We studied the early development of zooxanthellae-containing eggs of the scleractinian corals Porites cylindrica and Montipora digitata to elucidate how zooxanthellae become localized to the endoderm of planulae during the course of development. In both species, zooxanthellae were distributed evenly in the oocytes and delivered almost equally to the blastomeres during cleavage. In P. cylindrica, gastrulation occurred via delamination or ingression, and blastomeres containing zooxanthellae dropped into the blastocoel during gastrulation. Thus, zooxanthellae were restricted to the endodermal cells at the gastrula or early planula stage in P. cylindrica. In M. digitata, gastrulation occurred by a combination of invagination and epiboly to form a somewhat concave gastrula. Zooxanthellae were present in both endodermal and ectodermal cells of early planulae, but they disappeared from the ectoderm as the planulae matured. In our previous study on two species of Pocillopora, we found that zooxanthellae were localized in eggs as well as in embryos, and that blastomeres containing zooxanthellae later dropped into the blastocoel to become restricted to the endoderm (Hirose et al., 2000). The timing and mechanism of zooxanthella localization and types of gastrulation differed among species belonging to the three genera. These results suggest that zooxanthella localization in the embryos reflects the timing of the determination of presumptive endoderm cells and/or specificity of zooxanthellae toward presumptive endoderm cells.
Reef-building corals harbor intracellular symbiotic dino-flagellates called zooxanthellae in their endodermal cells. Although the vast majority of reef-building corals spawn gametes that lack zooxanthellae (Babcock et al., 1986), some hermatypic corals spawn eggs containing zooxanthellae (Kojis and Quinn, 1982; Babcock and Heyward, 1986; Heyward et al., 1987; Tomascik and Sander, 1987; Yeemin, 1988; Glynn et al., 1991, 1994; Kinzie, 1993, 1996; Sier and Olive, 1994; Kruger and Schleyer, 1998; Neves, 2000; Hirose et al., 2001). It is not known how zooxanthellae delivered to oocytes become restricted to the endodermal cells during the course of development. Although the early development of scleractinian corals has been described in various species (e.g., Szmant-Froelich et al. (1980), 1985; Babcock and Heyward, 1986; Harrison and Wallace, 1990), the early development of zooxanthella-containing eggs has been reported only in spawning species, such as Montipora effusa (Yeemin, 1998), M. verrucosa (Maté et al., 1998), and Pocillopora verrucosa and P. eydouxi (Hirose et al., 2000), and in a brooder, Porites porites (Tomascik and Sander, 1987).
Although zooxanthellae are generally restricted to endodermal cells in adult corals, zooxanthellae are at least temporarily observed in the ectoderm of planulae in some stony and soft corals (Szmant-Froelich et al., 1985; Benayahu et al., 1988; Benayahu, 1997; Benayahu and Schleyer, 1998; Schwarz et al., 1999). This is probably because algal infection first occurred in the ectodermal cells of embryos or early planulae (Szmant-Froelich et al., 1985) or because dividing cells at these stages transferred the multiplying symbionts to their daughter cells, including presumptive ectodermal cells (Benayahu, 1997; Benayahu and Schleyer, 1998). In the latter case, zooxanthellae were transferred from the ectoderm to the endoderm across the mesoglea before the larvae developed into mature planulae (Benayahu, 1997; Benayahu and Schleyer, 1998). In the scyphozoan Linuche unguiculata, Montgomery and Kremer (1995) observed symbiotic algae mainly in the ectodermal cells of early embryos, but the number of algae in the endoderm increased with time. They suggested that ectodermal cells containing symbiotic algae might migrate to the endoderm of planulae.
The corals P. verrucosa and P. eydouxi release eggs containing zooxanthellae. In these corals, zooxanthellae are concentrated in one hemisphere of the egg and are not equally delivered to all daughter cells. Furthermore, blastomeres containing zooxanthellae drop into the blastocoel during gastrulation, and zooxanthellae become restricted to the endoderm in the gastrula or early planula (Hirose et al., 2000). In a temperate zooxanthellate sea anemone, Anthopleura ballii, blastomeres containing zooxanthellae were observed to invaginate at one end of the embryo, leading to the restriction of zooxanthellae to the planula endoderm (Davy and Turner, 2003). Thus, the larvae of these corals and this sea anemone do not transfer the algae from the ectoderm to the endoderm.
The corals Porites cylindrica and Montipora digitata spawn zooxanthellate eggs, but the zooxanthellae are distributed evenly within the eggs. We studied the early development of these corals and compared the timing and mechanism of zooxanthella localization to the endoderm of planulae in these two corals with those reported for two Pocillopora species, in which a localized distribution of zooxanthellae occurs in oocytes (Hirose et al., 2001). We describe differences in the timing and mechanism of zooxanthella localization among the embryos of species belonging to the three genera. We also discuss the way in which gastrulation and the distribution pattern of zooxanthellae in eggs are related to the timing of zooxanthella localization to the endoderm in coral larvae.
MATERIALS AND METHODS
Collection of coral colonies
Branches were collected from colonies of M. digitata and P. cylindrica a few days before the full moon in May, June, and/or July from 1995 to 2000, from the reefs at Sesoko Island and Bise, northern Okinawa Island, Japan. The branches were maintained in an outdoor tank supplied with unfiltered running seawater.
Observation of early development
While the gonochoric colonies of P. cylindrica released individual eggs or sperm, the hermaphroditic colonies of M. digitata released egg–sperm bundles at night a few days after the full moon (Heyward et al., 1987). To collect gametes, the branches were placed separately into plastic containers before the expected spawning time, at about 1930 h (M. digitata) and 2230 h (P. cylindrica). Released gametes were collected by sucking up seawater from the container with a large plastic pipette. The gametes from each species were placed in separate plastic beakers, with gametes from two or three colonies mixed in a single beaker (100–300 ml suspension each). Filtered (0.45 μm) seawater (FSW) was added to the beaker to a final volume of 1 or 2 L. For M. digitata, the concentration of gametes was about 1 bundle/ml (sperm: 10−4– 10−5 cells/ml). Fertilized eggs were kept in FSW at room temperature (28–30°C). Eggs and embryos were sampled and observed under a light microscope at intervals of 30 min to 1 h, and photomicrographs were taken (Nikon Microphot; Nikon, Tokyo, Japan).
Histology and transmission electron microscopy
Eggs and embryos were put into a microtube and allowed to sink to the bottom. The supernatant was discarded, and the specimens were fixed for at least 2 h in 2.5% glutaraldehyde in 0.1 M cacodylate buffer (pH 7.4) containing 3% NaCl. The specimens were rinsed three times in the same buffer and post-fixed in 1% osmium tetroxide in the same buffer for 1 h on ice. The specimens were then dehydrated in a graded series of acetone, immersed in n-butyl glycidyl ether (QY1), and embedded in Spurr's resin. For light microscopic observation, sections (0.5–1 μm thick) were stained with 1% methylene blue and 1% azur II in 1% borax. For electron microscopy, silver or gold sections were stained with uranyl acetate and lead citrate and observed under a JEM-2000EX electron microscope (JEOL, Tokyo, Japan) at an acceleration voltage of 100 kV.
Early development of Porites cylindrica
Fifteen of 16 branches of P. cylindrica spawned gametes repeatedly for 2 to 6 days after the full moon in June 2000. The branches began releasing gametes 3 h after sunset (2230 h), and spawning persisted for 1 to 1.5 h. Eleven of these 15 branches were female. The female colonies released buoyant eggs, which were about 220 μm in diameter and contained about 400 zooxanthellae (Table 1). Zooxanthellae occupied about 2.6% of the egg volume. Male colonies released sperm at almost the same time.
Various parameters of spawned eggs of Pocillopora verrucosa, P. eydouxi, Porites cylindrica and Montipora digitata in 1998–2000. Means±SD.
Zooxanthellae were located evenly in spawned eggs of P. cylindrica (Fig. 1A). Cleavage occurred at intervals of 30 to 40 min. The zooxanthellae in the embryo were delivered almost equally to the blastomeres (Fig. 1B). At the morula stage, each daughter cell contained only a few zooxanthellae (Fig. 1C). At 6 h after fertilization, a hollow blastula was formed (Fig. 1D). During gastrulation, blastomeres possessing zooxanthellae accumulated around the blastocoel, and the surface layer of the embryo was occupied by relatively large, columnar cells without zooxanthellae (Fig. 1E, F). At about 10 h after fertilization, a stereogastrula was formed (Fig. 1G). Blastomeres containing zooxanthellae and/or lipid droplets filled the inner part of the embryo and the endoderm, and the ectoderm without zooxanthellae appeared to differentiate, although the mesoglea was not clearly seen (Fig. 1H). A mouth opening was observed as an invagination of the ectoderm (Fig. 1H). Ciliated larvae started to swim 24 h after fertilization. At this stage, the ectoderm and endoderm were demarcated by the mesoglea, and zooxanthellae were located in the endoderm (Fig. 1I). A gastrovascular cavity was formed.
Early development of Montipora digitata
Seven of 29 branches of M. digitata spawned egg–sperm bundles on the third to fifth night after the full moon during May–June 1999, and 11 of 40 branches spawned for 7 days after the full moon in June–July 1999. In 2000, all 12 branches collected spawned gametes for 10 days, beginning 1 day before the full moon in June. The spawning started about 1 h after sunset (between 2000 and 2100 h) and persisted for 1 to 1.5 h. Each egg–sperm bundle consisted of 10 to 15 eggs and sperm. The eggs were about 380 μm in diameter and contained about 1,400 zooxanthellae (Table 1), which occupied about 2.0% of the egg volume.
Zooxanthellae were distributed more or less homogeneously in the cytoplasm of M. digitata eggs (Fig. 2A). The first cleavage began about 2.5 h after fertilization. From the early cleavage stage to the morula stage, each blastomere contained an almost equal number of zooxanthellae (Fig. 2B, C). A hollow blastula was formed, although the blastocoel was filled with some acellular material (Fig. 2D). The blastula was composed of columnar blastomeres, and zooxanthellae were located in the basal region of the columnar cells. Gastrulation occurred by invagination, and one side of the embryo became concave (Fig. 2E, F). Two types of blastomeres (Fig. 2F) were observed, one containing a few large lipid droplets (arrows) and another containing many small lipid droplets (small arrows). The blastomeres of the latter type were aggregated on one side of the blastopore lip. At a later stage, the blastocoel disappeared, and the ectoderm and endoderm could be discerned (Fig. 2G). The ectoderm consisted of columnar cells, and the endoderm comprised cells containing large lipid droplets. Zooxanthellae were observed in both the ectoderm and endoderm. The ectoderm layer appeared to envelop the endodermal cells (Fig. 2G, H). The embryos then became spherical in shape again and formed a stereogastrula (Fig. 2I). Zooxanthellae were present in both the ectoderm and the endoderm, even in the late gastrula or early planula (Figs. 2J, 3A). Most of the zooxanthellae in the ectoderm were located in the basal region of the ectodermal cells (Fig. 3B). The mesoglea was very thin (about 0.3 μm) and was obscured or lost in some places (Fig. 3C, D). In these cases, cellular processes of ectodermal cells had entered the endoderm through a gap in the mesoglea (Fig. 3C), or zooxanthellae were located very close to the obscured mesoglea (Fig. 3D). As the planulae matured, zooxanthellae disappeared from the ectoderm, and were then seen only in the endoderm of mature planulae (Fig. 3K). Planulae settled and metamorphosed 3–4 days after fertilization. At about 6 days after settlement, the polyps began secreting a skeleton, and tentacles were discernible (Fig. 2L).
In adult colonies of hermatypic corals, only endodermal cells contain zooxanthellae, although some hermatypic corals release oocytes that contain them. Hence, zooxanthellae must become restricted to the endoderm at a certain stage of development. We studied the early development of zooxanthellate eggs in P. cylindrica and M. digitata to investigate how the timing of zooxanthella localization to the endoderm differs among these two species and two species of Pocillopora characterized previously (Hirose et al., 2000).
Early cleavage and distribution of zooxanthellae
In P. verrucosa and P. eydouxi, zooxanthellae moved toward the animal pole 1–2 days before spawning and became concentrated in one hemisphere of the oocytes (Hirose et al., 2001). The first cleavage apportioned zooxanthellae more or less equally between the first two blastomeres. At the second cleavage, however, two of the four blastomeres received almost all of the zooxanthellae, and the other two had few or none. This uneven distribution of zooxanthellae persisted until the zygotes developed into gastrulae (Hirose et al., 2000). In contrast to the two species of Pocillopora, P. cylindrica and M. digitata exhibited an even distribution of zooxanthellae in their eggs, and the zooxanthellae were delivered almost equally to the blastomeres until blastula formation.
Blastoderm differentiation in the two species
In both P. cylindrica and M. digitata, fertilized eggs developed into a hollow blastula, as occurs in the two species of Pocillopora (Hirose et al., 2000). In P. cylindrica, when the blastocoel disappeared and an acoel embryo was formed (Fig. 1G, H), the ectoderm and endoderm appeared to be at least partially differentiated, although they were not clearly demarcated by a mesogleal layer. The embryo at this stage might be considered a gastrula rather than a blastula. A mouth later formed by invagination of the ectoderm, and the ectoderm and endoderm became separated by a mesogleal layer. This process is considered to be planula development rather than gastrulation (Harrison and Wallace, 1990).
In M. digitata, the blastocoel was filled with some acellular material, which probably consisted of lipid droplets that had been expelled from blastomeres and accumulated in the blastocoel. This material was likely incorporated later by endodermal cells in the course of invaginating gastrulation. When the blastocoel disappeared upon invagination, the ectoderm and endoderm differentiated, and the endodermal cells became filled with large lipid droplets.
Gastrulation and zooxanthella relocation
In P. cylindrica as well as the two species of Pocillopora, blastomeres containing zooxanthellae and/or lipid droplets detached from the outer layer and dropped into the blastocoel until it was completely filled, resulting in a stereogastrula. In these corals, zooxanthella localization to the endoderm occurred during gastrulation (Fig. 4). Gastrulation was by multipolar ingression or delamination rather than invagination. Gastrulation through delamination has been suggested in Astrangia danae (Szmant-Froelich et al., 1980), Favia fragum (Szmant-Froelich et al., 1985), and M. verrucosa (Maté et al., 1998).
In M. digitata, gastrulation appeared to occur via the combined processes of invagination and epiboly. Zooxanthellae were present in both the endoderm and ectoderm of the late gastrula or early planula. Some presumptive endoderm cells may remain in the ectoderm of M. digitata planulae, or the specificity of zooxanthellae toward presumptive endoderm cells may be low or develops at a later stage in M. digitata compared to P. cylindrica and the two Pocillopora species. The zooxanthellae gradually disappeared from the ectoderm as the planulae matured.
The timing of zooxanthella localization differed among P. cylindrica, M. digitata, and the two species of Pocillopora (Fig. 4). This difference in timing may be related to the mode of gastrulation. However, in the temperate sea anemone Anthopleura ballii, which produces zooxanthellate eggs, zooxanthellae are localized at one end of the blastula, and the blastomeres containing zooxanthellae invaginate during gastrulation, restricting the zooxanthellae to the endoderm of planulae (Davy and Turner, 2003). The uneven distribution of zooxanthellae in embryos could indicate an earlier determination of presumptive endodermal cells. Zooxanthellae might be delivered more or less exclusively to presumptive endodermal cells during early developmental stages in the two Pocillopora species and the sea anemone A. ballii. It is likely that zooxanthella localization in embryos is related to the timing of the determination of presumptive endodermal cells, with the localization of zooxanthellae to the endoderm occurring no earlier than gastrulation.
Possible mechanisms of zooxanthella translocation from ectoderm to endoderm
Zooxanthellae have been observed in the ectoderm of the early planulae of some corals (F. fragum: Szmant-Froelich et al., 1985; Fungia scutaria: Schwarz et al., 1999), soft corals (Xenia umbellata: Benayahu et al., 1988; Litophyton arboreum: Benayahu et al., 1992, Benayahu, 1997; A. glauca: Benayahu and Schleyer, 1998), and the scyphozoan L. unguiculata (Montgomery and Kremer, 1995). In these cnidarians, zooxanthellae generally reside only in the endoderm in later stages of the life cycle; thus, zooxanthellae in the ectoderm might be translocated to the endoderm or removed from the ectoderm. Several mechanisms have been suggested for the translocation of zooxanthellae from the ectoderm to the endoderm (Montgomery and Kremer, 1995; Benayahu, 1997; Benayahu and Schleyer, 1998). In the jellyfish L. unguiculata, ectodermal cells infected by zooxanthellae can migrate to the endoderm of planulae (Montgomery and Kremer, 1995). In the soft coral A. glauca, zooxanthellae within a vacuole in the detached ectodermal cytoplasm pass through temporarily opened gaps in the mesoglea toward the endoderm (Benayahu, 1997; Benayahu and Schleyer, 1998). In M. digitata, we could not observe zooxanthellae in the process of entering the endoderm through a gap in the mesoglea. Some zooxanthellae in the endoderm, however, were located just below the mesogleal layer, which was more or less obscured, and appeared to have been translocated from the ectoderm (Fig. 3C). Most zooxanthellae in the ectoderm were contained in cellular processes and were located close to the mesogleal layer. It is not clear whether a whole zooxanthella-containing cell migrates to the endoderm, or whether only the zooxanthellae, together with small amounts of cytoplasm, move to the endoderm, as suggested in some soft corals (Benayahu, 1997; Benayahu and Schleyer, 1998).
We observed no degraded zooxanthella particles in the surface layer of embryos or early planulae of M. digitata. In contrast, Davy and Turner (2003) only occasionally found zooxanthellae in the ectoderm of the late gastrula or early planula and suggested that these stray zooxanthellae may degrade in the host or be expelled. It has been proposed that degraded zooxanthella particles are produced by partial digestion of zooxanthellae by the host (Titlyanov et al., 1996, 1998; Mise and Hidaka, 2005). Few zooxanthellae were observed at the bottom of beakers incubating embryos of M. digitata. The absence or scarcity of degraded and expelled zooxanthellae may indicate that those in the ectodermal layer of planulae are transferred to the endoderm during the maturation of planulae.
Vertical transmission via zooxanthellate eggs
The three genera Porites, Montipora, and Pocillopora, which contain species that spawn zooxanthellate eggs, belong to different families and occur in different lineages in phylogenetic trees of scleractinian corals (Veron, 1995; Romano and Palumbi, 1996, 1997; Romano and Cairns, 2000; Fukami et al., 2000; Kerr, 2005). Furthermore, the same family can contain some species that release zooxanthella-containing eggs and others that produce zooxanthella-free eggs. For example, most Montipora species produce eggs containing zooxanthellae, whereas Acropora species, which belong to the same family, release eggs without zooxanthellae (Fadlallah, 1983). This together with the present finding that the timing of zooxanthella localization and types of gastrulation were different among the three genera (Pocillopora, Porites, and Montipora) suggests that vertical transmission of symbionts via zooxanthellate eggs evolved independently in the three genera studied.
We thank the late T. Yamasu, Professor Emeritus, for technical advice, and the staff of Sesoko Station, Tropical Biosphere Research Center, University of the Ryukyus, where part of this study was conducted. We appreciate the advice and help of Dr. E. Hirose. This study was partly supported by the 21st Century COE Program of the University of the Ryukyus and by a Sasagawa Scientific Research Grant from the Japan Science Society.