The Ferredoxin-dependent Conversion of Glyceraldehyde-3-phosphate in the Hyperthermophilic ArchaeonPyrococcus furiosus Represents a Novel Site of Glycolytic Regulation*

The fermentative conversion of glucose in anaerobic hyperthermophilic Archaea is a variant of the classical Embden-Meyerhof pathway found in Bacteria and Eukarya. A major difference of the archaeal glycolytic pathway concerns the conversion of glyceraldehyde-3-phosphate. In the hyperthermophilic archaeonPyrococcus furiosus, this reaction is catalyzed by an unique enzyme, glyceraldehyde-3-phosphate ferredoxin oxidoreductase (GAPOR). Here, we report the isolation, characterization, and transcriptional analysis of the GAPOR-encoding gene. GAPOR is related to a family of ferredoxin-dependent tungsten enzymes in (hyper)thermophilic Archaea and, in addition, to a hypothetical protein in Escherichia coli. Electron paramagnetic resonance analysis of the purified P. furiosus GAPOR protein confirms the anticipated involvement of tungsten in catalysis. During glycolysis in P. furiosus, GAPOR gene expression is induced, whereas the activity of glyceraldehyde-3-phosphate dehydrogenase is repressed. It is discussed that this unprecedented unidirectional reaction couple in the pyrococcal glycolysis and gluconeogenesis gives rise to a novel site of glycolytic regulation that might be widespread among Archaea.

Ever since the classification of living organisms into the three domains of Bacteria, Archaea, and Eukarya (1), comparative studies have revealed that many archaeal features are in fact variations of previously established themes found in bacterial or eukaryal counterparts (2). In some cases, however, the monophylogenetic origin of the Archaea has been corroborated by unique properties that are not encountered among Bacteria or Eukarya (3). Here, we report the characterization of a unique glycolytic enzyme that may represent another typical feature within the archaeal domain.
The pathways of glucose degradation in saccharolytic, hyperthermophilic Archaea have been reported to resemble either the Embden-Meyerhof pathway found in both Bacteria and Eukarya or the bacterial Entner-Doudoroff pathway, albeit with some significant deviations (4,5). The modified Entner-Doudoroff pathway seems to be a main route of glucose catab-olism in (facultative) aerobic hyperthermophilic Archaea, like Sulfolobus acidocaldarius and Thermoplasma acidophila (5,6). A number of variants of the Embden-Meyerhof pathway, on the other hand, have been reported in anaerobic hyperthermophiles in both archaeal kingdoms, the Crenarchaeota and the Euryarchaeota (4,5). In contrast to the ATP-dependent kinases involved in the classical Embden-Meyerhof pathway, ADP-dependent glucokinase and phosphofructokinase activities have been demonstrated in the hyperthermophilic genera belonging to the Euryarchaeota, Pyrococcus and Thermococcus (5,7,8). Likewise, a pyrophosphate-dependent phosphofructokinase has been isolated from the hyperthermophilic crenarchaeote Thermoproteus tenax (9).
An even more dramatic difference in hyperthermophilic Archaea concerns the glycolytic conversion of glyceraldehyde-3phosphate (GAP). 1 In Bacteria and Eukarya, the enzyme-couple glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and 3-phosphoglycerate kinase (PGK) catalyzes the two-step conversion of GAP to 3-phosphoglycerate in a bidirectional manner, i.e., during glycolysis as well as gluconeogenesis. The corresponding enzymes in Archaea, however, have been proposed to be mainly involved in the anabolic gluconeogenesis process (10,11). The glycolytic conversion of GAP in the hyperthermophilic Archaeon Pyrococcus furiosus has recently been reported to be phosphate-independent and is catalyzed by an unprecedented enzyme glyceraldehyde-3-phosphate ferredoxin oxidoreductase (GAPOR) (12).
In the present study, we have isolated the P. furiosus GA-POR-encoding gene (gor). The derived primary structure suggests a structural conservation with a family of oxidoreductases, including the biochemically well characterized tungsten enzyme aldehyde ferredoxin oxidoreductase (AOR) from P. furiosus (13). The anticipated involvement of tungsten in GAPOR catalysis has been analyzed by electron paramagnetic resonance (EPR) analysis. Physiological implications of the possession of an alternative GAP-converting system has been addressed by the analysis of enzyme activities and gene expression of both GAPOR and GAPDH under different conditions. In addition, the gor promoter region has been analyzed in detail. It is concluded that these enzymes constitute a novel site of regulation of glycolysis/gluconeogenesis that may turn out to be characteristic for Archaea.

EXPERIMENTAL PROCEDURES
Organisms, Growth Conditions, Vectors, and Enzymes-P. furiosus (DSM 3638) was grown at 95°C in 330 or 1000 ml of artificial seawater medium, in 1-or 3-liter serum bottles, respectively, as described previously (14). The medium was supplemented with 0.25 g/liter yeast extract, 10 M Na 2 WO 4 , and either 10 mM cellobiose or 40 mM sodium pyruvate. For protein purification from P. furiosus, starch (5 g/liter) was used as substrate. Escherichia coli TOP10FЈ (Stratagene) was grown in LB medium and processed using standard procedures (15). The phagemid pTZ18R was obtained from Amersham Pharmacia Biotech.
Isolation and Sequencing the GAPOR-encoding Gene-Based on the previously published N-terminal amino acid sequence of GAPOR (10,12), two degenerate oligonucleotides were designed with the sequences 5Ј-ATGAARTTYTCNGTNYTNAA and 5Ј-YTCRTTRTGTCKCATDAT-NCC (R ϭ A/G; Y ϭ T/C; K ϭ T/G; D ϭ A/G/T; N ϭ A/C/G/T). P. furiosus genomic DNA (10 ng) was isolated as described previously (16) and used as template in a PCR reaction with these primers to amplify the 5Ј fragment of the GAPOR-encoding gene (gor). The product of the expected size (114 bp) was used to screen a genomic lambda EMBL3 library of P. furiosus (17). A positive clone was selected and subcloned in pTZ18R, and the inserts were analyzed on an automated sequencer (LiCor 4000L) using an Amersham cycle sequence kit with infraredlabeled universal forward and reverse primers. Analysis of the sequences has been performed with the Gapped Blast protein data base search program (24), and alignments have been performed using the Clustal method and the MegAlign program (DNAStar, Inc.) with a PAM250 residue weight table.
RNA Isolation, Induction Experiments, and Northern Analysis-Total RNA was isolated from an actively growing P. furiosus culture in the late log phase by using 4 M guanidinium isothiocyanate and acid phenol extraction (15). For the induction experiments, P. furiosus cells were grown until mid log phase on 30 mM sodium pyruvate. The actual induction of the gor transcription was accomplished by addition of 5 mM cellobiose (final concentration), and RNA was isolated as described (15).
For Northern blot analysis, 10 g of total RNA was separated on a denaturing gel containing 10% formaldehyde (15) and subsequently transferred to a Hybond-Nϩ membrane (Amersham) by capillary blotting according to the manufacturer's instructions. 32 P-Labeled (Amersham) probes were generated by nick translation (15). A 1.2-kb AccI restriction fragment of the gor gene ( Fig. 1) was used for this purpose. The GAPDH-encoding gene (gap) was amplified by PCR using P. furiosus genomic DNA (10 ng) and primers with the following sequence: 5Ј-AAAGGTCGGAATTAATGGATATGGG and 5Ј-TAAGTATGCCGAG-ACTCTTGTTTG, according to the Pyrococcus woesei gap sequence (18), yielding the expected 1.0-kb PCR product. The pyruvate ferredoxin oxidoreductase-encoding gene (por) was obtained with a similar approach, using the following primer pair: 5Ј-GAAGATGATAGAAGTT-CGC and 5Ј-AGCTTCTCTTGCAGCCC, yielding a 530-bp PCR product (19). After verification of the gap and por PCR products by restriction analysis, these fragments were labeled and used as probe (data not shown). Hybridizations were performed using standard protocols (15) at 65°C, and blots were washed in 2 ϫ SSC and 1% SDS at 65°C.
Cell-free Transcription System and Primer Extension-In vitro transcription was performed with the pTZ18R vector carrying a 1.2-kb HindIII-EcoRI insert containing the gor promoter and 0.7 kb of the coding sequence (Fig. 1). The plasmid was digested with EcoRV or HincII ( Fig. 1) for in vitro generation of run-off transcripts (20). For primer extension, a [␥-32 P]ATP end-labeled oligonucleotide (5Ј-TCCG-TAGTCTACAACTCCGAAAATGTCCTCCC) was applied to in vitrogenerated transcript, as well as to isolated total RNA (in vivo), as described (20). Total RNA (1 and 10 g) isolated from P. furiosus cells grown on pyruvate or cellobiose was used as in vivo template.
In cell-free extracts and at different stages of its purification, GAPOR activity was determined spectrophotometrically at 50°C using benzyl viologen as artificial electron acceptor. Stoppered anaerobic quartz cuvettes contained 50 mM Tris (pH 8.0) and 1.0 mM benzyl viologen. In the cuvette, the buffer was slightly prereduced with titanium-citrate before addition of the enzyme. The reaction was started by adding 1.5 mM GAP, and the reduction of benzyl viologen followed at 560 nm (molar absorbance 8.0 mM Ϫ1 ). The activity of GAPDH and pyruvate ferredoxin oxidoreductase have been determined as described (5,10). One unit of enzyme activity is defined as the amount of enzyme required to convert 1 mol of substrate per minute.
Electron Paramagnetic Resonance Analysis-A Bruker 200 D spectrometer with cryogenics, peripheral equipment, and data acquisition/ manipulation facilities was used for EPR spectroscopy (21). The obtained tungsten (W) signal was simulated using the spin Hamiltonian H ϭ ␤.g.SϩS.A.I, with the 183 W hyperfine interaction taken as a perturbation, up to second order, to the electronic Zeeman interaction. Angular dependence of the effective line width tensor was as that of a first-order, colinear hyperfine interaction.

RESULTS
Cloning of the GAPOR-encoding gor Gene-Using P. furiosus DNA as template, a 114-bp PCR product was obtained with primers based on the N terminus of GAPOR. The translated nucleotide sequence of this product exactly matched the 38 N-terminal residues of GAPOR (12). This PCR product was used as a probe to screen a genomic lambda library of P. furiosus (17), and subsequent Southern blot analysis, subcloning, and sequence analysis (not shown) allowed for the identification of the 1959-bp gor gene, coding for the 73,942-Da GAPOR apo-enzyme (Fig. 1).
Primary Structure of P. furiosus GAPOR-Protein data base search with the translated gor sequence revealed significant homology with two recently characterized tungsten proteins from hyperthermophilic Archaea, the AOR from P. furiosus and the formaldehyde ferredoxin oxidoreductase (FOR) from Thermococcus litoralis (22), 15.0 and 15.1% identity at amino acid level, respectively. The recently solved crystal structure of AOR revealed a catalytic site with one tungsten atom that is coordinated by two pterin molecules (W-dipterin site) in close proximity to a [4Fe-4S] cluster (13). Although the overall homology between GAPOR and AOR is relatively low, the residues involved in the coordination of the cofactors in AOR are well conserved in GAPOR, most likely suggesting a similar design of their catalytic sites (Fig. 2).
The P. furiosus GAPOR sequence was found to share the highest level of homology with an unidentified open reading frame from Methanococcus jannaschii (23) (39.0% identity). A dendrogram based on the alignment of the amino acid sequences (Fig. 2) indicates clustering of P. furiosus GAPOR and its homologs from M. jannaschii and of P. furiosus AOR and Thermococcus FOR (Fig. 3). Unlike the rather broad range of substrates that can be oxidized by AOR and FOR, the catalytic capacity of GAPOR seems to be restricted to GAP (12,22).
Regulation of GAPOR and GAPDH-To investigate the roles of GAPOR and GAPDH during glycolysis and gluconeogenesis, we have analyzed their gene expression as well as their enzyme activity during growth on cellobiose and pyruvate, respectively. It was found that the activity of GAPOR is 4 -5-fold higher during growth on cellobiose (glycolysis; 1.8 units⅐mg Ϫ1 ) as compared with pyruvate (gluconeogenesis; 0.4 units⅐mg Ϫ1 ). The modulation of GAPOR activity seemed to be regulated at the transcriptional level, as the gor transcription was significantly stimulated within 10 min after addition of cellobiose to a pyruvate-grown culture of P. furiosus (Fig. 4). This reflects a rapid response to the changed conditions, because under these conditions the generation time of P. furiosus is approximately 60 min. The activity of GAPDH, on the other hand, was significantly lower in extracts of P. furiosus cells grown on cellobiose (0.01 units⅐mg Ϫ1 ) than in extracts of pyruvate-grown cells (0.07 units⅐mg Ϫ1 ). Analysis of gene expression by Northern blot analysis indicated that gap expression is independent of the available carbon source, cellobiose or pyruvate (Fig. 4), indicating that the regulation of the GAPDH activity is at the posttranscriptional level. As a control, we analyzed the transcription of the por gene, coding for the pyruvate ferredoxin oxidoreductase, an enzyme that is involved in the conversion of pyruvate to acetyl-CoA (and eventually to acetate) during growth on pyruvate as well as on cellobiose (17). The results (Fig. 4) indicate that the por gene is not subject to transcription regulation. In addition, relatively little variation of the pyruvate ferredoxin oxidoreductase activity was detected under these conditions, in cellobiose extracts 1.2 units⅐mg Ϫ1 and in pyruvate extracts 0.6 units⅐mg Ϫ1 . Addition of elemental sulfur, a potential electron acceptor in P. furiosus, did not have a detectable effect on the transcription of the gor, gap, and por genes in cellobiose-grown cells (not shown). In Vitro Transcription and Primer Extension of the gor Gene-Because of the observed transcriptional regulation, which in Archaea has received little attention, we analyzed the promoter of the isolated gor gene by mapping its transcription start using primer extension. A single major transcriptional start of the P. furiosus gor gene was identified (Fig. 5A), located 27 bases downstream a putative TATA-box (Fig. 1, TTTTTA) (25) that closely resembles the concensus described for Thermococcales (TTTATA) (26). Two linearized plasmids containing the gor promoter and different parts of its coding region resulted in run-off transcripts of the expected size, 302 and 405 nucleotides (Fig. 5B), in a P. furiosus cell-free transcription system (20). Primer extension studies show that the transcription starts in vivo and in vitro are identical. In addition, an extension product was only observed when RNA from a cellobiose culture was included, confirming the aforementioned cellobiose-dependent transcription activation of the GAPOR gene (Fig. 5A).
Analysis of Purified GAPOR-The purified GAPOR enzyme had a specific activity of 25 units⅐mg Ϫ1 and was purified 35-fold compared with the activity in cell-free extract. The enzyme was approximately 95% pure as judged by gel electrophoresis (not shown). Analysis by gel filtration and SDS-polyacrylamide gel electrophoresis suggested a molecular mass of GAPOR of 56 and 64 kDa, respectively, an apparent underestimation of the actual mass based on the sequence (74 kDa). The final enzyme preparation was concentrated to 2.4 mg⅐ml Ϫ1 and used for EPR analysis. The GAPOR protein as isolated did not exhibit any EPR signals, probably because it was in the oxidized state. When the GAPOR preparation was made anaerobic and then incubated with 10 mM GAP for 3 min at 80°C, a typical signal was obtained when analyzed at 45 K (Fig. 6). An anticipated iron-sulfur signal was not observed, also not after an attempt to further reduce the enzyme by light irradiation in the presence of deazaflavin (27) (not shown). Most likely, the cluster was lost at some stage during the GAPOR purification and was not noticed since the capacity to reduce benzyl viologen was not affected. The obtained spectrum is characteristic for tungsten (W), 5d1 (Fig. 6), specifically with respect to its relatively slow spin-lattice relaxation (no broadening at 45 K), its g values (all less than 2.00), and the presence of hyperfine splitting from the I ϭ 1/2 isotope 183 W (14.4% natural abundance). The experimental spectrum was simulated and shows that the reaction with GAP affords a single W environment (Fig. 6). Subsequent incubation with 10 mM sodium dithionite at ambient temperature did not significantly change the W spectrum (not shown). DISCUSSION The fermentative conversion of glucose in hyperthermophilic Archaea differs from the classical Embden-Meyerhof pathway found in Bacteria and Eukarya. A major difference of the archaeal glycolytic pathway concerns the oxidation of GAP by an unprecedented tungsten enzyme, GAPOR, as first detected in P. furiosus (12). Here, we present a molecular analysis of the GAPOR enzyme from P. furiosus. The corresponding gor gene (Fig. 1) has been isolated, and its deduced primary structure indicates that it is related to the well characterized AOR, a tungsten-containing, ferredoxin-dependent enzyme with a rather broad substrate specificity (13,22). The alignment of GAPOR and AOR indicates that the ligands for the cofactors are well conserved (Fig. 2), suggesting a similar catalytic mechanism with a tungsten pterin in the substrate binding site, and a [4Fe4S] cluster that is involved in transfer of electrons to the physiological acceptor ferredoxin (13).
Purification of the GAPOR enzyme from P. furiosus resulted in a highly active, monomeric enzyme that was analyzed by EPR. After incubation at 80°C with its physiological substrate GAP, the enzyme was reduced, and a typical tungsten EPR signal appeared (Fig. 6). We conclude that tungsten is directly involved in this novel type of glycolytic GAP oxidation. A similar catalytic role of tungsten has previously been reported for AOR, but in this system the physiological substrate remains to be established (13,22).
A 5-fold stimulation of GAPOR activity was detected in lysates of P. furiosus cultures grown on cellobiose, as compared with the activity in lysates of pyruvate cultures. A reversed modulation was observed when GAPDH activity was analyzed, confirming earlier measurements (10). In Northern blot analysis, no significant fluctuation of the GAPDH mRNA could be detected, suggesting that the activity of GAPDH is regulated at a post-transcriptional level (Fig. 4). On the contrary, induction of the gor expression indicates regulation of GAPOR activity at the transcriptional level (Fig. 4). Estimation of the size of the latter signal suggests that the gor gene is transcribed as a monocistronic messenger. Subsequent promoter analysis was performed to gain insight in the molecular basis of the regulation of gor expression. The gor transcription start site has been determined by primer extension, both on isolated total RNA from P. furiosus and on an in vitro generated run-off transcript (Fig. 5). Both experiments indicate that the gor transcription start is located 27 bp downstream a potential TATA box (Fig. 1). The efficiency of in vitro transcription of the gor gene was much lower than that recently reported for the well characterized glutamate dehydrogenase-encoding gene from P. furiosus (20). This might reflect the involvement of some enhancing transcription regulator that is required for optimal gor expression. No obvious cis elements could be identified in the gor promoter region (Fig. 1).
The enzymes GAPDH and PGK catalyze the reversible twostep conversion of GAP to 3-phosphoglycerate and are present in many Bacteria, Archaea, and Eukarya. Well established sites of regulation of the Embden-Meyerhof pathway in Bacteria and Eukarya concern two irreversible steps that introduce FIG. 4. Northern blot analysis of P. furiosus gor, gap, and por transcripts. Cells were grown on pyruvate to which cellobiose was added. RNA was isolated from the culture just before (0) the addition of cellobiose and after 10, 25, 40, and 55 min. An RNA ladder (Life Technologies, Inc.) was used as marker.
specific deviations of the glycolysis and the gluconeogenesis, i.e., the interconversion of fructose-6-phosphate/fructose-1,6bisphosphate and phosphoenol pyruvate/pyruvate. In P. furiosus, the presence of GAPOR in addition to GAPDH/PGK introduces a potential site of regulation at this level of the pathway. Indeed, we have demonstrated that regulation does occur at this level. An analogous situation has recently been described in T. tenax (11). In addition to GAPDH/PGK, this hyperthermophilic archaeon expresses an alternative enzyme that is homologous to the GAPN-type aldehyde dehydrogenase of some Bacteria and photosynthetic Eukarya. The reaction catalyzed by GAPN resembles that of GAPOR: a single-step conversion of GAP to 3-phosphoglycerate that does not result in ATP generation. Unlike the ferredoxin-dependent GAPOR reaction, oxidation of GAP by GAPN is coupled to the reduction of NAD ϩ (11).
Apart from P. furiosus, GAPOR activity has recently been detected in T. litoralis and Desulfurococcus amylolyticus (5); in addition, homologous genes that encode either GAPOR or related tungsten enzymes (AOR, FOR) are present in M. jannaschii (23), Archaeoglobus fulgidus (28), and Pyrodictium occultum. 2 An explanation for the presence of tungsten enzymes in these marine hyperthermophilic Archaea might be the relatively high concentration of tungsten that has been detected in (deep sea/shallow) hydrothermal vents (22).
The tungsten enzymes discussed here constitute a growing family of archaeal aldehyde-converting, ferredoxin-dependent oxidoreductases, which we propose to refer to as the AOR family. Comparative analysis of the P. furiosus GAPOR primary structure and that of its apparent homolog from M. jannaschii indicates that GAPOR constitutes a new, though relatively distant branch within the AOR family (Figs. 2 and 3). Interestingly, a single non-archaeal sequence with a significant homology with the AOR family could be identified. This concerns a hypothetical protein from E. coli in which the ligands for the [4Fe4S] cluster are completely conserved, whereas the residues that are involved in the hydrogen bonding network that holds the W-dipterin in AOR are less conserved (Fig. 2). No additional sequences of other bacterial or eukaryal homologs are available yet, with the possible exception of some partial sequences, including the N terminus of the 2-hydroxycarboxylate viologen oxidoreductase from the enterobacterium Proteus vulgaris (Fig. 2). The latter enzyme is an 80-kDa protein, containing one molybdenum, four iron, and four sulfur atoms, as well as a mononucleotide (pterin) molybdenum cofactor (29). Although tungsten proteins are present in some Bacteria (22), one might speculate that the unidentified E. coli oxidoreductase possesses a molybdopterin cofactor.
The apparent coexistence of alternative systems for the oxi-  1 and 2) and a cellobiose culture (lanes 3 and 4), as well as on in vitro generated transcript (lanes 5 and 6). Obtained products were loaded next to a sequencing reaction (lanes G, A, T, and C) in which the same primer on the noncoding strand of gor was used. B, in vitro transcription of gor gene fragments in a P. furiosus cell-free transcription system. Labeled run-off transcripts of the expected size indicated were obtained.
FIG. 6. W EPR from P. furiosus GAPOR after incubation with the substrate GAP at 80°C. The enzyme preparation was made anaerobic by repeated vacuum/argon cycling of the frozen sample in an EPR tube. Then, 10 mM GAP was added, and the tube was placed under argon in an 80°C water bath for 3 min. After freezing in liquid N 2 , the spectrum EXP was obtained. EPR conditions were as follows: microwave frequency, 9395 MHz; microwave power, 8 mW; modulation frequency, 100 kHz; modulation amplitude, 0.5 mT; temperature, 45 K. Trace SIM is a simulation of the experimental spectrum based on 51 ϫ 51 molecular orientations and using the parameters g x,y,z ϭ 1.829, 1.885, 1.946; 183 W hyperfine splitting A x,y,z ϭ 6, 8, 7 mT; linewidth W x,y,z ϭ 2.2, 1.3, 1.5 mT. dation of GAP and the reduction of 3-phosphoglycerate has hitherto only been found in hyperthermophilic Archaea. In certain Bacteria and higher plants, the NADP ϩ -dependent GAPN enzyme provides an alternative/additional system for reductant supply (30,31). In the hyperthermophilic Archaea, it seems that the GAPDH enzyme is active mainly during gluconeogenesis, whereas either GAPOR or GAPN drives the glycolytic conversion of GAP. With the current knowledge, a thermodynamic advantage of such a variation of the Embden-Meyerhof pathway is not obvious. Apparently, rather than utilizing a glycolytic pathway in which the reducing power of GAP is used for substrate-level phosphorylation, Archaea possess an energetically less efficient glycolysis variant in which the interconversion of glyceraldehyde-3-phosphate and 3-phosphoglycerate developed into a novel regulatory site of the glucose metabolism.