The Preparation and Characterization of Pure Rat Liver Glucokinase*

Abstract Glucokinase from rat liver has been purified over 10,000-fold and appears homogeneous by the criterion of sodium dodecyl sulfate gel electrophoresis. The enzymatically active protein represents 0.007 % of the total liver protein and possesses a specific activity of 80 units per mg for purified enzyme. The use of potassium phosphate buffer gradients greatly enhances enzyme resolution during ion-exchange chromatographic procedures. In addition, use of substrate and sulfhydryl protecting reagents in all buffers prevents significant losses of activity. To our knowledge this is the highest degree of purity ever achieved for a mammalian hepatic ATP:d-hexose-6-phosphotransferase. Rat liver glucokinase has been characterized with respect to stability, Km, Vmax, and phosphorylation coefficient toward several substrates as well as behavior in the presence of several inhibitors. Spectral analysis fails to reveal the presence of a prosthetic group. The molecular weight determined by sodium dodecyl sulfate acrylamide gel electrophoresis and Sephadex G-100 gel filtration is 53,000 and 57,000, respectively. The turnover number is calculated to be 4,346 moles of d-glucose phosphorylated by ATP per min per mole of enzyme at 25°.


Glucokinase
from rat liver has been purified over lO,OOOfold and appears homogeneous by the criterion of sodium dodecyl sulfate gel electrophoresis.
The enzymatically active protein represents 0.007% of the total liver protein and possesses a specific activity of 80 units per mg for purified enzyme.
The use of potassium phosphate buffer gradients greatly enhances enzyme resolution during ion-exchange chromatographic procedures.
In addition, use of substrate and sulfhydryl protecting reagents in all buffers prevents significant losses of activity.
To our knowledge this is the highest degree of purity ever achieved for a mammalian hepatic ATP:D-hexose-6-phosphotransferase.
Rat liver glucokinase has been characterized with respect to stability, K,, V,,,, and phosphorylation coefficient toward several substrates as well as behavior in the presence of several inhibitors.
Spectral analysis fails to reveal the presence of a prosthetic group. The molecular weight determined by sodium dodecyl sulfate acrylamide gel electrophoresis and Sephadex G-100 gel filtration is 53,000 and 57,000, respectively.
The turnover number is calculated to be 4,346 moles of D-glucose phosphorylated by ATP per min per mole of enzyme at 25".
Several attempts have been made to obtain purified rat liver glucokinase (EC 2.7.1.2) (l-3). This enzyme is of particular interest not only because of its key role in metabolism but also because of its response to hormones and diet (4) as Iyell as its deficiency in certain hepatomas (5). Of the reported purifications, none gives a protocol which results in a homogeneous preparation, although the recent report by Pilkis (3) gives a procedure for obtaining enzyme enriched by a factor of almost 3,000. However, as judged by the criterion of disc gel electrophoresis, this preparation still contains several other proteins.
The difficulties encountered in purification are derived from the instability of the crude prepartion and the small amount of glucokinase in rat liver.
Our initial interest in glucokinase was stimulated by the possi-* This work was supported in part by Training Grant CA-07175 and Project Grant CA-07175 from the National Cancer Institute.
$ National Institutes of Health Postdoctoral Fellow (lF02-CA-53168-o). bility of obtaining fluorescent antibodies against the enzyme in order to study its occurrence in several types of hepatic cells. Accordingly, we have developed a procedure for the purification of glucokinase to apparent homogeneity.
In this paper we report the purification procedure and partial characterization.
To our knowledge this is the highest degree of purity ever achieved for a mammalian liver ATI' : I>-hesose-6-phosphotransferase.
Actin, purified from the slime mold Physarum Assay of Glucoki?lase-Glucokinase activity was assayed at 25" by recording the increase in absorbance at 340 nm of a reaction mixture containing: 50 mM glycylglycine (pH 7.5), 100 mM KCl, 7.5 mM MgSOd, 0.5 mM NADP, 5 rnhr ATP (pH 7.0), 0.9 unit of glucose-6-P dehydrogenase, 0.03 unit of 6.P-gluconate dehydrogenase, 100 mM glucose, and enzyme in a total volume of 3.0 ml. When hexokinase having a low Zcm (EC 2.7.1.1) was assayed, the glucose was omitted in the reaction mixture since the enzyme preparation contained sufficient glucose (approximately 0.50 mM) to saturate the hexokinases but well below that necessary to demonstrate glucokinase activity.
Hexokinase activity and glucose dehydrogenase activity were subtracted from the total activity observed with 100 mM glucose as substrate to give the glucokinase activity.
All reaction mixtures were incubated at 25" for 5 min prior to assay. One unit of glucokinase activity represents 1 pmole of glucose-6-P formed from glucose and ATP per min at 25" (1).
When substrate specificity was studied, the activity was measured by recording at 25" the decrease in absorbance at 340 nm of a mixture containing 100 mM triethanolamine-HCl (pH 7.5), 100 mM KCl, 7.5 mM MgS04, 0.75 mM P-enolpyruvate, 5 mM ATP, 0.25 mM NADH, 40 units of lactate dehydrogenase, 3 units of pyruvate kinase, and substrate.
The mixture was incubated for 10 min at 25" before initiating the reaction by the addition of enzyme to a final volume of 2.5 ml (1).

Protein
Determination-When protein concentrations were greater than 100 pg per ml and phosphate concentration below 0.1 M, protein was determined by measurement of the absorbance at 280 nm assuming 1 mg of protein corresponded to an optical density of 1. At low protein concentrations, or when column fractions containing increasing amounts of potassium phosphate were analyzed, the fluorescamine procedure (G) was used, with bovine serum albumin as the standard.

SDS Gel Acrylamide
Eleclrophoresis-Analytical SDS gel electrophoresis was performed according to the procedures of Laemmli

RESULTS
Enzyme Purification-Sprague-Dawley rats weighing between 250 and 300 g were maintained on a 60% carbohydrate diet ad Zibitum for at least 1 week prior t.o sacrifice.
All procedures were carried out at 4'. The crude homogenate was immediately centrifuged for 1 hour at 105,000 x g. The resulting high speed super- Enzyme was adsorbed onto a column (3 X GO cm) equilibrated with buffer consisting of 0.1 M potassium phosphate, 1 mM MgSOd, 1 mM EDTA, 50 mM glucose, and 10 mM mercaptoethanol, pH 7.0. Enzyme was eluted by a linear gradient of potassium phosphate prepared from 1 liter of 0.1 M potassium phosphate, pH 7.0, in the mixing chamber and 1 liter of 0.5 M potassium phosphate, pH 7.0, in the reservoir chamber each with 1 mM MgS04, 1 mM EDTA, 50 mM glucose, and 10 mM mercaptoethanol.
Eighteen-milliliter fractions were collected. The inset shows a SDS acrylamide gel of Fraction 92 stained for protein. Approximately 15 rg of protein were run. Glucokinase, as determined from the Rp value of the purified enzyme (Fig. 4), is indicated by the arrow. The column (3.5 X 100 cm) was equilibrated and developed with buffer consisting of 0.1 M potassium phosphate, 0.1 M KCI, 5 mM EDTA, 5 mM MgEOr, 2 mM dithiothreitol, and 0.5 M glucose, pH 7.0. Eighteen-milliliter fractions were collected. The inset shows an SDS acrylamide gel of Fraction 34 stained for protein.
Approximately 40 rg of protein were run. Glucokinase, as determined from the RF value of the purified enzyme ( Fig. 4), is indicated by the arrow.
The column (1 X 100 cm) was equilibrated and developed with the same buffer as used in Fig. 2 except that 2 liters of each gradient buffer were used. Six-milliliter fractions were collected. The inset shows a SDS acrylamide gel of purified glucokinase from Fraction 151 stained for protein.
Approximately 6 pg of protein were run. natant was then brought to 0.42 saturation in ammonium sulfate by the slow addition of a saturated solution of ammonium sulfate in Buffer 1' previously adjusted to pH 7.0 with concentrated ammonium hydroxide.
The supernatant obtained after ccntrifugation at 22,000 x g (15 min) was then made 0.68 saturated in ammonium sulfate. The precipitate was collected by centrifugation and dissolved in a minimum amount (approximately 150 ml) of Buffer 1'. This solution was then divided in half and subjected to gel filtration on two Sephadex G-100 columns (Fig. 1). The fractions (57 to 72) containing more than 0.05 units per ml were combined and treated with approximately 100 ml of packed DEAE-Sephades equilibrated with Buffer 1'. The suspension was stirred for 20 min, then allowed to settle, and the supernatant was drawn off. Assay of the supernatant generally showed that less than 3 units were unadsorbed to the gel. The gel was poured into a glass column (5 x 30 cm) and washed with Buffer 1' until the eluate was free of protein.
The column then was developed with an increasing gradient of potassium phosphate obtained with 1 liter of Buffer 1' in the mixing chamber and 1 liter of 0.5 M potassium phosphate, 1 rnhf MgSOd, 1 mM EDTA, 50 mM glucose, and 10 mM mercaptoethanol, pII 7.0, in the reservoir chamber.
The fractions with an activity over 0.20 units per ml were combined, concentrated in several batches to a final volume of 200 ml with an Amicon model 202 Ultrafiltration Cell equipped with a PM-30 membrane, then dialyzed against two l-liter volumes (10 hours each) of buffer consisting of 0.1 M potassium phosphate, 1 rnhf MgSO+ 1 mM EDTA, 50 mM glucose, and 10 m&i mercaptoethanol, pH 7.0. The enzyme was then adsorbed onto a DEAE-cellulose column (Fig. 2). The fractions (82 to 92) containing more than 0.4 units per ml were combined, concentrated as before with the ultrafiltration apparatus to 20 ml, and dialyzed against two 250-ml volumes (5 hours each) of buffer consisting of 0.1 M potassium phosphate, 0.1 by KCl, 5 rnM EDTA, 5 m&f MgS04, 2 mM dithiothreitol, and 0.5 N glucose, pH 7.0. The protein solution then was chromatographed on a Sephadex G-100 column (Fig. 3). Fractions exhibiting activity were monitored for protein concentration by absorbance at 280 nm and for protein purity by sodium dodecyl sulfate acrglamide gel electrophoresis.
Those fractions for which the specific activity was greater than 10 were combined (30 to 35) and dialyzed against two l-liter volumes (10 hours each) of 0.1 bf potassium phosphate buffer, 1 rnM h1gS04, 1 mM EDTA, 50 m&I glucose, and 10 mM mercaptoethanol, pH 7.0. The dialyzed enzyme solution was adsorbed onto a column of DEAE-Sephadex equilibrated with the same buffer as that used for the preceding enzyme dialysis (Fig. 4). The enzyme (Fractions 151 to 154) obtained from this column represent a 10,526-fold purification and a yield of about 2.5%. The purification is summarized in Table I.
Notes on Putification-The entire purification of glucokinase 3057 beginning with 200 rats, can be completed in about 3 weeks. Generally, we killed between 20 and 30 rats in the morning for 8 consecutive days. By early afternoon the dissolved ammonium sulfate paste was layered on two Sephadex G-100 columns. Generally, the active fractions were eluted by early afternoon the following day. It was soon observed that collection of every fraction was unnecessary since glucokinase activity was always eluted between fast moving green material and a slower moving red material. The active fractions were combined and were immediately treated with DEAE-Sephadex and the gradient elution was completed by the following day. At this point the purification was stopped until all the initial liver extracts had been processed to this stage, generally 9 to 11 days after the initial sacrifice. All the material obtained from the batchwise DEAE-Sephadex chromatography was combined. The purification was completed in an additional 7 days.
In the course of this purification, several other procedures were investigated in attempts to improve the protocol. Commercially available frozen livers or acetone powders were about 50% lower in initial activity than freshly excised livers.
Chromatography with CM-Sephadex, phosphocellulose, hydroxyapatite and alumina gel Cy proved less favorable than the procedures described above. In addition, significant purification or retention of activity was not obtained after fractionation by ethanol, acetone, or polyethylene glycol.
Although high concentrations of glucose (see "Thermal Stability") afforded the enzyme protection against thermal inactivation, it apparently did so for proteins in general since only minimal improvements in specific activity were observed when crude enzyme was heated at 48.5" for 30 min. Finally, electrophoretic methods including starch gel electrophoresis, preparative disc gel electrophoresis, and isoelectric focusing were also found unfavorable owing to the extremely high losses of activity.
Stability during Purification and Storage-Glucokinase activity in the high speed supernatant extract falls off rapidly below pH 7.0 (Fig. 5). Of the several buffers tried at pH 7.0, potassium phosphate was found to give the highest yield of the enzyme in the initial extract and the greatest stability.
Although all previously reported purification procedures have employed Tris-HCl buffer, we found both lower activity immediately after extraction and upon storage at 4", when Tris-HCl was compared with potassium phosphate.
Ammonium sulfate fractionation of glucokinase activity also falls off at pH below 7.0. In addition, we have consistently obtained better yields when fractionation was performed by the addition of a saturated solution of neutralized salt rather than by the addition of the crystalline form of ammonium sulfate.
The ammonium sulfate paste can be stored for at least several days at -20" without appreciable loss of activity.
Supernatants obtained from thirty-three per cent arations were made 0.5 M in glucose and heated as indicated, cooled rat liver homogenates (w/v) were prepared and stored at 4' using in ice, and assayed. Initial high speed supernatant from 33% the buffers indicated.
All cellulose column : minus glucose, A---A ; plus glucose, A---A. FIG. 6 (right). Effect of glucose on the rate of heat inactivation combined active fractions from the first gel filtration step must be batchwise treated immediately with DEAE-Sephadex because at this stage of purification a delay of one day results in a 75% or more loss of activity. Enzyme obtained from the batchwise DEAE-Sephadex step and all subsequent steps is very stable. We have had such preparation stored at 4" for several weeks with only a 10% loss of activity.
It is important that the purified enzyme be stored in 50 mM glucose, 100 InM potassium ion, and 10 m&f mercaptoethanol since omission of any of these results in rapid losses of activity.
V,,X Thermal Stability-At any stage of the purification of glucokinase, the activity decreases rapidly when the enzyme is heated for 10 min at temperatures above 42". In the absence of glucose all activity is lost when enzyme is held at 52" for 10 min. How ever, 0.5 M glucose has a significant effect in protecting glucokinase against thermal inactivation at 48.5", and the effect increases with increased enzyme purity (Fig. 6). In fact, purified glucokinase can be heated for 60 min at 48.5" in the presence of 0.5 M glucose without any detectable loss in activity.
Km and V,,, -The Michaelis constants for glucokinase activity towards several substrates are given in Table II. Included are the values for the phosphorylation coefficient defined as: enzyme (9). Therefore, o-mannose and n-glucose act as substrates, while glucosamine and N-acetylglucosamine function as competitive inhibitors.
The phosphorylation coefficient is a useful indicator of the relative susceptibility of a substrate to phosphorylation (8). The values for K, that we report are somewhat lower than those of Parry and Walker (1) although our K, value for glucose is in agreement with that reported by l'ilkis (3).
Spectral Analysis-Purified glucokinase exhibits the absorption spectrum expected from a protein devoid of the usual coenzymes. Absorption maxima were observed at 278 and 231 nm. The sample of enzyme used in this experiment had a protein concentration of 200 pg per ml when determined by the fiuorescamine assay for protein and an optical density reading of approsimately 0.20 units at 278 nm.
With yeast hexokinase only the 2-hydroxyl group of the carbohydrate substrate can be modified and still permit binding to the Molecular Weight and Turnover Number-Glucokinase has a molecular weight of 57,000 when determined by Sephadex G-100 gel filtration (Fig. 8). As seen in Fig. 9, glucokinase and glutamate dehydrogenase (subunit mol wt 53,000), if mixed, migrate as a single protein band when subjected to SDS acrylamide gel electrophoresis.
The difference in molecular weight obtained by A column (2.5 X 28 cm, Ire = 137 ml) of Sephadex G-100 was equilibrated with 0.1 M potassium phosphate, pH 7.0, plus 1 mM EDTA, 1 mM MgS04, 10 mM mercaptoethanol, and 50 mM glucose. The void volume (VO) was 38 ml. One milligram of each protein standard in 0.5 ml of buffer plus 0.05 ml of 3 M glucose was layered on the column. When glucokinase was examined, approximately 1 unit of activity was added and the elution volume (V,) determined by activity assay. K,, = we -Vo)I(V, -Vol. the two methods is probably due to the hydrodynamic properties of the native enzyme, which become important during gel filtration. Nevertheless, comparison of the values obtained by the two procedures, suggests that glucokinase is a monomer. The turnover number, assuming one catalytically active site, a molecular weight of 53,000, and a V,,,,, of 82 units per mg for pure enzyme, is 4,346 moles of n-glucose phosphorylated by ATP per min per mole of glucokinase.

DISCUSSION
Glucokinase is difficult to purify because of its instability during initial fractionation procedures and the small amount in the liver.
We have calculated that protein with glucokinase activity comprises approximately 0.007% of the total liver protein and 0.018% of the soluble liver protein.
It is clear from the value for the over-all yield presented in this paper that sufficient material for characterization and immunological studies can be obtained only after the purification procedure has been repeated several times. However, owing to the stability of the purified enzyme this task is not beyond the scope of possibility.
In our laboratory, we have repeated this procedure three times to date and are in the process of accumulating milligram quantities of pure glucokinase.
Our purification protocol differs from the three previously published methods in several respects.
In the purification reported by Gonzalez el al. (2), CM-Sephadex and hydroxyapatite were employed.
We have been unable to achieve significant purification or improvement of stability by using either of these materials.
In addition, these workers reported precipitation of glucokinase activity between 0.55 and 0.80 saturation with ammonium sulfate, whereas our results show maximum increases in specific activity are obtained between 0.42 and 0.68 saturation. Inclusion of glucose in our procedure during the fractionation or differences in protein concentration may account for the effectiveness of a lower ammonium sulfate concentration.
The purification reported by Parry and Walker (1) results in material purified approximately l,lOO-fold, with a specific activity of 11 units per mg. According to our calculation, this enzyme preparation is only 14% pure. At no point in their chromatographic separations did they obtain a protein peak which corresponded to the activity profile.
In contrast to their use of Tris-HCl buffer throughout the purification, we found potassium phosphate buffer to be superior because it enhanced enzymic stability.
Phosphate buffer gradients gave sharp separations during ion exchange chromatography.
In addition the KC1 requirement for stability could be met by potassium phosphate.
The purification protocol presented by Pilkis (3) is a modification of that of Parry and Walker (1). A significant purification and yield were obtained using preparative starch gel electrophoresis, a procedure, which in the hands of many investigators, including ours, has proven less than satisfactory.
Use of gel filtration with Sephadex G-100 is, however, a major reason for the success of his purification (2940-fold; specific activity, 29.4 units per mg) and ours. We have found that use of large gel filtration columns early in the purificat.ion achieves significant purification and removes the ammonium sulfate from the previous step, making dialysis unnecessary before batchwise ionexchange chromatography.
As seen in Fig. 3 after the second filtration of glucokinase through Sephadex G-100, a considerable amount of contaminating protein is still present, even though the specific activity represents a purification of 2,520.fold.
Special reference is made to the two closely migrating, heavily staining protein bands at the bottom of the gel. Disc gel electrophoresis of this preparation in the absence of sodium dodecyl sulfate allows the detection of only one protein band and it is our conclusion that this band represents the undenatured protein resulting in the double band (molecular weight 26,000 f. 1,000 each) appearing in the sodium dodecyl sulfate gel. It therefore appears to us, that prior to the last step of our purification, a dimeric protein of molecular weight 52,000 is being co-purified with glucokinase.
Use of a long, slowly running DEAE-Sephadex column with a shallow potassium phosphate gradient has been used in our procedure to effect separation of this impurity from the glucokinase.
We have confirmed the need for inclusion of 50 mM glucose, 100 InM potassium ion, and 10 InM mercaptoethanol in all buffers during purification and storage of glucokinase. As indicated, potassium phosphate buffer at pH 7.0 was found most suitable for the initial liver homogenizing medium and subsequent steps. The enzyme is most labile in the high speed rat liver supernatant and after the first gel filtration chromatography.
However, it is quite stable during all subsequent steps of the purification and during storage at 4" with the stabilizers present.
In addition, the ability of glucokinase to resist thermal inactivation in the presence of 0.5 M glucose increases with increasing enzyme purity.
Determination of the phosphorylat,ion coefficient reveal the following order of capacity to be phosphorylated by ATP: glucose > mannose > 2-deoxyglucose > fructose.
Of the several other compounds tested only those with an altered C-2 hydroxyl group would act as inhibitors confirming the observation that hexokinases require unaltered C-3, C-4, C-5, and C-6 hydroxyls in the pyranose configuration of the compound for binding (10). Like brain hexokinase, glucokinase is also inhibited by mnitrobenzoylglucosamine (11). From spectral analysis, glucokinase appears to be devoid of prosthetic groups, exhibiting absorption maxima at 278 and 231 nm.
The molecular weight of glucokinase, when determined by Sephadex G-100 gel filtration is 57,000 and when determined by sodium dodecyl sulfate acrylamide gel electrophoresis is 53,000. These results are in disagreement with previous findings (3) in which the molecular weight was reported to be 48,000 in the presence of 0.15 M KC1 and 65,000 in the absence of salt when examined by the method of gel filtration with Sephadex G-100 (3). At the same time, it was reported that glucokinase behaves like a protein with a molecular weight of 68,000 by gel filtration on Bio-Gel P-100 and by sucrose density gradient centrifugation.
The cause for these discrepancies is uncertain but it has been suggested (3) that glucokinase undergoes a conformat,ional change during gel filtration in buffers of lower ionic strength.
We have carefully examined the behavior of pure glucokinasc in sodium dodecyl sulfate acrylamide gel electrophoresis wit,11 reference proteins having molecular weights near that suspected for glucokinase.
Since glucokinase and glutamate dehydrogenasc (subunit mol wt 53,000) exhibit identical RF values when examined by sodium dodecyl sulfate acrylamide gel electrophoresis, we report a molecular weight of 53,000 for rat liver glucokinase with some certainty. It is difficult to compare our result obtained 011 gel filtration with those reported by others since our buffer system and ionic strength differed from theirs, even though our result, molecular weight 57,000 with this technique, falls between the reported values of 48,000 and 68,000. Comparison of the molecular weight values obtained by gel filtration and gel electrophoresis here suggest that glucokinase is a monomer in its native state.
At the outset of this work it was our intention to obtain sufficient pure glucokinase to produce fluorescent antibodies which could be used to ascertain whether individual liver cells contain widely differing amounts of glucokinase depending 011 their state of development.
At present it is unclear whether the low or questionable levels of glucokinase found in fetal liver and most hepatomas (5) means that all cells produce a reduced amount of enzyme or that some cells are producing normal amounts while others fail to do so. It is possible that slowly growing hepatomas contain a high proportion of cells that have differentiated function, therefore exhibiting glucokinase and pyruvate kinase type I (or L) while rapidly growing hepatomas, almost devoid of glucokinase, are essentially all enzymatically undifferentiated (12). The availability of pure glucokinase for the preparation of fluorescent antibodies makes possible a direct approach to the problem.