Tracing carbon assimilation in endosymbiotic deep-sea hydrothermal vent Mytilid fatty acids by 13 C-fingerprinting

Introduction Conclusions References


Introduction
Deep-sea hydrothermal vents host peculiar ecosystems fueled by methane, sulfide, iron or even hydrogen (e.g., Perner et al., 2009). The symbiotic association of bacteria with marine invertebrate hosts provides the former with access to the chemical substrates necessary for their metabolism and the latter with a source of organic carbon (and nitrogen). One of the two species of endosymbiotic bathymodioline Mytilids occurring along the Mid-Atlantic Ridge (MAR), Bathymodiolus azoricus, is found at the northernmost sites Menez Gwen, Lucky Strike and Rainbow (Desbruyères et al., 2001). Transmission electron microscopy has demonstrated the presence of two distinct Gram-negative bacterial endosymbionts inside specialized gill epithelial cells (bacteriocytes). One of these symbionts has centrally stacked intracytoplasmic membranes characteristic for methane-oxidising gamma-proteobacteria (Fiala-Médioni et al., 2002). Analysis of 16S rRNA sequences followed by fluorescence in situ hybridization (FISH) evidenced that B. azoricus endosymbiotic 16S rRNA phylotypes cluster with natural symbiotic and cultured sulfide-oxidising (SOX) and methane-oxidising gamma-proteobacteria (Won et al., 2003;Duperron et al., 2006;Spiridonova et al., 2006). Enzymatic and physiological assays using gills of B. azoricus from the Lucky Strike and Menez Gwen vent sites revealed the presence of active enzymes of the metabolic cycles of inorganic C assimilation and sulfide oxidation and of the C 1 carbon assimilation pathway (Fiala-Médioni et al., 2002). Furthermore, Riou et al. (2008) report active 13 C incorporation from bicarbonate in the presence of sulfide as well as from methane within the gills of B. azoricus. This incorporation was followed by carbon transfer to the host's aposymbiotic muscle tissue.
In order to understand the modes of matter and energy transfer to the host, the phenotypes of B. azoricus symbionts' need to be characterized. Classification of phospholipid ester-linked fatty acids (PLFA) profiles has proven useful to clarify bacterial genus and species interrelationships established by DNA-based phylogeny (Bodelier et al., 2009). PLFA profiles thus allow to distinguish between different methane-oxidising bacteria (MOB) species as evidenced for example by the fact that methane-oxidising gammaproteobacteria (formerly named after "Type I MOB") mainly contain fatty acids with 14 and 16 carbon atoms, while methane-oxidising alpha-proteobacteria (formerly "Type II MOB") PLFA are mainly composed of 18 carbon atoms . In addition, MOB bacteria possess fatty acids that are not found in any other known microorganism (methane-oxidising gamma-proteobacteria: 16:1(n − 8) and 16:1(n − 5)t; alpha-proteobacteria: 18:1(n − 8)) and these compounds therefore represent valuable biomarkers (e.g. Nichols et al., 1985).
The main difficulty consists in the impossibility to grow B. azoricus' symbionts in pure cultures. Tracer experiments represent powerful tools enabling the detection of the bacterial fatty acid signal inside the host. However, in a tissue total lipid extract, the symbionts' fatty acid molecules are "diluted" by the fatty acids of the eukaryotic cells since the bacterial symbionts are hosted inside mussel cells. Unlike plant and animal cells, bacteria do not contain acyl lipid stores (lipids containing O-or N-ester or ether linked fatty acids), and it is believed that their acyl lipids are confined to membranes. Phospholipids represent 90 to 98% of bacterial lipids (King et al., 1977;White et al., 1979) and around 50% of eukaryotic lipids (Vestal and White, 1989). Studying PLFA thus somehow "concentrates" the signal of the bacteria, and is more adapted than a total fatty acid study in the perspective of identifying the fatty acids that are produced by the symbionts from the assimilation of CH 4 and CO 2 (in the presence of sulfide).
In the present study, we thus investigated the incorporation of 13 C-labeled HCO − 3 (in the presence of H 2 S) or CH 4 into gill-extracted PLFA from B. azoricus in order to assess the metabolic activity of the endosymbionts. This enabled us to establish specific fatty acid (FA) patterns which provide insights into the symbiont phenotypes. On the contrary, a technique with a high yield of tissue total fatty acid (tFA) recovery was preferred to retain most of the information relative to both the symbionts and the host's fatty acids when analysing the assimilation of 13 C-labeled amino acids (dissolved into seawater) into B. azoricus gill fatty acids. This last experiment was indeed designed to observe (i) the potential for heterotrophic growth by the symbionts, and/or (ii) the occurrence of lipogenesis from the assimilation of amino acids by the Mytilid cell machinery (which could also use free amino acids as osmotic regulators of the cells, metabolic fuel, protein synthesis or in the glucogenesis). Close examination of the labeled FA patterns helped interpreting physiological processes occurring in B. azoricus.

Sampling and aquarium experiments
During the MOMARETO cruise (R/V Pourquoi Pas?) in August 2006, a cage was deployed at the Menez Gwen site (37 • 51 N-32 • 31 W, 817 m) and loaded with around 400 mussels. The cage was retrieved in May 2007 by the Portuguese vessel R/V Arquipélago. Bathymodiolus azoricus specimens were transferred to cooled seawater (9 • C) for a 14 h transit to Horta, Faial Island (Azores). Their valves were scrubbed clean of visible material, rinsed in chilled seawater and transferred to the Azorean land-based hydrothermal vent laboratory, LabHorta (Colaço and Santos, 2003). Before the start of the experiments, mussels were kept for 38 days in aerated seawater amended with methane and hydrogen sulfide, as described in Riou et al. (2008).
Three mussels from each experiment were selected for fatty acid analysis, and dissected into gill, mantle, muscle, and remaining tissues. Mussel tissues were immediately stored at −20 • C till they were freeze-dried, a few days after dissection.

PLFA and tFA preparation
Prior to use, all glassware was precombusted 4 h at 450 • C. Lipids were extracted from approximately 100 mg dry tissue (ground to a fine powder right before extraction, using a mortar and pestle) by a modified Bligh and Dyer protocol (Boschker et al., 1998), whereby water in the first extraction step was replaced by Sörens phosphate buffer pH 7.4 (White et al., 1979). Lipid extracts were fractionated on silicic acid columns into different polarity classes by sequential elution with chloroform, acetone and finally methanol. The phospholipids collected in the methanol fraction were derivatized using mild-alkaline methanolysis (using a methanolic KOH solution) to yield fatty acid methyl esters (hereafter refered to as PLFA since they were derived from the phospholipid fraction).
Total fatty acid methyl esters (tFA) were obtained from 25 mg tissue powder by direct acid methanolysis catalysed for 2 h at 90 • C (modified from Lewis et al., 2000) in 3 mL of a fresh solution of methanol/hydrochloric acid 37%/chloroform (10:1:1 volume). After cooling down to room temperature chloroform (1 mL) and water (1 mL) were added to the samples which were then well mixed before the biphasic system was allowed to separate. The chloroform fraction was transferred to a fresh tube and the upper phase was reextracted with fresh chloroform (1 mL). Chloroform extracts were pooled, cleaned with 2 mL water, dried under a mild nitrogen flow and the tFA were re-suspended in 300 µL hexane.

FAME 13 C content analyses and identification
The isotopic composition of individual FAME was analysed by gas chromatography-combustion-isotope ratio mass spectrometry (GC-c-IRMS) using an HP6890 coupled to a Thermo Finnigan delta+XL via a GC/C III interface. The FAME mix and Ag + -SPE fractions were resolved on a fused-silica capillary column (100 m × 0.25 mm) coated with 0.20 µm CP-Sil 88 (100% cyanopropyl polysiloxane, Varian BV, The Netherlands) after injection in splitless mode at 270 • C. Helium was used as a carrier gas at a flow rate of 1 mL min −1 and the following temperature program was applied: 4 min at 45 • C; an increase at 10 • C min −1 to 135 • C and a plateau of 90 min; an increase at 5 • C min −1 to 170 • C and a plateau of 25 min; an increase at 10 • C min −1 to 195 • C and a plateau of 15 min; an increase at 10 • C min −1 to 235 • C and a plateau of 5 min. A careful selection of the type of chromatographic column and the optimization of the Bathymodiolus azoricus (n = 3 specimens, mean ± minmax) gill FAME carbon isotopic signatures. The δ 13 C signatures were corrected for the methyle group added during the transesterification using formula (1), and weighed averages were obtained using formula (2) (A) Gill PLFA after 15 days with 28 µM 25% 13 CH 4 (black bars) or 20 days with 6 µM H 2 S + 16% H 13 CO − 3 (grey bars). (B) Gill total fatty acids (tFA) after 20 days in control conditions (striped bars) or 20 days with 300 µM dissolved 98% 13 C-amino acids (white bars). Sat: saturated, MUFA: mono-unsaturated, DUFA: di-unsaturated, PUFA: poly-unsaturated chains.
temperature program (enabling the detection of FAME 10:0 to 24:0) ensured baseline resolution of the IRMS m/z 44 trace for the majority of the components (including 16:1 and 18:1 positional isomers) and enabled an accurate identification of labeled PLFA. The δ 13 C ratios of each FAME was corrected for the addition of one methanol carbon per molecule to obtain the isotopic signature of the fatty acid as in Abrajano et al. (1994) using the mass balance equation taking a measured δ 13 C CH 3 OH value of −40.3‰: 2594 V. Riou et al.: Fatty acid 13 C-fingerprinting in Mytilid-bacteria symbiosis Where x is the fractional carbon contribution of the free fatty acid to the methyl ester (e.g.: 18/19 for FA 18:1(n − 7)). The weighed δ 13 C signature of fatty acid methyl esters, grouped according to their degree of unsaturation ( Fig. 1), was obtained using the following formula (with monounsaturated fatty acids -MUFA-taken as an example of fatty acid class): where %MUFA x is the area obtained from MUFA x on the chromatogram relative to the sum of the areas obtained from all fatty acids, and δ 13 C MUFA x is the carbon isotopic signature of MUFA x . For each PLFA, the incorporation of 13 C (I, expressed as micrograms of 13 C per gram of total PLFA) was calculated as in Knief et al. (2003): where A x is the peak area of PLFA x divided by the sum of the peak areas of all of the PLFA. F is the fraction of 13 C in PLFA x of samples incubated with 13 C (F l ) or in tFA x of control unlabelled samples (F u ): F = 13 C/( 13 C + 12 C) = R/(R + 1). The carbon isotope ratio (R) was derived from the measured δ 13 C values as follows: R = (δ 13 C/1000 − 1) · R VPDB , with R VPDB = 0.0112372. FAME identification was achieved by GC-mass spectrometry (GC-MS) using a Thermo Finnigan TRACE GC-MS system, applying the same GC conditions as described for the GC-IRMS analyses. The mass spectrometer was operated under mild conditions of electron impact ionization (EI + ; 40 eV) and recorded the mass spectra in the scan mode (m/z = 50-400). Aliquots of the PLFA or tFA mixture and of the MUFA, DUFA and PUFA fractions were derivatised further by a one step reaction into dimethyloxazolines (DMOX) to locate the unsaturations by GC-MS (Fay and Richly, 1991). We used a temperature gradient similar to the one used for FAME analyses, but since DMOX derivatives are less volatile (see also Precht and Molkentin, 2000) we increased the temperature of the isotherm plateaus at 135 • C, 170 • C and 195 • C for the temperature program described above, by 10 • C to 145 • C, 180 • C and 205 • C, respectively. The International Union of Pure and Applied Chemistry PLFA nomenclature used here is described by Guckert et al. (1985): the (n−) notation indicates the position of the carbon-carbon double bond in the FA aliphatic chain starting from the methyl end carbon. The notation, giving the location of FA unsaturation from the carbon at the carboxyl end, is only used in the next paragraphs to discuss fatty acid synthesis.
Phylogenetic analyses on sequences of the genes encoding the 16S rRNA subunit and the particulate methane monooxygenase (pmoA gene) revealed that the B. azoricus methaneoxidising endosymbiont is a gamma-proteobacterium related to free living MOB and to MOB symbionts from other Bathymodiolids (Duperron et al., 2006;Nakagawa and Takai, 2008;Spiridonova et al., 2006). Although Colaço et al. (2007) did not identify any MOB PLFA biomarkers in wild B. azoricus specimens from the Menez Gwen site they did reveal the presence of large amounts of i19:0, 18:1(n − 13) and 18:1(n − 9), which were attributed to the endosymbionts. Our tracer experiment with 13 CH 4 revealed that 18:1(n − 13) was indeed slightly labeled but this was not the case for 18:1(n − 9) (not enriched in the presence of H 13 CO − 3 +H 2 S either) or for i19:0 which was not detected in our specimens.

Fig. 2.
Bathymodiolus azoricus gill tissue PLFA (n = 3, average ± min-max) 13 C incorporation per total PLFA content (as calculated using Eq. 3) after a 15 day supply with 13 CH 4 (black bars), or a 20 day supply with H 13 CO − 3 in the presence of H 2 S (grey bars).
Phylogenetically the B. azoricus SOX endosymbiont stands far from any known cultured SOX bacteria. The closest strains such as the obligate chemolithoautotrophic thiodenitrifying gamma-proteobacteria Thiohalomonas nitratireducens (Sorokin et al., 2007) only reach around 88% 16S rRNA sequence homology (S. Duperron, personal communication, 2009). Furthermore, the PLFA labeled in our experiments represented only 69% of the fatty acid (FA) extracted from T. nitratireducens. It is thus difficult to relate the phenotype of B. azoricus SOX symbiont to any free-living bacteria. Conway and MacDowell Capuzzo (1991) suggested that all SOX bacteria desaturate their FA via the anaerobic pathway. The anaerobic desaturation pathway produces long chain MUFA by the elongation of medium chain length cis-3-unsaturated intermediates, with 18:1(n − 7) as the major end-product and 16:1(n − 7) as a secondary product. The O 2 -dependent MUFA synthesis pathway is known to produce a large variety of fatty acids (FA) with a double bond insertion occurring mainly in the 9 position (Conway and MacDowell Capuzzo, 1991). Note that the delta ( ) nomenclature which imposes carbon atoms numbering to start from the carboxylic acid end of the acyl chain, is used for describing biochemical reactions. Our results show that 18:1(n − 7) and 16:1(n − 7), which are the main products of the anaerobic desaturation pathway, were significantly more labeled (−19.8 and +182.2‰, respectively) than the main product of aerobic desaturation, 9 MUFA 18:1(n − 9) (−29.6‰, Fig. 2). However, the latter fatty acid was less abundant, and therefore incorporation of similar levels of label in the 3 fatty acids could have resulted in a higher δ 13 C signature in PLFA 18:1(n − 9). The fact that MUFA 16:1(n − 7), which was approximately 3 and 9 fold more abundant than 18:1(n−7) and 18:1(n − 9), respectively (Fig. 4), was the most labeled compound together with saturated PLFA (12:0, 14:0 and 15:0; see Fig. 2), supports the contention of Conway and Mac-Dowell Capuzzo that B. azoricus SOX symbiont desaturates its FA via the anaerobic pathway. Wright (1982) reports that Mytilus and Modiolus mussels take up free amino acids dissolved in seawater. Once inside the cell, free amino acids can be incorporated unchanged into proteins (Eccleston and Kelly, 1972) or used for the synthesis of other macromolecules. Excess amino acids (not being incorporated into proteins) cannot be stored in the cells and are generally used as metabolic fuel for the production of FA, ketone bodies or glucose (Berg et al., 2002). The ketogenic amino acids leucine, lysine, isoleucine, phenylalanine, Table 1. PLFA displaying significant enrichement after the tracer experiments with H 2 S+ 13 CO 2 (SOX) or with 13 CH 4 (MOB). X = 13 C labeled. % PLFA = proportion of the PLFA content in cultured strains of Methylosphaera hansonii and Thiohalomonas nitratireducens.
The incorporation of amino acid 13 C into SOX and MOB FA biomarkers (12:0, and 16:1(n − 8) and (n − 6), respectively) might indicate that the symbionts could also be assimilating the amino acids directly. Bacteria indeed possess effective metabolic mechanisms to survive long periods of low food supply and to react rapidly to available suitable nutrients (including low concentrations of dissolved organic matter; Sepers, 1977). The SOX Beggiatoa bacteria, for instance, can grow facultatively or mixotrophically on inorganic and soluble organic compounds (Zhang et al., 2005). However, in general, organic compounds do not stimulate the growth of specialist phototrophs, lithotrophs, or methylotrophs. Although some of the isolated bacteria strains using methane and other C l compounds also grow on sugars and acids (Patt et al., 1974), many isolates capable of growth on methane have proven incapable of growth on conventional organic media (Whittenbury et al., 1970). Several factors may explain the lack of capacity to assimilate organic compounds such as an inhibition effect by these compounds (although balanced mixtures of amino acids can cancel the inhibition by one or the other amino acid, Smith and Hoare, 1977), the impermeability of the cell (e.g. Eccleston and Kelly, 1972) or the loss of enzymes from the main assimilation pathways (Theisen et al., 2005;Smith and Hoare, 1977). In obligate methane-oxidising gamma-proteobacteria, the chemical conversion of carbohydrates, lipids and proteins into CO 2 and H 2 O for energy production is blocked at the level of the tricarboxylic acid (TCA) cycle, due to the concurrent absence of one of its key enzymes, α-ketoglutarate dehydrogenase, and of isocitrate lyase and malate synthase from the glyoxylate shunt (Trotsenko, 1983). In some cases, succinyl CoA synthetase, another enzyme from the TCA cycle, is also missing, like in Thiobacillus denitrificans or M. capsulatus (Smith and Hoare, 1977). However, our finding that some Fig. 4. Bathymodiolus azoricus gill tissue PLFA or tFA content (n = 3, average ± standard deviation) after a 15 day supply with 13 CH 4 (black bars), or a 20 day supply with H 13 CO − 3 in the presence of H 2 S (grey bars), or with 13 C-amino acids (white bars); striped bars represent control mussels. of the MOB (16:1(n − 8) and (n − 6)) and SOX (12:0) symbionts biomarkers were labeled after incubation with 13 Camino acids indicates that the symbionts had at least access to acetyl coA (or acetate) produced during the degradation of the amino acids by the host, and/or that they have the capacity to absorb and metabolise external amino acids. Further experiments are needed to ascertain these hypotheses.

Conclusions
PLFA and total FA 13 C-fingerprinting in Bathymodiolus azoricus gill tissues allowed us to trace the assimilation of 13 C-enriched HCO − 3 (in the presence of H 2 S) and CH 4 into the endosymbionts and host. Based on the selective labeling pattern of PLFA in each of the experimental treatments we could establish qualitative FA profiles of MOB and SOX endosymbiotic bacteria living in association with B. azoricus. Additionally, carbon from dissolved free amino acids was found to be incorporated into host specific FA and also into some of the symbiont biomarkers.
The tracer uptake experiments in the present study could only be performed under conditions of atmospheric pressure. It therefore needs to be verified through future experiments whether or not the activated metabolic pathways are the same under high, in situ pressure conditions and atmospheric pressure. While more experimentation is needed to better understand the physiology of the ensosymbionts, this experiment has shown its usefulness for positioning the endosymbionts among described strains, as well as for the direct identification of the symbiont biomarkers, for which until now only assumptions had been made.