OUP user menu

Microbial diversity associated with the hydrothermal shrimp Rimicaris exoculata gut and occurrence of a resident microbial community

Lucile Durand, Magali Zbinden, Valérie Cueff-Gauchard, Sébastien Duperron, Erwan G. Roussel, Bruce Shillito, Marie-Anne Cambon-Bonavita
DOI: http://dx.doi.org/10.1111/j.1574-6941.2009.00806.x 291-303 First published online: 24 December 2009


Rimicaris exoculata dominates the megafauna of several Mid-Atlantic Ridge hydrothermal sites. Its gut is full of sulphides and iron-oxide particles and harbours microbial communities. Although a trophic symbiosis has been suggested, their role remains unclear. In vivo starvation experiments in pressurized vessels were performed on shrimps from Rainbow and Trans-Atlantic Geotraverse sites in order to expel the transient gut contents. Microbial communities associated with the gut of starved and reference shrimps were compared using 16S rRNA gene libraries and microscopic observations (light, transmission and scanning electron microscopy and FISH analyses). We show that the gut microbiota of shrimps from both sites included mainly Deferribacteres, Mollicutes, Epsilon- and Gammaproteobacteria. For the first time, we have observed filamentous bacteria, inserted between microvilli of gut epithelial cells. They remained after starvation periods in empty guts, suggesting the occurrence of a resident microbial community. The bacterial community composition was the same regardless of the site, except for Gammaproteobacteria retrieved only in Rainbow specimens. We observed a shift in the composition of the microbiota of long-starved specimens, from the dominance of Deferribacteres to the dominance of Gammaproteobacteria. These results reinforce the hypothesis of a symbiotic relationship between R. exoculata and its gut epibionts.

  • Deferribacteres
  • midgut epibiosis
  • Mollicutes
  • Proteobacteria
  • Rimicaris exoculata
  • starvation experiment


Gut microbial communities are rather ubiquitous both in vertebrates and in invertebrates. Symbioses between host and microorganisms range from pathogenic to mutualistic, facultative to obligate relationships. Gut microbiotas have been characterized for insects such as termites (Breznak, 1982; Chaffron & Von Mering, 2007), isopods such as Porcellio scaber (Wang et al., 2004a, b, 2007) and other invertebrates (Harris, 1993). These gut-associated microbial communities play a major role in the metabolism of the host, in particular, in the case of low-level nutrient supply. Extreme environments such as deep-sea hydrothermal vents are oligotrophic and low oxygenated ecosystems enriched in numerous toxic compounds. Life under these conditions requires physiological adaptations of the fauna to low levels of nutrients and toxic environments. In such ecosystems, life is based on chemosynthetic primary production and microbial–invertebrate associations are widespread (Dubilier et al., 2008). Endosymbioses where bacteria are enclosed within bacteriocytes are well described in mussels such as Bathymodiolus spp. (Duperron et al., 2007) or vestimentiferan worms such as Riftia pachyptila (Dubilier et al., 1995), the latest even being deprived of an open digestive tract.

Rimicaris exoculata (Williams & Rona, 1986, Crustacea, Decapoda, Alvinocarididae) dominates the megafauna at several Mid-Atlantic Ridge (MAR) hydrothermal vent sites characterized by contrasted geochemical settings. End-member fluids from ultramafic-hosted sites, such as Rainbow, are usually enriched in methane (2.5 mM), hydrogen (16 mM) and iron (24.05 mM), but relatively depleted in sulphides (1.2 mM), in contrast to basalt-hosted sites such as Trans-Atlantic Geotraverse (TAG, 0.124–0.147 mM CH4, 0.15–0.37 mM H2, 1.640 mM Fe, 6.7 mM H2S; Charlou et al., 2002).

Rimicaris exoculata can form dense aggregates, of thousands individuals per square metre, living close to the active chimney walls, at temperatures between 10 and 20 °C (Segonzac et al., 1993; Zbinden et al., 2004; Copley et al., 2007; Schmidt et al., 2008). They seem to ‘graze’ the chimney walls, and are continuously in motion. Rimicaris exoculata has an enlarged gill chamber and hypertrophied mouthparts covered by thick microbial layers that could contribute to the shrimp nutrition and/or detoxification (Van Dover et al., 1988; Gebruk et al., 1993, 1997; Segonzac et al., 1993; Polz & Cavanaugh, 1995; Polz et al., 1998; Zbinden et al., 2004, 2008; Komaï & Segonzac, 2008). Cephalothoracic epibionts of R. exoculata from Snake Pit were first affiliated to a single phylotype of Epsilonproteobacteria and assumed to be sulphide-oxidizers (Polz & Cavanaugh, 1995). In a recent study, shrimps from Rainbow displayed a broader diversity of epibionts that may indicate an adaptation to the different geochemical conditions prevailing at this site (Zbinden et al., 2008).

Earlier observations have shown that R. exoculata is neither a predator nor a scavenger. Although its gut seems to be functional (Casanova et al., 1993; Segonzac et al., 1993), the nutrition of this shrimp thus remains unclear. Three main nutrition strategies are currently suggested: (1) Microorganisms from the chimney walls and the environment could constitute the main nutritional source (Van Dover et al., 1988), although δ13C analyses revealed that it was unlikely (Polz et al., 1998); (2) The gill chamber epibionts could be the host's main nutritional source, either by grazing on them (Casanova et al., 1993; Segonzac et al., 1993; Polz & Cavanaugh, 1995; Gebruk et al., 1997, 2000; Rieley et al., 1999) or by ingesting the bacteria and the exuviae after moult, or by transepidermal transfer of organic matter (Zbinden et al., 2004, 2008; Corbari et al., 2008); (3) Shrimps harbour a specific gut microbiota that could constitute an alternative nutritional source (Polz et al., 1998; Pond et al., 2000; Zbinden & Cambon-Bonavita, 2003). In a previous study of shrimps from Rainbow, no intact microbial cell was detected in the stomach, and the digestive microbial communities displayed a low diversity at the phylum level with mainly Epsilonproteobacteria (1/2), Entomoplasmatales (1/4) and Deferribacterales (1/4) (Zbinden & Cambon-Bonavita, 2003). However, as the gut was not empty, it was difficult to determine whether microbial communities were specific to the gut and/or ingested with minerals and surrounding seawater. Although the metabolisms and characteristics of these microbial communities are unknown, high carbon fixation rates have been measured in the gut (Polz et al., 1998), suggesting they could be involved in nutrition and/or detoxification processes.

The present study describes and characterizes the composition of R. exoculata gut-associated microbial communities, using 16S rRNA gene sequence analysis and microscopic observations. Shrimps from two MAR sites (TAG and Rainbow), representing contrasted geochemical settings and depths, were compared in order to test the influence of the hydrothermal habitat over the microbial gut community. To investigate the presence of a resident microbial community, starvation experiments were performed, in order that shrimps completely expelled the bolus, by incubating live R. exoculata specimens for 8, 22 or 72 h in sterile seawater within a pressurized incubator (IPOCAMP, Shillito et al., 2001).

Materials and methods

Sample collection

Samples were obtained during the EXOMAR cruise in 2005 at the TAG (26°8′N–44°50′W, 3650 m depth) and Rainbow (36°14′N–33°54′W, 2320 m depth) hydrothermal vent sites on the MAR. Shrimps were collected using the slurp-gun of the ROV Victor 6000, operated from the RV L'Atalante. Before each dive, the bowls of the slurp-gun used for collecting the shrimps were aseptically washed with ethanol (96%) before being filled with sterile seawater. Once on board, live shrimps were either immediately dissected under sterile conditions and processed as reference samples or placed into the pressurized incubator. The specimens were of the same size and developmental stage (they were in the late anecdysis stage). For DNA analyses, the stomach (foregut) was discarded to keep only the intestine-associated microorganisms.

In vivo experiments

Live shrimps were incubated in a pressurized incubator (IPOCAMP). The stainless-steel vessel (approximately 19 L) is a flow-through pressure system (Shillito et al., 2001). Pressure oscillations arising from pump strokes (100 r.p.m.) were <1 bar at working pressure. The temperature of the flowing filtered (0.4 μm) seawater was constantly measured and regulated, in the inlet and outlet lines (±1 °C).

Less than 2 h after sampling, live and active animals were repressurized at in situ pressure: 230 bars at the Rainbow site and 300 bars at the TAG site (slightly less than the TAG in situ pressure, i.e. 360 bars, due to instrumental limitations). The average incubation temperature was 15 °C (Segonzac et al., 1993; Desbruyères et al., 2001; Zbinden et al., 2004; Schmidt et al., 2008). There were 20 shrimps per pressurized chamber. Previous in vivo experiments showed that the shrimps were under good physiological conditions when repressurized following this procedure (Ravaux et al., 2003; Zbinden et al., 2008). According to previous experiments, the specimens were starved in sterile seawater for 8, 22 or 72 h, to eliminate the bolus and associated microorganisms (Zbinden et al., 2008). The 8-h starvation experiment was performed on Rainbow and TAG specimens, whereas the 22- and 72-h starvation experiments were only conducted for Rainbow shrimps. Live shrimps were dissected immediately after removal from the vessel and digestive tracts (Fig. 1a) were stored for DNA analyses (frozen at −80 °C), FISH analyses and microscopic observations. No faeces could be collected in these experiments as they were completely dissolved in seawater.


Representation of Rimicaris exoculata digestive tract (a). Bacterial gut epibionts inserted between microvilli of the gut epithelium of a Rainbow long-starved shrimp observed in FISH (b, c), semi-thin section (d), SEM (e) and TEM (f). Hybridizations have been performed using the Eub338 Cy3-labelled probe (in red, Amann et al., 1990) and DNA was stained with DAPI (in blue). m, mouth; pc, pyloric chamber; s, stomach. In circles: b, bacteria; mv, microvilli; n, nucleus; bl, basal lamina.

16S rRNA gene sequence analysis

Phylogenetic analysis was performed on six R. exoculata gut clone libraries (Rainbow site: two intestines from pooled reference samples, one from 8-h-starved, one from 22-h-starved, one from 72-h-starved shrimps; TAG site: one from reference sample, one from 8-h-starved shrimp) as follows: DNA was extracted on board on full-length intestines using the FastDNA® SPIN kit for soil (Qbiogen, Santa Ana, CA) following the manufacturer's instructions and kept at 4 °C. Amplification of the bacterial 16S rRNA gene was performed with universal primers E8F/U1492R (respectively 5′-AGAGTTTGATCATGGCTCAG-3′ and 5′-GTTACCTTGTTACGACTT-3′, 1484 bp, annealing temperature 49 °C) or E338F/U1407R (respectively 5′-ACTCCTACGGGAGGCAGC-3′ and 5′-GACGGGCGGTGWGTRCAA-3′, 1069 bp, annealing temperature 54 °C) and for Archaea with A24F/A1492R (respectively 5′-CGGTTGATCCTGCCGGA-3′ and 5′-GGCTACCTTGTTACGACTT-3′, 1468 bp, annealing temperature 49 °C). PCRs were performed using a GeneAmp PCR System 9700 (Applied Biosystems, Forster City, CA) under the following conditions: 3 min at 94 °C, then 30 cycles including 1 min at 94 °C, 1.5 min at the annealing temperature and 2 min at 72 °C, and a final step of 6 min at 72 °C. The PCR reaction mix (50 μL) was composed of 1 × Taq buffer, 0.8 μM dNTP mix (Qbiogen), 10 pmol of each primer (Eurogentec, Liège, Belgium), 2.5 U Taq polymerase (Qbiogen) and approximately 100 ng DNA template. PCR products were cloned using the TOPO® TA Cloning kit (Invitrogen, Carlsbad, CA) following the manufacturer's instructions. Positive clones were sequenced at the ‘Plateforme Biogenouest’ (Roscoff, France, http://www.sb-roscoff.fr/SG/) on an Abi prism 3100 GA using the Big-Dye Terminator V3.1 chemistry (Applied Biosystems).

Sequences were analysed using the NCBI blast search program within the GenBank database (Altschul et al., 1990). They were aligned using clustalw (Thompson et al., 1994) and edited using seaview (Galtier et al., 1996). Phylogenetic trees were constructed using phylo-win (Galtier et al., 1996). The robustness of inferred topologies was tested using 500 bootstrap resampling of the trees (Felsentein, 1985) calculated using the neighbour-joining algorithm (Saitou & Nei, 1987) with the Kimura two-parameter correction matrix. Sequences displaying over 98% similarity were considered to belong to a single phylotype [operation taxonomic unit (OTU)] and were clustered together. Only homologous positions were included in the final alignment.

Sequences were named as: R or T for Rainbow or TAG specimens, respectively, the number in the clone library, and R for reference shrimp, SW, S and LS for 8-, 22- and 72-h seawater starvation experiments, respectively. They are available from the European Molecular Biology Laboratory nucleotide sequence database under accession numbers FM863726FM863780 and FM865857FM865858.

For each library (one reference sample from TAG, two references pooled from Rainbow, one 8-h-starved shrimp from each TAG and Rainbow, one 22-h and one 72-h-starved specimens from Rainbow only), rarefaction curves were drawn using the rarfac program available online at http://www.icbm.de/pmbio/downlist.htm. The library clone coverage was estimated using the formula [1−(n1/N)] (Good, 1953), where n1 is the number of OTUs represented by only one clone and N is the total number of clones. The Shannon index was calculated using the formula

Embedded Image

where S is the number of OTUs, N is the total number of clones and pi is the relative abundance of each OTU (calculated as the proportion of clones of a given OTU to the total number of clones in the community) (Shannon, 1948; Krebs, 1989).


In situ hybridization analyses were performed in order to study the distribution of phylotypes identified in the clone libraries. Samples (n=8: one reference and one 8-h-starved shrimps from TAG; two references, one 8-h-starved, one 22-h-starved specimen and two 72-h-starved shrimps from Rainbow) were fixed for 2 h in formaldehyde 3%–sterile seawater solution and rinsed with phosphate-buffered saline (PBS) 2 ×–sterile seawater buffer (1 : 1). Samples were stored in absolute ethanol-PBS 2 × solution (1 : 1) at −20 °C until use. Samples were embedded in polyethylene glycol distearate-1-hexadecanol (9 : 1) blocks (Sigma, St. Louis, MO) after being dehydrated and soaked (water–ethanol and ethanol–resin series at 37 °C) (Duperron et al., 2007). Blocks were cut into 6–10-μm sections using an RM 2165 microtome (Reichert-Jung, Germany). Resin was eliminated in ethanol and rehydrated sections were hybridized in a reaction mix containing 0.5 μM of each probe in a 10%, 20%, 30% or 40% formamide hybridization buffer [0.9 M NaCl, 0.02 M Tris-HCl, 0.01% sodium dodecyl sulphate (SDS), 10%, 20%, 30% or 40% deionized formamide; see Table 1] and hybridized for 3 h at 46 °C. Sections were washed at 48 °C for 15 min in a washing buffer (respectively 0.450, 0.215, 0.102 or 0.046 M NaCl, 0.02 M Tris-HCl, 0.005 M EDTA, 0.01% SDS) and rinsed briefly. Sections were covered with Slow Fade® Gold antifade reagent containing 4′-6-diamidino-2-phenylindole (DAPI) (Invitrogen), and a cover slip. The universal probes (Eurogentec) were Eub338 (targeting most of the Eubacteria, Amann et al., 1990), Arch915 (targeting Archaea, Stahl & Amann, 1991), GAM42a (targeting Gammaproteobacteria, Manz et al., 1992) and EPSY549 (targeting Epsilonproteobacteria, Lin et al., 2006). New probes were also designed according to our gut clone sequences: Molli352 (targeting gut-associated Mollicutes of R. exoculata, modified from Wang et al., 2004a), Def576 (targeting gut-associated Deferribacteres of R. exoculata, modified from Kumaraswamy et al., 2005), clo4/Epsi653 and clo15/Epsi653 (targeting gut-associated Epsilonproteobacteria of R. exoculata, modified from Polz & Cavanaugh, 1995) and LBI32/130 (targeting R. exoculata cephalothorax methanotrophic Gammaproteobacteria, modified from Duperron et al., 2008) (Table 1). Each probe was used on every sample. Observations were performed on an Olympus BX61 microscope (Olympus Optical Co., Tokyo, Japan) equipped with a U-RFL-T UV light (Olympus Optical Co.) and using a Retiga 2000R camera (Qimaging, Surrey, BC, Canada). Micrographs were analysed using the qcapture pro program (Qimaging).

View this table:

Fluorescent probes used in this study

PhylotypeProbeProbe sequence (5′–3′)Fluorescent dyePosition (rRNA genes)% FormamideReferences
ArchaeaArch915GTGCTCCCCCGCCAATTCCTCy3915 (16S)10–20–30Stahl & Amann (1991)
EubacteriaEub338GCTGCCTCCCGTAGGAGTCy3 or Cy5 or ATTO488338 (16S)10–20–30–40Amann et al. (1990)
Mollicutes R. exoculata gut clonesMollic352GTGAAAAATTCCTTACTGCTGCy3 or ATTO488352 (16S)10–20–30–40This study
Deferribacteres R. exoculata gut clonesDef576CACTGACTTGACAAACCTCy3576 (16S)10–20–30–40This study
EpsilonproteobacteriaEPSY549CAGTGATTCCGAGTAACGCy3549 (16S)20–30Lin et al. (2006)
Epsilonproteobacteria R. exoculata gut clonesclo4/Epsi653ATCTTCCCCTCCCAGACTCTCy3653 (16S)10–20–30–40This study
Epsilonproteobacteria R. exoculata gut clonesclo15/Epsi653ATCTTCCTCTCCCTCACTCTCy5 or ATTO488653 (16S)10–20–30–40This study
GammaproteobacteriaGAM42aGCCTTCCCACATCGTTTCy31027 (23S)20–30Manz et al. (1992)
Methanotrophic Gammaproteobacteria R. exoculata cephalothoracic clonesLBI32/130TCCTGGCTATCCCCCACTACATTO488130 (16S)10–20–30This study

Light and electron microscopic observations

Samples for light microscopy (LM, n=3 digestive tracts: Rainbow reference, 22-h-starved and 72-h-starved shrimps), scanning electron microscopy (SEM, n=5: Rainbow reference, 8-h-starved and 72-h-starved shrimps and two TAG reference shrimps) and transmission electron microscopy (TEM, n=4: Rainbow reference, 8-h-starved and 72-h-starved shrimps and TAG reference shrimp) microscopic observations were fixed for 16 h in 2.5% glutaraldehyde-sterile seawater solution and stored at 4 °C in a NaN3-sterile seawater buffer (Sigma, final concentration 6.7 mM). Samples for SEM were then dehydrated by ethanol series and desiccated with hexamethyldisilazane (Sigma) for 45 min and 5 h in a critical-point dryer CPD 020 (Balzers union, Balzers, Liechtenstein). The digestive tracts cut longitudinally were displayed on a specimen stub, before desiccation, and gold-coated with an SCD 040 (Balzers union). Observations were performed using a Quanta 200 MK microscope (FEI, Hillsboro, OR) and the scandium acquisition program (Soft Imaging System, Munster, Germany). For LM and TEM observations, samples were postfixed in 1% osmium tetroxide, dehydrated in ethanol and polypropylene oxide series before they were embedded in an epoxy resin (Serlabo, Paris, France). Semi-thin (800 nm) and ultrathin (50–75 nm) sections were cut with an ultramicrotome Ultracut E (Reichert-Jung) using a diamond knife. Semi-thin sections were stained with toluidine blue for observations by LM (Nikon Optiphot-pol microscope and Zeiss Opton photomicroscope). Ultrathin sections were displayed on copper grids and contrasted with uranium acetate and lead citrate. Observations were carried out on a Philips 201 electron microscope, operating at 80 kV.

Results and discussion

In vivo experiments and sample considerations

The shrimps analysed were actively swimming before, during and after the incubation experiments. 16S rRNA genes were successfully amplified and cloned from the total DNA extracted from the midgut and the hindgut (Fig. 1a). Semi-thin sections and macroscopic observations of the digestive tracts indicated that the 8- or 22-h starvation incubations were not long enough to completely expel the bolus, as all the guts observed were still full of minerals (Supporting Information, Fig. S2a). However, after the 72-h starvation experiment, the guts were empty. In order to identify a possible resident microbial community, the gut microbial composition of reference shrimps (TAG and Rainbow) and animals incubated for 8 and 22 h was therefore compared with 72-h-starved shrimps.

Microscopic observations of the R. exoculata gut anatomy

The anatomy of the R. exoculata digestive tract showed a shorter foregut (mouth, stomach and pyloric chamber) and hindgut (Fig. 1a) compared with other crustaceans (Komaï & Segonzac, 2008). The gastric mill usually has a crushing function. The microscopic observations showed that spinules and setae on the stomach wall of R. exoculata were less abundant than for scavenger hydrothermal shrimps such as Chorocaris chacei and Mirocaris fortunata (data not shown). Rimicaris exoculata had more developed setae at the beginning of the pyloric chamber than C. chacei and M. fortunata. This suggests that R. exoculata may have a low mechanical digestive activity.

The midgut of R. exoculata, the central digestive absorption zone deprived of a cuticle, represented two-thirds of its total gut length, while it represents only between one sixth and one third of the total gut length in most crustaceans (Milne-Edwards, 1840). Cross-sections of R. exoculata midgut also revealed large exchange surfaces for all specimens (reference and starved animals), as the epithelium displayed numerous invaginations and cells with dense microvilli. These cells were typical active digestive cells, with large nuclei (Fig. 1d) and many mitochondria (MET observations, Fig. S2c). The microscopic observations of the bolus revealed that it was mainly composed of organic matter (few cuticle fragments and probably degraded microorganisms) and minerals (Fig. S2a; no total organic matter analysis of the bolus was performed), which may also suggest a seawater intake activity.

Microscopic observations of the microbial communities

In LM, SEM and TEM observations, the numerous mineral particles representing most of the bolus made the detection of microorganisms difficult. All our observations of the bolus have shown rare disc-shaped bacteria, which are reported on the vent chimney walls (Van Dover et al., 1988) and in the gill chamber (Polz & Cavanaugh, 1995; Zbinden et al., 2004). These rare disc-shaped cells were intact, apparently not subjected to rapid digestion. In a previous work, it was shown that the cephalothoracic filamentous mat was not scraped (Zbinden et al., 2004; Corbari et al., 2008). Moreover, the seawater flow in the gill chamber enters the chamber towards the gills and exits bathing the filamentous epibiotic bacteria (Casanova et al., 1993; Zbinden et al., 2004). Hence, the seawater flow is at the opposite of the mouth, which probably does not allow epibiont ingestion. The observation of microbial cells associated with the bolus was easier using fluorescence microscopy. A positive signal using the Eub338 probe (Eubacteria-specific) revealed the bacterial cells within the bolus (single rods or rods in small aggregates, cocci and rare disc-shaped cells). Part of the rods and a few cocci were detected using the Gammaproteobacteria-specific probe Gam42a. A few cocci hybridized using the LBI32/130 probe (data not shown). This probe was designed to target methanotrophic Gammaproteobacteria associated with R. exoculata gill chamber epibionts (Zbinden et al., 2008). A few rods and all disc-shaped bacteria successfully hybridized using the Epsilonproteobacteria-specific probe EPSY549. These rods were also positively labelled using the two R. exoculata gut clones-specific probes clo4/Epsi653 and clo15/Epsi653. The rod-shaped bacteria in small aggregates successfully hybridized using the Deferribacteres-specific probe Def576 (Fig. S2d and e). A few Deferribacteres were also retrieved associated with the bolus along the entire gut, regardless of the site. No signal was ever obtained using the Molli352 probe on the gut content.

In the midgut, the microscopic observations showed numerous long (up to 15 μm) and thin (0.2–0.3 μm) filaments, corresponding to individual cells without any visible septum, inserted between the microvilli of the epithelial gut cells (Fig. 1d–f). These filamentous bacteria were observed in dense populations within the gut of all specimens (Fig. 1b–e), with no visible difference in their distribution and abundance, regardless of the collection site and starvation treatment. In situ hybridizations using the Eubacteria-specific probe Eub338 indicated that all these filaments were active bacteria (Fig. 1b and c). No Deferribacteres were observed associated with the gut epithelium. No clear in situ hybridization signals were obtained from the filaments using Gammaproteobacteria-specific probes (listed in Table 1) or using the Molli352 probe. A weak positive signal was observed from these epithelium-associated communities using the Epsilonproteobacteria-specific probe EPSY549, even after the 72-h starvation experiment. These bacterial communities were clearly separated from the bolus by the peritrophic membrane (Fig. S2b), a natural barrier that preserves the gut epithelium of arthropods from mechanical abrasion and microbial infections (Mercer & Day, 1952; Brandt et al., 1978). The filamentous morphology of these bacteria associated with the large surfaces they occupy, thanks to the numerous invaginations of the gut epithelium, yields a large exchange surface between microbial communities and their environment. As the midgut is not subject to exuviation, this probably favours long-term microbial colonization and interactions of a resident microbial community with its host.

This community may play a significant role in the detoxification of compounds present in the bolus, such as minerals or heavy metals. Detoxification has been described for Limnoria tripunctata, a wood-boring marine isopod (Zachary & Colwell, 1979; Zachary et al., 1983). The resident gut microbiota of L. tripunctata was only observed in specimens inhabiting creosote-treated wood, but it was absent if reared on nontreated wood. Therefore, it suggested that these gut microbial communities could contribute to the creosote resistance of the isopod. Analogously, as the L. tripunctata bacterial community was also in close association with the intestinal epithelium within the peritrophic space, the R. exoculata gut microbial communities might participate in mineral/metal detoxification.

Archaeal communities of the gut

The composition of archaeal communities from the gut of reference and starved R. exoculata at the TAG and Rainbow sites was investigated using 16S rRNA gene clone libraries (detailed in Tables 2, 3 and S1). The archaeal sequences from the gut of TAG shrimps (34) were affiliated to the Euryarchaeota lineages DHVE 2 (deep-sea hydrothermal vent Euryarchaeota, 23), Thermococcales (6) and to Crenarchaeota Marine Group I (5). In contrast, all archaeal-related clone sequences from the gut of Rainbow shrimps (39) were affiliated to Methanococcales. Although these archaeal lineages are usually found at hydrothermal vents (e.g. Takai & Horikoshi, 1999; Reysenbach et al., 2000; Nercessian et al., 2003, 2005; Schrenk et al., 2003; Roussel & Cambon-Bonavita, unpublished data), the TAG and the Rainbow gut-associated archaeal populations were clearly different, probably due to the contrasted geochemical conditions characterizing both sites. Interestingly, no archaeal cells were ever detected using in situ hybridization with the general Archaea-specific probe Arch915 (Stahl & Amann, 1991), whatever the condition used (Table 1), and no archaeal sequences were retrieved after long starvation incubations. These results suggest that these archaeal communities were probably rare and/or not active. Therefore, they probably do not belong to the gut microbiota, but more likely have been ingested with chimney particles and fluids.

View this table:

Distribution of the bacterial 16S rRNA gene clones from Rainbow and TAG reference and starved shrimps. The main phylogenetic group per sample is shown in bold

Phylogenetic groupsNumber of clones
Reference samplesStarvation 8 hStarvation 22 hStarvation 72 hTotal
Rainbow (two guts)TAGRainbowTAGRainbowRainbow
View this table:

Closest match of representative 16S rRNA gene clone sequences

Phylogenetic groupRepresentative clone sequencesHit of blast (accession no.)Similarity (%)No. of clones
BacteriaR36S, R28SW, R82RR. exoculata gut clone 62 (AJ515723)99238
GammaproteobacteriaR67LSEndosymbiont of Alviniconcha sp. type 1 clone SyA1-P1 (AB235235)954
R57LSPseudomonas entomophila strain L48 (CT573326)981
R32RR. exoculata gill chamber clone LBI32 (AM412518)964
R48LSGammaproteobacterium clone Belgica2005/10-ZG-8 (DQ351804)932
R53LSAlteromonas sp. strain SHY1-1 (AB078014)993
R16LSZebrafish gut clone aab28h07 (DQ819366)9911
R68LSIron-reducing enrichment clone Cl-A2 (DQ676994)991
R56LSPhotobacterium phosphoreum strain RHE-01 (AY435156)989
R19LSEndosymbiont of Acanthamoeba sp. Ac309 (AY549549)981
EpsilonproteobacteriaR3R, R30LSHydrothermal vent gastropod clone SF_C23-F4 (AY531582)98–976
R62LSHydrothermal vent gastropod clone SF_C23-C8_shell (AY531600)962
R69LSR. exoculata gut clone 4 (AJ515714)9912
R28R, R23LSR. exoculata gut clone 11 (AJ515717)9913
MollicutesR67SW, R2RR. exoculata gut clone 42 (AJ515720)9924
T8SWR. exoculata gut clone 69 (AJ515722)9916
T28SWGillichthys mirabilis gut clone C13 (DQ340200)864
FirmicutesT17RMammals gut clone Saki_aaj62c07 (EU461853)871
R8R, R2LS, T17SW, R26LSGillichthys mirabilis gut clone C13 (DQ340200)81–918
R5LSCandidatus Bacilloplasma’ isopod gut clone P10 (DQ485976)802
CFBR21R, R21SW, R39SWArctic sediment clone SS1_B_01_16 (EU050905)94–974
T70RHydrothermal sediments clone p816_b_3.23 (AB305587)921
R14RBacteroidetes clone C319a-R8C-D4 (AY678514)941
VerrucomicrobiaeT72RHydrothermal Verrucomicrobia clone pItb-vmat-33 (AB294943)871
T20RParalvinella palmiformis clone P. palm C 136 (AJ441224)92–964
BetaproteobacteriaR64LSRalstonia sp. clone EMP_AD31 (EU794311)991
DeltaproteobacteriaR65SWGeothermobacter sp. Fe30-MC-S (AB268315)931
MethanococcalesR5RMethanococcus aeolicus strain Nankai-3 (DQ195164)9739
DHVE 2T48R, T4RHydrothermal clone met43 (DQ082955)93–9823
ThermococcalesT38Thermococcus siculi strain DSM 12349 (AY099185)996
Marine group 1T24RHydrothermal clone pEPR624 (AF526982)995

The gut bacterial distribution and composition

Six bacterial 16S rRNA gene clone libraries were constructed (Table 2), representing a total of 376 clone sequences from reference shrimps (154) and from starved shrimps (222). Although the analysis of clone libraries is only partially quantitative, as the sampling methods and molecular techniques introduce biases (e.g. Bent & Forney, 2008; Quince et al., 2008), the phylogenetic diversity of the microbial communities associated with R. exoculata midgut for each starvation incubation time can be compared, although caution is needed. As the number of clones per sample was low, rarefaction analyses were conducted for each library. Results indicated that clone libraries adequately represented the composition of the communities in the gut contents (see Fig. S1 for rarefaction curves), because curves reached a plateau for the number of clones investigated. The clone coverage was satisfactory for the six clone libraries: 92% for Rainbow and TAG reference shrimps, 96% for the Rainbow 8-h-starved specimen, 97% for the TAG 8-h-starved specimen, 95% for the 22-h-starved shrimp and 94% for the 72-h-starved shrimp. On average, the six 16S rRNA gene bacterial clone libraries were mostly dominated by four phyla affiliated to Deferribacteres (63%), Mollicutes (12%), Gammaproteobacteria (10%) and Epsilonproteobacteria (9%). Although sequences affiliated with Firmicutes, Cytophaga–Flavobacter–Bacteroides (CFB), Verrucomicrobiae, Delta- and Betaproteobacteria were also detected, they represented a small fraction of the clones (6%). The overall bacterial communities' composition, detailed in Tables 2 and S1, was consistent with a previous molecular analysis of the R. exoculata gut (Zbinden & Cambon-Bonavita, 2003). Clone libraries from the Rainbow and TAG reference shrimps showed a similar bacterial community composition. In both, the clones were mainly related to Deferribacteres (75%), Mollicutes (10%), Epsilonproteobacteria (6%), Gammaproteobacteria (3%) and other groups (CFB and Firmicutes, 6%). The compositions of gut bacterial clone libraries after the 8-h starvation experiment (on both sites) and the 22-h starvation incubation (on Rainbow site) were also similar and still dominated by Deferribacteres (81% and 96%, respectively). In contrast, the 72-h starvation clone library was dominated by Gammaproteobacteria (40%), Mollicutes (23%), Epsilonproteobacteria (23%) and other minor lineages (11%), whereas the Deferribacteres represented <3% of the sequences (Tables 2 and S1).

All sequences affiliated to the Deferribacteres phylum (238) represented a single phylotype previously detected from R. exoculata gut (Table S1 and Fig. 2a; R. exoculata gut clone 62 and 91, Zbinden & Cambon-Bonavita, 2003). To date, the R. exoculata gut clone-related genera, i.e. Mucispirillum and Geovibrio, have only been detected from gut microbiota, sediments and oil reservoirs and were never reported from hydrothermal vents, suggesting that this phylotype could be specific to R. exoculata. Moreover, the high similarity level between all Deferribacteres-related sequences (>99%) associated with shrimps, regardless of the site, suggests a long-term, specific association between these bacteria and their host. Hence, the Deferribacteres from this cluster probably are part of R. exoculata-specific gut microbial community. The difference in the proportion of Deferribacteres-related sequences between the 72-h-starved specimen clone library (<3%) and all the other clone libraries (at least 70%) may be explained by several hypotheses: (1) Deferribacteres could have been free-living in the gut (not attached to the gut wall) and eliminated with the bolus after the 72-h starvation experiment. (2) Some Deferribacteres use iron and sulphides as energy sources (Miroshnichenko et al., 2003), and therefore would not survive if deprived of these compounds during long starvation periods. (3) Some Deferribacteres are heterotrophic bacteria, therefore using organic matter of the bolus, such as ingested microorganisms or cuticle fragments. (4) All Deferribacteres known so far are strict anaerobes. During the starvation experiment, the reduced minerals are evacuated with the bolus. The gut is then probably under aerobic conditions, which could impair the Deferribacteres growth and maintenance.


Phylogenetic trees based on 16S rRNA gene sequences from the digestive tract of reference and starved Rimicaris exoculata from TAG and Rainbow hydrothermal sites. They represent the main bacterial phylogenetic groups of microorganisms associated with the shrimp gut: Deferribacteres (a, calculated on 882 bp), Mollicutes (b, calculated on 796 bp), Epsilonproteobacteria (c, calculated on 841 bp) and Gammaproteobacteria (d, calculated on 842 bp). The robustness was tested using 500 bootstrap resampling of the trees calculated using the neighbour-joining algorithm with the Kimura two-parameter correction matrix. Sequences were named as: R or T for Rainbow or TAG specimens, respectively, the number in the clone library, and R for reference shrimp, SW, S and LS for an 8-, 22- and 72-h seawater starvation experiments, respectively. Our clones are shown in bold.

Sequences related to Mollicutes also represented a significant proportion of the clone libraries (13% from reference shrimps) and this even after a 72-h starvation experiment (23%, Fig. 2b and Table 2). Three clusters were identified and affiliated to Entomoplasmatales and Mycoplasmatales. Clusters B and C were detected in specimens from both TAG and Rainbow sites, and after all starvation experiments (Table S1), suggesting that they were not a site-specific community and so reinforcing the hypothesis of a specific symbiotic relationship. Entomoplasmatales and Mycoplasmatales bacteria are usually heterotrophic pathogens, commensals or symbiotic bacteria associated with vertebrates (mammalians or fishes), insects, crustaceans or plants (Razin, 1978, 1998; Clark, 1984; Regassa & Gasparich, 2006). The cluster A (24 sequences, Fig. 2b) was closely related to a sequence detected from a starved Mediterranean shrimp (98% similarity), Pestarella tyrrhena, suggesting that Mollicutes are ubiquitous of crustaceans' gut and so could also be part of the resident digestive tract microbiota. The cluster A-affiliated sequences were retrieved only in Rainbow specimens (Table S1). This microbial population could be site-dependant or this could be due to the number of sequences analysed. The cluster B (16 sequences, Fig. 2b) was closely related to R. exoculata gut clones within the Mycoplasmatales order (99% similarity; Zbinden & Cambon-Bonavita, 2003). Cluster C (four sequences, Fig. 2b) was affiliated to epibiotic phylotypes from crustaceans (isopods) and fish guts, suggesting they could also represent resident microbial phylotypes. Interestingly, all the R. exoculata gut sequences related to Mollicutes were affiliated to epibionts from other crustaceans such as P. scaber, which harbours long or spherical stalked microorganisms inserted between the microvilli of midgut glands (Wang et al., 2004a, 2007). Although microscopic observations showed that the morphology and size of R. exoculata gut epibionts (long thin filamentous single cells) were similar to some Spiroplasmas (long thin bacteria, 0.1–0.35 μm diameter, up to 5 μm length, Garnier et al., 1981), suggesting they could be Mollicutes (Fig. 1b–f), no clear in situ hybridization signal was ever detected using the Mollicutes-specific probe for each stringency condition and sample tested (Table 1). However, Wang et al. (2004a) reported a very low hybridization signal on the Mollicutes symbionts of the midgut glands of P. scaber. Moreover, as even the DAPI staining of the filaments was difficult to observe, the very low Mollicutes hybridization signals could also be due to the size of these very thin bacteria (∼0.3 μm diameter). Mollicutes are polymorphic microorganisms (spheroid, filamentous, ramified, helicoïdal) characterized by an absence of a cell wall, and therefore poorly resistant to extreme environments. Usually, Mollicutes have a reduced genome and depend on host nutrients (Maniloff & Morowitz, 1972; Razin, 1978, 1998; Clark, 1984; Regassa & Gasparich, 2006). The midgut of R. exoculata does not moult and is relatively independent of environmental conditions, suggesting that it could be a sufficiently stable habitat to harbour Mollicutes. Therefore, Mollicutes could be a part of the resident microbial community associated with the gut wall, able to survive to a long-term starvation.

Sequences related to Epsilon- and Gammaproteobacteria from the gut of the reference specimens were dominant in the clone library of the 72-h starvation experiment (63%) (Tables 2 and S1, Fig. 2c and d). The values of the Shannon index are low after an 8- and a 22-h starvation experiment (1.01 for Rainbow and 0.65 for TAG 8-h-starved shrimps and 0.27 for Rainbow 22-h-starved shrimp). This may be explained by the low number of clones treated per sample. However, we also suggest that microbial density is low and so, molecular approaches favour only the dominant population. After the 72-h starvation experiment, the Shannon index value is high (4.97) as Gammaproteobacteria-related sequences were highly diverse. Their dominance could be a possible consequence of the lesser quantity of detectable 16S rRNA genes related to Deferribacteres. We also suggest that the Gammaproteobacteria were in a latent state in reference shrimps and between 8 and 22 h of starvation, and were actively dividing at 72-h starvation. This could be explained by the switch of the physiological conditions in the empty digestive tract. The elimination of the bolus led to the evacuation of reduced minerals and intake of oxygen. This could have been favourable to the Proteobacteria and deleterious to the Deferribacteres, which are strict anaerobe microorganisms. Regarding the Gammaproteobacteria-related sequences retrieved in the Rainbow reference shrimp clone library, they were all affiliated to the gill chamber clone LBI32 (AM412518, Zbinden et al., 2008). This sequence clusters in the methanotrophic Gammaproteobacteria symbionts group. Positive in situ hybridization signals were observed on methanotrophic-like bacteria. Hence, these Gammaproteobacteria-related sequences are probably site-dependant, which could be due to the geochemical conditions prevailing on this ultramafic site.

One-third of the epsilonproteobacterial sequences from starved shrimps were affiliated to R. exoculata gut sequences (99% similarity, Zbinden & Cambon-Bonavita, 2003) and to the R. exoculata ectosymbiont thought to be a sulphide-oxidizing chemoautotroph (Polz & Cavanaugh, 1995). Other Epsilonproteobacteria and part of the Gammaproteobacteria sequences from starved shrimps were affiliated to autotrophic epibiont-like microorganisms mainly associated with a hydrothermal gastropod (Goffredi et al., 2004). Other Gammaproteobacteria were affiliated to heterotrophic microorganisms (e.g. Pseudomonas entomophila CT573326 and Alteromonas sp. AB078014). Moreover, the sequences closely clustering within the Epsilonproteobacteria cluster A (99% similarity) were detected in the reference shrimps and also in the 22- and 72-h-starved shrimps (Fig. 2c). Hence, as suggested previously, specific Proteobacteria phylotypes could also be part of the local midgut resident microbiota (Polz et al., 1998; Zbinden & Cambon-Bonavita, 2003). From all our libraries, no sequence was affiliated to the gill chamber clones, except one (clone R32R) related to clone LBI32 (96% of similarity). This indicates that in our experiments, on both sites, the gill chamber epibionts were little ingested.

New insights in the shrimp nutrition and role of the microbial communities

Observations of the shrimps' behaviour indicated that they were healthy before and after treatments. Microscopic observations of brush cells showed that they were intact in all specimens, suggesting that the long filaments inserted between microvilli are probably not pathogenic.

Part of the community associated with the bolus was probably ingested nonspecifically with minerals from chimney walls and surrounding seawater (Archaea and some Proteobacteria identified within the bolus) and expelled in the faeces. The intake of environmental seawater and chimney particles could play a role in the diet of the host.

The gill epibiont communities are thought to represent the second major nutritive contribution (Gebruk et al., 1993, 1997; Segonzac et al., 1993; Polz & Cavanaugh, 1995). In our analyses, cephalothoracic epibionts do not seem to be that much ingested. Hence, transepidermal exchanges could be the main nutritional source from the gill epibionts as proposed before (Zbinden et al., 2004, 2008).

Microscopic observations, 16S rRNA gene clone libraries analyses and in situ hybridizations indicated that gut-associated bacterial communities were rather similar in all specimens from TAG and Rainbow, except for Gammaproteobacteria, and presented a limited diversity of phylotypes. Most of the bacterial sequences were related to eukaryote-associated bacteria usually considered as symbiotic, rather than to free-living vent environmental bacteria. Taken together, these data clearly reinforce the hypothesis of a symbiotic relationship between R. exoculata and at least certain members of gut-associated bacterial communities.

A previous study had measured autotrophic carbon intake in the gut (Polz et al., 1998). Some of the sequences retrieved in this study were affiliated to the Proteobacteria lineage and clustered with sequences from autotrophic bacteria. Mollicutes are usually heterotrophic bacteria and may be involved in organic matter degradation. The Deferribacteres species are usually heterotrophic microorganisms involved in sulphur compounds and iron cycles. Thus, they may be implicated in the nutrition of the shrimp and in detoxification processes. These results suggest that several metabolic pathways likely co-occur within the epibiotic community.

Culturing attempts are in progress in the laboratory. However, as the epibionts are usually refractory to cultures, metagenomic approaches will be conducted in order to better understand the epibionts' roles.


We profusely thank Philippe Crassous (DEEP/LEP, Ifremer) and Isabel Le Disquet (IFR 83 de Biologie Integrative – CNRS/Paris VI) for advice and work at the scanning electron microscope. TEM micrographs were taken by the Service de Microscopie Electronique, IFR 83 de Biologie Integrative – CNRS/Paris VI. Thanks to ‘Plateforme Biogenouest’ for sequencing work. We thank Anne Godfroy, chief scientist of the EXOMAR cruise, the Captain and crew of R/V L'Atalante and ROV Victor 6000 team for their efficiency. Finally, we are indebted to several colleagues for helpful comments and suggestions. This work was supported by Ifremer, Region Bretagne, GDR ECCHIS and ANR DEEPOASES.


  • Editor: Julian Marchesi


View Abstract