Compound-Specific Isotopic Fractionation Patterns Suggest Different Carbon Metabolisms among Chloroflexus-Like Bacteria in Hot-Spring Microbial Mats

Abstract

Stable carbon isotope fractionations between dissolved inorganic carbon and lipid biomarkers suggest photoautotrophy by Chloroflexus-like organisms in sulfidic and nonsulfidic Yellowstone hot springs. Where co-occurring, cyanobacteria appear to cross-feed Chloroflexus-like organisms supporting photoheterotrophy as well, although the relatively small ¹³C fractionation associated with cyanobacterial sugar biosynthesis may sometimes obscure this process.

Chloroflexus aurantiacus and its phylogenetic relatives, which comprise a deeply branching kingdom-level lineage in the domain Bacteria (24), are major components of photosynthetic microbial mats in both sulfidic and nonsulfidic hot springs in Yellowstone National Park, Wyo. (3, 4, 5, 12, 23, 47). C. aurantiacus, the most studied representative of the green nonsulfur bacteria available in pure cultures, can grow heterotrophically by aerobic respiration, photoheterotrophically (using light to incorporate prereduced organic compounds), and photoautotrophically (using light to fix inorganic carbon) (26). Photoautotrophic metabolism by an obligately phototrophic relative of C. aurantiacus, the predominant phototroph in sulfidic hot-spring microbial mats, has been reported (12). However, based on culture (25, 26) and radiolabeling studies (2, 32), it has been hypothesized (44) that Chloroflexus-like organisms are mainly photoheterotrophic in mats where they live together with cyanobacteria. In such mats cyanobacteria are thought to be the main primary producers, which cross-feed reduced organic compounds to Chloroflexus-like organisms, thus supporting their photoheterotrophic metabolism. We exploited the distinctive lipid biomarkers of cyanobacteria and Chloroflexus-like organisms to test this hypothesis (39). If Chloroflexus-like organisms were purely photoheterotrophic in cyanobacterial mats, their biomarkers should have a δ¹³C signature, similar to biomarkers of cyanobacteria. Lipids derived from cyanobacteria in hot-spring mats were found to have δ¹³C values typical of the −20 to −25‰ fractionations relative to the inorganic carbon source expected of the Calvin cycle (27, 31). However, long-chain polyunsaturated alkenes (e.g., the C_31:3 alkene hentriacontatriene) and C_30-37 wax esters that are typical of Chloroflexus (17, 35, 40) and mats containing Chloroflexus relatives (9, 35, 37, 39) were approximately 10 to 15‰ enriched in ¹³C relative to cyanobacterial lipids (37, 39). Such values are expected based on autotrophic metabolism by Chloroflexus, which is known to use an inorganic carbon fixation pathway, the 3-hydroxypropionate pathway (15), that imparts an unusually heavy δ¹³C signature to both biomass and lipids (16, 40). Thus, the enrichment in ¹³C of Chloroflexus lipids relative to cyanobacterial lipids in microbial mats pointed towards autotrophic growth of Chloroflexus-like organisms in hot-spring microbial mats (39).

Here, we examine an alternative explanation—Chloroflexus-like biomarkers could have heavy δ¹³C signatures because they are cross-fed isotopically heavy fixed organic matter from cyanobacteria. Radiolabeling studies have shown that light-driven CO₂ fixation in hot-spring cyanobacterial mats leads mainly to production of polysaccharides (18, 22), presumably by cyanobacteria, and the polysaccharides are fermented during darkness to short-chain fatty acids known to be photoassimilated by Chloroflexus-like organisms (2, 23, 32). It has been shown that sugars may be significantly enriched in ¹³C relative to lipids in cyanobacteria and other Calvin cycle photoautotrophs (7, 43). Hence, we now have significantly expanded our previous study by investigating not only the δ¹³C signatures of lipid biomarkers and biomass but also sugars and dissolved inorganic carbon (DIC) species to enable determination of isotopic fractionations. Furthermore, we included more representatives of mats where cyanobacteria and Chloroflexus-like organisms live together in different environments, thereby enabling us to better observe differences in isotopic fractionation patterns.

MATERIALS AND METHODS

Samples were taken from five different hot-spring microbial mats containing Chloroflexus-like bacteria with and without cyanobacteria located in Yellowstone National Park, Wyo. (Table 1). Samples for lipid analysis were frozen in the field and were kept frozen until lyophilization and lipid extraction. Samples for microscopy were stored on ice and directly analyzed after returning from the field. The presence of cyanobacteria and Chloroflexus-like bacteria was determined by phase contrast and autofluorescence microscopy. Water samples for sulfide analysis were collected from above each mat, preserved in zinc acetate, and analyzed by the method of Cline (6). Inorganic carbon in water overflowing mats was trapped as BaCO₃ by increasing the pH of 200 ml of spring water to approximately pH 11 by adding a saturated NaOH solution (pH 13) and solid BaCl₂ (36). From the stable carbon isotopic composition of the BaCO₃, the isotopic composition of the CO₂ in the spring water was calculated using the temperature-dependent isotopic equilibrium equation of Mook et al. (21).

TABLE 1.

General description of mat samples

Mat samples	Geothermal area	Temp (°C)	pH	[S2−] (μM)	Chloroflexus-like	Cyanobacteria	Sampling dated
NMA sourcea	Mammoth Terraces	58.9	6.1	133	+	−	05-12-1995
NMA downstreama	Mammoth Terraces	56.3	6.4	19	+	+	05-12-1995
Tangerine Springb	Mammoth Terraces	60	6.4	40	+	+	26-08-1999
Mushroom Springc	Lower Geyser Basin	64	8.3	BDe	+	+	25-08-1997
Octopus Springc	Lower Geyser Basin	58-64	8.3	BD	+	+	27-08-1997

^aPreviously reported by van der Meer et al. (39).

^bOn the shoulder of Orange Mound Spring along the Mammoth Upper Terrace loop road.

^cSee Ward et al. (45) and Ramsing et al. (28) for location.

^dDay-month-year.

^eBD, below detection.

Lipids were extracted, derivatized, and analyzed by using gas chromatography, gas chromatography-mass spectrometry, and isotope-ratio-monitoring gas chromatography-mass spectrometry (33). Cell-associated sugars were analyzed by using approximately 20 to 160 mg of microbial mat residue after lipid extraction. This material was hydrolyzed, and sugar monomers were derivatized and analyzed by gas chromatography, gas chromatography-mass spectrometry, and isotope-ratio-monitoring gas chromatography-mass spectrometry (42). Stable carbon isotopic compositions of the bulk cell material and BaCO₃ were determined by automated on-line combustion (Carlo Erba CN analyser 1502 series) followed by conventional isotope ratio-mass spectrometry (Fisons optima [11]). The stable carbon isotope compositions are reported in the delta notation relative to the Vienna PeeDee Belemnite ¹³C standard.

RESULTS AND DISCUSSION

The mat samples studied are compared in Table 1. The New Mound Annex (NMA) mat, found in a sulfidic, high-carbonate hot spring at Mammoth Terraces, is comprised of Chloroflexus without cyanobacteria (39). The sulfide is thought to poison cyanobacteria but is utilized by Chloroflexus spp. in a photoautotrophic metabolism (12). All other mats are comprised of both cyanobacteria (Synechococcus) and Chloroflexus-like filamentous bacteria. The Tangerine Spring mat, like the downstream NMA mat (39), occurs in a slightly acidic, high-carbonate spring in the Mammoth Terraces group, where sulfide levels have been reduced due to its use by Chloroflexus spp. and other upstream sulfide-utilizing organisms. The Octopus Spring and Mushroom Spring mats occur in low-sulfide, alkaline silica-rich springs of the Lower Geyser Basin.

Lipid and sugar composition.

The lipid compositions of the five mat samples are compared in Fig. 1 and Table 2, which also includes information for cultures of C. aurantiacus and its phylogenetic relative, Roseiflexus castenholzii (13), for reference. Lipids characteristic of Chloroflexus-like organisms were abundant in all mats. For instance, wax esters ranging from C₃₁ to C₃₇ were among the predominant lipids in all mats, with small-scale variation in distribution pattern among mat samples. Differences in the carbon skeletons of the wax esters between the environmental samples and the cultures may reflect physiological differences between cultivated and natural populations. For instance, the different mat systems do not contain monounsaturated wax esters as does C. aurantiacus but rather contain iso-branched wax esters (35, 39). Also, the mats contain C₃₁ to C₃₇ wax esters, whereas R. castenholzii produces C₃₇ to C₄₀ wax esters (41). Long-chain (C_29-32) alkenes, predominantly hentriacontatriene (C_31:3) (38), were abundant in mats in Mammoth Terraces springs but were present at only trace levels in mats from the Lower Geyser Basin. This may reflect the predominance of Chloroflexus spp. in Mammoth Terraces (46) and more distantly related Roseiflexus-like organisms in springs of the Lower Geyser Basin (23, 30), since the latter organism lacks hentriacontatriene (41). C₁₇ n-alkane, a biomarker for cyanobacteria (34), was found in all mats containing cyanobacteria but was absent from the NMA source mat. More common lipids, such as C_15-18 fatty acids and C_17-18 alcohols, were also detected in all mats.

FIG. 1.

Partial gas chromatograms of the total lipid extracts from the investigated microbial mats. 1, C₁₇ n-alkane; 2, C₁₅ fatty acid; 3, C₁₆ fatty acid; 4, C₁₇ fatty acid; 5, C₁₇ alkanol; 6, C_18:1 fatty acid; 7, C₁₈ fatty acid; 8, C₁₈ alkanol; 9, C_31:3 alkene; 10, C₃₁ wax ester; 11, C₃₂ iso-wax ester; 12, C₃₂ wax ester; 13, C₃₃ iso-wax ester; 14, C₃₃ wax ester; 15, C₃₄ iso-wax ester; 16, C₃₄ wax ester; I.S., internal standard.

TABLE 2.

Relative abundance of different lipids within mat samples

Category or compound	Peak no.a	Relative abundanceb
		NMA sourcec	NMA downstreamd	Tangerine Spring	Octopus Spring	Mushroom Spring	C. aurantiacusd	R. castenholziie
Chloroflexus biomarkers
C31:3 alkene	9	++	+	+	Trace	Trace	++	−
n-C31 wax ester	10	++	+	+	++	++	−
iso-C31 wax ester		+	+	+	+	+
n-C32 wax ester	12	+++	++	++	+++	+++	+
iso-C32 wax ester	11	+	+	+	+	+
C32:1 wax ester							+
n-C33 wax ester	14	+++	++	++	+++	+++	+
iso-C33 wax ester	13	+	+	+	+	+
n-C34 wax ester	16	+++	++	++	+++	+++	++
iso-C34 wax ester	15	+	+	+	+	+
C34:1 wax ester							+
n-C35 wax ester		+	+	+	++	++	++
n-C36 wax ester		+	−	+	+	+	++
C36:1 wax ester							+
n-C37 wax ester		+	−	+	+	+	+	+
n-C38 wax ester								++
n-C39 wax ester								+
n-C40 wax ester								++
Cyanobacterial biomarker
C17n-alkane	1	−	++	++	+	+	−	−
Nondiagnostic lipids
C15 fatty acid	2	++	++	+	+	++	+	−
C16 fatty acid	3	++	+++	+++	+	++	+++	−
C17 fatty acid	4	+	+	+	+	++	+++	−
C17 alkanol	5	+++	++	++	++	+++	+	−
C18:1 fatty acid	6	−	+++	++	−	−	+++	−
C18 fatty acid	7	+	++	++	+	+	+++	−
C18 alkanol	8	+++	+	++	+	++	++	−

^aSee Fig. 1.

^b+, detected; ++, ∼50% of base peak; +++, base peak; trace; detected in very low abundance; −, not detected.

^cPreviously reported by van der Meer et al. (39).

^dKnudsen et al. (17), Shiea et al. (35), and van der Meer et al. (40).

^evan der Meer et al. (41).

The sugar fractions of all mats contained arabinose, xylose, rhamnose, and glucose, the latter being dominant (approximately 30 to 80% of the sugar fraction). The sugar distribution was similar to that reported for C. aurantiacus (40).

Stable carbon isotopic compositions.

The isotopic compositions of DIC, bulk biomass, lipids, and sugars for all mats are reported in Table 3. The values for DIC species determined for Tangerine Spring was within the range previously reported for other Mammoth Terraces hot springs (19). The values for Mushroom Spring were somewhat lower. Isotopic composition of bulk biomass ranged from −13 to −17‰, except for the NMA downstream and Tangarine Spring mats, which were isotopically lighter (−24 and −25‰, respectively). Isotopic compositions of specific lipids ranged from −8.9 to −36.3‰. Compounds known from or possibly contributed by cyanobacteria (e.g., C₁₇ n-alkane and C₁₆ and C₁₈ fatty acids, respectively) were isotopically lighter (−21.3 to −36.3‰) than biomarkers of Chloroflexus-like organisms (C_31:3 and wax esters; −8.9 to −27.0‰). The isotopic composition of glucose ranged from −5.1 to −21.8‰. Isotopic compositions in cyanobacterial mats were generally more depleted in ¹³C in springs from Mammoth Terraces (Tangerine and NMA downstream) than in springs from the Lower Geyser Basin (Octopus Spring and Mushroom Spring).

TABLE 3.

Stable carbon isotopic compositions of bicarbonate, CO₂ (calculated from bicarbonate), bulk biomass, Chloroflexus and cyanobacterial biomarkers, nondiagnostic lipids, and glucose in mat samples expressed in per mille relative to the PeeDee Belemnite standard

Category or compound	Peak no.a	δ13C values of samples
		NMA sourceb	NMA downstreamb	Tangerine Spring	Octopus Spring	Mushroom Spring	C. aurantiacusc
Bicarbonate		−2 to −4d	−2 to −4d	−3.3	∼6.8e	−6.8	0
CO2f		∼9	∼9	∼9	∼12	∼11
Bulk biomass		−14.9	−23.5	−24.8	−16.9	−13.2	−13.0
Chloroflexus biomarkers
C31:3 alkene	9	−8.9	−15.1	−20.0	−18.9g	NMh	−12.5
C32 wax ester	12	−17.7	−22.7	−26.0	−18.3	−16.5	NM
C33 wax ester	14	−17.9	−24.1	−27.0	−18.5	−16.4	−15.0
C34 wax ester	16	−17.4	−23.4	−26.7	−18.5	−16.5	−14.4
Cyanobacterial biomarker
C17 n-alkane	1		−36.3	−36.3	−34.1	−29.6
General lipids
C15 fatty acid	2	−19.4	−23.4	−27.1	−20.2	−17.5	NM
C16 fatty acid	3	−19.3	−34.5	−35.3	−21.3	−21.3	−13.8
C17 fatty acid	4	−18.9	−22.1	−27.6	−18.7	−16.8	−14.8
C17 alkanol	5	−18.0	−22.0	−27.9	−19.4	−15.5	NM
C18:1 fatty acid	6		−32.7	−35.5			−12.5
C18 fatty acid	7	−16.0	−33.9	−35.1	NM	−24.4	−13.6
C18 alkanol	8	−16.1	−24.3	−29.1	−22.2	−21.5	−12.5
Sugar
Glucose		−5.1	−21.8	−20.0	−11.0	−11.5	−5.7

^aSee Figure 1.

^bPreviously reported by van der Meer et al. (39), except for the glucose isotope values.

^cFrom van der Meer et al. (40). Isotope values for organic compounds are reported relative to bicarbonate.

^dMadigan et al. (19).

^eAssuming similar δ¹³C value as for Mushroom Spring.

^fCalculated based on temperature-dependent isotope equilibrium equation (21).

^gPartial coelution.

^hNM, not measured.

Comparison of ¹³C fractionation in mats and cultures.

Isotopic fractionations between bicarbonate, the principle DIC species, and bulk biomass, lipids, and glucose for all mat samples and C. aurantiacus are compared in Fig. 2. The ¹³C fractionations between biomass, lipids, and glucose relative to DIC are similar for the photoautotrophically grown C. aurantiacus culture and the Chloroflexus spp.-dominated NMA source pool microbial mat (Fig. 2). The smallest ¹³C fractionation relative to DIC (Δδ¹³C_DIC) was observed for glucose, followed by hentriacontatriene, the bulk cell material, the fatty acids, and wax esters. This confirms earlier observations of Chloroflexus photoautotrophy in these high-sulfide, Chloroflexus-dominated microbial mats (12, 39) and shows that carbon isotopic observations made in culture experiments can be extended to environmental settings.

FIG. 2.

Depletions in ¹³C of biomass, key biomarkers, and glucose relative to DIC. ▹, biomass; ◊, hentriacontatriene; ○, wax esters (average); ×, C₁₇ n-alkane; □, glucose.

Chloroflexus biomarkers showed much larger fractionation relative to DIC in the two high-carbonate mats (NMA downstream and Tangerine Spring) where they live together with cyanobacteria (Fig. 2). This indicates that Chloroflexus-like organisms are not purely autotrophic in these cyanobacterial mats. Some isotopically lighter carbon must be obtained from an alternative source. The large Δδ¹³C_DIC for the cyanobacterial biomarker C₁₇ n-alkane suggests cyanobacteria as a possible source for this light carbon. We considered whether cyanobacterial glucose synthesis and fermentation coupled with cross-feeding could explain the observed isotopic fractionation patterns. The Δδ¹³C_DIC of glucose in these two high-carbonate mats is smaller than that observed for the cyanobacterial biomarker, C₁₇ n-alkane, consistent with the possibility that cyanobacterial glucose biosynthesis imparts a heavier isotopic signature than lipid biosynthesis (7, 43). However, even in the unlikely event that all of the glucose detected was from cyanobacteria and Chloroflexus derived all of its carbon from cyanobacterial sugar fermentation, the δ¹³C value for all Chloroflexus biomarkers should be lower than the δ¹³C of glucose, due to isotopic effects of subsequent lipid biosynthesis pathways (1, 14, 20). The fact that the δ¹³C value of the C_31:3 alkane is higher than (NMA downstream) or similar to (Tangarine Spring) that of glucose and that the δ¹³C value of wax esters (NMA downstream) is similar to that of glucose therefore suggests that cyanobacterial cross-feeding alone is unlikely to explain the observed results. Apparently, photoautotrophy is also occurring in Chloroflexus spp. in these mats. This could reflect either the inputs of separate heterotrophic and autotrophic Chloroflexus populations or mixotrophic carbon metabolism in a single Chloroflexus population. Since both cyanobacteria and Chloroflexus-like organisms contribute organic compounds to the mats, the δ¹³C values for the bulk biomass and glucose must be intermediate to cyanobacterial and autotrophic Chloroflexus isotope signatures.

Relative to the Mammoth Terraces cyanobacterial mats, the two mats in alkaline siliceous springs show smaller Δδ¹³C_DIC values (Fig. 2). The fractionation pattern of the bulk biomass, glucose, Chloroflexus biomarkers, and more general lipids resembles that of the autotrophically grown C. aurantiacus culture and the NMA source pool mat, suggesting the possibility of photoautotrophic metabolism by Chloroflexus relatives. However, in these mats the δ¹³C values of glucose are sufficiently heavy to support the hypothesis that cyanobacterial sugar biosynthesis imparts a heavier isotopic signature to sugars than to lipids; cross-feeding of sugar fermentation products could then impart the heavier signatures of Chloroflexus biomarkers. A complicating factor in Octopus Spring and Mushroom Spring could be the effect of CO₂ limitation on stable carbon isotope fractionation by cyanobacteria in these much more alkaline and lower-DIC settings (8). The Mammoth Terraces mats occur in carbonate-depositing springs that are high in DIC and, at pH 6.4, are poised near the pK_a of H₂CO₃/HCO₃⁻ (i.e., CO₂ is readily available). In contrast, the mats in alkaline silicious springs of the Lower Geyser Basin have midday pHs as high as 9.4, approaching the pK_a of HCO₃⁻/CO₃²⁻ (29). CO₂ limitation in these alkaline silicious hot springs may decrease the degree of isotopic fractionation by cyanobacteria, resulting in higher δ¹³C values for all organic compounds produced by cyanobacteria, including the C₁₇ n-alkane and glucose. However, the very large differentials in δ¹³C values of C₁₇ n-alkane and glucose (18 to 22.4‰) exceed those observed so far in Calvin cycle organisms (1 to 16‰ [43]). This large isotopic difference between the cyanobacterial biomarker and glucose, especially in combination with very small or no isotopic fractionation between CO₂ and glucose, makes it unlikely that all of the glucose is derived from cyanobacteria even when the possible effect of CO₂ limitation on the stable carbon isotope ratios of cyanobacterial products is considered. The high abundance of Chloroflexus biomarkers (i.e., wax esters) relative to cyanobacterial biomarkers (i.e., C₁₇ n-alkane) (Fig. 1) and the heavier isotopic signatures of C₁₆ and C₁₈ fatty acids (Table 3), which could be contributed by either type of phototroph (10, 17, 35, 40), indeed suggest that a large fraction of the total biomass might be Chloroflexus derived. This might be due to our analysis of thicker mat samples (i.e., >1 cm for Lower Geyser Basin mats versus 1 to 2 mm for Mammoth Terraces mats) and the persistence of Chloroflexus carbohydrates in deeper layers of the mats in alkaline siliceous springs (39).

By comparing isotopic compositions of compound classes from replicate cyanobacterial mats from different hot-spring settings, we were able to observe that both autotrophic and heterotrophic carbon metabolisms are employed by Chloroflexus-like bacteria. Heavier isotopic signatures of Chloroflexus-like bacteria may be due in part to their unique autotrophic biochemistry and in part to the differences in isotopic fractionation in sugar and lipid biosynthetic pathways of cyanobacteria. Further work will be necessary, however, in order to observe the degree to which cyanobacterial sugar biosynthesis affects isotopic compositions and, thus, the relative importance of heterotrophy (via cross-feeding from cyanobacteria) and autotrophy in the carbon metabolism of Chloroflexus-like bacteria in these mats. This is especially true of mats in alkaline siliceous springs, where CO₂ limitation effects make resolution of the two types of carbon metabolism more difficult.