|Home | About | Journals | Submit | Contact Us | Français|
Roseiflexus sp. strains were cultivated from a microbial mat of an alkaline siliceous hot spring in Yellowstone National Park. These strains are closely related to predominant filamentous anoxygenic phototrophs found in the mat, as judged by the similarity of small-subunit rRNA, lipid distributions, and genomic and metagenomic sequences. Like a Japanese isolate, R. castenholzii, the Yellowstone isolates contain bacteriochlorophyll a, but not bacteriochlorophyll c or chlorosomes, and grow photoheterotrophically or chemoheterotrophically under dark aerobic conditions. The genome of one isolate, Roseiflexus sp. strain RS1, contains genes necessary to support these metabolisms. This genome also contains genes encoding the 3-hydroxypropionate pathway for CO2 fixation and a hydrogenase, which might enable photoautotrophic metabolism, even though neither isolate could be grown photoautotrophically with H2 or H2S as a possible electron donor. The isolates exhibit temperature, pH, and sulfide preferences typical of their habitat. Lipids produced by these isolates matched much better with mat lipids than do lipids produced by R. castenholzii or Chloroflexus isolates.
We have investigated laminated cyanobacterial mats in alkaline siliceous hot springs of Yellowstone National Park (Octopus Spring and Mushroom Spring) as models for understanding principles of microbial community ecology (49, 51, 54) and as models of stromatolites (50, 55), which are important fossilized microbial communities of the Precambrian. These mats are dominated by unicellular cyanobacteria (Synechococcus spp.) and filamentous anoxygenic phototrophic bacteria (FAPs) (e.g., Chloroflexus spp. and Roseiflexus spp.); they also contain newly described anoxygenic phototrophic bacteria (8) and many organisms involved in mat decomposition (49). We have studied the composition and structure of the mat community using both nucleic acid and lipid biomarker approaches (50), as well as the functional contributions of and interactions among community members (34, 46, 47).
A major finding from molecular analyses was that the predominant Synechococcus spp. of the mat are distantly related to readily cultivated strains (15, 49). Careful cultivation of predominant strains was necessary before we could observe phenotypic properties of Synechococcus sp. isolates that were useful for understanding the in situ ecology of native mat populations (2). Genomic analysis of relevant isolates has also vastly improved our ability to understand metagenomic (5, 25; C. G. Klatt, J. M. Wood, D. B. Rusch, M. M. Bateson, J. F. Heidelberg, A. R. Grossman, D. Bhaya, F. M. Cohan, M. Kühl, D. A. Bryant, and D. M. Ward, unpublished data) and metatranscriptomic (Z. Liu, C. G. Klatt, J. M. Wood, D. B. Rusch, M. Ludwig, N. Wittekindt, L. P. Tomsho, S. C. Schuster, D. M. Ward, and D. A. Bryant, unpublished data) databases, which have the potential to provide a holistic view of the community and to permit observation of how metabolic networking may occur among its members.
Similarly, based on small-subunit rRNA (SSU rRNA) cloning and sequencing and fluorescent in situ hybridization (FISH) probing, the dominant FAPs of the community were found to be distantly related to readily cultivated Chloroflexus spp. but closely related to Roseiflexus castenholzii (35), a bacteriochlorophyll a-containing anoxygenic phototroph lacking chlorosomes, which was initially cultivated from a Japanese hot spring (20). Only one SSU rRNA sequence somewhat distantly related to that of Heliothrix oregonensis, the only other FAP known to possess only bacteriochlorophyll a (37), has been detected in these mats (35, 56) (Fig. (Fig.1).1). However, SSU rRNA sequences closely related to that of the chlorosome-containing Chloroflexus aurantiacus were also detected in the mat at a much lower abundance than Roseiflexus-like SSU rRNAs, although the contribution of Chloroflexus-like organisms was greater at 70°C than at 60°C (35). The SSU rRNA sequences of R. castenholzii and cultivated Chloroflexus spp. have diverged substantially from Roseiflexus- and Chloroflexus-like SSU rRNA sequences detected in mats (35), and thus it is unclear how representative the phenotypic properties of these isolates are compared to mat FAPs.
One of the main questions is how the metabolism of cultivated FAPs observed in laboratory studies and/or inferred from genomic analyses relate to the metabolism of in situ FAP populations. Both R. castenholzii and C. aurantiacus strains have been shown to be capable of photoheterotrophic and dark aerobic metabolisms (20, 30, 36), while only some Chloroflexus sp. strains have been shown to be photoautotrophic (18, 28). Furthermore, Castenholz and Pierson (11) pointed out that Chloroflexus isolates grew well on inorganic media in dual culture with Synechococcus isolates. These observations support the notion that FAPs are cross-fed small organic compounds produced by cyanobacterial photorespiration (3) and fermentation (34) and thus might grow photoheterotrophically in the mat. Although a recent pyrosequencing study argued against this producer-consumer relationship (33), we have observed evidence that 13CO2 fixed by Synechococcus spp. into polyglucose is transferred to FAPs via assimilation of [13C]acetate derived from cyanobacterial fermentation (46). Compound-specific stable carbon isotope studies of FAP biomarkers also suggested that FAPs may be photoautotrophic (42, 45). FAPs that assimilate carbon which has been fixed and then excreted by cyanobacteria should have stable carbon isotope signatures typical of those resulting from the Calvin cycle. In contrast, the stable carbon isotopic compositions of these FAP biomarkers showed that they were much heavier than cyanobacterial lipid biomarkers, suggesting the possibility that in situ FAP populations are photoautotrophic or photomixotrophic by using the 3-hydroxypropionate pathway. This pathway, which is known to be used by Chloroflexus spp., results in isotopically heavier organic carbon than that fixed in the Calvin cycle (21, 22, 40, 43). This photoautotrophic biochemistry of FAPs in the mat was confirmed in labeling experiments using 13C-labeled bicarbonate, which suggested that the 13C label was incorporated into FAP biomarkers and FAP biomass (46, 47). However, Roseiflexus spp., not Chloroflexus spp., are the dominant mat FAPs, and R. castenholzii has not been shown to grow photoautotrophically (20).
Another concern is the differences in the lipids of FAP isolates compared to lipids that we and others have detected in Octopus Spring and Mushroom Spring mats (14, 39, 41, 42, 55, 58, 59). In addition to complex polar lipids (53), these mats contain abundant long-chain (C31 to C37) normal and branched wax esters (14, 55, 58, 59) (Table (Table1).1). C. aurantiacus strains produce C31 to C37 saturated normal or monounsaturated wax esters, as well as long-chain alkenes dominated by hentriacontatriene (C31:3) (26, 39, 43). R. castenholzii produces C37 to C40 normal wax esters and glycosides and fatty glycosides consisting of an alkane-1-ol-2-alkanoate (mainly branched C20 alkane-1,2-diol/C14 fatty acid and branched C20 alkane-1,2-diol/C16 fatty acid) bonded by glycosidic linkage to a C6 sugar. It does not produce long-chain polyunsaturated alkenes (44). However, the lipid biomarkers in the mat and those produced by these cultivated FAPs do not correspond well (Table (Table1).1). The low level of mat alkenes (45) is consistent with the low relative importance of Chloroflexus spp.; however, the low abundance of long-chain diols and diol glycosides is inconsistent with their abundance in R. castenholzii (44). Although the chain lengths of mat wax esters match those of Chloroflexus spp., the mat does not contain unsaturated wax esters typical of Chloroflexus spp., and mat wax ester chain lengths do not match well with those of wax esters produced by R. castenholzii. Furthermore the mat contains branched wax esters, which neither of these FAP isolates produces.
We hypothesized that our ability to understand (i) the functional roles that different FAPs play in the mat community and (ii) the microbial sources of major mat lipid biomarkers is only as good as the genetic, and thus metabolic and physiological, similarities between the FAPs that have been cultivated and native FAP populations. Here, we describe the cultivation of Roseiflexus sp. strains which are genetically relevant compared to native FAPs of Yellowstone hot spring microbial mats and which have heretofore only been described in a very preliminary way (29).
Hot spring microbial mats were sampled on 27 July 2002 at an average temperature of ~60°C from Octopus Spring, Yellowstone National Park, Wyoming (7). Mat material was stored in a 50-ml tube completely filled with spring water; transported to Southern Illinois University, Carbondale, IL, within 3 days; and stored at 4°C until further processing.
On 4 September 2002 a subsample was homogenized with a glass tissue homogenizer in 10 ml Castenholz medium D (10) supplemented with 1 g liter−1 sodium acetate, 1 g liter−1 NaHCO3, 0.2 g liter−1 NH4Cl, 100 mg liter−1 yeast extract, 100 mg liter−1 NaH2SO3, 3 g liter−1 HEPES buffer, and trace elements (48) and adjusted to a final pH of 8.17 after autoclaving. Bicarbonate and/or sulfide was added to the mineral medium in some cases (see below) from separately sterilized solutions before pH adjustment. A dilution series was prepared from 10−2 to 10−12, and each dilution was filtered through a 0.2-μm-pore-size polycarbonate filter. Each filter was incubated as a “floating” filter (12) placed above support filters saturated with approximately 1 ml of medium in small petri dishes, and the medium was periodically replenished. Petri dishes were sealed in a Gas-Pak jar (Becton-Dickenson) using an H2-plus-CO2 generator to remove O2 and incubated at 58°C with incandescent light of 40 μmol photons m−2 s−1. Tiny red colonies on the filters were observed within a week, but these did not grow following transfer to fresh filters. After 4 months, the original filters had completely dried out but “red-orange growth” was noticed in between the old colonies (10−4 dilution filters); several discrete colonies and this “red-orange growth” were transferred to a variety of media for cultivation of anoxygenic phototrophs and incubated under light anoxic or dark oxic conditions. Colonies that developed on 0.2% yeast extract agar plates (PE medium  with 0.2% yeast extract replacing sodium succinate, acetate, and glutamate) incubated aerobically in darkness at 50°C were transferred regularly to fresh medium incubated under both oxic dark and anoxic light conditions. Tiny single colonies were streaked for isolation several times, and the purity of the cultures was checked by streaking on tryptic soy and nutrient agar plates with oxic and anoxic incubation at 50°C. To grow biomass sufficient for DNA extraction and sequencing, purified cultures were transferred to 0.2% yeast extract liquid medium containing approximately 30 μM sulfide and incubated in completely filled tubes anoxically in the light. R. castenholzii strain HL08, which had been kindly provided by Satoshi Hanada (Research Institute of Biological Resources, Higashi, Japan), was also grown in 0.2% yeast extract liquid medium and incubated anoxically in the light.
Cells pregrown in liquid medium D containing 0.2% (wt/vol) yeast extract under anoxic light conditions were inoculated (10%, vol/vol) into fresh medium in which yeast extract was replaced by various single carbon sources or mixtures of carbon sources (0.1%, wt/vol or vol/vol), as described below. Because of yeast extract carryover in the inoculum, we compared growth on media containing single carbon sources (or the mix) with growth in 0.02% yeast extract medium (actually up to 0.04% yeast extract due to carryover). After incubation in the dark overnight to allow cells to acclimate to their new culture conditions, these cultures were incubated under anoxic conditions in the light or dark, as indicated, and growth was measured turbidometrically. In some cases, transfers (10%, vol/vol) were made into media containing single carbon sources (or a mixture of carbon sources) to ensure that growth was not due to carryover of yeast extract from the initial inoculum. Optimization of temperature (at pH 8 and without sulfide), pH (at 50°C and without sulfide), and sulfide (at 50°C and pH 8) was investigated in medium D containing 0.2% yeast extract with incubation under anoxic conditions in the light. Except in the case of sulfide optimization, experiments were done in triplicate.
Biomass scraped from growth on 0.2% yeast extract plates was either (i) resuspended in aqueous 30% bovine serum albumin or (ii) centrifuged and then extracted in methanol. Absorption spectra were recorded using a Hitachi U-2000 spectrophotometer scanning from 400 to 1,200 nm.
Phase-contrast micrographs of cells immobilized on water agar slides were taken with an Olympus B-Max 60 photomicroscope. Transmission electron micrographs were taken of cells fixed, stained, and examined as previously described (24).
For PCR analysis of FAPs, cell material from growing colonies was added directly to the PCR mixture and amplification of Chloroflexus sp. and Roseiflexus sp. SSU rRNA gene segments was performed according to methods and using primers reported by Nübel et al. (35).
A neighbor-joining phylogenetic tree was constructed based on 873 nucleotides (nt) between positions 418 and 1291 in the Escherichia coli SSU rRNA sequence using the software package ARB (available at http://www.mikro.biologie.tu-muenchen.de) with Jukes-Cantor correction, the pos_var_Bacteria_100 masking filter, and bootstrapping with 1,000 replicates. Sequences shorter than this 973-nt alignment were added using the ARB parsimony tool (27).
In genomic and metagenomic analyses, we used metagenomes for Octopus Spring and Mushroom Spring microbial mat samples obtained from sites averaging ~60°C and ~65°C (Klatt et al., unpublished data) and 20 representative genomes of organisms suspected to be major inhabitants of these mats, including the dominant FAP Roseiflexus sp. strain RS1 described here (accession no. CP000686; http://genome.jgi-psf.org/finished_microbes/ros_r/ros_r.home.html) and the draft genome of Chloroflexus sp. strain 396-1 (9), the Chloroflexus isolate most closely related to those in the mat and for which a genome sequence is available (Fig. (Fig.1)1) (52). We conducted a BLAST analysis (WU-BLASTn, using the settings M = 3, N = −2, wordmask = seg, and default values for all other parameters) to identify the metagenomic sequences with the best alignments to the 20 genomes. We separately created BLAST databases for the individual genomes of the three FAP isolates we have used in lipid biomarker analysis: Chloroflexus sp. strain Y-400-fl (accession no. CP001364; http://genome.jgi-psf.org/draft_microbes/chl_y/chl_y.home.html), R. castenholzii (accession no. CP000804; http://img.jgi.doe.gov/cgibin/pub/main.cgi?section=TaxonDetail&taxon_oid=639857015), and Roseiflexus sp. strain RS1. The metagenomic sequences that could be confidently assigned to mat FAPs from the 20-genome analysis (i.e., >80% nucleotide identity to the Roseiflexus sp. strain RS1 and Chloroflexus sp. strain 396-1 genomes) were then searched against each of the Roseiflexus and Chloroflexus genomes individually to identify the percent nucleotide identity of isolate genes to the homologous FAP genes in the mat.
The Roseiflexus sp. strain RS1 genome data were subjected to the integrated microbial genomes (IMG) annotation pipeline of the Joint Genome Institute (JGI) (31). In some cases the functional predictions assigned to genes by IMG were adjusted after more detailed inspection (see below).
Lipids were ultrasonically and sequentially extracted from freeze-dried harvested cells with methanol (three times), dichloromethane (DCM)-methanol (1:1) (three times), and DCM (three times) to obtain a total lipid extract (TLE). Part of this TLE fraction was separated into apolar (containing wax esters) and polar fractions by eluting over a small aluminum oxide column in hexane-DCM (1:1) and ethyl acetate, respectively. Base hydrolysis of the apolar fraction was performed to analyze the fatty acids and alcohols that comprise the wax esters (see reference 44 for details). The TLE and hydrolyzed apolar fractions were derivatized before analysis using gas chromatography (GC) and gas chromatography-mass spectrometry (GC-MS) (see reference 38 for details).
Discrete red colonies arising early in the attempt to cultivate Roseiflexus sp. strains contained predominantly bacteriochlorophyll c. However, red-orange colonies obtained from the transfer of the red-orange growth that arose later in between discrete red colonies on 0.2% yeast extract agar plates contained only bacteriochlorophyll a, indicating that these could be Roseiflexus-type organisms. This was confirmed by PCR amplification with Roseiflexus- but not Chloroflexus-specific SSU rRNA primers. Two strains of Roseiflexus sp., RS1 and RS2, were isolated, both of which exhibited a filamentous morphology (Fig. (Fig.2A).2A). No evidence of chlorosomes, which are characteristically found in green sulfur and green nonsulfur bacteria, was observed in thin sections by transmission electron microscopy (Fig. 2B to D). Bright inclusions were common in sectioned cells of strain RS1 (Fig. (Fig.2D),2D), and these are likely to be sites of polyhydroxyalkanoate storage. Although intact cells of both isolates exhibited absorption maxima at 795 nm, they differed in absorption maxima at higher wave lengths (i.e., 900 nm for strain RS1 and 860 nm for strain RS2); these wavelengths are indicative of antenna complexes containing bacteriochlorophyll a (Fig. (Fig.3).3). Absorbance of methanol extracts of cells at 770 nm confirmed that the cells contained bacteriochlorophyll a but did not contain detectable amounts of bacteriochlorophyll c (Fig. (Fig.33).
Both strains grew best under photoheterotrophic conditions with 0.2% yeast extract and much more slowly or not at all with 0.02% yeast extract medium (Table (Table2;2; see Fig. S1A and E in the supplemental material). For strain RS1, acetate, pyruvate, a mixture of organic acids (containing acetate, glutamate, glycolate, lactate, and malate), glucose, fructose, and possibly malate (but not glutamate, glycolate, butyrate, or lactate) supported growth in light at a lower rate than that on 0.2% yeast extract and at a higher rate than that observed on 0.02% yeast extract medium (see Fig. S1A to C in the supplemental material). Continuous growth was not observed upon further transfer in medium containing 0.02% yeast extract plus 0.1% acetate or malate (see Fig. S1D in the supplemental material), suggesting that in addition to acetate or malate, other nutrients or substrates supplied by yeast extract were required. For strain RS2, acetate, lactate, and the organic acid mixture (but not glycolate, malate, or glutamate) appeared to stimulate growth in light to a rate that was slightly higher than that observed on 0.02% yeast extract medium, but only after an extended lag phase (see Fig. S1E in the supplemental material). As with strain RS1, strain RS2 did not grow better in subculture in medium containing acetate, lactate, or malate than in medium containing 0.02% yeast extract (see Fig. S1F in the supplemental material).
No growth was observed under photoautotrophic conditions with 0.02% yeast extract medium supplemented with ~1.5 mM sulfide or in aqueous media bubbled with H2 gas and incubated at 50°C and pH 8 in the light.
Roseiflexus sp. strains RS1 and RS2 both grew at temperatures of between 45 and 60°C but not at 65°C, with an apparent optimum at 55°C to 60°C (Fig. (Fig.4,4, top panel). Strain RS1 grew at pH 7 to 9, and growth was optimal at pH 8.1; strain RS2 grew at pH 6 to 9, and growth was optimal at pH 8.1 to 8.5 (Fig. (Fig.4,4, middle panel). For strain RS1, sulfide was optimal at 120 to 240 μM and was inhibitory at higher levels (Fig. (Fig.4,4, bottom panel).
SSU rRNA sequences for strains RS1 and RS2 exhibited 99% nucleotide identity to each other. The SSU rRNA of strain RS1 formed a separate clade together with a large number of Roseiflexus-like clones previously detected at 51 to 63°C in the Octopus Spring and Mushroom Spring microbial mats (35), but to the exclusion of the SSU rRNA sequence of R. castenholzii (Fig. (Fig.1).1). Because we have obtained the complete genomic sequences of Roseiflexus sp. strain RS1, R. castenholzii, and C. aurantiacus Y-400-fl, as well as metagenomic data from these microbial mats (9; Klatt et al., unpublished data), we could compare the genomic relevance of these new Roseiflexus sp. isolates and FAPs that had been used in prior lipid biomarker studies to native FAP populations. As shown in Fig. Fig.5,5, mat metagenomic sequences exhibited the highest percent nucleotide identity to homologs in the Roseiflexus sp. strain RS1 genome, with lower percent nucleotide identity to R. castenholzii homologs and the lowest percent nucleotide identity to C. aurantiacus Y-400-fl homologs.
Roseiflexus sp. strain RS1 has a single circular chromosome of 5,801,598 bp, which is slightly larger than the 5,723,298-bp genome of R. castenholzii. Bioinformatics analyses provide evidence for a glycolytic pathway and a tricarboxylic acid cycle, which, together with type-1 NADH dehydrogenase, alternative complex III, and cytochrome oxidase, can account for the dark aerobic metabolism observed (see Table S1 in the supplemental material). Likewise, the presence of genes encoding type 2 photosynthetic reaction centers, light-harvesting complex I, and bacteriochlorophyll a biosynthesis genes is consistent with the phototrophic capability and pigmentation of this organism. The presence of genes encoding the 3-hydroxypropionate autotrophic pathway may allow this organism to acquire at least some of its carbon from carbon dioxide via this pathway (25). The Roseiflexus sp. strain RS1 genome contains an Ni-Fe hydrogenase encoded by genes hydAB as well as a complete suite of hyp genes potentially involved in the biosynthesis and maturation of the hydrogenase enzyme. Collectively, these may confer the ability to oxidize hydrogen as a source of electrons. The genome does not contain homologs of well-characterized genes involved in dissimilatory sulfur metabolism, such as dsr, fcc, or sox genes or sqr, which would confer the capability of oxidizing reduced sulfur compounds. A homolog to a type II sulfide:quinone oxidoreductase gene (sqr) is found in the genome, but this gene may be involved in sulfide detoxification instead of using sulfide as a source of electrons for photosynthesis (9). The presence of a putative molybdopterin-containing carbon monoxide dehydrogenase gene in the Roseiflexus sp. strain RS1 genome could confer the ability to respire CO aerobically as a potential source of energy in a mechanism similar to that of carboxydotrophic bacteria (23, 32). This capability has recently been observed in a nonphototrophic member of the kingdom Chloroflexi found in the same microbial mats, Thermomicrobium roseum, which is distantly related to phototrophic members of this kingdom (57).
Comparisons of the Roseiflexus sp. strain RS1 genome to that of R. castenholzii revealed large sets of genes that are common to both and some that are unique to each organism. Both Roseiflexus sp. genomes contain a cluster of four colocalized genes, with an order of nifHBDK, that are predicted to encode the structural genes of an Mo-nitrogenase. However, both genomes lack any evidence of many other genes typically involved in biosynthesis and maturation of a functional nitrogenase apoprotein; thus, it is unclear if Roseiflexus spp. have the ability to fix nitrogen. In addition, both Roseiflexus sp. genomes contain genes potentially allowing for the transport of both ferric and ferrous iron and regulation of iron homeostasis. Because both organisms have genes encoding phosphate ABC transporters and regulatory enzymes, R. castenholzii and Roseiflexus sp. strain RS1 can probably obtain phosphorous in the form of phosphate. Additionally, the Roseiflexus sp. strain RS1 genome encodes a phosphonate transporter and associated regulatory genes; this suggests that this organism may be able to utilize organophosphates in addition to phosphate as nutrient sources. Finally, the Roseiflexus sp. strain RS1 genome encodes a bacteriorhodopsin, which the R. castenholzii genome lacks. This chromophore-containing protein may be used to harvest light to produce proton motive force for the cell or may be used as a mechanism to regulate ions in response to changing light conditions (17). A more complete analysis of the genomes of Roseiflexus sp. strain RS1 and R. castenholzii will be published elsewhere.
Roseiflexus sp. strains RS1 and RS2 showed nearly identical lipid profiles (Fig. (Fig.6),6), which are dominated by C30 to C36 wax esters, including n/iso- and iso/iso-alcohol-fatty acid combinations, but not unsaturated forms (Table (Table1).1). We did not detect any long-chain alkenes. Both strains also contained C13 to C17 fatty acids, C15 to C18 alcohols, and C19 to C21 diols, but diol glycosides were not detected.
The Roseiflexus sp. strains that we obtained from Yellowstone hot spring microbial mats closely resemble R. castenholzii, which was isolated from microbial mats at Nakabusa hot springs in Japan. These isolates are similar in filamentous morphology, the presence of bacteriochlorophyll a, the absence of chlorosomes and bacteriochlorophyll c found in green sulfur bacteria and Chloroflexus spp., and their photoheterotrophic and dark aerobic metabolism (20) (Table (Table2).2). However, genomic analyses revealed that despite the inability of microbiologists to grow R. castenholzii and Roseiflexus sp. strain RS1 photoautotrophically, both of these isolates have all of the known genes for enzymes of the 3-hydroxypropionate pathway for CO2 fixation (25) and thus the potential to acquire at least some of their cellular carbon from this metabolism. The two genomic sequences revealed possible shared nutrient acquisition strategies of Roseiflexus spp. (e.g., iron and phosphate uptake and possible nitrogen fixation) but also possible differences, in particular that Roseiflexus sp. strain RS1 may also utilize organophosphates as a phosphorous source. The acquisition of phosphonates may be important in these alkaline siliceous hot springs, as there is evidence that Synechococcus sp. strain B′ contains phosphonate uptake and utilization genes as well (1, 5).
The Roseiflexus sp. isolates that we obtained are adapted to slightly higher temperatures (45 to 60°C; optimum, 55°C to 60°C) than R. castenholzii (45 to 55°C; optimum, 50°C), consistent with the temperatures of the mats from which they came. The possible existence of temperature-adapted FAP strains was suggested by Bauld and Brock (4) on the basis of ecophysiological results. Nübel et al. (35) suggested the possibility of temperature-adapted strains of Roseiflexus spp. based on differences in phylogeny and temperature distribution of SSU rRNA sequences cloned from hot spring mats (6, 16). Differences in light absorption between strains RS1 and RS2 may also reflect differential adaptation that might help explain the cooccurrence of many Roseiflexus-like SSU rRNA sequences at a single mat location. The Yellowstone and Japanese strains of Roseiflexus spp. are also optimally adapted to the pH of the mats from which they were cultivated. Yellowstone Roseiflexus sp. strain RS1 was found to be adapted to sulfide levels of 100 to 200 μM, which are typical of these mats (13). Our inability to demonstrate sulfide-associated photoautotrophic growth might have been due to the use of inhibitory concentrations of sulfide; however, genes typically associated with sulfide metabolism were not detected in genomic analyses. R. castenholzii was not tested for sulfide tolerance, except that the isolate was unable to grow photoautotrophically in the presence of 200 to 400 μM sulfide (20).
Small differences in carbon source nutrition were noted among R. castenholzii and the two new strains we studied (Table (Table2).2). Genetically, however, the Yellowstone isolates were quite divergent from R. castenholzii and were much more closely related to the native Roseiflexus species populations that predominate in the Yellowstone hot spring mats from which they were cultivated, in terms of both SSU rRNA sequence similarity (Fig. (Fig.1)1) and the match between genomic homologs (Fig. (Fig.5).5). Surprisingly, this situation is distinctly different for two C. aurantiacus strains, J-10-fl from Japan and Y-400-fl from Octopus Spring, which differ in gene content by only four genes and which are also nearly identical (99.98%) in overall DNA sequence (9).
The mats from which the new Roseiflexus sp. strains were obtained also contain Roseiflexus-like 3-hydroxypropionate genes (25), consistent with the detection of carbon dioxide fixation activity associated with FAPs on the basis of sulfide- and hydrogen-stimulated incorporation of 13CO2 into wax esters and biomass during an early morning low-light period (46, 47). However, the association between these lipid biomarkers and specific FAPs was unclear due to the poor match between the lipids of FAP isolates available at the time and mat lipids. As shown in Table Table1,1, the new genetically relevant Roseiflexus sp. isolates produce wax esters that are nearly identical to those found in the mats. In particular, these isolates produce branched-chain wax esters for which no source organism was previously known. Furthermore, they do not appear to produce unsaturated wax esters, which are known to be produced by Chloroflexus spp. (39, 43) but which are below detection limits in the mats. Likewise, the new Roseiflexus sp. isolates do not produce the long-chain, polyunsaturated alkenes typical for Chloroflexus spp. or the diol glycosides that are abundant in R. castenholzii. In agreement, both alkenes and diol glycosides are also only minor compounds in these mats (44, 45).
The results of this study show that our new isolates are representative of predominant FAPs in Yellowstone mats from which they were isolated, and genomic analysis of one isolate supports the inference from labeling experiments that Roseiflexus spp. are capable of fixing carbon dioxide during low-light periods of the diel cycle via the 3-hydroxypropionate pathway (46). In general, our findings demonstrate the importance of having genetically relevant isolates to understand lipid biomarkers and complex metabolic networks in microbial communities. In this regard, the Roseiflexus sp. strain RS1 genome has been very useful in analyses of mat metagenomes (Klatt et al., unpublished data) and metatranscriptomes (Liu et al., unpublished data).
We thank W. I. C. Rijpstra for analytical assistance. We thank the U.S. National Park Service for permission to conduct research in Yellowstone National Park and for their cooperation in facilitating the work.
This study was supported by U.S. National Aeronautics and Space Administration (NASA) grants NAG5-8824, 13468, and NX09AM87G and NASA's support of the Montana State University Thermal Biology Institute (NAG5-8807). Genomic sequences were obtained in collaboration with the Joint Genome Institute (Walnut Creek, CA) and with support from Department of Energy grant DE-FG02-94ER20137 and National Science Foundation grant MCB-0523100 (D.A.B.). The metagenomics database employed in this work was created under the auspices of grant EF-0328698 from the Frontiers in Integrative Biology Program for the National Science Foundation to D.M.W.
Published ahead of print on 2 April 2010.
†Supplemental material for this article may be found at http://jb.asm.org/.