Gill-symbiosis in mytilidae associated with wood fall environments
- First Online:
- Cite this article as:
- Gros, O., Guibert, J. & Gaill, F. Zoomorphology (2007) 126: 163. doi:10.1007/s00435-007-0035-3
- 93 Views
Bivalves belonging to the genera Idas and Adipicola were collected from wood fall environments in the west Pacific (Vanuatu islands) between 300 and 890 m depths in 2004. Bacterial symbionts were checked by three complementary techniques: histological and DAPI staining, in situ hybridization (FISH), and TEM. No bacteria were detected inside the gills of the two species, rejecting the endosymbiosis hypothesis. However, results from our study demonstrated the existence of ectosymbionts colonizing microvilli differentiated at the apical surface of the cells constituting the lateral zone of gill filaments. These ectosymbionts are γ-Proteobacteria due to their strong hybridization with the specific probe GAM42; in contrast no hybridization was obtained from either gills or other host tissues by using the oligonucleotide probes specific to α- β- and δ-Proteobacteria. Based on TEM observations, these Gram-negative bacterial symbionts are not methanotrophic due to the lack of concentric stacking of intracellular membranes in their cytoplasm. Such ectosymbionts may represent thioautotrophic bacteria as already described in various Mytilidae from hydrothermal vents and cold seeps. Unfortunately, no phylogenetic analysis could be done in this study to compare their DNA sequence to that of other marine invertebrate symbionts described to date.
Interactions between prokaryotic symbionts and marine invertebrate hosts have been described since the early 1980s (Felbeck et al. 1981; Cavanaugh et al. 1981). They were then reported to occur in various environments from shallow water to hydrothermal vents and from various taxa including annelids, bivalves, etc. Gill-endosymbiosis in Mytilidae was first described in Bathymodiolus thermophilus [Kenk and Wilson, 1985] in hydrothermal vents (Fiala-Médioni et al. 1986). Then, small symbiotic mussels belonging to the genus Idasolas were reported to be associated with whale bones (Deming et al. 1997). Thus, symbiont-containing mytilids are commonly found at hydrothermal vents and cold seep environments throughout the world (for review, see Duperron et al. 2005). Distel et al. (2000) have suggested, based on phylogenetic analyses on various Mytilidae, that decomposing organic substrates such as wood and bone may serve as steps for the introduction of mytilids into vent and seep environments. Moreover, they suppose that Mytilidae found in such environments may harbor chemoautotrophic bacterial symbionts.
The aim of this study was to look for bacterial symbionts in mussel specimens collected from wood fall environments, to test the hypothesis of chemoautotrophy initially proposed by Distel et al. (2000). Histological analyses were performed to obtain a general view of the mytilid tissues, and to allow an overall in situ detection of bacteria in whole animal sections by DAPI staining. Complementary analyses focused on the gills, organs known to harbor symbionts in bivalves (Reid 1990), by using fluorescent in situ hybridization and TEM analyses.
Materials and methods
Sampling of the animals
Bivalves were collected during the BOA0 cruise in November 2004, using a beam trawl between depths of 300–890 m from wood fall environments in the Vanuatu area. Samples were collected between 15°41.78–15°44.05S latitude and 167°03.83–167°01.83W longitude attached to the exterior of small pieces of submerged wood. Samples were processed on the ship within 1 h after collection. Based on shell characters only, the three individuals examined in this study belong to two genera, Idas and Adipicola. As these two species are undescribed species (Von Cosel, personal communication), we gave them the nomen nudum name of Idas woodia and Adipicola sunkenia. The single individual of A. sunkenia was fixed specifically for histology, whereas complementary analyses were performed on I. woodia.
An overall view and histological information were obtained from paraffin sections. After breaking the shell, one single whole individual of Idas and Adipicola were fixed in Bouin’s fluid (Gabe 1968) for 24 h at room temperature before being embedded in paraplast. Sections (7 μm thick) were stained by Goldner’s trichrome for morphological information, according to Gabe (1968).
Fluorescent in situ hybridization experiments
One single whole individual of I. woodia was fixed for 1–3 h at 4°C in 4% paraformaldehyde in 1× PBS buffer. The specimen was then washed three times for 10 min each at 4°C in 1× PBS, then stored in 0.5× PBS/50% ethanol at 4°C until it was embedded in paraplast. Four micrometer-thick sections were placed on precoated slides from Sigma before hybridization. Five oligonucleotide probes were used: Eub338 (5′-GCTGCCTCCCGTAGGAGT-3′), targeting most members of the eubacteria (Amann et al. 1990a, b), and more specific probes such as ALF968 for α-proteobacteria (5′-GGTAAGGTTCTGCGCGTT-3′), GAM42 for γ-bacteria (5′-GCCTTCCCACATCGTTT-3′), BET42a for β-proteobacteria (5′-GCCTTCCCACTTCGTTT-3′), and NON338 (5′-ACTCCTACGGGAGGCAGC-3′) as a negative control. Hybridization experiments were similar to those previously described by Dubilier et al. (1995, 1999). To visualize total symbionts as a control, gill sections, after FISH hybridization, were counter-stained with DAPI (1 μg/ml).
Electron microscopy preparations
One single whole individual of I. woodia was prefixed on ship for 1 h at 4°C in 2.5% glutaraldehyde in 0.1 M pH 7.2 caccodylate buffer, adjusted to 900 mOsM with NaCl and CaCl2 in order to improve membrane preservation. Following a brief rinse, it was stocked in the same buffer at 4°C until it was brought to the laboratory where gills were dissected. One was fixed for 45 min at room temperature in 1% osmium tetroxide in the same buffer, then rinsed in distilled water and post-fixed with 2% aqueous uranyl acetate for one more hour before embedding and observation as described previously (Gros et al. 2003). The other gill was dehydrated in an ascending series of acetone dilutions and critical point dried using CO2 as transitional fluid for the SEM procedure. Samples were then sputter coated with gold before observation in a Hitachi S-2500 SEM at 20 kV.
In situ hybridization
Gill filament ultrastructure
The lateral zone forms the main part of the filament and accounts for the thickness of the gill. This lateral zone appears quite simple because of it is composed of only three cell types (Figs. 8, 9). The most prevalent one consists of cells containing bacteria at their apical surface. These bacteria are interspersed between microvilli differentiated by the host cells (Figs. 8, 9). Such a proximity, associated with the bacterial density, lead us to define them as bacteriocytes, as are named cells harboring endocellular bacteria in other bivalve families (Fisher 1990). In the two specimens analyzed here, the bacteriocytes, which have a basally-located nucleus are much wider than high, and possess long microvilli, among which bacteria are located (Figs. 8, 9). The bacteriocytes organelles consist of few mitochondria and numerous lysosome-like structures characterized by whorls of membranes as large as the nucleus (Figs. 9, 12). These large secondary lysosomes contain putative remains of bacteria (Figs. 9, 12) suggesting a possible digestion of bacterial ectosymbionts by the invertebrate host. Host cells, using plasmic membrane expansions, can enclose bacteria in phagocytic vacuoles at their apical pole (Fig. 10). Thus, the aspect of the lysosome-like structures depends on their maturity from early endosomes just after the phagocytosis of extracellular bacteria (Fig. 11) to classical secondary lysosomes (Fig. 12) and even residual bodies (Fig. 13).
There are no intercalary cells interspersed between bacteriocytes, as usually described in bivalves known to harbor bacteria (Figs. 7, 8, 9). Two secretory cell types were observed through the lateral zone of the gill filament of the Idas sample examined in this study. The first one has a cytoplasm filled with large, membrane bound, osmiophilic inclusions which are PAS positive on histological sections (not shown). Only a few of these granule cells were observed in the lateral zone of each gill filament even if they may be present throughout the length of the gills. The second secretory cell type consists of mucocytes, which are scarce in the lateral zone of the gill filaments. Most of the mucocytes observed are those which are located at the frontal crest of the filaments (not shown) inside the ciliated zone. Both cell types secrete acid mucopolysaccharides (stained in blue with alcian blue) and the mucous contents seem to be present along the surface of each gill filament, indicating that particle capture may occur in this species. Another possibility is that such mucous presence could help to protect gill cells against toxins in the environment.
Wood falls represent a massive input of food into a marine environment at depths where food is typically scarce due to the absence of sunlight. Wood debris can provide a huge amount of organic matter, drifting down from surface waters before being transported by submarine currents through various canyons at greater depths. Such organic matter could be used by microbial mats or by eukaryotic macrofauna directly (i.e. xylophagous organisms) or indirectly (heterotrophic organisms eating the microbial mats degrading the wood). Moreover, Leschine (1995) has shown that sulfide represents the principal product of the cellulose degradation in marine environment. Thus, such a continuous production of hydrogen sulfide could support chemosynthetic communities as previously described with bacterial decomposition of lipids from whale bones in the deep sea (Deming et al. 1997).
Gill structure and bacterial location
The bacteria described in the specimens of the two genera presented here (Adipicola and Idas) are located extracellularly while remaining in close contact with the apical pole of gill cells along the entire lateral zone of the gill filaments. By comparison with previous descriptions of gill endosymbiosis in bivalves or of ectosymbiosis as observed in individuals belonging to the families Thyasiridae (Dando and Southward 1986; Dufour 2005; Dias Passos et al. 2007) and Bathymiodiolinae (McKiness et al. 2005), such cells establishing preferential interactions with extracellular bacteria will be called bacteriocytes. The lateral zone consists of a thin, simple epithelium, suggesting that exchanges can occur between the bacteriocytes and the environment. Moreover, bacteriocytes develop microvilli all along their apical pole probably to (1) increase the surface of the cellular membrane, which is an important site of uptake of dissolved organic compounds present in the water and (2) offer the largest surface to harbor ectosymbiotic bacteria. Thus, bacteria live in close contact with host cells and with the flow of water circulating through the lateral zone of gill filaments. Because of their particular location, such bacteria, could protect host cells by taking up reduced sulfur compounds from the environment. Moreover, these ectosymbionts are probably regularly endocytosed by bacteriocytes and digested inside lysosomal structures providing the host cells with an input of organic compounds as previously described in various marine symbiosis models (Fiala-Médioni et al. 1994; Liberge et al. 2001). TEM analysis has revealed that in the Idas woodia individual, bacteriocytes contain numerous lysosome-like structures as large as nuclei. These structures appear to be filled with membrane whorls. Such structures have been previously reported in symbiotic bivalves (for review see Frenkiel et al. 1996; Dufour 2005) with various interpretations. They usually are considered to be lysosomes involved in the digestion of a limited portion of bacterial endosymbionts (Giere 1985; Distel and Felbeck 1987; Frenkiel et al. 1996). In our case, they may be involved in the digestion of ectosymbionts after their phagocytosis by host cells. This phenomenon may also be involved in a larger process: the control of the proliferation of the bacterial symbiont community on the cell surface. However, such structure with membrane whorls have also been described as sulfide-oxidizing bodies (SOBs) in marine invertebrates from sulfide-rich environments (Powel and Somero, 1985; Liberge et al. 2001). Thus, further cytochemical analyses of gill tissues will be necessary to conclude as well as stable isotope analyses and/or enzyme assays to detect of sulfur-based chemoautotrophic enzymes and to the ribulose-1,5-biphosphate carboxylase/oxygenase (Rubisco) to clarify the mode of nutrition of the bacteria and bivalves.
Are the bacterial ectosymbionts chemoautotrophic?
This is the first study to document gill-symbiosis in marine invertebrates associated with sunken wood environments. Mytilidae from cold seeps or hydrothermal vents usually harbor gill endosymbionts which are either sulfur-oxidizing symbionts (Fiala-Médioni et al. 1986; Distel and Cavanaugh 1994; Fujiwara et al. 2000) or methanotrophic bacteria (Childress et al. 1986; Cavanaugh et al. 1987; Fujiwara et al. 2000), while in some cases a dual symbiosis has been reported, with a mixed population composed of methanotrophic and thiotrophic bacteria inhabiting the same tissue (Fisher et al. 1993; Duperron et al. 2005). Recently, McKiness et al. (2005) have described extracellular bacterial symbionts in a Bathymiodolinae collected at 2,200 m depth at the Juan de Fuca hydrothermal vents. The 16S rDNA sequence obtained from the gills clustered with other mytilid chemoautotrophic symbionts previously described. Thus, species belonging to the family Mytilidae can harbor extracellular chemoautotrophic bacteria. Chemoautotrophic bacteria also occur as ectosymbionts in most of the members of the family Thyasiridae (Dufour, 2005). The distribution of the bacteria throughout the lateral zone of each gill filament and their specific location between microvilli of the host cells below the glycocalyx strongly suggest that these bacteria are not ordinary fouling microorganisms that could be found on any inert surface associated with sunken wood environment. They have established particular relationships with the bivalve, which are probably symbiotic, as described in marine invertebrates colonizing sulfide-rich habitats.
The two types of symbionts (thioautotrophic and methanotrophic) belong to the γ-proteobacteria (Cavanaugh 1994) and it is difficult to determine if the ectosymbionts found in here are chemoautotrophic based only on their ultrastructure. Ectosymbionts described in the bivalve family Thyasiridae are sulfur-oxidizing bacteria (Dufour 2005) while in shrimps the bacterial community described in the gill cavity of Rimicarisexoculata [Williams & Rona, 1986] belongs to the ε-proteobacteria (Madrid et al. 2001; Zbinden et al. 2004). In our case, we can rule out the methanotrophic metabolism due to the lack of concentric stacks of intracellular membranes in the bacterial ectosymbionts; however, further investigations will be necessary (phylogenetic or metabolic studies) to determine their metabolic pathways.
Symbiont transmission mode
The fact that these symbionts are located outside the bacteriocytes should indicate that the transmission mode of such ectosymbionts is environmental. Thus, symbionts come from an environmental stock of free-living, symbiosis-competent bacteria which colonize aposymbiotic juveniles of the new host generation as described in the family Lucinidae (Gros et al. 1996, 1998). This is supported by the recent suggestion, based on genetic data, that thioautotrophic gill endosymbionts are environmentally acquired in some Mytilidae (Won et al. 2003). New investigations are in progress in our lab to check for the presence of symbionts in members of the family Mytilidae associated with sunken wood collected at deeper sites, between 500 and 2,000 m depth. The localization of the gill symbionts in such specimens may be important in understanding the history of colonization of vent and seep environments. In fact, all Mytilidae described to date harbor endocellular bacteria which may be environmentally transmitted to the new host generation, based on genetic and cytological observations obtained from deep-sea mussels of the genus Bathymodiolus (Won et al. 2003). However, endosymbiosis represents a more integrated state between two partners than ectosymbiosis; the best integration is represented by intracellular symbionts vertically transmitted to the new host generations. In this case, the prokaryotic symbionts usually have lost a part of their genome and are dependent on the eukaryotic host cells, as previously described for eukaryotic organelles such as mitochondria (for review see Douglas 1994). If the hypothesis that decomposing wood may serve as a step for the introduction of mytilids to vents and seeps, species found at intermediate depths (500–800 m) could progressively internalize their symbionts in deeper environments as an adaptation to increasing environmental constraints. Once internalized, gill-endosymbionts would be initially environmentally transmitted before being vertically transmitted through the host gametes as in a more integrated symbiosis, such as the one described for bivalves belonging to the family Vesicomyidae (Cary and Giovannoni 1993; Hurtado et al. 2003).
In conclusion, this study represents the first report of gill-symbiosis in metazoan species associated with sunken wood environments, strengthening the hypothesis that these sites could serve as steps for the colonization of chemosymbiotic species into either vents or shallow water environments. However, new investigations are needed to improve our knowledge of the symbiotic relationships occurring in bivalve species inhabiting this specific environment, especially regarding symbiont phylogeny. We look forward to getting new specimens in order to pursue this investigation.
We gratefully thank M. Zbinden from UMR 7138 for collecting mytilids during the BOA0 cruise and R. von Cosel from the Museum National d’Histoire Naturelle for determination of the genera. We thank the captain and crew of Alis and the IRD team from Noumea, especially B. Richer de Forges. Part of this work was funded by the GDRE DIWOOD (CNRS SDV).