Val. 268, No.24, Issue of August 25, PP. 18259-18266,1993
Printed in U.S. A.
THEJOURNAL
OF BIOLOGICAL
CHEMISTRY
0 1993 by The American Society for Biochemistry and Molecular Biology, Inc.
Structure/Function Relationships in Hexokinase
SITE-DIRECTED MUTATIONAL ANALYSES AND CHARACTERIZATION OFOVEREXPRESSED
FRAGMENTSIMPLICATEDIFFERENTFUNCTIONS
FOR THE N- AND C-TERMINAL
HALVES OF THE ENZYME*
(Received for publication, January 13, 1993, and in revised form, April 9, 1993)
Krishan K. Arora, Charles R. FilburnS, and
Peter L. Pederseng
From the Laboratoryfor Molecular and Cellular Bioenergetics, Department of Biological Chemistry, The Johns Hopkins
University School of Medicine, Baltimore, Maryland 21205
Hexokinases are comprised of two highly homolo- in tumors has properties in common with both the Types I
gous -50-kDa halves and are product-inhibited
by glu- and I1 isozymes (5). The Types I, TI, and I11 isozymes, and
cose-6-P. Fouraminoacid
residues, SerSos,Asp”’,
the tumor enzyme, consist of a single polypeptide chain with
G ~ U “ ~located
,
in the C-terminal half of the
Glu708, and
a molecular mass of -100 kDa, and all four are inhibited by
tumor mitochondrial enzyme have been shown
to be the product Glc-6-P (6). The Type IV hexokinase (also known
essential for enzyme function (Arora, K. K., Filburn, as glucokinase), similar to yeast hexokinase, is -50 kDa and
C. R., and Pedersen, P. L. (1991) J. Biol. Chem. 266, is insensitive to inhibition by Glc-6-P (6).
5359-5362). Here we have assessed also the role of
Recent studies have resulted in thecloning and sequencing
the N-terminal half of the same enzyme.
of
hexokinases from a variety of tissues including yeast (7,8),
Site-directed mutagenesis of residues predicted to
interact with glucose in the N-terminal
half, Le. SerlS6, human kidney (9), rat brain (IO),rat skeletal muscle (I]), rat
AspaoB,and G1u260,to Ala,have no effect on hexokinase liver (12,13), and,from this laboratory, mouse hepatoma (14).
activity. In addition, inhibition by hexose mono- and The deduced amino acid sequence data indicate a significant
identity between the mammalian forms and the tumor enbisphosphates is unchangedforeach of themutant
enzymes. Significantly, the overexpressed N-terminal zyme, as well as strong similarity to theyeast hexokinase and
polypeptide is devoid ofcatalytic activity but doeshave rat liver glucokinase (12, 14, 15). Thus, these studies provide
the capacityto bind ATP-agarose and be released
with evidence in support of the gene duplication-gene fusion hyATP and glucose-6-P. In contrast, the overexpressed pothesis for the evolution of 100-kDa hexokinases from the
C-terminalpolypeptide is catalytically active and 50-kDa yeast-like enzymes. It has been speculated (16, 17)
shows the same product inhibition pattern
as the com- that, asa result of such a duplication event, one of the halves
plete 100-kDa parent enzyme.
may have evolved a regulatory function, whereas the other
These results emphasize that the N-terminal half of
tumor hexokinase is essential neither for catalysis nor half retained the catalytic function of the enzyme. Signifiproduct modulation. Rather, the N-terminal halfmay cantly, the C-terminal half of the brain enzyme has been
play another role, perhaps in modulation of the ATP/ shown to comprise all its catalytic activity (18),whereas the
glucose-6-P-dependent binding of the enzyme
tumor
to N-terminal half is predicted to be involved in product inhibition by Glc-6-P (10, 19).
mitochondriaorbyacting
as a spacer between the
Based on amino acid sequence comparisons of tumor and
outer mitochondrial membrane and the C-terminal
catalytic unit.
other hexokinases with that of the yeast enzyme, the x-ray
structure of which is known (20, 21), putative glucose and
ATP binding domains have been predicted (12, 15). The
glucose binding domains, found in boththe C- and N-terminal
Hexokinase (ATP:D-hexose 6-phosphotransferase, EC halves, include residues thought tointeract directly with
2.7.1.1) commits glucose to catabolism by catalyzing its phos- glucose in the yeast enzyme (21) (see Fig. 1, A and B, and
phorylation to Glc-6-P utilizing MgATP. In rapidly growing accompanying legends).
cancer cells, hexokinase is markedly elevated, and a large
In our most recent studies(22), we have used the technique
fraction (5040%)of the total activity is bound to the outer of site-directed mutagenesis for the firsttime to evaluate the
mitochondrial membrane (1,2), where it haspreferred access role of amino acids predicted to interact with glucose in the
to mitochondrially generated ATP (3). In mammals, there are C-terminal half of the tumor enzyme. We showedthat all four
four isozymes ofhexokinase, i.e. Types I, 11,111,and IV, which conserved residues (Ser603,Aspm7,Glu708,and G ~ U ~in
~ ’the
)
vary in their tissue distribution and kinetic properties (4).Of C-terminal half of tumor hexokinase, predicted from x-ray
the four known isozymes of hexokinase, the major form found studies of the yeast enzyme to be within “contact” distanceof
* This work was supported by National Institutes of Health Grant glucose, are required for enzyme function. Interestingly, these
CA 32742 (to P. L. P.), The costs of publication of this article were 4 residues predicted to be within “contact” distanceof glucose
defrayed in part by the payment of page charges. This article must are also conserved in the N-terminal half of the enzyme.
therefore be hereby marked ‘‘advertisement” in accordance with 18 These correspond to SeP5,Aspzw,G1uZ6O,and GluZg4
(see Fig.
U.S.C. Section 1734 solely to indicate this fact.
1B).
$ Permanent address: Laboratory of Biological Chemistry, GeronIn this report, we have evaluated the role of the N-terminal
tology Research Center, National Institute on Aging, Baltimore, MD
half of tumor hexokinase using two different approaches: (a)
21224.
f To whom correspondence should be addressed. Tel.: 410-955- by assessing the requirement of conserved amino acid residues
3827; Fax: 410-955-5759.
predicted to interactwith glucose using the technique of site-
18259
Structure-Function
Studies
18260
directed mutagenesis and( b ) by characterizing overexpressed
N- and C-terminal halves of the enzyme.
EXPERIMENTAL PROCEDURES
Materials
Adenine and pyridine nucleotides, dithiothreitol, and glucose-6phosphate dehydrogenase (Grade I) were obtained from Boehringer
Mannheim. 1,5-Anhydro-~-glucitol, 5-thio-D-glucose-6-P, glucose1,6-bisphosphate were from Sigma. Sources of restriction and DNAmodifying enzymes and other molecular biological chemicals have
been described previously (14). Oligonucleotide primers for site-directed mutagenesis were synthesized in the Protein/Peptide/DNA
facility in the Department of Biological Chemistry, The Johns Hopkins University School of Medicine. The oligonucleotide-directed in
vitro mutagenesis kit and theenhanced chemiluminescence detection
kit for Western blotting were obtained from Amersham Corp. and
were used according to the manufacturer’s instructions.
Methods
Recombinant DNA Procedures-Unless specified, standard molecular biological protocols described by Sambrook et al. (23) were used.
Oligonucleotide-directedMutagenesis within the N-terminalHalf of
Tumor Hexokinase-Site-directed mutagenesis was performed according to themethod of Taylor et al. (24) using a kitfrom Amersham.
A 2.84-kb’ XbaI-XbaI fragment containing the entire coding region
of tumor hexokinase and 50 base pairs of the 3”untranslated region
was subcloned into the polylinker region of MI3 mp19 at the XbaI
site and wasused as a template to make site-directed changes as
described previously by Arora et al. (22). Site-directed mutations
reported in this study and the sequence of the mutagenic primers,
with nucleotide mismatches underlined, are as follows.
S e r lS5
S +A
ASP^^'
D -+ A
G1UZ6’
E
*A
5’ - GTGAATGCCACAGTGGGG - 3‘
5‘ CACGTTTGCCTTCCCGTGC- 3‘
5‘-GTATTAACACGGsATGGGGAGCC-3’
Mutations were identified by sequencing of single strand DNA from
M13 transformants. DNA sequencing was carried out by the dideoxy
chaintermination method of Sangeret al. ( 2 5 ) using Sequenase
(version 2.0).
In order to transfer the mutants to the
expression plasmid pKAHX
(14) and to ensure that the site-directed mutations reported in this
study were confined to the N-terminal half of the molecule, a replicating form DNA of M13 mutants was cut at two unique restriction
sites, EcoRV (at base 118) and NcoI (at base 1365). The 1247-base
pair EcoRV-NcoI fragment was gel-purified and then recloned into
the plasmid pKAHX from which the wild type EcoRV-NcoI fragment
had been removed (see Fig. 2 A ) . The desired mutations in the expression plasmid were reconfirmed by DNA sequence analysis.
Plasmid Construction for Expression of the N- and C-terminal
Halves of Tumor Hexokinase-The scheme used for construction of
C-terminal halves of hexokinase
plasmids for expression of the N- and
is shown in Fig. 2B. For the N-terminal half (molecular mass approximately 46 kDa), the 6.5-kb plasmid pKAHX used previously for the
overexpression of intact tumor hexokinase (14) was digested with
NcoI and HindIII, and the 4.7-kb fragment was gel-purified. The 5 ’ overhangs generated were “filled in” with Klenow fragment in the
presence of four deoxynucleotides, ligated, and then transformed in
Escherichia coli strain DH5.
For the C-terminal half (molecular mass approximately 48.5 kDa),
the plasmid pKAHX (6.5 kb) was digested with Narl and BamHI,
and the1.5-kb fragment was gel-purified and ligated into theplasmid
pFOG 402 previously linearized with BamHI, after “filling in” with
Klenow fragment and thefour deoxynucleotides. The ligation mixture
was transformed in E. coli. strain DH5.
Overexpression and Partial Purification of Wild Type (Intact), Nand C-terminal Halves, and MutantHexokinase Proteins-Wild type
(intact), N- andC-terminal halves, and mutant hexokinase proteins
were overproduced in E. coli under the control of the alkaline phosphatase (phoA) promoter system as described previously in detail by
of Hexokinase
Arora et al. (14). “Particulate” fractions were prepared as described
previously also by Arora et al. (22) and in the case of the wild type
enzyme had specific activities ranging from 200-280 nmol of Glc-6-P
formed per min/mg of protein.
HexokinaseActivity Assay-Hexokinase activity assays for the wild
type (intact), N- and C- terminal halves, and mutant hexokinase
protein preparations were performed spectrophotometrically in the
presence of 0.25% Lubrol WX essentially as described previously (2)
in a coupled assay containing 2.2 mM glucose,6.4 mM ATP, and
glucose-6-phosphate dehydrogenase and NADP+. In inhibition experiments, all components of the assay mixture and the inhibitor were
added to a cuvette. The reaction was then initiated by the addition
of either wild type, mutant hexokinase enzymes, or N- and C-terminal
halves.
Synthesis of l,5-Anhydro-~-glucitol-6-P-1,5-Anhydro-~-glucitol6-P was synthesized by phosphorylating 1,5-anhydro-D-glucitl using
yeast hexokinase and ATP and was purified essentially according to
the procedure described by Ferrari et al. (26). The concentration of
the analog reported in this study is based on the determination of
phosphate released upon acid hydrolysis. Inorganic phosphate was
quantified by a modified method of Fiske and SubbaRow as described
by Bartlett (27). 1,5-Anhydroglucitol-6-P,aGlc-6-P analog, was
preferred ininhibitionstudies
reported here, because it is nota
substrate for the glucose-6-phosphate dehydrogenase-catalyzed reaction, and thus the convenience of the glucose-6-phosphate dehydrogenase-coupled assay for hexokinase activity determination is
retained.
Functional Analysis of the N-terminal Half of Hexokinase-The
isolated overexpressed N-terminal half of hexokinase exhibits no
catalytic activity (see “Results” section). Therefore, its function was
analyzed by its binding to ATP affinity agarose. Preliminary experiments indicated that the particulate fraction of the N-terminal half
solubilized in the presence of0.02 and 0.05% SDS gave the best
results in its binding to ATP-agarose. Protein solubilized with 0.1%
SDS did not bind to ATP-agarose, indicating that thehigher concentration of the detergent is disrupting the structure of the protein,
thus rendering it unable to bind to theaffinity matrix. In experiments
reported here, a 0.02% SDS-solubilized protein sample wasused.
Briefly, the following procedure wasused. A 50-pl suspension of
adenosine 5’-triphosphate-agarose (in which ATP was attached to
the matrix via C8 of the adenine ring with a nine carbon spacer,
Sigma) was added to an Eppendorf tube and washed in 1 ml of 10
mM Tris, pH 8.0, four times before suspending in 200 pl of Tris
buffer, pH 8.0. A 100-p1 aliquot of the solubilized fraction (0.2 mg/
ml) in HET medium (0.05 M HEPES, 0.5 mM EDTA, 10 mM monothioglycerol, pH 7.5) was then added to the agarose, and the tubes
were rotated at 4 “C for 30 min. After incubation, samples were
centrifuged, and the agarose pellets were washed in 1 ml of 10 mM
Tris, pH 8.0, four times. The final pellets were either heated in 150
pl of Laemmli sample loading buffer at 95 “C for 10 min or treated
with 100 mM ATP or 100 mM Glc-6-P for 15 min at room temperature
(with occasional vortexing) to release the bound N-terminal half
polypeptide. The supernatantsfrom the latter treatments
were mixed
with 100 p1 of sample loading buffer. Released protein samples were
subjected to electrophoresis on a 10% SDS-polyacrylamide gel, transferred onto a polyvinylidene difluoride membrane, and probed with
anti-hexokinase antibody (see below).
SDS-PAGE and Protein Determination-SDS-polyacrylamide gel
electrophoresis was carried out in 1.5-mm-thick 10% acrylamide gels
according to the method of Laemmli (28). Protein concentrations
were determined by the bicinchoninic acid assay (Pierce Chemical
Co.) using bovine serum albumin as thestandard.
Western Blotting-Immunoblots were performed on polyvinylidene
difluoride membranes (Immobilon’”) using anti-hexokinase antibodies raised against purified AS-30D hepatoma hexokinase and
visualized with an enhanced chemiluminescence detection protocol
(Amersham).
RESULTS
Site-directed Mutations of Ser‘55,Aspzo9,and G ~ inuthe~
Putative Glucose Binding Domain within the N-terminal Half
of Hexokinase Have Little Effect on Catalytic Actiuity-Previously, we showed that the 4 conserved amino acid residues
pre&PO3,
G1u708,
and Glu742intheC-terminalhalf,
‘The abbreviations used are: kb, kilobase(s); Glc-6-P, glucosedicted to interact with
glucose based on x-ray crystallographic
1,5-anhydro-~-glucitol-6-phosphate;
6-phosphate;1,5-AnG-6-P,
studies of the yeast enzyme (20, 21), are required for enzyme
PAGE, polyacrylamide gel electrophoresis.
~
Structure-Function Studies of Hexokinase
function (22). Significantly, these amino acid residues are
conserved also within the N-terminal half of tumor hexokinase (Fig. 1B). In order to evaluate their role during the
hexokinase-catalyzed reaction, 3 of the 4 conserved residues,
SerI5’,
and G1u2@’, were mutatedto Ala as described
18261
under “Methods” (see also Fig. 2 A ) . Alanine was chosen for
substitution,because it does not participate inhydrogen bonding to the hydroxyl groups of glucose. Hexokinase activity
determined for both the overexpressed wild type and mutant
enzymes is presented in Fig. 3A. It is clear that these point
”-
Dl89
C
IN-TERMINAL
HALF ,+-C-TERMINAL
HALF.-=
N
1
FIG. 1. A, computer graphics representation of the predicted structure of a mammalian hexokinase. The proposed structure is based on
the fact that 100-kDa mammalian hexokinases, including tumor hexokinase, are twice the size
of 50-kDa yeast-like enzyme. The two halves
of tumor hexokinase designated as N- and C-terminal and shown in the figure in
magenta and blue, respectively, share 68% amino acid
sequence homology (14). Theabove structure was obtained by fusing the
C terminus of one complete yeast hexokinase molecule with a second
truncated molecule lacking the
region corresponding to the first helix
of the yeast enzyme (which is present in the N-terminal half but absent
in the C-terminal half of the tumor and brain enzymes) and then rotating the
newly joined molecule to eliminate any steric effects. The
molecular model was generated using the
RIBBONS program (42) with the Personalized system using Silicon Graphics
IRIS 4D series
workstation. The x-ray coordinates for yeast hexokinase were obtained from the Brookhaven Protein Data Bank.
The figure illustrates the
location of various “direct” contact residues believed to interact with glucose based on the x-ray studies of the yeast enzyme. The locations
of these residues Ser’”, Asp”’, G ~ U ” ~ , aGlu’”
n d in both halves are indicated. Numbers refer to the location in the sequence as determined
by x-ray studies (21) and do not correspond directly to positions in the sequence as deduced from cDNA sequencing
of the cloned gene (7,8).
Also, shown is the predicted direct “contact” lysine residue for ATP (Lys”). B, schematic representation of the primary structure of tumor
hexokinase illustrating putative glucose and ATP binding domains and residues proposed to be within contact distance
of glucose and ATP.
Four amino acid residues believed to interact with glucose in the two halves of tumor hexokinase correspond to Ser’”, Asp2*, G1uZ6”,and
GluZg4 in the N-terminal half and Ser603, Asp657,Glu708, and Glu“’ in the C-terminal half. In the tumor enzyme, the amino acid residue
corresponding toLys’l of the yeast enzyme isLys558in the C-terminal half. However, the corresponding lysine residue in the N-terminal
half
is replaced with a glutamic acid residue which is not shown.
Structure-Function Studies of Hexokinase
18262
A.
E. Coli
B.
pKAHX
I)DQat with
I)Di~estwith
Ncol & Bindlll
Nul k B a n HI
2) Gel-purify
1.5kb
Nul-Ban HI
fragment
2) Gel-purify
4.7kb
Ncol-Hindlll
frapent
I)"Fill in"
2) litate
3) Transform 6: Coli
I
1) "Fill in"
21 lirate
3) Transform 6: Coli
4.7kb
CONSTRUCT
TO E X P R E S S
C-TERMINAL
CONSTRUCT
TO E X P R E S S
S-TERMINAL
I
1
w;-(
+
S-HALF
I
455
3 COO-
433
NH;~
+
C-HALF
'
918
+COO-
FIG. 2. A, strategy employed for transfer of the site-directed mutations in tumor hexokinase cDNA from the M13 vector to the
expression plasmid pKAHX (see "Methods" for details). B, construction of plasmids for the expression of N- and C-terminal halves of
tumor hexokinase. Illustrated here are the steps performed to construct plasmids for the expression of N- and C-terminal halves of
hexokinase (see "Methods" for details).
mutations have no major effect on the catalytic activity of
hexokinase.
The S155A mutation consistently induced a slight activation (12-15%) of catalytic activity, whereas the D209A and
E260A mutations induced a slight inhibitory effect (15-22%).
These results remained the same in studies not presented
here over a wide range of glucose and ATP concentrations.
For comparison, resultspresented in Fig. 3B confirm the
effect on catalytic activity reported in our earlier study (22)
when the corresponding residues (Ser603,
and Glu7OS),
as well as G ~ u ~were
~ ' , mutated in the C-terminal half of the
observed
enzyme. In this case, and in sharp contrast to results
for the N-terminal half, catalytic activity is reduced by greater
than 90% in all cases. Finally, these mutants were characterized also for their effect on the apparentK , for ATP. Values
were0.98,0.92, and 1.11 mM, respectively, for the S155A,
D209A, and E260A mutants revealing no significant differences from the wild type enzyme ( K , = 1.02 mM).
In control experiments all recombinant proteins bearing
mutationsin the N-terminal half, i.e. S155A,D209A, and
E260A, were shown to be stable in E. coli. This question was
addressed by examining the overexpressed protein products
of the wild type and mutant genes in SDS-PAGE (Fig. 4). All
mutant proteins (lanes 2 4 ) show a band near 100 kDa with
almost the same intensity as that for the wild type protein
(lanes 1 and 6 ) .These resultswere confirmed for each mutant
protein by using Western blot analysis where predominantly
one immunoreactive band comparable with that of the wild
type was observed (results not presented).
Site-directed Mutations of Ser'55, Aspzm, and G1u2@' within
the Putative Glucose Binding Domain in the N-terminal Half
of Hexokinase Also Have Little Effect on Its Capacity to be
Inhibited by Glc-6-P Analogs-Three proteins bearing mutations in theN-terminal half were tested for their capacity to
be inhibited by Glc-6-P analogs. The Glc-6-P analog 1,5AnG-6-P, which is a linear competitive inhibitor versus ATP,
and a linearuncompetitive inhibitor versus glucose (Fig. 5, A
and B ) , was used initially for the reasons given under "Methods." Results presented in Table I show that no appreciable
differences in Ki(apparent) areseen between the mutant and
the wild type enzymes in the presence of 1,5-AnG-6-P as
inhibitor.
In our previous study (22) Lyd5*, the "invariant" lysine
within the C-terminal half of tumor hexokinase predicted to
interact with ATP, was mutated to Met, and this mutation
decreased the relative V,,, by 70% of the wild type enzyme.
We thought that thislysine may contribute also to thebinding
of the product Glc-6-P. For this reason, the K558M mutant
protein was also tested for inhibition by 1,5-AnG-6-P in the
present study. Again a Ki(apparent) value similar to that of
the wild type was obtained (Table I ) , indicating that the
product inhibitory site within the C-terminal half does not
involve LYS~~'.
In addition to Glc-6-P and 1,5-AnG-6-P, other analogs of
Glc-6-P, viz. 5-thio-Glc-6-P and Glc-1,6-Pz,have been reported to be potent linear competitive inhibitors (versus ATP)
of mammalian hexokinases (29). Therefore, all four mutant
enzymes, S155A, D209A, E260A,and K558M, weretested for
theirinhibition by these hexose mono- and bisphosphate
analogs of Glc-6-P. In every case, both at low and high ATP
concentrations, essentially no difference in the inhibition
pattern from that of the wild type enzyme was observed
(results not presented).
Results described above on the intact enzyme suggest that
the C-terminal half of tumor hexokinase may be responsible
both for catalysis and product inhibition and furtherindicate
that Lys55swithin the catalytic half is not involved in product
inhibition.
The Overexpressed C-terminal Half of Tumor Hexokinase
RetainsCatalytic Activity and the Capacity to Be Productinhibited, whereas the Overexpressed N-terminal Half Has NO
Detectable Activity-In order to evaluate the role of the iso-
Structure-Function Studies of Hexokinase
A. N-TERMINAL HALF
18263
B. C-TERMINAL HALF
1
I
WILD- SlSSA D209A E260A
TYPE
FIG. 3. Effect of site-directedmutations within the N- and C- terminal halves of tumor hexokinase on its catalytic activity.
Procedures for the wild type and mutant enzyme preparations, and hexokinase activity determinations, are described under “Methods.” The
activity was measured a t saturating concentrations of glucose (2.2 mM) and ATP (6.4 mM). Activities expressed as percentage of the wild
type are shown as bar graphs. The datashown in A are means of a t least four independent experiments. The error bars represent the standard
error of mean. The average specific activity for the wild type enzyme was 270 +- 10 nmol of Glc-6-P formed per min/mg of protein ( n = 6).
For comparison, the relative activities are also shown for the corresponding site-directed mutants within the C-terminal half of tumor
hexokinase.
teins areindeed fragments of tumor hexokinase as both bands
react with an anti-hexokinase antibody.
200
The overexpressed C-terminal half of tumor hexokinase
exhibits significant catalytic activity, whereas the N-terminal
z
half has no detectable activity. Although the specific activity
92.5 of the C-terminal half is reduced relative to thewild type (i.e.
69
112 uersus 220 nmol of Glc-6-P formed per min/mg of pro<
tein), it wouldseemon the basis of the above mutational
46
analysis that thisreduction may result because of differences
in stability. In any case, it is important to note that the Cterminal half exhibits essentially the same kinetic patterns
with and without product inhibitor (Fig. 6, A and B ) as the
intact enzyme (Fig.5, A and B ) . Kinetic constants determined
30
from these plots aresummarized in Table 11, where it is clear
that the C-terminal half of tumor hexokinase interacts with
glucose and ATP with very similar, and in most cases nearly
21.5 identical, apparent affinities as theintact enzyme.
Taken together, these results not only indicate that the CFIG. 4. SDS-PAGE analysis of wild type (intact), mutant, terminal half of the tumor enzyme contains the catalytic site
and N- and C-terminal halves of hexokinase proteins. Wild
that thecapacity of the intactenzyme
type, N- and C-terminal halves, and mutanthexokinase proteins were but further demonstrate
overexpressed and subjected to SDS-PAGEexactly as described under to undergo product inhibition is due to its interaction with
“Methods.” Approximately 10-14 pg of protein from the particulate the C-terminal andnot the N-terminal half of the enzyme.
fraction was loaded onto each lane. Lanes I and 6, wild type; lane 2,
The Isolated Overexpressed N-terminal Half of Tumor HexS155A; lane 3, D209A; lam 4, E260A; lane 5, K558M; lam 7, Cterminal half; lane 8, N-terminal half. The position of a major single okinase Interacts with ATP and Glc-6-P-Results presented
band corresponding to that of hexokinase near 100 kDa is indicated above, both on mutant forms of the intact enzyme and on its
by an arrowhead to the left of lane 1. The position of a major single overexpressed halves, indicate that the N-terminal half of
band corresponding to that of the N- andC-terminal halves near 50 hexokinase is neither catalytic nor involved in product inhikDa is indicated by an arrowhead to the right of lane 8. (The two bition. This raised the question about the role of the Nadditional bands near 46 kDa in lane 8 are immunoreactive with
hexokinase antibody and arebelieved to be proteolytic products.) On terminal half of hexokinase and prompted us to examine
the left are shown the positions for molecular size markers (in kDa): whether it can bind ligands like ATP and Glc-6-P. These
myosin (ZOO), phosphorylase b (92.5). bovine serum albumin (69), experiments were done by mixing the N-terminal half with
ovalbumin (46), carbonic anhydrase (30), and trypsin inhibitor (21.5). an ATPaffinity agarose matrix as described under “Methods”
and then using heat, ATP, or Glc-6-P to establish whether
lated C- and N-terminal halves of tumor hexokinase, both binding had occurred by monitoring any released protein with
polypeptides were overexpressedin E. coli and partially puri- a hexokinase antibody. Fig. 7 demonstrates that the N-terfied following the same strategy used previously in this labo- minal half of hexokinase does bind to theATP-agarose matrix
ratory for the complete enzyme (14.22) (see also “Methods”). as immunoreactive species are released by heating (lanes 1
Fig. 4 shows a Coomassie-stained SDS-PAGE gel of protein and 2 ) and by adding ATP (lane 3) or Glc-6-P (lane 4 ) .
samples representing overexpressed N- and C-terminal Although someproteolysis has occurred under the conditions
halves. Bands near 50 kDa as expected for these polypeptides used, these experiments indicate that the isolated N-terminal
are seen in lanes 7 and 8. In experiments not presented here, half of tumor hexokinase does exhibit the capacity to interact
Western blot analysis confirmed that the overexpressed pro- with both ATP and Glc-6-P. Moreover, this binding activity
1 2 3 4 5 6
KDA
-
7 8
18264
Structure-Function Studies of Hexokinase
A.
1
TABLEI
Kinetic constant of the wild type and mutant hexokinase enzymes
Procedures for the wild type and mutantenzyme preparations and
hexokinase activity determinations are described under “Methods.”
The apparent Kivalues were determined with ATP as the varied
substrate (0.2-6.4 mM) at 2.2 mM glucose concentration by replotting
of slopes, obtained from the double-reciprocal plots, versus inhibitor
concentrations. The inhibition patterns in all cases were linear competitive. The specific activity of the wild type particulate preparation
was 270 nmol of Glc-6-P formed per min/mg of protein. The values
shown are averages of two independent experiments and the percent
deviation from the mean did not exceed 10%. app, apparent.
PM
’
OO
I
I
1
I
2
3
Wild type
S155A
D209A
E260A
K558M
88
72
95
78
75
I/[ATP] mM-1
terminal half of the molecule is less well conserved in the Nterminal half (10, 12, 14). This suggested different functions
for the two halves of the molecule. Although evidence was
obtained from biochemical studies of the brain enzyme that
the C-terminal half is catalytic, and theN-terminal half binds
Glc-6-P (18, 19), the finding that Glc-6-P also inhibits the
catalytic activity of the C-terminal half raised doubts about
the role of the N-terminal half in product inhibition.
Data presented in this report evaluate the role of the N‘? 32
terminal half of hexokinase by using two different molecular
biological approaches. Site-directed mutagenesis was the first
approach used to assess the role of the N-terminal half of
>
5 16
hexokinase. Theseexperiments revealed that site-directed
changes of the amino acid residues believed to interact with
8
glucose within the N-terminal half affect neither thecatalytic
activity (Fig. 3A) nor the capacity of the enzyme to be
)
1
1
1
1
)
)
analogs.
inhibited by 1,5-AnG-6-P (TableI) and other Glc-6-P
‘0
IO 20 30 40 50
The second approach involved preparation and characterizaI/[Glucose]
mM”
tion of the C- and N-terminal halves of the enzyme. Results
FIG. 5. Inhibition of the intact tumor hexokinase by 1,5- on the two isolated halves demonstrate that, not only does
anhydroglucitol-6-P withATP ( A )or glucose ( B )as the vari- the catalytic activity reside in the C-terminal half, but this
able substrate.The procedure for preparing the particulate fraction activity exhibits essentially the same inhibition patterns by
of the intact enzyme and for assaying hexokinase activity are de1,5-AnGlc-6-P as thecomplete 100-kDa parent enzyme (Figs.
scribed under “Methods.” The specific activity of the particulate
fraction was 220 nmol of Glc-6-P formed per min/mg of protein. The 5 and 6). Moreover, the N-terminal half is devoid of any
concentration of fixed substrates, glucose (in Fig. 5 A ) and ATP (in catalytic activity. These results provide direct evidence that
Fig. 5B), were 2.2 and 6.4 mM, respectively. 1,5-Anhydroglucitol-6-P the C-terminal half of the enzyme is specialized not only for
was 0 (0),0.0174 (o), 0.0348 (A), or 0.087 (0)mM when ATP was catalysis but for product inhibition as well. Taken together,
variable substrate and 0 (0),0.174 (o), 0.348 (A), or 0.696 (0)mM
when glucose was variable substrate. Inset, a secondary plot of the data obtained from site-directed mutagenesis and from overslopes (for ATP) andof the intercepts (for glucose) derived from the expressed C- and N-terminal halves of hexokinase, indicate
that the role of the N-terminal half of the enzyme is not to
double-reciprocal plot versus inhibitor concentrations.
provide a locus for an “allosteric” product inhibitory site.
To furtherinvestigate whether the N-terminal region binds
appears to be specific as pretreatment of the N-terminal
ATP and Glc-6-P, the isolated overexpressed fragment was
fragment with ATP abolishes the binding.
subjected to further analysis and shown to interact with both
DISCUSSION
ligands (Fig. 7). These findings lead us to believe that the NOne of the most important properties of 100-kDa mam- terminal half, whichisknown to contain a 12-amino acid
malian hexokinases is their capacity to be product inhibited region at its N terminus,believed to be involvedin membrane
by Glc-6-P, in contrastto the50-kDa yeast and liver enzymes binding (14), may be essential for other reasons. One possi(29). The extensive amino acid sequence similarity between bility is that the N-terminal half is required so that the
the two halves of the 100-kDa hexokinases (designated as N- relative levels of soluble and membrane bound forms can be
and C-) have suggested that these proteins evolved from an regulated in the cell by Glc-6-P and/or ATP. Significantly, it
ancestral 50-kDa yeast-like hexokinase by a process of gene is well established that both Glc-6-P and ATP can release
duplication and gene fusion (16,17). Significantly, both halves hexokinase from cellular membranes (2, 30). Moreover, the
have predicted glucose and ATP binding domains. In partic- soluble enzyme is more sensitive to product inhibition by Glcular, there is a very highconservation of amino acids ascribed 6-P (31).It is also possible that theN-terminal half serves as
to glucose binding in the two halves of the molecule (10, 12, a spacer between the membrane and the catalytic C-terminal
14). However, the putative ATP binding domain in the C- half of the enzyme, thus allowing the C-terminal half to
Structure-Function
Studies
of Hexokinase
TABLE
I1
Kinetic constants of the intact and the C-terminal h l f of
tumor hexokinase
Procedures for preparation of the intact and the C-terminal half
of hexokinase particulate fractions and hexokinase activity determinations are described under "Methods." The specific activity of the
intact and C-terminal half preparations were respectively, 220 and
112 nmol of Glc-6-P formed per min/mg of protein. The values shown
are averages of two independent experiments. The percent deviation
from the mean did not exceed 10%. app, apparent.
45
T
Q
X
18265
30
1
Intact
hexokinase
Kinetic constant
15
0.95
u
OO
2
I
3
Km(.ppP:
For Glucose ( p
For
1.02
ATP (mM)
u
-
'0
0
01
Id
0.4
nY
&
( w ) :,
,
b
1,5-AnG-6-P us. ATP
88
118
1,5-AnG-6-P us. Glc
180
165
"The apparent K,,, values were calculated from the double-reciprocal (l/u Versus l/s) plots shown in Figs. 5A and 6A (for ATP) and
Figs. 5B and 6B(for glucose), respectively.
* The apparentKi values were calculated from the secondary plots
of the slope (for ATP) and of the intercept (for glucose) versus
inhibitor concentrations shown as insets in Figs. 5A and 6 A , and 5B
and 6B, respectively.
1
200
-
92.5
-
69
-
46
-
30
-
21.5
-
Y
>
66
58
~ )
KDA
,"64
I
C-terminal half
of hexokinase
+
2
3
4
f
132
16
x48@
OO
I;,
2b
i oi o
sb
I / [Glucose] mM-'
'
FIG.6. Inhibition of the C-terminal half of tumor hexokinase by 1,s-anhydroglucitol-6-P with ATP (A) or glucose
( B ) as the variable substrate. The procedure for preparing the
particulate fraction of the C-terminal half, and for assaying hexokinase activity, are described under "Methods." The specific activity of
FIG. 7. Binding of the N-terminal half
of hexokinase to ATP
the particulate fraction was 112 nmol of Glc-6-P formed per min/mg
of protein. The concentration of fixed substrates, glucose (in Fig. 5A) affinity agarose. The experimental details for incubating the Nand ATP (in Fig. 5 B ) , were 2.2 and 6.4mM, respectively. 1,5- terminal half of hexokinase with ATP-agarose is described under
Anhydroglucitol-6-P was 0 (O),
0.0174 (O),0.0348 (A), or 0.087 (0) "Methods." The proteins run on a 10% SDS-PAGE gel were immumM when ATP was variable substrate and0 (O),
0.174 (O),0.348 (A), noblotted and probed also as detailed under "Methods." Lanes 1 and
or 0.696 (0)mM when glucose was variable substrate. Inset, a sec- 2, proteins elutedfrom the matrix by heating at 95 "C; lone 3, proteins
ondary plot of the slopes (for ATP) andof the intercepts (for glucose) eluted with ATP; lone 4, proteins eluted with Glc-6-P. The position
of a major single band corresponding to that of the N-terminal half
derived from the double-reciprocal plot uersus inhibitor concentranear 50 kDa is indicated by arrowheads. Protein recovery from the
tions.
ATP affinity matrix was greater than 90%. On the left are shown the
positions for molecular size markers (in kDa): myosin (200), phosinteract more readily with the next glycolytic enzyme, i.e. phorylase b (92.5), bovine serum albumin (69), ovalbumin (46), carbonic anhydrase (30), and trypsin inhibitor (21.5).
phosphoglucose isomerase.
In this study, the complete hexokinase molecule and its Cand N-terminal halves were derived by overexpressing the
respective proteins in E. coli and sedimenting the particulate
fraction. It couldbe argued that in such preparations the
presence of improperly folded molecules
within the population
may affect the interpretations described above. We believe
this unlikelyfortworeasons.
First, mutations in the Nterminal half of hexokinase were studied on the complete
enzyme and anactivity assay was usedto monitor the results.
Therefore, the effect of mutations only on active enzymes
within the population were being monitored. Second, the
isolated N-terminal fragment did bind ATP and Glc-6-P, a
finding that would not be expected if this fragment were
unfolded. However, in the latter experiment we cannot rule
out thepossibility that theN-terminal fragment was partially
unfolded.
Consistent with some of the observations made in this
study, White and Wilson (18)have shownusing a biochemical
approach that the C-terminal half of brain hexokinase has
catalytic activity, whereas the N-terminal half is inactive.
The
mutational approach described here supports this conclusion
as site-directed changes of conserved amino acidresidues
within the N-terminal half of the tumor enzyme, predictedto
18266
Structure-Function
Studies
of Hexokinase
interact with glucose, have very little effect on catalytic activ- presented in Fig. LA.Also, we thank Drs. Philip Thomas,Young Hee
ity of the enzyme, and the N-terminal half alone is inactive. KO,Michael Baumann, and Michael Lebowitz for helpful suggestions
Significantly also, while this work was being processed for while this work was in progress.
REFERENCES
publication, Baijal and Wilson (32), using COS-1 cells to
express the Type I brain enzyme, confirmed our earlier site1. Bustamante, E., Morris, H. P., and Pedersen, P. L. (1981) J. Bwl. Chem.
266,8699-8704
directed mutagenesis work on the C-terminal half of hexoki2. Parry, D.M., and Pedersen, P. L. (1983) J. Biol. Chem. 268,10904-10912
3. Arora, K. K., and Pedersen, P. L. (1988) J. Biol. Chem. 263,17422-17428
nase (22). They showed also that conversion of the conserved
4. Katzen. H.M.. and Schimke, R. T. (1965) Proc. Nutl. Acad. Scr. U. S. A.
serine (Ser155)in the N-terminal region of the brain enzyme
64, i218-1225
5. Nakashima, R. A,, Pappi, M. G., Scott, L. J., and Pedersen, P. L. (1988)
has littleor no effect on catalysis or product inhibition. These
Cancer Res. 48,913-919
workers did not examine the effect of mutations on other
6. Grossbard, L., and Schimke R. T. (1966) J. Biol. Chem. 241,3546-3560
7.
Frohlich,
K.-U., Entian, KID., and Mecke, D. (1985) Gene (Amst.) 36,
amino acid residues within the N-terminal half of the enzyme.
105-111
Also consistent with the work reported here are the recent
8. Kqtezki, E., Entian, K-D., and Mecke, D. (1985) Gene (Amst.) 3 9 , 95IOZ
studies of Magnani et ul. (33), demonstrating that the C9. Nishi, S., Seino, S., and Bell, G. I. (1988) Biochem. Biophys. Res. Commun.
terminal half of the Type I human placental hexokinase is
167,937-943
10. Schwab, D. A., and Wilson, J. E. (1989) Proc. Nutl. Acad. Sci. U. S. A. 86,
product-inhibited by Glc-6-P.
2563-2567
Finally, two other points seem relevant to the work pre- 11. Th!j_len,A. P., and Wilson, J. E. (1991) Arch. Biochem. Biophys. 286,645631
sented here. First, there hasbeen controversy in the literature 12. Andreone,
T.L., Printz, R. L., Pilkis, S. J., Magnuson, M. A., and Granner,
D. K. (1989) J. Biol. Chem. 264,363-369
for many years as to whether mammalian hexokinases exhibit
D.A., and Wilson, J. E. (1991) Arch. Biochem. Biophys. 286,365an allosteric site for product inhibition by Glc-6-P or whether 13. Schwab,
370
inhibition by this product can be accounted for by binding to 14. Arora, K. K., Fanciulli, M., and Pedersen, P. L. (1990) J. Biol. Chem. 2 6 6 ,
6481
- -- -S -A-" W
the catalytic site (29, 34-37). Certainly, the results presented 15. Griffin,
L. D., Gelb, B.D., Wheeler, D. A., Davison, D., Adams, V., and
McCabe,
E. R. B. (1991) Genomrcs 1 1 , 1014-1024
here emphasize that the C-terminal half of the molecule
16. Colowick, S. P. (1973) in The Enzymes (Boyer, P. D., ed) 3rd Ed., Vol. 9,
separated from the N-terminal half contains both a catalytic
pp. 1-48, Academic Press, New York
Ureta, T. (1982) Comp. Biochem. Physiol. 7 1 B , 549-555
site and a site for product inhibition and that mutations in 17.
18. White, T.K., and Wilson, J. E. (1989) Arch. Biochem. Biuphys. 274,375293
the N-terminal half of the complete enzyme have little effect
T.K., and Wilson, J. E. (1987) Arch. Biochem. Biophys. 2 6 9 , 402both on catalysis and product inhibition. Moreover, kinetic 19. WE@,
PAL
analyses reveal that the Glc-6-P analog 1,5-AnG-6-P is lin- 20. Anderson, C. M.,Stenkamp, R. E., McDonald, R. C., and Steitz, T. A.
(1978) J. Mol. Bml. 123,207-219
early competitive with the substrate ATP both in the intact 21. Bennett,
W. S., Jr., and Steitz, T. A. (1980) J. Mol. Biol. 1 4 0 , 211-230
enzyme (Fig. 5A) and in theC-terminal half (Fig. 6A). These 22. Arora, K. K., Filburn, C. F., and Pedersen, P. L. (1991) J. Biol. Chem. 2 6 6 ,
5359-5362
data, together with recent NMR data on the brain enzyme 23. Sambrook,
J., Fritsch, E. F., and Maniatis, T.(1989) Molecular Cloni
Laboratory Manual, Cold Spring Harbor Laboratow, Cold Spring Ha%$
indicating that glucose and Glc-6-P bind very nearly to the
NY
same region of the enzyme (38,39), renderunlikely the earlier 24. Taylor, J. W., Ott, J., and Eckstein, F. (1985) Nucleic Acids Res. 13,8764suggestions (18,19,34) thatmammalian 100-kDa hexokinases 25. S,;a; 11711.5 F., Nicklen, S., and Coulson, A. R. (1977) Proc. Nutl. Acud. Sci.
exhibit an allosteric site involved in product inhibition which
U. S. A. 74,5463-5467
R. A., Mandelstam, P., and Crane, R. K. (1959) Arch. Biochem.
is completely separate and distinct from the catalytic site. 26. Ferrari,
Biophys. 80,372-377
The second point relatedto the
first concernsthe evolutionary 27. Bartlett G. R. (1959) J. Bwl. Chem. 234,466-468
28. Laemmd. U. K. (1970) Nuture 227,680-685
events that led tothe mammalian 100-kDa hexokinases. 29.
Wilson, J. E. (1984) in Re ulution oiCurbohydrate Metabolism (Beitner, R.,
ed) Vol. 1, pp. 45-85, C%C Press, Inc., Boca Raton, FL
Clearly, the results presented in this study support the view
Parry, D. M., and Pedersen P. L. (1990) J. Biol. Chem. 2 6 6 , 1059-1066
that these gene duplication products did not arise from 50- 30.
31. Bustamante, E., and Pederien, P. L. (1980) Biochemistry 19,4972-4977
kDa unregulated yeast-like precursors which duplicated to 32. Baijal, M., and Wilson, J. E. (1992) Arch. Biochem. Bio hys 289,271-278
33. Magnani M., Bianchi, M., Casabianca, A., Stocchi, V.,&aniele, A., Altruda,
provide a separate product inhibitory site. Rather, they seem
F., Ferrone, M., and Silengo, L. (1992) Biochem. J. 286,193-199
34.
Crane, R. K., and Sols, A. (1954) J. Bml. Chem. 210,597-606
to have arisen from 50-kDa hexokinase species like those 35. Fromm,
H. J., and Zewe, V. (1962) J. Biol. Chem. 237,1661-1667
found in marine organisms (40, 41) which had acquired al- 36. Ellison, W. R.,Lueck, J. D., and Fromm, H. J. (1975) J. Biol. Chem. 2 6 0 ,
1864-1871
ready the capacity for product regulation, the purpose of gene 37. Solheim, L. P., andFromm, H.J. (1981) Arch. Biochem. Biophys. 211,9299
duplication in this case then being to serve another role,
38. Mehta, A., Jarori, G. K., and Kenkare, U. W. (1988) J . Biol. Chem. 2 6 3 ,
perhaps in regulating membrane binding as indicated above.
15492-15497
Acknowledgments-We are grateful to Drs. L. Mario Amzel amd
Mario Bianchet for help in obtaining the computer graphics representation of the predicted structure of a mammalian hexokinase
39. Jarori, G. K., Iyer, S. B., Kasturi, S. R., and Kenkare, U. W. (1990) Eur. J.
Biochem. 188,9-14
40. Mochizuki, Y., and Hori, S. H. (1977) J. Biochem. (Tokyo) 81,1849-1855
41. Mochizuki, Y. (1981) Comp. Biochem. Physiol. 70B,745-751
42. Carson, M. (1987) J. Mol. Graphics 6 , 103-106