|Home | About | Journals | Submit | Contact Us | Français|
The molecular and isotopic compositions of lipid biomarkers of cultured Aquificales genera have been used to study the community and trophic structure of the hyperthermophilic pink streamers and vent biofilm from Octopus Spring. Thermocrinis ruber, Thermocrinis sp. strain HI 11/12, Hydrogenobacter thermophilus TK-6, Aquifex pyrophilus, and Aquifex aeolicus all contained glycerol-ether phospholipids as well as acyl glycerides. The n-C20:1 and cy-C21 fatty acids dominated all of the Aquificales, while the alkyl glycerol ethers were mainly C18:0. These Aquificales biomarkers were major constituents of the lipid extracts of two Octopus Spring samples, a biofilm associated with the siliceous vent walls, and the well-known pink streamer community (PSC). Both the biofilm and the PSC contained mono- and dialkyl glycerol ethers in which C18 and C20 alkyl groups were prevalent. Phospholipid fatty acids included both the Aquificales n-C20:1 and cy-C21, plus a series of iso-branched fatty acids (i-C15:0 to i-C21:0), indicating an additional bacterial component. Biomass and lipids from the PSC were depleted in 13C relative to source water CO2 by 10.9 and 17.2‰, respectively. The C20–21 fatty acids of the PSC were less depleted than the iso-branched fatty acids, 18.4 and 22.6‰, respectively. The biomass of T. ruber grown on CO2 was depleted in 13C by only 3.3‰ relative to C source. In contrast, biomass was depleted by 19.7‰ when formate was the C source. Independent of carbon source, T. ruber lipids were heavier than biomass (+1.3‰). The depletion in the C20–21 fatty acids from the PSC indicates that Thermocrinis biomass must be similarly depleted and too light to be explained by growth on CO2. Accordingly, Thermocrinis in the PSC is likely to have utilized formate, presumably generated in the spring source region.
Based on phylogenetic analysis of small-subunit rRNA sequences, hyperthermophilic organisms proliferate in the deepest branches of the Bacterial and Archaeal domains. The branch lengths of these hyperthermophilic lineages tend to be short, which further suggests that such organisms are the closest known extant descendants of the last common ancestor and retain many ancestral phenotypic properties (49). The recent discovery of filamentous microfossils preserved in a 3,235-million-year-old submarine volcanogenic deposit lends considerable weight to the theory that hydrothermal vent organisms have had a very long history on Earth (41). Hyperthermophilic microbes are also attracting astrobiological and biogeochemical interest because of their potential role in the formation of many kinds of mineral deposits and the generation of rock textures and mineral assemblages that may be diagnostic for extant or extinct life beyond Earth (5).
A well-known example of a hyperthermophilic chemolithotrophic ecosystem is the pink filamentous streamers found at Octopus Spring in Yellowstone National Park (YNP), United States that were described by Brock in 1965 (3, 4). Similar streamer communities were first reported by Setchell in 1903 (45) and have subsequently been identified in neutral to alkaline springs of geothermal areas in Iceland, Japan, and Kamchatka, Russia (21, 48, 56) and, more recently, as distinct black streamers at Calcite Springs, YNP (42).
Molecular analysis of the small-subunit 16S rRNA sequences of the filamentous pink streamer community (PSC) indicates dominance of the domain Bacteria, in particular two deeply diverging phylotypes affiliated with the Aquificales and Thermotogales (43). From the PCS, the first pink streamer isolate, Thermocrinis ruber, was recently brought into culture (18). T. ruber forms a separate lineage within the order Aquificales and shares many features with the two previously isolated genera, Aquifex and Hydrogenobacter (22, 28, 29).
The use of lipid biomarkers for revealing microbial community structure is well established. Branched-chain fatty acids are not uncommon in thermophilic organisms (31) and have been identified in pink streamer samples from Octopus Spring previously (1). Other more distinctive lipids are now recognized as valuable biomarkers in some thermophiles. Thermomicrobium roseum contains internally methyl-branched C18 fatty acid and long-chain 1,2-diols as major components (38). The core lipids of members of the order Thermotogales are composed of unusual dicarboxylic fatty acids and a recently discovered ether lipid, 15,16-dimethyl-30-glyceroloxytriacontanoic acid (20). Other novel mono- and dialkyl glycerol ether (GME and GDE, respectively) lipids have been described in Thermodesulfobacterium commune (30) and Aquifex pyrophilus (22).
In addition to carrying distinctive chemical structures, lipid biomarkers also encode the stable isotopic signature that provides information about the physiologies of the source organisms (10, 26). However, interpretation of these isotopic signatures requires specific knowledge about C isotopic discriminations associated with the biochemical pathways involved in carbon fixation and lipid synthesis. Only a limited amount of information is available on the bulk isotopic fractionation factors for cultured Aquificales (17), and to our knowledge, nothing has been reported on the isotopic composition of lipids.
In this study, we initially set out to examine the microbial composition of the Octopus Spring PSC and nearby vent biofilms through a comprehensive lipid analysis. The resultant data revealed a more complex situation than was apparent from genomic analysis alone and also indicated a need for appropriate supporting data from pure-culture studies. A comparison of the lipid profiles of several genera within the Aquificales as well as measurements of the carbon isotopic fractionation associated with autotrophic and heterotrophic growth of T. ruber formed a framework for improved understanding of the population structure of the Octopus Spring PSC and associated vent microbiota.
Biomass consisting of the PSC was collected using forceps from an 87°C, pH 8.3 site in the main outflow just below the source pool vent of Octopus Spring in May 1997. The filaments were placed in glass tubes, sealed with Teflon-lined caps, frozen on dry ice within 3 h, and maintained so in transit to the National Aeronautics and Space Administration (NASA) Ames Research Center. A vent wall geyserite sample, approximately 25 cm2, was removed in 1996 from the shallower main pool that contains the main effluent of Octopus Spring (≈92°C, pH 8.0). The geyserite sample was also kept frozen until it was prepared for analysis.
Working in a glove box, the topmost 1 to 2 mm of the frozen geyserite biofilm was carefully removed by scraping with a sterilized spatula. The remaining material was then crushed in a sapphire mortar and pestle that had been cleaned with methanol and transferred to sterilized glass vials. PSC and vent geyserite samples were lyophilized and then ground to a powder in a glass mortar previously cleaned with sequential solvent washes of dichloromethane, methanol, and acetone. All glassware and metal implements used in our procedures were baked at 450°C for a minimum of 4 h. Only Teflon stoppers and/or Teflon-lined screw caps were used in analyses.
Thermocrinis ruber OC 1/4 (DSM 12173), Aquifex pyrophilus Kol5a (DSM 6858), Hydrogenobacter thermophilus TK-6 (IAM 12695), Aquifex aeolicus VF5 (21), and Thermocrinis sp. strain HI 11/12 (18) were obtained from the culture collection of the Lehrstuhl für Mikrobiologie, Universität Regensburg, Regensburg, Germany. Cell masses of the Aquificales strains were grown at 85°C (70°C for H. thermophilus and 80°C for T. ruber with formate) with stirring (up to 400 rpm) in a 300-liter enamel-protected fermentor (Bioengineering, Wald, Switzerland) as described in Table Table1.1. For growth of T. ruber in experiment 1 (isotope study), the cell titer was monitored and the culture was gassed with increasing flow rates (2, 5, 7.5, and 10 liter min−1) to maintain the growth rate.
For the analyses, an alignment of about 11,000 homologous full primary sequences available in public databases (ARB project [32, 33]) was used. The Aquificales 16S rRNA gene sequences were fitted in the 16S rRNA tree by using the automated tools of the ARB software package (33). Distance matrix (Jukes and Cantor correction), maximum parsimony, and maximum likelihood (fastDNAml) methods were applied as implemented in the ARB software package (34).
Lipids were extracted from lyophilized ground sinter or Aquificales biomass using a single-phase modification of the Bligh and Dyer procedure, and water-soluble contaminants were removed as previously reported (24). Elemental sulfur was removed by passing the total lipid extract over activated copper powder. The total lipid extract (TLE) was dried under nitrogen and then maintained in a vacuum desiccator over Drierite until it reached a constant weight.
Fatty acid methyl esters (FAME) and glycerol ethers (GME and GDE) were prepared by two procedures. In procedure I, FAME were prepared by subjecting a portion of the TLE to mild alkaline methanolysis (36) with heating at 37°C for 1 h. FAME were separated from the remaining polar ether lipids (GME and GDE) by thin-layer chromatography (TLC) using a methylene chloride mobile phase as previously reported (24). The FAME (Rf = 0.80) were recovered by eluting the silica gel with methylene chloride, and the ether-linked components were recovered from the origin of the TLC plate by Bligh and Dyer extraction of the silica gel zone. The polar ether components were hydrolyzed in 1 ml of chloroform-methanol-concentrated HCl (1:10:1) by heating to 100°C for 2 h (36), and the glycerol ethers were separated by TLC using hexane-diethyl ether-acetic acid (70:30:1) into GME (Rf = 0.04), GDE (Rf = 0.40), and diphytanylglycerol ether (Rf = 0.49) using reference compounds 1-O-hexadecyl-glycerol and 1,2-di-O-hexadecyl-glycerol (Sigma, St. Louis, Mo.) and diphytanylglycerol ether isolated from a Halobacterium sp. (52).
Procedure II was used in an attempt to analyze small samples such as the Octopus Spring vent geyserite. In this approach, the TLE was directly hydrolyzed with acid as described above, followed by trimethylsilyl (TMS) derivatization of the resulting free glycerol ethers, and gas chromatography-mass spectrometry (GC-MS) of the treated TLE. Some TLE samples were also analyzed for free fatty acids and glycerides by preparation of TMS derivatives. Abundance calculations were based on comparison of peak areas to internal standards, methyl tricosanoate (C23) for FAME and cholestanol for glycerol ethers. Weight percent of FAME (Table (Table2)2) was calculated based on flame ionization detector (FID) response of individual fatty acids (C14 to C22) relative to C23; for GME and GDE, values are based on the areas of the total ion chromatographs and should be considered semiquantitative.
Some of the PSC and T. ruber (experiment 1) TLE were also preparatively separated into a polar lipid (phospholipids) and a neutral lipid (glycolipids and glycerides) fraction by precipitation in cold acetone (27). The components of the polar and neutral fractions were then separated by thin layer chromatography on Silica gel G plates (Merck) using acetone-benzene-water (91:30:8) (37) or, in some cases, chloroform-methanol-water (65:25:4) (27). Preliminary characterization of TLC zones was made based on migration of standard diacyl compounds (phosphatidylcholine, phosphatidylethanolamine, and di- and monogalactosyl diglycerides) and staining with specific detection reagents, phosphomolybdic acid, ninhydrin, and α-naphthol (27). TLC zones for lipid analysis were detected by UV fluorescence with rhodamine 6G and recovered by the Bligh and Dyer elution. FAME and glycerol ethers were prepared from each fraction as described above. The double bond positions of the monounsaturated FAME were determined by preparing the dimethyl disulfide adducts (57). TMS derivatives of the glycerides were prepared using N,O-bis(trimethylsilyl)trifluoroacetamide with 1% trimethylchlorosilane (1:1 in pyridine).
Alkyl moieties were released from the glycerol ether compounds by reaction with BBr3 as reported previously (50).
FAME were analyzed by using a Perkin-Elmer Sigma 3B gas chromatograph equipped with an FID and 30-m megabore columns (J & W Scientific), either a DB-5ms programmed to increase at 4°C/min from 160 to 280°C, or a DB-27 programmed to increase at 4°C/min from 120 to 220°C. Compound identification was based on retention times on the nonpolar and the polar columns and on mass spectral analysis (see below).
GC-MS analyses of fatty acids as FAME or TMS esters and glycerol ethers as TMS derivatives, were performed using an HP 6890 gas chromatograph equipped with a J&W DB-5 (60 m by 0.32 mm, 0.25-μm film) capillary column and an HP 5973 mass-selective detector operated at 60°C for 10 min, then programmed at 10°C/min to 320°C, and held for 60 min (Fig. (Fig.1).1). The bond positions of the monounsatured FAME were determined by analyzing their dimethyl disulfide adducts as previously described (25).
The dissolved inorganic carbon (DIC) was measured by taking three 40-ml water samples from the outflow site using a syringe and immediately filtering through Whatman GF/F filters into preevacuated 130-ml serum bottles sealed with silicone stoppers and containing a few drops of saturated HgCl2 to inhibit bacterial growth. The bottles were kept chilled until analysis. At Ames, samples were withdrawn and acidified, and the CO2 gas was collected on a vacuum line for isotopic analysis on a Nuclide 6-60RMS mass spectrometer modified for small samples (9, 14). Analysis of the DIC composition of the Thermocrinis culture medium was similar except that samples were collected by filling three glass tubes with the gassed culture medium prior to inoculation and immediately stoppering them with crimp seals before shipment by air to Ames from Germany. Biomass and total lipid were determined at Ames using a Carlo Erba CHN EA1108 elemental analyzer interfaced to a Finnigan Delta Plus XL isotope ratio mass spectrometer (EA-IRMS). Compound specific isotope analyses were done at the Australian Geological Survey Organisation (AGSO) as previously described using a Finnigan MAT 252 mass spectrometer equipped with a CuO/Pt microvolume combustion furnace and a Varian 1400 gas chromatograph with DB-5 column (25). Reported δ values for FAME and GME were the averages of two runs which agreed within ±1.2% and have been corrected for the presence of carbon added during derivatization (−43‰ in the case of TMS-C and −50‰ in the case of methanol carbon used to form FAME).
The total organic carbon (TOC) recovered from the vent geyserite surface was only 0.24%, compared with 7.2% for PSC and 36.7% for T. ruber biomass (experiment 1). Procedure II, direct acid hydrolysis, was used to minimize loss of material during preparation of the vent sample (Fig. (Fig.1)1) and to allow comparison with PSC and T. ruber samples. Although this method does degrade the cyclopropyl FAME present in these extracts (6), it provides valuable comparative information.
The fatty acids of vent communities and PSC were predominantly C17 to C22 chain lengths, distinguished by iso-homologues of the C16 to C21 acids plus a diminished (see above) amount of cy-C21. The TMS derivatives of the free fatty acids present in the vent TLE confirmed that high amounts of cy-C21 (25% of total) were characteristic of the Octopus Spring vent biofilm community. The fatty acid composition of vent and PSC extracts, while qualitatively similar, differed quantitatively. Streamers had more abundant branched-chain fatty acid (41% in PSC versus 13% in Octopus Spring vent) with a somewhat shorter chain distribution: the ratio of i-C17:0 to i-C19:0 was 1:2.1 in PSC but 1:3.7 in the Octopus Spring vent. The anteiso analogues ai-C17:0 and ai-C19:0 were present in PSC (1.3%) but almost absent in the Octopus Spring vent. The Octopus Spring vent TLE also contained small amounts of even-carbon-numbered chain fatty acids, n-22:0 to n-30:0, and a mid-chain branched octadecanoic acid (1%), possibly 10- or 12-methyl-C18:0. Alkyl glycerol ethers were abundant in both vent and PSC, however, the vent GDE represented a higher proportion (43%) of total ether lipids than in PSC (29%). Significant amounts of archaeal biomarkers, 1,2-di-O-phytanylglycerol (archaeol) and both C20- and C25-isoprenoid glycerol monoethers, were present in the vent biofilm and PSC (≈5% of total ether lipids). BBr3 cleavage of the ether alkyl chains showed that, apart from the isoprenoid moieties, only straight-chain compounds were present, with carbon numbers consistent with the GC-MS characterization of the intact glycerol ethers. C18 chains dominated both GME and GDE (Fig. (Fig.11).
The acid hydrolysate of the T. ruber TLE (procedure II) was composed primarily of saturated and monounsaturated C20 fatty acids and a C18-GME (Table (Table2).2). No iso-branched fatty acids or GDE were detected. Additional analyses of the lipid using an alkaline methanolysis procedure (I) confirmed the high proportion of C20 fatty acids with C20:1, which together comprised over 49% of the total for T. ruber OC1/4 (Table (Table2,2, experiment 1). Large amounts of cy-C21 were also recovered in T. ruber and PSC by this procedure, and all subsequent FAME analyses were so carried out (Tables (Tables22 and and3).3).
T. ruber biomass from a variety of growth conditions, as shown in Table Table1,1, was analyzed. In experiment 1, T. ruber was grown using thiosulfate with a constant gassing of H2-CO2-O2 to maintain high substrate levels for measurement of the carbon isotopic discrimination associated with CO2 fixation (results below). These conditions resulted in accumulation of relatively large amounts of intracellular sulfur (≈15% of dry weight). S0 accumulation was not apparent in T. ruber grown with thiosulfate but without H2 (experiment 2) or in PSC extracts. The lipid composition of additional Aquificales cultures (Table (Table1)1) was also analyzed to assess the potential use of Thermocrinis-like lipids as group biomarkers to characterize the pink streamer community (Table (Table2).2). All Aquificales cultures contained GME, and no iso-branched fatty acids were detected (Table (Table2).2). GDE were only present in the lipids of the PSC and the two marine Aquifex cultures, A. aeolicus VF5 and A. pyrophilus Kol 5a. BBr3 cleavage of the Aquificales glycerol ethers showed only straight-chain alkyls with chain length distributions similar to those characterized for the intact molecules.
Using the methods of Rohmer et al. (44), no hopanoids were detected in the PSC total lipid extract, either as the polar bacteriohopanepolyol or as the free lipids, diplopterol or diploptene.
A preliminary attempt was made to separate the complex lipids of the PSC extract to aid in biomarker identification and as a preparative step to carbon isotope analysis. The diversity of polar head groups and the occurrence of both diacyl and dialkyl moieties made separation by one-dimensional TLC difficult. TLC separation of the polar lipid (PL) fraction using the chloroform-methanol-water system recovered 92% of PL-FAME in two zones, PL-2 (Rf = 0.35 to 0.25) and PL-4 (Rf = 0.60 to 0.56) that roughly comigrated with diacylphosphatidylcholine (Rf = 0.33) and diacylphosphatidylethanolamine (Rf = 0.66) (Table (Table3).3). All PL zones showed the presence of multiple components, and PL-4 was dominated by an aminolipid similar to that observed from A. pyrophilus (27).
FAME present in the neutral lipid (NL) fraction were separated using the acetone-benzene-water system. Both a glycolipid zone that migrated closely with a digalactosyl diglyceride standard (Rf = 0.43) and a rapidly migrating component (Rf = 0.95), a probable free glyceride designated NL-2, accounted for 12 and 82% of recovered NL-FAME, respectively (Table (Table3).3). The GDE present in the PL fraction were recovered primarily in PL-4 and those in the NL fraction in NL-2. Together these two zones accounted for 60 and 25% of total GDE, respectively. While the recoveries of FAME and GDE were in good agreement with analyses made using the TLE, the recovery of GME was poor (see below).
Lipid fractions as described above were also prepared from the total lipid of T. ruber (experiment 1). In this case, most of the FAME were recovered from two fractions equivalent to PL-2 and PL-4 with 24 and 70%, respectively. No GDE were detected in any of the isolated fractions, and although most of the GME was also recovered in PL-2 and PL-4, the recovery was not significant (0.2 μmol/g [dry weight]) relative to the amount measured by the direct acid hydrolysis procedure (Table (Table2).2). This discrepancy appears to be associated with the use of TLC to separate the polar compounds and not procedure I, as relatively large amounts of GME were recovered during analysis of the additional Aquificales cultures using procedure I.
The Octopus Spring water had a relatively high DIC content, 5.3 mM, with a δ13C value of −1.5‰ ± 0.3‰ (n = 3). At the temperature (87°C) and pH (8.3) of the pink streamer site, a dissolved CO2 concentration ([CO2]) of 71 μM and δ13C of −4.7‰ can be calculated based on a pKa for carbonic acid at 90°C of 6.42 and the equation of Mook et al. (35) in which the fractionation between dissolved CO2 (d) and dissolved bicarbonate (b) is given by Εd/b = 24.12‰ − 9866/T, where T is the absolute temperature. The streamer biomass and TLE from this site are depleted in 13C relative to [CO2] by 10.9 and 17.2‰, respectively (Table (Table4).4).
In laboratory cultures (Table (Table1,1, experiment 1), a sample of medium removed prior to inoculation measured 14.1 mM DIC with a δ13C of −25.5‰ ± 0.9‰ (n = 3). At the temperature and pH of the fluid, the [CO2] was 6.2 mM, with a δ13C of −27.4‰. The carbon available from the continuously flowing gas mixture far exceeded that assimilated by the culture. The T. ruber biomass and lipids were depleted in 13C relative to the [CO2] by 3.3 and 2.1‰, respectively. As shown in Table Table4,4, this depletion is much smaller than that observed for the PSC. Moreover, the patterns of depletion differ, with lipids depleted relative to biomass in the PSC and enriched in T. ruber.
In a second experiment where T. ruber was grown with 0.1% formate (Table (Table1,1, experiment 4), biomass was much more strongly depleted in 13C relative to carbon source (19.7‰). The yield from 250 liters of medium was 5 g (dry weight) (39.2% carbon), which accounted for 0.36% of available carbon. As with growth on CO2, lipids were slightly enriched relative to biomass.
The carbon isotopic compositions of several individual lipid biomarkers were also determined (Table (Table5).5). In the PSC fractions, fatty acids clustered into two isotopically distinct groups. Among the peaks with sufficient material for isotopic analysis, the iso-branched fatty acids (i-C17:0, i-C18:0, and i-C19:0) in PL-2, PL-4, and NL-2 were more depleted in 13C than the longer-chain C20 and cy-C21 fatty acids. Depletions relative to CO2 averaged 22.6‰ ± 0.4‰ (n = 8) for iso-branched and 18.4‰ ± 1.4‰ (n = 8) for the C20 and C21 acids. Bulk fractions varied in parallel. PL-2 and NL-2 contained greater proportions of iso acids relative to long-chain acids and were depleted in 13C relative to PL-4, in which longer-chain acids were more abundant. Alkyl chains from glycerol ethers analyzed after BBr3 cleavage yielded δ values of −23.8, −21.8, and −23.8‰ for C18, C19, and C20, respectively.
A previous study of the pink filamentous streamers at Octopus Spring identified three phylotypes, EM3, EM17, and EM19, from amplification of the mixed-population DNA (43). A phylogenetic tree was constructed to take advantage of a current, more extensive sequence database (Fig. (Fig.2)2) and confirms that the EM17 gene sequence clusters among the Aquificales and is closely related (99% sequence identity) to the pink streamer isolate T. ruber (18), whereas the EM3 sequence is related to the Thermotogales. EM19, however, constitutes a separate, more deeply diverging lineage, well outside the Aquificales and Thermotogales. In Reysenbach's study (43), the EM17 sequence represented the majority of clones examined (26 of 35) and a fluorescently labeled oligonucleotide probe complementary to EM17 hybridized in situ to the pink filaments. No hybridization was noted for EM3 or EM19 probes even though all morphotypes in the PSC did bind to universal and bacterial probes (43).
The pink streamers of the Octopus Spring outflow channel were characterized by high levels of iso- and cyclopropane ester-linked fatty acids and straight-chain ether-linked alkyl lipids (Table (Table2).2). Similar lipids were associated with the biofilm growing on the siliceous sinter walls of the vent pool (Fig. (Fig.1).1). The presence of two fatty acid pools, the C15-C19 iso-branched fatty acids and the C20-cy-C21 fatty acids (Tables (Tables22 and and3),3), together with the absence of iso-branched fatty acids in Aquificales cultures (Table (Table2),2), indicated that the PSC contained more than one distinct bacterial population.
To date, T. ruber OC 1/4 is the only cultivated isolate from the PSC (18, 21). Its fatty acid composition is similar to that reported previously for a number of Hydrogenobacter thermophilus strains (28) and to those of the additional Aquificales cultures analyzed here (Table (Table2).2). The fatty acids of these Aquificales were dominated by n-C18:0, n-C20:1, and cy-C21 (Table (Table2).2). Two sets of monounsaturated isomers, C18:1Δ9 and C18:1Δ11, and their chain elongation products, C20:1Δ11 and C20:1Δ13, were detected. Since cyclopropane fatty acids are formed by the addition of a methylene group from S-adenosylmethionine across the double bond of a monounsaturated fatty acid, the two cy-C21 isomers are probably derivatives of the n-C20:1 isomers. C20 fatty acids are rare in bacteria, and the presence of large amounts of the n-C20:1 and cy-C21, with lesser amounts of n-C20:0, n-C21:0, n-C22:0, and n-C22:1 in representatives of four distinct subclusters within the Aquificales (Fig. (Fig.2,2, Table Table2),2), demonstrates a phylogenetic clustering for these membrane lipids and suggests that these fatty acids can serve as taxonomic marker signatures for this order. Notably, the pink streamers also contained similar C20, C21, and C22 fatty acids.
Nonisoprenoid alkyl glycerol ethers are being increasingly recognized as bacterial membrane lipids. In addition to the GME and GDE with the n-C16–18 alkyl chains previously described in A. pyrophilus (22), GME and GDE with iso- and anteiso-branched chains have also been identified as major membrane lipids in two anaerobic thermophiles, Thermodesulfobacterium commune (30) and Ammonifex degensii (19). An unusual glycerol monoether with a dimethyltriacontanyl chain has been identified in another thermophile, Thermotoga maritima (8). Additionally, small amounts of glycerol monoethers with normal or methyl-branched chains have been detected in mesophilic and thermophilic clostridia. These presumably are derived from 1-O-alkylglycerols in which an ester-linked fatty acid is initially present at position 2 (31). Environmental analyses have identified small amounts of n-C18 and br-C17 1-O-alkyl ethers and a C15,C15 1,2-O-alkyl diether in hot spring cyanobacterial mats (58, 59), and more recently, relatively abundant n- and br-C14–18 1-O-alkylglycerols have been found in association with anaerobic methane-oxidizing consortia in marine sediments (15).
All of the Aquificales cultures in our study synthesize at least some alkyl glycerol ether lipids (Table (Table2).2). The A. pyrophilus ether lipids most closely approximated the distribution observed for the PSC and vent alkyl ether lipids. To date, however, the only identified Aquifex spp. are marine bacteria. T. ruber does not appear to be the source of the PSC ether lipids. The relative abundance of GME and GDE and the alkyl chain distribution in the PSC suggest that an additional Aquificales-like organism was present in these thermophilic communities or that some environmental condition is responsible for the presence of GDE. Although we attempted to grow T. ruber under a variety of conditions in our study, we cannot preclude this latter point. Simulation of natural flow conditions in the laboratory results in growth of T. ruber as filaments rather than as the individual cells characteristic of our batch cultures (18). It is interesting to speculate that the physical environment of more natural growth conditions might allow expression of GDE synthesis.
Although limited by the small amount of biomass available, our analyses suggest that the community of organisms present in the biofilm of the Octopus Spring vent differs somewhat from the PSC. Specifically, the organisms producing the iso-fatty acids appear to be less abundant than the Aquificales. Iso-C17:0 and iso-C19:0 are abundant in the PSC, and, while present in the vent biofilm, the iso-C19:0 is now the only major branched acid and is present in much lower amounts relative to the C20–22 fatty acids and the glycerol ethers representing the Aquificales community members. The GDE also make up a much higher proportion of the glycerol ether lipids in the vent biofilm than in the PSC, which may reflect an environmental effect (i.e., higher temperature of vent water) or possibly a different Aquificales population.
Four CO2 fixation pathways have been described for bacteria (see reference 13 for a review). Carbon isotopic fractionation varies widely, depending on CO2 assimilation pathway. Generally, the enzymes of the reductive acetyl-coenzyme A (CoA) pathway express the largest 13C discriminations, with δ13C values for biomass relative to CO2 ranging from 20 to 36‰ (11, 39). Organisms using the Calvin-Benson cycle and ribulose-1,5-bisphosphate carboxylase (Rubisco) for CO2 incorporation display somewhat less discrimination, with Δδ13C from 11 to 26‰ (12, 39, 40, 47). A much broader range of values is found for the metabolically diverse organisms using the hydroxypropionate cycle (Δδ13C values from 2 to 13‰ relative to DIC [16, 53, 55]), and those using the reductive citric acid cycle (3 to 13‰ [39, 40, 47]). A particular hallmark of organisms of this last group, reductive tricarboxylic acid (TCA), is enrichment of 13C in lipids relative to biomass (54).
The enzymes of the reductive citric acid cycle are present in A. pyrophilus, A. aeolicus, and H. thermophilus, whereas Rubisco is absent (2, 7, 46). However, no information has been available about the C isotopic discrimination associated with growth of these obligately autotrophic bacteria. Moreover, the enzymology and carbon fixation pathways of T. ruber which can grow either autotrophically or organotrophically (18), have not been investigated. Our results indicate that T. ruber cells grown autotrophically at 85°C are depleted in 13C relative to CO2 by 3.3‰ and that the lipids are enriched in 13C relative to biomass by 1.2‰. Although both of these results are consistent with fixation of inorganic carbon by the reductive citric acid cycle, a novel metabolism cannot be precluded.
The pink streamer community overall is strongly depleted relative to CO2 by 10.9‰ (Table (Table4).4). Moreover, as shown in Table Table3,3, the individual fatty acids are even more strongly depleted (by 15.7 to 23.2‰). Neither the relatively large depletion in biomass nor the depletion in fatty acids relative to biomass is consistent with dominance of the PSC by T. ruber or any other organism utilizing the reductive citric acid cycle for the assimilation of CO2.
The isotopic composition of the T. ruber-like biomass in the PSC can be estimated from the δ13C value of the relevant biomarkers, namely the C20 and C21 fatty acids. As shown in Table Table4,4, whether grown on CO2 or on formate, the lipids in T. ruber are enriched in 13C by 1.2 to 1.8‰ relative to biomass. Accordingly, the ≈23‰ value of the C20-C21 fatty acids (Table (Table3)3) corresponds to a δ13C biomass of ≈−24.5‰ (Fig. (Fig.3).3). This would be consistent with a CO2 carbon source with δ13C ≈ −21‰ or a formate carbon source with δ13C ≈ −5‰ (Fig. (Fig.3).3). The former can be excluded. The isotope composition of the dissolved CO2 in the pink streamer community is known to be −4.7‰ (Table (Table4).4).
It is not known whether formate was present in the waters around the pink streamer community at the time these samples were collected. In subsequent investigations, Shock (E. Shock, personal communication) has found formate downstream but not at the location of the pink streamers and also measured H2 and CO in the gas phase of the vent source. It is in any case possible that the organisms within the pink streamer communities both produced (from CO2 + H2 or from CO + O2 + H2O) and quantitatively consumed HCO2H. Given the misfit between isotopic composition of DIC and T. ruber biomass, this provides the explanation most consistent with the available evidence. T. ruber is the only member of the Aquificales to grow either autotrophically or as a chemoorganotroph using formate or formamide (18). The nature of this metabolic capacity is unknown (18).
The δ13C value for the PSC organic carbon overall is −15.6‰ (Table (Table4).4). The difference between this value and the −24.5‰ estimated for T. ruber biomass requires that the remaining organic carbon be enriched in 13C (Fig. (Fig.3).3). If its abundance is equal to T. ruber, by mass balance, it must have δ13C ≈ −6‰ in order to produce an average δ13C ≈ −15.6‰.
The nature of this heavy portion of the bulk organic carbon is then problematic. The δ13C of two additional, independently collected PSC samples (September 1994 and 1999) agreed within ±0.5‰ (data not shown), indicating that the isotopic composition of this community is stable. It does not appear to arise from the iso-fatty acids. These have δ13C values near −27‰. Based on our current knowledge, the biomass to which they are related must have a δ13C value of −15‰ or lighter (the greatest observed depletion of lipids relative to biomass in microorganisms is ≈12‰ ). The total lipid fraction represents a relatively small portion (<10%) of total bulk organic carbon in the PSC and is much too light (−22.6‰), in itself, to account for all of the heavy carbon. Some other nonlipid component, synthesized within the community or brought in from the surrounding environment, must be present to balance the isotopic abundances.
Information is needed about the concentration and isotopic composition of formate in Octopus Spring water, the potential nature of an isotopically heavy component associated with the PSC filaments, the apparent lack of GDE in T. ruber, and the physiological mechanisms leading to the novel fractionations associated with its growth on CO2 and formate in order to fully assess the PSC on physiological, structural, and ecosystem levels.
Hydrogen-oxidizing members of the Aquificales are widely distributed in hot springs and thought to play a major role in the biogeochemical processes in these ecosystems (29). The carbon isotopic composition of the biomarker lipids in the PSC points to the expression of a novel metabolic potential by the Thermocrinis-like organisms in this hyperthermophilic community, but leaves unanswered the role and/or relationship of the population represented by the iso-fatty acids. Aquificales are considered the prevalent phylotype in filamentous bacterial communities found in geothermal springs (29, 42, 43, 51), but identification of Aquificales signature lipids associated with this vent geyserite suggests a broader ecological role for this group. Electron microscopic images of Octopus Spring geyserite indicate that bacterial communities readily colonize the vent walls and contribute to geyserite morphogenesis (5). Because of their phylogenetic position and their potential microfossil record, this group of organisms is particularly important to a better understanding of Earth's earliest microbial life.
Tsege Embaye (ARC), Kendra Turk (ARC), Sean Sylva (WHOI), and Thomas Hader (Regensburg) provided technical assistance. We thank Mitch Schulte and Everett Shock for useful discussions on the geothermal chemical mechanisms. We are grateful to Karl O. Stetter for stimulating and critical discussions. We are also grateful for support from the staff of the Research Division of YNP. L.J. thanks Jack Farmer for logistic and field support in YNP.
The work of Linda Jahnke and David Des Marais was supported by grants from NASA's Exobiology Program and the NASA Astrobiology Institute. The work of Wolfgang Eder was supported by the Fonds der Chemischen Industrie (to K.O.S.). Work by Sherry Cady was supported by the NASA Exobiology and the NSF Life in Extreme Environments Programs. Roger Summons and Janet Hope publish with the permission of the CEO, AGSO.