Coassimilation of Organic Substrates via the Autotrophic 3-Hydroxypropionate Bi-Cycle in Chloroflexus aurantiacus^,

Abstract

Chloroflexus aurantiacus is a facultative autotrophic green nonsulfur bacterium that grows phototrophically in thermal springs and forms microbial mats with cyanobacteria. Cyanobacteria produce glycolate during the day (photorespiration) and excrete fermentation products at night. C. aurantiacus uses the 3-hydroxypropionate bi-cycle for autotrophic carbon fixation. This pathway was thought to be also suited for the coassimilation of various organic substrates such as glycolate, acetate, propionate, 3-hydroxypropionate, lactate, butyrate, or succinate. To test this possibility, we added these compounds at a 5 mM concentration to autotrophically pregrown cells. Although the provided amounts of H₂ and CO₂ allowed continuing photoautotrophic growth, cells immediately consumed most substrates at rates equaling the rate of autotrophic carbon fixation. Using [¹⁴C]acetate, half of the labeled organic carbon was incorporated into cell mass. Our data suggest that C. aurantiacus uses the 3-hydroxypropionate bi-cycle, together with the glyoxylate cycle, to channel organic substrates into the central carbon metabolism. Enzyme activities of the 3-hydroxypropionate bi-cycle were marginally affected when cells were grown heterotrophically with such organic substrates. The 3-hydroxypropionate bi-cycle in Chloroflexi is unique and was likely fostered in an environment in which traces of organic compounds can be coassimilated. Other bacteria living under oligotrophic conditions acquired genes of a rudimentary 3-hydroxypropionate bi-cycle, possibly for the same purpose. Examples are Chloroherpeton thalassium, Erythrobacter sp. strain NAP-1, Nitrococcus mobilis, and marine gammaproteobacteria of the OM60/NOR5 clade such as Congregibacter litoralis.

INTRODUCTION

The phototrophic Chloroflexus aurantiacus (32) and related Chloroflexi are facultative autotrophic green nonsulfur bacteria. They grow phototrophically in thermal springs, forming microbial mats with cyanobacteria (9, 35). During the day cyanobacteria produce and excrete glycolate as a product of photorespiration. At night they may excrete fermentation products due to oxygen limitation, as do anaerobes underneath the microbial mat. C. aurantiacus uses the 3-hydroxypropionate bi-cycle for autotrophic carbon fixation (44). Many intermediates of this CO₂ fixation cycle are closely related to the central carbon metabolism. Thus, the pathway appears to be ideally suited for coassimilating various organic compounds, which are channeled through these intermediates. The 3-hydroxypropionate bi-cycle starts from acetyl coenzyme A (acetyl-CoA), and conventional ATP and biotin-dependent acetyl-CoA and propionyl-CoA carboxylases act as carboxylating enzymes. In a first glyoxylate-forming cycle, two molecules of bicarbonate are fixed (Fig. 1). (S)-Malyl-CoA cleavage results in the formation of glyoxylate and acetyl-CoA. The latter can initiate another round of carbon fixation. Glyoxylate is assimilated in a second cycle (therefore the term bi-cycle), in which glyoxylate is combined with propionyl-CoA, an intermediate of the first cycle, to form β-methylmalyl-CoA. This condensation is followed by a series of C₅-transforming reactions yielding (S)-citramalyl-CoA, which is then cleaved into acetyl-CoA and pyruvate. The whole bi-cyclic pathway results in pyruvate formation from three molecules of bicarbonate and involves 19 steps but only 13 enzymes.

Fig. 1.

The autotrophic 3-hydroxypropionate bi-cycle in Chloroflexus aurantiacus and potential entry sites for various organic cosubstrates. 1, acetyl-CoA carboxylase; 2, malonyl-CoA reductase; 3, propionyl-CoA synthase; 4, propionyl-CoA carboxylase; 5, methylmalonyl-CoA epimerase; 6, methylmalonyl-CoA mutase; 7, succinyl-CoA:(S)-malate-CoA transferase; 8, succinate dehydrogenase; 9, fumarate hydratase; 10a/b/c, (S)-malyl-CoA/β-methylmalyl-CoA/(S)-citramalyl-CoA lyase; 11, mesaconyl-C1-CoA hydratase (β-methylmalyl-CoA dehydratase); 12, mesaconyl-CoA C1:C4 CoA transferase; 13, mesaconyl-C4-CoA hydratase.

Chloroflexi, like many autotrophic Bacteria and Archaea living in aquatic habitats, encounter carbon oligotrophic conditions and may grow as mixotrophs or facultative autotrophs. Coassimilation of traces of organic compounds together with CO₂ is advantageous under these conditions. The carbon fixation strategy of Chloroflexus is perfectly cut out for feeding a variety of organic compounds that are excreted by other microorganisms directly into the central carbon metabolism (Fig. 1). Furthermore, under anoxic conditions the usage of organic substrates that are more reduced than the average cell carbon requires CO₂ fixation as a sink for excess reducing equivalents that arise during the assimilation of those substrates. Even rudimentary elements of the 3-hydroxypropionate bi-cycle may function in the assimilation of mixtures of small organic molecules under oligotrophic conditions or in phototrophic microbial mats in which Chloroflexi live (4, 30, 41).

Potential organic cosubstrates are numerous, as depicted in Fig. 1. Glycolate formed by photorespiration in algae and cyanobacteria may be oxidized to glyoxylate and assimilated via the second part of the 3-hydroxyproionate bi-cycle. Similarly, fatty acids, alcohols, aromatics, and many other compounds may be assimilated after being metabolized to acetyl-CoA or propionyl-CoA. The same applies to fermentation products such as lactate, propionate, acetone, and acetoin. Polyhydroxyalkanoate utilization also results in acetyl-CoA (and sometimes propionyl-CoA) formation. Finally, beta-alanine and the abundant osmoprotectant dimethylsulfoniopropionate can be metabolized via 3-hydroxypropionate (2, 3, 28, 39, 43). Interestingly, various widespread bacteria appear to have acquired genes of the first part of the 3-hydroxypropionate bi-cycle, possibly for the same purpose of coassimilation. Genes operating in the second part of the bi-cycle are found predominantly in proteobacteria. The purposes of the elements of the rudimentary 3-hydroxypropionate bi-cycle and where they were obtained are open questions.

This work aimed to study the response of C. aurantiacus to different organic substrates provided in addition to CO₂. We show that the enzymes of the 3-hydroxypropionate bi-cycle are not significantly downregulated in this situation. Moreover, the bacterium readily assimilates many different organic acids by using the enzymes of the autotrophic 3-hydroxypropionate bi-cycle. Rudimentary elements of this bi-cycle may be used by various heterotrophic bacteria for the same purpose.

MATERIALS AND METHODS

Materials.

Chemicals were obtained from Fluka (Neu-Ulm, Germany), Sigma-Aldrich (Deisenhofen, Germany), Merck (Darmstadt, Germany), Serva (Heidelberg, Germany), or Roth (Karlsruhe, Germany). Biochemicals were from Roche Diagnostics (Mannheim, Germany), Applichem (Darmstadt, Germany), or Gerbu (Craiberg, Germany). [U-¹⁴C]acetate was obtained from Hartmann Analytic (Braunschweig, Germany).

Syntheses.

3-Hydroxypropionate was synthesized chemically from β-propiolactone. β-Propiolactone (1.25 ml, 25 mmol) was added dropwise with stirring to 6 ml of 5 M NaOH. The solution was lyophilized, and the dry powder was stored at room temperature. Acetyl-CoA and propionyl-CoA were synthesized from the anhydrides of the respective carbonic acids by the method of Stadtman (37). (S)-Malyl-CoA was synthesized enzymatically. The reaction mixture (1 ml) contained 200 mM 3-(N-morpholino)propanesulfonic acid (MOPS)-K⁺ buffer (pH 7.8), 5 mM MgCl₂, 30 mM glyoxylate, 10 mM acetyl-CoA, and 1.5 U of recombinant (S)-malyl-CoA/β-methylmalyl-CoA lyase/(S)-citramalyl-CoA (MMC) lyase. The reaction was carried out at 55°C for 15 min and stopped on ice by addition of 50 μl of formic acid, and precipitated protein was removed by centrifugation. Malyl-CoA was purified via preparative high-pressure liquid chromatography (HPLC) using a reversed-phase C₁₈ column (end capped, 5 μm, 125 by 4 mm; LiChrospher 100 [Merck]). CoA-thioesters were detected by UV absorbance at 260 nm with a Waters 996 photodiode array detector (Waters, Eschborn, Germany). A 15-ml gradient from 2 to 6% acetonitrile in 40 mM ammonium formate (pH 4.0) with a flow rate of 1 ml min⁻¹ was applied, the corresponding peak was collected on ice and lyophilized, and the dry thioester was stored at −20°C. (S)-Malyl-CoA was dissolved in water or buffer directly before use and kept on ice.

Organisms and cultivation.

C. aurantiacus strain OK-70-fl (DSMZ 636) was grown anaerobically and phototrophically under autotrophic conditions on minimal medium with H₂ and CO₂ (80:20, vol/vol) at 55°C (19). For photoheterotrophic growth (55°C) on minimal medium in the presence of organic substrates, the minimal medium was supplemented from anaerobic sterile stocks with either sodium acetate (2 g liter⁻¹) and sodium bicarbonate (3 g liter⁻¹), sodium 3- hydroxypropionate (2 g liter⁻¹) and sodium bicarbonate (3 g liter⁻¹), or sodium succinate (2 g liter⁻¹). Under these conditions the fermentors were not gassed during growth; thus, no hydrogen was present. Chloroflexus was also grown aerobically in the dark on minimal medium supplemented with sodium acetate (2 g liter⁻¹) as the sole carbon source at 55°C; the fermentor was aerated under stirring. Growth was determined by measuring the optical density at 600 nm (OD₆₀₀) (1-cm light path) of the cultures. Cells were harvested during exponential growth at an OD₆₀₀ of ∼2 by centrifugation (10 min, 6,000 × g, 25°C). For enzyme assays, cells were stored at −20°C.

Preparation of cell extracts.

For preparation of cell extracts with a mixer mill, cells (0.2 to 0.3 g [wet mass]) were suspended in a 3-fold volume of 20 mM MOPS-K⁺ buffer (pH 7.5) in microtubes (1.5 ml). After addition of 1.2 g of glass beads (0.1 to 0.25 mm), the cooled cell suspension was treated in a mixer mill (type MM2; Retsch, Haare, Germany) for 10 min at 30 Hz. The supernatant obtained after centrifugation (15 min, 16,000 × g, 4°C) was used for enzyme assays. For preparation of cell extracts with a French press, cells were suspended in a 2-fold volume of 20 mM MOPS-K⁺ buffer (pH 7.5) with 0.1 mg DNase I ml⁻¹. The cell suspensions were passed twice through a chilled French pressure cell at 137 MPa. The cell lysate was ultracentrifuged (1 h, 100,000 × g, 4°C), and the supernatant was used immediately or stored at −20°C in the presence of 20% (vol/vol) glycerol. Extracts were freshly prepared prior to enzyme activity measurements.

Experiments with suspensions of autotrophically grown cells.

To exclude contaminations, centrifuge tubes and bottles were autoclaved before use. Centrifuged fresh cells (2.6 g to 3.2 g [wet mass]) were suspended in 100 ml of the cell-free supernatant using 600-ml bottles with a rubber stopper. The 500-ml headspace was exchanged with H₂-CO₂ (80%:20%, vol/vol) gas through a sterile filter, and the pH was adjusted to 8.0 with sterile NaOH. The bottle was incubated with stirring in a transparent water bath (55°C) illuminated with light bulbs from two sides (together 120 W). Four different suspension experiments were performed using sterile stock solutions of organic acids (the final concentration of each compound was 5 mM), as follows: sodium acetate, sodium 3-hydroxypropionate, and disodium succinate; sodium lactate, sodium propionate, and sodium butyrate; sodium glycolate; and sodium acetate with 80 kBq [U-¹⁴C]acetate in 100 ml medium. Samples of 1 ml were taken at the beginning and at various incubation times, passed through a sterile filter to separate the cells, and frozen at −20°C. The samples were thawed directly before HPLC analysis. An ion exclusion column (Aminex HPX-87H, 300 by 7.8 mm; Bio-Rad Laboratories GmbH, Munich, Germany) was used under isocratic conditions with 12.5 mM H₂SO₄ at a flow rate of 0.6 ml min⁻¹. Standard compounds were applied before the samples were analyzed. Organic acids were detected at 205 nm using a Waters 996 photodiode array detector (Waters, Eschborn, Germany). In case of the [¹⁴C]acetate experiment, an additional aliquot (1 ml) of the cell suspension was withdrawn at each time point and centrifuged (10 min, 6,000 × g, 4°C). The sedimented cells were washed in 1 ml of water and centrifuged down. To determine the amount of incorporated ¹⁴C, the cells were resuspended in 1 ml water and 200 μl of the suspension was analyzed by liquid scintillation counting. Similarly, the amount of ¹⁴C in the medium supernatant was determined after acidification by adding 50 μl of acetic acid to 200 μl of the supernatant. Volatile ¹⁴CO₂ was then removed by agitation for 60 min at room temperature.

Enzyme Assays.

All assays were performed at 55°C. (S)-Malyl-CoA lyase activity was monitored spectrophotometrically at 324 nm (ε₃₂₄ of glyoxylate-phenylhydrazone, 17,000 M⁻¹ cm⁻¹). The assay mixture (0.5 ml) contained 200 mM MOPS-K⁺ buffer (pH 7.5), 5 mM MgCl₂, 2.5 mM phenylhydrazine, 0.2 mM (S)-malyl-CoA, and recombinant enzyme or cell extract. The reaction could be started with either (S)-malyl-CoA or enzyme.

Malonyl-CoA reductase was measured as described previously (19). Note that two NADPHs are oxidized per one malonyl-CoA added. The assay mixture contained 200 mM MOPS-K⁺ buffer (pH 7.5), 5 mM MgCl₂, 0.4 mM NADPH, 4 mM dithioerythritol (DTE), 1 mM malonyl-CoA, and cell extract. The reaction was started by addition of malonyl-CoA.

Propionyl-CoA synthase catalyzes three reactions: 3-hydroxypropionyl-CoA formation, dehydration to acryloyl-CoA, and NADPH-dependent reduction to propionyl-CoA. Propionyl-CoA synthase was measured as described previously (1). The assay mixture (0.5 ml) contained 100 mM MOPS-K⁺ buffer (pH 7.8), 0.4 mM NADPH, 2 mM 3-hydroxypropionate, 20 mM KCl, 3 mM ATP, 0.5 mM CoA, 4 mM DTE, and cell extract. The reaction was started by addition of 3-hydroxypropionate.

Isocitrate lyase activity was monitored spectrophotometrically at 324 nm (ε₃₂₄ of glyoxylate-phenylhydrazone, 17,000 M⁻¹ cm⁻¹). The assay mixture (0.5 ml) contained 100 mM MOPS-K⁺ buffer (pH 7.5), 5 mM MgCl₂, 5 mM phenylhydrazine, 1 mM dl-isocitrate, and cell extract. The reaction was started with either isocitrate or cell extract.

Malate synthase activity was measured using a discontinuous spectrophotometric assay. Formation of free CoA was detected at 412 nm using dithionitrobenzoic acid (DTNB) (ε₄₁₂ = 14,000 M⁻¹ cm⁻¹) (10). The reaction mixture (200 μl) contained 100 mM MOPS-K⁺ buffer (pH 7.5), 5 mM MgCl₂, 0.6 mM acetyl-CoA, 5 mM glyoxylate, and cell extract. The reaction was started by addition of glyoxylate. Samples of 40 μl were taken after 0, 3, 6, and 9 min of incubation and immediately mixed with 360 μl of 100 mM MOPS-K⁺ buffer (pH 7.5) containing 15 mM EDTA and 2.5 mM DTNB. Malate synthase as well as (S)-malyl-CoA lyase activity is dependent on divalent metal ions, and therefore the addition of excess EDTA stops the reaction. As a control, acetyl-CoA was incubated with cell extract for 9 min in the absence of glyoxylate.

Acetyl-CoA carboxylase and propionyl-CoA carboxylase were determined by a radiochemical assay. The assay mixture (1 ml) contained 100 mM MOPS-K⁺ (pH 7.8), 5 mM MgCl₂, 5 mM DTE, 10 mM NaHCO₃, 200 kBq Na_2¹⁴CO₃, 2 mM ATP, 0.4 mM acetyl-CoA or propionyl-CoA, and cell extract. Samples of 100 μl were withdrawn after 0, 2, and 5 min, their reaction was stopped in a 2-fold volume of 6% trichloroacetic acid, and then the mixture was incubated at room temperature while shaking for 12 h to remove all nonincorporated radioactive CO₂. In control experiments, acetyl-CoA or propionyl-CoA was omitted, respectively. The remaining acid-stabile radioactivity in the samples was measured by liquid scintillation counting. Acetyl-CoA carboxylase was also measured using a spectrophotometric assay (23). The assay mixture (500 μl) contained 200 mM MOPS-K⁺ (pH 7.8), 5 mM MgCl₂, 4 mM DTE, 10 mM NaHCO₃, 4 mM ATP, 0.4 mM NADPH, 0.1 U of recombinant malonyl-CoA reductase from C. aurantiacus, and cell extract. The reaction was started by addition of 0.4 mM acetyl-CoA. Two NADPHs were oxidized per one malonyl-CoA added, which was monitored at 365 nm. Recombinant malonyl-CoA reductase was produced and purified as described previously (23).

Biotin staining.

The soluble proteins of cell extracts (30 μg protein in each lane) were subjected to 12.5% SDS-PAGE (26) and blotted onto a nitrocellulose membrane. Streptavidin-coupled peroxidase was used to label the biotin carrier subunits of the biotin-dependent carboxylases. Therefore, the membrane was incubated for 90 min with 25 ml of 20 mM Tris-HCl (pH 7.5) containing 500 mM NaCl (TBS) to which 0.2 mg avidin-peroxidase was added. After being washed twice with TBS, the membrane was incubated with a mixture of staining solution A (20 ml cold methanol containing 60 mg of 4-chloronaphthol) and staining solution B (100 ml TBS containing 60 μl of 30% H₂O₂). Both solutions were freshly prepared and mixed right before use.

Liquid scintillation counting.

Samples were mixed with 3 ml of liquid scintillation cocktail (Rotiszint Eco Plus; Roth, Karlsruhe, Germany) and analyzed using a scintillation counter (Tri Carb 2100TR; Packard, Meriden, CT). The counting efficiency (75 to 85%) was determined by the channel ratio, and measured values were corrected accordingly.

Other methods.

Protein was determined using the Bradford method (7). DNA and amino acid sequences were analyzed with the BLAST network service at the National Center for Biotechnology Information (Bethesda, MD). BLAST searches were performed using the genome sequence of Chloroflexus aurantiacus strain J-10-fl. Phylogenetic trees were constructed using the neighbor-joining method (36) and the program MEGA 4.0.2 (24).

RESULTS

Regulation of enzymes of the central carbon metabolism in response to organic carbon sources.

The activities of characteristic enzymes of the 3-hydroxypropionate bi-cycle and the two key enzymes of the glyoxylate bypass, isocitrate lyase and malate synthase, were measured in extracts of cells that grew exponentially in the presence of different carbon sources. These analyses aimed at determining whether the bi-cycle or parts of it may be used for the coassimilation of small organic compounds and CO₂. In this case one would not expect a strict downregulation of the enzyme activities of the 3-hydroxypropionate bi-cycle in the presence of organic carbon sources. Therefore, C. aurantiacus was grown photoautotrophically under anaerobic conditions with H₂ plus CO₂, photoheterotrophically with bicarbonate plus acetate, with bicarbonate plus 3-hydroxypropionate, with succinate (only endogenously produced CO₂ was available), and aerobically in the dark with acetate alone. No hydrogen was present during any heterotrophic cultivation.

Activities of the assayed enzymes could be detected in all the cell extracts (Table 1). Interestingly, both activities of the key enzymes of the glyoxylate bypass were detected as well. During anaerobic growth there was only a modest response on the level of the overall enzyme activities to the different organic carbon sources. However, anaerobic utilization of organic compounds resulted in a significantly lower activity of malyl-CoA lyase and thus slowed down the regeneration of acetyl-CoA. The strongest changes were observed for aerobic growth with acetate in the dark. These cells showed a significant decrease in malonyl-CoA reductase activity (first partial cycle). The decrease in malyl-CoA lyase activity was even more drastic (second partial cycle), whereas malate synthase was severalfold more active (glyoxylate cycle) under these conditions.

Table 1.

Activities of characteristic enzymes of the 3-hydroxypropionate bi-cycle and of the glyoxylate bypass in extracts of C. aurantiacus cells that were grown phototrophically under different conditions

Enzyme	Sp act (nmol min−1 mg−1 protein)a in extracts of C. aurantiacus grown in the presence of:
	H2-CO2 without O2	HCO3−-acetate without O2	HCO3−-3-hydroxypropionate without O2	Succinate without O2	Acetate with O2
Acetyl-CoA carboxylase	7	6	3	7	4
Propionyl-CoA carboxylase	20	20	15	10	10
Malonyl-CoA reductase	60	60	50	30	10
Propionyl-CoA synthase	60	50	70	70	30
(S)-Malyl-CoA lyase	280	70	140	160	20
Isocitrate lyase	30	80	50	60	70
Malate synthase	400	310	370	400	1,800

^aMean values (deviations were <20 %) were obtained from at least two determinations at 55°C. Key genes of the glyoxylate pathway are indicated by a gray background.

Coassimilation of small organic compounds by autotrophically grown cells.

Cell suspension experiments were performed to investigate whether autotrophically grown C. aurantiacus cells are able to immediately take up organic substrates or whether there is a delay due to the preceding expression of substrate importers and enzymes of assimilatory pathways. Exponentially grown cultures harvested at an OD₆₀₀ of 2 (1.3 g [dry mass] liter⁻¹) were used for these studies. The generation time was 25 to 30 h, which corresponds to a specific growth rate of 0.028 h⁻¹ to 0.023 h⁻¹ and results in a specific autotrophic carbon fixation rate of 38 to 32 nmol min⁻¹ mg⁻¹ protein.

In the experiments, harvested cells were resuspended in a small volume of the same medium (380 to 480 mg [dry cell mass] in 100 ml supernatant from the harvest). Mixtures of different carbonic acids were provided at a 5 mM final concentration, and their consumption over time was analyzed by HPLC. Based on calculations, the provided H₂ and CO₂ in the headspace of the cell suspension bottles would have allowed for further autotrophic growth for more than 6 h. The results are illustrated in Fig. 2 a to c.

Fig. 2.

Assimilation of organic substrates by thick suspensions of autotrophically grown cells. The cells were pregrown autotrophically and harvested by centrifugation. Cells were suspended in a small aliquot of the culture supernatant. The concentrated cell suspension was then incubated at 55°C (pH 8) in the light with stirring in the presence of 80% H₂ and 20% CO₂. Simultaneously the following organic compounds were added at a 5 mM final concentration each: a mixture of acetate, succinate, and 3-hydroxypropionate (a); a mixture of lactate, propionate, and butyrate (b); glycolate (c); or [U-¹⁴C]acetate (d). In panel d, the acetate concentration and the amount of label in the supernatant were determined, as well as the amount of label incorporated into cell mass. Volatile ¹⁴CO₂ was removed by acidifying and shaking the samples. Decreasing concentrations in panels a to c were taken as indication of assimilation of the respective compounds.

All substrates tested except succinate were readily consumed by these autotrophically pregrown cells. Succinate, however, is a dicarboxylic acid, and therefore its import may be regulated differently than the import of monocarboxylic acids. The consumption of acetate, glycolate, 3-hydroxypropionate, propionate, lactate, and butyrate started without a lag phase. The consumption of C₂ and C₃ compounds was generally faster than that of the C₄ substrate butyrate, which is consistent with its higher carbon content. Only glycolate (6 nmol min⁻¹ mg⁻¹ cell protein) was assimilated more slowly than acetate (10 to 12 nmol min⁻¹ mg⁻¹ cell protein). The consumption rates of the substrate mixtures were 48 and 66 nmol organic carbon min⁻¹ mg⁻¹ cell protein, respectively (Table 2). These values are even higher than the previous CO₂ fixation rate of the autotrophically grown cells of 32 to 38 nmol min⁻¹ mg⁻¹ protein.

Table 2.

Consumption of organic carbon from the medium by thick suspensions of autotrophically grown cells of C. aurantiacus^a

Substrate	Rate (nmol min−1 mg−1 protein)
	Avg consumption of compound	Carbon consumption per compound	Total organic carbon consumption
Acetate	10	20	48
3-Hydroxypropionate	8	24
Succinate	1	4
Propionate	11	33	66
Lactate	7	21
Butyrate	3	12
Glycolate	6	12	12
[14C]acetate	12	24	24

^aThe dry cell mass of the cultures was determined to evaluate the contribution to carbon fixation. Rates were determined assuming that one-half of the cell dry mass is protein. These rates need to be compared with the CO₂ fixation rate of autotrophically growing cells of 32 to 38 nmol min⁻¹ mg⁻¹ protein.

Test of whether the coassimilated substrates were partly oxidized.

To test whether these compounds were partly oxidized anaerobically in the light in the presence of H₂ and CO₂, we exemplarily used uniformly ¹⁴C-labeled acetate (Fig. 2d). Disappearance of acetate and ¹⁴C from the supernatant proceeded almost in parallel; only a small fraction (20%) of nonvolatile, ¹⁴CO₂ free label remained after 3 h of incubation, when all of the acetate was used up. Half of the ¹⁴C that was originally added was incorporated into cell mass. Hence, approximately 30% of the added acetate carbon was converted to volatile products, most likely ¹⁴CO₂, and 20% was covered in soluble products (which may be partly due to cell lysis). The majority of acetate served as a carbon source, and the rate of carbon assimilation from acetate was consistent with the rate of autotrophic carbon fixation. This shows that, despite the presence of CO₂ and H₂, the cells preferentially used acetate as a carbon source. Only a minor part of the acetate was oxidized to CO₂, probably in the course of the assimilation process. Complete oxidation of acetate is unlikely since no external electron acceptor was added.

Biotin stain for carboxylases.

The measured activities of the biotin-dependent carboxylases in cell extracts were generally quite low. These enzyme complexes are notorious for disassembling in vitro, resulting in inactivation. The genome of C. aurantiacus harbors genes for three different biotin-dependent carboxylases and for two membrane-associated sodium ion-translocating biotin-dependent decarboxylases (presumably methylmalonyl-CoA decarboxylase) (see Table S1 in the supplemental material). The genes encoding different subunits of the carboxylases are not clustered in the C. aurantiacus genome. The question was whether all of these carboxylases are expressed and whether acetyl-CoA and propionyl-CoA are carboxylated by the same enzyme. Furthermore, there may be different isozymes of acetyl-CoA carboxylase, one for CO₂ fixation and another one for fatty acid biosynthesis. Thus, we used a highly specific biotin stain (avidin coupled to peroxidase) to detect biotin-containing proteins in the soluble protein fraction in cell extracts after SDS-PAGE and blotting onto a nitrocellulose membrane.

As shown in Fig. 3, there are at least three different biotinylated proteins present, which corresponds to what was found in the C. aurantiacus genome. These bands correlate with genes encoding biotin carboxyl carrier proteins/protein domains (Caur_1378, 64.3 kDa; Caur_2832, 69.4 kDa; and Caur_3739, 19.3 kDa) of acetyl-CoA/propionyl-CoA carboxylases. However, all the biotin carboxyl carrier proteins migrated as if their molecular mass was increased, presumably due to some kind of modification. The smaller, faint bands may correspond to the biotin/lipoyl attachment proteins (Caur_3053, 13.6 kDa; and Caur_3433, 16.2 kDa) of sodium ion-translocating decarboxylases (see Table S1 in the supplemental material). Although a slightly varying pattern of expression was observed, this variation does not correlate with the measured activities (Table 1). Note that only the carboxyl transferase subunits dictate substrate specificity. Also, the genes encoding biotin carboxyl carrier proteins are not clustered together with genes of other subunits of the carboxylases. Hence, the regulation of transcription may differ for all subunits. This approach did not discriminate genes encoding acetyl-CoA carboxylase, propionyl-CoA carboxylase, or a less specific carboxylase acting on both CoA thioesters.

Fig. 3.

Biotin staining of different extracts of C. aurantiacus cells grown under different conditions. Streptavidin-coupled peroxidase was used for the detection of biotin carrier subunits. The two upper bands correspond to fusion proteins Caur_1378 (64.3 kDa) and Caur_2832 (69.4 kDa), each comprising a biotin carboxyl carrier protein and a biotin carboxylase. The lower band corresponds to a protein with only the biotin carboxyl carrier domain, Caur_3739 (19.3 kDa). The PageRuler unstained protein ladder (Fermentas, St. Leon-Rot, Germany) was used as a molecular mass standard.

Occurrence of characteristic genes of the 3-hydroxypropionate bi-cycle in other bacteria.

BLAST searches in the NCBI database (http://blast.ncbi.nlm.nih.gov/) were used for a survey of the distribution among prokaryotes of those genes/enzymes that are related to the 3-hydroxypropionate bi-cycle. Some of these enzymes also belong to other common metabolic pathways. Acetyl-CoA carboxylase catalyzes the first committed step of fatty acid biosynthesis. Propionyl-CoA carboxylase, methylmalonyl-CoA epimerase, and methylmalonyl-CoA mutase take part in propionate and odd-chain fatty acid metabolism. Other enzymes, such as succinate dehydrogenase and fumarase, belong to the citric acid cycle. Hence, only a limited set of genes was chosen for this search, i.e., in the first partial cycle the key genes for malonyl-CoA reductase and propionyl-CoA synthase and in the second glyoxylate-assimilating partial cycle the genes for the characteristic C₅-transforming enzymes. The occurrence of these genes in representative other bacteria is summarized in Table 3. The organization of the genes is shown in Fig. 4.

Table 3.

Occurrence of characteristic genes/enzymes of the 3-hydroxypropionate bi-cycle in bacteria

Organism	Occurrence of indicated gene/enzymea
	Mcr	Pcs	Smt	Mcl	Mch	Mct	Meh
	2	3	7	10	11	12	13
Chloroflexus aurantiacus	X	X	X	X	X	X	X
Chloroflexus aggregans	X	X	X	X	X	X	X
Roseiflexus sp. strain RS-1	X	X	X	X	X	X	X
Roseiflexus castenholzii	X	X	X	X	X	X	X
Oscillochloris trichoides	X	X	x		x
Gammaproteobacterium NOR5-3	X	X
Gammaproteobacterium NOR51-B	X	X		x	x
Gammaproteobacterium HTCC2080	X	X
Congregibacter litoralis		X
Nitrococcus mobilis		x
Chloroherpeton thalassium		X
Erythrobacter sp. strain NAP1	X	X		x	x	X	X
“Candidatus Accumulibacter phosphatis”			X	X	X	X	X
Alphaproteobacteria				X	X	X	X
Betaproteobacteria			X	(X)	(X)	X	X
Actinobacteria				X	(X)	X	X

^aNumbers correspond to the reaction numbers in Fig. 1. Key genes of the autotrophic pathway are indicated by a gray background and genes required for glyoxylate assimilation by a white background. Lowercase represents genes with low identities (less than 40 or 50%, respectively; see the supplemental material). Parentheses correspond to genes that are present in only a very few members of the respective phylogenetic group. Enzymes are malonyl-CoA reductase (Mcr), propionyl-CoA synthase (Pcr), succinyl-CoA:(S)-malate-CoA transferase (Smt), (S)-malyl-CoA/β-methylmalyl-CoA lyase/(S)-citramalyl-CoA lyase (Mcl), mesaconyl-C1-CoA hydratase (Mch), mesaconyl-CoA C1:C4 CoA transferase (Mct), and mesaconyl-C4-CoA hydratase (Meh).

Fig. 4.

Organization of characteristic genes of the 3-hydroxypropionate bi-cycle in different bacteria. The enzymes are mesaconyl-C4-CoA hydratase (meh), succinyl-CoA:(S)-malate CoA transferase (smtAB), mesaconyl-CoA C1:C4 CoA transferase (mct), (S)-malyl-CoA/β-methylmalyl-CoA/(S)-citramalyl-CoA lyase (mcl), mesaconyl-C1-CoA hydratase (mch), propionyl-CoA synthase (pcs), malonyl-CoA reductase (mcr), acetyl-CoA carboxylase (accACD), beta-ketothiolase (bkt) (not required for the bi-cycle), and acyl-CoA dehydrogenase (acd) (not required for the bi-cycle). White arrows show open reading frames (ORFs) of unknown function. For catalyzed reactions, see Fig. 1.

Only members of the family of Chloroflexaceae (Chloroflexus aggregans and Roseiflexus spp.) harbor all the genes required to establish the complete 3-hydroxypropionate bi-cycle (22, 40, 44). Interestingly, Oscillochloris trichoides (also Chloroflexi) possesses only genes encoding malonyl-CoA reductase and propionyl-CoA synthase but lacks genes of the second glyoxylate assimilation cycle (21, 25). Note that this genome is still in draft form, and it is not certain that these genes are missing. However, Oscillochloris uses the Calvin-Benson cycle for autotrophic growth (5, 20), and malonyl-CoA reductase and propionyl-CoA synthase were not found to be active in cell ex- tracts (5). Genes for malonyl-CoA reductase and propionyl-CoA synthase were also found in various ubiquitous marine proteobacteria (gammaproteobacteria, unnamed strains NOR5-3, NOR51-B, and HTCC2080; alphaproteobacteria, Erythrobacter sp. strain NAP1). The propionyl-CoA synthase gene alone is present in the marine gammaproteobacteria Congregibacter litoralis and Nitrococcus mobilis as well as in Chloroherpeton thalassium (Chlorobiaceae), whereas malonyl-CoA reductase is lacking. Note that the Nitrococcus propionyl-CoA synthase lacks the CoA ligase domain, but this function may be taken over by another CoA ligase. These bacteria may use a rudimentary cycle for the mixotrophic assimilation of acetate, 3-hydroxypropionate, and/or propionate.

DISCUSSION

Functioning of 3-hydroxypropionate bi-cycle enzymes under mixotrophic and heterotrophic growth conditions.

Autotrophically grown cells of C. aurantiacus readily took up organic carbon compounds without any lag phase. This observation corroborates the idea that the 3-hydroxypropionate bi-cycle is a means of coassimilating organic carbon (Fig. 1). The rates of organic carbon uptake were consistent with the assumption that these compounds are completely assimilated. Not unexpectedly, the rate of carbon acquisition from organic substrates was even higher than the autotrophic carbon assimilation rate. Mixotrophy thus enables Chloroflexus to increase growth rate and yield compared to those in purely autotrophic growth. The 3-hydroxypropionate bi-cycle makes a balanced redox state of the cell possible, since CO₂ fixation consumes electrons (up to 10 electrons in one complete turn). Even an incomplete bi-cycle may be a necessity for anaerobic growth on substrates that are more reduced than the average cell carbon. A similar mechanism has been described for the Calvin-Benson cycle functioning in redox homeostasis under anaerobic photoheterotrophic conditions (27, 29, 34).

The characteristic enzymes of the 3-hydroxypropionate bi-cycle were active in cells grown under mixotrophic (in the presence HCO_3⁻) or purely heterotrophic conditions. In addition, the key enzymes of the glyoxylate cycle were active and malate synthase was induced under aerobic-dark conditions with acetate, indicating that the glyoxylate cycle may be responsible mainly for aerobic acetate assimilation. In fact, the incomplete 3-hydroxypropionate bi-cycle may represent another strategy of acetyl-CoA assimilation, as an alternative to the glyoxylate cycle and the ethylmalonyl-CoA pathway. Other products such as 3-hydroxypropionate and propionate can be assimilated simultaneously along with CO₂ and acetate. C₃ compounds need to generate some acetyl-CoA, which requires malyl-CoA lyase and glyoxylate assimilation. Otherwise, formation of acetyl-CoA as well as pyruvate would involve the unfavorable decarboxylation of C₄ compounds. Coassimilation of glycolate/glyoxylate via the second half of the 3-hydroxypropionate bi-cycle would result in additional CO₂ fixation, in contrast to the case for other glycolate salvage pathways, which are associated with a loss of CO₂. This feature may hold some potential for bioengineering.

The general anabolic use of the bi-cycle is reflected by the minor regulation of the key enzymes. The original purpose of the 3-hydroxypropionate bi-cycle may have been the assimilation of reduced organic compounds under conditions where oxygen is only sometimes available. Autotrophic CO₂ fixation via this pathway might have been just a secondary development, which was beneficial for Chloroflexus because it could exploit CO₂ as a carbon source as long as an electron donor was available. Thus, the complete bi-cycle may be a late and singular invention in the Chloroflexi.

Acquisition of genes of the 3-hydroxypropionate bi-cycle.

The organization of genes encoding characteristic enzymes of the pathway may be disadvantageous for gene transfer. Although the genes necessary for the glyoxylate assimilation part are clustered in the C. aurantiacus genome, the genes encoding malonyl-CoA reductase and propionyl-CoA synthase are located (each separately) far away from that gene cluster (Fig. 4). Therefore, a number of gene transfer events would be required to establish the whole bi-cycle in other organisms. However, in two closely related Roseiflexus species these two genes form another cluster together with genes probably encoding acetyl-CoA carboxylase (22).

Interestingly, the gammaproteobacterium HTCC2080 (38) seems to possess the genes for a chimeric 3-hydroxypropionate/4-hydroxybutyrate cycle that may allow even for autotrophic growth: genes required for the conversion of acetyl-CoA plus two bicarbonate molecules to succinyl-CoA are of the Chloroflexus type. In contrast, the regeneration of acetyl-CoA from succinyl-CoA may proceed as in the pathways found in autotrophic Sulfolobales, Thermoproteales, and Desulfurococcales (6, 18, 33). There, succinyl-CoA is reduced to 4-hydroxybutyrate, which is activated and further converted to two molecules of acetyl-CoA.

Besides Chloroflexi, only a few bacteria possess all genes required for the glyoxylate-assimilating partial cycle. One outstanding example is “Candidatus Accumulibacter phosphatis” (17), a betaproteobacterium that appears to harbor the whole Chloroflexus-like glyoxylate assimilation gene cluster. Under anoxic conditions “Candidatus Accumulibacter phosphatis” uses its glycogen and polyphosphate storages, while taking up acetate and propionate from the surroundings and producing polyhydroxybutyrate/polyhydroxyvalerate (8, 15, 16, 42). Remobilization of polyhydroxyvalerate yields acetyl-CoA and propionyl-CoA. The Chloroflexus-type enzymes of the glyoxylate assimilation cycle may be associated with the assimilation of propionyl-CoA.

Phylogenetic trees for characteristic enzymes.

To address the origin of key genes of the 3-hydroxypropionate bi-cycle, we constructed phylogenetic trees for bifunctional malonyl-CoA reductase, trifunctional propionyl-CoA synthase, promiscuous (S)- malyl-CoA/beta-methylmalyl-CoA/(S)-citramalyl-CoA lyase, and enzymes of the glyoxylate assimilation cycle (see Fig. S1 to S5 in the supplemental material). The neighbor-joining trees for malonyl-CoA reductase and propionyl-CoA synthase look very similar, suggesting that both genes were transferred or gained to- gether. The (S)-malyl-CoA/beta-methylmalyl-CoA/(S)-citramalyl- CoA lyase family forms three clusters: the Chloroflexus-type lyase cluster of a 360-amino-acid enzyme; a Rhodobacter-like cluster of a 320-amino-acid enzyme, Mcl1, that plays a role in the ethylmalonyl-CoA pathway for acetyl-CoA assimilation in many bacteria (11–14): and a distinct Mcl2 cluster representing a specific (S)-malyl-CoA hydrolase (thioesterase), also associated with the ethylmalonyl-CoA pathway (12). Interestingly, Methylobacterium spp. harbor all three types of enzymes (31).

All enzymes required for the glyoxylate assimilation part of the 3-hydroxypropionate bi-cycle form clades with similar topologies. Each of these clades comprises the respective enzymes of Chloroflexaceae and “Candidatus Accumulibacter phosphatis.” This strongly indicates that the whole Chloroflexus-like cluster of genes for glyoxylate assimilation was transferred in a singular event to “Candidatus Accumulibacter.” Considering each enzyme separately, one will find that they are all widely spread among bacteria, whereas only a few species possess more than two or three of the genes together. This may be an example of the modularity of metabolic pathways, where enzymes are gained through lateral gene transfer, sometimes mutated in order to achieve new substrate specificities, and combined to pathways serving completely different purposes.