Role of 4-Hydroxybutyrate-CoA Synthetase in the CO2 Fixation Cycle in Thermoacidophilic Archaea*

Background: Thermoacidophilic Sulfolobales contain a novel CO2 fixation pathway; all enzymes but one have been accounted for in Metallosphaera sedula. Results: Enzymes encoded in Msed_0394 and Msed_0406 each exhibit 4-hydroxybutyrate-CoA synthetase activity, consistent with transcriptomic evidence. Conclusion: Msed_0406 is likely the physiologically relevant enzyme in the cycle. Significance: All enzymes are now accounted for in the CO2 fixation cycle of M. sedula. Metallosphaera sedula is an extremely thermoacidophilic archaeon that grows heterotrophically on peptides and chemolithoautotrophically on hydrogen, sulfur, or reduced metals as energy sources. During autotrophic growth, carbon dioxide is incorporated into cellular carbon via the 3-hydroxypropionate/4-hydroxybutyrate cycle (3HP/4HB). To date, all of the steps in the pathway have been connected to enzymes encoded in specific genes, except for the one responsible for ligation of coenzyme A (CoA) to 4HB. Although several candidates for this step have been identified through bioinformatic analysis of the M. sedula genome, none have been shown to catalyze this biotransformation. In this report, transcriptomic analysis of cells grown under strict H2-CO2 autotrophy was consistent with the involvement of Msed_0406 and Msed_0394. Recombinant versions of these enzymes catalyzed the ligation of CoA to 4HB, with similar affinities for 4HB (Km values of 1.9 and 1.5 mm for Msed_0406 and Msed_0394, respectively) but with different rates (1.69 and 0.22 μmol × min−1 × mg−1 for Msed_0406 and Msed_0394, respectively). Neither Msed_0406 nor Msed_0394 have close homologs in other Sulfolobales, although low sequence similarity is not unusual for acyl-adenylate-forming enzymes. The capacity of these two enzymes to use 4HB as a substrate may have arisen from simple modifications to acyl-adenylate-forming enzymes. For example, a single amino acid substitution (W424G) in the active site of the acetate/propionate synthetase (Msed_1353), an enzyme that is highly conserved among the Sulfolobales, changed its substrate specificity to include 4HB. The identification of the 4-HB CoA synthetase now completes the set of enzymes comprising the 3HP/4HB cycle.

Metallosphaera sedula is an extremely thermoacidophilic archaeon that grows heterotrophically on peptides and chemolithoautotrophically on hydrogen, sulfur, or reduced metals as energy sources. During autotrophic growth, carbon dioxide is incorporated into cellular carbon via the 3-hydroxypropionate/4hydroxybutyrate cycle (3HP/4HB). To date, all of the steps in the pathway have been connected to enzymes encoded in specific genes, except for the one responsible for ligation of coenzyme A (CoA) to 4HB. Although several candidates for this step have been identified through bioinformatic analysis of the M. sedula genome, none have been shown to catalyze this biotransformation. In this report, transcriptomic analysis of cells grown under strict H 2 -CO 2 autotrophy was consistent with the involvement of Msed_0406 and Msed_0394. Recombinant versions of these enzymes catalyzed the ligation of CoA to 4HB, with similar affinities for 4HB (K m values of 1.9 and 1.5 mM for Msed_0406 and Msed_0394, respectively) but with different rates (1.69 and 0.22 mol ؋ min ؊1 ؋ mg ؊1 for Msed_0406 and Msed_0394, respectively). Neither Msed_0406 nor Msed_0394 have close homologs in other Sulfolobales, although low sequence similarity is not unusual for acyl-adenylate-forming enzymes. The capacity of these two enzymes to use 4HB as a substrate may have arisen from simple modifications to acyl-adenylate-forming enzymes. For example, a single amino acid substitution (W424G) in the active site of the acetate/propionate synthetase (Msed_1353), an enzyme that is highly conserved among the Sulfolobales, changed its substrate specificity to include 4HB. The identification of the 4-HB CoA synthetase now completes the set of enzymes comprising the 3HP/4HB cycle.
Carbon dioxide is chemically stable and unreactive and must be reduced to enable its incorporation into biological molecules. Autotrophic microorganisms are able to utilize carbon dioxide as their sole carbon source, and a variety of pathways are known to activate and incorporate it into biomolecules essential for growth and replication. Recently, carbon dioxide fixation pathways have received interest for biotechnological applications, since this could provide biological routes for de novo generation of fuels and small organic molecules (1).
There are currently at least six natural pathways for the incorporation of inorganic carbon dioxide into cellular carbon (2,3). The most recently discovered of these are found exclusively in extremely thermophilic archaea as follows: the 3-hydroxypropionate/4-hydroxybutyrate (3HP 3 /4HB) carbon fixation cycle, which operates in members of the crenarchaeal order Sulfolobales (2, 4 -6), and the dicarboxylate/4-hydroxybutyrate (DC/4HB) cycle, which is used by anaerobic members of the orders Thermoproteales and Desulfurococcales (4,7). In both cycles, two carbon dioxide molecules are added to acetyl-CoA (C2) to produce succinyl-CoA (C4), which is subsequently rearranged to acetoacetyl-CoA and cleaved into two molecules of acetyl-CoA. These pathways differ primarily in regard to their tolerance to oxygen and the co-factors used for reducing equivalents as follows: NAD(P)H for the 3HP/4HB cycle and ferredoxin/NAD(P)H for the DC/4HB cycle (3,8). The two archaeal pathways also differ in how they link the CO 2 fixation cycle to central metabolism. In the DC/4HB pathway, pyruvate is synthesized directly from acetyl-CoA using pyruvate synthase. In the 3HP/4HB pathway, another half-turn is required to make succinyl-CoA, which is then oxidized via succinate to pyruvate (2,9,10).
There are 13 enzymes proposed to catalyze the 16 reactions in the 3HP/4HB pathway. The first three enzymes convert acetyl-CoA (C2) to 3HP (C3) via an ATP-dependent carboxylation step. Next, 3HP is converted and reduced to propionyl-CoA, carboxylated a second time, and rearranged to make succinyl-CoA (C4). Succinyl-CoA is reduced to 4HB, which is converted to two molecules of acetyl-CoA in the final reactions of the cycle. Flux analysis and labeling studies have confirmed the operation of this pathway in Metallosphaera sedula (4,10).
All of the enzymes that comprise the first portion of the cycle up to the formation of 4HB have been identified and characterized biochemically in their native or recombinant form, mostly from the extremely thermoacidophilic archaeon M. sedula (T ϭ 70°C, pH 2.0) (see Table 1) (4,5,(11)(12)(13). The enzymes involved in the conversion of 4HB to two molecules of acetyl-CoA have not been characterized to the same extent (Fig. 1). Activities corresponding to 4-hydroxybutyryl-CoA dehydratase and acetoacetyl-CoA ␤-ketothiolase have been detected in cell extracts (4,14), although neither enzyme has been purified in its native form or recombinantly produced. Identification of candidates for both of these enzymes has been made based on genome annotation and transcriptomic analysis of autotrophic growth compared with heterotrophy (8,9). Although neither of the candidate genes for these enzymes has so far been confirmed biochemically, their identity is not in dispute because of strong homology to known versions in less thermophilic organisms. The corresponding gene products in M. sedula are Msed_1321 for the 4HB-CoA dehydratase and Msed_0656 for the acetoacetyl-CoA ␤-ketothiolase.
The identity of the crotonyl-CoA hydratase and the (S)-3hydroxybutyryl-CoA dehydrogenase was recently confirmed when it was discovered that both reactions were catalyzed by a single bifunctional fusion protein (9). In the same work, Ramos-Vera et al. (9) tested three different candidates for the 4HB-CoA synthetase, but they all failed to show activity on 4HB. In fact, the primary candidate suggested by the autotrophic transcriptome analysis (Msed_1422) showed no enzymatic activity on short chain linear unsubstituted or hydroxy acids, specifically acetate, propionate, 3HP, 3-hydroxybutyrate, 4HB, and crotonate. Two other candidates were selected, based on homology to 4HB-CoA synthetase from Thermoproteus neutrophilus (Tneu_0420) and 3HP-CoA synthetase from M. sedula: Msed_1353 and Msed_1291 were recombinantly produced and tested for ligase activity. Msed_1353 was active on propionate and acetate, but not on 4HB. Furthermore, Msed_1291 had no activity on any of the previously mentioned organic acids. Thus, although cycle function has been confirmed by metabolic flux analysis, and although 4HB-CoA synthetase activity has been measured in cell extracts of autotrophically grown M. sedula, the enzyme responsible for ligation of CoA to 4HB remains unclear.
To identify the missing link in the 3HP/4HB cycle, new methods for semi-continuous cultivation of M. sedula in a gas-intensive fermentation system were developed to tease out differential transcriptional response of autotrophy-related genes. Strict carbon dioxide limitation was used to drive increased operational efficiency of the CO 2 fixation enzymes, which hypothetically would increase transcriptional levels of genes encoding key enzymes to maximize carbon incorporation. Using these conditions for transcriptional analysis, a much clearer picture emerged concerning the global regulatory changes in M. sedula as its cellular metabolism switches from autotrophy to heterotrophy. This strategy produced new leads for the genes and corresponding enzymes responsible for the 4HB-CoA ligation step. The enzymes were recombinantly produced and shown to catalyze the ligation of CoA to 4HB.

EXPERIMENTAL PROCEDURES
Growth of M. sedula in a Gas-intensive Bioreactor-M. sedula (DSMZ 5348) was grown aerobically at 70°C in a shaking oil bath (90 rpm) under autotrophic or heterotrophic conditions on DSMZ medium 88 at pH 2. Heterotrophically grown cells were supplemented with 0.1% tryptone. Cell growth was scaled up from 300 ml in sealed 1-liter bottles (see Ref. 8) to 2 liters in a stirred bench-top glass fermentor (Applikon), also on DSMZ medium 88, pH 2, at 70°C, and agitated at 250 rpm. Two separately regulated gas feeds were used such that flow rates were held constant for all conditions at 1 ml/min for the hydrogen/CO 2 gas mixtures (composition varied) and 100 ml/min for air (composition: 78% N 2 , 21% O 2 , 0.03% CO 2 ). For the autotrophic carbon-rich (ACR) condition, the gas feed contained H 2 (80%) and CO 2 (20%); for the autotrophic carbonlimited (ACL) condition, the feed was changed to H 2 (80%) and N 2 (20%); for the heterotrophic condition (HTR), the medium was supplemented with 0.1% tryptone, and the gas feed composition was N 2 (80%) and CO 2 (20%). Tandem fermentors were run simultaneously with the same inoculum to generate biological repeats (Fig. 2). Cells were harvested at mid-exponential phase by rapid cooling with dry ice and ethanol and then centrifuged at 6000 ϫ g for 15 min at 4°C. M. sedula Oligonucleotide Microarray Transcriptional Response Analysis-A spotted whole-genome oligonucleotide microarray, based on 2256 protein-coding open reading frames (ORFs), was used, as described previously (15). Total RNA was extracted and purified (RNeasy; Qiagen), reverse-transcribed (Superscript III; Invitrogen), re-purified, labeled with either Cy3 or Cy5 dye (GE Healthcare), and hybridized to the microarray slides (Corning). Slides were scanned on a GenePix 4000B Microarray Scanner (Molecular Devices, Sunnyvale, CA), and raw intensities were quantitated using GenePix Pro version 6.0. Normalization of data and statistical analysis were performed using JMP Genomics 5 (SAS, Cary, NC). In general, significant differential transcription was defined to be a relative change at or above 2 (where a log 2 value of Ϯ1 equals a 2-fold change) with significance values at or above the Bonferroni correction; for these data, this was 5.4 (equivalent to a p value of 4.0 ϫ 10 6 ). Microarray data are available through the NCBI Gene Expression Omnibus (GEO) under accession number GSE39944.
Enzyme Assays for 4-Hydroxybutyrate-CoA Synthetase-Two assays were used to measure ligase activity, one spectro-photometric and one using high performance liquid chromatography (HPLC). A discontinuous assay was used to measure substrate-dependent disappearance of CoA at 75°C. The reaction mixture (600 l) contained 100 mM MOPS/KOH, pH 7.9, 5 mM MgCl 2 , 2.5 mM ATP, 0.15 mM CoA, and purified enzyme. At each time point, 80 l of reaction mixture was added to 80 l of cold 5,5Ј-dithiobis-(2-nitrobenzoic acid). A time point (0 min) was taken before heating. The reaction mixture was incubated for 2 min at 75°C, followed by addition of substrate. Additional time points were taken at 30, 60, 90, 120, and 180 s after addition of substrate. Absorbance was measured at 412 nm to determine free CoA concentration, based on the concentration of 2-nitro-5-thiobenzoate dianion (⑀ 412 ϭ 14,150 M Ϫ1 cm Ϫ1 ) (1,14). Enzymes were kinetically characterized by varying the concentration of the acyl-CoA substrate from 0.05 to 12 mM, although the other substrate concentrations were held constant. Measurements for specific activity were taken under saturating substrate concentrations (10 mM). Formation of the CoA ester was also confirmed using HPLC (Waters). The reaction mixture (0.15 ml) contained 100 mM potassium phosphate, pH 7.9, 10 mM MgCl 2 , 2 mM ATP, 0.5 mM CoA, 10 mM substrate, and purified enzyme. The reaction was incubated for 3 min at 75°C, quenched with 15 l of 1 M HCl, filtered with a 10-kDa spin column (Amicon YM-10) to remove the protein,  and loaded onto a reversed-phase C18 silica-based column (Shodex C18 -4E, 4.6 ϫ 250 mm). The mobile phase was 50 mM sodium phosphate buffer, pH 6.7, with 2% methanol.
Heterologous Expression of M. sedula Genes in E. coli-M. sedula genes encoding acyl-CoA synthetases were amplified from genomic DNA using primers synthesized by Integrated DNA Technologies (Coralville, IA). Msed_0394 and Msed_ 0406 were ligated into pET46-Ek/LIC, although Msed_1353 was ligated into pET21b using NdeI and XhoI restrictions sites. All constructs were designed to express with an N-terminal His 6 tag. Plasmids containing gene inserts were cloned into NovaBlue GigaSingles E. coli competent cells and selected by growth on LB-agar supplemented with ampicillin (100 g/ml). Plasmid DNA was extracted using a QIAprep spin miniprep kit. Sequences were confirmed by Eton Biosciences, Inc. (Durham, NC). For protein expression, the plasmids were transformed into Escherichia coli Rosetta 2 (DE3) cells and selected by growth on LB-agar, supplemented with ampicillin (100 g/ml) and chloramphenicol (50 g/ml). Cells harboring the recombinant plasmid were induced with isopropyl 1-thio-␤-D-galactopyranoside (final concentration 0.1 mM) at A 600 0.4 -0.6 and cultured for 3 h before harvest.
Purification of Recombinant Proteins-Cells were harvested by centrifugation at 6000 ϫ g for 15 min at 4°C. Cell yields ranged from 1.6 to 3.8 g of cells/liter of LB medium (wet weight). Cell pellets were resuspended in lysis buffer (50 mM sodium phosphate, 100 mM NaCl, 0.1% Nonidet P-40, pH 8.0) containing DNase and lysozyme at final concentrations of 10 and 100 g/ml, respectively. Cells were lysed with a French press (two passes at 18,000 p.s.i.), and the lysate was centrifuged at 22,000 ϫ g for 15 min at 4°C to removed insoluble material. Soluble, cell-free extract was heated to 65°C for 20 min to precipitate mesophilic proteins. Streptomycin sulfate (1% w/v) was added to precipitate nucleic acids, followed by a 1-h incubation at 4°C. A final centrifugation was performed at 22,000 ϫ g for 15 min at 4°C to collect the soluble, heat-treated cell-free extract, which was sterile-filtered (0.22 m) and purified using a 5 ml HisTrap TM nickel column (GE Healthcare). Proteins were bound to the HisTrap TM column using binding buffer (50 mM sodium phosphate, 500 mM NaCl, 20 mM imidazole, pH 7.4) and eluted using elution buffer (50 mM sodium phosphate, 500 mM NaCl, 300 mM imidazole, pH 7.4). SDS-PAGE was then performed on the immobilized metal affinity chromatography fractions to qualitatively determine the purity of the protein before further purification. Chromatography fractions containing the protein were concentrated, exchanged into phosphate buffer (50 mM potassium phosphate, 150 mM NaCl, pH 7.0) using an Amicon YM10 (Millipore) centrifugal filter membrane, and centrifuged at 4000 ϫ g and 4°C. To quantify the amount of protein, a Bradford assay was performed on the concentrated immobilized metal affinity chromatography fractions using known serial dilutions of bovine serum albumin (BSA) by taking absorbance readings at 595 nm. Protein was further purified using a Superdex 200 10/300 GL (GE Healthcare) gel filtration column. The proteins were eluted from the gel filtration column using elution buffer (50 mM potassium phosphate, 150 mM NaCl, pH 7.0). Proteins were dialyzed into 100 mM MOPS-KOH (pH 7.9) and either stored at 4°C or mixed with glycerol to 20% and stored at Ϫ20°C.
Structural Modeling of Acyl-CoA Synthetases-Three-dimensional structural models for M. sedula acyl-CoA synthetases were made using the iterative threading assembly refinement (I-TASSER) on-line server (2,3,16). The server first generates three-dimensional atomic models from multiple threading alignments and iterative structural assembly and then infers function by structural matching to other known proteins. All structures were generated using the Protein Database entry for Salmonella enterica ACS (STM4275, 1PG4) as a threading template for additional restraint specification. Amino acid sequence alignments were generated using the UCSF Chimera package by superposition of I-TASSER threedimensional structural models with the Protein Data Bank structure for S. enterica ACS.
Materials-Plasmid vectors and strains were obtained from Novagen (San Diego) and Stratagene (La Jolla, CA). Chemicals, devices, and reagents were obtained from Fisher, ACROS Organics (Geel, Belgium), Sigma, New England Biolabs (Ipswich, MA), Qiagen (Valencia, CA), Millipore (Billerica, MA), and Invitrogen. Gases were purchased from Airgas National Welders (Charlotte, NC). Protein purification columns were obtained from GE Healthcare. The Bradford assay reagent was obtained from Bio-Rad. Site-directed mutagenesis kit was obtained from Invitrogen.

M. sedula Autotrophic Growth Is
Hydrogen-limited-To explore the optimal growth conditions for H 2 -CO 2 autotrophy in M. sedula, a fermentation system was designed to allow controlled definition of the gas feed. Previous autotrophic work with M. sedula was done in batch cultures in an orbital shaking bath at 70°C (2, 4 -6, 8). In that case, gas-fed cultures were grown by replacing the air in a sealed volume with a gaseous mixture of a known composition. Mass transfer of H 2 , CO 2 , and O 2 into the culture medium was limited to diffusion across the vapor-liquid interface. Gas limitation presumably affected these cultures, and led to suboptimal growth, as evidenced by the slow doubling time that resulted for M. sedula under these conditions (t d ϭ 11-13 h).
To grow M. sedula autotrophically with more optimal delivery of gaseous substrate to the liquid medium, a semi-continuous fermentation system was developed using a 3L bioreactor. The system was modified to have two separate gas feeds that sparged directly into the media (sparging stone, 2-m pore size). Microbubble sparging stones were used to promote dissolution of sparingly soluble gases, in particular H 2 . The biore-actor and console were situated inside a modified fume hood, with an airflow monitoring system in place to detect hood failure. Tandem fermentors were seeded with the same inoculum and run simultaneously to provide a biological repeat.
Growth of M. sedula in an aerobic, autotrophic fermentation system was expected to be H 2 -and not O 2 -limited. Below saturating conditions, growth rates varied according to the amount of H 2 fed to the culture. For high H 2 supply rates (i.e. 30 ml/min), the growth rates were comparable with the fastest growth rates previously observed under heterotrophy (t d ϭ 4.8 h); concomitantly, the culture reached a cell density of 2 ϫ 10 9 cells/ml, the highest observed under autotrophic conditions. At an H 2 supply rate of 15 ml/min, the growth rate slowed (t d ϭ 6 h), although the final density was comparable with the 30 ml/min case (1.5 ϫ 10 9 cells/ml). A 30-fold reduction in H 2 flow rates (1 ml/min) caused the growth rate to decrease by half (t d ϭ 9.7 h) and the cells to enter stationary phase at 8 ϫ 10 8 cells/ml.
A similar trend emerged in response to limiting levels of CO 2 . When CO 2 was supplemented in the gas feed (referred to here as "rich" autotrophy), the growth rate was faster than observed for cells grown with air as the only source of CO 2 (t d ϭ 6.8 h versus 9.4 h, respectively). The growth rate for heterotrophically grown cells (t d ϭ 6.7 h) was comparable with the rich autotrophy condition. This suggests that, under the rich autotrophy condition, the cells were not limited by any one particular gaseous substrate and were doubling at or near their maximal rate. The decrease in growth rate for the carbon-limited autotrophy arises from the limiting amounts of CO 2 available in the medium.
Optimized H 2 -CO 2 Autotrophy Conditions Lead to Enhanced Transcriptomic Response-The optimized autotrophic growth conditions enhanced the global transcriptional response compared with previous work (4,7,8). Of the 2293 protein-coding genes in the 2.2-kb M. sedula genome, nearly half (984 genes) exhibited changes in transcription (either up-or down-regulation) of 2-fold or greater, when comparing heterotrophy (HTR) with the ACL condition (see Table 2). The number of genes that were differentially transcribed was twice as high as observed previously (3,8,8), which could be attributed to the refined conditions for autotrophic growth. Also, in the experiments reported here, it should be mentioned that the improved sensitivity of new equipment used for scanning microarray slides improved the resolution and dynamic response.
New Candidates for 4-Hydroxybutyrate-CoA Synthetase Identified from Refined Transcriptomic Data-The refined transcriptomic data provided new insights into the putative candidates for 4-hydroxybutyrate-CoA synthetase (Fig. 3). Based on bioinformatic analysis, there are nine candidate genes encoding acyl-CoA synthetases (not including Msed_1456, which was confirmed as a 3HP-CoA synthetase). The high upregulation of Msed_1422 under autotrophy (13-fold increase) that was observed in this work is consistent with previous transcriptomic studies. On the basis of that initial study, Msed_1422 was chosen for recombinant expression and testing (2,9,9,10). In the same study, recombinant forms of Msed_1291 and Msed_1353 were also produced, which were chosen based on homology to a confirmed 4HB-CoA synthetase from T. neutrophilus (Tneu_0420). None of these enzymes showed activity on 4HB. Msed_1422 and Msed_1291 showed no activity on acetate, propionate, 3HP, 3HB, 4HB, or crotonate, and Msed_1353 had activity only on acetate and propionate but not 4HB. Thus, it appears that Msed_1353 is a promiscuous acetate/propionate synthetase, although the substrate specificities of Msed_1422 and Msed_1291 remain unknown.
Among the other potential candidates that were annotated as acetate-CoA synthetases or medium chain fatty acid-CoA synthetases (Fig. 3), most showed no transcriptional response, had average or low levels of transcription, or were clearly downregulated under autotrophy. The new transcriptomic data were consistent with the expression of two previously unexamined candidates, Msed_0406 and Msed_0394, that are annotated as an acetyl-CoA synthetase (ACS) and AMP-dependent synthetase and ligase, respectively. Although Msed_0406 and Msed_0394 were both constitutively transcribed, with less than a 2-fold change in transcription levels between the conditions tested, both of them were in the top 25% of the transcriptome. This served as the basis to investigate these two genes by recombinant expression and activity assays, given that no other promising candidates for this step had emerged.
Kinetic Analyses of Msed_0394 and Msed_0406-Recombinant forms of Msed_0394 and Msed_0406 were produced in E. coli and purified to electrophoretic homogeneity. For both enzymes, the production of 4HB-CoA from 4HB and CoA was confirmed using reversed-phase HPLC. Msed_0394 and Msed_0406 were active on a range of small organic acids (see Table 3 for a summary of kinetic data). Fig. 4 shows the relative specific activities on different substrates for Msed_0394 and Msed_0406, along with reported data for 3HP-CoA synthetase (Msed_1456) for comparison (4,5,10). Note that the calculated molecular mass for these three enzymes varies slightly, 62 kDa for Msed_0394, 64 kDa for Msed_0406, and 74 kDa for Site-directed Mutagenesis of Msed_1353-Msed_1353, a highly conserved gene among the Sulfolobales, was previously reported to have activity only on acetate and propionate (4,5,5,9,11,12,15). Initial efforts to identify the unknown 4HB-CoA synthetase in M. sedula involved purification of native enzyme activity and analysis of multiple SDS-polyacrylamide gel bands using mass spectrometry. Msed_1353 was detected in these experiments, and based on the very large up-regulation of Msed_1353 under autotrophy, it was recombinantly produced to confirm its activity. Our results confirmed previous reports; under saturating substrate concentrations, Msed_1353 had the highest activity on acetate (8.9 mol min Ϫ1 mg Ϫ1 , 100%) and propionate (99%) but also on 3HP (8%) and butyrate (16%). However, no activity was found on 4HB or longer organic acid substrates (see Fig. 5A).
Structural modeling of the binding pocket of Msed_1353 revealed a conserved tryptophan residue, similar to that seen in acetate-CoA synthetase (ACS) from S. enterica (4,14,17). This tryptophan forms the bottom surface of the binding pocket and limits the size of substrate that can be accommodated within the active site. To test the importance of this residue in determining substrate specificity, Trp 424 in Msed_1353 was mutated to a glycine to produce Msed_1353-G424. The single substitution mutant (W424G) was predicted to contain a larger interior binding pocket for the hydrophobic end of the substrate. Accordingly, it showed a dramatic change in specificity (Fig.  5B). Activity for the mutant on acetate and propionate decreased by 60%, from 8.9 to 3.6 and 8.8 to 3.5 mol min Ϫ1 mg Ϫ1 , respectively. However, Msed_1353-G424 also showed activity on C4-C8 substrates, including 4HB (1.8 mol min Ϫ1 mg Ϫ1 ).
To compare the activity of these three enzymes on 4HB, the Michaelis-Menten curves are shown in Fig. 6. From this figure, it is clear that there is a large difference in catalytic rate for the three enzymes, and this difference holds over the entire range of substrate concentration, including when [S]/K m Ͻ Ͻ1. Therefore, although it is possible that both Msed_0394 and Msed_0406 are catalytically active on 4HB in vivo, it is likely that Msed_0406 is more physiologically relevant in terms of catalytic performance. Additionally, the single point mutation of Msed_1353 to Msed_1353-Gly 424 produces an enzyme that is active on 4HB at even higher rates for all substrate concentrations.

DISCUSSION
The semi-continuous gas-intensive bioreactor system developed here was successfully used to refine the transcriptional response of autotrophy-related genes in M. sedula. This system  provided better delivery of gases with low solubility and allowed more precise regulation of gas composition than could be achieved in serum bottles. At 70°C and 1 atm, the solubility of oxygen and hydrogen are comparable (0.6 mM), although the solubility of carbon dioxide is about 20-fold higher (12 mM) (8,9,18). For these experiments, the low solubility of H 2 was offset by the use of microbubbler sparing stones (2 m pore size) to increase the gas phase surface area and increase delivery of H 2 to the medium. Stoichiometrically, at least four H 2 molecules are required for every carbon atom fixed. Assuming that ATP generation requires the oxidation of two hydrogen molecules, then each turn of the cycle requires 12 molecules of hydrogen for every two molecules of carbon dioxide. As such, the limiting growth factor for M. sedula in a bioreactor is likely acquisition of the electron donor, in contrast to most aerobic microbial fermentations where acquisition of the final electron acceptor, oxygen, limits growth. In its natural environment, the picture may be somewhat different. Hydrogen measurements from the (largely anoxic) acidic hot springs at Yellowstone indicate that gaseous hydrogen may be quite abundant, with concentrations ranging between 10 and 300 nM (8,19). The source of this hydrogen gas is primarily geochemical; although the mechanism is not well understood, it probably arises from subsurface interaction of water with Fe[II] (15,20). For most subsurface environments, oxygen is probably limiting (21). However, M. sedula was isolated from aerobic (surface) samples of a hot water pond at Pisciarelli Solfatara, Naples, Italy (22). Thus, both hydrogen and oxygen may be available in abundance for autotrophic growth.
The regulation of growth modes in M. sedula involves massive transcriptional changes between heterotrophic and autotrophic growth. Nearly half the genome (984 genes out of 2293) responded with transcriptional changes of 2-fold or greater when comparing heterotrophy to carbon dioxide-lim-  ited autotrophy. Not much is known about the regulation strategies employed by archaea to control gene transcription, but between different forms of chemolithoautotrophy (reduced metals, H 2 , etc.) and heterotrophy, M. sedula can utilize a broad range of metabolic substrates for growth.
The missing step in the 3HP/4HB pathway has been the acyl-CoA synthetase that utilizes 4HB. Previous attempts to identify the gene that encodes this enzyme were unsuccessful, and the candidate enzymes had no activity on 4HB (9). In this work, two previously unexamined synthetases from M. sedula, consistent with the new transcriptomic evidence, were recombinantly produced and characterized. Both Msed_0394 and Msed_0406 showed activity on 4HB as well as other small organic acids. Based on the lack of other synthetase candidates suggested by the transcriptomic analysis and previous biochemical evidence ruling out Msed_1422 and Msed_1291, we conclude that one or both of these enzymes are necessary for autotrophic growth in M. sedula.
Acetyl-CoA synthetases belong to the class I superfamily of adenylate-forming enzymes that includes acyl-and aryl-CoA synthetases, the adenylation domains of nonribosomal peptide synthetases, and firefly luciferase (23). These enzymes use a two-step mechanism in which first an acyl-AMP intermediate is formed (with release of pyrophosphate) followed by displacement of AMP by CoA (24). Most acetyl-CoA synthetases have a limited substrate range. Archaeal acyl-CoA synthetases, which form a phylogenetic cluster distinct from other bacterial subgroups (25), have been reported to exhibit broader substrate preferences. The acetyl-CoA synthetase from Pyrobaculum aerophilum can work on acetate, propionate, butyrate, and isobutyrate (26); another acetyl-CoA synthetase from Archaeoglobus fulgidus was active on acetate, propionate, and butyrate (27). Both Msed_0394 and Msed_0406 were found to have activity on a broad range of small organic acid substrates of up to five carbons in length.
Activity of both purified Msed_0394 and Msed_0406 on 4HB was well above the reported activity measured in autotrophic cell extract (0.3 mol min Ϫ1 mg Ϫ1 ) (4). It appears that Msed_0406 is primarily a promiscuous propionate-CoA synthetase. Msed_0394, by contrast, has nearly equal levels of activity on acetate, propionate, and 4-HB. Although the overall activity for Msed_0394 is lower by comparison, when taking into account the different substrate specificities, this enzyme shows a preference for C5-C6 linear unsubstituted organic acids. By comparison, the homologous 4-HB-CoA synthetase from T. neutrophilus (Tneu_0420), an anaerobic archaeon that contains the DC/4HB carbon fixation cycle, was recombinantly produced and shown to have maximal activity on 4HB, followed by crotonate, acetate, 3HP, and 3HB (9). The reported K m value for Tneu_0420 was about 3-fold lower than that found for Msed_0406 (700 M versus 2000 M), with comparable activity (1.6 versus 1.8 mol min Ϫ1 mg Ϫ1 ), which suggests that the catalytic activities on 4HB are also comparable.
It is likely that Msed_0406 is more effective at catalyzing the ligation of CoA to 4HB in vivo than Msed_0394. Perhaps these enzymes have evolved from highly specific acetate/propionate synthetases to be sufficient for catalyzing the necessary reaction on 4HB for the 3HP/4HB fixation cycle. It is not clear why two synthetases would be required, or whether both of them are necessary for autotrophic growth. However, they are so far the only ligases in M. sedula that have been shown to activate 4HB with CoA.
Genes with high homology to Msed_0394 and Msed_0406 exist in the genome of the closely related M. cuprina (67 and 73% amino acid identity, respectively), but it is less clear whether homologs exist in the genomes of other Sulfolobales, such as the Sulfolobus and Acidianus spp. Members of the acyladenylate-forming enzyme family may share little identity or similarity in amino acid sequence apart from a few highly conserved core motifs (27). There are homologs of Msed_0406 in other species of Sulfolobales that have 30 -35% identity and one homolog in S. acidocaldarius with 61% identity. But the effort to find the M. sedula 4HB-CoA synthetase has shown that substrate specificity cannot be inferred from amino acid sequence homology alone. However, the low homology of the M. sedula 4HB-CoA synthetase gene does stand out among all the other genes in the 3HP/4HB cycle, which have distinct homologs in Sulfolobus spp. that range from 50 to 80% identity.
Because 4HB is a metabolite unique to butyrate metabolism (28), including ␥-aminobutyrate fermentation (29) and polyhydroxyalkanoate production (30), it is unlikely to have any other role in crenarchaeal metabolism outside of carbon fixation. Recent work with metabolic flux analysis has shown there is another exit route for carbon flux from the cycle through succinyl-CoA to succinate (10). In this study the authors estimate that two-thirds of the cycle carbon flux passes to succinate via succinyl-CoA or succinic semialdehyde, although one-third of the cycle carbon flux passes through the latter part of the cycle (via 4HB) to regenerate acetyl-CoA. Of course, this flux distribution may be highly dependent on growth conditions and could shift more to the 4HB branch depending on substrate availability. It is clear that all members of the Sulfolobales order have a homolog for 4hbd and therefore should have a complete set of enzymes for carbon fixation. However, previous studies have been mixed as to which Sulfolobus spp. are capable of autotrophic growth. Early reports on Sulfolobus acidocaldarius isolates claimed that they could grow chemolithoautotrophically on elemental sulfur (31,32). Subsequent reports claim that neither S. solfataricus nor S. acidocaldarius can grow autotrophically on elemental sulfur alone (33), although it is unclear whether they simply lost the ability to grow chemolithoautotrophically or were selected from what were originally mixed cultures (34). Recent reports have shown autotrophic growth of Sulfolobus metallicus on sulfur and Sulfolobus tokodaii on both sulfur and iron (35). The only other member of the Sulfolobales that has been reported to grow on hydrogen is Acidianus ambivalens, a sulfur-reducing acidophile (36). Genes encoding for hydrogenase and maturation enzymes with homology to M. sedula hydrogenase genes are present in one strain of Sulfolobus islandicus (HVE10/4), but this is predicted to be involved in anaerobic fermentation (37). Clearly, some Sulfolobus spp. must have a functional carbon fixation pathway, but others seem to possess an incomplete or nonfunctional pathway. It may be that the CoA-activating ligase that can operate on 4HB is essential for complete cycle function, and loss of 4HB-CoA synthetase activity renders the carbon fixation cycle inoperable.
To investigate the issue of substrate specificity, de novo structural predictions of M. sedula acyl-CoA synthetases were compared with crystal structures of other known synthetases, including acetyl-CoA synthetase from both S. enterica (17) and Saccharomyces cerevisiae (38) and 4-chlorobenzonate-CoA synthetase from Alcaligenes sp. (39). The structure for ACS from S. enterica revealed that there are four residues that form the acetate binding pocket as follows: Val 310 , Thr 311 , Val 386 , and Trp 414 (17). The conserved tryptophan residue cuts the binding pocket short and precludes activity on longer substrates (Fig. 7). Extensive mutagenesis of binding pocket residues in yeast ACS showed that mutation of W416G was sufficient to lengthen the binding pocket to accommodate C4-C8 organic acids (40). Amino acid sequence alignments show that Msed_1353 has a tryptophan in the same position (Trp 424 ) (Fig. 8) and should therefore only work on acetate and propionate, a fact that has been confirmed biochemically (9). Here, there was some activity with Msed_1353 on 3HP and butyrate but no activity on 4HB. Msed_0394 and Msed_0406 both have a glycine in this position, Gly 333 and Gly 346 , respectively. However, the rest of the genes annotated as acyl-CoA synthetases in M. sedula also have a glycine in this position, so this glycine residue alone is not sufficient to indicate activity on C3-C5 unsubstituted linear organic acids. Both Msed_1422 and Msed_1291 were recombinantly expressed and shown to be inactive on C2-C4 linear organic acids (9).
A mutant of Msed_1353 with a glycine in place of the conserved tryptophan (W424G) was made by site-directed mutagenesis and expressed in E. coli (Msed_1353-Gly 424 ). The native enzyme was active only on acetate and propionate, but the mutant showed activity on 3HP, 4HB, valerate, hexanoate, and even octanoate (Fig. 5). The activity was just as high on C5-C8 substrates as on acetate and propionate but lower on 3HP and 4HB. This suggests that the polar hydroxyl group destabilizes the interaction between the substrate and the residues of the enlarged binding pocket. A similar trend is evident with Msed_0406 (Fig. 4). However, Msed_0394 has nearly equal levels of activity on propionate, butyrate, and 4HB, suggesting that it can stabilize the hydroxyl group on 4HB better than that of 3HP. Similarly, Msed_1456, which catalyzes the ligation of CoA to 3HP in the 3HP/4HB pathway, has equal activity on propionate and 3HP, and therefore it might have residues in the active site that help stabilize the hydroxyl group of 3HP.  Amino acid sequence alignment of active site residues in putative acyl-CoA ligases reveals a conserved glycine (shown in red) except for Msed_1353, which has a tryptophan indicative of acetate-propionate CoA ligases. Alignment was generated using Chimera by superposition of I-TASSER three-dimensional structural models.
In Msed_1456, Val 386 , which makes contacts with the ␥-carbon of the propyl moiety in the S. enterica ACS structure, is replaced with Asn 390 , whose polar amide nitrogen could hydrogen bond with the hydroxyl group of 3HP to stabilize substrate binding. As for Msed_0406, both valine residues in the acetate binding pocket are replaced with alanine (Ala 249 and Ala 321 ) and Thr 311 is replaced with a lysine (Lys 250 ). In Msed_0394, all three of these residues are alanine (Ala 240 , Ala 241 , and Ala 309 ). Potential candidate residues for stabilizing the hydroxyl group of 4HB in Msed_0394 include His 341 and Tyr 338 .
This work helps to close the gaps on the missing piece of the 3HP/4HB pathway in M. sedula. It is still unclear why only certain members of the Sulfolobales operate the 3HP-4HB cycle, but this may reflect the environmental history of specific species. Furthermore, along with other recent successes obtaining recombinant versions of difficult to produce enzymes from the pathway (13), complete characterization of all cycle enzymes is close at hand. The information obtained for cycle function will be invaluable for the creation of a metabolically engineered platform capable of producing chemicals and fuels from carbon dioxide (1).