A 3-Hydroxypropionate/4-Hydroxybutyrate Autotrophic Carbon Dioxide Assimilation Pathway in Archaea

See allHide authors and affiliations

Science  14 Dec 2007:
Vol. 318, Issue 5857, pp. 1782-1786
DOI: 10.1126/science.1149976


The assimilation of carbon dioxide (CO2) into organic material is quantitatively the most important biosynthetic process. We discovered that an autotrophic member of the archaeal order Sulfolobales, Metallosphaera sedula, fixed CO2 with acetyl–coenzyme A (acetyl-CoA)/propionyl-CoA carboxylase as the key carboxylating enzyme. In this system, one acetyl-CoA and two bicarbonate molecules were reductively converted via 3-hydroxypropionate to succinyl-CoA. This intermediate was reduced to 4-hydroxybutyrate and converted into two acetyl-CoA molecules via 4-hydroxybutyryl-CoA dehydratase. The key genes of this pathway were found not only in Metallosphaera but also in Sulfolobus, Archaeoglobus, and Cenarchaeum species. Moreover, the Global Ocean Sampling database contains half as many 4-hydroxybutyryl-CoA dehydratase sequences as compared with those found for another key photosynthetic CO2-fixing enzyme, ribulose-1,5-bisphosphate carboxylase-oxygenase. This indicates the importance of this enzyme in global carbon cycling.

The dominant extant autotrophic pathway is the Calvin-Bassham-Benson cycle, which may have evolved late in evolution (1). Three additional autotrophic pathways are known: the reductive citric acid cycle (Arnon-Buchanan cycle), the reductive acetyl-CoA pathway (Wood-Ljungdahl pathway), and the 3-hydroxypropionate cycle. These pathways differ in several ways [e.g., with respect to energy demand, available reducing compounds, requirement for metals (Fe, Co, Ni, and Mo), usage of coenzymes, and oxygen sensitivity of enzymes]. These criteria determine the distribution of the pathways in autotrophic organisms in different habitats, in addition to their phylogeny.

In the thermoacidophilic Archaea Sulfolobus, Acidianus, and Metallosphaera spp., CO2 fixation appears to proceed via a modified, but until now unresolved, variant of the 3-hydroxypropionate cycle (26) (Fig. 1). The 3-hydroxypropionate cycle was originally discovered in the phototrophic bacterium Chloroflexus aurantiacus (7, 8); it involves the carboxylation of acetyl-CoA to malonyl-CoA by the biotin-dependent acetyl-CoA carboxylase. Because Archaea contain only trace amounts of fatty acids, if any, in their lipids, acetyl-CoA carboxylase cannot serve as the key enzyme of fatty acid synthesis. It seems that this enzyme functions in another metabolic pathway. Malonyl-CoA is reduced via malonate semialdehyde to 3-hydroxypropionate, which is further reductively converted to propionyl-CoA. Propionyl-CoA is carboxylated to (S)-methylmalonyl-CoA by a carboxylase, which is similar or identical to acetyl-CoA carboxylase. (S)-Methylmalonyl-CoA is isomerized to (R)-methylmalonyl-CoA, followed by carbon rearrangement to succinyl-CoA by coenzyme vitamin B12–dependent methylmalonyl-CoA mutase. In Chloroflexus, succinyl-CoA is convertedto(S)-malyl-CoA, which is cleaved by (S)-malyl-CoA lyase to acetyl-CoA (thus regenerating the CO2-acceptor molecule) and glyoxylate (8). However, this succinyl-CoA conversion via (S)-malyl-CoA was not verified in Sulfolobales (3, 4, 6), leaving the problem of how acetyl-CoA is regenerated in this group of organisms.

Fig. 1.

Proposed autotrophic 3-hydroxypropionate/4-hydroxybutyrate cycle in M. sedula. Reactions of the cycle are shown. Enzymes: 1, acetyl-CoA carboxylase; 2, malonyl-CoA reductase (NADPH); 3, malonate semialdehyde reductase (NADPH); 4, 3-hydroxypropionyl-CoA synthetase (AMP-forming); 5, 3-hydroxypropionyl-CoA dehydratase; 6, acryloyl-CoA reductase (NADPH); 7, propionyl-CoA carboxylase; 8, methylmalonyl-CoA epimerase; 9, methylmalonyl-CoA mutase; 10, succinyl-CoA reductase (NADPH); 11, succinate semialdehyde reductase (NADPH); 12, 4-hydroxybutyryl-CoA synthetase (AMP-forming); 13, 4-hydroxybutyryl-CoA dehydratase; 14, crotonyl-CoA hydratase; 15, 3-hydroxybutyryl-CoA dehydrogenase (NAD+); 16, acetoacetyl-CoA β-ketothiolase. The proposed pathway of glyceraldehyde 3-phosphate synthesis from acetyl-CoA and CO2 is also shown. Enzymes: 17, pyruvate synthase; 18, pyruvate, water dikinase [phosphoenolpyruvate (PEP) synthase]; 19, enolase; 20, phosphoglycerate mutase; 21, 3-phosphoglycerate kinase; 22, glyceraldehyde 3-phosphate dehydrogenase. The activities of pyruvate synthase and PEP synthase at 75°C were 10 to 25 nmol min–1 mg–1 protein (6) and 25 nmol min–1 mg–1 protein, respectively.

A comparison of the genomes of autotrophic Archaea that have acetyl-CoA carboxylase genes revealed several groups, which differ with respect to the pattern of other characteristic genes and enzymes (table S1). One group comprising autotrophic members of the genera Metallosphaera, Sulfolobus, Archaeoglobus, and Cenarchaeum contains the genes for acetyl-CoA/propionyl-CoA carboxylase and methylmalonyl-CoA mutase. These organisms differ considerably in their energy metabolism and habitats. Metallosphaera and Sulfolobus spp. metabolize sulfur, pyrite, or hydrogen in volcanic areas (9); Archaeoglobus spp. are anaerobic sulfate reducers (10); and Cenarchaeum sp. is a symbiont of marine animals (11), which is closely related to Nitrosopumilus sp., a marine ammonia oxidizer (12). All these organisms contain the gene for 4-hydroxybutyryl-CoA dehydratase, a [4Fe–4S] cluster and flavin adenine dinucleotide–containing enzyme, which catalyzes the elimination of water from 4-hydroxybutyryl-CoA by a ketyl radical mechanism yielding crotonyl-CoA (13, 14). This reaction is known to play a role in a few anaerobic bacteria such as Clostridium aminobutyricum, which ferment 4-aminobutyrate via 4-hydroxybutyryl-CoA (15).

We suspected that this group of Archaea is able to reduce succinyl-CoA to 4-hydroxybutyrate and convert it with the participation of 4-hydroxybutyryl-CoA dehydratase into two molecules of acetyl-CoA. Cell extracts of M. sedula, indeed, catalyzed a rapid (500 nmol min–1 mg–1 protein at 65°C) reduction of succinyl-CoA via succinate semialdehyde to 4-hydroxybutyrate with NADPH, the reduced form of nicotinamide adenine dinucleotide phosphate (NADP+) (16). Extracts also converted 4-hydroxy[1-14C]butyrate to [14C]acetyl-CoA at 50 nmol min–1 mg–1 protein at 65°C in the presence of Mg–adenosine triphosphate (MgATP), CoA, and nicotinamide adenine dinucleotide (NAD+) (Fig. 2A). High-performance liquid chromatography (HPLC) analysis of the reaction course revealed labeled 4-hydroxybutyryl-CoA, crotonyl-CoA, and 3-hydroxybutyryl-CoA as intermediates (Fig. 2B). A plot of the relative amounts of radioactivity in the individual products versus time indicated that the order of intermediates was as expected and that acetyl-CoA was the end product (Fig. 2C). The organism grows at 75°C in a H2/O2/CO2 mixture with a minimum generation time of 8 hours, which corresponds to a specific carbon fixation rate in vivo of 120 nmol min–1 mg–1 protein (17). Because two molecules of carbon are fixed in the proposed pathway, the minimal in vivo specific activity of enzymes in the cycle is 60 nmol min–1 mg–1 protein. All enzymes required for the postulated succinyl-CoA conversion to two acetyl-CoA were detected, and their specific activities met this expectation (see Table 1, which also summarizes the other enzyme activities; for cofactor specificity, see Fig. 1).

Fig. 2.

Conversion of 4-hydroxy [14C]butyrate to [14C]acetyl-CoA at 65°C by cell extracts of M. sedula. The reaction mixture (0.5 ml) contained 100 mM Mops/KOH (pH 7.2), 3 mM MgCl2, 5 mM dithiothreitol, 1 mM 4-hydroxy[1-14C]butyrate (160 kilobecquerels ml–1), 3 mM ATP, 2 mM CoA, and 2 mM NAD+. The reaction was started by the addition of 10 μl of cell extract (to the final concentration of 0.4 mg protein ml–1 in the assay). (A) HPLC separation of labeled substrate and products and 14C detection by flow-through scintillation counting. The figure shows samples (50 μl) taken after 30-s, 60-s, 150-s, and 20-min incubation, as well as a control experiment lacking ATP after 20-min incubation. The radioactive peak at 3.4 min most likely represents γ-butyrolactone, which forms spontaneously from 4-hydroxybutyryl-CoA at neutral pH or from 4-hydroxybutyrate at acidic pH. (B) Time course of substrate consumption and product formation. (C) Percentage of radioactivity present in the individual products, as compared with the total radioactivity in all labeled products, versus time. The strong negative slope for 4-hydroxybutyryl-CoA (4-OH-butyryl-CoA) indicates that it is the first intermediate. The strong positive slope for acetyl-CoA indicates that it is the end product. Crotonyl-CoA and 3-hydroxybutyryl-CoA (3-OH-butyryl-CoA) behave like intermediates between 4-hydroxybutyryl-CoA and acetyl-CoA.

Table 1.

Enzymes of the proposed 3-hydroxypropionate/4-hydroxybutyrate cycle in M. sedula. Because of the use of mesophilic coupling enzymes and the instability of some substrates, the indicated assay temperatures were used; the growth temperature was 75°C. The values are average values of several determinations, the mean deviations are about 10%, depending also on the batch of cells. The genes for 4-hydroxybutyryl-CoA synthetase, acryloyl-CoA reductase, succinate semialdehyde reductase, and one of the copies of 3-hydroxybutyryl-CoA dehydrogenase form a cluster in the M. sedula genome. ND, not determined.

View this table:

Some of the enzymes, for which no likely candidate gene could be identified in the genome, were purified, and the relevant genes were deduced from internal peptide sequences (Table 1). Succinyl-CoA reductase (NADPH) with a 39-kD subunit is identical to malonyl-CoA reductase, a paralog of aspartate semialdehyde dehydrogenase (17). Succinate semialdehyde reductase (NADPH) with a 40-kD subunit belongs to Zn-containing alcohol dehydrogenases. 4-Hydroxybutyryl-CoA synthetase [adenosine monophosphate (AMP)–forming] with a 50-kD subunit belongs to acyl-CoA synthetases. Similar genes were present in the sequenced Sulfolobales genomes (figs. S1 to S3).

Hence, the proposed 3-hydroxypropionate/4-hydroxybutyrate cycle, illustrated in Fig. 1, can be divided into two parts. The first transforms acetyl-CoA and two bicarbonate molecules to succinyl-CoA, and the second part converts succinyl-CoA to two acetyl-CoA molecules. In most Archaea, only one copy of genes for the acetyl-CoA/propionyl-CoA carboxylase subunits is present, which suggests that this is a promiscuous enzyme that acts on both acetyl-CoA and propionyl-CoA, as shown for Acidianus brierleyi (18) and M. sedula (19) and proposed for Cenarchaeum symbiosum (20).

A comparison of the proposed 3-hydroxypropionate/4-hydroxybutyrate cycle with the Calvin-Bassham-Benson cycle reveals important differences. The formation of one molecule of glyceraldehyde 3-phosphate from three molecules of CO2 follows the equation: 3 CO2 +6 NAD(P)H + 9 ATP → 1 triosephosphate + 6 NAD(P)+ + 9 adenosine diphosphate (ADP) + 8 inorganic phosphate (Pi). This does not account for the loss of energy and reducing power from the oxygenase activity of ribulose-1,5-bisphosphate carboxylase-oxygenase (RuBisCO). Triosephosphate formation via the proposed cycle (Fig. 1) follows the equation: 2 bicarbonate + 1 CO2 +5 NAD(P)H + 2 reduced ferredoxin + 6 ATP → 1 triosephosphate + 5 NAD(P)+ + 2 oxidized ferredoxin + 3 ADP + 4 Pi + 3 AMP + 2 inorganic pyrophosphate (PPi). Hence, in energetic terms, the two processes are comparable [because ATP cleavage to AMP is equivalent to the loss of two energy-rich compounds, and reduced ferredoxin has a stronger reduction potential than NAD(P)H], but oxygenase activity of RuBisCO renders the Calvin-Bassham-Benson cycle energetically inferior.

The active “CO2” species is CO2 in the Calvin-Bassham-Benson cycle, whereas it is bicarbonate as cosubstrate for acetyl-CoA/propionyl-CoA carboxylase in the 3-hydroxypropionate/4-hydroxybutyrate cycle, with CO2 as cosubstrate for pyruvate synthase (Fig. 1). The turnover number of RuBisCO (average: 5 s–1) reflects a low catalytic efficiency that requires large amounts of enzyme (21, 22). M. sedula acetyl-CoA/propionyl-CoA carboxylase catalyzing both carboxylation reactions has a turnover number at 65°C of 28 s–1 (19). The Michaelis constant (Km) value of RuBisCO for dissolved CO2 [the average value is 0.05 mM, but it may be as high as 0.34 mM in some marine cyanobacteria (21)] is lower than or even comparable to the apparent Km value of archaeal acetyl-CoA/propionyl-CoA carboxylase for bicarbonate [0.3 mM (19)]. However, the CO2 concentration in water equilibrated with air (1 atm pressure, 20°C, 370 parts per million CO2) is 0.012 mM at pH 7.4 (which is the assumed intracellular pH), and the bicarbonate concentration under the same conditions is 0.26 mM. The concentration of bicarbonate will be even higher at pH 7.8 to 8.2 of seawater (apparent dissociation constant pKapp for CO2/HCO3 is 6.3, 20°C). Thus, the acetyl-CoA carboxylase affinity for the active “CO2” species appears not to be inferior to that of RuBisCO. In addition, in volcanic gases and in animal hosts, the CO2 concentration is much higher than in ambient air. Although 4-hydroxybutyryl-CoA dehydratase is slowly inactivated by oxygen, it may be sufficiently stable at low oxygen tensions to maintain activity, especially in the protected environment of the cell. There is no additional clue from the genome as to how these Crenarchaeota protect one of their key enzymes of CO2 fixation from oxygen inactivation.

The 4-hydroxybutyryl-CoA dehydratase gene was not found in heterotrophic Crenarchaea, yet it is present in the genomes of autotrophic Ignicoccus and Pyrobaculum spp. (table S1), although these genera lack the genes required for part 1 of the proposed pathway. I. hospitalis forms succinyl-CoA from acetyl-CoAvia pyruvate, phosphoenolpyruvate, and oxaloacetate (23) and may use the second half of the 3-hydroxypropionate/4-hydroxybutyrate cycle for the regeneration of the CO2-acceptor acetyl-CoA from succinyl-CoA. Pyrobaculum spp. use a reductive citric acid cycle for CO2 fixation, which results in acetyl-CoA formation from citrate (6); however, pyruvate synthase activity appears to be lacking in autotrophically grown cells (24). This enzyme is normally responsible for the reductive carboxylation of acetyl-CoA to pyruvate in autotrophic anaerobes. In the absence of pyruvate synthase, the second half of the proposed cycle may be used in the reverse direction to assimilate two molecules of acetyl-CoA into succinyl-CoA, to give a modified reductive citric acid cycle. Several heterotrophic Thermococcales and Halobacteriales harbor the characteristic genes for the first part of the cycle but lack those for the second part (table S1). Because their genomes appear to lack the genes for isocitrate lyase and malate synthase, which are key enzymes of the glyoxylate cycle, these organisms may thus use the first part of the cycle for the assimilation of acetyl-CoA, instead of the glyoxylate cycle, as well as for propionate assimilation.

4-Hydroxybutyryl-CoA dehydratase was originally described in a few strict anaerobes that use it in fermentation (15, 25). These strictly anaerobic Bacteria form a separate “anaerobe cluster” in the 4-hydroxybutyryl-CoA dehydratase phylogenetic tree (Fig. 3). Two independent clusters are formed by hyperthermophilic Crenarchaea that are unable to ferment, with type 1 being more common than type 2. Another cluster is associated with mesophilic Crenarchaea such as Cenarchaeum sp., a member of the “marine group”–1 Crenarchaeota, as well as with many marine environmental metagenomes from the Global Ocean Sampling (GOS) database (26, 27). The Basic Local Alignment Search Tool (BLAST) search identified 189 specific genes (18 complete genes and 171 fragments) of 4-hydroxybutyryl-CoA dehydratase in the GOS database, most of which (84%) group in the “marine cluster”–1 (Fig. 3). It is unlikely that the bearers of this gene conduct fermentation in the oxic seawater. The number of identified 4-hydroxybutyryl-CoA dehydratase gene sequences in the GOS database is comparable with that of the gene of RuBisCO large subunit (344 specific genes, including 26 complete genes and 318 fragments). This indicates that a group of abundant mesophilic autotrophic Crenarchaea in the sea (2830) may use the proposed 3-hydroxypropionate/4-hydroxybutyrate pathway for CO2 fixation; yet, such comparisons depend on the cloning efficiency of the required genes. Because these Crenarchaea probably use ammonia as an electron donor (12, 20, 30), their contribution to the carbon cycle should be much lower than that of phototrophs [∼1% of annual marine production (29)].

Fig. 3.

Phylogenetic tree of representative 4-hydroxybutyryl-CoA dehydratase proteins. The tree is based on amino acid sequence analysis and rooted with 4-hydroxyphenylacetate 3-monooxygenase (HpaB) from Escherichia coli, which is a flavoprotein that resembles the flavin-binding domain of 4-hydroxybutyryl-CoA dehydratase. Tree topography and evolutionary distances are given by the neighbor-joining method with Poisson correction. The phylogenetic trees constructed by other algorithms (maximum parsimony, distance matrix, and maximum likelihood) resulted in identical topologies for principal clade grouping. The scale bar represents a difference of 0.1 substitutions per site. Numbers at nodes indicate the percentage bootstrap values for the clade of this group in 1000 replications. Only values above 70% were considered. The tree represents all full gene sequences from the GOS database and the sequences belonging to the major clusters on the tree. The accession numbers are listed in table S2, and the alignment is shown in fig. S4.

It should be stressed that, in other Archaea, not all individual reactions of the proposed cycle may be catalyzed by the same kind of enzyme. For instance, various alcohol dehydrogenases, aldehyde dehydrogenases, acyl-CoA synthetases, or enoyl-CoA hydratases may fulfill the same function; yet, each superfamily of those enzymes comprises different enzyme families, from which genes may have been recruited. Also, the pathway of acetyl-CoA assimilation and phosphoenolpyruvate formation may differ. At last, one may be able to trace back the roots of the individual enzymes and come to tentative conclusions regarding the evolution of this cycle. Carboxyphosphate, which is used in the biotin-dependent carboxylase reactions, is an attractive model for carbon fixation during chemoevolution.

Several questions remain. What is the function of two different copies of 4-hydroxybutyryl-CoA dehydratase genes in autotrophic Sulfolobales? What are the functions of acetyl-CoA carboxylase and the enzymes of part 1 of the previously unrecognized pathway in various heterotrophic Archaea? Does the second part of the pathway, when operated in the reverse direction, serve as an acetyl-CoA assimilation route in some Archaea that lack the first part of the cycle? Which of the potential autotrophic pathways is operating in Archaeoglobus, which harbors the genes for the reductive acetyl-CoA pathway, for the proposed new pathway, and for RuBisCO (31, 32)?

Supporting Online Material

Materials and Methods

Figs. S1 to S7

Tables S1 and S2


References and Notes

View Abstract

Navigate This Article