An Unusual High-K m Hexokinase Is Expressed in the mhAT3F Hepatoma Cell Line*

In most hepatoma cells, the high-K m GLUT2/glucokinase proteins are replaced by the ubiquitous low-K m GLUT1/hexokinase type I proteins. In the mhAT3F hepatoma cells, the stimulatory effect of glucose on gene expression and glycogen accumulation was not maximal at 5 mmol/liter glucose. This response to high glucose is observed in mhAT3F cells, where GLUT2 was expressed, but not glucokinase (assessed by Northern blotting and reverse transcription-polymerase chain reaction). A low-K m hexokinase activity (19.6 ± 3.8 milliunits/mg of protein) was present, but a high-K m (40 mmol/liter) hexokinase activity (13.9 ± 2.5 milliunits/mg) was also detected in mhAT3F cells. The high-K m hexokinase activity was dependent on both ATP (or PPi) and glucose in the assay and was recovered in a 10–50-kDa fraction after filtration. A 30-kDa protein was detected using an anti-glucokinase antibody and localized by confocal microscopy at the same sites as glucokinase in hepatocytes. In FAO cells, the high-K m hexokinase activity and 30-kDa protein were not found. We conclude that a high-K m hexokinase activity is present in mhAT3F cells. This might explain why the effects of glucose on gene expression were not maximal at a glucose concentration of 5 mmol/liter. A 30-kDa protein identified using an anti-glucokinase antibody may be responsible for this activity present in mhAT3F cells.

Glucose regulates the expression of several genes, including liver pyruvate kinase (1), fatty-acid synthase (2), insulin (3), transforming growth factor-␣ (4) glucose-6-phosphatase (5), and the facilitative glucose transporter type 2 (Glut2) (6,7). Glucose must be metabolized to stimulate the transcription of these different genes (8,9). In the context of liver genes, the effect of glucose is potentiated by all stimuli able to favor glucose metabolism. For most genes, insulin, by increasing glucokinase gene expression, stimulates glucose phosphorylation and metabolism and potentiates the effect of glucose on liver pyruvate kinase, fatty-acid synthase, and glucose-6-phosphatase gene transcription (5,10,11). For the Glut2 gene, the stimulatory effect of glucose is not potentiated by insulin. Indeed, liver Glut2 gene expression is inhibited by insulin both in vivo (7,12) and in cultured rat hepatocytes (7). Moreover, the presence of glucokinase does not seem to be required for glucose stimulation of Glut2 gene transcription since the Glut2 gene is stimulated in primary cultures of newborn rat hepatocytes that lack glucokinase (13).
In most hepatoma cells, the effect of glucose on gene expression is lost (14). This could be attributed to a partial dedifferentiation of hepatoma cells since the liver-specific proteins glucokinase and GLUT2 are replaced by the ubiquitous proteins hexokinase type I and GLUT1. By contrast, GLUT2 is still expressed in the mhAT3F cell line, whereas the phosphorylation of glucose is mediated by hexokinase type I and not by glucokinase (15,16). Despite the absence of glucokinase, the maximal effect of glucose on gene expression is reached at glucose concentrations Ͼ10 mmol/liter (17). This was surprising because hexokinase type I is inhibited by glucose 6-phosphate, and its maximal activity is reached at low glucose concentrations (1-2 mmol/liter) (18). In addition, GLUT2 is not a rate-limiting step for liver glucose metabolism. This suggested that in mhAT3F cells, the phosphorylation of glucose was mediated by a high-K m hexokinase. The aim of this work was to elucidate the mechanisms displayed in the mhAT3F cell line to stimulate gene transcription in response to glucose concentrations exceeding the K m of hexokinase type I.
When the cultures were performed in the absence of glucose, 10 mmol/liter lactate and 1 mmol/liter pyruvate were added as oxidative substrates. Nevertheless, when the cells were cultured in the absence of glucose in the medium, hepatocytes produced glucose from lactate, pyruvate, and amino acids present in the culture medium, and the glucose concentration in the culture medium was 1.2 mmol/liter after 24 h of culture.
Hepatoma Cell Culture-The mhAT3F hepatocyte cell line was de-* 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  rived from the tumorous livers of transgenic mouse expressing simian virus 40 early genes under the control of the liver-specific antithrombin III promoter (15). Cells were grown in Dulbecco's modified Eagle's medium/Ham's F-12 Glutamax medium (Life Technologies, Inc.) supplemented with 100 nmol/liter insulin, 1 mmol/liter dexamethasone, 1 mmol/liter triiodothyronine (Sigma), 30 nmol/liter Na 2 S 2 O 3 (Sigma, St. Quentin Fallavier, France), and 10% (v/v) fetal calf serum. To study the glycogen content, cells were cultured overnight without glucose to deplete the glycogen stores, and then glucose was added in medium as required.
Cellular Extracts-The medium was removed after 24 h of culture for the hepatocytes or after confluency for the mhAT3F cells; the cells were then washed two times in ice-cold 0.9% NaCl. Cells were scrapped into 1 ml of homogenization buffer containing 0.15 mol/liter KCl, 10 mmol/ liter Tris, 1 mmol/liter EDTA, 0.5 mmol/liter NADP ϩ , and 2 mmol/liter ␤-mercaptoethanol (pH 8) at 4°C. The cell suspension was homogenized for 2 min in a Potter homogenizer surrounded by ice. The cell homogenate was centrifuged at 1000 ϫ g for 1 h at 4°C. Western blotting was performed with 100 g of protein from the supernatants.
The cellular extracts prepared for hexokinase activity determination followed the same protocol, except that the buffer contained 100 mmol/ liter KCl, 25 mmol/liter Hepes, 7.5 mmol/liter MgCl 2 , and 4 mmol/liter dithiothreitol (pH 7.4). Homogenates were vortexed and centrifuged at 1000 ϫ g for 1 h at 4°C.
Hexokinase Activities-Hexokinase activity was measured in 100 l of cellular extracts prepared as described above, in the scraping buffer supplemented with 1 mmol/liter NAD, 4 mmol/liter ATP, and 100 or 0.5 mmol/liter glucose at pH 7.4. The reaction was started by the addition of 5 microunits of glucose-6-phosphate dehydrogenase from Leuconostoc mesenteroides (Boehringer Mannheim), and the activity was followed at 340 nm for 15-20 min at 30°C. One unit of glucose-6-phosphate dehydrogenase activity represents the production of 1 mol of NADH/min at 30°C. Enzyme activity is expressed as milliunits/mg of protein. The activity of high-K m hexokinase was calculated as the difference between activities calculated at 100 and 0.5 mmol/liter glucose (20). The K m value for glucose was determined by measuring the hexokinase activities as a function of added glucose (12.5, 25, 50, and 100 mmol/liter), and the results were drawn according to a Lineweaver-Burk doublereciprocal plot. The phosphoryl donor was determined at pH 7.4 by measuring the activity of high-K m hexokinase in the presence of 4 mmol/liter ATP, ADP, GTP, or PP i .
Hexokinase Activity in Cellular Extracts Containing Proteins with a Molecular Mass between 50 and 10 kDa-Cells were washed twice in ice-cold PBS 1 and frozen in liquid nitrogen. Cells were thawed and scrapped in homogenization buffer at room temperature. Cells were homogenized in a Potter homogenizer for 2 min at 4°C, and the homogenate was centrifuged at 1000 ϫ g for 1 h at 4°C. Five ml of the supernatant was filtered on Centriplus 50 (Amicon, St Quentin Yvelines, France) and centrifuged at 2000 ϫ g for 40 min at 4°C. According to the manufacturer, the proteins were separated by their molecular mass with a molecular mass cutoff at 50 kDa. The recovered sample was filtered again on Centriplus 10 (Amicon) with a molecular mass cutoff at 10 kDa by centrifugation at 2000 ϫ g for 1 h at 4°C to concentrate the preliminary filtrate. The hexokinase assay was performed on 100 l. To verify the molecular mass of the proteins in the sample, proteins were separated by 14% SDS-PAGE, and the blots were stained with Ponceau red.
Western Blots-Ten g of protein extracts prepared as described above was separated by Western blotting using SDS-PAGE (14% acrylamide). After an overnight incubation in 5% (w/v) albumin and Trisbuffered saline (TBS), the blot was washed for 20 min in TBS and 0.1% (v/v) Triton X-100. A polyclonal antibody raised against a recombinant glutathione S-transferase-glucokinase fusion protein produced in sheep was a kind gift from Dr. M. A. Magnuson (21). The blot was incubated for 90 min at room temperature with the antibody diluted 1:1000 (v/v) in TBS, 0.05% (v/v) Triton X-100, 0.01% (v/v) Tween 20, and 3% (w/v) albumin. Then the blot was washed four times in TBS and 0.1% Tween and incubated for 60 min with the secondary antibody (an anti-sheep antibody) coupled to horseradish peroxidase (Pierce). This secondary antibody was diluted 1:2000 (v/v) in TBS, 0.1% Tween, and 5% (w/v) nonfat milk. Again the blot was washed four times in TBS and 0.1% (v/v) Tween 20. The peroxidase activity was revealed with the Amersham Pharmacia Biotech detection system as described by the manu-facturer. Blots were exposed for 5 min (Hyperfilm, Amersham Pharmacia Biotech, Les Ulis, France). This antibody can detect mouse and rat glucokinases as well.
The anti-hexokinase type III antibody was a kind gift from Dr. J. Wilson (Michigan State University, East Lansing, MI). It was raised against a purified enzyme from rat Novikoff hepatoma (22). Ten g of total protein extracts prepared as described above for hexokinase activities was separated by SDS-PAGE (10% acrylamide) and transferred to a membrane. After an overnight incubation with 5% (w/v) albumin and TBS, the blot was washed for 20 min in TBS and 0.1% (v/v) Triton X-100. The blot was incubated for 3 h at room temperature with the antibody against hexokinase type III diluted 1:200 in PBS, 0.1 mmol/ liter EDTA, and 1% Triton X-100. The washing and the detection procedures were identical to those described above, except that an anti-rat antibody was used as the secondary antibody.
RT-PCR-Poly(A) ϩ RNAs extracted from liver and mhAT3F and FAO hepatoma cells were reverse-transcribed and amplified with appropriate controls (23). Using 0.5 g of mRNA as template, primer extension was performed with 3 units of Thermus thermophilus DNA polymerase (Promega) from 15 pmol of the 3Ј-primers in the N terminus and catalytic site of glucokinase and in hexokinase type I. The reaction proceeded for 10 min at 25°C, for 2 min at 57°C, and for 20 min at 70°C and then was stopped by 5 min at 99°C. The amplifications were performed with 2.5 l of reverse-transcribed RNA and with 15 pmol of the 5Ј-primers of the domains to amplify. For each set of primers, controls without RNA and without T. Thermophilus DNA polymerase were performed.
Specific primer sequences were in the N terminus of liver glucokinase (5Ј-GAGCCCAGTTGTTGACTCTG and 5Ј-TTGTCTTCACGCTC-CACTGC), yielding an amplified fragment of 285 bp. Another set of primers selected in the catalytic region of glucokinase (5Ј-GCATCA-GATGAAGCACAAGAA and 5Ј-AGAGGGGACTTTGAGATGGAT) gave a fragment of 190 bp (GenBank TM accession number M25807). The primers selected in hexokinase type I (GenBank TM accession number J04526) were 5Ј-GGACCATGATGACCTGCGGT and 5Ј-GTCAGCATG-GAGTCTGAGATC; they gave an amplified fragment of 340 bp. The cycling profile of the PCR was 2 min at 95°C, followed by 1 min at 94°C; 1 min, 30 s at 65°C; and 1 min, 30 s at 72°C for 10 cycles. The amplified cycles were 30 s at 94°C; 1 min at 60°C; and 1 min, 30 s at 72°C for 25 cycles in a GeneAmp Thermocycler (Perkin-Elmer, Roissy, France). Elongation was performed at 72°C for 5 min. The amplified fragments were separated on a 3% agarose gel containing ethidium bromide.
Glycogen Concentration-Glycogen content was measured in confluent cells on 60-mm dishes after 24 h of culture under different experimental conditions. Cells were washed twice with ice-cold 0.9% NaCl and scrapped into 0.2 mol/liter sodium acetate (pH 4.5). The cell homogenates were sonicated for 1 min (1 pulse/s). Cell homogenates (100 l) were incubated for 1 h at 55°C in the presence of amyloglucosidase to hydrolyze glycogen into glycosyl units. The reaction was stopped by centrifugation for 10 min at 3000 ϫ g, and glucose assay was performed on the supernatant. Results are expressed in micrograms of glycogen/mg of protein.
Glut2 mRNA Concentration-Total RNA was isolated by the method of Chomczynski and Sacchi (24). Northern blotting and hybridization were performed with 20 g of total RNA as described previously (7). The Glut2 cDNA probe was kindly supplied by Dr. B. Thorens (25). The glucokinase cDNA probe was provided by Dr. M. A. Magnuson (26). The poly(A) ϩ RNAs were obtained using the mRNA purification kit from Amersham Pharmacia Biotech.
Immunolocalization-Hepatocytes and mhAT3F and FAO cells were grown on Permanox four-chamber slides (Lab-Tek, Nunc) overnight in the presence of 20 mmol/liter glucose as described above. The glucokinase was detected using the polyclonal antibody raised against a recombinant glutathione S-transferase-glucokinase fusion protein produced in sheep. Cells were washed three times in PBS and permeabilized for 4 min in PBS containing 0.1% (p/v) Triton X-100. Cells were incubated for 40 min in diluted antibody (1:2000 dilution in PBS and 0.2% gelatin) at room temperature, washed three times in PBS and 0.2% gelatin, and treated for 1 h with fluorescein isothiocyanateconjugated rabbit anti-sheep IgG (1:128 dilution; Sigma). The slides were mounted in glycerol/PBS mounting medium (Cytifluor), and confocal laser scanning microscopy was performed using a Leica confocal imaging system (TCS-4D) and an immersion lens (63ϫ, numerical aperture 1.4 plan Apochromat). Micrographs were printed directly from the computer on a dye sublimation printer (Colorease, Eastman Kodak Co.).
Statistical Analysis-Results are expressed as means Ϯ S.E. Statistical analysis was performed using the Wilcoxon test for unpaired data.

RESULTS
Glucose Stimulates Glut2 mRNA and Glycogen Accumulation in mhAT3F Cells in a Dose-dependent Manner-When mhAT3F cells were cultured for 24 h in the presence of increasing glucose concentrations (0, 5, and 17 mmol/liter), an accumulation of Glut2 mRNA was observed (Fig. 1). Furthermore, glycogen accumulated in mhAT3F cells with a dose-response curve similar to the one observed for Glut2 mRNA (Fig. 1).
When mhAT3F cells were cultured in the absence of glucose, the concentration of glucose left in the medium after 24 h was 0.03 Ϯ 0.01 mmol/liter. When mhAT3F cells were cultured in the presence of an initial concentration of glucose of 17 mmol/ liter, the concentration of glucose left in the medium after 24 h was 8.65 Ϯ 0.1 mmol/liter. Hexokinase Activities-The maximal level of Glut2 mRNA in mhAT3F cells was reached at glucose concentrations higher than 5 mmol/liter, suggesting the presence of a high-K m hexokinase. To determine the type of hexokinase present in mhAT3F cells, we performed assays of hexokinase activity at two glucose concentrations: 0.5 and 100 mmol/liter. Two hepatoma cell types were used: mhAT3F, in which a stimulation of Glut2 gene expression in response to glucose was present (Fig.  1); and FAO, in which no response to glucose was detected (data not shown). Both hepatoma cell lines showed an elevated hexokinase activity (activity measured at 0.5 mmol/liter glucose) compared with rat hepatocytes (Fig. 2A). The highest activity was observed in mhAT3F cells (19.6 Ϯ 3.8 milliunits/mg of protein), and the lowest activity in hepatocytes (4.5 Ϯ 0.09 milliunits/mg of protein).
Unexpectedly, we detected a high-K m hexokinase activity in mhAT3F hepatoma cells (activity measured at 100 mmol/liter glucose minus activity at 0. 5 mmol/liter glucose). The high-K m hexokinase activity measured in mhAT3F cells was similar to the one attributed to glucokinase in rat hepatocytes (14.4 Ϯ 3 versus 15.6 Ϯ 0.3 milliunits/mg of protein) (Fig. 2A).
This hexokinase showed a K m for glucose of ϳ40 mmol/liter (Fig. 2B). It catalyzed the phosphorylation of glucose using either PP i or ATP as phosphoryl donor (Table I). Other nucleotides like GTP and ADP did not serve as substrates (Table I).
The enzyme was found to utilize PP i much more efficiently than it does ATP (Table I).
RT-PCR-To determine if glucokinase was expressed in mhAT3F cells at a level that was not detected by conventional Northern blotting (Fig. 3A), we performed a RT-PCR from poly(A) ϩ RNA (Fig. 3B). We used isoform-specific primers hybridizing with specific regions coding for the amino-terminal and catalytic domains of liver glucokinase and not present in other hexokinase mRNAs. We used mRNA from mhAT3F (upper panel), rat liver (middle panel), and FAO hepatoma (lower panel) cells, as presented in Fig. 3B. As shown in Fig. 3B, with RNA extracted from livers, we detected a 285-bp band corre-   ). B, to determine the K m , the hexokinase activities were measured at 0.5 and then at 12.5, 25, 50, and 100 mmol/liter glucose, and the high-K m activity results (calculated as described above) were plotted according to Lineweaver-Burk. The intercept of Ϫ1/K m was obtained using linear regression calculation.
sponding to N-terminal domain and a 190-bp band to the catalytic domain of liver glucokinase. We did not detect these bands with RNA extracted from mhAT3F cells. The catalytic domain of glucokinase was not present in FAO mRNA. We did detect a 340-bp band specific for hexokinase type I with RNA extracted from mhAT3F and FAO cells. This demonstrated that glucokinase was not expressed in mhAT3F cells (Fig. 3).
Hexokinase Proteins-Using an anti-hexokinase type III antibody, we found a large signal located at ϳ100 kDa in lung extracts used as a positive control (Fig. 4). The hexokinase type III protein was not detected in mhAT3F cells and hepatocytes (Fig. 4).
Using an antibody against glucokinase, we observed an immunoreactive band at 50 kDa in extracts from rat hepatocytes (Fig. 4) and mouse liver (data not shown). We did not obtain a 50-kDa signal with extracts from mhAT3F and FAO cells, suggesting that glucokinase was not present in these cell lines (Fig. 4). Nonetheless, a band around 30 kDa was detected with the polyclonal anti-glucokinase antibody in cellular extracts from mhAT3F cells. This band was not found in rat hepatocytes. Such a band was not observed in extracts from FAO cells, suggesting that this protein was specific for the mhAT3F cell line.
Hexokinase Activity in 10 -50-kDa Proteins from mhAT3F Extracts-To determine whether the 30-kDa protein recognized by the anti-glucokinase antibody in mhAT3F cells possessed a glucose phosphorylation activity, we measured the activity of hexokinase in fractions of protein selected by their molecular masses. To eliminate proteins with a molecular mass higher than 50 kDa, cellular extracts were filtered. We performed hexokinase assay on filtered extracts in the presence of 0.5 and 100 mmol/liter glucose (Fig. 5). A low-K m hexokinase activity was detected in the cellular extract; however, this activity was no longer found in the proteins under 50 kDa. A high-K m hexokinase activity, reminiscent of a glucokinase activity, was measured both in the cellular extract and in the 10 -50-kDa mhAT3F proteins.
We verified that this activity can be attributed to a hexokinase by performing the assay under the following conditions: 1) in the absence of ATP and in the presence of 100 mmol/liter glucose (0.3 Ϯ 0.02 milliunits/mg of protein) and 2) in the presence of ATP and in the absence of glucose (1.5 Ϯ 0.7 milliunits/mg of protein). Under these conditions, we detected very low hexokinase activities, suggesting that the enzyme required both ATP and glucose as substrates. We did not measure any hexokinase activity in the presence of glucose-6-phosphate dehydrogenase alone, without cellular extracts, or in the presence of fructose (data not shown). Thus, the activity measured in the presence of 100 mmol/liter glucose was in accordance with the activity measured in crude extracts (14.6 Ϯ 3.8 versus 13.9 Ϯ 2.5 milliunits/mg of protein). This enzyme activity could explain why the transcriptional effect of glucose on Glut2 was not maximal at glucose concentrations lower than 5 mmol/liter.
Immunolocalization of the Protein with an Antibody against Glucokinase-The cellular localization of the 30-kDa protein in mhAT3F cells was determined by immunofluorescence using the antibody against glucokinase previously used for the Western blot experiments (Fig. 4). Images were analyzed by confocal laser scanning microscopy. As expected, glucokinase was mainly localized in the vicinity of the plasma membrane in cultured rat hepatocytes; glucokinase was also detected in the nucleus in other hepatocytes (Fig. 6). Glucokinase was not detected in FAO hepatoma cells (Fig. 6). On the contrary, the mhAT3F hepatoma cells did show a positive signal in the cytoplasm, near the plasma membrane, in some cells and in the nucleus in other cells (Fig. 6). This confirmed that the 30-kDa protein detected by the anti-glucokinase antibody was specific for mhAT3F cells. An identical localization was observed in hepatocytes and mhAT3F cells. DISCUSSION In the liver, glucose phosphorylation is the limiting step for glucose metabolism since the glucokinase activity is 100-fold lower than the glucose transport capacity mediated by GLUT2 (6,27). In mhAT3F cells, which express GLUT2 but not glucokinase (16), Glut2 mRNA levels depended on glucose concentrations, as previously reported in hepatocytes (6,7,13). Thus, the presence of GLUT2 could be necessary for the stimulation of gene transcription in response to glucose. This is in agreement with a recent study demonstrating that pancreatic beta cells from Glut2 null mice had lost their capacity for stimulating insulin gene transcription in response to glucose (28). Nevertheless, the existence of a high-K m hexokinase in mhAT3F cells was hypothesized to explain the transcriptional response of GLUT2 to glucose concentrations higher than 5 mmol/liter and was tested.
An additional argument favoring the presence of the high-K m hexokinase in mhAT3F cells was the dose-dependent accumulation of glycogen in response to glucose. In hepatocytes that overexpressed hexokinase type I after adenovirus infection, glycogen synthesis was unrelated to glucose concentration in the culture medium (29). By contrast, in hepatocytes that overexpressed glucokinase after adenovirus infection, glycogen synthesis and lactate production were related to glucose concentrations in the medium (29). These differences could be attributed to the cellular localization of these two hexokinases. A number of studies have shown that liver glucokinase is translocated from a bound to a free compartment in response to glucose, fructose, and sorbitol (30 -32). The translocation occurs in response to metabolic signals that cause dissociation of glucokinase from its regulatory protein (33) The synthesis of glycogen is very sensitive to a small increase in glucokinase activity and correlates more closely with the free glucokinase activity (34). This suggested that mhAT3F cell expressed a high-K m hexokinase that shared some common characteristics with liver glucokinase. Moreover, a K m for glucose of 40 mmol/ Rat lung extracts were used as positive controls for immunodetection of hexokinase type III (HKIII). As expected, the antibody against glucokinase detected a single band at 50 kDa in hepatocytes, and the antibody against hexokinase type III detected a major band at 100 kDa in lung.
FIG. 5. Hexokinase activity in mhAT3F cellular extracts containing proteins between 10 and 50 kDa. mhAT3F cellular extracts were prepared as described under "Materials and Methods" and assayed for hexokinase activity. Protein contained in the filtrate exhibited a molecular mass between 10 and 50 kDa on SDS-PAGE. Hexokinase activity was measured in 100 l of filtrate in the presence of 0.5 and 100 mmol/liter glucose. Results are expressed in milliunits/mg of protein and are the means Ϯ S.E. for three different experiments. ***, significant differences (p Ͻ 0.0001) from values obtained in cellular extracts. The inset shows a blot of 10 g of proteins immunodetected with an anti-glucokinase antibody. Total proteins extracted from hepatocytes (lanes 1), mhAT3F cells (lanes 2), and FAO cells (lanes 3) were submitted to a filtration procedure to recover the proteins with molecular masses between 50 and 10 kDa (10ϽProtϽ50 kDa). They were then loaded on the 14% SDS-polyacrylamide gel.
liter and an ATP dependence showed common kinetics with liver glucokinase (35).
A high-K m hexokinase activity was found in mhAT3F cells, but not in FAO cells, which is another hepatoma cell line. Using Northern blotting and RT-PCR, we showed that liver glucokinase was not expressed in mhAT3F and FAO cells. Thus, glucokinase was not responsible for the high-K m hexokinase activity detected in mhAT3F cells. Two other enzymes could possibly be responsible for glucose phosphorylation. The presence of a high-K m glucose phosphorylation was reported previously in extrahepatic tissue (36). It was attributed to N-acetylglucosamine kinase, an enzyme that catalyzes the phosphorylation of glucose at high glucose concentrations (K m ϭ 370 mM) (36). This enzyme has a molecular mass of 55 kDa, close to the 50 kDa of glucokinase (36). In diabetic rats, the activity of liver glucokinase decreased, whereas the liver Nacetylglucosamine kinase activity remained steady (37). To rule out the possibility that the activity we measured in mhAT3F cells was N-acetylglucosamine kinase, filtration experiments were performed. After elimination of proteins with molecular masses higher than 50 kDa and lower than 10 kDa, we still detected a glucose phosphorylation activity that was thus not related to N-acetylglucosamine kinase. We also observed a lower glucose phosphorylating activity in cells cultured in the presence of 17 mM glucose than in cells cultured in absence of glucose. Thus, the high-K m hexokinase activity observed in mhAT3F cells cannot be attributed to N-acetylglucosamine kinase because the activity of this enzyme was not expected to be altered under these experimental conditions. The high-K m hexokinase activity observed in mhAT3F cells could also be attributed to the phosphotransferase activity of the glucose-6-phosphatase system (38). This seemed very unlikely because the glucose-6-phosphatase is located in the microsomes, whereas the high-K m activity was recovered in a soluble fraction after filtration. Furthermore, the enzyme activity was measured at pH 7.4, a pH at which the phosphotransferase was unable to use ATP as a phosphoryl donor (39).
The high-K m hexokinase activity was recovered in the proteins with molecular masses ranging from 50 to 10 kDa in mhAT3F extracts. In liver extracts, a major band at 50 kDa and a minor band at 30 kDa (considered as a degradation product of glucokinase) were detected with an anti-glucokinase antibody (40). This degradation product was not detected here in rat hepatocytes, but a 30-kDa band was present in mhAT3F cells. We were unable to amplify by PCR any N-terminal glucokinase sequences in mhAT3F cells, contrary to what was observed in rat liver. Thus, the presence of an N-terminal truncated glu-cokinase protein is unlikely. This band at 30 kDa cannot be attributed to the murine origin of mhAT3F cells since the antibody against rat glucokinase readily detects mouse liver glucokinase. Moreover, rat liver glucokinase cDNA can hybridize on Northern blotting with mouse glucokinase mRNA (data not shown). The ATP-and glucose-binding sites of glucokinase are localized within 272 amino acids (41), and these domains are likely to be epitopes that can be recognized by an antibody against purified recombinant glucokinase. Moreover, in mhAT3F cells, this protein was localized in the cytoplasm with accumulations in the vicinity of the plasma membrane in some cells and in the nucleus of other cells. These images were very similar to those observed in hepatocytes.
Taken together, these experiments suggest that a new protein whose function is to phosphorylate glucose at a K m for glucose of ϳ40 mmol/liter in an ATP-dependent manner is present in mhAT3F cells. This protein has a molecular mass of 30 kDa, shares epitopes with glucokinase, and displays a similar cellular localization in hepatocytes and mhAT3F cells.