Cell 7, Kim-Ha, J., Smith, J.L., Macdonald, P.M., oskar mrna is localized to the posterior pole of the Drosophila oocyte. Cell 66, Kimmel, C.B., Ballard, W.W., Kimmel, S.R., Ullmann, B., Schilling, T.F., Stages of embryonic development of the zebrafish. Dev. Dyn. 203, King, M.L., Messitt, T.J., Mowry, K.L., Putting RNAs in the right place at the right time: RNA localization in the frog oocyte. Biol. Cell 97, Kloc, M., Bilinski, S., Pui-Yee Chan, A., Etkin, L.D., The targeting of Xcat2 mrna to the germinal granules depends on a cis-acting germinal granule localization element within the 3 0 UTR. Dev. Biol. 217, Kloc, M., Etkin, L.D., Two distinct pathways for the localization of RNAs at the vegetal cortex in Xenopus oocytes. Development 121, Kloc, M., Etkin, L.D., RNA localization mechanisms in oocytes. J. Cell Sci. 118, Knaut, H., Pelegri, F., Bohmann, K., Schwarz, H., Nüsslein-Volhard, C., Zebrafish vasa RNA but not its protein is a component of the germ plasm and segregates asymmetrically before germline specification. J. Cell Biol. 149, Knaut, H., Steinbeisser, H., Schwarz, H., Nüsslein-Volhard, C., An evolutionary conserved region in the vasa 3 0 UTR targets RNA translation to the germ cells in the zebrafish. Curr. Biol. 12, Köprunner, M., Thisse, C., Thisse, B., Raz, E., A zebrafish nanosrelated gene is essential for the development of primordial germ cells. Genes Dev. 15, Krøvel, A.V., Olsen, L.C., Expression of a vas::egfp transgene in primordial germ cells of the zebrafish. Mech. Dev. 116, Machado, R.J., Moore, W., Hames, R., Houliston, E., Chang, P., King, M.L., Woodland, H.R., Xenopus Xpat protein is a major component of germ plasm and may function in its organisation and positioning. Dev. Biol. 287, Maegawa, S., Yamashita, M., Yasuda, K., Inoue, K., Zebrafish DAZ-like protein controls translation via the sequence GUUC. Genes Cells 7, Maegawa, S., Yasuda, K., Inoue, K., Maternal mrna localization of zebrafish DAZ-like gene. Mech. Dev. 81, Mahowald, A.P., Hennen, S., Ultrastructure of the germ plasm in eggs and embryos of Rana pipiens. Dev. Biol. 24, Mishima, Y., Giraldez, A.J., Takeda, Y., Fujiwara, T., Sakamoto, H., Schier, A.F., Inoue, K., Differential regulation of germline mrnas in soma and germ cells by zebrafish mir-430. Curr. Biol. 16, Onichtchouk, D., Aduroja, K., Belting, H.G., Gnugge, L., Driever, W., Transgene driving GFP expression from the promoter of the zona pellucida gene zpc is expressed in oocytes and provides an early marker for gonad differentiation in zebrafish. Dev. Dyn. 228, Rupp, R.A., Snider, L., Weintraub, H., Xenopus embryos regulate the nuclear localization of XMyoD. Genes Dev. 8, Selman, K., Wallace, R.A., Sarka, A., Qi, X., Stages of oocyte development in the zebrafish, Brachydanio rerio. J. Morphol. 218, Suzuki, H., Maegawa, S., Nishibu, T., Sugiyama, T., Yasuda, K., Inoue, K., Vegetal localization of the maternal mrna encoding an EDEN-BP/Bruno-like protein in zebrafish. Mech. Dev. 93, Theusch, E.V., Brown, K.J., Pelegri, F., Separate pathways of RNA recruitment lead to the compartmentalization of the zebrafish germ plasm. Dev. Biol. 292, Wang, C., Lehmann, R., Nanos is the localized posterior determinant in Drosophila. Cell 66, Weidinger, G., Stebler, J., Slanchev, K., Dumstrei, K., Wise, C., Lovell- Badge, R., Thisse, C., Thisse, B., Raz, E., dead end, a novel vertebrate germ plasm component, is required for zebrafish primordial germ cell migration and survival. Curr. Biol. 13, Williams, M.A., Smith, L.D., Ultrastructure of the germinal plasm during maturation and early cleavage in Rana pipiens. Dev. Biol. 25, Wylie, C., Germ cells. Cell 96, Yoon, C., Kawakami, K., Hopkins, N., Zebrafish vasa homologue RNAis localizedtothe cleavageplanesof2- and4-cell-stageembryosand is expressed in the primordial germ cells. Development 124, Zearfoss, N.R., Chan, A.P., Wu, C.F., Kloc, M., Etkin, L.D., Hermes is a localized factor regulating cleavage of vegetal blastomeres in Xenopus laevis. Dev. Biol. 267, Zhou, Y., King, M.L., Localization of Xcat-2 RNA, a putative germ plasm component, to the mitochondrial cloud in Xenopus stage I oocytes. Development 122,