Observations of the tissue-skeleton interface in the scleractinian coral Stylophora pistillata
Recent micro-analytical studies of coral skeletons have led to the discovery that the effects of biology on the skeletal chemical and isotopic composition are not uniform over the skeleton. The aim of the present work was to provide histological observations of the coral tissue at the interface with the skeleton, using Stylophora pistillata as a model, and to discuss these observations in the context of skeletal ultra-structural organization and composition. Several important observations are reported: (1) At all scales of observation, there was a precise morphological correspondence between the tissues and the skeleton. The morphological features of the calicoblastic ectoderm correspond exactly to the shape of individual crystal fiber bundles in the underlying skeleton, indicating that the calicoblastic cell layer is in direct physical contact with the skeletal surface. This is consistent with the previously observed chemical and isotopic composition of the ultra-structural components in the skeleton. (2) The distribution and density of desmocyte cells, which anchor the calicoblastic ectoderm to the skeletal surface, vary spatially and temporally during skeletal growth. (3) The tissue above the coenosteal spines lack endoderm and consists only of ectodermal cell-layers separated by mesoglea. These findings have important implications for models of vital effects in coral skeletal chemistry and isotope composition.
Minerals are formed by a large variety of micro- and macro-organisms, ranging from bacteria to vertebrates (e.g., Lowenstam and Weiner 1989). These biologically produced solids are referred to as “biominerals” and the processes by which they form are referred to as “biomineralization”. A fundamental scientific goal is to uncover common mechanisms of biomineralization adopted by different types of organisms and to understand how these mechanisms have evolved since biominerals first appeared in the geological record about 3.5 billion years ago (e.g., Lowenstam and Weiner 1989). To achieve this broad scientific goal, biominerals from a wide range of organisms must be investigated and unifying properties identified. At the same time, the unique characteristics of each organism must be taken into account, so that important differences in biomineralization strategy can be identified and linked with physiological and genetic differences. From this perspective, reef-building (e.g., hermatypic) corals represent a fascinating and highly suitable case study. Coral reefs are the most prolific biomineralizing ecosystems in nature with a calcification rate of about 2–6 kg CaCO3 m−2 y−1 (Barnes and Devereux 1984). Corals calcify faster than most other animals and outpace inorganic calcification rates on the reef by a factor of more than 100 (Cohen and McConnaughey 2003). In doing so, they control the tempo of the biomineralization in reef communities (Gattuso et al. 1999).
A special feature of hermatypic corals is that they host unicellular dinoflagellate symbionts, commonly called zooxanthellae, in their tissues. Zooxanthellae are believed to influence calcification by a process referred to as “light-enhanced” (or alternatively as “dark-repressed”) calcification (for review see Gattuso et al. 1999). Thus, hermatypic corals constitute a good case for studying biomineralization, symbiosis and the interactions between both processes. Even though our knowledge of coral biomineralization has increased in recent years due to combined efforts of biologists, geochemists, geologists, paleoclimatologists and medical surgeons, the intimate mechanisms of coral skeletogenesis are still poorly understood (for recent reviews see Gattuso et al. 1999; Cohen and McConnaughey 2003; Allemand et al. 2004; Cuif and Dauphin 2005). While information on coral biology is accumulating rapidly, this has not been accompanied by corresponding research on the functional anatomy of calcification (Fautin and Mariscal 1991). Part of this gap may be attributed to technical difficulties in studying the intimate association between the soft parts and the hard skeleton of corals.
Understanding the mechanisms of biomineralization in hermatypic scleractinians requires knowledge of both the organism and the calcareous skeleton. Since the 1970s, researchers have understood the importance of comparative studies on the morphology of the skeleton and the histology of the skeletogenic tissues (Barnes 1970; Johnston 1980; Brown et al. 1983; Gladfelter 1983; Isa 1986; Le Tissier 1987; Fautin and Mariscal 1991). Recent studies, however, have focused either on the properties of the calcifying tissue or on the properties of the skeleton, and only three studies have linked these two aspects at the microscopic level of the calcifying interface (Goldberg 2001a; Clode and Marshall 2002; Raz-Bahat et al. 2006).
Micro-scale studies of coral skeletons have led to the discovery that the effects of biology on the skeletal, chemical and isotopic compositions are not uniform over the skeleton. Early mineralizing zones, which are often referred to as centers of calcification (COC), show very different stable isotope (e.g., C, O and B) or trace element (e.g., B, Mg, Sr, Ba, N, S) composition compared with other parts of the skeleton, and large chemical and isotopic variations have been documented in the fibrous aragonite part of the skeleton (Allison 1996; Cuif and Dauphin 1998, 2005; Cohen et al. 2001; Adkins et al. 2003; Clode and Marshall 2003; Cuif et al. 2003, 2004; Rollion-Bard et al. 2003a, b; Allison et al. 2005; Meibom et al. 2003, 2004, 2006, 2007; Raz-Bahat et al. 2006; Sinclair et al. 2006). These observations are difficult to reconcile with the traditional view of skeletogenesis, whereby the skeleton is precipitated from a liquid reservoir, assumed to exist between the overlying tissue, the calicoblastic ectoderm, and the growing surface of the skeleton. These micro-analytical observations suggest a close relationship between the structure and the organization of the tissues, and the different ultra-structural components of the skeleton.
The Indo-Pacific scleractinian Stylophora pistillata has become a model for laboratory studies (Tambutté et al. 1995b) and numerous physiological data related to calcification are now available for this species (Rinkevich and Loya 1984; Gattuso 1987; Tambutté et al. 1995a, b, 1996; Yamashiro 1995; Tambutté 1996; Muscatine et al. 1997; Allemand et al. 1998; Gattuso et al. 1999; Zoccola et al. 1999; Puverel et al. 2004, 2005). However, the species is poorly described at the microscopic level. The aim of the present work was to present the first systematic study of the coral tissue and the associated skeleton in S. pistillata, in order to better understand the organization and the structure of the skeletal forming tissue associated with different skeletal ultra-structural components.
Materials and methods
Coral colonies of S. pistillata (Cœlenterata: Anthozoa: Scleractinia) were maintained at the Centre Scientifique de Monaco under the following conditions: semi-open circuit, Mediterranean seawater heated to 26 ± 0.2°C (mean ± range), pH 8.1–8.2, illuminated with 96 W full spectrum fluorescent bulbs (Custom Sealife) at a constant irradiance of 175 μmol photons m−2 s−1 on a 12 h:12 h day:night light-cycle, and fed with Artemia nauplii twice a week. Some corals were cut into small fragments to obtain micro-colonies, which were grown on slides for microscopy purposes (Tambutté 1996; Muscatine et al. 1997).
Sample fixation and decalcification
Samples were fixed overnight at 4°C with 4% glutaraldehyde in 0.085 M Sorensen phosphate buffer at pH 7.8 with 0.5 M sucrose. Decalcification was achieved by transferring the samples to a mixture of 0.085 M Sorensen phosphate buffer, 0.5 M sucrose containing 2% glutaraldehyde and 0.5 M ethylenediaminetetraacetic acid (EDTA) at pH 7.8 and 4°C. This solution was renewed twice daily until decalcification was complete. Decalcified samples were rinsed in Sorensen buffer, then post-fixed for 1 h at ambient temperature with 1% osmium tetroxide in Sorensen phosphate buffer. Samples were dehydrated by transfer through a graded series of ethanol ending with a concentration of 100%.
Field emission scanning electron microscopy (FESEM)
Samples were processed as described above. After dehydration, they were incubated for 5 min in Hexamethyldisilazane (HMDS) that was subsequently evaporated under a fume hood. For skeletal observations the soft tissue was removed from the skeleton with NaOH 1 N and the skeleton was then rinsed with distilled water.
All samples were coated with gold-palladium and observed at 3 kV in a JEOL JSM 6700 F field emission scanning electron microscope at the Centre Commun de Microscopie Appliquée at the University of Nice-Sophia Antipolis.
In the case of frozen hydrated preparations, some samples were fixed with 4% glutaraldehyde in 0.085 M Sorensen phosphate buffer at pH 7.8 with 0.5 M sucrose, then blotted on wet filter paper and rapidly frozen by plunging into a pasty nitrogen (−210°C). Freeze fracturing was done in a cryo-preparation system (Alto 2500, Galtan), followed by freeze-drying and coating with gold-palladium. Samples were viewed at 7.0 kV in JEOL JSM 6700 F field scanning electron microscope at the Centre Commun de Microscopie Appliquée at the University of Nice-Sophia Antipolis.
Transmission electron microscopy (TEM)
Samples were fixed and decalcified as described above and after dehydration, they were embedded in Epon resin. For observation in bright light, 1 μm-thick sections were cut with a diamond-knife and stained briefly in Methylene blue. For transmission electron microscopy, ultrathin sections were cut with a diamond knife, transferred to copper grids coated with Formvar, stained with uranyl acetate and lead citrate, rinsed in distilled water, and observed with a CM12 Phillips TEM at 80 kV at the Centre Commun de Microscopie Appliquée at the University of Nice-Sophia Antipolis.
Light and fluorescence microscopy
Whole samples were observed in bright light or fluorescent light with a Leica Z16APO (Leica™ microsystems). Fluorescent samples were excited with blue light and emission was filtered at 530 nm. The Leica Z16APO was connected to an Imaging Qicam camera. Decalcified samples were fixed for 2 h in 3% paraformaldehyde in S22 buffer (450 mM NaCl, 10 mM KCl, 58 mM MgCl2, 10 mM CaCl2, 100 mM Hepes, pH 8) at 4°C, overnight. They were then transferred in calcium-free S22 buffer containing 3% paraformaldehyde and 0.5 M EDTA at pH 8.2 and 4°C. The solution was renewed twice daily until decalcification was complete. The samples were rinsed and then observed in S22 buffer. Some samples were also processed for sectioning. After fixation and decalcification, these samples were dehydrated in an ethanol series, processed in toluene and embedded in Paraplast. Cross-sections (7 μm thick) were cut and mounted on silane-coated glass slides. Sections were stained with Haemalum-Eosine and and acetified Aniline blue or methylene blue-Azur blue.
Organization of the tissues
Relationship between tissues and skeleton
Attachment of tissue to skeleton
Calicoblastic ectoderm (calicodermis)
Vesicles within the calicoblastic cells or exocytotic vesicles on the apical membrane in S. pistillata were never observed. However, extra-cellular spheres of unknown origin appear sometimes in the intercellular spaces (Fig. 4d). These structures look like vesicles, but they are located in the baso-lateral region rather than in the apico-lateral region of the epithelium. These structures may equate to the “hollow or compressed circular vesicles” observed by Johnston (1980).
Calicoblastic ectoderm as a cast for fiber bundles
Special features of the tissues and skeleton
Thus, depending on the zone of the coenosarc, the histology is very different. Typically, as described in the beginning of this section, the tissue has four epithelial layers with numerous zooxanthellae in the oral endoderm and few in the aboral endoderm. However, there are only two epithelial layers, the oral and aboral ectoderms, in the area at the tip of coenosteal spines and no zooxanthellae (Fig. 7c, d). In these zones, the oral ectoderm is only separated from the calicoblastic ectoderm by the mesoglea and the whole tissue appears very thin.
Histology of the calicoblastic ectoderm
Characteristics of the calicoblastic cells
A coral colony can be thought of as a one big calcifying organism. However, as illustrated by the results of the present study, the tissues cannot be considered as one, uniform structure from which the skeleton is precipitated. The various cell-layers are structured differently over different skeletal zones or components. The calicoblastic ectoderm is always the thinnest cell layer. The shape of calicoblastic cells varies between long/thin/flat to thick and cup-like, and this shape correlates with the calcification activity of the cells. Flat calicoblastic cells appear to be associated with low calcification activity, and cup-like cells with high calcification activity (Duerden 1902). Junctions between calicoblastic cells are always observed on the apical side of the cells (Fig. 4). Similar junctions were initially described in Pocillopora damicornis as zonular and presumed to completely encircle each ectodermal cell, forming an occlusive band (Johnston 1980). Later, they were described at the apical region of calicoblastic cells of Galaxea fascicularis and labelled “septate junctions” (Clode and Marshall 2002) because of the similarity with other anthozoan junctions (Green and Flower 1980; Holley 1985). The junctions on the apical side of the calicoblastic ectoderm (calicodermis) of S. pistillata also have the characteristics of invertebrate septate junctions. The presence of junctions between cells strongly suggests that cells are asymmetric, or polarized, with apical and basolateral cell surfaces showing different features. It is known from similarly polarized cells in vertebrates that the membrane at each pole has different permeability properties and different patterns in the distribution of membrane proteins such as carriers and ion channels (Nelson et al. 2000; Anderson et al. 2004; Aijaz et al. 2006). In invertebrates, septate junctions are considered to form “tight” epithelia and thus constitute an impermeable barrier to the diffusion of ions and molecules (Clode and Marshall 2002; Faivre-Sarrailh et al. 2004). In S. pistillata and in G. fascicularis, observations suggest that the major transepithelial step for calcium transport occurs at the aboral level and that the calicoblastic ectoderm is responsible for active transport (Wright and Marshall 1991; Tambutté et al. 1995a, 1996). Such a transcellular pathway for calcium at the calicoblastic level is consistent with the presence of septate junctions between the cells.
Vesicles of the calicoblastic ectoderm
Intracellular vesicles in calicoblastic cells have been described in detail (Johnston 1980; Le Tissier 1990, 1991; Clode and Marshall 2002). In the present study, such vesicles were not observed in S. pistillata. Based on these observations, it is difficult to establish whether intracellular vesicles are indeed present, but cannot always be observed with the techniques used (e.g., TEM and FESEM), or whether the presence of intracellular vesicles is species dependent. Consistent with the findings of the present study, Clode and Marshall (2002) studied G. fascicularis and failed to find evidence for exocytotic vesicles, suggesting that indeed there is no vesicular activity associated with the apical membranes of calicoblastic ectodermal cells. Still, it cannot be completely excluded that exocytotic vesicles exist. In S. pistillata, calicoblastic cells have been shown to be the site of organic matrix synthesis (Puverel et al. 2004) and the inhibition of organic matrix synthesis leads to the inhibition of calcification (Allemand et al. 1998). Because there is no evidence of the presence/absence of exocytotic vesicles on the apical side of calicoblastic cells, the pathway of the organic matrix from the calicoblastic cells to the skeleton is not understood. However, experiments performed with agents inhibiting cytoskeleton activity prevent calcification, suggesting that exocytosis is involved in this process. Observations of organic matrix between the calicoblastic ectoderm and the skeleton have been performed using TEM on P. damicornis (Johnston 1980; Le Tissier 1987) and on frozen-hydrated preparations of G. fascicularis (Clode and Marshall 2002) and Mycetophyllia reesi tissue (Goldberg 2001a). The latter two studies have demonstrated that thin organic matrix “fibrils” are present in the sub-calicoblastic space. In S. pistillata, even on frozen-hydrated samples, neither sub-calicoblastic space nor organic matrix fibrils extending into the skeleton were observed.
In the present study, extracellular material was observed between calicoblastic cells (at the base), appearing as spherical structures that resemble the “vesicles” reported by Johnston (1980). Even though this study did not allow a more precise determination of these structures, it is possible that they could be exocytotic vesicles responsible for the secretion of components of the mesoglea, because it has been shown in the jellyfish Aurelia aurita that the ectoderm is responsible for part of the synthesis of elastic fibers of the mesoglea (Shaposhnikova et al. 2005).
Desmocyte cells of the calicoblastic ectoderm
The calicoblastic ectoderm is firmly attached to the skeleton by anchoring cells, the desmocytes, the morphology of which, and also the mechanism of anchorage to the skeleton have been described in detail (Muscatine et al. 1997). In the present study, the distribution of desmocyte cells in S. pistillata is variable depending on the zones and the stage of tissue growth. Desmocyte cells are ubiquitous at the interface with the skeleton. There is a transition in the spatial distribution of desmocyte cells from immature to mature polyps. In mature polyps, the desmocyte cells are more numerous and well organized in linear patterns on each polyp associated with the septa as a ‘finger glove’. In earlier studies, Muscatine et al. (1997) found high densities of desmocytes in the calicoblastic ectoderm of S. pistillata, whereas Clode and Marshall (2002) rarely observed desmocytes in frozen-hydrated polyps of G. fascicularis and therefore suggested that the density of desmocytes depends on the region of the skeleton: a high density of desmocytes in low-calcifying regions and a low density in high-calcifying regions. Contrary to these observations of Clode and Marshall (2002), in the present study a high density of desmocytes was seen in polyps of S. pistillata. The distribution of desmocytes may, therefore, not only vary among species, but also within the growth zones of an individual colony. It seems that desmocytes are most abundant once skeletal morphological development is complete.
The cell-skeleton interface
The present study shows that there is a strong morphological correspondence between the soft tissues and the calcareous skeleton from the macroscopic to the microscopic level of observation. At a macroscopic level, the skeleton is the perfect fingerprint of the soft tissues. At a microscopic level, calicoblastic cells correspond perfectly to the shape of fiber bundles of the skeleton. It seems highly unlikely that such a perfect, sub-cellular morphological match could be a simple artifact of sample preparation, i.e., by ‘collapse’ of a substantial, liquid-filled space between the tissue layer and the skeletal surface when the animal was killed. Similar observations have been made for other coral species, based on different preparation techniques, which strongly indicates that the observation made in this and previous studies are not artifacts of sample preparation. Furthermore, and consistent with the present observation for S. pistallata, a direct correspondence between calicoblastic ectoderm and crystal fiber bundles was also observed for P. damicornis (Johnston 1980; Brown et al. 1983) and Acropora cervicornis (Gladfelter 1983). Goldberg (2001a) noted that the “compartment-like structure” of the calicoblastic ectoderm corresponds roughly to bundles of crystals in Mycetophyllia reesi. In P. damicornis, Brown et al. (1983) observed that skeletal spines display either “fasciculate” or “smooth” surfaces, and that these features corresponded to different sections of the colony, fasciculate at the apex and smooth at the base, and were dependent on growth rates. In the present study, fasciculate and smooth surfaces described as ‘cup-like’ or ‘flat’ surfaces, respectively, effectively correspond to two different stages of growth, e.g., of the spines. When the spine is actively growing, its surface is comprised of curved crystal fiber bundles oriented perpendicular to the surface of the skeleton, whereas when the spine growth is nearly complete, the crystal fiber bundles become oriented parallel to the surface, i.e., the surface appears flat. Moreover, the shape of a crystal fiber bundle is the exact same shape as the calicoblastic ectoderm (Fig. 6). Cup-like cells are directly associated with curved crystal fiber bundles, whereas flat cells are associated with flat crystal fiber bundles. Combining the observations of Fig. 4b and Fig. 6c, d, it appears that each crystal fiber bundle is associated with more than one calicoblastic cell, suggesting the need for cell co-ordination during skeletogenesis, probably facilitated by gap junctions. With each fiber bundle being isolated from its neighbor during rapid growth, cell co-ordination is required throughout the entire calicoblastic ectoderm. At the scale of individual crystals (<<1 μm), further investigations are necessary to determine how the surface of calicoblastic cells complements the surface of each crystal. It is possible that, at this nano-metric length scale, there could be a small space between the skeletal surface and the calicoblastic cell-layer in which crystal-growth can occur (in a manner similar to that suggested by, e.g., Barnes 1970). However, at the scale of magnification (in freeze-fractured FESEM samples) used in this study, no such space was observed, and it may therefore be concluded that it must be so narrow as to serve no practical purpose.
Models in which calcification primarily occurs within pockets created where calicoblastic ectoderm is lifted away from the skeletal surface imply that the coral tissue plays a less direct role in crystal growth (Barnes 1970, 1972). In contrast, the observation of a physically tight interface between tissue and skeleton in S. pistillata clearly favors a model in which the calicoblastic ectoderm controls the calcification process and is consistent with recent micro-analytical observations of dramatic chemical and isotopic variations in the skeleton at ultra-structural length scales, as discussed above. In any case, the calcifying activity and shape of the calicoblastic cells vary depending upon the skeletal zone. This implies a direct impact of the calicoblastic ectoderm on the calcification process.
The role of zooxanthellae in coral biomineralization
Since the early observations of an intimate relationship between light level and the calcification rate in corals (Yonge 1931; Kawaguti 1937), many researchers have investigated the role of zooxanthellae in coral biomineralization, usually referred to as “light-enhanced calcification” (Goreau and Goreau 1959). Numerous studies have shown that zooxanthellate corals calcify at higher rates than azooxanthellate corals (for review see Gattuso et al. 1999). Several hypotheses have been proposed to explain this (for review see Allemand et al. 2004): (1) symbionts synthesize molecules that are essential to the calcification process (Muscatine and Cernichiari 1969), (2) symbionts affect the dissolved inorganic carbon (DIC) equilibrium within coral tissues by taking up CO2 for photosynthesis (Goreau and Goreau 1959; McConnaughey and Whelan 1997) or by secreting hydroxyl ions that are the product of Carbon Concentrating Mechanisms (Furla et al. 2000), (3) symbionts produce energy and O2 that can accelerate the calcification process (Chalker and Taylor 1975; Rinkevich and Loya 1984), and (4) symbionts play a role in the removal of substances that would otherwise inhibit calcification (Simkiss 1964).
Each hypothesis has its merits and none can be completely excluded, but each model needs to be evaluated against the most recent histological and physiological observations. Perhaps the most important observation in this regard is the general observation that the tissues which calcify at the highest rates, or which initiate calcification, do not possess zooxanthellae. For example, zooxanthellae are rare in zones with coenosteal spines, especially at the tip where the oral ectoderm is separated from the calicoblastic ectoderm only by a thin layer of mesoglea (present study for S. pistillata and Brown et al. 1983 for P. damicornis). Similarly, the calcification rate is higher at the tip of colony branches where there are fewer symbionts than at the base (Goreau 1959; Pearse and Muscatine 1971; Barnes and Crossland 1978). These observations hold whether the corals are grown in nature, on cover slips in the laboratory (Raz-Bahat et al. 2006) or on slides (personal observation), namely, at the growing front of the tissues, in the zones where the mineralization starts, there are no zooxanthellae. Similarly, in the cell tissue over the exert septa, there are essentially no zooxanthellae, and it has been demonstrated by autoradiography that calcium incorporation is higher here than in axial polyps (Marshall and Wright 1998). So what role, if any, do zooxanthellae play for the coral skeleton biomineralization process?
Recently, Raz-Bahat et al. (2006) questioned the concept of “light-enhanced calcification” and refuted the idea of diurnal changes in calcification patterns and rates. Instead they suggested that several competing mechanisms might control the calcification process. A similar suggestion was put forward by Goreau nearly 50 years ago (Goreau 1959) who speculated that “Although the zooxanthellae seem to play an important role in determining calcification rates in reef-building corals, certain, as yet unknown, physiological factors operate to control the basic mineralization process in a manner which bears no obvious relationships to the number of algae present in a given species”. It can be proposed, in light of the existing literature and the histological data of the present work, that “light-enhanced calcification” should be considered differently depending on the scale of observation: macro-scale (at the level of the organism) or microscale (at the level of the cells). At the macro-scale level, zooxanthellae might globally enhance calcification during the day by providing, in high-quantity, some precursors of organic matrix (Muscatine and Cernichiari 1969; Cuif et al. 1999; Muscatine et al. 2005). Other important parameters, such as ATP supply and cœlenteric pH regulation also vary on a diurnal cycle (Gattuso et al. 1999; Allemand et al. 2004). At the microscale level, the organization of the cell tissue and the presence or absence of zooxanthellae, must be taken into account and related to the organization and composition of the skeletal ultra-structure components. It is observed that over the tips of the coenosteal spines, which consist of center of calcification, the tissues systematically lack endoderms and consist only of two ectodermal layers separated by the mesoglea (and are thus much thinner than elsewhere, Fig. 7d). In these zones, calicoblastic cells are also far from zones containing endodermal cells with zooxanthellae, and the formation of center of calcification, which is a continuous process, seems to be taking place essentially in the absence of zooxanthellae.
This paper is dedicated to the memory of our dear friend and colleague Len Muscatine. Thanks are due to Dominique Desgré for coral maintenance. Thanks are also due to Jean-Pierre Laugier, Sophie Pagnotta and Pierre Gounon from the Centre Commun de Microscopie Appliquée at the University of Nice-Sophia Antipolis. This study was conducted as part of the Centre Scientifique de Monaco research program, supported by the Government of the Principality of Monaco, by the Agence Nationale de la Recherche and by IFREMER. This paper has greatly benefited from the highly constructive comments by Dr. Lasker and two anonymous reviewers.