Purification and characterization of a novel ADP-dependent glucokinase from the hyperthermophilic archaeon Pyrococcus furiosus.

Pyrococcus furiosus uses a modified Embden-Meyerhof pathway during growth on poly- or disaccharides. Instead of the usual ATP-dependent glucokinase, this pathway involves a novel ADP-dependent (AMP-forming) glucokinase. The level of this enzyme and some other glycolytic enzymes appeared to be closely regulated by the substrate. Growth on cellobiose resulted in a high specific activity of 0.96 units mg-1, whereas on pyruvate a 10-fold lower activity was found. The ADP-dependent glucokinase was purified 1350-fold to homogeneity. The oxygen-stable enzyme had a molecular mass of 93 kDa and was composed of two identical subunits (47 kDa). The glucokinase was highly specific for ADP, which could not be replaced by ATP, phosphoenolpyruvate, GDP, PPi, or polyphosphate. D-Glucose could be replaced only by 2-deoxy-D-glucose, albeit with a low efficiency. The Km values for D-glucose and ADP were 0.73 and 0.033 mM, respectively. An optimum temperature of 105 degrees C and a half-life of 220 min at 100 degrees C are in agreement with the requirements of this hyperthermophilic organism. The properties of the glucokinase are compared to those of less thermoactive gluco/hexokinases.

detected in cell-free extracts (5). Remarkably, two novel sugar kinases were recently discovered that required ADP instead of ATP (9). In contrast to the key enzymes of the pyroglycolysis, the specific activities of both kinases were sufficiently high to envisage a major catabolic role. Instead of a classical NAD-dependent glyceraldehyde-3-phosphate dehydrogenase, P. furiosus was recently shown to harbor a novel tungsten-containing glyceraldehyde-3-phosphate:ferredoxin oxidoreductase (11). The presence of an enzyme that converts glyceraldehyde-3phosphate instead of glyceraldehyde also substantiates the operation of a modified Embden-Meyerhof pathway instead of the pyroglycolysis. The discovery of the novel type of kinases and the tungsten proteins in P. furiosus supports the idea that life at elevated temperatures may involve different metabolic steps as a result of an altered biochemistry or a decreased stability of biomolecules.
In this paper, we describe the purification and characterization of the novel ADP-dependent glucokinase. The results show that ADP-dependent conversion of glucose is not only found in crude P. furiosus extracts but is catalyzed by a single enzyme that shows a characteristic specificity. The properties of the Pyrococcus enzyme are compared to those of glucokinases from less thermophilic sources.
Mass culturing (200 liters) was performed on the same medium except that Na 2 S was omitted, the fermenter was sparged with N 2 , and potato starch was used as substrate (5 g/liter).
Preparation of Cell-free Extracts-To determine the effect of the substrate on enzyme levels, cell extracts were prepared aerobically from * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. the 50-ml cultures. The contents of each bottle was centrifuged for 20 min at 22,800 ϫ g. The supernatant was discarded, and the cell pellet was resuspended in 1 ml of distilled water. The cell suspension was sonicated three times for 30 s. Cell debris were removed by centrifugation, and the supernatant was used as cell-free extract.
For use in enzyme purifications, cells and cell extracts were handled aerobically. Cells were suspended in 50 mM Tris/Cl Ϫ buffer (pH 7.8) (0.5 g of cells/ml of buffer) containing DNase (10 g/ml), and the suspension was passed twice through a French press at 138 MPa. Cell debris were removed by centrifugation for 1 h at 100,000 ϫ g. The cell extract, containing 35-45 mg protein/ml, was stored at Ϫ20°C until use.
Protein was determined with Coomassie Brilliant Blue G250 as described by Bradford (12), using bovine serum albumin as a standard. Occasionally, a modified more sensitive Coomassie Brilliant Blue G250 method was used as described by Löffler and Kunze (13).
Determination of Enzyme Activity-The enzyme assays were performed aerobically in stoppered 1-ml quartz cuvettes as described before (9). Specific enzyme activities were calculated from initial rates and expressed in units mg Ϫ1 protein. 1 unit was defined as that amount of enzyme required to convert 1 mol of glucose per min.
ADP-dependent glucokinase was determined by measuring the formation of NADPH in a coupled assay with yeast glucose-6-phosphate dehydrogenase. The assay was performed at 50°C. At this temperature, the yeast enzyme remained active, and the Pyrococcus enzyme was sufficiently active to measure its activity. The assay mixture contained 100 mM Tris/Cl Ϫ , pH 7.8, 10 mM MgCl 2 , 0.5 mM NADP, 15 mM D-glucose, 2 mM ADP, 0.35 units of D-glucose-6-phosphate dehydrogenase, and 5-50 l of enzyme preparation. The absorbance of NADPH was followed at 334 nm (⑀ ϭ 6.18 mM Ϫ1 cm Ϫ1 ). Care was taken that the activity of the auxiliary enzyme was always in excess of the glucokinase activity.
Substrate Specificity-The substrate specificity was tested using purified glucokinase. The use of 2-deoxy-D-glucose and D-galactose as possible substrates for the glucokinase was tested using the standard enzyme assay because the auxiliary enzyme from yeast is also able to use galactose-6-phosphate. For the determination of D-fructose as a possible substrate, phosphoglucose isomerase (1.4 units) was added to the standard assay mixture. D-Mannose was tested by adding phosphomannose isomerase (0.6 units) and phosphoglucose isomerase (1.4 units) as auxiliary enzymes. Proper functioning of the assay was tested using yeast glucokinase instead of P. furiosus glucokinase. All sugars were tested at a concentration of 15 mM. As possible phosphoryl group donor, ATP, GDP, PP i , phosphoenolpyruvate (each 2 mM), and polyphosphate (sodium phosphate glass, type 35; 0.2 g liter Ϫ1 ) were used instead of ADP. The divalent cation requirement was tested by adding 10 mM of MgCl 2 , MnCl 2 , CaCl 2 , ZnCl 2 , or CoCl 2 to the standard assay mixture containing 2 mM disodium EDTA.
Purification of the Glucokinase-All purification steps were performed without protection against oxygen. To prevent microbial contamination, all buffers contained 0.02% sodium azide. The standard buffer used was 100 mM Tris/Cl Ϫ , pH 7.8 (buffer A). Cell-free extract (50 ml) was first brought to 58% ammonium sulfate saturation (2 h, 0°C). After centrifugation, the pellet fraction was discarded, and the supernatant was loaded on a phenyl-Sepharose 6 Fast Flow (high sub) column (3.2 ϫ 4 cm), equilibrated in buffer A containing 2.5 M ammonium sulfate. The column was developed using two successive linear gradients from 2.5 to 0.75 M (NH 4 ) 2 SO 4 (120 ml) and from 0.75 to 0 M (NH 4 ) 2 SO 4 (360 ml). The glucokinase eluted at 0.5 M (NH 4 ) 2 SO 4 . Active fractions were pooled and desalted by ultrafiltration (Amicon YM-5) using buffer A, supplemented with 5 mM CHAPS. 1 The desalted glucokinase pool was applied to a Mono-Q HR 5/5 column equilibrated in buffer A containing 1 mM CHAPS. The glucokinase eluted during a 60-ml linear gradient (0 -0.5 M NaCl) at 0.18 M NaCl. Active fractions were pooled, and CHAPS was added up to 5 mM. The enzyme pool was loaded on a hydroxy apatite column (2 ϫ 20 cm) equilibrated in 1 mM potassium phosphate buffer (pH 6.8) containing 1 mM CHAPS. The column was developed using two successive linear gradients from 0 to 0.25 M potassium phosphate (140 ml) and from 0.25 to 0.5 M potassium phosphate (50 ml). Glucokinase-containing fractions eluted at 0.35 M potassium phosphate. The buffer of the active pool was exchanged for 50 mM potassium phosphate buffer (pH 7.0) containing 1.7 M (NH 4 ) 2 SO 4 by Amicon YM-5 ultrafiltration. The concentrated pool was loaded on a phenyl-Superose HR 5/5 column equilibrated in the same buffer. Glucokinase eluted from the column at 1.2 M (NH 4 ) 2 SO 4 during a linear 30-ml gradient from 1.7 to 0 M (NH 4 ) 2 SO 4 . Active fractions were combined, and the buffer was exchanged for 50 mM Pipes/Cl Ϫ (pH 6.2) by ultrafiltration. The enzyme pool was applied to a Mono-Q HR 5/5 column, equilibrated in 50 mM Pipes/Cl Ϫ (pH 6.2). The glucokinase eluted during a linear 40-ml gradient (0 -1 M NaCl) at 0.2 M NaCl. The enzyme pool was desalted and concentrated with Macrosep (30K) concentrators (Filtron). Complete purification was accomplished by continuous elution electrophoresis on a Prep Cell apparatus (Bio-Rad). A 1-ml sample from the concentrated enzyme pool was loaded on the gel (8% acrylamide), and electrophoresis of proteins was performed according to the instructions of the manufacturer. Protein was eluted from the gel in Tris (25 mM), glycine (192 mM) buffer, pH 8.3. Only those fractions that gave one single band on a native gel were combined.
Purity of the enzyme was checked by native and denaturing SDS-PAGE as described before (6). For determination of the subunit composition by SDS-PAGE, protein samples were diluted in sample buffer, containing 2% SDS (w/v) and 5% 2-mercaptoethanol, and subsequently heated at 100°C. In some cases, 2-mercaptoethanol was omitted from the sample buffer, and the sample was not heated. Silver staining was performed using the reagent kit from E. Merck (Darmstadt, Germany). Activity staining was performed on native PAGE gels by coupling the glucokinase activity to the reduction of nitro blue tetrazolium. Therefore, the gel was incubated at 37°C in the dark for 30 min in a staining mix with the following composition: 100 mM Tris/Cl Ϫ , pH 7.8, 0.001% phenazin methosulfate, 0.035% nitro blue tetrazolium, 15 mM MgCl 2 , 0.5 mM NADP, 3 mM ADP, 15 mM D-glucose, 185 mM NaCl, and D-glucose-6-phosphate dehydrogenase (1.75 units).
pH Optimum-The pH optimum was determined at 50°C in 200 mM Tris/maleate buffer over the pH range 5.5-9.0. Care was taken that the auxiliary enzyme was not limiting at the various pH values.
Temperature Effect and Thermostability-The effect of temperature on the activity was determined by incubating an appropriate amount of purified enzyme in 1-ml crimp-sealed vials containing 200 mM Tris/ maleate buffer (pH 8.5), 20 mM MgCl 2 , and 20 mM D-glucose. The vials were submerged in an oil bath at temperatures varying from 30 to 110°C, and the enzyme reaction was started by injecting 10 l of 100 mM ADP. After 15-30 min, the reaction was stopped by putting the vials on ice, and the amount of glucose-6-phosphate formed was determined spectrophotometrically by measuring the reduction of NADP (334 nm) in an assay with glucose-6-phosphate dehydrogenase.
Thermostability of the glucokinase was determined by incubating purified glucokinase in 200 mM Tris/maleate buffer (pH 8.5) at 100°C in crimp-sealed vials, submerged in an oil bath. At certain time intervals, 50-l aliquots were withdrawn and analyzed for activity in the standard assay.
Kinetic Analysis-Kinetic parameters were determined at 50°C by varying the concentration of glucose or ADP in the presence of a saturating concentration of ADP (4 mM) or glucose (15 mM), respectively. The 980-fold purified enzyme was used for these determinations.
N-terminal Amino Acid Sequence Analysis-The N-terminal sequence of the purified glucokinase was determined according to the Edman degradation method and was performed on two independent enzyme preparations by the sequencing facilities of Eurosequence (Groningen, The Netherlands) and SON (Leiden, The Netherlands). Because of the presence of Tris and glycine in the final preparations, the samples were subjected to PAGE and electroblotted on a polyvinylidene difluoride membrane prior to analysis.

Glucokinase Levels on Different Carbon
Sources-To discern the inducible or constitutive nature of the ADP-dependent glucokinase, cells of P. furiosus were grown on various carbon sources (Fig. 1). The level of the ADP-dependent glucokinase was found to vary from almost zero (0.003 units mg Ϫ1 ) during growth on peptone to 0.96 units mg Ϫ1 during growth on cellobiose. The non-glycolytic substrate pyruvate showed a relatively low glucokinase activity (0.074 units mg Ϫ1 ). Starch and maltose gave intermediate values of 0.43 and 0.49 units mg Ϫ1 , respectively. The activity of the subsequent enzymes in the glycolysis, viz. phosphoglucose isomerase and the ADP-dependent phosphofructokinase, were also determined. The level of both enzymes appeared to vary in a similar way as the glucokinase, i.e. the highest activity on cellobiose and lower activities on starch, maltose, and pyruvate.
Purification of the Glucokinase-The glucokinase was purified aerobically because no enzyme activity was lost upon storage of cell-free extracts at 4°C under air. After fractionation of the broken cell suspension at 100,000 ϫ g, most of the total amount of activity (92%) was recovered in the supernatant, indicating that the enzyme is located in the cytoplasm. During initial purification attempts, PAGE showed that the enzyme copurified with several other proteins, suggesting that the enzyme adhered to these proteins. Therefore, the zwitterionic detergent CHAPS was added to the buffers, which did not affect the activity negatively.
The use of affinity chromatographic techniques, like ADPagarose (either ribose-linked or N6-linked) or various dyeligand-agaroses (Dyematrex screening kit, Amicon), was unsuccessful because the enzyme did not bind to any of the ligands, even in the presence of 10 mM MgCl 2 . Therefore, a series of seven sequential purification steps were required to obtain a homogeneous preparation as judged by a silver-stained PAGE gel (Table  I). The colorless enzyme was 1346-fold purified with 2.1% recovery and showed a specific activity of 307 units mg Ϫ1 at 50°C. The identity of the band was confirmed by activity staining.
Physical Properties-The molecular mass of the native enzyme as determined by PAGE at various acrylamide concentrations was 93 kDa (not shown). SDS-PAGE of the 980-fold purified protein gave a single band of 47 kDa, irrespective the time of heating in sample buffer or the presence of 2-mercaptoethanol (Fig. 2). Apparently, the 93-kDa native enzyme easily disintegrates into two identical 47-kDa subunits. This result is in accordance with the immediate and complete inhibition of glucokinase activity that was found upon addition of 5 mM SDS (data not shown).
Catalytic Properties-The ADP-dependent glucokinase exhibited a high activity (Ͼ65% of maximum) between pH 6 and 9, with an optimum at pH 7.5. As all other kinases, the enzyme required divalent cations for activity (Table II). MgCl 2 was most effective, followed by MnCl 2 , which resulted in 77% of the activity found with MgCl 2 . No activity was found in the absence of divalent cations in the presence of EDTA. With respect to the phosphoryl group donor, the glucokinase was highly specific for ADP. ATP, GDP, phosphoenolpyruvate, PP i , or polyphosphate were unable to replace ADP (Table II). The glucokinase was also rather specific for the type of sugar. D-Fructose, D-mannose, and D-galactose could not be phosphorylated, and only 2-deoxy-D-glucose was able to replace glucose to a limited (9.2%) extent (Table II).
Kinetic Parameters-Michaelis-Menten constants were determined according to Lineweaver-Burk. A K m value of 0.73 Ϯ 0.06 and 0.033 Ϯ 0.003 mM was found for glucose and ADP, respectively. Apparent V max values were 249 Ϯ 18 and 194 Ϯ 15 units mg Ϫ1 for glucose and ADP, respectively.
Thermostability and Temperature Optimum-The thermostability of the purified glucokinase was determined at 100°C and 110°C. At 110°C, all activity was lost after 30 min of incubation. Addition of MgADP, glucose, or both did not affect the stability. Therefore, no attempts were made to determine the half-life of the enzyme at this temperature. At 100°C, however, the glucokinase was remarkably stable. Inactivation followed first-order kinetics with a half-life value of 220 min (not shown).
The temperature dependence of the activity is shown in Fig.  3. The optimum temperature was found at 105°C (15-min incubation period). Because of a rapid denaturation above this temperature, this optimum value may increase or decrease depending on the time of incubation (shorter or longer incuba-  tion time, respectively). An Arrhenius plot of the data (Fig. 3,  inset) showed a breakpoint at 60°C, resulting in activation energy values of 54.3 kJ mol Ϫ1 between 30 and 60°C and 37.4 kJ mol Ϫ1 between 60 and 105°C. N-terminal Amino Acid Sequence Analysis-Two independent attempts to determine the N-terminal sequence did not give an unambiguous and ungapped sequence, indicating that the N terminus may be blocked. Those amino acids that were identified as identical by both analyses gave the following sequence (first 10 residues, X ϭ ambiguous residue): NH 2 , MTXEX-LYKN(I/A). This sequence did not show similarity with any sequence given in the SWISSPROT data base. DISCUSSION P. furiosus has recently been shown to utilize a modified Embden-Meyerhof pathway, which involves a glucokinase and a phosphofructokinase that are both ADP-dependent (9). Here, the ADP-dependent glucokinase was purified and characterized. Cell-free extracts of P. furiosus contained high levels of this enzyme, especially when the organism was grown on cellobiose. This high level of the glucokinase and also of the phosphoglucose isomerase and the phosphofructokinase in cellobiose-grown cells as compared to maltose-grown cells clearly shows that at least the first steps of the glycolysis are closely regulated in this hyperthermophilic Archaeon. This also follows from the low activity of these enzymes on pyruvate-and peptone-grown cells.
The cytoplasmic and oxygen-stable glucokinase was purified more than 1000-fold to homogeneity. Thus, the glucokinase constitutes less than 0.1% of the total cellular protein. This seems a rather low value for such a key enzyme. However, using the experimentally determined relationship between activity and temperature, the specific activity at 100°C amounts to 2,233 units mg Ϫ1 (k cat ϭ 3,500 s Ϫ1 ), which is the highest reported (Table III). Moreover, it has been calculated before that the specific activity in cell-free extracts is more than satisfactory to sustain the glucose flux (9).
The ADP-dependent glucokinase had a native molecular mass of approximately 93 kDa and consisted of two identical subunits of approximately 47 kDa. This ␣ 2 composition is observed also for bacterial glucokinases and eukaryotic hexokinases, but it differs from the eukaryotic glucokinases (bakers' yeast, rat liver), which show a monomeric structure (Table III). Furthermore, the P. furiosus glucokinase differs from most bacterial enzymes by its native molecular mass, which is about twice the usual size of about 50 kDa. In this respect, it resem-  bles more the hexokinases from Eukarya. All the glucokinases described in detail (with sequence information) can be grouped into three evolutionary disconnected clusters, i.e. the mammalian glucokinases, the yeast glucokinases, and the bacterial glucokinases (25). However, based on the partial N-terminal sequence obtained for the P. furiosus enzyme, the latter does not group within any of these. Also, no similarity was found within other kinase families (25). Nevertheless, many of these kinases, if not all, exhibit a striking structural feature; each subunit contains two lobes separated by a cleft (26). Upon binding of the substrate the two lobes come together. Whether the Pyrococcus enzyme also shows this substrate-induced cleft closing remains to be elucidated. A comparison with respect to the substrate specificity is difficult since for many enzymes these data are incomplete. Nevertheless, the glucokinase from P. furiosus showed a high specificity for the type of sugar as well as the phosphoryl group donor. From previous NMR experiments, it can be concluded also that the glucokinase is able to use both the ␣and ␤-anomer of D-glucose (9). Thus, the enzyme is a true glucokinase and highly specific for ADP. The specificity for ADP is also reflected in the high affinity found for this compound, i.e. the K m value of 0.033 mM is the among the lowest reported (15)(16)(17)(18)(19)(20)(21)(22)(23)(24). The reason for the ADP dependence may lie in the ability to activate sugars at conditions of low energy charge, e.g. after a period of starvation. The ATP level is then probably very low, and the relatively high ADP-level may still enable the phosphorylation of sugars.
The dependence of the activity on temperature showed an optimum at 105°C, which is in accordance with the requirements of the organism. The breakpoint in the Arrhenius plot at 60°C has been observed before for other thermophilic enzymes and may reflect a conformational change of the protein (27,28).
The purified glucokinase showed a high thermostability (half-life of 220 min) at the physiological growth optimum of 100°C. As has been found for most other enzymes from (hyper)thermophiles (29), this thermostability is apparently not determined by extrinsic factors, like high salt concentrations or specific compatible solutes.
It is evident that P. furiosus harbors an exceptional ADPspecific glucokinase, perfectly fit for catalyzing the phosphorylation of glucose at elevated temperatures. Up to now, the ADP dependence is unique for P. furiosus. Other sugar-converting thermophiles examined thus far use either the classical Embden-Meyerhof pathway (30, 31) or a non-phosphorylated version of the Entner-Doudoroff pathway (32,33). Further re-search is focussing on the distribution of this type of enzyme among other (hyper)thermophiles and on the determination of their primary structure, which both will shed light on the evolutionary position of the ADP-dependent glucokinase. Moreover, the function of the ADP-dependent enzymes in relation to the bioenergetics of the organism is under present investigation.